Hydrophilic Interaction Liquid Chromatography PDF

© 2011 by Taylor and Francis Group, LLC
CHROMATOGRAPHIC SCIENCE SERIES
A Series of Textbooks and Reference Books
Editor:
NELU GRINBERG
Founding Editor:
JACK CAZES
1. Dynamics of Chromatography: Principles and Theory,

J. Calvin Giddings
2. Gas Chromatographic Analysis of Drugs and Pesticides,
Benjamin J. Gudzinowicz
3. Principles of Adsorption Chromatography: The Separation of Nonionic
Organic Compounds, Lloyd R. Snyder
4. Multicomponent Chromatography: Theory of Interference,
Friedrich Helfferich and Gerhard Klein
5. Quantitative Analysis by Gas Chromatography, Josef Novák
6. High-Speed Liquid Chromatography, Peter M. Rajcsanyi
and Elisabeth Rajcsanyi
7. Fundamentals of Integrated GC-MS (in three parts),
Benjamin J. Gudzinowicz, Michael J. Gudzinowicz,
and Horace F. Martin
8. Liquid Chromatography of Polymers and Related Materials, Jack Cazes
9. GLC and HPLC Determination of Therapeutic Agents (in three parts),
Part 1 edited by Kiyoshi Tsuji and Walter Morozowich, Parts 2 and 3
edited by Kiyoshi Tsuji
10. Biological/Biomedical Applications of Liquid Chromatography,
edited by Gerald L. Hawk
11. Chromatography in Petroleum Analysis, edited by Klaus H. Altgelt
and T. H. Gouw
12. Biological/Biomedical Applications of Liquid Chromatography II,
13. Liquid Chromatography of Polymers and Related Materials II,
edited by Jack Cazes and Xavier Delamare
14. Introduction to Analytical Gas Chromatography: History, Principles,
and Practice, John A. Perry
15. Applications of Glass Capillary Gas Chromatography, edited by
Walter G. Jennings
16. Steroid Analysis by HPLC: Recent Applications, edited by
Marie P. Kautsky
17. Thin-Layer Chromatography: Techniques and Applications,
Bernard Fried and Joseph Sherma
18. Biological/Biomedical Applications of Liquid Chromatography III,
19. Liquid Chromatography of Polymers and Related Materials III,
edited by Jack Cazes

20. Biological/Biomedical Applications of Liquid Chromatography,
21. Chromatographic Separation and Extraction with Foamed Plastics
and Rubbers, G. J. Moody and J. D. R. Thomas
22. Analytical Pyrolysis: A Comprehensive Guide, William J. Irwin
23. Liquid Chromatography Detectors, edited by Thomas M. Vickrey
24. High-Performance Liquid Chromatography in Forensic Chemistry,
edited by Ira S. Lurie and John D. Wittwer, Jr.
25. Steric Exclusion Liquid Chromatography of Polymers, edited by
Josef Janca
26. HPLC Analysis of Biological Compounds: A Laboratory Guide,
William S. Hancock and James T. Sparrow
27. Affinity Chromatography: Template Chromatography of Nucleic Acids
and Proteins, Herbert Schott
28. HPLC in Nucleic Acid Research: Methods and Applications,
edited by Phyllis R. Brown
29. Pyrolysis and GC in Polymer Analysis, edited by S. A. Liebman
and E. J. Levy
30. Modern Chromatographic Analysis of the Vitamins, edited by
André P. De Leenheer, Willy E. Lambert, and Marcel G. M. De Ruyter
31. Ion-Pair Chromatography, edited by Milton T. W. Hearn
32. Therapeutic Drug Monitoring and Toxicology by Liquid
Chromatography, edited by Steven H. Y. Wong
33. Affinity Chromatography: Practical and Theoretical Aspects, Peter Mohr
and Klaus Pommerening
34. Reaction Detection in Liquid Chromatography, edited by Ira S. Krull
35. Thin-Layer Chromatography: Techniques and Applications,
Second Edition, Revised and Expanded, Bernard Fried
and Joseph Sherma
36. Quantitative Thin-Layer Chromatography and Its Industrial
Applications, edited by Laszlo R. Treiber
37. Ion Chromatography, edited by James G. Tarter
38. Chromatographic Theory and Basic Principles, edited by
Jan Åke Jönsson
39. Field-Flow Fractionation: Analysis of Macromolecules and Particles,
Josef Janca
40. Chromatographic Chiral Separations, edited by Morris Zief
and Laura J. Crane
41. Quantitative Analysis by Gas Chromatography, Second Edition,
Revised and Expanded, Josef Novák
42. Flow Perturbation Gas Chromatography, N. A. Katsanos
43. Ion-Exchange Chromatography of Proteins, Shuichi Yamamoto,
Kazuhiro Naka-nishi, and Ryuichi Matsuno
44. Countercurrent Chromatography: Theory and Practice,
edited by N. Bhushan Man-dava and Yoichiro Ito
45. Microbore Column Chromatography: A Unified Approach
to Chromatography, edited by Frank J. Yang
46. Preparative-Scale Chromatography, edited by Eli Grushka
47. Packings and Stationary Phases in Chromatographic Techniques, edited
by Klaus K. Unger
48. Detection-Oriented Derivatization Techniques in Liquid
Chromatography, edited by Henk Lingeman and Willy J. M. Underberg
49. Chromatographic Analysis of Pharmaceuticals, edited by
John A. Adamovics

50. Multidimensional Chromatography: Techniques and Applications,
edited by Hernan Cortes
51. HPLC of Biological Macromolecules: Methods and Applications,
edited by Karen M. Gooding and Fred E. Regnier
52. Modern Thin-Layer Chromatography, edited by Nelu Grinberg
53. Chromatographic Analysis of Alkaloids, Milan Popl, Jan Fähnrich,
and Vlastimil Tatar
54. HPLC in Clinical Chemistry, I. N. Papadoyannis
55. Handbook of Thin-Layer Chromatography, edited by Joseph Sherma
and Bernard Fried
56. Gas–Liquid–Solid Chromatography, V. G. Berezkin
57. Complexation Chromatography, edited by D. Cagniant
58. Liquid Chromatography–Mass Spectrometry, W. M. A. Niessen
and Jan van der Greef
59. Trace Analysis with Microcolumn Liquid Chromatography,
Milos KrejcI
60. Modern Chromatographic Analysis of Vitamins: Second Edition,
edited by André P. De Leenheer, Willy E. Lambert, and Hans J. Nelis
61. Preparative and Production Scale Chromatography, edited by
G. Ganetsos and P. E. Barker
62. Diode Array Detection in HPLC, edited by Ludwig Huber
and Stephan A. George
63. Handbook of Affinity Chromatography, edited by Toni Kline
64. Capillary Electrophoresis Technology, edited by Norberto A. Guzman
65. Lipid Chromatographic Analysis, edited by Takayuki Shibamoto
66. Thin-Layer Chromatography: Techniques and Applications:
Third Edition, Revised and Expanded, Bernard Fried
and Joseph Sherma
67. Liquid Chromatography for the Analyst, Raymond P. W. Scott
68. Centrifugal Partition Chromatography, edited by Alain P. Foucault
69. Handbook of Size Exclusion Chromatography, edited by Chi-San Wu
70. Techniques and Practice of Chromatography, Raymond P. W. Scott
71. Handbook of Thin-Layer Chromatography: Second Edition,
Revised and Expanded, edited by Joseph Sherma and Bernard Fried
72. Liquid Chromatography of Oligomers, Constantin V. Uglea
73. Chromatographic Detectors: Design, Function, and Operation,
Raymond P. W. Scott
74. Chromatographic Analysis of Pharmaceuticals: Second Edition, Revised
and Expanded, edited by John A. Adamovics
75. Supercritical Fluid Chromatography with Packed Columns: Techniques
and Applications, edited by Klaus Anton and Claire Berger
76. Introduction to Analytical Gas Chromatography: Second Edition,
Revised and Expanded, Raymond P. W. Scott
77. Chromatographic Analysis of Environmental and Food Toxicants, edited
by Takayuki Shibamoto
78. Handbook of HPLC, edited by Elena Katz, Roy Eksteen,
Peter Schoenmakers, and Neil Miller
79. Liquid Chromatography–Mass Spectrometry: Second Edition,
Revised and Expanded, Wilfried Niessen
80. Capillary Electrophoresis of Proteins, Tim Wehr,
Roberto Rodríguez-Díaz, and Mingde Zhu
81. Thin-Layer Chromatography: Fourth Edition, Revised and Expanded,
Bernard Fried and Joseph Sherma

82. Countercurrent Chromatography, edited by Jean-Michel Menet
and Didier Thiébaut
83. Micellar Liquid Chromatography, Alain Berthod
and Celia García-Alvarez-Coque
84. Modern Chromatographic Analysis of Vitamins: Third Edition,
Revised and Expanded, edited by André P. De Leenheer,
Willy E. Lambert, and Jan F. Van Bocxlaer
85. Quantitative Chromatographic Analysis, Thomas E. Beesley,
Benjamin Buglio, and Raymond P. W. Scott
86. Current Practice of Gas Chromatography–Mass Spectrometry,
edited by W. M. A. Niessen
87. HPLC of Biological Macromolecules: Second Edition, Revised
and Expanded, edited by Karen M. Gooding and Fred E. Regnier
88. Scale-Up and Optimization in Preparative Chromatography:
Principles and Bio-pharmaceutical Applications, edited by
Anurag S. Rathore and Ajoy Velayudhan
89. Handbook of Thin-Layer Chromatography: Third Edition,
Revised and Expanded, edited by Joseph Sherma and Bernard Fried
90. Chiral Separations by Liquid Chromatography and Related
Technologies, Hassan Y. Aboul-Enein and Imran Ali
91. Handbook of Size Exclusion Chromatography and Related Techniques:
Second Edition, edited by Chi-San Wu
92. Handbook of Affinity Chromatography: Second Edition, edited by
David S. Hage
93. Chromatographic Analysis of the Environment: Third Edition,
edited by Leo M. L. Nollet
94. Microfluidic Lab-on-a-Chip for Chemical and Biological Analysis
and Discovery, Paul C.H. Li
95. Preparative Layer Chromatography, edited by Teresa Kowalska
and Joseph Sherma
96. Instrumental Methods in Metal Ion Speciation, Imran Ali
and Hassan Y. Aboul-Enein
97. Liquid Chromatography–Mass Spectrometry: Third Edition,
Wilfried M. A. Niessen
98. Thin Layer Chromatography in Chiral Separations and Analysis,
edited by Teresa Kowalska and Joseph Sherma
99. Thin Layer Chromatography in Phytochemistry, edited by
Monika Waksmundzka-Hajnos, Joseph Sherma, and Teresa Kowalska
100. Chiral Separations by Capillary Electrophoresis,edited by
Ann Van Eeckhaut and Yvette Michotte
101. Handbook of HPLC: Second Edition, edited by Danilo Corradini and
consulting editor Terry M. Phillips
102. High Performance Liquid Chromatography in Phytochemical Analysis,
edited by Monika Waksmundzka-Hajnos and Joseph Sherma
103. Hydrophilic Interaction Liquid Chromatography (HILIC) and Advanced
Applications, edited by Perry G. Wang and Weixuan He

Boca Raton London New York
CRC Press is an imprint of the

Taylor & Francis Group, an informa business

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1
International Standard Book Number: 978-1-4398-0753-8 (Hardback)
This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copy-
right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-
vides licenses and registration for a variety of users. For organizations that have been granted a pho-
tocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com

Contents
Preface.................................................................................................................... xiii
Editors....................................................................................................................... xv
Contributors............................................................................................................xvii
Chapter 1 Aqueous Normal-Phase Chromatography: The Bridge between

Reversed-Phase and HILIC................................................................... 1
Joseph J. Pesek and Maria T. Matyska
Chapter 2 Application of the Calculated Physicochemical Property, Log D,

to Hydrophilic Interaction Chromatography in the Bioanalytical
Laboratory........................................................................................... 27
Eugene P. Kadar, Chad E. Wujcik, and Olga Kavetskaia
Chapter 3 HILIC Behavior of Reversed-Phase/Ion-Exchange Mixed-Mode

Stationary Phases and Their Applications.......................................... 47
Xiaodong Liu and Christopher A. Pohl
Chapter 4 Mechanisms Controlling the Separation and Retention of

Proanthocyanidins and Other Plant Polyphenols in Hydrophilic
Interaction Chromatography............................................................... 77
Akio Yanagida and Yoichi Shibusawa
Chapter 5 Applications of Hydrophilic Interaction Chromatography

to Food Analysis.................................................................................. 89
Leticia Mora, Aleida S. Hernández-Cázares, M-Concepción
Aristoy, Fidel Toldrá, and Milagro Reig
Chapter 6 Hydrophilic Interaction Liquid Chromatography–Mass

Spectrometry (HILIC–MS) of Paralytic Shellfish Poisoning
Toxins, Domoic Acid, and Assorted Cyanobacterial Toxins............ 105
Carmela Dell’Aversano
ix

x Contents
Chapter 7 Application of HILIC for Polar Environmental Contaminants

(Including Pharmaceuticals) in Aquatic Systems.............................. 133
Alexander L.N. van Nuijs, Isabela Tarcomnicu, Hugo Neels,
and Adrian Covaci
Chapter 8 Hydrophilic Interaction Chromatography–Mass Spectrometry:

An Environmental Application......................................................... 157
Kerry M. Peru, Sandra L. Kuchta, John V. Headley,
and Allan J. Cessna
Chapter 9 Clinical Applications of Hydrophilic Interaction Liquid

Chromatography................................................................................ 177
Ping Wang
Chapter 10 Dimethylarginine Analysis by HILIC–MS/ MS and Its

Applications in Clinical Chemistry and In Vivo Animal Models.... 203
O. D’Apolito, G. Paglia, M. Zotti, F. Tricarico, L.Trabace,
and G. Corso
Chapter 11 Polar Functional Groups for HILIC Method..................................... 225

Zhigang Hao
Chapter 12 Analysis of Pharmaceutical Impurities Using Hydrophilic

Interaction Liquid Chromatography.................................................. 259
Mingjiang Sun and David Q. Liu
Chapter 13 Fast In-Process Method for the Determination of Ioversol

and Related Polar Compounds by Hydrophilic Interaction
Chromatography (HILIC) and UPLC............................................... 291
Yuming Chen, Xinqun Huang, Shaoxiong Huang,
and Michael Matchett
Chapter 14 Retention Behavior of Hydrazine Derivatives in HILIC Mode........309

Juris Hmelnickis, Igors Susinskis, and Kaspars Kokums
Chapter 15 HILIC Retention Behavior and Method Development for Highly

Polar Basic Compounds Used in Pharmaceutical Synthesis............. 345
Minhui Ma and Min Liu

Contents xi
Chapter 16 Retention of Polar Acidic and Basic Drugs by HILIC...................... 373

Mohammed Shahid Ali and Aamer Roshanali Khatri
Chapter 17 Retention and Selectivity of Polar Stationary Phases for

Hydrophilic Interaction Chromatography......................................... 401
Yong Guo
Chapter 18 HILIC-MS/MS for the Determination of Polar Bioactive

Substances: Representative Applications in the Fields
of Pharmacokinetics and Metabolic Profiling................................... 427
Yannis L. Loukas and Yannis Dotsikas
Chapter 19 Method Development and Analysis of Mono- and

Diphosphorylated Nucleotides by HILIC HPLC-ESI-MS................ 445
Samuel H. Yang, Hien P. Nguyen, and Kevin A. Schug
Chapter 20 Hydrophilic Interaction Liquid Chromatography–Based

Enrichment Protocol Coupled to Mass Spectrometry
for Glycoproteome Analysis..............................................................469
Cosima Damiana Calvano
Chapter 21 Development and Application of Methods for Separation

of Carbohydrates by Hydrophilic Interaction Liquid
Chromatography..............................................................................491
Goran Karlsson
Chapter 22 Hydrophilic Interaction Liquid Chromatography in the

Characterization of Glycoproteins.................................................... 523
Joanne E. Nettleship
Chapter 23 Analysis of Protein Glycosylation and Phosphorylation Using

HILIC-MS......................................................................................... 551
Morten Thaysen-Andersen, Kasper Engholm-Keller,
and Peter Roepstorff
Index....................................................................................................................... 577

Preface
This book on hydrophilic interaction liquid chromatography (HILIC) applications is
a result of the increasing interest among the separation science community in the lat-
est chromatography technologies for isolation and purification of various chemicals.
HILIC is a variation of normal-phase chromatography with the added advantage that
organic solvents that are miscible in water can be used. It uses polar materials, such
as amino, cyano, diol, and silanol, as its stationary phase and, thus, is sometimes
referred to as “reverse reversed-phase” or “aqueous normal phase” chromatography.
HILIC provides the following benefits compared to other chromatographic separa-
tion modes:
• High organic content mobile phases result in a lower operating back pres-
sure, allowing higher flow rates to be used in high-throughput analysis.
• The high organic solvent concentration in the mobile phase leads to a higher
sensitivity for LC/MS analyses due to significantly increased ionization.
• Significant improvement of peak shape and sensitivity results in more
accurate and precise quantitation of polar compounds such as peptides and
nucleic acids.
• Polar analytes that would be un-retained by reversed-phase chromatogra-
phy are retained by HILIC.
• Capability of direct injection of solid phase extraction (SPE) or liquid–
liquid extracts into the HILIC columns increases analytical throughput
and facilitates sample preparation.
The HILIC concept was first introduced by Dr. Andrew Alpert in his 1990 paper.
A large number of scientific papers on this subject have been published since then.
This book comprehensively and systematically describes the new technology and
provides detailed information and discussion on the most advanced HILIC appli-
cations in the fields of environmental sciences, food analysis, clinical chemistry,
pharmaceutical research, and biotechnology discovery. Although the theory behind
HILIC is not fully understood and commercial HILIC columns have limited avail-
ability, the extensive applications that we witness today have already made HILIC a
unique device in the chromatography toolbox for most separation scientists. We hope
that our readers will find this book to be a valuable resource for their projects rang-
ing from academic research to industrial applications.
We are indebted to several people who assisted us during the preparation and
editing of this book. We were extremely fortunate to have had the cooperation of
dedicated, well-known contributing authors. Their relentless efforts and sincere
scientific drive have made this book possible.
xiii

Editors
Dr. Perry G. Wang is a research chemist at the Office of Regulatory Science, Center
for Food Safety and Applied Nutrition, Food and Drug Administration (FDA). His
interests include analytical method development and validation for drugs and con-
stituents in foods and cosmetic products using advanced instrumentation. His exper-
tise focuses on high-throughput drug analysis by LC/MS/MS for the pharmaceutical
industry.
Dr. Wang has recently published two books: High-Throughput Analysis in the
Pharmaceutical Industry (published by CRC Press in October 2008) and Monolithic
Chromatography and Its Modern Applications (published by ILM Publications in
October 2010). He is currently coediting another book entitled Identification and
Analysis of Counterfeit and Substandard Pharmaceuticals with ILM Publications,
which is scheduled for publication in May 2011.
Dr. Wang has been invited to prepare, organize, and preside over symposia for
the Pittsburg Conference (PittCon) and American Chemistry Society (ACS) annual
meetings. He has been an invited speaker at numerous international conferences
including the PittCon, the Federation of Analytical Chemistry and Spectroscopy
Societies (FACSS), the Beijing Conference and Exhibition on Instrumental Analysis
(BCEIA), and the International Symposium on Chemical Biology and Combinatorial
Chemistry (ISCBCC). He has also been invited to teach short courses on high-Â�
throughput method development for drug analysis by LC/MS/MS at the PittCon,
ACS, and at the Eastern Analytical Symposium (EAS). His current research focuses
on developing analytical methods for the determination of constituents in cosmetics
and dietary supplements.
Dr. Wang received his BS in chemistry from Shandong University, and his MS
and PhD in environmental engineering from Oregon State University, Corvallis,
Oregon.
Dr. Weixuan He is currently an associate director, Product and Process Development,

at Meda Pharmaceuticals (formerly MedPointe Pharmaceuticals) located in North
Brunswick, New Jersey. He is responsible for designing and conducting analytical
investigations and experiments to support Meda Pharmaceuticals’ product develop-
ment, as well as providing technical services to ensure commercial manufacturing
processes are in compliance with regulatory requirements. He also focuses on devel-
oping analytical methods using a variety of new technologies to support pharmaceu-
tical product development.
Prior to joining Meda Pharmaceuticals in 2002, Dr. He was a senior research
investigator at Bristol Myers Squibb, specializing in process development for isola-
tion and purification of active pharmaceutical ingredient (API) from reaction mix,
natural products from fermentation or biotransformation media, and diastereomer
from stereoisomer mixtures. He successfully isolated and characterized various
xv

xvi Editors
process impurities, degradants, Â�metabolites, and natural products using modern iso-
lation technologies and spectroscopic methods.
Before starting his career in the pharmaceutical industry in 1994, Dr. He earned
his doctorate degree in organic chemistry from Oregon State University in Corvallis,
Oregon, and completed his postdoctoral training in biomedical research at The Johns
Hopkins University in Baltimore, Maryland.

Contributors
Mohammed Shahid Ali O. D’Apolito
Jamjoom Pharmaceuticals Company Department of Biomedical Sciences
Limited University of Foggia
Jeddah, Saudi Arabia Foggia, Italy
M-Concepción Aristoy Carmela Dell’Aversano

Consejo Superior de Investigaciones Dipartimento di Chimica delle Sostanze
Científicas Naturali
Instituto de Agroquímica y Tecnología Università di Napoli Federico II
de Alimentos Napoli, Italy
Valencia, Spain
Yannis Dotsikas
Cosima Damiana Calvano Department of Pharmaceutical Chemistry
Department of Chemistry University of Athens
University of Bari Aldo Moro Athens, Greece
Bari, Italy
Kasper Engholm-Keller
Department Biochemistry and
Allan J. Cessna
Molecular Biology
Agriculture and Agri-Food Canada
University of Southern Denmark
and Odense, Denmark
Environment Canada
Yong Guo
National Hydrology Research Centre
The Janssen Pharmaceutical Companies
Saskatoon, Saskatchewan, Canada
Johnson & Johnson
Raritan, New Jersey
Yuming Chen
Forest Laboratories, Inc. Zhigang Hao
Hauppauge, New York Global Analytical Science Department
Colgate-Palmolive Company
G. Corso Piscataway, New Jersey
Department of Biomedical Sciences
University of Foggia John V. Headley
Foggia, Italy Environment Canada
Water Science and Technology
Adrian Covaci Directorate
Toxicological Centre Aquatic Ecosystem Protection Research
University of Antwerp Division
Antwerp, Belgium Saskatoon, Saskatchewan, Canada
xvii

xviii Contributors
Aleida S. Hernández-Cázares Sandra L. Kuchta

Consejo Superior de Investigaciones Health Canada
Científicas Ottawa, Ontario, Canada
Instituto de Agroquímica y Tecnología
de Alimentos David Q. Liu
Valencia, Spain GlaxoSmithKline
King of Prussia, Pennsylvania
Juris Hmelnickis
Grindeks Min Liu
Amgen, Inc.
and Thousand Oaks, California
Faculty of Chemistry
University of Latvia Xiaodong Liu
Riga, Latvia Dionex Corporation
Sunnyvale, California
Shaoxiong Huang
Covidien, Inc. Yannis L. Loukas
St. Louis, Missouri Department of Pharmaceutical Chemistry
University of Athens
Athens, Greece
Xinqun Huang
Covidien, Inc.
Minhui Ma
St. Louis, Missouri
Amgen, Inc.
Thousand Oaks, California
Eugene P. Kadar
Pfizer, Inc. Michael Matchett
Groton, Connecticut Covidien, Inc.
St. Louis, Missouri
Goran Karlsson
Octapharma AB Maria T. Matyska
Stockholm, Sweden Department of Chemistry
San Jose State University
Olga Kavetskaia San Jose, California
Abbott Laboratories
Abbott Park, Illinois Leticia Mora
Consejo Superior de Investigaciones
Científicas
Aamer Roshanali Khatri Instituto de Agroquímica y Tecnología
Jamjoom Pharmaceuticals Company de Alimentos
Limited Valencia, Spain
Jeddah, Saudi Arabia
Hugo Neels
Kaspars Kokums Toxicological Centre
Grindeks University of Antwerp
Riga, Latvia Antwerp, Belgium

Contributors xix
Joanne E. Nettleship Kevin A. Schug

Oxford Protein Production Facility Department of Chemistry and
Wellcome Trust Centre for Human Biochemistry
Genetics The University of Texas at Arlington
University of Oxford Arlington, Texas
Oxford, United Kingdom
Yoichi Shibusawa
Hien P. Nguyen Division of Pharmaceutical and
Department of Chemistry and Biomedical Analysis
Biochemistry Tokyo University of Pharmacy and
The University of Texas at Arlington Life Sciences
Arlington, Texas Tokyo, Japan
Mingjiang Sun
G. Paglia
GlaxoSmithKline
Department of Biomedical Sciences
King of Prussia, Pennsylvania
University of Foggia
Foggia, Italy Igors Susinskis
Grindeks
Kerry M. Peru Riga, Latvia
Environment Canada
Water Science and Technology Isabela Tarcomnicu
Directorate Toxicological Centre
Aquatic Ecosystem Protection Research University of Antwerp
Division Antwerp, Belgium
Saskatoon, Saskatchewan, Canada
Morten Thaysen-Andersen
Joseph J. Pesek Department of Chemistry and
Department of Chemistry Biomolecular Sciences
San Jose State University Macquarie University
San Jose, California Sydney, New South Wales, Australia
Fidel Toldrá
Christopher A. Pohl
Consejo Superior de Investigaciones
Dionex Corporation
Científicas
Sunnyvale, California
Instituto de Agroquímica y Tecnología
de Alimentos
Milagro Reig
Valencia, Spain
Instituto de Ingeniería de Alimentos
para el Desarrollo L. Trabace
Universidad Politécnica de Valencia Department of Biomedical Sciences
Valencia, Spain University of Foggia
Foggia, Italy
Peter Roepstorff
Department of Biochemistry and F. Tricarico
Molecular Biology Department of Biomedical Sciences
University of Southern Denmark University of Foggia
Odense, Denmark Foggia, Italy

xx Contributors
Alexander L.N. van Nuijs Akio Yanagida

Toxicological Centre Division of Pharmaceutical and
University of Antwerp Biomedical Analysis
Antwerp, Belgium Tokyo University of Pharmacy
and Life Sciences
Tokyo, Japan
Ping Wang
The Methodist Hospital
Samuel H. Yang
Houston, Texas
Department of Chemistry and
and Biochemistry
The University of Texas at Arlington
Weill Cornell Medical College
Arlington, Texas
New York, New York
M. Zotti
Chad E. Wujcik Department of Biomedical Sciences
Monsanto University of Foggia
St. Louis, Missouri Foggia, Italy

1 Aqueous Normal-Phase
Chromatography:
The Bridge between
Reversed-Phase
and HILIC
Joseph J. Pesek and Maria T. Matyska
Contents
1.1 Fundamental Features of Aqueous Normal-Phase Retention............................ 2
1.2 Modes of ANP................................................................................................... 3
1.2.1 ANP 1.................................................................................................... 3
1.2.2 ANP 2.................................................................................................... 4
1.2.3 ANP 3.................................................................................................... 6
1.3 Chromatographic Properties in ANP................................................................ 7
1.3.1 Efficiency...............................................................................................7
1.3.2 Temperature Behavior...........................................................................7
1.3.3 Repeatability.......................................................................................... 9
1.3.4 Re-Equilibration.................................................................................. 10
1.4 Conditions for Aqueous Normal Phase Retention........................................... 11
1.4.1 Column................................................................................................ 11
1.4.2 Organic Component of Mobile Phase.................................................. 11
1.4.3 Mobile Phase Additive......................................................................... 12
1.4.4 Sample Solvent Additive...................................................................... 13
1.4.5 Gradients.............................................................................................. 14
1.5 Applications of Aqueous Normal Phase Retention......................................... 15
1.5.1 Amino Acids........................................................................................ 15
1.5.2 Carbohydrates...................................................................................... 16
1.5.3 Organic Acids...................................................................................... 17
1.5.4 Nucleotides..........................................................................................20
1.5.5 Other Analytes..................................................................................... 22
1.5.6 Universal Analytical Method for Hydrophilic Metabolites................. 25
1.6 Conclusions...................................................................................................... 25
Acknowledgment......................................................................................................25
References.................................................................................................................26

2 Hydrophilic Interaction Liquid Chromatography and Advanced Applications
1.1 F undamental Features of Aqueous

Normal-Phase Retention
Reversed-phase (RP) high-performance liquid chromatography (HPLC) has been
the predominant method for a vast array of separations over the last 30 years. Its
principles have been applied to the analysis of compounds that include small organic
and inorganic molecules as well as high molecular weight species such as proteins
and peptides. However, RP HPLC is not a universal approach for all separation prob-
lems. While many molecules have sufficient hydrophobicity to result in retention
on typical RP materials such as octyl (C8) or octadecyl (C18), others are too hydro-
philic so that under most circumstances they elute at or near the column void vol-
ume.1 This is especially the case when molecules are very polar or ionic. In order to
enhance the retention of hydrophilic analytes, several alternative chromatographic
meÂ�thods have been developed. The simplest is to make a derivative of the hydro-
philic compound so that the product is more hydrophobic in nature.2,3 This can often
be a time-consuming process that has the potential for error if the reaction is not
quantitative. Another technique is to add ion-pairing reagents to the mobile phase in
the case of ionic species so that the solutes being analyzed are neutral.4 Both of these
approaches utilize RP as the chromatographic technique since the resultant analyte
becomes more hydrophobic and thus can be retained on stationary phases like C8 or
C18. Another strategy for retention of polar/ionic compounds is ion-exchange.5–7 In
this case, the stationary phase retains a charge that can provide retention capabili-
ties for hydrophilic species. Unfortunately, these methods often result in the use of
mobile phases that are not compatible with mass spectrometry, a detection method
that is becoming increasingly more prevalent.
These restrictions and inconveniences have led to the development of hydrophilic
interaction liquid chromatography (HILIC) as a possible solution.8–14 Since most
highly polar and ionic species are generally soluble only in solvents such as water
or perhaps methanol, a chromatographic method must be developed that can utilize
these solvents. The retention of the analytes must be accomplished by utilizing a
polar stationary phase. Thus HILIC encompasses mobile phases with some polar
solvent (usually water) component and stationary phases that also are polar in nature
such as silica, amino, diol, or cyano-bonded organic compounds, and zwitterionic
materials that are bonded to support surfaces or are part of a polymer. Attraction and
hence retention is facilitated by using low amounts of the polar solvent and a high
amount of the organic component (typically acetonitrile [ACN]) in the mobile phase.
However, in many cases, additives which are not compatible with mass spectrometry
must be included in the mobile phase, retention times between runs are not reproduc-
ible, or the equilibration time between runs for gradient analyses (usually necessary
for complex mixtures) is excessively long.
A new and even more promising approach to solving the problem of hydrophilic
retention in HPLC is obtained with silica hydride-based materials.15 The essential
differences between ordinary silica (used for all commercial silica-based separation
materials) and silica hydride is shown in Figure 1.1. The presence of Si–H moieties
on the surface instead of silanols leads to a fundamentally different material with
chromatographic properties that result in a range of selectivities and capabilities not

Aqueous Normal-Phase Chromatography 3
O O O
O Si OH O Si O Si H
O O O
O Si OH O Si O Si H
O O O
O Si OH O Si O Si H
O O O
(A) (B)
Figure 1.1 Surface structures of ordinary silica (A) and silica hydride (B) materials used
for HPLC stationary phases.
available on other modified silicas. Among the most important features of hydride-
based phases are the absence of strong water adsorption on the surface leading to rapid
equilibration of the separation media when used in gradient applications; the ability
of all phases synthesized to date to operate in RP, aqueous normal phase (ANP), and
organic normal phase modes; retention of both polar and nonpolar compounds simul-
taneously; operation in 100% aqueous mobile phases without any dewetting (for-
merly known as phase collapse) resulting in subsequent loss of retention; and a high
level of reproducibility for both the retention of hydrophobic (RP) and hydrophilic
(ANP) compounds. Some of these significant differences are described later as they
relate to the retention and selectivity of polar compounds, the focus of this review.
1.2 Modes of ANP

In order to distinguish between the fundamental properties and retention behavior
of silica hydride-based materials in comparison to HILIC phases, the term aqueous
normal phase (ANP) retention is used. The term ANP has been used occasionally in
the past in the same context as HILIC. Some recent examples include the separation
of polar pharmaceutical compounds,16 the analysis of drugs in human plasma,17 the
determination of acrylamide in food samples,18 and a general review on the analysis
of a wide range of basic compounds.19 In reality, these terms should not be used
interchangeably. There are several types of ANP behavior observed with the silica
hydride stationary phases which clearly distinguish this mode from HILIC. These
will be briefly described with respect to the retention of hydrophilic compounds.
1.2.1 ANP 1
Figure 1.2 illustrates the retention map for two solutes (one retained by RP and the
other by a normal phase mechanism) run on a stationary phase that possess ANP prop-
erties.20 In this case, the two mechanisms are clearly evident but there is no region on
the retention map where the two solutes both have significant retention. Under these
circumstances, this phase can be used interchangeably for either RP or normal phase
retention, depending on whether a high or low aqueous mobile phase composition is
selected. With a high water percentage, the hydrophobic compound(s) will be retained
while the hydrophilic analyte(s) will elute at the void volume. For high organic con-
tent mobile phases, the opposite occurs with the polar compound(s) being retained

16
14 ANP 1
12
Retention time (min)
10
ANP compound
8 Reversed-phase compound
0
0 20 40 60 80 100
% Organic in mobile phase
Figure 1.2 Retention map (tR vs. % organic in mobile phase) for both a hydrophobic compound
(▪) and a hydrophilic compound (♦) on a silica hydride-based stationary phase with no overlap of
RP and ANP mechanisms. (Adapted from Pesek, J.J. and Matyska, M.T., LCGC, 25, 480, 2007.)
and the nonpolar specie(s) eluting at or near the dead volume. In contrast, a typical
HILIC material has only the behavior shown by the ANP compound (♦) in Figure 1.2.
Most HILIC phases show little or no RP behavior (▪) under any high aqueous mobile
phase conditions. This example provides the first clear distinction between silica
hydride-based stationary phases and HILIC materials. With water being the common
component in the mobile phase, it can be seen how a transition can be made from RP
to normal phase behavior by simply varying the aqueous content of the eluent.
1.2.2 ANP 2
Figure 1.3A shows a second type of retention map that can be obtained for two com-
pounds (one retained by RP and the other by a normal phase mechanism) on a silica
hydride phase having ANP properties. In this situation, there is a distinct range of
mobile phase compositions where both compounds display measurable retention. Under
these conditions, it is possible to separate mixtures of polar and nonpolar compounds,
often under isocratic conditions. The other possibility suggested by this retention map
is the analysis of hydrophobic and hydrophilic compounds using gradient elution with
the option of running in either direction, i.e., from low to high organic content (standard
gradient) or from high to low percent organic (inverse gradient). In contrast to a HILIC
phase (only polar compounds retained at high organic compositions) or an ANP 1 situ-
ation (retention of the type of compound determined by % organic in mobile phase),
the ANP 2 scenario offers unique separation capabilities that are available from only a
limited number of commercial products, mainly those based on silica hydride.
Figure 1.3B presents an example of the separation of a mixture that illustrates
the ANP 2 elution pattern. In this chromatogram, one polar (1) and one nonpolar (2)
compound are tested on a silica hydride-based C18 column. In the top chromatogram
at 40:60 ACN/water, the RP mechanism is dominant and the hydrophobic solute is
retained more strongly than the hydrophilic compound. The middle chromatogram

45
40 ANP 2
35
ANP compound
30 Reversed-phase compound
25
20
15
10
0
0 20 40 60 80 100
(a) % Organic in mobile phase
1 12 1
2
8 10
2 8
6
mAU
mAU
6
4 4
2 2
B
A 0
0
0 2 0 2
Time (min) Time (min)
2
10
1
8
mAU
2
C
0
0 2
(b) Time (min)
Figure 1.3 (a) Retention map (t R vs. % organic in mobile phase) for both a hydrophobic
compound (◾) and a hydrophilic compound (♦) on a silica hydride-based stationary phase
with overlap of RP and ANP mechanisms. (b) ChromatoÂ�grams for a polar (1) and a nonpolar
(2) compound on a silica hydride-based stationary phase at (A) 40:60 ACN/water; (B) 60:40
ACN/water and (C) 70:30 ACN/water. (Adapted from Pesek, J.J. and Matyska, M.T., LCGC,
25, 480, 2007.)

is obtained when the mobile phase has been changed to 60:40 ACN/water. Under
these conditions, the normal-phase mechanism has become slightly stronger than
the RP mode and compound 1 elutes just after compound 2. If the ACN content is
increased further (70:30), then the normal-phase mechanism becomes very domi-
nant and 1 is then retained significantly longer than 2.
1.2.3 ANP 3
A third possibility can also be obtained on a true ANP stationary phase and it also pro-
vides a clear distinction in comparison to a HILIC material. This situation is illustrated
by the retention map shown in Figure 1.4A. This elution pattern occurs for certain types
20
ANP 3
16
12
0
0 10 20 30 40 50 60 70 80 90 100
(A) % Organic in mobile phase
0.02 90% ACN

AU
0.00
0.04 80% ACN
AU
0.02
0.00
0.04 70% ACN
AU
0.02
0.00
0.04 60% ACN
AU
0.02
0.00
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
(B) Minutes
Figure 1.4 (A) Retention map (t R vs. % organic in mobile phase) for a compound with both
hydrophobic and hydrophilic components on a silica hydride-based stationary phase showing
both RP and ANP mechanisms. (B) Chromatograms for such a compound at various amounts of
ACN in the mobile phase. (Adapted from Pesek, J.J. and Matyska, M.T., LCGC, 25, 480, 2007.)

of compounds; those that have both hydrophobic and hydrophilic properties.21 The sol-
utes that result in this elution behavior are typically larger molecules with one or more
polar functionalities as well as a significantly hydrophobic portion. Some peptides and
proteins are examples of molecules that fit this description but some compounds with
molecular weights below 1000 having both polar and nonpolar regions also behave in
a similar manner. Under these circumstances, the chromatographer has a choice of
operating in the RP or ANP mode, depending on the type(s) of other molecules in the
mixture, sample compatibility with the mobile phase, or the means of detection uti-
lized. This unusual capability provides experimental flexibility that is not available on
typical stationary phases fabricated through the use of organosilane chemistry.
Figure 1.4B presents a series of chromatograms for a compound that is retained in
both the RP and the normal-phase modes, depending on the mobile phase composi-
tion selected. The chromatograms shown follow the retention profile predicted by the
map in Figure 1.4A. As expected retention goes through a minimum as a function
of ACN concentration, indicating that the mechanism is changing from RP to ANP.
This pattern is in clear contrast to a typical HILIC material where retention would be
observed only when the mobile phase contained a high proportion of ACN.
1.3 Chromatographic Properties in ANP

1.3.1 Efficiency
Since the exact mechanism of retention for polar compounds on the hydride-based
phases has yet to be elucidated, a variety of factors might affect efficiency, particularly
pH. Figure 1.5 shows some examples of efficiency data obtained on diamond hydride
(DH) columns.22 Figure 1.5A is a plot of flow rate vs. height equivalent to a theoretical
plate (HETP) for two carbohydrates on a 2.1â•›mm × 150â•›mm hydride-based column in the
ANP mode. The interesting feature here is that efficiency is only slightly lower (~15%)
while the flow rate has increased by a factor of 2.5. When making the same comparison
for these solutes on a 1â•›mm column over the same flow rates (linear flow rates more than
four times higher), the decrease in efficiency was only around 20%. Another interesting
result is shown in Figure 1.5B where the efficiency of uracil under ANP conditions is
measured as a function of flow rate on a 4.6â•›mm × 150â•›mm hydride-based column. In this
case, there is an increase in efficiency as the flow rate increases from 0.3 to 1.0â•›mL/min.
This result is in contrast to a commercial HILIC column of the same dimensions where
efficiency drops by more than 30% over this same range of flow rates. It should be noted
that linear flow rates are more than six times greater in the 2.1â•›mm column than in the
4.6â•›mm column, but direct comparison was limited due to restrictions of flow rates for
the mass spectrometer used for detection. These results indicate that the silica hydride-
based phases operating in the ANP mode can be used at high flow rates with little or no
deterioration in column efficiency in contrast to the HILIC phase tested.
1.3.2 Temperature Behavior

Figure 1.6 shows the retention behavior of 19 amino acids as a function of tem-
perature (log k vs. 1/T, van’t Hoff plot) in a mobile phase containing 25% water/75%
ACN with 0.1% formic acid. In all cases, retention increased as the temperature was

0.06
Glucose Sorbitol
0.05
0.04
HETP (mm)
0.03
0.02
0.01
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
(A) Flow rate (mL/min)
0.015
0.014 HILIC DH
0.013
0.012
HETP (mm)
0.011
0.01
0.009
0.008
0.007
0.006
0 0.2 0.4 0.6 0.8 1 1.2
(B) Flow rate (mL/min)
Figure 1.5 Plots of efficiency (HETP) vs. flow rate. (A) Glucose and sorbitol on DH
column (2.1â•›m m × 150â•›m m, particle size 4â•›μm) in a 80:20 ACN/DI water + 0.1% formic
acid mobile phase. (B) Comparison of commercial HILIC (4.6â•›m m × 150â•›m m, particle size
3.5â•›μm) and DH (4.6â•›m m × 150â•›m m, particle size 4.0â•›μm) columns for ANP retention of uracil.
(Adapted from Pesek, J.J. et al., J. Sep. Sci., 2200, 32, 2009.)
increased, indicating either a positive enthalpy for interaction of the solute with the
stationary phase and/or substantial entropy contributions.23 This temperature effect
is opposite of what is typically observed under RP conditions in HPLC, i.e., decreas-
ing retention with increasing temperature. This result provides other routes for
improving selectivity via temperature control. Increasing temperature can lengthen
retention, thus improving R values (resolution). Also, higher temperatures result in
lower viscosity and faster mass transfer which can decrease peak widths to provide
another means for improving resolution, depending on the extent of retention and
diffusion in the stationary phase. However, for a few solutes such as some carbohy-
drates, retention decreased with temperature. Thus, a single mechanism for the ANP
behavior on the hydride phases is probably not likely.

1.250
1.150
1.050
0.950
0.850
0.650
Log retention time
0.600
0.550
0.500
0.450
0.400
0.350
0.0028 0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035 0.0036 0.0037
1/T (K)
L-Alanine L-Glycine L-Phenylalanine

L-Asparagine L-Histidine L-Proline
L-Aspartic acid L-Isoleucine L-Serine
L-Arginine L-Leucine L-Threonine
L-Glutamic acid L-Lysine L-Tryptophan
L-Glutamine L-Methionine L-Tyrosine
L-Valine
Figure 1.6 Plot of log k vs. 1/T (K) for 19 amino acids on the silica hydride column.
(Adapted from Pesek, J.J. et al., J. Chromatogr. A, 1204, 48, 2008.)
1.3.3 Repeatability
Another advantage of the silica hydride phases is the small variation in retention (as
measured by % relative standard deviation [RSD]) that occurs when successive injec-
tions are run for the same analysis in the ANP mode. This consistency is observed for
both isocratic and gradient elution methods.22–24 Some examples of repeatability for
several types of small molecules are shown in Table 1.1. This is a representative sam-
pling from a larger group of amino acids, carbohydrates, and organic acids that were
reported in a study of hydrophilic metabolites on a silica hydride stationary phase.23
The % RSD values are typical for a broad range of hydrophilic compounds that have

Table 1.1
Repeatability Results for Some Hydrophilic
Compounds on the Hydride Stationary Phasea
Compound Retention Time (min) No. of Injections %RSD
Alanine 3.72 50 0.53
Arginine 11.32 50 0.44
Glutamine 6.37 50 0.62
Histidine 12.22 50 0.20
Proline 4.69 50 0.43
Glucose 3.11 54 0.53
Fumaric acid 6.64 10 0.08
Citric acid 9.75 10 0.10
a Mobile phases and gradients vary from compound to compound.
been tested to date on the DH column. Some further examples of repeatability will
be shown in some of the specific applications described later in this review.
1.3.4 Re-Equilibration
An important feature of hydride-based materials is their ability to re-equilibrate rap-
idly after changing mobile phase compositions, whether to new conditions for an
isocratic run or between runs in a gradient method.23 This feature is almost assuredly
a result of the hydride surface in contrast to silanol functional groups on ordinary
silica. Extensive investigations of gradient re-equilibration have been done in the
RP mode and the results of a typical study are shown in Table 1.2. It can be clearly
seen that with the proper experimental setup, the column has attained equilibrium
within 1â•›min of the end of the gradient for all of the solutes tested. Such short times
between runs are neither typical nor practical for most methods. All of the gradient
ANP methods cited in this review have an equilibrium time of 5â•›min. This is con-
siderably less than many HILIC methods reported which often have re-equilibration
Table 1.2
Equilibration Times for a Hydride-Based
Stationary Phase
Retention Time (min)
Solute/equilibration time 25 10 1
Benzene 7.30 7.35 7.25
Naphthalene 11.10 1.07 11.01
Phenanthrene 14.39 14.37 14.37
Anthracene 14.81 14.80 14.80
Pyrene 16.52 16.51 16.56

times of 30–40â•›min. Thus, ANP utilizing hydride-based separation materials often

can shorten the analysis time in comparison to other hydrophilic separation methods
such as HILIC or methods that require additives or derivatization in order to achieve
retention in the RP mode.
1.4 Conditions for Aqueous Normal Phase Retention

The presence and extent of ANP behavior is controlled by a number of chromatographic
conditions. These features are reviewed next in order to provide a better understanding
of the ANP phenomena and as a guide for developing and optimizing ANP methods.
1.4.1 Column
A number of silica hydride-based columns have been tested for ANP retention capa-
bilities. It has been demonstrated that each one (silica hydride without any modifica-
tion, silica hydride with a very low percentage of bonded organic moiety referred to
as DH, C8, C18, cholesterol, and undecenoic acid [UDA]) can function in the ANP
mode.20–27 Figure 1.7 shows the retention for the amino sugar glucosamine on a silica
hydride stationary phase with a bonded C18 moiety. This column is generally one
that would be used for RP applications, and as expected it has substantial retention
for a broad range of hydrophobic compounds as well. All three of the ANP retention
scenarios described earlier have been observed on the hydride-based C18 column.
The exact characteristics of retention as a function of the amount of organic constitu-
ent in the mobile phase depend on the specific solute(s) tested.
1.4.2 Organic Component of Mobile Phase

The most common solvent used in ANP, and frequently in HILIC as well, is ACN.
However, it has been demonstrated that other solvents can function in the ANP
7 Glucosamine
6
0
40 50 60 70 80 90 100
% Acetonitrile in DI water + 0.5% formic acid
Figure 1.7 Retention map for the amino sugar glucosamine on a silica hydride-based C18
stationary phase. Mobile phase: ACN/water with 0.1% formic acid added.

8 L-Alanine
L-Arginine
7 L-Aspartic acid
L-Cystine
L-Glutamic acid
6
L-Glycine
L-Histidine
5 L-Isoleucine
L-Leucine
4 L-Lysine
L-Methionine
L-Phenylalanine
3
L-Proline
L-Serine
2 L-Thrionine
L-Tyrosine
1 L-Valine
0
40 50 60 70 80 90 100
% ACN
Figure 1.8 Retention map for 17 common amino acids on a silica hydride-based phase with
minimal carbon on the surface (DH) using acetone as the organic component in the mobile phase.
Additive: 0.1% formic acid. (Adapted from Pesek, J.J. et al., J. Chromatogr. A, 1204, 48, 2008.)
mode under certain conditions.23 For example, with methanol being a more polar
solvent than ACN, it would be expected to induce less ANP retention. This has been
observed for all solutes’ test. Many compounds which can be retained very well in
the range of 70%–90% ACN show little or no retention with methanol. Only very
basic analytes, such as the drug tobramycin, have shown any appreciable retention in
methanol. In addition, the amount of methanol required to observe ANP retention is
always considerably higher than the percentage of ACN needed for a comparable
k value. Thus in most cases, methanol is of limited use in ANP.
In contrast, acetone, which has approximately the same polarity as ACN, is a very
useful mobile phase component in ANP. Figure 1.8 shows the retention map for a
number of amino acids as a function of the amount of acetone in the mobile phase.
Unfortunately, for many applications using UV detection, acetone has significant
absorption that precludes its use. However, when using MS or MS/MS, this is not
a restriction. Thus, when ACN is not a suitable mobile phase or cannot be easily
obtained, acetone may be a viable substitute for many applications.
1.4.3 Mobile Phase Additive

In most ANP applications, the use of an additive in the mobile phase enhances the
retention of analytes under conditions of high organic content. When using mass spec-
trometry for detection, these additives must be sufficiently volatile to be compatible with
the MS system.22–24 For basic compounds, weakly acidic species usually function best.
Both formic acid and acetic acid, usually at 0.1% or 0.2%, are the most often used. These
additives have the proton donating ability to create a positively charged species that pro-
vides both ANP retention and a suitable ion for MS detection. For acidic compounds, it
is generally required to have a higher pH additive in the mobile phase. Compounds that

fulfill this requirement and are compatible with MS detection include ammonium for-
mate and ammonium acetate usually at a concentration level of about 10–20â•›mM, con-
centrations compatible with MS detection. Under these conditions, the mobile phase pH
is usually higher than the pKa of most organic acid compounds, resulting in the forma-
tion of a negatively charged species to promote ANP retention and a suitable ion for MS
detection. Other polar compounds such as carbohydrates can be retained by ANP on a
silica hydride-based stationary phase even though they are not charged. For these ana-
lytes, the pH is not as crucial and in most cases any of the additives discussed earlier are
suitable. To enhance MS response, a small amount (~20â•›μM) of a second additive such
as sodium acetate can be used. The presence of this additive results in the formation of
a sodium adduct which provides the ionic species analyzed by the mass spectrometer.
A more universal additive would be desirable such that hydrophilic compounds
ranging in properties from acidic to basic as well as neutral could be analyzed with
a single mobile phase. Such a situation is desirable for the analysis of metabolites
in physiological samples. One possible choice for this additive is pyridine. Pyridine
containing formic or acetic acid will result in an effective hydrogen ion concentration
that will protonate most basic species and result in anions for most acidic compounds.
A further refining of the hydrogen activity is achieved by adding either formic acid
or acetic acid to the aqueous component of the mobile phase. Gradients can then be
designed which will result in the retention of a broad range of hydrophilic compounds.
In some cases, either strong interactions between the analyte and the stationary
phase or slow mass transfer effects can lead to tailing peaks. It has been demon-
strated that the addition of very small amounts (0.0001%–0.001%) of trifluoroacetic
acid (TFA) to the mobile phase can greatly improve peak symmetry.23 Such concen-
trations of TFA are low enough so that signal suppression is not a significant problem
when using mass spectrometry for detection.
1.4.4 Sample Solvent Additive

Just as the composition of the mobile phase has a profound effect on elution and
peak shape characteristics in ANP, the sample solvent can also influence the chro-
matographic results obtained for hydrophilic compounds.23,24 In some cases, it is not
necessary or even advantageous to add TFA to the mobile phase. However, peak
shape can be improved by adding a small amount of TFA (0.001%–0.005%) in the
sample solvent. This amount of TFA is not sufficient to cause any changes in reten-
tion and does not suppress the signal when using MS detection. Another additive that
has been identified as being beneficial for the improvement of peak shape for cer-
tain compounds is ammonia. Typically 1–2â•›μL of concentrated ammonia per mL of
sample was shown to improve peak symmetry for nucleotides.24 Ammonia is added
to the sample when the pH of the mobile phase is above 6. At this point, it is not cer-
tain what causes the improved peak shape for either the addition of TFA or ammonia.
An even more substantial improvement in peak shape can be obtained by adjust-
ing the ratio of organic to aqueous components in the sample solvent. Virtually all
ANP methods will start at 70% ACN or higher. For gradient methods, many begin
at 90%–95% ACN. Therefore, using a sample composition that is 50:50 ACN/DI
water results in the sample being in an excess of the strong solvent with respect to the

mobile phase into which it is being injected. Under these circumstances, a focusing
effect occurs that results in both narrower and more symmetrical peak shapes.23 This
effect has been demonstrated for many RP applications so it is not a unique feature
of ANP but it can be used as a means of providing the best resolution and efficiency
for the separation of hydrophilic compounds.
1.4.5 Gradients
As in other modes of HPLC separation, gradients can be used to shorten the analysis
time when a broad range of chromatographic retention factors (k) are present in a
single sample.22–24 In ANP, the gradient profile used to achieve a reduction in analysis
time for samples with a wide variation in k values is opposite in direction to an RP gra-
dient, i.e., the mobile phase starts at high organic content and then the amount of water
is increased to shorten the elution of the most highly retained species (most polar).
Water is the strong solvent in ANP and, thus, reduces the retention of hydrophilic com-
pounds. In addition, the presence of a gradient usually leads to improved peak shape
as well as enhanced efficiency. A good example of the efficiencies that can be achieved
in ANP analysis under gradient conditions is shown in Figure 1.9 for the separation
×104 –ESI EIC (506.0000–506.2000)

6 2
5
4
3
2 1
1
0
(A)
2
×105
–ESI EIC (346.0000–346.3000)
5
4
3
2 1
1
0
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8
(B) Counts vs. acquisition time (min)
Figure 1.9 LC-MS EIC for isobaric nucleotide species on the DH column: (A) m/z = 506
solutes: 1 = adenosine triphosphate and 2 = deoxycytidine-5′-monophosphate; (B) m/z = 346 sol-
utes: 1 = adenosine-3′-monophosphate and 2 = adenosine-5′-monophosphate. Column: 2.1â•›mm ×
150â•›mm. Mobile phase: (A) DI water + 15â•›mM ammonium acetate and (B) 90% ACN + 10%
water + 15â•›mM ammonium acetate. Gradient: 0.00â•›min 95% B; 0.00–1.00â•›min to 90% B; 1.00–
3.00â•›min to 80% B; 3.00–4.00â•›min hold 80% B; 4.00–5.00â•›min to 50% B; 5.00–6.00â•›min hold 50%
B; 6.00–7.00â•›min to 20% B. (Adapted from Pesek, J.J. et al., J. Chromatogr. A, 1216, 1140, 2009.)

of isobaric nucleotides. In this example, the efficiency for each of the four compounds
shown in these two gradient analyses is in excess of 200,000 plates/m. Similar sharp
symmetrical peaks have been obtained for many amino acids, carbohydrates, and
small organic acids by selecting the appropriate additives and gradient conditions.
1.5 Applications of Aqueous Normal Phase Retention

ANP chromatography has some features in common with HILIC, particularly those
that are related to the retention of polar compounds, but as explained earlier, there
are some notable differences. Because in all cases the separation material itself is
different, it has been found in many cases that the capabilities of the silica hydride-
based stationary phases in ANP offer a number of distinct advantages for practical
applications. Some examples of analyses involving a variety of polar compounds are
described next to illustrate how the fundamental aspects and conditions of ANP,
as described previously, result in these new capabilities. In these applications, many
of the solute molecules belong to the same general category of molecules as those
used in the previous section.
1.5.1 Amino Acids

One of the more challenging analyses that has importance in many metabolomic
studies is the separation of underivatized amino acids. Many methods have been
developed over the years for amino acids. The most common is the separation of
derivatized amino acids which simultaneously reduces the polarity to make these
compounds amenable to retention and separation by RP chromatography and
also providing a chromophore for UV detection. While certainly successful, these
meÂ�thods add a step to the analysis that results in significantly longer analysis times
than would be desired in many instances. The lack of a chromophore for most amino
acids necessitates analysis using detection by mass spectrometry or by light scatter-
ing methods. However, the underivatized amino acids are too polar to be retained by
RP methods on columns like C18 or C8 even at very high water content in the mobile
phase. Ion exchange is another alternative but these methods usually require mobile
phases that are not compatible with mass spectrometry. HILIC is thus a reasonable
choice for the retention and separation of amino acids.
It has already been demonstrated that ANP is a very suitable method for amino
acid analysis.23 Figure 1.8 demonstrates the ANP capability of the moderately modi-
fied silica hydride material (diamond hydride) for the retention of amino acids. As
stated, with a proper choice of gradient, a mixture of 19 amino acids can be ade-
quately separated for use with mass spectrometry detection. An example of a method
suitable for MS is shown in Figure 1.10 where all analytes with different mass as well
as M + 1 are separated. Only the isobaric compounds leucine and isoleucine are not
baseline separated with this gradient. A slight modification can separate these two
compounds as shown in the inset. In general, amino acids are best separated in an
acidic medium. Thus, formic acid or acetic acid provides a good choice of an additive
for amino acids since they are easily compatible with mass spectrometry detection.
Under acidic conditions using electrospray ionization, the highest abundance ion is

2
×102
1
1
0.95
0.9 Le
0.85
0.8
0.75 He
Pr Gl
0.7
0.65
0.6 P
0.55
0.5 Ar Hi
0.45
0.4
0.35 Tr
0.3 Va
0.25 Ty M
0.2
Ly
0.15 Th S
0.1 Gl As
Al
0.05 As Gl
1.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17

Counts vs. acquisition time (min)
Figure 1.10 EICs for 19 common amino acids on silica hydride stationary phase (DH)
using ACN as the organic solvent in the mobile phase. Inset: optimized separation of leucine
and isoleucine under slightly different gradient conditions. (Adapted from Pesek, J.J. et al.,
J. Chromatogr. A, 1204, 48, 2008.)
the (M + H)+ species. Since a weaker acid can often give better sensitivity, acetic acid
will sometimes provide a higher S/N than formic acid.
While running standard samples can demonstrate the ability of the ANP method
to retain amino acids, real physiological samples are more challenging and successful
analyses of these mixtures prove the ultimate usefulness of this approach to hydro-
philic molecule analysis. Figure 1.11 shows the analysis of several samples of human
saliva for the amino acids glutamine and lysine (m/z 147). These composite extracted
ion chromatograms (EICs) illustrate some important aspects regarding both the analy-
sis of physiological samples and the ANP method. First, analysis of metabolites in real
samples involves a range of concentrations for any particular analyte, as shown by the
varying peak heights in the overlaid EICs. Second, despite the complex matrix and
the variable concentration for these two compounds, the retention time is remarkably
reproducible for each peak. Thus, ANP is a rugged and reliable method for the analy-
sis of polar metabolites such as amino acids in physiological matrices.
1.5.2 Carbohydrates
Another class of solutes that can be difficult to analyze by RP as well as some
meÂ�thods designed for the retention of polar compounds is carbohydrates. The mul-
tiple hydroxyl groups lead to high solubility in aqueous mobile phases so that elution

×104 +ESI ×104 +ESI EIC (147.1134)

2
8 2
7 8 1 1
6 6
5 4
4 2
3 0
2 15.6 15.8 16 16.2 16.4 16.6 16.8
1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Figure 1.11 EICs for amino acids in saliva on the DH column: m/z 147, glutamine (1) and
lysine (2). (Adapted from Pesek, J.J. et al., J. Sep. Sci., 2200, 32, 2009.)
under typical RP conditions occurs at or near the void volume. Thus, ANP provides
the ideal mode for analysis of these hydrophilic compounds utilizing its strong reten-
tion in mobile phases containing a high percentage of organic solvent such as ACN.23
An example of carbohydrate retention is shown in Figure 1.12 for glucose when
using varying amounts of ACN in the mobile phase on a hydride-based material.
These results also illustrate another important advantage of the ANP method when
using mass spectrometry for detection. First, as expected, the retention of glucose
increases as the amount of ACN in the mobile phase increases. Second, the S/N ratio
improves as the amount of ACN in the mobile phase is increased from 60% to 90%.
A decrease in S/N is seen at 95% organic. This is most likely due to a low amount of
water that causes poorer ionization efficiency for the analyte. This example clearly
proves the usefulness of ANP for the retention of low molecular carbohydrates which
cannot be analyzed as obtained directly from metabolic samples using traditional
RP methods.
The detection of carbohydrates in physiological samples (urine and saliva) using
ANP has also been demonstrated.22 The consistency of the retention times over a
broad range of concentrations in these complex matrices further substantiates the rug-
gedness of the ANP approach using silica hydride-based stationary phases. Typical
RSD values for repeated injections of the same sample were 0.5% or less for all ana-
lytes tested. Each of these analyses utilized a gradient method with Â�re-equilibration
times between runs of 5â•›min. Thus, the transfer of protocols from standard samples
used for method development to more complex matrices generally requires little or
no adjustment of the original experimental parameters.
1.5.3 Organic Acids

Organic acids are some of the more difficult metabolites to analyze when the pH
of the mobile phase is above the pKa values. Therefore, some acids have been ana-
lyzed in the RP mode at low pH. However, in the ANP better retention is obtained
when the acid is ionized so the eluent is generally buffered close to neutral pH.
These conditions lead to a wide range of elution times because ANP retention is
strongly dependent on both the inherent acidity of the compound and the molecular

18
Hydrophilic Interaction Liquid Chromatography and Advanced Applications
×104 +EIC (203.00000 – 203.20000)
5
6 4
5.5 3
5 2
4.5
4 Mobile phase:
3.5
6 1. 60% B
3
2. 70% B
2.5 3. 80% B
2 4. 85% B
1.5 1 5. 90% B
6. 95% B
1
0.5
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.8
Figure 1.12 Comparison of the signal intensity for glucose at various compositions of ACN in the mobile phase containing 0.1% acetic acid. DH
column, 2.1â•›m mâ•›×â•›150 mm. Sample: glucose 0.3 μg/mL. (Adapted from Pesek, J.J. et al., J. Chromatogr. A, 1204, 48, 2008.)

×102
1 1 2 5 6
0.9
0.8
3
0.7
0.6
0.5
0.4
0.3
0.2
0.1
4
0
1 2 3 4 5 6 7 8 9 10 11 12
Counts (%) vs. acquisition time (min)
Figure 1.13 EICs of organic acids on a DH column. Detection in negative ion mode.
Maleic acid (1), m/z = 115; aconitic acid (2,3,4), m/z = 173; fumaric acid (5), m/z = 115; cit-
ric acid (6), m/z = 191. Mobile phase: A = DI water + 0.1% ammonium formate; B = 90%
ACN/10% DI water + 0.1% ammonium formate. Gradient: 0–3.0â•›m in 90% B; 3.0–6.0â•›m in
to 70% B; 6.0–7.0â•›m in 70% B; 7.0–7.1â•›m in to 30% B; 7.1–8.0â•›m in 30% B. (Adapted from
Pesek, J.J. et al., J. Chromatogr. A, 1204, 48, 2008.)
structure.23 Since the organic acids that have been identified as metabolites can
contain one, two, or three acidic functional groups, this feature alone provides for
significant differences in chemical properties that can be utilized to control their
retention.
A particularly striking example of the power of the ANP method for organic acids
is shown in Figure 1.13. The EICs for a number of small organic acids are presented
in a single chromatogram. The mobile phase is ACN/water with ammonium formate
as the additive. Thus, the pH of the mobile phase is higher than the pKa values of
the individual acids. All peak shapes are very symmetric and the efficiency is very
good. The most remarkable of the separations is for compounds 1 and 5, maleic and
fumaric acids. These are isobaric C4 acids with the only difference being the cis
(maleic) and trans (fumaric) configuration at the double bond. Fumaric acid has pKa
values of approximately 3.0 and 4.4 while maleic acid has pKa values of approxi-
mately 1.8 and 6.1. Thus, the longer retention for fumaric acid most likely is the
result of two ionized sites at the mobile phase pH while maleic acid only has one of
its acidic groups ionized under these conditions.
An example of the analysis of some small organic acids in a physiological sample
is shown in Figure 1.14. The overlaid EICs in the negative ion mode are for three
m/z values taken from several donor saliva samples. Three of the acids identified in
these samples are aconitic acid (m/z 173), succinic acid (m/z 117), and adipic acid
(m/z 145).22 The inset shows expanded EICs for adipic acid. In each sample, the

×105 –ESI
1
1.1 ×103 –ESI EIC (145.0506)
1 2.5 3
0.9 2
0.8 1.5
0.7 1
0.6 0.5
0.5 0
0.4 2 8 9 10 11 12 13
0.3 Counts vs. acquisition time (min)
0.2
0.1 3
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Figure 1.14 EICs for organic acids from saliva samples on the DH column. Peak identity:
(1) aconitic acid (m/z 173); (2) succinic acid (m/z 117); and (3) adipic acid (m/z 145). (Adapted
from Pesek, J.J. et al., J. Sep. Sci., 2200, 32, 2009.)
compounds were verified from the exact mass measurement. The retention times
for all three acids are very reproducible over the concentration ranges found in the
saliva samples. An interesting observation was made in comparing samples from
normal individuals to cancer patients. Aconitic acid (peak 1) was observed in only
2 of 10 normal samples while it was present in all of the samples obtained from
cancer patients. These preliminary results indicate that further investigations, both
qualitative and quantitative, could prove promising in identifying biomarkers for
clinical analyses.
1.5.4 Nucleotides
Nucleotides are important phosphate-containing compounds that are found in all liv-
ing cells and are involved in a broad spectrum of metabolic and biological processes.
They have a key role in the synthesis of DNA and RNA, are part of signal transduc-
tion pathways, can be coenzymes in biosynthetic processes, and function as energy
reservoirs in biological systems.28–33 The development of improved methods for the
analysis for these compounds is of continued interest, especially techniques which
can distinguish among the nucleotides based on their degree of phosphorylation.
An example of the separation of a mixture of mono- and triphosphate nucleotides
is shown in Figure 1.15. In this case, detection was done by UV absorption which
is generally not possible for each of the other categories of compounds described
previously. The stationary phase used in this separation consists of an 11-carbon
carboxylic acid bonded to the hydride surface. The retention capabilities of the UDA
column are the result of both hydrophilic (hydride surface and carboxylic acid func-
tionality) and hydrophobic (alkyl chain of bonded acid) properties.24 At 80% ACN
in the mobile phase, good retention was obtained but one pair of analytes (thymidine
triphosphate and adenosine triphosphate) was not resolved. If the organic content of
the mobile phase is raised to 85% (Figure 1.15), then retention became longer and the
two previously unresolved components were separated. When using the DH column,

mAU 5
20
17.5
15
2
12.5
10
7.5 1
5
3
2.5 4 6 7
0 10
Time (min)
Figure 1.15 Gradient separation of seven-component mixture of nucleotides on a silica

hydride-based UDA column. 0.0–0.5â•›min 95% B, 0.5–20â•›min to 30% B, 20.0–25.0â•›min hold
30% B. Mobile phase: (A) DI water + 0.1% ammonium formate; (B) 90% ACN/10% DI
water + 0.1% ammonium formate. Detection at 254â•›nm. Solutes: 1 = adenosine-3′,5′-cyclic
monophosphate; 2 = adenosine-5′-monophosphate; 3 = adenosine-5′-triphosphate; 4 =
thymidine-5′-triphosphate; 5 = uridine-5′-triphosphate; 6 = cytosine-5′-triphosphate; 7 =
guanosine-5′-triphosphate.
a complete resolution of this mixture was not successful under isocratic conditions
and required a gradient. The improved separation capabilities of the UDA column is
most likely due to the enhanced interaction of the nitrogen sites on the nucleotides
with the carboxylic acid moiety on the stationary phase.
In Figure 1.9B, a challenging pair of solutes, adenosine-3′-monophosphate and
adenosine-5′-monophosphate, with an m/z of 346 for the (M – H)− ion are separated.
The identity of each species was confirmed by injection of the individual compounds
since the molecular masses of these two analytes are identical and thus indistinguish-
able using a TOF™ analyzer. The necessity of having a suitable chromatographic
method is illustrated in this example because positive identification is facilitated by
their separation. These chromatograms demonstrate another advantage of MS detec-
tion: the flat baseline which is usually obtained in the EIC for a gradient elution. The
experimental conditions used in the LC-MS analysis shown in Figure 1.9 result in
both good peak shape and high efficiency (~200,000 plates/m). The ANP method
provides high sensitivity, as demonstrated earlier, for glucose since a better S/N ratio
is obtained for high percentages of organic in the mobile phase.23

1.5.5 Other Analytes

The above sections for specific classes of compounds are a representative cross-
section of hydrophilic compounds where a significant number of challenging analy-
ses occur, particularly in the field of metabolomics. Considerable progress has been
made with the ANP method in addressing many of the issues involved in the analysis
of these compounds. However, there are still improvements which can be made,
especially in trying to find a method that might encompass all hydrophilic com-
pounds, as discussed in Section 1.5.6. A number of other compounds and classes
have been evaluated by the ANP procedure. A few representative examples are pre-
sented here.
A relevant and practical problem in food analysis is shown in Figure 1.16 for the
analysis of melamine and cyanuric acid, two contaminants that can be difficult to
analyze by RP HPLC because of their highly polar properties. In the method devel-
oped for these two compounds, the pair is easily separated and identification is facili-
tated through MS detection. Cyanuric acid is detected in the negative ion mode and
melamine in the positive ion mode. The high degree of separation and the negligible
matrix effects make quantitation relatively easy down to the ppb level or less if using
an MS/MS system. The method utilized in Figure 1.16 is a simple linear ANP gradi-
ent over 15â•›min. The analysis time can be reduced further by increasing the steepness
of the gradient in order to decrease the retention time of melamine or decreasing the
column length, i.e., from 150 to 50â•›mm. Since the ANP column equilibrates rapidly,
2
×105
+ESI
2.6
Melamine - POS (M+H)+, 127.0727 m/z
2.4
2.2 OH
NH2
2 N N
N N
HO N OH
1.8 H2N N NH2
1.6 1. Cyanuric acid 2. Melamine
1.4 Cyanuric acid - NEG (M–H)–, 128.0102 m/z
1.2
1
1
0.8
0.6
0.4
0.2
0
1 2 3 4 5 6 7 8 9 10 11 12 13
Figure 1.16 Analysis of melamine and cyanuric acid on the DH stationary phase. Mobile
phase: (A) DI water + 0.1% ammonium acetate; (B) 90% ACN/10% DI water + 0.1% ammo-
nium acetate.

typically no more than 5â•›min, laboratories could easily analyze a relatively large
number of samples in a day. The column ruggedness and reproducibility add to the
attractiveness of the ANP approach for this analysis. Similar analyses have been
developed for herbicides such as glyphosate as well as paraquat and a number of
related compounds.
A physiologically relevant pair of analytes is choline and acetylcholine. These
polar compounds are difficult to analyze by RP so they are perfect candidates for the
ANP method. With a mobile phase consisting of ACN/water containing 0.5% formic
acid and using mass spectrometry for detection, a mixture of the two compounds
was tested on two hydride-based cholesterol columns. Using 90:10 ACN/water on a
4.6â•›mm × 75â•›mm column, the selectivity factor (α) was 1.21. On a 2.1â•›mm × 20â•›mm
column with a 92:8 ACN/water mobile phase, the α value was 1.28. In both cases,
the two analytes were baseline separated so that good quantitative measurements
could be made. Thus, quaternary amine compounds can be easily retained on the
hydride-based materials. This example also illustrates the advantage for ANP with
amine compounds in general. These analytes can be determined in acidic solution
as opposed to the very basic conditions that are required for RP methods. The acidic
experimental conditions are more favorable for long lifetimes of the column as well
as the instrument.
Another interesting potential use of the ANP method is shown in Figure 1.17
where the retention of a series of hydrophilic peptides is plotted as a function of
mobile phase composition.34 These results suggest that the use of hydride-based
phases in the ANP mode could be a complementary method to RP in such applica-
tions as proteomics. Just as in other applications where the sample contains a large
number of compounds with a very broad range of polarities, in a typical RP analysis
of a protein digest the hydrophilic peptides elute at or near the void volume while the
25
XP3 RBP 3/14

XP2 RBP 2/10
20 XP2 RBP 1/7
XP1 RBP 1/7
XP1 RBP 1/6

15 XP1 RBP 1/8
10
0
40 45 50 55 60 65 70 75 80
% ACN + 0.1% formic acid
Figure 1.17 Retention maps for various structurally related peptides on a silica hydride-
based C18 column. Mobile phase: ACN/water + 0.1% formic acid.

hydrophobic species are well retained and can be separated. In the complementary
analysis by ANP, the hydrophobic species would elute at or near the void volume
while the hydrophilic species would be well retained and could be separated. Thus
by two analyses of a protein digest, one RP and one ANP, the complete range of
peptides could be determined. A similar possibility also exists for samples such as
cell extracts where a sample containing a large number of proteins could also be
analyzed by the same complementary two-method approach.
A final example of the use of ANP involves the retention scenario described in
Figure 1.4. Here one compound can be retained by either RP or ANP, depending on
the mobile phase composition.21 A series of cytidine-related compounds were tested
for their ANP retention on a hydride-based stationary phase. Unsubstituted cytidine
is sufficiently polar so that it can be easily retained with high amounts of organic
in the mobile phase and an additive such as formic or acetic acid. As hydrophobic
substituents are added to the molecule, a more nonpolar moiety is created and these
molecules have RP retention. This effect is seen in Figure 1.18 for three substituted
cytidines that have alkyl side chains of varying length. The retention map focuses on
mobile phase compositions normally required for ANP behavior, i.e., >50% organic.
All of the substituted cytidines studied clearly display ANP behavior. The retention
of the compound shown as cytidine-R1 has the shortest alkyl chain of the three
molecules tested and does not have any RP retention as low as 60% ACN in the
mobile phase. Cytidine-R2 has a longer alkyl chain attached to the parent molecule
and shows a slight increase in retention as the amount of organic is decreased from
70% to 60%. Cytidine-R3 has the longest alkyl chain of the three compounds and it
clearly shows the greatest increase in retention as the amount of ACN is decreased
from 70% to 60%. This example illustrates a unique aspect of the hydride-based
phases. Typical RP materials would only have retention for this compound at high
amounts of water and HILIC columns would only have retention at high organic
compositions. For some analyses, the ability to utilize two modes of retention pro-
vides the greatest opportunity to find an optimal separation for a particular mixture.
2.5
Cytidine-R1
Cytidine-R2
2
Retention factor (k)
Cytidine-R3
1.5
0.5
55 60 65 70 75 80 85 90 95
% Acetonitrile
Figure 1.18 Retention map for substituted cytidines on a silica hydride-based choles-
terol column. Mobile phase: ACN/water + 0.1% formic acid. (Adapted from Pesek, J.J. et al.,
J. Sep. Sci., 30, 637, 2007.)

1.5.6 Universal Analytical Method for Hydrophilic Metabolites

Many of the classes of compounds described in the applications earlier are found
in metabolite samples. While each of these separations typically cites the optimal
conditions for separation, there is a practical need to limit the number of determi-
nations done for each sample. Ideally, there would be one method for hydrophobic
compounds based on RP conditions and another method that would encompass all
of the hydrophilic metabolites. Due to its ruggedness and versatility, ANP appears
an attractive choice to fill the role of analyzing for polar metabolites. The examples
presented previously while covering a range of mobile phase options are not that dis-
parate from each other. The conditions range from mildly acidic (formic acid) to near
neutral pH (ammonium acetate or ammonium formate). Thus, it might be possible to
conceive a gradient mobile program, either binary, tertiary, or quaternary, that could
sufficiently retain a reasonable number of metabolites and separate those with identi-
cal masses or differing by only 1â•›amu. Somewhat relaxed separation requirements
would be needed if using a mass spectrometer capable of exact mass determinations
at a level of 50â•›ppm or less, or with an MS/MS system. The total analysis time should
be minimized in developing the gradient program in order to generate as many deter-
minations as possible within a given time frame. The hydride-based stationary phases
can significantly aid in this respect since the equilibration time between runs is so
short, typically no more than 5â•›min. At present, this goal has not yet been attained but
continued investigations into understanding the exact mechanism of retention on the
hydride phases may facilitate achieving such an operating format.
1.6 Conclusions
Silica hydride-based stationary phases represent a new class of materials with fun-
damentally different properties than ordinary silica. One of the consequences of this
new surface is the ability to retain polar compounds in a manner similar to HILIC,
but with other features not possessed by stationary phases operating in the HILIC
format. For the stationary phases that combine a silica hydride surface with a bonded
organic moiety, retention is obtained for both polar and nonpolar compounds. The
hydride-based stationary phases are also characterized by rapid equilibration after
changing mobile phase compositions which can be particularly important after gra-
dient analyses. For hydride materials with a minimal amount of bonded organic moi-
ety, the retention of hydrophilic compounds is enhanced. Separation strategies can
be designed for either specific target analytes or to span a range of polarities utilizing
the variety of stationary phases based on silica hydride surfaces currently available.
Additional facets of these materials are being investigated and new phases are being
developed which should lead to further applications utilizing all the capabilities of
silica hydride, especially with respect to the ANP mode.
Acknowledgment
The authors would like to thank Microsolv Technology Corporation, Eatontown,
NJ, for assistance in providing data and columns to obtain some of the experimental
results described in this review.

References
1. Theodoridis, G., Gika, E., and Wilson, I.D., Trends Anal. Chem., 2004, 27, 251.
2. Yokoyama, T., Sokai, A., and Zenki, M., Chromatographia, 2008, 67, 535.
3. Concha-Herrera, V., Torres-Lapasio, J.R., Vivo-Truyols, G., and Garcia-Alvarez-Coque,
M.C., J. Liq. Chromatogr. Relat. Technol., 2006, 29, 2521.
4. Piraud, M., Vianey-Saban, C., Bordin, C., Acquaviva-Bordain, C., Boyer, S., Elfakir, C.,
and Bouchu, D., Rapid Commun. Mass Spectrom., 2005, 19, 3287.
5. Tomiya, N., Alior, E., Lawrence, S.M., and Betenbaugh, M.J., Anal. Biochem., 2001,
293, 129.
6. Yun, J., Shen, S., Chen, F., and Yao, K., J. Chromatogr. B, 2007, 860, 57.
7. Dorion, S. and Rivoal, J., Anal. Biochem., 2003, 323, 188.
8. Schlichterle, H., Affolter, M., and Cerny, C., Anal. Chem., 2003, 75, 2349.
9. Coulier, L., Bas, R., Jespersen, S., Verheij, E., van der Werf, M.J., and Hankemeier, T.,
Anal. Chem., 2007, 79, 6573.
10. Bajad, S.U., Lu, W., Kimball, E.H., Yuan, J., Peterson, C., and Rabinowitz, J.D.,
J. Chromatogr. A, 2006, 1125, 76.
11. Kind, T., Tolstikov, V., Fiehn, O., and Weiss, R.H., Anal. Biochem., 2007, 363, 185.
12. Tolstikov, V.V., Lommen, A., Nakanishi, K., Tanaka, N., and Fiehn, O., Rapid Commun.
Mass Spectrom., 2005, 19, 3031.
13. Pan, J., Song, Q., Shi, H., King, M., Junga, H., Zhou, S., and Naidong, W., Rapid
Commun. Mass Spectrosc., 2003, 17, 2549.
14. Cubbon, S., Bradbury, T., Wilson, J., and Thomas-Oates, J., Anal. Chem., 2007, 7, 8911.
15. Pesek, J.J. and Matyska, M.T., J. Liq. Chromatogr. Relat. Technol., 2006, 29, 1105.
16. Naidong, W., Shou, W.Z., Addison, T., Maleki, S., and Jiang, X., Rapid Commun. Mass
Spectrom., 2002, 16, 1965.
17. Shen, J.X., Xu, Y., Tama, C.I., Merka, E.A., Clement, R.P., and Hayes, R.N., Rapid
Commun. Mass Spectrom., 2007, 21, 3145.
18. Rosen, J., Nyman, A., and Hellenaes, K.-E., J. Chromatogr. A, 2007, 1172, 19.
19. McCalley, D.V., Advances in Chromatography, Vol. 46, CRC Press, Boca Raton, FL,
2008, pp. 305–350.
20. Pesek, J.J. and Matyska, M.T., LCGC, 2007, 25, 480.
21. Pesek, J.J., Matyska, M.T., and Larrabee, S., J. Sep. Sci., 2007, 30, 637.
22. Pesek, J.J., Matyska, M.T., Loo, J.A., Fischer, S.M., and Sana, T.R., J. Sep. Sci., 2009,
32, 2200.
23. Pesek, J.J., Matyska, M.T., Fischer, S.M., and Sana, T.R., J. Chromatogr. A, 2008, 1204, 48.
24. Pesek, J.J., Matyska, M.T., Hearn, M.T.W., and Boysen, R.I., J. Chromatogr. A, 2009, 1216,
1140.
25. Pesek, J.J. and Matyska, M.T., LC/GC, 2006, 24, 296.
26. Pesek, J.J., Matyska, M.T., Gangakhedkar, S., and Siddiq, R., J. Sep. Sci., 2006, 29, 872.
27. Pesek, J.J., Matyska, M.T., and Sharma, A., J. Liq. Chromatogr. Relat. Technol., 2008,
31, 134.
28. Sebestik, J., Hlavacek, J., and Stibor, I., Curr. Protein Pept. Sci., 2005, 6, 133.
29. von Ballmoos, C., Brunner, J., and Dimroth, P., Proc. Natl. Acad. Sci. USA, 2004, 101, 11239.
30. Ataullakhanov, F.I. and Vitvitsky, V.M., Biosci. Rep., 2002, 22, 501.
31. Begley, T.P., Kinsland, C., Mehl, R.A., Osterman, A., and Dorrestein, P., Vitam. Horm.,
2001, 61, 103.
32. Francis, S.H. and Corbin, J.D., Crit. Rev. Clin. Lab. Sci., 1999, 36, 275.
33. Ashcroft, S.J., Adv. Exp. Med. Biol., 1997, 426, 73.
34. Pesek, J.J., Matyska, M.T., Hearn, M.T.W., and Boysen, R.I., J. Sep. Sci., 2007, 30, 1150.

2 Application of
the Calculated
Physicochemical
Property, Log D, to
Hydrophilic Interaction
Chromatography in the
Bioanalytical Laboratory
Eugene P. Kadar, Chad E. Wujcik,
and Olga Kavetskaia
Contents
2.1 Introduction..................................................................................................... 27
2.2 Theoretical Introduction and Definitions........................................................ 29
2.3 Experimental Design....................................................................................... 32
2.4 Results and Discussion....................................................................................34
2.4.1 Relationship between Experimentally Determined k′
and Calculated Log DpH 3.0...................................................................34
2.4.2 Discussion of Assumptions Made to Formulate Hypothesis............... 35
2.4.3 Identification of Internal Standards..................................................... 43
2.5 Conclusions...................................................................................................... 43
Acknowledgment......................................................................................................44
References.................................................................................................................44
2.1 Introduction
Liquid chromatography combined with tandem spectrometric detection (LC-MS/
MS) is the preferred technique for quantitative bioanalysis due to its intrinsic selec-
tivity, sensitivity, and speed.1–4 These qualities provide optimum utility in the analy-
sis of drugs and their metabolites in complex biological matrices such as cerebral
27

spinal fluid, plasma, serum, tissue, urine, and whole blood in support of animal and
human studies during drug development.
Hydrophilic interaction chromatography (HILIC) has become an increasingly
popular choice as the chromatographic component in quantitative bioanalysis due to
its unique selectivity, reduced back pressure, and enhanced mass spectrometric sen-
sitivity relative to traditional reversed-phase chromatography.1,2,5 In 2003, Naidong
published a comprehensive literature review on the utilization of HILIC in bioan-
alytical LC-MS/MS.2 This review focused on the use of unmodified silica as the
stationary phase and included topics such as retention mechanism, sensitivity com-
parison with a reversed-phase, direct injection of SPE extracts, ultrafast LC-MS/MS,
column stability, and future perspectives. A tabulation of HILIC-MS/MS methods
for representative pharmaceutical compounds in various biological matrices is pro-
vided. Hsieh recently published a review article on the potential of HILIC-MS/MS in
the quantitative bioanalysis of drugs and metabolites.1 This chapter provides a tabu-
lation of HILIC-MS/MS applications for pharmaceutical bioanalysis and presents
examples of the utilization of sulfobetaine (ZIC) and amino-modified silica station-
ary phases for HILIC-MS/MS. Hsieh also discusses the topic of matrix effects, in the
form of ionization suppression or enhancement, commonly observed in bioanalysis.
Ionization suppression or enhancement is a matrix effect in which endogenous
components co-elute with the analyte of interest, causing a decrease or increase in
the level of analyte ionization in the MS source and subsequent mass spectrometric
signal. This phenomenon is common during the analysis of complex matrices such
as biological fluids by LC-MS/MS and occurs more frequently with electropsray
ionization (ESI) techniques.1,6–10 Typically, extensive and elaborate extraction proce-
dures, such as solid phase or liquid/liquid, are employed to alleviate matrix effects
that may be encountered during LC-MS/MS using reversed-phase chromatography.
For more polar compounds, the alternative selectivity of HILIC may be utilized to
remediate matrix effects, allowing the analyst to use less involved and more cost-
effective extraction methods like protein precipitation. The high level of orthogonal-
ity and the unique selectivity of HILIC relative to reversed-phase chromatography
may also provide the means of separating isobaric compounds and other interfer-
ences that cannot be achieved by reversed-phase chromatography. This alternative
selectivity is very advantageous during the early stages of drug candidate develop-
ment when in vivo metabolic transformation knowledge is limited. Another factor,
the reduced back pressure afforded by HILIC, allows the use of higher flow rates that
significantly decreases analysis time.1,2 The typical loss in column efficiency from
increased flow rates is not as prevalent for HILIC as compared to reversed-phase
chromatography.2 This benefit is believed to be a result of the differences in diffusion
and mass transfer as a result of the lower viscosity of HILIC relative to reversed-
phase chromatography.2
HILIC typically uses 5%–50% water in the mobile phase with the remainder
composed of a volatile organic solvent such as acetonitrile.5 The relatively high
organic solvent concentration is favored in the desolvation process of ESI, which
is extremely beneficial considering that this ionization technique is commonly
employed in the LC-MS/MS bioanalysis of polar compounds.1,2,5 Additionally,
HILIC allows the direct injection of many biological extracts obtained by protein

Application of the Calculated Physicochemical Property 29
precipitation, liquid/liquid, and solid phase extraction, without the need for further
compositional modification (e.g., evaporation with reconstitution).1–4
Bioanalysis plays an important role in drug development, in which the speed of
method development is a critical factor. The ability to predict and select appropriate
separation conditions can significantly expedite method development. A consider-
ation of the physicochemical properties of an analyte can provide great insight into
this process.
The distribution coefficient, D, is used to quantitatively express lipophilicity
under a specific pH condition. It is a physicochemical property that is defined as
the equilibrium concentration ratio of a substance between an aqueous buffer at a
specified pH and octanol.11–15 The commercial software program ACD/Logâ•›D Suite
is capable of calculating this physicochemical property from the input of the chemi-
cal structure of a compound. Recent literature has demonstrated the dependence of
the HILIC capacity factor, k′, on D.16–18 This chapter is based on a recent publication
in which a linear relationship between log D and the HILIC log k′ was demonstrated
for a series of representative pharmaceutical compounds under generic HILIC condi-
tions.16 This relationship was employed to rapidly assess the applicability of HILIC
as well as to identify suitable internal standards for LC-MS/MS bioanalysis.
2.2 Theoretical Introduction and Definitions

In 1990, Alpert coined the term “hydrophilic interaction chromatography” (HILIC)
and suggested that the mechanism of HILIC involves analyte partitioning between
a hydrophobic mobile phase and a partially immobilized layer of water on the sta-
tionary phase.19 Several recent review articles have been published on the topic of
HILIC, all of which conclude that the mechanism is in fact multidimensional includ-
ing additional components such as adsorption and electrostatic interactions.1,2,20–23
The following theoretical discussion is based on the hypothesis that partitioning
is the primary mechanism defining retention under a generic HILIC system that
includes an unmodified silica stationary phase and a relatively high mobile-phase
salt concentration of 10â•›m M ammonium formate, pH adjusted to 3.0.
The term capacity factor (k′) is used to indicate the magnitude of retention of an
analyte on a chromatographic column. This term is defined as
t R − t0
k′ = (2.1)
t0
where
t R is the retention time of the analyte
t0 is the column dead time or retention time of nonretained components24–28
The Center for Drug Evaluation and Research (CDER) recommends a minimum k′
value of >2 for separations that will be used to support regulatory submissions.25
This criterion typically provides adequate retention, allowing a separation between
the analyte of interest and the unretained hydrophobic matrix components. This sep-
aration is critical to prevent possible ionization suppression or enhancement when

using atmospheric pressure ionization mass spectrometers as detectors, especially

with the use of protein precipitation as the extraction technique.
The following partitioning relationship describes the dependence of the fraction
of organic modifier on k′ for a reversed-phase system:20,24,26,27
log k ′ = log k w′ − S ϕ (2.2)

where
kw′ â•›is the theoretical value of k′ for exclusively water as the mobile phase
φ is the volume fraction of the organic in the binary mobile phase
S is a constant representing the slope of the log k′ versus the φ line20,24,26,27
The value of S is approximately 4 for low-molecular weight compounds (<500â•›Da).24

This molecular weight range is representative of most of the drugs and drug metabo-
lites analyzed by LC-MS/MS.
The partition mechanism observed in revered-phase chromatography is compa-
rable to the partitioning of a compound between octanol and water.24,26–30 Solute
partitioning between octanol and water is a physicochemical property and is defined
as the partition coefficient, P30
γ O CO
P= (2.3)
γ W CW
where
γO and γW are the activity coefficients of the compound in octanol and water,
respectively
CO and CW are the concentrations of the compound in octanol and water,
respectively
In the case of dilute solutions, such as the chromatographic mobile phase, this equa-
tion can be reduced to30
CO
P= (2.4)
CW
The partition coefficient is typically expressed as the logarithm and is used to rep-
resent the lipophilicity of a compound. This physicochemical property is routinely
applied in the pharmaceutical industry as an indicator of the absorption and per-
meation characteristics of a compound.31 This property is usually determined using
the shake-flask method.26,27,29,30,32,33 However, this method is not sensitive to impuri-
ties and degradation products and has a low throughput and limited range.26,27,30,32,33
In order to improve these deficiencies, several laboratories have employed reversed-
phase chromatography and the above-described relationship between k′ and P to
determine log POct values of several test compounds.11,26,27,29,30,32,33 Predictive com-
mercial software programs, such as ACD/Logâ•›PDB, can also be used to calculate log
POct values.34 However, the use of log P is limited, especially when applied to chro-
matography, because this physicochemical property considers only compounds in the

unionized state.12,13 A majority of drugs and drug metabolites are typically present in
their ionized forms during the chromatographic phase of LC-MS/MS analysis.
The distribution coefficient, D, is a physicochemical property that takes into
consideration the ionization state of a compound and is also typically expressed as
the logarithm. This property is the equilibrium concentration ratio of a compound
between the octanol and aqueous buffer at a specified pH and is expressed as12,13,15
 1 
log D bases = log P + log  pKa − pH ) 
(2.5)
 (1 + 10 
The above equation pertains to monoprotic bases. This equation can be applied to mono-
protic acids by replacing the exponent pKa–pH with pH–pKa.12,13 This equation can also
be expanded to include polyprotic acids, bases, and zwitterions. Lombardo et al. suc-
cessfully utilized the relationship between log k′ and log D at the physiological pH of 7.4
7.4
to determine log DOct values of a set of basic and neutral pharmaceutical compounds.13
Quiming et al. recently developed HILIC retention prediction models for adrenore-
ceptor agonist and antagonist compounds utilizing multiple linear regression (MLR)
and artificial neural networks (ANN).17,18 Although HILIC retention prediction models
were determined using both unmodified silica and diol columns, the work using the
unmodified silica column will be discussed. MLR is commonly used for the statisti-
cal analysis of quantitative structure–retention relationships. ANN is another method
used in chromatographic optimization that has recently gained popularity. MLR was
used to derive the following predictors: the percentage acetonitrile in the mobile
phase, log D, the number of hydrogen bonding acceptors, the magnitude of the dipole
moment, and the total absolute atomic charge. Standardization coefficients (β) of the
predictors indicate the relative effect of each predictor on the logarithm of the reten-
tion factor. The percentage of acetonitrile in the mobile phase produced the largest
magnitude β (positive in sign), which can be expected considering the known depen-
dence of HILIC on organic mobile-phase concentration. Log D produced the second
largest magnitude β (negative sign), representing an inversely proportional relationship
to retention. The relatively high magnitude of β is indicative of the contribution of
partitioning to the overall mechanism of HILIC utilizing an unmodified silica column.
ACD/LogD Suite Version 9.0 is capable of calculating log D values for poly-
protic acids, bases, and zwitterions over the pH range of 0–14 in increments of 0.1â•›pH
units.14,15 Additionally, log D can be calculated with or without the incorporation of
ion-pairing. The chemical structure of the compound of interest is entered and the
software takes the proposed equilibrium scheme for all of the species in the organic
and aqueous phases into account for the log D calculation.15 Nozawa et al. success-
fully applied ACD Labs Log D as a measure of hydrophobicity in the advanced
Marfey’s method.35 The advanced Marfey’s method is a derivatization-based tech-
nique used to determine the absolute configuration of l- and d-amino acids in a
peptide using LC-MS.35 This technique is based on the hydrophobicity difference
between the derivatized l- and d-amino acids.
The capability to calculate the pH-dependent distribution coefficient using ACD/
LogD Suite allowed the following hypothesis to be explored: A relationship exists

between the HILIC capacity factor, k′, of an analyte and the calculated log DpH 3.0
in a generic HILIC system. This system was designed to minimize adsorption and
electrostatic interactions, thereby allowing the partition to act as the primary reten-
tion mechanism. The following assumptions were made for the evaluation of this
hypothesis: (1) partitioning between the hydrophobic mobile phase and a hydro-
philic partially immobilized layer of water on the stationary phase is the primary
chromatographic retention mechanism of the above-defined generic HILIC system,
(2) the pH of the immobilized layer of water on the silica stationary phase is 3.0,
(3) ion-pairing is insignificant, and (4) the ACD/LogD Suite can provide a reasonable
estimate of log DpH 3.0.
2.3 Experimental Design

A set of 30 probe compounds synthesized by Pfizer Inc., were selected to cover an
extensive calculated log DpH 3.0 range of −8.68 to 8.16. These probe compounds are
representative of the following therapeutic areas: anti-infectives, cancer, cardiovas-
cular and metabolic diseases, and the central nervous system. Five chemical analog
pairs were included in this set based on structure and log D similarity, in order to
determine the utility of using calculated log D as a criterion for analog internal stan-
dard selection. A tabulation of the probe compounds, including their physicochemi-
cal properties is presented in Table 2.1. The chemical structure of probe compound
No. 1 is illustrated in Figure 2.1.
This compound was utilized as a representative basic drug in order to develop the
generic HILIC system, particularly with respect to ammonium formate concentration.
ACD/LogD Suite, Version 9.0 was used to calculate the log DpH 3.0 of each probe
compound. Stock solutions were prepared at approximately 100â•›μg/mL in 50:50 (v/v)
water:acetonitrile. Working solutions were prepared at approximately 100â•›ng/mL in
acetonitrile by dilution of the corresponding stock solution, with the exception of
compound No. 30. An increased concentration of this compound, approximately
10,000â•›ng/mL, was required to obtain an adequate peak response.
The chromatographic system included a Shimadzu SCL-10A controller,
Shimadzu LC-10AD pumps, and a CTC Analytics (LEAP) HTLS PAL autosampler.
An unmodified silica column, Atlantis™ HILIC silica column, 2.1 × 50â•›mm, 5â•›μm,
100â•›Å, P/N 186002012 from Waters Corporation (Milford, MA), was employed at
ambient temperature. HPLC grade water and acetonitrile were combined with a
200â•›m M aqueous ammonium formate, pH 3.0 stock solution to produce three sepa-
rate mobile-phase solutions with acetonitrile concentrations of 85%, 90%, and 95%
(v/v) and a total ammonium formate concentration of 10â•›m M. These acetonitrile con-
centrations were chosen because they can typically provide an adequate k′ range of
2–5 for HILIC applicable compounds. k′ values of greater than 5 are typically not
utilized in LC-MS/MS bioanalysis.
The detector was an Applied Biosystems API 4000 tandem quadrupole mass
spectrometer equipped with a TurboIonspray™ source. The source was operated in
the positive-ion electrospray mode except for probe compound Nos. 24 and 27, for
which negative-ion electrospray was employed due to acidic functional groups. The
mass spectrometer was tuned for source conditions and precursor ion/product ion

Table 2.1
Physicochemical Properties of Probe Compounds
Compound Molecular
Identification No. Weight (Da) Acidic pKaa Basic pKaa Log DpH 3.0b Log P b
1 366 — 9.37, 5.65, 4.25 −8.68 −0.02
2 448 — 8.71, 7.48, 2.98 −7.44 3.09
3c 363 11.8 10.1, 4.11 −6.75 1.69
4c 377 — 10.1, 3.94 −6.36 1.64
5d 439 — 7.67, 5.46 −6.32 0.88
6d 453 — 7.67, 5.46 −5.79 1.41
7 227 — 9.95 −4.72 2.23
8e 322 — 9.16 −4.38 1.77
9 258 — 8.86, 2.75 −4.20 1.83
10e 318 — 9.35 −4.06 2.28
11 355 — 8.17, 4.83 −3.37 3.20
12 510 — 6.99, 5.60 −3.35 2.74
13f 378 — 9.43 −2.76 3.39
14f 366 — 9.41 −2.44 3.69
15 508 — 4.36, 1.31 −2.10 −0.71
16 470 12.1 8.18, 4.44 −1.08 5.12
17 300 — 9.04 −1.06 4.99
18 335 — 8.49 0.33 5.81
19 448 — 7.67, 2.61 0.67 5.14
20g 489 — 4.10 0.85 2.02
21g 474 — 4.10 0.94 2.11
22 325 9.25 — 1.51 1.51
23 382 — 5.23, 4.85 1.71 6.07
24 454 3.10 1.44 2.73 3.00
25 559 4.29 — 4.10 4.13
26 469 3.14 2.33 4.46 4.78
27 447 3.33, 11.5 — 5.88 6.08
28 639 10.5, 12.4 1.78 7.14 7.16
29 595 — — 7.27 7.28
30 600 — — 8.16 8.16
Source: Reproduced from Kadar, E.P. et al., J. Chromatogr. B, 863, 1, 2008.

a Calculated using ACD Labs pK Suite, version 9.0.
a
b Calculated using ACD Labs Log D Suite, version 9.0.
c–g Chemical analog pairs.
pair selection by the infusion of individual solutions of each probe compound. Data
acquisition and chromatographic review were performed using Applied Biosystems/
MDS SCIEX Analyst, Version 1.4.
The chromatographic system was allowed to equilibrate with each mobile-phase
condition (10â•›m M ammonium formate, pH 3.0 with acetonitrile concentrations of
85%, 90%, or 95% [v/v]). Exactly 3â•›μL of each probe compound working solution

was injected in triplicate with a run time of 7â•›min N F F

under isocratic conditions for each of the mobile-
phase compositions investigated. The retention N N
time of compound No. 30 was utilized as the col- N N
umn dead time (or the retention time of nonretained
N O
components), t0. This compound was chosen for
this purpose based on its high log P, 8.16, and its Figure 2.1 Chemical structure
neutral character. Additionally, experiments using of compound No. 1. (Reproduced
the same generic HILIC system with an expanded from Kadar, E.P. and Wujcik,
acetonitrile concentration range of 75%–95% (v/v) C.E., J. Chromatogr. B, 877, 471,
produced comparable retention times for com- 2009.)
pound No. 30 (0.14–0.15â•›min), suggesting that it is
not retained and therefore an appropriate t0 marker.36 The mean k′ (n = 3) for each
probe compound was calculated. Linear regression analysis of mean log k′ versus
calculated log DpH 3.0 was performed for each of the three mobile-phase acetonitrile
concentrations.
2.4 Results and Discussion

2.4.1 Relationship between Experimentally Determined k′
and Calculated Log DpH 3.0
The relationship between log k′ and log DpH 3.0 was satisfactorily described by a
linear function, with linear correlation coefficients of 0.751, 0.696, and 0.689, cor-
responding to mobile-phase acetonitrile concentrations of 85%, 90%, and 95% (v/v),
respectively. Plots are presented in Figure 2.2.
Equations 2.6 through 2.8 express the linear relationship between experimentally
determined log k′ and calculated log DpH 3.0 for mobile-phase acetonitrile concentra-
tions of 85%, 90%, and 95% (v/v), respectively:
log k ′ = −0.132 (log D) + 0.034 (2.6)

log k ′ = −0.132 (log D) − 0.234 (2.7)

log k ′ = −0.139 (log D) − 0.008 (2.8)

A minimum k′ value of 2 is recommended by CDER and is critical to ensure separa-

tion between the analyte of interest and unretained matrix components that could
possibly produce ionization or enhancement of the analyte signal.25 Equations 2.6
through 2.8 were utilized to produce log DpH 3.0 cut-off (k′ ≥ 2) values of −4.0, −2.0,
and −2.2 for mobile-phase acetonitrile concentrations of 85%, 90%, and 95% (v/v),
respectively. The mean experimentally determined k′ values are presented in Table
2.2. Additionally, the mean predicted k′ values using Equations 2.6 through 2.8 are
included in Table 2.2. Overall, the applicability of HILIC (k′ ≥ 2) was correctly pre-
dicted for approximately 90% of all compounds tested for each of the three generic
HILIC systems evaluated.

2.0 2.0
y = –0.132 x – 0.234 y = –0.132x + 0.034
1.5 R2 = 0.751 1.5 R2 = 0.696
1.0 1.0
0.5 0.5
Log k΄
Log k΄
0.0 0.0
–0.5 –0.5
–1.0 –1.0
–1.5 –1.5
–2.0 –2.0
–10.0 –5.0 0.0 5.0 10.0 –10.0 –5.0 0.0 5.0 10.0
(A) Log D (B) Log D
2.0 y = –0.139x – 0.008

1.5 R2 = 0.689
1.0
0.5
Log k΄
0.0
–0.5
–1.0
–1.5
–2.0
–10.0 –5.0 0.0 5.0 10.0
(C) Log D
Figure 2.2 Experimentally determined log k′ versus calculated log DpH 3.0 obtained using
generic HILIC systems (Atlantis™ HILIC silica column [5â•›μm], mobile phase: water and ace-
tonitrile were combined with a 200â•›m M aqueous ammonium formate, pH 3.0, stock solution
to produce three separate mobile-phase solutions with acetonitrile concentrations of 85%,
90%, and 95% (v/v) and a total ammonium formate concentration of 10â•›m M, with a flow rate
of 1.0â•›m L/min). Linear regression analysis provided correlation coefficients (R 2) of 0.751 (A),
0.696 (B), and 0.689 (C). (Reproduced from Kadar, E.P. et al., J. Chromatogr. B, 863, 1, 2008.)
2.4.2 Discussion of Assumptions Made to Formulate Hypothesis

The correlation coefficients produced from linear equations (2.1 through 2.3) clearly
deviate from unity. These results can be expected considering the assumptions that
were made regarding the formulation of our hypothesis. With a clear understanding
that the overall mechanism of HILIC is multimodal and is comprised of partitioning,
adsorption, and electrostatic mechanisms, our generic HILIC systems were designed
to minimize the latter two mechanisms and simplify the model.
The first assumption designates partitioning between the hydrophobic mobile
phase and a hydrophilic, partially immobilized layer of water on the stationary
phase as the primary chromatographic retention mechanism of our generic HILIC
systems. Consideration was given to the following three principal types of inter-
molecular forces involved in a chromatographic separation: dispersion forces, polar

36
Table 2.2
Tabulation of Experimental k′, Predicted k′, and Prediction of HILIC Applicability Utilizing Log D
Mobile Phase
Condition 85% Acetonitrile 90% Acetonitrile 95% Acetonitrile
Compound Mean (n = 3) Correct Appl. Mean (n = 3) Correct Appl. Mean (n = 3) Correct
ID No. Exptl. k ′ Pred. k ′ Pred.a Exptl. k ′ Pred. k ′ Pred.a Exptl. k′ Pred. k ′ Appl. Pred.a
1 3.56 8.20 Y 7.44 15.13 Y 10.73 15.92 Y
2 12.71 5.62 Y 30.64 10.38 Y 18.09 10.70 Y
3 7.18 4.56 Y 12.51 8.41 Y 18.94 8.57 Y
4 7.62 4.05 Y 11.71 7.47 Y 11.45 7.56 Y
5 4.07 4.00 Y 9.16 7.38 Y 14.67 7.47 Y
6 3.51 3.40 Y 7.89 6.28 Y 12.03 6.30 Y
7 1.36 2.46 N 2.67 4.54 Y 3.18 4.47 Y
8 1.80 2.22 N 3.40 4.09 Y 3.06 4.01 Y
9 7.98 2.10 Y 18.40 3.88 Y 18.52 3.78 Y
10 2.00 2.01 Y 3.80 3.71 Y 2.73 3.61 Y
11 1.04 1.63 Y 2.00 3.01 Y 2.52 2.90 Y
12 0.91 1.62 Y 1.11 2.99 N 0.73 2.88 N
13 1.07 1.35 Y 2.13 2.50 Y 2.06 2.38 Y
14 1.00 1.23 Y 2.04 2.27 Y 1.97 2.15 Y
15 0.20 1.11 Y 0.27 2.05 N 0.55 1.93 Y

Application of the Calculated Physicochemical Property
16 3.40 0.81 N 7.31 1.50 N 2.91 1.39 N
17 0.80 0.81 Y 1.40 1.49 Y 0.82 1.38 Y
18 0.80 0.53 Y 1.33 0.98 Y 0.64 0.88 Y
19 0.93 0.48 Y 1.67 0.88 Y 0.73 0.79 Y
20 0.33 0.45 Y 0.47 0.83 Y 0.27 0.75 Y
21 0.36 0.44 Y 0.47 0.81 Y 0.30 0.73 Y
22 0.07 0.37 Y 0.09 0.68 Y 0.18 0.60 Y
23 0.40 0.35 Y 0.53 0.64 Y 0.18 0.57 Y
24 0.40 0.25 Y 0.73 0.47 Y 1.27 0.41 Y
25 0.18 0.17 Y 0.40 0.31 Y 0.64 0.26 Y
26 0.38 0.15 Y 0.67 0.28 Y 1.00 0.23 Y
27 0.07 0.10 Y 0.27 0.18 Y 0.27 0.15 Y
28 0.00 0.07 Y 0.00 0.12 Y 0.00 0.10 Y
29 0.07 0.06 Y 0.13 0.12 Y 0.00 0.10 Y
30 0.00 0.05 Y 0.00 0.09 Y 0.00 0.07 Y
% Correct 90 90 93
applicability
predictions

a Correctly predicted if a k′ ≥ 2 would be achieved by HILIC.
37
forces, and ionic forces.37 The unmodified silica stationary phase utilized in this
work can exhibit all three of these intermolecular forces to a degree. Dispersion
forces, also referred to as “London dispersion forces,” are due to charge fluctuations
from electron/nuclei vibrations. This type of intermolecular force is responsible for
hydrophobic interactions. Unmodified silica columns contain siloxane bonds that are
hydrophobic and may give rise to dispersive interactions with relatively hydrophobic
analytes.38 However, any dispersive interactions should be minimized by the high
acetonitrile content of the generic HILIC system mobile phase.
Polar forces are due to dipole–dipole or dipole-induced dipole interactions.37 Polar
interaction between an analyte and the unmodified stationary phase should be very
minimal, considering the shielding effect of the partly immobilized water layer on
the stationary phase. Conversely, polar interactions, such as hydrogen bonding, occur
between the analyte and the water layer. This type of interaction is incorporated into
the log DpH 3.0 value of a compound and is therefore integrated into our hypothesis.
Ionic forces are interactions between molecules having opposite net charge.37
Possible interaction between positively charged analytes and negatively charged
silanols can occur in our generic HILIC system.20,38 This undesired interaction gives
rise to enhanced retention relative to partitioning alone. Additionally, this interac-
tion is responsible for peak tailing. The majority of the probe compounds used for
this work were protonated basic amines at the mobile-phase pH of 3.0. Refer to
Table 2.1 for a tabulation of the physicochemical properties of the probe compounds,
including pKa. Previous work performed utilizing compound No. 1 demonstrated the
influence of mobile-phase ammonium formate concentration on retention as well as
peak shape.36 Refer to Figure 2.3 for a chromatographic illustration of the influence
of ammonium formate concentration retention time and peak shape.
Figure 2.4 presents a graphical representation of the dependence of the tailing
factor calculated as
A+ B
T= (2.9)
2A
where
A is the width from the leading edge of the analyte peak to the peak maximum at
5% of peak height
B is the width from the peak maximum to the tailing edge of the analyte peak at
5% of peak height24
The tailing factor decreased from a mean value of 1.95 to 0.939 as the total ammo-
nium formate concentration of the mobile phase was increased from 1 to 20â•›m M. It
is apparent that a total ammonium formate mobile-phase concentration of at least
10â•›m M is required to achieve adequate peak symmetry. It was concluded that the
ammonium formate was responsible for the alleviation of electrostatic interactions
between the protonated drug, compound No.1, and the ionized silanol groups of the
unmodified silica stationary phase.36 Another experiment employing compound
No. 1, in which the percentage of acetonitrile was varied from 75% to 95% (v/v) while
maintaining a total ammonium formate concentration of 10â•›m M, was performed in

1
1.7e6
1.6e6
1.5e6
3
1.4e6
2
1.3e6 1. 20 mM ammonium formate
3. 10 mM ammonium formate
5. 1 mM ammonium formate
Intensity, cps
1.0e6
9.0e5
8.0e5
7.0e5
6.0e5
4
5.0e5
4.0e5
3.0e5
2.0e5 5
1.0e5
0.0
1 2 3 4 5 6 7 8 9
Time (min)
Figure 2.3 Influence of buffer concentration on retention time and peak shape.
Chromatography was performed using an Atlantis™ HILIC silica column (5â•›μm) at a flow
rate of 1.0â•›m L/min. Pump A delivered the required make-up volume of water, Pump B pro-
vided acetonitrile at a constant percent of 90% (v/v), and Pump C delivered the appropriate
molarity ammonium formate, pH 3, stock buffer (i.e., 10, 50, or 200â•›m M) to produce a net
mobile-phase concentration of 1, 5, 10, 15, and 20â•›mM. An ammonium formate concentration of
10â•›mM or greater produced symmetrical peaks. (Reproduced from Kadar, E.P. and Wujcik, C.E.,
J. Chromatogr. B, 877, 471, 2009.)
order to determine if electrostatic ionic interactions can be adequately remediated,

allowing partitioning to act as the primary mechanism under an elevated ammo-
nium formate concentration.36 A plot of the logarithm of the mean experimentally
determined k′ versus the acetonitrile percentage is provided in Figure 2.5. A linear
relationship between log k′ and the percentage of acetonitrile in the mobile phase
(correlation coefficient of 0.996) was observed. This linear relationship is suggestive
of a partitioning mechanism.20
Based on these results, a total of 10â•›mM ammonium formate concentration was
employed in our generic HILIC system. An examination of Table 2.2 reveals that
the predicted k′ of compound Nos. 2–5, 9, and 16 were significantly underestimated,
which can possibly be attributed in part to secondary interactions since these com-
pounds all contained ionized amines at the mobile-phase pH of 3.0. Consideration
must be made that the scope of this work was to develop a generic HILIC system in
which the applicability of HILIC (k′ ≥ 2) could readily be determined. It should also
be mentioned that compounds that possess a negative charge may experience electro-
static repulsion with the ionized silanols of the unmodified silica stationary phase.20

2.5
2.0
Tailing factor (T )
1.5
1.0
0.5
0.0
0 5 10 15 20 25
Ammonium formate concentration (mM)
Figure 2.4 Influence of ammonium formate concentration on tailing factor (T). Error bars
represent ±1 standard deviation (n = 3). Chromatography was performed using an Atlantis™
HILIC silica column (5â•›μm). Pump A delivered the required make-up volume of water, Pump B
provided acetonitrile at a constant percent of 90% (v/v), and Pump C delivered the appropriate
molarity ammonium formate, pH 3, stock buffer (i.e., 10, 50, or 200â•›m M) to produce a net
mobile-phase concentration of 1, 5, 10, 15, and 20â•›m M. An ammonium formate concentration
of 10â•›m M or greater produced symmetrical peaks. (Reproduced from Kadar, E.P. and Wujcik,
C.E., J. Chromatogr. B, 877, 471, 2009.)
The second assumption is that the pH of the immobilized layer of water on the
silica stationary phase is in fact 3.0. ACD/LogD Suite calculates log D values over
the pH range of 0–14 in 0.1 unit increments. The log D value used for each com-
pound in the above relationship is specific for pH 3.0. Rosés and Bosch published a
review on the influence of mobile-phase acid-base equilibria on the chromatographic
behavior of protolytic compounds.39 Their article provides insight into the effect of
an organic solvent on mobile-phase pH and concludes that an organic solvent, such
as acetonitrile, in the mobile phase can change the pKa of the buffer as well as that
of the analyte. In their review, Rosés and Bosch present a thorough discussion of the
relationship between the pKa of a solute measured in pure water versus the pKa of a
solute in mobile phase containing an organic modifier.39 They found the pKa devia-
tion to be compound specific, which can be explained by a preferential solvation of
the compound by the components of the solvent mixture.39 This finding is important
with respect to our generic HILIC system, considering the relatively high acetonitrile
concentration (85%–95% [v/v]) of the mobile phase. The following generalizations
can be applied to the relationship between the experimentally determined k′ and the
calculated log DpH 3.0 for the generic HILIC system: (1) acids generally demonstrate
increasing pKa with increasing organic modifier concentration, thereby increasing
the pH of the buffer; (2) the formate, the buffer utilized in this work, would relate
to this trend; and (3) the pKa values of the protonated basic compounds, including
those investigated in this chapter, decrease with the addition of an organic solvent

1.4
y = 0.048x – 3.4768
1.2 R2 = 0.996
1.0
0.8
Log k΄
0.6
0.4
0.2
0.0
70 75 80 85 90 95 100
Acetonitrile percentage
Figure 2.5 Influence of acetonitrile mobile-phase concentration on capacity factor, k′.

Error bars represent ±1 standard deviation (n = 3). Chromatography was performed using an
Atlantis™ HILIC silica column (5â•›μm). The mobile phase was prepared by combining water
and acetonitrile with a 200â•›m M aqueous ammonium formate, pH 3.0, stock solution to pro-
duce five separate mobile-phase solutions with acetonitrile concentrations of 75%, 80%, 85%,
90%, and 95% (v/v) and a total ammonium formate concentration of 10â•›m M, with a flow rate
of 1.0â•›m L/min. The linear relationship demonstrates that a partitioning mechanism is present.
(Reproduced from Kadar, E.P. and Wujcik, C.E., J. Chromatogr. B, 877, 471, 2009.)
(minimum pKa obtained at approximately 60% acetonitrile) and then increase to a

pKa value for a pure organic solvent higher than for water.39 Currently, information
with respect to pKa values at the acetonitrile concentration used in generic HILIC
systems are unavailable in the literature. However, it is assumed that the buffer
capacity of the 10â•›m M ammonium formate employed in the mobile phase is adequate
to provide a consistent overall pH, thereby providing a minimum impact on the util-
ity of the relationship to rapidly predict the applicability of HILIC.
The third assumption is that ion-pairing does not make a significant contribution
to the relationship between the experimentally determined k′ and the calculated
log DpH 3.0 for our generic HILIC system. Formate has been demonstrated to serve
as the counterion for protonated amines in some reversed-phase chromatography
applications.40,41 The total formate concentration of the mobile phase in our generic
HILIC system was approximately 40â•›m M, considering the formic acid used to adjust
the pH of the 200â•›m M ammonium formate stock solution to 3.0. The extent of ion-
pair formation or salt precipitation is unknown for our generic HILIC system, espe-
cially given the high acetonitrile concentration of the mobile phase. However, it
can be assumed to be related to the pKa of a specific compound.42 The log DpH 3.0
calculations were made with the ion-pairing function of ACD/LogD Suite disabled
because this function is designed to simulate physiological conditions in which the

42
Table 2.3
Chemical Analog Internal Standard Selection Evaluation Results
85% Acetonitrile 90% Acetonitrile 95% Acetonitrile
Compound No. Exptl. k′a Pred. k′b % Diff. k′ Ratioc Exptl. k′a Pred. k′b % Diff. k′ Ratioc Exptl, k′a Pred. k′b % Diff. k′ Ratioc
3 7.18 4.56 12.51 8.41 18.94 8.57
4 7.62 4.05 11.71 7.47 11.45 7.56
k′ ratio 0.94 1.13 −19.5 1.07 1.13 −5.4 1.65 1.13 31.5
5 4.07 4.00 9.16 7.38 14.67 7.47
6 3.51 3.40 7.89 6.28 12.03 6.30
k′ ratio 1.16 1.18 −1.5 1.16 1.18 −1.2 1.22 1.19 2.8
8 1.80 2.22 3.40 4.09 3.06 4.01
10 2.00 2.01 3.80 3.71 2.73 3.60
k′ ratio 0.90 1.10 −22.7 0.89 1.10 −23.2 1.12 1.11 0.6
13 1.07 1.35 2.13 2.50 2.06 2.38
14 1.00 1.23 2.04 2.27 1.97 2.15
k′ ratio 1.07 1.10 −2.6 1.04 1.10 −5.5 1.05 1.11 −5.9
20 0.33 0.45 0.47 0.83 0.27 0.75
21 0.36 0.44 0.47 0.81 0.30 0.73
k′ ratio 0.92 1.02 −11.6 1.00 1.02 −2.5 0.90 1.03 −14.2

a Experimentally determined k′ (mean of n = 3).
b Predicted k′ utilizing respective Equations 2.6 through 2.8.
c Percent difference between ratios of experimentally determined and predicted k′.

ionic strength is 0.15â•›M.15 The ionic strength of the mobile phase used in our generic
HILIC system was considerably less (approximately 0.025â•›M).
The fourth assumption is that ACD/LogD Suite can provide accurate calculations
of log DpH 3.0. This calculation incorporates calculated log P and aqueous pKa values.
The error assumed for the calculated log P and aqueous pKa is ±0.3 and ±0.2 units,
respectively.15 It is possible that the calculated pKa error is actually greater since it
is calculated based on an aqueous environment and the generic HILIC system has a
relatively high acetonitrile content. Consideration should also be given to the impact
of the mobile-phase pH with respect to the high acetonitrile content that is discussed
in the review article by Rosés and Bosch.39 In our work, the calculated log DpH 3.0 has
been demonstrated to be adequate by providing clear linear relationships with the
experimentally determined k′.
From the assumptions made to formulate our hypothesis, it must be realized that
successful HILIC applicability predictions utilizing our generic HILIC system rely
on consistency and control of the solvent system to drive partitioning as the primary
mechanism.
2.4.3 Identification of Internal Standards

Chemical analog internal standards are typically utilized in LC-MS/MS bioanalysis
when stable label internal standards are unavailable. Matching the internal standard
and analyte retention time is desirable in order to minimize matrix effects such as
ionization suppression or enhancement. Typically, analyte/analog pairs are chosen
based on structural similarity without consideration for physicochemical properties,
such as log D. The selection of an analog with a similar structural character can be
beneficial by providing similar ionization efficiency. The analyte/analog pairs evalu-
ated in this work were selected based on both a similar structure and log D. Table 2.3
provides a tabulation of mean experimentally determined and predicted k′ ratios
for chemical analog pairs. Percent difference values between the experimentally
determined and predicted k′ ratios ranged from −1.5% to −22.7%, −1.2% to −23.2%,
and −14.2% to 31.5% for acetonitrile mobile-phase concentrations of 85%, 90%, and
95%, respectively. These relatively minor differences demonstrate the utility of using
log DpH 3.0 similarity as an internal standard selection criterion for HILIC.
2.5 Conclusions
A relationship between the experimentally determined k′ of a compound and its
calculated log DpH 3.0, utilizing a generic HILIC system, was established. The mini-
mization of possible secondary interactions such as adsorption and ionic interaction
was achieved by the careful selection of the stationary phase and buffer concentra-
tion. This work was in agreement with the HILIC partition mechanism theory of
Alpert.19 The above discussion of the assumptions made to formulate our hypothesis
clearly indicates the complexity of a HILIC system. However, the goal of this work
was to provide a tool to rapidly predict the applicability of a generic HILIC sys-
tem by simply entering the chemical structure of interest into the ACD/LogD Suite.
The accuracy rate of approximately 90% for the HILIC applicability predictions

clearly demonstrates the utility of the above-defined relationship between the experi-
mentally determined k′ and calculated log DpH 3.0 under a generic HILIC system.
Additionally, the identification of an appropriate chemical internal standard was
found to be predicted by a log DpH 3.0 similarity. The practical application of the
calculated physicochemical property, log D, as a predictive “in-silico” tool allows a
reduction in the bioanalytical method development time without compromising qual-
ity and, hence, improves the cost and resource efficiency of the bioanalytical labora-
tory. This approach adds considerable value to the area of LC-MS/MS bioanalysis.
Acknowledgment
The authors would like to acknowledge Douglas M. Fast, Covance Inc., Madison,
WI, for his support and review of this chapter.
References
1. Hsieh, Y., J. Sep. Sci., 31, 1481–1491, 2008.
2. Naidong, W., J. Chromatogr. B., 796, 209–224, 2003.
3. Xu, R.N., Fan, L., Rieser, M.J., El-Shourbagy, T.A., J. Pharm. Biomed. Anal., 44,
342–355, 2007.
4. Zhou, S., Song, Q., Tang, Y., Naidong, W., Curr. Pharm. Anal., 1, 3–14, 2005.
5. Nguyen, H.P., Schug, K.A., J. Sep. Sci., 31, 1465–1480, 2008.
6. Avery, M., Rapid Commun. Mass Spectrom., 17, 197–201, 2003.
7. Brown, S.D., White, C.A., Bartlett, M.G., Rapid Commun. Mass Spectrom., 16, 1871–
1876, 2002.
8. Cappiello, A., Famiglini, G., Pierangela, P., Pierini, E., Termopoli, V., Trufelli, H., Anal.
Chem., 80, 9343–9348, 2008.
9. Matuszewski, B.K., Constanzer, M.L., Chavez-Eng, C.M., Anal. Chem., 75, 3019–3030,
2003.
10. Sangster, T., Spence, M., Sinclair, P., Payne, R., Smith, C., Rapid Commun. Mass
Spectrom., 18, 1361–1364, 2004.
11. Mannhold, R., Molecular Drug Properties—Measurement and Prediction, Wiley-VCH
Verlag GmbH & Co. KGaA, Weinheim, Germany, 2008.
12. Scherrer, R.A., Howard, S.M., J. Med. Chem., 20, 53–58, 1977.
13. Lombardo, F., Shalaeva, M.Y., Tupper, K.A., Gao, F., J. Med. Chem., 44, 2490–2497,
2001.
14. ACD/LogD, ACD/Labs Release: 9.00, Product Version: 9.03, Advanced Chemistry
Development, Inc., Toronto, Ontario, Canada, Copyright © 1997–2005.
15. ACD/LogD, Suite Version 9.0 Reference Manual, Advanced Chemistry Development,
Inc., Toronto, Ontario, Canada, Copyright © 1994–2005.
16. Kadar, E.P., Wujcik, C.E., Wolford, D.P., Kavetskaia, O., J. Chromatogr. B, 863, 1–8,
2008.
17. Quiming, N.S., Denola, N.L., Siato, Y., Jinno, K., Anal. Bioanal. Chem., 388, 1693–
1706, 2007.
18. Quiming, N.S., Denola, N.L., Ueta, I., Siato, Y., Tatematsu, S., Jinno, K., Anal. Chim.
Acta, 598, 41–50, 2007.
19. Alpert, A.J., J. Chromatogr., 499, 177–196, 1990.
20. Hemström, P., Irgum, K., J. Sep. Sci., 29, 1784–1821, 2006.

21. Ikegami, T., Tomomatsu, K., Takubo, H., Horie, K., Tanaka, N., J. Chromatogr. A, 1184,
474–503, 2008.
22. Hao, Z., Xiao, B., Weng, N., J. Sep. Sci., 31, 1449–1464, 2008.
23. Jandera, R., J. Sep. Sci., 31, 1421–1437, 2008.
24. Snyder, L.R., Kirkland, J.J., Glajch, J.L., Practical HPLC Method Development, 2nd
edn., John Wiley & Sons, Inc., New York, 1997, pp. 29, 237.
25. Center for Drug Evaluation and Research, U.S. Food and Drug Administration, Reviewer
Guidance, Validation of Chromatographic Methods, FDA, Rockville, MD, November
1994.
26. Liu, X., Tanaka, H., Yamauchi, A., Testa, B., Chuman, H., Helv. Chim. Acta, 87, 2866–
2876, 2004.
27. Liu, X., Tanaka, H., Yamauchi, A., Testa, B., Chuman, H., J. Chromatogr. A, 1091,
51–59, 2005.
28. Vailaya, A., J. Liq. Chromatogr. Relat. Technol., 28, 965–1054, 2005.
29. Kaliszan, R., Quant. Struct.-Act. Relat., 9, 83–87, 1990.
30. Piraprez, G., Herent, M., Collin, S., Flavour Fragrance J., 13, 400–408, 1998.
31. Lipinski, C., Lombardo, F., Dominy, B.W., Feeney, P.J., Adv. Drug. Deliv. Rev., 23,
3–25, 1997.
32. Lombardo, F., Shalaeva, M.Y., Tupper, K.A., Gao, F., Abraham, M.H., J. Med. Chem.,
43, 2922–2928, 2000.
33. Stella, C., Galland, A., Liu, X. et al., J. Sep. Sci., 28, 2350–2362, 2005.
34. ACD/LogP, ACD/Labs Release 11.00, Product Version 11.01, Advanced Chemistry
Development, Inc., Toronto, Ontario, Canada, Copyright © 1994–2007.
35. Nozawa, Y., Kawashima, A., Hashimoto, E.H., Kato, H., Harada, K., J. Chromatogr. A,
1216, 3807–3811, 2009.
36. Kadar, E.P., Wujcik, C.E., J. Chromatogr. B, 877, 471–476, 2009.
37. Cazes, J., Scott, R.P.W., Chromatography Theory, Marcel Dekker, Inc., New York, 2002,
pp. 62–73.
38. Nawrocki, J., J. Chromatogr. A, 779, 29–71, 1997.
39. Rosés, M., Bosch, E., J. Chromatogr. A, 982, 1–30, 2002.
40. Bianchi, F., Careri, M., Corradini, C., Elviri, L., Mangia, A., Zagnoni, I., J. Chromatogr.
A, 825, 193–200, 2005.
41. Roberts, J.M., Diaz, A.R., Fortin, D.T., Friedle, J.M., Piper, S.D., Anal. Chem., 74,
4927–4932, 2002.
42. Stahl, P.H., Wermuth, C.G., Handbook of Pharmaceutical Salts: Properties, Selection,
and Use, VHCA, Verlag Helvetica Chimica Acta, Zürich, Switzerland and Wiley-VCH,
Weinheim, Federal Republic of Germany, 2002.

3 HILIC Behavior of
Reversed-Phase/Ion-
Exchange Mixed-Mode
Stationary Phases and
Their Applications
Xiaodong Liu and Christopher A. Pohl
Contents
3.1 Introduction..................................................................................................... 48
3.2 Hydrophilic Interaction Liquid Chromatography............................................ 48
3.2.1 Introduction......................................................................................... 48
3.2.2 Retention Mechanism.......................................................................... 49
3.2.3 Column Chemistry.............................................................................. 49
3.2.4 Applications......................................................................................... 50
3.3 Reversed-Phase/Ion-Exchange Mixed-Mode Stationary Phases..................... 50
3.3.1 Introduction......................................................................................... 50
3.3.2 Column Chemistry.............................................................................. 51
3.3.3 Separation Mechanism........................................................................ 52
3.3.4 Applications......................................................................................... 53
3.4 HILIC Behavior of Reversed-Phase/Ion-Exchange Mixed-Mode
Stationary Phases............................................................................................. 55
3.4.1 Retention Behavior of Xanthins, Nucleosides,
and Water- Soluble Vitamins on RP/WAX Mixed-Mode
Columns in HILIC Mode.................................................................... 56
3.4.2 Retention Behavior of Fungal Metabolites on RP/WAX
Mixed-Mode Columns in HILIC Mode.............................................. 57
3.4.3 HILIC Behavior Comparison of RP/IEX Bimodal Mixed-Mode
Columns............................................................................................... 58
3.4.4 HILIC Behavior of RP/WAX/SCX Trimodal Mixed-Mode
Columns............................................................................................... 62
3.5 Applications of Reversed-Phase/Ion-Exchange Mixed-Mode Columns
in the HILIC Mode.......................................................................................... 62
3.5.1 Pharmaceutical Active Ingredients and Counterions.......................... 62
3.5.2 Melamine and Cyanuric Acid..............................................................64
47

3.5.3 Buffer Salts for Biopharmaceutical Applications................................ 68

3.5.4 Determination of Ethanol Consumption Markers............................... 70
3.6 Concluding Remarks....................................................................................... 71
Acknowledgments..................................................................................................... 72
References................................................................................................................. 73
3.1 Introduction
Reversed-phase/ion-exchange mixed-mode chromatography has gained increasing
attention in recent years. The main driver is that such techniques provide control of
selectivity through changes in the mobile-phase ionic strength, the pH, the organic
solvent content, and/or the salt additives. As a result, options are now available to
achieve optimal separations when compared to other separation modes, namely,
reversed-phase liquid chromatography (RPLC), normal-phase liquid chromatog-
raphy (NPLC), hydrophilic interaction liquid chromatography (HILIC), and ion-
exchange liquid chromatography (IEX-LC). The adjustable selectivity feature results
from the fact that mixed-mode chromatography combines multiple retention modes
in a single stationary phase, including RP/IEX, RP/HILIC, and IEX/HILIC combi-
nations, each of which can be controlled independently.
Analytes in a mixture can be different in hydrophobicity, polarity, charge, and
size. RPLC is the most commonly used separation mode, accounting for approxi-
mately 80%–90% of all HPLC analytical separations. A major limitation of RPLC
is its lack of adequate retention for polar molecules. HILIC is widely used for the
separation of highly polar substances including biologically active compounds, such
as pharmaceutical drugs, neurotransmitters, nucleosides, nucleotides, amino acids,
peptides, proteins, oligosaccharides, carbohydrates, etc. HILIC is an important sepa-
ration mode complementary to RPLC. In addition, IEX-LC is often employed to
separate charged molecules. Although HILIC and IEX-LC can retain polar mol-
ecules, their low hydrophobicity makes them unsuitable for separating hydrophobic
molecules. RPLC, HILIC, and IEX-LC combined together provide complementary
solutions to most HPLC separations. But very often, more than one separation mode
is needed to achieve desirable results. By comparison, RP/IEX mixed-mode columns
can operate in various separation modes and provide broader application ranges.
This chapter discusses RP/IEX mixed-mode stationary phases, including column
chemistry, chromatographic features, and applications. The HILIC behavior of these
stationary phases is also discussed in detail. Some applications of RP/IEX mixed-
mode columns in the HILIC mode are provided. As HILIC is thoroughly discussed
in other chapters, only a brief review of this subject is given in this chapter.
3.2 Hydrophilic Interaction Liquid Chromatography

3.2.1 Introduction
While RPLC is the most powerful separation technique for a wide range of mol-
ecules, a major limitation of RPLC is its lack of adequate retention of polar mol-
ecules. Normal-phase liquid chromatography (NPLC), often with silica as the
stationary phase, is used to separate compounds according to their polar features,

HILIC Behavior of Reversed-Phase/Ion-Exchange 49
usually with nonaqueous mobile phases, such as hexane, ethyl acetate, chloroform,
etc. Therefore, NPLC is a useful separation technique as it provides selectivity com-
plementary to RPLC. However, it is difficult to dissolve hydrophilic compounds,
such as peptides, in nonaqueous mobile phases, thus limiting its application. In
1975, Linden reported the work on carbohydrate separations using an unmodified
silica stationary phase in the NPLC mode but in an aqueous-containing mobile
phase.1 In these separations, the mobile phase had a relatively high water content,
which helps polar molecules dissolve into the mobile phase. This technique was
later named as hydrophilic-interaction liquid chromatography (HILIC) by Alpert
in his 1990 paper on the separation of hydrophilic substances, such as proteins,
peptides, and nucleic acids.2 Since then, HILIC has gained much attention because
of the increasing need to analyze polar compounds in complicated mixtures, as
well as the widespread use of LC-MS.3,4
3.2.2 Retention Mechanism

HILIC is a type of NPLC that uses mobile phases typically containing 5%–40%
water. In HILIC, the retention increases with an increasing polarity of the stationary
phase and the analyte, and with the decreasing polarity of the predominantly organic
solvent in the mobile phase, a trend opposite to that of RPLC.
The retention mechanism of HILIC is postulated as an analyte partitioning
between a water-rich layer of stagnant solvent sequestered on a hydrophilic station-
ary phase and a mobile phase rich in organic solvent, with the main components
typically composed of 5%–40% water in acetonitrile. The more polar the analyte,
the more it associates with this stagnant aqueous phase and the later it elutes—a
typical NPLC behavior. This partitioning mechanism can satisfactorily explain the
retention of very hydrophilic and uncharged analytes. The mobile phase comprises
a high percentage of an organic solvent (typically acetonitrile), which is modified by
a small percentage of the water/volatile buffer. The water-rich liquid layer is estab-
lished within the stationary phase, thus partitioning solutes from the mobile phase
into the hydrophilic layer. The primary mechanism of separation is partitioning
based on hydrophilic interaction, and a secondary mechanism, which also influences
selectivity, is the electrostatic interactions between charged analytes and charged
stationary phases. Elution is enabled by increasing the mobile-phase polarity (i.e.,
the content of the water component). Several studies on the HILIC retention mecha-
nism have been published.2,3,5–7
3.2.3 Column Chemistry

In principle, any stationary phase that has the polar functionality to form a water-rich
layer on its surface could be used for HILIC applications. Suitable polar functionalities
include silanols on unmodified silica, neutral hydrophilic moieties (e.g., hydroxyl,
cyano, amide), as well as anionic (e.g., amino), cationic (e.g., carboxylate), zwitter-
ionic (e.g., sulfoalkylbetaine and phosphorylcholine), and amphoteric moieties. A wide
variety of column chemistries are available for HILIC separations, which have been
summarized in a recent review.4

3.2.4 Applications
HILIC is a complementary technique to RPLC with several benefits over RPLC. First,
it retains polar analytes that cannot be retained on an RP column. For electrospray
LC/MS applications with very polar compounds, the organic solvent-rich mobile
phase used in HILIC provides a 10- to 20-fold sensitivity improvement. Moreover, by
eliminating the need for an evaporation and reconstitution of a sample dissolved in a
nonaqueous solvent, the sample analysis throughput can be greatly increased.
HILIC is useful for an analysis of polar solutes as RPLC is for nonpolar sol-
utes. Since 1990, HILIC has been applied to a wide variety of applications includ-
ing carbohydrates,8,9 peptides,6,10–15 proteins,16–19 pharmaceuticals,20–25 as well as
other highly polar molecules, such as urea,26,27 biurea,28 choline and butyrobetaine,29
tromethamine,30 ascorbic acid and related compounds,31,32 folic acid and its metab-
olites,33 nicotine and its metabolites,34 saponins,35 aminoglycoside antibiotics,36 glu-
cosinolates,37 ionic liquids,38 organophosphonate nerve agent metabolites,39 etc.
3.3 Reversed-Phase/Ion-Exchange Mixed-Mode
Stationary Phases
3.3.1 Introduction
RPLC is most commonly used for analyzing molecules of intermediate or higher
hydrophilicity, but it often fails to retain highly hydrophilic analytes, such as cat-
echolamines, organic acids, and inorganic ions (e.g., Na+ and Cl−). Although ion-
pairing chromatography improves the retention and selectivity of highly hydrophilic
ionizable analytes, it requires a long equilibration time and a column dedicated to
use with ion-pairing mobile phases. Additionally, the mobile phase is not compatible
with MS. Ion-exchange (IEX) liquid chromatography is used to separate charged
molecules, but it is not generally suitable for separating neutral molecules because
IEX columns normally provide inadequate hydrophobic retention for neutral com-
pounds. HILIC complements the aforementioned techniques and may help overcome
retention and selectivity problems for polar acidic solutes. However, this approach
has limitations when compounds with a wide range of physicochemical properties
have to be analyzed within the same analysis. Other drawbacks include low solubil-
ity of analytes in a high organic solvent and greater organic solvent consumption.
Mixed-mode columns come in various forms depending on different combina-
tions of interactive mechanisms and different geometrical and spatial arrangements
of functional groups. The reversed-phase/ion-exchange (RP/IEX) mixed-mode is
the most common and effective combination because of the orthogonality of the
hydrophobic and ion-exchange retention mechanisms. Such stationary phase materi-
als separate analytes by hydrophobicity and charge differences.
RP/IEX mixed-mode liquid chromatography combines both RPLC and IEX-LC.
A typical RP/IEX mixed-mode stationary phase has both reversed-phase and ion-
exchange properties, and facilitates the independent control of retention for both ionizable
and neutral molecules. Therefore, many application challenges involving hydrophilic
ionizable compounds that are difficult for an RP column can be easily resolved on a
mixed-mode column. In method development, column selectivity is one of the most

important factors determining the success of a separation. Therefore, it is highly desir-

able to have a column that has adjustable selectivity so that an optimal separation can be
easily achieved. This contrasts with conventional RPLC columns (i.e., C18 or C8) where
the extent of selectivity control is rather limited. Moreover, with such a mixed-mode
column, it is possible to separate several different types of molecules in a single run.
3.3.2 Column Chemistry

RP/IEX mixed-mode chromatography has been known for more than 20 years and
this concept has also been applied to solid-phase extraction-based sample prepara-
tion.40 Its use as an HPLC column material has been widely reported.41–62 Depending
on the combination of functional groups, they can be grouped into RP/anion-exchange
(AEX) and RP/cation-exchange (CEX) bimodal, as well as RP/AEX/CEX trimodal
materials. According to the arrangement of functional groups in the packing mate-
rials, bimodal columns may be classified into four categories (Figure 3.1). Type I
materials contain a mixed-mode bed by blending two different stationary phases, such
as RP and IEX (either AEX or CEX materials) in a single column. Such columns are
commercially available from Thermo-Fisher under the trade name Hypersil Duet.63
Type II materials consist of bonded phases modified with a mixture of RP and IEX
ligands. It is likely that the C8/Anion and C8/Cation mixed-mode columns offered by
Grace belong to this category.64 It should be noted that both Type I and Type II materials
have the inherent drawback of undesirable selectivity drift due to the difference in hydro-
lytic stability between the RP ligand-bonded sites and the IEX ligand-bonded sites.42
Both Type III and Type IV mixed-mode materials are prepared by functionalizing
the silica substrate with single silyl ligands containing both RP and IEX domains.
Type III materials are modified with ion-exchange functionality embedded in the
alkyl ligands. Type IV materials are bonded with ion-exchange functionality at the
terminus of the alkyl ligand. As a result, Type III materials are generally more hydro-
phobic and can be viewed as IEX-modified RP packings. Type IV materials exhibit
unique IEX property supplemented by RP characteristic so that it may be considered
as RP-modified IEX materials. Both Type III and Type IV packings represent state-
of-the-art mixed-mode column technology, and provide improved selectivity rugged-
ness because of the constant ratio between the RP and IEX bonded sites. Various Type III
Silica gel Silica gel Silica gel

I. Blend bed II. Mixed ligand III. “Embedded” IV. “Tipped”
RP–black; IEX–gray
Figure 3.1 Four types of reversed-phase/ion-exchange bimodal mixed-mode columns

classified by the arrangement of functional groups. (Courtesy of Marketing Communication
Group at Dionex Corporation, Sunnyvale, CA.)

mixed-mode columns have been commercialized by SIELC under the trade name
Primesep.65 Recently, Type IV mixed-mode stationary phases have received consid-
erable attention by both academia and industrial research organizations. Several RP/
weak anion-exchange (WAX) materials consisting of a selector immobilized onto
thiol-modified silica gel have been reported.53,54,56–59 In these materials, the WAX site
is located on the outer surface of the lipophilic layer and is linked to the hydrophilic
silica support via a lipophilic spacer with polar-embedded amide and sulfide groups.
Meanwhile, Dionex has introduced two Type IV mixed-mode columns—Acclaim®
Mixed-Mode WAX-161,66 and Acclaim Mixed-Mode WCX-1.62,67
Because of the complexity and variety of HPLC analytes in terms of hydrophilicity
and ionization, it is highly desirable to separate anionic, cationic, and neutral molecules
within a single analysis. This necessitates a trimodal stationary phase that can pro-
vide cation-exchange, anion-exchange, and reversed-phase retention simultaneously.
Packing materials with amphoteric or zwitterionic functionality have been developed.
However, because anion-exchange and cation-exchange functionalities on these mate-
rials neutralize one another due to their proximity if both functional groups are ion-
ized, they may retain and separate both anions and cations via salt exchange rather than
anion-exchange and cation-exchange processes. Consequently, anion-exchange and
cation-exchange retentions cannot be independently controlled. Recently, a trimodal
column based on Nanopolymer Silica Hybrid (NSH) technology was commercialized
by Dionex under the trade name of Acclaim Trinity™ P1.68 This material consists of
high-purity porous spherical silica gel coated with charged nanopolymer particles. The
inner-pore area of the silica gel is modified with an organic layer that provides both
reversed-phase and anion-exchange properties. The outer surface is modified with
cation-exchange functionality (Figure 3.2). This chemistry ensures the distinctive spa-
tial separation of the anion-exchange and cation-exchange regions, which allows both
retention mechanisms to function simultaneously and be controlled independently.
A number of commercial silica-based63–68 as well as polymer-based RP/IEX
mixed-mode columns69 are available from various suppliers and are listed in Table 3.1.
3.3.3 Separation Mechanism

The retention properties of IEX-based mixed-mode stationary phases are the result
of the complex interactions of reversed-phase, ion-exchange, and ion-exclusion
retention mechanisms. The relative contribution of each mechanism depends on the
hydrophobicity and charge character of solutes and chromatographic conditions,
such as mobile phase ionic strength, pH, and organic solvent composition. A sim-
ple empirical stoichiometric displacement model can be used for discussion.70–72
According to this model, plots of log k versus log counterion concentration (C) are
linear, and according to the following equation:70
log k = log Kz − Z log C
where in
Kz is a constant related to the ion-exchange equilibrium constant and the ion-
exchange capacity
Z is the ratio of the valencies of the solute ion (s) and counterion (c) (Z = s/c)

(c)
(b)
(a)
Figure 3.2 Column chemistry of Acclaim Trinity P1—the first reversed-phase/anion-

exchange/cation-exchange trimodal mixed-mode column based on NSH technology: (a) overview
of a silica particle coated with nanopolymer beads, (b) enlarged view of the silica surface coated
with negatively charged nanopolymer beads, (c) inner-pore area consists of reversed-phase and
weak anion-exchange functionalities and the outer surface provides strong cation-exchange inter-
action. (Courtesy of Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)
The empirical linear relationship between log k and log C is useful to assess the exis-
tence of an ion-exchange process. An examination of the slope Z gives the number of
charges involved in the ion-exchange process, where, for a monovalent counterion,
the slope is representative of the charge on the solute. In general, all mechanisms
function independently, and can be modified as needed by adjusting the mobile
phase ionic strength, the salt type, the pH, and the organic solvent content.53,56–59
3.3.4 Applications
RP/IEX mixed-mode liquid chromatography is not a new concept. However, the
fact that multiple retention mechanisms provide a valuable alternative or comple-
ment to RP, IEX, and HILIC phases in a variety of HPLC applications has not been
fully recognized until recently. These stationary phases are especially useful for
retaining highly hydrophilic-charged analytes that are difficult for an RP column to
retain. Furthermore, the increasing necessity of complementary analysis methods for
validation of RPLC assay specificity required by regulatory authorities for pharma-
ceutical impurity profiling and stability-indicating assays has driven the need for the
use of new stationary phases with complementary selectivity. Moreover, to analyze

Table 3.1
Commercial Reversed-Phase/Ion-Exchange Mixed-Mode Columns
Column Name Manufacturer Type of Column Chemistry Substrate References
Hypersil™ Duet Thermo-Fisher Type I, RP/SAX, bimodal Silica gel, 5â•›μm, [63]
C18/SAX mixed-mode column 120â•›Å, 170â•›m2/g
Hypersil Duet Thermo-Fisher Type I, RP/SCX, bimodal Silica gel, 5â•›μm, [41,42,63]
C18/SCX mixed-mode column 120â•›Å, 170â•›m2/g
Mixed-Mode Grace Type II, RP/anion-exchange, Silica gel, 7â•›μm, [47,64]
C8/anion bimodal mixed-mode 100â•›Å, 350â•›m2/g
column
Mixed-Mode Grace Type II, RP/cation- Silica gel, 5â•›μm, [47,64]
C8/cation exchange, bimodal 100â•›Å, 350â•›m2/g
mixed-mode column
Mixed-Mode Grace Type II, RP/cation- Silica gel, 5 or [64]
C18/cation exchange, bimodal 7â•›μm, 100â•›Å,
mixed-mode column 350â•›m2/g
Primesep• 100 SIELC Type III, RP/SCX, bimodal Silica gel, 5 or [65]
mixed-mode column 10â•›μm, 100â•›Å
Primesep B SIELC Type III, RP/SAX, bimodal Silica gel, 5 or [65]
mixed-mode column 10â•›μm, 100â•›Å
Acclaim• Dionex Type IV, RP/WAX, bimodal Silica gel, 3 or [61,66]
Mixed-Mode mixed-mode column 5â•›μm, 120â•›Å,
WAX-1 300â•›m2/g
Acclaim Dionex Type IV, RP/WCX, bimodal Silica gel, 3 or [62,67]
Mixed-Mode mixed-mode column 5â•›μm, 120â•›Å,
WCX-1 300â•›m2/g
Obelisc• R SIELC Amphoteric, RP/anion- Silica gel, 5 or [65]
exchange/cation-exchange, 10â•›μm, 100â•›Å
mixed-mode column
Acclaim Trinity Dionex RP/WAX/SCX, trimodal Silica gel, 3â•›μm, [68,80]
P1 mixed-mode column based 300â•›Å, 100â•›m2/g
on NSH technology
OmniPac• Dionex RP/SAX, mixed-mode Ethylvinylbenzene/ [69]
PAX column by coating the divinylbenzene
polymer beads with alkanol polymer beads
quaternary amine- (55% cross-
functionalized latex linked), 8.5â•›μm
OmniPac PCX Dionex RP/SCX, mixed-mode Ethylvinylbenzene/ [69]

column by coating the divinylbenzene
polymer beads with polymer beads
sulfonic acid- (55% cross-
functionalized latex linked), 8.5â•›μm

complex samples successfully, multidimensional separation and analysis methods

are often needed, which requires that the columns used in different multidimen-
sional LC assays are orthogonal or complementary, and that the mobile phases in
these dimensions are compatible. In this sense, RP/IEX mixed-mode columns may
be a viable alternative or replacement to the conventional columns for established
multidimensional LC methods.
Mixed-mode stationary phase materials separate analytes by both hydrophobic-
ity and charge differences. The combination of orthogonal retention mechanisms of
mixed-mode columns gives rise to a host of advantages, such as adequate retention of
charged analytes, selectivity complementary to RP, IEX, and/or HILIC columns, and
adjustable selectivity. As a result, greater flexibility for retention and selectivity tun-
ing during method development and an expanded application range can be achieved.
Applications that have been reported in literature using RP/IEX mixed-mode
columns include separations of biomolecules (e.g., proteins,73–75 peptides,52,53,76–78
oligonucleotides,43,44 and RNAs45,46), metabolism studies (e.g., organophosphate54
and catecholamines79 in urine), ethanol consumption marker analysis in urine,56
chromatographic profiling with mycotoxins,59 separation of anthocyanin compounds
from grapes,48 as well as various pharmaceuticals applications.41,42,49–51
Compared to RP/AEX or RP/CEX bimodal columns, an RP/AEX/CEX trimodal
column should provide a broader application range since anionic, cationic, and neu-
tral analytes could be separated within a single analysis due to the combined effects
of reversed-phase, anion-exchange, and cation-exchange retention mechanisms.
This feature is important for the determination of active pharmaceutical ingredi-
ents (APIs) and counterions. Traditionally, APIs and counterions are analyzed in
two separate assays using different LC columns and even different instruments. For
example, Na+ and Cl− are most commonly used counterions in drug formulation and
are unretained by any RP columns. It has been demonstrated that both commonly
used pharmaceutical counterions and their respective APIs can be separated within
a single analysis on a commercial RP/AEX/CEX trimodal column.80
3.4 HILIC Behavior of Reversed-Phase/Ion-Exchange

Mixed-Mode Stationary Phases
RP/IEX mixed-mode stationary phases promise great versatility and capability for
retaining and separating a variety of charged polar compounds in addition to more
apolar ionic and nonionic analytes. The combination of an alkyl chain with embed-
ded or terminus ion-exchange functionality generates hydrophobic, electrostatic, as
well as hydrophilic interactive domains. Because of their unique column chemistries,
these mixed-mode materials can be operated in a variety of chromatographic modes,
including the RP mode (e.g., for neutral molecules), the IEX mode (for solutes bear-
ing the opposite charge), the ion exclusion chromatography mode (for solutes having
the same charge), and the HILIC mode (polar neutral, basic, amphoteric, and acidic
compounds), depending on the chromatographic conditions and the characteristics of
the analytes. While a number of studies on retention behavior and applications in RP
and IEX modes have been reported, investigations of RP/IEX mixed-mode columns

in the HILIC mode have been scarce. Recently, retention properties of mixed-mode
columns under HILIC conditions (mobile phases with high acetonitrile content) have
been the subject of research of both academia and industries.53,56,58,59,81
3.4.1 Retention Behavior of Xanthins, Nucleosides, and Water-Soluble

Vitamins on RP/WAX Mixed-Mode Columns in HILIC Mode
Lämmerhofer et al. reported a comparative HILIC evaluation58 of several in-house
developed RP/WAX mixed-mode phases along with commercial RP/WAX mixed-
mode columns, as well as a reference HILIC column (TSK-Gel Amide-80) having a
neutral surface. Column chemistries of selected stationary phases in this study are
illustrated in Figure 3.3. Three different commonly used HILIC test mixtures were used
in the study, including xanthins (caffeine, theobromine, and theophylline), nucleosides
(adenosine, cytidine, guanosine, thymidine, and uridine), and water-soluble vitamins
(ascorbic acid, nicotinic acid, pyridoxine, riboflavin, and thiamine). HILIC mode
separation was demonstrated for xanthins because caffeine (trimethyl-xanthine)
eluted before the more polar 1,3-dimethyl xanthins (theobromine and theophylline)
on all of the stationary phases tested. However, while theobromine eluted before the-
ophylline on the TSK-Gel Amide-80 phase, the elution order of these two analytes
was reversed on the basic RP/WAX phases, presumably because theophylline (pKa,
theophylline = 8.6) is more acidic than theobromine (pKa, theobromine = 9.9) so that
electrostatic interaction was superimposed upon HILIC interaction.
For nucleosides, none of the columns followed the elution order predicted based
on log D values that should be in the order of adenosine < thymidine < uridine <
guanosine < cytidine. This deviation, including the elution order and the elution pro-
files cannot be explained by a partition mechanism alone, thus specific interactions
between analytes and ion-exchange functional groups of the stationary phase are
most likely to be involved in the retention.
O
–
N S O NH2
+ O
(a) (b)
H H2N O
S N
R H
N
O H n m
(c)
R
O
CH3
N N
H CH3
(d) (e)
Figure 3.3 Column chemistries of stationary phases in the study of Ref. [58]. (a) ZIC-
HILIC (Merck), (b) Luna NH2 (Phenomenex), (c) RP/WAX-AQ, (d) Acclaim Mixed-Mode
WAX-1 (Dionex), and (e) TSK-Gel Amide-80 (Tosoh).

1.8 1.8
Nicotinic acid
Ascorbic acid
1.4 1.4 Theophylline
Nicotinic acid
Log k (ZIC-HILIC)
Pyridoxine
Log k (Luna NH2)

1.0 1
Uridine
0.6 Thiamine 0.6
0.2 0.2 Riboflavine
–0.2 Riboflavine
–0.2 Thiamine
Theophylline
–0.6 –0.6
–0.4 0 0.4 0.8 1.2 1.6 2 –0.4 0 0.4 0.8 1.2 1.6 2
(a) Log k (TSK-Gel Amide-80) (b) Log k (TSK-Gel Amide-80)
1.9
Ascorbic acid 2.0
Nicotinic acid Ascorbic acid
1.4 Log k (Acclaim MM WAX-1)
Log k (RP/WAX-AQ360)
1.5 Nicotinic acid

Theophylline
0.9 Theophylline
Pyridoxine 1.0
Thymidine
0.4 0.5
–0.1 0.0 Riboflavine
–0.6 –0.5
Thiamine
Thiamine Pyridoxine
–1.1 –1.0
–0.4 0 0.4 0.8 1.2 1.6 2 –0.4 0 0.4 0.8 1.2 1.6 2
(c) Log k (TSK-Gel Amide-80)
(d) Log k (TSK-Gel Amide-80)
Figure 3.4 Orthogonality plots of log k values from HILIC tests of TSK-Gel Amide-80
versus (a) ZIC-HILIC, (b) Luna NH2, (c) RP/WAX-AQ360, and (d) Acclaim Mixed-Mode
WAX-1. Besides the fitted model, the plots show the confidence (inner lines) and prediction
limits (outer lines). (Reproduced from Lämmerhofer, M. et al., J. Sep. Sci. 31, 2572, 2008.
Copyright Wiley-VCH Verlag GmbH & Co. KGaA. With permission.)
The retention profile of the vitamins test mix was even more varied among dif-
ferent column chemistries because the ionic analytes in this test mix were expected
to show strongly deviating retention profiles on stationary phases with cationic,
anionic, zwitterionic, and neutral chromatographic bonding chemistries (Figure 3.4).
Nevertheless, it appears that the retention profile differences observed are an indi-
cation of the importance of specific interaction forces between ion-exchange sites
of the stationary phase and complementary solute functional groups. A partition
mechanism alone fails to explain the altered retention profiles and shifts in elution
orders with different stationary phases.
3.4.2 Retention Behavior of Fungal Metabolites on RP/WAX

Mixed-Mode Columns in HILIC Mode
Apfelthaler et al. investigated the retention pattern profiling of fungal metabo-
lites using the in-house developed Type IV RP/WAX mixed-mode phases.59 In
their study, retention factors k of a neutral solute (aflatoxinB1), a hydrophilic

acid (moniliformin), a lipophilic acid (gibberellic acid), a base (lysergol), and an

amphoteric compound (lysergic acid) were mapped in dependence of the organic
modifier percentage (between 40% and 80% acetonitrile or methanol) using 3D
plots (Figure 3.5). It was found that while nonionic aflatoxin B1 (log P = 0.45) fol-
lowed a typical RP retention mechanism, retention of the hydrophilic acid monili-
formin (log DpH3 = −3.75; log DpH7 = −4.22; pKa = −0.21) increased slightly with the
mobile-phase organic solvent increase, suggesting a HILIC mechanism. In addi-
tion, a strong dependency of retention on mobile phase pH indicated the presence
of anion-exchange stationary phase characteristics. For gibberellic acid (log D pH3 =
−0.02; log DpH7 = −2.81; pKa = 4.13), a slight U-shaped organic solvent dependency
with a minimum of around 50%–60% was observed for acetonitrile-based mobile
phases, indicating an anion-exchange process along with an RP mechanism below
50%–60% acetonitrile together with a HILIC mechanism at higher solvent per-
centages. Retention maps for the amphoteric lysergic acid (log D pH3 = −0.60; log
DpH7 = −0.37; pKa1 = 3.09; pKa2 = 9.18) generally revealed a maximum retention
at pH 7.5 for both acetonitrile and methanol, approximately 1â•›pH unit above the
calculated isoelectric point of the solute in an aqueous medium. Loss of retention
at a higher pH may be explained by a reduced ionization of the stationary phase
ion-exchange groups, and at a lower pH by a diminished anionic character of the
lysergic acid. Increasing retention with an increasing acetonitrile content at high
mobile phase acetonitrile percentages indicated a HILIC-type retention mecha-
nism. Retention profiles of the basic lysergol (log D pH3 = −1.34; log DpH7 = 0.41;
pKa1 = 8.33; pKa2 = 14.61) can be largely explained by RP behavior, along with
repulsive electrostatic interactions.
3.4.3 HILIC Behavior Comparison of RP/IEX Bimodal

Mixed-Mode Columns
A recent study on the retention behavior of hydrophilic neutral (uracil), anionic
(nitrate), and cationic (1,1-dimethylbiguanide) probes for both Type III (ion-exchange
embedded alkyl chain) and Type IV (alkyl chain with an ion-exchange terminus) RP/
IEX mixed-mode columns (column chemistries illustrated in Figure 3.6) showed
that at 90% acetonitrile, the retention of uracil was in the order of Acclaim Mixed-
Mode WAX-1 > Lichrosorb Diol > Acclaim Mixed-Mode WCX-1> Primesep B >
Primesep 100.81 It is clear that both Type III and Type IV mixed-mode columns
possess HILIC properties, and the latter are generally more hydrophilic because of
their more hydrophilic surface. Therefore, Type IV mixed-mode columns are better
suited to HILIC applications.
For ionic or ionizable polar analytes, such as nitrate and 1,1-dimethylbigua-
nide, the retention behavior is more complicated. For an inorganic anion, nitrate,
two RP/AEX mixed-mode columns (Mixed-Mode WAX-1 and Primesep B) show
strong electrostatic attractions and retention decreases continuously with an
increasing solvent content in the mobile phase, up to approximately 80%–90%
acetonitrile. A further increase in the solvent content results in a retention increase
on Mixed-Mode WAX-1 column—a typical HILIC behavior, but virtually no
change for the Primesep B column (Figure 3.7). Under the same chromatographic

HILIC Behavior of Reversed-Phase/Ion-Exchange
20 50 To front
40
40
15
Retention factor
Retention factor
Retention factor
30
30
10
20
20
5
10 10
80 80 80
0 75 0 75 0 75
65 difier
8.0 70 er 8.0 70 r 8.0 70
7.5 65 ifi 7.5 65 ifie 7.5
7.0 60 od 7.0 60 od 7.0 60 om
pH
6.5
6.0 50 55 nic m pH
6.5
6.0 50 55 nic m pH
6.5
6.0 50 55 anic
(a) 5.5 40 45 Orga (b) 5.5 40 45 Orga (c) 5.5 40 45 Org
% % %
20
15
15
Retention factor
Retention factor
10
10
5
5
80 0 80
0 75 8.5 75
8.0 70 8.0 70 er
65 er 65
7.5
7.0 60 o difi 7.5
7.0 60 o difi
6.5
6.0 50 55 nic m pH
6.5
6.0 50 55 nic m
pH
(d) 5.5 40 45 Orga (e) 5.5 40 45 Orga
% %
Figure 3.5 Retention pattern of (a) neutral (aflatoxin B1), (b) acidic (moniliformin), (c) acidic with hydrophobic properties (gibberellic acid; note:
40% MeOH spike pointing to front), (d) amphoteric (lysergic acid), and (e) basic (lysergol) analytes on AQ-RP/WAX as a function of organic modifier
content and pH of the eluent under isocratic elution conditions (constant amount of 25â•›m mol L−1 HOAc). Darker shade (semitransparent): methanol,
lighter shade: acetonitrile. (Reproduced from Apfelthaler, E. et al., J. Chromatogr. A, 1191, 171, 2008. Copyright Elsevier. With permission.)
59
O
CH3
N N
H CH3
(a)
O
OH
N
H O
(b)
+
B
(c)
–
A
(d)
OH
O
OH
(e)
Figure 3.6 Column chemistries of stationary phases in the study of Ref. [81]. (a) Acclaim
Mixed-Mode WAX-1 (Dionex), (b) Acclaim Mixed-Mode WCX-1 (Dionex), (c) Primesep B
(SieLC), (d) Primesep 100 (SieLC), (e) Lichrosorb Diol (Phenomenex).
50
40
k΄(1,1-dimethyl biguanide)
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
CH3CN (% v/v)
Figure 3.7 Comparison of the dependency of the retention factor of 1,1-dimethylbigu-

anide on the acetonitrile content in the mobile phase among Acclaim Mixed-Mode WAX-1
(square), Acclaim Mixed-Mode WCX-1 (triangle), Primesep B (diamond), Primesep 100
(cross), and Lichrosorb Diol (circle). (Courtesy of Marketing Communication Group at
Dionex Corporation, Sunnyvale, CA.)

200
180
160
140
120
k΄(nitrate)
100
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100
CH3CN (% v/v)
Figure 3.8 Comparison of the dependency of the retention factor of nitrate ion on the acetoni-
trile content in the mobile phase among Acclaim Mixed-Mode WAX-1 (square), Acclaim Mixed-
Mode WCX-1 (triangle), Primesep B (diamond), Primesep 100 (cross), and Lichrosorb Diol
(circle).
Â� (Courtesy of Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)
conditions, RP/CEX mixed-mode columns (Mixed-Mode WCX-1, Primesep 100)

exhibit little retention, presumably due to the electrostatic repulsion between the
stationary phase and the nitrate ion.
As shown in Figure 3.8, for the cationic probe, 1,1-dimethylbiguanide, Primesep
100 (Type III RP/SCX mixed-mode column) exhibited high retention throughout the
whole solvent range indicating strong electrostatic attraction, which decreased with
an increasing acetonitrile content to a minimum at 80% acetonitrile. Meanwhile,
compared to Primesep 100, the Acclaim Mixed-Mode WCX-1, a Type IV RP/WCX
packing material, showed weaker electrostatic retention in the 5%–80% solvent
region (k′ 1.5–2.6). Above 80% acetonitrile, the retention of 1,1-dimethylbiguanide
sharply increased with a solvent content for both Primesep 100 and Acclaim Mixed-
Mode WCX-1 due to a combination of HILIC and electrostatic interaction mech-
anisms. Under high organic conditions (acetonitrile >80%), retention of the basic
probe increases with an increasing solvent content in the mobile phase, as the result
of the combined effects of electrostatic attraction and HILIC retention. By com-
parison, Acclaim Mixed-Mode WAX-1 and Primesep B showed little or no retention
from 5% to 80% solvent content due to electrostatic repulsion. However, above 80%
solvent, retention increased with solvent content, a clear indication of HILIC interac-
tion overcoming electrostatic repulsion.
The relative contributions of HILIC interaction and electrostatic interaction to
the retention of a charged analyte on an RP/IEX mixed-mode stationary phase in
the HILIC mode (in the high organic solvent mobile phase) are complex. On the
weak cation-exchange surface, the ionization of the stationary phase’s weak cation-
exchange sites and the analyte, 1,1-dimethylbiguanide, weakens with the increase
in the mobile-phase solvent content, resulting in a reduced electrostatic attraction
between the stationary phase and the analyte, or lower retention. However, since

the ionization of the mobile phase ions, ammonium and acetate, also decreases with
an increasing solvent content, this effect weakens the mobile phase ionic strength
and increases retention time. Conversely, for the weak anion-exchange surface, the
ionization of weak anion-exchange sites and 1,1-dimethylbiguanide weakens with an
increasing solvent content resulting in reduced electrostatic repulsion between the
stationary phase and the analyte, or the retention increase. Meanwhile, the ionization
of the mobile phase ions, ammonium and acetate, also decreases as solvent levels
increase, reducing the suppression of electrostatic repulsion between the stationary
phase and the basic analyte, thereby reducing retention. For an anionic analyte, the
opposite observations can be made, but with the same reasoning.
3.4.4 HILIC Behavior of RP/WAX/SCX Trimodal Mixed-Mode Columns

For a better understanding of the contribution of hydrophilic interaction and electro-
static interaction of RP/IEX mixed-mode materials in HILIC conditions, methanol
was used to replace acetonitrile. As methanol is a strong solvent for HILIC and
minimizes the hydrophilic interaction between the stationary phase and the analyte,
hydrophilic retention is considered negligible in a methanol/aqueous mobile phase.
Thus, comparing the retention–solvent dependency for uracil, Na+, and NO3− with
that generated from acetonitrile will provide a better understanding of its retention
mechanism in HILIC conditions. Liu and Pohl in their recent work investigated the
retention behavior of a commercial trimodal mixed-mode column—Acclaim Trinity
P1, using three model probes—uracil (neutral), sodium ion (cationic), and nitrate
(anionic), in both acetonitrile/ammonium acetate and methanol/ammonium acetate
buffer systems.81 Because Acclaim Trinity P1 column simultaneously provides RP,
AEX, and CEX functionalities on the same material, it is an ideal model to study
the HILIC behavior of mixed-mode stationary phases for highly hydrophilic neu-
tral, anionic, and cationic analytes. As shown in Figure 3.9, at higher solvent levels
(>80%), acetonitrile exhibited significantly higher retention for uracil and sodium
ion compared to methanol. For nitrate ion, a similar observation was noted at a
90% solvent level. Because no HILIC interaction is expected in a methanol/aque-
ous mobile phase, the observed retention increase with increasing acetonitrile in
the mobile phase is a clear indication of a HILIC effect. At 90% acetonitrile, the
presence of an ion-exchange process was also demonstrated, along with the HILIC
mechanism, noting that retention decreased for charged analytes, such as sodium
and nitrate, with increasing mobile-phase buffer concentrations.
3.5 Applications of Reversed-Phase/Ion-Exchange

Mixed-Mode Columns in the HILIC Mode
3.5.1 Pharmaceutical Active Ingredients and Counterions
Determinations of active pharmaceutical ingredients (APIs) and counterions are
important assays in pharmaceutical drug development. Due to the wide variety
of charges and hydrophobicities of these pharmaceutical-related molecules, it is
highly challenging to analyze them via a simultaneous separation of the API and its

1.00
0.80
k΄ of uracil
0.60
0.40
0.20
0
0 10 20 30 40 50 60 70 80 90 100
(a) Organic solvent (% v/v)
40
35
30
25
k΄ of Na+
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
(b) Organic solvent (% v/v)
50
45
40
35
k΄ of NO3–
30
25
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
(c) Organic solvent (% v/v)
Figure 3.9 Comparison of the dependency of the retention factor of uracil (a), Na+ ion (b),
and NO3− ion (c) on the mobile phase solvent content [acetonitrile (diamond) versus metha-
nol (square)] on the Acclaim Trinity P1—a RP/WAX/SCX trimodal column. (Courtesy of
Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)

respective counterion. The most common pharmaceutical salt forms are sodium salts
of acidic APIs and hydrochloride salts of basic APIs. Although RPLC is often used
for analyzing drug molecules with intermediate or higher hydrophobicity, it often
fails to retain highly hydrophilic analytes, such as catecholamines, organic acids,
and inorganic ions (e.g., Na+ and Cl− ions). Ion-pairing chromatography improves the
retention and selectivity of highly hydrophilic ionizable analytes, but it requires long
equilibration times and a dedicated column. Furthermore, ion-pairing mobile phases
are generally MS-incompatible. HILIC complements the aforementioned techniques
and is suitable for analyzing highly polar analytes. However, it faces challenges, such
as inadequate retentions for APIs with lower hydrophilicity and poor solubility for
some highly polar analytes.
Penicillin G is an antibiotic medicine belonging to the beta-lactam family and
is often formulated in the potassium salt form. Because of the highly hydrophilic
nature of both the API and the counterion, it is impossible to separate them within
the same analysis on any RP column. A commercial RP/WAX/SCX trimodal col-
umn (Acclaim Trinity P1) provides baseline separation of Penicillin G and K+ ion
in both the RP/IEX mode and the HILIC mode with an excellent peak shape and
adequate retention (Figure 3.10).81 The “U”-shaped retention time versus the solvent
dependency of Penicillin G with a minimum at approximately 80% acetonitrile level
indicates an anion-exchange process accompanied by an RP mechanism below 80%
acetonitrile and by a HILIC mechanism at higher solvent levels. By comparison, the
retention of K+ increases with the mobile phase acetonitrile increase from 60% to
90% solvent, suggesting a cation-exchange mechanism superimposed with a HILIC
mechanism. It should be noted that compared to the RP/IEX mode, an elution order
reversal is observed in the HILIC mode separation. Therefore, depending on the spe-
cific application requirements, a suitable separation mode can be chosen for optimal
results. In this case, 70% acetonitrile provided a fast separation with the smaller K+
peak eluting before the larger Penicillin G peak. It is always desirable to avoid elu-
tion of small peaks after large peaks, which is prone to quantification inaccuracies
due to the tailing of the larger peak.
1,1-Dimethylbiguanide hydrochloride (metformin), a highly hydrophilic basic
drug formulated in the chloride salt form, is an antidiabetic agent that reduces blood
glucose levels and improves insulin sensitivity. Figure 3.11 illustrates separations of
the drug substance and its counterion using Acclaim Trinity P1 at four acetonitrile
levels. Due to both the hydrophilic nature of the analytes and the multiple reten-
tion mechanisms facilitated by this stationary phase, separation in both the RP/IEX
mode and HILIC mode provides baseline resolution, good peak shape, and adequate
retention.80
3.5.2 Melamine and Cyanuric Acid

Melamine is an organic base and a trimer of cyanamide, with a 1,3,5-triazine skele-
ton. It is widely used in plastics, adhesives, countertops, dishware, fire retardant, etc.
Cyanuric acid is a structural analogue of melamine, and may be found as an impurity
of melamine. Cyanuric acid is an FDA-accepted component of feed-grade biuret, a
ruminant feed additive. Melamine alone is known to form bladder stones in animal

O
H
NH
S CH3
N CH3
O
O – K+
O
2000
90% CH3CN (v/v)
2
80% CH3CN (v/v) 2
1
mV
70% CH3CN (v/v) 2
60% CH3CN (v/v) 2
0 1 2 3 4
min
Figure 3.10 Analysis of Penicillin G potassium salt on a RP/WAX/SCX trimodal column.

Column, Acclaim Trinity P1, 3.0 × 50-mm format; mobile phase, acetonitrile/ammonium
acetate buffer, pH 5.2 (20â•›m mol L−1 total concentration) (90:10, 80:20, 70:30, and 60:40â•›v/v);
flow rate, 0.6â•›m L min−1; injection volume, 2.0â•›μL; temperature, 30°C; and detection, evapo-
rative light scattering detector. Sample (0.5â•›mg mL−1): (1) K+, (2) Penicillin G. (Courtesy of
Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)
tests, and when combined with cyanuric acid, it can form crystals that can give rise
to kidney stones. Structures of melamine and cyanuric acid are given in Figure 3.12.
Recent investigations of adulterated pet foods and infant formula products
demanded a reliable chromatographic method for monitoring the presence of
melamine and cyanuric acid in food products and raw materials as well as in the tis-
sues of animals suspected of ingesting melamine-tainted feeds. Both gas chromatog-
raphy (GC) and liquid chromatography (LC) methods have been developed for the
quantitative determination of melamine and cyanuric acid. The GC method requires

NH NH • HCl
CH3
H2N N N
2500 H
1 CH3
2
1
2
mV
1
t0
2
0
0 1 2 3 4 5
min
Figure 3.11 Analysis of 1,1-dimethylbiguanide hydrogen chloride on an RP/WAX/SCX

trimodal column. Column, Acclaim Trinity P1, 3.0 × 50 mm format; mobile phase, ace-
tonitrile/ammonium acetate buffer, pH 5.2 (20â•›m mol L−1 total concentration) (90:10, 70:30,
50:50, and 30:70â•›v/v); flow rate, 0.5â•›m L min−1; injection volume, 2.5â•›μL; temperature, 30°C;
and detection, ELS detector. Sample (0.2â•›mg mL−1): (1) 1,1-DimethylbiguanideH+, (2) Cl−.
(Courtesy of Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)
NH2 OH O
N N N N HN NH
H 2N N NH2 HO N OH O N O
H
Melamine Cyanuric acid
Figure 3.12 Structures of melamine and cyanuric acid.
derivatization and is thus labor intensive. LC methods, sometimes combined with

mass spectrometer (MS), are more often used. RP and cation-exchange LC columns
are commonly used for melamine analysis. However, both types of columns are
unsuitable for cyanuric acid, due to either inadequate retention of the high hydro-
philicity analyte or electrostatic repulsion between the analyte and the stationary

phase. Recently, an LC-MS method was reported for the simultaneous separation
of melamine and cyanuric acid on a commercial Type IV RP/WAX mixed-mode
column (Acclaim Mixed-Mode WAX-1, Dionex).82 Unlike RP or cation-exchange
stationary phases on which cyanuric acid often elutes in or close to the void, this RP/
WAX mixed-mode phase offers good retention with a k′ greater than 3. It is expected
that melamine cannot be adequately retained on an anion-exchanger with a posi-
tively charged surface due to electrostatic repulsion between the analyte and the
stationary phase surface. However, when operated at a high acetonitrile level (90%
v/v), the retention generated by the HILIC mechanism outstrips electrostatic repul-
sion, enhancing melamine retention for improved quantification accuracy. In addi-
tion, the RP/WAX mixed-mode column offers unique selectivity (different elution
order) compared to its RP and cation-exchange counterparts, which may be adjusted
by altering the mobile phase pH (Figure 3.13a) and the buffer concentration (ionic
strength) (Figure 3.13b)—behavior typical of ion-exchangers.
Column: Acclaim Mixed-Mode WAX-1, 5 µm

Dimensions: 2.1 × 150 mm
Melamine-1.62 A = CH3CN,
Mobile phase:
B = 55 mM NH4OAc buffer
pH 4 Condition 1 (pink traces): A:B (pH 6) = 90:10
Condition 2 (blue traces): A:B (pH 5) = 90:10
Melamine-1.47 Condition 3 (purple traces): A:B (pH 4) = 90:10
Flow rate: 0. 50 mL min–1
pH 5
Column temperature: 30°C
Inj. volume: 10 µL
Melamine-1.47
Detection: MSQ plus™
pH 6 Melamine: pos. ESI SIM m/z = 127
Cyanuric acid: neg. ESI SIM m/z = 128
0 1 2 3 4 5 6 7
Cyanuric acid-3.34
Cyanuric acid-5.40
pH 4
pH 5
Cyanuric acid-5.50
pH 6
0 1 2 3 4 5 6 7
(a) Retention time (min)
Figure 3.13 LC-MS analysis of melamine and cyanuric acid on Acclaim Mixed-Mode
WAX-1—a Type IV RP/WAX mixed-mode column. Effects of mobile phase pH (a) and ionic
strength (b). (Reproduced from Wang, L. et al., Simultaneous Determination of Melamine
and Cyanuric Acid Using LC-MS with the Acclaim Mixed-Mode WAX-1 Column and
Mass Spectrometric Detection, http://www.dionex.com/en-us/webdocs/62283_LPN%
201991–01%20N_Melamine.pdf. With permission; Courtesy of Marketing Communication
Group at Dionex Corporation, Sunnyvale, CA.)
(continued)

Column: Acclaim Mixed-Mode WAX-1,5 µm

6 × 105 Melamine Dimensions: 2.1 × 150 mm
Mobile phase: A = CH3CN, B = H2O,
C = 55 mM NH4OAc buffer pH 5.2
Condition 1 (pink traces): A:B:C (11 mM) = 80:00:20
Intensity (counts)
Condition 2 (blue traces): A:B:C (8.25 mM) = 80:05:15

Condition 3 (purple traces): A:B:C (5.5 mM) = 80:10:10
Condition 4 (green traces): A:B:C (2.75 mM) = 80:15:05
Flow rate: 0.25 mL min–1
Inj. volume: 10 µL
Detection: MSQ plus
Melamine: pos. ESI SIM m/z = 127
Cyanuric acid: neg. ESI SIM m/z = 128
0 3 6 9 12 15
2 × 105 12.42
Cyanuric acid
Intensity (counts)
9.68
(NH4OAc) = 2.75 mM 8.27

(NH4OAc) = 5.5 mM 7.40
(NH4OAc) = 8.25 mM
(NH4OAc) = 11 mM
0 3 6 9 12 15
(b) Retention time (min)
Figure 3.13 (continued)
3.5.3 Buffer Salts for Biopharmaceutical Applications

Good’s buffers are a group of buffers first described in the research of Dr. Norman
Good et al. in 1966.83 These buffers were selected because they display character-
istics that were valuable to research in biology and biochemistry. The characteristics
associated with these buffers include a pKa value between 6.0 and 8.0, high solu-
bility, low toxicity, limited effect on biochemical reactions, very low absorbance
between 240 and 700â•›nm, enzymatic and hydrolytic stability, minimal pH changes
due to temperature and concentration, limited effects due to ionic strength or salt
composition of the solution, limited interaction with mineral cations, and limited
biological membrane permeability.83
Cell culture media is a complex blend containing amino acids, vitamins, car-
bohydrates, and salts with a biological buffer. All these components need to be
assayed in order to ensure proper media formulation. Most of the buffers used
are either inorganic buffers or Good’s buffers. Tests for inorganic buffers by
ICP or Ion Chromatography already exist. The conventional method for deter-
mination of Good’s buffers is titration, which is inapplicable for a complex bio-
logical matrix. Due to the hydrophilic nature of Good’s buffer salts, RPLC is
not an appropriate approach for this separation. Figure 3.14 illustrates struc-
tures of seven commonly used Good’s buffer salts, and Figure 3.15 demonstrates

H H O O H O
OH OH OH OH
N O N S S N S
S N
O OH O O O
O
CHES CAPS MES CAPSO
HO HO O
H H H3C
O O
N N S OH HO
HO S HO N S
H3C OH
OH O H OH O
O
OH OH
TES TAPS AMPSO
Figure 3.14 Structures of selected Good’s buffer salts. (1) 2-(Cyclohexylamino)

ethanesulfonic acid (CHES), (2) 3-(cyclohexylamino)-1-propanesulfonic acid (CAPS),
(3) 3-(cyclohexylamino)-2-hydroxy-1-propanesulfonic acid (CAPSO), (4) N-(1,1-dimethyl-
2-hydroxyethyl)-3-amino-2-hydroxypropanesulfonic acid (AMPSO), (5) 2-(N-morpholino)
ethanesulfonic acid (MES), (6) N-[Tris(hydroxymethyl)methyl]-2-aminoethanesulfonic acid
(TES), and (7) N-[Tris(hydroxymethyl)methyl]-3-aminopropanesulfonic acid (TAPS).
200
5
mV
6 7
0 2 4 6 8
min
Figure 3.15 Separation of Good’s buffer salts on the RP/WAX/SCX trimodal column.
Column, Acclaim Trinity P1, 3.0 × 100 mm format; mobile phase, acetonitrile/ammonium
acetate buffer, pH5.2 (12â•›m mol L−1 total concentration) (85:15â•›v/v); flow rate, 0.6â•›m L min−1;
injection volume, 2.5â•›μL; temperature, 30°C; and detection, ELS detector. Sample (0.15â•›mg
mL−1): (1) CHES, (2) CAPS, (3) CAPSO, (4) AMPSO, (5) MES, (6) TES, and (7) TAPS.
(Courtesy of Marketing Communication Group at Dionex Corporation, Sunnyvale, CA.)

the separation of these salts on a commercial RP/WAX/SCX trimodal column

(Acclaim Trinity P1 in 3 × 50â•›m m format) under the HILIC mode (85% acetoni-
trile/15% ammonium acetate buffer, v/v). This separation can be further “fine-
tuned” by adjusting the buffer concentration as well as the organic solvent level
in the mobile phase.81
3.5.4 Determination of Ethanol Consumption Markers

Confirmation of ethanol intake is of great importance in forensic toxicology, psy-
chiatry, and workplace monitoring. A number of different markers, from short
term to long term, are available, each of which with a specific time window of
detection. Among all, the ethanol phase II metabolites ethyl glucuronide (EtG)
and ethyl sulfate (EtS) have attracted particular attention lately because both com-
pounds are sensitive, specific, and stable intermediate-term markers of acute etha-
nol consumption in urine and other matrices. Bicker et al. developed an LC–MS/
MS method using an in-house developed Type IV RP/WAX mixed-mode column
for the simultaneous analysis of EtG and EtS as well as ethyl phosphate (EtP) (see
structures in Figure 3.16),56 a potential new marker with distinct nonspecific or
nonenzymatic formation in human urine. A reliable quantification of these highly
polar acidic metabolites in the polar urinary matrix poses a unique challenge, par-
ticularly considering the direct injection methodology chosen for diluted urine.
While an optimized RP/WAX separation was developed in which the two singly
charged analytes EtG and EtS eluted before the doubly charged EtP (which was
eluted in the back-flush mode to speed up the analysis), as expected from the pri-
mary anion-exchange process, the elution order of EtG and EtS, in contrast, fol-
lows the order of increasing hydrophilicity. Moreover, the retention properties of
EtG and EtS as a function of acetonitrile content and the pH of the mobile phase
showed that as the acetonitrile fraction increases, the retention factors increase and
the more hydrophilic EtG responds more strongly to the change in solvent content
than the less hydrophilic EtS (Figure 3.17). These results strongly suggest that in
an acetonitrile-rich mobile phase, a HILIC mechanism is superimposed upon the
primary anion-exchange process.
OH CH3
O
O O HO O
H O O S P
H
HO O HO O
OH H
OH OH CH3 CH3
OH OH
EtG EtS EtP
Figure 3.16 Structures of ethyl glucuronide (EtG), ethyl sulfate (EtS), and ethyl phos-
phate (EtP).

60 EtG
Retention factor
40
20
EtS
0
80
7.5 75
70
7.0 65
pH 6.5 60 CN
a
55 H3
6.0 50 %C
Figure 3.17 Dependency of the retention factor of EtG and EtS on the acetonitrile content
and pH of the mobile phase using a Type IV RP/WAX stationary phase. (Reproduced from
Bicker, W. et al., Anal. Chem., 78, 5884, 2006. Copyright American Chemical Society. With
permission)
3.6 Concluding Remarks

Reversed-phase/ion-exchange mixed-mode stationary phases promise great versatil-
ity and capability for retaining and separating a variety of charged polar compounds
in addition to more apolar ionic and nonionic analytes. Compared to Type I and
Type II materials, Type III and Type IV bimodal (reversed-phase/anion-exchange
or reversed-phase/cation-exchange) stationary phases represent the state-of-the-art
in mixed-mode column technology because of a better-defined chemistry and a
more rugged selectivity of the resulting products. A reversed-phase/anion-exchange/
cation-exchange trimode column based on NSH technology, a recent development
in mixed-mode chromatography column technology, further extends the application
range of mixed-mode phases to an even broader base.
The combination of a hydrophobic moiety with ion-exchange functionality gener-
ates hydrophobic, electrostatic, as well as hydrophilic interactive domains. Because
of their unique column chemistries, these mixed-mode materials can be operated in
a variety of chromatographic modes, including the RP mode (e.g., for neutral mole-
cules), the IEX mode (for solutes bearing the opposite charge), the ion-exclusion chro-
matography mode (for solutes having the same charge), and the HILIC mode (polar
neutral, basic, amphoteric, and acidic compounds), depending on the chromato-
graphic conditions and the characteristics of the analytes. These stationary phases
are especially useful for retaining highly hydrophilic charged analytes, which would
otherwise be difficult to retain with an RP column. Furthermore, the increasing need

for complementary analysis methods for validation of the RPLC assay specificity
required by regulatory authorities in pharmaceutical impurity profiling and stability
indicating assays has driven the need for new stationary phases with complementary
selectivity. Moreover, to analyze complex samples successfully, multidimensional
methods are often needed for greater method development flexibility. Consequently,
facile adjustment of selectivity as a result of multi-mode retention mechanisms is
highly beneficial. It might be interesting to note that for those who do not have
access to MS/MS systems, retention behavior as a function of mobile phase changes
can be used to help confirm peak identity from complex samples.
Due to the presence of polar ion-exchange functionality, modern reversed-phase/
ion-exchange mixed-mode columns are capable of operating in high solvent (aceto-
nitrile) conditions (in HILIC mode). Compared to their Type III counterparts, Type
IV bimodal mixed-mode columns are better suited to HILIC applications. Although
more studies are needed to thoroughly understand the retention mechanism of
reversed-phase/ion-exchange mixed-mode columns in HILIC conditions, the exist-
ing evidence supports a retention mechanism consisting of an electrostatic interac-
tion superimposed on a HILIC mechanism for highly hydrophilic ionic analytes and
a purely HILIC retention for highly hydrophilic neutral molecules. Owing to the
presence of a hydrophobic moiety, the polarity of the reversed-phase/ion-exchange
surface is of intermediate reversed-phase interaction strength. However, substantial
retention for charged analytes can be realized via ion-exchange interaction. As the
result, the retention of charged analytes can be tuned by adjusting the mobile phase
pH and ionic strength relative to neutral analytes whose retention is governed solely
by the mobile phase solvent content. Such flexible selectivity can be achieved not
only in reversed-phase conditions, but also in the HILIC mode. HILIC applications
of mixed-mode columns offer high-sensitivity LC-MS capability with greater flex-
ibility in method development.
Although not fully explored, it is evident that reversed-phase/ion-exchange
mixed-mode columns have great potential for a variety of HPLC applications, under
both reversed-phase and HILIC conditions. These stationary phases bridge the gaps
between traditional reversed-phase, ion-pairing, ion-exchange, and HILIC chroma-
tography. From the chromatographers’ perspective, the unique chromatographic
features of such materials stimulate innovative application method development.
Besides the applications described in this chapter, many other applications will
benefit from what reversed-phase/ion-exchange mixed-mode columns have to offer,
including in-line SPE, multi-dimensional-LC, preparative-LC, etc. We expect that
these interesting stationary phases will find their way to an increasingly broad range
of applications in the coming years.
Acknowledgments
The authors are thankful to Dr. Jeff Rohrer (Dionex Corporation) for the technical
review and edits, Dr. Wolfgang Bicker and Dr. Michael Lämmerhofer (both from
the University of Vienna) for providing Figures 3.4 and 3.5, respectively, and Scott
Lefferts and Lynn Sison for preparing all the figures in this chapter.

References
1. Linden, J. C. and Lawhead, C. L., J. Chromatogr., 1975, 105, 125.
2. Alpert, A. J., J. Chromatogr., 1990, 499, 177.
3. Hemström, P. and Irgum, K., J. Sep. Sci., 2006, 29, 1784.
4. Ikegami, T., Tomomatsu, K., Takubo, H., Horie, K. and Tanaka, N., J. Chromatogr. A,
2008, 1184, 474.
5. Zhu, B. Y., Mant, C. T. and Hodges, R. S., J. Chromatogr., 1991, 548, 13.
6. Yoshida, T., Anal. Chem., 1997, 69, 3038.
7. Churms, S. C. In: Journal of Chromatography Library, El Rassi, Z., Ed.; Vol. 58. New
York: Elsevier; 1995, Chap. 3.
8. Alpert, A. J., Shukla, M., Shukla, A. K., Zieske, L. R., Yuen, S. W., Ferguson, M. A. J.,
Mehlert, A., Pauly, M. and Orlando, R., J. Chromatogr. A, 1994, 676, 191.
9. Churms, S. C., J. Chromatogr. A, 1996, 720, 75.
10. Yoshida, T. J. Chromatogr. A, 1998, 808, 105.
11. Yoshida, T. J. Biochem. Biophys. Methods, 2004, 60, 265.
12. Zhang, J. and Wang, D., J. Chromatogr. B, 1998, 712, 73.
13. Furuya, K., Schegg, K. M., Wang, H., King, D. S. and Schooley, D. A., Proc. Natl Acad.
Sci. USA, 1995, 92, 12323.
14. Boutin, J. A., Ernould, A.-P., Ferry, G., Genton, A. and Alpert, A. J., J. Chromatogr.,
1992, 583, 137.
15. Faull, K. F., Feistner, G. J., Conklin, K., Roepstorff, P. and Andrews, P. C., Neuropeptides,
1998, 32, 339.
16. Jenö, P., Scherer, P. E., Manning-Krieg, U. and Horst, M., Anal. Biochem., 1993, 215,
292.
17. Schmerr, M. J., Cutlip, R. C. and Jenny, A., J. Chromatogr. A, 1998, 802, 135.
18. Schmerr, M. J. and Alpert, A. In: Prions and Mad Cow Disease, Nunnally, B. K. and
Krull, I. S., Eds.; Marcel Dekker: New York, 2004, pp. 359–377.
19. Carroll, J., Fearnley, I. M. and Walker, J. E., Proc. Natl Acad. Sci. USA, 2006, 103,
16170.
20. Strege, M. A., Anal. Chem., 1998, 70, 2439.
21. Strege, M. A., Am. Pharm. Rev., 1999, 2(3), 53.
22. Olsen, B. A., J. Chromatogr. A, 2001, 913, 113.
23. Wang, X., Li, W. and Rasmussen, H. T., J. Chromatogr. A, 2005, 1083, 58.
24. Ali, M. S., Ghori, M., Rafiuddin, S. and Khatri, A. R., J. Pharm. Biomed. Anal., 2007,
43, 158.
25. Hmelnickis, J., Pugovičs, O., Kažoka, H., Viksna, A., Susinskis, I. and Kokums, K.,
J. Pharm. Biomed. Anal., 2008, 48, 649.
26. Dallet, Ph., Labat, L., Kummer, E. and Dubost, J. P., J. Chromatogr. B, 2000, 742, 447.
27. Doi, T., Kajimura, K., Takatori, S., Fukui, N., Taguchi, S. and Iwagami, S., J. Chromatogr.
B, 2009, 877, 1005.
28. Mulder, P. P. J., Beumer, B. and Van Rhijn, J. A., Anal. Chim. Acta., 2007, 586, 366.
29. Wang, Y., Wang, T., Shi, X., Wan, D., Zhang, P., He, X., Gao, P., Yang, S., Gu, J. and Xu,
G., J. Pharm. Biomed. Anal., 2008, 47, 870.
30. Guo, Y. and Huang, A., J. Pharm. Biomed. Anal., 2003, 31, 1191.
31. Tai, A. and Gohda, E., J. Chromatogr. B, 2007, 853, 214.
32. Nováková, L., Solich, P. and Solichová, D., Trends Analyt. Chem., 2008, 27, 942.
33. Garbis, S. D., Melse-Boonstra, A., West, C. E. and van Breemen, R. B., Anal. Chem.,
2001, 73, 5358.
34. Moyer, T. P., Charlson, J. R., Enger, R. J., Dale, L. C., Ebbert, J. O., Schroeder, D. R. and
Hurt, R. D., Clin. Chem., 2002, 48, 1460.
35. Soltysik, S., Bedore, D. A. and Kensil, C. R., Ann. N.Y. Acad. Sci., 1993, 690, 392.

36. McGrane, M., Keukens, H. J., O’Keeffe, M., van Rhijn, J. A. and Smyth, M. R. In:
EuroResidue IV Veldhoven, The Netherlands; van Ginkel, L. A., Ruiter, A., Eds., 2000,
pp. 765–770.
37. Troyer, J. K., Stephenson, K. K. and Fahey, J. W., J. Chromatogr. A, 2001, 919, 299.
38. Le Rouzo, G., Lamouroux, C., Bresson, C., Guichard, A., Moisy, P. and Moutiers, G.,
J. Chromatogr. A, 2007, 1164, 139.
39. Mawhinney, D. B., Hamelin, E. I., Fraser, R., Silva, S. S., Pavlopoulos, A. J. and
Kobelski, R. J., J. Chromatogr. B, 2007, 852, 235.
40. Fontanals, N., Marce, R. M. and Borrull, F., J. Chromatogr. A, 2007, 1152, 14.
41. Walshe, M., Kelly, M. T., Smyth, M. R. and Ritchie, H., J. Chromatogr. A, 1995, 708, 31.
42. Bergqvist, Y. and Hopstadiua, C., J. Chromatogr. B, 2000, 714, 189.
43. Crowther, J. B., Fazio, S. D. and Hartwick, R. A., J. Chromatogr., 1983, 282, 619.
44. Bischoff, R. and McLaughlin, L. W., J. Chromatogr., 1983, 270, 117.
47. Muenter, M. M., Stokes, K. C., Obie, R. T. and Jezorek, J. R., J. Chromatogr. A, 1999,
844, 39.
48. McCallum, J. L., Yang, R., Young, J. C., Strommer, J. N. and Tsao, R., J. Chromatogr. A,
2007, 1148, 38.
49. Hsieh, Y., Duncan, C. and Liu, M., J. Chromatogr. B, 2007, 854, 8.
50. Li, J., Shao, S., Jaworsky, M. S. and Kurtulik, P. T., J. Chromatogr. A, 2008, 1185, 185.
51. Liu, X., Fang, J., Cauchon, N. and Zhou, P., J. Pharm. Biomed. Anal., 2008, 46, 639.
52. Abbood, A., Smadja, C., Herrenknecht, C., Alahmad, Y., Tchapla, A. and Taverna, M.,
J. Chromatogr. A, 2009, 1216, 3244.
53. Nogueira, R., Lämmerhofer, M. and Lindner, W., J. Chromatogr. A, 2005, 1089, 158.
54. Bicker, W., Lämmerhofer, M. and Lindner, W., J. Chromatogr. B, 2005, 822, 160.
55. Nogueira, R., Lubda, D., Leitner, A., Bicker, W., Maier, N. M., Lämmerhofer, M. and
Lindner, W., J. Sep. Sci., 2006, 29, 966.
56. Bicker, W., Lämmerhofer, M., Keller, T., Schuhmacher, R., Krska, R. and Lindner, W.,
Anal. Chem., 2006, 78, 5884.
57. Bicker, W., Lämmerhofer, M. and Lindner, W., Anal. Bioanal. Chem., 2008, 390, 263.
58. Lämmerhofer, M., Richter, M., Wu, J., Nogueira, R., Bicker, W. and Lindner, W., J. Sep.
Sci., 2008, 31, 2572.
59. Apfelthaler, E., Bicker, W., Lämmerhofer, M., Sulyok, M., Krska, R., Lindner, W. and
Schuhmacher, R., J. Chromatogr. A, 2008, 1191, 171.
60. Liu, X., Pohl, C. and Weiss, J., J. Chromatogr. A, 2006, 1118, 29.
61. Liu, X. and Pohl, C., Am. Lab., 2007, 39, 22.
62. Liu, X. and Pohl, C., Am. Lab., 2009, 41, 26.
63. Thermo-Fisher website on Duet columns: http://www.selectscience.net/products/
hypersil-duet-c18+sax-columns/?prodID=11596 (accessed on December 9, 2010) and
http://www.selectscience.com/products/hypersil-duet-c18+scx-columns/?prodID=7515
(accessed on December 9, 2010)
64. Grace website on mixed-mode columns: http://www.discoverysciences.com/Â�
uploadedFiles/Site_for_Catalog_2008/hplcic/HPLC_Columns/Small_Molecule/
alltech_specialty.pdf (Accessed on December 9, 2010)
65. SIELC website on Primesep columns: http://www.sielc.com/Technology_
NovelStationaryPhases.html (accessed on December 9, 2010)
66. Dionex website on Acclaim Mixed-Mode WAX-1: http://www.dionex.com/en-us/
webdocs/48829-DS_Acclaim_Mixed-Mode_WAX1_released021407.pdf (Accessed on
December 9, 2010)
67. Dionex website on Acclaim Mixed-Mode WCX-1: http://www.dionex.com/en-us/
columns-accessories/specialty/cons65659.html (accessed on December 9, 2010)

68. Dionex website on Acclaim Trinity P1: http://www.dionex.com/en-us/webdocs/70761-

DS-Acclaim-Trinity-12Feb2010-LPN2239-02.pdf (Accessed on December 9, 2010)
69. Dionex website on OmniPac: http://www.dionex.com/en-us/products/columns/lc/
mixed-mode/omnipac/lp-71739.html (Accessed on December 9, 2010)
70. Kopaciewicz, W., Rounds, M. A., Fausnaugh, F. and Regnier, F. E., J. Chromatogr.,
1983, 266, 3.
71. Sellergren, B. and Shea, K. J., J. Chromatogr. A, 1993, 654, 17.
72. Millot, M., Debranche, T., Pantazaki, A., Gherghi, I., Sebille, B. and Vidal-Madjar, C.,
Chromatographia, 2003, 58, 365.
73. Hansen, M. B., Lihme, A., Spitali, M. and King, D., Bioseparation, 1999, 8, 189.
74. Gao, D., Lin, D.-Q. and Yao, S.-J., J. Chromatogr. B, 2007, 859, 16.
75. Rios, M., Pharm. Technol., 2007, 31(5), 40, 42, 44, 46, 48.
76. Mant, C. T., Kondejewski, L. H. and Hodges, R. S., J. Chromatogr. A, 1998, 816, 79.
77. Mant, C. T., Litowski, J. R. and Hodges, R. S., J. Chromatogr. A, 1998, 816, 65.
78. Zhu, B. Y., Mant, C. T. and Hodges, R. S., Pept. Chem. Biol., Proc. Am. Pept. Symp.,
1992, 12, 546.
79. Mashige, F., Ohkubo, A., Matshushima, Y., Takano, M., Tsuchiya, E., Kanazawa, H.,
Nagata, Y., Takai, N., Shinozuka, N. and Sakuma, I., J. Chromatogr. B, 1994, 658, 63.
80. Liu, X., Pohl, C. and Woodruff, A. (manuscript in preparation).
81. Liu, X. and Pohl, C. J. Sep. Sci., 2010, 33, 779.
82. Wang, L., Henday, S. M., Liu, X., Tracy, M. and Schnute, W. C., Simultaneous
Determination of Melamine and Cyanuric Acid Using LC-MS with the Acclaim Mixed-
Mode WAX-1 Column and Mass Spectrometric Detection (http://www.dionex.com/
en-us/webdocs/62283_LPN%201991–01%20N_Melamine.pdf).
83. Good, N. E. et al., Biochemistry, 1966, 5(2), 467.

4 Mechanisms Controlling
the Separation
and Retention of
Proanthocyanidins and
Other Plant Polyphenols
in Hydrophilic Interaction
Chromatography
Akio Yanagida and Yoichi Shibusawa
Contents
4.1 Introduction..................................................................................................... 77
4.2 Polyphenolic Compounds................................................................................ 78
4.3 Chromatography Challenges for Separation of Various PAs.......................... 79
4.4 HILIC Conditions for Separation of PAs.........................................................80
4.5 HILIC Separation Profiles of Apple PA Mixtures Eluted
Using a Gradient Method................................................................................. 81
4.6 Retention Behavior of Oligomeric PA Standards in HILIC............................ 81
4.7 Retention Mechanism of Oligomeric PAs in HILIC.......................................84
4.8 HILIC Separation of Other Polyphenolic Compounds................................... 86
References................................................................................................................. 88
4.1 Introduction
Since the term “hydrophilic interaction chromatography (HILIC)” was proposed by
Alpert in 1990,1 a large number of highly polar (or hydrophilic) substances have been
separated efficiently using the HILIC approach. HILIC requires an aqueous organic
mobile phase and a polar stationary phase, such as bare silica or silica modified with
ionic, neutral, or complex functional groups. HILIC is a variant of normal-phase
liquid chromatography and is suitable for the separation of water-soluble polar sub-
stances that are poorly retained on conventional reversed-phase stationary phases,
such as ODS-silica. Most recent HILIC applications have focused on the analyses of
77

biologically active polar compounds such as saccharides (in glycoproteins), amino

acids, peptides, nucleotides, and their pharmaceutical derivatives. HILIC is a pow-
erful method for separating compounds that contain multiple hydrogen-bonding
groups. Hydrogen bonding between the solute and the polar stationary phase is the
essential mechanism for differential retention in most HILIC separations. Plant sec-
ondary metabolites, and in particular polyphenolic compounds and their glycosides,
are excellent candidates for HILIC due to the presence of multiple hydroxyl groups
in their structures. In this chapter, we review our recent experimental data obtained
from investigating the mechanisms underlying the retention and separation of plant
proanthocyanidins (oligomeric catechins) and other polyphenols by HILIC.
4.2 Polyphenolic Compounds

Polyphenols are a group of secondary metabolites widely distributed throughout the
plant kingdom. Almost all foods derived from edible plants contain polyphenolic com-
pounds. Plant polyphenols are well-known natural antioxidants, and recent studies have
suggested that the polyphenols in food play an important role as functional food factors
in the prevention of human diseases and cancers. The molecular mass of polyphenols
generally ranges from 100 to 10,000â•›Da. In the plant cell, the majority of polyphenols are
below 500â•›Da and exists in the form of glycosides and/or organic acid esters, rather than
as free aglycones. The separation and identification of these low-molecular-mass poly-
phenols are easily performed by conventional reversed-phase HPLC (high-Â�performance
liquid chromatography), because the aglycones usually contain hydrophobic aromatic
groups. On the other hand, the majority of high-molecular-mass polyphenols (over
500â•›Da) are present as condensed or hydrolysable tannins. Condensed tannins are
oligomeric flavan-3-ols, so-called proanthocyanidins (PAs), while the hydrolysable
tannins are polygalloyl and related hexahydroxydiphenoyl esters. In general, the high-
molecular-mass polyphenols are highly hydrophilic due to their unique 3D structures
associated with the degree of polymerization (DP). These compounds are not only the
water-soluble antioxidants but also strong inhibitors of numerous physiological enzy-
matic reactions because of their protein binding ability (often nonspecific). However, the
separation and identification of these high-molecular-mass polyphenols by conventional
reversed-phase HPLC are very difficult because of the heterogeneity of the monomeric
units and the different polymerization linkages in their structures.
Among the high-molecular-mass polyphenols, PAs are the common macromol-
ecules biosynthesized by enzymes in the plant cell. The monomeric units in the PA
structure are neutral flavan-3-ols (catechins). Because the general structure of flavan-
3-ol has two asymmetrical centers at C-2 and C-3 on the heterocyclic ring, two pairs of
diastereoisomers are theoretically present (see Figure 4.1). In addition to these isomers,
analogous compounds with a galloyl ester (gallate) at the C-3 hydroxyl group of each
flavan-3-ol isomer have also been identified. In most plants, these isomeric flavan-3-ols
and their gallates constitute the monomeric units of PAs. Plant PAs are a mixture of
oligomers consisting of different chain lengths of flavan-3-ol units linked most com-
monly through C-4 to C-8 (or C-6) interflavan bonds (type-B PA). For example, the
structures of six types of monomeric flavan-3-ols (1a–1f) and nine types of oligomeric
PAs (2a–5a) are shown in Figure 4.1. The numeric number assigned to each compound

Mechanisms Controlling the Separation and Retention 79
OH R1
OH
8 1a H : (+)-catechin C
HO O 2
R1
3 1b OH : (+)-gallocatechin GC
6
4 OH
OH
R1 R2
OH 1c H H : (–)-epicatechin EC
OH
8
1d OH H : (–)-epigallocatechin
HO O 2
R1 1e H galloyl : (–)-epicatechin gallate
3
6 O
4 R2 1f OH galloyl : (–)-epigallocatechin gallate
OH
2a EC C 3a EC EC EC
(4β 8) 2b EC EC 3b EC EC EC
(4β 6) 2c C C 3c EC EC EC
(4α 8) 2d GC EC 4a EC EC EC EC
5a EC EC EC EC EC
Figure 4.1 Structures of monomeric flavan-3-ols and oligomeric PAs used as experimen-
tal standards. (Reprinted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153, 2007. With
permission
Â� from Elsevier.)
is identical to its DP. The oligomeric PAs from dimer (except 2c and 2d) to pentamer
shown in Figure 4.1 were isolated from an immature apple, as described elsewhere.2–5
Furthermore, the MALDI-TOF MS study in our laboratory confirmed that the PA-rich
fraction obtained from immature apples contained oligomeric PAs ranging from
dimers (2-mer) to pentadecamers (15-mer).2 A number of PAs with similar monomeric
units but different DPs have also been identified from various plant sources, such as
fruit, seeds, bark, and leaves, by other MALDI or ESI MS analyses.
4.3 C
hromatography Challenges for Separation
of Various PAs
As mentioned above, reversed-phase LC is not the best separation method for plant
PAs. HPLC using an ODS column eluted with an aqueous acetonitrile (or metha-
nol) mobile phase has been the general method used for the separation of flavan-
3-ols and small DP oligomers such as dimers and trimers.3,6–8 However, the elution
order of these monomers and small DP oligomers was not based on their DPs, and
the broad unresolved peaks derived from coexisting highly oligomeric PAs over-
lapped the separation profiles of these monomers and small DP oligomers.3 For a
more efficient separation of highly oligomeric PAs, several useful methods based
on other LC modes have recently been reported. Size-exclusion chromatography
(SEC) using dimethylformamide9–12 or aqueous-organic mobile phases,3,13,14 and
high-speed counter-current chromatography (HSCCC) using immiscible two-phase

solvent systems15–17 enable the partial separation of oligomeric PAs based on their DP.
Furthermore, normal-phase HPLC using the combination of a bare silica column and
a nonaqueous organic mobile phase resulted in a more efficient DP-based separation
compared with the SEC and HSCCC methods. Using an optimized HPLC method
with a normal-phase silica column, oligomeric PAs from different plant sources were
clearly separated up to decamers with a gradient elution of dichloromethane–metha-
nol (containing a small volume of acidic water)18–22 and PAs from an apple were also
separated up to pentamers with a gradient elution of hexane–acetone.4,5
However, the above nonaqueous organic mobile phases of silica HPLC are
associated with some practical drawbacks. Mobile phases containing dichloromethane
must be handled carefully and need to be disposed of appropriately (dichlorometh-
ane is an organic solvent regulated and monitored by the Water Pollution Prevention
Law in Japan). In the case of hexane–acetone mobile phases, highly oligomeric
PAs (larger than pentamers) are barely eluted from the silica column due to their
low solubility and strong adsorption to the silica stationary phase. Safer and water-Â�
compatible mobile phases are therefore necessary for these silica HPLC methods.
In addition to these practical drawbacks, the mechanisms underlying the separation
and retention of oligomeric PAs in this mode have not been adequately elucidated.
We therefore decided to investigate HILIC using more aqueous conditions for the
separation of oligomeric PAs.23
4.4 HILIC Conditions for Separation of PAs

A PA-rich mixture from immature apples (apple PA mixture) was prepared, and
consisted of monomeric flavan-3-ols (1a and 1c in Figure 4.1) and oligomeric PAs
ranging in size from dimers to pentadecamers. The contents of this mixture were
divided into two fractions by methyl acetate extraction.4 The majority of monomers
and small DP oligomers (below hexamers) were selectively extracted with methyl
acetate, while highly oligomerized PAs remained in the residue. The methyl acetate
extracts and the residue fraction were each subjected to HILIC separation, and in
addition, the 15 different standards of flavan-3-ols and oligomeric PAs shown in
Figure 4.1 were also analyzed by HILIC.
HILIC separation of these polyphenolic samples was conducted on a TSK-Gel
Amide-80 column (250â•›mm × 4.6â•›mm ID, 5â•›μm particle size, Tosoh, Japan).24 This
silica-based stationary phase was chosen for its acrylamide nonionic (neutral) car-
bamoyl termini, since the solute PAs are neutral polyphenols and lack ionic groups.
The HPLC system consisted of a Model L-7100 pump (Hitachi, Tokyo, Japan), a
Rheodyne 7166 injector, and a Model L-7455 diode-array detector (Hitachi).
Chromatograms were recorded on a PC using Model D-7000 chromatography data
station software (Hitachi).
HILIC separation of oligomeric PAs in the methyl acetate extracts and residue
fraction of the apple PA mixture were carried out using a linear gradient. The mobile
phases were acetonitrile–water mixtures at volume ratios of (a) 90:10 and (b) 50:50,
respectively. The Amide-80 column was initially conditioned with solvent A at a
flow rate of 1.0â•›m L/min. Lyophilized methyl acetate extracts or apple PA residue
mixture were dissolved in solvent A (final conc.: 10â•›mg/mL) and a portion (20â•›μL)

was injected and eluted with a linear gradient from 100% A (0% B) to 0% A (100% B)
over 60â•›min at 1.0â•›m L/min at ambient temperature. The absorbance of the effluent
was monitored using a diode-array detector.
The retention characteristics on the Amide-80 column of the 15 standards of flavan-
3-ols (1a–f) and PA oligomers (2a–d, 3a–c, 4a, and 5a) as shown in Figure 4.1 were
also examined by the isocratic elution of acetonitrile–water (84:16, v/v). A 10 μL ali-
quot of each sample (100â•›μg/mL in the mobile phase) or the mixed sample solution
(100â•›μg each/mL) was injected and eluted at a flow rate of 1.0â•›mL/min. The retention
factor (k) of each compound was calculated using the following equation:
(t R − t 0 )
k=
t0
where
t R is the retention time (min) of the solute
t0 is the elution time of the non-retained hydrophobic compound (anthracene)
through the column
4.5 HILIC Separation Profiles of Apple PA Mixtures

Eluted Using a Gradient Method
HILIC chromatograms of the apple PA mixture of methyl acetate extracts and resi-
due fraction are shown in Figure 4.2A and B, respectively. The major peaks in both
chromatograms are numbered according to their elution order. The separation profile
of the apple PA mixture of methyl acetate extracts (Figure 4.2A) was very similar
to that obtained previously by silica HPLC using an organic hexane–acetone mobile
phase.4 The DPs of each oligomeric PA in the chromatogram were characterized
by off-line MALDI-TOF-MS analysis and revealed that the oligomeric PAs in the
extract eluted in the order of monomer to pentamer, in accordance with the increase
in the number of their DPs (data not shown). Compared with Figure 4.2A of methyl
acetate extracts, the separation profile of the residue from the apple PA mixture
(Figure 4.2B) showed more separated peaks (up to 10). An LC-ESI-MS study using
a semimicro Amide-80 column (2.0â•›mm ID) confirmed the DP-based separation of
oligomeric PAs by HILIC, i.e., the number of separated peaks (up to 10) in Figure
4.2B directly corresponds to the DP of each oligomer (up to decamers).23
4.6 Retention Behavior of Oligomeric

PA Standards in HILIC
Since the first report by Regaud et al. in 1993,18 several PA separation methods using
silica HPLC with organic mobile phases have been published. Although the separa-
tion methods were described as “normal-phase HPLC” in these reports, the sepa-
ration and retention mechanisms of oligomeric PAs have not been elucidated. To
clarify the precise mechanism of PA separation in these LC modes, HILIC studies
of the 15 standards shown in Figure 4.1 were carried out using isocratic elution

2.0
1 2
1.5 3
1.0 * 4
*
0.5 5
Abs. at 230 nm
(A)
0.5
1 2 3 4
5
0.4 6
7
8
0.3 * 9
* 10
0.2
0.1
0 10 20 30 40 50 60
(B) Retention time (min)
Figure 4.2 HILIC separation profiles of an apple PA mixture. (A) Chromatogram of

oligomeric PAs in the methyl acetate extract. (B) Chromatogram of oligomeric PAs in the
extraction residue. Individual peak in both chromatograms is numbered in accordance with
their elution order (up to 10). The two peaks marked (*) are from impurities (caffeoylquinic
acid and phloretin glycoside). (Reprinted from Yanagida, A. et al., J. Chromatogr. A, 1143,
153, 2007. With permission from Elsevier.)
(acetonitrile–water 84:16), and the retention factor (k) of each standard was calcu-
lated from its retention time.
The isocratic HILIC elution profile of the 15 standards is shown in Figure 4.3,
and their k values are listed in Table 4.1. The chromatogram confirmed the trend of
the separation of these standard oligomers based on their DP. However, the retention
times of oligomers with the same DP did not match perfectly. We initially believed
that this variation was due to differences in hydrophobicity. In general, the hydropho-
bicity of an analyte is evaluated on a logarithmic scale of its octanol–water partition
coefficient, log Po/w. Thus, we measured log Po/w values for the 15 different standards
by HSCCC using an octanol–water two-phase solvent system.23,25 A scatter diagram
of the DP values of the 15 standards against their log Po/w values is shown in Figure
4.4. The diagram revealed a tendency for log Po/w values to decrease significantly
with increasing DP values. A minus log Po/w value indicates that the compound
is highly hydrophilic. Because these compounds are strongly partitioned into the
water-enriched stationary phase in general normal-phase partition chromatography,
differences in the hydrophilicity of the standards shown in Figure 4.4 actually con-
tribute to the DP-based separation as shown in Figure 4.3. However, a more detailed
analysis showed that the log Po/w values of compounds eluting at approximately the

1a
1c
0.025
1b
1d 3b
1e 2c 3c
2a
Abs. at 230 nm
2b
1f
3a
2d 4a
5a
0
0 10 20 30 40 50 60
Figure 4.3 HILIC separation profile of 15 standards of flavan-3-ols and oligomeric PAs,
eluted isocratically. The numbering scheme for the standards is identical to that in Figure 4.1.
(Reprinted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153, 2007. With permission
from Elsevier.)
Table 4.1
Comparison of Structural and Chromatographic Parameters of 15 Standard
Flavan-3-Ols and Oligomeric PAs
Chromatographic
Structural Parameters Parameters
Number
Degree of of OH k in
Compounds Polymerization Groups log Po/w HILIC
1a C 1 5 0.31 0.89
1b GC 1 6 −0.31 1.82
1c EC 1 5 0.10 0.89
1d EGC 1 6 −0.53 1.82
1e ECg 1 7 1.10 1.33
1f EGCg 1 8 0.50 2.44
2a EC(4b-8)C 2 10 −1.35 5.24
2b EC(4b-8)EC 2 10 −0.90 5.24
2c C(4a-8)C 2 10 −0.91 4.51
2d GC(4a-8)C 2 11 −1.52 6.80
3a EC(4b-8)EC(4b-8)EC 3 15 −1.32 8.43
3b EC(4b-8)EC(4b-6)EC 3 15 −1.49 7.81
3c EC(4b-6)EC(4b-8)EC 3 15 −1.13 7.81
4a EC(4b-8)EC(4b-8)EC(4b-8)EC 4 20 −1.79 10.12
5a EC(4b-8)EC(4b-8)EC(4b-8)EC(4b-8)EC 5 25 −2.21 11.50

1.5
1 1e
0.5 1f
1a
0 1c
Log Po/w
1b
–0.5 1d
–1 2b, 2c
3c
2a 3a
–1.5 2d 3b
4a
–2
5a
–2.5
1 2 3 4 5
DP
Figure 4.4 Scatter diagram of the degree of polymerization (DP) of the 15 standards
against their log Po/w values. (Reprinted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153,
2007. With permission from Elsevier.)
ttsame time in Figure 4.3 were not identical in Figure 4.4 (e.g., 1c and 1a and 1e, or
2c and 2b and 2a). This inconsistency could not be explained by a normal-phase
partitioning mechanism based on the hydrophilicity of the solute PA. Therefore, we
believed that another important separation factor (or interaction) contributes to the
PAs separation by HILIC.
4.7 Retention Mechanism of Oligomeric PAs in HILIC

Based on the structural attributes (DP and the number of hydroxyl groups) and the
chromatographic parameters (log Po/w, t R, k, and log k in the HILIC) listed in Table 4.1,
we examined the relationship between the log k and log Po/w values of the 15 different
standards. As shown in Figure 4.5, the relationship between log k and log Po/w was
expressed as the linear regression equation “y = −1.8177x − 0.4767.” This relationship
revealed that the hydrophilicity (i.e., log Po/w values) of the oligomer standards actually
contributes to DP-based separation. However, the square of the correlation coefficient
of this relationship is not very high (r2 = 0.7949). To gain insight into another factor
affecting separation by HILIC, we examined the relationship between the log k values
of the15 standards and their number of hydroxyl groups (OH). A high correlation was
observed, as shown in Figure 4.6. The relationship was expressed as the linear regres-
sion equation “y = 12.341x + 9.2541,” and the square of the correlation coefficient (r2 =
0.9501) was larger than the value (0.7949) obtained from the correlation between log k
and log Po/w values. This high correlation between log k and the number of OH groups
revealed that hydrogen bonding between a carbamoyl terminal on the Amide-80 col-
umn and the OH group of the solute strongly contributes to PA separation by HILIC.
The data shown in Figures 4.5 and 4.6 confirmed that two effects, hydrogen bonding
and solute hydrophilicity, both contribute to DP-based PA separation, but the hydro-
gen bonding effect dominates HILIC separations using an aqueous acetonitrile mobile

1.5
1e
1
y = –0.2379x + 0.4533
0.5 1f r2 = 0.7998
1a
0 1c
Log Po/w
1b
–0.5 1d
3c
–1
2c 2b
3a
–1.5 2a 4a
2d 3b
–2 5a
–2.5
0 2 4 6 8 10 12
k΄ in HILIC
Figure 4.5 Relationship between log k of 15 standards in HILIC and their log Po/w values.
The inset regression equation was calculated from the linear least-square fit of all data (n = 15).
(Reprinted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153, 2007. With permission from
Elsevier.)
30
y = 1.5922x + 3.0577
25 r2 = 0.9289 5a
Number of OH groups
4a
20
15 * 3a
3b
3c
10 *
1e 1f 2c 2a 2d
* 2b
5 *
1b
1a 1d
1c
0
0 2 4 6 8 10 12
k΄ in HILIC
Figure 4.6 Relationship between log k of 15 standards in HILIC and the number of
hydroxyl groups in their structures. The inset regression equation was calculated from the lin-
ear least-square fit of all data (n = 15). *—two plots are stacked up. (Reprinted from Yanagida, A.
et al., J. Chromatogr. A, 1143, 153, 2007. With permission from Elsevier.)
phase. Hydrogen bond formation between two compounds is generally enhanced under
hydrophobic conditions compared to that in aqueous conditions. Therefore, in previous
LC methods using a bare silica column with nonaqueous organic mobile phases,4,5,18–22
strong hydrogen bonding between the silanol groups of the stationary phase and the
hydroxyl groups of the solute PAs might have been the major factor favoring retention.
If this is correct, this mode of PA separation on silica columns should be referred to
as “adsorption.” On the other hand, in PA separations using an amide–silica column

with an aqueous mobile phase, solute hydrophilicity is an important separation factor.

In this case, “hydrophilic interaction chromatography (HILIC)” is the most appro-
priate term to describe mixed-mode separations relying on adsorption (due to hydro-
gen bonding) and normal-phase partitioning (due to solute hydrophilicity). Recently,
another HILIC method using a diol-silica column has been used for the DP-based
separation of oligomeric PAs (up to decamers) in cacao and related foods.26 At present,
HILIC methods using either amide or diol columns are the best LC approach for the
DP-based separation of oligomeric PAs.
4.8 HILIC Separation of Other Polyphenolic Compounds

For most researchers, reversed-phase HPLC using an ODS column is the first choice
for chromatographic separations of plant polyphenols. HILIC has only been used in a
few examples mentioned in this chapter for the separation of oligomeric PAs. However,
HILIC is a potentially useful LC mode for the separation of plant polyphenols since
these compounds have multiple hydroxyl and/or carboxyl groups that can interact with
the HILIC stationary phase by hydrogen bonding. In this section, several examples of
HILIC separation of low-molecular-mass polyphenols are discussed.
The HILIC separation profiles of six neutral phenolics (Figure 4.7A) and five
phenolic acids (Figure 4.7B) on a TSK-Gel Amide-80 column (250â•›mm × 4.6â•›mm
ID, 5â•›μm particle size) were obtained with an aqueous acetonitrile mobile phase.
1 2
3
6
4
7
Abs. at 210 nm
(A)
10
1
8 11
9
12
0 5 10 15 20 25
(B) Retention time (min)
Figure 4.7 HILIC separation profiles of low-molecular-mass polyphenols. (A) Chromatogram

of six neutral phenolics. (B) Chromatogram of five phenolic acids. Peak number 1: (not
retained) anthracene; 2: catechol; 3: kaempferol; 4: phloretin; 5: phloroglucinol; 6: epicat-
echin; 7: phloridzin; 8: p-coumaric acid; 9: caffeic acid; 10: 3,4-dihydroxybenzoic acid;
11: gallic acid; and 12: chlorogenic acid.

Q-3-gal
Q
Abs. at 210 nm
Q-3-glu-rha
0 5 10 15 20
Figure 4.8 HILIC separation profile of quercetin (Q), its monoglycoside (Q-3-gal), and
diglycoside (Q-3-glu-rha).
Both separations were achieved isocratically using acetonitrile-2.5â•›mM phosphoric

acid (95:5, v/v) at a flow rate of 1.0â•›mL/min. In both chromatograms, unretained
hydrophobic anthracene eluted in the void volume (t0 about 3â•›min). On the other hand,
all the solute phenolics and phenolic acids were retained and separated on the column,
and their retention time (and thus the elution order) generally corresponded with the
number of hydrogen bond donors (hydroxyl and carboxyl groups) in their structures.
The HILIC method is an effective method for the separation of polyphenolic gly-
cosides. The HILIC separation profile of a flavonol, quercetin (Q), and its glycosides
(Q-3-gal and Q-3-glu-rha) using an Amide-80 column (250â•›mm × 4.6â•›mm ID, 5â•›μm
particle size) is shown in Figure 4.8. The separation was achieved with a linear gra-
dient elution of acetonitrile–water at ratios of 95:5 to 80:20 (v/v) over 20â•›min at a
flow rate of 1.0â•›m L/min. In the chromatogram, the most hydrophobic aglycone (Q)
eluted first, its monoglycoside (Q-gal) eluted second, and its diglycosides (Q-3-glu-
rha) eluted last. This elution order corresponds to the increasing number of glycoside
moieties in the structure of the solute. Therefore, HILIC may be useful for determin-
ing the number of glycosides in an unknown flavonoid structure and may represent
a general approach for the analysis of sugar chains in glycoproteins. Furthermore,
similar separation profiles were observed from HILIC analyses of some phenolic
acids and their organic acid esters (data not shown).
The separation of the polyphenolic compounds as demonstrated in Figures 4.7 and
4.8 can also be achieved by conventional reversed-phase HPLC using an ODS column.
However, the elution order of the HILIC separation differs from that of a reversed-phase
HPLC separation. The HILIC profile provides additional information on solute structure,
such as molecular hydrophilicity, and the number of functional groups in the structure.
Another advantage of HILIC is the mobile-phase composition. The acetonitrile-rich
mobile phase used for HILIC can effectively solubilize moderately hydrophobic solutes
(such as quercetin) compared with the water-rich mobile phase used for reversed-phase
HPLC. The efficient and effective use of the above advantages will lead to the wide
application of HILIC for separations and structural determination of polyphenols and
other plant secondary metabolites, such as alkaloids, glycosides of steroids, or terpenes.

References
1. Alpert, A.J., J. Chromatogr., 1990, 499, 177.
2. Ohnishi-Kameyama, M., Yanagida, A., Kanda, T., and Nagata, T., Rapid Commun. Mass
Spectrom., 1997, 11, 31.
3. Yanagida, A., Kanda, T., Shoji, T., Ohnishi-Kameyama, M., and Nagata, T., J. Chromatogr.
A, 1999, 855, 181.
4. Yanagida, A., Kanda, T., Takahashi, T., Kamimura, A., Hamazono, T., and Honda, S.,
J. Chromatogr. A, 2000, 890, 251.
5. Shoji, T., Mutsuga, M., Nakamura, T., Kanda, T., Akiyama, H., and Goda, Y., J. Agric.
Food Chem., 2003, 51, 3806.
6. Lea, A.G.H., J. Chromatogr., 1982, 238, 253.
7. Delage, E., Bohuon, G., Baron, A., and Drilleau, J.-F., J. Chromatogr., 1991, 555, 125.
8. Rohr, G.E., Meier, B., and Sticher, O., J. Chromatogr. A, 1999, 835, 59.
9. Bae, Y.S., Foo, L.Y., and Karchecy, J.J., Holzforschung, 1994, 48, 4.
10. López-Serrano, M. and Barceló, A.R., J. Chromatogr. A, 2001, 919, 267.
11. Kennedy, J.A. and Taylor, A.W., J. Chromatogr. A, 2003, 995, 99.
12. Kurumatani, M., Fujita, R., Tagashira, M., Shoji, T., Kanda, T., Ikeda, M., Shoji, A.,
Yanagida, A., Shibusawa, Y., Shindo, H., and Ito, Y., J. Liq. Chromatogr. Relat. Technol.,
2005, 28, 1971.
13. Yanagida, A., Shoji, T., and Kanda, T., Biosci. Biotechnol. Biochem., 2002, 66, 1972.
14. Bourvellec, C.Le., Picot, M., and Renard, C.M.G.C., Anal. Chim. Acta., 2006, 563, 33.
15. Shibusawa, Y., Yanagida, A., Isozaki, M., Shindo, H., and Ito, Y., J. Chromatogr. A,
2001, 915, 253.
16. Shibusawa, Y., Yanagida, A., Shindo, H., and Ito, Y., J. Liq. Chromatogr. Relat. Technol.,
2003, 26, 1609.
17. Köhler, N., Wray, V., and Winterhalter, P., J. Chromatogr. A, 2008, 1177, 114.
18. Rigaud, J., Escribano-Bailon, M.T., Prieur, C., Souquet, J.-M., and Cheynier, V.,
J. Chromatogr. A, 1993, 654, 255.
19. Hammerstone, J.F., Lazarus, S.A., Mitchell, A.E., Rucker, R., and Schmitz, H.H.,
J. Agric. Food Chem., 1999, 47, 490.
20. Natsume, M., Osakabe, N., Yamagishi, M., Takizawa, T., Nakamura, T., Miyatake, H.,
Hatano, T., and Yoshida, T., Biosci. Biotechnol. Biochem., 2000, 64, 2581.
21. Kennedy, J.A. and Waterhouse, A.L., J. Chromatogr. A, 2000, 866, 25.
22. Gu, L., Kelm, M., Hammerstone, J.F., Beecher, G., Cunningham, D., Vannozzi, S., and
Prior, R.L., J. Agric. Food Chem., 2002, 50, 4852.
23. Yanagida, A., Murao, H., Ohnishi-Kameyama, M., Yamakawa, Y., Shoji, A., Tagashira, M.,
Kanda, T., Shindo, H., and Shibusawa, Y., J. Chromatogr. A, March 2, 2007, 1143, 153.
25. Shibusawa, Y., Shoji, A., Yanagida, A., Shindo, H., Tagashira, M., Ikeda, M., and Ito, Y.,
J. Liq. Chromatogr. Relat. Technol., 2005, 28, 2819.
26. Kelm, M.A., Johnson, J.C., Robbins, R.J., Hammerstone, J.F., and Schmitz, H.H.,
J. Agric. Food Chem., 2006, 54, 1571.

5 Applications of
Chromatography
to Food Analysis
Leticia Mora, Aleida S. Hernández-Cázares,
M-Concepción Aristoy, Fidel Toldrá,
and Milagro Reig
Contents
5.1 Introduction..................................................................................................... 89
5.2 Application of Hydrophilic Interaction Chromatography to the Analysis
of Quality-Related Compounds in Foods........................................................90
5.2.1 HILIC with Conventional Detectors....................................................90
5.2.2 HILIC Coupled to Mass Spectrometry................................................ 95
5.3 Application of Hydrophilic Interaction Chromatography to the Analysis
of Safety-Related Compounds in Foods.......................................................... 96
Acknowledgments................................................................................................... 104
References............................................................................................................... 104
5.1 Introduction
There are many substances in foods that must be analyzed because of their contri-
bution to the final quality (i.e., protein compounds, lipid compounds, nucleotides),
nutrition (i.e., vitamins, essential amino acids, minerals, etc.), or safety (i.e., growth
promoters or antimicrobial residues). Different methodologies are being used for
their analysis. High-performance liquid chromatography (HPLC) is one of the most
widely used analytical techniques due to its versatility, speed, and high resolving
power. Reversed-phase high-performance liquid chromatography (RP-HPLC) with
and without ion-pairing agents and ion-exchange HPLC have been the methods of
choice for the determination of these compounds during the last decades. However,
limitations of RP-HPLC, such as the poor retention of polar molecules, show the
need for alternative separation modes. Hydrophilic interaction chromatography
(HILIC) represents a viable alternative for the separation of polar compounds in
foods for many reasons. This separation mode uses mobile phases with similar
89

organic solvents as those used in RP-HPLC with the consequent advantage of ana-
lyte solubility. Due to good retention of polar analytes, neither derivatization nor
ion-pair reagents are necessary for the analysis of polar or ionic compounds. An
additional important advantage of HILIC is its compatibility with mass spectro-
metric detection, eliminating the desalting step needed when using ion-exchange
chromatography. Furthermore, HILIC has proven to be a simple, fast, and reliable
alternative to the existing methods.
This chapter focuses on the recent applications of HILIC to the analysis of com-
pounds that are present in foods and are relevant for either food quality or safety.
Examples where HILIC is used for food quality testing including, among oth-
ers, small peptides and key compounds like creatine or nucleotides are detailed.
Examples using HILIC for the analysis of residues of compounds affecting the safety
of foods such as relevant growth promoters and polar antimicrobials in foods of ani-
mal origin are also presented.
5.2 Application of Hydrophilic Interaction

Chromatography to the Analysis of
Quality-Related Compounds in Foods
5.2.1 HILIC with Conventional Detectors
The analysis of nucleotides and nucleosides in meat is very important to follow the
postmortem process just after slaughter. In fact, the evolution of such compounds
at early postmortem can give a good indication of the final quality of meat.1–3 The
simultaneous analysis of adenosine triphosphate (ATP), adenosine diphosphate
(ADP), adenosine monophosphate (AMP), inosine monophosphate (IMP), inosine
(Ino), hypoxanthine (Hx), nicotinamide adenine dinucleotide (NAD+), and creati-
nine (Cn) in pork loin muscle was developed using a zwitterionic polymeric column
(ZIC-pHILIC from SeQuant). A comparison between the pork meat concentrations
of the compounds analyzed by HILIC and by ion-pair reversed-phase chromatog-
raphy (IP-RP-HPLC), using a Zorbax Eclipse XDB-C18 column (Agilent), was
performed.4 The good agreement obtained between the two sets of data shows that
HILIC is well suited to the separation and quantification of nucleotides in complex
matrices such as meat and provides an interesting alternative to other methodologies
(see Figure 5.1A and B). This same method, with conditions as listed in Table 5.1,
also showed good performance in the separation of guanosine (G), guanosine mono-
phosphate (GMP), guanosine diphosphate (GDP), guanosine triphosphate (GTP),
and uric acid (UA).4
Other compounds present in skeletal muscle and with relevance in postmortem
meat have also been analyzed with a ZIC-pHILIC column. This is the case for cre-
atine that provides the necessary energy for muscle contraction. Its amount in muscle
is kept in equilibrium with phosphocreatine since creatine is reversibly converted
into phosphocreatine by creatine kinase. This reaction involves the transfer of the
γ-phosphate group of ATP onto the guanidine group of creatine. In postmortem
meat, creatine is converted into creatinine through a nonenzymatic reaction that is
enhanced at increased temperature.

Applications of Hydrophilic Interaction Chromatography to Food Analysis 91
350
6
Absorbance at 254 nm (mAU) 300
250 3
200
150
100
5
2 4
50 1 8
0 7
–50
0 5 10 15 20 25
(A) Time (min)
850
6
750
Absorbance at 254 nm (mAU)
650 3
550
450
350
250 2
5
150 4
7 8
50
–50
0 5 10 15 20 25
(B) Time (min)
Figure 5.1 Chromatogram of pork meat sampled at 9â•›h postmortem. The column
was (A) ZIC-pHILIC, 150â•›mm × 4.6â•›mm, 5â•›μm (SeQuant) and (B) Zorbax Eclipse XDB-C18,
150 mm × 4.6â•›m m, 5â•›μm (Agilent). Numbers 1 to 8 correspond to Cn, Hx, Ino, AMP, NAD+,
IMP, ADP, and ATP, respectively.
Other compounds of interest are the histidine-containing dipeptides—carnosine,

anserine, and balenine which are extensively distributed in skeletal muscle. They
have been reported to act as intracellular buffers, modulators for some enzymes, neu-
rotransmitters, metal–ion chelators, antioxidants, and free-radical scavengers and
are thus very important functional compounds in meat.5 The simultaneous deter-
mination of carnosine, anserine, balenine, creatine, and creatinine in pork loin and
chicken breast meat was recently developed using an Atlantis silica column (Waters).6
Samples were extracted, deproteinized, and centrifuged and were then injected into
the HPLC (see Table 5.1 for further details). The separation was monitored using
a diode array detector at a wavelength of 214â•›nm for creatine, carnosine, anserine,
and balenine, whereas a wavelength of 236â•›nm was used for creatinine detection (see
Figure 5.2). This chromatographic method was subsequently applied in seven pork
muscles of different metabolic types. The results showed that the content of creatine and
creatinine were significantly higher in glycolytic muscles such as Semimembranosus,

92
Table 5.1
Performance of Some Recent HILIC-Based Methods for the Analysis of Compounds of Interest in Foods
Compound
to Analyze Food Matrix Extraction Type of Column Elution Conditions Detector Reference
ATP, ADP, AMP, IMP, Pork meat 0.6â•›M perchloric ZIC-pHILIC (SeQuant) Solvent A: 22.5â•›mM NH4Ac, DAD at 254â•›nm [4]
Ino, Hx, and NAD+ acid and 150â•›mm × 4.6â•›mm, 5â•›μm pH 3.5 in water/ACN (20:80);
neutralization with Solvent B: 50â•›mM NH4Ac,
solid potassium pH 7 in water/ACN (60:40)
carbonate Step gradient from 0% to 100%
of B in 3â•›min
Flow rate = 0.5â•›mL/min
Carnosine, anserine, Pork loin and 0.01N HCl and Atlantis silica (Waters) Solvent A: 0.65â•›mM NH4Ac, DAD at 214 and [6]
and balenine chicken breast deproteinization 150â•›mm × 4.6â•›mm, 3â•›μm pH 5.5 in water/ACN (25:75); 236â•›nm
dipeptides, creatine with ACN solvent B: 4.55â•›mM NH4Ac
and creatinine pH 5.5 in water/ACN (70:30)
Linear gradient from 0% to
100% of B in 13â•›min
Polar compounds Mushrooms Soaking in drinking Semipreparative Solvent A, ACN and aqueous ESI+ and [11]
contributing to water for 12â•›h at TSK-Gel Amide 80 NH4-formate (7â•›mM; pH 5.5), ESI− LCQ MS
umami-taste and room temperature (Tosoh BioSep), (80/20); solvent B, ACN and
mouth-drying oral 300â•›mm × 7.8â•›mm aqueous NH4-formate (7â•›mM;
sensation pH 5.5) (20:80)
Step gradient to 35% B

Applications of Hydrophilic Interaction Chromatography to Food Analysis
Free amino acids, Wheat gluten Wheat gluten was TSK-Gel Amide 80 In positive ionization mode: ESI+ and [12]
di- and tri-peptides and Parmesan dehydrolyzed with (Tosoh BioSep), NH4Ac buffer (pH 5.5) and an ESI− LCQ ion
and glycoconjugates cheese 0.1â•›M HCl 250â•›mm × 1.5â•›mm, aqueous gradient (10%–40% trap MS
Parmesan cheese 5â•›μm, 80â•›Å in ACN)
was defatted and For analysis in the negative
extracted with ionization mode, NH4Ac
water buffer (pH 7.0) and same
gradient
Flow rate = 40â•›μL/min
Dityrosine (DiTyr) Milk powder Hydrolysis with 6N TSK-Gel Amide 80 Solvent A, water and solvent B, ESI+ triple Q [14]
HCl and recovery (Tosoh BioSep), ACN
of DiTyr by SPE 250â•›mm × 1.5â•›mm, Both adjusted at 2.7 with
5â•›μm, 80â•›Å formic acid
The gradient was: 20% A for
3â•›min, 70% A at 10â•›min and
isocratic at 70% A
Flow rate = 200â•›μL/min
Taurine and Beverages Drinks were diluted Astec apHera™ NH2 Methanol–water (60/40) ESI− triple Q [15]
methionine rich in with the mobile polymer, Flow rate = 0.6â•›mL/min
carbohydrates phase (methanol/ 150â•›mm × 4.6â•›mm, 5â•›μm
water, 60:40, v/v)
93
1400 80
2
1200
Absorbance (mAU)
Absorbance (mAU)
1000 1 3 60
800
40
600
400 20
200 4 5
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13
(A) Time (min)
1200 25
2 5
1000 20
1
Absorbance (mAU)
Absorbance (mAU)
800
15
600
3 10
400
200 5
4
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13
(B) Time (min)
Figure 5.2 Chromatograms corresponding to pork loin (A) and chicken breast (B).
Numbers 1, 2, 3, 4, and 5 correspond to creatinine, creatine, carnosine, baline, and anserine,
respectively. The column was an Atlantis silica, 150â•›m m × 4.6â•›m m, 3â•›μm (water). Left scale
indicates 214â•›nm measures for all compounds except creatinine (solid line), whereas right
scale indicates 236â•›nm absorbance (dotted line) for creatinine.
Biceps femoris, Gluteus maximus, and Longissimus dorsi, whereas Masseter, a red oxi-
dative muscle, was characterized by the lowest contents of these compounds.7 HILIC
methodology was also employed to evaluate the effect of different cooking procedures
on the concentrations of creatine and creatinine in cooked ham. The results allowed
the establishment of a correlation between the creatine/creatinine ratio in different sec-
tions of the ham and the effectiveness of the heat treatment in cooked ham processes.8
The ZIC-pHILIC column has also been used by the current authors to study the evolu-
tion of creatine and creatinine during the processing of Spanish dry-cured ham obtain-
ing good results in the isolation of these compounds (see Figure 5.3).
Another application demonstrating the benefits of HILIC separations was the detec-
tion of 1-deoxynojrimycin in mulberry leaves. This compound is a natural alkaloid
with some biological activity in vivo. Mulberry leaves containing such compound are
used as ingredients in functional foods in Japan.9 Functional foods claim to have a
health-promoting or disease-preventing property beyond the basic function of sup-
plying nutrients. The method consisted of extraction with a mixture of acetonitrile
and water containing 6.5â•›mM ammonium acetate, pH 5.5 (81:19, v/v) followed by

Applications of Hydrophilic Interaction Chromatography to Food Analysis 95
1200 2 250
1

1000 200
800
150
600
100
400
200 50
0 0
0 2 4 6 8 10 12 14
Time (min)
Figure 5.3 Chromatogram of a 50 day post-salted dry-cured ham sampled at the semimem-
branosus muscle. The column was a ZIC-pHILIC, 150 mmâ•›×â•›4.6â•›mm, 5â•›μm (SeQuant). Creatine
(1): solid line corresponding to 210â•›nm. Creatinine (2): dotted line corresponding to 236â•›nm.
separation using an amide-bonded silica column (TSK-Gel Amide-80 from Tosoh) and
detection with an evaporative light-scattering detector (see Table 5.1). The detection
limit was 100â•›ng (0.613â•›nmol) on the column. Thus, this analysis is useful to determine
the content of mulberry leaves as ingredient in functional foods. The analysis of water
soluble vitamins in energy drinks after HILIC separation with the TSK-Gel Amide-80
(Tosoh) column and UV detection has also been recently reported.10
5.2.2 HILIC Coupled to Mass Spectrometry

HILIC mode separations provide many advantages when coupled to mass spectrom-
etry (MS) and this combination has been widely used in food analysis. Compounds
extracted from mushroom were analyzed using HILIC-MS to determine the key
compounds contributing to the umami-taste and mouth-drying oral sensation of
morel mushrooms.11 These polar compounds were analyzed using the TSK-Gel
Amide-80 (Tosoh) column coupled to an electrospray ionization (ESI) quadrupole
mass spectrometer (MS) as described in Table 5.1.
HILIC in combination with ESI-MS has also been employed to separate and
characterize low-molecular-weight compounds in some foods. Free amino acids,
glutamyl di- and tri-peptides, their glycoconjugates, and organic acids have been
separated and identified without the need for derivatization in complex food
matrices such as wheat gluten hydrosylate and Parmesan cheese using a TSK-Gel
Amide-80 (Tosoh) column.12 These amino acids, peptides, and glycoconjugates
eluted at the void volume in an RP-HPLC but could be separated and character-
ized by HILIC-ESI-MS in the positive ionization mode using an ammonium acetate
buffer (pH 5.5). ESI in the negative ionization mode was used for the analysis of
glycoconjugates with glutamate due to the negative charge at the second carbox-
ylic acid group.12 A complex mixture of glycopeptides and N- and O-glycopeptides
have also been efficiently separated with a capillary ZIC-HILIC (SeQuant) column
coupled to ESI-MS.13

A TSK-Gel Amide-80 (Tosoh) column was also used for the development of an
HILIC-MS method to accurately quantify dityrosine (DiTyr) in milk powder sam-
ples using d4-DiTyr as the internal standard.14 Milk proteins were first hydrolyzed
by hydrochloric acid, and DiTyr further recovered from the amino acid hydrolyzates
by a solid-phase extraction (see Table 5.1). Results demonstrated that DiTyr could be
potentially used as a good chemical marker of milk protein oxidation.
HILIC-MS analysis has been also extended to the field of water and energy drinks.
A validated procedure based on HILIC coupled to tandem mass spectrometry was
developed for the simultaneous determination of underivatized taurine and methio-
nine in beverages rich in carbohydrates such as energy drinks.15 Sample prepara-
tion only required an appropriate dilution before injection, and the chromatographic
separation of the compounds of interest was carried out using an Astec apHera NH2
polymer column.
5.3 Application of Hydrophilic Interaction

Chromatography to the Analysis of
Safety-Related Compounds in Foods
Compounds posing health threats have received much attention in recent years. One
of these substances is melamine and its determination as a contaminant in milk,
fish, and other food products has to be analyzed due to its presence from adultera-
tion or misuse.16 Melamine is a triazine-based chemical and is typically used in the
manufacture of plastics. Melamine has been detected in wheat flour, milk, fish, feed,
etc. The main health concern with melamine is the formation of bladder stones with
some studies reporting microcrystalluria.17 Furthermore, insoluble melamine cyan-
urate salt crystals can also be formed in the kidney if cyanuric acid is present in the
diet.18 A HILIC-based method using bare silica has been recently developed for the
analysis of melamine in fish and shrimp edible tissues with a detection limit as low
as 3.2â•›ng/g.16 Further chromatographic details are shown in Table 5.2. The authors
surveyed 105 market-ready shrimp, catfish, tilapia, salmon, eel, and other fishes
finding melamine concentrations exceeding 50â•›ng/g for those raised with feeds con-
taining low amounts of melanine, 6.7â•›mg/kg. Other authors have used a ZIC-HILIC
column for the analysis of melanine in milk powder with UV detection, reaching a
limit of quantification of 15â•›ng/mL.17
Dichloroacetic acid was detected in drinking water using ion-exchange and
HILIC on an amino column in conjunction with ESI-MS.18 A TSK-Gel Amide-80
(Tosoh) column in combination with ESI-MS/MS was successfully employed to sep-
arate polar mushrooms toxins including amanitins, phallotoxins, and muscarine.19
These authors studied the impact of different mobile phases on the separation, con-
cluding that the addition of methanol as a modifier in the mobile phase (comprising
acetonitrile, methanol, and ammonium formate, pH 3.5) could further enhance the
HILIC separation for the studied analytes.
HILIC-ESI-MS methodology was also effective when applied to roasted coffee
powder for the analysis of two polar compounds, 4(5)-methylimidazole and 2-acetyl-
4(5)-1,2,3,4)-tetrahydroxybutyl-imidazole. These undesirable compounds, which are

Table 5.2
Performance of Some Recent HILIC-HPLC-Based Methods for the Analysis of Contaminant Residues in Foods
Compounds Food LODa LOQa
to Analyze Matrix Main Extraction Type of Column Elution Conditions Detector (ng/mL) (ng/mL) Reference
Melamine Fish and Dry ice Atlantis silica Gradient from 95% ESI+ triple Q 3.2 — [18]
shrimp homogenization, (Waters), to 50% of
extraction with 50â•›mm × 3.0â•›mm, acetonitrile and
acetonitrile, water, 3â•›μm 20â•›mM aqueous
and HCl ammonium formate
Oasis MCX SPE Flow rate =
0.35â•›mL/min
Melamine Milk Aqueous solution SeQuant ZIC-HILIC Acetonitrile/ UV at — 15 [19]
powder containing 2.5% (Merck), 150â•›mm × ammonium acetate 210â•›nm
ammonia or 4.6â•›mm, 5â•›μm, 25â•›mN (85/15%,
perchloric acid 200â•›Å v/v), pH 6.8
Centrifugation Flow rate =
2â•›mL/min
Dichloroacetic acid Drinking Water samples were Luna Amino Solvent A, ACN ESI− triple Q 1 5 [20]
water dried under vacuum (Phenomenex) and solvent B,
and reconstituted in 150â•›mm × 2.1â•›mm, 40â•›mM NH4
60:40 ACN:water 5â•›μm formate in water
Linear gradient from
90% to 30% of
solvent A in 5â•›min
Flow rate =
0.7â•›mL/min
(continued)
97
98
Table 5.2 (continued)
Performance of Some Recent HILIC-HPLC-Based Methods for the Analysis of Contaminant Residues in Foods
Compounds Food LODa LOQa
to Analyze Matrix Main Extraction Type of Column Elution Conditions Detector (ng/mL) (ng/mL) Reference
Polar toxins
Amanitins Mushroom Extraction with TSK-Gel Amide 80 Different ESI MS/MS 20 30 [21]
acetonitrile and (Tosoh BioSep), proportions of
formic acid 250â•›mm × 2.0â•›mm, 2â•›mM NH4-
Oasis MAX anion- 5â•›μm, 80â•›Å formate + 5â•›mM
exchange cartridge formic acid at pH
3.5 acetonitrile
and MeOH were
tested
Flow rate =
0.2â•›mL/min
Phallotoxins SAA SAA SAA SAA SAA 20 30 SAA
Muscarine SAA SAA SAA SAA SAA 5 5 SAA
4(5)-methylimidazole Roasted Supercritical CO2 Atlantis silica Mobile phase. LC-DAD at 0.0015 0.0044 [22]
(4-MeI) coffee modified with (Waters) 150â•›mm × MeOH (80%) and 290â•›nm
powder aqueous MeOH and 2.1â•›mm, 3â•›μm 0.01â•›M NH4-
SPE using SCX Disc formate in water
cartridges (20%)
Isocratic mode
Flow rate =
0.2â•›mL/min

2-Acetyl-4(5)-(1,2,3,4) SAA SAA SAA SAA LC-DAD at 0.002 0.0063 SAA
tetrahydroxybutyl- 215â•›nm
imidazole (THI)
Glycoalkaloids Potatoes Sample extracted with Polyhydroxyethyl A Solvent A, 5mM ESI+ LC — 20 [23]
α-chaconine and chloroform:methanol: (PolyLC) NH4Ac pH 5.5 and MS/MS
α-solanine water (2:5:2, v/v/v) 100â•›mm × 2.1â•›mm, solvent B, ACN
3â•›μm Step gradient from
20% A to 100% A
Biogenic amines
Cadaverine Cheese Hydrolysis with 0.1â•›M Atlantis silica Acetonitrile and ESI+ MS/MS 3.5 10 [26]
HCl C-18 SPE (Waters), ammonium
150 × 2.1â•›mm, 3â•›μm formate 50â•›mM in
water, pH 4.0
Histamine SAA SAA SAA SAA SAA 1.5 5 SAA
Putrescine SAA SAA SAA SAA SAA 1.2 4 SAA
Spermidine SAA SAA SAA SAA SAA 1.0 3 SAA
Spermine SAA SAA SAA SAA SAA 1.2 6 SAA
Tryptamine SAA SAA SAA SAA SAA 1.3 3.5 SAA
Tyramine SAA SAA SAA SAA SAA 1.1 3 SAA
Note: SAA, same as above.

Units in ng/mL or ng/g.
a
99
neÂ�urotoxic, are formed during the coffee roasting process. The analytes were extracted
with supercritical carbon dioxide modified with aqueous methanol, and subsequently
analyzed using HILIC-MS with an Atlantis HILIC silica (Waters) column.20
HILIC has become popular recently (in the last 4–5 years) for the analysis of
polar and highly hydrophilic compounds. In relation to this, a classical RP column
(Hyperclone ODS C18 column, 100 × 2â•›mm, 3â•›μm, from Phenomenex) was com-
pared to a HILIC column (polyhydroxyethyl A column, 100 × 2.1â•›mm, 3â•›μm, from
PolyLC) for the detection and quantification of α-chaconine and α-solanine, the
two major glycoalkaloids in potatoes, using LC-ESI-MS/MS.21 Both methods proved
to be powerful additions to standard metabolomic approaches although according
to these authors, the HILIC-based method somehow disclosed its lack of rugged-
ness with respect to chromatographic peak shapes when hundreds of samples were
analyzed for this application. In spite of this, HILIC methodology has extensively
proved to be very robust in high throughput laboratories.22,23
The determination of biogenic amines in cheese is of interest because their
presence and relative amounts give useful information about freshness, level of
maturing, quality of storage, and cheese typification. A method based on HILIC
MS/MS was developed for the determination of seven biogenic amines (cadav-
erine, histamine, putrescine, spermidine, spermine, tryptamine, and tyramine) in
cheese.24 The method used an Atlantis silica (Waters) column (see Table 5.2) and
was reported to improve both the chromatographic separation and the mass spec-
trometry detection with respect to an HPLC-MS/MS method developed in that
laboratory.
The presence of residues of growth promoters or veterinary drugs in foods of
animal origin, which are generally used in farm animals for therapeutic and prophy-
lactic purposes, makes necessary the control of these substances in order to assure
consumer protection against any potentially harmful effects. Numerous analytical
methods exist for determining such substances25–28 in foods of animal origin, but
there is a recent trend toward the use of HILIC to separate the more polar compounds
like antibiotics. The main aspects of recent HILIC methods developed for the detec-
tion of drugs residues in foods are shown in Table 5.3.
Sulfonamide antibacterial residues have been successfully analyzed in milk
and eggs using a HILIC-MS method and a Luna NH2 (Phenomenex) column with
previous polymer monolith microextraction.29 The use of a ZIC-HILIC (SeQuant)
column allowed good detection of aminoglycosides antibiotics in kidney,30 neomy-
cin in serum,31 and cytostatics in wastewaters.32 The use of an Altima HP HILIC
column (Alltech) allowed the determination of the antibiotics spectinomycin and
lincomycin in hog manure33 (see Table 5.3). Carbadox and olaquindox have been
analyzed in feed using an Acquity BEH HILIC (Waters) column followed by
DAD.34
In summary, the use of HILIC coupled to DAD or MS detectors has been growing
rapidly in food applications in recent years. In fact, a good number of food quality
or food safety related applications are already available as reported in this chapter.
Furthermore, new types of HILIC phases and columns are being developed that will
facilitate the further expansion of HILIC to a wider range of applications in the field
of food analysis for quality and safety purposes.

Table 5.3
Performance of Some Recent HILIC-HPLC-Based Methods for the Analysis of Drugs Residues in Foods
Compounds LODa LOQa
to Analyze Food Matrix Main Extraction Type of Column Elution Conditions Detector (ng/mL) (ng/mL) Reference
Sulfonamides Whole milk PMME (poly(MAA- Luna NH2 column, Solvent A, ESI MS
+ 0.4–5.7 1.8–19.0 [31]
EGDMA) Phenomenex acetonitrile + 0.05%
monolithic column (Torrence, California) formic acid, and
150â•›mm × 2.0â•›mm, solvent B, water
3â•›μm gradient from 100%
to 95% ACN
SAA Eggs SAA SAA SAA SAA 0.9–9.8 3.2–32.8 SAA
Gentamicin Swine kidney Weak cation ZIC• HILIC, Sequant Solvent A, 1% formic ESI+ MS/ 6 19 [32]
exchange SPE (Umea, Sweden) acid + 150â•›mM MS
2.1â•›mm × 100â•›mm, ammonium acetate
5â•›μm and solvent B
acetonitrile
Gradient from 20%
to 95% of solvent A
Flow rate =
0.3â•›mL/min, 32°C
Spectinomycin SAA SAA SAA SAA SAA 19 63 SAA
Dihydrostreptomycin SAA SAA SAA SAA SAA 16 52 SAA
Kanamycin SAA SAA SAA SAA SAA 6 58 SAA
Apramycin SAA SAA SAA SAA SAA 16 52 SAA
Streptomycin SAA SAA SAA SAA SAA 18 90 SAA
Neomycin SAA SAA SAA SAA SAA 33 108 SAA
(continued)
101
Performance of Some Recent HILIC-HPLC-Based Methods for the Analysis of Drugs Residues in Foods
Compounds LODa LOQa
to Analyze Food Matrix Main Extraction Type of Column Elution Conditions Detector (ng/mL) (ng/mL) Reference
Neomycin Human serum Oasis MCX SPE ZIC-HILIC, Sequant Solvent A, mixture of ESI MS/ — 100 33
cartridges (Umea, Sweden) acetonitrile, 10â•›mM MS
100â•›mm × 2.1â•›mm, ammonium acetate
5â•›μm and formic
(5/95/0.2, v/v/v),
and solvent B
(95/5/0.2, v/v/v)
Gradient from 20%
to 80% of solvent A
Flow rate =
0.6â•›mL/min
5-Fluorouracyl Wastewater Isolute ENV + SPE ZIC-HILIC in 30â•›mM ammonium LTQ — 0.005 [34]
cartridge and polyetheretherketone, acetate and orbitrap
Speedisk Sequant acetonitrile (2/3, v/v)
H2O-philic (Umea, Sweden) Flow rate =
SA-DVB disc 150â•›mm × 2.1â•›mm, 0.2â•›mL/min
cartridge 3.5â•›μm
Gemcitabine SAA SAA SAA — 0.0009 SAA

Spectinomycin Hog manure Oasis HLB + Oasis Altima HP HILIC Solvent A, 90/10 APCI+ — 6 [35]
WCX SPE (Alltech Assoc, IL) Acetonitrile/water + triple Q
cartridges 150â•›mm × 2.1â•›mm, 0.1% formic acid,
3â•›μm and solvent B,
90/10 water/
acetonitrile + 0.1%
formic acid
Isocratic conditions
of 65% solvent A
and 35% solvent B
Flow rate =
0.2â•›mL/min
Lincomycin SAA Oasis HLB SPE SAA SAA SAA — 0.04 SAA
cartridge
Carbadox Feed C18 SPE Acquity UPLC BEH Acetonitrile–water DAD at 20 70 [36]
HILIC column (95/5, v/v) 307â•›nm
100â•›mm × 2.1â•›mm, containing 10â•›mM
1.7â•›μm ammonium acetate
Flow rate =
0.5â•›mL/min 30°C
Olaquindox SAA SAA SAA SAA DAD at 30 100 SAA
384â•›nm
Note: SAA, same as above.

a Units in ng/mL or ng/g.
103
Acknowledgments
Project A-001/09 from Platform on Food Safety, Conselleria de Sanitat, Generalitat
Valencia (Spain) is acknowledged.
References
1. Batlle, N. et al. J. Food Sci. 2000, 65: 413–416.
2. Batlle, N. et al. J. Food Sci. 2001, 66: 68–71.
3. Toldrá, F. In: Handbook of Food Science, Technology and Engineering (Y.H. Hui, Ed.),
CRC Press, Boca Raton, FL, 2006, pp. 28-1–28-18.
4. Mora, L. et al. Food Chem. 2010, 123: 1282–1288.
5. Wu, H.C. et al. J. Food Drug Anal. 2003, 11: 148–153.
6. Mora, L. et al. J. Agric. Food Chem. 2007, 55: 4664–4669.
7. Mora, L. et al. Meat Sci. 2008, 79: 709–715.
8. Mora, L. et al. J. Agric. Food Chem. 2008, 56: 11279–11284.
9. Kimura, T. et al. J. Agric. Food Chem. 2004, 52: 1415–1418.
10. Roemling, R. et al. The Applications Book 2009, March: 21–23.
11. Rotzoll, N. et al. J. Agric. Food Chem. 2005, 53: 4149–4156.
12. Schlichtherle-Cerny, H. et al. Anal. Chem. 2003, 75: 2349–2354.
13. Tagekawa, Y. et al. J. Sep. Sci. 2008, 31: 1585–1593.
14. Fenaille, F. et al. J. Chromatogr. A 2004, 1052: 77–84.
15. De Pearson, M. et al. J. Chromatogr. A 2005, 1081: 174–181.
16. Food and Drug Administration. 72 FR 30014. May 30, 2007.
17. World Health Organisation, 2009. Report 1–66.
18. Andersen, W.C. et al. J. Agric. Food Chem., 2008, 56: 4340–4347.
19. Jiang, W. and Ihunegoo, F.N. The Applications Book 2009, March: 40–41.
20. Dixon, A.M. et al. J. Liq. Chromatogr. Relat. Technol. 2004, 27: 2343–2355.
21. Chung, W.C. et al. J. Chromatogr. Sci. 2007, 45: 104–111.
22. Lojkova, L. et al. Food Addit. Contam. 2006, 23: 963–973.
23. Zywicki, B. et al. Anal. Biochem. 2005, 336: 178–186.
24. Naidong, W. J. Chromatogr. B, 2003, 796: 209–224.
25. Strege, M. Anal. Chem. 1998, 70: 2439–2445.
26. Gianotti, V. et al. J. Chromatogr. A 2008, 1185: 296–300.
27. Reig, M. and Toldrá, F. In: Handbook of Muscle Foods Analysis (L.M.L. Nollet and
F. Toldrá, Eds.), CRC Press, Boca Raton, FL, 2009, pp. 837–854.
28. Reig, M. and Toldrá, F. In: Handbook of Processed Meats and Poultry Analysis (L.M.L.
Nollet and F. Toldrá, Eds.), CRC Press, Boca Raton, FL, 2009, pp. 647–664.
29. Reig, M. and Toldrá, F. In: Safety of Meat and Processed Meats (F. Toldrá, Eds.),
Springer, New York, 2009, pp. 365–390.
30. Verdon, E. In: Handbook of Muscle Foods Analysis (L.M.L. Nollet and F. Toldrá, Eds.),
CRC Press, Boca Raton, FL, 2009, pp. 855–947.
31. Zheng, M.M. et al. Anal. Chim. Acta 2008, 625: 160–172.
32. Ishii, R. J. Food Addit. Contam. 2008, 20: 1–11.
33. Oertel, R. et al. J. Pharm. Biomed. Anal. 2004, 35: 633–638.
34. Kovalova, L. et al. J. Chromatogr. A. 2009, 1216: 1100–1108.
35. Peru, K.M. et al. J. Chromatogr. A. 2006, 1107: 152–158.
36. Kesiunaité, G. et al. J. Chromatogr. 2008, 1209: 83–87.

6 Hydrophilic Interaction
Liquid Chromatography–
Mass Spectrometry
(HILIC–MS) of Paralytic
Shellfish Poisoning
Toxins, Domoic
Acid, and Assorted
Cyanobacterial Toxins
Carmela Dell’Aversano
Contents
6.1 Introduction................................................................................................... 105
6.2 Paralytic Shellfish Poisoning Toxins............................................................. 108
6.2.1 HILIC-MS Determination of PSP Toxins......................................... 109
6.2.2 Application to Plankton and Shellfish Samples................................. 120
6.3 Domoic Acid.................................................................................................. 122
6.3.1 HILIC-MS Determination of DA...................................................... 122
6.3.2 Application to Shellfish Samples....................................................... 124
6.4 Assorted Cyanobacterial Toxins.................................................................... 126
6.4.1 HILIC-MS Determination of Assorted Cyanobacterial Toxins........ 126
6.4.2 Applications to Cyanobacterial Samples........................................... 128
6.5 Conclusions.................................................................................................... 130
References............................................................................................................... 131
6.1 Introduction
Marine toxins produced by harmful algae are listed among the most important caus-
ative agents of poisoning episodes involving seafood consumers.1,2 Toxic incidents are
usually associated with blooms of toxigenic plankton species in shellfish-producing
105

regions. The phenomenon may take many forms, ranging from massive “red tides”
or blooms of cells that discolor the water to dilute, inconspicuous concentrations of
plankton cells that get noticed only because of the harm caused by the highly potent
toxins they produce.3 Filter-feeding mollusks consume the plankton, accumulate
toÂ�xins in their edible tissues, and thus become the part of the food chain responsible
for transmission of toxicity to humans.
The poisoning syndromes have been named paralytic, diarrhetic, neurotoxic, and
amnesic shellfish poisoning (PSP, DSP, NSP, and ASP) based on major symptoms
they induce in humans.4 The causative toxins range from polar, low-molecular-weight
(MW) compounds to high-MW lipophilic substances; most are thermally labile and
some are pH, oxygen, and light sensitive. Basically, toxicity is the only common fac-
tor among different toxins. So it is not surprising that the AOAC mouse bioassay is
the most widely used method for toxin detection in a regulatory setting.5 It provides
two different extraction procedures for either lipophilic or hydrophilic toxins and
subsequent intraperitoneal injection of aliquots of each potentially toxic shellfish
extract into three 20â•›g mice. An observation time follows to determine symptoms
and time-to-death, which correlates with the amount of toxin contained in the sam-
ple. Although this approach provides a single integrated response from all the toxins,
it suffers from poor reproducibility, low sensitivity, and interferences from other
components in the extract. Most importantly, it cannot single out the causative toxins
and variation in toxin profiles cannot be monitored. Thus, instrumental methods are
required to identify and monitor the toxins.
The analytical technique must provide low detection limits (μg/kg), high-
resolution separation, and/or high detection selectivity to deal with complicated
sample matrices. In addition, since a contaminated sample may contain several
toÂ�xins from within a class or even different classes of toxins, analytical methods that
allow the combined analysis of assorted toxins are highly desirable. The combina-
tion of liquid chromatography and mass spectrometry (LC-MS) meets all the above
requirements and has shown great potential both in regulatory and research settings.
It is one of the most powerful tools for detection and quantitation of toxins in plank-
ton and shellfish samples, even at trace levels, the identification of new compounds,
and the investigation of toxin production and metabolism.6 A number of LC-MS
methods have been developed for all of the known toxins.7,8 Most of them are based
on the use of reversed phase liquid chromatography (RPLC), which works well for
the analysis of lipophilic toxins9 but suffers from a number of drawbacks when very
polar and hydrosoluble compounds have to be analyzed (see the following discus-
sion). A new approach based on the use of hydrophilic interaction liquid chroma-
tography (HILIC) coupled with electrospray ionization tandem mass spectrometry
(ESI-MS/MS) has been recently proposed for sensitive and selective detection of
very polar and ionic compounds belonging to the paralytic shellfish poisoning (PSP)
toxin class,10 even in combination with domoic acid (DA),11 the causative toxin of
the amnesic shellfish poisoning (ASP) syndrome, and assorted cyanobacterial toxins
(Figure 6.1).12 Details on HILIC-MS method development will be provided in the
following sections.

Hydrophilic Interaction Liquid Chromatography–Mass Spectrometry 107
R1 R2 R3 R4 Toxin
H H H O NH2 STX
H H OSO3– GTX2
R4 H OSO3– H O GTX3
OH H H NEO
R1 H OH H OSO3– GTX1
6 NH OH OSO3– H GTX4
N1 7
NH2+ H H OH 11α-OH-STX
9
3
NH H OH H 11β-OH-STX
H2N+ N
10 12 OH H H H B1 (GTX5)
OH H H OSO3– C1
11 O NHSO3–
R2 R3 H OSO3 –
H C2
OH H H O B2 (GTX6)
OH H OSO3– C3
STX = Saxitoxin OH OSO3– H C4
NEO = Neosaxitoxin
GTX = Gonyautoxins H H H dcSTX
H H OSO3– dcGTX2
H OSO3 – H OH dcGTX3
OH H H dcNEO
OH H OSO3– dcGTX1
OH OSO3– H dcGTX4
H O Me
COOH N
N COOH ATX-a
COOH H
DA
O H
N O
Me H R
N Me N H
R H
H H HO2C O
O OSO3– H Me HN O
N+ HN H
HN N HN O
Me H CO2H Me
H Me H HN O
OH N N
H H Me R= Me
H
R = OH CYN O Microcystin-LR
R = H doCYN OMe H
Me
H HN N NH2
Me R=
H HN NH2 NH
Microcystin-RR
Figure 6.1 Structures of principal paralytic shellfish poisoning (PSP) belonging to saxi-
toxin (STX), neosaxitoxin (NEO), and gonyautoxin (GTX) series, domoic acid (DA), anatoxin-
a (ATX-a), cylindrospermopsin (CYN) and its deoxy derivative (doCYN), and microcystins.

6.2 Paralytic Shellfish Poisoning Toxins

Saxitoxin (STX) and its analogues, also known as paralytic shellfish poisoning (PSP)
toxins, are fast-acting neurotoxins that pose serious risks to public health and have
a significant economic impact on the shellfish industry throughout the world. They
are produced by marine dinoflagellates belonging to Alexandrium, Pyrodinium, and
Gymnodinium genera13 and, generally, cause persistent problems to humans due to
their accumulation in filter-feeding mollusks. The PSP syndrome is characterized by
neurological distress, which typically appears within 15–30â•›min following consump-
tion of contaminated shellfish and can result in death, if the toxin concentration is
sufficiently high.
Structurally, PSP toxins are tetrahydropurine derivatives and can be divided into
three groups: (1) carbamoyl (STX, GTX2, GTX3, NEO, GTX1, GTX4, 11α,β-OH-
STX), (2) N-sulfocarbamoyl (B1, C1, C2, B2, C3, C4), and (3) decarbamoyl toxins
(dcSTX, dcGTX2, dcGTX3, dcNEO, dcGTX1, dcGTX4), based on the nature of
the side chain (Figure 6.1). They are potent, reversible blockers of voltage-activated
sodium channels on excitable cells,14 but, due to the differences in charge state and
substitution groups to the basic STX structure, they bind with different affinities
to site 1 of sodium channels resulting in different toxicities. The carbamoyl toxins
are the most toxic and the N-sulfocarbamoyl derivatives are the least toxic.15 Thus,
health risks can be reliably assessed just if the level of each toxin is individually
determined.
The AOAC mouse bioassay is widely used for detecting the presence of PSP
toxins in routine monitoring of shellfish.5 In the extraction procedure used, the
N-sulfocarbamoyl toxins are converted to the corresponding carbamoyl toxins.
Alternative assays based either on in vitro cell toxicity, receptor binding, or immu-
nological response have been developed.8 None of the above assays provides detailed
information on the toxin profile in samples, nor can they be used for the precise
quantitative analysis of samples containing variable levels of individual toxins. So,
positive results generally require instrumental confirmation in critical situations.
PSP toxins pose a real challenge to the development of instrumental methods for
their detection and quantitation since they include a great variety of closely related
structures with three different charge states (0, +1, and +2). They are nonvolatile and
thermally labile and lack any useful chromophore, which eliminate most traditional
chromatographic techniques such as gas chromatography (GC) and liquid chroma-
tography (LC) with ultraviolet (UV) detection.
The most used technique for the analysis of PSP toxins is ion-pair LC coupled with
postcolumn oxidation and fluorescence detection (LC-ox-FLD),16 which provides
conversion of PSP toxins into fluorescent derivatives under alkaline conditions. The
technique is highly sensitive, but interfering compounds can make difficult interpre-
tation of results.17 An alternative method is the precolumn oxidation approach (ox-
LC-FLD).18 This method is rapid, sensitive, and fully automated but interpretation of
quantitative results is complex because some toxins give the same oxidation product,
while others give two or three products.19
Because of the efficient ionization of STX and its analogues when they are injected
into a mass spectrometer (MS) equipped with an electrospray ionization (ESI)

source,20 a number of methods that use ESI-MS as detector for either a capillary
electrophoresis (CE) or a liquid chromatography (LC) device have been developed.
Although CE-MS is ideally suited to the analysis of the highly charged PSP
toÂ�xins,21 it is not possible to analyze all the toxins in a single analysis due to their
different charge states. In addition, the technique suffers from interference from
coextracted salts.
The nature of LC stationary phase is critical for developing an efficient LC-MS
method for the determination of PSP toxins. An efficient approach based on ion
exchange liquid chromatography with MS detection has been recently proposed by
Jaime et al.22 but the most common reverse-phase LC provides sufficient retention of
the highly polar PSP toxins only if ion-pairing reagents (heptanesulfonic acid, hepta-
fluorobutyric acid, or tetrabutylammonium sulfate) are added to the mobile phase.16
Such agents result in serious interference with MS detection by causing suppression
of ionization and ion source contamination. In addition, the neutral C toxins have to
be analyzed in a separate run from mono- and bi-charged toxins.
The combination of HILIC with ESI-MS/MS detection demonstrated to help
overcome drawbacks of the above-mentioned assays and instrumental methods. The
technique allows simultaneous determination of all PSP toxins with a high degree
of selectivity and eliminates the need for further confirmation. The mobile phase
does not use ion-pair agents, so that it does not reduce ionization efficiency. Further,
the mobile phase has high amounts of organic modifier so that ionization yield is
enhanced. All details of the HILIC-MS method are summarized in the following
section along with its application to plankton and shellfish samples.
6.2.1 HILIC-MS Determination of PSP Toxins

A key role in setting up the HILIC-MS method for detection of PSP toxins is played
by MS behavior of each saxitoxin-like compound.
In Table 6.1, MS and MS/MS data for most PSP toxins are summarized. Such
data were acquired on three PE-Sciex MS instruments, namely a single and a triple
quadrupole MS equipped with an ESI source (API-165 and API-III+, respectively)
and a triple quadrupole MS equipped with a Turbospray• interface (API-4000).
STX, NEO, and their decarbamoyl derivatives (dcSTX and dcNEO), which exist
as dications in solution, did not produce doubly charged ions in ESI-MS but gave
[M + H]+ ions as the base peak in their MS spectra, with a small fragment ion cor-
responding to the loss of a water molecule.
A different fragmentation pattern was observed with the two epimeric pairs of
gonyautoxins, GTX1 and GTX4 and GTX2 and GTX3. The [M + H]+ ion was still
the most abundant ion in the MS spectra of GTX1 and GTX2, which present an
hydroxysulfate group at C-11 in an α-orientation, whereas an [M + H-80]+ frag-
ment ion, corresponding to loss of SO3, dominated the spectra of 11β-hydroxysulfate
gonyautoxins (GTX3 and GTX4). The same behavior was observed for the decar-
bamoyl gonyautoxins, dc-GTX1-4.
The spectra of N-sulfocarbamoyl derivatives (B1, C1, C2, B2, C3, C4) var-
ied strongly with instrument type. Orifice voltage settings of 10 and 50â•›V on the
API-165 and API-III+ instruments, respectively, and a declustering potential of

Table 6.1
Mass Spectral Data of Paralytic Shellfish Poisoning (PSP) Toxins
MS Spectra (%RI)a MS/MS Spectra (%RI)b
Toxin [M + NH4â•›]+ [M + H]+ Fragment Precursor Product Loss of
STX 300 (100) 282 (10) 300 (40) 204 (100) − 2H2O − NH3 − NHCO
138 (75) − H2O − NH3 − CO2 − HNCNH − H2C2NH
179 (60) − H2O − NH3 − CO2 − HNCNH
186 (45) − 3H2O − H3 − NHCO
282 (40) − H2O
221 (25) − H2O − NH3 − CO2
GTX2 396 (5) 316 (100) 396 (0) 316 (100) − SO3
298 (10) − SO3 − H2O
GTX3 396 (100) 298 (38) 396 (0) 298 (100) − SO3 − H2O
316 (27) 316 (10) − SO3
378 (23) 220 (10) − SO3 − 2H2O − NH3 − NHCO
NEO 316 (100) 298 (12) 316 (100) 220 (68) − 2H2O − NHCO − NH3
138 (65) − H2O − NH3 − CO2 − HNCNH − H2C2NOH
298 (62) − H2O
177 (60) − 2H2O − NH3 − CO2 − HNCNH
237 (42) − H2O − NH3 − CO2
GTX1 412 (8) 332 (100) 412 (0) 332 (100) − SO3
314 (15) − SO3 − H2O
GTX4 412 (100) 394 (55) 412 (0) 314 (100) − SO3 − H2O
332 (12) 332 (10) − SO3
314 (9) 253 (5) − SO3 − H2O − NH3 − CO2
11OH-STX 316 (100) 316 (27) 148 (100) Not assigned
108 (50) Not assigned

Hydrophilic Interaction Liquid Chromatography–Mass Spectrometry
220 (45) − 2H2O − NHCO − NH3
298 (40) − H2O
196 (38) − H2O − NHCO − NH3 − NHCNH
237 (10) − 2H2O − NHCO
B1 (GTX5) 380 (100) 300 (98) 380 (0) 300 (100) − SO3
282 (10) 282 (38) − SO3 − H2O
257 (8) 204 (30) − SO3 − 2H2O − NHCO − NH3
221 (18) − SO3 − H2O − CO2 − NH3
C1 493 (50) 476 (18) 396 (100) 493 (0) 316 (100) − NH3 − 2SO3
413 (38) 396 (20) − NH3 − SO3
316 (25) 298 (2) − NH3 − 2SO3 − H2O
C2 493 (45) 476 (15) 396 (100) 493 (0) 298 (100) − NH3 − 2SO3 − H2O
378 (50) 316 (68) − NH3 − 2SO3
413 (42) 378 (30) − NH3 − SO3 − H2O
316 (5) 396 (12) − NH3 − SO3
B2 (GTX6) 396 (100) 316 (30) 396 (0) 316 (100) − SO3
220 (5) − SO3 − 2H2O − NH3 − NHCO
298 (4) − SO3 − H2O
237 (3) − SO3 − 2H2O − NHCO
177 (3) − SO3 − 2H2O − NH3 − HNCNH − CO2
C3 509 (12) 492 (17) 412 (100) 509 (0) 332 (100) − NH3 − 2SO3
332 (11) 412 (22) − NH3 − SO3
314 (13) − NH3 − 2SO3 − H2O
394 (2) − NH3 − SO3 − H2O
C4 509 (10) 492 (8) 412 (100) 509 (0) 314 (100) − NH3 − 2SO3 − H2O
332 (1) 394 (95) − NH3 − SO3 − H2O
332 (77) − NH3 − 2SO3
412 (38) − NH3 − SO3
111
(continued)

Mass Spectral Data of Paralytic Shellfish Poisoning (PSP) Toxins
MS Spectra (%RI)a MS/MS Spectra (%RI)b
Toxin [M + NH4] + [M + H] + Fragment Precursor Product Loss of
dcSTX 257 (100) 239 (17) 257 (20) 126 (100) Not assigned
138 (65) − H2O − NH3 − 2NHCNH
222 (50) − H2O − NH3
180 (48) − H2O − NH3 − NHCNH
156 (45) − NH3 − 2NHCNH
239 (25) − H2O
dcGTX2 353 (4) 273 (100) 353 (0) 273 (100) − SO3
255 (15) − SO3 − H2O
238 (8) − SO3 − H2O − NH3
196 (8) SO3 − H2O − NH3 − NHCNH
dcGTX3 353 (100) 335 (55) 353 (0) 255 (100) − SO3 − H2O
273 (42) 196 (25) − SO3 − H2O − NH3 − NHCNH
255 (30) 273 (12) − SO3
238 (12) − SO3 − H2O − NH3
335 (7) − H2O
dcNEO 273 (100) 255 (20) 273 (40) 126 (100) Not assigned

Hydrophilic Interaction Liquid Chromatography–Mass Spectrometry
255 (31) − H2O
220 (25) − 2H2O − NH3
dcGTX1 369 (5) 289 (100) 369 (0) 289 (100) − SO3
271 (13) − SO3 − H2O
dcGTX4 369 (100) 289 (1) 369 (0) 271 (100) − SO3 − H2O
289 (15) − SO3
351 (10) − H2O
Source: Modified from Dell’Aversano, C. et al., J. Chromatogr. A, 1081, 190, 2005. With permission from Elsevier.
a MS spectra were obtained using a PE-SCIEX API-165 mass spectrometer. Percentage relative intensities (%RI) are reported in brackets.
b MS/MS spectra were obtained using a PE-SCIEX API-4000 mass spectrometer. Percentage relative intensities (%RI) are reported in brackets.
113
50â•›V on the API-4000 instrument provided minimum fragmentation of analytes

while still maintaining a good background spectrum. The API-165 instrument
gave a higher degree of fragmentation than the API-III+ and the API-4000, which
resulted in a lower sensitivity. The Turbospray interface, which applies heated
nitrogen to assist nebulization, gave more fragmentation if too high temperatures
were used. Thus, a minimum temperature (275°C) had to be used to minimize
temperature-induced fragmentation. Under such conditions, B1 gave a spectrum
with abundant [M + H]+ and [M + H − 80]+ ions, while C1-4 gave both protonated
and ammoniated ions, as well as associated fragment ions due to loss of SO3 and/
or H2O molecules.
The protonated molecules or adduct ions and the main fragment ions of each toxin
were selected for select ion monitoring (SIM) experiments, taking into account that,
for some toxins, fragment ions caused by elimination of SO3 from [M + H]+ ion could
potentially interfere with protonated molecules of other compounds (e.g., B1 with
STX; C1, C2, GTX2, GTX3 and B2 with each other and NEO). This actually reduced
the number of ions required for SIM, thus increasing sensitivity, but it also posed the
need for a good chromatographic separation of the toxins.
Since multiple reaction monitoring (MRM) experiments can provide more selec-
tive detection of compounds, MS/MS spectra of PSP toxins were also examined. The
spectra of selected toxins are shown in Figure 6.2 and the results are summarized in
Table 6.1 for all compounds tested.
Significant differences were observed between fragmentation patterns of the epi-
meric pairs GTX2 and GTX3, GTX1 and GTX4, dcGTX2 and dcGTX3, dcGTX1
and dcGTX4, C1 and C2, and C3 and C4. Thus, more than one transition for each
of these toxins had to be selected for MRM experiments. The recommended MRM
transitions for all compounds are shown in Table 6.2.
The HILIC behavior of PSP toxins was investigated by analyzing a standard mix-
ture containing STX, GTX2, GTX3, NEO, GTX1, GTX4, B1, B2, C1, C2, C3, C4,
11-OH-STX, dcSTX, dcGTX2, dcGTX3, dcNEO, dcGTX1, and dcGTX4, in com-
parable amounts under various conditions. The most abundant ions (Table 6.1) and
transitions (Table 6.2) were selected for SIM or MRM experiments, respectively.
Two HILIC stationary phases were considered, namely TSK-Gel® Amide-80 and
polyhydroxyethyl aspartamide (PHEA), since they had been reported to retain gua-
nine, a compound very similar to PSP toxins, at various buffer concentrations.23
The PHEA column provided a good separation between epimeric pairs (C1 and C2,
GTX2 and GTX3, GTX1 and GTX4) when using water as eluent A and acetonitrile/
water (95:5) as eluent B, both containing 2.0â•›m M ammonium formate, 3.6â•›m M for-
mic acid, pH 3.5. However, selective detection of the major PSP toxins was possible
only over a 120â•›min analysis time with poor peak shape (gradient: 90% B for 80â•›min,
90%–65% B over 15â•›min and hold 25â•›min). Shorter analysis time (35â•›min) resulted in
poor separation between potentially interfering compounds, namely C1/GTX2 and
STX/B1 (gradient: 80%–65% B over 20â•›min and hold 15â•›min). Several mobile phase
systems were tried but separation selectivity could not be improved. Furthermore,
extreme column bleed was observed at pH 2.5 to 3.5 in the mass range 200–500,
which greatly hampered the determination of PSP toxins. Due to background and
selectivity problems, no further development on the PHEA column was considered.

100 STX 204

138 300 [M + H]+
179
50 186 282 (precursor)
221
0
100
NEO 316 [M + H]+
138 220 298
177 (precursor)
50 237
0
100
GTX2 316
298 396 [M + H]+
50
(precursor)
0
100 298
GTX3
396 [M + H]+
50
(precursor)
196 220 316
0
Relative intensity (%)
100 332
GTX1
50 412 [M + H]+
(precursor)
314
0
100 314
GTX4
50 412 [M + H]+
(precursor)
196 253 332
0
100 300
B1
380 [M + H]+
50 204 282 (precursor)
221
0
100
C1 316
493 [M + NH4]+
298
50 (precursor)
396
273
0
100 298
C2
316 493 [M + NH4]+
50
378 (precursor)
396
0
100 150 200 250 300 350 400 450 500
m/z
Figure 6.2 MS/MS spectra of selected PSP toxins obtained on the API-4000 system
using a collision energy of 35â•›V. The [M + H]+ ions of STX, NEO, GTX2, GTX3, GTX1,
GTX4, B1, and the [M + NH4]+ ions of C1 and C2 were used as precursor ions. Assignments
of labeled product ions are shown in Table 6.1. (Modified from Dell’Aversano, C. et al.,
J. Chromatogr. A, 1081, 190, 2005. With permission from Elsevier.)

Table 6.2
Recommended MRM Transitions for HILIC-MS Analysis of Major PSP Toxinsa
Toxin RT (min)b RRT (min)b m/z > m/z m/z > m/z m/z > m/z m/z > m/z m/z > m/z
STX 20.3 2.82 300 > 282 (100) 300 > 204 (80)
GTX2 9.6 1.33 396 > 316 (100) 316 > 298 (30) 316 > 220 (10) 396 > 298 (5)
GTX3 10.7 1.49 396 > 298 (100) 396 > 316 (20) 316 > 298 (10)
NEO 21.0 2.91 316 > 298 (100) 316 > 220 (50)
GTX1 9.8 1.36 412 > 332 (100) 412 > 314 (5)
GTX4 10.9 1.51 412 > 314 (100) 412 > 332 (10)
11OH-STX 24.9 3.46 316 > 298 (100) 316 > 220 (30)
B1 (GTX5) 13.1 1.82 380 > 300 (100) 300 > 282 (5) 300 > 204 (5)
C1c 7.2 1.00 493 > 316 (100) 396 > 316 (65) 396 > 298 (15) 316 > 298 (5) 493 > 298 (3)
C2c 8.0 1.11 396 > 298 (100) 493 > 298 (99) 493 > 316 (45) 316 > 298 (5) 396 > 316 (3)
B2 (GTX6) 14.6 2.03 396 > 316 (100) 396 > 298 (20) 316 > 298 (5) 316 > 220 (5)
C3c 7.9 1.10 509 > 332 (100) 412 > 332 (50) 412 > 314 (13) 509 > 314 (8)
C4c 8.8 1.22 412 > 314 (100) 509 > 314 (46) 509 > 332 (19) 412 > 332 (1)
dcSTX 21.1 2.93 257 > 239 (100)
dcGTX2 10.2 1.33 353 > 273 (100) 273 > 255 (30) 353 > 255 (15)
dcGTX3 11.3 1.57 353 > 255 (100) 353 > 273 (15) 273 > 255 (5)
dcNEO 20.8 2.89 273 > 255 (100)
dcGTX1 10.1 1.40 369 > 289 (100) 369 > 271 (15)
dcGTX4 11.4 1.58 369 > 271 (100) 369 > 289 (10)
a Reported data for most PSP toxins were obtained using a PE-SCIEX API-III+ MS. Percentage relative intensity is provided for each transition.
The most abundant transitions (100) are recommended for quantitative studies.
b Retention times (RT) and retention time relative to C1 (RRT) are referred to the optimized chromatographic conditions (Table 6.3).
c Data for C1-4 toxins were obtained using a PE-SCIEX API-4000 mass spectrometer. The [M + H]+ or [M + NH4]+ ions for C toxins were not
sufficiently abundant on the API-III+ MS system, in which case transitions associated with the [M − SO3 + H]+ ions were used for quantitation.

The 5â•›μm Amide-80 column was reported to have stability from pH 2 to 7.5 and,
most importantly, showed no significant bleed at low pH. This column provided supe-
rior retention within the chromatographic window and was therefore employed for
further optimization studies. The eluting system proposed by Strege23 for the Amide-
80 column was used initially (gradient: 90%–60% B over 20â•›min, hold for 60â•›min,
with A being water and B acetonitrile/water (95:5), both containing 6.5â•›m M ammo-
nium acetate, pH 5.5). Under these conditions, the early eluting peaks of GTX1-4
showed tailing and later eluting peaks of STX and NEO showed fronting. A num-
ber of eluting systems were tested on this column by paying particular attention to
organic modifier character and percentage, buffer character and percentage, and pH.
As found by Yoshida for the separation of peptides,24 the percentage of organic
modifier in the mobile phase played a key role for the absolute retention time of
PSP toxins. HILIC behaves like normal phase chromatography, so retention times
increase proportional to the percentage of organic modifier and to the polarity of the
solute. The neutral C toxins eluted first, followed by the single-charged gonyautoÂ�xins,
then the double-charged STX and NEO, and finally the decarbamoyl derivatives.
Both acetonitrile and methanol were tested as possible organic modifiers.
Acetonitrile provided sharper peaks than methanol and was thus preferred. Methanol
resulted in a dramatic change in the relative order of elution and poorer separation.
The ammonium formate buffer concentration significantly affected retention time
in the range 0–2â•›m M (>3â•›min) while about 1â•›min shifts were observed for buffer con-
centration 2–10â•›m M. Generally, in agreement with findings by Strege for the reten-
tion of guanine,23 retention times decreased as the concentration of buffer increased.
However, when no aqueous buffers were used, PSP toxin retention times exceeded
180â•›min and the peak widths were unacceptably broad. Therefore, an aqueous buffer
was required to modify the mobile phase. Both ammonium acetate and formate were
tested. The latter resulted in better peak shape and was thus preferred.
The pH of the mobile phase had the greatest influence on the separation of PSP
toxins. As pH increased, retention times and separation selectivity increased. At
low pH, the peaks were broader. The neutral C toxins did not change their absolute
retention time when the pH was changed whereas the other toxins showed large
changes in retention time (1–2â•›min) for relatively small changes in pH (0.1–0.2â•›pH
units). Therefore, the retention times relative to C1 were also calculated (Table 6.2).
The HILIC column efficiency was dependent on flow rate and varied between
toxins. Initial work was performed on a 250 × 4.6â•›mm i.d. Amide 80 column. The
highest plate number was generally obtained at flow rates 0.8–1â•›m L/min. The later
eluting toxins, namely STX and NEO, showed higher performance at 0.6–0.8â•›m L/
min but the greatly increased analysis time was considered a serious drawback.
Therefore, a flow rate of 1â•›m L/min was selected. For the 250 × 2â•›mm i.d. column, a
0.2â•›m L/min flow was selected.
The effect of column temperature on retention of PSP toxins in the range 10–45°C
was slight. In particular, retention times of monocharged (GTX1-4 and their decar-
bamoyl derivatives, B1 and B2) and bicharged (STX, NEO, and their decarbamoyl
derivatives) toxins increased as a function of the temperature. No effect was observed
upon retention of the neutral C toxins. Similarly, selectivity increased linearly with

Table 6.3
Optimized Chromatographic Conditions for the HILIC-MS Analysis of PSP
Toxins, DA, and Assorted Cyanobacterial Toxinsa
Column 5â•›μm amide-80 column (250 × 2â•›mm i.d.)
Column T (°C) 20
Flow rate (mL/min) 0.2
Mobile phase A = water, 2.0â•›mM ammonium formate, 3.6â•›mM formic acid (pH 3.5)
B = acetonitrile/water (95:5), 2.0â•›mM ammonium formate, 3.6â•›mM formic acid
Elution system A Isocratic, 65% B
Elution system B Isocratic, 75% B
Elution system C Gradient
75% B for 5â•›min
75%–65% B over 1â•›min, hold 13â•›min
65%–45% B over 4â•›min, hold 10â•›min
a Elution system A is recommended for PSP toxins analysis, whereas elution systems B and C are indi-
cated for determination of DA, CYN, doCYN, and ATX-a as single compounds or in combination with
PSP toxins, respectively.
temperature for all PSP toxins except neutral ones. These results supported a pre-
sumed role of ion exchange in the HILIC separation of PSP toxins, as retention and
selectivity are known to be temperature dependent in ion exchange chromatogra-
phy.25 In addition, as temperature increased, resolution of most peaks increased but
the peak shapes of later eluting toxins broadened. A column temperature of 20°C
was selected as optimum for retention, selectivity, and peak shape.
On the basis of all the above findings, the best chromatographic conditions were
selected (Table 6.3). Elution system A (65%B) allowed determination of all STX-like
compounds in a single chromatographic run of 30â•›min. HILIC-MS analyses in SIM
and MRM modes of a standard mixture of PSP toxins under the optimized HILIC
conditions are shown in Figure 6.3.
Some compounds could not be chromatographically resolved: GTX1 and GTX2,
GTX4 and GTX3, and the associated decarbamoyl derivatives eluted in a 2â•›min
range; C1 and C2 substantially coeluted with C3 and C4; B1 partially coeluted with
B2; STX partially coeluted with NEO, and dcSTX with dcNEO. However, the addi-
tional detection selectivity provided by different channels of detection in SIM or
MRM made it possible to individually detect all PSP toxins in a reasonable period
of time, namely 25–30â•›min.
As for the mechanism of separation, HILIC is considered to behave similarly
to normal phase partition chromatography, where a stagnant mobile phase (mostly
aqueous) is in contact with the stationary phase and a dynamic mobile phase
(mostly organic) is separated from the stationary phase.26 The analyte partitions
between the two mobile phases and may be orientated in space to interact with func-
tions of the stationary phase. Ion exchange or electrostatic interactions can also have
a role.27,28 As for PSP toxins, the mechanism of separation appears to be primarily

dcSTX dcSTX
m/z m/z > m/z

257 257 > 239
dcGTX2 dcGTX2
dcNEO
dcGTX3 dcGTX3 dcNEO
273 353 > 273
353 273 > 255
dcGTX1 dcGTX4
dcGTX1
dcGTX4 ?
289 369 > 289
369
GTX1
GTX1
GTX4
C3 C4 GTX4
412 > 314 C3C4
332
412 412 > 332
STX B1
B1 STX
380 > 300
300 300 > 282
380 300 > 204
NEO GTX2
GTX2
C1 GTX2
NEO
11-OH-STX
B2 C2 B2
C1 C2 GTX3 396 > 316
316 396 > 298
396 316 > 298
0 5 10 15 20 25 0 5 10 15 20 25 30
(a) Time (min) (b) Time (min)
Figure 6.3 HILIC-MS analyses of a standard mixture of assorted PSP toxins in SIM
(a) and MRM (b) modes under the optimized HILIC conditions (Table 6.3, elution system A).
SIM experiment was carried on an API-165 MS system by selecting protonated and/or frag-
ment ions (Table 6.1). MRM experiment was carried out on API-III+ MS system by selecting
transitions consistent with the fragmentation pattern of each toxin (Table 6.2). (Reprinted from
Dell’Aversano, C. et al., J. Chromatogr. A, 1081, 190, 2005. With permission from Elsevier.)
electrostatic in nature. This is supported by the short retention times of the neutral C
toxins, the intermediate retention times of the single-charged gonyautoxins, and the
long retention times of the double-charged toxins. The excellent separation achieved
between epimeric pairs such as GTX1 and GTX4, or GTX2 and GTX3, may be
explained by considering the effect that α- or β-orientation of the 11-hydroxysulfate
function has on the charge states of individual functional groups. Molecular model-
ing showed that when the 11-hydroxysulfate group is in the α-orientation (GTX1 and
GTX2), it can establish an intramolecular interaction with the guanidinium func-
tion at C-8. This would reduce the number of positively charged functions on the
molecule available for interaction with the stationary phase. On the other hand, both
guanidinium groups are available for interaction with the stationary phase when the
11-hydroxysulfate function is β-oriented (GTX3 and GTX4). It should, also, be noted
that 11(α,â•›β)-OH-STX do not resolve under the same conditions. This supports the
argument that the 11-hydroxysulfate group and its interaction with the guanidinium
function is important for separation of epimers.

Table 6.4
Limits of Detection (LOD, S/N = 3) for Major PSP Toxins on Different
LC-MS Systems in the SIM and MRM Modes
SIMa SRMa
API-165 API-III+ API-4000 LC-ox-FLD16
LOD LOD LOD LOD
Toxin m/z (nM) m/z > m/z (nM) m/z > m/z (nM) (nM)
STX 300 800 300 > 282 7000 300 > 204 20 60
GTX2 316 300 396 > 316 1000 396 > 316 20 20
GTX3 396 400 396 > 298 300 396 > 298 10 5
NEO 316 900 316 > 298 7000 316 > 220 30 60
GTX1 332 200 412 > 332 800 412 > 332 10 20
GTX4 412 800 412 > 314 400 412 > 314 5 30
B1 380 1000 380 > 300 700 380 > 300 10 100
C1 396 50 396 > 316 40 493 > 316 20 30
C2 396 60 396 > 298 50 396 > 298 10 20
Source: Modified from Dell’Aversano, C. et al., J. Chromatogr. A, 1081, 190, 2005. With
permission from Elsevier.
a Optimized chromatographic conditions (Table 6.3) were used.
Limits of detection (LOD) of the HILIC-MS method for matrix-free toxins were
obtained for STX, GTX2, GTX3, NEO, GTX1, GTX4, B1, C1, and C2, and ranged
from 50 to 1000â•›nM in SIM and 40 to 7000â•›nM in MRM mode on the API-III+
instrument (Table 6.4). Unfortunately, a single collision energy was used for the
entire group of MRM ions on the API-III+ system. The compromise value of 20â•›V
was too low for good fragmentation of STX and NEO, resulting in poor detection
limits. LOD values for such toxins could be lowered by at least fivefold by using a
30â•›V collision energy on [M + H]+ ions of STX and NEO, which is possible if time
programming of MRM transitions is used. Much better results in terms of LOD were
obtained on the API-4000 MS system, which presents better sensitivity and the abil-
ity to use optimized collision energies for each ion transition. Particularly LOD val-
ues ranged from 5 to 100â•›nM (Table 6.4) and excellent linearity was obtained, with
R2 values of 0.999 or greater. A comparison with LOD for the LC-ox-FLD method16
(Table 6.4) reveals that the HILIC-MS method on the API-4000 system has better
sensitivity than the LC-ox-FLD method and should therefore be suitable for analysis
of shellfish samples near the maximum acceptable regulatory limits for PSP toxins
(0.8â•›mg of saxitoxin equivalent per kg of edible tissue).
6.2.2 Application to Plankton and Shellfish Samples

The HILIC-MS method for PSP toxins was tested on field plankton and mussel sam-
ples collected during an intense bloom of Alexandrium tamarense, which occurred in

June 2000 in a Nova Scotian harbor.29 Extracts of the samples were analyzed in both
SIM and MRM modes. Although SIM mode proved suitable for detection of most of
the PSP toxins present in the samples, unambiguous interpretation of the results was
hampered by the presence of many interfering peaks, a high background signal in
some ion traces, and a matrix-related hump in the chromatograms at about 20â•›min. The
higher selectivity of the MRM mode made interpretation of the results much easier due
to elimination of signals from other coextractives. Anyway, a slight shift of retention
times for the sample extract versus those for the standards could be observed. This
was due to matrix effect that got worse as more concentrated crude extracts were used.
Figure 6.4 shows the results of the HILIC-MRM analyses of the crude extracts of
the plankton sample (Figure 6.4a) and the mussel extract (Figure 6.4b). The plankton
sample showed a complex array of toxins with the major toxins being GTX4, C2, C4,
GTX3, B1, NEO, and STX. Only low levels of the corresponding epimeric C and
GTX toxins were observed. The same toxins were observed in the mussel extract,
which was not surprising, as these mussels had consumed the plankton material.
Interestingly, some new saxitoxin analogues were observed in the mussel sample.
The new compounds, labeled as M1, M2, and M3, were absent in the plankton sample
and, thus, were likely metabolites and/or degradation products formed in the mus-
sel. The M2 peak had an exact match of retention time and product ion spectrum for
11-hydroxy-STX. Isolation work, which included the use of preparative HILIC-MS,
followed by in-depth MS and NMR investigation allowed to elucidate the structure
of M1 and M3, as 11β-hydroxy-N-sulfocarbamoyl saxitoxin and 11,11-dihydroxy-N-
sulfocarbamoyl saxitoxin.30
GTX4
GTX4
GTX1 GTX1
m/z > m/z C4 m/z > m/z C4 M3
412 > 314
×1 412 > 314 ×1
412 > 332 412 > 332
B1 B1
STX STX
380 > 300 ×4 380 > 300 ×4

300 > 282 380 > 282
300 > 204 300 > 204
C2 M1
GTX2
GTX2 GTX3
GTX3 C2 M2
C1 NEO
NEO (11-OH-STX)
396 > 316 ×2 C1 ?
396 > 316 ×2
396 > 298 396 > 298
316 > 298 316 > 298
0 5 10 15 20 25 0 5 10 15 20 25
Figure 6.4 HILIC-MS analyses in MRM mode of an Alexandrium tamarense extract

(a) and of a Mytilus edulis extract (b) containing various PSP toxins. Experiments were car-
ried out on API-III+ MS under the optimized HILIC conditions (Table 6.3, elution system A).
Some traces are plotted with expanded scale as indicated. (Reprinted from Dell’Aversano, C.

6.3 Domoic Acid

Domoic acid (DA) (Figure 6.1) is the causative toxin of amnesic shellfish poisoning
(ASP) syndrome.31 Symptoms of the intoxication include nausea, vomiting, gastro-
enteritis, cramps, and diarrhoea, all within 24â•›h after consumption of contaminated
shellfish. Neurological complaints develop within 48â•›h and include ataxia, confusion,
persistent short-term memory loss, disorientation, and even coma.
DA is produced by Pseudo-nitzschia pungens f. multiseries,32 a diatom widely
distributed in coastal waters all around the world, as well as by several species of
Pseudo-nitzschia (Pseudo-nitzschia australis, Pseudo-nitzschia pseudo-delicatissima,
Pseudo-nitzschia galaxiae),33–35 although there is considerable variability in toxicity
among clones and over time in culture.
The AOAC mouse bioassay for PSP toxins can be employed for detection of DA’s
unique toxicity,36 since this method involves boiling shellfish tissue with an equal
volume of 0.1â•›N HCl, which provides a common extract for screening of both DA
and PSP toxins. Intraperitoneal injection of DA in mice induces a very characteristic
symptomology, namely, scratching of shoulders by the hind leg, sedation-akinesia,
rigidity, loss of posteral control, convulsions, and death.37 However, such symptoms
may be observed only when domoic acid is present at levels >40â•›μg/g edible tissue,
whereas the regulatory limit for DA is 20â•›μg/g shellfish tissue. Therefore, the relative
insensitivity of this assay precludes its use for regulatory purposes and instrumental
confirmation is always needed.
Liquid chromatography with UV detection (LC-UV) is the preferred analytical
technique for determination of DA in shellfish.38,39 Mollusk tissues can be extracted
by the above AOAC procedure, or by aqueous methanol (1:1, v/v),39 which gives a
better yield and a more stable extract. LOD are in the range 0.1–1â•›μg DA/g tissue,
depending on the sensitivity of the UV detector, which is suitable for regulatory
purposes. However, interferences are commonly encountered that can result in false
positives. Particularly, tryptophan and its derivatives elute close to DA under the
chromatographic conditions used, thus making interpretation of the results ambigu-
ous. A strong anion exchange (SAX) solid phase extraction (SPE) cleanup is gener-
ally recommended.39
Alternative methods based on LC with MS detection9,40–43 have been developed
for monitoring DA in marine matrices. In this frame, the suitability of HILIC-MS
technique for analysis of PSP toxins has also been examined for detection and quan-
titation of domoic acid.11 The method could be used for combined analysis of DA and
PSP toxins, which are coextracted with any of the extraction procedures currently
used. Details on the developed method and its application to shellfish samples are
provided below.
6.3.1 HILIC-MS Determination of DA

The presence of both COOH and NH functionalities in the domoic acid molecule
allows MS detection of the toxin by both positive and negative ionization modes.
Full-scan mass spectra (MS) of DA were acquired on a PE-Sciex triple quaÂ�
drupole MS equipped with a Turbospray source (API-2000) and showed the presence

in positive ion mode of [M + H]+ and [M + Na]+ ions at m/z 312 and 334, respectively,
while the [M − H]− ion at m/z 310 dominated the spectrum in negative ion mode. No
significant fragmentation was observed under the conditions used. These ions were
chosen for SIM experiments.
The [M + H]+ and [M − H]− ions were selected as precursor ions for MS/MS product
ion scan experiments. The fragmentation pattern produced in the MS/MS spectrum of
[M + H]+ consisted mainly of H2O and CO or HCOOH losses. In particular, the follow-
ing ions were observed: m/z 294 [M + H − H2O]+, m/z 266 [M + H − HCOOH]+, m/z 248
[M + H − HCOOH − H2O]+, m/z 220 [M + H − HCOOH − H2O − CO]+, m/z 193 [M +
H − HCOOH − H2O − CO − HCN]+, and m/z 175 [M + H − HCOOH − 2H2O − CO −
HCN]+. The negative MS/MS spectrum yielded fragments due to successive losses
of CO2 and H2O from the [M − H]− ion. The following ions were observed: m/z 266
[M − H − CO2]−, m/z 248 [M − H − CO2 − H2O]−, m/z 222 [M − H − 2CO2]−, m/z 204
[M − H − 2CO2 − H2O]−, m/z 160 [M − H − 3CO2 − H2O]−, and m/z 82 [M − H − 228]−.
The above MS/MS transitions of the [M + H]+ and [M − H]− precursor ions were
selected for MRM experiments.
Under the HILIC conditions optimized for PSP toxins (Elution system A) (Table
6.3), domoic acid eluted at the chromatographic front. A higher percentage of organic
modifier, 75% B (Elution system B), was necessary to achieve effective retention of
the toxin, which then eluted at 4.96â•›min. By way of example, MRM chromatogram
in positive ion mode for a 1.1â•›μg/mL standard solution of domoic acid is shown in
Figure 6.5a. Similar results were obtained in negative ion mode. Combined analysis
of DA and PSP toxins (Figure 6.5b) was possible using the gradient elution (Elution
system C in Table 6.3).
Limits of detection (LOD, S/N = 3) and of quantitation (LOQ, S/N = 10) for DA in
SIM and MRM modes are reported in Table 6.5. In all cases, correlation coefficients
were >0.9994, indicating a high degree of linearity of the plots within the tested
concentration range.
In order to establish whether there was any ion enhancement or suppression for
DA due to the presence of matrix, MRM experiments of a spiked extract and of a
pure standard solution of DA (20â•›μg/mL) were run in triplicate. Comparison of the
results showed a 53% and 3% ion suppressions in signal intensity in positive and
negative ion modes, respectively. The matrix effect observed in positive ion mode
was significant while the matrix interference in negative ion mode appeared negligi-
ble, suggesting that quantitation of DA in 1â•›g/mL crude extract was reliable using a
pure standard solution as an external standard. However, the peak shape was broad
and showed significant peak tailing, suggesting that some improvement was advis-
able, even in negative ion mode. The matrix concentration that would induce mini-
mum suppression effect was established to be 0.0625 and 0.125â•›g/mL for positive
and negative ion modes, respectively. Toxin-free sample extracts at such matrix con-
centrations were used to prepare matrix matched standards at four levels of DA con-
centration. Subsequently, calibration curves for pure and matrix-matched standards
were generated in positive and negative MRM modes. Good linearity was observed
in all cases. In negative ion mode, the slope of the curve for matrix matched stan-
dard was similar to that of pure standard, with a possible slight enhancement effect
in the spiked extract. In positive ion mode, a noteworthy suppression of signal was

GTX1
GTX4
m/z 412 > 332
DA m/z 412 > 314
GTX2
GTX3
NEO
m/z 396 > 316
m/z 396 > 298
m/z 316 > 298
m/z 312 > 266 GTX5 STX
m/z 312 > 248

m/z 380 > 300
m/z 312 > 220 m/z 300 > 282
m/z 312 > 193 DA
m/z 312 > 294

×3 m/z 312 > 266
m/z 312 > 175 ×3 m/z 312 > 248
0 1 2 3 4 5 6 7 8 9 10 0 10 20 30
Figure 6.5 HILIC-MS analyses in positive MRM mode of (a) a 1.1â•›μg/mL standard solu-
tion of domoic acid (RT = 4.96â•›m in) by using isocratic elution (75% B) and (b) a standard mix-
ture containing domoic acid and selected PSP toxins (GTX2, GTX3, GTX1, GTX4, GTX5,
STX, and NEO), by using gradient elution (Table 6.3, elution system C). Under the used con-
ditions, the following toxins were determined: DA (RT = 4.69â•›m in), GTX1 (RT = 14.17) and
GTX4 (RT = 15.79), GTX2 (RT = 14.04) and GTX3 (RT = 15.51), NEO (RT = 28.33), GTX5
(RT = 19.00), and STX (RT = 27.86). (Modified from Ciminiello, P. et al., Rapid Commun.
Mass Spectrom., 19, 2030, 2005. With permission from Wiley.)
observed despite the additional dilution of matrix, suggesting that matrix matched
standard should be used for accurate quantitation in positive mode. The minimum
detection level for DA in tissue was found to be 63 and 190â•›ng/g in positive and
negative MRM experiments, respectively.
6.3.2 Application to Shellfish Samples

In order to test suitability of the developed method for the analysis of shellfish, a num-
ber of samples of Mytilus galloprovincialis were collected in the period 2000–2004
in the Adriatic Sea, in coincidence with the presence in sea-water of appreciable
amounts of Pseudo-nitzschia spp.
The samples were separately extracted with MeOH/H2O (1:1) and with 0.1â•›M ace-
tic acid; both procedures allow extraction of water-soluble PSP and DA, if they cooc-
cur. The crude extracts were directly analyzed by HILIC-MS for both PSP toxins
and DA.
As already observed for PSP toxins, SIM experiments showed a high background
signal in the selected ion traces and extra peaks from other components in the crude

Table 6.5
Limits of Detection (LOD, S/N = 3) and Limits of
Quantitation (LOQ, S/N = 10) for Domoic Acid as
Determined by HILIC-MSa in SIM and MRM Experiments
Using an API 2000 Triple Quadrupole System
Experiment Ion LOD (ng/mL) LOQ (ng/mL) R2
SIM+ 312 3 11 0.9998
334 14 47
MRM+ 312 > 266 3 11 0.9999
312 > 248 5 18 0.9999
312 > 220 8
312 > 193 10
312 > 175 23
312 > 294 30
SIM− 310 2 7 0.9999
MRM− 310 > 222 10 32 0.9994
310 > 266 10 34 0.9996
310 > 160 79
310 > 248 49
310 > 204 82
310 > 82 75
Source: Modified from Ciminiello, P. et al., Rapid Commun. Mass Spectrom.,

19, 2030, 2005. With permission from Wiley.
a Optimized chromatographic conditions reported in Table 6.3 were used
(Elution system B).
extract, which made interpretation of the results difficult. The more selective and
sensitive (relative to S/N) MRM experiments indicated the presence of DA in some
of the analyzed samples basing on peak retention time (4.90â•›min), the presence of six
diagnostic fragments for the analyte, and the ion abundance ratios.
The domoic acid content in the M. galloprovincialis samples was determined by
direct comparison to individual standard solutions of domoic acid at similar concen-
trations injected in the same experimental conditions. The most abundant transitions
(m/z 310 > 266 and 310 > 222) were used for quantitative studies, which were carried
out on the methanol/water (1:1) extract (official extraction procedure), since DA in
0.1â•›M acetic acid extract seemed to have poorer stability over a long storage period.
Concentrations of DA in the analyzed samples were in the range 0.25–5.5â•›μg/g,
levels well below to the regulatory limit (20â•›μg/g in edible tissue). However, this was
the first time that domoic acid had been detected in Adriatic sea and represented a
warning for domoic acid as one of the toxins that needs to be carefully monitored in
Adriatic shellfish.

6.4 Assorted Cyanobacterial Toxins

Cyanobacteria are found in fresh and brackish water throughout the world and can
present a public safety hazard through contamination of drinking water supplies.
They are known to produce a number of toxins including neurotoxins, such as PSP
toxins and anatoxin-a, and hepatotoxins, such as cylindrospermopsins and micro-
cystins (Figure 6.1).
Among cyanobacteria, PSP toxins are produced by Aphanizomenon flos-aquae,44,45
Anabaena circinalis,46 Lyngbya wollei,47 and Cylindrospermopsis raciborskii.48
Anatoxin-a (ATX-a) is a potent depolarizing neuromuscular blocking agent49
produced by different strains of Anabaena, Planktothrix, and Aphanizomenon.50
Cylindrospermopsin and its deoxy-derivative are alkaloids produced by C. racibor-
skii,51 Umezakia natans,52 Aphanizomenon ovalisporum,53 and Raphidiopsis cur-
vata.54 Microcystins, produced mainly by Microcystis sp., are heptapeptides55,56 and
show tumor-promoting activity on rat liver with inhibition of protein phosphatases
1 and 2A.57
The monitoring of drinking water supplies for the presence of these toxins is of
critical importance for the assessment of environmental and health risks. So, the
HILIC-MS method for detection of PSP toxins10 was extended to the analysis of
ANTX-a, CYN, and its deoxy-derivative as well as microcystins LR and RR,12 with
the aim to develop an analytical method that could provide simultaneous detection
and unambiguous identification of different cyanobacterial toxins.
6.4.1 HILIC-MS Determination of Assorted Cyanobacterial Toxins

As first step for method development, ESI MS behavior of CYN, doCYN, and ATX-a
was investigated.
The MS spectrum of CYN (Figure 6.6a) contained an abundant [M + H]+ ion
at m/z 416, as well as the ammonium, sodium, and potassium adduct ions at m/z
433, 438, and 454, respectively. The MS/MS spectrum of the protonated ion of
CYN (Figure 6.6c) showed fragment ions corresponding to the loss of SO3 and H2O
from [M + H]+, at m/z 336 and 318, respectively. Another fragment ion, [M + H −
142]+ at m/z 274, resulted from loss of the [6-(2-hydroxy-4-oxo-3-hydropyrimidyl)]-
hydroxymethinyl moiety of the molecule and this ion subsequently underwent loss
of SO3 and H2O to afford ions at m/z 194 and 176, respectively. MS and MS/MS
spectra of doCYN (Figure 6.6b and d, respectively) were very similar, the only
difference being the shift of ions at m/z 336 and 318 in CYN down 16 mass units
in doCYN. The proposed fragmentation of CYN and doCYN is summarized in
Figure 6.6e.
The MS spectrum of ATX-a showed an abundant [M + H]+ ion at m/z 166 and
fragment ions at m/z 149 and 131 due to sequential elimination of NH3 and H2O.
The MS/MS spectrum showed the same ions, as well as prominent ion at m/z 91 (not
assigned) and m/z 43 corresponding to CH3CO+ from the acetyl function.
The protonated and ammonium-adduct ions of CYN and doCYN as well as the
[M + H]+ and the [M + H − NH3]+ ions of ATX-a were selected as ions to monitor

100 [M+H]+ 100 [M+H]+

416 400
80 80
60 60
40 433 40 417
438 422
438
20 454 20
% Relative intensity
0 0
100 200 300 400 500 100 200 300 400 500
(a) (b)
100 194 100

194
416 400
80 80
60 60
40 176 336 40 176

274 320
20 20 274
318
416 302 400
0 0
100 200 300 400 500 100 200 300 400 500
(c) m/z (d) m/z
–SO3 –H2O
m/z 274 m/z 194 m/z 176
–SO3 m/z 336 (R = OH)

R H O m/z 320 (R = H)
H H
O O S O –H2O
HN N HN N + O
CH3 m/z 318 (R = OH)
H
OH m/z 302 (R = H)
N
H
[M + H]+
m/z 416 (R = OH)
m/z 400 (R = H)
(e)
Figure 6.6 Electrospray mass spectra in positive ion mode of (a) cylindrospermopsin
(CYN) and (b) its deoxy derivative (doCYN). MS/MS product ion spectra of the [M + H]+
ions of CYN and doCYN are shown in (c) and (d), respectively. MS/MS spectra were carried
out on API-III+ system using a collision energy of 30â•›V. Assignments of labeled fragment ions
are shown (e). (Modified from Dell’Aversano, C. et al., J. Chromatogr. A, 1028, 155, 2004.
With permission from Elsevier.)

in SIM experiments. The most abundant fragment ions contained in the MS/MS
spectra of each toxin were selected for MRM experiments.
Under the same chromatographic conditions used for saxitoxin and its analogues
(65% B), ATX-a, CYN and doCYN eluted too quickly. As observed for domoic acid,
a higher percentage of solvent B was required for effective retention of these toxins.
In particular, with 75% B isocratic (Elution system B), ATX-a, CYN, and doCYN
eluted at 5.8, 7.1, and 6.2â•›min, respectively. The three-step gradient (Elution system
C) developed for combined analysis of PSP toxins and DA allowed determination
also of the above cyanobacterial toxins in combination with STX-like compounds
in one 32â•›min run. The results are shown in Figure 6.7 for a mixture of standard
compounds.
A gradient elution (90%–65% B over T = 13â•›min and hold 5â•›min) was required
for analysis of standard solutions of microcystin-LR and -RR, which eluted at 7.9
and 13.5â•›min, respectively. Mono- and bi-charged protonated molecules (m/z 995
and 498 for microcystin-LR and m/z 1038 and 520 for microcystin-RR, respectively)
were selected as ions to monitor in selected ion monitoring experiments. HILIC-MS
technique proved to selectively detect the microcystins, but the peak shapes were not
satisfactory, showing increased broadening and tailing compared to the other toxins.
Likely, an adjustment of mobile phase, pH, or buffer could improve the peak shape,
but many RPLC methods for the analysis of a large range of microcystins have been
reported in literature,58–60 which offer robustness and better chromatographic perfor-
mance than HILIC.
6.4.2 Applications to Cyanobacterial Samples

Field and cultured cyanobacterial samples were analyzed to test suitability of the
HILIC-MS method to the analysis of real samples. A simple extraction method was
used and no cleanup was performed on the crude extracts in order to demonstrate
rapid analysis.
A sample of A. circinalis, collected in October 1997 from the Coolmunda Dam,
South East Queensland (Australia), was acquired and analyzed by HILIC-MS for
the presence of saxitoxin-related compounds. A preliminary RPLC-ox-FLD analy-
sis had shown that STX, GTX2, GTX3, C1, C2, and dcSTX were present in the
sample at concentrations in the range of 75–1000â•›μg/g. HILIC-MS analysis of the
crude extract was carried out in both SIM and MRM modes. The latter gave better
results. All of the toxins detected by RPLC-ox-FLD, as well as GTX5, dcGTX2,
and dcGTX3, were easily confirmed by matching ion transition signals and reten-
tion times with those of standards. In addition, a significant peak was present in the
chromatogram for the m/z 380 > 300 transition at 9.7â•›min whose identity could not
be assigned just on the basis of MS data.
The Coolmunda Dam A. circinalis extract was also analyzed for ATX-a. The
detection limit for the ATX-a standard on the m/z 166 ion was estimated (S/N = 3)
to be 1.4â•›pmol injected on column in SIM mode and the sample was found to contain
no detectable ATX-a. Interestingly, European strains of A. circinalis produce ATX-a
while Australian strains have never been observed to do so.61

dcSTX
m/z
257 (×4)
STX
B1
380, 300 (×4)

dcGTX2
dcNEO
dcGTX3
273 (×10)
GTX1
C3 C4 GTX4
492, 412, 332 (×10)

C1 C2
GTX2 NEO
GTX3
316 (×2)
476, 396 (×15)
CYN
416 (×6)
ATX-a
166 (×1)
0 5 10 15 20 25 30
Time (min)
Figure 6.7 HILIC-MS analyses of a standard mixture containing most STX-like com-
pounds, ATX-a and CYN. Experiments were carried out in SIM mode on the API-III+ sys-
tem. Some traces are plotted with expanded scales as indicated. Gradient elution was used
(Table 6.3, elution system C). (Modified from Dell’Aversano, C. et al., J. Chromatogr. A,
1028, 155, 2004. With permission from Elsevier.)

doCYN
CYN
m/z > m/z

416 > 194
416 > 176
416 > 336
416 > 274
400 > 194

400 > 320
400 > 274
0 5 10
Time (min)
Figure 6.8 HILIC-MS analyses in positive MRM mode of a Cylindrospermopsis racibor-

skii extract containing cylindrospermopsin and its deoxy derivative. Experiments were carried
out on the API-III+ system using isocratic elution (Table 6.3, elution system B). (Modified from
Dell’Aversano, C. et al., J. Chromatogr. A, 1028, 155, 2004. With permission from Elsevier.)
A cultured sample of C. raciborskii, known to contain CYN and doCYN at lev-

els as high as 2â•›mg/g, was acquired for testing. A HILIC-MS analysis of the crude
extract based on SIM of the [M + H]+ ions provided sufficient selectivity and sen-
sitivity, detection limit (S/N = 3) for the CYN standard being 3.2â•›pmol injected on
column. An MRM experiment was also performed on the sample for added confir-
mation. The results (Figure 6.8) confirmed the SIM results and also provided excel-
lent sensitivity, the detection limit being (S/N = 3) 0.3â•›pmol injected on column. The
C. raciborskii extract was also analyzed for saxitoxin-related toxins but none was
detected in this strain.
6.5 Conclusions
HILIC-MS is a technique suitable for a number of polar hydrosoluble toxins,
including saxitoxin and its analogues, most commonly known as PSP toxins,
domoic acid, the causative toxin of ASP syndrome, and some assorted cyanotoxins,
such as anatoxin-a, cylindrospermopsin, and its deoxy derivative. Microcystins
could also be chromatographed but they are best analyzed by existing RPLC-MS
methods.
For saxitoxin and its analogues as well as for domoic acid, MRM is the preferred
method due to its higher selectivity. Indeed, a cleanup step is needed if SIM is the

only acquisition mode available. For cylindrospermopsin and anatoxin-a, the SIM
method is adequate, but MRM can provide additional selectivity for confirmatory
analyses. Multitoxin determination is possible, thus allowing the rapid, simultaneous
screening of an entire range of toxins.
The HILIC-MS technique provides high sensitivity which is basically due to
mobile phase character, which does not use ion-pair agents and is high in organic
modifier, which results in high ionization efficiency. LOD for PSP toxins in MRM
mode can vary between instruments and MS tuning, but for a sensitive instrument,
such as the triple quadrupole API-4000, LOD as low as 5–100â•›nM have been dem-
onstrated and compete favorably with those achieved by the LC-ox-FLD method. So,
the analysis of shellfish samples with a PSP toxin content near the regulatory limit is
possible. LOD for domoic acid were below the regulatory limit and allow the screen-
ing of DA in shellfish tissues even at trace levels.
The developed HILIC-MS methods have shown to be sensitive, straightforward,
readily automated, and robust with similar results obtained in different laboratories
using different instrumentation.
References
1. Hallegraeff, G. M., Phycologia, 1993, 32, 79.
2. Tibbetts, J., Environ. Health Persp., 1998, 106, A326.
3. Anderson, D. M., Red Tides: Biology Environmental Science and Toxicology, Okaichi, T.,
Anderson, D. M., Nemoto, T., Eds., Elsevier, New York, 1989, p. 11.
4. Aune, T., Seafood and Freshwater Toxins—Pharmacology, Physiology, and Detection,
Botana, L. M., Ed., CRC Press Taylor & Francis Group, Boca Raton, FL, 2nd edn.,
2008, p. 21.
5. Horwitz, W., Ed., Official Methods of Analysis of AOAC International, 17th edn., AOAC
International, Gaithersburg, MD, 2000.
6. Quilliam, M. A., Applications of LC-MS in Environmental Chemistry, Barcelo, D., Ed.,
Elsevier Science BV, Amsterdam, the Netherlands, 1996, p. 415.
7. Quilliam, M. A., J. Chromatogr. A, 2003, 1000, 527.
8. Botana, L. M., Ed., Seafood and Freshwater Toxins–Pharmacology, Physiology, and
Detection, 2nd edn., CRC Press Taylor & Francis Group, Boca Raton, FL, 2008.
9. Quilliam, M. A., Hess, P., Dell’Aversano, C., Mycotoxins and Phycotoxins in Perspective
at the Turn of the Millennium, deKoe, W. J., Samson, R. A., Van Egmond, H. P., Gilbert, J.,
Sabino, M., Eds., Wageningen, the Netherlands, 2001, p. 383.
10. Dell’Aversano, C., Hess, P., Quilliam, M. A., J. Chromatogr. A, 2005, 1081, 190.
11. Ciminiello, P. et al., Rapid Commun. Mass Spectrom., 2005, 19, 2030.
12. Dell’Aversano, C., Eaglesham, G. K., Quilliam, M. A., J. Chromatogr. A, 2004, 1028,
155.
13. Mortensen, A. M., Toxic Dinoflagellates, Anderson, D. M., White, A. W., Baden, D. G.,
Eds., Elsevier, New York, 1985, p. 163.
14. Narahashi, T., Fed. Proc., 1972, 31, 1124.
15. Genenah, A. A., Shimizu, Y., J. Agric. Food Chem., 1981, 29, 1289.
16. Oshima, Y., J. AOAC Int., 1995, 78, 528.
17. Sato, S., Shimizu, Y., Harmful Algae, Proceedings of the VIII International Conference
on Harmful Algae, Reguera, B., Blanco, J., Fernandez, M. L., Wyatt, T., Eds., IOC,
UNESCO, Vigo, Spain, 1997, p. 465.
18. Lawrence, J. F., Menard, C., J. AOAC Int., 1991, 74, 1006.

19. Quilliam, M. A., Janecek, M., Lawrence, J. F., Rapid Commun. Mass Spectrom., 1993,
7, 482.
20. Quilliam, M. A. et al., Rapid Commun. Mass Spectrom., 1989, 3, 145.
21. Thibault, P., Pleasance, S., Laycock, M. V., J. Chromatogr., 1991, 542, 483.
22. Jaime, E. et al., J. Chromatogr., 2001, 929, 43.
23. Strege, M. A., Anal. Chem., 1998, 70, 2439.
25. Dorsey, J. G. et al., Anal. Chem., 1998, 70, 591R.
26. Alpert, A. J., J. Chromatogr., 1990, 499, 177.
27. Tolstikov, V. V., Fiehn, O., Anal. Biochem., 2002, 301, 298.
28. Yoshida, T., J. Biochem. Biophys. Methods, 2004, 60, 265.
29. Cembella, A. D. et al., Harmful Algae, 2002, 1, 313.
30. Dell’Aversano, C. et al., J. Nat. Prod., 2008, 71, 1518.
31. Quilliam, M. A., Wright, J. L. C., Anal. Chem., 1989, 61, 1053.
32. Bates, S. S. et al., Can. J. Fish. Aquat. Sci., 1989, 46, 1203.
33. Garrison, D. L. et al., J. Phycol., 1992, 28, 604.
34. Martin, J. L. et al., Marine Ecol.: Progress Series, 1990, 67, 177.
35. Cerino, F. et al., Harmful Algae, 2005, 4, 33.
36. Lawrence, J. F. et al., J. Chromatogr., 1989, 462, 349.
37. Tasker, R. A. R., Connell, B. J., Strain, S. M., Can. J. Physiol. Pharmacol., 1991, 69, 378.
38. Quilliam, M. A. et al., Int. J. Environ. Anal. Chem., 1989, 36, 139.
39. Quilliam, M. A., Xie, M., Hardstaff, W. R., J. AOAC Int., 1995, 78, 543.
40. Lawrence, J. F. et al., J. Chromatogr., 1994, 659, 119.
41. Hummert, C. et al., Chromatographia, 2002, 55, 673.
42. Hess, P. et al., J. AOAC Int., 2001, 84, 1657.
43. Furey, A. et al., J. Chromatogr. A, 2001, 938, 167.
44. Ferriera, F. M. B. et al., Toxicon, 2001, 39, 757.
45. Li, R. et al., Hydrobiologia, 2000, 438, 99.
46. Humpage, A. R. et al., Aust. J. Mar. Freshwater Res., 1994, 45, 761.
47. Onodera, H. et al., Nat. Toxins, 1997, 5, 146.
48. Lagos, N. et al., Toxicon, 1999, 37, 1359.
49. Carmichael, W. W., Biggs, D. F., Peterson, M. A., Toxicon, 1979, 17, 229.
50. Sivonen, K., Mycotoxins and phycotoxins—Developments in chemistry, toxicology and
food safety, Proceedings of the International IUPAC Symposium on Mycotoxins and
Phycotoxins, M. Miraglia, Ed., Rome, Italy, 1998, p. 547.
51. Ohtani, I., Moore, R. E., Runnegar, M. T. C., J. Am. Chem. Soc., 1992, 114, 7941.
52. Harada, K.-I. et al., Toxicon, 1994, 32, 73.
53. Banker, R. et al., J. Phycol., 1997, 33, 613.
54. Li, R. H. et al., J. Phycol., 2001, 37, 1.
55. Carmichael, W. W. et al., Toxicon, 1988, 26, 971.
56. Rinehart, K. L. et al., J. Am. Chem. Soc., 1988, 110, 8557.
57. Matsushima, R. et al., Biochem. Biophys. Res. Commun., 1990, 171, 867.
58. Meriluoto, J., Anal. Chim. Acta, 1997, 352, 298.
59. Lawton, L., Edwards, C., Codd, G. A., Analyst, 1994, 119, 1525.
60. Harada, K.-I., Kondo, F., Tsuji, K., J. AOAC Int., 2001, 84, 1636.
61. Fitzgerald, J., Cunliffe, D. A., Burch, M. D., Environ. Toxicol., 1999, 14, 203.

7 Application of HILIC
for Polar Environmental
Contaminants (Including
Pharmaceuticals) in
Aquatic Systems
Alexander L.N. van Nuijs, Isabela Tarcomnicu,
Hugo Neels, and Adrian Covaci
Contents
7.1 Introduction................................................................................................... 134
7.2 Marine Toxins................................................................................................ 134
7.2.1 Generalities........................................................................................ 134
7.2.2 Paralytic Shellfish Poisoning Toxins................................................. 135
7.2.3 Tetrodotoxin and Analogs................................................................. 140
7.2.4 Domoic Acid...................................................................................... 142
7.2.5 Concluding Remarks......................................................................... 142
7.3 Pharmaceuticals............................................................................................. 143
7.3.1 Estrogens............................................................................................ 143
7.3.2 Cytostatic Drugs................................................................................ 146
7.3.3 Antibiotics.......................................................................................... 147
7.3.4 Miscellaneous Pharmaceuticals........................................................ 147
7.3.5 Drugs of Abuse.................................................................................. 149
7.4 Pesticides....................................................................................................... 150
7.4.1 Organophosphorus Pesticides............................................................ 151
7.4.2 Phenylurea Herbicides....................................................................... 151
7.4.3 Quaternary Ammonium Salt Herbicides........................................... 152
7.5 Miscellaneous................................................................................................ 153
7.5.1 Dichloroacetic Acid........................................................................... 153
7.5.2 Oligosaccharides................................................................................ 153
References............................................................................................................... 154
133

7.1 Introduction
Reversed-phase liquid chromatography (RPLC) has been the method of choice for
the analysis of various classes of compounds not amenable to gas chromatography
(GC). A large number of compounds were preferably measured by RPLC and C18-
based silica stationary phases. Yet, highly polar compounds undergo early elution
on traditional RP stationary phases, leading to lower sensitivity of the mass spec-
trometric (MS) detection due to high water percentage in the mobile phase at the
beginning of the run. The analysis of highly hydrophilic, ionic, and polar compounds
by hydrophilic interaction liquid chromatography (HILIC) coupled to MS has been
demonstrated as a valuable complementary approach to RPLC.1 The use of a low
aqueous and high polar organic mobile phase in HILIC separation is almost ideal for
electrospray ionization in many cases, resulting in increased sensitivity.1
In this review, various applications of HILIC separation coupled with MS detec-
tion recently developed for a number of environmental contaminants present in
aquatic systems, such as marine toxins, pharmaceuticals, drugs of abuse, and pes-
ticides, are discussed together with detailed description of the methodologies used.
7.2 Marine Toxins

7.2.1 Generalities
Cyanobacteria and dinoflagellates in freshwater and brackish water produce a wide
range of toxins that can accumulate in drinking water supplies or in aquatic orgaÂ�
nisms. Because of the human consumption of these contaminated aquatic organisms
(shellfish, puffer fish) or this water, these compounds are of concern for public health
and safety.
Paralytic shellfish poisoning (PSP) toxins can induce dangerous intoxications in
humans if contaminated shellfish is consumed. PSP toxins act as potent, reversible
sodium channel blockers, resulting in paralytic symptoms such as respiratory insuf-
ficiency and can even lead to death.2
The amnesic shellfish poisoning (ASP) syndrome causing toxin, domoic acid
(DA), binds predominantly to N-methyl-d-aspartate receptors in the central nervous
system and causes a depolarization of neurons and cell dysfunction or death.3 The
ASP syndrome is firstly characterized by gastrointestinal symptoms (nausea, diar-
rhea, and gastroenteritis) followed by neurological complaints (ataxia, confusion,
disorientation, and coma).4
The toxin tetrodotoxin (TTX) and its analogs are responsible for puffer fish poison-
ing (PFP). TTX inhibits reversibly voltage-activated sodium channels, which results in
a series of paralytic symptoms comparable with the effects of PSP toxins.5
The overall toxicity of TTX, PSP toxins, and DA in seafood is currently tested
by an Association of Official Analytical Chemists (AOAC) official mouse bioassay
method.6 A prepared extract of shellfish or puffer fish tissue is injected in mice and
the presence of the compounds is confirmed if a symptom pattern specific for these
toxins is observed. Quantification is based on the time of death of the mouse. This
mouse bioassay leads to severe ethical problems and does not offer very accurate and
precise results. Moreover, the knowledge of a toxin profile (different analogs) is not

Application of HILIC for Polar Environmental Contaminants 135
possible with this bioassay. Because of the dangerous effects of these toxins, there is
a need for techniques that allow fast and sensitive detection of these toxins in fresh-
water and aquatic organisms and that can offer knowledge of toxin profiles. Methods
based on LC with fluorescence (FL) or MS detection have been gaining interest.
These methods are often based on RPLC separations,7–12 but because of the highly
polar character of the toxins, ion-pair agents have to be used to obtain retention in
RPLC. The problem with these agents is that they interfere with the MS detection
(ion suppression or enhancement) and, because of the different charge of the toxin
and its analogs, a single run analysis is not possible. FL detection after oxidation is
an alternative, but the complex setup and the difficult interpretation of quantitative
results are major drawbacks. In this part, the application of HILIC combined with
FL or MS detection for the analysis of these toxins is discussed.
7.2.2 Paralytic Shellfish Poisoning Toxins

PSP toxins are produced by marine dinoflagellates belonging to the genera
Alexandrium, Pyrodinium, and Gymnodinium, as well as by cyanobacteria preÂ�
sent in freshwater, such as Aphanizomenon flos-aquae, Anabaena circinalis, and
Lyngbya wollei.13 PSP toxins all consist of a tetrahydropurine backbone and can be
divided into three groups based on their side chain (carbamoyl, N-sulfocarbamoyl,
and decarbamoyl) (Figure 7.1). Saxitoxin (STX), gonyautoxin 1, 2, 3, and 4 (GTX1,
GTX2, GTX3, GTX4), neosaxitoxin (NEO), 11α-OH-saxitoxin (11α-OH-STX), and
11β-OH-saxitoxin (11β-OH-STX) are members of the carbamoyl group. The
N-sulfocarbamoyl group consists of B1-gonyautoxin (B1), B2-gonyautoxin (B2),
R1 R2 R3 R4 Toxin
H H H O NH2 STX
H H OSO3– GTX2
R4 H OSO3– H O GTX3
OH H H NEO
R1 H OH H OSO3– GTX1
NH OH OSO3– H GTX4
N
NH+2 H H OH 11α-OH-STX
NH H OH H 11β -OH-STX
H2N+ N
OH H H H B1(GTX5)
OH H H OSO3– C1
O NHSO3–
R2 R3 H OSO3– H C2
OH H H O B2 (GTX6)
OH H OSO3– C3
STX = saxitoxin OH OSO3– H C4
NEO = neosaxitoxin
GTX = gonyautoxin H H H dcSTX
H H OSO3– dcGTX2
H OSO3– H OH dcGTX3
OH H H dcNEO
OH H OSO3– dcGTX1
OH OSO3– H dcGTX4
Figure 7.1 Chemical structure of PSP toxins.

C1-gonyautoxin (C1), C2-gonyautoxin (C2), C3-gonyautoxin (C3), and C4-gonyautoxin

(C4). Decarbamoylsaxitoxin (dcSTX), decarbamoylneosaxitoxin (dcNEO), and decar-
bamoylgonyautoxin 1, 2, 3, and 4 (dcGTX1, dcGTX2, dcGTX3, dcGTX4) belong to
the decarbamoyl group.
The aim of the developed procedures was to analyze seafood (shellfish, puffer
fish) or cell cultures for the presence of PSP toxins, DA and TTX, and analogs.
Before LC analysis, extracts from the different organisms have to be prepared. For
the analysis of PSP toxins in dinoflagellates and cyanobacteria, the sample prepara-
tion consists of a suspension of the organism culture in an acidic aqueous or aceto-
nitrile solution followed by homogenization, centrifugation, and filtration.14–17 The
resulting extract is then injected in the LC system. For tissues from puffer fish and
shellfish, the sample preparation consists of an acidic extraction of the toxins from
the tissues followed by homogenization, centrifugation, and in most cases further
cleanup with solid-phase extraction (SPE) or liquid–liquid extraction (LLE).14,16,18–20
The resulting extract is then injected in the LC system.
To have an unambiguous and fast identification and quantification of all PSP
toxins and to have knowledge of toxin profiles, methods are required that allow the
analysis of all PSP toxins (Figure 7.1) in a single analysis. Chromatographic separa-
tion coupled to MS detection is the obvious choice for such analyses because of the
high specificity and sensitivity of this detection. Because of the thermolabile nature
of PSP toxins, LC is the preferred chromatographic technique. However, the use of
RPLC reveals a major drawback in the analysis of PSP toxins: They are highly polar,
resulting in poor retention. The employment of ion-pair agents can overcome the
problem, but such agents interfere with the MS detection. HILIC can separate these
hydrophilic toxins, while the ionization efficiency in MS detection is enhanced with
HILIC due to the use of a high amount of organic solvent (methanol or acetonitrile)
in the mobile phase.
Dell’Aversano et al.16 optimized a method based on HILIC coupled to MS detec-
tion for the simultaneous analysis of all major PSP toxins in a single 30â•›min run
with a high degree of selectivity and no need for further confirmation (Table 7.1).
A TSKgel® Amide-80 stationary phase was used together with a mobile phase con-
sisting of 65% eluent (B), where (A) was water and (B) was acetonitrile/water (95:5),
both containing 2.0â•›m M ammonium formate and 3.6â•›m M formic acid (pH = 3.5). The
use of a PolyHydroxyEthyl Aspartamide™ column was also tested by this group,
but resulted in longer run time and bad peak shapes. Mobile phase, flow rate, and
column temperature were further optimized for the amide column. The method gave
excellent linearity and had a limit of detection of 30â•›nM, which should allow the
analysis of shellfish samples near the regulatory limit of 0.8â•›mg STX equivalent
per kg edible tissue.
Diener et al.14 have separated PSP toxins with HILIC using two detection sys-
tems, FL and tandem MS (MS/MS), building on the HILIC separation developed
by Dell’Aversano et al.,16 which were not interested in FL detection because of
co-elutions and inconstant retention times. The use of a zwitterionic ZIC®-HILIC
stationary phase in combination with gradient elution with a mobile phase consisting
of (A) 10â•›m M ammonium formate and 10â•›m M formic acid in water and (B) 80% ace-
tonitrile and 20% water with a final concentration of 5â•›m M ammonium formate and

Application of HILIC for Polar Environmental Contaminants
Table 7.1
Overview of Studies Concerning Marine Toxins: Studied Compounds, Chromatographic Conditions, and Analyzed Samples
Toxins Studied Stationary Phase Mobile Phase Samples References
PSP toxins
STX, NEO, GTX1, GTX2, GTX3, GTX4, ZIC-HILIC (A) 10â•›mM Ammonium formate and Alexandrium catanella [14]
B1, B2, C1, C2, C3, C4, dcSTX, dcNEO, (250 × 4.6â•›mm, 5â•›μm) 10â•›mM formic acid in water; (dinoflagellate)
dcGTX1, dcGTX2, dcGTX3, dcGTX4 (B) 80% Acetonitrile and 20% water with Gymnodinium catenatum
a final concentration of 5â•›mM ammonium (dinoflagellate)
formate and 2â•›mM formic acid—gradient
STX, NEO, GTX1, GTX2, GTX3, GTX4, TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Anabaena circinalis [15]
B1, B2, C1, C2, C3, C4, dcSTX, dcNEO, (250 × 2â•›mm, 5â•›μm) in water, both containing 2.0â•›mM (cyanobacterium)
dcGTX1, dcGTX2, dcGTX3, dcGTX4 ammonium formate and 3.6â•›mM formic Cylindrospermopsis
acid—gradient raciborskii
(cyanobacterium)
STX, NEO, GTX1, GTX2, GTX3, GTX4, TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Alexandrium tamarense [16]
B1, B2, C1, C2, C3, C4, dcSTX, dcNEO, (250 × 2â•›mm, 5â•›μm) in water, both containing 2.0â•›mM (dinoflagellate)
dcGTX1, dcGTX2, dcGTX3, dcGTX4, PolyHydroxyEthyl ammonium formate and 3.6â•›mM formic Mytilus edulis (mussel)
11α-OH-STX, 11β-OH-STX Aspartamide™ acid—isocratic
(200 × 2â•›mm, 5â•›μm)
(continuedâ•›)
137
Overview of Studies Concerning Marine Toxins: Studied Compounds, Chromatographic Conditions, and Analyzed Samples
Toxins Studied Stationary Phase Mobile Phase Samples References
STX, NEO, GTX1, GTX2, GTX3, GTX4, TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Alexandrium catenella [17]
dcGTX1, dcGTX2, dcGTX3, dcGTX4 ammonium formate and 3.6â•›mM formic
acid—isocratic
STX, NEO, GTX1, GTX2, GTX3, GTX4, TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Mytilus galloprovincialis [19]
B1, B2, C1, C2, C3, C4, dcSTX, dcNEO, (250 × 2â•›mm, 5â•›μm) in water, both containing 2.0â•›mM (mussel)
dcGTX1, dcGTX2, dcGTX3, dcGTX4, ammonium formate and 3.6â•›mM formic
11α-OH-STX, 11β-OH-STX, DA acid—isocratic
STX, NEO, GTX1, GTX2, GTX3, GTX4, TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Alexandrium ostenfeldii [21]
dcGTX1, dcGTX2, dcGTX3, dcGTX4, ammonium formate and 3.6â•›mM formic
11α-OH-STX, 11β-OH-STX acid—isocratic
Domoic acid
DA TSK-Gel Amide-80 (A) Deionized water; (B) 95% acetonitrile Mytilus galloprovincialis [18]
(250 × 2â•›mm, 5â•›μm) in water, both containing 2.0â•›mM (mussel)
ammonium formate and 3.6â•›mM formic
acid—gradient

TTX and analogs
STX, NEO, GTX1, GTX2, GTX3, GTX4, B1, ZIC-HILIC (A) Deionized water; (B) 95% acetonitrile Mytilus edulis (mussel) [20]
dcSTX, dcNEO, dcGTX2, dcGTX3 (150 × 2.1â•›mm, 3.5â•›μm) in water, both containing 2.0â•›mM
TSK-Gel Amide-80 ammonium formate and 3.6â•›mM formic
(250 × 2.1â•›mm, 5â•›μm) acid—gradient
TTX, 11-deoxyTTX, 6,11-dideoxyTTX, TSK-Gel Amide-80 16â•›mM Ammonium formate buffer and Fugu pardalis (puffer fish) [23]
5-deoxyTTX, 5,6,11-trideoxyTTX (150 × 2â•›mm, 5â•›μm) acetonitrile (3/7, v/v)—isocratic
TTX, 4-epiTTX, 11-deoxyTTX, 5-deoxyTTX, TSK-Gel Amide-80 16â•›mM Ammonium formate buffer and Fugu pardalis (puffer fish) [22]
5,6,11-trideoxyTTX, 4,9-anhydroTTX, (150 × 2â•›mm, 5â•›μm) acetonitrile (3/7, v/v)—isocratic
4-CysTTX, STX, dcSTX
TTX, 4-epiTTX, 6-epiTTX, 11-deoxyTTX, ZIC-HILIC (A) 10â•›mM Ammonium formate and Takifugu oblongus (puffer [24]
norTTX-6(S)-ol, norTTX-6(R)-ol, (150 × 2.1â•›mm, 5â•›μm) 10â•›mM formic acid in water; (B) 80% fish)
nor-TTX-6,6-diol, anhydroTTX, acetonitrile and 20% water with a final
6-epianhydroTTX, 5-deoxyTTX, concentration of 5â•›mM ammonium
trideoxyTTX formate and 2â•›mM formic acid—gradient
TTX, 4-epiTTX, 6-epiTTX, 11-deoxyTTX, TSK-Gel Amide-80 16â•›mM Ammonium formate buffer and Fugu poecilonotus (puffer [25]
5-deoxyTTX, 5,6,11-trideoxyTTX, (150 × 2â•›mm, 5â•›μm) acetonitrile (3/7, v/v)—isocratic fish)
4,9-anhydroTTX, 6-epi-4,9-anhydroTTX, Cynops ensicauda (newt)
norTTX-6(S)-ol, norTTX-6(R)-ol
139
2â•›m M formic acid was proposed (Table 7.1). The application of this gradient provided
a complete separation of all regulated PSP toxins in a single run. FL detection was
possible, but MS/MS detection proved to be a better choice because of two rea-
sons: (1) the HILIC-MS/MS method led to shorter run times (40â•›min for MS/MS vs.
70â•›min for FL detection) because the FL detection required a slower gradient for the
effective separation of the PSP toxins and (2) a more selective detection with MS/MS
was observed compared with FL. Moreover, the HILIC-FL method needed a post-
column oxidation before FL detection was possible, making this method also more
complicated than the HILIC-MS/MS method. The method had good linearity and
the limit of detection for the compounds ranged between 0.03 and 0.80â•›ng injected,
depending on the compound and the type of detection.
Turrell et al.20 evaluated two stationary phases for the separation of PSP toxins
with HILIC coupled to MS and MS/MS detection (Table 7.1). Using a gradient elu-
tion with a mobile phase consisting of (A) deionized water and (B) 95% acetonitrile
in water, both containing 2.0â•›m M ammonium formate and 3.6â•›m M formic acid for
the TSK-Gel Amide-80 and ZIC-HILIC column, they concluded that the ZIC-HILIC
column provides a faster separation (35 vs. 43â•›min) and better peak shape for the late
eluting toxins, in accordance with the findings of Diener et al.14
The application of the HILIC-MS, HILIC-MS/MS, or HILIC-FLD methods to
extracts of dinoflagellate cultures showed the presence of a wide range of PSP toÂ�xins
in such samples (Table 7.1). Several studies revealed the presence of most of the
PSP toxins in prepared extracts of different species of dinoflagellates from China,
Mexico, Canada, Italy, and Australia.14–17 Samples in Australia revealed concentra-
tions of dcGTX2 up to 430â•›μg/g.15 An accurate knowledge of the toxin profiles of
these organisms could be obtained by HILIC and provided new insights into pat-
terns and sources of toxin accumulation in marine food webs. Extracts of mussels
from Canada and Scotland and extracts from puffer fish of Japan, analyzed with the
developed HILIC methods, showed the presence of PSP toxins.16,20–22 A toxicity cor-
responding with 178â•›μg STX/100â•›g shellfish was detected in Scottish mussels with
HILIC-MS.20
7.2.3 Tetrodotoxin and Analogs

TTX is a toxin that is probably produced by microalgae and is considered to be
the most lethal toxin coming from the marine environment.23 It is named after the
Tetraodontidae family of fish (puffer fish), but is also found in other fish and mol-
lusks and even in amphibians and land animals.24 An overview of the structures of
TTX and its analogs is given in Figure 7.2.
Sample preparation for the analysis of TTX and its analogs in puffer fish consists
of an acidic extraction of the toxins from the tissues followed by homogenization and
centrifugation. Diener et al.25 apply a direct injection of the supernatant, while other
papers report further clean-up on reversed-phase resins (Table 7.1).22,26,27
Nakagawa et al.27 described a method based on HILIC-MS for the detection of
TTX and its analogs in puffer fish. A TSK-Gel Amide-80 column was used in com-
bination with a mixture of 16â•›m M ammonium formate buffer and acetonitrile (3/7,
v/v) as mobile phase. This method allowed the separation of 11 TTX analogs. A 12th

OH R1 R2 R3 R4
TTX H OH OH CH2OH
O O
HO R4 4-epiTTX OH H OH CH2OH
R1
6-epiTTX H OH CH2OH OH
OH 11-deoxyTTX H OH OH CH3
R2 norTTX-6(S)-ol H OH OH H
HN NH R3
norTTX-6(R)-ol H OH H OH
+H norTTX-6, 6-(diol) H OH OH OH
2N
OH
O O
O R2 R1 R2
anhydroTTX OH CH2OH
OH 6-epianhydroTTX CH2OH OH
H
HN NH R1
+H N
2
OH
H O
HO R2 R1 R2
H
5-deoxyTTX OH CH2OH
OH trideoxyTTX H CH3
HO
HN NH R1
+H
2N
Figure 7.2 Chemical structure of TTX and its analogs.
TTX analog, 6,11-dideoxytetrodotoxin, was detected and reported for the first time
in puffer fish samples by applying the already optimized HILIC-MS method.
Diener et al.25 published a HILIC-MS study for the determination of TTX and ana-
logs in puffer fish. A ZIC-HILIC stationary phase was used for the separation of 11
toxins using a mobile phase consisting of (A) 10â•›mM ammonium formate and 10â•›mM
formic acid in water and (B) 80% acetonitrile and 20% water with a final concentration
of 5â•›mM ammonium formate and 2â•›mM formic acid. This setup resulted in enhanced
sensitivity and accuracy compared with the previous LC-MS methods.11,27 A further
advantage of this method was that the ZIC-HILIC stationary phase and mobile phase
can also be used for the analysis of PSP toxins.27 This allowed a combined determina-
tion of TTX and analogs and PSP toxins in one single run and has major advantages
over the method published by Jang and Yotsu-Yamashita,22 which needed two separate
runs for the analysis of TTX and analogs and PSP toxins in puffer fish.
Toxin profiles of TTX and its analogs could be revealed by analyzing extracts
of different tissues of puffer fish from Japan and Bangladesh with HILIC-MS and
HILIC-MS/MS.22,25–27 Samples from puffer fish from Bangladesh revealed concen-
trations up to 2929â•›μg/g of 5,6,11-trideoxyTTX in ovaries25 and up to 94â•›μg/g TTX
could be quantified in ovaries of puffer fish from Japan.22

COOH
COOH
COOH N
H
Figure 7.3 Chemical structure of DA.
7.2.4 Domoic Acid

DA is a naturally occurring, but rare, amino acid (Figure 7.3) produced by the dia-
tom Pseudo-nitzschia pungens f. multiseries.28 Other species of Pseudo-nitzschia,
namely, Pseudo-nitzschia australis, Pseudo-nitzschia pseudo-delicatissima, and
Pseudo-nitzschia galaxiae, are also found to produce DA.29 This compound is toxic
to marine mammals and seabirds, as well as to humans.
The sample preparation for the analysis of DA in shellfish samples is comparable
with the preparation of extracts for the analysis of PSP toxins in shellfish.18 Several
efforts have been made to analyze DA in shellfish with LC-MS because of its high
sensitivity and specificity. Recently, different RPLC-MS methods have been pro-
posed for the analysis of DA in seafood.30–32
Ciminiello et al.18 underlined the urgent need for multi-toxin determination
methods, since shellfish can be contaminated with different classes of toxins (DA
and PSP toxins). They presented a method based on HILIC-MS that simultaneously
analyzes DA and PSP toxins in shellfish. A TSK-Gel Amide-80 stationary phase was
used and a gradient elution with a mobile phase consisting of (A) deionized water
and (B) 95% acetonitrile in water, both containing 2.0â•›m M ammonium formate and
3.6â•›m M formic acid was applied. The method was based on an earlier described pro-
cedure,16 but in this method, a gradient elution was used, because DA was eluting at
the chromatographic front if the original isocratic conditions were applied. By using
a gradient elution starting with a higher amount of organic solvent, good retention
for DA in combination with an efficient separation of PSP toxins was obtained. DA
was detected in mussel samples from Italy (4 out of 11 samples were positive for DA)
by HILIC-MS and HILIC-MS/MS.18 Concentrations ranged from 0.25 to 2.5â•›μg/g,
which are below the regulatory limit of 20â•›μg/g.
7.2.5 Concluding Remarks

The presence of PSP toxins, DA, and TTX and analogs in seafood can have severe
implications for human health. Routine monitoring for these toxins is necessary
to prevent human intoxications. It is clear that there is a need for replacement
of the official mouse bioassay, which is now used for the evaluation of the toxi-
city of contaminated seafood. The application of HILIC coupled to MS detection
is a good alternative and deliver simple (one analysis run), sensitive, and accu-
rate measurements of toxins in seafood samples. In the future, methods based on
HILIC-MS can serve as scientific basis for the validation and implementation of
analytical techniques in (shell)fish monitoring programs to replace the traditional
mouse bioassay.

7.3 Pharmaceuticals
Pharmaceuticals, drugs of abuse (DOA), and their metabolites are contaminants of
increasing concern in the aquatic environment due to their toxicity, various endo-
crine-disrupting effects, and their potential accumulation in the aquatic ecosystem.
Several studies pointed out that excreted human or veterinary drugs, personal care
products, or DOA end up in the environment through the sewage system.33–35 A large
interest has been nowadays focused on their detection and quantification in wastewa-
ter and surface water, and HILIC is a method that is gaining interest due to the high
polarity of these pollutants. Table 7.2 shows a brief list of the molecules analyzed in
HILIC mode from aquatic systems.
7.3.1 Estrogens
Estrogens are steroid hormones produced by the endocrine system that are important
for the human brain, reproduction, and bone development. They are also present in
the environment from natural sources (human and animal excretion, plants, fungi) or
as by-products of synthetic chemicals. In aquatic systems, estrogens are a risk factor
as they act as endocrine disruptors producing fish feminization.36 They may influ-
ence plant growth or even human health, and the monitoring of these compounds has
therefore become imperious for more and more institutions. The analysis of estro-
gens (e.g., estrone) and their glucuronide and sulfate conjugates (Figure 7.4) at low
ng/L levels in environmental samples can be done by immunoassays, GC/MS, or
LC/MS. Immunoassay methods are sensitive and specific, but only few antibodies
are available and cross-reactions remain a problem. GC/MS requires derivatization
and hydrolysis of the estrogen conjugates, so the substitution group is lost before
analysis. LC-MS/MS is the most advantageous technique and is widely used today.
HILIC chromatography is useful for the separation of estrogen sulfates and glucu-
ronides, which are highly hydrophilic compounds.
To detect estrogens (estrone, estriol, estradiol) and their glucuronide and sulfate
conjugates, in a single run, Qin et al.37 used a column-switching method involving a
C18 and a HILIC stationary phase. They collected water samples from River North
Saskatchewan (Canada), adjusted them to pH = 2, and stored them at 4°C prior to
analysis. The analytes were isolated and pre-concentrated from 500â•›mL water by
SPE on Oasis™ HLB cartridges. Free estrogens were eluted with ethyl acetate, while
estrogen conjugates were eluted with methanol containing 2% ammonium hydroxide.
After elution of the SPE cartridge, the estrogen fraction was further derivatized with
dansyl chloride and the reaction mixture was purified once more on an Oasis HLB
cartridge. The last methanolic eluate was mixed then with the estrogen conjugates
fraction, the solvent was evaporated, and the analytes were reconstituted with mobile
phase and injected in the LC/MS system. A column switching set-up, employing
a binary pump connected to a quadrupole-linear ion trap mass spectrometer with
an electrospray (ESI) interface operated alternatively in positive (for the detection
of the dansylated estrogens) and negative (for estrogen conjugates) ion mode in the
same run. A 10-port 2-positions switching valve served to link up the two columns,
Luna® C18(2) (100 × 2.0â•›mm, 3â•›μm) for the separation of dansylated estrogens and

Table 7.2
Pharmaceuticals, Drugs of Abuse, and Pesticides Analyzed in HILIC Mode from Aquatic Systems—A Brief Review
of Employed Columns, Mobile Phases, and Type of Samples
Compound/Class
of Compounds Matrix Column Mobile Phase References
Pharmaceuticals
Estrogen conjugates River water ZIC-pHILIC (100 × 2.1â•›mm, 5â•›μm) (A) 5â•›mM Ammonium acetate 5â•›mM in water; [36]
(B) acetonitrile—gradient
Cytostatics Wastewater ZIC-HILIC (150 × 2.1â•›mm, 3.5â•›μm) (A) 30â•›mM Ammonium acetate in water; [37]
Antibiotics Liquid manure; Alltima HP HILIC (150 × 2.1â•›mm, 3â•›μm) (A) 0.1% Formic acid in water; [38]
rainfall runoff (B) acetonitrile—isocratic
Metformin Wastewater; ZIC-HILIC (150 × 2.1â•›mm, 3.5â•›μm) (A) 10â•›mM Ammonium formate in water [52]
surface water (pH 3 with formic acid); (B) acetonitrile—gradient
Albuterol, cimetidine, Water, sludge 1. Zorbax HILIC Plus (100 × 2.1â•›mm, 3.5â•›μm) 1. (A) 10â•›mM Ammonium acetate in water; [39]
ranitidine, metformin (B) acetonitrile—gradient
2. Atlantis HILIC (100 × 2.1â•›mm, 3â•›μm) 2. (A) 0.1% Acetic acid/ammonium acetate in [40]
water; (B) acetonitrile—gradient

Drugs of abuse
Cocaine and metabolites Wastewater Zorbax RX-Sil (150 × 2.1â•›mm, 5â•›μm) (A) 2â•›mM Ammonium acetate in water (pH 4.5 [41]
with acetic acid); (B) acetonitrile—gradient
Drugs of abuse Wastewater Luna HILIC (150 × 3â•›mm, 5â•›μm) (A) 5â•›mM Ammonium acetate in water; [59]
Pesticides
Organophosphorus Water Atlantis HILIC (150 × 2â•›mm, 5â•›μm) Acetonitrile/isopropanol/ammonium formate [42]
pesticides 200â•›mM (pH 3)—isocratic
Phenylurea pesticides and Drinking water Nucleosil Diol (150 × 4â•›mm, 5â•›μm) (A) 20â•›mM Ammonium formate pH 3.3; [43]
degradation products (B) acetonitrile—gradient
Diquat, paraquat Drinking water Atlantis HILIC (150 × 2.1â•›mm, 3.5â•›μm) (A) 250â•›mM Ammonium acetate pH 3; [44]
(B) acetonitrile—isocratic
Atlantis HILIC (150 × 2.1â•›mm, 3â•›μm) (A) 10â•›mM Ammonium formate pH 3.7; [45]
(B) acetonitrile—isocratic
145
CH3 O
O CH3 O
CH3 O
NaO C
O
HO O
HO NaO3SO
HO OH
Estrone Estrone-3-glucuronide sodium salt Estrone sulfate sodium salt
Figure 7.4 Estrone and its conjugates.
ZIC-pHILIC (100 × 2.1â•›mm, 5â•›μm) for the estrogen conjugates. The mobile phase
consisted of (A) acetonitrile/aqueous ammonium acetate 5â•›mM pH = 6.8 (75/25,
v/v) and (B) acetonitrile/aqueous ammonium acetate 5â•›mM pH = 6.8 (95/5, v/v).
The sample extract was injected with an initial mobile phase composition of 40%
A and the valve connected both columns. In this way, estrogens are trapped on the
C18 stationary phase and their conjugates transferred to the HILIC stationary phase.
Then, with an appropriate gradient and valve program, the two groups of compounds
were separated and detected. Using this approach, the method detection limits were
up to 10-fold better than in the classical RP approach, and the linearity range was
0.2–200â•›ng/L for estrogens, and 0.5–500â•›ng/L for conjugates (except for estriol glucuÂ�
ronides with a range of 2–2000â•›ng/L), with an injection volume of 10â•›μL. However,
strong matrix effects (up to 50%) were observed and these should be compensated
with deuterated internal standards. Only estrone and estrone-3-sulfate were detected
in river water samples.
7.3.2 Cytostatic Drugs

Cytostatic drugs are used in cancer therapy as chemotherapeutic drugs. Their action
is mainly to inhibit cell growth and cell division by disrupting DNA replication. The
cytotoxicity is not limited to tumor cells, but also affects other fast-dividing cells.
Thus, the chemotherapeutic drugs pose an intrinsic carcinogenic potential, and
their release and accumulation in the environment (they are poorly biodegraded) is
a risk factor for all organisms.38 The presence of chemotherapeutic drugs in waste-
water and river water has recently received increased attention. Anti-metabolites
are the most used cytostatic agents and have polar structures (5-fluorouracil, cyta-
rabine, and gemcitabine shown in Figure 7.5). They were analyzed from biologi-
cal and environmental samples, at very low concentrations, by GC/MS, often with
NH2 NH2
F N N
HN
N O N O
O N O O O
H
HO HO
F
HO OH HO F
5-Fluorouracil Cytarabine Gemcitabine
Figure 7.5 Chemical structures of some anti-metabolites.

derivatization,39 and by HPLC with ion-pairing on RP columns or using adequate

stationary phases as porous graphite or HILIC.
Cytostatics and metabolites (5-fluorouracil, cytarabine, gemcitabine, α-fluoro-β-
alanine, uracil 1-β-d-arabinofuranoside, 2,2′-dihydrodeoxyuridine) were extracted
by Isolute® ENV+ cartridges.40 Water samples (50â•›m L) were adjusted to pH = 6, loaded
onto the cartridges and eluted with methanol. The eluate was evaporated to dryness
and reconstituted with initial mobile phase before injection. For the investigation of
the cytostatics in hospital wastewater, samples were collected from a Swiss hospital,
filtered through glass fiber and a cellulose acetate membrane, and stored at −20°C
until analysis. In the method development, wastewater samples from the University
Hospital in Aachen (Germany) were used. The system employed by Kovalova et al.40
to separate cytostatics and their human metabolites was composed of a binary pump,
autosampler, and column thermostat, connected to a triple quadrupole mass spec-
trometer equipped with an ESI interface, operated in positive or negative ion mode
(in two separate runs). The extracted compounds were chromatographed in gradient
on a ZIC-HILIC column (150 × 2.1â•›mm, 3.5â•›μm), with a mobile phase consisting
of (A) aqueous ammonium acetate 30â•›m M/acetonitrile (2/3, v/v) and (B) acetoni-
trile. Method limits of quantification were in the range of 0.9–9â•›ng/L. To confirm
the masses of the analytes and elucidate the interfering peak observed in hospital
wastewater, a high-resolution Orbitrap mass analyzer was used. The method was
applied to hospital wastewater samples and 5-fluorouracil was detected in 76% of
samples, while gemcitabine and 2,2′-difluorodeoxyuridine were present in 65% and
88% of the samples, respectively. The analysis of cytostatics in wastewater is impor-
tant because of their possible negative effects on the aquatic environment. Therefore,
it is necessary to have knowledge about their occurrence in wastewater and surface
water and their fate during wastewater treatment.
7.3.3 Antibiotics
Antibiotics, compounds that kill or inhibit the growth of bacteria, have the potential
to affect the microbial community in sewage systems. The inhibition of wastewater
bacteria may seriously affect organic matter degradation and, therefore, effects of
antibacterial agents on the microbial population are of great interest. Antibiotics
have also been evidenced in surface water where they may affect organisms of dif-
ferent trophic levels.41 Antibiotics have various (polar) structures and were deter-
mined by bioassays, GC/MS, and RPLC, often with derivatization, with ion-pair
agents in the mobile phase, or even by ion-exchange techniques. Nowadays, HILIC
applications are expanding and include the analysis of aminoglycoside antibiotics
from biological samples,42 analysis of lincomycin and spectinomycin in liquid hog
manure supernatant, and simulated rainfall runoff from manure-treated cropland
(discussed in detail in Chapters 7 and 8 of this book).43
7.3.4 Miscellaneous Pharmaceuticals

The high concern of measuring pharmaceuticals as an important group of emerging
contaminants in the environment was expressed also by the U.S. Environmental

NH NH N NH2 O
N N NH2 S N+
H O N O–
Metformin Ranitidine
OH
H
N NH NH
HO N
S
N N N
HO H
Albuterol Cimetidine
Figure 7.6 Chemical structures of the pharmaceuticals analyzed by HILIC with EPA
1694 method.
Protection Agency (EPA) that issued a method for the analysis of pharmaceuticals
and personal care products in water, soil, sediments, and biosolids by HPLC-MS/
MS.44 Several classes of pharmaceuticals, 71 compounds in total, and 19 internal
standards were determined mainly by RPLC, but for the most polar of them, albuterol,
metformin, cimetidine, and ranitidine (Figure 7.6), a HILIC approach was preferred,
after sample pretreatment by SPE.
Metformin is a widely used anti-diabetic drug from the biguanide class, and one of
the most prescribed overall. It is not metabolized by the human body, being excreted
in the urine. Albuterol (salbutamol) is also widely used as bronchodilator prescription,
while ranitidine and cimetidine are histamine H2-receptor antagonists that inhibit
stomach acid production, being in the top 20 list of most prescribed drugs in different
European countries.45 Therefore, it is very important to evaluate the environmental
risk of these pharmaceuticals. SPE was also involved in the extraction of water and
sludge samples.44 An ultrasonic extraction with acetonitrile was performed for the
soil samples or the solid particles filtered from water. A volume of 500–1000â•›mL
filtered water was spiked with labeled internal standards, adjusted to pH = 10, and
loaded on an Oasis HLB cartridge for isolation of metformin, albuterol, ranitidine,
and cimetidine. The elution was done with methanol and then with 2% formic acid
solution in methanol. The extract was concentrated to nearly dryness, then 3â•›mL
of methanol was added and spiked with instrument-labeled internal standards, and
finally brought to 4â•›mL with 0.1% formic acid in methanol solution. For the determi-
nation of metformin in wastewater and surface water, Scheurer et al.46 have published
a method based on sample preparation with Strata™ X-CW SPE cartridges followed
by a separation on a ZIC-HILIC column (150 × 2.1â•›mm, 3.5â•›μm) with a mobile phase
consisting of (A) 10â•›m M ammonium formate in water set to pH = 3 with formic acid
and (B) acetonitrile, in gradient starting with 95% B. The chromatographic separa-
tion was coupled to a 4000 Q-Trap mass spectrometer equipped with an ESI ion
source for detection. Concentrations up to 129â•›μg/L in influent wastewater, 21â•›μg/L
in effluent wastewater, and 1.7â•›μg/L in surface water were measured.
The EPA method44 was set up using an HPLC system connected to a triple
quadrupole mass spectrometer, equipped with an ESI ion source operated in positive
mode. The chromatographic separation was carried out on a Waters Atlantis® HILIC

column (100 × 2.1â•›mm, 3â•›μm) eluted in gradient with a mobile phase consisting of (A)
ammonium acetate/acetic acid buffer 0.1% in water and (B) acetonitrile. The acquisi-
tion was done in multiple reaction monitoring (MRM) mode using one transition per
compound. An alternative to the EPA setup was proposed by Agilent.47 The sample
pretreatment is the same as in the EPA method, and the separation is performed on a
Zorbax® HILIC Plus column (100 × 2.1â•›mm, 3.5â•›μm) with a gradient of acetonitrile/
aqueous ammonium acetate 10â•›m M. A second MRM transition was added for con-
firmation of each compound.
7.3.5 Drugs of Abuse

One of the dark sides of modern society is the spread of DOA, which are consumed
to alter mood, thought, and feeling through their actions on the central nervous sys-
tem (brain and spinal cord). In addition, they are capable to produce dependence,
either physical or psychological.48 Even if these substances are severely regulated
and there are many efforts toward prevention, the underground consumption remains
quite high and their presence in the environment has been proven.
The analysis of DOA residues in wastewater and surface water is not only a preoc-
cupation of scientists but has become in recent years a tool to estimate the consump-
tion of these illicit substances in different countries.34,35,49–51 Using the excretion
patterns and considering the stability of the metabolites and of the unchanged DOA
in the environment, an overall estimation of the consumption can be made.35,51
HPLC methods for DOA (including cocaine) in RP mode have been published.52,53
Considering the highly polar structures, a method for detection and quantification of
cocaine and its principal metabolites (Figure 7.7) in wastewater and surface water
using HILIC was developed, validated, and applied to samples collected from 41
wastewater treatment plants (WWTPs) and rivers across Belgium.49–51,54 A signifi-
cant increase in sensitivity for all analytes was found compared to the reversed-
phase results. The target compounds were isolated by SPE on Oasis HLB cartridges.
Before pretreatment, water samples (100â•›m L for wastewater, 500â•›m L for river water)
were filtered through a glass microfiber filter and adjusted to pH = 6 with ammonium
hydroxide. Then the deuterated internal standards were added and the samples were
loaded on SPE columns preconditioned with methanol and water. After a washing
step with water/methanol (95/5, v/v) and the subsequent drying, the analytes and
internal standards were eluted with 2 × 4â•›m L methanol. The eluate was evaporated
until dryness and the samples were reconstituted with 150â•›μL acetonitrile/methanol
mixture (3/1, v/v).
O O
N N
OH N O
OMe
O OMe
O
O OH
O
Cocaine Benzoyl ecgonine Ecgonine methyl esther
Figure 7.7 Cocaine and its principal metabolites.

Analysis has been carried out on a system comprised of an autosampler, binary

pump, and ion trap detector with electrospray ionization (ESI).54 The interface was
operated in positive ion mode and the quantitative data were acquired in MRM
mode, measuring the specific transitions for each compound and its deuterated inter-
nal standard. HILIC separation was performed on a Zorbax Rx-SIL column (150 ×
2.1â•›mm, 5â•›μm) with a Zorbax Rx-Sil (12.5 × 2.1â•›mm, 5â•›μm) guard column, eluted
in gradient with a mobile phase composed by (A) ammonium acetate 2â•›m M/acetic
acid buffer (pH 4.5) and (B) acetonitrile, and the flow was 0.25â•›m L/min. The pump
program was as follows: 0–1â•›min 80% B, 1–10â•›min linear gradient up to 40% B,
10–18â•›min constant on 40% B, and 21–31â•›min re-equilibration of the column with
the initial conditions. A seven-point calibration curve based on deuterated internal
standards was generated for each analyte, and linearity was obtained for the follow-
ing ranges: 35–3300â•›pg for benzoylecgonine, 10–870â•›pg for cocaine, and 15–780â•›pg
for ecgonine methyl ester injected on column. Matrix effects were also evaluated by
comparing the analyte signal in samples (WWTP, river or tap water) with the ana-
lyte signal recorded for Milli-Q water. The ion suppression varied between 30% and
48% for wastewater and 12% and 22% for surface and tap water. To easily overcome
these matrix effects, deuterated internal standards have to be used for quantification.
With this method, 162 samples have been analyzed and the concentrations mea-
sured in Belgian WWTP samples ranged from 9 to 753â•›ng/L and 37 to 2258â•›ng/L
for cocaine and benzoylecgonine, respectively.49–51 Ecgonine methyl ester concen-
trations were below the method limit of quantification. A new HILIC method was
recently developed for the simultaneous determination of nine DOA and metabolites
(amphetamine, methamphetamine, MDMA, methadone, EDDP, 6-monoacetyl mor-
phine, cocaine, benzoylecgonine, and ecgonine methyl ester) in wastewater, with
improved quantification limits (down to 1â•›ng/L for all compounds, except for ecgo-
nine methyl ester, amphetamine, and 6-monoacetyl morphine for which quantifica-
tion limits were 2â•›ng/L).55 This method involves a chromatographic separation on a
Luna HILIC column (150 × 3â•›mm, 5â•›μm) and gradient with (A) ammonium acetate
5â•›m M in water and (B) acetonitrile, at a flow rate of 0.4â•›m L/min and starting with
95% B. The sample pretreatment was performed by SPE on Oasis MCX cartridges.
All compounds, except 6-monoacetyl morphine, could be quantified in a set of 12
wastewater samples from Belgium.
7.4 Pesticides
On its official Web site, the European Commission states that there is a certain
accepted risk associated with the use of pesticides, due to their direct benefits
generated for agriculture.56 Pesticides are employed on a large scale in modern
farms, mainly for plant protection and control of harmful organisms, such as
pests, but they also have unwanted dangerous effects on the human health and
environment if misused, carelessness, or even if used according to label direc-
tions. Therefore, these substances are regulated for a long time in the developed
countries, their environmental fate is studied, and the risks closely evaluated.56
Some pesticides or pesticide artifacts are particularly difficult to determine by
conventional analytical techniques, because of their highly hydrophilic character

O O O O O O
H
P P N P P
N S O S O O N S O
H O O O H
O NH2
Acephate Omethoate Monocrotophos Methamidophos
Figure 7.8 Chemical structures of various polar organophosphorus pesticides.
(that complicates their isolation from an aqueous compatible matrix), lack of

chromophore or fluorescent groups, and low molecular weights.
7.4.1 Organophosphorus Pesticides

Organophosphorus pesticides (OPPs) are cholinesterase inhibitors widely used for
crop protection. The possible contamination of drinking water is monitored; in the
European Union for instance, a maximum concentration of 0.1â•›μg/L is allowed.57
Most OPPs are easily analyzed by gas chromatography, but a few of them are very
polar and thermolabile (Figure 7.8).
Hayama et al.58 analyzed polar OPPs from water samples by HILIC-MS/MS.
Water samples (50â•›mL) were directly loaded to a preconditioned GL-Pak activated
carbon cartridge which was back-flush eluted with 5â•›mL of 0.2% (v/v) formic acid in
acetonitrile/isopropanol (95/5, v/v). The eluate was fortified with an internal standard
(2H6-acephate) and injected into the LC-MS system. For the separation of OPPs, an
Atlantis HILIC column (150 × 2.0â•›mm, 5â•›μm) was used with a mixture of acetonitrile/
isopropanol/200â•›mM ammonium formate in water (pH = 3) (92:5:3, v/v) in isocratic
conditions. Good retention times (range 3.4–4.9â•›min) were obtained for the target
compounds. The best sensitivity was achieved in acidic conditions (buffer pH = 3),
while the addition of isopropanol also improved the sensitivity. Though, only small
percentages of buffer and isopropanol ensured a good retention of OPPs on the HILIC
column. Hayama et al.58 applied this method to river water samples collected from
Chikugo River (Fukuoka, Japan). The spiked samples showed recoveries between
76% and 99%, similar to those obtained with distilled water.
7.4.2 Phenylurea Herbicides

Phenylurea herbicides are one of the most important classes of herbicides, mostly
used in cotton, fruit, and cereal production. Diuron is also used as total herbicide
in urban areas and as algicide in antifouling paints. Because of their extensive use
and poor biodegradation, several phenylurea herbicides have been often detected as
water pollutants.59 This class of compounds also has high polarity and water solu-
bility. Chemical degradation is negligible under normal pH and temperature con-
ditions, while microbial and photodegradation may help for the decontamination
of soil. Farre et al.60 evaluated the intermediates generated during the degradation
of diuron and linuron (Figure 7.9) by the photo-Fenton reaction using HILIC–MS/
MS. They purified wastewater and spiked it with linuron or diuron and with Fenton
reagent (FeSO4 and H2O2), and then irradiated the mixture for 60â•›min with a 6W
Philips black light with a measured intensity of 0.21â•›mW/cm2. The photodegradation

CH3
O
H H
Cl N N Cl N N
O O
Cl Cl
Diuron Linuron
Figure 7.9 Chemical structures of diuron and linuron.
products of linuron and diuron were investigated on a Nucleosil® Diol (150 × 4â•›mm,
5â•›μm) column, with a mobile phase composed by (A) 20â•›m M ammonium formate
buffer (pH = 3.3) and (B) acetonitrile, in gradient conditions. Farre et al.60 identified
by HILIC various amounts of methylurea and 1,1 dimethylurea in the photo-treated
solutions of herbicides, together with chlorinated structures which were not com-
pletely elucidated. This justifies, according to the authors, the different biodegra-
dability of the effluents.
7.4.3 Quaternary Ammonium Salt Herbicides

Diquat and paraquat, quaternary ammonium salt derivatives, are widely used herbi-
cides. In the United States alone, the estimated use of paraquat and diquat in 1997
was over 1500 and 100 tons, respectively.61 These compounds are particularly dif-
ficult to analyze by RPLC, due to their ionic structures (Figure 7.10), usually an ion-
pair reagent being added in the mobile phase. HILIC was reported as an alternative
by Makihata et al.62 and in a Waters Application Note.61
The isolation procedure for paraquat and diquat presented by Waters was based
on SPE using Oasis WCX cartridges.61 After loading, the columns were washed with
phosphate buffer pH = 7, water, and methanol, and then the herbicides were eluted
with ACN/water/TFA 84:14:2. The dried extracts were reconstituted with mobile
phase and 20â•›μL of the final solution were injected. The analysis of diquat and
paraquat was performed on an Atlantis HILIC (150 × 2.1â•›mm, 3.5â•›μm) column, the
mobile phase consisting of 40% acetonitrile, 60% aqueous buffer pH = 3.7 (250â•›m M
ammonium formate).
Makihata et al.62 have used a similar approach to determine diquat from Japanese
surface water (tap, river, lake, shallow well, and deep well water samples) collected
from Hyogo prefecture. An Atlantis HILIC column (150 × 2.1â•›mm, 5â•›μm) was used and
the mobile phase was acetonitrile/10â•›mM ammonium formate (pH = 3.7) (50/50, v/v).
In these conditions, the achieved limit of detection was 50â•›ng/L, 100 times lower than
2 Br– 2 Cl–
N+ N+
N+ N+
Diquat Paraquat
Figure 7.10 Chemical structures of the two common quaternary ammonium salt
herbicides.

the target value for diquat residues in Japan, and the herbicide was found in one of the
shallow well water samples, at a concentration of 2â•›μg/L.
7.5 Miscellaneous
A number of other applications which cannot be categorized in the previous chapters
have been found involving HILIC and contaminants of aquatic systems.
7.5.1 Dichloroacetic Acid

Dichloroacetic acid (DCA) is a small, polar compound that individuals may be
exposed to as a result of drinking water consumption. DCA is found in drinking
water as a disinfection by-product of chlorination or deriving from metabolization
of chlorinated solvents, such as trichloroethylene. The occurrence of DCA in drink-
ing water is of concern because DCA has been shown to be carcinogenic in labora-
tory animals. Most published methods use a derivatizing reagent that can artificially
increase the levels of DCA during the GC-MS analysis. Recently, a LC-MS/MS method
was validated for the quantitative analysis of DCA in drinking water.63 No other sam-
ple preparation than dilution was performed with the water samples. A Luna Amino
column (150 × 2.1â•›mm, 5â•›μm) was used for HILIC-ion exchange (the retention is
based on the affinity of the polar analyte for the charged end group on the stationary
phase) at a flow of 0.7â•›m L/min. The two components of the mobile phase were (A)
40â•›m M of ammonium formate and (B) acetonitrile. The gradient run was as follows:
90% B at 0â•›min, 30% B at 5â•›min, and 90% B at 6â•›min, followed by a re-equilibration
period of 9â•›min. The triple quadrupole mass spectrometer was run in ESI using the
MRM mode to monitor the transition from m/z 127 to 83. Method limits of quantifi-
cation as low as 5â•›ng/mL water could be established with this method. Bottled water
samples contained much less DCA than tap water samples. This is not a surprising
finding, as DCA is commonly found in tap water as a disinfection by-product of
chlorination.
7.5.2 Oligosaccharides
The massive accumulation of organic matter, which periodically occurs in the northern
Adriatic Sea, and in other locations worldwide, is presently thought to be the result
of the aggregation of dissolved organic matter (DOM) into particulate organic mat-
ter (POM). This phenomenon is the result of human activities and weather condi-
tions. Although many aspects are well understood, the trigger mechanisms leading
to mucilage formation have not been clarified yet, probably as a consequence of
inadequate analytical approaches. In this context, the new advancements in LC-MS
interfacing might contribute in clarifying the mechanism of mucilage formation.
Recently, HILIC coupled with ESI-MS/MS was proposed as an innovative method
for the investigation of underivatized oligosaccharides in mucilage samples.64
Although amino columns have been extensively used for derivatized sugar analysis,
HILIC represents a rare application in LC-ESI-MS/MS for the analysis of undeÂ�
rivatized oligosaccharides.64 After defrosting, the mucilage samples were isolated from

its surrounding seawater by decantation and next centrifuged at 4000â•›rpm for 20â•›min
at ambient temperature. After centrifugation, the precipitate was immediately dia-
lyzed against Milli-Q water overnight. The dialyzed sample was then lyophilized and
subsequently extracted with water. The aqueous suspension was then centrifuged at
4000â•›rpm for 20â•›min, and the supernatant, containing the water-soluble fraction of the
macro-aggregate, was finally subjected to LC-ESI-MS/MS analysis. Separations were
performed on an Alltech Alltima™ Amino column (250 × 2.1â•›mm, 5â•›μm) at a flow rate
of 0.150â•›mL/min. This approach allowed the efficient separation of the oligosaccharides
from the complex mucilage matrix without derivatization. A gradient elution program
was applied with (A) 10â•›mM sodium acetate and (B) acetonitrile. Solvent composition
varied from 10% to 70% A in 40â•›min. The injection volume was of 10â•›mL. The column
was connected to an ESI-quadrupole ion trap mass spectrometer for MS/MS analysis.
Mass spectra of oligosaccharides sodium adduct ions were acquired in positive ion mode.
Recent findings suggest that the significant presence of these compounds in sea-
water can play an important role in the initial steps of the agglomeration processes
forming gelatinous material. Furthermore, the presence of several maltodextrins was
evidenced in the water-soluble fraction of mucilage macro-aggregates, collected in
various locations of the northern Adriatic Sea.64
Acknowledgments
Alexander van Nuijs and Dr. Adrian Covaci are grateful to the Flanders Scientific
Funds for Research for financial support. Dr. Isabela Tarcomnicu acknowledges a
postdoctoral fellowship from the University of Antwerp.
References
1. Nguyen, H.P. and Schug, K.A., J. Sep. Sci., 2008, 31, 1465.
2. McFarren, E.F., Schafer, M.L., Campbell, J.E., Lewis, K.H., Jensen, E.T., and Schantz,
E.J., Adv. Food Res., 1960, 10, 135.
3. Berman, F.W. and Murray, T.F., J. Neurochem., 1997, 69, 693.
4. Quilliam, M.A. and Wright, J.L.C., Anal. Chem., 1989, 61, 1053.
5. Kao, C.Y., Pharmacol. Rev., 1966, 18, 997.
6. Horwitz, W., Ed., Official Methods of Analysis of AOAC International, AOAC
International, Gaithersburg, MD, 2000.
7. Harada, K.-I., Nagai, H., Kimura, Y., and Suzuki, M., Tetrahedron, 1993, 49, 9251.
8. Oshima, Y., J. AOAC Int., 1995, 78, 528.
9. Eaglesham, G.K., Norris, R.L., Shaw, G.R. et al., Environ. Toxicol., 1999, 14, 151.
10. Pietsch, J., Fichtner, S., Imhof, L., Schmidt, W., and Brauch H.-J., Chromatographia,
2001, 54, 339.
11. Shoji, Y., Yotsu-Yamashita, M., Miyazawa, T., and Yasumoto, T., Anal. Biochem., 2001,
290, 10.
12. Lawrence, J.F., Niedzwiadek, B., and Menard, C., J. AOAC Int., 2004, 87, 83.
13. Hall, S. and Strichartz, G., Eds., Marine Toxins: Origin, Structure, and Molecular
Pharmacology, American Chemical Society, Washington, DC, 1990.
14. Diener, M., Erler, K., Christian, B., and Luckas, B., J. Sep. Sci., 2007, 30, 1821.
15. Dell’Aversano, C., Eaglesham, G.K., and Quilliam, M.A., J. Chromatogr. A., 2004,
1028, 155.

16. Dell’Aversano, C., Hess, P., and Quilliam, M.A., J. Chromatogr. A., 2005, 1081, 190.
17. Krock, B., Seguel, C.G., and Cembella, A.D., Harmful Algae, 2007, 6, 734.
18. Ciminiello, P., Dell’Aversano, C., Fattorusso, E. et al., Rapid Commun. Mass Spectrom.,
2005, 19, 2030.
19. Ciminiello, P., Dell’Aversano, C., Fattorusso, E. et al., Toxicon, 2006, 47, 174.
20. Turrell, E., Stobo, L., and Lacaze, J.-P., J. AOAC Int., 2008, 91, 1372.
21. Ciminiello, P., Dell’Aversano, C., Fatturosso, E. et al., Toxicon, 2006, 47, 597.
22. Jang, J. and Yotsu-Yamashita, M., Toxicon, 2006, 48, 980.
23. Halstead, B.W., Ed., Dangerous Marine Animals: That Bite-Sting-Shock-Are-Non-
Edible, Cornell Maritime Press, Cambridge, MD, 1959.
24. Kim, Y.H., Brown, G.B., and Mosher, F.A., Science, 1975, 189, 151.
25. Diener, M., Christian, B., Ahmed, M.S., and Luckas, B., Anal. Bioanal. Chem., 2007,
389, 1997.
26. Jang, J. and Yotsu-Yamashita, M., Toxicon, 2007, 50, 947.
27. Nakagawa, T., Jang, J., and Yotsu-Yamashita, M., Anal. Biochem., 2006, 352, 142.
28. Bates, S.S., Bird, C.J., De Freitas, A.S.W. et al., Can. J. Fish. Aquat. Sci., 1989, 46, 1203.
29. Cerino, F., Orsini, L., Sarno, D., Dell’Aversano, C., Tartaglione, L., and Zingone, A.,
Harmful Algae, 2005, 4, 33.
30. Hess, P., Gallacher, S., Bates, L.A., Brown, N., and Quilliam, M.A., J. AOAC Int., 2001,
84, 1657.
31. de Koe, W.J., Samson, R.A., van Egmond, H.P., Gilbert, J., and Sabino, M., Eds.,
Mycotoxins and Phycotoxins in Perspective at the Turn of the Millenium, Wageningen,
the Netherlands, 2001.
32. Furey, A., Lehane, M., Gillman, M., Fernandez-Puente, P., and James, K.J., J.
Chromatogr. A., 2001, 938, 167.
33. Zuccato, E., Calamari, D., Natangelo, M., and Fanelli, R., The Lancet, 2000, 355, 1789.
34. Zuccato, E., Chiabrando, C., Castiglioni, S. et al., Environmental Health: A Global
Access Science Source, 2005, 4, 14.
35. Zuccato, E., Chiabrando, C., Castiglioni, S., Bagnati, R., and Fanelli, R., Environ.
Health Perspect., 2008, 116, 1027.
36. Routledge, E.J., Sheahan, D., Desbrow, C., Brighty, G.C., Waldock, M., and Sumpter,
J.P., Environ. Sci. Technol., 1998, 32, 1559.
37. Qin, F., Zhao, Y.Y., Sawyer, M.B., and Li, X.F., Anal. Chim. Acta, 2008, 627, 91.
38. Johnson, A.C., Juergens, M.D., Williams, R.J., Kümmerer, K., Kortenkamp, A., and
Sumpter, J.P., J. Hydrol., 2008, 348, 167.
39. Tauxe-Wuersch, A., De Alencastro, L.F., Grandjean, D., and Tarradellas, J., Int. J.
Environ. Anal. Chem., 2006, 86, 473.
40. Kovalova, L., McArdell, C., and Hollender, J., J. Chromatogr. A, 2009, 1216, 1100.
41. Kümmerer, K., Antibiotics in the aquatic environment—A review—Part I, Chemosphere,
2009, 75, 417.
42. Oertel, R., Neumeister, V., and Kirch, W., J. Chromatogr. A, 2004, 1058, 197.
43. Peru, K.M., Kuchta, S.L., Headley, J.V., and Cessna, A.J., J. Chromatogr. A, 2006, 1107,
152.
44. EPA method 1694, Pharmaceuticals and Personal Care Products in Water, Soil,
Sediment, and Biosolids by HPLC/MS/MS, 2007. http://www.epa.gov/waterscience/
methods/method/files/1694.pdf (accessed on October 15, 2010)
45. Fent, K., Weston, A.A., and Caminada, D., Aquat. Toxicol., 2006, 76, 122.
46. Scheurer, M., Sacher, F., and Brauch, H., J. Environ. Monit., 2009, 11, 1608.
47. Ferrer, I., Thurman, E.M., and Zweigenbaum, J., Agilent’s 6410A LC/MS/MS Solution
for Pharmaceuticals and Personal Care Products in Water, Soil, Sediment, and Biosolids
by HPLC/MS/MS, 2009. Agilent application note, http://www.chem.agilent.com/
Library/applications/5989–9665EN.pdf (accessed on October 15, 2010)

48. U.S. Drug Enforcement Administration (DEA), DEA Briefs & Background, Drugs
and Drug Abuse, Drug Descriptions, Drug Classes, 2010. http://www.usdoj.gov/dea/Â�
concern/drug_classes.html (accessed on October 15, 2010)
49. van Nuijs, A.L.N., Pecceu, B., Theunis, L. et al., Environ. Pollut., 2009, 157, 123.
50. van Nuijs, A.L.N., Pecceu, B., Theunis, L. et al., Water Res., 2009, 43, 1341.
51. van Nuijs, A.L.N., Pecceu, B., Theunis, L. et al., Addiction, 2009, 104, 734.
52. Castiglioni, S., Zuccato, E., Crisci, E., Chiabrando, C., Fanelli, R., and Bagnati, R.,
Anal. Chem., 2006, 78, 8421.
53. Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., and Ventura, F., Sci. Total
Environ., 2008, 397, 31.
54. Gheorghe, A., van Nuijs, A., Pecceu, B. et al., Anal. Bioanal. Chem., 2008, 391, 1309.
55. van Nuijs, A.L.N., Tarcomnicu, I., Bervoets, L. et al., Anal. Bioanal. Chem., 2009, 395,
819.
56. EU Commission. Sustainable Use of Pesticides, 2009. http://ec.europa.eu/environment/
ppps/home.htm (accessed on October 15, 2010)
57. Strosser, P., Pau Vall, M., and Plotscher, E., Water and Agriculture: Contribution to an
Analysis of a Critical but Difficult Relationship, 1999. http://ec.europa.eu/agriculture/
envir/report/en/eau_en/report.htm (accessed on October 15, 2010)
58. Hayama, T., Yoshida, H., Todoroki, K., Nohta, H., and Yamaguchi, M., Rapid Commun.
Mass Spectrom. 2008, 22, 2203.
59. Sorensen, S., Bending, G., Jacobsen, C., Walker, A., and Aamand, J., FEMS Microbiol.
Ecol., 2003, 453, 1.
60. Farre, M.J., Brosillon, S., Domenech, X., and Peral, J., J. Photochem. Photobiol. A:
Chem., 2007, 189, 364.
61. Young, M.S. and Jenkins, K.M., Waters application note, http://www.waters.com/
waters/library.htm?locale = en_us&lid = 10077196
62. Makihata, N., Yamasaki, T., and Eiho, J., Bunseki Kagaku, 2007, 56, 579–585.
63. Dixon, A.M., Delinsky, D.C., Bruckner, V.J., Fisher, J.W., and Bartlett, M.G., J. Liq.
Chromatogr. Relat. Technol., 2004, 27, 2343.
64. Cappiello, A., Trufelli, H., Famiglini, G. et al., Water Res., 2007, 41, 2911.

Chromatography–Mass
Spectrometry: An
Environmental
Application
Kerry M. Peru, Sandra L. Kuchta,
John V. Headley, and Allan J. Cessna
Contents
8.1 Introduction................................................................................................... 158
8.1.1 Pharmaceuticals in the Environment................................................. 158
8.1.2 Antimicrobials in the Environment................................................... 158
8.1.3 Analysis of Veterinary Antimicrobials.............................................. 160
8.1.4 Analysis of Spectinomycin and Lincomycin..................................... 160
8.1.5 Spectinomycin................................................................................... 161
8.1.6 Lincomycin........................................................................................ 162
8.1.7 Objective............................................................................................ 162
8.2 Experimental................................................................................................. 163
8.2.1 Chemicals and Reagents.................................................................... 163
8.2.2 Preparation of Standards and Solutions............................................. 163
8.2.3 Environmental Samples..................................................................... 163
8.2.3.1 Sample Collection............................................................... 163
8.2.3.2 Sample Preparation............................................................. 164
8.2.4 Solid-Phase Extraction...................................................................... 164
8.2.4.1 Spectinomycin..................................................................... 164
8.2.4.2 Lincomycin......................................................................... 165
8.2.5 Fortification Studies........................................................................... 166
8.2.6 LC/MS/MS Analysis......................................................................... 166
8.2.6.1 HILIC LC System............................................................... 166
8.2.6.2 Mass Spectrometer.............................................................. 166
8.3 Results and Discussion.................................................................................. 167
8.3.1 Extraction Method............................................................................. 167
157

8.3.2 Recovery Studies............................................................................... 167

8.3.2.1 Liquid Manure.................................................................... 167
8.3.2.2 Soil...................................................................................... 168
8.3.2.3 Water................................................................................... 168
8.3.3 Mass Spectrometry............................................................................ 168
8.3.4 Chromatography................................................................................ 170
8.3.5 Quantification.................................................................................... 173
8.3.6 Application of Method to Analysis of Environmental Matrices....... 173
8.4 Conclusions.................................................................................................... 174
References............................................................................................................... 174
8.1 Introduction
8.1.1 Pharmaceuticals in the Environment
Pharmaceuticals are used extensively in both human and veterinary medicine. Many
are excreted in the feces and urine as the nonmetabolized parent compound and,
consequently, may be present in municipal biosolids and livestock manure. Because
biosolids and manure are valuable sources of nutrients for crop growth, the use of
these materials as a fertilizer is widespread in many agricultural areas. Consequently,
use of biosolids and manure as fertilizer may be a potential source of pharmaceutical
contamination of surface and ground water via surface runoff and leaching, respec-
tively. Other sources of water contamination by pharmaceuticals include municipal
sewage discharge, aquaculture, and inappropriate disposal1 (Figure 8.1). Some phar-
maceuticals detected in environmental waters are known to exhibit hormonal activ-
ity with the potential to disrupt normal endocrine function.2 There is also concern
that the presence of veterinary antimicrobials in environmental waters may acceler-
ate the development of antimicrobial-resistant bacteria. While there is no clear evi-
dence that the transfer of antimicrobial resistance and that the selection for resistant
microbial agents occur at antimicrobial concentrations found in the environment, the
spread of resistant bacteria and resistance genes by use of biosolids and manure on
the agricultural landscape is not yet fully understood. Thus, in recent years, concern
over the impact, fate, and transport of both human and veterinary pharmaceuticals
entering the environment has gained much attention.3 In Canada, water quality has
become an important issue with the public both in terms of safety of drinking water
and in terms of protecting and conserving aquatic ecosystems.
8.1.2 Antimicrobials in the Environment

One class of pharmaceuticals commonly used in both human and veterinary medi-
cine is antimicrobials that are generally used to treat bacterial infections. However,
in livestock production, antimicrobials are routinely used not only at therapeutic
levels for the treatment of disease but also at subtherapeutic levels (milligrams per
kilogram of feed) to improve feed efficiency and to promote growth. Increasingly,
swine, poultry, and cattle are being produced in large confined animal feeding
operations that are also increasing in size. The number of animals housed in these

Hydrophilic Interaction Chromatography–Mass Spectrometry 159
Manufacturing
Livestock treatments process
Treatment of Aquaculture
companion animals treatments
Inappropriate disposal of
used containers and Wastewater
unused medicine treatment
Storage of
manure and slurry
Soil Receiving water

Manure/slurry spreading
Figure 8.1 Pharmaceutical routes of entry into the environment. (Reprinted from European
Molecular Biology Organization (EMBO), EMBO Rep., 5, 12, 1110, 2004. With permission.)
facilities increased by approximately two to three times in the United States over
14 years.4 A similar increase in swine numbers from 1990 to 2000 was observed in
Canada.5 There has been a corresponding increase on the reliance of antimicrobial
therapy with animals housed under high-density confined conditions that are condu-
cive for the transmission of infectious disease. Estimates of the total antimicrobial
use in North American agriculture are highly variable, ranging from 9 to 16 million kg
annually with between 10% and 70% being used subtherapeutically.6
Most antimicrobials are poorly absorbed through the gut and only partially
metabolized in the animal and, as a consequence, can be excreted up to 75% or more
in the feces and urine.7 Excretion rates for antimicrobials commonly used as feed
additives in beef cattle8,9 and swine production10,11 have been reported. There is also
considerable evidence that antimicrobials can also persist for prolonged periods in
livestock manure.10–15 Concentrations of antimicrobials present in lagoon manure
can be at milligrams per liter levels.12,16–18
The fact that residual antimicrobials persist in manure raises the possibility that,
after land application of the manure, they may persist in soil15,16,19–21 and potentially
come in contact with surface16,17,22–25 and ground water.22,26,27 Antimicrobials used in
the livestock industry have been detected in surface waters in Canada,28,29 the United
States,30–32 Europe,33,34 and Asia.35 Veterinary antimicrobials have been detected in
ground water following manure application16 and in ground water in the vicinity of
confined animal feeding operations in the United States, albeit in a limited number of
samples.12,32,36 There is limited information available on the biologic effect of antimi-
crobials in the environment. They have been shown to affect the composition of micro-
bial communities in soil37,38 and to affect bacterial populations in wetlands.39 However,

there are still major knowledge gaps in the environmental occurrence, transport, and
fate of antimicrobials and their biologic effects in terrestrial and aquatic ecosystems;
consequently, continued research in these areas of study is an ongoing requirement.
8.1.3 Analysis of Veterinary Antimicrobials

The study of the environmental occurrence, fate, and transport of antimicrobials
and other pharmaceutical compounds and the determination of the biologic impact
from short- and long-term exposure to these chemicals, generally requires the analy-
sis of environmental matrices (soil, sediment, water, plant and animal tissues, and
manure). To analyze and quantify these compounds at environmental concentrations
in such complex matrices requires sensitive and robust analytical methodologies.
Most pharmaceutical compounds are polar in nature and are soluble in water. Mass
spectrometry is well suited for the analysis of polar compounds because greater sen-
sitivity is generally achieved with pre-charged molecules when using atmospheric
pressure ionization (API) techniques. However, pre-charged molecules lack retention
on traditional reverse-phase (RP) liquid chromatography column packing materials
and analytes of interest often elute in or near the void volume of the column along
with interfering matrix components. This may lead to significant enhancement or
suppression of ionization in the mass spectrometer source, as well as possible mass
interference, by co-eluting matrix co-extractives. A great deal of time and expense
by column manufacturers and researchers has been invested in developing RP pack-
ing materials to achieve adequate polar analyte retention, often in conjunction with
additives (e.g., ion pairing reagents, pH adjustment) to enhance analyte retention
and separation. Success has been achieved, but sometimes at the expense of reduced
sensitivity due to the ionization suppression effects of these additives.40,41
8.1.4 Analysis of Spectinomycin and Lincomycin

Confined animal feeding operations require that infections be minimized for
profitable production. Escherichia coli infections causing diarrhea or edema dis-
ease in weanling pigs are estimated to cause up to 50% of the economic losses.42
Spectinomycin and lincomycin (Figure 8.2) are two commonly administered anti-
microbials used to control postweaning diarrhea. Spectinomycin is a member of the
aminocyclitol group of antimicrobials produced by a species of soil microorganism
designated as Streptomyces spectabilis. It is effective against a broad spectrum of
gram-positive and gram-negative bacteria by inhibiting protein synthesis through
interactions with the 30â•›S ribosomal subunit of bacterial cells.43 Spectinomycin is
excreted mostly unchanged in the feces and urine.44,45 Lincomycin is a member of the
lincosamide group of antimicrobials produced from the actinomyces S. lincolnensis.
It is effective against a variety of anaerobic and gram-positive bacteria by also inhib-
iting protein synthesis.45 Lincomycin undergoes some hepatic metabolism, however,
up to 32% can be excreted in the bile and urine.11,44,45
Waste management in confined animal feeding operations can be daunting. It
is estimated that each pig produces just over 2â•›m3 of waste annually and the liquid
manure management system results in a total of 2.5â•›m3 of manure per pig to process.46

CH3
N
CH3
HO C H
H2C C NH C H
OH
CH2 O
CH3HN O O CH3 O
H3C
HO SCH3
HO O
OH HO OH
(a) CH3HN O (b)
Figure 8.2 Chemical structures of (a) spectinomycin and (b) lincomycin. (Reprinted from
Peru, K.M. et al., J. Chromatogr. A, 1107, 152, 2006. With permission from Elsevier.)
As an economic means to waste management, liquid manure is frequently utilized

as a nutrient source on cropland. The applied manure is thus a potential source of
spectinomycin and lincomycin to the environment; however, little information has
been established on their environmental fate.
8.1.5 Spectinomycin
Previously, methods have been reported for the determination of spectinomycin using
microbial assays,47–52 gas chromatographic assay,53 thin layer chromatography/den-
sitometry,54 high-performance liquid chromatography (HPLC) with electrochemical
detection,55–57 HPLC with amperometric detection,58 and HPLC methods requiring
either pre- or post-column derivatization.59–65 These methods have mainly focused
on residues in animal tissues and milk for human consumption. Trifluoroacetic acid
(25â•›m M) has been used as an ion pairing agent to establish the retention of spectino-
mycin on an Agilent SB-C18 column.66 Detection was with evaporative light scatter-
ing, however, the detection limit (6â•›mg L−1 in aqueous samples) was insufficient for
an environmental level analysis.
Analysis of spectinomycin by LC-MS-MS has proved challenging largely because
of the high polarity and basic properties of the analyte. Chromatographic retention
is difficult to achieve and thus poor separation from co-eluting matrix components
leads to API enhancement and/or suppression interference.41,67 Previously, two meth-
ods have been reported for the determination of spectinomycin in tissues using ion-
pairing reagents with detection by LC-MS68 and LC-MS-MS.69 More recently, the
utility of heptafluorobutyric acid (HFBA) ion-pairing to aid in LC retention and
solid-phase extraction was reported.70,71 However, it has been observed that the
HFBA concentrations required for sufficient retention of spectinomycin and sepa-
ration leads to compromised sensitivity and detection limits when using API tech-
niques.41,67 An LC/MS/MS method utilizing a pH gradient was employed to better
retain and resolve spectinomycin from interfering co-eluting compounds.67 This
method provided improved detection limits compared to the HFBA method, elimi-
nated or significantly reduced the ionization inference in the MS source, and gave

reproducible results for samples of simulated rainfall runoff. However, the method
did not provide the required retention and reproducibility for analysis of the more
complex matrix of liquid hog manure.
8.1.6 Lincomycin
Methods have been reported for the determination of lincomycin in animal tissues
using GC with derivatization,72,73 HPLC with electrochemical detection in milk
and animal tissues;74 ion-pair HPLC with electrochemical detection in salmon tis-
sues,75 and HPLC with pulsed electrochemical detection in animal feed.76 Recently,
LC-MS-MS with electrospray ionization (ESI) has been used to determine linco-
mycin in surface waters,31,33 swine tissues,77 and milk.78 One method utilized both
a radioimmunoassay and LC-MS with ESI to determine lincomycin residues in
samples collected from swine lagoons.12 Ultra performance liquid chromatography
(UPLC) methods have also been reported using UV detection with success, however,
mass spectrometry was additionally required to exclude the misidentification of low
lincomycin levels.79 The use of LC-MS-MS with APCI has also been described for
the determination of lincomycin in honey.80 However, applications using APCI in
combination with HILIC separations are sparse.
Recently, research using HILIC separations has been increasing steadily, along
with various stationary phases developed for HILIC.40,81 HILIC was explored in this
work to improve the LC retention of the two antimicrobials investigated, particu-
larly spectinomycin. HILIC, based on silica columns, is an alternative to normal-
phase chromatography but utilizes traditional RP mobile phases.82 HILIC aids in
the retention of basic, polar compounds that are not retained well on RP columns.
Mobile phases containing a high organic solvent content (typically acetonitrile) and
a low water content are used to retain analytes depending on their hydrophilicity.
Retention is proportional to the polarity of the solute and inversely proportional to
the polarity of the mobile phase.83 Retention is therefore, the opposite of what is nor-
mally observed using RP chromatography. Several retention mechanisms, namely,
hydrophilic interaction, ion-exchange, and RP retention occur simultaneously when
using silica as the stationary phase. The combination of these retention mechanisms
offers unique selectivity and provides retention of polar solutes. These attractive fea-
tures of HILIC have been used to separate peptides84 and histones,85,86 and, more
recently, some polar pharmaceuticals,87–89 including the antimicrobials neomycin,90
avoparcin,91 and tetracycline, chlortetracycline, and oxytetracycline.92
8.1.7 Objective
As part of an investigation to fill knowledge gaps on the environmental fate and
transport of veterinary antimicrobials, we report a robust, specific, and sensitive ana-
lytical method for the determination of spectinomycin and lincomycin in liquid hog
manure, manure-treated soil, and snowmelt runoff and ground water from manure-
treated cropland. The APCI LC/MS/MS method described in this work uses hydro-
philic interaction chromatography (HILIC) to provide sufficient resolution to reduce
interference from the complex liquid swine manure matrix while providing a baseline

resolution between lincomycin and spectinomycin. This was achieved without using
mobile phase additives and, consequently, without compromising sensitivity.
8.2 Experimental
8.2.1 Chemicals and Reagents
Spectinomycin dihydrochloride (purity ≥98%) and lincomycin hydrochloride (purity
≥90%) were obtained from Sigma-Aldrich Co., (St. Louis, Missouri). HPLC grade
acetonitrile and methanol, and certified A.C.S. grade formic acid, ammonium
hydroxide, trisodium citrate, citric acid, and 20- to 30-mesh Ottawa sand were
obtained from Fisher Scientific (Edmonton, AB, Canada). ASE-prep diatomaceous
earth was obtained from Dionex (Sunnyvale, California). Deionized water (18â•›MΩ)
containing less than 4â•›μg L−1 total organic carbon was obtained using a Millipore
Milli-Q Gradient A10 (with a total organic carbon detector) purification system
(Millipore Corp., Billerica, Massachusetts).
8.2.2 Preparation of Standards and Solutions

Stock standard solutions (1000â•›mg L−1) of spectinomycin and lincomycin were pre-
pared by weighing and dissolving each antimicrobial in 50% aqueous acetonitrile.
The stock solutions were stored in the dark at 4°C for a maximum of 1 month. Nine
calibration standards were prepared (0.1–500â•›μg L−1) in mobile phase (acetonitrile/
H2O, 75:25, v/v containing 0.1% formic acid). Due to differences in detection limits
for spectinomycin and lincomycin, a subset of five calibration solutions (10–500â•›μg L−1)
was used to generate the calibration curve for spectinomycin, while another subset of
five calibration solutions (0.1–10â•›μg L−1) was used for lincomycin.
8.2.3 Environmental Samples

8.2.3.1 Sample Collection
Details of field sites, sampling procedures, and timelines are described elsewhere.16,17
8.2.3.1.1â•… Liquid Swine Manure

Liquid swine manure was collected from two commercial swine barns in
Saskatchewan, Canada. The liquid manure was typically a brown, opaque slurry
containing an average of 15% solids.11 In one barn, control manure was collected
from manure pits located under the pens of grower-finisher pigs that were not cur-
rently being administered antimicrobials. These samples were used as a matrix
blank and for fortification experiments. As part of a study to determine the excre-
tion of spectinomycin and lincomycin by weanling pigs, liquid swine manure was
also collected from manure pits located under the pens of weanling pigs housed in
the nursery area of the barn.11 These pigs were being administered approximately
equal amounts of spectinomycin and lincomycin in their feed to prevent and control
postweaning diarrhea. Details of liquid manure sampling procedures are described
elsewhere.11

Liquid swine manure from storage lagoons at both barns was applied in the fall to
cropland and pasture in 2003 and/or 2004 at rates ranging from 60,000 to 110,000â•›L
ha−1.16,17 Liquid manure samples were collected during manure application and would
permit the determination of the rates at which the antimicrobials were applied to the
agricultural landscape.
8.2.3.1.2â•… Soil
Manure-amended soil was sampled immediately after liquid manure application and
then monthly after snowmelt runoff the following spring.16 These samples would
permit the determination of the half-lives of lincomycin and spectinomycin in soil
and their presence in the upper layer of soil would confirm their availability for
transport in surface runoff.
8.2.3.1.3â•… Snowmelt Runoff

Spring snowmelt runoff from manure-amended soil was sampled from ephemeral
wetlands and closed basin depressions immediately after snowmelt was complete
and then weekly until the snowmelt had completely infiltrated.16,17 These samples
would confirm whether the management practice of utilizing liquid swine manure
as a plant nutrient source on crop and pasture land would result in the transport of
lincomycin and spectinomycin to aquatic ecosystems and would permit the determi-
nation of their persistence in the water column of the ephemeral wetlands.
8.2.3.1.4â•… Ground Water

Ground water from piezometers installed in manure-amended fields was collected
immediately after manure application and then monthly after spring snowmelt run-
off.16 These samples would confirm whether the management practice of utilizing
liquid swine manure as a plant nutrient source on crop and pasture land would result
in leaching of lincomycin and spectinomycin to ground water.
8.2.3.2 Sample Preparation

Liquid swine manure samples were separated into solids and liquid components as
described previously.11,41 Briefly, samples (10â•›m L) were centrifuged at 3000â•›g for
15â•›min followed by the decanting of the supernatant. This initial supernatant was
centrifuged two more times to maximize solids removal. The solids components
were combined for subsequent extraction.
8.2.4 Solid-Phase Extraction

8.2.4.1 Spectinomycin
8.2.4.1.1â•… Manure Liquid Component
A Waters Oasis HLB cartridge was stacked on top of a Waters Oasis WCX col-
umn for extraction and cleanup purposes. Cartridges were conditioned by sequen-
tially washing with 10â•›m L methanol followed by 10â•›m L deionized water. The clear
supernatant (2â•›m L ca) was diluted to 200â•›m L with deionized water and thoroughly
mixed. Dilution was required to aid in the extraction process and to minimize

potential losses from sample handling. A subsample (100â•›m L) was passed through
the stacked cartridges at approximately 1â•›m L min−1 (the remaining 100â•›m L was
used for lincomycin analysis). After sample extraction, the HLB cartridge was dis-
carded while the WCX cartridge was washed using 5â•›m L of 25â•›m M citrate buffer
(pH 5) followed by 5â•›m L of methanol. The cartridge was then eluted with 10â•›m L
of acetonitrile containing 3% formic acid. The eluate was taken to dryness using a
stream of N2 gas and the extract residue dissolved in acetonitrile (1.0â•›m L) for LC/
MS/MS analysis.
8.2.4.1.2â•… Manure Solids Component

The pressurized liquid extraction (PLE) procedure has been described previously.11
Briefly, a subsample (1â•›g) of the air-dried manure solids component was mixed with
diatomaceous earth and Ottawa sand to fill a 33 mL PLE cell. The packed PLE cell
was then extracted as follows: the cell was heated to 100°C and extracted for 5â•›min
with deionized water at a pressure of 1500â•›psi. The cell was then flushed with a
30% volume of deionized water and purged with nitrogen gas for 1.5â•›min. Optimal
extraction was achieved by extracting each cell twice. The resulting aqueous extracts
(∼30â•›m L each) were combined and diluted to 200â•›m L with deionized water and
extracted as described for the manure liquid component.
8.2.4.1.3â•… Manure-Amended Soil

Air-dried soil (2â•›g) was extracted by PLE as described for the manure solids com-
ponent. The resulting aqueous extracts (∼30â•›m L each) were combined and diluted
to 200â•›m L with deionized water and extracted as described for the manure liquid
component.
8.2.4.1.4â•… Surface and Ground Water

Surface or ground water (100â•›m L) was extracted as described for the manure liquid
component.
8.2.4.2 Lincomycin
8.2.4.2.1â•… Manure Liquid Component
An Oasis HLB cartridge was conditioned by sequentially washing with 10â•›m L ace-
tonitrile followed by 10â•›m L deionized water. The clear supernatant (2â•›m L ca) was
diluted to 200â•›m L with deionized water, and a subsample (100â•›m L) was adjusted to
pH 9 with 1â•›M ammonium hydroxide solution and passed through the cartridges at
approximately 1â•›m L min−1 followed by elution with 10â•›m L of acetonitrile. The eluate
was taken to dryness using a stream of N2 gas and the extract residue dissolved in
acetonitrile (1.0â•›m L) for LC/MS/MS analysis.
8.2.4.2.2â•… Manure Solids Component

Manure solids samples were subjected to the same PLE procedure as that used for
the analysis of spectinomycin. The combined PLE extracts (∼30â•›m L each) were
diluted to 200â•›m L with deionized water and extracted as described for the manure
liquid component.

8.2.4.2.3â•… Manure-Amended Soil

Air dried soil (2â•›g) was extracted by PLE as described for the manure solids compo-
nent. The combined PLE extracts (∼30â•›m L each) were diluted to 200â•›m L with deion-
ized water and extracted as described for the manure liquid component.
8.2.4.2.4â•… Surface and Ground Water

Surface or ground water (100â•›m L) was extracted as described for the manure liquid
component.
8.2.5 Fortification Studies

Fortification studies were carried out with all matrices studied (liquid manure,
manure liquid component, manure solids component, soil, and water) in order to
determine the reliability of the analytical methods. Liquid manure was fortified with
both spectinomycin and lincomycin at 100 (n = 6) and 10 (n = 6) μg L−1, manure
liquid component at 100 (n = 3), 10 (n = 3), and 1 (n = 3) μg L−1 and manure solids
component at 100 (n = 5), 50 (n = 4), and 1 (n = 5) μg kg−1. Soil was fortified with
both antimicrobials at 100 (n = 5) and 1 (n = 5) μg kg−1. South Saskatchewan River
water (as a surrogate for snowmelt runoff and ground water) and deionized water
were fortified with spectinomycin at 1â•›μg L−1 (n = 9) and lincomycin at 0.01â•›μg L−1
(n = 9). The fortification procedures have been described previously.11,16
8.2.6 LC/MS/MS Analysis

8.2.6.1 HILIC LC System
An Alliance 2695 Separations Module (Waters Corp., Milford, Massachusetts) con-
sisting of an LC pump, column oven, solvent degasser, and auto-sampler was used for
all chromatographic separations. HILIC separations were carried out using a silica-
based Altima HP hydrophilic interaction column (2.1â•›mm i.d. × 150â•›mm, 3â•›μm, Alltech
Associates Inc., Deerfield, Illinois), maintained at 36°C. The mobile phases were, A:
90/10 acetonitrile/water + 0.1% formic acid and B: 90/10 water/acetonitrile + 0.1%
formic acid. The sample injection volume was 15â•›μL. Separation was achieved with
isocratic conditions (200â•›μL min−1) of 65% mobile phase A and 35% mobile phase B.
8.2.6.2 Mass Spectrometer

All experiments were conducted on a Micromass Quattro Ultima (Waters Corp.,
Milford, Massachusetts) triple quadrupole mass spectrometer equipped with an
APCI interface operated in the positive ion mode. Optimized instrumental settings
were as follows: corona 0.20â•›m A, cone 60â•›V, source temperature 120°C, desolvation
temperature 550°C, cone gas flow rate of 153â•›L h−1 N2, desolvation gas flow rate of
199â•›L h−1 N2, nebulizer gas N2 maximum, collision energy of 30â•›V, and multiplier
700â•›V. Multiple reaction monitoring (MRM) was used for quantitative analysis with
the following settings: 0.5â•›s dwell time and 0.1â•›s interchannel delay while monitoring
two channels. Argon was used as the collision gas at a pressure sufficient to increase
the Pirani gauge of the collision cell to a reading of 2.41 × 10 −4 mbar.


8.3.1 Extraction Method
Methods for the extraction of lincomycin are numerous in the literature,31,33,72–75,80
as it is commonly included in multiresidue methods for the determination of other
antimicrobials. Lincomycin is amenable to solid-phase extraction (SPE) using tradi-
tional packings (C18, C8, etc.) and good recoveries are obtained from a variety of
matrices. In contrast, only one SPE extraction method along with data on the levels
of spectinomycin has been reported.70 This may be due in part to the difficulty of the
analysis. The analyte is poorly retained on RP packings leading to poor SPE recover-
ies and inadequate chromatographic retention on traditional RP LC packing materi-
als. Our approach was to use stacked SPE cartridges for all sample matrices, such
that the Oasis HLB cartridge was used for cleanup (removal of nonpolar compounds)
followed by the trapping of spectinomycin on a second cartridge containing a weak
cation-exchange sorbent (Oasis WCX). The WCX sorbent contains carboxylic acid
groups that allow protons to be exchanged with basic ionic compounds at pH 5 or
higher. Acetonitrile containing 3% formic acid is then passed through the WCX
cartridge that re-protonates the carboxylic acid groups, resulting in the elution of the
sorbed basic analytes. Improved chromatography (with reproducible retention times
and fewer matrix effects) was observed when the stacked cartridges were used versus
use of the WCX cartridge alone. This can be attributed to the retention of nonpolar
co-extractives on the HLB cartridge, resulting in less fouling of the WCX sorbent
and providing a cleaner final extract.
In this study, lincomycin was extracted from snowmelt runoff, ground water,
diluted liquid manure supernatant, diluted manure solids extract, and diluted soil
extract using an Oasis HLB cartridge. In order to suppress protonation of the amino
nitrogen and to achieve acceptable recovery, it was necessary to adjust water sam-
ples and diluted matrix extracts to pH = 9 using 25% ammonium hydroxide solu-
tion. This required step, unfortunately, circumvented the use of the HLB cartridge
used for cleanup during spectinomycin analysis for the simultaneous extraction of
lincomycin.
8.3.2 Recovery Studies

Method recoveries were determined by the extraction of fortified liquid swine
manure (liquid and solids components and whole manure), soil, deionized water, and
river water. The river water was used as a control sample because it contained no
detectable levels of the antimicrobials investigated and served as a matrix similar to
snowmelt runoff. Because similar recoveries of lincomycin or spectinomycin were
obtained from each matrix regardless of the fortification level, the recovery for each
analyte was determined as an average recovery for a given matrix.
8.3.2.1 Liquid Manure

The average recoveries of spectinomycin and lincomycin were 84 ± 2% and 78 ± 7%,
respectively, from the manure liquid component (n = 9). Recovery of both analytes
from the manure solids component was unreliable at the 1â•›μg kg−1 fortification level.

Average recoveries for the 100 and 10â•›μg kg−1 levels (n = 9) were 62% ± 12% and
70% ± 10%, respectively, for spectinomycin and lincomycin.11 Average recoveries
from liquid manure (liquid plus solids components, n = 12) were 69% ± 9% and
77% ± 14% for spectinomycin and lincomycin, respectively.11
8.3.2.2 Soil
The average recoveries of spectinomycin and lincomycin from fortified soil (n = 10)
were similar to those for the manure solids component and were 66% ± 15% and
74% ± 8%, respectively.16
8.3.2.3 Water
Average recoveries from deionized water (n = 9) were 108% ± 7% and 103% ± 10%
for spectinomycin and lincomycin, respectively,40 whereas corresponding recoveries
from river water (n = 9) were 95% ± 8% and 91% ± 4%.16,17,41
8.3.3 Mass Spectrometry

Prior to developing an alternative to ion-pairing for sufficient LC separation, MS
parameters were evaluated to determine the most sensitive ionization technique
available in our laboratory. This evaluation was performed by infusing a 1â•›mg L−1
solution of spectinomycin and lincomycin in 50:50 methanol/water, and for com-
parison in 50:50 acetonitrile/water. Both ESI and APCI in the positive ion mode
using either solvent produced an (M + H)+ ion for lincomycin (m/z 407) of a similar
intensity (Figure 8.3). However, in contrast to previously reported ion trap data,70
both ESI and APCI produced a very weak (M + H)+ at m/z 333 for spectinomycin
(Figure 8.4a and b) when using either solvent. Varying instrumental operating condi-
tions and a changing solvent composition did not improve the (M + H)+ abundance
100 407
Relative abundance (%)
0
315 335 355 375 395 415 435 455
m/z
Figure 8.3 Mass spectrum of lincomycin obtained using APCI using acetonitrile/water
as the solvent. (Reprinted from Peru, K.M. et al., J. Chromatogr. A, 1107, 152, 2006. With
permission from Elsevier.)

+
365 (M + H + CH3OH)
100
191
(M + H-H2O-NH2CH3)+
284
(M + H + H2O)+
351
385
(M + H)+ 396
236 333
0
100 140 180 220 260 300 340 380 420 460
(a) m/z
365 (M + H + CH3OH)+
100 351
(M + H + H2O)+
(M + H)+
333
0
235 255 275 295 315 335 355 375 395
(b) m/z
Figure 8.4 Mass spectrum of spectinomycin obtained using (a) ESI and (b) APCI using
methanol/water as the solvent. (Reprinted from Peru, K.M. et al., J. Chromatogr. A, 1107,
for spectinomycin. The corresponding ESI mass spectrum (Figure 8.4a) was noisy
with the ion current being averaged over many ions. Thus, monitoring the (M + H)+
did not provide adequate sensitivity for the quantitative analysis of spectinomycin
at levels present in environmental matrices. However, intense adduct ions (M + H +
H2O)+ and (M + H + CH3OH)+ were observed with APCI (Figure 8.4b) when using
methanol/water as the solvent and similarly (M + H + H2O)+ and (M + H + CH3CN)+
when using acetonitrile/water as the solvent. APCI was thus the ionization technique
of choice for both antimicrobials because it produced the most intense signals and,
consequently, the lowest detection limits. Acetonitrile/water was used as the mobile

407
100
126
0
50 90 130 170 210 250 290 330 370 410
(a) m/z
351
100
333
315
0
225 245 265 285 305 325 345
(b) m/z
Figure 8.5 Product ion mass spectrum of (a) lincomycin (M + H)+, m/z 407 and (b) water
adduct of spectinomycin (M + H + H2O)+, m/z 351.
phase for the liquid chromatographic separation of the antimicrobials because it pro-
vided a lower chromatographic system pressure and improved separation.
Product ion scans were completed to evaluate the best MRM transitions
(Figure 8.5a and b). The optimal MRM transitions monitored were m/z 351 > 333 for
spectinomycin and m/z 407 > 126 for lincomycin. Although monitoring the loss of
H2O for spectinomycin is not a highly specific transition, this compromise was made
to optimize detection limits.
8.3.4 Chromatography
Initial results using C8 or C18 HPLC column packing materials with a variety of
mobile phase compositions generated little to no retention of spectinomycin (Figure
8.6a) and resulted in matrix components affecting the ionization within the source

1.29
100
% Lincomycin
0
1.10
100
Spectinomycin
0
0.20 0.60 1.00 1.40 1.80 2.20 2.60 3.00
(a) Time (min)
3.63
100
Lincomycin
%
0
100 6.32
Spectinomycin
%
0
0.50 1.50 2.50 3.50 4.50 5.50 6.50
(b) Time (min)
Figure 8.6 LC chromatograms illustrating retention of spectinomycin and lincomycin

using (a) C18 analytical column and (b) HILIC analytical column. (Reprinted from Peru,
K.M. et al., J. Chromatogr. A, 1107, 152, 2006. With permission from Elsevier.)
of the mass spectrometer. We observed up to a 200% recovery for spectinomy-

cin due to ionization enhancement when elution of the analyte was at or near the
void volume. By utilizing HILIC, analyte retention was increased (Figure 8.6b)
while ionization enhancement and/or matrix interference was significantly reduced
or eliminated. Even under isocratic conditions, HILIC provided very good reten-
tion and separation from matrix components while providing baseline resolution
of lincomycin and spectinomycin (Figure 8.6b). Furthermore, the addition of for-
mic acid to the mobile phase to promote better source ionization, increased the

overall sensitivity, provided better peak shape, and improved the reproducibility of
the retention times of the analytes. The retention order of the two antimicrobials is
reversed from that on RP columns because more polar compounds are retained lon-
ger using HILIC. Figure 8.7 illustrates a chromatogram observed from each MRM
channel for both the WCX extract and the HLB extract (fortified liquid hog manure
supernatant).
6.27
100
351 > 333 Spectinomycin
channel
%
0
100
407.1 > 125.9 Lincomycin
channel
%
0
0.50 1.50 2.50 3.50 4.50 5.50 6.50
(a) Time (min)
3.68
100 Lincomycin
channel
407.1 > 125.9

%
0
100
351 > 333 Spectinomycin

channel
%
0
0.50 1.50 2.50 3.50 4.50 5.50 6.50
(b) Time (min)
Figure 8.7 Chromatogram of each MRM channel for (a) WCX cartridge cleanup of a
manure liquid component extract for spectinomycin analysis and (b) HLB cartridge cleanup
of a manure liquid component extract for lincomycin analysis. (Reprinted from Peru, K.M.

100 3.52
Lincomycin (m/z 407 > 125)
(from Oasis HLB extract)
0
100 6.43
Spectinomycin (m/z 351 > 333)
(from Oasis WCX extract)
0
0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00
Time (min)
Figure 8.8 Typical chromatogram of the lowest fortification level (1â•›μg L−1) of lincomy-
cin and spectinomycin in the manure liquid component. (Reprinted from Peru, K.M. et al.,
8.3.5 Quantification
Five-point calibration curves were generated for both spectinomycin and lincomy-
cin. Over the tested concentration ranges, the linear regression of observed peak
areas versus concentration gave excellent linearity with R2 values of 0.998 or greater.
Limits of quantitation were determined to be 5â•›μg L−1 (whole liquid manure),11 0.5â•›μg kg−1
(soil),16 0.2â•›μg L−1 (river water; spectinomycin),41 and 0.005â•›μg L−1 (river water; lin-
comycin).16 A typical chromatogram of the lowest fortification level of liquid swine
manure liquid component (1â•›μg L−1) is illustrated in Figure 8.8.
8.3.6 Application of Method to Analysis of Environmental Matrices

Over the course of several studies, more than 300 samples (including liquid swine
manure, manure-amended soil, snowmelt runoff, and ground water samples) have been
analyzed.11,16,17,41 The method has proven to be robust and required minimal ion-source
cleaning. Depending on the sample type and matrix loading, slight shifts in retention
times (up to ±0.4â•›min) for both spectinomycin and lincomycin were sometimes observed.
Both antimicrobials were excreted by weanling pigs and were detected in the liquid
manure collected from the nursery area of a commercial-scale barn.11 Concentrations
ranged from 173 to 686â•›μg L−1 for spectinomycin and 2520 to 9780â•›μg L−1 for lincomy-
cin.11 Liquid manure from the storage lagoon associated with this barn was applied
in the fall to crop and pasture land. Lincomycin was detected in manure samples
(25.1–231â•›μg L−1)16,17 collected during manure application, but spectinomycin was
not. Consequently, spectinomycin was not detected in any soil, snowmelt runoff,
or ground water samples. Lincomycin, however, was detected in manure-amended

soil (<0.5–117â•›μg kg−1),16 snowmelt runoff (<0.005–3.2â•›μg L−1),16,17 and ground water
(generally <0.005â•›μg L−1).16
8.4 Conclusions
In order to overcome the effects of matrix components on analyte ionization within
the source of the mass spectrometer, the separation of analytes from matrix com-
ponents must be achieved. Such a separation could not be achieved with traditional
reverse-phase liquid chromatographic packing materials for the analysis of spec-
tinomycin and lincomycin without sacrificing overall detection limits. In contrast,
hydrophilic interaction chromatography provided an excellent retention of spectino-
mycin and lincomycin resulting in separation of both antimicrobials from interfer-
ing matrix components. In addition, HILIC provided a baseline separation of the
two antimicrobials without the use of ion-pairing reagents. HILIC also facilitated
the use of formic acid as a mobile-phase additive that was compatible with API
techniques and provided good ionization efficiency. APCI provided intense ions that
were conducive to trace analysis for both antimicrobials using MS-MS. Weak cation-
exchange solid-phase extraction provided good recoveries for spectinomycin from
complex matrices such as liquid hog manure. Stacking an Oasis HLB cartridge on
top of the WCX cartridge provided the necessary cleanup prior to analysis in order to
reduce or eliminate the matrix effects and to provide reproducible chromatography.
The method was used to demonstrate that both spectinomycin and lincomycin
were present in liquid swine manure collected from the nursery area of a commercial-
scale barn. Likewise, it was shown that lincomycin was detectable in lagoon manure
and, after fall liquid manure application, in manure-amended soil, snowmelt runoff,
and ground water collected from manure-treated agricultural fields. These results
indicate that liquid swine manure applied to agricultural land as a nutrient source is
a potential source of veterinary antimicrobials to aquatic ecosystems.
Acknowledgments
The authors thank David Gallén and Sumith Priyashantha for field technical assistance,
Jonathan Bailey for laboratory technical assistance, Prairie Swine Centre Incorporated
and River Lakes Stock Farm barn personnel for information on lincomycin use and times
and rates of manure application, the collaborating producers for permission to sample on
their farms, and the operators of the manure applicators for manure sample collection.
The study was funded, in part, by the Saskatchewan Agriculture Development Fund.
References
1. European Molecular Biology Organization (EMBO), the Environmental Side Effects of
Medication, EMBO Rep., 2004, 5, 12, 1110.
2. Ternes, T.A., Joss, A., and Siegrist, H., Environ. Sci. Technol., 2004, 38, 393.
3. Richardson, S.D., Anal. Chem., 2008, 80, 4373.
4. United States General Accounting Office (GAO), Animal Agriculture: Information on Waste
Management and Water Quality Issues, GAO/RCED-95-200BR, Washington, DC, 1995.
5. Statistics Canada, Number of Farms Reporting Pigs and Average Number of Pigs per
Farm, 2002, Livestock Statistics, Cat. No. 23–603-UPE, http://www.statcan.ca.

6. Sarmah, A.K., Meyer, M.T., and Boxall, A.B.A., Chemosphere, 2006, 65, 725.
7. Kumar, K. et al., Adv. Agron., 2005, 87, 1.
8. EMEA, Tylosin Summary Report (Parts 1–5), 1994–2002,
Part 1: http://www.emea.europa.eu/pdfs/vet/mrls/tylosin1.pdf;
Part 2: http://www.emea.europa.eu/pdfs/vet/mrls/tylosin2.pdf;
Part 3: http://www.emea.europa.eu/pdfs/vet/mrls/020597en.pdf;
Part 4: http://www.emea.europa.eu/pdfs/vet/mrls/073200en.pdf;
Part 5: http://www.emea.europa.eu/pdfs/vet/mrls/082902en.pdf.
9. Halling-Sørensen, B. et al., Worst-case estimates of predicted environmental soil con-
centrations (PEC) of selected veterinary antibiotics and residues used in Danish agri-
culture. In: Kűmmerer, K., Ed., Pharmaceuticals in the Environment. Springer Verlag,
Berlin, Germany, 2001, p. 143.
10. Winckler, C. and Grafe, A., J. Soils Sediments, 2001, 1, 66.
11. Kuchta, S.L. and Cessna, A.J., Arch. Environ. Contam. Toxicol., 2009, 57, 1.
12. Campagnolo, E.R. et al., Sci. Total Environ., 2002, 299, 89.
13. Schlusener, M. et al., Arch. Environ. Contam. Toxicol., 2006, 51, 21.
14. Arikan, O.A. et al., Bioresour. Technol., 2006, 98, 169.
15. De Liguoro, M. et al., Chemosphere, 2003, 52, 203.
16. Kuchta, S.L. et al., J. Environ. Qual., 2009, 38, 1719.
17. Kuchta, S.L. and Cessna, A.J., Chemosphere, 2009, 76, 439.
18. Haller, M.Y. et al., J. Chromatogr. A, 2002, 952, 111.
19. Aga, D.S. et al., J. Agric. Food Chem., 2005, 53, 7165.
20. Jacobsen, A.M. et al., J. Chromatogr. A, 2004, 1038, 157.
21. Christian, T. et al., Acta Hydroch. Hydrob., 2003, 31, 36.
22. Dolliver, H. and Gupta, S., J. Envrion. Qual., 2008, 37, 1227.
23. Stoob, K. et al., Environ. Sci. Technol., 2007, 41, 7349.
24. Burkhardt, M. et al., J. Environ. Qual., 2005, 34, 1363.
25. Kay, P., Blackwell, P.A., and Boxall, A.B.A., Chemosphere, 2005, 59, 951.
26. Kay, P., Blackwell, P.A., and Boxall, A.B.A., Environ. Toxicol. Chem., 2004, 23, 1136.
27. Boxall, A.B.A. et al., Toxicol. Lett., 2002, 131, 19.
28. Forrest, F. et al., Livestock Pharmaceuticals in Agricultural Streams: A Scoping Study
for Alberta. Alberta Agriculture, Food and Rural Development, Edmonton, Alberta,
2006, p. 35.
29. Lissemore, L. et al., Chemosphere, 2006, 64, 717.
30. Yang, S. and Carlson, K., Water Res., 2003, 37, 4645.
31. Kolpin, D.W. et al., Envrion. Sci. Technol., 2002, 36, 1202.
32. Lindsey, M.E., Meyer, M., and Thurman, E.M., Anal. Chem., 2001, 73, 4640.
33. Calamari, D. et al., Environ. Sci. Technol., 2003, 37, 1241.
34. Hirsch, R. et al., Sci. Total Environ., 1999, 225, 109.
35. Managaki, S. et al., Environ. Sci. Technol., 2007, 41, 8004.
36. Batt, A.L., Snow, D.D., and Aga, D.S., Chemosphere, 2006, 64, 1963.
37. Vaclavik, E., Halling-Sørensen, B., and Ingerslev, F., Chemosphere, 2004, 56, 667.
38. Westergaard, K. et al., Soil Biol. Biochem., 2001, 33, 2061.
39. Sura et al., Presented at SETAC North America 29th Annual Meeting, Tampa, FL,
November 2008.
40. Rubio, S. and Pérez-Bendito, D., Anal. Chem., 2009, 81, 4601.
41. Peru, K.M. et al., J. Chromatogr. A, 2006, 1107, 152.
42. Stahl, C.H., Iowa State University Animal Industry Report, A.S. Leaflet R2025, Ames,
IA, 2005.
43. Murray, L., Ed., Physicians Desk Reference, 45th edn., Mont Vale, NJ, Medical
Economies, 1991, pp. 2260.
44. Burrows, G.E., J. Am. Vet. Med. Assoc., 1980, 176, 1072.

45. Aiello, S.E., Ed., The Merck Veterinary Manual, 8th edn., Merck, Whitehouse
Station, NJ, 1998.
46. Health Canada, Canadian Handbook on Health Impact Assessment—Volume 4: Health
Impacts by Industry Sector, Cat. No.: H46-2/04-363E, 2004.
47. Stahl, G.L. and Zaya, M.J., J. Assoc. Off. Anal. Chem., 1991, 74, 471.
48. El-Sayed, M.G.A. et al., Dtsch. Tierarztl. Wochenschr., 1995, 102, 446.
49. Tanaka, T., Ikebuchi, H., and Sawada, J.I., J. AOAC Int., 1996, 79, 426.
50. Shaikh, B. et al., J. AOAC Int., 1999, 8, 1002.
51. Medina, M.B. J. Agric. Food Chem., 2004, 52, 3231.
52. Thacker, J.D., Casale, E.S., and Tucker, C.M., J. Agric. Food Chem., 1996, 44, 2680.
53. Hoebus, J., Yun, L.M., and Hoogmartens, J., Chromatographia, 1994, 39, 71.
54. Krzek, J. et al., J. AOAC Int., 2000, 83, 1502.
55. Elrod, L., Bauer, J.F., and Messner, S.L., Pharm. Res., 1988, 5, 664.
56. Schermerhorn, P.G., Chu, P., and Kijak, P.J., J. Agric. Food Chem., 1995, 43, 2122.
57. Debremaeker, D. et al., J. Chromatogr. A, 2002, 953, 123.
58. Phillips, J.G. and Simmonds, C., J. Chromatogr. A, 1994, 675, 123.
59. Tsuji, K. and Jenkins, K.M., J. Chromatogr., 1995, 333, 365.
60. Myers, H.M. and Rindler, J.V., J. Chromatogr., 1979, 176, 103.
61. Burton, S.D. et al., J. Chromatogr., 1991, 571, 209.
62. Haagsma, N., Keegstra, J.R., and Scherpenisse, P., J. Chromatogr., 1993, 615, 289.
63. Haagsma, N. et al., J. Chromatogr. B, 1995, 672, 165.
64. Hornish, R.E. and Wiest, J.R., J. Chromatogr. A, 1998, 812, 123.
65. Bergwerff, A.A., Scherpenisse, P., and Haagsma, N., Analyst, 1998, 123, 2139.
66. Wang, J. et al., J. Chromatogr. B, 2006, 834, 178.
67. Peru, K.M. et al., Presented at 52nd ASMS Conference on Mass Spectrometry, Nashville,
TN, June 2004.
68. McLaughlin, L.G. and Henion, J.D., J. Chromatogr., 1992, 591, 195.
69. McLaughlin, L.G., Henion, J.D., and Kijak, P.J., Biol. Mass Spectrom., 1994, 23, 417.
70. Carson, M.C. and Heller, D.N., J. Chromatogr. B, 1998, 718, 95.
71. Zhu, W.X. et al., J. Chromatogr. A, 2008, 1207, 29.
72. Farrington, W.H. et al., Food Addit. Contam., 1987, 5, 67.
73. Luo, W. et al., J. Chromatogr. B, 1996, 687, 405.
74. Moats, W.A., J. Agric. Food Chem., 1991, 39, 1812.
75. Luo, W. et al., J. AOAC Int., 1996, 79, 839.
76. LaCourse, W.R. and Dasenbrock, C.O., J. Pharm. Biomed. Anal., 1999, 19, 239.
77. Sin, D.W.M. et al., Anal. Chim. Acta, 2004, 517, 39.
78. Crellin, K.C., Sible, E., and Van Antwerp, J., Int. J. Mass Spectrom., 2003, 222, 281.
79. Olsovska, J. et al., J. Chromatogr. A, 2007, 1139, 214.
80. Thompson, T.S. et al., J. Chromatogr. A, 2003, 1020, 241.
81. Ikegami, T. et al., J. Chromatogr. A, 2008, 1184, 474.
83. Peru, K.M. et al., Presented at 53rd ASMS Conference on Mass Spectrometry, San
Antonio, TX, June 2005.
85. Lindner, H. et al., J. Chromatogr. A, 1996, 743, 137.
86. Lindner, H., Sarg, B., and Helliger, W., J. Chromatogr. A, 1997, 782, 55.
87. Strege, M.A., Anal. Chem., 1998, 70, 2439.
88. Strege, M.A., Stevenson, S., and Lawrence, S.M., Anal. Chem., 2000, 72, 4629.
89. Olsen, B.A., J. Chromatogr. A, 2001, 913, 113.
90. Oertel, R., Renner, U., and Kirch, W., J. Pharm. Biomed. Anal., 2004, 35, 633.
91. Curren, M.S.S., and King, J.W., J. Chromatogr. A, 2002, 954, 41.
92. Valette, J.C. et al., Chromatographia, 2004, 59, 55.

9 Clinical Applications of
Liquid Chromatography
Ping Wang
Contents
9.1 Characteristics of Clinical Samples and Clinical Analysis........................... 178
9.2 Clinical Applications of HILIC..................................................................... 179
9.2.1 Sample Preparation and Concentration............................................. 179
9.2.2 Biomarker Profiling........................................................................... 179
9.2.3 Quantitation of Specific Analytes...................................................... 180
9.2.3.1 Serum/Plasma..................................................................... 180
9.2.3.2 Urine................................................................................... 186
9.2.3.3 Breath Condensate.............................................................. 190
References............................................................................................................... 191
Hydrophilic interaction liquid chromatography (HILIC) has found many applications

in clinical diagnostics and clinical research in recent years. HILIC is suitable for clini-
cal specimens such as plasma and urine, as many metabolites and biomarkers are preÂ�
sent in these matrices as polar forms. Using HILIC obviates the needs for ion-pairing
reagents necessary for polar analytes in reverse phase liquid chromatography (RPLC).
The relatively high concentration of organic solvents in HILIC mobile phase leads to
increased ionization signal in the subsequent mass spectrometry analysis. The compo-
sition of HILIC mobile phases is also friendlier to the HPLC pump seals than that of
the classical normal phase chromatography. Compared with gas chromatography (GC),
HILIC eliminates the hydrolysis and derivatization steps, and can preserve the intact-
ness of heat-sensitive molecules. HILIC may be interfaced with single or tandem mass
spectrometry, electrochemical detectors, and nuclear magnetic resonance spectroscopy
for high sensitivity and specificity. These characteristics make HILIC well suited for
the detection and quantitation of low-level biomarkers for early disease detection and
monitoring. HILIC may also be used for other clinical purposes, such as the speciation
and quantitation of contrast agents, metabolites of abused drugs, and toxic agents.
This chapter focuses on applications of HILIC in clinical diagnostics and clini-
cal research analyzing human clinical samples. Applications in animal studies will
not be included. Use of HILIC for pharmaceutical analysis will also not be covered,
although human samples may be used in these assays. The unique characteristics of
177

clinical samples and clinical analysis will be discussed. HILIC applications in three
categories will then be introduced. These include sample preparation and concentra-
tion, biomarker profiling, and quantitation of specific analytes. Examples encom-
passing applications in a diverse array of sample types such as plasma, urine, tissue,
and breath condensate will be given. A comparison will be made between the HILIC
method and other methods when applicable.
9.1 C
haracteristics of Clinical Samples
and Clinical Analysis
Various types of human fluid and tissue samples are analyzed in clinical laboratories
for the purpose of disease diagnosis or prognosis. These include serum, plasma, cere-
brospinal fluid, urine, dialysis fluid, hair, sweat, saliva, and solid tissues. Biomarkers
are parent forms or metabolites of endogenous proteins or small molecules, whose
levels in the above matrices change when diseases are present. Detection and quan-
titation of biomarkers provide valuable information for laboratorians and clinicians.
As most samples that are easy to procure for clinical analysis are aqueous, many
biomarkers of interest are present as polar forms in these matrices. Historically, both
GC and RPLC have been widely used in the clinical laboratories for the analysis
of polar biomarkers. This is because both technologies are well developed and the
expertise is readily available in the clinical laboratory environment. However, both
methods have limitations when it comes to clinical sample analysis. GC requires
extraction and derivatization of nonvolatile compounds, and may lead to heat-
induced fragmentation of analytes. High temperature in the injection port may
also cause heat-induced artifactual formation of derivatives. On the other hand,
ion-Â�pairing reagents are needed to retain polar analytes in order to be separated by
RPLC. Assays are often optimized individually with the optimal ion-pairing reagent
for each analyte. This may have practical concerns in clinical laboratories, espe-
cially when many different markers are analyzed on the same instrument. Switching
between different ion-pairing reagents and mobile phases may cause interference
between assays, increased run time, and erroneous results. Furthermore, the rela-
tively high concentration of salt in RPLC mobile phase suppresses electrospray ion-
ization, resulting in decreased detection sensitivity in interfaced mass spectrometry.
HILIC methods separate compounds by forming a water-rich layer on the surface
of the polar stationary phase. When analytes are introduced in mobile phases rich
in organic solvents, hydrophilic molecules are extracted from the mobile phase and
retained in the water-rich layer due to hydrogen bonding and dipole–dipole interac-
tions. Because of this mechanism, HILIC is well suitable for the analysis of polar
compounds such as carbohydrates, amino acids, polar metabolites, and peptides.1–3
The high organic content of the mobile phase in HILIC is favorable for electrospray
ionization efficiency when HILIC is interfaced with MS, which leads to enhanced
sensitivity compared to RPLC-MS.
As a result, HILIC is increasingly being used in clinical sample analysis in recent
years. New methods using HILIC have been developed for analytes that have tradi-
tionally been measured using GC or RPLC. At the same time, HILIC has enabled
clinical applications that have not been accomplished before.

Clinical Applications of Hydrophilic Interaction Liquid Chromatography 179
9.2 Clinical Applications of HILIC

9.2.1 Sample Preparation and Concentration
The ability of the polar stationary phase in HILIC to retain hydrophilic molecules
is exploited to concentrate hydrophilic components in human clinical samples. One
example is the enrichment of glycoproteins in plasma for N-glycosylation pattern
profiling.4–14 Protein glycosylation is an important posttranslational modification
that has diverse biological roles. Clinically, changes in glycosylation pattern have
ramifications in immune system diseases, cancers, and congenital disorders.15–17
The concentration of plasma glycoproteins is relatively low compared to other high-
abundance proteins such as albumin and immunoglobulins. To enhance the detec-
tion signal of glycans, HILIC solid phase extraction columns are used to concentrate
N-glycans, which are released from glycoproteins by peptide-N-glycosidase diges-
tion. N-glycans, either native or fluorescent tag-labeled, bind to the HILIC station-
ary phase, while hydrophobic peptides and other high-abundance matrix proteins
are removed in the flow-through fractions. Bound glycans are later eluted by water.
A variety of stationary phases have been used in glycoprofiling, including micro-
crystalline cellulose and Sepharose, synthetic polymer Diacovery DPA-6S, and resin
with a covalently bound zwitterionic sulfobetaine functional group. The HILIC col-
umns demonstrate great selectivity for the hydrophilic N-glycans. Eluted glycans can
be analyzed by matrix-assisted laser desorption ionization time-of-flight mass spec-
trometry (MALDI-TOF-MS), high-performance liquid chromatography (HPLC)
with fluorescence detection, capillary electrophoresis mass spectrometry (CE-MS),
or liquid chromatography tandem mass spectrometry (LC-MS/MS).
9.2.2 Biomarker Profiling

HILIC is a technique gaining increasing popularity for urinary metabolomics. Urine
is a commonly analyzed matrix for metabolomics studies. It can be obtained non-
invasively and contains a wealth of metabolites of endogenous as well as exogenous
molecules. A significant portion of the metabolites are hydrophilic to be soluble in
urine and excreted out of the body. This feature greatly facilitates the use of HILIC
in urinary metabolomics studies.
In a comparison of HILIC with RPLC to correctly classify mass spectrometry
data based on sex, diurnal variation, and age, statistically comparable results were
observed.18 To generate a complete picture, both positive and negative electrospray
ionization modes were used. It was concluded that HILIC is a comparable, if not bet-
ter separation method for LC-MS metabolomics studies. One should keep in mind
that urine samples used in the above work were from healthy volunteers. Under clini-
cal circumstances when the analytes of interest are likely to exist in polar forms in
the urine samples, one may opt for HILIC for metabolomic analysis of these sam-
ples, especially when the volume of sample is limited. Several method development
studies have demonstrated that very polar endogenous metabolites or xenobiotics not
resolved in RP separations can be separated and studied using HILIC coupled to MS
and nuclear magnetic resonance (NMR) spectroscopy.19–21 The stationary phase may
be either silica gel or silica modified with zwitterionic hydrocarbon chains.

The high organic solvent concentration in HILIC mobile phase is advantageous

for retention of analytes introduced in organic solvents. As an example, urinary
nucleosides in 100% acetonitrile were preconcentrated using an online solid phase
extraction column packed with aprotic boronic acid. Eluted nucleosides were sepa-
rated by HILIC to avoid the possibility that any acetonitrile entering the analytical
column may compromise retention of nucleosides on RP stationary phases. A HILIC
column packed with polyhydroxyethyl aspartamide provided robust separation of
22 urinary nucleosides, including structural isomers.22
In comprehensive urinary metabolomic analysis, HILIC, RPLC, and GC are usu-
ally used as complementary techniques to cover the majority of urine metabolites.
GC-MS may generate data for unconjugated metabolites with masses up to 500â•›Da,
while LC-MS covers conjugated metabolites up to 4000â•›Da.23 Thousands of features
may be identified using each of the separation method, the combination of which
provides a good coverage of individual metabolomic profiles. The approach may be
used for disease marker discovery or basic disease mechanism research.
9.2.3 Quantitation of Specific Analytes

9.2.3.1 Serum/Plasma
9.2.3.1.1â•… S-Adenosylmethionine
S-adenosylmethionine (SAM) has been suggested as a diagnostic marker for
Pneumocystis jirovecii pneumonia (PCP) in HIV-positive patients. It is synthesized
from methionine and ATP by SAM synthetase and serves as a universal methyl
donor in biometabolism. SAM quantitation requires a sensitive method. Plasma or
serum levels 60–160â•›nM (24–64â•›ng/mL) have been reported for healthy people; these
levels are further decreased in PCP-infected patients. Plasma SAM can be measured
using HPLC after derivatization with a fluorescent reagent.24 This method has the
drawback of requiring an expensive derivatization reagent and fluorescence detec-
tors are not routinely used or available in clinical laboratories. A sensitive coulo-
metric electrochemical method was also reported but had the same disadvantage of
not being amenable to routine clinical use.25 Methods using RPLC-MS/MS require
ion-pairing reagents in order for the polar SAM molecule to be retained by C18
columns.26–28 The presence of ion-pairing reagents can affect the performance of
other LC assays. In contrast, HILIC is ideal to retain and quantify the highly polar
SAM molecule.29 In one application, serum SAM was first extracted using solid-
phase extraction and separated on a silica column. The limit of quantitation (LOQ)
(intraassay coefficient of variation [CV] <20%) for SAM was 10â•›ng/mL. Using this
method, it was found that serum SAM concentration was correlated with fasting sta-
tus, especially methionine intake. Consistent with previous studies, SAM concentra-
tions in convalescent serum samples were significantly increased compared to acute
levels only in patients with PCP.29
9.2.3.1.2â•… Free Metanephrines

Plasma-free metanephrines are metabolic products of catecholamines, which are
excreted from pheochromocytoma cells in a continuous manner. Compared to
catecholamines, metanephrines are better markers for diagnosis and follow-up of

pheochromocytoma, because the plasma levels of metanephrines are less subjected

to environmental changes and metanephrines are more stable. Sensitive and specific
assays are demanded to quantify the low levels of plasma metanephrines and to
distinguish among the structurally similar metanephrine family members. HPLC
with electrochemical detection is currently most frequently used.30–32 These methods
invariably require laborious sample preparation and are subjected to interferences
of coeluting compounds. A cyano column-based LC-MS/MS method increases the
specificity but still requires manual offline solid-phase extraction.33 The total analy-
sis times for the above methods are long. Immunoassays provide rapid results but
have the disadvantage of cross-reactivity and nonspecific binding.34,35 The polar
basic nature of metanephrines is well suited for HILIC separation. Using a silica
stationary phase and gradient elution, a method was able to achieve lower limits of
detection (LODs) and LOQs than electrochemical methods with a plasma volume
of 50â•›μL.36 The use of online solid-phase extraction also reduced the total analysis
time compared to other assays.
9.2.3.1.3â•… Gadolinium-Based MRI Contrast Agents

Gadolinium-based contrast agents are used in magnetic resonance imaging to
enhance the contrast of images. In these compounds, gadolinium is complexed with
chelators to prevent the toxic effects of free gadolinium. The agents exhibit good
safety profile in most patients. However, a recently described nephrogenic systemic
fibrosis syndrome may arise in some terminally ill and renally impaired patients.
Decrease in renal function leads to increased accumulation of contrast agents, which
may result in skin and inner organ fibrosis. To better elucidate the pathogenesis
of the disease, sensitive and specific quantification of gadolinium-based contrast
agents is necessary. The total amount of contrast agents is quantifiable with element-
selective techniques such as inductively coupled plasma-optical emission spectros-
copy (ICP-OES), inductively coupled plasma-mass spectrometry (ICP-MS), and
atomic absorption spectroscopy. The individual contrast agent has been determined
using capillary electrophoresis,37–39 size exclusion chromatography with ICP-MS,40
ion exchange chromatography with ultraviolet detection,41 and RPLC.42–45 None of
the above methods are able to simultaneously detect several contrast agents. Most
provide only qualitative results. The ionic contrast agent species can be elegantly
separated by a ZIC-HILIC column and detected by electrospray mass spectrom-
etry, which provides good correlation with ICP-OES with regard to total amount of
gadolinium.46 Five different gadolinium-related contrast agents can be simultane-
ously quantitated using this method, with LODs (signal to noise ratio [S/N] = 3) in
the range 0.1–1â•›nM and LOQs in the range 0.5–5â•›nM. The drawback of this method
is the long chromatographic run time (gradient elution time of 30â•›m in and column
equilibration time of 20â•›m in).
9.2.3.1.4â•… Methylmalonic Acid

Cobalamin (vitamin B12) deficiency is a common cause of macrocytic anemia.
Methylmalonic acid is a nonvolatile aliphatic dicarboxylic acid that is related to the
catabolism of valine, isoleucine, and propionic acid. The conversion of methylmalo-
nyl coenzyme A to succinyl coenzyme A requires cobalamin. Therefore, cobalamin

deficiency causes increase in the level of methylmalonic acid, which can be used as a
biomarker for cobalamin deficiency. Due to the nonvolatile and polar nature of meth-
ylmalonic acid, GC-MS assays for methylmalonic acid require lengthy extraction and
derivatization steps,47 and RPLC does not show good retention. The physiological
concentration of plasma methylmalonic acid is less than 0.2â•›μM, which is much lower
than that of its structural isomer succinic acid. The low molar absorption coefficient
of methylmalonic acid makes direct ultraviolet detection of physiological concen-
tration of methylmalonic acid difficult. To increase assay sensitivity and selectiv-
ity, methylmalonic acid is derivatized with fluorescent agents in HPLC or capillary
electrophoresis methods, and detected by fluorescence detectors.48,49 Derivatization
is also usually required for LC-MS/MS methods,50–52 with some exceptions.53
A recently developed HILIC-MS assay took advantage of the bonded zwitterionic
stationary phase ZIC-HILIC to separate methylmalonic acid and succinic acid with-
out derivatization.54 Using an isocratic mobile phase of 4 volumes acetonitrile and
1 volume 100â•›m M ammonium acetate adjusted to pH 4.5 with formic acid, succinic
acid eluted in the void volume in the first 0.5â•›min. The LOQ (10 × standard deviation
[SD]) was 0.09â•›μM, and the LOD (3 × SD) was 0.03â•›μM, which was low enough for
the quantification of physiological methylmalonic acid levels. The HILIC method
has several advantages. First, labor-intensive and time-consuming derivatization is
avoided, which makes the assay suitable for automation and decreases the labor and
consumable costs. Secondly, succinic acid is eluted in the void volume, leaving only
methylmalonic acid, which can be detected by simple single-ion monitoring. Finally,
the high acetonitrile concentration in the mobile phase yields high electrospray ion-
ization efficiency in the mass spectrometer, leading to high analytical sensitivity.
9.2.3.1.5â•… Morphine and Metabolites

Morphine is an opioid drug used for pain relief. The main metabolites are morphine-3
glucuronide and morphine-6-glucuronide, of which morphine-6-glucuronide is bio-
logically active. For therapeutic monitoring and pharmacokinetic research purposes,
sensitive and specific quantitation methods that can be applied to various biological
matrices are desirable. RPLC methods with or without ion-pairing reagents have
been developed for plasma and microdialysis samples, with either MS or electro-
chemical detection.55–59 Most of these methods require solid-phase extraction to
clean up sample matrix. Desalting is especially important when analyzing micro-
dialysis samples using MS detectors, since the perfusion fluid contains a high con-
centration of salts in order to prevent fluid loss due to osmotic gradient. Electrospray
ionization is known to be susceptible to the ionization suppression effect of high
Â�concentration of salts. Since morphine glucuronides are hydrophilic compounds,
normal phase or HILIC-MS/MS methods have also been developed.60–62 The com-
parison of three normal phase or HILIC methods is listed in Table 9.1.
9.2.3.1.6â•… Ornithine
Ornithine is a basic amino acid that is an important component of the urea cycle,
which converts ammonia, the protein catabolic product, into the less toxic com-
pound urea. Ornithine is synthesized from arginine by arginase, as urea is generated.
Arginine can also be synthesized from ornithine and citrulline. Because arginine is

Clinical Applications of Hydrophilic Interaction Liquid Chromatography
Table 9.1
Comparison of the Three Normal Phase or HILIC-MS Methods for Morphine and Morphine Glucuronides
Sample Analytical Chromatographic
Method Sample Type Preparation Column Mobile Phase Run Time LOQ
Bengtsson et al.62 Microdialysis Online desalting ZIC-HILIC 70% acetonitrile + 30% 10â•›min M, M3, M6: 0.5, 0.22, 0.55â•›ng/mL
Fluid (5â•›μL), with C18 column ammonium acetate (microdialysis fluid); 0.78, 1.49,
plasma (100â•›μL) 0.53â•›ng/mL (plasma)
Naidong et al.60 Plasma (1.0â•›mL) Solid-phase Silica 1% formic acid + 10% 3â•›min M, M3, M6: 0.5, 10, 1â•›ng/mL
extraction with water + 90% acetonitrile
C18 column
Shou et al.61 Plasma (250â•›μL) Solid-phase Silica 11% water + 89% 48â•›s M, M3, M6: 1, 20, 2â•›ng/mL
extraction with acetonitrile, supplemented
Oasis HLB plate with 0.01% TFA
183
the source of nitric oxide generated in endothelial cells, ornithine is an indirect fac-
tor that influences nitric oxide generation rate and endovascular function. Ornithine
is also marketed as a nutritional supplement with the claim to increase anabolic
hormone levels and expedite healing process. Ornithine is traditionally analyzed
together with other amino acids. Amino acids in biological fluids are either sepa-
rated by ion exchange chromatography and then detected using the ninhydrin reac-
tion, or are first derivatized with o-phthalaldehyde and then separated by RPLC and
detected by fluorescence detectors.63–66 An RPLC-MS/MS method used ion-pairing
reagent to help retain underivatized amino acids on an RP column.67 The run times
of chromatography in the above methods are typically 30–120â•›min in order to sepa-
rate many different amino acids. A HILIC-MS/MS method dedicated to ornithine
shortened the run time to 4â•›min and required only a protein precipitation step for
sample preparation,68 which made the method amenable to high-throughput analysis.
The use of a silica column resulted in efficient ornithine retention. The LOQ was
determined to be 7.5â•›μmol/L, and the LOD (S/N = 3) was 0.1â•›μmol/L with 100â•›μL of
plasma. The method was used successfully to study the pharmacokinetics of orni-
thine after arginine supplementation.68
9.2.3.1.7â•… Quaternary Ammonium Compounds

The quanternary ammonium compounds, choline, betaine, and dimethylglycine, are
important precursors of the neurotransmitter acetylcholine and membrane phospho-
lipids. Betaine is linked to homocysteine and folate metabolism. The highly polar
quaternary ammonium compounds are well suited to be detected by HILIC-MS.
Methods using HILIC have been developed for acetylcholine, choline, and butyr-
obetaine in human liver tissues,69 and for choline, betaine, and dimethylglycine in
human plasma.70 Comparison of the two HILIC methods is presented in Table 9.2.
Historically, radioenzymatic assays,71,72 GC,73 GC-MS,74,75 HPLC with electro-
chemical detection (ECD)76–82 and RP HPLC-MS83–86 have been used to measure
choline or acetylcholine. The radioenzymatic assays are cumbersome, whereas GC
applications require extraction and derivatization. HPLC-ECD assays usually pro-
vide high sensitivity and specificity, but may produce unstable results from electrode
to electrode. In order to detect the hydrophilic quaternary ammoniums by HPLC-MS,
highly aqueous mobile phase, ion-pairing reagents, or ion exchange columns are used,
which lead to reduced ionization and decreased sensitivity. An RP HPLC method
for dimethylglycine and betaine requires derivatization,87 while GC-MS for beta-
ine necessitates an enzymatic conversion step.88 The HILIC-MS methods described
above provide simple and elegant solutions to detect the quaternary ammonium com-
pounds simultaneously.69,70
9.2.3.1.8â•… Tobacco-Specific Nitrosamine Metabolites

The tobacco-specific nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone
(NNK) is a suspected lung carcinogen. NNK is metabolized to 4-(methylnitrosamino)-
1-(3-pyridyl)-1-butanol (NNAL), which is then conjugated and excreted in urine as
NNAL-O-glucuronide and NNAL-N-glucuronide. Plasma and urine methods for
determining NNAL and NNAL–glucuronide concentrations have been developed
using GC-thermoelectric analysis.89–91 The methods typically require large volumes

Clinical Applications of Hydrophilic Interaction Liquid Chromatography
Table 9.2
Comparison of the Two HILIC Methods for Quaternary Ammonium Compounds
Compounds HILIC Stationary
Method Sample Type Detected Phase Linearity LOD LOQ Recovery
Wang et al.69 Liver tissue Acetylcholine (Ach), Polyhydroxyethyl A Ach:1.0–200â•›ng/mL; S/N = 3 S/N = 10 94%–105.7%
choline (Ch), Ch: 100–10,000â•›ng/mL; Ach: 0.2â•›ng/mL; Ach: 0.6â•›ng/mL;
butyrobetaine (Bb) Bb: 20–5,000â•›ng/mL Ch: 30â•›ng/mL; Ch: 80â•›ng/mL;
Bb: 2.0â•›ng/mL Bb: 15â•›ng/mL
Holm et al.70 Serum or plasma Choline, betaine, Hypersil silica 0.4–400â•›μM S/N = 5 N/A 87%–105%
dimethylglycine (40.8–46,800â•›ng/mL) 0.3â•›μM (30–35â•›ng/mL)
185
of plasma or urine, and have a chromatographic run time of 100â•›min. Extraction

and derivatization are also required as sample preparation steps. An RPLC-MS/MS
assay has an LOQ of 20â•›pg/mL using 15â•›m L of urine.92 Because of the accumulation
of basic compounds on the RP column, the method requires acetonitrile flushing of
the analytical column every 20 injections. Using a silica column and an isocratic
mobile phase, a HILIC-MS/MS assay requires only a simple liquid–liquid extraction
and has a short run time of 1â•›min.93 With a total plasma volume of 1.0â•›m L, both total
and free NNAL can be quantified with an LOQ of 5â•›pg/mL.
9.2.3.2 Urine
9.2.3.2.1â•… Acrylamide Metabolites
Low concentration of acrylamide is generated during heat preparation of carbo-
hydrate-rich foods.94 Concerns have been raised regarding increased cancer risks
caused by long-term exposure to acrylamide from food. Therefore, efforts have been
devoted to quantitate acrylamide metabolites in human urine to improve risk assess-
ment. Although RPLC-MS/MS methods have been developed,95–98 the highly polar
metabolites are not well retained on RP columns and usually elute close to or within
the void volume, which contains coeluting salts and matrix proteins. This causes ion
suppression when samples are introduced into the mass spectrometer. To avoid this
effect, interfering substances need to be cleaned up from urine samples using solid
phase extraction in RPLC-MS/MS methods.
Recently a HILIC-MS/MS method using zwitterionic sulfoalkylbetaine station-
ary phase was developed to bypass the problems associated with the RP methods.99
An online trap column was used to preconcentrate the acrylamide metabolites,
including glycidamide, iso-glycidamide, mercapturic acids, and mercapturic acid-
sulfoxide. ZIC-HILIC was able to separate the above metabolites well, including the
regioisomers of glycidamide and diastereomers of mercapturic acid-sulfoxide. The
LOD (S/N = 3) ranged 0.1–1â•›μg/L and the LOQ (S/N = 10) was 0.5–2.0â•›μg/L.
9.2.3.2.2â•… 1,5-Anhydroglucitol
Hemoglobin A1c and fructosamine are glycemic markers that reflect the glycemic
control over a period of 2–6 weeks. 1,5-Anhydroglucitol, on the other hand, is a
short-term marker that changes almost real-time with plasma glucose levels in dia-
betic patients. Renal tubule reabsorption of 1,5-anhydroglucitol is inhibited when
glucose concentration is high in glomerular filtrate.100 This causes decreased plasma
level and increased urinary level of 1,5-anhydroglucitol.101–103 In contrast, the plasma
concentration of 1,5-anhydroglucitol is very stable in euglycemic individuals.101,104,105
1,5-Anhydroglucitol is an extremely polar compound that does not retain on RP col-
umns. An enzymatic assay was developed to measure serum 1,5-anhydroglucitol,
with an LOD of 0.3â•›mg/L based on 0.5â•›m L of serum.106,107 Serum and urine assays
were also developed with GC-MS, which required sample preparation and derivatiza-
tion.105,108,109 The lowest LOD in urine by GC-MS was 0.06â•›μg/mL with 0.1–0.2â•›m L
of sample.110 Ion exchange HPLC methods had detection limits between 100 and
200â•›ng/mL for 1,5-anhydroglucitol in serum or urine.111,112 Cation-exchange chroma-
tography coupled to mass spectrometry had a similar detection limit of 200â•›ng/mL
with 0.5â•›m L of serum.113 More sensitive and specific detection for 1,5-anhydroglucitol

was achieved using a HILIC-MS3 method with a linear ion trap.114 Using this method,
only minimal sample preparation was involved and an LOQ of 50â•›ng/mL could be
achieved with 50â•›μL of human urine.
9.2.3.2.3â•…Arginine, Synthetic Dimethylarginine, and Asymmetric

Dimethylarginine
The importance of asymmetric dimethylarginine (ADMA) as a marker in cardiovas-
cular disease, renal failure, diabetes mellitus, essential hypertension, and hypercho-
lesterolemia is gradually being revealed in recent years.115–118 ADMA is a product of
arginine methylation during protein posttranslational modification.119 The methyla-
tion reaction is catalyzed by two types of protein arginine methyltransferases. Both
types catalyze the formation of monomethylarginine. Type I enzyme forms ADMA,
whereas type II enzyme also generates symmetric dimethylarginine (SDMA). Both
ADMA and SDMA are released into cytosol during protein catabolism. Most of the
ADMA is metabolized intracellularly and excreted into the urine. A small amount
of ADMA enters the plasma and may subsequently enter the endothelial cells.120
Recent studies suggest that ADMA is a competitive inhibitor of the endothelial
nitric oxide synthase,121,122 which generates nitric oxide (NO) from arginine. NO has
important physiological functions as a powerful vasodilator and inhibitor of vascular
wall adhesion of proinflammatory cells, platelet aggregation, and smooth muscle cell
proliferation. Although SDMA does not inhibit endothelial nitric oxide synthase, it
shares the same cell entry pathway with ADMA and therefore may influence the NO
production rate indirectly.123,124
Various methods have been developed for the determination of arginine, ADMA,
and SDMA in biological fluids. A commercial ELISA assay that measures only
ADMA is available, but shows no correlation with chromatographic results.125
Importantly, chromatographic methods offer better sensitivity, specificity, and
simultaneous detection of all the three compounds. The earliest RPLC methods
required derivatization with o-phthalaldehyde and detection by fluorescence.66,126–128
The run times for these methods are usually long in order to separate the structur-
ally similar ADMA and SDMA. Mass spectrometric detection obviates the need for
complete chromatographic resolution of the two compounds by taking advantage
of unique molecular transitions. Nonetheless, RPLC-MS methods require precol-
umn derivatization in order to retain the polar amino acids on RP columns.129,130
Two HPLC-MS assays achieved complete chromatographic separation of ADMA
and SDMA using a hypercarb column and an octadecylsilyl column,131,132 respec-
tively, but did not use isotopic labeled analogs as internal standards, which may
cause incorrect quantitation due to matrix effect.
Two HILIC-MS methods have been developed to quantitate arginine, ADMA,
and SDMA simultaneously.133,134 Both methods require only protein precipitation as a
minimal sample preparation step, use isotopic labeled analogs as internal standards,
and employ isocratic mobile phases and silica columns for separation. A detailed
parameter optimization study was carried out to explore the optimal mobile phase
composition, percentage of trifluoroacetic acid, and column temperature.135 Both
methods were simple and robust enough to be applied to the analysis of clinical
samples. Statistically similar results were reported for plasma arginine, ADMA, and

Table 9.3
Comparison of the Two HILIC-MS Methods for Arginine
and Its Dimethylated Derivatives
Complete
Chromatographic
Internal Resolution of
Method Sample Type Mobile Phase Standards ADMA and SDMA
Martens- Plasma (100â•›μL), 10% water + 90% 13 C6-Arginine, No
Lobenhoffer urine (50â•›μL), cell acetonitrile, 2 D6-ADMA
et al.133 culture supernatant supplemented with
(100â•›μL) 0.025% TFA and
1% propionic acid
D’Apolito Plasma (0.02â•›mL) 10% water + 90% 13 C6-Arginine Yes
et al.134 acetonitrile,
supplemented with
0.025% TFA and
1% acetic acid
SDMA levels in healthy individuals using the two methods. One method detected
statistically significant differences between arginine derivative levels in healthy sub-
jects and those of type 2 diabetic patients.134 The comparison of the two methods is
listed in Table 9.3.
9.2.3.2.4â•… Cocaine and Metabolites

Cocaine (benzoylmethyl ecgonine) is one of the most often abused recreational drugs.
Many cocaine-related deaths are reported every year. Cocaine is a strong central ner-
vous system stimulant that acts by blocking the reuptake of dopamine and serotonin.
Physiological responses to cocaine include constricted blood vessels, increased body
temperature, increased heart rate, and blood pressure. When ingested in large doses,
cocaine can cause irritability, paranoid psychosis, myocardial infarction, stroke, and
sudden death.
Cocaine is rapidly hydrolyzed by plasma and liver carboxylesterases to benzoy-
lecgonine and ecgonine methylester. If cocaine is administered by smoking, pyrol-
ysis of cocaine results in the formation of methylecgonidine or anhydroecgonine
methylester. Cocaine also undergoes metabolism in the liver by the Cytochrome
P450 enzymes. N-demethylation of cocaine produces norcocaine, while aromatic
hydroxylation produces hydroxycocaine and hydroxybenzoylecgonine. When
cocaine is coingested with ethanol, cocaethylene is formed from the two compounds.
Cocaethylene is further metabolized into norcocaethylene, benzoylecogonine, and
ecgonine ethylester. Cocaine and its metabolites form a family of basic compounds
with a wide range of polarity, which are traditionally detected using GC-MS meÂ�thods
with derivatization. One of the drawbacks of GC-MS methods is the possibility of
artifactual formation of anhydroecgonine methylester. This possibility is avoided in
RPLC-MS methods, which require solid-phase extraction of urine samples before

chromatography.136–139 A HILIC-MS method was reported that used a silica col-

umn to separate cocaine and eight of its metabolites.140 Blood, bile, urine as well as
brain tissue could be analyzed following chloroform/isoporanol extraction, evapora-
tion, and reconstitution. The LOQ was 5â•›ng/mL and the LOD (S/N = 3) was below
0.5â•›ng/mL for most of the metabolites analyzed. This method was used successfully
to detect and quantitate cocaine and metabolites in a case of fatal overdose.140
9.2.3.2.5â•… Estrogen Metabolites

Estrogen is metabolized through conjugation with sulfate and glucuronide moieties.
Various estrogen conjugates are then excreted in the urine. The concentration of
free estrogen in urine is usually extremely low. Estrogen metabolism varies signifi-
cantly between individuals, suggesting the profile of urinary estrogen metabolites
may provide important physiological and pathological information. Immunoassays
are widely used in the clinical laboratory for estrogen measurement. However,
these assays are usually subjected to cross-reactivity and poor precision at low
concentrations. Labor-intensive sample preparation steps including hydrolysis and
derivatization are needed in GC-MS assays, as estrogen metabolites are highly
polar.141–144 HPLC assays have been developed with electrochemical detection for
estrogens and metabolites in human tissues,145 but suffer from low sensitivity and
background interference. Several RPLC-MS methods determine estrogens with or
without derivatization with toluenesulfonhydrazide or dansyl chloride. Estrogen
conjugates are hydrolyzed and converted to estrogen for quantification in these
methods.
Â� 146–152 Besides time-consuming derivatization and hydrolysis, the disad-
vantage of these assays is the inability to quantify individual estrogen conjugates

and isomers. RPLC-MS/MS methods have also been developed to quantify different
estrogen conjugates.153,154 The RPLC-MS/MS method with the highest sensitivity has
an LOQ of 0.02–0.6â•›ng mL for seven estrogen conjugates based on 5â•›m L of urine.154
A HILIC-MS/MS assay provides 2- to 10-fold higher sensitivity for the same metab-
olites than that of the RPLC-MS/MS method, with 1â•›m L of urine sample.155 This is
due to the higher content of organic solvent in the mobile phase, which enhances
the ionization efficiency in the electrospray ion source. The assay uses an analyti-
cal column packed with derivatized silica (carbamoyl groups attached to silica via
an aliphatic carbon chain). Urine samples undergo a solid-phase extraction using a
Waters Oasis HLB cartridge. Complete chromatographic separation of all conjugates
is not necessary due to the unique MRM transitions. Using this HILIC method, the
unique profiles and changes of urinary estrogen conjugates during different stages of
pregnancy were obtained. The assay was also able to demonstrate individual-specific
urinary estrogen conjugate profiles in postmenopausal breast cancer patients before
and after aromatase inhibitor treatment.155
9.2.3.2.6â•… Organophosphorus Nerve Agent Metabolites

The organophosphorus nerve agents are extremely toxic substances that could be
used as terrorist weapons. The agents include sarin (GB), soman (GD), tabun (GA),
cyclohexylsarin (GF), VX, and rVX (VR). These compounds act by binding and
reacting with the active site of acetylcholinesterase, inhibiting the hydrolysis of
acetylcholine in the neural synapse. Excess acetylcholine results in overstimulation

of the acetylcholine receptors in the neuromuscular junction, leading to paralysis

of muscles. Death may result from the paralysis of the respiratory muscles. It is
thought that these agents are metabolized within the human body by hydrolysis
breakdown. Typical metabolites analyzed to assess human exposure to the organo-
phosphorus nerve agents are the alkyl methylphosphonic acids. Various GC-MS
methods were developed to quantify the alkyl methylphosphonic acids.156–162
Methods without chromatographic separation have also been developed.163–165 In
clinical and forensic analysis, isotope dilution GC-MS or GC/MS/MS methods
are usually used, with solid-phase extraction and derivatization required. The
polar acidic nature of the metabolites makes HILIC an attractive alternative. In an
effort to maximize the signal-to-noise ratio generated by the mass spectrometer,
HILIC was also found to be optimal in order to maintain the high acetonitrile
content of the mobile phase.166 Using a solid-phase extraction procedure with non-
bonded silica and a silica analytical column, the LOD in the range 30–240â•›pg/mL
was achieved for five alkyl methylphosphonic acid metabolites. When compared
to an established GC-MS/MS method160 using the same blind unknown urine
samples distributed to six different laboratories, the HILIC-MS/MS method had
comparable accuracy for three metabolites and better accuracy for the other two
metabolites.166
9.2.3.2.7â•… Pheomelanin Hydrolysis Products

Pheomelanin is a skin pigment thought to be responsible for the higher incidence
of malignant melanomas in individuals with light colored skin than those with dark
skin. It is often excreted in the urine of human melanoma patients and can be used
as a disease marker. As pheomelanin is formed by oxidative polymerization of dif-
ferent bezothiazine derivatives, analysis of pheomelanin is usually accomplished by
measurement of the reductive hydrolysis products. Two major hydrolysis products,
4-amino-3-hydroxyphenylalanine (4-AHP) and 3-amino-4-hydroxyphenylalanine
(3-AHP), are formed after hydriodic acid reduction of pheomelanin.167 Of the two
compounds, 4-AHP is a more specific indicator of pheomelanin, as 3-AHP can be gen-
erated from reduction of other compounds.168,169 Early HPLC methods cannot resolve
the peaks,170 and therefore are not optimal for quantitation of pheomelanin. RPLC
methods can separate the two peaks by using ion-pairing reagents,168,171–173 which
may cause problems in subsequent ionization in mass spectrometer. A Â�silica-based
HILIC stationary phase with covalently bonded zwitterionic sulfabetaine groups
separates 4-AHP and 3-AHP well and is suitable for mass spectrometric analysis.174
When HILIC is coupled to an electrochemical detector, the simple mobile phases,
long column life, and robust performance of the method enable quantitation of phe-
omelanin in urine, hair and melanoma tissues, structural analysis of pheomelanin,
and stability study of hydrolysis products. The LOD (S/N = 3) was as low as 73â•›pg
for 4-AHP and 51â•›pg for 3-AHP, respectively.174
9.2.3.3 Breath Condensate

9.2.3.3.1â•… 3-Nitrotyrosine, Tyrosine, Hydroxyproline, and Proline
Monitoring of biomarkers in breath condensate is a noninvasive method for detec-
tion of diseases related to the lung. Nonvolatile compounds such as amino acids are

exhaled in droplets of the bronchoalveolar lining fluid in the airflow. As the breath
condensates on a cold surface, the analytes are collected in the aqueous phase.
3-Nitrotyrosine is formed from the reaction of tyrosine with peroxynitrite, which is a
reactive form of nitric oxide especially during inflammation. It has been discovered that
the ratio between 3-nitrotyrosine and tyrosine is a good marker of asthmatic inflamma-
tion. On the other hand, hydroxyproline is supposed to be a marker for collagen turnover.
The ratio between hydroxyproline and its precursor proline is hypothesized to be a good
marker for asbestosis, fibrotic reaction of the lung. The extremely low concentration of
these amino acids in breath condensate calls for a highly sensitive method. Although
determination of 3-nitrotyrosine alone in exhaled breath condensate has been achieved
using enzyme immunoassay,175 GC-MS/MS,176 GC-MS(NCI),177 and LC-MS/MS178,179
after extensive sample preparation, the simultaneous measurement of nitrotyrosine,
tyrosine, proline, and hydroxyproline is beneficial. This is because the variation of ana-
lyte dilution by exhaled water vapor may be internally corrected for by the measure-
ment of tyrosine and proline. The first method that simultaneously quantifies the four
amino acids was developed using a ZIC-HILIC column, without any need for derivatiza-
tion.180 The condensate sample was lyophilized and reconstituted. The LOD (S/N = 3)
was 5â•›ng/L for hydroxyproline and nitrotyrosine and 0.05â•›μg/L for proline and tyrosine,
which was favorable compared to previous detection limits in breath condensate.
References
1. Alpert, A. J. 1990. Hydrophilic-interaction chromatography for the separation of pep-
tides, nucleic acids and other polar compounds. J. Chromatogr. 499: 177–196.
2. Ikegami, T.; Tomomatsu, K.; Takubo, H.; Horie, K.; Tanaka, N. 2008. Separation effi-
ciencies in hydrophilic interaction chromatography. J. Chromatogr. A 1184: 474–503.
3. Hemstrom, P.; Irgum, K. 2006. Hydrophilic interaction chromatography. J. Sep. Sci. 29:
1784–1821.
4. Anumula, K. R.; Dhume, S. T. 1998. High resolution and high sensitivity methods for
oligosaccharide mapping and characterization by normal phase high performance liq-
uid chromatography following derivatization with highly fluorescent anthranilic acid.
Glycobiology 8: 685–694.
5. Hagglund, P.; Bunkenborg, J.; Elortza, F.; Jensen, O. N.; Roepstorff, P. 2004. A new strat-
egy for identification of N-glycosylated proteins and unambiguous assignment of their
glycosylation sites using HILIC enrichment and partial deglycosylation. J. Proteome.
Res. 3: 556–566.
6. Omaetxebarria, M. J.; Hagglund, P.; Elortza, F.; Hooper, N. M.; Arizmendi, J. M.;
Jensen, O. N. 2006. Isolation and characterization of glycosylphosphatidylinositol-
anchored peptides by hydrophilic interaction chromatography and MALDI tandem
mass spectrometry. Anal. Chem. 78: 3335–3341.
7. Prater, B. D.; Anumula, K. R.; Hutchins, J. T. 2007. Automated sample preparation
facilitated by PhyNexus MEA purification system for oligosaccharide mapping of gly-
coproteins. Anal. Biochem. 369: 202–209.
8. Royle, L.; Campbell, M. P.; Radcliffe, C. M. et al. 2008. HPLC-based analysis of serum
N-glycans on a 96-well plate platform with dedicated database software. Anal. Biochem.
376: 1–12.
9. Ruhaak, L. R.; Huhn, C.; Waterreus, W. J. et al. 2008. Hydrophilic interaction chroma-
tography-based high-throughput sample preparation method for N-glycan analysis from
total human plasma glycoproteins. Anal. Chem. 80(15): 6119–6126.

10. Tajiri, M.; Yoshida, S.; Wada, Y. 2005. Differential analysis of site-specific glycans on
plasma and cellular fibronectins: Application of a hydrophilic affinity method for glyco-
peptide enrichment. Glycobiology 15: 1332–1340.
11. Thaysen-Andersen, M.; Thogersen, I. B.; Nielsen, H. J. et al. 2007. Rapid and
Â�individual-specific glycoprofiling of the low abundance N-glycosylated protein tissue
inhibitor of metalloproteinases-1. Mol. Cell. Proteomics 6: 638–647.
12. Wada, Y.; Tajiri, M.; Yoshida, S. 2004. Hydrophilic affinity isolation and MALDI
multiple-stage
Â� tandem mass spectrometry of glycopeptides for glycoproteomics. Anal.
Chem. 76: 6560–6565.
13. Wuhrer, M.; Koeleman, C. A.; Hokke, C. H.; Deelder, A. M. 2005. Protein glycosylation
analyzed by normal-phase nano-liquid chromatography–mass spectrometry of glyco-
peptides. Anal. Chem. 77: 886–894.
14. Yu, Y. Q.; Gilar, M.; Kaska, J.; Gebler, J. C. 2005. A rapid sample preparation method
for mass spectrometric characterization of N-linked glycans. Rapid Commun. Mass
Spectrom. 19: 2331–2336.
15. Daniels, M. A.; Hogquist, K. A.; Jameson, S. C. 2002. Sweet ‘n’ sour: The impact of
differential glycosylation on T cell responses. Nat. Immunol. 3(10): 903–910.
16. Dwek, M. V.; Ross, H. A.; Leathem, A. J. 2001. Proteome and glycosylation mapping
identifies post-translational modifications associated with aggressive breast cancer.
Proteomics 1: 756–762.
17. Freeze, H. H. 2001. Update and perspectives on congenital disorders of glycosylation.
Glycobiology 11: 129R–143R.
18. Cubbon, S.; Bradbury, T.; Wilson, J.; Thomas-Oates, J. 2007. Hydrophilic interaction
chromatography for mass spectrometric metabonomic studies of urine. Anal. Chem.
79(23): 8911–8918.
19. Godejohann, M. 2007. Hydrophilic interaction chromatography coupled to nuclear
magnetic resonance spectroscopy and mass spectroscopy—A new approach for the
separation and identification of extremely polar analytes in bodyfluids. J. Chromatogr.
A 1156(1–2): 87–93.
20. Idborg, H.; Zamani, L.; Edlund, P. O.; Schuppe-Koistinen, I.; Jacobsson, S. P. 2005.
Metabolic fingerprinting of rat urine by LC/MS Part 2. Data pretreatment methods for
handling of complex data. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 828:
14–20.
21. Idborg, H.; Zamani, L.; Edlund, P. O.; Schuppe-Koistinen, I.; Jacobsson, S. P. 2005.
Metabolic fingerprinting of rat urine by LC/MS Part 1. Analysis by hydrophilic interac-
tion liquid chromatography–electrospray ionization mass spectrometry. J. Chromatogr.
B Analyt. Technol. Biomed. Life Sci. 828: 9–13.
22. Tuytten, R.; Lemiere, F.; Van Dongen, W. et al. 2008. Development of an on-line
SPE-LC-ESI-MS method for urinary nucleosides: Hyphenation of aprotic boronic
acid chromatography with hydrophilic interaction LC-ESI-MS. Anal. Chem. 80(4):
1263–1271.
23. Kind, T.; Tolstikov, V.; Fiehn, O.; Weiss, R. H. 2007. A comprehensive urinary metabo-
lomic approach for identifying kidney cancer. Anal. Biochem. 363(2): 185–195.
24. Merali, S.; Vargas, D.; Franklin, M.; Clarkson, A. B., Jr. 2000. S-Adenosylmethionine
and Pneumocystis carinii. J. Biol. Chem. 275: 14958–14963.
25. Melnyk, S.; Pogribna, M.; Pogribny, I. P.; Yi, P.; James, S. J. 2000. Measurement of
plasma and intracellular S-adenosylmethionine and S-adenosylhomocysteine utilizing
coulometric electrochemical detection: Alterations with plasma homocysteine and pyri-
doxal 5′-phosphate concentrations. Clin. Chem. 46: 265–272.
26. Struys, E. A.; Jansen, E. E.; de Meer, K.; Jakobs, C. 2000. Determination of
S-adenosylmethionine and S-adenosylhomocysteine in plasma and cerebrospinal fluid
by stable-isotope dilution tandem mass spectrometry. Clin. Chem. 46: 1650–1656.

27. Gellekink, H.; van Oppenraaij-Emmerzaal, D.; van Rooij, A.; Struys, E. A.; den
Heijer, M.; Blom, H. J. 2005. Stable-isotope dilution liquid chromatography–elec-
trospray injection tandem mass spectrometry method for fast, selective measurement
of S-adenosylmethionine and S-adenosylhomocysteine in plasma. Clin. Chem. 51:
1487–1492.
28. Delabar, U.; Kloor, D.; Luippold, G.; Muhlbauer, B. 1999. Simultaneous determina-
tion of adenosine, S-adenosylhomocysteine and S-adenosylmethionine in biological
samples using solid-phase extraction and high-performance liquid chromatography.
J. Chromatogr. B Biomed. Sci. Appl. 724(2): 231–238.
29. Wang, P.; Huang, L.; Davis, J. L. et al. 2008. A hydrophilic-interaction chromatogra-
phy tandem mass spectrometry method for quantitation of serum S-adenosylmethionine
in patients infected with human immunodeficiency virus. Clin. Chim. Acta 396(1–2):
86–88.
30. Lenders, J. W.; Pacak, K.; Walther, M. M. et al. 2002. Biochemical diagnosis of pheo-
chromocytoma: Which test is best? JAMA 287: 1427–1434.
31. Pallant, A.; Mathian, B.; Prost, L.; Theodore, C.; Patricot, M. C. 2000. Determination of
plasma methoxyamines. Clin. Chem. Lab. Med. 38(6): 513–517.
32. Lenders, J. W.; Eisenhofer, G.; Armando, I.; Keiser, H. R.; Goldstein, D. S.; Kopin, I. J.
1993. Determination of metanephrines in plasma by liquid chromatography with elec-
trochemical detection. Clin. Chem. 39: 97–103.
33. Lagerstedt, S. A.; O’Kane, D. J.; Singh, R. J. 2004. Measurement of plasma free meta-
nephrine and normetanephrine by liquid chromatography–tandem mass spectrometry
for diagnosis of pheochromocytoma. Clin. Chem. 50: 603–611.
34. Manz, B.; Kuper, M.; Booltink, E.; Fischer-Brugge, U. 2004. Development of enantiose-
lective immunoassays for free plasma metanephrines. Ann. N Y Acad. Sci. 1018: 582–587.
35. Lenz, T.; Zorner, J.; Kirchmaier, C. et al. 2006. Multicenter study on the diagnostic
value of a new RIA for the detection of free plasma metanephrines in the work-up for
pheochromocytoma. Ann. N Y Acad. Sci. 1073: 358–373.
36. de Jong, W. H.; Graham, K. S.; van der Molen, J. C. et al. 2007. Plasma free metaneph-
rine measurement using automated online solid-phase extraction HPLC–tandem mass
spectrometry. Clin. Chem. 53: 1684–1693.
37. Vetterlein, K.; Buche, K.; Hildebrand, M.; Scriba, G. K.; Lehmann, J. 2006. Capillary
electrophoresis for the characterization of the complex dendrimeric contrast agent
Gadomer. Electrophoresis 27(12): 2400–2412.
38. Vetterlein, K.; Bergmann, U.; Buche, K. et al. 2007. Comprehensive profiling of the
complex dendrimeric contrast agent Gadomer using a combined approach of CE, MS,
and CE-MS. Electrophoresis 28(17): 3088–3099.
39. Campa, C.; Rossi, M.; Flamigni, A.; Baiutti, E.; Coslovi, A.; Calabi, L. 2005. Analysis
of gadobenate dimeglumine by capillary zone electrophoresis coupled with electro-
spray–mass spectrometry. Electrophoresis 26(7–8): 1533–1540.
40. Loreti, V.; Bettmer, J. 2004. Determination of the MRI contrast agent Gd-DTPA by
SEC-ICP-MS. Anal. Bioanal. Chem. 379(7–8): 1050–1054.
41. Vora, M. M.; Wukovnig, S.; Finn, R. D.; Emran, A. M.; Boothe, T. E.; Kothari, P. J. 1986.
Reversed-phase high-performance liquid chromatographic determinationof gadolinium-
diethylenetriaminepentaacetic acid complex. J. Chromatogr. 369: 187–192.
42. Hagan, J. J.; Taylor, S. C.; Tweedle, M. F. 1988. Fluorescence detection of gadolinium
chelates separated by reversed-phase high-performance liquid chromatography. Anal.
Chem. 60(6): 514–516.
43. Hvattum, E.; Normann, P. T.; Jamieson, G. C.; Lai, J. J.; Skotland, T. 1995. Detection
and quantitation of gadolinium chelates in human serum and urine by high-performance
liquid chromatography and post-column derivatization of gadolinium with Arsenazo III.
J. Pharm. Biomed. Anal. 13(7): 927–932.

44. Behra-Miellet, J.; Briand, G.; Kouach, M.; Gressier, B.; Cazin, M.; Cazin, J. C. 1998.
On-line HPLC-electrospray ionization mass spectrometry: A pharmacological tool for
identifying and studying the stability of Gd3+ complexes used as magnetic resonance
imaging contrast agents. Biomed. Chromatogr. 12(1): 21–26.
45. Arbughi, T.; Bertani, F.; Celeste, R.; Grotti, A.; Sillari, S.; Tirone, P. 1998. High-
performance liquid chromatographic determination of the magnetic resonance imaging
contrast agent gadobenate ion in plasma, urine, faeces, bile and tissues. J. Chromatogr.
B Biomed. Sci. Appl. 713(2): 415–426.
46. Kunnemeyer, J.; Terborg, L.; Nowak, S. et al. 2008. Speciation analysis of gadolinium-
based MRI contrast agents in blood plasma by hydrophilic interaction chromatography/
electrospray mass spectrometry. Anal. Chem. 80(21): 8163–8170.
47. Windelberg, A.; Arseth, O.; Kvalheim, G.; Ueland, P. M. 2005. Automated assay for the
determination of methylmalonic acid, total homocysteine, and related amino acids in
human serum or plasma by means of methylchloroformate derivatization and gas chro-
matography–mass spectrometry. Clin. Chem. 51: 2103–2109.
48. Schneede, J.; Ueland, P. M. 1993. Automated assay of methylmalonic acid in serum
and urine by derivatization with 1-pyrenyldiazomethane, liquid chromatography, and
fluorescence detection. Clin. Chem. 39: 392–399.
49. Schneede, J.; Ueland, P. M. 1995. Application of capillary electrophoresis with laser-
induced fluorescence detection for determination of methylmalonic acid in human
serum. Anal. Chem. 67(5): 812–819.
50. Kushnir, M. M.; Komaromy-Hiller, G.; Shushan, B.; Urry, F. M.; Roberts, W. L. 2001.
Analysis of dicarboxylic acids by tandem mass spectrometry. High-throughput quantitative
measurement of methylmalonic acid in serum, plasma, and urine. Clin. Chem. 47: 1993–2002.
51. Schmedes, A.; Brandslund, I. 2006. Analysis of methylmalonic acid in plasma by liquid
chromatography–tandem mass spectrometry. Clin. Chem. 52: 754–757.
52. Magera, M. J.; Helgeson, J. K.; Matern, D.; Rinaldo, P. 2000. Methylmalonic acid mea-
sured in plasma and urine by stable-isotope dilution and electrospray tandem mass spec-
trometry. Clin. Chem. 46: 1804–1810.
53. Blom, H. J.; van Rooij, A.; Hogeveen, M. 2007. A simple high-throughput method for
the determination of plasma methylmalonic acid by liquid chromatography–tandem
mass spectrometry. Clin. Chem. Lab. Med. 45(5): 645–650.
54. Lakso, H. A.; Appelblad, P.; Schneede, J. 2008. Quantification of methylmalonic acid in
human plasma with hydrophilic interaction liquid chromatography separation and mass
spectrometric detection. Clin. Chem. 54: 2028–2035.
55. Ekblom, M.; Hammarlund-Udenaes, M.; Paalzow, L. 1993. Modeling of tolerance
development and rebound effect during different intravenous administrations of mor-
phine to rats. J. Pharmacol. Exp. Ther. 266: 244–252.
56. Toyo’oka, T.; Yano, M.; Kato, M.; Nakahara, Y. 2001. Simultaneous determination of
morphine and its glucuronides in rat hair and rat plasma by reversed-phase liquid chro-
matography with electrospray ionization mass spectrometry. Analyst 126(8): 1339–1345.
57. Projean, D.; Minh Tu, T.; Ducharme, J. 2003. Rapid and simple method to determine
morphine and its metabolites in rat plasma by liquid chromatography–mass spectrom-
etry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 787(2): 243–253.
58. Whittington, D.; Kharasch, E. D. 2003. Determination of morphine and morphine glucuÂ�
ronides in human plasma by 96-well plate solid-phase extraction and liquid chroma-
tography–electrospray ionization mass spectrometry. J. Chromatogr. B Analyt. Technol.
Biomed. Life Sci. 796(1): 95–103.
59. Mabuchi, M.; Takatsuka, S.; Matsuoka, M.; Tagawa, K. 2004. Determination of mor-
phine, morphine-3-glucuronide and morphine-6-glucuronide in monkey and dog plasma
by high-performance liquid chromatography–electrospray ionization tandem mass spec-
trometry. J. Pharm. Biomed. Anal. 35(3): 563–573.

60. Naidong, W.; Lee, J. W.; Jiang, X.; Wehling, M.; Hulse, J. D.; Lin, P. P. 1999.
Simultaneous assay of morphine, morphine-3-glucuronide and morphine-6-glucuronide
in human plasma using normal-phase liquid chromatography–tandem mass spectrome-
try with a silica column and an aqueous organic mobile phase. J. Chromatogr. B Biomed.
Sci. Appl. 735(2): 255–269.
61. Shou, W. Z.; Chen, Y. L.; Eerkes, A. et al. 2002. Ultrafast liquid chromatography/tandem
mass spectrometry bioanalysis of polar analytes using packed silica columns. Rapid
Commun. Mass Spectrom. 16(17): 1613–1621.
62. Bengtsson, J.; Jansson, B.; Hammarlund-Udenaes, M. 2005. On-line desalting and
determination of morphine, morphine-3-glucuronide and morphine-6-glucuronide in
microdialysis and plasma samples using column switching and liquid chromatogra-
phy/tandem mass spectrometry. Rapid Commun. Mass Spectrom. 19(15): 2116–2122.
63. Fekkes, D.; van Dalen, A.; Edelman, M.; Voskuilen, A. 1995. Validation of the deter-
mination of amino acids in plasma by high-performance liquid chromatography using
automated pre-column derivatization with o-phthaldialdehyde. J. Chromatogr. B
Biomed. Appl. 669(2): 177–186.
64. Le Boucher, J.; Charret, C.; Coudray-Lucas, C.; Giboudeau, J.; Cynober, L. 1997.
Amino acid determination in biological fluids by automated ion-exchange chromatogra-
phy: Performance of Hitachi L-8500A. Clin. Chem. 43: 1421–1428.
65. Patchett, M. L.; Monk, C. R.; Daniel, R. M.; Morgan, H. W. 1988. Determination of
agmatine, arginine, citrulline and ornithine by reversed-phase liquid chromatography
using automated pre-column derivatization with o-phthalaldehyde. J. Chromatogr.
425(2): 269–276.
66. Zhang, W. Z.; Kaye, D. M. 2004. Simultaneous determination of arginine and seven
metabolites in plasma by reversed-phase liquid chromatography with a time-controlled
ortho-phthaldialdehyde precolumn derivatization. Anal. Biochem. 326(1): 87–92.
67. Piraud, M.; Vianey-Saban, C.; Petritis, K.; Elfakir, C.; Steghens, J. P.; Bouchu, D. 2005.
Ion-pairing reversed-phase liquid chromatography/electrospray ionization mass spec-
trometric analysis of 76 underivatized amino acids of biological interest: A new tool for
the diagnosis of inherited disorders of amino acid metabolism. Rapid Commun. Mass
Spectrom. 19(12): 1587–1602.
68. Martens-Lobenhoffer, J.; Postel, S.; Troger, U.; Bode-Boger, S. M. 2007. Deter-
mination of ornithine in human plasma by hydrophilic interaction chromatographyÂ�–
tandem mass spectrometry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci.
855(2): 271–275.
69. Wang, Y.; Wang, T.; Shi, X. et al. 2008. Analysis of acetylcholine, choline and butyrob-
etaine in human liver tissues by hydrophilic interaction liquid chromatography–tandem
mass spectrometry. J. Pharm. Biomed. Anal. 47(4–5): 870–875.
70. Holm, P. I.; Ueland, P. M.; Kvalheim, G.; Lien, E. A. 2003. Determination of choline,
betaine, and dimethylglycine in plasma by a high-throughput method based on normal-
phase chromatography–tandem mass spectrometry. Clin. Chem. 49: 286–294.
71. Wang, F. L.; Haubrich, D. R. 1975. A simple, sensitive, and specific assay for free cho-
line in plasma. Anal. Biochem. 63(1): 195–201.
72. Eckernas, S. A.; Aquilonius, S. M. 1977. Free choline in human plasma analysed by
simple radio-enzymatic procedure: Age distribution and effect of a meal. Scand. J. Clin.
Lab. Invest. 37(2): 183–187.
73. McMahon, K. E.; Farrell, P. M. 1985. Measurement of free choline concentrations in
maternal and neonatal blood by micropyrolysis gas chromatography. Clin. Chim. Acta
149(1): 1–12.
74. Pomfret, E. A.; daCosta, K. A.; Schurman, L. L.; Zeisel, S. H. 1989. Measurement of
choline and choline metabolite concentrations using high-pressure liquid chromatogra-
phy and gas chromatography–mass spectrometry. Anal. Biochem. 180(1): 85–90.

75. Tsai, T. H. 2000. Separation methods used in the determination of choline and acetyl-
choline. J. Chromatogr. B Biomed. Sci. Appl. 747(1–2): 111–122.
76. Guerrieri, A.; Palmisano, F. 2001. An acetylcholinesterase/choline oxidase-based
amperometric biosensors as a liquid chromatography detector for acetylcholine and
choline determination in brain tissue homogenates. Anal. Chem. 73(13): 2875–2882.
77. Huang, T.; Yang, L.; Gitzen, J.; Kissinger, P. T.; Vreeke, M.; Heller, A. 1995. Detection
of basal acetylcholine in rat brain microdialysate. J. Chromatogr. B Biomed. Appl.
670(2): 323–327.
78. Niwa, T. 1996. Organic acids and the uremic syndrome: Protein metabolite hypothesis
in the progression of chronic renal failure. Semin. Nephrol. 16(3): 167–182.
79. Tsai, T. R.; Cham, T. M.; Chen, K. C.; Chen, C. F.; Tsai, T. H. 1996. Determination of ace-
tylcholine by on-line microdialysis coupled with pre- and post-microbore column enzyme
reactors with electrochemical detection. J. Chromatogr. B Biomed. Appl. 678(2): 151–155.
80. Yang, S.; Khaledi, M. G. 1995. Chemical selectivity in micellar electrokinetic chro-
matography: Characterization of solute-micelle interactions for classification of surfac-
tants. Anal. Chem. 67(3): 499–510.
81. You, T.; Niwa, O.; Tomita, M.; Hirono, S. 2003. Characterization of platinum nanopar-
ticle-embedded carbon film electrode and its detection of hydrogen peroxide. Anal.
Chem. 75(9): 2080–2085.
82. Fossati, T.; Colombo, M.; Castiglioni, C.; Abbiati, G. 1994. Determination of plasma
choline by high-performance liquid chromatography with a postcolumn enzyme reactor
and electrochemical detection. J. Chromatogr. B Biomed. Appl. 656(1): 59–64.
83. Hows, M. E.; Organ, A. J.; Murray, S. et al. 2002. High-performance liquid chromatog-
raphy/tandem mass spectrometry assay for the rapid high sensitivity measurement of
basal acetylcholine from microdialysates. J. Neurosci. Methods 121(1): 33–39.
84. Reubsaet, J. L.; Ahlsen, E.; Haneborg, K. G.; Ringvold, A. 2003. Sample preparation
and determination of acetylcholine in corneal epithelium cells using liquid chromatog-
raphy–tandem mass spectrometry. J. Chromatogr. Sci. 41(3): 151–156.
85. Zhang, M. Y.; Hughes, Z. A.; Kerns, E. H.; Lin, Q.; Beyer, C. E. 2007. Development
of a liquid chromatography/tandem mass spectrometry method for the quantitation of
acetylcholine and related neurotransmitters in brain microdialysis samples. J. Pharm.
Biomed. Anal. 44(2): 586–593.
86. Koc, H.; Mar, M. H.; Ranasinghe, A.; Swenberg, J. A.; Zeisel, S. H. 2002. Quantitation
of choline and its metabolites in tissues and foods by liquid chromatography/electro-
spray ionization-isotope dilution mass spectrometry. Anal. Chem. 74(18): 4734–4740.
87. Laryea, M. D.; Steinhagen, F.; Pawliczek, S.; Wendel, U. 1998. Simple method for the
routine determination of betaine and N,N-dimethylglycine in blood and urine. Clin.
Chem. 44: 1937–1941.
88. Allen, R. H.; Stabler, S. P.; Lindenbaum, J. 1993. Serum betaine, N,N-dimethylglycine
and N-methylglycine levels in patients with cobalamin and folate deficiency and related
inborn errors of metabolism. Metabolism 42(11): 1448–1460.
89. Hecht, S. S.; Carmella, S. G.; Ye, M. et al. 2002. Quantitation of metabolites of
4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone after cessation of smokeless tobacco
use. Cancer Res. 62: 129–134.
90. Carmella, S. G.; Akerkar, S.; Hecht, S. S. 1993. Metabolites of the tobacco-specific
nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone in smokers’ urine. Cancer
Res. 53: 721–724.
91. Hecht, S. S.; Carmella, S. G.; Chen, M. et al. 1999. Quantitation of urinary metabolites
of a tobacco-specific lung carcinogen after smoking cessation. Cancer Res. 59: 590–596.
92. Byrd, G. D.; Ogden, M. W. 2003. Liquid chromatographic/tandem mass spectrometric
method for the determination of the tobacco-specific nitrosamine metabolite NNAL in
smokers’ urine. J. Mass Spectrom. 38(1): 98–107.

93. Pan, J.; Song, Q.; Shi, H. et al. 2004. Development, validation and transfer of a hydro-
philic interaction liquid chromatography/tandem mass spectrometric method for the
analysis of the tobacco-specific nitrosamine metabolite NNAL in human plasma at
low picogram per milliliter concentrations. Rapid Commun. Mass Spectrom. 18(21):
2549–2557.
94. Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; Tornqvist, M. 2000. Acrylamide:
A cooking carcinogen? Chem. Res. Toxicol. 13(6): 517–522.
95. Bjellaas, T.; Stolen, L. H.; Haugen, M. et al. 2007. Urinary acrylamide metabolites as
biomarkers for short-term dietary exposure to acrylamide. Food Chem. Toxicol. 45(6):
1020–1026.
96. Boettcher, M. I.; Angerer, J. 2005. Determination of the major mercapturic acids of
acrylamide and glycidamide in human urine by LC-ESI-MS/MS. J. Chromatogr. B
Analyt. Technol. Biomed. Life Sci. 824(1–2): 283–294.
97. Urban, M.; Kavvadias, D.; Riedel, K.; Scherer, G.; Tricker, A. R. 2006. Urinary mer-
capturic acids and a hemoglobin adduct for the dosimetry of acrylamide exposure in
smokers and nonsmokers. Inhal. Toxicol. 18(10): 831–839.
98. Fuhr, U.; Boettcher, M. I.; Kinzig-Schippers, M. et al. 2006. Toxicokinetics of acryl-
amide in humans after ingestion of a defined dose in a test meal to improve risk assess-
ment for acrylamide carcinogenicity. Cancer Epidemiol. Biomarkers Prev. 15: 266–271.
99. Kopp, E. K.; Sieber, M.; Kellert, M.; Dekant, W. 2008. Rapid and sensitive HILIC-
ESI-MS/MS quantitation of polar metabolites of acrylamide in human urine using col-
umn switching with an online trap column. J. Agric. Food Chem. 56(21): 9828–9834.
100. Buse, J. B.; Freeman, J. L.; Edelman, S. V.; Jovanovic, L.; McGill, J. B. 2003. Serum
1,5-anhydroglucitol (GlycoMark): A short-term glycemic marker. Diabetes Technol.
Ther. 5(3): 355–363.
101. Yamanouchi, T.; Akanuma, H.; Asano, T.; Konishi, C.; Akaoka, I.; Akanuma, Y. 1987.
Reduction and recovery of plasma 1,5-anhydro-D-glucitol level in diabetes mellitus.
Diabetes 36: 709–715.
102. Yamanouchi, T.; Akanuma, H.; Nakamura, T.; Akaoka, I.; Akanuma, Y. 1988. Reduction
of plasma 1,5-anhydroglucitol (1-deoxyglucose) concentration in diabetic patients.
Diabetologia 31(1): 41–45.
103. Yoshioka, S.; Saitoh, S.; Negishi, C. et al. 1983. Variations of 1-deoxyglucose(1,5-anhy-
droglucitol) content in plasma from patients with insulin-dependent diabetes mellitus.
Clin. Chem. 29: 1396–1398.
104. Yamanouchi, T.; Minoda, S.; Yabuuchi, M. et al. 1989. Plasma 1,5-anhydro-D-glucitol
as new clinical marker of glycemic control in NIDDM patients. Diabetes 38: 723–729.
105. Yamanouchi, T.; Tachibana, Y.; Akanuma, H. et al. 1992. Origin and disposal of 1,5-anhy-
droglucitol, a major polyol in the human body. Am. J. Physiol. 263: E268–E273.
106. McGill, J. B.; Cole, T. G.; Nowatzke, W. et al. 2004. Circulating 1,5-anhydroglucitol
levels in adult patients with diabetes reflect longitudinal changes of glycemia: A U.S.
trial of the glycomark assay. Diabetes Care 27: 1859–1865.
107. Fukumura, Y.; Tajima, S.; Oshitani, S. et al. 1994. Fully enzymatic method for determin-
ing 1,5-anhydro-D-glucitol in serum. Clin. Chem. 40: 2013–2016.
108. Yoshioka, S.; Saitoh, S.; Fujisawa, T.; Fujimori, A.; Takatani, O.; Funabashi, M. 1982.
Identification and metabolic implication of 1-deoxyglucose (1,5- anhydroglucitol) in
human plasma. Clin. Chem. 28: 1283–1286.
109. Yamanouchi, T.; Akaoka, I.; Akanuma, Y.; Akanuma, H.; Miyashita, E. 1990. Mechanism
for acute reduction of 1,5-anhydroglucitol in rats treated with diabetogenic agents. Am.
J. Physiol. 258: E423–E427.
110. Akanuma, Y.; Morita, M.; Fukuzawa, N.; Yamanouchi, T.; Akanuma, H. 1988. Urinary
excretion of 1,5-anhydro-D-glucitol accompanying glucose excretion in diabetic
patients. Diabetologia 31(11): 831–835.

111. Tanabe, T.; Tajima, S.; Suzuki, T. et al. 1997. Quantification of 1,5-anhydro-D-glucitol
in urine by automated borate complex anion-exchange chromatography with an immo-
bilized enzyme reactor. J. Chromatogr. B Biomed. Sci. Appl. 692(1): 23–30.
112. Tanaka, S.; Nakamori, K.; Akanuma, H.; Yabuuchi, M. 1992. High performance liquid
chromatographic determination of 1,5-anhydroglucitol in human plasma for diagnosis
of diabetes mellitus. Biomed. Chromatogr. 6(2): 63–66.
113. Niwa, T.; Tohyama, K.; Kato, Y. 1993. Analysis of polyols in uremic serum by liquid
chromatography combined with atmospheric pressure chemical ionization mass spec-
trometry. J. Chromatogr. 613(1): 9–14.
114. Onorato, J. M.; Langish, R. A.; Shipkova, P. A. et al. 2008. A novel method for the deter-
mination of 1,5-anhydroglucitol, a glycemic marker, in human urine utilizing hydro-
philic interaction liquid chromatography/MS(3). J. Chromatogr. B Analyt. Technol.
Biomed. Life Sci. 873(2): 144–150.
115. Zoccali, C.; Bode-Boger, S.; Mallamaci, F. et al. 2001. Plasma concentration of asym-
metrical dimethylarginine and mortality in patients with end-stage renal disease: A pro-
spective study. Lancet 358(9299): 2113–2117.
116. Abbasi, F.; Asagmi, T.; Cooke, J. P. et al. 2001. Plasma concentrations of asymmetric
dimethylarginine are increased in patients with type 2 diabetes mellitus. Am. J. Cardiol.
88(10): 1201–1203.
117. Surdacki, A.; Nowicki, M.; Sandmann, J. et al. 1999. Reduced urinary excretion of nitric
oxide metabolites and increased plasma levels of asymmetric dimethylarginine in men
with essential hypertension. J. Cardiovasc. Pharmacol. 33(4): 652–658.
118. Boger, R. H.; Bode-Boger, S. M.; Szuba, A. et al. 1998. Asymmetric dimethylarginine
(ADMA): A novel risk factor for endothelial dysfunction: Its role in hypercholesterol-
emia. Circulation 98: 1842–1847.
119. Gary, J. D.; Clarke, S. 1998. RNA and protein interactions modulated by protein argi-
nine methylation. Prog. Nucleic Acid Res. Mol. Biol. 61: 65–131.
120. Teerlink, T. 2007. HPLC analysis of ADMA and other methylated L-arginine analogs in
biological fluids. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 851(1–2): 21–29.
121. Bode-Boger, S. M.; Boger, R. H.; Kienke, S.; Junker, W.; Frolich, J. C. 1996. Elevated
L-arginine/dimethylarginine ratio contributes to enhanced systemic NO production by
dietary L-arginine in hypercholesterolemic rabbits. Biochem. Biophys. Res. Commun.
219(2): 598–603.
122. Cooke, J. P. 2004. Asymmetrical dimethylarginine: The Über marker? Circulation 109:
1813–1818.
123. Fleck, C.; Janz, A.; Schweitzer, F.; Karge, E.; Schwertfeger, M.; Stein, G. 2001. Serum
concentrations of asymmetric (ADMA) and symmetric (SDMA) dimethylarginine in
renal failure patients. Kidney Int. Suppl. 78: S14–S18.
124. Lluch, P.; Mauricio, M. D.; Vila, J. M. et al. 2006. Accumulation of symmetric dimethy-
larginine in hepatorenal syndrome. Exp. Biol. Med. (Maywood) 231(1): 70–75.
125. Valtonen, P.; Karppi, J.; Nyyssonen, K.; Valkonen, V. P.; Halonen, T.; Punnonen, K.
2005. Comparison of HPLC method and commercial ELISA assay for asymmetric
dimethylarginine (ADMA) determination in human serum. J. Chromatogr. B Analyt.
Technol. Biomed. Life Sci. 828(1–2): 97–102.
126. Meyer, J.; Richter, N.; Hecker, M. 1997. High-performance liquid chromatographic
determination of nitric oxide synthase-related arginine derivatives in vitro and in vivo.
Anal. Biochem. 247(1): 11–16.
127. Pettersson, A.; Uggla, L.; Backman, V. 1997. Determination of dimethylated arginines in
human plasma by high-performance liquid chromatography. J. Chromatogr. B Biomed.
Sci. Appl. 692(2): 257–262.

128. Teerlink, T.; Nijveldt, R. J.; de Jong, S.; van Leeuwen, P. A. 2002. Determination of
arginine, asymmetric dimethylarginine, and symmetric dimethylarginine in human
plasma and other biological samples by high-performance liquid chromatography. Anal.
Biochem. 303(2): 131–137.
129. Martens-Lobenhoffer, J.; Krug, O.; Bode-Boger, S. M. 2004. Determination of arginine
and asymmetric dimethylarginine (ADMA) in human plasma by liquid chromatogra-
phy/mass spectrometry with the isotope dilution technique. J. Mass Spectrom. 39(11):
1287–1294.
130. Schwedhelm, E.; Tan-Andresen, J.; Maas, R.; Riederer, U.; Schulze, F.; Boger, R. H.
2005. Liquid chromatography–tandem mass spectrometry method for the analysis of
asymmetric dimethylarginine in human plasma. Clin. Chem. 51: 1268–1271.
131. Kirchherr, H.; Kuhn-Velten, W. N. 2005. HPLC–tandem mass spectrometric method for
rapid quantification of dimethylarginines in human plasma. Clin. Chem. 51: 249–252.
132. Huang, L. F.; Guo, F. Q.; Liang, Y. Z.; Li, B. Y.; Cheng, B. M. 2004. Simultaneous deter-
mination of L-arginine and its mono- and dimethylated metabolites in human plasma
by high-performance liquid chromatography–mass spectrometry. Anal. Bioanal. Chem.
380(4): 643–649.
133. Martens-Lobenhoffer, J.; Bode-Boger, S. M. 2006. Fast and efficient determination of
arginine, symmetric dimethylarginine, and asymmetric dimethylarginine in biological
fluids by hydrophilic-interaction liquid chromatography–electrospray tandem mass
spectrometry. Clin. Chem. 52: 488–493.
134. D’Apolito, O.; Paglia, G.; Tricarico, F. et al. 2008. Development and validation of a fast
quantitative method for plasma dimethylarginines analysis using liquid Â�chromatography–
tandem mass spectrometry. Clin. Biochem. 41(16–17): 1391–1395.
135. Paglia, G.; D’Apolito, O.; Tricarico, F.; Garofalo, D.; Corso, G. 2008. Evaluation of
mobile phase, ion pairing, and temperature influence on an HILIC-MS/MS method for
L-arginine and its dimethylated derivatives detection. J. Sep. Sci. 31(13): 2424–2429.
136. Van Bocxlaer, J. F.; Clauwaert, K. M.; Lambert, W. E.; Deforce, D. L.; Van den Eeckhout,
E. G.; De Leenheer, A. P. 2000. Liquid chromatography–mass spectrometry in forensic
toxicology. Mass Spectrom. Rev. 19(4): 165–214.
137. Tatsuno, M.; Nishikawa, M.; Katagi, M.; Tsuchihashi, H. 1996. Simultaneous deter-
mination of illicit drugs in human urine by liquid chromatography–mass spectrometry.
J. Anal. Toxicol. 20(5): 281–286.
138. Marquet, P. 2002. Progress of liquid chromatography–mass spectrometry in clinical and
forensic toxicology. Ther. Drug Monit. 24(2): 255–276.
139. Weinmann, W.; Svoboda, M. 1998. Fast screening for drugs of abuse by solid-phase
extraction combined with flow-injection ionspray–tandem mass spectrometry. J. Anal.
Toxicol. 22(4): 319–328.
140. Giroud, C.; Michaud, K.; Sporkert, F. et al. 2004. A fatal overdose of cocaine associated
with coingestion of marijuana, buprenorphine, and fluoxetine. Body fluid and tissue
distribution of cocaine and its metabolites determined by hydrophilic interaction chro-
matography–mass spectrometry (HILIC-MS). J. Anal. Toxicol. 28(6): 464–474.
141. Ding, W. H.; Chiang, C. C. 2003. Derivatization procedures for the detection of estro-
genic chemicals by gas chromatography/mass spectrometry. Rapid Commun. Mass
Spectrom. 17(1): 56–63.
142. Gibson, R.; Tyler, C. R.; Hill, E. M. 2005. Analytical methodology for the identification
of estrogenic contaminants in fish bile. J. Chromatogr. A 1066(1–2): 33–40.
143. Zuo, Y.; Zhang, K.; Lin, Y. 2007. Microwave-accelerated derivatization for the simulta-
neous gas chromatographic-mass spectrometric analysis of natural and synthetic estro-
genic steroids. J. Chromatogr. A 1148(2): 211–218.

144. Quintana, J. B.; Carpinteiro, J.; Rodriguez, I.; Lorenzo, R. A.; Carro, A. M.; Cela, R.
2004. Determination of natural and synthetic estrogens in water by gas chromatography
with mass spectrometric detection. J. Chromatogr. A 1024(1–2): 177–185.
145. Rogan, E. G.; Badawi, A. F.; Devanesan, P. D. et al. 2003. Relative imbalances in estro-
gen metabolism and conjugation in breast tissue of women with carcinoma: Potential
biomarkers of susceptibility to cancer. Carcinogenesis 24: 697–702.
146. Draisci, R.; Palleschi, L.; Ferretti, E.; Marchiafava, C.; Lucentini, L.; Cammarata, P.
1998. Quantification of 17 beta-estradiol residues in bovine serum by liquid chroma-
tography–tandem mass spectrometry with atmospheric pressure chemical ionization.
Analyst 123(12): 2605–2609.
147. Xu, X.; Ziegler, R. G.; Waterhouse, D. J.; Saavedra, J. E.; Keefer, L. K. 2002. Stable
isotope dilution high-performance liquid chromatography–electrospray ionization
mass spectrometry method for endogenous 2- and 4-hydroxyestrones in human urine.
J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 780(2): 315–330.
148. Xu, X.; Keefer, L. K.; Waterhouse, D. J.; Saavedra, J. E.; Veenstra, T. D.; Ziegler, R. G. 2004.
Measuring seven endogenous ketolic estrogens simultaneously in human urine by high-
performance liquid chromatography–mass spectrometry. Anal. Chem. 76(19): 5829–5836.
149. Nelson, R. E.; Grebe, S. K.; O’ Kane, D. J.; Singh, R. J. 2004. Liquid chromatography–
tandem mass spectrometry assay for simultaneous measurement of estradiol and estrone
in human plasma. Clin. Chem. 50: 373–384.
150. Xia, Y. Q.; Chang, S. W.; Patel, S.; Bakhtiar, R.; Karanam, B.; Evans, D. C. 2004. Trace
level quantification of deuterated 17 beta-estradiol and estrone in ovariectomized mouse
plasma and brain using liquid chromatography/tandem mass spectrometry following
dansylation reaction. Rapid Commun. Mass Spectrom. 18(14): 1621–1628.
151. Xu, X.; Veenstra, T. D.; Fox, S. D. et al. 2005. Measuring fifteen endogenous estrogens
simultaneously in human urine by high-performance liquid chromatography–mass spec-
trometry. Anal. Chem. 77(20): 6646–6654.
152. Xu, X.; Roman, J. M.; Issaq, H. J.; Keefer, L. K.; Veenstra, T. D.; Ziegler, R. G. 2007.
Quantitative measurement of endogenous estrogens and estrogen metabolites in human
serum by liquid chromatography–tandem mass spectrometry. Anal. Chem. 79(20):
7813–7821.
153. Volmer, D. A.; Hui, J. P. 1997. Rapid determination of corticosteroids in urine by com-
bined solid phase microextraction/liquid chromatography/mass spectrometry. Rapid
Commun. Mass Spectrom. 11(17): 1926–1933.
154. D’Ascenzo, G.; Di Corcia, A.; Gentili, A. et al. 2003. Fate of natural estrogen conjugates
in municipal sewage transport and treatment facilities. Sci. Total Environ. 302(1–3):
199–209.
155. Qin, F.; Zhao, Y. Y.; Sawyer, M. B.; Li, X. F. 2008. Hydrophilic interaction liquid chro-
matography–tandem mass spectrometry determination of estrogen conjugates in human
urine. Anal. Chem. 80(9): 3404–3411.
156. Shih, M. L.; Smith, J. R.; McMonagle, J. D.; Dolzine, T. W.; Gresham, V. C. 1991.
Detection of metabolites of toxic alkylmethylphosphonates in biological samples. Biol.
Mass Spectrom. 20(11): 717–723.
157. Nakajima, T.; Sasaki, K.; Ozawa, H. et al. 1998. Urinary metabolites of sarin in a patient
of the Matsumoto sarin incident. Arch. Toxicol. 72(9): 601–603.
158. Driskell, W. J.; Shih, M.; Needham, L. L.; Barr, D. B. 2002. Quantitation of organophos-
phorus nerve agent metabolites in human urine using isotope dilution gas chromatogra-
phy–tandem mass spectrometry. J. Anal. Toxicol. 26(1): 6–10.
159. Kataoka, M.; Seto, Y. 2003. Discriminative determination of alkyl methylphosphonates
and methylphosphonate in blood plasma and urine by gas chromatography–mass spec-
trometry after tert.-butyldimethylsilylation. J. Chromatogr. B Analyt. Technol. Biomed.
Life Sci. 795(1): 123–132.

160. Barr, J. R.; Driskell, W. J.; Aston, L. S.; Martinez, R. A. 2004. Quantitation of metabo-
lites of the nerve agents sarin, soman, cyclohexylsarin, VX, and Russian VX in human
urine using isotope-dilution gas chromatography–tandem mass spectrometry. J. Anal.
Toxicol. 28(5): 372–378.
161. Barr, J. R. 2004. Biological monitoring of human exposure to chemical warfare agents.
J. Anal. Toxicol. 28(5): 305.
162. Riches, J.; Morton, I.; Read, R. W.; Black, R. M. 2005. The trace analysis of alkyl
alkylphosphonic acids in urine using gas chromatography–ion trap negative ion tan-
dem mass spectrometry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 816(1–2):
251–258.
163. Joshi, K. A.; Prouza, M.; Kum, M. et al. 2006. V-type nerve agent detection using a
carbon nanotube-based amperometric enzyme electrode. Anal. Chem. 78(1): 331–336.
164. Mulchandani, P.; Chen, W.; Mulchandani, A. 2006. Microbial biosensor for direct deter-
mination of nitrophenyl-substituted organophosphate nerve agents using genetically
engineered Moraxella sp. Anal. Chim. Acta 568(1–2): 217–221.
165. Shu, Y. R.; Su, A. K.; Liu, J. T.; Lin, C. H. 2006. Screening of nerve agent degradation
products by MALDI-TOFMS. Anal. Chem. 78(13): 4697–4701.
166. Mawhinney, D. B.; Hamelin, E. I.; Fraser, R.; Silva, S. S.; Pavlopoulos, A. J.; Kobelski,
R. J. 2007. The determination of organophosphonate nerve agent metabolites in human
urine by hydrophilic interaction liquid chromatography tandem mass spectrometry.
J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 852(1–2): 235–243.
167. Ito, S.; Fujita, K. 1985. Microanalysis of eumelanin and pheomelanin in hair and mela-
nomas by chemical degradation and liquid chromatography. Anal. Biochem. 144(2):
527–536.
168. Wakamatsu, K.; Ito, S.; Rees, J. L. 2002. The usefulness of 4-amino-3-hydroxyphenyl-
alanine as a specific marker of pheomelanin. Pigment Cell Res. 15(3): 225–232.
169. Crowley, J. R.; Yarasheski, K.; Leeuwenburgh, C.; Turk, J.; Heinecke, J. W. 1998.
Isotope dilution mass spectrometric quantification of 3-nitrotyrosine in proteins and tis-
sues is facilitated by reduction to 3-aminotyrosine. Anal. Biochem. 259(1): 127–135.
170. Ito, S.; Jimbow, K. 1983. Quantitative analysis of eumelanin and pheomelanin in hair
and melanomas. J. Invest. Dermatol. 80(4): 268–272.
171. Kolb, A. M.; Lentjes, E. G.; Smit, N. P.; Schothorst, A.; Vermeer, B. J.; Pavel, S. 1997.
Determination of pheomelanin by measurement of aminohydroxyphenylalanine isomers
with high-performance liquid chromatography. Anal. Biochem. 252(2): 293–298.
172. Borges, C. R.; Roberts, J. C.; Wilkins, D. G.; Rollins, D. E. 2001. Relationship of mela-
nin degradation products to actual melanin content: Application to human hair. Anal.
Biochem. 290(1): 116–125.
173. Takasaki, A.; Nezirevic, D.; Arstrand, K.; Wakamatsu, K.; Ito, S.; Kagedal, B. 2003.
HPLC analysis of pheomelanin degradation products in human urine. Pigment Cell Res.
16(5): 480–486.
174. Nezirevic, D.; Arstrand, K.; Kagedal, B. 2007. Hydrophilic interaction liquid chro-
matographic analysis of aminohydroxyphenylalanines from melanin pigments.
J. Chromatogr. A 1163(1–2): 70–79.
175. Bodini, A.; Peroni, D. G.; Zardini, F. et al. 2006. Flunisolide decreases exhaled nitric
oxide and nitrotyrosine levels in asthmatic children. Mediators Inflamm. 2006(4):
31919.
176. Larstad, M.; Soderling, A. S.; Caidahl, K.; Olin, A. C. 2005. Selective quantification of
free 3-nitrotyrosine in exhaled breath condensate in asthma using gas chromatography/
tandem mass spectrometry. Nitric Oxide 13(2): 134–144.
177. Celio, S.; Troxler, H.; Durka, S. S. et al. 2006. Free 3-nitrotyrosine in exhaled breath
condensates of children fails as a marker for oxidative stress in stable cystic fibrosis and
asthma. Nitric Oxide 15(3): 226–232.

178. Goen, T.; Muller-Lux, A.; Dewes, P.; Musiol, A.; Kraus, T. 2005. Sensitive and accu-
rate analyses of free 3-nitrotyrosine in exhaled breath condensate by LC-MS/MS.
J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 826(1–2): 261–266.
179. Baraldi, E.; Giordano, G.; Pasquale, M. F. et al. 2006. 3-Nitrotyrosine, a marker of nitro-
sative stress, is increased in breath condensate of allergic asthmatic children. Allergy
61(1): 90–96.
180. Conventz, A.; Musiol, A.; Brodowsky, C. et al. 2007. Simultaneous determination of
3-nitrotyrosine, tyrosine, hydroxyproline and proline in exhaled breath condensate by
hydrophilic interaction liquid chromatography/electrospray ionization tandem mass
spectrometry. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 860(1): 78–85.

10 Dimethylarginine
Analysis by HILIC–MS/ MS
and Its Applications in
Clinical Chemistry and
In Vivo Animal Models
O. D’Apolito, G. Paglia, M. Zotti,
F. Tricarico, L. Trabace, and G. Corso
Contents
10.1 Introduction................................................................................................... 203
10.2 Arginine and Its Dimethyl Derivatives..........................................................204
10.3 MS Setup.......................................................................................................205
10.4 Chromatographic Evaluations.......................................................................207
10.4.1 Implementation of Two HILIC Methods...........................................207
10.4.2 Effect of ACN Percentage on HILIC and Comparison between
Organic-Rich and Water-Rich Mobile Phase on Silica Column
Chromatography................................................................................207
10.4.3 Effect of TFA Percentage on the HILIC Method.............................. 211
10.4.4 Effect of Column Temperature on the HILIC Method...................... 212
10.5 Plasma Method.............................................................................................. 214
10.6 Collection of CSF Samples from Rat Brain by Microdialysis...................... 219
10.7 Method Development for the Analysis of CSF.............................................. 219
References............................................................................................................... 222
10.1 Introduction
Suitable methods are needed to separate polar compounds but small polar analytes
are not well retained during a conventional reversed-phase liquid chromatography
(RPLC) even with high aqueous mobile phases. On the other hand, normal-phase
liquid chromatography (NPLC), using nonaqueous mobile phase, is not generally
suitable for routine application because of poor reproducibility and difficulty in
interfacing with mass spectrometry (MS) detection.1
203

Hydrophilic interaction liquid chromatography (HILIC), a method where polar

stationary phase is used in conjunction with a low aqueous/high organic mobile
phase,2 is a useful alternative to RPLC for applications involving polar compounds.3
This feature helps to eliminate the problem associated with low solubility in high
organic mobile phase of polar compounds, which are usually more soluble in aceto-
nitrile (ACN)/water mobile phase than in hexane/acetate based mobile phase; more-
over, interfacing with electrospray ionization-mass spectrometry (ESI-MS) works
very well with the typical low salt acetonitrile/water HILIC eluents whereas efficient
ionization is not as easily achieved with normal phase solvents.1 HILIC is often more
reproducible and eluent preparation is simpler since there is no need for controlling
over a low water content in the organic solvents.
HILIC coupled with tandem MS (HILIC-MS/MS) has been utilized as a quantita-
tive method for a number of bioanalytical applications with complex matrices.
As reported by Yong Guo and Sheetal Gaiki, HILIC applications have been ini-
tially focused on carbohydrates and peptides and then broadened to include many
small polar compounds, such as pharmaceuticals, toxins, plant extracts, and other
compounds important for analyses in food and pharmaceutical industries.1 Many
bioanalytical applications of HILIC with and without extraction procedures are
reported in literature. In 2003, Naidong reviewed bioanalytical LC-MS/MS methods
using underivatized silica columns and low aqueous/high organic mobile phases for
the analysis of various types of polar analytes, extracted by using protein precipita-
tion (PP), liquid/liquid extraction (LLE), or solid-phase extraction (SPE) and then
analyzed using LC-MS/MS on the silica columns.4 Kadar et al. reported many exam-
ples of bioanalytical applications of HILIC on different biological samples (human
and animals plasma/serum, extracellular fluid of rat, mouse brains, and swine kid-
ney) with and without extraction procedures.5
Here, we describe the application of a HILIC-MS/MS for separation and simul-
taneous quantification of arginine (Arg), asymmetric NG, −NG-dimethylarginine
(ADMA), and symmetric NG, −NG′-dimethylarginine (SDMA), with a short run time
(less than 5â•›min), using a small sample volume of human plasma (0.02â•›m L), use-
ful for routine applications. Furthermore, we report the evaluation of some chro-
matographic aspects (mobile phase, ion pairing, and temperature) influencing the
HILIC-MS/MS method for the detection and quantitation of l-arginine and its
dimethylated derivatives. Finally, we describe the application of our HILIC-MS/MS
method for the determination of dimethylarginine levels in rats’ cerebrospinal fluid
(CSF), for an in vivo microdialysis study, by a fast chromatography method using a
shorter silica column.
10.2 Arginine and Its Dimethyl Derivatives

As reported in literature, lysine, histidine, and arginine N-methylation of pro-
teins is a posttranslational modification that allows the cell to expand its proteome
repertoire.6 Arginine methylation in proteins is catalyzed by two types of pro-
tein arginine methyltransferases (PRMTs) which transfer the methyl group from
S-adenosyl-l-methionine to the guanidine group of arginine residue. Both types of
enzymes catalyze the formation of NG -monomethylarginine (MMA) in proteins.7

Dimethylarginine Analysis by HILIC–MS/MS and Its Applications 205
Type I PRMT enzyme produces ADMA, whereas type II enzyme produces SDMA,
a constitutional isomer of ADMA.7–9
In 1992, Vallance et al.10 first described ADMA as an endogenous competitive
inhibitor of endothelial nitric oxide synthase (eNOS), competing with the natural
substrate l-arginine (Arg), thus decreasing nitric oxide (NO) synthesis. NO is an
ubiquitous intracellular messenger involved in many cellular events. It is formed by
the enzymatic oxidation of l-arginine to l-citrulline in the presence of oxygen and
NADPH using flavin adenine dinucleotide, flavin mononucleotide, heme, thiol, and
tetrahydrobiopterin as cofactors.11,12 In vascular endothelial cells, NO is essential
for the regulation of vascular homeostasis; it is a powerful endogenous vasodila-
tor and inhibits the adhesion of inflammatory cells to the vascular wall, the aggre-
gation of platelets, and the proliferation of smooth muscle cells.13 NO synthesized
by the neuronal NOS (nNOS) is involved in many processes such as learning and
memory.14 The immune responses of the organism also induces an other isoform of
NOS (iNOS).13
SDMA has no inhibitory effect on NOS but may interfere with NO synthesis indi-
rectly by competing with arginine for transport across cell membrane.15
10.3 MS Setup

The described methods used a high performance liquid chromatography (HPLC)
system with an autosampler (Alliance 2695, Waters, Milford, MA). The HPLC was
coupled with a Triple Quadrupole Mass Spectrometer (Micromass Quattromicro,
Waters, Milford, MA) equipped with an electrospray ion source and operated in
positive mode. The optimal conditions of the method were obtained using the flow
injection analysis of aqueous working solutions directly pumped into the electro-
spray probe. Working solutions were obtained by diluting the standard solutions
containing Arg, ADMA, and SDMA (Sigma–Aldrich-Steinheim, Germany), and
13C -l-Arginine (Cambridge Isotope Laboratories, Inc., Andover, MA), as inter-
6
nal standard (IS). Standards solutions were obtained by dissolving Arg, ADMA,
SDMA, and 13C6 -l-Arginine (13C6 -l-Arg) in water and diluted to 0.010â•›mg/mL for
Arg and IS and 0.001â•›mg/mL for dimethylarginines.
Optimal mass conditions were set with a capillary voltage of 3.3â•›kV and a cone
voltage of 25â•›V. Source temperature and desolvation gas temperature for human
plasma method were set at 150°C and 450°C, respectively, while for CSF method
were set at 110°C and 320°C, respectively. The optimum collision energy was
observed at 18â•›eV for Arg, 15â•›eV for IS, 16â•›eV for ADMA, and 14â•›eV for SDMA.
Selected reaction monitoring (SRM) experiments were used to obtain higher spec-
ificity for the target molecules, selecting the molecular ion on the first analyzer and
monitoring the specific product ion, from fragments induced in collision cell, on the
second analyzer. Both ADMA and SDMA showed the same protonated molecule
[M + H]+ at m/z 203, while specific product ions of m/z 46 and 172, respectively, have
been selected for their analysis. Arg and IS showed the protonated molecule [M + H]+
at m/z 175 and 181, respectively; the selected product ions from the fragmentation pat-
tern were of m/z 70 and 74, respectively (Figure 10.1). No interferences were observed
during analytes detection and quantitation.

Daughters of 203ES+
70 4.81e4
100 116
ADMA
158
88
46
203
%
71
72 115
112133
0 m/z
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
100 116
Daughters of 203ES+
1.73e4
203 SDMA
172
%
70 133
88 115
158
71
126
72 113 128 161
0 m/z
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
70
100
Daughters of 175ES+
1.75e5
Arg
%
60 116
130
72 112 158 175
0 m/z
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
100 74
Daughters of 181ES+
8.35e6
13C -Arg
6
61
121
%
135 181
164
76 117
0 m/z
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
Figure 10.1 Full-scan positive-ion mass spectra (fragmentation spectra) of ADMA,

SDMA, Arginine, and IS (13C6 -l-Arginine).

10.4 Chromatographic Evaluations

10.4.1 Implementation of Two HILIC Methods
Two different chromatographic methods were evaluated in our laboratory. Firstly,
the chromatographic method for the quantification of free l-Arginine and dimethyl-
arginines in human plasma was developed and validated using a silica column (Luna
Silica, 3â•›μm, 100â•›mm × 2â•›mm i.d., Phenomenex, Torrance, CA).
We have evaluated the impact of some experimental parameters (mobile phase,
ion pairing, and temperature) that could influence the retention time and peak shape
of the studied compounds in order to achieve the best compromise to obtain a robust
and useful HILIC method for the routine application.
Secondly, a faster chromatographic method was developed using a shorter silica
column (Zorbax RX-SIL, 1.8â•›μm, 50â•›mm × 2.1â•›mm i.d., Agilent Technologies, Santa
Clara, CA) and applied to the protein-free fluids analysis, such as rats’ CSF obtained
by in vivo microdialysis.
Both HILIC methods were performed with an isocratic mobile phase of 90%
Solution A (ACN/trifluoroacetic acid (TFA)/acetic acid; 1000:0.25:10) and 10%
Solution B (water/TFA/acetic acid; 1000:0.25:10) at a flow rate of 0.4â•›m L/min. The
HPLC eluent was split by a Valco valve and approximately only one-third of flow
was introduced into the stainless steel capillary probe. The column temperature was
kept at about 22°C for plasma analysis and at 25°C for CSF analysis.
Using these chromatographic conditions, a short run time and a good peak separation
were obtained, although it was reported that it might not be necessary to completely sep-
arate all the peaks using the MS-MS technology.16–18 In fact, Arg, ADMA, and SDMA
can be unambiguously identified by their specific fragmentation pattern (Figure 10.1).
The robustness of the chromatographic method was evaluated by varying the col-
umn temperatures and the proportion of organic solvent and TFA content in the
mobile phase. For chromatographic method optimization we used two work solu-
tions, containing 50â•›ng/mL of Arg, ADMA, and SDMA; standard solution 1 was
obtained by diluting the standard solutions in ACN/water (90:10) and the standard
solution 2 by using ACN/water (95:5).
Column efficiency value (N) was calculated using the full width at half height
of the peak. The asymmetric factor (As) was calculated using the ratio of the half
widths of the rear and front sides of the peak at 10% of the peak height. SDMA/Arg
and ADMA/SDMA selectivities were measured by the retention time ratios between
SDMA and Arg, and between ADMA and SDMA, respectively.
10.4.2 Effect of ACN Percentage on HILIC and Comparison

between Organic-Rich and Water-Rich Mobile Phase
on Silica Column Chromatography
A typical HILIC application uses mobile phases containing 5%–40% of water in

ACN. The chromatographic effects due to different percentages of ACN in an aqueous
mobile phase, while keeping both TFA (0.025%) and acetic acid (1%) constant, were
evaluated by injecting 0.01â•›mL of the standard solution 1 at a flow rate of 0.4â•›mL/min.

As shown in Figures 10.2 and 10.3, the decrease of ACN in mobile phase causes
reduction in retention time and column efficiency.19 In fact, when ACN percentage
decreased from 95% to 80%, the retention time of Arg changed from 9.56 to 2.12â•›min,
SDMA retention time from 11.45 to 2.21â•›min, and ADMA retention time from 12.15
to 2.29â•›min. Although higher ACN content in HILIC mobile phases should improve
peak shape, we observed a negative effect on peak shape, analytes retention, and
selectivity.
Arginine
3.57 min
175 > 70
ACN:H2O
(90:10)
9.56 min
ACN:H2O
2.12 min (95:5)
ACN:H2O
(80:20)
+0.025% TFA
+1% Acetic acid
4.06 min SDMA

ACN:H2O 203 > 172
(90:10)
2.21 min
11.45 min
ACN:H2O
ACN:H2O
(80:20)
(95:5)
+0.025% TFA
+1% Acetic acid
4.29 min ADMA

ACN:H2O 203 > 46
(90:10) 12.15 min
2.29 min ACN:H2O
(95:5)
ACN:H2O
(80:20)
+0.025% TFA
+1% Acetic acid
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00
Time
Figure 10.2 Effects of increasing ACN in mobile phase. (Reprinted from Paglia, G. et al.,
Rapid Commn. Mass Spectrom., 22, 3809, 2008. With permission from Wiley-VCH Verlag
Gmbh & Co. KGaA.)

3500 Efficiency
3000
2500
Arg
(plates m–1)
2000 ADMA
N
1500 SDMA
1000
500
5)
7)
5)
)
10
15
20
30
40
50
60
:9
:9
5:
0:
5:
0:
0:
0:
0:
0:
(5
(3
(9
(9
(8
(8
(7
(6
(5
(4
(ACN:H2O) : 0.025 TFA : 1 acetic acid
14 Retention time
12
10
8 Arg
min
ADMA
6 SDMA
4
0
5)
5)
7)
)
)
)
)
30
40
50
60
10
15
20
5:
:9
:9
0:
0:
0:
0:
0:
5:
0:
(9
(5
(3
(7
(6
(5
(4
(9
(8
(8
(ACN:H2O) : 0.025 TFA : 1 acetic acid
Figure 10.3 Effects of water percentage in mobile phase. (Reprinted from Paglia, G.
et al., Rapid Commn. Mass Spectrom., 22, 3809, 2008. With permission from Wiley-VCH
Verlag Gmbh & Co. KGaA.)
To exclude the hypothesis of the peak broadening that was caused by sample
overloading, different injection volumes and different water contents in sample dilu-
ent were tested. Volumes of 2, 5, 10, 15, 20, 25, and 30â•›μL of standard solution 1
with a mobile phase containing ACN 90% were injected. No overloading effect was
detected. Moreover, injections of 2, 5, and 10â•›μL, for both standard solutions 1 and 2,
using 90% and 95% ACN/water mobile phases showed peak broadening when mobile
phase was changed from 90:10 ACN/water to 95:5 of ACN/water.
Figure 10.4 shows that the same elution pattern was obtained using a water-rich
mobile phase compared to an ACN-rich mobile phase.19 In fact, with a mobile phase
of water/ACN/methanol/formic acid (95:2.5:2.5:0.1) and 10â•›mM ammonium formate, a
good separation of the studied analytes and high selectivity were achieved, especially
between SDMA and Arg (see also Table 10.1). Nevertheless, the use of a water-rich
mobile phase was not suitable to the silica columns as it would decrease the limit of

3.57 Arginine Arginine

2.71
175 > 70 175 > 70
ACN:H2O (90:10) H2O:ACN:MeOH

+0.025% TFA (95:2.5:2.5)
+1% Acetic acid +0.1% Formic acid
+10 mM Am. formate
SDMA
SDMA
4.06 203 > 172 4.71
203 > 172
4.29 ADMA 5.55 ADMA

203 > 46 203 > 46
1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
(a) (b)
Figure 10.4 Combined effects by mobile phase composition. (Reprinted from Paglia, G.
et al., Rapid Commn. Mass Spectrom., 22, 3809, 2008. With permission from Wiley-VCH
Verlag Gmbh & Co. KGaA.)
detection, efficiency, and asymmetric factor and considerably increase column back
pressure (Table 10.1).19 All of these would lead to a high stress on HPLC hardware
(valve, seals, etc.), a faster impairment of the column, and a decreased ESI efficiency
because of worse desolvation and increased surface tension. Alternatively, by using an
organic-rich mobile phase, it is possible to inject the same organic solvent used for protein

Table 10.1
Comparison of Some Chromatographic Variables
H2O:ACN:MeOH (95:2.5:2.5) +
0.1% Formic Acid + Ammonium ACN:H2O (90:10) + 0.025% TFA +
Formate 10â•›mM 1% Acetic Acid
Flow rate (μL/min) 300 400
Column back 2900 1150
pressure (psi)
Selectivity SDMA/ 1.738 1.137
ARG (α)
Selectivity ADMA/ 1.178 1.057
SDMA (α)
Retention Time (min) Arginine SDMA ADMA Arginine SDMA ADMA
Efficiency, N 2.71 4.71 5.55 3.57 4.06 4.29
(plates m−1)
Asymmetric 374 318 414 1515 1645 1863
factor (As)
Limit of detection 1.79 1.6 1.72 1.25 1.34 1.28
(ng/mL)
4.53 10.42 8.44 0.32 1.61 1.42
Source: Extracted from Paglia, G. et al., Rapid Commn. Mass Spectrom., 22, 3809, 2008. With
permission from Wiley-VCH Verlag Gmbh & Co. KGaA.
precipitation, liquid/liquid extraction, or SPE, directly in chromatographic system with-

out evaporation and reconstitution, resulting in a reduction of sample workup.20–23
10.4.3 Effect of TFA Percentage on the HILIC Method

TFA is commonly used as an additive in HPLC and LC-MS for analysis of basic
compounds. It is also routinely added to aqueous–organic mobile phases utilized in
the HILIC-ESI/MS/MS16,24 to improve peak shape. According to Shou and Naidong,24
acetic acid was also added to alleviate sensitivity loss caused by TFA. The effect of
different percentages of TFA and acetic acid (0.025% TFA—1% acetic acid; 0.020%
TFA—0.8% acetic acid; 0.015% TFA—0.6% acetic acid; 0.010% TFA—0.4% acetic
acid; 0.005% TFA—0.2% acetic acid) in the mobile phase consisting of ACN/water
(90:10) on the retention time, selectivity, and efficiency for Arg, ADMA, and SDMA
was investigated. Figure 10.5 shows that by decreasing the TFA percentage from
0.025% to 0.005%, and correspondingly acetic acid amounts, the retention time (min)
increases from 3.57 to 5.94 for Arg, from 4.05 to 7.15 for SDMA, and from 4.28 to 7.76
for ADMA.19 The selectivity is almost constant between ADMA and SDMA while
between SDMA and Arg increases a little bit, reducing the amount of TFA and acetic
acid. We have not observed significant differences of selectivity and efficiency when
TFA content changed from 0.025% to 0.015% and acetic acid from 1% to 0.6%. In fact,
the proper amounts of the acids used would optimize the chromatographic method.

Retention time
8.50
7.50 Arg
6.50 ADMA
SDMA
min
5.50
4.50
3.50
2.50
0.025 0.020 0.015 0.010 0.005
TFA %
Selectivity
1.250
SDMA/Arg selectivity
1.200
SDMA/ADMA selectivity
1.150
α
1.100
1.050
1.000
0.025 0.020 0.015 0.010 0.005
TFA %
Efficiency
Arg
2400
ADMA
(plates m–1)
SDMA
2000
N
1600
1200
0.025 0.020 0.015 0.010 0.005
TFA %
Figure 10.5 Effects of different TFA percentages in mobile phase. (Reprinted from
Paglia, G. et al., Rapid Commn. Mass Spectrom., 22, 3809, 2008. With permission from Wiley-
VCH Verlag Gmbh & Co. KGaA.)
10.4.4 Effect of Column Temperature on the HILIC Method

Temperature may affect liquid chromatographic physical parameters such as solu-
bility, vapor pressure, viscosity, and diffusivity and subsequently on retention, col-
umn efficiency, selectivity, and column back pressure. Then, the effects of different
column temperatures on efficiency, retention time, and selectivity of HILIC, using
ACN/water/TFA/acetate (90:10:0.025:1) as mobile phase were investigated.
As shown in Figure 10.6, the temperature increase would cause little increase
in both the retention and the efficiency.19 The selectivity remained nearly constant

van’t Hoff plot

4.3
ln k΄ 4.2
4.1
Arg ADMA SDMA
4
0.00305 0.00315 0.00325 0.00335 0.00345
1/Temp. (K)
Selectivity
1.140
1.100 SDMA/Arg selectivity

α
ADMA-SDMA selectivity
1.060
1.020
20 25 30 35 40 50
Column temperature (°C)
2400 Efficiency
Arg
2200 ADMA
2000 SDMA
(plates m–1)
1800
N
1600
1400
1200
20 25 30 35 40 50
Column temperature (°C)
Figure 10.6 Effects of different column temperatures. (Reprinted from Paglia, G. et al.,
Rapid Commn. Mass Spectrom., 22, 3809, 2008. With permission from Wiley-VCH Verlag
Gmbh & Co. KGaA.)
between ADMA and SDMA, but decreases between SDMA and Arg when higher
temperatures were used. The van’t Hoff plot depicted in Figure 10.6 shows a
remarkable endothermic adsorption process.19 According to Dong and Huang,25
the amount of adsorbed solvent on bare silica might change with the tempera-
ture, influencing the partitioning of the analytes between a water-enriched layer
of semi-immobilized eluent on a hydrophilic stationary phase and a relatively
hydrophobic bulk eluent. As expected, the column back pressure would decrease
when higher column temperature was applied. Nevertheless, using a constant flow
rate and an ACN-rich mobile phase, the pressure changed from 1500 to 1100â•›psi by

varying temperature from 20°C to 50°C. Setting the column temperature between
25°C and 30°C seems to be a good compromise.
The main objective of our analytical laboratory is to perform high-throughput
analyses, which requires the analytical method to be highly robust and reproduc-
ible. Thus, many potential sources of variability need to be accurately evaluated
during the development. Endogenous compounds or salts always present in bio-
logical matrices together with high amount of ion pairing agent (TFA, etc.) could
affect the reproducibility, influencing the ionization efficiency of ESI source. Using
diluted samples (1:30–1:50) and a mobile phase with acetic acid, to alleviate the
ionization loss caused by TFA, could be a solution to overcome problems due to
the matrix salts.
As described in the text, Arg, ADMA, and SDMA are well retained using HILIC,
and their chromatographic behavior is mainly influenced by the amount of ACN and
TFA content in the mobile phase. The optimal chromatographic conditions for stan-
dard solutions of the method are: column temperature at 25°C, injection sample vol-
ume of 0.01â•›m L, and mobile phase containing 90% ACN/water, with a fixed amount
of 0.025% TFA and 1% acetic acid, and flow rate of 0.4â•›m L/min. Nevertheless, many
other parameters (different organic modifiers, stationary phases, mobile phase pH,
etc.) should be evaluated as a potential source of chromatographic variability.
10.5 Plasma Method

After evaluation and optimization, a simple, fast, selective, and sensitive HILIC-MS/
MS method has been developed and validated. It allows the simultaneous quanti-
tative analysis of plasma Arg and its dimethylated forms, ADMA and SDMA, in
less than 5â•›min. This is a micromethod, with a short chromatographic run time, that
allows to work with just 0.02â•›m L or less of biological samples. The method is suitable
for routine analysis of human plasma samples for clinical studies.
To evaluate the sample preparation, six different protein precipitation solu-
tions were tested to obtain a rapid and efficient plasma sample preparation:
(a) methanol; (b) methanol:acetonitrile (50:50); (c) methanol:acetonitrile (25:75);
(d) methanol:acetonitrile (25:75) to which 250â•›μL/L TFA and 10â•›m L/L acetic acid
have been added; (e) acetonitrile to which 250â•›μL/L TFA and 10â•›mL/L acetic acid have
been added; (f) acetonitrile. These solutions have been tested on blank and on two
spiked plasma samples at Arg concentration of 105 and 128â•›μM, at ADMA con-
centration of 0.63 and 0.72â•›μM, and at SDMA concentration of 0.60 and 0.67â•›μM.
Based on the best signal intensity, peak shape, and retention time, solution
(d) methanol:acetonitrile (25:75) was chosen for sample preparation.
Blood samples were collected in EDTA tubes from 30 apparently healthy sub-
jects and 33 type 2 diabetic patients, without (n = 24) or with (n = 9) renal dysfunc-
tion. Plasma samples, obtained after centrifugation at 2500â•›r pm for 10â•›min, were
stored at −20°C until the analysis. Plasma (0.02â•›m L) was mixed with 0.02â•›m L of
IS (0.56â•›nmol) and incubated at room temperature (10â•›min). To precipitate proteins,
0.16â•›m L of methanol:acetonitrile mixture (25:75) was added, vortex mixing for 3â•›min
and centrifuging at 3000â•›r pm for 10â•›min. Then, 0.015â•›m L of supernatant was injected
into LC-MS/MS.

The method validation study evaluated a few critical method parameters, such as
calibration curves, QCs, recovery, and carryover.
Calibration standards (calibrators) were prepared by spiking blank human plasma
with Arg and ADMA and SDMA standard solutions and diluting to give concentra-
tions of (μM) 42, 84, 94, 105, 128, 169, and 254 for Arg, of (μM) 0.22, 0.47, 0.57,
0.63, 0.72, 0.91, and 1.29 for ADMA, and of (μM) 0.20, 0.42, 0.53, 0.60, 0.67, 0.87,
and 1.27 for SDMA. They were prepared fresh for each assay and analyzed along
with plasma and QCs. The concentrations of blank plasma for all analytes were cal-
culated using the intercept value of a first calibration curve obtained on theoretical
concentrations of spiked plasma with standard solutions. The lowest concentration
point of the calibration curve was obtained diluting the blank plasma with a saline
buffer (1:2). The linearity of calibration curves was estimated by the coefficients of
correlation (r) which ranged from 0.9926 to 0.9984; slopes ranged from 0.0397 to
0.0685; and intercept values ranged from −0.2466 to −0.0009.
Three levels of plasma quality control (QC) samples, used in the validation
study, were prepared as follows. The lowest QC level consisted of a pooled blank
plasma different from that used for the preparation of the calibration curve. This
pool was analyzed five times and the mean concentrations of 48.6â•›μM for Arg,
0.41â•›μM for ADMA, and 0.33â•›μM for SDMA were calculated using the calibration
curves; this QC level has been used just for imprecision study. The other two QC
levels were obtained by spiking the new pooled human plasma with diluted stock
standard solutions to provide concentrations of (μM) 94.2 and 144.2 for Arg, of
(μM) 0.88 and 1.38 for ADMA, and of (μM) 0.80 and 1.30 for SDMA. The QC
samples were used to evaluate the within-day and between-day imprecision while
inaccuracy and recovery were calculated using the last two spiked plasma QC lev-
els. All QC samples were aliquoted and stored at −20°C with the clinical samples
until use.
QC samples were analyzed after a sequence of unknown samples. The within-day
imprecision (CV%) and inaccuracy (%) were calculated in the same analytical run,
each level of QC samples 15 times. The imprecision (CV%) ranged from 2.3 to 3.9
for Arg, from 5.7 to 8.4 for ADMA, and from 5.8 to 7.5 for SDMA; the inaccuracy
(%) ranged from 1.8 to 2.3 for Arg, from 6.1 to 8.7 for ADMA, and from 5.3 to 6.3
for SDMA.
The between-day imprecision (CV%) and inaccuracy (%) were calculated by ana-
lyzing each level of QC samples once a day for 15 days. The imprecision (CV%)
ranged from 2.3 to 3.0 for Arg, from 6.9 to 8.4 for ADMA, and from 6.1 to 8.0 for
SDMA; the inaccuracy (%) ranged from −0.1 to 0.4 for Arg, from −0.8 to 1.8 for
ADMA, and from −0.1 to 0.9 for SDMA.
Recovery was calculated according to the IUPAC recommendations25 using the
following formula: RA = [QA(O + S) − QA(O)]/QA(S), in which RA is the recovery;
QA(S) the quantity of analyte A added (spiked value); QA(O) is the quantity of analyte
A in the original sample; and QA(O + S) is the quantity of analyte A recovered from
the spiked sample. Recovery was obtained from three different experiments in which
the two pooled blank spiked QC samples were repeated five times (n = 15). The mean
recovery percentage for Arg, using the spiked amounts of (μM) 45.6 and 95.6, was
104% and 101%, respectively; the mean recovery, using the spiked amounts of (μM)

0.47 and 0.97, for ADMA was 106% and 102%, respectively, and for SDMA was
105% and 102%, respectively.
Carryover was evaluated as the percentage of the peak areas ratio between the
mobile phase and the highest concentration of QC plasma sample injected just before
the mobile phase. Each experiment was performed three times. Carryover was less
than 1.2% compared to plasma sample levels. Data from calibration curves and QCs
showed that the method was accurate and precise and it could be used for the deter-
mination of Arg, ADMA, and SDMA in patient samples.
A typical chromatogram obtained from human plasma is depicted in Figure 10.7.26
The retention time was of 2.93â•›min for Arg and IS, 3.30 and 3.53â•›min for SDMA and
ADMA, respectively.
Listed in Table 10.2 are the analytical results of Arg, ADMA, and SDMA from 30
plasma samples of apparently healthy subjects (group 1), and from 33 type 2 diabetic
patients without kidney dysfunction (group 2, n = 24) and with kidney dysfunction
(group 3, n = 9).26 Plasma samples at concentration greater than 250â•›μM for Arg and
1.5â•›μM for ADMA and SDMA were opportunely diluted and reanalyzed.
The plasma Arg, ADMA, and SDMA concentrations calculated by the validated
method are comparable to those reported in literature17,28–30 even though our results
are generally a little bit higher, most likely because of different selection of sub-
jects and of the small number of patients. However, recent reviews suggest that the
mean levels for ADMA measured in plasma of healthy subjects by HPLC or mass
spectrometry-based analytical methods range between 0.4 and 0.6â•›μM,31 which are
in agreement with the results as we reported here.
Comparison among three groups of studied patients was performed using the
analysis of variance (ANOVA) with Bonferroni correction; p < .05 was considered
statistically significant. Arg and ADMA levels of group 2 (diabetic) and group 3 (dia-
betic + KD) are statistically different (p < .05) from group 1 (healthy subjects), and
SDMA levels of group 3 are statistically different from groups 1 and 2. Besides, Arg
and ADMA levels of group 2 vs. group 3 and SDMA levels of group 1 vs. group 2
were not statistically different. In particular, ADMA levels of diabetic patients are
significantly elevated both in groups 2 (+11%) and 3 (+25%) compared to healthy sub-
jects while SDMA levels of group 3 are significantly elevated (2.25-fold) compared to
groups 2 and 1 (Table 10.2). These values are in agreement with those of other stud-
ies,32,33 given the same general observation that plasma SDMA is a sensitive marker
of reduced renal function.33
In conclusion, a practical method is developed for routine analytical applications
and it allows a short analytical cycle of 5â•›min or less and requires minimum sample
volume and preparation.
In literature are reported many chromatographic–mass spectrometric methods
for the measurement of l-Arginine and its dimethyl derivatives in biological fluids;
they have been discussed in detail by Martens-Lobenhoffer and Bode-Böger who
stated that there is an uncertainty in the measurement of Arg, ADMA, and SDMA
due to a few factors.34 The implementation of this method was aimed to overcome
some of the analytical problems linked to these published methods. As regards
sample preparation, most methods include sample extraction by ion-exchange solid
phase, but, as known, small changes in conditioning protocol of the extraction columns

3.30
100 MRM of 4 channels ES+
203 > 172
1.04e4
SDMA
%
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00
3.53
203 > 46
5.12e3
ADMA
%
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00
2.93
181 > 74
1.90e5
13C
6-L-Arginine
%
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00
2.93
175 > 70
1.25e6
Arginine
%
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00
Time
Figure 10.7 The figure depicts a chromatography result of a human plasma sample.
(Reprinted from D’Apolito, O. et al., Clin. Biochem., 41, 1391, 2008. With permission from
Elsevier.)

Table 10.2
Concentrations (μM) of Arginine, ADMA, and SDMA in Plasma
from Apparently Healthy Subjects and Diabetic (Type 2) Patients
with and without Kidney Dysfunction (KD) Reported as Mean
Values (SD)
ARG ADMA SDMA
Subjects n Mean (SD) Mean (SD) Mean (SD)
Apparently healthy (group 1) 30 49.9 (19.0) 0.57 (0.11) 0.52 (0.10)
Diabetic without KD (group 2) 24 64.2 (20.5)a 0.63 (0.10)a 0.53 (0.12)
Diabetic with KD (group 3) 9 66.2 (17.3)a 0.71 (0.24)a 1.17 (0.68)b
Source: Extracted from D’Apolito, O. et al., Clin. Biochem., 41(16–17), 1391, 2008. With
a p < .05 group 2 vs. group 1; group 3 vs. group 1 (Arg and ADMA of group 2 vs. group 3
were not statistically significant).

b p < .05 group 3 vs. group 1 and group 2 (SDMA of group 1 vs. group 2 were not statisti-
cally significant).
(e.g., pH values of the solutions utilized, extraction and elution speed, and other
factors) can affect the extraction efficiency.28 The sample preparation procedure
presented here consisting of a fast and simple protein precipitation could be com-
pleted in 13â•›m in; this procedure is also well tolerated to MS detection and together
with a fast chromatographic separation produces a very short total analytical run
time.26 To evaluate protein precipitation procedure, we tested six different solu-
tions, including the one used by Martens-Lobenhoffer and Bode-Böger16 (acetic
acid is used rather than propionic acid with no remarkable differences in pH value).
The methanol:acetonitrile (25:75) solution was chosen because it gave best peak
intensity and peak shape.
As shown in Figure 10.7, using the HILIC-MS/MS method with SRM quantita-
tive analysis and with 13C6 -l-Arginine as an internal standard, we did not observe
endogenous interferences in the separation of the analytes, as reported by others.34
Indeed, the mass spectrometric detection provides a high specificity for qualitative
and quantitative analyses despite the fact that the sample composition after a simple
extraction step remains very complex.17
In addition, the SRM analysis generally does not require high chromatographic
resolution. Using an HPLC separation of underivatized analytes, the peaks were
unaffected by a noisy baseline as occurs in the methods reported by others in which
the peaks were also not well separated from endogenous interferences.34,35
The compounds were analyzed in their underivatized forms,16,29,30,35 using an
isocratic mobile phase for separation rather than a gradient elution, as reported by
others.29,30 In addition, the best chromatographic performances were obtained by a
controlled low column temperature.
This method could present some minimal limitations, mainly the mobile phase
which has pH 1.38, whereas the suggested optimal working pH range of the column

was from 2 to 8; thus, we would expect that the acidity of the mobile phase could, over
time, impair the efficiency of the column. However, we did not observe any altera-
tions of the chromatographic reproducibility during all steps of this study. Another
consideration is that we used just 13C6 -l-Arginine as an internal standard. As known,
the isotope dilution analysis with labeled analogue is undoubtedly the gold standard
and the labeled ADMA and/or SDMA should be recommended for the analysis.
10.6 C
ollection of CSF Samples from Rat Brain
by Microdialysis
Microdialysis technique is one of the best methodological approaches to evaluate
extracellular concentrations of several endogenous compounds. It is based on the
diffusion of these substances down a concentration gradient from the extracellular
fluid compartment to the dialysis fluid compartment, separated through a semi-
permeable microdialysis membrane, of the microdialysis probe.36,37 Experiments
were conducted in conscious, freely moving Wistar rats (Harlan, S., Pietro al
Natisone, Udine, Italy) weighing 250–300â•›g, housed at a constant room temperature
(22°C ± 1°C) and relative humidity (55% ± 5%) under a 12â•›h light/dark cycle. Food
and water were freely available. For the experiments, the fibers of microdialysis
probe, inserted in the prefrontal cortex (PFC), were perfused, at a constant flow rate
of 2â•›μL/min, with an artificial CSF, buffered at pH 7.3 with a 0.6â•›m M NaH 2PO4,
containing 145â•›m M NaCl, 3â•›m M KCl, 1.26â•›m M CaCl2, 1â•›m M MgCl2, and 1.4â•›m M
Na2HPO4 in distilled water. The microdialysis membrane was allowed to stabilize
for 2â•›h, collecting samples at the end of the stabilization period at 20â•›m in intervals
for 1â•›h for the evaluation of basal levels. The method reported here is an extension
of the above-described HILIC-MS/MS method adapted to the new matrix (CSF).
The aim was to determine the endogenous ADMA and SDMA in CSF samples and
if their basal levels are stable over time.
10.7 Method Development for the Analysis of CSF

This method was optimized using working solutions prepared fresh daily by dilut-
ing aqueous stock standard solutions (1â•›g/L) of all analytes with water to provide
concentrations of 57.4â•›μM for IS and 4.9â•›μM for ADMA and SDMA.
Calibration standards were prepared by diluting the stock standard solutions with
a Krebs–Ringer buffer containing CSF-like electrolyte solution of (mM) 138 NaCl,
11 KCl, 1.5 CaCl2, 1.0 MgCl2, 11 NaHCO3 at pH 7.4, to obtain concentrations of
(nM) 256, 128, 64, 32, 16 for ADMA and SDMA before the analysis. The calibra-
tion solutions were prepared fresh for each assay and analyzed twice along with CSF
and QC samples. Correlation coefficients of calibration curves (r) were greater than
0.998 in all cases for the concentration range chosen.
QC samples were prepared for the validation study by spiking the Krebs–Ringer
solution with working standard solutions to give two levels of concentrations (20
and 100â•›nM) both for ADMA and SDMA, then aliquoted and stored at −20°C until
use. The imprecision of the method was evaluated by analyzing daily the two levels
of QC samples in triplicate for 3 days, together with calibration and CSF samples.

The imprecision (CV%, n = 9) ranged from 10% to 16% for ADMA and from 13%
to 14% for SDMA. The inaccuracy (%) was less than 15% on average for ADMA
and SDMA.
Microdialysis samples were stored at −20°C until the analysis. The analytical
samples were prepared by mixing 0.03â•›m L of CSF sample with 0.03â•›m L of mobile
phase containing 13C6 -l-Arginine (15â•›nmol) as labeled internal standard. The mix
was vortexed vigorously and incubated for 10â•›min and further diluted with 0.12â•›m L
of acetonitrile. An aliquot (0.02â•›m L) of this solution was automatically injected into
the LC-MS/MS system. CSF samples with concentrations over the range of calibra-
tion curves were opportunely diluted and re-assayed.
Compared to the other approaches, the MS detector is a highly selective tool
due to its ability of a structure-based analysis. The direct separation of analytes
will enhance the efficiency of analysis and will also reduce the time needed for
the sample preparation and ultimately enhances the analytical samples’ throughput.
Due to the advantage of HILIC separation and MS detection, the coupling between
these two tools starts to seem attractive for the analyses of neurotransmitters and
related compound.
Figure 10.8 depicts the selected ion chromatograms acquired for a typical in vivo
CSF sample in MRM mode for ADMA, SDMA, and IS.38 The peak at retention
time of 1.87â•›min was for IS, 1.98â•›min for SDMA, and 2.03â•›min for ADMA. To avoid
ion suppression, a switch valve was used to divert the eluent to waste so that signals
acquisition occurs only from 1.3 to 2.8â•›min. Moreover, ion suppression was strongly
minimized by diluting the sample and using the chromatographic parameters and
column, as reported in Section 10.3.
Analytes values (mean ± SD) measured in CSF samples of rats PCF (n = 5) col-
lected over time intervals of 20, 40, and 60â•›min were of (nmol/L) 72 ± 9, 74 ± 40,
and 55 ± 35, respectively, for ADMA and of (nmol/L) 54 ± 6, 55 ± 30, and 40 ± 16,
respectively, for SDMA. Data were analyzed by using one-way ANOVA for repeated
measures. Statistical computations were performed with Sigma Stat software pack-
age version 3.1.
These results showed that both ADMA and SDMA levels were detectable in the
PFC of freely moving rats by combining microdialysis technique and HILIC-MS/MS.
Furthermore, the present study showed that ADMA and SDMA extracellular concen-
trations did not differ among the three time points considered. Since NO is a cell sig-
naling substance of paramount importance in various tissues and its altered generation
has been implicated in the pathogenesis of several diseases, it is of great importance to
determine the concentration of circulating methylated arginine analogues that occur
free in the brain and can exert a control on signal transduction through the nitrergic
system. As mentioned earlier, ADMA and SDMA are the major circulating forms
of methylarginines in human plasma and the elevated ADMA concentrations are an
established risk marker for various diseases. However, little information is available
on the levels of this molecule in CSF. Direct CSF determinations are essential for the
estimation of neuronal NO production and the involvement of oxidative stress pro-
cesses. These experiments showed that ADMA and SDMA levels were detectable in
vivo in the PFC of rats by using a method which allows the simultaneous determina-
tion of both endogenously produced compounds using only a fast dilution as sample

2.03
203 > 46
905
ADMA
%
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00
1.98
203 > 172
1.29e3
SDMA
%
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00
1.87
181 > 74
1.01e4
13C
6-Arginine
%
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00
Time
Figure 10.8 Chromatograms of a CSF sample. (Reprinted from Zotti, M. et al., J. Sep.
Sci., 31, 2511, 2008. With permission from Wiley-VCH Verlag Gmbh & Co. KGaA.)

preparation and with a total chromatographic run time of 3â•›min or less.38 In this study,
the combined advantages of this technique with microdialysis is useful to realize an in
vivo approach; the present findings suggest that HILIC-MS/MS represents a sensitive
technique for the simultaneous monitoring of dimethylarginines in CSF of rat PFC.
Interestingly, these results could have important implications for studying the role of
ADMA and SDMA in several diseases of the central nervous system.
References
1. Guo, Y. and Gaiki, S. 2005. Retention behavior of small polar compounds on polar sta-
tionary phases in hydrophilic interaction chromatography. Journal of Chromatography
A 1074:71–80.
2. Alpert, A.J. 1990. Hydrophilic-interaction chromatography for the separation of peptides,
nucleic acids and other polar compounds. Journal of Chromatography 499:177–196.
3. Olsen, B.A. 2001. Hydrophilic interaction chromatography using amino and silica
columns for the determination of polar pharmaceuticals and impurities. Journal of
Chromatography A 913:113–122.
4. Naidong, W. 2003. Bioanalytical liquid chromatography tandem mass spectrometry
methods on underivatized silica columns with aqueous/organic mobile phases. Journal
of Chromatography B 796:209–224.
5. Kadar, E.P. et al., 2008. Rapid determination of the applicability of hydrophilic inter-
action chromatography utilizing ACD labs log D suite: A bioanalytical application.
Journal of Chromatography B 863:1–8.
6. Teerlink, T. 2007. HPLC analysis of ADMA and other methylated L-arginine analogs in
biological fluids. Journal of Chromatography B 851:21–29.
7. Gehrig, P.M. et al. 2004. Fragmentation pathways of N(G)-methylated and unmodi-
fied arginine residues in peptides studied by ESI-MS/MS and MALDI-MS. Journal of
American Society of Mass Spectrometry 15:142–149.
8. Bedford, M.T. and Richard, S. 2005. Arginine methylation an emerging regulator of
protein function. Molecular Cell 18:263–272.
9. Valtonen, P. et al. 2005. Comparison of HPLC method and commercial ELISA assay
for asymmetric dimethylarginine (ADMA) determination in human serum. Journal of
Chromatography B 828:97–102.
10. Vallance, P. et al. 1992. Accumulation of an endogenous inhibitor of nitric oxide synthe-
sis in chronic renal failure. Lancet 339:572–575.
11. Giraldi-Guimarães, A. Bittencourt-Navarrete, R.E., and Mendez-Otero, R. 2004.
Expression of neuronal nitric oxide synthase in the developing superficial layers
of the rat superior colliculus. Brazilian Journal of Medical and Biological Research
37:869–877.
12. Akyol, O. et al. 2004. Nitric oxide as a physiopathological factor in neuropsychiatric
disorders. In Vivo 18:377–390.
13. Cooke, J.P. and Dzau, V.J. 1997. Nitric oxide synthase: Role in the genesis of vascular
disease. Annual Review of Medicine 48:489–509.
14. Fedele, E., and Raiteri, M. 1999. In vivo studies of the cerebral glutamate receptor/NO/
cGMP pathway. Progress in Neurobiology 58:89–120.
15. Lluch, P. et al. 2006. Accumulation of symmetric dimethylarginine in hepatorenal syn-
drome. Experimental Biology and Medicine (Maywood) 231:70–75.
16. Martens-Lobenhoffer, J. and Bode-Böger, S.M. 2006. Fast and efficient determination
of arginine, symmetric dimethylarginine, and asymmetric dimethylarginine in biologi-
cal fluids by hydrophilic-interaction liquid chromatography-electrospray tandem mass
spectrometry. Clinical Chemistry 52(3):488–493.

17. Kushnir, M.M. et al. 2005. Assessing analytical specificity in quantitative analysis using
tandem mass spectrometry. Clinical Biochemistry 38:319–327.
18. Schwedhelm, E. et al. 2005. Liquid chromatography-tandem mass spectrometry method
for the analysis of asymmetric dimethylarginine in human plasma. Clinical Chemistry
51:1268–1271.
19. Paglia, G., D’Apolito, O., and Corso, G. 2008. Precursor ion scan profiles of acylcar-
nitines by atmospheric pressure thermal desorption chemical ionization tandem mass
spectrometry. Rapid Communication in Mass Spectrometry 22(23):3809–3815.
20. Song, Q. and Naidong, W. 2006. Analysis of omeprazole and 5-OH omeprazole in
human plasma using hydrophilic interaction chromatography with tandem mass spec-
trometry (HILIC-MS/MS)—Eliminating evaporation and reconstitution steps in 96-well
liquid/liquid extraction. Journal of Chromatography B 830:135–142.
21. Naidong, W. et al. 2004. Direct injection of 96-well organic extracts onto a hydro-
philic interaction chromatography/tandem mass spectrometry system using a silica sta-
tionary phase and an aqueous/organic mobile phase. Rapid Communication in Mass
Spectrometry 18:2963–2968.
22. Li, A.C. et al. 2004. Direct injection of solid-phase extraction eluents onto silica columns
for the analysis of polar compounds isoniazid and cetirizine in plasma using hydrophilic
interaction chromatography with tandem mass spectrometry. Rapid Communication in
Mass Spectrometry 18:2343–2350.
23. Naidong, W. et al. 2002. Liquid chromatography/tandem mass spectrometric bioanalysis
using normal-phase columns with aqueous/organic mobile phases—A novel approach of
eliminating evaporation and reconstitution steps in 96-well SPE. Rapid Communication
in Mass Spectrometry 16:1965–1975.
24. Shou, W.Z. and Naidong, W. 2005. Simple means to alleviate sensitivity loss by tri-
fluoroacetic acid (TFA) mobile phases in the hydrophilic interaction chromatography-
electrospray tandem mass spectrometric (HILIC-ESI/MS/MS) bioanalysis of basic
compounds. Journal of Chromatography B 825:186–192.
25. Dong, L. and Huang, J. 2007. Effect of temperature on the chromatographic behav-
ior of epirubicin and its analogues on high purity silica using reversed-phase solvents.
Chromatographia 65:519–526.
26. D’Apolito, O. et al. 2008. Development and validation of a fast quantitative method
for plasma dimethylarginines analysis using liquid chromatography–tandem mass spec-
trometry. Clinical Biochemistry 41:1391–1395.
27. Burns, D.T., Danzer, K., and Townshend, A. 2002. Use of the terms “recovery” and
“apparent recovery” in analytical procedures (IUPAC recommendations 2002). Pure
and Applied Chemistry 74:2201–2205.
28. Martens-Lobenhoffer, J. and Bode-Böger, S.M. 2006. Measurement of asymmetric
dimethylarginine (ADMA) in human plasma: From liquid chromatography estima-
tion to liquid chromatography–mass spectrometry quantification. European Journal of
Clinical Pharmacology 62:61–68.
29. Kirchherr, H. and Kühn-Velten, W.N. 2005. HPLC–tandem mass spectrometric method
for rapid quantification of dimethylarginines in human plasma. Clinical Chemistry
51:249–252.
30. Bishop, M.J. et al. 2007. Direct analysis of un-derivatized asymmetric dimethylarginine
(ADMA) and L-arginine from plasma using mixed-mode ion-exchange liquid chroma-
tography–tandem mass spectrometry. Journal of Chromatography B 859:16416–16419.
31. Bode-Böger, S.M., Scalera, F., and Ignarro, L.J. 2007. The L-arginine paradox:
Importance of the L-arginine/asymmetrical dimethylarginine ratio. Pharmacology &
Therapeutics 114:295–306.
32. Fleck, C. et al. 2001. Serum concentrations of asymmetric (ADMA) and symmetric
(SDMA) dimethylarginine in renal failure patients. Kidney International 59:S14–S18.

33. Fleck, C. et al. 2003. Serum concentrations of asymmetric (ADMA) and symmetric
(SDMA) dimethylarginine in patients with chronic kidney diseases. Clinica Chimica
Acta 336:1–12.
34. Martens-Lobenhoffer, J. and Bode-Böger, S.M. 2007. Chromatographic–mass spectro-
metric methods for the quantification of L-arginine and its methylated metabolites in
biological fluids. Journal of Chromatography B 851:30–41.
35. Vishwanathan, K. et al. 2000. Determination of arginine and methylated arginines
in human plasma by liquid chromatography–tandem mass spectrometry. Journal of
Chromatography B 748:157–166.
36. Benveniste, H. et al. 1989. Determination of brain interstitial concentrations by micro-
dialysis. Journal of Neurochemistry 52:1741–1750.
37. Lindefors, N., Amberg, G., and Ungerstedt, U. 1989. Intracerebral microdialysis:
I. Experimental studies of diffusion kinetics. Journal of Pharmacological Methods
22:141–156.
38. Zotti, M. et al. 2008. Determination of dimethylarginine levels in rats using HILIC-MS/
MS: An in vivo microdialysis study. Journal of Separation Science 31:2511–2515.

11 Polar Functional Groups
for HILIC Method
Zhigang Hao
Contents
11.1 Introduction................................................................................................... 225
11.2 Ionic Functional Groups in HILIC Separation.............................................. 227
11.2.1 Effect of Stationary Phase Composition............................................ 228
11.2.2 Effect of Buffer Concentration (Ionic Strength)................................ 233
11.2.3 Effect of Column Temperature.......................................................... 239
11.3 Nonionic Polar Functional Groups in HILIC Separation..............................244
11.3.1 Effect of Mobile Phase Composition.................................................244
11.3.2 Effect of Column Temperature.......................................................... 252
References............................................................................................................... 256
11.1 Introduction
Using polar stationary phases such as bare silica paired with aqueous-organic mobile
phases like water-acetonitrile (H2O-ACN) to separate hydrophilic ingredients can
be traced back to the 1970s.1,2 The name hydrophilic interaction chromatography
(HILIC) was coined by Alpert in 1990.3 Dr. Weng initiated the first HILIC review
on bioanalytical application with underivatized silica column and mass spectrometer
(MS) detection in 2003.4 Bare silica has no secondary modification on its surface, and
it is still the most popular stationary phase in HILIC applications due to column stabil-
ity and low back-pressure. Different types of silica materials (types A, B, and C) have
been developed for method selectivity. The rapid growth of HILIC chromatography
began around 2003 and since then hundreds of applications with many new station-
ary phases have been successfully applied to polar analyte separation. Hemstrom and
Irgum constructed review of the HILIC field in 2006 and attempted to ascertain the
extent to which partition or adsorption accounted for the separation mechanism.5
In 2008, Prof. Laemmerhofer edited a special HILIC issue in Journal of Separation
Science. Prof. Jandera reviewed HILIC stationary phases and differentiated the selec-
tivity between HILIC and reverse phase modes.6 It was postulated that the mobile
phase similarity in both HILIC and reverse phase systems provided an ideal com-
bination for two-dimensional (2D) chromatographic separation. Polar analytes are
separated in HILIC mode, usually in the first dimension, whereas less hydrophilic
ones in reverse phase mode. Univariate and multivariate approaches in HILIC method
development were summarized by Prof. Vander Heyden’s group.7 Parameters including
225

stationary phase, column temperature, mobile phase composition, and flow rate for the
detection of biological and nonbiological samples were discussed in this review. Our
group discussed the importance of column temperature and mobile phase for HILIC
selectivity.8 Drs. Nguyen and Schug highlighted the advantages of HILIC separation
coupled with ESI-MS detection.9 The HILIC-MS method application in biological
assays was reviewed by Dr. Hsieh.10 Several new HILIC studies were also included in
this special issue. These articles gave readers an updated and representative overview
of what can be accomplished using HILIC mode, the typical application fields, and the
benefits for polar analytes as compared to traditional reverse phase chromatography.
The need to identify novel analytes such as those found in biological fluid metabolites
and synthetic mixtures leads to analytical method development. With research advances
in genomics, proteomics, and metabolomics, more and more small polar components
have appeared in the analytical laboratory. These compounds are usually not retained
well enough in a reverse phase high performance liquid chromatography (HPLC) sys-
tem to satisfy the recommendation from the Center for Drug Evaluation and Research
(CDER), which suggests a minimum capacity factor (k′) value of >2.0 to ensure ade-
quate separation of the analytes from un-retained matrix components.11 However, these
analytes are retained more efficiently in HILIC mode, and the use of a high ratio of
organic solvent such as ACN in the mobile phase is compatible with MS detection.
For reverse phase chromatography, retention is considered to be (ideally) controlled
by partition since most functional groups such as silanols are covered/shielded by carbon
chain materials (C8 or C18) in the stationary phase. Compared to reverse phase HPLC,
columns for HILIC have a much wider variety of functional groups such as silanol,
amine, amide, diol, cyano, poly-succinimide, sulfoalkylbetaine, and zirconia. All polar
functionalities including anionic and cationic moieties on the HILIC packing material
surface can absorb some water (0.5%–1.0%) to form a stagnant water-enriched layer
between the mobile and stationary phases, especially when the water ratio is low (usu-
ally less than 40%) in the mobile phase. This layer is immobilized and can be considered
as a portion of the stationary phase. The transition between adsorption and partition
mechanisms is probably continuous as the water content in the mobile phase gradually
increases.6,12 The term “HILIC” refers to practical application possibilities rather than
to a special retention mechanism. Of course, this is the simplest way to consider the
HILIC separation compared to reverse phase chromatography. The more hydrophilic an
analyte, the more it associates with the stagnant water-rich layer and the later it elutes
out. The functional moieties in the structure that convey this property to highly polar
compounds are either charged groups or groups capable of entering strong dipolar or
hydrogen bonds. An empirical formula can be used to describe analyte polarity:
(number of polar group)

Polarity = (11.1)
(number of carbon)
where the polar group can be ionic, protic, CN, C=O, and CONH. The analyte can
become very hydrophilic (the polarity is high) when a large number of polar groups
relative to the number of carbons are present in a structure. A significant character-
istic of polar analytes is that they contain multiple polar groups. If only partition
dominates the HILIC separation, we can use the empirical formula (11.1) to predict

Polar Functional Groups for HILIC Method 227
the elution order, which will be reversed compared to that in reverse phase chro-
matography. However, the HILIC stationary phase is different from reverse phase
materials in that polar functional groups bound on a HILIC column are not usually
highly covered or shielded. They can be exposed directly to the mobile phase and
even to the polar functional groups on analyte structures inside the stagnant liquid
layer. The variety of polar functional groups bound on a HILIC phase exhibit more
significant, strong, and different interactions with the hydrophilic functional groups
on analyte structures. Some authors consider these secondary interactions undesir-
able,13 whereas others utilized them for better method selectivity.14,15 The structural
diversity of HILIC surface chemistries provides analysts with a lot of opportunity for
better separation of polar analytes in terms of column selection and the manipulation
of the experimental conditions. So far, less attention has been paid to electrostatic (or
called as ionic), hydrogen-bond, and hydrophobic interactions for HILIC retention
capability. More importantly, these interactions specifically correspond to the differ-
ent functional group moieties in both analyte and HILIC phase structures. An elec-
trostatic interaction (attraction or repulsion) is produced primarily by either cations
or anions from both analytes and stationary phases. Experimentally, mobile phase
pH, buffer species, concentration, and column temperature can impact upon this
type of interaction. The hydrogen bond usually can be generated by protic groups,
and its interaction strength can be affected by varying the mobile phase composition
(either protic or aprotic organic solvents). The hydrophobic interaction comes mainly
from nonpolar moieties in an analyte structure, and it becomes relatively strong only
when a high ratio of water content (>40%) is present in the mobile phase. It is not a
typical HILIC separation condition and will not be discussed in this chapter.
When polar analytes from different projects are sent to an analytical laboratory,
columns are selected based on analyte structures. When an ionic group (either posi-
tive or negative) is present in an analyte, logically the anionic or cationic stationary
phase under a designed experimental condition should be selected to enhance or
reduce the analyte retention due to electrostatic attraction or repulsion. If multiple
protic groups are presented in an analyte such as a carbohydrate, a protic station-
ary phase such as an amide column might be used for retention on the column due
to strong hydrogen-bond interaction. Different polar functional groups from both
analytes and stationary phases can exhibit varying levels of interaction strength in
a HILIC separation. The most polar analytes usually contain more than one polar
functional group and each group can contribute to some degree. The individual inter-
action strength can be manipulated with different experimental conditions such as
column selection, pH, buffer concentration, mobile phase composition, and column
temperature. This chapter focuses on specific interactions between polar functional
groups in both analytes and the stationary phase. For better understanding, ionic and
nonionic groups will be reviewed and discussed in some detail separately.
11.2 Ionic Functional Groups in HILIC Separation

An ionic compound is a chemical compound in which ions are held together by the
strong electrostatic force between oppositely charged groups. Small ionic compounds
tend to dissolve in polar solvents like water and form strong electrostatic bonds. Usually,

the positive-to-positive or negative-to-negative ions exhibit an electrostatic repulsive

interaction. In a HILIC system, the ionic functional groups, like other polar groups,
prefer
Â� to stay inside the stagnant water-rich layer and can interact with ions either inside
the stagnant layer (usually from buffer) or bound on the stationary phase by either elec-
trostatic attraction or repulsion. The partition retaining power of an analyte can be deter-
mined by the empirical formula (Equation 11.1), as described earlier. The more polar
groups an analyte possesses, the stronger its retaining power. The retention strength from
ionic interaction can be determined by several factors. The first factor is the presence of
ionic functional group(s) (either cationic or anionic) in the analyte structures or bound
on the stationary phase under a specific pH condition. Certainly, the silanol, amine,
and zwitterion chromatography (ZIC) stationary phases should be considered first in the
HILIC separation of ionic analytes. The secondary factor is inclusion of buffer salts in
the separation system. Different buffer species and their concentration could affect the
ionic interaction between analytes and the stationary phase. The third factor is column
temperature when a significant analyte transferring enthalpy is involved between the
dynamic mobile phase and the stagnant water-rich layer. These three factors for ionic
analyte separation will be reviewed and discussed in the following sections. The organic
solvent compositions of the mobile phase can impact on ionic group retention in a HILIC
system. However, this is more relevant to the hydrogen-bonding retention mechanism
and will be reviewed in the section of nonionic functional groups in HILIC separation.
11.2.1 Effect of Stationary Phase Composition

The role and classification of the stationary phase in HILIC separation efficiencies
have been reviewed by Prof. Tanaka’s group.16 Numerous ionic analytes, including
acids, bases, amino acids, peptides, nuclei bases, and nucleosides, have been separated
on bare silica, amino-silica, amide-silica, poly(succinimide), sulfoalkylbetaine, diol,
cyano, cyclodextrin, and triazol columns. The application of the cyano-silica column
to the HILIC mode is still limited probably because the cyano group does not have
sufficient association with water molecules to generate an effective water-rich layer for
partitioning. The residual silanol groups under the cyano-group perhaps are unable to
stimulate accumulation of water due to the surface shielding by organic ligands.16–18
In general, different HILIC stationary phases exhibit enough variability to retain a
wide spectrum of analytes. For example, bare silica columns have been successfully
applied for the separation of many nitrogen-containing analytes such as atenolol,19 carve-
dilol,20 levofloxacin,21 doxazosin,22 donepezil,23 and glycyl-sarcosine.24 These nitrogen-
containing analytes usually provide a positive charge under an acidic condition. However,
electrostatic interaction between basic analytes and the different bare silica materials has
not been discussed intensively enough to make an informed decision on column selec-
tion. Type-A silica materials usually have a lower average pKa value and should provide
a stronger electrostatic attraction to the positive charged group on polar analytes under
most pH conditions. When this attraction is the secondary retaining power in the HPLC
separation, tailing phenomenon can be observed. To eliminate interaction with the strong
acidic silanol groups on the silica surface, newer silica materials, so-called types B and
C, were developed. Silanol groups on type-B and type-C silica surfaces usually have
a higher average pKa value. However, when the primary partition in HILIC separation

0.275
0.220
Nicotinic acid - 6.824
Picolinic acid - 7.840

0.165
AU
0.110
0.055
0.000
0.00 2.00 4.00 6.00 8.00 10.00 12.00
min
Figure 11.1 Representative chromatogram of nicotinic and picolinic acids at 70:30 ACN/
buffer with a Phenomenex Luna NH2 column. The buffer was made by equal volume of 200â•›m M
ammonium acetate and 100â•›mM glacial acetic acid in water. (Adapted from Christopherson, M.J.
et al., J. Liq. Chromatogr. Relat. Technol., 29, 2545, 2006. With permission.)
cannot provide enough retention capability, the electrostatic attraction from type-A sil-
ica materials can provide a very powerful retention for basic analytes compared with
type-B and type-C silica materials. Of course, type-A silica may provide a stronger elec-
trostatic repulsion to acidic functional groups, resulting in very weak analyte retention.
A comparison of nicotinic acid separation on different columns is a good example of
electrostatic interaction in HILIC chromatography. Different types of stationary phases,
Thermo Hypersil silica column and Phenomenex Luna NH2, were used to separate
nicotinic acid. Nicotinic acid retention on the amino phase (Luna NH2) column shown
in Figure 11.1 is much stronger than its retention on the Hypersil silica column shown in
Figure 11.2, even though a much stronger mobile phase containing 30% water was used
with the amino phase column. The different retentions from Figures 11.1 and 11.2 can
be explained by electrostatic interaction of functional groups between analytes and the
stationary phase. In Figure 11.2, the carboxyl group (RCOOH) on analytes was deprot-
onated and the amine group (RNH2) on column was protonated under the experimental
pH condition.15 The negative ion (RCOO−) can be electrostatically attracted by the posi-
tive ion (NH3+). Nicotinic acid and picolinic acid can be retained longer, even though a
high ratio of water (30%) was presented in the mobile phase. However, Hypersil silica
column in Figure 11.2 is a more acidic silica type-A material. The silanol group (Si-OH)
bound on this material surface can be deprotonated into a negative ion (Si-O−) under
the pH condition employed. The negative silanol ions give a repulsive interaction to the
negative carboxyl group (RCOO−) on analytes and result in very short retention time
even though a low ratio of water (2%) was presented in the mobile phase.
The repulsive interaction between silanol and carboxyl groups is further elucidated
by the data in Table 11.1.14 A Hypersil silica column was used. The chemical structures
of glycine (G), diglycine (DG), triglycine (TG), N-[1-deoxy-d-glycose-1-yl]-glycine
(GG), N-[1-deoxy-d-glycose-1-yl]-diglycine (GDG), and N-[1-deoxy-d-glycose-1-yl]-
triglycine (GTG) are shown in Figure 11.3. The equilibrium reactions for ionic func-
tional groups in both analytes and stationary phase are shown in the following formula:
Si Si-OH RCOO RCOOH RNH 2 RNH 3+

1.5e5
NiAc
1.5e5
NiUAc
Intensity, cps
1.5e5
NiNH2
1.5e5
ISTD
0.2 0.6 1.0 1.4 1.8 2.2 2.6 3.0 3.4 3.8 4.2 4.6
Time, min
Figure 11.2 HILIC-APCI-MS/MS chromatograms for nicotinamide (NiNH2), nicotinic

acid (NiAc), 6-methylnicotinic acid (ISTD), and nicotinuric acid (NiUAc), using high organic
mobile phase containing 2% (solid line) or 5% (dotted line) water with a Thermo Hypersil
silica column. (Adapted from Hsieh, Y. and Chen, J., Rapid Commun. Mass Spectrom., 19,
3031, 2005. With permission.)
All equilibrium reactions will move to the right-hand side when mobile phase acidity
increases. Amine NH(1–2) groups in the analyte structures should be fully protonated
to positive ammonium NH(2–3)+ groups based on the pH values of different formic
acid concentrations in the mobile phase (from 0.1% to 0.7%) and the pKa values of G,
DG, TG, GG, GDG, and GTG listed in Table 11.1, even though the real pH values
in aqueous organic mobile phase are slightly different from those in water media.
Because their pKa values are 3.5, the neutral form of carboxyl groups in the DG, TG,
GDG, and GTG structures will be dominant under mobile phase pH conditions from
2.20 to 2.68. The ionic form of silanol groups on the silica surface will decrease
when the formic acid concentration increases because these groups usually have a
broad range of pKa values. As expected, DG, TG, and their Amadori compounds,
GDG and GTG, are retained less when formic acid content is increased because
less ionic interaction between Si-O − and NH(2–3)+ results in a weaker retention. It
was interesting to note that retention times of G and GG were completely different
from the other four analytes. They were retained longer when the formic acid con-
tent was increased from 0.1% to 0.7% in the mobile phase. The observation can be
explained by the equilibrium reactions described earlier. The pKa of silanol groups

Table 11.1
Retention Times of Six Analytes, G, DG, TG, GG, GDG, and GTG under
the HPLC Condition of Hypersil Silica Column, 100 × 1â•›mm, Particle
Size: 3â•›𝛍m with Mobile Phase 2% Water and 98% Methanol and the
Variable Formic Acid Contents in Both Solvents from 0.1% to 0.7%
and Column Temperature at 30°C
0.1%FA 0.2%FA 0.3%FA 0.4%FA 0.5%FA 0.6%FA 0.7%FA
(pH:2.68)a (pH:2.50) (pH:2.41) (pH:2.34) (pH:2.28) (pH:2.25) (pH:2.20)
G 3.92 4.74 5.15 5.46 6.69 6.90 6.80
(pKa: 2.34, 9.6)
DG 10.62 10.52 10.31 9.90 8.87 8.67 8.26
(pKa: 3.5, 8.0)
TG 13.62 13.11 12.49 11.87 10.43 10.12 9.61
(pKa: 3.5, 8.1)
GG 3.05 3.15 3.35 3.35 4.38 4.69 4.79
(pKa: 3.5, 8.5)b
GDG 8.20 7.59 7.48 7.28 6.87 6.56 6.35
(pKa: 3.5, 8.0)b
GTG 10.28 9.15 8.63 8.22 7.29 7.09 6.99
(pKa: 3.5, 8.0)b
Source: Adapted from Hao, Z. et al., J. Chromatogr. A, 1147, 165, 2007. With permission.
a The pH values in this row were measured in water media and the real pH values in aqueous organic
mobile phase could be slight different in water media.

b The pK data for GG, GDG, and GTG were not measured and the data were just assumed to be similar
a
to G, DG, and TG, respectively.
O O O O OH
H H
NH2 N N
HO HO N OH
H
O OH OH
G GTG
O O OH OH
H H
N N OH
HO NH2 HO N
O H
DG O O OH
GDG
O O O O OH
H H
N NH2 N
HO N HO OH
H OH OH
O GG
TG
Figure 11.3 Structures of G, DG, TG, GG, GDG, and GTG.

is the average value of all silanols present on the silica surface. When the pH falls
below 2.1, the ionization of silanols is suppressed with the exception of the most
acidic ones.25 The silanol groups on type-A silica provide a lower average pKa value.
A significant amount of the ionic form, Si-O −, is presented on the stationary phase
surface under the pH conditions listed in Table 11.1. At a pH of 2.68, the ionic form
of carboxyl group, COO −, is dominant in the G and GG structures due to their pKa
values of 2.34 and 2.5, respectively. The electrostatic repulsion between COO − and
Si-O − results in less retention. When formic acid content is increased, both carboxyl
acid and silanol ionizations were suppressed. The decreased repulsive force provided
a longer retention time. Indeed, the retention times were increased from 3.92 and
3.05 to 6.80 and 4.79â•›min for G and GG, respectively, when the formic acid concen-
tration was changed from 0.1% to 0.7% (Table 11.1). A similar situation was found in
sodium cromoglicate separation on an Atlantis HILIC-Si column. Its retention time
became longer when the mobile phase pH increased from 4.2 to 5.8.26 Usually, a bare
silica column cannot provide a very strong retention capability to acidic groups in
polar analytes under an acidic condition without buffer applications.27,28
The repulsive interaction is not only present in the negative-to-negative groups
but also in the positive-to-positive groups between analytes and the stationary phase.
Dr. Liu et al. found that only a Zorbax-NH2 column caused four hydrazines to be
eluted out before the void volume compared to three other columns (YMC-Pack
Diol-120-NP, Amide-80, and ZIC-HILIC; see Figure 11.4).29 The Zorbax-NH2 col-
umn also showed a weak retention to basic 4-(aminomethyl)pyridine and related
compounds when compared with six other columns.18
2
1 T0 2/3
0.75 1
0.5 2
1 3
0.25 31 4
4 (d)
0
mV
T0
–0.25 (c)
–0.5 T0
4 (b)
–0.75
–1 1/2/3/4 T0
(a)
–1.25
0 5 10 15 20 25 30 35 40
min
Figure 11.4 The separation of hydrazines on different columns. Parameters for the studies
included 30°C column temperature, 0.4â•›mL/min flow rate with splitter, CLND detector system
set at 10, 50°C combustion furnace, 50â•›mL/min argon, 280â•›mL/min oxygen, 75â•›mL/min makeup
(argon), 30â•›mL/min ozone, 5°C cooler, gain x1, and 750â•›V on PMT. The analyte concentrations
were about 30–70â•›μg/mL in water/ethanol (20/80, v/v). Injection volume was 10â•›μL. Mobile phase
was formic acid/water/ethanol (0.5/20/80, v/v/v). 0.1% acetonitrile (v/v) in ethanol was used as
a void volume marker. 1: 1,2-dimethylhydrazine, 2: 1,1-dimethylhydrazine, 3: methylhydrazine,
and 4: hydrazine. (a) Zorbax NH2, (b) Diol, (c) Amide-80, and (d) ZIC HILIC. (Adapted from Liu,
M. et al., J. Chromatogr. A, 1216, 2362, 2009. With permission.)

The secondary retaining force can be an undesired interference in HPLC separa-

tion causing tailing or broad peak issues.13 But it can also be a very important reten-
tion mechanism if we understand which column is able to provide such a retaining
force for our analytes of interest. Switching from bare silica to an amino-phase for
acidic or even basic functional groups should introduce a significant separation capa-
bility. More recently, Alpert has coined another chromatographic acronym, ERLIC,
for electrostatic repulsion hydrophilic interaction chromatography.30 The HILIC
mobile phase (a high ratio of organic solvents) is combined with an ion-exchange
column (a charged stationary phase) in this methodology. The separation strategy
in ERLIC superimposes a second mode of chromatography that selectively reduces
the retention of analytes that are usually the most strongly retained. Actually, many
HILIC columns have this ion-exchange capability. For example, negative residual
silanol groups are present in all silica-based HILIC columns,31 which not only pro-
vide an extra retention capability to the basic or positive-charged functional groups
but can also generate an electrostatic repulsion to acidic or negative-charged func-
tional groups. The negative residual silanol groups are more significant in type-A
silica material. For the amine column, the positive charge on the outer layer of the
stationary phase can give a strong electrostatic attraction to acidic analytes, even
though the residual silanols are present in the inner layer of the stationary phase. In
contrast, the ZIC-HILIC stationary phase has an inner positive quaternary ammo-
nium ion and an outer negative sulfonate ion separated by a three-methylene group,
which has been reported for the separation of basic analytes such as hydrazines,29
mildronates,17 and acetylcholine.32 In addition, some degree of electrostatic interac-
tion was found between ionizable analytes and the residual silanol groups under non-
ionic (neutral) polar stationary phase surfaces such as diol,33 sorbitol methacrylate,34
and amide-silica.35 These initial studies are promising, but more studies are required
to better understand the role of electrostatic interaction in HILIC separation.
11.2.2 Effect of Buffer Concentration (Ionic Strength)

The buffer is a very important component when considering ionic interaction. The buf-
fer can be used to control mobile phase pH, and pH can modify the ionization status
not only for analytes but for stationary phases as well. Optimizing buffer concentra-
tion can also generate the ion strength necessary to mediate the electrostatic inter-
action between analytes and stationary phase.36 In addition, buffer composition can
affect analyte retention or transition enthalpy between mobile and stationary phases.
Retention enthalpy is an important parameter if column temperature needs to be con-
sidered for improved analyte separation, which will be discussed in next section.
In general, at a low buffer concentration or a buffer-free condition, ionic analytes
are effectively retained on stationary phases containing counter ionic groups. As the
buffer concentration increases, a high level of organic composition in the mobile
phase could force the buffer ions inside the stagnant water-rich liquid layer. Higher
buffer ion concentration would drive more solvated salt ions into this liquid layer
and result in an increase in volume or hydrophilicity of the liquid layer. The strength
of the electrostatic interaction between analytes and the stationary phase would be
weakened with an increase of the water-rich layer volume, thus resulting in a weak

electrostatic interaction. If the electrostatic interaction is with counter ions, the elec-
trostatic attraction would be weakened, resulting in a longer retention times. If the
charge–charge interaction is with co-ions, the electrostatic repulsion would also be
weakened, resulting in shorter retention times. Most experimental data, especially
with ammonium acetate or formate buffer, support this rationale.29,37–39 For example,
room temperature ionic liquid (RTIL) imidazolium cations were not eluted (a very
strong capacity factor) on diol stationary phase with a salt-free mobile phase due to
their strong electrostatic interactions with the negative residual silanols under the
diol stationary phase surface.40 A decrease in capacity factor was observed when the
ammonium acetate concentration was raised from 5 to 20â•›mmol/L (shown in Figure
11.5). The negative residual silanols on a diol phase column was also reported for a
uric acid separation.41 The negative silanol groups exert electrostatic repulsion on the
negatively charged acids under the experimental conditions. Increasing the concen-
tration of salt in the mobile phase would reduce this electrostatic repulsion leading
to stronger retention.
The retention of anionic nicotinic and picolinic acids on an amine-silica column
were reduced by increasing the buffer concentration, as shown in Figure 11.6.15 In
another example, the retention time of acidic analytes (aspirin and salicylic acid)
increased by about 20%–40% on amide-silica, HILIC silica, and ZIC-HILIC col-
umns but decreased sharply on the amine-silica column when the ammonium acetate
concentration was increased from 5 to 20â•›mM (shown in Table 11.2).36 The buffer
concentration can contribute to both electrostatic repulsion and attraction between
analytes and stationary phases. The ionized residual silanol groups on amide-silica
and silica columns can repulse the ionized carboxyl groups on analytes under pH 6.9.
1.80
1.60
1.40 5
4
1.20 3
1.00
ln k΄
0.80
0.60
0.40
0.20
0.00
0 5 10 15 20 25
Ammonium acetate concentration (mM)
Figure 11.5 Plots of ln k′ vs. ammonium acetate concentration in the mobile phase for three
RTILs: (3) 1-butyl-2-methyl-3-methyl imidazolium bis-(trifluoromethylsulfonyl)-imide, (4)
1-hexyl-3-methyl imidazolium chloride, and (5) 1-methyl-3-octyl imidazolium chloride. Stationary
phase: Uptisphere OH; mobile phase: ACN/H2O with 10â•›mM ammonium acetate at 0.2â•›mL/min.
(Adapted from Rouzo, G.L. et al., J. Chromatogr. A, 1164, 139, 2007. With permission.)

7.0
6.0
5.0
k΄
4.0
3.0
2.0
50 90 130 170 210
Ammonium hydroxide concentration (mM)
Figure 11.6 Effect of buffer concentration on the retention of nicotinic acid (▪) and picolinic
acid (○). (Adapted from Christopherson, M.J. et al., J. Liq. Chromatogr. Relat. Technol.,
29, 2545, 2006. With permission.)
Table 11.2
Retention Time of the Model Compounds at Different Ammonium
Acetate Concentrations in the Mobile Phasea
Column Concentration (mM) Salicylic Acid Aspirin Cytosine
TSK-Gel Amide-80 5 2.07 3.06 6.84
10 2.39 3.65 7.19
20 2.61 4.14 8.01
YMC-Pack NH2 5 7.59 20.21 6.03
10 4.72 11.50 6.10
20 3.56 7.17 6.45
HILIC Silica 5 1.78 2.94 5.51
10 2.06 3.51 5.78
20 2.49 4.21 6.62
ZIC-HILIC 5 2.16 2.78 5.52
10 2.44 3.22 5.59
20 2.64 3.55 5.98
Source: Adapted from Guo, Y. and Gaiki, S., J. Chromatogr. A, 1074, 71, 2005. With
a Mobile phase: acetonitrile/ammonium acetate solution (85/15, v/v). Column tempera-
ture: 30°C. Flow rate: 1.5â•›mL/min.
An increase in buffer ions would eliminate such repulsive interaction, resulting in a

longer retention time. The ZIC-HILIC stationary phase has an inner positive qua-
ternary ammonium ion and an outer negative sulfonate ion separated by a three-
methylene group.29 The ascorbic acid (vitamin C) retention time on this column became
longer when the ammonium acetate concentration increased from 10 to 100â•›mM.42

The electrostatic repulsion between the analyte carboxyl group and sulfonate group
on the outer layer of the stationary phase seems to be predominant. Analyte size may
block or reduce the electrostatic attraction between analyte carboxyl ions and qua-
ternary ammonium ions on the inner layer of the stationary phase. The ZIC-HILIC
column has been reported for the separation of inorganic anions by ion chromatog-
raphy.43 The retention behavior of four hydrazine analogues in different ammonium
formate concentrations is shown in Figure 11.7. Since the HCl salts of hydrazine and
1,2-dimethylhydrazine were used and RI is a universal detector, the chloride peak
was also observed. The retention of positively charged hydrazines decreased with
increasing ionic strength, whereas the retention behavior of negatively charged chlo-
ride was just the opposite (Figure 11.7a).
In conventional ion-exchange chromatography (IEC) with water media, the rela-
tionship of the retention factor k′ of an analyte and the buffer concentration [C] is as
described below44,45
log k ′ = − s · log [C ] + constant

where s is a constant slope, which is dependent on the overall charge of the ana-
lytes and counter-ions. For a singly charged analyte and univalent counterion, the
slope should be −1. Ammonium formate concentrations in the range of 5–30â•›m M
were used for drawing the plots of log k′ vs. log [C] for hydrazines on a ZIC-HILIC
column. The linear relationships observed for all hydrazines and for chloride (coef-
ficient of determination r 2 ≥ 0.97) are shown in Figure 11.7b. All hydrazines in this
Cl 1 3
2 4 5 mM
Cl 1 3 4 10 mM
10,000 2
1 Cl 2 3 4 20 mM
8,000 1 Cl 30 mM
2 3 4
1 2/Cl
nRIU
3 4 40 mM
6,000 1
2 Cl 3 4 50 mM
4,000 1 Cl
2 3 4 70 mM
1 2 Cl/3 4 90 mM
2,000
5 10 15 20 25 30 35 40
(a) min
Figure 11.7 (a) The effect of ionic strength on the retention of hydrazines on a ZIC-HILIC
column. Column temperature was set at 30°C. Isocratic runs with a mobile phase of 5–90â•›m M
ammonium formate buffer pH 3.0/ethyl alcohol (20/80, v/v) (the buffer concentrations refer
to the concentration before mixing with organic solvent). Flow rate was 0.4â•›m L/min with
RI detection. The analyte concentrations were about 0.8–1.2â•›mg/mL in water/ethyl alcohol
(20/80, v/v). The injection volume was 1â•›μL. 0.5% toluene (v/v) in ethyl alcohol was used as a
void volume marker. Chloride is from the HCl salts of hydrazine and 1,2-dimethylhydrazine.
1: 1,2-dimethylhydrazine, 2: 1,1-dimethylhydrazine, 3: methylhydrazine, and 4: hydrazine.

0.7
0.6
0.5
0.4
R2 = 1.000
Log k΄
0.3
0.2 R2 = 1.000
0.1 R2 = 1.000
R2 = 0.964
0
R2 = 1.000
–0.1
0.6 0.8 1.0 1.2 1.4
(b) Log [Ammonium formate]
R2 = 0.973
5
4
R2 = 0.975
3
k΄
R2 = 0.976
2 R2 = 0.978
1
R2 = 0.946
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
(c) 1/[Ammonium formate]
Figure 11.7 (continued) (b) Plot of logarithm of retention factor (k′) against logarithm of buf-
fer concentration (5–30â•›mM). (c) Plot of k′ against the inversed buffer concentration (5–30â•›mM).
Hydrazine (⦁), methylhydrazine (▴), 1,1-dimethylhydrazine (◼), 1,2-dimethylhydrazine (⬥), and
chloride (⚬). (Adapted from Liu, M. et al., J. Chromatogr. A, 1216, 2362, 2009. With permission.)
study have negative slopes while chloride has a positive slope, which are indicative of
the net electrostatic attraction and repulsion with the stationary phase, respectively.
The absolute slope values in Figure 11.7b (∼0.3 for hydrazines and 0.1 for chloride)
indicated other interactions, and mobile phase composition could also contribute
to analyte retention. Positive intercepts were observed for all components, which
provide evidence for the existence of additional retention mechanisms at “infinite”
buffer concentration. Cumulatively, data in Figure 11.7 indicate that both ionic inter-
action and hydrophilic interaction were involved in the separation of hydrazine ana-
logues and their counterions under an aqueous organic mobile phase condition.
An exceptional example was found with the positive analytes (metformin hydrochlo-
ride [MFH], cyanoguanidine [CGD], and melamine [MLN]) separated on an Atlantis
HILIC-Si column.46 Ammonium acetate and ammonium formate buffers proved to

14
CGD
12 MLN
MFH
10
0
10 20 30 40
Phosphate buffer (mM)
Figure 11.8 Effect of sodium buffer strength on the retention of MFH, CGD, and MLN
on a 5â•›μm, 250 × 4.6 Atlantis HILIC-Si column. Acetonitrile/buffer (84/16, v/v) with a pH of 3.
Flow rate: 2.0â•›m L/min. (Adapted from Ali, M.S. et al., Chromatographia, 67, 517, 2008. With
be unsuitable for UV detection at 218â•›nm. The retention of metformin increased when

the concentration of sodium phosphate increased from 10 to 40â•›mM (Figure 11.8). It is
not entirely clear whether sodium phosphate buffer has an impact upon ionic analyte
separation in HILIC since the use of this buffer is limited with HILIC method.
Determining how buffer type modifies electrostatic interaction between analytes
and stationary phases can be complicated. When ammonium acetate was replaced
with ammonium formate in Drs. Guo and Gaiki study described previously,36 no
significant change was found for the neutral analyte, cytosine. Little retention change
was observed on amide-silica, HILIC silica, and ZIC-HILIC columns for acidic ana-
lytes (aspirin and salicylic acid). A significant retention increase was obtained on
an amine-silica column for acidic analytes, which might be related to formic acid
size since these smaller sized ions might penetrate through the outer amine layer to
suppress silanol ionization in the inner layer of stationary phase. Surprisingly, when
ammonium acetate was replaced with ammonium bicarbonate, the latter caused
drastic decreases in retention on all four HILIC columns and destroyed the separa-
tion of the acids.
The study of buffer species is limited because some buffers that are typically
used in reverse phase HPLC may not be suitable for HILIC due to their poor solu-
bility in mobile phase containing a high level of ACN.36 In addition, the use of buf-
fers is Â�relatively limited when HILIC separation is coupled with mass spectrometry
detection. Most nonvolatile buffers produce a very strong ionic suppression on MS
detection signals. Therefore, the most common buffer solutions used in HILIC-MS

methods are ammonium acetate or formate. Practical HILIC applications coupled

with MS detection only use organic acid to control the mobile phase pH because
most HILIC stationary phases are stable under an acidic condition.
11.2.3 Effect of Column Temperature

Column temperature has long been recognized as an important parameter in HPLC sep-
aration, especially after column and mobile phase selections have been made. Column
temperature has a significant impact on analyte diffusivity, mobile phase viscosity, and
analyte transferring enthalpy between mobile and stationary phases. In general, tem-
perature elevation increases the diffusion coefficient and decreases the mobile phase
viscosity, resulting in narrower peaks. But its drawback is shorter retention times. The
total resolution could not be improved much under such a scenario. Only if increased
column temperature would simultaneously retain various analytes longer on the column,
then retention differences between individual analytes with the narrowed peaks would be
enlarged and overall resolution could be improved. To achieve such a scenario, a positive
transferring enthalpy (ΔH value), or the negative slope in the van’t Hoff plot, needs to be
present. Small ionic analytes combined with a counterionic stationary phase in HILIC
has provided such an opportunity. When ionic functional groups are present in small
polar analytes, column temperature should therefore be considered in HILIC separations.
For example, when anionic aspirin (acetylsalicylic acid) was separated on four
different HILIC columns, YMC-pack NH 2, TSK-Gel Amide-80, Atlantis HILIC sil-
ica (type-B), and SeQuant ZIC, only the positive amino stationary phase (YMC-pack
NH2) obtained negative slopes or positive ΔH values (Figure 11.9).36 The positive
retention enthalpy indicated an endothermic process of transferring analytes from
the mobile phase to the stationary phase. Guo et al. further investigated six organic
acids, acetylsalicylic acid, salicylic acid, gentisic acid, hippuric acid, salicyluric
acid, and α-hydroxyhippuric acid, on the five columns (Figure 11.10b), and found
2.5
2
ln k΄
1.5
0.5
0
0.0028 0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035
1/T (K)
Figure 11.9 The van’t Hoff plots for aspirin on (♦) TSK-Gel Amide-80, (◾) YMC-Pack
NH2, (▴) HILIC Silica, and (×) ZIC HILIC columns. Mobile phase: ACN/water (90:10, v/v)
containing 10â•›mM ammonium acetate. (Adapted from Guo, Y. and Gaiki, S., J. Chromatogr. A,
1074, 71, 2005. With permission.)

ACN content (%) Salt concentration (mM)

65 70 75 80 85 40 30 20 10
25 25
(min)

20 20
Retention time
15 15
10 10
5 5
40 85
Sal
t c 30 80 )
on
cen 75 t (%
tra 20 nten
tio 70 co
n( N
mM) 10 65 AC
(a)
Salt concentration (mM) Column temperature (°C)

10 20 30 40 60 50 40 30 20 10
16 16
in)
(min)
14 14
Retention time (m
Retention time
12 12
10 10
8 8
6 6
60 40
Co 50 30 M)
lum 40 (m
nt n
em 30 20 tio
per tra
atu 20 cen
re ( 10
t con
(b)
°C) 10 Sal
Figure 11.10 3D response surfaces for the amino phase (YMC-pack NH2 column) gener-
ated by DOE software. (a) Influence of ACN and ammonium acetate concentration on acidic
analyte retention. (b) Influence of ammonium acetate concentration and column temperature
on acidic analyte retention. (Adapted from Guo, Y. et al., Chromatographia, 66, 223, 2007.
With permission.)

1.4
1.2 y = –887.79x + 4.0107

r 2 = 0.9739
ln k΄
0.8
y = –919.26x + 3.9214
r 2 = 0.9922
0.6
0.003 0.0031 0.0032 0.0033 0.0034 0.0035
1/T
Figure 11.11 The van’t Hoff curves for nicotinic acid (◾) and picolinic acid (○) on a
Phenomenex Luna NH2 column using 70:30 (acetonitrile:buffer). (Adapted from ChristoÂ�
pherson, M.J. et al., J. Liq. Chromatogr. Relat. Technol., 29, 2545, 2006. With permission.)
the amino phase had very different response surfaces. The retention time increased
when the column temperature was elevated from 10°C to 60°C. Similarly, when pic-
olinic and nicotinic acids were separated using the positive Phenomenex Luna NH2
stationary phase shown in Figure 11.1, negative slopes from both anionic analytes
were obtained in the van’t Hoff plots (shown in Figure 11.11).
In contrast, when cationic analytes are separated with an anionic stationary phase,
the analytes which transfer enthalpy might also exhibit a better separation. The cat-
ionic analytes, G, DG, TG, GG, GDG, and GTG (their structures are in Figure 11.3),
were separated by four silica type-A columns (Figure 11.12). Their retention times
increased when the column temperature was elevated no matter if methanol or ACN
was used in the mobile phase.
When these six analytes were separated on bare silica type-B and type-C col-
umns, much weaker retentions were observed for all six analytes (Figure 11.13). More
importantly, the positive slopes or negative transferring enthalpies were obtained in
van’t Hoff plots when a column temperature program from 5°C to 80°C was applied
for analyte separation. A possible rationale is higher average pKa values attributed to
silanol groups on type-B and type-C silica material surfaces.
However, negative slopes were reported for the separation of the basic analogue epiru-
bincin on a Kromasil KR100-5SIL bare silica column (Figure 11.14).47 More importantly,
the column plate number (N) was increased from 32 to 55â•›K plates/m by elevating the
column temperature from 25°C to 40°C (about 70% improvement in column efficiency).
For comparison, we also used this type of column to test the impact of column
temperature on the separation of six analytes in our laboratory. In contrast, slopes
for all six analytes were positive in the van’t Hoff plots (Figure 11.15). Buffer solu-
tion was not used in our experiments. It is not clear whether the different outcomes
are the result of a difference in analyte characteristics or the sodium formate buffer
used in epirubicin analogue separation. More studies with Kromasil KR100-5SIL
are needed for a more definitive conclusion.

40.00
GTG
35.00 GDG
R2 = 0.984
GG
R2 = 0.992 TG
30.00 R2 = 0.981 DG
R2 = 0.995 G
25.00
Log k΄
20.00
R2 = 0.993
15.00
R2 = 0.980
10.00
5.00
0.00
0.00270 0.00280 0.00290 0.00300 0.00310 0.00320 0.00330 0.00340 0.00350 0.00360
(a) 1/T (K–1)
18.00
GTG
16.00 R2 = 0.981 GDG
GG
14.00 TG
R2 = 0.985
DG
G
12.00
R2 = 0.987
10.00 R2 = 0.994
Log k΄
8.00
R2 = 0.979
6.00
4.00 R2 = 0.892
2.00
0.00
0.00270 0.00280 0.00290 0.00300 0.00310 0.00320 0.00330 0.00340 0.00350 0.00360
(b) 1/T (K–1)
Figure 11.12 The van’t Hoff plots for G, DG, TG, GG, GDG, and GTG under HILIC
condition of Hypersil silica column, 100 × 1.0â•›m m, particle size of 3â•›μm with a flow rate of
100â•›μL/min, the column temperature varying from 5°C to 80°C, mobile phase for top: water/
ACN (25:75, v/v) containing 0.4% formic acid and mobile phase for bottom: water/MeOH
(2:98, v/v) containing 0.4% formic acid. (Adapted from Hao, Z. et al., J. Sep. Sci., 31, 1449,
2008. With permission from Wiley-VCH Verlag Gmbh & Co. KGaA.)

25.00
GTG
GDG
20.00 GG
TG
DG
G
15.00
Log k΄
10.00
5.00
0.00
0.00270 0.00280 0.00290 0.00300 0.00310 0.00320 0.00330 0.00340 0.00350 0.00360
1/T (K–1)
Figure 11.13 The van’t Hoff plots for six analytes, G, DG, TG, GG, GDG, and GTG
under HILIC condition of Atlantis silica column, 50 × 2.1â•›m m, particle size of 5â•›μm with
a flow rate of 100â•›μL/min, column temperature varying from 5°C to 80°C, mobile phase of
water/ACN (10:90, v/v) containing 0.4% formic acid. (Adapted Hao, Z. et al., J. Sep. Sci., 31,
1449, 2008. With permission from Wiley-VCH Verlag Gmbh & Co. KGaA.)
2.3 Epidaunorubicin
Daunorubicin
2.2 Epirubicin
Doxorubicin
2.1
2.0
ln k΄
1.9
1.8
1.7
1.6
3.22 3.24 3.26 3.28 3.30 3.32 3.34 3.36
103/T (K–1)
Figure 11.14 Effect of temperature upon retention with acetonitrile as modifier.

Conditions: Kromasil KR100-5SIL (5â•›μm); mobile phase: 90% (v/v) ACN in sodium for-
mate buffer (30â•›m M, pH 2.9) and UV detection at 254â•›nm for epirubicin and its analogues;
flow rate: 1.0â•›m L/min; injection volume: 20â•›μL. (Adapted from Dong, L. and Huang, J.,
Chromatographia, 65, 519, 2007. With permission.)

24.00 GTG
GDG
GG
TG
22.00 DG
G
20.00
Log k΄
18.00
16.00
14.00
12.00
0.00270 0.00280 0.00290 0.00300 0.00310 0.00320 0.00330 0.00340 0.00350 0.00360
1/T (K–1)
Figure 11.15 The van’t Hoff plots for six analytes, G, DG, TG, GG, GDG, and GTG, under
HILIC condition of Kromasil KR100-5SIL silica column, 100 × 2.0â•›mm, particle size of 5â•›μm
with a flow rate of 100â•›μL/min, the column temperature varying from 5°C to 80°C, mobile
phase of water/acetonitrile (10:90, v/v) containing 0.4% formic acid. (Adapted Hao, Z. et al.,
J. Sep. Sci., 31, 1449, 2008. With permission from Wiley-VCH Verlag Gmbh & Co. KGaA.)
11.3 Nonionic Polar Functional Groups

in HILIC Separation
The nonionic polar functional group has a permanent dipole moment but lacks
a complete electric charge. The permanent dipole occurs when two atoms in a
functional group have substantially different electronegativity, one atom attracts
an electron more than the other becoming more negative, while the other atom
becomes more positive. The major and strongest interaction between nonionic
polar functional groups is hydrogen bonding. The hydrogen bond can also provide
powerful retaining power in HILIC, even though it is weaker than ionic interac-
tion. Hydrogen-bond energy is within 5–30â•›k J/mol while ionic bond energy is
usually larger than 100â•›k J/mol. Mobile phase composition and column temperature
can dramatically impact upon hydrogen bonding between analytes and the station-
ary phase.
11.3.1 Effect of Mobile Phase Composition

In HILIC, the secondary hydrogen-bond retention mechanism can be distinguished
from the primary partition retention mechanism. Hydrogen-bond strength can be

explained by the neutral polar functional groups, including protic or aprotic. The
polar protic groups can be both donors and acceptors of hydrogen bonds whereas
aprotic solvents can be only hydrogen-bond acceptors. Partition is usually deter-
mined by the entire analyte polarity or hydrophilicity, represented by the log P value,
which is the logarithm of the octanol–water partition coefficient. Log P values are
unavailable and an extra independent experiment is needed to determine them. For
practical HILIC method development, the hydrogen-bond retention mechanism is
more suitable to predict the retention capability of neutral functional groups within
analytes and stationary phase. A separation study of neutral oligomeric proanthocy-
anidins (structures in Figure 11.16) on an amide-silica column (TSK-Gel Amide-80)
is a good example showing the contributions of hydrogen bonding and partitioning
in a HILIC separation.48 The correlation between the logarithm of retention fac-
tors (log k) and the number of hydroxyl groups in Figure 11.17 is even better than
between log k and log P values in Figure 11.18 (r 2 = 0.9501 vs. 0.7949, respectively).48
When hydrogen bonding is strongly involved in HILIC retention, switch-
ing organic components in the mobile phase can be a very important strategy to
improve analyte separation. These organic components can be subdivided into polar
protic and aprotic solvents. Methanol, ethanol, isopropanol, and acetic acid are rep-
resentative polar protic solvents. The typical polar aprotic solvents are ACN and
tetrahydrofuran (THF). Because of their strong ability to hydrogen bond, polar protic
solvents can more effectively compete for polar active sites on the HILIC phase sur-
face, perturbing the formation of water layers by replacing water molecules, thus
R1
OH
OH 1a H : (+)-catechin C
8
HO O 2
R1 1b OH : (+)-gallocatechin GC
3
6
4 OH
OH R1 R2
OH 1c H H : (–)-epicatechin EC
OH
8
1d OH H : (–)-epigallocatechin
HO O 2
R1 1e H galloyl : (–)-epicatechin gallate
3
6 O
4 R2 1f OH galloyl : (–)-epigallocatechin gallate
OH
2a EC C 3a EC EC EC
(4β 8) 2b EC EC 3b EC EC EC
(4β 6) 2c C C 3c EC EC EC
(4α 8) 2d GC EC 4a EC EC EC EC
5a EC EC EC EC EC
Figure 11.16 Chemical structures of monomeric flavan-3-ols and oligomeric proanthocy-

anidins. (Adapted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153, 2007. With permission.)

30
5a
25 y = 12.341x + 9.2541
r 2 = 0.9501
Number of OH groups
4a
20
3b3c
15 3a
10 2a 2d
1f
1e 2c2b
1a1c
5 1b1d
0
–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Log k
Figure 11.17 Relationship between log k′ of 15 standards in amide HILIC column

(TSK-Gel Amide-80) and the number of hydroxyl groups (–OH) in their structures. Inset
regression formula was calculated from the linear least-square fit of all data (n = 15).
(Adapted from Yanagida, A. et al., J. Chromatogr. A, 1143, 153, 2007. With permission.)
1 1e y = –1.8177x – 0.4767
r 2 = 0.7949
1a 1f
0 1c
1b
Log Po/w
1d
–1 3c
2c 2b 3a
2a
4a
2d 3b
–2 5a
–3
–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Log k
Figure 11.18 Relationship between logâ•›k ′ of 15 standards in amide HILIC column

(TSK-Gel Amide-80) and log P values measured by HSCCC. Inset regression formula was
calculated from the linear least-square fit of all data (n = 15). (Adapted from Yanagida,
A. et al., J. Chromatogr. A, 1143, 153, 2007. With permission.)
producing a more hydrophobic stationary phase.49 As a consequence, analytes with

strong hydrogen-bond capability are poorly retained.
Li and Huang compared the ability of various organic modifiers to retain epiru-
bicin analogues using a Kromasil KR100-5SIL column (Figure 11.19).50 Methanol
caused all four analytes to be eluted with no retention. Isopropanol has a longer
alkyl chain and less hydrophilic character. It competes less strongly for the active

140
1+3+2+4
1+3
120 2+4
100
Absorbance (mAU)
Methanol
80 1
3
24
60 Isopropanol
40
1
THF 3
2 4
20
Acetonitrile
0
0 5 10 15 20 25 30
Time (min)
Figure 11.19 Effect of organic modifier on separation of epirubicin and its analogues.
Conditions: Kromasil KR100-5SIL (5â•›μm); mobile phase: sodium formate buffer (20â•›m M,
pH 2.9) modified with various organic solvents (10:90, v/v). Peaks: (1) epidaunorubicin, (2)
daunorubicin, (3) epirubicin, (4) doxorubicin. (Adapted from Li, R.P. and Huang, J.X.,
J. Chromatogr. A, 1041, 163, 2004. With permission.)
sites and left more time for analytes to be retained. In contrast to methanol and
isopropanol, the aprotic solvents ACN and THF provided more effective retention
of the analytes on the column. The analytes were retained more strongly with ACN
than THF because the latter is a better hydrogen-bond acceptor. Interestingly, the
elution order of epirubicin and daunorubicin with ACN is different from THF and
isopropanol (Figures 11.19 and 11.20). The major differences between these two
structures are an extra hydroxyl group by the keto group in epirubicin and an inner
hydrogen bond between the hydroxyl group and amine group in daunorubicin.
Epirubicin was retained longer in the mobile phase containing ACN, where stron-
ger hydrogen bonding occurs between analyte and the stationary phase. When
ACN was replaced by THF or isopropanol, the retention contribution from such
hydrogen bonding became weaker and ion-exchange interactions between the
analyte and the stationary phase became stronger. Daunorubicin, with a higher
pKa due to such inner hydrogen bonding, was retained longer. The four hydrazine
analogues mentioned earlier were successfully separated by an aqueous ethanol
mobile phase on a ZIC-HILIC column because ACN cannot be used with a chemi-
luminescent nitrogen detector (CLND).36
A systematic comparison of selectivity (log α values) between G vs. GG, DG
vs. GDG, and TG vs. GTG was investigated on a Hypersil silica column with

O OH CH3 O OH CH2OH
O O
OH OH
OH O OH OH O OH
O H
H O O
O
O O
H CH3 H CH3
N N
H H
(2) (3)
Figure 11.20 Chemical structures of daunorubicin (2) and epirubicin (3).
different organic solvents used in the mobile phase (Figure 11.21).14 When the
methanol content was low and log α < 0, G, DG, and TG eluted before GG, GDG,
and GTG, respectively. When the methanol content was increased and log α = 0,
no separation occurred for each of the individual pairs. When the methanol con-
tent was further increased and log α > 0, G, DG, and TG eluted after GG, GDG,
0.100
0.080
G-GG
DG-GDG
0.060
TG-GTG
0.040
0.020
Log α
0.000
0 5 10 15 20 25 30 35 40 45
–0.020
–0.040
–0.060
–0.080
–0.100
%MeOH relative to ACN in mobile phase
Figure 11.21 The plots of log α (selectivity) vs. %methanol relative to acetonitrile in the
mobile phase for three pairs of compounds, G vs. GG, DG vs. GDG, and TG vs. GTG, under
HILIC condition on a Hypersil silica column, 100 × 1â•›m m, particle size of 3â•›μm, 0.1% formic
acid in all mobile phase solvents used, and column temperature at 30°C. Total organic content
(acetonitrile + methanol) in the mobile phase remained at 75%. (Adapted from Hao, Z. et al.,

and GTG, respectively. The active silanol groups on the stationary phase surface
can effectively be competed with either the hydroxyl groups from methanol or the
neutral polar hydroxyl groups from Amadori compounds. The more active sides
are occupied by methanol, the less are left for Amadori compounds, thus result-
ing in a reversed elution order. A similar phenomenon was found in an analysis of
morphine and its metabolite, morphine-3-glucuronide, and their elution order was
switched when ACN was replaced by methanol in the mobile phase on the Inertsil
silica and ZIC-HILIC columns.39,51
Switching a protic organic solvent with aprotic can improve separation selectivity
not only for similar but also for completely different structures. For example, 50%
ACN-H2O containing 0.1% formic acid as mobile phase was used to separate choline
and arginine on a Hypersil bare silica column (Figure 11.22a). If the aprotic organic
solvent ACN was replaced by the protic organic solvent methanol, the elution order
in Figure 11.22a was switched to those shown in Figure 11.22b.8 The rationale behind
this switching is hydrogen-bond competition. Abundant protic functional groups like
RT: 0.00–29.99
5.43
100 5.36
5.56
80
5.70
60
5.84
40
6.01 16.96 17.10
20 6.15 17.68
6.45 6.90 8.13 16.72 18.26
0.13 1.29 3.39 5.22 10.2812.33 13.26 16.58 19.16 21.4
0 5.43
100
Relative abundance
80
Choline
60
40
20
0.93 2.10 3.40 4.40 7.11 7.65 8.99 11.19 13.65 15.47 16.84 17.56 20.89 22
0 16.96 17.10
100
80 Arginine
60
40
20 13.26 18.85
0.13 1.29 3.39 4.52 5.24 6.92 8.94 10.28 11.58 13.70 21.45
0
0 2 4 6 8 10 12 14 16 18 20 22
(a) Time (min)
Figure 11.22 HILIC-MS/MS chromatograms of choline and arginine under HILIC con-
dition on a Hypersil silica column, 100 × 1â•›m m, particle size of 3â•›μm. (a) The mobile phase is
consistent of 50% water and 50% ACN and 0.4% formic acid in both solvents.
(continued)

RT: 0.00–29.99
11.66
100
11.90
80 11.42 11.94
12.08
60
12.25
11.29
40 12.39
12.52
12.69
20 9.69 9.80 13.00
1.14 1.76 3.75 4.60 6.32 7.00 9.39 13.86 14.6816.1618.21 20.75 21
0
11.66
100
Relative abundance
80 11.90
Choline
60
40
20
13.3114.20 20.75
1.14 1.76 3.75 4.60 6.32 7.00 8.34 9.30 15.57 17.25 21
0
100 9.80
80 Arginine
10.11
60 10.17
40
10.62
20 1.19 1.33 10.89
2.25 4.89 6.03 7.98 11.4812.78 15.38 17.1018.47 20.05 22.4
0
0 2 4 6 8 10 12 14 16 18 20 22
(b) Time (min)
Figure 11.22 (continued) (b) The mobile phase is consistent of 30% water and 70%
methanol and 0.4% formic acid in both solvents. (Adapted from Hao, Z. et al., J. Sep. Sci., 31,
−NH(1–2) and −OH in the arginine structure can form hydrogen bonds with silanol
groups on the silica surface or hydroxyl groups from the immobilized water-rich
layer on the silica surface, contributing to arginine retention on the stationary phase.
However, this hydrogen bond could be eliminated by a protic solvent such as metha-
nol, thus resulting in a decrease in retention of arginine. The choline polarity may be
higher or lower than arginine but its retention was less affected by ACN replacement
with methanol compared to arginine.
Methanol was also preferred for the separation of taurine and methionine in a
beverage matrix relative to ACN. Many carbohydrates such as glucose, fructose, and
saccharose present in the beverage solution could be strongly retained on a HILIC
column through a hydrogen-bonding retention mechanism if an ACN–water mobile
phase was used, whereas with a methanol–water mobile phase they are not retained
(Figure 11.23).52
Methanol has not always been successfully used to replace ACN for selecti-
vity improvement. In general, ACN is often selected over methanol since it has
many advantages including nearly ideal spectroscopic qualities, low viscosity, and

Met
b
Tau
0 5 10 15
Time (min)
Figure 11.23 HILIC-ELSD analysis of a beverage diluted to 1/10 with mobile phase.
(a) Beverage containing Met and Tau, (b) amino acid-free beverage. Column: apHera NH2
(150â•›m m × 4.6â•›m m I.D., 5â•›μm). Column temperature is at 37°C. Mobile phase: MeOH/H2O
(60:40, v/v) under isocratic elution mode; flow rate 0.6â•›m L/min; injection volume: 10â•›μL.
(Adapted from de Person, M. et al., J. Chromatogr. A, 1081, 174, 2005. With permission.)
unique chromatographic properties. Acetonitrile is basically a polar solvent and it

is miscible with water in all proportions. More importantly, ACN does not asso-
ciate strongly with water and ACN–water mixtures remain binary in character.
This simplifies the interactive theory and allows a more simple prediction of
retention based on the ACN concentration in the partition retention mechanism.
Methanol, on the other hand, not only interacts with analytes and the stationary
phase but also forms a strong association with water so that at a high concentra-
tion of water, the mobile phase behaves as a binary mixture of water and water–
methanol associate. At high concentrations of methanol, the converse applies,
that is, the mobile phase consists of a mixture of methanol and water–methanol
associate. Between these extremes the mobile phase consists of a complex ternary
mixture of methanol, water–methanol associate, and water. A review by Scott
showed that there was some association between water with ACN and water–THF,
but not nearly to the same extent as water with methanol.53 At the point of maxi-
mum association in water/methanol mixtures, the solvent contained nearly 60%
of the water–methanol associate. In contrast, the maximum amount of water–
THF associate formed was only 17%, and that for water–ACN as little as 8%.
That is why methanol was considered to be too strong an elution solvent, leading
to poor retention, 54 and ACN usually provided sharper peaks than methanol.55

The hydrogen-bonding interaction between methanol and analytes may introduce

extra resonance structures and cause broad or tailing peaks.56,57
It is worthy to note that hydrogen bonding can become unstable at elevated tem-
peratures.58 The broad and tailing peaks with a methanol mobile phase could be
improved if the column is running at an elevated temperature, which will be dis-
cussed in the next section. The effects of mobile phase composition and temperature
are often complementary and a simultaneous optimization of these parameters could
become a useful approach to control analyte retention and improve selectivity and
peak shape.59 One should be aware, however, that polar analytes are poorly soluble
in mobile phases with high concentrations of ACN.37
11.3.2 Effect of Column Temperature

Column temperature can change not only the analyte transferring enthalpy from
mobile phase to stationary phase but can also have an effect on analyte structure.
For example, when crystalline glucose, which is a single compound, is dissolved in
an aqueous solution, tautomerization occurs. Ultimately an equilibrium mixture of
at least five compounds is formed: the α-pyranose, the β-pyranose, the α-furanose,
the β-furanose, and the aldehyde form.60 The transformation rate between different
isomers is temperature dependent. The isomers α- and β-glucopyranoses have been
separated at lower temperature on Ca2+-form Aminex HPX-87C column.61 Partially
resolved double peaks were also observed for l-fucose on a carbamoyl-silica HILIC
column (TSK-Gel Amide-80).35 Elevated temperature can preclude the existence of
specific isomers of carbohydrate analytes. When isomers of the same analyte cannot
be distinguished at higher temperatures, broad or split peaks will be narrowed down
into a single peak.35,62
Amadori compounds can also exist as tautomers due to the carbohydrate por-
tion of their structures. The different α-, β-, and acyclic anomers of Amadori
compounds in Figure 11.24 have been confirmed by nuclear magnetic resonance.63
OH
O O
OH N COOH OH N
H COOH
H
OH
OH OH
(a) (b)
OH O
H
HO N COOH
OH
(c)
Figure 11.24 Chemical structures of open-chain and cyclic forms of N-(1-deoxy-d-

xylose-1-yl)-glycine: (a), α-anomer; (b), β-anomer; (c), acyclic form. (Adapted from Davidek, T.
et al., Anal. Chem., 77, 140, 2005. With permission.)

Within the six structures of G, DG, TG, GG, GDG, and GTG seen in Figure 11.3,
only three Amadori compounds contain a carbohydrate (glucose) portion. At a low
temperature of 5°C, broad/split peaks were observed when these compounds were
separated on bare silica columns and became narrower single peaks at an elevated
column temperature (Figure 11.25).14
Elevated column temperatures can narrow down the peak widths of analytes con-
taining nonionic polar functional groups, especially for carbohydrate rings, but it
also shortens their retention times. The overall resolution of these types of com-
pounds is still primarily analyte dependent.
RT: 0.00–25.00
18.18
100 7.67
4A, TIC 18.38
7.77 17.87 20.87
50 16.48 21.28
7.47 9.99 9.88 22.72
0.69 2.85
0
9.99
100
G
50
16.67
0.73 2.27 5.36 7.72 11.43 19.45 23.66
0
16.69
100
DG
50
0.75 2.29 6.71 9.59 13.81 19.36 20.49 23.99
0
Relative abundance
19.38
100
TG
50
0.76 2.20 6.01 10.02 11.36 13.93 17.63 20.82 24.01
0
7.67
100
GG
50
0.68 2.94 4.17 9.63 11.58 13.74 16.31 19.09 24.13
0
18.18
100
GDG
50
0.69 2.85 6.66 9.95 11.19 12.42 16.22 19.52 22.09
0
20.87
100
GTG
50
0.71 4.11 5.34 6.68 9.97 15.01 16.24 19.64 22.72
0
0 5 10 15 20 25
(a) Time (min)
Figure 11.25 HILIC-MS/MS chromatograms of G, DG, TG, GG, GDG, and GTG under
HILIC condition on a Hypersil silica column, 100 × 1â•›m m, particle size of 3â•›μm with mobile
phase of 25% water and 75% ACN and 0.4% formic acid in both solvents. Column tem-
perature was at 80°C for (a), 30°C for (b), and 5°C for (c). (Adapted from Hao, Z. et al.,
(continued)

RT: 0.00–26.00
100
5.82 12.47 14.07
4B, TIC 5.92 14.37
16.24 20.03
50 5.61 6.03
7.52 12.16 16.96
0.75 4.40 20.56 22.83
0
7.52
100
G
50
12.45
0.73 3.40 7.00 9.06 14.41 18.32 22.33 23.77
0
12.47
100
DG
50
0.75 3.94 6.92 8.15 9.90 14.53 15.25 18.75 22.45 23.68
0
Relative abundance
14.55
100
TG
50
0.76 4.36 5.70 9.71 12.18 16.09 18.04 21.23 25.6
0
5.82
100
GG
50
2.02 4.38 7.16 8.91 11.17 13.74 16.83 18.78 21.56 22.69
0
14.07
100
GDG
20.03
50
19.52
0.70 4.40 7.59 8.72 12.63 21.78 25.51
0 16.24
100
GTG
50 20.05
0.71 4.41 7.50 9.35 12.33 20.56 24.81
0
0 5 10 15 20 25
(b) Time (min)

RT: 0.00–26.00
9.80
100 4.89 9.69 11.56
4.79 12.21 4C, TIC
50 4.59 9.59
5.97 14.39
0.75 16.24 20.87 22.10
0
5.97
100
G
50
9.78
0.73 2.27 7.21 11.73 14.31 17.49 20.68 22.84
0
9.80
100
DG
50
0.75 3.83 5.89 8.36 11.65 12.57 15.76 18.54 21.62 23.99
0
Relative abundance
11.56
100
TG
50
0.76 2.20 5.29 6.83 10.02 13.00 15.47 19.79 22.05 24.83
0
4.89
100
GG
50
0.78 2.32 6.23 8.39 9.73 12.09 15.80 18.88 21.45 24.02
0
12.01
100
GDG
10.98
50
1.31 2.96 5.94 9.54 13.04 15.61 17.67 21.16 24.9
0
14.08
100
13.47 GTG
50
1.64 3.69 6.78 8.84 15.42 16.76 20.87 24.26
0
0 5 10 15 20 25
(c) Time (min)

Acknowledgments
The author would like to thank Mark Storton and Kate Jackson from our Global
Analytical Science Department for their helpful comments on my manuscript!
References
1. Linden, J.C. and Lawhead, C.L., J. Chromatogr., 1975, 105, 125–133.
2. Palmer, J.K., Anal. Lett., 1975, 8, 215–224.
3. Alpert, A.J., J. Chromatogr., 1990, 499, 177–196.
4. Naidong, W., J. Chromatogr. B, 2003, 796, 209–224.
5. Hemstron, P. and Irgum, K., J. Sep. Sci., 2006, 29, 1784–1821.
6. Jandera, P., J. Sep. Sci., 2008, 31, 1421–1437.
7. Dejaegher, B., Mangelings, D., and Vander Heyden, Y., J. Sep. Sci., 2008, 31, 1438–1448.
8. Hao, Z., Xiao, B., and Weng, N., J. Sep. Sci., 2008, 31, 1449–1464.
9. Nguyen, H.P. and Schug, K.A., J. Sep. Sci., 2008, 31, 1465–1480.
10. Hsieh, Y., J. Sep. Sci., 2008, 31, 1481–1491.
11. FDA. Center for Drug Evaluation and Research, U.S. Food and Drug Administration,
Reviewer Guidance, Validation of Chromatographic Method, FDA, Rockville, MD,
November, 1994.
12. Kovalova, L., McArdell, C.S., and Hollender, J., J. Chromatogr. A, 2009, 1216,
1100–1108.
13. Kadar, E.P. and Wujcik, C.E., J. Chromatogr. B, 2009, 877, 471–476.
14. Hao, Z., Lu, C. Y., Xiao, B., Weng, N., Parker, B., Knapp, M., and Ho, C.T., J. Chromatogr.
A, 2007, 1147, 165–171.
15. Christopherson, M.J., Yoder, K.J., and Hill, J.T., J. Liq. Chromatogr. Relat. Technol.,
2006, 29, 2545–2558.
16. Ikegami, T., Tomomatsu, K., Takubo, H., Horie, K., and Tanaka, N., J. Chromatogr. A,
2008, 1184, 474–503.
17. Hmelnickis, J., Pugovics, O., Kazoka, H., Viksna, A., Susinskis, V., and Kokums, K.,
J. Pharm. Biomed. Anal., 2008, 48, 649–656.
18. Liu, M., Chen, E.X., Ji, R., and Semin, D., J. Chromatogr. A, 2008, 1188, 255–263.
19. Li, W., Li, Y., Francisco, D.T., and Naidong, W., Biomed. Chromatogr., 2005, 19,
385–393.
20. Jeong, D.W., Kim, Y.H., Ji, H.Y., Youn, Y.S., Lee, K.C., and Lee, H.S., J. Pharm. Biomed.
Anal., 2007, 44, 547–552.
21. Ji, H.Y., Jeong, D.W., Kim, Y.H., Kim, H.H., Sohn, D.R., and Lee, H.S., J. Pharm.
Biomed. Anal., 2006, 41, 622–627.
22. Ji, H.Y., Park, E.J., Lee, K.C., and Lee, H.S., J. Sep. Sci., 2008, 31, 1628–1633.
23. Park, E.J., Lee, H.W., Ji, H.Y., Kim, H.Y., Lee, M.H., Park, E.S., Lee, K.C., and Lee, H.S.,
Arch. Pharm. Res., 2008, 31, 1205–1211.
24. Sun, Y., Sun, J., Liu, J., Yin, S., Chen, Y., Zhang, P., Pu, X., Sun, Y., and He, Z., J. Chromatogr.
B, 2009, 877, 649–652.
25. Nawrocki, J., J. Chromatogr. A, 1997, 779, 29–71.
26. Ali, M.S., Rafiuddin, S., Al-Jawi, D.A., Al-Hetari, Y., Ghori, M., and Khatri, A.R.,
J. Sep. Sci., 2008, 31, 1645–1650.
27. Hsieh, Y. and Chen, J., Rapid Commun. Mass Spectrom., 2005, 19, 3031–3036.
28. Chen, C.Y., Chang, S.N., and Wang, G.S., J. Chromatogr. Sci., 2009, 47, 67–74.
29. Liu, M., Ostovic, J., Chen, E.X., and Cauchon, N., J. Chromatogr. A, 2009, 1216,
2362–2370.
30. Alpert, A.J., Anal. Chem., 2008, 80, 62–76.

31. Neue, U.D., VanTran, K., Iraneta, P.C., and Alden, B.A., J. Sep. Sci., 2003, 26, 174–186.
32. Schebb, N.H., Fischer, D., Hein, E.M., Hayen, H., Krieglstein, J., Klumpp, S., and
Karst, U., J. Chromatogr. A, 2008, 1183, 100–107.
33. Wu, J.Y., Bicker, W., and Linder, W., J. Sep. Sci., 2008, 31, 1492–1503.
34. Persson, J., Hemstron, P., and Irgum, K., J. Sep. Sci., 2008, 31, 1504–1510.
35. Karlsson, G., Winge, S., and Sandberg, H., J. Chromatogr. A, 2005, 1092, 246–249.
36. Guo, Y. and Gaiki, S., J. Chromatogr. A, 2005, 1074, 71–80.
37. Ali, M.S., Ghori, M., Rafiuddin, S., and Khatri, A.R., J. Pharm. Biomed. Anal., 2007,
43, 158–167.
38. Guo, Y., Srinivasan, S., and Gaiki, S. Chromatographia, 2007, 66, 223–229.
39. Vikingsson, S., Kronstrand, R., and Josefsson, M., J. Chromatogr. A, 2008, 1187, 46–52.
40. Rouzo, G.L., Lamouroux, C., Bresson, C., Guichard, A., Moisy, P., and Moutiers, G.,
J. Chromatogr. A, 2007, 1164, 139–144.
41. Quiming, N.S., Denola, N.L., Saito, Y., Catabay, A.P., and Jinno, K., Chromatographia,
2008, 67, 507–515.
42. Novakova, L., Solichova, D., Pavlovicova, S., and Solich, P., J. Sep. Sci., 2008, 31,
1634–1644.
43. Jiang, W. and Irgum, K., Anal. Chem., 1999, 71, 333–344.
44. Walton, H.F. and Rochlin, R.D., Ion Exchange in Analytical Chemistry, CRC Press,
Boca Raton, FL, 1990.
45. Yang, X.Q., Dai, J., and Carr, P.W., J. Chromatogr. A, 2003, 996, 13–31.
46. Ali, M.S., Rafiuddin, S., Ghori, M., Rafiuddin, S., and Khatri, A.R., Chromatographia,
2008, 67, 517–525.
47. Dong, L. and Huang, J., Chromatographia, 2007, 65, 519–526.
48. Yanagida, A., Murao, H., Ohnishi-Kameyama, M., Yamakawa, Y., Shoji, A., Tagashira,
M., Kanda, T., Shindo, H., and Shibusawa, Y., J. Chromatogr. A, 2007, 1143, 153–161.
49. Scott, R.P.W., Adv. Chromatogr., 1982, 20, 169–187.
50. Li, R.P. and Huang, J.X., J. Chromatogr. A, 2004, 1041, 163–169.
51. Naidong, W., Lee, J.W., Jiang, X., and Wehling, M., J. Chromatogr. B, 1999, 735,
255–269.
52. de Person, M., Hazotte, A., Elfakir, C., and Lafosse, M., J. Chromatogr. A, 2005, 1081,
174–181.
53. Scott, R.P.W., J. Chromatogr. A, 1993, 656, 51–68.
54. Tai, A. and Gohda, E., J. Chromatogr. B, 2007, 83, 214–220.
55. Aversano, C.D., Hess, P., and Quillian, M.A., J. Chromatogr. A, 2005, 1081, 190–201.
56. Olsen, B.A., J. Chromatogr. A, 2001, 913, 113–122.
57. Valette, J.C., Demesmay, C., Rocca, J.L., and Verdon, E., Chromatographia, 2004, 59,
55–60.
58. Linden, H.V.D., Herber, S., Olthuis, W., and Bergveld, P., Sens. Mater., 2002, 14,
129–139.
59. Guillarme, D. and Heinisch, S., Sep. Purif. Rev., 2005, 34, 181–216.
60. Angyal, S.J., Angew. Chem. Int. Ed. Engl., 1969, 8, 157–166.
61. Nishikawa, T., Suzuki, S., Kubo, H., and Ohtani, H., J. Chromatogr. A, 1996, 720,
167–172.
62. Alpert, A.J., Shukla, M., Shukla, A.K., Zieske, L.R., Yuen, S.W., Ferguson, M.A.J.,
Mehlert, A., Paulv, M., and Orlando, R., J. Chromatogr. A, 1994, 676, 191–202.
63. Davidek, T., Kraehenbuehl, K., Devaud, S., Robert, F., and Blank, I., Anal. Chem., 2005,
77, 140–147.

12 Analysis of
Pharmaceutical
Impurities Using
Mingjiang Sun and David Q. Liu
Contents
12.1 Introduction...................................................................................................260
12.2 Trace Analysis of Genotoxic Impurities........................................................ 261
12.2.1 Alkyl Sulfonates................................................................................ 263
12.2.2 Alkyl Halides..................................................................................... 267
12.3 Analysis of Polar Impurities in Nonpolar APIs or Intermediates................. 273
12.3.1 Imidazole........................................................................................... 273
12.3.2 (3S)-3-Morpholinemethanol.............................................................. 273
12.3.3 Acetamide.......................................................................................... 275
12.3.4 Organic Acids.................................................................................... 275
12.4 Purity and Impurity Analysis of Polar APIs and Intermediates................... 276
12.4.1 5-Fluorouracil in 5-Fluorocytosine................................................... 276
12.4.2 Guanine in Acyclovir......................................................................... 277
12.4.3 Epirubicin.......................................................................................... 278
12.4.4 Chiral Compounds............................................................................. 279
12.4.5 Carbamates........................................................................................280
12.4.6 Sodium Cromoglicate........................................................................ 282
12.4.7 Cytosine............................................................................................. 283
12.5 Counterion Analysis...................................................................................... 283
12.6 Conclusions and Future Prospects................................................................. 285
References............................................................................................................... 287
259

12.1 Introduction
Organic impurities in active pharmaceutical ingredients (APIs) are derived from many
different sources during manufacturing and storage. They can be starting materials,1
intermediates,2 reagents, side-reaction products,3 packaging extractables, degradants,4
etc. The presence of such impurities could pose risks to the safety of the general pub-
lic, and thus their levels need to be controlled according to International Conference
on Harmonization (ICH) guidelines.5–8 Impurity analysis and control are the key
activities in process analytical development, and thus adequate analytical methods
are needed for the accurate determination of those impurities.9 Some API impurities,
however, are so polar that they cannot be retained on the conventional reversed-phase
high-performance liquid chromatographic (RP-HPLC) columns; therefore their analy-
ses constitute a real challenge to the analytical community. Research has been devoted
to the development of general strategies for the accurate analysis of very polar ana-
lytes. For instance, some polar compounds containing reactive functional groups such
as hydroxyl and amino can be analyzed after chemical derivatization to reduce the
polarity.10 Some polar analytes are charged, therefore ion chromatography or capillary
electrophoresis has been used for their analysis. Nonetheless, the application of these
methods has been limited because of the relatively poor reproducibility, complexity in
method development, and the limitation in choosing detection methodologies.
The polar pharmaceutical impurities discussed in this chapter refer to those that
are poorly retained on conventional RP-HPLC columns even if high aqueous or pure
aqueous mobile phases are used. It is well known that typical RP-HPLC columns
such as C18 columns may collapse in the high aqueous mobile phase by de-wetting,
resulting in poor retention, low selectivity, and irreproducibility.11 Newer RP-HPLC
columns that are more amenable to high aqueous mobile phases are commercially
avaliable.12 For the use of such columns, high aqueous content in sample diluent is
required in order to prevent peak fronting caused by solvent mismatch. Therefore,
its application in pharmaceutical analysis is constrained by the low aqueous solu-
bility of samples. Poor separation efficiency and reproducibility are also potential
issues for this approach. Furthermore, some compounds are so polar that they do not
retain on conventional RP-HPLC columns at all, even when the pure aqueous mobile
phase is used. Alternatively, ion pairing RP-HPLC employs ion pairing agents to
improve retention of polar compounds. Ion pairing reagents, however, are gener-
ally not compatible with mass spectrometry due to their ion suppression effects.13
Polar compounds very often have low molecular weight and lack a chromophore
for UV detection. Therefore, universal detectors such as evaporative light scatter-
ing (ELSD), conduct activity (CAD), and reflective index (RID) detectors have been
explored. Nonetheless, issues such as low specificity and narrow dynamic ranges
were experienced.14–16 Because of superior sensitivity, specificity, and versatility,
LC/MS has become the technique of choice for the low-level detection of many types
of pharmaceutical compounds. As such, the use of mass spectrometry compatible
mobile phases has become an important attribute of HPLC methods.
Hydrophilic interaction liquid chromatography (HILIC) is a complementary tool
for the analysis of polar compounds.17 HILIC uses polar stationary phases including
bare silica, polar functional group modified silica, or modified polymers, while the

Analysis of Pharmaceutical Impurities 261
mobile phases contain water and an organic solvent, mostly acetonitrile. The interac-
tion between water and the polar stationary phase forms a water-enriched layer on
the surface of the polar stationary phase. The partition of analytes between the bulk
of the organic-enriched mobile phase and the water-enriched layer on the surface of
the stationary phase results in their chromatographic separation. Polar compounds
will partition between the two phases, while nonpolar compounds may run through
the column in the mobile phase directly. As a consequence, polar compounds are
better retained than less polar ones. To date, various types of stationary phases
including bare silica, amino, amide, diol, cyano, poly(succinamide), sulfoalkylbe-
taine, and cyclodextrin-modified silica have been developed and commercialized.
Ikegami et al. provided an excellent review on such column types and their applica-
tions.18 The stationary phases can be acidic, basic, or neutral, which may interact
with analytes by ionic interaction and/or hydrogen bonding. Thus, the separation
of analytes on HILIC columns could be a mix of several different mechanisms.19–22
The applications of HILIC in the analysis of amino acids, peptides, proteins, sugars,
toxins, and drugs have been demonstrated in the literature.17 The recent develop-
ment in column technology has facilitated the rapid growth of HILIC applications in
various industries, such as pharmaceutical, food, and environmental analyses. In the
pharmaceutical industry, the technique has been used in drug discovery,23,24 phar-
macokinetics,13 quality and process controls of APIs and drug products.25,26 The fact
that many pharmaceutical APIs are highly soluble in organic solvents makes HILIC
a viable option for the determination of polar impurities in APIs.
The aim of this chapter is to provide a review on the applications of HILIC in the
analysis of polar impurities in APIs. In particular, the strategies to convert genotoxic
alkylating agents to positively charged quaternary amines to enhance mass spec-
trometry detection sensitivity are highlighted. Practical examples from both unpub-
lished data from the authors’ laboratory and recent literature will be presented.
12.2 Trace Analysis of Genotoxic Impurities

Potential genotoxic impurities are defined as impurities that may react with DNA
and potentially increase cancer risk in patients. The European Medicines Agency
(EMEA) issued guidelines regarding the control of potential genotoxic impuri-
ties in drug substances or drug products.8 The U.S. Food and Drug Administration
(FDA) also drafted a guidance document and made it available online recently.
Pharmaceutical manufacturers are actively seeking strategies to monitor and control
impurity levels in APIs according to the regulatory requirements.27 When a new
manufacturing process is developed, one of the goals is to avoid the usage and gen-
eration of genotoxic impurities.28 Nonetheless, it is almost impossible to completely
eliminate the risk of introducing genotoxic impurities into a particular manufactur-
ing process. Therefore, the levels of genotoxic impurities in APIs are managed by
a controlled risk approach. The acceptable risk caused by pharmaceutical products
for patients was defined as no greater than 10 −5 based on lifetime exposure. As such,
the daily intake of a specific genotoxin for a person cannot be greater than 1.5â•›μg.
Consequently, the concentration limits of genotoxic impurities in drug substances or
drug products can be derived based on the maximum daily dose. For instance, if the

H N R
O N N
NO2
R R
Epoxide Aziridine Aromatic nitro Aromatic azo
R X H
Cl, Br, I N O R NH2
(SH) R N N R OOH
H
Alkyl halides N-nitroso Hydrazines Hydroperoxide
NH2 O
O O
R O O OR
S
R1 O R S R
O O
R = methyl, ethyl, Acrylates
Alkyl sulfonates propyl, isopropyl Primary aniline (Michael acceptors)
Figure 12.1 Structures with high genotoxic concerns.
daily dose of a drug is 1â•›g, the allowable concentration of a genotoxic impurity in

the drug substance should be no more than 1.5â•›ppm (ppm, μg genotoxic impurity
per g API). Regulatory agencies also advocate applying the limit to the total of all
genotoxic impurities in an API that act through the same mechanism. Given such a
low ppm concentration limit, developing sensitive and robust methodology for their
detection poses great challenges on analytical technologies.
The EMEA issued a list of alerted structures for potential genotoxins, as illustrated
in Figure 12.1, which must be controlled in APIs at a trace level unless they are demon-
strated to be nongenotoxic by other experiments such as the Ames test.8 Alkyl sulfonates
and alkyl halides are among the alerted structures of potential genotoxic impurities.29,30
They could be reagents used during manufacturing processes, or the by-products from
side reactions. For example, common alcohols such as methanol, ethanol, and isopro-
panol can be converted to alkyl chlorides or alkyl sulfonates in the presence of HCl
or sulfonic acid, respectively, during API salt formation processes.31,32 They should be
analyzed and controlled at trace levels depending on the daily doses. The challenges of
developing robust analytical methods with desired low ppm sensitivity for those genotoÂ�
xins are well known.33 The instability of some of the analytes poses additional difficulties
to method development, and thus the sensitivity and reproducibility of direct analysis
methods of the unstable compounds are often unsatisfactory.34–37
Both alkyl sulfonates and alkyl halides are strong alkylators, which can readily react
with electron rich nucleophiles. This property can be advantageous for derivatization
reactions. By reacting with alkyl amines, permanently positively charged quater-
nary amines could be produced (Scheme 12.1), which are excellent candidates for
electrospray (ES) mass spectrometry detection. Therefore, converting to quaternary
amines followed by LC/MS detection seems to be a very effective strategy for ana-
lyzing alkyl sulfonates and alkyl halides. However, HPLC separation of the resulting
very polar quaternary amines is a new challenge. Previous reports demonstrate that
some quaternary amines can be analyzed by HPLC after derivatization or by ion

R1 R1
+
LG R2 + N(R3)3 R2 N(R3)3 + LG
LG: Leaving group
Scheme 12.1 Alkylators react with trialkyl amines producing quaternary ammoniums.
chromatography directly.10,38 The advances in HILIC made it possible to directly

analyze these compounds by HPLC. The high organic mobile-phase compositions
of HILIC methods are ideal for electrospray ionization (ESI), resulting in improved
ionization effficiency.39 Guo et al. explored the analysis of several quaternary amines
using amino or amide HILIC columns. It was found that these compounds were
better retained on an amide column than on an amino column.40 The interfacing
of HILIC with mass spectrometry lowered the limit of detection (LOD) of choline
by 75 times in comparison to the use of an RP-HPLC method.40 Mckeown et al.
analyzed several basic compounds with silica HILIC columns using a mobile phase
with a pH from 2 to 9.41 The quaternary amines are permanently positively charged,
while the surface silanol groups of the stationary phase of silica HILIC columns are
negatively charged in the neutral and even weak acidic mobile phase.42 Thus, silica
HILIC columns may offer an ion-exchange separation mechanism in addition to the
simple partition mechanism of HILIC, and thus the quaternary amines may be bet-
ter retained and resolved on silica columns.43 Aqueous ammonium formate buffers
are typically used as the strong solvent for HILIC silica columns. The peak shape
and retention factor of an analyte are affected by the concentration of ammonium
ions, and higher buffer concentrations reduce peak broadening and tailing.41 The
retention factors are also strongly affected by the mobile phase pH values. At pH 6,
the percentage of the aqueous solvent in the mobile phase may need to be as high as
30% to elute the quaternary amines derivatives with a desirable retention time. Since
a higher percentage of the organic mobile phase is desirable for better ionization
(sensitivity) in ESI, the aqueous mobile phase should be adjusted to pH 4 by add-
ing formic acid for optimal detection. The analytes can be eluted by using less than
20% of the aqueous mobile phase typically. The nonpolar APIs, on the other hand,
are generally eluted in the solvent front and diverted to waste stream before entering
the mass spectrometer to avoid interfering with the detection of the target analytes
(Figure 12.2). The isocratic gradient can often be implemented to reduce turn round
time by eliminating the extra column equilibrium time. The modified conditions are
optimal for mass spectrometry detection, thus providing superior method specificity
and sensitivity. Integrating HILIC/MS with derivatization offers a generic approach
for the analysis of the class of genotoxins, alkyl sulfonates, and alkyl halides.
12.2.1 Alkyl Sulfonates

The derivatization-HILIC/MS strategy was initially demonstrated for the determi-
nation of a group of 16 commonly encountered alkyl esters of sulfonates or sulfates
(Figure 12.3) in drug substances at low ppm levels by An et al.43 The method uses
trimethylamine as the derivatization reagent for ethyl, propyl, and isopropyl esters, and
triethylamine for the methyl esters. The sulfonates readily react with the derivatization

mAU
2500
2000 API
1500
1000
500
0
(a) 0 1 2 3 4 5 6 min
7000
6000
5000 HN N+
4000
3000
2000
1000
0
(b) 0 1 2 3 4 5 6 min
Figure 12.2 Separation of derivatized bis(2-chloroethyl)amine from API by HILIC.

(a) UV trace, (b) MS trace. The first 3 min of the LC flow was diverted to waste after UV
detection. See Tables 12.1 and 12.2 for experimental details.
O O O O
O O S S O O
O O TMA
S O
S
O
N+
O
EMS EBS ETS DES

O O
O O O O
O O S
S O S
S O O O TMA
O N+
PMS PBS PTS DPS

O O
O O O O TMA
O O
S
S
O N+
S O S
O O O
IPMS IPBS IPTS DIPS

O O O O
S O O TEA
O O O S
O N+
S S
O O O
MMS MBS MTS DMS
Figure 12.3 Structures of sixteen alkyl sulfonates and their trialkylamine derivatives.
reagents, producing four common derivatives: N,N,N-triethylmethanaminium, N,N,N-

trimethylethanaminium, N,N,N-trimethylpropanaminium, and N,N,N-trimethylisoÂ�
propanaminium. The resulted quaternary ammonium derivatization products are
highly polar (ionic) and can be retained by a silica HILIC column and readily separated
from the main interfering peak, e.g., the less polar API peak, which is usually
present at a very high concentration. The mobile phase used 15% of 50â•›mM ammo-
nium formate buffer (pH 4) in acetonitrile (Table 12.1). The chromatograms of all four

Table 12.1
HILIC Separation Conditions of Pharmaceutical Compounds
Matrices Impurities/Analytes Method Details References
Choline, chlormequat, HPLC-MS with an amide HILIC column [40]
Mepiquat, (TSK-Gel Amide-80, 250â•›mm × 2.0â•›mm,
acetylcholine, 3â•›μm), mobile phase consisting of a
betaine mixture of water (10â•›mM ammonium
acetate) and acetonitrile (19/81). Flow
rate 1â•›mL/min
Seven APIs Sixteen sulfonates Analytes were converted to quaternary [43]
amine before analysis. HPLC-MS with a
silica HILIC column (Atlantis, 50â•›mm ×
2.1â•›mm, 3â•›μm), mobile phase consisting
of a mixture of water (50â•›mM ammonium
formate and 0.1% formic acid, pH 4) and
acetonitrile (15/85/v/v). Flow rate 0.3â•›mL/
min. SIM: m/z 88, m/z 102, or m/z 116
Six APIs Seven alkyl halides Analytes were converted to quaternary [44] and [45]
amine before analysis. HPLC-MS with a
silica HILIC column (Atlantis, 50â•›mm ×
2.1â•›mm, 3â•›μm), mobile phase consisting
of a mixture of water (50â•›mM ammonium
formate and 0.1% formic acid, pH 4) and
acetonitrile (with varied ratio). Flow rate
0.3â•›mL/min
API Imidazole HPLC with a silica HILIC column
(Atlantis, 100â•›mm × 2.1â•›mm, 3â•›μm),
mobile phase consisting of a mixture of
water (50â•›mM ammonium formate and
0.1% formic acid) and acetonitrile (1/99).
Flow rate 1â•›mL/min. λ: 220â•›nm
Ethanol cleaning (3S)-3- HPLC-MS with a silica HILIC column
solution Morpholinemethanol (Atlantis, 50â•›mm × 2.0â•›mm, 3â•›μm),
mobile phase consisting of a mixture of
water (20â•›mM ammonium formate) and
acetonitrile (15/85). Flow rate 0.3â•›mL/
min. SIM: m/z 118
Intermediate Acetamide HPLC with an amino HILIC column [49]
(Zorbax NH2, 250â•›mm × 4.6â•›mm, 5â•›μm),
mobile phase acetonitrile-water (90/10);
Flow rate 1.0â•›mL/min. λ: 205â•›nm
API Oxamide, oxamic HPLC with an amino HILIC column [49]
acid, oxalic acid (Zorbax NH2, 250â•›mm × 4.6â•›mm, 5â•›μm),
mobile-phase acetonitrile-water (60/40),
50â•›mM potassium phosphate, pH 7.0;
(continued)


5-Fluorocytosine 5-Fluorouracil HPLC with a silica HILIC column [49]
(Zorbax SIL, 250â•›mm × 4.6â•›mm, 5â•›μm),
5â•›mM phosphoric acid; Flow rate 1.0â•›mL/
min. λ: 275â•›nm
Or
HPLC with an amino HILIC column
25â•›mM potassium phosphate pH 6.5;
Acyclovir Guanine HPLC with a silica HILIC column [49]
(Zorbax SIL, 250â•›mm × 4.6â•›mm, 5â•›μm),
5â•›mM phosphoric acid; Flow rate 1.0â•›mL/
min. λ: 275â•›nm
Epirubicin Epidaunorubicin HPLC with a silica HILIC column [55]
Daunorubicin (Kromasil KR100-5SIL, 250â•›mm ×
Doxorubicin 4.6â•›mm, 5â•›μm), mobile-phase acetonitrile,
30â•›mM sodium formate, pH 2.9 (70/30);
LY293558 LY293559 HPLC with a Chirobiotic T• column [57]
(250â•›mm × 4.6â•›mm), mobile-phase
acetonitrile-water (65/35); Flow rate
1.0â•›mL/min. ELSD detection
API Three polar carbamate HPLC with a diol column (YMC-pack [2]
impurities Diol-120-NP, 250â•›mm × 4.6â•›mm, 5â•›μm),
mobile phase 10â•›mM NH4Cl in
acetonitrile-water (95/5); Flow rate
1.5â•›mL/min. λ: 215â•›nm
Sodium Unspecified HPLC with a silica HILIC column [60]
cromoglicate degradants (Atlantis, 250â•›mm × 4.6â•›mm, 5â•›μm),
mobile phase 30â•›mM NH4Ac in
acetonitrile-water (86/14, pH 5.6); Flow
rate 1.5â•›mL/min. λ: 326â•›nm
Cytosine Uracil HPLC with an amide column (TSK-Gel [61]
7-Aminopyrido[2,3-d ] Amide-80, 250â•›mm × 4.6â•›mm, 5â•›μm),
pyrimidin-2-ol mobile phase A: sodium hydroxide/
formic acid (pH 3.0), mobile phase B:
acetonitrile; Gradient: 5% A for 5â•›min,
then linear increase to 25% A in 10â•›min,
and hold for 5â•›min; Flow rate 1.0â•›mL/min


10 APIs Inorganic ions: Na+, HPLC with a zwitterionic column [68]
K+, Zn2+, Ca2+, Mg2+, (ZIC-HILIC, 250â•›mm × 4.6â•›mm, 5â•›μm),
Cl−, Br−, NO3−, mobile phase A: acetonitrile/buffer
SO42−, PO43−; (85/15), mobile phase B: acetonitrile/
Organic ions: lysine, buffer (10/90); Buffer: ammonium
diethanolamine, acetate, pH adjusted by acetic acid;
trizma, piperazine, Gradient: 0% B for 2â•›min, then linear
choline, esylate, increase to 100% A in 20â•›min, and hold
mesylate, edisylate, for 5â•›min; Flow rate 1.0â•›mL/min.
isethionate, citrate, Detector: ELSD
glucuronate,
mandelate, tartrate,
fumarate, glycolate,
glutarate, maleate,
malate, tosylate,
napadisylate,
benzylamine,
arginine
API Tromethamine HPLC with an amino HILIC column [67]
mobile-phase acetonitrile-water (80/20);
Flow rate 1.0â•›mL/min. Detector: RID
derivatives are depicted in Figure 12.4. The cationic analytes can be retained on a silica
HILIC column through the interaction of the cations with the ionized silanol groups.
The method gave excellent sensitivity for all the alkyl esters at the targeted analytical
level of 1–2â•›ppm when the API samples were prepared at 5â•›mg/mL with a single quaÂ�
drupole mass detector. The recoveries at 1–2â•›ppm were generally above 85% for all the
alkyl esters in various APIs. The injection precisions of the lowest standards ranged
from 0.4% to 4% (%RSD). A linear range from 0.2 to 20â•›ppm concentrations was estab-
lished, with R2 ≥ 0.99. The generic method has been applied to the development of
several investigational drugs, and there was no need for method alteration for different
APIs. However, the method is unable to distinguish the alkyl donors if multiple sources
coexist in the sample. On the other hand, the method can be used to quantify the total
methyl donors when multiple reactive methyl impurities coexist, and this also applies
to the ethyl, propyl, or isopropyl alkylating impurities.
12.2.2 Alkyl Halides

Alkyl halides are another group of common alkylators that need to be controlled in
APIs. Low molecular weight volatile alkyl halides are typically analyzed by head-
space gas chromatography (HSGC) with mass spectrometry or an electron capture

10,000 Dimethyl sulfate at 1 ng/mL
+
8,000 N
6,000
4,000
(a) 2.5 3 3.5 4 4.5 5 5.5 6 6.5 min
2,400 Ethyl methanesulfonate at 1 ng/mL

2,200 +
2,000 N
1,800
1,600
1,400
(b) 2.5 3 3.5 4 4.5 5 5.5 6 6.5 min
12,000 Dipropyl sulfate at 1 ng/mL +

10,000 N
8,000
6,000
(c) 2.5 3 3.5 4 4.5 5 5.5 6 6.5 min
35,000
30,000 Diisopropyl sulfate at 10 ng/mL +
25,000 N
20,000
15,000
10,000
(d) 2.5 3 3.5 4 4.5 5 5.5 6 6.5 min
Figure 12.4 Separation of derivatized alkyl sulfonates on a silica HILIC column. See Table 12.1 for experimental details.

detector (ECD).29 Direct injection GC methods are generally not desirable because
of potential instrument contamination issues due to the presence of a large amount
of API. Furthermore, for some unstable or nonvolatile polar alkyl halides, the HSGC
approach is not an option because of the long incubation duration (typically 10â•›min)
at the elevated temperature (near boiling point of analytes). And these compounds
may not survive such conditions especially in the presence of API. A group of alkyl
halides that fall into this category is listed in Table 12.2. They are either unstable or
not volatile enough for HSGC analysis, nor suited for direct HPLC analysis.44,45 For
example, bis(2-chloroethyl)amine seems a good candidate for direct LC/MS analy-
sis. However, this compound is readily hydrolyzed to a series of degradants in an
aqueous solution.46,47 Therefore, an analysis of this compound after derivatization not
only stabilized the compound but also improved its detectability.45 3-Chloropropane-
1,2-diol, has been analyzed by headspace solid-phase microextraction (HS-SPME)
coupled with the GC/MS method following derivatization with phenylboronic acid.48
The detection limit of the method was 3.78â•›ppm, which was insufficient for the appli-
cation in the analysis of pharmaceutical genotoxins.
Following the same strategy developed for the analysis of alkyl sulfonates, alkyl
halides can also be converted to polar amines that can be readily separated from APIs
by HILIC and detected by electrospray mass spectrometry.44,45 Based on the reactiv-
ity of the alkyl halides, they can be derivatized with either dimethylamine (DMA)
or trimethylamine (TMA). The derivatization conditions and structures of deriva-
tives are listed in Table 12.2. For the majority of the compounds, the derivatization
reactions proceed quickly. Alkyl chlorides, on the other hand, are generally not as
reactive as their bromide and iodide analogues. 4-Chlorobutan-1-ol was derivatized
with dimethylamine catalyzed by NaI. Its derivative is not permanently charged,
but is protonated in the slightly acidic mobile phase. A separation of the deriva-
tives from API and other interferences was achieved on a silica HILIC column with
the mobile phase containing acetonitrile and an ammonium formate buffer of pH 4
(Table 12.1). Because of the diverse structures of the alkyl halides, a single gradient
may not be sufficient to cover the separation of all the analytes. For optimal separa-
tion and detection sensitivity, the percentage of the aqueous buffer in the mobile
phase is adjusted in a typical range from 13% to 30% (Table 12.2). The retention of
the derivatives appears to be related to both their polarity and their ability to gener-
ate ionic interactions with the stationary phase (Figure 12.5). Nonetheless, further
investigations seem warranted to understand how the molecule structures and chro-
matographic conditions impact the retention factor.
Excellent sensitivity was achieved for all these trace levels of analytes in the
presence of API. The limits of quantitation (LOQs) of the various analytes in the range
of 0.1–0.2â•›ppm (ca) were successfully achieved. In the case of 3-chloropropane-1,2-
diol,44 the HILIC/MS approach was about 20 times more sensitive than the derivatiza-
tion-HS-SPME-GC/MS method, which had an LOQ of 3.78â•›ppm.48 The recovery
tests were performed by spiking alkyl halides into various investigational APIs where
their manufacturing process involved the use of the corresponding alkyl halides. It
was observed that the recovery of the analytes ranged from 50% to 109% depending
on the API sample matrices (Table 12.2). Since these methods are limit tests only, the

Table 12.2
Structures of Seven Alkyl Halides, and Key Analytical Validation Results of Derivatization HILIC/MS Methods
Genotoxic Reaction Temp/ Mobile LOQ
Impurity Reagent Derivative Time (°C)/h Diluent Phase A/B (ppm) (ca) Recovery (%)
H DMA 60/1 ACN/acetic acid (99/1) 30/70 0.2 98% at 1â•›ppm

N +
HN N with %RSD 1.6%
m/z 115
Cl Cl
O DMA 35/1 ACN/H2O (85/15) 13/87 0.1 89% at 2â•›ppm
O N+ with %RSD 0.4%
Cl Cl m/z 116
Cl DMA 60/24 2â•›mg/mL NaI in ACN/H2O (9/1) 14/86 0.2 50% at 2â•›ppm
N with %RSD 5.1%
HO H+
HO
m/z 118
DMA OH 75/2 ACN/H2O (4/1) 17/83 0.2 58% at 3.5â•›ppm
OH HO
Cl with %RSD 3.0%
H+
N
HO
m/z 120

Analysis of Pharmaceutical Impurities
TMA 60/1.5 ACN/H2O (4/1) 15/85 0.1 87% at 5â•›ppm
Cl Cl
with %RSD 3.3%
Br N+
m/z 136
TMA 60/1 ACN/H2O (4/1) 15/85 0.1 56% at 2â•›ppm
Br OH N+ OH with %RSD 2.1%
m/z 104
TMA 70/4 ACN/H2O (4/1) 15/85 0.1 109% AT 3â•›ppm
I OH N+ OH with %RSD 3.2%
m/z 104
271
Intens.

6000
5000
4000 Bis(2-chloroethyl)amine, 5 ng/mL
3000 (a)
2000
×104
2.0
1.5 2-Chloroethyl ether, 10 ng/mL
1.0
0.5 (b)
0.0
4000
2000 4-Chloro-1-butanol, 5 ng/mL
(c)
5000
4000
3000
2000 3-Chloro-1,2-propanediol, 10 ng/mL
1000 (d)
0
×104
1.5
1.0
0.5 (e) 1-Bromo-3-chloropropane, 20 ng/mL
0.0
×104
4
2-Bromoethanol, 10 ng/mL System peak
2
(f )
0
×104
3
2 System peak
2-Iodoethanol, 10 ng/mL
1 (g)
0
1 2 3 4 5 6 7
Time [min]
Figure 12.5 Separation of derivatized alkyl halides by HILIC. See Tables 12.1 and 12.2 for details.

relatively low recoveries are acceptable per regulatory method validation guidelines.
The relatively low recovery of certain analytes in certain sample matrices requires
further investigation. One possible reason could be that the derivatization reaction
conditions for those compounds were not optimal and APIs may have interfered with
the derivatization reaction. In summary, the HILIC/MS approach provides an alter-
native strategy for the sensitive detection of some unstable alkyl halides.
12.3 Analysis of Polar Impurities in Nonpolar

APIs or Intermediates
Certain polar impurities in nonpolar APIs or intermediates cannot be quantified using
a single impurity profile method because they may not be retained on the RP-HPLC
columns. In such a situation, a second method is usually required to quantify the
specific polar impurity. HILIC has been successfully applied to the analysis of polar
impurities in the presence of APIs or intermediates, of which the polar impurities are
retained on HILIC columns, while the nonpolar components are eluted earlier than
the targeted polar impurities.
12.3.1 Imidazole
N,N′-Carbonyldiimidazole is a common coupling reagent for the synthesis of amides
and carbamates. Imidazole (Figure 12.6) is a by-product of the coupling reactions,
which needs to be controlled as low as 0.05% w/w. Imidazole was not retained on the
C18 column that was used for the impurity profile analysis. An ion pairing chromato-
graphic method was also explored. However, imidazole was barely retained by a C8
column with 1-hexanesulfonic acid as the ion pairing agent in a high aqueous mobile
phase. Ultimately, a HILIC method was developed for the accurate determination of
imidazole using a silica HILIC column. The mobile phase was a combination of the
ammonium formate buffer and acetonitrile. It was observed that the retention factor
of the imidazole peak was greatly affected by the percentage of the aqueous buffer,
as expected for HILIC. The desired retention of the analyte was achieved with an
isocratic elution using an ammonium formate buffer in acetonitrile (Table 12.1). The
API peak eluted in the column void for about 0.5â•›min while the imidazole peak had
a retention time of about 2.0â•›min with a good peak shape and a tailing factor of 1.2
(Figure 12.7). The method was validated according to ICH guidelines. The %RSD of
injection precisions is less than 0.01%. LOQ can be as low as 0.0004% (w/w) based
on a 50â•›mg/mL sample.
12.3.2 (3S)-3-Morpholinemethanol
In regulated API manufacturing facilities, reaction vessels or drying equipment must
be cleaned and verified before being released for producing a different drug or inter-
mediate. Depending on the dose of the compounds, the levels of impurities in the
cleaning solution are usually controlled at ppm (μg/mL) levels, and must be verified
by validated limit test analytical methods. (3S)-3-Morpholinemethanol (Figure 12.6)
is a building block of an experimental drug in early development. After production

Cl N+ N+ O N+
HO N+
O
Choline Chlomequat Mepiquat Acetylcholine
O O
HN
–O N+ N
NH OH
Betaine Imidazole (3S)-3-Morpholinemethanol
O O O
O NH2 OH OH
NH2 H2N O HO
O NH2 O
Acetamide Oxamide Oxamic acid Oxalic acid
Figure 12.6 Structures of representative pharmaceutical compounds and impurities

analyzed by HILIC.
0.40
0.35
0.30
API
0.25
0.20
AU
0.15
Imidazole
0.10
0.05
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00
min
Figure 12.7 Separation of imidazole from API on an Atlantis Silica HILIC column. See
Table 12.1 for details.
of the API, the equipment was cleaned with 200 proof ethanol. The levels of the
compound need to be controlled to less than 8â•›ppm. The process monitoring GC
method cannot provide sufficient sensitivity. Considering the polar nature of this
compound, a HILIC method was developed. A silica HILIC column was used for
the separation with 15% of 20â•›m M ammonium formate buffer (pH 6) in acetoni-
trile as the mobile phase. The separation of the peaks is illustrated in Figure 12.8.

1,000,000
800,000 Blank
600,000
400,000
200,000
0
1.5 2 2.5 3 3.5 4 4.5 min
1,000,000
8000,00 (3S)-3-morpholinemethanol at 8 ppm in ethanol
6000,00
4000,00
2000,00
0
1.5 2 2.5 3 3.5 4 4.5 min
Figure 12.8 Analysis of (3S)-3-morpholinemethanol by HILIC/MS. See Table 12.1 for

details.
An Agilent LC/MSD was used, and the analyte was detected in the SIM mode by
monitoring the ion at m/z 118. The S/N at 8â•›ppm was better than 50. The injection
precision was 8% RSD. The relatively high value is because m/z 118 is one of the ions
used for instrument calibration.
12.3.3 Acetamide
Olsen explored the application of silica and amino columns for the analysis of low
molecular weight acetamide (Figure 12.6).49 Both the columns seemed able to retain
the compound. To determine the acetamide level in a pharmaceutical intermedi-
ate, an amino HILIC column with the acetonitrile/water (90/10) mobile phase was
used. As expected, the bulk intermediate peak eluted earlier than the acetamide peak
(Figure 12.9). The acetamide was sufficiently retained on the column to allow detec-
tion at a 0.05% level. However, the sensitivity was undermined by the tailing of the
large intermediate peak due to the insufficient specificity of UV detection.
12.3.4 Organic Acids

When neutral or acidic mobile phases are used, the stationary phase of amino col-
umns are protonated. Anionic analytes such as organic acids can be retained via
ion-exchange mechanisms in addition to hydrophilic partition. Oxamic acid, oxalic
acid, and neutral oxamide, the possible impurities in an API (Figure 12.6), could not
be resolved by RP-HPLC but were separated by an ion-exclusion chromatography
previously.50 On an amino column with acetonitrile/phosphate buffer at pH 7 as the
mobile phase, the compounds were retained in the following order: di-anionic oxalic
acid > mono-anionic oxamic acid > neutral oxamide. The elution order appears to
be in correlation with acidity.49 Their retentions seem dependent on the buffer con-
centrations. Increasing the buffer concentration resulted in less retention of the two

605.00
602.90
600.80
%Acetamide
598.70
added
Response (mV)
596.60
594.50
0.1
592.40
590.30 0.05
588.20 0.0
586.10
584.00
0 48 96 144 192 240 288 336 384 432 480
Time (s)
Figure 12.9 Determination of acetamide in a pharmaceutical intermediate. Conditions:

Zorbax NH2 column; acetonitrile–water (90:10) mobile phase; 205â•›nm detection; sample con-
centration: 4â•›mg/mL in tetrahydrofuran. (Reprinted from Olsen, B.A., J. Chromatogr. A, 913,
acids but did not affect the retention of the neutral oxamide. The experiments dem-
onstrated that the ion-exchange mechanism is responsible for the retention of the
acids on the HILIC columns. A successful determination of the three compounds
in a nonpolar API was achieved on an amino column using acetonitrile and 50â•›m M
potassium phosphate pH 7 as the mobile phase (Table 12.1). The two acids were well
separated from the API peak while oxamide eluted at the tail of the API peak. Oxalic
acid gave a relatively broad peak so its sensitivity was no better than that of the
ion-exclusion chromatography method.50 Other factors such as the pH of the mobile
phases could be further optimized to improve the separation.
12.4 P
urity and Impurity Analysis of Polar APIs
and Intermediates
12.4.1 5-Fluorouracil in 5-Fluorocytosine
5-Fluorouracil (Figure 12.10), an impurity in 5-fluorocytosine, needs to be controlled
below the limit of 0.1%.49 Both silica and amino columns were explored for the deter-
mination of 5-fluorouracil in 5-fluorocytosine. The less polar 5-fluorouracil was eluted
in front of 5-fluorocytosine on both columns with good resolution (Figure 12.11). Two
limit test methods were developed using the two columns respectively. One utilized
a silica HILIC column eluted with an acidic mobile phase and 5â•›mM phosphoric acid
(25%) in acetonitrile. Another used the amino HILIC column with a near neutral
mobile phase and 25â•›mM potassium phosphate in acetonitrile with a pH of 6.5 (Table
12.1). Recently, other studies suggest that a silica-based amino stationary phase is not

NH2
O HO OH
NH N N
H2N NH HO N N NH2 HO
O N O
H
Uracil Cytosine 7-ADOP Tromethamine
O O
F H
N F N N
NH HN HN OH
H2N N O O N O H2N N N H2N N N
H H
O
5-Fluorocytosine 5-Fluorouracil Guanine Acyclovir
O
O OH O R H H H
O OH O R
OH
OH OH NH
HN N
H
N N
O O OH O O OH
LY293558
OH O O O O O
CH3 CH3 H H H
OH OH
NH2 NH2 NH
HN N
H
N N
Epirubicin (R = OH) Doxorubicin (R = OH)
Epidaunorrubicin (R = H) Daunorrubicin (R = H) LY293559
NaOOC O O COONa
OH
O O O O
Sodium
cromoglicate
Figure 12.10 Structures of representative pharmaceutical compounds and impurities

analyzed by HILIC continued.
stable under these chromatographic conditions, and a polymer-based amino column is

a better choice in terms of chromatographic stability and column bleeding.51,52
12.4.2 Guanine in Acyclovir

A similar limit test method was developed for the determination of guanine (Figure
12.10), an impurity in acyclovir with an LOQ of 0.7%.49 The mobile phase contained
acetonitrile and 5â•›mM phosphoric acid (Table 12.1). Both guanine and acyclovir were
well retained and resolved with the guanine peak eluting after the acyclovir peak.
The silica HILIC column performed better than the amino column in terms of method
selectivity, and thus it was selected over the amino column where guanine was not
baseline resolved from the acyclovir peak. Furthermore, a zwitterionic HILIC column
may be potentially evaluated as an alternative approach for the similar purposes.53

1040.0
1033.5
1027.0
1020.5
Response (mV)
1014.0 5-FU
5-Fluorocytosine +
1007.5 0.1% 5-fluorouracil
1001.0
994.50
988.00
5-Fluorocytosine
981.50
975.00
0 120 240 360 480 600 720 840 960 1080 1200
(a) Time (s)
630.00
625.00
620.00
615.00
5-Fluorocytosine +
Response (mV)
610.00
5-FU 0.08% 5-fluorouracil
605.00
600.00
595.00
5-Fluorocytosine
590.00
585.00
580.00
0 72 144 216 288 360 432 504 576 648 720
(b) Time (s)
Figure 12.11 Determination of 5-fluorouracil in 5-fluorocytosine. (a) Zorbax SIL column;

mobile-phase acetonitrile–water (75:25), 5â•›m M phosphoric acid; 275â•›nm detection; 5-fluoro-
cytosine (1.0â•›mg/mL). (b) Zorbax NH2 column; mobile-phase acetonitrile–25â•›m M potassium
phosphate, pH 6.5 (80:20); 275â•›nm detection; 5-fluorocytosine (1.0â•›mg/mL). (Reprinted from
Olsen, B.A., J. Chromatogr. A, 913, 113, 2001. With permission from Elsevier.)
12.4.3 Epirubicin
Epirubicin, an oncology drug, contains doxorubicin, daunorubicin, and epidaunorubi-
cin as the major impurities (Figure 12.10). They all contain a very polar daunosamine
sugar moiety and have been analyzed by ion pairing HPLC.54 Li and Huang stud-
ied the separation of these compounds on a bare silica HILIC column and obtained

10
Epidaunorubicin
Daunorubicin
9
Epirubicin
Doxorubicin
8
Retention factor (k΄)
2 3 4 5 6 7
Buffer pH
Figure 12.12 Effect of buffer pH on the retention factor (k′) of epirubicin and its ana-
logues. Conditions: Kromasil KR100-5SIL (5â•›μM); mobile phase: sodium formate buffer
(20â•›m M) modified with acetonitrile (10:90, v/v). Peaks: (▪) epidaunorubicin, (•) daunorubicin,
(▴) epirubicin, (▾) doxorubicin. (Reprinted from Li, R. and Huang, J., J. Chromatogr. A, 1041,
promising results.55 A mixture of acetonitrile and sodium formate buffer was used
as the mobile phase. The pH value of the mobile phase was screened between pH 2.4
and 6.5. It was observed that the pH value of the buffer had a significant effect on the
retention factors of the analytes (Figure 12.12). At a low pH, the retention of these
molecules increased with the increase of the mobile phase pH. The best retention was
obtained at about pH 4.2 for all compounds. However, a small change of the mobile
phase pH around pH 4.2 may change the elution order of daunorubicin and epirubi-
cin. A further increase of the mobile phase pH caused a decrease of the retention fac-
tors, and the analytes displayed strong irreversible binding to the silica column at pH
6.5 or higher, which was consistent with the instability of the dihydroquinone type of
compounds at higher pH. Based on the experimental results, the optimal pH value of
the mobile phase should be in the range from 2.9 to 3.6 or 5 to 6.5. In these pH ranges,
the compounds were well resolved and the pH effects on the retention factor and elution
order were relatively small. A higher pH should be avoided because of the irreversible
adsorption of the analytes on the column. A sodium formate buffer (30â•›m M) at pH
2.9 was selected for the batch analysis of epirubicin containing the related impurities
(Table 12.1). All impurities were resolved from the major peak with resolutions bet-
ter than 2.0 (Figure 12.13). Under similar conditions, the temperature effects on the
separation were studied separately.56 A higher column temperature at 40°C improved
column efficiency resulting in a slightly better separation of the analytes.
12.4.4 Chiral Compounds

An experimental drug LY293558 cannot be easily separated from its enantiomeric
impurity LY293559 (Figure 12.10) by a normal-phase chiral method without involving

70
60
4
50
Absorbance (mAU)
40
30
20
10
3
1 5
2
0
0 5 10 15 20 25
Time (min)
Figure 12.13 Chromatogram of the analysis of epirubicin hydrochloride in a raw material.

Conditions: Kromasil KR100-5SIL (5â•›μm); mobile phase: sodium formate buffer (30â•›m M,
pH 2.9)–acetonitrile (10:90, v/v). Peaks: (1) epidaunorubicin, (2) unknown compound,
(3) daunorubicin, (4) epirubicin, (5) doxorubicin. (Reprinted from Li, R. and Huang, J.,
deravitization.57,58 Guisbert et al. reported the direct separation of the enantiomers

using a Chirobiotic T• column, which displayed classic HILIC characteristics when
the water–acetonitrile mobile phase was used. The retention factor of LY293558 was
improved from 2.0 to 5.2 when the acetonitrile content was increased from 40%
to 80% (Figure 12.14). At a higher composition of acetonitrile (80%) in the mobile
phase, the resolution of the two enantiomers was improved to 7.3 from 3 at a lower
organic composition (5%). The pH value in a narrow range (4.0–6.5) did not seem
to have significant effects on the retention and resolution of the two enantiomers.
The optimized separation of the two enantiomers was accomplished with a mobile
phase of aqueous acetonitrile (65%) with an ELSD for detection (Figure 12.15). To
overcome the narrow dynamic range of the detector, a “high–low” injection strategy
was applied. That is, high concentration samples were injected for the determination
of the low level impurity, while a further diluted sample was injected for the deter-
mination of the major peak. The method was validated in terms of linearity, preci-
sion, accuracy, recovery, and sensitivity. At a sample concentration of 10â•›mg/mL of
LY293558, the quantification limit of LY293559 could be as low as 0.1%.
12.4.5 Carbamates
Wang et al. developed a stability-indicating method for the analysis of a proprietary
experimental drug.2 Three impurities or degradants were structurally closely related to
the API. The structures of the analytes were not disclosed. Compound 1 has amino and

6.0
5.0
4.0
Capacity factor
3.0
2.0
1.0
0.0
5 10 15 20 25 30 35 40 50 60 70 80
%Acetonitrile
Figure 12.14 Effect of increasing the organic modifier on the capacity factor. (Reprinted
from Guisbert, A.L. et al., J. Liq. Chromatogr. Relat. Technol., 23, 1019, 2000. With permission
from Elsevier.)
30
27
24
21 1
Response (mV)
18
15 2
12
9
6
3
0
0 90 180 270 360 450 540 630 720 810 900
Time (s)
Figure 12.15 Sample chromatogram for the separation of LY293558 (1) and LY293559
(2). (Reprinted from Guisbert, A.L. et al., J. Liq. Chromatogr. Relat. Technol., 23, 1019, 2000.
hydroxyl functional groups. The polar API (Compound 2) is aromatic, and contains
amino and carbamate functional groups. Compound 3 has urea and alcohol functional
groups. Compound 4 contains a carbamate functional group. The polarity and basic-
ity of Compound 1 and API were obviously higher than that of compounds 2 and 3
because of their amino functional groups. In an RP-HPLC analysis, their retention
factor was in the order of compound 1 < compound 2 < compound 3 < compound 4

0.50
Comp 2
0.45
0.40
0.35
0.30
0.25
AU
Comp 3
0.20
0.15
Comp 4
Comp 1
0.10
0.05
(a) 0.00
0.50
0.45
0.40
Comp 2
0.35
0.30
0.25
AU
0.20
Comp 3
Comp 4
0.15
Comp 1
0.10
0.05
0.00
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0
(b) Minute
Figure 12.16 Chromatograms of the specificity solution on (a) RP-HPLC and (b) HILIC.
RP-HPLC conditions: column 100â•›m m × 4.6â•›m m 5â•›μm Xterra MS C18; gradient elution with
(A) 0.09% phosphoric acid and (B) acetonitrile (B, 2%–25% in 20â•›m in); column temperature
35°C; flow rate 1â•›m L/min; UV detection 215â•›nm. HILIC conditions: column 250â•›m m × 4.6â•›m m
5â•›μm YMC-pack Diol-120 NP; mobile phase 10â•›m M NH4Cl in acetonitrile/water (95:5, v/v);
column temperature 30°C; flow rate 1.5â•›m L/min; UV detection 215â•›nm. (Reprinted from
Wang, X. et al., J. Chromatogr. A, 1083, 58, 2005. With permission from Elsevier.)
(Figure 12.16a). An orthogonal HILIC method using a diol column was also devel-
oped. Under HILIC conditions, the retention factor was changed to the order of com-
pound 4 < compound 3 < compound 1 < compound 2 (Figure 12.16b). The addition
of ammonium chloride to the water/acetonitrile mobile phase did not assert significant
effects on the separation. A mobile phase containing acetonitrile and 10â•›mM NH4Cl
was selected for the final method. The method was validated according to ICH guide-
lines, and LOQs of the impurities were estimated to be 0.05%. The HILIC method
served as an orthogonal method for the RP-HPLC method.
12.4.6 Sodium Cromoglicate

Sodium Cromoglicate (SCG, Figure 12.10), a polar drug for the treatment of allergy,
had been analyzed by RP-HPLC with a high aqueous mobile phase.59 SCG (pKa 1.9)

is negatively charged under basic or weakly acidic conditions. Alternatively, the

compound could be analyzed by HILIC. Ali et al. developed a stability-indicating
impurity profile method for the compound.60 SCG was analyzed by a silica HILIC
column using ammonium acetate buffered acetonitrile/water as the mobile phase
(Table 12.1). Increasing the pH of the mobile phase from 4.0 to 5.8 reduced its reten-
tion factor from 8.7 to 3.7, probably due to the increased repulsion between the nega-
tively charged silanol and the negatively charged analyte. Lowering the mobile phase
pH increased the retention of SCG, and thus a buffer of pH 3.0 was selected. At pH
3.0, increasing the concentration of ammonium acetate slightly improved the reten-
tion from 3.1 to 4.4, presumably because the ammonium ion would interact with the
negative charges of the silanols and thus reduce the repulsion between SCG and ion-
ized silanol. The method was validated with respect to linearity, specificity, accuracy,
precision, and sensitivity. Forced degradation samples of SCG were analyzed by the
method, and the degradants were well separated from the API peak (Figure 12.17).
A limited method robustness test was performed. Furthermore, as demonstrated by
Yang et al., amino columns might be an alternative for acidic compounds.50
12.4.7 Cytosine
Cytosine (Figure 12.10) is a common building block for the synthesis of many drug
substances. 7-ADOP and uracil are two major impurities observed in commercial
supplies.61 Cytosine and other pyrimidines have been analyzed either by GC after
derivatization or ion-pairing HPLC methods.62,63 One amide and two amino col-
umns were screened for the separation of the three compounds using a gradient
mobile phase from 5% to 25% aqueous sodium formate (pH 3.5)/acetonitrile in
10â•›min. Cytosine and 7-ADOP were retained on both columns, while uracil was
only retained on amide columns. Thus, a HILIC method using an amide column
was developed. The method was fully validated, demonstrating acceptable linear-
ity, precision, repeatability, and sensitivity. During robustness evaluation, a range
of parameters including flow rate, column temperature, injection volume, mobile
phase pH, buffer concentration, values of retention factor, tailing factor, and resolu-
tion were shown to be satisfactory. Intermediate precision and sample stability were
also evaluated. Should there be a need to convert the method to a mass spectrometry
compatible method, ammonium formate buffer could be used.
12.5 Counterion Analysis

In order to improve physicochemical properties, drug substances are often produced
as salts. The counterions could be inorganic ions, polar organic bases, or acids. The
stoichiometry of the salt form is a critical attribute of APIs and thus must be deter-
mined as part of batch release tests before the usage. These counterions are generally
not retained on conventional RP-HPLC columns, and had been analyzed by several
special techniques including inductively coupled plasma-atomic emission spectrom-
etry (ICP-AES),64 capillary electrophoresis (CE),65 and ion chromatography (IC).66
While IC is the most commonly used approach in the pharmaceutical industry, indi-
vidual IC method developments are necessary for each assay. HILIC-based HPLC

0.15
0.15
0.10 0.10
V
V
0.05 0.05
0.00 0.00
SCG
SCG
0 5 10 0 5 10
(a) min (b) min
0.15
0.10
0.10
V
V
0.05
0.05
0.00 0.00
SCG
SCG
0 5 10 0 5 10
(c) min (d) min
Figure 12.17 Chromatograms of SCG obtained at various stress conditions. (a) Standard
(no stress), (b) Oxidative (5% H2O2, 5 mL, 80°C, 5 h), (c) Photolytic-254 nm (24 h) and (d) Basic
0.1N NaoH, 5 mL, RT, 2 h). (Reprinted from Ali, M.S. et al., J. Sep. Sci., 31, 1645, 2008. With
permission from Wiley-VCH Verlag GmbH & Co. KGaA.)
methods have been shown as a promising approach to simplify the method develop-
ment.67–69 Risley and Pack developed a method for the simultaneous analysis of 12 cat-
ions and 21 anions using HILIC interfaced with an ELSD (Table 12.1).68 The method
used a zwitterionic HILIC column with a mixture of acetonitrile and an ammonium
formate buffer as the mobile phase, which gave satisfactory separation of all the ions.
It was believed that the analytes were separated by an ion-exchange mechanism. The
mobile-phase pH had significant effects on the retention of the ions in the range from
pH 3.1 to 6.6. The ionic interaction between the analytes and the stationary phase
dominated the separation mechanism; thus, increasing the pH resulted in a decreased
retention of anions and an improved retention of cations. For the same reason, buffer

concentrations affected the retention and the peak shape and a buffer concentration
ranging from 50 to 100â•›mM was recommended. The low buffer concentration caused
peak tailing, while the high concentration resulted in peak fronting. Satisfactory vali-
dation data including linearity, precision, and accuracy were obtained. The accuracy
for the majority of the ions was within ±2.5% of the theoretical values except for
PO4−3, which was believed to be impaired by the limited solubility in the organic
diluent. Limited solubility of some API salts in the organic-rich diluent is the limiting
factor for the application of the method. Huang et al. explored charged aerosol detec-
tion (CAD) for the analysis of inorganic counterions.69
A group of amino HILIC columns was also explored for the analysis of organic
counterions. Guo and Huang67 determined tromethamine (Figure 12.10) in an API
using an amino column interfaced with a refractive index (RID) detector. The effects
of ammonium acetate buffer on the separation were examined. The presence of
ammonium acetate increases the retention, probably due to the interaction between
the amino groups of the stationary phase and the negatively charged acetate ions
that reduce the repulsion between the analytes and the stationary phase. However,
increasing the buffer concentration caused the fronting of the analyte peak. The
authors attributed the phenomenon to the potential changes of the water-enriched
layer on the surface of the stationary phase that may be disturbed by the higher salt
concentration. As demonstrated in the analysis of other basic compounds, ZIC or
silica HILIC columns appear to be better choices for the task.4,68 The method was
validated with regard to the specificity, repeatability, linearity, and sensitivity. The
%RSD of six injections was 1.9%, and the detection limit was 0.03â•›mg/mL. The
application was demonstrated with an actual formulated API (Figure 12.18).
The method robustness was satisfied by slightly changing the mobile phase composi-
tion, flow rate, and column temperature. Three columns from the same manufacturer
were used for testing the repeatability, with about 10% variation observed in the
retention time and the separation efficiency.
12.6 Conclusions and Future Prospects

The above examples demonstrate the representative applications of HILIC in the anal-
ysis of polar impurities in various pharmaceutical products in pharmaceutical R&D.
When compared to its applications in the bioanalysis of drugs and drug metabolites,
HILIC is to some extent underutilized in pharmaceutical analysis.13 The majority of
the published applications appears to be in the R&D phases rather than in the com-
mercial production stage. As described earlier, HILIC has been a useful tool in the
analysis of polar impurities in polar APIs and intermediates as well as in the analysis
of polar impurities in nonpolar pharmaceuticals. In some cases, HILIC methods were
developed as orthogonal methods to the most popular RP-HPLC methods, either as
a substitution for ion-pairing methods or as mass spectrometry compatible methods.
Most small molecule drugs in pharmaceutical pipelines are nonpolar, and thus
RP-HPLC methods are well suited for their analyses. Typically, their impurity pro-
file methods are modified to accommodate all the impurities including the polar ana-
lytes whenever possible to minimize the need for an additional method. Thus, HILIC
is only used as a supplement method for the determination of very polar compounds

mAU
175 UV
150
125
100
75
50
25 Tromethamine
0
(a) 0 2 4 6 8 10 min
nRIU
15,000 Salt (RI detection) Tromethamine
10,000
5,000
0
–5,000
–10,000
(b) 0 2 4 6 8 10 min
nRIU
15,000 Free acid (RI detection)
10,000
5,000
0
–5,000
–10,000
(c) 0 2 4 6 8 10 min
Figure 12.18 Chromatograms for the tromethamine salt (a and b) and free acid (c) forms
of investigational drug, from UV (a) and RI detectors (b and c). Column: Zorbax NH 2,
4.6 × 150â•›m m, 5â•›m m particle size. Column temperature: 25°C. Mobile phase: acetonitrile/
water (80/20, v/v). Flow rate: 1â•›m L/min. Sample: the investigational API (salt form/2â•›mg/mL
in the mobile phase). Injection volume: 50â•›m L. (Reprinted from Guo, Y. and Huang, A.,
J. Pharm. Biomed. Anal., 31, 1191, 2003. With permission from Elsevier.)
when conventional RP-HPLC approaches fail to deliver the desired separation or

when facing other considerations such as mass spectrometry detection.4 Among the
process analytical methods to be transferred to manufacturing facilities, UV detec-
tion is the most popular detection method for HPLC analysis, while advanced tech-
niques such as mass spectrometry are undesirable. Therefore, mass spectrometry
compatible advantages of HILIC would not be a key factor regarding choosing either
an HILIC method or an ion-pairing RP-HPLC method, unless mass spectrometry
detection is used.
In light of the increasing attention on trace level genotoxic impurities in phar-
maceuticals, analyte-specific trace analysis methods are being developed regularly.
In order to achieve the required sensitivity and specificity, they are often mass

spectrometry-based approaches.70 HILIC/MS-based methods for the determina-

tion of genotoxic alkyl sulfonates and alkyl halide have been successfully devel-
oped. Furthermore, some of the low molecular weight polar genotoxic impurities
such as small aromatic amines (Figure 12.1) can be potentially analyzed by HILIC.
Therefore, increasing the usage of HILIC in pharmaceutical analytical development
and even in commercial settings can be foreseen.
HILIC stationary phases are of different types of chemical structures, and the
concerning polar analytes have diverse structures that could be acidic, basic, or
neutral molecules. Depending on the mobile-phase composition, the analytes not
only partition between the mobile phase and the water-rich layer on the surface of
the stationary phase, but also interact with the stationary phase directly.18 Thus, the
retention mechanism of analytes on stationary phases could be a combination of
hydrophilic, ionic, hydrogen bonding, and even hydrophobic factors.20 Depending
upon the conditions, one mechanism could dominate the others. Selecting the
stationary phase and mobile phase for the target analytes is critical to the separa-
tion and could be challenging during method development. A systematic under-
standing of the retention mechanisms of analytes on different stationary phases
has gained increased attention recently.71 Hao et al. suggested taking advantage of
the ion-exchange mechanism to separate ionizable analytes: selecting acidic col-
umns for cationic analytes and basic columns for anionic analytes, respectively.20
For example, the silica stationary phase can be used for separating bases, while
the amino column can be used for separating acids, although this may not always
work. Understanding of factors such as mobile-phase composition, buffer pH and
concentration, column temperature that affects the retention factor, efficiency, reso-
lution, peak shape, and peak tailing has been reviewed in the literature.17,18,20,41,71
Knowledge in terms of structure and retention relationship, however, remains very
limited. Column selectivity-based systematic column screening strategies that have
been adopted in pharmaceutical analysis for RP-HPLC method development have
not been reported for HILIC method development.72 Further understanding of reten-
tion mechanisms and factors are imperative, which should help facilitate HILIC
method development.
Acknowledgments
The authors would like thank Dr. Jianguo An, Lin Bai, and Josephine Vega for
providing
Â� analytical data and thoughtful discussions.
References
1. Liu, M. et al., J. Chromatogr. A, 2009, 1216, 2362.
2. Wang, X., Li, W., and Rasmussen, H.T., J. Chromatogr. A, 2005, 1083, 58.
3. Hmelnickis, J. et al., J. Pharm. Biomed. Anal., 2008, 48, 649.
4. Liu, M. et al., J. Chromatogr. A, 2008, 1188, 255.
5. International Conference on Harmonisation (ICH) Guidance for Industry: Impurities in
New Drug Substances Q3A (R2); U.S. Department of Health and Human Services, Food
and Drug Administration, Center for Drug Evaluation and Research (CDER): Rockville,
MD, June 2008, http://www.fda.gov/RegulatoryInformation/Guidances/ucm127942.htm

6. International Conference on Harmonisation (ICH) Guidance for Industry: Impurities

in New Drug Products Q3B (R2); U.S. Department of Health and Human Services,
Food and Drug Administration, Center for Drug Evaluation and Research (CDER):
Rockville, MD, July 2006, http://www.fda.gov/RegulatoryInformation/Guidances/
ucm128032.htm
7. International Conference on Harmonisation (ICH) Guidance for Industry: Impurities:
Guideline for Residue Solvents Q3C and Q3C(M); U.S. Department of Health and
Human Services, Food and Drug Administration, Center for Drug Evaluation and
Research (CDER): Rockville, MD, July 1997 and September 2002, http://www.fda.gov/
downloads/RegulatoryInformation/Guidances/ucm128317.pdf
8. Guideline on the Limits of Genotoxic Impurities, CPMP/SWP/5199/02, EMEA/CHMP/
QWP/251344/2006; Committee for Medicinal Products (CHMP), European Medicines
Agency (EMEA): London, 28 June 2006, http://www.emea.europa.eu/pdfs/human/
swp/519902en.pdf
9. Argentine, M.D., Owens, P.K., and Olsen, B.A., Adv. Drug Deliv. Rev., 2007, 59, 12.
10. Tsai, T.H., J. Chromatogr. B: Biomed. Sci. Appl., 2000, 747, 111.
11. Przybyciel, M. and Majors, R.E., LCGC North Am., 2002, 20, 516.
12. Majors, R.E., and Przybyciel, M., LCGC North Am., 2002, 20, 584.
13. Hsieh, Y., J. Sep. Sci., 2008, 31, 1481.
14. Ramos, R.G. et al., J. Chromatogr. A, 2008, 1209, 88.
15. Vervoort, N., Daemen, D., and Török, G., J. Chromatogr. A, 2008, 1189, 92.
16. de Villiers, A. et al., J. Chromatogr. A, 2007, 1161, 183.
17. Hemstroem, P. and Irgum, K., J. Sep. Sci., 2006, 29, 1784.
18. Ikegami, T. et al., J. Chromatogr. A, 2008, 1184, 474.
19. Ali, M.S. et al., Chromatographia, 2008, 67, 517.
20. Hao, Z., Xiao, B., and Weng, N., J. Sep. Sci., 2008, 31, 1449.
21. Jin, G. et al., Talanta, 2008, 76, 522.
22. Strege, M.A., Stevenson, S., and Lawrence, S.M., Anal. Chem., 2000, 72, 4629.
24. Boyman, L. et al., Biochem. Biophys. Res. Commun., 2005, 337, 936.
25. Beilmann, B. et al., J. Chromatogr. A, 2006, 1107, 204.
26. Risley, D.S., Yang, W.Q., and Peterson, J.A., J. Sep. Sci., 2006, 29, 256.
27. Genotoxic and Carcinogenic Impurities in Drug Substances and Products:
Recommended Approaches; U.S. Department of Health and Human Services, Food and
Drug Administration, Center for Drug Evaluation and Research (CDER): Silver Spring,
MD, December 2008; http://www.fda.gov/downloads/Drugs/Guidance-Compliance
RegulatoryInformation/Guidances/ucm079235.pdf
28. Delaney, E.J., Regul. Toxicol. Pharmacol., 2007, 49, 107.
29. Elder, D.P., Lipczynski, A.M., and Teasdale, A., J. Pharm. Biomed. Anal., 2008, 48, 497.
30. Elder, D.P., Teasdale, A., and Lipczynski, A.M., J. Pharm. Biomed. Anal., 2008, 46, 1.
31. Snodin, D.J., Regul. Toxicol. Pharmacol., 2006, 45, 79.
32. Teasdale, A. et al., Org. Process Res. Dev., 2009, 13, 429.
33. Jacq, K. et al., J. Pharm. Biomed. Anal., 2008, 48, 1339.
34. Li, W., J. Chromatogr. A, 2004, 1046, 297.
35. Ramjit, H.G., Singh, M.M., and Coddington, A.B., J. Mass Spectrom., 1996, 31, 867.
36. Raman, N.V.V.S. et al., J. Pharm. Biomed. Anal., 2008, 48, 227.
37. Ramakrishna, K. et al., J. Pharm. Biomed. Anal., 2008, 46, 780.
38. Salamoun, J., Nguyen, P.T., and Remien, J., J. Chromatogr., 1992, 596, 43.
40. Guo, Y., J. Liq. Chromatogr. Relat. Technol., 2005, 28, 497.
41. McKeown, A.P. et al., J. Sep. Sci., 2001, 24, 835.
42. Weng, N., J. Chromatogr. B, 2003, 796, 209.

43. An, J. et al., J. Pharm. Biomed. Anal., 2008, 48, 1006.

44. Bai, L. et al., J. Chromatogr. A, 2010, 1217, 302.
45. Sun, M. et al., J. Pharm. Biomed. Anal., 2010, 52, 30.
46. Hemminki, K. et al., Arch. Toxicol., 1987, 61, 126.
47. Watson, E., Dea, P., and Chan, K.K., J. Pharm. Sci., 1985, 74, 1283.
48. Huang, M. et al., Anal. Sci., 2005, 21, 1343.
50. Yang, L. et al., J. Pharm. Biomed. Anal. 2000, 22, 487.
51. Kosovec, J.E. et al., Rapid Commun. Mass Spectrom., 2008, 22, 224.
52. Pisano, R. et al., J. Pharm. Biomed. Anal., 2005, 38, 738.
53. Kamleh, M.A. et al., FEBS Lett., 2008, 582, 2916.
54. Nicholls, G., Clark, B.J., and Brown, J.E., J. Pharm. Biomed. Anal., 1992, 10, 949.
55. Li, R. and Huang, J., J. Chromatogr. A, 2004, 1041, 163.
56. Dong, L. and Huang, J., Chromatographia, 2007, 65, 519.
57. Guisbert, A.L. et al., J. Liq. Chromatogr. Relat. Technol., 2000, 23, 1019.
58. Risley, D.S. and Strege, M.A., Anal. Chem., 2000, 72, 1736.
59. Ozoux, M.L. et al., J. Chromatogr. B: Biomed. Sci. Appl., 2001, 765, 179.
60. Ali, M.S. et al., J. Sep. Sci., 2008, 31, 1645.
61. Strege, M. et al., LCGC North Am., 2008, 26, 632.
62. Schram, K.H., Taniguchi, Y., and McCloskey, J.A., J. Chromatogr., 1978, 155, 355.
63. Schilsky, R.L. and Ordway, F.S., J. Chromatogr. Biomed. Appl., 1985, 337, 63.
64. Wang, L. et al., J. Pharm. Biomed. Anal., 2003, 33, 955.
65. Wiliams, R.C. et al., J. Pharm. Biomed. Anal., 1997, 16, 469.
66. Kotinkaduwe, R.P. and Kitscha, R.A., J. Pharm. Biomed. Anal., 1999, 21, 105.
68. Risley, D.S. and Pack, B.W., LCGC North Am., 2006, Suppl., 82, 776–785.
69. Huang, Z. et al., J. Pharm. Biomed. Anal., 2009, 50, 809.
70. Borman, P.J. et al., J. Pharm. Biomed. Anal., 2008, 48, 1082.
71. Dejaegher, B., Mangelings, D., and Vander Heyden, Y., J. Sep. Sci., 2008, 31, 1438.
72. Krisko, R.M. et al., J. Chromatogr. A, 2006, 1122, 186.

13 Fast In-Process Method
for the Determination
of Ioversol and Related
Polar Compounds by
Chromatography
(HILIC) and UPLC
Yuming Chen, Xinqun Huang,
Shaoxiong Huang, and Michael Matchett
Contents
13.1 Overview of HILIC/UPLC............................................................................ 291
13.1.1 HILIC and HILIC Stationary Phases................................................ 291
13.1.2 Overview of UPLC and HILIC-UPLC Applications........................ 293
13.2 Case Study: Fast In-Process Analysis of Ioversol by HILIC-UPLC............. 295
13.2.1 Introduction....................................................................................... 295
13.2.2 Experimental..................................................................................... 296
13.2.3 Results and Discussion...................................................................... 297
13.2.3.1 Univariate Method Development........................................ 297
13.2.3.2 Multivariate Method Optimization.....................................300
13.2.4 Conclusions........................................................................................304
References...............................................................................................................304
13.1 Overview of HILIC/UPLC

13.1.1 HILIC and HILIC Stationary Phases
HILIC is used primarily for the separation of very polar compounds that are poorly
retained by reversed-phase liquid chromatography (RPLC).1–23 Alpert24 described
HILIC as a variant of normal-phase liquid chromatography (NPLC), where the sta-
tionary phase is more polar than the mobile phase, and analytes typically elute in an
291

order opposite to that of RPLC. Different from traditional NPLC, a HILIC system
allows for water as well as buffering agents to be present in the mobile phase. It is
commonly believed that in HILIC separations, water in the mobile phase is preferen-
tially adsorbed to the polar stationary phase, resulting in a layer of liquid enriched in
water near the stationary phase surface vs. a layer of liquid enriched in organic in the
bulk mobile phase. Analytes can then partition between these two layers to achieve
chromatographic separation. Polar compounds have higher solubility in the water layer
and are thus retained longer than less polar compounds. The retention mechanism may
include hydrogen bonding and electrostatic interactions between the analytes and func-
tional groups on the surface of the stationary phase. Therefore, the separation mecha-
nisms in HILIC may be characterized as mixed modes of partition and absorption.
Compared with RPLC, HILIC provides an orthogonal separation that can be
valuable in many applications, particularly for complex mixtures. In addition, HILIC
often uses more volatile mobile phases relative to RPLC, so it offers higher sensitiv-
ity when combined with mass spectrometer detection. 5,25,26
In recent years, an increasing number of HILIC stationary phases (SPs) have
become commercially available. The selection of a proper column is the first step to
assure a successful analytical method. A brief overview is given here to categorize
the most widely used HILIC SPs.
Bare silica: Bare silica is one of the classical and still often used HILIC SPs.
Zorbax SIL,27 Supelcosil LC-Si,28 Hypersil Silica,29,30 Nucleosil SIL, and Atlantis
HILIC27,31–35 all belong to this category. The siloxane and silanol groups, in some
cases with trace amounts of metal ions on the surface of the silica particles, provide
a polar surface to retain polar compounds. The operating pH range for such SPs
is usually 2–6. These SPs may offer unique selectivity,36 but in some cases bear
the common problem of poor reproducibility. Columns with 5,37 4,38 3,31,32,39–43 and
1.7â•›μm 2–4,9,14,44 particle sizes are commercially available. Bare silica columns are
mainly used for the separation of small polar compounds.
Polar neutral bonded silica: The most commonly used SPs in this group include
diol- and amide-modified silica.
Diol SPs usually provide improved reproducibility and different selectivity when
compared with bare silica (e.g., YMC-Pack Diol,45 Uptisphere Diol, Nucleosil Diol).
Because hydrogen bonding with the diol functional groups is not as strong as with
the silanol groups on a bare silica surface, the retention is usually less with diol SPs.
Diol columns are mostly used for the separation of small organic compounds.46,47
Amide-bonded SPs are one of the most popular SPs in HILIC separation (e.g.,
TSK-Gel Amide 80). The columns are packed with spherical silica-based material to
which carbamoyl functional groups are covalently bonded. These SPs have been used
to separate oligosaccharides,48–51 peptides,52–55 and small organic molecules.18,56–58
The TSK-Gel Amide 80 is also available in a 3â•›μm particle size.59
Silica-based ion-exchange SPs: Ion-exchange SPs can be further categorized into
positively charged and negatively charged SPs.
Amine-bonded silica columns, such as YMC-Pack NH2, Luna Amino, Alltima
Amino, Zorbax NH2, and Hypersil NH2 columns, are typical positively charged

Fast In-Process Method for the Determination of Ioversol 293
columns and have been used for the separation of carbohydrates.60,61 They are not
recommended for peptide separations due to low analyte recovery.62 In some cases,
these SPs demonstrated poor reproducibility and short column life because of chemical
instability of the amine group.
Negatively charged cation-exchange columns have been successfully used for
peptide and protein separations. PolyCAT A column, a weak cation-exchange column
formed by poly(aspartic acid) bonded to silica, was reported for the separation of phos-
phorylated H1 histone.63 Synchropak CM 300, a weak cation-exchange column with
carboxymethyl functionalities, was reported for the successful separation of histone
with different degrees of acetylation.64 Polysulfoethyl A, a strong cation-exchange
column, was used for peptide separations.65–67
Zwitterionic SPs: Over the past few years, HILIC columns have progressed to sec-
ond- and third-generation embodiments, most of which involve mixed or multiple-
interaction modes. A good example is the zwitterion SPs, which has sulfoalkylbetaine
functional groups that strongly absorb water by hydrogen bonding. This type of SPs
offers multiple analyte-SP interactions including hydrogen bonding, dipole–dipole
interactions, and electrostatic interactions for the separation of charged and neutral
compounds. Zwitterionic SPs are commercially available either on silica gel support
(e.g., ZIC HILIC) or polymer support (e.g., ZIC pHILIC). ZIC HILIC columns have
demonstrated impressive effectiveness in separations of complicated protein68 and
peptide samples.69–71 They have also become one of the most popular choices for
small organic molecules.72–75
13.1.2 Overview of UPLC and HILIC-UPLC Applications

There is always a drive to pursue high efficiency in the field of liquid chromatog-
raphy (LC). The underlying principle of LC can be explained by the van Deemter
equation, which describes the kinetics of a chromatographic separation19 using
three terms:
B
H = A+ +C × u
u
where
H is the plate height (column efficiency)
A is the Eddy diffusion
B is the longitudinal diffusion
C is the mass transfer kinetics of the analyte between the mobile and station-
ary phase
u is the linear velocity
Among these three terms, the A term is proportional to dp (particle size of packing
material), whereas the C term is proportional to d 2p. Based on the van Deemter equa-
tion, high column efficiency can be attained over a wide range of linear velocity if
the particle size of the packing is smaller than 2â•›μm (Figure 13.1). Ultra-performance

HPLC
30
25 10 μm
20
5 μm
HETP: μm
15
10
1.7 μm
5
0
0 1 2 3 4 5 6 7
Linear velocity: mm/s
Figure 13.1 van Deemter plots showing relationships between HETP and the linear
velocity of the mobile phase for columns with stationery phases of different particle sizes.
liquid chromatography (UPLC) has been developed to take advantage of these chro-
matographic principles, so the separation can be conducted using sub-2â•›μm columns
at higher linear velocity, which results in increased speed for the separation with
superior resolution and higher sensitivity.
Currently, commercially available UPLC systems include the Waters Acquity
UPLC system, the Agilent 1200 series Rapid Resolution liquid chromatographic sys-
tem, the Shimadzu UFLC system, and the Thermo Fisher Accela UHPLC system.
All these systems are built specifically for delivering accurate flow at very high back
pressure (up to 18,000â•›psi).
Literature reports on HILIC-UPLC applications are limited simply because of
the limited availability of suitable columns. HILIC-UPLC methods so far are based
on the bridged ethylene hybrid (BEH) SP.2–4,9,14,20 The BEH HILIC column is packed
with underivatized 1.7â•›μm BEH particles. Because the ethylene bridged groups are
embedded in the silica matrix, nearly one-third of the surface silanols are removed,
therefore causing decreased retention attributed to the reduced number of silanols.76
Speed is one of the major drivers for HILIC-UPLC applications. Kesiunaite et al.9
reported the separation of cabadox and olaquindox, two synthetic antibacterial
drugs for animal feeding, within 1â•›min on a 1.7â•›μm BEH HILIC column by isocratic
elution with a mobile phase of 10â•›m M ammonium acetate in acetonitrile-water
(95:5, v:v). The resolution was comparable to that obtained by an HPLC method,77 but
the separation was 10 times faster. Demacker et al.3 developed an UPLC-HILIC-MS
method to analyze citrulline, a nonessential amino acid in plasma samples. With a
simple protein precipitation/extraction step, 12 samples could be analyzed in an hour
by the method.
HILIC-UPLC has also been utilized for the mapping of complex biosamples. Gika
et al.4 evaluated a HILIC-UPLC-MS method for profiling metabolites in Zucker rat
urine samples. Both HILIC-UPLC and RP-UPLC methods were used to generate
complementary profiles. HILIC conditions were optimized to separate highly polar
metabolites found in these samples, which included simple organic acids, amino

acids, and amphoteric basic compounds. Compared to the RPLC profiles, the HILIC
profiles showed higher signal intensity in total ion current (TIC) chromatograms, pre-
sumably due to higher ionization efficiency. The HILIC-UPLC-MS results revealed
different biomarker information from the RP-UPLC data and these biomarkers
were from polar metabolites of the urine samples. Such findings demonstrate the
value of HILIC as an orthogonal method to RPLC by providing critical information
in biomarker identification and characterization for complex biosamples.
In other HILIC-UPLC applications, Chen et al.2 compared HPLC with a BetaMax
acid column and HILIC-UPLC with a BEH HILIC column for the detection of 10
haloacetic acids in drinking water. The HILIC-UPLC method provided lower on-
column detection limits. New and Chan14 evaluated BEH C18, BEH HILIC, and the
high-strength silica (HSS) T3 C18 column for the analysis of glutathione, glutathi-
one disulfide, and ophthalmic acid. A gradient elution was used along with MS/MS
detection. However, BEH HILIC did not show an advantage over the other SPs.
13.2 C
ase Study: Fast In-Process Analysis
of Ioversol by HILIC-UPLC
13.2.1 Introduction
Ioversol is an x-ray imaging enhancing reagent widely used in clinical diagnostics,
such as angiocardiography,78 urography,79 herniography,80 and arthrography.81 The
chemical structure of Ioversol is shown in Figure 13.2. Each Ioversol molecule con-
tains three iodine atoms, which are responsible for x-ray opacity. Ioversol is highly
water-soluble, and its safety has been demonstrated in a number of animal and clini-
cal studies.82–85 In clinical applications, a single dose of Ioversol can be as high as
40â•›g,86 and the same dosage can be repeated if needed.80 Because of the high dosing
amount, a stringent control of the impurity levels is required to ensure the safety of
the product.
During the large-scale production of Ioversol, up to seven process impurities were
monitored and controlled. Speed and specificity are the two most important attributes
OH
H OH
O N
I I
OH
O
H OH
HO N
N
I O
OH Ioversol
C18H24N3O9I3 807.11 g/mol
Figure 13.2 Structure of Ioversol.

for an in-process method. Ioversol and its process impurities are highly polar ana-
logues, which are difficult to separate by RPLC. Assay87 and inprocess88 methods for
Ioversol based on RPLC have been reported, but proved to be unsuitable for resolv-
ing all the impurities in our case. So a method using HILIC-UPLC was developed
and optimized to separate Ioversol and its process impurities in less than 10â•›min. To
the author’s knowledge, the separation of Ioversol and its related impurities using a
HILIC-UPLC method has not been attempted before.
The HILIC-UPLC method was developed and optimized by both univariate and
multivariate approaches. Method parameters including buffer concentration, aceto-
nitrile percentage, temperature, and flow rate were investigated in univariate ways.
Based on the results of univariate studies, the multivariate approach was used for
the first time89 for a HILIC method to demonstrate the correlations among different
parameters.
13.2.2 Experimental
Reagents and materials: HPLC grade acetonitrile was purchased from Fisher
Scientific (Fair Lawn, NJ). Acetic acid (Glacial) was purchased from J. T. Baker
(Philipsburg, NJ). HPLC grade water was obtained from an in-house Milli-Q water
purification system (Millipore, Bedford, MA).
Standards of Ioversol and seven process impurities (RS-1 to RS-7) were supplied
by Covidien’s Department of Imaging Research and Development. Ioversol con-
sists of several stereo-isomers90 and often shows as a group of unresolved peaks in
chromatographic analysis. The process impurities are structural analogs of Ioversol.
Each of them also contains three iodine atoms, and some of them also have one
or more structural isomers. The functional groups and physical properties of these
compounds are listed in Table 13.1.
Preparation of solutions: The stock solution for each process impurity was prepared
at a concentration of 0.5â•›mg/mL in 95% acetonitrile. About 0.2â•›mg of Ioversol stan-
dard and 1â•›m L of each impurity stock solution were added into the same 50â•›m L
Table 13.1
Ioversol and Its Process Impurities
Name Molecular Weight Functional Group pKa
Ioversol 807 –OH (6) 11.35
RS-1 763 –OH (5) 11.48
RS-2 705 –OH (4) 11.58
Aromatic –NH2 (1)
RS-3 791 –OH (5) 11.35
RS-4 807 –OH (5) 11.58
RS-5 851 –OH (7) N/A
RS-6 747 –OH (4) N/A
RS-7 781 –OH (4) N/A
–Cl (1)

volumetric flask and diluted to volume with 95% acetonitrile. This resolution solu-
tion contained about 0.4â•›mg/mL of Ioversol and 0.01â•›mg/mL of each impurity.
UPLC conditions: An Acquity UPLC system (Waters, Milford, MA) equipped
with a binary solvent delivery system, an autosampler, and an Acquity photodiode
array detector was used. System control and data processing were performed by the
Empower 2 software (Waters).
An Acquity BEH HILIC column (50â•›mm × 2.1â•›mm I.D., 1.7â•›μm particle size) was
used for the separation. The optimized method has a column temperature of 35°C, a
flow rate of 0.4â•›mL/min, and a mobile phase composed of acetonitrile:water (95:5, v/v)
with 10â•›m M ammonium acetate. The injection volume was 2â•›μL using a partial
loop with the needle overfill injection mode. Data was collected at 254â•›nm.
13.2.3 Results and Discussion

In the method development stage, a one-variable-at-a-time approach was used. Four
parameters including acetonitrile concentration, buffer concentration, column tem-
perature, and flow rate were investigated to identify a suitable range for each of
the parameters. In the method optimization stage, three parameters including buffer
concentration, flow rate, and column temperature were selected for further optimiza-
tion using a multivariate design-of-experiment approach.
13.2.3.1 Univariate Method Development
Acetonitrile concentration: The effect of acetonitrile concentration on the retention of
Ioversol and its impurities was investigated by varying the percentage of acetonitrile in
the mobile phase while keeping the ammonium acetate concentration (10â•›mM), flow rate
(0.5â•›mL/min), and column temperature (25°C) constant. As can be seen in Figure 13.3,
all analytes exhibited the typical retention behavior of a HILIC system, i.e., decreas-
ing retention with increasing water content in the mobile phase. The retention was
8 Retention time vs. acetonitrile percentage
7 Ioversol
RS-1
6 RS-2
Retention time: min
RS-3
5
RS-4
4 RS-5
RS-6
3 RS-7
0
75 80 85 90 95 100
ACN%
Figure 13.3 Retention time vs. acetonitrile concentration in the mobile phase.

poor for all components in the acetonitrile range of 80%–90%. Significant increases in
retention were observed as the acetonitrile concentration increased from 90% to 95%.
Similar retention behavior was also observed by Grumbach et al.5 who evaluated the
performance of BEH HILIC columns. The mobile phases with greater than 95% ace-
tonitrile were not tested because about 5% of water was needed to maintain the desired
buffer concentration. The best selectivity was also observed with 95% acetonitrile in
the mobile phase.
Buffer concentration: Buffer (or salt) is an essential component in a HILIC method.
A buffer solution is needed even though the analyte(s) of interest may not have an ioniz-
able functional group. A suitable buffer is also critical for achieving desirable efficiency
and sometimes selectivity. In some cases, the choice of a buffer or salt is limited to its
solubility in the high-organic-content mobile phases. In developing the HILIC method
for Ioversol and its impurities, ammonium acetate was selected for its relatively high
solubility in 95% acetonitrile. To study the effect of buffer concentration on retention,
the ammonium acetate concentration was varied from 2.5 to 15â•›mM while holding
the flow rate (0.5â•›mL/min), column temperature (25°C), and acetonitrile concentration
(95%) constant. Buffer concentration was found to have a significant impact on both
the retention and selectivity of the analytes. As shown in Figure 13.4, the logarithms of
retention for Ioversol and its impurities vs. buffer concentration show a linear relation-
ship. More interestingly, there is a difference in slope among these linear plots, indicat-
ing that the variation in buffer concentration causes a change in selectivity. Ioversol,
RS-1, and RS-5 show steeper slopes compared with the rest. Based on the hypothesis
that ammonium acetate prefers the water-enriched surface layer of the polar stationary
phase, increasing the ammonium acetate levels thereby promotes the formation of the
water-rich layer, thus increasing the retention of polar analytes. Because Ioversol and its
impurities did not retain well at low buffer concentrations, higher levels of 10, 12.5, and
15â•›mM ammonium acetate were later investigated for method optimization.
Flow rate: The effect of flow rate on column efficiency for this group of compounds
is presented in Figure 13.5. Compared with the van Deemter plot in Figure 13.1, the
1.2 Log (retention time) vs. buffer concentration
1 Ioversol
RS-1
0.8
Log (retention time)
RS-2
RS-3
0.6
RS-4
0.4 RS-5
RS-6
0.2 RS-7
0
0 2 4 6 8 10 12 14
–0.2
Buffer concentration: mM
Figure 13.4 Effect of buffer concentration on retention.

45.00 HETP vs. linear velocity
40.00
35.00 Ioversol
30.00 RS-1
HETP: μm
25.00 RS-2
RS-3
20.00
RS-4
15.00 RS-5
10.00 RS-6
RS-7
5.00
0.00
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
Linear velocity: mm/s
Figure 13.5 Effect of linear velocity on column efficiency.

abnormal shape and trend of the Ioversol curve were due to the partial separation of
Ioversol stereo-isomers. Curves of six out of the seven impurities show characteristics
of HETP-flow relationships typically observed with 5â•›μm particle size HPLC columns.
The curve of RS-4 is closer to the 1.7â•›μm curve in Figure 13.1 with a steeper slope. This
observation may be related to the presence of multiple retention mechanisms in the
HILIC mode, i.e., partition as well as hydrogen bonding interactions, which may slow
down mass transfer between the SP and the mobile phase.
Column temperature: The effect of column temperature on retention and column
efficiency was explored across the range of 15°C–55°C while holding the other con-
ditions constant. Column temperature had a significant impact on both retention
and selectivity (Figure 13.6). RS-1 and RS-3 reversed their elution order when the
Retention time vs. temperature
10
9 Ioversol
RS-1
8
RS-2
7
Retention time: min
RS-3
6 RS-4
RS-5
5
RS-6
4 RS-7
3
2
1
0
10 20 30 40 50 60
Temperature: °C
Figure 13.6 Effect of column temperature on retention time.

60.00 HETP vs. temperature

Ioversol
RS-1
50.00
RS-2
RS-3
40.00
Retention time: min
RS-4
RS-5
RS-6
30.00
RS-7
20.00
10.00
0.00
10 20 30 40 50 60
Temperature: °C
Figure 13.7 Relationship between column temperature and column efficiency.
temperature increased from 15°C to 25°C. Ioversol and RS-4 co-eluted at 55°C but
were well separated at lower temperatures. Temperature may affect the separation in
multiple ways. Higher temperature facilitates mass transfer, which increases column
efficiency. On the other hand, higher temperature increases longitudinal diffusion,
which decreases column efficiency. Secondary interactions, such as hydrogen bond-
ing and electrostatic interactions between analytes and the stationary phase may be
reduced at high temperature, which then changes the selectivity. Figure 13.7 shows
that maximum column efficiency was achieved around 45°C for Ioversol and some of
the impurities. The impact of column temperature on column efficiency was further
analyzed using multivariate analysis.
13.2.3.2 Multivariate Method Optimization

The one-variable-at-a-time studies were necessary to identify suitable operating ranges
for the four method parameters investigated. Figure 13.4 implies that acetonitrile con-
centration in the mobile phase should be kept at 95% to obtain sufficient retention for
this group of compounds. The buffer concentration may be varied from 10 to 15â•›mM,
whereas column temperature showed a promising range between 25°C and 35°C based
on Figures 13.6 and 13.7 considering both selectivity and column efficiency. On the other
hand, these results did not demonstrate relationships among those parameters, and the
relative importance of these factors for the method was also unknown. Therefore, a mul-
tivariate method optimization study was conducted to further optimize the method.
A simple three-factor by two-level design with an additional center point (Table 13.2)
was used. After execution of the designed experiments, multiple responses were
collected, which included the resolution of all peaks, the column efficiency for all com-
ponents, and the tailing factor for all components. To limit the discussions within the
scope of this article, only two examples regarding the data analysis are given here.

Fast In-Process Method for the Determination of Ioversol
Table 13.2
Factors in Designed Experiments and Responses
Factors Response 1 Response 2—Peak Width (min)
Temperature Buffer Flow Rate Resolution
Run (°C) (mM) (mL/min) (Peaks Resolved) RS-2 RS-7 RS-6 RS-3 RS-1 RS-4 Ioversol RS-5
1 25 10 0.4 Most 0.095 0.098 0.163 0.168 0.171 0.158 0.241 0.323
2 35 10 0.4 Some 0.049 0.051 0.091 0.091 0.086 0.165 0.129 0.165
3 25 15 0.4 Some 0.133 0.133 0.228 0.235 0.287 0.225 1.030 0.490
4 35 15 0.4 All 0.064 0.068 0.111 0.114 0.119 0.113 0.243 0.228
5 25 10 0.6 Some 0.068 0.070 0.116 0.120 0.122 0.112 0.153 0.228
6 35 10 0.6 Some 0.037 0.039 0.060 0.068 0.063 0.064 0.093 0.123
7 25 15 0.6 Some 0.094 0.100 0.171 0.167 0.197 0.173 0.691 0.367
8 35 15 0.6 All 0.049 0.051 0.082 0.084 0.088 0.084 0.119 0.164
9 30 12.5 0.5 Most 0.068 0.071 0.117 0.119 0.126 0.115 0.190 0.232
301
0.018
Ioversol
0.016
0.014
0.012
RS-7
0.010 RS-2
AU
RS-6
0.008
RS-4
RS-3
RS-1
0.006
0.004
RS-5
0.002
0.000
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
min
Figure 13.8 Typical chromatogram showing the separation of Ioversol from seven process
impurities.
Data analysis by visual inspection of results: The first example involved tabulating
the resolution results and analyzing them by visual inspection. The resolution results
under different experimental conditions are summarized in Table 13.2. A complete
resolution of all components was achieved in two out of nine runs. A chromatogram
showing the separation is presented in Figure 13.8.
Multivariate data analysis: The second example of data analysis utilized a partial
least squares (PLS) algorithm. In this case, experimental condition values were used
as x-variables (n = 3) and the peak width values (Table 13.2) for all components mea-
sured at half peak height were used as y-variables (n = 8). The data was normalized
using the equation of 1/(standard deviation). Results of the PLS modeling are graphi-
cally presented in Figure 13.9. The PLS algorithm uses latent variables to describe
variances within the two sets of experimental data. A latent variable may repre-
sent a particular correlation that can have specific physical meanings. The residual
validation variance plot (bottom-left) in Figure 13.9 indicates that only one latent
variable is needed to account for 87% of the variances in the y-variables. The cor-
relation loadings plot (top-right), characterized by two ellipses, demonstrates the correla-
tion between the x- and y-variables. The outer ellipse is the unit circle and indicates
100% explained variance. The inner ellipse indicates 50% of explained variance.
The x- and y-variables that are closest to +1 (positive correlation) or −1 (negative
correlation) along the PC1 axis have the strongest correlation. The most significant
correlations are observed between temperature (in blue) and the eight y-variables
(in red). In other words, column temperature plays the most important role in defin-
ing peak width among the studied parameters and within the experimental condi-
tions. The other two plots in Figure 13.9 also support this conclusion. The scores

Fast In-Process Method for the Determination of Ioversol
Scores Correlation loadings (x and y)
PC2 PC2
2 1.0
1 0.5
0 0
–1 –0.5
–2 –1.0
PC1 PC1
–2.0 –1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 –1.0 –0.5 0 0.5 1.0
Efficiency_PLS2, X-expl: 33%, 33% Y-expl: 87%, 2% Efficiency_PLS2, X-expl: 33%, 33% Y-expl: 87%, 2%
Y-variance Residual validation variance Predicted Y
1.0 0.15
Elements: 9
Slope: 0.937822
Offset: 0.004539
0.8 Correlation: 0.968412
0.10 R-Square: 0.937822
RMSEC: 0.007006
SEC: 0.007431
0.6 Bias: –3.104e–09
0.05
0.4
0.2 0
PCs Measured Y
PC 00 PC 01 PC 02 PC 03 0.02 0.04 0.06 0.08 0.10 0.12 0.14
Efficiency_PLS2, Variable: v. Total Efficiency_PLS2, (Y-var, PC): (RS-2, 1)
303
Figure 13.9 Multivariate data analysis using a PLS model.

plot (top-left) shows a pattern that is highly column-temperature dependent along the
PC1 axis. The measured vs. predicted peak width plot (bottom-right) demonstrates a
quantitative relationship between the measured and predicted peak widths as a func-
tion of column temperature. Even though the plot shows a curvature, a correlation
coefficient of 0.968 and an R2 of 0.938 were obtained for impurity RS-2. Similar
correlation and trends were observed for the other seven compounds (not presented
here). For simplicity of data analysis, the peak width data was not corrected for reten-
tion time differences, which may cause the curvature in the plots.
The above analyses effectively identified the optimized HILIC conditions, and
revealed the controlling factors with regard to one study outcome. The approach can
also be used for studying method robustness, which is sometimes a more serious
concern for a HILIC method.
13.2.4 Conclusions
Method development and optimization were conducted by a two-stage process.
The univariate approach defined the suitable operation ranges for four of the most
important method parameters. The experimental results showed that small changes
in any one of the parameters (acetonitrile concentration, buffer concentration, or col-
umn temperature) would cause a significant change in retention, column efficiency,
and resolution. The design-of-experiment approach was effective for the identifica-
tion of the conditions to achieve complete separation of all components. The chro-
matographic conditions were optimized using a flow rate of 0.4â•›m L/min, a column
temperature at 35°C, and with a mobile phase of 15â•›m M ammonium acetate in 95%
acetonitrile.
References
1. L.T. Andersen, H. Schlichtherle-Cerny, Y. Ardo, Dairy Science & Technology 88
(2008) 467.
2. C.-Y. Chen, S.-N. Chang, G.-S. Wang, Journal of Chromatographic Science 47
(2009) 67.
3. P.N.M. Demacker, A.M. Beijers, H. van Daal, J.P. Donnelly, N.M.A. Blijlevens, J.M.W.
van den Ouweland, Journal of Chromatography, B: Analytical Technologies in the
Biomedical and Life Sciences 877 (2009) 387.
4. H.G. Gika, G.A. Theodoridis, I.D. Wilson, Journal of Separation Science 31 (2008)
1598.
5. E.S. Grumbach, D.M. Diehl, U.D. Neue, Journal of Separation Science 31 (2008) 1511.
6. J. Guitton, S. Coste, N. Guffon-Fouilhoux, S. Cohen, M. Manchon, M. Guillaumont,
Journal of Chromatography, B: Analytical Technologies in the Biomedical and Life
Sciences 877 (2009) 149.
7. D.N. Heller, C.B. Nochetto, Rapid Communications in Mass Spectrometry 22 (2008)
3624.
8. E.P. Kadar, C.E. Wujcik, Journal of Chromatography, B: Analytical Technologies in the
Biomedical and Life Sciences 877 (2009) 471.
9. G. Kesiunaite, E. Naujalis, A. Padarauskas, Journal of Chromatography, A 1209 (2008) 83.
10. L. Kovalova, C.S. McArdell, J. Hollender, Journal of Chromatography, A 1216 (2009)
1100.

11. H.-A. Lakso, P. Appelblad, J. Schneede, Clinical Chemistry (Washington, DC) 54

(2008) 2028.
12. S.-M. Liang, D.-M. Liu, W.-J. Zeng, Y. Zeng, Y.-H. Luo, Xiandai Shipin Keji 24 (2008)
1180.
13. K.T. Myint, K. Aoshima, S. Tanaka, T. Nakamura, Y. Oda, Analytical Chemistry
(Washington, DC) 81 (2009) 1121.
14. L.-S. New, E.C.Y. Chan, Journal of Chromatographic Science 46 (2008) 209.
15. R. Oertel, N. Arenz, J. Pietsch, W. Kirch, Journal of Separation Science 32 (2009) 238.
16. M. Strege, C. Durant, J. Boettinger, M. Fogarty, LC-GC Europe 21 (2008) 608.
17. Y. Sun, J. Sun, J. Liu, S. Yin, Y. Chen, P. Zhang, X. Pu, Y. Sun, Z. He, Journal of
Chromatography, B: Analytical Technologies in the Biomedical and Life Sciences 877
(2009) 649.
18. E. Turrell, L. Stobo, J.-P. Lacaze, S. Piletsky, E. Piletska, Journal of AOAC International
91 (2008) 1372.
19. J.J. van Deemter, F.J. Zuiderweg, A. Klinkenberg, Chemical Engineering Science 5
(1956) 271.
20. J.-H. Wang, X.-Y. Lu, M. Huang, Z. Wu, G.-P. Ma, C.-Y. Xu, Fenxi Huaxue 35 (2007)
1509.
21. D. Xie, J. Mattusch, R. Wennrich, Engineering in Life Sciences 8 (2008) 582.
22. L. Yan, M. Wu, Z. Zhang, Y. Zhou, L. Lin, E. Fang, D. Xu, L. Chen, Sepu 26
(2008) 759.
23. N.L. Young, M.D. Plazas-Mayorca, B.A. Garcia, American Biotechnology Laboratory
27 (2009) 23.
24. A.J. Alpert, Journal of Chromatography 499 (1990) 177.
25. E.S. Grumbach, D.M. Diehl, J.R. Mazzeo, LC-GC Europe 22 (2004) 1010.
26. F. Qin, Y.-Y. Zhao, M.B. Sawyer, X.-F. Li, Analytical Chemistry (Washington, DC) 80
(2008) 3404.
27. V. Goekmen, H.Z. Senyuva, J. Acar, K. Sarioglu, Journal of Chromatography, A 1088
(2005) 193.
28. M.S. Ali, S. Rafiuddin, M. Ghori, A.R. Khatri, Chromatographia 67 (2008) 517.
29. Z. Hao, C.-Y. Lu, B. Xiao, N. Weng, B. Parker, M. Knapp, C.-T. Ho, Journal of
Chromatography, A 1147 (2007) 165.
30. Y. Hsieh, J. Chen, Rapid Communications in Mass Spectrometry 19 (2005) 3031.
31. T.M. Baughman, W.L. Wright, K.A. Hutton, Journal of Chromatography, B: Analytical
Technologies in the Biomedical and Life Sciences 852 (2007) 505.
32. D.W. Jeong, Y.H. Kim, H.Y. Ji, Y.S. Youn, K.C. Lee, H.S. Lee, Journal of Pharmaceutical
and Biomedical Analysis 44 (2007) 547.
33. T. Ohmori, M. Nakamura, S. Tada, T. Sugiyama, Y. Itoh, Y. Udagawa, K. Hirano, Journal
of Chromatography, B: Analytical Technologies in the Biomedical and Life Sciences 861
(2008) 95.
34. H. Zhang, Z. Guo, F. Zhang, Q. Xu, X. Liang, Journal of Separation Science 31 (2008)
1623.
35. Y. Iwasaki, M. Hoshi, R. Ito, K. Saito, H. Nakazawa, Journal of Chromatography, B:
Analytical Technologies in the Biomedical and Life Sciences 839 (2006) 74.
36. B.A. Olsen, Journal of Chromatography, A 913 (2001) 113.
37. W. Naidong, Journal of Chromatography, B: Analytical Technologies in the Biomedical
and Life Sciences 796 (2003) 209.
38. M. Godejohann, Journal of Chromatography, A 1156 (2007) 87.
39. H.Y. Ji, D.W. Jeong, Y.H. Kim, H.-H. Kim, Y.-S. Yoon, K.C. Lee, H.S. Lee, Rapid
Communications in Mass Spectrometry 20 (2006) 2127.
40. H.Y. Ji, D.W. Jeong, Y.H. Kim, H.-H. Kim, D.-R. Sohn, H.S. Lee, Journal of
Pharmaceutical and Biomedical Analysis 41 (2006) 622.

41. S. Cohen, F. Lhuillier, Y. Mouloua, B. Vignal, P. Favetta, J. Guitton, Journal of

(2007) 165.
42. J. Vacek, B. Klejdus, J. Petrlova, L. Lojkova, V. Kuban, Analyst (Cambridge, U.K.) 131
(2006) 1167.
43. T. Kawamoto, M. Yano, N. Makihata, Analytical Sciences 22 (2006) 489.
44. E.S. Grumbach, T.E. Wheat, J.R. Mazzeo, LCGC North America 24 (2006) 70.
45. X. Wang, W. Li, H.T. Rasmussen, Journal of Chromatography, A 1083 (2005) 58.
46. P. Uutela, R. Reinila, P. Piepponen, R.A. Ketola, R. Kostiainen, Rapid Communications
in Mass Spectrometry 19 (2005) 2950.
47. A.E. Karatapanis, Y.C. Fiamegos, C.D. Stalikas, Journal of Separation Science 32
(2009) 909.
48. G. Karlsson, S. Winge, H. Sandberg, Journal of Chromatography A 1092 (2005) 246.
49. V.V. Tolstikov, O. Fiehn, Analytical Biochemistry 301 (2002) 298.
50. R.R. Towsend, P.H. Lipniumas, C. Bigge, A. Ventom, R. Parekh, Analytical Biochemistry
239 (1996) 200.
51. S. Inagaki, J.Z. Min, T. Toyo’oka, Biomedical Chromatography 21 (2007) 338.
52. T. Yoshida, Journal of Chromatography A 811 (1998) 61.
53. T. Yoshida, Journal of Chromatography A 808 (1998) 105.
54. T. Yoshida, Analytical Chemistry 69 (1997) 3038.
55. D.E. McNulty, R.S. Annan, Molecular and Cellular Proteomics 7 (2008) 971.
56. T. Kimura, K. Nakagawa, Y. Saito, K. Yamagishi, M. Suzuki, K. Yamaki, H. Shinmoto,
T. Miyazawa, BioFactors 22 (2004) 341.
57. D.S. Risley, W.Q. Yang, J.A. Peterson, Journal of Separation Science 29 (2006) 256.
58. C. Dell’Aversano, G.K. Eaglesham, M.A. Quilliam, Journal of Chromatography
A 1028 (2004) 155.
59. C. Tomasek, LCGC North America: The Application Notebook 25 (2007) 21.
60. B. Beilmann, P. Langguth, H. Haeusler, P. Grass, Journal of Chromatography A 1107
(2006) 204.
61. S.-C. Lin, W.-C. Lee, Journal of Chromatography A 803 (1998) 302.
62. T. Yoshida, T. Okada, Journal of Chromatography A 840 (1999) 1.
63. H. Lindner, B. Sarg, W. Helliger, Journal of Chromatography A 782 (1997) 55.
64. H. Lindner, B. Sarg, C. Meraner, W. Helliger, Journal of Chromatography A 743
(1996) 137.
65. C.T. Mant, J.R. Litowski, R.S. Hodges, Journal of Chromatography A 816 (1998) 65.
66. C.T. Mant, L.H. Kondejewski, R.S. Hodges, Journal of Chromatography A 816 (1998)
79.
67. R.S. Hodges, Y. Chen, E. Kopecky, C.T. Mant, Journal of Chromatography A 1053
(2004) 161.
68. C. Viklund, A. Sjoegren, K. Irgum, I. Nes, Analytical Chemistry 73 (2001) 444.
69. M.J. Omaetxebarria, P. Haegglund, F. Elortza, N.M. Hooper, J.M. Arizmendi, O.N.
Jensen, Analytical Chemistry 78 (2006) 3335.
70. W. Jiang, G. Fischer, Y. Girmay, K. Irgum, Journal of Chromatography A 1127 (2006) 82.
71. Y. Takegawa, K. Deguchi, H. Ito, T. Keira, H. Nakagawa, S.-I. Nishimura, Journal of
Separation Science 29 (2006) 2533.
72. P. Appelblad, P. Abrahamsson, LCGC Europe, Suppl., Mar. (2005) 47.
73. J. Bengtsson, B. Jansson, M. Hammarlund-Udenaes, Rapid Communications in Mass
Spectrometry 19 (2005) 2116.
74. H. Idborg, L. Zamani, P.-O. Edlund, I. Schuppe-Koistinen, S.P. Jacobsson, Journal of
(2005) 9.
75. K.L. Wade, I.J. Garrard, J.W. Fahey, Journal of Chromatography A 1154 (2007) 469.

76. K.D. Wyndham, J.E. O’Gara, T.H. Walter, K.H. Glose, N.L. Lawrence, B.A. Alden, G.S.
Izzo, C.J. Hudalla, P.C. Iraneta, Analytical Chemistry 75 (2003) 6781.
77. G. Kesiunaite, A. Padarauskas, Chemija 18 (2007) 30.
78. S. Morimoto, T. Kozuka, M. Takamiya, K. Kimura, S. Matsuyama, S. Kuribayashi,
A. Shigeta, J. Umemura, J. Harada, Y. Yamada, Nihon Igaku Hoshasen Gakkai
Zasshi. Nippon Acta Radiologica 50 (1990) 1087.
79. H. Jahn, R. Muller-Spath, Annales de Radiologie 35 (1992) 297.
80. http://www.drugs.com/mmx/ioversol.html
81. http://www.drugs.com/mmx/optiray-350.html
82. W.H. Ralston, M.S. Robbins, P. James, Investigative Radiology 24 (1989) S16.
83. R.A. Wilkins, J.R. Whittington, G.S. Brigden, A. Lahiri, M.E. Heber, L.O. Hughes,
Investigative Radiology 24 (1989) 781.
84. M. Akagi, S. Masaki, K. Kitazumi, M. Mio, K. Tasaka, Methods and Findings in
Experimental and Clinical Pharmacology 13 (1991) 449.
85. M. Akagi, K. Tasaka, Methods and Findings in Experimental and Clinical Pharmacology
13 (1991) 377.
86. http://www.rxlist.com/optiray-injection-drug.htm, The Internet Drug Index.
87. P. Zou, S.-N. Luo, Y.-L. Liu, M.-H. Xie, Y.-J. He, Z. Yang, Zhongguo Xiandai Yingyong
Yaoxue 22 (2005) 68.
88. E. Collins, S. Muldoon, B. O’Callaghan, D. Carolan, Mallinckrodt Medical Imaging-
Ireland, Ire., Application: WO (1997) p. 25.
89. B. Dejaegher, D. Mangelings, Y. Vander Heyden, Journal of Separation Science 31
(2008) 1438.
90. J. Dunn, Y. Lin, D. Miller, M. Rogic, W. Neumann, S. Woulfe, D. White, Investigative
Radiology 25 (1990) S102.

14 Retention Behavior
of Hydrazine Derivatives
in HILIC Mode
Juris Hmelnickis, Igors Susinskis,
and Kaspars Kokums
Contents
14.1 Introduction...................................................................................................309
14.2 Effect of Organic Solvent.............................................................................. 312
14.3 Effect of Acetonitrile Content on Retention with Different
Stationary Phases..........................................................................................316
14.4 Effect of pH................................................................................................... 321
14.5 Effect of Buffer Concentration with Various Stationary Phases................... 326
14.6 Temperature Effect with Various Stationary Phases..................................... 332
14.7 Application Results and Method Validation.................................................. 338
14.8 Summary....................................................................................................... 341
References............................................................................................................... 341
14.1 Introduction
In pharmaceutical research and development, separation and analysis of a variety of
raw materials, intermediates, and excipients from the active pharmaceutical ingredi-
ent (API) is a hard task, and often demands unusual tools. The nature of impurities
is difficult to predict and analytical methods must be able to detect a wide range of
physicochemical properties, depending on the type of analysis.
A good separation method should combine sufficient retention, selectivity, and
sensitivity. Reversed-phase liquid chromatography (RPLC) is currently the most
popular method in the field of HPLC. Although RPLC is a powerful separation
mode, it has a major limitation: the lack of adequate retention for polar molecules.
A suitable analytical method should avoid both too weak retention that usually
leads to poor separation and too strong retention that results in the longer run and
often is accompanied with unsatisfactory peak shape.
It is hypothesized that analytes are not retained by traditional RPLC on C18 pack-
ing with aqueous mobile phase, containing ammonium acetate pH 5.5 buffer and 2%
ACN, then these compounds would represent highly polar analytes.1
309

Some chromatographic methods, such as size-exclusion chromatography (SEC),

ion-exchange chromatography (IEC), hydrophobic-interaction chromatography
(HIC), and immobilized metal-affinity chromatography (IMAC), are not compat-
ible for mass-spectrometric detection due to high salt concentrations in the mobile
phases. Traditional RPLC can work without salts, but separation of small hydrophilic
compounds is generally poor, because of weak retention. However, derivatization of
those compounds is laborious and prone to introduce errors in the overall analytical
scheme. Side reactions and slow kinetics at trace concentrations are also the factors
limiting the usefulness of derivatization. Hydrophilic-interaction chromatography
(HILIC) can solve this problem by retaining polar compounds because such separa-
tion mode is based on the hydrophilicity of the analyte.
HILIC as an analytical tool for the analysis of polar compounds has been studied
extensively in the past few years. HILIC mode is based on the combination of hydro-
philic stationary phases and hydrophobic, mostly organic, mobile phases. HILIC is
effective for separation of low molecular weight polar and ionizable compounds.2–5
This mode was first discussed in detail in the works of Alpert.6 It should be pointed
out that chromatography with hydrophilic sorbents has been routinely used for anal-
ysis of sugars since 1975, and the use of low amounts of water in organic solvents for
chromatographic elution was described by Samuelson in the 1950s—but Alpert was
the first who used abbreviation “HILIC.”7,8 A comprehensive review on the scope
and limitations of HILIC approach for analysis of polar analytes was published
recently by Hemström and Irgum.9 HILIC method has been used for pharmaceutical
product analysis and is more suitable to mass-spectrometric detection and since it
gives better mass-spectrometric sensitivity compare to the RPLC.1,10–18
In HILIC mode, the mobile phase usually is an aqueous/organic mixture with
the water content in less than 60%. Polar compounds always have good solubil-
ity in aqueous mobile phase, which overcome poor solubility of these compounds
in normal-phase chromatography. Polar analytes usually are retained stronger than
nonpolar ones. The retention mechanism for HILIC is a partitioning process between
the bulk mobile phase and a partially immobilized water-rich layer on the stationary
phase. Some researchers call such conditions as “the aqueous normal phase” (point-
ing out that “traditional” normal-phase chromatography is based on adsorption onto
a polar stationary phase, not a water layer); others state that HILIC and the “aqueous
normal phase” are the same. But “pure HILIC” separation mode appears, however,
quite rare.9 Depending on the surface chemistry of the stationary phase, ion exchange
in particular can contribute significantly to the retention of ionic analytes.2,3,9,19–23
Yoshida, same as Alpert, considered that HILIC retention encompassed both
hydrogen bonding (which depends on Lewis acidity/basicity) and dipole–dipole
interactions (dependent on dipole moments and polarizability of molecules).24
Clearly, the mechanism of HILIC separations—partition, ion exchange, and hydro-
phobic retention—contribute to various degrees, depending on the particular condi-
tions employed.25
In this chapter, we describe the retention behavior of a hydrazine-based pharma-
ceutical compound mildronate and its analogs (see Table 14.1) under HILIC mode on
various polar stationary phases (silica, amino, cyano, amide, and zwitterionic sul-
fobetaine). Mildronate or 3-(2,2,2-trimethylhydrazinium)propionate dihydrate (also

Retention Behavior of Hydrazine Derivatives in HILIC Mode 311
Table 14.1
Studied Compounds
Compound
Formula No.
NH+ Br – 1
H2N N+ Br– 2
N
N+ O–
*H2O 3
O
*2H2O
N+ N O– 4
O
Br–
N+ N O 5
O
Br–
N+ N O 6
O
Br–
N+ N O 7
known as Mildronatum, Meldonium hydrate, MET-88, 3-TMHP) is an anti-ischemic

drug developed at the Latvian Institute of Organic synthesis over 20 years ago. The
impact of separation conditions (organic solvent content, buffer pH, temperature, salt
concentration in mobile phase) on retention has been systematically studied.
As mildronate and its related substances are small polar molecules, weak retention
of these substances under the reversed-phases mode can be expected, and applica-
tion of ion-pair chromatography (on RPLC sorbents) seems to be the easiest alter-
native. However, there is another restriction for the above compounds—the lack of
distinguished chromophores, which makes mass spectrometry (MS) the appropriate
detection alternative. Unfortunately, ion-pair agents as well as a high water content in
the mobile phase usually reduce the sensitivity of mass-spectrometric detection.26,27
HPLC method with evaporative light-scattering detection (ELSD) was used for
the determination of mildronate previously.28,29 It is known that stripping voltamme-
try, HPLC-MS-MS, or UV detection (with derivatization) were used for mildronate
detection in the plasma and urine.30–32 In all these methods, polar stationary phases
(columns Silasorb 600 Silica, Inertsil NH2, Inertsil CN-3) and water-rich (more than
60% water) mobile phases were used. In our opinion, such chromatographic conditions

suffer from insufficient sensitivity, even HPLC-MS being employed. Impurities men-
tioned in Table 14.1 have not been studied chromatographically before, except for a
brief discussion by the same researchers as this article (determination of these com-
pounds is important for purity control of the active pharmaceutical compound).33
Latest works described the detection of 1,1-dimethylhydrazine (this compound,
as well as hydrazine are classified as a carcinogen requiring control) in HILIC mode
by chemiluminescent nitrogen detector (CLND). Acetonitrile in mobile phase was
replaced by ethanol.34 The same scientists refer to other studies, mentioning ion
chromatography with conductivity detection.35 Other researchers used derivatization
followed by GC or HPLC.36–42 But derivatization is complicated, time consuming,
and sometimes not suitable for detection of hydrazine in compounds with amine
functional groups.
14.2 Effect of Organic Solvent

The selectivity and sensitivity of an LC-MS method depends not only on the ioniza-
tion technique but also on the LC technique. One of the most important parameters
is the mobile phase composition. Variation of organic solvent (acetonitrile, methanol,
ethanol, isopropanol, acetone, tetrahydrofuran) concentration is commonly used for
optimizing retention, selectivity, and sensitivity.
Typical HILIC mobile phases consist of 40%–97% acetonitrile in water (with
volatile buffer, if necessary). At least 3% of water (some researchers talk about “non-
water HILIC”—using another polar solvent instead) is necessary to ensure sufficient
hydration of surface of stationary phase.
Acetonitrile is the most popular organic solvent for HILIC; nevertheless it is pos-
sible to use several other polar, water-miscible solvents. Solvent strength is inverted
to what is observed in RPLC, and relatively:
Tetrahydrofuran < acetone < acetonitrile < isopropanol < ethanol < methanol <
water43 (other sources mention the sequence “acetonitrile < tetrahydrofuran < isopropa-
nol < methanol”19).
The organic components of the HILIC mobile phase can be subdivided into polar
protic and aprotic solvents. Methanol, ethanol, and isopropanol represent polar protic
solvents. Examples of polar aprotic solvents are acetonitrile and tetrahydrofuran
(THF). Protic solvents can be both donors and acceptors of hydrogen bonds, and
aprotic solvents can be only hydrogen-bond acceptors. Due to their ability to make
hydrogen bond, protic solvents can compete for polar active sites on the surface
of the stationary phase, affecting the formation of water layer on the surface, thus
changing stationary phase to less hydrophilic. As a consequence, the analytes with
strong hydrogen-bonding potentials are poorly retained.
According to Snyder, there are three major contributions to solvent selectivity, i.e.,
solvent strength, solvent–analyte localization, and solvent–analyte hydrogen bond-
ing.44 The solvent strengths (ε°) is following the order of: methanol (0.73), isopropa-
nol (0.60), THF (0.48), acetonitrile (0.50), and acetone (0.50).
Because of its strong hydrogen-bonding ability, methanol can perturb the for-
mation of water layer on silica surface, replacing water molecules and producing a
more hydrophobic stationary phase.45 As a consequence, the ionic analytes are poorly

retained. Compared to methanol, a decrease of the hydrophilic character of isopro-

panol is caused by the longer alkyl chain, which competes less strongly for the active
sites, and leaves more space for analyte molecules to be absorbed. The hydrogen-
bonding interactions between analytes and solvent molecules arise in both the mobile
phase and the stationary phase, significantly affecting the analytes’ retention and
solvent selectivity. In contrast to methanol and isopropanol, the hydrogen-bond donor
strengths (denoted α) of acetonitrile and THF are weak, and their α values are 0.15
and 0.00, respectively. The elution strengths of acetonitrile and THF are approximate
equal, but the analytes are retained more strongly with acetonitrile than THF. This is
because THF is a better hydrogen-bond acceptor (the corresponding strength being
denoted β) than acetonitrile, and the solvent–analyte localization effect is correlated
with β of the solvent.44 The higher β of the solvent, the lower the pKa value of the
basic analyte.46 The β values of THF and acetonitrile are 0.49 and 0.25, respectively.
It was found that retention factors of analytes follow the values of pKa of amino
groups of these analytes when methanol is used in mobile phase.20 The results have
shown again that the ion-exchange interactions dominate the retention of the ana-
lytes because of the ionization of silica silanols at elevated pH values, causing weak-
ened adsorption interactions between methanol and silica surface.
Some researchers stated that even sample diluent can influence the peak shape
of the analytes.47,48 The diluent should be as close to the initial mobile phase as pos-
sible. The peak shape improves as the aqueous content of the diluent is decreased.
However, if water is removed completely from diluent and the sample is prepared
in, for example, 100% methanol, wide peaks are observed and the resolution of the
compounds decreases. Castells et al. determined that the best compromise for solu-
bility and peak shape would be 75:25 (v/v) mixture of acetonitrile and methanol. It
should be noted that the solubility in various organic solvent combinations is analyte
dependent and should be optimized accordingly.
It is important to be aware that organic solvents in the mobile phase can change
pKa of a buffer as well as that of the analyte. Consequently, that shifts protolytic
equilibrium and changes conditions of chromatography.
Other researchers tested replacing water in the mobile phase by other polar
solvents, mainly alcohols.2 In such conditions, hydrophilic interaction was still
observed, but a greater content (approximately doubled) of methanol was required to
keep retention the same as for a water-containing mobile phase. Some compounds
were not eluted by 100% n-propanol at all, apparently due to the decreasing hydro-
philic character caused by the longer alkyl chain.
Ideally, samples should be dissolved in the same solvent as the mobile phase.
Differences in viscosity between the sample and mobile phase can lead to peak
distortion, while differences in the solvent strength can lead to peak broadening. The
strong injection solvent prevents the analyte from absorbing at the head of the column.
As the solvent injected is diluted by the mobile phase, the analyte retention gradually
approaches that characteristic of the mobile phase. However, as the rate of dilution is
finite, additional broadening occurs.49
All experiments were carried out with mobile phases containing from 60% to
95% of the organic part, with addition of 0.1% of formic acid in aqueous part, and
Atlantis HILIC column.

First, retention behavior of the above hydrazine derivatives in various organic

solvents was compared by varying the amount of water in mobile phase and chang-
ing different organic modifiers as well. The main tendency is obvious: retention is
constantly rising with the increase of the organic part of the mobile phase. It seems
that mobile phases with a lower content of organics are too strong: no compounds
were retained, separation did not occur.
Retention becomes significant only when the mobile phase contains at least 80%–
85% of organics. On the other hand, there was no use to employ more than 95% of
organics: retention was too strong, and one chromatography run took more than 2â•›h.
It was not surprising that in the line of homologs of methanol, ethanol, and isopro-
panol (as shown in Figures 14.1 through 14.3), the greatest retention of analytes was
provided by the one with the lowest polarity—isopropanol, and other members of
this line presented predictable change of properties. It is found no clear connection
7 7 6 5 4 3 2 1
6
0
50 60 70 80 90 100
Methanol, %
Figure 14.1 Retention of hydrazine derivatives in methanol-containing mobile phase.
30
25 7 6 5 4 3 2 1
20
15
10
0
50 60 70 80 90 100
Ethanol, %
Figure 14.2 Retention of hydrazine derivatives in ethanol-containing mobile phase.

90
80 7 6 5 4 3 2 1
70
60
50
40
30
20
10
0
60 70 80 90 100
Isopropanol, %
Figure 14.3 Retention of hydrazine derivatives in isopropanol-containing mobile phase.

between the retention of analytes and any of organic solvent’s properties; probably
there is not such a simple law or formula due to the complex mechanism of retention
on HILIC sorbents.
Acetone was chosen for its low viscosity and good miscibility with water. As pre-
dicted, acetone turned out as weak mobile phase. There was no literature found about
acetone as the mobile phase—maybe due to its UV-absorbing properties (UV cutoff
at 340â•›nm). But the authors hope that this example will encourage other researchers
to look for unusual mobile phases—suitable for their specific applications.
Acetonitrile—a solvent very popular in HPLC—showed a behavior (see Figure
14.4) similar to that of ethanol (Figure 14.2). THF—a solvent sometimes used in
RP-HPLC—showed a behavior similar to that of acetone (see Figures 14.5 and 14.6).
The solvent strength for hydrazine compounds is as follows:
Tetrahydrofuran < acetone < isopropanol < acetonitrile < ethanol < methanol < water
40
35 7 6 5 4 3 2 1
30
25
20
15
10
0
50 60 70 80 90 100
Acetonitrile, %
Figure 14.4 Retention of hydrazine derivatives in acetonitrile-containing mobile phase.

100
90 7 6 5 4 3 2 1
80
70
60
50
40
30
20
10
0
50 60 70 80 90 100
Acetone, %
Figure 14.5 Retention of hydrazine derivatives in acetone-containing mobile phase.
100
90 7 6 5 4 3 2 1
80
70
60
50
40
30
20
10
0
50 60 70 80 90 100
Tetrahydrofuran, %
Figure 14.6 Retention of hydrazine derivatives in tetrahydrofuran-containing mobile phase.
14.3 Effect of Acetonitrile Content on Retention

with Different Stationary Phases
Structurally, the studied substances (see Table 14.1) can be divided into three differ-
ent groups: compounds 1–2 (cations), compounds 3–4 (zwitterions), and compounds
5–7 (cationic esters).
The columns 1–6 (Table 14.2) were tested with respect to their ability of retaining
the analytes 1–7 in typical HILIC conditions. The percentage of acetonitrile in the
mobile phase was studied in the range between 60% and 95%, with deionized water
and 0.1% FA (formic acid).
It was found that the polar bonded phase Discovery Cyano appeared to be the least
retentive (Figure 14.7). At acetonitrile concentrations up to 90%, the retention was

Table 14.2
Columns Used and Properties of Stationary Phases
Pore Particle Surface Dimension
No. Column Name Phase Type Size (Å) Size (μm) Area (m2/g) (mm)
1 Discovery® Cyano Cyano 180 5 200 2.1 × 100
2 Hypersil APS-1 Amino 120 3 170 3.2 × 100
3 Atlantis HILIC Silica Silica 100 3 330 2.1 × 150
4 Alltima HP Silica Silica 100 3 450 2.1 × 150
5 Spherisorb® Silica Silica 80 3 220 2.1 × 100
6 ZIC®-HILIC Sulfobetaine 200 5 135 2.1 × 100
7 Spherisorb® NH2 Amino 80 3 220 2.1 × 100
8 TSK-Gel Amide-80 Amide 80 5 313 2.0 × 100
15
14
7 6 5 4 3 2 1
13
12
11
10
9
8
7
6
5
4
3
2
1
0
60 65 70 75 80 85 90 95
Acetonitrile content, %
Figure 14.7 Relationship between the retention factor (k) and the acetonitrile content.
Column: Discovery Cyano. Mobile phases: 0.1% FA in water and acetonitrile.
negligible. Even at 95% of acetonitrile, the retention factors (k) were only about 3.
This indicates that under the conditions used, the volume of dynamically generated
water-rich stationary phase is small. Probably, the residual silanol groups are unable
to accumulate water because of surface shielding by organic ligands.
An interesting phenomenon was the fact that there was almost complete absence of
structural selectivity on this cyano column. All analytes irrespective of their molecular
size, functionalities, and ionized sites present elute within an extremely narrow k range.
Obviously, such behavior is not attractive from a purely analytical point of view. Still, its
mechanism might be a subject of a more thorough investigation in the future.
The amino-modified stationary phase (Hypersil APS-1 column) showed a dif-
ferent behavior (Figure 14.8). Cationic substances were retained very weakly in the
entire range of acetonitrile concentrations studied (60%–95%). On the other hand,

60
7 6 5 4 3 2 1
50
40
30
20
10
0
60 65 70 75 80 85 90 95
Column: Hypersil APS-1. Mobile phases: 0.1% FA in water and acetonitrile.
zwitterions 3–4 were strongly retained which indicates electrostatic interaction

between the protonated amino groups of the stationary phase and the negatively
charged moieties of zwitterions.
Specialized HILIC silica (Atlantis HILIC Silica column) showed an increase of
retention with the increase of acetonitrile content in the mobile phase (Figure 14.9).
The cationic esters 5–7 differ by one carbon atom while having identical function-
ality. Retention behavior within such a structurally close group of compounds is
mainly defined by the balance of hydrophilic and hydrophobic parts. The linear
40
7 6 5 4 3 2 1
35
30
25
20
15
10
0
60 65 70 75 80 85 90 95
Column: Atlantis HILIC Silica. Mobile phases: 0.1% FA in water and acetonitrile.

relationships between the number of carbon atoms and retention is a common phe-
nomenon in such cases, indicative of the role of partition mechanism between the
stationary and mobile phases. The cationic compounds 1–2 cannot be separated on
this column, even when retention is strong, at 95% of acetonitrile. One could expect
that compound 4 should be retained much more strongly than cationic ester 5. In fact,
data show only slight difference between these two compounds. Most likely, in the
acidic conditions used in these experiments, the carboxylic group does not influence
partition values. The role of this group seems to be so insignificant that compounds
5–7 and 4 belonging to esters and a betaine behave like homologs (Figure 14.9).
Compound 3 is retained much stronger than 4–7 and falls out of the common line.
This can be explained by a more pronounced contribution of ion exchange in the case
of this cyclic structure.
It was observed that the common silica (Alltima HP Silica column) when used
with 70%–80% acetonitrile mobile phase showed similar retention (Figure 14.10)
and similar elution order to the specialized HILIC silica (Atlantis HILIC Silica col-
umn). This suggests the retention mechanisms are similar on both silica packings.
On the other hand, at higher (90%–95%) acetonitrile concentrations, the retention on
the common silica is much stronger than that on HILIC silica. A possible explana-
tion for this phenomenon is the larger surface area of Alltima HP Silica (see Table
14.2). It can retain a larger volume of a water-rich layer acting as the stationary phase.
Spherisorb silica column generally is about three times (Figure 14.11) more retentive
than Alltima HP silica if the mobile phase with 70% acetonitrile is used. At higher
concentrations of ACN, the difference is even more pronounced. Retention behav-
iors within homolog series 5–7 are similar to what was observed on Alltima HP and
Atlantis HILIC silica. At the same time, the profile of structural selectivity is dif-
ferent. For example, zwitterions are retained relatively weak on this stationary phase.
110
100 7 6 5 4 3 2 1
90
80
70
60
50
40
30
20
10
0
60 65 70 75 80 85 90 95
Column: Alltima HP Silica. Mobile phases: 0.1% FA in water and acetonitrile.

140
7 6 5 4 3 2 1
120
100
80
60
40
20
0
60 65 70 75 80 85 90 95
Column: Spherisorb Silica. Mobile phases: 0.1% FA in water and acetonitrile.
Compound 3 is the least retained, and the behavior of mildronate (compound 4) is

different as compared to esters (compounds 5–7). A possible explanation lies in the
properties of Spherisorb silica. It is one of the earliest types of HPLC silica (type
A silica), most likely manufactured according to an old technology and contain-
ing metal impurities, capable of secondary interactions with nitrogen-containing
compounds.
Type A silica gels, prepared by precipitation from solutions of silicates, are
acidic because of contamination with certain metals, which activate surface silanol
groups and can form complexes with some chelating analytes, causing strong reten-
tion of asymmetrical peaks. Newer, highly purified and less-acidic spherical silica
particles, “sil-gel” (type B silica) are formed by the aggregation of silica sols in
the air, contain very low amounts of metals and are more stable at intermediate
and higher pH than the xerogel-type materials, to at least pH 9. Atlantis HILIC
Silica column is made out of type B silica. At higher pH, silanol groups are ionized
and cation exchange plays a significant role in retention, especially of positively
charged basic compounds. Suppression of silanol ionization by the addition of tri-
fluoroacetic acid (TFA) may promote the ion-pairing mechanism. Occasionally,
irreversible adsorption may occur on bare silica in the HILIC mode. Silica gel of
type C having hydrosilated surface populated with nonpolar silicon hydride (Si-H)
groups instead of silanol groups may have up to 95% of original silanols removed,
so that it is less polar than silica gels with higher population of silanol groups. An
important characteristic of the hydride phases is little attraction for water and con-
sequently improved reproducibility of retention in both the organic and aqueous
normal phases (NPs).50
Some researchers found that columns of ultra-high purity silica gave lower reten-
tion and less selectivity than the columns made from metal containing silica.49 This
was caused by higher acidity of the silanols activated by metal impurities in their
vicinity, mainly aluminum and iron. McKeown et al.51 explained that as combined
RP and ion exchange interactions, but Hemström and Irgum insist that these findings
might be a mixed mode of cation exchange and HILIC, as well.9

30
7 6 5 4 3 2 1
25
20
15
10
0
60 65 70 75 80 85 90 95
Column: ZIC-HILIC. Mobile phases: 0.1% FA in water and acetonitrile.
The fact that considerable retention is observed even at 60% of acetonitrile (when
the volume of the aqueous-rich stationary phase is relatively low) suggests a direct
interaction with surface groups of the stationary phase.
Retention of test analytes under HILIC conditions on ZIC-HILIC column is rep-
resented in Figure 14.12. Average retention values on ZIC-HILIC are lower than
those observed for silica columns, but the general character of plots is similar. On
the other hand, the structural factors are more visible on ZIC-HILIC column. First,
like it was on silica columns, the retention factors of cations 1 and 2 show very
similar values. On the other hand, the retention of cationic esters 5–7 is much lower
than that observed on silica. Zwitterions behave differently. At intermediate content
of acetonitrile (70%–80%), their behavior is similar to that of other analytes, but at
above 80% they show a more rapid increase in retention than other compounds do.
A similar, still less pronounced effect was observed on HILIC silica.
14.4 Effect of pH
The concept of electrostatic ion chromatography (EIC), or zwitterionic ion chroma-
tography (ZIC) as it was named later, with a zwitterionic stationary phase for the
separation of ions, was first proposed by Hu et al. in 1993.52 The separation uses
zwitterionic stationary phase that maintains a fixed positive and negative charge
in close proximity to each other. Analyte ions can access both the fixed positive
charge, in the case of anions, and the fixed negative charge, in the case of a cation.
As a result of the proximity of the charges, the analyte ions will be repulsed and
attracted at the same time. Thus, a unique and sometimes complicated selectiv-
ity is obtained. Many studies have been performed to attempt outlining the charge
interactions on a molecular level. Hu and Haddad reported the formation of an elec-
trical double layer to explain retention mechanisms.53,54 Okada and Patil modeled

zwitterionic retention based upon Poisson–Boltzmann theory.55 The formation of a

Donnan membrane combined the previous theories of Hu (electric double layer) and
Patil (charged surfaces) to explain both elution order and the effect that mobile com-
position has no retention.56,57 There have been just a few applications reported that
take advantage of the separating power of this unique stationary phase.58 Jonsson
and Appelblad demonstrated the separation of polar and hydrophilic compounds
with a sulfobetaine-type zwitterionic stationary phase. The article focused on the
selectivity from a HILIC perspective, where the effect of acetonitrile and methanol
was evaluated for the retention of RNA–DNA bases in an ammonium formate buf-
fer system.59
The combination of separation mechanisms between HILIC and EIC can the-
oretically be complicate and difficult to understand; however, the performance of
zwitterionic column is greatly enhanced with the addition of organic to the mobile
phase to take advantage of the HILIC effect. It already has been demonstrated that
acetonitrile will promote the HILIC effect more than methanol.
Across the pH range 3.1–6.6, sulfobetaine-type zwitterionic stationary phase
retains its permanent positive and negative charges. Increasing pH (within this range),
the retention of all the cations increases and the retention of the anions decreases.
The effect on cation retention is presumably due to the H+ interacting with the nega-
tively charged part of the zwitterions (SO3−), which ultimately shields the cation
from having a strong interaction at a lower pH. The anions are following standard
ion-exchange theory. The same studies showed that while increasing the buffer con-
centration from 10 to 200â•›m M ammonium acetate both the peak shape and retention
times of the ions are drastically affected. The peaks of cations had tailing at lower
buffer concentration, then peak fronting at higher buffer concentrations. From these
experiments, a range of 50–100â•›m M buffer concentration was recommended for the
experiments. Such concentration of the buffer is doubtful for mixtures of aqueous
buffer with organic part (acetonitrile) of the mobile phase: the solubility of ammo-
nium salts in a mobile phase containing high concentration of organics is limited.
Besides, high buffer concentration is unacceptable for MS detection.
Change of retention can be explained by a two-part mechanism. The main mecha-
nism of interaction in the HILIC mode is based on a partitioning of the ions into
stagnant layer of water on stationary-phase surface, so the decrease of the retention
time might be best understood by a shift of equilibrium concentrations. As the buf-
fer cation’s (NH4+) concentration increases preferentially in the aqueous layer, there
is less opportunity for the analyte counterions to partition into the aqueous layer.
Thus, the ions are swept through the column (mainly in the organic layer) with less
interaction with the column and the aqueous phase. In addition, as the NH4+ interacts
strongly with the SO3− fixed negative charges as the buffer concentration increases,
access to these fixed charges is diminished. As a result, cations do not interact with
SO3− and are not significantly retained; anion retention is affected in the opposite
manner. The anions do not experience the typical repulsion forces of the SO3− func-
tionality and can then access the tertiary amine for ion exchange. This ion-exchange
interaction causes the anions to be retained more strongly.58
Peptides are amphoteric molecules whose charges change with pH of the sur-
rounding medium. Their net charge is zero at pH = pI and increases with decreasing

pH of buffer solution and vice versa, owing to protonation and dissociation of weekly
basic and acidic side chains of the peptide, and of the amino and carboxyl terminals.
In this study, hydrophilic interaction and elution power of ionic interaction were kept
constant (changing pH of buffers, while maintaining the acetonitrile content and buf-
fer concentration constant). The retention will mainly depend on the charge change
of peptides and stationary phase.60 Thus, the retention data can be correlated with the
estimated pH-dependent charge of the peptides. The k of small peptides decreased
slightly when pH was increased from 3 to 7. It can be explained as increased posi-
tive charge accompanied by an increased hydrophilicity of the peptides at lower pH,
which in turn increased retention factors in the HILIC separation mode. An opposite
corresponding dependence has been found by Guo et al., i.e., that the positive charge
at lower pH results in large decrease in retention times of peptides in RP-HPLC.60,61
Small peptides were well separated based on a mix of hydrophilic interaction and
ion interaction between the stationary phase and analyte. Compared to native silica,
zwitterionic sorbent showed more stable retention for basic peptides within the pH
range 3–7 and higher retention capacity for acidic peptides. The contribution from
ionic interaction from ionic interaction on zwitterionic sorbent was also lower than
on silica at pH > 5, and can be eliminated when the buffer concentration in mobile
phase is above 20â•›m M.
Some researchers were surprised when they encountered dramatic loss of reso-
lution or theoretical plates while significantly increasing mobile-phase flow rate
(as described, from 1 to 5â•›m L/min).62 It seems they had forgotten conclusions from
van Deemter’s equation—that there is an optimal flow rate with minimal theo-
retical plate height, and at conditions far away from such plates would not be so
ideal. The same researchers, working with a silica column in HILIC mode, came
to a conclusion that increasing organic content in the mobile phase substantially
increases retention times of cations. In addition, this column could exhibit ion-
exchange characteristics as a part of the retention mechanism; thus, the retention
of the cations should be affected by a change in buffer concentration and pH. It
is experimentally proved: increasing buffer concentration from 10 to 200â•›m M
ammonium acetate (pH 6.65), the retention times of cations monitored dropped
about 2–4 times. When pH decreased from 6.65 to 3.55 at constant buffer concen-
tration of 50â•›m M ammonium acetate, a decrease of retention times of the cation
was observed—for about one-third. All these observations can be attributed to the
active silanol sites and ion exchange.
Effect of pH on retention and selectivity of seven compounds was examined in pH
range of 2.6–5.8 for Atlantis HILIC Silica column (Figure 14.13) and in pH 2.5–7.6
for ZIC-HILIC column (Figure 14.14).
The retention factor of cations 1–2 and cationic esters 5–7 on Atlantic HILIC
Silica column increased steadily in the pH range between 2.8 and 5.8. The most
probable reason of this is ion exchange. At higher pH, the degree of ionization of
surface silanol group increases, thus creating more favorable conditions for such
interactions. Mildronate (4) behaves in a similar way as cations 1–2. Another zwit-
terion 3 shows a completely different pattern: its retention is independent of pH.
While changes of retention on Atlantis HILIC are monotonous, similar series of
experiments on ZIC-HILIC column, performed at a wider pH range, showed maxima

7 6 5 4 3 2 1
60
50
40
30
20
10
0
2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
pH
Figure 14.13 Effect of buffer pH on the retention factor (k). Column: Atlantis HILIC.
Mobile phase: acetonitrile–5â•›m M ammonium formate (85/15, v/v).
7 6 5 4 3 2 1
35
30
25
20
15
10
0
2.5 3.5 4.5 5.5 6.5 7.5
pH
Figure 14.14 Effect of buffer pH on the retention factor (k). Column: ZIC-HILIC. Mobile
phase: acetonitrile–5â•›m M ammonium formate (85/15, v/v).
at pH about 6. Again, the compound 3 is different, its retention steadily decreasing in

the whole pH range studied.
Chromatograms of the seven test compounds are shown in Figure 14.15a for
Atlantic HILIC Silica column, pH 5.0; and Figure 14.15b for ZIC-HILIC column at
pH 5.0. ZIC-HILIC column shows the best selectivity for the seven test compounds.

Retention Behavior of Hydrazine Derivatives in HILIC Mode
7 7
ES+ TIC
TIC 3.73 × 104
9.05 × 104
2
4
4
6 5
6
Intensity
Intensity
3
2
3 1
1
20.00 40.00 60.00 80.00 100.00 120.00 140.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00
(a) Time, min (b) Time, min
â•…
Figure 14.15 Separation of test compounds 1–7 (see Table 14.1) under HILIC mode: (a) Atlantis HILIC silica; (b) ZIC HILIC column. Mobile phase:
325
acetonitrile–5â•›m M ammonium formate pH 5.0 (85:15, v/v).

14.5 Effect of Buffer Concentration

with Various Stationary Phases
Many salts typically used in reversed-phase (RP) chromatography may not be suit-
able for HILIC due to poor solubility in the mobile phases containing high level
of acetonitrile. In addition to ammonium formate, other salts with relatively high
solubility at high organic levels have also been used for HILIC, such as ammonium
acetate and bicarbonate salts, triethylamine phosphate and sodium perchlorate, but
the last two are not compatible with MS detection.
It is reported in the literature that low concentration of buffer salts in the mobile
phase improves peak shape and separation efficiency in HILIC.1 The use of buff-
ered mobile phase is crucial for achieving reproducible chromatographic separations
of charged species, since electrostatic interactions between the analyte and station-
ary phase impacting retention are influenced and controlled by the buffer. However,
although volatile buffering agents are compatible with ESI-MS, it is known that elec-
trospray ionization is adversely affected by buffer salts due to ionization suppression.
It has also been determined that HILIC retention is inversely proportional to mobile
phase ionic concentration, since increases in salt concentration result in increases of
solution hydrophilicity. Therefore, an investigation of the effects of volatile buffer
salt concentration at low levels required for optimal retention and ESI-MS sensitivity
needs to be performed. Even the presence of 3.3â•›m M of ammonium acetate (pH 5.5)
significantly improves the peak shape and efficiency. The factors influencing sample
HILIC retention in extremely low buffer salt environments are complex and may
include effects such as ion exclusion and/or ion exchange, in addition to the impact
of the salt upon mobile phase hydrophilicity. Other set of experiments proved that the
increase in both sample solution pH and ionic strength can by minimized through
the use of the ammonium acetate buffer.1
Mobile phases containing TFA have been successfully employed for the nor-
mal phase HILIC chromatography of polar compounds. But at the conditions of
this study, TFA provided unacceptably weak retention, probably due to ion-pairing
effects which serve to decrease the hydrophilicity of the analytes. In general, in
the buffered mobile phases retention of the test compounds appeared to follow the
pattern of “acetate ≈ formate > bicarbonate,” which corresponds to a relationship
similar to the differences in anion polarity. Similar results have been reported for the
HILIC separation of peptides, where retention in the presence of TFA, formic acid,
and acetic acid was compared and the differences were attributed to the effectiveness
of these acids in minimizing ion-exchange interactions. Since both the acidic and
basic test compounds behaved in a relatively similar fashion from one buffer system
to another, the results suggested that the effect of the pH upon analyte retention were
minimal in comparison to the changes occurring within the stationary base directed
adsorption/desorption processes, as influenced by the character of the buffer anion.
HILIC columns contain hydrophilic and often charged stationary phases at lower
pH range. The main part of chromatographic separation occurs in water layer on the
surface of the stationary phase. But when the stationary phase is charged, it is neces-
sary to keep in mind electrostatic or ionic interactions—or to use salts or buffers in
the mobile phase to eliminate such interactions.

Three types of mobile phases are commonly used in the HILIC mode, including
acetonitrile/water, acetonitrile/salt solution, and acetonitrile/buffer. The mobile phase
should be chosen on the basis of chemical properties of the analytes. For example,
for neutral molecules, the acetonitrile/water mobile phase is preferred but any of the
three types can be used. For organic salts of strong bases, the acetonitrile/salt solution
mobile phase can be used. The acetonitrile/buffer type of the mobile phase is the best
for weak acids or weak bases, as well as for samples containing unknown components.
If the mobile phase with salt solution is used, the observed effects are consistent with
the HILIC retention mechanism. It has been proposed that there may be a layer of
stagnant water on the surface of the polar stationary phase to facilitate the partition
interaction for analytes. In the absence of a salt or buffer, the charged analytes have a
much greater tendency to reside in the water layer. Therefore, the analytes have long
retention time or are permanently retained on the column. As the salt concentration
increases, the presence of excess amount of counter ions will promote the formation of
ion pairs for the charged analytes. The formed ion pairs will have better solubility in
the mobile phase, which results in shorter retention times.63
There are a number of publications dealing with studies of salt-concentration effects
in HILIC, but no information was found on compounds shown in Table 14.1.1,3,4,62,63
Six commercially available silica-based columns 3–8 (see Table 14.2) representing dif-
ferent polar stationary phases were selected for this study. The effect of ammonium
formate concentration in the mobile phase of acetonitrile/pH 5.0 ammonium formate
buffer (85/15, v/v) on the retention of the seven test compounds (see Table 14.1) was
investigated. For all seven compounds, samples of 0.1â•›ng/mL concentrations were pre-
pared. The sample solvent was a binary mixture of acetonitrile–water (90:10, v/v). The
concentration range of ammonium formate was from 1 to 30â•›mM (the buffer concen-
tration and pH values refer to the aqueous part only). A further increase in the salt
concentration was not possible due to solubility limitations in the mobile phase.
As shown in Figure 14.16a through c, there are significant retention differences
between silica columns of different manufacturers. The Alltima HP stationery phase
yielded the highest retention while both Atlantis HILIC and Waters Spherisorb show
similar retention factor values. Although purity of the silica used in the prepara-
tion of the stationary phase may cause such differences,16 one cannot exclude other
reasons like pore size, surface area, and even different manufacturing technologies.
As shown in Figure 14.16a through c, the increase of ammonium formate con-
centration from 1 to 30â•›m M, on all silica phases results in a significant decrease
of retention for compounds 1 and 2, as well as 5, 6, and 7. Retention of betaine-
type substances 3 and 4 seems not affected by the change of ammonium formate
concentration.
We found that the retention factor of unmodified silica sorbents is proportional to
the relative surface area of stationary phase (see Figures 14.17 and 14.18). Sorption
capacity per unit of surface is constant; the same is the percentage of ion exchange
and partition.
One can speculate that the increase in ammonium formate concentration may
cause a certain saturation of the active silanol sites by excess of ammonium ions thus
decreasing the ability of the positively charged analytes to interact with these sites.14
Our opinion is that the presence of a salt in the mobile phase suppresses the solution

120
7 6 5 4 3 2 1
100
80
60
40
20
0
0 5 10 15 20 25 30
(a) Ammonium formate, mM
70
7 6 5 4 3 2 1
60
50
40
30
20
10
0
0 5 10 15 20 25 30
(b) Ammonium formate, mM
70
7 6 5 4 3 2 1
60
50
40
30
20
10
0
0 5 10 15 20 25 30
(c) Ammonium formate, mM
Figure 14.16 Effect of ammonium formate buffer solution concentration in mobile

phase on retention of mildronate and related substances in sorbents. (a) Alltima HP silica,
(b) Atlantis HILIC silica, (c) Spherisorb silica.

20
15
7 6 5 4 3 2 1
10
0
0 5 10 15 20 25 30
(d) Ammonium formate, mM
70
60 7 6 5 4 3 2 1
50
40
30
20
10
0
0 5 10 15 20 25 30
(e) Ammonium formate, mM
60
7 6 5 4 3 2 1
50
40
30
20
10
0
0 5 10 15 20 25 30
(f ) Ammonium formate, mM
Figure 14.16 (continued) (d) Spherisorb NH2, (e) TSK-Gel Amide-80, (f) ZIC-HILIC.

100
y = 0.246249x – 20.663123
90
R2 = 0.992040
80
70
60
50
40
30
20
10
0
150 200 250 300 350 400 450 500
Surface area, m2/g
Figure 14.17 Retention factor (k) depending on surface area of silica packing. Measured
compound: 4. Mobile phase: acetonitrile–15â•›m M ammonium formate pH 5.0 (85/15, v/v).
25
y = 0.0609x – 5.6998
R2 = 0.9985
20
15
10
0
150 200 250 300 350 400 450 500
Surface area, m2/g
Figure 14.18 Retention factor (k) depending on surface area of silica packing. Measured
compound: 3. Mobile phase: acetonitrile–15â•›m M ammonium formate pH 5.0 (85/15, v/v).
of analytes. The retention of the betaine-like compounds 3 and 4 then is determined

by a mechanism different from the affinity of positively charged particles toward
silanol groups of the stationary phase.
The average retention values are the highest for Alltima column, followed by
Atlantis and Spherisorb ones. It correlates the surface areas of these stationary phases
and suggests that the surface chemistry factor plays a similar role on these three silicas.
The Waters Spherisorb NH2 column has an aminopropyl group without endcap-
ping. Aminopropyl-bonded stationary phases are protonated at neutral and acidic

mobile-phase pH values. For compounds that are anions in the mobile phase, this
offers the possibility of an anion-exchange mechanism in addition to hydrophilic
interactions. The ion-exchange nature of the separation was demonstrated in Figure
14.16d, the aqueous–acetonitrile ratio of the mobile phase was constant but the buf-
fer concentration was varied.
The formate counter ions might be adsorbed onto the positively charged amino
phase via electrostatic interaction. This could reduce electrostatic repulsion of the
positively charged analytes from the positively charged stationary phase, thus caus-
ing an increased retention.
TSK-Gel Amide-80 stationary phase bears carbamoyl groups attached to the
silica surface through an aliphatic carbon chain. The amide group is less reactive
compared to the amine and lacks its basicity.64
As shown in Figure 14.16e, the increase of ammonium formate concentration
from 1 to 5â•›m M results in a significant decrease of retention of compounds 1 and 2
as well as 5, 6, and 7. The retention of betaine-type substances 3 and 4 has little
or not affected by the change of ammonium formate concentration with Amide-80
stationary phase.
Similar to a silica phase, the decrease of retention time for compounds 1–2
and 5–7 could be attributed to the ion-exchange effect of surface silanol groups.6
Retention of betaine-like compounds 3 and 4 then is determined by a mechanism
different from the affinity of positively charged particles toward silanol groups of
the stationary phase.
ZIC-HILIC phase includes a –CH2N+(Me)2CH2CH2CH2SO3− ligand covalently
attached to porous silica. This ligand includes two oppositely charged functional
groups. Close proximity of these two charged moieties preclude their independent
interaction with analytes.
Zwitterionic stationary phase has the negatively charged sulfonic acid group
exposed to the mobile phase and the positively charged quaternary ammonium
group “hidden” inside the alkyl chain. Hence, the attractive interaction of any posi-
tive charge of the analyte with the sulfonic acid group of the zwitterionic station-
ary phase is stronger than the respective interaction between a negative charge of
the analyte with the positively charged group of the zwitterionic stationary phase.
Similarly, negatively charged analytes are more strongly repelled by the sulfonic
acid groups than the “hidden” ammonium groups are attracted. Altogether, this leads
to higher retention times for positively charged analytes, and shorter retention times
for negatively charged analytes.18
The dependence of retention on the ammonium formate concentration is shown
in Figure 14.16f. The retention of cations 1–2 and cationic esters 5–7 follows a very
similar pattern: an increase in buffer concentration leads to a decrease of retention.
Usually, it is explained by increasing competition for sorption sites when the number
of counterions in the mobile phase increases.
On the other hand, zwitterions behave differently. Retention of these compounds is
slightly influenced by the salt concentration. This indicates that on conditions of the
experiment ion-exchange role in the retention mechanism of zwitterions is insignificant.
The buffer concentration is of primary importance for adjusting retention of cat-
ions and cationic esters, but has little effect on zwitterions. The retention of the

cations and the cationic esters follows a dual (partition and ion exchange) mecha-
nism. Retention data of zwitterionic compounds suggest that the role of ion exchange
is less significant for this group of compounds. Concentration of ammonium formate
buffer makes selective impact on retention of the compounds, and the most signifi-
cant change is observed at lower buffer concentrations of less than 10â•›m M.
14.6 Temperature Effect with Various Stationary Phases

Mobile phase composition and column temperature are commonly used to control
adsorption behavior in RP chromatography. Several studies have been performed to
compare the effect of a modulation of the mobile phase composition to that of the
temperature.65–67 It is found that the temperature is a less effective operating parameter
for reducing retention compared to the solvent strength. Strong temperature effects are
evidenced only for ionizable compounds; therefore, manipulating of the temperature
is a useful tool mainly in ion-exchange and ion-pair chromatography.67 For neutral
components, a change of 4°C–5°C in temperature corresponds to approximately 1%
changes in composition of the mobile phase.65,66 But the temperature range of inter-
est is usually limited from the ambient temperature up to 50°C–60°C since classical
bonded silica stationary phases are thermally unstable and operating at higher temper-
ature requires robust, thermally stable, but expensive new generation columns. The use
of higher temperature can also cause decomposition of temperature-labile analytes.
However, combining solvent strength and temperature can have a powerful
effect on the separation in RP chromatography in terms of selectivity and effi-
ciency. The acceptance of importance of temperature is based on well-known
effects of the decrease of viscosity of the mobile phase and the increase in analyte
diffusivity. The first effect leads to lowering backpressure over the column, allow-
ing increased flow rates. This gives improvement of the process productivity,
which is directly related to the speed of the separation process. The consequence
of increased diffusivity is faster mass-transport kinetics, which is reported to
result in improvement of column efficiency.
The relationship between retention factor (k) and column temperature (T) in
RPLC is often described by the van’t Hoff equation:
∆H 0 ∆S 0
ln k = − + + ln φ
RT R
where
ΔH0 and ΔS 0 are retention enthalpy and entropy
R is gas constant
ϕ is phase ratio
If the retention of polar compounds in HILIC is through partitioning between the

mostly organic mobile phase and the water-rich liquid layer on the stationary phase
surface as proposed by Alpert,6 the van’t Hoff equation should apply to HILIC.
These propositions are fulfilled just when ϕ = constant.

There are studies, where the relationship between retention factor (k) and water
content in the mobile phase explains partitioning and/or adsorptive retention of ana-
lytes.68 Besides, slopes of the relation graph of analytes are higher for analytes hav-
ing more polar interactive sites, and typically negative enthalpy values are higher for
more polar phases and more polar analytes.
In ion-exchange HPLC, enthalpy changes generally are large and dominate the
retention interaction. Enthalpy changes can be either positive or negative (exothermic
or endothermic reactions), and in ion exchange, generally cause significant changes
in retention. Thus, separation at higher temperature in ion-exchange chromatogra-
phy is quite commonly found in the literature reports.
In the RP-HPLC, retention interactions are usually exothermic with relatively
small changes in enthalpy. This usually manifests itself in only modest decreases of
retention time as temperature increases in contrast to the changes in ion-exchange
chromatography.69–71
When elevated temperature is discussed in chromatographic circles, it is focused
on reduction of system backpressure followed by various operational benefits. Lower
pressure causes less stress on the hardware (valves, seals, etc.) and also enables the
use of higher flow rates, if desired. This becomes important when some of the mod-
ern columns packed with smaller sized stationary phase particles are used that have
a higher backpressure than the more commonly used 5â•›μm material.
Column temperature can tamper analyte transferring enthalpy from mobile to sta-
tionary phases, especially when ionic interactions are involved.66 Dong and Huang
studied basic compounds, having negative slopes of van’t Hoff curves.72 This fact
means that retention times behave non-normally, increasing with raising of tempera-
ture. The authors observed simultaneous increase in column plate number, thus rais-
ing column temperature improved the resolution of the tested compounds.
We want to present a study of the relationship between the column temperature
and retention of both mildronate and its related impurities. There is a number of pub-
lications dealing with studies of column temperature effects in HILIC,3,4,20,73,74 but no
information was found for the compounds as described in Table 14.1.
This chapter presents an investigation on the relationship between chromato-
graphic retention and column temperature effect on various stationary phases (silica
and zwitterionic sulfobetaine, columns 3, 5, 6 in Table 14.2).
Temperature effects were determined for three mobile phases: (1) 10% water with
0.1% formic acid/90% acetonitrile; (2) 10% water with 0.1% formic acid/90% metha-
nol; and (3) 15% 5â•›m M ammonium formate (pH 5)/85% acetonitrile. In this study,
the temperature effect on retention was investigated by varying column temperature
from 30°C to 55°C, and the retention data for the test compounds were used to con-
struct van’t Hoff plots for the three columns mentioned above. Figures 14.19 through
14.21 show the retention behavior of compounds 1–7 as a function of temperature
(log k vs. 1/T).
With a silica column and acetonitrile mobile phases, the retention increased as the
temperature was increased which indicates either a positive enthalpy for interaction
of the analytes with the stationary phase and/or substantial entropy contributions.
This temperature effect is opposite to what is typically observed under RP conditions
in HPLC, i.e., decreasing retention with increasing temperature. This result provides

20
15
ln k
10
7 6 5 4 3 2 1
5
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(a) 1/T
3.8
3.6
ln k
3.4
3.2
7 6 5 4 3 2 1
3
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(b) 1/T
3.00
2.50
2.00
ln k
1.50
1.00
0.50
7 6 5 4 3 2 1
0.00
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(c) 1/T
Figure 14.19 Retention as function of temperature. Mobile phase: 90% ACN. Stationary
phases: (a) Atlantis HILIC Silica, (b) Waters Spherisorb Silica, and (c) ZIC HILIC.

1.4
1.2 7 6 5 4 3 2 1
0.8
ln k
0.6
0.4
0.2
0
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(a) 1/T
5
7 6 5 4 3 2 1
4
ln k
2
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(b) 1/T
2.0
7 6 5 4 3 2 1
1.5
ln k
1.0
0.5
0.0
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(c) 1/T
Figure 14.20 Retention as function of temperature. Mobile phase: 90% MeOH. Stationary
phases: (a) Atlantis HILIC Silica, (b) Waters Spherisorb Silica, and (c) ZIC HILIC.
other routes for improving selectivity via temperature control. Raising temperature
can increase retention, thus improving R values. Also, higher temperatures result in
lower viscosity and faster mass transfer which can decrease peak widths to provide
another means for improving resolution depending on the extent of retention and dif-
fusion in the stationary phase.75

5
7 6 5 4 3 2 1
4
3
ln k
0
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(a) 1/T
5
7 6 5 4 3 2 1
4
3
ln k
0
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(b) 1/T
5
7 6 5 4 3 2 1
3
ln k
0
0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
(c) 1/T
Figure 14.21 Retention as function of temperature. Mobile phase: 15% 5â•›m M ammo-
nium formate (pH 5), 85% ACN. Stationary phases: (a) Atlantis HILIC Silica, (b) Waters
Spherisorb Silica, and (c) ZIC HILIC.

Linear relationship between k and temperature was observed in the temperature

range of 30°C–55°C for the test compounds on all columns with R2 ranging from
0.950 to 0.998. This is similar to what is usually observed in RPLC,76,77 even though
nonlinear van’t Hoff plots have also been reported and attributed to the tempera-
ture dependency of both enthalpy (ΔH0) and phase ratio (ϕ) in RPLC.78,79 Generally,
when nonlinear van’t Hoff plots are observed, it is assumed that the enthalpy and
entropy change with temperature. This study did not find any nonlinear behavior of
HILIC sorbents.
References to temperature in HPLC may refer to the temperature of the column
packing, the mobile phase, or the column-heating device. The temperature inside the
column is of primary interest, but difficult to be measured without physically affect-
ing the column. A simpler way to determine the column temperature is to use a flow-
through thermocouple to measure the temperature of the mobile phase entering and/
or exiting the column. An average of the inlet and outlet mobile-phase temperature
will give a reasonable value for the average column temperature and information
about the temperature gradient along the column. The column oven temperature gen-
erally will differ from the temperature inside the column, but as long as the column
oven temperature is constant, this measurement will provide a suitable reference
value for most applications.
MeOH is a protic solvent as an organic component in the mobile phase. When
MeOH was employed as the organic part of mobile phase instead of ACN, the reten-
tion of seven test compounds exhibited disorderly changes with increasing tem-
perature, as shown in Figure 14.20a through c. Some of them displayed increasing
retention with elevated temperature, others showed the opposite. The different tem-
perature effects are the result of a multiple retention mechanisms. In addition to
hydrophilic and ion-exchange interactions, solvent–analyte and solvent–stationary-
phase hydrogen bonding are also involved.19
All these results show that temperature plays a more complex role on the reten-
tion of analytes in MeOH than in ACN due to the multiplicity and complexity of the
retention mechanisms.
With temperature increase, the trend of retention increase was observed at pH
2.1; further the retention values of the analytes significantly decrease of retention
was achieved at pH 5.0. These results reflect a transitional retention mechanism with
the pH change. As the buffer pH varies from 2.1 to 5.0, the ionization of the silanols
on the silica surface gradually increases and the analytes are kept in a protonated
state,80 offering more opportunity for cation exchange. The major mechanism gov-
erning the retention is transformed from hydrophilic to ion-exchange interaction.
When ion exchange becomes the dominant mechanism, the pKa of the analyte plays
an important role in affecting the retention and selectivity. According to the previous
work,81,82 an increase in temperature can cause a reduction of protonated species.
Thus, the retention of the test compounds at higher temperature would decrease as
observed in Figures 14.19 and 14.21 due to the weakened ion-exchange contribution.
The impact of temperature on retention for various analytes could be different.
As discussed above, at pH 5, because ion-exchange interaction is stronger, the pKa
of the bases becomes an important parameter governing retention. As proposed in
previous literature,81,82 the pKa of analytes decreases with increasing temperature,

and the smaller the pKa of the base is, the more it decreases. Since the pKa value of
compound 1 (pKa 9.80) is smaller than that of compound 2 (pKa 6.9), a greater reduc-
tion of the protonated species must be obtained for compound 1. This implies that the
ion-exchange contribution to the retention factor is smaller.
It has been found that temperature plays a different role when the type and content
of the organic modifier or pH in the mobile phase are changed. For the seven test
compounds examined in an organic-rich environment, temperature response varied
with the type of the organic solvent.
Column temperature is an important parameter that affects the retention of polar
compounds in HILIC. The effect of column temperature on the retention of hydrazine-
based compounds was investigated on silica and zwitterionic sulfobetaine columns in
the temperature range of 30°C–55°C with different organic solvents and different pHs.
At an increased temperature on silica column, the retention factor depends (increases
or decreases) on the organic solvent (ACN, MeOH) and pH. But on zwitterionic sul-
fobetaine sorbent, the retention factor changes significantly depending on the type of
organic solvents and pH.
14.7 Application Results and Method Validation

As a result of this systematic investigation of different columns and mobile phases,
the HILIC method using a zwitterionic sulfobetaine stationary phase was practically
used to determine six impurities in the hydrazine-based pharmaceutical compound
mildronate. The validation was performed with ZIC-HILIC column with mobile
phase 85% ACN/15% 5â•›m M ammonium formate (pH 5.0) with LC-MS-MS accord-
ing to the ICH Guidelines.83–86 The validation parameters included specificity, accu-
racy, linearity, precision, limit of detection (LOD), and limit of quantitation (LOQ).
The method specificity has been demonstrated by fair separation of all the known
and potential degradation products. At the start of the study, an assumption was
made that any impurity may be present at concentrations not more that 0.1% of assay
concentration (the preliminary quality specification). The concentration of the main
compound (mildronate) 1.0â•›mg/mL was used during the validation.
Validation results for the impurities method are summarized in Table 14.3. The
method was tested on real samples of mildronate, produced in JSC “Grindeks.”
Table 14.3
Summary of Results of Validation
Precision, Mean LOD, LOQ,
Compound RSD% (n = 6) Accuracy, % R2 % of Assay % of Assay
1 9.4 96.9–104.7 0.991 0.003 0.01
2 3.3 105.7–114.0 0.997 0.0003 0.001
3 7.7 111.2–116.4 0.998 0.0005 0.002
5 2.7 102.1–114.4 0.992 0.0004 0.001
6 2.1 104.2–109.6 0.998 0.0001 0.0004
7 2.7 103.1–108.4 0.998 0.00006 0.0002

MRM of 7 Channels ES+

189 > 58
100
%

100 175 > 58
%
0
100 161 > 59
%

100 147 > 59
46.15 3.77 × 106
%
100 13.59
115 > 72
7.12 × 103
%
100 20.40 MRM of 7 Channels ES+

75 > 59
1.53 × 105
%

60 > 45
6.15 × 103
%
0
10.00 20.00 30.00 40.00 50.00
Time
Figure 14.22 MRM chromatogram of technical batch 2374-06 of mildronate. Column:

ZIC-HILIC; Concentration of mildronate 1â•›mg/mL. Mobile phase: acetonitrile–5â•›m M ammo-
nium formate pH 5.0 (85/15, v/v). Specific transitions of compounds 1–7 (from bottom to top).
The validated method was used to analyze technical batch (Figure 14.22) and
commercial batch (Figure 14.23). Figure 14.22 shows that technical product contains
visible quantities of related substances, but Figure 14.23 shows only traces of them.
Comparative results of analysis are presented in Table 14.4. The developed method
is suitable to control the quality of commercial mildronate API.

100 MRM of 7 Channels ES+

189 > 58
%

175 > 58
%

100
161 > 59
%
0
50.00

100
45.91 147 > 59
3.70 × 106
%

100 115 > 72
%
20.61
75 > 59
795
%

60 > 45
1.84 × 103
%
0
10.00 20.00 30.00 40.00 50.00
Time
Figure 14.23 MRM chromatogram of production batch 990208 of mildronate. Column:

ZIC-HILIC; Concentration of mildronate 1â•›mg/mL. Mobile phase: acetonitrile–5â•›m M
ammonium formate pH 5.0 (85/15, v/v). Specific transition of compound 1–7 (from bottom
to top).

Table 14.4
Quantitative Results of Analyses of Technical
and Commercial Mildronate, in Percent
Technical, Commercial,
Compound Batch 2374-06 Batch 990208
1 0.20 0.01
2 0.35 Traces (∼0.0003)
3 0.02 Not found
5 Not found Not found
14.8 Summary
HILIC is a valuable alternative to RP and ion-pair chromatography. It has been shown
that HILIC separations of quaternary hydrazine derivatives are possible using both
specialized HILIC type and common silica columns. Specialized columns show
weaker retention and are more suitable for separations at high concentrations of aceto-
nitrile. Variation of polar stationary phase is used as a tool to change the selectivity of
separation for the test analytes. It was shown that the variation of acetonitrile concen-
tration is effective to adjust retention but does not significantly influence the selectivity.
The pH and buffer concentration are of primary importance for adjusting reten-
tion of cations and cationic esters, but has little effect on zwitterions. The retention
of cations and cationic esters follow regularities expected for a dual (partition—ion
exchange) mechanism. Retention data of zwitterionic compounds suggest that the
role of ion exchange is less significant for this group of substances. HILIC is an
efficient tool in method development for extremely polar and ionic hydrazine deriva-
tives. ZIC-HILIC column was tested even more, the method validation having been
performed for quantification of related substances. The validated method is suitable
for purity analysis of mildronate drug substance.
Acknowledgments
The successful completion of this work owes a great deal to cooperation provided by
Imants Davidsons and Dr. Osvalds Pugovics, Dr. Helena Kazoka, JSC “Grindeks,”
University of Latvia and Latvian Organic Institute, and we have pleasure in express-
ing our gratitude to them.
References
1. Strege, M.A., Anal. Chem., 70 (1998), 2439.
2. Olsen, B.A., J. Chromatogr. A, 913 (2001), 113.
3. Guo, Y. and Gaiki, S., J. Chromatogr. A, 1074 (2005), 71.
4. Guo, Y., J. Liq. Chromatogr. Relat. Technol., 28 (2005), 497.
5. Chrums, S.A., J. Chromatogr. A, 720 (1996), 75.

6. Alpert, A.J., J. Chromatogr., 499 (1990), 177.

7. Rückert, H. and Samuelson, O., Sven. Kem. Tidskr., 66 (1954), 337.
8. Samuelson, O. and Sjöström, E., Sven. Kem. Tidskr., 64 (1952), 305.
9. Hemström, P. and Irgum, K., J. Sep. Sci., 29 (2006), 1784.
10. Strege, M.A., Am. Pharm. Rev., 2 (1999), 53.
11. Strege, M.A., Stevenson, S., and Lawrence, S.M., Anal. Chem., 72 (2000), 4629.
12. Garbis, S.D., Melse-Boonstra, A., West, C.E., and vanBreemen, R.B., Anal. Chem., 73
(2001), 5358.
13. Tolstikov, V.V. and Fiehn, O., Anal. Biochem., 301 (2002), 298.
14. Person, M., Hazotte, A., Elfakir, C., and Lafosse, M., J. Chromatogr. A, 1081 (2005),
174.
15. Schlichtherle-Cerney, H., Affloter, M., and Cerny, C., Anal. Chem., 75 (2003), 2349.
16. Baughman, T.M., Wright, W.L., and Hutton, K.A., J. Chromatogr. B, 852 (2007), 505.
17. Peru, M., Kuchta, L., Headley, V., and Cessna, J., J. Chromatogr. A, 1107 (2006), 152.
18. Xuan, Y., Scheuermann, B., Meda, R., Hayen, H., Wiren, N., and Weber, G.,
J. Chromatogr. A, 1136 (2006), 73.
19. Li, R.P. and Huang, J.X., J. Chromatogr. A, 1041 (2004), 163.
20. Liu, M., Chen, E., Ji, R., and Semin, D., J. Chromatogr. A, 1188 (2008), 255.
21. Alpert, A., J. Anal. Chem., 80 (2008), 62.
22. Boersema, P.J., Mohammed, S., and Heck, A.J.R., Anal. Bioanal. Chem., 391 (2008),
151.
23. Vikingsson, S., Kronstrand, R., and Josefsson, M., J. Chromatogr. A, 1187 (2008), 46.
24. Yoshida, T. Anal. Chem., 69 (1997), 3038.
25. McCalley, D.V., J. Chromatogr. A, 1171 (2007), 46.
26. Gustavsson, S., Samskog, J., Markides, K., and Langstrom, K., J. Chromatogr. A, 937
(2001), 41.
27. Dunphy, R. and Burinsky, D.J., J. Pharm. Biomed. Anal., 31 (2003), 905.
28. Guo-ying, C., Run-feng, L., Xin, H., and Chun-hua, S., Chin. Pharm. J., 40 (2005), 864.
29. Run-feng, L., Guo-ying, C., Xin, H., and Jun-ren, Z., Chin. J. Pharm. Anal., 26 (2006),
358.
30. Ivanovskaya, E.A., Anisimova, L.S., Belikhmaer, Ya.A., Koshelskaya, O.A., and
Sokolova, A.A., Pharm. Chem. J., 29 (No. 3) (1995), 219.
31. Yun-Feng, L., Xin, H., and Kai-Shun, B., J. Chromatogr. B, 852 (2007), 35.
32. Sahartova, O., Shatz, V., and Kalvins, I., J. Pharm. Biomed. Anal., 11 (1993), 1045.
33. Hmelnickis, J., Pugovics, O., Kazoka, H., Viksna, A., Susinskis, I., and Kokums, K.
J. Pharm. Biomed. Anal., 48 (2008), 69.
34. Liu, M., Ostovic, J., Chen, E.X., and Cauchon, N., J. Chromatogr. A, 1216 (2009), 2362.
35. Jagota, N.K., Chetram, A.J., and Nair, J.B., J. Pharm. Biomed. Anal., 16 (1998), 1083.
36. Carlin, A., Gregory, N., and Simmons, J., J. Pharm. Biomed. Anal., 17 (1998), 885.
37. Gyllenhaal O., Groenberg, I., and Vessman, J., J. Chromatogr., 511 (1990), 303.
38. Selim, S. and Warner, C.R., J. Chromatogr., 166 (1978), 507.
39. Timbrell, J.A., Wright, J.M., and Smith, C.M., J. Chromatogr., 138 (1977), 165.
40. Seifart, H.I., Gent, W.I., Parkin, D.P. et al., J. Chromatogr. B, 674 (1995), 269.
41. Abdou, H.M., Medwick, T., and Bailey, L.C., Anal. Chim. Acta, 93 (1977), 221.
42. Kirchherr, H., J. Chromatogr., 617 (1993), 157.
43. A Practical Guide to HILIC. SeQuant AB, Umea, Sweden, April 2006.
44. Snyder, L.R., J. Chromatogr., 255 (1983), 3.
45. Scott, R.P.W., Adv. Chromatogr., 20 (1982), 169.
46. Vervoort, R.J.M., Ruyter, E., Debets, A.J.J. et al., J. Chromatogr. A, 931 (2001), 67.
47. Grumbach, E.S., Wagrowski-Diehl, D.M., Mazzeo, J.R. et al., Application notes
Hydrophilic Interaction Chromatography Using Silica Columns for the Retention of Polar
Analytes and Enhanced ESI-MS Sensitivity, Waters Corporation, www.waters.com

48. Castells, R.C., Castells, C.B., and Castillo, M.A., J. Chromatogr. A, 775 (1997), 73.
49. Zhou, T. and Lucy, C.A., J. Chromatogr. A, 1123 (2008), 8.
50. Jandera, P., J. Sep. Sci., 31 (2008), 1421.
51. McKeown, A.P., Euerby, M.R., Lomax, H. et al., J. Sep. Sci., 24 (2001), 835.
52. Hu, W., Takeuchi, T., and Haraguchi, H., Anal. Chem., 65(17) (1993), 2204.
53. Hu, W., Langmuir, 15(21) (1999), 7168.
54. Hu, W. and Haddad, P.R., Trends Anal. Chem., 17(2) (1998), 73.
55. Okada, T. and Patil, J.M., Langmuir, 14(21) (1998), 6241.
56. Cook, H.A., Hu, W., Fritz, J.S. et al., Anal. Chem., 73 (2001), 3022.
57. Cook, H.A., Diconoski, G., and Haddad, P.R., J. Chromatogr. A, 997(1–2) (2003), 13.
58. Risley, D.S. and Pack, B.W., LC/GC North America, 24(8) (2006), 12.
59. Jonsson, T. and Appelblad, P., LCGC, 17(Suppl.) (2004), 72.
60. Jiang, W., Fischer, G., Girmay, Y. et al., J. Chromatogr. A, 1127 (2006), 82.
61. Guo, D.C., Mant, C.T., Taneja, A.K. et al., J. Chromatogr., 359 (1986), 499.
62. Pack, B.W. and Risley, D.S., J. Chromatogr. A, 1073 (2005), 269.
63. Wang, X., Li, W., and Rasmussen, H., J. Chromatogr. A, 1083 (2005), 58.
64. Normal Phase Chromatography: TSKgel Amide-80, Tosoh Bioscience 2008 Laboratory
Products Catalogue, www.separations.us.tosohbioscience.com
65. Chen, M.H., J. Chromatogr. A, 788 (1991), 51.
66. Dolan, J.W., J. Chromatogr. A, 965 (2002), 195.
67. Kiriden, W., Poole, C.F., and Koziol, W.W., J. Chromatogr. A, 1060 (2004), 177.
68. Wu, J.Y., Bicker, W., and Lindner, W., J. Sep. Sci., 31 (2008), 1492.
69. Melander, W.R., Stoveke, J., and Horvath, Cs., J. Chromatogr., 185 (1979), 111.
70. Melander, W.R., Nahum, A., and Horvath, Cs., J. Chromatogr., 185 (1979), 129.
71. Nahum, A. and Horvath, Cs., J. Chromatogr., 203 (1981), 53.
72. Dong, L. and Huang, J., Chromatographia, 65 (2007), 519.
73. Guo, Y. and Huang, A., J. Pharm. Biomed. Anal., 31 (2003), 1191.
74. Tanaka, H., Zhou, X., and Masayoshi, O., J. Chromatogr. A, 987 (2003), 119.
75. Pesek, J., Matyska, T., Fisher, M., and Sana, R., J. Chromatogr. A, 1204 (2008), 48.
76. Philipsen, A., Claessens, A., Lind, H., Klumperman, B., and German, L., J. Chromatogr.
A, 790 (1997), 101.
77. Kayillo, S., Dennis, R., Wormell, P., and Shalliker, A., J. Chromatogr. A, 967 (2002),
173.
78. Cole, L. and Dorsey, J., Anal. Chem., 64 (1992), 1317.
79. Chester, T. and Coym, J., J. Chromatogr. A, 1003 (2003), 101.
80. Nawrocki, J., Rigney, R.P., McCormick, A., and Carr, W., J. Chromatogr. A, 657 (1993),
229.
81. Buckenmaier, S.M.C., McCalley, D.V., and Euerby, M.R., J. Chromatogr. A, 1060
(2004), 117.
82. Buckenmaier, S.M.C., McCalley, D.V., and Euerby, M.R., J. Chromatogr. A, 1026
(2004), 251.
83. ICH Guideline Q2A: Text on Validation of Analytical Procedures: Terms and defini-
tions, International Conference on Harmonization, Red. Reg. (60 FR 11260), March 1,
1995.
84. ICH Guideline Q3A; Impurities in New Drug Substances, International Conference on
Harmonization, Fed. Reg. (68 FR 6924), February 11, 2003.
85. ICH Guideline Q2B: Validation of Analytical Procedures: Methodology, International
Conference on Harmonization, Fed. Reg. (62 FR 27463), May 19, 1997.
86. ICH Guideline Q3B: Impurities in New Drug Products, International Conference on
Harmonization, Fed. Reg. (62 FR 27454), May 19, 1997.

15 HILIC Retention
Behavior and Method
Development for
Highly Polar Basic
Compounds Used in
Pharmaceutical Synthesis
Minhui Ma and Min Liu
Contents
15.1 Introduction................................................................................................... 345
15.2 HILIC Method for Purity and Stability Monitoring of 4-AMP.................... 347
15.2.1 Method Development......................................................................... 347
15.2.1.1 HILIC-Column Screening.................................................. 347
15.2.1.2 Mobile-Phase Components................................................. 348
15.2.1.3 Column Temperature.......................................................... 356
15.2.2 Method Validation............................................................................. 358
15.3 HILIC-CLND Method for Hydrazine Residues in an Intermediate............. 360
15.3.1 Method Development.........................................................................360
15.3.1.1 Alcohol as a Weak Eluent...................................................360
15.3.1.2 Combined ZIC and HILIC Modes...................................... 361
15.3.1.3 Type of Acid Additive......................................................... 362
15.3.1.4 Type of Buffer and Ionic Strength...................................... 365
References............................................................................................................... 369
15.1 Introduction
A large proportion of pharmaceuticals are basic or alkaline organic compounds.
The analysis of basic compounds has traditionally presented significant challenges
to reversed-phase (RP) HPLC due to severe peak tailing caused by secondary inter-
actions between charged basic analytes and active residual silanols on RP station-
ary phases.1 The peak-tailing problem has been largely addressed in the 1990s with
345

the development of the new generation of RP-HPLC columns based on high-purity

Type B silica.2,3 RP-HPLC methods with high efficiency and good peak shape can
now be readily developed for basic compounds. The exception is a group of basic
compounds that are highly polar and consequently poorly retained on RP columns
even with 100% aqueous mobile phase. In pharmaceutical analysis, many sub-
stances, such as starting materials, reaction reagents, intermediates, process impuri-
ties, or drug substance may fall into this group.
Hydrophilic interaction chromatography (HILIC) is an alternative technique for
the analysis of polar compounds.4 It uses a polar stationary phase (such as silica or
polar stationary phases covalently bonded on silica) in conjunction with a mobile
phase containing a higher percentage of organic solvent (often acetonitrile). In
HILIC, the elution order is analogous to nonaqueous normal-phase LC. It can be
regarded as a variant of normal-phase LC. Early applications of HILIC started in the
1970s and focused mainly on the analysis of sugars and oligosaccharides.5–8 HILIC
applications have since expanded to include peptides,4,9–12 proteomics,13 and small
polar compounds in biological fluids,14–16 natural products,16,17 pharmaceuticals,18–30
and others. Examples of pharmaceutical applications included the analyses of start-
ing materials with their impurities,18–20 intermediates or drug substances with their
reaction reagent residues,21 process impurities,22–26 and counterions.27–29 Since 2003
a number of review articles on both the fundamentals and the applications of HILIC
have been published.11–16,31–37 A large number of new HILIC columns have also been
introduced by column manufacturers in recent years.38
The analysis of basic compounds by mixed HILIC and cation exchange mode was
first reported by Jane in 1975,39 in which the separation of a wide range of basic drugs
of forensic interest was achieved on an unmodified silica column using methanol–
aqueous ammonia/ammonium nitrate as mobile phase. Since then, similar methods
using silica columns and methanol–aqueous buffer at neutral or alkaline pH as mobile
phase have been utilized as an alternative to RP-HPLC for the analysis of basic phar-
maceutical compounds.30,40–43 More recent applications of HILIC methods in phar-
maceutical analysis use not only silica but also other polar stationary phases, and the
mobile phases are mostly acetonitrile and aqueous buffer at acidic or neutral pH.44,45
In this chapter, we present two case studies of systematic HILIC method devel-
opment for the analysis of highly polar basic compounds. The first method was
developed for purity and stability monitoring of a pharmaceutical starting material
4-(aminomethyl)pyridine (4-AMP)19 and the second for simultaneous determination
of genotoxic hydrazine and 1,1-dimethyl hydrazine residues in a pharmaceutical inter-
mediate.21 Although the retention of polar compounds in HILIC has been proposed
to mainly involve a partition mechanism between the largely organic mobile phase
and a water-enriched layer partially immobilized on the surface of the stationary
phase,4,8,46 the exact retention mechanism for HILIC is still open to debate.11,31,32,34,36
As a result, method development guidelines for HILIC are generally unavailable.
Therefore, a systematic approach to HILIC method development was carried out,
which involved the evaluation and optimization of stationary phase, mobile phase
strength, type of organic modifier, buffer type, buffer pH, ionic strength, and column
temperature. Also, many highly polar basic compounds contain nitrogen but lack UV
chromophore. In the second case study, alcohol is exploited to replace acetonitrile as

HILIC Retention Behavior and Method Development 347
an organic modifier in the mobile phase, which enables the use of chemiluminescent
nitrogen detection (CLND). Both methods have been validated and used to support
pharmaceutical synthesis.
15.2 HILIC Method for Purity and Stability

Monitoring of 4-AMP
4-AMP is a commonly used starting material for the synthesis of active pharmaceu-
tical ingredient (API). A stability-indicating analytical method is required to monitor
its purity and stability. The purity method originally used for lot release by suppliers
was a gas chromatography method with flame ionization detection (GC-FID). But
the suitability of this GC-FID method for stability monitoring had not been properly
evaluated or challenged until discoloration and solidification were noticed after the
starting material was stored for an extended period of time at room temperature. The
GC-FID method failed to provide any meaningful information for the investigation.
Therefore, the degradation products were suspected to be nonvolatile. The separa-
tion effort using RP-HPLC was not successful because 4-AMP and its related com-
pounds are highly polar and poorly retained. An ion-pair HPLC method developed
in our lab was able to detect the degradation products, but was not compatible with
mass detector for the identification of the unknown degradation products. For this
reason, a MS compatible HILIC method was developed.
15.2.1 Method Development

15.2.1.1 HILIC-Column Screening
As in RP-LC, different column chemistries are available for HILIC, which provide
different selectivity. An extensive number of HILIC columns were screened. Table
15.1 shows the physical characteristics of the HILIC columns used in this study. For
the column screening, a suitable sample mixture as shown in Table 15.2 was prepared
by spiking potential impurities 4-methylpyridine (4-MP), 2-(aminomethyl)pyridine
(2-AMP), and 3-(aminomethyl)pyridine (3-AMP) at 1% level (5â•›μg/mL) into an aged
4-AMP sample that contained degradation products (Degradant-1 and Degradant-2).
All standards and samples were prepared in acetonitrile/water (90/10, v/v). The con-
centration for the main component 4-AMP was 0.5â•›mg/mL. All experiments for this
study were carried out on an Agilent 1100 series HPLC system with UV detection at
254â•›nm. Toluene was used as a void volume marker.
Column-screening experiments were performed under the same conditions and
the chromatograms are shown in Figures 15.1 and 15.2 for the seven bonded polar
stationary phases and the five unmodified silica stationary phases, respectively.
Under the screening conditions used, only two bonded polar phases (TSK-Gel
Amide-80 and ZIC-HILIC) showed complete separation of all potential impurities
and degradation products from the main peak (4-AMP) while satisfactory separation
were obtained from all five unmodified silica columns. Those results can partially
explain why many published separations of strongly basic small molecules in HILIC
mode were achieved on unmodified silica phases.14,23,24,30,39–43,45,47

Table 15.1
Physical Characteristics of the HILIC Columns Used in This Study
Dimension Particle Surface Pore
Column Name Functionality (mm) Size (μm) Area (m2/g) Size (Å)
TSK-Gel Amide-80 Carbamoyl 250 × 4.6 5 350 80
ZIC-HILIC Sulfoalkylbetaine 150 × 4.6 5 140 200
zwitterionic
Develosil 100 Dihydroxypropane 250 × 4.6 5 350 100
Diol-5
Polyhydroxyethyl A Poly(2-hydroxyethyl 100 × 4.6 3 N/A 100
aspartamide)
Luna NH2 3-Aminopropyl 150 × 4.6 3 400 100
Luna CN Propylcyano 150 × 4.6 3 400 100
Hypersil Gold PFP Pentafluorophenylpropyl 150 × 4.6 3 220 170
Luna Silica (2) Bare silica 150 × 4.6 3 400 100
Kromasil Silica Bare silica 150 × 4.6 3.5 340 100
Atlantis HILIC Bare silica 150 × 4.6 3 330 100
Alltima HP HILIC Bare silica 150 × 4.6 3 230 120
HyPurity Silica Bare silica 150 × 4.6 3 200 190
Source: Adapted from Liu, M. et al., J. Chromatogr. A, 1188, 255, 2008. With permission from Elsevier.
It is interesting to note that similar separation profiles were observed with all
unmodified silica phases as well as the two bonded polar phases (TSK-Gel Amide-80
and ZIC-HILIC) except for the poor separation between the two degradants. The
retention times increased in the order of analyte hydrophilicity: 4-MP, 2/3/4-AMP,
and Degradant-1/-2, as predicted from the log D values calculated using the ACD
software as shown in Table 15.2. The much shorter retention of 2-AMP compared
with the regioisomers 3- and 4-AMP was somewhat surprising because similar log D
values should result in similar retention if liquid–liquid partition is the only retention
mechanism. Such difference may imply the significant contribution of specific sur-
face interaction to their retention. The separations on all five unmodified (Type B)
silica columns are very comparable. The increase in retention time correlated well
with the increase in specific surface area. High comparability among Type B silica
columns in the HILIC mode has been reported recently for the separation of a mix-
ture of acidic, neutral, and strongly basic compounds.45 Judging from the slightly
improved peak shape and separation between the two degradants, the Atlantis HILIC
column was selected for further method development.
15.2.1.2 Mobile-Phase Components

Similar to RP-HPLC method development, mobile-phase components in the HILIC
mode can also significantly influence analyte retention and selectivity. In this study,
mobile phase strength, pH, and ionic strength are evaluated.

HILIC Retention Behavior and Method Development
Table 15.2
Structure and Physical Properties of 4-AMP and Its Related Compounds
Compound 1 2 3 4 5 6
4-MP 2-AMP 3-AMP 4-AMP Degradant-1 (degradant) Degradant-2 (degradant)
(synthetic (isomeric (isomeric (main
impurity) impurity) impurity) compound)
Molecular 93.13 108.14 108.14 108.14 305.38 305.38
weight
Structure CH3 H2N N N
H2N
H2N
N N
N N H2N H2N
N N
H H
N N
N N
pKa 5.94 8.70 8.34 7.81 5.65 5.65
2.18 2.27 2.91 5.38 5.38
5.32 5.32
3.71 3.71
0.23 0.23
Log D at −1.18 −3.56 −3.56 −3.70 −4.76 −4.76
pH 3.0
Note: The pKa (i.e., ww pKa ) and log D values predicted by ACD software version 8.0 are aqueous based and thus can be used as a reference to
polarity and ionization only. Compounds 5 and 6 are diastereomers which were trimerized from 4-AMP.
349
mAU
a
4 5/6
25 3
2
1 b
6/5
0 2 3 4
1
c
–25 4 Missing 5/6
2 3
1
4/6 d
–50 3
1 2 5
e
–75 2 3 4/5
1 6 f
–100 3/4 6 5
1 2
g
–125 5/6
1 2/3/4
–150
0 2 4 6 8 10 12 14 16 18 min
Figure 15.1 Separation of 4-AMP and its related compounds on different HILIC sta-
tionary phases. Column temperature is at 30°C. Mobile phases A and B are acetonitrile
and 50â•›m M ammonium formate (pH 3.0), respectively. Gradient is from 10% B to 50% B
in 20â•›m in. Flow rate is 1.5â•›m L/min and detection is at 254â•›nm. Compounds 1–6 are 4-MP,
2-AMP, 3-AMP, 4-AMP, Degradant-1, and Degradant-2, respectively. Columns: (a) TSK-
Gel Amide-80, (b) SeQuant ZIC-HILIC, (c) Developsil 100 Diol-5, (d) PolyHydroxyEthyl A,
(e) Phenomenex Luna-NH2, (f) Phenomenex Luna CN, (g) Hypersil Gold PFP. (Reproduced
from Liu, M. et al., J. Chromatogr. A, 1188, 255, 2008. With permission from Elsevier.)
mAU
40 a
5 6
3 4
20 2
1 6 b
3 4 5
0 2
1
3 5 6 c
–20 4
2
1 6 d
3 5
–40 4
2
1 5 6 e
–60 3 4
2
–80 1
2 4 6 8 10 12 14 16 18 min
Figure 15.2 Separation of 4-AMP and its related compounds on bare silica HILIC columns
from different manufactures. Chromatographic conditions are as in Figure 15.1. Compounds
1–6 are 4-MP, 2-AMP, 3-AMP, 4-AMP, Degradant-1, and Degradant-2, respectively. Columns:
(a) Phenomenex Luna Silica(2), (b) Phenomenex Kromasil Silica, (c) Waters Atlantis HILIC,
(d) Alltech Alltima HP HILIC, (e) ThermoElectron HyPurity Silica. (Reproduced from Liu, M.
15.2.1.2.1â•…Mobile-Phase Strength
The mobile-phase strength and the gradient program used for column screening as shown
in Figure 15.2c are acceptable for routine pharmaceutical analysis in terms of peak sepa-
ration and overall run time. To help understand the retention mechanisms of polar basic
compounds on unmodified silica in HILIC mode, mobile-phase strength was evaluated

over a range of 5%–40% water (95%–60% acetonitrile) with isocratic elution. The effect
of the stronger solvent concentration (volume fraction of water) in the mobile phase on
the retention factors (k′) showed the typical HILIC mode behavior (Figure 15.3A).
As mentioned in the introduction, the retention mechanisms in HILIC mode are
still not well understood. In RP-LC, retentions are believed to follow a partition
mechanism and can be described by a semi-empirical equation:3,31,34,48,49
log k ′ = log kw′ − Sϕ (15.1)

where
kw′ is theoretically equal to the retention factor in pure weaker solvent of the
mobile phase
φ is the concentration (volume fraction) of the stronger solvent in mobile phase
S is the slope of log k′ versus φ when fitted to a linear regression model
When Equation 15.1 is applied to the HILIC mode, the stronger solvent is water.
On the other hand, the retention of polar analytes in nonaqueous normal phase
systems containing a nonpolar and a polar solvents can be described by the mecha-
nism of localized adsorption through displacement of the adsorbed polar solvent (the
Snyder–Soczewinski model).3,31,34,50
A 
log k ′ = log kB′ –  S  log N B (or log k ′ = log kB′ – S′ log ϕ)) (15.2)
 nB 

where
NB is the mole fraction of the polar (or stronger) solvent B in a binary organic
mobile phase
kB′ is the retention factor if pure polar solvent B is used as mobile phase
AS and nB are the cross-sectional areas occupied by the solute molecule (analyte)
on surface and the polar solvent B, respectively
The ratio (AS/nB) is the number of solvent B molecules necessary to displace one
adsorbed molecule of the analyte. This equation can be written in terms of volume
fraction (φ) as well.
Figure 15.3B and C show the log–linear (i.e., log k′ versus volume fraction of
water) and the log–log (i.e., log k′ versus logarithm of water volume fraction) plots
corresponding to the partition and the adsorption models, respectively. Due to the
difficulties in determining very low retention factors with accuracy, 4-MP was not
included in the plots. For 4-AMP and its regioisomers 2-AMP and 3-AMP, good lin-
ear relationships were observed in the log–log plot (R2 > 0.998), indicating that adsorp-
tion contributes to their retention. However, linear relationships were not observed
for the two degradants in either of the two plots over the range of water concentra-
tions evaluated in this study. If we assume that liquid–liquid partition is indeed one
of the retention mechanisms involved in the HILIC mode and that it can be described
by Equation 15.1, the observed results for the two degradants suggest that neither
partition nor adsorption dominates their retention processes over the entire range of

200
150
100
k΄
50
0
0 10 20 30 40
(A) Volume fraction of water in eluent (%)
3.0
2.0
Log k΄
1.0
0.0
0 10 20 30 40
–1.0
(B) Volume fraction of water in eluent (%)
3.0
2.0
Log k΄
1.0
0.0
0.6 0.8 1.0 1.2 1.4 1.6
–1.0
(C)
Log (volume fraction of water in eluent)
Figure 15.3 (A) Plot of retention factor (k ′) versus the volume fraction of water in eluent.
(B) Plot of log k ′ versus the water volume fraction in eluent. (C) Plot of log k ′ versus logarithm
of the water volume fraction in eluent. An Atlantis HILIC column is used. Column tem-
perature is at 30°C. Isocratic runs with the mobile-phase compositions of acetonitrile/50â•›m M
ammonium formate (pH 3.0) from 60/40 to 95/5. Flow rate is 1.5â•›m L/min and detection is
at 254â•›nm. 4-MP (—♦—), 2-AMP (—◾—), 3-AMP (—▴—), 4-AMP (—▵—), Degradant-1
(— ○ —), Degradant-2 (—⦁—). (Reproduced from Liu, M. et al., J. Chromatogr. A, 1188, 255,

water concentrations evaluated in this study. Such nonlinear relationships for both
the log–linear and the log–log plots have been previously reported.31,51 This is not
surprising considering the complexity and multiplicity of the retention mechanisms
that may be involved in the HILIC mode. As a result, multiple linear regressions based
on quantitative structure–retention relationship models have been recently applied to
predict HILIC retentions.37,51 On the other hand, a closer examination of the results
from current study and those published by others31,44,51,52 suggests that strongly basic
compounds or weakly basic compounds but under low buffer concentrations are more
likely to follow a linear log–log relationship, probably due to stronger ionic interaction
with HILIC stationary phases.
15.2.1.2.2â•…Mobile-Phase pH
For basic and acidic compounds, their retention depends on both the mobile-phase pH
and their pKa values because the mobile-phase pH controls the ionization of the basic
and acidic compounds as well as the silanols on silica surface. In this study, mobile
phase A was pure acetonitrile and mobile phase B was 50â•›m M ammonium formate,
which was adjusted to the desired pH using formic acid. Because the mobile-phase pH
and pKa values of acidic or basic compounds are strongly influenced by the organic
content in the mobile phase, there is a need to distinguish the following terms53–56:
s
aqueous-buffer pH (ww pH) and mobile-phase buffer pH (w pH and ss pH) after mixing
with organic solvent. There are two different pH scales (sw pH and ss pH) for mobile
phase because pH electrode may be calibrated in water (W) or in the same aqueous–
organic solvent (S). The ss pH has the ordinary physical meaning and is directly related
to the ionized fraction of an analyte in the mobile phase, but it is more difficult to
measure than sw pH. To distinguish the variation of pKa, the same notations used for
s
pH should be applied, i.e., ww pK a , sw pK a , and s pK a .53
In this study, gradient elution was used and the mobile-phase sw pH values were not
measured. The effect of the measured aqueous buffer ww pH (in the range of 2.5–5.0)
on analyte retention and separation is shown in Figure 15.4. When ww pH was raised,
the retention times of 4-AMP and its regioisomers, 2-AMP and 3-AMP, increased,
whereas the retention times of 4-MP and the two degradants decreased. To have
a meaningful discussion about the pH effect, we need to make an estimate of the
s
more meaningful ss pH and s pK a values based on relevant information available in
the literature. In a recent study, McCalley measured sw pH of 0.1â•›M ammonium for-
mate buffer in 85% acetonitrile over the aqueous buffer ww pH range of 3.0–4.5.45 He
observed significant differences between ww pH and sw pH: ww pH 3.0 (sw pH ~ 5.2), ww pH 3.5
(sw pH ~ 5.7), and ww pH 4.5 (sw pH ~ 6.5). Under specific mobile-phase composition and
temperature, the sw pH value is related to ss pH by a constant value (δ), δ = ws pH − ss pH.
It has been reported that at 30°C the δ values are −1.709, −0.923, −0.580, and −0.388
for 90%, 80%, 70%, and 60% (v/v) acetonitrile, respectively.55 As shown in Figure
15.2c, 4-MP, the 2-/3-/4-AMP isomers, and the two degradants eluted at about 90%,
80%–75%, and 65%–60% acetonitrile of the gradient, respectively. Therefore, a sig-
nificant difference between ww pH and ss pH is expected especially for early eluting
peaks. On the other hand, it has been reported that the ss pK a value of pyridinium first
decreased gradually from the original aqueous ww pK a 5.23 to reach a minimum value
of 4.5 at 60% acetonitrile and then increased to a level close to the original aqueous

1
2
1
0
8
6
k΄
0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
Buffer
Figure 15.4 Effect of pH on the retention of 4-AMP and its related compounds on an
Atlantis HILIC column. Column temperature is 30°C. Mobile phase A is ACN while mobile
phase B is 50â•›m M ammonium formate with a pH ranging from 2.5 to 5.0. The gradient is
from 10% B to 50% B in 20â•›m in. Flow rate is 1.5â•›m L/min and UV detection at 254â•›nm.
4-MP (—♦—), 2-AMP (—▪—), 3-AMP (—▴—), 4-AMP (—▵—), Degradant-1 (— ○ —),
Degradant-2 (—⦁—). (Reproduced from Liu, M. et al., J. Chromatogr. A, 1188, 255, 2008.
w
wpK a value at about 90% acetonitrile.53,54 Therefore, the differences between ww pK a
and ss pK a are expected to be minimal to moderate for the analytes.
Based on the above information, the insensitivity of the 2-/3-/4-AMP retention
times to mobile-phase pH may be explained by the fact that they are stronger bases
with ss pK a values still significantly larger than the mobile-phase ss pH over the range
studied (ww pH 2.5–5.0). The small increase in retention over the pH range may be
due to the increase in ionization of those more acidic silanol groups because a large
increase in (bulk) silanol ionization on the Atlantis Silica stationary phase occurs
w s
only above w pH 9.0 (or w pH ~ 8.0) as reported by McCalley.45 On the other hand,
for the observed decrease in retention times of 4-MP and the two degradants, one
may assume that they are weaker bases with ss pK a values close to the mobile-phase
s w
s pH over the range studied (w pH 2.5–5.0). Thus, an increase in the mobile-phase pH
reduced their ionization, which resulted in shorter retention. The retention factor of
4-MP reached a minimum at ww pH 4.0 or higher while the retention factors of the two
degradants continued to decrease. This observation may be explained by the sig-
nificant difference in the mobile-phase ss pH experienced between 4-MP and the two
degradants during gradient elution. At ww pH 4.0 or higher, it is likely that almost all
4-MP exist in the neutral form in a mobile phase of ∼90% acetonitrile and with a ss pH
value much higher than ww pH.
15.2.1.2.3â•…Ionic Strength of Mobile Phase

For ionizable compounds, buffer concentration or ionic strength may significantly
influence their retention, if the retention mechanism involves ionic interaction.
Figure 15.5 shows the effect of ammonium formate buffer concentration at pH 3.0.

mAU
80
60
5 6
40 4
3
2
20 1
c 5 6
3 4
2
0
1
b
–20 3 4 5 6
1 2
a
–40
0 2 4 6 8 10 12 14 16 18 min
Figure 15.5 Effect of buffer ionic strength on the separation of 4-AMP and its related
compounds on Atlantis HILIC column. Column temperature is 30°C. Mobile phase A is ACN
and mobile phase B is 10â•›m M (a), 50â•›m M (b), and 100â•›m M (c) ammonium formate pH 3.0,
respectively. Gradient is from 10% B to 50% B in 20â•›m in. Flow rate is 1.5â•›m L/min and UV
detection at 254â•›nm. Compounds 1–6 are 4-MP, 2-AMP, 3-AMP, 4-AMP, Degradant-1, and
Degradant-2, respectively. (Reproduced from Liu, M. et al., J. Chromatogr. A, 1188, 255,
As buffer concentration varied from 10 to 100â•›m M, different retention behaviors were

observed for the analytes. No change in retention time was noticed for 4-MP over the
buffer concentration range, which may suggest that the molecule exists mostly in the
neutral form. This is consistent with what has been discussed in the previous section.
For 2-AMP and 3-AMP, their retention times were reduced with the increase in buf-
fer concentration. This is the typical behavior of ion-exchange retention mechanism,
which has been widely reported and well understood.3,24,40,45 The retention time
change for 2-AMP was larger than 3-AMP. Such retention time change appeared
to be minimal for 4-AMP. These results suggested that the contribution of ionic
interaction to their retention decreased in the order: 2-AMP > 3-AMP > 4-AMP,
which correlated to their basicity (Table 15.2). The retention behavior of the two
degradants was different. Their retention times increased with the increase in buffer
concentration, which has been referred to as a salting-out effect in the literature.57
Similar results were obtained in the isocratic mode (data not shown). This unusual
behavior has been previously reported by Guo and Gaiki for the retention of salicylic
acid, aspirin, and cytosine on four different columns in the HILIC mode.44 They
hypothesized that the buffer is preferentially dissolved in the water-rich liquid layer
on the stationary phase, which would result in an increase in volume or hydrophilic-
ity of the liquid layer and lead to stronger retention of the analytes by the partition
mechanism.
When buffer concentration was reduced, it was noticed that the tailing of 2-AMP
peak increased significantly, indicating a secondary retention mechanism. The tail-
ing factors at 100, 50, and 10â•›m M were 1.7, 1.9, and 2.2, respectively. After taking

into consideration of baseline drift, peak tailing, and selectivity, we selected the
50â•›m M buffer concentration for the final method.
15.2.1.3 Column Temperature

In liquid chromatography, column temperature has been increasingly recognized as
one of the important parameters for optimizing chromatographic method to increase
efficiency and to reduce analysis time. Column temperature can also influence ana-
lyte retention and selectivity because the distribution equilibrium constant (Keq) of
an analyte between the mobile phase and the stationary phase is temperature depen-
dent. For isocratic retention, the dependence of analyte retention factor, k′, on col-
umn temperature, T, is well known:
− ∆H o ∆S o
ln k ′ = + + ln β (15.3)
RT R
where
ΔHo and ΔSo are the enthalpy and entropy changes, respectively, associated with
the transfer of the analyte from the mobile phase to the stationary phase
R is the gas constant
β is the phase ratio of the column
The van’t Hoff plot of ln k′ versus 1/T is linear if the phase ratio, ΔHo and ΔSo do not
change significantly with temperature and if the retention mechanism remains the
same. The slope of the plot will give the enthalpy change of the transfer.
In RP-HPLC, the effect of temperature on retention and selectivity has been
widely studied.58–61 For nonionizable analytes, temperature generally has a minor
effect on retention and selectivity. It has been shown that on silica-based C18 col-
umn, a temperature increase of 5°C has roughly the same effect on retention as a
1% increase in acetonitrile concentration.62 For ionizable analytes, strong tempera-
ture effects on retention and selectivity are evidenced at mobile-phase pH close
to analyte pKa.63–66 In this case, the apparent enthalpy change obtained from the
van’t Hoff plot is not only determined by the enthalpies of transfer for the neutral
and the ionized species, but also by the enthalpies of the secondary equilibrium
processes (e.g., protonations for both analyte and buffer, ion pairing), as well as
the ratio of the conjugate base and acid forms of the buffering agent. As a conse-
quence, the van’t Hoff plots of the retention factors may not yield straight lines
depending on the relative magnitude of the pertinent enthalpy changes. Even a
negative slope was predicted by the theoretical model for certain combinations of
enthalpy changes.63,67 Under the limit conditions when the analyte is present only
as the neutral or the ionized form, the theoretical model predicted linear van’t Hoff
plots.63 Recent studies have taken the effect of organic solvents on buffer pKa and
pH into the consideration.67–69
In the HILIC mode, studies have been performed on temperature effect.32,43,44,47,70,71
Like in RP-LC, temperature can be a significant variable that assists in optimizing
chromatographic conditions. Such studies may also help understand retention mech-
anism. Similar to the findings in RP-LC, linear and nonlinear van’t Hoff plots with

mAU
6 5
3/4
175 2
1
70°C
150
60°C
125
50°C
100
40°C
75
30°C
50
20°C 5 6
25 3 4
2
1
10°C
0
(A) 0 2 4 6 8 10 12 14 16 18 min
4.0
3.0
2.0
ln k΄
1.0
0.0
–1.0
–2.0
0.0028 0.0030 0.0032 0.0034 0.0036
(B) 1/T (K)
Figure 15.6 (A) Effect of column temperature on the separation of 4-AMP and its related
compounds on Atlantis HILIC column. Column temperature is in the range of 10°C–70°C.
Mobile phases A and B are ACN and 50â•›m M ammonium formate pH 3.0, respectively.
Gradient is from 10% B to 50% B in 20â•›m in. Flow rate is 1.5â•›m L/min and UV detection is
at 254â•›n m. (B) Plot of the logarithm of retention factor (ln k′) versus the reciprocal of col-
umn temperature (1/T) under isocratic conditions. Mobile phase is 80/20 (v/v) ACN/50â•›m M
ammonium formate pH 3.0. Other conditions are as in (A). 4-MP (—♦—), 2-AMP (—▪—),
3-AMP (—▴—), 4-AMP (—▵—), Degradant-1 (— ○ —), Degradant-2 (—⦁—). (Modified
from Liu, M. et al., J. Chromatogr. A, 1188, 255, 2008. With permission from Elsevier.)
both positive and negative slopes have been observed in the HILIC mode.32,44,43,47,70,71
In this study, the temperature effect was evaluated over the range from 10°C to 70°C.
Figure 15.6A shows the temperature effect on the retention and separation of 4-AMP
and its related compounds using the Atlantis HILIC column with a gradient elution.
The most noticeable changes in the chromatograms over the temperature range are
the drastic decrease in the retentions of 4-MP and the two degradants as well as
the switch of elution order between the two degradants. In addition, the resolution
between 3- and 4-AMP is totally lost at temperatures above 50°C.

In order to construct the van’t Hoff plots as shown in Figure 15.6B, the experi-
ment was also performed under isocratic conditions. For all analytes, the slopes
(or the apparent enthalpy changes) are negative, demonstrating that the overall
chromatographic process is an exothermic process. As expected, 4-MP and the
two degradants have much larger slopes than 2-, 3-, and 4-AMP. This observa-
tion is consistent with the fact that the mobile-phase ss pH is close to the ss pK a
values of 4-MP and the two degradants but still significantly lower than those
of 2-, 3-, and 4-AMP as discussed in previous section. It has been reported that
amine compounds have relatively large enthalpies of ionization while carboxylic
acids (e.g., acetic acid) and inorganic acid (e.g., H3PO 4 and H 2PO 4 −) have neg-
ligible enthalpies of ionization.63,67,68,72 Therefore, all the analytes in this study
are expected to have decreased ss pK a values at elevated temperatures while the
s s
s pK a value (and the mobile-phase s pH) for the formate buffer remains almost the
same over the temperature range studied. As a result, the ratio of the neutral to
the cationic forms of 4-MP as well as the two degradants will be increased at
elevated temperatures, leading to reduced retention. On the other hand, since
the ss pK a values of 2-, 3-, and 4-AMP are much higher than the mobile-phase
s
s pH, such significant changes in retention is not expected. Since all the plots are
linear, the apparent enthalpy changes are estimated from the slopes of the van’t
Hoff plot, which are 19, 2.4, 4.1, 5.1, 20, and 22â•›k J/mol for 4-, 2-, 3-, 4-AMP,
Degradant-1, and Degradant-2, respectively. Considering the overall separation
profile, column temperature 30°C was a good compromise and was selected for
the final method.
15.2.2 Method Validation

Based on the evaluation of all the above chromatographic parameters, final method
conditions are summarized as follows:
Column: Water Atlantis HILIC, 3â•›μm, 150 × 4.6â•›mm

Mobile phase A: 100% acetonitrile
Mobile phase B: 50â•›m M ammonium formate (pH 3.0)
Gradient: 10% B–50% B in 20â•›min
Flow rate: 1.5â•›m L/min
UV detection: 254â•›nm
Sample diluent: Acetonitrile/water (90/10, v/v)
Sample concentration: 0.5â•›mg/mL 4-AMP
System suitability sample: Aged 4-AMP sample spiked with 1% 4-MP, 2-,
â•… and 3-AMP
This method has been fully validated for lot release and stability testing of 4-AMP,
which included specificity, accuracy, precision, linearity, limit of detection (LOD)
and limit of quantitation (LOQ), and robustness. The method validation results are
summarized in Table 15.3.

HILIC Retention Behavior and Method Development
Table 15.3
Validation Summary for the Determination of 4-AMP Assay and Related Impurities
Experiment Results
Specificity of the method with High degree of specificity to 4-AMP demonstrated by absence of interference from all known potential impurities and degradation
respect to potential impurities products. Peak purity verified by photo diode array and mass spectrometry.
of synthesis and degradation
Accuracy and precision for Acceptable recovery and precision demonstrated at 80%, 100%, and 120% of the nominal assay concentration:
4-AMP assay Level 80% 100% 120%
%Mean recovery 100.5 99.9 100.3
%RSD (n = 3): 0.64 0.88 0.60
Accuracy and precision for Acceptable recovery and precision demonstrated for all impurities at 0.05%, 0.5%, and 0.1% level:
impurities and degradants Level 0.05% 0.5% 1.0%
4-MP %Mean recovery 91.1 101.3 99.1
%RSD (n = 3) 5.7 1.3 0.1
2-AMP %Mean recovery 94.0 98.9 101.1
%RSD (n = 3) 6.7 0.7 1.1
3-AMP %Mean recovery 105.8 98.7 100.5
%RSD (n = 3) 4.1 2.0 1.1
Degradant-1 %Mean recovery 102.0 100.6 100.6
%RSD (n = 3) 3.3 0.9 0.8
Degradant-2 %Mean recovery 102.2 101.7 100.6
%RSD (n = 3) 4.0 0.7 0.7
Linearity Acceptable linearity of 4-AMP demonstrated over a range of 0.02%–120% of nominal with an R value of 1.0000.
Acceptable linearity demonstrated for each impurity and degradant over the range of LOQ to 1% with R above 0.9997.
LOD/LOQ Acceptable LOD/LOQ demonstrated for each impurity and degradant: LOD = 0.02% (S/N ≥ 3), LOQ = 0.05% (S/N ≥ 10).
Robustness Acceptable robustness demonstrated with respect to mobile phase ratios, buffer pH, column lot, flow rate, column temperature,
and UV wavelength.
359
15.3 HILIC-CLND Method for Hydrazine Residues

in an Intermediate
Hydrazine and methyl-substituted hydrazines are commonly used as reducing agents
in pharmaceutical synthesis. Hydrazine is also a degradation product in some phar-
maceutical preparations such as hydralazine and isoniazid.73 Since hydrazine and
substituted hydrazine are genotoxic and potentially carcinogenic,74 their residues in
synthetic intermediate or API must be monitored and controlled. Due to the lack
of UV chromophore in all hydrazine-related molecules, most analytical procedures
involve derivatization followed by GC,73,75 HPLC,76 or TLC.77 To avoid derivatization,
ion chromatography with electrochemical detector has been used for direct analysis
of hydrazine residue in a pharmaceutical intermediate.78 Here, we present a HILIC
method using a CLND79–82 for direct analysis of hydrazine and 1,1-dimethylhydra-
zine in a pharmaceutical intermediate. The method development was focused on the
use of alcohol rather than acetonitrile as a weak eluent in the mobile phase with sys-
tematic evaluation of method parameters. To aid the exploration of retention mecha-
nism, methylhydrazine and 1,2-dimethylhydrazine were also included in the study
because they provide additional variability in analyte hydrophilicity (Table 15.4).
15.3.1 Method Development

15.3.1.1 Alcohol as a Weak Eluent
As mentioned in the introduction, methanol as an organic modifier in the mobile phase
has been used mostly with aqueous buffers at neutral or alkaline pH for separation of
basic drugs on silica columns.30,39–43 Under such mobile-phase conditions, the silanol
groups of the silica stationary phase are at least partially ionized. The retention of
basic compounds involves ion exchange although other retention mechanisms (such
as interaction with surface silanol and siloxane groups) also play a role, depending on
pH, methanol concentration, and the type of buffer in the mobile phase.40,42,43,57,83–85
At acidic pH, the retention of basic compounds on silica is significantly reduced
Table 15.4
Structure and Physical Properties of Hydrazines
Compound 1 2 3 4
1,2-Dimethylhydrazine 1,1-Dimethylhydrazine Methylhydrazine Hydrazine
MW 60.10 60.10 46.07 32.05
Structure H H3C H H2N NH2
H3C N N
N CH3 N H3C NH2
H NH2
H3C
w
w pK a 7.52 7.21 7.87 8.07
w
Source: w pK a values are Adapted from Hinman, R.L., J. Org. Chem., 23, 1587, 1958.

due to the decrease in the degree of silanol ionization.40,42,57,84 With the reduction of
the electrostatic interaction, the eluotropic strength of a mobile phase constituted of
methanol and aqueous buffer becomes too strong to be useful.19,24 Therefore, in the
recent HILIC applications of silica columns, the less polar solvent acetonitrile is the
most commonly used organic modifier.14,19,24,45,86–88 Acetonitrile has much weaker
hydrogen-bonding capability and will not compete as strongly as methanol with ana-
lytes for hydrogen-bond interaction with the stationary phase. Nevertheless, the use
of methanol as a weak eluent has been shown to offer different selectivity.24,32,47,89
Another advantage of using methanol or other short-chain alcohols as a weak elu-
ent is the enhanced buffer solubility and in some cases sample solubility as well.30
The use of alcohol as a weak eluent also makes it possible to use a nitrogen-specific
detector CLND.
To increase the retention of basic compounds in acidic mobile phase with alcohol
as a weak eluent, it requires that the stationary phase possesses some electrostatic
interaction under acidic conditions. Based on our column screening, the sulfoalkyl-
betaine bonded zwitterionic stationary phase, ZIC-HILIC, provides adequate reten-
tion of hydrazines. Because of the concurrent presence of a quaternary ammonium
group and a sulfonate group in close proximity of 1:1 ratio, the ZIC-HILIC column
does not show strong ion exchange interaction. But the zwitterionic stationary phase
maintains a certain level of unique electrostatic interaction with both anionic and
cationic analytes and is capable of simultaneously separating anions and cations.29,90
This ZIC-HILIC stationary phase also has a slight electrostatic contribution from the
sulfonic group and this low negative excess charge is independent of pH.31,90 A recent
study has shown decent retention and separation of opioids and their glucuronides
on the ZIC-HILIC column with a mobile phase containing 80/20 (v/v) methanol/
ammonium formate (1â•›m M, pH 4).89 Figure 15.7 shows the retention of hydrazines on
a ZIC-HILIC column with different alcohols as a weak eluent. The retention times
of hydrazines increased drastically as the less polar ethanol or isopropyl alcohol was
used. Methanol is a much stronger elution solvent because of its strong hydrogen-
bond interactions with both stationary phase and the analytes. This result is consis-
tent with the findings in the literature.24,30 In this study, ethanol was selected to be
the weak eluent because it provided adequate retention and separation of hydrazines
with acceptable peak shape.
15.3.1.2 Combined ZIC and HILIC Modes

Since the introduction of zwitterion ion chromatography (ZIC) (also called electro-
static ion chromatography or EIC) by Hu et al. in 1993,91 the retention mechanism
of ZIC has been a subject of continuous debate.92–96 Until the introduction of the
covalently bonded zwitterionic stationary phases (more specifically the silica-based
ZIC-HILIC phase) by Irgum and colleagues,90,97–99 ZIC applications were mostly
performed on zwitterionic stationary phases that were formed by physical adsorp-
tion of zwitterionic surfactants onto a C18 stationary phase. Almost all studies on
ZIC-retention mechanism were focused on the separation of inorganic anions using
such dynamically coated zwitterionic stationary phases under fully or highly aque-
ous mobile-phase conditions.91–96 The introduction of sulfoalkylbetaine-bonded
zwitterionic stationary phase ZIC-HILIC eliminated the need for the hydrophobic

mV
MeOH
0
EtOH
2
1 3 4 IPA
–2
0 20 40 60 80 100 120 min
Figure 15.7 Effect of type of alcohol on the retention of hydrazines on a ZIC-HILIC

column. For each separation, isocratic run was performed in the mobile phase of TFA/
water/alcohol (0.1/10/90, v/v/v), column temperature at 30°C, flow rate at 0.4â•›m L/min with
a splitter and CLND. The CLND system was set at 1050°C combustion furnace, 50â•›m L/min
argon, 280â•›m L/min oxygen, 75â•›m L/min makeup (argon), 30â•›m L/min ozone, 5°C cooler,
gain ×1, 750â•›V on PMT. The analyte concentrations were about 30–70â•›μg/mL in water/etha-
nol (20/80, v/v). Acetonitrile at 0.1% (v/v) level in ethanol was used as a void time marker.
Injection volume was 10â•›μL. 1: 1,2-dimethylhydrazine, 2: 1,1-dimethylhydrazine, 3: meth-
ylhydrazine, 4: hydrazine. (Reproduced from Liu, M. et al., J. Chromatogr. A, 1216, 2362,
C18 substrate and also made it possible to use high organic content in the mobile
phase. Consequently, the ZIC-HILIC phase has found more applications in the
HILIC mode with its unique combined zwitterionic and HILIC interactions. The
retention mechanism on ZIC-HILIC stationary phase is complex and very limited
mechanistic studies have been performed so far.31 Most applications of ZIC-HILIC
used an acetonitrile–water mobile phase with organic buffers such as ammonium
acetate or format. Nevertheless, typical HILIC retention behavior of hydrazines
was also observed on the ZIC-HILIC column using an alcohol–water mobile phase
under acidic conditions. Hydrazines were eluted in the order of increased hydro-
philicity (i.e., 1,2-dimethylhydrazine, 1,1-dimethylhydrazine, methylhydrazine, and
hydrazine) and their retentions decreased with the increase in mobile-phase strength
(water content). Because the retention mechanism on ZIC-HILIC column is not well
understood, in addition to the effect of mobile-phase strength other critical method
parameters such as type of acid additive, ionic strength, buffer type, and column tem-
perature were systematically evaluated during the hydrazine method development.21
15.3.1.3 Type of Acid Additive

The effects of three different acids, TFA, formic acid, and acetic acid, and their
concentrations in the mobile phases on the retention and separation of hydrazines
are shown in Figure 15.8 and Table 15.5, respectively. In all cases, the elution orders
of the hydrazines were the same, but their retentions were strongly influenced by
the acid type. The shortest retention and best peak shape were observed with the

mV
1.5
2
1 3
1
4
0.5 2
TFA
0
3
1
4 HCOOH 2
–0.5 3
1 4 HOAc
10 20 30 40 50 60 min
Figure 15.8 Effect of type of acid additives on the retention of hydrazines on a ZIC-
HILIC column. For each separation, isocratic elution was performed with acid/water/ethyl
alcohol (0.1/30/70, v/v/v) as mobile phase, column temperature at 30°C, flow rate at 0.4â•›m L/
min with a splitter and CLND. 1: 1,2-dimethylhydrazine, 2: 1,1-dimethylhydrazine, 3: meth-
ylhydrazine, 4: hydrazine. Other conditions as in Figure 15.7. (Reproduced from Liu, M.
strongest acid, TFA. Similar effect had been reported when peptides were analyzed
on the TSK-Gel Amide-80 column using the three different acids.10 In addition, as
shown in Table 15.5, minimal change in retention was observed with the change of
TFA concentrations, while the most significant change in retention was noticed with
the increase in acetic acid concentration. These observations may be interpreted by
the differences in acidity and ion-pairing capability among the three different acids.
The pH values shown in Table 15.5 were measured before mixing with the organic
modifier. Considering the relatively high aqueous ww pK a values of 1,2-dimethylhy-
drazine (7.52), 1,1-dimethylhydrazine (7.21), methylhydrazine (7.87), and hydrazine
(8.07),100 all hydrazines are expected to be protonated in the ethanol–water mobile
phase with TFA (ww pK a 0.52), and likely with formic acid (ww pK a 3.75) as well, even
though the ss pK a values of TFA and formic acid in 80% ethanol are expected to be
higher than their ww pK a. With acetic acid (ww pK a of 4.75), however, its ss pK a in 80%
ethanol might increase to a level close to the ss pK a values of the hydrazines, resulting
in only partial protonation of the hydrazines. It has been reported that in 80/20 (v/v)
methanol/water the ss pK a values of formic acid and acetic acid are increased to about
5.2 and 6.5, respectively, while the ss pK a values of protonated neutral bases such as
ammonium and anilinium are about 0.7 units lower than ww pK a.56
In ZIC, the retention of ionic analytes on sulfoalkylbetaine stationary phase is
strongly affected by mobile-phase composition.92–96,98 Without electrolyte in the
mobile phase, the positive and negative charges of one sulfoalkylbetaine zwitterion
are likely paired with the negative and positive charges, respectively, of adjacent
zwitterions. A small amount of salt in the mobile phase may breakup the self-
association and results in an abrupt change in surface morphology. The retention of

Table 15.5
Effect of Acid Modifier on the Retention Time (Minutes) of Hydrazines
TFA
Analyte 0.05% (pH 2.26) 0.1% (pH 1.93) 0.15% (pH 1.77) 0.2% (pH 1.65)
Hydrazine 21.86 20.74 20.45 20.03
Methylhydrazine 15.33 15.15 15.24 14.98
1,1-Dimethylhydrazine 13.20 13.07 13.17 12.98
1,2- Dimethylhydrazine 11.61 11.50 11.59 11.42
HCOOH
0.1% (pH 2.65) 0.25% (pH 2.44) 0.5% (pH 2.29) 0.75% (pH 2.19)
Hydrazine 39.78 36.18 36.03 35.44
HOAc
0.1% (pH 3.24) 0.25% (pH 3.04) 0.5% (pH 2.89) 0.75% (pH 2.79)
Hydrazine 119.16 79.40 63.96 56.47
1,2- Dimethylhydrazine 48.35 34.34 28.49 25.57
Notes: (1) The pH (i.e., ww pH) was measured before mixing with organic solvent. Retention is in minutes.
(2) Chromatographic conditions: the mobile phase was acid/water/ethanol (various/20/80, v/v/v).
Other chromatographic conditions were the same as those described in Figure 15.8.
ionic analytes is strongly influenced by the eluent anion present in the mobile phase
and its ion-pairing capability with the quaternary ammonium group of the sulfoal-
kylbetaine stationary phase. In general, strong ion-pairing eluent (or chaotropic ions
in the Hofmeister series) such as ClO4− will impart negative charge to the stationary
phase, resulting in decreased retention of analyte anions or increased retention of
analyte cations. In addition, analyte cations especially monovalent inorganic cations
are much weakly retained due to their weak interaction with the sulfonate group.92–96
Considering the pKa values of the three different acids and their ion-pairing capa-
bility, formic acid and acetic acid are less likely than TFA to form strong ion pairs
with either the quaternary ammonium group of the stationary phase or the positively
charged hydrazines in the mobile phase. Stronger ion pair in the mobile phase would
lead to weaker electrostatic interaction with the sulfoalkylbetaine stationary phase
as well as decreased hydrophilic partition into the water-rich layer on the stationary
phase, resulting in shorter retention. However, this cannot explain the drastic differ-
ence in retention between formic acid and acetic acid as additive because the two
acids are not expected to have a big difference in ion-pairing capability.101,102 Other
factors might play important roles. For example, with acetic acid, it is possible to

have strong HILIC mode retention of the neutral hydrazines, which may partition
into the immobilized water-rich layer without experiencing electrostatic repulsion
from the positively charged quaternary ammonium group of the stationary phase.
15.3.1.4 Type of Buffer and Ionic Strength
Under acidic conditions, the retention of protonated hydrazines on the ZIC-HILIC
stationary phase involves electrostatic interaction. Therefore, the effects of buffer
type and buffer concentration are expected to be significant. In ZIC, the effect of
salt type and concentration on the retention behavior of inorganic anions has been
extensively studied in 100% aqueous mobile phase.90,92–99 The information available
for the ZIC-HILIC phase in the HILIC mode is limited to the effect of ammonium
acetate buffer on the retention of acidic compounds44,103 and inorganic anions and
cations.29 The use of alcohol as the weak eluent in this study made it possible to eval-
uate five different types of commonly used buffers, ammonium formate, ammonium
acetate, ammonium phosphate, sodium phosphate, and triethylamine phosphate, in a
ethanol–aqueous (80/20) mobile phase.
The effect of five different buffers on the retention of the hydrazines is sum-
marized in Table 15.6. Because CLND requires nitrogen-free eluent and is not
compatible with high salt concentration in the mobile phase, a refractive index
(RI) detector was used for this experiment. Since the HCl salts of hydrazine and
1,2-dimethylhydrazine were used and RI is a universal detector, the chloride peak
was also observed because both cation and anion can be retained on the ZIC-
HILIC column. The results in Table 15.6 showed that among the three ammonium
salts, the retention time increased for hydrazines and decreased for chloride ion in
the following order: formate < acetate < phosphate (H2PO4 −). This retention order
is difficult to explain based on the relative strength of the interactions between
the buffer anions and the quaternary ammonium group of the sulfoalkylbetaine
stationary phase because the interactions with formate and acetate are expected
Table 15.6
Effect of Buffer Type on the Retention Time (Minutes)
Ammonium Ammonium Ammonium Sodium Triethylamine
Name Formate Acetate Phosphate Phosphate Phosphate
Hydrazine 26.00 27.87 44.94 37.39 >60
Methylhydrazine 20.02 21.53 33.03 29.42 46.75
1,1-Dimethylhydrazine 18.05 19.27 27.39 25.01 32.91
1,2-Dimethylhydrazine 14.96 16.27 22.23 20.85 29.18
Chloride 14.58 14.04 10.57 12.99 8.71
Source: Adapted from Liu, M. et al., J. Chromatogr. A, 1216, 2362, 2009. With permission from
Elsevier.
Note: Chromatographic conditions: the mobile phase was 50â•›mM buffer pH 4/ethanol (20/80, v/v); other
chromatographic conditions were the same as those described in Figure 15.9.

to be stronger than phosphate (H2PO4 −) according to the Hofmeister series.101,102

We again hypothesize that the observed order of retention is related to the effect in
the mobile phase where the interaction of protonated hydrazines with formate or
acetate is stronger than with phosphate, resulting in weaker electrostatic interac-
tions with the stationary phase. The results in Table 15.6 also showed the effect
of buffer cations on the retention of hydrazines and chloride. Shorter retention
of hydrazines was observed with sodium phosphate while the longest retention
was observed with triethylammonium phosphate (Na+ < NH4+ < TEA+). This elu-
tion order is consistent with the affinity sequence observed for conventional strong
cation exchanger.104 TEA+ is a weaker eluent because it contains larger and more
hydrophobic groups, which results in a more diffuse positive charge and weaker
interaction with the sulfonate group.96
The effect of buffer concentration/ionic strength on the retention of hydra-
zines is shown in Figure 15.9. The retention of the positively charged hydrazines
decreased when ammonium formate buffer concentration was increased from 5 to
90â•›m M, whereas the retention of the negatively charged chloride ion increased
but quickly reached a plateau at about 30â•›m M ammonium formate. Similar results
have been reported for the effect of ammonium acetate buffer concentration on
the retention of inorganic cations on the ZIC-HILIC column by Risley and Pack.29
They explained the effect by a two-part competition mechanism, i.e., increasing
ammonium concentration in the mobile phase reduces the interaction of analyte
cations with the sulfonate group of the stationary phase and also decreases the
opportunity for analyte cations to partition into the enriched aqueous layer in the
HILIC mode. However, such competition for interaction with the sulfonate group
of the ZIC-HILIC phase is not as strong as the effect of salt concentration on a
conventional strong cation exchanger. In conventional cation-exchange chroma-
tography, a plot of the logarithm of the analyte retention factor (log k′) versus
the logarithm of buffer cation concentration is linear with a slope of −1 for singly
charged analyte and monovalent buffer cation.104,105 Figure 15.9B shows the plot of
log k′ versus log (buffer concentration) in the range of 5–30â•›m M. Within this range,
linear relationships were observed for all hydrazines and for chloride. However, the
observed slopes (about −0.3 for hydrazines and +0.1 for chloride) are much smaller
than −1 or +1. Such decreased slopes have been reported for inorganic anions and
are believed to be typical in ZIC mode because both electrostatic attraction and
repulsion are present.95,106–108 In ZIC-HILIC mode, the effect of buffer concentra-
tion on HILIC interaction may also play a role.
The retention behavior of chloride ion at low ammonium formate buffer con-
centrations (5–30â•›m M) is similar to the increased retention of iodide in an eluent
with less chaotropic anions such as SO42− in the ZIC mode as reported by Cook
et al.95 Formate is less chaotropic than chloride in the Hofmeister series.95,101,102 Guo
and Gaiki observed an increase in the retention of salicylic acid on three differ-
ent types of HILIC columns (TSK-Gel Amide-80, HILIC Silica, and ZIC-HILIC)
when ammonium acetate concentration in mobile phase was increased from 5 to
20â•›m M.44 They hypothesized that the increased retention was due to weakened
electrostatic repulsion at higher salt concentrations. Similar interpretations have
been made by Takegawa et al.103 and Alpert.109

nRIU Cl 1
2 3 4 5 mM
Cl 1
10,000 2 3 4 10 mM
1 Cl 2 3 4 20 mM
1 Cl
8,000 2 3 4 30 mM
1 2/Cl
3 4 40 mM
6,000 1 Cl
2 3 4 50 mM
1 Cl
4,000 2 3 4 70 mM
1 Cl/3
2 4
2,000 90 mM
(A) 5 10 15 20 25 30 35 40 min
0.7
0.6
0.5
0.4
R2 = 1.000
Log k΄
0.3
0.2 R2 = 1.000
0.1 R2 = 1.000
R2 = 0.964
0
R2 = 1.000
–1.0
0.6 0.8 1.0 1.2 1.4
(B) Log [ammonium formate]
Figure 15.9 (A) The effect of ionic strength on the retention of hydrazines on a ZIC-HILIC
column. Column temperature was 30°C. Isocratic runs with the mobile phase of 5–90â•›m M
ammonium formate buffer pH 3.0/ethanol (20/80, v/v). (The buffer concentrations refer to the
concentration before mixing with organic solvent.) Flow rate was 0.4â•›m L/min with RI detec-
tion. The analyte concentrations were about 0.8–1.2â•›mg/mL in water/ethanol (20/80, v/v). The
injection volume was 1â•›μL. 0.5% Toluene (v/v) in ethanol was used as a void volume marker.
Chloride is from the HCl salts of hydrazine and 1,2-dimethylhydrazine. 1: 1,2-dimethylhy-
drazine, 2: 1,1-dimethylhydrazine, 3: methylhydrazine, 4: hydrazine. (B) Plot of logarithm of
retention factor (k′) against logarithm of buffer concentration (5–30â•›m M), hydrazine (—⦁ —),
methyl hydrazine (—▴—), 1,1-dimethyl hydrazine (— ▪ —), 1,2-dimethyl hydrazine (—⬩ —),
chloride (— ○ —). (Reproduced from Liu, M. et al., J. Chromatogr. A, 1216, 2362, 2009. With


Based on method development, the final method conditions are summarized as follows:
Column: ZIC-HILIC, 5â•›μm, 250 × 4.6â•›mm

Mobile phase: 0.1/30/70 (v/v) TFA/water/ethanol
Flow rate: 0.4â•›m L/min (0.2â•›m L/min to CLND by post-column
â•… flow splitter)
Detector parameters: CLND, 1050°C combustion furnace, 50â•›mL/min argon,
280â•›m L/min oxygen, 75â•›m L/min makeup (argon),
30â•›m L/min ozone, 5°C cooler, gain ×1, 750â•›V on PMT
Injection volume: 20â•›μL
Sample diluent: DMSO/ethanol (30/70, v/v)
Sample concentration: 10â•›mg/mL intermediate
Standard: 2â•›μg/mL each of 1,1-hydrazine and hydrazine
System suitability: 10â•›mg/mL intermediate spiked with 0.02% 1,1-hydrazine
â•… and hydrazine
Column temperature was evaluated during method development over the range of
10°C–60°C. Only a slight decrease in retention times was observed. This is under-
standable because no significant temperature effect is expected for ionizable com-
pounds if mobile phase pH is not close to their pKa values. Column temperature 30°C
was selected for the final method. The intermediate is hydrophobic and very soluble
in DMSO. In HILIC mode, the sample diluent DMSO/ethanol (30/70â•›v/v) is a weaker
solvent compared with the mobile phase.
The optimized method has been validated for its performance and suitability for
simultaneous analysis of trace amounts of hydrazine and 1,1-dimethylhydrazine in
Table 15.7
Method Validation Summary for the Determination of 1,1-Dimethylhydrazine
and Hydrazine Residues
Experiment Results
Specificity No interference from the intermediate and other potential impurities
Accuracy Acceptable recovery demonstrated for 1,1-dimethylhydrazine and hydrazine at 0.02%
and 0.04% level spiked into 10â•›mg/mL intermediate:
1,1-Hydrazine Hydrazine
Spiking level 0.02% 0.04% 0.02% 0.04%
%Mean recovery (n = 2) 101 103 84 103
Linearity Acceptable linearity demonstrated for 1,1-dimethylhydrazine and hydrazine in the range
of 0.01%–0.1% with coefficients of determination 0.9998 and 0.9996, respectively
Repeatability Acceptable injection repeatability of 1,1-dimethylhydrazine and hydrazine at 0.02%
level: 3.9% and 5.0%, respectively
LOD/LOQ Acceptable signal-to-noise ratio for reporting limit at 0.02% for 1,1-dimethylhydrazine
(S/N = 28) and hydrazine (S/N = 9)

mV
4 2
4
Intermediate-spiked 0.02%
3
Pharmaceutical intermediate
2
2 4
STD-0.02%
1
Blank
0
0 2 4 6 8 10 12 14 16 min
Figure 15.10 Simultaneous analysis of 1,1-dimethylhydrazine and hydrazine in a phar-

maceutical intermediate on a ZIC-HILIC column. Mobile phase was TFA/water/ethanol
(0.1/30/70, v/v/v). Flow rate was 0.4â•›m L/min with a splitter and CLND. Injection volume
was 20â•›μL. Pharmaceutical intermediate sample contained 10â•›mg/mL of the intermediate in
DMSO/ethanol (30/70, v/v). The spiked intermediate sample consisted of 10â•›mg/mL interme-
diate, 2â•›μg/mL 1,1-dimethylhydrazine and 2â•›μg/mL hydrazine in DMSO/ethanol (30/70, v/v).
The standard (STD) solution had 2â•›μg/mL each of 1,1-dimethylhydrazine and hydrazine in
water/ethanol (30/70, v/v). 2: 1,1-dimethylhydrazine, 4: hydrazine. Other conditions are as
in Figure 15.7. (Reproduced from Liu, M. et al., J. Chromatogr. A, 1216, 2362, 2009. With
a pharmaceutical intermediate to support process development for an early phase

program. The method validation results are summarized in Table 15.7 and typical
chromatograms are shown in Figure 15.10. The method sensitivity is adequate for
early-phase process development. But for late-phase process development and drug-
substance release testing, higher sensitivity is needed to meet the requirement of the
recently implemented EMEA guideline on genotoxic impurities. A separate method
based on ion chromatography with electrochemical detection has been developed in
our laboratory for this purpose.110
References
1. McCalley, D.V., J. Sep. Sci., 2003, 26, 187.
2. McCalley, D.V., Liquid chromatographic separations of basic compounds. In Advances
in Chromatography, ed. E. Grushka and N. Grinberg. CRC Press, Boca Raton, FL, 2008,
pp. 305–350.
3. Snyder, L.R., Kirkland, J.J., and Giajch, J.L., Practical HPLC Method Development,
2nd. Wiley-Interscience, New York, 1997.
5. Havlicek, J. and Samuelson, O., Anal. Chem., 1975, 47, 1854.
6. Linden, J.C. and Lawhead, C.L., J. Chromatogr., 1975, 105, 125.
7. Palmer, J.K., Anal. Lett., 1975, 8, 215.
8. Churms, S.C., J. Chromatogr., 1996, 720, 75.

9. Zhu, B.Y., Mant, C.T., and Hodges, R.S., J. Chromatogr., 1991, 548, 13.
12. Mant, C.T. and Hodges, R.S., J. Sep. Sci., 2008, 31, 2754.
13. Boersema, P.J., Mohammed, S., and Heck, A.J.R., Anal. Bioanal. Chem., 2008, 391, 151.
14. Naidong, W., J. Chromatogr. B, 2003, 796, 209.
15. Iwasaki, Y., Ishii, Y., Ito, R. et al., J. Liq. Chromatogr. Relat. Technol., 2007, 30, 2117.
18. Gavin, P.F., Olsen, B.A., Wirth, D.D., and Lorenz, K.T., J. Pharm. Biomed. Anal., 2006,
41, 1251.
19. Liu, M., Chen, E.X., Ji, R., and Semin, D., J. Chromatogr. A, 2008, 1188, 255.
20. Strege, M.A., Durant, C., Boettinger, J., and Fogarty, M., LCGC North Am., 2008, 26, 632.
21. Liu, M., Ostovic, J., Chen, E.X., and Cauchon, N., J. Chromatogr. A, 2009, 1216, 2362.
23. McKeown, A.P., Euerby, M.R., Lomax, H. et al., J. Sep. Sci., 2001, 24, 835.
24. Li, P. and Huang, J., J. Chromatogr. A, 2004, 1041, 163.
25. Wang, X., Li, W., and Rasmussen, H.T., J. Chromatogr. A, 2005, 1083, 58.
26. Hmelnickis, J., Pugovičs, O., Kažoka, H. et al., J. Pharm. Biomed. Anal., 2008, 48, 649.
28. McClintic, C., Remick, D.M., Peterson, J.A., and Risley, D.S., J. Liq. Chromatogr.
Relat. Technol., 2003, 26, 3093.
29. Risley, D.S. and Pack, B.W., LCGC N. Am., 2006, 24, 776.
30. Ali, M.S., Ghori, M., Rafiuddin, S. et al., J. Pharm. Biomed. Anal., 2007, 43, 158.
31. Hemström, P. and Irgum, K., J. Sep. Sci., 2006, 29, 1784.
32. Hao, Z., Xiao, B., and Weng, N., J. Sep. Sci., 2008, 31, 1449.
33. Ikegami, T., Tomomatsu, K., Takubo, H. et al., J. Chromatogr. A, 2008, 1184, 474.
34. Jandera, P., J. Sep. Sci., 2008, 31, 1421.
35. Dejaegher, B., Mangelings, D., and Heyden, Y.V., J. Sep. Sci., 2008, 31, 1438.
36. Wang, C., Jiang, C., and Armstrong, D.W., J. Sep. Sci., 2008, 31, 1980.
37. Jinno, K., Quiming, N.S., Denola, N.L. et al., Anal. Bioanal. Chem., 2009, 393, 137.
38. Majors, R.E. LCGC North Am., 2008, 26, 238.
39. Jane, I., J. Chromatogr., 1975, 111, 227.
40. Bidlingmeyer, B.A., Del Rios, J.K., and Korpi, J., Anal. Chem., 1982, 54, 442.
41. Richardson, H. and Bidlingmeyer, B.A., J. Pharm. Sci., 1984, 73, 1480.
42. Schmid, R.W. and Wolf, Ch., Chromatographia, 1987, 24, 713.
43. Bidlingmeyer, B.A. and Henderson, J., J. Chromatogr. A, 2004, 1060, 187.
44. Guo, Y. and Gaiki, S., J. Chromatogr. A, 2005, 1074, 71.
45. McCalley, D.V., J. Chromatogr. A, 2007, 1171, 46.
46. McCalley, D.V. and Neue, U.D., J. Chromatogr. A, 2008, 1192, 225.
47. Dong, L. and Huang, J., Chromatographia, 2007, 65, 519.
48. Snyder, L.R., Dolan, J.W., and Gant, J.R., J. Chromatogr., 1979, 165, 3.
49. Jandera, P., Churáčk, J., and Svoboda, L., J. Chromatogr., 1979, 174, 35.
50. Snyder, L.R. and Poppe, H., J. Chromatogr., 1980, 184, 363.
51. Jin, G., Guo, Z., Zhang, F., Xue, X., Jin, Y., and Liang, X., Talanta, 2008, 76, 522.
52. Yoshida, T., J. Chromatogr. A, 1998, 811, 61.
53. Rosés, M. and Bosch, E., J. Chromatogr. A, 2002, 982, 1.
54. Espinosa, S., Bosch, E., and Rosés, M., Anal. Chem., 2000, 72, 5193.
55. Gagliardi, L.G., Castells, C.B., Rafols, C., Rosés, M., and Bosch, E., Anal. Chem., 2007,
79, 3180.
56. Bosch, E., Bou, P., Allemann, H., and Rosés, M., Anal. Chem., 1996, 68, 3651.
57. Cox, G.B. and Stout, R.W., J. Chromatogr. A, 1987, 384, 315.

58. Greibrokk, T. and Anderson, T., J. Chromatogr. A, 2003, 1000, 743.

59. Dolan, J.W., J. Chromatogr. A, 2002, 965, 195.
60. Heinisch, S. and Rocca, J.-L., J. Chromatogr. A, 2009, 1216, 642.
61. Rosés, M., Subirats, X., and Bosch, E., J. Chromatogr. A, 2009, 1216, 1756.
62. Chen, M.H. and Horváth, C., J. Chromatogr. A, 1997, 788, 51.
63. Melander, W.R., Stoveken, J., and Horváth, C., J. Chromatogr., 1979, 185, 111.
64. Zhu, P.L., Dolan, J.W., Snyder, L.R. et al., J. Chromatogr. A, 1996, 756, 51.
65. Li, J., Anal. Chim. Acta, 1998, 369, 21.
66. McCalley, D.V., J. Chromatogr. A, 2000, 902, 311.
67. Castells, C.B., Gagliardi, L.G., Ràfols, C., Rosés, M., and Bosch, E., J. Chromatogr. A,
2004, 1042, 23.
68. Gagliardi, L.G., Castells, C.B., Ràfols, C., Rosés, M., and Bosch, E., J. Chromatogr. A,
2005, 1077, 159.
69. Heinisch, S., Puy, G., Barrioulet, M.-P., and Rocca, J.-L., J. Chromatogr. A, 2006, 1118,
234.
70. Hao, Z., Lu, C.Y., Xiao, B., Weng, N. et al., J. Chromatogr. A, 2007, 1147, 165.
71. Guo, Y., Srinivasan, S., and Gaiki, S., Chromatographia, 2007, 66, 223.
72. Castells, C.B., Ràfols, C., Rosés, M., and Bosch, E., J. Chromatogr. A, 2003, 1002, 41.
73. Matsui, F., Robertson, D.L., and Lovering, E.G., J. Pharm. Sci., 1983, 12, 948.
74. Hydrazine/Hydrazine sulfate (CASRN 302-01-2) in EPA’s online Integrated Risk
Information System (IRIS). http://www.epa.gov/iris/subst/0352.html (accessed June 25,
2009).
75. Carlin, A., Gregory, N., and Simmons, J., J. Pharm. Biomed. Anal., 1998, 17, 885.
76. Monograph: Hydralazine Hydrochloride. In United States Pharmacopeia. United States
Pharmacopeial Convention, Rockville, MD, 2007, USP 30, p. 2284.
77. Monograph: Povidone. In United States Pharmacopeia. United States Pharmacopeial
Convention, Rockville, MD, 2007, USP 30, p. 2994.
78. Jagota, N.K., Chetram, A.J., and Nair, J.B., J. Pharm. Biomed. Anal., 1998, 16, 1083.
79. Fujinari, E.M. and Courthaudon, L.O., J. Chromatogr. A, 1992, 592, 209.
80. Yan, X., J. Chromatogr. A, 1999, 842, 267.
81. Bhattachar, S.N., Wesley, J.A., and Seadeek, C., J. Pharm. Biomed. Anal., 2006, 41, 152.
82. Yan, B., Zhao, J., Leopold, K., Zhang, B., and Jiang, G., Anal. Chem., 2007, 79, 718.
83. Flanagan, R.J. and Jane, I., J. Chromatogr. A, 1985, 323, 173.
84. Lingeman, H., van Munster, H.A., Beynen, J.H. et al., J. Chromatogr. A, 1986, 352, 261.
85. Law, B., J. Chromatogr. A, 1987, 407, 1.
86. Naidong, W., Lee, J.W., Jiang, X. et al., J. Chromatogr. A, 1999, 735, 255.
87. Naidong, W., Shou, W., Chen, Y.-L., and Jiang, X., J. Chromatogr. A, 2001, 754, 387.
88. Grumbach, E.S., Wagrowski-Diehl, D.M., Mazzeo, J.R. et al., LCGC North Am., 2004,
22, 1010.
89. Vikingsson, S., Kronstrand, R., and Josefsson, M., J. Chromatogr. A, 2008, 1187, 46.
90. Jiang, W. and Irgum, K., Anal. Chem., 2002, 74, 4682.
91. Hu, W., Takeuchi, T., and Haraguchi, H., Anal. Chem., 1993, 65, 2204.
92. Hu, W., Haddad, P.R., Hasebe, K. et al., Anal. Chem., 1999, 71, 1617.
93. Okada, T. and Patil, J.M., Langmuir, 1998, 14, 6241.
94. Patil, J.M. and Okada, T., Anal. Commun., 1999, 36, 9.
95. Cook, H.A., Hu, W., Fritz, J.S., and Haddad, P.R., Anal. Chem., 2001, 73, 3022.
96. Fritz, J.S., J. Chromatogr. A, 2005, 1085, 8.
98. Viklund, C., Sjögren, A., Irgum, K., and Nes, I., Anal. Chem., 2001, 73, 444.
100. Hinman, R.L., J. Org. Chem., 1958, 23, 1587.
101. Roberts, J.M., Diaz, A.R., Fortin, D.T. et al., Anal. Chem., 2002, 74, 4927.

102. Dai, J., Mendonsa, S.D., Bowser, M.T., Lucy, C.A., and Carr, P.W., J. Chromatogr. A,
2005, 1069, 225.
103. Takegawa, Y., Deguchi, K., Ito, H. et al., J. Sep. Sci., 2006, 29, 2533.
104. Ståhlberg, J., J. Chromatogr. A, 1999, 855, 3.
105. Rocklin, R.D., Pohl, C.A., and Schibler, J.A., J. Chromatogr. A, 1987, 411, 107.
106. Hu, W., Haddad, P.R., Tanakar, K., and Hasebe, K., Analyst, 2000, 125, 241.
107. Hu, W., Hasebe, K., Tanakar, K., and Haddad, P.R., J. Chromatogr. A, 1999, 805, 161.
108. Jiang, W., Fischer, G., Girmay, Y., and Irgum, K., J. Chromatogr. A, 2006, 1127, 82.
109. Alpert, A.J., Anal. Chem., 2008, 80, 62.
110. Lai, S., Wang, S., and Ostovic, J., unpublished data.

16 Retention of Polar Acidic
and Basic Drugs by HILIC
Mohammed Shahid Ali and
Aamer Roshanali Khatri
Contents
16.1 Introduction................................................................................................... 373
16.2 Analysis of Pharmaceutical Ingredients........................................................ 374
16.2.1 Basic Amines in Cough–Cold Formulations..................................... 375
16.2.2 MFH and Related Compounds in Tablets......................................... 375
16.2.3 BT in Ophthalmic Solution................................................................ 375
16.2.4 SCG in Ophthalmic Solution............................................................. 375
16.3 Retention Mechanism.................................................................................... 377
16.3.1 Effect of Organic Solvent.................................................................. 378
16.3.1.1 Choice of Organic Solvent.................................................. 378
16.3.1.2 Use of Alternate Solvent..................................................... 379
16.3.2 Mobile-Phase pH and Ion Exchange................................................. 380
16.3.3 Ionic Strength.................................................................................... 386
16.3.4 Stationary Phase................................................................................ 387
16.3.5 Diluent and Injection Volume............................................................ 391
16.3.6 Validation of Methods....................................................................... 392
16.4 Advantages..................................................................................................... 393
16.5 Conclusion..................................................................................................... 397
References............................................................................................................... 397
16.1 Introduction
Drug analysis has evolved to cater to the requirements of drug discovery and phar-
maceutical quality control. High-performance liquid chromatography (HPLC) is the
most widely used separation technique for qualitative and quantitative determina-
tion of various chemicals in pharmaceutical industry. Prior to the 1970s, few reli-
able HPLC methods were available to the laboratory scientist. Till then, hydrophilic
stationary phase and hydrophobic mobile phase were used as normal phase (NP)
chromatography. Discovery of reversed-phase (RP) liquid chromatography in the
late 1970s allowed improved separation for a wide variety of compounds. Since a
substantial number of drugs in pharmaceutical research and discovery are ionic and
polar hydrophilic, RP-HPLC methods may be difficult for highly polar compounds
because of inadequate retention due to their highly polar nature. Achieving retention
373

and separation of polar compounds is an ongoing challenge for chromatographers.1–3

Such compounds need highly or 100% aqueous mobile phases but this can lead to
wetting of stationary phases (e.g., C18).4,5 Attempts to reduce the risk of dewetting
of C18 phases by means of polar embedding or polar end capping is not success-
ful since these stationary phases provide lower retention than standard C18 phases.2
Lack of retention for hydrophilic compounds in RP-HPLC is largely due to solo-
vophilic factors, where polar functional groups have an ability to form favorable
dipolar bonds with the solvent to become solvated. Since nonpolar stationary phase
cannot offer similar bonding, the solutes stay in solution and elute in void volume.6
In these cases, ion-exchange chromatography is the obvious choice where retention
is accomplished by coulombic interaction mechanism. Another approach is the use
of ion-pairing chromatography (IPC) where the ion-pair reagents in mobile phase
mediate the retention of analytes on RP columns. Ion-pair reagents are soap-like
ionic molecules having a charge opposite to that of molecule of interest, as well as a
hydrophobic region to interact with stationary phase. The counter ion combines with
the ions of the eluents, becoming ion pairs in the stationary phase, which results in
retention. Several IPC methods have been reported for polar compounds but its use
is limited due to expensive ion-pair reagents and its incompatibility with electro-
spray ionization mass spectrometry (ESI-MS).7 To retain compounds that are highly
hydrophilic and uncharged, derivatization techniques are used where one or more
of the functional groups are converted into hydrophobic groups by a chemical reac-
tion.8 Derivatization may be limited to single-component analysis and not suitable
for multicomponents where one or a few of the solutes may not be derivatized to
give detectable molecules under similar conditions. Also, derivatization may not be
suitable for closely eluting related compounds and the procedure is time consuming.
Hydrophilic-interaction chromatography (HILIC) has been used for the analysis of
small polar and basic compounds. A high-organic, low-aqueous mobile phase is used
to retain analytes with increasing orders of hydrophilicity. HILIC has emerged as a
viable option to analyze polar hydrophilic compounds. Although the first application
of HILIC is associated with the analysis of sugars about 30 years ago, HILIC was later
recognized as a subset of chromatography. In 1990, Andrew Alpert first described the
term HILIC, also called reverse reversed-phase or aqueous NP chromatography to
distinguish it from the NP chromatography.9 According to Alpert, retention is propor-
tional to the polarity of the solute and inversely proportion to the polarity of the mobile
phase.9 HILIC has been proved to be a useful technique for the separation of polar and
basic compounds because of its complementary selectivity to that of RP-HPLC.
Therefore, the HILIC approach was applied to separate neutral, acidic, and basic
pharmaceutical compounds in different formulations. Methods were validated as per
ICH Guidelines. The retention mechanism was studied and the effects of chromato-
graphic parameters were evaluated.
16.2 Analysis of Pharmaceutical Ingredients

We have developed a few analytical methods suitable for drug substance like pseu-
doephedrine hydrochloride (PSH), diphenhydramine hydrochloride (DPH), and dex-
tromethorphan hydrobromide (DXH) in cough–cold liquid formulation;10 metformin

Retention of Polar Acidic and Basic Drugs by HILIC 375
hydrochloride (MFH) and related compounds cyanoguanidine (CGD) and melamine

(MLN) in tablets11; brimonidine tartrate (BT),12 sodium cromoglycate (SCG) in oph-
thalmic solutions13; and brimonidine tartrate (BT), maleic acid (MAC), and timo-
lol maleate (TIM) in ophthalmic solutions. The unique selectivity on silica column
caused due to hydrophilic interaction, ion exchange, and RP retention9,14–17 facilitates
the analysis of simple single-component polar analytes to multicomponent analytes
of varying nature. The detailed chromatographic conditions for these compounds are
described in Table 16.1. The composition of the selected formulations for detailed
study is described herein.
16.2.1 Basic Amines in Cough–Cold Formulations

The test formulation was syrup with active ingredients as PSH (30.0â•›mg), DPH
(12.5â•›mg), and DXH (15.0â•›mg) in each 5â•›m L. Other active ingredients were menthol
and sodium citrate with excipients as saccharin sodium, flavor, and sodium benzo-
ate (as preservative). Standard and test solutions for injection in LC system were
prepared in diluent (water and methanol, 50:50, v/v) to obtain final concentrations of
analytes as PSH: 300â•›μg/mL, DPH: 125â•›μg/mL, and DXH: 150â•›μg/mL.
16.2.2 MFH and Related Compounds in Tablets

Tablet formulations containing 500, 850, and 1000â•›mg of MFH were evaluated. The
inactive components present in the formulation were excipients like hypromellose,
magnesium stearate, Povidone K 30, polyethylene glycol, talc, and croscarmellose
sodium. Solutions of MFH were prepared in the diluent (acetonitrile and water,
50:50, v/v) to provide a concentration of 150â•›μg/mL for injection in LC system.
16.2.3 BT in Ophthalmic Solution

Brimo ophthalmic solution manufactured by Jamjoom Pharmaceuticals Company
Limited, Jeddah, Saudi Arabia, was used to evaluate determination of BT. The
formulation comprised of BT 2.0â•›mg/mL as active ingredient. Other ingredients
included citric acid monohydrate, polyvinyl alcohol, sodium chloride, sodium citrate,
water, and benzalkonium chloride (as preservative). Standard and test solutions were
prepared in diluent (acetonitrile and water, 92:8, v/v) to contain BT 100â•›μg/mL for
injection in LC system.
16.2.4 SCG in Ophthalmic Solution

Test samples were Optidrin ophthalmic solution manufactured by Jamjoom
Pharmaceuticals Company Limited with composition as SCG (20â•›mg), benzal-
konium chloride as preservative (0.1â•›mg) per mL. Other components were poly-
sorbate 80, disodium edetate, sodium chloride, hypromellose (hydroxypropyl
methylcellulose), and water. Standard and test solutions were prepared in diluent
(acetonitrile and water, 86:14, v/v) to contain 200â•›μg of SCG per mL for injection
in LC system.

Table 16.1
Summary of Chromatographic Conditions
Flow Rate Injection
Analyte Buffer Mobile Phase Column (mL/min) Volume (μL) Detection
PSH, DPH, 6.0â•›g ammonium acetate and Methanol + buffer (95:5, Supelcosil™ LC–Si, 25â•›cm × 1.2 10 254â•›nm for PSH and
and DXH 10â•›mL triethylamine per liter v/v), apparent pH of 4.6â•›mm, packed with 5â•›μm DPH, changed to
water, pH adjusted to 5.2 mobile phase ≈7.2. silica particles (Supelco, 280â•›nm at 9.0â•›min
with orthophosphoric acid Switzerland; Part No. 58295) for DXH
85% (OPA)
MFH, CGD, 25â•›mM NaH2PO4, pH adjusted Acetonitrile + buffer Atlantis® HILIC–Si, 25â•›cm × 2.0 10 218â•›nm
and MLN to 3.0 with OPA (84:16, v/v) 4.6â•›mm, packed with 5â•›μm
silica particles (Waters,
Ireland; Part No. 18600203)
BT 5.0â•›g ammonium acetate per Acetonitrile + buffer (92:8, Supelcosil™ LC–Si, 25â•›cm × 2.0 10 254â•›nm
liter water v/v), pH adjusted to 7.1 4.6â•›mm, packed with 5â•›μm
with glacial acetic acid. silica particles (Supelco,
Buchs, Switzerland; Part No.
58295)
SCG 30â•›mM ammonium acetate, Acetonitrile + buffer Atlantis® HILIC–Si, 25â•›cm × 2.0 10 326â•›nm
pH adjusted to 3.0 with (86:14, v/v), apparent pH 4.6â•›mm, packed with 5â•›μm
glacial acetic acid of mobile phase ≈5.6. silica particles (Waters,
BT, MAC, and 5â•›g ammonium acetate per Acetonitrile + buffer Atlantis® HILIC–Si, 25â•›cm × 2.0 10 280â•›nm
TIM liter water (88:12, v/v), (no pH 4.6â•›mm, packed with 5â•›μm
adjusted) silica particles (Waters,

16.3 Retention Mechanism

HILIC is a variation of NP chromatography.18 The retention mechanism is multimodal
on silica14 and is believed to be governed by hydrophilic interaction (partitioning).
Superimposed electrostatic interaction between analytes and silica (ion exchange) func-
tions as another mode of retention and is controlled by buffer (e.g., ammonium acetate).
Besides, hydrophilic interaction may also include adsorption of analytes by silica gel.
HILIC uses polar stationary phases such as underivatized silica,1,19–23 or silica
packings modified with various functionalities, such as aminopropyl,24–26 polyhy-
droxyethyl aspartamide,9 cyclodextrin,27 3-cyanopropyl,28 2,3-dihydroxypropyl,29,30
amide,31,32 poly(succinimide).33 Besides, a wide variety of silica packings with
functionalities like polyamine, carboxymethyl, polyaspartic acid, poly(2-sulfoethyl
aspartamide), polymeric sulfoalkylbetaine zwitterionic are also available. In addition,
polar packings with functionalities like 2,3-dihydroxypropyl on methacrylic copoly-
mer support, amine on poly(vinyl alcohol) gel support, polyamine on PVA copolymer
support, mono-, di-, and triamine bonded by a propyl spacer on zirconia support,
sulfonic acid on styrene/divinyl benzene support are used for the analysis of carbohy-
drates.6 We used underivatized silica for the development of HILIC methods.
In HILIC, the stationary phase is polar material such as silica and the mobile phase
is highly organic (e.g., acetonitrile or methanol or a combination of one or more organic
solvents) with a small amount of aqueous solvent and counter ion (e.g., ammonium
acetate or sodium dihydrogen phosphate) where compounds elute in order of increasing
hydrophilicity.9 Nonpolar compounds elute from the column first, while polar solutes
show strong interaction with silanol groups on the silica surface that results in enhanced
retention. This polar selectivity of silica is a helpful tool in achieving the separation of
polar amine compounds from weak acids, e.g., cough–cold formulation matrix.
Mobile-phase condition optimized for analysis of a basic analyte was found
unsuitable for an acidic or another basic analyte, and vice versa in RP-HPLC due to
resolution problem and peak shape. Resolution between PSH, DPH, and DXH peaks
could not be achieved with different mobile phases having combination of phosphate
buffer, ammonium acetate buffer with methanol, or acetonitrile in conjunction with
different stationary phases such as octadecylsilane, octylsilane, and nitrile groups
chemically bonded to porous silica particles. Cough–cold formulations usually
require mixtures of acidic and basic compounds for their effectiveness and therefore
are difficult to separate by RP-HPLC using RP columns (e.g., C18 column). Due to
above limitations of RP-HPLC in dealing with such complex formulations, inter-
ferences from other matrix components like saccharin sodium, sodium benzoate,
citric acid, flavor, and sodium citrate present in formulation were also observed.
Therefore, separation of active components from matrix components is a key for
developing a rugged and robust analytical method. These interferences in RP-HPLC
are due to ionization constant of active components having basic pKa (PSH, DPH,
and DXH) and other matrix components having acidic pKa like saccharin sodium
(pKa = 1.8), citric acid (pKa = 3.15, 4.77, and 6.40), and sodium benzoate (pKa = 4.2).
However, the pKa profiles of active ingredients and matrix components were proved
to be very useful in present HILIC procedure employing slightly basic mobile phase
because of their ionization characteristics.10 This resulted in unique selectivity of

these compounds in HILIC mode and unambiguous separation of PSH, DPH, and
DXH were achieved with elution of matrix components in dead volume region. Like
RP-HPLC and NP-HPLC, selectivity and separation of mixtures in HILIC tech-
nique too, depends on the mobile phase pH, buffer salts, aqueous or organic solvents,
sample-solution media type, and the components themselves. Critical parameters
were studied and evaluated and described as follows.
16.3.1 Effect of Organic Solvent

Under HILIC conditions, solvent strength from weakest to strongest is tetrahydro-
furan (THF) < acetone < acetonitrile < isopropanol < ethanol < methanol < water,
where water is the strongest eluting solvent. The retention time (RT) of the described
compounds was found to be increased with the increase in organic solvent concen-
tration in the mobile phase. Therefore, it is necessary to consider the proportion of
different organic solvents while optimizing mobile phase.
16.3.1.1 Choice of Organic Solvent

Selection of organic solvent depends on the nature of molecule. For example,
methanol was used for the analysis of basic amines (PSH, DPH, and DXH) since
acetonitrile provided much higher retention and resulted broad peak shape and tail-
ing. While the analysis of MFH, BT, and SCG required acetonitrile to be used as
the organic solvent. Figure 16.1 shows impact of acetonitrile and methanol on the
16.0 Different ratio of acetonitrile/methanol and pH 3.0

ammonium acetate buffer (%), 2.0 mL/min,
Atlantis HILIC-Si column 4.6 × 250 mm
14.0
12.0
Retention factor (k)-SCG
10.0
8.0
6.0
4.0 Acetonitrile
Methanol
2.0
0.0
82 83 84 85 86 87 88 89 90
Organic solvent (%) in mobile phase
Figure 16.1 Effect of acetonitrile and methanol on the retention of SCG. (Reproduced
from Ali, M.S. et al., J. Sep. Sci., 31, 1645, 2008. With permission from Wiley-VCH Verlag
Gmbh & Co. KGaA.)

14 35
CGD CGD-MLN
12 MLN MLN-MFH
30
MFH
Capacity factor (k ΄)
10
25
Resolution (Rs)
8
20
6
15
4
2 10
0 5
81 82 83 84 85 86 87 81 82 83 84 85 86 87
(a) Acetonitrile (%) in mobile phase (b) Acetonitrile (%) in mobile phase
Figure 16.2 Effect of acetonitrile on (a) retention and (b) resolution of MFH, MLN, and
CGD. (From Ali, M.S. et al., Chromatographia, 67, 517, 2007. With kind permission from
Springer Science + Business Media.)
retention factor of SCG, where SCG is found to be unretained with methanol as

organic solvent. In general, adequate retention was achieved using the appropriate
organic solvent concentration for all subject compounds irrespective of basic and
acidic nature of the drug substances. The impact of organic solvent change may be
different for compounds in a given mixture. For example, related compounds CGD
and MLN are analyzed together with MFH. The effect of acetonitrile is shown in
Figure 16.2, where insignificant change in retention factor (k) of CGD was observed
as compared to MLN and that of MFH. This, in turn, affects the resolution between
CGD and MLN, and that between MLN and MFH. In order to maintain an enriched
water layer around the silica particles inside the HILIC column, 5% water is mini-
mum requirement in the mobile phase. Among the described methods, least organic
concentration was that for simultaneous determination of MFH, MLN, and CGD
where 84% acetonitrile was used while highest organic concentration was employed
for the simultaneous determination of PSH, DPH, and DXH which required 95%
methanol as organic modifier in the mobile phase.
16.3.1.2 Use of Alternate Solvent

Many HILIC procedures use acetonitrile as part of mobile phase. Acetonitrile is a
by-product during the manufacture of acrylonitrile. Acrylonitrile is co-polymerized
with butadiene and styrene to make ABS, a plastic used in cars etc, and the market
for it collapsed in 2008–2009. This has resulted in the shortage of acetonitrile and
its price went very high, and affected the analytical activities worldwide. One of the
most feasible options is to redevelop the analytical method using another solvent or
combination of different solvents depending on the solvent strengths, e.g., analysis
of sucrose in iron–sucrose injection as described in USP 2009. Chromatographic
conditions for this analysis is to use a mixture of acetonitrile and water (79:21, v/v)
at a flow rate of 2.0â•›m L/min as mobile phase, a 4â•›mm × 25â•›cm column that contains
packing USP L8 (aminopropylsilane), and refractive index detector. This is a HILIC

Sucrose
0 2 4 6 8 10 12 14
Minutes
Figure 16.3 Chromatogram of iron–sucrose solution showing separation of sucrose.
method since it uses of 79% organic solvent on a polar amino column and the analyte
being hydrophilic sucrose. When the chromatograms are recorded for standard and
test solution under prescribed conditions, typical RT for sucrose is about 8â•›min. In
order to replace acetonitrile from the mobile phase, we need a mobile phase that pos-
sesses similar strength as described in the USP method. Mobile phase was, therefore,
optimized using THF, methanol, and water with the consideration of column back
pressure increase due to the use of highly viscous solvents. Mobile phase consist-
ing of THF, methanol, and water (60:35:5, v/v/v) at a flow rate of 1.5â•›m L/min using
a Spherisorb 5â•›μm NH2 4.6â•›mm × 25â•›cm column (Waters, USA) provided an RT of
about 8â•›min upon injection of 10â•›μL standard and test solution (Figure 16.3). The
correlation coefficient (R2) was found to be 0.999 for the standard solutions of 13, 16,
18, 21, and 23â•›mg of sucrose per mL. Injection volume 20â•›μL provided distorted peak
shapes for solutions with higher concentration of sucrose (>21â•›mg/mL). Acetonitrile
is considered a stronger solvent than THF under HILIC conditions. Mobile-phase
mixture containing THF will require another solvent which is stronger than ace-
tonitrile to bring the cumulative solvent strength similar to that described in USP
2008 for sucrose analysis in iron–sucrose injection. For other HILIC methods having
mobile phases with significant proportion of acetonitrile, similar experiments can be
used to redevelop alternative methods for long-term use.
16.3.2 Mobile-Phase pH and Ion Exchange

In HILIC, more hydrophilic compounds are retained longer on a silica column
whereas hydrophobic compounds elute out of a silica column first. It is important
to control mobile-phase pH since ionic interactions vary with pH. It is expected
that PSH, DPH, and DXH would also elute out of a silica column in the same order
based on their hydrophilicity or hydrophobicity. It must be noted that PSH is highly

soluble followed by DPH which is freely soluble and DXH is sparingly soluble in
water. The elution pattern found was not based on their solubilities as expected.
PSH eluted first followed by DPH and DXH. The retention behavior of these com-
pounds was found to be influenced by the organic modifier. With every increase
in the percentage volume of organic modifier, here methanol, the individual RT
of these compounds increased indicating involvement of hydrophilic interactions
in the separation mechanism. Another factor that influences the separation in this
case is the ion-exchange interaction. This interaction is very strong and influences
largely the separations achieved. All the compounds in the present example are
ionizable compounds and the ionized species that carried the maximum charge
(ionization degree) elutes out first. Mobile-phase pH selection is a technique that
works well for ionizable compounds, because the retention characteristics of ion-
izable compounds are a function of mobile-phase pH.4,34 In HILIC, selection of
an appropriate pH of the mobile phase is very critical while achieving separation
of ionizable compounds. Therefore, it is important to understand the process of
ionization with varying pH. It is useful to have an idea of the pKa values of the
sample components, because changes in mobile pH have the largest effect on ana-
lyte retention when the pH is near the pKa of the analyte. Changes in pH during a
separation attempt involving ion-exchange phenomena can make a great difference
in the chromatogram. PSH, DPH, and DXH are salts of weak bases and get com-
pletely ionized when dissolved in water. They exist as conjugate acids of respective
weak bases and halide counter ions namely Cl−, Cl−, and Br−. PSH is a salt of the
weak base pseudoephedrine and strong acid HCl; DPH is a salt of the weak base
diphenhydramine and strong acid HCL; and DXH is a salt of the weak base dex-
tromethorphan and strong acid HBr. The pH of the aqueous solution in which they
exist is the most influential factor for their ionization.
Percentage of ionization at a given pH can be calculated by35
100
% Ionization = (16.1)
[1 + antilog( pH − pK w + pK b )]
where
pH is the value at which % ionization is calculated
pKw is the ionic product of water (14.00 at 25°C)
pKb is the basicity or dissociation constant
Conjugate acid–base pairs are linked by the expression
pK w = pK a + pK b (16.2)
Hence, at 25°C, Equation 16.1 can be written as
100
% Ionization = (16.3)
[1 + antilog(pH − pK a )]

We can calculate the degree of ionization for a given compound with its known
pKa value by substituting the value of antilog, which can be obtained by a scien-
tific calculator. For example, % ionization for DXH at pH 7.3, can be calculated
as follows:
100
% Ionization (DXH at pH 7.3) =
[1 + antilog(pH − pKa )]
100
=
[1 + antilog(7.3 − 8.3)]
100
=
[1 + antilog( −1.0)]
= 90.91%

It is well known that basic compounds are completely ionized at pH value 2 units
below their pKa value and are completely unionized at pH values 2 units higher
than pKa values. The degree of ionization for PSH, DPH, and DXH at different
mobile-phase pH values is shown in Table 16.2. If the pH value of the mobile phase
is lower than the pKa value of the compounds being analyzed, they will exhibit
an elution pattern based on the ionization behaviors. The positively charged spe-
cies undergo cation exchange with negatively charged silanol groups.15 Basic com-
pounds like PSH, DPH, and DXH will exist as cations and undergo ion-exchange
interactions with the anionic silanol groups originating from the silica of the sta-
tionary phase. PSH is the most charged ion among the analytes and hence shows
the least RT, eluting after matrix peaks. This is an indicator of high ion-exchange
Table 16.2
Ionization of Compounds at Different Mobile-Phase pH
PSH (pKa 9.5) DPH (pKa 9.1) DXH (pKa 8.3)
Buffer
pH MP pHa pH–pKa % Ionization pH–pKa % Ionization pH–pKa % Ionization
5.5 7.5 −2.0 99.01 −1.6 97.55 −0.8 86.32
5.3 7.3 −2.2 99.37 −1.8 98.44 −1.0 90.91
5.2 7.2 −2.3 99.50 −1.9 98.76 −1.1 92.64
5.0 7.0 −2.5 99.68 −2.1 99.21 −1.3 95.23
4.6 6.7 −2.8 99.84 −2.4 99.60 −1.6 97.55
4.2 6.3 −3.2 99.94 −2.8 99.84 −2.0 99.01
4.0 6.1 −3.4 99.96 −3.0 99.90 −2.2 99.37
Source: Ali, M.S. et al., J. Pharm. Biomed. Anal., 43, 158, 2007. With permission from Elsevier.
a Measured pH of mobile phase consisting of methanol and buffer (6.0â•›g ammonium acetate and 10â•›mL
triethylamine per liter, at specified pH) in the ratio 95:5, v/v.

activity of the cation of PSH and the residual silanol groups of the surface of silica
column. Another important factor in understanding the separation mechanism is
the presence of buffer salts, which create the counter ion, present in the mobile
phase. The counter ions promote ion-exchange mechanism responsible for lower-
ing the retention of analytes. Use of appropriate buffers in mobile phase with pH at
least 2 units apart of the pKa values of the compounds, here lower than compound
pKa because of limitation of use of high mobile-phase pH in conjunction with
the used silica column, can help in shifting the equilibrium of ionization by sup-
pression of the ionization process. The counter ions in the mobile phase compete
with the ions of compounds and thus influence the retention behavior through the
ion-exchange interactions and lowering of the RTs. The increasing RT of PSH,
DPH, and DXH, can therefore, be attributed to the electrostatic interaction with the
silanol of silica and counterions.
Effect of pH of the retention of these compounds is shown in Figure 16.4, which
is a direct result of the positively charged ions interacting with the silanol groups
influenced by varying degree of ionization. The study of this retention behavior
pH 7.5
pH 7.3
pH 7.2
pH 7.0
pH 6.7
pH 6.3
pH 6.1
0 2 4 6 8 10 12 14 16
Minutes
Figure 16.4 Overlaid chromatograms of test solution at different mobile-phase pH show-

ing their retention behaviors. (From Ali, M.S. et al., J. Pharm. Biomed. Anal., 43, 158, 2007.

Effect of pH on retention
10
Capacity factor (k΄)

6
DXH
4 DPH
PSH
2
0
6.0 6.4 6.8 7.2 7.6 8.0
Mobile phase pH
Figure 16.5 Plot of mobile-phase pH vs. capacity factor of PSH, DPH, and DXH. (From
Ali, M.S. et al., J. Pharm. Biomed. Anal., 43, 158, 2007. With permission from Elsevier.)
was carried out with a mobile phase comprising of 6.0â•›g /L ammonium acetate and
10â•›m L/L triethylamine.10 The pH range studied using this buffer was 4.0–5.5. The
final mobile phase comprised of an organic modifier methanol and the above buf-
fer in volume ratio of 95:5. The pH of the final mobile phase was as per the studied
pH values as mentioned in Table 16.2. Change in the pH of mobile phase caused
DXH RTs to vary more significantly as compared to PSH and DPH in Figure 16.5.
Lower pH values of 6.1 in the study led to higher ionization of PSH (99.96%),
DPH (99.90%), and DXH (99.37%); thus, resulting in lower resolution and RTs.
At the pH values of 6.1 and lower, the ionization of the compounds will lead to
100% ionization and more of ion exchange in the column leading to lower RTs.
The peaks are not well resolved, some times co-eluting and thus making it difficult
and unsuitable for quantitative determinations. However, at pH values higher than
7.5, the ionization of DXH is least. Due to low ionization, DXH shows higher RT
and same is the case with PSH and DPH as seen in Figure 16.5. Mobile phases of
values higher than 7.5 resulted in longer retention and peak broadening and hence
making them unsuitable for quantitative applications. Operating on silica columns
at pH values of 7.5 and above is not suggested due to loss of silica, creation of
voids leading to changes in RT, resolution loss, and peak symmetry. The pH of the
mobile phase is a very critical parameter and needs to be optimized for any appli-
cation based on the desired separation and the ionization degree of compounds to
provide the necessary chromatographic behaviors. The optimum chromatographic
conditions for the separation of PSH, DPH, and DXH are as summarized in Table
16.1. This was achieved at mobile phase pH 7.2 (buffer pH 5.2) that provided
optimal selectivity with fair resolution and retention. The compounds at thisÂ� pH
show Â�adequate retention and resolution with acceptable peak shapes, because of

their ionization degree (PSH-99.50%, DPH-98.76%, and DXH-92.64%) suitable

for optimum cation exchange interactions.
An increasing trend in the individual retention of PSH, DPH, or DXH is seen
in the studied mobile phase pH range 6.1–7.5. At lower mobile phase pH, e.g., pH
6.1, the positive charge densities of the analytes is highest. As the mobile-phase pH
increases (toward pH 7.5), the %ionization decreases, thereby causing weakening of
electrostatic repulsive force and increase in retention.
BT is positively charged and shows decrease of RT as pH of mobile phase is increased
between the ranges of 6.5 and 7.5.12 The pKa value of BT is 7.4 and at the pH values
between 6.5 and 7.5, it exists as a cation with decreasing charge density which leads to
lower electrostatic attraction as mixed-mode effect and hence lower RTs are observed.
Based on the separation mechanism explained above the behavior of the com-
pounds like MFH, CGD, and MLN can also be described. The charge density
of these compounds will vary depending on the ionization state and it is highly
influenced and controlled by the mobile phase pH. This tool is very helpful in
controlling the desired RT and achieving the optimum separation goal. Molecules
having lesser degree of ionization (positive) have been shown to possess higher
capacity factor.10 The pKa value of MFH is reported to be 2.8 and 11.5.36 MFH pos-
sess a charge of +2 at pH values below 2.8 and a charge of +1 between pH values
of 2.8 and 11.5. Due to this varying charge exhibited by MFH, the RT of MFH is
explainable due to the increase in electrostatic attraction of MFH with the silanol
groups in a mixed-mode effect over the pH range 3.7–5.3 (Figure 16.6) due to
increase in negatively charged silanols of silica surface, although slight decrease
in its charge density shall occur. MFH will exist as an un-ionized species at pH
9
CGD
8 MLN
MFH
7
6
0
3.7 4.2 4.7 5.2 5.7
Mobile phase pH
Figure 16.6 Effect of mobile-phase pH on the retention of MFH, CGD, and MLN.
(From Ali, M.S. et al., Chromatographia, 67, 517, 2007. With kind permission from
Springer Science + Business Media.)

values above 5.8 and hence RT of MFH is not influenced significantly beyond this
pH value since charge density (degree of ionization) will not change significantly
and extent of ionization of MFH will approach zero as mobile-phase pH increases.
Experimental conditions of pH value beyond 6.0 are not recommended due to the
limitation of the silica column as described earlier.37 The pKa value of MLN is
reported to be 5.1.38 Over the experimental range of pH from 3.7 to 5.3, MLN
becomes less ionized. On loss of its charge, MLN seems to be less influenced by
the negative charge carried by the silanol groups of silica column. Figure 16.6
shows the decrease in RT of MLN over the pH range used in experimentation. It
must be noted for the present separation method that the ionization degree influ-
ences the retention for MFH while the retention of MLN is dependent on the ion-
ization state. CGD found to be the least impacted by the changes in the pH of the
mobile phase which shows that its ionization is unaffected by changes in pH in the
range 3.7–5.3. Higher selectivity of the method is mainly due to the hydrophilic
interactions exhibited by the different polarity of the analytes. Hydrophilicity of
the compounds namely, MFH, MLN, and CGD dictates the elution pattern in this
example of separation.
Another example that is worth looking at to understand the mechanism is that
of SCG, an anion with pK a value of 1.9. Over the pH range of the experimentation,
increase of negative charge leading to complete ionization is observed followed
by the decrease in capacity factor in the pH range of 4.2–5.9.13 The limitation of
the operating pH values for the column used in the experiments apply as before
and hence maximum pH value studied is 5.9. The lowering of capacity factor for
SCG with increase in mobile phase pH from 4.2 to 5.9 is characterized by the
increase in the negative charge of surface silanol groups in the silica column lead-
ing to strengthening of electrostatic repulsion between SCG and silanols. For the
retention behavior of anionic species dependent on degree of their ionization, the
impact on retention by change in mobile phase pH is therefore opposite to that of
cationic species.
16.3.3 Ionic Strength

Correct choice of buffer is crucial for the maintenance of mobile phase pH because
buffers themselves are ionic and may take part in the ion-exchange process. It is
important to add a buffer in the mobile phase to control the ionization states of
both the analyte and the stationary phase, which affects the acid–base equilibrium
between analyte and stationary phase. Decreased retention was observed for PSH,
DPH, and DXH with increase in ionic strength (ammonium acetate or triethylamine)
in the mobile phase. Since the main mechanism in HILIC is believed to be the par-
titioning of the analytes into the aqueous phase that forms a stagnant layer on the
stationary-phase surface, the decrease in RT might be best understood by a shift in
equilibrium concentrations. As the NH4+ concentration increases preferentially in
the aqueous layer, there is less opportunity for the analyte counter ions to partition
into the aqueous layer.39 Thus, the cations are swept through the column (mainly
in the organic layer) with less interaction with the negatively charged silanols of
column and the aqueous phase. Decreased resolution, capacity factor, and peak

tailing were observed with buffer strength increase in the mobile phase. Increase
in the ratio of buffer in mobile phase (e.g., methanol and buffer 90:10, instead of
95:5, v/v) showed decreased retention and resolution. This retention behavior seen
above is indicative of HILIC because of opposite chromatographic effect than that
found in RP-HPLC.10 Therefore, the decrease in retention of PSH, DPH, and DXH
with increased buffer concentration is attributed to the increase in the electrostatic
repulsion of these compounds due to the presence of NH4+ ions.
Similar behavior was observed for BT where increased ammonium acetate
strength resulted in decreased retention of BT. As buffer strength increases, the elec-
trostatic attractive interactions arising from negatively charged silanols and cationic
analyte are weakened, which causes a decrease in the retention.
Buffer plays an important role in peak shape and retention of ionic compounds.
Ammonium acetate and ammonium formate buffers usually are advantageous for
extension of a HILIC method to ESI-MS analysis because of their volatility. These
buffers provide good solubilities in organic solvents like methanol and acetonitrile
at high proportions required in HILIC. Precipitation due to buffer solubility in the
mobile phase is generally not a problem with acetate buffers. At low wavelength,
these buffers may interfere with analysis and decrease the sensitivity of the method
by providing noisy baseline. Hence, use of these buffers at low wavelength (less
than 225â•›n m) should be limited. At wavelengths above 225â•›n m, acetate buffer can
be used conveniently and safely. Because of the above factor, sodium dihydrogen
phosphate buffer, which provides low absorption, was selected for the analysis of
MFH, CGD, and MLN at 218â•›n m (Figure 16.7). Significant increase in retention of
MFH was observed compared to MLN and CGD with increase in buffer strength
(Figure 16.8) at the optimized mobile phase pH. These compounds shall be influ-
enced by the increase in ionic strength depending on their ionization degree.
Significant increase in the retention of MFH in buffer strength range 20–40â•›m M
phosphate buffer was observed because of availability of negatively charged buffer
ions in the stagnant aqueous layer of silica column.
It has been previously established that ionic concentration has drastic effect on
the retention and peak shape of ions.34 Effect of ionic strength on the retention of
SCG was studied using 20, 25, 30, 35, and 40â•›m M solutions of ammonium acetate.13
The pH of each solution was adjusted to pH 3.0 with glacial acetic acid. Mobile
phase comprising acetonitrile and each buffers (86:14, v/v) under similar chromato-
graphic condition were employed and retention of SCG was noted. Figure 16.9 shows
increased retention with the increase in ammonium acetate buffer strength (mM). As
buffer strength (NH4+) increases, repulsive electrostatic interactions which are aris-
ing from negatively charged silanols and anionic cromoglicate are weakened. This
accounts for the increased retention of SCG. Buffer strength of 30â•›m M was consid-
ered the optimal choice (k value ≈ 4.2).
16.3.4 Stationary Phase

A wide variety of HILIC stationary phases are available involving underivatized
silica and derivatized silica.6 We used bare silica columns from Waters (Atlantis
HILIC silica) and Supelco (LC-Si) for the method development. One of the most

MLN
0.06
0.04
Voltage (V)
CGD
0.02
MFH
0.00
0 2 4 6 8 10 12
Time (min)
0.06
0.04
Voltage (V)
MFH
MLN
0.02
CGD
0.00
0 2 4 6 8 10 12
Time (min)
Figure 16.7 Chromatogram showing separated peaks of CGD, MLN, and MFH at differ-
ent wavelengths (top) 218â•›nm (bottom) 232â•›nm. (From Ali, M.S. et al., Chromatographia, 67,
517, 2007. With kind permission from Springer Science + Business Media.)
important advantages of using bare silica columns is the low column back pres-
sure exhibited by using the high viscous organic solvents like methanol and THF,
which facilitates operating chromatography at high flow rates. Another property
of bare silica packing is the absence of ligands that may detach from the packings
and appear as spurious peaks in the MS-spectra. Silica columns from different
manufacturers can provide a slight/significant difference in retention behavior and
resolution for the particular analytes.6 This impact can be due to the presence of
metal ions which affects the acidity of the silanol groups, or the purity of silica
due to differences in manufacturing procedures. Specific silica columns for HILIC
applications are commercially available which are designed to provide better peak
shape, adequate retention, and resolution. Presence of water shall cause formation
of negatively charged silanol groups which in turn attracts cationic analytes (e.g.,
BT, MFH) for ion-exchange mechanism. However, the same silanol group shall

14
CGD
12 MLN
MFH
10
0
10 20 30 40
Phosphate buffer (mM)
Figure 16.8 Effect of ionic strength on the retention of MFH, CGD, and MLN. (From Ali,
M.S. et al., Chromatographia, 67, 517, 2007. With kind permission from Springer Science +
Business Media.)
4.4
4.2
Retention factor (k)-SCG
4.0
3.8
3.6
3.4
Acetonitrile and different molarities ammonium
acetate buffer pH 3.0 (86:14, v/v), 2.0 mL/min,
3.2 Atlantis HILIC-Si column 4.6 × 250 mm.
3.0
20 25 30 35 40
Ammonium acetate buffer (mM) pH 3.0
Figure 16.9 Effect of ionic strength on the retention of SCG. (Reproduced from Ali, M.S.
et al., J. Sep. Sci., 31, 1645, 2008. With permission from Wiley-VCH Verlag Gmbh & Co.
KGaA.)

exhibit electrostatic repulsion with anionic analytes (e.g., SGC) and ion exchange
shall not take place. The use of electrolytic buffer is required to control the electro-
static interactions due to mixed-mode separation of cationic and anionic analytes
on silica columns.
To promote ion exchange for anionic analytes, silica modified with triazole has
been used. Positively charged triazole is suitable to retain acidic compounds via
anion exchange. The impact of pH and ionic strength on retention behavior of cat-
ionic and anionic analytes shall be opposite to that of bare silica column packing.
Therefore, selection of column should take consideration of the analyte type(s), and
separation goal. For example, separation of BT (cationic analyte), TIM (cationic ana-
lyte), and MAC (anionic analyte) (as impurity of TIM) was achieved on Atlantis
HILIC-Si column 4.6 × 250â•›mm using acetonitrile and buffer (ammonium acetate,
5â•›g/L) in the ratio 88:12, v/v, at a flow rate of 2.0â•›m L/min with detection wavelength
at 280â•›nm (Figure 16.10).
Nonpolar mobile phase comprising of solvents like hexane, toluene, etc. are
used in NP LC with silica columns. Commercially available silica columns, e.g.,
Suplecosil™ LC-Si are shipped in nonpolar solvents like hexane and ethyl acetate
(98:2). To use such columns in HILIC separation, they need to be flushed with com-
paratively less polar solvents like isopropyl alcohol followed by acetonitrile and
methanol to avoid immiscibility. Subsequently, the column can be hydrated with
5% water. Columns that are designated for HILIC mode need not be subjected to
this conversion process as they are shipped in solvents like acetonitrile and little
amount of water. The operating pH range of commercially available columns also
Timolol maleate
Brimonidine tartrate
Maleic acid
0 2 4 6 8 10 12
Minutes
Figure 16.10 Chromatogram showing separated peaks of maleic acid, brimonidine tar-
trate, and timolol maleate.

vary depending on the column material, e.g., Suplecosil™ LC-Si, HALO HILIC col-
umn (2–8), Agilent Zorbax HILIC plus column (1–8), Atlantis HILIC-Silica column
(2–6) as recommended by column manufacturers.
16.3.5 Diluent and Injection Volume

Solvents used for the sample preparation influence the peak shape and sensitivity
of determination in chromatographic analysis.34 Solvents that resemble the mobile
phase composition are known to provide excellent results. The mobile phase used in
the analysis of SCG is acetonitrile and 30â•›m M ammonium acetate pH 3.0 in volume
ratio of 86:14. Similar sample diluent composition applies to other methods too.
Normally, injection volumes of 5–50â•›μL provides sharp peak shapes and enhanced
sensitivity as long as the diluent composition is maintained similar to mobile phase.
Various combinations of diluents were used to examine the impact of diluent on
peak shape. Sharp and symmetrical peak shapes were obtained on injection vol-
ume of 5â•›μL with sample diluents having acetonitrile and water 70:30, 60:40, 50:50,
40:60, 30:70 (v/v). Peak distortion was seen with these diluents following injection
volume higher than 10â•›μL. It is important to note that water alone as diluent pro-
vided peak distortion in all the described methods. Diluent comprising of methanol
and water (50:50, v/v) for PSH, DPH, and DXH determination with injection volume
of 5–30â•›μL showed good peak shapes for all compounds as shown in Figure 16.11.
30 μL
20 μL
10 μL
5 μL
0 2 4 6 8 10 12 14
Minutes
Figure 16.11 Overlaid chromatograms of test solution following injection volume of 5,

10, 20, and 30â•›μL. (From Ali, M.S. et al., J. Pharm. Biomed. Anal., 43, 158, 2007. With

16.3.6 Validation of Methods

Validation of all the methods developed in author’s lab was done as per the require-
ments of ICH Guidelines.40 Parameters such as specificity, accuracy, precision range,
robustness, and system suitability were examined and found to be acceptable for the
intended analyses.
Specificity is defined as the ability of the analytical procedure to assess unequivo-
cally the analyte in the presence of component like impurities, degradants, matrix,
etc. that may be present in the analytical samples. Forced degradation, as part of
specificity evaluation, is necessary while developing stability-indicating assay meth-
ods. Specificity is the most challenging parameter to achieve for a stability-indicating
method. Blank matrix solution should not provide interfering peak(s) and impurity
peak(s) should be separated from the main analyte peaks.
Degradations of PSH, DPH, and DXH were simulated by subjecting the solutions
to acidic hydrolysis (HCl), alkaline hydrolysis (NaOH), oxidation (H2O2), photolytic
(UV/sunlight), and heat as recommended by the ICH Guidelines. The purity of ana-
lyte peaks were evaluated using photodiode array detector and found to be homo-
geneous. Figure 16.12 shows the chromatogram of oxidation study with maximum
degradation. Two impurities namely CGD and MLN were separated from the main
MFH peak with fair resolution (Figure 16.7) and could be quantified using the same
method. Chromatograms of the stressed samples revealed complete separation of
Dextromethorphan HBr
Pseudophedrine HCl
Diphenhydramine HCl
0 2 4 6 8 10 12 14
Minutes
Figure 16.12 Chromatogram of test solution subjected to forced degradation by oxidation

(H2O2) showing degradation peaks. (From Ali, M.S. et al., J. Pharm. Biomed. Anal., 43, 158,

Oxidative (5 mL H2O2 30%, 70°C, 1 h) Photolytic (sunlight, 2 h)

1.50
0.20
1.25
0.15
1.00
Voltage (V)
Voltage (V)
0.75 0.10
0.50
0.05
BT
0.25
0.00 0.00
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
0.30
Acidic (2 mL HCl 37%, 70°C, 6 h) Alkaline (2 mL NAOH 50%, 70°C, 20 h)
0.35
BT
0.25
0.30
0.25 0.20
Voltage (V)
Voltage (V)
0.20 0.15
0.15
0.10
0.10
0.05
0.05
0.00 0.00
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
Figure 16.13 Chromatograms of BT obtained after various stress conditions. (From Ali,
M.S. et al., Chromatographia, 70, 539, 2009. With kind permission from Springer Science +
Business Media.)
secondary peak(s) from the parent peak BT (Figure 16.13) with peak purity of 100%.
Similarly, complete resolution of SCG from degradation compounds was observed
in stress studies (Figure 16.14). The HILIC method used to quantify SCG could not
retain its two impurities namely SCG impurity 1 (Cromolyn diethyl ester) and SCG
impurity 2 (1,3-bis(2-acetyl-3-hydroxyphenoxy)-2-propanol) because of their non-
polar and hydrophobic nature. These impurities eluted in dead volume and did not
interfere with SCG quantification. Method-validation data are presented in Tables
16.3 and 16.4. The data suggests that the methods are specific, accurate, precise, and
robust for the intended analyses.
16.4 Advantages
By utilization of the unique selectivity of HILIC, an analyst is now able to separate
a wide variety of difficult to retain polar compounds. High organic mobile phase is
ideal for efficient desolvation and compound ionization in ESI-MS. It can provide

0.15
0.15
0.10 0.10
Volts
Volts
0.05 0.05
0.00 0.00
SCG
SCG
0 5 10 0 5 10
(a) Minutes (b) Minutes
0.15
0.10
0.10
Volts
Volts
0.05
0.05
0.00 0.00
SCG
SCG
0 5 10 0 5 10
(c) Minutes (d) Minutes
Figure 16.14 Chromatograms of SCG obtained due to various stress conditions.

(a) Standard (no stress), (b) Oxidative (5% H2O2, 5â•›m L, 80°C, 5â•›h), (c) Photolytic-254â•›nm
(24â•›h), and (d) Basic (0.1â•›N NaOH, 5â•›m L, RT, 2â•›h). (Reproduced from Ali, M.S. et al., J. Sep.
Sci., 31, 1645, 2008. With permission from Wiley-VCH Verlag Gmbh & Co. KGaA.)
enhanced sensitivity and lower limits of detection in ESI-MS. The methodology has
the potential to replace many RP-HPLC methods by providing superior peak shapes
and good resolution. HILIC can be operated at high flow rate due to less viscous ace-
tonitrile used in the mobile phase, which provided lower column back pressure and
could be useful in high-throughput analysis.

Retention of Polar Acidic and Basic Drugs by HILIC
Table 16.3
Summary of Method-Validation Data
Results and Observations
Determination of PSH, DPH, and DXH
Validation Parameters Data Element PSH DPH DXH Determination of BT
Specificity Peak purity Spectrally pure peaks Spectrally pure peaks Spectrally pure peaks Spectrally pure BT peak
with degraded solutions with degraded solutions with degraded solutions with degraded solutions
Resolution >2.0 between secondary >2.0 between secondary >2.0 between secondary >2.0 between secondary
peak(s) and main peaks peak(s) and main peaks peak(s) and main peaks peak(s) and BT peak
Linearity R2 0.99974 (50%–200%) 0.99965 (50%–200%) 0.99985 (50%–200%) 0.99998 (20%–200%)
Accuracy Recovery 99.0%–101.6% (n = 5) 99.4%–100.6% (n = 5) 100.1%–101.8% (n = 5) 99.3%–101.2% (n = 8)
(50%–150%) (50%–150%) (50%–150%) (20%–200%)
Average 100.7% (n = 5) 100.1% (n = 5) 100.8% (n = 5) 100.3% (n = 8)
RSD 0.99% (n = 5) 0.75% (n = 5) 0.61% (n = 5) 0.31% (n = 8)
Instrumental precision RSD 0.56% (n = 6) 0.66% (n = 6) 0.59% (n = 6) 0.22% (n = 6)
Repeatability RSD <1.0% (n = 6) <1.0% (n = 6) <1.0% (n = 6) <0.53% (n = 5)
Intermediated precision RSD 0.66% (n = 20) 0.64% (n = 20) 0.91% (n = 20) 0.62% (n = 12)
Recovery 100.5% (n = 20) 100.2% (n = 20) 101.1% (n = 20) 100.63% (n = 12)
Robustness T 1.90–1.93 1.92–1.95 1.95–1.96 1.28–1.33
N >4300 >8000 >4500 >7500
k >2.4 >3.4 >6.3 >3.0
Rs N/A 5.0–5.4 9.2–9.8 N/A
Peak area (RSD) 0.25% (n = 6) 0.33% (n = 6) 0.18% (n = 6) <0.4% (n = 6)
Limit of detection (LOD) S/N ratio 8.5 (0.75â•›μg/mL) 3.3 (0.3125â•›μg/mL) 15.7 (0.375â•›μg/mL) 3.2 (0.005â•›μg/mL)
Limit of quantitation (LOQ) S/N ratio 16.5 (1.5â•›μg/mL) 10.1 (0.625â•›μg/mL) 32.7 (0.750â•›μg/mL) 12.7 (0.02â•›μg/mL)
RSD (at LOQ) 3.3% (n = 6) 2.8% (n = 6) 3.5% (n = 6) 1.9% (n = 6)
Solution stability (after 24â•›h) Assay (%) No significant change No significant change No significant change No significant change
Secondary peaks Not found Not found Not found Not found
395
Table 16.4
Summary of Method-Validation Data
Results and Observations
Determination of MFH, CGD, and MLN
Validation Parameters Data Element MFH (Assay) CGD (as Impurity) MLN (as Impurity) Assay of SCG
Specificity Peak purity Spectrally pure peaks with Spectrally pure peaks with Spectrally pure peaks Spectrally pure SCG peak
degraded solutions degraded solutions. with degraded solutions. with degraded solutions.
Resolution >2.0 between secondary, >2.0 between secondary, >2.0 between secondary, >2.0 between secondary
MFH, MLN, and CGD MFH, MLN, and CGD MFH, MLN, and CGD peak(s) and SCG peak
peak(s) peak(s) peak(s)
Linearity (25%–200%) R2 0.9999 0.99985 0.99985 0.99999
Accuracy Recovery 99.8%–101.0% (n = 8) 96.4%–102.3% (n = 4) 96.7%–101.9% (n = 4) 99.8%–100.7% (n = 7)
(20%–200%) (50%–200%) (50%–200%) (25%–200%)
Average 100.27% (n = 8) 99.7% (n = 12) 99.6% (n = 12) 100.4% (n = 7)
RSD 0.56% (n = 8) 1.66% (n = 12) 1.79% (n = 12) 0.31% (n = 7)
Instrumental precision RSD 0.65% 1.76% 1.58% 0.17% (n = 6)
Repeatability RSD <1.35% (50%–200%) <1.35% (50%–200%) <1.35% (50%–200%) <0.2% (n = 3)
(50%–150%)
Intermediate precision RSD 0.71% (n = 12) 1.58% (n = 12) 1.50% (n = 12) 0.64% (n = 12)
Recovery 100.3% (n = 12) 100.2% (n = 12) 100.4% (n = 12) 100.12% (n = 12)
Robustness T 1.05–1.12 1.11–1.13 1.03–1.06 0.90–1.12
N >15,000 >6,500 >12,000 >3,500
k >7.0 >0.6 >3.5 >2.0
Rs N/A 22.0–28.0 (CGD–MLN) 14.1–17.6 (MLN–MFH) N/A
Peak area (RSD) 0.14% (n = 6) 0.55% (n = 6) 0.51% (n = 6) <1.0% (n = 6)
Limit of detection (LOD) S/N ratio 2.9 (100â•›ng/mL) 2.8 (5.0â•›ng/mL) 3.8 (25â•›ng/mL) 2.9 (25â•›ng/mL)
Limit of quantitation (LOQ) S/N ratio 10.4 (350â•›ng/mL) 10.7 (25â•›ng/mL) 11.2 (75â•›ng/mL) 12.3 (100â•›ng/mL)
RSD (at LOQ) 1.5% (n = 6) 1.9% (n = 6) 2.0% (n = 6) 2.9% (n = 6)
Solution stability (after 24â•›h) Assay (%) No significant change No significant change No significant change No significant change
Secondary peaks Not found Not found Not found Not found

16.5 Conclusion
The described HILIC methods are suitable for the analysis of drug substances, such
as PSH, DPH, DXH, MFH, CGD, MLN, and BT in various pharmaceutical products.
The methods are superior to those GC or RPLC methods described previously. The
use of silica column has been proved to be beneficial to retain polar compounds.
Unique selectivity is resulted due to mixed-mode retention mechanisms. Various
polar acidic and basic analytes can be separated using HILIC. Diluent has a signifi-
cant impact on analyte peak shape and should be kept similar to the mobile-phase
composition if possible. Yet, water as a solvent proved to be another useful entity
because of its affinity to silica and creating an enriched layer for retention of polar
analytes. It is believed that many analytical problems posed by small and polar com-
pounds can be solved by HILIC.
References
1. Olsen, B. A. 2001. Hydrophilic interaction chromatography using amino and silica
columns for the determination of polar pharmaceuticals and impurities. J. Chromatogr.
A 913:113–122.
2. Layne, J. 2002. Characterization and comparison of the chromatographic performance
of conventional, polar-embedded, and polar-endcapped reversed-phase liquid chroma-
tography stationary phases. J. Chromatogr. A 957:149–164.
3. Hanai, T. 2003. Separation of polar compounds using carbon columns. J. Chromatogr.
A 989:183–196.
4. Neue, U. D., Grumbach, E. S., Mazzeo, J. R., Tran, K., Wagrowski-Diehl, D. M. 2003.
Method development in reversed-phase chromatography, Bioanalytical Separations, In
Handbook of Analytical Separations, ed. I. D. Wilson. Elsevier Science B.V., New York,
pp. 185–214.
5. Reid, T. S., Henry, R. A. 1999. Am. Lab. 31:24–28.
6. Hemstroem, P., Irgum, K. 2006. Review. Hydrophilic interaction chromatography.
J. Sep. Sci. 29:1784–1821.
7. Gustavsson, S. A., Samskog, J., Markides, K., Langstrom, B. 2001. Studies of signal
suppression in liquid chromatography-electrospray ionization mass spectrometry using
volatile ion-pairing reagents. J. Chromatogr. A 937:41–47.
8. Lunn, G., Hellwig, L. C. 1998. Handbook of Derivatization Reactions for HPLC. Wiley &
Sons, Inc., New York.
9. Alpert, A. J. 1990. Hydrophilic interaction chromatography (HILIC): A new method
for separation of peptides, nucleic acids and other polar solutes. J. Chromatogr.
499:177–196.
10. Ali, M. S., Ghori, M., Rafiuddin, S., Khatri, A. R. 2007. A new hydrophilic interaction
liquid chromatographic (HILIC) procedure for the simultaneous determination of pseu-
doephedrine hydrochloride (PSH), diphenhydramine hydrochloride (DPH) and dextro-
methorphan hydrobromide (DXH) in cough-cold formulations. J. Pharm. Biomed. Anal.
43:158–167.
11. Ali, M. S., Rafiuddin, S., Ghori, M., Khatri, A. R. 2007. Simultaneous determination
of metformin hydrochloride, cyanoguanidine and melamine in tablets by mixed-mode
hydrophilic interaction liquid chromatography. Chromatographia 67:517–525.
12. Ali, M. S., Khatri, A. R., Munir, M. I., Ghori, M. 2009. A stability-indicating assay of
brimonidine tartrate in ophthalmic solution and stress testing by hydrophilic interaction
chromatography (HILIC). Chromatographia 70:539–544.

13. Ali, M. S., Al-Jawi, D. A., Al-Hetari, Y., Ghori, M., Khatri, A. R. 2008. Stability-
indicating assay of sodium cromoglicate in ophthalmic solution using mixed-mode
hydrophilic interaction chromatography. J. Sep. Sci. 31:1645–1650.
14. Naidong, W. 2003. Bioanalytical liquid chromatography tandem mass spectrometry
methods on underivatized silica columns with aqueous/organic mobile phases.
J. Chromatogr. B 796:209–224.
15. Bidlingmeyer, B. A., Del Rios, J. K., Korpi, J. 1982. Separation of organic amine com-
pounds on silica gel with reversed-phase eluents. Anal. Chem. 54:442–447.
16. Cox, G. B., Stout, R. W. 1987. Study of the retention mechanism for basic compounds
on silica under ‘pseudo-reversed-phase’ conditions. J. Chromatogr. 384:315–336.
17. Strege, M. A., Stevenson, S., Lawrence, S. M. 2000. Mixed mode anion-cation
exchange/hydrophilic interaction liquid chromatography-electrospray mass spectrom-
etry as an alternative to reversed phase for small molecule drug discovery. Anal. Chem.
72:4629–4633.
18. Rabel, F. M., Caputo, A. G., Butts, E. T. 1976. Separation of carbohydrates on a new
polar bonded phase material. J. Chromatogr. 126:731–740.
19. Naidong, W., Shou, W., Chenn,Y. L., Jiang, X. 2001. Novel liquid chromatographic-tandem
mass spectrometric methods using silica columns and aqueous-organic mobile phases
for quantitative analysis of polar ionic analytes in biological fluids. J. Chromatogr. B
754:387–399.
20. Naidong, W., Shou, W. Z., Addison, T., Maleki, S., Jiang, X. 2002. Liquid
chromatography-tandem mass spectrometric bioanalysis using normal-phase columns
with aqueous-organic mobile phases—A novel approach of eliminating evaporation and
reconstitution steps in 96-well SPE. Rapid Commun. Mass Spectrom. 16:1965–1975.
21. Grumbach, E. S., Diehl, D. M., McCabe, D. R., Mazzeo, J. R., Neue, U. D. 2003.
Hydrophilic interaction chromatography for enhanced ESI-MS sensitivity and retention
of polar basic analytes. LCGC N. Am. Suppl. S:53–54.
22. Grumbach, E. S., Diehl, D. M., Mazzeo, J. R. 2004. A sensitive ESI-MS HILIC method
for the analysis of acetylocholine and choline. LCGC N. Am. Suppl. S:74–75.
23. Su, J., Hirji, R., Zhang, L., He, C., Selvaraj, G., Wu, R. 2006. Evaluation of the stress-
inducible production of choline oxidase in transgenic rice as a strategy for producing the
stress-protectant glycine betaine. J. Exp. Bot. 57:1129–1135.
24. Neue, U. D. 1997. HPLC Columns Theory, Technology and Practice. John Wiley &
Sons, Hoboken, NJ, pp. 217–223.
25. Tolstikov, V. V., Fiehn, O. 2002. Analysis of highly polar compounds of plant origin:
Combination of hydrophilic interaction chromatography and electrospray ion trap mass
spectrometry. Anal. Biochem. 301:298–307.
26. Beumer, J. H., Joseph, E., Egorin, M. J., Covey, J. M., Eiseman, J. L. 2006. Quantitative
determination of zebularine (NSC 309132), a DNA methyltransferase inhibitor, and
three metabolites in murine plasma by high-performance liquid chromatography cou-
pled with on-line radioactivity detection. J. Chromatogr. B Analyt. Technol. Biomed. Life
Sci. 831:147–155.
27. Strege, M. A. 1998. Hydrophilic interaction chromatography-electrospray mass spec-
trometry analysis of polar compounds for natural product drug discovery. Anal. Chem.
70:2439–2445.
28. McClintic, C., Remick, D. M., Peterson, J. A., Risley, D. S. 2003. Novel method for the
determination of piperazine in pharmaceutical drug substances using hydrophilic inter-
action chromatography and evaporative light scattering detection. J. Liq. Chromatogr.
Relat. Technol. 26:3093–3104.
29. Tanaka, H., Zhou, X., Masayoshi, O. 2003. Characterization of a novel diol column for
high-performance liquid chromatography. J. Chromatogr. A 987:119–125.

30. Uutela, P., Reinila, R., Piepponen, P., Ketola, R. A., Kostiainen, R. 2005. Analysis of
acetylcholine and choline in microdialysis samples by liquid chromatography/tandem
mass spectrometry. Rapid Commun. Mass Spectrom. 19:2950–2956.
31. Higley, T. J., Yoshida, T. 2003. Separation of peptides by hydrophilic interaction chro-
matography using TSKgel Amide-80. LCGC N. Am., June: Suppl. S:20–21.
32. Takahashi, N. 1996. Three-dimensional mapping of N-linked oligosaccharides using
anion-exchange, hydrophobic and hydrophilic interaction modes of high-performance
liquid chromatography. J. Chromatogr. A 720:217–225.
33. Alpert, A. J., Shukla, M., Shukla, A. K., Zieske, L. R. et al. 1994. Hydrophilic-interaction
chromatography of complex carbohydrates. J. Chromatogr. A 676:191–202.
34. Neue, U. D., Pheobe, C. H., Tran, K., Cheng, Y. F., Lu, Z. 2001. Dependence of reversed-
phase retention of ionizable analytes on pH, concentration of organic solvent and silanol
activity. J. Chromatogr. A 925:49–67.
35. Florence, A. T., Attwood, D. 2006. Physicochemical Principles of Pharmacy, 4th edn.,
Pharmaceutical Press, London, U.K., pp. 75–82.
36. Van de Merbel, N. C., Wilkens, G., Fowless, S., Oosterhuis, B., Jonkman, J. H. G.
1998. LC phases improve, but not all assays do: Metformin bioanalysis revisited.
Chromatographia 47:542–546.
37. Column Care and Use Instructions—Atlantis™ Columns, Waters Corporation, Milford,
MA, p. 8.
38. Hirt, R. C., Schmitt, R. G. 1958. Ultraviolet absorption spectra of derivatives of symmet-
ric triazine—II: Oxo-triazines and their acyclic analogs. Spectrochim. Acta 12:127–138.
39. Hu, W., Tao, H., Haraguchi, H. 1994. Evaluation of sulfobetaine-type zwitterionic sta-
tionary phases for ion chromatographic separation using water as a mobile phase. Anal.
Chem. 66:2514–2520.
40. ICH Harmonized Tripartite Guideline, Validation of Analytical Procedures: Text and
Methodology, Q2(R1), November 2005, ICH Steering Committee, ICH Secretariat,
Geneva, Switzerland.

17 Retention and
Selectivity of Polar
Stationary Phases for
Chromatography
Yong Guo
Contents
17.1 Introduction................................................................................................... 401
17.2 Polar Stationary Phases for HILIC................................................................402
17.3 Selectivity of Various Polar Stationary Phases in HILIC..............................406
17.4 Retentivity of Various Polar Stationary Phases in HILIC............................. 413
17.5 Factors Affecting Retention and Selectivity in HILIC.................................. 415
17.6 Separation of Positional Isomers in HILIC................................................... 422
17.7 Conclusions.................................................................................................... 423
Acknowledgment.................................................................................................... 424
References............................................................................................................... 424
17.1 Introduction
Small polar compounds, although representing only a small portion of therapeu-
tic drugs, are often very challenging to pharmaceutical analysis. Lack of sufficient
retention on reversed-phase columns makes it difficult to analyze polar compounds
by reversed-phase liquid chromatography (RPLC). Hydrophilic interaction chroma-
tography (HILIC) provides a viable alternative to ion-pairing or normal phase meth-
ods, which are typically used for the analysis of polar and/or ionizable compounds.
Recent reviews on HILIC reveal increasing popularity of HILIC in pharmaceutical
analysis of small polar drugs (Hemstrom and Irgum 2006, Dejaegher and Heyden
2010, Jian et al. 2010). The term “hydrophilic interaction chromatography” (HILIC)
was first coined by Andrew Alpert in 1990 (Alpert 1990); however, similar separa-
tion technique had been used for sugar analysis much earlier (Linden and Lawhead
1975, Palmer 1975). Alpert also postulated the retention mechanism for HILIC as
hydrophilic partitioning of polar solutes between the organic solvent-rich mobile
phase and a water-rich liquid layer immobilized on the surface of polar stationary
401

phases (Alpert 1990). McCalley and Neue (2008) recently demonstrated possible
presence of a water layer on porous silica surface by measuring the elution volume of
benzene using a mobile phase containing high levels of acetonitrile. More published
studies seem to support hydrophilic partitioning as the major retention mechanism
for HILIC, but there are mounting evidences that other interactions (e.g., adsorption,
electrostatic, hydrogen-bonding, dipole–dipole interaction) also play very important
roles in HILIC separation (Olsen 2001, Li and Huang 2004, Guo and Gaiki 2005,
Hao et al. 2008).
Most HILIC separation was performed on the polar stationary phases that were
typically used in normal-phase chromatography (e.g., amino, cyano, and silica
phases) in the early days. The normal-phase columns had to be first converted into
the reversed-phase conditions and then used for HILIC separation. With increasing
popularity of HILIC, more and more polar phases with diverse functional groups
have been designed and developed for HILIC in recent years (Dejaegher et al. 2008,
Ikegami et al. 2008, Jandera 2008). The availability of various polar phases provides
method development chemists with the opportunity to find an appropriate phase for
the desired separation. At the same time, it also presents a big challenge to select the
optimal phase during method development in a systematic manner in a short time.
As in RPLC, the choice of a stationary phase should be based on a thorough under-
standing of the retention and selectivity characteristics of the available stationary
phases. This chapter focuses on the retention and selectivity of various polar phases
commercially available.
17.2 Polar Stationary Phases for HILIC

Underivatized silica is a popular phase for HILIC application, particularly in bio-
analytical fields (Wen 2003, Jian et al. 2010). Irreversible adsorption of solutes is not
as problematic in HILIC as in normal-phase chromatography due to the presence
of a significant level of water in the mobile phase. The silica columns for normal-
phase chromatography have been used for HILIC separation (Olsen 2001, Hao et al.
2007); however, the columns need to be washed extensively to remove normal-phase
solvents. Significant difference in retention has been reported among the silica col-
umns from different manufactures (Olsen 2001, Hao et al. 2008). A new superficially
porous silica column has been shown to have greater resistance to overloading for
ionized basic compounds than the reversed-phase columns (McCalley 2008). There
are more and more silica columns on the market that are developed exclusively for
HILIC application, such as Atlantis HILIC silica column. The silica columns pro-
moted for HILIC application are typically packed and stored in aqueous organic
solvents instead of normal-phase solvents.
In addition to the silica phase, a wide variety of polar functional groups have been
employed as stationary phases in HILIC (Hemstrom and Irgum 2006, Ikegami et al.
2008, Jandera 2008). Most bonded phases used for HILIC can be classified into three
major categories based on the charge characteristics of the functional groups, namely,
neutral, cationic, and zwitterionic groups as listed in Table 17.1. The polar phases in
Table 17.1 are presented by the conventional names of the functional groups.

Retention and Selectivity of Polar Stationary Phases
Table 17.1
Polar Stationary Phases Commonly Used in HILIC
Phase Type Phase Name Functional Group
Neutral Amide R1 R3 NH2
Si
n m
R2
R4 O
Aspartamide OH
HO
HN
O NH
O O O
NH
N N
H H n
H
Si N O
O
Diol
Si OH
OH
(continued)
403
Polar Stationary Phases Commonly Used in HILIC
Phase Type Phase Name Functional Group
Cationic Amino
Si NH2
Imidazole
Si N
N
Triazole N NH
Si
N
Zwitterionic Sulfobetaine +
Si N SO3–
Phosphocholine O
O
Si P N+
–O O

Retention and Selectivity of Polar Stationary Phases 405
The neutral group includes many popular HILIC phases, such as amide, diol,
and cross-linked diol phases. The amide phase was originally developed for car-
bohydrate analysis under “aqueous normal phase” condition in 1988, but has
become one of the most popular phases for HILIC. The functional group on the
amide phase is carbamoyl moiety positioned at the distal end of an aliphatic chain
attached to the silica surface. The diol phase is the same as the one used in normal-
phase chromatography. Some diol columns are packed in reversed-phase solvents,
but some are packed in normal-phase solvents, which need extensive washing to
convert to aqueous conditions. The cross-linked diol phase (structure not shown) is
covalently bonded to the silica surface and the diol groups are highly cross-linked
through ether linkages (Expolore Luna HILIC 2007). Similar to the cross-linked
diol phase, both polyhydroxy and polyvinyl alcohol (PVA) phases contain a layer
of polyhydroxylated and PVA polymer bound to the silica surface, respectively,
and have many hydroxyl groups (–OH) present on the polymer surface, which
render these phases polar. The aspartamide phase in this group is worth special
mention since it was the first stationary phase specifically developed for HILIC
separation. The aminopropylated silica was first coated with a layer of polysuccin-
imide, which was then treated with ethanolamine to generate the final stationary
phase. It is not as commonly used as other phases possibly due to relatively lower
efficiency (Hemstrom and Irgum 2006).
The second category includes the stationary phases that can be positively
charged in the pH range typically used for HILIC separation (pH 3–8). The amino,
imidazole, and triazole phases are bonded to the silica surface through an ali-
phatic linker. The amino phase has been used for HILIC separation even before
the term HILIC was invented, and still remains very popular for various applica-
tions. Many amino columns are prepared for normal-phase separation, and have
significant differences in retention for polar solutes when run in HILIC mode (Guo
and Huang 2003). The amino phase is also not very stable over a long period of
time. Polyamine phases containing secondary and tertiary amine groups are more
stable, but have different selectivity than the conventional amino phase. The cat-
ionic group also includes the imidazole and triazole phases specially designed for
HILIC application. The imidazole phase is attached to the silica surface through
the secondary amine moiety on the imidazole ring; however, the point of attach-
ment on the triazole phase is unknown. The positive charges on these phases can
have direct impact on retention and selectivity for charged solutes through electro-
static interactions.
The third group is the zwitterionic phases specially designed for HILIC. Two
representative zwitterionic phases are sulfobetaine and phosphocholine phases,
which differ in the placement of charged groups (Jiang 2003, Jiang et al. 2006).
The quaternary amine group bearing positive charge is at the proximal end of
the sulfobetaine phase, but at the distal end of the phosphocholine phase. The
sulfobetaine phase has been gaining popularity in HILIC rapidly in recent years;
however, the phosphocholine phase has not been commercialized at the time of
writing.

17.3 Selectivity of Various Polar Stationary

Phases in HILIC
In RPLC, stationary phase chemistry (e.g., C18 vs. phenyl phase) has been widely used
to change selectivity and achieve desired separation (Snyder et al. 1997, Kazakevich
and Lobrutto 2007). Although a variety of functional groups have been used as the
stationary phase (Table 17.1), there is a lack of critical understanding of the selectiv-
ity of various polar phases in HILIC. With the availability of more and more HILIC
phases, this has become particularly important since the knowledge about various
HILIC phases is useful to select appropriate columns during method development.
The selectivity of the polar phases presented in Table 17.1 has been experimentally
investigated under the same chromatographic conditions. A group of nucleosides and
nucleic acid bases were used as the model compounds for this investigation (Guo and
Gaiki 2005). The model compounds were eluted isocratically using a mobile phase of ace-
tonitrile and ammonium acetate solution (85/15, v/v). The selectivity of the amide, aspar-
tamide, sulfobetaine, and silica phases is well illustrated by the chromatograms shown
in Figure 17.1. Major selectivity difference is observed among the selected phases. For
example, adenosine and uridine are not separated on the amide phase, but almost baseline
resolved on the aspartamide phase. The selectivity of the sulfobetaine phases is similar to
that of the aspartamide phase. In comparison, the silica phase has a very different selectiv-
ity for the model compounds than the amide, aspartamide and sulfobetaine phase.
Figure 17.2 shows the separation of the model compounds on the diol, cross-linked
diol, polyhydroxy, and PVA phases, which all have –OH groups on the packing sur-
face. These phases share relatively similar selectivity for the model compounds. The
only difference is that some phases (e.g., cross-linked diol and polyhydroxy phase)
have better selectivity for cytidine and guanosine. In addition, it is interesting to note
that the silica phase (Figure 17.1) seems to be very different from the phases shown
in Figure 17.2. This indicates that the silica phase may have some specific interac-
tions with the model compounds.
The separation of the model compounds on the cationic phases (e.g., amino,
imidazole, and triazole phase) is shown in Figure 17.3. Different selectivity of the
three phases is well depicted by the elution patterns. For examples, adenosine and
uridine barely resolved on the amino phase, but well separated on the imidazole and
triazole phases with opposite elution order. It is worth noting that the model com-
pounds are not charged at the mobile-phase pH used in this study. The difference
in selectivity most likely results from specific interactions (e.g., hydrogen bonding,
dipole interaction) between the stationary phases and solutes.
The selectivity of the various polar phases has also been evaluated using a group
of organic acids, which include acetylsalicylic acid and some of its metabolites (Guo
et al. 2007). Figures 17.4 through 17.6 show the separation of the acidic compounds
on various polar phases under the same chromatographic conditions. As shown in
Figure 17.4, the amide and aspartamide phases share similar selectivity for the acidic
compounds, but is different than the sulfobetaine phase. For example, hippuric acid
and α-hydroxyhippuric acid have opposite elution order on the sulfobetaine phase
compared to the amide and aspartamide phases. In addition, the silica phase also
shows very different selectivity than the other phases, particularly for gentisic acid.

0.60
0.50 1 Amide phase
0.40 2, 3
0.30
AU
4
0.20 6
0.10 5
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(A) Minutes
0.60 1 Aspartamide phase

0.50
0.40
AU
0.30 2
3 4
0.20 6
0.10 5
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(B) Minutes
0.60 Sulfobetaine phase

1
0.50
0.40
AU
0.30 2 4
0.20 3 6
5
0.10
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(C) Minutes
1.20
1 Silica phase
1.00
0.80
AU
0.60
3 2 6 4
0.40
5
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(D) Minutes
Figure 17.1 Separation of neutral model compounds on (A) amide, (B) aspartamide,
(C) sulfobetaine, and (D) silica phase. Mobile phase, acetonitrile/water (85/15, v/v) contain-
ing 10â•›m M ammonium acetate. Column dimension, 250â•›m m × 4.6â•›m m ID, 5â•›μm particle size.
Column temperature, 30°C. Flow rate, 1.5â•›m L/min. UV detection at 248â•›n m. Peak label:
(1) uracil, (2) adenosine, (3) uridine, (4) cytosine, (5) cytidine, and (6) guanosine.

1.00
1 Diol phase
0.80
0.60
4
AU
0.40 2 3 6
5
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(A) Minutes
1.00 Cross-linked diol phase

1
0.80
0.60 4
AU
6
2 3
0.40 5
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(B) Minutes
1.00
1 Polyhydroxy phase
0.80
0.60
AU
0.40 3 4
2 6
0.20 5
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(C) Minutes
1.00
1 Polyvinyl alcohol phase
0.80
0.60
AU
4
0.40 2 3 5, 6
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(D) Minutes
Figure 17.2 Separation of neutral model compounds on (A) diol, (B) cross-linked diol,
(C) polyhydroxy, and (D) PVA phase. Other conditions are the same as Figure 17.1.

0.80 Amino phase

1
0.70
0.60
0.50
AU
0.40 2
4
0.30 3 6
0.20 5
0.10
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(A) Minutes
1.00
1 Imidazole phase
0.80
0.60
AU
0.40 2 4
3 6
0.20 5
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(B) Minutes
1.00 Triazole phase

1
0.80
0.60
AU
4
0.40 3
2 6
5
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(C) Minutes
Figure 17.3 Separation of neutral model compounds on (A) amino, (B) imidazole, and
(C) triazole phase. Other conditions are the same as Figure 17.1.
Figure 17.5 shows the separation of the acidic compounds on the diol, cross-linked
diol, polyhydroxy, and PVA phases. The diol, polyhydroxy, and PVA phases display
similar selectivity for the acidic compounds with a small difference in the resolution
of hippuric acid and α-hydroxyhippuric acid. In contrast, the cross-linked diol phase
seems to be much less selective for the acids with two pairs co-eluting, possibly due
to insufficient retention.
The separation of the acids on the amino, imidazole and triazole phase is shown
in Figure 17.6. The elution order of the acids on the amino phase bears some

0.90
0.80 1 Amide phase
0.70
0.60
0.50 2
5
AU
0.40
0.30 3 4 6
0.20
0.10
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(A) Minutes
1.00 1 Aspartamide phase

0.80
2
0.60
AU
5
0.40 3 4
6
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(B) Minutes
1.80
1.60 1 Sulfobetaine phase
1.40
1.20 2
1.00
5
AU
0.80 6
0.60 3 4
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
(C) Minutes
1.00 Silica phase

1
0.80
0.60 6
AU
0.40 3 4 2 5
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(D) Minutes
Figure 17.4 Separation of acidic model compounds on (A) amide, (B) aspartamide, (C) silica,
and (D) sulfobetaine phase. Mobile phase, acetonitrile/water (85/15, v/v) containing 20â•›mM ammo-
nium acetate. Column dimension, 250â•›mm × 4.6â•›mm ID, 5â•›μm particle size. Column temperature,
30°C. Flow rate, 1.0â•›mL/min. UV detection at 228â•›nm. Peak label: (1) salicylic acid, (2) gentisic
acid, (3) acetylsalicylic acid, (4) salicyluric acid, (5) hippuric acid, and (6) α-hydroxyhippuric acid.

1.40
1.20 1 Diol phase
1.00
0.80 2 5
6
AU
0.60 4
3
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(A) Minutes
1.40
2, 3 Cross-linked diol phase
1.20 5, 6
1.00 1
0.80
AU
0.60 4
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(B) Minutes
1.20 Polyhydroxy phase

1
1.00
0.80 2
5
AU
0.60 6
3 4
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(C) Minutes
1.40
1 PVA phase
1.20
1.00
2 5
0.80
AU
3 4 6
0.60
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(D) Minutes
Figure 17.5 Separation of acidic model compounds on (A) diol, (B) cross-linked diol,
(C) polyhydroxy, and (D) PVA phase. Other conditions are the same as Figure 17.4.

0.80
Amino phase
0.70 1
0.60
0.50
0.40 2
AU
0.30
4 5
0.20 3 6
0.10
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00 30.00
(A) Minutes
Imidazole phase
0.80 1
0.60
AU
0.40 2 4, 5
3 6
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00 30.00
(B) Minutes
1.20
Triazole phase
1.00 1
0.80
2 5
0.60 6
AU
0.40 3 4
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00 30.00
(C) Minutes
Figure 17.6 Separation of acidic model compounds on (A) amino, (B) imidazole, and
(C) triazole phase. Other conditions are the same as Figure 17.4.
resemblance to that on the amide and aspartamide phases as shown in Figure 17.4,
but α-hydroxyhippuric acid has a significant tailing. The selectivity of the imidazole
and triazole phases is very different from the amino phase, for example, gentisic
acid and acetylsalicylic acid are well resolved on the amino phase, but only partially
resolved on the imidazole and triazole phases. It is also interesting to compare the
selectivity of the imidazole and triazole phases. On the imidazole phase, salicyluric
acid co-elutes with hippuric acid, but is separated from α-hydroxyhippuric acid. In
comparison, salicyluric acid is separated from hippuric acid, but nearly co-elutes
with α-hydroxyhippuric acid on the triazole phase. It is important to point out that

the acids were negatively charged and the amino and imidazole groups were posi-
tively charged in the mobile phase (pHâ•›∼â•›6); however, it is not clear whether the tri-
azole phase was charged or not. The difference in selectivity among the cationic
phases is at least partially related to the electrostatic interactions between the nega-
tively charged solutes and positively charged functional groups.
17.4 Retentivity of Various Polar Stationary

Phases in HILIC
Stronger retention for polar compounds is the primary reason for the increasing pop-
ularity of HILIC. In addition to selectivity, retentivity is another factor important to
achieve desired resolution according to Equation 17.1 (Snyder et al. 1997):
1 α − 1 k′
Rs = ⋅ N⋅ ⋅ (17.1)
4 α 1 + k′
where
N is the number of theoretical plates
α and k′ are the separation and retention factors, respectively
Since HILIC separation is based on hydrophilic partitioning of the solutes between

the mobile phase and a water-rich liquid layer immobilized on the surface of pack-
ing materials, weak retentivity of the stationary phases can limit the content of the
organic solvents in the mobile phase to a narrow range, thus limiting selectivity.
In RPLC, toluene is usually used as the reference compound to evaluate relative
retention of different stationary phases. However, there has not been a compound
widely accepted as the reference compound to evaluate the retention of the station-
ary phases in HILIC. Cytosine was used as the reference compound to compare the
relative retentivity of various polar phases listed in Table 17.1. The capacity factors
of cytosine on the selected polar phases, in the order of increasing retention, are
graphically presented in Figure 17.7. It is worth noting that hydrophilic partitioning
was likely the primary retention mechanism since cytosine is uncharged under the
experimental conditions. The bar chart in Figure 17.7 clearly demonstrates different
retentivity of various polar phases in HILIC. The cross-linked diol phase has the
least retention for cytosine. The amide and aspartamide phases, on the other hand,
have stronger retention. Interestingly, most phases have relatively similar retention
for cytosine, such as PVA, triazole, silica, diol, amino sulfobetaine, polyhydroxy, and
imidazole phases. The diversity of the chemical structure in these phases does not
seem to make a significant difference in the retentivity. This may be a reflection of
hydrophilic partitioning as the underlining retention mechanism for these stationary
phases as opposed to strong specific interactions.
Salicyluric acid was also used as the reference compound to evaluate the rela-
tive retention of the polar phases. As shown in Figure 17.7, there are some similar-
ities, and also noticeable differences in the relative retention of the polar phases
when salicyluric acid, instead of cystosine, is used as the reference compound.

8
7 Cytosine
Capacity factor (k΄) 6 Salicyluric acid
5
4
3
2
1
0
ol
le
lfo iol
e
e
e
y
id
in
id
lic
ol
ox
in
PV
zo
di
az
m
ta
m
Si
dr
ia
d
be
ta
id
A
ke
hy
Tr
ar
Im
lin
ly
sp
Po
Su
s-
A
os
Cr
Figure 17.7 Capacity factors of cytosine and salicyluric acid on various HILIC phases.
Conditions are the same as Figure 17.1 for cytosine and Figure 17.4 for salicyluric acid.
The cross-linked diol phase is the least retentive phase for both cytosine and
salicyluric acid. The other neutral phases (e.g., the diol, polyhydroxy, and PVA
phases) also show similar retentivity for salicyluric acid. However, the cationic
phases, the amino and imidazole phase in particular, have much stronger reten-
tion for the acid, which can be attributed to the electrostatic attraction between
the negatively charged acid and positively charged functional groups under the
experimental conditions. In contrast, the silica, amide, and aspartamide phases
seem to have reduced retention for the acid compared to the other phases. All
these phases have negative charges on the packing surface, silica from deproton-
ated silanol groups, amide from the residual silanol groups, and aspartamide from
residual negative charges from C-terminus. It is possible that the reduced reten-
tion of the acid on these phases is related to electrostatic repulsion of the nega-
tively charged acid. However, there is not sufficient evidence to indicate that the
electrostatic repulsion is the only factor contributing to the decrease in retention.
The above examples demonstrate that the electrostatic interactions between
the charged stationary phases and solutes are very important to the retention and
selectivity in HILIC. The cationic phases (e.g., amino phase) are particularly sub-
jective to the electrostatic interactions. The retention can be strengthened for the
negatively charged compounds, but weakened for the positively charged com-
pounds on the cationic phases. The negative charges found in the silica, amide,
and aspartamide phases can also induce electrostatic interactions with charged
solutions. However, the electrostatic interactions on these phases are not over-
whelmingly strong, in comparison to the cationic phase. On the other hand, some
neutral phases (e.g., diol, cross-linked diol, polyhydroxy, and PVC phases) are less
vulnerable to the electrostatic interactions, thus have relatively similar retentivity
regardless of the charge state of the solutes. In RPLC, it is rather straightforward to
determine the relative retentivity of the stationary phase based on the hydrophobic-
ity of the functional groups (Snyder et al. 1997, Kazakevich and Lobrutto 2007).

For example, the C18 phase has stronger retention than the C8 or phenyl phase.
In HILIC, however, it is difficult to link the relative retentivity to the functional
group. Empirical comparison can provide useful guidance for column selection in
HILIC method development.
17.5 F actors Affecting Retention

and Selectivity in HILIC
In addition to stationary phase chemistry, other chromatographic parameters, such
as organic solvents, mobile-phase pH, salt concentration, and column temperature
can also have direct impact on the retention and selectivity in HILIC. The mobile
phase for HILIC separation typically contains organic solvents (e.g., acetonitrile,
ethanol, and THF) at high levels (above 60%, v/v), which is needed to effect hydro-
philic interaction (Guo and Gaiki 2005, Hemstrom and Irgum 2006). The effect of
organic content in the mobile phase on retention is depicted in Figure 17.8, which
shows the curves of log k′ vs. acetonitrile content in the mobile phase for acetylsali-
cylic acid and cytosine on four different stationary phases (i.e., amino, amide, silica,
and sulfobetaine phases). The retention of the model compounds typically increases
with the acetonitrile content in a nonlinear fashion in the range of 65%–95% (v/v).
At lower acetonitrile content (<85%), log k′ increases almost linearly with the ace-
tonitrile content in the mobile phase for both the neutral and acidic compounds on
the selected phases, except acetylsalicylic acid on the amino phase. However, the
retention increases much faster at the acetonitrile content above 85%, indicating that
the retention is very sensitive to small changes in the organic content. Interestingly,
the retention of acetylsalicylic acid on the amino phase remains almost unchanged
in the acetonitrile range of 65%–80%, but increases as the acetonitrile content goes
above 80%. This observation is likely related to the electrostatic interaction between
the negatively charged acid and positively charged amino phase. The dominant
retention mechanism might be the electrostatic attraction at low acetonitrile lev-
els. As the acetonitrile content increases, the retention might be more controlled
by hydrophilic partitioning, thus more influenced by the acetonitrile content in the
mobile phase.
Mobile-phase pH is another chromatographic parameter critical to the reten-
tion of ionizable compounds in HILIC, which is subjected to the influence of the
mobile-phase pH through ionization. Typically, charged solutes are more retained
than uncharged ones in HILIC (Guo and Gaiki 2005, Hemstrom and Irgum 2006).
The plots in Figure 17.9 show the variation of the retention time of acetylsalicylic
acid and cytosine with the mobile-phase pH in the range of 3.3–6.5 on the amino,
amide, silica, and sulfobetaine phases. For acetylsalicylic acid, the retention time
is relatively unchanged between pH 6.5 and 4.8. As the mobile-phase pH decreases
from 4.8 to 3.3, there is a gradual decrease in retention time on the amide, silica,
and sulfobetaine phase, but a large drop in retention time at pH 3.3 on the amino
phase. Acetylsalicylic acid has a pK aâ•›∼â•›3.5 and is predominantly charged at pH 6.5
and 4.8, but is partially uncharged at pH close to its pKa through protonation. The
uncharged acid has less retention, resulting in a gradual decrease in retention time
at lower mobile-phase pH on the amide, silica, and sulfobetaine phases. On the

1.5
Acetylsalicylic acid
0.5
ln k΄
–0.5
–1
60 70 80 90 100
(A) Acetonitrile content (%, v/v)
1.4
Cytosine
1.2
1
0.8
0.6
ln k΄
0.4
0.2
0
–0.2
–0.4
60 70 80 90 100
(B) Acetonitrile content (%, v/v)
Figure 17.8 The plots of log k′ vs. acetonitrile content (%) for (a) acetylsalicylic acid
and (b) cytosine on (♦) amino, (◾) amide, (▴) silica, and (×) sulfobetaine phase. Column
temperature 30°C. The mobile phase contains 5â•›m M ammonium acetate. (From Guo, Y. and
Gaiki, S., J. Chromatogr. A, 1074, 71, 2005. With permission from Elsevier.)
other hand, the stationary phase can also change its charge states at different pH,
thus introducing electrostatic interaction with the charged solutes. The amino phase
is positively charged in the pH range from 3.3 to 6.5, which can induce the electro-
static attraction with the negatively charged acid and influence the retention. When
acetylsalicylic acid becomes partially charged at pH 3.3, the electrostatic attraction
is significantly reduced, thus resulting in a big drop in retention time, as shown
in Figure 17.9. In comparison, cytosine has a very different behavior at different
mobile-phase pHs. The retention time of cytosine only varies slightly on the silica,
amide, and sulfobetaine phases in the pH range 3.3–6.8. On the amino phase, how-
ever, the retention time of cytosine decreases slightly from pH 6.5 to 4.8, and drops
significantly below pH 4.8. Cytosine has two pKa values: pK a 1 ∼ 4.6 and pK a 2 ∼ 12.2
in water. At the mobile-phase pH below pK a 1, cytosine becomes positively charged,

12
10
Retention time (min) 8
0
2 3 4 5 6 7
(A) Mobile phase pH
20
Cytosine
16
12
4
2 3 4 5 6 7
(B) Mobile phase pH
Figure 17.9 The effect of mobile-phase pH on the retention of (a) acetylsalicylic acid
and (b) cytosine on (♦) amino, (◾) amide, (▴) silica, and (×) sulfobetaine phase. Column
temperature 30°C. Mobile phase, acetonitrile/water (90/10, v/v) containing 10â•›m M
ammonium formate. Mobile-phase pH is the pH values of ammonium acetate solutions.
inducing electrostatic repulsion from the positively charged amino phase, and lead-
ing to reduced retention time below pH 4.8.
The electrostatic interactions between the charged solutes and stationary
phases can be modulated by adding buffer salts to the mobile phase. The salt
ions can reduce the electrostatic interaction between the charged solutes and
stationary phase. Therefore, the salt concentration in the mobile phase can have a
direct impact on the retention in HILIC. Figure 17.10 shows the effect of salt con-
centration on the capacity factor of salicyluric acid and cytosine on the amino,
amide, silica, and sulfobetaine phases. On the amino phase, the capacity factor
of salicyluric acid decreases drastically as the salt concentration increases from
5 to 40â•›m M. This is a direct result of reduced electrostatic attraction between

5
Salicyluric acid

3
0
0 10 20 30 40 50
(A) Salt concentration (mM)
4
Cytosine
1
0 5 10 15 20 25
(B) Salt concentration (mM)
Figure 17.10 The effect of salt concentration on the retention of salicyluric acid and
cytosine on (♦) amino, (◾) amide, (▴) silica, and (×) sulfobetaine phase. Column tempera-
ture 30°C. (a) The mobile phase for salicyluric acid is acetonitrile/water (80/20, v/v) contain-
ing 5–40â•›m M ammonium acetate. (b) The mobile phase for cytosine is acetonitrile/water
(85/15) containing 5–20â•›m M ammonium acetate.
the negatively charged acid and positively charged amino groups. In contrast,
a small, but significant increase in the retention of salicyluric acid is observed
on the silica, amide, and sulfobetaine phases as the salt concentration increases,
especially from 5 to 15â•›m M. This could be due to reduced electrostatic repulsion
between the negatively charged acid and the negative charges on the silica and
amide phases. Furthermore, the increase in retention with the salt concentration
on the sulfobetaine phase indicates that there is not any significant electrostatic
attraction between the negatively charged acid and positive quaternary amine
moiety. Additionally, a similar trend is also observed for cytosine on all the four
phases, as shown in Figure 17.10. The salt concentration seems to have more
effect on the amide and silica phases than on the amino and sulfobetaine phases.

The electrostatic interaction cannot explain the effect of salt concentration on

the retention for cytosine, which is not charged under the experimental condi-
tions. The increase in retention may be related to some changes of the stationary
phase induced by the salt ions. It is possible that higher salt concentrations might
increase the polarity of the stagnant liquid layer on the stationary phase; how-
ever, the exact mechanism for increased retention is not clear.
In addition to retention, selectivity can also be affected by the salt concentration.
Figure 17.11 shows the separation of the acidic model compounds on the silica phase
with 10 and 20â•›mM ammonium acetate in the mobile phase. As expected, the retention
of the acids increases at 20â•›m M salt concentration. More interestingly, there is also a
change in the selectivity for acetylsalicylic acid, salicyluric acid, and α-hippuric acid.
The three acids elute close to each other at 10â•›m M ammonium acetate, but are well
separated at 20â•›mM. This is another piece of evidence that salt concentration might
induce changes in the stationary phase, thus leading to selectivity change.
Column temperature is another parameter that can alter retention and selectivity
in RPLC, and can have similar effect in HILIC. Increasing temperature generally
1.80
10 mM ammonium acetate
1.60 1
1.40
1.20 2
1.00
5
AU
0.80 46
0.60 3
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(A) Minutes
1.80
1.60 20 mM ammonium acetate
1
1.40
1.20 2
1.00
AU
0.80 5
4 6
0.60 3
0.40
0.20
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

(B) Minutes
Figure 17.11 Separation of the acidic model compounds on the silica phase. Mobile phase,
acetonitrile/water (85/15, v/v) containing (a) 10 and (b) 20â•›mM ammonium acetate. Column
dimension, 250â•›mm × 4.6â•›mm ID, 5â•›μm particle size. Column temperature, 30°C. Flow rate,
1.0â•›mL/min. UV detection at 228â•›nm. Peak label: (1) salicylic acid, (2) gentisic acid, (3) acetyl-
salicylic acid, (4) salicyluric acid, (5) hippuric acid, and (6) α-hydroxyhippuric acid.

2.5
2
ln k΄
1.5
0.5
0
0.0028 0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035
(A) 1/T (K)
2.2
Cytosine
1.7
ln k΄
1.2
0.7
0.0028 0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035
(B) 1/T (K)
Figure 17.12 Effect of column temperature on the retention of (a) acetylsalicylic acid and
(b) cytosine on (♦) amino, (◾) amide, (▴) silica, and (×) sulfobetaine phase. Mobile phase,
acetonitrile/water (90/10, v/v) containing 10â•›m M ammonium acetate. (From Guo, Y. and
Gaiki, S., J. Chromtogra. A, 1074, 71, 2005. With permission of Elsevier.)
leads to reduced retention, and can often improve peak shape due to faster diffusion
at higher temperatures. Figure 17.12 shows the van’t Hoff plots for acetylsalicylic
acid and cytosine on the amino, amide, silica, and sulfobetaine phases. A general
trend of decreasing retention at higher temperature is observed on the amide, sul-
fobetaine and silica phases, and the amide phase seems to be more sensitive to the
temperature change than the other phases judging from the slope of the van’t Hoff
plots. On the amino phase, however, an increase in retention is observed for ace-
tylsalicylic acid with increasing temperatures. Similar retention increase at higher
temperatures has also been reported for basic compounds on the silica phase (Hao
et al. 2007). The amino phase is known to have significant electrostatic interactions
with the negatively charged acid at low salt concentrations. The non-classic van’t
Hoff behavior may be associated with the electrostatic interactions.
The retention behavior of the acidic compound at different temperatures has also
been found to change with the salt concentration in the mobile phase. Figure 17.13

0.6 Amide phase

0.5
0.4
0.3
ln k΄
0.2
0.1
0
–0.1
–0.2
0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035 0.0036
1/T
0.55 Aspartamide phase
0.5
0.45
0.4
0.35
ln k΄
0.3
0.25
0.2
0.15
0.1
0.0028 0.0030 0.0032 0.0034 0.0036
1/T
1.2 Amino phase
0.8
ln k΄
0.6
0.4
0.2
0
0.0029 0.003 0.0031 0.0032 0.0033 0.0034 0.0035 0.0036
1/T
Figure 17.13 Effect of salt concentration on van’t Hoff plots on amide, aspartamide, and
amino phase. Column temperature, 30°C. Mobile phase, acetonitrile/water (80/20, v/v)
containing (•) 40 (▴) 25, and (◾) 15â•›m M ammonium acetate.

shows the van’t Hoff plots of salicyluric acid on the amino, amide, and aspartamide
phases with 15, 25, and 40â•›m M ammonium acetate in the mobile phase. On the
amide phase, a linear van’t Hoff plot is observed for salicyluric acid at 15â•›m M, but
a small curvature in the van’t Hoff plots appears at 25 and 40â•›m M ammonium ace-
tate. In comparison, the curvature in the van’t Hoff plots is more pronounced on the
aspartamide phase even at 15â•›m M, but the curvature remains the same at higher salt
concentrations. The deviation of the van’t Hoff plots from linearity indicates that
the retention might be under the control of multiple forces in addition to hydrophilic
interaction. On the amino phase, the van’t Hoff plot for salicyluric acid also indicates
an increase in retention at higher temperatures similar to acetylsalicylic acid, but a
small curvature is observed at low temperatures even at 15â•›m M ammonium acetate.
At higher salt concentration (25 and 40â•›m M), the curvature in the van’t Hoff plots
is more pronounced. The van’t Hoff plot at 40â•›m M curves upward in the low tem-
perature range, indicating that the acid returns to “normal” temperature behavior
possibly due to sufficient suppression of the electrostatic interaction between the acid
and amino phase at high salt concentration.
17.6 Separation of Positional Isomers in HILIC

The separation of positional isomers is of great interest to analytical chemists
since many synthetic impurities and metabolites of drugs are positional isomers,
and is also very challenging due to the similarity of the positional isomers not
only in structure, but also in chromatographic properties. Reversed-phase and
normal-phase chromatography have been applied to the separation of position
isomers with certain success (Kazakevich and Lobrutto 2007). However, HILIC
has not been well established as a viable technique for the separation of posi-
tional isomers.
Caffeine metabolites include some mono-methylated, dimethylated xanthines,
and dimethylated uric acids. Caffeine and its metabolites (including 3-mono-
methylated xanthines, 3-dimethylated xanthines, and 3-dimethylated uric acids)
were separated on the amide and silica phases. As shown in Figure 17.14, an
interesting feature of HILIC separation is that the positional isomers eluted as
clusters are based on the degree of methylation, i.e., trimethylated xanthine (caf-
feine) eluting first, then dimethylated xanthines, followed by mono-methylated
xanthines. Dimethylated uric acids are strongly retained in HILIC and a gradient
was needed to elute them in a reasonable time. The amide phase shows nearly
baseline separation of all the positional isomers. However, only dimethylated uric
acid isomers were resolved on the silica phase. The isomers of mono- and dimeth-
ylated xanthines essentially co-eluted under the same conditions. Comparing to
reversed-phase separation, HILIC separation has totally different selectivity in
terms of eluting order, particularly for dimethylated uric acids (Safranow and
Machoy 2005).

0.070
Amide
0.060
1
0.050
9
0.040
10
23
AU
0.030
4 5 8
0.020 7
6
0.010
0.000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00
(A) Minutes
Silica
0.08 2, 3, 4
0.06 1
AU
0.04 5, 6, 7
9
8 10
0.02
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00
(B) Minutes
Figure 17.14 Separation of positional isomers of mono-methylated, dimethylated xan-

thines, and dimethylated uric acids on (a) amide and (b) silica phase. Column dimension,
150â•›mm × 4.6â•›mm ID, 3â•›μm particle size. Column temperature, 30°C. Flow rate, 1.2â•›m L/min.
Mobile phase, acetonitrile/water (90/10, v/v) containing 10â•›m M ammonium acetate. UV detec-
tion at 240â•›nm. Peak labels: (1) caffeine, (2) 1,3-dimethylxanthine, (3) 1,7-dimethylxanthine,
(4) 3,7-dimethylxanthine, (5) 1-methylxanthine, (6) 3-methylxanthine, (7) 7-methylxanthine,
(8) 1,3-dimethyluric acid, (9) 1,7-dimethyluric acid, (10) 3,7-dimethyluric acid.
17.7 Conclusions
More and more polar phases with diverse chemistry become available for HILIC
applications. Most polar phases have been shown to have very different selectivity
and retentivity for polar compounds. The cross-linked diol phase has been found to
have the weakest retentivity, and the amide and aspartamide phases have stronger
retentivity than other phases. Most polar phases show very different selectivity for
the polar compounds, but the diol, polyhydroxy, and PVA phases seem to have rather
similar selectivity. The retention and selectivity of some phases with ionizable func-
tional groups (e.g., the amino and silica phase) is highly subjective to electrostatic

interactions with charged solutes. It is important to point out that there has been no
correlation found between the structure of the functional groups and the retentivity
or selectivity. In addition, the retentivity and selectivity are also highly influenced
by the chromatographic parameters, such as organic solvent, mobile-phase pH, salt
concentration, and column temperature. The retention of the polar solutes is very
sensitive to the change in the content of organic solvent in the mobile phase, espe-
cially when the organic solvent content is above 85% (v/v). Mobile-phase pH can
influence the retention and selectivity of polar solutes through the ionization of both
the ionizable compounds and stationary phases. Comparing to reversed-phase sepa-
ration, the salt concentration seems to have a more significant effect on the retention
and selectivity in HILIC. In addition to attenuating the electrostatic interaction, the
salt concentration seems to be related to the retention of even neutral compounds.
However, the mechanism by which the salt exerts its effect is not fully understood.
The temperature study on the amide and aspartamide phases indicates that the reten-
tion on the polar phases might be under multiple mechanisms depending on the salt
concentration.
Acknowledgment
The author would like to recognize the contribution of Sheetal Gaiki, Emma Huang,
and other colleagues for many of the findings reported in this chapter.
References
Alpert, A.J. 1990. Hydrophilic-interaction chromatography for the separation of peptides,
nucleic acids and other polar compounds. J. Chromatogr., 499: 177–196.
Dejaegher, B. and Heyden, Y.V. 2010. HILIC methods in pharmaceutical analysis. J. Sep. Sci.,
33: 698–715.
Dejaegher, B., Mangelings, D., and Heyden, Y.V. 2008. Method development for HILIC
assays. J. Sep. Sci., 31: 1438–1448.
Guo, Y. and Gaiki, S. 2005. Retention behavior of small polar compounds on polar stationary
phases in hydrophilic interaction chromatography. J. Chromatogr. A, 1074: 71–80.
Guo, Y. and Huang, A.H. 2003. A HILIC method for the analysis of tromethamine as the
counter ion in an investigational pharmaceutical salt. J. Pharm. Biomed. Anal., 31:
1191–1201.
Guo, Y., Srinivasan, S., and Gaiki, S. 2007. Investigating the effect of chromatographic condi-
tions on retention of organic acids in hydrophilic interaction chromatography using a
design of experiment. Chromatographia, 66: 223–229.
Hao, Z., Lu, Y., Xiao, B., and Weng, N. 2007. Separation of amino acids, peptides and correspond-
ing Amadori compounds on a silica column at elevated temperature. J. Chromatogr. A,
1147: 165–171.
Hao, Z., Xiao, B., and Weng, N. 2008. Impact of column temperature and mobile phase com-
ponents on selectivity of hydrophilic interaction chromatography (HILIC). J. Sep. Sci.,
31: 1449–1464.
Hemstrom, P. and Irgum, K. 2006. Hydrophilic interaction chromatography. J. Sep. Sci., 29:
1784–1821.
Ikegami, T., Tomomatsu, K., Takubo, H., Horie, K., and Tanaka, N. 2008. Separation efficien-
cies in hydrophilic interaction chromatography. J. Chromatogr. A, 1184: 474–503.

Jandera, P. 2008. Stationary phases for hydrophilic interaction chromatography, their character-
ization and implementation into multidimensional chromatography concepts. J. Sep. Sci.,
31: 1421–1437.
Jian, W., Edom, R.W., Xu, Y., and Weng, N. 2010. Recent advances in application of hydrophilic
interaction chromatography for quantitative bioanalysis. J. Sep. Sci., 33: 681–697.
Jiang, W. 2003. Zwitterionic Separation Materials for Liquid Chromatography and Capillary
Electrophoresis, PhD Thesis, Umea, Sweden: Umea University.
Jiang, W., Fischer, G., Girmay, Y., and Irgum, K. 2006. Zwitterionic stationary phase with
covalently bonded phosphorylcholine type polymer grafts and its applicability to sepa-
ration of peptides in the hydrophilic interaction chromatography mode. J. Chromatogr.
A, 1127: 82–91.
Kazakevich, Y. and Lobrutto, R. Eds. 2007. HPLC for Pharmaceutical Scientists. Hoboken,
NJ: John Wiley & Sons, Inc.
Li, R.P. and Huang, J.X. 2004. Chromatographic behavior of epirubicin and its analogues on
high-purity silica in hydrophilic interaction chromatography. J. Chromatogr. A, 1041:
163–169.
Linden, J.C. and Lawhead, C.L. 1975. Liquid chromatography of saccharides. J. Chromatogr.,
105: 125–133.
McCalley, D.V. 2008. Evaluation of the properties of a superficially porous silica stationary
phase in hydrophilic interaction chromatography. J. Chromatogr. A, 1193: 85–91.
McCalley, D.V. and Neue, U.D. 2008. Estimation of the extent of the water-rich layer associ-
ated with the silica surface in hydrophilic interaction chromatography. J. Chromatogr.
A, 1192: 225–229.
Olsen, B.A. 2001. Hydrophilic interaction chromatography using amino and silica columns for
determination of polar pharmaceuticals and impurities. J. Chromatogr. A, 913: 113–122.
Palmer, J.K. 1975. A versatile system for sugar analysis via liquid chromatography. Anal.
Lett., 8: 215–224.
Safranow, K. and Machoy, Z. 2005. Simultaneous determination of 16 purine derivatives in
urinary calculi by gradient reversed-phase high-performance liquid chromatography
with UV detection. J. Chromatogr. B, 819: 229–235.
Snyder, L.R., Kirkland, J.J., and Glajch, J.L. 1997. Practical HPLC Method Development, 2nd
Ed. Hoboken, NJ: John Wiley & Sons, Inc.
Wen, N. 2003. Bioanalytical liquid chromatography tandem mass spectrometry methods on
underivatized silica column with aqueous/organic mobile phases. J. Chromatogr. B, 796:
209–224.
Expolore Luna HILIC, www.phenomenex.com, Phenomenex, 2007.

18 HILIC-MS/MS for the
Determination of Polar
Bioactive Substances:
Representative
Applications in the Fields
of Pharmacokinetics and
Metabolic Profiling
Yannis L. Loukas and Yannis Dotsikas
Contents
18.1 Introduction.................................................................................................. 428
18.2 Application of HILIC-MS/MS in the Field of Pharmacokinetics–
Bioequivalence Studies................................................................................. 428
18.3 Bioequivalence Study of Carvedilol............................................................. 429
18.4 Pharmacokinetics Studies of Donepezil, Loratadine, and Cetirizine.......... 429
18.5 Determination of Gabapentin in Human Plasma Using HILIC
with Tandem Mass Spectrometry................................................................. 434
18.6 Pharmacokinetic Study of the Peptide Drug Taspoglutide.......................... 434
18.7 HILIC-ESI-MS/MS Method for the Quantitation of Polar Metabolites
of Acrylamide in Human Urine.................................................................... 436
18.8 HILIC-MS/MS Technique for Sensitive Monitoring of the Changes
of Urinary Estrogen Conjugates................................................................... 437
18.9 Combination of HILIC-MS and RPLC-MS for Profiling Polar Urine
Metabolites................................................................................................... 438
18.10 Quantification of Methylmalonic Acid and Homocystein in Serum
and Urine with HILIC-MS........................................................................... 441
References............................................................................................................... 441
427

18.1 Introduction
HILIC or hydrophilic interaction liquid chromatography (originally called hydro-
philic interaction chromatography) is a liquid chromatographic technique for the
separation of polar and hydrophilic compounds. The expression “aqueous normal
phase” is another term that sometimes is used for this technique. Thus, in HILIC, we
should have a column with a hydrophilic stationary phase and a solvent (eluent) com-
posed of water, buffer, and a high concentration of a water-miscible organic solvent
like acetonitrile or methanol. Typically, in HILIC applications, the mobile phase
consists of acetonitrile at a concentration between 50% and 95% in an aqueous buf-
fer such as ammonium formate or ammonium acetate. These buffers are soluble in
organic solvents and are volatile, compatible with mass spectrometry (MS) or evap-
orative light scattering (ELSD) detectors that are used mostly in HILIC applications.
In HILIC, compounds elute in the reverse order to that of reversed-phase liquid
chromatography (RPLC). A compound that elutes first in an RPLC column should
have the highest retention in HILIC, and vice versa. Hydrophilic compounds are
problematic to separate in RPLC, such as, acids, bases, ions, sugars, and other
charged and neutral compounds, could have a better and easier separation in HILIC.
The applications of HILIC have rapidly increased in the last few years and have
gained increasing popularity, especially in the separation of small organic, polar,
and hydrophilic molecules in the area of pharmaceutics (pharmacokinetics, bio-
equivalence studies), metabolomics, biomarkers discovery, pesticides, etc.
Mass spectrometry is one of the most popular detectors of HPLC methods today.
Electrospray ionization (ESI) sensitivity in an HILIC-MS method could increase
dramatically (10–100 times) compared to that in an RPLC-MS method due to the
higher content of organic solvent used in the mobile phase, which lowers surface
tension, thereby simplifying droplet formation during the ESI process and signifi-
cantly improving the formation of ions in the gas phase. Moreover, when solid-phase
extraction (SPE) is used during sample preparation, the analytes that are eluted with
solvents compatible with HILIC columns would allow direct injections into MS
without any evaporation and reconstitution, simplifying the whole procedure.
18.2 Application of HILIC-MS/MS in the Field of

Pharmacokinetics–Bioequivalence Studies
Pharmaceutical drug discovery and development has been extremely benefited from
the use of high-performance liquid chromatography (HPLC) coupled with triple
quadrupole detectors (MS/MS). A search of the keywords LC-MS/MS in scien-
tific databases produces more than 20,000 records showing the applicability of this
technique. Most of these studies include two main parts: (1) sample preparation and
(2) the development and validation of the LC-MS method. HILIC offers advantages
to both parts. During sample preparation, the most commonly used procedures are
liquid–liquid extraction (LLE), solid-phase extraction (SPE), and protein precipita-
tion (PP). Despite the wide application of LLE in plasma sample preparation, this
procedure is hampered by time-consuming steps, along with decreased sensitivity in
some methods. The direct injection of LLE overcomes both drawbacks and offers an

HILIC-MS/MS for the Determination of Polar Bioactive Substances 429
alternative for plasma samples analysis. Moreover, a typical SPE protocol involves
labor-Â�intensive and time-consuming steps of column conditioning, sample loading,
washing, elution, evaporation, and final reconstitution. It has been estimated that
about half of the sample preparation time is consumed during sample evaporation
and reconstitution.1 However, the last two steps might be eliminated for compounds
with certain physicochemical properties by the replacement of the usually employed
RPLC methods with HILIC. As far as the PP procedure is concerned, successful com-
binations with HILIC-MS/MS have been previously reported for the quantification of
several drug candidates in biological matrices.2–4 However, all the reported methods
involved mixing plasma samples with the crash solvent (most often acetonitrile), fol-
lowed by the time-consuming steps of centrifugation, transfer of the supernatant to
new vials or plates, as well as in some cases, evaporation and reconstitution. The
applicability of an HILIC procedure could be summarized in the following examples.
18.3 Bioequivalence Study of Carvedilol

In the pharmacokinetic study of carvedilol, using cisapride as an internal standard,
HILIC-MS/MS with an Atlantis HILIC Silica Column was applied.5,6 The multiple
reaction monitoring (MRM) mode was employed for the quantification: m/z 407.2 →
99.9 for carvedilol and m/z 466.1â•›→â•›183.8 for cisapride as shown in Figure 18.1.
As the mobile phase was a mixture of acetonitrile–ammonium formate (50â•›m M,
pH 4.5) (90:10, v/v), the higher organic content resulted in sensitivity improvement
via the enhancement of ionization efficiency. Because of the higher sensitivity of the
HILIC–MS/MS method compared to that of RPLC-MS/MS, the plasma sample vol-
ume (50â•›μL) used in this study was smaller than that (200â•›μL) in RPLC-MS/MS6 with
the same lower limit of quantitation (LLOQ) (0.1â•›ng/mL). The described method was
successfully applied to the bioequivalence study of orally administered carvedilol
after a single oral dose of carvedilol (25â•›mg tablet).
18.4 P
harmacokinetics Studies of Donepezil,
Loratadine, and Cetirizine
Another example describes the behavior of three pharmaceutical molecules of
medium polarity—Donepezil (DNP), Loratadine (LOR), and Cetirizine (CTZ) as
shown in Figure 18.2, and the superior applicability of HILIC-MS/MS for quan-
titation during their bioequivalence studies.7 The MRM was performed at m/z
380.6 → 91.2 for DNP, 389.0 → 201.2 for CTZ, and 383.3 → 337.2 for LOR. DNP,
CTZ, and LOR solutions in acrylonitrile (ACN) were injected in triplicate onto C18
and silica YMC analytical columns under various mobile-phase compositions. In
particular, the mobile phase consisted of an ACN/formic acid 10â•›m M mixture in
ratios ranging from 95% ACN to 90% aqueous buffer (v/v). The ACN percentage
of the mobile-phase composition using either a C18 or a YMC HILIC silica column
resulted in the following results: at lower ACN concentrations (20%–40%), the C18
column revealed much higher retention than the silica column for CTZ and LOR,
while similar results were obtained for DNP as shown in Figure 18.3. The retention

99.9
3.5e5 221.9
m/z 100
3.0e5 O
O N
H
OH H3CO
2.5e5
N
Intensity
2.0e5 H
1.5e5
180.2 MH+
283.0 407.2
1.0e5
5.0e4
100 200 300 400 500

(a) m/z
1.9e6 183.8
1.8e6 MeO NH2

OMe
1.6e6 H
N
Cl
1.4e6 O N O
m/z 184
1.2e6
F
Intensity
1.0e6
8.0e5 MH+
466.1
6.0e5
233.8
4.0e5
2.0e5
434.1
96.1
100 200 300 400 500
(b) m/z
Figure 18.1 Product ion mass spectra of (a) carvedilol and (b) cisapride (IS). (From
Jeong, D.W. et al., J. Pharm. Biomed. Anal., 44, 547, 2007. With permission from Elsevier.)

O COOH O
N
N N
H3 C
H3C
Cetirizine Donepezil
Cl
N
N
O O
CH3
Loratadine
Figure 18.2 Chemical structures of donepezil, loratadine, and cetirizine. (From Apostolou, C.
et al., Biomed. Chromatogr., 22, 1393, 2008. With permission from John Wiley and Sons.)
12
C18
10 DNP
LOR
8 CTZ
t (min)
0
30 40 50 60 70 80 90 95
% ACN
7
Silica
6
DNP
LOR
5 CTZ
4
t (min)
0
20 30 40 50 60 70 80 90
% ACN
Figure 18.3 Retention time vs. ACN concentration in the mobile phase for C18 and silica
columns. (From Apostolou, C. et al., Biomed. Chromatogr., 22, 1393, 2008. With permission
from John Wiley and Sons.)

mechanism under these conditions favors the employment of the C18 column as
silica only presents a pseudo-reversed-phase profile.8 However, the lower sensitivity
and incompatibility with ESI-MS discourage the employment of such chromato-
graphic conditions. At medium ACN concentrations (50%–70%), silica presented
a slightly higher retention than C18 for all three compounds. A relatively lower
retention at the silica column under these conditions was due to the mixed modes
of separation and the complicated retention mechanisms. However, when the ACN
concentration was higher than 70%, all three compounds were retained much more
efficiently with silica than with the C18 column. The retention mechanism with
the silica column was mainly hydrophilic interactions, while C18 column retention
was mainly the interactions of the analytes with the remaining uncapped silanol
bases of the packing material. The retention times from C18 were similar to those
produced from silica with a medium ACN concentration, however, the increased
sensitivity favored the application of HILIC conditions of the LC-MS/MS analysis
for all three compounds. In particular, ACN concentrations higher than 80% in the
mobile phase were proven to be suitable for efficient retention under HILIC mecha-
nism. As a result, in all HILIC methods developed, the mobile phase consisted
of at least 80%–85% ACN, hydrophilic interactions being the main mechanism
governing the retention of all compounds. HILIC-MS/MS was superior compared
100 100
n-hexane LOR Toluene
% Intensity
% Intensity
DNP
LOR
DNP
0 0
0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0
(a) t (min) (b) t (min)
100
MTBE LOR
% Intensity
DNP
0
0.5 1.0 1.5 2.0
(c) t (min)
Figure 18.4 LLE organic solvent optimization for direct injection on the silica column.
(From Apostolou, C. et al., Biomed. Chromatogr., 22, 1393, 2008. With permission from
John Wiley and Sons.)

to RPLC during the LC-MS/MS development phase of the above three molecules.
Furthermore, the results of direct injections to MS/MS after LLE, SPE, and PP
are shown in Figures 18.4 through 18.7. As a result, HILIC is proven to be a highly
attractive alternative and complementary to the RPLC technique for pharmaceuti-
cals of medium to high polarity.
100 100
Silica LOR Amino
DNP
% Intensity
% Intensity
LOR
DNP
0 0
0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0
100
C18
% Intensity
DNP LOR
0
0.5 1.0 1.5 2.0
(c) t (min)
Figure 18.5 Column selection for direct injection of LLE with methyl butyl ether (MTBE)
as the organic solvent. (From Apostolou, C. et al., Biomed. Chromatogr., 22, 1393, 2008.
With permission from John Wiley and Sons.)
100 100
HILIC LOR RPLC LOR
% Intensity
% Intensity
CTZ
CTZ
0 0
0.5 1.0 1.5 2.0 0.6 1.2 1.8 2.4
Figure 18.6 HILIC and RPLC chromatograms comparison for CTZ determination
employing direct injection of SPE. (From Apostolou, C. et al., Biomed. Chromatogr., 22,
1393, 2008. With permission from John Wiley and Sons.)

100 100
DNP
DNP
HILIC CTZ
CTZ RPLC
% Intensity
% Intensity
0 0
0.5 1.0 1.5 2.0 0.8 1.6 2.4 3.2
Figure 18.7 HILIC and RPLC chromatograms comparison for CTZ determination
employing direct injection of PP or PPT. (From Apostolou, C. et al., Biomed. Chromatogr.,
22, 1393, 2008. With permission from John Wiley and Sons.)
18.5 Determination of Gabapentin in Human Plasma

Using HILIC with Tandem Mass Spectrometry
Gabapentin, 2-(aminomethyl-1-cyclohexyl)acetic acid, was originally developed as a
structural analogue of the inhibitory neurotransmitter–aminobutyric acid (GABA) to
reduce spinal reflex during the treatment of spasticity and was later found to have anti-
convulsant activity in various seizure models.9 Gabapentin is an amino acid and lacks
a chromophore that would allow trace level analysis by absorption spectrophotometry.
Moreover, HPLC with UV10 or fluorescence11 detection, capillary electrophoresis with
UV/fluorescence detection,12 and gas chromatography (GC) with flame ionization13
and mass spectrometry (MS) detection14 have been used, but these methods require
derivatization and protein precipitation. The described method presents the successful
application of HILIC-MS/MS for the determination of polar gabapentin using metformin
as an internal standard. The zwitterionic characteristic of gabapentin makes it extremely
difficult to be retained in RP-HPLC15 and to be extracted from biological samples. The
HILIC method with a silica column and a low-aqueous–high-organic mobile phase is
used for the retention of polar gabapentin and metformin. By increasing the content of
water, a stronger elution solvent in the HILIC mode reduced the retention times of gaba-
pentin and metformin on the silica column. The higher sensitivity of HILIC-MS/MS
(ca 20â•›pg on-column) was achieved using a smaller sample volume (i.e., 10 μL of human
plasma) in comparison to those (ca 200–2000â•›pg on-column) of RPLC-MS/MS with
100–200 μL of human plasma [15]. The described method was used successfully for a
bioequivalence study of orally administered gabapentin in healthy volunteers.
18.6 P
harmacokinetic Study of the Peptide
Drug Taspoglutide
Another alternative to the SPE is the online cleanup, where the solid-phase extrac-
tion is performed online by using two different types of columns operating simulta-
neously through a divert 10-port valve16,17 as shown in Figure 18.8. The first column
is the trapping column (TC) for cleaning the biological sample and retaining the

Configuration A Configuration B
Inlet Inlet
1 1
6 2 Waste 6 2 Waste
5 3 5 3
Sample Sample
4 loop 4 loop
Pump 1 Pump 1
Pump 2 Pump 2
1 1
10 2 10 2
9 3 9 3
Jumper Jumper
HLB 1
HLB 1
HLB 2
HLB 2
8 4 8 4
7 5 7 5
6 6
Waste Waste
Analytical
Analytical
Column
(a) Column
Pump 1 AS TC
T 4
Pump 2 5 3 Waste
6 2
L 1
MS AC Pump 3
(b)
Figure 18.8 Scheme of column switching LC-MS/MS system. (a) Two trapping columns
and one analytical column. (From Heinig, K. and Wirz, T., Anal. Chem., 81, 3705, 2009. With
permission from American Chemical Society.) (b) One trapping column and one analytical
column. (From Kousoulos, C. et al., Talanta, 30, 360, 2007. With permission from Elsevier.)

15N 15N
N N
13C
N 15N
13C
13C
13C
N 13C
N
R N 15N 13C
R
O O
Taspoglutide Internal standard
Figure 18.9 Structures of taspoglutide (drug) and internal standard (IS), R = His-Aib-
Glu-Gly-Thr-Phe-Thr-Ser-Asp-Val-Ser-Ser-Tyr-Leu-Glu-Gly-Gln-Ala-Ala-Lys-Glu-Phe-
Ile-Ala-Trp-Leu-Val-Lys-Aib. (From Kousoulos, C. et al., Talanta, 30, 360, 2007. With
analytes of interest, while the second one should be the analytical column (AC) for
the separation of analytes. During the cleaning phase of the procedure, the valve
directs the mobile phase to waste, usually with a high flow rate, while during the
analytical phase, the valve diverts the flow to the detector. A very impressive and
novel application of an HILIC column as a TC was used in the study of the pep-
tide drug, taspoglutide, MW 3339.7â•›g/mol, and [13C6,15N4-taspoglutide] (IS), MW
3347.7â•›g/mol, synthesized at F. Hoffmann-La Roche as shown in Figure 18.9. The TC
was a 50â•›mm × 2.1â•›mm Atlantis HILIC Si, 3â•›μm with precolumn filter (Waters). The
AC consisted of two Zorbax Poroshell 300-SB C18, 75â•›mm × 2â•›mm, 5â•›μm columns
(Agilent) in series. The HILIC TC was an excellent choice for online SPE followed
by RP chromatography because (1) large volumes of highly organic sample solu-
tions could be directly applied without prior dilution and (2) the polar analyte and
the IS were selectively enriched, while the lipophilic constituents were rinsed off.
According to the authors, the described method performed reliably with respectable
LLOQs of 10.0 and 50.0â•›pg/mL in human and animal plasma, respectively. The low
required human plasma volume of only 250â•›μL compared to the 1.5â•›m L plasma in the
previously employed method provided the possibility for repeat analyses or reduced
blood volumes taken from volunteers or patients.
18.7 HILIC-ESI-MS/MS Method for the Quantitation of

Polar Metabolites of Acrylamide in Human Urine
The carcinogen acrylamide (AA) is formed during food processing.18 AA is metabo-
lized to mercapturic acids, which are excreted with urine with the pathway described
in Figure 18.10. A HILIC-MS/MS method using a zwitterionic stationary phase
(ZIC-HILIC) was developed and validated to quantitate the mercapturic acids of AA
(AAMA) and glycidamide (GAMA), and AAMA-sulfoxide in human urine. In con-
trast to RPLC, the application of ZIC-HILIC resulted in the efficient retention and
separation of these highly polar compounds. The analysis was based again on online
cleanup with the use of a Stability BS-C17 AC (5â•›μm, 3â•›mm × 33â•›mm, Ammerbuch,

O O
CYP 2E1 O *
NH2 NH2
Acrylamide Glycidamide
1. Conjugation with glutathione

2. Degradation by transpeptidases
3. N-acetylation
O O H2N O
O
O O+ O * O
S NH2 S NH2 S NH2 S *
–
HN HN O HN OH HN OH
COOH COOH COOH COOH

AAMA AAMA-sulfoxide GAMA iso-GAMA
Figure 18.10 Biotransformation of acrylamide in the human body. (From Kopp, E.K. et al.,
J. Agric. Food Chem., 56, 9828, 2008. With permission from American Chemical Society.)
Germany) and the AC (ZIC-HILIC, 3.5â•›μm, 2.1â•›mm × 150â•›mm, SeQuant AB, Umeå,
Sweden) with a mobile phase consisting of a 14% ammonium acetate buffer and 86%
acetonitrile.
Urine samples were obtained from healthy human subjects (three female and
three male subjects, body weights between 52 and 75â•›kg, and ages between 23 and
28 years). Urine samples from the subjects were collected over a predefined time
frame of 72â•›h in intervals of 8â•›h. All subjects were nonsmokers and did not drink
alcoholic beverages 72â•›h before and during the study. Because of the baseline separa-
tion of the isobaric AAMA-sulfoxide and GAMA, which share a fragmentation of
m/z 249.2 to m/z 120.0, the interferences between these metabolites could be avoided
(Figure 18.11). Peaks representing AAMA-sulfoxide may easily be misinterpreted
as GAMA if the peaks were not clearly separated. An inefficient separation of
GAMA and AAMA-sulfoxide may explain some of the high GAMA concentrations
reported in human urine samples.19 Glycidamide, the precursor of GAMA, is the
DNA-reactive AA-metabolite supposedly responsible for tumor induction after AA
administration in rodents.20 Therefore, an overestimation of GAMA excretion due to
interference with AAMA-sulfoxide in human urine may result in an overestimation
of the potential risk of health effects due to AA exposures for humans.
18.8 HILIC-MS/MS Technique for Sensitive Monitoring

of the Changes of Urinary Estrogen Conjugates
Estrogens play important roles in the development and maintenance of secondary
sexual characteristics—pregnancy and long-bone maturation.21 They also have been
found to be associated with the development of breast cancer. Conjugation to sul-
fate and glucuronide is one of the major estrogen metabolism pathways, and these
conjugates are excreted mainly through urine. Free estrogens in urine usually occur

10.72 10.80
1.5e4 8000 2
AAMA: H3-AAMA (IS):
1.0e4 m/z 233 104 5000 m/z 236 104
5000
0 0
5 10 15 20 5 10 15 20
12.23
2000 12.15 8000 2
H3-GAMA (IS):
Intensity, cps
GAMA:
m/z 249 120 5000 m/z 252 120
1000
0 0
5 10 15 20 5 10 15 20
16.44 16.52
800 15.22
300 AAMA-sulfoxide: 15.15 2
H3-AAMA-
m/z 249 116 500 sulfoxide (IS):
200
m/z 252 119
100
0 0
5 10 15 20 5 10 15 20
Figure 18.11 HILIC-ESI-MS/MS in the multiple reaction monitoring mode (MRM).

Chromatogram showing the mass traces for the mercapturic acids of acrylamide excreted
in the urine of a nonexposed human subject. The corresponding mass traces for the internal
standards are shown on the right-hand side. (From Kopp, E.K. et al., J. Agric. Food Chem.,
56, 9828, 2008. With permission from American Chemical Society.)
at extremely low levels and are not detectable. Estrogen metabolism varies signifi-
cantly from person to person suggesting that individual profiles of estrogens and
their metabolites may provide information on variations in estrogen metabolism and
cancer risk. The accurate measurement of these estrogen conjugates may further the
study of the roles they play in estrogen-related physiological processes.
This study demonstrates the superiority of an HILIC-MS/MS method for the
determination of seven estrogens in human urine. Their structures appear in Figure
18.12. HILIC separation of the analytes was performed on a TSK-Gel Amide-80
(2.0â•›mm × 150â•›mm, 5â•›μm, 80â•›Å; Tosoh Bioscience, Montgomeryville, PA) with a
guard cartridge (2.0â•›mm × 10â•›mm) at room temperature. Isocratic elution was used
and the mobile phase was acetonitrile/aqueous ammonium acetate (5â•›m M, pH 6.80)
(85/15, v/v). The eluent from the LC column was directly transferred into the ion
source of the mass spectrometer without postcolumn splitting. The advantages of the
HILIC-MS/MS compared to that of RPLC-MS/MS, concerning the limits of quanti-
tation, appear in Table 18.1.
18.9 C
ombination of HILIC-MS and RPLC-MS
for Profiling Polar Urine Metabolites
Complex biological samples, such as urine, contain a very large number of endog-
enous metabolites reflecting the metabolic state of an organism (nicotine, cre-
atinine, tryptophane, lidocaine, β-hydroxyethyl theophylline, caffeine, salicylic

O O OH
HO3S O glucuronide O HO3S O

Estrone-3-sulfate (E1-3S) Estrone-3-glucuronide (E1-3G) Estradiol-3-sulfate (E2-3S)
OH OH OH
OH OH
glucuronide O HO3S O glucuronide O

Estradiol-3-glucuronide (E2-3G) Estriol-3-sulfate (E3-3S) Estriol-3-glucuronide (E3-3G)
H
OH HOOC
H O
O-glucuronide HO
glucuronide=
HO
HO H OH
Estriol-16-glucuronide (E3-16G) H H
Figure 18.12 Structures of the seven estrogen conjugates and of glucuronide. (From Qin, F.
et al., Anal. Chem., 80, 3404, 2008. With permission from American Chemical Society.)
Table 18.1
Comparison of the Limits of Quantification or Detection (LOQ or LOD)
Obtained Using the HILIC-MS/MS with Those Obtained Using the
RPLC-MS/MS Methods
HILIC-MS/MSa RPLC-MS/MSb RPLC-MS/MSc RPLC-MS/MSd
Analytes (ng/mL) (ng/mL)22 (ng/mL)23 (ng/mL)24
E1-3S 0.002 0.02 0.2 30
E2-3S 0.005 0.03 0.2
E3-3S 0.02 0.052 0.2 20
E1-3G 0.02 0.035
E2-3G 0.02 0.07
E3-16G 0.2 0.24
E3-3G 1.0 0.6
Source: Qin, F. et al., Anal. Chem., 80, 3404, 2008. With permission from American Chemical Society.
a LOQ is based on 1.0â•›mL of urine sample.
b LOQ is based on 5.0â•›mL of urine sample.
c LOQ is based on 0.1â•›mL of urine sample.
d LOD is based on 3.0â•›mL of urine sample.
acid, p-nitrophenol and cholic acid, hydroxyproline and uric acid).25 Metabolite
patterns can provide a comprehensive signature of the physiological state of an
organism as well as insights into specific biochemical processes. Although the
metabolites excreted in urine are commonly highly polar, the samples are gen-
erally analyzed using reversed-phase liquid chromatography mass spectrometry

(RP-LC/MS). In this published work, a method for detecting highly polar metabo-
lites by HILIC-ESI-MS is described as a complement to RP-LC/ESI-MS. SPE
was used for sample preparation. The adsorbent was activated and conditioned
first with 1â•›m L methanol and then with 1â•›m L ammonium acetate buffer (10â•›m M,
pH 4). An aliquot of 0.5â•›m L of rat urine was loaded onto the SPE column (30â•›mg,
Waters Oasis HLB). In the washing step, a 0.5â•›m L ammonium acetate buffer
(10â•›m M, pH 4) was used and 0.5â•›m L methanol was used for elution. Both the wash
fraction and the eluate were collected separately and filtered through syringe fil-
ters before injection.
For the HILIC analysis, a ZIC®-HILIC column (3.5â•›μm, 2.1â•›mm × 100â•›mm)
from SeQuant AB (Umeå, Sweden) was used together with a C4 precolumn (5â•›μm,
2.1â•›mm × 10â•›mm) from Thermo. Mobile phase A consisted of 5â•›m M ammonium
acetate (adjusted to pH 4 with formic acid), while mobile phase B consisted of ace-
tonitrile and 0.025% formic acid. The SPE sorbent is a hydrophilic–lipophilic based
(HLB) copolymer that is more suitable for polar compounds. It would be advanta-
geous to use the same stationary phase both for the sample preparation and for the
analytical column. (When the wash fraction was injected and RPLC was used, the
amount of nonretained compounds was higher compared to what was obtained after
analyzing the eluate.) By using these two complementary setups with a ZIC-HILIC
column for analysis of the wash fraction and a C18 column for the analysis of the
eluate, the number of metabolites to be detected could be increased. As shown in
Table 18.2, some of the substances examined retained on the RP C18 column and
others on the HILIC column.
Table 18.2
Retention Times for 11 Compounds Using HILIC and RP-HPLC
Retention Timea
Compound HILIC RP-HPLC
Hydroxyproline (C5H9NO3) 12.7 1.9
Uric acid (C5H4N4O3) 10.5 2.06
Nicotine (C10H14N2) 14.6 2.6
Creatinine (C4H7N3O) 11.5 2.2
Tryptophane (C11H12N2O2) 13.4 4.3
Lidocaine (C14H22N2O) 11.2 9.3
β-Hydroxyethyl theophylline (C9H12N4O3) 3.4 5.4
Caffeine (C8H10N4O2) 3.3 7.5
Salicylic acid (C7H6O3) 2.7 7.5
ρ-Nitrophenol (C6H5NO3) 3.2 11.7
Cholic acid (C24H40O5) 3.3 13.5
Source: Idborga, H. et al., J. Chromatogr. B, 828, 9, 2008. With permission from Elsevier.
a The void volume corresponding time (t ) was calculated to be 2.3â•›min for ZIC®-HILIC
0
and 2.1â•›min for RPLC that might differ slightly from experimental results.

18.10 Quantification of Methylmalonic Acid and

Homocystein in Serum and Urine with HILIC-MS
Homocysteine (Hcy) and methylmalonic acid (MMA) are the most specific markers
for inborn errors of methionine and propionic acid metabolism.26–28 The different
methods that have been developed for measuring Hcy and MMA in serum, plasma,
urine, and cerebrospinal fluid include GC-MS, LC-MS/MS, and capillary electro-
phoresis. The main obstacles to overcome are related to the low physiological con-
centrations of MMA in human serum and the fact that both markers are hydrophilic,
nonvolatile compounds. Finally, MMA is a structural isomer of succinic acid (SA)
that may cause interference. Concentrations of SA in the serum are usually consider-
ably higher than MMA. Current methods require extraction and derivatization steps
to yield MMA derivatives that are compatible either with GC-MS or RP-HPLC/MS
techniques.
Recently, all the above problems have been overcome by using HILIC-MS and
specifically by using the ZIC-HILIC column. A 50â•›mm, 4.6â•›mm, 5â•›μm, baseline sep-
aration was achieved within 3â•›min. The analysis was performed with the following
conditions: (1) column temperature: 30°C, (2) mobile phase: acetonitrile/ammonium
acetate (100â•›m M, pH 6.8); 75/25 (v/v), and (3) flow-rate: 1.0â•›m L/min with a split:
100â•›μL/min to MS. The ZIC®-HILIC column is indeed a suitable tool for separa-
tion of Hcy, MMA, and SA. Combined with MS detection, a physiological relevant
concentration can easily be quantified with the possibility of processing up to 20
samples per hour.
In conclusion, the combination of polar stationary phases and aqueous/organic
mobile phases could significantly enhance HILIC-MS/MS method sensitivity. The
low column backpressure due to high-organic mobile phases has made higher flow
rates feasible. Furthermore, a direct injection of the organic solvent extracts from the
application of LLE, SPE, and PPT onto the hydrophilic columns is compatible with
HILIC-MS/MS systems. All the above make HILIC a useful partner or an alterna-
tive to RP LC-MS/MS, especially for the analysis of polar compounds.
References
1. Zhou W, Zhou S, Pelzer M, Liu C, Jiang X, and Weng N. 2004. Impact of high flow rate
on sensitivity and resolution in quantitative bioanalytical LC-MS/MS using monolithic
and HILIC columns. Proceedings of the 52nd ASMS Conference on Mass Spectrometry
and Allied Topics, May 23–27, Nashville, Tennessee.
2. Li W, Li Y, Francisco DT, and Naidong W. 2005. Hydrophilic interaction liquid chro-
matographic tandem mass spectrometric determination of atenolol in human plasma.
Biomedical Chromatography 19: 385–393.
3. Shou WZ, Bu H, Addison T, and Jiang X. 2002. Development and validation of a liquid
chromatography/tandem mass spectrometry (LC/MS/MS) method for the determination
of ribavirin in human plasma and serum. Journal of Pharmaceutical and Biomedical
Analysis 29: 83–94.
4. Brown SD, White CA, and Bartlett MG. 2002. Hydrophilic interaction liquid chroma-
tography/electrospray mass spectrometry determination of acyclovir in pregnant rat
plasma and tissues. Rapid Communications in Mass Spectrometry 16: 1871–1876.

5. Jeong DW, Kima YH, Ji HY, Youn YS, Lee KC, and Lee HS. 2007. Analysis of carve-
dilol in human plasma using hydrophilic interaction liquid chromatography with tandem
mass spectrometry. Journal of Pharmaceutical and Biomedical Analysis 44: 547–552.
6. Borges NCC, Mendes GD, Silva DO, Rezende VM, Barrientos-Astigarraga RE, and
Nucci GD. 2005. Quantification of carvedilol in human plasma by high-performance
liquid chromatography coupled to electrospray tandem mass spectrometry: Application
to bioequivalence study. Journal of Chromatography B 822: 253–262.
7. Apostolou C, Kousoulos C, Dotsikas Y, and Loukas YL. 2008. Comparison of hydro-
philic interaction and reversed-phase liquid chromatography coupled with tandem
mass spectrometric detection for the determination of three pharmaceuticals in human
plasma. Biomedical Chromatography 22: 1393–1402.
8. Cox GB and Stout RW. 1987. Study of the retention mechanism for basic compounds on
silica under ‘pseudo-reversed-phase’ conditions. Journal of Chromatography 384: 315.
9. Ji HY, Jeong DW, Kim YH, Kim HH, Yoon YS, Lee KC, and Lee HS. 2006. Determination
of gabapentin in human plasma using HILIC-MS/MS. Rapid Communications in Mass
Spectrometry 20: 2127–2132.
10. Zhu Z and Neirinck L. 2002. High-performance liquid chromatographic method for
the determination of gabapentin in human plasma. Journal of Chromatography B 779:
307–312.
11. Chollet DF, Goumaz L, Juliano C, and Anderegg G. 2000. Fast isocratic high-perfor-
mance liquid chromatographic assay method for the simultaneous determination of gab-
apentin and vigabatrin in human serum. Journal of Chromatography B 746: 311–314.
12. Garcia LL, Shihabi ZK, and Oles K. 1995. Determination of gabapentin in serum by
capillary electrophoresis. Journal of Chromatography B 669: 157–162.
13. Hooper WD, Kavanagh MC, and Dickinson RG. 1990. Determination of gabapentin in
plasma and urine by capillary column gas chromatography. Journal of Chromatography
B 529: 167–174.
14. Gambelunghe C, Mariucci G, Tantucci M, and Ambrosini MV. 2005. Gas chroma-
tography–tandem mass spectrometry analysis of gabapentin in serum. Biomedical
Chromatography 19: 63–67.
15. Ifa DR, Falci M, Moraes ME, Bezerra FA, Moraes MO, and de Nucci G. 2001.
Gabapentin quantification in human plasma by high-performance liquid chromatogra-
phy coupled to electrospray tandem mass spectrometry. Application to bioequivalence
study. Journal of Mass Spectrometry 36: 188–194.
16. Heinig K and Wirz T. 2009. Determination of taspoglutide in human and animal plasma
using liquid chromatography–tandem mass spectrometry with orthogonal column-
switching. Analytical Chemistry 81: 3705–3713.
17. Kousoulos C, Dotsikas Y, and Loukas YL. 2007. Turbulent flow and ternary column-
switching on-line clean-up system for high-throughput quantification of risperidone and
its main metabolite in plasma by LC-MS/MS Application to a bioequivalence study.
Talanta 30: 360–367.
18. Kopp EK, Sieber M, Kellert M, and Dekant W. 2008. Rapid and sensitive HILIC-
ESI-MS/MS quantitation of polar metabolites of acrylamide in human urine using col-
umn switching with an online trap column. Journal of Agricultural and Food Chemistry
56: 9828–9834.
19. Bjellaas T, Janak K, Lundanes E, Kronberg L, and Becher G. 2005. Determination and
quantification of urinary metabolites after dietary exposure to acrylamide. Xenobiotica
35: 1003–1018
20. Manjanatha MG, Aidoo A, Shelton SD, Bishop ME, McDaniel LP, Lyn-Cook LE, and
Doerge DR. 2006. Genotoxicity of acrylamide and its metabolite glycidamide adminis-
tered in drinking water to male and female Big Blue mice. Environmental and Molecular
Mutagenesis 47: 6–17

21. Qin F, Zhao Y, Sawyer B, and Li X. 2008. Hydrophilic interaction liquid chromatogra-
phy–tandem mass spectrometry determination of estrogen conjugates in human urine.
Analytical Chemistry 80: 3404–3411.
22. D’Ascenzo G, Di Corcia A, Gentili A, Mancini R, Mastropasqua R, Nazzari M, and
Samperi R. 2003. Fate of natural estrogen conjugates in municipal sewage transport and
treatment facilities. The Science of the Total Environment 302: 199–209.
23. Zhang HW and Henion J. 1999. Quantitative and qualitative determination of estro-
gen sulfates in human urine by liquid chromatography/tandem mass spectrometry using
96-well technology. Analytical Chemistry 71: 3955–3964.
24. Volmer DA and Hui JPM. 1997. Rapid determination of corticosteroids in urine by com-
bined solid phase microextraction/liquid chromatography/mass spectrometry. Rapid
Communications in Mass Spectrometry 11: 1926–1933.
25. Idborga H, Zamania L, Edlunda P, Schuppe-Koistinenb I, and Jacobsson S. 2008.
Metabolic fingerprinting of rat urine by LC/MS: Part 1. Analysis by hydrophilic inter-
action liquid chromatography–electrospray ionization mass spectrometry. Journal of
Chromatography B 828: 9–13.
26. Lakso H, Appelblad P, and Schneede J. 2008. Quantification of methylmalonic acid in
human plasma with hydrophilic interaction liquid chromatography separation and mass
spectrometric detection. Clinical Chemistry 54: 2028–2035.
27. Rinaldo P, Hahn S, and Matern D. 2006. Inborn errors of amino acid, organic acid
and fatty acid metabolism. In: Tietz Textbook of Clinical Chemistry and Molecular
Diagnostics 5th Edition. Burtis CA, Ashwood ER, and Burns DE, Eds. Elsevier Sanders
Company, St. Louis, MO, pp. 2207–2247.
28. Watkins D and Rosenblatt DS. 2001. Cobalamin and inborn errors of cobalamin absorp-
tion and metabolism. The Endocrinologist 11: 98–104.

19 Method Development
and Analysis of Mono-
and Diphosphorylated
Nucleotides by HILIC
HPLC-ESI-MS
Samuel H. Yang, Hien P. Nguyen,
and Kevin A. Schug
Contents
19.1 Introduction................................................................................................... 445
19.2 Experimental Section.................................................................................... 447
19.2.1 Chemicals and Materials................................................................... 447
19.2.2 Instrumentation.................................................................................. 451
19.2.3 Physiochemical Parameters............................................................... 451
19.3 Results and Discussion.................................................................................. 452
19.3.1 Optimal pH........................................................................................ 452
19.3.2 Column Chemistries.......................................................................... 456
19.3.3 Gradient Method Development.......................................................... 456
19.3.4 Solubility Issues................................................................................. 457
19.3.5 Separation Results............................................................................. 457
19.4 Concluding Remarks.....................................................................................464
Acknowledgments...................................................................................................464
References...............................................................................................................465
19.1 Introduction
Reversed-phase high-performance liquid chromatography (RP-HPLC) has been
widely used as the standard for the high-throughput analysis of soluble polar com-
pounds. In RP-HPLC, separation of polar compounds is achieved giving an elution
order based on increasing hydrophobicity. Although a powerful technique, RP-HPLC
is not ideal for highly hydrophilic compounds as they are poorly retained and are
445

often eluted at the dead volume; this behavior subjects components to coelution
with other matrix components and limits quantitative analysis. Compounds that are
poorly retained in RP-HPLC can potentially be evaluated using normal-phase liquid
chromatography (NP-HPLC), where a nonpolar mobile phase and a polar stationary
phase are employed. However, poor solubility of hydrophilic compounds in nonpolar
solvents severely limits the application of NP-HPLC to these analytes. Also, when
coupling NP-HPLC to a detection technique such as electrospray ionization–mass
spectrometry (ESI-MS), additional problems arise. For instance, nonpolar solvents
cause a drastic loss in sensitivity due to their poor ionization efficiency1; some non-
polar solvents can also potentially damage plastic (e.g., PEEK) tubing commonly
used with HPLC-ESI-MS instruments.
Hydrophilic interaction chromatography (HILIC) is a mode of chromatography
that conveniently overcomes the limitations of both RP-HPLC and NP-HPLC for
analysis of hydrophilic compounds.2–5 HILIC can separate polar analytes while
using a mobile phase system ideally suited for ESI-MS, specifically a hydroorganic
solvent mixture with high polar organic solvent content.1 HILIC mobile phases dif-
fer from RP-HPLC in elution strength. Generally, HILIC uses a lower aqueous and
higher organic content in order to achieve retention of polar analytes on stationary
phases with polar interaction sites. In this mode, water plays the role of the stronger
eluting solvent, and thus, the lower aqueous content facilitates enhanced interaction
between the analyte and the polar stationary phase, and enables retention.6 With the
development of HILIC, a wide range of hydrophilic compounds, which had previ-
ously been difficult to separate, can now be rapidly analyzed by HPLC-ESI-MS with
sufficient capacity and resolving power.7–15
A large amount of interest currently focuses on the development of analytical
techniques to quantitate and characterize biomolecules.16–21 Nucleotides are a class
of biomolecules that hold an utmost importance in the living cell and play a critical
role in cell proliferation and metabolism regulation. For example, cyclic adenosine
monophosphate (cAMP) is a key regulator of many protein kinases, such as protein
kinase A (PKA), which performs signal transfer initiated by specific hormones in a
wide variety of cellular functions (e.g., in glycogen and lipid metabolism).22–26 Even
more so, single nucleotides act as a form of energy storage in the form of adenosine
triphosphate (ATP) where cleavage of the gamma phosphate gives adenosine diphos-
phate (ADP) and releases energy that drives enzymatic reactions.27–30 Nucleotides
can also act as regulators of enzymes based on their degree of phosphorylation, such
as in the case of small guanosine triphosphates (GTPases) in the G-protein signal-
ing pathway. Binding of these GTPases to guanosine diphosphate (GDP)/guanosine
triphosphate (GTP) causes inactivation/activation of another regulation mechanism
for the subsequent enzyme downstream in the signaling cascade.31–34
Despite such significance of nucleotides in cellular functions, a method involving
facile sample preparation with rapid separation and detection of phosphorylated nucle-
otides by HILIC-ESI-MS has had minimal progress when compared to other classes of
biomolecules. Separation of the nitrogen bases of nucleosides has been reported and is
tenable.35 The same success cannot be claimed, however, when considering the nucleo-
tide molecule consisting of all three of its major components, specifically, a nitro-
gen base, a 5-membered sugar ring, and an attached phosphate tail. Chromatographic

Method Development and Analysis of Nucleotides 447
methods generally involve the addition of a hydrophobic ion-pairing reagent into the
mobile phase that interacts with the nucleotides resulting in retention on a standard
RP column. One such common additive is N,N-dimethylhexylamine (DMHA).36–38
Although appreciable retention can be obtained from this method, it also has some
limitations. The overwhelming interference from the ion-pairing reagent in the mobile
phase causes a dramatic loss of sensitivity as well as a high background signal when
coupled to ESI-MS detection. The interaction of the ion-pairing reagent with the nucle-
otide, which can access a variety of protonation states, can be unpredictable leading to
sporadic results when sampling a wide range of nucleotides and nucleotide derivatives.
Additionally, the nature of these ion-pairing reagents causes heavy contamination of
the capillary lines in the HPLC-MS system as well as long column equilibration times.
Although some work has successfully applied HILIC for the quantification of a single
nucleotide,39 there is still a lack of a method to rapidly perform separations on a wide
range of nucleotides.
HILIC is a logical choice in building a method for the analysis of highly hydro-
philic nucleotides. Keeping the themes of rapid and reproducible analysis in mind, we
have focused on the development of a set of optimal conditions for the separation and
quantification of a mixture of mono- and diphosphorylated nucleotides. A comparative
study was also conducted between multiple types of commercial HILIC columns and
their ability to resolve various nucleotides within the mixture. The data allow insights
to be drawn into some of the interactions between the nucleotides and the various
HILIC stationary phases that are available to the scientific community.
19.2 Experimental Section

19.2.1 Chemicals and Materials
LC-MS grade water and acetonitrile were supplied by Burdick and Jackson
(Muskeegon, MI) and formic acid was obtained from J.T. Baker (Phillipsburg, NJ).
Ammonium formate was obtained from Acros Organics (Morris Plains, NJ). Six
of the nucleotides, adenosine 5′-diphosphate (ADP), guanosine 5′-monophosphate
(GMP), guanosine 5′-diphosphate (GDP), thymine 5′-monophosphate (TMP), cyto-
sine 5′-monophosphate (CMP), and uridine 5′-monophosphate (UMP), were pur-
chased from Sigma-Aldrich (Milwaukee, WI). The other two nucleotides, adenosine
5′-monophosphate (AMP) and uridine 5′-diphosphate (UMP), were purchased
from Fluka Analytical (Buchs, Switzerland). Table 19.1 shows a list of the eight
nucleotides, including their structures, monoisotopic masses, and calculated log P
values. Six different HILIC columns were tested—Phenomenex (Torrance, CA)
Luna HILIC (10â•›cm × 2.0â•›mm, 3â•›μm particle diameter (dp), 180â•›Å pore size); Tosoh
Bioscience (Kyoto, Japan) TSK-Gel Amide-80 (10â•›cm × 2.0â•›mm, 3â•›μm dp, 80â•›Å pore
size); Varian (Palo Alto, CA) Polaris 3 Amide (10â•›cm × 2.0â•›mm, 3â•›μm dp, 200â•›Å pore
size); Varian Pursuit XRs 3 Diol (10â•›cm × 2.0â•›mm, 3â•›μm dp, 100â•›Å pore size); Varian
Polaris 3 Amide-C18 (10â•›cm × 2.0â•›mm, 3â•›μm dp, 200â•›Å pore size); and Thermo-Fisher
Scientific, Inc. (West Palm Beach, FL) BETASIL Cyano (10â•›cm × 1â•›mm, 5â•›μm dp,
pore size not available). In-depth comparisons of column chemistries have been
reserved for the discussion section.

Table 19.1
List of Nucleotides Used in This Study
Compound
Number Compound Name Compound Structure Massa (amu) Log D (pH 7.0)b
1 Adenosine 5′-monophosphate (AMP) NH2 346.06 −4.28
N N
O N N
HO P O O
O– H H
H H
OH OH
2 Adenosine 5′-diphosphate (ADP) NH2 425.01 −7.71
N N
O O N N
HO P O P O O
O– O– H H
H H
OH OH
3 Guanosine 5′-monophosphate (GMP) O 362.05 −4.49
N NH
O N N NH2
HO P O O
O– H H
H H
OH OH

Method Development and Analysis of Nucleotides
4 Guanosine 5′-diphosphate (GDP) O 441.01 −7.92
N NH
O O N N NH2
HO P O P O O
O– O– H H
H H
OH OH
5 Uridine 5′-monophosphate (UMP) O 323.03 −5.67
NH
O N O
HO P O O
O– H H
H H
OH OH
6 Uridine 5′-diphosphate (UDP) O 401.99 −9.09
NH
O O N O
HO P O P O O
O– O– H H
H H
OH OH
(continued)
449
List of Nucleotides Used in This Study
Compound
Number Compound Name Compound Structure Massa (amu) Log D (pH 7.0)b
7 Thymine 5′-monophosphate (TMP) O 321.05 −5.67
NH
O N O
HO P O O
O– H H
H H
OH H
8 Cytidine 5′-monophosphate (CMP) NH2 322.04 −5.63
O N O
HO P O O
O– H H
H H
OH OH
a Monoisotopic.
b Log P calculated with ACD/labs log P calculator for compound in its neutral unionized form.

Samples were made by first preparing 10â•›mM stock solutions of each nucleotide in
water followed by dilution to make tested experimental concentrations at 50, 100, and
200â•›μM. Samples were premixed in a composition of water and ACN (30:70) prior to
injection onto the column in order to emulate starting mobile-phase compositions. All
samples were prepared at room temperature. Water/ACN compositions in excess of
80% organic content were not possible due to precipitation of the nucleotides at a high
organic content. A constant flow rate of 200â•›μL/min was used throughout all chro-
matographic separations for 2â•›mm I.D. columns and a flow rate of 50â•›μL/min was used
for the BETASIL cyano column with 1â•›mm I.D. Mobile phase A consisted of water
with 20â•›mM ammonium formate and 0.05% formic acid (v/v) (pH 3.98), while mobile
phase B was composed of a mixture of water/ACN (10:90) with 20â•›mM ammonium
formate and 0.05% formic acid (v/v). The best separation (on the Tosoh Amide-80
column) was achieved using a gradient program consisting of a 5â•›min 80% ACN iso-
cratic period in the beginning, then an acetonitrile gradient of 80% ACN to 50% ACN
in a 15â•›min period, and finally followed by a final 5â•›min isocratic period at 50% ACN.
Optimal gradient profiles were also obtained for the Phenomenex Luna HILIC and
amide-C18 columns in which only a single change was made where the gradient pro-
grams are 80% ACN to 50% ACN in 25â•›min, and 87.5% ACN to 50% ACN in 15â•›min,
respectively. Capacity factors (k′) were calculated based on the equation (tR – t0)/t0
where tR was the retention time of the sample analyte and t0 was the estimated dead
time. Injection volumes of 15â•›μL were made for all sample solutions.
19.2.2 Instrumentation
All measurements were performed on a Shimadzu LCMS 2010 (Shimadzu Scientific
Instruments, Inc., Columbia, MD), which included a dual high-pressure-mixing
LC-20AD pump system and an SIL-20A HT autosampler coupled to a quadrupole
mass analyzer, equipped with a conventional ESI source. Electrospray ionization
was performed in both the positive and the negative modes with a spray capillary
voltage of 4.5 and −3.5â•›kV, respectively. Scan speeds used over the entire runtime
were set at 250 amu/s at a range of 320–450 (m/z). A positive-/negative-ionization
mode switching time interval of 1â•›Hz was used. Other ion source parameters are as
follows: nebulizing gas, 1.2â•›L/min; drying gas flow rate, 6000â•›k PA; curved desolva-
tion line temperature, 250°C; and CDL heat block temperature, 250°C. The detector
voltage was set to 1.60â•›kV. Data analysis was performed using LCM Solutions (ver-
sion 3.4) software.
19.2.3 Physiochemical Parameters

Calculations of octanol–water partition coefficients (log P), pH-dependent octanol–
water partition coefficients (log D), and pKa values were performed for each nucleo-
tide using Advanced Chemistry Development, Inc. (ACD/Labs) log P, log D, and pKa
calculators (version 9), respectively. Values presented for log P (Table 19.1) and log D
(Table 19.2) are expected to contain approximately 10% relative uncertainty. The
calculated uncertainties in pKa values range from 0.1 to 0.5â•›pKa units according to
the output of the program.

Table 19.2
Comparison of Log D Values, Charge States (z), and % Species of Charge
State z Present at pHs 4.7, 7.0, and 10.0
pH 4.7 pH 7.0 pH 10.0
Log D z % Species Log D z % Species Log D z % Species
AMP −3.0 −1 89.5 −4.3 −2 81.6 −4.7 −2 99.8
ADP −7.2 −1 90.8 −7.7 −2 56.7 −8.4 −3 99.8
GMP −3.2 −1 97.5 −4.5 −2 81.2 −5.6 −2 85.8
GDP −7.7 −2 99.0 −7.9 −3 55.9 −9.3 −4 86.0
UMP 4.4 −1 97.7 −5.7 −2 78.3 −7.0 −2 98.0
UDP −8.8 −2 99.2 −9.1 −3 54.5 −10.7 −4 98.1
TMP −4.4 −1 97.7 −5.7 −2 81.7 −6.7 −2 85.0
CMP −4.3 −1 94.2 −5.6 −2 81.6 −6.1 −2 99.0

Sample preparation followed the same protocol used in method development. For
the establishment of calibration curves, 10 different concentrations of nucleotide
mixtures were tested in triplicate with a range of 0–350â•›μM. Multiple curve fitting
approaches were investigated to find the best fit to the experimental data. A set of
seven replicate measurements at 50â•›μM was used for the calculation of the limit of
detection for each nucleotide.

19.3.1 Optimal pH
Method development for the separation of nucleotides with HILIC begins with the
determination of an optimal pH for the system. Manipulation of the hydrogen ion
activity directly affects analyte selectivity and retention; significant changes in
these chromatographic characteristics can result with only small changes in pH.
For HILIC HPLC-ESI-MS of nucleotides, control over the pH of the system holds
even greater importance because of the existence of different protonation states of
each nucleotide at different pHs. Depending on the type of nucleotide, whether it is
(a) purine- or pyrimidine-based, or (b) mono-, di-, or triphosphate, the pH causes
populations of different protonation states in varying percentages. Knowing that this
phenomenon occurs, it becomes crucial to find an optimal pH that isolates all nucleo-
tides in the mixture with a dominant protonation state. The presence of a mixture of
different protonation states would obscure chromatographic peaks for each nucleo-
tide possibly leading to severe peak broadening, multiple and inconsistent retention
times, less-than-optimal ionization efficiency, or other unpredictable results. Thus,
an appreciation of the pKa and log D values for each nucleotide can be helpful when
selecting a suitable pH that accommodates all the nucleotides.
A representation of the calculations obtained for the nucleotide, UMP (5), is
shown in Figure 19.1A, where the percent compositions of protonation states are

(a) O
a c NH
100 O
Log D = –4.1
HO P O N O
b O z = –1
O–
80 OHOH
Percent composition
(b) O
NH
60 O
O– P O N O Log D = –5.8
O z = –2
O–
40 OHOH
(c) O
20 N–
O
O– P O N O Log D = –7.0
O z = –3
O–
0
0 2 4 6 8 10 12 14 OHOH
(A) pH
(a) not pictured
Log D = –5.0, z = –2
b d O
100 (b) N NH
O O
N N NH2
O– P O P O O
O– O–
80
c OH OH
Percent composition
Log D = –8.7, z = –3
a
60 (c)
O
N N
–
O O
N N NH2
O– P O P O O
40 –
O O
–
OH OH
Log D = –9.3, z = –4
20 O
(d)
N N–
O O
– N N NH2
OPOPO
0 O– O –
O
0 2 4 6 8 10 12 14
OH O–
(B) pH Log D = –10.8, z = –5
Figure 19.1 Percent composition of predicted protonation states for (A) UDP (5) and (B)
UDP (6) as a function of pH. z denotes charge (protonation) state.
plotted against the pH. From these data, it is clear that two pH settings are the most
feasible to be experimentally tested; specifically, pH 4.1 and pH 11.1 are predicted
to place over 90% of the species population in a single protonation state (−1 and −3,
respectively). In comparison, for UDP (6) (Figure 19.1B), the most optimal pHs for
separation are predicted to be 4.5 and 10.6 (z = −3 and −5, respectively). The different
nucleotides each have their own optimal pH values based on similar analyses. After
a comparison between all eight nucleotides, it was concluded that the most suitable
pHs that satisfy the optimal ranges for all nucleotides, and where all exist predomi-
nantly in a single ion form, were 4.7 and 10.0, as shown in Table 19.2. An operating
pH of 10.0 was ruled out due to the increase in the solubility of the silica-based
stationary phases at this high pH.1 Thus, acidic mobile phases were investigated to
optimize the separations.
It is important to note that control (and measurement) of the pH during HILIC
separations is not straightforward. While it is relatively simple to accurately set the
pH of the aqueous mobile phase, to perform the same pH control on the organic

mobile phase is impractical with a standard pH meter. The large presence of acetoni-
trile, a weak base with a low dielectric constant, in the organic mobile phase reduces
charge separation and the ionizability of compounds.40,41 This phenomenon results
in a pH shift from the desired pH of 4.7 to higher values, when a large proportion
of acetonitrile is present. Adding more acid to the organic mobile phase until the
pH reaches 4.7 is not a viable option as it causes more complications (i.e., a variable
ionic strength) when using a gradient profile in the separation scheme, and it cannot
be reliably measured. Due to the change in organic composition during the gradient
chromatographic elutions, the exact pH cannot be determined during the run. Still,
it is reasonable that most acidic species will be affected in a similar fashion when
the dielectric constant of the medium is changed. Percentages of formic acid (0.01%,
0.05%, 0.1%, and 0.5%) were varied to find the optimal pH (data not included) at
which the best separation and the strongest ESI-MS signal intensities were recorded.
The addition of 0.05% (v/v) formic acid with 20â•›m M ammonium formate buffer solu-
tion, equating to a pH of 3.96 in the aqueous mobile-phase reservoir, was found to be
optimal; an equal amount of acid was added to the organic mobile-phase mixture to
ensure consistency of the acid content throughout the experimental testing.
While operating under acidic conditions, logic would normally dictate for the
analyte to be preferentially detected in the positive-ionization mode. This gener-
ally accepted idea is a misconception, as it is not always the case, and the HILIC
separation of nucleotides represents one such exception. According to Boyd, the
ESI-MS analysis of amino acids can follow a “wrong-way-round” electrospray ion-
ization pattern in that the amino acids can still be detected in the negative-ionization
mode despite working under acidic conditions.42 Conversely, amino acid detection
can also be achieved in the positive-ionization mode while operating under basic
conditions.42,43 In this study, the detection of nucleotides under acidic conditions was
generally most sensitive with the positive-ionization mode. However, the negative-
ionization mode shows appreciable signal intensity for the nucleotides as well. For
certain nucleotides, the signal intensity in the negative-ionization mode equaled or
even surpassed that of the positive-ionization mode. Figure 19.2 shows extracted ion
chromatograms of GMP (3) and UDP (6), retained on the amide column, along with
their respective mass spectra in both the positive- and negative-ionization modes.
Both positive-mode spectra exhibit [M + H]+ signals, with matching [M – H]− sig-
nals in the negative mode; the differences lie within their signal strengths. While the
GMP [M + H]+ signal was nearly an order of magnitude greater than its correspond-
ing [M – H]− signal, the UDP [M + H]+ and [M – H]− signals were within the same
order of magnitude.
The rest of the nucleotides that were analyzed also exhibited either of the two
patterns between the positive- and negative-ionization modes. Interestingly, UMP
(5), UDP (6), and TMP (7) are the three nucleotides that follow the latter trend that
were mentioned, previously. An explanation for such behavior can be found in the
structure of their nearly identical nucleoside bases, which are unique from other
nucleoside bases in that they contain two carbonyl groups attached to the pyrimi-
dine ring as opposed to incorporation of at least one amine group onto the pyrimidine
ring. The presence of an amine group facilitates sensitive detection in the positive-
ionization mode. It is also important to note the presence of an [UDP + NH4]+ adduct

Method Development and Analysis of Nucleotides
1.5
6.0
EIC, GMP
Ion intensity/105
EIC, UDP
Ion intensity/105
1.0 362.00 m/z 403.00 m/z
4.0
0.5 2.0
0.0 0.0
0.0 5.0 10.0 15.0 20.0 0.0 5.0 10.0 15.0 20.0
362.00 403.00
2.0
Ion intensity/104
Ion intensity/104
6.0
[GMP – H]– 1.5 [UDP – H]–
4.0
322.05 1.0
2.0 346.05 0.5
0.0 0.0
320 340 360 380 400 420 440 m/z 320 340 360 380 400 420 440 m/z
364.00
2.0 5.0 422.00
Ion intensity/105
Ion intensity/104
[GMP + H]+
1.5 4.0
3.0
1.0 324.00 [UDP + H]+
2.0 404.95
0.5 348.10 1.0
0.0 0.0
(A) 320 340 360 380 400 420 440 m/z (B) 320 340 360 380 400 420 440 m/z
Figure 19.2 (A) Extracted ion chromatogram of GMP (3) on Tosoh TSK-Gel Amide-80 column with gradient separation. Mass spectra of the peak
in the negative- and positive-ionization modes show signals for [GMP − H]− and [GMP + H]+, respectively. (B) Extracted ion chromatogram of UDP
(6) under the same conditions. Mass spectra of the peak in the negative- and positive-ionization modes show signals for [UDP − H]− and [UDP + H]+,
455
respectively.

in the mass spectra, which is unique to UDP as it is not seen on any other of the
analyzed nucleotides. The presence of this ammonium adduct must be taken into
consideration as it represents a significant fraction of the UDP ions and will affect
the calculated sensitivity values in method validation. Therefore, due to the appar-
ent viability of both the positive- and negative-ionization modes in the detection
of the nucleotides, both modes were scanned during each chromatographic run. In
addition, both modes were considered and individually analyzed in further method
validation calculations.
19.3.2 Column Chemistries

A total of five types of HILIC columns were tested in developing a method for the
separation of nucleotides. The columns were obtained from four different manu-
facturers allowing a comparison between some of the different types of columns
that are currently commercially available. The types of stationary phases include
polar functional groups such as amide- (TSK-Gel Amide and Polaris Amide-C18)
or dihydroxy-bonded (diol) variants (Luna HILIC and XRs Pursuit 3 Diol). Polaris
Aminopropyl and BETASIL Cyano columns were also studied. Theoretically,
the amide- and amino-functionalized stationary phases should be the most suited
to perform separations on acidic analytes. Hydrogen bonding and electrostatic
forces increase the interaction between the amide/amino groups with the nega-
tively charged phosphate of the nucleotides yielding significant retention. The two
bridged diol columns and the cyano column can also interact with the negatively
charged analytes through hydrogen bonding. The Polaris Amide-C18 has a polar
embedded amide group underneath a C18 chain that could allow for both hydrogen
bonding and hydrophobic interactions to contribute to analyte retention.
19.3.3 Gradient Method Development

Multiple chromatographic parameters were optimized leading to the best separa-
tions from each type of column. Although buffer pH has already been considered
as previously mentioned, mobile phase compositions and gradients as well as ionic
strengths were extensively tested. Isocratic experiments were first performed on each
of the columns at different mobile-phase compositions in order to give a rough esti-
mate of retention times and selectivity on each of the columns. From these data, the
optimal gradient profiles were designed. A range of different buffer ionic strengths
(10, 20, 30, 40, and 100â•›m M ammonium formate) were tested (data not included) to
find the optimal buffer concentration, which was determined to be 20â•›m M ammo-
nium formate. Other buffer systems were also tested (i.e., ammonium carbonate,
ammonium bicarbonate, and ammonium acetate), but did not provide acceptable
performance (data not included).
The best separations, which were most effective on the Tosoh TSK-Gel Amide-
80 column, were achieved using a gradient program consisting of a 5â•›min isocratic
period in the beginning, followed by a gradient of 80% ACN to 50% ACN in 15â•›min,
and then a 5â•›min isocratic period at 50% ACN. Separation was achieved on the
Phenomenex Luna HILIC using a similar gradient program as that of the Tosoh

amide, except with a gradient of 80% ACN to 50% ACN in 25â•›min. Other gradients
profiles were tested with slightly different modifications to the gradient for opti-
mization of retention on other columns. For the amide-C18 column, the optimal
conditions for retention required a gradient profile that started at 87.5% ACN to 50%
ACN in 15â•›min following the same program as before with the 5â•›min isocratic peri-
ods before and after the gradient. The other three columns (aminopropyl, diol, and
cyano) were also tested with different gradient profiles containing different varia-
tions, but the retention and separation achieved from these columns were not as suc-
cessful as those for the first three columns described earlier.
19.3.4 Solubility Issues

Ideally, an analyte sample should be premixed in the initial mobile-phase com-
position of the separation to reduce the equilibration of the sample upon injection
onto the column. Preparation of the nucleotide samples in the initial mobile-phase
compositions (such as 20:80 water/ACN and 12.5:87.5 water/ACN) was not pos-
sible due to the precipitation of the nucleotides at a high organic content. Being
negatively charged and thus highly polar, the nucleotides have low solubility in
organic solvents such as acetonitrile. The precipitation of solids not only obscures
the chromatographic results, but also causes clogging of the electrospray needle and
leads to exceedingly high back pressures, which requires immediate maintenance
to resolve. As a result, a composition of 30:70 water/ACN was determined to be the
highest amount of organic solvent that can be premixed with nucleotides without
causing precipitation. Some higher contents of acetonitrile would initially dissolve
the nucleotides (such as a 25:75 water/ACN composition), but precipitation would
still occur after the passage of time within the autosampler vial as well as after the
elution of the nucleotides from the column, again causing the electrospray needle to
become blocked.
19.3.5 Separation Results

Figure 19.3 shows the extracted ion chromatograms of each nucleotide (1–8) obtained
following separation with the Tosoh Amide-80 column. Separation was observed
for all eight of the nucleotides with distinct retention times except for compounds
1 and 5. The purine diphosphates 2 and 4 exhibit broadened peaks and low signal
intensities, a recurring theme in this investigation. A further discussion on these
results with the purine diphosphates is given in the following text. The extracted
ion chromatogram for CMP (8) at m/z 322 contains two different peaks at different
retention times due to the proximity of its molecular mass to that of TMP (7). The
first peak corresponds to the +1 isotope of TMP (7) whose molecular mass is exactly
one proton less than CMP (8), whereas the second peak corresponds to CMP (8). All
assigned signals for the respective extracted ion chromatograms were assigned by
mass as well as by analysis of single component mixtures.
Extracted ion chromatograms of each nucleotide (1–8) collected from separa-
tion on the Phenomenex Luna HILIC column are shown in Figure 19.4. The Luna
HILIC column also provided appreciable retention and selectivity for many of the

EIC m/z
AMP
346.0
ADP
426.0
GMP
352.0
442.0 GDP
UMP
323.0
UDP
403.0
TMP
321.0
CMP
322.0
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5
Figure 19.3 Extracted ion chromatograms of AMP (1), ADP (2), GMP (3), GDP (4), UMP
(5), UDP (6), TMP (7), CMP (8) obtained by separation on a Tosoh TSK-Gel Amide-80 col-
umn with a gradient program.
EIC m/z AMP
346.0
ADP
426.0
GMP
362.0
GDP
442.0
UMP
323.0
UDP
403.0
TMP
321.0
CMP
322.0
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5
Figure 19.4 Extracted ion chromatograms of AMP (1), ADP (2), GMP (3), GDP (4), UMP
(5), UDP (6), TMP (7), CMP (8) obtained by separation on a Phenomenex Luna HILIC col-
umn with a gradient program.

compounds, comparable to that of the Tosoh Amide-80 column. The peaks do how-
ever, suffer in shape and quality exhibiting a high degree of peak broadening and
peak tailing in all extracted ion chromatograms. Asymmetrical peaks are typically
indicative of multiple interaction modes between the stationary phase and the ana-
lyte.44–46 The same trend with a diminished sensitivity of the purine diphosphates
(2, 4) is also observed as well as the same dual peaks are seen from the chromato-
gram of CMP (8), which can again be attributed to the +1 isotope of TMP (7), which
has the same mass-to-charge ratio as CMP (8).
The Varian Polaris 3 Amide-C18 column also provided some separation of the
nucleotides, although it was significantly less pronounced than that for the Tosoh
amide or the Phenomenex Luna HILIC. The majority of the monophosphates (1, 5, 7,
and 8) eluted near the dead volume peak giving capacity factors less than 1. However,
the diphosphates (2, 4, and 6) and GMP (3) were minimally retained by the unique
stationary phase yielding capacity factors within a range of 1.0–1.2. Whereas the other
two columns were able to successfully resolve the nucleotides, the amide-C18 was
only able to separate the nucleotides into two major peaks with overall lower retention
compared to the Tosoh Amide-80 and Phenomenex Luna HILIC columns. Once again,
the purine diphosphates (2 and 4) showed a loss in sensitivity with broadened peaks.
Table 19.3 displays calculated capacity factors (k′) of each nucleotide on each of
the three columns that exhibited retention. From the results of this investigation, it
is clear that the Tosoh amide was the most effective for the analysis of the mixture
of nucleotides and gave the best retention, resolution, and quality of peak shapes out
of the three columns. The Varian XRs 3 Diol and the Thermo BETASIL Cyano col-
umns did not provide appreciable retention of the nucleotide analytes, and thus their
results were not reported in the comparison. The results from the Varian aminopro-
pyl column were unique in that elution of the nucleotides was not observed, which
leads to the speculation that they may still be retained on the column.
An explanation of the results can be made based on the different column stationary
phases and their ability to separate the nucleotide mixture. As mentioned previously,
Table 19.3
Comparison of Capacity Factors (k′) of Each Nucleotide with Different
HILIC Columns
AMP (1) ADP (2) GMP (3) GDP (4) UMP (5) UDP (6) TMP (7) CMP (8)
Tosoh TSK-Gel 12.5 16.0 13.7 16.9 12.4 15.4 10.8 14.4
Amide-80
(t0 = 1.0â•›min)
Phenomenex 10.4 14.3 11.3 15.0 7.3 12.7 4.4 11.8
Luna HILIC
(t0 = 1.25â•›min)
Varian Polaris 3 0.5 1.2 1.1 N.D. 0.3 1.1 0.3 0.4
Amide-C18
(t0 = 1.23â•›min)

the amide-and amino-functionalized stationary phases should theoretically perform

the best in the separation of these negatively charged analytes. This conjecture has
been verified by the results obtained with the Tosoh Amide-80 column as it indeed
performed the most effective separation. One obvious question that arises is why the
Varian aminopropyl column did not have as much success as the Tosoh amide col-
umn. The gradient elutions of the aminopropyl column, as well as the washings (data
not shown), did not contain ion signals corresponding to the nucleotide masses and
thus indicated that the nucleotides may not have been eluted off the column under the
typical HILIC mobile-phase conditions used in this study. Perhaps the acidic mobile-
phase conditions caused strong binding interactions between the amino group and
the nucleotide phosphate groups and thus prevented the elution of the nucleotides.
Retention of nucleotides on the Polaris aminopropyl column has been reported,39
but the basic mobile-phase conditions (pH 9.45) approach the solubility limit of the
silica-based stationary phase and may cause damage to the column over prolonged
usage.1 Interestingly, the Varian Polaris 3 amide-C18 column was still capable of
achieving minimal retention of the nucleotides despite the presence of the large,
hydrophobic C18 chain attached to the amide nitrogen, which would intuitively hin-
der the hydrophilic interaction between the amide group and the nucleotides. The
Thermo cyano and Varian diol columns did not provide appreciable retention of the
nucleotides under the experimental conditions investigated. Both types of columns
have also been reported to be well suited for the separation of basic compounds,47 so
the findings for these columns are not surprising, except to note that the Luna HILIC
(also, a diol stationary phase) did indeed provide adequate retention. Without an in-
depth knowledge of the exact stationary-phase chemistry, no further explanation of
this disparity can currently be made.
Throughout the analysis, there was an obvious distinction between the monophos-
phates and the diphosphates. The monophosphates eluted before the diphosphates
on a consistent basis, which was expected. The presence of two phosphate groups
increases the hydrophilicity of the diphosphate nucleotides relative to the monophos-
phates, and increases retention. The calculated values for log D (see Figure 19.1 and
Table 19.2) support this assertion. Within the subgroups of mono- and diphosphates,
the nucleotides containing purine bases (1–4) were retained more than the pyrimidine
nucleotides (5–8). The N-7 nitrogen in the imidazole ring of purine bases can act as an
additional polar group accentuating the interaction between the polar stationary phase
and the nucleotide. The imidazole nitrogen is also suspect in causing the pattern of
low sensitivity and broadened peaks seen throughout all analyses. Protonation on this
imidazole nitrogen is possible and could provide a positive charge that neutralizes the
negative charge from the phosphate group, leading to poor ionization efficiency.48,49
This deleterious effect is even more pronounced in the case of guanine-based com-
pounds 2 and 4, because of an additional protonation site on N-1 nitrogen, which is
not present in adenosine nucleotides 1 and 3. Although the purine nucleotides consis-
tently had lower signal intensities than the pyrimidine nucleotides, the diphosphate
purines suffered a much greater loss in sensitivity and quality of peak shape than their
monophosphate counterparts due to the adverse effects caused by both the second
phosphate and the purine base. The same negative effects were not seen in the pyrimi-
dine diphosphate, UDP (6); it exhibited both sharp peaks and high ion-signal intensity.

Among the pyrimidines, TMP (7) had the least retention because of the presence
of a methyl group on the C-5 carbon of the nucleoside base as well as the absence of a
hydroxyl group at the C-2 carbon on the ribose sugar ring. TMP (7) is the only tested
nucleotide with a deoxyribose sugar because it naturally exists solely in a deoxy-
form, one of the clear distinctions between deoxyribonucleic acids (DNA) and ribo-
nucleic acids (RNA). It is also important to note that UMP (5) and CMP (8), which
differ between the attachment of a carbonyl group and an amino group, respectively,
on the C-4 carbon of the nucleoside base, switch elution orders when separated on
the Tosoh Amide-80 column and the Phenomenex Luna HILIC column. The amide
group retains like-amine groups present on the CMP (8) nucleoside base more than
the carbonyl group on the UMP (5).

The establishment of calibration curves for each nucleotide in both the positive- and
negative-ionization mode was achieved by use of seven chosen concentration points
within a range of 0–350â•›μM. A fitting of the resultant plots was then performed and
lead to the discovery that not all generated calibration curves were uniformly linear,
the details of which are reported in Table 19.4. Some nucleotide curves were best fit
to a polynomial (x2) regression or even a segmented, linear regression containing two
distinct linear trends within the same calibration curve. It is also interesting to note that
this nonlinear fit did not carry through between both positive- and negative-ionization
modes. An analysis of the calibration curves lead to the finding that nucleotide curves
that were nonlinear in one specific ionization mode would be linear in the opposite
ionization mode with the exception of ADP (2), which produced nonlinear calibration
curves in both modes. Figure 19.5 presents the notable calibration curves of nucle-
otides that illustrate these results. In the positive-ionization mode, CMP (8) clearly
fit a square-polynomial with an R2-value of 0.999, whereas when the analysis was
performed in the negative-ionization mode, a linear regression was the best fit to the
curve with an R2-value of 0.991. Conversely, in the case of UDP (6), the opposite was
observed where the linear fit was best achieved on the calibration curve generated
from the positive-ionization mode, while the curve from the negative-ionization mode
was best fit by a square polynomial regression. Both regression fits of UDP calibration
curves had correlation coefficients better than R2 = 0.988. Further questions arise from
the two calibration curves obtained from the method validation of ADP (2), which
not only displayed a square-polynomial fit for its curve from the negative-ionization
mode, but also produced a segmented, linear fit for the calibration curve from the posi-
tive mode as seen from Figure 19.5. There are clearly two separate linear functions in
the ADP positive-mode calibration curve where the first line retains linearity from 0
to 150â•›μM, and the second line begins at 150â•›μM and maintains linearity to 300â•›μM.
Currently, the cause of these nonlinear regressions is difficult to surmise. It can only
be speculated that some mechanism occurs during the chromatographic elution that
reduces the signal detection of certain nucleotides at low concentrations up to a certain
threshold, after which the signal intensities of the nucleotide rise sharply. It is pos-
sible that poor peak shape for some of the compounds could also account for this odd
behavior. Fortunately, with the exception of ADP (2), the issue of nonlinearity can be

Table 19.4
Selected Method Validation Parameters for Positive- and Negative-Mode Ionization of Eight Nucleotides
Positive Mode Negative Mode
tR ± SD (min) Best Fit Range (R2) LOD (ng) Best Fit Range (R2) LOD (ng)
AMP 14.4 ± 0.1 Linear 20–300â•›μM (0.991) 96 Linear 10–300â•›μM (0.991) 74
ADP 17.80 ± 0.06 Linear, segmented 10–150â•›μM (0.994); 116 Polynomial (x2) 10–300â•›μM (0.998) ∼60
150–300â•›μM (0.987)
GMP 15.45 ± 0.08 Linear 10–300â•›μM (0.994) 92 Linear 10–300â•›μM (0.980) 74
GDP 17.92 ± 0.06 Linear 20–300â•›μM (0.970) 162 Polynomial (x2) 10–300â•›μM (0.972) ∼70
UMP 14.3 ± 0.1 Polynomial (x2) 10–300â•›μM (0.999) ∼50 Linear 10–300â•›μM (0.982) 82
UDP 16.73 ± 0.06 Linear 20–300â•›μM (0.988) 127 Polynomial (x2) 10–300â•›μM (0.998) ∼60
TMP 13.0 ± 0.2 Linear 10–300â•›μM (0.993) 71 Linear 10–300â•›μM (0.961) 110
CMP 16.04 ± 0.08 Polynomial (x2) 10–300â•›μM (0.999) ∼50 Linear 10–300â•›μM (0.991) 103

3.0 2.5
CMP (+) CMP (–)
2.5
2.0
y = 262.09x2 + 11506x + y = 7572.6x + 62450
Ion intensity/107
Ion intensity/106
2.0 101960 R2 = 0.9914
R2 = 0.9998 1.5
1.5
1.0
1.0
0.5
0.5
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
(A) Concentration (μM) (B) Concentration (μM)
3.0 14.0 ADP (–)

ADP (+)
12.0
2.5
y = 136.67x2 – 1389.9x +
10.0
Ion intensity/106
Ion intensity/107
2.0 yb = 157997xb – 2E + 07 167258

R2 = 0.9982
R2 = 0.9857 8.0
1.5
6.0
ya = 26365xa + 4115.6
1.0 R2 = 0.9947
4.0
0.5 2.0
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
(C) Concentration (μM) (D) Concentration (μM)
16.0 UDP (+)

UDP (–)
14.0 10.0
y = 959.38x2 – 3134.7x – 96506
12.0 9.0 R2 = 0.998
Ion intensity/103
y = 4320.8x – 43137 8.0

Ion intensity/107
10.0 R2 = 0.9878 7.0

8.0 6.0
5.0
6.0
4.0
4.0 3.0
2.0
2.0
1.0
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
(E) Concentration (μM) (F) Concentration (μM)
Figure 19.5 Calibration curves obtained for CMP (8) performed in the (A) positive-
ionization mode and (B) negative-ionization mode; for ADP (2) performed in the
(C) positive-ionization mode and (D) negative-ionization mode; and for UDP (6) performed
in the (E) positive-ionization mode and (F) negative-ionization mode. All analyses were
performed with the optimized method on the Tosoh Amide-80 column.

circumvented with an analysis of the nucleotide in the opposite ionization mode, which
does retain linearity within the reported range. The calibration curves maintained
correlation to their best fit regression primarily within a range of 10–300â•›μM with
some exceptions (i.e., the positive-ionization mode of ADP (2), GDP (4), and UDP (6)).
At a concentration of 350â•›μM (data not shown), all nucleotide calibration curves devi-
ated from their best fit and were thus excluded from the calibration curves.
The data indicate good reproducibility of the developed method with regard to
retention times (<2% RSD). The limit of detection (LOD) determinations for each
nucleotide in each ionization mode indicate moderate levels of sensitivity ranging
in the detection of 48–162â•›ng of analyte in a 15â•›μL injection volume. The detection
limits of the nonlinear plots were only roughly determined by successive decreases
of analyte concentration until the nucleotide mass spectral signal was unable to be
detected. In calculating the LOD for uridine 5′-diphosphate (UDP) in the positive-
ionization mode, both the signal intensities from [UDP + H]+ and the [UDP + NH4]+
ions were taken into account by plotting the summation of both signal intensities
against concentration after which the LOD was then calculated as mentioned, previ-
ously. The division of the total signal intensity between the [UDP + H]+ ion and the
[UDP + NH4]+ adduct leads to an expected decrease in sensitivity for this specific
nucleotide. According to the table, the best ionization mode for optimal LOD is
analyte dependent. AMP, ADP, GMP, GDP, and UDP show the lowest LOD in the
negative-ionization mode, whereas UMP, TMP, and CMP are detected with the lowest
LOD in the positive-ionization mode.
19.4 Concluding Remarks

From this investigation, an HILIC-ESI-MS method was developed for the separation
of eight different nucleotides. In addition, six different commercial columns were
tested. It is clear that the Tosoh Amide-80 column is most suited for achieving the
separation of nucleotides, whereas the Phenomenex Luna HILIC column showed the
capability in doing so as well. With the use of ESI-MS detection, overlapping signals
from unresolved analytes can be deconvoluted for nucleotide detection, however,
poor peak shapes in some cases, arising from multiple interactions modes between
the analyte and the stationary phase, were difficult to avoid.
The optimization of pH was crucial in the development of the method to control
the protonation states of the different nucleotides. The described method for the
nucleotides is expected to lead to a further development of rapid enzyme kinetic
assays. Quantification of nucleotide enzyme substrates or enzyme products can now
be achieved using a more simplified HPLC-ESI-MS protocol that does not require
the use of ion-pairing reagents. However, the investigation of additional stationary-
phase chemistries (e.g., zwitterionic) may be necessary to achieve the best possible
analysis conditions.
Acknowledgments
The authors wish to acknowledge support from the University of Texas at Arlington,
Shimadzu Scientific Instruments, Inc., and Varian, Inc. in performing this work.

References
1. Nguyen, H. P.; Schug, K. A. The advantages of ESI-MS detection in conjunction
with HILIC mode separations: Fundamentals and applications. J. Sep. Sci. 2008, 31,
1465–1480.
2. Shou, W. Z.; Weng, N. Simple means to alleviate sensitivity loss by trifluoroacetic
acid (TFA) mobile phases in the hydrophilic interaction chromatography–electrospray
tandem mass spectrometric (HILIC–ESI/MS/MS) bioanalysis of basic compounds.
J. Chromatogr. B 2005, 825, 186–192.
3. Li, W.; Li, Y.; Li, A. C.; Zhou, S.; Weng, N. Simultaneous determination of norethindrone
and ethinyl estradiol in human plasma by high performance liquid chromatography with
tandem mass spectrometry—Experiences on developing a highly selective method using
derivatization reagent for enhancing sensitivity. J. Chromatogr. B 2005, 825, 223–232.
4. Song, Q.; Junga, H.; Li, A. C. et al. Automated 96-well solid phase extraction and hydro-
philic interaction liquid chromatography–tandem mass spectrometric method for the
analysis of cetirizine (ZYRTEC®) in human plasma—With emphasis on method rugged-
ness. J. Chromatogr. B 2005, 814, 105–114.
5. Hao, Z.; Xiao, B.; Weng, N. Impact of column temperature and mobile phase com-
ponents on selectivity of hydrophilic interaction chromatography (HILIC). J. Sep. Sci.
2008, 31, 1449–1464.
6. Alpert, A. J. Hydrophilic-interaction chromatography for the separation of peptides,
nucleic acids and other polar compounds. J. Chromatogr. A 1990, 499, 177–196.
7. Gianotti, V.; Chiuminatto, U.; Mazzucco, E.; Gosetti, F.; Bottaro, M.; Frascarolo, P.;
Gennaro, M. C. A new hydrophilic interaction liquid chromatography tandem mass
spectrometry method for the simultaneous determination of seven biogenic amines in
cheese. J. Chromatogr. A 2008, 1185, 296–300.
8. Wang, Y.; Lu, X.; Xu, G. Development of a comprehensive two-dimensional hydro-
philic interaction chromatography/quadrupole time-of-flight mass spectrometry system
and its application in separation and identification of saponins from Quillaja saponaria.
J. Chromatogr. A 2008, 1181, 51–59.
9. Zhang, H.; Guo, Z.; Zhang, F.; Xu, Q.; Liang, X. HILIC for separation of co-eluted
flavonoids under RP-HPLC mode. J. Sep. Sci. 2008, 31, 1623–1627.
10. Zhang, H.; Wei, Z. G.; Jiatao, L. et al. Purification of flavonoids and triterpene saponins
from the licorice extract using preparative HPLC under RP and HILIC mode. J. Sep. Sci.
2009, 32, 526–535.
11. Appelblad, P.; Jonsson, T.; Jiang, W.; Irgum, K. Fast hydrophilic interaction liquid chro-
matographic separations on bonded zwitterionic stationary phase. J. Sep. Sci. 2008, 31,
1529–1536.
12. Dernovics, M.; Lobinski, R. Speciation analysis of selenium metabolites in yeast-based
food supplements by ICPMS assisted hydrophilic interaction HPLC hybrid linear ion
trap/orbitrap MS. Anal. Chem. 2008, 80, 3975–3984.
13. Qin, F.; Zhao, Y.-Y.; Sawyer, M. B.; Li, X.-F. Hydrophilic interaction liquid chromatog-
raphy–tandem mass spectrometry determination of estrogen conjugates in human urine.
Anal. Chem. 2008, 80, 3404–3411.
14. Jiang, Z.; Smith, N. W.; Ferguson, P. D.; Taylor, M. R. Hydrophilic interaction chroma-
tography using methacrylate-based monolithic capillary column for the separation of
polar analytes. Anal. Chem. 2007, 79, 1243–1250.
15. Wang, C.; Jiang, C.; Armstrong, D. W. Considerations on HILIC and polar organic
solvent-based separations: Use of cyclodextrin and macrocyclic glycopetide stationary
phases. J. Sep. Sci. 2008, 31, 1980–1990.
16. Park, S. K.; Liao, L.; Kim, J. Y.; Yates, J. R. A computational approach to correct arginine-
to-proline conversion in quantitative proteomics. Nat. Methods 2009, 6, 184–185.

17. Ebhardt, H. A.; Xu, Z.; Fung, A. W.; Fahlman, R. P. Quantification of the post-translational
addition of amino acids to proteins by MALDI-TOF mass spectrometry. Anal. Chem.
2009, 81, 1937–1943.
18. Zuberovic, A.; Wetterhall, M.; Hanrieder, J.; Bergquist, J. CE MALDI-TOF/TOF MS
for multiplexed quantification of proteins in human ventricular cerebrospinal fluid.
Electrophoresis 2009, 30, 1836–1843.
19. Xia, B.; Feasley, C. L.; Sachdev, G. P.; Smith, D. F.; Cummings, R. D. Glycan reductive
isotope labeling for quantitative glycomics. Anal. Biochem. 2009, 387, 162–170.
20. Ding, L.; Cheng, W.; Wang, X.; Ding, S.; Ju, H. Carbohydrate monolayer strategy
for electrochemical assay of cell surface carbohydrate. J. Am. Chem. Soc. 2008, 130,
7224–7225.
21. Boersema, P. J.; Mohammed, S.; Heck, A. J. R. Hydrophilic interaction liquid chroma-
tography (HILIC) in proteomics. Anal. Bioanal. Chem. 2008, 391, 151–159.
22. Ren, Y. L.; Garges, S.; Adhya, S.; Krakow, J. S. Cooperative DNA binding of heterolo-
gous proteins: Evidence for contact between the cyclic AMP receptor protein and RNA
polymerase. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 4138–4142.
23. Ling, E.; Danilov, Y. N.; Cohen, C. M. Modulation of red cell band 4.1 function by
cAMP-dependent kinase and protein kinase C phosphorylation. J. Biol. Chem. 1988,
263, 2209–2216.
24. Yamano, S.; Tanaka, K.; Matsumoto, K.; Toh-e, A. Mutant regulatory subunit of
3′,5′-cAMP-dependent protein kinase of yeast Saccharomyces cerevisiae. Mol. Gen.
Genet. 1987, 210, 413–418.
25. Bubis, J.; Neitzel, J. J.; Saraswat, L. D.; Taylor, S. S. A point mutation abolishes binding
of cAMP to site A in the regulatory subunit of cAMP-dependent protein kinase. J. Biol.
Chem. 1988, 263, 9668–9673.
26. Corbin, J. D.; Cobb, C. E.; Beebe, S. J. et al. Mechanism and function of cAMP- and
cGMP-dependent protein kinases. Adv. Second Messenger Phosphoprotein Res. 1988,
21, 75–86.
27. Swierczynski, J.; Klimek, J.; Zelewski, L. Correlation between the malate dependent
progesterone and citrate biosynthesis in the mitochondrial fraction of human term pla-
centa. The stimulatory effect of ADP and ATP. J. Steroid Biochem. Mol. Biol. 1986, 24,
591–595.
28. Wittenberg, B. A.; Wittenberg, J. B. Myoglobin-mediated oxygen delivery to mito-
chondria of isolated cardiac myocytes. Proc. Natl. Acad. Sci. U.S.A. 1987, 84,
7503–7507.
29. Kun, E. Kinetics of ATP-dependent Mg2+ flux in mitochondria. Biochemistry 1976, 15,
2328–2336.
30. Sarkar, N. K. The effect of ions and ATP on myosin and actomyosin. Enzymologia 1950,
14, 237–245.
31. Geyer, M.; Schweins, T.; Herrmann, C.; Prisner, T.; Wittinghofer, A.; Kalbitzer, H. R.
Conformational transitions in p21ras and in its complexes with the effector protein Raf-
RBD and the GTPase activating protein GAP. Biochemistry 1996, 35, 10308–10320.
32. Valencia, A.; Chardin, P.; Wittinghofer, A.; Sander, C. The ras protein family:
Evolutionary tree and role of conserved amino acids. Biochemistry 1991, 30, 4637–4648.
33. Klebe, C.; Bischoff, F. R.; Ponstingl, H.; Wittinghofer, A. Interaction of the nuclear GTP-
binding protein Ran with its regulatory proteins RCC1 and RanGAP1. Biochemistry
1995, 34, 639–647.
34. Riese, M. J.; Wittinghofer, A.; Barbieri, J. T. ADP ribosylation of Arg41 of Rap by ExoS
inhibits the ability of Rap to interact with its guanine nucleotide exchange factor, C3G.
Biochemistry 2001, 40, 3289–3294.
35. Guo, Y.; Gaiki, S. Retention behavior of small polar compounds on polar stationary
phases in hydrophilic interaction chromatography. J. Chromatogr. A 2005, 1074, 71–80.

36. Tuytten R.; Lemiere, F.; Van Dongen, W.; Esmans. E. L.; Slegers, H. Short capillary ion-
pair high-performance liquid chromatography coupled to electrospray (tandem) mass
spectrometry for the simultaneous analysis of nucleoside mono-, di- and triphosphates.
Rapid Commun. Mass Spectrom. 2002, 16, 1205–1215.
37. Xing, J.; Adrienne, A. A.; Zhao, T. N. Liquid chromatographic analysis of nucleosides
and their mono-, di- and triphosphates using porous graphitic carbon stationary phase
coupled with electrospray mass spectrometry. Rapid Commun. Mass Spectrom. 2004,
18, 1599–1606.
38. Cordell, R. L.; Hill, S. J.; Ortori, C. A.; Barrett, D. A. Quantitative profiling of nucleo-
tides and related phosphate-containing metabolites in cultured mammalian cells by liq-
uid chromatography tandem electrospray mass spectrometry. J. Chromatogr. B 2008,
871, 115–124.
39. Vincenzo, P.; Giuliano, C.; Zhang, R. et al. HILIC LC-MS for the determination of
2′-C-methyl-cytidine-triphosphate in rat liver. J. Sep. Sci. 2009, 32, 1275–1283.
40. Subirats, X.; Bosch, E.; Rosés, M. Retention of ionisable compounds on high-performance
liquid chromatography XVIII: pH variation in mobile phases containing formic acid,
piperazine, tris, boric acid or carbonate as buffering systems and acetonitrile as organic
modifier. J. Chromatogr. A 2009, 1216, 2491–2498.
41. Kostiainen, R.; Kauppila, T. J. Effect of eluent on the ionization process in liquid chro-
matography–mass spectrometry. J. Chromatogr. A 2009, 1216, 685–699.
42. Mansoori, B. A.; Volmer, D. A.; Boyd, R. K. “Wrong-way-round” electrospray ioniza-
tion of amino acids. Rapid Commun. Mass Spectrom. 1997, 11, 1120–1130.
43. Zhou, S.; Cook, K. D. Protonation in electrospray mass spectrometry: Wrong-way-
round or right-way-round? J. Am. Soc. Mass Spectrom. 2000, 11, 961–966.
44. De Schutter, J. A.; De Moerloose, P. Polar contributions of the stationary phase to the
reversed-phase ion-pair high-performance liquid chromatographic separation of quater-
nary ammonium drugs. J. Chromatogr. A 1988, 437, 83–95.
45. Sýkora, D.; Tesarová, E.; Popl, M. Interactions of basic compounds in reversed-phase
high-performance liquid chromatography influence of sorbent character, mobile phase
composition, and pH on retention of basic compounds. J. Chromatogr. A 1997, 758,
37–51.
46. Gilar, M.; Yu, Y.-Q.; Ahn, J.; Fournier, J.; Gebler, J. C. Mixed-mode chromatography for
fractionation of peptides, phosphopeptides, and sialylated glycopeptides. J. Chromatogr.
A 2008, 1191, 162–170.
47. Liu, M.; Chen, E. X.; Ji, R.; Semin, D. Stability-indicating hydrophilic interaction liquid
chromatography method for highly polar and basic compounds. J. Chromatogr. A 2008,
1188, 255–263.
48. Enke, C. G. A predictive model for matrix and analyte effects in electrospray ionization
of singly-charged ionic analytes. Anal. Chem. 1997, 69, 4885–4893.
49. Sherman, C. L.; Brodbelt, J. S. An equilibrium partitioning model for predicting response
to host-guest complexation in electrospray ionization mass spectrometry. Anal. Chem.
2003, 75, 1828–1836.

Liquid Chromatography–
Based Enrichment
Protocol Coupled to
Mass Spectrometry for
Glycoproteome Analysis
Cosima Damiana Calvano
Contents
20.1 Introduction...................................................................................................469
20.2 Protein Modifications: Glycosylation............................................................ 471
20.3 Troubles in Glycoproteome Analysis............................................................. 472
20.4 Enrichment Methods for Glycoproteome...................................................... 473
20.5 HILIC: Mechanism and Material.................................................................. 474
20.6 HILIC: From Glycomics to Glycoproteomics............................................... 477
20.7 Glycoproteomics by HILIC and Mass Spectrometry.................................... 478
20.8 Application to Biomarker Discovery............................................................. 482
References............................................................................................................... 485
20.1 Introduction
Hydrophilic interaction liquid chromatography (HILIC) was first introduced in the
1970s1 for carbohydrates analysis and subsequently defined by Alpert in 19902 as a
variant of normal-phase liquid chromatography (NPLC). However, HILIC offers a
series of advantages compared to NPLC. In the latter, stationary phase is polar and
the mobile phase consists of apolar solvents, resulting in increased retention with
increased polarity of the analyzed samples and/or stationary phase and/or decreased
polarity of the mobile phase.3,4
Nevertheless, the nonpolar solvents used as mobile phases are often dangerous
for environment, expensive, and quite toxic and not completely suitable to solubi-
lize polar and hydrophilic compounds. In HILIC, a hydrophilic stationary phase
469

and a partly aqueous organic solvent mobile phase are used. Similarly to NPLC,
the retention increases with increased polarity of the analyzed compounds and/or
stationary phase and/or decreased polarity of the mobile phase. However, contrary
to NPLC, HILIC uses aqueous-organic solvents with a high organic-solvent frac-
tion as mobile phase, which demonstrate an increased solubility for the polar and
hydrophilic compounds. For this reason, HILIC applications have seen a consider-
able increase of interest5–10 for the analysis of many categories of polar compounds,
charged as well as uncharged, in complex sample mixtures.11
Another reason for the increase in popularity over NPLC is the great compat-
ibility of the buffer conditions used in the liquid chromatography–electrospray
ionization–mass spectrometry (LC–ESI–MS) technique.
The interfacing with ESI–MS is often a problem with NPLC, since totally organic,
nonpolar eluents do not enable an efficient ionization contrary to HILIC where the
presence of water as strongly eluting solvent favors the ESI process.12,13
The high organic content of the buffers gives HILIC two added advantages: high
sensitivity in ESI–MS12,14,15 and faster separations due to the low viscosity of HILIC
eluents.16 In fact, higher portions of organic modifier within the elution solvents
could result in better electrospray sensitivity due to the decreasing of surface ten-
sion and solvation energies for polar compounds. Further, solvents with higher vis-
cosities such as those used in reverse phase tend to be less volatile and have higher
surface tension. These solvents will originate larger and less efficiently charged
droplets at the capillary tip and desolvatation of the electrosprayed eluent droplets
will require higher interface drying gas temperatures. Then, the higher volatility
of the organic modifier permits a more efficient desolvation of the electrosprayed
droplet. As a result, HILIC works best for solutes that are challenging in reverse
phase and can be used as “orthogonal” separation to reverse phase, which allows
a multidimensional separation of complex samples17–19 saving time and preventing
sample losses.20
This approach can be useful to catch and separate solutes that have retention on
both RP and HILIC-type stationary phases such as peptides or proteins, as shown by
the increased number of HILIC application in the proteomics field.21 In particular,
due to the selectivity toward polar group, HILIC has shown to be a useful tool for
the enrichment of posttranslational modifications (PTMs) such as glycosylation,22
N-acetylation,18 and phosphorylation23 in proteomics applications.
In this chapter, an introduction to the most diffuse PTM, i.e., glycosylation,
will be effort together with a summary of HILIC mechanism, materials, and
applications for polar compounds. Moreover, a focus on the use of HILIC in
glycoproteomics for the enrichment of PTMs at both peptide and protein levels
will be presented. It will be described how SPE/HILIC scheme is among the
preferred combinations when the aim is to enrich glycocompounds. The cou-
pling of this enrichment strategy to different spectrometric techniques such as
matrix-assisted laser desorption ionization (MALDI) and ESI will be illustrated
for the analysis of complex sample mixtures such as biological fluids. A compari-
son with some other chromatographic materials for the enrichment yield will be
shortly illustrated.

HILIC-MS for Glycoproteome Analysis 471
20.2 Protein Modifications: Glycosylation

The characterization of the different PTMs is receiving much attention in proteomics
research since they have a significant effect on protein function.24 Glycosylation is
one of the most abundant PTMs of proteins25,26 known to be a highly complex and
dynamic process associated with membrane-bound and secreted proteins. The gly-
cosylation of proteins is involved in many different biological roles including sig-
naling, transcription, apoptosis, differentiation, adhesion, and infiltration, oncogenic
transformation, and metastasis.27–30
Alterations of protein glycosylation sites, alteration in the behavior of peptide
backbone due to glycan composition, size and shape, branching and heterogeneity
of proteins on the cell surface, different protein folding, and the assembly of protein
complexes in body fluids are implicated in several inflammatory diseases, cancers,
and congenital disorders.31–41
Glycoproteins generally exist as populations of glycosylated variants, also known
as glycoforms, of a single homogeneous peptide at one glycosylation site bringing
more than 100 alternative glycans with variable composition, linkage, branching, and
anomericity. This glycan heterogeneity contributes to the structural complexity and
diversity of glycoproteins. There are five known types of protein glycosylations42–44:
N-Glycosylation, O-Glycosylation, C-Mannosylation, phosphoglycosylation, and
glypiation. In O-glycosylation oligosaccharide chains are covalently attached to ser-
ine or threonine residues, in C-glycosylation to tryptophan residues, in phosphogly-
cosylation to serine residues, in glypiation to proline residues, and in N-glycosylation
to asparagine residues. However, N-glycosylation is characterized in greatest detail.
In this case, glycosylation occurs only at the sequence Asn-X-Ser/Thr, where X
can be any amino acid except proline. Figure 20.1 shows some typical structures of
N-linked and O-linked oligosaccharide chains. A suitable method to verify whether
a candidate glycoprotein is N-glycosylated is to treat the protein with enzymes called
N-glycanase that cleave the glycan residues and examine the product by SDS-PAGE.
N-linked
α2-6 β1-4 β1-2 Fuc
NeuAc Gal GlcNAc Man α1–6
β1-4 β1-4 α1–6
Man GlcNAc GlcNAc Asn
α2-6 β1-4 β1-2
NeuAc Gal GlcNAc Man α1–3
O-linked
α2-3 β1-4
NeuAc Gal GlcNAc β1–6
α
GalNAc Thr
α2-3
NeuAc Gal β1–3
Figure 20.1 Some typical structures of N-linked and O-linked oligosaccharide chains
found attached to proteins.

If a decrease in molecular weight is observed for the product of the N-glycanase

reaction in comparison to the undigested glycoprotein, then the removal of N-linked
carbohydrate has occurred and the protein was glycosylated.
N-linked glycans attached to glycoproteins fall into three major classes, i.e., high-
mannose, complex, and hybrid, all of which have a common core structure consist-
ing of two N-acetylglucosamine (GlcNAc) residues, three glucose (Glc), and nine
mannose (Man) monosaccharides. In high-mannose type, several mannose residues
ranging from 5 to 9 are attached to the core structure. Complex-type N-glycans may
contain branching structures originating from the GlcNAc residues known as anten-
nae. Each core mannose can have several galactose, GlcNAc, sialic acid, and fucose
residues attached to the core structure giving rise to more than two antennae. Hybrid-
type glycans contain structural elements of both high-mannose- and complex-type
glycans. The full characterization of all glycoproteins in a complex sample is a very
difficult task involving the determination of glycan size, branching points, and link-
ages of the carbohydrate monomers for each glycosylation site.
20.3 Troubles in Glycoproteome Analysis

The characterization of posttranslationally modified proteins can be approached in
two ways. In fact, it is possible to investigate the modification while still attached to
the target protein or following its release from the target protein. Each of the strate-
gies for different types of glycosylation uses different experimental protocols.
However, in all cases, variable composition, heterogeneity, linkage, branching,
and anomericity of the constituent sugars contribute to the structural complexity
of glycoprotein glycans. As a consequence, a large number of techniques are often
necessary to fully determine the structure of a glycan or a glycoprotein. These tech-
niques may include 1H and 13C-nuclear magnetic resonance (NMR) spectroscopy
for the determination of the structure and configuration of the carbohydrates,45,46
electrophoresis,47 gas chromatography48 (GC), and liquid chromatography (LC),49,50
possibly combined with mass spectrometry (MS).51–53 However, the study of protein
glycosylation had limited applications in biomedical and glycobiological fields since
each of these strategies has its challenges inherent to the technique itself and to the
complexity of the modifications, the sensitivity with which it can be analyzed,
the relatively extensive and time-consuming analytical procedures employed.
As far as NMR is concerned, it is less sensitive than other techniques and
requires rather homogeneous glycoprotein samples, resulting in a few suitable tech-
niques for most proteomic studies, where the amount of sample is a limiting factor.
Electrophoresis has shown to be unable to generate a quantitative or even qualitative
separation for glycoprotein and GC is often associated with very time-consuming
sample preparation protocols including derivatization.
Indeed, LC–MS technique overcomes most of these problems even if three major
drawbacks still remain.54 First, glycosylations can be highly labile, overall on threo-
nine or serine residues, complicating analyses and reproducibility. Second, glycosyl-
ated peptides have poor ionization efficiencies with respect to unmodified peptides.
Third, glycan modifications are extremely heterogeneous complicating the analysis,
as a single peptide may be modified by several forms of glycans with different

structures and masses, resulting in a signal dilution of glycopeptide ions overall in

the case of complex protein mixtures.
Although many complications, recent advances in sample preparation and MS
have demonstrated the ability to isolate and identify both the peptide backbone and
glycan. It has been demonstrated that it is always advantageous and often necessary
to enrich for glycopeptides prior to MS analysis. For this reason, there is growing
interest in methods that can selectively isolate glycopeptides from proteolytic digests
of complex protein mixtures for subsequent analysis by LC–MS.55–59
20.4 Enrichment Methods for Glycoproteome

Several recent studies highlight that a particular modification in proteins or pep-
tides could be used as target to develop enrichment strategies using affinity chro-
matography methods.60–62 For example, in glycoproteins, the glycan moiety can be
used to select the modified proteins/peptides in complex peptide mixtures using
glycan-specific structural features (e.g., affinity enrichment by antibodies and lec-
tins63–70) or glycan-specific chemical properties (e.g., derivatization methods71–75).
Among affinity enrichment techniques, increasing interest has grown toward
lectin-based chromatography used extensively to purify glycoproteins, glycopep-
tides and oligosaccharides from different biological samples. Lectins are usually
plant glycoproteins that specifically bind to certain oligosaccharide structures on
glycoconjugates. Different types of lectins are easily immobilized on solid sup-
ports showing a variety of selectivities and binding specificities to different soluble
carbohydrates or to carbohydrate moieties,67,68 which are a part of glycoproteins or
glycolipids. However, each lectin species isolates only a subset of glycan species
making difficult an analysis of a complex unknown sample. Besides the specific
lectin-affinity enrichment methods, approaches based on physical and chemical
features of glycopeptides and glycoproteins are most valuable. First, the diffuse
characteristic that tryptic glycopeptides arising from a complex peptide/glycopep-
tide mixture have higher masses than nonglycosylated permits a significant enrich-
ment by size exclusion chromatography.76–79 Recently, a chemical derivatization
followed by the immobilization using periodate oxidation and hydrazide chem-
istry58 has been used to selectively and efficiently isolate, identify, and quantify
N-linked glycopeptides. However, in this workflow sugars are destructed and
removed from glycan moieties causing a loss of structural information even if a
relative quantitation of glycopeptides can be attempted by the inclusion of stable
isotope labeling.80 The problem frequently met in these chemistry-based meth-
ods relates to the large amounts of starting material and to the generation of side
products. It has recently demonstrated that complex sample fractionation, purifica-
tion, and preconcentration can be achieved by using “custom-made” miniaturized
solid-phase extraction (SPE)81,82 made with a variety of chromatographic func-
tionalities and selectivity. These kinds of SPE columns, cheap, easy to prepare,
simple to use usually require very little sample material. For example, SPE col-
umns using graphite powder show a good recovery of very hydrophilic peptides,
such as glycopeptides71; SPE columns using titanium dioxide microparticles show
a good recovery of sialic acid modified peptides.83 HILIC has mainly been used

for the analysis of small, polar molecules such as carbohydrates, oligonucleotides,

and amino acids, and more recently for the separation of peptides and glycopep-
tides.84–86 It has established to be a useful method for purification of glycosylated
peptides due to the retention determined by the hydrophilic character and size of
the glycan moiety. HILIC could reduce the complexity of peptide/glycopeptide
mixtures allowing depletion of hydrophobic peptides, and retention of hydrophilic
glycopeptides.
20.5 HILIC: Mechanism and Material

As stated in the introduction, HILIC can be considered a variant of normal-phase
chromatography, where the retention mechanism is supposed to be partition-
ing of the analyte between a water-enriched layer covering the hydrophilic sta-
tionary phase and a relatively hydrophobic bulk eluent, typically composed of
5%–40% water in acetonitrile.2 Gregor et al.87 described the existence of a water-
enriched layer on the surface of hydrophilic stationary phases, while Rückert and
Samuelson88 proposed an absorption of nonelectrolytes in ion exchange resins by
means of a stagnant water layer.
Several studies demonstrated that analytes are retained on the hydrophilic sta-
tionary phase by hydrogen bonding, ionic interactions, and dipole–dipole interac-
tions. Yoshida89 proposed an adsorptive retention mechanism for peptides in HILIC
explaining that elution patterns were similar to the normal-phase separation and
suggested hydrogen bonding as the principal interaction mode. Guo and Gaiki13,90
supposed a partitioning mechanism since they found two experimental evidences
that could not be explained otherwise: (a) no strong specific interaction between the
solutes and four different HILIC stationary phases and (b) the increase in retention
time with salt concentration. However, it is always more diffuse and recognized the
definition introduced by Alpert2 who stated that the term HILIC should be used in
the case of (a) strongly eluting solvent consisting of water and (b) retention mecha-
nism based on partitioning.
Recently, several stationary phases have been specifically introduced for HILIC
approaches each of them displaying different retention characteristics, separation
selectivity, and requiring distinct buffer constitutions for each application.91 The
use of buffered eluents is highly recommended to reduce electrostatic interactions
between charged analytes and charged silanol groups or other charged species on
the stationary phase. The preferred buffer salts are ammonium salts of formate or
acetate due to their suitable buffering range, high solubility in acetonitrile rich elu-
ents, and compatibility with MS techniques.
It is important to highlight that the type of the stationary phase together with the
composition of the mobile phase can be plainly tuned to solve specific separation
problems in the mixed-mode HILIC. The most diffuse phases used include neat silica
(Atlantis) containing functional groups such as siloxanes, derivatized silica carrying
polymeric coatings like poly(2-hydroxyethyl aspartamide) (PolyHydroxyethyl A),
cation exchanger polysulfoethyl A,92 weak cation exchanger Polycat A,93 weak anion
exchanger PolyWAX,94,95 TSK-Gel amide 80,98,99 and zwitterionic (ZIC)-HILIC.22
Figure 20.2 illustrates some structures of typical HILIC phases.

SO3H
HO3S
H2N R4
O
HN
NH
OO O H H
O m n
H R3
N *
N N n
R1 R2
H H
O
O
NH
PolySulfoethyl A TSK-Gel Amide-80
O–
O
O
H H H
N N N *
N n
* H
O O O
HN O –O O
SO–3 O
N+
NH
ZIC–HILIC PolyCAT A
Figure 20.2 Chemical structures of the functional groups in common HILIC station-
ary phases. (Reproduced from Boersema, P.J. et al., Anal. Bioanal. Chem., 391, 151, 2008.
With permission.)
The stationary phase of silica(-based) columns contains negative charge at

basic pH due to deprotonation of silanol groups, then the HILIC retention is
increased when pH is below pKa of basic compounds. In such cases, negatively
charged surface electrostatically repels negatively charged compounds like sialic
acids requiring an acidic pH and/or ion pairing reagent to avoid unwanted inter-
actions between basic compounds and deprotonated silanol groups. In the case
of amine and aminopropyl phases an opposite effect is observed, where at pH
below 9, positively charged amino groups act as weak anion-exchangers requiring
a basic pH and/or ion pairing reagent. It is easy to understand as a pH shift could
have a great effect on the ionic interactions by reducing or introducing charged
groups on both analyte and stationary phase. Moreover, another critical factor
when changing the mobile phase is that too much increased ionic strength results
in a reduction of ionic interactions between analyte and stationary phase, leaving
the hydrogen bonding as the major retentive effect. The presence of ions in the
mobile phase will also weaken ionic interactions since they can act as counterions
for both charged analytes and groups of the stationary phase and will decrease
the ionization efficiency of ESI. Further disadvantages of this kind of bare col-
umns are the possibility to observe phenomena of dissolution and hydrolysis that
limited stability in aqueous solvents and phenomena of irreversible adsorption
of compounds onto stationary phases due to their highly polar character (silica
columns) and to high reactive stationary phases (aminopropyl columns). More
recently, zwitterionic sulfoalkylbetaine stationary phases71 have been introduced
for HILIC separations of small organic ionic compounds, inorganic salts and

proteins.98,99 These charged stationary phases exhibit permanent negative and

positive charges given by strongly acidic sulfonic acid groups and strongly basic
quaternary ammonium groups separated by a short alkyl spacer acting as a weak-
cation exchanger.5
The chromatographic properties of zwitterionic materials significantly dif-
fer from other HILIC phases. The 1:1 molar ratio of the two oppositely charged
groups assures a very low net negative surface charge on the bonded layer, only
little affected by pH, and ascribed to large distance of the sulfonic groups from the
silica gel surface. Therefore, for zwitterionic stationary phase, both ion-exchange
and electrostatic interactions are considered weaker than typical ion exchangers and
bonded amino phases, or “bare” silica, amide and other bonded phases, respectively.
In fact, in this case, the residual silanol groups are efficiently protected by self-
association of oppositely charged functional groups. As schematized in Figure 20.3,
the sulfoalkylbetaine-bonded phases strongly adsorb water by hydrogen bonding
and the bulk layer of water becomes part of the stationary phase. This layer plays
an important role in determining the retention mechanism, in part explicated by
partitioning of the analytes between the bulk mobile phase and the adsorbed water
layer, in part by polar interactions (such as hydrogen-bonding and dipole–dipole
interactions) in the stationary phase, and in smaller extent by weak electrostatic
interactions between analytes carrying either positive or negative charges. In any
case, the retention increases with increasing sample polarity. Different trades are
commercially available for zwitterionic columns such as ZIC–HILIC (on silica gel
support) and ZIC–pHILIC (on polymer support). Takegawa et al.100 demonstrated
the potential of ZIC–HILIC by the separation of 2-aminopyridine-labeled N-glycans
of human serum IgG.
CH3
–
+ Hydrophilic
C N CH2 CH2 CH2 SO3
H2 partitioning
H2O
CH3 Analyte Analyte
CH3
H2O
+ –
C N CH2 CH2 CH2 SO3
H2
Electrostatic
CH3 H2 O
interactions Analyte
ZIC–HILIC
CH3 H2O
+ –
C N CH2 CH2 CH2 SO3
H2
CH3 H 2O
H2O
CH3
+ –
C N CH2 CH2 CH2 SO3
H2
CH3
Figure 20.3 Mechanism of HILIC separation including partition and adsorption driven
by hydrophilic and electrostatic interactions on a sulfoalkylbetaine-bonded phase.

20.6 HILIC: From Glycomics to Glycoproteomics

The emergence of glycoproteomics has extended the applicability of HILIC from
only being used for carbohydrate analysis, via peptides, to small molecules and even
whole proteins. Many applications of HILIC–ESI–MS for neutral oligosaccharides
have been described in literature such as the analysis of plant sugar residues,101 the
analysis of reduced O-glycans,102,103 of fluorescently labeled N-glycans,104–109 and the
stereoisomers (anomers) of the reducing end carbohydrate ring of N-glycans.107,108
A particular force of this technique in the field of oligosaccharides analyses is the
tolerance of the chromatographic media to modifications: since the retention is
mainly established by the hydrogen bonding of the multiple glycan moiety, polar
substituents as sulfates or various hydrophobic tags remain all compatible with
HILIC separation of oligosaccharides. This strength explains also why the separa-
tions of monosaccharides (mannose, fucose, galactose, glucuronic acid, N-acetyl-
glucosamine, N-acetylneuraminic acid) present in N-linked glycoproteins110 are also
determined with HILIC. Many other applications are found in pharmaceutical field
such as the determination of mannitol, gemcitabine, and sodium cation in various
formulations111 or fine particle dose of lactose in a dry inhalation product,112 in agri-
cultural field such as the determination of spectinomycin and lincomycin in run-off
waters from lands assigned to crop cultivation,113 or in biological field such as the
determination of 2-aminopyridine derivatized N-glycans and N-glycopeptides of
human serum immunoglobulin114 and of α-1-acid glycoprotein.100
As seen in the overview above, the more recent applications of hydrophilic sepa-
ration materials referred to protein separations. At the present, the separation of
proteins in their native and overall of protein or peptide posttranslational modified
state remains a very challenging task in LC difficult to run in a routine fashion.
The translation of this technique from small polar compounds to proteomics has
been quite simple because no significant differences between the chromatographic
behavior of small molecules and large macromolecules as proteins are observed
if HILIC interactions are dominating. In theory, the large size of proteins means
small diffusion coefficients and lower number of interactions with the stationary
phase per unit of time. Decreased diffusion that is on the base of mass transfer
mechanism in chromatography, results in decreased efficiency in the separation
of biological macromolecules compared to small molecules in the same experi-
mental conditions. Protein separations therefore rely on selectivity. However, the
chromatographic behavior of oligosaccharides and glycopeptides can be similar if
hydrophilic interactions dominate on other interactions correlated to analyte size
such as in size exclusion chromatography. This phenomenon has been observed in
the separation of oligosaccharides and glycopeptides varying the acetonitrile con-
centration in the mobile phase.87 In fact, at acetonitrile concentrations of up to 40%
many silica-based and polymeric stationary phases show conventional gel filtration
behavior for oligosaccharides in size exclusion chromatography. If the acetonitrile
concentrations is up to or higher than 50%, an inversion of the elution sequence is
observed and decreasing polarity of the mobile phase results in increased reten-
tion. At these acetonitrile concentrations, the hydrophilic interaction of oligosac-
charides with the stationary phase is favored compared to the gel filtration effects.

This observation was confirmed for N-glycopeptides: the separation of a tryptic

digest of horseradish peroxidase in 50% acetonitrile resulted in the separation of
glycoforms. In this case, it was noted that glycoforms with disaccharides attached
to the N-glycosylation site eluted earlier than the glycoforms with a heptasac-
charide N-glycan structure, following HILIC behavior rather than size exclusion
chromatography. The same considerations can be carried out for HILIC applied
as SPE for oligosaccharides and glycopeptides in the complex mixtures contain-
ing salts, nonglycosylated peptides and proteins. Recently, more global proteomics
approaches were reported that targeted protein glycosites using SPE–HILIC to
enrich glycopeptides.22,115–118
20.7 Glycoproteomics by HILIC and Mass Spectrometry

MS has proved to be the election technique for glycan analysis due to its high sensi-
tivity, selectivity, and possibility to automation. First applications in this area were
run out using GC–MS still used today for the detection of monosaccharide composi-
tion.119,120 Laborious protocols for oligosaccharides derivatization, scarce volatility
of bigger molecules determined a significant expansion of soft ionization techniques
such as fast atom bombardment, electrospray ionization, and matrix-assisted laser
desorption/ionization in glycosylation analysis. Additionally, since ESI and MALDI
methods typically do not fragment the precursor ions and can hyphenate with LC
online and offline, respectively, they are well suited for analyzing complex mixtures.
MALDI and ESI MS analyses of the recovered glycopeptides are very sensitive and
rapid means to generate a glycan profile for individual glycosylation sites in pro-
teins.121 Almost all kinds of N-linked, O-linked, and GPI-anchored peptides have
been purified and characterized by MALDI or ESI tandem MS.22,98,115 Even better
results can be achieved by using techniques allowing to achieve high mass accu-
racy in the MS and MS/MS mode such high-resolution mass spectrometers like quaÂ�
drupole time of flight (Q-TOF), ion trap Fourier transform ion cyclotron resonance,
or ion trap Orbitrap instruments. A more straightforward and detailed assignment
of glycan species and structure elucidation122 can be reached thanks to these tech-
niques.123 The ion-trap-type instruments offer MS capabilities, sometimes useful
for detailed analysis of glycans.
There are two strategies of characterizing glycans using MS. In the first strat-
egy, glycans are cleaved from the glycoprotein chemically or enzymatically and
subjected to purification before mass spectral analysis.124,125 This approach reveals
scarce results when more than one glycosylation site is present on the glycoprotein,
because it does not permit to correlate glycan composition with the different attach-
ment sites. The second approach characterizes glycopeptides after subjecting the
glycoprotein to proteolysis. In this strategy, called also “glycosylation site-specific
analysis,” the glycan composition can be correlated to the attachment site on glyco-
peptide, due to the presence of the peptide portion on each glycopeptides. In this kind
of studies, N-linked glycans have received more attention compared with O-linked
glycans, probably because the former contain a consensus sequence easily recog-
nized for enzymatic treatments. However, glycopeptides analysis can also be applied
to O-linked glycopeptides. In this chapter, the protocol proposed is more suitable for

Lectin Protease
Extraction
affinity digestion
Complex sample Glycoproteins Glycopeptides +

Peptides
LC-MS/MS
1. PNGase F MALDI TOF Glycopeptides
2. MALDI HILIC
TOF
Figure 20.4 Strategy showing glyco-capture procedure for the identification of glycoproteins.
N-linked glycopeptides analysis. A basic approach involves the use of different forms
of proteolytic digestion, combined with the use of affinity enrichment material in
microcolumns and sensitive MS detection and characterization. The digestion proto-
col can be carried out on complex mixtures composed of proteins and glycoproteins
or can be performed on simpler sample mixtures where a previous purification at
proteins level has been carried out. Then a variant to a basic approach consists of a
selective capture of glycoproteins by means of gel electrophoresis or affinity chro-
matography (Figure 20.4).
For a protein-specific analysis, to isolate a broad range of glycoproteins, columns
having immobilized lectins are employed. Some examples of lectins include con-
canavalin A (ConA), wheat germ agglutinin and jacalin that can be used alone in
a single affinity column or mixed in different columns in series or combined to get
better coverage.126,127
After protein enrichment, it can be possible to isolate glycoproteins by SDS-PAGE
and perform an in-gel digestion or to perform directly an in-solution digestion on the
enriched fraction. Peptides and glycopeptides are obtained and a further enrichment
for glycosylated peptides can be made by using ZIC–HILIC microcolumns. The
relative abundance of glycopeptides was significantly increased following enrich-
ment by HILIC. It was observed for a standard protein as fetuin that using reverse
phase SPE for purification generated a spectrum dominated by nonglycosylated pep-
tides while, in comparison, SPE with HILIC resulted in a spectrum full of glyco-
sylated peptides. Moreover, for different proteins a different mass spectrum profile
was obtained after HILIC, but, in all cases, the higher mass region was enriched by
glycosylated peptides. An example is illustrated in Figure 20.5 where MALDI TOF
spectra of the enriched portion of different protein digests (fetuin, α-glycoprotein,
and RNase B) are shown.
Small diversions from this protocol have been reported and include lectin-
mediated affinity capture at the peptide level116 and the further separation of glyco-
peptides by strong cation exchange.115 Figure 20.6 shows as example a comparison
between spectra resulting from the enriched fraction of α-1-glycoprotein tryptic

Relative intensity Fetuin
α-glycoprotein
RNaseB
1,000 5,000 10,000

(m/z)
Figure 20.5 MALDI TOF mass spectra obtained after HILIC purification on different
protein tryptic digests.
(A) HILIC
(B) SNA
Relative intensity
(C) conA
(D) UE
A
(E) Jacalin
(F) WGA
1,000 5,000 10,000
Figure 20.6 MALDI TOF MS spectra obtained from an α-1-glycoprotein digest followed
by (A) HILIC purification, (B) Sambucus nigra lectin (SNA) purification, (C) Concanavalin
(ConA) purification, (D) Urex europaeus (UE) purification, (E) Jacalin purification, and
(F) wheat germ agglutinin (WGA) purification.
digest by HILIC and different lectins microcolumns. As seen, similar results are
obtained, even if HILIC has a broader versatility also on different proteins.
The recovered glycopeptides can be directly analyzed using either MALDI MS/
MS or LC–ESI–MS/MS or can be subject to an enzymatic deglycosylated prior to
MS analysis.
The common enzyme used for deglycosylation is N-glycosidase F (Glycopeptidase F,
GPase F) that removes N-linked oligosaccharides from the polypeptide backbone.
The cleavage occurs between the innermost residue of the oligosaccharide and the
asparagine residue to which the oligosaccharide is linked. In this enzymatic reaction,

After HILIC before PNGase F

Intensity
PNGase F
Δm
+0.984 Da
N D +
(Δ) (Δ)
VV HAVE VAL AT FN AE S DG S YL QL VE IS R
RP TG EV YDI EI DT LET TC HV LD PT PL ADCS VR
Intensity
LC PDCP LL APL DDSR

After PNGase F (1 mU in NH4HCO3, overnight)
1000 3000 5000 7000 9000
Figure 20.7 MALDI TOF MS spectra obtained before and after PNGase F digestion on
fetuin protein tryptic digest.
the asparagine is converted to an aspartate residue with a concomitant mass increase

of 0.98â•›Da. Figure 20.7 shows the results obtained after PNGase F digestion on
fetuin protein digest. One drawback of this approach is that it does not distinguish
between enzymatic asparagine-to-aspartate conversion and other causes of deami-
dation, in vivo or in vitro. For this reason, in the effort to unambiguous localize
glycosylation sites it is opportune tagging the site using 18O isotope labeling dur-
ing PGNase digestion115,116 or to perform enzymatic deglycosylation using endo-δ-
N-acetylglucosaminidases, which cleave the glycosidic bond between two proximal
GlcNAc residues in the chitobiose core leaving one GlcNAc residue attached to the
peptide.
If this approach is followed in conjunction with mass spectrometric analysis of
glycopeptides,126 it is possible to observe in the mass spectrum an oxonium ion of
GlcNAc at m/z 204.08 a marker indicating the presence of an N-glycan attachment
site in a peptide.
The oxonium ion helps in identifying by a simple full scan analysis whether or
not the peaks in the mass spectrum correspond to glycopeptides. After verification,
the glycopeptides should be characterized in the peptide portions and in the glycan
portions and the assignment should be validated with MS/MS data.
If a PNGase F digestion has been performed instead, a 0.98â•›Da mass shift is
observed in the spectrum. However, when glycopeptides are subjected to collision-
induced dissociation (CID), other marker ions at m/z 163 (Hex+), m/z 204 (HexNAc+),
m/z 366 (Hex-HexNAc+), and m/z 292 (NeuNAc+) can be observed.128,129 These ions
can be used as diagnostic ions to verify the presence of glycopeptides in the mass

spectrum. When the peptide is unknown, a rapid way to identify marker ions is to
look for a pair of peaks that are distant 120 or 266â•›Da. This pair of peaks corresponds
to [Peptide + Glc-NAc + H]+ and [Peptide + C2H2NHAc]+ or [Peptide + GlcNAc +
Fucose + H]+ and [Peptide + C2H2NHAc]+, respectively.
After determining which ions correspond to glycopeptides in the mass spectrum,
it is necessary to proceed with the compositional assignment of the peptide por-
tion usually separately from the glycan portion recurring to MS/MS data analysis.
Usually MS/MS data obtained by the CID of glycopeptides are dominated by the
glycosidic cleavage of the sugar moieties and y- and b-type ions from the peptide
portion. One way to recognize the peptide is by identifying the glycan composition
from MS/MS experiments first and then deducting this mass from the precursor ions
so that the remaining mass can be used to obtain the peptide mass.
Finally, the glycopeptide compositions can be characterized using databases such
as GlycoPep DB (http://hexose.chem.ku.edu)130–132 or GlycoMod (http://us.expasy.
org/tools/glycomod/).133,134 To use these tools, the mass spectrum is converted into a
peak list and searched against these databases for possible matches.
20.8 Application to Biomarker Discovery

In a survey of the SWISS-PROT database results that more than 50% of proteins
are glycosylated. Biological fluids are a rich source of glycoproteins and are very
informative since N-glycosylated proteins arise from tissues and organs consti-
tuting useful biomarkers of their status. In fact, diseased tissues are expected to
release disease-specific proteins and the detection and characterization of such
disease-specific biomarkers is of great importance. Then glycan heterogeneity
and biological variations as well as glycan-related differences between healthy
and diseased individuals can be discovered in a biological fluid such as serum as
a diagnostic and preventing tool. The main problem in glycoproteins/peptides
analysis in serum is their very low concentrations. In fact, 99% of serum protein
content is constituted by albumin, transferrins, immunoglobulins, and comple-
ment factors represent, while the remaining 1% is represented by lower abundance
circulatory proteins, shed proteins, and apoptotic and necrotic cells.134 It is neces-
sary to isolate glycoproteins from the most abundant background serum proteins
prior to protein and peptide analysis by MS.
The proposed strategy135 here uses different immobilized lectins for glycoproteins
extraction and HILIC resin for the enrichment of glycopeptides prior to MALDI
MS and LC–ESI–MS/MS. HILIC material was packed into GELoader tips (Figure
20.8) in microcolumns previously for improving sensitivity and sequence coverage
for proteins. Enzymatic release of N-linked glycans with peptide N-glycosidase F
was used for the determination of glycosylation sites. A protocol for HILIC prepara-
tion136 can be easily reproduced for serum samples:
1. ZIC–HILIC chromatographic media, (ZIC–HILIC, silica 10â•›μm, 200â•›Å,

Sequant AB, Umeå, Sweden), suspended in acetonitrile or methanol.
2. C8 StageTips made from either GELoader tips or 10â•›μL disposable syringe tips.
3. Disposable 1â•›mL syringe, fitted to the GELoader with a cut down 200â•›μL tip.

Packing
Packing of microcolumn
Washing resin
Washing of resin with 80% ACN, 0.5% FA
Load sample in 80% ACN, 0.5% FA
Load sample
Washing with 80% ACN, 0.5% FA Washing sample
Elute sample in 0.5% FA

Elution
Figure 20.8 Illustration of all steps for HILIC SPE preparation.
4. HILIC wash: 80% acetonitrile, 19.5% water, 0.5% formic acid—can be

stored at 4°C for up to 1 week.
5. HILIC elute: 0.5% aqueous formic acid—can be stored at 4°C for up to
1 week.
6. 1–20â•›pmol of the digested sample is made up in 10–20â•›μL of HILIC wash
solution.
7. Prepare the HILIC microcolumns. Vortex the ZIC–HILIC bead–containing
solution and deposite a few microlitres of the resulting slurry into 10â•›μL of
HILIC wash solution that was loaded into a StageTip. The HILIC beads
are packed on top of the C8 plug by applying gentle air pressure with the
plastic syringe. The length of the column is dependent on the amount of
peptides you wish to analyze, with a 3–5â•›m m column sufficient for up to
20â•›pmol.
8. Clean the microcolumn with 15â•›μL of HILIC elution solution, flushing the
solution through with gentle pressure from the syringe.
9. Condition the column by flushing with 30â•›μL of HILIC wash solution.
10. Load the sample containing glycopeptides onto the column using the
syringe.
11. Wash unbound material from the column with 20–40â•›μL of HILIC wash
solution.
12. Elute the glycosylated peptides from the HILIC material with 7–15â•›μL of
HILIC elute solution—retain eluate.

13. Elute these glycopeptides with relatively hydrophobic peptides from the
plug with 3â•›μL of HILIC—wash solution—pool eluate.
14. Dry the glycopeptides down in a vacuum centrifuge and store until required
for mass spectrometric analysis.
A comparison between HILIC and lectin-based techniques on glycopeptides mix-

tures is also assessed. Four different sample preparation protocols (Figure 20.9) were
tested in terms of glycoprotein identification efficiency, N-linked glycan site deter-
mination, and reproducibility. The four glycopeptide enrichment strategies provided
a comparable number of identified glycoproteins (Figure 20.10). The main differ-
ence between the four strategies is in the number of identified glycopeptides. If very
complex samples are analyzed without a prepurification (Strategies A and B), a lower
number of glycosylation sites are recovered. The presence of very abundant non-
glycosylated serum proteins and their derived tryptic peptides exceeds the capacity
of the HILIC resin (Strategy B, Figure 20.9). In contrast, more informative results
were obtained when introducing an enrichment step at the glycoprotein level as well
as at the glycopeptide level. Many of the glycopeptides and N-glycosylation sites
were determined by both lectin (Strategy C, Figure 20.9) and HILIC (Strategy D,
Figure 20.9) and only few peptides derived from nonglycosylated human serum
albumin were detected.
These studies highlight that HILIC works very well on simple protein mixtures or
even single proteins, but an unambiguous characterization of peptide glycan struc-
tures at the proteome level such as in serum is still rather challenging. HILIC then
can be used as an effective and relatively simple tool for the targeted analysis of
Serum Serum Serum Serum
Tryptic digestion Tryptic digestion Lectin mix Lectin mix
Lectin mix HILIC Tryptic digestion Tryptic digestion
Deglycosylation Deglycosylation Lectin mix HILIC
MALDI/MS MALDI/MS Deglycosylation Deglycosylation

and/or and/or
LC-MS/MS LC-MS/MS
MALDI/MS MALDI/MS
and/or and/or
LC-MS/MS LC-MS/MS
(A) (B) (C) (D)
Figure 20.9 Purification methods including all preparative steps. The strategies are indi-
cated as (A) lectin only, (B) HILIC only, (C) lectin-lectin, and (D) lectin-HILIC. (Taken from
Calvano, C.D. et al., J. Proteomics, 71, 304, 2008. With permission from Elsevier.)

86
90 81
80
65
70
60 51
50 45 44
39
40 31
30
20
10
0
A B C D
Identified glycopeptides
Identified glycoproteins
Figure 20.10 Graphical comparison of the results obtained from the employed methods
for N-glycosylation analysis as described in Figure 20.9. (Taken from Calvano, C.D. et al.,
J. Proteomics, 71, 304, 2008. With permission from Elsevier.)
protein glycosylation, but further studies are necessary to assess if HILIC might
also have an important role to play in not only glycosite elucidation but also in the
compositional analysis.
References
1. Linden, J.C., Lawhead, C.L., J. Chromatogr. A, 1975, 105, 125.
2. Alpert, A.J., J. Chromatogr. A, 1990, 499, 177.
3. Rizzi, A., Retention and selectivity, in: Katz, E., Eksteen, R., Schoenmakers, P., Miller, N.
(Eds.), Handbook of HPLC, Marcel Dekker, New York, 1998, p. 1.
4. Caude, M., Jardy, A., Normal-phase liquid chromatography, in: Katz, E., Eksteen, R.,
Schoenmakers, P., Miller, N. (Eds.), Handbook of HPLC, Marcel Dekker, New York,
1998, p. 325.
5. Hemström, P., Irgum, K. J Sep. Sci., 2006, 29, 1784.
6. Strege, M.A., Stevenson, S., Lawrence, S.M. Anal. Chem., 2000, 72, 4629.
7. Risley, D.S., Strege, M.A., Anal. Chem., 2000, 72, 1736.
8. Sarg, B., Koutzamani, E., Helliger, W., Rundquist, I., Lindner, H.H., J. Biol. Chem.,
2002, 277, 39195.
9. Sarg, B., Helliger, W., Hoertnagl, B., Puschendorf, B., Lindner, H., Arch. Biochem.
Biophys., 1999, 372, 333.
10. Lindner, H., Sarg, B., Grunicke, H., Helliger, W., J. Cancer Res. Clin. Oncol., 1999,
125, 182.
11. Xu, R.N., Fan, L., Rieser, M.J., El-Shourbagy, T.A., J. Pharm. Biomed. Anal., 2007, 44, 342.
12. Naidong, W., J. Chromatogr. B, 2003, 796, 209.
13. Guo, Y., Gaiki, S., J. Chromatogr. A, 2005, 1074, 71.
14. Grumbach, E.S., Wagrowski-Diehl, D.M., Mazzeo, J.R., Alden, B., Iraneta, P.C., LCGC N.
Am., 2004, 22, 1010.

15. Shou, W.Z., Naidong, W., J. Chromatogr. B, 2005, 825, 186.

16. Shou, W., Chen, Y., Eerkes, A., Tang, Y. et al., Rapid Commun. Mass Spectrom., 2002,
16, 1613.
17. Gilar, M., Olivova, P., Daly, A.E., Gebler, J.C., Anal Chem., 2005, 77, 6426.
18. Boersema, P.J., Divecha, N., Heck, A.J.R., Mohammed, S., J. Proteome Res., 2007,
6, 937.
19. Wang, X.D., Li, W.Y., Rasmussen, H.T., J. Chromatogr. A, 2005, 1083, 58.
20. Weng, N., Shou, W., Addison, T., Maleki, S., Jiang, X., Rapid Commun. Mass Spectrom.,
2002, 16, 1965.
21. Boersema, P.J., Mohammed, S., Heck, A.J.R., Anal. Bioanal. Chem., 2008, 391, 151.
22. Hagglund, P., Bunkenborg, J., Elortza, F., Jensen, O.N., Roepstorff, P., J. Proteome Res.,
2004, 3, 556.
23. McNulty, D.E., Annan, R.S., in: 55th ASMS Conference on Mass Spectrometry,
Indianapolis, IN, June 3–7, 2007.
24. Jensen, O.N., Nat. Rev. Mol. Cell. Biol., 2006, 7, 391.
25. Bertozzi, C.R., Kiessling, L.L., Science, 2001, 291, 2357.
26. Rudd, P.M., Elliott, T., Cresswell, P., Wilson, I.A., Dwek, R.A., Science, 2001, 291, 2370.
27. Gewinner, C., Hart, G., Zachara, N., Cole, R., Beisenherz-Huss, C., Groner, B., J. Biol.
Chem., 2004, 279, 3563.
28. Alikhani, Z., Alikhani, M., Boyd, C.M., Nagao, K., Trackman, P.C., Graves, D.T.,
J. Biol. Chem., 2005, 280, 12087.
29. Phan, U.T., Waldron, T.T., Springer, T.A., Nat. Immunol., 2006, 7, 883.
30. Partridge, E.A., Le Roy, C., Di Guglielmo, G.M., Pawling, J., Cheung, P., Granovsky,
M., Nabi, I.R., Wrana, J.L., Dennis, J.W., Science, 2004, 306, 120.
31. Durand, G., Seta, N., Clin. Chem., 2000, 46, 795.
32. Bakry, N., Kamata, Y., Simpson, L.L., J. Pharmacol. Exp. Ther., 1991, 258, 830.
33. Becker, J.W., Reeke, G.N., Jr., Wang, J.L., Cunningham, B.A., Edelman, G.M., J. Biol.
Chem., 1975, 250, 1513.
34. Yang, Z., Hancock, W.S., J. Chromatogr. A, 2004, 1053, 79.
35. Turner, G.A., Clin. Chim. Acta, 1992, 208, 149.
36. Hellwage, J., Kuhn, S., Zipfel, P.F., Biochem. J., 1997, 326, 321.
37. Junnikkala, S., Jokiranta, T.S., Friese, M.A., Jarva, H., Zipfel, P.F., Meri, S., J. Immunol.,
2000, 164, 6075.
38. Yang, Z., Harris, L.E., Palmer-Toy, D.E., Hancock, W.S., Clin. Chem., 2006, 52, 1897.
39. Zhao, J., Patwa, T., Qiu, W., Shedden, K., Hinderer, R., Misek, D., Anderson, M.,
Simeone, D., Lubman, D., J. Proteome Res., 2007, 6, 1864.
40. Donate, F., Juarez, J.C., Guan, X., Shipulina, N.V., Plunkett, M.L., Tel-Tsur, Z., Shaw, D.E.,
Morgan, W.T., Mazar, A.P., Cancer Res., 2004, 64, 5812.
41. Olsson, A.-K., Larsson, H., Dixelius, J., Johansson, I., Lee, C., Oellig, C., Bjork, I.,
Claesson-Welsh, L., Cancer Res., 2004, 64, 599.
42. Aebersold, R., J. Proteome Res., 2005, 4, 1104.
43. Kannagi, R., Izawa, M., Koike, T., Miyazaki, K., Kimura, N., Cancer Sci., 2004, 95, 377.
44. Varki, A. (Eds.), Essentials of Glycobiology, Cold Spring Harbour Press, Cold Spring
Harbour, NY, 1999.
45. Voisin, S., Houliston, S.R., Kelly, J., Brisson, J.R., Watson, D., Bardy, S.L., Jarrell, K.F.,
Logan, S.M., J. Biol. Chem., 2005, 280, 16586.
46. Yamaguchi, Y., Takizawa, T., Kato, K., Arata Y., Shimada I., J. Biomol. NMR, 2000, 18, 357.
47. Zhuang, Z., Starkey, J.A., Mechref, Y., Novotny, M.V., Jacobson, S.C., Anal. Chem.,
2007, 79, 7170.
48. Priem, B., Gitti, R., Bush, C.A., Gross, K.C., Plant Physiol., 1993, 102, 445.
49. Maslen, S., Sadowski, P., Adam, A., Lilley, K., Stephens E., Anal. Chem., 2006, 78, 8491.

50. Hashii, N., Kawasaki, N., Itoh, S., Hyuga, M., Kawanishi, T., Hayakawa, T., Proteomics,
2005, 5, 4665.
51. Siluka, D., Kima, H.S., Colea, T., Wainera, I.W., J. Pharm. Biomed. Anal., 2008, 48, 960.
52. Liu, Z., Mutlib, A.E., Wang, J., Talaat, R.E., Rapid Commun. Mass Spectrom., 2008,
22, 3434.
53. Jauregui, O., Sierra, A.Y., Carrasco, P., Gratacos, E., Hegardt, F.G., Casals, N., Anal.
Chim. Acta, 2007, 599, 1.
54. Raman, R., Raguram, S., Venkataraman, G., Paulson, J.C., Sasisekharan, R., Nat.
Methods, 2005, 2, 817.
55. Lewandrowski, U., Moebius, J., Walter, U., Sickmann, A., Mol. Cell Proteomics, 2006,
5, 226.
56. Wada, Y., Tajiri, M., Yoshida, S., Anal. Chem., 2004, 76, 6560.
57. Tajiri, M., Yoshida, S., Wada, Y., Glycobiology, 2005, 15, 1332.
58. Zhang, H., Li, X.J., Martin, D.B., Aebersold, R., Nat. Biotechnol., 2003, 21, 660.
59. Ghosh, D., Krokhin, O., Antonovici, M., Ens, W., Standing, K.G., Beavis, R.C., Wilkins,
J.A., J. Proteome Res., 2004, 3, 841.
60. Kakehi, K., Kinoshita, M., in: N.G. Karlsson and N.H. Packer (Eds.), Glycomics:
Methods and Protocols, Methods in Molecular Biology, Springer, 2009, vol. 534, 1.
61. Tomana, M., Niedermeier, W., Mestecky, J., Schrohenloher, R.E., Porch, S., Anal.
Biochem., 1976, 72, 389.
62. Kubota, K., Sato, Y., Suzuki, Y., Goto-Inoue, N., Toda, T., Suzuki, M., Hisanaga, S.,
Suzuki, A., Endo, T., Anal. Chem., 2008, 80, 3693.
63. Nawarak, J., Phutrakul, S., Chen, S.T., J. Proteome Res., 2004, 3, 383.
64. Uematsu, R., Furukawa, J., Nakagawa, H., Shinohara, Y., Deguchi, K., Monde, K.,
Nishimura, S., Mol. Cell Proteomics, 2005, 4, 1977.
65. Qiu, R., Regnier, F.E., Anal. Chem., 2005, 77, 2802.
66. Gabius, H.J. et al., Biochim. Biophys. Acta, 2002, 1572, 165.
67. Wang, Y., Wu, S.L., Hancock, W.S., Biotechnol. Prog., 2006, 22, 873.
68. Drake, R.R., Schwegler, E.E., Malik, G., Diaz, J., Block, T., Mehta, A., Semmes, O.,
Mol. Cell Proteomics, 2006, 5.10, 1957.
69. Wang, L., Li, F., Sun, W., Wu, S., Wang, X., Zhang, L., Zheng, D., Wang, J., Gao, Y.,
Mol. Cell Proteomics, 2006, 5.3, 560.
70. Monzo, A., Bonn, G.K., Guttman, A., Trends Anal. Chem., 2007, 26, 423.
71. Larsen, M.R., Cordwell, S.J., Roepstorff, P., Proteomics, 2002, 2, 1277.
72. Gerard, C., Methods Enzymol., 1990, 182, 529.
73. Brittain, S.M., Ficarro, S.B., Brock, A., Peters, E.C., Nat. Biotechnol., 2005, 23, 463.
74. Mirzaei, H., Regnier, F., Anal. Chem., 2005, 77, 2386.
75. Zhang, Q., Tang, N., Brock, J.W.C., Mottaz, H.M., Ames, J.M., Baynes, J.W., Smith, R.D.,
Metz, T.O., J. Proteome Res., 2007, 6, 2323.
76. Nice, E.C., Rothacker, J., Weinstock, J., Lim, L., Catimel, B., J.Chromatogr. A, 2007,
1168, 190.
77. Madera, M., Mechref, Y., Novotny, M.V., Anal. Chem., 2005, 77, 4081.
78. Qiu, R., Regnier, F.E., Anal. Chem., 2005, 77, 7225.
79. Thaysen-Andersen, M., Hojrup, P., Am. Biotechnol. Lab., 2006, 24, 14.
80. Kaji, H., Saito, H., Yamauchi, Y., Shinkawa, T., Taoka, M., Hirabayashi, J., Kasai, K.,
Takahashi, N., Isobe, T., Nat. Biotechnol., 2003, 21, 667.
81. Callesen, A.K., Mohammed, S., Bunkenborg, J., Kruse, T.A., Cold, S., Mogensen,
O., Christensen, R., Vach, W., Jørgensen, P.E., Jensen, O.N., Rapid Commun. Mass
Spectrom., 2005, 19, 1578.
82. Aresta, A., Calvano, C.D., Palmisano, F., Zambonin, C.G., Monaco, A., Tommasi, S.,
Pilato, B., Paradiso, A., J. Pharm. Biomed. Anal., 2008, 46, 157.

83. Larsen, M.R., Jensen, S.S., Jakobsen, L.A., Heegaard, N.H., Mol. Cell Proteomics,
2007, 6, 1778.
84. Churms, S.C., J. Chromatogr. A, 1996, 720, 75.
85. Kieliszewski, M.J., Oneill, M., Leykam, J., Orlando, R., J. Biol. Chem., 1995, 270, 2541.
86. Zhang, J., Wang, D.I., J. Chromatogr. B, 1998, 712, 73.
87. Gregor, H.P., Collins, F.C., Pope, M., J. Colloid Sci., 1951, 6, 304.
88. Rückert, H., Samuelson, O., Sven. Kem. Tidskr., 1954, 66, 337.
90. Guo, Y., Huang, A.H., J. Pharm. Biomed. Anal., 2003, 31, 1191.
91. Kind, T., Tolstikov, V., Fiehn, O., Weiss, R.H., Anal. Biochem., 2007, 363, 185.
92. Zhu, B.-Y., Mant, C.T., Hodges, R.S., J. Chromatogr. A, 1991, 548, 13.
93. Lindner, H., Sarg, B., Meraner, C., Helliger, W., J. Chromatogr. A, 1996, 743,137.
94. Ytterberg, J.A., Ogorzalek-Loo, R.R., Boontheung, P., Loo, J.A., in: 55th ASMS
Conference on Mass Spectrometry, Indianapolis, IN, June 3–7, 2007.
95. Alpert, A.J., Anal. Chem., 2008, 80, 62.
97. Tomiya, N., Awaya, J., Kurono, M., Endo, S., Arata, Y., Takahashi, N., Anal. Biochem.,
1988, 171, 73–90.
98. Omaetxebarria, M.J., Hagglund, P., Elortza, F., Hooper, N.M., Arizmendi, J.M., Jensen,
O.N., Anal Chem., 2006,78, 3335.
99. Callesen, A.K., Vach, W., Jørgensen, P.E., Cold, S., Tan, Q., Depont Christensen, R.,
Mogensen, O., Kruse, T.A., Jensen, O.N., Madsen, J.S., J. Proteome Res., 2008, 4, 1419.
100. Takegawa, Y., Deguchi, K., Ito, H., Keira, T., Nakagawa, H., Nishimura, S., J. Sep. Sci.,
2006, 29, 2533.
101. Alpert, A.J., Shukla, M., Shukla, A.K., Zieske, L.R., Yuen, S.W., Ferguson, M.A.,
Mehlert, A., Pauly, M., Orlando, R., J. Chromatogr. A, 1994, 676, 191.
102. Thomsson, K.A., Karlsson, N.G., Hansson, G.C., J. Chromatogr. A, 1999, 854, 131.
103. Thomsson, K.A., Karlsson, H., Hansson, G.C., Anal. Chem., 2000, 72, 4543.
104. Charlwood, J., Birrell, H., Bouvier, E.S., Langridge, J., Camilleri, P., Anal. Chem., 2000,
72, 1469.
105. Saba, J.A., Shen, X., Jamieson, J.C., Perreault, H., J. Mass Spectrom., 2001, 36, 563.
106. Saba, J.A., Kunkel, J.P., Jan, D.C., Ens, W.E., Standing, K.G., Butler, M., Jamieson,
J.C., Perreault, H., Anal. Biochem., 2002, 305, 16.
107. Wuhrer, M., Koeleman, C.A.M., Deelder, A.M., Hokke, C.H., Anal. Chem., 2004, 76, 833.
108. Wuhrer, M., Koeleman, C.A.M., Hokke, C.H., Deelder, A.M., Int. J. Mass Spectrom.,
2004, 232, 51.
109. Geyer, H., Wuhrer, M., Resemann, A., Geyer, R., J. Biol. Chem., 2005, 280, 40731.
110. Karlsson, G., Winge, S., Sandberg, H., J. Chromatogr. A, 2005, 1092, 246.
111. Risley, D.S., Yang, W.Q., Peterson, J.A., J. Sep. Sci., 2006, 29, 256.
112. Beilmann, B., Langguth, P., Häusler, H., Grass, P., J. Chromatogr. A, 2006, 1107, 204.
113. Peru, K.M., Kuchta, S.L., Headley, J.V., Cessna, A.J., J. Chromatogr. A, 2006, 1107, 152.
114. Takegawa, Y., Deguchi, K., Keira, T., Ito, H., Nakagawa, H., Nishimura, S.-I.,
J. Chromatogr. A, 2006, 1113, 177.
115. Hagglund, P., Matthiesen, R., Elortza, F., Hojrup, P., Roepstorff, P., Jensen, O.N.,
Bunkenborg, J., J. Proteome Res., 2007, 6, 3021.
116. Kaji, H., Yamauchi, Y., Takahashi, N., Isobe, T., Nat. Protoc., 2006, 1, 3019.
117. Thaysen-Andersen, M., Thogersen, I.B., Nielsen, H.J., Lademann, U., Brunner, N.,
Enghild, J.J., Hojrup, P., Mol. Cell Proteomics, 2007, 6, 638.
118. Kaji, H., Kamiie, J., Kawakami, H., Kido, K., Yamauchi, Y., Shinkawa, T., Taoka, M.,
Takahashi, N., Isobe, T., Mol. Cell Proteomics, 2007, 6, 2100.
119. DeHoffman, E., Stroobant, V., Oligosaccharides in Mass Spectrometry: Principles and
Applications, 2nd edn., John Wiley & Sons Ltd., West Sussex, U.K., 2001, pp. 292–320.

120. Montilla, A., van de Lagemaat, J., Olano, A., del Castillo, M.D., Chromatographia,
2006, 63, 453.
121. Mortz, E., Sareneva, T., Julkunen, I., Roepstorff, P., J. Mass Spectrom., 1996, 31, 1109.
122. Harvey, D.J., Proteomics, 2005, 5, 1774.
123. Harvey, D.J., Expert Rev. Proteomics, 2005, 2, 87.
124. Mechref, Y., Novotny, M.V., Chem. Rev., 2002, 102, 321.
125. Zaia, J., Mass Spectrom. Rev., 2004, 23, 161.
126. Wuhrer, M., Catalina, M.I., Deelder, A.M., Hokke, C.H., J. Chromatogr. B, 2007, 849, 115.
127. Yamamoto, K., Tsuji T., Osawa, T., Mol. Biotechnol., 1995, 3, 25.
128. Conboy, J.J., Henion, J.D., J. Am. Soc. Mass Spectrom., 1992, 3, 804.
129. Huddleston, M.J., Bean M.F., Carr, S.A., Anal. Chem., 1993, 65, 877.
130. Dalpathado, D.S., Irungu, J., Go, E.P., Butnev, V.Y., Norton, K., Bousfield, G.R.,
Desaire, H., Biochemistry, 2006, 45, 8665.
131. Go, E.P., Rebecchi, K.R., Dalpathado, D.S., Bandu, M.L., Zhang, Y., Desaire, H., Anal.
Chem., 2007, 79, 1708.
132. Bykova, N.V., Rampitsch, C., Krokhin, O., Standing, K.G., Ens, W., Anal. Chem., 2006,
78, 1093.
133. Larsen, M.R., Hojrup, P., Roepstorff, P., Mol. Cell Proteomics, 2005, 4, 107.
134. Roberts, G., Keyser, J.W., Baum, M., Br. J. Surg., 1975, 62, 816.
135. Calvano, C.D., Zambonin, C.G., Jensen, O.N., J. Proteomics, 2008, 71, 304.
136. Selby, D.S., Larsen, M.R., Calvano, C.D., Jensen, O.N., Glycosylation. Methods in
Molecular Biology: Functional Proteomics, 2008, 484, 263.

21 Development and
Application of Methods
for Separation of
Carbohydrates by
Goran Karlsson
Contents
21.1 Introduction................................................................................................... 492
21.1.1 Carbohydrates.................................................................................... 492
21.1.2 Techniques for Separation and Detection of Carbohydrates............. 492
21.1.3 HILIC in the Separation of Carbohydrates....................................... 494
21.2 Development and Optimization of HILIC Methods for Separating
and Analyzing Carbohydrates....................................................................... 498
21.2.1 Use of Different Stationary Phases.................................................... 498
21.2.2 Mobile-Phase Composition................................................................ 499
21.2.3 Temperature and Other Variables...................................................... 505
21.2.4 Sample Preparation Strategy............................................................. 508
21.2.5 Detection of Carbohydrates...............................................................509
21.2.6 Short Outline for the Development of an HILIC Method................. 511
21.3 HILIC Applications for the Separation of Carbohydrates............................. 512
21.3.1 HILIC Separation of Mono-, Di-, and Oligosaccharides by
Cyclodextrin Columns in Combination with UV and RI
Detection.....................................................................................512
21.3.2 HILIC Separation by an Amide-80 Column and MS Analysis
of Carbohydrates from Plants............................................................ 513
21.3.3 HILIC Separation by a ZIC-HILIC Column and MS Analysis
of Tryptic Glycopeptides from Human Immunoglobulin................. 514
491

21.3.4 Analysis of 2-Aminobenzoic Acid-Derivatized Oligosaccharides

by an NH2 Column and Fluorescence Detection............................... 515
21.3.5 Separation of Glucosinolates by a ZIC-HILIC Column
and Detection by Absorbance............................................................ 515
21.3.6 Separation of GAGs by an Amide-80 Column Combined
with MS Analysis.............................................................................. 517
References............................................................................................................... 520
21.1 Introduction
21.1.1 Carbohydrates
Carbohydrates (sugars or saccharides) are a large group of organic polar compounds
that contain aldehydes or ketones, or their derivatives, which have many hydroxyl
groups in their chemical structure.1 Monosaccharides are the simplest type of car-
bohydrates, which can assume cyclic forms (hemiketals or hemiacetals) and which
can also react with other carbohydrates to form di-, oligo-, and polysaccharides.
Carbohydrates are the most abundant class of biomolecules, having diverse functions
such as the storage of energy (e.g., starch in plants and glycogen in animals) or the
formation of structural components (e.g., cellulose in plants and chitin and cartilage
in animals). Additionally, carbohydrates and their derivatives play major roles in
biochemical systems, as in the vital parts of proteins (including enzymes, antibodies,
and hormones), and are thus crucial for the functioning of physiological processes
such as the immune system, fertilization, pathogenesis, blood clotting, and develop-
ment. The inappropriate glycosylation of proteins, for example, can lead to a reduced
biological activity, decreased half-life in circulation, or unwanted immunogenicity.
21.1.2 Techniques for Separation and Detection of Carbohydrates

For more than 30 years, carbohydrates have been separated by aqueous normal-phase
liquid chromatography (NPLC), using amino-propyl columns with about 75% ace-
tonitrile and 25% water for isocratic elution.2 Thus, these methods can be included
in the liquid chromatography group now denoted HILIC (hydrophilic interaction
liquid chromatography), which is a variant of NPLC. In contrast to conventional
NPLC, HILIC uses water for elution in a mostly organic mobile phase. NPLC,
including HILIC, in contrast to reversed-phase liquid chromatography (RPLC),
which is the most used liquid chromatography method, provides good retention of
highly polar compounds, both charged and uncharged, including carbohydrates. The
elution order in HILIC is roughly reversed when compared with that of RPLC,3
and, thus, HILIC gives strong retention of compounds that are weakly or not at all
bound in RPLC, which allows them to be separated by this technique. HILIC is used
primarily for the separation of very polar compounds with a negative value of log
octanol–water partition coefficient (log P).
The direct detection of carbohydrates by UV absorbance or fluorescence is usu-
ally not possible, and, during the 1970s and 1980s, the universal refractive index
(RI) detector was often used. Later, this rather insensitive detector was replaced by
evaporative light scattering detection (ELSD), mass spectrometry (MS), and other

Development and Application of Methods for Separation 493
detection techniques. Gas chromatography (GC) of trimethylsilyl (TMS) derivatized

monosaccharides was introduced more than 50 years ago4 and later was successfully
combined with mass spectrometry (GC-MS),5 which was a major step in determin-
ing the structure and concentration of carbohydrates. Nuclear magnetic resonance
(NMR) is a powerful technique for determining carbohydrate structure and provides
complimentary information to the MS data. Another approach for carbohydrate
detection uses derivatization of the reducing end of the mono-, di-, or oligosaccha-
ride, where a hydrophobic chromophore6 or fluorophore7 is attached, which is fol-
lowed with RPLC for the separation and detection by absorbance or fluorescence,
respectively. Derivatization of carbohydrates for fluorescence detection is especially
useful for detecting carbohydrate derivates with a very high sensitivity.
Ligand-exchange chromatography (LEC) has been successfully used for the sepa-
ration of monosaccharides.8 With this method, hydroxyl groups of carbohydrates
form a complex with multivalent metal ions (such as Pb2+), and separation follows on
a cation-exchange chromatography column. Additionally, ordinary anion-exchange
chromatography can be used, for example, for sialic acid9 and glucuronic acid10 con-
taining carbohydrates. Boronate easily forms esters with cis-diols in carbohydrates,
which was used by Bourne more than 40 years ago to enhance the separation of
carbohydrates by paper chromatography.11 Boronate affinity chromatography, with
boronate covalently coupled to the gel matrix, was later, for example, developed
for the separation between hemoglobin and a glycosylated form of hemoglobin.12
Fluorophore-assisted carbohydrate electrophoresis, where a charged fluorophore is
covalently bound to the carbohydrate, is used for the separation of simple and com-
plex carbohydrates.13 Other electrophoretic techniques for the separation of carbohy-
drates, such as capillary electrophoresis (CE)14 and capillary electrochromatography
(CEC),15 which achieve high efficiency and resolution, are powerful alternatives
to liquid chromatography methods for analyzing complex carbohydrates derived
from biological samples. Many other techniques have been used for the separation
of different types of simple and complex carbohydrates, including affinity chroma-
tography using columns coupled with a suitable lectin (e.g., concavalin A, which
has a high affinity for mannose-containing carbohydrates),16 and supercritical fluid
chromatography.17
High-performance anion-exchange chromatography (HPAEC) is a high-resolution
technique for analyzing carbohydrates that uses a strongly alkaline mobile phase
(pH about 13), where hydroxyl groups of the carbohydrates are ionized and nega-
tively charged and thus can be bound to an anion-exchange chromatography column,
and the elution is with a buffer of increasing ionic strength. For example, a sodium
acetate gradient may be used for the elution of complex carbohydrates, or in an
isocratic mode can be used to separate mono- and disaccharides. HPAEC is often
combined with pulsed amperometric detection (PAD) for the very sensitive detec-
tion of carbohydrates.18 This technique is now widely used and considered to have
excellent separating and detection capabilities for analyzing carbohydrates. Besides
the high resolution and selectivity of HPAEC-PAD chromatography, this system has
an advantage, compared to fluorescence-labeling techniques, for example, in that
derivatization of the carbohydrates is not required for PAD detection. Nevertheless,
PAD requires relatively more maintenance (particularly for the gold membrane in

the working electrode) to avoid baseline drift and other distortions of the chromatogram.
A comprehensive description and evaluation of all useful methods and techniques
for separation and analysis of carbohydrates is beyond the scope of this book. For
review articles that cover some of the methods and techniques, see Cataldi et al.
(HPAEC-PAD),19 Campa et al. (CE),20 Klampfl (CEC-MS),21 and Zaia (MS).22
21.1.3 HILIC in the Separation of Carbohydrates

HILIC was defined by Alpert in 19903 as a variant of normal-phase chromatog-
raphy, where a hydrophilic stationary phase is used in combination with a mostly
organic aqueous mobile phase, often acetonitrile (sometimes propanol or acetone),
and elution is usually performed by increasing the polarity of the mobile phase by
increasing the water concentration. Elution by increasing the ionic strength can also
be used, especially when running ion-exchange chromatography (IEX) columns in
HILIC mode (i.e., using about 60%–90% organics), according to Alpert.3 Aqueous
NPLC, now defined as HILIC, has been used since 1975 for separating simple and
complex carbohydrates.2 In those days, RI detectors were often used in analytical
chromatography, so that gradient elution was not possible. In addition to acetonitrile
and water, a volatile buffer, such as ammonium acetate, which is compatible with an
electrospray ionization (ESI) interface MS, is often included in the mobile phase to
avoid electrostatic interaction with ionized silanols in the silica solid phase. The high
concentration of acetonitrile facilitates the detection by ESI MS, a technique which
is widely used today, and which is used for numerous purposes due to high specific-
ity and sensitivity. HILIC can be readily combined with an ESI interface in a liquid
chromatography-mass spectrometry (LC-MS) system. This is in contrast to HPAEC
systems, where some kind of post-column ion-suppression device is required before
the ESI interface,23 and in contrast to ordinary NPLC that uses 100% organics, which
does not facilitate ionization.
In HILIC, the retention depends on the distribution of the analytes between a
partially immobilized aqueous layer, bound to the gel matrix surface, and the mobile
phase. In addition, hydrogen bonding and electrostatic interactions between the ana-
lyte and the stationary phase are also believed to be more important than comparable
mixed-mode retention mechanisms in other liquid chromatography techniques, for
example, RPLC.3,24,25 This is in contrast to conventional NPLC, where mixtures of
organics, without water, are used in the mobile phase, and where the separation
mainly depends on surface adsorption. A higher number of monosaccharides in
more complex carbohydrates usually give an increased retention in HILIC, for exam-
ple, the disaccharide sucrose elutes after its monosaccharides glucose and fructose
(Figure 21.1). Fucose, lacking one hydroxyl group, is thus equivalent to 6-deoxy-l-
galactose and has a lower retention compared to corresponding monosaccharides, for
instance, galactose and glucose (Figure 21.2), which both have an intact 6′ hydroxyl
group. In addition, aminosugars (e.g., N-acetyl-d-glucosamine) have a stronger reten-
tion on HILIC columns, compared to corresponding nonaminosugars.24
Using water in HILIC mobile phases is advantageous because it increases the
solubility of polar compounds, and the water content does not need to be completely
controlled, as is the case in ordinary NPLC. An increased interest in HILIC has been

600.00
Sucrose
400.00
200.00
0.00
(A) 0.00 2.00 4.00 6.00 8.00 10.00
600.00
Light scattering (mV)
Glucose
400.00
200.00
0.00
(B) 0.00 2.00 4.00 6.00 8.00 10.00
600.00
400.00
Fructose
200.00
0.00
0.00 2.00 4.00 6.00 8.00 10.00
(C) Time (min)
Figure 21.1 Hydrophilic interaction chromatography of (A) sucrose, (B) glucose, and
(C) fructose. A PolyHydroxyethyl A column was used and the elution was isocratically per-
formed with 25% water and 75% acetonitrile, at ambient temperature (about 22°C). The
reducing sugars glucose and fructose show typical mutarotation double-peaks. ELS was used
for detection. (Reproduced from Karlsson, G. et al., J. Chromatogr. Sci., 42, 361, 2004, Fig. 1.
With permission from Preston Publications, a Division of Preston Industries, Inc.)
seen for more than a decade due to the desire to analyze more complex biological
samples, such as in studies of proteomics and metabolomics. During this period,
several liquid chromatography column manufacturers have introduced new HILIC
columns with different types of ligands (poly-succinimide-derived, amide, diol, and
ion-exchange ligands, including the zwitterionic type, etc.) as well as the bare silica
columns, suitable for carbohydrate analysis. See the work by Hemström and Irgum
for a comprehensive review of HILIC.25
In the following sections, method development and applications of HILIC methods
for separating and analyzing carbohydrates are discussed.

500.00
450.00 40°C, 5 mM AmForm
400.00
350.00
300.00
250.00
M c
lc
200.00
an
N
A
150.00
Fu
N
c
eu
100.00
G
5A
G
lc
al
c
50.00
A
0.00
–50.00
(A) 0.00 10.00 20.00 30.00 40.00 50.00
500.00
450.00 50°C, 5 mM AmForm
400.00
350.00
300.00
250.00
200.00
150.00
100.00
50.00
0.00
–50.00
(B) 0.00 10.00 20.00 30.00 40.00 50.00
500.00
450.00 60°C, 5 mM AmForm
400.00
350.00
300.00
250.00
200.00
150.00
100.00
50.00
0.00
–50.00
0.00 10.00 20.00 30.00 40.00 50.00
(C) Time (min)
Figure 21.2 Optimization of the separation of monosaccharides by HILIC. The influence

of temperature and ammonium formate (AmForm, pH 5.5) concentration on the separation is
shown in A–E, and the selectivity results are given in F. A TSK-Gel Amide-80 column, and
82% acetonitrile, in water, was used in all experiments: (A) 40°C, 5â•›m M AmForm; (B) 50°C,
5â•›m M AmForm; (C) 60°C, 5â•›m M AmForm;

500.00
450.00 60°C, 15 mM AmForm
400.00
350.00
300.00
250.00
200.00
150.00
100.00
50.00
0.00
–50.00
(D) 0.00 10.00 20.00 30.00 40.00 50.00
500.00
450.00 60°C, 10 mM AmForm
400.00
350.00
300.00
250.00
200.00
150.00
100.00
50.00
0.00
–50.00
(E) 0.00 10.00 20.00 30.00 40.00 50.00
500.00
G alN uc
450.00
lc A
60°C, 5 mM AmForm
G
an G G
N c
Ac
400.00
al lc
F
350.00
N
G
eu
lc
300.00
5A
A
c
250.00
200.00 GlcA
Neu5Ac
150.00 Gal
Glc
100.00 Man
GlcNAc
50.00 GalNAc
Fuc
0.00
–50.00
0.00 10.00 20.00 30.00 40.00 50.00
(F) Time (min)
Figure 21.2 (continued) (D) 60°C, 15â•›m M AmForm; (E) 60°C, 10â•›m M AmForm; and
(F) 60°C, 5â•›m M AmForm. Mixtures of all six monosaccharides [l-fucose (Fuc), d-mannose
(Man), d-galactose (Gal), N-acetyl-d-glucosamine (GlcNAc), N-acetylneuraminic acid
(Neu5Ac, the most common sialic acid), and d-glucuronic acid (GlcA)] were injected in A–E.
In F, which shows the overlay of eight chromatograms, the same six monosaccharides and
N-acetyl-d-galactosamine (GalNAc) and d-glucose (Glc) were analyzed separately. Increasing
the temperature gave an increased efficiency and resolution, while increasing the ammonium
formate concentration gave increased retention for the carboxylic acid-containing monosac-
charides (N-acetylneuraminic acid and d-glucuronic acid). The conditions in C and F repre-
sent the optimized method. ELS was used for detection. (Reproduced from Karlsson, G.

21.2 Development and Optimization of HILIC Methods

for Separating and Analyzing Carbohydrates
In the development of HILIC methods for separating and analyzing carbohydrates
and other compounds, many variables should be considered, including column type,
mobile-phase composition, temperature, flow rate, sample preparation, sample con-
centration and volume, and mode of detection. Different approaches and aims of the
analysis can be used, depending on whether the analysis is for quantification, struc-
ture determination, impurity determination, or obtaining a chromatographic profile
of a mixture of complex carbohydrates. For example, if the aim is to quantify one
specific carbohydrate, then a baseline resolved peak (Rs > 1.5) of this compound is
preferred, but if the aim is to obtain a chromatographic profile of a mixture of very
complex carbohydrates, the baseline resolution of all peaks is usually not possible.
In addition, the mobile-phase composition must be compatible with the detector,
which is important for MS detection and characterization. Recent review articles
that cover the method development of HILIC methods include the works by Ikegami
et al.,26 Dejaegher et al.,27 and Hao et al.28 The excellent book by Snyder et al.29 is also
recommended for its coverage of general high-performance liquid chromatography
(HPLC) method development.
21.2.1 Use of Different Stationary Phases

Amino-propyl was the first HILIC stationary phase and was used much during the
1970s and 1980s. In the past decade, as several new stationary phases for HILIC
have been commercially available, amino-propyl columns are less frequently used.
This can be partially explained by the fact that amino-propyl columns are unstable
and that a reactive Schiff’s base can be formed by the amino group. Schiff’s bases
can bind to aldehydes, including reducing carbohydrates, which may change the
functionality of the HILIC column, though this has not been shown to occur for
reducing carbohydrates in the HILIC conditions. In 1990, Alpert defined HILIC
and introduced a new column material of the succinimide type (polyhydroxyethyl
aspartamide silica, PolyHydroxyethyl A, from PolyLC), for separating various polar
compounds.3 A few years later, a similar succinimide-based column (PolyGlycoplex,
from PolyLC), intended mainly for analyzing carbohydrates, including oligosaccha-
rides, was described by the same author.24 Bare silica columns, adapted for HILIC,30
have also been commercialized in the last decade (e.g., Atlantis, manufactured by
Merck). This column type gives a low retention, compared to the other types of
HILIC columns, and thus needs a higher concentration of organics in the mobile
phase to bind and separate carbohydrates. Diol columns, which readily give hydro-
gen bonding, are more stable than bare silica and have been mostly used for protein
separation, though this column type has also been occasionally used for HILIC of
carbohydrates. Cyano-propyl columns, to some extent, have been used in HILIC;
however, these columns give a low retention, due to the inability to form hydrogen
bonds. They also suffer from low stability in aqueous mobile phases.25
The amide column is more stable than several other HILIC column types and
has been frequently used to separate several polar substances, including simple

and complex carbohydrates.31–35 Reports suggest that amide columns have a higher
efficiency than that of PolyHydroxyethyl A columns.34 HILIC columns with zwit-
terionic (sulfoalkylbetaine) ligands were introduced by Irgum and have been used
to separate different types of carbohydrates.25,36,37 Columns with zwitterionic
groups are very stable and have been previously used for conventional cation-
exchange chromatography. When the organic (acetonitrile) concentration exceeds
about 50%, the hydrophilic forces mainly dominate, in comparison to electrostatic
interactions, when using an ion-exchange column in HILIC mode (i.e., with about
60%–90% organics in the mobile phase). The ionic interactions that are present
at more than 50% acetonitrile will be less prominent and will contribute to the
mixed-mode retention. The amide (e.g., the Amide-80 column, from Tosoh) and
zwitterionic (e.g., the ZIC HILIC column, from SeQuant/Merck) types of HILIC
columns are now among the most popular stationary phases for HILIC separa-
tions. One difference between the two is that a higher flow rate can be used for the
amide-type column. User-made monolithic HILIC columns, with polyacrylamide-
coated silica, for achieving a high flow rate and high resolution, have been used
for separating galactose, sucrose, and lactose.38 Bare silica monolithic columns
are available from Merck and can be used for HILIC. Commercially available
cyclodextrin HILIC columns have been used for separating mono-, di-, and oli-
gosaccharides.39 Cyclodextrins, which are composed of five or more 1-4 linked
α-d-glucopyranoside units, are relatively hydrophobic in the interior of their toroid
structure, which is used for RPLC separations. The outside, however, is more
hydrophilic and is used for HILIC separations. A column length of 15â•›cm, with an
inner diameter (ID) of 4.6â•›m m, is generally recommended, but shorter columns,
with an ID of 2.1â•›m m, are more suitable for MS detection, because a lower flow rate
can be used and that it gives less dilution of the sample. The particle size of 3â•›μm,
which increases the resolution compared to the more traditionally 5â•›μm particles,
and a pore size of about 100â•›Å (10â•›n m) is often used, except for larger molecules
where a larger pore size is recommended. For example, in the case of hyaluronic
acid with a high molecular mass (up to 5 million Da), 4000â•›Å pore size has been
used.10 A 1.7â•›μm particle size bare silica HILIC column, Acquity (Waters), is avail-
able for use in ultrahigh-performance liquid chromatography (UPLC). Generally,
silica-based columns tolerate a pH of 2–7, while polymeric-based HILIC columns
may tolerate a pH up to 10 (or higher). See Table 21.1 for information about some
different HILIC column types and manufacturers.
21.2.2 Mobile-Phase Composition

The first choice for an organic for HILIC is acetonitrile, which has excellent chro-
matographic properties, at least partly because of its low viscosity and therefore high
diffusion rate and efficiency, so that the resolution will be superior to that of other
organics (such as propanol) that are used in HILIC. The eluotropic strength of solvents
used in HILIC is water > methanol > ethanol > propanol > acetonitrile > acetone >
tetrahydrofuran (THF). Alcohols (e.g., methanol) more easily form hydrogen bonds
with analytes and also with water than acetonitrile. This will affect the partially
immobilized aqueous layer that is bound to the gel matrix surface, which will lower

Table 21.1
Applications for HILIC Separation of Carbohydrates
Column Brand Name
Application/Analyte and Manufacturer Support and Functionality Mobile Phasea Detection Reference
Mono-, di-, and Cyclobond III and I, Silica, α- and β-cyclodextrin 92%–70% acetonitrile UV and RI 39
oligosaccharides Advanced Separation Appl. 3.1
Technologies
Highly polar compounds Amide-80, Tosoh Silica, amide 85%–45% acetonitrile and ESI MS-MS 34
from plant samples 1.0–3.6â•›mM ammonium acetate Appl. 3.2
(including mono- and (pH 5.5) dual gradient
oligosaccharides)
Monosaccharides and Amide-80, Tosoh Silica, amide Isocratic 80% or 60% acetonitrile Fluorescence of 33
oligosaccharides and 80%–60% acetonitrile post-column
gradient derivatization
with
benzamidine
Monosaccharides in Amide-80, Tosoh Silica, amide 82% acetonitrile, 5â•›mM ELSD 32
N-linked ammonium formate, pH 5.5
oligosaccharides
Mono-, di-, and User-made Monolithic silica, modified 80% acetonitrile, 13â•›mM ESI MS/MRM 38
trisaccharides with polyacrylamide ammonium acetate
Mono- and disaccharides Luna Amino, Phenomenex Silica, amino 80% acetonitrile, 40â•›μM cesium MS/MRM 40
acetate
Highly polar compounds ZIC HILIC, SeQuant/Merck Silica, sulfoalkylbetaine 90%–10% acetonitrile and MS-MS 41
from plants (mostly zwitterionic 0.5–4.5â•›mM ammonium acetate
mono- and disaccharides) dual gradient, in 0.1% formic
acid, pH 4

Development and Application of Methods for Separation
Glucuroconjugate Atlantis, Merck Bare silica 87% acetonitrile, 12â•›mM ESI MS-MS 30
of propofol ammonium acetate, pH 5
Sucrose in a protein PolyHydroxyethyl A, Silica/poly-succinimide, 75% acetonitrile ELSD 42
formulation PolyLC polyhydroxyethyl aspartamide
Oligosaccharides, PolyGlycoplex, PolyLC Silica/poly-succinimide derived 10â•›mM TEAP, 80% methanol, MS and UV 24
complex carbohydrates pH 4.4 200/230â•›nm
N-linked glycans and ZIC-HILIC, SeQuant/Merck Silica, sulfoalkylbetaine 77%–63% acetonitrile, in 5â•›mM ESI MS and UV 36
N-glycopeptides zwitterionic ammonium acetate 215â•›nm, of Appl. 3.3
peptides and
2-AP glycans
Oligosaccharides Polymer-NH2, Astec Polymer, amino 70%–5% acetonitrile, 2.9%–4.9% Fluorescence of 35
acetic acid/0.9%–2.9% TEA, 2-AB derivatives Appl. 3.4
dual gradient, in 1% THF
Glucosinolates ZIC-HILIC, SeQuant/Merck Silica, polymeric 70% acetonitrile, 15â•›mM UV (235â•›nm) 37
sulfoalkylbetaine zwitterionic ammonium formate, pH 4.5 Appl. 3.5
Glucosaminoglycans Amide-80 capillary column, Polymeric, amide 82%–46% acetonitrile and MS-MS 43
Tosoh 10–30â•›mM formic acid, pH 4.4, Appl. 3.6
dual gradient
N-linked oligosaccharides Amide-80, Tosoh Silica, amide 64%–52% acetonitrile, 5 or 25â•›mM Fluorescence of 31
from fetuin ammonium acetate, pH 5.5 2-AB derivatives
Oligosaccharides GlycoSep N, ProZyme Silica, amide 75%–58% acetonitrile and MS and 44
12–29â•›mM ammonium acetate, fluorescence of
pH 4.4, dual gradient 2-aminoacridone
derivatives
a Water or aqueous buffer comprises the remaining part of the mobile phases.
501
the polar retention of carbohydrates to the column. Changing from an aprotic organic
solvent (e.g., acetonitrile) to a protic organic solvent (e.g., propanol) can sometimes
change the selectivity. However, the efficiency in the chromatographic system, when
using propanol or other alcohols, will unfortunately decrease, compared to that
when using acetonitrile, and, usually, it is best to stick to acetonitrile. An alternative,
though seldom used, is to use a ternary mobile phase, for example, 70% acetoni-
trile/15% methanol/15% water, in ammonium acetate buffer, to change the selectiv-
ity without losing much efficiency. Water is sometimes completely replaced with the
very polar organic methanol, by using 75% acetonitrile/25% methanol, for example.
However, water is usually included in the HILIC mobile phases.
To obtain good chromatographic results, the water content in an HILIC mobile
phase should be no less than 5% (or 3%). For an initial investigative analysis, a scout-
ing gradient, for example, with 85%–50% acetonitrile, in water (i.e., 15%–50% water)
with a buffer component, may be used for most HILIC columns, in a volume that cor-
responds to about 25 column volumes. If all peaks elute in a minor part of the gradient,
it would usually be best to then try an isocratic run, with a concentration of acetonitrile
corresponding to about the middle part where the peaks eluted in the gradient. If the
peaks have a more spread-out elution, in the larger part of the gradient, then a gradi-
ent should be used. A more time-consuming alternative would be to use a series of
isocratic runs with increasing organic concentrations, where it would be best to start
with a very strong mobile phase, for example, 50% acetonitrile in a buffered aqueous
solution, and then stepwise (initially 10% steps, followed by 5% steps) increase the
acetonitrile concentration as necessary, usually to about 60%–80% for most carbohy-
drates. For an optimized isocratic system, the retention factor k (formerly called the
capacity factor k′) is ideally 2–10 (1–20 may also be acceptable, especially for more
complex samples) for the separated compounds. If early or late eluters (peaks with
very low or high k values, i.e., outside the 1–20 range) appear in the chromatogram,
a gradient would likely be necessary for the separation. For determining a suitable
mobile-phase composition for an isocratic run, as judged from the scouting gradient,
compensating adjustments should be made for the gradient delay volume (also called
the dwell volume) and the column dead volume. If a gradient is chosen, the gradient
range can be decreased (e.g., to 80%–60% acetonitrile), provided that all peaks appear
in that range. A number of gradient variations, with different concentration steepness,
gradient times, and acetonitrile concentration ranges should be tested to obtain the
desired resolution, peak shape, and analysis time. If a small difference occurs in the
polarity between the carbohydrates, for example, if just a number of monosaccharides
need to be separated, the first choice would be to try an isocratic separation. In con-
trast, if mixtures of carbohydrates that contain components of substantially different
size and/or differences in the number of carboxyl- or other charged groups, i.e., a large
difference in total polarity, are to be separated, then a gradient is usually necessary.
Using a gradient will avoid peak broadening for peaks with long retention times and
will accomplish peak compression, giving high, narrow peaks with high efficiency
and improved detection sensitivity over the whole gradient. After an 80%–60% aceto-
nitrile linear gradient, in water, the column should generally be equilibrated in about
5–10 column volumes with the initial conditions (80% acetonitrile). See Figure 21.3
for a comparison between isocratic and gradient elutions of carbohydrates.

2
200
1
3
100
4
5 6
0
(A) 0 30 60 90
300 2
Fluorescence intensity (mV)
200 1
4
100 5
6 7
8 9
0
(B) 0 5 10 15
2
200
1 3
100 4
5 6 7 8 9
0
0 15 30 45
(C) Time (min)
Figure 21.3 Simultaneous separation of three monosaccharides and six oligosaccharides.

A TSK-Gel Amide-80 column was used at 80°C together with post-column labeling by ben-
zamidine followed by fluorescence detection. (A) Isocratic elution with acetonitrile:water
(80:20); (B) isocratic elution with acetonitrile:water (60:40); and (C) linear gradient, 80%–60%
acetonitrile, in water. A mixture of (1) d-ribose, (2) d-fructose, (3) d-glucose and the lami-
narioligosaccharides (4) laminaribose, (5) laminaritriose, (6) laminaritetraose, (7) lami-
naripentaose, (8) laminarihexaose, and (9) laminariheptaose was separated. By using the
gradient, narrow, high, and well-resolved peaks were obtained for each of the separated
carbohydrates. (Reproduced from Kakita, H. et al., J. Chromatogr. A, 961, 77, 2002. With

As an alternative to ordinary linear gradients, a segmented gradient may also be

considered, which can be successfully used in cases where an adjusted (more shal-
low) gradient does not give the desired resolution. If, for example, several crowded
low-resolved peaks appear in the middle of a 80%–60% acetonitrile gradient, in
addition to some well-separated peaks before and after the crowded part, it might
help to first run an 80% to about 70% acetonitrile linear gradient, as before, followed
by a very shallow gradient part, or an isocratic plateau, to achieve an improved reso-
lution in the crowded part, and finally a continuation of the linear gradient, down
to 60% acetonitrile. Furthermore, the flow rate can also, if required, be optimized
(0.5–1.0 and about 0.2â•›m L/min are usually recommended for 4.6 and 2.1â•›mm ID col-
umns, respectively, see column information from the respective manufacturers). In
addition to a high concentration of acetonitrile and some water, a buffer (5–20â•›m M,
up to 200â•›m M), primarily formic acid, acetic acid, ammonium formate, ammonium
acetate, ammonia, or ammonium carbonate, depending on the desired pH, is rec-
ommended for addition to the mobile phases. Buffers are recommended by many
column manufacturers to avoid electrostatic interaction with charged free silanols in
the silica stationary phase. The pH of the mobile phase can influence the retention
of ionizable carbohydrates in HILIC. The charged state of the ionizable carbohy-
drate depends on whether the pH is below or above the pKa, which in turn affects
the hydrophilicity and thereby the retention of the analyte. Thus, the pH should be
changed if the desire is to change the selectivity due to the charge of ionizable ana-
lytes. In addition, when the pH increases, for example, from 3 to 6, more silanols in
the stationary phase are ionized, and an increased retention of positively charged
compounds is usually obtained. Even if only neutral carbohydrates are analyzed,
ionized contaminants might be present in low concentrations in the sample. Hence,
buffers in the mobile phase are beneficial for reproducibility and performance in
HILIC. The buffer capacity of the mobile phase should also be of concern; it is rec-
ommended to set the pH to ±1.0â•›pH unit from the pKa, preferably as close as possible
to the pKa of the used buffer, to obtain an acceptable buffer capacity. This is not
strictly followed, however, and trifluoroacetic acid, for example, has a pKa of 0.3, but
is often used as a “buffer” (or “additive” in this case) at pH 2 in RPLC, even though
the buffer capacity is very low at that pH. The pH should be adjusted, if necessary,
before the organics are added to the mobile phase.
The buffer component (e.g., ammonium acetate), in addition to the buffering
effect, can influence retention and selectivity in HILIC in several ways. An increased
salt concentration can increase the polarity of the partly immobilized aqueous layer
on the surface of the column material, thereby increasing the retention of polar
compounds in an HILIC system, which supports the partitioning theory of reten-
tion in HILIC.3,24,25 On the contrary, for charged column HILIC materials (e.g.,
NH2 columns at neutral or acidic pH), an increased salt concentration will decrease
electrostatic interactions and thereby decrease the retention of negatively charged
carbohydrates, giving the elution due to an ion-exchange effect. Elution in a simulta-
neously dual mode, where an increasing water gradient gives the main elution, and
an increasing buffer concentration (e.g., ammonium acetate) is used in the gradient,
has been used for separating various carbohydrates (Table 21.1). By adding a salt

gradient, an improved selectivity and resolution can sometimes be obtained, espe-

cially for charged HILIC columns, for example, amino columns in combination with
negatively charged carbohydrates.35 This elution mode has been called HILIC-IEX
(or HILIC weak anion-exchange [WAX] chromatography). In HILIC-IEX, the elu-
tion is sometimes used in a segmented mode; first, the acetonitrile concentration is
decreased to a suitable level in the gradient, which elutes uncharged carbohydrates;
and then the ion strength is increased to elute negatively charged carbohydrates
which are more strongly bound to the column.
For amide columns, an increased concentration of ammonium acetate or ammo-
nium formate (10–25â•›m M instead of 5â•›m M) increases the retention of ionized mono-
and oligosaccharides that contain carboxylic groups, possibly due to increased
binding of these carbohydrates to enriched salts in the partly immobilized aqueous
layer on the column gel matrix surface (Figures 21.2 and 21.4). Inorganic buffer
salts (e.g., phosphates) should generally be avoided due to their low solubility in
HILIC mobile phases. Nevertheless, phosphates have the advantage of a high detec-
tion sensitivity at low UV wavelengths (about 210â•›nm), and the solubility of phos-
phates in acetonitrile is higher at acidic pH than at alkaline pH, due to a less charged
state. Buffer salts that have been used in HILIC include ammonium acetate, ammo-
nium formate, cesium acetate, triethylammonium phosphate (TEAP), and sodium
perchlorate (see Table 21.1).
21.2.3 Temperature and Other Variables

In HILIC, column oven temperature of about 30°C–60°C (occasionally up to
100°C) is often used. However, the possibility of temperature-dependent degrada-
tion of the carbohydrates should be considered, and elevated temperatures should
be avoided for labile compounds. A higher temperature is often preferred because
it decreases the viscosity of the mobile phase and increases the diffusion and the
chromatographic efficiency, giving a high resolution. Mutarotation, i.e., the specific
rotation of cyclic carbohydrates as they reach equilibrium between their α- and
β-anomeric forms of reducing sugars, can give typical disturbing double-peaks
in HILIC. An elevated temperature and a more alkaline pH,45 as well as organic
amines46 (e.g., triethylamine),24 accelerate the rate of the anomer interconversion
sufficiently to reach a dynamic equilibrium of the anomers that avoids the double-
peak appearance. In Figure 21.1, typical anomeric mutarotation double-peaks are
shown for the reducing monosaccharides glucose and fructose, but not for the non-
reducing disaccharide, sucrose, in an HILIC system run at ambient temperature
(about 22°C), with a PolyHydroxyethyl A column. Later, an amide column was used
at temperatures up to 60°C, with a similar mobile phase, to avoid the mutarotation
double-peaks (Figure 21.2).
In HILIC, as in most other chromatographic techniques, the retention time is
usually decreased by elevated temperatures, with other conditions being unchanged,
which can be compensated by increasing the percentage of organics. Figure 21.2
shows an optimized separation of monosaccharides, where different temperatures
and concentrations of ammonium formate were evaluated on an amide column.

I) 5mM AmAc II) 25mM AmAc III) Oligosaccharide structure
200.00 Oligomannose 5 (MAN-5)

200.00 MAN-5 MAN-5
150.00 Manα1
150.00
6
100.00 100.00 Manα1
3
Manα1 6
50.00 50.00 Manβ1 4 GlcNAcβ1 4GlcNAc
3
0.25 0.25 Manα1
(a) 0.00 10.00 20.00 30.00 40.00 50.00 (a) 0.00 10.00 20.00 30.00 40.00 50.00
200.00 Oligomannose 9 (MAN-9)
200.00 MAN-9 MAN-9
150.00 Manα1 2 Manα1
150.00
Fluoroscence intensity (a.u.)
6
100.00 100.00 Manα1 2 Manα1 3 Manα1 6
4GlcNAcβ1 4GlcNAc
50.00 3 Manβ1
50.00 Manα1 2 Manα1 2 Manα1
0.25 0.25
(b) 0.00 10.00 20.00 30.00 40.00 50.00 (b) 0.00 10.00 20.00 30.00 40.00 50.00
200.00
200.00 NA2 NA2 Asialo galactosylated biantennary complex (NA2)
150.00
150.00
100.00 100.00 Galβ1 4GlcNAcβ1 2Manα1
6
50.00 50.00 Manβ1 4GlcNAcβ1 4 GlcNAc
3
(c) 0.00 10.00 20.00 30.00 40.00 50.00 (c) 0.00 10.00 20.00 30.00 40.00 50.00
200.00
200.00 NA3 NA3 Asialo galactosylated triantennary complex (NA3)
150.00
150.00
6
50.00 4GlcNAcβ1 4GlcNAc
50.00
Galβ1 4GlcNAcβ1 3 Manβ1
4 Manα1
0.25 0.25 2
(d) (d) 0.00 10.00 20.00 30.00 40.00 50.00 Galβ1 4GlcNAcβ1
0.00 10.00 20.00 30.00 40.00 50.00

200.00
Development and Application of Methods for Separation

200.00 A1 A1
150.00
Monoslatylated galactosylated biantennary complex (A1)
150.00
100.00 100.00 3/6 Galβ1 4GlcNAcβ1 2Manα1
6
NeuSAco 2 Manβ1 4GlcNAcβ1 4GlcNAc
50.00 50.00 3
3/6 Galβ1 4GlcNAcβ1 2Manα1
0.25 0.25
(e) 0.00 10.00 20.00 30.00 40.00 50.00 (e) 0.00 10.00 20.00 30.00 40.00 50.00
200.00
200.00
A2 150.00 A2 Disialylated galactosylated biantennary complex (A2)
Fluoroscence intensity (a.u.)
150.00
100.00 100.00
NeuSAco2 3/6 Galβ1 4GlcNAcβ1 2Manα1
50.00 50.00
6 Manβ1 4GlcNAcβ1 4GlcNAc
0 0 3
(f ) 0.00 10.00 20.00 30.00 40.00 50.00 (f ) 0.00 10.00 20.00 30.00 40.00 50.00 NeuSAco2 3/6 Galβ1 4GlcNAcβ1 2Manα1
200.00
200.00
A3 Trisialylated galactosylated triantennary complex (A3)
150.00 150.00
100.00 100.00 NeuSAco2 3/6Galβ1 3/4 GlcNAcβ1 2 Manα1

50.00 NeuSAco2 3/6Galβ1 3/4 GlcNAcβ1 6 Manβ1 4GlcNAcβ1 4GlcNAc
50.00
3
0 4
0 Manα1
(g) 0.00 10.00 20.00 30.00 40.00 50.00 (g) 0. 00 10.00 20.00 30.00 40.00 50.00 2
NeuSAco2 3/6Galβ1 3/4 GlcNAcβ1
Time (min)
Figure 21.4 Influence of ammonium acetate (AmAc) concentration on the retention of 2-aminobenzamide (2-AB)-labeled oligosaccharides. A mix-
ture of seven 2-AB-labeled glycan standards was separated by the Amide-80 column at different concentrations of ammonium acetate, using a segmented
acetonitrile gradient (64%–52%) in ammonium acetate (pH 5.5) and water for elution at 45°C. (I) 5â•›mM ammonium acetate; (II) 25â•›mM ammonium acetate;
(III) structures of the seven, N-linked glycan standards purchased from ProZyme; abbreviated MAN-5, MAN-9, NA2, NA3, A1, A2, and A3. N-acetyl-d-
glucosamine (GlcNAc), d-mannose (Man), d-galactose (Gal), and N-acetylneuraminic acid (Neu5Ac) are indicated in the structures. The higher concentration
of ammonium acetate (25â•›m M) increased the retention for N-acetylneuraminic acid-containing oligosaccharides. (Reproduced from Karlsson, G. et al.,
J. Chromatogr. Sci., 46(1), 68, 2008, Figs. 1,3,4. With permission from Preston Publications, a Division of Preston Industries, Inc.)
507
Autosampler- and needle-wash solutions should be used, as they are for many
HPLC systems, that contain 80% acetonitrile, 20% water, and no buffer, to avoid
possible disturbances from very small amounts of washing solution that contain a
high percentage of water that could otherwise give peak broadening.
21.2.4 Sample Preparation Strategy

Ideally, the injected sample should have the same percentage of organics as in the
initial condition of the HILIC analysis, and never less than 50% organics, to avoid
peak broadening and decreased chromatographic resolution. However, when inject-
ing very small sample volumes, a higher water concentration is tolerated, which can
be useful because carbohydrates, especially glycans, often have low solubilities in
high concentrations of organic solvents. Usually, samples with about 70% acetoni-
trile are sufficient for avoiding peak broadening and precipitation. In addition, the
injected sample should not contain particles or interfering substances and should
be injected in the lowest possible volume, preferably 5–25 and 0.5–5â•›μL for 4.6 and
2.1â•›m m ID columns, respectively. For samples containing a lot of salts or proteins,
precipitation will usually occur when the acetonitrile is added. The precipitate
should then be removed by centrifugation (e.g., 10,000 × g for 10â•›m in), and the clear
supernatant can be injected on the HILIC system. The recovery of carbohydrates
may decrease when the relative volume of the pellet increases, since pellets often
contain approximately 50% liquid, and, in addition, some binding forces may occur
between the carbohydrates and the precipitate. To remove particles from the HILIC
samples, a recommended approach would be to use centrifugation (see above) or
filtration with a syringe filter (based on poly-tetrafluoroethene or similar chemical-
inert materials) having a 0.45â•›μm pore size. The sample concentration depends on
the loading capacity of the selected column, and on the complexity and solubility
of sample components. The sample solubility should be tested in the mobile phase
prior to injection.
The removal of large amounts of proteins, lipids, and other nonpolar substances,
which may disturb the chromatographic performance or the detection, in a sample,
can easily be achieved by RPLC solid-phase extraction (SPE) cartridges (often with
100â•›mg dry gel per cartridge), usually with C-18 groups bound to the gel matrix. By
combining RPLC with an HILIC analysis, a so-called orthogonal separation, where
the two separation mechanisms are totally independent of each other, is achieved,
which is analytically advantageous. The dry gel SPE cartridges are (i) first, acti-
vated with methanol (or sometimes acetonitrile), for wetting and cleaning of the gel;
(ii) conditioned with, for example, ammonium acetate buffer (or water) to remove
the methanol and equilibrate the cartridge; (iii) loaded with sample; and finally
(iv) washed with buffer (or water). During steps (iii and iv), all polar substances,
including carbohydrates, are eluted. These should be collected and concentrated, for
example, by centrifugal evaporation, and the dry carbohydrates can be dissolved in
a small volume of water, followed by an addition of acetonitrile to a final concentra-
tion of about 70%, before being injected on the HILIC column. If the nonpolar sub-
stances are desired, they can easily be eluted from the SPE cartridge by increasing

the organic concentration. Cartridges of porous graphitic carbon are useful for bind-
ing and separating underivatized oligosaccharides, by RPLC, from fluorescent dyes
and other interfering compounds.
Another approach is to collect fractions from a previous RPLC analysis and
directly inject these on an HILIC system, or more conveniently, use an automatic
column-switching system to perform the procedure. However, it is often necessary
to concentrate the sample and to add more organics before the sample is injected on
the HILIC system. Sample preparation with ion-exchange SPE cartridges can also be
used before performing analytical HILIC, and HILIC SPE cartridges are also com-
mercially available. In addition, liquid–liquid extraction and other extraction tech-
niques are used for sample preparation. For the specific cleavage of N-linked glycans
from glycoproteins, peptide N-glycosidase F (PNGase F), or hydrazine, which also
cleaves O-linked carbohydrates, can be used, followed by an RPLC SPE step, for
example, to separate glycans and protein. See the book by David and Moldoveanu
for a comprehensive review of sample preparation techniques and strategies in
chromatography.47
21.2.5 Detection of Carbohydrates

The detectors for analytical liquid chromatography can be divided into universal
detectors, which respond to all compounds independent of their physicochemical
properties, and selective detectors, which respond to just some of these properties.
Although direct detection of carbohydrates by UV absorbance or fluorescence is usu-
ally not possible, some carbohydrates that have a high number of carboxylic groups,
for example, hyaluronic acid, can be detected by absorbance at 210â•›nm, using a buf-
fer and salt of low absorbance (e.g., sodium phosphate buffer and sodium sulfate in a
salt gradient) to avoid interference.10
The RI detector, which is a common universal detector, has been used for a long
time for carbohydrate analyses. However, RI detectors suffer from disadvantages
such as low sensitivity, noncompatibility with gradient elution, and being affected by
small fluctuations in temperature and pressure. In the past two decades, ELSD has
been used as a semiuniversal HPLC detector, as it is compatible with gradient elution
and is widely used for detecting nonvolatile compounds such as carbohydrates.42,48,49
However, ELSD is not sensitive enough for detecting low concentrations of carbo-
hydrates. ELSD is based on the light scattering detection of nonvolatile compounds
after nebulization and evaporation of the mobile phase.
The derivatization of reducing carbohydrates with chromophoric reagents, for
example, p-aminobenzoic ethyl ester,6 followed by absorbance detection was a
great improvement for detecting carbohydrates. Derivatization with fluorescent
groups (e.g., 2-aminobenzamide (2-AB), 2-anthranilic acid (2-AA), benzamidine, or
2-aminoacridone) is often performed to detect glycans with very high sensitivity.50
However, disadvantages of the derivatization procedure include the possibility of
carbohydrate degradation, fluctuating yield of derivates, and toxicity of the reagents.
Figure 21.5 shows a schematic fluorescence-labeling procedure, using 2-AB.51
The labeling works effectively for reducing carbohydrates, and the procedure can

O
Dye Schiff’s base
H2N OH
H2N O
RO OH
H 2N HO N
+ NHAc
OH Glycan
OH Reductant
RO
HO O
NHAc
Cyclic/acyclic OH
H2N O
in equilibriuim OH
RO H
HO N
NHAc
Labeled glycan
Figure 21.5 Labeling of carbohydrates by 2-aminobenzamide (2-AB). The Schiff’s base

formation requires a glycan with a free reducing sugar, which, in equilibrium, is between
the ring closed (cyclic) and ring open (acyclic) forms. The primary amino group of the dye
performs a nucleophilic attack on the carbonyl carbon of the acyclic reducing sugar residue
to form a partially stable Schiff’s base. The Schiff’s base imine group is then chemically
reduced by cyanoborohydride to give a stable, labeled glycan. The fluorescent-labeled gly-
can has a distinct fluorescence with maximum excitation and emission at 330 and 420â•›nm
wavelengths, respectively. (Reproduced from Signal 2-AB labeling kit, Prod. No. GKK-404.
Product information from ProZyme, San Leandro, CA. With permission from ProZyme.)
become more complicated for nonreducing carbohydrates, possibly involving the

post-column hydrolysis of complex carbohydrates and labeling of the generated
reducing sugars. After labeling, a clean-up step is usually necessary to remove most
of the unreacted dye, to avoid a strong fluorescent interfering signal. See the review
by Anumula for a comprehensive description and evaluation of fluorescence labeling
in the detection of carbohydrates.50
NMR is a powerful and complicated technique for detecting and characterizing
carbohydrates. NMR requires expertise and intricate equipment and operates on the
principle of detecting which nuclei distort a magnetic field owing to their property of
spin. NMR provides complimentary information to MS in obtaining comprehensive
structural data, and NMR and MS can be used in parallel with liquid chromatogra-
phy. For a review of LC-NMR-MS, refer to Corcoran and Spraul.52
MS is close to being a universal LC detector, and different MS systems, including
triple quadrupoles, ion traps, and quadrupoles coupled to time-of-flight instruments,
are increasingly used for analysis and identification of carbohydrates.26,34,40,41,43 The
principle of MS is that the analytes are first ionized and then a mass analyzer dif-
ferentiates the ions according to their mass-to-charge ratio (m/z), with the ion beam
being measured by a detector. The ESI interface, between the LC column and the
MS inlet, is commonly used and fully compatible with HILIC mobile phases with a
high percentage of organics, water, and low concentrations of volatile buffers. ESI
MS is usually performed in the positive mode, generating positively charged carbo-
hydrates by the addition of a proton [M + H]+ (or another cation such as sodium ion
[M + Na]+), or may be performed in the negative mode, for carbohydrates containing

anionic groups, such as deprotonized carboxylic acid groups [M − H]−. In simple

MS, where one MS analyzer is used, only m/z is detected and the analytes with same
m/z cannot be discriminated. In tandem mass spectrometry (MS-MS), two MS ana-
lyzers operate in discrete zones and ions can be fragmented by energy transfers (usu-
ally by collisions with gas molecules) in the region between the zones. MS-MS is a
powerful technique for determining structure and for quantifying carbohydrates and
other compounds. Sequential enzymatic digestion, with exoglucosidases that have a
narrow specificity, in combination with LC-MS, is also a useful tool for determin-
ing complex carbohydrate structures. One drawback in using glycosidases (or other
digesting enzymes) is in the risk of incomplete cleavage, due to the steric hindrance
or suboptimal incubation conditions. Multiple reaction monitoring (MRM), also
called single reaction monitoring, is an MS-MS technique, where both analyzers are
set to monitor specified pairs of parent (precursor) and product ions, which is use-
ful for quantifying carbohydrates.38,40 MRM is usually performed with an internal
standard that has exactly the same structure as the analyte but is labeled with a few
atoms of a stable heavy isotope (e.g.,2H,13C,15N) to increase accuracy and precision.
Short descriptions of a few HILIC-MS/MRM methods are given in Table 21.1. For a
review of the analysis of carbohydrates by MS, see Zaia.22
Thus, MS is often the preferred HILIC detection technique, because of its high
accuracy, precision, and sensitivity, and the peaks from contaminating compounds
can be directly identified. Fluorescence labeling and detection is very sensitive and the
method can be used for low concentrations of reducing carbohydrates. ELSD is also
convenient if high concentrations of carbohydrates are to be measured. All of these
detection techniques involve some optimization, for example, MS may need adjust-
ments of the cone voltage, capillary voltage, scan time, and other settings, and the fluo-
rescence labeling and clean-up procedures may also need some optimization. ELSD
should be optimized by testing different temperatures and gas flows.
21.2.6 Short Outline for the Development of an HILIC Method

This outline has been successfully used to develop HILIC methods for several years
at the author’s laboratory. The following discussion should be considered as general
input for developing an HILIC method for separating and analyzing carbohydrates.
In developing an HILIC method, many aspects should be considered. For example,
what is the aim of the work? Is it for quantifying one compound or many com-
pounds, for determining the structure of a complex carbohydrate, or for obtaining
the chromatographic profile of a mixture of glycans? How complex is the sample
and what interfering substances are known to be present? Is a suitable sample prep-
aration method available? Which detection technique should be used: an ESI MS
(or ESI MS-MS) system on-line to the LC or fluorescence labeling and detection?
The chemical characteristics of the analyte (structure, hydrophilicity, ionizable
groups, stability, solubility in possible mobile phases, etc.) should be considered,
and published analytical methods should be investigated with regard to the ana-
lyte (or similar substances), including sample preparation procedures and detection
techniques. In developing an HILIC method, especially when little work has been
done in the area, the best approach is to begin with a column with a documented

good selectivity and the resolution necessary for the type of separation being devel-
oped, together with an 85%–50% acetonitrile scouting gradient, in water and 10â•›m M
ammonium acetate (or ammonium formate), at a column temperature of about 40°C,
with a high but suitable flow rate for the selected column. The organic percentage
for an isocratic separation, or possibly gradient, should be optimized for the analy-
sis. If the desired resolution, selectivity, peak shape, detection response, or analy-
sis time is not obtained, additional variables should be optimized, for example, by
testing increased temperatures, changes in pH, buffer type and concentration, and
slightly altered flow rates. In addition, sample preparation procedures and detection
settings may need to be optimized. Further optimization for very complex or dif-
ficult samples can be derived by testing segmented and dual gradients, other types
of HILIC columns, or other organics. In the next section, some applications for the
HILIC analysis of carbohydrates are described, as valuable inputs to developing a
new method.
21.3 HILIC Applications for the Separation

of Carbohydrates
In this section, a number of HILIC applications for the separation and analysis
of different types of carbohydrates, using different detection techniques and col-
umns, are described, which may be used to identify suitable initial conditions in
developing a method. Some HILIC analyses of several classes of carbohydrates
are briefly described in Table 21.1. HILIC of different types of carbohydrates is
covered in several reviews, including the separation of low-molecular-weight car-
bohydrates in food and beverages,53 marine samples,54 glycoalkaloids in potatoes
and tomatoes,55 glucosaminoglycans (GAGs),56 and for the assessment of protein
glycosylation.57
21.3.1 HILIC Separation of Mono-, Di-, and Oligosaccharides

by Cyclodextrin Columns in Combination with UV
and RI Detection
Cyclodextrins, which are made of starch, are a family of cyclic oligosaccharides.

The most used cyclodextrins are α-, β-, and γ-cyclodextrin, which are composed
of six, seven, and eight monosaccharides, respectively. Cyclodextrin has been used
in liquid chromatography to obtain the specific separation of chiral substances,
and cyclodextrin columns are often used for various RPLC applications and for the
HILIC separation of carbohydrates. Figure 21.6 shows a chromatogram of the HILIC
analysis of a mixture of mono-, di-, and oligosaccharides, with high resolution of the
separated components in the analyzed mixture of the commercially obtained carbo-
hydrates. This example is from Armstrong et al.,39 where both α- and β-cyclodextrin
columns, Cyclobond III and I (250 × 4.6â•›mm ID, 5â•›μm particle size, from Advanced
Separation Technologies), respectively, were tested with isocratic and gradient elu-
tion, using acetonitrile–water or acetone–water, together with UV or RI for detec-
tion. The retention was related to the size and the number of hydroxyl groups of the

5 Water %
30
3
20
10
6 8
0
0 10 20 30 40 min
4
1 10
16
9
13
12 17 19
14
18
15
11
0 10 20 30 40
Time (min)
Figure 21.6 The gradient separation of carbohydrates on a β-cyclodextrin column

(Cyclobond I), used at 22°C. An acetonitrile–water gradient, flow rate 1.0â•›m L/min, was used,
and the change in the mobile-phase composition is shown in the figure inset: (1) Phenyl-
β-d-glucopyranoside, (2) 2-deoxy-d-ribose, (3) ribose, (4) xylose, (5) talose, (6) sorbose,
(7) glucose, (8) sorbitol, (9) sucrose, (10) turanose, (11) maltose, (12) lactose, (13) melibiose,
(14) melezitose, (15) maltotriose, (16) stachyose, (17) α-cyclodextrin, (18) β-cyclodextrin, and
(19) γ–cyclodextrin. (Reproduced from Armstrong, D.W. and Jin, H.L., J. Chromatogr., 462,
carbohydrate. The efficiency and stability of the cyclodextrin columns were found to
be superior to amino-alkyl columns.
21.3.2 HILIC Separation by an Amide-80 Column

and MS Analysis of Carbohydrates from Plants
In this example, from Tolstikov and Fiehn,34 an Amide-80 column (250 × 2â•›mm
ID, 5â•›μm particle size, from Tosoh) at ambient temperature, was used to separate

5
100
8 7
50
9
12
3
2 6
4 11
1
10
0 10 20 30 40 50
Time (min)
Figure 21.7 HILIC-MS base peak chromatograms of a mixture of standards. A TSK-Gel

Amide-80 column was used with a dual gradient (10–60â•›min), 85%–45% acetonitrile and
1.0–3.6â•›m M ammonium acetate (pH 5.5), in water, at a flow rate of 0.15â•›m L/min. A 10â•›μ L
sample was injected. Identified peaks: (1) N-acetyl-d-glucosamine, (2) sucrose, (3) D(+) raf-
finose, (4) l-methionine, (5) N-methyl-l-deoxynojirimycin, (6) l-alanyl-l-alanine, (7) 1,4-dideoxy-
1,4-imino-d-arabinitol, (8) uridine-5-diphosphoglucose, (9) stachyose, (10) glucosaminic acid,
(11) 2-amino-2-deoxy-d-glucose, and (12) maltoheptaose. An electrospray interface coupled
to an ion trap mass spectrometer, with stepwise fragmentation, was used for identification.
(Reproduced from Tolstikov, V.V. and Fiehn, O., Anal Biochem. 301, 298, 2002. With permis-
sion from Elsevier.)
plant-related substances (oligosaccharides, glycosides, amino sugars, amino acids,

and sugar nucleotides) from Cucurbita maxima leaves (Figure 21.7). Samples
from phloem exudates from the petioles of the herb were analyzed. The goal of
the study was metabolomic, i.e., quantification of every metabolite in a biological
system. HILIC is very useful for metabolomics since the metabolites are generally
very polar. For identification, ESI (both in positive and in negative modes, using
continuous polarity switching) MS-MS was used on-line with the HILIC system.
Quantification of the analytes was performed using external standards. The Amide-
80 column gave a superior resolution compared to another tested HILIC column, a
PolyHydroxyethyl A column (PolyLC).
21.3.3 HILIC Separation by a ZIC-HILIC Column and MS Analysis

of Tryptic Glycopeptides from Human Immunoglobulin
Most proteins are glycosylated, and the linked carbohydrates are important in vari-
ous biological functions. Thus, glycosylation must be analyzed and evaluated, for
example, when producing recombinant glycoproteins for pharmaceutical use. The
N-linked glycans from glycoproteins have complex structures and, because of the
many possible isomers, require sophisticated analyses. Usually, glycopeptides are

analyzed by RPLC, the most often applied liquid chromatography technique, but
sometimes this does not give the desired separation, especially for small and hydro-
philic peptides, which have low or no retention in RPLC. On the contrary, these
hydrophilic peptides, regardless of being glycosylated, usually have a strong reten-
tion in HILIC systems, so that a combination of RPLC and HILIC gives a broader
sequence coverage of the protein, compared to the analysis with RPLC alone.
In the example from Takegawa et al.,36 a zwitterionic type of column, ZIC-
HILIC (150 × 2.1â•›m m ID, 3.5â•›μm particle size, from SeQuant/Merck) was used at
40°C and combined with MS analysis of the tryptic glycopeptides from human
immunoglobulin G (IgG) (Figure 21.8). Trypsination was performed according
to a standard procedure, and the samples were injected in 20â•›μ L 80% acetoni-
trile. Elution was performed by a gradient (0–120â•›m in), 77%–63% acetonitrile, in
water and 5â•›m M ammonium acetate, at a flow rate of 0.2â•›m L/min. N-glycans and
tryptic peptides were analyzed after 2-aminopyridine derivatization (PA). The
HILIC system was run on-line with an ESI MS system for detection and identifi-
cation. The ZIC HILIC column gave a high resolution and a high selectivity for
glycopeptides, including the isomeric separation of different glycans on the same
peptide.
21.3.4 Analysis of 2-Aminobenzoic Acid-Derivatized Oligosaccharides

by an NH2 Column and Fluorescence Detection
In this application, from Anumula and Dhume,35 2-aminobenzoic acid (2-AA)-

derivatized oligosaccharides from the well-characterized standard glycoprotein
bovine fetuin were analyzed. An amino polymeric column (Polymer-NH2, 250 ×
4.6â•›m m ID, 5â•›μm particle size, from Astec) was used at 50°C and a dual gradient
(0–80â•›m in), with decreasing organics (70%–5% acetonitrile), and increasing ionic
strength (2.9%–4.9% acetic acid and 0.9%–2.9% triethylamine; TEA), in water
and 1% THF, at a flow rate of 1.0â•›m L/min, was used for elution. The sample prep-
aration of the glycans was performed by enzymatic cleavage by PNGase F and
hydrazionolysis, and detection of the 2-AA derivates was by fluorescence. By com-
bining elution in HILIC mode (increasing water) and IEX mode (increasing ion
strength), sometimes called HILIC-IEX (also known as NP-IEX or NP-HPAEC),
a very high resolution can be obtained. The same sample was also analyzed by the
reference high-resolution technique, HPAEC-PAD, and the results were similar
with regard to resolution and selectivity. The HILIC-IEX method, for separa-
tion of fluorescence-labeled glycans, gave even more defined peaks than did the
HPAEC-PAD method. In both methods, oligosaccharides containing zero to four
sialic acids were eluting in separate regions, with consecutively increased reten-
tion times, as shown for the HILIC-IEX method in Figure 21.9.
21.3.5 Separation of Glucosinolates by a ZIC-HILIC

Column and Detection by Absorbance
Glucosinolates, which are synthesized in plants, are a group of glucosides with
structures that include a sulfonated oxime, a thio-linked glycone, and an additional

UV chromatogram (220 nm)

300 Glycopeptides
200 Peptides
100
0
0 10 20 30 40 50 60 70 80 90 100
(A) Retention time (min)
Mass spectrum (0–60 min)

6000
838
Peptides
905
1275
994
1037
Intensity
1174
1384
4000
1069
587
962
2000 660
0
200 400 600 800 1000 1200 1400 1600 1800 2000
(B) m/z
Mass spectrum (60–100 min) Hex Hex
Glycopeptides Hex Hex

Intensity
5000
0
200 400 600 800 1000 1200 1400 1600 1800 2000
(C) m/z
Mass chromatograms
a-2
c-2
b-2 d-2 m/z 13022+ ± 1
m/z 13832+ ± 1
IgG-2
m/z 14642+ ± 1 Peptide sequence
e-2 2+
Intensity
m/z 1403 ± 1 = EEQFNSTFR

f-2 g-2 h-2 b-1
m/z 14842+ ± 1
a-1 d-1 m/z 15652+ ± 1
m/z 13182+ ± 1
c-1 m/z 13992+ ± 1
IgG-1
m/z 14802+ ± 1
e-1 2+
Peptide sequence
m/z 1419 ± 1 = EEQYNSTYR
f-1 m/z 15002+ ± 1
h-1 m/z 15812+ ± 1
60 65 70 75 80 85 90 95 100
(D) Retention time (min)
Figure 21.8 ZIC-HILIC separation and mass spectrometry analysis of tryptic glycopep-
tides from human immunoglobulin G (IgG). (A) UV (220â•›nm) chromatogram, (B) accumu-
lated mass spectrum, 0–60â•›m in, (C) accumulated mass spectrum, 60–100â•›m in, and (D) mass
chromatograms of molecular ions of major N-glycopeptides of IgG-1. The table shows struc-
ture, exact mass values, and observed integer mass values of PA-labeled N-glycans and major
N-glycopeptides from human IgG. (Reproduced from Takegawa, Y. et al., J. Chromatogr. A,
1113, 177, 2006. With permission from Elsevier.)

Summary of structures, annotations, exact mass values, and observed integer mass values
of PA N-glycans and major N-glycopeptides from human serum IgG
IgG-1 IgG-2
Structures PA N-glycans
N-glycopeptides N-glycopeptides
(n = 1–3) (R1 = PA)
(R2 = EEQYNSTYR) (R3 = EEQFNSTER)
a a-1 a-2
Rn E.m. 1540.6 E.m. 2633.0 E.m. 2601.1
Obs. 1541 (z = 1) Obs. 1318 (z = 2) Obs. 1302 (z = 2)
Rn b/c b-1/c-1 b-2/c-2

E.m. 1702.7 E.m. 2795.1 E.m. 2763.1
Obs. 1703 (z = 1) Obs. 1399 (z = 2) Obs. 1383 (z = 2)
Rn
d
d-1 d-2
E.m. 1864.7
Rn E.m. 2957.1 E.m. 2925.2
Obs. 933 (z = 2),
Obs. 1480 (z = 2) Obs. 1464 (z = 2)
1865 (z = 1)
e e-1 e-1
E.m. 1743.7
Rn E.m. 2836.1 E.m. 2804.1
Obs. 873 (z = 2),
Obs. 1419 (z = 2) Obs. 1403 (z = 2)
1744 (z = 1)
Rn f/g
f-1 f-2/g-2
E.m. 1905.7
E.m. 2998.2 E.m. 2966.2
Obs. 954 (z = 2),
Obs. 1500 (z = 2) Obs. 1484 (z = 2)
Rn 1906 (z = 1)
h h-1 h-2
Rn E.m. 2067.8 E.m. 3160.2 E.m. 3128.2
Obs. 1035 (z = 2) Obs. 1581 (z = 2) Obs. 1565 (z = 2)
( ) Mannose (Man); ( ) Galactose (Gal); ( ) Fucose (Fuc); ( ) N-acetyl-glucosamine (GlcNAc);

PA: 2-aminopyridine; E.m.: exact mass; Obs.: Observed mass; z: number of charges.
side-group R. Glucosinolates are precursors of the cancer-protective isothiocyanates.

Wade et al.37 used a zwitterionic ZIC-HILIC column (150 × 4.6â•›mm ID, 5â•›μm particle
size, from SeQuant/Merck) at 25°C with elution by 70% acetonitrile, in water and
15â•›mM ammonium formate, pH 4.5, flow rate 0.5â•›mL/min, to separate glucosinolates
(Figure 21.10). Detection was performed by absorbance at 235â•›nm. The results from
the ZIC-HILIC column were more reproducible than those from a PolyHydroxyethyl
A column.
21.3.6 Separation of GAGs by an Amide-80 Column

Combined with MS Analysis
GAGs are composed of long, unbranched polysaccharides of a repeating disaccha-
ride unit and are the important components in connective tissue, including connec-
tive joint tissues. The analysis of GAGs is important to research in several fields,
including studies of osteoarthritis.

AA
Artifacts
3a
2
Fluorescence
1
4
4a
0SA 1SA 2SA 3SA 4SA
0 20 40 60 80
Time (min)
Figure 21.9 Fetuin oligosaccharide map of 2-aminobenzoic acid (2-AB)-labeled oligo-

saccharides separated by HILIC on an amino polymeric column. Bars indicate the regions
where glycans with zero to four sialic acids (SA) elute. (Reproduced from Anumula, K.R. and
Dhume, S.T., Glycobiology, 8, 685, 1998. With permission from Oxford University Press.)
GR GI
0 5 10 15 20
Minutes
Figure 21.10 The separation of glucosinolates by a ZIC-HILIC column using 70% ace-
tonitrile in water for elution. GR and GI denote glucoraphanin (4-(methylsulfonyl) butyl glu-
cosinolate) and glucoiberin (3-(methylsulfonyl) butyl glucosinolate), respectively. Detection
by absorbance at 235â•›nm. (Reproduced from Wade, K.L. et al., J. Chromatogr. A, 1154, 469,
Hitchcock et al.43 used an Amide-80 capillary column (150 × 0.25â•›m m ID,

5â•›μm, from Tosoh) together with a dual gradient (0–50â•›m in), 82%–46% acetonitrile
in water and 10–30â•›m M formic acid, pH 4.4, at a flow rate of 0.1â•›m L/min for elu-
tion (Figure 21.11). MS-MS was used for detection and identification. GAGs were
extracted from joint tissues and partly depolamirized, using chondroitinase. The
oligosaccharide products were labeled with differently stable isotopes, using (d0 or
d4) 2-anthranilic acid before the HILIC MS-MS analysis. By this method, the struc-
tures of different regions of chondroitin sulfate from human and bovine joint tissues
were determined.

×105 Δ(2,2,2)
Δ(3,3,3) + (3,3,3)
1.5
Δ(1,1,1)
Intens. 1.0
Δ(4,4,4)
(1,2,2)
Δ(2,2,1)
Δ(3,3,2)
Δ(5,5,5)
(2,3,3)
Δ(6,6,6)
0.5
0.0
(A) 20 25 30 35 40 45 50 55 60
Time (min)
Δ(2,2,2)
1000 520.6 Δ(1,1,1)
583.0
800
605.0 Δ(1,1,1)+Na
759.1 (3,3,3)
Intens.
563.0 Δ(1,1,1)-tris
600
740.1 Δ(3,3,3)-tris
499.8 Δ(3,3,3)
441.5 (1,2,2)
529.5 (2,2,2)
710.1 Δ(3,3,2)
Δ(2,2,2)-tris
480.5 Δ(2,2,1)
400
459.0
Δ(4,4,4)
200
0
(B) 400 450 500 550 600 650 700 750
m/z
(1,2,2) Δ(3,3,3)
(2,2,2) Δ(4,4,4)
(2,3,3) Δ(5,5,5)
(3,3,3) Δ(6,6,6)
Δ(1,1,1) L4(1,2,1,0)
Δ(2,2,1) L4(1,2,1,1)
Δ(2,2,2) L6(1,2,2,1,0)
Δ(2,2,3) L6(1,2,2,1,1)
Δ(3,3,2) L6(1,2,2,1,2)
(C)
Δ-unsaturated Hexuronic acid
glucuronic acid
N-acetylgalactosamine Galactose
Xylose Sulfate
Figure 21.11 The separation of glycosaminoglycans (GAGs) by an Amide-80 capillary

column, using a negative acetonitrile gradient, in water, and a positive formic acid gradient,
combined with MS detection. (A) Base peak chromatogram of chondroitin lyase depolami-
rized chondroitin sulfate and dermatan sulfate from d4-2-AA-juvenile bovine cartilage mixed
with d0-2-AA chondroitin sulfate internal oligosaccharide standard. GAG oligosaccharide
chains ranging from disaccharide to dodecasaccharide elute at 15–55â•›m in. (B) Average mass
spectrum of all eluted oligosaccharides in the sample mixture. (C) Oligosaccharide structures
found in connective tissue samples. (Reproduced from Hitchcock, A.M. et al., Proteomics, 8,

References
1. Campbell MK, Farrell SO. Biochemistry, 6th edn. Thomson Brooks & Cole, Belmont,
CA, 2009.
2. Linden JC, Lawhead CL. 1975. Liquid chromatography of saccharides. J. Chromatogr.
105:125–133.
3. Alpert AJ. 1990. Hydrophilic-interaction chromatography for the separation of peptides,
nucleic acids and other polar compounds. J. Chromatogr. 499:177–196.
4. Langer SH, Pantages P, Wender I. 1958. Gas–liquid chromatographic separation of
phenols as trimethylsilyl ethers. Chem. Ind. (Lond.) 1664–1665.
5. Hansson GC, Karlsson H. 1993. Gas chromatography and gas chromatography-mass
spectrometry of glycoprotein oligosaccharides. Methods Mol. Biol. 14:47–54.
6. Kwon H, Kim J. 1993. Determination of monosaccharides in glycoproteins by reverse-
phase high-performance liquid chromatography. Anal. Biochem. 215:243–252.
7. Ambrosius M, Kleesiek K, Götting C. 2008. Quantitative determination of the gly-
cosaminoglycan delta-disaccharide composition of serum, platelets and granulo-
cytes by reversed-phase high-performance liquid chromatography. J Chromatogr. A
1201:54–60.
8. Gey MH, Unger KK. 1996. A strategy for chromatographic and structural analysis of
monosaccharide species from glycoproteins. Fresenius J. Anal. Chem. 356:488–94.
9. Nakagawa H, Kawamura Y, Kato K, Shimada I, Arata Y, Takahashi N. 1995. Identification
of neutral and sialyl N-linked oligosaccharide structures from human serum glycopro-
teins using three kinds of high-performance liquid chromatography. Anal. Biochem.
226:130–138.
10. Karlsson G, Bergman R. 2003. Determination of the distribution of molecular masses
of sodium hyaluronate by high-performance anion-exchange chromatography.
J. Chromatogr. A 986:67–72.
11. Bourne EJ, Lees EM, Weigel H. 1963. Paper chromatography of carbohydrates and
related compounds in the presence of benzeneboronic acid. J. Chromatogr. 11:253–257.
12. Herold DA, Boyd JC, Bruns DE, Emerson JC, Burns KG, Bray RE et al. 1983.
Measurement of glycosylated hemoglobins using boronate affinity chromatography.
Ann. Clin. Lab. Sci. 13:482–8.
13. Buzzega D, Maccari F, Volpi N. 2008. Fluorophore-assisted carbohydrate electropho-
resis for the determination of molecular mass of heparins and low-molecular-weight
(LMW) heparins. Electrophoresis 29:4192–4202.
14. Soga T, Heiger DN. 1998. Simultaneous determination of monosaccharides in glycopro-
teins by capillary electrophoresis. Anal. Biochem. 261:73–78.
15. Liu CY, Chen TH, Misra TK. 2007. A macrocyclic polyamine as an anion receptor
in the capillary electrochromatographic separation of carbohydrates. J. Chromatogr.
A 1154:407–415.
16. Santori F, Hubble J. 2003. Isocratic separation of monosaccharides using immobilized
Concanavalin A. J. Chromatogr. A 1003:123–6.
17. Herbreteau B, Lafosse M, Morin-Allory L, Dreux M. 1990. Analysis of sugars by super-
critical fluid chromatography using polar packed columns and light-scattering detection.
J. Chromatogr. 505:299–305.
18. Panagiotopoulos C, Sempéré R, Lafont R, Kerhervé P. 2001. Sub-ambient temperature
effects on the separation of monosaccharides by high-performance anion-exchange
chromatography with pulsed amperometric detection. Application to marine chemistry.
J. Chromatogr. A 920:13–22.
19. Cataldi TR, Campa C, De Benedetto GE. 2000. Carbohydrate analysis by high-
performance anion-exchange chromatography with pulsed amperometric detection:
The potential is still growing. Fresenius J. Anal. Chem. 368:739–758.

20. Campa C, Coslovi A, Flamigni A, Rossi M. 2006. Overview on advances in capillary

electrophoresis-mass spectrometry of carbohydrates: A tabulated review. Electrophoresis
27:2027–2050.
21. Klampfl CW. 2004. Review coupling of capillary electrochromatography to mass spec-
trometry. J. Chromatogr. A 1044:131–144.
22. Zaia J. 2008. Mass spectrometry and the emerging field of glycomics. Chem. Biol.
15:881–892.
23. Schols HA, Mutter M, Voragen AG, Niessen WM, van der Hoeven RA, van der Greef J
et al. 1994. The use of combined high-performance anion-exchange chromatography-
thermospray mass spectrometry in the structural analysis of pectic oligosaccharides.
Carbohydr. Res. 261:335–342.
24. Alpert AJ, Shukla M, Shukla AK, Zieske LR, Yuen SW, Ferguson MA et al. 1994.
Hydrophilic-interaction chromatography of complex carbohydrates. J. Chromatogr. A
676:191–202.
25. Hemström P, Irgum K. 2006. Hydrophilic interaction chromatography. J. Sep. Sci.
29:1784–1821.
26. Ikegami T, Tomomatsu K, Takubo H, Horie K, Tanaka N. 2008. Separation efficiencies
in hydrophilic interaction chromatography. J. Chromatogr. A 1184:474–503.
27. Dejaegher B, Mangelings D, Vander Heyden Y. 2008. Method development for HILIC
assays. J. Sep. Sci. 31:1438–1448.
28. Hao Z, Xiao B, Weng N. 2008. Impact of column temperature and mobile phase com-
ponents on selectivity of hydrophilic interaction chromatography (HILIC). J. Sep. Sci.
31:1449–1464.
29. Snyder LR, Barker J, Ando DJ, Kirkland JJ, Glajch JL. 2002. Practical HPLC Method
Development: With Mass Spectrometry, 2nd revised edn. Wiley Interscience, New York.
30. Cohen S, Lhuillier F, Mouloua Y, Vignal B, Favetta P, Guitton J. 2007. Quantitative
measurement of propofol and in main glucuroconjugate metabolites in human plasma
using solid phase extraction-liquid chromatography-tandem mass spectrometry.
J. Chromatogr. B 854:165–172.
31. Karlsson G, Swerup E, Sandberg H. 2008. Combination of two hydrophilic interaction
chromatography methods that facilitates identification of 2-aminobenzamide-labeled
oligosaccharides. J. Chromatogr. Sci. 46:68–73.
32. Karlsson G, Winge S, Sandberg H. 2005. Separation of monosaccharides by hydrophilic
interaction chromatography with evaporative light scattering detection. J. Chromatogr.
A 1092:246–249.
33. Kakita H, Kamishima H, Komiya K, Kato Y. 2002. Simultaneous analysis of monosac-
charides and oligosaccharides by high-performance liquid chromatography with post-
column fluorescence derivatization. J. Chromatogr. A 961:77–82.
34. Tolstikov VV, Fiehn O. 2002. Analysis of highly polar compounds of plant origin:
Combination of hydrophilic interaction chromatography and electrospray ion trap mass
spectrometry. Anal. Biochem. 301:298–307.
35. Anumula KR, Dhume ST. 1998. High resolution and high sensitivity methods for
oligosaccharide mapping and characterization by normal-phase high performance
liquid chromatography following derivatization with highly fluorescent anthranilic
acid. Glycobiology 8:685–694.
36. Takegawa Y, Deguchi K, Keira T, Ito H, Nakagawa H, Nishimura S. 2006. Separation of
isomeric 2-aminopyridine derivatized N-glycans and N-glycopeptides of human serum
immunoglobulin G by using a zwitterionic type of hydrophilic-interaction chromatogra-
phy. J. Chromatogr. A 1113:177–181.
37. Wade KL, Garrard IJ, Fahey JW. 2007. Improved hydrophilic interaction chromatogra-
phy method for the identification and quantification of glucosinolates. J. Chromatogr.
A 1154:469–472.

38. Ikegami T, Horie K, Saad N, Hosoya K, Fiehn O, Tanaka N. 2008. Highly efficient
analysis of underivatized carbohydrates using monolithic-silica-based capillary hydro-
philic interaction (HILIC) HPLC. Anal. Bioanal. Chem. 391:2533–2542.
39. Armstrong DW, Jin HL. 1989. Evaluation of the liquid chromatographic separation
of monosaccharides, disaccharides, trisaccharides, tetrasaccharides, deoxysaccharides
and sugar alcohols with stable cyclodextrin bonded phase columns. J. Chromatogr.
462:219–232.
40. Rogatsky E, Jayatillake H, Goswami G, Tomuta V, Stein D. 2005. Sensitive LC MS
quantitative analysis of carbohydrates by Cs+ attachment. J. Am. Soc. Mass Spectrom.
16:1805–1811.
41. Antonio C, Larson T, Gilday A, Graham I, Bergström E, Thomas-Oates J. 2008.
Hydrophilic interaction chromatography/electrospray mass spectrometry analysis of
carbohydrate-related metabolites from Arabidopsis thaliana leaf tissue. Rapid Commun.
Mass Spectrom. 22:1399–1407.
42. Karlsson G, Hinz AC, Winge S. 2004. Determination of the stabilizer sucrose in a
plasma-derived antithrombin process solution by hydrophilic interaction chromatogra-
phy with evaporative light-scattering detection. J. Chromatogr. Sci. 42:361–365.
43. Hitchcock AM, Yates KE, Costello CE, Zaia J. 2008. Comparative glycomics of connec-
tive tissue glycosaminoglycans. Proteomics 8:1384–1397.
44. Charlwood J, Birrell H, Bouvier ES, Langridge J, Camilleri P. 2000. Analysis of oli-
gosaccharides by microbore high-performance liquid chromatography. Anal. Chem.
72:1469–1474.
45. Eliasson AC (ed.). 2006. Carbohydrates in Food, 2nd edn. Taylor & Francis, London, U.K.
46. Brons C, Olieman C. 1983. Study of the high-performance liquid chromatographic sepa-
ration of reducing sugars, applied to the determination of lactose in milk. J. Chromatogr.
259:79–86.
47. David V, Moldoveanu SC (eds). 2002. Sample preparation in chromatography. Journal
of Chromatography Library, Vol. 65. Elsevier Science, Amsterdam, the Netherlands.
48. Morin-Allory L, Herbreteau B. 1992. High-performance liquid chromatography and
supercritical fluid chromatography of monosaccharides and polyols using light-scattering
detection: Chemometric studies of the retentions. J. Chromatogr. 590:203–213.
49. Wei Y, Ding M. 2000. Analysis of carbohydrates in drinks by high-performance liquid
chromatography with a dynamically modified amino column and evaporative light scat-
tering detection. J. Chromatogr. A 904:113–117.
50. Anumula KR. 2006. Advances in fluorescence derivatization methods for high-performance
liquid chromatographic analysis of glycoprotein carbohydrates. Anal. Biochem. 350:1–23.
51. Signal 2-AB labeling kit, Prod. No. GKK-404. Product information from ProZyme, San
Leandro, CA.
52. Corcoran O, Spraul M. 2003. LC-NMR-MS in drug discovery. Drug Discov. Today
8:624–631.
53. Montero CM, Dodero MC, Sanchez DA, Barroso CG. 2004. Analysis of low molecular
weight carbohydrates in food and beverages: A review. Chromatographia 59:15–30.
54. Panagiotopoulos C, Sempéré R. 2005. Analytical methods for the determination of sug-
ars in marine samples: A historical perspective and future directions. Limnol. Oceanogr.
Methods 3:419–454.
55. Friedman M. 2004. Analysis of biologically active compounds in potatoes (Solanum
tuberosum), tomatoes (Lycopersicon esculentum), and jimson weed (Datura stramo-
nium) seeds. J. Chromatogr. A 1054:143–155.
56. Sasisekharan R, Raman R, Prabhakar V. 2006. Glycomics approach to structure-function
relationships of glycosaminoglycans. Annu. Rev. Biomed. Eng. 8:181–231.
57. Brooks SA. 2006. Protein glycosylation in diverse cell systems: Implications for modi-
fication and analysis of recombinant proteins. Expert Rev. Proteomics 3:345–359.

in the Characterization
of Glycoproteins
Joanne E. Nettleship
Contents
22.1 Introduction to Glycoproteins........................................................................ 524
22.1.1 Importance of Glycosylation in Biological Systems and
How Variation Can Lead to Disease................................................. 524
22.1.2 Types of Glycosylation and Glycan Patterns..................................... 525
22.1.3 Challenges in Glycosylation Analysis by MS.................................... 527
22.1.3.1 Mass Spectrometry............................................................. 527
22.1.3.2 MALDI-MS........................................................................ 527
22.1.3.3 ESI-MS............................................................................... 528
22.2 Characterization of Glycoproteins................................................................. 528
22.2.1 Glycoprotein Enrichment from Complex Solutions.......................... 528
22.2.1.1 Lectin Affinity Chromatography........................................ 529
22.2.1.2 Capillary Electrophoresis................................................... 530
22.2.1.3 Other Separation Techniques.............................................. 530
22.2.2 Glycopeptide Enrichment.................................................................. 531
22.2.2.1 Hydrophilic Interaction Liquid Chromatography............... 531
22.2.2.2 Lectin Affinity Chromatography........................................ 534
22.2.2.3 Hydrazine Chemistry.......................................................... 534
22.2.2.4 Ion-Pairing Normal-Phase Chromatography...................... 535
22.2.2.5 Others.................................................................................. 535
22.2.3 Tandem Methods............................................................................... 536
22.3 Examples of Glycoprotein Characterization Using HILIC........................... 536
22.3.1 Glycoprotein Enrichment................................................................... 537
22.3.2 Characterization of Glycoforms........................................................ 537
22.3.3 Characterization of Glycosylation Site Occupancy........................... 538
22.3.4 Full Characterization of Glycoproteins.............................................540
22.4 Summary....................................................................................................... 542
References............................................................................................................... 542
523

22.1 Introduction to Glycoproteins

Glycosylation is one of the most common posttranslational modifications of proteins—
it is estimated that over 50% of human proteins are glycosylated.1 Glycoproteins have
great importance in biological systems with the glycan moiety acting as a key to the
correct folding and action of the protein.2 The characterization of glycoproteins is
important in the diagnosis and treatment of human diseases. For biopharmaceutical
industry, consistent generation of the same glycoforms in different batches of recombi-
nant protein is essential to retain the drug activity; thus glycoprotein analysis plays an
important role in the production process. In addition, characterization of glycoproteins
from patients with a specific disease can lead to the discovery of new biomarkers or
targets for drug development.
This review discusses the importance of glycoproteins in biological systems in
Section 22.1.1 along with a review of glycan classes in Section 22.1.2. Some of the
challenges faced in characterizing glycoproteins and the use of mass spectrometry
(MS) for this analysis are given in Section 22.1.3. Various enrichment and separation
techniques for both glycoproteins and glycopeptides are discussed in Section 22.2.
Finally, some examples in the use of hydrophilic interaction liquid chromatography
(HILIC) for glycoprotein analysis are given in Section 22.3.
22.1.1 Importance of Glycosylation in Biological Systems

and How Variation Can Lead to Disease
Carbohydrates have been shown to be involved in many biological processes such

as inter- and intracellular signaling, immune functions, cell adhesion and division,
viral replication, parasitic infections, and protein regulation.3–7 It is the complexity of
the glycan chains (discussed in Section 22.1.2) that enables glycoproteins to encode
a large amount of information for specific molecular recognition. Glycosylation also
determines protein folding, stability, and kinetics.2 For example, Limjindaporn et al.
have observed the interaction of the glycosylated envelope protein of dengue virus
with chaperones in the human host, namely, immunoglobulin heavy chain–binding
protein, calnexin, and calreticulin.8 These interactions were shown to be important
in virion production since decreasing expression of the chaperones reduced the pro-
duction of infectious virions.8 In addition, Rudd et al. comment in their review that
almost all the key proteins involved in immune response are glycoproteins.5 For
example, the folding and assembly of both peptide-loaded major histocompatibil-
ity complex antigens and the T cell receptor complex involve the use of specific
glycoforms.5
Glycosylation is found in all eukaryotes and has also been observed in bacteria
and archaea.9,10 For instance, the intestinal bacterium Bacteroides fragilis uses an
O-glycosylation system for the incorporation of exogenous fucose into proteins in
order to successfully colonize the mammalian gut.11 In addition, O-glycosylation
of pilin has been shown to occur in Neisseria gonorrhoeae whereas glycosylation-
deficient mutants show the attenuated virulence.12,13
Glycosylation has been implicated in many disease states, such as cancer, athero-
sclerosis and rheumatoid arthritis, inflammatory diseases, and congenital disorders.14–19

HILIC in the Characterization of Glycoproteins 525
In neurodegenerative diseases, for example, Creutzfeldt–Jakob disease, Alzheimer’s

disease and Parkinson’s disease, glycosylation modification of vital enzymes has been
shown to be important.18 A key enzyme, acetylcholinesterase has been shown to take
altered glycosylation in the postmortem brain and cerebrospinal fluid of Alzheimer’s
patients.20 In addition, there is altered glycosylation of a related enzyme, butylcho-
linesterase, in cerebrospinal fluid.20 Unfortunately, the sensitivity is too low to use as
a biomarker for diagnosing Alzheimer’s disease from other types of dementia using
these alterations.21
Glycosylation has also been implicated in the transmissibility of influenza A
viruses. The ability of the virus to adapt to a human host is governed by the binding
specificity of the viral surface hemagglutinin to long α2–6 sialylated glycan recep-
tors in the tissues of human upper respiratory tract.22 Srinivasan et al. showed that
a single amino acid mutation in the viral hemagglutinin led to a change in the bind-
ing specificity from α2–6 sialylated to mixed α2–3/α2–6 sialylated glycans, which
resulted in inefficient transmission of the virus.22 The subtle difference in glycan
binding changes the human-to-human transmissibility of the virus and could act as
an indication of the probability of an influenza pandemic.
In addition to having a role in disease, glycosylation is being used as a target for
vaccine development, for example, in AIDS23,24 and in cancer.25 For instance, Ni et al.
are examining the possibility of a carbohydrate-based HIV-1 vaccine by synthesizing
oligomannose-containing glycoconjugates, which resemble a cluster on HIV-1â•›gp120.24
In the field of cancer immunotherapy, several carbohydrate antigens are vaccination
targets.25 Slovin et al. have used Thomsen–Friedenreich (TF) self-antigen, a core
disaccharide of O-glycosylated complex glycoproteins expressed on malignant cells,
in clinical trials as a target for specific delivery of a synthetic TF cluster-keyhole limpet
hemocyanin conjugate vaccine in the treatment of prostate cancer.26
22.1.2 Types of Glycosylation and Glycan Patterns

The glycans are attached to glycoproteins through specific amino acids, namely,
asparagine of N-glycosylation, and serine or threonine of O-glycosylation. There are
some rare examples of other glycan linkages such as covalent O-glycosidic linkages
with hydroxylysine in collagens27 and through lysine residues as seen in diabetes.28
N-linked glycans mostly contain an N-acetylglucosamine (GlcNAc) joined via
an amide bond to the asparagine residues of a protein (Figure 22.1), whereas, most
commonly (e.g., mucins), O-glycosylation occurs via an N-acetylgalactosamine
(GalNAc) moiety linked to the hydroxyl group of either serine or threonine (Figure
22.2). There are other, rare linkages such as O-GlcNAc in cytosolic and nuclear
proteins,29 fucose (Fuc)-serine linkages in epidermal growth factor domains,30 and
mannose (Man)-serine linkages found in proteins from Saccharomyces cerevisiae31
and Pichia pastoris32 as well as glycoproteins found in the brain.33
N-linked glycans consist of a core unit, which becomes elaborated with variable
antennae regions (Figure 22.1) that are formed by addition or deletion of sugar units
to a common precursor. There are three subgroups of N-glycans classified according
to the type and location of the sugar residues added to the core. Glycans trimmed
by α-glucosidases and α-mannosidases without any additions contain only mannose

Man Man
Man
Man Man Man GlcNAc GlcNAc Asn
Man
Man Man
(A)
NeuAc Gal GlcNAc Man Fuc
Man GlcNAc GlcNAc Asn
NeuAc Gal GlcNAc Man
(B)
Man
Man
Man Man GlcNAc GlcNAc Asn
NeuAc Gal GlcNAc Man
(C)
Figure 22.1 Examples of the three classes of N-glycans showing (A) high mannose,
(B) complex, and (C) hybrid-type glycans.
GalNAc Ser/Thr GalNAc Ser/Thr

Gal GlcNAc
GlcNAc GlcNAc
GalNAc Ser/Thr GalNAc Ser/Thr
Gal GlcNAcl
Figure 22.2 Four of the common O-glycan core structures.
attached to the core structure with the composition of Man5–9GlcNAc2. These glycans
are commonly known as “high-mannose type” (Figure 22.1A): Glycans containing
N-acetyllactosamine (LacNAc composed of Gal followed by GlcNAc) in their anten-
nal region are classified as “complex type” (Figure 22.1B). These complex glycans
have a variable number of antennae formed by addition of GlcNAc to the α-mannose
residues of the core structure. In mammalian proteins, the antennae are often com-
posed of tandem repeats of LacNAc formed by sequential addition of GlcNAc and
Gal residues. The third class of glycans is the “hybrid type,” which contains both a
mannose and a LacNAc attached to the glycan core (Figure 22.1C). Complex-type
glycans are usually capped by an α-linked sialyl or fucosyl group—the backbone
sugars are generally β-linked. In addition to these subclasses of N-glycosylation pat-
terns, a GlcNAc residue can be attached to the 4-position of the β-mannose resulting
in a bisecting residue. A fucose (Fuc) moiety can also be added to the core GlcNAc
residue. The presence or absence of this core-fucosylation further adds to the diver-
sity of glycan structures (Figure 22.1B).
The structures of O-linked glycans are even more diverse than N-linked glycans34
with sizes ranging from single monosaccharides to large polysaccharides similar
in size to the complex-type N-glycans. Large O-glycans can be divided into two
domains, the core and the antennae. There are four mucin-type O-glycan core struc-
tures, which are the most widespread in mammalian glycoproteins (Figure 22.2)

though these are by no means the limit of O-glycan core structure diversity.35 This
structural variation continues with elongation of the core structures with a variety of
sugar residues such as Gal, Fuc, GlcNAc.
In addition to the size and structure of the glycan, glycoproteins show variation in
the occupancy of glycosylation sites. A potential glycosylation site in a protein may
be unoccupied due to steric limitations at the site, possibly as a result of another gly-
cosylation site being in close proximity. Moreover, more than one glycan structure
may exist at a glycosylation site within a glycoprotein population. This variation in
the structure of the attached glycan gives rise to protein glycoforms.
22.1.3 Challenges in Glycosylation Analysis by MS

Compared to the analysis of non-glycosylated proteins, glycoprotein analysis faces
a greater challenge due to the complexity and heterogeneity of the oligosaccharide
chains. This challenge is highlighted by the fact that over 50% of proteins in the
Swiss-Prot database are predicted to be glycosylated, though only around 10% of
these glycoproteins have been characterized.1 Several steps are usually needed for
the complete structural analysis of a glycoprotein: identification of the protein and
the analysis of protein amino acid backbone, determination of the glycosylation sites
and the occupancy of these sites, and analysis of the glycan structure and quantita-
tion of each glycan attached to a particular glycosylation site.
N-glycans can be released from the protein backbone by treatment with endogly-
cosidases such as peptide-N-glycosidase F (PNGase F), which removes the glycan and
converts the asparagine residue to an aspartic acid,36 or a combination of endo-β-N-
acetylglucosaminidases that leave one GlcNAc residue at the glycosylation site.37 The sit-
uation is more challenging with O-glycosylation sites as there are no universal enzymes
for the removal of all the O-glycans. However, the glycans can be removed by alkaline
β-elimination38 or by base-catalyzed β-elimination in the presence of ethylamine.39
22.1.3.1 Mass Spectrometry

MS in combination with a separation technique is one of the most powerful tools for
characterization of glycoproteins because it provides structural information about
the glycosylation with low sample consumption.40 Although MS responses of glyco-
proteins and glycopeptides are normally weaker than non-glycosylated analytes,35,41
MS is the principle analysis technique used in glycoproteomics. There are two main
mass spectrometric ionization methods used in the characterization of glycoproteins:
matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization
(ESI). Dalpathado and Desaire have recently reviewed the use of MS for glycosyl-
ated proteins,42 and hence only a brief discussion is included in this chapter.
22.1.3.2 MALDI-MS
MALDI was developed for the analysis of large molecular weight molecules and can
be used for the analysis of proteins as well as oligosaccharides. The method is very
sensitive, and samples can be analyzed at femtomol to picomol levels. MALDI can
tolerate a wide variety of sample compositions including most biological buffers.40 For
analysis by MALDI, the sample is mixed with a matrix such as 2,5-dihydroxybenzoic

acid, which absorbs photon energy at an irradiation laser wavelength (often a nitrogen
laser at 337â•›nm). Ionization occurs during laser pulsing in which the matrix transfers
the absorbed energy to the analyte. This results in predominantly singly charged ana-
lyte ions, which are subsequently directed toward the mass analyzer.
MALDI-TOF (MALDI time-of-flight) analysis of intact glycoproteins generally
yields information on average carbohydrate content43 due to the glycan heterogene-
ity, although useful spectra have been obtained.44–46 Isotopic resolution of glycopro-
teins can be achieved for proteins with a molecular mass under about 10â•›k Da.40 The
correct choice of matrix is essential to gain the best result possible for intact glyco-
proteins; for instance, larger glycoproteins give enhanced signals when sinapinic
acid is used as the matrix.47
The characterization of glycoproteins often involves digestion of the protein using
a protease such as trypsin and analysis of the resultant glycopeptides. These glyco-
peptides can be analyzed using MALDI-MS in a number of ways: the intact glyco-
peptide can be measured to facilitate the determination of the existence of a glycan,
the peptide backbone can be fragmented giving information on the peptide sequence
and glycosylation site, and the oligosaccharide can be fragmented to reveal the gly-
can structure. In some cases, multidimensional chromatography is needed prior to
the analysis of glycopeptides from a digestion mixture containing several glycopro-
teins, for example, cellular extracts, in order to reduce sample complexity.
22.1.3.3 ESI-MS
The other ionization method for MS, which is used extensively in protein analysis, is ESI.
This is a soft ionization technique, which normally generates multiple charged ion species
for the mass measurement of higher molecular weight proteins. Usually, ESI produces
little or no fragmentation in the ionization source. In ESI, the sample is ionized through a
metal capillary at a high voltage. The electric field generates a mist of charged droplets as
the sample emerges from the capillary. The solvent is gradually evaporated with the aid
of heat and gas flow, thus the droplet size is constantly decreased until naked sample ions
are generated, which are then directed into the analyzer for mass measurement.
As with MALDI, analysis of intact glycoproteins via ESI-MS results in an average
molecular weight. Glycoforms become more difficult to be completely resolved when
protein mass and/or glycosylation sites increase, although smaller glycoproteins have
been fully analyzed by this method.48–50 The characterization of glycopeptides can be
performed using ESI-MS with or without fragmentation to yield information on the pep-
tide backbone as well as the attached glycans. The ESI-MS is usually coupled with a
chromatography system allowing for separation of peptides immediately before analysis.
22.2 Characterization of Glycoproteins

22.2.1 Glycoprotein Enrichment from Complex Solutions
The heterogeneity of oligosaccharide chains in glycoproteins leads to difficulties in
resolving the structural differences of glycoproteins and in particular of glycoforms
of the same protein. The challenge is increased when the glycoprotein is present
in cellular extracts or fluids, as the protein of interest is often available in minute

Table 22.1
Examples of the Separation of Glycoproteins from Complex Solutions
Source of
Enrichment Technique Glycoprotein(s) Glycoprotein Reference
Concanavalin A lectin Urine glycoproteins Human urine 51
affinity chromatography
Mixed lectin affinity Glycoproteome analysis Human serum 65
chromatography
Concanavalin A magnetic RNase B (spiked into sample) Human serum 63
beads
Capillary zone Trisialotransferrin Human serum 70
electrophoresis
Capillary isoelectric Alpha-1-acid glycoprotein Human serum 135
focusing
Capillary gel Influenza A virus Mammalian cell culture 136
electrophoresis glycoproteins
Micellar electrokinetic Antithrombin Human plasma 137
chromatography
2D SDS-PAGE Liver cancer glycoproteins Human serum 138
Hydrophilic interaction IgE-FcεRIα Mammalian cell culture See Section
chromatography 22.3.1
Phenylboranate Erythropoietin Mammalian cell culture 84
chromatography
quantities.51,52 Conventional chromatographic techniques for protein separation such

as ion exchange, size exclusion, and reversed-phase chromatography can be used
for glycoprotein analysis.53 Some of the common separation techniques specific for
glycoproteins are discussed in this section, with Table 22.1 giving some examples.
22.2.1.1 Lectin Affinity Chromatography

The use of lectin affinity chromatography for glycoprotein separation was first
described by Donnelly and Goldstein in 197054 and remains one of the most widely
used methods for the purification of glycoproteins. There are several recent reviews
on this topic55–57 and only a brief description is included here. The sugar chains of a
glycoprotein are bound to lectin, which is immobilized on a matrix; unbound material
is then washed away prior to the elution of the retained glycoprotein. Numerous plant
lectins, in immobilized format, are commercially available, which can be used for gly-
can binding. However, a majority of the lectins are selective for a certain sugar type.
A mixture of glycopeptides can be purified using a lectin with broad specificity such as
wheat germ agglutinin or concanavalin A,58 or a specific lectin can be used for targeted
purification of glycopeptides. For example, lentil lectin only binds high mannose gly-
cans with core fucosylation.59 Broad specificity is of particular use in the initial steps
of a purification protocol or in glycoproteomics where the aim is to identify all the gly-
coproteins from a given species.60,61 The choice of lectins with specific sugar binding

affinity has to be made through screening exercises where the most appropriate immo-
bilized lectin for isolation of a particular glycoform can be discovered.62 This method
of lectin glycan profiling can be miniaturized and has proved useful in determining the
types of oligosaccharide attached to a protein.60 In addition to miniaturization, lectins
are also attached to magnetic beads to facilitate automation of such protocols.63
Lectin affinity chromatography may be used in series to increase the specificity
of a purification protocol. For example, Qiu and Regnier55 coupled a concanavalin A
column to a Sambucus nigra agglutinin column to study the degree of sialylation in
human serum glycoproteins. In addition, different lectins may be mixed together to
achieve the specificity required.64 For example, Yang and Hancock used a mixture of
the agarose-bound lectins concanavalin A, wheat germ agglutinin and Jacalin lectin,
to study the glycoproteome from human serum.65
Although a selected lectin can be specific to a certain type of glycan, lectin affin-
ity chromatography often lacks the desired specificity for targeted glycans because
other oligosaccharides may interact weakly with the matrix and are co-purified
with the glycoprotein of interest.66 Therefore, only limited information on structure
of the glycan chain can be obtained using lectin affinity chromatography. Despite
this disadvantage, lectin chromatography is of great importance in the enrichment
of glycoproteins from complex solutions and is often the first technique employed
in a series of experiments for the analysis of glycoproteins (Section 22.2.3).
22.2.1.2 Capillary Electrophoresis

High-performance capillary electrophoresis (CE) is a useful technique for the sepa-
ration of glycoforms of a particular protein due to its high resolving power. Several
modes of separation can be used for different applications, such as capillary zone
electrophoresis (CZE), capillary isoelectric focusing (CIEF), capillary gel electro-
phoresis (CGE), and micellar electrokinetic chromatography (MEKC), with some
examples given in Table 22.1. In general, CE consists of a fused silica or glass cap-
illary that is often modified with a coating such as a covalently bonded polymer.67
The glycoforms are separated using either a positive or negative voltage depending
on the coating and background electrolyte. For instance, a negative voltage would be
used for glycoprotein separation with a fused silica capillary containing a polybrene
coating with 20% MeOH, 1â•›M acetic acid as the background electrolyte.68 CE is also
used in a microchip format, thus making it amenable for the separation of very low
abundance glycoproteins and offering a shorter analysis time.69
CE has been used extensively to monitor glycoform populations in the production of
recombinant glycoproteins for therapeutic use.67,70 CE can also be used as a diagnostic
tool.71 In addition to the separation of glycoforms, CE has been used to separate glyco-
proteins from complex sample solutions.72,73 For example, CE has been applied to the
separation of transferrin from albumin-depleted human serum samples obtained either
from healthy patients or those with congenital disorder of glycosylation.74
22.2.1.3 Other Separation Techniques

In addition to lectin affinity chromatography and CE, other methods that are often
employed for glycoprotein enrichment, including 2D SDS-PAGE, HILIC and phenyl-
boronate chromatography, are discussed here.

Two-dimensional (2D) electrophoresis is often used to separate proteins for analy-

sis by in-gel digestion followed by MS.75,76 When applied to glycoproteins, two prob-
lems with this technique must be taken into consideration: glycoproteins bind less
SDS than non-glycosylated proteins due to their carbohydrates and therefore have a
different mobility within the gel.77 Furthermore, glycoproteins tend to resolve into
more than one spot because of the differences in molecular weight and/or isoelectric
point caused by various glycoforms.35,77 The identification of glycoproteins from an
SDS gel image is aided by glycoprotein-specific stains such as periodic acid-Schiff
and Pro-Q Emerald® by Molecular Probes.78,79
Although HILIC can be used for the separation and concentration of intact glyco-
proteins (see Section 22.3.1),80,81 it is more commonly associated with glycopeptide
separations and will be discussed extensively in Section 22.2.2.1.
Glycoproteins can be enriched using an agarose-based resin with a phenylbo-
ronate ligand such as the m-aminophenylboronic acid beads commercialized by
Clontech.82 This ligand binds to the cis-diol groups of glycans such as mannose by
forming a 5-membered ring (Figure 22.3A). The protein can be released by lower-
ing the pH or using an elution buffer containing Tris or sorbitol.83 For example, this
technique has been used by Zanette et al. for the purification of erythropoietin from
mammalian cell cultures.84
22.2.2 Glycopeptide Enrichment

Glycopeptide separations face similar challenges as those for glycoprotein as both
contain oligosaccharide chains with great heterogeneity. One of the main issues
with glycopeptide identification is the low abundance of these glycopeptides—only
about 2%–5% of the peptides produced from a glycoprotein are glycosylated.85,86 In
addition, glycopeptides normally produce lower MS response compared with the
non-glycosylated peptides.35,41 The situation becomes much more difficult when a
complex sample such as human serum is digested and analyzed using a shotgun
proteomics approach—99% of the total protein mass of human serum is made up of
only 22 proteins.87 Thus glycopeptide information is masked by the heavy presence
of abundant non-glycosylated peptides.
Several methods of glycopeptide enrichment are described in the following text.
These methods can be used either for shotgun proteomics or as the second step in a
tandem approach to glycoprotein analysis.
22.2.2.1 Hydrophilic Interaction Liquid Chromatography

HILIC is a useful tool for the separation of very polar or hydrophilic compounds. The
stationary phase is hydrophilic and may be charged depending on the type of ligands
used as well as mobile-phase pH values. The aqueous component of the mobile phase
is an integral part of the matrix as a water-enriched liquid layer is established within
the stationary phase (Figure 22.4). The mobile phase for HILIC typically contains
water and acetonitrile, and is therefore amenable to mass-spectrometric analysis.
Separation by HILIC is achieved by peptides partitioning into the water-enriched sta-
tionary phase through both hydrogen bonding and dipole–dipole interactions.88 For
charged HILIC stationary phases such as zwitterionic interaction chromatography

R1 R1 CH2OH CH2OH
HO R2 O CH2OH
– R2 O Oxidation Coupling O
N B OH + 2 H 2O OH OH O
HO R3 H R3
R4 O
R4 HO O HO O
HO O N N Glycan
(A) Glycan O O Glycan
HN HN
H2N NH2
HN HN
(B)
OH O
HO OH OH
O
HO Y289L O O HN NH
HO O GalT O
NHAc Protein HO HO O O
Protein + H
NHAc
N O
HO OH + O S N NH2
O H
O O
HO NH
O O O O
O P P N O HO OH OH
O OO O O O O
– – HN NH
O
O HO HO O
OH OH H NHAc Protein
N N O N
S
H
(C) O
OH
HO COOH H COOH
NalO4 ONH2 COOH
O O Protein O H
R HO O Protein R NH
NH N O Protein
HO R NH
HO
HO
O
O O
(D)
Figure 22.3 Chemical modifications of glycoproteins and peptides. (A) Reversible interaction of 1,2-cis-diol sugars with phenylboronate beads. (B) Oxidation
of a carbohydrate to an aldehyde followed by coupling to hydrazine resin. (C) Scheme for the detection of O-GlcNAc glycosylated proteins where the final product
binds to streptavidin-HRP. (D) Selective oxidation of sialic acid–containing glycoproteins and oxime bond coupling to a solid support (R = H or OH).

H2O H2O Glycopeptide

– H2O
– SO3 –
O3S Glycopeptide SO3
CH2
CH2 H2O H2C
H2O CH2
H2O CH2 H2C
CH
CH2 + 2 H2C H 2O
+ CH3 H3 C N CH3 H3C +
H3C N CH
N CH
3
H2C H 2 O 2 H2 O
CH2
H2O
Figure 22.4 The structure of ZIC-HILIC resin showing the water-enriched liquid layer
and interactions with glycopeptides by both hydrophilic partitioning (left-hand side) and elec-
trostatic interactions (right-hand side).
(ZIC)-HILIC, the retention of analytes is influenced by the electrostatic interactions

imposed by the matrix on the water layer (Figure 22.4). Because the stationary phase
is zwitterionic, salts or buffers are not needed in the mobile phase since the electro-
static forces of the stationary phase are counterbalanced by the proximity of an ion
of the opposite charge (Figure 22.4).89
Glycopeptides are more hydrophilic than non-glycosylated peptides due to their
sugar chains; therefore, glycopeptides elute later than non-glycosylated peptides
during HILIC separation. As a result, HILIC is becoming a powerful technique in
the enrichment of glycopeptides in a sample. The retention of a glycopeptide on a
HILIC column depends on the hydrophilicity and size of the glycan chain but not on
the glycan type.90 However, some glycoforms can be resolved, such as the separation
of differentially sialylated glycopeptides from interferon-γ where the number of sialyl
groups correlated to the retention time of the glycopeptides on the HILIC column.91
In general, ZIC-HILIC has proved to be more useful than standard HILIC for
the separation of glycopeptides as the zwitterionic nature of ZIC-HILIC eliminates
the use of buffers in the mobile phase and is therefore easily interfaced with MS.92
This technique is now widely used in glycoproteomics37,93–97 to significantly increase
the relative abundance of glycopeptides over non-glycosylated peptides.98 Hagglund
et al. first used this method in glycoproteomics to reduce the complexity in a glyco-
protein digest sample by removal of non-glycosylated peptides. Upon the separation
of glycopeptides using the HILIC method, the analysis of the N-glycosylation sites
was successfully accomplished after the removal of the complex sugars to leave
one GlcNAc residue.37 Totally, 62 glycosylation sites were identified using HILIC
enrichment from a digest of the glycoproteins in human plasma.37 Kaji et al. ana-
lyzed the diversity of glycoproteins in Caenorhabditis elegans using an initial lec-
tin capture of glycoproteins followed by enrichment of glycopeptides by HILIC.95
In these experiments, 829 glycoproteins were identified, which contained 1465
N-glycosylation sites.95
For recombinant glycoproteins or simple protein mixtures, HILIC may be used
for the enrichment36,99 or for the separation of glycopeptides followed by the analysis
of the product with or without glycan removal.100 For example, offline ZIC-HILIC

enrichment of glycopeptides allows for the fast analysis of glycosylation site occu-
pancy of recombinant glycoproteins by ESI-MS.36 Full mass spectrometric analysis
of peptides with complex glycans has been performed on small glycopeptides from
a nonspecific enzymatic digestion (e.g., pronase). Using HILIC, these glycopeptides
were separated according to the size of the glycan chain.99,100 Information about
the glycans was obtained from MS/MS data, and the identification of the peptide
backbone and the glycan attachment site were made based on MS3 spectra.100 This
method was used to identify two glycans with different structures that were attached
alternatively to one glycosylation site of Dolichos biflorus lectin.100
As an alternative to the HILIC resins and columns described above, glycopep-
tides can be enriched based on hydrophilic interaction with carbohydrate gel matri-
ces such as cellulose or sepharose.101,102 This method has been used to analyze the
N-glycosylation sites of apolipoprotein B-100 showing that 17 out of the 19 potential
glycosylation sites were occupied.102 It has also been used for the differential analysis
of site-specific glycans from cellular fibronectin.102
22.2.2.2 Lectin Affinity Chromatography

Similar to glycoprotein enrichment, glycopeptides can be purified using lectin affin-
ity chromatography. This can be done either using a broad specificity lectin such
as the widely used concanavalin A resin to capture N-glycosylated peptides,58,103,104
or using sequential or multi-lectin columns to capture all glycopeptides including
ones with O-linked glycans,61,105 or using a specific lectin to capture a particular
group of glycopeptides.106,107 Lectin affinity chromatography is often used in the
“glyco-catch” method where the lectin column is used in an initial enrichment of
the glycoprotein and then the same column is used again for the enrichment of the
glycopeptides after glycoprotein digestion.61 This method has been used to identify
soluble glycopeptides from a C. elegans lysate.108
22.2.2.3 Hydrazine Chemistry

Both glycoproteins and glycopeptides can be trapped using hydrazine chemistry. The
vicinal hydroxyl groups of carbohydrate moiety of N-linked glycopeptides are oxidized to
aldehydes with periodate first. These aldehyde groups are then covalently bound to hydra-
zine resin (Figure 22.3B).109 After the glycoprotein is captured using hydrazine resin, it
is then digested in situ using trypsin to remove non-glycosylated peptides. The peptide
portion of N-linked glycopeptides is then released by PNGase F treatment, leaving the
oligosaccharides bound to the resin. A scheme for the modification of glycopeptides by
hydrazine chemistry is shown in Figure 22.3B. One of the main advantages of this method
is its ability to immobilize glycoproteins, digest in situ, and then remove all unwanted
peptides before elution of the glycopeptides, thus abolishing the need for both glycopro-
tein and glycopeptide enrichment. Sun et al. have used hydrazine resin to capture the gly-
copeptides from a tryptic digest of the microsomal fraction of an ovarian cancer cell line.
The approach yielded a 19- to 45-fold enrichment of glycopeptides in the sample.85 This
method has also been used for the identification of glycoproteins in the plasma membrane
from HeLa cell lysates,110 and for the analysis of glycoproteins in human serum109 and
in plasma.111 In these cases, the glycoprotein was immobilized on the hydrazine matrix
before digestion and then the glycopeptides were released by PNGase F treatment.

22.2.2.4 Ion-Pairing Normal-Phase Chromatography

Although standard normal phase chromatography is not as widely used as HILIC
chromatography due to reproducibility problems, the separation of glycosylated and
non-glycosylated peptides by this technique can be improved by the addition of inor-
ganic monovalent ions such as NaCl, LiCl, or NaOH.112 This ion-pairing interaction
increases the hydrophobicity difference between peptides with glycans and the ones
without glycans, and consequently results in a more efficient separation than with a
normal-phase column used under standard conditions. The improvement has been
exemplified by the analysis of glycopeptides from breast cancer cells.112
22.2.2.5 Others
In addition to the methods discussed above, glycopeptides may also be enriched
from complex peptide solutions by techniques such as size-exclusion chromatog-
raphy (SEC), graphitized carbon and CE. The logic behind SEC is based on the
fact that glycopeptides tend to have a higher mass than non-glycosylated peptides,113
whereas the use of graphitized carbon is mainly because of the high hydrophobicity
of the materials for retaining small glycopeptides.114
CZE is a useful technique for the enrichment of glycopeptides because it can be
highly selective. By carefully choosing the conditions for electrophoretic and elec-
trokinetic countermigration, this technique can be used to enrich small and highly
hydrophilic glycopeptides, which otherwise are difficult to analyze using a standard
reversed-phase HPLC approach. These glycopeptides normally elute near the void
volume in HPLC separation.73 Bindila et al. used CZE in the analysis of glycopep-
tides and amino acids from human urine of a patient with N-acetylhexosaminiclase
deficiency.115 CZE showed good separation efficiency and resolution according to the
degree of sialylation and the type of amino acids allowing structural identification of
single components of the complex sample.115
Most methods described above do not alter the glycan structure. However,
highly specific enrichment of glycopeptides can be achieved using chemical
modifications.116 For example, glycopeptides with glycans containing cis-diol can
be isolated using phenylboranic acid by the method discussed for glycoprotein
enrichment in Section 22.2.1.3. The scheme shown in Figure 22.3A illustrates
this approach.63,117 This method is particularly useful because it binds both N- and
O-glycopeptides.
O-GlcNAc-containing peptides can be analyzed using modification of the gly-
can with galactosyltransferase, which introduces a ketone-labeled galactose to the
peptide. This ketone group is then biotinylated using a Schiff-base reaction (Figure
22.3C); thus glycopeptides can be purified using streptavidin affinity chromatogra-
phy.118 This method has been used to identify glycoproteins from the mammalian
brain119 and HeLa cell lysates.120
Sialylated glycopeptides can be enriched by selective oxidation of the terminal
sialic acid residue with periodate followed by ligation of the newly formed aldehyde
group with a polymer substrate such as aminooxy-functionalized polyacrylamide
(Figure 22.3D). After washing to remove non-sialylated peptides, the sialylated gly-
copeptides are eluted with trifluoroacetic acid, which cleaves the bond between the

sialic acid and adjacent galactose residue.121 The authors exemplified the method by
enriching glycopeptides from human α-fetoprotein, bovine pancreas fibrinogen and
human erythropoietin.121
22.2.3 Tandem Methods

For many glycoproteomic applications, the protein samples are rather complex.35,86
Hence more than one enrichment method is needed in practice. Generally, the tandem
enrichment method consists of initial glycoprotein selection followed by tryptic diges-
tion and subsequent glycopeptide enrichment, although in some cases more than two
methods are used in series. For example, Kaji et al. used an initial lectin capture of
glycoproteins followed by tryptic digestion and purification of the glycopeptides by both
lectin affinity chromatography and HILIC in their analysis of C. elegans glycoproteins.95
Two techniques that are widely used in the most common tandem methods are lec-
tin affinity chromatography (for glycoprotein capture) coupled to an additional step of
enrichment of the glycopeptides either using lectin or HILIC. If the same lectin affin-
ity column is used for both glycoprotein and glycopeptide enrichment, the technique
is known as the glyco-catch method, which is described briefly in Section 22.2.2.2.61
The protocol used by Kaji et al. combined lectin affinity chromatography and
HILIC for the identification of glycoproteins and their glycosylation sites from
mouse liver.94 The purity of the glycopeptides was increased to 70%–100% after the
HILIC step. Totally, around 2500 N-glycosylation sites in 1200 proteins were identi-
fied in the work.94 The analysis was finished within a week, highlighting the high-
throughput nature of the technique. A similar tandem strategy was also described by
Bunkenborg et al. in which two lectin steps of enrichment for both glycoproteins and
glycopeptides were employed for glycoproteomic analysis of human body fluids.122
The method was exemplified using the model proteins α-1-acid glycoprotein, fetuin,
ovalbumin, and ribonuclease B.122
Recently, Calvano et al. assessed these tandem techniques for the characteriza-
tion of serum glycoproteins and found 86 N-glycosylation sites in 45 proteins by the
glyco-catch method and 81 N-glycosylation sites in 44 proteins by lectin affinity
followed by HILIC.80 The methods were shown to be complementary as different
glycopeptides were isolated using the differing tandem approaches. An overlap of 63
N-glycosylation sites from 38 proteins were demonstrated in the report.80
In addition to those using lectin as the initial step, any of the methods described
here can be used in series to yield a tandem method for glyco-enrichment. For exam-
ple, more than one type of CE can be used in tandem.123 It should be noted, however,
that compatibility between the various techniques and the mass spectrometer needs
to be taken into consideration.
22.3 Examples of Glycoprotein

Characterization Using HILIC
The examples in this chapter all use human IgE-FcεRIα as the test glycoprotein.
IgE-FcεRIα is the α-chain ectodomain of the high-affinity immunoglobulin-E
(IgE) receptor FcεRI, which is found on the surface of effector cells within the

immune system. IgE-FcεRIα is associated with allergic response, anaphylaxis and

antiparasitic immunity.124,125 Two glycosylation variants have been used: the wild
type, denoted wt-IgE-FcεRIα, which contains six N-glycosylation sites (N21, N42,
N50, N74, N135, and N166) and one O-glycosylation site (T142), and a triple mutant
referred to as mut3-IgE-FcεRIα in this chapter. Based on the work of Garman et al.,126
the triple mutant lacks glycosylation sites at residues N74, N135, and T142, all of
which are variably occupied (see Section 22.3.3).
The glycans attached to IgE-FcεRIα were simplified either by expressing the
protein in human embryonic kidney (HEK) 293T cells in the presence of kifunen-
sine, which inhibits N-glycosylation processing leaving sugars in the form Man5–9
GlcNAc2, or by producing the protein in ricin-resistant HEK 293S cells, which have
no GnTI activity giving predominantly Man5GlcNAc2 glycans.127,128
22.3.1 Glycoprotein Enrichment

In the proof-of-principle example described below, ZIC-HILIC was used to
enrich the recombinant glycoprotein mut3 -IgE-FcεRIα from a mixture containing
three other non-glycosylated proteins. The purified protein has about 1â•›mg/mL
concentration with about 25% of the protein being glycoprotein (Figure 22.5,
lane 1). After the glycoproteins in the sample were bound to the ZIC-HILIC
resin and the resin washed; the eluted glycoprotein
was estimated to have approximately 70% purity 1 2
kDa
(Figure 22.5, lane 2). The method is of particular
use in recombinant protein production when protein 84
yield is low; therefore, enrichment of glycoproteins 66
is needed prior to analysis by SDS-PAGE or MS. It
is to be noted, however, that most media used for
45
mammalian cell culturing contains other glycosyl-
36
ated proteins, especially from components of fetal
calf serum. These proteins are also enriched using 29
this method along with the glycoprotein of interest.
24
The method can be extended to enrich glycoproteins
present in cellular extracts or fluids. 80,81 The prod- 20
uct after glycoprotein enrichment can then be used
for further analysis, for example, determination of
glycosylation site occupancy (Section 22.3.3) and Figure 22.5 SDS-PAGE
characterization of glycan structure (Section 22.3.4). gel showing enrichment of
mut3-IgE-FcεRIα (lane 2)
from a mixture also contain-
22.3.2 Characterization of Glycoforms ing three non-glycosylated
Although HILIC is not commonly used for the anal- proteins (lane 1). Enrichment
is from ∼25% to 70% mea-
ysis of glycoforms, it can be useful when studying
sured using densitrometry
proteins with low degree of glycan heterogeneity. In of the bands on the gel. The
the example described here, mut3 -IgE-FcεRIα was position of mut3-IgE-FcεRIα
grown in HEK 293T cells in the presence of kifu- is shown by the arrow on the
nensine128 producing only Man5–9GlcNAc2 sugars.127 right-hand side.

Table 22.2
Glycoform Separation of the Tryptic Peptide Containing
N166 from mut3-IgE-FcεRI𝛂 Grown in HEK 293T Cells
in the Presence of Kifunensine
Retention Time (Minutes) Glycoform Expected m/z Observed m/z
11.59 Man9GlcNAc2 1956.35 1957.31
11.48 Man8GlcNAc2 1875.33 1873.18
11.41 Man7GlcNAc2 1794.30 1793.60
Oligosaccharide structures are given along with the expected and observed m/z
values for the [M + 2H]2+ ions observed.
After tryptic digest of the protein, the resulting glycopeptides were separated
using a ZIC-HILIC column with isocratic elution of 55% acetonitrile, 45% water,
10â•›m M ammonium acetate buffer pH 8.0, and analyzed by ESI-MS. [M + 2H]2+
ions detected for the tryptic glycopeptide containing N166 are given in Table 22.2.
The difference in retention time between various glycoforms was not as great as
the results for sialylated glycopeptides by Zhang and Wang91 and by Takegawa
et al.129 This was probably caused by the greater hydrophobicity introduced by
the large glycan-containing peptide (17 amino acids, average hydrophilicity130 is
equal to −0.2 with values below 0 indicating a hydrophobic peptide). Glycoform
analysis by this method is particularly useful in tracking batch-to-batch glycoform
variation in the production of a recombinant glycoprotein. In the current example,
glycoform characterization was used to analyze the glycan heterogeneity prior to
crystallization of the protein.126
22.3.3 Characterization of Glycosylation Site Occupancy

Several methods use HILIC for N-glycosylation site occupancy analysis. These
methods differ in the labeling of the glycosylation site prior to peptide analysis. If
the glycoprotein is purified and the sequence is known, the glycosylation site can be
readily determined by simply identifying the peptides whose molecular weight sees
a 1â•›Da increase when compared to the theoretical calculation. The mass change in
peptide molecular weight is brought about by PNGase F treatment, which converts
the asparagine residue to an aspartic acid.36 HILIC glycopeptide enrichment can also
be undertaken offline in a batch mode to enrich all glycopeptides prior to PNGase F
treatment. Table 22.3 shows the results from glycosylation site occupancy analysis
for recombinant wt-IgE-FcεRIα produced in mammalian cells. The data indicates
that four of the N-glycosylation sites are always occupied, with two being variably
occupied and no sites being unoccupied (Figure 22.6). The results verify mutation
studies.126,131 The O-glycosylation site occupancy cannot be investigated by this
method. However, the identification of relevant non-glycosylated peptide implies that
the O-glycosylation site could be either not occupied at all or only variably occupied

Table 22.3
Results from Glycosylation Site Analysis
of wt-IgE-FcεRIα
Glycosylation Site Occupancy
N21 Fully occupied
N42 Fully occupied
N50 Fully occupied
N74 Variably occupied
N135 Variably occupied
N166 Fully occupied
T142 Either variably or not occupied
1025.120
1024.622
1025.618
1026.115
1026.631
1020 1022 1024 1026 1028 1030

(A) m/z
1330.676
1329.698
1331.634
1330.695
1331.634 1332.650
1332.688 1333.669
1326 1328 1330 1332 1334 1336 1326 1328 1330 1332 1334 1336
(B) m/z m/z
Figure 22.6 MS data for (A) the [M + 2H]2+ ion of the tryptic peptide containing N166,
which is fully occupied, and (B) the [M + H]+ ions for the peptide containing N74, which is
variably occupied (+0â•›Da and +1â•›Da) generated by digestion with chymotrypsin.
(Table 22.3). Mutation studies have shown the O-linked glycan was not essential for
cell growth.126,131
If the glycoprotein is in a complex mixture and the sequence is unknown, unam-
biguous assignment of the glycosylation sites is performed by either stable isotope
labeling with 18O,95 or by leaving one GlcNAc residue attached to the site using a
combination of endo-β-N-acetylglucosaminidases instead of PNGase F.37

22.3.4 Full Characterization of Glycoproteins

For complete characterization of a glycoprotein, the protein backbone, the occu-
pancy of the glycosylation sites and the glycan structures need to be thoroughly
analyzed. In the case of a purified recombinant glycoprotein containing only
N-linked glycans, the protein backbone can be analyzed simply by the treat-
ment of the sample with PNGase F followed by MS analysis of the intact protein
(Figure 22.7A). It is to be noted that the expected mass should be 1â•›Da less for
21571
1000 1200 1400 1600 1800 2000 2200

m/z
20,500 21,000 21,500 22,000 22,500

(A) Mass
[N]-GlcNAc2Man41+
[TN]-GlcNAc2Man21+
[NS]-GlcNAc2Man31+
[TNS]-GlcNAc2Man51+
GlcNAc2Man31+
[TN]-GlcNAc2Man31+
GlcNAcMan51+
800 900 1,000 1,100 1,200 1,300 1,400 1,500 1,600

(B) m/z
Figure 22.7 Characterization of mut3-IgE-FcεRIα. (A) Deconvoluted intact protein mass

spectrum with combined raw data shown in the inset. (B) Fragment ion spectrum (MS/MS) of the
pronase-generated glycopeptide TNS of mut3-IgE-FcεRIα with a Man5GlcNAc2 glycan moiety
attached. MS/MS data obtained from the doubly protonated precursor ion at m/z = 769.3.

each fully occupied N-glycosylation site because of the conversion of asparagine

to aspartic acid upon PNGase F treatment. If identification of the glycoprotein
is needed, for example, the protein has been enriched from a complex sample,
the protein is digested with trypsin and subjected to MS/MS analysis followed
by the identification of the peptides using a database searching software such
as Mascot.132 Because the removal of O-linked sugars presents some difficulty
(Section 22.1.3), tryptic digestion is recommended for proteins containing these
glycans. Digestion is followed by analysis of the non-glycosylated peptides in
order to identify the protein.
In general, glycosylation site occupancy and analysis of glycan structure are
performed separately. However, both analyses can be efficiently combined using
the method of Wuhrer et al.100 This combined method analyzes glycopeptides
directly to obtain information on both glycan structure and glycosylation site.
This method can be used for both N- and O-linked glycans. Such information
is lost when the glycans are removed before analysis. After the identification
of the protein backbone (if needed) using either intact protein MS or analy-
sis of tryptic peptides, the glycoprotein is subjected to digestion with pronase.
Nonspecific proteolysis with pronase results in small peptides of two to eight
amino acids, which are amenable to detection by MS.133 The glycopeptides
are separated using normal phase chromatography or HILIC before analysis
by multistage MS fragmentations. Dodds et al. have improved the method by
immobilization of pronase to eliminate auto-proteolytic fragments from the
final peptide sample.134
An example is shown here where mut3-IgE-FcεRIα with Man5GlcNAc2 gly-
cans was digested with pronase followed by HILIC and mass spectrometric analy-
sis (Figure 22.7B). The HILIC column effectively separated the pronase-generated
peptides using isocratic elution in 55% acetonitrile, 45% water, 10â•›m M ammonium
acetate buffer pH 8.0. The glycopeptides elute in the retention time window from
13.7 to 18.0â•›min. Glycopeptides were readily identified by screening for neutral loss
of terminal monosaccharide units—a neutral loss of 81 is indicative of a doubly
charged precursor ion containing a terminal hexose. Figure 22.7B shows the MS/MS
spectrum for the doubly charged ion at m/z = 769.3, which corresponds to the tripep-
tide TNS containing the glycosylation site N50. The identification of the peaks in the
spectrum in Figure 22.7B was facilitated by prior knowledge of the glycan pattern,127
which was of particular use as the MS/MS data contained ions from fragmenta-
tion of both glycosidic linkages and peptide bonds. The data obtained, however,
did confirm that only Man5GlcNAc2 glycan was attached to mut3-IgE-FcεRIα as
expected due to the protein being produced from HEK 293S cells.127,128 As indicated
by Wuhrer et al., this technique gives most information when used in combination
with other glycan analysis methods including conventional glycan release followed
by MS of the oligosaccharides.100
Although full characterization of glycoproteins by this method is complex, it has
been successfully applied to model glycoproteins such as ribonuclease B,100,134 horse
radish peroxidise,99,100 κ-casein,134 and chicken egg albumin.134 In addition, it is also
used to characterize a previously uncharacterized protein, D. biflorus lectin.100

22.4 Summary
In this chapter, the challenges associated with glycoprotein characterization are
shown to be associated with the diversity and heterogeneity of the glycan moieties
attached to a protein. Various general methods for the isolation and separation of
glycoproteins and glycopeptides are presented along with their benefits and limita-
tions. In many cases, glycoprotein characterization is seen to require more than one
step of enrichment in order to gain the maximum amount of information from each
sample. These tandem methods usually consist of a step of isolation of the glycopro-
tein from solution followed by digestion and enrichment or separation of the result-
ing glycopeptides.
The use of HILIC in glycoprotein characterization was highlighted by four worked
examples using the human-secreted protein IgE-FcεRIα. In the first experiment, the
intact glycoprotein was enriched from a mixture also containing non-glycosylated
proteins using a batch method. Several glycoforms attached to an N-glycosylation
site were separated using column chromatography where the elution volume was
related to the size of the glycan. The N-glycosylation sites were characterized for
occupancy using a simple batch method for the enrichment of glycopeptides before
PNGase F treatment and MS analysis. In the final example, both the occupancy of
a glycosylation site and the composition of the attached glycan were characterized
using digestion with pronase followed by HILIC and MS/MS analysis.
Acknowledgments
Within the Oxford Protein Production Facility, I wish to thank Nahid Rahman for
production and purification of mut3-IgE-FcεRIα and Ray Owens for critical reading
of this manuscript. I would also like to thank Jens Loebermann and Rebecca Beavil
at King’s College, London, for the gift of wt-IgE-FcεRIα.
References
1. Apweiler, R., Hermjakob, H., and Sharon, N., On the frequency of protein glycosyl-
ation, as deduced from analysis of the SWISS-PROT database, Biochim. Biophys. Acta
1473 (1), 4–8, 1999.
2. Mitra, N., Sinha, S., Ramya, T. N., and Surolia, A., N-linked oligosaccharides as
outfitters for glycoprotein folding, form and function, Trends Biochem. Sci. 31 (3),
156–63, 2006.
3. Helenius, A. and Aebi, M., Intracellular functions of N-linked glycans, Science 291
(5512), 2364–2369, 2001.
4. Huang, Y., Huang, J. H., Xie, Q. J., and Yao, S. Z., Carbohydrate-protein interactions,
Prog. Chem. 20 (6), 942–950, 2008.
5. Rudd, P. M., Elliott, T., Cresswell, P., Wilson, I. A., and Dwek, R. A., Glycosylation and
the immune system, Science 291 (5512), 2370–2376, 2001.
6. Rabinovich, G. A. and Toscano, M. A., Turning ‘sweet’ on immunity: Galectin-glycan
interactions in immune tolerance and inflammation, Nat. Rev. Immunol. 9 (5), 338–352,
2009.
7. Hann, S. R., Role of post-translational modifications in regulating c-Myc proteolysis, tran-
scriptional activity and biological function, Semin. Cancer Biol. 16 (4), 288–302, 2006.

8. Limjindaporn, T., Wongwiwat, W., Noisakran, S., Srisawat, C., Netsawang, J., Puttikhunt, C.,
Kasinrerk, W., Avirutnan, P., Thiemmeca, S., Sriburi, R., Sittisombut, N., Malasit, P., and
Yenchitsomanus, P. T., Interaction of dengue virus envelope protein with endoplasmic
reticulum-resident chaperones facilitates dengue virus production, Biochem. Biophys. Res.
Commun. 379 (2), 196–200, 2009.
9. Abu-Qarn, M., Eichler, J., and Sharon, N., Not just for Eukarya anymore: Protein glyco-
sylation in Bacteria and Archaea, Curr. Opin. Struct. Biol. 18 (5), 544–550, 2008.
10. Upreti, R. K., Kumar, M., and Shankar, V., Bacterial glycoproteins: Functions, biosyn-
thesis and applications, Proteomics 3 (4), 363–379, 2003.
11. Fletcher, C. M., Coyne, M. J., Villa, O. F., Chatzidaki-Livanis, M., and Comstock, L. E.,
A general O-glycosylation system important to the physiology of a major human intestinal
symbiont, Cell 137 (2), 321–331, 2009.
12. Szymanski, C. M. and Wren, B. W., Protein glycosylation in bacterial mucosal patho-
gens, Nat. Rev. Microbiol. 3 (3), 225–237, 2005.
13. Aas, F. E., Vik, A., Vedde, J., Koomey, M., and Egge-Jacobsen, W., Neisseria gonor-
rhoeae O-linked pilin glycosylation: Functional analyses define both the biosynthetic
pathway and glycan structure, Mol. Microbiol. 65 (3), 607–624, 2007.
14. Daniels, M. A., Hogquist, K. A., and Jameson, S. C., Sweet ‘n’ sour: The impact of dif-
ferential glycosylation on T cell responses, Nat. Immunol. 3 (10), 903–910, 2002.
15. Dwek, M. V., Ross, H. A., and Leathem, A. J., Proteome and glycosylation mapping
identifies post-translational modifications associated with aggressive breast cancer,
Proteomics 1 (6), 756–762, 2001.
16. Freeze, H. H., Update and perspectives on congenital disorders of glycosylation,
Glycobiology 11 (12), 129R-143R, 2001.
17. Marquardt, T. and Denecke, J., Congenital disorders of glycosylation: Review of their
molecular bases, clinical presentations and specific therapies, Eur. J. Pediatr. 162 (6),
359–379, 2003.
18. Hwang, H., Zhang, J., Chung, K. A., Leverenz, J. B., Zabetian, C. P., Peskind, E. R.,
Jankovic, J., Su, Z., Hancock, A. M., Pan, C., Montine, T. J., Pan, S., Nutt, J., Albin, R.,
Gearing, M., Beyer, R. P., Shi, M., and Zhang, J., Glycoproteomics in neurodegenerative
diseases, Mass Spectrom. Rev. 8, 8, 2009.
19. Saldova, R., Royle, L., Radcliffe, C. M., Abd Hamid, U. M., Evans, R., Arnold, J. N.,
Banks, R. E., Hutson, R., Harvey, D. J., Antrobus, R., Petrescu, S. M., Dwek, R. A., and
Rudd, P. M., Ovarian cancer is associated with changes in glycosylation in both acute-
phase proteins and IgG, Glycobiology 17 (12), 1344–1356, 2007.
20. Saez-Valero, J. and Small, D. H., Acetylcholinesterase and butyrylcholinesterase glyco-
forms are biomarkers of Alzheimer’s disease, J. Alzheimers Dis. 3 (3), 323–328, 2001.
21. Saez-Valero, J., Fodero, L. R., Sjogren, M., Andreasen, N., Amici, S., Gallai, V.,
Vanderstichele, H., Vanmechelen, E., Parnetti, L., Blennow, K., and Small, D. H.,
Glycosylation of acetylcholinesterase and butyrylcholinesterase changes as a function
of the duration of Alzheimer’s disease, J. Neurosci. Res. 72 (4), 520–526, 2003.
22. Srinivasan, A., Viswanathan, K., Raman, R., Chandrasekaran, A., Raguram, S., Tumpey,
T. M., Sasisekharan, V., and Sasisekharan, R., Quantitative biochemical rationale for
differences in transmissibility of 1918 pandemic influenza A viruses, Proc. Natl. Acad.
Sci. U.S.A. 105 (8), 2800–2805, 2008.
23. Wang, L. X., Toward oligosaccharide- and glycopeptide-based HIV vaccines, Curr.
Opin. Drug Discov. Dev. 9 (2), 194–206, 2006.
24. Ni, J., Song, H., Wang, Y., Stamatos, N. M., and Wang, L. X., Toward a carbohydrate-
based HIV-1 vaccine: Synthesis and immunological studies of oligomannose-containing
glycoconjugates, Bioconjug. Chem. 17 (2), 493–500, 2006.
25. Slovin, S. F., Keding, S. J., and Ragupathi, G., Carbohydrate vaccines as immunother-
apy for cancer, Immunol. Cell Biol. 83 (4), 418–428, 2005.

26. Slovin, S. F., Ragupathi, G., Musselli, C., Fernandez, C., Diani, M., Verbel, D.,
Danishefsky, S., Livingston, P., and Scher, H. I., Thomsen-Friedenreich (TF) antigen as
a target for prostate cancer vaccine: Clinical trial results with TF cluster (c)-KLH plus
QS21 conjugate vaccine in patients with biochemically relapsed prostate cancer, Cancer
Immunol. Immunother. 54 (7), 694–702, 2005.
27. Hart, G. W., Glycosylation, Curr. Opin. Cell. Biol. 4 (6), 1017–1023, 1992.
28. Rajeswari, P., Natarajan, R., Nadler, J. L., Kumar, D., and Kalra, V. K., Glucose
induces lipid peroxidation and inactivation of membrane-associated ion-transport
enzymes in human erythrocytes in vivo and in vitro, J. Cell. Physiol. 149 (1),
100–109, 1991.
29. Wells, L., Vosseller, K., and Hart, G. W., Glycosylation of nucleocytoplasmic proteins:
Signal transduction and O-GlcNAc, Science 291 (5512), 2376–2378, 2001.
30. Harris, R. J. and Spellman, M. W., O-linked fucose and other post-translational modifi-
cations unique to EGF modules, Glycobiology 3 (3), 219–224, 1993.
31. Mormeneo, S., Zueco, J., Iranzo, M., and Sentandreu, R., O-linked mannose composi-
tion of secreted invertase of Saccharomyces cerevisiae, FEMS Microbiol. Lett. 48 (3),
271–274, 1989.
32. Duman, J. G., Miele, R. G., Liang, H., Grella, D. K., Sim, K. L., Castellino, F. J., and
Bretthauer, R. K., O-Mannosylation of Pichia pastoris cellular and recombinant pro-
teins, Biotechnol. Appl. Biochem. 28 (Pt 1), 39–45, 1998.
33. Kogelberg, H., Chai, W., Feizi, T., and Lawson, A. M., NMR studies of mannitol-
terminating oligosaccharides derived by reductive alkaline hydrolysis from brain
glycoproteins, Carbohydr. Res. 331 (4), 393–401, 2001.
34. Hounsell, E. F., Davies, M. J., and Renouf, D. V., O-linked protein glycosylation struc-
ture and function, Glycoconj. J. 13 (1), 19–26, 1996.
35. Geyer, H. and Geyer, R., Strategies for analysis of glycoprotein glycosylation, Biochim.
Biophys. Acta 1764 (12), 1853–1869, 2006.
36. Nettleship, J. E., Aplin, R., Aricescu, A. R., Evans, E. J., Davis, S. J., Crispin, M., and
Owens, R. J., Analysis of variable N-glycosylation site occupancy in glycoproteins by
liquid chromatography electrospray ionization mass spectrometry, Anal. Biochem. 361 (1),
149–151, 2007.
37. Hagglund, P., Bunkenborg, J., Elortza, F., Jensen, O. N., and Roepstorff, P., A new strat-
egy for identification of N-glycosylated proteins and unambiguous assignment of their
glycosylation sites using HILIC enrichment and partial deglycosylation, J. Proteome
Res. 3 (3), 556–566, 2004.
38. Greis, K. D., Hayes, B. K., Comer, F. I., Kirk, M., Barnes, S., Lowary, T. L., and Hart,
G. W., Selective detection and site-analysis of O-GlcNAc-modified glycopeptides by
beta-elimination and tandem electrospray mass spectrometry, Anal. Biochem. 234 (1),
38–49, 1996.
39. Hanisch, F. G., Jovanovic, M., and Peter-Katalinic, J., Glycoprotein identification and
localization of O-glycosylation sites by mass spectrometric analysis of deglycosylated/
alkylaminylated peptide fragments, Anal. Biochem. 290 (1), 47–59, 2001.
40. Morelle, W. and Michalski, J. C., The mass spectrometric analysis of glycoproteins and
their glycan structures, Curr. Anal. Chem. 1 (1), 29–57, 2005.
41. Annesley, T. M., Ion suppression in mass spectrometry, Clin. Chem. 49 (7), 1041–1044,
2003.
42. Dalpathado, D. S. and Desaire, H., Glycopeptide analysis by mass spectrometry, Analyst
133 (6), 731–738, 2008.
43. Sei, K., Nakano, M., Kinoshita, M., Masuko, T., and Kakehi, K., Collection of alpha1-
acid glycoprotein molecular species by capillary electrophoresis and the analysis of
their molecular masses and carbohydrate chains. Basic studies on the analysis of glyco-
protein glycoforms, J. Chromatogr. A 958 (1–2), 273–281, 2002.

44. Gimenez, E., Benavente, F., Barbosa, J., and Sanz-Nebot, V., Towards a reliable molec-
ular mass determination of intact glycoproteins by matrix-assisted laser desorption/
ionization time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 21 (16),
2555–2563, 2007.
45. Zaia, J., Boynton, R., Heinegard, D., and Barry, F., Posttranslational modifications
to human bone sialoprotein determined by mass spectrometry, Biochemistry 40 (43),
12983–12991, 2001.
46. Ongay, S., Puerta, A., Diez-Masa, J. C., Bergquist, J., and de Frutos, M., CIEF and
MALDI-TOF-MS methods for analyzing forms of the glycoprotein VEGF 165,
Electrophoresis 30 (7), 1198–1205, 2009.
47. Beavis, R. C. and Chait, B. T., Cinnamic acid derivatives as matrices for ultraviolet
laser desorption mass spectrometry of proteins, Rapid Commun. Mass Spectrom. 3 (12),
432–435, 1989.
48. Hui, J. P., White, T. C., and Thibault, P., Identification of glycan structure and glycosyl-
ation sites in cellobiohydrolase II and endoglucanases I and II from Trichoderma reesei,
Glycobiology 12 (12), 837–849, 2002.
49. Gong, B., Cukan, M., Fisher, R., Li, H., Stadheim, T. A., and Gerngross, T.,
Characterization of N-linked glycosylation on recombinant glycoproteins produced
in Pichia pastoris using ESI-MS and MALDI-TOF, Methods Mol. Biol. 534,
213–223, 2009.
50. Karas, M., Bahr, U., and Dulcks, T., Nano-electrospray ionization mass spectrometry:
Addressing analytical problems beyond routine, Fresenius J. Anal. Chem. 366 (6–7),
669–676, 2000.
51. Wang, L., Li, F., Sun, W., Wu, S., Wang, X., Zhang, L., Zheng, D., Wang, J., and Gao, Y.,
Concanavalin A-captured glycoproteins in healthy human urine, Mol. Cell Proteomics
5 (3), 560–562, 2006.
52. White, K. Y., Rodemich, L., Nyalwidhe, J. O., Comunale, M. A., Clements, M. A.,
Lance, R. S., Schellhammer, P. F., Mehta, A. S., Semmes, O. J., and Drake, R. R.,
Glycomic characterization of prostate-specific antigen and prostatic acid phospha-
tase in prostate cancer and benign disease seminal plasma fluids, J. Proteome Res.
7, 7, 2009.
53. Bonner, P. L. R., Protein Purification, Taylor & Francis, New York, 2007.
54. Donnelly, E. H. and Goldstein, I. J., Glutaraldehyde-insolubilized concanavalin A:
An adsorbent for the specific isolation of polysaccharides and glycoproteins, Biochem.
J. 118 (4), 679–680, 1970.
55. Qiu, R. and Regnier, F. E., Use of multidimensional lectin affinity chromatography in
differential glycoproteomics, Anal. Chem. 77 (9), 2802–2809, 2005.
56. Monzo, A., Bonn, G. K., and Guttman, A., Boronic acid-lectin affinity chromatogra-
phy. 1. Simultaneous glycoprotein binding with selective or combined elution, Anal.
Bioanal. Chem. 389 (7–8), 2097–2102, 2007.
57. Mechref, Y. and Novotny, M. V., Structural investigations of glycoconjugates at high
sensitivity, Chem. Rev. 102 (2), 321–369, 2002.
58. Bunkenborg, J., Pilch, B. J., Podtelejnikov, A. V., and Wisniewski, J. R., Screening for
N-glycosylated proteins by liquid chromatography mass spectrometry, Proteomics 4 (2),
454–465, 2004.
59. Chen, H., Xu, X., Lin, H. H., Chen, S. H., Forsman, A., Aasa-Chapman, M., and Jones,
I. M., Mapping the immune response to the outer domain of a human immunodeficiency
virus-1 clade C gp120, J. Gen. Virol. 89 (Pt 10), 2597–2604, 2008.
60. Hirabayashi, J., Concept, strategy and realization of lectin-based glycan profiling,
J. Biochem. 144 (2), 139–147, 2008.
61. Hirabayashi, J. and Kasai, K., Separation technologies for glycomics, J. Chromatogr. B
Anal. Technol. Biomed. Life Sci. 771 (1–2), 67–87, 2002.

62. Wiener, M. C. and van Hoek, A. N., A lectin screening method for membrane glycopro-
teins: Application to the human CHIP28 water channel (AQP-1), Anal. Biochem. 241 (2),
267–268, 1996.
63. Sparbier, K., Koch, S., Kessler, I., Wenzel, T., and Kostrzewa, M., Selective isolation of
glycoproteins and glycopeptides for MALDI-TOF MS detection supported by magnetic
particles, J. Biomol. Tech. 16 (4), 407–413, 2005.
64. Wang, Y., Wu, S. L., and Hancock, W. S., Approaches to the study of N-linked glycopro-
teins in human plasma using lectin affinity chromatography and nano-HPLC coupled to
electrospray linear ion trap—Fourier transform mass spectrometry, Glycobiology 16 (6),
514–523, 2006.
65. Yang, Z. and Hancock, W. S., Approach to the comprehensive analysis of glycoproteins
isolated from human serum using a multi-lectin affinity column, J. Chromatogr. A 1053
(1–2), 79–88, 2004.
66. Mega, T., Oku, H., and Hase, S., Characterization of carbohydrate-binding specificity of
concanavalin A by competitive binding of pyridylamino sugar chains, J. Biochem. 111 (3),
396–400, 1992.
67. Kamoda, S. and Kakehi, K., Evaluation of glycosylation for quality assurance of anti-
body pharmaceuticals by capillary electrophoresis, Electrophoresis 29 (17), 3595–3604,
2008.
68. Balaguer, E. and Neususs, C., Glycoprotein characterization combining intact protein
and glycan analysis by capillary electrophoresis-electrospray ionization-mass spectrom-
etry, Anal. Chem. 78 (15), 5384–5393, 2006.
69. Dolnik, V. and Liu, S., Applications of capillary electrophoresis on microchip, J. Sep.
Sci. 28 (15), 1994–2009, 2005.
70. Ramdani, B., Nuyens, V., Codden, T., Perpete, G., Colicis, J., Lenaerts, A., Henry,
J. P., and Legros, F. J., Analyte comigrating with trisialotransferrin during capillary
zone electrophoresis of sera from patients with cancer, Clin. Chem. 49 (11), 1854–1864,
2003.
71. Mischak, H., Coon, J. J., Novak, J., Weissinger, E. M., Schanstra, J. P., and Dominiczak,
A. F., Capillary electrophoresis-mass spectrometry as a powerful tool in biomarker dis-
covery and clinical diagnosis: An update of recent developments, Mass Spectrom. Rev.
30, 30, 2008.
72. Mechref, Y. and Novotny, M. V., Glycomic analysis by capillary electrophoresis-mass
spectrometry, Mass Spectrom. Rev. 30, 30, 2008.
73. Amon, S., Zamfir, A. D., and Rizzi, A., Glycosylation analysis of glycoproteins and pro-
teoglycans using capillary electrophoresis-mass spectrometry strategies, Electrophoresis
29 (12), 2485–2507, 2008.
74. Sanz-Nebot, V., Balaguer, E., Benavente, F., Neususs, C., and Barbosa, J., Characterization
of transferrin glycoforms in human serum by CE-UV and CE-ESI-MS, Electrophoresis
28 (12), 1949–1957, 2007.
75. Issaq, H. and Veenstra, T., Two-dimensional polyacrylamide gel electrophoresis
(2D-PAGE): Advances and perspectives, Biotechniques 44 (5), 697–698, 700, 2008.
76. Carrette, O., Burkhard, P. R., Sanchez, J. C., and Hochstrasser, D. F., State-of-the-art
two-dimensional gel electrophoresis: A key tool of proteomics research, Nat. Protoc. 1 (2),
812–823, 2006.
77. Loster, K. and Kannicht, C., 2-dimensional electrophoresis: Detection of glycosylation
and influence on spot pattern, Methods Mol. Biol. 446, 199–214, 2008.
78. Miller, I., Crawford, J., and Gianazza, E., Protein stains for proteomic applications:
Which, when, why?, Proteomics 6 (20), 5385–5408, 2006.
79. Wu, J., Lenchik, N. J., Pabst, M. J., Solomon, S. S., Shull, J., and Gerling, I. C.,
Functional characterization of two-dimensional gel-separated proteins using sequential
staining, Electrophoresis 26 (1), 225–237, 2005.

80. Calvano, C. D., Zambonin, C. G., and Jensen, O. N., Assessment of lectin and HILIC
based enrichment protocols for characterization of serum glycoproteins by mass spec-
trometry, J. Proteomics 71 (3), 304–317, 2008.
81. Schneider, U., Protein concentration by hydrophilic interaction chromatography com-
bined with solid phase extraction, Methods Mol. Biol. 424, 63–9, 2008.
82. Clontech, Glycoprotein enrichment and detection, Clontechniques 23 (3), 6–8, 2008.
83. Brena, B. M., Batista-Viera, F., Ryden, L., and Porath, J., Selective adsorption of immuno-
globulins and glucosylated proteins on phenylboronate-agarose, J. Chromatogr. 604 (1),
109–115, 1992.
84. Zanette, D., Soffientini, A., Sottani, C., and Sarubbi, E., Evaluation of phenylboronate
agarose for industrial-scale purification of erythropoietin from mammalian cell cultures,
J. Biotechnol. 101 (3), 275–287, 2003.
85. Sun, B., Ranish, J. A., Utleg, A. G., White, J. T., Yan, X., Lin, B., and Hood, L., Shotgun
glycopeptide capture approach coupled with mass spectrometry for comprehensive gly-
coproteomics, Mol. Cell Proteomics 6 (1), 141–149, 2007.
86. Liu, X., Ma, L., and Li, J. J., Recent developments in the enrichment of glycopeptides
for glycoproteomics, Anal. Lett. 41 (2), 268–277, 2008.
87. Issaq, H. J., Xiao, Z., and Veenstra, T. D., Serum and plasma proteomics, Chem. Rev.
107 (8), 3601–3620, 2007.
88. Alpert, A. J., Hydrophilic-interaction chromatography for the separation of pep-
tides, nucleic acids and other polar compounds, J. Chromatogr. 499, 177–196,
1990.
89. Viklund, C., Sjogren, A., Irgum, K., and Nes, I., Chromatographic interactions between
proteins and sulfoalkylbetaine-based zwitterionic copolymers in fully aqueous low-salt
buffers, Anal. Chem. 73 (3), 444–452, 2001.
90. Wuhrer, M., Catalina, M. I., Deelder, A. M., and Hokke, C. H., Glycoproteomics
based on tandem mass spectrometry of glycopeptides, J. Chromatogr. B Anal. Technol.
Biomed. Life Sci. 849 (1–2), 115–128, 2007.
91. Zhang, J. and Wang, D. I., Quantitative analysis and process monitoring of site-specific
glycosylation microheterogeneity in recombinant human interferon-gamma from Chinese
hamster ovary cell culture by hydrophilic interaction chromatography, J. Chromatogr.
B Biomed. Sci. Appl. 712 (1–2), 73–82, 1998.
92. SeQuant, A Practical Guide to HILIC, Umea, Sweden, 2008.
93. Hagglund, P., Matthiesen, R., Elortza, F., Hojrup, P., Roepstorff, P., Jensen, O. N., and
Bunkenborg, J., An enzymatic deglycosylation scheme enabling identification of core
fucosylated N-glycans and O-glycosylation site mapping of human plasma proteins,
J. Proteome Res. 6 (8), 3021–3031, 2007.
94. Kaji, H., Yamauchi, Y., Takahashi, N., and Isobe, T., Mass spectrometric identification
of N-linked glycopeptides using lectin-mediated affinity capture and glycosylation site-
specific stable isotope tagging, Nat. Protoc. 1 (6), 3019–3027, 2006.
95. Kaji, H., Kamiie, J., Kawakami, H., Kido, K., Yamauchi, Y., Shinkawa, T., Taoka, M.,
Takahashi, N., and Isobe, T., Proteomics reveals N-linked glycoprotein diversity in
Caenorhabditis elegans and suggests an atypical translocation mechanism for integral
membrane proteins, Mol. Cell Proteomics 6 (12), 2100–2109, 2007.
96. Thaysen-Andersen, M., Thogersen, I. B., Nielsen, H. J., Lademann, U., Brunner, N.,
Enghild, J. J., and Hojrup, P., Rapid and individual-specific glycoprofiling of the low
abundance N-glycosylated protein tissue inhibitor of metalloproteinases-1, Mol. Cell
Proteomics 6 (4), 638–647, 2007.
97. Omaetxebarria, M. J., Hagglund, P., Elortza, F., Hooper, N. M., Arizmendi, J. M., and
Jensen, O. N., Isolation and characterization of glycosylphosphatidylinositol-anchored
peptides by hydrophilic interaction chromatography and MALDI tandem mass spec-
trometry, Anal. Chem. 78 (10), 3335–3341, 2006.

98. Boersema, P. J., Mohammed, S., and Heck, A. J., Hydrophilic interaction liquid chroma-
tography (HILIC) in proteomics, Anal. Bioanal. Chem. 391 (1), 151–159, 2008.
99. Yu, Y. Q., Fournier, J., Gilar, M., and Gebler, J. C., Identification of N-linked glycosyl-
ation sites using glycoprotein digestion with pronase prior to MALDI tandem time-of-
flight mass spectrometry, Anal. Chem. 79 (4), 1731–1738, 2007.
100. Wuhrer, M., Koeleman, C. A., Hokke, C. H., and Deelder, A. M., Protein glycosylation
analyzed by normal-phase nano-liquid chromatography—mass spectrometry of glyco-
peptides, Anal. Chem. 77 (3), 886–894, 2005.
101. Wada, Y., Tajiri, M., and Yoshida, S., Hydrophilic affinity isolation and MALDI multiple-
stage tandem mass spectrometry of glycopeptides for glycoproteomics, Anal. Chem.
76 (22), 6560–6565, 2004.
102. Tajiri, M., Yoshida, S., and Wada, Y., Differential analysis of site-specific glycans on
plasma and cellular fibronectins: Application of a hydrophilic affinity method for glyco-
peptide enrichment, Glycobiology 15 (12), 1332–1340, 2005.
103. Demelbauer, U. M., Zehl, M., Plematl, A., Allmaier, G., and Rizzi, A., Determination
of glycopeptide structures by multistage mass spectrometry with low-energy collision-
induced dissociation: Comparison of electrospray ionization quadrupole ion trap and
matrix-assisted laser desorption/ionization quadrupole ion trap reflectron time-of-flight
approaches, Rapid Commun. Mass Spectrom. 18 (14), 1575–1582, 2004.
104. Uematsu, R., Furukawa, J., Nakagawa, H., Shinohara, Y., Deguchi, K., Monde, K., and
Nishimura, S., High throughput quantitative glycomics and glycoform-focused pro-
teomics of murine dermis and epidermis, Mol. Cell Proteomics 4 (12), 1977–1989, 2005.
105. Yang, Z. and Hancock, W. S., Monitoring glycosylation pattern changes of glycopro-
teins using multi-lectin affinity chromatography, J. Chromatogr. A 1070 (1–2), 57–64,
2005.
106. Schwientek, T., Mandel, U., Roth, U., Muller, S., and Hanisch, F. G., A serial lectin
approach to the mucin-type O-glycoproteome of Drosophila melanogaster S2 cells,
Proteomics 7 (18), 3264–3277, 2007.
107. Xiong, L. and Regnier, F. E., Use of a lectin affinity selector in the search for unusual
glycosylation in proteomics, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 782
(1–2), 405–418, 2002.
108. Hirabayashi, J., Hayama, K., Kaji, H., Isobe, T., and Kasai, K., Affinity capturing and
gene assignment of soluble glycoproteins produced by the nematode Caenorhabditis
elegans, J. Biochem. 132 (1), 103–114, 2002.
109. Zhang, H., Li, X. J., Martin, D. B., and Aebersold, R., Identification and quantification
of N-linked glycoproteins using hydrazide chemistry, stable isotope labeling and mass
spectrometry, Nat. Biotechnol. 21 (6), 660–666, 2003.
110. McDonald, C. A., Yang, J. Y., Marathe, V., Yen, T. Y., and Macher, B. A., Combining
results from lectin affinity chromatography and glycocapture approaches substantially
improves the coverage of the glycoproteome, Mol. Cell Proteomics 8 (2), 287–301,
2009.
111. Liu, T., Qian, W. J., Gritsenko, M. A., Camp, D. G., II, Monroe, M. E., Moore, R. J., and
Smith, R. D., Human plasma N-glycoproteome analysis by immunoaffinity subtrac-
tion, hydrazide chemistry, and mass spectrometry, J. Proteome Res. 4 (6), 2070–2080,
2005.
112. Ding, W., Hill, J. J., and Kelly, J., Selective enrichment of glycopeptides from glycopro-
tein digests using ion-pairing normal-phase liquid chromatography, Anal. Chem. 79 (23),
8891–8899, 2007.
113. Alvarez-Manilla, G., Atwood, J., III, Guo, Y., Warren, N. L., Orlando, R., and Pierce, M.,
Tools for glycoproteomic analysis: Size exclusion chromatography facilitates identifica-
tion of tryptic glycopeptides with N-linked glycosylation sites, J. Proteome Res. 5 (3),
701–708, 2006.

114. Larsen, M. R., Hojrup, P., and Roepstorff, P., Characterization of gel-separated glyco-
proteins using two-step proteolytic digestion combined with sequential microcolumns
and mass spectrometry, Mol. Cell Proteomics 4 (2), 107–119, 2005.
115. Bindila, L., Peter-Katalinic, J., and Zamfir, A., Sheathless reverse-polarity capillary
electrophoresis-electrospray-mass spectrometry for analysis of underivatized glycocon-
jugates, Electrophoresis 26 (7–8), 1488–1499, 2005.
116. Bond, M. R. and Kohler, J. J., Chemical methods for glycoprotein discovery, Curr. Opin.
Chem. Biol. 11 (1), 52–58, 2007.
117. Xu, Y. W., Wu, Z. X., Zhang, L. J., Lu, H. J., Yang, P. Y., Webley, P. A., and Zhao, D. Y.,
Highly specific enrichment of glycopeptides using boronic acid-functionalized mesopo-
rous silica, Anal. Chem. 81 (1), 503–508, 2009.
118. Khidekel, N., Arndt, S., Lamarre-Vincent, N., Lippert, A., Poulin-Kerstien, K. G.,
Ramakrishnan, B., Qasba, P. K., and Hsieh-Wilson, L. C., A chemoenzymatic approach
toward the rapid and sensitive detection of O-GlcNAc posttranslational modifications,
J. Am. Chem. Soc. 125 (52), 16162–16163, 2003.
119. Khidekel, N., Ficarro, S. B., Peters, E. C., and Hsieh-Wilson, L. C., Exploring the
O-GlcNAc proteome: Direct identification of O-GlcNAc-modified proteins from the
brain, Proc. Natl. Acad. Sci. U.S.A. 101 (36), 13132–13137, 2004.
120. Tai, H. C., Khidekel, N., Ficarro, S. B., Peters, E. C., and Hsieh-Wilson, L. C., Parallel
identification of O-GlcNAc-modified proteins from cell lysates, J. Am. Chem. Soc. 126 (34),
10500–10501, 2004.
121. Kurogochi, M., Amano, M., Fumoto, M., Takimoto, A., Kondo, H., and Nishimura, S.,
Reverse glycoblotting allows rapid-enrichment glycoproteomics of biopharmaceuticals
and disease-related biomarkers, Angew. Chem. Int. Ed. Engl. 46 (46), 8808–8813, 2007.
122. Bunkenborg, J., Hagglund, P., and Jensen, O. N., Modification-specific proteomic
analysis of glycoproteins in human body fluids by mass spectrometry, in Proteomics
of Human Body Fluids, ed. Thongboonkerd, V., Humana Press, Totowa, NJ, 2007,
pp. 107–128.
123. Kasicka, V., Recent developments in CE and CEC of peptides, Electrophoresis 29 (1),
179–206, 2008.
124. Gould, H. J. and Sutton, B. J., IgE in allergy and asthma today, Nat. Rev. Immunol. 8 (3),
205–217, 2008.
125. Kraft, S., Rana, S., Jouvin, M. H., and Kinet, J. P., The role of the FcepsilonRI beta-
chain in allergic diseases, Int. Arch. Allergy Immunol. 135 (1), 62–72, 2004.
126. Garman, S. C., Wurzburg, B. A., Tarchevskaya, S. S., Kinet, J. P., and Jardetzky, T. S.,
Structure of the Fc fragment of human IgE bound to its high-affinity receptor Fc epsil-
onRI alpha, Nature 406 (6793), 259–266, 2000.
127. Chang, V. T., Crispin, M., Aricescu, A. R., Harvey, D. J., Nettleship, J. E., Fennelly, J. A.,
Yu, C., Boles, K. S., Evans, E. J., Stuart, D. I., Dwek, R. A., Jones, E. Y., Owens, R. J.,
and Davis, S. J., Glycoprotein structural genomics: Solving the glycosylation problem,
Structure 15 (3), 267–273, 2007.
128. Nettleship, J. E., Rahman-Huq, N., and Owens, R. J., The production of glycopro-
teins by transient expression in Mammalian cells, Methods Mol. Biol. 498, 245–263,
2009.
129. Takegawa, Y., Deguchi, K., Ito, H., Keira, T., Nakagawa, H., and Nishimura, S., Simple
separation of isomeric sialylated N-glycopeptides by a zwitterionic type of hydrophilic
interaction chromatography, J. Sep. Sci. 29 (16), 2533–2540, 2006.
130. Hopp, T. P. and Woods, K. R., Prediction of protein antigenic determinants from amino
acid sequences, Proc. Natl. Acad. Sci. U.S.A. 78 (6), 3824–3828, 1981.
131. Letourneur, O., Sechi, S., Willette-Brown, J., Robertson, M. W., and Kinet, J. P.,
Glycosylation of human truncated Fc epsilon RI alpha chain is necessary for efficient
folding in the endoplasmic reticulum, J. Biol. Chem. 270 (14), 8249–8256, 1995.

132. Perkins, D. N., Pappin, D. J., Creasy, D. M., and Cottrell, J. S., Probability-based
protein identification by searching sequence databases using mass spectrometry data,
133. An, H. J., Peavy, T. R., Hedrick, J. L., and Lebrilla, C. B., Determination of
N-glycosylation sites and site heterogeneity in glycoproteins, Anal. Chem. 75 (20),
5628–5637, 2003.
134. Dodds, E. D., Seipert, R. R., Clowers, B. H., German, J. B., and Lebrilla, C. B., Analytical
performance of immobilized pronase for glycopeptide footprinting and implications for
surpassing reductionist glycoproteomics, J. Proteome Res. 8 (2), 502–512, 2009.
135. Lacunza, I., Kremmer, T., Diez-Masa, J. C., Sanz, J., and de Frutos, M., Comparison of
alpha-1-acid glycoprotein isoforms from healthy and cancer patients by capilliary IEF,
136. Schwarzer, J., Rapp, E., and Reichl, U., N-glycan analysis by CGE-LIF: profiling influ-
enza A virus hemagglutinin N-glycosylation during vaccine production, Electrophoresis
29 (20), 4203–4214, 2008.
137. Romisch, J., Donges, R., Stauss, H., Inthorn, D., Muhlbayer, D., Jochum, M., and
Hoffmann, J. N., Quantification of antithrombin isoform proportions in plasma samples
of healthy subjects, sepsis patients, and in antithrombin concentrates, Pathophysiol.
Haemost. Thromb. 32 (3), 143–150, 2002.
138. Block, T. M., Camunale, M. A., Lowman, M., Steel, L. F., Romano, P. R., Fimmel, C.,
Tennant, B. C., London, W. T., Evans, A. A., Blumberg, B. S., Dwek, R. A., Mattu, T. S.,
and Mehta, A. S., Use of targeted glycoproteins that correlate with liver cancer in wood-
chucks and humans, Proc. Natl. Acad. Sci. U.S.A. 102 (3), 779–784, 2005.

23 Analysis of Protein
Glycosylation and
Phosphorylation
Using HILIC-MS
Morten Thaysen-Andersen, Kasper
Engholm-Keller, and Peter Roepstorff
Contents
23.1 Introduction................................................................................................... 552
23.2 Background.................................................................................................... 552
23.3 Analysis of Glycans and Glycopeptides Using HILIC.................................. 554
23.3.1 HILIC SPE for Sample Preparation of Glycans and Glycopeptides.....554
23.3.1.1 Selective Enrichment of Glycopeptides from Purified/
Semi-Purified Glycoprotein................................................ 555
23.3.1.2 Enrichment and Desalting of Released Glycans................. 557
23.3.1.3 Quantitative Desalting of Glycopeptides and Glycans
Using HILIC SPE............................................................... 559
23.3.1.4 Enrichment of Glycopeptides from Complex Mixture
Using HILIC SPE............................................................... 559
23.3.2 HILIC Separation of Glycans and Glycopeptides............................. 563
23.3.2.1 HILIC Separation of Glycans with Off-Line
MS Detection................................................................ 563
23.3.2.2 HILIC Separation of Glycopeptides with Off-Line
MS Detection......................................................................564
23.3.2.3 Online HILIC-MS of Glycans............................................ 565
23.3.2.4 Online HILIC-MS of Glycopeptides.................................. 566
23.4 Analysis of Phosphopeptides Using HILIC................................................... 567
23.4.1 Phosphopeptide Enrichment Methods............................................... 567
23.4.1.1 Immobilized Metal Affinity Chromatography................... 567
23.4.1.2 Strong Cation Exchange...................................................... 567
23.4.1.3 Titanium Dioxide................................................................ 568
23.4.1.4 Other Phosphopeptide Enrichment Methods...................... 568
23.4.2 HILIC in Multidimentional LC-MS.................................................. 568
551

23.4.3 HILIC in Phosphorylation Analysis.................................................. 568

23.4.3.1 Early Applications of HILIC in Phosphoprotein/
Peptide Analysis.................................................................. 568
23.4.3.2 Application in Phosphoproteomics..................................... 569
23.4.4 Electrostatic Repulsion Hydrophilic Interaction Chromatography......570
23.4.5 Multiphosphopeptide Enrichment Prior to HILIC Fractionation...... 571
23.5 Future Prospects............................................................................................ 572
Acknowledgment.................................................................................................... 572
References............................................................................................................... 572
23.1 Introduction
Glycosylation and phosphorylation represent the majority of posttranslational protein
modifications and it is widely accepted that they are substantially involved in numer-
ous essential cellular processes such as inter- and intracellular signaling, metabo-
lism, protein synthesis and degradation, and cell survival. For this reason, functional
glycomics, glycoproteomics, and phosphoproteomics are rapidly growing research
areas. One of the most prominent challenges associated with these disciplines is the
substoichiometric presence of these modifications, resulting from substantial het-
erogeneity of glycosylation as well as frequently a low degree of phosphorylation
of a given site. Fractionation or enrichment of the modified proteins is consequently
essential to alleviate this problem. However, the significant hydrophilicities associ-
ated with these biomolecules limit the use of traditional purification techniques such
as reversed-phase liquid chromatography (RPLC) setups. In contrast, hydrophilic
interaction chromatography (HILIC), which features the opposite characteristics
of RPLC, is an attractive technique for the analysis of hydrophilic compounds and
applications usually involve detection by mass spectrometry (MS) to benefit from
its high sensitivity, accuracy, and resolution as well as its high throughput potential.
In this chapter, we describe various HILIC applications for detailed studies of
glycans, glycopeptides, and phosphopeptides. The applications range from sample
preparations using HILIC in solid-phase extraction (SPE) formats to chromato-
graphic separation using analytical and capillary-scale HILIC columns with both
off- and online detection. The advantages of combining HILIC strategies with down-
stream MS detection are emphasized.
23.2 Background
Following synthesis, proteins are often modified by covalent attachment of various
chemical groups. These posttranslational modifications (PTMs) can potentially alter
the physicochemical properties of the original proteins and thereby modulate their
activities. In particular, glycosylation and phosphorylation, which occur frequently
on proteins, are known to be involved in a multitude of cellular processes.
Protein glycosylation is the attachment of carbohydrate moieties (glycans) to the
polypeptide backbone and includes two major glycosylation types although sev-
eral other less abundant types are known. In N-linked glycosylation, the glycans
are linked to asparagine residues in the restricted sequence Asn-Xaa-Thr/Ser/Cys,

Analysis of Protein Glycosylation and Phosphorylation Using HILIC-MS 553
where Xaa is any residue except for proline. In contrast, no consensus sequence is
known for O-linked glycans, which are linked to the polypeptide chain through ser-
ine and threonine residues. Often substantial heterogeneity arises from the linkage
of numerous glycan structures to a given glycosylation site. It is estimated that more
than half of all proteins are glycosylated1 and the biological roles of glycosylation
are extremely diverse, spanning the spectrum from conformational stability and pro-
tection against degradation to molecular and cellular recognition in development,
growth, and cellular communication.2
Protein phosphorylation is the covalent coupling of a phosphate group to an amino
acid side chain in a polypeptide backbone and is catalyzed by protein kinases, which
accounts for ∼1.7% of the genes in the human genome.3 It has been estimated that
up to 30% of the proteins in eukaryotes are phosphorylated.4,5 In mammalian cells,
phosphorylation occur on the amino acids serine, threonine, and tyrosine, while the
basic amino acids histidine, arginine, and lysine are also prone to phosphorylation
in prokaryotes. The relative abundances of phosphoserine, phosphothreonine, and
phosphotyrosine, have been estimated to be approximately 90%, 10%, and 0.05%,
respectively.6 Reversible phosphorylation of proteins is an essential regulatory mech-
anism, which can alter catalytic activity, stability, and interaction with other bio-
molecules and is thus involved in most cellular processes such as inter-/intracellular
signaling, metabolism, protein synthesis and degradation, and apoptosis.7
Considering the significant involvement of protein glycosylation and phosphory-
lation in molecular/cellular processes and their abundances, it is no surprise that
these two PTMs have been intensely investigated. Although, understanding of the
role of the modifications is beginning to emerge on the individual protein level, the
complete involvement of the PTMs is far from known. One of the main challenges in
deciphering the PTM code arises from the fact that the modifications are produced
by enzymatic processing (non-template driven), meaning that the level and type of
modification of a given protein can vary depending on cell-type and the physio-
logical condition of the cell. The heterogeneity and substoichiometry of the PTMs
is another obstacle limiting the characterization, which makes sample enrichment
essential before PTM proteins analysis. Furthermore, the hydrophilicity associated
with the peptides bearing phosphorylations or glycosylations is posing yet another
challenge, as most traditional techniques such as reversed-phase liquid chromatog-
raphy are designed for analyzing molecules based on hydrophobicity. Together these
challenges call for sensitive, robust and reproducible techniques that allow separa-
tion and enrichment of peptides containing hydrophilic PTMs.
HILIC in conjunction with MS represents such a technique. The reversed reten-
tion characteristics compared to RPLC,8 makes HILIC ideal for the analysis of polar
and hydrophilic compounds such as glycosylated and phosphorylated species. An
additional benefit is that hydrophilic analytes have relatively high solubility in the
polar aqueous/organic mobile phases used in HILIC. Furthermore, the use of high
concentration of organic solvent, mainly acetonitrile (ACN) and volatile buffer sys-
tems is ideal for downstream MS detection in an on- or off-line configuration. HILIC
has experienced a sudden increase in popularity,9 which has been promoted by the
need to analyze hydrophilic analytes in mixtures and the increased use of LC-MS.
This, in turn, has been induced by some advances on the MS instrumental side

e.g., development of fast and accurate mass spectrometers with high resolving power
(e.g., Orbitrap) and the introduction of new fragmentation techniques (e.g., electron
transfer dissociation [ETD]), which has been shown to be very useful for fragmenta-
tion of peptides containing labile PTMs such as phosphorylation and glycosylation.
As a consequence, the HILIC-MS combination represents an attractive approach
to study hydrophilic and labile PTMs, and a number of new applications have been
published over the last 5–6 years. These include applications for studying protein
glycosylation and phophorylation spanning from sample enrichment and cleanup
during sample preparation to traditional chromatographic setups in which modified
peptides are separated using HILIC in an off- or online combination with MS.
23.3 Analysis of Glycans and Glycopeptides Using HILIC

23.3.1 HILIC SPE for Sample Preparation of Glycans and Glycopeptides
The combination of a relative poor ionization efficiency of glycopeptides with
respect to the unmodified peptides and the “dilution” of signals resulting from the
distribution of multiple glycoforms at the linkage sites, makes glycopeptide enrich-
ment essential prior to MS detection. Among various other sample preparation
techniques i.e., lectins,10,11 graphitized carbon,12,13 titanium dioxide (TiO2) (for nega-
tively charged compounds),14 hydrazide chemistry15 and boronic acids,16 which are
described in detail in recent reviews.17,18 HILIC SPE has been used increasingly to
enrich for glycopeptides. The rationale is that the hydrophilic contribution from the
glycan moiety is often sufficient to generate a rather unique overall hydrophilic-
ity among the glycopeptides, which can be used as a parameter to separate them
from the less hydrophilic non-glycosylated peptides on polar stationary phases using
hydrophilic interactions. The hydrophilicity of glycans is primarily a result of the
large number of hydroxy groups, and, as the hydrophilic contribution increases with
glycan size (and charge), applications have mainly been presented for N-linked gly-
copeptides, since N-glycans tend to be larger than O-linked glycans. For traditional
glycomic approaches, HILIC SPE also represents an ideal technique for clean-up
(desalting) of released glycans prior to MS.
The HILIC enrichment has been performed utilizing a variety of stationary phases
and SPE formats on samples ranging from purified glycoproteins to extremely com-
plex matrix, such as serum and other body fluids. Examples of various commercial
and noncommercial SPE formats are shown in Figure 23.1.
When choosing a format, it is important to consider the type of stationary phase,
column/membrane compatibility, and loading capacity/speed as well as mobile
phase. Ideally, it is preferred that the non-glycosylated peptides appear in the flow-
through, while the glycopeptides are selectively retained on the column and can be
eluted separately. As salts and detergents are eluted in the flow-through/wash frac-
tions, desalting of the sample is simultaneously performed. Desalting is important,
as salts interfere with the MS detection if present. Solvents used for HILIC SPE
purifications are usually similar to the ones used in regular HILIC separation (see
Section 23.3.2 HILIC separation of glycans and glycopeptides), but vary slightly
depending on the specific experiment and the downstream detection methods.

E.
A.
B.
C.
D.
Self-made ZIC -TIPS SPE prototype ZIC -HILIC LudgerCleanTM S

micro-column w. ZIC-HILIC ProteaTips cartridge
Figure 23.1 Examples of HILIC SPE formats. (A) Self-made micro-column packed in
GELoader tips (Eppendorf, Germany). See Thaysen-Andersen et al.20 for details on column
preparation and handling. Various HILIC SPE materials and column volumes can be used,
which are the major advantages of this format when optimizing the purifications for indi-
vidual experiments. (B) The format of the commercial available ZIC®-TIPS (Merck Sequant,
Umeå, Sweden) is similar to the self-made micro-columns; however, the zwitter-ionic HILIC
SPE material (ZIC-HILIC) forms a column higher in the tip. (C) This SPE prototype from
Merck Sequant (discontinued format) has a higher column volume and consequently a higher
capacity. (D) ZIC®-HILIC ProteaTips (Protea Biosciences, Morgantown, WV) is a variant
of the ZIC®-TIPS format using the same stationary phase. The tip is packed in standard P10
pipette tips and specific washing/loading and elution solvents can be purchased together with
the columns. (E) LudgerClean™ (Ludger, Oxfordshire, U.K.) is a commercial purification
cartridge containing a hydrophilic binding membrane. The size of the membrane enables
high capacity (up to 20â•›μg glycan).
Therefore, it is recommended to consult the listed references for details. Following

column equilibration, load and wash HILIC SPE in a suitable mobile phase of high
organic content, a one-step elution is normally performed by switching to an aque-
ous mobile phase either with or without a low concentration of organic solvent (typi-
cally 0%–30% ACN). Thus, water is the stronger eluting solvent, and a small amount
of weak acid (e.g., 0.5%–2% formic acid) is often included to generate protonated
species for MS. The one-step elution is carried out since fractionation of the retained
compounds often is not required. The use of MS-friendly solvents allows subsequent
analysis of the enriched glycopeptide/glycan fractions by electrospray ionization MS
(ESI-MS) or matrix-assisted laser desorption/ionization MS (MALDI-MS).
23.3.1.1 Selective Enrichment of Glycopeptides

from Purified/Semi-Purified Glycoprotein
One very useful HILIC SPE application is the enrichment of glycopeptides from rela-
tively simple mixtures i.e., tryptic digests of purified or semi-purified glycoproteins.

100 Non-glycosylated peptides

No purification
* Glycopeptides
%
* * * * *
0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800
(A) m/z
100 R2
%
0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800
(B) m/z
100
* HILIC
* *
* * *
*
%
* *
* *
* *
0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800
(C) m/z
Figure 23.2 MALDI-Q-TOF MS of a peptide mixture generated from purified recom-

binant plasminogen-activator inhibitor 1 (PAI-1). Three different sample preparations are
compared: (A) the peptide mixture spotted directly on the target as dried droplet with
no purification, (B) R2 (hydrophobic) purification in micro-column format, and (C) ZIC-
HILIC micro-column purification. The glycopeptides (marked with *) were barely observ-
able without purification and completely absent in the R2 purification. In contrast, the
glycopeptides were highly enriched in the HILIC SPE enabling easy characterization of
the glyco-conjugates based on their mass and on the subsequent fragmentation pattern
using MS/MS. Recombinant PAI-1 was kindly provided by Professor Peter Andreasen,
University of Aarhus, Denmark.
Here, HILIC has been shown to be capable of selectively retaining the glycopeptides.
The resulting depletion of non-glycosylated peptides is essential when MALDI-MS
or infusion ESI-MS is used, where no further analyte separation is performed prior
to MS. This is exemplified in Figure 23.2, where a peptide mixture obtained by tryp-
tic digestion of the N-glycosylated protein plasminogen-activator inhibitor 1 (PAI-1)
was prepared using three different approaches, all with MALDI-quadrupole time-of-
flight (Q-TOF) MS (and MS/MS) as detection.
When the sample was spotted directly on the target without any purification, the
signals for the non-glycosylated peptides were dominating, and the glycopeptides
were barely detectable due to suppression effects and signal dilution. Upon desalting

using the conventional hydrophobic resin (R2), the signals for the glycopeptides
were completely absent, indicating that the hydrophilic glycopeptides either were
non-retained on the hydrophobic stationary phase or were fully suppressed by the
non-glycosylated peptides. Upon HILIC purification on micro-columns, in contrast,
the signals for the glycopeptides were clearly present in the high mass region as
well as some fragmentation products in the lower mass region, whereas most of
the non-glycosylated peptides were depleted. This demonstrates the importance of
glycopeptide enrichment. The enrichment was performed using self-made micro-
columns in GELoader tips packed with zwitter-ionic material (ZIC-HILIC, Sequant/
Merck, Uppsala, Sweden) (see Figure 23.1A), but other HILIC phases and formats
are expected to give similar results.
The method for rapid and sensitive site-specific glycoprofiling of N-glycosylated
proteins is illustrated with some examples shown in Figure 23.3. Salivary glycopro-
teins were separated by SDS-PAGE and their glycoprofiles were determined using
MALDI-TOF MS and MS/MS following HILIC SPE enrichment of the glycopeptides.
Specifically, salivary tissue inhibitor of metalloproteinases-1 (TIMP-1), which
has two occupied N-glycosylation sites, was found to be extremely heterogeneously
glycosylated and it was possible to map 54 glycopeptide forms to the two sites. The
glycosylation pattern of extra parotid glycoprotein (EP-GP) containing a single
occupied N-linked site was also characterized to be heterogeneous. Both profiles
were rich in glycans containing terminal fucose residues, which is a known feature
of saliva glycoproteins.19 Recently, the sensitivity of the approach was illustrated by
the glycoprofiling of low abundant glycoproteins in plasma20 and the strategy could
be used to investigate the potential of glycosylated proteins as biomarkers.21 The
same approach using ZIC-HILIC SPE has also been used for the selective purifica-
tion and enrichment of glycophosphatidylinosito (GPI)-anchored peptides.22
Other formats than the HILIC GELoader tips have been proven useful for purifi-
cation. For example, Yu et al. presented an approach where a 96-well micro-elution
plate was packed with 5â•›mg HILIC stationary phase i.e., aminopropyl silica per well.23
In this setup, both released N-linked glycans and pronase-generated glycopeptides
were shown to be selectively enriched and desalted using a mobile phase consisting
of ACN and aqueous ammonium citrate (10â•›m M).24 In another study, hydrophilic
affinity separation of oligosaccharides and glycopeptides was obtained by partition-
ing with cellulose or Sepharose in a microcentrifuge tube in a batch mode.25,26 Using
a mobile phase consisting of 1-butanol/ethanol/water (4:1:1, v/v) impressive separa-
tion was achieved.
23.3.1.2 Enrichment and Desalting of Released Glycans

Enzymatic or chemical release of N- and O-linked glycans from glycoprotein are
commonly used approaches in glycomics to determine the global glycan profile.
Due to lower mass and the hydrophilic nature, the released glycans are easily puri-
fied from the proteins. Following isolation using high molecular mass cutoff filters/Â�
membranes or graphitized carbon columns, the glycans are often derivatized using
a variety of methods e.g., fluorescence labeling27 or permethylation28 in order to
enhance detection and identification of the glycans. The introduction of salts and
detergents in these derivatization steps calls for thorough desalting of the sample

100
3667.5
kDa -HNR- -HNR- -HNR- -VNQ- -VNQ- -VNQ- -VNQ- -HNR- -VNQ- -VNQ-
1 2 3 4 5
55 11 12 13 14
45
-HNR- -VNQ- -HNR- -VNQ- -VNQ- -VNQ- -VNQ- -VNQ- -HNR-
6 7 8 9 10
35
15 16 17 18
-HNR- -VNQ- -VNQ-
3521.4
3813.6
3402.3
TIMP-1
Intensity
Proteolysis +
25 extraction HILIC -HNR-
SPE -HNR-
of peptides -VNQ-
-VNQ- -HNR-
-VNQ-
3443.4
-VNQ-
2994.3
15 -VNQ-
3256.3
-VNQ-
4178.8
-VNQ- -VNQ- 4
-HNR-
-VNQ-
3505.4
3156.3
3548.4
-VNQ- -VNQ- -VNQ-
3140.3
-HNR- 6
3913.6
-HNR- -VNQ-
10
3651.5
3708.5
2 -HNR-
3359.3
3589.4
3240.3
-HNR- 3
3302.3
3767.5
3297.3
-HNR- -HNR- -HNR- 5 -VNQ-
3094.3
-VNQ- 11 12
4325.0
3484.4
4032.7
2891.3
-HNR-
3037.2
3959.7
7
2729.2
8 9
2832.3
3562.4
1
4016.7
-VNQ- -VNQ-
4059.7
2875.2
3854.6
2848.2
2932.2
-HNR- -VNQ-
3692.5
-HNR-
3870.6
10
3751.5
13
SDS-PAGE of
3078.3
2629.2
14 16
3975.7
18
4105.8
17
4425.0
4162.8
15
4471.2
4544.2
4278.9
4382.0
affinity purified
4219.8
4528.2
4571.1
4366.1
saliva
0
2600 m/z 4500
3327.3
-TNR
-TNR
EP-GP
3473.4
Intensity
-TNR
2816.2
-TNR -TNR
3165.2
-TNR -TNR
3187.2
-TNR
3165.2
-TNR -TNR
3019.2
3838.5
-TNR
2962.2
3692.4
-TNR
3619.4
3222.3
-TNR
-TNR
3530.4
-TNR -TNR
3984.6
-TNR
TNR
3384.3
3546.4
-TNR
3765.5
-TNR -TNR
3676.5
3311.3
-TNR
3368.3
-TNR
3108.2
4130.7
4276.8
4422.8
0
2800 m/z 4600
Figure 23.3 Rapid and sensitive site-specific glycoprofiling of affinity purified salivary N-glycoproteins separated on SDS-PAGE using MALDI-MS.
The stained gel bands were excised, digested with trypsin, and the extracted peptide mixtures were purified using ZIC-HILIC SPE as previously
described.20 The signals for glycopeptides could be assigned based on MS and MS/MS data. The upper gel band (and mass spectrum) was found to
contain tissue inhibitor of metalloproteinases-1 (TIMP-1) and the lower band contained extra parotid glycoprotein (EP-GP). Monosaccharide code:
HexNAc: filled square, Hex: filled circle, Fucose: open triangle.

prior to MS. Together with graphitized carbon purification, HILIC SPE represents a
widely used method for glycan cleanup and, as presented next, desalting of glycans
can be performed without loss of quantitative information.
23.3.1.3 Quantitative Desalting of Glycopeptides and Glycans

Using HILIC SPE
HILIC, in contrast to lectins, is considered to be a non-biased matrix that do not
select for certain structural classes or subclasses of the glycome/glycoproteome.
However, all sample handling, including purification, can potentially generate a
bias by Â�changing the composition of the species in a given mixture resulting in loss
of quantitative information. In a recent study, a number of SPE materials includ-
ing these with HILIC character were investigated for bias when glycopeptides and
released (free) glycans of ribonuclease B (RNase B) were desalted.29 Here, a fixed
amount of sample (glycans/glycopeptides) was applied to HILIC columns of vary-
ing volume and the glycoprofile of the retained fraction was subsequently deter-
mined using MALDI-TOF-MS. The obtained profiles were then evaluated against
a reference profile obtained by fluorescence detection of 2-aminobenzamide
(2-AB) labeled glycans after separation by HPLC. For ZIC-HILIC SPE desalt-
ing of both glycans and glycopeptides, biases were introduced, when the column
capacities were exceeded due to the higher ratios of sample amount versus column
volume (>50â•›fmol/nL), as shown in Figure 23.4A,B. This was a result of competi-
tive binding where the normally abundant Man 5–6 glycopeptides/glycans were
outcompeted by the more hydrophilic Man 7–9 glycopeptides/glycans. However,
using columns of sufficient capacities for the applied amount of glycan and glyco-
peptides (50â•›fmol/nL), the Â�purifications were essentially non-biased.
The same trend was observed for a variety of other HILIC stationary phases i.e.,
PolyHydroxyEthyl A, TSK-Gel amide 80, PolySulfoEthyl A, and LudgerClean S,
as shown in Figure 23.4C. This observation stresses the need for sufficient column
capacity when performing HILIC SPE purification. For enrichment from complex
mixtures, however, it should be noted that columns with capacities far greater than
needed may increase the binding of non-glycosylated peptides (Thaysen-Andersen,
unpublished data). Thus, the column capacity should be optimized to the analyte
amount and the sample complexity.
23.3.1.4 Enrichment of Glycopeptides from Complex Mixture

Using HILIC SPE
Self-made ZIC-HILIC SPE micro-columns have also been used to enrich glycopep-
tides from complex mixtures. This approach has mostly been used to determine
N-glycosylation sites of glycoproteins and a general workflow, based on the available
literature, is presented in Figure 23.5.
In 2004, Hagglund et al. enriched glycopeptides from lectin-purified plasma using
HILIC SPE. Following the removal of the glycan moiety except for the innermost
N-acetylglucosamine (GlcNAc) residue using endo-β-N-acetylglucosaminidase,
62 glycosylation sites were unambiguously identified from 37 glycoproteins using
LC-ESI-MS/MS.30 The unambiguous assignment was elicited by the unique mass

60.00%
60.00% ZIC-HILIC glycopeptide 600 fmol/nL
ZIC-HILIC glycan 600 fmol/nL
300 fmol/nL
300 fmol/nL
50.00% 50.00% 200 fmol/nL
200 fmol/nL
125 fmol/nL 125 fmol/nL
50 fmol/nL 40.00% 50 fmol/nL
40.00%
Rel. abundance
NP-HPLC
Rel. abundance
NP-HPLC
30.00% 30.00%
20.00% 20.00%
10.00% 10.00%
0.00% 0.00%
(A) Man 5 Man 6 Man 7 Man 8 Man 9 (B) Man 5 Man 6 Man 7 Man 8 Man 9
â•…
Figure 23.4 Test for bias of ZIC-HILIC SPE self-made micro-columns when used for desalting of bovine pancreatic RNase B (A) glycans and (B)
glycopeptides. Fixed sample amounts were applied to columns of different volumes (see insert for sample amount/column volume ratios and color/pat-
tern coding of the bars) and the retained fractions were glycoprofiled using MALDI-TOF MS. These profiles were compared to a reference HPLC profile
generated using fluorescence labeled (2-AB) N-glycan (black outline). The correct glycoprofile as evaluated by the reference could be obtained for both
the glycans and glycopeptides MS quantitation using rather low sample amount/column volume ratios (50â•›fmol/nL). Significant biases as illustrated by
an overrepresentation of more hydrophilic glycoforms (Man 7–9) were introduced when this ratio was exceeded (>50â•›fmol/nL). Thus, columns of suf-
ficient capacities were needed to avoid loss of quantitative information.

Analysis of Protein Glycosylation and Phosphorylation Using HILIC-MS
60.00%
HILIC type comparison (glycopeptide) No desalting
50.00% ZIC-HILIC (50 fmol/nL)
PolyHydroxyethyl A (50 fmol/nL)
PolySulfoethyl A (220 fmol/nL)
40.00% TSK-Gel Amide-80 (50 fmol/nL)
Rel. abundance
LudgerClean S (35 fmol/nL)

30.00% NP-HPLC
20.00%
10.00%
0.00%
(C) Man 5 Man 6 Man 7 Man 8 Man 9
Figure 23.4 (continued) (C) The same tendency to introduce a bias has been shown for other HILIC SPE materials29; however, the correct glyco-
profile could be obtained for all the materials tested when columns of sufficient capacities were used. The ratios used for obtaining non-biased profiles
are listed for the individual HILIC types.
561
Complex sample
e.g., Serum, bodyfluids
Proteolysis e.g., trypsin, pronase
Reduce sample complexity e.g., with lectin
(before or after proteolysis)
Mixture of peptides +
glycopeptides
HILIC-SPE
Glycopeptides
(+hydrophilic peptides)
Enzymatic deglycosylation e.g., PNGase F/A, Endo H/D

18O-labeling possible
Deglycosylated glycopeptide
(+hydrophilic peptides)
Use mass increment to validate glycosylation site
ESI-LC-MS/MS
Figure 23.5 General workflow for the identification of N-linked glycosylation sites from
complex samples. The enrichment of glycopeptides from the peptide mixture is essential
and can be performed using HILICSPE. Following partial or complete deglycosylation using
endoglycosidase H/D or PNGase F/A deglycosylated peptides can be identified in an ESI-
LC-MS/MS setup. The identification is based on the mass increment associated with the
deglycosylation and can be combined with O18 labeling.
increment of the (glyco)peptide moiety of 203 and 349â•›Da corresponding to the mass
of the innermost GlcNAc without and with an attached fucose residue, respectively.
Using the similar strategy, 103 N-glycosylation sites were identified from Cohn IV
fraction of human plasma.31 Lately, 63 N-glycosylation sites were mapped out from
32 glycoproteins obtained from human breast milk using complete enzymatic degly-
cosylation with N-glycosidase F and the mass increment of 0.98â•›Da as being the
positive identifier of a glycosylation site in the LC-ESI-MS.32 The major pitfall of
using the asparagine-to-aspartate conversion is that the approach does not distin-
guish between deglycosylation-induced conversion and other causes of deamidation,
either in vivo or in vitro. Deglycosylation in heavy isotopic water (H218O) has been
performed as the +2.98â•›Da mass shift is more easily recognized,33 but the drawback
here is that the 18O also can be incorporated in the C-terminus of tryptic peptides in
case of residual tryptic activity, thereby complicating the identification and increas-
ing the false positive rate. In another study utilizing ZIC-HILIC SPE, glycopeptides

were enriched from an extremely complex mixture of tryptic peptides obtained from
serum with or without previous lectin purification.34 Following enzymatic degly-
cosylation, the N-glycosylation sites were identified using both MALDI-MS and
LC-ESI-MS/MS. In this study, 86 (with lectin purification) and 81 (without lectin
purification) sites were identified. A limited overlap between the identified sites led
to the conclusion that these two approaches were complementary.
23.3.2 HILIC Separation of Glycans and Glycopeptides

Although the separation efficiency of HILIC may not be as good as that of RP-HPLC
separations (wider peaks and consequently lower resolution), HILIC is still an effi-
cient method for retention and separation of polar compounds, such as glycans and
glycopeptides. Applications include separations of different glycoforms of partially
purified glycan/glycopeptide fractions with off- or online MS detection as well as
more crude separations of non-glycosylated species from glyco-conjugates in pre-
fractionation approaches. In contrast to RP-HPLC, which represents a rather homoge-
neous type of chromatography, the retention mechanism of HILIC varies depending
on the functional groups of the stationary phase and the analyte, as well as the nature
of the mobile phase. Thus, the stationary and mobile phases are essential parameters
to consider for optimal analyte retention and separation. As reviewed recently, a
selection of HILIC stationary phases are available and will not be covered in details
here.35 For the mobile phase, solvents with a high content of organic component
(40%–97%) in water are generally used. Dissolving glycans and glycopeptides in
high organic mobile phases could be difficult, and as a consequence, an 80% organic
solvent water mixture is typically used in the mobile phase as starting condition.
ACN is by far the most popular organic modifier that hydrates the stationary phase
and generates reproducible results. At least 3% water should be present. Suitable
buffers include ammonium salts of acetate and formate due to their volatibility and
excellent solubility in organic solvent. These salts have been shown to minimize
the electrostatic interactions (repulsions or attractions) between charged stationary
phases and analytes (i.e., sialic acids).36 Formate and other weak acids are often used
as mobile phase additives for adjusting pH when using online HILIC-MS detection.
23.3.2.1 HILIC Separation of Glycans with Off-Line MS Detection

Separation and profiling of released glycans using HILIC (alternatively called
NP-HPLC by other authors in literature) coupled with a fluorescence detector has been
the standard technique for decades together with high-performance anion-exchange
chromatography with pulsed amperometric detection.37–39 Often analytical-scale
HILIC columns are used and detection limits around 50–100 fmol for 2-AB labeled
glycans have been reported.40 In a typical setup, separation of glycans is performed
on a TSK-Gel amide 80 column (TOSOH, Japan) in a mixture of high ACN amount
and 50â•›m M ammonia formate in water, pH 4.4, as the eluent. Gradient elution is nor-
mally performed by slowly decreasing the concentration of the organic solvent, and a
120–180â•›min run is frequently used. Although this approach is limited in resolution,
sensitivity, and throughput, it benefits from easy quantitation of the separated gly-
cans by measuring the peak area of the eluting analytes. Another advantage is that

Man 5
20
15 2-AB 2-AB 2-AB 2-AB 2-AB

Man 6
Man 5 Man 6 Man 7 Man 8 Man 9
LU
10 (47%) (28%) (8%) (13%) (4%)

Man 8
5 Man 7
Man 9
0
34 36 38 40 42 44 46 48
min
Figure 23.6 HILIC separation (traditionally named as NP-HPLC in the literature for
this type of application) of released and 2-AB labeled RNase B N-glycans using a TSK-Gel
amide 80 column (2 × 150â•›m m, 5â•›μm). The five glycoforms (Man 5–9), which are illustrated,
could easily be separated using gradient elution. Isomeric glycan structures were observed for
Man 7, which eluted as two peaks. Based on fluorescence detection, the relative abundances
of the glycoforms could be determined using the peak area of the eluting analytes. Starting
mobile phase of 80% ACN in aqueous 50â•›m M ammonium formate, pH 4.4, was used and a
gradient of decreasing ACN content was introduced (not shown).
not only the size but also the structure of the glycans affect the retention time often
providing information of isomeric structures, that cannot be distinguished in MS due
to identical m/z. The identification of the eluting compounds is usually performed by
matching retention times to an existing library and to a reference dextran ladder or
simply by collection of the eluted fractions with subsequent MS detection. The five
glycoforms of RNase B (not counting isomeric structures) could easily be separated
using a TSK-Gel amide 80 HILIC column and a relatively short gradient of 60â•›min
(Figure 23.6). The retention pattern illustrates longer retention times for the larger
glycans in HILIC mode due to increased hydrophilicity.
More complex mixtures can be separated by extending the gradient or includ-
ing an additional separation step. For example, HILIC fractions can be applied to
a second chromatographic dimension with online MS detection. This was recently
performed in a study where HILIC-separated glycans were collected and applied
to RP-nano-LC-ESI-MS.41 In another variant, capillary-scale HILIC separation of
glycans has been combined with automatic spotting on a MALDI target plate with
subsequent automated MS acquisition.42,43
23.3.2.2 HILIC Separation of Glycopeptides with Off-Line MS Detection

Although not widely used, glycopeptides can be separated by HILIC and detected
by their UV absorbance. Since glycans have almost no UV absorbance, the detection
is based on the absorbance of the peptide moiety. The eluted fractions are then col-
lected and characterized using MS in an off-line setup. Figure 23.7 shows an exam-
ple of this approach, where an RNase B tryptic digest was separated on a TSK-Gel
amide 80 column. The non-glycosylated peptides of the mixture eluted early in the

Salts + detergents
NLTK
Non-glycosylated peptides
NLTK
Absorbance (220 nm)
NLTK
NLTK
NLTK
Glycopeptides
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 min
Figure 23.7 HILIC separation of a peptide mixture generated by trypsin digestion of

bovine pancreatic RNase B. A TSK-Gel amide 80 column (2 × 150â•›m m, 5â•›μm) was used and
the gradient of decreasing ACN concentration is indicated with a dotted line. Salts, deter-
gents, and very hydrophobic peptides eluted in flow-through or very early in the gradient,
whereas the majority of the non-glycosylated peptides and glycopeptides were retained fairly
well on the HILIC column. The glycopeptides were fractionated from the non-glycosylated
peptides by late elution and the five glycopeptides representing the individual RNase B gly-
coforms could be separated (see insert).
gradient, whereas the glycopeptides were better retained on the HILIC column and
eluted separately. In addition, the five glycopeptides (Man 5–9) bearing the five high
mannose glycoforms were resolved enabling relative quantitation of the glycoforms.
This approach has been used to quantitatively monitor variations in the micro-
heterogeneity of glycoproteins i.e., antennary and sialylation profile of recombinant
human interferon-γ from Chinese hamster ovary cell cultures.44 Here, an initial
separation of glycopeptides and non-glycosylated peptides was performed using
RP-HPLC. Off-line HILIC-MS can also be used for prefractionation of glycopep-
tides from non-glycosylated peptides originating from crude mixtures, since gly-
copeptides in general will be more hydrophilic and therefore better retained on
HILIC. However, due to the likely presence of hydrophilic non-glycosylated peptides
in the glycopeptide fractions, a secondary separation step is required (e.g., online
RPLC-MS).
23.3.2.3 Online HILIC-MS of Glycans

The use of high-polar organic/low aqueous solvent content in the HILIC mobile
phase is ideal for direct ESI-MS detection, and this online approach has been used
increasingly in recent years (a review by Nguyen and Schug in 2008 summarized

fundamentals and applications of the technique45). The analysis of released glycans

has been achieved with great success using online HILIC-ESI-MS with various sta-
tionary phases (e.g., ZIC-HILIC, TSK-Gel amide 80, and PolyHydroxyEthyl A), gly-
can types (e.g., N- and O-linked), and glycan derivatization levels (e.g., free, 2-AB
or ANTS labeled)6–48 as recently reviewed.49 Regular HILIC conditions are used for
online approaches: high concentration of organic solvent and low concentration of
volatile salts and weak acids (i.e., formic acid or acetic acid). It is important, that the
amount of ion pairing agents like TFA should be minimized, because it reduces the
MS detection sensitivity significantly.50 The elution of glycans is normally observed
at above 50% ACN. The detection is typically performed in positive ionization
mode; however, negative ionization mode has also proved efficient in particular for
sialylated and sulfated glycans.36,51 Most studies using online HILIC-ESI-MS use
flow rates of 40–200â•›μL/min, but nanoflows (300â•›nL/min) have also been reported,
where detection limits in the low femtomol range have been achieved.46,48 Lately, a
chip-based HILIC-ESI-MS platform has been introduced, where glycosaminogly-
cans are separated and profiled in negative ion mode.52
23.3.2.4 Online HILIC-MS of Glycopeptides

In the last few years in the field of functional glycomics and glycoproteomics,
reports of glycopeptide separation using HILIC with online MS detection are mov-
ing toward site-specific approaches. Glycopeptide profiling is expected to be used
increasingly in the future. Takegawa et al. showed that N-glycopeptides of human
serum IgG could be profiled from a peptide mixture without previous enrichment
using ZIC-HILIC on an analytical-scale column.51 This was possible due to the
strong retention of glycopeptides compared to the non-glycosylated peptides, which
eluted early in the gradient. The same investigators tested the approach by applying a
tryptic peptide mixture of α-1-acid glycoprotein to a similar HILIC-ESI-MS setup.36
The sialylated and neutral glycopeptides were well-separated at the end of the gra-
dient, and the glycopeptides were efficiently characterized using ESI-ion trap-MS
in a negative ionization mode. Recently, capillary HILIC-ESI-MS has been used
to profile 8 O- and 105 N-linked Glu-C digested glycopeptides from recombinant
human erythropoietin starting from as low as 150â•›ng protein.53 In a different strat-
egy, pronase, which is a mixture of bacterial proteases, has been used to generate
very small glycopeptides. Pronase is known to cleave the proteins to single amino
acids and small peptides, but due to the steric hindrance of the glycan, the pep-
tide part around the glycosylation site usually contains 3–5 amino acid residues. To
demonstrate the potential of this approach, Wuhrer and coworkers showed that pro-
nase digested RNase B glycopeptides could efficiently be analyzed by nano-HILIC-
ESI-MS to identify both the peptide and the glycan.54 In order to obtain peptide
sequence information of glycopeptides using collision induced dissociation (CID)
fragmentation, multistage MS/MS, is required since the first fragmentation (MS2)
often gives information only about the glycan structure. A subsequent fragmentation
of the deglycosylated fragment in MS3 or MS4 results in peptide fragmentation infor-
mation. To avoid multiple ion isolation/fragmentation cycles, ETD has been demon-
strated to provide valuable peptide sequence information directly at the MS2 stage
and can consequently be used in combination with CID to yield complimentary

ions in MS/MS.17 O-linked glycopeptides generated from pronase digestion of the

β-chain of human choriogonadotropin have similarly been applied to nano-HILIC-
ESI-MS, which allowed identification of the four O-glycosylasion sites of the protein
as well as the structure of the attached glycans.49 Finally, an automated setup using
the pronase treatment approach has recently been presented, where glycoproteins are
digested on-column with pronase and the resulting glycopeptides are subsequently
collected on a graphitized carbon column followed by separation and detection by
HILIC-ESI-MS.55
23.4 Analysis of Phosphopeptides Using HILIC

Despite HILIC being utilized in glycopeptide analysis for a number of years, this
chromatographic technique has only quite recently been extensively applied in the
field of phosphopeptide analysis and fractionation. Like glycosylation, phosphoryla-
tion increases the hydrophilicity of a given peptide, making peptides containing this
PTM suitable for HILIC. Due to their sub-stoichiometric abundance and generally
lower ionization efficiencies compared to their non-phosphorylated counterparts,56
phosphopeptides are rarely detected in complex mixtures, and specific enrichment
strategies have therefore been developed to circumvent these problems.
23.4.1 Phosphopeptide Enrichment Methods

23.4.1.1 Immobilized Metal Affinity Chromatography
Immobilized metal affinity chromatography (IMAC) is one of the most extensively
used enrichment methods prior to mass spectrometric analysis of protein phos-
phorylation. It was first introduced for separation of phosphoproteins,57 but was
later adapted in phosphopeptide enrichment.58 IMAC is based on the affinity of
the anionic phosphoryl group for metal ions, e.g., Fe3+ or Ga3+. To facilitate an easy
enrichment procedure, these metal ions are chelated to a solid resin such as nitri-
lotriacetate coated beads. Although effectively capturing phosphopeptides, IMAC
suffers from co-enrichment of acidic peptides, but have nevertheless successfully
been applied to phosphopeptide enrichment in several large-scale phosphopro-
teomics studies.59,60
23.4.1.2 Strong Cation Exchange

Traditionally, strong cation exchange (SCX) has been a popular first dimension in
2D LC-MS setups, whether off- or online, as in the so-called MudPIT strategy.61
Most typical tryptic peptides only contain a single basic residue (lysine or arginine).
In the acidic buffers used for SCX fractionation, the phosphorylated forms of these
peptides will therefore overall be neutral or negatively charged due to the negatively
charged phosphoryl group. These phosphopeptides thus elute in the flow-through
or very early in the SCX gradient. This effect has successfully been exploited for
phosphopeptide enrichment by Beausoleil et al.62 However, all peptides containing
histidines or missed cleavages will be positively charged and will elute later in the
gradient, reducing the specificity of the method.

23.4.1.3 Titanium Dioxide

Recently, TiO2 was introduced as a chromatographic resin with phosphopeptide-
specific affinity in an online 2D-LC-MS setup.63 As for IMAC, the resin binds acidic
peptides as well as phosphopeptides, but this unspecific binding was shown to be
abrogated by the use of very acidic buffers (<pH 1) and addition of specific organic
acids (e.g., 2,5-dihydroxybenzoic acid).64 Therefore, TiO2 has gained widespread use
in large-scale proteomic experiments.65
23.4.1.4 Other Phosphopeptide Enrichment Methods

Antibodies targeting phosphorylated amino acid residues have been successful in
specific enrichment of tyrosine-phosphorylated peptides.66,67 A phosphopeptide pre-
cipitation method using calcium phosphate prior to IMAC has also been devised.68
Finally, phosphoproteomic researchers have developed various chemical derivatiza-
tion strategies based on phosphoamidate chemistry69 or beta elimination of the
phosphoryl group on serines and threonines followed by coupling to a stationary
phase.70,71 The chemical reactions performed in these strategies, however, usually
introduce further complexity into the already very complex phosphopeptide sam-
ples, and enrich only for primarily monophosphorylated peptides, and have thus had
limited use in large-scale phosphoproteomics.
23.4.2 HILIC in Multidimentional LC-MS

As previously mentioned, up to 30% of all eukaryotic proteins are prone to phos-
phorylation at one or multiple sites. Consequently, the phosphoproteome may be as
complex as the “standard” proteome. Efficient fractionation strategies, which can
sort the proteomic samples according to specific physicochemical properties, are
therefore required to reduce the overall complexity of the sample. Multidimensional
LC peptide fractionation coupled to MS has thus become the method of choice
in phosphoproteomics. In 2005, Gilar et al. investigated the orthogonality of dif-
ferent 2D LC-MS setups by comparing the elution profiles of a peptide mixture
containing ∼200 different species using size exclusion chromatrography, RPLC
under acidic and basic conditions, SCX, and HILIC on a bare silica column.72 The
combination of HILIC and RPLC showed a higher peak capacity and orthogonality
compared to SCX-RPLC, in which peptides eluted in distinct clusters according to
their charge. Despite having not actually coupled the two chromatographic dimen-
sions experimentally, this study showed the potential of HILIC as the first dimension
in a 2D-LC setup in proteomics.
23.4.3 HILIC in Phosphorylation Analysis

23.4.3.1 Early Applications of HILIC in Phosphoprotein/Peptide Analysis
One of the first applications of HILIC for the separation of phosphorylated peptides
was published in 1992. Here, HILIC was used to separate a 32P-labeled phosphory-
lated peptide from its unreacted, non-phosphorylated counterpart, γ-32P-ATP, and
32P-orthophosphate after an in vitro tyrosine kinase reaction.73 The different species

in the reaction mixture were separated on a PolyHydroxyEthyl A column in triethyl-

ammonium phosphate (TEAP) buffers at pH 2.8. A gradient from 90% ACN, 4â•›m M
TEAP to 0% ACN, 10â•›m M TEAP was first applied. During this gradient, the phos-
phorylated product could be resolved from the non-phosphorylated substrate and
the 32P-orthophosphate. The elution of the γ-32P-ATP required an increase in TEAP
concentration up to 800â•›m M.
Five years later, in 1997, histone H1.1 variants were fractionated on a PolyCat
A, a weak cation exchange column under HILIC conditions (a methanephosphonic
acid/triethylamine buffer with 70% ACN at pH 3.0).74 Elution with an increasing
sodium perchlorate gradient facilitated the separation of the histones into their
distinct non-, mono-, doubly, and triply phosphorylated forms. It was shown that
this separation could not be achieved running the column under regular cation
exchange conditions (sodium phosphate buffer with increasing sodium perchlorate
gradient). No mass spectrometric detection was used for the identification of the
phosphorylation state of the histones—instead the histones were treated with alka-
line phosphatase to remove the phosphate groups. By comparison of the chromato-
graphic behavior before and after the treatment, the phosphorylation states could
be deduced.
23.4.3.2 Application in Phosphoproteomics

The high potential of HILIC in fractionation of phosphopeptides was first demon-
strated by McNulty and Annan.75 Phosphopeptides were separated on a TSK-Gel
amide 80 carbamoyl-derivatized silica column and analyzed by RPLC-MS on a lin-
ear ion trap MS. This fractionation resulted in a similar number of phosphopeptide
identifications as obtained using SCX-RPLC-MS, but had the advantage of using vol-
atile salt-free solvents and buffers (e.g., ACN and TFA), which were directly compat-
ible with the next IMAC-based phosphopeptide enrichment. A reduction in sample
complexity was achieved by eluting a significant fraction of the non-phosphorylated
peptides in the flow through from the HILIC column, and furthermore fractionating
the later eluting phosphopeptides using a decreasing organic gradient. IMAC is
generally known to co-purify acidic peptides, but the reduction in sample complex-
ity resulted in very pure phosphopeptide fractions (>99%). However, less than 10%
of the phosphopeptides with multiple phosphorylation sites were identified, which is
significantly less than generally observed in large-scale experiments using SCX.60,65
In total, more than 1000 unique phosphorylation sites were identified from 300â•›μg of
calyculin A-treated HeLa cells.
The combination of HILIC and IMAC prior to RPLC-MS was also employed in a
very similar study, to investigate the phosphoproteome in the yeast Saccharomyces
cerevisiae after DNA damage.76 Here, the analysis resulted in the identification of
more than 8000 unique phosphopeptides. However, these numbers cannot be com-
pared directly with McNulty and Annan’s, as the type of sample, sample amounts,
database search algorithms, and validation criteria were not the same. The potential
of the strategy was further emphasized by the analysis of two low-abundant proteins
(Rad9 and Mrc1). Whereas immunoprecipitation of these specific proteins from
500â•›mg of yeast sample followed by phosphopeptide analysis by LC-MS (without
HILIC fractionation) resulted in 40 phosphopeptide identifications from Rad9 and

31 from Mrc1, respectively. LC-MS analysis of HILIC fractionation of 6â•›mg sample

(80 times less than that without HILIC fractionation) allowed for the identification
of 21 phosphopeptides from Rad9 and 14 from Mrc1, respectively. For comparison,
SCX fractionation of 50â•›mg of sample only resulted in five and two phosphopeptide
identifications, respectively.
The buffers for HILIC and RPLC are not directly compatible. Therefore, auto-
mated online coupling of these two chromatographic dimensions cannot be achieved.
However, Boersema et al. recently devised a semi-automated HILIC-RPLC-MS
setup.77 A ZIC-HILIC microliter flow system was coupled to a microliter flow-scale
fraction collector to facilitate mixing of the HILIC output with formic acid buf-
fer. Hereby, the organic concentration of HILIC fractions is decreased to a level
compatible with RPLC. The distribution of peptides containing different PTMs was
evaluated in this study, and while selectively enriching for N-acetylated peptides,
no enrichment of phosphorylated peptides was achieved. As the authors specu-
lated, the zwitter-ionic character of the functional groups on the chromatographic
material might cause this effect: The negative sulfate group could potentially repel
the phosphate group on the peptide, making the phosphopeptides elute along with
the unmodified peptides.
23.4.4 Electrostatic Repulsion Hydrophilic Interaction Chromatography

Unless the analytes and/or the chromatographic material are completely uncharged
at the pH used for the LC separation, the separation of the analytes in HILIC will
be influenced by electrostatic interactions. The use of ion exchange material in
HILIC mode for the retention of analytes with the same charge as the stationary
phase through hydrophilic interactions has been termed electrostatic repulsion-
hydrophilic interaction chromatography (ERLIC) by Alpert in 2008.78 In this pub-
lication, ERLIC was performed on a PolyWAX LP column separating a peptide
standard mixture. On this weak anion exchange column, the type of salt in the
buffers strongly affected the retention of the different types of peptides in the
standard mixture. Basic peptides were retained strongly in TEAP-containing buf-
fers, while eluting very early in sodium methylphosphonate (Na-MePO4) buffers.
This was explained by the phosphate ion in the TEAP buffer acting as a counter-
ion to the positive stationary phase. The second negative charge on the phosphate
ion is free to attract positively charged basic amino acids, retaining the peptides
analytes. Exchanging TEAP with Na-MePO4 as the buffer salt enables negative
amino acids to interact directly with the stationary phase, because methylphos-
phonate only has a single negative charge. Single phosphorylated peptides could
be isocratically separated from non-Â�phosphorylated peptides and eluted using 70%
ACN, 20â•›m M Na-MePO4 at pH 2.0. Multiply phosphorylated peptides, however,
required a gradient of increasing salt (TEAP) and decreasing ACN concentration
to elute. Hereby, peptides varying in phosphorylation number could be separated
in discrete fractions.
Ytterberg et al. demonstrated the utility of HILIC and ERLIC in a micro-column
setup by packing PolyHydroxyEthyl A and PolyWAX LP material in pipette tips.79
In this work, samples from standard proteins as well as a digest of saliva proteins

were selectively enriched for phosphopeptides. These were retained efficiently on

the micro-columns using standard HILIC buffers. The PolyWAX LP material also
used in Alpert’s work showed the highest potential for phosphopeptide fractionation.
The interaction between phosphopeptides and the resin could be gradually abro-
gated using decreasing pH, followed by decreasing ACN concentration, and finally
increasing salt concentration. In this micro-column setup, Ytterberg et al. used step-
wise elution to elute peptides containing different numbers of phosphorylations in
separate fractions.
23.4.5 Multiphosphopeptide Enrichment Prior to HILIC Fractionation

The work of McNulty and Annan showed the high potential of HILIC as the first
dimension in an off-line 2D-LC-MS setup, but also resulted in the identification
of a low number of multiply phosphorylated peptides. Therefore, the evaluation
of combining HILIC with a preceding multiphosphopeptide enrichment step was
initiated. Such an enrichment technique, sequential elution from IMAC (SIMAC)
has previously been published.80 This technique is based on the observation that
mainly monophosphorylated peptides are eluted from the IMAC resin under acidic
conditions, while multiply phosphorylated peptides are retained and first eluted
when using highly basic buffers. Being able to selectively separate multiphosphory-
lated from monophosphorylated peptides allows for separate sample preparation
and mass spectrometric analysis for the two types of samples. This is advanta-
geous because the multiphosphorylated species are particularly difficult to ana-
lyze. Adsorption to surfaces during sample handling,81 poor ionization efficiencies,
or limited peptide fragmentation in CID due to multiple neutral losses of H3PO4
decreases the number of identifications of these peptides. By reducing the num-
ber of handling steps and analyzing the sample with LC-MS setups specifically
optimized for multiply phosphorylated peptides (e.g., multistage activation82), more
identifications can be obtained.80
HILIC combined with an initial multiphosphopeptide enrichment step generating
two fractions—one containing mainly multiply phosphorylated peptides and one
containing non- and monophosphorylated peptides—was therefore very recently
investigated.83 The HILIC method was very similar to McNulty and Annan’s except
that TiO2 was used for enrichment of phosphopeptides from the HILIC fractions
instead of IMAC. Furthermore, an LTQ-Orbitrap XL capable of performing mul-
tistage activation was used for the LC-MS analysis. This method was applied to a
sample from epidermal growth factor stimulated HeLa cells, which had been sub-
jected to stable isotope labeling with amino acids in cell culture (SILAC).84 The
cells were lysed and the proteins were fractionated into membrane associated and
soluble prior to proteolysis and subsequent peptide fractionation by SIMAC and
HILIC. From less than 500â•›μg of total protein, more than 4600 unique phosphopep-
tides were identified. With 21% of the peptides having two or more phosphoryla-
tions, the recovery of multiple phosphorylated peptides was significantly improved.
These results, along with the above-mentioned studies, emphasize the considerable
potential for HILIC as an alternative to SCX as a prefractionation step in large-scale
phosphoproteomics.

23.5 Future Prospects

The proteomic field has a strong focus on PTM characterization, because protein
modification involves modulation of many protein functions. Therefore, a consider-
able effort has been put into expanding the classical proteomics strategies to enable
PTM mapping, the so-called modification-specific proteomics. Taking the comple-
mentary nature of HILIC compared to standard chromatographic techniques into
account, as well as its MS compatibility, it is not a surprise that HILIC has gained
popularity in proteomics. HILIC is especially suitable for separation and enrich-
ment of hydrophilic PTMs such as glycosylation and phosphorylation. The ability to
retain peptides containing these hydrophilic modifications is a very attractive feature
of HILIC and it has greatly expanded the list of biomolecules that can be analyzed
using MS. However, it seems that the full potential of HILIC is not yet exploited and
significant improvements must still be obtained before the glycopeptides and phos-
phopeptides can be routinely analyzed by HILIC-MS. In particular, it is important
to establish the retention mechanisms of different types of PTMs (i.e., glycopep-
tides and phosphopeptides) on different HILIC phases and the influence of various
solvents and additives. Understanding the binding mechanisms, the selectivity of
HILIC can be optimized by designing new stationary phases and/or by using altered
mobile phase conditions. For enrichment purposes, this would result in fractions
with higher concentrations of the desired PTM proteins or modified peptides, and for
chromatographic purposes, this knowledge would allow prediction of the conditions
for optimal separation. In addition, full implementation of HILIC into the proteomic
field will require development of methods with the potential for automatization/high
throughput as well as high sensitivity, reproducibility, and robustness. It is evident
from the data already available that HILIC is a worthy complement to RPLC. Thus,
it is expected that HILIC gradually will move into the proteomic field as a new and
complementary technique in the separation tool-box that will be used primarily for
the analysis of hydrophilic PTMs like glycosylation and phosphorylation, but also
for unmodified hydrophilic peptides that are difficult to analyze with the current
existing techniques.
Acknowledgment
We would like to thank Tina Nielsen and Jimmy Ytterberg for contribution of data.
References
1. Apweiler, R.; Hermjakob, H.; Sharon, N. Biochim. Biophys. Acta 1999, 1473, 4–8.
2. Varki, A. Glycobiology 1993, 3, 97–130.
3. Manning, G.; Plowman, G. D.; Hunter, T.; Sudarsanam, S. Trends Biochem. Sci. 2002,
27, 514–520.
4. Manning, G.; Whyte, D. B.; Martinez, R.; Hunter, T.; Sudarsanam, S. Science 2002, 298,
1912–1934.
5. Hubbard, M. J.; Cohen, P. Trends Biochem. Sci. 1993, 18, 172–177.
6. Hunter, T.; Sefton, B. M. Proc. Natl. Acad. Sci. U.S.A. 1980, 77, 1311–1315.

7. Hunter, T. Cell 2000, 100, 113–127.

8. Alpert, A. J. J. Chromatogr. 1990, 499, 177–196.
9. Hao, Z.; Xiao, B.; Weng, N. J. Sep. Sci. 2008, 31, 1449–1464.
10. Hirabayashi, J. Glycoconj. J. 2004, 21, 35–40.
11. Bunkenborg, J.; Pilch, B. J.; Podtelejnikov, A. V.; Wisniewski, J. R. Proteomics 2004,
4, 454–465.
12. Packer, N. H.; Lawson, M. A.; Jardine, D. R.; Redmond, J. W. Glycoconj. J. 1998,
15, 737–747.
13. Larsen, M. R.; Hojrup, P.; Roepstorff, P. Mol. Cell. Proteomics 2005, 4, 107–119.
14. Larsen, M. R.; Jensen, S. S.; Jakobsen, L. A.; Heegaard, N. H. H. Mol. Cell. Proteomics
2007, 6, 1778–1787.
15. Zhang, H.; Li, X. J.; Martin, D. B.; Aebersold, R. Nat. Biotechnol. 2003, 21,
660–666.
16. Sparbier, K.; Koch, S.; Kessler, I.; Wenzel, T.; Kostrzewa, M. J. Biomol. Tech. 2005,
16, 407–413.
17. Wuhrer, M.; Catalina, M. I.; Deelder, A. M.; Hokke, C. H. J. Chromatogr. B Anal.
Technol. Biomed. Life Sci. 2007, 849, 115–128.
18. Liu, X.; Ma, L.; Li, J. J. Anal. Lett. 2008, 41, 268–277.
19. Guile, G. R.; Harvey, D. J.; O’Donnell, N.; Powell, A. K.; Hunter, A. P.; Zamze, S.;
Fernandes, D. L.; Dwek, R. A.; Wing, D. R. Eur. J. Biochem. 1998, 258, 623–656.
20. Thaysen-Andersen, M.; Thogersen, I. B.; Nielsen, H. J.; Lademann, U.; Brunner, N.;
Enghild, J. J.; Hojrup, P. Mol. Cell. Proteomics 2007, 6, 638–647.
21. Thaysen-Andersen, M.; Thogersen, I. B.; Lademann, U.; Offenberg, H.; Giessing, A. M.;
Enghild, J. J.; Nielsen, H. J.; Brunner, N.; Hojrup, P. Biochim. Biophys. Acta 2008,
1784, 455–463.
22. Omaetxebarria, M. J.; Hagglund, P.; Elortza, F.; Hooper, N. M.; Arizmendi, J. M.;
Jensen, O. N. Anal. Chem. 2006, 78, 3335–3341.
23. Yu, Y. Q.; Gilar, M.; Kaska, J.; Gebler, J. C. Rapid Commun. Mass Spectrom. 2005, 19,
2331–2336.
24. Yu, Y. Q.; Fournier, J.; Gilar, M.; Gebler, J. C. Anal. Chem. 2007, 79, 1731–1738.
25. Wada, Y.; Tajiri, M.; Yoshida, S. Anal. Chem. 2004, 76, 6560–6565.
26. Tajiri, M.; Yoshida, S.; Wada, Y. Glycobiology 2005, 15, 1332–1340.
27. Anumula, K. R.; Dhume, S. T. Glycobiology 1998, 8, 685–694.
28. Ciucanu, I.; Costello, C. E. J. Am. Chem. Soc. 2003, 125, 16213–16219.
29. Thaysen-Andersen, M.; Mysling, S.; Hojrup, P. Anal. Chem. 2009, 81, 3933–3943.
30. Hagglund, P.; Bunkenborg, J.; Elortza, F.; Jensen, O. N.; Roepstorff, P. J. Proteome Res.
2004, 3, 556–566.
31. Hagglund, P.; Matthiesen, R.; Elortza, F.; Hojrup, P.; Roepstorff, P.; Jensen, O. N.;
Bunkenborg, J. J. Proteome Res. 2007, 6, 3021–3031.
32. Picariello, G.; Ferranti, P.; Mamone, G.; Roepstorff, P.; Addeo, F. Proteomics 2008, 8,
3833–3847.
33. Gonzalez, J.; Takao, T.; Hori, H.; Besada, V.; Rodriguez, R.; Padron, G.; Shimonishi, Y.
Anal. Biochem. 1992, 205, 151–158.
34. Calvano, C. D.; Zambonin, C. G.; Jensen, O. N. J. Proteomics 2008, 71, 304–317.
35. Hemstrom, P.; Irgum, K. J. Sep. Sci. 2006, 29, 1784–1821.
36. Takegawa, Y.; Deguchi, K.; Ito, H.; Keira, T.; Nakagawa, H.; Nishimura, S. I. J. Sep. Sci.
2006, 29, 2533–2540.
37. Anumula, K. R. Anal. Biochem. 2006, 350, 1–23.
38. Karlsson, N. G.; Hansson, G. C. Anal. Biochem. 1995, 224, 538–541.
39. Wang, W. T.; Erlansson, K.; Lindh, F.; Lundgren, T.; Zopf, D. Anal. Biochem. 1990, 190,
182–187.

40. Rudd, P. M.; Colominas, C.; Royle, L.; Murphy, N.; Hart, E.; Merry, A. H.; Hebestreit, H. F.;
Dwek, R. A. Proteomics 2001, 1, 285–294.
41. Wuhrer, M.; Koeleman, C. A.; Deelder, A. M.; Hokke, C. H. FEBS J. 2006, 273,
347–361.
42. Maslen, S.; Sadowski, P.; Adam, A.; Lilley, K.; Stephens, E. Anal. Chem. 2006, 78,
8491–8498.
43. Maslen, S. L.; Goubet, F.; Adam, A.; Dupree, P.; Stephens, E. Carbohydr. Res. 2007,
342, 724–735.
44. Zhang, J.; Wang, D. I. J. Chromatogr. B Biomed. Sci. Appl. 1998, 712, 73–82.
45. Nguyen, H. P.; Schug, K. A. J. Sep. Sci. 2008, 31, 1465–1480.
46. Wuhrer, M.; Koeleman, C. A.; Deelder, A. M.; Hokke, C. H. Anal. Chem. 2004, 76,
833–838.
47. Charlwood, J.; Birrell, H.; Bouvier, E. S.; Langridge, J.; Camilleri, P. Anal. Chem. 2000,
72, 1469–1474.
48. Wuhrer, M.; Koeleman, C. A. M.; Hokke, C. H.; Deelder, A. M. Int. J. Mass Spectrom.
2004, 232, 51–57.
49. Wuhrer, M.; de Boer, A. R.; Deelder, A. M. Mass Spectrom. Rev. 2009, 28, 192–206.
50. Naidong, W. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 2003, 796, 209–224.
51. Takegawa, Y.; Deguchi, K.; Keira, T.; Ito, H.; Nakagawa, H.; Nishimura, S. J. Chromatogr.
A 2006, 1113, 177–181.
52. Staples, G. O.; Bowman, M. J.; Costello, C. E.; Hitchcock, A. M.; Lau, J. M.;
Leymarie, N.; Miller, C.; Naimy, H.; Shi, X.; Zaia, J. Proteomics 2009, 9, 686–695.
53. Takegawa, Y.; Ito, H.; Keira, T.; Deguchi, K.; Nakagawa, H.; Nishimura, S. J. Sep. Sci.
2008, 31, 1585–1593.
54. Wuhrer, M.; Koeleman, C. A.; Hokke, C. H.; Deelder, A. M. Anal. Chem. 2005, 77,
886–894.
55. Temporini, C.; Perani, E.; Calleri, E.; Dolcini, L.; Lubda, D.; Caccialanza, G.; Massolini,
G. Anal. Chem. 2007, 79, 355–363.
56. Gropengiesser, J.; Varadarajan, B. T.; Stephanowitz, H.; Krause, E. J. Mass Spectrom.
2009, 44, 821–831.
57. Andersson, L.; Porath, J. Anal. Biochem. 1986, 154, 250–254.
58. Neville, D. C.; Rozanas, C. R.; Price, E. M.; Gruis, D. B.; Verkman, A. S.; Townsend, R. R.
Protein Sci. 1997, 6, 2436–2445.
59. Gruhler, A.; Olsen, J. V.; Mohammed, S.; Mortensen, P.; Faergeman, N. J.; Mann, M.;
Jensen, O. N. Mol. Cell. Proteomics 2005, 4, 310–327.
60. Villen, J.; Beausoleil, S. A.; Gerber, S. A.; Gygi, S. P. Proc. Natl. Acad. Sci. U. S. A.
2007, 104, 1488–1493.
61. Washburn, M. P.; Wolters, D.; Yates, J. R., 3rd. Nat. Biotechnol. 2001, 19, 242–247.
62. Beausoleil, S. A.; Jedrychowski, M.; Schwartz, D.; Elias, J. E.; Villen, J.; Li, J.; Cohn, M. A.;
Cantley, L. C.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12130–12135.
63. Pinkse, M. W.; Uitto, P. M.; Hilhorst, M. J.; Ooms, B.; Heck, A. J. Anal. Chem. 2004,
76, 3935–3943.
64. Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.; Roepstorff, P.; Jorgensen, T. J. Mol.
Cell. Proteomics 2005, 4, 873–886.
65. Olsen, J. V.; Blagoev, B.; Gnad, F.; Macek, B.; Kumar, C.; Mortensen, P.; Mann, M. Cell
2006, 127, 635–648.
66. Rush, J.; Moritz, A.; Lee, K. A.; Guo, A.; Goss, V. L.; Spek, E. J.; Zhang, H.; Zha, X. M.;
Polakiewicz, R. D.; Comb, M. J. Nat. Biotechnol. 2005, 23, 94–101.
67. Schmelzle, K.; Kane, S.; Gridley, S.; Lienhard, G. E.; White, F. M. Diabetes 2006, 55,
2171–2179.
68. Zhang, X.; Ye, J.; Jensen, O. N.; Roepstorff, P. Mol. Cell. Proteomics 2007, 6, 2032–2042.
69. Zhou, H.; Watts, J. D.; Aebersold, R. Nat. Biotechnol. 2001, 19, 375–378.

70. Oda, Y.; Nagasu, T.; Chait, B. T. Nat. Biotechnol. 2001, 19, 379–382.
71. Adamczyk, M.; Gebler, J. C.; Wu, J. Rapid Commun. Mass Spectrom. 2001, 15,
1481–1488.
72. Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. Anal. Chem. 2005, 77, 6426–6434.
73. Boutin, J. A.; Ernould, A. P.; Ferry, G.; Genton, A.; Alpert, A. J. J. Chromatogr. 1992,
583, 137–143.
74. Lindner, H.; Sarg, B.; Helliger, W. J. Chromatogr. A 1997, 782, 55–62.
75. McNulty, D. E.; Annan, R. S. Mol. Cell. Proteomics 2008, 7, 971–980.
76. Albuquerque, C. P.; Smolka, M. B.; Payne, S. H.; Bafna, V.; Eng, J.; Zhou, H. Mol. Cell.
Proteomics 2008, 7, 1389–1396.
77. Boersema, P. J.; Divecha, N.; Heck, A. J.; Mohammed, S. J. Proteome Res. 2007, 6,
937–946.
78. Alpert, A. J. Anal. Chem. 2008, 80, 62–76.
79. Ytterberg, J. A.; Ogorzalek-Loo, R. R.; Boontheung, P.; Loo, J. A. Poster Presentation
at the 55th ASMS Conference on Mass Spectrometry and Allied Topics June 3–7, 2007,
Indianapolis, IN.
80. Thingholm, T. E.; Jensen, O. N.; Robinson, P. J.; Larsen, M. R. Mol. Cell. Proteomics
2008, 7, 661–671.
81. Jensen, S. S.; Larsen, M. R. Rapid Commun. Mass Spectrom. 2007, 21, 3635–3645.
82. Schroeder, M. J.; Shabanowitz, J.; Schwartz, J. C.; Hunt, D. F.; Coon, J. J. Anal. Chem.
2004, 76, 3590–3598.
83. Engholm-Keller, K.; Jensen, S. S.; Larsen, M. R. Poster Presentation at the 57th
ASMS Conference on Mass Spectrometry and Allied Topics May 31–June 4, 2009,
Philadelphia, PA.
84. Ong, S. E.; Blagoev, B.; Kratchmarova, I.; Kristensen, D. B.; Steen, H.; Pandey, A.;
Mann, M. Mol. Cell. Proteomics 2002, 1, 376–386.

Hydrophilic Interaction Liquid Chromatography PDF

Uploaded by

Copyright:

Available Formats

Hydrophilic Interaction Liquid Chromatography PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hydrophilic Interaction Liquid Chromatography PDF

Uploaded by

Copyright:

Available Formats

© 2011 by Taylor and Francis Group, LLC

CHROMATOGRAPHIC SCIENCE SERIES

A Series of Textbooks and Reference Books

1. Dynamics of Chromatography: Principles and Theory,

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

CRC Press is an imprint of the

© 2011 by Taylor and Francis Group, LLC

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

International Standard Book Number: 978-1-4398-0753-8 (Hardback)

© 2011 by Taylor and Francis Group, LLC

Chapter 1 Aqueous Normal-Phase Chromatography: The Bridge between

Chapter 2 Application of the Calculated Physicochemical Property, Log D,

Chapter 3 HILIC Behavior of Reversed-Phase/Ion-Exchange Mixed-Mode

Chapter 4 Mechanisms Controlling the Separation and Retention of

Chapter 5 Applications of Hydrophilic Interaction Chromatography

Chapter 6 Hydrophilic Interaction Liquid Chromatography–Mass

© 2011 by Taylor and Francis Group, LLC

Chapter 7 Application of HILIC for Polar Environmental Contaminants

Chapter 8 Hydrophilic Interaction Chromatography–Mass Spectrometry:

Chapter 9 Clinical Applications of Hydrophilic Interaction Liquid

Chapter 10 Dimethylarginine Analysis by HILIC–MS/ MS and Its

Chapter 11 Polar Functional Groups for HILIC Method..................................... 225

Chapter 12 Analysis of Pharmaceutical Impurities Using Hydrophilic

Chapter 13 Fast In-Process Method for the Determination of Ioversol

Chapter 14 Retention Behavior of Hydrazine Derivatives in HILIC Mode........309

Chapter 15 HILIC Retention Behavior and Method Development for Highly

© 2011 by Taylor and Francis Group, LLC

Chapter 16 Retention of Polar Acidic and Basic Drugs by HILIC...................... 373

Chapter 17 Retention and Selectivity of Polar Stationary Phases for

Chapter 18 HILIC-MS/MS for the Determination of Polar Bioactive

Chapter 19 Method Development and Analysis of Mono- and

Chapter 20 Hydrophilic Interaction Liquid Chromatography–Based

Chapter 21 Development and Application of Methods for Separation

Chapter 22 Hydrophilic Interaction Liquid Chromatography in the

Chapter 23 Analysis of Protein Glycosylation and Phosphorylation Using

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

Dr. Weixuan He is currently an associate director, Product and Process Development,

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

M-Concepción Aristoy Carmela Dell’Aversano

© 2011 by Taylor and Francis Group, LLC

Aleida S. Hernández-Cázares Sandra L. Kuchta

© 2011 by Taylor and Francis Group, LLC

Joanne E. Nettleship Kevin A. Schug

© 2011 by Taylor and Francis Group, LLC

Alexander L.N. van Nuijs Akio Yanagida

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

1.1 F undamental Features of Aqueous

© 2011 by Taylor and Francis Group, LLC

1.2 Modes of ANP

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

© 2011 by Taylor and Francis Group, LLC

1.1 F undamental Features of Aqueous