Protein Glycosylation

Provided for non-commercial research and educational use. Not for reproduction, distribution or commercial use. This article was originally published in Comprehensive Biotechnology, Second Edition, published by Elsevier, and the attached copy is provided by Elsevier for the author's benefit and for the benefit of the author's institution, for non-commercial research and educational use including without limitation use in instruction at your institution, sending it to specific colleagues who you know, and providing a copy to your institution's administrator. All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing copies or access, or posting on open internet sites, your personal or institution's website or repository, are prohibited. For exceptions, permission may be sought for such use through Elsevier's permissions site at: http://www.elsevier.com/locate/permissionusematerial Kattla JJ, Struwe WB, Doherty M, Adamczyk B, Saldova R, Rudd PM, and Campbell MP (2011) Biologics | Protein Glycosylation. In: Murray Moo-Young (ed.), Comprehensive Biotechnology, Second Edition, volume 3, pp. 467–486. Elsevier. © 2011 Elsevier B.V. All rights reserved. Author's personal copy 3.41 Protein Glycosylation JJ Kattla, WB Struwe, M Doherty, B Adamczyk, R Saldova, and PM Rudd, Dublin-Oxford Glycobiology Laboratory, The National Institute for Bioprocessing Research and Training, Dublin, Ireland MP Campbell, Dept. of Chemistry and Biomolecular Sciences, Macquarie University, Sydney, NSW, Australia © 2011 Elsevier B.V. All rights reserved. 3.41.1 3.41.1.1 3.41.2 3.41.2.1 3.41.2.2 3.41.3 3.41.3.1 3.41.3.2 3.41.3.2.1 3.41.3.2.2 3.41.3.3 3.41.3.4 3.41.3.4.1 3.41.3.4.2 3.41.3.4.3 3.41.3.4.4 3.41.3.5 3.41.3.5.1 3.41.3.5.2 3.41.3.6 3.41.3.7 3.41.3.8 3.41.3.9 3.41.4 3.41.4.1 3.41.4.2 3.41.5 References Introduction Glycoanalytical Methods Analysis of Intact Glycoproteins and Glycopeptides CZE Coupled with MS Lectin Array Technology Analysis of Free Glycans Sample Preparation Glycan Release Hydrazinolysis Enzymatic release Glycan Labeling and Purification High-Performance Liquid Chromatography Anion exchange chromatography and separation by charge Hydrophilic interaction liquid chromatography Reversed-phase HPLC High-pH anion exchange chromatography Glycan Sequencing Using Exoglycosidases Glycan analysis Applications of oligosaccharide sequencing in glycan analysis of therapeutic glycoproteins High-Throughput Glycan Analysis Based on a Robotics Platform Data Analysis and Glycobioinformatics Mass Spectrometry Capillary Gel Electrophoresis with Laser-Induced Fluorescence Analysis of Monosaccharides Analysis of Neutral Monosaccharides Analysis of Sialic Acids Glycan Analysis Design for Therapeutic Glycoproteins Glossary biologic A protein-based therapeutic such as vaccine, somatic cell, or recombinant therapeutic protein made or derived from a living organism that is used in diagnosis, treatment, and prevention of disease. biopharmaceutical glycosylation Refers to the type and distribution of glycans of therapeutic glycoproteins. glycan analysis The process of isolating and characterizing complex carbohydrate structures, sequence, and linkage from a variety of sources. glycan fingerprinting The analysis of free glycans used to elucidate N- or O-linked glycan structures attached to the 468 468 469 469 469 469 470 470 471 471 472 474 474 475 475 475 477 477 478 479 480 481 482 482 482 482 483 484 protein backbone of glycoproteins, including their oligosaccharide sequence and linkage. It is also referred to as analysis of released glycans, oligosaccharide profiling/ sequencing, or mapping. glycan macroheterogeneity A term used to describe glycoprotein variants resulting from variations in glycosylation site occupancy. glycan microheterogeneity A term used to describe glycoprotein variants as a result of alternative oligosaccharide sequences at individual N- or O-linked sites of glycosylation of the polypeptide chain. glycoforms Glycosylated variants of glycoproteins that differ only in oligosaccharide sequence and linkage. 467 Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 468 Biologics 3.41.1 Introduction Many biological drugs such as monoclonal antibodies and therapeutic proteins are posttranslationally modified during their synthesis. Glycosylation is the most common type of posttranslational modifications (PTMs) and almost 70% of human proteins are glycosylated [1]. There are different types of protein glycosylation, including (1) N-glycans (attached to nitrogen in Asn or Arg side chains); (2) O-linked glycans (attached to hydroxyl oxygen of Ser, Thr, hydroxyproline side chains); (3) phosphoserine glycans (attached to phosphate of a phosphoserine); (4) C-linked glycans (rare form of glycans added to a carbon on Trp side chain); (5) glypiated glycans (addition of a glycosylphosphatidylinositol (GPI) anchor that links proteins to lipids through glycan linkages); and (6) other O-linked glycans found in cytoplasmic/neucleoplasmic proteins, such as O-GlcNAc, O-fucose, and O-mannose [2]. N-glycans can be further classified into four main groups based on the structure of the glycan residues added to the trimannosyl core (oligomannose, complex, hybrid, and poly-N-acetyllactosamine glycans) [3–5]. Glycosylation is dependent upon a series of glycosyltransferases and glycosidases that act upon a consecutively processed oligosaccharide chain as a glycoprotein passes through the endoplasmic reticulum (ER) and Golgi apparatus. Given that not all the glycosyltransferases may act on every oligosaccharide, a mixture of glycoforms with different glycosylation patterns is generated for most proteins. Glycoproteins can have multiple glycosylation sites and exhibit variable site occupancy. Incorporation of carbohydrate moieties (glycans) into the therapeutic protein backbone not only affects physiochemical properties, such as stability and solubility but also plays a pivotal role in cellular location, biological function, receptor interaction, transport, immunogenicity, circulating half-life, and the safety and efficacy of biologics [2, 6–13]. Glycans are the most abundant, complex, and diverse class of molecules found in biological systems and glycoproteins are almost always heterogeneous in nature. Macroheterogeneity results from variations in glycosylation site occupancy, while microheterogeneity is the term used to describe alternative oligosaccharide sequences at a single amino acid position of the polypeptide chain. The development of recombinant technology in the 1990s led to the large-scale manufacture of biological drugs in mammalian cell culture systems. Recombinant biological drugs also exhibited significant structural divergence in glycan pattern. The precise glycans that are present on the secreted proteins depend on the cell type, density, and age of the cells. In addition, factors in the manufacturing process such as media, pH, temperature, concentration of nutrients, and the type of fermentation process can also cause appreciable alterations to the glycan structures [14]. Therefore, it is vital to identify the glycosylation pattern of recombinant glycoproteins at all stages of bioproduction including clone screening, clone selection, optimization of cell culture media, fermentation, purity testing, and finally quality control for product release. In view of the fact that different glycoforms have diverse biological functions, the ability to monitor and control glycosylation is important for maintaining the safety and efficacy of a therapeutic protein. These links between the complex nature of the glycans and their activity have raised serious concerns within the drug regulatory authorities across the world. Current regulatory directives delineate increasingly stringent guidelines on the glycan analysis of biologics. By law, it is now essential to determine (1) carbohydrate content (neutral sugars, amino sugars, and sialic acids); (2) extent of mannosylation, galactosylation, fucosylation, and sialylation; (3) the ratio of G0, G1, and G2 galactosylated glycans; (4) presence or absence of glycosylation in additional glycosylation sites; and (5) extensive analysis of glycan structure including the antennary profile (www.emea.europa.eu). Additionally, the Food and Drug Administration (FDA)’s process analytical technologies (PAT) and quality-by-design (QbD) proposals encourage biomanufacturers to screen and analyze glycosylation of biologics at all stages of bioproduction (www.fda.gov). When compared to the analytical methods in the field of genomics and proteomics, analysis of protein glycosylation is invariably more complicated and lacks a universal and robust method to unravel the complete structure and quantity of glycans. 3.41.1.1 Glycoanalytical Methods For nearly all major therapeutic glycoproteins, a key feature is the identification of glycosylation critical quality attributes (GCQAs). Considering the heterogeneous nature of glycoproteins, different experimental approaches are employed to analyze different features such as sialylation, glycoform analysis, detailed structure of carbohydrate chains, oligosaccharide pattern (antennary profile), and glycosylation site(s). Glycan analysis can be generally tackled in four different ways: (1) analysis of glycans on intact glycoproteins (while still attached to the protein), (2) glycopeptides (glycans attached to the peptides after protease digestion), (3) free glycans (oligosaccharide released from the proteins), and (4) monosaccharide analysis (after being broken down into their component monosaccharides). There are several glycoanalytical methods employed to elucidate the glycan structure of biological drugs, including high-performance liquid chromatography (HPLC), mass spectrometry (MS), high-pH anion exchange chromatography coupled to analysis with a pulsed amperometric detector (HPAE-PAD), capillary electrophoresis (CE), nuclear magnetic resonance (NMR), lectin arrays, and gas–liquid chromatography (GC). However, there is no universal method available to date that can characterize the complete glycan structure with their oligosaccharide sequence and linkage information. It is therefore essential to apply several orthogonal methods to measure individual parameters such as glycosylation site analysis, oligosaccharide sequence, and monosaccharide content of a therapeutic glycoprotein. This article reviews various glycoanalytical approaches and strategies for examining the overall glycan structure of therapeutic glycoproteins. Its aim is to provide information on current state-of-the-art techniques from analysis of intact glycans, glycopeptides, release of glycans, and labeling of glycans to ultimately elucidate the complete glycan structure. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 469 3.41.2 Analysis of Intact Glycoproteins and Glycopeptides Analysis of intact glycoproteins and glycopeptides is used to elucidate the glycan structures at individual sites. There are several methods including capillary zone electrophoresis (CZE), matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS), electrospray ionization mass spectrometry (ESI-MS), lectin arrays, and NMR. Furthermore, ion mobility MS enables the rapid separation and characterization of molecules by their size, mass, and shape illustrated as drift time in the gas phase and chargeto-mass ratio (m/z) in the mass analyzer and represents an innovative technique for glycoprotein analysis. 3.41.2.1 CZE Coupled with MS CZE coupled with MS (CZE-MS) is used for both glycoprotein and glycopeptide analyses. Typically, the separation mechanism is based on differences in the m/z ratio of the analytes. In general, glycoproteins or glycopeptides in aqueous solution are injected electrochemically and separations are carried out at 10 kV using 50-cm fused-silica capillaries and buffers compatible with MS detection. The separated glycans are monitored by an ultraviolet (UV) detector at 214 nm. Both derivatized and nonderivatized glycoproteins can be analyzed with MALDI and ESI in both positive and negative MS modes. CZE-MS is an efficient tool that offers faster analysis for glycoprotein macroheterogeneity (CZE-ESI-MS) [15] and characterization of oligosaccharide structure (CZE-MS-MS) [16]. These methods resolve glycoproteins with high mass accuracy and resolution, identifying even minor modifications to glycan structures such as differences in sialic acid content (sialoforms) and other PTMs commonly present, such as acetylation, sulfation, phosphorylation, or deamination [17]. Though CZE is a desirable method for intact glycoprotein and glycopeptide analysis, detailed information on individual glycans and their site specificities is hard to obtain. In addition, the use of fused-silica capillaries causes adsorption of proteins to negatively charged silanol groups of the capillary wall. The addition of zwitter ions to the protocol suppresses the protein adsorption and makes this disadvantage less of an issue [18]. In common with most glycan analysis, the interpretation of CZE-MS data tends to be time consuming and requires considerable expertise. Nevertheless, glycoform analysis by CZE-MS can provide the total and collective information on monosaccharide composition and other modifications that is crucial for monitoring the consistency and stability of glycoproteins. CZE-MS is now routinely used by industry at all stages of biopharmaceutical production including downstream processing, purity testing, batch-to-batch consistency, stability testing, and glycosylation site analysis. Furthermore, CZE is also used to compare the glycosylation patterns of recombinant and natural proteins. 3.41.2.2 Lectin Array Technology Animal lectins are a family of glycan-recognizing proteins that are classified into different groups based on their specificity for individual monosaccharide units, for example, Galanthus nivalis agglutinin (GNA) lectin specifically binds terminal α(1–3)-linked mannose [2]. There is significant interest in the use of lectins in array formats to analyze intact therapeutic glycoproteins [19–21]. Typically, a set of 20–30 different types of lectins with different specificities are printed on either aldehyde- or epoxide-derivatized glass slides to produce an array. In general, the glycoproteins applied on the lectin arrays are detected with a labeled probe to produce a fingerprint using an array scanner. The fingerprints are further deconvoluted to provide information on glycan structures using computer-assisted data analysis algorithms. Lectin arrays can be performed on small volumes of crude samples from a bioreactor such as cell culture media, in a relatively short period of time (approximately 4–5 h) [22]. This technique can be used to monitor the glycosylation pattern of therapeutic glycoproteins at all stages of the biomanufacturing process. For example, it is possible to monitor the ratio of G0, G1, and G2 glycans present in recombinant monoclonal antibodies at different time points throughout a typical fermentation run. Lectin arrays are also used in the identification of desired glycosylation patterns during clone screening and selection. In addition, the use of lectin arrays during process development allows optimization of media choice/effect on glycosylation of therapeutics. This method is used to screen batch-to-batch consistency with regard to glycosylation patterns, in determination of batch feeding schedules and harvest times, and to allow termination of batches due to aberrant glycosylation. Furthermore, the use of lectin biosensors enables rapid detection of label-free glycans based on glycan–lectin interactions [23]. This involves interaction of encapsulated glyconanoparticles with lectins such as peanut agglutinin (PNA) or Sambucus nigra agglutinin (SNA) immobilized to printed circuit board (PCB). This technique allows differentiation of glycan microheterogeneity [23]. Despite several advantages, lectin arrays do not provide detailed information on (1) glycoforms, (2) sialic acid speciation, and (3) oligosaccharide sequence and linkage. Moreover, the data are semiquantitative and have to take into account that lectins have different affinities and may not be stoichiometric, given that lectins are large in comparison with the spaces between terminal glycan residues on a multiantennary glycan. 3.41.3 Analysis of Free Glycans Analysis of free glycans (also referred to as analysis of released glycans, oligosaccharide profiling/sequencing, mapping, or glycan fingerprinting) is used to elucidate N- and O-linked glycan structures attached to glycoproteins, including oligosaccharide sequence and linkage. The initial step is to release the glycans in high yield, nonselectively and unmodified, from the molecules to which they are attached. Following release, glycans can be either derivatized (labeled) for HPLC, CE, or MS analysis or underivatized (unlabeled) for MS analysis. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 470 Biologics 3.41.3.1 Sample Preparation Sample preparation is a very important aspect of glycan analysis. Glycoprotein samples can be prepared and deglycosylated in two ways: (1) in-solution digest and (2) in-gel or membrane digestion (protein of interest cut from a sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gels or protein sample immobilized to a gel block or a membrane (see Figure 1). In-solution deglycosylation requires greater amounts of enzyme when compared to in-gel digestion. There are many advantages to gel methods such as (1) the protein can be purified and separated from contaminants with minimal sample loss, (2) higher recovery of glycans is possible when the sample is immobilized with less enzyme, and (3) the protein remains in the gel after removal of glycans, and following trypsin digestion, can be used in peptide analysis by MS. The selection of the sample preparation method depends on the glycoprotein and the aim of the study. In the case of heavy chain glycan analysis, the immunoglobulin G (IgG) sample can be denatured, reduced, and separated by SDS-PAGE. IgG can be resolved into heavy and light chains. On the other hand, IgG can be purified using columns or solid phase extraction (SPE) plates derivatized with protein A or G to obtain pure sample. It is also important to note that protein A does not bind to IgG3. However, by this method the complete glycan analysis from both Fab and Fc glycosylation is possible. 3.41.3.2 Glycan Release The most common glycan release methods are based on chemical or enzymatic deglycosylation strategies. Therapeutic glycoprotein Tryptic digest Glycopeptides in solution Removing N-glycans from glycoproteins (in-solution PNGase A deglycosylation) Purified protein solution IgG purification using protein A or G columns Protein separation (SDS-PAGE) Removing N-glycans from glycoproteins (in-solution PNGase F deglycosylation) Removal of N-glycans from glycopeptides using C18 columns Proteins immobilized to gel block or PVDF Removing N-glycans from glycoproteins (in-gel PNGase F deglycosylation) Extraction of N-glycans Oligosaccharides Protein Nano-HPLC-nano-ESI-MS MS Unlabeled glycans MALDI-MS N-Glycan profiling by HPAE-PAD Protein sequence confirmation Fluorescent labeling or N-glycans (2-AB, 2-AA, 2-AP, or APTS) Removal of excess dye and desalting N-Glycan profiling by HILIC, RP-HPLC, WAX, HPAE-FD, CGE-LIF Glycan fingerprinting using exoglycosidases Figure 1 Workflow for identification of N-linked free glycans in therapeutic glycoproteins. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 3.41.3.2.1 471 Hydrazinolysis Hydrazinolysis is a chemical method for releasing both O- and N-linked glycans from glycoproteins whereby anhydrous hydrazine (N2H4) is used to generate the free reducing end required for fluorescent tagging. Glycan release involves three steps. First, addition of N2H4 to glycoproteins under controlled conditions of temperature and time (see Figure 2), after which excess hydrazine is evaporated. Hydrazinolysis results in the formation of a de-N-acetylated hydrazone derivative that then converts to the hydrazide form. Second, primary amino groups generated during hydrazinolysis are re-N-acetylated by addition of sodium bicarbonate and acetic anhydride to form β-acetohydrazide derivatives. Third, the acid-labile β-acetohydrazide derivatives are hydrolyzed to produce glycan with reducing ends for labeling [24]. The structures of the labeled glycans can be analyzed by chromatography and/or MS. Hydrazinolysis cleaves glycans effectively but leads to destruction of the protein element [25]. Moreover, hydrazinolysis leads to undesirable side reactions such as peeling. Peeling is chemical degradation of O-linked glycans resulting in the release of sugars (peeled glycans) from their reducing ends [26]. The complexity of the above method and the use of toxic chemicals raise some safety issues; in addition, the N-acetyl and N-glycolyl groups are also cleaved. The glycans can be re-N-acetylated, but then the distinction between acetyl and glycolyl groups is lost [27]. This loss could affect acetyl or glycolyl group and can raise serious concerns over biological and immunogenic properties of analyzed glycoproteins. Other traditional methods used in release of O-linked glycans such as reductive β-elimination also destroy peptide bonds and are accompanied by peeling. 3.41.3.2.2 Enzymatic release A large number of enzymes are available for cleaving various oligosaccharide chains from the protein backbone. The most commonly used enzyme for cleaving all common classes of Asn-linked oligosaccharides is peptide-N-glycosidase F, also known as PNGase F (see Figure 3). PNGase F is an endoglycosidase, specific for cleaving the amide bond between the innermost GlcNAc and the asparagine residue [28]. However, oligosaccharides containing an α(1–3)-linked fucose to the asparagine-linked GlcNAc, commonly found in glycoproteins from plants, Caenorhabditis elegans, and parasitic worms, are resistant to PNGase F [29]. Therapeutic glycoproteins Incubation of lyophilized, salt-free glycoproteins with anhydrous hydrazine Hydrazinolysis (cleavage of glycans) O-Glycans 4–6 h at 60 °C N-Glycans 5–16 h at 95 °C N- and O-Glycans 4 h at 95 °C Remove excess hydrazine by evaporation Reacetylation Recovery of released glycans Re-N-acetylation of released glycans (acetic anhydride with sodium carbonate buffer) Desalting the above mixture on a cation + exchange resin H and purifying the released glycans Fluorescent labeling of N-glycans (2-AB, 2-AA, 2-AP, or APTS) Analysis of glycans N-Glycan profiling by HILIC, RP-HPLC, WAX, HPAE-FD, CGE-LIF Figure 2 Workflow of N- and O-glycan analysis by hydrazinolysis. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 472 Biologics Glycan Protein (a) aa aa PNGase A NH aa Asn Asn X aa PNGase F Ser / Thr aa aa CO aa (b) aa NH aa Asn Asn PNGase A X aa Ser / Thr aa aa (c) CO aa aa NH aa Asn Asn X aa Endo H Ser / Thr aa aa CO Figure 3 Comparison between specificity of common enzymes for releasing oligosaccharides. (a) PNGase F cleaves all asparagine-linked oligosaccharides unless α(1–3) core fucosylated. (b) PNGase A hydrolyzes all types of N-glycan chains from glycopeptides, even those carrying α(1–3) bound core fucose residues present in insect and plant glycoproteins. (c) Endoglycosidase H cleaves between the N-acetylglucosamine (GlcNAc) residues of the chitobiose core of N-linked glycans; the specificity of this enzyme is such that oligomannose and most hybrid types of glycans are cleaved, whereas complex glycans are not released. Furthermore, biphosphorylated glycopeptides are resistant to PNGase F until dephosphorylated [30]. In such situations, N-glycosidase A (PNGase A) can be successfully used on trypsin-digested glycoproteins to release all types of complex glycans [31]. In addition to the above enzymes, there are two other enzymes (endo D and endo H) that cleave the bond between two GlcNAc residues, leaving one GlcNAc attached to the asparagine. Endo D cleaves all types of N-linked glycans, but Endo H selectively cleaves the oligomannose and hybrid-type structures [24, 32]. Finally, there is a group of enzymes that includes the endo-β-N-acetylglucosaminidases F1, F2, and F3 that cleave within the chitobiose core [GlcNAc(β1–4)GlcNAc-Asn]. F1 catalyzes the cleavage of oligomannose and hybrid structures, F2 cleaves oligomannose and biantennary structures, and F3 is specific for bi- and triantennary structures [24]. Regarding O-linked glycans, there are no broad specificity of O-glycanase enzymes currently available; hence, generic cleavage of O-glycans is possible only through chemical methods. 3.41.3.3 Glycan Labeling and Purification Fluorescent tagging of glycans is essential for quantitative glycan analysis using HPLC or CE. There are many labeling methods to choose from, depending on the objectives of analysis. Currently, there is no ideal label for all analytical applications, so careful consideration should be given to selecting the appropriate label to suit the purpose. Small fluorophores conjugated to the reducing end of glycans using reductive amination increase the spectral absorption of glycans, improving the detection of labeled glycans by HPLC and other methods (see Table 1). An appropriate fluorescent label should be non selective and highly efficient, allowing the detection of glycans with high sensitivity (in the femtomolar range) during HPLC and CE. The fluorescent label at the reducing end of glycans must be stable to allow Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 473 quantitation based on the fluorescence signal. There is a 1:1 stoichiometry between the glycan and the label; therefore, the fluorescence signal is directly related to number of moles of labeled glycans. The most important characteristic of the label is nonselectivity so that glycans are detected in their correct molar proportions. 2-Aminobenzamide (2-AB) is the most commonly used fluorophore to label glycans. A classic labeling method involves resuspension of purified and dried glycans with free reducing terminus in 1 mol dm−3 of 2-AB and sodium cyanoborohydride in 30% acetic acid and 70% dimethyl sulfoxide (DMSO) (v/v). The reaction mixture is then incubated at 65 °C for 2 h [33]. 2-AB is a neutral fluorophore and therefore the order of glycan eluted from the HPLC column is directly related to size/ shape, hydrophilicity/hydrophobicity, and the number of sugar residues present in the glycans [31]. 2-AB-labeled glycans are well suited to a variety of glycoanalytical methods including hydrophilic interaction liquid chromatography (HILIC), weak anion exchange (WAX) HPLC, reversed-phase HPLC (RP-HPLC), MALDI-MS, and ESI-MS as it is predictive. On the other hand, negatively charged labels, such as 2-anthranilic acid (2-AA), are less suitable for predictive HPLC methods; in contrast to neutral glycans, it gives a negative incremental value for sialic acids, making the data interpretation complicated unnecessarily. However, 2-AA is preferred for analytical electrophoretic analyses [34]. Other labels commonly used include 2-aminopyridine (2-AP), (3-acetylamino)-6-aminoacridine (AA-Ac) for HPLC analysis, and 8-aminopyrene-1,3,6-trisulfonate (APTS) for profiling by CE [35] (see Table 1). In 2005, Kamoda and colleagues found an alternative method of labeling with 9-fluorenylmethyl chloroformate (Fmoc-Cl) [36]. Fmoc-Cl binds to the reducing end of sialo, asialo, high-mannose, and hybrid-type glycans. In addition, there are fluorescence dyes such as 1,2-diamino-4,5-methylenedioxybenzene (DMB) that can be used to label nonreducing glycans including sialic acids released from glycoproteins [37]. Most of the analytical separations require prior glycan purification to remove excess of labels, reagents, salts, and other contaminants. A variety of cleanup procedures have been developed. The traditional paper chromatography cleanup method has been replaced with new generations of tips and cartridge technologies. Typically, the protocol involves purification of labeled Table 1 Fluorescent labels for glycoprofiling Flourescent tag Structure of labeled product Λexcitation Λemission Separation (nm) (nm) method 2-AB (2-aminobenzamide) 356 450 HPLC, MS 2-AA (anthranilic acid) 330 420 PAGE, HPLC, MS 2-AP (2-aminopyridine) 320 400 HPLC, CE AA-Ac ((3-acetylamino)-6-aminoacridine) 445 525 HPLC, MS APTS (8-aminopyrene-1,3,6-trisulfonate) 488 520 CE DMB (1,2-diamino-4,5-methylenedioxybenzene) 373 448 HPLC Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 474 Biologics glycans using normal-phase polyamide resin in custom-made pipette tips (PhyNexus®) [38]. On the other hand, two types of cartridge technology are used. The first technology is based on hydrophilic interaction, where glycans bind to a hydrophilic matrix of solid-phase extraction (SPE) resin and hydrophobic contaminants are washed off (LudgerCleanTMD1). The purified glycans are then recovered by elution with an aqueous solvent. Alternative technology is based on low-affinity columns, where the glycans pass through the matrix while other reagents are retained (Glyko® GlycoCleanTM R cartridge, Prozyme). Furthermore, it is vital to check that large, complex glycans such as tri- and tetra-antennary glycans are completely eluted and are not retained to paper or purification cartridges. These methods offer faster purification times and increased throughput. 3.41.3.4 High-Performance Liquid Chromatography HPLC coupled with fluorescence detection is a robust and commonly used quantitative glycoanalytical method. Separation using HPLC allows detection and preliminary assignment of specific fluorophore-labeled glycans based on their retention time. A typical HPLC profile of a glycan mixture consists of several glycan peaks containing different sugars, and the percentage area of each glycan peak represents the quantity of the glycans in the peak. A variety of HPLC systems and separation principles can be used to differentiate glycans based on size or charge such as HILIC or normal-phase HPLC, RP-HPLC, WAX, and HPAE-PAD or high-pH anion exchange with fluorescence detection (HPAE-FD). 3.41.3.4.1 Anion exchange chromatography and separation by charge Anion exchange HPLC separates glycans on the basis of charged groups. The separation can be performed under the conditions of either a strong anion exchange (SAX) or WAX chromatography. Typically, the principle of this method involves the application of labeled glycans to an anion exchange column (GlycoSepTM C, Prozyme) and elution with a salt gradient (500 mM ammonium acetate, pH 4.5). Negatively charged glycans such as sialic acids bind to the anion exchanger, while neutral sugars pass through the column. The separated sialylated (or other charged) glycans can be quantified compared to fetuin oligosaccharides and classified as mono-, di-, tri-, and tetrasialylated glycans (see Figure 4). This method is limited to identification of glycans based on the number of charged residues on glycans. Further glycan characterization is achieved by coupling WAX analysis with normal-phase HILIC. This involves collection of individual peaks from WAX profiles and analyzing them using HILIC. (a) Human serum glycome Disialylated glycans EU 60.00 40.00 Monosialylated glycans Trisialylated glycans Tetrasialylated glycans 20.00 0.00 80.00 Disialylated glycans (b) Fetuin standard Trisialylated glycans EU 60.00 Monosialylated glycans 40.00 Tetrasialylated glycans 20.00 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 Minutes Figure 4 Weak anion exchange profiles for 2-AB-labeled N-glycans obtained from (a) human serum and (b) fetuin. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 3.41.3.4.2 475 Hydrophilic interaction liquid chromatography HILIC on an amide 80 column is a widely used method for glycan fingerprinting and for assigning detailed structures of fluorescently labeled glycans. The principle of this method involves interaction of the charged silanol groups on the amide column with the hydroxyl groups on the carbohydrates in a gradient of decreasing organic solvent (typically, acetonitrile) [39]. In general, 20% aqueous glycan solution and 80% acetonitrile (v/v) solution are applied to amide columns such as 3 or 5 µm GlycoSep N columns (Prozyme) and eluted with an increasing aqueous gradient (65–35% (v/v) acetonitrile in 250 mM ammonium formate buffer, pH 4.4). The glycans are eluted based on the hydrophilic surface area exposed to resin (see Figure 5). The retention time of glycan peaks can be compared with an external 2-AB-labeled dextran standard for structural elucidation. The dextran standard is a glucose homopolymer obtained by controlled acid hydrolysis of dextran resulting in homopolymers between 1 and 22 glucose residues. The retention times of peaks standardized with dextran are referred in glucose units (GUs) and are more stable and reproducible on different machines compared with retention time alone. The GU value of the glycan peak is calculated by fitting a fifth-order polynomial distribution curve to the dextran ladder [24]. The GU value is directly related to the number and linkage of its component monosaccharide units. Hence the higher the GU value, the larger the glycan structure [24]. The glycan peaks with their corresponding GU value can be compared with a glycan database to assign structures [33, 40]. There are several advantages for the use of HILIC in glycan fingerprinting. (1) It is a highly sensitive, reproducible, and robust method that can be used to quantify individual glycans and their isobaric glycan stereoisomers; (2) it has low detection limits, that is, it can detect glycans at low femtomolar range; (3) it separates both neutral and charged glycans of therapeutic glycoproteins in the same HPLC run [39, 41, 42]; (4) it provides information on both linkage and monosaccharide sequence; (5) it has high precision with relative standard deviation and coefficient of variation for inter/intra-batch reproducibility of GU values for all the peaks [43, 44]; (6) it has high throughput and automation; (7) it has well-established databases for structural assignments; and (8) it is a simple method and does not require experienced human resources. The major disadvantage of HILIC separations are as follows: (1) glycans cannot be separated based on charge, which is best done by WAX chromatography (see Section 3.41.3.4.1), (2) different glycans co-elute, and (3) data interpretation can be time consuming (see Section 3.41.3.7 for data analysis and glycobioinformatics). The use of ultra-performance liquid chromatography (UPLC) improves the quality, resolution, and sensitivity of glycan analysis [45, 46]. UPLC systems use 2 µm columns and are capable of delivering mobile phases at high pressures with lower dispersion generating sharper peaks. Furthermore, UPLC increases the throughput, decreases the run time, and reduces the solvent use without compromising the quality of results. 3.41.3.4.3 Reversed-phase HPLC Reversed-phase (RP) chromatography is another method used for glycan analysis [36, 42, 47]. Traditionally, this method uses resins incorporating alkyl chains that separate glycans based on hydrophobicity. The principle involves the application of glycans in 100% aqueous solvent to a reversed-phase resin column (C18) and elution with increasing concentrations of organic solvent (50% acetonitrile (v/v) in 50 mM formate, adjusted to pH 5 with triethylamine) [24]. In contrast to HILIC, the columns are calibrated with an arabinose ladder rather than a glucose homopolymer such as dextran. The retention time standardized with arabinose is represented as arabinose unit (AU) [47]. This method is useful for identifying differences in the linkage positions of component sugars, particularly for distinguishing sugars with bisecting GlcNAc from those with same composition but with the GlcNAc on the antenna [24]. This method also works well with LC-MS applications [48]. Although RP-HPLC is a desirable approach for HPLC methods, there is a drawback for MS application. The use of C18 columns results in retention of neutral underivatized glycans. However, the use of graphite columns have given better results as they retain underivatized glycans better than reversed-phase resins [49]. Furthermore, the use of ion-pairing chromatography in reversed phase increases the retention of the charged species [50]. 3.41.3.4.4 High-pH anion exchange chromatography High-pH anion exchange chromatography (HPAEC) involves a high-pH separation of derivatized and underivatized oligosaccharides [51, 52]. Hydroxyl groups of glycans are converted to oxyanions by a high-pH buffer followed by separation of the oxyanion derivatives on an anion exchange column using a high salt gradient. Mobile phases include 150 mM sodium hydroxide and 0–250 mM sodium acetate. Eluted glycans are then detected by an HPAE-PAD or HPAE-FD. The retention time of the glycans depends on the interaction of oxyanions on glycans with the anion exchange column. Separations of glycans are based on charge, size, composition, branching, and linkage isomerism. There are several advantages in using HPAE-PAD: (1) this technique allows analysis of monosaccharides and large oligosaccharides; (2) it allows simultaneous separation of both neutral and charged glycans; (3) it is a versatile method allowing both isocratic and gradient elution; and (4) both amino acids and carbohydrates can be detected in a single run. Although PAD is a desirable method for glycan analysis, each sugar gives a different signal and can be highly incomparable with the different PAD cells raising concerns for day-to-day reproducibility. In addition, it is difficult to quantify glycans using PAD. Fortunately, recent advances in the technology have made these disadvantages less of an issue. The quantification issue can be overcome by labeling glycans with 2-AB and analyzing with fluorescence detectors (HPAE-FD) [53]. HPAE-PAD and HPAE-FD are now widely used orthogonal methods in the biopharmaceutical industry for glycan analysis of therapeutic glycoproteins. In summary, HPLC techniques allow the analysis of glycans based on their size, charge, composition, structure, topology, and branching. Additionally, these methods can also be combined with MS to give more detailed structural information. A key Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 476 Biologics Non-galactosylated glycans (a) Galactosylated glycans Sialylated glycans Fucosylated glycans 3 7 13 8 4 17 9 2 6 1 5 10 11 14 12 16 15 18 GU (b) 5 Glycan peak Name 1 A2 2 A2B 6 Structure 8 7 Glycan peak Name 9 Glycan peak Structure Structure Name F(6)A2B[6]G1 F(6)A2BG2 14 9 A2[3]G1S1 F(6)A2B[3]G1S1 3 F(6)A2 4 F(6)A2B A2[6]BG1S1 15 A2G2S1 16 A2BG2S1 17 F(6)A2G2S1 18 F(6)A2G2S2 F(6)A2B[3]G1 10 5 A2[6]G1 A2B[3]G1S1 6 A2[3]G1 A2G2 11 F(6)A2[6]G1 F(6)A2[6]G1S1 7 A2B[6]G1 A2BG2 12 F(6)A2[3]G1 F(6)A2B[6]G1S1 8 A2B[3]G1 F(6)A2G2 13 A2[6]G1S1 F(6)A2[3]G1S1 (c) Symbol and Name N-Acetylgucosamine Galactose Fucose (deoxygalactose) N-Aacetylneuraminic acid Linkage position 6 8 Linkage β-Linkage α-Linkage Mannose 4 Unknown β-linkage 3 2 Unknown α-linkage Figure 5 (a) HILIC chromatogram of 2-aminobenzamide (2-AB)-labeled N-glycans obtained from human IgG. Peaks in blue represent nongalactosylated glycans, in yellow represent galactosylated glycans, in pink represent sialylated glycans, and in red represent fucosylated glycans that are distributed in all the peaks. (b) N-glycan structures identified on human IgG. The identities of all major peaks were confirmed by GU values and exoglycosidase digestion data. (c) Standard symbols for drawing N- and O-linked carbohydrates and designation of linkage position by the angle linking two monosaccharides shown for glucose [1]. Structural abbreviations: All N-glycans have two-core GlcNAcs; F(6) at the start of the abbreviation specifies a core fucose α(1–6) attached to the inner GlcNac; A2, biantennary with both GlcNAc linked to trimannosyl core; B, bisecting GlcNAc-linked β(1–4) to β(1–3) mannose; Gx, number (x) of β(1–4)-linked galactose on antenna; [3] G1 and [6] G1 indicates that the galactose is on the antenna of the α(1–3) or α(1–6) mannose; Sx, number (x) of sialic acids linked to galactose. Symbolic diagrams: N-Linked structures are represented according to Oxford notation for N- and O-linked glycans [65]. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 477 advantage of all HPLC-based methods is that they are exceptionally versatile and can be used preparatively for follow-on analyses. For example, the released glycan pool can be first separated on WAX HPLC by charge, and the individual peaks can be further analyzed by range of methods including HILIC, RP-HPLC, HPAE, and MS. The major disadvantages of the HPLC method are as follows: (1) several glycans co-elute and (2) elucidation of sequence requires orthogonal technologies such as MS or exoglycosidase array digestions. 3.41.3.5 Glycan Sequencing Using Exoglycosidases The highly branched complex structures of glycans and their isobaric stereoisomers make it difficult to elucidate specific sequence and linkage. Specific exoglycosidase enzymes that cleave glycosidic bonds of individual monosaccharide units or small oligosaccharides from the terminal residue can provide this information. Furthermore, a combination of HPLC, CE, or MS with exoglycosidase array digestions makes it feasible to elucidate the glycan structure with respect to their sequence, number of monosaccharide units, and linkage information. The glycan sequencing is undertaken using exoglycosidase enzymes that systematically cleave off terminal monosaccharide residues depending on their linkage and anomericity (see Figure 6). Generally, one-pot exoglycosidase digestion involves dissolution of dried 2-AB-labeled glycans in enzyme buffer followed by incubation with the enzymes at 37 °C for 16–18 h [31]. It is important to note that buffer and enzyme concentrations are critical to obtain maximum enzyme efficiency. The released glycans are then separated from the exoglycosidase mixture using centrifugal filtration devices and analyzed using HPLC, CE, or MS. In addition, arrays of different exoglycosidases can be used to progressively digest monosaccharide units to obtain the oligosaccharide sequence. After each digestion, the residual enzymes are removed by a centrifugal filtration device and released glycans are reanalyzed by HILIC. The shifts in GU values after each digestion facilitate the determination of structure and linkage information since individual monosaccharide unit displays characteristic GU shifts [31]. A classical set of exoglycosidase arrays is shown in Figure 7 for human serum glycans. 3.41.3.5.1 Glycan analysis Figure 7 shows a series of exoglycosidase array digestions of the N-glycome from human serum. The principle is illustrated on three selected N-glycans: A2G2S2-disialylated biantennary glycan, F(6)A2G2S2-disialylated biantennary glycan with α(1–6)-linked fucose (core fucose), and A3F(3)1G3S3-trisialylated triantennary glycan with α(1–3)-linked fucose (outer arm fucose). After addition of Arthrobacter ureafaciens sialidase (ABS), which releases α(2–3)-, α(2–6)-, and α(2–8)-linked nonreducing terminal sialic acids, A2G2S2 digests to A2G2, F(6)A2G2S2 to F(6)A2G2, and A3F(3)1G3S3 to A3F(3)1G3. After subsequent digestion with ABS ABS SPG BTG JBM BKF CBG BTG JBM AMF SPG BTG GUH ABS/NAN1 BKF Figure 6 Commonly used exoglycosidases in glycan analysis: ABS (Arthrobacter ureafaciens sialidase) releases α(2–6)-, α(2–3)-, and α(2–8)-linked nonreducing terminal acids N-acetylneuraminic acid (NANA) and N-glycolylneuraminic acid (NGNA). NAN1 (recombinant sialidase) releases α(2–3)-linked nonreducing terminal sialic acids (NANA and NGNA). AMF (almond meal α-fucosidase) releases α(1–3)- & α(1–4)-linked nonreducing terminal fucose residues. Not core α(1–6) fucose. BKF (bovine kidney alpha-fucosidase) releases α(1–2)- and α(1–6)-linked nonreducing terminal fucose residues more efficiently than α(1–3)- and α(1–4)-linked fucose. Digests core α(1–6) fucose. BTG (bovine testes β-galactosidase) hydrolyzes nonreducing terminal galactose β(1–3) and β(1–4) linkages. SPG (Streptococcus pneumoniae beta-galactosidase) hydrolyzes nonreducing terminal galactose β(1–4) linkages. CBG (coffee bean alpha-galactosidase) hydrolyzes α(1–3) and α(1–4) galactose. GUH (Streptococcus pneumoniae hexosaminidase, recombinant in E. coli) will digest βGlcNAc, but not a bisecting GlcNAc β(1–4) linked to mannose. JBM (jack bean mannosidase) removes mannose α1-2, 6 > 3. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 478 Biologics A2G2S2 F(6)A2G2S2 A3F(3)1G3S3 Undigested A2G2 F(6)A2G2 A3F(3)1G3 + Sialidase (ABS) A2 F(6)A2 + β-Galactosidase (BTG) A2 A3 A3F(3)1G1 A3F(3)1G1 + α-Fucosidase (BKF) A3 + α-Fucosidase (AMF) M3 + β-Hexosaminidase (GUH) Dextran GU 4 5 6 7 8 9 10 11 12 Figure 7 HILIC profiles of 2-AB-labeled N-glycans from human serum before and after exoglycosidase digestions. The bottom panel shows glucose unit (GU) values from 2-AB-labeled dextran ladder. and bovine testes β-galactosidase (BTG), which hydrolyzes nonreducing terminal β(1–3) and β(1–4) galactose, A2G2 further digests to A2, F(6)A2G2 to F(6)A2, and A3F(3)1G3 to A3F(3)1G1. After subsequent digestion with ABS, BTG, and bovine kidney α-fucosidase (BKF), which releases nonreducing terminal α(1–2) and α(1–6) fucose (core fucose) more efficiently than α(1–3)- and α(1–4)-linked fucose, F(6)A2 digests to A2 and A3F(3)1G1 partially digests to A3. If instead of BKF, almond meal α-fucosidase (AMF) is used, A3F (3)1G1 fully digests to A3. AMF releases α(1–3)- and α(1–4)-linked nonreducing terminal fucose. After ABS + BTG + BKF digests, N-glycan pool can be further digested with Streptococcus pneumoniae hexosaminidase (GUH), expressed in recombinant E. coli, which digests β-linked antennary GlcNAc, but not a bisecting GlcNAc β(1–4) linked to mannose. Both A2 and A3 digest to M3 glycan. 3.41.3.5.2 Applications of oligosaccharide sequencing in glycan analysis of therapeutic glycoproteins Glycan fingerprinting and detailed glycan analysis of therapeutic glycoproteins is an extremely useful tool for regulatory submissions and patent applications of glycotherapeutics. In Chapter 1.32, we reviewed the importance of various glycans with respect to biological function. For example, different glycoforms of recombinant monoclonal antibodies can potentially initiate different biological effects: (1) the presence of fucosylated IgG glycoforms can decrease FcγRIII-mediated antibody-dependent cellular toxicity (ADCC) activity [54] and (2) glycans incorporating N-glycolylneuraminic acid (NGNA) or α(1–3)-linked galactose are immunogenic raising concerns of therapeutic safety [11, 12]. It is now possible to identify specific glycosylation features of recombinant monoclonal antibodies that affect efficacy, circulating half-life, and immunogenicity by using combinatorial exoglycosidase digestions coupled with HILIC analysis. There are several advantages of using oligosaccharides. First, this technology provides relative quantification of different glycans and information on their structures. Second, the use of ABS and Streptococcus pneumonia sialidase (NAN1) exoglycosidases on labeled glycans can differentiate between the linkages of sialic acids. ABS releases α(2–6)-, α(2–3)-, and α(2–8)-linked nonreducing terminal sialic acids, whereas NAN1 releases α(2–3)-linked nonreducing terminal sialic acids. Third, combinatorial exoglycosidase digestions using ABS, BKF, and AMF provide information on the relative levels and Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 479 ratio of G0, G1, and G2 glycans. Fourth, comparison of HILIC profiles from ABS + BTG + BKF and ABS + BTG + AMF reveals the relative levels of core and antennary fucosylation. Finally, immunogenic glycan epitopes such as α(1–3) galactose can be identified using exoglycosidases such as ABS, BTG, and CBG. 3.41.3.6 High-Throughput Glycan Analysis Based on a Robotics Platform The rapidly growing field of recombinant therapeutic medicine has provided the impetus for the development of high-throughput sample preparation and analysis. Glycan analysis has become critical at every stage in the bioprocessing industry from QbD to PAT and critical feature analysis. It is essential to maintain reproducible carbohydrate processing in monitoring recombinant therapeutics for safety and efficacy [55]. In disease, specific changes in the glycosylation of serum proteins have been associated with numerous pathologies such as rheumatoid arthritis [56], ovarian cancer [56], and Congenital Disorders of Glycosylation (CDGs) [57]. In each case, high-throughput glycan analysis has provided both diagnostic and prognostic information. Biomarker discovery and validation requires the analysis of large sample cohorts necessitating high-throughput methods for glycan analysis. The absence of a chromophore and the hydrophilic properties inherent in the glycan structures make oligosaccharide characterization particularly challenging. One of the most important and initial steps in glycan analysis is the development of a robust and reproducible method for glycan release and preparation for analysis which is capable of large sample numbers. To that end, we have developed a high-throughput 96-well platform automated through the use of a robotic liquid handling system (see Figure 8). Previously in our laboratory, a reproducible and robust high-throughput method was developed for N-glycan 2. PNGase F digestion: 1 h Bioreactor 4. Formic acid treatment: 15 min Protein A plate 1. Purify IgG from crude supernatant 5. 2-AB label: 2 h 3. Release and elute glycans Off-line: Speed Vac 6.Transfer labeled glycans to LS-A plate for excess label removal: 15 min 9. Structural assignment using database and quantitation 8. UPLC-profile of the released glycan pool: 20 min 7. Elute 2-AB-labeled glycans, dry, redissolve in set volume Hamilton STAR robotics platform Figure 8 Schematic time line for high-throughput glycan release, fluorescent labeling, UPLC profiling, structural assignments, and relative quantification using Hamilton STAR robotics platform. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 480 Biologics analysis [33]. This method has now been further developed, optimized, and fully automated on a robotic platform. Furthermore, the time taken to complete the glycan removal and analysis has also been significantly reduced, with a current turnaround time of approximately 8 h. Interpretation of the particular oligosaccharide pool present in the samples is accomplished by utilizing GlycoBase in structural assignment [33]. This automated system has a direct application in the production of glycosylated therapeutics where rapid prescreening of producer cells is required, product consistency needs to be controlled, and the implications of glycan heterogeneity on product efficacy and function require assessment [58, 59]. Depending on the sample to be analyzed, the sample can be immobilized in a gel block, applied directly to protein A affinity resin (crude cell culture broth) or immobilized on PVDF membrane. With the aid of integrated incubators on the platform, samples may be rapidly deglycosylated using PNGase F at 37 °C and subsequently labeled with 2-AB at 60 °C. Removal of excess derivatized material is facilitated by a 96-well SPE plate. Integral to this system is a vacuum manifold, which permits eluants to be washed away by applying a gentle vacuum. When collection is required, this is facilitated by drip directors from each well, ensuring that samples are retained in a collection block positioned in the vacuum underdrain while eliminating sample crossover. This high-throughput method is currently operated utilizing a Hamilton STAR platform (Reno, Nevada); however, this optimized method is adaptable to other robotic platforms. In the bioprocessing industry, this robotic platform may be directly interfaced with a bioreactor and LC to ensure a streamlined analytical process. In the instance of certain bioreactor samples, the process of reduction and alkylation may be omitted to further expedite the overall process. Certain analyses may require different outcomes, in which case the platform can be easily adapted to suit individual requirements. This automated platform coupled to LC provides a powerful high-throughput tool for rapid and detailed glycan composition of the therapeutic glycoproteins at all stages of bioproduction. It is anticipated that this adaptable technology will accelerate glycodiagnostics and complement the evolving technologies available for glycomic analysis. 3.41.3.7 Data Analysis and Glycobioinformatics Recent developments in high-throughput HPLC-glycan platforms combined with improved cycle speeds and robotic handling require new methods for processing, analyzing, and disseminating data. The sheer complexity and volume of data routinely generated necessitates glycoinformatic solutions in the form of data repositories and analytical tools to facilitate data interpretation. In Chapter 4.47, we briefly reviewed the current status of glycan-related databases and tools. Unlike genomics and proteomics, glycomics lacks database resources for integrating structural and experimental data. Notably, the field comprises many disconnected and incompatible collections of experimental data and proprietary applications. This not only hinders the development of bioinformatic tools necessary for large-scale studies but can also lead to database sparseness and redundancy. However, recent collaborations between publicaly available databases under the guidance of the working group on glycan database standard (WGGDS) aim to establish protocols for cross-database referencing, querying, and the development of an accurate and user-friendly suite of tools and databases. The EUROCarbDB design study funded by the sixth European Union (EU) framework program established the technical foundations for implementing such a platform, including an introduction and/or recommendation of formats and nomenclatures; user-curated structural and experimental data; and access to software programs and libraries to support continued development (http://www.eurocarbdb.org). The partners have developed a collection of tools and workflows to assist the interpretation of HPLC, MS, and NMR experimental data and by utilizing/extending this it would be possible to develop a global collection of tools for data sharing and analysis. The inherent complexity of glycans and the techniques used to elucidate their structures are bottlenecks to the development of integrated software and database packages. However, there is an increasing number of tools and databases to support glycomics, especially for HPLC [33, 40, 60], MS [61–63], and NMR [64]. Recently, an innovative suite of recent HPLC-based tools has been developed in conjunction with EUROCarbDB to help populate databases and to assist the process of data analysis and annotation to support high-throughput technologies (see Section 3.41.3.6), inclusive of a robust database framework for storing and retrieving glycan structural and experimental data such as GlycoBase, autoGU [40], and GlycoExtractor [60] (http://glycobase.nibrt.ie). GlycoBase is an experimental database that contains the elution positions for 2-AB-labeled glycan structures (expressed in the form of GU values), the predicted products of exoglycosidase digestions, supporting literature information; and a listing of subgroups in which the glycan has been identified. Each structure was fully characterized using normal-phase HPLC with exoglycosidase sequencing and, where required, verified by MS techniques. The pictorial representations of the carbohydrate structures can be dynamically converted to a series of supported nomenclatures using the interfaces created by EUROCarbDB including the Oxford University notation [65] and CFG black-and-white and color formats and textual representation (see Chapter 1.32). The interpretation of HPLC data, including exoglycosidase digestions, can be time consuming, and software (autoGU) is available to assist the assignment of HPLC profiles by listing potential structures for each integrated peak. When the tool is used in combination with data generated from exoglycosidase digestions, a refined list of structures based on the digest footprint, that is, shifts in GU values due to cleavage of terminal monosaccharides, is generated. GlycoExtractor is a recent web-based tool for extracting large volumes of HPLC data from proprietary chromatography databases that store locally acquired sample runs. It allows users to extract sample data generated by high-throughput methods and improves existing methods for processing and exporting large volumes of information. The tool eliminates the requirement to manually Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 481 export data to multiple disconnected files, by querying multiple experimental projects and exporting all necessary data, for example, GU values and peak areas, to a single file in human-readable form that will be supported by autoGU for automated data analysis. 3.41.3.8 Mass Spectrometry The development of MS has enabled the detailed analysis of glycotherapeutics. MS is the chosen method for glycan analysis because of its reproducibility, sensitivity, and speed. Glycan characterization via MS is accomplished through various means: analysis of positively or negatively charged analytes; coupling online separation methods prior to MS such as CE, HPLC, or GC; and derivatization including permethylation. Modern techniques for glycan analysis focus primarily on the released intact glycan. The main disadvantage to this approach is identification of glycosylation site occupancy to the protein backbone that is unattainable. Alternatively, analysis of intact or digested glycoproteins does not provide detailed structural information on the glycan component. Here we highlight the various preferred MS-based methods for glycan analysis. MS-based glycan analyses employ both MALDI and ESI ion sources. Each is capable for oligosaccharide analysis, but different ionization methods allow for distinctive analysis. MALDI involves the mixing of glycans to be analyzed with a matrix, commonly 2,5-dihydroxybenzoic acid (DHB). DHB absorbs the energy from a UV laser and transfers its charge to the analyte molecules. Matrixassisted laser desorption time of flight (MALDI-TOF) instruments are ideal for quick and compositional analysis. Additionally, MALDI-TOF-MS requires minimal sample expenditure and is more tolerant of sample impurities such as salts. Moreover MALDITOF primarily generates singly charged ions that simplify data interpretation, whereby each peak represents a single mass. However, structural isomers or isobars are not distinguishable from a single MALDI-TOF-MS spectrum. Therefore, the reported structural assignment, most significant of which is linkage between residues, is a conjecture that is based on the N-glycan biosynthetic pathway. A second disadvantage of MALDI-TOF-MS analysis is that charged residues (N-acetylneuraminic acid (NANA), N-glycolylneuraminic acid (NGNA), sulfated, or phosphorylated residues) are fragmented during the ionization process. This fragmentation leads to inaccurate glycan assignments. This problem can be met with permethylation, where hydroxyl groups are converted to methyl ethers, which stabilizes charged species and allows for analysis on MALDI-TOF instruments [66]. ESI provides continuous flow of sample, unlike MALDI where ionized analyte molecules are generated during bursts when ablated by the UV laser. The advantage of ESI is that the continuous flow of the glycans in solution allows for longer analysis time and the ability to combine chromatographic interfaces. ESI-based instruments may be used in conjunction with a wider range of mass analyzers and therefore enables various functions that tailor to a specific analyte. For example, sequential MS (MSn) following permethylation requires an ion trap for the multiple rounds of ion isolation and fragmentation [67, 68]. MSn strategies are distinctive in MS glycan analysis in their ability to detect structural isomers, isobars, and biomarkers [66, 69–71]. Analysis of permethylated glycans is considered the method of choice among MS glycomics laboratories [72]. Mass spectrometers can analyze glycans as positively or negatively charged ions. In both MALDI and ESI ion sources, protonated and sodiated ions predominate in positive mode analysis and deprotonated ions in negative mode (see Figure 9), especially when sialic acids are present. In addition to sodium ions, lithium and potassium ions are also likely in positive mode analysis. Negative mode tandem MS, which has received less attention in the past, is particularity powerful to [M–3H]3– 958.7 Relative abundance 100 [M–2H]2– 1293.5 [M–4H]4– 791.8 50 [M–2H]2– 1439.0 [M–3H]3– 1056.1 2– [M–2H] 1110.4 [M–2H]2– 1584.6 0 0 800 1000 1200 1400 1600 [M–H]– 1930.7 1800 2000 [M–H]– 2295.8 2200 m/z Figure 9 Negative-mode tandem MS analysis of N-glycans from fetuin (fetal calf). N-Glycan structures are present as [M – H]−, [M – 2H]2−, [M – 3H]3−, and/or [M – 4H]4− ions. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 482 Biologics investigate acidic glycans (i.e., glycans with sialic acid or sulfate groups) [73–75]. Negative mode analysis is useful because fragmentation of glycans via collision-induced dissociation (CID) results in cross-ring cleavages (A-type) that are more informative than fragments between residues (B- and Y-type) [76]. A-type fragments are more useful in defining N-glycan structures than B- and Y-type cleavages between residues that predominate in positive mode analysis. It is important to note that the workflow and sample preparation largely influences the type of ion/adduct present and should be largely considered during the wet-lab workflow prior to MS. Typically, ESI-MS analyses require larger sample than MALDI-MS. Therefore, ESI analyses typically couple LC, using porous graphitized carbon (PGC), amide-80 resins, ion exchange, or reverse C18 stationary phases. A consequence of online MS approaches is the degree to which glycans can be analyzed via MSn. For MSn applications, a continuous injection of an entire sample is required so a single m/z precursor can be analyzed for extended periods of time. During LC-MS, the sample flow is limited to its relevant retention time, which limits the time to which a composition can be analyzed. Generally, a comprehensive MS analysis of N- or O-glycans will incorporate both MALDI and ESI ion sources, where MALDI-MS confirms the glycan compositions in a sample (i.e., a glycan mass map) and ESI-MS or LC-ESI-MS in positive or negative mode determines the topology and branching of the structures in the sample. Above all, sample preparation and purity is crucial in any MS approach; therefore, online chromatography interfaces can be advantageous. The field of MS glycomics relies heavily on the development of instrumentation, but the future of MS-based glycomics will aim to characterize the glycosylated peptide backbone while determining the fine structure of the glycan itself. 3.41.3.9 Capillary Gel Electrophoresis with Laser-Induced Fluorescence In recent years, capillary gel electrophoresis with laser-induced fluorescence (CGE-LIF) has gained high popularity with the biopharmaceutical industry for its ability to analyze structure and linkage information from heterogeneous mixtures of glycoforms. Typically, the principle involves separation of derivatized glycans (APTS-labeled free glycans) in capillary gel matrix under the influence of electric field using classical CE instrumentation [77], capillary array electrophoresis, or microfluidics platform [78–80]. In general, APTS-labeled glycans in aqueous solution are injected electrochemically (2 kV for 20 s) and separations are carried out at 2–8 kV using 10–40 cm capillaries filled with Carbohydrate (CARB) separation solution (eGene). The separated glycans are detected by a photodiode using a 520-nm long-pass filter [43]. The samples elute in 10 min. The separation times are significantly shorter than HPLC approaches because the sulfate groups present in APTS label increase the mobility of the glycans. The glycan peaks can be compared with an external 2-APTS-labeled dextran standard for structural elucidation. CGE-LIF is a robust and reproducible method that offers high separation efficiency, separations based on linkage, smaller sample volume, faster analysis time, multicapillary options [81], high-throughput analysis [79, 80], and automation. Furthermore, CE can be coupled with exoglycosidase digestion to obtain changes in glycosylation pattern and structure [82, 83]. Though CGE-LIF is a desirable and fast method for glycan analysis, there are currently no glycan databases available for structural assignments. Therefore, the data have to be compared with other orthogonal methods such as HILIC or MS. 3.41.4 Analysis of Monosaccharides Quantitative monosaccharide composition analysis is a powerful technique that provides specific insights into the levels of various monosaccharides such as neutral, N-acetylated, and sialic acid species. Monosaccharide analysis allows detection of both desirable glycans (fucose, galactose, mannose, GlcNAc, GalNAc, and NANA) and aberrant glycans such as NGNA. This information can be used at all facets of bioproduction to determine the type of glycosylation (N- or O-linked glycans) and the extent of glycosylation. There are several methods to analyze monosaccharide composition such as GC, HPAE-PAD, HPLC, and CE. The most commonly used methods, HPLC and CE, require release of monosaccharides by mild acid hydrolysis followed by derivatization [84, 85]. 3.41.4.1 Analysis of Neutral Monosaccharides Typically, this involves hydrolysis of glycoproteins in 2–4 mol dm−3 trifluoroacetic acid (TFA) for 3–6 h at 100 °C followed by re-N-acetylation of amino sugars using aqueous sodium bicarbonate buffer and desalting on cation exchange resins [84, 85]. The released monosaccharides are derivatized with 2-AB or 2-AA and analyzed by either RP-HPLC or HPAE-FD. The HPLC profiles are compared against external standards that are matched to the analytes. 3.41.4.2 Analysis of Sialic Acids Distribution of sialic acids on therapeutic glycoproteins significantly modulates the circulating half-life, safety, and biological function of these drugs. Under the current regulatory guidelines International Conference on Harmonisation (ICH) Q6B and Q5E, it is now required to characterize sialic acid speciation during and after bioproduction process. Sialic acids are a family of nine carbon terminal monosaccharides attached to galactose residues at the nonreducing termini of both N- and O-linked glycans. The Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 483 sialic acids are usually linked to galactose by α(2–3) or α(2–6) linkage. There are more than 20 different naturally occurring sialic acid species reported [86]. The two most common types of sialic acid residues found in recombinant glycoproteins are NANA and NGNA [87]. NGNA is a nonhuman glycoform that differs by a glycolyl group instead of an acetyl group at C-5 amino group [12] and hence is crucial to control the ratio of NANA to NGNA during bioproduction. In addition, other sialic acids containing O-acetyl, methyl, lactyl, phosphate, and sulfate substituents have also been reported [86]. Analysis of DMB-labeled sialic acids is a widely used method for sialic acid speciation. Typically, the method involves release of sialic acids using mild acid hydrolysis (2 mol dm−3 acetic acid for 2 h at 80 °C) followed by DMB labeling. The labeled sialic acids are then stabilized by sodium dithionite. Finally, the labeled sialic acids are analyzed by RP-HPLC with C18 columns. The sialic acids are calibrated with three standards including a mixture of sialic acids found in humans and animals, and NANA and NGNA standards [37]. Information from monosaccharide analysis can be used at all stages of drug development, particularly in (1) preliminary investigation of glycosylation, (2) screening of glycoproteins produced using different cell lines, (3) determination of nonhuman glycans such as NGNA, (4) identification of carbohydrate impurities, and (5) demonstration of the consistency between the batches during fermentation. The major downside of monosaccharide analysis is loss of oligosaccharide branching information, and destruction of released monosaccharides by acid hydrolysis. In addition, hydrolysis does not work on all glycan linkages or substituted glycans. 3.41.5 Glycan Analysis Design for Therapeutic Glycoproteins Setting up a scheme for glycan analysis of a therapeutic glycoprotein is not a straightforward task. Glycans consist of monosaccharide building blocks linked in a structurally diverse manner, often forming complex branched structures. Unlike the proteome, glycosylation is not directly template driven which makes it hard to predict the glycan structure based on gene expression levels. Depending on the number of glycosylation sites and glycan macro- and microheterogeneity, a typical sample from a bioreactor will consist of a range of glycoforms. This means that same glycoprotein backbone can contain different glycans that differ in monosaccharide linkage and composition. Figure 10 displays the classic glycoanalytical techniques employed to analyze the glycosylation pattern of therapeutic glycoproteins. The design consists of four different modules: (1) analysis of monosaccharides, (2) analysis of intact glycoproteins, (3) analysis of glycopeptides, and (4) analysis of free glycans. Each module provides detailed insights into different glycosylation parameters required by regulatory authorities such as monosaccharide profile, sialic acid speciation, glycosylation site profiles, glycopeptide profiles, and oligosaccharide sequence profiles. A combination of the various analytical methods allows us to analyze the complete glycan architecture of glycoproteins and to identify GCQAs. In addition, it is essential to use orthogonal methods to confirm the glycan structure. The analytical approach depends on the nature of the therapeutic glycoprotein, its manufacturing process parameters, critical quality attributes, and especially the predicted glycan structure, oligosaccharide pattern of the natural molecule, cell culture, and expression system. For example, erythropoietin (EPO) produced in mammalian expression systems relies on fully sialylated glycans for stability and regulatory function of erythroid progenitors. Many studies show that sialic acids and their speciation play a critical role in drug safety and efficacy. EPO isoforms having fewer sialic acid residues showed a reduction in circulating half-life from 2 h to less than 10 min [8, 9]. It is important to consider the levels and distribution of sialic acid residues: extent of sialic acid speciation, relative levels of human versus nonhuman sialylation, and detailed oligosaccharide sequences and their linkage. Appropriate glycan analysis to measure GCQAs for EPO includes sialic acid profile (NANA vs. NGNA), glycosylation site analysis, analysis of released Analysis of intact glycoprotein Analysis of monosaccharides Monosaccharide profile Glycosylation site profile MS, CE, LC-MS Acid hydrolysis Therapeutic Proteolysis glycoproteins DMB labeling RP-HPLC Sialic acid profile Oligosaccharide MS profile Glycosylation site profile Glycan sequencing profile Oligosaccharide LC-MS profile HPAE-PAD, CE Free monosaccharides Analysis of free glycans Analysis of glycopeptides MS, CE, LC-MS Glyopeptides MS Glycan release HPAE-PAD Lectin array oligosaccharide profile Glycopeptide profile Free glycans (underivatized) HPAE-PAD Oligosaccharide HPAE profile Exoglycosidase LC-MS digestion HILIC Oligosaccharide HILLIC profile Labeled glycans (derivatized) WAX Charge profile CE Oligosaccharide CE profile Figure 10 Different strategies employed in analysis of biopharmaceutical glycosylation. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 484 Biologics glycans using a set of orthogonal methods including HILIC amide profile, CE profile, WAX HPLC profile plus MS (including MALDI-MS and MS-MS), and exoglycosidase sequencing for detailed structure analysis. In addition, the relative quantification of individual glycan species and detailed sequence and linkage information is achieved by HILIC with fluorescence detection along with exoglycosidase digestion [33]. Although considerable effort has been put into improving glycoanalytical methods, there is no single method available that can characterize the complete glycan structure. It is therefore essential to apply several orthogonal methods to measure individual parameters such as glycosylation site occupancy, oligosaccharide sequence, and monosaccharide content of a glycoprotein. A combination of glycoanalytical methods provides the rational tools needed to establish the detailed glycan structures that are best suited for therapeutic use. References [1] Apweiler R, Hermjakob H, and Sharon N (1999) On the frequency of protein glycosylation, as deduced from analysis of the SWISS-PROT database. Biochimica et Biophysica Acta 1473(1): 4–8. [2] Varki A, Cummings RD, Esko JD, et al. (2009) Essentials of Glycobiology. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. [3] Brooks SA, Dwek MW, and Schumacher U (2002) Functional & Molecular Glycobiology P. Dines. Oxford: BIOS Scientific Publishers Limited. [4] Kornfeld R and Kornfeld S (1985) Assembly of asparagine-linked oligosaccharides. Annual Review of Biochemistry 54: 631–664. [5] Yamashita K, Kamerling JP, and Kobata A (1982) Structural study of the carbohydrate moiety of hen ovomucoid. Occurrence of a series of pentaantennary complex-type asparagine-linked sugar chains. The Journal of Biological Chemistry 257(21): 12809–12814. [6] Mimura Y, Sondermann P, Ghirlando R, et al. (2001) Role of oligosaccharide residues of IgG1-Fc in Fc gamma RIIb binding. The Journal of Biological Chemistry 276(49): 45539–45547. [7] Nimmerjahn F and Ravetch JV (2008) Fc[gamma] receptors as regulators of immune responses. Nature Reviews. Immunology 8(1): 34–47. [8] Erbayraktar S, Grasso G, Sfacteria A, et al. (2003) Asialoerythropoietin is a nonerythropoietic cytokine with broad neuroprotective activity in vivo. Proceedings of the National Academy of Sciences of the United States of America 100(11): 6741–6746. [9] Fukuda M, Sasaki H, Lopez L, and Fukuda M (1989) Survival of recombinant erythropoietin in the circulation: The role of carbohydrates. Blood 73(1): 84–89. [10] Arnold JN, Wormald MR, Sim RB, et al. (2007) The impact of glycosylation on the biological function and structure of human immunoglobulins. Annual Review of Immunology 25(1): 21–50. [11] Chung CH, Mirakhur B, Chan E, et al. (2008) Cetuximab-induced anaphylaxis and IgE specific for galactose-alpha-1,3-galactose. The New England Journal of Medicine 358(11): 1109–1117. [12] Muchmore EA, Milewski M, Varki A, and Diaz S (1989) Biosynthesis of N-glycolyneuraminic acid. The primary site of hydroxylation of N-acetylneuraminic acid is the cytosolic sugar nucleotide pool. The Journal of Biological Chemistry 264(34): 20216–20223. [13] Martı́nez-Maza R, Poyatos I, López-Corcuera B, et al. (2001) The role of N-glycosylation in transport to the plasma membrane and sorting of the neuronal glycine transporter glyt2. Journal of Biological Chemistry 276(3): 2168–2173. [14] Hossler P, Khattak SF, and Li ZJ (2009) Optimal and consistent protein glycosylation in mammalian cell culture. Glycobiology 19(9): 936–949. [15] Balaguer E and Neusüss C (2006) Glycoprotein characterization combining intact protein and glycan analysis by capillary electrophoresis-electrospray ionization-mass spectrometry. Analytical Chemistry 78(15): 5384–5393. [16] Bateman KP, White RL, Yaguchi M, and Thibault P (1998) Characterization of Protein Glycoforms by Capillary-Zone Electrophoresis-Nanoelectrospray Mass Spectrometry, p. 442. Amsterdam, PAYS-BAS: Elsevier. [17] Amon S, Zamfir AD, and Rizzi A (2008) Glycosylation analysis of glycoproteins and proteoglycans using capillary electrophoresis-mass spectrometry strategies. Electrophoresis 29(12): 2485–2507. [18] Kakehi K, Kinoshita M, and Nakano M (2002) Analysis of glycoproteins and the oligosaccharides thereof by high-performance capillary electrophoresis – significance in regulatory studies on biopharmaceutical products. Biomedical Chromatography 16(2): 103–115. [19] Pilobello KT and Mahal LK (2007) Lectin microarrays for glycoprotein analysis. Methods in Molecular Biology 385: 193–203. [20] Pilobello KT and Mahal LK (2007) Deciphering the glycocode: The complexity and analytical challenge of glycomics. Current Opinion in Chemical Biology 11(3): 300–305. [21] Pilobello KT, Slawek DE, and Mahal LK (2007) A ratiometric lectin microarray approach to analysis of the dynamic mammalian glycome. Proceedings of the National Academy of Sciences of the United States of America 104(28): 11534–11539. [22] Rosenfeld R, Bangio H, Gerwig GJ, et al. (2007) A lectin array-based methodology for the analysis of protein glycosylation. Journal of Biochemical and Biophysical Methods 70(3): 415–426. [23] La Belle JT, Gerlach JQ, Svarovsky S, and Joshi L (2007) Label-free impedimetric detection of glycan-lectin interactions. Analytical Chemistry 79(18): 6959–6964. [24] Royle L, Dwek RA, and Rudd PM (2006) Determining the structure of oligosaccharides N- and O-linked to glycoproteins. Current Protocols in Protein Science Chapter 12: Unit 12.6. [25] Patel T, Bruce J, Merry A, et al. (1993) Use of hydrazine to release in intact and unreduced form both N- and O-linked oligosaccharides from glycoproteins. Biochemistry 32: 679–693. [26] Merry AH, Neville DCA, Royle L, et al. (2002) Recovery of intact 2-aminobenzamide-labeled O-glycans released from glycoproteins by hydrazinolysis. Analytical Biochemistry 304(1): 91–99. [27] Deberire P, Montreuil J, Moczar E, et al. (1985) Primary structure of two major glycans of bovine fibrinogen. European Journal of Biochemistry 151(3): 607–611. [28] Kuhn P, Tarentino AL, Plummer TH, Jr., and Van Roey P (1994) Crystal structure of peptide-N4-(N-acetyl-beta-D-glucosaminyl)asparagine amidase F at 2.2-A resolution. Biochemistry 33(39): 11699–11706. [29] Tretter V, Altmann F, and Marz L (1991) Peptide-N4-(N-acetyl-β-glucosaminyl)asparagine amidase F cannot release glycans with fucose attached α1 → 3 to the asparagine-linked N-acetylglucosamine residue. European Journal of Biochemistry 199(3): 647–652. [30] Zhao KW, Faull KF, Kakkis ED, and Neufeld EF (1997) Carbohydrate structures of recombinant human alpha-L-iduronidase secreted by Chinese hamster ovary cells. The Journal of Biological Chemistry 272(36): 22758–22765. [31] Royle L, Radcliffe CM, Dwek RA, and Rudd PM (2006) Detailed structural analysis of N-glycans released from glycoproteins in SDS-PAGE gel bands using HPLC combined with exoglycosidase array digestions. In: Brockhausen I (ed.) Methods in Molecular Biology. Totowa, NJ: Humana Press Inc. [32] Trimble RB and Maley F (1984) Optimizing hydrolysis of N-linked high-mannose oligosaccharides by endo-beta-N-acetylglucosaminidase H. Analytical Biochemistry 141(2): 515–522. [33] Royle L, Campbell MP, Radcliffe CM, et al. (2008) HPLC-based analysis of serum N-glycans on a 96-well plate platform with dedicated database software. Analytical Biochemistry 376(1): 1–12. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy Protein Glycosylation 485 [34] Bigge JC, Patel TP, Bruce JA, et al. (1995) Nonselective and efficient fluorescent labeling of glycans using 2-amino benzamide and anthranilic acid. Analytical Biochemistry 230(2): 229–238. [35] Shilova NV and Bovin NV (2002) Fluorescent labels for the analysis of mono- and oligosaccharides. Russian Journal of Bioorganic Chemistry 29(4): 339–355. [36] Kamoda S, Nakano M, Ishikawa R, et al. (2005) Rapid and sensitive screening of N-glycans as 9-fluorenylmethyl derivatives by high-performance liquid chromatography: A method which can recover free oligosaccharides after analysis. Journal of Proteome Research 4(1): 146–152. [37] Fernandes D (2006) Biopharmaceutical sialylation. European Biopharmaceutical Review Spring: 100–104. [38] Prater BD, Anumula KR, and Hutchins JT (2007) Automated sample preparation facilitated by PhyNexus MEA purification system for oligosaccharide mapping of glycoproteins. Analytical Biochemistry 369(2): 202–209. [39] Guile GR, Rudd PM, Wing DR, et al. (1996) A rapid high-resolution high-performance liquid chromatographic method for separating glycan mixtures and analyzing oligosaccharide profiles. Analytical Biochemistry 240(2): 210–226. [40] Campbell MP, Royle L, Radcliffe CM, et al. (2008) Glycobase and autogu: Tools for HPLC-based glycan analysis. Bioinformatics 24(9): 1214–1216. [41] Anumula KR (2000) High-sensitivity and high-resolution methods for glycoprotein analysis. Analytical Biochemistry 283(1): 17–26. [42] Anumula KR (2006) Advances in fluorescence derivatization methods for high-performance liquid chromatographic analysis of glycoprotein carbohydrates. Analytical Biochemistry 350(1): 1–23. [43] Domann PJ, Pardos-Pardos AC, Fernandes DL, et al. (2007) Separation-based glycoprofiling approaches using fluorescent labels. Proteomics 7(Suppl. 1): 70–76. [44] Knezević A, Polasek O, Gornik O, et al. (2008) Variability, heritability and environmental determinants of human plasma N-glycome. Journal of Proteome Research 8(2): 694–701. [45] Guillarme D, Grata E, Glauser G, et al. (2009) Some solutions to obtain very efficient separations in isocratic and gradient modes using small particles size and ultra-high pressure. Journal of Chromatography A 1216(15): 3232–3243. [46] Yang Y and Hodges CC (2005) Assay Transfer from HPLC to UPLC for Higher Analysis Throughput. www.Chromatographyonline.com (accessed 27th Jan 2011). [47] Guile GR, Harvey DJ, O’Donnell N, et al. (1998) Identification of highly fucosylated N-linked oligosaccharides from the human parotid gland. European Journal of Biochemistry 258(2): 623–656. [48] Pabst M, Bondili JS, Stadlmann J, et al. (2007) Mass + retention time = structure: A strategy for the analysis of N-glycans by carbon LC-ESI-MS and its application to fibrin N-glycans. Analytical Chemistry 79(13): 5051–5057. [49] Koizumi K (1996) High-performance liquid chromatographic separation of carbohydrates on graphitized carbon columns. Journal of Chromatography A 720: 119–126. [50] Gennaro LA, Harvey DJ, and Vouros P (2003) Reversed-phase ion-pairing liquid chromatography/ion trap mass spectrometry for the analysis of negatively charged, derivatized glycans. Rapid Communications in Mass Spectrometry 17(14): 1528–1534. [51] Chen L, Yet M, and Shao M (1988) New methods for rapid separation and detection of oligosaccharides from glycoproteins. The FASEB Journal 2(12): 2819–2824. [52] Townsend RR and Hardy MR (1991) Analysis of glycoprotein oligosaccharides using high-pH anion exchange chromatography. Glycobiology 1(2): 139–147. [53] Kotani N and Takasaki S (1998) Analysis of 2-aminobenzamide-labeled oligosaccharides by high-pH anion-exchange chromatography with fluorometric detection. Analytical Biochemistry 264(1): 66–73. [54] Okazaki A, Shoji-Hosaka E, Nakamura K, et al. (2004) Fucose depletion from human IgG1 oligosaccharide enhances binding enthalpy and association rate between IgG1 and Fc[gamma]RIIIa. Journal of Molecular Biology 336(5): 1239–1249. [55] Walsh G and Jefferis R (2006) Post-translational modifications in the context of therapeutic proteins. Nature Biotechnology 24(10): 1241–1252. [56] van de Geijn FE, Wuhrer M, Selman MH, et al. (2009) Immunoglobulin G galactosylation and sialylation are associated with pregnancy-induced improvement of rheumatoid arthritis and the postpartum flare: Results from a large prospective cohort study. Arthritis Research & Therapy 11(6): R193. [57] Haeuptle MA and Hennet T (2009) Congenital disorders of glycosylation: An update on defects affecting the biosynthesis of dolichol-linked oligosaccharides. Human Mutation 30(12): 1628–1641. [58] Blow N (2009) Glycobiology: A spoonful of sugar. Nature 457(7229): 617–620. [59] Kawasaki N, Itoh S, Hashii N, et al. (2009) The significance of glycosylation analysis in development of biopharmaceuticals. Biological & Pharmaceutical Bulletin 32(5): 796–800. [60] Artemenko NV, Campbell MP, and Rudd PM (2010) GlycoExtractor – a web-based interface for high throughput processing of HPLC-glycan data. Journal of Proteome Research 9: 2037–2041. [61] Goldberg D, Sutton-Smith M, Paulson J, and Dell A (2005) Automatic annotation of matrix-assisted laser desorption/ionization N-glycan spectra. Proteomics 5(4): 865–875. [62] Ceroni A, Maass K, Geyer H, et al. (2008) GlycoWorkbench: A tool for the computer-assisted annotation of mass spectra of glycans. Journal of Proteome Research 7 (4): 1650–1659. [63] Maass K, Ranzinger R, Geyer H, et al. (2007) ‘Glyco-peakfinder’ – de novo composition analysis of glycoconjugates. Proteomics 7(24): 4435–4444. [64] Loss A, Stenutz R, Schwarzer E, and von der Lieth CW (2006) GlyNest and CASPER: Two independent approaches to estimate 1H and 13C NMR shifts of glycans available through a common web-interface. Nucleic Acids Research 34(Web Server issue): W733–W737. [65] Harvey DJ, Merry AH, Royle L, et al. (2009) Proposal for a standard system for drawing structural diagrams of N- and O-linked carbohydrates and related compounds. Proteomics 9(15): 3796–3801. [66] Ashline DJ, Lapadula AJ, Liu YH, et al. (2007) Carbohydrate structural isomers analyzed by sequential mass spectrometry. Analytical Chemistry 79(10): 3830–3842. [67] Ashline D, Singh S, Hanneman A, and Reinhold V (2005) Congruent strategies for carbohydrate sequencing. 1. Mining structural details by MSn. Analytical Chemistry 77(19): 6250–6262. [68] Sheeley DM and Reinhold VN (1998) Structural characterization of carbohydrate sequence, linkage, and branching in a quadrupole ion trap mass spectrometer: Neutral oligosaccharides and N-linked glycans. Analytical Chemistry 70(14): 3053–3059. [69] Prien JM, Huysentruyt LC, Ashline DJ, et al. (2008) Differentiating N-linked glycan structural isomers in metastatic and nonmetastatic tumor cells using sequential mass spectrometry. Glycobiology 18(5): 353–366. [70] Hanneman AJ, Rosa JC, Ashline D, and Reinhold VN (2006) Isomer and glycomer complexities of core GlcNAcs in Caenorhabditis elegans. Glycobiology 16(9): 874–890. [71] Prien JM, Ashline DJ, Lapadula AJ, et al. (2009) The high mannose glycans from bovine ribonuclease B isomer characterization by ion trap MS. Journal of the American Society for Mass Spectrometry 20(4): 539–556. [72] Zaia J (2008) Mass spectrometry and the emerging field of glycomics. Chemistry & Biology 15(9): 881–892. [73] Harvey DJ (2005) Fragmentation of negative ions from carbohydrates: Part 2, fragmentation of high-mannose N-linked glycans. Journal of the American Society for Mass Spectrometry 16: 631–646. [74] Harvey DJ (2005) Fragmentation of negative ions from carbohydrates: Part 1. Use of nitrate and other anionic adducts for the production of negative ion electrospray spectra from N-linked carbohydrates. Journal of the American Society for Mass Spectrometry 16: 622–630. [75] Harvey DJ (2005) Fragmentation of negative ions from carbohydrates: Part 3. Fragmentation of hybrid and complex N-linked glycans. Journal of the American Society for Mass Spectrometry 16: 647–659. [76] Domon B and Costello CE (1988) A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates. Glycoconjugate Journal 5: 397–409. [77] Guttman A (1996) High-resolution carbohydrate profiling by capillary gel electrophoresis. Nature 380(6573): 461–462. [78] Callewaert N, Contreras R, Mitnik-Gankin L, et al. (2004) Total serum protein N-glycome profiling on a capillary electrophoresis-microfluidics platform. Electrophoresis 25(18–19): 3128–3131. [79] Khandurina J, Anderson AA, Olson NA, et al. (2004) Large-scale carbohydrate analysis by capillary array electrophoresis: Part 2. Data normalization and quantification. Electrophoresis 25(18–19): 3122–3127. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0 Author's personal copy 486 Biologics [80] Khandurina J, Blum DL, Stege JT, and Guttman A (2004) Automated carbohydrate profiling by capillary electrophoresis: A bioindustrial approach. Electrophoresis 25(14): 2326–2331. [81] Olajos M, HajĂłs P, Bonn GK, and Guttman A (2008) Sample preparation for the analysis of complex carbohydrates by multicapillary gel electrophoresis with light-emitting diode induced fluorescence detection. Analytical Chemistry 80(11): 4241–4246. [82] Edge CJ, Rademacher TW, Wormald MR, et al. (1992) Fast sequencing of oligosaccharides: The reagent-array analysis method. Proceedings of the National Academy of Sciences of the United States of America 89(14): 6338–6342. [83] Guttman A and Ulfelder KW (1997) Exoglycosidase matrix-mediated sequencing of a complex glycan pool by capillary electrophoresis. Journal of Chromatography A 781(1–2): 547–554. [84] Fan JQ, Namiki Y, Matsuoka K, and Lee YC (1994) Comparison of acid hydrolytic conditions for Asn-linked oligosaccharides. Analytical Biochemistry 219(2): 375–378. [85] Hardy MR, Townsend RR, and Lee YC (1988) Monosaccharide analysis of glycoconjugates by anion exchange chromatography with pulsed amperometric detection. Analytical Biochemistry 170(1): 54–62. [86] Higa HH and Paulson JC (1985) Sialylation of glycoprotein oligosaccharides with N-acetyl-, N-glycolyl-, and N-O-diacetylneuraminic acids. Journal of Biological Chemistry 260(15): 8838–8849. [87] Baker KN, Rendall MH, Hills AE, et al. (2001) Metabolic control of recombinant protein N-glycan processing in NS0 and CHO cells. Biotechnology and Bioengineering 73(3): 188–202. Comprehensive Biotechnology, Second Edition, 2011, Vol. 3, 467-486, DOI: 10.1016/B978-0-08-088504-9.00230-0