Rapid Development of New Protein Biosensors Utilizing

Peptides Obtained via Phage Display

Jun Wu1¤a, Jong Pil Park1¤b, Kevin Dooley1, Donald M. Cropek2, Alan C. West1, Scott Banta1*
1 Department of Chemical Engineering, Columbia University, New York, New York, United States of America, 2 United States Army Engineer Research and Development
Center, Construction Engineering Research Laboratory (CERL), Champaign, Illinois, United States of America

Abstract
There is a consistent demand for new biosensors for the detection of protein targets, and a systematic method for the rapid
development of new sensors is needed. Here we present a platform where short unstructured peptides that bind to a
desired target are selected using M13 phage display. The selected peptides are then chemically synthesized and
immobilized on gold, allowing for detection of the target using electrochemical techniques such as electrochemical
impedance spectroscopy (EIS). A quartz crystal microbalance (QCM) is also used as a diagnostic tool during biosensor
development. We demonstrate the utility of this approach by creating a novel peptide-based electrochemical biosensor for
the enzyme alanine aminotransferase (ALT), a well-known biomarker of hepatotoxicity. Biopanning of the M13 phage
display library over immobilized ALT, led to the rapid identification of a new peptide (ALT5-8) with an amino acid sequence
of WHWRNPDFWYLK. Phage particles expressing this peptide exhibited nanomolar affinity for immobilized ALT
(Kd,app = 85620 nM). The newly identified ALT5-8 peptide was then chemically synthesized with a C-terminal cysteine for
gold immobilization. The performance of the gold-immobilized peptides was studied with cyclic voltammetry (CV), QCM,
and EIS. Using QCM, the sensitivity for ALT detection was 8.960.9 Hz/(mg/mL) and the limit of detection (LOD) was 60 ng/
mL. Using EIS measurements, the sensitivity was 142612 impedance percentage change %/(mg/mL) and the LOD was
92 ng/mL. In both cases, the LOD was below the typical concentration of ALT in human blood. Although both QCM and EIS
produced similar LODs, EIS is preferable due to a larger linear dynamic range. Using QCM, the immobilized peptide
exhibited a nanomolar dissociation constant for ALT (Kd = 20.160.6 nM). These results demonstrate a simple and rapid
platform for developing and assessing the performance of sensitive, peptide-based biosensors for new protein targets.

Citation: Wu J, Park JP, Dooley K, Cropek DM, West AC, et al. (2011) Rapid Development of New Protein Biosensors Utilizing Peptides Obtained via Phage
Display. PLoS ONE 6(10): e24948. doi:10.1371/journal.pone.0024948
Editor: Richard C. Willson, University of Houston, United States of America
Received March 14, 2011; Accepted August 24, 2011; Published October 7, 2011
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for
any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This study was funded by the United States Army Corps of Engineers Applied Research program. The funders had no role in study design, data
collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: [email protected]
¤a Current address: Atotech USA, Inc., Albany, New York, United States of America
¤b Current address: Department of Herbal Pharmaceutical Engineering, Daegu Haany University, Gyeongsan, Korea

Introduction hindered by their high cost and the significant time necessary to
develop new antibodies to emerging targets [4]. Several efforts
A biosensor is an analytical device that combines a have been made to address these limitations on the biomolecular
recognition element with a transducer (detection element) for recognition element, including the use of nucleic acid-based
the detection of a biological analyte (target) [1,2]. The aptamers [5] alternative protein scaffolds [6], and short unstruc-
recognition process utilizes the affinity of the recognition tured peptides [4,7]. Compared to more complex protein-based
element to the analyte and the interaction information is affinity scaffolds, short unstructured peptides have several potential
transmitted as a measurable signal (electrical, optical, etc.) by advantages that can be exploited for biosensor development: 1)
the transducer. The overall selectivity and the sensitivity of the peptides are stable and resistant to harsh environments, 2) peptides
biosensor are dependent on both the recognition element and can be synthesized easily and inexpensively, and 3) peptides can be
the transducer. In this work, we demonstrate a general pathway more amenable than antibodies to engineering at the molecular
for the development of new biosensors utilizing unstructured level [8,9,10]. In addition, the immobilization of short peptides on
peptides selected using M13 phage display as the recognition gold electrodes for use as the recognition element has been well
element, QCM as a diagnostic tool during development, and characterized since the early 80’s [11,12].
electrochemical techniques (CV, EIS) as the detection elements. Biopanning of phage displayed peptide libraries is a widely
This procedure is fast and can be applied to almost any desired utilized method that allows for the rapid selection of peptides that
protein target (Figure 1A). bind to desired protein targets. Several groups have reported
Immunosensors are commonly used biosensors that rely on biosensors where the entire phage particles from these selections
antibodies as the biomolecular recognition element and require (featuring multiple copies of the peptides) are employed as the
multi-step processing and labeling of the samples [3]. The sensing probes in the biosensors [13,14,15]. Although phage
widespread use of antibody-based immunoassays has been display has been widely applied to identify peptides or proteins

PLoS ONE | www.plosone.org 1 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

Figure 1. Schematic diagram illustrating the general process of biosensor development. (A) Work flow diagram for biosensor
development: 1. Target protein selection, 2. Phage display selection, 3. Peptide synthesis, 4. QCM diagnosis, 5. Biosensor detection. B) The basic
principle of QCM where the binding of the target protein to the immobilized peptides causes a frequency change in the oscillation of the quartz
crystal. C) The basic principle of EIS where the binding of the target protein to the immobilized peptides causes increased resistance to the reaction of
an added redox couple. See text for details.
doi:10.1371/journal.pone.0024948.g001

with selective binding capabilities, the application of free (non- Electrochemical techniques can also be used for the creation of
phage-bound) peptides in the development of biosensors has been label-free biosensors [21,22,23,24]. They have the potential to be
less frequently reported. Using free peptides can be advantageous inexpensive, fast responding, and low maintenance. The most
as this can simplify the electrochemical detection techniques. As well-known example of an electrochemical biosensor is the
an example of this approach, we have recently described a new commercial glucose sensor for diabetes [25,26], where the reaction
biosensor for the detection of troponin I (a cardiac biomarker) product of glucose oxidase with glucose converts ferricyanide to
using peptides isolated by M13 phage display [16,17]. the electroactive species ferrocyanide that is monitored electro-
The QCM has been widely used in biosensor development chemically. Other applications also benefit from the fact that
because it is a label-free technique [18,19,20]. The QCM is based biosensing electrodes can be easily integrated into microfabricated
on a piezoelectric material (quartz), where an alternating electrical systems and used as portable test tools for field measurements
field across the quartz creates an alternating shear motion of the [27,28].
crystal. Through appropriate circuitry, the change in the Cyclic voltammetry (CV) and electrochemical impedance
resonance frequency, Df, of the crystal is tracked in real time. At spectroscopy (EIS) are two frequently used techniques in these
constant temperature, this resonant frequency responds primarily applications. Both techniques measure either the current or the
to the change in mass associated with the crystal surface, Dm, and impedance obtained by varying the applied potential (voltage).
changes in the viscoelastic properties of the fluid adjacent to the When applied here, the method detects binding of an analyte on
crystal (Figure 1B). The relationship between Df and Dm is linear the surface of an electrode by the amount it hinders the access of
according to the Sauerbrey equation [16]. an electroactive redox couple (ferri/ferrocyanide, in this case) to

PLoS ONE | www.plosone.org 2 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

the electrode, therefore decreasing the redox current in CV or Biopanning of Phage-Displayed Peptide Library
increasing impedance in EIS. CV provides rapid evidence for ALT (2.6 mg/mL) was dissolved in NaHCO3 buffer (100 mM,
binding but it is less quantitative. In EIS, the sinusoidal current pH 8.3) and transferred to the wells of polystyrene microplates.
obtained in response to small amplitude, sinusoidal perturbations After overnight incubation with mild agitation at 4uC, the well
of the potential is measured. The current response has an surfaces were coated with ALT. The well surfaces were then
amplitude relative to the potential amplitude and is also out of blocked with blocking buffer (0.1 M NaHCO3 containing 2% BSA
phase from the perturbation. For this reason, results are reported and 0.02% of NaN3) for 1 hr at 4uC, and washed 6 times with
as complex impedance. It is common to plot EIS data in a Nyquist TBST (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Tween 20)
plot, and fit them using an equivalent circuit model to extract the to remove weakly bound ALT. The Ph.D.-12 phage displayed
parameter of interest. Here it is the resistance of charge transfer of random peptide library (1.561011 pfu) in 100 mL of TBS buffer
the redox couple, which is impacted by bound analyte (Figure 1C). (50 mM Tris-HCl pH 7.5, 150 mM NaCl) was added to the ALT
In this paper, we describe a novel biosensor for the detection of coated wells, and the plate was shaken gently for 1 h at room
alanine aminotransferase (ALT, EC 2.6.1.2, also called glutamate temperature. Wells were washed 10 times with TBST to remove
pyruvate transaminase, GPT). ALT catalyzes the reversible unbound phage. The bound phage were eluted with 100 mL of
transamination of L-alanine and a-ketoglutarate to pyruvate and 0.2 M glycine-HCl (pH 2.2) and the elution was immediately
L-glutamate with pyridoxal 59-phosphate as a coenzyme. ALT is neutralized with 15 mL Tris-HCl (1 M, pH 9.1) to prevent phage
found mainly in the liver, but is also found in red blood cells, heart killing. The eluted phage were amplified in E. coli ER2738 and the
cells, muscle tissue and other organs, such as the pancreas and phage were harvested by precipitation with NaCl/polyethylene
kidneys [29,30]. Serum ALT levels are an indicator for liver glycol (20% (w/v) polyethylene glycol-8000 with 2.5 M NaCl).
damage and its detection is considered the gold standard This biopanning process was repeated for five rounds. The
biomarker of hepatotoxicity [30,31,32]. The normal concentration concentration of Tween 20 in the washing step was increased from
of ALT in blood serum ranges from 5–35 U/L (0.1–0.7 mg/mL), 0.1% in the first round of panning to 0.3% in the second round of
[29] and serum ALT levels increase up to 50-fold in connec- panning and 0.5% in the final three rounds of panning. After each
tion with a variety of liver conditions, including viral infection, round of panning, the phage were titered using Luria-Bertani (LB)
cirrhosis, non-alcoholic steatohepatitis, and drug toxicity [31,33, broth plates containing isopropyl b-D-thiogalactopyranoside
34]. Most ALT biosensors are based on the detection of the (IPTG) and X-gal. After the fifth round of selection, twelve blue
enzymatic activity of the ALT enzyme as opposed to the detection monocolonies were randomly picked and amplified for DNA
of the protein itself [29,35,36,37]. sequencing.
In this article we present the selection of ALT binding peptides
from an M13 phage displayed peptide library and the character- ELISA for Peptide Binding to ALT
ization of the binding affinities of the peptides. Once the most
To test whether or not the selected phage clones from the
promising peptide was selected, a cysteine-modified free peptide
biopanning procedure could bind to ALT, ELISA assays were
was synthesized and transferred to the biosensor platform. The
performed. Microplates were coated with ALT (2.6 mg/mL)
binding activity was monitored in situ by QCM, and the peptide-
overnight at 4uC, blocked with blocking buffer, and washed 6
modified gold electrode was used to detect ALT quantitatively
times with TBST solution (containing 0.1% Tween 20). Varying
using EIS. The general approach (Figure 1) can be extended to
amounts of amplified phage clones were added to each well and
develop biosensors for a wide variety of target analytes. The
incubated for 1 hr at room temperature. After washing, HRP-
methodology can be easily applied in a relatively short period of
conjugated anti-M13 monoclonal antibody (diluted 1:5000 in
time at low cost.
blocking buffer) was added and incubated at room temperature for
1 hr. Following incubation and washing again with the same
Materials and Methods buffer, ABTS and H2O2 (30%, v/v) were added and the
Materials absorbance was measured after a 1 hr incubation with mild
agitation at 405 nm with a microplate spectrophotometer
All reagents used were analytical grade or higher unless otherwise
(Molecular Devices, Sunnyvale, CA).
stated. Escherichia coli strain ER2738, a host for M13 phage, and a
polyvalent M13 phage display kit (Ph.D.-12) which contained In order to estimate the apparent dissociation constant for the
random 12-mer peptides were obtained from New England Biolabs ALT5-8 clone, further ELISA experiments were performed with
(Ipswich, MA). ALT (human liver) and lactate dehydrogenase (LDH) extended phage concentrations. The data were fit to a Langmuir
were obtained from Lee Biosolutions, Inc. (St. Louis, Missouri). isotherm using non-linear regression (SigmaPlot, San Jose, CA)
Horseradish peroxidase (HRP) conjugated anti-M13 monoclonal and the Kd,app was determined.
antibody was from GE Healthcare (Piscataway, NJ). Tween 20, 2,29-
Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt Kinetic Studies for Measuring Peptide Inhibition of the
(ABTS), streptavidin (SA) from Streptomyces avidinii, bovine serum Enzyme
albumin (BSA), L-cysteine hydrochloride (98%) and all other ALT enzyme kinetics were measured to determine the impact of
chemicals were purchased from Sigma-Aldrich (St. Louis, MO). the ALT5-8 peptide on the activity of the enzyme. Kinetic
Polystyrene microplates were from Pierce Biotechnology (Rockford, experiments were performed in a coupled assay where the
IL). The ALT5-8 peptide modified with a C-terminal cysteine pyruvate generated by ALT was reduced to lactate by LDH with
(sequence: WHWRNPDFWYLKC) was synthesized (.90% purity) the concomitant oxidation of NADH. The depletion of NADH
by Genscript (Piscataway, NJ). Quartz crystals (AT-cut, 5 MHz) with was monitored by following its absorbance at 340 nm in UV-
Au deposited on both sides were obtained from Inficon (East transparent microplates. L-alanine and a-ketoglutarate concentra-
Syracuse, New York). Buffer solutions (pH = 7.3) containing 0.01 M tions were varied from 2.5 to 10 mM and 0.5 to 10 mM
sodium phosphate and 0.1 M NaCl were used to make all protein (1– respectively. Initial concentrations of pyridoxal 59-phosphate,
10 mg/mL) and peptide (0.1 mM) solutions for the QCM and EIS NADH, and LDH were 0.1 mM, 0.25 mM and 47 U/L
measurements. All solutions were prepared using Milli-Q water. respectively. The reactions were initiated by the addition of 330

PLoS ONE | www.plosone.org 3 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

U/L ALT and all measurements were made in triplicate. All stock viscosity differences between the sample solutions. CV experi-
solutions were prepared in 100 mM Tris-HCl buffer (pH 7.8), and ments were performed at a potential scan rate of 100 mV/s.
all kinetic trials were conducted 300 mL at 37 uC. Impedance spectra were recorded over a frequency range of 0.1–
The kinetic rates were fit to the bi-bi ping-pong mechanism 105 Hz. A single sinusoidal AC voltage of 10 mV was
(Eqn. 1), to obtain estimates for kcat and the Michaelis constants superimposed on the open-circuit potential (typically 0.2 V vs.
(KM) for L-alanine (A) and a-ketoglutarate (B). Ag/AgCl). The impedance was recorded and a Nyquist plot was
obtained. A Randles circuit was used to fit the semi circular curves
Et kcat ½A"½B" in order to calculate the charge transfer resistance (RCT). The
Rate~ ð1Þ change percentage of RCT is defined as
½A"½B"zKM,A ½B"zKM,B ½A"

RCT ALT{RCT ref

DRCT (%)~ |100%
In order to determine if the peptide inhibited the enzyme as well RCT ref
as the nature of the inhibition, the kinetics were repeated with
varying L-alanine concentrations at a constant concentration of a- where RCT ref and RCT ALT is the charge transfer resistance before
ketoglutarate (1 mM). Free ALT5-8 peptide inhibitor (I) was added and after ALT binding.
to the reactions at concentrations from 0–0.25 mM. The rates
plotted on a Lineweaver-Burk plot to understand the mode of Results
inhibition and the rates were also fit to the bi-bi ping-pong
equation with a competitive inhibitor (Eqn. 2): Selection and characterization of an ALT-binding peptide
A large commercially available phage display library was
biopanned over the immobilized ALT target protein. The yield of
Et kcat ½A"½B"
Rate~ ! " ð2Þ phage binding to the plate increased each round after the second
I (Table S1), even as the stringency of the selections was increased by
½A"½B"zKM,A ½B" 1z zKM,B ½A"
KI the addition of Tween 20. Twelve phage clones randomly selected
after the fifth round of selection were sequenced and the library was
found to have converged (10/12) to one predominant sequence,
ALT5-8, with a sequence of WHWRNPDFWYLK (Table 1). This
QCM measurements clone along with two other sequences obtained in the sequencing
The QCM system consisted of a phase lock oscillator (PLO-10i, (ALT4-3 and ALT5-9) were investigated using an ELISA assay to
Maxtek, East Syracuse, New York) and a frequency counter assess the affinity of the selected phage clones for immobilized ALT.
(53131A, Agilent Technologies, Santa Clara, CA). The operation The ALT4-3 and ALT5-9 clones were found to have a low affinity
of the setup was described previously [16]. Typically, a stable for ALT similar to that of the M13 control phage, However, the
baseline was achieved 1 hr after flow injection on a clean crystal at ALT5-8 clone exhibited a dose-dependent affinity for the ALT
a flow rate of 50 mL/min. After a stable baseline was obtained in protein (Figure 2) and was selected for further characterization.
buffer, a peptide solution in the buffer was flowed to the cell. Further ELISA experiments were performed with the ALT5-8
Usually an abrupt change in the signal occurs with the valve clone in order to determine the apparent affinity of the phage
switching. In about 5 min, the new solution arrived at the cell and particles for the ALT protein. A fit of this data to the Langmuir
caused a frequency change due to peptide binding to the crystal isotherm (Figure 3) resulted in an apparent dissociation constant,
surface. After the binding reached a steady state, buffer rinse was Kd,app of 85620 nM. Since the phage particles have approx-
then applied and no further change in frequency was observed. imately 5 copies of the peptide per particle, it is possible that the
Similar operations were performed for the addition of the cysteine true affinity of the free peptide for the ALT protein will be
and ALT solutions, as well as solutions of other protein different from the apparent value due to the loss of the avidity
competitors such as BSA and SA. All experiments were repeated effect.
three times and similar results were obtained.
Measurement of peptide inhibition constant, KI
Electrochemical measurements The ALT target protein has a large surface area for peptide
All electrochemical measurements were performed with a binding and the binding of the peptide in or near the active site of
mAUTOLAB potentiostat (Type III, FRA2, Metrohm Autolab, the enzyme could possibly affect the catalytic properties of the
Netherlands). A gold wire (dia. 1 mm) sealed in epoxy was used as enzyme. Therefore, the impact of the peptide on the kinetics of the
a working electrode and a platinum mesh was used as a counter ALT enzyme was investigated. In the absence of the peptide, the
electrode. All potentials measured are reported versus a Ag/AgCl kinetic data fit well to the bi-bi ping-pong equation (Eqn. 1) and the
reference electrode. Both cyclic voltammetry and EIS measure- following kinetic parameters were obtained: kcat = 5.860.2 s21,
ments were conducted in a solution of 1 mM ferro/ferricyanide in
0.1 M sodium perchlorate.
CV and EIS were performed as previously described [16]. Table 1. Sequences of the 12 selected phage clones obtained
Briefly, the gold electrodes were prepared in 0.1 mM peptide following the fifth round of biopanning against ALT.
solution for 1 hr, followed by backfilling in 1 mM cysteine solution
for 2 hr to obtain a Au-Pep-Cys electrode. The Au-Pep-Cys
Name Amino acid sequence Frequency
electrodes were incubated in ALT solution for 1 hr before CV and
EIS measurements. After each electrode preparation step, the ALT4-3 SNNLYPQRAVST 1/12
electrode is removed and placed into solutions of 1 mM ferro/ ALT5-8 WHWRNPDFWYLK 10/12
ferricyanide in 0.1 M NaClO4for analysis. Therefore, all the ALT5-9 LETEWDSLWYAP 1/12
electrochemical measurements were conducted in the ferro/
ferricyanide solution which eliminates the effect of any potential doi:10.1371/journal.pone.0024948.t001

PLoS ONE | www.plosone.org 4 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

Figure 4. Lineweaver-Burk plots for the determination the of

the ALT enzyme inhibition by the ALT5-8 peptide. Varying
concentrations of peptide [I] were added as a function of initial ALT
concentration [E]. All data were collected in triplicate and R2 values are
Figure 2. Characterization of the selected ALT binding phage shown for the best fit line for each inhibitor concentration. A KI value of
clones using ELISA. ALT binding of the selected phage clones ALT5-8 71617 nM was obtained by fitting the entire data set simultaneously to
(&), ALT4-3 (%), ALT5-9 (m) and M13 control phage (n) with varying the bi-bi ping pong rate equation with competitive inhibition (Eqn. 2).
concentrations of phage particles ranging from 105 pfu to 1011 pfu. All doi:10.1371/journal.pone.0024948.g004
measurements were performed in triplicate and error bars represent
standard deviations.
doi:10.1371/journal.pone.0024948.g002
Sensor preparation
In order to use the phage selected ALT peptide in the
KM,alanine = 8.060.9 mM, and KM,ketoglutarate = 0.09060.030 mM development of an ALT biosensor, the peptide was modified with
(Figure S1) which are in agreement with previously published values a C-terminal cysteine for immobilization on gold. The preparation
for the ALT enzyme [38,39,40,41]. of the sensor consisted of two steps: the immobilization of the
Lineweaver-Burk plots of pyruvate production at varying peptide and blocking of empty sites with free cysteine. Both steps
alanine concentrations are shown in Figure 4. The data suggests were monitored by QCM and electrochemical (CV and EIS)
that the ALT binding peptide is a competitive inhibitor for the measurements as shown in Figure 5. The immobilization of the
active site, with respect to alanine. The data were fit to the bi-bi peptide resulted in a frequency drop in the QCM signal (Figure 5A)
ping-pong mechanism with competitive inhibition (Eqn. 2) using due to mass loading, a current decrease in the CV (Figure 5B), and
the kcat and KM values obtained in the absence of the peptide, and an impedance increase in the EIS (Figure 5C) due to surface
the inhibition constant for the ALT5-8 peptide was found to be blocking. The abrupt changes in the QCM noted by arrows
KI = 71617 nM. (Figure 5A) are due to pressure changes caused by valve switching.
This phenomenon was observed in all the experiments. All the
experiments were repeated three times with similar results each
time. All three techniques show the successful immobilization of
the peptides on the gold surface. Backfilling of the surface cysteine
did not cause any mass change in QCM since both the amount
and molecular weight of the blocking cysteine is small. However,
the backfilling did result in a current increase and a prominent
decrease in the impedance signal (smaller semi-circles). A similar
effect was observed in our previous studies [16] and in studies of
other modified surfaces [42]. The impedance decrease upon
backfilling can be explained by the easier charge transfer through
tunneling across the shorter cysteine. The charge transfer between
Fe(CN)64-/3- and the electrode occurs either by tunneling of
electrons through the surface self-assembled monolayer (SAM)
(peptide or cysteine), or through defects (unblocked sites) on the
surface [43]. The packing density of the immobilized peptide can
be calculated using the Sauerbrey equation [16]:

Dfm ~{Cf |Dm ð3Þ

Figure 3. Determination of the apparent dissociation constant
of ALT5-8 for ALT. ELISA experiments were performed in triplicate at
eight different phage concentrations and the data were fit by a where Dfm is the frequency change due to mass loading, Dm is the
Langmuir isotherm (R2 = 0.88) in order to estimate the apparent mass change at the QCM crystal surface and Cf = 0.0566 Hz/(ng/
equilibrium dissociation constant (Kd,app = 85620 nM.). cm2) for a 5 MHz AT-cut crystal at 20uC. The frequency drop of
doi:10.1371/journal.pone.0024948.g003 the QCM due to peptide immobilization was about 21 Hz, which

PLoS ONE | www.plosone.org 5 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

corresponds to a packing density of 2610210 mol/cm2 or

1.261014 molecule/cm2 based on a peptide molecular weight of
1851 g/mol. The surface density of a similar sized troponin-
binding peptide immobilized on a gold surface has previously been
reported to be 7.461013 molecule/cm2 [16]. Inamori et al. report a
surface density of 1.861012 to 1.861013 molecules/cm2, depend-
ing on the attachment methods for an immobilized peptide on
gold [44]. The average cross-section of each peptide was
,0.8 nm2 based on the surface density, which indicates a dense
packing mode of the immobilized peptides. The subsequent buffer
rinse caused a small frequency increase (the vertical drop observed
at the injection of buffer was caused by the sudden pressure
change due to valve switching), possibly due to the removal of
loosely bound peptide or the viscosity/density change of solutions.

ALT sensor operation

QCM, CV and EIS measurements validated the formation of
the peptide-based sensing layer. The same techniques were used to
demonstrate the affinity of the immobilized peptide for ALT, as
shown in Figure 6. Upon binding of ALT by the immobilized
peptide, a frequency decrease (Df = -31 Hz) in QCM (Figure 6A),
a current decrease in CV (Figure 6B), and a large impedance
increase in EIS (Figure 6C) were observed. No change was
observed for the buffer control in all three cases. All three
techniques indicate binding of ALT to the immobilized peptide.
To begin to study the selectivity of the peptide, similar
experiments were performed with streptavidin (SA) and bovine
serum albumin (BSA). Figure 7 shows the frequency change of Au-
Pep-Cys binding to SA, BSA and ALT. A small frequency change
of about -3 Hz was observed upon addition of SA and BSA, while
a frequency change of -31 Hz was observed with ALT. The small
frequency change seen with the control proteins might be due to
nonspecific binding of SA and BSA with unoccupied sites on the
surface or non-specific interactions with the peptide.
Although all three techniques demonstrate a signal change upon
the binding of ALT, the CV technique is a poor candidate for
quantitative measurement since the detection range is limited by
the fact that the current decreases upon target binding. For QCM
and EIS, response curves over a range of ALT concentrations are
shown in Figure 8. Both QCM (Figure 8A) and EIS (Figure 8B)
show a quantitative response to a range of ALT concentrations
and thus can be used as a sensing technique in quantitative ALT
measurements. The LOD obtained by QCM is 60 ng/mL, while
EIS had a LOD of 92 ng/mL. The sensitivity of the QCM system
was 8.960.9 Hz/(mg/mL), while the sensitivity of the EIS system
was 142612 impedance percentage change %/(mg/mL).
Based on the QCM response curve, the kinetic binding
constants can be deduced. As described previously [16], the
equilibrium dissociation constant, Kd, can be obtained from a plot
of frequency change, Dfeq, vs. Dfeq/C. Here, a dissociation constant
of Kd = 2.01+0.06 mg/mL or 20.1+0.6 nM was calculated for
ALT binding on QCM.

Discussion
Figure 5. Sensor preparation. A) the change in frequency during the
immobilization of peptide (Pep, 0.1 mM), rinsing (Rinse) and subse- Our overall goal is to develop a streamlined platform for the
quent blocking with cysteine (Cys, 1 mM) on gold using QCM; B) CV of rapid development of new biosensors that can be used to detect
the bare gold electrode (Au), the electrode with the peptide (Au-Pep), virtually any desired protein target. Phage display is a commonly
and the blocked electrode (Au-Pep-Cys), C) EIS of bare gold (Au), the used method for identifying new binding motifs, and commercially
peptide modified electrode (Au-Pep) and the blocked electrode (Au- available kits make this process broadly accessible [4,7]. Biopan-
Pep-Cys) in a 1 mM solution of Fe(CN)64-/3- in 0.1 M NaClO4.
ning procedures are well documented, and new binding peptides
doi:10.1371/journal.pone.0024948.g005
can usually be identified in a short period of time (1–2 weeks).
There are several detection methods available that can be used to
create a biosensor from a recognition peptide. We have chosen

PLoS ONE | www.plosone.org 6 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

Figure 7. Specificity of the ALT biosensor. The shift in QCM

frequency in response to injections of SA, BSA or ALT on the Au-Pep-Cys
modified crystal. The concentrations of all three solutions were 10 mg/
mL.
doi:10.1371/journal.pone.0024948.g007
electrochemical techniques since they are generally simple,
inexpensive, and can be readily incorporated into microfluidic
devices. In this manuscript, we demonstrate this process from start
to finish to create a new biosensor for the detection of ALT.

Figure 6. Sensor operation. A) QCM frequency change of Au-Pep-

Cys electrode in Buffer alone and in 10 mg/mL ALT, B) CV of Au-Pep-Cys
electrode before (Baseline) and after Buffer alone or ALT binding, C) EIS
of Au-Pep-Cys before (Baseline) and after Buffer alone or ALT binding. Figure 8. Response curves of ALT binding. Measurements were
After incubation in Buffer or 10 mg/mL ALT, the electrodes were made using QCM (shown as absolute value of frequency change) (A)
transferred to 1 mM solution of Fe(CN)64-/3- in 0.1 M NaClO4 for the EIS and EIS (B). Each run was performed using a single electrode with
measurements. successive tests in ALT solutions from low to high concentration. The
doi:10.1371/journal.pone.0024948.g006 apparent amount of bound ALT (NALT) for the QCM measurements was
calculated using Eqn. 3 and is shown on the right ordinate. Error bars
represent standard deviations obtained from triplicate measurements.
doi:10.1371/journal.pone.0024948.g008

PLoS ONE | www.plosone.org 7 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

ALT is an enzyme, and one approach for its detection is to can require complex sample labeling and modification techniques
monitor its enzymatic activity [29,35,36]. This approach can be [45]. The unstructured peptides used in this approach can be
very sensitive, but it is not a general platform for biosensor easily selected against virtually any target and are inexpensive to
development, and this approach needs to be redeveloped for each synthesize, but they may suffer from poor selectivity in complex
new enzyme to be detected. We have chosen instead to directly solutions. Several transduction techniques are also available,
detect the presence of the protein using an evolved binding peptide, especially optical methods such as surface plasmon resonance
and in this way the detection method is generic and does not depend (SPR) [46] and Förster resonance energy transfer (FRET) [47], but
on the enzymatic activity of the target protein. The phage display EIS has the potential to be one of the most powerful and cost-
library converged to a new binding sequence after 5 rounds of effective tools for biosensor development [48].
biopanning (ALT5-8) and this peptide exhibited nanomolar affinity In summary, we have developed a general method for the
for ALT when displayed on phage particles. We further explored creation of new biosensors using short synthetic peptides obtained
whether the peptide was binding near the active site of the enzyme via biopanning of a phage displayed library. The whole process
through enzyme inhibition studies. The peptide was found to be a includes: target selection and immobilization, phage display to
competitive inhibitor for ALT with a nanomolar inhibition select binding peptides, peptide synthesis with a terminal thiol,
constant. However, the enzyme was still quite active even when QCM in-situ monitoring of peptide immobilization, and sensor
the peptide to protein ratio was at 2:1. Therefore, it is possible that detection using electrochemical techniques (Figure 1). We have
the enzyme is still active in the bound state, and future work will be demonstrated this approach by creating a new biosensor for a well-
necessary to determine if a biosensor can be created based on known biomarker for hepatocellular toxicity, ALT. The new
detecting the bound ALT activity instead of monitoring the mass biosensor had a LOD value just below the physiological
change that occurs upon binding of the enzyme. concentration of the target protein in human blood. Since both
Another concern with this approach is that peptides identified phage display and EIS are widely used techniques, this general
using phage display may lose affinity for their target when they are approach to biosensor development is straightforward and can be
displayed on a different platform. There are, on average, 5 copies readily applied for the development of biosensors to detect almost
of the ALT5-8 peptide on the surface of the M13 phage, and this any desired protein target.
avidity effect gives rise to an apparent nanomolar dissociation
constant. This proved not be a significant problem as, when Supporting Information
peptides were immobilized on a solid surface, a true dissociation
constant for the ALT binding peptide remained in the nanomolar Figure S1 Uninhibited ALT enzyme kinetic data. Kinetic
scale range. experiments were performed in a coupled assay where the
As we previously found with Troponin I, EIS appears to be the pyruvate generated by the ALT reaction was reduced to lactate
optimal method for the electrochemical detection of the bound by LDH with the concomitant oxidation of NADH. Varying
target [16]. The procedure is straightforward, and high sensitivities concentrations of L-alanine (A) and a-ketoglutarate (B) were used,
and low LOD values are obtained. The only downside to the EIS and the data were fit to the bi-bi ping-pong mechanism (Eqn. 1)
technique is the need to measure the impedance signal in the using non-linear regression software (SigmaPlot). The Michaelis
presence of a redox couple after the sample has been exposed to constants for L-alanine and a-ketoglutarate were determined to be
the peptide sensing layer. But, the EIS technique produces a larger 8.060.9 mM and 0.09060.030 mM respectively with
linear sensing range than can be obtained with QCM and it is kcat = 5.860.2 s21. All data were collected in triplicate and error
more robust than QCM as it is less sensitive to environmental bars represent standard errors.
perturbations. (TIF)
The new ALT biosensor made from the ALT5-8 peptide Table S1 Enrichments obtained by selecting the phage
coupled with EIS detection had a LOD (0.092 mg/mL) close to the displayed peptide library over immobilized ALT.
normal level of ALT in human blood (0.1–0.7 mg/mL) [29]. (DOCX)
Although we have demonstrated that the ALT5-8 peptide is
selective for ALT as compared to BSA and SA, further work will
be required to explore the performance of the biosensor in real Acknowledgments
samples such as tissue culture media or plasma, and in the The authors would like to thank Ms. Victoria Sun for assistance with the
presence of other potential interferents. collection of the enzyme kinetic data. We also thank Mr. Oren Shur, Dr.
Although this approach will require further development before Elliot Campbell, and Ms. Asli Sahin for thoughtful reviews of the
it can be used to create a biosensor for clinical use, the use of manuscript.
unstructured peptides obtained via phage display coupled with
electrochemical detection methods has several advantages com- Author Contributions
pared to other biosensor platforms described in the literature. Conceived and designed the experiments: JW JPP KD DMC ACW SB.
Alternative recognition elements such as antibodies are time Performed the experiments: JW JPP KD. Analyzed the data: JW JPP KD
consuming and expensive to produce, and aptamer-based sensors DMC ACW SB. Wrote the paper: JW JPP KD DMC ACW SB.

References
1. Newman JD, Setford SJ (2006) Enzymatic biosensors. Mol Biotechnol 32: 6. Binz HK, Amstutz P, Pluckthun A (2005) Engineering novel binding proteins
249–268. from nonimmunoglobulin domains. Nat Biotechnol 23: 1257–1268.
2. Ronkainen NJ, Halsall HB, Heineman WR (2010) Electrochemical biosensors. 7. Nilsson F, Tarli L, Viti F, Neri D (2000) The use of phage display for the
Chem Soc Rev 39: 1747–1763. development of tumour targeting agents. Adv Drug Deliv Rev 43: 165–196.
3. Luppa PB, Sokoll LJ, Chan DW (2001) Immunosensors - principles and 8. Dover JE, Hwang GM, Mullen EH, Prorok BC, Suh SJ (2009) Recent advances
applications to clinical chemistry. Clinica Chimica Acta 314: 1–26. in peptide probe-based biosensors for detection of infectious agents. Journal of
4. Kehoe JW, Kay BK (2005) Filamentous Phage Display in the New Millennium. Microbiological Methods 78: 10–19.
Chemical Reviews 105: 4056–4072. 9. Iqbal SS, Mayo MW, Bruno JG, Bronk BV, Batt CA, et al. (2000) A review of
5. Cho EJ, Lee JW, Ellington AD (2009) Applications of aptamers as sensors. Annu molecular recognition technologies for detection of biological threat agents.
Rev Anal Chem (Palo Alto Calif) 2: 241–264. Biosensors & Bioelectronics 15: 549–578.

PLoS ONE | www.plosone.org 8 October 2011 | Volume 6 | Issue 10 | e24948

 Peptide-Based Biosensors

10. Luong JHT, Male KB, Glennon JD (2008) Biosensor technology: Technology 30. Amacher DE (2002) A toxicologist’s guide to biomarkers of hepatic response.
push versus market pull. Biotechnology Advances 26: 492–500. Hum Exp Toxicol 21: 253–262.
11. Barker PD, Digleria K, Hill HAO, Lowe VJ (1990) Electron-Transfer Reactions 31. Ozer J, Ratner M, Shaw M, Bailey W, Schomaker S (2008) The current state of
of Metalloproteins at Peptide-Modified Gold Electrodes. European Journal of serum biomarkers of hepatotoxicity. Toxicology 245: 194–205.
Biochemistry 190: 171–175. 32. Kaplan MM (2002) Alanine aminotransferase levels: what’s normal? Ann Intern
12. Digleria K, Hill HAO, Lowe VJ, Page DJ (1986) Direct Electrochemistry of Med 137: 49–51.
Horse-Heart Cytochrome-C at Amino Acid-Modified Gold Electrodes. Journal 33. Amacher DE (2002) A toxicologist’s guide to biomarkers of hepatic response.
of Electroanalytical Chemistry 213: 333–338. Human and Experimental Toxicology 21: 253–262.
13. Archer MJ, Liu JL (2009) Bacteriophage T4 Nanoparticles as Materials in 34. Amacher DE (1998) Serum Transaminase Elevations as Indicators of Hepatic
Sensor Applications: Variables That Influence Their Organization and Injury Following the Administration of Drugs. Regulatory Toxicology and
Assembly on Surfaces. Sensors 9: 6298–6311. Pharmacology 27: 119–130.
14. Liu FF, Luo ZF, Ding X, Zhu SG, Yu XL (2009) Phage-displayed protein chip 35. Chang KS, Chang CK, Chou SF, Chen CY (2007) Sequential measurement of
based on SPR sensing. Sensors and Actuators B-Chemical 136: 133–137. aminotransferase activities by amperometric biosensors. Biosensors & Bioelec-
15. Nanduri V, Sorokulova IB, Samoylov AM, Simonian AL, Petrenko VA, et al. tronics 22: 2914–2920.
(2007) Phage as a molecular recognition element in biosensors immobilized by 36. Ohgami N, Upadhyay S, Kabata A, Morimoto K, Kusakabe H, et al. (2007)
physical adsorption. Biosensors & Bioelectronics 22: 986–992. Determination of the activities of glutamic oxaloacetic transaminase and
16. Wu J, Cropek DM, West AC, Banta S (2010) Development of a troponin I glutamic pyruvic transaminase in a microfluidic system. Biosensors &
biosensor using a peptide obtained through phage display. Anal Chem 82: Bioelectronics 22: 1330–1336.
8235–8243. 37. Xuan GS, Oh SW, Choi EY (2003) Development of an electrochemical
17. Park JP, Cropek DM, Banta S (2010) High affinity peptides for the recognition of immunosensor for alanine aminotransferase. Biosensors & Bioelectronics 19:
the heart disease biomarker troponin I identified using phage display. Biotechnol 365–371.
Bioeng 105: 678–686. 38. Bergmeyer HU, Scheibe P, Wahlefeld AW (1978) Optimization of methods for
18. Marx KA (2007) The quartz crystal microbalance and the electrochemical aspartate aminotransferase and alanine aminotransferase. Clin Chem 24: 58–73.
QCM: applications to studies of thin polymer films, electron transfer systems, 39. Gruber W, Mollering H, Perras L (1977) Isolation of pH-optima and apparent
biological macromolecules, biosensors, and cells. Springer Series on Chemical Michaelis constants of highly purified enzymes from human and animal sources.
Sensors and Biosensors 5: 371–424. Comparison of enzymes of human and animal origin, I. J Clin Chem Clin
19. Laricchia-Robbio L, Revoltella RP (2004) Comparison between the surface Biochem 15: 565–573.
plasmon resonance (SPR) and the quartz crystal microbalance (QCM) method 40. Hafkenscheid JC, Dijt CC (1979) Determination of serum aminotransferases:
in a structural analysis of human endothelin-1. Biosensors & Bioelectronics 19: activation by pyridoxal-59-phosphate in relation to substrate concentration. Clin
1753–1758. Chem 25: 55–59.
20. Casero E, Vazquez L, Parra-Alfambra AM, Lorenzo E (2010) AFM, SECM and 41. Jung K, Grutzmann KD, Egger E, Fechner C (1977) Suitability of commercial
QCM as useful analytical tools in the characterization of enzyme-based enzyme control sera for the quality control of activity determinations of L-
bioanalytical platforms. Analyst 135: 1878–1903. aspartate aminotransferase and L-alanine aminotransferase in human serum.
21. Lisdat F, Schafer D (2008) The use of electrochemical impedance spectroscopy Clin Chim Acta 79: 515–526.
for biosensing. Analytical and Bioanalytical Chemistry 391: 1555–1567. 42. Mantzila AG, Strongylis C, Tsikaris V, Prodromidis MI (2007) Assessment of the
22. Sadik OA, Aluoch AO, Zhou A (2009) Status of biomolecular recognition using interaction between a synthetic epitope of troponin C and its specific antibody
electrochemical techniques. Biosens Bioelectron 24: 2749–2765. using a label-free faradaic impedimetric immunosensor and alpha-Keggin
23. Privett BJ, Shin JH, Schoenfisch MH (2008) Electrochemical sensors. Anal silicotungstic heteropolyacid as a redox probe. Biosensors & Bioelectronics 23:
Chem 80: 4499–4517. 362–369.
24. Chen XJ, West AC, Cropek DM, Banta S (2008) Detection of the superoxide 43. Shen Z, Stryker GA, Mernaugh RL, Yu L, Yan HP, et al. (2005) Single-chain
radical anion using various alkanethiol monolayers and immobilized cytochrome fragment variable antibody piezoimmunosensors. Analytical Chemistry 77:
C. Anal Chem 80: 9622–9629. 797–805.
25. Wang J (2001) Glucose biosensors: 40 years of advances and challenges. 44. Inamori K, Kyo M, Matsukawa K, Inoue Y, Sonoda T, et al. (2008) Optimal
Electroanalysis 13: 983–988. surface chemistry for peptide immobilization in on-chip phosphorylation
26. Wang J (2008) Electrochemical glucose biosensors. Chemical Reviews 108: analysis. Analytical Chemistry 80: 643–650.
814–825. 45. Sadik OA, Aluoch AO, Zhou AL (2009) Status of biomolecular recognition
27. Ceriotti L, Ponti J, Colpo P, Sabbioni E, Rossi F (2007) Assessment of using electrochemical techniques. Biosensors & Bioelectronics 24: 2749–2765.
cytotoxicity by impedance spectroscopy. Biosensors & Bioelectronics 22: 46. Homola J (2003) Present and future of surface plasmon resonance biosensors.
3057–3063. Analytical and Bioanalytical Chemistry 377: 528–539.
28. Curtis TM, Widder MW, Brennan LM, Schwager SJ, van der Schalie WH, et al. 47. Ai HW, Hazelwood KL, Davidson MW, Campbell RE (2008) Fluorescent
(2009) A portable cell-based impedance sensor for toxicity testing of drinking protein FRET pairs for ratiometric imaging of dual biosensors. Nature Methods
water. Lab on a Chip 9: 2176–2183. 5: 401–403.
29. Huang XJ, Choi YK, Im HS, Yarimaga O, Yoon E, et al. (2006) Aspartate 48. Kara P, de la Escosura-Muniz A, Maltez-da Costa M, Guix M, Ozsoz M, et al.
aminotransferase (AST/GOT) and alanine aminotransferase (ALT/GPT) (2010) Aptamers based electrochemical biosensor for protein detection using
detection techniques. Sensors 6: 756–782. carbon nanotubes platforms. Biosensors & Bioelectronics 26: 1715–1718.

PLoS ONE | www.plosone.org 9 October 2011 | Volume 6 | Issue 10 | e24948