Identification of sequence-structure RNA binding motifs for SELEX-derived aptamers.

Abstract

MOTIVATION: Systematic Evolution of Ligands by EXponential Enrichment (SELEX) represents a state-of-the-art technology to isolate single-stranded (ribo)nucleic acid fragments, named aptamers, which bind to a molecule (or molecules) of interest via specific structural regions induced by their sequence-dependent fold. This powerful method has applications in designing protein inhibitors, molecular detection systems, therapeutic drugs and antibody replacement among others. However, full understanding and consequently optimal utilization of the process has lagged behind its wide application due to the lack of dedicated computational approaches. At the same time, the combination of SELEX with novel sequencing technologies is beginning to provide the data that will allow the examination of a variety of properties of the selection process. RESULTS: To close this gap we developed, Aptamotif, a computational method for the identification of sequence-structure motifs in SELEX-derived aptamers. To increase the chances of identifying functional motifs, Aptamotif uses an ensemble-based approach. We validated the method using two published aptamer datasets containing experimentally determined motifs of increasing complexity. We were able to recreate the author's findings to a high degree, thus proving the capability of our approach to identify binding motifs in SELEX data. Additionally, using our new experimental dataset, we illustrate the application of Aptamotif to elucidate several properties of the selection process.

Full text

Page 1

Aptamers as a Sensitive Tool to Detect Subtle
Modifications in Therapeutic Proteins
Ran Zichel1¤, Wanida Chearwae1, Gouri Shankar Pandey1, Basil Golding2, Zuben E. Sauna1*
1Laboratory of Hemostasis, Division of Hematology, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, Maryland, United States of
America, 2Laboratory of Plasma Derivatives, Division of Hematology, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, Maryland,
United States of America
Abstract
Therapeutic proteins are derived from complex expression/production systems, which can result in minor conformational
changes due to preferential codon usage in different organisms, post-translational modifications, etc. Subtle conformational
differences are often undetectable by bioanalytical methods but can sometimes profoundly impact the safety, efficacy and
stability of products. Numerous bioanalytical methods exist to characterize the primary structure of proteins, post
translational modifications; protein-substrate/protein/protein interactions and functional bioassays are available for most
proteins that are developed as products. There are however few analytical techniques to detect changes in the tertiary
structure of proteins suitable for use during drug development and quality control. For example, x-ray crystallography and
NMR are impractical for routine use and do not capture the heterogeneity of the product. Conformation-sensitive antibodies
can be used to map proteins. However the development of antibodies to represent sufficient epitopes can be challenging.
Other limitations of antibodies include limited supply, high costs, heterogeneity and batch to batch variations in titer. Here
we provide proof-of-principle that DNA aptamers to thrombin can be used as surrogate antibodies to characterize
conformational changes. We show that aptamers can be used in assays using either an ELISA or a label-free platform to
characterize different thrombin products. In addition we replicated a heat-treatment procedure that has previously been
shown to not affect protein activity but can result in conformational changes that have serious adverse consequences. We
demonstrate that a panel of aptamers (but not an antibody) can detect changes in the proteins even when specific activity
is unaffected. Our results indicate a novel approach to monitor even small changes in the conformation of proteins which
can be used in a routine drug-development and quality control setting. The technique can provide an early warning of
structural changes during the manufacturing process that could have consequential outcomes downstream.
Citation: Zichel R, Chearwae W, Pandey GS, Golding B, Sauna ZE (2012) Aptamers as a Sensitive Tool to Detect Subtle Modifications in Therapeutic Proteins. PLoS
ONE 7(2): e31948. doi:10.1371/journal.pone.0031948
Editor: John J. Rossi, Beckman Research Institute of the City of Hope, United States of America
Received September 9, 2011; Accepted January 16, 2012; Published February 27, 2012
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 work was supported by funds from the Laboratory of Hemostasis and the Center for Biologics Evaluation and Research, Food and Drug
Administration’s Modernization of Science program (ZES). 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: zuben.sauna@fda.hhs.gov
¤ Current address: Department of Biotechnology, Israel Institute for Biological Research, Ness-ziona, Israel
Introduction
Therapeutic proteins now represent a significant segment of the
pharmaceutical industry [1] and include some of the most
innovative products which are on the cutting edge of clinical
care. This class of therapeutics is clearly different from small
molecule synthetic drugs. Protein-drugs are 100 to 1,000 times
larger, have complex secondary and tertiary structures, and cannot
be synthesized by chemical processes and have to be manufactured
in living cells. Compared to small-molecule entities, the manufac-
ture of biopharmaceuticals involves far larger numbers of batch
records (.250 vs. ,10), product quality tests (.2,000 vs. ,100),
critical process steps (.5,000 vs. ,100), and process data entries
(.60,000 vs. ,4,000) [2]. Analytical testing is an indispensible
part of the pre-clinical development as well as the routine
manufacture of any pharmaceutical and recent trends make such
testing even more critical. The globalization of the industry means
that the different steps in the manufacture of a single product
occur at several locations and even in different countries. This
poses challenges in quality control and has seen the emergence
(and adoption) of approaches like quality by design [3], which rely
heavily on exhaustive analytical testing. The lack of analytical
techniques to comprehensively characterize large molecule
biotherapeutics also lay at the heart of the debate on whether or
not to permit the development and licensure of biosimilars [4].
Legislative authorities in Europe [5] as well as the US [6] have now
ratified pathways for the approval of biosimilars. An examination
of the EMA experience [7] shows that the paradigm used for small
molecule generics cannot be used for biosimilars. The classical
generic approach worked well for chemically derived products
because characterization by analytical methods was determined to
be a good predictor of the biological and clinical properties of the
drug. The lack of suitable techniques to effectively compare the
biosimilar with the reference product necessitates more extensive
clinical trials than would otherwise be warranted [8].
Table 1 lists the techniques used to characterize protein
therapeutics. A number of technologies can be used to
characterize lower levels of protein organization (such as the
PLoS ONE | www.plosone.org1 February 2012 | Volume 7 | Issue 2 | e31948

Page 2

primary and secondary structures) with a high degree of accuracy
and sensitivity. Significant improvements in mass spectrometry
over the last decade allow the determination of variations in post-
translational modifications. Similarly the biophysicist’s toolkit
offers a choice of technologies to quantify protein-protein and
protein-substrate interactions. Moreover robust assays to measure
the biochemical activity of most protein products have been well
established. A significant gap, however, remains in monitoring the
tertiary and quaternary structures of proteins during drug
development and manufacture [9]. Techniques currently used to
determine the structures of proteins such as X-ray crystallography
and NMR do not lend themselves to routine use during the
manufacturing process. Issues of cost, time and technical skills
aside these techniques fail to capture the heterogeneity of the
product, which is a hallmark of biotherapeutics. It has, for
example, been estimated that as many as 108possible product-
variants and impurities for a monoclonal antibody product can
occur [2].
Epitope mapping using antibodies is one of the few methods
currently available to monitor the conformation of a protein
product [10–12]. In recent years synthetic nucleic acid reagents,
called aptamers have emerged as surrogates to antibodies and
appear to be particularly suited for bioanalytical applications [13].
Aptamers are engineered through repeated rounds of in vitro
selection, SELEX (systematic evolution of ligands by exponential
enrichment) to bind virtually any molecular target such as small
molecules, proteins, nucleic acids, and even cells, tissues and
organisms. The in vitro selection process confers significant
advantages to aptamers over antibodies. The three topical
thrombins currently marketed in the US were not developed as
biosimilars, nonetheless these products are used for the same
indications and are generally regarded as interchangeable in a
clinical setting [14]. However, these products are derived from
different sources and have different manufacturing processes [15]
and offer a good real life model system to evaluate the panel of
aptamers for discriminating between different potential confor-
mations of similar proteins. Here we present proof-of-principle
that aptamers can be used to characterize conformational epitopes
on therapeutic proteins. We have used six anti-human thrombin
aptamers to characterize epitopes on three thrombin products
currently marketed. The aptamers clearly distinguish between the
human and bovine thrombin products. However the two human
thrombin products (one purified from human plasma and the
other manufactured by recombinant DNA technology) show
comparable binding to all six aptamers. We also heat treated the
human thrombin products, using a procedure similar to that
resulting in enhanced immunogenicity in Factor VIII (FVIII)
[16,17]. This treatment did not affect thrombin activity or the
binding to an anti-thrombin antibody. The panel of anti-thrombin
aptamers however detected significant alterations in the confor-
mation of thrombin epitopes.
Results
Using aptamers as surrogate antibodies in an ELISA assay
We identified six anti-thrombin DNA aptamers previously
described to bind different thrombin epitopes with varying
affinities [18,19]. The affinities of these anti-thrombin aptamers
for thrombin have been determined using different assays and
there have been some discrepancies (see [18–20]). For consistency
of results it is very important that the conditions for the generation
of aptamers be rigorously adhered to. Moreover the assay we
developed uses biotinylated aptamers, which could affect the
affinity. The primary sequences of the six polynucleotides used in
this study are depicted in Table 2 and the conditions for
generating the aptamers are given in the Materials and Methods.
The aptamers were first evaluated using a sandwich ELISA. This
assay was standardized using an anti-thrombin monoclonal
capture antibody and a biotinylated polyclonal anti-thrombin
detector antibody (Fig. 1A). In modifying this assay, we used the
biotinylated anti-thrombin aptamers in lieu of the detector
antibody (Fig. 1B). Both assays use the same monoclonal antibody
for capture and identical reagents, NutraAvidin-HRP and TMB.
Using recombinant thrombin expressed in CHO cells (Reco-
throm), we demonstrate that the limit of detection for thrombin is
approximately 10 ng/ml using either the antibody or the aptamer
(Figs. 1 C&D). We therefore used a fixed concentration of the
thrombin (5 mg/ml) to determine the relative affinities of the six
aptamers used in this study (Fig. 2 and Table 3). The assay is
robust, reproducible and the six aptamers bind to the thrombin
with apparent Kd values ranging from 0.5860.01 nM to
2.7560.36 nM (Table 3). Prior to each experiment the oligonu-
cleotides were denatured and then allowed to refold under
controlled conditions (see Materials and Methods for details) to
form the aptamers. The SD thus represents not only the variation
in the assay per se but also the variability in process of generating
the aptamers.
Comparing the binding of aptamers to thrombin
products
There are three topical thrombin products in the market. Prior
to 2007 the only topical thrombin available was one derived from
bovine plasma (Thrombin JMI), since then a human-plasma
derived thrombin (Evithrom) and a human recombinant thrombin
(Recothrom) have been approved [14]. Bovine thrombin is
produced by extracting prothrombin from bovine plasma. After
activation to thrombin, the product undergoes a chromatographic
Table 1. Analytical techniques used to characterize protein therapeutics.
Composition, primary
structure
Peptide mapping (mass spectroscopy), peptide mass fingerprint (mass spectroscopy), amino
acid sequencing
Higher-order structureSpectroscopy, thermal stability testing
Conformation ELISA, epitope mapping
Post-translational modifications (e.g., glycosylation)Mass spectroscopy, ion-exchange
chromatography
Size, detection of aggregates Gel electrophoresis, size-exclusion chromatography, analytical ultracentrifugation
BindingCell assays, spectroscopy, SPR, ELISA
Biological activityCell assays, animal models
doi:10.1371/journal.pone.0031948.t001
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org2 February 2012 | Volume 7 | Issue 2 | e31948

Page 3

purification process that includes ion exchange and viral filtration
[21]. Human thrombin is derived from human plasma [22] while
human recombinant thrombin is produced via recombinant DNA
technology from genetically modified chinese hamster ovary
(CHO) cells [23]. Studies have indicated that the efficacies of
the human thrombins are comparable to that of the bovine
thrombin which has been used in the clinic for decades [24,25].
Fig. 3A shows a colloidal blue stained gel with the three thrombin
products. All three products are highly purified, however the
major protein component of the plasma derived human product is
human serum albumin (HSA) which is used as a stabilizing agent
[22]. No proteins are added to the bovine thrombin or the human
recombinant thrombin drug-products [21,23]. Despite these
differences in manufacturing processes and excipients, the binding
of the Aptamer 30-8 to the two human thrombin products is
almost indistinguishable (Fig. 3B). Furthermore when we com-
pared the kinetics of the binding of all six aptamers to the two
human thrombin products we found an extremely high correlation
(r=0.996, P,0.05) (Fig. 3C). Interestingly none of the six
aptamers binds to the bovine thrombin (data not given).
Aptamers can detect subtle conformational changes in
human thrombin products
There are many instances where a small change in manufac-
turing process significantly impacts a protein therapeutic (see [2]
for numerous examples). For example, a manufacturing change
involving heat treatment of FVIII resulted in a significant increase
Figure 1. ELISA based binding assays using aptamers as surrogate antibodies. (A & B) Illustrations showing assay platforms using either
polyclonal antibody (A) or aptamers (B) to detect thrombin. Both assays use an identical anti-thrombin monolonal antibody as the capture antibody
and identical reagents for detection (NutrAvidin-HRP and TMB). Assays used in Fig. 6 however use a polyclonal antibody as the capture antibody, one
of four monoclonal antibodies as detectors and anti-mouse-HRP as a secondary antibody (C) ELISA assay showing the limit of detection for purified
human thrombin using the antibody based assay illustrated in (A). (D) ELISA assay showing the limit of detection for purified human thrombin using
the aptamer based assay illustrated in (B).
doi:10.1371/journal.pone.0031948.g001
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org3 February 2012 | Volume 7 | Issue 2 | e31948

Page 4

in the frequency of inhibitors even though other product
parameters (such as specific activity) remained unaffected [26]. It
was subsequently determined that the heat treatment likely
resulted in subtle alterations in the conformation leading to the
exposure of some epitopes [17]. We therefore investigated whether
a comparable treatment of the human thrombins would affect the
binding of the panel of aptamers used in this study. Evithrom and
Recothrom were incubated at 60uC for a maximum of 17 h in the
presence of 60% sucrose buffer as has been described previously
(see Materials and Methods and [16] for details). Consistent with
previous studies [16] we found that heating in the presence of 60%
sucrose afforded protection and there was no significant effect on
the specific activity of either product (Fig. 4A). On the other hand
incubation at 60uC in PBS alone resulted in both protein
degradation (Fig. 4B) as well as complete abrogation of thrombin
activity (data not given). Similarly the antibody based binding
assay (see Fig. 1A for assay format) showed no detectable
difference in the binding of the polyclonal antibody, to thrombin
following up to 17 h of incubation at 60uC in the presence of 60%
sucrose buffer (Fig. 4C). On the other hand the positive control
(heat treatment in the presence of PBS alone) results in antibody
binding being almost complete abolished. The aptamer based
assay shows time-dependent changes in the binding kinetics when
the thrombin is incubated in the presence of sucrose (Fig. 4D). We
compared the effect of heat treatment of both the plasma derived
and recombinant thrombin on the binding of all six aptamers
(Fig. 5). The binding of five out of six aptamers showed a
statistically significant decrease in binding compared to untreated
Figure 2. Affinities of the six aptamers used in this study determined using the aptamer based ELISA. A fixed concentration of thrombin
(Evithrom), 150 nM was captured by the anti-thrombin monoclonal antibody, 5 mg/ml. Binding of increasing concentrations of the six anti-thrombin
aptamers (A, Aptamer 30-14, B, Aptamer 30-8, C, Aptamer 60-18/38, D, Aptamer 30-16/27, E, Aptamer 30-38/27 and F, Aptamer G15D) was
determined using the assay depicted in Fig. 1B. Data presented shows the mean (6 SD) of three experiments where the aptamers were generated
independently for each experiment.
doi:10.1371/journal.pone.0031948.g002
Table 2. Sequences of aptamers used in this study.
AptamerSequence
30-1459AGATGCCTGTCGAGCATGCTGTTGTGGTAGGGTTAGGGATGGTAGCGGTTGTAGCTAAACTGCTTTGTCGACGG93
30-16/2759TACCGTGGTAGGGAAGGTTGGAGTGTA93
30-38/2759ACCCGTGGTAGGGTAGGATGGGGTGGT93
60-18/3859CAGTCCGTGGTAGGGCAGGTTGGGGTGACTTCGTGGAA93
G15D59GGTTGGTGTGGTTGG93
30-859AGATGCCTGTCGAGCATGCTGTGAATAGGTAGGGTCGGATGGGCTACGGTGTAGCTAAACTGCTTTGTCGACGGG93
doi:10.1371/journal.pone.0031948.t002
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org4 February 2012 | Volume 7 | Issue 2 | e31948

Page 5

recombinant thrombin. However, the binding of only two
aptamers was affected when the plasma derived thrombin
preparation was heat treated.
We evaluated four commercially available anti-thrombin
monoclonal antibodies for their capacity to distinguish between
control and heat-treated thrombin (Fig. 6). In the experiment
depicted in Fig. 3B a monoclonal antibody was used to capture
thrombin and a polyclonal antibody used as the detector. In the
ELISA assays shown in Fig. 6, a sheep anti-human thrombin
polyclonal was used to capture the thrombin and a panel of mouse
anti-thrombin monoclonal antibodies was used to detect the
protein. The monoclonal antibody EST-2 recognizes thrombin
alone as well as the thrombin-antithrombin complex [27]. The
EST-4 on the other hand binds thrombin but not the thrombin-
antithrombin complex and it has been suggested that this
monoclonal antibody recognizes an epitope near the active site
[27]. Both these antibodies were obtained by the fusion of a-
thrombin immunized spleen cells with a myeloma cell line. We
also used two additional mouse anti-thrombin antibodies (sc-59717
and AH-5020) [28,29] raised against purified full-length native
human thrombin in these assays (Fig. 6C&D). We selected
monoclonal antibodies that were raised using protein molecules
rather than peptides as these are more likely to be conformation
sensitive. However all the four monoclonal antibodies showed
comparable binding kinetics for the control and heat treated
thrombin. In fact the monoclonal antibody sc-59717 continued to
bind denatured thrombin (heated at 60uC for 17 h in PBS).
Using aptamers in label free characterization of the
human thrombin products
ELISA based assays are simple, and do not require much
specialized equipment or skills and have found wide application in
research, clinical laboratories as well as during drug development.
These assays however require capture and detection antibodies,
are subject to wash steps, multiple incubations, and serial dilutions
for each sample; factors that affect precision, accuracy and turn
around time. Moreover, in recent years label free technologies
have been found to be advantageous over methods where the
kinetic constants are derived indirectly and are assay dependent
[30]. Several technologies are available that permit the label-free
determination of kinetic constants including the association rate
constant (kon), dissociation rate constant (kdis), and equilibrium
dissociation constant (KD). Here we utilized BioLayer Interfer-
ometry, to measure in real-time the aptamer-thrombin interac-
tions using streptavidin. In this assay, the biotinylated aptamer is
bound to the biosensor and the rate of binding to thrombin was
found to be concentration dependent as has been shown
previously for protein-protein interactions such as those between
an antibody and its ligand [30]. We used this data for base-line
correction and to obtain the konand kdisand KD values using the
Octet Software (data not given). Using the recombinant human
thrombin, Recothrom as an example we found that the Kdisvalues
obtained in the label-free assay correlate (r=0.842, P,0.05) with
the apparent Kds estimated using the ELISA based assay (Fig. 7A).
A similar correlation was obtained when we compared the KD
values from the label free assay with the Kdvalues from the ELISA
based assay (data not given). The advantage of using the kdisvalues
is that while the apparent Kd and KD determinations are
concentration dependent and require accurate estimations of
protein and nucleotide the kdisis independent of the concentration
of the ligand. This can be useful as the concentration of samples
can change during processing. The heat treatment described in the
Table 3. Kinetic parameters for the binding of aptamers to
Recothrom.
AptamerVmax
Apparent Kd
60-18/381.5860.01 1.4860.05
30-141.3860.05 2.7560.36
30-81.2960.010.5860.01
30-16/270.81960.01961.5460.42
30-38/27 0.1860.031.2860.03
G15D 0.6860.021.4860.15
doi:10.1371/journal.pone.0031948.t003
Figure 3. Comparing three topical thrombin products using binding assays with aptamers as detector reagents in an ELISA format.
(A) The three topical thrombin products were electrophoresed on a 12% NuPAGE gel and stained using Colloidal BlueH. The thrombin and HSA (used
as an exipient) bands are indicated with red and blue arrows respectively. (B) Kinetics of binding of the Aptamer 30-8 to the purified human thrombin
products Recothrom (squares) and Evithrom (circles). (C) The affinities (apparent Kds) of all six aptamers to both Recothrom and Evithrom were
determined as shown in (B). The apparent Kdvalues for Recothrom (X-axis) were plotted against those for Evithrom (Y-axis) and were shown to be
highly correlated (r=0.996, P,0.05).
doi:10.1371/journal.pone.0031948.g003
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org5February 2012 | Volume 7 | Issue 2 | e31948

Page 6

previous section involves dilution and incubation in the 60%
sucrose medium. We therefore compared the change in the
dissociation rate constants (kdis) for the six aptamers following heat
treatment of Recothrom (Fig. 7B). A statistically significant change
in the kdisis observed in 4 of the 6 aptamers as a consequence of
heat treatment, consistent with our results using the ELISA based
assay (Fig. 5).
Discussion
In this study we characterize aptamers as surrogate antibodies
for the identification of changes in the conformation of a protein
during the manufacturing process and during storage. This is
important because the inability to detect changes in the
conformation of a protein-product can have consequences even
if the activity is unaffected. For example even subtle changes in
conformation can result in new epitopes on the protein, leading to
immunogenicity [31]. Immunogenicity is currently a significant
impediment to the successful development of therapeutic proteins,
as the development of inhibitory antibodies can reduce the efficacy
and sometimes even result in life threatening safety issues [32].
Similarly, comparison of biosimilars to reference standards using
bioanalytical techniques is a key component of the licensure of
such products [5]. Here we have used a panel of aptamers to
compare three different thrombin products vis-a `-vis the confor-
mational epitopes.
The affinity of a panel of six anti-thrombin aptamers for the
protein was determined using an ELISA like assay which is
illustrated in Fig. 1B. A single monoclonal antibody was used to
capture the thrombin molecules which were then detected using
either a polyclonal anti-thrombin antibody or one of the six anti-
thrombin aptamers. In another set of experiments, the anti-
thrombin polyclonal antibody was used to capture the thrombin
molecules and the binding kinetics of four anti-thrombin
monoclonal antibodies were evaluated. By standardizing the assay
we were able to obtain excellent signal to noise ratio for all six
aptamers, comparable to that for the antibody based assay.
Moreover the limit of detection for thrombin is also comparable
regardless of whether an aptamer or antibody was used and we
obtained consistent results for binding kinetics where the per-cent
CV ranged from 9 to 14% with an average of 12.4% (Table 3).
Both the polyclonal detector antibody and all six aptamers
bound the Evithrom with high affinity (Fig. 2 & 3B) but showed
Figure 4. Conformational changes in thrombin following heat treatment in the presence of 60% sucrose. The two purified human
topical thrombin products were heated for up to 17 h at 60uC in the presence of 60% sucrose in duplicate. (A) The catalytic activity of the Recothrom
(clear bars) and Evithrom (grey, filled bars) samples was determined at time 0 and 17 h; the mean activity (6 SD) of three independently treated
samples is depicted. (B) Following heat treatment both samples were electrophoresed on a 12% NuPAGE gel and stained using Colloidal BlueH. As a
control (Con) the thrombin samples were heated at 60uC in PBS the absence of sucrose. (C) The binding kinetics of the heat-treated thrombin at time
0 h (Blue open circles) and 17 h (purple open triangles) were determined using the antibody based ELISA illustrated in Fig. 1A. Samples heated in the
presence of PBS alone are shown as red filled circles. (D) The binding kinetics of the heat-treated thrombin at time 0 h (Blue open circles), 1 h (green
open squares) and 17 h (purple open triangles) were determined using the aptamer based ELISA illustrated in Fig. 1B. Samples heated in the
presence of PBS alone are shown as red filled circles.
doi:10.1371/journal.pone.0031948.g004
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org6 February 2012 | Volume 7 | Issue 2 | e31948

Page 7

almost no binding to Thrombin JMI. Both these thrombin
products are purified from plasma but the former is of human
origin and the latter of bovine origin. This demonstrates significant
specificity as the bovine and human thrombins are 89%
homologous and show 81% sequence identity [33]. Moreover in
bioassays as well as clinical studies the bovine and human
thrombins have been shown to exhibit comparable activity. A
comparison of the binding affinities of the aptamers for Evithrom
and Recothrom (Fig. 3C) showed a very high correlation
(r=0.996, P,0.05). Evithrom and Recothrom are both human
thrombin products, however while the former is purified from
plasma the latter is a recombinant protein expressed in CHO cells
[22,23]. This result would suggest that these two products while
obtained by two very different processes have remarkably similar
conformations given that the aptamers are extremely sensitive to
subtle changes (Figs. 4D, 5 & 7B).
We have alluded to a case study from the 1990s where an
additional step of pasteurization was used to inactivate non-lipid-
enveloped viruses in the manufacture of FVIII led to the
development of inhibitory antibodies at a very high frequency
leading to the withdrawal of the product from the market [26].
Subsequent studies have suggested that the change in the
manufacturing process exposed sites in the C2 domain of the
FVIII leading to immunogenicity [17]. This case vividly illustrates
the significance of the need for techniques that can routinely detect
subtle changes in conformation because not all changes affect
product activity in in vitro assays. When we used a similar heat-
treatment for the two human thrombin products we found no
significant change in activity even after 17 h (Fig. 4A). Nor could
assays using the antibody-based ELISA platform detect any
changes in the affinity of the anti-thrombin antibody (Figs. 4C & 6)
following heat-treatment when the protein is protected with 60%
sucrose. As a control we demonstrate that heat-treatment in the
absence of sucrose completely abolishes antibody binding (Fig. 4C
& 6) in all but one of the antibodies screened. Aptamers on the
other hand show small but significant differences in binding during
heat-treatment of Recothrom in the presence of 60% sucrose and
these changes are time dependent (Fig. 4D). It is interesting to note
(a) that the magnitude of change varies for different aptamers and
(b) the binding of only some aptamers is affected (Fig. 5A). This
suggests that the aptamers recognize different epitopes or
conformations on the protein. We obtained similar results in a
label-free platform using BioLayer Interferometry (Fig. 7B). This
panel of aptamers also showed alterations in binding to Evithrom
following heat treatment, except that the binding of fewer
aptamers is affected (Fig. 5B). It is noteworthy that all the
aptamers affected by the heat-treatment of Evithrom are also
affected in Recothrom. It is plausible that the high concentration
of HSA in the Evithrom provides additional protection to the
protein during heat treatment. Taken together these results suggest
that a suitable panel of aptamers can be used to detect small
changes in protein conformation not detectable by other methods.
Figure 5. Heat treatment of human thrombins in the absence and presence of 60% sucrose. The two purified human topical thrombin
products were heat-treated as described in Fig. 3. Control samples were heated in for 17 h at 60uC in PBS (in the absence of 60% sucrose). Binding of
either the antibody or each of the six aptamers was monitored to (A) Recothrom or (B) Evithrom at time 0 and 17 h. The percent of the binding at
time 0 (mean 6 SD, n=3) is represented. The aptamers used are shown on the figure.
doi:10.1371/journal.pone.0031948.g005
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org7 February 2012 | Volume 7 | Issue 2 | e31948

Page 8

We used a panel of six aptamers in this study; the nature,
number and choice of antibodies/aptamers is likely to be an
important factor in the success of epitope mapping for detecting
subtle changes in the conformation of individual protein products.
Depending on the complexity of a therapeutic protein the panel of
aptamers used to map a protein can readily be expanded, the same
is not true of antibodies. In addition protein-specific aptamers are
selected from a very large pool of conformations [34–36].
Consequently a larger repertoire of aptamers representing more
epitopes can be generated for a target protein. Aptamers are also
likely to be more useful in a quality control environment [37–41].
In addition, we determined in this study that the presence of
excipients (including proteins other than the active ingredient) has
no significant effect on the binding of aptamers to the thrombins.
This observation could potentially be of consequence in
comparing biosimilars to their reference standards. The primary
hurdle for the developer of any biosimilar is that there is normally
no direct access to originator companies’ proprietary data. Thus,
the developer of a biosimilar has to retrieve the reference protein
as a finished drug product, purify the drug substance and reverse
engineer the process [42]. This in turn has raised concerns that the
reverse engineered ‘reference standard’ could itself be altered from
the original active ingredient manufactured by the innovator. As
aptamer binding is unaffected by the excipients, the assays
described here provide a means of comparing the active ingredient
in the finished drug product with the reverse engineered reference
standard.
This study demonstrates that aptamers can be used to
characterize changes in the tertiary structure of proteins. This
has potential applications as an analytical test during the
development of therapeutic proteins and to monitor the effect of
manufacturing changes and storage. A range of biochemical and
biophysical tests are available to characterize protein-drugs,
however analytical methods to rapidly and routinely monitor
subtle changes in the tertiary structure represent a critical lacuna.
The need for such methods is particularly acute in the
development of biosimilars. Here we have provided proof-of-
principle that aptamers are stable reagents that can be used to
design sensitive and robust analytical tests to detect small changes
in the conformation of therapeutic proteins.
Figure 6. Effect of heat treatment of thrombin in the presence of 60% sucrose monitored using anti-thrombin monoclonal
antibodies. The binding kinetics of the heat-treated thrombin (Recothrom) at time 0 h (Green open squares) and 17 h (purple open triangles) were
determined using an antibody based ELISA. The polyclonal anti-thrombin antibody, PAHT-S-B was used as the capture antibody and a panel of
monoclonal antibodies as detector antibodies as described in the Materials & Methods. The binding kinetics of the mouse anti-thrombin antibodies
EST-2 (A), EST-4 (B), sc-59717 (C) and AH-5020 (D) are depicted.
doi:10.1371/journal.pone.0031948.g006
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org8 February 2012 | Volume 7 | Issue 2 | e31948

Page 9

Materials and Methods
Materials
Recombinant human thrombin was from Zymogenetics (Seat-
tle, WA), and plasma-derived human thrombin was from Omrix
(Israel). Aptamers were synthesized by the standard phosphor-
amidite method on an ABI model 394 oligonucleotide synthesizer
(Facility for Biological Research, CBER, FDA), purified by
reversed phase HPLC and lyophilized. Aptamers were used either
unmodified or in the biotynilated forms as indicated in the text.
Table 2 lists the aptamers used in the study.
Antibody based binding assay
ELISA plates (MAxisorp, Nunc) were coated with 50 ml mouse
anti-human thrombin monoclonal antibody (Hemtech AHT-5020)
diluted to 3.9 mg/ml in coating buffer (50 mM Na2CO3, pH 9.6),
and incubated overnight at 4uC. Plates were washed in PBST and
blocked for 1 hr at 37uC with 300 ml per well of 2% (W/V) BSA in
Tris-NaCl pH 7.6 (TSTA). After washing, plates were incubated
with 5 mg/ml of the indicated thrombin preparation diluted in
PBST (50 ml, in duplicate) for 1 hr at 37uC. Plates were washed 4
times with PBST and incubated with biotinylated sheep anti-
human thrombin polyclonal antibody (50 ml, Hemtech PAHT-S-
B) diluted 1:4000 in PBST for 1 hr at 37uC. After additional
washing with PBST, the plates were incubated with 50 ml of HRP-
Nutravidin (Pierce) diluted 1:10000 in PBST for 1 hr at 37uC.
Finally, plates were washed 4 times with PBST, and the
chromogenic reaction was started using 100 ml 3,39,5,59-tetra-
methylbenzidine (TMB). Reaction was stopped with 100 ml
0.25 M H2SO4and absorbance was measured at 450 nm. To
determine the binding kinetics of a panel of anti-thrombin
monoclonal antibodies, the ELISA plates (MAxisorp, Nunc) were
coated with 50 ml sheep anti-human thrombin polyclonal antibody
(Hemtech PAHT-S-B) diluted to 10 mg/ml in coating buffer
(50 mM Na2CO3, pH 9.6), and incubated overnight at 4uC. Plates
were processed as described above and incubated with 5 mg/ml of
the indicated thrombin preparation diluted in PBST (50 ml, in
duplicate) for 1 hr at 37uC. Plates were then washed 4 times with
PBST and incubated for 1 hr at 37uC with serial dilutions of one
of four mouse anti-human thrombin monoclonal antibodies (EST-
2, EST-4 both from American Diagnostica Inc; sc-59717, Santa
Cruz Biotechnology, Inc. or AHT-5020, Haematologic Technol-
ogies, Inc). After additional washing with PBST, the plates were
incubated with 50 ml of HRP-conjugated anti-mouse antibody
(Rockland) diluted 1:10000 in PBST for 1 hr at 37uC. Finally,
plates were washed 4 times with PBST, and the chromogenic
reaction was started using 100 ml TMB. Reaction was stopped
with 100 ml 0.25 M H2SO4 and absorbance was measured at
450 nm.
Aptamer based binding Assay
ELISA plates (MAxisorp, Nunc) were coated with 50 ml mouse
anti-human thrombin monoclonal antibody (Hemtech AHT-5020)
diluted to 3.9 mg/ml in coating buffer (50 mM Na2CO3, pH 9.6),
and incubated overnight at 4uC. Plates were washed in PBST and
blocked for 1 hr at 37uC with 300 ml per well of 2% (W/V) BSA in
Tris-NaCl pH 7.6 (TSTA). After washing, plates were incubated
with 5 mg/ml of the indicated thrombin preparation diluted in
PBST (50 ml, in duplicate) for 1 hr at 37uC. Biotinylated aptamers
were heated to 95uC for 10 minutes and allowed to refold at room
temperature (RT) for another 20 minutes. Folded aptamers
serially diluted in folding buffer (25 mM Tris, 10 mM NaCl,
1 mM MgCl2pH=7.4) were loaded to the plate and incubated at
RT protected from light for 1 hr. The plate was washed 4 times
with aptamer folding buffer containing 0.005% Tween-20. The
plates were incubated with 50 ml of HRP-Nutravidin (Pierce)
diluted 1:10000 in dilution buffer for 1 hr at RT. After a final
wash the chromogenic reaction was started using 100 ml TMB.
Reaction was stopped with 100 ml 0.25 M H2SO4and absorbance
was measured at 450 nm.
Figure 7. Estimating the binding of aptamers to thrombin in a label free assay. The binding of aptamers to Recothrom was monitored in
real time using BLI and the kinetic parameters obtained using the Octet data analysis software version 7.0. (A) Correlation between Kdisvalues
obtained in the label-free assay with the apparent Kds estimated using the ELISA based assay (r=0.842, P,0.05). (B) Effect of heat-treatment of
human thrombins on the kdisfor all 6 aptamers. Recothrom was heat-treated as described in Fig. 3 and the kdisdetermined following heat treatment
for 0 and 17 h. The percent of the binding at time 0 is represented. Data presented is the mean (6 SD) of three independent experiments.
doi:10.1371/journal.pone.0031948.g007
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org9 February 2012 | Volume 7 | Issue 2 | e31948

Page 10

Stability Assay
Recombinant and plasma derived thrombin were diluted to
30 mg/ml in sucrose buffer (60% w/v sucrose, 0.1 M glycine,
pH=7.3). Control samples were diluted accordingly in PBS.
Samples (1.4 ml) were incubated at 60uC for 0, 1 and 17 hr and
maintained at 280uC until analyzed.
Activity assay
Two-fold dilution of calibrator thrombin (25 ml, Thrombino-
scope, Netherlands) in the range of 1000-0.1 nM or unknown
samples were added to 96-well plates pre-loaded with 25 ml Z-
GGR-AMC substrate at 800 mM in reaction buffer (20 mM
Hepes, 150 mM NaCl, 0.1% BSA, pH7.4). The increase in
fluorescence intensity, which is proportional to thrombin activity,
was monitored continuously at 37uC by automatic reading every
30 sec up to 30 min using fluorescence reader (Tecan) with an
excitation wavelength of 380 nm and an emission wavelength of
430 nm. Thrombin concentrations were expressed as the rate
development of fluorescence intensity [arbitrary fluorescence units
(FU)], calculated for every reading (FU min21).
Determining the Binding Affinity of aptamers using Bio-
Layer Interferometry
Bio-Layer Interferometry (BLI), a label-free technology, was
used for measuring the binding of aptamers with thrombin
samples. The affinity measurements were performed with ForteBio
Octet RED96 equipped with streptavidin (SA) biosensor tips
(ForteBio, Inc., Menlo Park, CA, USA). The assays were
maintained at a temperature of and the speed of 30uC and
1000 rpm respectively. Streptavidin-coated biosensor tips were
pre-wet for 15 min. Then the tips were loaded with 250 nM–
500 nM of biotinylated aptamer for 5 min. The typical resulting
capture levels of eight biosensors were in the range of 0.4–0.5 nm.
The association (kon) and dissociation (kdis) were then established
by dipping the biosensors for 10 mins in various concentrations of
thrombin samples dispensed in 96-microwell plates (Fisher
Scientific, Turnburry Drive, Hanover Park, IL, USA) at a volume
of 200 ml per well. Data were processed and analysed using the
Octet data analysis software version 7.0 (ForteBio). The binding
profile of each aptamer was shown as ‘‘nm’’ shift. This shift is a
comparison of the shift in the interference patterns of light
reflected from a reference layer within the biosensor versus
molecules as the bind to the biosensor tip. The results were
summarized as KD which was calculated from ‘‘KD=kon/kdis’’,
where konis the ‘on rate’ or association and kdisis the ‘off rate’ or
dissociation.
Statistical analysis
Stability experiment was conducted in three independent
repeats. Statistical analysis was carried out using repeated analysis
of variance (ANOVA) performed with GraphPad InStat 3
software. Differences were considered significant when P was
,0.05.
Acknowledgments
We thank Drs. Chava Kimchi-Sarfaty (FDA), Adam Friedman (FDA),
Teresa Przytycka (NCBI, NIH) and Jan Hoinka (NCBI, NIH) for helpful
discussions and Ms. Divya Jain for technical assistance.
Disclaimer The findings and conclusions in this article have not been
formally disseminated by the Food and Drug Administration and should
not be construed to represent any Agency determination or policy.
Author Contributions
Conceived and designed the experiments: RZ BG ZES. Performed the
experiments: RZ WC GSP. Analyzed the data: RZ WC ZES. Wrote the
paper: RZ ZES.
References
1. Belsey MJ, Harris LM, Das RR, Chertkow J (2006) Biosimilars: initial
excitement gives way to reality. Nat Rev Drug Discov 5: 535–536.
2. Rathore AS (2009) Follow-on protein products: scientific issues, developments
and challenges. Trends Biotechnol 27: 698–705.
3. Rathore AS, Winkle H (2009) Quality by design for biopharmaceuticals. Nat
Biotechnol 27: 26–34.
4. Schellekens H (2004) When biotech proteins go off-patent. Trends Biotechnol
22: 406–410.
5. Anonymous (2006) Guideline on similar biological medicinal products
conataining biotechnology-derived proteins as active substance. European
Medicines Agency.
6. Kozlowski S, Woodcock J, Midthun K, Sherman RB (2011) Developing the
nation’s biosimilars program. N Engl J Med 365: 385–388.
7. Schellekens H, Moors E (2010) Clinical comparability and European biosimilar
regulations. Nat Biotechnol 28: 28–31.
8. Roger SD (2010) Biosimilars: current status and future directions. Expert Opin
Biol Ther 10: 1011–1018.
9. Chirino AJ, Mire-Sluis A (2004) Characterizing biological products and assessing
comparability following manufacturing changes. Nat Biotechnol 22: 1383–1391.
10. Rockberg J, Lofblom J, Hjelm B, Uhlen M, Stahl S (2008) Epitope mapping of
antibodies using bacterial surface display. Nat Methods 5: 1039–1045.
11. Barlow DJ, Edwards MS, Thornton JM (1986) Continuous and discontinuous
protein antigenic determinants. Nature 322: 747–748.
12. Pizzi E, Cortese R, Tramontano A (1995) Mapping epitopes on protein surfaces.
Biopolymers 36: 675–680.
13. Tombelli S, Mascini M (2009) Aptamers as molecular tools for bioanalytical
methods. Curr Opin Mol Ther 11: 179–188.
14. Kessler CM, Ortel TL (2009) Recent developments in topical thrombins.
Thromb Haemost 102: 15–24.
15. Lomax C, Traub O (2009) Topical Thrombins: Benefits and Risks.
Pharmacotherapy 29: 8S–12S.
16. Dichtelmuller H, Rudnick D, Breuer B, Kotitschke R, Kloft M, et al. (1996)
Improvement of virus safety of a S/D-treated factor VIII concentrate by
additional dry heat treatment at 100 degrees C. Biologicals 24: 125–130.
17. Raut S, Di Giambattista M, Bevan SA, Hubbard AR, Barrowcliffe TW, et al.
(1998) Modification of factor VIII in therapeutic concentrates after virus
inactivation by solvent-detergent and pasteurisation. Thromb Haemost 80:
624–631.
18. Tasset DM, Kubik MF, Steiner W (1997) Oligonucleotide inhibitors of human
thrombin that bind distinct epitopes. J Mol Biol 272: 688–698.
19. Macaya RF, Schultze P, Smith FW, Roe JA, Feigon J (1993) Thrombin-binding
DNA aptamer forms a unimolecular quadruplex structure in solution. Proc Natl
Acad Sci U S A 90: 3745–3749.
20. Tsiang M, Jain AK, Dunn KE, Rojas ME, Leung LL, et al. (1995) Functional
mapping of the surface residues of human thrombin. J Biol Chem 270:
16854–16863.
21. Thrombin-JMI package insert. (2007) Jones Pharma Incorporated, Bristol, VA.
22. Evithrom Package Insert. (2007) J&J Wound Management, Somerville, NJ.
23. Recothrom Package Insert. (2007) Zymogenetics Inc, Seatle, WA.
24. Doria C, Fischer CP, Wood CG, Li PM, Marra S, et al. (2008) Phase 3,
randomized, double-blind study of plasma-derived human thrombin versus
bovine thrombin in achieving hemostasis in patients undergoing surgery. Curr
Med Res Opin 24: 785–794.
25. Chapman WC, Singla N, Genyk Y, McNeil JW, Renkens KL, Jr., et al. (2007) A
phase 3, randomized, double-blind comparative study of the efficacy and safety
of topical recombinant human thrombin and bovine thrombin in surgical
hemostasis. J Am Coll Surg 205: 256–265.
26. Peerlinck K, Arnout J, Di Giambattista M, Gilles JG, Laub R, et al. (1997)
Factor VIII inhibitors in previously treated haemophilia A patients with a double
virus-inactivated plasma derived factor VIII concentrate. Thromb Haemost 77:
80–86.
27. Dawes J, James K, Micklem LR, Pepper DS, Prowse CV (1984) Monoclonal
antibodies directed against human alpha-thrombin and the thrombin-anti-
thrombin III complex. Thromb Res 36: 397–409.
28. Modesti PA, Gamberi T, Bazzini C, Borro M, Romano SM, et al. (2009)
Response of serum proteome in patients undergoing infrarenal aortic aneurysm
repair. Anesthesiology 111: 844–854.
29. Byeon JY, Limpoco FT, Bailey RC (2010) Efficient bioconjugation of protein
capture agents to biosensor surfaces using aniline-catalyzed hydrazone ligation.
Langmuir 26: 15430–15435.
30. Concepcion J, Witte K, Wartchow C, Choo S, Yao D, et al. (2009) Label-free
detection of biomolecular interactions using BioLayer interferometry for kinetic
characterization. Comb Chem High Throughput Screen 12: 791–800.
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org10 February 2012 | Volume 7 | Issue 2 | e31948

Page 11

31. Schellekens H (2005) Factors influencing the immunogenicity of therapeutic
proteins. Nephrol Dial Transplant 20: 3–9.
32. Shankar G, Pendley C, Stein KE (2007) A risk-based bioanalytical strategy for
the assessment of antibody immune responses against biological drugs. Nat
Biotechnol 25: 555–561.
33. Schoenecker JG, Johnson RK, Fields RC, Lesher AP, Domzalski T, et al. (2003)
Relative purity of thrombin-based hemostatic agents used in surgery. J Am Coll
Surg 197: 580–590.
34. Carothers JM, Oestreich SC, Davis JH, Szostak JW (2004) Informational
complexity and functional activity of RNA structures. J Am Chem Soc 126:
5130–5137.
35. Ruigrok VJ, Levisson M, Eppink MH, Smidt H, van der Oost J (2011)
Alternative affinity tools: more attractive than antibodies? Biochem J 436: 1–13.
36. Jhaveri S, Rajendran M, Ellington AD (2000) In vitro selection of signaling
aptamers. Nat Biotechnol 18: 1293–1297.
37. Iliuk AB, Hu L, Tao WA (2011) Aptamer in bioanalytical applications. Anal
Chem 83: 4440–4452.
38. Soukup GA, Breaker RR (1999) Nucleic acid molecular switches. Trends
Biotechnol 17: 469–476.
39. Famulok M (2005) Allosteric aptamers and aptazymes as probes for screening
approaches. Curr Opin Mol Ther 7: 137–143.
40. Rentmeister A, Mayer G, Kuhn N, Famulok M (2007) Conformational changes
in the expression domain of the Escherichia coli thiM riboswitch. Nucleic Acids
Res 35: 3713–3722.
41. Vinkenborg JL, Karnowski N, Famulok M (2011) Aptamers for allosteric
regulation. Nat Chem Biol 7: 519–527.
42. Schneider CK, Kalinke U (2008) Toward biosimilar monoclonal antibodies. Nat
Biotechnol 26: 985–990.
Epitope Mapping Using Aptamers
PLoS ONE | www.plosone.org11 February 2012 | Volume 7 | Issue 2 | e31948