http://informahealthcare.com/autISSN: 0891-6934 (print), 1607-842X (electronic)

Autoimmunity, 2014; 47(7): 438–444! 2014 Informa UK Ltd. DOI: 10.3109/08916934.2014.914176

ORIGINAL ARTICLE

Anti-b2-glycoprotein I paratopes and b2-glycoprotein I epitopescharacterization using random peptide libraries

Urska Zager1, Tanja Kveder1, Sasa Cucnik1, Borut Bozic1,2, and Mojca Lunder2

1Department of Rheumatology, University Medical Centre Ljubljana, Ljubljana, Slovenia and 2Faculty of Pharmacy, University of Ljubljana,

Ljubljana, Slovenia

Abstract

Studies concerning interactions between anti-b2-glycoprotein I antibodies (anti-b2GPI) andb2-glycoprotein I (b2GPI) suggest relevance of charge interactions and hydrogen bonds.However, paratope of diagnostically and clinically relevant anti-b2GPI and epitope character-istics of b2GPI, still remain unclear. The aim of our study was to determine paratopecharacteristics of various anti-b2GPI antibodies and epitope characteristics of b2GPI usingphage display. Monoclonal IgG anti-b2GPI, purified polyclonal high avidity and low avidity IgGanti-b2GPI derived from plasma of APS patients were used to screen phage display libraries.The affinity and competition ability of selected clones were evaluated. Various heptapeptidespresenting putative paratopes of anti-b2GPI and specific heptapeptides presenting putativeepitopes of b2GPI were determined. Epitope presenting peptides bind to the respectiveanti-b2GPI and consequently interrupt antibody–antigen interaction. The amino acid compos-ition of selected peptides confirmed the importance of hydrogen bonds and chargeinteractions in the binding of anti-b2GPI to the antigen. Epitopes recognized by high avidityanti-b2GPI predominately contain hydrogen bond forming side chains, while in low avidityanti-b2GPI epitope the charged side chains prevail. The alignment of selected sequences tothree-dimensional antigen structure revealed that polyclonal high avidity anti-b2GPI recognizenative epitopes that are accessible regardless of b2GPI’s conformation whereas the epitoperecognized by low avidity anti-b2GPI is cryptic and cannot be accessed when b2GPI takes theclosed plasma conformation.

Keywords

Antiphospholipidsyndrome, antibody–antigen interaction,epitope mimetics, paratope mimetics,phage display

History

Received 20 February 2014Revised 20 March 2014Accepted 08 April 2014Published online 9 May 2014

Introduction

Bacteriophage peptide library is a pool of phage clones each

displaying a unique peptide on its surface [1]. Random

peptide libraries are generated by inserting the peptide-

encoding oligonucleotide into the gene coding for phage coat

protein, which results in a most advantageous aspect of phage

display: the genotype–phenotype linkage [2]. These libraries

are very convenient tool for studying protein interactions and

identification of new ligands for various biological targets [3].

Namely, by sequential affinity selections of phage library with

target macromolecule, phage displayed peptides with high

binding affinity towards target molecule can be recovered and

their primary structure determined. Characterization of anti-

body molecule by peptide libraries usually refers to the

epitope mapping. The epitope presenting or mimicking

peptides can be applied toward identification of unknown

antigens, development of more specific diagnostic immuno-

assays or can serve as lead compounds for novel therapeutics

and vaccines. The paratope presenting peptides can

conversely provide an insight into antibody–antigen inter-

actions, reveal distinct differences among different antibodies

targeting the same antigen and present potential triggers of

anti-idiotypic response.

Antibodies against b2-glycoprotein I (anti-b2GPI) are

commonly found in the sera of patients with antiphospholipid

syndrome (APS), an autoimmune disorder that manifests

clinically as recurrent vascular thrombosis and/or fetal loss

[4,5]. The reported pathogenicity of the anti-b2GPI in the

experimental animal models and their correlation with the

disease manifestations in humans indicate that these anti-

bodies have pathogenic role and are therefore potential

therapeutic target [6–9]. Especially high avidity (HAv) anti-

b2GPI were demonstrated to associate with thrombosis and

obstetric complications in APS [10]. We used linear and

constrained heptamere libraries with an attempt to character-

ize anti-b2GPI paratopes and b2GPI epitopes. For the

paratope determination, the libraries were screened with

b2GPI, whereas for the epitope determination the libraries

were screened with different anti-b2GPI subgroups. Various

washing and elution protocols were applied to select paratope

and epitope mimicking peptides.

Correspondence: Mojca Lunder, Faculty of Pharmacy, University ofLjubljana, Askerceva 7, 1000 Ljubljana, Slovenia. Tel: +386 1 42 69570. E-mail: mojca.lunder@ffa.uni-lj.si

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

Materials and methods

Patients’ sera and monoclonal antibodies

High avidity (HAv) and low avidity (LAv) IgG anti-b2GPI

were isolated from sera of two male patients with primary

APS, who suffered initially from venous thrombosis and were

positive for IgG anti-b2GPI [11]. Monoclonal chimeric IgG

anti-b2GPI antibody (HCAL, Inova Diagnostics Inc., San

Diego, CA), was used as control material.

Isolation of anti-b2GPI antibodies

First, total IgG were isolated from the sera by protein G

affinity chromatography (MabTrapTM Kit; Amersham, GE

Healthcare, Little Chalfont, UK), according to the manufac-

turer’s instructions. Subsequently, LAv and HAv anti-b2GPI

were isolated from IgG fractions by an in-house b2GPI

affinity column. Both IgG fractions (from patient A and

patient B) were applied separately to the b2GPI affinity

column and circulated for 90 min at 4 �C. After extensive

washing with phosphate buffered saline (PBS) at pH 7.4, LAv

anti-b2GPI were eluted with 0.5 M NaCl PBS and HAv anti-

b2GPI with 0.1 M glycine/4 M NaCl PBS, pH 2.5. Eluates

were immediately neutralized, dialyzed overnight against PBS

at pH 7.4 and concentrated (Amicon ultra centrifugal unit;

Millipore, Billerica, MA) [12]. The concentrations of active

LAv and HAv IgG anti-b2GPI were determined by anti-

b2GPI ELISA [13]. Three polyclonal fractions, the LAv anti-

b2GPI of patient B, the HAv anti-b2GPI of patient A and the

HAv anti-b2GPI of patient B were used in following

experiments.

Immobilization of target molecules

For paratope characterization, microtitre plates (High

Binding; Costar, Cambridge, MA) were coated with 2 mg/

well of un-nicked b2GPI [13] in PBS pH 7.4 and incubated

for 2 h at room temperature (RT). After rinsing, the wells were

blocked with 250 ml/well of 1% bovine serum albumin (BSA)

in PBS and incubated for additional 2 h at (RT). Microtitre

plates for subtractive selection were coated only with 1% BSA

in PBS.

For epitope characterization, microtitre plates (Nunc-

immuno Maxisorp, Thermo Fisher Scientific,

Langenselbold, DE) coated with 10 mg/well of protein G

(Sigma–Aldrich, Taufkirchen, Germany) in 0.1 M carbonate

buffer pH 9.6 were incubated overnight at 4 �C. After washing

with 250 ml/well of PBS, 100 ml/well of monoclonal (2.2 mg/l),

polyclonal HAv or polyclonal LAv IgG anti-b2GPI in

blocking buffer (1% BSA in PBS) were applied and incubated

for 2 h at RT. Microtitre plates for subtractive selection were

coated with protein G and 1% BSA.

Phage selection

Commercially available linear heptamer (Ph.D.-7) and cyclic

heptamer (Ph.D.-C7C) libraries (New England Biolabs,

Beverly, MA) were used for selection of b2GPI and anti-

b2GPI binding peptides. The selection was carried out

according to manufacturer’s instructions. Briefly, phage

library (2� 1011 pfu in 0.1% Tween-20 in PBS (PBST))

was incubated with immobilized target (b2GPI or anti-b2GPI)

at RT with gentle shaking. Unbound phages were removed by

intensive washing. Bound clones were eluted either with

solutions of targets’ ligands (anti-b2GPI or b2GPI) or buffers

for non-specific disruption of binding interactions (0.1 M

glycine buffer pH 2.2 with 4 M NaCl) in combination with

ultrasound (30 kHz, 10 min, RT). Recovered phages were

amplified for subsequent rounds of selection. In subsequent

rounds, the selection conditions were intensified by prelim-

inary subtractive selections, by shortening the binding period

and by intensifying the washing steps. Washing was

intensified by increasing ionic strength of the basic washing

buffer (1% PBST with 0.5 M NaCl) or using 0.1 M glycine

buffer pH 2.5 with 0.5% BSA. After the second, third or

fourth round of selection 20–25 random plaques were picked,

amplified and purified by PEG precipitation to perform

preliminary phage ELISA (as described below). Phage clones

giving rise to absorbance40.2 and target to background

ratio42 were considered positive binders. A single-stranded

DNA from selected phage clones was isolated and sequenced

(MWG Biotech, Munich, Germany).

Phage ELISA

Phage clones with target to background ratio in preliminary

ELISA44 were further characterized in quantitative ELISA

where equal amount of each phage clone determined by

titration was used. Microtitre plates (Nunc-immuno Maxisorp

or High binding Costar) were coated with target (b2GPI or

anti-b2GPI) in PBS and incubated overnight at 4 �C. After

one wash with PBS, blocking buffer (1% BSA in PBS) was

added and incubated for 2 h at RT. For negative control, a

separate set of wells was incubated only with blocking buffer

without previous target immobilization. About 5� 109 pfu of

each selected phage clone in 0.1% PBST was applied to

coated wells and incubated for 1 h at RT. After washing with

0.1% PBST horseradish peroxidase-labeled mouse anti-M13

monoclonal antibody (Amersham Biosciences, Little

Chalfont, UK) in 1% BSA in 0.1% PBST (dilution 1:5000)

was added for 1 h. Finally, the plate was rinsed four times

(0.1% PBST) and the substrate 3,30,5,50-Tetramethylbenzidine

(TMB, Sigma–Aldrich, St Louis, MO) was added. After color

development, the stop solution (2M H2SO4) was added and

the absorbance (450 nm) was measured.

Competitive ELISA

b2GPI-coated plates (High binding Costar) were blocked with

1% BSA in PBS. The ability of phage clones presenting

putative anti-b2GPI’ paratopes to occupy b2GPI epitopes and

thus prevent the anti-b2GPI binding to the antigen was tested

by applying mixtures of anti-b2GPI and individual phage

clones (1� 1010 or 2� 1010 pfu/well) to the b2GPI-coated

plates. The ability of phage clones presenting putative

epitopes to inhibit the anti-b2GPI binding to immobilized

b2GPI was determined the same way, except the mixtures

were incubated for 2 h at RT before applying them to the

b2GPI-coated plates. The detection of anti-b2GPI bound to

b2GPI was the same as in routine anti-b2GPI ELISA [14].

Phage clones without fusion peptide or selected on unrelated

target were used as negative controls.

DOI: 10.3109/08916934.2014.914176 Anti-b2GPI paratopes and b2GPI epitopes characterization 439

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

Alignment of putative epitope sequences to thethree-dimensional structure of b2GPI

To map each of the selected peptide sequences presenting

putative epitopes onto the surface of the antigen, we used the

Pepitope server (Tel Aviv, Israel), a web-based graphical tool

that allows viewing the predicted epitopes [15].

Results

Anti-b2GPI’ paratope characterization

Various selection protocols in terms of rinsing and elution

conditions were used to select phage clones mimicking

paratopes of anti-b2GPI antibodies. Rinsing conditions were

intensified by applying the rinsing buffer with higher salt

concentration or lower pH which contributes to the disruption

of electrostatic bonds. Ultimately phage clones were obtained

in elution step with HCAL, polyclonal HAv, polyclonal LAv

anti-b2GPI antibodies or non-specifically with acidic buffer.

Positive binders were determined in preliminary phage

ELISA (Table 1). Phage clones with target to background

ratio44 were further characterized in quantitative ELISA

(Figure 1). Selections among cyclic heptapeptides were

largely unsuccessful. Only two positive binders were obtained

during non-specific elution (peptides 40 and 41, Table 1).

As expected, common motif Q/N, L/P/A/I, W, L/P

was obtained during elution with HCAL (peptides 16–21,

Table 1). Some displayed peptides (peptides 22–24) did not

share that motif despite strong affinity determined in ELISA.

To some extend that motif emerged also during elution with

Table 1. Heptapeptide sequences selected by various selection protocols in an attempt tocharacterize the anti-b2GPI paratopes.

Gray shades mark peptide motifs found in selections using various antibodies or acidic buffer forelution. Common features among selected motifs obtained during elution strategies areunderlined.

Figure 1. Binding of selected clones to the corresponding b2GPI in quantitative ELISA. BSA-coated wells were used for negative control (white bars).

440 U. Zager et al. Autoimmunity, 2014; 47(7): 438–444

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

isolated polyclonal anti-b2GPI (peptides 1–3 and 25–26) and

even during non-specific elution with low pH (peptide 37).

During elution with HAv anti-b2GPI, other motifs also

emerged and some shared common features with motifs

obtained in elution with LAv anti-b2GPI (Table 1, under-

lined). The greatest variability of peptides and lack of

common motifs was seen in elution with LAv anti-b2GPI.

Despite significant and selective binding of the isolated

clones to the b2GPI (Figure 1), none of them was able to

diminish interactions between b2GPI and corresponding anti-

b2GPI subgroup, as revealed by competitive ELISA (data not

shown). Namely, anti-b2GPI ELISA signals did not decrease

upon applying the antibodies mixed with 2� 1010 pfu/well of

each selected phage clone. This indicates that affinity of the

antibodies was greater than that of the phage displayed

peptide. Since antibodies were intended to displace bound

displayed peptides from b2GPI, the inability of peptides to

diminish antibody–antigen interaction was expected.

However, even by non-specific acidic elution peptides that

could prevent binding of anti-b2GPI to its target were not

obtained.

b2GPI’ epitope characterization

To define epitopes targeted by different anti-b2GPI anti-

bodies, three biopanning rounds of linear and constrained

heptamere libraries over corresponding anti-b2GPI were

performed. Only selections with linear library yielded

clones that gave rise to absorbance40.2 and target back-

ground ratio42 in preliminary ELISA (data not shown).

During the selection of HCAL binding phage clones, the

elution was carried out either by b2GPI in solution or non-

specifically using acidic buffer. Fifty phage clones from each

selection were randomly picked and their affinity towards

HCAL was determined. Three phage clones exhibited signifi-

cantly higher binding to HCAL than to background in

preliminary ELISA and were identical in their primary

structure, as revealed by nucleotide sequence analysis

(Figure 2). The three clones emerged during non-specific

elution. Therefore, selection of HAv and LAv anti-b2GPI

binding phage clones was performed by employing just non-

specific acidic elution. HAv anti-b2GPI from patient A and

patient B and LAv anti-b2GPI from patient B were used as

targets. From each selection experiment, 20 phage clones were

randomly picked and tested for their affinity towards corres-

ponding polyclonal anti-b2GPI. Only one phage clone with

displayed peptide KMDGNHP significantly and selectively

bound to polyclonal LAv anti-b2GPI fraction. The selections

on HAv anti-b2GPI from patient A and from patient B resulted

in isolation of 13 and 6 positive binders, respectively

(Figure 2). Sequence analysis revealed two displayed peptides

FNPYWYV and QGPAHSK selected on HAv anti-b2GPI of

patient A and the clone displaying peptide FNPYWYV

selected on HAv anti-b2GPI of patient B. Interestingly, the

majority of HAv anti-b2GPI phage clones were identical

despite different origins of antibodies (Figure 2).

The ability of identified phage clones to bind to the

respective anti-b2GPI and consequently to interrupt

their interaction with b2GPI was tested by inhibition

ELISA. The inhibition capacity of phage clones was

expressed as the percentage of inhibition, calculated from

the decrease in anti-b2GPI ELISA signal due to the addition

of each phage clone to the anti-b2GPI in reference to anti-

b2GPI signal without phage clone addition. The addition of

1� 1010 pfu/well of HCAL specific clone SLDSDRS to

12.5 mg/ml HCAL resulted in 56.5% inhibition of HCAL

binding to b2GPI. Two unrelated control clones (displaying

peptide GLIDRIA or QTLNTIK) caused only 1.5% and 3.2%

of inhibition, respectively. The SLDSDRS inhibition activity

depended on HCAL concentration and reached 78.5% at

6.25 mg/ml of HCAL and 39% at 50 mg/ml of HCAL.

The inhibition capacity of LAv and HAv anti-b2GPI

specific clones was detected at �12 mg/ml of anti-b2GPI.

HAv anti-b2GPI specific clones FNPYWYV and QGPAHSK

caused 34.5% and 36.9% inhibition of HAv anti-b2GPI

binding to b2GPI, respectively and the control clone without

heptapeptide caused 8.9% inhibition. LAv anti-b2GPI binding

clone KMDGNHP caused 35.0% inhibition of polyclonal LAv

anti-b2GPI binding to b2GPI, whereas with clone without

heptapeptide only 5.3% inhibition was achieved.

Since none of the selected sequences corresponded to the

primary sequence of the b2GPI, the Pepitope server was used

to align the selected sequences to the surface of three-

dimensional b2GPI model. Based on the required input of

peptide sequences and a PDB identifier of the antigen, the

algorithm efficiently searches virtually all possible three-

dimensional paths (i.e. a sequence of neighboring residues on

the surface of the antigen structure) for those that exhibit high

similarity to the peptide sequences [15]. HCAL specific clone

SLDSDRS corresponds very well to the cluster289S323V322D321S319D317K in the fifth b2GPI domain

(Figure 3). The HAv anti-b2GPI specific clone QGPAHSK

Figure 2. Primary sequences of displayedpeptides presenting putative epitopesrecognized by various anti-b2GPI. Blackbars represent binding of the specific phageclones to the corresponding anti-b2GPI inquantitative ELISA. White bars representbinding to background (BSA).

DOI: 10.3109/08916934.2014.914176 Anti-b2GPI paratopes and b2GPI epitopes characterization 441

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

matches well with the cluster 143N142G139P141A144N140S138K

on the third domain while the FNPYWYV clone partially

corresponds to 96Y98N116P115L111W81F64V cluster in the

second domain of the b2GPI. The LAv anti-b2GPI specific

clone KMDGNHP corresponds to the cluster between the

third and fourth domain consisting of 231K161M165D163G164N

(Figure 3).

Discussion

Anti-b2GPI represents one of the main subgroups of

antiphospholipid antibodies which are encountered in

patients with APS [16]. Due to their diagnostic value and

putative pathogenicity, anti-b2GPI has been the subject of

extensive research. However, the reports concerning anti-

b2GPI features are very diverse and indicate an enormous

heterogeneity among anti-b2GPI in terms of their avidity,

epitopic specificity and their involvement in pathogenesis

[10,17,18].

The aim of the present study was to determine character-

istics of anti-b2GPI’s fine specificity by revealing the

paratopes of the monoclonal HCAL, polyclonal HAv and

LAv anti-b2GPI from APS patients and by identifying the

epitopes on b2GPI. HCAL is a HAv chimeric antibody that

contains variable regions of WB-CAL-1, a monoclonal

antibody established from an APS-prone mouse which has

specificity similar to that of antiphospholipid antibodies in

sera of humans with APS [19]. The HAv and LAv anti-b2GPI

are polyclonal patients’ derived IgG fractions that were

obtained by separating patients’ total anti-b2GPI by affinity

chromatography using gradient elution. Linear and con-

strained heptamere peptide libraries were used to characterize

paratopes and epitopes. Constrained heptamers contain

terminal cysteine residues forming a disulfide bridge which

limits the number of degrees of freedom and favors certain

conformations. They are considered to have a higher affinity

and specificity [20]. However, in our selections, the majority

of selected peptides with high affinity toward the target were

linear, probably due to their advantageous flexibility and

adaptability to various structures.

Within characterization of anti-b2GPI paratopes common

motif Q/N, L/P/A/I, W, L/P emerged during elution with

HCAL. Partially, that motif emerged also during the elution

with isolated polyclonal anti-b2GPI and even during the

non-specific elution indicating that fraction of HAv as well as

LAv anti-b2GPI antibodies displays similar paratope as

monoclonal HCAL. Common motifs that emerged during

elution with HAv and LAv anti-b2GPI also suggest similar

pararatopes in both groups of antibodies. The overall high

content (32%) of polar residues (Q, N, H, S, T) capable of

forming hydrogen bonds observed in selected sequences is in

agreement with observations of Jurgec et al. [21], who

emphasized the importance of hydrogen bonds for b2GPI–

anti-b2GPI interaction. Furthermore, the sequences also

contain an excess (9%) of arginine (R), lysine (K) and

asparagine (N) residues in comparison (1%) to aspartic acid

(D) or glutamic acid (E) as already observed for anti-b2GPI’

contact sites [22].

The use of synthetic peptides that would neutralize

patients’ pathogenic anti-b2GPI antibodies could alleviate

symptoms of APS and represents a possible new therapeutic

approach to APS [23]. Since antibodies were used in elution

protocol to displace bound phages from b2GPI, selected

displayed peptides lacked the ability to interfere with anti-

b2GPI binding to b2GPI. That was also the case for peptides

selected using non-specific acidic elution. Due to significant

and selective binding, we assumed that displayed peptides did

bind to the epitopes or at least to the parts of epitopes

recognized by anti-b2GPI, but did not possess adequate

affinity to compete for the binding spot with the anti-b2GPI.

The latter could be due to the inadequate length of phage

peptides which account only for the residues that make a

major contribution to the binding energy [24] and not for the

antibody’s framework residues that also directly and indir-

ectly contribute to the antigen binding [25].

Regarding characterization of epitopes on b2GPI numer-

ous reports indicated that epitope mapping with random

peptide libraries is an effective approach to determine the fine

specificity of monoclonal and even polyclonal antibodies

[26]. To assure the uniform orientation of antibodies, t.i. with

exposed paratopes facing upwards, we applied the anti-b2GPI

to the protein G-coated plates. In this way, the chances of

applied phage particles to encounter a paratope instead of

other antibody’s parts were increased. The selection involving

specific elution with b2GPI turned out to be unsuccessful.

Figure 3. Epitopic regions (black) recognized by HCAL positioned inthe b2GPI’s fifth domain (SVDSDK), by HAv anti-b2GPI in the second(KMDGN) and third domain (YNPLWFV) and LAv anti-b2GPIpositioned between the third and the fourth b2GPI’s domain(NGPANSK).

442 U. Zager et al. Autoimmunity, 2014; 47(7): 438–444

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

Evidently, b2GPI in a solution could not displace peptides

from the binding site and consequently strong affinity clones

were not enriched. In subsequent selections, acidic

elution buffer to disrupt phage – target as well as protein G

– antibody’s Fc interaction was used and thus facilitated

isolation of high affinity clones.

The SLDSDRS heptapeptide presenting putative epitope

recognized by HCAL almost completely corresponded to the

SVDSDK cluster on b2GPI’s fifth domain (Figure 3),

considering that L and V both have non-polar aliphatic side

chains, and R and K are both basic amino acids, exhibiting

positive charge at pH 7.4. The significant inhibitory capacity

of the selected clone confirmed that we identified the true

conformational epitope recognized by HCAL. This result also

agrees with our conclusions regarding the binding of HCAL

to the b2GPI in a solution and its capacity to disrupt the inter-

domain bond [11]. Also, according to Agar et al. [27] b2GPI

in a solution takes the closed circular conformation that can

only be recognized by high affinity antibodies, capable of

disrupting the hydrophilic bond between the first and the fifth

domain and thus exposing their otherwise cryptic epitopes.

The extremely hydrophilic nature of epitope indicates

that HCAL establishes a hydrophilic interaction with its

epitope on fifth domain which could cause the disruption

of b2GPI’s inter-domain bond. However, the selected

sequence also slightly resembles to the dodecapeptide

(SLWQDRLNAMQS) that Kasahara et al. [28] selected

over WB-CAL-1, a source of HCAL’s variable regions.

Using domain deletion mutants of b2GPI, authors concluded

that WB-CAL-1 binds to the b2GPI’s fourth domain,

presumably to the cryptic amino acid cluster which becomes

exposed upon b2GPI binding to negatively charged surface.

The discrepancies could be due to the framework differences

between HCAL and WB-CAL-1, however firm conclusion

cannot be made, since inhibition capacity of dodecapeptide

bearing clone was not evaluated.

The selections of heptapeptides presenting putative

epitopes recognized by the polyclonal anti-b2GPI resulted

in the isolation of one potential mimotope (KMDGNHP)

recognized by LAv anti-b2GPI, and two different potential

mimotopes (QGPAHSK and FNPYWYV) recognized by HAv

anti-b2GPI. Considering the polyclonality of anti-b2GPI, we

have expected a much more diverse set of sequences.

However, the inhibitory capacity of selected sequences

indicated that considerable part of polyclonal anti-b2GPI

recognizes these potential epitopes. Therefore, we suspect that

the selections were guided by the predominant paratopes in

the polyclonal group. All selected sequences contained a

considerable amount of polar residues capable of forming

hydrogen (N, Q, S, Y, W) or electrostatic bonds (K, D, H).

Comparison of the LAv anti-b2GPI binding sequence and the

HAv anti-b2GPI binding sequences revealed that the former

contain more charged residues, while the latter contain more

hydrogen donor residues. According to these results, the

binding of LAv anti-b2GPI could be predominantly mediated

by electrostatic interactions while HAv binding requires the

presence of hydrogen bonds. Our observation is also in

agreement with the fact that during the affinity isolation of

anti-b2GPI the LAv fraction was eluted using ionic strength

gradient which disrupted the electrostatic bonds and the HAv

fraction was eluted with high ionic strength acidic buffer that

could also disrupt the hydrogen bonds [29,30].

A five-amino acid cluster completely corresponding to the

LAv anti-b2GPI binding peptide was identified on the

b2GPI’s third and fourth domains, above on the inner side

of fishhook curvation (Figure 3). We suspect that KMDGN

epitope could not be accessed when b2GPI took the closed

circular conformation (i.e. in solution) which explained the

inability of the LAv anti-b2GPI to bind to the antigen in a

solution [11]. After binding to the negatively charged surface

(microtitre plate) b2GPI opened (converted into fish-hook

like conformation), the LAv anti-b2GPI gained access to the

epitope and could thus be detected in ELISA [27].

Furthermore, terminal amino acids of LAv anti-b2GPI

binding heptapeptide can be found in clusters close proximity

(corresponding to H159P157 of b2GPI sequence), suggesting

they could also be a part of epitope.

The clusters resembling the putative conformational epi-

topes recognized by HAv anti-b2GPI were identified on the

third and second domain (Figure 3). The QGPAHSK peptide

almost completely corresponded to the NGPANSK cluster,

considering that Q and N both have amide group in their side

chains. The FNPYWYV peptide, on the other hand, was best

aligned to the YNPLWFV cluster, with the mismatch in the

fourth residue and alternative aromatic residues on first and

sixth residue (Y4F). Both clusters were accessible

even when b2GPI was in a circular conformation, which

explained how HAv anti-b2GPI recognized antigen in a

solution [11].

To sum up, various heptapeptides presenting putative

paratopes of anti-b2GPI and specific heptapeptides presenting

putative epitopes of b2GPI were determined. Within charac-

terization of anti-b2GPI paratopes, one common motif was

obtained regardless of the elution agent used, indicating that

fraction of HAv as well as LAv anti-b2GPI antibodies

displays similar paratope as monoclonal HCAL. Even more

common motifs were found, when comparing peptides eluted

with HAv and LAv anti-b2GPI (Table 1), which suggests

additional similar paratopes in both groups of antibodies.

In reversed conditions, when identifying b2GPI epitopes, a

narrow set of sequences was obtained even when targeting

polyclonal anti-b2GPI. Both groups HAv and LAv anti-b2GPI

seem to contain prevalent paratopes that guided the selection

into a restricted set of epitope mimetics. Those recognized by

HAv anti-b2GPI predominately contain more hydrogen donor

residues, whereas those recognized by LAv anti-b2GPI

contain more charged side chains. Moreover, the alignment

of epitope mimicking peptides to three-dimensional antigen

structure revealed that HAv anti-b2GPI recognize native

epitopes that are accessible regardless of b2GPI’s conform-

ation whereas the epitope of LAv anti-b2GPI is cryptic and

cannot be accessed when b2GPI takes the closed plasma

conformation. Our study confirmed the heterogeneity of anti-

b2GPI and represents an additional insight to the numerous,

sometimes oppositional reports concerning the anti-b2GPI

specificity.

Declaration of interest

The authors report no conflicts of interest.

DOI: 10.3109/08916934.2014.914176 Anti-b2GPI paratopes and b2GPI epitopes characterization 443

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.

References

1. Dani, M. 2001. Peptide display libraries: design and construction.J. Recept. Signal. Transduct. Res. 21: 469–488.

2. Bottger, V., and A. Bottger. 2009. Epitope mapping using phagedisplay peptide libraries. Methods Mol. Biol. 524: 181–201.

3. Dani, M. 2001. Biological libraries. J. Recept. Signal. Transduct.Res. 21: 447–468.

4. Atsumi, T., O. Amengual, S. Yasuda, et al. 2005. Research aroundbeta 2-glycoprotein I: a major target for antiphospholipidantibodies. Autoimmunity. 38: 377–381.

5. Tincani, A., G. Balestrieri, E. Danieli, et al. 2003. Pregnancycomplications of the antiphospholipid syndrome. Autoimmunity. 36:27–32.

6. Blank, M., D. Faden, A. Tincani, et al. 1994. Immunizationwith anticardiolipin cofactor (beta-2-glycoprotein I) inducesexperimental antiphospholipid syndrome in naive mice.J. Autoimmun. 7: 441–455.

7. Danowski, A., T. S. Kickler, and M. Petri. 2006. Anti-beta2-glycoprotein I: prevalence, clinical correlations, and importanceof persistent positivity in patients with antiphospholipid syn-drome and systemic lupus erythematosus. J. Rheumatol. 33:1775–1779.

8. Tincani, A., C. Casu, S. Cartella, et al. 2010. Antiphospholipidantibody: laboratory, pathogenesis and clinical manifestations.Reumatismo. 62: 65–75.

9. Alijotas-Reig, J., R. Ferrer-Oliveras, M. J. Rodrigo-Anoro, et al.2010. Anti-beta(2)-glycoprotein-I and anti-phosphatidylserineantibodies in women with spontaneous pregnancy loss. Fertil.Steril. 93: 2330–2336.

10. Cucnik, S., T. Kveder, Z. Ulcova-Gallova, et al. 2011. The avidityof anti-beta2-glycoprotein I antibodies in patients with or withoutantiphospholipid syndrome: a collaborative study in the frame ofthe European forum on antiphospholipid antibodies. Lupus. 20:1166–1171.

11. Zager, U., S. Irman, M. Lunder, et al. 2011. Immunochemicalproperties and pathological relevance of anti-beta(2)-glycoprotein Iantibodies of different avidity. Int. Immunol. 23: 511–518.

12. Cucnik, S., T. Kveder, I. Krizaj, et al. 2004. High avidity anti-beta2-glycoprotein I antibodies in patients with antiphospholipidsyndrome. Ann. Rheum. Dis. 63: 1478–1482.

13. Cucnik, S., I. Krizaj, B. Rozman, et al. 2004. Concomitant isolationof protein C inhibitor and unnicked beta2-glycoprotein I.Clin. Chem. Lab. Med. 42: 171–174.

14. Cucnik, S., A. Ambrozic, B. Bozic, et al. 2000. Anti-beta2-glycoprotein I ELISA: methodology, determination of cut-offvalues in 434 healthy Caucasians and evaluation of monoclonalantibodies as possible international standards. Clin. Chem. Lab.Med. 38: 777–783.

15. Mayrose, I., O. Penn, E. Erez, et al. 2007. Pepitope: epitope mappingfrom affinity-selected peptides. Bioinformatics. 23: 3244–3246.

16. Marai, I., A. Tincani, G. Balestrieri, and Y. Shoenfeld. 2005.Anticardiolipin and anti-beta-2-glycoprotein I antibodies.Autoimmunity. 38: 33–38.

17. Ioannou, Y., C. Pericleous, I. Giles, et al. 2007. Binding ofantiphospholipid antibodies to discontinuous epitopes on domain Iof human beta(2)-glycoprotein I: mutation studies includingresidues R39 to R43. Arthritis Rheum. 56: 280–290.

18. Inanc, M., E. L. Radway-Bright, and D. A. Isenberg. 1997. beta 2-Glycoprotein I and anti-beta 2-glycoprotein I antibodies: where arewe now? Br. J. Rheumatol. 36: 1247–1257.

19. Ichikawa, K., A. Tsutsumi, T. Atsumi, et al. 1999. A chimericantibody with the human gamma1 constant region as a putativestandard for assays to detect IgG beta2-glycoprotein I-dependentanticardiolipin and anti-beta2-glycoprotein I antibodies. ArthritisRheum. 42: 2461–2470.

20. Castel, G., M. Chteoui, B. Heyd, and N. Tordo. 2011. Phage displayof combinatorial peptide libraries: application to antiviral research.Molecules. 16: 3499–3518.

21. Jurgec, I., M. Lunder, T. Bratkovic, et al. 2009. Hydrogen bonds arecrucial for the anti-beta 2-glycoprotein I antibody bindings to theantigen. In European Workshop for Rheumatology Research. T.K.Kvien, ed. BMJ Publ. Group, Warsaw–Poland. p. A4.

22. Giles, I., A. Lambrianides, and A. Rahman. 2008. Examining thenon-linear relationship between monoclonal antiphospholipid anti-body sequence, structure and function. Lupus. 17: 895–903.

23. Blank, M., Y. Shoenfeld, S. Cabilly, et al. 1999. Prevention ofexperimental antiphospholipid syndrome and endothelial cellactivation by synthetic peptides. Proc. Natl. Acad. Sci. USA. 96:5164–5168.

24. Janeway, C. 2005. Immunobiology: the Immune System in Healthand Disease. New York: Garland Science.

25. Lo, B. K. 2004. Antibody Engineering: Methods and Protocols.Totowa, NJ: Humana Press.

26. Bottger, V. 2001. Epitope Mapping with Random Peptide Libraries.Heidelberg: Springer Verlag.

27. Agar, C., G. M. van Os, M. Morgelin, et al. 2010. Beta2-glycoproteinI can exist in 2 conformations: implications for our understanding ofthe antiphospholipid syndrome. Blood. 116: 1336–1343.

28. Kasahara, H., E. Matsuura, K. Kaihara, et al. 2005. Antigenicstructures recognized by anti-beta2-glycoprotein I auto-antibodies.Int. Immunol. 17: 1533–1542.

29. Lieberman, M. A. 2006. Marks’ Basic Medical Biochemistry: aClinical Approach. Philadelphia, PA: Lippincott Williams &Wilkins.

30. Elgert, K. D. 2009. Immunology: Understanding the ImmuneSystem. Hoboken, NJ: Wiley-Blackwell.

444 U. Zager et al. Autoimmunity, 2014; 47(7): 438–444

Aut

oim

mun

ity D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cgill

Uni

vers

ity o

n 12

/17/

14Fo

r pe

rson

al u

se o

nly.