Original Paper

Vox Sang 1997;72:52–60

Human Monoclonal Fab FragmentsRecovered from a Combinatorial LibraryBind Specifically to the Platelet HPA-1aAlloantigen on Glycoprotein IIb–IIIa

Chantal Proulx a

Pierre Chartrand b, c

Valérie Roy b

Mindy Goldman a

Francine Décary a, b

Aline Rinfret a, b

a Canadian Red Cross Society,Québec Transfusion Centre-Montréal,

b Department of Microbiology andImmunology, University of Montréal and

c Cancer Institute, Notre-Dame Hospital,Montréal, Canada

E-Mail karger karger.chFax +41 61 306 12 34http://www.karger.ch

1997 S.Karger AG, Basel0042–9007/97/0721–0052 $12.00/0

Dr. Chantal ProulxCanadian Red Cross Society, Blood ServicesQuébec Transfusion Centre-Québec2535, boul. LaurierSte-Foy, Qué. G1V 4M3 (Canada)

Received: March 22, 1996Revised manuscriptreceived: August 29, 1996Accepted: Sept. 2, 1996

.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .AbstractBackground and objectives: Certain clinical conditions are related to the pres-ence of platelet-specific alloantibodies in the patient’s serum. We studied themolecular diversity of HPA-1a antibodies to analyze some peculiarities of thisantibody response. Materials and methods: Human antibody Fab fragmentsthat bind to the platelet alloantigen HPA-1a on glycoprotein IIb–IIIa (GPIIbIIIa)were generated by using a recombinant phage display system. We established animmunoglobulin G1, kappa combinatorial library from the peripheral blood lym-phocytes of a person undergoing a severe posttransfusion purpura. Results:Characterization of Fab clones selected from the fifth round of antigen-specificpanning of this library demonstrates a highly specific reactivity to the HPA-1aalloantigen. The nucleotide sequence analysis of representative HPA-1a-specificclones reveals at least 3 distinct VL and 3VH gene segments that present an exten-sive degree of mutation as demonstrated by comparison of gene usage and ho-mologies to the nearest germline genes. Conclusions: These human HPA-1a-specific Fab reagents should allow us to better understand the molecular mecha-nism involved in HPA-1a alloimmunization... . . . . . . . . . . . . . . . . . . .

Introduction

Clinical syndromes such as posttransfusion purpura(PTP) and neonatal alloimmune thrombocytopenia(NAITP) are consequences of an immunization induced byplatelet alloantigens following a blood transfusion or duringpregnancy, respectively. These manifestations are related tothe presence of platelet-specific alloantibodies in the pa-tient serum. In a large number of cases of NAITP and mostcases of PTP, these alloantibodies are reactive with theHPA-1a epitope present at the N-terminal portion of the pla-telet membrane glycoprotein IIb–IIIa (GPIIbIIIa). TheHPA-1 specificity is associated with a single base mutation

resulting in a leucine (HPA-1a) to proline (HPA-1b) substi-tution at residue 33 [1].

The immune response of HPA-1b homozygous individu-als to the HPA-1a alloantigen is rather intriguing. HPA-1a isnot an immunodominant epitope considering the significantdifference between the expected and observed frequencies(only 10–30% of the expected frequency [2, 3]) of immuni-zation induced in HPA-1b homozygous mothers giving birthto HPA-1a-positive babies. Yet, the HPA-1a immune re-sponse is quite diverse with respect to the isotype and lightchain type distribution and the specific antibody titer pro-duced [4]. In contrast to the Rh alloimmunization (red celldisorder hemolytic disease of the newborn), the first epi-

Human HPA-1a-Specific Fab Fragmentsfrom Phage Display

Vox Sang 1997;72:52–60 53

sode of NAITP occurs in 60% of the cases in the first preg-nancy. Over 90% of subsequent HPA-1a-positive fetuseswill be thrombocytopenic. Moreover, the antibody responsecan be long-lasting as demonstrated by the detection of spe-cific HPA-1a antibodies in sera of mothers in their late 60s[our unpubl. observations]. Furthermore, only a minority ofHPA-1a-immunized mothers will give birth to thrombocy-topenic babies [5, 6]. Although our study on the titer andisotype distribution of the HPA-1a antibody response in im-munized mothers showed interesting features, no predictionof the severity or the occurrence of thrombocytopenia innewborns could be made [4]. As well, no significant differ-ences in titer or isotype distribution were observed in theanitbody response of NAITP and PTP cases in spite of thefact that the route of immunization (via placental transfervs. blood transfusion) and the outcome (destruction of het-erologous platelets vs. hetero- and autologous platelets) arequite different. Previous investigators have suggested theexistence of a mixture of specificities in sera of HPA-1a al-loimmunized individuals. Heterogeneous forms of anti-body differing by their functionnal or qualitative character-istics (Ig class) have been demonstrated in the same serumof acute phase PTP patients [7–9]. Other evidence for heter-ogeneity in the humoral response to HPA-1a was providedby results of studies on the interaction of complement withplatelets pretreated with HPA-1a antibodies [10, 11], on thestructure of the alloantigen HPA-1a [12] and on the inhib-itory activity of antibodies directed against fibrinogen bind-ing to GPIIbIIIa [13].

We were interested in further analyzing this intriguingantibody response by studying more directly the moleculardiversity of HPA-1a antibodies. We used an antibody phagedisplay expression system [14] to isolate monoclonalHPA-1a-specific antibodies from the peripheral blood lym-phocytes of a PTP patient. In doing so, we generated a panelof human monoclonal antibody Fab fragments directedagainst HPA-1a alloantigen on the platelet GPIIbIIIa by an-tigen selection from a random combinatorial library ex-pressed on the surface of filamentous phage. This is the firstreport of the isolation and sequence analysis of humanHPA-1a antibodies obtained from both VH and VL gene rep-ertoires of an immunized individual.

Materials and Methods

Blood SampleThe library was prepared from peripheral blood lymphocytes ob-

tained from an individual undergoing a severe PTP. The HPA-1b ho-mozygous woman had a strong HPA-1a antibody titer in her serum asfirst determined in platelet suspension immunofluorescence test

(PSIFT) and in monoclonal antibody-specific immobilization of pla-telet antigen assays [data not shown]. Indirect ELISA using purifiedHPA-1a GPIIbIIIa indicated that these specific HPA-1a antibodieswere immunoglobulins G1, kappa [IgG1 (k)] [4].

Library Construction and PanningThe construction of an IgG1 (k) Fab library on the surface of phage

was performed by use of the pComb3 system as described elsewhere[14]. The plasmid was kindly provided by the Scripps Institute. A sec-ond PCR amplification was performed as described by Williamson etal. [15] in order to increase the efficiency of restriction enzyme cuttingof PCR-amplified products. This amplification is done using appropri-ate extension primer pairs which contain poly(GA) tail 5 to the se-quence of the original primers increasing the number of bases betweenthe cutting site and the end of the molecule. PCR products obtainedfrom this second amplification were pooled and the heavy or lightchain band was cut, gel-purified, and subjected to restriction enzymedigestion prior to ligation into the pComb3 vector. Ligated DNA (lightchain and Fd ligated into the vector sequentially) was transformed intoEscherichia coli XLI-Blue. The library was amplified and the bacte-rial culture was infected with helper phage VCSM13.

The panning procedure was performed as described by Barbas et al.[14]. Fifty microliters of purified HPA-1a GPIIbIIIa (2.5 µg/ml in0.05 M carbonate buffer pH 9.6) were used to coat two wells of a micro-titer plate (Nunc, polysorp) overnight at 4 C. After 1wash with distilledwater, wells were filled with 50 µl of purified Fab-presenting phage for2 h at room temperature. Wells were then thoroughly washed with Tris-buffered saline/0.5% Tween and bound phages were recovered after a10-min incubation at room temperature with the elution buffer (0.1 MHCl-adjusted with glycine to pH 2.2/BSA 1 mg/ml). Eluted Fab-pre-senting phage and helper phage were used for reinfection of bacteriaand reamplification of the library for subsequent rounds of panning.

Preparation and Characterization of Soluble Fab FragmentsThe production of soluble Fab fragments from phagemid DNA was

performed as described by Barbas et al. [14]. Briefly, phagemid DNAfrom each selected clones was isolated and digested with SpeI andNheI to remove the gIII-encoding fragment. The cut vector was gel-purified, self-ligated and transformed into XL1-Blue. Each clone wasgrown under selection at 37 C in SuperBroth medium for 6 h afterwhich isopropyl â-D-thiogalactopyranoside (1 mM) was added andthe cultures were further incubated O/N at 30 C. Supernatants con-taining soluble Fab were prepared and stored at –20 C until use.

Supernatants containing soluble Fab fragments were tested in in-direct ELISA as reported in Proulx et al. [4]. Microtiter plates werecoated overnight at 4 C with 50 µl of purified HPA-1a or HPA-1bGPIIbIIIa (2.5 µg/ml of carbonate buffer, pH 9.6), tetanus toxoid (5 µg/ml) and a preparation of major outer membrane proteins from Hae-mophilus influenzae type b (OMP Hib, 5 µg/ml). Wells were washedonce with 0.05% Tween 20 in phosphate-buffered saline (PBS) andblocked with 100 µl of 1% BSA/0.05% Tween in PBS for 1 h at 37 C.Supernatants containing Fab fragments diluted 1/5 in PBS, positive an-ti-HPA-1a (PTP patient) and negative normal control sera diluted1/20,000 and 1/100 in PBS, respectively, and murine monoclonal anti-bodies specific to GPIIbIIIa (kindly provided by Réal Lemieux, Cana-dien Red Cross Society, Québec Centre), GPIIb (clone SZ.22, Bio/CanScientific) and GPIIIa (clone SZ.21, Bio/Can Scientific) diluted1/1,000 in PBS were added to blocked wells in 50-µl volumes for 1 h at37 C. Plates were washed 3 times and wells were filled with 3% BSA-PBS containing alkaline phosphatase-conjugated antihuman kappa-

54 Vox Sang 1997;72:52–60 Proulx/Chartrand/Roy/Goldman/Décary/Rinfret

specific antibody (Southern Biotech) or antimouse immunoglobulins(Jackson Immunoresearch) diluted 1/250 and 1/5,000, respectively,and incubated for 1 h at 37 C. After 3 washes, 100 µl of substrate wereadded to each well. The optical density was read after a 30-min in-cubation with the substrate at room temperature using the MicroplateReader II, Multiskan MCC/340 (DuPont).

Western blots were performed after separation of purified HPA-1aor HPA-1b GPIIbIIIa by SDS-PAGE in 1.5-mm-thick slab 6% poly-acrylamide gels under nonreducing conditions. The resolved proteinswere transferred electrophoretically onto nitrocellulose paper (Hy-bond, Amersham). The nitrocellulose paper was soaked in 1% milk(Carnation)/1% polyvinyl-pyrrolidone (PVP-360, Sigma) in PBS for1 h at room temperature with agitation. The blots were incubated withsupernatants containing Fab fragments diluted 1/2 in 1% powderedskim milk-PBS, positive and negative sera as mentioned above diluted1/100 and murine anti-GPIIbIIIa mAb diluted 1/250 for 1 h at 37 C in aMini-PROTEAN Multiscreen Apparatus (Bio-Rad). Following thor-ough washing with 0.05% Tween-PBS, each channel was filled eitherwith peroxidase-conjugated antihuman kappa-specific antibody(Southern Biotech) or antimouse immunoglobulins (Jackson Immu-noresearch Lab.) diluted 1/4,000 in 3% milk-PBS and incubated for 1 hat 37 C. The blots were taken out of the multiscreen apparatus andsoaked in 0.05% Tween-PBS for several washes over a period of 1 h atroom temperature. The nitrocellulose papers were then drained off andcovered with ECL Western blotting detection reagents (AmershamLife Science) for 1 min. The papers were drained of excess reagent,placed between SaranWrap sheets and exposed for 15 s at room tem-perature to a Kodak XAR-5 film in a Kodak X-Omatic cassette fittedwith intensifying screens.

Supernatants containing HPA-1a antibodies prepared from repre-sentative clones were confirmed in PSIFT [16]. Briefly, platelet sus-pensions were prepared from blood samples of HPA-1a-positive andHPA-1a-negative (HPA-1b/1b) donors. The platelet-rich plasma waswashed 3 times in PBS-EDTA, incubated for 2 min at room temper-ature in paraformaldehyde-EDTA 1% followed by 2 washes in PBS-EDTA. One hundred microliters of a platelet supension adjusted to

300 106 platelets/ml were incubated with an equal volume of un-diluted HPA-1a-positive serum (PTP patient), normal control serum orsoluble Fab fragments for 30 min at 37 C. Following 3 washes withPBS-EDTA, platelets were incubated with FITC-conjugated antihu-man kappa-specific antibodies (Southern Biotech) diluted 1/20 inPBS-EDTA for an additional 30 min at 37 C. Washed (2 ) plateletswere mounted in a drop of glycerol-PBS and preparations were ob-served under a Zeiss fluorescent microscope.

Nucleic Sequence AnalysisNucleic acid sequencing was performed only for a limited number

of clones using an automated sequencing facility (Perkin-Elmer’s ABI373 DNA sequencer, using the PRISM Ready Reaction DyeDeoxyTerminator Cycle Sequencing kit). These clones were selected by thepattern of bands produced from PCR reamplification of the VH/VL

genes followed by digestion with the frequent-cutting enzyme BstNI.To obtain the heavy-chain sequences, primers from the 5 vector se-quence (VH-5 primer: 5 GGATTGTTATTACTCGCTGC 3 ) and theCH1 constant region (CH3 primer: 5 GTCGTTGACCAGGCAGCC-CAG 3 ) were used. For the light chain sequences, primers from the 5vector sequence (VK-5 primer: 5 CACTGGCTGGTTTCGCTAC 3 )and the CK primer (CK-3 primer: 5 CAACTGCTCATCAGATGG 3 )were used. Sequence analysis was performed using the GCG (Genet-ics Computer Group) package, version 8.0 [17].

Results

Isolation and Analysis of Anti-HPA-1a Fab FragmentsAn IgG1(k) library with a repertoire of 1 107 represen-

tatives was prepared from a patient undergoing a PTP. Thechoice of Fd class and light chain type was made on the ba-sis of results obtained from a previous study on the class,subclass and light chain-type analysis of HPA-1a-specificantibodies present in PTP and NAITP cases [4]. Five roundsof panning of this library were performed against purifiedHPA-1a GPIIbIIIa. After the fourth round, a significant in-crease in the relative yield ( 50 ) could be seen [data notshown]. Since the fifth panning did not give better yieldsthan the fourth one, the panning procedure was stopped atthis point and soluble Fab fragments were obtained for anal-ysis.

Soluble Fab fragments were prepared from 24 colonieschosen at random from plates of bacterial cultures infectedwith phage-Fab recovered after the 5th panning. All theseclones produced from 10 to 50 µg Fab fragments/ml as dem-onstrated by results obtained from a quantitative ELISA us-ing a known concentration of human IgG [data not shown].Results from indirect ELISA analysis of supernatants con-taining soluble Fab fragments showed a reactivity of 20 outof 24 Fab-producing clones with the purified protein. Table1 shows data obtained from the reactivity of 6 representativeclones against different antigens in indirect ELISA, Westernblots and PSIFT. Results in indirect ELISA demonstratedthat clones were highly specific to the HPA-1a alloantigenas shown by the lack of reactivity against HPA-1b alloanti-gen, tetanus toxoid or OMP Hib. The 4 remaining clonesshowed a reactivity near background with all antigens usedincluding the HPA-1a alloantigen (represented by clone FabNo. 8). The positive control serum (HPA-1a serum from thePTP patient) showed strong reactivity against purifiedHPA-1a GPIIbIIIa but not against HPA-1b. The murine anti-GPIIbIIIa mAb reacted similarly against both alleles of thepurified GPIIbIIIa but not against tetanus toxoid or OMPHib as expected. The reactivity of the negative serumagainst purified GPIIbIIIa (HPA-1a and HPA-1b) was nearbackground level. Both negative and positive sera reactedagainst tetanus toxoid and OMP Hib due to the expectedpresence of specific antibodies from previous immuniza-tions (in the case of tetanus toxoid) and of probable cross-reactive antibodies (in the case of OMP Hib).

In immunoblotting, representative clones specific toHPA-1a GPIIbIIIa and the HPA-1a-positive PTP serum re-acted with a 96-kD band corresponding to the GPIIIa sub-unit of the HPA-1a GPIIbIIIa but not with GPIIIa on HPA-1bGPIIbIIIa (table 1; fig.1, lanes b–g). The murine mAb reac-

Human HPA-1a-Specific Fab Fragmentsfrom Phage Display

Vox Sang 1997;72:52–60 55

Table 1. Results of the reactivity of representative Fab-producing clones against purified GPIIbIIIa or unrelated proteins in indirect ELISAand Western blots and against fixed HPA-1a-positive and HPA-1a-negative platelets in PSIFT

Antibodies ELISA (OD 405 nm) Western blots PSIFT

HPA-1a HPA-1b TT OMP Hib HPA-1a HPA-1b HPA-1a/1aplatelets

HPA-1b/1bplatelets

Soluble Fab fragmentsFab No. 8 0.02 0.02 0.02 0.02 negative negative – –Fab No. 6, 14, 23, 24, 19 1.50 0.02 0.02 0.02 band at 96 kD negative +++ –

Human seraPTP patient 1.80 0.06 0.80 1.00 band at 96 kD negative ++ –Normal control 0.05 0.07 1.80 1.50 negative negative – –

Murine mAbsAnti-GPIIbIIIa 1.40 1.00 0.05 0.05 band at 96 kD band at 96 kD n.d. n.d.Anti-GPIIb 1.90 1.20 0.05 0.05 n.d. n.d. n.d. n.d.Anti-GPIIIa 1.90 0.20 0.05 0.05 n.d. n.d. n.d. n.d.

TT = Tetanus toxoid; OMP Hib = preparation of major outer membrane proteins from Haemophilus influenzae type b; n.d. = not done.

Table 2. Comparison of gene usage andhomologies for heavy (VH) V regions ofHPA-1a Fab fragments

Fab fragments VH family Nearest VH Homology withVH DNA, %

D JH

Group A/No. 6–14 VH 3 DP-29 84.8 D .21.9 andPA-4

4a

Group B/No. 23–24 VH 4 VIV-4 91.7 D4 5bGroup C/No. 19 VH 3 DP-31 87.8 DA5 5b

Table 3. Comparison of gene usage and homologies for light (VL)V regions of HPA-1a Fab fragments

Fab fragments VL family Nea-rest VL

Homology withVL DNA, %

JL

Group A/No. 6–14 kIII Vg 89.7 k4Group B/No. 23–24 kI L12a 90.8 k2Group C/No. 19 kIII Vg 89.0 k5

ted with both proteins, as expected (table 1, fig.1, lane i).The representative clone (Fab No. 8) which was negative inindirect ELISA using different antigens and the negativenormal control serum were also included as negative con-trols (table 1, fig.1, lanes a and h).

Immunofluorescence tests confirmed the results on thespecificity of representative clones (table 1, fig. 2) as shownby the reactivity of HPA-1a Fab fragments againstHPA-1a/1a platelets but not against HPA-1b/1b platelets.

DNA Sequencing AnalysisTo determine how many clones isolated from the HPA-1a

GPIIbIIIa were identical or closely related, plasmid DNAwas prepared, VH and VL sequences were PCR-amplifiedand analyzed by the BstNI digestion. At least three differentpatterns were discerned with a ratio of 10:8:2 (groups A, Band C). Figure 3 shows the pattern of 3 representativeclones: Fabs No. 6, 24 and 19. Nucleic acid sequencing of theheavy and light chain V regions of Fab fragments was per-formed on 2 representative clones of the first two groups (A

and B) and 1 of group C. Sequence analysis showed that allthree groups of clones were using different VDJ and VJ seg-ments (tables 2, 3, fig. 4). For instance, Fab clones 6 and 14(group A) used the VH3 family which has DP-29 as the clos-est germline gene, associated with a light chain from theVkIII family, which has Vg as the closest germline gene.Clones 24 and 23 (group B) showed a completely differentusage of VH/VL family combination: VH4 family with thenearest germline gene VIV-4 in association with a light

56 Vox Sang 1997;72:52–60 Proulx/Chartrand/Roy/Goldman/Décary/Rinfret

Fig. 1. Luminescent immunoblots. Puri-fied HPA-1a (A) and HPA-1b (B) GPIIbIIIawere probed with soluble Fab fragments byWestern blot. Lanes a–f = Fab fragments pro-duced by clones 8, 6, 14, 23, 24, and 19, respec-tively; lane g = HPA-1a serum from the PTPpatient (diluted 1/100); lane h = serum from anegative normal control individual (diluted1/100); lane i = murine anti-GPIIbIIIa mAb(diluted 1/250). Numbers on the sides showsize in kilodaltons which corresponds to therelative molecular weigth of the GPIIIa sub-unit.

A B

Fig. 2. Immunofluorescence of HPA-1a/1a platelets. Platelet suspensions were in-cubated with soluble Fab fragments fromclone 6 (A) or with an HPA-1a serum fromthe PTP patient (B). A similar reactivity wasobserved with clones 14, 24, 19 and 23. No flu-orescence was observed when usingHPA-1b/1b platelet suspensions incubatedwith the clones mentioned above or when us-ing HPA-1a/1a platelet suspensions withclone 8 [data not shown].

A

B

Human HPA-1a-Specific Fab Fragmentsfrom Phage Display

Vox Sang 1997;72:52–60 57

Fig. 3. BstNI fingerprinting of representative Fab clones. The VH

(lanes 1–3) and VL sequences (lanes 7–9) were amplified from plasmidDNA, the products were digested with BstNI (VH: lanes 4–6; VL: lanes10–12) and analyzed on an agarose gel. M = 1 kb DNA Ladder (GibcoBRL). Representative clone 6 in lanes 1, 4, 7 and 10, clone 24 in lanes 2,5, 8 and 11 and clone 19 in lanes 3, 6, 9 and 12.

Fig. 4. Predicted amino acid sequences of HPA-1a-specific FabVH and VL domains in comparison to closest knowngermline. A VH sequences. B VL sequences.

A

B

chain from the VkI family with L12a as the closest germlinegene. The smallest group (group C) represented by clone 19used a combination V gene somewhat apparented to the oneused by group A clones, VH3/DP-31 and VKIII/Vg (Jk5 in-stead of Jk4 for group A) being the closest germline genes.Percentage of homologies of cloned V region genes withnearest VH or VL germline genes indicated the presence ofintermediate to high degrees of point mutations (tables 2, 3)resulting in more than 20 amino acid substitutions per chain(fig. 4).

Discussion

The combinatorial approach has proven useful for thegeneration of highly specific human monoclonal antibodiesagainst certain protein antigens otherwise difficult to obtain(anti-RhD antibodies [18]; anti-self antibodies [19]; anti-blood group antigen antibodies [20]). Whether or not anti-bodies isolated from this approach have the same VH/VL

pairing as found in ‘natural’ antibodies has still not beenresolved. Some groups report that antibodies generated

58 Vox Sang 1997;72:52–60 Proulx/Chartrand/Roy/Goldman/Décary/Rinfret

from combinatorial libraries use different V region genesthan the ones obtained by cellular fusion [21]. Conversely,results from other studies indicated a similar VH/VL combi-nation in both sets of antibodies [22]. In any case, as pointedout elsewhere, the fact that libraries are constructed fromimmunologically amplified mRNA and not DNA raises theprobability of obtaining the original pairings of heavy andlight chains [23]. We were interested in further analyzingthe HPA-1a immune response which can lead to severe clin-ical syndromes such as NAITP and PTP. The mechanismsinvolved in the pathophysiology of these diseases are yet tobe elucidated. Because of difficulties in obtaining stableHPA-1a-producing hybridomas we considered phage dis-play technology as the best alternative to generate mono-clonal antibodies to the HPA-1a alloantigen from peripheralblood lymphocytes of an immunized individual. We be-lieved that these antibodies would be useful reagents to fur-ther delineate mechanisms of effector functions involved inPTP and NAITP.

This study reports the generation of a panel of humanHPA-1a-specific Fab fragments isolated from an antibodyphage library constructed from peripheral blood lympho-cytes of an acute phase PTP patient. Twenty out of 24 ran-domly selected Fab-producing clones showed a strong reac-tivity against purified HPA-1a GPIIbIIIa in ELISA. As dem-onstrated in indirect ELISA using different antigens, theseclones were highly specific to the HPA-1a alloantigen. Re-sults from Western blots and PSIFT confirmed the ELISAdata regarding the specificity of PTP monoclonal Fab frag-ments towards the HPA-1a alloantigen.

The generation of human antiplatelet antibodies havebeen reported previously by other groups [24–28]. Inter-estingly, Denomme et al. [24, 25] obtained and character-ized a human monoclonal autoantibody specific to plateletGPIIb derived from normal human lymphocytes. Similarly,Kunicki et al. [26] generated an anti-GPIIb mAb with acomparable restricted reactivity, from a patient who hadautoreactive anti-GPIb/IX in the serum. These results in-dicated that although autoantibody-producing B cells spe-cific to platelet antigens were not found to be actively syn-thesizing antibody at the time they were harvested, thesecells are nevertheless present in peripheral blood. Whetherthese B cells play a role in the pathogenicity of immunethrombocytopenia remains to be determined. In compari-son, none of the clones selected from the antigen-specificpanning of our phage library reacted with epitopes otherthan HPA-1a on GPIIbIIIa. A possible explanation may bethat although all present in the library, HPA-1a-specificclones are of high affinity compared to other GPIIbIIIaclones which could have been missed during the stringent

panning procedure. Conversely, considering that Fab frag-ments generated from our library are exclusively of theIgG1 class, autoreactive antibodies frequently of the IgMclass could be absent.

DNA sequence analysis of 5 representative PTP clonesindicated the isolation of 3 unique clones which used 3 dis-tinct light chains associated with 3 different heavy chains.Comparison between heavy and light V regions of HPA-1aFab fragments and the nearest germline genes showed theoccurrence of a relatively high degree of mutation. PTP oc-curs primarily in HPA-1b homozygous women who havehad previous immunologic sensitization to the antigenthrough pregnancy or through prior transfusion [29]. In thisparticular case, the PTP woman had received multiple trans-fusions over a 4-year period. This history suggests an anam-nestic response which is characterized by the selection ofhigh-affinity antibodies. Our results from DNA sequencingof V regions of PTP clones are certainly compatible with asecondary response. Interestingly, another group had ob-tained one HPA-1a-specific single chain Fv (scFv) clonefrom the combination of two VH libraries from the lympho-cytic mRNA of an HPA-1a-alloimmunized individual with anonimmune VL gene repertoire derived from the lympho-cytic mRNA of two normal individuals [27]. Results indi-cated that a different VH family had been used in the case ofthis HPA-1a-specific clone. The VH gene showed 19 pointmutations when compared to VH1family DP75 (which is thenearest germline gene) although only 8 of these point muta-tions produced amino acid substitutions. Their results couldindicate that they have isolated a less mutated clone whencompared to some of the clones we obtained [which showedaround 40 point mutations (data not shown) resulting inabout 20 amino acid substitutions, fig. 4].

PTP and NAITP are the consequences of an alloimmuni-zation induced by a relatively simple antigen, resulting froma single amino acid change. Strong evidence from previousstudies indicates that the alloantibodies induced by this sim-ple antigen are hetergeneous by their functional or qualita-tive characteristics [7–11]. For example, Valentin et al. [12]have recently shown a differential reactivity of antibodiespresent in sera of cases of PTP and NAITP against a recom-binant GPIIbIIIa which lacked one of the disulfide bridge(Cys5-Cys435). Other studies demonstrated the capacity ofmost PTP sera but of no NAITP sera to inhibit fibrinogenbinding to platelets [13]. Our results confirm these studiesshowing additional evidence of heterogeneity of theHPA-1a response, at least qualitatively, since we were ableto isolate three distinct groups of antibodies specific toHPA-1a alloantigen. Ongoing studies on the reactivity ofthese specific reagents in competitive ELISA using purified

Human HPA-1a-Specific Fab Fragmentsfrom Phage Display

Vox Sang 1997;72:52–60 59

Fab fragments and polyvalent PTP and NAITP sera as wellas more detailed sequencing of these and other selectedHPA-1a clones will provide more information on theHPA-1a epitope(s) on GPIIbIIIa. The biological activity ofthese reagents will also be analyzed in an attempt to betterunderstand the mechanisms underlying the pathophysiolo-gy of PTP and NAITP.

In conclusion, we have been successful in producinghighly specific human HPA-1a Fab fragments from a combi-natorial library generated from an alloimmunized individu-al undergoing a severe PTP. We hope that these antibodieswill serve to further delineate this intriguing antibody re-sponse.

Acknowledgments

We thank Dr. G. Boucher and Dr. R. Bazin from the Québec Trans-fusion Centre at Québec for their precious help in the analysis of DNAsequences. We thank J. Cotton-Montpetit for the production of puri-fied HPA-1a and HPA-1b GPIIbIIIa. We also wish to thank techniciansSylvain Chiroux and Madeleine Malette, Immunobiology Departmentof the Québec Transfusion Centre at Montreal, for their great help inPSIFT analysis. C. Proulx is the recipient of a fellowship from theMedical Research Council of Canada. A. Rinfret is the recipient of aBayer/Canadian Red Cross Society/Medical Research Council ofCanada scholarship. V. Roy is the recipient of a studentship from theFonds pour la Formation de Chercheurs et l’Aide à la Recherche(FCAR). This work was supported by a grant from the Canadian RedCross Research and Development Fund (MO 01.93).

.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .References

1 Newman RJ, Derbes RS, Aster RH: The humanplatelet alloantigens, PlA1 and PlA2, are associ-ated with a leucine33/proline33 amino acid poly-morphism in membrane glycoprotein IIIa, andare distinguishable by DNA typing. J Clin In-vest 1989;83:1778–1781.

2 Mueller-Eckhardt C, Mueller-Eckhardt G, Wil-len-Ohf H, Horz A, Kuenzlen E, O’Neill GJ,Schendl D: Immunogenicity of and immune re-sponse to the human platelet antigen is stronglyassociated with HPA.B8 and DR3. Tissue Anti-gens 1985;26:71–76.

3 Blachette VC, Chen L, De Friedberg ZS, HoganVA, Tendel E, Décary F: Alloimmunization tothe PlA1 platelet antigen: Results of a prospec-tive study. Br J Haematol 1990;74:209–215.

4 Proulx C, Filion M, Goldman M, Bradley A,Devine D, Décary F, Chartrand P: Analysis ofimmunoglobulin class, IgG subclass and titre ofHPA-1a antibodies in alloimmunized mothersgiving birth to babies with or without neonatalalloimmune thrombocytopenia. Br J Haematol1994;87:813–817.

5 Mueller-Eckhardt C, Grubert A, Weisheit M,Mueller-Eckhardt G, Kiefel V, Kroll H, SchmidtS, Santoso S: 348 cases of suspected neonatalalloimmune thrombocytopenia. Lancet 1989;i:363–366.

6 Décary F: Is HLA-DR3 a risk factor in P1A1-negative pregnant women? First CanadianWorkshop and Conference on Platelet Serolo-gy; in Décary F, Rock G (eds): Current Studiesin Haematology and Blood Transfusion. Basel,Karger, 1986, vol 52, pp 78–86.

7 Pegels JG, Bruynes ECE, Engelfriet CP, vondem Borne AEGK: Posttransfusion purpura: Aserological and immunochemical study. Br JHaematol 1981;49:521–530.

8 Berney SI, Metlcalfe P, Wathen NC, WatersAH: Post-transfusion purpura: A serologicaland immunochemical study. Br J Haematol1985;61:627–632.

9 Tanning E, Skov F: Elution of anti-Zwa (PlA1)from autologous platelets after normalization ofplatelet count in post-transfusion purpura. VoxSang 1991;60:40–44.

10 Cines DB, Schreiber AD: Effect of anti-PlA1 an-tibody of human platelets. I. The role of com-plement. Blood 1979;53:567–577.

11 Aster RH: The immunologic thrombocytope-nias; in Kunicki TJ, George JN (eds): Platelet Im-munobiology. Philadelphia, Lippincott, 1989,p 387.

12 Valentin N, Visentin GP, Newman PJ: Involve-ment of the cysteine-rich domain of glycopro-tein IIIa in the expression of the human plateletalloantigen, PlA1: Evidence for heterogeneity inthe humoral response. Blood 1995;85:3028–3033.

13 Beadling MV, Herman JH, Stuart MJ, Keashen-Schnell M, Miller JL: Fetal bleeding in neonatalalloimmune thrombocytopenia mediated by an-ti-PlA1 is not associated with inhibition of fibri-nogen binding to platelet GPIIbIIIa. Am J ClinPathol 1995;103:636–641.

14 Barbas CF, Kang AS, Lerner RA, Benkovic SJ:Assembly of combinatorial antibody librairieson phage surfaces: The gene III site. Proc NatlAcad Sci USA 1991;88:7978–7982.

15 Williamson RA, Burioni R, Sanna PP, PartridgeLJ, Barbas CF, Burton D: Human monoclonalantibodies against a plethora of viral pathogensfrom single combinatorial libraries. Proc NatlAcad Sci USA 1993;90:4141–4145.

16 von dem Borne AEG, van Leeuwen EF, vonRiesz LE, van Boxtel CJ, Engelfriet CP: Neona-tal alloimmune thrombocytopenia: Detectionand characterization of the responsible antibod-ies by the platelet immunofluorescence test.Blood 1981;57:649–656.

17 Devereux J, Haeberli P, Smithies O: A compre-hensive set of sequence analysis programs forthe VAX. Nucleic Acids Res 1984;12:387–395.

18 Dziegiel M, Nielsen LK, Andersen PS, Blanch-er A, Dickneiss E, Engberg J: Phage displayused for gene cloning of human recombinantantibody against the erythrocyte surface anti-gen, rhesus D. J Immunol Methods 1995;182:7–19.

19 Griffiths AD, Malmqvist M, Marks JD, ByeJM, Embleton MJ, McCafferty J, Baier M, Hol-liger KP, Gorick BD, Hughes-Jones NC, Hoo-geboom HR, Winter G: Human anti-self anti-bodies with high specificity from phage displaylibraries. EMBO J 1993;12:725–734.

20 Marks JD, Ouwehand WH, Bye JM, Finnern R,Gorick BD, Voak D, Thorpe SJ, Hughes-JonesNC, Winter G: Human antibody fragments spe-cific for human blood group antigens from aphage display library. Biotechnology 1993;11:1145–1149.

21 Ames RS, Tornetta MA, McMillan LJ, KaiserKF, Holmes SD, Appelbaum E, Cusimano DM,Theisen TW, Gross MS, Jones CS, Silverman C,Porter TG, Cook RM, Bennett D, Chaiken IM:Neutralizing murine monoclonal antibodies tohuman IL-5 isolated from hybridomas and a fil-amentous phage Fab display library. J Immunol1995;154:6355–6364.

22 Caton AJ, Koprowski H: Influenza virus he-magglutinin-specific antibodies isolated from acombinatorial expression library are closely re-lated to the immune response of the donor. ProcNatl Acad Sci USA 1990;87:6450–6454.

23 Burton DR, Barbas CF, Persson MAA, KoenigS, Chanock RM, Lerner RA: A large array ofhuman monoclonal antibodies to type 1 humanimmunodeficiency virus from combinatorial li-braries of asymptomatic seropositive individu-als. Proc Natl Acad Sci USA 1991;88:10134–10137.

60 Vox Sang 1997;72:52–60 Proulx/Chartrand/Roy/Goldman/Décary/Rinfret

24 Denomme GA, Smith JW, Kelton JG, Bell DA:A human monoclonal autoantibody to plateletglycoprotein IIb derived from normal humanlymphocytes. Blood 1992;79:447–451.

25 Denomme GA, Mahmoudi M, Cairns E, BellDA: Immunoglobulin V region sequences oftwo human antiplatelet monoclonal autoantibo-dies derived from B cells of normal origin. J Au-toimmun 1994;7:521–535.

26 Kunicki TJ, Furihata K, Kekomaki R, Scott JP,Nugent DJ: A human monoclonal autoantibodyspecific from human platelet glycoprotein IIb(integrin αIIb) heavy chain. Hum AntibodiesHybridomas 1990;1:83–89.

27 Griffin H, Ouwehand WH: A human monoclo-nal antibody specific for the leucine-33 (PlA1,HPA-1a) form of platelet glycoprotein IIIa froma V gene phage display library. Blood 1995;86:4430–4436.

28 Ishida F, Gruel Y, Brojer E, Nugent DJ, KucickiTJ: Repertoire cloning of a human IgG inhibitorof alpha IIB beta 3 function. The OG idiotype.Mol Immunol 1995;32:613–622.

29 Kunicki TJ, Beardsley DS: The alloimmunethrombocytopenias: Neonatal alloimmunethrombocytopenic purpura and post-transfu-sion purpura; in Coller BS (ed): Progress in He-mostasis and Thrombosis. Philadelphia, Saun-ders, 1989, pp 203–232.