intl.elsevierhealth.com/journals/thre

REGULAR ARTICLE

Highly potent anti-human GPVI monoclonalantibodies derived from GPVI knockoutmouse immunizationB

Yutaka Matsumoto a, Hisao Takizawa a, Xiaoqi Gong a, Sang Le a, SimonLockyer a, Keiji Okuyama b, Michinori Tanaka c, Masuhiro Yoshitake a,Narendra N. Tandon a, Junichi Kambayashi a,*

a Otsuka Maryland Medicinal Laboratories, 9900 Medical Center Drive, Rockville, MD 20850, USAb Otsuka GEN Research Institute, Otsuka Pharmaceutical Co., Ltd., 463-10 Kagasuno, Kawauchi-cho,Tokushima, 771-0192, Japanc Medicinal Chemistry Research Institute, 463-10 Kagasuno, Kawauchi-cho, Tokushima, 771-0192, Japan

Received 12 October 2005; received in revised form 17 January 2006; accepted 27 January 2006Available online 6 March 2006

0049-3848/$ - see front matter D 200doi:10.1016/j.thromres.2006.01.023

Abbreviations: CHO, Chinese hamplatelet rich plasma; TBS-T, Tris-buffeB A part of this study was presented inin Sydney, Australia.* Corresponding author. Tel.: +1 240

E-mail address: junichik@otsuka.co

KEYWORDSAntibodies;Glycoprotein VI;Knockout mouse;Platelet;Collagen

Abstract Recent progress in the understanding of thrombus formation hassuggested an important role for glycoprotein (GP) VI in this process. To clarify theexact role in detail, it is necessary to use specific, high affinity inhibitory antibodies.However, possibly due to the conserved structure of GPVI among species, it has beendifficult to obtain potent antibodies. In this study, we developed highly potent anti-human GPVI monoclonal antibodies using GPVI knockout mice for immunization. Fabfragments of these antibodies, named OM1 and OM2, potently inhibit collagen-induced platelet aggregation. The IC50 values for OM1 and OM2 are 0.6F0.05 and1.7F0.5Ag/mL, respectively, showing potency greater than, or equal to that ofabciximab (1.7F0.3Ag/mL), an anti-GPIIb/IIIa antibody. Fab fragments of OM1 andOM2 also potently inhibit collagen-induced ATP release, thromboxane A2 formation,and platelet adhesion to immobilized collagen under static and flow conditions.Interestingly, platelet aggregation induced with collagen-related peptide waspotently inhibited by OM2 but not OM1, indicating that OM1 recognizes an epitopethat is different from collagen-related peptide-binding site on GPVI. These resultssuggest that OM1 and OM2 may be useful tools to understand the role of GPVI in

Thrombosis Research (2007) 119, 319—329

6 Elsevier Ltd. All rights reserved.

ster ovary; CRP, collagen-related peptide; PMBCs, peripheral mononuclear blood cells; PRP,red saline supplemented with Tween 20; TX thromboxane; vWf, von Willebrand factor.the 20th Congress of the International Society on Thrombosis and Haemostasis on Aug, 11, 2005

683 3360; fax: +1 301 721 7360.m (J. Kambayashi).

Y. Matsumoto et al.320

thrombus formation. Furthermore, these antibodies have the potential to bedeveloped as a new class of therapeutic tool.D 2006 Elsevier Ltd. All rights reserved.

Introduction

Arterial thrombosis is thought to be triggered bythe adhesion and activation of platelets on exposedsubendothelium, following the rupture of an ath-erosclerotic plaque [1,2]. Although several suben-dothelial macromolecules such as fibronectin,vitronectin, laminin and von Willebrand factor(vWf) can provide a suitable surface for plateletadhesion, collagen is considered to be the mostimportant constituent, supporting not only plateletadhesion, but also acting as a strong plateletactivator, resulting in platelet aggregation andsubsequent thrombus formation [3,4]. Althoughintegrin a2h1 (GPIa/IIa) and GPVI were identifiedas platelet surface components in the early 1980s,their role as major collagen receptors was onlyrecently established. Based upon studies usingplatelets from patients lacking a specific receptor[5—13] and recent studies with platelets fromknockout mice [14—16], a2h1 and GPVI are nowconsidered to be major platelet collagen receptors.

Conflicting hypotheses have been proposed re-garding the relative contribution of the two recep-tors [17,18]. A central role for GPVI in platelet—collagen interactions has been proposed in arevised two-site model in which primary contactis mediated primarily by GPIb—vWf and GPVI—collagen interactions. In the second step, theGPVI—collagen interaction initiates cellular activa-tion, followed by the conversion of integrins from alow- to a high-affinity state and the release ofsecondary mediators such as ADP and thromboxane(TX) A2, which further potentiate the activationprocess and the formation of a stable thrombus[18]. Our earlier studies provide further evidence infavor of a model in which GPVI plays a major role inprimary adhesion to fibrillar collagen and subse-quent activation and secretion; we demonstratedthat blockade of GPVI by polyclonal anti-GPVIantibody Fab fragments results in the inhibition ofdivalent cation-independent platelet adhesion totype I fibrillar collagen, subsequent formation ofTXA2 and the expression of activated GPIIb/IIIacomplex on the platelet surface [19,20].

There is evidence to suggest a crucial role forGPVI in arterial thrombosis and hyperplasia from invivo studies using GPVI-depleted [21] and FcRgchain knockout mice also lacking GPVI [22]. These

observations indicate that GPVI may be a potentialtarget for the treatment of cardiovascular disease.Since GPVI is expressed only on megakaryocytesand platelets [23], anti-GPVI antibodies should besafe to use for therapeutic purposes. Additionally,since GPVI deficiency is not associated with anymajor bleeding risk in humans or mice [8—13,24],anti-GPVI therapy may provide a better risk/benefit ratio. So far, there are a few reports thatdescribe the development of anti-human GPVIantibodies using wild-type mice [25] or phagedisplay techniques [26,27]. However, their poten-cies as GPVI antagonists, as measured by theinhibition of collagen-induced platelet aggregationare not comparable with the commercially avail-able anti-GPIIb/IIIa antibody abciximab, which mayreflect the limitation of methodologies availablefor antibody production. This report describes thedevelopment of two high affinity monoclonal anti-bodies, produced by immunization of GPVI knock-out mouse. The Fab fragments of these antibodiespotently inhibit collagen-induced platelet aggrega-tion and attenuate both primary adhesion andthrombus formation on immobilized collagen underflow conditions.

Materials and methods

Chemicals

Horm collagen was purchased from Nycomed,Germany. Protein G-Sepharose, MonoQ (source15beads) and size exclusion matrix Superdex 75were purchased from Amersham. Papain, iodoace-tamide, cysteine and FITC-labeled goat anti-mouseIgG (g chain specific) and goat anti-rabbit IgG werepurchased from Sigma Chemicals. Ultra low IgGfetal calf serum was purchased from GIBCO. TheLDH assay kit was purchased from Promega.Recombinant hirudin was obtained from HoechstPharmaceuticals. Collagen-related peptide (CRP)was synthesized as previously reported [28]. Con-vulxin was purified from lyophilized snake venom(Crotalus durissus terrificus, Soerensen Laborato-ries, Brazil) by size exclusion chromatography asdescribed earlier [29]. Hybridoma for IV.3, an anti-FcgRII monoclonal antibody was obtained fromATCC.

Anti-GPVI antibodies from GPVI knockout mice 321

Animals

GPVI knockout mice were developed by a targeteddisruption of the 2nd and 3rd exons of the GPVI gene[16]. These animals are fertile and show no abnor-mality other than the complete lack of plateletresponsiveness to collagen in assays such as colla-gen-induced platelet aggregation and platelet ad-hesion to immobilized collagen. In addition, wefound that a proportion of GPVI knockout mice wasprotected against lethal thromboembolism, inducedby the infusion of a mixture of collagen andepinephrine [16]. These animals were procuredand bred in our animal facility under our Institution-al Animal Care and use Committee (IACUC) protocol.

Establishment of CHO cells stably expressingGPVI and FcR;

Chinese hamster ovary (CHO) cells stably express-ing GPVI and FcRg (GPVI/FcRg-CHO cell) wereestablished by transfection of pTarget vector(Promega) containing full-length human GPVI cDNAand pcDNA3.1(+)zeocin vector (Invitrogen) contain-ing full-length FcRg cDNA using Lipofectamine 2000(Invitrogen). Cells expressing both receptors wereselected by culturing in medium supplementedwith G418 and zeocin. Expression of GPVI wasconfirmed by FACS analysis with an Epic Altra FACSanalyzer (Beckman Coulter), using a polyclonalhuman anti-GPVI antibody provided by Dr. Okuma[8]. The presence of GPVI was further confirmed byimmunoblotting using biotinylated convulxin. De-tection of FcRg expression was performed byimmunoblotting using commercially available anti-FcRg polyclonal antibody (Upstate Biotechnology).

Preparation of monoclonal antibodies

GPVI knockout mice were immunized intraperito-neally with GPVI/FcRg-CHO cell. Monoclonal anti-bodies were produced by conventional hybridomamethods using P3U1 cells as the fusion partner.Screening was performed by FACS analysis usingGPVI/FcRg-CHO cell. FACS analysis was also per-formed with non-transfected wild-type CHO cells toeliminate clones against epitopes other than GPVI.Positive clones were further characterized by FACSanalysis andWestern blotting using human platelets.

Purification of IgG and preparation of Fabfragments

Hybridomas producing anti-GPVI antibodies weregrown in DMEM/F12 medium containing 10% fetal

calf serum (containing negligible amounts of bovineIgG: b1Ag/mL). IgG was purified by protein G-Sepharose affinity chromatography on a Waters 650protein separation system. For functional assays,protein G-bound IgG was eluted by low pH glycine—HCl (pH 2.75), collected directly into basic solutionand dialyzed against saline.

F(abV)2 fragments were prepared by papaindigestion: a solution of IgG (5mg/mL) in 100mMcitric acid, pH 6.5 and 5.0mM EDTAwas digested for15hrs at 378C with cysteine-free pre-activatedpapain at an enzyme/IgG ratio of 1:50 (w/w).Digestion was quenched with freshly preparediodoacetamide and F(abV)2 and Fab fragments wereseparated from undigested IgG and Fc fragment byion exchange chromatography on MonoQ. Fab andF(abV)2 eluted from the column at low salt (50—80mM) while undigested IgG and Fc were elutedbetween 120—200mM salt. F(abV)2-containing frac-tions were reduced/alkylated to obtain monovalentFab fragments [30]. Finally, Fab fragments werepurified to homogeneity by size exclusion chroma-tography on Superdex 75.

Washed platelet preparation

Blood was collected in 1/10volume of 3.8% triso-dium citrate anticoagulant and platelet rich plasma(PRP) was obtained by centrifugation at 180�g for15min at room temperature. Washed plateletswere prepared from PRP by citrate wash proceduredescribed previously [19]. Briefly, citric acid andPGE1 were added to PRP to final concentration of4mM and 1Ag/mL, respectively. The PRP was spunat 1520�g for 10min to sediment platelets. Plate-let pellet was washed twice with citrate washbuffer (4.26mM NaH2PO4, 7.46mM Na2HPO4,4.77mM trisodium citrate, 2.5mM citric acid,128mM NaCl, 3.5mg/mL bovine serum albumin,pH 6.5 and 100ng/mL PGE1). The pellet was thensuspended at 1�109/mL in HEPES-Tyrode buffer(10mM HEPES, 136.7mM NaCl, 5mM glucose,2.6mM KCl, 13.8mM NaHCO3, 0.26mM NaHPO4,3.5mg/mL bovine serum albumin, pH 7.35 and100ng/mL PGE1) for FACS analysis, Western blot-ting, affinity assay and static platelet adhesionassay.

FACS analysis

Washed platelets in HEPES-Tyrode buffer (1�109/mL) were supplemented with 1mM EDTA and EGTA.Platelets (50AL) were incubated with OM1, OM2 orcontrol IgG (10Ag/ml) for 60min at room temper-ature. Samples were diluted with 1mL wash buffer

Y. Matsumoto et al.322

(HEPES-Tyrode buffer supplemented with EDTA/EGTA) and centrifuged at 1520�g for 5min. Plate-let pellets were suspended into a final volume of50AL wash buffer containing FITC-labeled goatanti-mouse IgG (1:50 dilution: Sigma ChemicalCo.) and incubated for 45min at room tempera-ture. Samples were diluted with 1 mL wash bufferand centrifuged at 1520�g for 5min to sedimentplatelets and to remove free FITC-labeled antibody.Platelet pellets were suspended into 1 mL washbuffer and analyzed in a flowcytometer (Beckman-Coulter Epic Altra, Miami, FL). Peripheral mononu-clear blood cells (PMBCs) were isolated from wholeblood anticoagulated with trisodium citrate usingFicoll-Paque Plus (Amersham) and treated similarly.

Western blotting

Washed platelets in HEPES-Tyrode buffer (1�109/mL) supplemented with 1 mM EDTA and EGTA werecentrifuged at 1520�g for 5min at room tempera-ture. Platelet pellet was washed twice with Tris-buffered saline (10mM Tris—HCl, pH 7.5 and 140mMNaCl) supplemented with 1mM EDTA and EGTA andsuspended in the same buffer at 2�109/mL.Platelets were lysed for 30min at room tempera-ture with an equal volume of lysis buffer (2% SDS,1mM EDTA, 1mM EGTA, 20Ag/mL leupeptin). Plate-let proteins (20AL lysate) were resolved by SDS-PAGE on a 4—20% acrylamide gel and transferred tonitrocellulose membrane. Membrane was blockedwith TBS-T (10mM Tris—HCl, 140mM NaCl, 0.5%Tween 20) supplemented with 5% skim milk powderfor 60min. Nitrocellulose membrane was incubatedwith OM1 or OM2 (2Ag/ml) at 48C for 15h, washedextensively with TBS-T and probed with HRPconjugated goat anti-mouse IgG for 1h at roomtemperature. Membrane was washed three timeswith large excess of TBS-T and immune reactivebands were visualized using an enhanced chemilu-minescence detection system (ECL-Amersham Bio-tech, Little Chalfont, UK).

Affinity assay of antibodies

Labeling of antibodies with 125I was performed bythe Iodo-beads method, as described by themanufacturer (Pierce, IL). Binding affinity deter-mination was performed by the method of theprevious report [31]. Briefly, washed platelets(5�108/mL) were incubated with 1nM 125I-labeledIgG in the presence and absence of variousconcentrations of unlabelled homologous IgG (0—500nM) for 1h. Free and bound radioactivities wereseparated by layering the mixture over BSA (10% in

saline) solution and centrifuged at 15,000�g for10min. After careful removal of both the superna-tant and the BSA cushion, the tip of the tube wascut off and the radioactivity of pellet was counted.Eight to ten competitive binding isotherms wereconstructed in triplicate to evaluate the binding ofindividual IgGs. The data were analyzed using non-linear regression analysis software (Prism, Graph-Pad Software, CA).

Measurement of platelet aggregation

Platelet number in PRP was counted and adjustedto 3�108 platelets/mL with platelet poor plasma.All measurements were performed within 4hrs ofblood collection. Aggregation studies were per-formed in an AG10 aggregometer (Kowa, Japan).PRP was incubated with antibody for 10min at 378Cprior to the addition of collagen or CRP andaggregation was monitored for an additional 5—10min. Since responsiveness to agonist variedamong individuals, the agonist dose required toinduce 70—90% aggregation was determined foreach donor prior to the evaluation of antibodies.

Measurement of ATP release andthromboxane A2 formation

ATP release and TXA2 formation were measured inparallel. ATP release from collagen-activated plate-lets was quantified by the increase in PRP fluores-cence in the presence of luciferin and luciferaseusing a lumi-aggregometer (Chrono-log). Tenminutes after collagen addition, 500AL of stopsolution (50mM EDTA, 2mM indomethacin in130mM NaCl) was added to 200AL of PRP toterminate TXA2 formation. The suspension wascentrifuged at 1000�g for 15min at 4 8C. Thesupernatant was sampled and frozen at �208C untilTXB2 measurement (a stable metabolite of TXA2).TXB2 was quantified using a commercially availablekit (Thromboxane B2 Biotrak Assay, Amersham).

Platelet adhesion to immobilized collagen

The effects of anti-GPVI antibody Fab fragments onplatelet adhesion to immobilized collagen understatic conditions were examined using the proce-dure described earlier [19], with slight modifica-tion. Briefly, washed platelets (4�108/mL) wereincubated with Fab fragments for 30min at roomtemperature prior to their addition (50AL/well) tocollagen-coated wells (2Ag/well) in 96 well-platesin the presence of PGE1 (100ng/mL) to inhibitplatelet aggregation. After 60min incubation,

Anti-GPVI antibodies from GPVI knockout mice 323

unadhered platelets were removed by gentlewashing with buffered saline and adhesion wasquantified by determining the LDH content ofadhered platelets.

Platelet adhesion to immobilized collagen underflow conditions was also measured. Collagen-coatedPetri dishes (35mm diameter) were prepared byincubating with collagen (40Ag/mL, sodium chlo-ride 150mM, glucose 5mM, acetic acid 5mM) for 4hat 48C. The dishes were then blocked with BSA (BSA2% (w/v), sodium chloride 140mM, Tris 10mM, pH7.4). Dynamic adhesion assays were performed at ashear rate approximating arterial flow (2600s�1)using a parallel plate flow chamber (Glycotech,Rockville, MD). Whole blood, anticoagulated withhirudin (50units/mL), was drawn through the flowchamber with a syringe pump (Harvard Apparatus,Holliston, MA) for 2min; adhered platelets wererinsed with saline, fixed with glutaraldehyde (0.5%v/v) and stained with toluidine blue/sodium borate(0.05% w/v, 5min). Stained platelets were rinsedwith phosphate-buffered saline and observed under400� magnification; ten consecutive non-overlap-ping fields from the same relative positions for eachexperiment were captured using a digital cameraand recorded to disk. Images were then subjectedto software analysis (SigmaScan Pro, SPSS, Chicago,IL) to determine the percentage surface areacovered by adhered platelets.

Results

Identification of anti-GPVI antibodies

We screened 3900 hybridoma clones obtained bythe fusion of P3U1 cells with splenocytes of GPVIknockout mice immunized with GPVI/FcRg-CHOcell. Among these, we obtained 8 clones thatproduced anti-GPVI monoclonal antibodies. Basedon the rank order of potency of the Fab frag-ments of these antibodies to inhibit collagen-induced platelet aggregation, we chose the 2most potent clones, named OM1 and OM2, forfurther characterization.

Specificity and affinity of anti-GPVIantibodies

OM1 and OM2 showed positive shifts in FACS analysisusing human platelets (Fig. 1A, left). These anti-bodies did not show any interaction with PMBCs,demonstrated by the lack of any rightward shift(Fig. 1A, right). In Western blotting using humanplatelet lysates, these antibodies clearly detected

a single band with a molecular mass of 58kDa innon-reducing conditions (Fig. 1B).

We examined the binding affinities of these 2anti-GPVI antibodies to human platelets using 125I-labeled IgGs. OM1 and OM2 bound to plateletswith high affinity with Kd values of 1.7F0.2�10�9M (0.26F0.03Ag/mL) and 0.7F0.1�10�9M(0.11F0.01Ag/mL), respectively. The number ofbinding sites on a single platelet for these anti-bodies was calculated from the Bmax value of thesaturating binding curve and the specific activityof the radio-labeled antibodies and is approxi-mately 1900 and 1500 for OM1 and OM2, respec-tively. There are some reports that quantifiedGPVI density on human platelets [32—34] and thenumber per platelet ranged from 1260 to 3730,probably reflecting the methodology used in eachstudy.

Effects on collagen- and CRP-induced humanplatelet aggregation

OM1 and OM2 whole IgG activated human plateletsand induced aggregation at low concentration(b1Ag/mL). We examined the effect of Fab frag-ment of IV.3, an anti-FcgRII antibody, on OM1 andOM2 whole IgG-induced platelet aggregations. Incase of OM1, the Fab fragment of IV.3 (20Ag/mL)had no effect on whole IgG-induced plateletaggregation, suggesting that OM1 IgG inducesplatelet aggregation by cross-linking of neighboringGPVI molecules. In contrast, the Fab fragment ofIV.3 partially inhibited OM2 IgG-induced plateletaggregation, suggesting that OM2 whole IgG inducesplatelet aggregation by both cross-linking of GPVI-GPVI and FcgRII-GPVI.

The Fab fragments of OM1 and OM2 potentlyinhibited collagen-induced human platelet aggre-gation in a concentration-dependent manner (Fig.2). OM1 is more potent than abciximab, while OM2is equipotent in inhibiting collagen-induced plate-let aggregation (Fig. 2, Table 1). Interestingly,platelet aggregation induced by CRP was inhibitedpotently by the Fab fragment of OM2 but not OM1(Table 1). In addition, the Fab fragments of OM1and OM2 also inhibited convulxin-induced humanplatelet aggregation, with IC50~1.0Ag/mL (datanot shown).

Effects on collagen-induced ATP release andthromboxane A2 formation in humanplatelets

The Fab fragments of OM1 and OM2 potentlyinhibited collagen-induced ATP release from hu-

OM

1O

M2

100 101 102 103

32

coun

ts

FL1-H

100 101 102 103

29

A

B

coun

ts

FL1-H

100 101 102 103

30

coun

tsFL1-H

100 101 102 103

28

coun

ts

FL1-H

PMBCsPlatelets

Figure 1 Specificity of anti-GPVI antibodies. (A) FACS charts of OM1 (upper panels) and OM2 (lower panels). Resultswith human platelets (left) and PMBCs (right) are shown. Dotted areas: control sample (treated with non-immune IgGat 10Ag/mL), Grey areas: anti-GPVI antibodies (10Ag/mL)-treated samples. Both antibodies show clear shifts inplatelets but not with PMBCs. (B) Western blot with human platelet lysates under non-reduced condition. Bothantibodies (2Ag/mL) detect a single band with molecular mass of 58kDa. On the left is shown the molecular mass ofmarker proteins.

Y. Matsumoto et al.324

man platelets (Table 2). OM1 showed an inhibi-tion of more than 90% at 1Ag/mL. OM2 achievedan inhibition of more than 90% at 3Ag/mL. Theeffect of abciximab on ATP release was weakerthan that of either of the anti-GPVI antibody Fabfragments.

Fab fragments of OM1 and OM2 potentlyinhibited collagen-induced TXA2 formation in hu-man platelets (Table 3). OM1 showed more than90% inhibition at 1Ag/mL. OM2 attained an in-hibition of more than 90% at 3Ag/mL. In con-trast, abciximab did not inhibit TXA2 formationat 3Ag/mL.

Effects on adhesion onto collagen understatic and flow conditions

Adhesion under static condition was carried out inthe presence (GPIa/IIa- and GPVI-dependent) andabsence (GPVI-dependent) of Mg2+ (1mM). In con-trol samples, the number of adhered platelets(measured by LDH activity) in the absence of Mg2+

was 15—20% of that in the presence of Mg2+. Mg2+-independent adhesion was inhibited by OM1 andOM2 Fab fragments at very low concentrations (Fig.3). OM1 inhibited adhesion by 50% at 0.1Ag/mL andOM2 showed 90% inhibition at 0.1Ag/mL. Mg2+-

-20

0

20

40

60

80

100

Agg

rega

tion

(%)

control

0.3 µg/mL

0.5 µg/mL

0.7 µg/ml

OM1 Fab

-20

0

20

40

60

80

100

Agg

rega

tion

(%) control

1 µg/mL

3 µg/mL

5 µg/mL

OM2 Fab

0 2 4 6 8 10 12 14-20

0

20

40

60

80

100

Time (min)

Agg

rega

tion

(%) control

1 µg/mL

3 µg/mL

5 µg/mL

abciximab

Figure 2 Effects of Fab fragments of anti-GPVI antibodies and abciximab on collagen-induced aggregation of humanplatelets. Typical aggregation traces obtained from 1 donor are shown. Collagen (2Ag/mL) was added to platelet richplasma at 5min of measurement (shown by arrow).

Anti-GPVI antibodies from GPVI knockout mice 325

dependent adhesion, however, which is also depen-dent on GPIa/IIa, required relatively higher con-centrations of the anti-GPVI antibody Fabfragments (Fig. 3). OM1 gave less than 50%inhibition even at 100Ag/mL. However, OM2caused approximately 50% inhibition of adhesionat 1Ag/mL.

Table 1 Effects of Fab fragments of anti-GPVI anti-bodies on collagen- and CRP-induced aggregation ofhuman platelets

Antibody IC50 (Ag/mL)

Collagen-inducedaggregation

CRP-inducedaggregation

OM1 0.6F0.0 N10OM2 1.7F0.5 0.5F0.1Abciximab 1.7F0.3 8.9F4.4

Efficacies are expressed by IC50 values. Values are meansFSDfrom 3 different donors. Platelet aggregation was inducedwith 2Ag/mL of collagen or 31.25—62.5ng/mL of CRP.

The effect of anti-GPVI antibody Fab fragmentson platelet adhesion was also tested under highshear (2600s�1) conditions. Fab fragments of OM1and OM2 reduced the percentage surface areacoverage by platelets on immobilized collagen

Table 2 Effects of Fab fragments of anti-GPVI anti-bodies on collagen-induced ATP release from humanplatelets

Antibody Concentration(Ag/mL)

%inhibition(meanFSD)

OM1 1 97.6F2.13 96.8F6.3

OM2 1 63.6F41.63 97.7F2.7

Abciximab 1 18.0F13.53 56.9F31.7

Results were obtained from 4 different donors and shown as %inhibition.

Depending on the aggregation response of individual donor,ATP release was induced with 1—4Ag/mL of collagen.

Table 3 Effects of Fab fragments of anti-GPVI anti-bodies on collagen-induced thromboxane A2 formationin human platelets

Antibody Concentration(Ag/mL)

ng/3�108 platelets(meanFSD)

Control 104.3F37.4OM1 1 5.0F3.4

3 2.9F2.7OM2 1 31.5F36.5

3 5.0F3.3Abciximab 1 77.8F35.1

3 106.7F70.9

Results were obtained from 4 different donors.

Depending on the aggregation response of each donor,thromboxane A2 formation was induced with 1—4Ag/mL ofcollagen.

Thromboxane B2, a stable metabolite of thromboxane A2,was quantified.

0

10

20

30

40

50

60

70

80

90

1 2 10 30 50 10 30 50 10 30 50

abciximab(µg/mL)

% S

urfa

ce C

over

age

8

4 4 4

control OM1(µg/mL)

OM2(µg/mL)

Figure 4 Effects of Fab fragments of anti-GPVI anti-bodies on platelet adhesion to immobilized collagenunder flow condition. Whole blood anticoagulated withhirudin was drawn through a collagen-coated flowchamber at high shear rate (2600s�1) for 2min. Thenumbers on top of each bar represents number of donorsfor each condition. Control measurements were con-ducted at the beginning (control 1) and the end (control

Y. Matsumoto et al.326

under flow conditions in a concentration-depen-dent manner (Fig. 4). The anti-GPVI antibody Fabfragments induced dramatic changes in the size andmorphology of the platelet aggregates compared tocontrol. Abciximab also prevented the formation ofaggregates and a uniform layer of single platelets

0

20

40

60

80

100

120

0.1 0.2 1 10 20 50 100 0.1 0.2 1 10 20 50 100

Fab (µg/ml)

Adh

esio

n (%

of

Con

trol

)

OM1 OM2

Mg2+

-independent

Mg2+

-dependent

Figure 3 Effects of Fab fragments of anti-GPVI anti-bodies on platelet adhesion to immobilized collagenunder static condition. Platelet adhesion was measuredin the absence (black column) or presence (grey column)of 1mM Mg2+. Adhesion was quantified by LDH activity ofadhered platelets and expressed as percent of controlsample. Saline was added to the control sample. Resultsare meansFSD of 3 different donors.

2) of each experiment, showing the similar surfacecoverage in the beginning and the end of each experi-ment. Results are meansFSD of % surface coverage.Representative adhesion pattern and morphology ofplatelet thrombus shown above each group is in thepresence of 30Ag/mL of Fab fragment.

could be seen associated with the collagen fibers(Fig. 4).

Discussion

Recent progress in the understanding of the mech-anism of thrombus formation has suggested apivotal role for GPVI in the activation of plateletsfollowing the adhesive process [16,35,36]. Howev-er, most of these results were obtained throughstudies in mice, whose hemostatic system issomewhat different from humans [37]. Therefore,it is necessary to obtain specific and potent anti-GPVI monoclonal antibodies that can block theinteraction between GPVI and collagen in humanand other larger mammals to better understand therole of GPVI.

So far, there have been a few reports thatdescribe anti-GPVI monoclonal antibodies obtainedby routine methods, using normal mice for immu-nization [25], or by phage display technology[26,27]. However, they are not very potent as

Anti-GPVI antibodies from GPVI knockout mice 327

GPVI antagonists when compared to GPIIb/IIIaantagonists such as abciximab, probably due tothe conserved sequences in GPVI among species.One way to circumvent this issue is to utilizeantigen-deficient animals for immunization. Infact, there are some reports where highly potentantibodies have already been obtained using thisstrategy [38,39]. In this study, we obtained highlyspecific and potent anti-human GPVI monoclonalantibodies using GPVI knockout mice for immuni-zation. The affinity of these antibodies is veryhigh, with Kd value ~10�9 M. Fab fragments ofOM1, one of our antibodies, inhibit collagen-induced platelet aggregation with an IC50 of0.6Ag/mL, which indicates that this antibody isat least 10 times more potent than previouslyreported anti-GPVI antibodies, though the mea-surement conditions are slightly different amongstudies [25—27]. The efficacy of OM1 Fab fragmentin inhibiting collagen-induced human platelet ag-gregation is even greater than that of abciximab(IC50 value: 1.7Ag/mL).

In this study, we compared the effects of Fabfragments of anti-GPVI antibodies and abciximab oncollagen-induced responses in human platelets.Although the efficacy of the Fab fragment of OM2to inhibit collagen-induced platelet aggregation iscomparable to that of abciximab, its effects oncollagen-induced ATP release and TXA2 formationare much greater. This phenomenon can beexplained by the difference in the modes ofinhibition by these antibodies. Anti-GPVI antibodiesinhibit the interaction of GPVI with collagen, thusshutting down the initial signaling events that leadto platelet activation and release reactions. How-ever, anti-GPIIb/IIIa antibodies inhibit the interac-tion of GPIIb/IIIa with fibrinogen/vWF, thusblocking aggregation and the subsequent outside-in signal. The collagen-induced release reactionand TXA2 formation are evoked mainly by GPVI—collagen interaction and possibly, in part, byoutside in signaling following GPIIb/IIIa-fibrino-gen/vWF interaction. These results indicate thatanti-GPVI antibodies may inhibit the cascade ofplatelet activation by various kinds of agonists thatare released from, or produced on plateletsfollowing initial activation, such as ADP, serotonin,TXA2 and thrombin [18,19,40].

Although the role of GPVI in adhesion understatic condition in the absence of Mg2+ is obvious,as shown in this study (Fig. 3), the precise role ofGPVI in the primary adhesion to immobilizedcollagen under flow conditions has been disputed.Nieswandt et al. reported that blockade of GPVI ordepletion of GPVI by JAQ1 antibody, inhibitsplatelet adhesion to collagen under flow in mice

in vitro [41]. Furthermore, Massberg et al., usingintravital fluorescence microscopy, demonstratedthat blockade or depletion of GPVI abolished stablearrest of platelets onto denuded subendothelium inmice in vivo [21]. In contrast, using GPVI knockoutmice, Kato et al. showed that GPVI deficiency doesnot affect the primary adhesion of platelets toimmobilized collagen under flow, although throm-bus formation is largely inhibited [15]. Siljander etal. also reported that blockade of GPVI by antibody10B12 does not inhibit primary platelet adhesion toimmobilized collagen under flow in human [42]. Inthis study, we showed that Fab fragments of OM1and OM2 decreased surface coverage by humanplatelets in a concentration-dependent manner(Fig. 4). However, we did not observe completeinhibition of the primary adhesion at concentra-tions up to 50Ag/mL, which greatly exceeds thatrequired for complete inhibition of aggregation,indicating the relatively lesser contribution of GPVIto primary adhesion under flow as compared to theessential role in platelet activation. These discrep-ancies between reports might be due to thedifferences in methods of GPVI blockade, flowconditions and systems employed for evaluatingadhesion, etc. Although there seems to be no cleardifferences between the effects of anti-GPVI anti-body Fab fragments and abciximab on the reduc-tion of surface coverage under flow, more detailedanalyses of the activation state of adhered plate-lets (i.e. P-selectin expression or annexin V bind-ing) may distinguish their effects.

Interestingly, the relative potency of the 2 anti-GPVI antibody Fab fragments was not the same foreach assay system. The Fab fragment of OM1inhibited collagen-induced platelet aggregation,ATP release and TXA2 synthesis more potently thanOM2. The Fab fragment of OM2, on the other hand,was slightly more potent than OM1 in inhibitingplatelet adhesion, especially under static condi-tions. These observations suggest that there may bedomains on the GPVI molecule that are function-specific. Further studies are required to completelycharacterize these two antibodies.

There have been some reports that suggestedmultiple collagen-binding sites on the GPVI mole-cule [43,44]. In this study, we showed that the Fabfragment of OM1 inhibited collagen-induced plate-let aggregation, but not CRP-induced plateletaggregation (Table 1). This result suggests thatGPVI possesses a collagen-binding site that doesnot interact with CRP. To our knowledge, OM1 isthe only anti-GPVI monoclonal antibody, amongthe antibodies reported so far, that inhibitscollagen-induced platelet aggregation, but notCRP-induced platelet aggregation. Identification

Y. Matsumoto et al.328

of the epitope recognized by OM1 may demon-strate the presence of a second collagen-bindingsite on GPVI.

The introduction of GPIIb/IIIa antagonists intoclinical use initiated a new era in the treatment ofcardiovascular complications [45—47]. However,there are still some issues to be resolved. Firstly,as exemplified by the result of the GUSTO IV-ACStrial [48], the use of GPIIb/IIIa antagonists in thetreatment of acute coronary syndromes [49] doesnot appear to be as beneficial as their use as anadjunct in percutaneous coronary intervention[45—47]. This suggests that other interactions, suchas the platelet—collagen interaction, may playsignificant roles in these clinical conditions inaddition to platelet—platelet interactions. Second-ly, there is an increased incidence of bleedingcomplications associated with GPIIb/IIIa antagonisttherapy, despite careful patient monitoring [46—49]. Patients lacking GPVI [8—13] and GPVI-defi-cient mice [15,16,24] both show a low incidence ofbleeding diatheses, therefore it is reasonable toassume that blockade of GPVI by anti-GPVI anti-bodies will not result in severe bleeding complica-tions compared to the inhibition of GPIIb/IIIa.Therefore, GPVI antagonists have the potentialfor development as a new class of therapeuticagents for thrombotic disorders.

Acknowledgements

We would like to thank Dr. Masashi Niimi (MolecularBiology, Otsuka Maryland Medicinal Laboratories)and Ms. Lian Tao (Molecular Biology, Otsuka Mary-land Medicinal Laboratories) for their valuablesuggestions in molecular biological techniques andMr. Maurice Guertin (Technical support, OtsukaMaryland Medicinal Laboratories) and Ms. MarinaSamuel (Technical support, Otsuka Maryland Me-dicinal Laboratories) for their help in antibodypurification and Fab fragment preparations. Wealso appreciate Dr. Bing Sun (Molecular Biology,Otsuka Maryland Medicinal Laboratories) and Dr.Naomasa Yamamoto (Tokyo Metropolitan Instituteof Medical Science) for providing vector for FcRgcDNA and GPVI cDNA, respectively.

References

[1] Fuster V, Badiman L, Badiman JJ, Chesebro JH. Thepathogenesis of coronary artery disease and the acutecoronary syndrome. N Engl J Med 1992;326:310–8.

[2] Conti CR, Mehta JL. Acute myocardial ischemia: role ofatherosclerosis, thrombosis, platelet activation, coronary

vasospasm and altered arachidonic acid metabolism. Cir-culation 1987;75:V84–95.

[3] Baumgartner HR. Platelet interaction with collagen fibrilsin flowing blood: reaction with human platelets with alphachymotrypsin-digested subendothelium. Thromb Haemost1977;31:1–16.

[4] Hawiger J. Macromolecules that link platelets followingvessel injury. Ann N Y Acad Sci 1987;509:131–41.

[5] Nieuwenhuis HK, Akkerman J, Houdljk WP, Sixma JJ. Humanplatelets showing no response to collagen failed to expressglycoprotein Ia. Nature 1985;318:470–2.

[6] Kehrel B, Balleisen R, Kokott R, Mesters R, Stenzinger W,Clemetson KJ, et al. Deficiency of intact thrombospondinand membrane glycoprotein Ia in platelets with defectivecollagen-induced aggregation and spontaneous loss ofdisorder. Blood 1988;71:1074–8.

[7] Handa M, Watanabe K, Kawai Y, Kamata T, Koyama T, NagaiH, et al. Platelet unresponsiveness to collagen: involve-ment of glycoprotein Ia—IIa (a2h1 integrin) deficiencyassociated with a myeloproliferative disorder. ThrombHaemost 1995;73:521–8.

[8] Sugiyama T, Okuma M, Ushikubi F, Senaaki S, Kanaji K,Uchino H. A novel platelet aggregating factor found in apatient with defective collagen-induced platelet aggre-gation autoimmune thrombocytopenia. Blood 1987;69:1712–20.

[9] Moroi M, Jung SM, Okuma M, Shinmyozu K. A patient withplatelets deficient in glycoprotein VI that lack bothcollagen-induced platelet aggregation and adhesion. J ClinInvest 1989;84:1140–5.

[10] Arai M, Yamamoto N, Moroi M, Akamatsu N, Fukutake K,Tanoue K. Platelets with 10% of the normal amounts ofglycoprotein VI have an impaired response to collagen thatresults in a mild bleeding tendency. Br J Haematol1995;89:124–30.

[11] Takahashi H, Moroi M. Antibody against platelet membraneglycoprotein VI in a patient with systemic lupus erythema-tosus. Am J Hematol 2001;67:262–7.

[12] Nurden P, Jandrot-Perrus M, Combrie R, Winckler J,Arocas V, Lecut C, et al. Severe deficiency of glycopro-tein VI in a patient with gray platelet syndrome. Blood2004;104:107–14.

[13] Boylan B, Chen H, Rathore V, Paddock C, Salacz M,Friedman KD, et al. Anti-GPVI-associated ITP: an acquiredplatelet disorder caused by autoantibody-mediated clear-ance of the GPVI/FcRgamma-chain complex from thehuman platelet surface. Blood 2004;104:1350–5.

[14] Chen J, Diacovo TG, Grenache DG, Santoro SA, Zutter MM.The a2 integrin subunit deficient mouse: a multifacetedphenotype including defects in branching morphogenesisand hemostasis. Am J Pathol 2002;161:337–44.

[15] Kato K, Kanaji T, Russell S, Kunicki TJ, Furihata K, Kanaji S,et al. The contribution of glycoprotein VI to stable plateletadhesion and thrombus formation illustrated by targetedgene deletion. Blood 2003;102:1701–7.

[16] Lockyer S, Okuyama K, Begum S, Le S, Sun B, Watanabe T,et al. GPVI-deficient mice lack collagen responses and areprotected against experimentally induced pulmonarythromboembolism. Thromb Res 2005 [Electronic publica-tion ahead of print].

[17] Chen H, Kahn M. Reciprocal signaling by integrin andnonintegrin receptors during collagen activation of plate-lets. Mol Cell Biol 2003;23:4764–77.

[18] Nieswandt B, Watson SP. Plateletcollagen interaction: isGPVI the central receptor? Blood 2003;102:449–61.

[19] Nakamura T, Jamieson GA, Okuma M, Kambayashi JI,Tandon NN. Platelet adhesion to native Type I collagen

Anti-GPVI antibodies from GPVI knockout mice 329

Fibrils: role of GPVI in divalent cation-dependent andindependent adhesion and thromboxane A2 generation.J Biol Chem 1998;273:4338–44.

[20] Nakamura T, Kambayashi JI, Okuma M, Tandon NN. Activa-tion of the IIb—IIIa complexs induced by platelet adhesionto collagen is mediated by both a2h1 integrin and GPVI.J Biol Chem 1999;274:11897–903.

[21] Massburg S, Gawaz M, Gruner S, Schulte V, Konard I,Zohlhffer D, et al. A crucial role of glycoprotein VI forplatelet recruitment to the injured arterial wall in vivo.J Exp Med 2003;197:41–9.

[22] Konoshi H, Katoh Y, Takaya N, Kashiwakura Y, Itoh S, Ra C,et al. Platelets activated by collagen through immunor-eceptor tyrosine based activation motif play pivotal role inthe initiation and generation of neointimal hyperplasiaafter vascular injury. Circulation 2002;105:912–6.

[23] Jandrot-Perrus M, Busfield S, Lagrue A-H, Xiong X, Debili N,Chickering T, et al. Cloning, characterization, and func-tional studies of human and mouse glycoprotein VI: aplatelet-specific collagen receptor from immunoglobulinsuperfamily. Blood 2000;96:1798–807.

[24] Nieswandt B, Schulte V, Bergmeier W, Mokhari-Nejad R,Rackebrandt K, Cazenave C, et al. Long-term antithrom-botic protection by in-vivo depletion of platelet glycopro-tein GPVI in mice. J Exp Med 2001;193:459–69.

[25] Lecut C, Feeney LA, Kingsbury G, Hopkins J, Lanza F,Gachet C, et al. Human platelet glycoprotein VI function isantagonized by monoclonal antibody-derived Fab frag-ments. J Thromb Haemost 2003;1:2653–62.

[26] Qian MD, Villeval J-L, Xiong X, Jandrot-Perrus M, NagashimaK, Tonara J, et al. Anti GPVI human antibodies neutralizingcollagen-induced platelet aggregation isolated from com-binatorial phage display library. Hum Antibodies 2002;11:97–105.

[27] Smethurst PA, Joutsi-Korhonen L, O’Connor MN, Wilson E,Jennings NS, Garner SF, et al. Identification of the primarycollagen binding surface on human glycoprotein VI by site-directed mutagenesis and by a blocking phage antibody.Blood 2004;103:903–11.

[28] Morton LF, Hargreaves PG, Farndale RW, Young RD, BarnesMJ. Integrin alpha 2 beta 1-independent activation ofplatelets by simple collagen-like peptides: collagen tertiary(triple-helical) and quaternary (polymeric) structures aresufficient alone for alpha 2 beta 1-independent plateletreactivity. Biochem J 1995;306:337–44.

[29] Polgar J, Clemetson JM, Kehrel BE, Wiedemann M, MagnenatEM, Wells TN, et al. Platelet activation and signal transduc-tion by convulxin, a C-type lectin from Crotalus durissusterrificus (tropical rattlesnake) venom via the p62/GPVIcollagen receptor. J Biol Chem 1997;272:13576–83.

[30] Parham P, Androlewicz MJ, Brodsky FM, Holms NJ, Ways JP.Monoclonal antibodies: purification and application tostructural functional studies of class I MHC antigens.J Immunol Methods 1982;53:133–73.

[31] Fujimura H, Altar CA, Chen R, Nakamura T, Nakahashi T,Kambayashi J, et al. Brain-derived neurotrophic factor isstored in human platelets and released by agonist stimu-lation. Thromb Haemost 2002;87:728–34.

[32] Jandrot-Perrus M, Lagrue A, Okuma M, Bon C. Adhesion andactivation of human platelets induced by convulxin involveglycoprotein VI and integrin a2h1. J Biol Chem 1997;272:27035–41.

[33] Chen H, Locke D, Liu Y, Liu C, Kahn M. The platelet receptorGPVI mediates both adhesion and signaling responses to

collagen in a density-dependent fashion. J Biol Chem2002;277:3011–9.

[34] Best D, Senis Y, Jarvis G, Eagleton H, Roberts D, Saito T, etal. GPVI levels in platelets: relationship to platelet functionat high shear. Blood 2003;102:2811–8.

[35] Clemetson KJ, Clemetson JM. Platelet collagen receptors.Thromb Haemost 2001;86:189–97.

[36] Watson SP, Asazuma N, Atkinson B, Berlanga O, Best D, BobeR, et al. The role of ITAM and ITIM-coupled receptors inplatelet activation by collagen. Thromb Haemost2001;86:276–88.

[37] Kahn ML, Zheng YW, Huang W, Bigornia V, Zeng D, Moff S, etal. A dual thrombin receptor system for platelet activation.Nature 1998;394:690–4.

[38] Declerck PJ, Carmeliet P, Verstreken M, Cock FD, Collen D.Generation of monoclonal antibodies against autologousproteins in gene-inactivated mice. J Biol Chem 1995;270:8397–400.

[39] Castrop J, Verbeek S, Hofhuis F, Clevers H. Circumventionof tolerance for the nuclear T cell protein TCF-1 byimmunization of TCF-1 knockout mice. Immunology1995;193:281–7.

[40] Siljander P, Farndale RW, Feijge MAH, Comfurius P, Kos S,Bevers EM, et al. Platelet adhesion enhances the glycopro-tein VI-dependent procoagulant response: involvement ofp38 MAP kinase and calpain. Arterioscler Thromb Vasc Biol2001;21:618–27.

[41] Nieswandt B, Brakebusch C, Bergmeier W, Schulte V,Bouvard D, Mokhtari-Nejad R, et al. Glycoprotein VI butnot a2h1 integrin is essential for platelet interaction withcollagen. EMBO J 2001;20:2120–30.

[42] Siljander PRM, Munnix ICA, Smethurst PA, Deckmyn H,Lindhout T, Ouwehand WH, et al. Platelet receptorinterplay regulates collagen-induced thrombus formationin flowing human blood. Blood 2004;103:1333–41.

[43] Schulte V, Snell D, Bergmeier W, Zirngibl H, Watson SP,Nieswandt B. Evidence for two distinct epitopes withincollagen for activation of murine platelets. J Biol Chem2001;276:364–8.

[44] Lecut C, Arocas V, Ulrichts H, Elbaz A, Villeval J-L, LacapereJ-J, et al. Identification of residues within human glyco-protein VI involved in the binding to collagen: evidence forthe existence of distinct binding sites. J Biol Chem2004;279:52293–9.

[45] Nguyen CM, Harrington RA. Glycoprotein IIb/IIIa receptorantagonists: a comparative review of their use in percuta-neous coronary intervention. Am J Cardiovasc Drugs2003;3:423–36.

[46] The EPIC Investigators. Use of a monoclonal antibodydirected against the platelet glycoprotein IIb/IIIa receptorin high-risk coronary angioplasty. The EPIC Investigation. NEngl J Med 1994;330:956–61.

[47] The EPISTENT Investigators. Randomised placebo-con-trolled and balloon-angioplasty-controlled trial to assesssafety of coronary stenting with use of platelet glycopro-tein-IIb/IIIa blockade. Lancet 1998;352:87–92.

[48] The GUSTO IV-ACS Investigators. Effect of glycoprotein IIb/IIIa receptor blocker abciximab on outcome in patients withacute coronary syndromes without early coronary revascu-larisation: the GUST IV-ACS randomised trial. Lancet2001;357:1915–24.

[49] The PURSUIT Trial Investigators. Inhibition of plateletglycoprotein IIb/IIIa with eptifibatide in patients with acutecoronary syndromes. N Engl J Med 1998;339:436–43.