Adhesion and Activation of Human Platelets Induced by Convulxin Involve Glycoprotein VI and Integrin α2β1 *

We analyzed the interaction of convulxin (Cvx), a 72-kDa protein isolated from the venom of Crotalus durissus terrificus, with human platelets. Cvx is a potent platelet agonist that induces an increase in the intracellular Ca2+concentration ([Ca2+] i ), granule exocytosis and aggregation. 125I-Labeled Cvx binds specifically and rapidly to platelets at binding sites of high and moderate affinity. Platelets adhere to immobilized Cvx in a time-dependent but cation-independent manner. Platelet exocytosis and aggregation induced by Cvx were inhibited by an anti-integrin α2β1 monoclonal antibody (6F1) and by the Fab fragments of a polyclonal anti-glycoprotein VI (GPVI) antibody. Both the adhesion of platelets to Cvx and the Cvx-induced increase in [Ca2+] i were inhibited by anti-GPVI Fab fragments but not by 6F1. Ligand blotting assay showed that 125I-Cvx binds to a 57-kDa platelet protein with an electrophoretic mobility identical to that of GPVI. In addition, we observed the following: (i)125I-Cvx binds to GPVI immunoprecipitated by the anti-GPVI antibody from a platelet lysate, and (ii) Cvx inhibits the binding of anti-GPVI IgG to GPVI. Taken together, these results demonstrate that GPVI behaves as a Cvx receptor and that the α2β1 integrin appears to be involved in the later stages of Cvx-induced platelet activation,i.e. exocytosis and aggregation.

Many snake venom proteins are known to interact with platelets. Some behave in vitro as cell agonists, and others are inhibitors of platelet activation induced by physiological agents (1). Several groups can be distinguished according to their mechanism of action and molecular structure: inhibitory and activating proteases, glycoproteins (aggregoserpentins) and lectins (thrombolectins) that stimulate platelets, or peptides that inhibit platelet aggregation (disintegrins) (2). The target of some of these compounds has been identified. Disintegrins interact with the platelet integrin ␣ IIb ␤ 3 corresponding to glycoprotein (GP) 1 IIb-IIIa and inhibit the binding of fibrinogen to this receptor and, hence, platelet aggregation. Another large group of venom proteins interact with membrane GPIb, result-ing in platelet agglutination and/or inhibition of the binding of von Willebrand factor to this receptor (3).
Convulxin (Cvx), a 72-kDa glycoprotein, isolated from the venoms of Crotalus durissus cascavella and Crotalus durissus terrificus (4), is an extremely potent platelet activator (5). Platelet aggregation induced by Cvx has been shown to be Ca 2ϩ -dependent but fibrinogen-, ADP-, and cyclooxygenaseindependent (6). The transduction pathway activated by Cvx involves the activation of a phospholipase C␥ and activation of tyrosine kinases (7). 2 Cvx also binds to rabbit platelets with a high affinity (8). Most studies performed on Cvx have been conducted on animal, and particularly rabbit, platelets, but platelet membrane components are better known in humans. We have therefore focused our studies on Cvx interaction with human platelets to elucidate its mechanism of action. We show that Cvx is a particularly potent activator of human platelets, inducing an increase in [Ca 2ϩ ] i , granule exocytosis, and aggregation, and that immobilized Cvx induces a cation-independent platelet adhesion. Cvx binds specifically and rapidly to washed and to fixed platelets with a high affinity. Use of specific antibodies showed that GPIb, GPIV, and GPV do not interact with Cvx but that both the platelet integrin ␣ 2 ␤ 1 (GPIa-IIa) and GPVI are involved in platelet activation induced by Cvx, and that GPVI acts as a platelet receptor for Cvx.

EXPERIMENTAL PROCEDURES
Reagents-Cvx was purified from the venom of C. d. terrificus (Pentapharm, Basel, Switzerland and Syntex Laboratories, Ribero Preto, Brazil) as described previously (8). Cvx was labeled with 125 I using the IODOGEN procedure (Pierce) and Na 125 I (Amersham, Les Ulis, France). Specific activity of the labeled compound was 9 MBq/mg, and labeling did not modify the biological properties of Cvx. The monoclonal anti-GPIb␣ antibody, SZ2, and the anti-GPV antibody, SW16, were purchased from Immunotech (Marseille, France). The anti-␣ 2 ␤ 1 integrin monoclonal antibodies used were 6F1 (9), a generous gift from Dr. Barry S. Coller (Mount Sinai Medical Center, New York, NY), and Gi9, purchased from Immunotech. The anti-GPIV antibody FA6-152 produced by Dr. Lena Edelman (Institut Pasteur, Paris, France) was kindly provided by Dr. C. Legrand (INSERM U-353, Paris, France). Purification of human polyclonal anti-GPVI IgG, from the serum of a patient with autoimmune thrombocytopenia (10), was performed by chromatography on protein A-Sepharose (Pharmacia Biotech Inc., Uppsala, Sweden). Fab fragments were prepared from purified IgG incubated for 90 min at 37°C, with papain (Sigma; 1 g/100 g of IgG), in 20 mM sodium phosphate, 150 mM NaCl (PBS), pH 7.4, containing 1 mM EDTA and 1.7% (v/v) ␤-mercaptoethanol. Iodoacetamide (30 mM) was then added and left for 15 min at 37°C, and the sample was dialyzed in PBS. Fc fragments were removed on protein A-Sepharose. The purity of Fab fragments and the absence of contaminating IgG, in particular, were controlled by SDS-PAGE. Purified anti-GPVI IgG induced platelet aggregation, whereas Fab fragments inhibited both collagen-and anti-GPVI IgG-induced platelet aggregation as reported previously (10). Control human IgG and Fab fragments used in these studies were prepared from the plasma of healthy donors according to the procedures described above. Anti-Cvx antibody was kindly provided by Dr. M. Leduc (Unité des Venins Institut Pasteur). Calf skin collagen type I, obtained from Stago (Asnières, France), was used according to the manufacturer's instructions.
Platelet Preparation-Blood from healthy human volunteers was collected by venipuncture on acid-citrate-dextrose anticoagulant (ACD-A) or on trisodium citrate. Platelet-rich plasma (PRP) was obtained by centrifugation at 110 ϫ g for 15 min. Platelet secretion and adhesion were determined using platelets labeled in PRP (ACD-A) incubated with 0.6 M [ 14 C]5-hydroxytryptamine (Amersham, Les Ulis, France) and 2 Ci/ml 51 Cr (CIS International, Gif-sur-Yvette, France), for 30 min at 37°C respectively. Platelets were sedimented at 1100 ϫ g for 15 min after acidification of the PRP to pH 6.5 with ACD-A and addition of 25 g/ml apyrase (Sigma) and 100 nM prostaglandin E1 (Sigma) and resuspended in washing buffer (103 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 5 mM glucose, 36 mM citric acid) at pH 6.5 containing 3.5 mg/ml BSA (Sigma), 25 g/ml apyrase, and 100 nM prostaglandin E1 (11). After sedimentation, the platelets were washed twice in this buffer and resuspended at 3 ϫ 10 8 /ml in the reaction buffer composed of 5 mM Hepes, 137 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 12 mM NaHCO 3 , 0.3 mM NaH 2 PO 4 , 5.5 mM glucose, pH 7.4, containing 3.5 mg/ml BSA. Formaldehyde-fixed platelets were prepared by incubating 1 volume of citrated PRP with 0.98 volume of 10 mM Tris, 150 mM NaCl, pH 7.0, and 0.02 volume of 30% (w/v) formaldehyde for 18 h at room temperature and in the dark. Fixed platelets were washed three times in PBS before resuspension in PBS.
Platelet Aggregation and Secretion-PRP or washed platelets were preincubated for 3 min at 37°C in the presence of buffer or antibodies before aggregation was initiated by Cvx or collagen. Experiments were performed under stirring conditions at 37°C in a Chrono-Log aggregometer (Chrono-Log Corp, Haverton, PA). Release of [ 14 C]5-HT was measured as described previously (11).
Platelet Adhesion-Platelet adhesion was measured on microtitration plates. Typical experiments were performed as follows. Collagen (2 g) in 100 l of 20 mM acetic acid, Cvx (1.4 g) in 100 l of PBS, or BSA (5 g) in 100 l of PBS were immobilized on Immulon II plates (Dynatech, St-Cloud, France) for 2 h at room temperature. Plates were then saturated with 2 mg/ml BSA in PBS for 1 h, and washed with PBS and with reaction buffer. 51 Cr-Labeled platelets (2 ϫ 10 8 /ml in reaction buffer, 100 l) were added to the wells in the presence or absence of 300 M Arg-Gly-Asp-Ser (RGDS) peptide (Bachem Biochimie, Voisins-le-Bretonneux, France) or 2 mM EDTA. A certain number of experiments were performed in the presence of antibodies, as indicated in the text. Wells were emptied and washed three times with reaction buffer after different incubation times at room temperature, One hundred l of 2% SDS (w/v) was subsequently added to each well, and the samples were counted for 51 Cr.
Intracellular Ca 2ϩ Concentration-Intracellular free calcium ([Ca 2ϩ ] i ) transients were monitored by fura-2 fluorescence. Platelets were resuspended in washing buffer after centrifugation of the PRP and loaded with 2 M fura 2-acetoxymethylester (fura 2-AM, Sigma) for 60 min at 37°C, centrifuged again, washed twice, and resuspended in reaction buffer. The platelets (2 ϫ 10 8 /ml) were then preincubated with 2 mM CaCl 2 in the absence or presence of different antibodies for 3 min at 37°C prior to addition of Cvx or collagen. Fluorescence was measured at 37°C using two excitation wavelengths of 340 and 380 nm and an emission wavelength of 510 nm on an Hitachi H-2000 spectrofluorimeter (Sciencetec, Les Ulis, France). Maximal and minimal fluorescence were determined after platelet lysis with 1% (v/v) Triton X-100 and the addition of 2 mM EGTA, respectively. Ca 2ϩ concentrations were calculated using a K d of 224 nM for the interaction between fura 2 and Ca 2ϩ (12).
Binding of 125 I-Labeled Cvx to Platelets-Washed or formaldehydefixed platelets (3 ϫ 10 8 /ml) were incubated at 37°C in 400 l with various amounts of 125 I-labeled Cvx. The platelets were transferred to tubes containing 500 l of 20% (w/v) sucrose in PBS after different incubation times, and centrifuged for 5 min at 12,000 ϫ g. Supernatants were aspirated, and the tips of the tubes were counted for 125 I to determine the fraction of Cvx bound to platelets. Nonspecific binding was determined in the presence of a 100 -500-fold excess of unlabeled Cvx.
Immunoprecipitation-Platelet lysates were obtained by solubilization of washed platelets (5 ϫ 10 9 /ml) with 1% (v/v) Nonidet P-40 in 20 mM Tris-HCl, pH 8.0, 150 mM NaCl containing 15 g/ml leupeptin (Sigma), 50 kallikrein-inactivating units of aprotinin, 1 mM phenylmethylsulfonyl fluoride (Sigma), 2 mM benzamidine HCl, and 2 mM EDTA (lysis buffer), at 4°C for 30 min followed by centrifugation at 13,000 ϫ g at 4°C for 30 min. Lysates were precleared by incubation with protein A-Sepharose for 30 min at 4°C and centrifugation to avoid nonspecific precipitation. Cleared lysates were incubated with 200 g/ml anti-GPVI IgG for 30 min at room temperature and then protein A-Sepharose at 4°C overnight. Samples were centrifuged at 2,000 ϫ g, and the immunoprecipitates were washed three times with the lysis buffer. Immunoprecipitated proteins were eluted by 2% SDS in Laemmli buffer, subjected to SDS-polyacrylamide gel electrophoresis, and transferred to nitrocellulose membranes, which were probed using 125 I-Cvx or anti-GPVI IgGs, as described above. (Fig. 1) as reported previously for rabbit platelets (6). Aggregation was also observed in PRP and was preceded by a change in cell shape. The effect of Cvx was tested on formaldehyde-fixed platelets to verify that Cvx induced true platelet aggregation rather than passive agglutination. No modification of light transmittance was observed for suspensions of fixed platelets, indicating that Cvx does not agglutinate platelets (data not shown). The threshold concentration of Cvx that typically induced platelet aggregation was between 15 and 35 pM. Important heterogeneity in platelet sensitivity to Cvx was observed with platelets from normal individuals however. Secretion of [ 14 C]5-HT from dense granules paralleled aggregation. The dose-response curve was sigmoid and characterized by a very abrupt slope ( Fig. 2A). Aggregation was prevented by 2 mM EDTA and by 300 M RGDS peptide, which block fibrinogen binding to integrin ␣ IIb ␤ 3 . In contrast, neither cell shape change nor [ 14 C]5-HT release were inhibited by EDTA or RGDS peptide (Fig. 2B). Along with granule exocytosis and cell aggregation, Cvx in- duced a dose-dependent increase in [Ca 2ϩ ] i . Following a lag phase, which could be shortened by increasing the Cvx concentration, [Ca 2ϩ ] i reached a plateau that remained stable for 3 min (Fig. 3).

Cvx Induces Activation of Human Platelets-Washed human platelets aggregated in response to Cvx
Human Platelets Adhere to Immobilized Cvx-51 Cr-Labeled platelets were found to bind to microtitration plates coated with Cvx (Fig. 4). Platelets also bound to immobilized collagen under similar static conditions. Platelet adhesion was a function of the coating concentration in both cases and reached a maximum for Cvx and collagen concentrations Ն 10 g/ml (data not shown). A total of 43 Ϯ 2% of the platelets were bound to immobilized Cvx after 1 h of incubation when the proportion of platelets bound to immobilized collagen was 2-fold lower and less than 0.5% were bound to immobilized BSA. The number of platelets bound to Cvx and collagen decreased to 12.7 Ϯ 0.5% and 8.6 Ϯ 0.8%, respectively, due to inhibition of platelet aggregation when experiments were performed in the presence of 300 M RGDS (Fig. 4). These results indicate that platelet aggregation is more important in Cvx-coated wells than collagen-coated wells and confirms that Cvx has a high potency for inducing platelet activation. The proportion of platelets bound to Cvx remained unchanged in the presence of EDTA (2 mM) as in the presence of RGDS, but platelet adhesion to collagen was prevented by EDTA (Fig. 4), as reported previously (14).

125
I-Cvx Binds to Human Platelets-125 I-Cvx bound to washed platelets in a time-and concentration-dependent manner (Fig. 5). Nonspecific binding measured in the presence of a 100-fold excess of unlabeled Cvx represented less than 10% of the total radioactivity added. Association of 125 I-Cvx with human platelets at 37°C was rapid and reached a steady state at 15 min (Fig. 5A). Addition of a 500-fold excess of unlabeled Cvx 15 min after the addition of 125 I-Cvx to platelets showed, in contrast, that dissociation was very slow, with less than 1% of the bound 125 I-Cvx being displaced per minute in the presence of a 100-fold excess of unlabeled Cvx (data not shown). Scatchard plot analysis of specific binding of 125 I-Cvx to human platelets indicated two classes of binding sites, one of high affinity and low capacity and one of moderate affinity and capacity (Table I). Comparable results were obtained when formaldehyde-fixed platelets were used. Affinities were slightly lower in this case, but receptor number was increased for both classes of sites (Table I). Additional experiments were per-formed in the presence of sugars such as 45 mM galactose or mannose or in the presence of 10 g/ml WGA, suspected previously to interact with the Cvx binding site on platelets (15), to rule out the possibility that Cvx binding to platelets might be due to a lectin-like activity. None of the above compounds modified Cvx binding to platelets (data not shown).
Integrin ␣ 2 ␤ 1 and GPVI Are Involved in Cvx-induced Platelet Activation-Specific antibodies to known membrane glycoproteins were tested for their effect on Cvx-induced platelet activation to discover whether these glycoproteins were involved in Cvx interaction with platelets. We tested antibodies to glycoproteins already identified as receptors for different cell agonists or ligands, such as the GPIb-V-IX complex (receptor for von Willebrand factor), the integrin ␣ 2 ␤ 1 and GPVI (receptors for collagen), and GPIV (CD 36), which has a less clearly defined function and may act as a collagen and/or thrombospondin receptor (16). Monoclonal antibodies with inhibitory activity against GPIb␣ (SZ2), GPIV (FA6-152), and GPV (SW16) had no effect on Cvx-induced platelet aggregation, platelet adhesion to Cvx, or Cvx binding to platelets (data not shown). Cvx-induced platelet aggregation and secretion were both inhibited, in contrast by 6F1, a monoclonal antibody against integrin ␣ 2 ␤ 1 and by polyclonal anti-GPVI Fab fragments (Fig. 6, A and B). Inhibition by 6F1 was concentration- dependent and reached a maximum after preincubation of the platelets with 1 g/ml antibody for 3 min at 37°C. Inhibition was total for low concentrations of Cvx (Ͻ50 pM) but could be overcome by increasing the concentration of Cvx above 0.1 nM. Another monoclonal anti-␣ 2 ␤ 1 antibody, Gi9, also inhibited Cvx-induced platelet aggregation but was less potent than 6F1, inhibition being 30% and 80% with 1 g/ml and 5 g/ml of Gi9, respectively. The anti-GPVI Fab fragments, at 0.2 mg/ml, produced a ϳ50% inhibition of platelet aggregation and secretion induced by 30 -40 pM Cvx. As with 6F1, inhibition of platelet aggregation and secretion by the anti-GPVI Fab fragments was overcome by increasing the concentration of Cvx. Control experiments showed that nonimmune human Fab had no effect on Cvx-induced platelet aggregation and secretion (data not shown). The concentrations of 6F1, Gi9, and anti-GPVI Fab fragments that inhibited Cvx-induced platelet aggregation also inhibited collagen-induced platelet aggregation (data not shown and Refs. 9 and 10). The increase in [Ca 2ϩ ] i concentration induced by Cvx was prevented by preincubation of platelets with the anti-GPVI Fab fragments (Fig. 6C). 6F1 (1 g/ml) did not decrease the signal induced by Cvx in contrast but completely inhibited Ca 2ϩ mobilization induced by collagen (Fig. 6C).
The effects of 6F1 and anti-GPVI Fab-fragments on platelet adhesion to immobilized Cvx and collagen were also analyzed. All experiments were performed in the presence of 300 M RGDS to avoid platelet aggregation. None of the antibodies modified the number of 51 Cr-labeled platelets bound to immobilized Cvx after a 60-min incubation, whereas 6F1 inhibited platelet adhesion to collagen by 90% (data not shown), as shown by others (9). We used shorter incubation times (5 min for Cvx and 15 min for collagen) to increase the chances of observing an inhibition since binding of Cvx to platelets is rapid, whereas dissociation is very slow. Approximately the same percentage of platelets adhered to the two proteins under these conditions. Platelet adhesion to Cvx at 5 min was not significantly reduced by the presence of 6F1 (even at concentrations up to 10 g/ml), but platelet adhesion to collagen at 15 min was clearly inhibited by 6F1 at a concentration as low as 0.5 g/ml (Fig. 7). Gi9 also failed to inhibit platelet adhesion to Cvx, but it inhibited platelet adhesion to collagen (data not shown). The anti-GPVI Fab fragments (0.2 mg/ml) inhibited platelet adhesion to Cvx significantly in contrast, and also inhibited platelet adhesion to collagen at 15 min (Fig. 7).
Cvx Binds to GPVI-The possibility that Cvx recognizes a specific platelet protein was tested using a ligand blotting assay, since Cvx has a very high affinity for platelets. Binding of 125 I-Cvx was observed to occur on a single band of M r ϭ 57,000 Ϯ 1000 (Fig. 8A) after separation of proteins from whole platelet lysates by SDS-polyacrylamide gel electrophoresis and transfer to nitrocellulose membranes. Labeling of this band was not observed when the incubation was performed in the presence of a 100-fold excess of unlabeled Cvx, or 50 g/ml anti-Cvx antibody, indicating that binding was specific. No binding of labeled Cvx was observed when the platelet proteins were reduced with 5% ␤-mercaptoethanol before electrophoresis. Binding was not inhibited in the presence of 5 g/ml WGA, and was still observed after pretreatment of platelets with 5 nM thrombin, or 3 g/ml protease of Serratia marcescens or 1 M ␣-chymotrypsin, respectively (data not shown). The latter findings indicated that this 57-kDa protein was neither a binding site for WGA nor degraded to any great extent by these proteases. We also observed a protein band migrating at the same position in membrane extracts obtained after cellular fractionation (data not shown). It appeared likely that this 57-kDa platelet protein corresponded to GPVI since the size of the protein recognized by 125 I-Cvx was similar to that reported for GPVI (10,15,16) and platelet activation by Cvx was inhibited by the anti-GPVI Fab fragments. Immunoprecipitates of platelet lysates obtained with the anti-GPVI IgG were probed with 125 I-Cvx to test this possibility. Fig. 8B shows that the immu-  noprecipitated platelet protein interacted with 125 I-Cvx. Competition experiments using nitrocellulose membranes incubated with the anti-GPVI IgG in the presence of cold Cvx were therefore performed. The intensity of the band labeled by the anti-GPVI IgG decreased as the concentration of Cvx increased, thus confirming that Cvx and the anti-GPVI IgG bind to the same protein, i.e. GPVI (Fig. 8C).

DISCUSSION
The present study demonstrates that Cvx is a very potent activator of human platelets and that the membrane glycopro-teins GPVI and integrin ␣ 2 ␤ 1 are both involved in the activation pathway. Cvx at picomolar concentrations induces extensive activation of stirred platelets, resulting in exocytosis of dense granules and aggregation. Although some authors (6) have reported that Cvx-induced activation of rabbit platelets is calcium-dependent, our results show that Cvx-induced platelet activation of human platelets occurs in the absence of external Ca 2ϩ and is in agreement with the findings of Sano-Martins and Daimon (15). However, Cvx induces an increase in [Ca 2ϩ ] i . The kinetic of Cvx-induced Ca 2ϩ mobilization is slow when compared with the very rapid response induced by other agonists such as thrombin (19), and the lag preceding Ca 2ϩ mobilization is more similar to that observed with collagen (see Fig.  6). It has been demonstrated recently that Cvx induces the activation of phospholipase C␥2 (PLC␥2) by a protein-tyrosine kinase-dependent pathway. 2 Tyrosine phosphorylation and activation of PLC␥2 have already been reported in collageninduced platelet activation (20,21), and a relationship between the mechanisms of collagen-and Cvx-induced platelet activation has been suspected previously since rabbit platelets exposed to convulxin become refractory to subsequent exposure to collagen (6). Our results provide further evidence for similarities between Cvx and collagen interaction with human platelets. Indeed, integrin ␣ 2 ␤ 1 and GPVI are already known to act as co-receptors for collagen-induced platelet activation. Thus, platelet adhesion to immobilized collagen, platelet aggregation induced by collagen fibers, and collagen-induced phosphorylation of the non-receptor tyrosine kinase Syk and of PLC␥2 are inhibited by anti-␣ 2 ␤ 1 monoclonal antibodies (9,22). Human polyclonal anti-GPVI IgG, on the other hand, activates normal platelets but not GPVI-deficient platelets, and the Fab fragments from these IgGs inhibit collagen-induced platelet acti- vation (10). In addition, platelets from patients deficient in ␣ 2 ␤ 1 or in GPVI do not aggregate in response to soluble collagen (23)(24)(25)(26). These results suggest that ␣ 2 ␤ 1 and GPVI cooperate in producing the full response to collagen.
The results of the present study indicate that GPVI behaves as a platelet receptor for Cvx; binding of Cvx to immunoprecipitated GPVI was demonstrated by ligand blotting experiments, and involvement of GPVI in Cvx-induced platelet activation was demonstrated by the inhibitory effect of anti-GPVI antibodies on platelet adhesion to Cvx, Cvx-induced [Ca 2ϩ ] i increase, platelet aggregation, and secretion. Cvx binding to GPVI may thus be essential for signaling, possibly by eliciting Syk phosphorylation as reported for platelet activation induced by the F(abЈ)2 fragments of the anti-GPVI IgG (27). The ␣ 2 ␤ 1 integrin is also important for Cvx-induced platelet activation since the anti-␣ 2 ␤ 1 antibodies, 6F1 and to a lesser extent Gi9, inhibited Cvx-induced platelet aggregation and exocytosis as well as collagen induced platelet aggregation (9). However, the role played by ␣ 2 ␤ 1 in platelet activation by Cvx appears to differ from that observed in collagen-induced platelet activation. Indeed, while platelet adhesion to collagen is mediated by ␣ 2 ␤ 1 in a Mg 2ϩ -dependent manner (9,14), platelet adhesion to Cvx is Mg 2ϩ -independent and is not affected by the anti-␣ 2 ␤ 1 antibody 6F1. In addition, 6F1 inhibited collagen-induced increase in [Ca 2ϩ ] i , but not Cvx-induced increase in [Ca 2ϩ ] i . The mechanism by which 6F1 prevents Cvx-induced aggregation and secretion is still unclear, and, as for collagen, the mechanism by which ␣ 2 ␤ 1 and GPVI cooperate in producing a total platelet response to Cvx remains to be determined.
Equilibrium binding studies have identified two types of Cvx-binding sites on human platelets. The K d value measured for the very high affinity binding sites (8 ϫ 10 Ϫ11 M) is consistent with that reported for rabbit platelets (8) and with the concentration of Cvx that induces platelet activation (see Fig.  1). The slight differences in K d and B max values noted between washed and fixed platelets may be due to modifications in the distribution of glycoproteins following Cvx-induced activation of washed platelets (28) or from modifications due to platelet fixation with formaldehyde. Data from the present study did not permit identification of the nature of the high and moderate affinity binding sites. The anti-␣ 2 ␤ 1 integrin monoclonal antibody 6F1 did not significantly modify 125 I-Cvx binding to platelets (data not shown) nor it did reduce platelet binding to immobilized Cvx, even under conditions of pre-equilibrium (short incubation) or at low Cvx concentrations (Ͻ100 pM). In addition, the scarcity of Fab fragments from the anti-GPVI IgG, which inhibited the binding of platelets to immobilized Cvx, prevented their use in equilibrium binding studies. The estimated number of surface ␣ 2 ␤ 1 sites varies among platelets from different individuals from 1000 to 2000 (29), i.e. within the range for the moderate affinity binding sites for Cvx. The number of surface GPVI sites has not been reported to date.
Platelet activation by Cvx may thus occur via activation of an unidentified signaling pathway coupled to GPVI and/or ␣ 2 ␤ 1 or from the cross-linking of these two membrane receptors by Cvx. Several recent observations support the second hypothesis. Indeed, clustering of platelet surface glycoproteins appears to be a main mechanism by which platelet activation occurs. Thus, homotypic cross-linking of the Fc␥ RII IgG receptor by specific antibodies (30), or GPVI by anti-GPVI F(abЈ) 2 fragments (27), as well as heterotypic cross-linking of ␣ 2 ␤ 1 with Fc␥-RII (22) activate protein-tyrosine kinases and PLC␥2. It has been reported previously, using a nondenaturing medium, that Cvx can be present in at least three forms: the 72-kDa species, which is a trimeric assembly with a ␣ 3 ␤ 3 structure consisting of ␣ (13.5 kDa) and ␤ (12.5 kDa) Cvx subunits linked by disulfide bonds; and the multimerized 144-kDa and 300-kDa forms of this trimer, organized in a ring-shaped structure (31,32). All of these molecular forms of Cvx are potentially multivalent and may engage in multiple interactions at the platelet surface, resulting in glycoprotein cross-linking and cell signaling.
Electron microscopy studies have also indicated that Cvxtreated platelets have a decreased capacity to bind WGA. This suggested that Cvx and WGA might share common binding site(s) on platelets (15). However, no modification in Cvx binding to platelets or GPVI, respectively, was observed in the presence of WGA. It is likely that the decrease in WGA binding, described in earlier studies, might have been due to an activation-FIG. 8. 125I-labeled-Cvx binds to GPVI. A, in a ligand blotting assay, SDS-solubilized platelet proteins were separated on 10% acrylamide slab gels and transferred to nitrocellulose membranes, which were incubated with 50 ng/ml 125 I-Cvx (lanes 1-3), 50 ng/ml 125 I-Cvx and 10 nM unlabeled Cvx (lane 4), or 125 I-Cvx and 50 g/ml anti-Cvx antibody (lane 5). In lane 2, platelet proteins were reduced with 5% (v/v) ␤-mercaptoethanol before electrophoresis. The arrow denotes the band labeled by 125 I-Cvx, as detected by autoradiography. The positions of molecular weight standard proteins are indicated on the left. B, Nonidet P-40-solubilized platelet proteins were immunoprecipitated with the anti-GPVI IgG and protein A-Sepharose as described under "Experimental Procedures." After SDS-PAGE of nonreduced immunoprecipitates in 10% acrylamide slab gels, samples were analyzed by ligand blotting with 125 I-Cvx (lanes 1 and 3) and by immunoblotting with the anti-GPVI IgG revealed by 125 I-protein A (lanes 2 and 4) followed by autoradiography. Samples are whole platelet lysates (lanes 1 and 2) and immunoprecipitates (lanes 3 and 4). The arrowhead denotes the band corresponding to GPVI.The high molecular mass protein corresponds to labeled IgG. C, whole platelet proteins were separated by SDS-PAGE under nonreducing conditions as in A and immunoblotted with the anti-GPVI (9 g/ml, lanes [1][2][3][4][5] in the absence (lanes 1 and 5) 4). Anti-GPVI IgG were revealed using peroxidase-labeled protein A and chemiluminescence. The arrowhead denotes the band corresponding to GPVI. dependent, Cvx-induced redistribution of platelet membrane glycoproteins. Such a mechanism would be comparable to the thrombin-induced internalization of GPIb, which is known to constitute a major binding site for WGA (28).
It is noticeable that the threshold concentration of Cvx required for platelet activation is very low (3 ng/ml), as compared with collagen (10 g/ml), or the triple helical collagen-like synthetic peptides (Ͼ20 ng/ml), which are the simplest structures known to induce platelet activation as collagen (33). Cvx does not appear to be structurally related to collagen. Furthermore, it is not a calcium-dependent lectin, since its interaction with platelets is Ca 2ϩ -and sugar-independent. However, Cvx shares structural characteristics with proteins of this family (8). As snake toxins are usually derived from parental physiologically active proteins, it is tempting to hypothesize that Cvx may be analogous to an as yet unidentified molecule in man, associated with a new platelet activation pathway, which might be of physiological importance.
In conclusion, the results of the present work demonstrate that platelet membrane GPVI is a platelet Cvx receptor and that integrin ␣ 2 ␤ 1 is required for Cvx-induced platelet aggregation. Definition of the respective roles played by these glycoproteins in the signaling events evoked by convulxin is therefore of interest.