Proteolytic Cleavage of the β1 Subunit of Platelet α2β1 Integrin by the Metalloproteinase Jararhagin Compromises Collagen-stimulated Phosphorylation of pp72 syk *

Early signaling events in the stimulation of platelets by collagen include the tyrosine phosphorylations of FcR γ-chain, pp72 syk and phospholipase Cγ2. These events are dependent on the main platelet collagen receptor, α2β1 integrin (glycoprotein Ia-IIa complex). We recently found that jararhagin, a 52-kDa snake venom metalloproteinase, selectively inhibits collagen-induced platelet secretion and aggregation in parallel with the cleavage of the β1 subunit of the α2β1integrin. The present study demonstrates that jararhagin also interferes with collagen-induced phosphorylation of the protein-tyrosine kinase pp72 syk . This effect is not observed when the platelet aggregation response to collagen is inhibited by two venom RGD-containing disintegrins, contortrostatin and echistatin. These disintegrins inhibit platelet aggregation through their high affinity binding to the platelet αIIbβ3integrin (glycoprotein IIb-IIIa complex). We also show that mild stimulation by ADP of jararhagin-treated platelets, but not of platelets treated with the RGD-containing disintegrins, restores the collagen-induced platelet aggregation. ADP also restored both pp72 syk and pleckstrin phosphorylation of jararhagin-treated platelets in response to collagen, presumably via interaction of collagen with ADP-activated αIIbβ3integrin. Thus, RGD-containing disintegrins do not interfere with agonist-induced pp72 syk phosphorylation but inhibit aggregation through occupancy of the αIIbβ3 integrin. Conversely, jararhagin affects early platelet signaling events in response to collagen through its effects on the α2β1 integrin without interfering with the function of the αIIbβ3 integrin. Our demonstration that the degradation of the β1 subunit of α2β1 by jararhagin results in the loss of pp72 syk phosphorylation, suggests that this subunit is critically involved in collagen-induced platelet signaling.

Activation of platelets by different adhesive proteins involves binding of these proteins to integrins. Integrins belong to a superfamily of heterodimeric (␣␤) cell surface glycoproteins that participate in cell-cell and cell-matrix interactions (1). Following adhesion, platelet activation is marked by the char-acteristic features of shape change, granule secretion, and aggregation, involving mechanisms which are not entirely understood. However, it is well established that these effects are accompanied by tyrosine phosphorylation of a number of proteins, demonstrating a role of tyrosine kinases in integrininduced stimulation signaling (2)(3)(4)(5)(6). The understanding of the precise sequence of events involved in collagen-induced signaling in platelets is further complicated by the presence of multiple collagen-binding receptor proteins on these cells. Thus, the platelet/collagen interaction is thought to be mainly mediated by the Mg 2ϩ -dependent ␣ 2 ␤ 1 integrin (7-9) but also involves the ␣ IIb ␤ 3 integrin (10), glycoprotein IV (CD36) (11), glycoprotein VI (12), and some 85-to 90-kDa glycoproteins not yet characterized (13). The proteins which become tyrosine phosphorylated upon platelet stimulation by collagen include the Fc receptor ␥-chain (14), the low affinity immunoreceptor Fc␥RII (15), the nonreceptor tyrosine kinase pp72 syk (5,16,17), and phospholipase C␥2 (15, 18 -20). The precise role of these different proteins in collagen-induced signaling via different platelet collagen receptors has been the subject of intense recent investigation (14,15,21).
Poor platelet response to fibrillar collagen with decreased protein tyrosine phosphorylation including that of pp72 syk is associated with deficiencies of the membrane glycoprotein VI (12,22,23) and ␣ 2 ␤ 1 integrin (24 -26). Platelet ␣ 2 ␤ 1 integrin is required for effective tyrosine phosphorylation of pp72 syk and downstream phospholipase C␥2 activation (15). Similarly, glycoprotein VI deficiency also results in the decreased phosphorylation of pp72 syk during stimulation of platelets by collagen, despite normal expression of ␣ 2 ␤ 1 integrin (21). Nevertheless, the activation of platelets by collagen involves rapid phosphorylation of pp72 syk independently of Ca 2ϩ fluxes, ADP release and platelet aggregation (5,17). The platelet Fc␥RII also becomes tyrosine phosphorylated by collagen in a reaction which is upstream of pp72 syk activation and independent of ␣ 2 ␤ 1 integrin (15). However, the precise role of this reaction in collagen-induced signaling in platelets is not clear.
Platelet aggregation can be inhibited by snake venom components which mainly belong either to a well known group of RGD-containing polypeptides (disintegrins) or to the group of hemorrhagic metalloproteinases. The RGD sequence in the disintegrins is recognized by ␣ IIb ␤ 3 integrin and their binding to platelets effectively inhibits adhesive protein-mediated cellcell interaction (27,28). This results in inhibition of platelet aggregation by agonists such as thrombin or collagen without interfering with either ATP/ADP release or rise in intracellular calcium (29,30). These inhibitors, therefore, do not appear to influence intracellular events during platelet/agonist interac-tion but rather interfere with the ligand binding to activated RGD-binding integrins. In contrast, venom metalloproteinases inhibit platelet function (31,32) by a mechanism which is less well defined.
We have investigated platelet interaction with the 52-kDa hemorrhagic enzyme from Bothrops jararaca venom, jararhagin, which belongs to the class of venom metalloproteinases possessing a disintegrin-like domain (33). This domain has a high degree of homology with the RGD-containing disintegrins but in jararhagin the RGD sequence is replaced by ECD sequence (33,34). It has been proposed that the disintegrin domain of jararhagin binds to the ␣ 2 subunit I domain of ␣ 2 ␤ 1 integrin (35,36). The proposal that jararhagin recognizes ␣ 2 ␤ 1 integrin is supported by our studies, which show that this enzyme specifically inhibits the interaction of platelets with collagen (37) and cleaves the ␤ 1 subunit of this main platelet collagen receptor (36). Moreover, we have shown that jararhagin selectively inhibits the secretion-dependent phase of collagen-induced platelet aggregation and the ability of collagen to induce serine/threonine phosphorylation of pleckstrin, indicating that the enzyme impairs signal transduction leading to protein kinase C activation (38). Since in jararhagin-treated platelets the ␣ IIb ␤ 3 integrin is functionally intact (36,39), the combined use of jararhagin and RGD-containing disintegrins provides an opportunity to assess the relative contribution of ␣ 2 ␤ 1 and ␣ IIb ␤ 3 integrins to collagen-induced platelet signaling.
Our results show that although both types of venom component inhibit platelet aggregation, only jararhagin abolishes collagen-induced phosphorylation of pp72 syk kinase. In contrast, the RGD-containing disintegrins do not interfere with pp72 syk activation by collagen, but inhibit platelet aggregation due to their occupancy of ␣ IIb ␤ 3 integrin. Our findings that ADP-activated jararhagin-treated platelets signal for pp72 syk phosphorylation and aggregate with collagen and that this can be inhibited by the RGDS peptide, show that signaling by collagen requires both ␣ 2 ␤ 1 and ␣ IIb ␤ 3 integrins. Since jararhagin cleaves the ␤ 1 subunit of ␣ 2 ␤ 1 integrin, we propose that this subunit is involved in the collagen-induced signaling resulting in pp72 syk activation.
Platelet Isolation and Radiolabeling-Venous blood from healthy donors was mixed with 3.8% trisodium citrate (9:1 v/v) and centrifuged at 125 ϫ g for 10 min at room temperature. The supernatant platelet-rich plasma was separated and further centrifuged for 30 min at 1,500 ϫ g on a 25/34% albumin gradient at room temperature. Platelets were collected between the two albumin layers and gel-filtered on a Sepharose 2B column equilibrated with Tyrode's Hepes buffer, pH 7.4 (138 mM NaCl, 3 mM KCl, 1 mM MgCl 2 , 1 mM glucose, 0.5 mM NaH 2 PO 4 , 20 mM Hepes) containing 0.35% albumin (41). For 32 P-labeling, a platelet suspension (3 ml), prepared as described above except that phosphate was omitted from the buffer, was incubated with 500 Ci of 32 P at 37°C for 60 min. 32 P-Labeled platelets were separated from unbound label by gel filtration as above. For 125 I-labeling, platelets in plateletrich plasma were treated with 1 g/ml prostaglandin E 1 and then washed in albumin-free Tyrode's Hepes buffer prior to iodination by the lactoperoxidase-catalyzed reaction as described previously (36).
Platelet Activation-Platelet aggregation (3 ϫ 10 8 cells/ml) was initiated by adding either 2 g/ml collagen, 0.05 unit/ml thrombin, 2 M ADP, or 10 g/ml IV.3 mAb subsequently cross-linked with 50 g/ml anti-mouse IgG. Platelet aggregation by ADP was carried out in the presence of 200 g/ml fibrinogen. Platelet aggregation by collagen does not require external fibrinogen because it is mediated by the binding of secreted ␣ granular proteins, including fibrinogen, to the ␣ IIb ␤ 3 integrin activated by secreted dense granule ADP. Platelet stimulation was also carried out using suboptimal amounts (0.2 M) of ADP, either alone or together with 2 g/ml collagen. All aggregation patterns were recorded over 4 min.
Platelet Protein Phosphorylation and Western Blotting-For analysis of phosphorylated proteins in serine/threonine residues, unstimulated or stimulated 32 P-labeled platelets (3 ϫ 10 8 cells/ml) were lysed with equal volumes of cold 2 ϫ radioimmunoprecipitation assay (RIPA) buffer 1 (25 mM Tris-HCl, pH 7.4, buffer containing 1% Triton X-100, 150 mM NaCl, 5 mM EDTA, 30 mM sodium pyrophosphate, 50 mM sodium fluoride, 100 M sodium vanadate, 1 mM phenylmethanesulfonyl fluoride, and 1 g/ml each of pepstatin and leupeptin) for 30 min at 4°C. After centrifugation at 13,000 ϫ g for 30 min at 4°C, the lysates (30 l) were mixed with equal volumes of double strength sample buffer containing SDS and 4% 2-mercaptoethanol and electrophoresed in 5-15% polyacrylamide gels (42). Dried gels were autoradiographed. For analysis of protein phosphorylated on tyrosine, unstimulated or stimulated unlabeled platelets were lysed with an equal volume of cold 2 ϫ RIPA buffer 2 (158 mM NaCl, 1 mM EGTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, in 10 mM Tris-HCl, pH 7.2, containing 1 mM sodium vanadate, 1 mM phenylmethanesulfonyl fluoride and 100 units/ml aprotinin) for 15 min at 4°C. The insoluble residues were removed by centrifugation at 13,000 ϫ g for 15 min at 4°C. Lysates (37 l) were mixed with the same volume of sample buffer, boiled for 3 min and electrophoresed on a 7.5% polyacrylamide gel under reducing conditions. After electrophoresis, the proteins were transferred for Western blotting to Immobilon-P membranes. Membranes were blocked with 5% albumin in blocking buffer (170 mM NaCl, 50 mM Tris-HCl pH 7.5, containing 0.2% Nonidet P-40) and incubated with PY20-HRP in blocking buffer containing 3% albumin for 1 h. Membranes were washed 4 times in blocking buffer; protein blots were developed with ECL reagent and then immediately exposed to film.
Immunoprecipitation-Platelet lysates (up to 10 9 platelets/ml) were precleared with protein G-Sepharose for 1 h, antibody (1 g) was added, and the mixtures were incubated overnight at 4°C. The immune complexes were subsequently precipitated on protein G-Sepharose for 1 h at 4°C. Beads were washed four times in 1 ϫ RIPA buffer 2, and the samples were electrophoresed in 7.5% polyacrylamide gels. After electrophoresis of immunoprecitates, proteins were analyzed by Western blotting as described above. For analysis of immunoprecipitated protein, PY20 immunoblots were treated for 30 min at 50°C with stripping buffer (62.5 mM Tris-HCl, pH 6.7, containing 2% SDS and 100 mM 2-mercaptoethanol). Membranes were then washed twice with large volumes of blocking buffer, blocked with 5% albumin in blocking buffer, and probed with an appropriate primary antibody for 1 h. They were then washed four times and probed with a secondary HRP-conjugated antibody for 30 min. Membranes were finally washed, and blots were developed with ECL reagent as above.
Chemical Cross-linking of Platelet Proteins-Unstimulated or collagen-stimulated platelets (10 9 cells/ml) were incubated with 500 M of the membrane-permeable cross-linker, dithiobis(succinimidylpropionate), for 25 min at room temperature. The reaction was quenched by adding 50 mM Tris-HCl, pH 7.5. After 15 min, cells were centrifuged, the pellet resuspended in Tyrode's Hepes buffer and finally lysed with cold 2 ϫ RIPA buffer 2 as above. Proteins were immunoprecipitated from the lysates by appropriate antibodies as described above. Immunoprecipitates were either analyzed under reducing conditions by SDS-PAGE and Western blotting as above or submitted to in vitro kinase assay.
In Vitro Kinase Assay-Protein G-Sepharose beads containing immunoprecipitated proteins were washed once with RIPA buffer 2 and once with kinase reaction buffer (20 mM Tris-HCl, pH 7.5, containing 10 mM MgCl 2 and 10 mM MnCl 2 ). Then, 25 l of kinase reaction buffer containing 2 M ATP, 10 Ci of [␥-32 P]ATP, and 10 g of acid-activated enolase were added to the beads for 10 min at room temperature. The reaction was stopped by the addition of boiling SDS sample buffer, and reduced proteins were resolved by SDS-PAGE in 7.5% gels. Proteins were then transferred to Immobilon-P membranes and treated with 1 M KOH for 60 min at 55°C. Dried membranes were exposed to film.

Jararhagin and RGD-containing Disintegrins Inhibit Platelet Aggregation by Distinct Mechanisms-We have confirmed that jararhagin inhibits platelet aggregation by collagen and
have also demonstrated that contortrostatin, an RGD-containing disintegrin, inhibits this platelet response. In contrast to the inhibition caused by jararhagin, the inhibition of the platelet response to collagen due to contortrostatin was never complete (Fig. 1).
␣ IIb ␤ 3 integrin, the major platelet receptor found in an inactive conformational state in unstimulated platelets, requires activation to bind ligands (43). Platelet agonists via their specific receptors transduce a cascade of signals ultimately leading to the activation of ␣ IIb ␤ 3 integrin to a high affinity state capable of binding soluble ligands (44). To verify whether ␣ IIb ␤ 3 integrin can be activated in either jararhagin-or contortrostatin-treated platelets, we stimulated these platelets with ADP.
The addition of suboptimal amounts of ADP (0.2 M) restored the aggregation of jararhagin-treated platelets by collagen; this was inhibited by the RGDS tetrapeptide ( Fig. 1) known to block ligand-␣ IIb ␤ 3 binding. This peptide also inhibited the collageninduced aggregation of normal platelets (not shown). Similar ADP stimulation of contortrostatin-treated platelets did not restore any response to collagen (Fig. 1). Fig. 2 shows that ADP-induced aggregation of jararhagintreated platelets in the presence of fibrinogen was similar to that of control platelets, demonstrating binding of fibrinogen to the activated ␣ IIb ␤ 3 integrin. As expected, contortrostatin inhibited this reaction. Platelets treated with jararhagin also aggregated upon stimulation by thrombin (38), whereas those treated with contortrostatin, echistatin, or RGDS peptide did not (data not shown). It thus appears that jararhagin and RGD-containing disintegrins use different mechanisms to inhibit platelet responses.
The RGD-containing venom disintegrins target ␣ IIb ␤ 3 integrin and, through the occupancy of this receptor, they inhibit platelet aggregation. In contrast, our findings that platelet aggregation by either thrombin, ADP alone, or ADP plus collagen is preserved in jararhagin-treated platelets clearly indicate that jararhagin does not target the ␣ IIb ␤ 3 integrin. The fact that platelet aggregation by collagen alone is inhibited by jararhagin points to the specificity of this enzyme for ␣ 2 ␤ 1 integrin whereby jararhagin degrades the ␤ 1 subunit of this collagen receptor as shown in Fig. 3. The data in Fig. 3 were obtained using B44 mAb which in these experiments precipitated the ␤ 1 subunit only, without associated ␣ 2 or ␣ 5 subunits as seen with other anti-␤ 1 mAbs (36). The 130-kDa band shown in this figure therefore contained only ␤ 1 protein without coprecipitated 135-kDa component of the reduced ␣ 5 chain. Fig. 3 also shows that part of ␤ 1 protein was not degraded by jararhagin. This can be explained by preservation of the fraction of ␤ 1 associated with ␣ 5 since, in contrast to ␣ 2 , ␣ 5 lacks the I domain required for jararhagin binding to the integrin heterodimer. We next investigated the serine/threonine and tyrosine phosphorylations of some proteins involved in platelet activation to gain further insight into the mechanisms of inhibition of platelet stimulation with collagen by these venom components.

Phosphorylation of Pleckstrin (p47) Indicates Different Effects of Contortrostatin and Jararhagin on Platelet Stimulation
Signaling-Following agonist stimulation, serine/threonine phosphorylation of pleckstrin (45) is associated with platelet granule secretion (46,47). We have already shown that in platelets treated by jararhagin, the secretion-associated p47 phosphorylation in response to collagen is diminished (38); Fig.  4 (i) shows the results of p47 phosphorylation, confirming that this platelet response to collagen is inhibited by jararhagin (lane 3). However, p47 phosphorylation was preserved when 32 P-labeled jararhagin-treated platelets were preactivated by ADP before exposure to collagen, although ADP alone (0.2 M) did not induce this phosphorylation. In contrast, in contortro-statin-treated platelets p47 was normally phosphorylated in response to collagen in the absence or presence of ADP. The phosphorylation of p47 was fully preserved in both jararhaginand contortrostatin-treated platelets stimulated with thrombin ( Fig. 4 (i)). These results thus show that the secretion-inducing pathway, when activated by collagen alone, was inhibited in jararhagin-but not in contortrostatin-treated platelets. The results of p47 phosphorylation in contortrostatin-treated platelets concur with preserved granule secretion recorded in platelets treated with other RGD-containing disintegrins (29,30). Fig. 5 (i) shows immunoblots of tyrosine-phosphorylated proteins from platelet lysates. Stimulation of untreated platelets with collagen caused increased phosphorylation of proteins in the 50 -76-kDa and 125-kDa ranges. Jararhagintreated platelets, either nonstimulated or stimulated by collagen, did not show an increase in protein phosphorylation compared with untreated controls. However, when jararhagintreated platelets were exposed to 0.2 M ADP, protein tyrosine phosphorylation clearly increased in response to platelet stimulation by collagen. These results demonstrate that jararhagin itself does not stimulate platelets and fully support the restoration of the platelet aggregation response to collagen with suboptimal ADP treatment (Fig. 1, trace 4).

Platelet Protein Tyrosine Phosphorylation in Response to Collagen Is Inhibited by Jararhagin but Not by RGD Disintegrins-
Similar experiments were carried out with platelets treated FIG. 3. Proteolysis of ␣ 2 ␤ 1 integrin by jararhagin. 125 I-Labeled platelet ␣ 2 ␤ 1 integrin was purified by immunoprecipitation with an anti-␤ 1 monoclonal antibody (B44 clone), as described previously (36). The purified integrin was treated with 100 nM either native or 5 mM 1,10-phenanthroline-inactivated jararhagin for 5 min at 37°C and analyzed by SDS-PAGE (7-12% gel) under reducing conditions. Dried gels were autoradiographed. with contortrostatin and echistatin. Results showed that, despite strong reduction of aggregation, protein tyrosine phosphorylation was fully preserved in contortrostatin-treated platelets when induced with either collagen alone or collagen and ADP (Fig. 5 (i)). However, the phosphorylation of a band of about 125 kDa (indicated by an arrowhead), which was preserved in contortrostatin-treated cells (Fig. 5 (i) lanes 7 and 8), was absent in echistatin-treated platelets. The 125-kDa band was identified as the focal adhesion kinase, pp125 FAK , by Western blotting (not shown).
To determine the most prominent substrates of tyrosine kinase(s) responding to collagen stimulation, 32 P-labeled platelets were also used. When these were treated with native jararhagin and then stimulated with collagen, analysis of PY20-immunoprecipitated proteins showed reduced protein tyrosine phosphorylation compared with the control. In particular, the absence of the major doublet labeled protein in the region of 70 kDa and strong reduction in phosphorylation of a 125-kDa protein (pp125 FAK ) were recorded (Fig. 6 (i)). Such a reduction in protein tyrosine phosphorylation in response to collagen is not observed in glycoprotein VI-deficient platelets expressing normal ␣ 2 ␤ 1 integrin (21) further confirming that the effect of jararhagin is restricted to ␣ 2 ␤ 1 signaling. We next examined the phosphorylation state of pp72 syk tyrosine kinase, an enzyme with key role in collagen-induced signaling.
The Phosphorylation of Tyrosine Kinase pp72 syk in Collagenstimulated Platelets Is Decreased by Jararhagin-In control platelets stimulated with collagen, pp72 syk was clearly phosphorylated (Figs. 4 (ii) and 6 (ii), lanes 2). However, when platelets were pretreated with jararhagin this response to collagen was markedly decreased. Thus, pp72 syk was not phosphorylated in jararhagin-treated platelets and this resembles the earlier observations in glycoprotein VI-deficient platelets by Ichinobe et al. (21). Platelets treated with RGD-disintegrins had phosphorylated pp72 syk despite the inhibition of aggregation when induced with either collagen alone or in association with ADP (Fig. 4 (ii) and 5 (ii)). Also, jararhagin, contortrostatin, or echistatin alone did not cause an increase in pp72 syk phosphorylation. The suboptimal amounts of ADP alone used in these studies did not cause substantial pp72 syk phosphorylation ( Fig. 4 (ii)). This implies that although the ␣ IIb ␤ 3 integrin had been activated, pp72 syk is phosphorylated only when collagen binds to this integrin. In platelets stimulated by thrombin, pp72 syk phosphorylation was not affected by either jararhagin or contortrostatin (Fig. 4 (ii)). The respective total immunoprecipitated pp72 syk blots are shown under the blots of phosphorylated pp72 syk (Fig. 5 (iii)).
These results in our experimental model demonstrate that the ability to cause pronounced inhibition of collagen-induced pp72 syk phosphorylation is a specific property of jararhagin. Since we have previously demonstrated (36), and confirmed here, that jararhagin cleaves the platelet ␣ 2 ␤ 1 integrin, this proteolysis could be the most likely cause of the inhibition of platelet interaction with collagen leading to pp72 syk phosphorylation.
Platelet ␤ 1 Subunit of ␣ 2 ␤ 1 Is Not Directly Associated with a Protein Kinase-Jararhagin cleaves the ␤ 1 subunit of ␣ 2 ␤ 1 integrin (Fig. 3) and, in other cell types, ␤ 1 has been shown to associate with cytoskeletal proteins (48,49) and to play an important role in signaling events (50,51). Therefore, we next searched for possible kinase activity associated with the ␤ 1 subunit, although attempts to demonstrate this have failed in the past (26). However, since in the latter report (26) immunoprecipitates were prepared with an anti-␣ 2 antibody, we carried out similar experiments using anti-␤ 1 antibody and employed a membrane-permeable chemical cross-linker in the search of other proteins that may be associated with the cytoplasmic tail of the immunoprecipitated protein. Although we examined either non-cross-linked or chemically cross-linked resting and collagen-activated platelets, we could detect neither a phosphoprotein nor protein-tyrosine kinase activity in these immunoprecipitates (not shown). These results therefore confirm the absence of protein kinase association with the ␣ 2 ␤ 1 integrin in platelets as previously reported by Asazuma et al. (26). DISCUSSION Here we have demonstrated that the venom metalloproteinase jararhagin inhibits an early stage of signaling in collagenstimulated platelets, markedly reducing the phosphorylation of the tyrosine kinase pp72 syk . The defective signaling in jararhagin-treated platelets can be attributed to the previously demonstrated proteolysis of ␣ 2 ␤ 1 integrin by jararhagin, clearly indicating the involvement of this integrin in signaling for pp72 syk phosphorylation by collagen. The activation of platelets by ADP restores the ability of jararhagin-treated platelets to generate this signal most likely because collagen binds to the activated ␣ IIb ␤ 3 integrin (10). In contrast, two RGD-containing venom disintegrins, contortrostatin and echistatin, inhibited platelet aggregation by collagen but did not impair collageninduced phosphorylation of pp72 syk . It is therefore likely that the occupancy of the major platelet adhesion/aggregation receptor ␣ IIb ␤ 3 integrin by these disintegrins, without any interference with signaling by ␣ 2 ␤ 1 integrin, is the main cause of inhibition of platelet aggregation in response to collagen. Thus, the present investigation shows that two distinct classes of venom-derived components, metalloproteinases and RGD-containing disintegrins, inhibit platelet function differently.
Jararhagin selectively inhibits collagen-induced platelet responses by first recognizing the I domain of ␣ 2 subunit of ␣ 2 ␤ 1 integrin (35) followed by proteolysis of the ␤ 1 subunit of the integrin (36). Both of these events are probably important for the resulting alterations in the functional properties of this integrin. Thus, we have found that upon inactivation of its FIG. 6. Immunoprecipitated phosphotyrosine proteins and pp72 syk from 32 P-labeled platelets. Untreated or jararhagin-treated platelets (100 nM jararhagin) were stimulated under stirring conditions with 2 g/ml collagen for 4 min. Cells were lysed and immunoprecipitation was carried out using either PY20 or an antibody against pp72 syk . Immunoprecipitated proteins were analyzed by SDS-PAGE (10% gel) and dried gels autoradiographed. i, phosphotyrosine proteins; ii, phosphorylated pp72 syk . J, jararhagin; Coll, collagen. The arrowhead indicates the position of the 70-kDa tyrosine-phosphorylated doublet. Data represent one of three similar experiments. catalytic site, jararhagin still binds to platelets (36) most probably via the part of its disintegrin domain containing the SECD sequence (38). Since we could not demonstrate an equilibrium binding of active jararhagin to platelets, it is likely that the proteolysis of the ␤ 1 chain affects the binding properties of the receptor for both the enzyme and collagen.
The demonstration that jararhagin-treated platelets respond to collagen in the presence of ADP suggests that platelet ␣ IIb ␤ 3 integrin and collagen retain their mutual binding capacities. ADP activates platelet ␣ IIb ␤ 3 integrin to bind RGD-containing ligands (43,44), including collagen (10), and this binding is presumably responsible for the restored signal transduction observed here. Since it is known that platelet stimulation with agonists such as thrombin increases the expression of platelet ␣ 2 ␤ 1 integrin by about 20% (52), it is possible that our observation of restored collagen-induced platelet aggregation in jararhagin-treated platelets in the presence of ADP might have been the result of such an increase. However, this seems unlikely because this newly expressed ␣ 2 ␤ 1 was also available for the attack by jararhagin during the incubation preceding platelet stimulation by collagen. Moreover, our demonstration that the restoration of platelet response to collagen by ADP is greatly reduced by the RGDS peptide, further supports our proposal that in ADP-stimulated jararhagin-treated platelets collagen binds to the activated ␣ IIb ␤ 3 integrin.
In contrast to jararhagin, the RGD-containing venom disintegrins impaired the platelet aggregation with either collagen alone or collagen and ADP, but did not interfere with pp72 syk activation. Since these disintegrins block ␣ IIb ␤ 3 integrin but do not interfere with the interaction of collagen with ␣ 2 ␤ 1 integrin, the latter interaction is presumably responsible for the preserved p72 syk phosphorylation observed in the presence of these agents. We also found that contortrostatin did not inhibit platelet pp72 syk activation in response to stimulation by thrombin as previously reported by Clark et al. (53). It has been reported that contortrostatin alone can trigger pp72 syk activation but at much higher concentrations (Ͼ500 nM) (53) than were used in the present study (100 nM). This effect, attributed to the dimeric structure of contortrostatin was not seen in our studies but, at equimolar concentrations, contortrostatin was a less effective platelet inhibitor than the monomeric disintegrin echistatin as judged by phosphotyrosine blots shown in Fig. 5. In platelets treated with the RGD-containing disintegrins, collagen-induced pp72 syk phosphorylation is followed by a series of downstream events including protein kinase C activation and phosphorylation of pleckstrin (Fig. 4), as well as granule secretion (29,30). In contrast, we have shown that in jararhagintreated platelets not only aggregation but also pleckstrin phosphorylation and granule secretion in response to collagen are absent. It appears therefore that all signals, with the exception of that responsible for myosin light chain phosphorylation and platelet shape change (38), are inhibited by the treatment of platelets with jararhagin.
Studies of integrin signaling in platelets have been mostly concerned with the ␣ IIb ␤ 3 integrin. It has been shown that the cytoplasmic tail of the ␤ 3 subunit is vital for the initiation of the outside-in signaling by this integrin (54,55) and that the ␤ 3 subunit coprecipitates with Src family kinases (55,56). Recent studies with other cells have demonstrated that the ␤ 1 subunit of ␤ 1 integrins also participates in signaling. During adhesion of epithelial cells to fibronectin, the ␤ 1 subunit of the ␣ 5 ␤ 1 integrin associates with the 59-kDa integrin-linked phosphoserine/phosphothreonine kinase, pp59 ILK , providing a downstream link with other cytoplasmic and cytoskeletal proteins (57). Upon fibroblast adhesion to the extracellular matrix, the ␤ 1 subunit of the ␣ 5 ␤ 1 integrin also coaggregates with pp125 FAK (58) and paxillin (50). Moreover, expression of variously modified transmembrane or cytoplasmic domains of the ␤ 1 subunit results in decreased activation of protein tyrosine phosphorylation in transformed cells (51). All this evidence, therefore, supports our proposal that the ␣ 2 ␤ 1 -dependent signal for activation of protein tyrosine kinases in platelets stimulated by collagen requires intact ␤ 1 subunit of the ␣ 2 ␤ 1 integrin. However, since association of kinase activity with this subunit could not be demonstrated either in our studies or in those of others (26), the nature of link between ␣ 2 ␤ 1 and the protein tyrosine kinase cascade in platelets remains to be established.
Although the ␣ 2 ␤ 1 integrin seems to be relevant for the platelet response to collagen both in our investigations and in those of others (7,9,15,25,26), evidence for the role of glycoprotein VI in collagen-induced platelet signaling cannot be ignored (12,21,23). The findings that platelet pp72 syk cannot be phosphorylated in the absence of either ␣ 2 ␤ 1 alone or glycoprotein VI alone, led to the suggestion that cooperation between these two receptors is necessary for collagen-induced signaling (21). Our preliminary experiments using chemical cross-linking have so far failed to demonstrate physical association of ␣ 2 ␤ 1 integrin and glycoprotein VI in normal platelets (not shown). Also we could not investigate whether jararhagin has any direct effect on glycoprotein VI in the present study because we could not obtain an antibody against this glycoprotein.
Apart from pp72 syk , Fc␥RII is also tyrosine phosphorylated at an early stage during the platelet-collagen interaction (15). It is known that binding of IV.3 mAb to Fc␥RII induces platelet aggregation (59 -61) and that this is accompanied by the phosphorylation of the receptor itself, pp72 syk and pp125 FAK (3, 62), but not of the FcR ␥-chain (14). We have found that jararhagintreated platelets fully aggregated with IV.3 mAb (not shown), indicating a preserved Fc␥RII receptor. Therefore, the absence of pp72 syk phosphrylation observed with jararhagin-treated platelets cannot be attributed to a change in Fc␥RII. This further reinforces our proposal that this defect in platelet signaling is caused by the action of jararhagin on the ␣ 2 ␤ 1 integrin.
Platelet glycoprotein IV (CD36) is also implicated as a primary receptor for platelet adhesion to collagen (11) and, when stimulated by antibodies, generates the signal for platelet aggregation (63). However, platelets lacking glycoprotein IV can be fully aggregated with collagen and show normal protein tyrosine phosphorylation (19,64). Our unpublished observation that a monoclonal antibody against glycoprotein IV (OKM5) promotes full aggregation of jararhagin-treated platelets, excludes a possible effect of jararhagin on this receptor.
In conclusion, our results demonstrate impaired signaling by collagen in platelets following proteolysis of the ␣ 2 ␤ 1 integrin by jararhagin. In particular, this is the first report of the loss of signal for tyrosine phosphorylation of pp72 syk attributable to an acquired defect in the function of ␣ 2 ␤ 1 integrin. Moreover, the signaling was largely restored by ADP stimulation apparently through collagen binding to activated ␣ IIb ␤ 3 . Thus, although the precise pathway of collagen-induced signaling in platelets still remains unresolved, our findings draw attention to the importance of both ␣ 2 ␤ 1 and ␣ IIb ␤ 3 integrins in this process.