INTRODUCTION

Platelet adhesion to collagen at sites of vascular injury initiates a cascade of signaling events leading to platelet activation and aggregation. Several collagen-binding proteins exist on the platelet surface that may mediate platelet activation by collagen, including the ₂₁ integrin (1, 2), glycoprotein (GP)¹ VI (3), GPIV (also known as GPIIIa or CD36 (4, 5)), and a 65-kDa protein (6, 7). Although multiple potential collagen receptors exist on platelets, the relative contribution of each in signaling collagen-induced platelet activation is unknown.

The ₂₁ integrin is necessary for the adhesion of platelets to collagen since patient platelets lacking this integrin do not adhere to or aggregate in response to soluble collagen fibers (8, 9). Additionally, antibodies against the ₂₁ integrin block platelet aggregation in response to soluble collagen (10) and adhesion to immobilized collagen (2, 11, 12). A more direct role for the ₂₁ integrin in early collagen-induced platelet signaling events, other than mediating the initial adhesion to collagen, has been assumed (13), but has not yet been clearly established. Platelet adhesion to immobilized collagen induces the phosphorylation of unknown 40-, 101-, and 105-kDa proteins and of pp125^FAK (14, 15, 16). However, the role of ₂₁ in these events is difficult to establish since anti-₂₁ integrin antibodies almost totally inhibit adhesion to immobilized collagen (12, 14). Therefore, it is not clear whether blocking the ₂₁ integrin disrupts specific signaling events or simply the initial adhesion of platelets to collagen.

A direct role for the ₂₁ integrin in mediating signal transduction in platelets would be consistent with emerging data linking integrins to various signaling events (for review, see Ref. 17). For example, the phosphorylation of numerous proteins during platelet aggregation has been shown to be dependent on the _IIb3 integrin (18, 19, 20), which becomes functionally active following platelet activation. Recently, Lin et al. (21) have found that engagement of ₁ integrins on monocytes induces Syk activation and NF-B-dependent gene expression. Additionally, integrins have been shown to mediate the phosphorylation of numerous proteins (22, 23, 24, 25, 26, 27), the activation of mitogen-activated protein kinases (28), the stimulation of metalloproteinase gene expression (29), and the activation of the Na⁺/H⁺ antiporter (30). The ₂₁ integrin has been implicated in signaling events in cells other than platelets since it mediates the activation of Ras in T-cells (31), the morphological organization of mammary cells in three-dimensional collagen gels (32), and collagenase expression in osteosarcoma cells (33). However, the mechanism by which the ₂₁ integrin or any other integrin transduces signals in cells is still poorly understood.

Although the ₂₁ integrin is likely to play a role in collagen-induced platelet activation, its role is confused by data suggesting that other collagen-binding proteins also contribute to platelet signaling. For example, it is clear that GPVI plays an important role in collagen-induced platelet activation since patient platelets deficient in GPVI are defective in collagen-induced aggregation and exhibit partially decreased cation-independent adhesion to collagen (3, 34, 35). GPVI directly contributes to signaling in the platelet, as shown by the recent finding that cross-linking of GPVI with antibodies causes the Fc-independent activation and phosphorylation of the non-receptor tyrosine kinase, Syk (36). A role for the p65 collagen-binding protein, which has a similar molecular mass compared with GPVI, is also suggested since an anti-p65 antibody inhibits collagen-induced platelet aggregation (7). A role for GPIV/CD36 is less clear since platelets from patients lacking GPIV (the Nak^a negative phenotype) aggregate and adhere normally to collagen (37, 38, 39, 40). In spite of this finding, GPIV may play some role since it is physically associated with the Src family kinases, Fyn, Yes, and Lyn (20). The presence of other collagen receptors on platelets may explain the finding that triple-helical collagen-like synthetic peptides are able to stimulate platelet aggregation in an ₂₁-independent manner (41).

Collagen stimulation of platelets leads to the phosphorylation of numerous proteins (Refs. 4, 13, 18, and 42; for review, see Ref. 43). Collagen induces the tyrosine phosphorylation and activation of phospholipase C2 (44, 45), which cleaves phosphatidylinositol 4,5-bisphosphate into diacylglycerol and inositol 1,4,5-trisphosphate. These second messengers, in turn, cause the activation of protein kinase C and the release of intracellular calcium, which are necessary for the subsequent events leading to granule secretion, activation of the _IIb3 integrin, aggregation, and actin polymerization (46, 47, 48, 49). The 72-kDa non-receptor tyrosine kinase, Syk, a member of the Zap70 kinase family, is also tyrosine-phosphorylated and activated in platelets following stimulation by collagen (50). Like collagen, stimulation of platelets through either the FcII receptor or CD9 also causes activation of both PLC2 and Syk (44, 51, 52, 53, 54, 55). Additionally, a number of other agonists (including thrombin (56) and the thromboxane A₂ mimetic, U46619 (57)) also stimulate Syk phosphorylation. The activation of platelets by thrombin causes the relocation of Syk to the cytoskeletal fraction (58, 59) and its association with phosphatidylinositol 3-kinase (60), suggesting that Syk has multiple roles in platelet signaling. Currently, the collagen-induced signaling pathways in platelets leading to the activation of either PLC2 or Syk have not been fully defined.

Because multiple collagen-binding proteins exist on platelets, but the relative contribution of each to collagen-induced platelet activation has not yet been determined, the purpose of this study was to determine which intracellular collagen-induced signaling events, if any, could be directly linked to the function of the ₂₁ integrin. Additionally, we sought to determine whether these signaling events contribute to collagen-induced platelet aggregation. We find that collagen stimulation of both Syk and PLC2 is dependent on the ₂₁ integrin. However, antibody cross-linking of ₂₁ is not sufficient to induce the phosphorylation of Syk and PLC2 unless the FcII receptor is also engaged, suggesting that ₂₁ functions in a co-stimulatory role with other receptors. Moreover, we find that blocking the ₂₁ integrin does not inhibit all collagen-induced phosphorylation events, which further suggests that multiple collagen receptors are simultaneously activated during platelet binding to collagen. We also identify one additional phosphorylation event induced by collagen, that of FcRII, even in the absence of added antibody. Finally, our data suggest that Syk kinase activity is upstream of PLC2; these signaling events had not previously been ordered on the same pathway in platelets. Moreover, inhibition of Syk activity suppresses collagen-induced platelet aggregation, suggesting an important role for Syk in the activation of platelets.

EXPERIMENTAL PROCEDURES

Piceatannol was a generous gift of Dr. Robert Geahlen (Purdue University, West Lafayette, IN). The anti-₂₁ integrin antibody JBS2 was a generous gift of Dr. John Wilkins (RDU Research Laboratory, Winnipeg, Manitoba, Canada). The anti-₂₁ integrin antibody P1H5 was a generous gift of Dr. William Carter (Fred Hutchinson Cancer Research Center, Seattle, WA), and P1E6 was purchased from Life Technologies, Inc. The anti-Src antibody 327 was a generous gift of Dr. Patricia Maness (University of North Carolina, Chapel Hill, NC). The polyclonal anti-Syk antibody was a generous gift of Dr. Andrew Chan (Washington University, St. Louis, MO) and was also purchased, along with PLC2, from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-phosphotyrosine antibody PY-20 and the recombinant RC-20 were purchased from Transduction Labs (Lexington, KY). Secondary F(ab)₂ anti-mouse IgG and horseradish peroxidase-conjugated anti-rabbit IgG were purchased from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Hybridomas for the anti-₁ integrin antibody (TS2/16), anti-FcRII antibody (IV.3), and isotype-matched control IgGs were obtained from American Type Culture Collection, and antibodies were prepared in the monoclonal core facility of the Department of Pharmacology (University of North Carolina, Chapel Hill). Protein G-Sepharose was purchased from Pharmacia (Uppsala). Tubulin was a generous gift of Dr. Michael Caplow (University of North Carolina, Chapel Hill). Collagen fibers were formed from bovine skin collagen (type I; Collaborative Biomedical Products, Becton Dickinson, Bedford, MA) by neutralizing the collagen solution in 10 mM Na₂HPO₄ to give a final concentration of 2 mg/ml. Prostaglandin E₁ and U46619 were obtained from Cayman Chemical Co., Inc. (Ann Arbor, MI). Deoxycholic acid was from Calbiochem. All other chemicals were from Sigma unless otherwise noted. Gel electrophoresis and transfers were performed using a Novex Xcell II system.

Blood was obtained from healthy consenting human donors using acid-citrate/dextrose as an anticoagulant, and platelets were prepared by centrifugation as described (61). Briefly, platelet-rich plasma was separated from whole blood at 800 rpm in a Beckman GP tabletop centrifuge for 25 min, treated with prostaglandin E₁ (50 ng/ml), and centrifuged at 1800 rpm for 20 min to pellet the platelets. Platelets were washed once in citrate/dextrose/NaCl (pH 7.0) and resuspended in buffer A (12 mM NaHCO₃, 138 mM NaCl, 5.5 mM glucose, 2.9 mM KCl, 50 mM HEPES (pH 7.3), 2 mM MgCl₂, and 1 mM CaCl₂). The platelet concentration was determined using a Thrombocounter (Coulter Corp., Hialeah, FL). Platelets were diluted in buffer A to 10⁹/ml and, where indicated, treated with 200 µg/ml bovine skin collagen fibers (Collaborative Research).

Platelet aggregation studies were performed to determine platelet function for every platelet preparation. Platelets were resuspended in buffer A at a concentration of 2 × 10⁸/ml, incubated at 37 °C under stirring conditions in an aggregometer (Chrono-Log Corp., Havertown, PA), and activated by the addition of collagen (200 µg/ml) or anti-integrin or control antibodies in the presence or absence of added fibrinogen. Incubation of isolated platelets in an aggregometer with added fibrinogen in the absence of agonist demonstrated that the platelets were not pre-activated.

Platelets were lysed by the addition of an equal volume of ice-cold 2 × RIPA buffer (50 mM HEPES (pH 7.3), 150 mM NaCl, 2% Nonidet P-40, 1% deoxycholate, 0.2% SDS, 2 mM sodium vanadate, 100 mM NaF, 20 mM sodium pyrophosphate, 10 µg/ml leupeptin, 20 µg/ml aprotinin, 5 mM Pefabloc (Boehringer Mannheim), 5 mM benzamidine HCl, and 2 mM EDTA) at 4 °C for 30 min, centrifuged at 15,000 × g for 15 min, and precleared for 1 h by the addition of protein G-Sepharose (GammaBind, Pharmacia). Cleared lysates were incubated overnight with the polyclonal anti-Syk or anti-PLC2 antibody or with the monoclonal antibody IV.3 to immunoprecipitate FcRII plus protein G-Sepharose at 4 °C and then washed three times with 1 × RIPA lysis buffer. Immunoprecipitated proteins were eluted in 1 × Laemmli sample buffer, reduced, subjected to SDS-polyacrylamide gel electrophoresis, and transferred to polyvinylidene difluoride membranes. To determine phosphotyrosine content, membranes were incubated with the anti-phosphotyrosine antibody (PY-20 or RC-20) conjugated to horseradish peroxidase and visualized using the ECL reagent (Amersham Corp.). To reprobe membranes for PLC2 or Syk, membranes were incubated overnight with NaN₃ in order to inhibit horseradish peroxidase activity, washed, and reprobed with the anti-PLC2 or anti-Syk antibody, followed by horseradish peroxidase-conjugated anti-rabbit IgG.

For antibody inhibition studies, platelets were incubated with the anti-₂ integrin antibody, P1E6, at a dilution of 1:1000 or isotype-matched IgG or without antibody for another 3 min, followed by stimulation with or without collagen for 2.5 min. For antibody stimulation experiments, platelets were preincubated with rabbit IgG for 3 min in order to block Fc receptors and incubated with anti-₁ integrin TS2/16 or IgG for 3 min, and where indicated, antibodies were cross-linked with anti-mouse F(ab)₂ for 5 min. It was later determined that treatment with rabbit IgG was not sufficient to completely block the Fc receptors (see below). F(ab)₂ fragments of TS2/16 were prepared by digesting the antibody with immobilized ficin and removing Fc fragments with a protein A column (Pierce). Gel electrophoresis confirmed that no contaminating intact IgG remained. Platelets were then lysed by the addition of 2 × RIPA buffer as described above. For peptide inhibition studies, the peptide GPAGKDGEAGA, representing residues 430-440 from the ₁ chain of collagen I (62), was synthesized at the University of North Carolina core protein synthesis facility (UNCCH-NIEHS). The peptide was diluted in buffer A, and platelets were preincubated with the indicated concentration of peptide for 5 min, stimulated with collagen, and then lysed in 2 × RIPA buffer.

For FcRII blocking experiments, platelets were preincubated for 5 min with 20 µg/ml IV.3 antibody and then incubated with anti-integrin antibodies that were biotinylated using the ECL protein biotinylation module (Amersham Corp.) according to the manufacturer's instructions. Biotinylated antibodies were dialyzed against phosphate-buffered saline prior to use. Biotinylated antibodies were cross-linked with 10 µg/ml avidin (Pierce) for 5 min prior to lysing the platelets in 2 × RIPA buffer.

Precleared lysates, prepared as described above, were immunoprecipitated with the anti-Syk or anti-Src antibody plus protein G-sepharose for 2 h and then washed three times with 1 × RIPA buffer without detergent. Immunoprecipitates were split into two samples: one for phosphotyrosine determination and one for the kinase assay, which was resuspended in 20 µl of kinase buffer (25 mM HEPES (pH 7.5), 10 mM MnCl₂, and 10 mM p-nitrophenyl phosphate) plus 1 µM ATP, 5 µCi of [-³²P]ATP (Amersham Corp.), and 10 µg of tubulin and incubated at 24 °C for 2 min. The assay was terminated by adding 3 × Laemmli sample buffer and boiling the samples for 3 min. Samples were electrophoresed, and the gels were stained with Coomassie Brilliant Blue, dried, and subjected to autoradiography.

RESULTS

Kinetics of the Collagen-induced Phosphorylation of PLC2 and Syk

Treatment of platelets with collagen (200 µg/ml) under non-stirring conditions caused the phosphorylation of Syk on tyrosine within 1 min (Fig. 1). This collagen-induced phosphorylation persisted for 20 min. Other investigators have also noted similar kinetics of Syk phosphorylation following collagen stimulation (48). Stimulation of platelets by thrombin (Fig. 1) or the thromboxane A₂ mimetic, U46619 (data not shown), also induced the tyrosine phosphorylation of Syk between 30 s and 2.5 min. Collagen-induced tyrosine phosphorylation of PLC2 lagged behind that of Syk since obvious phosphorylation did not occur until 2.5 min, with only a faint phosphorylated PLC2 band visible at 1 min (Fig. 1). Stimulation of Syk and PLC2 phosphorylation was independent of platelet aggregation since it occurred in the absence of stirring or when aggregation was blocked by the addition of a synthetic RGDW-containing peptide, which blocks fibrinogen binding to the _IIb3 integrin (data not shown). We also observed the phosphorylation of PLC2 after treatment with thrombin (Fig. 1) or U46619 (data not shown) for 2.5 min. This result is in contrast to the results of Daniel et al. (45) or Blake et al. (44), who did not find PLC2 phosphorylation following thrombin treatment, but is consistent with the results of Tate and Rittenhouse (63).

Kinetics of collagen-induced Syk and PLC2 tyrosine phosphorylation.

Tyrosine Phosphorylation of PLC2 and Syk Is Dependent on the ₂₁ Integrin

To determine the role of the ₂₁ integrin in collagen-induced platelet signaling events, platelets were pretreated with an anti-₂₁ integrin mAb (P1E6) that blocks ₂₁-mediated binding of platelets to collagen (11). As shown in Fig. 2A, P1E6 treatment dramatically decreased the phosphorylation of both Syk and PLC2. In control experiments, an isotype-matched IgG did not inhibit the phosphorylation of Syk or PLC2 (Fig. 2A). Inhibition of Syk and PLC2 phosphorylation was confirmed using a different function-blocking anti-₂ integrin antibody (P1H5) (data not shown). These results suggest that collagen interaction with ₂₁ is necessary for the phosphorylation of PLC2 and Syk.

An anti-₂₁ integrin-blocking antibody inhibits Syk and PLC2 phosphorylation.

We also examined the effect of the anti-₂₁ integrin mAb on the total phosphotyrosine content of collagen-treated platelets. Antibody blocking of the ₂₁ integrin caused the loss of several phosphotyrosine bands, including a band just larger than 100 kDa. However, not all signaling events were affected since the collagen-induced tyrosine phosphorylation of several other unidentified proteins, including at least three major bands at about 46, 50, and 65 kDa, was not inhibited by the presence of the anti-₂₁ integrin mAb (Fig. 2B), suggesting that collagen interacts with the platelet surface even when the ₂₁ integrin has been blocked.

In aggregation assays, the anti-₂₁ integrin mAb (P1E6) inhibited collagen-induced platelet aggregation, while control IgG had no effect (Fig. 2C). This inhibition was specific since P1E6 did not inhibit platelet aggregation stimulated by thrombin (data not shown).

As an additional approach to block the ₂₁ integrin, we used a synthetic peptide from the 1(I) collagen chain, GPAGKDGEAGA (KDGEA), a putative ligand of ₂₁ that blocks the cation-dependent adhesion of platelets and breast cells to collagen (62). Pretreatment of platelets with a 3 mM concentration of the KDGEA peptide resulted in the partial inhibition, and pretreatment with 10 mM resulted in the total inhibition of both Syk and PLC2 tyrosine phosphorylation (data not shown). The concentration of the KDGEA peptide that was necessary for inhibition is in close agreement with the concentration necessary to inhibit platelet adhesion to collagen, which was partial at 3 mM and total at 6 mM (62). As noted for treatment with the anti-₂₁ integrin antibody, pretreatment of platelets with KDGEA did not inhibit all tyrosine phosphorylation events (data not shown). Thus, inhibition of Syk and PLC2 phosphorylation with both an ₂₁-specific antibody and peptide suggests that occupancy of the ₂₁ integrin is required for collagen-induced Syk and PLC2 phosphorylation.

Antibody Cross-linking of the ₂₁ Integrin Results in the Phosphorylation of Syk and PLC2 in an Fc Receptor-dependent Manner

To determine whether cross-linking of ₂₁ is sufficient to induce tyrosine phosphorylation of Syk and PLC2, platelets were pretreated with either a stimulatory anti-₁ integrin mAb (TS2/16) or an isotype-matched control IgG, followed by cross-linking of the primary antibody with a secondary anti-mouse F(ab)₂ antibody to cluster the integrin receptors. Treatment of platelets in suspension with the TS2/16 antibody, but not control IgG, caused the phosphorylation of both Syk and PLC2 (Fig. 3A, left panel). Interestingly, cross-linking TS2/16 with the secondary antibody was not necessary since TS2/16 alone also induced the phosphorylation of Syk and PLC2. However, there was an enhancement following cross-linking with the secondary antibody (compare lanes ± secondary). These results are consistent with the results of Lin et al. (21), who found that TS2/16 alone, in the absence of secondary antibody, could stimulate Syk phosphorylation in a monocytic cell line.

Cross-linking the ₂₁ integrin with antibody induces Syk and PLC2 phosphorylation in an FcRII-dependent manner.

To determine whether cross-linking of the ₂ subunit stimulates signaling in platelets, we used JBS2, an anti-₂ integrin mAb that stimulates collagen binding and Ras activation in T-cells (31). Treatment of platelets in suspension with JBS2 stimulated the tyrosine phosphorylation of PLC2 (Fig. 3A, right panel). In contrast to the findings with TS2/16, it was necessary to cross-link the JBS2 antibody in order to induce PLC2 phosphorylation. This difference may reflect a difference in the way that the TS2/16 and JBS2 antibodies bind to and stimulate the ₂ or ₁ integrin subunits. It is possible that the TS2/16 antibody cross-links two integrin heterodimers even in the absence of the secondary antibody. We also found that the level of JBS2-induced phosphorylation was less than that noted when TS2/16 was used. This difference may be due to the very strong stimulatory activity of the TS2/16 antibody compared with the JBS2 antibody. Furthermore, since platelets have ₅₁ and ₆₁, stimulation of these ₁ integrins may also contribute to the stronger signal noted with TS2/16. Our results with JBS2 demonstrate that specifically cross-linking the ₂₁ integrin with intact mAb induces signaling in platelets.

Although control IgG did not result in the phosphorylation of Syk or PLC2, we investigated whether Fc receptors contributed to signaling following cross-linking with anti-integrin antibodies. F(ab)₂ fragments of the TS2/16 antibody failed to stimulate the phosphorylation of either Syk or PLC2 (Fig. 3B), demonstrating that the Fc portion of the TS2/16 antibody contributes to its activity. To check the activity of the F(ab)₂ fragments, we took advantage of the fact that the TS2/16 antibody enhances binding of the integrin to collagen and thus enhances platelet aggregation in response to suboptimal collagen concentrations. We found that the F(ab)₂ fragments of TS2/16 retained this ability (data not shown), indicating that the failure of F(ab)₂ fragments to stimulate platelets is not due to a loss of their activity.

To determine more directly whether FcRII, which is the only Fc receptor on platelets, contributes to anti-integrin antibody signaling, FcRII was blocked with 20 µg/ml IV.3 antibody prior to stimulation with intact antibodies. To avoid using a secondary antibody to cluster anti-integrin antibodies, which would stimulate the platelets by also cross-linking the IV.3 antibody, the anti-integrin and control IgG antibodies were biotinylated and cross-linked using avidin. Pretreatment of platelets with IV.3 blocked stimulation of Syk and PLC2 phosphorylation by biotinylated TS2/16 (Fig. 3C), once again suggesting a contribution of FcRII. Similar results were obtained with biotinylated JBS2 (data not shown). Anti-integrin antibody stimulation led not only to the phosphorylation of Syk and PLC2, but also to the tyrosine phosphorylation of FcRII itself (data not shown). This phosphorylation of FcRII was also blocked by the IV.3 antibody. It is unlikely that anti-integrin antibody signaling is due solely to a nonspecific effect of antibody binding to FcRII since control IgG antibodies consistently failed to activate Syk or PLC2 phosphorylation. These results suggest that antibody clustering of the ₂₁ integrin alone is insufficient to induce Syk and PLC2 phosphorylation and that FcRII co-stimulates platelets treated with anti-integrin antibodies. Cross-linking the ₂₁ integrin on the platelet surface was not sufficient to induce aggregation since treatment of platelets with intact TS2/16 or JBS2 antibodies with or without the secondary antibody did not cause the aggregation of platelets (data not shown).

Collagen Causes an ₂₁-independent Phosphorylation of FcRII

Because FcRII was necessary for anti-₂₁ integrin antibody-stimulated phosphorylation of Syk and PLC2, we determined whether FcRII was also a target for tyrosine phosphorylation following collagen stimulation. Collagen treatment of platelets induced the phosphorylation of FcRII within 1 min (Fig. 4A). This phosphorylation proceeded for at least 5 min following collagen stimulation. Unlike the collagen-induced phosphorylation of Syk and PLC2, the phosphorylation of FcRII was not blocked by pretreatment of platelets with the anti-₂ integrin antibody (P1E6) (Fig. 4B). In fact, the experiment shown in Fig. 4B (third lane)) suggests that P1E6 may cause a slight stimulation of FcRII phosphorylation even in the absence of collagen. Even with this background level of phosphorylation, there is still a further stimulation of FcRII phosphorylation upon collagen addition (Fig. 4B, compare the third and fourth lanes), indicating that P1E6 does not inhibit collagen stimulation of FcRII phosphorylation. Thus, although antibody stimulation of ₁ integrins can induce FcRII phosphorylation, collagen stimulation of FcRII phosphorylation is ₂₁ integrin-independent.

Collagen induces the ₂₁-independent phosphorylation of FcRII.

Collagen stimulation of FcRII phosphorylation was not inhibited by pretreating the platelets with the blocking IV.3 anti-FcRII antibody (Fig. 4C). This implies that collagen probably does not bind directly to the FcII receptor. Additionally, this result suggests that collagen stimulation of FcRII phosphorylation does not occur by inducing platelet release of immune complexes that subsequently bind the FcII receptor, but more likely by an internal signaling pathway. Pretreatment of platelets with IV.3 also did not inhibit the collagen-induced phosphorylation of Syk or PLC2 (Fig. 4D) or collagen-induced platelet aggregation (data not shown), suggesting that if FcRII plays a role in collagen stimulation of platelets, this role is not mediated by its ability to bind Fc ligands. Thus, stimulation of Syk and PLC2 phosphorylation by collagen differs from stimulation by anti-₂₁ integrin antibodies, which was blocked by preincubation with the IV.3 antibody. This difference suggests that stimulation by collagen does not share the same requirement for FcRII binding as stimulation by anti-integrin antibodies. In the case of antibody stimulation, FcRII may substitute for a co-receptor that is normally stimulated by collagen.

Inhibition of Syk Kinase Activity Also Inhibits the Phosphorylation of PLC2

To determine whether Syk is on the same pathway and upstream of PLC2, platelets were treated with the Syk-selective kinase inhibitor, piceatannol (64). Platelets were pretreated for 10 min with various concentrations of piceatannol or Me₂SO, followed by treatment with collagen for 2.5 min. Syk phosphorylation was dramatically reduced following treatment with 10 µg/ml piceatannol and completely eliminated following treatment with 30 µg/ml piceatannol (Fig. 5A). The tyrosine phosphorylation of PLC2 was closely correlated with that of Syk since the phosphorylation of PLC2 was also decreased following treatment with 10 µg/ml piceatannol and completely eliminated following treatment with 30 µg/ml piceatannol (Fig. 5A). This finding is consistent with a model placing Syk activity upstream of PLC2. Piceatannol was less effective in inhibiting the collagen-induced phosphorylation of FcRII since treatment with 10 µg/ml piceatannol did not inhibit, and 30 µg/ml piceatannol only partially inhibited FcRII phosphorylation (Fig. 5A, bottom panel). This is consistent with results suggesting that FcRII is phosphorylated by Src in platelets (65). Pretreatment of platelets with piceatannol also inhibited the tyrosine phosphorylation of several other (but not all) proteins (Fig. 5B), implying that Syk may be involved in several phosphorylation events in stimulated platelets.

Piceatannol inhibits both Syk and PLC2 phosphorylation.

Syk is known to have autophosphorylating activity, and its level of tyrosine phosphorylation correlates with its kinase activity in platelets (58). To verify that piceatannol inhibited Syk kinase activity as well as the phosphotyrosine content of Syk, immune complex kinase assays were performed. Pretreatment of platelets with piceatannol inhibited Syk kinase activity as measured by autophosphorylation and by the phosphorylation of an exogenous substrate, tubulin (Fig. 5C). Treatment of platelets with 30 µg/ml piceatannol inhibited Syk kinase activity by 72% (Table I). In contrast, Src kinase activity was not significantly affected by pretreatment of platelets with piceatannol (Fig. 5C and Table I). Treatment of platelets with 10 µg/ml piceatannol did not significantly inhibit Syk kinase activity (Fig. 5C and Table I), even though this concentration partially inhibited the tyrosine phosphorylation of Syk (Fig. 5A). The reason for this difference is not clear, but may be due to the removal of piceatannol in washing steps before the kinase assay; kinase activity may therefore appear to be less sensitive to inhibition by piceatannol than in the direct phosphotyrosine blots shown in Fig. 5A. Our results are consistent with those of Oliver et al. (64), who found that pretreatment of mast cells with 50 µg/ml piceatannol inhibited Syk kinase activity, but not the Src family kinase Lyn. Thus, the concentrations of piceatannol used here seem to be relatively specific for Syk and do not affect Src family kinases, which suggests that the inhibition of PLC2 phosphorylation by piceatannol is due to an effect on Syk activity. However, the possibility still exists that piceatannol also affects other kinases in platelets as well.

Table I. Syk and Src kinase activities following piceatannol treatment of platelets [Piceatannol] Kinase activitya Sykb Srcc % of control 1 µg/ml 101 103 3 µg/ml 82 90 10 µg/ml 76 84 30 µg/ml 28 81 a  A quantification of kinase activity was determined by excising the phosphorylated bands from the dried gel shown in Fig. 5C and counting them in a scintillation counter. Percent of control is the ratio of the dpm in the sample treated with piceatannol to the dpm in the Me2SO control band. b  Syk kinase activity was determined by excising the tubulin band from the dried gel shown in Fig. 5C. c  Src kinase activity was determined by excising the Src band from the dried gel shown in Fig. 5C.

The effects of piceatannol on platelet aggregation were also studied. Collagen-induced platelet aggregation was partially inhibited by pretreatment of platelets with 10 µg/ml piceatannol and totally inhibited by pretreatment with 30 µg/ml piceatannol (Fig. 5D). Collagen-induced platelet shape change was also inhibited by 30 µg/ml piceatannol. Piceatannol also inhibited platelet aggregation in response to thrombin and U46619. As with collagen-induced aggregation, 10 µg/ml piceatannol caused a partial inhibition, and 30 µg/ml piceatannol caused a total inhibition of thrombin-induced aggregation (Fig. 6). Unlike collagen stimulation, platelets pretreated with 30 µg/ml piceatannol still exhibited shape change in response to thrombin or U46619 (Fig. 6). This suggests that the mechanisms leading to collagen-induced platelet shape change are different from those stimulated by G-protein-coupled receptors. The inhibition of platelet aggregation by piceatannol suggests an important role for Syk in platelet activation by a number of agonists.

DISCUSSION

We have shown that occupancy of the ₂₁ integrin is necessary, but not sufficient, for the collagen-induced tyrosine phosphorylation of Syk and PLC2 in platelets. Blocking of the ₂₁ integrin using either inhibitory antibodies or a collagen-derived peptide that is a putative ₂₁ ligand inhibited the phosphorylation of both Syk and PLC2. Additionally, cross-linking of the ₁ or ₂ subunits with stimulatory antibodies in solution co-stimulated, along with FcRII, the phosphorylation of both Syk and PLC2. These early platelet signaling events had not previously been linked directly to occupancy of the ₂₁ integrin. Furthermore, we find evidence for a possible role for the FcII receptor in collagen stimulation since collagen induced the phosphorylation of FcRII, a finding that, to our knowledge, had not previously been reported. By using the Syk-selective inhibitor, piceatannol, we are able to place Syk activation upstream of PLC2 in platelets. Our results begin to define a pathway whereby collagen binding to the ₂₁ integrin contributes to the activation of Syk and the subsequent phosphorylation of numerous proteins including PLC2, leading to platelet activation and aggregation.

Although Syk activation has been assumed by some investigators to occur upstream of PLC2, this order of events in platelets was not previously established. Several pieces of evidence support the positioning of Syk activation upstream of PLC2. First, inhibition of Syk by the Syk-selective kinase inhibitor, piceatannol, inhibited not only Syk kinase activity and tyrosine phosphorylation, but also PLC2 phosphorylation. In our hands, in agreement with Oliver et al. (64), concentrations of piceatannol that inhibited Syk activity had no significant effect on Src family kinase activity, suggesting that piceatannol specifically inhibits Syk. However, it remains possible that other kinases are also affected by piceatannol. Second, kinetic studies demonstrated that Syk phosphorylation occurred before PLC2 phosphorylation. This agrees with the observation that, in B-cells, a knockout of the syk gene results in cells that no longer phosphorylate PLC2 in response to B-cell receptor cross-linking (66). It remains to be determined whether Syk directly phosphorylates PLC2 or whether there are other kinases in between Syk and PLC2.

Our data demonstrate that both the ₂₁ integrin and the subsequent activation of Syk and PLC2 are necessary for collagen-induced platelet aggregation since either anti-₂₁ integrin antibodies or piceatannol inhibited aggregation. Other investigators have also noted that collagen-induced platelet aggregation is inhibited by some anti-₂₁ integrin antibodies (10) or in platelets missing the ₂₁ integrin (2, 8). Although the ₂₁ integrin/Syk pathway was required for collagen-induced platelet aggregation, it was not sufficient since intact antibody cross-linking of the ₂₁ integrin and subsequent Syk activation did not induce platelet aggregation. This suggests that multiple signaling events, most likely mediated through additional collagen-binding proteins, are necessary to cause full platelet activation. Antibody cross-linking of one putative collagen-binding protein, GPVI, induces the Fc-independent activation of Syk and platelet aggregation (36). It is of interest that both GPVI and the ₂₁ integrin seem to be necessary to elicit the full range of platelet responses to collagen since patients deficient in either receptor have platelets that are defective in collagen-induced aggregation (3, 8, 9). The potential necessity for both receptors to elicit the full range of platelet responses to collagen suggests that there may be other signals independent of the Syk/PLC2 pathway that contribute to platelet responses to collagen, some of which are linked to the ₂₁ integrin and some to GPVI, such that eliminating either receptor might eliminate a portion of the collagen-induced signaling pathway in platelets. Evidence that stimulation of platelets by collagen occurs through two different sites, both of which are necessary for platelet activation by collagen, has also been suggested by Santoro et al. (67). It remains to be determined just how the ₂₁ integrin and other collagen receptors, including GPVI, cooperate in producing the total platelet response to collagen.

Our observation that some collagen-induced phosphorylation events occur even when the ₂₁ integrin is blocked by antibodies or peptides suggests, first, that blocking the ₂₁ integrin blocks specific signaling events rather than totally inhibiting all interactions of platelets with soluble collagen. Second, this observation suggests that collagen binding to other receptors occurs independent of, and in addition to, collagen binding to the ₂₁ integrin. This finding differs from the observation that platelets deficient in the ₂₁ integrin do not adhere to collagen (8, 9) and that anti-₂₁ integrin mAbs almost completely block adhesion to immobilized collagen (11). It is possible that soluble collagen simultaneously binds the ₂₁ integrin and other collagen receptors on platelets, but that the ₂₁ integrin is required for stable adhesion to immobilized collagen. The other collagen-binding proteins on platelets may be too weak or interact too transiently to mediate stable adhesion to collagen in the absence of the ₂₁ integrin, but are probably necessary to elicit the full range of platelet responses to collagen.

The notion that the ₂₁ integrin works in conjunction with other receptors is consistent with FcRII and the ₂₁ integrin acting together in antibody cross-linking experiments. Anti-integrin antibodies did not induce phosphorylation of Syk and PLC2 when the Fc portion of the antibody was removed or when FcRII was blocked with the IV.3 antibody. However, antibody-induced tyrosine phosphorylation was not solely due to FcRII since only the anti-₁ integrin antibody, and not control IgG, caused the phosphorylation of Syk and PLC2. It is likely that the anti-integrin antibody caused a clustering of the integrin with FcRII. This heterotypic clustering differs from homotypic clustering of FcRII alone since anti-integrin antibodies did not induce the platelet aggregation that is observed upon cross-linking with the anti-FcRII antibody (65). Functional cooperation between integrins and the FcII receptor has also been described for ₂ integrins in neutrophils (68, 69).

Our results are not the first in which co-stimulation of FcRII contributes to antibody stimulation of platelets. Most notably, antibodies against the tetraspanin CD9 strongly stimulate platelet aggregation, but only in conjunction with FcRII (51, 70). It is thus of interest that CD9 has recently been co-immunoprecipitated with ₁ integrins from various cells (71, 72) and modulates ₁ integrin function in B-cells (73). An interaction between CD9 and the ₂₁ integrin was specifically not found by Berditchevski et al. (72), but remains to be tested in platelets.

Unexpectedly, collagen stimulation of platelets, in the absence of antibody, led to the ₂₁-independent phosphorylation of FcRII. Previously, the tyrosine phosphorylation of an unidentified 40-kDa protein following collagen stimulation had been reported (15, 16). Our results suggest that this 40-kDa protein might be FcRII. Interestingly, this phosphorylation was not blocked by an anti-₂₁ integrin antibody, suggesting that the phosphorylation of FcRII might occur by collagen binding through a different platelet receptor. This receptor might be GPVI since anti-GPVI antibodies stimulate the phosphorylation of a 40-kDa protein (34). Additionally, collagen-induced FcRII phosphorylation was also not blocked by the IV.3 anti-FcRII antibody, implying that it did not occur as a result of binding to an Fc ligand and may, instead, be due to internal signaling events following collagen stimulation. Collagen is not known to bind FcRII directly. Although collagen binding directly to FcRII is unlikely, if it does occur, it would have to bind at a site not blocked by IV.3. Our finding differs from that of Yanaga et al. (55), who observed no phosphorylation of FcRII following collagen stimulation. However, their experiments were performed in the presence of EGTA, which would affect the divalent cation-dependent functioning of the ₂₁ integrin and possibly other collagen receptors and alter the overall response of the platelets to collagen.

In spite of these observations, the role of FcRII in collagen signaling, if any, is unclear. Blocking FcRII with the IV.3 antibody had no effect on either collagen-induced FcRII, Syk, and PLC2 phosphorylation or subsequent platelet aggregation. One possibility is that FcRII plays no role in collagen-induced platelet activation, but substitutes for the role of a co-receptor or co-stimulator in the case of anti-₂₁ integrin antibody stimulation. In the more physiologic case where ₂₁ is bound by collagen, this co-receptor may instead be one of the other collagen-binding proteins such as GPVI. Alternatively, it is possible that the presence of phosphorylated FcRII, in the absence of Fc ligand binding, is sufficient to contribute to collagen-induced platelet activation. FcRII contains a tyrosine-based activation motif that, when tyrosine-phosphorylated, binds and activates Syk in platelets (55, 74), consistent with the role of tyrosine-based activation motifs in T- and B-cell receptor signaling (75, 76, 77, 78). Integrin  and  subunits lack such a motif, although the ₁ subunit does have two tyrosine residues with similar spacing as those found in tyrosine-based activation motifs. An attractive model is that FcRII becomes phosphorylated as a result of collagen activation and then serves as a docking molecule to localize Syk to receptor complexes that might include the ₂₁ integrin and perhaps other collagen receptors. Because FcRII phosphorylation was less sensitive to inhibition by piceatannol than was Syk or PLC2 phosphorylation, it is likely that FcRII is phosphorylated upstream of Syk by another kinase. This kinase may be Src since the phosphorylation of tyrosine-based activation motifs is thought to occur through the action of Src family kinases (79, 80), which are not inhibited by the concentrations of piceatannol used in our assays (64). Furthermore, Src is implicated as the kinase that phosphorylates FcRII following cross-linking of FcRII in platelets (65). The determination of whether FcRII contributes to collagen activation, or how it does so, awaits further study.

While other platelet agonists, like collagen, stimulate downstream signaling events that include protein kinase C activation and calcium second messenger signaling, there is evidence that part of the signaling pathway stimulated by either collagen or the FcII receptor diverges from that stimulated by G-protein-coupled agonists such as thrombin or thromboxane A₂. Both collagen- and FcRII-induced platelet activation are insensitive to elevated cAMP (15, 48). In contrast, cAMP is a potent inhibitor of platelet activation induced by thrombin, the thromboxane mimetic (U46619), or ADP (48). Additionally, stimulation by collagen or via the FcII receptor is similarly sensitive to inhibition by phenylarsine oxide, unlike thrombin stimulation of platelets (16, 54). Finally, stimulation of Syk phosphorylation by collagen and via FcRII is insensitive to the combined inhibition of protein kinase C and intracellular calcium chelation, also unlike stimulation by thrombin (55).

Some investigators have proposed that the difference between collagen versus G-protein-coupled receptors could be explained by the fact that collagen (and FcRII) activates the PLC2 isoform, whereas thrombin activates PLC. However, conflicting reports exist as to whether thrombin also stimulates PLC2 (44, 45, 63, 65). Our finding that thrombin and U46619 also stimulated PLC2 phosphorylation is consistent with the findings of Tate and Rittenhouse (63), but is in contrast to the findings of Blake et al. (44) and Daniel et al. (45). This may relate to the fact that these investigators examined PLC2 phosphorylation only up to 2 min following stimulation by thrombin or U46619, while in this study, we noted PLC2 phosphorylation beginning at 2.5 min.

Aggregation studies following treatment of platelets with piceatannol suggest the general importance of Syk activity to platelet aggregation. Piceatannol inhibited platelet aggregation not only in response to collagen, but also in response to thrombin and U46619. Interestingly, piceatannol completely inhibited platelet shape change, as determined by aggregation traces, in response to collagen, but not in response to thrombin or U46619. These results imply a role for Syk in the initial collagen-induced shape change, a role that may not be shared by G-protein-coupled receptors. A role for Syk upstream of platelet aggregation is further supported by the findings that Syk activation is rapid (50, 58, 60) and occurs even in the absence of platelet aggregation and _IIb3 integrin engagement (58). Thus, we found that pretreatment of platelets with the RGDW peptide, which blocks fibrinogen binding to the _IIb3 integrin, did not inhibit collagen-induced Syk activation. In spite of our finding that Syk is important to platelet aggregation, recent studies of Syk null mice demonstrate no obvious platelet defects (81, 82). Further studies will be needed to fully determine the role of Syk in platelet aggregation. Additionally, although Syk activation might be required for platelet aggregation, activation of Syk is not sufficient to cause platelet aggregation since cross-linking the ₂₁ integrin by intact antibodies caused phosphorylation of Syk, but not platelet aggregation.

A role for Syk in integrin-mediated signaling was also found by Lin et al. (21) in studies of the ₅₁ integrin in the monocytic cell line THP-1. Thus, Syk activation may be downstream of a number of integrin receptors in hematopoietic cells. In support of this, about half of the Syk that becomes activated in aggregating platelets occurs following engagement of the _IIb3 integrin during platelet aggregation (58), although Syk is activated prior to _IIb3 engagement as well. It will be interesting to determine whether this represents different subpopulations or localization of Syk in the platelet.

Activation of focal adhesion kinase differs from Syk since its activation in stimulated platelets occurs only after platelet aggregation and requires _IIb3 integrin engagement (19, 83, 84, 85). Focal adhesion kinase phosphorylation is also known to be downstream of the ₂₁ integrin in the case of platelet adhesion to immobilized collagen (14). Haimovich et al. (14) also found a role for the ₂₁ integrin in the phosphorylation of two other unidentified 100- and 105-kDa proteins. In our studies, treatment of platelets with soluble collagen also caused the phosphorylation of a similar sized protein (100-110 kDa) that was inhibited by blocking the ₂₁ integrin with antibody. Additionally, treatment of platelets with piceatannol also caused the loss of a 100-110-kDa phosphorylated band. The identity of this protein and whether it is the same as that noted previously (14) are not known.

In summary, we have elucidated some steps of ₂₁ integrin signaling in platelets by linking specific events to the receptor. It will be of future interest to determine the proximal signaling events that lead to Syk activation and how these steps are regulated by the ₂₁ integrin. Additionally, it will be of interest to identify the signaling events unrelated to Syk that may also have a role in collagen-induced aggregation. Finally, it will be important to determine how other collagen-binding proteins contribute to the overall response of platelets to collagen.