Integrin-independent Tyrosine Phosphorylation of p125 fak in Human Platelets Stimulated by Collagen*

Collagen fibers or a glycoprotein VI-specific collagen-related peptide (CRP-XL) stimulated tyrosine phosphorylation of the focal adhesion kinase, p125 fak (FAK), in human platelets. An integrin a 2 b 1 -specific triple-helical peptide ligand, containing the sequence GFOGER (sin-gle-letter nomenclature, O 5 Hyp) was without effect. Antibodies to the a 2 and b 1 integrin subunits did not inhibit platelet FAK tyrosine phosphorylation caused by either collagen fibers or CRP-XL. Tyrosine phosphorylation of FAK caused by CRP-XL or thrombin, but not that caused by collagen fibers, was partially inhibited by GR144053F, an antagonist of integrin a IIb b 3 . The in- tracellular Ca 2 1 chelator, BAPTA, and the protein kinase C inhibitor, Ro31-8220, were each highly effective inhibitors of the FAK tyrosine phosphorylation caused by collagen or CRP-XL. These data suggest that, in human platelets, 1) occupation or clustering of the integrin a 2 b 1 is neither sufficient nor necessary for activation of FAK, 2) the fibrinogen receptor a IIb b 3 is not required for activation of FAK by collagen fibers, and 3) both intracellular Ca 2 1 and protein kinase C activity are essential intermediaries of FAK activation.

ent cells such as fibroblasts and platelets (33). FAK is of particular interest, because it is considered a key intermediary of signaling through integrins (34 -36).
Phosphorylation of FAK occurs at five tyrosine residues and correlates with an increase in FAK tyrosine kinase activity. Autophosphorylation of tyrosine 397 allows it to bind the c-Src family member Fyn (37,38), whereas phosphorylation of tyrosine 407 and the C-terminal tyrosine 861 may support the interaction of FAK with other signaling molecules (39). Tyrosines 576 and 577 are phosphorylated by c-Src (40) and contribute to the regulation of the catalytic activity of FAK. Another possible regulatory mechanism for FAK is proteolytic cleavage by the Ca 2ϩ -dependent protease calpain, which leads to a reduction in its autophosphorylation (41).
Evidence from fibroblasts suggests that occupation of ␤ 1 integrins is a sufficient stimulus to activate FAK (35,42,43). Collagen binding to ␣ 2 ␤ 1 in T cells protects them from apoptosis in a FAK-dependent manner (36). Adhesion of platelets to monomeric collagen occurs through ␣ 2 ␤ 1 . A causal relationship has been proposed between ␣ 2 ␤ 1 and FAK activation in platelets adherent to monomeric collagens (44 -46). This does not exclude the operation of the two-step, two-site model (6), because other (lower affinity) collagen receptors may come into play only after platelet adhesion via ␣ 2 ␤ 1 . Secondary binding of sequences within collagen, such as that of the GPO motif to platelet GpVI, is increasingly viewed as an obligatory activatory event (10,22,23).
The platelet fibrinogen receptor, integrin ␣ IIb ␤ 3 , is required for FAK activation in platelets stimulated with thrombin (46), although some stimuli, such as cross-linking of Fc␥RIIA, may activate FAK or other focal adhesion-associated proteins without involving ␣ IIb ␤ 3 (47,48). FAK phosphorylation in platelets has been dissociated from both ␣ IIb ␤ 3 occupancy and focal adhesion formation centering on ␣ IIb ␤ 3 in the absence of fibrinogen binding (49,50).
We considered that FAK tyrosine phosphorylation in platelets might be an early event after occupancy of ␣ 2 ␤ 1 by collagen fibers, and that we could determine the relative importance of ␣ 2 ␤ 1 and GpVI in this process by comparing the capacity of collagen, CRP-XL, and GFOGER-GPP-XL to elicit FAK tyrosine phosphorylation. Synthetic triple-helical peptides have not hitherto been examined in this context. We have applied these ligands to human platelets, immunoprecipitated FAK with specific anti-FAK antibodies, and determined the tyrosine phosphorylation state of the enzyme as an index of its activity.
Furthermore, we have examined the role of intracellular Ca 2ϩ and PKC in platelet FAK activation using the Ca 2ϩ ionophore, ionomycin, to increase the internal platelet Ca 2ϩ concentration and the Ca 2ϩ chelator, BAPTA, to buffer platelet cytosolic Ca 2ϩ ; PKC was either directly activated using the phorbol ester, TPA, or specifically inhibited using Ro31-8220.
Platelet Preparation-Platelet concentrates, less than 24 h-old, pooled from four donors, were obtained from the National Blood Service, Long Road, Cambridge, UK, centrifuged at 250 ϫ g for 15 min to remove red blood cells, leaving platelet-rich plasma, from which the platelets were centrifuged at 700 ϫ g for 15 min. The platelet pellet was resuspended in loading buffer (LB; 145 mM NaCl, 5 mM KCl, 10 mM glucose, 1 mM MgSO 4 , 0.5 mM EGTA, 10 mM HEPES, pH 7.36). The platelets were pelleted at 700 ϫ g for 10 min and resuspended in LB at 10 9 /ml for immunoprecipitation and at 5 ϫ 10 8 /ml for other work. Aspirin (100 M) and apyrase (0.25 units/ml) were used where indicated.
Immunoprecipitation-Platelet-agonist suspensions (500 l) were mixed with an equal volume of 2 ϫ radioimmune precipitation buffer (2% Triton X-100, 2% sodium deoxycholate, 0.2% SDS (each w/v), 316 mM NaCl, 2 mM EGTA, 20 mM Tris/HCl, pH 7.2, with 10 mg/ml leupeptin, 10 mM benzamidine, 2 mM phenylmethylsulfonyl fluoride, and 2 mM Na 3 VO 4 ), and incubated on ice for 30 min before centrifugation (13,000 ϫ g) for 5 min at 4°C. Pansorbin (60 l/ml lysate) was added to each sample tube and rotated at 4°C for 60 min. Before use, stock Pansorbin was centrifuged (13,000 ϫ g) for 1 min at 4°C, resuspended in the original volume of 1 ϫ radioimmune precipitation buffer, and allowed to stand at room temperature for 15 min. It was then centrifuged again and resuspended in 1ϫ radioimmune precipitation buffer containing bovine serum albumin (1% w/v). Samples were centrifuged (13,000 ϫ g) for 1 min at 4°C, the supernatant was removed, and to it anti-FAK polyclonal antibody (Santa Cruz) was added at 1 g per ml of platelet lysate, and the sample was rotated for 20 h at 4°C. Pansorbin (60 l) was then added to each sample, and the samples were mixed and rotated at 4°C for 60 min before being centrifuged (13,000 ϫ g) for 1 min at 4°C. Pellets were washed three times with 800 l of ice-cold 1ϫ radioimmune precipitation buffer before being resuspended in 80 l of 1ϫ SDS sample buffer (10% glycerol, 0.002% bromphenol blue, 2% SDS, each w/v, 70 mM Tris/HCl, pH 7.2, with 1% 2-mercaptoethanol, v/v) and boiled for 5 min. Samples were divided in 2ϫ 40 l, and proteins were separated by 8% SDS-polyacrylamide gel electrophoresis then blotted to nitrocellulose (2 h at 1 mA/cm 2 , Hoefer TE77 semi-dry blotter). Uniform protein transfer was verified by Ponceau S staining. One blot was incubated with 4G10 (1:2500) and washed with TBST (20 mM Tris/HCl, 136 mM NaCl, 0.1% (w/v) Tween 20, pH 7.6), and anti-phosphotyrosine was detected using horseradish peroxidase-linked anti-mouse antibody (1:10,000) and enhanced chemiluminescence (1.24 mM luminol, 1.63 mM 4-iodophenol, 2.71 mM H 2 O 2 ). Phosphorylation was quantitated densitometrically using a Leica Q500 image analyzer (51) and is expressed as a percentage change relative to control values. The other blot was probed with monoclonal anti-FAK (Affiniti, 1:1000), to verify uniform recovery of FAK. Each experiment was performed using a different platelet preparation on three separate occasions.
Platelet Aggregation-Platelets were prepared as for immunoprecipitation and resuspended to 10 9 /ml in LB. 150 l of suspension was stirred (1100 rpm) in an aggregometer at 30°C as described (19), and inhibitors or solvent were added and followed 5 min later by ligand in a volume of 3 l as indicated.
To verify the inhibitory properties of the anti-␤ 1 antibody, 2A4, washed platelets were prepared from whole blood (19), and preincubated in the aggregometer as above for 1 min with 2A4 (20 g/ml), before addition of just sufficient collagen fibers to cause maximal aggregation.
Protein Kinase C Activity-Platelets were prepared as for immunoprecipitation and resuspended to 10 9 /ml in LB. They were labeled with 32 P i at 100 Ci/ml for 1 h at 30°C, centrifuged to remove excess radiolabel, and resuspended to 10 9 /ml. Samples (20 l) were treated with ligand (5 l), and the reaction was stopped after 2 min using 25 l of Laemmli buffer (51). Proteins were separated on a 10% polyacrylamide gel, and the phosphorylation of a 47-kDa protein band (p47, presumed to be pleckstrin, the major protein kinase C substrate in platelets (52)) was detected by autoradiography.
Intracellular Ca 2ϩ Measurement-Platelet concentrates were centrifuged to remove red cells as above and loaded with 2 M Fura2-AM at room temperature for 45 min. Platelets were pelleted by centrifugation as above and resuspended to 10 8 /ml in LB. The platelet suspensions were transferred to a Spex Fluoromax DM3000CM fluorimeter, and fluorescence excited at 340 and 380 nm was used to calculate the intracellular calcium concentration as described previously (53). Where indicated, BAPTA-AM (20 M) was preincubated after Fura2 loading for 20 min.

RESULTS
Immunoprecipitation of FAK-FAK, immunoprecipitated from platelets activated by collagen at 25 g/ml, showed a time-dependent increase in tyrosine phosphorylation (Fig. 1a). The increase in tyrosine phosphorylation at 60 s was detectable but small; therefore, 5-min incubation, causing a substantial increase in FAK tyrosine phosphorylation, was chosen for subsequent assays. Fig. 1b shows equal recovery of FAK in each sample.
Immunoprecipitation of FAK from platelets activated by CRP-XL at 5 g/ml showed a time-dependent increase in tyrosine phosphorylation of FAK (Fig. 1c). Fig. 1d shows immunoprecipitated FAK from CRP-XL-activated platelets, using the Affiniti monoclonal anti-FAK for immunodetection. Again, equal amounts of FAK were demonstrated in each sample.
Effect of Ligand Concentration- Fig. 2, a and d, shows a concentration-dependent increase in the tyrosine phosphorylation of FAK immunoprecipitated from platelets activated by different concentrations of collagen and CRP-XL. Equal recovery of FAK was demonstrated in all cases (data not shown). Collagen fibers at 25 g/ml and CRP-XL at 5 g/ml caused near-maximal increases in FAK tyrosine phosphorylation. Some experiments (data not shown) were performed in the presence of apyrase, which scavenges ADP secreted by activated platelets, and aspirin, which blocks the conversion of arachidonate to thromboxane A 2 . The inhibitors had no marked effect, indicating that tyrosine phosphorylation of FAK does not depend upon these processes in platelets stimulated by collagen or CRP-XL.
In some experiments the basal level of FAK tyrosine phosphorylation was detectable, whereas others (e.g. Fig. 2e) showed negligible FAK tyrosine phosphorylation. This may reflect variation between donors, or in the activation state of resting platelets between experiments, as well as in the immunodetection procedure. Conclusions throughout this study are therefore based on comparisons made within an experiment, and where possible, within immunoblots rather than between blots.
Role of ␣ 2 ␤ 1 in FAK Tyrosine Phosphorylation by Collagen and CRP-XL-When FAK was immunoprecipitated from platelets preincubated with anti-␣ 2 P1E6 or anti-␤ 1 2A4 for 5 min for the indicated times were lysed, and FAK was precipitated and immunodetected using 4G10 anti-phosphotyrosine (a) or anti-FAK (b) as described. c and d, platelets were treated with CRP-XL (5 g/ml) and handled otherwise as for a and b.

FIG. 2. Activation of FAK by collagen fibers and CRP-XL is
dose-dependent and independent of ␣ 2 ␤ 1 occupancy. Platelets were treated with ligand, as indicated, for 5 min, then FAK tyrosine phosphorylation was determined in immunoprecipitates as described in the legend to Fig. 1. Platelets were treated in parallel experiments with collagen, up to 100 g/ml, in the absence (a) or presence (c) of 50 M Ca 2ϩ in excess of the 500 M EGTA in the buffer. In b, platelets were preincubated with the anti-␣ 2 , P1E6 (2 g/ml), or anti-␤ 1 , 2A4 (20 g/ml) as indicated, for 5 min, then treated with collagen fibers (25 g/ml) for 5 min. In d, platelets were treated with the indicated levels of CRP-XL and were otherwise handled exactly as in a; in e platelets were treated with CRP-XL (5 g/ml) after preincubation with P1E6 or 2A4 as for b.
before activation with collagen for 5 min, there was no diminution, confirmed by densitometry, in the level of tyrosine phosphorylation of FAK induced by either ligand (Fig. 2b). Similar data were obtained using the anti-␤ 1 mAb13 (data not shown) or the anti-␣ 2 , 6F1 (Fig. 5b).
CRP-XL induces platelet activation without involvement of ␣ 2 ␤ 1 and caused substantial tyrosine phosphorylation of FAK. This shows that ligation of GpVI, the receptor for CRP-XL, induces phosphorylation of FAK. As anticipated, the anti-␣ 2 and anti-␤ 1 antibodies had no effect on the tyrosine phosphorylation of FAK by CRP-XL.
Recent work in this laboratory has shown that the affinity of platelet ␣ 2 ␤ 1 is dependent upon the presence of micromolar Ca 2ϩ in the suspending medium (54). For this reason, the experiments shown above for collagen were repeated in the presence of a small excess of Ca 2ϩ over EGTA in the buffer, conditions that support ␣ 2 ␤ 1 -dependent platelet adhesion to immobilized collagens. FAK tyrosine phosphorylation was not enhanced by the presence of Ca 2ϩ compared with the parallel incubation in the absence of Ca 2ϩ (Fig. 2c).
We have recently shown the peptide sequence GFOGER to be a recognition motif in type I collagen for the ␣ 2 ␤ 1 I domain (25). Application of the cross-linked triple-helical peptide, GFOGER-GPP-XL, to platelets at up to 50 g/ml caused no discernible increase in FAK tyrosine phosphorylation (Fig. 3a). In contrast, in this experiment as in Fig. 2a, collagen fibers caused substantial FAK tyrosine phosphorylation. The addition of micromolar Ca 2ϩ to the medium did not support FAK phosphorylation stimulated by even high levels of the peptide (200 g/ml).
Functional Verification of the Anti-␣ 2 ␤ 1 Antibodies-6F1 as used in the present study completely blocked platelet adhesion to monomeric collagen (54,55). Similar experiments showed both P1E6 and mAb13 to be effective inhibitors of platelet adhesion to monomeric collagen. The anti-␣ 2 , P1E6, blocked the capacity of reconstituted type I collagen fibers to induce platelet aggregation (7). We verified here that both P1E6 and the anti-␤ 1 , 2A4, could attenuate the platelet aggregation stimulated by threshold concentrations of native collagen fibers (data not shown). Together, these data confirm the functional activity of the antibodies used here.
Functional Verification of the ␣ IIb ␤ 3 Antagonist GR144053F- Fig. 4 shows that CRP-XL (5 g/ml) or thrombin (1 unit/ml), levels of agonist consistent with the rest of the study, aggregated platelets suspended in medium containing 0.5 mM EGTA, but that collagen fibers (25 g/ml) caused minimal platelet aggregation. Preincubation with the fibrinogen receptor antagonist GR144053F (1 M) reduced the extent of aggregation to Ͻ15% of control values in platelets stimulated by CRP-XL or thrombin. Fig.  5a shows that preincubation of platelets with 1 M GR144053F, a level which causes complete blockade of ␣ IIb ␤ 3 (54), caused a substantial reduction in FAK tyrosine phosphorylation in platelets subsequently stimulated by CRP-XL (77% reduction over four trials) or thrombin (63% over two trials). This effect was of similar order to the inhibition (ϳ85%) of aggregation by GR144053F for CRP-XL or thrombin. In contrast, there was little observable inhibition (15%; five trials) of the action of collagen by GR144053F, even when used in conjunction with ␣ 2 -blockade by 6F1 (Fig. 5b). Basal phosphorylation of FAK was also inhibited to some extent (30%; four trials), perhaps indicating a degree of activation of platelets under resting conditions, consistent with the suggestion, above, that the basal platelet preparations might to some extent be activated. This effect of GR144053F was minor compared with the marked inhibition of the action of CRP-XL or thrombin.

FIG. 4. Platelet aggregation, which occurs in the presence of EGTA with CRP-XL and thrombin, is sensitive to GR144053F.
Platelets were prepared as for immunoprecipitation studies and suspended in LB at 1 ϫ 10 9 per ml. They were preincubated as indicated with GR144053F (GR; 1 M) for 5 min, stirring at 1100 rpm in the aggregometer, then ligands were added to elicit aggregation. a, thrombin (Thr) 1 units/ml was added; b, CRP-XL (CRP) was added at 5 g/ml; c, collagen fibers (Col) were added at 25 g/ml, at times indicated by arrows.

FIG. 3. GFOGER-GPP-XL does not elicit FAK phosphorylation in the presence or absence of Ca 2؉ .
Platelets were treated with the indicated levels of the ␣ 2 ␤ 1 -specific peptide, GFOGER-GPP-XL, for 5 min in the absence (a) or presence (b) as indicated, of an excess of Ca 2ϩ as in Fig. 2c. FAK phosphorylation was determined as for Fig. 1. The lane marked col represents a control using collagen fibers at 25 g/ml.

Role of Protein Kinase C in FAK Tyrosine
Phosphorylation-To investigate signaling pathways required for FAK tyrosine phosphorylation, we examined the role of protein kinase C. FAK was immunoprecipitated from platelets stimulated for 5 min with collagen (25 g/ml), CRP-XL (5 g/ml), or TPA (400 nM). As before, marked tyrosine phosphorylation of FAK was induced by collagen and CRP-XL, whereas control levels were undetectable (Fig. 6a). TPA caused a very minor increase in FAK tyrosine phosphorylation: densitometry showed that CRP-XL and collagen were each about 20 times more effective than TPA.
Effect of Ro31-8220 or BAPTA on Tyrosine Phosphorylation of FAK Stimulated by Collagen, CRP-XL, or Ionomycin- Fig.  6b shows complete inhibition of tyrosine phosphorylation of FAK immunoprecipitated from platelets after pretreatment with the PKC inhibitor Ro31-8220 (5 M) prior to activation by collagen (25 g/ml) or CRP-XL (5 g/ml) for 5 min. We have shown 5 M Ro31-8220 to cause complete inhibition of PKC, measured as p47 phosphorylation (56). The calcium ionophore, ionomycin, also caused substantial tyrosine phosphorylation of FAK (Fig. 6c), suggesting a role for calcium signaling in FAK activation, and again, as for collagen and CRP-XL, this action was substantially attenuated by Ro31-8220.
Preincubation of platelets with the Ca 2ϩ -chelating agent, BAPTA-AM, to buffer rises in intracellular Ca 2ϩ , markedly attenuated the ability of both CRP-XL and collagen fibers to stimulate tyrosine phosphorylation of FAK (Fig. 6d). For comparison, in Western blots prepared from whole platelet lysates there was an inhibition of overall tyrosine phosphorylation stimulated by collagen, CRP-XL and in the control samples of 12, 21, and 7%, respectively (Fig. 6e), when platelets were preincubated with BAPTA-AM. This inhibition indicated that the effects on FAK are highly specific. In contrast, one band of about 38 kDa increased in intensity significantly after BAPTA pretreatment in both collagen-and CRP-stimulated platelets. Note that the effects of BAPTA on the 120-kDa region are minor for collagen, although much more apparent for CRP, suggesting that other bands insensitive to BAPTA comigrate with FAK.
Effect of Ionomycin on PKC Activity- Fig. 7a shows that ionomycin, from 500 to 2000 nM, was an effective activator of PKC, determined from the phosphorylation of p47. Higher ionomycin levels caused no further increase in phosphorylation of p47 (data not shown).
Effect of BAPTA-AM on PKC Activity- Fig. 7b shows the effect of BAPTA-AM (20 M) on platelets activated by collagen (25 g/ml) or CRP-XL (5 g/ml). With or without BAPTA loading, both effectors caused marked activation of PKC, indicated by p47 phosphorylation. The action of TPA (not shown) or CRP-XL, was largely insensitive to the presence of BAPTA; only the action of collagen was noticeably attenuated, but PKC FIG. 5. FAK phosphorylation, stimulated by CRP-XL and thrombin, but not collagen, is sensitive to GR144053F. Platelets were preincubated with GR144053F (1 M) for 5 min. Ligands were then added for a further 5 min, and FAK was immunoprecipitated as described. a, platelets were stimulated with CRP-XL (5 g/ml) or thrombin (1 unit/ml) as indicated (bas represents basal controls). b, platelets were stimulated with collagen fibers (25 g/ml) as indicated. The presence of GR144053F or of the anti-␣ 2 monoclonal antibody, 6F1 at 2 g/ml in the preincubation is denoted by ϩ beneath the relevant lanes.

FIG. 6. A role for PKC activity and [Ca 2؉
] i in the tyrosine phosphorylation of FAK. Tyrosine phosphorylation was determined in immunoprecipitates as described, from (a) platelets stimulated with collagen (25 g/ml), CRP-XL (5 g/ml), or TPA (400 nM) for 5 min, as indicated. For b, platelets were preincubated for 10 min with Ro31-8220 (5 M), denoted by ϩ, then treated with collagen (25 g/ml) or CRP-XL (5 g/ml) for 5 min. For c FAK tyrosine phosphorylation was determined in platelets treated with 1 M ionomycin (Io) with or without Ro31-8220 preincubation as above. For d, platelets were preincubated for 20 min with BAPTA-AM (20 M), then stimulated with collagen or CRP-XL as indicated. e, a Western blot of whole platelet lysates, treated with CRP or collagen after preincubation with BAPTA as indicated, then probed for phosphotyrosine. The position where FAK is expected to run in this blot is indicated. activity persisted at greater than 50% despite BAPTA loading. This suggests the presence of Ca 2ϩ -sensitive and -insensitive PKC isoforms activated by collagen receptors in human platelets.
Effect of Ro31-8220 on [Ca 2ϩ ] i - Fig. 8 shows time courses for the rise in [Ca 2ϩ ] i evoked by collagen (a) or ionomycin (b), with and without pre-incubation of platelets with the PKC inhibitor, Ro31-8220. Inhibition of PKC caused a marked increase in both the peak amplitude and duration of Ca 2ϩ signals observed under these conditions. Parallel measurement of [Ca 2ϩ ] i using Fura2 showed that calcium signaling was abolished by BAPTA loading (data not shown). DISCUSSION Our aim in this study was to explore the capacity of the platelet integrin ␣ 2 ␤ 1 to activate FAK, comparing the efficacy of the synthetic analogue of collagen, CRP-XL, with that of native type I collagen fibers and with the ␣ 2 ␤ 1 -specific peptide, GFOGER-GPP-XL. Thus, we intended to determine whether ␣ 2 ␤ 1 acts as a signaling receptor for collagen in platelets, working from the premise that FAK phosphorylation is an event likely to be integrin-dependent in platelets as well as in other cells.
The first part of the present work addresses the role of the collagen receptor ␣ 2 ␤ 1 in the regulation of FAK. Both collagen and CRP-XL activate FAK, as indicated by its tyrosine phosphorylation state, in a concentration-and time-dependent manner. The failure of antibodies against the ␣ 2 and ␤ 1 inte-grin subunits, which prevent adhesion to collagen (validated as described under "Results") to block FAK activation demonstrates that ␣ 2 ␤ 1 occupancy by collagen fibers does not regulate FAK. This result contrasts with the proposed general role of ␤ 1 integrins in FAK activation (43). These experiments were performed in the presence of micromolar Ca 2ϩ , conditions where the integrin is known to be competent to bind collagen (54). The potency of CRP-XL, which does not bind ␣ 2 ␤ 1 , in stimulating tyrosine phosphorylation of FAK suggests that another collagen receptor, GpVI, initiates FAK activation in platelets.
Recently, we have identified the sequence GFOGER within collagen type I, which binds to the I domain of the integrin ␣ 2 subunit (25). This peptide sequence, in triple-helical conformation, binds platelets in ␣ 2 ␤ 1 -dependent manner and supports purified ␣ 2 ␤ 1 binding. By co-crystallization with the recombinant ␣ 2 I domain, we have shown that the E residue of the peptide coordinates the divalent cation in the metal ion-dependent adhesion site of the integrin ␣ 2 subunit (26). This indicates that the peptide properly replicates the ␣ 2 ␤ 1 -binding properties of collagen. However, even at levels up to 200 g/ml with or without micromolar Ca 2ϩ (Fig. 3b), GFOGER-GPP-XL caused no discernible increase in tyrosine phosphorylation of platelet FAK. This shows that neither ␣ 2 ␤ 1 occupancy nor clustering by the cross-linked peptide is sufficient to activate FAK in platelets in suspension.
The fibrinogen receptor ␣ IIb ␤ 3 has attracted most attention as a means of regulating FAK activity in platelets. Collagen   FIG. 7. Ionomycin, collagen, and CRP-XL stimulate PKC activity. a, platelets were labeled with 32 P i as described then treated with increasing levels of ionomycin, as indicated, and an autoradiograph was prepared as under "Experimental Procedures." The position of p47 is indicated on the left. For b and c, platelets were prepared as above, then preincubated with 20 M BAPTA-AM for 20 min, as indicated, before stimulating with collagen (25 g/ml), CRP-XL (5 g/ ml), ionomycin (Io; 1 M), or TPA (200 nM) as shown. Autoradiographs were prepared as above. and CRP-XL were added to unstirred suspensions of platelets in the presence of EGTA, conditions where ␣ IIb ␤ 3 is not competent and aggregation is not anticipated (57); therefore, we did not expect ␣ IIb ␤ 3 to regulate FAK in these experiments. To verify this, we added collagen fibers to platelets stirred in an aggregometer, causing, as anticipated, no significant aggregation. However, despite the presence of EGTA, both CRP-XL and thrombin under similar conditions caused some aggregation, which was highly sensitive to the ␣ IIb ␤ 3 antagonist, GR144053F (Fig. 4). This indicates that GR144053F as used here is a good antagonist of ␣ IIb ␤ 3 occupancy.
Partial inhibition of FAK phosphorylation by GR144053F in platelets treated with either CRP-XL or thrombin showed that ␣ IIb ␤ 3 activation is important in the regulation of FAK, as has been shown previously for thrombin (44). But GR144053F had little effect on the activation of FAK by collagen, consistent with collagen's failure to cause much aggregation under these conditions. (It should be noted that aggregation is more likely to occur during stirring in the aggregometer than in all other components of the study, where platelets were not stirred for more than a second after the addition of ligand.) The inclusion of both GR144053F and 6F1 (Fig. 5b), to provide simultaneous blockade of ␣ 2 ␤ 1 and ␣ IIb ␤ 3 , had little effect on FAK tyrosine phosphorylation stimulated by collagen fibers, which is therefore shown to proceed in platelets without the involvement of either integrin under these conditions. Recently, the use of mutant ␣ IIb ␤ 3 showed that FAK activation could be dissociated from ␣ IIb ␤ 3 occupancy (49), as we propose here for the regulation of FAK by collagen fibers. Integrin-independent activation of FAK has also been reported in platelets activated using immobilized human IgG (48), an event that depends instead on Fc␥RIIA.
The identity of the collagen receptors responsible for FAK activation remains to be resolved. Our experiments demonstrate that GpVI occupancy alone, resulting from treating platelets with CRP-XL, is not sufficient to elicit full tyrosine phosphorylation of FAK that is independent of ␣ IIb ␤ 3 . In this respect, CRP-XL shows some similarity to thrombin. Using specific antibodies to cross-link CD36, a candidate receptor along with GpVI, others have discounted CD36 as a regulator of FAK (58), although this technique provides clustering only of CD36 populations rather than of CD36 with other receptors, as we expect will occur with the native collagen fibers used here. Investigation of whether GpVI acts as a co-receptor in regulating FAK in platelets stimulated with collagen fibers, and the possible role of CD36 in these events, must await the development of receptor-specific antagonists.
We sought to identify intracellular signaling events that are involved in the regulation of FAK activity during platelet activation. PKC has been implicated in FAK activation in both platelets (48,59) and other cells (60,61) that were adherent to non-collagenous substrates. We found that TPA stimulated only a slight increase in tyrosine phosphorylation of FAK in platelet suspensions. However, pretreatment of platelets with the PKC inhibitor, Ro31-8220, virtually abolished tyrosine phosphorylation of FAK caused by collagen or CRP-XL. This suggests that, although direct stimulation of PKC itself is insufficient to cause major stimulation of FAK, PKC is an important mediator of the tyrosine phosphorylation of FAK stimulated by either collagen or CRP-XL.
Next, we showed that [Ca 2ϩ ] i is also important in the control of FAK tyrosine phosphorylation, by using ionomycin to elicit Ca 2ϩ mobilization, and BAPTA-AM loading to buffer [Ca 2ϩ ] i . Ionomycin stimulated tyrosine phosphorylation of FAK, whereas BAPTA-AM completely abolished tyrosine phosphorylation of FAK in platelets stimulated with collagen or CRP-XL, confirming a role for Ca 2ϩ . These results contrast with the work of Haimovich et al. (48) who showed that, in IgG-adherent platelets, exposure to BAPTA-AM caused, if anything, increased FAK phosphorylation. The same group reported no effect of BAPTA-AM on FAK phosphorylation in platelets adherent to fibrinogen (59), but in the same paper, they show that BAPTA-AM abolishes the action of thrombin in stimulating FAK in fibrinogen-adherent platelets. A requirement for increased [Ca 2ϩ ] i in the regulation of FAK was similarly proposed for epinephrine-stimulated platelet suspensions (59). Possibly, the role of Ca 2ϩ in regulating FAK activity may be ligand-specific.
Ionomycin treatment also activated PKC. To resolve the roles of [Ca 2ϩ ] i and PKC, platelets were first preincubated with Ro31-8220 to inactivate PKC and then treated with ionomycin. FAK tyrosine phosphorylation was virtually abolished, as in platelets stimulated with collagen or CRP-XL after PKC blockade. It is important to note that Ro31-8220 enhanced the increase in [Ca 2ϩ ] i stimulated by either collagen or ionomycin, very likely as a consequence of inhibiting PKC-dependent Ca 2ϩ ATPases, which export Ca 2ϩ from the cytosol. Reciprocal experiments showed that, although BAPTA blocks FAK phosphorylation, it had little effect on PKC activity. Hence, neither FIG. 8. Ro31-8220 enhances Ca 2؉ signals elicited by collagen and ionomycin. Platelets were loaded with Fura2-AM, as described, then stirred in the cuvette of a fluorimeter, after preincubation with Ro31-8220 for 10 min, where indicated. Collagen (a, 90 g/ml) or ionomycin (b, 1 M) were added as indicated by the arrows. Intracellular calcium levels were calculated as described under "Experimental Procedures." elevated [Ca 2ϩ ] i or increased PKC activity is sufficient to support FAK phosphorylation, but each is necessary for FAK activation by collagen, CRP-XL, or ionomycin. Such a role for Ca 2ϩ has been proposed for endothelial cell FAK activation consequent to spreading on type IV collagen (62).
In conclusion, our data suggest that the regulation of platelet FAK by native collagen fibers is independent of integrins, occurring despite blockade of ␣ 2 ␤ 1 or ␣ IIb ␤ 3 or both. The activation of phospholipase C␥2 via GpVI (29,63) causes Ca 2ϩ and PKC signals essential for the regulation of FAK. Yet these signals together, activated by either CRP-XL or thrombin, are not sufficient to elicit full FAK phosphorylation without ␣ IIb ␤ 3 occupancy. Coordination of signals from ␣ IIb ␤ 3 and other receptors have been proposed to regulate FAK (34). Our data suggest that collagen, perhaps because it is recognized by different platelet receptor populations in addition to GpVI and ␣ 2 ␤ 1 , is able to bypass the requirement for integrins. The role and identity of these co-receptors for collagen remain to be elucidated.