Directly Activating the Integrin αIIbβ3 Initiates Outside-In Signaling by Causing αIIbβ3 Clustering*

αIIbβ3 activation in platelets is followed by activation of the tyrosine kinase c-Src associated with the carboxyl terminus of the β3 cytosolic tail. Exogenous peptides designed to interact with the αIIb transmembrane (TM) domain activate single αIIbβ3 molecules in platelets by binding to the αIIb TM domain and causing separation of the αIIbβ3 TM domain heterodimer. Here we asked whether directly activating single αIIbβ3 molecules in platelets using the designed peptide anti-αIIb TM also initiates αIIbβ3-mediated outside-in signaling by causing activation of β3-associated c-Src. Anti-αIIb TM caused activation of β3-associated c-Src and the kinase Syk, but not the kinase FAK, under conditions that precluded extracellular ligand binding to αIIbβ3. c-Src and Syk are activated by trans-autophosphorylation, suggesting that activation of individual αIIbβ3 molecules can initiate αIIbβ3 clustering in the absence of ligand binding. Consistent with this possibility, incubating platelets with anti-αIIb TM resulted in the redistribution of αIIbβ3 from a homogenous ring located at the periphery of discoid platelets into nodular densities consistent with clustered αIIbβ3. Thus, these studies indicate that not only is resting αIIbβ3 poised to undergo a conformational change that exposes its ligand-binding site, but it is poised to rapidly assemble into intracellular signal-generating complexes as well.

␣IIb␤3 activation in platelets is followed by activation of the tyrosine kinase c-Src associated with the carboxyl terminus of the ␤3 cytosolic tail. Exogenous peptides designed to interact with the ␣IIb transmembrane (TM) domain activate single ␣IIb␤3 molecules in platelets by binding to the ␣IIb TM domain and causing separation of the ␣IIb␤3 TM domain heterodimer. Here we asked whether directly activating single ␣IIb␤3 molecules in platelets using the designed peptide anti-␣IIb TM also initiates ␣IIb␤3-mediated outside-in signaling by causing activation of ␤3-associated c-Src. Anti-␣IIb TM caused activation of ␤3-associated c-Src and the kinase Syk, but not the kinase FAK, under conditions that precluded extracellular ligand binding to ␣IIb␤3. c-Src and Syk are activated by trans-autophosphorylation, suggesting that activation of individual ␣IIb␤3 molecules can initiate ␣IIb␤3 clustering in the absence of ligand binding. Consistent with this possibility, incubating platelets with anti-␣IIb TM resulted in the redistribution of ␣IIb␤3 from a homogenous ring located at the periphery of discoid platelets into nodular densities consistent with clustered ␣IIb␤3. Thus, these studies indicate that not only is resting ␣IIb␤3 poised to undergo a conformational change that exposes its ligand-binding site, but it is poised to rapidly assemble into intracellular signal-generating complexes as well.
␣IIb␤3 activation in platelets is followed by ␣IIb␤3 clustering (1) and ␣IIb␤3-mediated "outside-in" signal transduction (2) that is initiated by activation of the tyrosine kinase c-Src associated with the carboxyl terminus of the ␤3 cytosolic tail (CT) 3 (3)(4)(5). Activated c-Src then initiates an intracellular signaling cascade culminating in the reorganization of the platelet cyto-skeleton, platelet spreading, and fibrin clot retraction, events important for efficient hemostasis in the hemodynamic environment of flowing blood (6).
Recently, we reported studies of the interaction of c-Src with the ␤3 CT (5). We detected little to no interaction in unstimulated platelets, but following platelet stimulation with thrombin, c-Src associated with ␤3 in a time-dependent manner and underwent transient activation. We also found that the ␤3 CT binds to the c-Src SH3 domain at a site that is occupied by the linker connecting the c-Src SH2 and kinase domains in inactive c-Src. Following platelet activation, c-Src is "unlatched," allowing ␤3 CT binding to the SH3 domain when the linker binding site is vacated. However, the full catalytic activity of c-Src requires phosphorylation of Tyr 419 (numbered according to the UniProtKB/Swiss-Prot entry P12931 for human c-Src), located in a loop between the two lobes of the c-Src kinase domain (7). In platelets, this occurs by trans-autophosphorylation when ␣IIb␤3 clustering brings ␤3 CT-bound c-Src into proximity (3). Thus, ␣IIb␤3 clustering is an essential step in the outside-in signaling that follows ␣IIb␤3 activation.
Transmembrane (TM) helix-helix interactions play an important role in maintaining ␣IIb␤3 in its inactive state, and disrupting these interactions is sufficient to cause ␣IIb␤3 activation (8,9). Membrane-soluble peptides corresponding to the ␣IIb TM domain (10) or designed to interact strongly with the ␣IIb TM domain (11) cause ␣IIb␤3 activation, as well as platelet aggregation, by directly disrupting the TM domain interactions, mimicking the effect of the physiologic ␣IIb␤3 activator talin ( Fig. 1) (12). Moreover, these peptides activate ␣IIb␤3 in platelets in the absence of inside-out signaling. Thus, they provide a way to test whether directly activating individual ␣IIb␤3 molecules can also cause ␣IIb␤3 clustering. Here we used the computationally designed peptide anti-␣IIb TM (11) to directly activate single ␣IIb␤3 molecules in platelets and used the transautoactivation of ␤3-associated c-Src as a probe for ␣IIb␤3 clustering. We found that anti-␣IIb TM caused ␣IIb␤3-dependent c-Src activation under conditions that preclude ligand binding to ␣IIb␤3 and platelet actin polymerization. These results suggest that not only is resting ␣IIb␤3 poised to undergo a conformational change that exposes its ligand binding site, but it is poised to rapidly oligomerize into intracellular signalgenerating complexes as well.

Experimental Procedures
Peptide Synthesis-The peptides anti-␣IIb TM (11) (KKAY-VMLLPFFIGLLLGLIFGGAFWGPARHLKK) and MS1 (13) (BQLLIAVLLLIAVNLILLIAVARLRYLVG, where B represents ␤-alanine) were synthesized with a Discover microwave peptide synthesizer (CEM, Matthews, NC) using the standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) protection strategy and on a Rink Amide AM resin (200 -400 mesh) (Nova Biochem) with a substitution level of 0.71 mmol/g. Five equivalents of HBTU coupling reagent (Pepnet) and five equivalents of amino acid (ChemPep) were used per coupling. Each coupling was performed in duplicate with a coupling time of 2 min. After cleavage and workup, peptides with an amidized C terminus and a free amino N terminus were characterized using a Perseptive Biosystems MALDI-TOF mass spectrometer. Purification of peptides was carried out on a Waters 600E HPLC equipped with a SepaxGP-C8 reverse-phase, 21.2 ϫ 250 mm column over a 50:50 to 0:100 (water, 0.1% TFA:acetonitrile, 0.1% TFA) gradient for 30 min. Fractions were characterized with MALDI mass spectrometry and lyophilized to dryness. The purity of each product was verified by HPLC.
Measurements of Platelet Function-Human platelet studies were approved by the Institutional Review Board of the University of Pennsylvania Office of Regulatory Affairs. Platelet aggregation and ATP secretion were measured using a Chrono-Log model 700 lumi-aggregometer (10). Human platelets, obtained from blood anticoagulated with sodium citrate (65 mM), citric acid (77 mM), and glucose (95 mM) (pH 4.4) were washed with 10 mM HEPES buffer (pH 6.5) containing 150 mM NaCl, 3 mM EDTA, 1 M PGE1, and 0.3 units/ml apyrase and resuspended in modified Tyrode's buffer (20 mM HEPES (pH 7.35) containing 135 mM NaCl, 2.7 mM KCl, 3 mM NaH 2 PO 4 , 5 mM glucose, and 0.1% BSA). 450-l aliquots of the platelet suspension supplemented with 200 g/ml human fibrinogen and 1 mM CaCl 2 were stirred at 1200 rpm in siliconized aggregometer cuvettes. Secreted ATP secretion was measured using 50 l of CHRONO-LUME TM reagent (firefly luciferin-luciferase) and analyzed using AGGRO/LINK8 software. Membrane-soluble peptides were dissolved in DMSO solution, and final DMSO concentrations were no greater than 0.5%.
FITC-conjugated fibrinogen binding to washed platelets was measured as described previously (14). Briefly, 500-l aliquots of the washed platelets were stimulated in the presence of 200 g/ml FITC-conjugated fibrinogen. The platelets were then fixed with 0.37% formalin in PBS buffer for 10 min, washed, and examined by FACS.
Immunostaining of Platelet ␣IIb␤3-To visualize ␣IIb␤3 on the surface of resting platelets and platelets incubated with anti-␣IIb TM, washed human platelets were resuspended at a concentration of 2.5 ϫ 10 8 platelets/ml in Tyrode's buffer containing 5 mM EDTA but without BSA. 500-l aliquots of the platelet suspension were then transferred to BSA-blocked centrifuge tubes and incubated for 1 min at 37°C in the presence or absence of 2 M anti-␣IIb TM (8). Following the incubation, the platelets were fixed by adding 500 l of 4% paraformaldehyde for 15 min at room temperature and transferred to poly-lysinecoated chamber glass slides. The platelet-containing chambers were then incubated sequentially with 10% goat serum for 1 h at room temperature and with the anti-␤3 mAb SSA6 labeled with Alexa Fluor 488 (Thermo Fisher Scientific) for 1 h at 37°C.  (52). Anti-␣IIb TM disrupts the TM domain heterodimer by binding to the face of the ␣IIb TM helix that would otherwise interact with the ␤3 TM helix (11,53).
After washing with PBS, the stained platelets were mounted with Fluoromount G (Southern Biotech) and visualized using a Zeiss Axioplan upright microscope with a Zeiss FluoArc mercury lamp. Images were captured using a Zeiss AxioCam HRm high-resolution monochrome charge-coupled device camera and AxioVision 4.5 software (Zeiss).

Role of c-Src in Anti-␣IIb TM-induced Platelet Aggregation
and Secretion-To address whether c-Src is involved in platelet function stimulated by anti-␣IIb TM, we compared the effects of the peptide to the effects of low and high concentrations of thrombin on the function of normal platelets in the absence and presence of the c-Src inhibitor PP2. As shown in Fig. 2A, 2 M anti-␣IIb TM caused both platelet aggregation and ATP secretion. However, 10 M PP2 attenuated anti-␣IIb TM-induced platelet aggregation and abrogated peptide-induced platelet ATP secretion. Likewise, both platelet aggregation and ATP secretion stimulated by 0.1 units/ml thrombin were inhibited by 10 M PP2. By contrast, when platelets were stimulated by thrombin at 5 units/ml, 10 M PP2 had no effect on ATP secretion, although platelet aggregation was modestly attenuated. Thus, as noted previously by Liu et al. (17), when platelets were stimulated by ␥-thrombin , c-Src activation appears to potentiate the effects of weaker platelet stimuli such as anti-␣IIb TM and 0.1 units/ml thrombin, but increasing the strength of the platelet stimulus, in this case increasing the thrombin concentration to 5 units/ml, largely bypasses the c-Src requirement.
Next we asked whether the effects of anti-␣IIb TM on platelets require the presence of ␣IIb␤3 by measuring ATP secretion from the platelets of a well characterized patient with Glanzmann thrombasthenia (GT) whose platelets lack ␣IIb␤3 (18). Although 0.1 units/ml thrombin caused brisk ATP secretion from the GT platelets, an anti-␣IIb TM concentration that caused robust ATP secretion from normal platelets, had no effect on GT platelets (Fig. 2B). On the other hand, c-Src in both FIGURE 2. Anti-␣IIb TM causes platelet aggregation and c-Src-dependent ATP secretion by causing ␣IIb␤3 activation. A, simultaneous platelet aggregation and ATP secretion stimulated by 2 M anti-␣IIb TM, 0.1 unit/ml thrombin, or 5 units/ml thrombin was measured in the presence or absence of the Src kinase inhibitor 10 M PP2. These experiments were repeated four times using platelets from four different blood donors with identical results. B, comparison of anti-␣IIb TM-and thrombin-induced ATP secretion from the platelets of a well characterized patient with GT. These experiments were repeated using platelets from another patient with GT with identical results. C, thrombin (Tb) causes c-Src phosphorylation in GT and WT platelets. GT and WT platelets were stimulated with 1 unit/ml thrombin in the presence or absence of PP2. c-Src was then immunoprecipitated from platelet lysates and immunoblotted for c-Src pTyr 419 . MW, molecular weight. D, platelet aggregation and ATP secretion following platelet incubation with the membrane-soluble peptide MS1. These experiments were performed five times with platelets from different blood donors with similar results. E, anti-␣IIb TM causes fibrinogen binding to ␣IIb␤3 on human platelets. Human platelets were incubated with 2 M anti-␣IIb TM for 1 min in the presence of 200 g/ml FITC-labeled human fibrinogen, after which fibrinogen binding to ␣IIb␤3 was measured by fluorescence-activated cell sorting. The specificity of fibrinogen binding to ␣IIb␤3 was verified using either 5 mM EDTA or 10 M eptifibatide. The contribution of secreted ADP or activated c-Src to fibrinogen binding was tested by adding either 10 units/ml apyrase or 10 M PP2 to the incubation. These measurements were repeated using platelets from a different platelet donor with identical results.
wild-type control and GT platelets undergoes phosphorylation when the platelets are stimulated by thrombin, confirming that c-Src in GT platelets can be activated by a stimulus that is independent of ␣IIb␤3 (Fig. 2C). Studies using platelets from another well characterized patient with GT (19) produced identical results (data not shown). Nonetheless, it is still possible that the functional effects of anti-␣IIb TM are due to its membrane solubility. To test this possibility, we studied the effects of an unrelated membrane-soluble peptide on platelet aggregation and secretion. The membrane-soluble peptide MS1 was designed using the two-stranded coiled-coil peptide GCN4-P1 as a template and is insoluble in water but readily solubilized in a variety of detergents (13) and inserts into lipid bilayers (20). As seen in Fig. 2D, incubating platelets with 2 M MS1 caused neither platelet aggregation nor ATP release. Thus, these experiments indicate that the ability of anti-␣IIb TM to initiate platelet function requires the presence of ␣IIb␤3.
To test this conclusion and to address whether anti-␣IIb TM-induced ␣IIb␤3 activation requires c-Src kinase activity, we measured anti-␣IIb TM-induced fibrinogen binding to platelet ␣IIb␤3 in the absence or presence of PP2. As shown by the flow cytometry histograms in Fig. 2E, 2 M anti-␣IIb TM caused FITC-fibrinogen binding to washed human platelets. The platelet-associated fibrinogen was bound to ␣IIb␤3 because binding was inhibited by the ␣IIb␤3 antagonist eptifibatide (21) and prevented by the divalent cation chelator EDTA (10,11). On the other hand, fibrinogen binding was only slightly decreased by the ADP-metabolizing enzyme apyrase, indicating that anti-␣IIb TM-stimulated ␣IIb␤3 activation is not mediated by secreted platelet ADP (10,11). The small attenuation is likely due to the lack of signal amplification caused by secondary stimulation from secreted ADP. Furthermore, fibrinogen binding was unaffected by 10 M PP2, indicating that it did not require c-Src kinase activity. Thus, these results indicate that anti-␣IIb TM initiates platelet activation by binding and altering the conformation of ␣IIb␤3 and that this effect on ␣IIb␤3 is independent of platelet secretion and c-Src activity. Nonetheless, the attenuated platelet aggregation and lack of ATP secretion we observed in the presence of PP2 demonstrates that c-Src-stimulated platelet ADP secretion potentiates the platelet aggregation induced by the peptide.
Anti-␣IIb TM Causes the Activation of ␤3-bound c-Src in Platelets-Platelet agonists like thrombin cause the rapid and transient activation of ␤3-bound c-Src (5). To determine whether anti-␣IIb TM has the same effect, we measured the anti-␣IIb TM-stimulated interaction of c-Src with the ␤3 CT. Although we found no detectable c-Src associated with ␤3 in unstimulated platelets, c-Src binding to ␤3 was detected within 10 s after platelet exposure to anti-␣IIb TM (Fig. 3). Similar to platelets stimulated by thrombin (5), the amount of ␤3-bound c-Src increased progressively as the duration of exposure to anti-␣IIb TM increased, reaching a maximum at 40 s, after which it declined. Phosphorylation of the ␤3-bound c-Src was first detected at 10 s. Like c-Src binding, it increased progressively, reaching a maximum at 40 -60 s, after which it also declined (Fig. 3). Thus, like conventional platelet agonists, anti-␣IIb TM causes rapid c-Src binding to ␤3 and its subsequent phosphorylation. c-Src represents 0.2-0.4% of total platelet protein (22), but only 3% is associated with the ␤3 CT (3). Thus, it is conceivable that the c-Src phosphorylation caused by anti-␣IIb TM could be an off-target effect, unrelated to the ability of anti-␣IIb TM to bind to ␣IIb, thereby releasing ␤3 to interact with cytoplasmic proteins (Fig. 1). To test this possibility, we compared anti-␣IIb TM-induced c-Src phosphorylation in GT and WT control platelets. Although c-Src was readily detected in both the GT and WT platelets, we did not detect phosphorylated Tyr 419 when the GT platelets were stimulated with either 10 g/ml collagen or 2 M anti-␣IIb TM (Fig. 4A). Thus, these results indicate that anti-␣IIb TM-induced phosphorylation of c-Src on Tyr 419 occurs in an ␣IIb␤3-specific manner. However, it remains possible that c-Src activation is a nonspecific consequence of simply incubating platelets with a hydrophobic membrane-soluble peptide. To address this possibility, we incubated platelets with the membrane-soluble peptide MS1 as a negative control. However, as shown in Fig. 4B, incubating platelets with MS1 did not cause c-Src phosphorylation. Taken together, these results confirm that anti-␣IIb TM-induced c-Src phosphorylation occurs when anti-␣IIb TM interacts with ␣IIb␤3.
Next, to address whether anti-␣IIb TM-induced c-Src activation requires ligand binding to ␣IIb␤3, we incubated washed platelets suspended in buffer lacking fibrinogen and containing ␣IIb␤3 antagonists with 2 M anti-␣IIb TM for 1 min. c-Src was then immunoprecipitated from platelet lysates and immunoblotted for the phosphorylated c-Src residue Tyr 419 . As shown in Fig. 4, C and D, anti-␣IIb TM induced c-Src phosphorylation despite the presence of EDTA or eptifibatide, implying that it occurred in the absence of ␣IIb␤3-bound ligand. It also occurred when platelets were preincubated with the actin polymerization inhibitors cytochalasin D or latrunculin, indicating that c-Src phosphorylation was independent of platelet cytoskeletal rearrangement. On the other hand, anti-␣IIb TM-induced c-Src phosphorylation was prevented by the c-Src kinase inhibitor PP2 but not by its inactive congener PP3 or by the Syk kinase inhibitor R406. Thus, these data indicate that anti-␣IIb TM induces c-Src phosphorylation by a process that is independent of both extracellular ligand binding to ␣IIb␤3 and platelet cytoskeletal rearrangement but requires c-Src enzymatic activity and likely results from c-Src trans-autophosphorylation.
Anti-␣IIb TM Causes Syk, but not FAK, Phosphorylation Independent of Ligand Binding to ␣IIb␤3-The tyrosine kinase Syk is rapidly and transiently phosphorylated following platelet stimulation in a c-Src-dependent reaction (23)(24)(25). Although Syk was initially reported to interact directly with the ␤3 CT of ligand-occupied ␣IIb␤3 (26), subsequent studies suggest that Syk binds instead to the c-Src phosphorylated tyrosines of immunoreceptor tyrosine-based activation motif (ITAM)-con- taining proteins such as platelet Fc␥RIIa (27). Regardless, we found that incubating human platelets with anti-␣IIb TM caused the phosphorylation of Syk on tyrosine residues (Fig. 5,  A and B). Anti-␣IIb TM-induced Syk phosphorylation was unaffected by EDTA, eptifibatide, cytochalasin D, and latrunculin A, indicating that it did not require ligand binding to ␣IIb␤3 or platelet cytoskeletal rearrangement, but it was prevented PP2, indicating that it was initiated by c-Src kinase activation (25,28). It was also diminished by the Syk inhibitor R406, indicating that it results in part from trans-autophosphorylation of Syk as well (25).
FAK also undergoes Src-mediated phosphorylation following platelet stimulation by agonists such as thrombin and collagen (29). However, in contrast to the rapidity of agonist-stimulated c-Src and Syk phosphorylation, FAK phosphorylation occurs minutes after platelet stimulation, coincident with the onset of platelet aggregation (24). Furthermore, although there is no conclusive evidence that FAK binds to ␤3 (30), agoniststimulated FAK phosphorylation is prevented by ␣IIb␤3 antagonists, implying that it depends on the binding of macromolecular multivalent ligands to ␣IIb␤3 (29). We found that incubating platelets with anti-␣IIb TM caused c-Src-dependent FAK phosphorylation (Fig. 5C). However, unlike anti-␣IIb TM-induced c-Src and Syk phosphorylation, anti-␣IIb TM-induced FAK phosphorylation was inhibited by EDTA and by the ␣IIb␤3 antagonists eptifibatide and abciximab, indicating anti-␣IIb TM-induced FAK phosphorylation requires ligand binding to ␣IIb␤3. Furthermore, although small-molecule ␣IIb␤3 antagonists such as eptifibatide have been shown to cause ␣IIb␤3 activation (31), incubating platelets with eptifibatide was not sufficient to cause FAK phosphorylation.

Effect of Mn 2ϩ on c-Src Phosphorylation in Platelets-Mn 2ϩ
activates integrins by perturbing the conformation of their extracellular domains (32). However, this appears to be a relatively local perturbation induced by the metal ion because it does not propagate to induce separation of the TM domains of the leukocyte integrin ␣L␤2 (33). We used this property of Mn 2ϩ to address whether anti-␣IIb TM-induced c-Src phosphorylation requires separation of the ␣IIb and ␤3 stalks. We found that, although thrombin and anti-␣IIb TM caused platelet aggregation and ATP secretion, incubating platelets with 2 mM MnCl 2 caused neither (Fig. 6A). Moreover, both thrombin and anti-␣IIb TM caused the phosphorylation of ␤3-bound c-Src but MnCl 2 did not (Fig. 6B). Thus, inducing an active conformation in the ␣IIb␤3 extracellular domain alone is insufficient to cause c-Src phosphorylation, and TM domain separation at a minimum is required for phosphorylation to occur.
Anti-␣IIb TM Causes Clustering of ␣llb␤3 in Platelets-Because both c-Src and Syk are activated by trans-autophosphorylation, it is likely that anti-␣IIb TM causes c-Src phosphorylation by inducing ␣IIb␤3 clustering. To test this possibility, we incubated washed human platelets for 60 s at 37°C in the presence or absence of 2 M anti-␣⌱Ib TM, after which the platelets were fixed with 4% paraformaldehyde and stained with the Alexa Fluor 488-labeled ␤3-specific monoclonal antibody SSA6. In the absence of anti-␣IIb TM, ␣llb␤3 was present in a homogenous ring at the periphery of the discoid platelets (Fig. 7, A and B). By contrast, when the platelets were incubated with anti-␣IIb TM, ␣IIb␤3 was redistributed into discrete nodular densities (Fig. 7, C and D). Classifying 270 resting platelets in 12 microscopic fields revealed that ␣IIb␤3 was homogeneously distributed in 80% of the platelets and as The experiment was performed three times. B, the bar graphs were generated from densitometry of the immunoblots from the three experiments and represent the mean Ϯ S.E. of phosphorylated Syk normalized using the densitometry of the corresponding Syk band. C, gel-filtered human platelets were incubated for 1 min with buffer, 10 g/ml collagen, 1 unit/ml thrombin, or 2 M anti-␣IIb TM, the latter in the absence or presence of 5 mM EDTA, 10 M PP2, 10 M PP3, 10 M eptifibatide, or 50 g/ml abciximab. FAK was immunoprecipitated from platelet lysates, and phosphorylated FAK was detected by immunoblotting using 4G10. nodular densities in the remaining 20%. Conversely, when 262 anti-␣IIb TM-incubated platelets were classified in six microscopic fields, ␣IIb␤3 was present exclusively as discrete nodular densities in 85% of the platelets and was homogeneously distributed in the remaining 15%. The differences in the distribution of ␣IIb␤3 in resting and anti-␣IIb TMincubated platelets were highly significant by chi-square testing (p ϭ 1.82E-50). Thus, incubating platelets with anti-␣IIb TM not only induces ␣IIb␤3 activation but causes the formation of ␣IIb␤3 clusters. In turn, ␣IIb␤3 clustering increases the local concentration of ␤3-bound c-Src (5), thereby facilitating c-Src trans-autophosphorylation.
Anti-␣IIb TM-induced ␣IIb␤3 Clustering Does Not Require Stable Binding of Talin-1 and Kindlin-3 to the ␤3 CT-How anti-␣IIb TM drives ␣IIb␤3 clustering is unclear, but it is possibly indirect, mediated by proteins bound to the ␤3 CT. Physiologic ␣IIb␤3 activation occurs when the 4.1/ezrin/radixin/ moesin domains of the focal adhesion protein kindlin-3 and the cytoskeletal protein talin-1 bind to the ␤3 CT (12,34,35). Talin-1 is predominantly present in platelets as an antiparallel homodimer (36,37). Thus, ␤3-bound talin-1 could cross-link adjacent activated ␣IIb␤3 molecules (35). It has also been suggested that ␤3-bound kindlin-3 supports talin-induced ␣IIb␤3 activation by promoting ␣IIb␤3 clustering (38). To test whether either protein mediates anti-␣IIb TM-induced ␣⌱Ib␤3 clustering, we incubated washed human platelets with 2 M anti-␣IIb TM or 1 unit/ml thrombin for specified intervals up to 180 s, after which the platelets were lysed using 1% Nonidet P-40. ␤3 was then immunoprecipitated from the lysates, and the immunoprecipitates were immunoblotted for talin-1 and kindlin-3. We detected little talin-1 associated with ␤3 in unstimulated platelets (Fig. 8, A and B). However, we found that the amount of ␤3-associated talin-1 increased 2.50 Ϯ 0.63-fold (n ϭ 4) after platelets were stimulated for 90 s with 1 unit/ml thrombin (Fig.  8C). By contrast, the amount of ␤3-associated talin-1 increased only 0.63 Ϯ 0.26-fold when platelets were exposed to 2 M anti-␣IIb TM for 90 s. Kindlin-3 appeared to be constitutively associated with ␤3 in unstimulated platelets (Fig. 8A), and there was a small, non-significant, 0.55 Ϯ 0.10-fold increase in the amount of ␤3-associated kindlin-3 when platelets were stimulated for 90 s with 1 unit/ml thrombin (p ϭ 0.12, t test for pair samples, n ϭ 4). On the other hand, the amount of ␤3-associated kindlin-3 actually decreased 0.42 Ϯ 0.11-fold below resting levels when platelets were incubated for 90 s with 2 M anti-␣IIb TM (Fig. 8C). In summary, although we cannot rule out a very weak or transient interaction with kindlin-3 and talin-1, these interactions appear much weaker when platelets are treated with anti-␣IIb TM than when platelets are activated by thrombin. Therefore, we consider it unlikely that either talin-1 and kindlin-3 binding to ␤3 accounts for the ␣IIb␤3 clustering that initiates c-Src trans-autophosphorylation when platelets are incubated with anti-␣IIb TM.

Discussion
Activation of the c-Src associated with the ␤3 CT in platelets triggers a protein phosphorylation cascade that results in outside-in platelet signaling (4). Here we report that this cascade can be initiated by activating individual ␣IIb␤3 molecules using a membrane-soluble peptide that binds to the ␣IIb TM domain. The membrane-soluble peptide, anti-␣IIb TM, was designed computationally to bind with high affinity and specificity to the ␣IIb TM domain (8,39). When added to suspensions of washed platelets, anti-␣IIb TM spontaneously inserts into the platelet plasma membrane, where it binds to the ␣IIb TM domain and causes ␣IIb␤3 activation by disrupting the TM domain heterodimer of resting ␣IIb␤3 (8,40). In this respect, anti-␣IIb TM mimics the physiologic ␣IIb␤3 activator talin, which causes ␣IIb␤3 TM domain separation when it binds to the membraneproximal region of ␤3 CT (Fig. 1) (12).
As we have reported recently (5), c-Src does not interact specifically with the ␤3 CT of resting platelets but binds to the CT and is phosphorylated within seconds following platelet stimulation by agonists such as thrombin. The time courses of anti-␣IIb TM-induced c-Src binding to the ␤3 CT and its subsequent phosphorylation mimic the response to thrombin, implying that the mechanisms may be the same. Nevertheless, it is possible that incubating platelets with a hydrophobic peptide like anti-␣IIb TM could have off-target effects that cause c-Src activation by a mechanism unrelated to ␣IIb␤3. To address this possibility, we added apyrase to platelet suspensions to metabolize any platelet ADP that might be nonspecifically released by anti-␣IIb TM, tested platelets from two patients with GT to confirm that ␣IIb␤3 must be present to observe an anti-␣IIb TM effect, and used an unrelated membrane-soluble peptide, MS1, to further control for nonspecific peptide effects. Because anti-␣IIb TM caused c-Src phosphor-ylation, despite the presence of apyrase, did not cause c-Src phosphorylation in GT platelets, and MS1 neither caused platelet aggregation, platelet secretion, or c-Src phosphorylation, anti-␣IIb TM appears to initiate platelet outside-in signaling by directly interacting with ␣IIb␤3.
Besides activating c-Src, the platelet agonists thrombin, collagen, and ADP cause phosphorylation and subsequent activation of the non-receptor tyrosine kinase Syk (24). Initially, Syk was thought to directly interact with the distal ␤3 CT at a site overlapping the c-Src binding site (26). However, more recent observations suggest that Syk is activated after binding via its SH2 domains to ITAM-containing proteins such as Fc␥RIIa previously phosphorylated by activated c-Src (27,41). We found that anti-␣IIb TM, like conventional platelet agonists, causes Syk phosphorylation on tyrosine residues. Moreover, anti-␣IIb TM-stimulated Syk phosphorylation was prevented by the Src-kinase inhibitor PP2, consistent with a proximal requirement for c-Src activation, and was attenuated by an inhibitor of Syk kinase activity, suggesting that, like c-Src, Syk undergoes trans-autophosphorylation (25).
The tyrosine kinase FAK is also phosphorylated by activated c-Src in agonist-stimulated platelets (29). Although it is clear that FAK phosphorylation in platelets requires ␣IIb␤3 (29) and can be induced by forced ␣IIb␤3 clustering (42), there is no conclusive evidence that FAK actually associates with the integrin (30). We found that anti-␣IIb TM initiates the c-Src-dependent phosphorylation of FAK, but like FAK phosphorylation induced by conventional agonists, it was prevented by ␣IIb␤3 antagonists and actin polymerization inhibitors (data not shown). Thus, although direct activation of single ␣IIb␤3 molecules alone appears sufficient to activate c-Src and Syk, events subsequent to ␣IIb␤3 activation appear to be required to cause FAK activation.
The observation that anti-␣IIb TM-induced ␣IIb␤3 activation causes c-Src trans-autophosphorylation implies that anti-␣IIb TM-activated ␣IIb␤3 undergoes oligomerization. Moreover, because anti-␣IIb TM-induced c-Src phosphorylation was not impaired by EDTA or by ␣IIb␤3 antagonists, it appears that the necessary ␣IIb␤3 oligomerization was not driven by extracellular ligand binding. The redistribution of ␣IIb␤3 into discrete nodular densities we observed by immunofluorescence when suspended platelets were incubated with anti-␣IIb TM in the presence of EDTA supports this possibility.
How activated ␣IIb␤3 might undergo spontaneous oligomerization is unclear. It is likely that oligomerization is indirect (43)(44)(45)(46), mediated by proteins bound to the ␤3 CT of activated ␣IIb␤3. Physiologic ␣IIb␤3 activation occurs when the head domain of the cytoskeletal protein talin binds to the membrane-proximal ␤3 CT (12,34). Talin, a 270-kDa protein, consists of a 50-kDa N-terminal 4.1/ezrin/radixin/moesin domain connected to a 200-kDa rod domain by a largely unstructured flexible linker (36). The C terminus of the talin rod also contains a dimerization domain, and, in platelets, talin is predominantly present as an antiparallel homodimer (36,37). Thus, ␤3-bound talin would seem a likely candidate to cross-link adjacent activated ␣IIb␤3 molecules (35). However, we found that, although thrombin stimulation caused an increase the amount of ␤3-associated talin-1, there was essentially no change in the amount of ␤3-associated when platelets were incubated with anti-␣IIb TM. This makes it improbable that talin-1 is responsible for the apparent ␣IIb␤3 oligomerization that follows ␣IIb␤3 activation by anti-␣IIb TM.
In addition to talin, kindlin-3, a member of the kindlin family of focal adhesion proteins, is required for agonist-stimulated ␣IIb␤3 activation in platelets (35). Kindlin-3 binds to the ␤3 CT via its 4.1/ezrin/radixin/moesin domain at a site distinct from talin (47). Although kindlin-3 binding to ␤3 alone is not sufficient to cause ␣IIb␤3 activation, agonist-stimulated ␣IIb␤3 activation in platelets does not occur when kindlin-3 is absent, despite the presence of talin (48). The mechanism by which kindlin-3 "primes" talin-induced ␣IIb␤3 activation is unknown, but it has been suggested that kindlin-3 increases the avidity of ␣IIb␤3 for multivalent, but not univalent, soluble ligands by promoting ␣IIb␤3 clustering (38). However, kindlins are monomeric (35), so their ability to cause ␣IIb␤3 clustering would have to be indirect, and additional kindlin binding partners such as migfilin (49) and integrin-linked kinase (ILK) (50,51) would be required. We detected the constitutive association of kindlin-3 with ␤3 in resting platelets. Although there was a small increase in the amount of ␤3-associated kindlin-3 when platelets were stimulated by thrombin, the amount of ␤3-associated kindlin-3 progressively declined when platelets were incubated with anti-␣IIb TM. Thus, like talin-1, these results indicate that kindlin-3 binding to the ␤3 CT does not account for anti-␣IIb TM-induced ␣IIb␤3 clustering.
In summary, we found that anti-␣IIb TM, a computationally designed peptide that binds with high affinity to the ␣IIb TM domain, causes ␣IIb␤3 activation, platelet aggregation, and ATP secretion as well as the phosphorylation of ␤3-associated c-Src.Because␤3-associatedc-Srcisactivatedbytrans-autophosphorylation, these observations imply that activation of individual ␣IIb␤3 molecules can initiate ␣IIb␤3 oligomerization even when ligand binding is absent. Thus, our studies indicate that not only is resting ␣IIb␤3 poised to undergo a conformational FIGURE 8. Anti-␣IIb TM-and thrombin-stimulated talin-1 and kindlin-3 binding to ␤3. Washed platelets were incubated with 2 M anti-␣IIb TM or 1 unit/ml thrombin for 30, 90, and 180 s, after which the platelets were lysed with 1% Nonidet P-40, ␤3 was immunoprecipitated from the platelet lysates, and the ␤3 immunoprecipitates were immunoblotted for talin-1 and kindlin-3. A, time course of talin-1 and kindlin-3 binding to ␤3 following platelet simulation by 1 unit/ml thrombin and 2 M anti-␣IIb TM. These experiments were repeated four times. The lower molecular weight (MW) band seen in several of the kindlin-3 lanes is nonspecific and is present sporadically when kindlin-3 is immunoblotted with polyclonal anti-kindlin-3 antibodies. B, densitometry using National Institutes of Health ImageJ software of the immunoblots shown in A. To facilitate comparisons, the densitometry for ␤3-bound talin-1 and kindlin-3 at each time point was normalized using the densitometry for the 0 time point. C, comparison of the amount of ␤3-associated talin-1 and kindlin-3 at 90 s after platelet stimulation by 1 units/ml thrombin or 2 M anti-␣IIb TM. The data shown are the mean Ϯ S.E. of the four separate experiments and were generated from the 90-s time point of the four experiments.
change that exposes its ligand binding site, but it is poised to rapidly assemble into intracellular signal-generating complexes as well.
Author Contributions-H. Z. and K. P. F. performed the majority of the experiments with the assistance of L. M. S., K. Y., and D. T. M. R. T. synthesized anti-␣IIb TM and MS1, supervised by H. Y. W. F. D. analyzed the data and assisted with the preparation of the manuscript: J. S. B. designed the experiments, analyzed the data, and prepared the manuscript.