The Platelet Cytoskeleton Regulates the Affinity of the Integrin αIIbβ3 for Fibrinogen*

Agonist-generated inside-out signals enable the platelet integrin αIIbβ3 to bind soluble ligands such as fibrinogen. We found that inhibiting actin polymerization in unstimulated platelets with cytochalasin D or latrunculin A mimics the effects of platelet agonists by inducing fibrinogen binding to αIIbβ3. By contrast, stabilizing actin filaments with jasplakinolide prevented cytochalasin D-, latrunculin A-, and ADP-induced fibrinogen binding. Cytochalasin D- and latrunculin A-induced fibrinogen was inhibited by ADP scavengers, suggesting that subthreshold concentrations of ADP provided the stimulus for the actin filament turnover required to see cytochalasin D and latrunculin A effects. Gelsolin, which severs actin filaments, is activated by calcium, whereas the actin disassembly factor cofilin is inhibited by serine phosphorylation. Consistent with a role for these factors in regulating αIIbβ3 function, cytochalasin D- and latrunculin A-induced fibrinogen binding was inhibited by the intracellular calcium chelators 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester and EGTA acetoxymethyl ester and the Ser/Thr phosphatase inhibitors okadaic acid and calyculin A. Our results suggest that the actin cytoskeleton in unstimulated platelets constrains αIIbβ3 in a low affinity state. We propose that agonist-stimulated increases in platelet cytosolic calcium initiate actin filament turnover. Increased actin filament turnover then relieves cytoskeletal constraints on αIIbβ3, allowing it to assume the high affinity conformation required for soluble ligand binding.

The function of integrins in circulating blood cells is regulated by limiting their access to ligands. The prototypical example of this form of regulation is the platelet integrin ␣ IIb ␤ 3 since agonist-generated "inside-out" signals are required to enable ␣ IIb ␤ 3 to bind soluble ligands such as fibrinogen (1). Because the binding of soluble fibrinogen to ␣ IIb ␤ 3 is a prerequisite for platelet aggregation (2), regulating the affinity of ␣ IIb ␤ 3 for fibrinogen in this way assures that only stimulated platelets aggregate. It is likely that intracellular molecules regulate the function of ␣ IIb ␤ 3 by interacting with its cytoplasmic domains (3), but the identity of these molecules and how they interact with ␣ IIb ␤ 3 are not known.
The actin cytoskeleton appears to play a role in regulating ␣ IIb ␤ 3 function. Thus, micromolar concentrations of cytochalasins, fungal metabolites that impair actin polymerization by binding to the barbed end of actin filaments (4), inhibit agonistinduced fibrinogen binding to ␣ IIb ␤ 3 on platelets (5-7), whereas nanomolar concentrations induce fibrinogen binding to recombinant ␣ IIb ␤ 3 expressed on the surface of B lymphocytes (8). In the work described in this paper, we used cytochalasin D (Cyto-D), 1 latrunculin A (Lat-A; another inhibitor of actin polymerization (9)), and jasplakinolide (a compound that stabilizes actin filaments (10)) to examine the role of actin filament turnover in regulating the affinity of ␣ IIb ␤ 3 for fibrinogen in platelets.
Assays of Platelet Function-To determine whether changes in actin filament turnover affect fibrinogen binding to platelets, unstirred suspensions of Ϸ3 ϫ 10 6 platelets were incubated at room temperature for various periods of time with either Cyto-D or Lat-A dissolved in dimethyl sulfoxide, with jasplakinolide dissolved in methanol, or with equivalent volumes of dimethyl sulfoxide or methanol alone. The platelets were then transferred to tubes containing 0.5 mM CaCl 2 and 200 g/ml human fibrinogen radiolabeled with 125 I by the iodine monochloride technique (2). Following a 5-min incubation at 37°C, unbound and platelet-bound fibrinogens were separated by sedimenting the platelets through silicone oil. The supernatant buffer and oil were then aspirated, and the pelleted platelets were counted for 125 I. The effect of intracellular calcium chelators on fibrinogen binding was measured after incubating gel-filtered platelets with 10 M BAPTA/AM or EGTA/AM for 30 min at room temperature. Turbidometric platelet aggregation was measured in a Chrono-Log dual-channel aggregometer (12).
Three methods were used to measure platelet secretion. First, secretion of platelet-dense granule serotonin was measured after loading platelets in plasma for 30 min with [ 14 C]serotonin at 0.038 Ci/ml. Following gel filtration, the platelets were incubated with Cyto-D, Lat-A, or jasplakinolide in the presence of 2 M imipramine for 20 min at room temperature and transferred to tubes containing 0.5 mM CaCl 2 and 200 g/ml human fibrinogen for a 3-min incubation at 37°C. The platelets were then sedimented through silicone oil, and the [ 14 C]serotonin content of a 100-l aliquot of the resulting platelet-free supernatant was measured and compared with the [ 14 C]serotonin content of a 100-l aliquot of the original platelet suspension. The ability of ADP and thrombin to induce [ 14 C]serotonin secretion was tested by incubating the platelets with either 10 M ADP or 2 units/ml thrombin for 5 min. Second, the secretion of platelet-dense granule ATP was measured using a Lumi-Aggregometer (Chrono-Log) as described previously (13). Third, the secretion-dependent translocation of the ␣-granule membrane protein P-selectin to the platelet surface was measured using the P-selectin-specific monoclonal antibody S12 and flow cytometry as described previously (14,15).
Measurement of Actin Nucleating Activity in Gel-filtered Platelets-Pyrenyl-actin assays of actin nucleation were performed as described previously (16). Briefly, 20-l aliquots of a suspension of gel-filtered platelets at 1 ϫ 10 8 cells/ml were diluted directly into 1 ml of 25 mM Tris buffer, pH 7.4, containing 1% Triton X-100, 0.14 M KCl, 2 mM MgCl 2 , 1 mM EGTA, and 1 mM ATP; 1.5 M pyrenyl-G-actin was added just before use. The rate of pyrenyl-actin polymerization between 100 and 300 s was determined from the increase in pyrenyl fluorescence at Ex 370 nm / Em 410 nm . All samples had the same final concentration of seeds, supernatant, and pyrenyl-actin. The presence of 2 M cytochalasin B decreased the rate of polymerization by Ͼ90% (data not shown), indicating that the polymerization was due to barbed-end elongation.
Measurement of the Platelet Content of F-actin-The platelet content of F-actin was measured using the rhodamine-phalloidin binding assay described by Cassimeris et al. (17). Briefly, platelets incubated with Cyto-D, Lat-A, or ADP were fixed in 1% glutaraldehyde, permeabilized using 1% Nonidet P-40, and incubated with 400 nM rhodamine-phalloidin for 60 min. The platelets were then pelleted at 1200 ϫ g, and incorporated rhodamine-phalloidin was extracted from the platelet pellet with 1 ml of methanol. Rhodamine fluorescence in the extract was measured in a fluorometer using an excitation wavelength of 540 nm and an emission wavelength of 575 nm.
Flow Cytometry-Flow cytometry measurements were made using a FACSCalibur flow cytometer (Becton Dickinson) formatted for one-color analysis using CellQuest Version 1.2.2 software as described previously (15). Binding of the ␣ IIb ␤ 3 -specific, activation-dependent mAb PAC1 (18) was measured by preincubating platelets with fluorescein isothiocyanate-labeled PAC1 (40 g/ml) for 30 min. PAC1 binding was then analyzed at selected time points after the addition of agonist. To measure the binding of the conformation-specific mAbs 10-758 (␤ 3 -specific), 15-758 (␤ 3 -specific), and 42-758 (␣ IIb -specific), gel-filtered platelets at a concentration of 2 ϫ 10 7 cells/ml were incubated sequentially at room temperature with either 10 M L-739,758 or 1 M Cyto-D for 30 min or with 20 M ADP for 5 min, with each mAb for 60 min, and with a 1:10 dilution of fluorescein isothiocyanate-labeled goat anti-mouse IgG for 60 min. Platelet cytosolic calcium was measured by loading platelets in platelet-rich plasma with 25 M fluo-3/AM for 60 min at 37°C. Following the addition of agonist, samples were analyzed for fluo-3 fluorescence over a period of 3 min using the time acquisition mode of the flow cytometer.

Effect of Cyto-D and
Lat-A on the Affinity of ␣ IIb ␤ 3 for Fibrinogen-To investigate the role of actin polymerization in regulating the affinity of ␣ IIb ␤ 3 for fibrinogen, we incubated unstimulated gel-filtered human platelets with various concentrations of Cyto-D and measured the amount of 125 I-labeled fibrinogen that specifically bound to the incubated platelets. As shown in Fig. 1a, exposing unstimulated platelets to Cyto-D for 30 min induced fibrinogen binding in a concentration-dependent manner. Maximum fibrinogen binding occurred at a Cyto-D concentration of 1 M, and fibrinogen binding decreased at Cyto-D concentrations Ͼ1 M or when incubations were prolonged beyond 30 min (data not shown). In 21 experiments, 1 M Cyto-D induced 43 Ϯ 4% as much fibrinogen binding as stimulating platelets with 10 M ADP. Furthermore, Cyto-D-induced fibrinogen binding was completely inhibited by the ␣ IIb ␤ 3 -specific mAb A2A9 (12), confirming that the fibrinogen was bound to ␣ IIb ␤ 3 (data not shown).
Cyto-D inhibits actin polymerization by binding to the barbed end of actin filaments (4). To confirm that Cyto-D induced fibrinogen binding to ␣ IIb ␤ 3 by impairing actin filament turnover, we used Lat-A, a potent marine toxin that impairs the polymerization of actin filaments by sequestering G-actin monomers (9). As shown in Fig. 1b, incubating platelets with Lat-A for 20 min also induced fibrinogen binding in a concentration-dependent manner. Maximum fibrinogen binding occurred at a Lat-A concentration of 1 M, and fibrinogen binding decreased at greater Lat-A concentrations and when the incubations were prolonged beyond 20 min (data not shown). In 14 separate experiments, Lat-A induced 51 Ϯ 2% as much fibrinogen binding as stimulating platelets with 10 M ADP.
To measure the affinity of platelet ␣ IIb ␤ 3 for fibrinogen induced by Cyto-D and Lat-A, we incubated unstimulated platelets with 1 M Cyto-D or Lat-A in the presence of increasing concentrations of 125 I-labeled fibrinogen. As shown in Fig. 2, the fibrinogen binding induced by Cyto-D and Lat-A was saturable. Dissociation constants for fibrinogen binding calculated from these data were 57 Ϯ 8 nM for Cyto-D and 110 Ϯ 10 nM for Lat-A, values comparable to the dissociation constants of 81 Ϯ 12 and 178 Ϯ 18 nM for ADP-and epinephrine-stimulated platelets, respectively (2). In addition, we found that fibrinogen binding induced by 1 M Cyto-D or Lat-A supported the aggregation of platelets stirred in an aggregometer nearly as well as stimulating platelets with ADP (data not shown).
The agonist-induced increase in ␣ IIb ␤ 3 affinity that results in fibrinogen binding is associated with an alteration in the con- The Platelet Cytoskeleton Regulates Ligand Binding to ␣ IIb ␤ 3 formation of ␣ IIb ␤ 3 that can be detected using conformationspecific mAbs (14). To determine whether incubating platelets with Cyto-D or Lat-A also produces a conformational change in ␣ IIb ␤ 3 , we compared the binding of a number of conformationspecific mAbs to ␣ IIb ␤ 3 on platelets exposed to Cyto-D and various platelet agonists. The PAC1 mAb binds to an epitope expressed exclusively by the activated conformation of ␣ IIb ␤ 3 at or near its fibrinogen-binding site (18). As shown in Fig. 3a, 1 M Cyto-D induced PAC1 binding to platelets that was comparable to the PAC1 binding induced by the thrombin receptoractivating peptide. Occupation of ␣ IIb ␤ 3 by ligands such as the Arg-Gly-Asp mimetic L-739,758 and, to a variable degree, agonist stimulation in the absence of ligand binding result in the exposure of neoepitopes called ligand-induced binding sites on ␣ IIb and ␤ 3 (19). As shown in Fig. 3b, the binding of two ␤ 3 -specific FЈ ligand-induced binding sites mAbs (10-758 and 15-758) and one ␣ IIb -specific FЉ ligand-induced binding sites mAb (42-217) to platelets exposed to 1 M Cyto-D was intermediate between that induced by L-739,758 and ADP. Comparable results were seen when platelets were incubated with Lat-A (data not shown). Thus, these experiments indicate that not only do Cyto-D and Lat-A increase the affinity of ␣ IIb ␤ 3 for ligands such as fibrinogen, they induce a conformational change in ␣ IIb ␤ 3 as well.
Effect of Jasplakinolide on Fibrinogen Binding to Platelets-As an additional test of the hypothesis that actin filaments regulate the affinity of ␣ IIb ␤ 3 for ligands, we examined the effects of jasplakinolide, a cell-permeable cyclic peptide that binds to and stabilizes actin filaments (10). Preincubating platelets with 5-10 M jasplakinolide for as short a period as 10 min completely prevented Cyto-D-and Lat-A-induced fibrinogen binding. Moreover, as shown in Fig. 4, preincubating platelets with jasplakinolide also progressively inhibited ADP-stimulated fibrinogen binding, such that following a 30-min incubation, ADP-stimulated fibrinogen binding was completely inhibited. The inhibitory effect of jasplakinolide on fibrinogen binding was at least partially specific for ␣ IIb ␤ 3 activation because preincubating platelets with 10 M jasplakinolide had no effect on thrombin-stimulated [ 14 C]serotonin secretion (Fig.  5b). Thus, these experiments indicate that stabilizing actin filaments in platelets prevents Cyto-D-, Lat-A-, and ADP-in-duced fibrinogen binding and provide additional evidence that the actin cytoskeleton plays a role in regulating the affinity of ␣ IIb ␤ 3 for soluble ligands.
Cyto-D-and Lat-A-induced Fibrinogen Binding Requires Subthreshold Concentrations of ADP-There appears to be little, if any, actin filament turnover in unstimulated platelets, although a rapid increase follows platelet stimulation (20 -22). Thus, it is likely that Cyto-D or Lat-A induced fibrinogen binding to ␣ IIb ␤ 3 by either interrupting a slow rate of spontaneous actin filament turnover in the submembranous lattice of unstimulated platelets or interrupting slow actin filament turnover initiated by a subthreshold platelet stimulus. Consistent with this hypothesis, we found that Cyto-D-induced fibrinogen binding was inhibited by impairing platelet ATP synthesis with 2-deoxyglucose and sodium azide or by exposing the platelets to the inhibitory prostaglandin E 1 (Fig. 5a). Moreover, we found that preincubating platelets with either of the two ADP scavengers apyrase and creatine phosphate/creatine phosphokinase, but not with the prostaglandin-endoperoxide H synthase inhibitor indomethacin, reduced fibrinogen binding induced by FIG. 2. Fibrinogen binding to human platelets induced by cytochalasin D and latrunculin A as a function of fibrinogen concentration. Fibrinogen binding to gel-filtered human platelets was measured after incubating platelets with 1 M cytochalasin D (gray circles) or 1 M latrunculin A (black circles) and the indicated concentrations of 125 I-labeled fibrinogen for 30 min at room temperature as described in the legend to Fig. 1 and under "Experimental Procedures." The binding isotherms were generated from the data using SigmaPlot software (Jandel Scientific).  Cyto-D to base-line levels (Fig. 5a). Thus, it is likely that subthreshold concentrations of extracellular ADP released into plasma or into the platelet suspension medium during gel filtration provided the stimulus for the actin filament turnover required to see the Cyto-D and Lat-A effects.
On the other hand, stimulating unstirred platelets with strong agonists such as thrombin results in the secretion of ADP stored in platelet-dense granules (12). Thus, it is conceivable that Cyto-D and Lat-A simply induced secretion of dense granule ADP that in turn was responsible for fibrinogen binding by ␣ IIb ␤ 3 . We addressed this possibility in three ways. First, we loaded the platelet-dense granules with [ 14 C]serotonin and measured [ 14 C]serotonin release into the medium during a 30-min incubation with Cyto-D or Lat-A. In contrast to thrombin stimulation, which resulted in the secretion of 80% of the total platelet serotonin pool, there was no serotonin release when platelets were incubated with Cyto-D or Lat-A (Fig. 5b). Second, we found that incubating platelets with Cyto-D and Lat-A did not result in the secretion of dense granule ATP (data not shown). Finally, neither Cyto-D nor Lat-A induced the translocation of the dense granule membrane protein P-selectin to the platelet surface (data not shown).
Effect of Apyrase on the Actin Nucleating Activity and F-actin Content in Gel-filtered Platelets-Agonist stimulation increases the ability of lysed platelets to nucleate actin polymerization (16,23). To determine whether the subthreshold concentrations of ADP responsible for Cyto-D-and Lat-A-induced fibrinogen binding actually affected the actin cytoskeleton in platelets, we measured the actin nucleating activity in platelets that had been incubated in the absence or presence of 5 units/ml apyrase. Actin nucleating activity was measured as the rate of pyrene-labeled G-actin polymerization after its addition to Triton X-100 lysates of platelets (16). We found that the rate of pyrenyl-actin polymerization induced by platelets that had been incubated in the absence of apyrase was 29 Ϯ 7% greater than when platelets were incubated in its presence (p ϭ 0.004). Adding apyrase after platelet lysis had no effect on pyrenylactin incorporation, whereas incorporation was inhibited completely by adding 0.2 M Cyto-D to the assay, indicating that the pyrenyl-actin polymerization was due primarily to barbedend elongation. On the other hand, nucleating activity increased Ϸ1.5-fold when the platelets were stimulated with 1 unit/ml thrombin for 30 s before lysis.
We also measured the effect of apyrase on the content of F-actin in platelets using a quantitative rhodamine-phalloidin binding assay (17). The F-actin content of platelets incubated in the absence of apyrase was 33% greater than the basal level observed in platelets that had been incubated in the presence of apyrase. Stimulating platelets with 1 unit/ml thrombin increased their F-actin content to ϳ200% of the basal level, and this was unaffected by the presence or absence of apyrase. Last, the presence of 5 M cytochalasin B reduced the F-actin content of both thrombin-stimulated and unstimulated platelets to that of the unstimulated platelets incubated with apyrase. Taken together, the measurements of actin nucleating activity and F-actin content support our hypothesis that actin filament turnover initiated by subthreshold concentrations of ADP enabled Cyto-D and Lat-A to induce fibrinogen binding to ␣ IIb ␤ 3 .

Effect of Cyto-D and Lat-A on ADP-stimulated Fibrinogen
Binding to ␣ IIb ␤ 3 -To determine whether Cyto-D and Lat-A also affect fibrinogen binding stimulated by suprathreshold concentrations of ADP, we incubated gel-filtered human platelets with increasing concentrations of Cyto-D or Lat-A and then compared unstimulated fibrinogen binding with fibrinogen binding stimulated by 10 M ADP. As shown in Fig. 6 and consistent with the data shown in Fig. 1, fibrinogen binding in Thus, these experiments suggest that concentrations of Cyto-D and Lat-A sufficient to induce maximal fibrinogen binding to platelets exposed to subthreshold concentrations of ADP were also sufficient to uncouple ADP stimulation and fibrinogen binding in platelets exposed to suprathreshold concentrations of this agonist.

Inhibition of Cyto-D-and Lat-A-induced Fibrinogen Binding by Intracellular Calcium Chelators-Platelet agonists like
ADP stimulate increases in the calcium concentration in the platelet cytosol (24). We addressed a role for cytosolic calcium in the platelet response to Cyto-D and Lat-A by examining the consequence of loading platelets with the intracellular calcium chelators BAPTA/AM and EGTA/AM. As shown in Fig. 7a, preincubating platelets with either 10 M BAPTA/AM or 20 M EGTA/AM completely inhibited fibrinogen binding stimulated by Cyto-D and Lat-A. Similarly, preincubating platelets with 10 M BAPTA completely inhibited ADP-stimulated fibrinogen binding, whereas preincubating platelets with 20 M EGTA/AM inhibited ADP-stimulated fibrinogen binding by 31%. Thus, these results indicate that free cytosolic calcium is required for Cyto-D-and Lat-A-induced fibrinogen binding.
To rule out the possibility that Cyto-D and Lat-A, by themselves, increase the calcium concentration in the platelet cytosol, we loaded platelets with the fluorescent calcium indicator fluo-3/AM and measured changes in fluo-3 fluorescence following platelet exposure to thrombin, ADP, and Cyto-D using flow cytometry. Whereas thrombin and, to a lesser extent, ADP induced a rapid increase in fluo-3 fluorescence indicative of an increase in cytosolic calcium, there was no change in the fluorescence of platelets incubated with Cyto-D (Fig. 7b), even when the incubation was extended to 20 min (data not shown).

Effect of Serine/Threonine Phosphatase Inhibitors on Cyto-D-and Lat-A-induced Fibrinogen
Binding-Agonist-stimulated platelet function is impaired in the presence of Ser/Thr phosphatase inhibitors (25). Moreover, Davidson and Haslam (26) observed that cofilin, an actin disassembly factor in platelets that is inhibited by serine phosphorylation, is dephosphorylated following the exposure of platelets to the calcium ionophore A23187. To address the possible role of a serine/ threonine phosphatase in the response of platelets to Cyto-D and Lat-A, we preincubated platelets with the serine/threonine phosphatase inhibitors okadaic acid and calyculin A and meas- The Platelet Cytoskeleton Regulates Ligand Binding to ␣ IIb ␤ 3 ured fibrinogen binding induced by ADP, Cyto-D, and Lat-A. As shown in Fig. 8, preincubating platelets with okadaic acid and especially calyculin A inhibited ADP-, Cyto-D-, and Lat-A-induced fibrinogen binding. Thus, these results suggest that a serine/threonine phosphatase is involved in the responsiveness of platelets to these agents. DISCUSSION Our data suggest that an increase in actin filament turnover in platelets induced by inside-out signaling relieves cytoskeletal constraints on the integrin ␣ IIb ␤ 3 , thereby increasing its affinity for soluble ligands. It was observed previously that releasing cytoskeletal constraints on integrins can increase the avidity of cell adhesion, presumably by increasing the likelihood of random encounters between integrins and immobilized ligands. Kucik et al. (27) reported that exposing B lymphocytes to either phorbol 12-myristate 13-acetate or submicromolar concentrations of Cyto-D increases the random movement of the integrin ␣ L ␤ 2 while, at the same time, substantially increasing the number of cells adherent to ICAM-1. Lub et al. (28) made similar observations, whereas Elemer and Edgington (29) found that exposing monocytes to cytochalasin B induces soluble fibrinogen and Factor X binding to the integrin ␣ M ␤ 2 , implying that disrupting cytoskeletal associations in leukocytes can also alter the affinity of ␤ 2 integrins for ligands. On the other hand, Stewart et al. (30) reported that although calpain-mediated cytoskeletal disassembly in lymphocytes induces ␣ L ␤ 2 clustering and cell adhesion to ICAM-1, it does so without increasing the affinity of ␣ L ␤ 2 for soluble ICAM-1.
Unlike leukocytes whose integrins interact with inducible membrane proteins like ICAM-1, circulating platelets undergo homotypic aggregation after soluble fibrinogen or von Willebrand factor binds to ␣ IIb ␤ 3 (31). Because the concentrations of fibrinogen and von Willebrand factor in plasma are substantially in excess of those required to saturate ␣ IIb ␤ 3 (2), circulating platelets maintain ␣ IIb ␤ 3 in an inactive state to prevent spontaneous platelet aggregation. However, platelet stimulation, either by inducing a change in the conformation of ␣ IIb ␤ 3 or by inducing ␣ IIb ␤ 3 clustering on the platelet surface or both, enables ␣ IIb ␤ 3 to bind soluble ligands (7,18,32). Which of these events is the predominate factor in controlling ␣ IIb ␤ 3 affinity is not clear, although Hato et al. (33) recently concluded, from experiments in which ␣ IIb ␤ 3 could be clustered in the absence of agonist stimulation, that clustering by itself makes only a modest contribution to the affinity of ␣ IIb ␤ 3 for soluble ligands.
A substantial fraction of the integrin ␣ IIb ␤ 3 in the plasma membrane of unstimulated platelets is associated with a submembranous cytoskeletal lattice containing short actin filaments, spectrin, talin, and vinculin (34,35). Platelet stimulation results in fragmentation of the actin filaments in this lattice (23), and coincident with platelet aggregation, ␣ IIb ␤ 3 redistributes to a detergent-insoluble cytoskeletal core (34), suggesting that either platelet stimulation or ligand binding disrupts its association with the membrane skeleton. We found that exposing unstimulated gel-filtered platelets to the actin polymerization inhibitors Cyto-D and Lat-A induced the binding of soluble fibrinogen to ␣ IIb ␤ 3 , suggesting that disrupting the association of ␣ IIb ␤ 3 with the membrane skeleton also alters its affinity for soluble ligands. The converse of these observations, stabilizing actin filaments with jasplakinolide, prevented the induction of fibrinogen binding by Cyto-D and Lat-A as well as by ADP and provides further support for this premise.
Since neither Cyto-D nor Lat-A by itself initiates actin filament turnover (4,9), it is likely that their effects on ␣ IIb ␤ 3 function resulted from perturbing either a slow rate of spontaneous actin filament turnover in unstimulated platelets or, more likely, actin turnover initiated by a subthreshold platelet stimulus. We found that Cyto-D-and Lat-A-induced fibrinogen binding was prevented by ADP scavengers, suggesting that the subthreshold concentrations of ADP that are present in platelet-rich plasma (13) or that likely are released during the gel filtration of platelets were the stimuli that initiated actin filament turnover. Consistent with this interpretation, we found that incubating gel-filtered platelets with the ADP scavenger apyrase decreased their content of F-actin and inhibited the actin nucleating activity of platelet lysates. Paradoxically, we also found that Cyto-D and Lat-A inhibited fibrinogen binding stimulated by suprathreshold concentrations of ADP, a finding consistent with previous reports that cytochalasins inhibit fibrinogen binding to agonist-stimulated platelets (5-7). However, neither agent was able to completely inhibit fibrinogen binding; indeed, there was no difference in the amount of fibrinogen bound at Cyto-D and Lat-A concentrations Ն1 M, regardless of whether the platelets were stimulated by subthreshold or suprathreshold concentrations of ADP. Thus, Cyto-D and Lat-A inhibit the additional ␣ IIb ␤ 3 activation induced by high concentrations of agonist.
It seems unlikely that a single explanation can account for both the enhancement and inhibition of fibrinogen binding by Cyto-D and Lat-A. It is possible that at subthreshold agonist concentrations, Cyto-D and Lat-A are able to relieve cytoskeletal constraints only on a portion of the integrin ␣ IIb ␤ 3 in the platelet membrane; but this would not explain that fact that in the presence or absence of agonist, the fibrinogen bound when Cyto-D and Lat-A concentrations exceed 1 M is the same. Most likely, when at least subthreshold concentrations of agonist are present, ␣ IIb ␤ 3 activation results from destabilizing actin filaments with Cyto-D and Lat-A. At higher agonist concentrations, ␣ IIb ␤ 3 activation results from actin turnover associated with net actin polymerization. In a similar situation, the lamellipodia of chemoattractant-stimulated leukocytes contain a labile pool of actin filaments that turn over rapidly, continually polymerizing at their barbed ends and depolymerizing at their pointed ends, in the presence of chemoattractant (17). Nevertheless, merely increasing the platelet content of F-actin is not sufficient to activate ␣ IIb ␤ 3 because jasplakinolide, which stabilizes F-actin, inhibited fibrinogen binding. Thus, it is likely that agonist stimulation reorganizes the membrane skeleton as well.
How might agonist stimulation reorganize the membrane skeleton? Platelets contain at least two proteins that specifically sever and/or depolymerize actin filaments and could be The Platelet Cytoskeleton Regulates Ligand Binding to ␣ IIb ␤ 3 involved in the platelet response to Cyto-D and Lat-A. One is gelsolin, a calcium-activated protein that severs actin filaments by rupturing noncovalent bonds between actin subunits, followed by capping the barbed end of the severed filaments (36). Gelsolin-deficient mice have prolonged bleeding times, consistent with defective platelet function. Moreover, there is decreased actin fragmentation following platelet activation and an accompanying decrease in actin nucleating activity. A second protein is cofilin, a ubiquitously expressed member of the cofilin/ADF family of small actin-binding proteins (37). The ability of cofilin to disassemble actin filaments is inhibited by LIM kinase-mediated phosphorylation of Ser-3 (38,39). Davidson and Haslam (26) found that Ϸ25% of the cofilin in unstimulated platelets is phosphorylated and is rapidly dephosphorylated following platelet exposure to the calcium ionophore A23187. We found that Cyto-D-and Lat-A-induced fibrinogen binding was inhibited by preincubating platelets with intracellular calcium chelators or with inhibitors of the Ser/Thr phosphatases PP1/PP2A (40). Thus, it is possible that the former could be acting, at least in part, to prevent the activation of gelsolin, whereas the effect of latter could be to impair the activation of cofilin. A third platelet protein, calpain, has been implicated in agonist-induced clustering of ␣ L ␤ 2 in T lymphocytes and increased avidity of lymphocyte adhesion (30). However, Fox et al. (41) reported that calpain activation in platelets requires ligand binding to ␣ IIb ␤ 3 ; and thus, it is downstream from ␣ IIb ␤ 3 activation. In summary, we have shown that interrupting actin filament turnover in platelets, either by capping actin filaments with Cyto-D or by sequestering actin monomers with Lat-A, enables a substantial proportion of the platelet integrin ␣ IIb ␤ 3 to bind soluble fibrinogen. Our results suggest that in unstimulated platelets, actin or actin-associated proteins such as talin or ␣-actinin in the membrane skeleton constrain ␣ IIb ␤ 3 in a low affinity state, perhaps by binding directly to the ␣ IIb ␤ 3 cytoplasmic tails (42,43) or by limiting access of ␣ IIb ␤ 3 -activating proteins such as ␤ 3 -endonexin (44) or CIB (45). We propose that platelet agonists initiate actin filament turnover in the membrane skeleton, perhaps by activating such actin disassembly factors as gelsolin and cofilin. In turn, the turnover of actin filaments relieves the constraint on ␣ IIb ␤ 3 , allowing it to assume the high affinity conformation required for fibrinogen binding and platelet aggregation. The biochemical reactions that control actin assembly and disassembly in platelets may be relevant pharmacologic targets for regulating ␣ IIb ␤ 3 function in vivo.