Essential role for phosphoinositide 3-kinase in shear-dependent signaling between platelet glycoprotein Ib/V/IX and integrin α_IIbβ₃

Platelet adhesion and aggregation at sites of blood vessel injury are essential for the arrest of bleeding and for the maintenance of vascular integrity. Under conditions of rapid blood flow, the formation of the primary hemostatic plug requires the synergistic contribution of multiple platelet receptor-ligand interactions,1,2 foremost of which involves the sequential binding of von Willebrand factor (VWF) to platelet adhesion receptors, glycoprotein (GP) Ib/V/IX and integrin α_IIbβ₃. The VWF-GP Ib/V/IX interaction is the first step in the hemostatic process that tethers platelets to the site of vascular injury. A characteristic feature of this adhesive interaction is its rapid reversibility such that platelets translocate (roll) on immobilized VWF. During surface translocation, GP Ib/V/IX transduces signals (outside-in signaling) to induce platelet activation, converting integrin α_IIbβ₃ from a low- to a high-affinity state (inside-out signaling) capable of engaging the C1 domain of VWF.3-5 Recent in vivo studies in mice lacking VWF or fibrinogen or both have demonstrated the central role for VWF in promoting platelet-vessel wall and platelet-platelet adhesive interactions during thrombus growth.6 However, despite its fundamental importance, the mechanisms linking the VWF-GP Ib/V/IX interaction to activation of integrin α_IIbβ₃, particularly under shear conditions, remain poorly defined.

An important signaling pathway operating in all mammalian cells involves the activation of one or more members of the phosphoinositide (PI) 3–kinase family. These enzymes phosphorylate membrane inositol phospholipids at the 3-OH position and have been classified into 3 distinct classes based on their primary structure and in vitro lipid substrate specificity.7,8 It is likely that all mammalian cells express representatives of each of the 3 types of PI 3–kinases; however, only the type 1 PI 3–kinases are capable of phosphorylating all 3 conventional lipids—PtdIns, PtdIns(4)P, PtdIns(4,5)P₂—in vitro, generating PtdIns(3)P, PtdIns(3,4)P₂ and PtdIns(3,4,5)P₃, respectively. PtdIns(3,4)P₂ and PtdIns(3,4,5)P₃are considered the primary output signals of type 1 PI 3–kinases in vivo, and have been implicated in regulating a diverse range of cellular processes, including cell growth, prevention of apoptosis, glucose transport, and cytoskeletal reorganization.

A great deal of information on the function of type 1 PI 3–kinases in cells has been based on the use of the structurally distinct pharmacologic inhibitors, LY294002 and wortmannin. These inhibitors have been used to study the role of type 1 PI 3–kinases in platelets and have demonstrated a potentially important role for these enzymes in regulating the adhesive and signaling function of integrin α_IIbβ_3.9 For example, evidence from the study of Glanzmann thrombasthenic platelets (congenitally deficient in integrin α_IIbβ₃) indicates that the generation of the PI 3–kinase lipid product, PtdIns(3,4)P₂, is primarily dependent on signaling downstream of integrin α_IIbβ_3.10 Furthermore, several integrin α_IIbβ₃–dependent functional responses, including platelet spreading11 and irreversible platelet aggregation,12 are dependent on PI 3–kinases. Although these studies have defined an important role for type 1 PI 3–kinases in integrin α_IIbβ₃ signaling (inside-out signaling) there is less convincing evidence for an absolute requirement for PI 3–kinases in inducing activation of integrin α_IIbβ₃ activation (outside-in signaling) by the majority of physiologic platelet stimuli studied to date.13 

A potential role for type 1 PI 3–kinases in GP Ib/V/IX signaling has been suggested by the observation that VWF binding to GP Ib/V/IX can induce the cytoskeletal association and activation of p85/p110 form of type 1 PI 3–kinase.14 More recently, VWF stimulation of platelets was shown to initiate complex formation between the p85 subunit of PI 3–kinase and GP Ib/V/IX.15 In this report we have investigated the potential involvement of type 1 PI 3–kinases in shear-dependent signaling between GP Ib/V/IX and integrin α_IIbβ₃.

LY294002, EGTA-am, and adenosine 5′-O-(1-Thio-triphosphate) (ATPαS) were from Calbiochem (La Jolla, CA). Adenosine 3′-phosphate 5′-phosphosulfate (A3P5PS) was purchased from Sigma (St Louis, MO). Wortmannin was purchased from Sapphire Bioscience (Crow's Nest, New South Wales, Australia). Oregon Green 488 BAPTA-1, am and Fura Red, am were from Molecular Probes (Eugene, OR). DiOC₆ was obtained from Sigma. Apyrase was purified from potatoes according to the method of Molnar and Lorand.16 Human VWF was purified to homogeneity from plasma cryoprecipitate according to the method of Montgomery and Zimmerman.17 AR-C69931MX was generously supplied by Astrazeneca R & D Charnwood (Leicestershire, England). All other reagents were obtained from sources described previously.14,18,19 

PAC-1 monoclonal antibody was from Becton Dickinson (Victoria, Australia) and fluorescein isothiocyanate (FITC)–conjugated antimouse IgM (α-IgM) antibody was from Southern Biotechnology Associates (Birmingham, AL).

For washed platelet and platelet reconstitution studies, blood was collected from healthy volunteers who had not received any antiplatelet medication in the preceding 2 weeks. Platelets were isolated and washed as described previously18 and resuspended in modified Tyrode buffer (10 mM Hepes, 12 mM NaHCO₃, pH 7.4, 137 mM NaCl, 2.7 mM KCl, 5 mM glucose), containing 1 mM CaCl₂ or 1 mM MgCl₂ or both where indicated. Autologous red blood cells were obtained by an initial centrifugation of anticoagulated whole blood at 200g for 30 minutes. The platelet-rich plasma was removed and red blood cells were washed 3 times with washing buffer (10 mM Hepes, pH 7.4, 140 mM NaCl, 5 mM glucose). Plasma was obtained from centrifugation of anticoagulated blood (15 mM trisodium citrate, pH 7.4) at 2000g for 10 minutes. Washed platelets were pretreated with vehicle alone (dimethyl sulfoxide [DMSO]), LY294002 (0-25 μM), or wortmannin (0-100 nM) for 15 minutes, or alternatively pretreated with apyrase (up to 16.5 U/mL) or the adenosine diphosphate (ADP) receptor antagonists, AR-C69931MX (100 nM) and A3P5PS (200 μM), and ATPαS (50-100 μM) for 30 minutes before static and flow-based assays were performed (150 s^-1 or 1800 s^-1) according to a modified method of Yuan et al20 and Cooke et al,21respectively. In control studies, we demonstrated that the concentrations of apyrase, ATPαS, AR-C69931MX, and A3P5PS used in our experimental assays abolished platelet aggregation or platelet shape change in washed platelets induced by 10 μM exogenous ADP. In some studies, washed platelets were reconstituted with either red blood cells alone (50% [vol/vol] autologous packed red blood cells), in the presence of 0.4 U/mL apyrase (ADPase activity), or red blood cells and plasma prior to perfusion through VWF-coated microcapillary tubes. Analysis of platelet adhesion and surface area was performed as described by Yap et al22 and expressed as a percent of control. For whole blood studies, anticoagulated blood (15 mM trisodium citrate, pH 7.4) was pretreated with vehicle alone (DMSO), LY294002 (0-200 μM), apyrase (8.25 U/mL), ATPαS (100 μM), or AR-C69931MX (100 nM) and A3P5PS (200 μM) for 10 minutes before perfusion through VWF-coated microcapillary tubes at 1800 s^-1. In control studies, we confirmed that the concentrations of apyrase, ATPαS, AR-C69931MX, and A3P5PS used in our experimental assays prevented platelet aggregation in whole blood induced by 10 μM exogenous ADP using a sensitive single-platelet detection assay.

In static adhesion assays, washed platelets were allowed to adhere to VWF-coated coverslips in the presence of PAC-1 antibody (1 μg/mL). In flow adhesion assays, washed platelets reconstituted with red blood cells were perfused through microcapillary tubes for 5 minutes prior to the perfusion of PAC-1 (1 μg/mL) over adherent platelets for 15 minutes. Adherent platelets were fixed, incubated with a FITC-conjugated anti-IgM antibody, and visualized using confocal microscopy (× 63W; Leica TCS SP; Leica, Heidelberg, Germany). For the imaging of platelet thrombi, adherent platelets were fixed and stained with DiOC₆ (1 μM). Thrombi were imaged using fluorescent confocal microscopy and reconstructed in Voxblast (Vaytek, Fairfield, IA).

Following formation of thrombi, red cells were lysed by perfusing 1% ammonium oxalate through the microcapillary tubes. Adherent platelets were then lysed with 0.1% Triton lysis buffer and lactate dehydrogenase (LDH) levels determined using the COBAS MIRA S Chemistry System (Roche, Somerville, NJ). This quantitative method was validated independently by examining thrombus dimensions by confocal imaging, according to the method of Kulkarni et al.23 

Platelets were labeled with inorganic ³²P (11.1 MBq [0.3 mCi/mL]) for 2 hours at 37°C and the level of phospholipids determined as described by Carter et al.24Platelets in suspension (resting) or adherent to VWF were lysed with radioimmunoprecipitation assay (RIPA) buffer (10 mM Tris HCl, pH 7.2, 1% Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS], 158 mM NaCl), and phospholipids extracted with a chloroform/methanol-based mixture (100 μL chloroform/methanol, 1:2, 75 μL 3.1 M HCl, 100 μL chloroform). Following deacylation with methylamine/butanol/methanol (42:9:47) phospholipids were identified by SAX high-performance liquid chromatography (HPLC) analysis using³H-labeled commercial standards.

Adherent platelets were fixed with 2% glutaraldehyde in 100 mM Na₂HPO₄/NaH₂PO₄, pH 7.4, for 60 minutes then incubated with 1% OsO₄ in 100 mM Na₂HPO₄/NaH₂PO₄, pH 7.4, for 30 minutes. The fixed platelets were dehydrated by successive immersions in increasing concentrations of ethanol followed by critical point drying. The coverslips or microcapillary tubes, which had been dissected laterally, were mounted on scanning electron microscopy stubs and coated with gold prior to imaging using an Hitachi S570 scanning electron microscope.

Analysis of intracellular calcium fluxes was performed according to the method of Yap et al.22 Briefly, platelets were loaded with calcium indicator dyes, Oregon Green 488 BAPTA-1,am (1 μM), and Fura Red, am (1.25 μM) for 30 minutes at 37°C, then washed and resuspended in Tyrode buffer containing either extracellular calcium (1 mM) or EGTA/Mg⁺⁺(1 mM each), prior to incubation with PI 3–kinase inhibitors. Changes in the concentration of cytosolic calcium were derived from the ratio of signal intensity in the Oregon Green and Fura Red channels using confocal microscopy. The calcium dynamics in individual platelets were monitored every 0.586 seconds over a 73.25-second or 37.5-second time interval for static or flow experiments, respectively, and recorded for off-line single cell analysis as described previously.22The number of platelets demonstrating oscillatory calcium transients over the time period examined was determined. Platelet translocation was determined based on the displacement of a tethered platelet from its original point over a 37.5-second period.

Significant differences were detected using Student ttest and one-way ANOVA, using the Prism software package (Graphpad Software for Science, San Diego, CA).

To investigate the potential involvement of type 1 PI 3–kinases in regulating platelet adhesion, integrin α_IIbβ₃ activation and spreading on immobilized VWF, washed platelets were pretreated with the well-characterized pharmacologic PI 3–kinase inhibitors, LY294002 or wortmannin. As demonstrated in Figure 1A, wortmannin or LY294002 exhibited a modest dose-dependent inhibitory effect on platelet adhesion under static conditions, with a maximal inhibition of 22% ± 3.45% (P < .01) using LY294002 (25 μM) and 30% ± 4.81% (P < .01) with wortmannin (100 nM). These inhibitors did not, however, inhibit integrin α_IIbβ₃ activation, as assessed by binding of the activation-specific monoclonal antibody, PAC-1 (Figure 1B). In control studies, we demonstrated that integrin α_IIbβ₃ activation under these experimental conditions was not dependent on released ADP, because pretreating platelets with apyrase (16.5 U/mL; Figure 1B) or with the ADP receptor antagonists, AR-C69931MX (α-P2Y₁₂) and A3P5PS (α-P2Y₁) or ATPαS (α-P2Y₁₂ and α-P2Y₁) (data not shown) did not prevent PAC-1 binding. Examination of platelet morphology demonstrated that pretreatment of platelets with wortmannin (0-100 nM) or LY294002 (0-25 μM) resulted in a minor, yet significant (P < .05), effect on the ability of cells to spread (∼11% reduction in platelet surface area in LY294002- and wortmannin-treated platelets [Figure2A]). High-resolution imaging of spread platelets demonstrated that a significant proportion (> 50%) of LY294002- and wortmannin-treated cells exhibited irregular cell margins and had not fully spread (Figure 2B). Real-time analysis of platelet spreading demonstrated an important role for PI 3–kinase in regulating the rate of platelet spreading. For example, control platelets adhering to VWF extended dynamic filopodial projections that extended and retracted from the cell surface. The formation of these projections was closely followed by the extension of lamellipodial sheets, resulting in full platelet spreading over 10 to 15 minutes (Figure 2C). Pretreating platelets with LY294002 (25 μM) or wortmannin (100 nM) slightly delayed the onset of filopodial formation (typically by < 3 minutes; Figure 2C). In contrast to cells pretreated with vehicle, there was a consistent delay in the subsequent formation of lamellipodial sheets (10-15 minutes), such that platelet spreading typically occurred over 20 to 30 minutes.

To investigate whether the inability of PI 3–kinase inhibitors to prevent integrin α_IIbβ₃ activation and platelet spreading was due to incomplete inhibition of PI 3–kinase, we examined the effects of these inhibitors on the generation of PI 3–kinase lipid products. As demonstrated in Figure3A, PtdIns(3,4)P₂ was undetectable in resting platelets. However, this lipid increased to at least 0.15% of total PI levels following platelet spreading on immobilized VWF (Figure 3A). This increase in PtdIns(3,4)P₂is similar to that previously reported for thrombin-stimulated platelets and platelets spread on fibrinogen.11,25 In contrast, the level of PtdIns(3)P was unaltered in these cells and PtdIns(3,4,5)P₃ was not detected (data not shown). Pretreatment of platelets with either LY294002 or wortmannin led to a dose-dependent inhibition of PtdIns(3,4)P₂ production in spreading platelets, with a maximal inhibition of about 95% and 100% using LY294002 (25 μM) and wortmannin (100 nM), respectively (Figure3A,B). The effects of these inhibitors were specific in that they did not inhibit the levels of PtdIns, PtdIns(4)P, and PtdIns(4,5)P₂ (Figure 3C). Further evidence that LY294002 (25 μM) and wortmannin (100 nM) effectively inhibited PI 3–kinase was obtained from studies demonstrating that both inhibitors prevented integrin α_IIbβ₃ activation and spreading of platelets on a fibrinogen matrix (Figure 3D). In addition, both inhibitors completely eliminated other PI 3–kinase–dependent events such as phosphorylation of the downstream PI 3–kinase target, PKB/Akt, and platelet aggregation induced by heat-aggregated IgG and CD-9 (data not shown).

Our recent studies have demonstrated that VWF-induced integrin α_IIbβ₃ activation is critically dependent on intracellular calcium mobilization.22 To examine the relationship between inhibition of PI 3–kinase and calcium release, cytosolic calcium levels were monitored during platelet spreading on VWF. As demonstrated in Figure 4, panels A and B, resting platelets exhibited a cytosolic calcium level below 50 nM, which did not undergo oscillations. Platelets firmly adherent to VWF underwent dynamic calcium oscillations in about 60% of cells (Figure 4A). The range of calcium concentrations varied widely between individual cells, with calcium oscillations typically fluctuating between 150 and 800 nM.22 Pretreatment of platelets with wortmannin (100 nM) delayed the onset of the oscillatory calcium transients by 5 to 10 minutes (Figure 4A); however, they did not affect the final magnitude of the response (Figure 4B). Similar results were obtained with LY294002 (25 μM; Figure 4A). The inhibitory effect of LY294002 and wortmannin on the initiation of the oscillatory calcium transients was consistent with the delayed spreading response in these cells. Because PI 3–kinase has been implicated in regulating calcium influx as well as the release of intracellular stores, we investigated the effect of LY294002 and wortmannin on intracellular calcium mobilization. Consistent with our recent findings,22chelating extracellular calcium with the calcium chelator, EGTA, did not prevent oscillatory calcium transients, but had a major inhibitory effect (∼75%) on the magnitude of the calcium response. Pretreatment of EGTA-treated platelets with LY294002 or wortmannin delayed the onset of oscillatory calcium transients by 5 to 10 minutes; however, the final magnitude of these oscillatory calcium responses was unaffected by these inhibitors (data not shown), suggesting that the primary function of PI 3–kinase is to regulate intracellular calcium mobilization.

To investigate the functional importance of PI 3–kinase under shear conditions, we examined the effect of LY294002 and wortmannin on platelet adhesion and spreading on immobilized VWF using an in vitro flow-based adhesion assay. In contrast to adhesion assays performed under static conditions, inhibition of PI 3–kinase with either wortmannin or LY294002 profoundly affected the ability of platelets to adhere to a VWF matrix under flow. As demonstrated in Figure5A, under low shear conditions (150 s^-1) pretreating platelets with LY294002 or wortmannin inhibited platelet adhesion in a dose-dependent manner, with maximal inhibition of about 70% (P < .05). Under high shear conditions (1800 s^-1), inhibiting PI 3–kinase reduced platelet adhesion by up to 90% (P < .001). Real-time analysis of platelet adhesion under flow revealed that LY294002 or wortmannin did not affect the ability of cells to tether or translocate on the VWF matrix but dramatically reduced the ability of these cells to form stationary adhesion contacts (data not shown). This reduction in stationary adhesion correlated with more than 95% reduction in PAC-1 binding (Figure 5D). Continuous observation of translocating platelets for prolonged periods (20 minutes) demonstrated that these cells were unable to form stable adhesion contacts at all times points examined (data not shown). Inhibition of PI 3–kinase had no effect on the ability of platelets to change shape and extend filopodia on immobilized VWF; however, lamellipodial formation was completely inhibited (Figure 5B,C). In further control studies, we confirmed that ADP was not essential for stationary platelet adhesion under high shear conditions, because platelets pretreated with the ADP receptor antagonists, ATPαS (100 μM) or AR-C69931MX/A3P5PS (100 nM and 200 μM, respectively), or with apyrase (8.25 U/mL) adhered normally to the VWF matrix (data not shown).

Analysis of cytosolic calcium during shear-dependent platelet adhesion demonstrated a critical role for PI 3–kinase in this process. As demonstrated in Figure 6A, pretreatment of platelets with wortmannin (100 nM) in the presence of EGTA abolished sustained calcium oscillations in more than 90% of platelets relative to control platelets. Whereas control platelets underwent strong oscillatory calcium transients and formed stationary adhesion contacts, wortmannin-treated platelets (100 nM) translocated continuously and exhibited minor calcium oscillations (Figure 6B,C). Similar results were obtained with LY294002 (data not shown). These studies demonstrate an important requirement for PI 3–kinase in promoting calcium mobilization under flow.

To investigate the potential importance of PI 3–kinase in regulating thrombus formation under experimental conditions more closely simulating those experienced by platelets in vivo, flow studies were performed on anticoagulated whole blood. As demonstrated in Figure7A, perfusing anticoagulated whole blood through VWF-coated microcapillary tubes (1800 s^-1) for 5 minutes resulted in the formation of platelet-rich thrombi. Analysis of these thrombi by confocal imaging demonstrated that they covered approximately 30% to 40% of the VWF-coated surface and had a height of 7 to 10 μm (Figure 7A). In contrast, pretreating whole blood with increasing concentrations of LY294002 resulted in a dose-dependent inhibition of thrombus growth (Figure 7A,B). Real-time analysis of LY294002-treated platelets (50-200 μM) revealed that the majority of tethered cells rolled along the VWF matrix and did not form stationary adhesion contacts, similar to that observed with washed platelets (data not shown). It should be noted that the requirement for higher LY294002 concentrations to inhibit platelet adhesion in whole blood is presumably due to plasma protein binding to LY294002, because a 5- to 10-fold increase in concentration of this reagent was required to inhibit CD-9–induced platelet aggregation in platelet-rich plasma relative to washed platelets (data not shown). Further evidence that PI 3–kinase was essential for platelet thrombus formation on VWF was derived from studies using wortmannin. In these studies, washed platelets were initially pretreated with wortmannin (100 nM) to irreversibly inhibit PI 3–kinase. The cells were subsequently reconstituted with red blood cells and plasma and examined for their ability to form stationary adhesion contacts on immobilized VWF. Similar to that observed with LY294002, wortmannin completely inhibited the ability of platelets to form stationary adhesion contacts and thrombi on VWF (data not shown).

The results presented here demonstrate for the first time a key signaling role for type 1 PI 3–kinases in linking the VWF-GP Ib/V/IX interaction to integrin α_IIbβ₃activation under physiologic flow conditions. This conclusion is primarily based on studies using the pharmacologic inhibitors, LY294002 and wortmannin. Wortmannin, a fungal metabolite that irreversibly inhibits the catalytic function of PI 3–kinase, has been well characterized to specifically inhibit PI 3–kinase at concentrations up to 100 nM.26 At higher concentrations, wortmannin also inhibits PI 4-kinase, myosin light-chain kinase, and phospholipase A₂. LY294002, a quercetin analogue that reversibly inhibits PI 3–kinase, has no inhibitory effects on these enzymes and has generally been regarded as a selective PI 3–kinase inhibitor at concentrations up to 25 μM.27 However, a recent study has demonstrated that casein kinase 2 (CK2), a serine-threonine kinase involved in promoting cell growth, is also inhibited by LY294002 at low micromolar concentrations.28 CK2 is not, however, inhibited by wortmannin at concentrations as high as 1 μM, suggesting that the inhibitory effects of LY294002 in our studies are unlikely to be primarily due to effects on CK2. To further strengthen our hypothesis that PI 3–kinase is the responsible enzyme for shear-dependent platelet adhesion, we have performed dose-response studies correlating the inhibitory effects on platelet adhesion with the cellular levels of PtdIns(3,4)P₂. In all studies we observed an excellent correlation between these events, thereby providing strong evidence for a key role for type 1 PI 3–kinases in this process.

Our observations that PI 3–kinase inhibitors did not block platelet spreading on VWF under static conditions was somewhat surprising given previous reports demonstrating an absolute requirement for this enzyme in promoting platelet spreading on a fibrinogen matrix.11 Several lines of evidence suggest that this difference was not due to incomplete inhibition of PI 3–kinase, but, in fact, reflects a specific difference between platelet spreading on fibrinogen versus VWF. First, we have demonstrated that both LY294002 and wortmannin effectively inhibited generation of the PI 3–kinase lipid product, PtdIns(3,4)P₂. Furthermore, inhibition of PI 3–kinase activity with both inhibitors completely blocked PKB/Akt phosphorylation (unpublished observations, February 2001). Second, we demonstrated that both LY294002 and wortmannin completely inhibited platelet spreading on immobilized fibrinogen under identical conditions to those used in our spreading studies on VWF. Finally, both inhibitors abolished other PI 3–kinase–dependent processes, including platelet aggregation induced by heat-aggregated IgG or anti-CD9 antibodies.

Our findings demonstrating an important role for PI 3–kinase in regulating calcium mobilization in adherent platelets supports a growing body of evidence for an important link between PI 3–kinase signaling and calcium fluxes. PI 3–kinase has been demonstrated to regulate both calcium influx and mobilization, although the extent of its involvement in these processes appears to be specific for both cell type and stimulus. A role for PI 3–kinase, in particular its PtdIns(3,4,5)P₃ lipid product, in calcium influx has been demonstrated through studies in platelets and T cells. For example, platelets deficient in src homology 2 (SH2)–containing inositol 5′ phosphatase (SHIP) exhibited a specific increase in the levels of PtdIns(3,4,5)P₃, resulting in enhanced transmembrane calcium flux.29 Similarly, addition of exogenous PtdIns(3,4,5)P₃ to T cells promotes calcium influx, without affecting intracellular calcium mobilization.30 In other cell types, including platelet-derived growth factor (PDGF)–stimulated NIH 3T3 cells31 and COS-1 cells,32 and antigen-stimulated B cells,33 inhibiting PI 3–kinase primarily leads to a decrease in IP₃ generation and calcium release from intracellular stores. PtdIns(3,4,5)P₃ has been implicated in this process because microinjection of pleckstrin homology (PH) domains that specifically bind this lipid decrease IP₃ formation and calcium mobilization in COS-1 cells32 and megakaryocytes,34 respectively. Although we were unable to detect a rise in PtdIns(3,4,5)P₃levels following platelet spreading on VWF, the level of increase in PtdIns(3,4)P₂ reported here is in agreement with that demonstrated for platelets adherent and spread on immobilized fibrinogen.11 The most likely explanation for only observing a rise in PtdIns(3,4)P₂ in our studies is that in vitro generation of PtdIns(3,4,5)P₃ occurs more rapidly and precedes the appearance of PtdIns(3,4)P_2,35suggesting that PtdIns(3,4,5)P₃ is likely to be detected at an earlier time point.

In all studies to date, PI 3–kinase has been demonstrated to primarily serve as a modulator of phospholipase Cγ (PLCγ) activation and IP₃ formation and was not found to be essential for calcium release. In contrast, in our studies PI 3–kinase appears to have an essential role for sustained calcium oscillations under shear conditions. Moreover, the effects of PI 3–kinase inhibitors on the time course and magnitude of the calcium responses in our static assays appears distinct from that previously described. For example, we have demonstrated that inhibiting PI 3–kinase prolongs the onset of calcium mobilization, but does not have any effect on the final magnitude of the calcium response. Thus, PI 3–kinase appears to be essential for inducing the “early” release of calcium from internal stores and does not appear to be required for subsequent calcium influx. These observations raise the possibility that PI 3–kinase inhibitors prevent the formation of irreversible platelet adhesion under shear conditions, because this “early” calcium response may be essential for activation of integrin α_IIbβ₃. In this context, it is of interest that prolonged interaction of translocating platelets with the VWF matrix (∼20 minutes) did not result in stationary platelet adhesion. It is therefore possible that PI 3–kinase participates in a shear-specific signaling pathway. This latter possibility would be in keeping with previous studies in endothelial cells in which shear-induced nitric oxide production was dependent on the activation of PI 3–kinase.36,37 

A key outstanding issue is the mechanism by which the VWF-GP Ib interaction induces PI 3–kinase activation. Previous studies examining shear-induced platelet activation have suggested that platelet activation induced by the VWF–GP Ib interaction occurs indirectly through the release of ADP.38-40 Dense granule ADP has a well-defined role in promoting platelet activation by a number of platelet agonists and has been demonstrated to induce PI 3–kinase activation.41 However, we do not believe that this is the major mechanism for PI 3–kinase activation during platelet adhesion to VWF, because ADP receptor antagonists do not prevent activation of integrin α_IIbβ₃ under these experimental conditions. A recent study by Canobbio et al42 has suggested that VWF-induced platelet activation is critically dependent on FcγRIIA and thromboxane₂ (TXA₂) generation. FcγRIIA has previously been postulated to be physically and functionally linked to GP Ib/V/IX and activates platelets through a PI 3–kinase–dependent mechanism.43-45 However, this signaling pathway appears distinct from those operating during platelet adhesion to VWF under flow22 and during shear-induced platelet aggregation,39 because VWF-dependent platelet activation under these conditions is insensitive to the inhibitory effects of aspirin.

A number of recent studies from several independent laboratories have suggested that GP Ib/V/IX can signal directly to regulate the affinity status of integrin α_IIbβ₃ independent of ADP, TXA₂, and Fc receptors.22,46,47 The precise mechanism by which GP Ib signals has not been elucidated, although one potential mechanism for PI 3–kinase activation involves direct binding to the 14.3.3ζ adapter protein. The 14.3.3ζ protein is physically linked to the cytoplasmic tail of GP Ibα and has recently been suggested to be important for GP Ib/V/IX-induced integrin α_IIbβ₃ activation, based on studies of Chinese hamster ovary (CHO) cell lines expressing GP Ib/IX and integrin α_IIbβ₃.47 Although there is evidence for complex formation between 14.3.3ζ and the p85/p110 form of PI 3–kinase in T cells,48 there is as yet no evidence that this association promotes kinase activation. A recent study by Munday et al15 has suggested a direct association between GP Ib/V/IX and PI 3–kinase; however, the significance of this association, given its low stoichiometry, remains unclear. It is possible that PI 3–kinase activation in VWF-stimulated platelets is in part regulated through integrin α_IIbβ₃outside-in signaling. Previous studies have highlighted the synergistic contribution of GP Ib/V/IX and integrin α_IIbβ₃ to cytoskeletal signaling complex formation18,49 and have also demonstrated that direct ligand binding to integrin α_IIbβ₃ is sufficient to induce PI 3–kinase activation and a selective increase in the cellular levels of PtdIns(3,4)P₂.12These observations, combined with the demonstration that ligand binding to integrin α_IIbβ₃ induces intracellular calcium mobilization, raises the interesting possibility that integrin α_IIbβ₃-dependent PI 3–kinase activation may play an important role in promoting calcium flux in VWF-stimulated platelets. Future studies are currently under way to examine the relative contribution of GP Ib/V/IX and integrin α_IIbβ₃ to PI 3–kinase activation and calcium mobilization in VWF-stimulated platelets.

In conclusion, our studies demonstrate for the first time an essential role for type 1 PI 3–kinases in regulating the earliest steps of the hemostatic process, namely, the shear-induced activation of integrin α_IIbβ₃ following platelet adhesion to VWF. In this context, it is of interest that wortmannin was originally described as a hemorrhagic factor,50,51 a finding that has not been easily reconciled with in vitro studies demonstrating a modest effect of wortmannin on platelet activation induced by the majority of physiologic agonists studied to date.9 Although it remains to be established whether the bleeding diathesis induced by wortmannin is primarily due to effects on platelets, and in particular GP Ib/V/IX signaling, our findings nonetheless provide important new insight into the functional role of PI 3–kinase in regulating the hemostatic function of platelets.

We would like to thank Suhasini Kulkarni for technical assistance.