Distinct glycoprotein Ib/V/IX and integrin αIIbβ3-dependent calcium signals cooperatively regulate platelet adhesion under flow

We have investigated the calcium signaling relationship between the two major platelet adhesion receptors, glycoprotein Ib/V/IX (GPIb/V/IX) and integrin αIIbβ3, involved in regulating platelet adhesion on von Willebrand factor (vWf) under flow. Our studies demonstrate that GPIb engagement of immobilized vWf elicits a transient calcium spike that may function to promote reversible arrest of translocating platelets. Subsequent integrin αIIbβ3 engagement of vWf promotes sustained calcium oscillations that are essential for the maintenance of irreversible adhesion. GPIb-induced calcium spikes appear distinct from those initiated by integrin αIIbβ3, in that the former are exclusively mediated through release of intracellular calcium stores via a signaling mechanism independent of PI 3-kinase. In contrast, integrin αIIbβ3-dependent calcium flux involves a PI 3-kinase-dependent signaling mechanism linked to intracellular calcium mobilization and subsequent transmembrane calcium influx. Studies employing the caged calcium chelator (o-nitrophenyl-EGTA) demonstrate that transient calcium spikes initiate a transient phase of platelet arrest that is converted to irreversible adhesion with the development of sustained oscillatory calcium flux. These studies demonstrate the existence of a dual step calcium signaling mechanism utilized by GPIb and integrin αIIbβ3 that serves to regulate the dynamics of platelet adhesion under flow.

The platelet is a specialized adhesive cell that plays a central role in the normal blood clotting process through its ability to rapidly adhere to subendothelial matrix proteins and to other activated platelets at sites of vascular injury. Platelet adhesion is a multistep process requiring an initial cell-tethering step, involving interaction between the glycoprotein Ib/V/IX (GPIb/ V/IX) 1 receptor complex with subendothelial von Willebrand factor (vWf) and a firm adhesion step mediated by one or more platelet integrins (1,2). Integrins themselves are generally inefficient at initiating platelet adhesion under conditions of rapid blood flow, due to their slow intrinsic binding kinetics. In contrast, the rapid formation and dissolution of bonds between the vWf A1 domain and GP Ib␣ supports efficient platelet tethering and translocation (rolling) under conditions of high shear (3). Understanding the mechanisms regulating platelet translocation is potentially important, since recent studies have demonstrated that the majority of platelets adhering to the injured vessel wall and to the surface of thrombi in vivo undergo a variable period of surface translocation prior to forming stationary adhesion contacts (4,5).
There is a considerable body of evidence demonstrating that in addition to mediating platelet tethering and translocation, GPIb transduces signals necessary for integrin ␣ IIb ␤ 3 activation. The mechanism by which GPIb transduces signals remains controversial and appears to be significantly influenced by the experimental conditions used to study this process. For example, based on studies of shear-induced platelet aggregation using a cone-and-plate viscometer, GPIb has been proposed to initiate platelet activation by inducing transmembrane calcium influx, leading to integrin ␣ IIb ␤ 3 activation through an indirect mechanism dependent on released ADP (6). In contrast, recent studies examining platelet adhesion to an immobilized vWf matrix have demonstrated the existence of a distinct GPIb signaling mechanism linked to intracellular calcium mobilization (7). GPIb-dependent calcium mobilization is a shear-sensitive signaling process promoting integrin ␣ IIb ␤ 3 activation directly, independent of released ADP.
An important unresolved issue is the relative contribution of GPIb and integrin ␣ IIb ␤ 3 outside-in signaling to cytosolic calcium flux initiated by the platelet-vWf interaction. For example, a recent study by Kuwahara et al. has suggested that calcium flux during shear-dependent platelet adhesion on vWf is exclusively mediated through integrin ␣ IIb ␤ 3 (8). The involvement of integrins in the regulation of cytosolic calcium levels has been well established in a range of cell types, including platelets, leukocytes, endothelial cells, fibroblasts, and osteoclasts (9). In general, calcium signaling is utilized by ␤ 2 leukocyte integrins as well as integrins that engage ligands containing the tripeptide sequence Arg-Gly-Asp (RGD), including platelet ␣ IIb ␤ 3 , ␣ v ␤ 5 , and ␣ v ␤ 3 (9 -14) . However, the relationship between integrins and calcium signaling is complex and in most cell types incompletely understood. For example, in platelets there is evidence that ␣ IIb ␤ 3 plays an important role in regulating calcium homeostasis in the resting cell and promotes calcium influx following platelet activation (15,16). Other studies have suggested that ligand binding to ␣ IIb ␤ 3 can either induce or inhibit calcium mobilization, depending on the nature of the primary platelet-activating stimulus (17)(18)(19).
In this study, we have examined the relative contribution of GPIb and integrin ␣ IIb ␤ 3 in promoting cytosolic calcium flux during shear-dependent platelet adhesion on immobilized vWf. We demonstrate the existence of a dual step calcium signaling mechanism initiated by vWf engagement of GPIb and integrin ␣ IIb ␤ 3 that serves to regulate platelet translocation dynamics and firm platelet adhesion under flow.

EXPERIMENTAL PROCEDURES
Materials-5,5Ј-Dimethyl-BAPTA, AM; Oregon Green 488 BAPTA-1, AM; Fura Red, AM; and NP-EGTA were from Molecular Probes, Inc. (Eugene, OR). Apyrase was purified from potatoes according to the method of Molnar and Lorand (20) and was a generous gift from Dr. Francois Lanza. Human vWf was purified to homogeneity from plasma cryoprecipitate according to the methods of Montgomery and Zimmerman (21). LY294002 was from Calbiochem. Wortmannin was purchased from Sapphire Bioscience P/L (New South Wales, Australia). All other reagents were obtained from sources described previously (22,23).
Platelet Isolation and Reconstitution with Red Blood Cells-Anticoagulated whole blood (15 mM trisodium citrate, pH 7.4) was collected from healthy volunteers who had not received any anti-platelet medication in the preceding 2 weeks. Platelet isolation was carried out according to Yuan et al. (22), and washed platelets were finally resuspended in modified Tyrode's buffer (10 mM HEPES, 12 mM NaHCO 3 , pH 7.4, 137 mM NaCl, 2.7 mM KCl, 5 mM glucose) supplemented with 1 mM CaCl 2 /MgCl 2 or 1 mM EGTA/MgCl 2 where indicated. Autologous red blood cells were prepared according to Yap et al. (24) prior to reconstitution with isolated platelets (50% (v/v) hematocrit) in the presence of 0.4 units/ml apyrase (ADPase activity) and 1 unit/ml hirudin.
Analysis of Platelet Calcium Flux-Platelet calcium flux was monitored as previously described by Yap et al. (24)  Off-line Calcium Analysis-Calcium flux recordings were analyzed off-line using Leica Physiology Software (Leica TCS SP; Leica, Heidleberg, Germany), in the case of single cell recordings. Alternatively, platelet ⌬[Ca 2ϩ ] c was analyzed at a population level using MCID TM Image analysis software (Imaging Research Inc., Ontario, Canada). The first 3 min of platelet flow was captured as sequential 37.5-s series (0.586 frames per second capture rate) via confocal microscopy, and individual frames were analyzed for fluorescence pixel intensities at 5.86-s intervals following background subtraction of small pixel targets of less than 10-pixel diameters. Data analyzed in this way were pooled and presented as a time-averaged picture as either a scatter plot or relative frequency histogram. It should be noted that population analysis done in this way represents a frequency profile of cytosolic calcium events rather than single-cell calcium recordings.
Flow-based Platelet Adhesion Assays-Flow assays were performed using glass microcapillary tubes (Microslides, Vitro Dynamics Inc., NJ) coated with Human vWf (100 g/ml) according to a modified method of Cooke et al. (25). Oregon Green BAPTA-1/Fura Red loaded platelets were reconstituted with washed red blood cells (50% hematocrit) and Tyrode's supplemented with either 1 mM Ca 2ϩ /Mg 2ϩ or EGTA/Mg 2ϩ and perfused through vWf microcapillary tubes blocked with 10% heatinactivated human serum at a shear rate of 1,800⅐s Ϫ1 or 10,000⅐s Ϫ1 .
Measurement of Platelet Translocation Velocity-Platelet translocation velocity was determined in vitro according to a modified method of Savage et al. (1). Platelets were considered to have translocated when their spatial displacement was greater than one cell diameter. Platelet motion was analyzed off-line utilizing Leica confocal software. Briefly, a sequential stack of images over a 37.5-s time frame were projected using Leica acquisition software, and the xy coordinates of the centroid of individual platelets were determined every 0.576 s. Mean platelet translocation velocities were determined over the entire 37.5-s scan.
NP-EGTA "Caged Calcium" Assay-Isolated human platelets were loaded with Oregon Green BAPTA and Fura Red as per established protocol (24). Calcium dye loaded platelets were subsequently incubated in PWB with 10 M NP-EGTA (Molecular Probes) for 30 min at 37°C. NP-EGTA-treated platelets were subsequently washed once with PWB and resuspended in PWB prior to experimentation. The potential effects of the NP-EGTA precursor loading (prior to photolysis) on platelet function were assessed using standard aggregometry following stimulation of the cells with 1-12.5 M ADP or 0.1-1 units/ml thrombin. Incubation with 10 M NP-EGTA did not modify the aggregation of platelets to these agonists relative to untreated controls. The behavior of NP-EGTA-loaded platelets was assessed under flow conditions (1,800⅐s Ϫ1 ) at the surface of immobilized human vWf (100 g/ml) as per established protocols. NP-EGTA-loaded platelets exhibited translocation and adhesive properties equivalent to untreated controls. NP-EGTA uncaging was carried out following 18 s of reconstituted blood flow via exposure of platelets to a near UV (300 -400-nm) light source generated by a 100-watt mercury lamp directed through the optical path of a Leica DMIRBE confocal microscope, for an interval of 0.6 s. Control studies were carried out with unloaded control platelets and demonstrated that the brief (0.6-s) UV exposure did not lead to photodynamic damage or activation of the platelets under flow.

Cytosolic Calcium Flux Regulates Platelet Translocation Be-
havior-Recent studies examining shear-dependent platelet adhesion on immobilized vWf under flow have demonstrated a role for both the GPIb/V/IX and integrin ␣ IIb ␤ 3 receptor complexes in regulating cytosolic calcium flux (7,8). However, the temporal relationship and mechanisms by which these receptors regulate cytosolic calcium (⌬[Ca 2ϩ ] c ) remain incompletely understood. In this study, we have employed a confocal-based dual dye ratiometric Ca 2ϩ assay to accurately quantitate realtime cytosolic Ca 2ϩ flux during platelet adhesion under flow (24). In initial studies, Oregon Green BAPTA and Fura Redloaded human platelets were perfused through human vWfcoated (100 g/ml) microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 , and the relationship between cytosolic calcium flux and platelet translocation behavior was examined. Fig. 1A demonstrates that the platelet population exhibited a broad range of ⌬[Ca 2ϩ ] c , ranging from 0 to 1,200 nM, and translocation velocities between 0 and 20 m⅐s Ϫ1 , similar to those reported by Savage et al. (1). Detailed single-cell analysis revealed that platelets exhibit three broad calcium response subclasses that differ with respect to the magnitude and dynamics of their calcium flux. The first subclass consists of platelets with a relatively low cytosolic calcium content (⌬[Ca 2ϩ ] c Ͻ20 nM). These platelets were characterized by a rapid rate of translocation across the vWF surface and minimal calcium oscillations (Fig. 1B). The second subclass exhibited a moderately elevated (intermediate) ⌬[Ca 2ϩ ] c , ranging from 20 to 65 nM, which underwent minor oscillations; these platelets displayed a reduced rate of translocation that was stop-start in nature (Fig. 1B). The third and final subclass of platelets displayed elevated oscillatory ⌬[Ca 2ϩ ] c , ranging from 65 up to 1200 nM. The defining quality of platelets falling into this high range calcium response subclass was the pulsatile nature of the calcium flux, with cells undergoing rapid base-line to peak oscillations. The translocation behavior of these platelets was characterized by extended periods of stationary adhesion on the matrix surface (Fig. 1B).
To examine in more detail the correlation between ⌬[Ca 2ϩ ] c and translocation behavior, platelets were pretreated with the high affinity calcium chelator DM-BAPTA prior to perfusion through vWf-coated microcapillary tubes. DM-BAPTA treat-ment resulted in an approximately 10-fold increase in platelet translocation velocity, from a mean velocity of 6.2 m⅐s Ϫ1 for controls up to 61.4 m⅐s Ϫ1 for DM-BAPTA-treated platelets (Fig. 1A). In contrast, when the cytosolic concentration of calcium was artificially elevated, by pretreating platelets with increasing concentrations of the sarco/endoplasmic reticulum Ca 2ϩ -ATPase inhibitor thapsigargin (Tg), there was a significant concentration-dependent decrease in mean platelet translocation velocity. Single cell analysis of thapsigargin-treated platelets revealed a close inverse correlation (R ϭ 0.94) between the cytosolic calcium level and the translocation velocity of individual platelets, with mean platelet translocation velocities approaching 0 m⅐s Ϫ1 as the mean ⌬[Ca 2ϩ ] c approached 100 nM (Fig. 1C). In control studies, we examined the potential role for ADP, thromboxane A 2 , or trace amounts of thrombin in platelet calcium signaling, by pretreating platelets with apyrase (2 units/ml), aspirin (1.5 mM), and/or hirudin (200 units/ml) prior to perfusion through vWf-coated microcapillary tubes. Consistent with previous findings (24), none of these inhibitors, either alone or in combination, modified the distribution of calcium events initiated by vWf engagement of GPIb and integrin ␣ IIb ␤ 3 at 1,800⅐s Ϫ1 (data not shown).
The Relative Roles of GPIb/V/IX and Integrin ␣ IIb ␤ 3 in Regulating Cytosolic Calcium Flux-Our recent studies have demonstrated that platelets forming irreversible adhesion contacts on a vWf matrix, under static or flow conditions, exhibit a sustained oscillatory calcium response (24). To investigate the relative roles of GPIb/V/IX and integrin ␣ IIb ␤ 3 in regulating these calcium changes, platelets were pretreated with vehicle, Aggrastat, or c7E3 Fab prior to performing adhesion studies on vWf. As demonstrated in Fig. 2A, platelets forming irreversible adhesion contacts with vWf under static conditions elicited sustained oscillatory calcium responses. However, blocking ligand binding to integrin ␣ IIb ␤ 3 completely abolished these sustained oscillations, with all cells exhibiting distinct calcium spikes ( Fig. 2A). Several lines of evidence demonstrate that these calcium spikes are elicited as a result of the vWf-GPIb interaction, rather than secondary to the release of ADP or thromboxane A 2 . First, pretreating platelets with apyrase or aspirin had no inhibitory effect on the generation of these calcium spikes (data not shown). Second, blocking ligand binding to GPIb abolished all calcium responses (data not shown). Third, pretreating platelets with ristocetin, a cationic modulator that increases the affinity of the vWf-GPIb interaction, increased the proportion of cells that elicited these isolated calcium spikes (Fig. 2C) but did not affect the amplitude of the Ca 2ϩ transients (mean ⌬[Ca 2ϩ ] c ϭ 305 Ϯ 146 nM; n ϭ 40) (Fig.  2B). These studies suggest that even in the absence of shear, GPIb binding to vWf is sufficient to induce a transient Ca 2ϩ signal.
We have previously demonstrated that shear increases the proportion of platelets exhibiting a GPIb-dependent calcium signal (7). To investigate the effects of shear on the magnitude and duration of calcium signals initiated by GPIb/V/IX and integrin ␣ IIb ␤ 3 , control or Aggrastat-treated platelets were perfused through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . Platelets forming stationary adhesion contacts under high shear exhibited a sustained oscillatory calcium response similar in magnitude and duration to that observed under static conditions (data not shown). Integrin ␣ IIb ␤ 3 blockade under these conditions resulted in almost complete inhibition of high range ⌬[Ca 2ϩ ] c (Fig. 3A), with a concomitant (5fold) increase in platelet translocation velocity, from a mean of 6.2 m⅐s Ϫ1 in control platelets up to 34.1 m⅐s Ϫ1 in Aggrastattreated platelets (data not shown). To examine specifically the effects of shear on GPIb-derived calcium signals, independent C, washed human platelets were perfused through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 following pretreatment with 0 -100 nM Tg. Isolated platelets were incubated with Tg in calcium-free buffer at 37°C for 2 min prior to perfusion. Platelets were reconstituted with red blood cells (50% hematocrit) in Tyrode's buffer supplemented with 1 mM EGTA/Mg 2ϩ (n ϭ 3). of platelet translocation, Aggrastat-treated platelets were perfused through vWf-coated microcapillary tubes in the presence of ristocetin. Under these experimental conditions, the vWf-GPIb interaction sustains stationary platelet adhesion under flow (1800⅐s Ϫ1 ), independent of integrin ␣ IIb ␤ 3 (data not shown). In control studies, we demonstrated that the presence of ristocetin did not have any significant effect on the overall magnitude and duration of the sustained oscillatory calcium response initiated by vWf engagement of GPIb/V/IX and integrin ␣ IIb ␤ 3 (compare Fig. 3C with Fig. 2, A and B). Analysis of Aggrastat-treated platelets at 1,800⅐s Ϫ1 revealed that ϳ67% of adherent cells displayed discrete calcium transients or "spikes," that exhibited a mean ⌬[Ca 2ϩ ] c of 391 Ϯ 182.9 nM (maximum ⌬[Ca 2ϩ ] c ϭ 896 nM; n ϭ 41) (Fig. 3, B and C). Thus, regardless of the experimental conditions, the vWf-GPIb interaction appears to elicit transient Ca 2ϩ spikes that are distinct from the complex oscillatory response initiated by integrin ␣ IIb ␤ 3 engagement of vWf.
The Role of Transient Calcium Spikes in Regulating Platelet Adhesion under Flow-To investigate the potential importance of transient calcium spikes in regulating platelet adhesion dynamics under flow, we examined calcium response profiles of individual translocating platelets. Detailed analysis of platelets undergoing high range Ca 2ϩ responses under shear conditions revealed a subset of cells that exhibited elevated but transient calcium responses (Fig. 4A). As demonstrated in Fig.  4A, these transient calcium spikes coincided with a brief period of stationary adhesion, which was followed by a return to surface translocation following a decline in ⌬[Ca 2ϩ ] c toward 100 nM. In contrast, all platelets exhibiting sustained oscilla-

FIG. 2. GPIb/V/IX engagement of vWF elicits transient spikes in intraplatelet [Ca 2؉ ] c .
A, representative single platelet recordings demonstrating integrin ␣ IIb ␤ 3 (Control) and GPIb/V/IX (ϩAggrastat)dependent intraplatelet Ca 2ϩ flux at the surface of immobilized human vWf under static conditions. Calcium dye-loaded platelets were allowed to settle on the surface of vWF-coated coverslips (blocked with 10% heat-inactivated human serum) for 30 min at 37°C. Individual platelets underwent an oscillatory Ca 2ϩ response after adhering to the vWf substrate. Ca 2ϩ flux was monitored at 1-s intervals over the entire 30-min incubation period (100-s time frame presented). vWF, control cells at the surface of immobilized vWF; ϩAggrastat, platelets were treated with 200 nM Aggrastat for 10 min prior to assay. B, quantitation of the proportion (percentage) of adherent platelets undergoing defined Ca 2ϩ spikes in the absence of integrin ␣ IIb ␤ 3 engagement (ϩ200 nM Aggrastat) at the surface of immobilized vWF (n ϭ 3). C, platelets were allowed to adhere to immobilized vWF in the presence of 1 mg/ml ristocetin. ϩRistocetin, control platelets; ϩRistocetin ϩAggrastat, platelets treated with 200 nM Aggrastat prior to adhesion in the presence of 1 mg/ml ristocetin. tory calcium responses maintained stationary adhesion contacts with the vWf matrix (Fig. 4A).
To examine more directly the role of calcium spikes in initiating transient platelet arrest, we developed a caged-calcium platelet activation assay. NP-EGTA is a caged calcium chelator that displays a marked increase in its K d for Ca 2ϩ upon photolysis with near UV light (300 -400 nm). This reagent can effectively be used to release a relatively large concentration of free Ca 2ϩ in the cytosol within milliseconds of UV activation. In control studies, we demonstrated that loading platelets with 10 M NP-EGTA alone, independent of UV exposure, had no effect on the ability of platelets to adhere to thrombogenic surfaces or to aggregate in response to soluble agonist stimulation. How-ever, following brief exposure to UV irradiation, these cells rapidly released caged calcium (data not shown). To investigate the effects of transient calcium spikes on platelet adhesion under flow, NP-EGTA-loaded platelets were perfused through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . Exposure of these cells to UV light for a period of 0.6 s resulted in the generation of a rapid spike in ⌬[Ca 2ϩ ] c , approaching 700 -1200 nM (Fig. 4B). Examination of the translocation properties of these platelets demonstrated that stationary adhesion was tightly controlled by the onset of the calcium spikes (Fig.  4B). Platelet adhesion under these experimental conditions was mediated through integrin ␣ IIb ␤ 3 , since it was completely prevented by pretreating platelets with c7E3 Fab or Aggrastat (Fig. 4B). In control studies, we demonstrated that the effects of uncaged calcium on stationary platelet adhesion were likely to be direct rather than secondary to release of ADP or thromboxane A 2 , since pretreating platelets with apyrase or aspirin had no effect on UV light-induced platelet adhesion (data not shown).
To investigate further the relationship between sustained calcium oscillations and firm cell adhesion, NP-EGTA-loaded platelets were pretreated with a suboptimal concentration of Tg (1 nM) 2 min prior to perfusion through vWf-coated microcapillary tubes. Pretreatment with 1 nM Tg had minimal effect on platelet translocation behavior at a shear rate of 1,800⅐s Ϫ1 prior to UV photolysis (Fig. 4C). Triggering of a Ca 2ϩ spike by UV photolysis of NP-EGTA led to an equivalent transient Ca 2ϩ elevation as that observed for control platelets not incubated with Tg (Fig. 4C). Tg pretreatment partially blocked the reuptake of the NP-EGTA-elicited Ca 2ϩ spikes and led to the onset of an oscillatory Ca 2ϩ response (⌬[Ca 2ϩ ] c Ͼ100 nM) (Fig.  4C) in the entire platelet population (data not shown). The prolonged oscillatory Ca 2ϩ response in the presence of Tg following UV exposure directly correlated with an increased duration of stationary adhesion (Fig. 4C). This stationary adhesion was mediated by integrin ␣ IIb ␤ 3 , since it was completely blocked by pretreating platelets with c7E3 or Aggrastat (Fig.  4C). Taken together, these data suggest that transient Ca 2ϩ spikes may serve to initiate integrin ␣ IIb ␤ 3 activation and transient cell arrest; however, irreversible adhesion appears to be dependent on sustained integrin ␣ IIb ␤ 3 -dependent calcium oscillations.
Extracellular Calcium and PI 3-kinase Requirement for GPIb and Integrin ␣ IIb ␤ 3 -derived Calcium Signals-Our recent studies have defined an important role for transmembrane calcium influx and PI 3-kinase in potentiating sustained Ca 2ϩ oscillations and integrin ␣ IIb ␤ 3 activation during platelet adhesion to immobilized vWf under flow (24,26). To investigate the contribution of transmembrane calcium influx to GPIb and integrin ␣ IIb ␤ 3 -dependent calcium signaling, in vitro flow studies were performed in the presence or absence of EGTA. As demonstrated in Fig. 5A, EGTA markedly reduced integrin ␣ IIb ␤ 3 -dependent calcium signals at 1,800⅐s Ϫ1 , resulting in an 86% reduction in the frequency of ⌬[Ca 2ϩ ] c events occurring above 100 nM in the platelet population. In contrast, chelating extracellular calcium had no inhibitory effect on the magnitude or duration of GPIb-dependent calcium spikes (n ϭ 40) (Fig. 5B). Significantly, even in the absence of extracellular calcium, integrin ␣ IIb ␤ 3 engagement of vWf was still able to induce a sustained oscillatory calcium response (data not shown), indicating that ligand binding to this receptor can initiate calcium release from internal stores.
To investigate the role of PI 3-kinase for GPIb and integrin ␣ IIb ␤ 3 -dependent calcium signaling, platelets were pretreated with concentrations of wortmannin (100 nM) or LY294002 (25 M) that effectively abolish PI 3-kinase activation in vWf-stim- Platelets (150 ϫ 10 9 /liter) were perfused through vWf-coated (100 g/ml) microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . Of the platelets observed to undergo high range calcium responses, 60% were found to undergo sustained Ca 2ϩ flux, whereas 40% were found to undergo transient responses. These cells are representative of over 100 cells analyzed. B, platelets were loaded with 10 M of the caged Ca 2ϩ compound NP-EGTA prior to perfusion through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . The platelets were allowed to translocate on the vWf surface for 18 s prior to exposure to a near UV light source generated by a 100-watt mercury burner directed through the optical path of a Leica DMIRBE inverted microscope. UV exposure was restricted to an exposure time of 0.6 s. Control, representative NP-EGTA-loaded platelet exposed to UV (300 -400 nm) for 0.6 s and concomitant displacement versus time graph; ϩc7E3 Fab, platelet was pretreated with 20 g/ml c7E3 Fab prior to perfusion at 1,800⅐s Ϫ1 . C, NP-EGTA-loaded platelets were treated with 1 nM Tg 2 min prior to perfusion through microcapillary tubes. Tg, representative Tg-treated cell displaying rapid surface translocation prior to near UV exposure and sustained oscillatory Ca 2ϩ flux following the initial NP-EGTA-induced Ca 2ϩ spike; ϩTg ϩc7E3 Fab, NP-EGTA-loaded platelets were treated with 100 g/ml c7E3 10 min prior to perfusion. ulated platelets (26). As demonstrated in Fig. 6A, both inhibitors significantly inhibited integrin ␣ IIb ␤ 3 -dependent high range calcium oscillations, resulting in a 98% reduction in the frequency of ⌬[Ca 2ϩ ] c events occurring above 100 nM in the platelet population. In contrast, neither wortmannin nor LY294002 had any significant effect on transient calcium spikes initiated by the vWf-GPIb interaction (Fig. 6B). Consistent with our previous findings, inhibition of sustained calcium oscillations resulted in an inability of the platelets to form irreversible adhesion contacts under flow (data not shown). These findings suggest that PI 3-kinase involvement in sheardependent platelet adhesion is primarily linked to integrin ␣ IIb ␤ 3 , not GPIb-dependent calcium signals. DISCUSSION The studies reported here have demonstrated the existence of two distinct, cooperative calcium signaling mechanisms utilized by GPIb/V/IX and integrin ␣ IIb ␤ 3 to regulate platelet- FIG. 5. Contribution of transmembrane calcium influx to GPIb/V/IX and integrin ␣ IIb ␤ 3 -dependent Ca 2؉ signals. A, relative frequency histogram demonstrating the distribution (percentage) of calcium flux events recorded in the platelet population following perfusion through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . Ca 2ϩ , platelets were perfused in Tyrode's buffer plus red blood cells (50%) supplemented with 1 mM Ca 2ϩ /Mg 2ϩ ; EGTA, platelets were perfused in Tyrode's buffer plus red blood cells (50%) supplemented with 1 mM EGTA/Mg 2ϩ . The data are representative of three independent experiments. B, platelets were perfused through vWfcoated microcapillaries at a shear rate of 1,800⅐s Ϫ1 in the presence of 1 mg/ml ristocetin following treatment for 10 min with 200 nM Aggrastat. Representative single platelet Ca 2ϩ flux recordings are shown representative of over 100.
FIG. 6. Effect of PI 3-kinase inhibition on GPIb/V/IX and integrin ␣ IIb ␤ 3dependent Ca 2؉ flux. A, relative frequency histogram demonstrating the distribution (percentage) of calcium flux events recorded in the platelet population following perfusion through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 . Me 2 SO, platelets were treated with 0.25% (v/v) Me 2 SO 10 min prior to perfusion; LY294002, platelets were treated with a 25 M concentration of the PI 3-kinase inhibitor LY294002 15 min prior to perfusion; Wortmannin, platelets were treated with a 100 nM concentration of the PI 3-kinase inhibitor wortmannin 15 min prior to perfusion. The data are representative of three independent experiments. B, platelets were perfused through vWf-coated microcapillary tubes at a shear rate of 1,800⅐s Ϫ1 in the presence of 1 mg/ml ristocetin following pretreatment for 15 min with LY294002 (25 M) or Wortmannin (100 nM). Single platelet Ca 2ϩ flux recordings are shown representative of over 100.
adhesive behavior under flow (Fig. 7). A significant finding from these studies is that GPIb/V/IX engagement of vWf elicits a transient calcium response that may serve to initiate integrin ␣ IIb ␤ 3 activation; however, this calcium signal appears insufficient to maintain sustained platelet adhesion in the shear field. Our studies suggest that subsequent integrin ␣ IIb ␤ 3 engagement of vWf triggers outside-in signaling events linked to the initiation and propagation of sustained oscillatory calcium flux that is necessary for stable platelet adhesion.
By examining real time changes in cytosolic calcium during platelet translocation, we have been able to establish a close correlation between cytosolic calcium fluctuations and the stopstart nature of platelet translocation. In particular, we have been able to identify platelets that exhibit transient calcium oscillations that are associated with intermittent phases of stationary adhesion, often lasting 10 -20 s or more. In all cells, the transition from stationary adhesion to surface translocation corresponded to a drop in the cytosolic calcium, approaching 100 nM. These studies suggest that a critical calcium threshold must be reached and maintained for platelets to sustain stable adhesion contacts in a shear field. Furthermore, we have demonstrated that this calcium-dependent regulation of platelet translocation behavior is mediated through the reversible activation of integrin ␣ IIb ␤ 3 , establishing for the first time an important role for this receptor in regulating platelet translocation dynamics under flow.
Our studies have also provided new insight into the relative contribution of GPIb and integrin ␣ IIb ␤ 3 outside-in signaling toward vWf-induced calcium flux. The traditional model for platelet activation by vWf, based primarily on studies of shearinduced platelet aggregation using a cone-and-plate viscometer, had suggested that GPIb-induced transmembrane calcium influx represented the critical proximal signaling step for subsequent ADP release and integrin ␣ IIb ␤ 3 activation. However, recent studies from our laboratory have demonstrated that vWf binding to GPIb is sufficient to induce intracellular calcium mobilization, independent of transmembrane calcium influx and ADP release, as a necessary event for subsequent integrin ␣ IIb ␤ 3 activation (7,24). In this report, we demonstrate the cooperative relationship between GPIb and integrin ␣ IIb ␤ 3 in regulating cytosolic calcium flux and, furthermore, demonstrate that the calcium signaling mechanisms operating downstream of GPIb/V/IX and integrin ␣ IIb ␤ 3 are distinct, in that the former are primarily due to calcium release from internal stores, whereas the latter are dependent on both intracellular calcium mobilization and transmembrane calcium influx. The reason for this difference remains to be established but may be related to differences in the duration and magnitude of calcium mobilization. Calcium store emptying is a major stimulus for transmembrane calcium influx, and it is conceivable that integrin ␣ IIb ␤ 3 -dependent sustained calcium flux leads to more extensive depletion of intracellular calcium stores in comparison with the transient calcium spikes initiated by GPIb binding. Alternatively, integrin ␣ IIb ␤ 3 engagement may directly activate an associated calcium influx pathway, as proposed by Brass (15).
This study demonstrates for the first time a role for PI 3-kinase in regulating integrin ␣ IIb ␤ 3 calcium signaling and, somewhat unexpectedly, does not support an important role for this kinase in GPIb signaling, at least under high shear conditions. There are two potential mechanisms by which PI 3-kinase may promote integrin ␣ IIb ␤ 3 activation and sustained calcium oscillations. First, while not essential for GPIb-dependent calcium flux, PI 3-kinase may function downstream of cytosolic calcium to initiate integrin ␣ IIb ␤ 3 activation. Alternatively, PI 3-kinase may participate in integrin ␣ IIb ␤ 3 outside-in signaling events associated with the initiation and propagation of intracellular calcium oscillations, necessary for sustained integrin ␣ IIb ␤ 3 activation. While we have not formally excluded the former hypothesis, several lines of evidence suggest that the latter hypothesis is more likely. First, to our knowledge there are no precedents for PI 3-kinase signaling downstream of calcium. Second, there is a considerable body of evidence demonstrating PI 3-kinase involvement in integrin ␣ IIb ␤ 3 outside-in signaling. For example, the production of 3-phosphorylated phosphoinositides in thrombin-stimulated platelets is regulated downstream of integrin ␣ IIb ␤ 3 (27). Furthermore, direct ligand binding to integrin ␣ IIb ␤ 3 is sufficient to induce PI 3-kinase activation and a selective increase in the cellular levels of phosphatidylinositol 3,4-bisphosphate (28). This integrin ␣ IIb ␤ 3dependent activation of PI 3-kinase has been proposed to sustain integrin ␣ IIb ␤ 3 activation as an important event for cytoskeletal reorganization, platelet spreading, and irreversible aggregation (28,29).
It remains to be established what the molecular mechanism is by which PI 3-kinase promotes integrin ␣ IIb ␤ 3 -dependent cytosolic calcium flux. Previous studies in platelets and a range of other cells have demonstrated that the PI 3-kinase lipid FIG. 7. Proposed model of the cooperative calcium signaling relationship operating between GPIb/V/IX and integrin ␣ IIb ␤ 3 . Initial outside-in signaling events mediated by GPIb/V/IX engagement of surface immobilized vWF triggers the initiation of an elementary calcium spike, leading to low level integrin ␣ IIb ␤ 3 activation. Subsequent intergin-␣ IIb ␤ 3 engagement of vWF initiates outside-in signaling events, driving further rounds of Ca 2ϩ mobilization and integrin ␣ IIb ␤ 3 activation. PI 3-kinase and Ca 2ϩ influx serve to reinforce the initial GPIb-dependent Ca 2ϩ signal by potentiating integrin ␣ IIb ␤ 3 -dependent outside-in signaling, driving platelet deceleration and adhesion in the shear field. Ca 2ϩ , free cytosolic calcium; GP Ib/V/IX, glycoprotein Ib/V/IX vWF adhesion receptor; ϩ, integrin ␣ IIb ␤ 3 activation; PI3K, phosphatidylinositol 3-kinase. product, phosphatidylinositol 3,4,5-trisphosphate, plays a significant role in the regulation of both intracellular calcium mobilization and transmembrane calcium flux (30 -34). However, given previous observations that direct ligation of integrin ␣ IIb ␤ 3 leads to a selective increase in phosphatidylinositol 3,4-bisphosphate, it remains to be determined whether phosphatidylinositol 3,4,5-trisphosphate plays a significant role in integrin ␣ IIb ␤ 3 -dependent calcium signaling.
Finally, much of our current conceptual understanding of integrin ␣ IIb ␤ 3 signaling in platelets is based on studies of platelets in suspension, in which the addition of soluble stimuli is required to initially induce integrin ␣ IIb ␤ 3 activation (insideout signaling). In this assay system, subsequent ligand binding events are required to generate outside-in signals, necessary to promote relatively "late" platelet functional responses, including irreversible platelet aggregation, clot retraction, and the shedding of procoagulant-rich microvesicles. Our studies suggest that the establishment of integrin ␣ IIb ␤ 3 outside-in calcium signaling plays a critical "early" role in the initial activation of platelets on vWf. These studies define a previously unrecognized role for integrin ␣ IIb ␤ 3 in driving a calcium-dependent positive feedback mechanism that plays an important role in regulating the affinity status of the integrin itself.