A selective role for phosphatidylinositol 3,4,5-trisphosphate in the Gi-dependent activation of platelet Rap1B.

The small GTP-binding protein Rap1B is activated in human platelets upon stimulation of a G(i)-dependent signaling pathway. In this work, we found that inhibition of platelet adenylyl cyclase by dideoxyadenosine or SQ22536 did not cause activation of Rap1B and did not restore Rap1B activation in platelets stimulated by cross-linking of Fcgamma receptor IIA (FcgammaRIIA) in the presence of ADP scavengers. Moreover, elevation of the intracellular cAMP concentration did not impair the G(i)-dependent activation of Rap1B. Two unrelated inhibitors of phosphatidylinositol 3-kinase (PI3K), wortmannin and LY294002, totally prevented Rap1B activation in platelets stimulated by cross-linking of FcgammaRIIA, by stimulation of the P2Y(12) receptor for ADP, or by epinephrine. However, in platelets from PI3Kgamma-deficient mice, both ADP and epinephrine were still able to normally stimulate Rap1B activation through a PI3K-dependent mechanism, suggesting the involvement of a different isoform of the enzyme. Moreover, the lack of PI3Kgamma did not prevent the ability of epinephrine to potentiate platelet aggregation through a G(i)-dependent pathway. The inhibitory effect of wortmannin on Rap1B activation was overcome by addition of phosphatidylinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P(3)), but not PtdIns(3,4)P(2), although both lipids were found to support phosphorylation of Akt. Moreover, PtdIns(3,4,5)P(3) was able to relieve the inhibitory effect of apyrase on FcgammaRIIA-mediated platelet aggregation. We conclude that stimulation of a G(i)-dependent signaling pathway causes activation of the small GTPase Rap1B through the action of the PI3K product PtdIns(3,4,5)P(3), but not PtdIns(3,4)P(2), and that this process may contribute to potentiation of platelet aggregation.

Rap1B is a small GTP-binding protein highly expressed in human platelets (1). In resting cells, it is mainly located at the membrane, but it translocates to the cytosol upon phosphorylation by protein kinase A (2). In activated platelets, Rap1B rapidly interacts with the reorganized actin-based cytoskeleton (3). As other GTPases, Rap1B is activated by binding of GTP. Platelet stimulation by different agonists, such as thrombin, collagen, and ADP, induces the rapid binding of GTP to Rap1B (4,5). An increase in the intracellular Ca 2ϩ concentration in stimulated platelets has been shown to be sufficient to promote Rap1B activation, and specific Ca 2ϩ /calmodulin-sensitive guanine nucleotide exchange factors for Rap1B have been identified (5,6). We (7) and others (8) have recently described a new pathway for Rap1B activation that is initiated by stimulation of membrane G i -coupled receptors and that is independent of intracellular Ca 2ϩ increases. In fact, the sole binding of ADP to the P2Y 12 receptor, as well as the interaction of epinephrine with the ␣ 2A -adrenergic receptor, is sufficient to trigger Rap1B activation. Moreover, we have found that agonists that activate platelets through stimulation of G q -coupled receptors, such as the thromboxane A 2 analog U46619, or through stimulation of a tyrosine kinase-based pathway, such as in the case of crosslinking of Fc␥RIIA, 1 totally rely on binding of secreted ADP to the G i -coupled P2Y 12 receptor to activate Rap1B (7). Finally, activation of Rap1B induced by ADP or epinephrine is prevented in G␣ i2 -and G␣ z -deficient mice, respectively (8).
During the last few years, activation of a G i -dependent signaling pathway has been recognized to represent a crucial event absolutely required to elicit full platelet activation. For instance, platelet responsiveness to the thromboxane A 2 analog U46619, to protease-activated receptor-1-activating peptide, or to cross-linking of Fc␥RIIA is strongly compromised when secretion is prevented by protein kinase C inhibitors or when extracellular ADP is neutralized by specific scavengers, such as apyrase or creatine phosphate/creatine phosphokinase (9 -12). It has also been clearly shown that, although ADP can bind to two different G-protein-coupled receptors on the platelet surface (the P2Y 1 receptor coupled to G q and the P2Y 12 receptor coupled to G i ), only the latter one is responsible for potentiation of platelet activation induced by other agonists (10 -12).
The exact mechanism for the G i -mediated potentiation of platelet activation is still unclear. The ␣-subunits of the G i family of heterotrimeric G-proteins are known to inhibit adenylyl cyclase, but several findings indicate that reduction of basal cAMP levels does not contribute to ADP-mediated potentiation of platelet activation (13,14). By contrast, several studies using specific cell-permeable inhibitors have suggested a crucial role for phosphatidylinositol 3-kinase (PI3K) in this event (11,12,14,15). The recent finding that stimulation of a G i -coupled receptor is sufficient to trigger activation of Rap1B also suggests that this small GTPase could be involved in potentiation of platelet aggregation.
In this work, we have investigated the mechanism of Rap1B activation downstream of stimulation of G i -coupled receptors in an attempt to reveal a correlation between this small GTPase and the potentiation of agonist-induced platelet activation. We have found that inhibition of adenylyl cyclase is not sufficient to stimulate GTP binding to Rap1B. By contrast, it is shown here that PI3K plays an important role in the G i -mediated activation of Rap1B. We also provide evidence suggesting that activation of Rap1B downstream of PI3K is associated with the G i -mediated potentiation of platelet activation.
Human Platelet Isolation and Stimulation-Human platelets were isolated by gel filtration on Sepharose CL-2B and eluted with HEPES buffer (10 mM HEPES, 137 mM NaCl, 2.9 mM KCl, and 12 mM NaHCO 3 , pH 7.4) as previously described (17). Platelet concentration was adjusted to 0.35 ϫ 10 9 platelets/ml. Platelet samples (0.5 ml) were incubated at 37°C in an aggregometer under constant stirring and typically stimulated with 10 M ADP and 1 M epinephrine or by cross-linking of Fc␥RIIA through addition of 2 g/ml mAb IV.3 for 2 min, followed by 30 g/ml sheep anti-mouse F(abЈ) 2 fragments. Platelet stimulation was typically performed for 1 min. Where indicated, 1 unit/ml apyrase, 5 mM creatine phosphate, 40 units/ml creatine phosphokinase, 500 M PAPS, or 100 nM AR-C69931MX was added to the platelet samples 2 min before stimulation. Preincubation with wortmannin or LY294002 was performed for 15 min at 37°C. Inhibition of adenylyl cyclase was achieved by incubation of platelets with 100 M dideoxyadenosine or 300 M SQ22536 for 30 min. PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 were dissolved in Me 2 SO and added to the platelet suspension at a final concentration of 30 M. Measurement of platelet aggregation was performed under the same conditions indicated above, and aggregation was monitored continuously over 10 min.
Rap1B Activation Assay-Activation of Rap1B was evaluated using GST-RBD immobilized on GSH-Sepharose, which is known to bind specifically and selectively the GTP-bound form of Rap1B from a platelet lysate. Platelet stimulation was stopped by addition of an equal volume of ice-cold modified 2ϫ RIPA buffer (100 mM Tris-HCl, pH 7.4, 400 mM NaCl, 5 mM MgCl 2 , 2% Nonidet P-40, 20% glycerol, 2 mM phenylmethylsulfonyl fluoride, 2 M leupeptin, 0.2 M aprotinin, and 0.2 mM Na 3 VO 4 ). Cell lysis was performed on ice for 10 min, and the insoluble material was eliminated by centrifugation at 13,000 rpm for 10 min at 4°C. Recombinant purified GST-RBD was coupled to GSH-Sepharose by incubating 200 g of the protein with 100 l of GSH-Sepharose (75% slurry) for 2 h at room temperature under constant tumbling and then added to the cleared platelet lysates (20 g of GST-RBD/sample). Precipitation of GTP-bound Rap1B was performed by incubation at 4°C for 45 min. The precipitates were collected by brief centrifugation, washed three times with modified 1ϫ RIPA buffer, and finally resuspended in 25 l of SDS sample buffer (25 mM Tris, 192 mM glycine, pH 8.3, 4% SDS, 1% dithiothreitol, 20% glycerol, and 0.02% bromphenol blue). Precipitated Rap1B was separated by SDS-PAGE on 10 -20% acrylamide gradient gels and transferred to nitrocellulose. The presence of active Rap1B in precipitates with GST-RBD was evaluated by staining the nitrocellulose filters with a specific polyclonal antiserum directed against Rap1B, used at a final dilution of 1:1000. Reactive proteins were detected by enhanced chemiluminescence reaction. Data in all figures are representative of at least three separate experiments.
Measurement of Akt Phosphorylation-Platelet samples (0.2 ml, 10 9 platelets/ml) were incubated at 37°C and stimulated as indicated in the figure legends for 1 min. Platelets were lysed in 2% SDS in HEPES buffer, and protein concentration was determined. Aliquots containing 80 g of total platelet lysates were heated at 96°C for 5 min in SDS sample buffer, separated on 12% acrylamide gel, and transferred to nitrocellulose. Blots were probed with anti-phospho-Akt Thr 308 antibody and then reprobed with anti-Akt antibody. Analysis of Akt phosphorylation in platelets stimulated by cross-linking of Fc␥RIIA was performed with immunoprecipitated Akt because preliminary experiments revealed a cross-reactivity of the anti-phospho-Akt Thr 308 antibody with the heavy chains of mAb IV.3, used to activate Fc␥RIIA, that compromised the interpretation of the results (data not shown). Platelet samples were lysed in ice-cold RIPA buffer and precleared with protein A-Sepharose. The precleared lysates that were also devoid of mAb IV.3 were used to immunoprecipitate Akt with mAb 5G3. Immunoprecipitates were then analyzed by immunoblotting with anti-phospho-Akt Thr 308 antibody and reprobed with anti-Akt antibody.
Studies with Mouse Platelets-PI3K␥-deficient mice were generated as previously described (18). Blood was collected from anesthetized mice from the inferior vena cava into syringes containing heparin solution (5 units/ml). Blood was centrifuged at 90 ϫ g for 10 min, and the platelet-rich plasma was collected. Aggregation studies were directly performed with the platelet-rich plasma upon adjustment of the platelet count to 2 ϫ 10 8 platelets/ml with autologous platelet-poor plasma. For analysis of Rap1B activation, washed platelets were prepared by centrifuging the platelet-rich plasma at 150 ϫ g for 10 min. The platelet pellet was washed once with 0.038% trisodium citrate, 0.6% glucose, and 0.72 NaCl, pH 7.0, containing 25 ng/ml prostaglandin E 1 and finally resuspended in HEPES buffer at a final concentration of 3 ϫ 10 8 platelets/ml. Stimulation of platelet samples (0.4 ml), lysis, and pull-down assay for Rap1B activation were performed as described above.

G i -dependent Activation of Rap1B Does Not Require Inhibition of Adenylyl Cyclase-
We have previously demonstrated that stimulation of a G i -dependent pathway by epinephrine or by binding of ADP to the P2Y 12 receptor is sufficient to trigger activation of the small GTPase Rap1B and that Rap1B activation promoted by cross-linking of Fc␥RIIA is completely dependent on the stimulation of the G i -coupled P2Y 12 receptor by secreted ADP (7). To investigate the signaling pathway linking G i to Rap1B, we first analyzed the possible role of G␣ i -mediated inhibition of adenylyl cyclase. Gel-filtered platelets were treated with two different membrane-permeable inhibitors of adenylyl cyclase (SQ22536 and dideoxyadenosine), and activation of Rap1B was evaluated upon precipitation of the GTPbound form of the protein with GST-RBD. Although SQ22536 and dideoxyadenosine were used at concentrations reported to maximally inhibit forskolin-stimulated adenylyl cyclase (13), neither compound by itself caused detectable activation of Rap1B (Fig. 1A). These results are in agreement with those recently reported by Woulfe et al. (8). However, we considered the possibility that, although not sufficient by itself to induce Rap1B activation, inhibition of adenylyl cyclase could contribute to this process in association with co-stimulation of other signaling pathways. In platelets stimulated by cross-linking of Fc␥RIIA, activation of Rap1B is suppressed by the ADP scavenger creatine phosphate/creatine phosphokinase, but is restored by the simultaneous addition of epinephrine (7). By contrast, in the presence of creatine phosphate/creatine phosphokinase, neither SQ22536 nor dideoxyadenosine was able to restore Rap1B activation upon cross-linking of Fc␥RIIA (Fig.  1A). This indicates that the contribution of the G i pathway to Rap1B activation does not require the inhibition of adenylyl cyclase by the G-protein ␣-subunit.
We also considered that the very low intracellular levels of cAMP in resting platelets may represent a permissive condition to allow agonist-induced Rap1B activation. To verify this possibility, human platelets were incubated with prostaglandin E 1 to stimulate adenylyl cyclase and to increase cAMP levels and then treated with agonists that activate Rap1B through stimulation of G i . Fig. 1B shows that, even when intracellular cAMP levels were increased by prostaglandin E 1 , activation of Rap1B induced by clustering of Fc␥RIIA, epinephrine, or ADP occurred normally. Taken together, these results indicate that the G i -mediated activation of Rap1B is completely independent of the modulation of intracellular cAMP levels.
G i -mediated Activation of Rap1B Is Regulated by PI3K-It is known that activation of G i is necessary to support full platelet secretion and aggregation as well as Rap1B activation induced by U46619 or by cross-linking of Fc␥RIIA (7,10,12). Moreover, it has been shown that PI3K plays a crucial role in the G imediated potentiation of platelet activation and is required for irreversible aggregation (11,12,14,15). Therefore, we investigated the role of PI3K in Fc␥RIIA-mediated activation of Rap1B using two structurally unrelated inhibitors of the enzyme, wortmannin and LY294002. Fig. 2A shows how activation of Rap1B induced by clustering of Fc␥RIIA was totally suppressed by both compounds. By contrast, Rap1B activation induced by thrombin was not significantly affected by LY294002 ( Fig. 2A) or by wortmannin (data not shown). We have previously shown that Fc␥RIIA-induced activation of Rap1B requires the binding of secreted ADP to the P2Y 12 receptor (7). To verify whether, upon Fc␥RIIA recruitment, PI3K signals to Rap1B downstream of the G i -coupled receptor for ADP, human platelets were stimulated by cross-linking of Fc␥RIIA in the presence of PAPS, a selective antagonist of the P2Y 1 receptor for ADP. Under these conditions, secreted ADP can bind exclusively to the G i -coupled P2Y 12 receptor. As shown in Fig. 2B, binding of secreted ADP to the P2Y 12 receptor was sufficient to allow Rap1B activation in response to Fc␥RIIA cross-linking. Moreover, under these conditions, inhibition of PI3K by wortmannin or LY294002 still completely suppressed activation of Rap1B (Fig. 2B).
To further demonstrate that PI3K lies downstream of the G i -coupled P2Y 12 receptor, we analyzed Rap1B activation in response to exogenous ADP. Fig. 3A shows that, when exogenous ADP was allowed to bind exclusively to the P2Y 12 receptor, i.e. in the presence of PAPS, activation of Rap1B was prevented by the PI3K inhibitors wortmannin and LY294002. Finally, we analyzed the role of PI3K in Rap1B activation induced by epinephrine, which binds exclusively to the G icoupled ␣ 2A -adrenergic receptor on the platelet surface. Fig. 3B shows that both wortmannin and LY294002 almost completely suppressed epinephrine-induced activation of Rap1B. These results confirm and extend previously reported data (8) and indicate that PI3K is a key element in the G i -dependent pathway for Rap1B activation.
PI3K␥ Is Not Involved in Either Rap1B Activation or Potentiation of Platelet Aggregation Mediated by G i -It is known that human platelets express different isoforms of PI3K (23). At least two members of the class I PI3K family, viz. PI3K␤ and PI3K␥, are activated by G-protein ␤␥-dimers (24 -26) and therefore may represent the isoforms linking G i to Rap1B. To investigate the possible role of PI3K␥, we compared agonistinduced activation of Rap1B in platelets from PI3K␥-deficient and wild-type mice. Because mouse platelets do not express Fc␥RIIA, these studies have been performed using ADP and epinephrine as Rap1B activators. As shown in Fig. 4, activation of Rap1B induced by ADP or epinephrine in PI3K␥-deficient platelets was almost identical to that observed in platelets from wild-type mice. This finding argues against a role for PI3K␥ in coupling activation of G i to Rap1B. However, because several intracellular messengers, including calcium, diacylglycerol, and tyrosine kinases, can mediate Rap1 activation (6), we considered that, in the absence of PI3K␥, other signaling pathways may become predominant and result in an almost equally efficient activation of Rap1B. By using a selective antagonist of the P2Y 12 receptor, we found that, in both wild-type and PI3K␥ Ϫ/Ϫ mice, ADP-induced activation of Rap1B was equally dependent on stimulation of the G i -coupled receptor (Fig. 4). Moreover, in both control and PI3K␥-deficient platelets, ADP-or epinephrine-induced activation of Rap1B was still prevented by the PI3K inhibitors LY294002 (Fig. 4) and wortmannin (data not shown). These results clearly indicate that a PI3K isoform different from PI3K␥ is involved in the G i -mediated activation of Rap1B.
Both PI3K and Rap1B have been hypothesized to be involved in the G i -dependent potentiation of platelet aggregation. Because activation of Rap1B downstream of G i occurs normally in PI3K␥ Ϫ/Ϫ mice, these platelets represent a good model to test the specific contribution of PI3K␥ versus Rap1B to epinephrineinduced potentiation of platelet aggregation. In mouse platelets, ADP caused a reversible platelet aggregation that required the concomitant stimulation of the G q -coupled P2Y 1 and G i -coupled P2Y 12 receptors. A small (but significant) reduction of ADP-induced platelet aggregation in PI3K␥ Ϫ/Ϫ mice was reported in an earlier study (32). This effect was more evident when low doses of ADP were used to stimulate washed platelets. In the present work, such inhibition was negligible because we used high doses of ADP to stimulate platelets in Platelets were isolated from wild-type (PI3K ϩ/ϩ ) or PI3K␥-deficient (PI3K Ϫ/Ϫ ) mice and stimulated with buffer (none), 10 M epinephrine (Epi), or 10 M ADP for 1 min. Preincubation with 25 M LY294002 was performed for 15 min before stimulation, whereas preincubation with 100 nM AR-C69931MX was for 2 min before addition of ADP. Active Rap1B was precipitated from platelet lysates with GST-RBD and revealed by immunoblotting with a specific polyclonal antiserum. The amount of total Rap1B in cell lysates was also monitored by immunoblotting.
platelet-rich plasma. However, Fig. 5 shows that selective blockade of the P2Y 12 receptor strongly inhibited platelet aggregation induced by high doses of ADP in both wild-type and PI3K␥ Ϫ/Ϫ mice. However, in the presence of the P2Y 12 receptor antagonist, addition of epinephrine, which activates a G i -dependent signaling pathway leading to stimulation of Rap1B activation, strongly, although not completely, restored ADPinduced platelet aggregation in both wild-type and PI3K␥ Ϫ/Ϫ platelets. This indicates that PI3K␥ is involved neither in Rap1B activation nor in the G i -dependent potentiation of platelet aggregation.

PI3K-dependent Activation of Rap1B Is Mediated by the Lipid Product PtdIns(3,4,5)P 3 , but Not by PtdIns(3,4)P 2 -Acti-
vation of PI3K leads to the accumulation of the lipid products PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 , which, in turn, promote stimulation of Akt. To investigate the role of 3-phosphorylated phosphoinositides in Rap1B activation, we analyzed their ability to relieve the inhibition of Rap1B activation in wortmannintreated platelets. Preliminary experiments with permeabilized platelets revealed that saponin treatment caused an almost total loss of platelet responsiveness to the analyzed agonists (data not shown). However, PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 have been shown to trigger biological responses even when added to whole cells (19 -21). Moreover, when dissolved in Me 2 SO and then added to intact platelets, both lipids have been found to be rapidly incorporated into the cell membrane (22). Fig. 6A shows that, in platelets stimulated by cross-linking of Fc␥RIIA, inhibition of Rap1B activation by wortmannin was partially reversed by addition of PtdIns(3,4,5)P 3 . By contrast, the related lipid PtdIns(3,4)P 2 was unable to relieve the wortmannin-induced inhibition of Rap1B activation. Similarly, even when activation of Rap1B was triggered by direct stimulation of G i -coupled receptors by ADP or epinephrine, PtdIns(3,4,5)P 3 was able to partially counteract the inhibitory effect of wortmannin (Fig. 6, B and C). These results indicate that the lipid product PtdIns(3,4,5)P 3 , but not PtdIns(3,4)P 2 , is directly involved in the PI3K-dependent activation of Rap1B downstream of stimulation of G i . We next investigated the ability of PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 to promote Akt phosphorylation using phosphospecific antibodies against phospho-Akt Thr 308 . As shown in Fig. 7A, platelet activation by cross-linking of Fc␥RIIA induced the phosphorylation of Akt at Thr 308 , which was prevented by preincubation with wortmannin. When analyzed using the A, gel-filtered platelets were stimulated by clustering (clust.) of Fc␥RIIA with 2 g/ml mAb IV.3 and 30 g/ml sheep antimouse F(abЈ) 2 fragments for 1 (1Ј) or 3 (3Ј) min or with 1 unit/ml thrombin (THR) for 1 min. Preincubation with 50 nM wortmannin (wt) was for 15 min before stimulation. Platelets were lysed in RIPA buffer, and Akt was immunoprecipitated (Ip) from the precleared lysates with mAb 5G3 and then analyzed by immunoblotting with anti-phospho-Akt Thr 308 antibody (p-Akt (Thr 308)). Blots were stripped and reprobed with anti-Akt antibody (Akt). B, gel-filtered platelets were treated with 30 M PtdIns(3,4,5)P 3 , 30 M PtdIns(3,4,5)P 3 , 10 M ADP, or 1 unit/ml thrombin at 37°C for 1 min. Samples were lysed in 2% SDS, and the levels of Akt phosphorylation (p-Akt) at Thr 308 were measured by immunoblotting with phospho-specific antibodies. Blots were then stripped and reprobed with anti-Akt antibody. C, similar samples were lysed in RIPA buffer and processed for measurement of Rap1B activation. same experimental approach, phosphorylation of Akt induced by thrombin was found to be comparable to that induced by stimulation of Fc␥RIIA (Fig. 7A). Fig. 7B shows that addition of PtdIns(3,4,5)P 3 or PtdIns(3,4)P 2 to intact platelets caused phosphorylation of Akt to a similar extent, as evaluated by immunoblotting of whole platelet lysates. The level of Akt phosphorylation induced by the 3-phosphorylated phosphoinositides was lower than that observed in platelets stimulated with thrombin, but comparable to that induced by ADP. Parallel analysis of Rap1B activation showed that addition of PtdIns(3,4,5)P 3 to platelets was sufficient to stimulate GTP binding to Rap1B (Fig. 7C). By contrast, PtdIns(3,4)P 2 , although able to promote Akt phosphorylation, did not stimulate activation of Rap1B. These results indicate that exogenous PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 can mimic the effects of endogenous lipids and that phosphorylation of Akt does not seem to be required for Rap1B activation.
In the presence of ADP scavengers, platelet aggregation induced by cross-linking of Fc␥RIIA is inhibited, but can be restored by the simultaneous addition of epinephrine (12). Under the same conditions, also activation of both PI3K and Rap1B is prevented and can be restored by epinephrine (7,12). Therefore, we wondered whether direct activation of Rap1B by PtdIns(3,4,5)P 3 could by-pass the need for PI3K activation to induce platelet aggregation. Fig. 8 shows that Fc␥RIIA-mediated platelet aggregation was totally prevented when platelets where stimulated in the presence of apyrase, but could be almost completely restored by epinephrine. Addition of PtdIns(3,4,5)P 3 at a concentration shown to restore Rap1B activation also resulted in a limited (but significant) recovery of platelet aggregation. At higher concentration of PtdIns-(3,4,5)P 3 , the recovery of platelet aggregation was even more evident, although clearly less marked than that induced by epinephrine. Taken together, these results indicate that, when ADP secreted upon cross-linking of Fc␥RIIA is neutralized, exogenous PtdIns(3,4,5)P 3 causes both activation of Rap1B and restoration of platelet aggregation. DISCUSSION During the last few years, a number of studies using scavenger systems and receptor antagonists demonstrated the essential role played by secreted ADP in potentiating platelet activation in response to many different extracellular agonists (9 -12). Although the biological effects of ADP on human platelets require concomitant activation of two different G-proteincoupled receptors, P2Y 1 (coupled to G q ) and P2Y 12 (coupled to G i ), potentiation of platelet activation induced by other agonists is mediated mainly by activation of a G i -dependent pathway through the P2Y 12 receptor (10,12). In an attempt to identify the intracellular messengers promoting G i -dependent potentiation of platelet activation, we (7) and others (8) have recently found that the small GTPase Rap1B is indeed activated upon the sole binding of ADP to the P2Y 12 receptor or upon activation of the G i -coupled ␣ 2A -adrenergic receptor by epinephrine. Moreover, in platelets from mice that lack G␣ i2 or G␣ z , activation of Rap1B by ADP or epinephrine is abolished (8). Rap1 is emerging as a modulator of cell adhesion and integrin function (6), and a very recent study by Bertoni et al. (27) demonstrates that active Rap1B regulates the affinity state of integrin ␣ IIb ␤ 3 in mouse megakaryocytes. Therefore, Rap1B may represent a link between activation of a G i pathway and stabilization of integrin-mediated platelet aggregation.
In this work, we investigated the biochemical mechanism of Rap1B activation downstream of stimulation of a G i signaling pathway. By using inhibitors and activators of adenylyl cyclase, we demonstrated that the reduction of the intracellular levels of cAMP promoted by the G i ␣-subunits is not required to induce activation of Rap1B. These results confirm and extend those recently reported by Woulfe et al. (8) and strengthen the idea that G␣ i -mediated inhibition of adenylyl cyclase does not directly contribute to the G i -promoted potentiation of platelet activation (13,14).
In addition to the activated ␣-subunits, stimulation of a G i -coupled receptor generates free ␤␥-dimers that are able to stimulate different intracellular effectors, including selective isoforms of the lipid-metabolizing enzyme PI3K, such as PI3K␤ and PI3K␥. In this work, we have actually demonstrated that activation of PI3K is an essential step in the G i -mediated signaling pathway leading to Rap1B activation. In fact, the PI3K inhibitors wortmannin and LY294002 have been found to totally suppress Rap1B activation upon clustering of Fc␥RIIA by acting downstream of P2Y 12 receptor stimulation by secreted ADP. Moreover, as recently described (8), we also found that PI3K inhibitors prevented activation of Rap1B induced by direct binding of ADP to the P2Y 12 receptor or of epinephrine to the ␣ 2A -adrenergic receptor. Therefore, PI3K appears to link G i -coupled receptors to the small GTPase Rap1B.
How activation of G i results in the stimulation of PI3K and how PI3K promotes activation of Rap1B are still unclear. Free ␤␥-dimers generated upon stimulation of a G i -coupled receptor may activate some class I PI3K isoforms, viz. PI3K␤ and PI3K␥ (24 -26). PI3K␥ is highly expressed in platelets, and it has been reported to play an essential role in ADP-induced platelet activation downstream of the G i -coupled P2Y 12 receptor (32).

FIG. 8. Inhibition of Fc␥RIIA-mediated platelet aggregation by apyrase is reversed by PtdIns(3,4,5)P 3 .
Gel-filtered platelets were placed in an aggregometer, and clustering of Fc␥RIIA was induced by addition of 2 g/ml mAb IV.3 and 30 g/ml sheep anti-mouse F(abЈ) 2 fragments as indicated by the arrows. Aggregation was monitored continuously. The representative traces shown refer to samples treated as follows. trace A, no treatment; trace B, 1 unit/ml apyrase; trace C, 1 unit/ml apyrase and 10 M epinephrine; trace D, 1 unit/ml apyrase and 30 M PtdIns(3,4,5)P 3 ; trace E, 1 unit/ml apyrase and 60 M PtdIns(3,4,5)P 3 . Apyrase was added together with mAb IV.3, whereas epinephrine or PtdIns(3,4,5)P 3 was added together with sheep antimouse F(abЈ) 2 fragments. Therefore, we hypothesized that this isozyme could link G i to activation of Rap1B. However, studies with platelets from PI3K␥ knockout mice clearly showed that the lack of expression of this isoform does not compromise the ability of epinephrine or ADP to induce activation of Rap1B. A variable decrease in agonist-induced Rap1B activation in PI3K␥ Ϫ/Ϫ mice has been recently reported (8). Although some small differences were occasionally observed in our experiments (see, for instance, Rap1B activation induced by epinephrine in the representative experiment reported in Fig. 4), the analysis of several determinations led us to conclude that these occasional differences were not significant and that the contribution of PI3K␥ to G i -mediated Rap1B activation, if any, is negligible.
Our results also demonstrate that the G i -mediated pathway for platelet aggregation, which can be monitored by the ability of epinephrine to replace ADP, is normally functional in PI3K␥ Ϫ/Ϫ mice. This is consistent with a role for activated Rap1B, but not for PI3K␥, in the signaling pathway from G i to platelet aggregation. We also have shown that, in PI3K␥ Ϫ/Ϫ mice, G i -mediated activation of Rap1B is still prevented by PI3K inhibitors. This clearly indicates that a different isoform of the enzyme is mainly involved. Interestingly, platelets express high levels of PI3K␤ (32), which is also activated by ␤␥-dimers (25,26). Moreover, a recent report demonstrates that PI3K␤, but not PI3K␥, represents a link between ␤␥dimers released upon stimulation of the G i -coupled receptor for lysophosphatidic acid and activation of the small GTPase Ras (33). Therefore, it is most likely that PI3K␤ represents the main PI3K isoform involved in the regulation of Rap1B.
In this work, we have further strengthened the functional correlation between PI3K and Rap1B activation by demonstrating that exogenous PtdIns(3,4,5)P 3 , but not PtdIns(3,4)P 2 , can restore wortmannin-inhibited activation of Rap1B in platelets stimulated with ADP or epinephrine or by cross-linking of Fc␥RIIA. This indicates that the action of PI3K on Rap1B activation is actually mediated, at least in part, by the main lipid product of this enzyme. It is interesting to note that, in stimulated platelets, accumulation of PtdIns(3,4,5)P 3 is rapid and transient, whereas synthesis of PtdIns(3,4)P 2 occurs mainly as a consequence of integrin ␣ IIb ␤ 3 -mediated platelet aggregation (28,29). Therefore, the ability of PtdIns(3,4,5)P 3 , but not PtdIns(3,4)P 2 , to support Rap1B activation is consistent with a role for this GTPase in an early phase of platelet stimulation that precedes integrin ␣ IIb ␤ 3 activation and cell aggregation. Evaluation of Akt activation by exogenous PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 revealed that both lipids were able to stimulate phosphorylation of Akt at Thr 308 . A previous report suggested that, in contrast to PtdIns(3,4)P 2 , addition of PtdIns(3,4,5)P 3 to intact platelets does not activate Akt (20). Although we cannot provide a definitive explanation for this discrepancy, our results are clearly in line with the current concept of PtdIns(3,4,5)P 3 being a major regulator of Akt (24). It is interesting to note that, although both 3-phosphorylated phosphoinositides were able to activate Akt, only PtdIns(3,4,5)P 3 caused activation of Rap1B. The finding that PtdIns(3,4)P 2 induces Akt phosphorylation, but not Rap1B activation, suggests a dissociation between the two events. This is also supported by the finding that ADP induced normal, PI3Kdependent activation of Rap1B in PI3K␥ Ϫ/Ϫ mouse platelets (Fig. 4) under conditions in which phosphorylation of Akt has been reported to be totally suppressed (32).
The data in Fig. 6 clearly show that the restoration of Rap1B activation by PtdIns(3,4,5)P 3 never returns the amount of active Rap1B to that observed in control platelets stimulated in the absence of wortmannin. This indicates either that generation of PtdIns(3,4,5)P 3 is not the only mechanism by which PI3K participates in Rap1B activation or that endogenously generated lipids act more efficiently. In this regard, it should be noted that stimulators of G i , such as ADP and epinephrine, are really weak stimulators of PI3K (12,30). For instance, epinephrine induces the accumulation of Ͻ20% of the amount of PtdIns(3,4,5)P 3 measured in thrombin-stimulated platelets (12). Despite this, the finding that PI3K inhibitors totally suppressed Rap1B activation clearly indicates that such a small amount of 3-phosphoinositides plays a crucial role. Thus, the endogenously generated PtdIns(3,4,5)P 3 seems to have a very high efficiency in stimulating Rap1B activation. Although the reason for this effect is not known, it may be hypothesized that a specific and optimal localization of endogenously generated PtdIns(3,4,5)P 3 may improve its effect on Rap1B. In this regard, it is interesting to note that a preferential distribution of the PI3K-generated lipids in platelet membrane rafts has been recently reported (31). In this work, we have also provided evidence that PtdIns(3,4,5)P 3 -mediated activation of Rap1B is an important step in the G i -dependent pathway for platelet aggregation. In fact, exogenous PtdIns(3,4,5)P 3 , in addition to partially restore Rap1B activation, can overcome the inhibition of Fc␥RIIA-mediated platelet aggregation promoted by ADP scavengers. In this context, the effect of PtdIns(3,4,5)P 3 resembles that of epinephrine, which activates a truly G i -dependent pathway. Our data show that PtdIns(3,4,5)P 3 is much less efficient than epinephrine in promoting restoration of Fc␥RIIAmediated platelet aggregation in apyrase-treated platelets. However, this fits well with the reduced ability of this lipid to restore Rap1B activation.
In conclusion, we have shown that stimulation of a G i signaling cascade leads to activation of the small GTPase Rap1B through the action of the PI3K lipid product PtdIns(3,4,5)P 3 , rather than through inhibition of adenylyl cyclase. These results also suggest that Rap1B represents a downstream effector for PI3K in the G i -dependent signaling pathway for potentiation of platelet aggregation.