Thrombopoietin Complements G i - but Not G q -dependent Pathways for Integrin (cid:1) IIb (cid:2) 3 Activation and Platelet Aggregation*

Binding of thrombopoietin (TPO) to the cMpl receptor on human platelets potentiates aggregation induced by a number of agonists, including ADP. In this work, we found that TPO was able to restore ADP-induced platelet aggregation upon blockade of the G q -coupled P2Y1 puri- nergic receptor but not upon inhibition of the G i -coupled P2Y12 receptor. Moreover, TPO triggered platelet aggregation upon co-stimulation of G z by epinephrine but not upon co-stimulation of G q by the thromboxane analogue U46619. Platelet aggregation induced by TPO and G i stim- ulation was biphasic, and cyclooxygenase inhibitors prevented the second but not the first phase. In contrast to ADP, TPO was unable to induce integrin (cid:1) IIb (cid:2) 3 activation, as evaluated by binding of both fibrinogen and PAC-1 monoclonal antibody. However, ADP-induced activation of integrin (cid:1) IIb (cid:2) 3 was blocked by antagonists of the G q - coupled P2Y1 receptor but was completely restored by the simultaneous co-stimulation of cMpl receptor by TPO. Inside-out activation of integrin (cid:1) IIb (cid:2) 3 induced by TPO and G i stimulation occurred independently of thrombox- ane A 2 production and was not mediated by protein ki- nase C, MAP kinases, or Rho-dependent kinase. Impor-tantly, TPO and G i activation of integrin (cid:1) IIb (cid:2) 3 was suppressed by wortmannin and Ly294002, suggesting a critical regulation by phosphatidylinositol 3-kinase. We found that TPO did not activate phospholipase C in human platelets and was unable to restore ADP-induced phospholipase C activation upon blockade of the G q -cou- pled P2Y1 receptor. TPO induced a rapid and sustained activation of the small GTPase Rap1B through a pathway dependent on phosphatidylinositol 3-kinase. In ADP-stim-ulated platelets, Rap1B activation was reduced, although not abolished, upon blockade of the P2Y1 receptor. How-ever, accumulation of GTP-bound Rap1B in platelets activated by co-stimulation of cMpl and P2Y12 receptor was identical to that induced by the simultaneous ligation of P2Y1 and P2Y12 receptor by ADP. These results indicate that TPO can integrate G i , but not G q , stimulation and can efficiently support integrin (cid:1) IIb (cid:2) 3 activation platelet ag- gregation by an alternative signaling pathway independent of phospholipase C but involving the phosphatidylinositol 3-kinase and the small GTPase Rap1B.

Thrombopoietin (TPO) 1 is the major regulator of megakaryopoiesis and platelet production (1). The TPO receptor is the product of the proto-oncogene c-Mpl and is expressed on CD34 ϩ stem cells, megakaryocytes, and circulating platelets (1,2). The intracellular domain of c-Mpl receptor associates with the cytosolic tyrosine kinase JAK2, which is activated by trans-phosphorylation upon ligand binding (3). Activated JAK2 initiates at least three different signaling pathways involving STAT3/ STAT5, phosphatidylinositol-3-kinase (PI-3K)/Akt, and p21 ras / MAPK, respectively, to promote megakaryocyte proliferation and maturation (3,4). The same signaling pathways have also been reported to be activated upon TPO binding to human platelets. A number of studies have shown that TPO promotes the tyrosine phosphorylation of several proteins in human platelets, including JAK2 (5-7), STAT3 (5,8), Cbl (9), p85 subunit of PI-3K (10), Vav (11), and cortactin (12). In addition, activation of PI-3K, p21 ras , and MAPK has also been reported (10,13,14). Despite this, TPO is unable to promote platelet functional responses, such as secretion or aggregation, in the majority of healthy subjects. However, TPO has the intriguing ability to potentiate platelet aggregation induced by a variety of stimuli, including thrombin, ADP, and collagen (5-7, 10, 15, 16), to prime platelet aggregation induced by shear stress, and to increase platelet adhesion under flow (17,18). Although the mechanism for platelet priming by TPO is unclear, a role for PI-3K-as well as for MAPK-dependent pathways has been proposed (10, 19 -22).
Recently, it has become evident that platelet aggregation induced by many agonists results from the concomitant signaling through both G q -and G i -coupled receptors. For instance, ADP-induced platelet aggregation requires the co-activation of both G q -coupled P2Y1 and G i -coupled P2Y12 receptors (23). The P2Y1 receptor induces activation of phospholipase (PLC) ␤, leading to calcium mobilization and protein kinase C (PKC) activation, and is required for shape change and initiation of aggregation (24 -26). The P2Y12 receptor, which is classically associated with the inhibition of adenylyl cyclase through G␣ i , is responsible for the amplification of platelet response and completion of platelet aggregation (24 -27). This effect is possibly mediated by the G-protein ␤␥ dimers-mediated activation of PI-3K and the small GTPase Rap1B (28 -31). Other agonists, such as the thromboxane A 2 (TxA 2 ), the receptors of which are coupled exclusively to G q to stimulate PLC␤, are able to induce aggregation as long as secreted ADP can stimulate the G icoupled P2Y12 receptor (32,33). Moreover, replacement of G q -dependent activation of PLC␤ by stimulation of PLC␥2 through tyrosine kinase-based pathways, such as in the case of Fc␥ receptor IIA (Fc␥RIIA) clustering or glycoprotein VI stimulation, can still result in platelet aggregation providing that a G i -dependent pathway is concomitantly stimulated (34,35). Finally, the requirement of G q stimulation to induce platelet aggregation can be bypassed by treatment with PKC activators (36). In conclusion, it may be assumed that concomitant stimulation of PLC and activation of G i are critical to support agonist-induced platelet aggregation.
In this work, we investigated the possible ability of TPO to integrate G q -or G i -dependent pathways for platelet activation. We found that platelet aggregation may be triggered by concomitant activation of cMpl receptor and G i -but not G q -coupled receptors. Aggregation in response to TPO and G i stimulation occurs in the absence of PLC activation and is sustained by inside-out activation of integrin ␣ IIb ␤ 3 . Therefore, stimulation of cMpl receptor can bypass the need for PLC activation and efficiently integrate G i -dependent signaling to trigger full platelet aggregation.
Platelet Preparation-Blood was drawn from healthy volunteers who abstained from drugs for at least 2 weeks. When PRP was used for platelet aggregation, 1:10 sodium citrate (3.8%) or hirudin (500 units/ ml) was added as anticoagulant. Conversely, to prepare gel-filtered platelets, blood was drawn in 1:10 ACD (152 mM sodium citrate, 130 mM citric acid, 112 mM glucose) as anticoagulant. PRP was prepared by centrifugation of whole blood at 120 ϫ g for 10 min at room temperature. Platelets were then recovered by centrifugation of the PRP at 300 ϫ g for 10 min at room temperature and resuspended in a small volume (0.5 ml) of autologous plasma. Platelets were isolated by gel filtration on a 10-ml column of Sepharose CL-2B and eluted with HEPES buffer (10 mM HEPES, 137 mM NaCl, 2.9 mM KCl, 12 mM NaHCO 3 , pH 7.4). Platelet count was typically adjusted to 0.3 ϫ 10 9 cells/ml unless otherwise stated. For 32 P labeling, the platelet suspension (10 9 cells/ml) was incubated with 0.2 mCi/ml 32 P for 90 min at 37°C and then centrifuged at 300 ϫ g for 10 min in the presence of 2 mM EDTA. 32 P-labeled platelets were resuspended in HEPES buffer and used at the final concentration of 0.5 ϫ 10 9 platelets/ml.
Measurements of Platelet Aggregation-Samples of PRP (0.4 ml) were prewarmed at 37°C in an aggregometer under constant stirring, incubated with buffer or increasing concentrations of TPO for 3 min, and then stimulated with 2 M ADP, 1 M U46619, or 1 M epinephrine. Aggregation was monitored for at least 5 min. When indicated, AR-C69931MX (1 M) or MRS2179 (200 M) was added 2 min before stimulation. Treatment with 1 mM aspirin or 10 M indomethacin was performed for 30 min before stimulation. The baseline was set using platelet-poor plasma obtained by centrifugation of a small volume of PRP at 500 ϫ g for 20 min.
Rap1B Activation Assay-Activation of Rap1B was evaluated essentially as described previously (29,37), using the GST-tagged Rap binding domain of RalGDS (GST-RalGDS-RBD), which specifically precipitates the GTP-bound form of Rap1B from a platelet lysate. Active Rap1B was identified by immunoblotting with a specific polyclonal antibody and quantified by densitometric scanning. In parallel, the total amount of Rap1B in the whole cell lysate was also analyzed by immunoblotting.
Measurement of Cytosolic Ca 2ϩ Concentration-Platelets were prepared essentially as described above with slight modifications. PRP was incubated with 3 M FURA-2-AM at 37°C for 30 min before further processing. Platelets were then isolated by gel filtration and eluted with HEPES buffer containing 0.5% bovine serum albumin and 5.5 mM glucose. Platelet count was then adjusted to 2 ϫ 10 8 cells/ml. Measurement of cytosolic Ca 2ϩ was performed on 0.4-ml samples prewarmed at 37°C under gentle stirring in a PerkinElmer Life Sciences LS3 spectrofluorimeter in the presence of 1 mM CaCl 2 . The fluorescence excitation and emission wavelengths were 340 and 510 nm, respectively. FURA-2 fluorescence signals were calibrated according to the method of Pollock et al. (38). All determinations were repeated at least three times with platelets from different donors.
Measurement of Pleckstrin Phosphorylation-Samples of 32 P-labeled platelets (0.1 ml) treated with buffer or MRS2179 (200 M) were stimulated with 10 M ADP, 100 ng/ml TPO, or 1 unit/ml thrombin for the times indicated in Fig. 4. The reaction was stopped by the addition of an equal volume of SDS sample buffer (25 mM Tris, 192 mM glycine, pH 8.3, 4% SDS, 1% dithiothreitol, 20% glycerol, and 0.02% bromphenol blue), and samples were heated for 5 min at 96°C. Aliquots of total platelet proteins (20 l) were separated by SDS-PAGE on a 5-15% acrylamide gradient gel followed by staining with Coomassie Brilliant Blue. Gels were then dried, and phosphorylation of pleckstrin was evaluated upon autoradiography for about 18 h at Ϫ80°C. Alternatively, PKC-dependent protein phosphorylation was evaluated by immunoblotting using an anti phospho-(Ser) PKC substrate antibody. Unlabeled platelet samples (0.1 ml, 0.3 ϫ 10 9 platelets/ml) were incubated at 37°C and stimulated for 1 min with 10 M ADP and/or 100 ng/ml TPO. When indicated, 200 M MRS2179 was added 2 min before stimulation. Platelets were lysed in 2% SDS in HEPES buffer, and protein concentration was determined. Aliquots containing 50 g of total platelet lysates were heated at 96°C for 5 min in SDS sample buffer, separated on a 5-15% acrylamide gradient gel, and transferred to nitrocellulose. Blots were probed with anti-phospho-(Ser) PKC substrate antibody diluted 1:1,000. Immunoreactive bands were visualized by enhanced chemiluminescence reaction.
Analysis of Fibrinogen and PAC-1 Fab Binding to Human Platelets-Platelets were isolated from PRP by centrifugation at 300 ϫ g in the presence of 1 M prostaglandin E 1 and 1 unit/ml apyrase. Pellets were washed with 15 ml of PIPES buffer (137 mM NaCl, 20 mM PIPES, pH 6.5) and then centrifuged in the presence of 1 unit/ml apyrase. Finally, platelets were gently resuspended in HEPES buffer containing 1 mM CaCl 2 , 1 mM MgCl 2 , 0.1% bovine serum albumin, 5.5 mM glucose at the final concentration of 2 ϫ 10 7 /ml. Platelet samples (50 l) were incubated with 200 g/ml biotin-fibrinogen and 10 g/ml phycoerythrinstreptavidin in the absence or presence of 200 M MRS2179 and then treated with 10 M ADP and/or 100 ng/ml TPO for 30 min at room temperature without stirring. For PAC-1 binding, platelet samples (50 l, 0.5 ϫ 10 7 cells/ml) were incubated with 1 g/ml PAC-1 Fab antibody and 10 g/ml goat anti-mouse antibody conjugated with fluorescein and then treated with agonists, as described for fibrinogen binding. When indicated, platelets were preincubated with selected inhibitors as fol- When appropriate, platelets were treated with a corresponding volume of Me 2 SO, used as solvent for many of the inhibitors analyzed, as a control. Reactions were stopped by the addition of 0.95 ml of 0.5% (w/v) paraformaldehyde in phosphate-buffered saline, and samples were placed in the dark. To assess nonspecific binding, parallel samples were incubated with 1 mM RGDS. Fibrinogen and PAC-1 Fab binding to platelets was quantified by flow cytometry.

RESULTS
Thrombopoietin and G i Stimulation Induce Activation of Integrin ␣ IIb ␤ 3 and Platelet Aggregation-TPO potentiates platelet aggregation induced by low doses of different agonists, including ADP (5-7). Fig. 1A shows that the reversible aggregation induced by 2 M ADP is progressively converted into full irreversible aggregation by increasing concentrations of TPO. The potentiating effect of TPO is already detectable at 50 ng/ml and is maximal at higher concentrations (200 ng/ml). Aggregation induced by ADP requires the concomitant activation of the G q -coupled P2Y1 and the G i -coupled P2Y12 receptors (23). We investigated whether co-stimulation of the two purinergic receptors by ADP was also strictly necessary for TPO-induced potentiation of platelet aggregation. When the P2Y1 receptor was blocked by the specific antagonist MRS2179, ADP-induced aggregation was prevented but could be restored by the addition of TPO in a dose-dependent manner (Fig. 1B). Interestingly, a residual shape change persisted in ADP-stimulated platelets in PRP despite the presence of MRS2179. Although we have not performed an accurate analysis to try to explain this event, we may exclude that this is due to an inefficient inhibition of P2Y1 receptor as we have observed a complete suppression of PLC activation under the same experimental conditions (see Figs. 3 and 4, and data not shown). By contrast, when the G i -coupled P2Y12 receptor was blocked, platelets did not undergo aggregation in response to ADP even in the presence of a high concentration of TPO (200 ng/ml) (Fig. 1C). Identical results were also obtained using gel-filtered platelets stimulated with 10 M ADP in the presence of exogenous fibrinogen (data not shown). Moreover, upon blockade of the P2Y1 receptor, platelet aggregation profiles very similar to those reported in Fig. 1B were recorded when ADP and TPO were added simultaneously or when treatment with ADP preceded the addition of TPO (data not shown), indicating that the observed effects are independent of the order of agonists addition. These findings suggest that TPO can replace stimulation of G q , but not of G i , to induce platelet aggregation. To confirm our results, we analyzed epinephrine-induced aggregation. Binding of epinephrine to ␣2A-adrenergic receptor (which is coupled to G z , a member of the G i family) is unable to induce platelet aggregation in hirudin-treated PRP. However, concomitant stimulation of ␣2A-adrenergic receptor by epinephrine and cMpl receptor by TPO resulted in a full platelet aggregation (Fig. 1D). By contrast, selective stimulation of the G qcoupled receptor for the TxA 2 (pursued by the addition of the stable analogue U46619 in the presence of the P2Y12 receptor antagonist AR-C69931MX) did not result in platelet aggregation, either in the absence or in the presence of a high concentration of TPO (Fig. 1E). Platelet aggregation induced by a combination of TPO-and G i -dependent signaling constantly showed a biphasic pattern, suggesting a possible contribution of secondary messengers such as TxA 2 . We therefore investigated the ability of TPO to restore ADP-induced aggregation upon blockade of the P2Y1 receptor in aspirin-or indomethacin-treated platelets. Fig. 1F shows that the inhibitors of cyclooxygenase prevented the second, but not the first, phase of platelet aggregation induced by co-stimulation of P2Y12 and cMpl receptors. Therefore, the rapid, initial phase of platelet aggregation is independent of TxA 2 generation. Although potentiation of ADP-induced platelet aggregation was evident at concentrations of TPO as low as 25 ng/ml (Figs. 1B and 1D), further studies had been performed with higher doses of TPO (typically 100 ng/ml) to better reveal and to more reliably detect even weakly activated signaling events.
Platelet aggregation is mediated by fibrinogen binding to activated integrin ␣ IIb ␤ 3 . To evaluate the ability of TPO to integrate G i -dependent signals for inside-out activation of integrin ␣ IIb ␤ 3 , we measured binding of the activation-dependent antibody PAC-1, as well as of biotinylated-fibrinogen, to stimulated platelets under non-aggregating conditions (i.e. in the absence of stirring). Fig. 2 shows that ADP, but not TPO alone, was able to activate integrin ␣ IIb ␤ 3 and to stimulate binding of both fibrinogen and PAC-1 antibody. ADP-induced activation of integrin ␣ IIb ␤ 3 required the stimulation of the P2Y1 receptor as it was strongly inhibited by MRS2179. However, in the presence of MRS2179, ADP-induced activation of integrin ␣ IIb ␤ 3 was completely restored by pretreatment with TPO. Therefore, TPO integrates signals from the G i -coupled P2Y12 ADP receptor to induce inside-out activation of integrin ␣ IIb ␤ 3 . Binding of fibrinogen induced by ADP plus TPO in the presence of the P2Y1 receptor antagonist MRS2179, as well as by ADP alone, was unaffected by cyclooxygenase inhibitors aspirin and indomethacin (see Fig. 6, and data not shown), indicating that the ability of TPO to restore ADP-induced integrin ␣ IIb ␤ 3 activation in the absence of G q signaling is independent of TxA 2 production.
In the Absence of G q -Stimulation, Thrombopoietin Does Not Restore Phospholipase C Activation Induced by ADP-Our results indicate that G q -dependent signaling for integrin ␣ IIb ␤ 3 activation and initial platelet aggregation can be replaced by the addition of TPO as long as a G i -dependent pathway is stimulated. The G q -coupled P2Y1 receptor contributes to ADPinduced platelet activation through the stimulation of PLC␤. We therefore verified whether TPO was able to restore ADPinduced activation of PLC upon blockade of the P2Y1 receptor, thus resulting in an increase of the intracellular Ca 2ϩ concentration and in the activation of PKC. The intracellular Ca 2ϩ concentration was measured in FURA-2-loaded platelets. As shown in Fig. 3A, TPO alone was not able to induce any Ca 2ϩ movement in platelets. By contrast, stimulation of platelets with 10 M ADP induced an evident increase in the intracellular Ca 2ϩ concentration, which was completely prevented by blockade of the P2Y1 receptor by MRS2179 (Figs. 3B and 3C). Fig. 3D shows that no intracellular Ca 2ϩ increase was detected in platelets preincubated with a high concentration of TPO (100 ng/ml) and then stimulated with ADP in the presence of the P2Y1 receptor antagonist MRS2179.
We next analyzed the phosphorylation of pleckstrin, the main platelet substrate for PKC. Using 32 P-labeled platelets, we detected a faint but evident phosphorylation of pleckstrin in platelets stimulated with 10 M ADP. ADP-induced phosphorylation of pleckstrin was totally prevented by inhibition of the P2Y1 receptor with MRS2179 and was not restored by pretreatment of platelets with TPO (Fig. 4A). We also found that prolonged incubation of platelets with a high concentration of TPO alone (100 ng/ml) was unable to induced pleckstrin phosphorylation. Because of the faint phosphorylation of pleckstrin detected in ADP stimulated 32 P-labeled platelets, we decided to confirm these results by using a different strategy. Pleckstrin phosphorylation was investigated by immunoblotting using an antibody specific for the serine-phosphorylated PKC substrates. This approach appeared to be a more reliable technique to detect the phosphorylation of pleckstrin (identified as a major reactive protein of about 47 kDa) and allowed the concomitant detection of other proteins phosphorylated by PKC in ADP-stimulated platelets (Fig. 4B). We confirmed that PKCdirected protein phosphorylation in ADP-stimulated platelets was prevented by the P2Y1 antagonist and was not restored by treatment with TPO (Fig. 4B). Taken together, these results indicate that platelet aggregation induced by stimulation of G i in the presence of TPO occurs independently of PLC activation.
Thrombopoietin Induces Activation of Rap1B-The small GTP-binding protein Rap1B has been proposed to play a role in integrin ␣ IIb ␤ 3 activation and platelet aggregation (39,40). We therefore investigated the activation of Rap1B induced by ADP and TPO in human platelets. We found that TPO alone was able to induce GTP binding to Rap1B. This effect was rapid, being evident after 30 s of stimulation and persisted at least for 10 min (Fig. 5A). Moreover, TPO-induced activation of Rap1B was independent of TxA 2 production (see Fig. 7A). TPO was revealed to be a weaker activator of Rap1B when compared with other platelet agonists such as thrombin, ADP, or epinephrine (Fig. 5B, and data not shown). It has been previously shown that ADP-induced activation of Rap1B is largely promoted by the P2Y12 receptor and is only partially regulated by the P2Y1 receptor (29,30,41). In agreement with these find-ings, Fig. 5, B and C, show the partial inhibition of ADPinduced activation of Rap1B by the P2Y1 receptor antagonists MRS2179 and A3P5P. Moreover, we found that activation of Rap1B induced by TPO and G i stimulation was additive since treatment of platelets with TPO and ADP in the presence of either MRS2179 or A3P5P resulted in a level of Rap1B activation comparable with that observed upon concomitant stimulation of P2Y1 and P2Y12 receptor by ADP alone (Fig. 5, B and  C). Therefore, TPO appeared to be able to completely restore Rap1B activation induced by ADP when the P2Y1 receptor was blocked. We have also found that TPO can potentiate the weak Rap1B activation induced by epinephrine (Fig. 5D), confirming a synergistic effect of TPO and G i stimulation on Rap1B regulation.
Role of PI-3K in Thrombopoietin-mediated Activation of Rap1B and Complementation of G i -dependent Pathway for integrin ␣ IIb ␤ 3 Activation-To get further insights into the signaling pathways involved in the complementation of G i signaling by TPO, we tested the ability of a number of pharmacological inhibitors to prevent fibrinogen binding to platelets upon concomitant stimulation of cMpl and P2Y12 receptors. Some of these inhibitors had been selected because of their ability to interfere with intracellular effectors that had been shown to be activated by TPO. Among these, there are the cyclooxygenase inhibitors aspirin and indomethacin, the MAP kinase inhibitors SB203580 and PD98058, and the PI-3K inhibitors Ly294002 and wortmannin. Other compounds, such as the Src kinase inhibitor PP2, the Rho kinase inhibitor Y27632, the PKC inhibitor Ro318220, and the intracellular Ca 2ϩ chelating agent BAPTA-AM, affect pathways that had not been previously reported to be activated by TPO. Although many of the drugs tested caused a modest reduction of fibrinogen binding to stimulated platelets, a marked inhibition was observed upon treatment with the intracellular Ca 2ϩ chelator BAPTA-AM and with the two structurally unrelated PI-3K inhibitors, wortmannin and Ly294002 (Fig. 6). Since TPO does not induce PLC activation and intracellular calcium increase, these results reveal a critical role for PI-3K in mediating the effects of TPO. We next investigated the ability of the same panel of inhibitors to affect the TPO-induced restoration of platelet aggregation upon stimulation with ADP in the presence of MRS2179. Results are reported in Fig. 6B. It is clear that inhibition of MAPK, Rho kinase, or PKC does not significantly alter platelet aggregation induced by TPO plus ADP upon blockade of the P2Y1 receptor. By contrast, platelet aggregation was totally suppressed by the two PI-3K inhibitors tested and significantly reduced by BAPTA-AM. Therefore, there is a general good correlation between the effect of the tested inhibitors on agonist-induced fibrinogen binding and platelet aggregation. It may be interesting to note that the Src kinase inhibitor PP2 displayed a more pronounced effect on platelet aggregation rather than on fibrinogen binding. This may be a consequence of the fact that fibrinogen binding has been measured under non-stirring conditions, whereas aggregation was monitored on stirred platelets. Moreover, Src kinases have been shown to be involved in the outside-in signaling through integrin ␣ IIb ␤ 3 , and may contribute to the consolidation of aggregation (42). Previous works have reported that the PI-3K inhibitors, as well as intracellular Ca 2ϩ chelators, inhibit ADP-induced activation of Rap1B (29,30). Fig. 7A shows that Ly294002, wortmannin, and BAPTA-AM, but not aspirin or indomethacin, inhibited Rap1B activation induced by TPO. This indicates that TPO stimulates Rap1B activation through a PI-3K-and Ca 2ϩ -dependent pathway. When platelets were stimulated with TPO and ADP in the presence of the P2Y1 receptor antagonist MRS2179, activation of Rap1B was prevented by PI-3K inhibitors wortmannin and Ly294002 and by the Ca 2ϩ chelating agent BAPTA-AM (Fig. 7B). By contrast, no reduction of Rap1B activation was observed in platelets treated with the cyclooxygenase inhibitors aspirin and indomethacin or with Y27632, Ro318220, or PD98058, whereas a modest inhibition was observed with the Src kinase inhibitor PP2. All together, these results strengthen the correlation between activation of Rap1B and integrin ␣ IIb ␤ 3 -dependent platelet aggregation upon stimulation of cMpl and P2Y12 receptors. DISCUSSION It is known that platelet aggregation induced by ADP requires the concomitant activation of G q -and G i -mediated signaling pathways (23)(24)(25)(26)(27). In the present work, we have demonstrated that co-stimulation of the G i -coupled P2Y12 receptor for ADP and the cMpl receptor for TPO induces integrin ␣ IIb ␤ 3 activation and platelet aggregation in the absence of G q -mediated signaling. Our results demonstrate that stimulation of G q is not strictly required for integrin ␣ IIb ␤ 3 activation in human platelets and confirm the essential role for G i stimulation. In addition, our work describes a new pathway for integrin ␣ IIb ␤ 3 -dependent platelet aggregation, based on the integration of G i -mediated signaling by stimulation of the cMpl receptor.
Stimulation of the cMpl receptor by TPO initiates a tyrosine kinases-mediated signaling pathway (1,4). Previous work has reported that stimulation of tyrosine kinases downstream engagement of immunoreceptors, such as Fc␥RIIA, can induce integrin ␣ IIb ␤ 3 activation and platelet aggregation as long as a G i -coupled receptor is concomitantly activated (34). There are, however, dramatic differences between Fc␥RIIA-and cMpl receptor-dependent signaling. Clustering of Fc␥RIIA stimulates the tyrosine kinase Syk, which phosphorylates and activates PLC␥2, and eventually leads to intracellular calcium increase and PKC activation. The same intracellular messengers are also produced upon G q stimulation through the activation of PLC␤ isoforms. Activation of different PLC enzymes through G q -or tyrosine kinase-based pathways occurs independently of G i stimulation, although activation of the G icoupled P2Y12 receptor by ADP has been shown to potentiate intracellular Ca 2ϩ increase promoted by stimulation of both Fc␥RIIA-or G q -coupled P2Y1 receptor (29,34,43). Therefore, it FIG. 4. Analysis of pleckstrin phosphorylation. A, 32 P-labeled platelets were stimulated with 10 M ADP, with 100 ng/ml TPO, or with both agonists in the absence or presence of 200 M MRS2179 for times ranging from 30 s to 6 min, as indicated. A control sample was stimulated with 1 unit/ml thrombin (THR) for 30 s. Platelet proteins were separated by SDS-PAGE on a 5-15% acrylamide gradient gel, and phosphorylated pleckstrin was identified by autoradiography. The positions of the molecular mass markers are indicated on the left. The arrow on the right indicates the phosphorylated pleckstrin. bas, resting. B, unlabeled platelet samples were incubated in the absence or presence of 200 M MRS2179 and then stimulated with 10 M ADP, 100 ng/ml TPO, or both agonists for 1 min. Upon platelet lysis, proteins were separated by SDS-PAGE on a 5-15% acrylamide gradient gel and transferred to nitrocellulose. Pleckstrin phosphorylation was analyzed by immunoblotting using a specific antibody against the serine-phosphorylated PKC substrates. The arrow on the right indicates the migration of phosphorylated pleckstrin. On the left, the positions of molecular mass markers are reported. may be considered that the tyrosine kinase-dependent pathway downstream Fc␥RIIA can integrate G i signaling for platelet aggregation because it can substitute G q -dependent signaling for PLC activation. By contrast, activation of cMpl receptor by TPO stimulates the tyrosine kinase JAK2 and leads to the phosphorylation of the transcriptional factor STAT3, and through a still poorly characterized pathway, to the activation of PI-3K and Ras (1, 4). TPO does not activate any PLC isoform, and we have shown in this work that TPO does not activate PLC even when G i signaling is concomitant activated. Therefore, the onset of intracellular mediators leading to integrin ␣ IIb ␤ 3 activation are different in platelets stimulated through cMpl and P2Y12 receptors than in cells stimulated through Fc␥RIIA/P2Y1 receptor and P2Y12 receptor. Our results indicate that when the P2Y1 receptor is blocked, activation of integrin ␣ IIb ␤ 3 may be stimulated by ADP in the presence of TPO, through an alternative intracellular pathway, which is independent of PLC-derived second messengers. Although previous works have proposed the essential role for PLC activation in platelet aggregation, recent evidence indicates that platelet aggregation can be triggered by co-stimulation of G i -and G 12 / 13 -coupled receptors, under conditions in which activation of PLC is not supposed to occur (44). Moreover, complementation of G i -dependent pathway for platelet aggregation can be  6). D, gel-filtered platelets were stimulated with buffer (none), 100 ng/ml TPO, 1 M epinephrine (Epi), or both agonists for 1 min at 37°C, and activation of Rap1B was analyzed by a pull-down assay followed by immunoblotting with anti-Rap1 antibody.
obtained by an antibody against glycoprotein VI in the absence of any detectable intracellular calcium rise (35).
The signaling pathways activated by TPO and able to integrate G i stimulation to trigger platelet aggregation are not completely clear. TPO has been reported to stimulate cPLA 2 in platelets and to lead to the production of TxA 2 (12,22), the receptor of which is coupled to both G q and G 12/13 , and could, thus, be responsible for integration of G i -dependent signaling (32,33,44). We have demonstrated in this work that TxA 2 is actually involved in the second phase of platelet aggregation induced by a combination of TPO and P2Y12 receptor stimulation. However, the initial, rapid phase of platelet aggregation occurred normally in platelets treated with the cyclooxygenase inhibitors aspirin and indomethacin. Moreover, concomitant stimulation of cMpl and P2Y12 receptors induce activation of integrin ␣ IIb ␤ 3 and fibrinogen binding under non-aggregating conditions independently of TxA 2 . It is therefore likely that the TxA 2 formation, responsible for the second irreversible phase of platelet aggregation in platelets activated by TPO and G i stimulation, is promoted by integrin ␣ IIb ␤ 3 -dependent outside-in signaling during the primary phase of aggregation, rather than by direct signaling through cMpl or P2Y12 receptors. Similarly, it is very well known that TxA 2 generation in ADP-stimulated platelets is a consequence of the primary aggregation rather than purinergic receptor-mediated cPLA 2 activation (45,46).
Because of the production of TxA 2 during platelet aggregation, studies on the intracellular messenger activated in platelets stimulated by TPO and G i have been performed under non-aggregating conditions. We have found that TPO can induce activation of the small GTPase Rap1B and that activation of Rap1B induced by concomitant stimulation of cMpl and P2Y12 receptors is similar to that induced by concomitant stimulation of P2Y1 and P2Y12 receptors by ADP. Rap1B has been demonstrated to be involved in integrin ␣ IIb ␤ 3 activation (39,40), and we propose that, upon blockade of G q -coupled P2Y1 receptor, restoration of the normal amount of activated Rap1B by TPO contributes to the re-establishment of a functional signaling pathway for integrin ␣ IIb ␤ 3 activation. It is known that the level of Rap1B activation in ADP-treated platelets is largely dependent on P2Y12 receptor stimulation and is regulated only by about 20 -30% through the P2Y1 receptor (29,30,41). However, it is possible that a threshold level of active GTP-bound Rap1B is required to support integrin ␣ IIb ␤ 3 activation and that such a threshold is not reached when the sole P2Y12 receptor is stimulated by ADP. In this regard, it is interesting to note that very high concentrations of ADP can promote some platelet aggregation in P2Y1-deficient mice (28). We may not exclude, of course, that in addition to the restoration of the normal level of GTP-bound Rap1B, other intracel-lular effectors stimulated by TPO participate in integrin ␣ IIb ␤ 3 activation in the presence of G i -dependent signaling. In fact, although Rap1B has been recognized to be important for integrin ␣ IIb ␤ 3 activation, studies with mouse megakaryocytes or with platelets from Rap1B-deficient mice indicate that this protein is not absolutely required for this event and that both fibrinogen binding and platelet aggregation are reduced but still occur in the absence of Rap1B signaling (39,47). However, we tested a number of pharmacological inhibitors able to affect pathways potentially activated downstream cMpl for their ability to prevent TPO-mediated complementation of G i signaling, and we detected a relevant inhibitory effect on integrin ␣ IIb ␤ 3 activation exclusively under conditions in which Rap1B activation was also prevented. In particular, our data suggest a role for PI-3K and for intracellular Ca 2ϩ in TPO-mediated activation of Rap1B and integrin ␣ IIb ␤ 3 . The observed inhibitory effect of the Ca 2ϩ chelating agent BAPTA-AM is quite intriguing since we have found that TPO does not induce activation of PLC and intracellular Ca 2ϩ increase. Similarly, an inhibitory effect of BAPTA-AM on Rap1B activation downstream G i stimulation has been shown previously, although G i is not able, per se, to induce any intracellular Ca 2ϩ rise (29,30). Although there is no definitive explanation for these observations, it is interesting to note that a recent study with knock-out mice has identified the Ca 2ϩ -dependent CalDAG-GEFI as the exchange factor responsible for Rap1B activation in ADP-stimulated platelets (48). It is possible to speculate that signaling through G i or through the cMpl receptor reduces the Ca 2ϩ requirement of CalDAG-GEFI, rendering this factor active at the basal level of Ca 2ϩ . By lowering the basal level of this ion inside the platelet, BAPTA-AM may therefore prevent CalDAG-GEFI activity upon stimulation with ADP or TPO. In this regard, we have found that the inositol 1,4,5-trisphosphate receptor antagonist 2-APB, which, like BAPTA-AM, efficiently prevents ADP-induced increase of intracellular Ca 2ϩ concentration, but in contrast to BAPTA-AM, does not lower the basal level of this ion, does not inhibit Rap1B activation induced by ADP or TPO (data not shown). This finding supports the idea that the apparent paradox of the effect of BAPTA-AM on Rap1B activation may be the consequence of the reduction of the basal level of Ca 2ϩ , rather than the inhibition of the increase of its concentration upon platelet stimulation. Whatever the mechanism for BAPTA-AM-mediated inhibition of Rap1B activation under conditions in which no increase of intracellular Ca 2ϩ is detected, it is interesting to note that BAPTA-AM also prevents fibrinogen binding and platelet aggregation upon stimulation through cMpl and P2Y12 receptors, strengthening the correlation between Rap1B and integrin ␣ IIb ␤ 3 activation.
We have found that the PI-3K inhibitors Ly294002 and wortmannin prevented both Rap1B and integrin ␣ IIb ␤ 3 activation in platelets treated with TPO and ADP. It should be noted, however, that upon inhibition of PI-3K, a reduced, but evident, level of fibrinogen binding was still detectable, despite the parallel total inhibition of Rap1B activation. This may be explained, considering that Rap1B, albeit important, is not the only regulator of integrin ␣ IIb ␤ 3 activation, as also indicated by previous studies (39,47). In addition, the observed effect of PI-3K inhibitors, as well as of BAPTA-AM, on platelet aggregation induced by TPO plus ADP is indicative of a much more complex cross-talk and interplay among different signal transduction pathways. In this work, we have also identified PI-3K as a key regulator of Rap1B activation induced by TPO alone. Previous works have reported that PI-3K also mediates Rap1B activation downstream of G i (30,31). Therefore, our results indicate that G i -coupled receptors and cMpl receptor exploit similar signaling pathways for Rap1B stimulation. The inhibitory effects of PI-3K antagonists on integrin ␣ IIb ␤ 3 activation may thus result from the blockade of events stimulated downstream both receptors. Although this consideration may rise concerns as to whether PI-3K is the real and only mediator of the potentiating effects of TPO, it clearly outlines the crucial role for this enzyme for integrin ␣ IIb ␤ 3 stimulation and Rap1B activation.
In conclusion, our results demonstrate that stimulation of cMpl receptor by TPO is able to integrate signals from the G i -coupled P2Y12 ADP receptor, leading to integrin ␣ IIb ␤ 3 activation and platelet aggregation. These effects occur in the absence of PLC activation, and thus, do not involve intracellular Ca 2ϩ increase or PKC activation but are critically regulated by PI-3K and by the small GTPase Rap1B. Therefore, the G q -dependent contribution for full platelet activation can be successfully replaced by an alternative signaling pathway activated downstream of a cytokine receptor. Moreover, these findings confirm the crucial role of G i stimulation in promoting platelet activation and highlight the versatility of anti-thrombotic strategies targeting the P2Y12 ADP receptor.