Costimulation of Gi- and G12/G13-mediated Signaling Pathways Induces Integrin αIIbβ3 Activation in Platelets*

Platelet activation is a complex process induced by a variety of stimuli, which act in concert to ensure the rapid formation of a platelet plug at places of vascular injury. We show here that fibrillar collagen, which initiates platelet activation at the damaged vessel wall, activates only a small fraction of platelets in suspension directly, whereas the majority of platelets becomes activated by mediators released from collagen-activated platelets. In Gαq-deficient platelets that do not respond with activation of integrin αIIbβ3 to a variety of mediators like thromboxane A2 (TXA2), thrombin, or ADP, collagen at high concentrations was able to induce aggregation, an effect that could be blocked by antagonists of the TXA2 or P2Y12 receptors. The activation of TXA2 or P2Y12 receptors alone, which in Gαq-deficient platelets couple to G12/G13 and Gi, respectively, did not induce platelet integrin activation or aggregation. However, concomitant activation of both receptors resulted in irreversible integrin αIIbβ3-mediated aggregation of Gαq-deficient platelets. Thus, the activation of G12/G13- and Gi-mediated signaling pathways is sufficient to induce integrin αIIbβ3 activation. Although Gq-mediated signaling plays an important role in platelet activation, it is not strictly required for the activation of integrin αIIbβ3. This indicates that the efficient induction of platelet aggregation through G-protein-coupled receptors is an integrated response mediated by various converging G-protein-mediated signaling pathways involving Gq and Gi as well as G12/G13.

Under normal conditions, platelets circulate freely in the blood and do not adhere to each other. At sites of vascular damage, however, platelets adhere to subendothelial surfaces and become activated. Platelet activation involves a rapid change of shape followed by degranulation and integrin ␣ IIb ␤ 3mediated aggregation. The activation of platelets is induced by various extracellular stimuli and involves positive feedback loops, which in a complex process ensure the rapid formation of a platelet plug. Collagen, exposed at subendothelial surfaces at sites of vascular injury, induces platelet activation initially by binding to glycoprotein VI, which through the associated FcR␥ chain signals in a G-protein-independent manner (1). Full platelet aggregation by collagen and subsequent recruitment of platelets into a growing platelet plug, however, requires the formation and release of mediators like thromboxane A 2 (TXA 2 ) 1 and ADP. The transmembrane receptors of TXA 2 and ADP couple to heterotrimeric G proteins, which represent central mediators of platelet activation. TXA 2 receptors couple to G q and to G 12 /G 13 (2,3), whereas ADP acts through two G-protein-coupled receptors, P2Y 1 , which couples to G q , and P2Y 12 , which couples to G i (4,5).
Some of the roles of individual G-protein-mediated signaling pathways in platelet activation have been described during recent years. It is generally believed that G q -mediated phospholipase C␤ activation plays an essential role in platelet activation as demonstrated by the phenotype of G␣ q -deficient platelets, which fail to aggregate and secrete in response to thrombin, ADP, and the TXA 2 mimetic U46619 because of a lack of agonist-induced phospholipase C activation (6). Evidence for a role of G q -independent processes in the regulation of platelet activation emerged from studies describing the mechanism of ADP-induced platelet activation. Platelets deficient in P2Y 1 or in which P2Y 1 was blocked pharmacologically do not aggregate in response to low and intermediate concentrations of ADP (7)(8)(9)(10)(11). Aggregation could be restored by serotonin, which induces G q -mediated phospholipase C activation but alone is not able to induce platelet aggregation (8,9,11). Platelet activation by ADP obviously requires an additional signal, which appears to be mediated by the G i -coupled P2Y 12 receptor (5). Platelets deficient in P2Y 12 or in which P2Y 12 was blocked did not aggregate in response to ADP unless the G imediated pathway was activated by adrenaline, which itself is unable to induce platelet aggregation (8,12). Thus, G q and G i can synergize to induce platelet aggregation, and full platelet activation in response to ADP at intermediate concentrations requires concomitant activation of both G-proteins.
G-proteins of the G 12 family, G 12 and G 13 , which are activated through TXA 2 and thrombin receptors (2,3), have been involved in the induction of the platelet shape change. This is mainly based on the finding that in G␣ q -deficient platelets in which TXA 2 receptors only couple to G 12 and G 13 , U46619 still induces a rapid shape change. This effect appears to be mediated by the Rho/Rho kinase pathway (3). Here we demonstrate a role for G 12 /G 13 -mediated signaling in platelet aggregation and show that in the absence of G␣ q , TXA 2 , and ADP or adrenaline can induce integrin ␣ IIb ␤ 3 -dependent platelet aggregation by concomitant activation of G 12 /G 13 through TXA 2 receptors and of G i via P2Y 12 or ␣ 2 -adrenergic receptors, whereas each of the stimuli given alone was without effect. Thus, G 12 /G 13mediated as well as G i -mediated signaling processes can synergize to activate the platelet fibrinogen receptor.

EXPERIMENTAL PROCEDURES
Animals-Mutant mice deficient in G␣ q were produced as described previously (6). G␣ q -deficient and wild-type littermates, which were of 129/Sv ϫ C57BL/6 genetic background were used for experiments.
Chemicals-The ATP analog AR-C69931MX was a generous gift from ASTRA Charnwood (Loughborough Leics, England). ADP, adrenaline, essentially fatty acid-free bovine serum albumin, and human fibrinogen were from Sigma. The thromboxane analog U46619 was purchased from Alexis Biochemicals (Grü nberg, Germany), and fibrillar type I collagen (Horm) from equine tendon was from Nycomed (Munich, Germany). The thromboxane A 2 receptor antagonist SQ29548 was from Biomol (Hamburg, Germany), and the Rho kinase inhibitor Y-27632 was kindly provided by Yoshitomi Pharmaceutical Industries, Ltd. (Saitama, Japan).
Antibodies-JON/A PE , which selectively binds to activated mouse ␣ IIb ␤ 3 integrin (13), was produced and modified in our laboratory. Fluorescein isothiocyanate-conjugated anti-P-selectin and control IgG antibodies were purchased from BD Pharmingen. Unlabeled JON/A was used to block ␣ IIb ␤ 3 in aggregometry as described previously (13).
Platelet Preparation-Mice were bled under ether anesthesia from the retroorbital plexus. Blood was collected in a tube containing 10% (v/v) 7.5 units/ml heparin, and platelet-rich plasma was obtained by centrifugation at 300 ϫ g for 10 min at room temperature. Platelet-rich plasma was removed and centrifuged at 1570 ϫ g for 10 min. The platelet pellet was washed twice in Tyrode's buffer (137 mmol/liter NaCl, 2 mmol/liter KCl, 12 mmol/liter NaHCO 3 , 0.3 mmol/liter NaH 2 PO 4 , 5.5 mmol/liter glucose, 5 mmol/liter Hepes, pH 7.3) containing 0.35% bovine serum albumin and finally resuspended at a density of 5 ϫ 10 5 platelet/l in the same buffer in the presence of 0.02 units/ml ADP scavenger apyrase, a concentration sufficient to prevent desensitization of platelet ADP receptors during storage. Platelets were kept at 37°C throughout all experiments. To block thromboxane A 2 or P2Y 12 receptors, platelets were incubated with 10 M SQ29548 or 10 M AR-C69931MX, respectively, for 10 min at 37°C before the start of the experiment. To inhibit Rho kinase activity, platelets were incubated with Y-27632 (10 M) for 30 min before the start of the experiment.
Platelet Aggregation-To determine platelet aggregation, light transmission was measured using washed platelets adjusted to a platelet concentration of 3 ϫ 10 5 platelets/l with Tyrode's buffer containing CaCl 2 (1 mM) and human fibrinogen (0.2 mg/ml). Transmission was recorded on a Fibrintimer 4-channel aggregometer (APACT Laborgerä te und Analysensysteme, Hamburg, Germany) and is expressed as arbitrary units with 100% transmission adjusted with plasma. Agonists were added as 20 -100-fold concentrates.
Measurement of Secretion and TXB 2 Formation-For determination of platelet degranulation, platelets were loaded with serotonin by incubation of platelet-rich plasma for 1 h at 37°C with 2 Ci/ml [ 3 H]serotonin (80 -130 Ci/mmol). Thereafter, platelets were washed once in Tyrode's buffer, and platelets were adjusted at 2 ϫ 10 5 /ml. Platelets were exposed for 3 min to the indicated stimuli and chilled. After centrifugation, the secreted serotonin was determined by scintillation counting of the supernatants. TXB 2 formation was determined using an enzyme immunoassay (Cayman, Ann Arbor, MI). 2.5 ϫ 10 6 washed platelets were incubated for 3 min in the presence of the indicated stimuli. After centrifugation, TXB 2 levels in supernatants were determined according to manufacturer's instructions.
Flow Cytometry-Washed platelets were adjusted to 2 ϫ 10 4 platelets/l with Tyrode's buffer, and 50 l of this dilution were stimulated with the indicated agonists for 5 min at 37°C. Samples were then stained with fluorophore-labeled monoclonal antibodies for 10 min at room temperature and directly analyzed on a FACScalibur (BD Pharmingen). Platelets were gated by forward/side scatter (FSC/SSC) characteristics.

RESULTS AND DISCUSSION
We have previously shown that G␣ q -deficient platelets display defective activation by collagen (6) although primary signaling through the activating platelet collagen receptor, GPVI/ FcR␥, is not affected by the absence of G␣ q (14). This may be because of the fact that fibrillar collagen is an insoluble macromolecule and, therefore, only a small fraction of platelets comes into direct contact with collagen (15), whereas the majority of platelets exposed to collagen may be activated secondarily by mediators like ADP and TXA 2 , which are released from the collagen-responsive fraction of platelets. To test this possibility, we performed flow cytometric analysis of single platelets in diluted suspensions under experimental conditions that largely exclude the accumulation of released mediators. Only 11% of the wild-type platelets were directly activated by collagen at high concentrations (50 g/ml) as shown by the activation of ␣ IIb ␤ 3 using the JON/A PE antibody (13) and by degranulation-dependent exposure of P-selectin (Fig. 1). In contrast, Ͼ99% of platelets are activated upon exposure to stimuli like ADP and TXA 2 under the same conditions (data not shown). The primary activation of a subpopulation of platelets by fibrillar collagen was not impaired in G␣ q -deficient platelets of which also ϳ11% responded to collagen. Thus, a small fraction of platelets is activated by collagen directly and subsequently releases soluble agonists like ADP, TXA 2 , and other mediators, which in turn activate the majority of platelets, resulting in the full aggregation response. This mechanism mediates not only collageninduced activation of platelets in suspension as studied here but most probably also underlies the initial activation of platelets by subendothelial collagen at the damaged vessel wall in vivo.
The amplification of collagen-induced platelet activation by mediators like ADP and TXA 2 is thought to be largely G q - , or the combination of both agonists for 2 min at 37°C, stained with fluorophore-labeled antibodies for 10 min, and directly analyzed on a FACScalibur. Platelets were gated by FSC/SSC characteristics. Where indicated, platelets were incubated with the P2Y 12 antagonist AR-C69931MX (10 M) for 10 min before the addition of the agonists. In G␣ q -deficient platelets, P-selectin was not expressed under any condition in these experiments (data not shown). C, the shaded area represents the staining obtained with JON/A PE in resting platelets, and the solid line represents those obtained on stimulation with the indicated agonists. The results shown are representative of 12 individual experiments. Note: the levels of ␣ IIb ␤ 3 activation in response to U46619 ϩ ADP are significantly lower in G␣ q -deficient platelets (black line) as compared with wild-type platelets (gray line). D, washed G␣ q -deficient platelets (3 ϫ 10 5 /l) were preincubated with vehicle or AR-C69931MX for 10 min and subsequently stimulated with U46619 (1 M) followed by ADP (5 M). Aggregation was recorded on an APACT 4-channel aggregometer. Where indicated, the experiment was performed in the presence of a blocking antibody against integrin ␣ IIb ␤ 3 (30 g/ml). The results shown are representative of 6 -9 individual experiments. dependent, because both ADP and TXA 2 fail to induce platelet aggregation in the absence of G␣ q (6). To test this hypothesis, we studied aggregation of G␣ q -deficient platelets in response to increasing concentrations of fibrillar collagen. Although low concentrations of collagen did not induce aggregation of platelets in the absence of G␣ q , collagen at concentrations of Ͼ10 g/ml induced robust aggregation of G␣ q -deficient platelets, which occurred with a pronounced shape change (as shown by the decrease in light transmission before the start of aggregation) (Fig. 2A). Whereas collagen was much less potent in G␣ q -deficient platelets compared with wild-type platelets, the maximal efficacy appeared to be only moderately affected by the absence of G q -mediated signaling. The aggregation of G␣ qdeficient platelets in response to high collagen concentrations was accompanied by serotonin release and the formation of TXA 2 (Fig. 2, B and C), whereas collagen at low concentrations was unable to induce significant release or formation of mediators in the absence of G␣ q . These results suggested that soluble agonists released from collagen-activated platelets can induce shape change and aggregation in the absence of G␣ q .
In G␣ q -deficient platelets, the receptors for ADP and TXA 2 couple to G i and G 12 /G 13 , respectively. Therefore, we studied the role of these signaling pathways in the overall process of platelet activation/aggregation. To test the significance of TXA 2 receptor-mediated activation of G 12 /G 13 for the observed shape change and aggregation response in collagen-stimulated G␣ qdeficient platelets, we inhibited the platelet TXA 2 receptor by the specific antagonist SQ29548 (16). As evaluated by light transmission, this treatment completely abolished the collageninduced shape change of G␣ q -deficient platelets (Fig. 2D). Thus, the activation of G 12 /G 13 through TXA 2 receptors plays a major role in this process. Unexpectedly, the inhibition of the TXA 2 receptor also markedly reduced the aggregation response, suggesting that G 12 /G 13 stimulation is involved in the activation of the platelet integrin ␣ IIb ␤ 3 . On the other hand, the inhibition of the G i -coupled P2Y 12 receptor by the specific antagonist AR-C69931MX (10 M) had no influence on the shape change but also abolished the aggregation response (Fig. 2D).
The finding that both TXA 2 and ADP are required for collagen-induced platelet aggregation in the absence of G␣ q sug- FIG. 4. Effect of U46619 and adrenaline on G␣ q -deficient and wild-type platelets. A, washed G␣ q -deficient platelets (2 ϫ 10 4 /l) were stimulated with U46619 (1 M), adrenaline (10 M) (Adr), or the combination of both agonists for 2 min at 37°C, stained with fluorophore-labeled antibodies for 10 min, and directly analyzed on a FACScalibur. The shaded area represents the staining obtained with JON/A PE in resting platelets, and the solid line represents staining obtained upon stimulation with the indicated agonists. P-selectin was not expressed under any condition on these platelets (data not shown). Note: the levels of ␣ IIb ␤ 3 activation in response to U46619 ϩ adrenaline are significantly lower in G␣ q -deficient platelets (black line) as compared with wild-type platelets (gray line). B, washed G␣ q -deficient platelets (3 ϫ 10 5 /l) were stimulated with adrenaline (10 M) alone or in combination with U46619 (1 M) under stirring conditions and aggregation was recorded on an APACT 4-channel aggregometer. Where indicated, the experiment was performed in the presence of irrelevant rat IgG (30 g/ml) or a blocking antibody against integrin ␣ IIb ␤ 3 (30 g/ml). The results shown are representative of 9 -12 individual experiments. C, washed wild-type platelets (2 ϫ 10 4 /l) were stimulated with vehicle, Adr (10 M), U46619 (1 M), or the combination of both agonists for 2 min at 37°C, stained with the indicated fluorophore-labeled antibodies for 10 min, and analyzed on a FACScalibur directly. Platelets were gated by FSC/SSC characteristics. The results shown are representative of 6 -9 individual experiments. gests that G 12 /G 13 -and G i -mediated signaling pathways act in concert to activate integrin ␣ IIb ␤ 3 . To study the role of G 12 /G 13 in platelet integrin activation, we used the TXA 2 mimetic U46619, which is known to specifically stimulate G q and G 12 / G 13 through the TXA 2 receptor. As shown previously, U46619 induced a rapid shape change but no aggregation in G␣ q -deficient platelets at concentrations up to 30 M, whereas robust aggregation was already seen at 0.3 M in wild-type platelets (Fig. 3A), confirming the crucial role of G q in this process. U46619-induced aggregation of wild-type platelets has been shown to largely depend on released ADP acting on the G icoupled P2Y 12 receptor (17). This finding suggests that the activation of G q and G 12 /G 13 is not sufficient to stimulate full integrin activation and, possibly, degranulation in platelets unless G i is concomitantly activated. This was confirmed when we examined U46619-stimulated wild-type platelets by flow cytometry for activated ␣ IIb ␤ 3 integrin and P-selectin expression. U46619 at concentrations of up to 30 M induced only very low levels of ␣ IIb ␤ 3 activation and P-selectin expression in wild-type platelets (Fig. 3B). Thus, the activation of G q and G 12 /G 13 is indeed not sufficient for efficient integrin activation and granule release in platelets. However, the addition of low concentrations of exogenous ADP (5 M) resulted in profound ␣ IIb ␤ 3 activation and P-selectin expression. At this concentration, ADP alone is known to induce only low levels of ␣ IIb ␤ 3 activation and no P-selectin expression in wild-type platelets, resulting in reversible platelet aggregation (Fig. 3B) (4). The potentiating effect of ADP was dependent on stimulation of the G i -coupled P2Y 12 receptor, because it was abrogated in the presence of the highly specific P2Y 12 antagonist AR-C69931MX (10 M).
This finding suggested that the failure of G␣ q -deficient platelets to aggregate in response to U46619 is not only because of the absence of the G q -mediated pathway leading to inside-out activation of integrin ␣ IIb ␤ 3 but is also based on defective release of stored ADP and subsequent G i activation. To test this hypothesis, we analyzed ␣ IIb ␤ 3 activation on U46619-stimulated G␣ q -deficient platelets in the absence or presence of exogenously added ADP. In these platelets, which do not degranulate in response to TXA 2 receptor activation, U46619 alone did not induce ␣ IIb ␤ 3 activation at concentrations up to 30 M (Fig. 3C). Strikingly, however, weak but consistent activation of ␣ IIb ␤ 3 was detectable upon the addition of 5 M ADP, which by itself had no effect because of the lack of P2Y 1mediated calcium mobilization (18). Similar to wild-type platelets, the potentiating effect of ADP was inhibited by the P2Y 12 antagonist AR-C69931MX (10 M) (Fig. 3C).
The relatively low levels of ␣ IIb ␤ 3 activation induced by U46619 and ADP in G␣ q -deficient platelets were sufficient to mediate robust and irreversible aggregation that was inhibited by a blocking antibody against ␣ IIb ␤ 3 integrin (Fig. 3D). Importantly, aggregation was also blocked in the presence of AR-C69931MX, confirming the critical role of the P2Y 12 receptor- mediated G i activation in this process (Fig. 3D). Similar results were obtained when G 12 /G 13 activation was induced by thrombin instead of U46619. Although thrombin (0.2 units/ml) alone induced shape change but no aggregation of G␣ q -deficient platelets, robust ␣ IIb ␤ 3 -dependent aggregation occurred in the presence of exogenously added ADP (5 M). This potentiating effect of ADP was again inhibited by AR-C69931MX (10 M) (data not shown).
It is known that G␣ q -deficient platelets do not aggregate in response to U46619 or ADP because of the lack of intracellular calcium mobilization (6). To test whether the combination of both agonists would be able to bypass this defect by undefined mechanisms, we measured intracellular calcium levels in G␣ qdeficient platelets in response to increasing concentrations of U46619 and ADP. However, although wild-type platelets responded with robust calcium mobilization to a combination of 1 M U46619 and 5 M ADP, no response was seen in G␣ qdeficient platelets, even at 10 M U46619 plus 50 M ADP (data not shown).
These results suggested that concomitant stimulation of G 12 / G 13 and G i results in integrin activation in platelets and that G q is not necessarily required to induce platelet activation through G-protein-coupled receptors. To test this hypothesis further, we stimulated wild-type and G␣ q -deficient platelets with U46619 in the absence or presence of the G i -selective agonist, adrenaline. Adrenaline alone is unable to promote platelet shape change or aggregation (14,19) but can potentiate various platelet responses through G i and the G i family member G z (20). Neither wild-type nor G␣ q -deficient platelets responded to adrenaline (10 M) with ␣ IIb ␤ 3 activation (Fig. 4, A and C) or P-selectin expression ( Fig. 4C and data not shown), and consequently, no aggregation occurred ( Fig. 4B and data not shown). Adrenaline (like ADP) potentiated the effect of U46619 in both mouse strains as shown by ␣ IIb ␤ 3 activation and P-selectin expression in wild-type (Fig. 4C) and ␣ IIb ␤ 3 activation in G␣ q -deficient platelets (Fig. 4A). The ␣ IIb ␤ 3 activation observed in G␣ q -deficient platelets in response to U46619 and adrenaline was sufficient to mediate irreversible platelet aggregation that was inhibited by a blocking antibody against the integrin (Fig. 4B). This finding demonstrates the critical involvement of both G 12 /G 13 and G i in full integrin activation and granule release in response to the TXA 2 mimetic U46619 and indicates that activation of G 12 /G 13 -and G i -mediated pathways is sufficient for platelet activation.
Recent evidence suggests that the G 12 /G 13 -mediated platelet shape change response is mediated by the Rho/Rho kinase pathway (3,21,22). In agreement with this finding, the TXA 2dependent shape change of G␣ q -deficient platelets in response to high collagen concentrations was abolished in the presence of the Rho kinase inhibitor Y-27632 (10 M) (23). Strikingly, however, the inhibition of Rho kinase had no significant influence on the aggregation response, indicating that G 12 /G 13 activation contributes to integrin activation by regulating a Rho kinase-independent mechanism (Fig. 5A). To test this hypothesis, we activated G␣ q -deficient platelets with U46619 alone or in combination with ADP or adrenaline in the absence or presence of Y-27632 (10 M). Indeed, whereas the inhibition of Rho kinase abrogated the shape change induced by U46619, it had no effect on ␣ IIb ␤ 3 activation (Fig. 5B) and platelet aggregation (Fig. 5C). Similarly, inhibitors of non-receptor tyrosine kinases like src, which have been shown to be activated through G 12 / G 13 in platelets (3), had no effect on the activation of G␣ qdeficient platelets in response to U46619 and ADP (data not shown). The mechanisms linking the G 12 /G 13 -mediated signaling pathway to integrin ␣ IIb ␤ 3 activation in platelets remains unclear.
To test the significance of the synergy between G 12 /G 13 -and G i -mediated signaling in normal platelets, we stimulated wildtype platelets with 0.03 M U46619 in the presence or absence of adrenaline. At this concentration, U46619 alone is known to induce a rapid shape change through G 12 /G 13 activation but no aggregation due to insufficient G q signaling (3,24,25) (Fig.  5D). However, the concomitant stimulation of G i by the addition of adrenaline (10 M) induced irreversible integrin ␣ IIb ␤ 3dependent aggregation (Fig. 5D). These results demonstrate that the contribution of G 12 /G 13 signaling to platelet activation also plays a significant role in wild-type platelets, at least at low agonist concentrations. This may reflect the in vivo situation where a variety of stimuli, which are present at submaximally active concentration, synergistically induce platelet activation during thrombus formation.
Taken together, our data show that the activation of G 12 /G 13and G i -mediated signaling pathways is sufficient to induce integrin ␣ IIb ␤ 3 activation. Although, G q -mediated signaling processes are essential for calcium mobilization and thus play an important role in platelet activation, they are not absolutely required for the activation of integrin ␣ IIb ␤ 3 . These data indicate that the activation of platelets through G-protein-coupled receptors is a highly integrated process, which in order to occur with high efficiency requires the activation of at least three G-protein-mediated pathways involving G q , G i , and G 12/13 .