Protein Kinase B Is Regulated in Platelets by the Collagen Receptor Glycoprotein VI*

Phosphoinositide 3-kinase (PI3K) is a critical component of the signaling pathways that control the activation of platelets. Here we have examined the regulation of protein kinase B (PKB), a downstream effector of PI3K, by the platelet collagen receptor glycoprotein (GP) VI and thrombin receptors. Stimulation of platelets with collagen or convulxin (a selective GPVI agonist) resulted in PI3K-dependent, and aggregation independent, Ser473 and Thr308 phosphorylation of PKBα, which results in PKB activation. This was accompanied by translocation of PKB to cell membranes. The phosphoinositide-dependent kinase PDK1 is known to phosphorylate PKBα on Thr308, although the identity of the kinase responsible for Ser473phosphorylation is less clear. One candidate that has been implicated as being responsible for Ser473 phosphorylation, either directly or indirectly, is the integrin-linked kinase (ILK). In this study we have examined the interactions of PKB, PDK1, and ILK in resting and stimulated platelets. We demonstrate that in platelets PKB is physically associated with PDK1 and ILK. Furthermore, the association of PDK1 and ILK increases upon platelet stimulation. It would therefore appear that formation of a tertiary complex between PDK1, ILK, and PKB may be necessary for phosphorylation of PKB. These observations indicate that PKB participates in cell signaling downstream of the platelet collagen receptor GPVI. The role of PKB in collagen- and thrombin-stimulated platelets remains to be determined.

Activation of platelets in vivo is tightly regulated, with circulating platelets remaining quiescent under normal biological conditions. However, upon vascular damage platelets are exposed to extracellular matrix proteins, such as sub-endothelial collagens, factors released or generated by activated platelets, and thrombin, which activate platelets. This leads to the formation of a platelet thrombus, preventing further blood loss. Although this mechanism is vital when tissue is damaged, inappropriate activation of platelets can lead to thrombosis. Alternatively, failure to become activated can lead to excessive bleeding. An understanding of the intracellular mechanisms that regulate the function of platelets is vital for the development of therapies to combat these disorders.
Platelets express a number of receptors on their cell surface that are important for activation and aggregation at sites of tissue injury. These include the collagen activatory receptor glycoprotein VI (GPVI) 1 (1,2), thrombin receptors (PAR1 and PAR4) (3,4), and the fibrinogen receptor integrin ␣ IIb ␤ 3 (5), among others. Damage to the blood vessel wall exposes platelets to sub-endothelial collagens. Following initial interactions of platelets with collagen via von Willebrand factor, the collagen receptors GPVI and integrin ␣ 2 ␤ 1 support firm adhesion. GPVI is non-covalently associated with the Fc-receptor ␥-chain, and activation of this complex by collagen leads to tyrosine phosphorylation of this protein (2,6). This leads to the recruitment and activation of various signaling molecules, which, in turn, and via a range of signaling pathways, result in platelet aggregation (7)(8)(9)(10). Thrombin receptors, on the other hand, are coupled to G proteins and elicit their effects through G proteinlinked signaling pathways. Although collagen and thrombin activate divergent signaling pathways, common to both is phosphoinositide 3-kinase (PI3K). PI3K has been shown to be central to the intracellular signaling pathways activated by agonist-induced stimulation of either of these receptors. Inhibition of PI3K with wortmannin blocks collagen-induced aggregation (11), whereas the thrombin receptor requires phosphoinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P 3 ), a product of PI3K activity, to elicit its effect (12). However, little has been shown in platelets with regard to the signaling that lies downstream of PI3K. In recent years attention has been focused on the enzyme protein kinase B (PKB/Akt) as one of the central downstream effectors of PI3K in many cell systems. Indeed, PKB has been shown to become activated in platelets following stimulation with thrombin and thrombopoietin (TPO), but it is not activated upon stimulation with other platelet agonists such as ADP and platelet-activating factor (13,14). PKB is a 57-kDa phospholipid-dependent serine/threonine kinase that has been implicated in regulating cell survival (15,16), cell cycle progression (17,18), as well as some aspects of cellular metabolism (19,20). It is a multidomain protein that contains a pleckstrin homology (PH) domain adjacent to a centrally located catalytic domain. The kinase domain is highly homologous to the catalytic domains of protein kinase A (PKA) and protein kinase C (PKC). The regulation of PKB activity has been shown to be critically dependent on phosphorylation of specific serine and threonine residues (21). Indeed the catalytic domain contains one such residue. The catalytic domain is connected to a short C-terminal tail, which displays consider-able similarity to the regulatory subunit of PKC and contains a second regulatory serine phosphorylation site (22).
Three isoforms of PKB have been identified in humans, PKB␣, -␤, and -␥. The isoforms are differentially distributed, but each tissue contains at least one isoform (22). The subtypes possess more than 80% sequence homology but differ with respect to their regulatory phosphorylation sites (PKB␣: Thr 308 /Ser 473 ; PKB␤: Thr 309 /Ser 474 ; PKB␥: Thr 305 ) (23,24). In resting cells PKB is predominantly cytoplasmic and exists as an inactive multimer (25). Activation requires both translocation to cell membranes and phosphorylation of the enzyme. Upon cellular stimulation, PKB is initially recruited to the plasma membrane via interaction of its PH domain with phosphoinositide products of PI3K, particularly PtdIns(3,4,5)P 3 (26). Once at the plasma membrane, PKB is believed to undergo conformational changes and is subsequently phosphorylated at its regulatory threonine and serine phosphorylation sites. Phosphorylation of the threonine site alone facilitates partial activation and occurs via another phospholipid-dependent kinase, namely PDK1 (27,28). However, with the exception of PKB␥ (which has only a threonine phosphorylation site due to C-terminal truncation), for maximal activation and protection against dephosphorylation by cytosolic phosphatases, PKB also requires phosphorylation at the regulatory serine residue (21). The mechanism by which this serine residue is phosphorylated remains contentious. Some researchers propose autophosphorylation (29) as the given mechanism, whereas others suggest that PDK1 is responsible for phosphorylating this residue as well as the threonine residue (30). Several reports indicate that Ser 473 of PBK␣ is phosphorylated by a second PDK, and recent studies have indicated that this may be integrin-linked kinase (ILK) (31,32).
In this study we have investigated the regulation of PKB in human platelets. PKB␣ and PKB␤ have been shown to be expressed in platelets (14). Of these two isoforms PKB␣ appears to be the most abundant and is activated in response to the platelet agonist thrombin (13,14). In this report we present data to show that PKB␣ is also phosphorylated in platelets following activation of the collagen receptor GPVI. This is demonstrated by stimulating platelets with collagen and convulxin (Cvx), a selective agonist for the platelet collagen receptor GPVI (33) purified from the venom of the rattlesnake Crotalis durrisus terrificus. GPVI-stimulated phosphorylation of PKB␣ occurs independently of aggregation but is dependent on the activation of PI3Ks. We have also examined the interaction of PKB␣ with ILK and PDK1 in resting and stimulated platelets, kinases that are implicated in the phosphorylation and regulation of PKB.
Preparation and Stimulation of Platelets-Human platelets were obtained on the day of experiments from drug-free volunteers and prepared by differential centrifugation as described previously (36). Platelets were then resuspended in Tyrode's-Hepes buffer (134 mM NaCl, 0.34 mM Na 2 HPO 4 , 2.9 mM KCl, 12 mM NaHCO 3 , 20 mM Hepes, 5 mM glucose, 1 mM MgCl 2 , pH 7.3) to a density of 8 ϫ 10 8 cells/ml. Platelets were stimulated for the appropriate duration with collagen, convulxin (Cvx), or thrombin at 37°C in an aggregometer with contin-uous stirring (1200 rpm). For inhibitor studies platelets were incubated with the appropriate inhibitor (dissolved in Me 2 SO, final concentration Ͻ0.2% (v/v)) for 15 min prior to stimulation. Control samples were incubated with the appropriate concentration of Me 2 SO alone. Where stimulations were performed under non-aggregating conditions, platelets were preincubated for 30 min with 1 mM EGTA and 10 M indomethacin.
Immunoblotting-Platelets were stimulated for the appropriate time, and the reaction was stopped by the addition of an equal volume of reducing Laemmli sample treatment buffer. Samples were separated by SDS-PAGE and transferred onto polyvinylidene difluoride membrane by semi-dry Western blotting. After blocking by incubation in Tris-buffered saline-Tween (TBS-T; 20 mM Tris, 137 mM NaCl, 0.1% (v/v) Tween 20, pH 7.6) containing 5% (w/v) milk protein, membranes were incubated overnight at 4°C with primary antibody, at a concentration of 1 g/ml in TBS-T containing 5% (w/v) bovine serum albumin or 5% (w/v) milk protein where appropriate. Subsequent to washing membranes in TBS-T, membranes were incubated with horseradish peroxidase-conjugated secondary antibody, at a concentration of 0.1 g/ml in 5% (w/v) bovine serum albumin dissolved in TBS-T, and then washed with TBS-T. Membranes were then incubated with an enhanced chemiluminescence substrate for horseradish peroxidase and exposed to x-ray film.
Subcellular Fractionation-Following stimulation, platelets were pulse-centrifuged (14,000 ϫ g), and the supernatant was removed. PBS containing 0.05% (w/v) digitonin was added to the platelet pellet and incubated for 10 min on ice with intermittent agitation. Samples were then centrifuged at 10,000 ϫ g for 10 min at 4°C, and the supernatant cytosolic fraction was removed. PBS containing 1% (v/v) Triton X-100 was added to the cell pellet and incubated on ice for a further 10 min with intermittent agitation. Samples were centrifuged as before, and the supernatant membrane fraction was harvested. The protein concentration for both membrane and cytosol fractions were determined using the Lowry protein assay and following the manufacturer's protocol.
Immunoprecipitation-After stimulation, cells were lysed using an equal volume of radioimmune precipitation buffer (4 M NaCl, 1.25 mM sodium deoxycholate, 1 M Tris, 10% (w/v) SDS, 10% (w/v) Nonidet P-40, pH 7.4 containing 1 g/ml pepstatin A, 10 g/ml leupeptin, 10 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 10 M okadaic acid, and 2 mM Na 3 VO 4 ) and incubated on ice for more than 15 min. Lysates were clarified by centrifugation at 14,000 ϫ g for 15 min at 4°C; the supernatants were then precleared by incubation with protein A-Sepharose (PAS; 20 l of a 50% (v/v) suspension in TBS-T) for 30 min. Following removal of the PAS by centrifugation, 1 g of the appropriate antibody was added to each sample and the samples were incubated for 1 h before addition of PAS (40 l of a 50% (v/v) suspension in TBS-T). After a further 1-h incubation, the samples were centrifuged and the PAS beads were washed three times in TBS-T. Each sample was resuspended in reducing Laemmli sample treatment buffer and boiled prior to loading on a polyacrylamide gel.

Collagen and Convulxin Induce PKB Phosphorylation in a
Time-dependent and Aggregation-independent Manner-It has been shown previously that PKB␣ is phosphorylated in platelets in response to the agonist thrombin (13,14). Because PI3K activity is required for platelet responses to both collagen and thrombin, and PtdIns(3,4,5)P 3 generation is required for activation of PKB, we investigated whether signaling generated by the platelet collagen receptor GPVI also leads to PKB␣ phosphorylation. Immunoblot analysis of platelet lysates using PKB␣ phosphospecific antibodies revealed that PKB␣ is phosphorylated on Ser 473 (Fig. 1) and Thr 308 (not shown) in response to stimulation with collagen, Cvx, and thrombin. This response was found to be both concentration-and time-dependent (not shown and Fig. 1, respectively). An increase in Ser 473 phosphorylation in response to collagen (100 g/ml) was detectable within 90 s stimulation, and the level of phosphorylation peaked at around 5 min. Phosphorylation was more rapid in response to Cvx and thrombin at the concentrations presented (125 ng/ml and 1 unit/ml, respectively) with phosphorylation levels reaching above basal levels within 20 s. Maximal phosphorylation in response to Cvx was achieved within 5 min, and for thrombin, 90 s. Similar profiles of Thr 308 phosphorylation were also detected (not shown). Differences in kinetics reflect the relative potencies of these agonists and the concentrations at which they were used. Each of these experiments was performed in the presence of 1 mM EGTA. Under these conditions aggregation is prevented (as confirmed using optical aggregometry (not shown)) due to inhibition of fibrinogen binding. The levels of PKB␣ phosphorylation observed were very similar in platelets that were allowed to aggregate through the omission of EGTA. This indicates that the phosphorylation of PKB␣ on Ser 473 and Thr 308 is not dependent on platelet aggregation.
It has been reported that thrombin stimulates dual signaling pathways leading to the phosphorylation and activation of PKB␣; one pathway being PI3K-dependent and the other independent of PI3K (14). The latter pathway was shown to involve calcium-dependent PKC isoforms (PKC␣/␤). In contrast, TPOstimulated PKB␣ phosphorylation is solely dependent on PI3K signaling (14). We therefore examined GPVI-stimulated PKB phosphorylation in the presence of PI3K inhibitors. Both LY294002 (Fig. 2) and wortmannin (not shown) completely inhibited collagen and Cvx-stimulated PKB␣ Ser 473 (Fig. 2) and Thr 308 (not shown) phosphorylation, indicating that GPVIstimulated PKB␣ phosphorylation is dependent on PI3K. In agreement with Kroner et al. (14) these inhibitors did not prevent PKB␣ phosphorylation at high concentrations of thrombin, although complete inhibition of PKB␣ phosphorylation was observed at all concentrations of collagen and Cvx tested (not shown).
Translocation of PKB to the Membrane Coincides with Its Phosphorylation-To examine the subcellular localization of PKB under resting and stimulated conditions, platelets were activated with collagen, Cvx, or thrombin, then separated into cytosolic and membrane fractions. Fractions were examined for the presence of PKB by immunoblot analysis using an anti-PKB antibody. Fig. 3 shows that under basal conditions PKB is present predominantly in the cytosolic fraction. However, when the platelets are stimulated with collagen (100 g/ml), Cvx (125 ng/ml), or thrombin (1 unit/ml), PKB translocates to the membrane fraction. Translocation is rapid, occurring within 20 s of stimulation and coincides (or precedes, in the case of collagen) with the onset of phosphorylation of PKB, as seen in Fig. 1. This indicates that the phosphorylation and, therefore, the activation, of PKB in platelets coincide with its association with cell membranes. This is consistent with observations in other cell systems and the dependence of PKB on PI3K activity and the generation of PtdIns(3,4,5)P 3 (25,37).
PKB Co-associates with PDK1 and ILK-We next looked at the interaction between PKB and reported upstream regulators PDK1 and ILK, which we have previously demonstrated to be expressed in platelets (38) 2 and are implicated in PKB phosphorylation. PDK1 and ILK were immunoprecipitated from resting platelets and platelets stimulated with either Cvx or thrombin. Isolated proteins were separated by SDS-PAGE and immunoblotted using anti-phospho-Ser 473 PKB␣ antibodies. Fig. 4a shows that phosphorylated PKB␣ is present in PDK1 immunoprecipitates, and that the level of phosphorylation of associated PKB␣ is increased on stimulation with Cvx (125 ng/ml) and thrombin (1 unit/ml). It is well established that PDK1 is responsible for Thr 308 phosphorylation (28), although controversy surrounds the identity of the kinase that phosphorylates Ser 473 . ILK is implicated in this role (31,32). Of particular interest, therefore, is the observation that phosphorylated PKB␣ is also present in ILK immunoprecipitates, and similarly the level of phosphorylation of associated PKB␣ is increased on stimulation with Cvx and thrombin (Fig. 4b). In addition we observed interaction between PDK1 and ILK in platelets (Fig. 4c). Furthermore, the level of this interaction was increased following stimulation of platelets with Cvx (125 ng/ml) or thrombin (1 unit/ml). Because all three enzymes contain PH domains, it was of concern that these may not be true associations but result from the incorporation into lipid microvesicles formed during immunoprecipitation. To verify this was not the case, all blots (Fig. 4, a-c) were reprobed with an antibody to an abundant PH-domain containing protein pleckstrin. No pleckstrin was detected in ILK and PDK1 immunoprecipitates. These results illustrate the potential impor-

FIG. 3. Agonist-induced platelet activation leads to PKB translocation.
Platelets were stimulated with collagen (100 g/ml) (a), convulxin (125 ng/ml) (b), or thrombin (1 unit/ml) (c) before being fractionated into cytosol and membrane (see "Experimental Procedures"). The protein concentration of each fraction was determined, and 10 g of each was separated by SDS-PAGE. Subsequent to electrophoresis, gels were Western-blotted, and membranes were probed with an anti-PKB antibody.
tance of both ILK and PDK1 in the regulation of PKB activity in collagen-and thrombin-stimulated platelets. DISCUSSION Human platelets express on their surface a plethora of receptors that are activated by a number of physiological factors leading, via multiple signaling pathways, to platelet activation. Central to the function of several of these receptors is the activation of PI3K. PI3K has previously been shown to be necessary for collagen-and thrombin receptor-mediated platelet aggregation, as well as Cvx-induced granule secretion, inositol phosphate production, and increase in [Ca 2ϩ ] i (11,12). However, little is known of the downstream effectors of PI3K in platelets. Because PKB has been shown to be a key effector of PI3K in other cell systems, we investigated this enzyme in collagen receptor (GPVI)-mediated platelet signaling.
We have observed that stimulation of platelets with a variety of agonists, including collagen, Cvx, and thrombin results in the phosphorylation of PKB␣ on Ser 473 and Thr 308 , and this phosphorylation is not dependent on aggregation. Kinetic analysis reveals that PKB␣ is rapidly phosphorylated in platelets stimulated with any of these agonists. Phosphorylation continues to increase for up to 5 min in response to collagen and Cvx, and is sustained above basal levels for up to 1 h in response to all of the agonists used.
PKB␣ has been shown previously to become phosphorylated and activated in platelets following stimulation with thrombin and TPO. Other platelet agonists such as ADP and platelet-activating factor have been shown not to stimulate PKB phosphorylation (34). This study extends the involvement of PKB in platelets to the signaling downstream of the GPVI-Fc receptor ␥-chain collagen receptor complex. PKB activity in platelets stimulated with thrombin has been shown to be biphasic. This mirrors the biphasic activation of PI3K and generation of 3Јphosphorylated inositol phospholipids, where PtdIns(3,4,5)P 3 is generated upon initial stimulation and PtdIns(3,4)P 2 is generated later on platelet aggregation (13). Biphasic activation of PKB was not observed in response to collagen, Cvx, or thrombin in these studies, because platelets were prevented from aggregating during kinetic studies of PKB phosphorylation. PKB activity is regulated by the 3Ј-phosphorylated inositol phospholipids PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 (39). These lipids show an affinity for the PH domain of signaling proteins; therefore, as the concentration of these phospholipids rise, PH domain-containing proteins are recruited to membranes. We have observed that stimulation of platelets with collagen, Cvx, or thrombin leads to the translocation of PKB to membranes, which is consistent with the rise in intracellular PtdIns (3,4,5)P 3 stimulated in platelets by these agonists (11,12). Fig.  3 shows that by 20 s PKB translocates from the cytosolic fraction of stimulated platelets to the membrane fraction, a time point that correlates with or precedes its phosphorylation. Indeed, phosphorylated PKB␣ can be visualized by indirect immunofluorescence at the plasma membrane of platelets adhered to collagen (not shown).
It is widely accepted that the regulation of PKB is dependent on the phospholipid products of PI3K activity. Recently, however, Kroner and colleagues (14) reported that activation of PKB in thrombin-stimulated platelets is a consequence of PI3K signaling and a second partially PI3K independent mechanism. This second mechanism involves the activation of phospholipase C␤ (PLC␤1), leading to the activation of calcium-dependent isoforms of PKC (PKC␣/␤) and ultimately the phosphorylation of Ser 473 . Indeed, the activation of PKC directly using phorbol 12-myristate causes the selective phosphorylation of Ser 473 . It has been suggested that in this case PI3K␥ activity may be involved in the phosphorylation of Thr 308 via activation of PDK1. This dual regulation of PKB in platelets is possible, because the thrombin receptors PAR1 and PAR4 are able to activate PLC␤1/2 (through the action of G ␣q and G ␤␥ protein subunits), p85/110 PI3K, and PI3K␥. Activation of the TPO receptor c-Mpl, on the other hand, leads to the activation of PKB by p85/110 PI3K-dependent signaling alone. This is responsible for regulating the phosphorylation of Thr 308 and Ser 473 , because phosphorylation is abolished completely in the presence of PI3K inhibitors. It was therefore important to establish whether a single or dual regulation system is in place in platelets to regulate PKB␣ phosphorylation and activation downstream of the collagen receptor GPVI. We have shown in this study that PKB phosphorylation in response to relatively high concentrations of collagen (100 g/ml) and Cvx (125 ng/ ml) is completely abrogated by the PI3K inhibitors LY294002 and wortmannin. This indicates that GPVI-stimulated PKB phosphorylation is solely dependent on PI3K activity and suggests that a single regulation system is operational downstream of this receptor. Consistent with the work of Kroner et al., at high concentrations of thrombin, Ser 473 phosphorylation was partially resistant to PI3K inhibitors.
The molecular links between PI3K activation and PKB phosphorylation are phosphoinositide-dependent kinases such as PDK1. PDK1 is a 63-kDa monomeric protein that is ubiquitously expressed in human tissues (28). In the presence of 3Ј phospholipids PDK1 phosphorylates PKB at its Thr 308 residue, which results in partial activation of the enzyme. The mecha- nism of phosphorylation of the Ser 473 site, however, is still hotly debated. The question remains whether this site is phosphorylated by an independent protein or by PKB itself (that is, PKB autophosphorylation of Ser 473 ). Evidence is mounting for the involvement of an independent protein in the phosphorylation of Ser 473 , and several candidates have been proposed for the role. Integrin-linked kinase is one such candidate, but the role of this 59-kDa protein in phosphorylating PKB, however, is contentious. Several studies report that ILK phosphorylates Ser 473 directly (31,32), whereas others propose that ILK acts as an adapter (40) thereby facilitating phosphorylation by an alternative kinase. We show here that PDK1 and ILK both co-associate with PKB in platelets (Fig. 4, a and b). The association of PKB with ILK is consistent with recent observations in DU145 cells (32) and implies that ILK may play a role in regulating PKB activity. Furthermore, Fig. 4c shows that PDK1 and ILK also co-associate with each other. We postulate that the formation of this tertiary complex (PKB⅐PDK1⅐ILK) may facilitate the phosphorylation of PKB on its Ser 473 site by changing the conformation of the enzyme such that its serine site becomes available for phosphorylation. Whether the Ser 473 phosphorylation occurs in platelets via PDK1, ILK, or an alternative mechanism, however, remains to be determined.
An additional layer of complexity to the regulation of PKB has been recently introduced with the observation that PKB can become tyrosine-phosphorylated. Two tyrosine residues have been identified near the activation loop of PKB␣, Tyr 315 and Tyr 326 , and phosphorylation of these residues in vitro effects PKB activity (41). The significance of PKB tyrosine phosphorylation has yet to be determined, but it is interesting to note that PKB becomes tyrosine-phosphorylated following stimulation of platelets with Cvx. 2 The importance of PI3K activity and the stimulation of platelet activation is well established. However the functional significance of PKB in platelet function is presently unknown. With the present lack of a specific PKB inhibitor, this is a difficult question to address. We have established that a kinase inhibitor, ML-9, which has been shown to inhibit PKB activity (42), is able to partially inhibit Cvx-stimulated platelets 2 ; however, the promiscuous nature of this inhibitor prevents solid conclusions to be drawn from such experiments. An alternative approach is to examine the role of effectors of PKB in platelets.
PKB has numerous substrates that have been identified, including forkhead transcription factors (43), CREB (44), phosphodiesterase 3b (45), 6-phosphofructose 2-kinase (46), and glycogen synthase kinase-3 (19). Many of these proteins are not present in platelets, although, we have found that glycogen synthase kinase-3 is present (47). Therefore, investigations are currently underway in our laboratory to determine if this is a functionally relevant substrate of PKB in these cells.