Glucose Uptake via Glucose Transporter 3 by Human Platelets Is Regulated by Protein Kinase B*

In insulin-responsive tissues, insulin is a potent activator of protein kinase B (PKB)-mediated glucose uptake through the facilitative glucose transporter GLUT4. In platelets, glucose uptake is mediated through GLUT3, which is present in plasma (15%) and intracellular α-granule (85%) membranes. Here we report the PKB-mediated glucose uptake by platelets by agents that do (thrombin) or do not (insulin) induce α-granule translocation to the plasma membrane. Both thrombin and insulin activate PKB and induce glucose uptake albeit with different kinetics. Inhibition of PKB by the pharmacological inhibitor ML-9 decreases thrombin-induced α-granule release and thrombin- and insulin-induced glucose uptake. At low glucose (0.1 mm), both agents stimulate glucose uptake by lowering the Km for glucose (thrombin and insulin) and increasing Vmax (thrombin). At high glucose (5 mm), stimulation of glucose uptake by insulin disappears, and insulin becomes an inhibitor of thrombin-induced glucose uptake via mechanisms independent of PKB. We conclude that in platelets glucose transport through GLUT3 is regulated by changes in surface expression and affinity modulation, which are both under control of PKB.

In insulin-responsive tissues, insulin is a potent activator of protein kinase B (PKB)-mediated glucose uptake through the facilitative glucose transporter GLUT4. In platelets, glucose uptake is mediated through GLUT3, which is present in plasma (15%) and intracellular ␣-granule (85%) membranes. Here we report the PKBmediated glucose uptake by platelets by agents that do (thrombin) or do not (insulin) induce ␣-granule translocation to the plasma membrane. Both thrombin and insulin activate PKB and induce glucose uptake albeit with different kinetics. Inhibition of PKB by the pharmacological inhibitor ML-9 decreases thrombin-induced ␣-granule release and thrombin-and insulin-induced glucose uptake. At low glucose (0.1 mM), both agents stimulate glucose uptake by lowering the K m for glucose (thrombin and insulin) and increasing V max (thrombin). At high glucose (5 mM), stimulation of glucose uptake by insulin disappears, and insulin becomes an inhibitor of thrombin-induced glucose uptake via mechanisms independent of PKB. We conclude that in platelets glucose transport through GLUT3 is regulated by changes in surface expression and affinity modulation, which are both under control of PKB.
Platelet aggregation and secretion depend on plasma glucose as a source for anaerobic resynthesis of metabolic ATP (1). The facilitative glucose transporter 3 (GLUT3) 2 is the major GLUT subtype in human platelets. At resting conditions, ϳ85% of GLUT3 is located intracellularly in ␣-granule membranes, whereas the remainder is present in the plasma membrane (PM) (2). Upon platelet activation, GLUT3 containing ␣-granules fuse with the PM, and glucose uptake increases 2-fold (2,3). GLUT3 is predominantly expressed in neuronal cells and platelets, but has also been found in granulocytes, monocytes, fibroblasts, pancreatic, endothelial, myocardial, and skeletal muscle cells (4 -8). GLUT3 has a K m for glucose of ϳ1.5 mM, and this parameter appears constant in different cell types (4 -9). This implicates that under normal (ϳ5 mM glucose) and pathological conditions (up to 20 mM glucose) this transporter functions near saturation. GLUT1 is a ubiquitously expressed PM transporter with a K m for glucose of 29 mM, which is well above (patho-) physiological levels (10,11). Platelets contain a minimal amount of GLUT1 (2).
GLUT3 belongs to the same subclass of facilitative glucose transporters as the insulin-responsive GLUT4 sharing a dual cellular location and a low K m for glucose (12). In L6 muscle cells, glucose uptake via GLUT3 and GLUT4 is stimulated by the serine-threonine (Ser/Thr) protein kinase B (PKB, also known as Akt kinase) (13). Glucose transport by GLUT4 is determined by copy number at the PM and phosphorylation state, which affects its affinity for glucose (14). Little is known about the regulation of GLUT3.
A key element in insulin-responsive tissues is insulin receptor substrate-1 (IRS-1), which controls the signal transduction pathways from the insulin receptor to phosphatidylinositol 3-kinase (PI3K) (15). In platelets, activation of IRS-1 induces binding of IRS-1 to the ␣ 2 subunit of the heterotrimeric GTP-binding protein (G-protein) G i (16). This G-protein is under control of the purinergic P2Y 12 receptor of ADP that supports platelet activation by thrombin by inhibiting adenylyl cyclase and accumulation of cAMP, which is a potent platelet inhibitor (17,18). The result of Tyr-phosphorylation of G i ␣ 2 is inactivation of the G-protein, preventing the decrease in cAMP and attenuating the mobilization of Ca 2ϩ ions and platelet aggregation (16).
An important signaling element in insulin-responsive tissues is PKB, which participates in insulin-induced glucose-uptake (15). Platelets contain the PKB ␣ and PKB ␤ isoforms that contribute to platelet aggregation (19). Animal models show that both PKB-subtypes are crucial for the formation and stabilization of a thrombus in an arterial injury model (20,21), suggesting that PKB is a critical component of PI3K-supported platelet activation. The activity of PKB depends on its phosphorylation at Ser 473/4 and Thr 308/9 for PKB ␣ and PKB ␤ , respectively (22), and is finely regulated by PI3K-independent mechanisms that include Ca 2ϩsensitive protein kinase C (19), integrin-linked kinase (23), phosphatidylserine translocation (24), and the Src-family kinase Lyn (25).
By analogy with insulin-responsive tissues one would expect that platelets respond to insulin with an increase in glucose uptake via a mechanism controlled by PKB. Up-regulation of glucose uptake requires an increase in surface expression of GLUT3 (2,3). Because this process depends on the fusion of ␣-granule membrane with the PM, and because secretion depends on an increase in cytosolic Ca 2ϩ , one would expect that insulin supports secretion by enhancing Ca 2ϩ mobilization. This appears not to be the case (16).
The present study was undertaken to understand the apparent contradiction in insulin responses inflicted in platelets. The results show that insulin indeed triggers PKB activation as well as glucose uptake despite its capacity to inhibit thrombin-induced Ca 2ϩ mobilization. Whether insulin acts as an activator or inhibitor of PKB-mediated glucose uptake depends on the extracellular glucose concentration, which shifts the balance between activation and inhibition of platelets by insulin.
Immunoblotting-Samples of washed platelets were collected in 3ϫ sample buffer and separated by SDS-PAGE on 10% gels. Proteins were transferred from the gel to nitrocellulose sheets and blocked in 4% bovine serum albumin/TBST 0.1%. Blots were probed separately with the primary antibody p-PKB a/b/g Ser 473/4 for PKB detection at 57 kDa. The proteins were detected by enhanced chemiluminescence with horseradish peroxidase-labeled secondary antibodies. For analysis of lane loading, the blots were reprobed with anti-PKB ␣/␤ . Quantification of the bands was performed with ImageQuant-TL software (Amersham Biosciences).
Flow Cytometry-Incubations of platelets were stopped by placing aliquots of 100 l in 700 l of Hepes/Tyrode buffer, pH 7.25, at different time intervals. Aliquots were preincubated with P-selectin/RPE (CD-62P) for 30 min, and samples were fixed with 1% formaldehyde. Fixed RPE-labeled platelets were analyzed in a FACScan (BD Biosciences, Immunocytometry Systems, San Jose, CA). The samples were gated for single platelets based on forward and side scatter profiles, and 10,000 events were recorded. The results were analyzed using CellQuest software (BD Biosciences) and expressed as percentage of thrombin-induced P-selectin expression (set at 100%).
Measurement of Platelet Aggregation-Aliquots of 0.5 ml of washed platelets were warmed to 37°C for 5 min, followed by stimulation with 0.25 unit/ml thrombin. Platelet aggregation was monitored continuously for 7 min at 900 rpm in an optical aggregometer (model 570 VS, Chrono-Log Corp., Havertown, PA).
RT-PCR-The megakaryocytic cell lines MEG-01, DAMI, and CHRF-288-11 resembling immature (26), intermediate (27), and mature megakaryocytes (28), respectively, were cultured as described previously (29). mRNA was isolated with the RNeasy mini kit from Qiagen and converted to cDNA using the SuperScript II Reverse Transcriptase kit from Invitrogen, according to the manufacturer's instructions. cDNA from human umbilical vein endothelial cells, Wistar rat brain, and 3T3-L1 adipocytes was kindly provided by R. T. Urbanus The reaction was performed in 20 l containing 2 l of 10ϫ PCR buffer II, 4 l of MgCl 2 (25 mM), and 0.2 l of AmpliTaq DNA polymerase (5 units/l) from the GeneAMP kit obtained from Applied Biosystems (Foster City, CA), 1 l of dNTP (dATP, dCTP, dGTP, and dTTP of each 25 M) obtained from Amersham Biosciences, 6.8 l of distilled water (9.8 l for GLUT1), 2 l of both sense and antisense primers (10 M; 1 l of each primer for GLUT1), and 2 l cDNA for GLUT3 and GLUT4 (1 l for GLUT1).
After heating at 94°C for 2 min, amplification proceeded for 43 (GLUT1) or 40 (GLUT3/4) cycles, with denaturation for 30 s at 94°C, annealing of primers for 1 min at 62°C (GLUT1) or 56°C (GLUT3/4), and extension for 1 min at 72°C. This was followed by a final extension step at 72°C for 5 min. After the RT-PCR, 10 l of the products was separated on a 2% agarose gel for GLUT1 and a 1% agarose gel for GLUT3 and GLUT4.
TaqMan-PCR-Real-time quantitative RT-PCR was performed using the ABI Prism 7700 sequence detection system (Applied Biosystems, Foster City, CA). Primers and TaqMan probes were designed with Primer Express software (Applied Biosystems, Foster City, CA). Primers designed for GLUT1 and GLUT3 amplified a 62-bp amplicon and a 63-bp amplicon, respectively, spanning the exon 4-exon 5 junction of each sequence. Both TaqMan probes were labeled with 6-carboxifluorecein (FAM) as a reporter dye and tetramethylrhodamine (TAMRA) as a quenching dye. The following primers were used: GLUT1 forward: 5Ј-CAT CGT CGT CGG CAT CCT-3Ј; GLUT1 reverse: 5Ј-CCT TGT TGC CCA TGA TGG A-3Ј; GLUT1 TaqMan probe: 5Ј-(FAM)-ATC GCC CAG GTG TTC-(TAMRA)-3Ј; GLUT3 forward: 5Ј-GGT GGC CCA GAT CTT TGG T-3Ј; GLUT3 reverse: 5Ј-CAG TAG CAG CGG CCA TAG C-3Ј; GLUT3 TaqMan probe: 5Ј-(FAM)-CTT CAG ACC CAA GGA TGA-(TAMRA)-3Ј. The relative expression of GLUT1 and GLUT3 in cDNA of the aforementioned megakaryocytic cell lines was determined by comparing the cycles at which the reporter dye fluorescence surpassed an arbitrary threshold. The same amount of cDNA was used in each PCR reaction.

Uptake of [ 3 H]DOG by Platelets-Washed
platelets were resuspended in HEPES/Tyrode buffer, pH 7.25, with or without 5 mM glucose, which is referred to as low and high glucose, respectively. Glucose uptake by platelets was initiated by addition of [ 3 H]DOG in the indicated concentrations prior to or after treatment with insulin or thrombin as outlined in the "Results." Glucose transport was stopped by adding 5 ml of stop buffer (phosphate-buffered saline, with 10 M cytochalasin B) as described previously (2,3). Samples were centrifuged (3000 ϫ g, 10 min, 20°C), and the supernatant was replaced by 5 ml of stop buffer. After three washing steps, the supernatant was replaced with 500 l of 0.1 M NaOH. The platelet pellet was dissolved overnight, and total protein content was determined to correct for variations in cell number. [ 3 H]DOG uptake was determined in a liquid scintillation counter and expressed as arbitrary units (AU) defined as disintegrations per minute corrected for protein content (disintegrations per minute/g of protein). Kinetic parameters for glucose transport were determined by a Hofstee plot with weighted linear regression analysis (30).
Measurement of Ca 2ϩ Mobilization-Platelet-rich plasma was preincubated with 3 M Fura-2-AM (45 min, 37°C, light-protected). After incubation, platelet-rich plasma was supplemented with acid-citratedextrose and prostaglandin I 2 , centrifuged again (330 ϫ g, 15 min, 20°C), and resuspended in HEPES/Tyrode buffer (pH 7.25) containing 5 mM glucose. The final platelet concentration was adjusted to 2.0 ϫ 10 11 cells/liter. Fura-2 fluorescence was recorded in 1.0-ml aliquots of platelets without additional Ca 2ϩ at 20°C in a F-4500 fluorescence spectrophotometer (Hitachi Ltd., Tokyo, Japan) with excitation wavelengths of 340 and 380 nm and emission at 510 nm. Changes in [Ca 2ϩ ] i were monitored using the Fura-2 fluorescence ratio and calibrated according to the method of Grynkiewicz et al. (31).
Statistical Analysis-Statistical analysis was performed using oneway analysis of variance with Tukey's multiple comparisons test as posttest for repeated measurements unless stated otherwise. Results are expressed as means Ϯ S.D. of five observations unless stated otherwise; ns indicates not significant. Differences were considered significant at p Ͻ 0.05. An asterisk indicates a significant difference with controls. Fig. 1 (A and B), thrombin induced a rapid phosphorylation of Ser 473/4 confirming earlier observations (19). Interestingly, stimulation with insulin also initiated the phosphorylation of Ser 473/4 illustrating that, in addition to inhibition of Ca 2ϩ mobilization by interfering with P2Y 12 signaling (16), insulin has potential activating properties by triggering the activation of PKB. ML-9 is a diazepine derivative that has been shown to inhibit PKB phosphorylation in adipocytes and myoblasts without interfering with the upstream PI3K (32,33). Fig. 1C shows that ML-9 also interfered with the phosphorylation of PKB Ser 473/4 induced by thrombin and insulin and thereby with the activity of PKB (19). Hence, ML-9 is an effective inhibitor of PKB in platelets.

ML-9 Inhibits Thrombin-and Insulin-induced PKB Phosphorylation-As illustrated in
PKB Regulates Surface Expression of P-selectin and Platelet Aggregation-Earlier studies showed that the uptake of glucose by platelets depends on an increase in GLUT3 surface expression that accompanies the secretion of ␣-granules (2,3). To address the question whether PKB participates in the translocation of ␣-granules and thereby in the surface expression of GLUT3, the expression of P-selectin was measured, which is a second component of the ␣-granule membrane. There was a slight P-selectin expression on non-stimulated platelets, which was unaf-fected by ML-9. In thrombin-stimulated platelets, ML-9 (100 M) reduced P-selectin expression by 37.3 Ϯ 3.1% ( Fig. 2A). These results suggest that PKB contributes to the secretion of ␣-granules thereby increasing surface expression of GLUT3.
The secretion of ␣-granules is part of the release reaction that accompanies platelet activation by thrombin. A second secretion response is the extrusion of dense granule contents, leading to liberation of ADP that facilitates aggregation by signaling through the P2Y 12 receptor (17). ML-9 (100 M) reduced thrombin-induced aggregation by 78 Ϯ 13% (Fig. 2, B and C). Thus ML-9 was a better inhibitor of aggregation than of secretion suggesting that apart from indirectly modulating aggregation via ADP release, PKB directly participates in the regulation of the aggregation response.
Glucose Transport Supports Platelet Aggregation-Apart from anaerobic and aerobic degradation of glucose as sources for resynthesis of metabolic ATP, platelets can maintain their energy content through oxidative phosphorylation of lipids (1). To assess the importance of glucose uptake in aggregation, platelets were preincubated with cytochalasin B and phloretin, which inhibit glucose transport by interfering with the intra-and extracellular glucose binding sites, respectively. These treatments induced a 30% fall in aggregation responses, illustrating that glucose transport is necessary for optimal platelet aggregation (Fig. 3, A and B).
GLUT Subtypes in Megakaryocytes-Platelets contain virtually all important elements of the insulin signaling pathway such as the insulin receptor, IRS-1, PI3K, and PKB. It is unclear if platelets contain GLUT4, which is the major glucose transporter under control of insulin. To address this issue, GLUT subtype distribution was determined by RT-PCR in the megakaryocytic cell lines MEG-01, DAMI, and CHRF, which represent different stages in the maturation of normal megakaryocytes (26 -28). Degenerated oligonucleotide primers against conserved sequence regions in GLUTs were used together with cDNA from the megakaryocytic cell lines, human umbilical vein endothelial cells, Wistar rat brain, and 3T3-L1 adipocytes. The primer combinations resulted in amplification of products with the expected sizes for GLUT1, GLUT3, and GLUT4. All megakaryocytic cell lines expressed GLUT1 and GLUT3 mRNA. No message was found for GLUT4 in the megakaryocytic cell lines in contrast to the positive controls (Fig. 4). The ratio of GLUT1 and GLUT3 mRNA was measured in the megakaryocytic cell lines MEG-01, DAMI, and CHRF by real-time quantitative RT-PCR using the TaqMan fluorogenic probe system. We observed 4.3 Ϯ 0.2-, 2.6 Ϯ 0.2-, and 13.2 Ϯ 2.0-fold increases in GLUT3 mRNA in the MEG-01, DAMI, and CHRF cell lines, respectively, compared with GLUT1 mRNA (n ϭ 3). These results suggest that platelets are deficient of GLUT4 and contain predominantly GLUT3.
Glucose Uptake by Platelets in Medium with Low Glucose-To investigate the involvement of PKB in glucose transport, platelets were preincubated in medium containing either 0 mM glucose or 5 mM glucose (defined as low and high glucose, respectively). Subsequent addition of 0.1 mM [ 3 H]DOG would reveal the regulation of both GLUT1 (K m for glucose 29.1 mM) and GLUT3 (K m for glucose 1.5 mM) under conditions of low glucose and predominantly of GLUT1 alone under conditions of high glucose concentration. Although the kinetic parameters of GLUTs in platelets have not been established, current literature data show these characteristics to be constant among the different cell types (4 -11). To study glucose transport at low glucose, platelets were isolated in a glucose-free medium and preincubated with [ 3 H]DOG for 3 min. Subsequently, thrombin, insulin, or vehicle was added, and glucose uptake was measured during a subsequent 30-min incubation. Basal glucose transport increased during the first 10 min following addition of vehicle and then reached a plateau (Fig. 5A). Thrombin increased the glucose uptake almost 2-fold and reached a plateau after 30 min. A similar increase was found following addition of insulin, but there was a clear delay of about 10 min before glucose uptake increased. A plot emphasizing the increase induced by thrombin and insulin revealed the rapid stimulation by thrombin in contrast to the slow enhancement by insulin (Fig. 5A, inset). These data show that both thrombin and insulin stimulate glucose transport in platelets albeit with different kinetics.
Because we reported earlier that insulin inhibits thrombin-induced increases in Ca 2ϩ , which would slow down secretion and thereby the surface expression of GLUT3 (16), a possible inhibition of thrombininduced glucose uptake by insulin was investigated. Platelets were preincubated with insulin for 5 min, stimulated with thrombin, and glucose uptake was assessed 30 min later. Separate addition of thrombin and insulin increased basal glucose uptake to a similar degree. Insulin did   not inhibit the thrombin-induced glucose uptake (Fig. 5B). These data suggest that the decrease in GLUT3 expression caused by the expected inhibition of thrombin-induced Ca 2ϩ signaling by insulin was compensated by a second mechanism that enhanced the uptake of glucose.
To understand the role of PKB in thrombin-and insulin-induced glucose uptake, platelets in the presence of 100 M ML-9 were preincubated with [ 3 H]DOG for 3 min and thereafter stimulated with thrombin or insulin for 30 min. Inhibition of PKB reduced basal glucose uptake by ϳ30%. In thrombin-stimulated platelets ML-9 induced a 30% decrease that is 60% of the extra glucose uptake induced by thrombin. In insulintreated platelets the inhibition was stronger and fully abolished the extra glucose uptake induced by insulin (Fig. 5C). Together, these results indicate that PKB contributes to glucose uptake both in resting platelets and in platelets stimulated with thrombin or insulin.
We reported earlier that in thrombin-stimulated platelets activation of PKB is mediated via Ca 2ϩ -dependent protein kinase C subtypes (19). This property together with the Ca 2ϩ -dependent secretion of ␣-granules would make the surface expression of GLUT3 strongly dependent on an increase in Ca 2ϩ . Fig. 6A shows that the Ca 2ϩ chelator BAPTA-AM reduced the basal uptake by ϳ70%, illustrating that in resting platelets glucose uptake is Ca 2ϩ -dependent. The additional uptake induced by thrombin was completely blocked by Ca 2ϩ chelation. Concurrent analysis of P-selectin expression showed a 50% fall induced by BAPTA-AM, which is in agreement with a previous publication (34) (Fig. 6B). Thus, in addition to supporting secretion, Ca 2ϩ contributes to glucose uptake via a second mechanism possibly reflecting the regulation of PKB.
Having established that Ca 2ϩ was important for PKB signaling to secretion and glucose uptake, the possibility that PKB had an effect on the regulation of Ca 2ϩ was investigated. Treatment with ML-9 did not change the Ca 2ϩ level in resting platelets, and the thrombin-induced increase in Ca 2ϩ was left undisturbed (Fig. 6C). These data suggest that Ca 2ϩ homeostasis is insensitive to changes in PKB activity.
The observation that both thrombin, an inducer of GLUT3 surface expression through ␣-granule release, and insulin, known not to induce ␣-granule release, stimulate glucose uptake indicated that changes in surface expression and in affinity of the transporter for glucose contribute to uptake regulation. Kinetic analysis of the uptake of [ 3 H]DOG in concentrations ranging between 0.1 and 2 mM revealed a K m for glucose of ϳ1.2 mM under conditions that both GLUT1 and GLUT3 were functional ( Fig. 7 and TABLE ONE). Initial velocities were determined and plotted against substrate concentrations according to Wools as advocated by Hofstee (30). Treatment with thrombin and insulin decreased the K m for glucose transport to ϳ0.5 mM, indicating that both agonists induced a strong increase in the affinity of the transporter for glucose. In addition, thrombin induced a 1.5-fold increase in V max , probably reflecting the increase in surface expression of GLUT3. In contrast, insulin did not change the V max in agreement with its inability to trigger secretion. Because GLUT1 is known to have a constant surface expression and is insensitive to regulation by intracellular signaling (35), the changes in kinetic parameters are likely to reflect the properties of GLUT3 (4 -11).
Glucose Uptake by Platelets in Medium with High Glucose-To further separate the regulation of GLUT1 and GLUT3, platelets were preincubated in a medium with 5 mM glucose. This is ϳ4 times the K m for glucose of GLUT3 making this transporter less susceptible to affinity regulation without disturbing the control of surface expression.
Platelets were preincubated with [ 3 H]DOG for 3 min, stimulated with thrombin and insulin, and glucose transport was monitored during a subsequent 30-min incubation (Fig. 8A). Basal glucose transport was constant throughout this period and remained the same upon addition of insulin. In contrast, thrombin induced about 3-fold increase in [ 3 H]DOG uptake. These results support the concept that an increase in glucose uptake at physiological glucose concentration is mainly mediated through an increase in surface expression of GLUT3. ML-9 and BAPTA-AM reduced glucose uptake by resting platelets as observed in a low glucose medium. In thrombin-stimulated cells both inhibitors completely blocked the extra glucose uptake induced by thrombin (Fig.  8B). These date illustrate the major role of PKB and Ca 2ϩ in thrombininduced surface expression of GLUT3.
To evaluate a possible inhibition by insulin under conditions that kept affinity changes in GLUT3 to a minimum, platelets were preincubated with insulin for 10 min to induce maximal inhibition of thrombininduced Ca 2ϩ increases (16). Again this treatment did not change the basal glucose uptake in a high glucose medium. Subsequent stimulation with thrombin increased [ 3 H]DOG uptake to 190.1 Ϯ 7.0%. This increase was dose-dependently inhibited by insulin reducing thrombinstimulated glucose transport to 18.8 Ϯ 14.4% at 100 nM insulin (Fig. 8C).
To confirm that the reduced glucose uptake following preincubation with insulin was caused by inhibition of surface expression mediated via ␣-granule release, the expression of P-selectin was determined. As expected, insulin inhibited thrombin-induced P-selectin expression reaching 17.9 Ϯ 4.2% inhibition at 100 nM insulin FIGURE 5. Glucose uptake by platelets in medium with low glucose. Platelets were isolated in a low glucose medium to investigate glucose uptake. A, platelets were preincubated with [ 3 H]DOG followed by stimulation with thrombin or insulin. The inset in A depicts thrombin-and insulin-induced glucose uptake corrected for basal uptake. Basal glucose uptake increased for 10 min after which it leveled off. Thrombin and insulin increased basal glucose uptake to 84.3 Ϯ 40.7 AU (n ϭ 5, p Ͻ 0.01) and 94.0 Ϯ 33.2 AU (p Ͻ 0.001) after 30 min and correction for basal levels. Thrombin-induced glucose uptake was faster than that induced by insulin. Insulin-induced glucose uptake showed a delay of 10 min. B, platelets were preincubated with [ 3 H]DOG followed by stimulation with thrombin and/or insulin for 30 min. Platelets were preincubated with insulin for 10 min prior to stimulation with thrombin. Basal glucose uptake after 30 min was expressed as 100%. Thrombin and insulin increased basal glucose uptake to 195.0 Ϯ 19.0% and 188.1 Ϯ 19.0% (n ϭ 5, p Ͻ 0.001). Co-treatment of thrombin and insulin did not further increase glucose uptake (200.0 Ϯ 10.0%; ns). C, platelets were preincubated with ML-9 and stimulated with thrombin or insulin for 30 min. Thrombin and insulin increased basal glucose uptake to 212.7 Ϯ 35.0% (p Ͻ 0.001) and 188.1 Ϯ 39.4% (p Ͻ 0.01), respectively. ML-9 inhibited basal, thrombin-and insulin-induced glucose uptake to 66.5 Ϯ 16.4% (p Ͻ 0.01), 149.1 Ϯ 17.2% (p Ͻ 0.01), and 115.4 Ϯ 36.1% (n ϭ 5, p Ͻ 0.05), respectively. (Fig. 8D). Together, these data show that, at physiological glucose levels, glucose uptake is mainly regulated via changes in surface expression as a result of ␣-granule release. They also demonstrate that relative small changes in surface expression can be accompanied by large changes in glucose uptake.
To determine the contribution of GLUT1 to glucose uptake in platelets, platelets were incubated with different concentrations of glucose (0 -30 mM) for 10 min, and glucose uptake was measured. At low glucose concentrations (0.5-2 mM) uptake was substantial. It leveled off at 5 mM glucose, which is about four times the K m of GLUT3. At 5-30 mM glucose, with 30 mM being close to the K m of GLUT1, a slight increase in glucose uptake was observed. The results were analyzed using non-linear regression. They showed a slightly better fit with a two-site binding model (R 2 ϭ 0.8079) than with a one-site binding model (R 2 ϭ 0.8061) indicating the presence of two glucose transporters (Fig. 9). Thus, although a contribution of GLUT-1 cannot be excluded, these data show that glucose uptake by platelets is predominantly via GLUT3.
PKB Does Not Regulate Platelet Inhibition by Insulin-The inhibition of thrombin-stimulated glucose uptake by insulin at physiological glucose concentration (Fig. 8C) agrees with the inhibition of Ca 2ϩ mobilization and aggregation by interference with signaling through P2Y 12 (16). Because insulin is also an activator of PKB, we investigated whether PKB plays a role in the effect of insulin on P2Y 12 signaling. Fig. 10 (A and  B) shows that the adenylyl cyclase inhibitor SQ22536 fully blocks the inhibition by insulin, an effect also seen with the P2Y 12 antagonist ARC69931MX (16,36) illustrating the importance of cAMP regulation through P2Y 12 . Direct analysis of Ca 2ϩ changes in thrombin-stimulated platelets confirmed the decrease induced by preincubation with insulin. This inhibition was unaffected by ML-9 (Fig. 10C). Thus, both throm-bin-induced Ca 2ϩ changes and the effect of insulin on that response are unaffected by PKB.

DISCUSSION
The present study shows that platelet stimulation with thrombin and insulin activate PKB, an enzyme intimately involved in the regulation of glucose uptake by insulin responsive tissues, such as adipocytes and muscle cells (15). Both agonists also stimulate the uptake of glucose by platelets. This has been demonstrated earlier for thrombin and shown to be caused by translocation of GLUT3 from ␣-granule membranes to the platelet surface upon induction of secretion (2,3). The finding that also insulin increases glucose uptake is therefore surprising, because it is incapable of inducing ␣-granule release. On the other hand, these findings agree with observations in many insulin-responsive tissues where insulin triggers a strong, PKB-mediated, glucose uptake via signaling through PI3K (15). In adipocytes, glucose uptake by GLUT4 is up-regulated by an increase in surface expression, together with an increase in affinity for glucose (14).
Data in resting platelets reveal a role for PKB in basal glucose transport. The PKB inhibitor ML-9 induces a 30% reduction suggesting that the basal activity of PKB-Ser 473/4 phosphorylation is important for glucose uptake by resting platelets. In theory, this uptake is the result of glucose transport through both GLUT1, a ubiquitously expressed transporter in the PM, and through GLUT3 of which ϳ15% of the cellular pool is present in the PM. Because GLUT1 is insensitive to control by PKB (35) and the contribution to glucose uptake in platelets is minimal, the inhibition induced by ML-9 probably reflects the participation of GLUT3. Furthermore, the low K m of basal glucose transport and 13-fold increase in GLUT3 mRNA compared with GLUT1 mRNA in the mature megakaryocytic cell line CHRF also support the predominant role of GLUT3 in glucose uptake by platelets.
The rapid increase in glucose uptake following stimulation by thrombin agrees with rapid activation of PKB and the immediate induction of ␣-granule secretion induced by a relatively high thrombin concentration (0.25 unit/ml). Again it is likely to reflect changes in the activity of GLUT3. The slower stimulation of glucose uptake by insulin agrees with the slower activation of PKB. Because insulin does not induce secretion, the increased uptake must reflect an affinity change in GLUT3. Indeed, insulin induced a 2-fold decrease in the K m for glucose without changing the V max . Apparently, insulin increases the GLUT3 affinity in platelets via a similar mechanism that induces the increase in GLUT4 affinity in adipocytes (14). GLUT3 is a 12-transmembrane (TM)-spanning protein and one has proposed that changes in key TM helices modulate glucose transport (37).
Studies on insulin-induced glucose transport in platelets are compli-  cated by the fact that insulin is an inhibitor of Ca 2ϩ mobilization through interference with P2Y 12 signaling, which supports platelet activation upon release of ADP. Interference by insulin is absent in the absence of a secretion-inducing agents, and insulin alone neither changes the basal cAMP level nor Ca 2ϩ homeostasis (16). In experiments with co-stimulation by thrombin and insulin, one might therefore expect inhibition of thrombin-induced glucose uptake. In low glucose medium such an inhibition is absent suggesting that a possible decrease in GLUT3 surface expression is compensated by an increase in transport affinity. In high glucose medium, such an inhibition by insulin is apparent leading to a dose-dependent reduction in thrombin-induced glucose uptake. At this condition GLUT3 functions near saturation, indicating that the fall in glucose uptake reflects a fall in GLUT3 surface expression.
The question whether in addition to an increase in GLUT3 expression thrombin induces an affinity change was addressed by incubating platelets in a low glucose medium where glucose uptake depends both on GLUT1 and predominantly on GLUT3. Kinetic analysis in thrombin-stimulated platelets revealed both an increase in V max and a decrease in the K m for glucose. Again, these properties likely reflect changes in GLUT3, because GLUT1 is not known to change its subcellular localization or its affinity to glucose (10,11). Assuming that the surface expression of P-selectin is a reliable marker for expression of GLUT3 located in the ␣-granule membrane, one might expect that changes in P-selectin expression and glucose transport go hand in hand. FIGURE 9. Glucose uptake at different glucose concentrations. Platelets were isolated in a medium without glucose. Glucose transport was initiated with different [ 3 H]DOG concentrations (0.5-30 mM) and continued for 10 min. Glucose transport at 0.5 mM was expressed as 100%. Glucose uptake at 0, 0.5, 1, 2, 5, 10, 20, and 30 mM was 11.1 Ϯ 0.2%, 100%, 132.3 Ϯ 9.2%, 311.2 Ϯ 83.8%, 371.2 Ϯ 101.2%, 429.3 Ϯ 104.8, 491.6 Ϯ 150.5%, and 535.7 Ϯ 169.7%, respectively. The increase in glucose uptake was pronounced in the low glucose concentration range (0.5-2 mM). Uptake at 1 mM glucose differed from uptake at 0.5 mM (p Ͻ 0.0001), and 2 mM from 1 mM (p Ͻ 0.006). Glucose uptake leveled off at 5 mM glucose (p ϭ 0.39). Uptake at 30 mM glucose did not differ from uptake at 5 mM (p ϭ 0.16). The graph was fitted by non-linear regression using a two-site binding model (R 2 ϭ 0.8079). Data are means Ϯ S.D., n ϭ 4. . D, platelets were preincubated with insulin and stimulated by thrombin, and the expression of P-selectin was determined by flow cytometry. Thrombin-induced P-selectin expression was set as 100%. Insulin alone did not increase basal P-selectin expression (7.5 Ϯ 4.6% and 7.9 Ϯ 3.4%, respectively; ns). Insulin inhibited thrombin-induced P-selectin expression to 82.1 Ϯ 4.2% (p Ͻ 0.001). Data are means Ϯ S.D., n ϭ 6.
This appears not to be the case. Compared with resting platelets, thrombin induced a 12-fold increase in P-selectin expression and only a 2-fold increase in glucose transport. Insulin inhibited thrombin-glucose uptake to levels of resting platelets while decreasing P-selectin expression by not more than 20%. A similar discrepancy between glucose uptake and GLUT expression has been observed for GLUT4 in muscle cells, which has been explained by assuming that additional factors modulate the full functional expression of GLUTs at the cell surface (14). The characteristics of GLUT3 regulation in platelets resemble those of GLUT4 in adipocytes and muscle cells. In addition to being stimulated by insulin, GLUT3 and GLUT4 share the soluble N-ethylmaleimide-sensitive attachment protein receptors (SNAREs) that mediate the fusion of GLUT3 and GLUT4 with the PM (34,38). These SNARE proteins are located in lipid rafts, which are detergent-resistant and cholesterol-and sphingolipid-rich membrane domains involved in important cellular processes such as signal transduction and intracellular trafficking (39). Interestingly, a major part of the proteins described in this study such as GLUT3 (39), the IR/IRS-1 complex (40), G i (39), and PKB (41) are associated with lipid rafts revealing a highly organized signaling complex that regulates glucose uptake.
Even though PKB is crucial in insulin-induced glucose uptake, inhibition of the kinase does not abolish the inhibitory properties of insulin. The PKB inhibitor ML-9 left thrombin-induced Ca 2ϩ mobilization and its interference by insulin undisturbed. The inhibition of P2Y 12 signaling through inactivation of G i ␣ 2 by insulin accords with the anti-lipolytic effect of insulin in 3T3-L1 adipocytes. Both effects are independent of signaling through PI3K but result from G i ␣ 2 inactivation by the IR⅐IRS-1 complex (16,42). Thus, the signaling events triggered by insulin can be separated in those directed at the maintenance of glucose homeostasis, in which activation of PKB is crucial, and those directed at the regulation of platelet activity, in which the activity of G i plays a vital role.
A comparison between the effects of ML-9 on the different platelets responses learns that in addition to glucose uptake, PKB plays a role in ␣-granule release and aggregation. However, a concentration that induces complete inhibition of PKB (100 M), leads to complete inhibition of aggregation but incomplete inhibition of glucose uptake and P-selectin expression. It is clear that in addition to roles in secretion response and glucose uptake, which are both important for aggregation, PKB has other effects that directly contribute to the aggregation response. Examples are a role in regulation of integrin ␤ 3 activity (43) and re-assembly of the actin cytoskeleton (44). The fact that insulin activates PKB without inducing aggregation illustrates that PKB activation alone is an insufficient stimulus to induce platelet-platelet interaction. A response that is clearly independent of PKB is the mobilization of Ca 2ϩ induced by thrombin. PKB is activated via PI3K and Ca 2ϩ -dependent protein kinase C subtypes (19). Recent studies show that inhibition of PI3K with LY294002 inhibited Ca 2ϩ mobilization (36) and that PI3K regulates the entry of Ca 2ϩ in the endoplasmic reticulum (45). Further studies are therefore required to clarify the roles of PKB and its upstream regulators in more detail.