The Rab GTPase-Activating Protein AS160 Integrates Akt, Protein Kinase C, and AMP-Activated Protein Kinase Signals Regulating GLUT4 Traffic Free

Plasmids encoding FLAG-tagged wild-type AS160 (WT-AS160), AS160 with four of its six predicted Akt phosphorylation sites (Ser³¹⁸, Ser⁵⁸⁸, Thr⁶⁴², and Ser⁷⁵¹) mutated to alanine (4P-AS160), AS160 with arginine 973 mutated to lysine (RK-AS160), and AS160 with mutated Akt phosphorylation sites and R973K (4PRK-AS160) (4) were generously provided by Dr. G.E. Lienhard (Dartmouth Medical School, Hanover, NH). Hemagglutinin (HA)-tagged kinase-inactive Akt (AAA-Akt) with point mutations at Lys¹⁷⁹, Thr³⁰⁸, and Ser⁴⁷³ (24) was a gift from Dr. J.R. Woodgett (Ontario Cancer Institute/Princess Margaret Hospital, Toronto, Ontario, Canada). HA-tagged α2-AMPK (α2-AMPK) was provided by Dr. L.A. Witters (Dartmouth Medical School). Chinese hamster ovary fibroblasts stably overexpressing the insulin receptor (CHO-IR) were from Dr. C. Yip (University of Toronto).

L6 myoblasts stably expressing GLUT4 with an exofacial myc epitope (L6 GLUT4myc) and CHO-IR cells were maintained in monolayer culture as previously described (24,25). Transfection of L6 GLUT4myc myoblasts grown on 25-mm glass coverslips and CHO-IR cells was performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). All experiments were carried out at 18–24 h posttransfection. In CHO-IR and L6 cells, the transfection efficiency of the FLAG-AS160 constructs tested was ∼80 and 40%, respectively. Expression levels in CHO-IR cells are shown in each figure. The expression of wild-type and mutant AS160 constructs in L6 cells was comparable, and surface GLUT4myc levels were quantified through single cell assays.

CHO-IR cells expressing FLAG WT-AS160 were left unstimulated (basal) or stimulated with 100 nmol/l insulin, 50 ng/ml PDGF, 2 mmol/l AICAR, 0.5 mmol/l DNP, 0.45 mol/l sucrose, and 1 μmol/l 4-phorbol-12-myristate-13-acetate (PMA) in serum-free α-minimal essential media or incubated in HEPES-buffered saline in which 120 mmol/l K⁺ gluconate isosmotically replaced NaCl at 37°C. Cells were lysed in buffer containing 30 mmol/l HEPES (pH 7.6), 150 mmol/l NaCl, 1% (vol/vol) Triton X-100, 1 mmol/l EDTA, 30 mmol/l Na₄P₂O₇, 25 mmol/l β-glycerophosphate, 10 mmol/l NaF, 1 mmol/l Na₃VO4, 100 nmol/l okadaic acid, and protease inhibitor cocktail (1 mmol/l benzamidine, 10 μmol/l E-64, 1 μmol/l leupeptin, and 0.2 mmol/l phenylmethylsulfonyl fluoride; Sigma-Aldrich, St. Louis, MO). AS160 was immunoprecipitated with a monoclonal anti-FLAG (1 μg antibody/mg protein; Sigma-Aldrich) rotating overnight at 4°C. A similar lysis and immunoprecipitation with anti-AS160 (Upstate Cell Signaling Solutions, Lake Placid, NY) was done with untransfected L6 cells. The immunocomplex and the supernatant were resuspended in Laemmli buffer containing β-mercaptoethanol and heated for 5 min at 95°C. Proteins were resolved on SDS-PAGE, transferred to PVDF membranes, and blocked with 5% milk. The immunocomplex was separated into two equal portions, each of which were immunoblotted with anti–phospho-Akt (pS/pT) substrate (Cell Signaling Technology, Beverly, MA) and anti-AS160 antibodies (a kind gift from Dr. G.E. Lienhard [2] or, in later experiments, from Upstate Cell Signaling Solutions). Membranes from the supernatants were incubated with anti-AS160, anti–phospho-Akt Ser⁴⁷³, anti–phospho-Akt Thr³⁰⁸, anti–phospho-AMPK Thr¹⁷² (Cell Signaling Technology), or anti-HA (BAbCO, Beverly, CA) antibodies. Membranes were washed in Tris-buffered saline with 0.02% Tween, incubated with horseradish peroxidase–conjugated secondary antibodies (Jackson Immunoresearch, West Grove, PA), visualized by enhanced chemiluminescence (Bio-Rad, Richmond, CA), and quantified using ImageJ software (National Institutes of Health, Bethesda, MD). Membranes were stripped by incubating in 62.5 mmol/l Tris-HCl, pH 6.7, 2% (wt/vol) SDS, and 100 mmol/l β-mercaptoethanol at 50°C, blocked in 5% milk, and reprobed with anti-AS160, anti-Akt1/2, or anti-AMPK antibodies.

Surface GLUT4myc, Akt phosphorylation, and actin remodeling were detected by indirect fluorescence microscopy as previously described (12) with slight modifications. Serum-starved L6 cells were unstimulated (basal) or stimulated with 100 nmol/l insulin, 50 ng/ml PDGF, 120 mmol/l K⁺ gluconate, 1 μmol/l PMA, 2 mmol/l AICAR, 0.5 mmol/l DNP, or 0.45 mol/l sucrose at 37°C. Surface GLUT4myc was detected by incubating intact cells with anti-myc antibody (1:200) (Sigma-Aldrich) followed by Alexa 488–conjugated goat anti-rabbit IgG (1:750) (Molecular Probes, Eugene, OR) fixed with 4% paraformaldehyde and permeabilized with 0.1% (vol/vol) Triton X-100. Cells expressing epitope-tagged proteins were identified by immunostaining with the appropriate primary antibodies (monoclonal anti-FLAG antibody [1:5,000], polyclonal anti-FLAG antibody [1:750], or monoclonal anti-HA antibody [1:3,000]) followed by Cy3- or Cy5-conjugated anti-mouse or anti-rabbit IgG (1:750) (Jackson Immunoresearch). To assess actin remodeling or Akt phosphorylation, cells were unstimulated (basal) or stimulated with insulin or PDGF, fixed with 4% paraformaldehyde, permeabilized with 0.1% (vol/vol) Triton X-100, and incubated with rhodamine-conjugated phalloidin (Molecular Probes) or anti–phospho-Akt Ser⁴⁷³ immunohistochemistry-compatible antibody (Cell Signaling Technology) followed by Alexa 488–conjugated goat anti-rabbit IgG (1:750) as previously described (12). Fluorescent images were obtained with a Zeiss LSM 510 laser-scanning confocal microscope. Whole cells were scanned along the z-axis, and a single composite image (collapsed xy projection) of the optical cuts per cell was generated using LSM5 Image software. Surface GLUT4myc was quantified using ImageJ software (National Institutes of Health).

Statistical analyses were performed with one-way ANOVA (SigmaStat, Chicago, IL). Tukey post hoc comparisons were made when statistical significance (P < 0.05) was found between observations. Data are presented as means ± SE.

While the increase in surface GLUT4 induced by PDGF occurs in parallel to Akt activation (12,13), it was not known if Akt activation is required for this outcome. Here, surface GLUT4myc was quantified in L6 myoblasts transfected with a kinase-inactive Akt mutant (AAA-Akt) (24). Surface GLUT4myc levels in unstimulated cells were unaffected by AAA-Akt, but PDGF-induced GLUT4myc exocytosis was reduced by 87 ± 2% (Fig. 1).

AS160 phosphorylation is typically detected from the reactivity of a 160-kDa band with the PAS antibody (3–6). Using this approach, we observed an insulin-dependent phosphorylation of a 160-kDa band in L6 cells (results not shown). To specifically define whether PDGF increases AS160 phosphorylation, we measured phosphorylation of transfected WT-AS160 in CHO-IR cells, which exhibit high transfection efficiency. PDGF and insulin treatments elevated AS160 phosphorylation by 3.2 ± 0.8- and 6.6 ± 0.9-fold, respectively (Fig. 2). These responses were prevented by preincubation with the PI3K inhibitor LY294002 (data not shown). We therefore explored whether AS160 is a component of the signaling cascade in PDGF-induced GLUT4 translocation. In control cells, insulin and PDGF raised surface GLUT4myc by 2.1 ± 0.1- and 2.4 ± 0.2-fold, respectively (Fig. 3E). In unstimulated cells, WT-AS160 reduced surface GLUT4myc (Fig. 3A and E); however, it did not affect the gain in surface GLUT4myc upon insulin or PDGF stimulation (Fig . 3A and E). In unstimulated cells expressing 4P-AS160, surface GLUT4myc was lower compared with control (Fig. 3B and E). However, 4P-AS160 fully prevented GLUT4myc translocation triggered by insulin or PDGF (Fig. 3B and E). The observation that 4P-AS160 expression reduced surface GLUT4myc levels in nontreated cells complicates interpretation of the effects of 4P-AS160 on the multiple stimuli to increase surface GLUT4myc. In light of this, we calculated the net gains in surface GLUT4myc by subtracting the increase in response to each stimulus from the respective basal. These calculations reveal an action of the mutant on the actual responses and no longer include the residual effect of 4P-AS160 in the unstimulated state. In cells expressing 4P-AS160, the net gains in surface transporters in response to insulin or PDGF remained markedly lower compared with control (Fig. 3E, inset). In contrast, the insulin- or PDGF-stimulated gains in surface GLUT4myc were unaffected by RK-AS160 or 4PRK-AS160. Moreover, in the basal state, surface GLUT4myc levels were not diminished by RK-AS160 or 4PRK-AS160 and were similar to those observed in control cells (Fig. 3C–E). The observed reduction in surface GLUT4myc in unstimulated cells expressing WT-AS160 or 4P-AS160 is likely due to the augmented cellular Rab-GAP activity from the overexpressed protein. This interpretation is supported by the reciprocal observation that knockdown of endogenous AS160 in 3T3-L1 adipocytes by siRNA increases plasma membrane GLUT4 levels by enhancing the basal GLUT4 recycling rate (3,6).

Additional experiments demonstrated that insulin- or PDGF-induced Akt phosphorylation was unaffected by 4P-AS160 transfection (Fig. 4A). Similarly, 4P-AS160 did not influence insulin- or PDGF-regulated actin remodeling (Fig. 4B). Thus, the inhibitory effects of 4P-AS160 on insulin- or PDGF-stimulated GLUT4myc translocation did not result from 4P-AS160 sequestering Akt ot disrupting insulin-induced actin remodeling, a phenomenon that is Akt independent yet essential for insulin-dependent GLUT4 translocation (23,24).

Muscle contraction/exercise stimulates glucose uptake, an effect that may rely upon activation of AMPK, c/n PKC, and/or CaMK (20). Similarly, K⁺ depolarization (14) and hyperosmolarity (16) also increase surface GLUT4myc in L6 cells. However, these effects of K⁺ depolarization and hyperosmolarity occur independently of the PI3K and AMPK pathways (14,16,17) and may involve the participation of c/n PKC instead (14,21). Moreover, phorbol esters, which activate c/n PKC, promote GLUT4 exocytosis (26) and glucose uptake (27,28). Thus, we assessed if these activators of c/n PKC would increase AS160 phosphorylation and if AS160 is downstream of these signaling cascades that raise surface GLUT4myc content.

In CHO-IR cells, AS160 phosphorylation was elevated by K⁺ depolarization (2.6 ± 0.4-fold), PMA (5.1 ± 0.6-fold), and hyperosmolar sucrose (6.7 ± 1.1-fold) (Fig. 5). Pretreatment of cells with bisindolylmaleimide (BIM), at concentrations (1 μmol/l) that only inhibit c/n PKC, reduced AS160 phosphorylation elicited by these stimuli (Fig. 5). Akt phosphorylation was unaffected by these stimuli (Fig. 5).

We next evaluated if the gain in surface GLUT4myc observed with these stimuli engages AS160 in L6 cells. Surface GLUT4myc levels were increased by K⁺ depolarization (1.6 ± 0.1-fold), PMA (2.0 ± 0.2-fold), or hyperosmolar sucrose (2.3 ± 0.1-fold) (Fig. 6A and E). WT-AS160, RK-AS160, or 4PRK-AS160 did not alter the increased membrane GLUT4myc elicited by these stimuli (Fig. 6A, d, and E). In contrast, 4P-AS160 fully prevented the gain in surface GLUT4myc stimulated by K⁺ depolarization and partly but significantly reduced the gains evoked by PMA and hyperosmolarity (Fig. 6B and E). The net gains in GLUT4myc in response to K⁺ depolarization and PMA, but not to hyperosmolar sucrose, remained significantly lower in cells expressing 4P-AS160 (Fig. 6E, inset).

In skeletal muscle or muscle cells, AMPK activation promotes GLUT4 exocytosis (17,29). AMPK exists as a heterotrimeric complex composed of α catalytic subunits with β and γ regulatory subunits (30). There are two isoforms of the catalytic subunit (α1 and -2); however, the stimulatory effects of AICAR (17,29) and hypoxia (19), but not hyperosmolarity (17), on glucose uptake requires α2-AMPK. Moreover, rodent skeletal muscle expresses more α2-AMPK, and a greater proportion of AMPK activity can be attributed to the α2-AMPK heterotrimeric complex (31).

We first assessed if α2-AMPK activation increases AS160 phosphorylation. In CHO-IR cells (which likely lack α2-AMPK), neither AICAR nor DNP caused WT-AS160 phosphorylation. However, upon transfection of α2-AMPK, WT-AS160 phosphorylation was increased by AICAR (2.7 ± 0.5-fold) or DNP (2.5 ± 0.3-fold) (Fig. 7). Conversely, hyperosmolar sucrose evoked WT-AS160 phosphorylation in CHO-IR cells, and this response was not further elevated by α2-AMPK transfection (Fig. 7). None of these stimuli affected Akt phosphorylation (Fig. 7).

We next explored if AS160 is a component of AMPK signaling leading to GLUT4 traffic. L6 cells express α2-AMPK, but total cellular AMPK activity is dominated by the α1-catalytic subunit (32). Since α2-AMPK mediates AICAR-stimulated glucose uptake in skeletal muscle (17,29) and α2-AMPK was required for AICAR- and DNP-stimulated AS160 phosphorylation in CHO-IR cells (Fig. 7), we cotransfected the α2-subunit with WT-AS160 or 4P-AS160 in L6 cells. In control cells, surface GLUT4myc density was raised by AICAR (2.0 ± 0.1-fold), DNP (1.6 ± 0.1-fold), and hyperosmolar sucrose (2.2 ± 0.1-fold) (Fig. 8E). The gains in surface transporters elicited by these stimuli were unaffected by WT-AS160, RK-AS160, or 4PRK-AS160 (Fig. 8A and C–E). In contrast, 4P-AS160 significantly reduced surface GLUT4myc in response to AICAR or DNP (Fig. 8B and E). In cells treated with hyperosmolar sucrose, 4P-AS160 modestly reduced surface GLUT4myc content (Fig. 8B and E) akin to that observed in the absence of α2-AMPK cotransfection (Fig. 6B and E). These data suggest that AS160 mediates GLUT4myc translocation in response to AICAR or DNP, especially in the context of higher α2-AMPK levels, but does not contribute to the rise in surface GLUT4 evoked by hyperosmolarity. The calculated net gains in GLUT4myc in response to AICAR, but not to either DNP or hyperosmolar sucrose, remained significantly lower in 4P-AS160–transfected cells (Fig. 8E, inset).

The identification of AS160 as an Akt substrate provides a link between insulin signaling and regulation of GLUT4 traffic. A current model proposes that the Rab-GAP activity of AS160 retains GLUT4 vesicles intracellularly in the basal state (3,6). This clamping action of AS160 on GLUT4 traffic is then relieved upon insulin stimulation through Akt phosphorylation, effectively uncoupling it from its regulation of a downstream Rab protein. The exact level or location at which insulin signaling pathways integrate with GLUT4 vesicle traffic is not clear, but recent studies suggest that in adipocytes, tethering and fusion of GLUT4 vesicles with the plasma membrane are a major regulated event (33,34) and that Akt activity is required for events beyond the arrival of vesicles near the membrane (34).

In addition to insulin, contraction increases the amount of GLUT4 at the plasma membrane (1,20). Intriguingly, these two stimuli require different early signal transduction pathways, since activation of class IA PI3K and Akt are not required for contraction-induced GLUT4 exocytosis or glucose uptake (20,35). Although contraction activates AMPK and activation of this enzyme by AICAR increases glucose uptake via GLUT4, there does not appear to be a causal role of AMPK in contraction-activated glucose uptake (17,19,20,29). Nonetheless, pathophysiological challenges such as hypoxia and DNP engage AMPK to stimulate glucose uptake (19). Moreover, AMPK has emerged as a viable pharmacological target in the treatment of type 2 diabetes (36,37), and chronic treatment of obese rodents with an AMPK activator improved glucose tolerance (38). The signals leading to the contraction-elicited GLUT4 exocytosis thus remain unknown, and there is circumstantial evidence suggesting that c/n PKC may participate in this response (28,39,40). Recently, we demonstrated that membrane depolarization, a prelude to muscle contraction, stimulates c/n PKC (21) and AMPK (14), but only the former mediates the depolarization stimulation of GLUT4 redistribution to the cell surface (14).

Given that such diverse stimuli increase surface GLUT4 levels in muscle cells, it was of interest to explore if the distinct signaling pathways engaged by the stimuli converge at a downstream step. Although AS160 was identified as an Akt substrate, examination of its linear sequence reveals putative phosphorylation sites by a variety of kinases, including AMPK and PKC. Hence, the purpose of this study was to examine if AS160 constitutes a convergence point in the increase in surface GLUT4 in muscle cells elicited by stimuli engaging three different signaling pathways (Akt, AMPK, and c/n PKC).

PDGF stands out due to its ability to activate Akt and GLUT4 translocation in cells that coexpress PDGF receptors and GLUT4 (12,13,41). Before the current research was undertaken, it was not known whether Akt is requisite to achieving GLUT4 externalization by PDGF. We show here that PDGF increased AS160 phosphorylation, and this effect was prevented by pretreatment with the PI3K inhibitor LY294002. Further, the gain in surface GLUT4 elicited by PDGF was abolished by a kinase-inactive Akt mutant. In addition, 4P-AS160 blocked the rise in surface GLUT4myc levels caused by PDGF, whereas RK-AS160 and 4PRK-AS160 (with inactive GAP activity) did not. Collectively, these results indicate that AS160 functionally participates not only in insulin- but also in PDGF-induced GLUT4 translocation in L6 cells downstream of PI3K.

AS160 phosphorylation was determined with the PAS antibody that recognizes the consensus Akt phosphorylation motif peptide sequence RXRXX(pS/pT). Intriguingly, this antibody also reacted with AS160 in cells treated with activators of α2-AMPK or c/n PKC, which do not stimulate Akt. Thus, AMPK and PKC can theoretically target the same AS160 sites phosphorylated by Akt and recognized by the PAS antibody. Indeed, of the six Akt phosphorylation sites, S318 (FRSRCSpSVTGV) and S588 (MRGRLGpSVDSF) have putative AMPK consensus phosphorylation motifs in the context φ XXRXX(pS/pT)XXXφ, where φ is a hydrophobic residue (42). In addition, S570 (KRSLTSpSLENI) has some features of an AMPK consensus phosphorylation motif, and five of the six Akt phosphorylation sites on AS160 fit the minimal PKC consensus phosphorylation motif, RXXpS (43). These observations prompted us to analyze the functional role of AS160 upon activation of AMPK and c/n PKC.

Muscle contraction triggers PKC translocation to the plasma membrane (44), and this has been argued to be causally related to the observed increase in glucose uptake (40). Here we found that K⁺ depolarization, PMA, and hyperosmolar sucrose all increased WT-AS160 phosphorylation. The c/n PKC inhibitor BIM prevented WT-AS160 phosphorylation evoked by K⁺ depolarization and PMA and reduced that elicited by hyperosmolar sucrose, suggesting that these stimuli lead to AS160 phosphorylation at least in part through c/n PKC. 4P-AS160 expression abolished the rise in surface GLUT4myc triggered by K⁺ depolarization. Intriguingly, however, 4P-AS160 only partially reduced the gain in surface GLUT4myc caused by PMA and minimally by hyperosmolar sucrose (Fig. 6E). Further studies will be required to identify the precise kinases and signals involved in PMA-induced GLUT4 translocation. However, our previous study (14) rules out AMPK and CaMKII as possible mediators of the K⁺ depolarization response.

We previously demonstrated that K⁺ depolarization largely increases surface GLUT4myc in L6 cells by inhibiting its endocytosis, possibly via activation of c/n PKC (14,21). Although AS160 is not required for insulin regulation of GLUT4 endocytosis in 3T3-L1 adipocytes (5), it is possible that AS160 might participate in regulating GLUT4 endocytosis in response to K⁺ depolarization. In L6 cells, hyperosmolar sucrose also inhibits GLUT4 endocytosis (12). However, 4P-AS160 did not significantly affect the net gains in surface GLUT4myc induced by hyperosmolar sucrose. These data imply that inhibition of GLUT4 endocytosis by hyperosmolar sucrose is not mediated by AS160. Instead it is likely caused by hyperosmolar sucrose preventing clathrin-coated pit formation (45).

AS160 phosphorylation is increased by AICAR or muscle contraction in rodent muscle (8), and when this article was being prepared for publication, two studies (46,47) showed that α2-AMPK is responsible for the AICAR and possibly the contraction effects. DNP elevates surface GLUT4 and glucose uptake (14,16), and in L6 cells, this process is dependent on AMPK activity (14). Here we show that AICAR and DNP increase AS160 phosphorylation in CHO-IR cells, but only when the cells expressed α2-AMPK. Moreover, in L6 cells expressing α2-AMPK, 4P-AS160 largely prevented the AICAR- and DNP-induced increase in surface GLUT4myc.

However, when net gains in surface GLUT4myc were calculated, the AICAR response remained significant, whereas the DNP response was not significantly altered by 4P-AS160 (Fig. 8E). One interpretation is that AS160 is not a component mediating the stimulatory actions of DNP via AMPK or other signaling proteins on GLUT4 exocytosis. Alternatively, in addition to activating AMPK, DNP also raises intracellular Ca²⁺ (48), and in L6 cells, chelating intracellular Ca²⁺ reduced DNP-stimulated glucose uptake (16). AS160 has a Ca²⁺/calmodulin binding domain, and its binding to AS160 has been proposed to disable its GAP activity (49). Thus, the inhibitory effect of 4P-AS160 on net gain in surface GLUT4myc in DNP-treated cells may have been opposed by Ca²⁺/calmodulin binding to AS160, resulting in partial deactivation of its GAP activity.

The results presented indicate that AS160 can be phosphorylated in response to agents activating Akt, c/n PKC, and AMPK. The specific sites phosphorylated in response to each stimulus must now be identified. The outcome of each of these phosphorylation events may depend on the site modified, as well as on the duration of the signal and localization of AS160. In this way, AS160 may integrate distinct signals that could conceivably control its GAP activity toward individual Rab proteins that may be associated with GLUT4 vesicle movement, tethering, or fusion with the plasma membrane (6,7).

While our results indicate that both insulin-dependent and -independent signaling cascades that mutually increase plasma membrane GLUT4 levels intersect at AS160, they do not suggest that AS160 is the end point in regulating GLUT4 traffic. Hence, the possibility exists that these signaling cascades may diverge from AS160 and subsequently target distinct downstream effectors to regulate GLUT4 traffic. It should also be considered that PKC and AMPK may phosphorylate AS160 on sites not recognized by the PAS antibody, and such phosphorylations may have additional functional effects on AS160 and GLUT4 traffic. Moreover, the fact that neither AS160 mutant with disabled GAP activity (RK or 4PRK) elevated surface GLUT4myc above levels observed in control cells suggests that AS160 is necessary but insufficient for GLUT4 traffic. It is also possible that these stimuli, like insulin (3,6), utilize both AS160-dependent and -independent mechanisms to promote a gain in surface transporters.

Exercise represents a viable mechanism to bypass insulin resistance of glucose uptake in obese and diabetic patients. Therefore, unraveling the signals elicited by exercise will be paramount in our ability to relieve insulin resistance. AMPK and c/n PKC have been implicated as partial contributors to exercise-stimulated glucose uptake. The present discovery (i.e., AS160 is downstream of c/n PKC and AMPK, leading to increases in surface GLUT4) makes AS160 a key target for modulation and improvement of insulin sensitivity. Future studies should consider strategies to deactivate AS160 for therapeutic purposes, specifically in the sites modified by Akt, c/n PKC, or AMPK.

This work was supported by grant MT12601 from the Canadian Institutes of Health Research (CIHR) to A.K. F.S.L.T. was supported by postdoctoral research fellowships from Natural Sciences and Engineering Research Council of Canada and CIHR.

We are grateful to Dr. G.E. Lienhard for the AS160 cDNAs and anti-AS160 antibody, Dr. J.R. Woodgett for the AAA-Akt cDNA, Dr. L.A. Witters for the α2-AMPK cDNA, and Dr. C.C. Yip for the CHO-IR cells used in this study.