Energy Depletion Inhibits Phosphatidylinositol 3-Kinase/Akt Signaling and Induces Apoptosis via AMP-activated Protein Kinase-dependent Phosphorylation of IRS-1 at Ser-794*

Energy depletion activates AMP-activated protein kinase (AMPK) and inhibits cell growth via TSC2-dependent suppression of mTORC1 signaling. Long term energy depletion also induces apoptosis by mechanisms that are not well understood to date. Here we show that AMPK, activated by energy depletion, inhibited cell survival by binding to and phosphorylating IRS-1 at Ser-794. Phosphorylation of IRS-1 at this site inhibited phosphatidylinositol 3-kinase/Akt signaling, suppressed the mitochondrial membrane potential, and promoted apoptosis. Of the treatments promoting energy depletion, glucose deprivation, hypoxia, and inhibition of ATP synthesis in the mitochondria stimulated phosphorylation of IRS-1 at Ser-794 via an LKB1/AMPK-dependent manner, whereas oxidative stress and 2-deoxyglucose stimulated phosphorylation at this site via a Ca2+/calmodulin-dependent protein kinase kinase β/AMPK axis. These data define a novel pathway that cooperates with other adaptive mechanisms to formulate the cellular response to energy depletion.

Energy depletion activates AMP-activated protein kinase (AMPK) and inhibits cell growth via TSC2-dependent suppression of mTORC1 signaling. Long term energy depletion also induces apoptosis by mechanisms that are not well understood to date. Here we show that AMPK, activated by energy depletion, inhibited cell survival by binding to and phosphorylating IRS-1 at Ser-794. Phosphorylation of IRS-1 at this site inhibited phosphatidylinositol 3-kinase/Akt signaling, suppressed the mitochondrial membrane potential, and promoted apoptosis. Of the treatments promoting energy depletion, glucose deprivation, hypoxia, and inhibition of ATP synthesis in the mitochondria stimulated phosphorylation of IRS-1 at Ser-794 via an LKB1/AMPK-dependent manner, whereas oxidative stress and 2-deoxyglucose stimulated phosphorylation at this site via a Ca 2؉ /calmodulin-dependent protein kinase kinase ␤/AMPK axis. These data define a novel pathway that cooperates with other adaptive mechanisms to formulate the cellular response to energy depletion.
Cells have developed adaptive mechanisms that sense energy depletion caused by a variety of insults including glucose deprivation, hypoxia, and oxidative stress. Triggering these adaptive mechanisms allows the cells to survive temporarily until the insult is removed. In response to energy depletion, AMPK, 2 a sensor of the cellular energy status, phosphorylates TSC2, which functions as a GTPase-activating protein for the small GTPase Rheb, an upstream activator of mTORC1 (1,2). Activation of TSC2 inhibits mTORC1 and limits energy consumption by down-regulating anabolic pathways such as protein synthesis (3,4).
AMPK is a heterotrimeric enzyme consisting of a catalytic subunit (␣1 or ␣2) and two regulatory subunits (␤1 or ␤2 and ␥1, ␥2, or ␥3). Both isoforms of the AMPK catalytic subunit form complexes with non-catalytic ␤ and ␥ subunits (5). AMPK senses the cellular energy status via four tandem pairs of cystathionine ␤-synthase domains, which are present in ␥ subunits and allosterically bind AMP and ATP (6). AMP binding is required for AMPK activation via phosphorylation by upstream kinases. At least two kinases can phosphorylate AMPK: LKB1, which transduces signals that are generated by changes in the cellular energy status (5,7), and Ca 2ϩ /calmodulin-dependent protein kinase kinase ␤ (CaMKK␤), which senses changes in intracellular calcium (5, 8 -11).
Energy depletion also affects the activity of the PI 3-kinase/ Akt pathway (12,13). The immediate effect of energy depletion is a transient activation of the pathway due to inhibition of the mTORC1/S6K1 axis, which in turn prevents the phosphorylation of IRS-1 at inhibitory phosphorylation sites (12)(13)(14). Following its immediate activation, the pathway is repressed because of subsequent inhibitory events, which are not well understood to date. Inhibition of the PI 3-kinase/Akt pathway by prolonged energy depletion promotes apoptosis.
IRS-1 plays a central role in the activation of the PI 3-kinase/ Akt pathway by insulin and insulin-like growth factor I signals.
In addition to its role in PI 3-kinase/Akt activation, IRS-1 also plays an important role in a feedback loop that inhibits the pathway. The feedback loop is triggered by mTORC1/S6K1, which is activated by insulin, insulin-like growth factor I, or nutrients (12)(13)(14)(15). The activated mTORC1/S6K1 pathway phosphorylates IRS-1 at several inhibitory sites (12)(13)(14)(15) and inhibits PI 3-kinase/Akt signaling. In addition to mTORC1/ S6K1, other kinases such as JNK and IB kinase-␤ also phosphorylate IRS-1 at inhibitory sites in the course of various pathological conditions such as inflammation and diabetes (16). The importance of IRS-1 in both the activation and the inhibition of the PI 3-kinase/Akt pathway by insulin and insulin-like growth factor I raised the question whether it may contribute to the inhibition of the PI 3-kinase/Akt pathway during energy stress.
Previous studies have shown that activation of AMPK by 5-aminoimidazole-4-carboxamide riboside in C2C12 myoblasts (17) or up-regulation of the AMPK family member saltinduced kinase-2 in dexamethasone-treated 3T3-L1 adipocytes (18) promotes the phosphorylation of IRS-1 at Ser-794. However, the natural stimuli that trigger phosphorylation of IRS-1 at Ser-794 remained unknown. Moreover the biological significance of this phosphorylation event remained undetermined with one study suggesting that it promotes (17) and other studies suggesting that it inhibits (18,19) PI 3-kinase/Akt signaling.
The work in this study focused on the nature, regulation, and biological significance of the events leading to the inhibition of the PI 3-kinase/Akt pathway during energy stress. Our data show that energy depletion induced by glucose deprivation, hypoxia, inhibition of ATP synthesis in mitochondria, oxidative stress, and 2-deoxyglucose (2-DG) activated AMPK, which binds to and phosphorylates IRS-1 at Ser-794. Glucose deprivation, hypoxia, and inhibition of ATP synthesis stimulated phosphorylation of IRS-1 at Ser-794 primarily via the ␣2 subunit of AMPK in an LKB1-dependent manner, whereas oxidative stress and 2-DG stimulated phosphorylation of IRS-1 at the same site primarily via the ␣1 subunit perhaps in a CaMKK␤dependent manner. Phosphorylation of IRS-1 at Ser-794 induced by energy depletion inhibited the PI 3-kinase/Akt pathway and promoted apoptosis by suppressing the mitochondrial membrane potential.
Short Interference RNA (siRNA) Transfection-siRNA against the human LKB1, the human AMPK ␣1 and ␣2 catalytic subunits, and the human IRS-1 were bought from Santa Cruz Biotechnology, and they were transfected using Lipofectamine 2000 (Invitrogen). Transfections were carried out using a final siRNA concentration of 80 nM. To knock down both the ␣1 and the ␣2 subunits of AMPK simultaneously, cells were transfected with a mixture of the two siRNAs (final concentration of 40 nM for each one, 80 nM combined). Transfection efficiency, measured with the use of a fluorescein-conjugated control nonspecific siRNA (Cell Signaling Technology catalog number 6201), was higher than 90%. Cells were grown for 48 -60 h before each experiment.
Antibodies-Anti-AMPK ␣1 and ␣2 antibodies were bought from Santa Cruz Biotechnology. Anti-AMPK antibody (it recognizes both the ␣1 and ␣2 catalytic subunits of AMPK) was bought from Cell Signaling Technology. Hif-1␣ antibody was bought from BD Transduction Laboratories. p85␣ antibody was bought from Upstate. All other antibodies, including secondary antibodies, were bought from Cell Signaling Technology.
Cell Lysis, Immunoprecipitation, and Western Blotting-Cells were washed in ice-cold PBS and solubilized in lysis buffer (50 mM Tris, pH 7.5, 200 mM NaCl, 1% Triton X-100, 10 mM Na 3 VO 4 , 50 mM NaF, 1 mM ␤-glycerophosphate, 1 mM sodium pyrophosphate, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 50 nM okadaic acid supplemented with a mixture of protease inhibitors). The lysates were cleared by centrifugation for 10 min at 14,000 ϫ g at 4°C. For immunoprecipitation, 0.5-1 mg of a given lysate was mixed overnight via gentle agitation with 1-2 g of specific or nonspecific antibodies coupled to protein A-or G-Sepharose beads. After extensive washing with lysis buffer, beads were resuspended in SDS sample buffer, supplemented with 5% ␤-mercaptoethanol, boiled for 5 min, and subjected to Western blot analysis using standard Western blotting protocols.
Measurement of Mitochondrial Membrane Potential and Annexin V Staining-To measure the mitochondrial membrane potential (⌬⌿m) we used the MitoProbe TM 1,1Ј,3,3,3Ј,3Ј-hexamethylindodicarbocyanine iodide (DiIC 1 ) (5) assay kit (catalog number M34151, Molecular Probes) as instructed by the manufacturer. Briefly cells (including floating cells) grown in 12-well plates were collected following mild trypsinization. Trypsinized cells were washed once with PBS, and the cells were resuspended in 500 l of PBS. Resuspended cells were labeled with 50 nM DiIC 1 (5) (excitation/emission, 638/658 nm) at 37°C in the dark for 20 -30 min. Labeled cells were washed once in PBS, and they were analyzed by fluorescence-activated cell sorting using a CyAn (Dako Cytomation) high performance flow cytometer.
To measure apoptosis we used the Annexin V fluorescein staining kit (catalog number A13199, Molecular Probes) as instructed by the manufacturer. Cells treated as described in the previous paragraph were collected following mild trypsinization. Trypsinized cells were washed once with PBS, and they were resuspended in 100 l of annexin binding buffer mixed with 5 l of Annexin V fluorescein conjugate (excitation/ emission, 494/518 nm). Resuspended cells were incubated at room temperature in the dark for 15 min. Labeled cells were analyzed by fluorescence-activated cell sorting. All flow cytometry data were analyzed using Summit Version 3.1 software.
Cell Proliferation Assay-To measure changes in the number of live cells, we used the 3-(4,5-dimethylthiazol-2-yl)-2,5-dimethyltetrazolium bromide (MTT; catalog number M5655, Sigma) assay. HEK293 cells stably transfected with the wildtype IRS-1 or the S794A IRS-1 mutant were seeded at 10 4 cells/ well in 24-well tissue culture plates. Sixteen hours later, cells were cultured in glucose-or serum-free media for an additional 48 h period as indicated under "Results" and in the figure legends. Then cells were placed in complete medium containing MTT at a final concentration of 500 g/ml for an additional 2-h period. Absorbance was measured with a Bio-Rad Benchmark microplate reader at 490 nm.
Statistical Analysis-Results are expressed as means Ϯ S.E. Differences between two groups were assessed using unpaired two-tailed t tests. Western blot band densitometry was done with IGOR Pro software (Wavemetrics, Inc.).

Energy Depletion Induces Phosphorylation of IRS-1 at Ser-794 and Down-regulates IRS-1 Expression-Energy depletion is
defined as an increase in the AMP:ATP ratio, and it can be caused by a variety of factors such as glucose deprivation, hypoxia, inhibition of ATP synthesis in the mitochondria, oxidative stress, and 2-DG. Fig. 1A demonstrates that in L6 myoblasts glucose deprivation induced phosphorylation of the AMPK kinase on Thr-172 and inhibited the mTORC1 pathway. In addition, glucose deprivation stimulated phosphorylation of IRS-1 at Ser-794 with kinetics similar to the kinetics of phosphorylation of AMPK. Phosphorylation in response to glucose deprivation was not affected by insulin stimulation (supplemental Fig. S1) and was amino acid-independent (supplemental Fig. S1 and data not shown). Long term glucose deprivation also down-regulated the expression of endogenous IRS-1 whose levels decreased with a half-life of ϳ9 h (Fig. 1A), whereas amino acid depletion alone did not affect IRS-1 expression (data not shown). Therefore, in L6 myoblasts a glucose-dependent pathway regulates both the phosphorylation and expression of IRS-1.
To explore the mechanism of IRS-1 phosphorylation at Ser-794 and to determine its functional significance we needed to establish a system in which glucose deprivation would not induce the down-regulation of IRS-1. To this end, we engineered HEK293 cells stably expressing IRS-1 under the constitutively active cytomegalovirus promoter. HEK293 cells normally express IRS-1 at very low levels. Fig. 1B shows that glucose deprivation did not down-regulate the ectopically expressed IRS-1. This suggests that IRS-1 down-regulation following glucose deprivation (Fig. 1A) may be regulated transcriptionally possibly by a mechanism that involves suppression of mTORC1 signaling and mitochondria dysfunction as reported previously (15,21). HEK293 cells stably expressing IRS-1 allowed us to measure the stoichiometry of phosphorylation of IRS-1 over time in cells cultured in media with decreasing concentrations of glucose. The results show that, upon glucose deprivation, IRS-1 phosphorylation at Ser-794 took place as early as 10 min from the start of the starvation period and persisted for 9 h similarly to phosphorylation of AMPK at Thr-172 (Fig. 1B). Moreover phosphorylation of IRS-1 at Ser-794 occurred within physiological glucose concentrations (4 -6 mM glucose; supplemental Fig. S2) in a manner similar to phosphorylation of AMPK at Thr-172.
Hypoxia increases the AMP:ATP ratio and activates AMPK (22). Fig. 1C shows that hypoxia, in addition of stabilizing Hif-1␣, stimulated phosphorylation of IRS-1 at Ser-794 in HEK293 cells stably expressing IRS-1 with kinetics that were similar to the kinetics of AMPK phosphorylation at Thr-172. Similarly the hypoxia-mimicking agent, cobalt chloride (CoCl 2 ), also stimulated AMPK and IRS-1 phosphorylation at these sites ( Fig. 1D). Lastly short term insulin stimulation did not affect hypoxia-induced phosphorylation of IRS-1 at Ser-794 (supplemental Fig. S3). Mitochondrial oxidative phosphorylation produces the majority of cellular ATP. To directly address the role of mitochondrial dysfunction in the regulation of IRS-1 phosphorylation at Ser-794 we examined the phosphorylation of IRS-1 at this site in cells treated with oligomycin and carbonyl cyanide 3-chlorophenylhydrazone (CCCP). Oligomycin and CCCP inhibit ATP production via inhibition of the F 0 F 1 ATP synthase and via irreversible dissipation of the electrochemical gradient, respectively. Fig. 1E demonstrates that inhibition of ATP synthesis in the mitochondria by both oligomycin and CCCP activated AMPK and stimulated IRS-1 phosphorylation at Ser-794.
Hydrogen peroxide-induced oxidative stress acutely activates AMPK by increasing the cellular AMP:ATP ratio (8,9,11,23). In HEK293 cells and L6 myoblasts hydrogen peroxide stimulated AMPK and IRS-1 phosphorylation at Thr-172 and Ser-794, respectively, in a dose-dependent manner within 2 min from the start of the treatment (Fig. 1, F and G, respectively). Lastly increase of the AMP:ATP ratio in cells treated with 2-DG (3) potently induced phosphorylation of AMPK at Thr-172 and IRS-1 at Ser-794 (supplemental Fig. S4).
Overall the data presented in Fig. 1 establish that energy depletion caused by a variety of treatments induces phosphorylation of IRS-1 at Ser-794 and that long term exposure to the same treatments reduces the expression of endogenous IRS-1 possibly transcriptionally (Figs. 1, A and B, and 4C and Refs. 15 and 21). Of note, energy depletion caused by all tested treatments completely eliminated previously described nutrientsensitive mTORC1/S6K1-mediated phosphorylations of IRS-1 at inhibitory sites (data not shown and Refs. [12][13][14]. IRS-1 Phosphorylation at Ser-794 in Response to Glucose Deprivation, Hypoxia, and Oxidative Stress Depends on the AMPK Catalytic Subunit ␣-To determine whether IRS-1 phosphorylation at Ser-794 in response to energy depletion is AMPKdependent we either overexpressed or knocked down the catalytic subunit of AMPK, and we examined the effects of these treatments on IRS-1 phosphorylation at this site. Fig. 2A demonstrates that overexpression of the wild-type form of the ␣2 (or the ␣1, not shown) catalytic subunit of AMPK significantly enhanced IRS-1 phosphorylation at Ser-794 following glucose deprivation, whereas overexpression of the kinase-deficient mutant of AMPK did not. Next we used siRNA that specifically knocked down the ␣1 and ␣2 catalytic subunits of the AMPK in HEK293 cells stably transfected with IRS-1. HEK293 cells express both the ␣1 and ␣2 catalytic subunits of AMPK with the ␣1 subunit being more abundant (5,24). Fig. 2B shows that knocking down either subunit efficiently suppressed the phosphorylation of IRS-1 at Ser-794 in response to glucose deprivation. Simultaneous knockdown of both subunits, using half the amount of siRNA for each subunit (see "Experimental Procedures"), completely eliminated basal as well glucose deprivation-induced phosphorylation of IRS-1 at Ser-794.
Next we addressed the role of AMPK in the hypoxia-induced phosphorylation of IRS-1 at Ser-794. Fig. 2C shows that overexpression of the wild-type form of AMPK enhanced the hypoxia-induced phosphorylation of IRS-1 at Ser-794, whereas the use of compound C, a specific AMPK inhibitor (25), reversed the effect of AMPK. Knocking down either of the two AMPK catalytic subunits suppressed hypoxia-induced Ser-794 phosphorylation, although simultaneous knockdown of both subunits had a synergistic effect, suggesting that both isoforms phosphorylate IRS-1 at Ser-794 in response to hypoxic stress (Fig. 2D).
Data presented in Fig. 1E suggest that inhibition of ATP synthesis in the mitochondria by oligomycin or CCCP induced phosphorylation of IRS-1 at Ser-794. Fig. 2E shows that compound C suppressed oligomycin-and CCCP-induced IRS-1 phosphorylation at Ser-794. Similarly siRNA against AMPK suppressed oligomycin-and CCCP-induced phosphorylation of IRS-1 at Ser-794 (data not shown) suggesting an important role for AMPK in mediating these effects.
To dissect the signaling pathway that mediates the phosphorylation of IRS-1 at Ser-794 in response to oxidative stress we used an array of known kinase inhibitors. Supplemental Fig. S5 shows that compound C suppressed IRS-1 phosphorylation induced by hydrogen peroxide treatment in a dose-dependent manner, whereas JNK, ERK, and PI 3-kinase/Akt inhibitors did not interfere with this event. Consistent with these data, knocking down the ␣1 catalytic subunit of AMPK suppressed hydrogen peroxide-induced IRS-1 phosphorylation at Ser-794 (Fig.  2F). However, knocking down the ␣2 subunit did not, suggesting that whereas both the ␣1 and the ␣2 subunits contribute to the phosphorylation of IRS-1 at Ser-794 in response to glucose deprivation and hypoxia, it is the ␣1 subunit that is primarily responsible for the phosphorylation of IRS-1 at this site in response to oxidative stress. This is in agreement with previous reports showing that hydrogen peroxide activates preferentially the AMPK ␣1, as opposed to the ␣2, catalytic subunit in vitro (23). Indeed knocking down the ␣1 subunit lowered the stoichiometry of AMPK phosphorylation at Thr-172, whereas knocking down the ␣2 subunit did not (Fig. 2F). Lastly supplemental Fig. S6 shows that siRNA against the AMPK ␣1 catalytic subunit also suppressed 2-DG-induced IRS-1 phosphorylation at Ser-794.
Previous studies have shown that energy depletion induces the phosphorylation of TSC2 by AMPK via a mechanism that involves direct interaction between AMPK and TSC2 (3). To this end we performed co-immunoprecipitation experiments to address whether AMPK, which is required for the phosphorylation of IRS-1 at Ser-794 upon energy depletion, interacts with IRS-1 and whether such an interaction can be induced by energy depletion. For these experiments we used HEK293 cells stably expressing IRS-1 to avoid its down-regulation upon energy depletion (Fig. 1A). Fig. 2G demonstrates that both AMPK ␣ catalytic subunits, but not LKB1, bound IRS-1 inducibly in response to glucose deprivation in HEK293 cells stably expressing IRS-1.
Overall the data presented in Fig. 2 suggest that AMPK is the major kinase that phosphorylates IRS-1 at Ser-794 in response to energy depletion. The AMPK ␣1 catalytic subunit has a prominent role in phosphorylating IRS-1 at this site in response to oxidative stress, whereas both the ␣1 and the ␣2 catalytic subunits phosphorylate IRS-1 in response to glucose deprivation and hypoxia. Because the ␣2 subunit is expressed at lower levels in HEK293 cells than the ␣1 subunit (5, 24), we conclude that the ␣2 catalytic subunit is the one that primarily phosphorylates IRS-1 in glucose-or oxygen-deprived cells.

Energy Depletion and PI 3-Kinase/Akt Signaling
LKB1 Is Required for the Phosphorylation of IRS-1 at Ser-794 upon Glucose Deprivation, Hypoxia, and Inhibition of ATP Synthesis in the Mitochondria-The LKB1 kinase, in complex with the STe20-related adaptor (STRAD)-␣/␤ and MO25-␣/␤ proteins, has been identified as an upstream AMPK kinase (7). Overexpression of wild-type LKB1 did not stimulate Ser-794 phosphorylation (data not shown). However, overexpression of the kinase-deficient mutant of LKB1 weakly inhibited IRS-1 phosphorylation at this site in glucose-or oxygen-deprived cells (data not shown). This observation prompted us to further address the role of LKB1 in LKB1 Ϫ/Ϫ knock-out 3T3 fibroblasts reconstituted with the wild-type or the kinase-deficient form of LKB1. The results show that the kinase activity of LKB1 was required for AMPK activation and IRS-1 phosphorylation at Ser-794 following glucose deprivation (Fig. 3A), hypoxia (Fig.  3B), and inhibition of ATP synthesis in the mitochondria (Fig.  3C). On the contrary, hydrogen peroxide (Fig. 3D) and 2-DG (supplemental Fig. S7) continued to partially activate AMPK and to induce phosphorylation of IRS-1 at Ser-794 in LKB1 Ϫ/Ϫ fibroblasts. Knocking down LKB1 in HEK293 cells stably expressing IRS-1 confirmed that LKB1 is selectively required for the phosphorylation of IRS-1 at Ser-794 following glucose deprivation or inhibition of mitochondrial oxidative phosphorylation but not following exposure to hydrogen peroxide or 2-DG (Fig. 3E). These findings suggest that glucose deprivation, hypoxia, and the inhibition of ATP synthesis in the mitochondria may transmit via LKB1 a mitochondrially derived signal that activates AMPK. In addition, the findings suggest that hydrogen peroxide and 2-DG may also activate AMPK and may stimulate phosphorylation of IRS-1 at Ser-794 via a kinase other than LKB1. Further studies indeed showed that hydrogen peroxide-induced phosphorylation of IRS-1 at Ser-794 in LKB1 knock-out fibroblasts was sensitive to STO-609 (Fig. 3F), suggesting that phosphorylation depends on the activity of the CaMKK␤ as reported previously (8 -11). Overall the data presented in Fig. 3 support the conclusion that energy depletion due to glucose deprivation, hypoxia, and inhibition of ATP synthesis in the mitochondria stimulate phosphorylation of IRS-1 at Ser-794 in an LKB1-depedent manner, whereas hydrogen peroxide and 2-DG stimulate phosphorylation of IRS-1 at the same site via CaMKK␤ and to a lesser extent LKB1.

Energy Depletion and Oxidative Stress Inhibit IRS-1-associated PI 3-Kinase/Akt
Signaling-Next we sought to determine the effect of energy depletion on IRS-1-associated PI 3-kinase/ Akt signaling. Fig. 4A shows that long term glucose deprivation (16 h) stimulated phosphorylation of IRS-1 at Ser-794 and suppressed insulin-induced IRS-1-associated PI 3-kinase activity (data not shown) and Akt phosphorylation (Fig. 4A) in L6 myoblasts. However, the observed inhibitory effect of chronic glucose removal on Akt phosphorylation in these cells was largely due to down-regulation of endogenous IRS-1 (Figs. 1A and 4A).
To determine the effects of IRS-1 phosphorylation at Ser-794 in the absence of IRS-1 down-regulation, we performed the same experiments in HEK293 cells stably transfected with IRS-1. Prior to that, we examined whether IRS-1 is required for Akt activation by insulin in HEK293 cells. Supplemental Fig. S8 shows that knocking down endogenous IRS-1 interfered with Akt phosphorylation at Thr-308 in response to insulin, suggesting that insulin-induced activation of Akt in HEK293 cells depends on IRS-1. Next Fig. 4B shows that long term glucose deprivation induced robust phosphorylation of IRS-1 at Ser-794 and suppressed insulin-stimulated IRS-1-associated PI 3-kinase activity and Akt phosphorylation in the absence of IRS-1 down-regulation. The inverse correlation between IRS-1 phosphorylation at Ser-794 and Akt phosphorylation suggests an inhibitory effect of this phosphorylation on PI 3-kinase signaling.

Phosphorylation of IRS-1 at Ser-794 Inhibits PI 3-Kinase/Akt
Signaling-To directly address the role of Ser-794 phosphorylation on PI 3-kinase/Akt signaling we engineered HEK293 cells stably expressing equal levels of wild-type IRS-1 or the IRS-1 mutants S794D and S794A (Fig. 5A). We then cultured these cells in media containing serum and physiological concentration of glucose (5 mM), and we asked whether IRS-1 undergoes phosphorylation at Ser-794 and whether phosphorylation at FIGURE 2. IRS-1 phosphorylation at Ser-794 in response to glucose deprivation, hypoxia, and oxidative stress depends on the AMPK catalytic subunit ␣. A, HEK293 cells engineered to stably express IRS-1 were transiently transfected with the wild-type or the kinase-deficient form of the AMPK ␣2 catalytic subunit. Twenty-four hours later, cells were serum-starved in the presence or absence of glucose for an additional 8-h period as indicated in the figure. Whole cell lysates were analyzed by Western blotting. B, HEK293 cells engineered to stably express IRS-1 were transiently transfected with siRNA against the AMPK ␣1 or ␣2 catalytic subunits or both as described under "Experimental Procedures." Forty-eight hours later, cells were serum-starved in the presence or absence of glucose (5 mM) for an additional 8-h period as indicated in the figure. Whole cell lysates were analyzed by Western blotting. C, HEK293 cells engineered to stably express IRS-1 were transiently transfected with the wild-type form of the AMPK ␣2 catalytic subunit. Twenty-four hours later, cells were serum-starved and maintained either in a normoxic or a hypoxic environment for an additional 8-h period in the presence or absence of the AMPK inhibitor compound C as indicated in the figure. Whole cell lysates were analyzed by Western blotting. D, the same cells shown in C were transfected with siRNA against the AMPK ␣1 or ␣2 catalytic subunits or both. Forty-eight hours later, cells were serum-starved and maintained either in a normoxic or hypoxic environment for an additional 8-h period. Whole cell lysates were analyzed by Western blotting. E, HEK293 cells stably expressing IRS-1 were pretreated with the AMPK inhibitor compound C. Thirty minutes later, they were treated with oligomycin or CCCP for an additional 30-min period. Whole cell lysates were analyzed by Western blotting. F, HEK293 cells stably expressing IRS-1 were transfected with siRNA against the AMPK ␣1 or ␣2 catalytic subunits or both. Forty-eight hours later, cells were treated with hydrogen peroxide for an additional 15-min period. Whole cell lysates were analyzed by Western blotting. G, HEK293 cells stably expressing IRS-1 were transfected with the indicated cDNAs. Twenty-four hours later cells were cultured for an additional 8-h period in serum-free media in the presence or absence of glucose. LKB1 and AMPK ␣1 and ␣2 catalytic subunits were immunoprecipitated with the anti-Myc antibody. IRS-1 in the immunoprecipitates was detected by Western blotting. GFP, green fluorescent protein; IP, immunoprecipitate. this site regulates the binding of IRS-1 to p85␣, the regulatory subunit of PI 3-kinase. The experiment in Fig. 5A showed that exogenous IRS-1 was weakly phosphorylated at Ser-794 under these conditions and that blocking its phosphorylation by mutating Ser-794 to alanine enhanced its binding to p85␣. The same experiment showed that the binding of the S794D mutant to p85␣ was impaired. When this experiment was repeated using unstimulated or insulin-stimulated cells cultured for 16 h in serum-free media containing low glucose concentrations (5 mM), we observed that the insulin-stimulated association of p85␣ with the S794D IRS-1 mutant was also impaired. The S794A mutation on the other hand restored the ability of IRS-1 to interact with p85␣ (Fig. 5B). We conclude that phosphorylation of IRS-1 at Ser-794 inhibits the IRS-1-p85␣ interaction.
Next we sought to determine the role of IRS-1 phosphorylation at Ser-794 on insulin-induced phosphorylation of Akt. To address this question we transfected HEK293 cells with wild-type IRS-1 or the IRS-1 mutants. Transfected cells were cultured in media containing low (4 mM) or high (25 mM) glucose concentrations. Following serum starvation for 16 h, cells were stimulated with insulin. Unstimulated and insulin-stimulated cells were harvested 15 min later. Fig. 5C shows that overexpression of the S794D but not the S794A mutant of IRS-1 reduced insulin-induced Akt phosphorylation at Thr-308, mimicking the effect of energy depletion on PI 3-kinase/Akt signaling. Of note, overexpression of either mutant had no effect on ERK phosphorylation in response to insulin. Lastly Fig. 5D shows the cumulative data on the -fold induction of Akt phosphorylation upon insulin stimulation in serum-starved cells growing in physiological glucose concentrations. Overexpression of the S794D mutant of IRS-1 reduced by ϳ50% the -fold phosphorylation of Akt at Thr-308, whereas the S794A mutant restored the ability of insulin to activate Akt. Therefore, we conclude that phosphorylation of IRS-1 at Ser-794 is induced by energy depletion and inhibits the activation of the PI 3-kinase/Akt pathway by insulin.
To further support this conclusion we examined the phosphorylation of Akt substrates in HEK293 cells expressing wildtype IRS-1 and IRS-1 S794D and S794A mutants before and after stimulation with insulin. Supplemental Fig. S9A shows that the phosphorylation of several Akt substrates detected with the phospho-Akt substrate antibody (Cell Signaling Technology catalog number 9614, clone 110B7) was impaired in HEK293 cells expressing IRS-1 S794D but not in HEK293 cells expressing wild-type IRS-1 or IRS-1 S794A. Based on the size of one of these proteins we hypothesized that it may be mTOR (molecular mass, ϳ289 kDa; apparent molecular mass, ϳ230 -250 kDa). We therefore probed the same lysates with an antibody that specifically detects mTOR phosphorylated at Ser-2448, an Akt-dependent phosphorylation (26 -29), that exhibits the characteristic Akt phosphorylation motif RXRXX(S/T). The results (supplemental Fig. S9B) confirmed that mTOR phosphorylation at Ser-2448 was impaired in cells expressing the IRS-1 S794D mutant but not in HEK293 cells expressing wild-type IRS-1 or IRS-1 S794A.

Phosphorylation of IRS-1 at Ser-794 in HEK293 Cells Growing under Energetically Unfavorable Conditions Promotes
Apoptosis-Inhibition of Akt activation by external signals may result in apoptosis (30 -32). We therefore hypothesized that apoptosis induced by energy depletion may depend on the inhibition of Akt via IRS-1 phosphorylation at Ser-794. To address this question, we seeded an equal number of HEK293 cells stably expressing the wild-type or the S794A mutant of IRS-1, and we cultured them in media containing serum and either low (1 mM) or high (25 mM) glucose concentrations. After 2 days we measured cell viability with the MTT assay (see "Experimental Procedures"). Fig. 6A shows that, in the presence of serum, glucose deprivation reduced the number of live HEK293 cells expressing wild-type IRS-1 to levels ϳ5.8-fold lower than the levels of live control cells growing in the presence of glucose (25 mM). Interestingly glucose deprivation reduced the number of live HEK293 cells expressing the S794A IRS-1 mutant to levels only 1.5-fold lower than the levels of live control cells growing in the presence of glucose (25 mM). However, the wild-type IRS-1-expressing and the IRS-1 S794A-expressing cells were equally sensitive to serum withdrawal (data not shown). Therefore, phosphorylation of IRS-1 at Ser-794 selectively regulates survival following glucose, but not serum, depletion.
To determine whether the decrease in the number of live cells in wild-type IRS-1-expressing HEK293 cells undergoing phosphorylation of IRS-1 at Ser-794 in response to energy depletion is due to apoptosis, we stained the cells with Annexin V, which specifically labels apoptotic cells. Supplemental Fig.  S10 shows that HEK293 cells stably expressing wild-type IRS-1 and growing in media containing low concentrations of glucose (1 mM) exhibit a statistically significant increase in Annexin V staining compared with HEK293 cells stably expressing the S794A IRS-1 mutant. Consistently Western blot analysis of lysates of energy-depleted cells revealed that energy depletion induced caspase-3 cleavage only in HEK293 cells expressing wild-type IRS-1 but not the S794A IRS-1 mutant (Fig. 6B). Therefore, the IRS-1 S794A mutant protects HEK293 cells from apoptosis induced by glucose deprivation.
Mitochondrial ⌬⌿m is often used as an indicator of cellular viability, and its disruption has been implicated in the initiation of the intrinsic pathway of apoptosis that culminates with the cleavage of caspase-3 (33). To measure ⌬⌿m we used the membrane potential-sensitive cyanine dye MitoProbe DiIC 1 (5). DiIC 1 (5) localizes in mitochondria in a ⌬⌿m-dependent manner. The intensity of fluorescence reflects the integrity of mitochondrial function in living cells. Fig. 6C shows that HEK293 cells ectopically expressing the S794A mutant of IRS-1 and grown in the presence of serum and a low (1 mM) concentration FIGURE 3. LKB1 is selectively required for IRS-1 phosphorylation at Ser-794 upon glucose deprivation, hypoxia, and inhibition of ATP synthesis in the mitochondria. LKB1 knock-out fibroblast reconstituted with the wild-type or the kinase-deficient (KD) form of LKB1 were cultured in serum-free media in the presence or absence of glucose for 8 h (A) or in serum-free, glucose-containing media in a normoxic or hypoxic environment for 16 h (B). C and D, the cells as in A and B were cultured in serum-free media, and they were treated either with CCCP for 30 min (C) or hydrogen peroxide for 15 min (D). Whole cell lysates were analyzed by Western blotting. E, HEK293 cells stably transfected with IRS-1 were treated with LKB1 siRNA. Forty-eight hours later, cells were treated as indicated in the figure. Whole cell lysates were analyzed by Western blotting. F, LKB1 knock-out fibroblast were pretreated with the CaMKK␤ inhibitor STO-609 for 1 h. Subsequently they were treated with hydrogen peroxide for an additional 15-min period. Whole cell lysates were analyzed by Western blotting. exp., exposure.
of glucose exhibited intact mitochondrial membrane potential even after 3 days of treatment. On the contrary the mitochondrial membrane potential of HEK293 cells expressing the wildtype IRS-1 was gradually reduced under similar conditions. Specifically Fig. 6D shows that HEK293 cells expressing the wildtype IRS-1 and grown for 2 days in the presence of serum and low (1 mM) glucose concentrations exhibited a statistically significant drop both in the mean (1.6 times) and peak (3.1 times) fluorescence intensities compared with HEK293 cells expressing the S794A mutant of IRS-1.

DISCUSSION
Data presented in this report show that energy depletion activated AMPK, which bound to and phosphorylated IRS-1 at Ser-794. The phosphorylation of IRS-1 at Ser-794 provides an entry into a novel and previously unsuspected mechanism by which glucose deprivation and hypoxia mediate their physiological effects via regulation of IRS-1-associated PI 3-kinase/Akt signaling. Indeed there has been no evidence to date that glucose availability, oxygen tension, and oxidative stress regulate the phosphorylation of IRS-1 at any site. Although the phosphorylation of IRS-1 at Ser-794 has been reported previously following 5-aminoimidazole-4-carboxamide riboside (17) or dexamethasone (18) treatment its biological significance has been a matter of controversy because it has been correlated both with inhibition (18,19) and activation (17) of PI 3-kinase/ Akt signaling by insulin. The studies reported here settled this controversy. By using L6 myoblasts and HEK293 cells engineered to stably express IRS-1 as model systems we demonstrated that energetically unfavorable conditions stimulate IRS-1 phosphorylation at Ser-794. In addition, we showed that the same conditions suppress IRS-1-associated PI 3-kinase activity and Akt phosphorylation in response to insulin (Fig. 4). . Energy depletion and oxidative stress inhibit IRS-1-associated PI 3-kinase/Akt signaling. A, L6 myoblasts cultured in media containing different concentrations of glucose were serum-starved. Sixteen hours later, the cells were stimulated with insulin for 15 min. Whole cell lysates were analyzed by Western blotting. B, HEK293 cells stably transfected with IRS-1 were maintained in serum-free media in the absence or presence of glucose. Twenty-four hours later, they were stimulated with insulin for an additional 15-min period. IRS-1 was immunoprecipitated, and the PI 3-kinase regulatory subunit p85␣ bound to IRS-1 was measured by probing the immunoprecipitates with an antibody specific for this subunit. The IRS-1-associated PI 3-kinase activity was measured by TLC (right panel). Left panel, Western blot of whole cell lysates was probed with the indicated antibodies. C, L6 myoblasts were cultured in serum-containing media in a normoxic or hypoxic environment for the indicated periods of time. Western blots of whole cell lysates were probed with the indicated antibodies. D, HEK293 cells stably expressing IRS-1 were maintained in the presence of serum in a normoxic or hypoxic environment. Forty-eight hours later, whole cell lysates were analyzed by Western blotting as indicated. IRS-1 was immunoprecipitated, and the amount of the associated p85␣ was determined by Western blotting (right panel). E, L6 myoblasts were serum-deprived for 6 h and then pretreated with hydrogen peroxide. Forty-five minutes later, cells were stimulated with insulin for an additional 15-min period. Endogenous IRS-1 was immunoprecipitated, and the associated PI 3-kinase activity was measured by TLC (right panel). Western blot of whole cell lysates was probed with the indicated antibodies (left panel). IP, immunoprecipitate; PI(3)P, phosphatidylinositol 3-phosphate. By using IRS-1 mutants at Ser-794, we demonstrated that the S794D mutant inhibits insulin-stimulated Akt phosphorylation, whereas the IRS-1 S794A mutant does not (Fig. 5). Moreover expression of IRS-1 mutant S794A in HEK293 cells prevented the loss of mitochondrial membrane potential and suppressed apoptosis following energy depletion (Fig. 6). Based on these findings we conclude that IRS-1 phosphorylation at Ser-794 upon energy depletion inhibits PI 3-kinase/Akt signaling and induces apoptosis (Fig. 7).
Previous studies have suggested that Akt requires glucose to exert its antiapoptotic function via regulation of the mitochondrial hexokinase (30 -32), the enzyme that controls the first step in glycolysis. The data presented here suggest that another reason for the glucose dependence of Akt survival signals may be the glucose-mediated suppression of IRS-1 phosphorylation at Ser-794, which is required for the activation of Akt by hormonal and growth factor signals.
Another interesting finding of this study is that phosphorylation of IRS-1 at Ser-794 by AMPK depended on the ␣1 or the ␣2 isoforms of the AMPK catalytic subunit in a stimulus-specific manner. The ␣2 subunit, which is the main subunit expressed in metabolically relevant tissues, was preferentially used to stimulate phosphorylation of IRS-1 at Ser-794 following glucose deprivation and hypoxia, whereas the ␣1 subunit exclusively mediated the effect of oxidative stress. The signal specificity of AMPK isoform utilization appears to reflect the speci-ficity by which various energy depletion signals regulate the upstream kinases involved in AMPK activation. Thus, in accordance with earlier studies showing that the LKB1/AMPK  ␣2 axis plays a prominent role in the regulation of metabolic homeostatic mechanisms (5, 34 -42), our study showed that the phosphorylation of IRS-1 at Ser-794 in glucose-or oxygendeprived cells depends on the LKB1/AMPK ␣2 axis. On the other hand, oxidative stress and 2-DG, which induced IRS-1 phosphorylation at Ser-794 via an ␣1 catalytic subunit-dependent manner, depend on a kinase other than LKB1. Because the phosphorylation of IRS-1 at Ser-794 in LKB1 Ϫ/Ϫ fibroblasts treated with hydrogen peroxide was inhibited by STO-609, we conclude that the upstream kinase activated by these treatments is CaMKK␤ (8,9,11).
Oxidative stress, induced by reactive oxygen species (ROS), is causally involved in mitochondrial dysfunction and aging as well as insulin resistance (43). ROS-induced insulin resistance has been attributed to the inhibitory effects of ROS on the activation of Akt by insulin (43). However, the molecular mechanism for this effect remains unknown. Our data show that phosphorylation of IRS-1 at Ser-794 was sensitive to oxidative stress and provide a molecular basis for the previously described inhibitory effect of oxidative stress on Akt activity in vitro and in vivo (43). Moreover our data show that this phosphorylation was preferentially mediated by the AMPK ␣1 catalytic subunit, suggesting a possible role for this isoform of AMPK in the pathogenesis of oxidative stress-induced insulin resistance.
ROS can be induced by numerous physiological and stress stimuli, including those produced by growth factors, cytokines, glucocorticoids, and phorbol esters. In the context of these stimuli, ROS contribute to signal transduction by functioning as second messenger molecules. ROS produced by any of these stimuli should be expected to promote the phosphorylation of IRS-1 at Ser-794. Indeed insulin, platelet-derived growth factor, and other growth factors generated signals that weakly phosphorylated IRS-1 at Ser-794 in different cell types like L6 myoblasts and mouse embryonic fibroblasts (Fig. 4E). 3 Current studies aim to define the kinase(s) responsible for the phosphorylation of IRS-1 at Ser-794 in response to these signals. Our preliminary data suggest that another kinase distinct from the AMPK family may be involved in IRS-1 phosphorylation at Ser-794 upon phorbol ester stimulation. 3 Along the same lines, dexamethasone induces the expression of salt-induced kinase 2 in 3T3-L1 adipocytes, promoting the phosphorylation of IRS-1 at Ser-794 (18).
The preceding studies focused on various immortalized but not transformed cell lines to show that energy depletion induced by glucose or oxygen deprivation and other insults leads to IRS-1 phosphorylation at Ser-794, which in turn inhibits Akt and cell survival. Cancer cells defy the homeostatic limits of normal cells and grow even in the presence of insufficient nutrient and oxygen supplies because they are specifically selected to survive these insults. It is therefore logical to expect that cancer cells may accumulate mutations that either prevent phosphorylation of IRS-1 at Ser-794 in response to such insults or suppress the downstream effects of this phosphorylation event once it takes place. In agreement with this prediction, loss of functional LKB1 has been observed in various neoplasms including lung and breast cancer (7,44). Moreover germ line mutations of LKB1 have been linked to the development of a dominantly inherited cancer predisposition syndrome, the Peutz-Jeghers syndrome (7,44). Finally consistent with our data previous studies have shown that long term pharmacological activation of AMPK suppresses proliferation of cancer cells by suppressing Akt activity (45)(46)(47). Deciphering the molecular circuitry that couples cellular energy status with cell growth and proliferation may provide new insights into the biology of cancer cells and into strategies for the translation of these insights into novel therapeutics.