Akt activates the mammalian target of rapamycin by regulating cellular ATP level and AMPK activity.

The serine/threonine kinase Akt is an upstream positive regulator of the mammalian target of rapamycin (mTOR). However, the mechanism by which Akt activates mTOR is not fully understood. The known pathway by which Akt activates mTOR is via direct phosphorylation and inhibition of tuberous sclerosis complex 2 (TSC2), which is a negative regulator of mTOR. Here we establish an additional pathway by which Akt inhibits TSC2 and activates mTOR. We provide for the first time genetic evidence that Akt regulates intracellular ATP level and demonstrate that Akt is a negative regulator of the AMP-activated protein kinase (AMPK), which is an activator of TSC2. We show that in Akt1/Akt2 DKO cells AMP/ATP ratio is markedly elevated with concomitant increase in AMPK activity, whereas in cells expressing activated Akt there is a dramatic decrease in AMP/ATP ratio and a decline in AMPK activity. Currently, the Akt-mediated phosphorylation of TSC2 and the inhibition of AMPK-mediated phosphorylation of TSC2 are viewed as two separate pathways, which activate mTOR. Our results demonstrate that Akt lies upstream of these two pathways and induces full inhibition of TSC2 and activation of mTOR both through direct phosphorylation and by inhibition of AMPK-mediated phosphorylation of TSC2. We propose that the activation of mTOR by Akt-mediated cellular energy and inhibition of AMPK is the predominant pathway by which Akt activates mTOR in vivo.

The serine/threonine protein kinase Akt, also known as protein kinase B, a downstream effector of phosphoinositide-3-OH kinase, has emerged as a critical mediator of the mammalian target of rapamycin (mTOR) 2 activity. Mammalian cells express three separate Akt proteins (Akt1-3), which share Ͼ80% amino acid sequence identity and are encoded by different genes. The rate-limiting step in Akt activation is the binding of phosphatidylinositol 3,4,5-trisphosphate to the pleckstrin homology domain of Akt and the subsequent translocation of Akt to the plasma membrane. Akt is then phosphorylated by 3-phosphoinositide-dependent kinase-1 and by another as yet unknown phosphoinositide-3-OH kinase-dependent kinase. Both phosphorylation events are required for full activation of Akt (for reviews see Refs. [1][2][3]. Biochemical and genetic data show that Akt is a positive regulator of mTOR that mediates the activation of mTOR by growth factors (reviewed in Ref. 4). mTOR controls mRNA translation by phosphorylating and activating S6 kinase 1 (S6K1) and by phosphorylating and inactivating the eukaryotic initiation factor 4E-binding proteins (4E-BPs), which repress mRNA translation. Thus, the phosphorylation status of S6K1 and one of the 4E-BPs members, 4E-BP1, is often used as readout for mTOR activity in vivo. mTOR is activated by the small GTPase Rheb, which is inhibited by its GAP protein TSC2 that heterodimerizes with tuberous sclerosis complex 1 (TSC1) (5)(6)(7). Genetic studies and biochemical analyses in mammalian cells (8 -12) and Drosophila (11,13), show that TSC2 is an upstream negative regulator of mTOR. Akt inactivates TSC2 by phosphorylating it on four residues, thereby activating mTOR (9,10,13). However, it is still not clear whether the phosphorylation of TSC2 by Akt is critical or sufficient for the activation of mTOR by Akt (14,15). mTOR activity also appears to be dependent on intracellular ATP level (16). ATP depletion activates AMP-activated protein kinase (AMPK), which in turn phosphorylates and activates TSC2 leading to the inhibition of mTOR activity (17). Currently, the activation of mTOR via Aktmediated phosphorylation of TSC2 and via inhibition of AMPK are viewed as two separate pathways leading to the activation of mTOR (4).
Here we provided evidence that Akt lies upstream of both pathways.
We provide genetic evidence to demonstrate that Akt modulates energy homeostasis by maintaining the level of ATP in cells so as to inhibit AMPK activity. Cells that are deficient for both Akt1 and Akt2 have reduced ATP and elevated AMPK activity and display impaired mTOR activity without a significant effect on Akt-mediated TSC2 phosphorylation. Expression of a dominant negative mutant of AMPK restores mTOR activity in Akt1/Akt2-deficient cells, implying that residual Akt3 activity is sufficient to phosphorylate TSC2 but is insufficient to inhibit AMPK and therefore to fully activate mTOR. We also found that in TSC2-deficient cells that have reduced Akt activity AMPK activity is consistently elevated. However, despite elevated AMPK activity, mTOR activity in TSC2-deficient cells is relatively refractive to ATP depletion, further supporting the conclusion that TSC2 mediates the inhibition of mTOR activity by AMPK. Expression of an Akt-phosphomimetic mutant of TSC2 in TSC2-deficient cells restores the sensitivity of mTOR to ATP depletion, demonstrating that TSC2 phosphorylation by Akt cannot prevent the activation of TSC2 by AMPK. However, co-expression of activated Akt in TSC2-deficient cells expressing the Akt-phosphomimetic TSC2 mutant reduces AMPK activity and renders mTOR activity resistant to ATP depletion. Taken together, these results demonstrate that Akt activates mTOR through both direct phosphorylation of TSC2 and by maintaining a high level of ATP with a concomitant decrease in the AMP/ATP ratio and inactivation of AMPK. We suggest that the activation of mTOR by Akt via inhibition of AMPK could be more relevant at the organismal level where cells do not always have access to nutrients for energy metabolism. This pathway by which Akt activates mTOR may also explain recent results showing that in Drosophila, TSC2 mutated in all Akt phosphorylation sites can still rescue the lethality and cell growth defect of TSC2 null mutant (14). Thus, Akt-phosphorylation mutants of TSC2 can still be activated by AMPK and be inhibited by Akt.
Adenine Nucleotides Analysis-Cultured cells were quickly harvested into phosphate-buffered saline and immediately centrifuged for 2 min at 1000 ϫ g (4°C). Pellets were resuspended in 150 l of perchloric acid, 4% v/v, and incubated on ice for 30 min. Within 1 h the lysates were adjusted to pH 6 -8 using a solution of 2 M KOH/0.3 M MOPS and incubated for 30 min on ice. Precipitated salt was separated from the liquid phase by centrifugation at 13,000 ϫ g for 10 min. Aliquots of samples were stored at Ϫ80°C. Adenine nucleotide measurements were conducted using HPLC (HPLC-Pro Star from Varian, Walnut Creek, CA) with a Spherisorb column (ODS II, 5 mm, 0.46 ϫ 25 cm, Z22.697-1, Sigma). The nucleotides analyzed, detected spectrophoto-metrically at 254 nm, eluted after ϳ17 min of isocratic elution at a flow rate of 1.0 ml min Ϫ1 . The order of eluted nucleotides was ATP, ADP, and AMP. Internal standards (7.5 M ATP, ADP, and AMP in ddH 2 O) were used to quantify the samples. The HPLC buffer contained 25 mM Na 4 P 2 O 7 -10H 2 O, 25 mM H 4 P 2 O 7 , adjusted to pH 5.75 with a saturated solution of Na 4 P 2 O 7 .

Akt Maintains the Intracellular Level of ATP and Regulates AMPK
Activity-Akt was shown to phosphorylate and inactivate TSC2 (9, 10, 13), thereby activating mTOR. mTOR activity, as measured by 4E-BP1 phosphorylation following serum stimulation (see Ref. 15 and Fig. 1A), is impaired in mouse embryo fibroblasts (MEFs) deficient for Akt1 and Akt2. However, this impairment did not correlate with a decrease in TSC2 phosphorylation (see Ref. 15 and Fig. 1B). These results suggest that the remaining Akt3 activity in Akt1/Akt2 double knock-out (DKO) cells ( Fig. 1A) is sufficient to substantially phosphorylate TSC2 and that TSC2 phosphorylation may not be sufficient for Akt to fully activate mTOR. Thus, these results prompted us to investigate whether there is an additional function of Akt that is required to fully activate mTOR and that is impaired in Akt1/Akt2 DKO cells.
mTOR activity is also dependent on intracellular ATP level and AMPK activity (16,17). Thus, we sought to determine whether Akt regulates intracellular ATP levels that could affect mTOR activity. We first determined the intracellular ATP level in Akt1/Akt2 DKO cells. Basal ATP level was lower in serum-deprived Akt1/Akt2 DKO cells compared with WT cells ( Fig. 2A). Following insulin or serum stimulation, the ATP level was increased but it was still retained markedly reduced in DKO cells, 2-to 3-fold lower than WT. Likewise, in cells expressing activated myristoylated Akt (Rat1a-mAkt), intracellular ATP was markedly higher (about 3-fold) than that measured in control cells (Fig. 2B). These results show that Akt mediates the insulin-and serum-dependent increase in intracellular ATP level.
AMPK is the sensor of ATP level in cells (27,28). AMPK activation inhibits mTOR activity (17,29) via direct phosphorylation of TSC2 (17). AMPK activity is dependent on the cellular AMP/ATP ratio (27,28). We found that, in the presence of growth factors, this ratio was markedly higher (about 2-to 3-fold) in Akt1/Akt2 DKO cells compared with WT cells (Fig. 2C), consistent with the lower ATP level in Akt1/Akt2 DKO cells ( Fig. 2A). Indeed, AMPK activity, as measured by its phosphorylation at Thr172 (30) and by the phosphorylation of the AMPK target acetyl-CoA carboxylase (ACC) at Ser-79 (31), was significantly higher in Akt1/Akt2 DKO cells compared with WT cells in the presence of serum (Fig. 2D). Moreover, introduction of a conditionally active Akt into Akt1/Akt2 DKO cells and modulating its activity decreased AMP/ ATP ratio in Akt-dependent manner (supplemental Fig. S1).
In the absence of serum, the AMP/ATP ratio in WT and Akt1/Akt2 DKO cells was comparable (Fig. 2C). Although p-ACC was substantially increased in WT cells following serum deprivation, it did not reach the level observed in DKO cells (Fig. 2D). One possible explanation for this apparent discrepancy between AMPK activity and AMP/ATP ratio is that Akt has an additional impact on AMPK, which is less dependent on AMP/ATP ratio (see "Discussion"). Consistent with the results observed in the Akt1/Akt2 DKO cells, in Rat1a cells expressing activated Akt (Rat1a-mAkt) the AMP/ATP ratio was about 3-fold lower than in control cells (Fig. 2E). When control Rat1a cells were deprived of serum, the AMP/ATP ratio markedly increased, concomitant with an increase in AMPK activity (Fig. 2, E and F). Although the AMP/ATP ratio also increased in serum-deprived Rat1a-mAkt cells, this ratio was comparable to the ratio in control cells in the presence of serum, and thus AMPK activity was not markedly increased in these cells (Fig. 2, E and F). The increase in the AMP/ATP ratio and AMPK activity in control cells correlated with a decrease in mTOR activity, as determined by S6K1 and 4E-BP1 phosphorylation, and by 4E-BP1-mobility shift (Fig.  2F). Thus, cells that maintained an AMP/ATP ratio below a certain threshold level also had high mTOR activity (Fig. 2, E and F). Taken together these results provide genetic evidence and demonstrate that Akt is a regulator of energy metabolism, which is required to maintain low AMPK activity in the cells.
ATP Depletion and Activation of AMPK Attenuate the Ability of Akt to Activate mTOR-To determine whether the Akt-mediated increase in the intracellular level of ATP and the decrease in the AMP/ATP ratio are required for Akt to activate mTOR, we first used inhibitors of glycolysis and oxidative phosphorylation to deplete ATP in cells expressing activated Akt. In Rat1a-mAkt cells, mTOR is constitutively active even in the absence of growth factors, as determined by 4E-BP1 phosphorylation and mobility shift (18) (Figs. 2F and 3A). However, ATP depletion by inhibition of glycolysis (using the glucose analogue 5-thioglucose, 5-TG) or the inhibition of oxidative phosphorylation (using rotenone) inhibited this Akt-mediated 4E-BP1 phosphorylation, as determined by p-4E-BP1 and by mobility shift, with no significant effect on either Akt activity or TSC2 phosphorylation by Akt (Fig. 3A). However, AMPK activity was elevated as measured by ACC phosphorylation (Fig. 3A,  right panel). Addition of 5-TG impaired the 4E-BP1 phosphorylation, induced by insulin in Rat1a cells, more strongly than the 4E-BP1 phosphorylation mediated by activated Akt in Rat1a-mAkt cells in the absence of insulin (Fig. 3, A and B). Similarly, 5-TG had a more profound effect in Akt1/Akt2 DKO cells than in WT cells (supplemental Fig. S2). Higher concentrations of 5-TG were required to impair mTOR activity in cells expressing activated Akt, which is also correlated with the more dramatic effect of 5-TG on AMPK activity in control cells (Fig. 3B, right  panel). This is correlated with the more substantial decline in ATP level in control cells in comparison with its decline in activated Akt expressing cells following addition of 5-TG (Fig. 3C). This could be due to increased glucose uptake and glycolysis in cells expressing activated Akt (see "Discussion"). We note that Akt-mediated mTOR activity also could be inhibited by adding a high level of 2-deoxyglucose (2-DOG, 100 mM) to cells expressing activated Akt (data not shown). However, this concentration of 2-DOG can also induce osmotic stress; moreover, 5-TG is probably a more effective inhibitor because, unlike 2-DOG, it is a competitive inhibitor that cannot be phosphorylated. Therefore, we used 5-TG for our experiments with cells expressing activated Akt to show that, even in these cells that have higher ATP level and lower AMP/ATP ratio, it is possible to decrease mTOR activity if ATP is depleted. Thus, these results suggest that the ability of Akt to mediate mTOR activity is dependent on its ability to increase the intracellular ATP level, which subsequently down-regulates AMPK.
To further assess the possibility that AMPK is a downstream effector of Akt leading to mTOR activation, we first exposed serum-deprived Rat1a-mAkt cells to AICAR, which activates AMPK and impairs insulin-mediated S6K1 phosphorylation (29). As shown in Fig. 4A, exposure of insulin stimulated Rat1a cells, to increasing concentrations of AICAR increased AMPK activity, as measured by AMPK and ACC phosphorylation, with a concomitant decrease in 4E-BP1 and S6 phosphorylation. Similar results were obtained in Rat1a-mAkt cells, in which mTOR was constitutively activated, although higher concentrations of AICAR were required (Fig. 4B). Thus, AICAR impairs the constitutive activation of mTOR in Rat1a-mAkt cells. We then examined whether an activated form of AMPK can alleviate the ability of Akt to activate mTOR as measured by 4E-BP1 phosphorylation.
For this purpose, HA-tagged 4E-BP1 was transiently co-transfected along with increasing amounts of an activated form of AMPK (CA-AMPK) into HEK293 cells stably expressing mAkt (18). As we previously showed, in contrast to control HEK293 cells, 4E-BP1 in mAkt-expressing cells was constitutively phosphorylated even in the absence of insulin stimulation (18) (Fig. 4C, lane 1). However, 4E-BP1 phosphorylation was impaired following expression of CA-AMPK (Fig. 4C, lanes 2-4). These results indicate that the Akt-mediated increase in the ATP level and the decrease in AMP/ ATP ratio are required for Akt to fully activate mTOR.
Dominant Negative AMPK Restores mTOR Activity in Akt1/Akt2 DKO Cells-If the ability of Akt to activate mTOR is dependent on its ability to increase ATP level and to inhibit AMPK kinase activity, then it is expected that the inhibition of AMPK activity in Akt1/Akt2 DKO cells would restore the impaired mTOR activity in these cells.
To explore this possibility we utilized WT and Akt1/Akt2 DKO MEFs immortalized by the expression of a dominant-negative form of p53 using retroviral infection (see "Experimental Procedures"). As in the primary cells, mTOR activity was impaired in immortalized Akt1/Akt2 DKO cells (Fig. 5). The immortalized WT and DKO cells were infected with retrovirus expressing dominant negative (DN) FIGURE 2. Akt regulates intracellular ATP level and AMPK activity. A, ATP level is reduced in Akt1/Akt2 DKO cells. Primary WT and Akt1/Akt2 DKO MEFs were plated in 10% FBS and deprived of serum for 24 h. Cells were stimulated with 10% FBS, 20% FBS, or 1 g ml Ϫ1 insulin for 60 min, and intracellular ATP, ADP, and AMP concentrations were analyzed by HPLC as described under "Experimental Procedures." Results represent the average of three independent experiments. B, ATP level is elevated in cells expressing activated Akt. Proliferating or serum-deprived control Rat1a and Rat1a-mAkt cells were analyzed for intracellular ATP level. Results represent the average of three independent experiments. C, AMP/ATP ratio is elevated in Akt1/Akt2 DKO cells. Primary WT and Akt1/Akt2 DKO MEFs were subjected to analysis as described in A, and the AMP/ATP ratio was determined in cells grown in 10% FBS (ϩserum), or 0.1% FBS (Ϫserum) for 24 h, and following serum and insulin stimulation. D, AMPK activity is elevated in Akt1/Akt2 DKO cells. Cell lysates from primary WT and Akt1/Akt2 DKO MEFs grown in 10% FBS (ϩserum) or 0.1% FBS (Ϫserum) for 24 h were subjected to immunoblotting using anti-pan Akt, anti-p-Akt-S473, anti-p-ACC-S79, anti-p-AMPK-T172, and anti-AMPK. E, AMP/ATP ratio is reduced in cells expressing activated Akt. Rat1a and Rat1a-mAkt cells grown in 10% FBS (ϩserum) or 0.1% FBS (Ϫserum) for 48 h were subjected to ATP and AMP analysis, and the AMP/ATP ratio was determined. F, AMPK activity is down-regulated in cells expressing activated Akt and is correlated with mTOR activity. Cell lysates from proliferating and serum-deprived Rat1a and Rat1a-mAkt cells were subjected to immunoblotting using anti-p-Akt-S473, anti-pan Akt, anti-p-TSC2-T1462, anti-TSC2, anti-pAMPK-T172, anti-TSC2, anti-AMPK, anti-p-S6K1-T389, and anti-p-4E-BB1-S65, anti-4E-BP1 and anti-␤-actin.
AMPK to generate polyclonal cell lines stably expressing DN-AMPK. DN-AMPK markedly decreased AMPK activity, as measured by ACC phosphorylation, and restored mTOR activity in the DKO cells as determined by the phosphorylation of S6K1, S6, and 4E-BP1 (Fig.   5). These results clearly demonstrate that mTOR activity in Akt1/ Akt2 DKO cells is impaired because of the inability to sufficiently increase the intracellular ATP level via insulin and growth factors and to sufficiently decrease AMPK activity in these cells. The Ability of Akt to Activate mTOR by Inhibiting AMPK Is Dependent on TSC2-TSC2 is phosphorylated and activated by AMPK, establishing one potential mechanism by which ATP and AMPK regulate mTOR activity (17). We thus examined TSC2Ϫ/Ϫ/p53Ϫ/Ϫ MEFs in comparison with TSC2ϩ/Ϫ/p53Ϫ/Ϫ MEFs and found that, consistent with previous results (17), ATP depletion had only a moderate effect on mTOR activity in TSC2Ϫ/Ϫ cells under these conditions, suggesting that the ATP level regulates mTOR activity mostly through TSC2 (Fig.  6A). In addition, we found that the AMP/ATP ratio and AMPK activity were markedly higher in TSC2Ϫ/Ϫ cells compared with TSC2ϩ/Ϫ cells (Fig. 6, B and C, lanes 1 and 3). Thus, mTOR is constitutively activated in TSC2-deficient cells despite the high AMP/ATP ratio and AMPK activity. The higher AMP/ATP ratio and the higher AMPK activity in TSC2 null cells could be due to reduced Akt activity (Fig. 6A,  lanes 1 and 2 and lanes 5 and 6) via a negative feedback loop mechanism (12,25). Indeed, when we stably expressed activated Akt in TSC2 KO cells the AMP/ATP ratio was restored with a concomitant decrease in AMPK activity, similar to that in TSCϩ/Ϫ cells (Fig. 6, B and C). These results further establish a role for Akt in regulating AMPK activity. The decrease in AMPK activity in TSC2 KO cells expressing activated Akt did not significantly increase mTOR activity (Fig. 6C), further supporting previous results that TSC2 is the major target for AMPK upstream of mTOR (17).
The results presented thus far strongly suggest that the phosphorylation of TSC2 by Akt is not sufficient to fully activate mTOR and that the inhibition of AMPK activity by Akt is also required for mTOR activation. To further assess this interpretation we utilized the Akt-phosphomimetic mutant of rat TSC2, TSC2 (S939D/S1086D/S1088E/T1422E) (TSC2 (2D,2E) ), in which four residues phosphorylated by Akt are substituted with acidic residues (10). HA-TSC2 (2D,2E) was cloned into a retroviral vector that we introduced into TSC2Ϫ/Ϫ cells to generate a polyclonal cell line (Fig. 7A). TSC2 (2D,2E) rendered mTOR activity in TSC2Ϫ/Ϫ cells more sensitive to ATP depletion (Fig. 7B, lanes 2, 5, and  8). Thus, the phosphorylation of TSC2 by Akt is not sufficient to render mTOR activity resistant to ATP depletion, indicating that the activation of TSC2 via its phosphorylation by AMPK is dominant over the inhibition of TSC2 via its phosphorylation by Akt. However, when mAkt was co-expressed with TSC2 (2D,2E) , it renders mTOR activity resistant to ATP depletion back to what was observed in the parental TSC2Ϫ/Ϫ cells (Fig. 7B, lanes 3, 6, and 9). These results demonstrate that Akt leads to the activation of mTOR through both direct phosphorylation and inactivation of TSC2 and through inhibition of AMPK activity. The  phosphorylation of TSC2 by Akt is not sufficient to fully activate mTOR, and Akt-mediated intracellular ATP level and the subsequent reduction in AMPK activity in conjunction with the direct phosphorylation of TSC2 by Akt is required to fully inhibit TSC2 and fully activate mTOR (Fig. 7C). It should be noted, however, that although mTOR activity in TSC2Ϫ/Ϫ cells is relatively resistant to ATP depletion it is still sensitive to ATP depletion, as determined by the phosphorylation of S6K1, 4E-BP1, and S6 (Figs. 6A and 7B). This observation suggests that the effect of ATP depletion on mTOR activity is not exclusively mediated by TSC2 and that ATP level and AMPK also can regulate mTOR activity through other unknown mechanisms.

DISCUSSION
Our present work provides genetic evidence and establishes that the serine/threonine kinase Akt is a key regulator of energy metabolism that inhibits AMPK. Akt-deficient cells have reduced ATP levels and elevated AMPK activity, whereas cells expressing activated Akt have markedly elevated ATP levels and reduced AMPK activity. The effect of Akt on the generation of ATP occurs via an increase in glycolysis and oxidative phosphorylation (32). Although the one or more exact mechanisms by which Akt affects these processes are not known, Akt can affect glycolysis through multiple mechanisms, including glucose transporters expression and translocation (33)(34)(35)(36), and the increased activity and expression of glycolytic enzymes (32,37,38). The ability of Akt to increase glycolysis also could ultimately affect oxidative phosphorylation in the mitochondria by increasing the availability of substrates for oxidative phosphorylation. The effect of Akt on ATP level causes a concomitant reduction in the AMP/ATP ratio and therefore reduces AMPK activity. AMPK is a heterotrimeric complex comprising a catalytic subunit (␣) and two regulatory subunits (␤ and ␥). AMP causes allosteric changes that activate AMPK and promote phosphorylation of Thr-172 in the activation loop of the ␣ subunit by AMPK kinase, recently identified as LKB1. This phosphorylation is required for full activation of AMPK (reviewed in Ref. 39). As shown by our present work, Akt not only decreases the phosphorylation of Thr-172 as well as the in vivo activity of AMPK as measured by the phosphorylation of ACC, but it is also required for the inhibition of AMPK activity by growth factors. We attributed this effect of Akt to its ability to regulate the intracellular AMP/ATP ratio. However, as a kinase, Akt also could potentially affect AMPK activity via phosphorylation of AMPK itself or its upstream regulator, LKB1. Indeed, Thr-366 in LKB1 lies within a consensus for the optimal phosphorylation motif for Akt. Although it  1 and 5), from insulin-stimulated cells (lanes 2 and 6), from insulin-stimulated/2-DOG-treated cells (lanes 3 and 7), and from insulin-stimulated/5-TG-treated cells (lanes 4 and 8) were subjected to immunoblotting using anti-p-Akt-S473, anti-pan-Akt, anti-p-TSC2-T1462, anti-pS6K1-T389, anti-S6K, anti-pS6-S235/236, anti-S6, anti-4E-BP1, and anti-␤-actin. B, TSC2Ϫ/Ϫ cells have an elevated AMP/ATP ratio, which is reduced by activated Akt. ATP and AMP levels were analyzed in TSC2Ϫ/Ϫ MEFs, a TSC2Ϫ/Ϫ polyclonal MEF cell line expressing mAkt, and TSC2ϩ/Ϫ MEFs as described in Fig. 2, and the AMP/ATP ratio was determined. The results represent the average of three independent experiments. C, TSC2Ϫ/Ϫ cells have elevated AMPK activity, which is reduced by activated Akt. Cell lysates from TSC2Ϫ/Ϫ MEFs, a TSC2Ϫ/Ϫ polyclonal MEF cell line expressing mAkt, and TSC2ϩ/Ϫ MEFs were subjected to immunoblotting using anti-TSC2, anti-pan-Akt, anti-p-Akt-S473, anti-p-AMPK-T172, anti-AMPK, anti-p-ACC-S79, anti-p-S6K1-T389, anti-p-4E-BP1-S65, anti-4E-BP1, and anti-␤-actin.
was reported that LKB1 is a poor substrate for Akt in vitro (40,41), it is still conceivable that Akt phosphorylates LKB1 in vivo under certain physiological conditions. AMPK impairs the induction of mTOR activity by growth factors (29) and directly phosphorylates and activates TSC2 thereby inhibiting mTOR activity (17). Here we show that in order for Akt to fully inhibit TSC2 and to activate mTOR it needs to directly phosphorylate TSC2 and to inhibit AMPK preventing it from activating TSC2. We showed that in cells deficient for Akt1 and Akt2, mTOR activity is impaired without a substantial effect on TSC2 phosphorylation by Akt. However, AMPK activity is elevated in these cells, suggesting that the residual Akt activity in Akt1/Akt2-deficient cells mediated by Akt3 is sufficient to phosphorylate TSC2 but insufficient to maintain normal ATP levels, thus leading to AMPK activation. Indeed, expression of DN-AMPK in Akt1/Akt2-deficient cells restores mTOR activity. Furthermore, expression of an Akt-phosphomimetic mutant of TSC2 in TSC2-deficient cells (in which mTOR activity is relatively refractive to ATP depletion) restores sensitivity of mTOR to ATP depletion. These data imply that TSC2 phosphorylation by Akt does not prevent the activation of TSC2 by AMPK. However, expression of activated Akt together with the Akt-phosphomimetic TSC2 mutant reverses the sensitivity of mTOR activity to ATP depletion to the same as observed in TSC2deficient cells. In addition, AMPK activity is elevated in TSC2-deficient cells, and the expression of activated Akt inhibits the elevated AMPK activity in these cells. Taken together these results clearly demonstrate that Akt, in addition to inhibiting TSC2 via direct phosphorylation, also inhibits TSC2 and activates mTOR through the inhibition of AMPK. This establishes an alternative mechanism for the activation of mTOR by growth factors and Akt. This alternative pathway by which Akt activates mTOR can explain, at least in part, why in Drosophila a TSC2, which is mutated in all Akt phosphorylation sites, can still rescue the lethality and cell growth defect of TSC2 null mutant (14). Thus, TSC2 mutants that are not directly phosphorylated by Akt can still be activated by AMPK and inhibited by Akt. This could be particularly of importance at the organismal level and in tumors where cells do not always have access to excess of nutrients for energy metabolism, raising the possibility that at the organism level the predominant effect of Akt on TSC2 is via the inhibition of AMPK. mRNA translation and ribosomal biogenesis, two processes that are mediated by mTOR, consume high levels of cellular energy. Thus the high consumption of ATP in TSC2-deficient cells together with the reduced Akt activity, due to a negative feed back loop (12, 25), could  1, 4, and 7), a polyclonal TSC2Ϫ/Ϫ MEF cell line expressing HA-TSC2 (2D,2E) (lanes 2, 5, and 8), and a polyclonal TSC2Ϫ/Ϫ MEF cell line expressing HA-TSC2 (2D,2E) and mAkt (lanes 3, 6, and 9) were subjected to immunoblotting using anti-p-Akt-S473, anti-p-S6K1-T389, anti-p-S6-S240/244, anti-p-4E-BP1-S65, anti-4E-BP1, and anti-␤-actin. C, a model showing that Akt inhibits TSC2 by direct by phosphorylation anti-S6K1 indirectly via inhibition of AMPK activity by Akt. Both of these processes are required for full activation of mTOR leading to the inhibition of two tumor-suppressors, TSC2 and LKB1.
contribute to the elevated AMP/ATP ratio and AMPK activity observed in these cells. However, because of TSC2 deficiency mTOR activity in these cells is resistant to the elevated AMPK activity.
Although mTOR activity in TSC2-deficient cells is relatively refractive to ATP depletion, it is still somewhat reduced in these cells in response to ATP depletion. This suggests that ATP level and AMPK activity can also affect mTOR activity in a TSC2-independent manner. One possibility is that ATP levels affects mTOR activity directly due to the relatively high K m of mTOR for ATP as was previously reported (16). Another possibility is that AMPK can directly phosphorylate and inactivate mTOR (Fig. 7C). It was recently shown that AMPK can phosphorylate Thr-2446 of mTOR (42), which resides in the putative negative regulatory domain of mTOR (43). Phosphorylation of Thr-2446 by AMPK inhibits the phosphorylation of Thr-2448 by Akt and thus can potentially inhibit mTOR activity (42). However, thus far it has not been demonstrated that Thr-2448 phosphorylation by Akt has any impact on mTOR activity (42), and we have not observed any reduction in Thr-2448 phosphorylation following ATP depletion. 3 Clearly more experiments are required to delineate the additional mechanism(s) by which ATP and/or AMPK affect mTOR activity.
Both TSC2 and LKB1 appear to act as tumor suppressors, and their deficiency leads to the development of benign tumors and hamartomas (44 -46). It is therefore possible that LKB1 exerts its tumor suppressor activity by activating AMPK and inhibiting TSC2 (47,48). Akt is frequently activated in human cancers mainly through the inactivation of the tumor suppressor phosphatase and tensin homolog deleted on chromosome ten (PTEN) whose deficiency also can lead to the development of benign tumors and hamartomas (49). Our results strongly suggest that the phosphorylation of TSC2 by Akt is not sufficient to overcome the activity of LKB1 as a tumor suppressor. However, the ability of Akt to negate AMPK activity could be sufficient to overcome the tumor suppressor activity of LKB1 (Fig. 7C). Thus, phosphatase and tensin homolog deleted on chromosome ten (PTEN) deficiency should be capable of overcoming the tumor suppressor activities of both TSC2 and LKB1.