ERK and Akt signaling pathways function through parallel mechanisms to promote mTORC1 signaling

Abstract

The mammalian target of rapamycin (mTOR) is a protein kinase that, when present in a complex referred to as mTOR complex 1 (mTORC1), acts as an important regulator of growth and metabolism. The activity of the complex is regulated through multiple upstream signaling pathways, including those involving Akt and the extracellular-regulated kinase (ERK). Previous studies have shown that, in part, Akt and ERK promote mTORC1 signaling through phosphorylation of a GTPase activator protein (GAP), referred to as tuberous sclerosis complex 2 (TSC2), that acts as an upstream inhibitor of mTORC1. In the present study we extend the earlier studies to show that activation of the Akt and ERK pathways acts in a synergistic manner to promote mTORC1 signaling. Moreover, we provide evidence that the Akt and ERK signaling pathways converge on TSC2, and that Akt phosphorylates residues on TSC2 distinct from those phosphorylated by ERK. The results also suggest that leucine-induced stimulation of mTORC1 signaling occurs through a mechanism distinct from TSC2 and the Akt and ERK signaling pathways. Overall, the results are consistent with a model in which Akt and ERK phosphorylate distinct sites on TSC2, leading to greater repression of its GAP activity, and consequently a magnified stimulation of mTORC1 signaling, when compared with either input alone. The results further suggest that leucine acts through a mechanism distinct from TSC2 to stimulate mTORC1 signaling.

the mammalian target of rapamycin (mTOR) is a serine-threonine protein kinase that is conserved across eukaryotic species. It was originally identified based on the growth-inhibiting properties of the macrolide immunosuppressant rapamycin in yeast (15), but subsequent studies have shown that rapamycin inhibits cell proliferation by suppressing both cell growth and G1- to the S-phase cell cycle progression (reviewed in Ref. 54). The kinase exists in two complexes referred to as mTOR complex (mTORC) 1 and 2 (27). The protein composition of the complexes is similar in that each contains mTOR and mammalian lethal with Sec13 (mLST)8 but differ in that mTORC1 additionally contains the regulatory-associated protein of mTOR (raptor), whereas mTORC2 contains the rapamycin-insensitive companion of mTOR (rictor). Recent studies (20, 43) have shown that rapamycin selectively inhibits mTORC1 and does not directly repress mTORC2, suggesting that the growth inhibitory properties of the drug are likely mediated through mTORC1 rather than mTORC2. Further support for this suggestion is provided by the observation that mice (48) or flies (33) with a chromosomal disruption in the gene encoding the ribosomal protein (rp) S6 kinase S6K1, a specific target of mTORC1 (36), are smaller than wild-type controls. Moreover, mice in which the S6K1 phosphorylation sites on rpS6 are changed to alanine to prevent phosphorylation exhibit a similar growth repression (40), indicating an important role for S6 phosphorylation in the growth stimulatory signal(s) originating from mTORC1.

Although mTORC1 signaling is critical in promoting growth and proliferation in unicellular and young multicellular organisms, its more important role in adults is in the control of metabolism. In this regard, mTORC1 upregulates the expression and/or activity of proteins such as the peroxisome proliferator-activated receptor-γ and the sterol regulatory element binding protein 1 that act to regulate lipid metabolism (12, 26), and HIF-1α that promotes glucose uptake and glycolysis (12). It is also an important regulator of mitochondrial biogenesis and metabolism (37). However, under certain circumstances, the growth regulatory function of mTORC1 is essential even in adults. As an example, rapamycin-induced inhibition of mTORC1 obviates the gain in muscle mass that occurs in a compensatory hypertrophy model of resistance exercise (4). Although the signaling pathway(s) involved in the activation of mTORC1 in response to resistance exercise is incompletely defined, activation of both the Akt (4) and extracellular-regulated kinase (ERK) signaling (7) pathways have been reported. Moreover, exogenous expression of either constitutively active Akt (caAkt) in skeletal muscle (35) or constitutively active mitogen-activated protein kinase kinase (MEK) in the heart (5, 52) leads to muscle hypertrophy. Both the Akt (10, 32, 38) and MEK/ERK (31) signaling pathways have been shown to promote mTORC1 signaling through phosphorylation of a repressor of mTORC1 activity, the tuberous sclerosis complex (TSC) consisting of TSC1 and TSC2. The TSC1/2 complex functions as a GTPase activating protein (GAP) toward a protein referred to as ras homolog enriched in brain (Rheb) (18). The Rheb·GTP binary complex activates mTORC1 through an incompletely defined mechanism, whereas Rheb·GDP has little, if any, stimulatory activity (25, 29). Akt and ERK phosphorylate distinct sites on TSC2 (31, 60), suggesting that the two pathways act independently to inhibit the GAP activity of TSC2. However, the relative contribution of the two pathways in regulating signaling through mTORC1 has not been examined. Moreover, the effect of simultaneous inputs through both pathways on mTORC1 signaling is as yet unexplored.

In the present study, we have employed a cell culture model that mimics skeletal muscle in regard to the regulation of mTORC1 by hormones and nutrients to compare and contrast the effect of the input signals through the Akt pathway compared with the ERK pathway. We find that maximal input signals through either pathway stimulates mTORC1 signaling to a similar extent, and that combined maximal input signals through both pathways leads to an even greater stimulation compared with either alone. Evidence for TSC2 being a point of convergence of the signaling inputs through the Akt and ERK pathways is provided by the observation that in cells lacking TSC2, mTORC1 signaling is insensitive to activation of either pathway. Overall, the results support a model in which Akt signaling promotes phosphorylation of multiple residues on TSC2, including S939 and T1462, whereas ERK signaling acts through a mechanism distinct from phosphorylation of those residues. In this model, phosphorylation of both the Akt- and ERK-regulated sites leads to additive effects on the inhibition of TSC2, and subsequently, leads to a greater stimulation of mTORC1 signaling compared with input signaling through either pathway alone.

MATERIALS AND METHODS

Materials and reagents.

Cell culture medium lacking leucine, histidine, and pyruvate was a custom formulation purchased from Atlanta Biologicals; histidine was added to the medium before use. Insulin (Novolin) was purchased from Novo Nordisk, leucine was purchased from Sigma-Aldrich, and 18:1 lysophosphatidic acid [LPA; 1-oleoyl-2-hydroxy-sn-glycero-3-phosphate (sodium salt)] was purchased from Avanti Polar Lipids. Anti-total and anti-phospho-ERK(T202/Y204), anti-total and anti-phospho-Akt(S473), anti-total and anti-phospho-TSC2(S939 and T1462), and anti-phospho-S6K1(T389) antibodies were purchased from Cell Signaling; anti-total S6K1 antibody was purchased from Bethyl Laboratories; and anti-GAPDH antibody was purchased from Santa Cruz. Anti-total and anti-phospho-PRAS40(T246) was purchased from Invitrogen. Akt1/2 kinase inhibitor (Akt1/2KI) and U0126 were purchased from Sigma-Aldrich and Promega, respectively. MEK1 cDNA (activated) in pUSEamp (S218D/S222D double mutant) and pUSEamp Empty Vector were purchased from Upstate Biotechnology, HA-Akt CA (myr-HA-Akt) in pCMV5 was purchased from Addgene (plasmid donated to Addgene by Dr. Mien-Chie Hung), and pCMV5 Empty Vector was graciously donated by Dr. David Russell (UT Southwestern). FuGENE HD Transfection Reagent was purchased from Roche.

Cell culture.

Rat2 fibroblasts, the experimental model for most of the studies reported herein, were purchased from American Type Culture Collection, and TSC2^−/− and p53^−/− (TSC2 KO) mouse embryo fibroblasts (MEF) were a gift from Dr. David J. Kwiatkowski (Harvard Medical School). All cell lines were maintained in DMEM lacking sodium pyruvate and containing high glucose (GIBCO/Invitrogen Life Sciences), 10% fetal bovine serum (FBS) (Atlas Biologicals), and 1% Pen Strep (GIBCO). On the day of the study, medium was replaced with one lacking leucine and pyruvate. The medium also contained FBS as noted in the figure legends. In studies involving inhibition of MEK and Akt, cells were incubated for 2.25 h in medium containing 0.5% FBS and lacking leucine. During the last 15 min, either U0126 or Akt1/2KI was added to the medium at final concentrations of 10 or 0.4 μM, respectively. The cells were then treated with 22 μM LPA, 10 nM insulin (Novolin), and/or 0.76 mM leucine for 2 h. For studies involving the overexpression of active forms of MEK1 and Akt, cells were incubated in medium containing 0.5% FBS and lacking leucine for 2 h before sample collection. TSC2 knockout and wild-type MEF cells were incubated in medium containing 0.5% FBS and lacking leucine for 2 h before the addition of 22 μM LPA and/or 10 nM insulin (Novolin) for an additional 2 h.

LPA treatments.

LPA was prepared as suggested by the manufacturer. Specifically, LPA powder was suspended 1:1 in water:ethanol, incubated at 37°C for 5 min, and then sonicated for 5 min using a water bath sonicator.

Transfections.

Cells were transfected using FuGENE HD Transfection Reagent according to instructions provided by the manufacturer. Briefly, cells at ∼80% confluency were incubated with transfection reagent and 2 μg total DNA for 6 h, after which the medium was removed and high-glucose DMEM containing 10% FBS and 1% PennStrep was added. Cells were then incubated for 16–18 h before the start of experimentation.

Western blot analysis.

For Western blot analysis, cells were scraped into 1× Laemmli buffer, and samples were resolved using Bio-Rad Criterion Tris·HCl 4–15% gels as described previously (23). Blots were probed with anti-phospho-S6K1(T389), anti-phospho-ERK(T202/Y204), anti-phospho-Akt(S473), anti-phospho-PRAS40(T246), anti-phospho-TSC2(S939), or anti-phospho-TSC2(T1462). After development, blots were stripped using buffer containing 62.5 mM Tris·HCl, 69.35 mM SDS, and 18.3 μM β-mercaptoethanol, pH 6.7, and then reprobed with either antibody recognizing the respective protein regardless of phosphorylation state or a monoclonal anti-GAPDH antibody. Values were normalized to GAPDH.

Statistics.

Data were analyzed by one-way ANOVA using the Prism 5 software program (GraphPad). If a significant difference was detected, data were analyzed further by unpaired t-test. P < 0.05 was considered statistically significant.

RESULTS

In the present study, Rat2 fibroblasts were used as an experimental model because, unlike C2C12 and L6 myotubes, but similar to skeletal muscle in vivo, mTORC1 signaling responds to physiological concentrations of nutrients such as leucine (24, 57). Moreover, when compared with muscle cell lines, Rat2 cells are relatively easy to transfect at high efficiency (24). Indeed, in the present study, the transfection efficiency was 89 ± 7% (n = 3), as assessed by the proportion of cells expressing green fluorescent protein (GFP) after transfection with a plasmid encoding the protein. In a recent study (57), we observed that LPA is a highly effective agent for activating the ERK signaling pathway in these cells. As illustrated in Fig. 1A, treatment with a concentration of LPA previously determined to be maximally effective in these cells (57), produced a sevenfold increase in ERK phosphorylation, whereas a supraphysiological concentration of insulin had little effect, as assessed by changes in ERK1/2(T202/Y204) phosphorylation. When LPA and insulin treatments were combined, ERK phosphorylation was not significantly different compared with LPA alone, indicating that the contribution of insulin was minimal (Fig. 1A). Additionally, while LPA treatment had only a minor effect on signaling through the Akt pathway, as assessed by Akt (S473) phosphorylation, insulin produced an ∼2.5-fold increase (Fig. 1B). To test whether the ERK and Akt signaling pathways function together to activate mTORC1, cells were treated with maximally effective concentrations of LPA or insulin individually or in combination, and S6K1 phosphorylation on T389, a site directly phosphorylated by mTORC1 (36), was assessed. Individually, LPA and insulin stimulated mTORC1 signaling by approximately twofold (Fig. 1C). However, when combined the treatments stimulated mTORC1 signaling approximately fourfold. Thus the ERK and Akt signaling pathways acted in an additive manner to promote mTORC1 signaling.

Fig. 1.

Effect of lysophosphatidic acid (LPA) and/or insulin on extracellular-regulated kinase (ERK)1/2, Akt, and S6K1 phosphorylation in Rat2 fibroblasts. Cells were incubated for 2 h in medium containing 0.5% FBS and lacking leucine, and then lysophosphatidic acid (LPA) and/or insulin were added at final concentrations of 22 μM and 10 nM, respectively. Two hours later, cells were harvested and homogenates were subjected to Western blot analysis for phosphorylation of ERK1/2 on T202/Y204 (A), Akt on S473 (B), and S6K1 on T389 (C), as described under materials and methods. The results represent the means ± SE of 6 (A and B) or 3 (C) experiments. Within each experiment 3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. P, phosphorylated protein; T, the respective total protein. *P < 0.05 vs. no additions; †P < 0.002 vs. LPA; ‡P < 0.005 vs. insulin.

To assess the combined and individual roles of the ERK and Akt signaling pathways in the stimulation of mTORC1 signaling, the MEK inhibitor U0126 and the Akt1/2 kinase inhibitor (KI) Akt1/2 KI were employed. As shown in Fig. 2, A and B, respectively, U0126, but not Akt1/2 KI, prevented LPA-induced ERK1/2 phosphorylation, whereas Akt1/2 KI, but not U0126, inhibited insulin-induced Akt phosphorylation. In confirmation of the results shown in Fig. 1B, combined treatment with LPA and insulin led to significantly greater phosphorylation of Akt on S473 compared with either agent alone (Fig. 2B). However, in cells treated with U0126, the combined effect was lost, and Akt(S473) phosphorylation in cells treated with U0126, LPA, and insulin was not different compared with cells treated with insulin alone. Thus activation of MEK is largely responsible for the combined effect of LPA and insulin on Akt(S473) phosphorylation.

Fig. 2.

Selectivity of U0126 and Akt1/2 KI in Rat2 fibroblasts. Cells were incubated for 2.25 h in medium containing 0.5% FBS and lacking leucine. During the last 15 min, either U0126 or Akt1/2KI was added to the medium to final concentrations of 10 or 0.4 μM, respectively. LPA and/or insulin were then added at final concentrations of 22 μM and 10 nM, respectively, and 2 h later, cells were harvested and homogenates were subjected to Western blot analysis for phosphorylation of ERK1/2 on T202/Y204 (A) or Akt on S473 (B). The results represent the means ± SE of 6 experiments. Within each experiment 3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. P, phosphorylated protein; T, the respective total protein. *P < 0.05 vs. no additions; †P < 0.0001 vs. LPA plus insulin; ‡P < 0.005 vs. insulin.

Having established the specificity of the inhibitors for the ERK and Akt signaling pathways, their respective effects on LPA- and insulin-induced stimulation of mTORC1 signaling was assessed. As shown in Fig. 3A, in the presence of U0126, mTORC1 signaling in cells treated with both LPA and insulin was the same as in cells treated with insulin alone. Similarly, in the presence of Akt1/2 KI, the combined effect of LPA and insulin was abolished, and mTORC1 signaling was statistically identical to that observed in cells treated with LPA alone (Fig. 3B). Combined, the two inhibitors prevented completely the LPA- and insulin-induced stimulation of mTORC1 signaling (Fig. 3C). Overall, the results shown in Figs. 1 and 3 demonstrate that the Akt and ERK signaling pathways are acting through parallel mechanisms to promote mTORC1 signaling. This conclusion is also supported by results from a recent report (9) showing that phosphorylation of ribosomal protein S6 on Ser240/244 was stimulated to a greater extent in human embryonic kidney (HEK)293 cells treated with a combination of insulin and phorbol 13-myristate 12-acetate, a potent activator of the ERK signaling pathway, compared with treatment with insulin alone.

Fig. 3.

LPA and insulin act through the mitogen-activated protein kinase kinase (MEK)/ERK and Akt signaling pathways to stimulate mammalian target of rapamycin complex 1 (mTORC1) signaling in Rat2 fibroblasts. Cells were incubated with U0126 (A), Akt KI (B), or U1026 and Akt KI (C), as described in Fig. 2, and homogenates were subjected to Western blot analysis for phosphorylation of S6K1 on T389. The results represent the means ± SE of 3 experiments. Within each experiment 3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. P, phosphorylated protein; T, the respective total protein. *P < 0.0005 vs. no additions; †P < 0.005 vs. LPA, insulin, or LPA plus insulin in the presence of U0126 and/or Akt KI; ‡P < 0.002 vs. LPA, insulin, or LPA plus insulin.

To further confirm that the ERK and Akt signaling pathways were acting through parallel mechanisms to stimulate mTORC1 signaling, the effect of exogenous expression of constitutively active MEK1 (S218D/S222D) and/or Akt (myr-HA-Akt) on S6K1 phosphorylation was assessed. The expression of caMEK1 led to a fourfold increase in ERK1/2 phosphorylation, whereas expression of caAkt had little effect (Fig. 4A). In contrast, caMEK had essentially no effect on phosphorylation of PRAS40, a direct target of Akt (14), whereas caAkt significantly increased PRAS40 phosphorylation (Fig. 4B). Importantly, in cells expressing both kinase variants, no combined effect on either ERK or PRAS40 was observed. In cells individually expressing caMEK or caAkt, mTORC1 signaling was increased ∼2.5-fold relative to cells transfected with a control plasmid (Fig. 4C). When caMEK1 and caAkt were expressed together, mTORC1 signaling was significantly greater than when either construct was expressed alone (Fig. 4C), a result consistent with the conclusion that LPA and insulin treatments acted through parallel mechanisms to activate mTORC1.

Fig. 4.

Effect of exogenous expression of constitutively active MEK1 and/or Akt1 on mTORC1 signaling in Rat2 fibroblasts. Cells were transfected with plasmids encoding constitutively active, hemagglutinin (HA)-tagged MEK1 (pUSE MEK1) and/or Akt1 (pCMV5 Akt), or control plasmids (pUSE or pCMV5, respectively) and were harvested ∼18 h later. Homogenates were analyzed for phosphorylation of ERK1/2 on T202/Y204 (A), PRAS40 on T246 (B), and S6K1 on T389 (C), as described in materials and methods. The results represent the means ± SE of 3 experiments. Within each experiment 2–3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. Top blot in each panel depicts a blot for the phosphorylated protein, whereas bottom blot depicts the respective total protein. *P < 0.05 vs. cells transfected with the respective control plasmid(s); †P < 0.001 vs. cells transfected with pUSE MEK1; ‡P < 0.005 vs. cells transfected with pCMV5 Akt.

Although ERK and Akt phosphorylate and thereby inactivate TSC2, both kinases can also act downstream of TSC2 to modulate mTORC1 activity. For example, through activation of p90RSK, ERK promotes phosphorylation of raptor on multiple sites, and exogenous expression of a raptor variant that cannot be phosphorylated by p90RSK attenuates ERK-induced mTORC1 (8). Moreover, ERK directly phosphorylates raptor on multiple residues, leading to the stimulation of mTORC1 signaling (9). Similarly, although Akt phosphorylates, and thereby inactivates TSC2, it also phosphorylates mTOR (34) and the mTORC1 repressor, proline-rich Akt substrate (PRAS) 40 (14). Thus activation of ERK and/or Akt could potentially activate mTORC1 through both TSC2-dependent and -independent mechanisms. To assess the contribution of TSC2-independent mechanisms in the regulation of mTORC1, the effect of LPA and insulin was assessed in MEFs lacking TSC2. Similar to the results from Rat2 cells, in wild-type MEFs LPA and insulin acted in an additive manner to stimulate mTORC1 signaling (Fig. 5A). The combined inhibition of MEK and Akt by U0126 and Akt1/2 KI, respectively, reduced S6K1(T389) phosphorylation to the control value, whereas inhibiting MEK or Akt individually led to only a partial reduction in S6K1(T389) phosphorylation (data not shown). However, in TSC2 knockout (KO) cells, mTORC1 signaling was constitutively high, and neither LPA nor insulin treatment had any additional effect (Fig. 5B). This finding suggests that in the absence of TSC2, ERK signaling to raptor and Akt signaling to PRAS40 to control the activation state of mTORC1 is rendered ineffective.

Fig. 5.

Effect of LPA and insulin on mTORC1 signaling in wild-type (A) or TSC2 KO mouse embryo fibroblasts (MEF) (B). Cells were incubated as described in Fig. 1, and homogenates were subjected to Western blot analysis for phosphorylation of S6K1 on T389. The results represent the means ± SE of 3 or 4 experiments. Within each experiment 3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. P, phosphorylated S6K1; T, total S6K1. *P < 0.0001 vs. no additions; †P < 0.0001 vs. either LPA or insulin alone.

One mechanism through which the ERK and Akt signaling pathways might act in concert to repress TSC2 function and thus stimulate mTORC1 signaling is through phosphorylation of specific residues that act in a complimentary manner to repress the GAP activity of the protein. To assess this possibility, the effect of LPA and insulin treatment on phosphorylation of S939 and T1462, sites phosphorylated by Akt was assessed. Insulin treatment increased phosphorylation of TSC2 on both S939 (Fig. 6A) and T1462 (Fig. 6B). In contrast, LPA alone had no effect on phosphorylation of either S939 or T1462, and LPA had no additional effect in insulin-treated cells, showing that LPA acts through a mechanism distinct from that of insulin.

Fig. 6.

Effect of LPA and insulin on TSC2 phosphorylation on S939 and T1462 in Rat2 fibroblasts. Cells were incubated as described in the legend to Fig. 1, and homogenates were subjected to immunoprecipitation using an anti-tuberous sclerosis complex (TSC2) antibody followed by Western blot analysis of the immunoprecipitates for phosphorylation of TSC2 on (A) S939 or (B) T1462. The results represent the means ± SE of 3 experiments. Within each experiment 3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. P, phosphorylated TSC2; T, total TSC2. *P < 0.002 vs. no additions or LPA alone.

Amino acids are thought to stimulate mTORC1 signaling through a pathway parallel to TSC1/2 (50) that involves the Rag GTPases (21, 42). If insulin and LPA are acting primarily through TSC2 to regulate mTORC1 signaling, then restoration of leucine to the culture medium would be expected to enhance the stimulation of mTORC1 signaling beyond that observed with just insulin and LPA. Indeed, as shown in Fig. 7, leucine acted in an additive manner to stimulate mTORC1 signaling when combined with either insulin or LPA. Moreover, in cells treated with insulin, LPA, and leucine, mTORC1 signaling was significantly greater compared with cells treated with any two agents, confirming that insulin and LPA were acting through a pathway distinct from that activated by leucine.

Fig. 7.

Effect of LPA, insulin, and/or leucine on mTORC1 signaling in Rat2 fibroblasts. Cells were incubated for 2 h in medium containing 0.5% FBS and lacking leucine, and then LPA, insulin, and/or leucine were added at final concentrations of 22 μM, 10 nM, and 0.76 mM, respectively. Two hours later, cells were harvested and homogenates were subjected to Western blot analysis for phosphorylation of S6K1 on T389. The results represent the means ± SE of 6 or 7 experiments. Within each experiment 2–3 wells of cells were independently analyzed. Representative blots are shown beneath the graphs. In the blot shown in the figure, all samples were run on the same gel, but not in contiguous lanes. Noncontiguous lanes are separated by white lines. P, phosphorylated S6K1; T, total S6K1. *P < 0.002 vs. no additions; ‡P < 0.01 vs. LPA, insulin, or leucine alone; †P < 0.05 vs. leucine plus insulin, LPA plus insulin, or leucine plus LPA.

DISCUSSION

In the present study, insulin and LPA were used to selectively stimulate the Akt and ERK signaling pathway, respectively. Individually, insulin and LPA stimulated mTORC1 signaling to a similar extent. However, together insulin and LPA acted in an additive manner, suggesting that they signal through parallel pathways to stimulate mTORC1 signaling. A caveat to this conclusion is that combined treatment with both insulin and LPA increased Akt phosphorylation to a greater extent compared with either agent alone, suggesting that MEK/ERK might act through the Akt pathway to stimulate mTORC1 signaling. If this caveat was correct, then inhibition of Akt should block LPA-induced stimulation of mTORC1 signaling. However, stimulation of mTORC1 signaling by a combination of insulin and LPA was only partially attenuated by inhibition Akt; inhibition of both the Akt and MEK/ERK pathways was required to completely block the stimulation of mTORC1 signaling. This finding provides support for the conclusion that the two pathways function in parallel to stimulate mTORC1 signaling. Results from studies utilizing constitutively active variants of Akt and MEK1 provide further support for this conclusion. Thus exogenous expression of caMEK1 had essentially no effect on Akt signaling, as assessed by phosphorylation of PRAS40, and coexpression of caMEK1 and caAkt did not increase Akt activation beyond that engendered by expression of caAkt alone. However, coexpression of the two constructs resulted in a significantly greater stimulation of mTORC1 signaling compared with that observed with expression of either one alone. Combined, the results strongly suggest that Akt and ERK act through parallel mechanisms to stimulate mTORC1 signaling.

A possible point of convergence of input stimuli through the Akt and ERK pathways is at the TSC1·TSC2 complex. The complex is a nexus through which a variety of positive and negative signals converge to control the activation state of mTORC1. For example, AMP-activated protein kinase (AMPK), which is activated in response to energy stress, phosphorylates, and thereby activates TSC2 (19). Akt and ERK also phosphorylate TSC2, but in contrast to AMPK, phosphorylation by either Akt (38) or ERK (30) represses TSC2 function. Thus phosphorylation of TSC2 can either activate or repress its function, depending on the sites that are modified. Interestingly, the sites phosphorylated by Akt and ERK are distinct (31, 60), providing a potential mechanism through which the two pathways might act in a cooperative manner to repress TSC2 function. In the present study, activation of Akt, but not MEK/ERK, increased the phosphorylation of TSC2 on S939 and T1462, two sites previously shown to be targeted by Akt (60). Unfortunately, we were unable to detect TSC2 phosphorylated on S664, a site known to be phosphorylated by ERK (30), using two different antibodies specific for that residue. However, it is tempting to speculate that phosphorylation of TSC2 by ERK on a residue(s) distinct from those phosphorylated by Akt mediates LPA-induced repression of TSC2, and that a combination of phosphorylation by the two kinases is required for maximal inhibition of TSC2 activity.

Further support for the conclusion that Akt and ERK act through TSC2 to regulate mTORC1 is provided by the finding that leucine restoration to deprived cells stimulated mTORC1 signaling in the presence of both insulin and LPA. Amino acids stimulate mTORC1 signaling through a TSC2-independent mechanism involving the Rag GTPases (22, 42). Although the details are incompletely defined, a recent study suggests that amino acids activate mTORC1 by causing it to be recruited to a complex consisting of the Rag GTPases and a trimeric complex consisting of mitogen-activated protein kinase (MAPK) scaffold protein 1 (MP1), p14, and p18 that is referred to as Ragulator (41). Based on the results of that study, a model has been proposed wherein association of mTORC1 with the Rag/Ragulator complex at the lysosomal membrane brings it into proximity to Rheb (1), resulting in partial activation of kinase. However, full activation of mTORC1 requires both amino acid-induced recruitment of mTORC1 to Rheb, as well as increased GTP loading on Rheb, an event that is mediated by repression of TSC2. The results of the present study are consistent with such a model. Thus leucine restoration to deprived cells resulted in partial activation of mTORC1, as did treatment with either insulin or LPA in the absence of leucine. Combined, treatment with leucine, insulin, and LPA caused a greater stimulation of mTORC1 signaling than did treatment with any pair.

The results of the present study suggest that efforts to increase and/or maintain muscle mass, e.g., in aging or sepsis, through activation of mTORC1 may be enhanced by simultaneous activation of the Akt, MEK/ERK, and Rag pathways. Thus activation of the Rag·Ragulator pathway by amino acids without simultaneously downregulating TSC2 activity, e.g., by increasing insulin concentrations, is relatively ineffective in activating mTORC1 in skeletal muscle. For example, in skeletal muscle of diabetic rats (2) or in rats administered diazoxide to block insulin secretion (3), the amino acid-induced stimulation of mTORC1 signaling is blunted compared with control animals. The results of the present study also suggest that activation of both the Akt and MEK/ERK signaling pathways leads to greater repression of TSC2 activity compared with activation of either alone. In this regard, both the Akt and the MEK/ERK signaling pathways have been implicated in muscle hypertrophy. In skeletal muscle, either exogenous expression of caAkt (35) or treatment with IGF-1 to activate Akt (39, 58) lead to increased fiber diameter. Similarly, expression of caMEK (5, 52) or treatment with clenbuterol to activate the MEK/ERK signaling pathway (47) promotes cardiac and skeletal muscle hypertrophy, respectively, whereas inhibition of ERK signaling abolishes clenbuterol-induced hypertrophy (46). In addition, exogenous expression of MAPK phosphatase 1, which dephosphorylates and represses ERK signaling, decreases fiber size in skeletal muscle (46), further demonstrating the importance of ERK signaling in mediating the hypertrophic response. Both the Akt- and ERK-induced hypertrophy of skeletal muscle are blocked by rapamycin, demonstrating a critical role for mTORC1 in the response. Similar effects are observed in adult rat ventricular cardiomyocytes in which expression of caMEK1 activates mTORC1 (53). Notably, the effect of caMEK1 on mTORC1 signaling is inhibited by rapamycin (16), a result consistent with MEK acting through ERK to inhibit TSC1/2. However, in contrast to the aforementioned studies, evidence supporting a direct role for ERK signaling in muscle hypertrophy under more physiological conditions, e.g., in response to exercise, is less clear. For example, muscle protein synthesis, ERK phosphorylation, and mTORC1 signaling are increased during recovery from resistance exercise (e.g., 6, 11, 56). Whether or not ERK activation is directly involved in the stimulation of protein synthesis and mTORC1 signaling under such conditions is unclear.

Combined activation of the Akt and MEK/ERK pathways may also be important in the development of pathophysiological conditions such as cancer. For example, in anaplastic thyroid cancer, simultaneous genetic alterations in components of both pathways are observed with a frequency of 81% (28), and concurrent activation of the Akt and ERK pathways is also observed in other types of cancer (28, 49, 51). A recent study (44) demonstrated that in tumor cell lines in which independent mutations occur in both the catalytic subunit of phosphoinositide-3 kinase (PI3K) and B-Raf, inhibition of either the Akt or ERK pathway alone is ineffective in repressing cell proliferation. Instead, simultaneous inhibition of both pathways is required to repress the proliferation of such cells, both in culture and in xenografts in mice. A similar requirement for combined inhibition of both pathways when both Akt and ERK are constitutively activated has been observed in earlier studies (13, 17, 45, 55, 59). One interpretation of the results of these studies is that both pathways target a common downstream effector that mediates upregulation of cell proliferation, and that inhibition of both pathways is therefore required to repress the function of the effector. Based on the results of the present study, the common effector may be the TSC1·TSC2 complex.

Overall, the results of the present study support a model in which activation of the Akt signaling pathway results in phosphorylation of TSC2 on specific residues, including S939 and T1462, leading to partial repression of its GAP activity. In contrast, activation of the MEK/ERK pathway has no effect on phosphorylation of either S939 or T1462, but instead leads to phosphorylation on separate residues, most likely S664. Combined, phosphorylation of the Akt- and MEK/ERK-directed sites on TSC2 results in a greater repression of GAP activity toward Rheb compared with phosphorylation of either alone. The results also provide further evidence supporting a model in which amino acids act in a pathway parallel to TSC2 to activate mTORC1.

GRANTS

The studies described herein were supported by National Institutes of Health Grant DK-15658 (to L. S. Jefferson).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).