Signaling Elements Involved in the Metabolic Regulation of mTOR by Nutrients, Incretins, and Growth Factors in Islets

Our initial approach was to assess the effects of forskolin-generated cAMP on S6K1 phosphorylation at basal glucose concentrations. In this experimental design, gel shift assays were used to detect mTOR-mediated phosphorylation of S6K1. As shown in Fig. 3A, incubation of isolated rat islets for 60 min in CMRL tissue culture medium in the presence of an elevated glucose concentration (20 mmol/l) resulted in enhanced phosphorylation of S6K1 (lane 2) as visualized by the appearance of a more slowly migrating upper band in comparison to 3 mmol/l glucose (lane 1). As shown in lanes 3–5, forskolin dose-dependently increased phosphorylation of S6K1 with optimal effects observed at 1 and 10 μmol/l forskolin. The ability of cAMP to enhance S6K1 phosphorylation was independent of insulin secretion and blocked by rapamycin (data not shown). Studies were also performed with MDL-12330A, an adenylyl cyclase inhibitor (21), to determine if the ability of 20 mmol/l glucose to robustly activate mTOR involved the production of cAMP. As shown in Fig. 3B, MDL-12330A dose-dependently reduced glucose-stimulated phosphorylation of S6K1 with a marked inhibition observed at 15 μmol/l (lane 5 vs. lane 2). Under similar conditions, MDL-12330A inhibited forskolin-induced cAMP production by ∼80% (data not shown). This adenylyl cyclase inhibitor had no effect on glucose-stimulated insulin secretion, suggesting that it was not exerting nonspecific effects at these concentrations. These studies also demonstrated that although cAMP increases insulin secretion in a glucose-dependent manner, and thus provides a synergistic, autocrine effect on S6K1 phosphorylation through the insulin-signaling pathway, it also exerts an effect on mTOR independent of insulin secretion. Additionally, cAMP analogs at basal glucose concentrations resulted in phosphorylation of S6K1 comparable to 20 mmol/l glucose (data not shown).

To further characterize the effects of Exenatide on the ability of glucose to stimulate S6K1 phosphorylation via mTOR activation, a dose-response of Exenatide was performed at an intermediate concentration of glucose. As shown in Fig. 4, 8 mmol/l glucose resulted in a partial activation of S6K1 (lane 2 vs. lane 3) and Exenatide at 1, 10, and 100 nmol/l (lanes 4–6) dose-dependently stimulated S6K1 phosphorylation at 8 mmol/l glucose, comparable to that attained at 20 mmol/l glucose. Exenatide (10 or 100 nmol/l) was ineffective in stimulating S6K1 phosphorylation at 3 mmol/l glucose and failed to further enhance S6K1 phosphorylation at 15 or 20 mmol/l glucose, and rapamycin (25 nmol/l) inhibited completely the enhanced phosphorylation of S6K1 at all glucose concentrations (data not shown). These studies demonstrate that the glucose-dependency of Exenatide to stimulate S6K1 phosphorylation is, in fact, mediated through activation of mTOR. Overall, these findings suggest that the ability of GLP-1 agonists to increase cAMP production results in synergistic effects with glucose that may contribute to β-cell growth and survival through mTOR.

GLP-1 increases cAMP production, induces Ca²⁺ influx through l-type voltage-gated Ca²⁺ channels, and causes Ca²⁺-induced Ca²⁺ release (CICR) from intracellular stores (22,23). To assess the effect of enhanced Ca²⁺ influx on mTOR activation, isolated rat islets were exposed to glyburide (1 μmol/l) at a basal glucose concentration (3 mmol/l). As shown in Fig. 5, inhibition of K_ATP channels by glyburide resulted in a partial stimulation of S6K1 phosphorylation (lane 4 vs. lane 1) as indicated by the appearance of a more slowly migrating upper band. This increased phosphorylation of S6K1 was completely blocked by rapamycin (lane 5). Glyburide also increased insulin secretion at basal glucose concentrations. Although insulin secretion in response to 1 μmol/l glyburide was similar to 20 mmol/l glucose, the level of phosphorylation of S6K1 was only partial. Thus, elevated glucose provides an additional signal to mTOR separate from the insulin secretory signal. Overall, these findings suggest that increases in both intracellular cAMP and Ca²⁺ may be involved in the metabolic activation of mTOR.

Our next approach was to examine the effects of preventing the opening of l-type Ca²⁺ channels with diazoxide, an activator of K_ATP channels, on glucose-stimulated S6K1 phosphorylation. Because glyburide, an inhibitor of K_ATP channels, resulted in a partial activation of mTOR as assessed by increased phosphorylation of S6K1, the expected result due to diazoxide exposure was an attenuation of enhanced S6K1 phosphorylation by glucose. An advantage of this model of hyperpolarization is that diazoxide prevents glucose-stimulated insulin secretion due to activation of K_ATP channels but does not alter glucose-stimulated metabolism (24). Thus, the ability of an elevated concentration of glucose to upregulate the mitochondria remains intact. As shown in Fig. 6, diazoxide treatment resulted in a partial inhibition of S6K1 phosphorylation by 20 mmol/l glucose (lane 3 vs. lane 2). Furthermore, cAMP derived from Exenatide (100 nmol/l) or forskolin (10 μmol/l), as shown in lanes 4 and 6, completely reversed this inhibition in a rapamycin-sensitive manner (lanes 5 and 7). Glucose-stimulated insulin secretion was blocked by diazoxide even when S6K1 phosphorylation was restored with Exenatide, and forskolin partially restored glucose-stimulated insulin secretion to ∼50% of control values obtained with 20 mmol/l glucose (data not shown). Because diazoxide prevents Ca²⁺ influx due to activation of K_ATP channels, the ability of cAMP to reverse inhibition of S6K1 phosphorylation is believed to be due primarily to the mobilization of intracellular Ca²⁺ stores.

To further support a role for cAMP-induced mobilization of intracellular Ca²⁺ as a mediator of mTOR activation, we examined the effect of chelating intracellular Ca²⁺ with BAPTA. In this experimental design, diazoxide was used to prevent the influx of extracellular Ca²⁺ in the presence of an elevated glucose concentration. As shown in Fig. 7A, diazoxide partially inhibited the ability of 20 mmol/l glucose to stimulate S6K1 phosphorylation (lanes 2 and 3). Exenatide reversed this inhibition (lane 4), and the presence of BAPTA (10 μmol/l) in combination with Exenatide resulted in nearly complete inhibition of S6K1 phosphorylation (lane 5).

Studies were also performed with nifedipine, an inhibitor of l-type voltage-gated Ca²⁺channels. In Fig. 7B, nifedipine (10 μmol/l) alone failed to inhibit glucose-stimulated S6K1 phosphorylation although glucose-stimulated insulin secretion was inhibited to basal levels (data not shown). These studies suggest that although nifedipine blocks the influx of extracellular Ca²⁺ sufficiently to inhibit glucose-stimulated insulin secretion, it failed to reduce cytosolic Ca²⁺ concentrations to a level necessary to block glucose-stimulated phosphorylation of S6K1. Furthermore, BAPTA alone or in combination with nifedipine resulted in the same magnitude of inhibition, suggesting that BAPTA sufficiently reduced cytosolic Ca²⁺ concentrations to inhibit glucose-stimulated S6K1 phosphorylation. The ability of nifedipine to block the influx of extracellular Ca²⁺ also revealed a protein kinase A (PKA)-sensitive component involved in the mobilization of intracellular Ca²⁺ stores, as unmasked by the PKA inhibitor H-89 (Fig. 7B, lane 3 vs. lane 6). Overall, these results support our hypothesis that the mobilization of Ca²⁺ from intracellular stores by Exenatide or forskolin is essential for the subsequent activation of mTOR by glucose under conditions when Ca²⁺ influx is inhibited.

The proposed mechanism for the metabolic regulation of mTOR is depicted schematically in Fig. 8. In this model, mTOR is activated optimally by integrating components of several pathways. First, the metabolism of glucose and amino acids, in particular the branched chain amino acid leucine, provides TCA intermediates that are used by the mitochondria to generate ATP. ATP also serves as a substrate for adenylyl cyclase. Second, ATP in combination with ADP inhibits K_ATP channels that result in membrane depolarization and influx of extracellular Ca²⁺. This increase in cytosolic Ca²⁺ facilitates CICR from intracellular stores as follows. Extracellular Ca²⁺ influx stimulates adenylyl cyclase, which uses ATP generated by glycolysis and mitochondrial metabolism to generate cAMP, which subsequently mobilizes intracellular Ca²⁺ stores by a PKA-dependent mechanism and, possibly, a PKA-independent mechanism. Overall, this enhanced increase in cytosolic Ca²⁺ derived from the extracellular media and intracellular stores optimally energizes the mitochondria by activating mitochondrial dehydrogenases.

Consistent with this proposed mechanism, glyburide provided partial activation of mTOR under basal glucose concentrations because of the lack of TCA intermediate provision to the mitochondria at 3 mmol/l glucose. Support for modulation of mTOR by cAMP is also provided by the ability of forskolin to promote S6K1 phosphorylation at basal glucose concentrations. Additionally, an adenylyl cyclase inhibitor partially blocked glucose-induced S6K1 phosphorylation. Further support for modulation of mTOR by cAMP is provided by the diazoxide-hyperpolarization model in which Ca²⁺ influx is inhibited although the metabolism of glucose is unaltered (Fig. 6). Inhibition of extracellular Ca²⁺ influx results in a lower generation of cAMP that does not allow for CICR and the upregulation of mitochondrial dehydrogenases for maximal ATP production. Thus, elevated glucose resulted in only partial activation of mTOR. However, the production of cAMP, generated by Exenatide or forskolin, mobilized intracellular Ca²⁺ stores under conditions of reduced extracellular Ca²⁺ influx and produced full activation of mTOR in a BAPTA-sensitive manner. This may be explained by the increase in mitochondria-derived ATP that opposes the effects of diazoxide on K_ATP channels. Exenatide may also be acting through the Gsβγ portion of the GLP-1 receptor to stimulate the PI3K and Akt pathway. There is also the possibility that cAMP may have a direct effect on K_ATP channels.

Although the mitochondrial mediators of mTOR activation are not well characterized, enhanced ATP production is a prime candidate. Our previous studies have indicated that agents that interfere with oxidative phosphorylation result in the inability of glucose to activate mTOR (11). Other studies have demonstrated that interactions of mTOR with the mitochondria in some manner permit mTOR to respond to changes in mitochondrial function, including conditions of osmotic stress and hypoxia (25,26). Recently, mTOR was reported to serve as a physiological homeostatic ATP sensor (27). However, arguments against this hypothesis include the relatively high cellular content of ATP and its maintenance over a narrow concentration range. Of particular relevance to this hypothesis, Tsuboi et al. (28) demonstrated that GLP-1 increases mitochondrial concentrations of both Ca²⁺ and ATP in the β-cell line MIN6 under basal glucose concentrations. They propose that GLP-1 evokes the release of intracellular Ca²⁺ stores in a cAMP-dependent manner that leads to increase in mitochondria-localized Ca²⁺ and to a stable 1.5- to 2.0-fold increase in ATP synthesis. Their results suggest that β-cells specialize in coupling changes in electrical activity to glucose metabolism to regulate insulin secretion and may depend on mitochondria-derived ATP, compartmentalized in a manner unique from other cell types. Although Tsuboi et al. (28) focused on the mechanism of GLP-1 potentiation of glucose-induced insulin secretion, their findings are also pertinent to the metabolic regulation of mTOR. Another potential candidate as mediator of mTOR activation is AMP, which is argued to serve as a more sensitive indicator of the β-cell energy status than ATP. Recent studies have proposed, as described in Fig. 1, that AMP-dependent protein kinase (AMPK) suppresses mTOR through the TSC1/2 complex. There may also be a role for Rheb/GTP signaling to mTOR (rev. in 1).

Although glucose-stimulated insulin secretion and mTOR activation utilize similar signaling pathways, the following significant differences exist: 1) glucose activation of mTOR is amino acid-dependent, whereas glucose-stimulated insulin secretion is not; 2) rapamycin at nanomolar concentrations completely inhibits glucose activation of mTOR but has no effect on glucose-stimulated insulin secretion; 3) mTOR activation is achieved at very low intracellular Ca²⁺ concentrations that are insufficient to support glucose-stimulated insulin secretion; and 4) glucose-stimulated activation of mTOR is partially blocked by the adenylyl cyclase inhibitor, MDL-12330A, whereas glucose-stimulated insulin secretion is unaltered.

Briaud et al. (29) have reported that glyburide at basal glucose concentrations did not stimulate S6K1 phosphorylation, and verapamil, an inhibitor of voltage-dependent Ca²⁺ channels, also failed to block glucose stimulation of S6K1 phosphorylation. An explanation for this lack of agreement with our findings is that these particular studies with glyburide were performed in a β-cell line, INS-1, as opposed to primary islets that exhibit a robust activation of S6K1 by glucose. It is also probable that cAMP production by INS-1 cells differs significantly from primary β-cells. Based on our studies with nifedipine alone and in combination with BAPTA to lower cytosolic Ca²⁺ concentrations by chelation, the inability of verapamil to produce partial inhibition of S6K1 phosphorylation by glucose may be due to the incomplete blockage of extracellular Ca²⁺ entry (compare lanes 3 and 4 in Fig. 7B).

Although our understanding of the regulation of mTOR in β-cells is not yet complete, some important regulatory sites are described in Fig. 8. Because activation of β-cell mitochondria exerts a significant role in nutrient stimulation of mTOR, upregulation of glycolysis and/or provision of TCA intermediates represents a potential strategy to further enhance mTOR-mediated growth and proliferation. This may be the mechanism, in part, whereby the branched chain amino acid leucine, which is exclusively metabolized by the mitochondria, in combination with glucose participates in the activation of mTOR. Our studies have also identified K_ATP channels as possible therapeutic targets by their ability to regulate membrane depolarization and the influx of extracellular Ca²⁺. It appears that the large oscillatory increases in cytosolic Ca²⁺ derived from only the extracellular media by metabolic inhibition of K_ATP channels are sufficient to facilitate both mTOR activation and insulin secretion. In contrast, the smaller increases in cytosolic Ca²⁺ achieved by the mobilization of only intracellular Ca²⁺ stores selectively activate mTOR, probably due to the very low cytosolic Ca²⁺concentrations required to upregulate the mitochondria, whereas higher cytosolic Ca²⁺ is necessary for glucose-stimulated insulin secretion. In our studies with primary islets, glyburide and agents that increase cAMP, including Exenatide and forskolin, were very effective in restoring full mTOR activation at suboptimal concentrations of glucose. These latter therapeutic sites would be advantageous under conditions in which the metabolism of glucose is impaired. The mobilization of intracellular stores of Ca²⁺ would be a site for therapeutic intervention under conditions in which sources of extracellular Ca²⁺ could not be utilized. In the event that adenylyl cyclase cannot function effectively to produce cAMP, cell-permeable analogs of cAMP may be utilized. In addition, Exenatide activation of the Gsβγ portion of the GLP-1 receptor and stimulation of the PI3K and Akt pathway may be useful in overcoming a defect in adenylyl cyclase. Insulin, IGF-1, and other growth hormones such as GH and PTH (growth hormone and parathyroid hormone) may also exert a positive role in the upregulation of the mTOR pathway in conjunction with nutrient signaling.

Our findings suggest that the glucose-dependency of incretins to increase growth and proliferation may, in part, involve mTOR activation. This is attributed primarily to the function of the Gsα subunit of the GLP-1 receptor that activates adenylyl cyclases to produce cAMP. The ability of cAMP to mobilize intracellular Ca²⁺ stores enhances mitochondrial activation, which in turn can modulate K_ATP channels in the presence of the nutrients—glucose and amino acids—and thereby contributes to optimal activation of mTOR. mTOR activation is regulated by both basal and elevated glucose concentrations that are due, in part, to metabolic modulation of K_ATP channels. Our inability to demonstrate an effect of incretins at basal glucose may be due to the production of relatively low levels of cAMP in comparison to pharmacologic approaches employing forskolin (30). The potential significance of this signaling pathway is that mTOR remains activated during a continuous stimulation with elevated glucose concentrations and mediates DNA synthesis by primary islets in a rapamycin-sensitive manner. The ability of rapamycin to block basal glucose-stimulated ³H-thymidine incorporation demonstrates a novel regulation of mTOR at low glucose concentrations. Based on indirect approaches with an adenylyl cyclase inhibitor, a significant component of glucose-mediated DNA synthesis may involve a role for cAMP. It is hoped that a more thorough knowledge of the mTOR signaling pathway will make it possible to use nutrients, incretins, and growth factors along with current diabetes treatments to circumvent points of dysfunction and promote β-cell survival, growth, and proliferation.

Exenatide was a kind gift from Amylin Pharmaceuticals (San Diego, CA) and Eli Lilly (Indianapolis, IN). This work was supported by National Institutes of Health Grant DK-55024 (M.L.M.), an American Diabetes Association Mentor-Based Postdoctoral Fellowship (K.L.P.), and an American Diabetes Association/Takeda Pharmaceuticals Mentor-Based Minority Postdoctoral Fellowship (M.S.R.).

All procedures performed on animals were approved by the Washington University Animal Studies Committee.