ReviewTranscription

Transcriptional regulation of the stress response by mTOR

See allHide authors and affiliations

Science Signaling  01 Jul 2014:
Vol. 7, Issue 332, pp. re2
DOI: 10.1126/scisignal.2005326

Abstract

The kinase mammalian target of rapamycin (mTOR) is a central regulator of cell growth and proliferation that integrates inputs from growth factor receptors, nutrient availability, intracellular ATP (adenosine 5′-triphosphate), and a variety of stressors. Since early works in the mid-1990s uncovered the role of mTOR in stimulating protein translation, this kinase has emerged as a rather multifaceted regulator of numerous processes. Whereas mTOR is generally activated by growth- and proliferation-stimulating signals, its activity can be reduced and even suppressed when cells are exposed to a variety of stress conditions. However, cells can also adapt to stress while maintaining their growth capacity and mTOR function. Despite knowledge accumulated on how stress represses mTOR, less is known about mTOR influencing stress responses. In this review, we discuss the capability of mTOR, in particular mTOR complex 1 (mTORC1), to activate stress-responsive transcription factors, and we outline open questions for future investigation.

mTOR Complexes in Mammalian Cells

The serine/threonine kinase target of rapamycin (TOR) belongs to the family of phosphatidylinositol 3-kinase (PI3K)–related kinases (PIKKs) and is a main activator of biosynthetic processes needed for cell growth in all eukaryotic organisms (1). TOR is the most frequently used acronym for this kinase, but earlier works also used the name FRAP, for FK506-binding protein 12–rapamycin–associated protein. The acronym mTOR refers to “mammalian TOR,” although recently “mechanistic target of rapamycin” is commonly used. The kinase mTOR functions in larger multiprotein complexes: mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). mTORC1 is defined by regulatory-associated protein of mTOR (Raptor) (2, 3), whereas mTORC2 contains the proteins rapamycin-insensitive companion of mTOR (Rictor) and stress-activated mitogen-activated protein kinase (MAPK)–interacting protein 1 (Sin1) (4, 5). These and additional mTOR-interacting proteins regulate its activity and specificity toward various substrates (6). mTORC1 is activated by growth factors, nutrients, and energy and promotes cell growth by enhancing the translation rate of diverse proteins by activating the ribosomal S6 subunit kinases (S6K) 1 and 2 (1, 7, 8) and inactivating the translation repressor protein eukaryotic translation initiation factor 4E (eIF4E)–binding protein 1 (4E-BP1) (9). mTORC1 also enhances protein synthesis indirectly by increasing the activity of RNA polymerases I and III, which transcribe genes encoding ribosomal and transfer RNAs (10).

In addition, mTOR influences diverse transcription factors and the expression of gene products involved in the control of metabolism, ribosomal biogenesis, growth, and proliferation (1116). mTORC2 was originally shown to be essential for the function of the actin cytoskeleton (4, 17), but is now also known to regulate cell growth, differentiation, proliferation, and lipid homeostasis (18, 19). At least part of the growth-promoting activity of mTORC2 is mediated by activating and stabilizing the kinase Akt [also known as protein kinase B (PKB)], which in turn enhances the activity of mTORC1 (20, 21). A defining feature of mTOR is its inhibition by rapamycin (22, 23), a compound originally isolated from the bacterium Streptomyces hygroscopicus that binds to the intracellular chaperone FK506-binding protein 1A, 12 kD (FKBP12). The rapamycin-FKBP12 complex in turn binds with high affinity to the FKBP12-rapamycin binding (FRB) domain in TOR. This domain is accessible to the rapamycin-FKBP12 complex in mTORC1, but not in mTORC2. Structural studies show that mTORC1 complexes are dimeric, with two molecules of mTOR and Raptor per complex (24). This work, together with earlier biochemical analysis, shows that rapamycin-FKBP12 alters the conformation of the mTOR-Raptor complex, weakening their interaction and destabilizing the mTORC1 dimer (2, 24). Notably, activation of mTORC1 by nutrients causes a conformational change in the mTOR-Raptor complex that makes it more sensitive to the destabilizing effect of rapamycin (2). Rapamycin-FKBP12 rapidly suppresses the activity of mTORC1 toward many, although not all, of its substrates (2527). However, when cells are incubated with rapamycin for long periods (hours to days), rapamycin-FKBP12 can bind to newly synthesized mTOR before it can assemble into new mTORC1 and mTORC2 complexes, precluding the regeneration of their pools and inhibiting both mTORC1- and mTORC2-dependent functions (28).

Stress-Sensitive Checkpoints in mTOR Activity

The activity of mTOR is sensitive to complex signaling networks, and the number of proteins that at some point can influence mTOR function is probably in the hundreds (29). From this perspective, it is conceivable that diverse sources of stress may affect mTOR complexes by acting on different components in these networks, and indeed, there are several well-characterized regulators and signaling circuits that inhibit the activity of mTOR under stress (3033) (Fig. 1, A and B).

Fig. 1 mTOR sensitivity to stressors, and stress-responsive transcription factors stimulated by mTOR.

(A) mTOR integrates signaling from growth factor, nutrient, and energy sensors to stimulate growth-promoting processes and can be inhibited by diverse types of stress conditions. (B) A schematic view of mTORC1 inhibition by the TSC1/2 complex in response to diverse stressors, by other negative regulators (such as AMPK, PRAK, and MARK4) under energy stress, or through deactivation of the Ragulator complex under amino acid deprivation. (C) Stress-responsive transcription factors that can be stimulated by mTORC1.

A substantial part of mTORC1 inhibitory inputs are channeled through the tuberous sclerosis (TSC) proteins TSC1 and TSC2 (34, 35). mTORC1 activity is primarily regulated by its associated protein Ras homolog enriched in brain (Rheb), a small guanosine triphosphatase (GTPase) that, in its guanosine triphosphate (GTP)–loaded state, activates mTORC1. Rheb enables the activation of mTORC1 in response to growth factors and, together with the Ragulator complex, to amino acids (36), a process that occurs at the lysosome surface (3739). TSC2 acts as a GTPase-activating protein (GAP) and promotes GTP hydrolysis to convert Rheb-GTP to Rheb-GDP (guanosine diphosphate), thereby inactivating mTORC1 (40). TSC2 exists in a heterodimeric complex together with TSC1, which has no GAP activity but is needed to stabilize TSC2 (41, 42). The GAP activity of the TSC1/TSC2 complex is assisted by a third component of TSC, Tre2-Bub2-cdc16 (TBC) 1 domain family, member 7 (TBC1D7) (43). The GAP activity of TSC2 is adjusted through inhibitory and activating phosphorylations. TSC2 phosphorylation by the growth factor–activated kinase Akt/PKB disrupts its association with TSC1 and causes the dissociation of the TSC complex from the lysosome, enabling the regeneration of Rheb-GTP and the assembly of mTORC1 with the Ragulator complex and Rheb (39, 44). TSC can also be inactivated through phosphorylation of TSC2 by the p90 ribosomal protein S6 kinase α-1 (RSK1) (45). On the other hand, TSC is activated by phosphorylations mediated by glycogen synthase kinase 3β (GSK3β) (46) and adenosine 5′ monophosphate (AMP)–activated kinase (AMPK) in response to energy stress and reactive oxygen species (ROS) (47, 48). AMPK is a major sensor of energy stress and is activated in response to the increase in the intracellular ratio of AMP and adenosine 5′-diphosphate (ADP) to adenosine 5′-triphosphate (ATP) (4951). Another mTORC1-suppressive mechanism under energy stress is the inhibitory phosphorylation of Rheb by the complex formed by p38β with the p38-regulated/activated protein kinase [PRAK, also known as the MAPK-activated protein kinase 5 (MK5)] (52). TSC2 activity can also be enhanced by the DNA damage and hypoxia-inducible protein regulated in development and DNA damage responses (REDD1), also known as DNA damage–inducible transcript 4 (DDIT4) (30, 53), which facilitates the dissociation of TSC2 from the scaffold protein 14-3-3 (54). TSC2 is also activated by the REDD1 homolog REDD2 (30, 53) and by sestrins 1 and 2 induced by the transcription factor p53 in response to DNA damage (55). The activity of mTORC1 can also be decreased under energy stress conditions through the inhibitory phosphorylation of Raptor at Ser722/792 by AMPK (56) and the AMPK-related kinase MAP/microtubule affinity-regulating kinase 4 (MARK4) (57).

TSC2-activating and inhibitory circuits are also regulated by changes in the abundance of their components. For instance, in response to DNA damage, the tumor suppressor and transcriptional regulator p53 can enhance the expression of genes encoding mTORC1 repressors, including TSC2, AMPK, REDD1, sestrin 2, and the PI3K antagonist phosphatase and tensin homolog (PTEN) (58). REDD1 is also induced in a p53-independent manner in response to energy stress (59) and by the hypoxia-inducible factor (HIF) during hypoxia (60, 61), in which mitochondrial ATP synthesis is compromised while production of mitochondrial ROS is enhanced.

Activation of mTORC2 depends on PI3K signaling (26, 62) and would be expected to be sensitive to perturbations of this pathway under stress. Notably, mTORC2 is partially repressed, although not fully inactivated, upon phosphorylation of its essential component Rictor by the mTORC1-activated S6K1 and by negative feedback through mTORC1 on insulin receptor signaling (6365). Another intriguing aspect of mTORC2 is its positive regulation by TSC2, an effect that is independent of the GAP activity of TSC1/2 and inhibition of mTORC1 (66). The finding that mTORC1 can attenuate mTORC2 activity does not mean, though, that mTORC2 is not functioning in cells with active mTORC1. Rather, both complexes are active in growing and proliferating cells (26, 67) and should be viewed as continuously communicating and balancing each other (Fig. 1B).

Regulation of Stress Response Transcription Factors by mTOR

Adaptive mechanisms to survive and maintain functionality under stress often involve changes in gene expression patterns. Although a large number of transcription regulators are directly or indirectly sensitive to perturbations of the intracellular milieu, which in itself could be used by cells to react to various stressors, certain transcription factors have critical roles in stress responses. For example, HIF-1α, ROS-activated nuclear factor (erythroid-derived 2)–like 2 (NFE2L2 or Nrf2), hypertonicity-activated nuclear factor of activated T cells 5 (NFAT5), heat shock response factor (HSF1), DNA damage–activated p53, and the homeostasis regulator forkhead box O (FOXO) proteins orchestrate stress-specific adaptation responses and are sensitive to mTOR activity (Fig. 1C).

mTORC1 and HIF-1α

The hypoxia response factor HIF is a heterodimer formed by a constitutive subunit, HIF-1β, and an inducible subunit, HIF-1α. HIF-1α is continuously synthesized, but under normoxic conditions, it is rapidly degraded (68). HIF-1β can also dimerize with the HIF-1α homolog HIF-2α, which is also regulated through continuous synthesis and degradation. Although both HIF-1α and HIF-2α can activate hypoxia-protective responses, they do not have entirely overlapping functions (69, 70). Under normoxic conditions, HIF-1α is inhibited by two oxygen-dependent, ROS-sensitive types of enzymes: prolyl hydroxylases (PHDs) and factor inhibiting HIF-1 (FIH). Both proteins use molecular oxygen and α-ketoglutarate to hydroxylate key residues in HIF-1α. PHD-mediated hydroxylation of Pro402 and Pro564 in HIF-1α targets it for ubiquitylation by the von Hippel–Lindau (VHL) E3 ubiquitin ligase complex and subsequent proteasomal degradation. FIH hydroxylates Asn803, a modification that inhibits HIF-1 transcriptional activity, although it does not cause its degradation (71, 72). Hypoxia causes the activation of HIF-1–dependent responses by suppressing PHD and FIH. Decreasing oxygen concentration to 0.5 to 2% causes a rapid accumulation of HIF-1α and expression of its target genes (73, 74). A major effect of activating HIF-1α is the switch in the mode of energy production in the mitochondria from respiration-dependent to glycolysis-dependent ATP synthesis (75). This aspect is quite interesting because it means that HIF-1α is an effective regulator of energy and metabolite resources. Indeed, it is now clear that some cell types, such as activated lymphocytes and tumor cells, also use HIF-1α to boost glucose uptake and glycolysis during normoxia as well to respond to increased biosynthetic and energetic demands (76, 77).

mTORC1 can enhance HIF-1α activity in both normoxia and hypoxia. During normoxia, and despite the activity of PHDs, mTORC1 facilitates the accumulation of HIF-1α in normal and tumor cells by increasing its translation rate (7680). mTORC1 can also enhance HIF-1α protein abundance during hypoxia and chemically induced hypoxia-like conditions (81, 82) and increase its transcriptional activity under hypoxia by facilitating the interaction of HIF-1α with the transcriptional coactivator p300 (83). mTORC1 activity can be inhibited during hypoxia; one mechanism by which this occurs is through HIF-1α–induced expression of the gene encoding mTORC1 inhibitor REDD1 (54, 60, 84), which suggests reciprocal regulation and a balance between mTOR activity and the intensity of the hypoxia response.

mTORC1 and Nrf2

The transcription factor Nrf2 responds to increases in ROS concentrations and enhances the expression of genes encoding diverse proteins involved in maintaining cellular redox balance, such as glutathione-synthesizing enzymes, thioredoxin reductases, or peroxiredoxins (85). The abundance of Nrf2 is controlled by a repressor, Kelch-like ECH-associated protein 1 (Keap1), which facilitates the ubiquitylation and subsequent proteasome-mediated degradation of Nrf2 (8688). Increased intracellular ROS induces the oxidation of various cysteines in Keap1, triggering a conformational change that causes it to dissociate from Nrf2, which then escapes the ubiquitylation-degradation cycle and induces antioxidant response genes (85). Activation of Nrf2 in the absence of overt oxidative stress can also have important effects, as seen in tumors bearing mutations in either KEAP1 or NFE2L2 that disrupt the repression of Nrf2 by Keap1 (85, 89). Enhanced Nrf2 activity in tumors increases survival not only to endogenous ROS but also to chemotherapeutic drugs and confers stronger antiapoptotic defenses and cell proliferation capability (90).

Nrf2 is sensitive to mTOR activity, although different studies show different effects of mTOR on Nrf2. Rapamycin represses Nrf2-dependent antioxidant responses in renal carcinoma cells (91), suggesting a positive regulation by mTORC1. Activation of Nrf2 by its release from Keap1 and induction of cytoprotective target genes are also enhanced by mTORC1 during selective autophagy (92). However, a study in HepG2 hepatocarcinoma cells showed that although rapamycin reduced basal abundance of Nrf2, it did not inhibit its accumulation in response to (R)-α-lipoic acid, a dithiol redox-active compound that induces Nrf2-dependent gene expression (93), which suggests that oxidative stress can increase Nrf2 independently from mTORC1. On the other hand, prolonged treatment of human fibroblasts with rapamycin reduced Keap1 abundance and enhanced Nrf2 accumulation (94). Therefore, mTORC1 can have both positive and negative effects on Nrf2 in different cell types and experimental settings. As will be discussed further in the following sections, mTORC1 can also enhance ROS production in epithelial stem cells and hematopoietic stem cells (HSCs) (95, 96), which suggests that the potential protective function of mTORC1 may be offset by its ROS-promoting effects in some cell types and conditions.

mTORC1 and HSF1

HSF1 is a heat stress–activated transcription factor that is ubiquitously expressed in mammalian cells and exists as a monomeric inactive form in unstressed cells (97). Heat, as well as heat-independent stressors, such as H2O2, low pH, and hypoosmotic and hyperosmotic stress, activates HSF1 by inducing its trimerization, which enables it to bind DNA (97). HSF1 induces the expression of genes encoding various heat shock proteins (HSPs), including HSP70, to chaperone other proteins at risk of aggregation and misfolding (98). In turn, accumulating HSP70 induces negative feedback on HSF1 by binding HSF1 and repressing its transcriptional activity (99, 100). HSF1 can also play substantial roles in addition to its stress-responsive function, as shown by the recent identification of numerous genes regulated by HSF1 that support oncogenic processes in tumor cells (101).

Short-term heat shock has been shown to activate the kinase S6K downstream of mTORC1 (102, 103) and enhance the transcriptional activity of HSF1 and its induction of HSPs in a human cancer cell line by phosphorylating it in Ser326 (103). The later study showed that heat shock–enhanced S6K1 phosphorylation and activation of HSF1 were rapamycin-sensitive, suggesting the involvement of mTORC1. Moreover, a recent screening for tumor-suppressing compounds showed that translation inhibitors, including rapamycin and other inhibitors of the PI3K-mTOR pathway, prevented HSF1 from binding DNA in various cancer cell types (104), which provides another mechanism by which mTOR activity may enhance HSF1 function. However, a study in Saccharomyces cerevisiae showed that sustained activation of yeast HSF1 inhibited TOR signaling (105). Although both studies were done in very different organisms, they suggest that mTORC1-activated HSF1 could negatively feedback on mTORC1 to modulate its activity.

mTORC1 and NFAT5

NFAT5 belongs to the family of proteins with a Rel-like DNA binding domain, which also comprises calcineurin-activated NFAT1, NFAT2, NFAT3, and NFAT4 and nuclear factor κB (NF-κB) (106, 107). One of the most extensively characterized functions of NFAT5 is its ability to protect cells from osmotic stress caused by excessive extracellular tonicity, which it does by inducing various chaperones, osmolite transporters, and enzymes that enable cells to survive and adapt to prolonged osmotic stress (108, 109). Osmotic stress can inhibit the activity of mTORC1 and mTORC2 complexes, although the degree of inhibition differs with stress intensity and duration. Exposure of diverse cell lines to intense, short-term osmotic stress (more than 600 mosmol/kg for up to 1 hour) causes a substantial decrease in the phosphorylation of major mTORC1 and mTORC2 targets, such as S6K1 or Akt (110112). Intriguingly, one work reported that a 30-min pulse of hyperosmotic stress (800 mosmol/kg) enhanced the activity of S6K1 in fibroblasts, although whether this was due to increased mTORC1 activity was not tested (113). Although the inhibitory effect of osmostress on mTOR has been known for years (30, 31), the mechanisms involved have not been fully elucidated. The ability of AMPK to respond to diverse stressors, including osmotic stress (51), would make it a possible candidate to inhibit mTORC1 under osmostress. However, recent studies do not support a main role for AMPK in regulating mTORC1 in cells exposed to hypertonicity (67, 114). Earlier studies showed that inhibition of S6K1 and Akt in cells exposed to high, short-term osmostress was mediated by a calyculin A–sensitive phosphatase (110, 111). In this regard, it was recently shown that short-term osmotic stress activated the intrinsic kinase activity of mTORC1 through the phosphorylation of Raptor by c-Jun N-terminal kinase (JNK), but still inhibited the phosphorylation of S6K1 and 4E-BP1 in cells (114). Calyculin A abrogated the inhibition of S6K1 and 4E-BP1 and enabled their robust phosphorylation by mTORC1 under osmotic stress (114). These findings indicate that high-intensity osmotic stress can interfere with signaling pathways downstream of mTORC1 without inhibiting its kinase activity. On the other hand, mammalian cells exposed to milder hypertonic conditions (500 mosmol/kg) exhibited a moderate inhibition of mTOR signaling (67), but maintained the function of mTOR complexes to sustain protein synthesis, cell growth, and cell proliferation under osmostress (67, 115).

An earlier study in yeast had already shown that TOR promoted survival under salt stress (116). In mammalian cells, mTORC1 stimulated the expression of several osmoprotective gene products under moderate osmostress and enhanced histone acetylation in their promoters and the recruitment of NFAT5 and RNA polymerase II (67). mTORC1 was found to influence the expression of a subset of NFAT5-regulated genes as well as several NFAT5-independent ones, suggesting that mTOR influences other mechanisms or factors in the osmotic stress response (67). One such NFAT5-dependent osmotic stress–responsive gene encodes REDD1, the abundance of which is increased at both the mRNA and protein levels by mTOR. REDD1 can inhibit mTORC1 activity in various stress contexts (54, 61), but intriguingly, its induction by osmotic stress does not appear to inhibit mTORC1 (67). Hypertonic stress can enhance mitochondrial ROS production (117, 118), which can inhibit mTORC1 (119). Because REDD1 attenuates the generation of mitochondrial ROS (120), it is possible that the mTORC1 inhibitory function of REDD1 under osmotic stress could be compensated by a protective antioxidant effect.

mTORC1 and p53

The tumor suppressor transcription factor p53 is well known for its responsiveness to DNA damage and its ability to prevent cellular transformation by inducing gene products that control the cell cycle, DNA repair, senescence, and apoptosis (121123). Among other targets, p53 induces the expression of various genes encoding proteins capable of inhibiting mTOR activity directly or indirectly [including PTEN, AMPK, TSC2, REDD1, sestrins, and TIGAR (TP53-induced glycolysis and apoptosis regulator) (58)], thus slowing down cell growth and attenuating biosynthetic processes that could favor the expansion of altered cells. Through AMPK-mediated phosphorylation of p53 at Ser15, p53 can also repress growth and proliferation under energy stress and in the absence of DNA damage (124). Besides these stress-activated functions, p53 can influence mTOR activity by modulating glucose utilization and energy production by inhibiting various steps in glycolysis and enhancing mitochondrial respiration (125) and by providing antioxidant defenses through the induction of proteins that are involved in ROS neutralization, such as sestrins (126, 127).

As found for other stress-responsive transcription factors, p53 is also sensitive to mTORC1 activity. mTORC1 can increase p53 activity by enhancing its translation rate (128, 129) or by activating S6K1 that then sequesters and neutralizes the E3 ubiquitin protein ligase mouse double minute 2 (MDM2), a p53 repressor (130). It is interesting that increased activation of mTOR in TSC- or REDD1-deficient cells increases their sensitivity to stressors that activate p53; thus, in a sense, mTORC1 can be viewed as an amplifier of the proapoptotic function of p53 under stress. A similar stress sensitization mechanism in TSC1 mutant cells functions through the enhanced mTORC1-dependent translation of alternative reading frame (ARF), a cell cycle repressor and p53 activator (131). In addition, mTORC1 can enhance p53-dependent induction of replicative senescence in cells subjected to moderate DNA damage and also in nonstressed cells (132, 133). The stimulatory effect of mTORC1 on p53 could be one reason why TSC-deficient cells that maintain p53 functionality mostly give rise to benign tumors (128, 134).

mTOR complexes and FOXO

Transcription factors of the FOXO family are important regulators of cellular homeostasis that play protective roles under diverse types of stress but can also enhance cell death mechanisms (135). In different contexts, FOXO transcription factors can induce the expression of genes encoding proteins that arrest the cell cycle, activate apoptosis, enhance antioxidant responses, promote autophagy, or modulate receptor signaling, including signaling in mTOR-mediated pathways (135).

Stress conditions that stimulate FOXO activity are generally inhibitory for mTOR complexes. Several kinases that activate FOXO, such as JNK in response to ROS (136), AMPK under energy stress (137, 138), and MK5 (or PRAK) in response to DNA damage (139), can either inhibit mTORC1 directly or inhibit upstream signaling to mTORC1 and mTORC2. FOXO can also oppose mTORC1 by inducing its negative regulator sestrin 3 (140). Conversely, by activating Akt (141, 142) or serum- and glucocorticoid-regulated kinase 1 (SGK1) (143, 144), mTORC2 represses FOXO activity. In glioblastoma tumors, mTORC2 also opposes FOXO by inhibiting its deacetylation by Sirt1 (145). Intriguingly, FOXO promotes the expression of the gene encoding PI3K and enhances PI3K signaling in certain tumor cells (146) as well as the expression of the gene encoding the mTORC2 component Rictor (140), which suggests that FOXO can induce its own negative feedback through the expression of several repressors.

However, although these observations reveal some degree of antagonism between FOXO and both mTOR complexes, there are also specific contexts in which FOXO can benefit from the ability of mTORC1 to attenuate mTORC2. In T lymphocytes, excess mTORC1 activation through deletion of TSC1 leads to reduced mTORC2 function along with a substantial reduction in mTORC2 and Akt-dependent repressive phosphorylation of FOXO1 and FOXO3 in response to T cell receptor (TCR) activation (147). TSC1-deficient T lymphocytes also exhibit a stronger induction of apoptosis upon TCR activation and greater expression of several FOXO-regulated genes encoding proteins that promote apoptosis and cell cycle arrest, such as Bcl-2–like protein 11 (BCL2L11) and cyclin-dependent kinase inhibitor 1A (CDKN1A, also known as p21Cip1) (147). Additionally, rapamycin treatment in human colon cancer cell lines inactivates FOXO concomitantly with increased Akt activity (148).

Negative effects of mTOR during stress responses

Although mTORC1 can be beneficial for the stress response, augmented mTORC1 activity can be detrimental. For instance, TSC2-deficient cells cannot shut down mTORC1 when their DNA is damaged and die faster than wild-type cells because mTORC1 enhances the accumulation of p53 (128). Likewise, TSC2-deficient cells poorly survive glucose deprivation because they cannot cope with the energy drain caused by mTORC1-driven protein synthesis (149). Studies in flies show that enhancing TORC1 activity by overexpressing Rheb increases their sensitivity to oxidative stress (150), and other studies in flies and mice reveal that excess TORC1 activity worsens neuronal deterioration caused by huntingtin aggregates, whereas inhibition of TORC1 activates an autophagic response that clears protein aggregates and improves neuron survival (151, 152). These findings indicate that excessive mTORC1 activity could overload mechanisms of protein quality control, as confirmed in a recent article showing that TSC2-deficient fibroblasts exhibit decreased mRNA translation fidelity (153).

Increased activation of mTORC1 by the suppression of TSC1 or overexpression of Rheb also impairs hematopoiesis and self-renewal of HSCs (96, 154). This repressive effect correlates with increased mitochondrial biogenesis and increased amounts of ROS, and the neutralization of ROS restores HSC function in Tsc1-deficient mice (96). mTORC1 also increases ROS in normal oral keratinocytes and epithelial stem cells by decreasing the abundance of mitochondrial manganese superoxide dismutase (MnSOD) (95). These effects seem at odds with the ability of mTORC1 to activate antioxidant defenses through Nrf2 described in other cells (92), suggesting that the balance between the ROS-protective and ROS-promoting activities of mTORC1 may differ substantially in different cell types and microenvironments.

If reducing mTORC1 activity can help cells to resist certain stresses, is it then paradoxical that mTORC1 can also stimulate protective stress responses? Both possibilities can be reconciled by proposing that partial suppression of mTORC1 may protect stressed cells by reducing the rate of biosynthesis and the cellular energy demand while enabling sufficient mTORC1 activity to support prosurvival stress responses (Fig. 2, A and B). As described above, examples where the sustained activation of mTORC1 reduces resistance to stress include its roles in increasing mitochondrial ROS production; enhancing p53’s antiproliferative, proapoptotic, and senescence-promoting functions; exacerbating ATP depletion in nutrient-restricted cells; or disrupting proteostasis through excessive protein translation (95, 128, 149, 153). A common theme in these effects may be the persistence of energy-draining mTORC1-driven biosynthetic processes and the overloading of mechanisms for eliminating endogenous noxious and waste products (Fig. 2C). On the other hand, the stimulation of stress resistance responses by mTORC1 is observed in cells where mTORC1 is not constitutively augmented, under conditions that can either maintain or transiently enhance its activity [such as hypoxia, xenophagy, and short-term heat shock (79, 82, 83, 92, 103)] or reduce it [such as hyperosmotic stress (67)]. These observations may suggest that sustained, increased activity of mTORC1 sensitizes cells to several types of stress, whereas moderate, regulated activity of mTORC1 would favor stress adaptation responses (Fig. 2, B and C). Here, it must be noted that stress conditions under which mTORC1 can stimulate transcriptional responses [such as hypoxia (82, 83), hyperosmotic stress (67), or DNA damage (130)] can also inhibit mTORC1 (60, 67, 112, 134). This suggests that the extent of mTORC1 activity under increasing stress intensity will be important for determining its capacity to enhance protective responses. These examples suggest that careful analysis of mTOR function under moderate stress conditions could uncover other mTOR-sensitive transcriptionally regulated processes. For instance, severe nutrient deprivation or intense endoplasmic reticulum (ER) stress can suppress mTOR, but a milder intensity of these stressors could allow enough mTOR activity to support relevant cell functions. In this regard, mTORC1 antagonizes the induction of the ER stress–responsive factors activating transcription factor 4 (ATF4), ATF6, and CCAAT/enhancer binding protein homologous protein (CHOP) (155), and emerging evidence suggests that a bidirectional crosstalk exists between mTOR complexes and the unfolded protein response (156).

Fig. 2 Potential scenario compatible with mTOR being partially inhibited under stress but still capable of enhancing adaptive stress responses.

(A) Moderate stress intensity can cause a partial suppression of mTOR activity and initiate stress responses with induction of stress-adaptation genes. Even if mTOR activity fell below the threshold required to sustain rapid proliferation or growth rate, cells could still have sufficient mTOR to enhance protective responses and facilitate a faster adaptation to stress. Pharmacological mTOR inhibitors such as rapamycin or Torin1 can prevent mTOR from enhancing stress responses, as shown for HIF-1α, HSF1, Nrf2, or NFAT5. High-intensity stress can suppress mTOR so that it cannot further stimulate some stress responses. In this model, lack of an enhanced stress response mediated by mTOR may not be necessarily detrimental for cells, and they could survive stress—despite inducing a less robust response—by decreasing high-energy expending and stress-sensitizing biosynthetic processes. (B) In cells with normally regulated mTORC1, partial suppression of mTORC1 by stress could attenuate processes that can negatively affect stress resistance. Remaining mTORC1 activity would contribute to enhance transcriptionally regulated stress protection programs. (C) By contrast, in cells unable to suppress mTOR activity, such as TSC mutants or cells overexpressing Rheb, the combination of stress with the inability to attenuate energy expenditure and the overloading of repair or disposal of damaged components would worsen sensitivity to stress. Even if some stress adaptation responses could be stimulated by mTORC1, they would not be sufficient to counteract stress. In addition, overactive mTORC1 can enhance the activation of p53 in response to various stressors, driving cells to senescence or death.

Toward Future Directions: A Place for mTOR Complexes in the Map of Stress Responses?

As discussed above, mTORC1 can stimulate a number of stress-activated transcriptional responses, and mTORC2 can also enhance certain stress responses, as shown in yeast and mammalian tumors. In S. cerevisiae, TORC2 enhances cell survival to low amounts of DNA damage by stimulating the activity of the SGK1-related kinases Ypk1 and Ypk2 (157), and suppression of TORC2 in Schizosaccharomyces pombe impairs survival under hydrogen peroxide stress, salt stress, and DNA damage (158160). Furthermore, in a positive feedforward loop with Akt and the atypical RIO kinases RIOK1 and RIOK2, mTORC2 increases survival of mouse glioblastoma cells to the DNA-damaging drugs doxorubicin and temozolomide (161).

Intensity and duration could be important variables in the effect of stress on mTOR activity and its capacity to enhance protective responses. A given quantity of stress is likely to have a different impact on cellular functions and elicit different responses depending on a cell’s state of biosynthesis and growth; for example, cells undergoing active growth and intense metabolic activity may benefit from the ability of mTOR to enhance stress survival responses. In this regard, it is interesting that different types of tumors often exhibit augmented mTOR signaling (162) together with enhanced activity of some stress response factors like HIF-1 (163165), Nrf2 (92, 166), or HSF (101, 103, 104). Also, mTOR hypomorphic mice have a lower incidence of spontaneous malignant tumors (167). Although this does not imply that stronger mTOR signaling necessarily enhances anti-stress capabilities in tumors, the coexistence of elevated activities of mTOR and various stress-responsive transcription factors in different tumors can provide an interesting scenario to study how mTOR signaling may contribute to cellular resilience—or fragility—under stress. These studies might also reveal an opportunistic weakness in tumors that are over-reliant on sustained mTOR activity to survive stresses.

Elucidating how mTOR complexes influence the balance between resistance and enhanced sensitivity to stress might also aid understanding their role in aging. Reduction of mTOR activity by pharmacological or genetic means increases life span in yeast, worms, flies, and mice, an observation that opens exciting challenges regarding the possibility of slowing aging by manipulating mTOR (6, 168171). The precise mechanisms underlying the longevity phenotype in mice with reduced mTOR activity are still being elucidated, but it is already apparent that some processes benefit more than others from lower mTOR activity (168). Mice with low mTOR expression throughout their entire life, and therefore reduced activity of both mTORC1 and mTORC2, show improved neuromotor and cognitive performance and a lower incidence of malignant tumors at old age (167). Decreasing mTOR activity in old mice with rapamycin also improves self-renewal capacity of HSCs (172), which agrees with the finding that overactive mTORC1 impairs HSC renewal (96, 154). However, complete lack of mTORC1 impairs HSC regeneration (173). It is also important to note that mTOR-null organisms are not viable (174, 175), and that decreasing global mTOR activity below a certain threshold can severely weaken immune defenses against infections (167), an observation that underscores extensive evidence of a fundamental role of mTOR complexes in immune responses (176180).

Prolonging longevity in an organism requires a good coordination between extending the life span of terminally differentiated cells, maintaining their proper function in tissues, and ensuring the replicative and differentiation potential of progenitor cells. These processes need not be equally sensitive to changes in mTOR activity. Because cells in any organism will endure different stressors to varying degrees throughout their life, it is possible that cellular longevity may be influenced by the ability of mTOR complexes to regulate sensitivity and resistance to stress. Reduced mTORC1 activity could attenuate cellular stress under aging-promoting conditions (such as DNA damage, perturbed proteostasis, or ROS) while simultaneously being sufficient to enhance specific stress adaptation responses. However, this is but one among various possibilities and will need to be experimentally tested.

Several conceptual and experimental questions arise regarding the relevance of mTOR-mediated regulation of stress responses and the mechanisms involved. To know whether mTOR activity is predictive of a better or worse outcome under stress, it will be important to determine three things: (i) how cells quantitate stress and balance this information against mTOR stimulatory inputs; (ii) which mTOR functions are more and less sensitive to defined stress conditions of varying intensity and duration; and (iii) how important it is for the cell to enhance specific response mechanisms in an mTOR-dependent manner to survive stress.

Identifying genes and proteins that are sensitive to varying degrees of stimulation or inhibition of mTOR activity in different stress scenarios and different cell types is technically feasible with current transcriptomic, genomic, metabolomics, and proteomic tools (11, 15, 67, 181183). It is also possible to identify and quantitate changes in posttranslational modifications and the abundance of individual proteins associated with stress responses that are regulated by mTOR (184). However, dissecting how mTOR’s contribution to stress responses may affect a cell’s behavior during and after stress has subsided poses several challenges. Manipulating mTOR-sensitive stress-responsive genes or proteins could provide clues about how changes in their expression or activity would influence cellular behavior under stress, but this could be difficult if the number of genes and proteins involved were relatively large, thus complicating conventional approaches that silence or overexpress candidate genes. However, this task could be simplified if mTOR-sensitive stress response networks could be delineated, and their key nodes identified. Characterization of functional nodes and links in these networks would enable the manipulation of key elements to dissect the contribution of specific pathways in mTOR-regulated stress responses. Also, devising tools that allowed reversible and quantitative modulation of mTOR activity at different stages along the stress response could be useful to understand how mTOR may influence cellular functions during and after stress. Alternating repression and activation of mTOR in a time-controlled manner could be accomplished by combining current chemical inhibitors, such as rapamycin or Torin1 (26), with inducible mTOR mutants engineered to be resistant to those inhibitors (Fig. 3). Rapamycin-resistant mTOR mutants already exist (185, 186), and the available structure of Torin1-bound mTOR (187) should facilitate the design of Torin1-resistant mutants.

Fig. 3 Schematic outline of approaches to address mTOR function in stress responses.

Potentially, mTOR could be reversibly and quantifiably manipulated in cells engineered to express inducible recombinant mTOR mutants resistant to either rapamycin or Torin1. Endogenous mTOR complexes could be inhibited with rapamycin or Torin1, and resistant mutants (represented with a red dot) could be either expressed from an inducible promoter or activated by release from a repressor (for instance, estrogen receptor) in a time-controlled manner. These approaches could use wild-type or TSC1/2-deficient cells, and one could design ectopic mTOR mutants that are capable of interacting only with Raptor or Rictor, to selectively reactivate mTORC1 or mTORC2 functions.

Concluding Remarks

The activity of mTOR complexes is sensitive to diverse forms of stress, which in general means that cells under stress often suppress mTOR signaling. At least in the case of mTORC1, attenuating its activity can improve cell survival under certain stress conditions, whereas abnormally increased mTORC1 function during stress can be detrimental for the cell. On the other hand, both mTORC1 and mTORC2 can also enhance cellular responses that support adaptation and survival to stress, and stress-activated transcriptional regulators such as HIF-1α, HSF1, NFAT5, p53, and FOXO are stimulated by mTORC1. Different findings suggest that the communication between mTOR and stress responses may play an important role in cellular functions, tumor biology, and perhaps long-term fitness of the organism. Aside from identifying relevant mTOR targets in stress response pathways and the regulatory mechanisms involved, it will be important to elucidate how these are regulated by the balance between mTORC1 and mTORC2 activities. It will also be necessary to dissect the role of mTOR complexes in stress separately from their other functions, and to determine how the contribution of mTOR to stress responses may affect longer-term outcomes in cell survival, proliferative capacity, maintenance of specialized functions, differentiation potential, or longevity.

REFERENCES AND NOTES

Acknowledgments: We thank the members of the J.A. and C.L.-R. groups for continued and inspiring discussions. We also thank the editors and reviewers for insight and helpful comments. Our apologies go to those authors whose work was not cited because of our oversight. Funding: J.A. and C.L.-R. are supported by grants from the Spanish Ministry of Economy and Competitiveness (SAF2011-24268 to J.A., SAF2012-36535 to C.L.-R.), by Fundació la Marató TV3 (122530), and by Generalitat de Catalunya (2009 SGR601, 2014 SGR1153). S.T. is supported by a predoctoral fellowship BES-2013-062670. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article