Cellular stresses can induce a wide range of biological responses, depending on the type of stress, the type of cell and the cellular environment. Stress-mediated changes in translational output cover a broad spectrum of potential responses, including an overall decrease in translation or an increase in the translation of specific mRNAs. Many of these changes involve post-translational modifications of components of the translational machinery. The mTOR (mammalian target of rapamycin) pathway is a critical regulator of growth and translation in response to a wide variety of signals, including growth factors, amino acids and energy availability. Through its kinase activity, mTOR activation results in the phosphorylation of translational components and an increase in translation. As stress-mediated changes in translational output are context-dependent, the interplay between stress and mTOR in the control of translation is also likely to depend on factors such as the strength and type of incident stress. In the present paper, we review mTOR-dependent and -independent translational responses, and discuss their regulation by stress.
mTOR (mammalian target of rapamycin)-dependent translation
Regulation of mTOR by growth factors and amino acids
mTOR is involved in two mutually exclusive complexes, mTORC1 and mTORC2. mTORC2 phosphorylates and activates Akt, whereas mTORC1 activates translation and controls cell growth . mTORC1, the complex involved in translation, is rapamycin-sensitive and defined by the presence of raptor (regulatory associated protein of mTOR) [2,3]. This review will discuss mTOR in the context of mTORC1. mTOR is activated by growth factors and nutrients, and as such is able to promote growth and translation in response to pro-growth stimuli . In response to growth factors, activation of Akt leads to the activation of mTOR through numerous overlapping mechanisms. Akt phosphorylates and inactivates TSC (tuberous sclerosis complex) 2, part of a heterodimeric GAP (GTPase-activating protein) for the small GTPase Rheb [4–6]. Since Rheb positively controls mTOR activation, the inactivation of TSC2 results in an increase in mTOR activity. Akt also activates mTOR through the phosphorylation of PRAS40 (proline-rich Akt substrate of 40 kDa), an mTOR inhibitor [7–11]. Phosphorylation of PRAS40 in response to growth factors leads to a disruption of the PRAS40–mTOR interaction and an increase in mTOR kinase activity [9,10].
mTOR is also activated by amino acids. This activation requires Rheb, but does not act through the TSC1–TSC2 complex . The mechanism through which amino acids activate mTOR remain unclear. Whereas Rheb is required for this activation, the activation status of Rheb itself is not altered in all cell types in response to amino acids . The greatest fraction of Rheb in most cell types is in its active GTP-bound form; despite this, amino acid starvation and stimulation affect mTOR activity. Previous studies have suggested that amino acids signal to mTOR through the endosome-associated phosphatidylinositol kinase Vps34 (vacuolar protein sorting 34) [14,15]. However, in Drosophila, Vps34 is involved in TOR (target of rapamycin)-induced autophagy, but is not required for TOR activation in response to amino acid modulation . Vps34 may therefore be required for mTOR activation, but is not part of the signalling machinery. Similarly, MAP4K3 (mitogen-activated protein kinase kinase kinase kinase 3), a kinase proposed to be involved in the activation of JNK (c-Jun N-terminal kinase), is activated by amino acids . Again, the signalling pathways downstream of MAP4K3 are not clear, and there is no described mechanism through which MAP4K3 could activate mTOR.
Regulation of translation downstream of mTOR
The regulation of translation by mTOR occurs largely through the mTOR-mediated phosphorylation of S6K (S6 kinase) and the eIF (eukaryotic initiation factor) 4E-binding protein, 4E-BP1 . In unstimulated cells, 4E-BP1 is bound to eIF4E. Phosphorylation of 4E-BP1 by mTOR prevents the interaction between 4E-BP1 and eIF4E, thereby releasing eIF4E to interact with eIF4G . This eIF4E–eIF4G complex binds to the 7-methylguanosine cap of mRNA and recruits additional components of the translational initiation complex, including eIF4B, to the cap. The rate-limiting step in this assembly is the formation of the eIF4E–eIF4G complex. mTOR therefore controls a critical step in the formation of the translational initiation complex through the phosphorylation of 4E-BP1.
mTOR also phosphorylates S6K. Phosphorylation of S6K by mTOR increases its kinase activity, thereby increasing the phosphorylation of the ribosomal protein S6. The functional consequence of S6 phosphorylation remains somewhat obscure; data from mice in which the phosphorylated residues on S6 are mutated to alanine suggest that S6 phosphorylation does not increase either the general rate of translation or the translation of 5′-TOP (terminal oligopyrimidine tract)-containing mRNAs, the subset of mRNAs encoding translational components . Despite this, cells from these mice are small. S6 phosphorylation is therefore an important mechanism of growth control, albeit through currently undefined mechanisms. In addition to S6, S6K phosphorylates a number of other targets, including eIF4B and eEF2k (eukaryotic elongation factor 2 kinase) [21–23]. The phosphorylation of eIF4B by S6K increases the association of eIF4B with the initiation complex, thereby contributing to the efficiency of translational initiation . S6K-mediated phosphorylation of eEF2k acts further along the translational process, at the level of translational elongation. S6K-mediated phosphorylation of eEF2k inhibits its kinase activity, thus decreasing the phosphorylation of eEF2 . This allows eEF2 to bind to ribosomes and promote translational elongation.
mTOR therefore controls translation at two steps: it controls the initiation of translation through the phosphorylation of 4E-BP1 and eIF4B as well as controlling the rate of translational elongation through the phosphorylation of eEF2k.
Regulation of translational initiation by ERK (extracellular-signal-regulated kinase)
In addition to mTOR, there are numerous other pathways which modify growth and translation in response to environmental stimuli. The ERK/MAPK (mitogen-activated protein kinase) pathway is able to control translation at a number of different levels, through both mTOR-dependent and -independent mechanisms. ERK phosphorylates and inactivates TSC2, leading to increased mTOR activity [24,25]. ERK can also activate mTOR through a second independent mechanism. One of the downstream ERK targets, p90RSK (p90 ribosomal S6 kinase), phosphorylates raptor . Raptor is a key component of mTORC1, and acts as a scaffolding protein to bring mTOR substrates into close proximity to mTOR. Phosphorylation of raptor by p90RSK increases the kinase activity of mTOR. Finally, p90RSK, as a member of the S6 kinase family, can phosphorylate S6 on Ser235 and Ser236 . Whereas S6K phosphorylates five residues, including both Ser235 and Ser236, the phosphorylation of Ser235 and Ser236 alone by p90RSK can still increase translation.
Although the interaction between 4E-BP1 and eIF4E is controlled by the phosphorylation of 4E-BP1, phosphorylation of eIF4E itself can also affect the ability of eIF4E to bind to capped mRNA. MNK (MAPK-interacting kinase; Lk6 in Drosophila) acts downstream of ERK and phosphorylates eIF4E directly [28–30]. In Drosophila, the result of this phosphorylation event appears to be dependent on nutritional conditions . Lk6-deficient Drosophila are small only when raised on low-nutrient food. One possible explanation is that the low level of 4E-BP1 phosphorylation that occurs when mTOR is inactive in such nutrient-poor conditions renders translational output sensitive to eIF4E phosphorylation status, whereas, under high-nutrient conditions, this phosphorylation is less important. eIF4E therefore serves as an integration point for numerous signals controlling growth and translation.
Like eIF4E, eEF2k serves as a hub for integrating the capacity and desirability for growth from multiple signalling networks. eEF2k is a negative regulator of translation; phosphorylation of eEF2k itself can be either stimulatory or inhibitory. Kinases such as AMPK (AMP-activated protein kinase) (see below) phosphorylate and activate eEF2k (thereby decreasing translation), whereas kinases such as S6K and mTOR phosphorylate and inactivate eEF2k. ERK phosphorylates eEF2k at Ser366, the same site that is phosphorylated by S6K. Thus, in a similar manner to mTOR signalling, ERK/MAPK signalling can activate translation at the levels of both initiation and elongation. Interestingly, the cell-cycle-regulated kinase Cdc2 (cell division cycle 2 kinase)/Cdk1 (cyclin-dependent kinase 1) has also recently been shown to phosphorylate and inhibit eEF2k [30a]. This phosphorylation event is postulated to enable protein synthesis to be maintained in mitotic cells. eEF2k is a key regulatory mechanism in protein translation, as reflected by the range of signalling pathways controlling its activity.
Stress and translation
Cellular stresses come in a number of different forms. Stresses such as DNA damage induce a different combination of cellular responses from stresses such as oxidative damage, despite activating an overlapping set of signalling pathways. Translational control is a key regulatory mechanism in an increasing number of cellular responses, including the stress response . The effects of stress on translation have been reported to be either activating or inhibitory. For example, hypoxic stress inhibits growth in many cell types [32,33], but indeed induces growth in vascular endothelial cells . Some cells need to conserve energy upon exposure to stress, to reduce resource consumption in order to survive stressful conditions. In other situations, an incident stress must increase translation in order for the cell to respond appropriately and combat the negative effects of the stress. In these situations, stress should increase translation, either through a general up-regulation of translation or through the increased translation of specific mRNAs. This is particularly prevalent in the immune system, where translation is modified to promote the production of inflammatory cytokines.
Translation is an energy-consuming process. Concentrations of ATP and oxygen, two important molecules that are required for numerous biosynthetic processes, signal their availability through stress pathways. Both oxygen and ATP must be present in order to provide the energy necessary for high rates of translation. One of the most well-described inactivators of translation is hypoxia. Hypoxia induces the transcription of REDD1 (regulated in development and DNA damage responses 1), REDD1 activates TSC2 by disrupting the interaction between TSC2 and 14-3-3 protein, thereby releasing TSC2 . Significantly, modulation of the REDD1 functional homologues in Drosophila (Scylla and Charybdis) affects both cell size and sensitivity to hypoxia .
Energy is stored in cells in the form of ATP. A decrease in available energy leads to a decrease in ATP and a corresponding increase in AMP. AMPK transmits this signal to mTOR. There are at least two mechanisms through which this occurs: an activating phosphorylation of TSC2 and an inhibitory phosphorylation of raptor [35,36]. Each of these phosphorylation events decreases mTOR activity, thereby reducing translation.
The p38 family is a group of stress-responsive kinases . These kinases become active in response to a number of external stresses, including osmotic shock, heat and UV irradiation. There are four mammalian p38 isoforms: α, β, γ and δ. p38α and p38β have similar substrate specificities and are inhibited by traditional p38 inhibitors such as SB202190. p38γ and p38δ have restricted expression patterns and slightly different substrate preferences, and are not inhibited by pyridinyl imidazole compounds such as SB202190 .
p38α was originally identified through its role in inducing pro-inflammatory cytokines such as IL-1 (interleukin-1) and TNFα (tumour necrosis factor α) in response to the bacterial cell wall component LPS (lipopolysaccharide) . In response to LPS stimulation, macrophages and other professional APCs (antigen-presenting cells) secrete IL-1 and TNFα. This recruits effector T-cells to the site of infection. p38 is also required for the production of cytokines such as interferon-γ in response to T-cell receptor engagement . Cytokine production in both APCs and T-cells is controlled by p38 at both the transcriptional and translational levels. p38 phosphorylates a number of transcription factors directly, leading to increased transcription of cytokine mRNAs. Many cytokine mRNAs contain 3′ AREs (AU-rich elements). Under normal conditions, ARE-containing mRNAs are degraded. However, stabilization of ARE-containing mRNAs occurs in response to p38 activation and plays a significant role in increased cytokine production . There may be additional mechanisms through which cytokine production is increased in response to p38 activation. For example, increased eIF4E phosphorylation is observed in a p38-dependent manner in response to murine coronavirus infection . This phosphorylation is concurrent with the phosphorylation of MNK, raising the possibility that MNK can also be activated by p38.
Interestingly, p38δ also phosphorylates and inactivates eEF2k . p38δ has a restricted expression pattern, but can be found in CD4+ T-cells . Interestingly, this p38δ-dependent phosphorylation of eEF2k can be found in keratinocytes, suggesting that non-immune cells are capable of up-regulating translation in response to stress . Thus there may be numerous mechanisms, both transcriptional and translational, through which p38 can induce the expression of cytokines. p38 has described roles in the phosphorylation of the translational initiation components eIF4E and 4E-BP1, as well as in translational elongation through the phosphorylation of eEF2k.
Context-dependent translational activation by stresses
Numerous growth factors, including known mTOR activators such as insulin and EGF (epidermal growth factor), induce the production of ROS (reactive oxygen species). As a by-product of mitochondrial respiration, ROS are also produced by metabolically active cells. Interestingly, chemical disruption of mitochondrial energetics (and therefore decreased ROS production) results in the dephosphorylation of mTOR targets; this can be reversed by simultaneously treating cells with oxidizing compounds . Consistently, in hepatic stellate cells, ROS are required for mTOR activation in response to amino acid treatment . Mitochondrial respiration results in the production of ROS; ROS may therefore act as a sensor for mitochondrial capacity. In this context, activation of mTOR by ROS may be a mechanism through which a cell is able to communicate its ability to generate energy for translation. Conflictingly, H2O2, which induces ROS formation, has been shown to reduce mTOR activity in HEK (human embryonic kidney)-293 cells . ROS can therefore activate or inactivate mTOR. The effects of ROS on mTOR activity may depend on the ROS source and levels, the cell type and micro-environment, or on pathways simultaneously activated by the incident stress.
The relationship between translation, cell growth and cell survival is complex. mTOR has consistently been shown to be activated in response to UV-induced DNA damage [45–47]. When mTOR is activated by UV, this induces the phosphorylation of 4E-BP1 as predicted, increasing translation . This 4E-BP1 phosphorylation, however, does not correlate to cell survival. Increased translation and cell growth are similarly distinct; mutation of the phosphorylation sites on S6 results in small cells, whereas overall translation is not decreased . Increased translation is therefore not necessarily synonymous with either increased cell growth or increased survival. Indeed, in cells with limited resources, increased translation might exhaust the energy supply, leading to cell death. The relationship between translation and cell survival, and between translation and cell growth, is not simply cause and effect.
The dual role of stresses such as ROS in the activation and inactivation of translation is reminiscent of the tumour suppressor p53. p53, as the guardian of the genome, is involved both in repairing DNA damage and in inducing apoptosis. Low levels of DNA damage lead to p53-induced cell-cycle arrest, allowing the cell to repair damaged DNA. High levels of DNA damage do not induce repair, but instead promote apoptosis . The p53-dependent response of a cell to DNA damage therefore depends on the amount of incident stress. In a similar vein, stress-induced signalling pathways may be able to either promote or inhibit translation in response to different amounts or types of incident stresses.
mTOR Signalling, Nutrients and Disease: Biochemical Society Focused Meeting held at Medical Sciences Teaching Centre, University of Oxford, U.K., 15–16 September 2008. Organized and Edited by Richard Boyd (Oxford, U.K.), Deborah Goberdhan (Oxford, U.K.) and Richard Lamb (Cancer Research UK, London, U.K.).
AMP-activated protein kinase
eukaryotic elongation factor 2 kinase
eukaryotic initiation factor
eIF4E-binding protein 1
mitogen-activated protein kinase kinase kinase kinase 3
mitogen-activated protein kinase
mammalian target of rapamycin
p90 ribosomal S6 kinase
proline-rich Akt substrate of 40 kDa
regulatory associated protein of mTOR
regulated in development and DNA damage responses 1
reactive oxygen species
tumour necrosis factor α
target of rapamycin
tuberous sclerosis complex
vacuolar protein sorting 34
We thank Caroline Treins and David Hancock for a critical reading of this manuscript.
Work is funded by Cancer Research UK.