Regulation of mRNA Translation by Signaling Pathways

Abstract

mRNA translation is the most energy consuming process in the cell. In addition, it plays a pivotal role in the control of gene expression and is therefore tightly regulated. In response to various extracellular stimuli and intracellular cues, signaling pathways induce quantitative and qualitative changes in mRNA translation by modulating the phosphorylation status and thus the activity of components of the translational machinery. In this work we focus on the phosphoinositide 3-kinase (PI3K)/AKT and the mitogen-activated protein kinase (MAPK) pathways, as they are strongly implicated in the regulation of translation in homeostasis, whereas their malfunction has been linked to aberrant translation in human diseases, including cancer.

mRNA translation may be the most energy-demanding cellular process. Therefore, it is tightly regulated by signaling pathways (e.g., TOR and MAPK) that induce quantitative and qualitative changes in translation.

Translation plays an evolutionarily conserved role in the regulation of gene expression (Mathews et al. 2007). In fact, recent findings indicate that translation plays a major part in determining protein expression levels in mammalian cells (Schwanhausser et al. 2011). Amperometric oxygen consumption measurements in rat thymocytes revealed that translation consumes ∼20% of cellular ATP, and is thus thought to be the most energy demanding cellular process (Buttgereit and Brand 1995). Therefore, it is not surprising that translation represents a tightly regulated cellular process, dysregulation of which contributes to diverse human diseases, including cancer (see Silvera et al. 2010; Braun et al. 2012).

Translation occurs in three distinct steps: initiation, elongation, and termination (Sonenberg et al. 2012). Although all of the steps are highly regulated, most of the translational control occurs at the rate-limiting initiation step, during which mRNAs and tRNA^Met_i are recruited to the ribosome through the highly orchestrated action of translation initiation factors (IFs) (reviewed by Sonenberg and Hinnebusch 2009). Various stimuli including environmental stress (e.g., heat-shock, UV irradiation), extracellular stimuli (e.g., nutrients, hormones, growth factors), and intracellular cues (e.g., energy status of the cell, intracellular availability of amino acids) (Wek et al. 2006; Ma and Blenis 2009; Sonenberg and Hinnebusch 2009) induce dramatic qualitative and quantitative changes in the translatome, i.e., the pools of cellular mRNAs that are being actively translated (Greenbaum et al. 2001). This is achieved in part via the phosphorylation of eukaryotic translation initiation factors (eIFs) and proteins that regulate their activity (e.g., 4E-BPs, PDCD4) by signaling pathways (reviewed by Sonenberg and Hinnebusch 2009; Jackson et al. 2010).

Here we summarize current knowledge on the role of cellular signaling pathways in translational control. In particular, we portray the role of the target of rapamycin (TOR) and the mitogen-activated protein kinase (MAPK) pathways in the regulation of translation, because these pathways regulate the phosphorylation and function of a multitude of eIFs and associated factors. Another important signaling node in translation involves the eIF2α kinases, discussed by Benham (2012).

TOR

TOR is an evolutionarily conserved Ser/Thr kinase, which regulates proliferation (increase in cell number) and growth (increase in volume/mass) in response to cellular energy status, growth factors, hormones, and nutrient availability (Wullschleger et al. 2006). TOR exists in two functionally and structurally distinct protein complexes referred to as TOR complex 1 and 2 (TORC1 and 2) (Wullschleger et al. 2006). In mammalian cells, mechanistic/mammalian TORC1 (mTORC1) consists of the catalytic component mTOR, the scaffolding protein Raptor (regulatory-associated protein of TOR, which is orthologous to KOG1 in yeast), the GTPase β-subunit like protein gβL (also known as mLST8), proline-rich AKT substrate of 40 kDa (PRAS40) and Deptor (disheveled, Egl-10, pleckstrin [DEP] domain containing mTOR interacting protein) (Fig. 1) (Guertin and Sabatini 2007; Peterson et al. 2009). mTOR, gβL, and Deptor are also found in mTORC2 (Guertin and Sabatini 2007; Peterson et al. 2009). In turn, Rictor (rapamycin-insensitive companion of TOR, which is orthologous to AVO3 in yeast), mSIN1 (mammalian stress-activated protein kinase (SAPK)-interacting protein), and PRR5 (Proline-rich protein 5, also known as Protor) (Frias et al. 2006; Jacinto et al. 2006; Pearce et al. 2007; Thedieck et al. 2007; Woo et al. 2007) are found exclusively in mTORC2 (Fig. 2). mTORC1 and mTORC2 regulate disparate cellular functions by phosphorylating distinct sets of substrates. Several substrates of mTORC1 have been identified in the past two decades including the eIF4E-binding proteins (4E-BPs), 70 kDa ribosomal S6 kinases 1 and 2 (S6Ks), PRAS40, Ser/Thr kinase Ulk1 (also known as hATG1), and growth factor receptor-bound protein 10 (Grb10) (reviewed by Caron et al. 2010; Yea and Fruman 2011; Zoncu et al. 2011b). By modulating their activity, mTORC1 regulates a variety of cellular processes including growth, proliferation, translation, autophagy, as well as its own activation (Zoncu et al. 2011b). The function, upstream regulators, and associated substrates of mTORC2 are less well understood (Oh and Jacinto 2011). mTORC2 phosphorylates AGC kinase family members, e.g., AKT, protein kinase C (PKC) and serum/glucocorticoid regulated kinase 1 (SGK1) and thereby controls cytoskeletal organization and cell survival (Sarbassov et al. 2004; Sarbassov et al. 2005; Guertin and Sabatini 2007; Garcia-Martinez and Alessi 2008). mTORC2 also associates with the ribosome (Zinzalla et al. 2011) where it phosphorylates residues in nascent polypeptide chains that are important for optimal protein folding (Fig. 2) (Oh et al. 2010).

Figure 1.

Schematic representation of mTORC1 signaling to the translational machinery. Growth factors and hormones stimulate mTORC1 by activating receptor tyrosine kinase (RTK) signaling via PI3K/AKT (light blue) and Ras/ERK (green) pathways. mTORC1 is also activated by amino acids via small Rag GTPases (aquamarine). Insufficient energy resources and hypoxia inactivate mTORC1 via the LKB1/AMPK pathway (purple) and REDD1 (brown), respectively. TSC1/2 (red) suppresses mTORC1 signaling by inhibiting GTPase activity of Rheb (red). mTORC1 modulates translation via the phosphorylation of downstream targets including 4E-BPs (olive), S6Ks (champagne), and their downstream effectors. In addition, mTORC1 stimulates ribosome biogenesis and tRNA synthesis by activating TIF-IA (gray) and inhibiting Maf1 (gray), respectively. mTOR, Deptor, and gβL (yellow) are found in both mTORC1 and mTORC2, whereas PRAS40 and Raptor are specific components of mTORC1 (pink). T bars represent inhibitory signals, whereas arrows indicate stimulatory signals. Abbreviations and detailed explanations are provided in the text.

Figure 2.

Schematic representation of mTORC2 signaling pathway. Cellular functions of mTORC2 are described less well than those of mTORC1. Protor, Rictor, and mSin1 (orange) are specific components of mTORC2. mTORC2 stimulates the activity of AGC kinase family members (blue) by phosphorylating residues localized in their hydrophobic motifs. Phosphorylation of PKC, AKT, and SGK1 by mTORC2 is thought to influence cytoskeletal organization and cell survival, respectively. In addition, it was recently reported that mTORC2 associates with the ribosome. It appears that ribosome-localized mTORC2 cotranslationally phosphorylates nascent polypeptide chains as they emerge from the ribosome thus facilitating their optimal folding. Upstream regulators of mTORC2 are still largely elusive. T bars represent inhibitory signals, whereas arrows indicate stimulatory signals. Abbreviations and detailed explanations are provided in the text.

mTORC1 and mTORC2 exhibit differential sensitivity to rapamycin. Rapamycin is a naturally occurring allosteric inhibitor of mTORC1 (Hay and Sonenberg 2004; Petroulakis et al. 2006; Guertin and Sabatini 2009). It forms a complex with its intracellular receptor, FKBP12 (FK506-binding protein of 12 kDa), which binds to the FRB (FKBP12-rapamycin binding) domain of mTOR and inhibits mTORC1 function (Chen et al. 1995). Although the precise mechanism of inhibition remains incompletely defined, rapamycin:FKBP12 binding to the FRB domain was shown to weaken the mTOR:Raptor interaction and partially reduce intrinsic mTORC1 catalytic activity, as measured by mTOR autophosphorylation at Ser2481 (Soliman et al. 2010; Yip et al. 2010). Recent studies, however, revealed that the inhibition of mTORC1 by rapamycin is incomplete, inasmuch as it efficiently suppresses the phosphorylation of S6Ks, but not of 4E-BPs (Choo et al. 2008). In contrast to mTORC1, mTORC2 appears to be largely insensitive to the effects of rapamycin in acute treatment. However, it was proposed that prolonged rapamycin treatment results in inhibition of mTORC2 signaling in a cell type-specific manner, likely by blunting its de novo assembly (Sarbassov et al. 2006).

mTORC1 plays a central role in the regulation of cell growth and proliferation (Fingar and Blenis 2004; Ma and Blenis 2009; Zoncu et al. 2011b), cellular processes that are directly proportional to translational activity (Brooks 1977; Zetterberg et al. 1995). Growth factors, hormones, branched-chain amino acids, and glucose stimulate mTORC1, up-regulate translation, and stimulate cellular growth and proliferation (reviewed by Zoncu et al. 2011b). Conversely, under conditions in which energy production, oxygen supply, and nutrients are inadequate, mTORC1 signaling is down-regulated, resulting in inhibition of translation, reduction in cellular growth proliferation, and induction of autophagy (reviewed by Zoncu et al. 2011b). Autophagy is the major catabolic process in the cell, during which cytoplasmic organelles and macromolecular complexes are degraded to replenish intracellular pools of amino acids (Stipanuk 2009).

Upstream of mTORC1

mTORC1 signaling to the translational apparatus is modulated via a multitude of cellular pathways including phosphoinositide 3-kinase (PI3K), Ras/MAPK, Rag GTPases, and adenosine monophosphate (AMP)-activated protein kinase (AMPK) (Fig. 1) (Hay and Sonenberg 2004; Laplante and Sabatini 2009).

Growth Factors and Hormones

Hormones and growth factors [e.g., insulin and insulin-like growth factors (IGFs)] activate PI3K via receptor tyrosine kinases (RTKs; such as insulin- or IGF-receptor) and associated adaptor molecules (e.g., IRS-1 and -2). PI3K converts phosphatidylinositol 4,5-bisphosphate (PIP₂) into phosphatidylinositol-3,4,5-triphosphate (PIP₃) (reviewed by Cantley 2002; Engelman et al. 2006), which is reversed by the tumor suppressor phosphatase and tensin homolog [PTEN; (Maehama and Dixon 1998)]. PIP₃ binds the pleckstrin homology domains of PDK1 and AKT resulting in their recruitment to the plasma membrane (Cantley 2002; Engelman et al. 2006). The Ser/Thr kinase PDK1 phosphorylates AKT and other AGC-kinases (e.g., PKC, RSK, and S6K) on residues localized in their activation loops (e.g., Thr308 in human AKT1) (reviewed by Pearce et al. 2010). In addition, mTORC2 modulates the activity of AKT by phosphorylating a serine residue in its hydrophobic motif (Ser473 in human AKT1) (Sarbassov et al. 2005). TSC1/2 is comprised of the scaffolding protein TSC1 (hamartin) and the GTPase activating protein (GAP) TSC2 (tuberin) (Kwiatkowski and Manning 2005). AKT phosphorylates TSC2 at multiple sites, which are thought to inhibit its GAP activity, thus reducing Ras homolog enriched in brain (Rheb)-GTP hydrolysis to its inactive GDP-bound form (Manning et al. 2002; Garami et al. 2003; Inoki et al. 2003). Rheb is a small GTPase (Yamagata et al. 1994) that activates mTORC1 through a poorly understood mechanism (Inoki et al. 2003). A single gene coding for Rheb is found in yeast and Drosophila, whereas mammalian cells express two variants of this protein, Rheb1 and Rheb2 (reviewed by Avruch et al. 2006).

PRAS40 is a recently identified suppressor of mTORC1 (Fonseca et al. 2007; Oshiro et al. 2007; Sancak et al. 2007; Vander Haar et al. 2007; Wang et al. 2007). PRAS40 is phosphorylated on Thr246 via PI3K/AKT, which suggests that it acts as an upstream inhibitor of mTORC1 (Sancak et al. 2007; Vander Haar et al. 2007). However, PRAS40 binds Raptor via a TOR signaling (TOS) motif, which is present in other mTORC1 substrates such as 4E-BPs and S6Ks (Schalm and Blenis 2002; Schalm et al. 2003) and is phosphorylated by mTORC1 at multiple sites, including rapamycin-sensitive (Ser183 and Ser221) and -insensitive (Ser212) residues (Oshiro et al. 2007; Wang et al. 2007). These findings suggest a model whereby PRAS40 acts as a downstream target of mTORC1, which competes with 4E-BPs and S6Ks for binding to Raptor (Fig. 1). Accordingly, phosphorylation and release of PRAS40 from mTORC1 increases substrate accessibility but does not increase its kinase activity (Rapley et al. 2011).

Growth factors also stimulate mTORC1 activity through a pathway involving the small Ras GTPases (reviewed by Shaw and Cantley 2006). Loss of the tumor suppressor gene NF1, which encodes a Ras GAP, results in increased mTORC1 signaling and tumorigenesis (Johannessen et al. 2005, 2008). Ras is an oncogene that triggers several signaling cascades, including the MAPK pathway, which consists of the sequential activation of the Raf, MEK, and ERK protein kinases (Rajalingam et al. 2007). At present, it is not clear how ERK1/2 stimulate mTORC1 activity, but TSC2 was shown to be a direct target of ERK1/2 and its downstream substrate, the 90 kDa ribosomal S6 kinase (RSK) (Roux et al. 2004; Ballif et al. 2005; Ma et al. 2005). Both protein kinases also phosphorylate Raptor at regulatory sites, correlating with increased mTORC1 activity and signaling to downstream substrates (Carriere et al. 2008a, 2011). Recent evidence suggests that RSK also phosphorylates Deptor within a degron sequence that is recognized by the E3-ubiquitin ligase SCF^βTrCP, thereby promoting its ubiquitination and proteosomal degradation (Zhao et al. 2011).

Nutrients, Oxygen, and Energy Status in the Cell

mTORC1 controls cellular proliferation and growth in response to nutrient availability and cellular energy balance (Fig. 1). Amino acids, in particular those with a branched side chain, are indispensable for mTORC1 signaling in cell culture (Hara et al. 1998; Wang et al. 1998). In yeast, amino acids stimulate Vam6/VPS39, which loads GTP onto Gtr1/Gtr2 GTPases (Binda et al. 2009). Gtr1/Gtr2 are components of the vacuolar-membrane-associated EGO complex, which binds to and activates TORC1 (Kim and Guan 2011). A similar mechanism has recently been described in mammals, whereby in response to amino acids Rag GTPases recruit mTORC1 to lysosomal membranes via interaction with Raptor (Sancak et al. 2008). This brings mTORC1 into close proximity of Rheb resulting in its activation (Kim et al. 2008; Sancak et al. 2008). Analogous to Gtr1/Gtr2, Rags form heterodimers comprising RagA or RagB bound to RagC or RagD, whose activity is governed by the scaffolding Ragulator complex (Sengupta et al. 2010), which consists of the p18, p14, and MP1 proteins and is required for the anchoring of Rags to the lysosome (Sancak et al. 2010). Vacuolar H⁺-ATPase (v-ATPase) associates with the Ragulator complex and plays an essential role in conducting amino acid signals from the lysosomal lumen to Rags (Zoncu et al. 2011a). In addition, it has been reported that signaling adaptor p62 interacts with Raptor, mediates the interaction of mTOR with Rag GTPases, and stimulates translocation of mTORC1 to the lysosome (Duran et al. 2011).

Glucose deprivation leads to a decrease in glucose flux and cellular ATP levels, thus inhibiting mTORC1 signaling (Fig. 1). Changes in cellular energy balance impact on mTORC1 signaling via AMPK, which is a Ser/Thr kinase consisting of a catalytic α-subunit and two regulatory subunits, β and γ (Kahn et al. 2005; Shaw 2009). AMPK is activated under conditions in which the intracellular AMP/ATP ratio is increased (e.g., lack of nutrients, mitochondrial dysfunction) (Kahn et al. 2005; Shaw 2009). AMP directly associates with the γ-subunit of AMPK, thus facilitating the phosphorylation of the α-subunit on Thr172 by upstream kinases such as Serine/Threonine kinase 11 (STK11/LKB1) (Shaw 2009). On activation, AMPK suppresses anabolic processes including protein synthesis and restricts proliferation and growth (Kahn et al. 2005; Shaw 2009). This is partly achieved via inhibition of mTORC1 (Shaw et al. 2004). AMPK inhibits mTORC1 by phosphorylating and activating TSC2 (Corradetti et al. 2004) and by phosphorylating Raptor resulting in its sequestration by 14-3-3 proteins (Gwinn et al. 2008). Inhibition of mTORC1 signaling appears to be required for the block in proliferation exerted by AMPK (Gwinn et al. 2008), which is paralleled by a decrease in global translation (Dowling et al. 2007). These results suggest that the AMPK/mTOR pathway connects the regulation of intracellular energy balance with translational control and cell proliferation.

Under conditions where the supply of oxygen is limited, mTORC1 signaling is suppressed via multiple mechanisms (Fig. 1). In addition to inhibiting oxidative phosphorylation and, as a result, activating AMPK, hypoxia represses mTORC1 signaling through regulated in development and DNA damage response 1 (REDD1) (Brugarolas et al. 2004). REDD1 expression is stimulated by several cellular insults (Ellisen et al. 2002) and was shown to inhibit mTORC1 by stabilizing the TSC1/2 complex (DeYoung et al. 2008). Additional inhibitory mechanisms were described to occur during hypoxia, including the direct interaction and inhibition of Rheb by the hypoxia-inducible proapoptotic protein BNIP3 (BCl2/adenovirus E1B 19 kDa protein-interacting protein 3) (Li et al. 2007).

It is worthwhile to note that because of nonphysiological conditions used in the aforementioned in vitro studies (i.e., glucose and amino acid deprivation followed by acute refeeding), the understanding of how mTORC1 is regulated by nutrients and alterations in the energy balance in vivo is still incomplete. Indeed, recent studies suggest that the regulation of mTORC1 signaling by nutrients at the organismal level is more complex than initially thought (reviewed by Howell and Manning 2011).

Emerging Mechanisms of mTORC1 Regulation

mTOR activity is regulated by the phosphorylation of residues located within its kinase domain (Ser2159 and Thr2164 in human mTOR) (Ekim et al. 2011). Phosphorylation of these residues stimulates mTOR autophosphorylation (on Ser2481 in human protein) (Soliman et al. 2010) and is required for its effects on cell growth and proliferation (Ekim et al. 2011). The identity of the mTOR Ser2159 andThr2164 kinase(s) remains unknown. Phosphorylation of mTOR at Ser1261, which lies in a centrally located HEAT (Huntington, Elongation Factor 3, PR65/A, TOR)-repeat, also promotes mTORC1 signaling, mTOR autophosphorylation at Ser2461 and cell growth (Acosta-Jaquez et al. 2009). mTOR also phosphorylates Raptor on several sites (including Ser863 in human Raptor) that, in turn, increases mTOR activity toward downstream substrates (Foster et al. 2010). Interestingly, ERK1/2 were also shown to phosphorylate some of these sites (Carriere et al. 2011), but the physiological relevance of this additional layer of regulation remains unknown.

mTORC1 Signaling to the Translational Machinery

mTORC1 stimulates global protein synthesis, as well as translation of a specific subset of mRNAs (Fig. 3). 4E-BPs and S6Ks are the most extensively studied and best-understood downstream effectors of mTORC1, which have been implicated in the regulation of translation (Hay and Sonenberg 2004).

Figure 3.

mTORC1 and MAPK pathways modulate translation of a specific subset of mRNAs by stimulating eIF4E activity. (A) Hierarchical phosphorylation of 4E-BPs (green) by mTORC1 (broken arrows) leads to their dissociation from eIF4E, thereby stimulating the interaction of eIF4E with eIF4G and the assembly of the eIF4F complex (red). eIF4E is the most limiting subunit of the eIF4F complex, which is critical for the recruitment of eIF4A to the 5′UTR of mRNA and unwinding of the secondary structure during scanning of the ribosome toward the initiation codon (AUG; red arrow). MAPK-interacting kinases 1 and 2 (MNK1/2) (blue) are recruited to eIF4E via eIF4G and phosphorylate eIF4E at a single Ser residue (Ser209 in human eIF4E). (B) mRNAs that contain long and highly structured 5′UTRs frequently encode proliferation, survival, and tumor promoting proteins (top panel). Translation of these mRNAs strongly depends on the unwinding activity of eIF4A, and is thus robustly stimulated by the increase in the amount of eIF4E, which is available for eIF4F complex assembly (further explanation provided in the text). In addition, phospho-eIF4E selectively stimulates translation of Mcl-1, MMP, and proinflammatory mRNAs by a hitherto unknown mechanism. In contrast, housekeeping proteins are typically encoded by mRNAs that bear short, unstructured 5′ UTRs (lower panel) and their translational activity is only marginally influenced by the changes in the availability of eIF4E or its phosphorylation status. The eIF4E diagram was generated using PyMOL software (http://www.pymol.org). eIF4E-PDB accession number 1L8B.

4E-BPs

The first step of cap-dependent translation initiation is the assembly of the eIF4F complex on the 5′-mRNA cap structure (Mathews et al. 2007; Sonenberg and Hinnebusch 2009; Jackson et al. 2010; Topisirovic et al. 2011). The eIF4F complex comprises the cap-binding subunit eIF4E, the large scaffolding protein eIF4G, and the DEAD-box RNA helicase eIF4A, which unwinds secondary structure within the 5′-untranslated region (5′UTR) of the mRNA (Fig. 3A) (Gingras et al. 1999b; Sonenberg and Hinnebusch 2009; Jackson et al. 2010). 4E-BPs are small-molecular weight translational repressors (4E-BP1, 2, and 3 in mammals), which interfere with the assembly of the eIF4F complex by competing with eIF4G for binding to eIF4E (Pause et al. 1994). On activation, mTORC1 phosphorylates residues corresponding to Thr37 and Thr46 on human 4E-BP1, which act as priming sites for the phosphorylation of Ser65 and Thr70 (Fig. 3A) (Gingras et al. 1999a, 2001). Phosphorylation of 4E-BPs on these four residues, leads to their dissociation from eIF4E, thus allowing the assembly of the eIF4F complex (Pause et al. 1994; Gingras et al. 1999a, 2001).

eIF4E is a general translation initiation factor required for cap-dependent translation of all cellular mRNAs (reviewed by Sonenberg and Hinnebusch 2009). Nonetheless, it is well established that alterations in eIF4E levels and/or activity affect translation of a specific pool of “eIF4E-sensitive” mRNAs, but do not have a major impact on global protein synthesis (Graff and Zimmer 2003; De Benedetti and Graff 2004; Graff et al. 2008; Sonenberg and Hinnebusch 2009). It is thought that “eIF4E-sensitivity” of mRNAs is determined by the complexity of their 5′UTRs. “eIF4E-sensitive” mRNAs, which frequently encode proliferation and survival promoting proteins (e.g., Bcl-xL, cyclins, ornithine decarboxylase, c-myc, vascular, and endothelial growth factor), possess long and highly structured 5′UTRs and thus are strongly dependent on the unwinding activity of the eIF4A subunit of eIF4F, whereas “eIF4E-insensitive” mRNAs such as those encoding housekeeping proteins (e.g., actins and tubulins) bear short 5′UTRs and are only minimally sensitive to alterations in eIF4F levels (Fig. 3B) (Koromilas et al. 1992; Svitkin et al. 2001; De Benedetti and Graff 2004; Sonenberg 2008). Because eIF4E is the most limiting factor among eukaryotic translation initiation factors, it controls the levels of eIF4F (reviewed by Sonenberg and Hinnebusch 2009). 4E-BPs impede eIF4F complex assembly (Pause et al. 1994; Gingras et al. 1999a, 2001). Correspondingly, alterations in the expression and/or phosphorylation status of 4E-BPs only marginally affect global protein synthesis, while strongly influencing translation of a subset of mRNAs (e.g., IRF-7, Gas2, cyclin D3, ornithine decarboxylase, and vascular and endothelial growth factor) (Lynch et al. 2004; Colina et al. 2008; Petroulakis et al. 2009; Dowling et al. 2010a). Thus, in addition to its effects on global protein synthesis, mTORC1 selectively stimulates translation of “eIF4E-sensitive” mRNAs by phosphorylating and inactivating 4E-BPs. The vast majority of “eIF4E-sensitive” mRNAs encode proliferation, survival, and tumor-promoting proteins, and accordingly, 4E-BPs act as major mediators of the effects of mTORC1 on proliferation (Dowling et al. 2010a). In Drosophila, dS6K and d4E-BP play overlapping roles in regulating cell size and proliferation downstream from dTOR (Montagne et al. 1999; Miron et al. 2001). In contrast, mammalian 4E-BPs do not appear to significantly influence cell size, which is thought to be largely achieved via S6Ks (Pende et al. 2004; Ohanna et al. 2005; Dowling et al. 2010a). The evolutionary advantage of this “division of labor” between 4E-BPs and S6Ks downstream from mTORC1 is still unclear.

S6Ks

In addition to 4E-BPs, TOR regulates translation by activating the S6Ks (Hay and Sonenberg 2004; Ma and Blenis 2009; Dowling et al. 2010b; Zoncu et al. 2011b). Although Drosophila expresses a single S6K protein (dS6K), mammals express two variants of S6K (S6K1 and S6K2; or S6Kα and S6Kβ, respectively), which are encoded by two distinct genes (RPS6KB1 and RPS6KB2) and share a high degree of homology (reviewed by Fenton and Gout 2011). S6K1 and S6K2 exist in two different isoforms (p70 and p85, and p54 and p56, respectively), which are generated via alternative translational initiation sites from a common mRNA (Grove et al. 1991; Gout et al. 1998). p70S6K1 is the more abundant isoform of S6K1 and is predominantly cytoplasmic, whereas p85S6K1, p54S6K2, and p56S6K2 are localized in the nucleus (reviewed by Fenton and Gout 2011).

S6Ks belong to the AGC kinase family and are activated by PDK1 and mTORC1 via the phosphorylation of Thr residues localized in their activation loop (Thr229 in human p70S6K1) and hydrophobic motif (Thr389 in human p70S6K1), respectively (reviewed by Fenton and Gout 2011). Recent findings indicate that GSK3 also contributes to the activation of S6Ks through the phosphorylation of their turn motif (Ser371 in human p70S6K1) (Shin et al. 2011a). Several S6Ks substrates have been implicated in the regulation of translation including ribosomal protein S6 (rpS6) (Banerjee et al. 1990; Kozma et al. 1990), eukaryotic initiation factor 4B (eIF4B) (Raught et al. 2004; Shahbazian et al. 2006), and programmed cell death 4 protein (PDCD4) (Dorrello et al. 2006).

rpS6 was the first identified S6K substrate. Five phosphorylation sites (Ser235, Ser236, Ser240, Ser244, and Ser247 in humans and rodents) are clustered in the carboxyl terminus of rpS6 (Meyuhas 2008). It has been proposed that S6Ks phosphorylate rpS6 in a sequential fashion, whereby the phosphorylation of Ser236 is followed by the phosphorylation of Ser235, Ser240, Ser244, and Ser247 (Krieg et al. 1988; Ferrari et al. 1991; Bandi et al. 1993). Although S6K1 contributes to rpS6 phosphorylation, S6K2 appears to be the predominant kinase that phosphorylates rpS6 on these residues (Meyuhas 2008). In contrast, the RSKs phosphorylate rpS6 only on Ser235 and Ser236 (Pende et al. 2004; Roux et al. 2007), but the physiological relevance for this specificity remains unknown. Experiments obtained using mice in which wild-type rpS6 is replaced by a nonphosphorylatable mutant revealed that the loss of rpS6 phosphorylation mirrors defects in cell growth observed in S6K1/2 knockout mice (Ruvinsky et al. 2005). Notwithstanding the overlap in their physiological roles, the understanding of how S6Ks and rpS6 impact translation remains obscure. Loss of S6Ks only modestly affects global translation rates, whereas the expression of the nonphosphorylatable mutant of the rpS6 results in a moderate up-regulation of overall protein synthesis rates (Pende et al. 2004; Ruvinsky et al. 2005). 5′-terminal oligopyrimidine tract (5′-TOP)-containing mRNAs encode components of the translational machinery and their translation is repressed under conditions where S6K activity and rpS6 phosphorylation are minimal, such as when cells are deprived of amino acids (Levy et al. 1991). Thus, it was proposed that S6Ks promote translation of the 5′-TOP-mRNAs via stimulation of rpS6 phosphorylation (Kawasome et al. 1998; Shima et al. 1998; Loreni et al. 2000). However, it was subsequently shown that neither the loss of S6Ks nor the phosphorylation status of rpS6 influences translation of 5′-TOP mRNAs (Tang et al. 2001; Pende et al. 2004; Ruvinsky et al. 2005).

PDCD4 plays an established role in apoptosis and has been suggested to possess tumor suppressor properties (reviewed by Lankat-Buttgereit and Goke 2009). PDCD4 binds to eIF4A via two conserved MA-3 domains (also found in eIF4G) and competes with eIF4G for eIF4A binding (Goke et al. 2002; Yang et al. 2003). This leads to the inhibition of eIF4A and the consequent repression of cap-dependent translation (Yang et al. 2003). S6Ks and AKT phosphorylate PDCD4 on Ser67 and Ser457 leading to its degradation by the E3-ubiquitin ligase SCF^βTrCP (Dorrello et al. 2006).

eIF4B and eIF4H are two auxiliary factors that stimulate the RNA unwinding activity of eIF4A (Grifo et al. 1984; Rozen et al. 1990; Pause et al. 1994; Richter-Cook et al. 1998; Rogers et al. 2001). eIF4B stimulates cellular proliferation and survival by selectively up-regulating translation of mRNAs such as those encoding Cdc25, ODC, XIAP, and Bcl-2 (Shahbazian et al. 2010). eIF4B is phosphorylated by several AGC kinases on Ser406 (likely by RSK and S6K) and Ser422 (by S6K, AKT, and RSK) in a stimulus- and cell type-dependent manner (Raught et al. 2004; Shahbazian et al. 2006; van Gorp et al. 2009). It was suggested that eIF4B phosphorylation promotes its association with eIF3 and correlates with increased translation initiation (Holz et al. 2005; Shahbazian et al. 2006). eIF3 is the multisubunit complex that was also identified as a dynamic scaffold for mTORC1 and S6K1 binding (Holz et al. 2005). On activation, mTORC1 is recruited to the eIF3 complex where it phosphorylates S6K1, leading to its dissociation from eIF3 and subsequent phosphorylation by PDK1 (Holz et al. 2005).

S6Ks also phosphorylate eukaryotic elongation factor 2 (eEF2) kinase at Ser366 (in human eEF2K). eEF2K impedes translation elongation by phosphorylating and inhibiting eEF2 (Wang et al. 2001a), a GTPase that promotes translocation of the nascent polypeptide chain from the A-site to the P-site of the ribosome (Skogerson and Moldave 1968; Bermek and Matthaei 1971). Finally, S6K1 has been shown to promote translational efficiency of newly spliced mRNAs on its recruitment to the exon-junction complex (EJC) by its substrate and binding partner SKAR (Richardson et al. 2004). Recruitment of S6K1 and SKAR to the EJC leads to the phosphorylation of numerous mRNA binding proteins and correlates with increased translational efficiency of spliced mRNAs (Ma et al. 2008).

Additional mTOR Targets Implicated in Translational Control

Besides 4E-BPs and S6Ks, mTORC1 has been suggested to modulate translational initiation via phosphorylation of eIF4G at multiple residues (Raught et al. 2000). mTORC1 also stimulates ribosome biogenesis and tRNA synthesis, via stimulation of TIF-IA (Mayer et al. 2004) and inhibition of the TF-IIIC repressor Maf1, respectively (Wei et al. 2009; Kantidakis et al. 2010; Michels et al. 2010; Shor et al. 2010).

MAPK SIGNALING TO THE TRANSLATIONAL MACHINERY

The MAPKs are Ser/Thr kinases that are among the most ancient signal transduction pathways and are widely used throughout evolution in many physiological processes (Widmann et al. 1999). All eukaryotic cells possess multiple MAPK pathways, which coordinate gene expression, mitosis, metabolism, motility, survival, apoptosis, and differentiation. In mammals, 14 MAPKs have been characterized, but the most extensively studied groups are the ERK1/2, JNKs, and p38 isoforms (reviewed by Chen et al. 2001; Kyriakis and Avruch 2001; Pearson et al. 2001). The wide range of functions regulated by these MAPKs is mediated through phosphorylation of several substrates, including members of a family of Ser/Thr kinases termed MAPK-activated protein kinase (MAPKAPK) (Gaestel 2006; Gaestel 2008; Cargnello and Roux 2011). Two MAPKAPKs have been directly implicated in the regulation of translation, namely the RSKs (Carriere et al. 2008b) and the MAPK-interacting kinases (MNKs) (Buxade et al. 2008). Although the RSK isoforms are strictly regulated by ERK1/2 downstream from growth factors and mitogens, the MNKs can be activated by either ERK1/2 or p38 isoforms, making them responsive to both mitogenic and stress stimuli (Fig. 4). The RSKs and MNKs become active on phosphorylation of a threonine residue followed by a proline, located in the activation loop of their kinase domain (Roux and Blenis 2004).

Figure 4.

Schematic representation of MAPK signaling to the translational machinery. The Ras/ERK (green) and p38MAPK (champagne) pathways impinge at different levels on the translational machinery. Although Ras/ERK signaling stimulates the activity of both RSK (green) and MNK (orange), the latter is also responsive to agonists of the p38MAPK pathway. MNK interacts with eIF4G and on activation phosphorylates eIF4E (orange) at Ser209, a site that increases its oncogenic potential and facilitates the translation of specific mRNAs. Following stimulation of the Ras/ERK pathway, activated RSK phosphorylates rpS6, eIF4B, and eEF2K (green), which are important translational regulators. RSK also participates in the regulation of mTORC1 by inhibiting TSC2 (red) and Deptor (yellow), which are negative regulators of mTORC1. ERK and RSK regulate LKB1-dependent (purple) and -independent phosphorylation of Raptor (pink), resulting in increased mTORC1 signaling. ERK and RSK also collaborate in the regulation of ribosome biogenesis by promoting TIF-1A phosphorylation (green). T bars represent inhibitory signals, whereas arrows indicate stimulatory signals. Abbreviations and detailed explanations are provided in the text.

MNKs

In addition to its regulation by the 4E-BPs through mTORC1 signaling, the activity of eIF4E is regulated by MNK1 and MNK2, which phosphorylate a single serine residue in its carboxyl terminus (Fig. 4) (Ser209 in human eIF4E) (Fukunaga and Hunter 1997; Waskiewicz et al. 1997). MNK1 and 2 exist in two isoforms (MNK1a and b, and MNK2a and b), which are generated by alternative splicing. MNK1a and MNK2a have long C termini harboring MAPK-binding motifs and show predominantly cytoplasmic localization, whereas MNK1b and MNK2b, which contain short carboxyl termini, are equally distributed between the nucleus and cytoplasm (reviewed by Buxade et al. 2008). In most cell lines, the basal activity of MNK2a is high, which is thought to be a consequence of its ability to sustainably associate with activated ERK (reviewed by Buxade et al. 2008). Conversely, MNK1a has a low basal activity and it is activated by ERK and p38MAPK in response to stimuli such as growth factors and phorbol esters or cytokines and environmental stress, respectively (Scheper et al. 2001; Wang et al. 2001a). MNK1b exhibits high basal activity, whereas MNK2b, which is the only MNK variant that lacks the MAPK-binding domain, has a very low basal activity (reviewed by Buxade et al. 2008).

MNKs are recruited to eIF4E through association with the carboxy-terminal part of eIF4G, via a polybasic region located in their amino termini (Pyronnet et al. 1999). Phosphorylation of eIF4E appears to be restricted to metazoans as yeast lack MNK orthologs and a MNK interaction domain in eIF4G. Although initial genetic studies in Drosophila revealed that the phosphorylation of eIF4E is required for normal development, growth, and viability (Lachance et al. 2002), MNK1/2 double knockout (DKO) mice, and mice in which wild-type eIF4E was replaced with a nonphosphorylatable mutant, do not exhibit any conspicuous phenotype (Ueda et al. 2004; Furic et al. 2010). Nonetheless, the phosphorylation of eIF4E appears to be critical for its tumorigenic activity, inasmuch as the nonphosphorylatable mutant of eIF4E exhibits dramatically lower oncogenic potential in vivo and in vitro as compared to wild-type (Topisirovic et al. 2004; Wendel et al. 2007).

The effects of eIF4E phosphorylation on cap-dependent translation are still poorly understood. Because MNKs are recruited to eIF4E via eIF4G (Pyronnet et al. 1999), it is most likely that the phosphorylation of eIF4E occurs during or shortly after the assembly of the eIF4F complex on the 5′-mRNA cap (Fig. 3B). Because Ser209 is located in the proximity of the entrance of the cap-binding pocket of eIF4E, it is plausible that the phosphorylation status of eIF4E alters its cap-binding activity (Topisirovic et al. 2011). It was initially predicted that a salt bridge between the phosphate group of Ser209 and Lys159, forms a “clamp” that stabilizes the eIF4E:5′mRNA cap complex (Marcotrigiano et al. 1997; Matsuo et al. 1997). However, subsequent studies revealed that the phosphorylation of eIF4E reduces its affinity for the cap (Scheper et al. 2002; Slepenkov et al. 2006). Depending on the experimental conditions, eIF4E phosphorylation was shown to correlate with increased (Kaspar et al. 1990; Manzella et al. 1991; Walsh and Mohr 2004; Worch et al. 2004) or decreased global translation rates (Knauf et al. 2001; Morley and Naegele 2002; Naegele and Morley 2004). In addition to the variable effects on global translation, it was shown that the phosphorylation of eIF4E selectively affects translation of a subset of mRNAs, such as Mcl-1 (Wendel et al. 2007). Indeed, a genome-wide comparison of translational activity observed in wild-type mouse embryonic fibroblasts and their counterparts in which wild-type eIF4E was replaced by the nonphosphorylatable mutant, revealed that the phosphorylation status of eIF4E selectively affects translation of a subset of mRNAs including those encoding proteins that play roles in inflammation (e.g., Ccl2 and Ccl7) and tumor progression (e.g., MMP3 and MMP9) (Furic et al. 2010). These findings suggest an intriguing possibility that phospho-eIF4E modulates the inflammatory response and stimulates tumorigenesis by selectively up-regulating translation of mRNAs that encode proteins critical for these processes.

RSKs

In addition to being involved in the upstream regulation of mTORC1, several studies have shown a significant role for RSKs in various stages of translational control (Fig. 4). The RSK family is comprised of four Ser/Thr kinases (RSK1-4) that are directly activated by ERK1/2-mediated phosphorylation (reviewed by Carriere et al. 2008b). RSK family members exist in all vertebrate species and not so distant RSK orthologs have been identified in Drosophila and C. elegans. All RSK isoforms are expressed at relatively high levels during development and in adult tissues, with the exception of RSK4, which is more abundant during embryogenesis (Zeniou et al. 2002). A notable structural feature of the RSK family came about during evolution, when the genes for two distinct protein kinases fused, generating a single kinase capable of receiving an upstream activating signal from ERK1/2 to the RSK carboxy-terminal kinase domain (CTKD), and transmitting, with high efficiency and fidelity, an activating input to the RSK N-terminal kinase domain (NTKD). Both the NTKD and CTKD are distinct and functional (Jones et al. 1988; Fisher and Blenis 1996), but only the NTKD has thus far been implicated in the phosphorylation of exogenous substrates. This domain belongs to the AGC family of protein kinases, which also includes AKT, PKC, and S6Ks. These kinases phosphorylate very similar consensus sequences and, consistent with this, several bona fide RSK substrates were also shown to be targeted by AKT, PKC, and/or S6Ks.

The first clue suggesting that RSK may be involved in translational control came 25 years ago when it was identified as an rpS6 kinase in maturating Xenopus laevis oocytes (Erikson and Maller 1985, 1986). A subsequent study showed that activated RSK translocates to polysomes and stimulates the phosphorylation of several ribosome-associated proteins (Angenstein et al. 1998). With the use of rapamycin, the S6Ks were later found to be the predominant rpS6 kinases operating in somatic cells (Chung et al. 1992). Studies from S6K1/2-knockout mice confirmed these findings but also showed that there was residual MEK1/2-dependent phosphorylation at Ser235/236 (Pende et al. 2004). In accordance with this observation, RSK was shown to specifically phosphorylate rpS6 on Ser235/236 in vitro and in cells in response to agonists of the MAPK pathway (Roux et al. 2007). RSK-mediated phosphorylation of rpS6 correlates with eIF4F complex assembly and cap-dependent translation, suggesting that RSK provides an mTOR/S6K-independent input linking MAPK signaling to the regulation of translation initiation.

It is now understood that the Ras/MAPK pathway impinges on the PI3K/mTOR pathway at various steps to regulate translation. Aside from its role upstream of mTORC1, RSK phosphorylates additional components of the translational machinery, such as eIF4B in vitro and in vivo (Shahbazian et al. 2006). eIF4B stimulates the RNA-helicase activity of eIF4A (Rozen et al. 1990) and eIF4B phosphorylation was found to promote its interaction with eIF3 (Holz et al. 2005; Shahbazian et al. 2006). This interaction correlates with increased translation rates (Shahbazian et al. 2006) and also agrees with the finding that phosphorylated eIF4B stimulates cap-dependent translation in vivo (Holz et al. 2005). As indicated above, S6Ks also regulate eIF4B phosphorylation, which may explain the biphasic pattern of eIF4B phosphorylation observed in response to certain mitogenic cues.

RSK might also regulate translation through the phosphorylation of GSK3β (Sutherland et al. 1993). As with AKT and S6Ks, RSK-mediated phosphorylation of GSK3β on Ser9 inhibits its kinase activity and thereby releases the inhibition of eIF2B (Cohen and Frame 2001). In collaboration with ERK1/2, RSK2 was shown to participate in rRNA synthesis and cell growth. On serum stimulation, RSK2 phosphorylates a residue important for the function of TIF-1A, a transcription initiation factor required for RNA Pol I activity and rRNA transcription (Zhao et al. 2003). Finally, RSK was also shown to phosphorylate eEF2 kinase (Wang et al. 2001a), thereby underscoring the involvement of RSK at multiple levels of the pathway that leads to protein synthesis. Thus, RSK activation seems to coordinate crucial processes that are associated with the proper regulation of gene expression and protein synthesis.

FUTURE PERSPECTIVES

Over the past two decades, significant progress has been made in understanding how translation is modulated by signaling pathways in response to various extracellular stimuli and intracellular cues. Notwithstanding the large body of recently garnered data, in particular those uncovering the molecular mechanisms that underlie the effects of MAPK, TOR (this article), and eIF2α-kinases (Benham 2012) on translation, several outstanding issues await to be addressed. For instance, it has been known for more than a decade that rapamycin suppresses translation of 5′-TOP mRNAs (Jefferies et al. 1994). How mTORC1 stimulates translation of these transcripts is still unknown, however. Moreover, emerging data suggest that mTOR and MAPK act as pivotal regulators of energy metabolism (Shaw and Cantley 2006; Howell and Manning 2011). Protein synthesis is the most energy consuming process in the cell (Buttgereit and Brand 1995), which begs the question of whether mTOR and MAPK act as central nodes of the cellular networks that coordinate changes in the translatome and metabolome by orchestrating the activity of cellular factors involved in their respective processes. Finally, several signaling pathways other than MAPK, TOR, and eIF2α-kinases (e.g., Pak2, GSK3, Cdk11, and CK2) have been implicated in the phosphorylation of the components of the translational apparatus and auxiliary factors (Table 1), but their physiological role in translational control is still largely unknown. In spite of these issues, there are good reasons to be optimistic as we enter a new era of research in signaling to translational machinery. Recent advances in pharmacological tools (e.g., compounds that specifically and potently target signaling to translational machinery such as active-site mTOR inhibitors [Malina et al. 2012]), and advancements in the technologies that enable monitoring of changes in the phosphoproteome (e.g., quantitative mass spectrometry), translatome (e.g., ribosome profiling/RNA-seq [Lasko 2012]) on a genome-wide scale will undoubtedly facilitate efforts to establish the role of signaling pathways in translational control in homeostasis and disease.

Table 1.

Phosphorylation sites in mammalian translation factors and associated proteins, regulatory kinases, and functional consequences of the phosphorylation

Protein	Phosphorylation sitesa (major kinase)	Function and referencesb
4E-BP1	Thr37,1 Thr461 (mTORC1)2	Priming sites3–5
	Ser651 (mTORC1),2 Thr701 (mTORC1?/CDK1?)	Dissociation from eIF4E3–5
	Ser841 (?), Ser1016 (?), Ser1127 (CDK?)	Unknown function
4E-BP2	Thr378, Thr468 (mTORC1)	Priming sites (by analogy with 4E-BP1)
	Ser65 (mTORC1), Thr70 (?mTORC1/?CDK1)	Dissociation from eIF4E (by analogy with 4E-BP1)
eIF4E	Ser2099 (MNK1/2)10	Unknown function10,11 Increases oncogenic activity and promotes translation of a subset of mRNAs (e.g., Mcl-1, MMPs, CCLs)12
eIF4GI	Ser1186 (PKCα)13	Modulates MNK binding13
	Ser1108, Ser1148, Ser1192 (mTORC1)14	Stimulation of translation of mRNAs containing upstream open reading frame (uORF)15 (?)
	Ser896 (Pak2)16	Inhibition of cap-dependent translation16
eIF2α	Ser5117 (HRI; PKR; GCN2, PERK) reviewed in Ref. 18 and see Pavitt and Ron (2012)	Stabilizes the eIF2/GDP/eIF2B complex, thus preventing recycling of eIF2 (reviewed in Ref. 18 and see Pavitt and Ron 2012)
rpS6	Ser23519 and Ser2368 (S6K1/2,8 RSK20), Ser240,8 Ser244,8 and Ser2478 (S6K1/2)8	Unknown function20–23 Global translation rates increased in MEFs expressing a nonphosphorylatable form of rpS624
PDCD4	Ser67, Ser457 (S6K1/2; AKT)25,26	Degradation by the ubiquitin-proteasome system and subsequent activation of eIF4A25,26
eIF4B	Ser406(?),27 Ser422 (S6K1/2,28 AKT,27 RSK29)	Increases binding to eIF329,30
eIF4H	Tyr12,31 Tyr45,31 Tyr101,31 Ser19331 (?)	?
eIF2Bε	Ser540 (GSK3)32	Inhibits recycling of eIF232
	Ser544 (DYRK)33	Priming site for GSK333
	Ser717/718 (CK2)34	Facilitates eIF2 binding34
eIF3 (29 phosphosites)35	eIF3b: Ser83,36 Ser85,36 Ser12536(?)	?
	eIF3c: Ser39,37 Ser166,36 Thr524,36 Ser90938 (?)	?
	eIF3f: Ser46, Thr119 (CDK11)39,40	Regulation of protein synthesis and apoptosis39,40
	eIF3g: Thr41,41 Ser4241 (?)	?
	eIF3h: Ser18342 (?)	Increased oncogenic activity42
	eIF3i: Tyr44531 (?)	?
eIF1	Tyr3031 (?)	?
eIF5	Ser389, Ser390 (CK2)43	Promotes cell cycle progression43
eIF5B	Ser107,36 Ser113,36 S135,36 S137,38 S164,36 S182,36 S183,38 S186,36 S190,36 S214,41 S116836 (?)	?
eIF6	Ser235 (PKCβII)44	Dissociation of eIF6 from the 60S; 80S assembly44
eEF1A	Thr431 (PKCδ)45 Ser51 (PKCβI)46 Ser300 (TβR-I)47	Activation (?)45 ? Inhibition of mRNA translation47
eEF2	Thr56 (eEF2K)48	Inhibits binding to the ribosome49
eEF2K	Ser78 (mTOR?)50	Inhibits CaM binding50
	Ser359 (SAPK/p38δ?)51	Inhibition (?)51
	Ser366 (S6K1; RSK)52	Inhibition52
	Ser398 (AMPK)53	Activation53
	Ser500 (PKA)54	Induces Ca2 +-independent activity54

This table includes selected phospho-acceptor sites identified in large-scale mass-spectrometry-based experiments that await functional characterization (e.g., eIF5B; unknown kinase/function indicated with ?), as well as phosphorylation sites with established role in translational control (e.g., 4E-BPs and eIF2α). Further information on the as-yet functionally noncharacterized phosphorylated residues of the components of the translational apparatus can be found on Phosphosite (www.phosphosite.org) or Uniprot (www.uniprot.org) Web sites.

^aAmino acid numbering is based on human proteins.

^bReferences: (1) Fadden et al. (1997); (2) Brunn et al. (1997); (3) Pause et al. (1994); (4) Gingras et al. (1999); (5) Gingras et al. (2001); (6) Wang et al. (2003); (7) Heesom et al. (1998); (8) Wang et al. (2005); (9) Joshi et al. (1995); (10) Waskiewicz et al. (1997); (11) Fukunaga et al. (1997); (12) Furic et al. (2010); (13) Dobrikov et al. (2011); (14) Raught et al. (2000); (15) Ramirez-Valle et al. (2008); (16) Ling et al. (2005); (17) Kudlicki et al. (1987); (18) Sonenberg and Hinnebusch (2009); (19) Krieg et al. (1988); (20) Roux et al. (2007); (21) Banerjee et al. (1990); (22) Kozma et al. (1990); (23) Pende et al. (2004); (24) Ruvinsky et al. (2005); (25) Dorrello et al. (2006); (26) Palamarchuk et al. (2005); (27) van Gorp et al. (2009); (28) Raught et al. (2004); (29) Shahbazian et al. (2006); (30) Holz et al. (2005); (31) Rush et al. (2005); (32) Welsh et al. (1998); (33) Woods et al. (2001); (34) Wang et al. (2001b); (35) Damoc et al. (2007); (36) Beausoleil et al. (2004); (37) Gevaert et al. (2005); (38) Kim et al. (2005); (39) Shi et al. (2003); (40) Shi et al. (2009); (41) Ballif et al. (2005); (42) Zhang et al. (2008); (43) Homma et al. (2005); (44) Ceci et al. (2003); (45) Kielbassa et al. (1995); (46) Piazzi et al. (2010); (47) Lin et al. (2010); (48) Nairn et al. (1987); (49) Price et al. (1991); (50) Browne et al. (2004); (51) Knebel et al. (2001); (52) Wang et al. (2001a); (53) Horman et al. (2002); (54) Diggle et al. (1998).

Abbreviations: CDK, cyclin-dependent kinase; PKC, protein kinase C; Pak2, p21-activated kinase 2; HRI, heme-regulated eIF2α kinase; PKR, double-stranded RNA-activated eIF2α kinase; GCN2, general control nonrepressed eIF2α kinase; PERK, double-stranded RNA-activated protein kinase-like ER kinase; DYRK, dual-specificity tyrosine phosphorylation-regulated kinase; CK2, protein kinase CK2 (formerly known as casein kinase II); TβR-I, TGF-β1 receptor; eEF2K, eukariotic translation elongation factor 2-kinase; PKA, protein kinase A; SAPK, stress-activated protein kinase. Additional abbreviations are provided in the text.