For most eukaryotic organisms, including Saccharomyces cerevisiae, the rapid inhibition of protein synthesis forms part of a response to stress. In order to balance the changing conditions, precise stress-specific alterations to the cell's proteome are required. Therefore, in the background of a global down-regulation in protein synthesis, specific proteins are induced. Given the level of plasticity required to enable stress-specific alterations of this kind, it is surprising that the mechanisms of translational regulation are not more diverse. In the present review, we summarize the impact of stress on translation initiation, highlighting both the similarities and distinctions between various stress responses. Finally, we speculate as to how yeast cells generate stress-responsive programmes of protein production when regulation is focused on the same steps in the translation pathway.
- processing body (P-body)
- Saccharomyces cerevisiae
- translation initiation
The success of an organism in a given environment is limited by its ability to adapt to ever-changing surroundings. As a single-celled non-motile eukaryote, Saccharomyces cerevisiae must adapt in order to survive environmental changes. The inhibition of translation initiation plays a pivotal role in this adaptation process. The down-regulation of such an energy-consuming biosynthetic process as protein synthesis rationalizes precious resources. In addition, more specific up-regulation of translation initiation leads to the induction of specific adaptation proteins.
The process of translation initiation is mediated by a set of highly conserved eIFs (eukaryotic initiation factors) and associated proteins (Figure 1). Translational regulation involves altered protein interactions, activities and stabilities as well as post-translational modifications.
Amino acids and nitrogen signalling to translation initiation
Adaptation to changes in the concentration of intracellular metabolites such as amino acids and other nitrogenous compounds has been widely studied in yeast. Two pathways known as the GAAC (general amino acid control) and the TOR (target of rapamycin) pathway are particularly important and both target translation initiation as part of their mode of operation [1–3].
The GAAC pathway responds to amino acid starvation leading to a global inhibition of translation initiation. Nonetheless, translation of the transcription factor Gcn4p is up-regulated during this process, leading to the transcriptional induction of a host of genes including those involved in amino acid biosynthesis [4,5].
Following amino acid depletion, an increase in uncharged tRNA activates the protein kinase Gcn2p, leading to phosphorylation of Ser51 on the α-subunit of the initiation factor eIF2 [2,5,6]. eIF2, in its active GTP-bound form, binds to methionyl initiator tRNA to form a TC (ternary complex). TC binds to the 40S small ribosomal subunit, with other eIFs, to form the 43S complex . The 43S complex then interacts with the mRNA to form the 48S complex and following scanning, start codon recognition and 60S joining, translation initiation is complete and the elongation/termination processes act to finish polypeptide production (Figure 1). However, when phosphorylated, eIF2 acts as a competitive inhibitor of its guanine-nucleotide-exchange factor, eIF2B, resulting in an accumulation of inactive eIF2-GDP, a decrease in TC and a lowering of the rate of translation initiation. Classically, two mutations have been described which short-circuit this pathway: the deletion of the Gcn2p protein kinase gene and expression of the non-phosphorylatable form of eIF2α, SUI2-S51A. These are commonly used when evaluating the role of the GAAC pathway in response to uncharacterized stress responses .
Another indicator of GAAC pathway activation is the GCN4 gene whose promoter and 5′-UTR (untranslated region) are fused to a reporter coding sequence, such as β-galactosidase. GCN4 translational induction is controlled by four uORFs (upstream open reading frames). Under non-starvation conditions, after translation of the uORF1 (first uORF), an unusual translation termination event allows a proportion of ribosomes to remain associated with the mRNA and reinitiate downstream at uORFs 2–4 . Translation termination at these uORFs leads to the release of the mRNA, such that the downstream GCN4 ORF is seldom translated. Under starvation conditions, TC is less abundant, therefore on the GCN4 mRNA a proportion of reinitiating ribosomes bypass uORFs 2–4 as TC is not regained by the ribosome in sufficient time . Hence these ribosomes reach the GCN4 ORF to translate it. Thus any stress that leads to reduced TC induces expression of Gcn4p, e.g. fusel alcohols, oxidative stress, purine starvation, salt stress and rapamycin.
Rapamycin is a macrolide drug that has proved invaluable in studying the connections between growth and protein synthesis [1,9–11]. Rapamycin inhibits the TOR pathway by interacting with FKBP12 (FK506-binding protein 12) to prevent Tor1/2p protein kinase activation . This pathway is usually involved in the maintenance of active translation initiation under nutrient replete conditions. In mammals, a key mechanism for such translational activation is the inhibition of 4E-BP1 (eIF4E-binding protein 1) . The cap binding protein eIF4E binds to eIF4G in the closed-loop complex facilitating interaction with the 43S complex (Figure 1). This interaction is inhibited by 4E-BP1. mTOR (mammalian TOR) phosphorylates the 4E-BP1 to decrease eIF4E binding, leading to increased protein translation. In contrast, the absence of nutrients inhibits mTOR leading to 4E-BP1 hypophosphorylation. This increases 4E-BP1 affinity for eIF4E and so lowers levels of translation initiation . In yeast, treatment with rapamycin can decrease global translation by more than 80% . In S. cerevisiae two 4E-BPs have been identified, Eap1p and Caf20p. Although eap1Δ mutants are partially resistant to rapamycin, the basis for this resistance is largely unknown: caf20Δ mutants are equally sensitive to wild-type strains . An observation that may be connected to the eap1Δ phenotype is that, after rapamycin treatment, eIF4G becomes degraded . However, a careful kinetic analysis suggests that there is a lag between the translational inhibition and eIF4G degradation . Therefore eIF4G degradation is likely to represent a consequence rather than a cause of translational inhibition. However, this degradation could have dramatic implications for the continued translation of specific mRNAs.
A variety of observations suggest connections between the TOR and GAAC pathways. Rapamycin leads to a rapid phosphorylation of eIF2α, Gcn4p expression is induced and an SUI2-S51A mutant exhibits no polysome run-off following rapamycin treatment [16–18]. It seems that TOR may have an impact on the GAAC pathway via effects on the protein phosphatase Sit4p. Rapamycin increases eIF2α phosphorylation leading to translational repression either by inhibiting eIF2α dephosphorylation or by activating Gcn2p [16,17]. Therefore both amino acid starvation and TOR inhibition globally repress translation via the same basic mechanism: inhibition of eIF2B activity by phosphorylated eIF2. However, the mRNAs that remain translated under these two stress regimes are unlikely to be the same; yet how such a precise differential regulation might be established is unknown.
Carbohydrate-responsive signal transduction pathways are both highly complex and conserved between yeast, plants and mammals [19–21]. In S. cerevisiae, glucose is the preferred carbon source and a major signalling molecule. Depletion of glucose causes a rapid inhibition of translation initiation [22,23], where the underlying mechanism is distinct from other nutrient starvation pathways: it is independent of eIF2α phosphorylation, the TOR pathway, eIF4G degradation and 4E-BPs [11,22–24]. The inhibition of translation has a number of consequences. It decreases the number of ribosomes  both as a result of reduced translation and increased protein/mRNA degradation of ribosome components and biogenesis factors. In addition, it leads to a redistribution of mRNA, mRNA decay factors and translation initiation factors into stress granules and mRNA P-bodies (processing bodies). However, the primary cause for this dramatic inhibition of translation initiation is still largely unknown. Recently, upon glucose depletion, the helicase eIF4A was observed to dissociate from the pre-initiation complex . This effect is concomitant with an increased association between eIF4G and eIF3 within the pre-initiation complex, and an increase in translation factors associated with the 40S subunit. These data suggest that the lack of eIF4A prevents the 43S pre-initiation complex from scanning the mRNA 5′-UTR and therefore leads to a stalled and hence stabilized pre-initiation complex .
Many mutants involved in the glucose repression/derepression pathway are translationally resistant to glucose stress, including reg1Δ, hxk2Δ, glc7-Q48K, ssn6-Δ6 and mig1Δ mig2Δ [22,25–27]. Since constitutively derepressed mutants from all stages of the pathway (including transcription factors) exhibit this phenotype, it seems likely that this pathway does not directly cause the translation inhibition. Instead the altered gene expression programme in these mutants probably explains the maintenance of translation initiation following glucose starvation. Therefore, metabolically, these mutants are more similar to cells growing on non-fermentable carbon sources, and when such carbon sources are removed from exponentially growing cells, little or no inhibition of translation is observed.
PKA (protein kinase A) also plays a central role in the response of S. cerevisiae to glucose . In yeast, there are three isoforms of the PKA catalytic subunit, Tpk1p, Tpk2p and Tpk3p. Mutant strains with weak PKA activity exhibit maintained translation following glucose depletion . Recent research has shown that the Tpk2p and Tpk3p catalytic subunits move to P-bodies following nutrient starvation , where they are thought to play a key role in P-body formation . PKA is thought to limit P-body size by phosphorylating the conserved P-body constituent Pat1p 
mRNA decay mutants also maintain translation following glucose starvation. More specifically, mutants affecting the 5′→3′ mRNA decay pathway at stages downstream of deadenylation continue translating following glucose depletion . Intriguingly, these same mutants exhibit large constitutive P-bodies  probably because they accumulate mRNA to such an extent that these bodies form bearing undegraded and partially degraded mRNA. It has been suggested that the excess of mRNA in these strains may swamp the translational regulatory machinery . It has also been suggested that the mRNA decay components, Dhh1p and Pat1p, can act as translational regulators .
Aerobic metabolism results in the production of ROS (reactive oxygen species) including peroxide, hydroxyl radicals and the superoxide anions. Redox active metals, such as iron and copper, and molecules that deplete or oxidize glutathione (e.g. cadmium) also cause oxidative stress. As part of a co-ordinated response to oxidative stress, there is a rapid and reversible inhibition of protein synthesis in S. cerevisiae [34,35]. Evidence suggests that phosphorylation of eIF2α by Gcn2p explains the effect of peroxide on translation initiation [35,36]. Cadmium and diamide (a thiol-oxidizing agent) also inhibit translation, yet Gcn2p kinase activation only partially explains the regulation, with the 4E-BP Eap1p also playing a key role . Similarly, although Gcn4p is needed for peroxide resistance, this transcription factor is not necessary for resistance to cadmium or diamide . Therefore oxidative stress targets translation initiation in complex ways in S. cerevisiae. What is more, inhibition of translation initiation alone does not explain the reduction in amino acid incorporation rates; therefore oxidative stress must target some other aspect of translation, perhaps elongation and/or termination, to provide a concerted down-regulation of protein production .
Salt, lithium and osmotic stress
S. cerevisiae can adapt to a variety of osmotic environments and the primary mechanism for osmoregulation is through the HOG (high osmolarity in glycerol) pathway. This pathway induces a programme of gene expression that restores the osmotic balance . High osmolarity also transiently inhibits global translation initiation [38,39], in a mechanism relying upon Gcn2p and eIF2α phosphorylation . For instance, NaCl addition leads to a 2 h translational inhibition, whereupon a return to normal translation levels requires the HOG pathway. However, the addition of high salt can have complex effects and act as both an osmotic and a cationic stress , for example, the gcn2Δ mutant strain is tolerant to high NaCl but sensitive to high KCl .
Other salts such as lithium act differently again: lithium specifically inhibits translation in yeast grown on galactose, due to its inhibition of phosphoglucomutase, a key enzyme in galactose metabolism. . Overexpression of either eIF4A or Sit4p confers lithium tolerance to S. cerevisiae grown in galactose, suggesting that translation up-regulation may be enough to counteract lithium toxicity .
eIF1A is another translation initiation factor with connections to osmotic stress . This factor normally functions to promote scanning of the 5′-UTR for the AUG start codon by the 43S pre-initiation complex. Overexpression of this protein tempers the translational inhibition in response to both NaCl and lithium.
Therefore, once again, there are complex inputs on the translational machinery that explain the translational effects of osmotic stress. So, even though eIF2α phosphorylation probably explains the global down-regulation, other factors clearly play a role in this complex response.
Fusel and volatile alcohols
In yeast, fusel alcohols result from the use of amino acids as a nitrogen source during nitrogen scarcity . Addition of various fusel alcohols to S. cerevisiae inhibits translation initiation in a Gcn2p-independent manner . Inhibition of translation also occurs in TOR, CAF20 and EAP1 mutants, indicating that this process is independent of these factors. Butanol up-regulates expression of Gcn4p through a more direct inhibition of eIF2B and subsequent reduction in TC levels, than is brought about by phosphorylation of eIF2α [46,47]. Mutations of eIF2B confer both enhanced sensitivity and resistance to fusel alcohols [46,47]. Perhaps most surprisingly, addition of fusel alcohols leads to eIF2α dephosphorylation in a Sit4p-dependent manner , but it seems highly unlikely that this effect is directly connected to the inhibition of translation initiation. Dephosphorylation of eIF2α is also seen upon the addition of 12% isoflurane, a volatile short-chain ether used in anaesthetics . As for the butanol, exposure to isoflurane increases Gcn4p expression and rapidly decreases translation initiation levels in a Gcn2p-independent manner. Mutations that cause resistance to isoflurane have also been mapped to the eIF2B proteins . Therefore it appears that a variety of small alcohol or ether molecules have the capacity to somehow inhibit eIF2B leading to reduced levels of TC and translational inhibition. Intriguingly, eIF2B localizes to a large cytoplasmic body that has been termed the eIF2B body . eIF2B is a resident feature of this body whereas eIF2 cycles through the body. The rate of eIF2 cycling correlates closely with the rate of guanine nucleotide exchange . Interestingly, the eIF2B body moves rapidly around the cell and inhibition of its movement correlates closely with the inhibition of translation initiation caused by fusel alcohols .
Heat shock and membrane stress
After a robust increase in temperature S. cerevisiae responds by decreasing its levels of protein translation and up-regulating the transcription of heat-shock-responsive genes via the transcription factors Hsf1p and Msn2p/4p [50,51]. Inhibition of translation initiation is observed and correlates with an increase in eIF2α phosphorylation. However, polysome run-off still occurs in a gcn2Δ strain showing that the eIF2α phosphorylation is not required for the translational inhibition . In contrast, deletion of the 4E-BP gene, EAP1, does appear to partially prevent the polysomal run-off associated with heat stress . In addition to denaturing proteins, heat stress also causes membrane disordering [54,55]. Indeed temperature-sensitive mutations in components of the vesicle trafficking pathway (which transport membrane components) and cells treated with the membrane stress inducer CPZ (chlorpromazine), also cause an inhibition of translation initiation [54,55]. These effects are not dependent on the membrane stress signalling complexes Wsc1p or Pkc1p , but instead they may be caused by deficiencies in sphingolipids and ceramides; the biosynthesis of which are also up-regulated by heat stress . Here, it seems a combination of effects at the level of eIF2α phosphorylation and the 4E-BP, Eap1p seem to explain the effects of CPZ on translation initiation .
Mutants in the sphingolipid production pathway are also hypersensitive to heat. One such mutation, lcb1–100, is found in a serine palmitoyltransferase gene that is involved in an initial step in sphingolipid synthesis. . Mutants in EAP1 restore translation initiation rates in lcb1–100 mutants following heat shock. Therefore these specific mutants and inhibitors in sphingolipid synthesis repress protein synthesis during heat stress via the 4E-BP Eap1p, rather than phosphorylation of eIF2α .
The present review summarizes work showing that diverse stresses appear to regulate translation initiation in a limited number of pathways. These include the regulation of eIF2B (either dependent or independent of eIF2α phosphorylation) and control of the closed-loop complex via 4E-BPs or at the level of eIF4A interaction. This raises an obvious question; that is, how is it possible to generate the diverse responses required to enable adaptation under these different stress conditions, if the stresses target protein synthesis in a similar manner? One possible answer to this lies in the complex multivalent impact of stress on cells. The effects on protein synthesis are just one facet of the impact of these stresses. The stress conditions may alter other aspects of gene regulation: transcription, mRNA stability and mRNA localization among others, and it is only evaluating all of these stages that a true representation of the stress-specific gene expression pattern can be established. Although this is almost certainly true, the situation may be even more complex than this. A characterization of translationally maintained mRNAs shortly after amino acid starvation or fusel alcohol treatment showed that there was virtually no overlap in the programme of translational maintenance even though the stresses both target eIF2B and the mRNAs are largely unaltered at the transcript level . Therefore it appears that there are other inputs at the level of translational regulation that are more specific in terms of the mRNAs that are regulated, or that a combination of global regulatory inputs produces different patterns of gene regulation. For example, the degradation of eIF4G may prove critical in defining the distinction between rapamycin and amino acid starvation. Similarly, a combination of effects on Eap1p and eIF2α phosphorylation may explain why cadmium treatment is distinct from peroxide. The challenge in studying stress responses now lies less in defining the major impact of the stress and more in dissecting these combinatorial and modulatory effects.
C.E.S. was funded by a Wellcome Trust project grant [grant number 088141/Z/09/Z] to M.P.A.
We thank members of the Ashe laboratory past and present who have contributed to the thoughts and ideas developed during the course of our work on various stresses.
RNA UK 2012: An Independent Meeting held at The Burnside Hotel, Bowness-on-Windermere, Cumbria, U.K., 20–22 January 2012. Organized and Edited by Raymond O'Keefe and Mark Ashe (Manchester, U.K.).
Abbreviations: CPZ, chlorpromazine; eIF, eukaryotic initiation factor; 4E-BP1, eIF4E-binding protein 1; GAAC, general amino acid control; HOG, high osmolarity in glycerol; mTOR, mammalian target of rapamycin; P-body, processing body; PKA, protein kinase A; TC, ternary complex; TOR, target of rapamycin; uORF, upstream open reading frame; UTR, untranslated region
- © The Authors Journal compilation © 2012 Biochemical Society