Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Histone acetylation and deacetylation wikipedia , lookup
Protein (nutrient) wikipedia , lookup
Magnesium transporter wikipedia , lookup
Signal transduction wikipedia , lookup
Phosphorylation wikipedia , lookup
Protein moonlighting wikipedia , lookup
Protein phosphorylation wikipedia , lookup
Gene expression wikipedia , lookup
Oncogene (1999) 18, 6121 ± 6128 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00 http://www.stockton-press.co.uk/onc Regulation of translation initiation following stress M Saeed Sheikh*,1 and Albert J Fornace Jr1 1 Division of Basic Sciences, National Cancer Institute, National Institutes of Health, Building 37, Room 5CO9, Bethesda, Maryland, MD 20892, USA Recent studies suggest that genotoxic and non-genotoxic stresses appear to invoke translational checkpoints in order to inhibit protein synthesis. Depending on the stress and/or cell type, this downregulation of protein synthesis may either (i) protect against the deleterious eects of noxious agents and ensure the conservation of resources that are needed to survive under adverse conditions or (ii) activate apoptosis. In this article, we have reviewed several lines of evidence which support the notion that regulation of translation initiation is an important component of the cellular stress response. While the stress-induced post-translational regulation of translation initiation factors (eIFs) has been well documented, stress-induced regulation of eIFs at the mRNA levels, as reviewed here, is only beginning to be elucidated. Thus, the stress-mediated regulation of eIFs occurs at multiple dierent levels involving, transcriptional, post-transcriptional and post-translational controls. Keywords: translation initiation; eIF2; eIF3; eIF5; eIF1A121/SUI1; translation checkpoints Eukaryotic cells have developed multiple mechanisms to mount a response against endogenous and exogenous stresses. Cellular responses to genotoxic or non-genotoxic stress involves transcriptional and posttranscriptional controls aecting various cellular processes (reviewed by Fornace, 1992; Kaufman, 1999). It is now widely held that DNA damage (genotoxic stress) activates cell cycle checkpoints at G1/S and G2/M which provide necessary time for cells to repair damaged DNA prior to progression into next phase of the cell cycle (reviewed by Fornace, 1992; Pellegata et al., 1996; Morgan and Kastan, 1997; Schwartz and Rotter, 1998). Translation regulation in response to stress is an area that has not been extensively studied. Several lines of evidence support the notion that the regulation of protein translation appears to be an important component of the cellular stress response. In general, it is conceivable that genotoxic and non-genotoxic stresses could invoke translational checkpoints to block protein synthesis, particularly that of toxic proteins containing errors. The downregulation of protein synthesis could be an adaptive response which may protect against the deleterious eects of toxic agents and ensure the conservation of resources that are needed to survive *Correspondence: MS Sheikh under such adverse conditions (reviewed by Hinnebusch, 1994). However, depending on the stress and cell type, the inhibition of protein synthesis may also activate apoptosis. In this article, we will review the current state of published knowledge and some of our recent data in this emerging area of cellular stress response. Protein translation Eukaryotic mRNAs undergo modi®cations in the nucleus that involve addition of a unique `cap' structure at the 5'-end and poly(A)+ tail at the 3'-end (reviewed by Sachs et al., 1997). The `cap' is an inverted 7-methylguanosine that is attached via a 5'-5' triphosphate bridge to the very ®rst nucleotide of mRNA. The `cap' appears to function in splicing, polyadenylation, nuclear export, RNA splicing, mRNA stability and is essential for ecient translation (reviewed by Lewis and Izaurralde, 1997). Recent evidence also supports an important role of 3'poly(A)+ tail in protein translation (reviewed by Sachs et al., 1997). Translation initiation requires the recruitment of the smaller (40S) ribosomal subunit to mRNA and involves protein-protein and protein-RNA interactions (reviewed by Hershey et al., 1996). All mRNAs are translated via free ribosomes or via membrane bound ribosomes. The former are unattached cytoplasmic ribosomes that are involved in the synthesis of a majority of the cellular proteins, while the latter are bound to the cytosolic side of the endoplasmic reticulum (ER) membrane and synthesize membrane proteins or those that are destined for secretion. The scanning model of translation initiation proposes that a 43S subunit complex containing the initiator methionyl-tRNAi (Met-tRNAi) coupled 40S ribosomal subunit, eIF2, eIF3 and eIF1A binds to the `cap' structure (reviewed by Clemens and Bommer, 1999). With the help of eIF4B and eIF4F whose helicase activities unwind the secondary structure, the 40S ribosome scans mRNA in the 3'-direction in search of the ®rst initiation codon (AUG) (reviewed by Clemens and Bommer, 1999). Once the ®rst AUG is identi®ed, the interactions between AUG and the anticodon of Met-tRNAi sets the stage for chain elongation which is the next step in protein synthesis. The polypeptide chain elongation occurs in three steps (reviewed by Clemens and Bommer, 1999; Clark, 1980; Caskey, 1980) (Figure 1). In step 1, the eIFs dissociate from the smaller ribosomal subunit, the larger ribosomal subunit binds to the smaller subunit and a second aminoacyl-tRNA molecule is recruited to the A-site of the ribosome (reviewed by Clemens and Translational controls following stress MS Sheikh and AJ Fornace 6122 Figure 1 An overview showing eukaryotic protein translation in progress Bommer, 1999; Clark, 1980). Step 2 involves a reaction catalyzed by peptidyl transferase in which the amino acid on the second tRNA is joined by a peptide bond to an amino acid linked to tRNA in the P-site. In step 3, the ribosome moves three nucleotides along mRNA in the 3'-direction, the tRNA in the A-site is translocated to the P-site, the A-site becomes vacant and ready to accept another amino acid-chargedtRNA. The process continues until one of the three stop codons are reached at which point the translation is terminated (reviewed by Clemens and Bommer, 1999; Clark, 1980; Caskey, 1980) (Figure 1). On average, the completion of protein synthesis can take from 20 ± 60 s. Post-translational regulation of eIF2 in response to stress Mammalian eIF2 is composed of three subunits (a, b, g) in 1 : 1 :1 stoichiometry (Pathak et al., 1988). The 43S preinitiation complex contains a small ribosomal subunit (40S), eIF1A, eIF1A121/SUI1(see below), eIF3 and a ternary complex composed of eIF2, GTP and Met-tRNAi (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). Within the 43S complex, eIF3 interacts with eIF4G, which in turn interacts with the `cap'-binding subunit eIF4E (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). Thus, 43S complex via various subunits of eIF4 (see below) binds to the `cap' at the 5'-end of mRNA and slides along mRNA towards 3'-end until the ®rst AUG is reached (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). As soon as the ®rst AUG is identi®ed, the GTP in the ternary complex is hydrolyzed to GDP, which results in the release of eIF2 and GDP as a binary complex (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). In a nucleotide-exchange reaction catalyzed by eIF2B, the GDP bound to eIF2 is replaced with GTP and it interacts with the Met-tRNAi, to once again, form a ternary complex (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). Phosphorylation of eIF2a Three dierent types of kinases that phosphorylated eIF2a have been identi®ed (reviewed by Haro et al., 1996). All these kinases are activated by various types of stresses (reviewed by Haro et al., 1996). HRI (heme regulated inhibitor of translation) was the ®rst eIF2a kinase that was identi®ed from rabbit reticulocytes and is activated in response to heme deprivation, heavy metals and heat shock proteins (reviewed by Chen and London, 1995). PKR (double-stranded-RNA-activated protein kinase) is activated by double-stranded-RNA species, particularly those produced following viral infection (Schmedt et al., 1995; reviewed by Haro et al., 1996). The expression of PKR is also induced by interferon in mammalian cells (Meurs et al., 1990). GCN2 is a third type of eIF2a kinase, which has been characterized from yeast Saccharomyces cerevisiae (reviewed by Hinnebusch, 1994). GCN2 is activated in response to nutrient deprivation, a condition that causes the accumulation of uncharged-tRNAs (reviewed by Hinnebusch, 1994). It has been reported that phosphorylation of eIF2 (Ser51 of a subunit) inhibits eIF2B such that eIF2B can no longer catalyze nucleotide exchange and thus, a majority of the eIF2a exists in binary complex with GDP (reviewed by Clemens and Bommer, 1999; Haro et al., 1996). It follows that the phosphorylated eIF2a has a high anity for eIF2B, which further prevents its ability to catalyze GDP ± GTP exchange on the unphosphorylated eIF2a (reviewed by Clemens and Bommer, 1999; Haro et al., 1996). Thus, the eIF2a phosphorylation on Ser51 is a highly conserved adaptive response that can cause downregulation of translation initiation under certain types of stresses (Figure 2). gadd34/MyD116 and virus connection HSV-1 (herpes simplex virus) infection is known to stimulate a host defense response that involves the activation of host PKR, phosphorylation of eIF2a, and inhibition of protein synthesis. However, HSV-1 has evolved a mechanism to overcome the host defense response (reviewed by Liebermann and Homan, 1998). For example, the HSV-1 g34.5 gene (also known as ICP34.5) product prevents the accumulation of phosphorylated eIF2a and thereby allows continuous protein synthesis throughout viral infection (Chou and Roizman, 1992; Chou et al., 1995) (Figure 2). The carboxyl-terminal domain of the g34.5 gene product, responsible for inducing dephosphorylation of eIF2a, shows a high degree of homology to the carboxyl-terminal domain of the Gadd34 protein and the virulence-associated gene from African swine fever virus (Figure 3) (Zhan et al., 1994; Chou and Roizman, 1994; reviewed by Liebermann and Homan, 1998). Interestingly, the gadd34 gene was originally cloned as Translational controls following stress MS Sheikh and AJ Fornace Figure 2 Stress-induced phosphorylation of eIF2a can cause inhibition of translation initiation. Stress activates HRI, PKR and GCN2 kinases. These kinases phosphorylate the a subunit of eIF2. Phosphorylated eIF2a (i) inhibits eIF2B and thereby prevents its ability to catalyze the exchange of GDP to GTP on eIF2a, (ii) interacts with eIF2B with a higher anity which further inhibits eIF2B's ability to catalyze GDP-GTP exchange on eIF2a. Starvation and DNA damage induce gadd34 which shows homology to HSV-1 viral g34.5 gene. g34.5 gene product (and perhaps Gadd34) appears to interact with PP1a to dephosphorylate eIF2a in order to derepress protein synthesis inhibition. It is also possible that Gadd34 and g34.5 gene product may play opposing roles since g34.5 gene product appears to block the growth inhibitory eects of Gadd34 a growth arrest and DNA damage-inducible gene from Chinese hamster ovary cells (Fornace et al., 1988 and see below). The murine homolog of human and hamster genes has also been cloned as the myeloid dierentiation primary response gene 116 (MyD116) (reviewed by Liebermann and Homan, 1998). The human, hamster and mouse Gadd34/MyD116 proteins are highly homologous (Figure 3) and recently the rat homolog named PEG3 has also been cloned (Su and Fisher, 1997). It has been shown that the HSV-1 carrying a deletion in the Gadd34/MyD116 homologous carboxyl-terminal domain of the g34.5 gene product is unable to propagate, most likely due to premature termination of protein synthesis (Chou and Roizman, 1994). Recent evidence demonstrates that the substitution of the Gadd34/MyD116 carboxyl-terminal domain with the corresponding mutant domain of the g34.5 gene product (Figure 3) is able to preclude the premature termination of protein synthesis thus suggesting that the carboxyl-terminal domain of mammalian Gadd34/ MyD116 appears to display a role comparable to that of viral g34.5 gene product (He et al., 1996; reviewed by Liebermann and Homan, 1998) (Figure 2). The molecular mechanism via which the conserved carboxyl-terminal domains of g34.5 gene product and Gadd34/MyD116 induce dephosphorylation of eIF2a remains unclear. A recent study has demonstrated that the g34.5 gene product can form a complex with protein phosphatase 1a (PP1a) and PP1a may be responsible for the dephosphorylation of eIF2a (He et al., 1997) (Figure 2). It still remains unclear, however, whether Gadd34/MyD116 plays a role similar to that of the g34.5 gene product. Viruses with the mutant g34.5 gene are unable to grow in the absence of exogenous Gadd34/MyD116 C-terminal region (reviewed by Liebermann and Homan, 1998). This could be because the endogenous Gadd34/MyD116 protein levels are very low and not sucient to complement the mutant g34.5 phenotype. However, this scenario appears paradoxical since overexpression of exogenous Gadd34 is incompatible with cell survival (Zhan et al., 1994). We have recently noted that the upregulation of endogenous Gadd34 is tightly coupled with the induction of apoptosis; overexpression of exogenous GADD34 cDNA also induces apoptosis in human neoplastic cells (Adler et al., in press and our unpublished results). Furthermore, enforced overexpression of g34.5 gene appears to protect from the growth inhibitory eects of Gadd34 (our unpublished results). It, therefore, appears as if Gadd34 and g34.5 gene product may play opposing roles i.e. Gadd34 tends to induce apoptosis while the g34.5 gene product inhibits apoptosis in the virally-infected cells. Whether indeed the Gadd34-induced apoptotic eects are mediated via alterations in protein translation, remains to be elucidated. Regulation of eIF4 in response to stress The binding of cap structure at the 5'-end is facilitated by various subunits of eIF4 including eIF4E, -4A, -4B, -4F and -4G (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). Together eIF4E, -4A and -4G form a large complex known as eIF4F. eIF4E is a capbinding protein, eIF4A has a RNA helicase activity and eIF4G is a scaolding subunit that promotes interactions with eIF3 (reviewed by Clemens and Bommer, 1999; Sachs et al., 1997). eIF4E is believed to be a critical subunit of eIF4 since not only does it promote the direct interaction of 43S complex with mRNA, but it also exists in limited molar amounts (reviewed by Clemens and Bommer, 1999; Raught and Gingras, 1999; Sachs et al., 1997). Regulation of eIF4E: a MAP kinase connection eIF4E is an important initiation factor that plays a critical role in controlling the rate of protein translation (reviewed by Raught and Gingras, 1999). Transcriptional upregulation of eIF4E or its phosphorylation via MAPK pathway promotes eIF4E binding to the `cap' and consequently enhances protein translation (Wang et al., 1998; Waskiewicz et al., 1999; reviewed by Clemens and Bommer, 1999; Raught and Gingras, 1999; Sachs et al., 1997). In general, eIF4E promotes cell growth and proliferation, inhibits apoptosis and cause cellular transformation (reviewed by Raught and Gingras, 1999). Protein kinase Mnk1 is a substrate for both ERK1 and p38 MAPK (Fukunaga and Hunter, 1997; Waskiewicz et al., 1997). ERK1 is 6123 Translational controls following stress MS Sheikh and AJ Fornace 6124 Figure 3 Homology of Gadd34 with viral proteins. The human, hamster and mouse Gadd34 proteins are homologous to each other and contain a carboxyl-terminal conserved region (dark blue) that is highly homologous to the corresponding regions of HSV1 g34.5 gene product (also known as ICP34.5) and African swine fever virus-NL protein. Lower panel shows the amino acid sequence alignment of the carboxyl-terminal domain of Gadd34 with the corresponding amino acid sequence from the HSV-1 g34.5 gene product and the African swine fever virus-NL protein. The human, hamster and mouse Gadd34 proteins also contain moderately conserved 30 ± 40 amino acids long repeats that are not present in viral proteins (Zhan et al., 1994) activated in response to growth promoting signals while, p38 is a stress-activated MAPK (Dhanasekaran and Reddy, 1998). Both ERK1 and p38 can promote eIF4E phosphorylation via Mnk1 (Wang et al., 1998; reviewed by Clemens and Bommer, 1999; Raught and Gingras, 1999). Evidence suggests that MAPKmediated phosphorylation of eIF4E appears to enhance its ability to bind to the `cap' and thereby increases the rate of translation initiation (reviewed by Raught and Gingras, 1999). However, p38 also promotes phosphorylation of eIF4E inhibitory proteins E4-BPs, which although do not aect eIF4E binding to the `cap', do nevertheless prevent its interaction with eIF4G and consequently with the 43S complex (Waskiewicz et al., 1999; reviewed by Clemens and Bommer, 1999; Raught and Gingras, 1999). Thus, the activity of eIF4E can be aected indirectly during stress and can ultimately cause inhibition of protein synthesis (Figure 4). Caspase-mediated cleavage of eIF4G In general, apoptosis can be induced due to overall shutdown of protein synthesis, which would suggest that apoptosis can occur in the absence of de novo protein synthesis. For example, the protein synthesis inhibitor cycloheximide has been shown to induce apoptosis (Ishii et al., 1995, 1997). PKR has also been shown to induce apoptosis by inhibiting protein synthesis via eIF2a phosphorylation (Lee and Este- ban, 1994; Srivastava et al, 1998; Gil et al., 1999). Overexpression of phosphorylation-de®cient eIF2a, which acts in a dominant-negative manner, has been shown to override the apoptotic as well as the translational inhibitory eects of PKR (Gil et al., 1999). A number of studies have reported that, in certain situations, de novo protein synthesis appears necessary for apoptosis to occur. For example, p53induced apoptosis has been shown to require new protein synthesis in order for the full program to be completed (Polyak et al., 1997). Similarly, nutrient deprivation (Bulera et al., 1996; Schulz et al., 1996), ionizing radiation (our unpublished results), and nonsteroidal anti-in¯ammatory drugs (NSAIDs) (Chan et al., 1998)-induced apoptosis can be inhibited by cycloheximide suggesting that ongoing protein synthesis is needed. Whether protein synthesis is absolutely required for apoptosis to occur, remains debatable and may vary based on the stress and cell type. However, it is becoming increasingly clear that a majority of the apoptotic signals engage the preexisting intracellular apoptotic machinery in the absence of de novo protein synthesis. Caspases are an integral component of the preexisting apoptotic machinery and exist in an inactive state as procaspases that are processed into catalytically active state in response to apoptotic stimuli (reviewed by Nunez et al., 1998; Li and Yuan, 1999). The caspase activation involves (i) oligomerization-mediated autoactivation of upstream Translational controls following stress MS Sheikh and AJ Fornace eIF4G in response to genotoxic and non-genotoxic stress appears to occur via caspase 3 (Marissen and Lloyd, 1998), also known as `the executioner' caspase which cleaves a number of other death substrates (reviewed by Nunez et al., 1998; Li and Yuan, 1999). Thus, eIF4G has been identi®ed as one of many death substrates and its cleavage during stress-induced apoptosis appears yet another mechanism which can modulate translation initiation (Figure 5). Regulation of eIF1A121/SUI1, eIF3 and eIF5 genes following stress Figure 4 Stress-induced phosphorylation of 4E-BPs can indirectly aect eIF4E activity to inhibit translation initiation. Stressinduced activation of p38 MAPK promotes eIF4E phosphorylation via Mnk1. Phosphorylation of eIF4E potentiates its interaction with `cap' and promotes translation initiation. p38 MAPK can also phosphorylate 4E-BPs, phosphorylated 4E-BPs prevent interaction between eIF4E and eIF4G and thus inhibit translation initiation caspases and (ii) cleavage of downstream procaspases by active upstream caspases (reviewed by Nunez et al., 1998; Li and Yuan, 1999). Therefore, the activation of the caspase cascade may occur in the complete absence of de novo protein synthesis. It is nevertheless possible that some apoptotic stimuli, albeit in a cell type-speci®c manner, may still require the synthesis of certain proapoptotic molecules that appear critical for the activation of the caspase cascade (Figure 5). This appears to be the case for p53, which may mediate its apoptotic eects, at least in part, by transcriptionally upregulating the expression of its downstream target genes including BAX, DR5, FAS and PIGs (p53inducible genes) (Sheikh et al., 1998 and references therein). Recently, it was shown that genotoxic and nongenotoxic stress-induced apoptosis involves the proteolytic cleavage of eIF4G (Marissen and Lloyd, 1998; Clemens et al., 1998). eIF4G plays a critical role in translation initiation by bridging the interactions between eIF4E and eIF3 (reviewed by Clemens and Bommer, 1999). Thus, the proteolytic cleavage of eIF4G during stress-induced apoptosis results in protein synthesis inhibition (Marissen and Lloyd, 1998; Clemens et al., 1998). Therefore, an overall decrease in the levels of various macromolecules including the eIF4G following complete shutdown of protein synthesis is conceivable. However, eIF4G is not cleaved following complete inhibition of protein synthesis (Marissen and Lloyd, 1998). Cleavage of We have recently reported that mammalian eIF5 is a DNA damage-inducible (DDI) gene (Sheikh et al., 1997). Expression of the human eIF5 gene is induced by UV-irradiation albeit in a cell type speci®c manner (Sheikh et al., 1997). We have also cloned the full length cDNA of another DDI transcript referred to as A121 (Sheikh et al., 1997, 1999). The expression of A121 was induced by UV-irradiation, methylmethane sulfonate (MMS) and agents that cause endoplasmic reticulum (ER)-stress (Sheikh et al., 1999). (The ER is a site for calcium storage, translation, post-translational modi®cations, protein folding, and lipid and sterol synthesis. ER-stress is induced by agents that alter ER homeostasis (reviewed by Kaufman, 1999)). The human A121 cDNA contains a 113 amino acid long open reading frame which bears a high degree of homology to yeast translation initiation factor Sui1 (Sheikh et al., 1999; Yoon and Donahue, 1992). To con®rm that A121 cDNA encodes an eIF that is equivalent to yeast Sui1p, the human A121 cDNA was expressed in yeast Saccharomyces cerevisiae harboring the mutant endogenous sui1 allele (Sheikh et al., 1999). Expression of exogenous human A121 cDNA corrected the mutant sui1 phenotype con®rming that human A121 encodes a bona ®de eIF (Sheikh et al., 1999). The predicted amino acid sequence of A121 exhibits perfect identity to the partial length sequence of the recently puri®ed human eIF1 (Pestova et al., 1998; Sheikh et al., 1999), further demonstrating that A121 encodes the human eIF1 (hereafter refer to as eIF1A121/SUI1). The identi®cation of eIF5 and eIF1A121/SUI1 as genotoxic stress-inducible genes (Sheikh et al., 1997, 1999) demonstrates, for the ®rst time, that the stressmediated regulation of eIFs can also occur at the mRNA level. Furthermore, the expression of human eIF1A121/SUI1 gene is regulated not only by genotoxic stress but also by ER-stress inducing agents (Sheikh et al., 1999). Further characterization of the human eIF1A121/SUI1 gene revealed that it exhibits two transcripts of approximately 0.65 kb and 1.3 kb (Sheikh et al., 1999). These transcripts contain a common coding region but dier in their 3'-untranslated region due to alternative usage of dierential poly(A)+ signals (Sheikh et al., 1999). Interestingly, the long and short eIF1A121/SUI1 transcripts were dierentially regulated by genotoxic and ER stress. For example, UV-irradiation and the alkylating agent MMS upregulated the expression of only the larger transcript whereas ERstress regulation of eIF1A121/SUI1 involved upregulation of both transcripts (Sheikh et al., 1999). The signi®cance of eIF1A121/SUI1 upregulation in response to genotoxic and ER-stress is an issue that 6125 Translational controls following stress MS Sheikh and AJ Fornace 6126 Figure 5 Stress-induced apoptosis may involve activation of caspase cascade; caspases can cleave cellular proteins that control basic cellular processes including replication, transcription and building of cellular architecture. Stress-induced caspase-dependent cleavage of eIF4G has recently been reported and can cause inhibition of protein synthesis. Generalized inhibition of any of the basic cellular processes can activate apoptosis. In general apoptosis can occur in the absence of de novo proteins synthesis. However, depending on the stimulus and cell type, the ongoing protein synthesis may also be required for apoptosis to occur is currently being investigated. Yeast sui1 and two other genes sui2 and SUI3 were originally identi®ed as translation initiation suppressor loci (Castilho-Valavicius et al., 1990; Cigan et al., 1989). The products of these genes have been implicated in controlling the ®delity of translation initiation, since mutations in any one of these loci could restore the expression of the HIS4 allele lacking the initiation codon AUG (Yoon and Donahue, 1992; Castilho-Valavicius et al., 1990; Cigan et al., 1989). SUI2 and SUI3 code for the a and b subunits of eIF2, respectively (Cigan et al., 1989) whereas SUI1 encodes the yeast equivalent of human eIF1A121/SUI1 (Sheikh et al., 1999). A recent report has described the puri®cation of human eIF1 and eIF1A (previously known as eIF4C) (Pestova et al., 1998). It has been reported that eIF1 and eIF1A in association with the 43S complex form a 48S complex at the initiation codon (Pestova et al., 1998). eIF1 and eIF1A act in concert, since the 43S complex lacking eIF1 and eIF1A does not reach the initiation codon (Pestova et al., 1998). Human eIF1 alone can recognize and destabilize the aberrantly formed complexes at the initiation codon (Pestova et al., 1998), a ®nding which is consistent with its role in controlling the ®delity of translation initiation. It is possible that the upregulation of eIF1A121/SUI1 expression following stress is a defensive response whereby cells try to control the translation of certain unwanted mRNAs whose products might otherwise be deleterious (Figure 6). Support for this hypothesis comes from recent ®ndings that eIF1A121/SUI1 not only controls the ®delity of translation initiation (Pestova et al., 1998; Yoon and Donahue, 1992; Cui et al., 1998) Figure 6 Genotoxic and endoplasmic reticulum stress may control translation initiation ®delity by upregulating the levels of eIF1A121/SUI1 but its overexpression negatively aects cell proliferation. For example, when overexpressed, eIF1A121/SUI1 inhibits the colony forming eciency of human cancer cells (Lian et al., 1999 and our unpublished results). Recently, it has been shown that the hepatitis B virus X antigen can downregulate the expression of eIF1A121/SUI1 (Lian et al., 1999). Analyses of tumors and matching normal tissues from HBV-infected Translational controls following stress MS Sheikh and AJ Fornace patients suering from hepatocellular carcinoma revealed the absence of eIF1A121/SUI1 gene expression only in the hepatic tumors but not in the matching normal hepatic tissues (Lian et al., 1999). It is interesting that a variety of RNA viruses can synthesize multiple proteins from a single transcript by ribosomal frameshifting (Cui et al., 1998 and references therein). It is not clear whether the hepatitis B virus can also direct ribosomal frameshifting, however since yeast sui1 has been implicated in blocking negative frameshifting (Cui et al., 1998), it is conceivable that certain viruses may downregulate the expression of eIF1A121/SUI1 (which might otherwise prevent negative frameshifting) and thereby gain a survival advantage. The molecular mechanisms underlying the DNA damage repair and replication in the higher eukaryotes are complex. It is expected that the DNA damage induced by genotoxic stress in mammalian cells may also be repaired by the error-prone DNA repair/ replication mechanisms (Hays et al., 1990; Viswanathan et al., 1999 and refs. therein) which can ®x mutations that may also alter the translation initiation codon. Recent evidence suggests that certain types of DNA base modi®cations can cause RNA polymerase to miscode at the site of lesions which can result in mutated transcripts (Viswanathan et al., 1999 and references therein). It is interesting that RNA polymerase-induced miscoding results in G to A base substitution (Viswanathan et al., 1999 and references therein) and may also mutate the start codon (AUG to AUU). Since eIFA121/SUI1 will not allow translation to initiate at a non-AUG codon, it is, therefore tempting to speculate that upregulation of eIF1A121/SUI1 following genotoxic stress may provide yet another form of checkpoint to control translation of mRNAs corresponding to genes that have a mutated start codon. Similarly, it is possible that the upregulation of eIF1A121/SUI1 following ER-stress may be critical in order to exert control over ER-coupled translation as well. In this article, we have reviewed several lines of evidence which support the notion that regulation of translation initiation is an important component of cellular stress response. While the stress-induced posttranslational regulation of eIFs has been well documented, stress-induced regulation of eIFs at the mRNA levels is beginning to be elucidated. Clearly, the stress-mediated regulation of eIFs occurs at multiple dierent levels involving, transcriptional, post-transcriptional and post-translational controls. In the future, it will be pertinent to investigate whether the activity and expression of other eIFs, not covered in this article, are also regulated (negatively or positively) in response to stress. Our initial evidence demonstrates that the expression of eIF1A, the interacting partner of eIF1, is not regulated by stress in a manner similar to that of eIF1A121/SUI1 (Sheikh et al., 1999 and our unpublished results). The expression of one of the subunits of eIF3, by contrast, is downregulated in response to ionizing radiation in certain cells (our unpublished results). Given that eIF1A121/SUI1 exhibits several putative phosphorylation sites (Sheikh et al., 1999), it will also be important to investigate (i) whether it is phosphorylated in response to stress and, if indeed it is, (ii) whether or not its activity is altered following stress-induced phosphorylation. Finally, the issue of how stress-induced overexpression of eIF1A121/SUI1 exactly aects translation initiation, will require more in depth studies. We have reviewed only a few examples of stress-induced regulation of eIFs, clearly, much more remains to be done. Continued investigations in this emerging area of cellular stress response promise to provide more exciting clues, which will certainly enhance our understanding of the molecular mechanisms underlying the regulation of protein translation in context to cellular stress response. Acknowledgments We thank Drs L Augeri, R. Kaufman and Y Huang for helpful discussions and comments. References Adler HT, Chinery R, Wu DY, Kussick SJ, Forance Jr AJ and Tkachuck DC. Mol. Cell. Biol., (in press). Bulera SJ, Sattler CA and Pitot HC. (1996). Hepatology, 23, 1591 ± 1601. Caskey CT. (1980). Trends Biochem. Sci., 5, 234 ± 237. Castilho-Valavicius B, Yoon HJ and Donahue T. (1990). Genetics, 24, 483 ± 495. Chan TA, Morin PT, Vogelstein B and Kinzler KW. (1998). Proc. Natl. Acd. Sci. USA., 95, 681 ± 686. Chen JJ and London IM. (1995). Trends Biochem. Sci., 20, 105 ± 108. Chou J and Roizman B. (1992). Proc. Natl. Acad. Sci. USA, 89, 3266 ± 3270. Chou J and Roizman B. (1994). Proc. Natl. Acad. Sci. USA, 91, 5247 ± 5251. Chou J, Chen JJ, Gross M and Roizman B. (1995). Proc. Natl. Acad. Sci. USA, 92, 10516 ± 10520. Cigan AM, Pabich EK, Feng L and Donahue TF. (1989). Proc. Natl. Acad. Sci. USA, 86, 2784 ± 2788. Clark B. (1980). Trends Biochem. Sci., 5, 207 ± 210. Clemens MJ, Bushell M and Morley SJ. (1998). Oncogene, 17, 2921 ± 2931. Clemens MJ and Bommer U-A. (1999). Int. J. Biochem. Cell Biol., 31, 1 ± 23. Cui Y, Dinman JD, Kinzy TG and Peltz S. (1998). Mol. Cell. Biol., 18, 1506 ± 1516. Dhanasekaran N and Reddy EP. (1998). Oncogene, 17, 1447 ± 1455. Fornace Jr AJ. (1992). Annu. Rev. Genet., 26, 507 ± 526. Fornace Jr AJ, Alamo IJ and Hollander MC. (1988). Proc. Natl. Acad. Sci. USA, 85, 8800 ± 8804. Fukunaga R and Hunter T. (1997). EMBO J., 16, 1921 ± 1933. Gil J, Alcami J and Esteban M. (1999). Mol. Cell. Biol., 19, 4653 ± 4663. Haro CD, Mendez R and Santoyo J. (1996). FASEB J., 10, 1378 ± 1387. Hays JB, Ackerman EJ and Pang QS. (1990). Mol. Cell. Biol., 10, 3505 ± 3511. 6127 Translational controls following stress MS Sheikh and AJ Fornace 6128 He B, Chou J, Liebermann DA, Hofman B and Roizman B. (1996). J. Virol., 70, 84 ± 90. He B, Gross M and Roizman B. (1997). Proc. Natl. Acad. Sci. USA, 94, 843 ± 848. Hershey JWB, Mathews MB and Sonenberg N. (1996). Translational control. Cold Spring Habor Laboratory Press: Cold Spring Harbor, New York. Hinnebusch AG. (1994). Trends Biochem. Sci, 19, 409 ± 414. Ishii H, Tamauchi H and Gob GC. (1997). Int. J. Exp. Pathol., 78, 123 ± 131. Ishii H, Etheridge MR and Gob GC. (1995). Immunol. Cell Biol., 73, 463 ± 468. Kaufman RJ. (1999). Genes Dev., 13, 1211 ± 1233. Lee SB and Esteban M. (1994). Virology, 199, 491 ± 496. Lewis JD and Izaurralde E. (1997). Eur. J. Biochem., 247, 461 ± 469. Liebermann DA and Homan B. (1998). Oncogene, 17, 3319 ± 3329. Lian Z, Pan J, Liu J, Zhang S, Zhu M, Arbuthnot P, Kew M and Feitelson MA. (1999). Oncogene, 18, 1677 ± 1687. Li H and Yuan J. (1999). Curr. Opin. Cell Biol., 11, 261 ± 266. Marissen WE and Lloyd RE. (1998). Mol. Cell. Biol., 18, 7565 ± 7574. Meurs E, Chong K, Galabru J, Thomas NSB, Kerr IM, Williams BRG and Hovanessian AG. (1990). Cell, 62, 379 ± 390. Morgan SE and Kastan MB. (1997). Adv. Cancer Res., 71, 1 ± 25. Nunez G, Benedict MA, Hu Y and Inohara, N. (1998). Oncogene, 17, 3237 ± 3245. Pathak VK, Nielsen PJ, Trachsel H and Hershey JWB. (1988). Cell, 54, 633 ± 639. Pellegata NS, Antoniono RJ, Redpath JL and Stanbridge EJ. (1996). Proc. Natl. Acad. Sci. USA, 93, 15209 ± 15214. Pestova TV, Borukhov S and Hellen CUT. (1998). Nature, 394, 854 ± 859. Polyak K, Xia Y, Zweier JL, Kinzler KW and Vogelstein B. (1997). Nature, 389, 300 ± 305. Raught B and Gingras A-C. (1999). Int. J. Biochem. Cell Biol., 31, 43 ± 57. Sachs AB, Sarnow P and Hentze MW. (1997). Cell, 89, 831 ± 838. Schmedt C, Green SR, Manche L, Talor DR, Ma Y and Mathews MB. (1995). J. Mol. Biol., 249, 29 ± 44. Schulz JB, Weller M and Klockgether T. (1996). J. Neurosci., 16, 4696 ± 4706. Schwartz D and Rotter V. (1998). Semin. Cancer Biol., 8, 325 ± 336. Sheikh MS, Fernandez-Salas E, Yu M, Hussain A, Dinman JD, Peltz SW, Huang Y and Fornace Jr AJ. (1999). J. Biol. Chem., 274, 16487 ± 16493. Sheikh MS, Burns TF, Huang Y, Wu GS, Amundson S, Brooks KS, Fornace Jr AJ and El-Deiry WS. (1998). Cancer Res., 58, 1593 ± 1598. Sheikh MS, Carrier F, Papathanasiou MA, Hollander MC, Zhan Q, Yu K and Fornace Jr AJ. (1997). J. Biol. Chem., 272, 26720 ± 26726. Srivastava SP, Kumar KU and Kaufman RJ. (1998). J. Biol. Chem., 273, 2416 ± 2423. Su ZZ and Fisher PB. (1997). Proc. Natl. Acad. Sci. USA, 94, 9125 ± 9130. Viswanathan A, You HJ and Doetsch PW. (1999). Science, 284, 159 ± 162. Wang X, Flynn A, Waskiewicz AJ, Webb BLJ, Vries RG, Baines IA, Cooper JA and Proud CG. (1998). J. Biol. Chem., 273, 9373 ± 9377. Waskiewicz AJ, Johnson JC, Penn B, Mahalingam M, Kimball SR and Cooper JA. (1999). Mol. Cell. Biol., 19, 1871 ± 1880. Waskiewicz AJ, Flynn A, Proud CG and Cooper JA. (1997). EMBO J., 16, 1909 ± 1920. Yoon HJ and Donahue TF. (1992). Mol. Cell. Biol., 12, 248 ± 260. Zhan Q, Lord KA, Alamo Jr I, Hollander MC, Carrier F, Ron D, Kohn KW, Homan B, Liebermann DA, Fornace Jr AJ. (1994). Mol. Cell. Biol., 14, 2361 ± 2371.