Environmental stresses inducing translation arrest are accompanied by the deposition of translational components into stress granules (SGs) serving as mRNA triage sites. It has recently been reported that, in Saccharomyces cerevisiae, formation of SGs occurs as a result of a prolonged glucose starvation. However, these SGs did not contain eIF3, one of hallmarks of mammalian SGs. We have analyzed the effect of robust heat shock on distribution of eIF3a/Tif32p/Rpg1p and showed that it results in the formation of eIF3a accumulations containing other eIF3 subunits, known yeast SG components and small but not large ribosomal subunits and eIF2α/Sui2p. Interestingly, under these conditions, Dcp2p and Dhh1p P-body markers also colocalized with eIF3a. Microscopic analyses of the edc3Δlsm4ΔC mutant demonstrated that different scaffolding proteins are required to induce SGs upon robust heat shock as opposed to glucose deprivation. Even though eIF2α became phosphorylated under these stress conditions, the decrease in polysomes and formation of SGs occurred independently of phosphorylation of eIF2α. We conclude that under specific stress conditions, such as robust heat shock, yeast SGs do contain eIF3 and 40S ribosomes and utilize alternative routes for their assembly.
Intracellular compartmentalization of specific mRNAs and components of the translation machinery is an important mode of regulation for gene expression in eukaryotic cells. In this respect, formation of various mRNA-containing assemblies, such as stress granules (SGs) or processing bodies (P-bodies), is a striking illustration of this regulation (Anderson and Kedersha, 2008; Bond, 2006; Parker and Sheth, 2007).
Various stresses cause a fast and transient redistribution of nontranslated mRNAs into SGs in mammalian as well as plant cells. This effect is a result of a rapid repression of general translation initiation, often mediated through phosphorylation of the alpha subunit of the translation initiation factor 2 (eIF2) (Holcik and Sonenberg, 2005; Kedersha et al., 1999). Therefore mammalian SGs are thought to represent abortive 48S complexes that include mRNA linked to poly(A)-binding protein 1, translation initiation factors (e.g. eIF3, eIF4A, eIF4G) and 40S ribosomal subunits (Anderson and Kedersha, 2008; Kedersha et al., 2002). SGs persist in equilibrium with polysomes and this can be affected by translation inhibitors, which either stabilize or destabilize the polysomes. It has been proposed that the SGs serve as triage sites redirecting mRNA to either translation, storage or degradation (Anderson and Kedersha, 2008). Interestingly, SG-like structures, containing eIF3, have been found in heat-shocked fission yeast Schizosacchomyces pombe (Dunand-Sauthier et al., 2002), but have not yet been reported in Saccharomyces cerevisiae.
P-bodies were originally described as sites of mRNA decapping and decay, which mainly contain the mRNA degradation machinery components including the decapping enzyme complex (Brengues et al., 2005). P-bodies and SGs share some proteins and mRNA components, but also contain a number of unique markers specific to each structure (Kedersha and Anderson, 2007). Unlike SGs, P-bodies are present even under non-stress conditions (Mollet et al., 2008) and never contain eIF3 and 40S ribosomal subunits (Anderson and Kedersha, 2008; Brengues and Parker, 2007; Parker and Sheth, 2007; Sheth and Parker, 2003; Sheth and Parker, 2006). P-bodies are therefore believed to be structurally distinct from SGs; however, their simultaneous appearance in stressed mammalian cells showed a close spatial connection perhaps indicating an intimate mutual communication (Kedersha et al., 2005; Wilczynska et al., 2005). In addition, because of these interactions, a role for pre-existing P-bodies in the nucleation of SG assembly has been suggested recently (Buchan et al., 2008; Mollet et al., 2008).
In the budding yeast S. cerevisiae, a number of small cytosolic foci of accumulated Dcp2p can be observed also in normal proliferating cells (Brengues et al., 2005; Teixeira et al., 2005). Under various stresses, e.g. glucose starvation, hyperosmotic stress and heat stress at 37°C, these foci increase in size to form enlarged P-bodies (Brengues et al., 2005). Recently, novel mRNA assemblies called EGP-bodies (for eIF4E, eIF4G and Pab1p) that are distinct from P-bodies were found to form under prolonged glucose starvation in S. cerevisiae (Hoyle et al., 2007). While this manuscript was under a review process in this journal, additional markers of mammalian SGs were identified within EGP-bodies that led the authors to rename them `yeast stress granules' (Buchan et al., 2008). In sharp contrast to mammalian SGs, however, the reported yeast SGs of glucose-deprived yeast cells do not contain eIF3 and 40S ribosomal subunits (Buchan et al., 2008; Hoyle et al., 2007).
In S. cerevisiae, numerous stresses including hyperosmolarity, glucose deprivation and robust heat shock at 46°C were demonstrated to stabilize many mRNAs (Hilgers et al., 2006). Whereas the impact of hyperosmolarity on formation of P-bodies seems to be identical to that of glucose deprivation (Brengues et al., 2005), it is not known whether robust heat shock also elicits similar effects. In this report, we show that robust heat shock at 46°C for 10 minutes applied to aerobically cultivated yeast cells results in formation of accumulations containing eIF3 subunits, mRNA, eIF4G2, Pab1p, Ngr1p, Pub1p, 40S ribosomal subunits and also typical P-body proteins such as Dcp2p and Dhh1p. Assembly of these transient protein accumulations is eIF2α-phosphorylation independent. We also demonstrate that they are not formed in cycloheximide-treated and energy-depleted cells. Because of their composition, irregular shapes and requirement for different scaffolding proteins than are those needed for P-bodies, we posit that these protein accumulations are yeast SGs specific for the robust heat shock. We further propose that they can only be detected in S. cerevisiae when transient severe stresses such as robust heat shock are applied and the period of yeast mRNA triage is relatively prolonged.
Robust heat shock induces transient formation of eIF3-containing SGs
We showed previously that eIF3a/Tif32p/Rpg1p colocalized with cytoplasmic microtubules in the fixed yeast cells (Hasek et al., 2000). In order to examine intracellular distribution of eIF3a in living cells, we constructed wild-type strains expressing eIF3a fused either with RFP or GFP from its chromosomal locus. In accord with previously published data (Brengues et al., 2005), we found that distribution of both fluorescing eIF3a fusions is uniformly cytosolic in cells cultivated in complete YPD medium at 30°C (Fig. 1A). In addition, this localization pattern was not affected by a mild heat shock at 42°C for 10 minutes (data not shown), which has been used to induce formation of SGs in fission yeast S. pombe (Dunand-Sauthier et al., 2002). However, we found that raising the temperature of the heat shock to over 42°C has a dramatic effect on the eIF3a distribution. In particular, incubation of cells in YPD medium at 46°C for 10 minutes resulted in formation of distinct eIF3a accumulations (Fig. 1B). We observed that this rearrangement of eIF3a was transient and reversible (Fig. 1C), since the eIF3a became uniformly cytosolic after cultivation of heat-shocked cells in YPD medium at 30°C for 30 minutes.
To investigate whether these protein accumulations are limited to eIF3a in isolation or whether it is a common characteristic of components of the eIF3 complex, we prepared strains coexpressing eIF3a-GFP with either eIF3b/Prt1p-RFP or eIF3c/Nip1p-RFP fusions from their chromosomal loci. As shown in Fig. 1D, the eIF3a-GFP foci always colocalized with the other eIF3 subunits, strongly suggesting that protein accumulations induced by robust heat shock contain core components of the eIF3 complex. As confirmed by high values of the Pearson's correlation coefficient (Rr), over 0.9 (see Materials and Methods), distribution for both eIF3b and eIF3c signals almost completely matched that of eIF3a.
In living yeast, accumulation of mRNA in P-bodies has been assessed using a specific PGK1 reporter mRNA containing multiple U1A-specific binding sites in its 3′ untranslated region to which U1A-GFP fusion protein binds (Brengues et al., 2005; Teixeira et al., 2005). We employed this detection system in order to find out if mRNA also accumulates in the eIF3a foci after robust heat shock. As shown in Fig. 1E, the PGK1 mRNA in unstressed cells was distributed across the cytosol, as has previously been observed (Brengues et al., 2005). In cells heat-shocked in a complete medium at 46°C for 10 minutes, the cytoplasmic GFP signal corresponding to the accumulated PGK1 mRNA significantly overlapped with that of eIF3a foci. Weak background fluorescence was observed in the cytoplasm of control cells expressing the U1A-GFP fusion protein only, without PGK1 mRNA under all tested conditions (data not shown).
eIF3a foci contain typical components of the mammalian SGs
Stalled 48S pre-initiation complexes containing mRNA, eIF3, eIF4G, PABP and 40S, but not 60S ribosomal subunits and eIF2, are the core constituents of the mammalian SGs (Kedersha et al., 2002). To determine whether yeast eIF3 foci are related to mammalian SGs, we constructed strains coexpressing various RFP and GFP fusion proteins, exposed the exponentially growing cells to robust heat shock in YPD medium at 46°C for 10 minutes and analyzed the distribution of fusion proteins using the Olympus Cell R microscopic system.
First, we found that in the heat-shocked cells eIF3a colocalized with eIF4G2 and Pab1p (Fig. 2A). Very high values of Pearson's correlation coefficient (Rr), over 0.8, confirmed a high degree of colocalization. In addition, we also analyzed changes in the distribution of Rps30A-GFP fusion protein that we used as a marker of the 40S ribosomal subunit. We found that robust heat shock induced formation of Rps30A-GFP foci, which clearly colocalized with eIF3a-RFP as illustrated by a high Rr (Fig. 2B). By contrast, exposing cells to robust heat shock did not induce accumulation of Rpl25-GFP that was used as a marker of the 60S ribosomal subunit (Fig. 2C). Consistently, the frequency scatter plot analysis as well as a small value of Rr confirmed a very weak colocalization of Rpl25p with eIF3a. These results imply that, in contrast to the 40S ribosomal subunit, the 60S subunit is not actually accumulated within the eIF3a foci. No fluorescent foci were observed in heat-shocked strains expressing eIF2α-RFP (Sui2p-RFP), GFP alone or GFP fusions of the cytosolic metabolic enzymes Pgk1, Pfk1 and Pfk2 (Fig. 2D). Thus, we conclude that cytosolic accumulations of eIF3a, induced by robust heat shock in S. cerevisiae, contain components typical of mammalian SGs. Therefore we refer to them as SGs throughout the rest of this study.
P-body proteins colocalize with eIF3a-containing SGs in heat-shocked cells
Glucose deprivation and mildly elevated temperature (37°C) were shown to induce formation of enlarged cytosolic P-bodies (Brengues et al., 2005). Despite the fact that no eIF3a accumulations were observed upon glucose deprivation, we decided to investigate the relationship of the SGs with P-bodies upon robust heat shock. To do this, we constructed yeast strains coexpressing eIF3a-RFP together with GFP fusions of known P-body proteins from their chromosomal loci. First, we analyzed colocalization of eIF3a and the decapping enzyme Dcp2p that is a core component of P-bodies. In contrast to uniform cytosolic distribution of eIF3a, Dcp2p accumulated in small cytosolic foci in control cells cultivated in YPD medium at 30°C (Fig. 3A, control) as described previously (Brengues et al., 2005). Surprisingly, when these cells were heat-shocked in YPD medium at 46°C for 10 minutes, Dcp2p colocalization with eIF3a in multiple enlarged cytosolic foci (Fig. 3A, HS). The very high degree of colocalization of these fusion proteins in heat-shocked cells was exemplified by high Rr values of over 0.8. A similar pattern of colocalization in heat-shocked cells was also obtained for another P-body-specific marker Dhh1p (Fig. 3B). These findings thus suggest that components of SGs and P-bodies form joint accumulations when the fermenting cells are exposed to robust heat shock.
Differential centrifugation confirmed localization data from microscopic examination
To confirm the in vivo localization data biochemically, we performed immunochemical analyses of samples prepared from control and the heat-shocked cells by differential centrifugation using a protocol described elsewhere (Teixeira et al., 2005). Protein accumulations from the non-stressed versus heat-shocked cells were collected from the lysates as pellets and together with the corresponding supernatants subjected to western blotting (see the Materials and Methods). As shown in Fig. 4, the pellet from the heat-shocked cells was specifically enriched for eIF3a, Dcp2p, eIF4G2 and Pab1p, as expected from our fluorescence microscopy data. In addition, we also confirmed the presence of 40S ribosomal subunits (Rps30Ap) in the pellets and the absence of 60S ribosomal subunits (Rpl25p). Similarly, eIF2α, Pgk1p and Pfk1p, which did not occur in the SGs, were not pelleted from lysates of the heat-shocked cells. Hence, these biochemical data fully comply with our microscopic observations.
Inhibition of translation initiation and formation of SGs upon robust heat shock is Gcn2p independent
It is well known that inhibition of general translation upon various types of stress is mediated by phosphorylation of eIF2α, which is sufficient for formation of mammalian SGs under various stress conditions (Kedersha et al., 1999). In S. cerevisiae, Gcn2p is the only eIF2α kinase (Proud, 2005). Interestingly, it was previously shown that formation of P-bodies is independent of eIF2α phosphorylation (Kedersha et al., 2005). Thus we next decided to examine whether eIF2α phosphorylation is required for the formation of SGs in heat-shocked cells. As shown in Fig. 5A, incubation of fermenting wild-type GCN2+ cells at 46°C for 10 minutes (HS) resulted in marked phosphorylation of the α-subunit of eIF2 in contrast to the heat-shocked gcn2Δ cells. Accordingly, the GCN2+ cells incubated at 46°C for 10 minutes (HS) displayed a dramatic polysome run-off, resulting in accumulation of the 80S ribosomal species called monosomes (Fig. 5B). This redistribution of ribosomes from polysomes to monosomes serves as a hallmark of inhibition of translation initiation (Hartwell and McLaughlin, 1969). Strikingly, robust heat shock (HS) inhibited translation initiation in gcn2Δ cells (unable to phosphorylate eIF2α) to the same extent as in the wild-type GCN2+ cells. In agreement with the translational arrest induced by robust heat shock in gcn2Δ cells, formation of SGs was also found to be independent of the Gcn2 kinase (Fig. 5C). To conclude, our results indicate that in S. cerevisiae, the inhibition of translation initiation and formation of SGs triggered by robust heat shock are independent of eIF2α phosphorylation.
Cycloheximide prevents assembly of SGs in heat-shocked cells
Cycloheximide (CYH) causes stalling of polyribosomes on mRNA and as such has been shown to prevent formation of SGs in mammalian cells (Kedersha et al., 1999) and assembly of P-bodies or SGs in yeast cells starved of glucose (Brengues et al., 2005; Buchan et al., 2008). Therefore we also analyzed its effect on assembly of SGs during robust heat shock using the strain coexpressing eIF3a-RFP and Dcp2p-GFP from their chromosomal loci. We treated exponentially growing cells with CYH (50 μg/ml) in YPD medium at 30°C for 10 minutes before glucose deprivation or robust heat shock. In accordance with previously published data (Brengues et al., 2005), we observed the CYH-related inhibition of the P-body assembly in glucose-deprived cells (Fig. 6A). Similarly, no enlarged accumulations of either eIF3a or Dcp2p were formed when the culture of CYH-treated cells was subsequently heat-shocked in YPD at 46°C for 10 minutes (Fig. 6B). Assuming that CYH prevented formation of SGs by sequestration of mRNAs into polysomes, these results suggest that the protein as well as mRNA composition of SGs of heat-shocked cells is dynamic, similar to P-bodies and SGs of glucose-deprived yeast cells (Buchan et al., 2008; Sheth and Parker, 2003) and mammalian SGs (Kedersha et al., 2000) and, moreover, that their core components are also in equilibrium with polysomes.
Assembly of SGs upon robust heat shock is an energy-dependent process
To examine whether formation of heat shock-induced SGs in S. cerevisiae depends on the energy supply, we analyzed changes in distribution of Dcp2p and eIF3a upon robust heat shock in energy-depleted cells. Energy depletion was achieved by inhibiting both glycolysis by 2-deoxy-D-glucose (Franzusoff and Cirillo, 1982), and oxidative phosphorylation, by sodium-azide that blocks cytochrome C oxidase (Duncan and Mackler, 1966). Under normal conditions in fermenting cells, Dcp2p was localized into small cytosolic foci and eIF3a was uniformly cytosolic (Fig. 7A; control). We found that simultaneous treatment of these cells with 20 mM 2-deoxy-D-glucose and 20 mM sodium azide for 20 minutes did not significantly affect distribution of eIF3a; however, small cytosolic foci of Dcp2p enlarged into obvious P-bodies (Fig. 7A, ED). In contrast to heat-shocked fermenting cells in which there is colocalization of foci of Dcp2p and eIF3a (Fig. 7B, HS), the energy-depleted and subsequently heat-shocked cells (Fig. 7B, ED and HS) were unable to form SGs, but they displayed enlarged P-bodies characterized by accumulated Dcp2p. Interestingly, growth rate analyses of the cells recovering from the robust heat shock revealed that the outgrowth of ED/HS cells was significantly delayed by several hours when compared with cells that were only heat-shocked (Fig. 7C). These striking findings suggest that assembly of SGs in heat-shocked S. cerevisiae is, in contrast to P-bodies, an active process that requires energy. In addition, depletion of the ATP and/or GTP pool in living cells significantly delays the recovery phase of heat-shocked cells hinting that SGs could play a key role in the adaptation and resilience of yeast to robust heat shock.
Robust heat shock elicits formation of Dcp2p accumulations even in the absence of typical P-body scaffolding proteins Edc3p and Lsm4p that overlap with eIF3-containing SGs
The fact that eIF3a and Dcp2p colocalize, prompted us to elucidate whether formation of SGs in cells upon robust heat shock depends on the same scaffolding proteins as P-bodies. To this end, in a strain coexpressing eIF3a-RFP and Dcp2p-GFP from their chromosomal loci, we generated edc3Δ and lsm4ΔC mutations. The deletion of the EDC3 gene and the deletion of 97 amino acids from the C-terminal part of Lsm4p have been previously shown to prevent formation of P-bodies (Decker et al., 2007) and SGs (Buchan et al., 2008) in glucose-deprived yeast cells. Consistently, we found that glucose starvation of cells at 30°C resulted in formation of several enlarged P-bodies in wild-type cells but not in the mutant cells (Fig. 8A, top panels). The edc3Δ lsm4ΔC mutant cells had a stronger Dcp2-GFP signal in the cytosol as well as an obvious accumulation of Dcp2-GFP in the nuclear region. As expected, distribution of eIF3a-RFP was uniformly cytosolic in both strains starved of glucose for 30 minutes (Fig. 8A, middle panels). In the presence of glucose, wild-type fermenting cells had small cytosolic foci of Dcp2p-GFP, as expected (Fig. 8B, left-handed top panel), whereas the edc3Δ lsm4ΔC mutant cells showed a dominant accumulation of the Dcp2-GFP signal in the nuclear region (Fig. 8B, right-handed top panel). Distribution of eIF3a-RFP was again uniformly cytosolic in both strains grown in glucose as expected (Fig. 8B, middle panels). Strikingly, when both strains were first starved of glucose for 30 minutes and subsequently heat-shocked at 46°C for 10 minutes in the YP medium (in the absence of glucose), mainly enlarged P-bodies developed in the wild-type cells (Fig. 8C, left-handed top panel), whereas in the edc3Δ lsm4ΔC mutant cells there were only distinct accumulations of Dcp2-GFP in the vicinity of nuclei (Fig. 8C, right-handed top panel). In both strains, eIF3a-RFP clearly colocalized with Dcp2-GFP accumulations (Fig. 8C, middle panel). This novel and rather surprising finding indicates that the robust heat shock induces formation of Dcp2p accumulations qualitatively distinct from P-bodies, at least with respect to scaffolding proteins, which overlap with eIF3a-containing SGs. In perfect agreement, Dcp2p-GFP accumulations readily formed also in fermenting edc3Δlsm4ΔC mutant cells (in the presence of glucose) subjected to robust heat shock (Fig. 8D, right-handed top panel).
Assembly of SGs upon robust heat shock requires different scaffolding proteins than assembly of SGs in glucose-deprived cells
Recently, it has been shown that a prolonged glucose starvation results in assembly of yeast SGs that do not contain eIF3 and 40S ribosomal subunits (Buchan et al., 2008). In fact, these SGs were proposed to be identical to previously described accumulations called EGP bodies (Hoyle et al., 2007). To compare SGs assembled in glucose-deprived cells with eIF3-containing accumulations evoked by robust heat shock we constructed several strains coexpressing known yeast TIAR and TIA protein orthologs that were shown to constitute yeast SGs such as Ngr1-GFP or Pub1-GFP with eIF3a-RFP (Buchan et al., 2008). We found that both of these yeast SG markers expressed from their chromosomal loci colocalized with eIF3a in cells exposed to robust heat shock at 46°C in YPD medium for 10 minutes (Fig. 9A). As confirmed by high values of the Pearson's correlation coefficient of over 0.8, distribution of both Ngr1-GFP and Pub1-GFP signals almost completely matched that of eIF3a-RFP. Interestingly, in addition to that, both yeast SG markers were also enriched in the nuclear region of the stressed cells.
To test whether deletions of the P-body scaffolding proteins Lsm4 and Edc3 would or would not have an effect on the accumulation of Ngr1p into robust heat shock-induced SGs (the deletions do not affect accumulation of eIF3a), we generated the edc3Δ lsm4ΔC double-deletion strain coexpressing eIF3a-RFP with Ngr1-GFP from chromosomal loci. As predicted, we found that the heat-shocked edc3Δ lsm4ΔC mutant cells accumulated Ngr1-GFP in the eIF3-containing SGs (Fig. 9B). In addition to this colocalization pattern, we also observed cells with the Ngr1-GFP signal in the nuclear region. This may suggest an increased level of degradation of this yeast SG marker upon robust heat shock.
Pub1p but not Ngr1p was implicated in playing a crucial role in yeast SG formation under glucose deprivation (Buchan et al., 2008). To examine the role of Ngr1 and Pub1 proteins in the assembly of SG upon robust heat shock, we deleted the NGR1 and PUB1 genes in the strain coexpressing Dcp2-GFP and eIF3a-RFP from the chromosomal loci. As shown in Fig. 9C, both heat-shocked mutants displayed unchanged colocalization pattern of Dcp2-GFP foci with eIF3a-RFP accumulations when compared with wild-type cells (see Fig. 3A) suggesting that neither Pub1p nor Ngr1 are required for assembly of yeast SGs upon robust heat shock, in contrast to glucose-deprivation-induced SGs (Buchan et al., 2008).
Glucose starvation of heat shocked cells results in separation of the Dcp2-GFP signal from eIF3a-containing SGs
The fact that formation of eIF3 accumulations in fermenting cells upon robust heat shock requires different scaffolding proteins than assembly of P-bodies and SGs in glucose-deprived cells prompted us to investigate links between eIF3-containing SGs in glucose-fed fermenting cells and P-bodies and/or SGs formed in glucose-deprived cells. To do that, we used the strains coexpressing eIF3a-RFP and Dcp2-GFP or Pab1-GFP from their chromosomal loci. We first exposed the exponentially growing cells to robust heat shock at 46°C for 10 minutes in the YPD medium, washed them once with YP medium without glucose and cultured them in YP medium at 30°C for an additional 90 minutes. In sharp contrast to recovery after heat shock in the presence of glucose (see Fig. 1C), the absence of glucose in the medium prevented dissolution of eIF3a accumulations (Fig. 10). In these cells, the Dcp2-GFP signal did not overlap with the persistent eIF3a accumulations after cultivation for 30 minutes in the absence of glucose (Fig. 10A). By contrast, the marker of the yeast SGs, Pab1-GFP, remained closely colocalized with eIF3a-RFP even after 90 minutes of subsequent cultivation (Fig. 10B). Therefore our results provide strong support for the idea that SGs and P-bodies are intertwined but distinct accumulations (Mollet et al., 2008).
In stressed eukaryotic cells, non-translating mRNAs are sequestered into various transient granules or bodies depending on the type of stress condition that the cells are subjected to and the cell type (Anderson and Kedersha, 2008; Bond, 2006; Brengues et al., 2005; Buchan et al., 2008; Hoyle et al., 2007; Kedersha and Anderson, 2007; Malagon and Jensen, 2008; Mollet et al., 2008; Parker and Sheth, 2007). In this investigation we analyzed the impact of a robust heat shock and employed live cell imaging fluorescence microscopy to demonstrate for the first time that robust heat shock in budding yeast, S. cerevisiae, induces assembly of SGs more similar in their content to mammalian SGs than those that were described in glucose-deprived S. cerevisiae (Buchan et al., 2008) while this manuscript was under editorial review.
It should be emphasized that all tested proteins that were fused to either green fluorescent protein (GFP) or red fluorescent protein (RFP), were the only forms of those proteins expressed in the test cells and they fully supported growth. In control experiments, neither GFP alone nor GFP fusions of several cytosolic enzymes formed any visible aggregates or accumulations upon robust heat shock, indicating that the formation of large fluorescent foci is not a general phenomenon of protein aggregation resulting from intense heat treatment. Hence, we consider the ability of translational components to accumulate in detectable accumulations to be part of their function in translational control of stressed cells. Whereas studies on mammalian cells showed that SGs can be either deficient in eIF2α (Kedersha et al., 2002) or not (Kimball et al., 2003), the fact that eIF2α is not part of SGs in heat-shocked cells of S. cerevisiae clearly indicates that SGs are not the same as the heat shock granules containing eIF2α that were found as a result of prolonged heat treatment (60 minutes) in plant cells (Weber et al., 2008). In accord with the logic applied by Buchan et al. (Buchan et al., 2008), we assume that eIF2α does not accumulate with eIF3-containing SGs because it is believed to form yeast-specific structures with its GTP/GDP exchange factor eIF2B under certain stress conditions (Campbell et al., 2005).
In agreement with previously published data (Brengues et al., 2005), PGK1 mRNA was found uniformly distributed in the cytosol of unstressed exponentially growing cells. However, upon robust heat shock this mRNA clearly accumulated in eIF3 foci. Consistent with this, cycloheximide prevents formation of SGs upon robust heat shock in a similar way to that described for the formation of P-bodies (Brengues et al., 2005) and SGs in glucose-deprived yeast cells (Buchan et al., 2008). This suggests that mRNA may be an essential component in the formation of SGs upon robust heat shock, presumably by decreasing its availability. This conclusion is in keeping with the idea that under robust heat shock, at least some yeast mRNAs are protected in sub-cellular structures that are inaccessible to deadenylases, retaining a portion of the cytoplasmic mRNA pool for later use during recovery from the stress (Hilgers et al., 2006).
SGs of mammalian cells have already been shown to interact with P-bodies suggesting they are distinct but closely intertwined assemblies sharing components, depending on cell type and type of stress (Kedersha et al., 2005; Mollet et al., 2008; Wilczynska et al., 2005). In addition, preexisting P-bodies have been only recently shown to directly promote assembly of SGs in glucose-deprived S. cerevisiae cells (Buchan et al., 2008). We show here that in contrast to glucose deprivation, robust heat shock does not result in formation of typical enlarged P-bodies but rather in multiple small Dcp2p-GFP accumulations containing other P-body markers as well. The key question is whether these accumulations of Dcp2-GFP in heat-shocked cells can still technically be considered as a different form of P-bodies? We observed that both formations in question depend on the availability of mRNA, but can be easily distinguished by their requirement for different scaffolding proteins; Edc3p and Lsm4p are required for P-body formation in glucose-deprived cells but not for the formation of Dcp2p accumulations upon robust heat shock (see Fig. 8). Interestingly, similar requirements were also found in glucose-deprivation-induced SGs (Buchan et al., 2008) but not by SGs elicited by robust heat shock. Finally, it is also worth noting that eIF3-containing SGs in heat-shocked cells had distinct morphologies and a different requirement for energy than P-bodies, and that we observed a separation of Dcp2p and eIF3a signals in heat-shocked cells recovering in the glucose-free medium. Taken together, the results presented herein clearly indicate that Dcp2p accumulations and eIF3-containing SGs induced by robust heat-shock are also two discrete but closely intertwined assemblies similar to but differing substantially from the P-bodies and SGs induced by glucose deprivation as shown previously (Buchan et al., 2008). Since SGs and P-bodies are generally considered to be dynamic assemblies of mRNA-protein systems (mRNPs) (Brengues et al., 2005; Hoyle et al., 2007), we suggest that these apparent differences most probably reflect distinctions in the rate-limiting steps in a given mRNP transition process that probably changes under various stress conditions.
In keeping with this suggestion, assembly of mammalian SGs was shown to be dependent on the RNA binding protein TIA-1. Its overexpression induces SG formation without stress. Moreover, the absence of TIA-1 results in a strong impairment of SG formation under various stresses (Gilks et al., 2004). S. cerevisiae contains potential TIA, TIAR and ATXN2 orthologs, namely Pub1p, Ngr1p and Pbp1p, respectively. Whereas pub1Δ and ngr1Δ strains showed a strong decrease in the number of SGs in glucose-deprived cells (Buchan et al., 2008), we found that the absence of either Pub1p or Ngr1p had no significant impact on the assembly eIF3-containing SGs following robust heat shock. These findings thus may suggest that either the function of the latter proteins is more redundant under robust heat shock conditions or that other non-homologous scaffolding proteins are involved.
Despite the fact that yeast SGs that are formed upon robust heat shock seem to contain most if not all components of mammalian SGs, their formation is eIF2α-phosphorylation independent in the budding yeast, unlike the situation in mammals (Kedersha et al., 1999). Actually, in this respect, yeast SGs formed in heat-shocked cells more closely resemble P-bodies of glucose-deprived cells, the assembly of which also does not require phosphorylation of eIF2 (Kedersha et al., 2005). Nevertheless, the most recent observation from trypanosomes showing that assembly of SGs in response to heat shock was also found to be independent of phosphorylation of eIF2α (Kramer et al., 2008) might indicate that either the biochemistry of the SG assembly induced by heat shock differs substantially from that induced by other stresses (perhaps in involvement of heat shock proteins) or that higher eukaryotes acquired some novel aspect(s) of this control mechanism that has imposed a need for eIF2α-phosphorylation. Since interfering with the activity of eIF4G (Mazroui et al., 2006) or eIF4A (Dang et al., 2006; Mazroui et al., 2006) seems to trigger SG formation independently of eIF2α phosphorylation also in mammalian cells, these factors could play some role in this process.
Along these lines, it was only recently shown that several components of the hexosamine biosynthetic pathway, which reversibly modifies proteins with O-linked N-acetylglucosamine (O-GlcNAc) in response to stress, are required for accumulation of untranslated messenger ribonucleoproteins in SGs in mammalian cells (Ohn et al., 2008). However, the hexosamine biosynthetic enzymes are lacking in budding yeast and the authors suggested that this fact may contribute to differences between mammalian SGs and their closest relatives known in yeast so far, the SGs without eIF3 and 40S ribosomal subunits. Hence, in analogy, it could be proposed that the absence of these proteins makes the yeast translational machinery somewhat more resistant to mild stresses, explaining why accumulation of eIF3a was not previously observed in S. cerevisiae cells heat-shocked at 42°C, or subjected to hyperosmotic stress (Brengues et al., 2005) or glucose deprivation (Buchan et al., 2008). It will be intriguing in the future to investigate what differentiates yeast cells from other cell types that produce typical SGs in response to robust heat shock, and what roles these accumulations play in stress survival.
Materials and Methods
Yeast strains and growth conditions
S. cerevisiae strains used in this study were derived either from the BY4742 background, S288C background (Huh et al., 2003) or SEY6210 background (Scott Emr, Cornell University, Ithaca, NY) and are listed in Table 1. Yeast cultures were grown in YPD medium (1% yeast extract, 2% peptone, 2% glucose) or SC medium (0.17% yeast nitrogen base medium without amino acids and ammonium sulfate, 0.5% ammonium sulfate, 2% glucose, supplemented with a complete or a strain-specific mixture of amino acids) at 30°C. Solid media contained 2% agar. Standard methods were used for all DNA manipulations (Sambrook and Russell, 2001). The cells coexpressing U1A-GFP and PGK1 mRNA from the corresponding plasmids were pre-cultured in an appropriate SC medium after several inoculation steps (1:10) followed by an 8-hour culture to increase the homogeneity of the U1A-GFP distribution pattern within the cell population. To subject yeast strains to robust heat shock, cells were resuspended in YPD medium preheated to 46°C and incubated for an additional 10 minutes while shaking. Cells were always washed with SC medium before being mounted for microscopic inspection. Cycloheximide (Sigma) was added at a final concentration of 50 μg/ml. Inhibitors 2-deoxy-D-glucose (Sigma) and sodium azide (Sigma) were added as stock aqueous solutions to a final concentration of 20 mM into cultivation media.
Construction of strains with chromosome-derived expression of GFP or RFP fusions
Integrative cassettes containing various RFP or GFP fusions were created by PCR using template DNA pRFPKanMX or pGFPKanMX, respectively, as described elsewhere (Malinska et al., 2003). Purified PCR fragments were transformed into SEY6210 wild-type cells and the corresponding transformants were selected on YPD plates containing 200 μg/ml of G418. The correct integration of the RFP or GFP fusions was confirmed by PCR. The strains were generated by mating, subsequent sporulation in liquid Fowel medium and spore dissection using the Singer micromanipulator.
The mutant strain edc1Δ lsm4ΔC expressing DCP-GFP and eIF3a-RFP from chromosomal sites was constructed by the repeated one-step gene disruption technique (Rothstein, 1991). The deletion cassettes loxP-URA3-loxP and loxP-LEU2-loxP were amplified from pUG72 and pUG73, respectively (Gueldener et al., 2002), using the ORF-specific primers. The lsm4ΔC designates a partial deletion of the C-terminal 97 amino acids of the Lsm4p (Decker et al., 2007).
Polysome profile analyses
Cells were grown to an OD600. Approximately 1 and 50 μg/ml cycloheximide (Sigma) was added to a culture 5 minutes before harvesting. Cells were chilled and washed in GA buffer (20 mM Tris-HCl pH 7.5, 50 mM KCl, 10 mM MgCl2, 1 mM DTT, 5 mM NaF) containing 50 μg/ml cycloheximide. Lysates were prepared in GA buffer (supplemented with 1 tablet/10 ml of Complete Mini Protease Inhibitor Mix EDTA-free (Roche) and 50 μg/ml cycloheximide) using glass beads and Fastprep Bio101 at speed 5 for 20 seconds. Lysates were pre-cleared twice by centrifugation at 826 × g for 5 minutes and 15,520 × g, for 15 minutes (Jouan AB 2.14, France) and loaded onto 5-45% sucrose gradient. Gradients were ultracentrifuged in a SW41 rotor for 2.5 hours at 260,000 × g at 4°C. Gradients were collected from the bottom and A254 was recorded. Fractions of the sucrose gradient were precipitated with trichloroacetic acid, washed several times with ethanol and used for western blot analysis.
Analyses of protein accumulations
Exponentially growing cells were heat-shocked at 46°C for 10 minutes and harvested. Cells were washed and re-suspended in the lysis buffer (Teixeira et al., 2005) containing 50 mM Tris-HCl (pH 7.6), 50 mM NaCl, 5 mM MgCl2, 0.1% NP-40, 1 mM β-mercaptoethanol and 1 tablet/10 ml of Complete Mini Protease Inhibitor Mix EDTA-free (Roche). Disruption of cells was carried out in Fastprep (Bio101-Thermo Savant, Savant Instruments, Waltham, MA) twice at speed 4 for 20 seconds. Cell debris were pelleted at 2296 × g for 10 minutes at 4°C. Supernatant was centrifuged at 18,000 × g (Jouan AB 2.14, Saint-Herblain, France) for 10 minutes at 4°C. Pellets and supernatants were analyzed by western blotting for the presence of translation factors and P-body markers.
Western blot analyses
Except for detection of eIF2α phosphorylation, whole cell lysates for western blot analyses with various antibodies were prepared according to a protocol as described elsewhere (Riezman et al., 1983). Proteins resolved by SDS-PAGE were transferred to Protran (Sigma) nitrocellulose membranes. The membrane blots were blocked with 5% non-fat dried milk and incubated overnight with the antibodies. The antibodies used were: mouse monoclonal anti-Rpg1 antibody PK1 (Jirincova et al., 1998) at 1:10,000; mouse monoclonal anti-GFP antibody (Unifect, Moscow, Russia) at 1:2000; rabbit anti-eIF2α/Sui2p antibody (kindly provided by Thomas Dever, NICHHD, Bethesda, MD) at 1:1000. As a secondary antibody, goat anti-mouse IgG antibodies or goat anti-rabbit IgG antibodies conjugated with a horseradish peroxidase (Amersham Pharmacia Biotech, Uppsala, Sweden) were used.
To detect eIF2α phoshorylation, cells were grown to mid-logarithmic phase (OD600 ∼0.5) in YPD medium and either kept at the permissive temperature or exposed to heat shock at 46°C for 10 minutes. Subsequently, the cells were collected by centrifugation, re-suspended in ice-cold breaking buffer (100 mM Tris-HCl, pH 8.0, 20% glycerol, 1 mM 2-mercaptoethanol, protease inhibitors (Complete Mini Protease Inhibitor Mix plus EDTA; Roche) and phosphatase inhibitors (Inhibitor Cocktail I and II, Sigma)) and lysed using glass beads in Fastprep (Bio101-Thermo Savant) at speed 5, four times 20 seconds at 4°C. Lysates were then pre-cleared at 2000 × g to remove glass beads and cell debris and centrifuged at 15,000 × g for 30 minutes at 4°C. Supernatants were collected and separated by electrophoresis on 12% acrylamide SDS-PAGE gels and transferred to PVDF membranes. Western blot analyses were carried out using the rabbit polyclonal phosphospecific antibody against eIF2α (Biosource, Invitrogen) and the rabbit anti-eIF2α antibody (kindly provided by T. Dever).
The cells were inspected after washing with SC or SC-glucose medium, mounting on coverslips and coating with a slice of 1% agarose in appropriate medium. Distribution of GFP and RFP fusion proteins was analyzed with a 100× PlanApochromat objective (NA 1.4) using an Olympus IX-71 inverted microscope equipped with a Hammamatsu Orca/ER digital camera and the Olympus Cell R detection and analyzing system (GFP filter block U-MGFPHQ, exc. max. 488, em. max. 507; RFP filter block U-MWIY2, exc. max. 545-580, em. max. 610). Images were processed and merged using Olympus Cell-R and Adobe CS2 software. The quantitative colocalization analyses were performed using the NIH ImageJ software with the Colocalization Finder plugin, available at http://rsb.info.nih.gov/ij/plugins/. This software was used to determination the Rr, which describes the extent of overlap between image pairs. It is a value between –1 and +1, with –1 being no overlap and +1 being perfect overlap of two images.
We are grateful to Mark Ashe, Anna Koffer and Michael Breitenbach for critical reading of manuscript and helpful comments. The technical assistance by J. Serbouskova and M. Spryngar is also gratefully acknowledged. The plasmid used for the PCR preparation of particular RFP integrative cassettes was kindly provided by Roger Tsien (Howard Hughes Medical Institute, University of California, San Diego, CA). The plasmid pRS316-L25e-GFP is a kind gift from Eduard Hurt (University of Heidelberg, Heidelberg, Germany); plasmids pRP2037 and pRP1187 for the PGK1 mRNA detection were kind gifts from Roy Parker (University of Arizona, Tuscon, AZ). The plasmid pUG23 was kindly provided by Johannes Hegemann (Heinrich Heine University, Düsseldorf, Germany). This work was supported by grants from the Czech Science Foundation 204/02/1424, 204/05/0838 and 204/09/1924, L545 (MSMT), ME 939, and also by the Institutional Research Concept No. AV0Z50200510. P.I. was supported by grant INTAS 05-109-4807 and L.V. was supported by a Jan E. Purkyne Fellowship from Academy of Sciences of the Czech Republic.
↵* Present address: A.N. Belozersky Institute of Physico-Chemical Biology MSU, Moscow, Russia
- Accepted March 18, 2009.
- © The Company of Biologists Limited 2009