The RNA Helicase Ded1 Regulates Translation and Granule Formation during Multiple Phases of Cellular Stress Responses

ABSTRACT

Ded1 is a conserved RNA helicase that promotes translation initiation in steady-state conditions. Ded1 has also been shown to regulate translation during cellular stress and affect the dynamics of stress granules (SGs), accumulations of RNA and protein linked to translation repression. To better understand its role in stress responses, we examined Ded1 function in two different models: DED1 overexpression and oxidative stress. DED1 overexpression inhibits growth and promotes the formation of SGs. A ded1 mutant lacking the low-complexity C-terminal region (ded1-ΔCT), which mediates Ded1 oligomerization and interaction with the translation factor eIF4G1, suppressed these phenotypes, consistent with other stresses. During oxidative stress, a ded1-ΔCT mutant was defective in growth and in SG formation compared to wild-type cells, although SGs were increased rather than decreased in these conditions. Unlike stress induced by direct TOR inhibition, the phenotypes in both models were only partially dependent on eIF4G1 interaction, suggesting an additional contribution from Ded1 oligomerization. Furthermore, examination of the growth defects and translational changes during oxidative stress suggested that Ded1 plays a role during recovery from stress. Integrating these disparate results, we propose that Ded1 controls multiple aspects of translation and RNP dynamics in both initial stress responses and during recovery.

INTRODUCTION

Organisms are frequently subjected to various adverse conditions, including a lack of nutrients, chemical imbalances, and exposure to toxic substances. To survive and adapt to these stresses, cells use a variety of responses, including highly coordinated changes in gene expression (1–4). Changes in translation play a particularly large role in stress responses because of their ability to rapidly alter the proteome and the need to reduce the outsized energy requirements of protein synthesis during stress (5, 6). Translation of most mRNAs, especially those associated with growth, is greatly diminished during stress conditions. On the other hand, “stress-responsive” genes are upregulated, and ribosome profiling of stressed cells has shown increases in occupancy at upstream open reading frames (uORFs) and non-AUG initiation (7, 8). These data suggest that the translational stress response is complex and specific, but the sources of this specificity are not fully clear.

A common feature of many types of stress conditions is the formation of stress granules (SGs), nonmembranous organelles composed of RNAs and proteins (9). SGs appear to form as a result of multiple interactions among these components, often including proteins with low-complexity, intrinsically-disordered regions (IDRs) that promote liquid-liquid phase separation (10). Despite close association with both stress conditions and translation repression, the function of SGs is not well understood, although mutations in genes required for SG formation also reduce cell survival in stress (9). SGs are often suggested to function as sorting and storage sites for mRNAs during stress (11). Consistent with this hypothesis, RNA-seq and localization studies have shown that some mRNAs are enriched in SGs while others are depleted, and mRNAs are able to actively exchange between SGs, the cytosol, and other structures (12–14). The mRNAs in SGs have generally been considered to be translationally repressed, although a recent study has suggested that this may not always be the case (15).

Ded1 is a translation factor in budding yeast that plays several different roles in translation initiation (16, 17). Ded1 (DDX3X in humans) is a member of the DEAD box RNA helicase family of proteins, which utilize ATP to alter RNA-RNA and RNA-protein interactions and are critical for many gene expression processes (18). Similar to many DEAD box proteins, the Ded1 domain structure consists of a central helicase core flanked by long extensions that are predicted to be IDRs (Fig. 1A). These N- and C-terminal regions mediate binding to other proteins, including members of the eIF4F translation complex, and oligomerization of Ded1 itself (19–23). Canonically, Ded1 stimulates translation initiation in steady-state growth conditions by unwinding secondary structure in 5′ untranslated regions (5′UTRs), UTRs, which facilitates start site scanning by the translation pre-initiation complex (PIC) and makes mRNAs with more complex, structured 5′UTRs UTRs hyperdependent on Ded1 activity (24, 25). Furthermore, ded1 mutation or depletion results in utilization of alternative translation initiation sites (ATIS) that may affect downstream translation or protein function (25). Ded1 also promotes assembly of the 43S PIC on mRNA, again in an mRNA-specific manner (19, 23, 26, 27).

FIG 1.

Growth inhibition by DED1 overexpression correlates with Ded1 protein levels and requires the C-terminal region. (A) Domain map of the Ded1 protein illustrating the N- and C-terminal low-complexity regions (NT and CT, respectively) and the boundaries of five sequential deletions of 14 residues within the CT analyzed in this study. (B) Fivefold serial dilutions of ded1-ΔCT cells with galactose-inducible wild-type DED1 spotted on selective medium containing either galactose or glucose. Cells harbored a single plasmid that encoded wild-type Ded1 with a His/HA/protein A C-terminal tag (GAL-DED1-HHA), a single plasmid encoding untagged Ded1 (GAL-DED1/HIS3 or GAL-DED1/TRP1), or two plasmids that each encoded untagged Ded1 (GAL-DED1/HIS3 + GAL-DED1/TRP1). (C) Fivefold serial dilutions of ded1-ΔCT cells with galactose-inducible DED1 or the indicated ded1 mutants spotted on selective medium containing either galactose or glucose. (D) Western blot analysis of protein extracts from ded1-ΔCT cells containing the indicated plasmid, induced for 7 h. Samples were probed with antibodies specific for Ded1 or Pgk1 (as a loading control). Ded1 levels were quantified by densitometry, normalized to Pgk1 levels in each sample. The mean from 3 or 4 biological replicates is shown.

Ded1 and its orthologs have also been implicated in translation repression during stress conditions and are associated with SGs in particular. Ded1 and DDX3X are major protein components of SGs and can affect SG assembly (19, 28, 29). Overexpression of DED1 alone can induce SG-like foci in cells (19), while Ded1 undergoes phase separation in vitro under conditions of high Ded1 concentration, elevated temperature, and/or the presence of RNA (30, 31). These effects are at least partially mediated by the N- and C-terminal disordered regions (19, 30, 31). Overall, Ded1 appears to promote SG formation, although the consequences of this stimulation have not been fully defined. Iserman et al. have proposed that sequestration of Ded1 into granules during heat shock causes a switch in translation to mRNAs with less complex 5′UTRs UTRs, but this model has not been fully tested (31).

Recently, we showed that Ded1 has a role in the translational response downstream of the target-of-rapamycin (TOR) pathway, which serves as a central integrator of growth and stress in cells (32). Specifically, when the TOR pathway was downregulated, Ded1 was required for efficient translation repression and growth inhibition. In particular, the low-complexity C-terminal region of Ded1 was necessary for these effects, which promoted degradation of its binding partner eIF4G1. We proposed that in these conditions, Ded1 remodeled translation complexes to cause dissociation and degradation of eIF4G1, thus reducing bulk translation during stress. However, it is currently unclear how Ded1’s role in SG dynamics may affect this model, particularly in different cellular stresses. Here we sought to address these gaps by examining Ded1-dependent mechanisms of translation regulation in two different conditions. First, upon DED1 overexpression, both growth inhibition and SG formation were dependent on Ded1 levels and the presence of the C-terminal region but not helicase activity. Furthermore, Ded1 interaction with eIF4G1 had only a modest effect, suggesting a contribution from Ded1 oligomerization. Second, when cells were subjected to oxidative stress through addition of peroxide to the media, the C-terminal region played critical roles both in survival as well as the ability of cells to recovery from the stress over time. Interestingly, deletion of the C-terminal region increased SG formation during oxidative stress, in a manner opposite to the overexpression results. Similar to overexpression, Ded1 interaction with eIF4G1 played a moderate role in responding to oxidative stress. Consistent with the effects on growth and SG formation, reporter assays revealed that Ded1 and its C-terminal region mediate changes in translation during a time course of peroxide treatment. To integrate these results, we propose a biphasic model for the function of Ded1 in the stress response wherein it has distinct functions on cell survival/growth, translation, and SGs in an early response phase and during a later recovery/adaptation phase.

RESULTS

DED1 growth inhibition is dependent on protein levels and the C-terminal region but not the helicase domain.

We recently showed that Ded1 plays a role in the translational response to TOR pathway inhibition in a manner dependent on its C-terminal domain and interaction with eIF4G1 (32). To examine whether this mechanism is similar during other conditions of translation repression, including SG induction, we first used overexpression of DED1. Initial experiments were performed in cells with a ded1 mutation deleting the C-terminal domain, ded1-ΔCT (Δ536-604), to minimize the effect of the endogenous protein. This mutant displays no growth phenotype in normal conditions at 30°C (32), and DED1 overexpression in wild-type cells showed similar results (see below). Hilliker et al. previously demonstrated that DED1 overexpression causes growth inhibition and formation of SG-like aggregates, dependent on the presence of conserved “assembly domains” (19). Likewise, we observed severe growth inhibition upon overexpression of tagged, wild-type DED1 from a galactose-inducible promoter on a high-copy-number plasmid (Fig. 1B, GAL-DED1-HHA). However, when untagged DED1 was expressed from the same plasmid backbone, the growth inhibition, while still present, was much less severe (Fig. 1B, GAL-DED1/HIS3). Western blotting of the expressed constructs showed that protein levels of the tagged Ded1 were higher than in the untagged version, suggesting that the C-terminal tag stabilizes the protein (Fig. 1D). This result further suggests that growth inhibition is quite sensitive to Ded1 protein levels. Supporting this idea, galactose-induced overexpression of untagged DED1 from two high-copy-number plasmids in the same cells caused a greater decrease in growth, similar to tagged DED1 (Fig. 1B; GAL-DED1/HIS3 + GAL-DED1/TRP1). Untagged DED1 was used for the remainder of this study unless otherwise noted.

We next sought to determine whether the functional requirements we identified in TOR pathway downregulation also affect DED1-mediated growth inhibition. Consistent with previous results (19), overexpression of the ded1-ΔCT mutant was largely unable to inhibit growth (Fig. 1C). In our prior study, a smaller deletion of the final 14 amino acids of the C-terminal region (ded1-Δ591-604), which greatly reduced binding to eIF4G1 in vitro, phenocopied the larger deletion (32). In contrast with these results after TOR inhibition, overexpressing the ded1-Δ591-604 mutant inhibited growth much more strongly than ded1-ΔCT (Fig. 1C), indicating that there are differences between the Ded1-dependent mechanisms. Furthermore, we constructed a mutant that deleted most of the central helicase domain, ded1-Δ190-497, leaving the N- and C-terminal domains fused together with only short flanking sequences (Fig. 1A). This mutant inhibited growth to about the same extent as wild-type DED1 (Fig. 1C). It was previously shown that growth inhibition did not require Ded1 activity, as an ATPase-deficient mutant had a similar phenotype to wild-type DED1 (19), and this result extends that conclusion by suggesting that the N- and C-terminal regions themselves are sufficient for these effects. Again, this differs from our prior results in TOR inhibition wherein Ded1 activity was required for the effects on translation repression (32). Western blotting showed that minor differences in expression of these mutants were not sufficient to explain these results (Fig. 1D).

To attempt to determine which regions of the Ded1 C-terminus are critical for the growth inhibition, we tested a set of serial 14 amino acid deletions in the C-terminal region (Fig. 1A), either alone and in combination with a wild-type GAL-DED1 plasmid (Fig. 2A and B). All of these ded1 mutants showed growth inhibition similar to wild-type DED1. This included the ded1-Δ591-604 mutant, which showed, at best, a minor difference compared to wild-type DED1 when expressed alone (Fig. 1C) but no difference when coexpressed with a wild-type copy (Fig. 2B). To further investigate the critical regions of the C terminus, we generated two additional mutants, ded1-Δ546-557 and -W603/604A, that target conserved residues in this region (Fig. 2C). Overexpression of these mutants showed growth inhibition similar to, and perhaps slightly greater than, wild-type DED1, unlike the full C-terminal deletion mutant (Fig. 2D). These results suggest that the mechanism of growth inhibition is distinct in DED1 overexpression and is dependent on the length of the C-terminal region rather than residue-specific interactions.

FIG 2.

Growth inhibition by I overexpression is largely unaffected by more specific mutations of the C terminus. (A) Fivefold serial dilutions of ded1-ΔCT cells containing galactose-inducible DED1 with or without a C-terminal tag (GAL-DED1-HHA or GAL-DED1, respectively) or the indicated ded1 mutant (untagged), spotted on selective medium containing either galactose or glucose. (B) Fivefold serial dilutions of ded1-ΔCT cells containing galactose-inducible DED1 (GAL-DED1/TRP1) plus a second galactose-inducible DED1 or ded1 mutant plasmid as indicated, spotted on selective medium containing either galactose or glucose. (C) Amino acid alignment of the C-terminal regions of Ded1 and orthologs in D. melanogaster (Belle) and H. sapiens (DDX3X). Conserved residues are denoted in yellow below. The conserved 546 to 557 and 603 to 604 regions are highlighted with a gray background. (D) Fivefold serial dilutions of ded1-ΔCT cells containing galactose-inducible untagged DED1 (GAL-DED1) or the indicated ded1 mutants, spotted on selective medium containing either galactose or glucose.

Ded1-induced SG formation is dependent on the C-terminal region but not the helicase domain.

Overexpression of DED1 has been shown to induce SG-like foci that contain a number of SG components, including mRNAs and translation factors, and formation of the Ded1-induced SGs correlates with growth inhibition (19). Therefore, we tested whether SG formation was affected by the ded1 mutations examined above using a PAB1-GFP reporter, a well-established yeast SG marker (33). After 7 h of galactose induction, DED1-overexpressing cells frequently contained one or more Pab1-GFP-positive foci, indicating SG formation in 24% of cells, while very few SGs were observed in control cells (Fig. 3A and B). Further increasing DED1 levels with a second galactose-inducible plasmid led to an additional increase in the percentage of cells with SGs, consistent with the inhibitory effects on growth. In contrast, cells expressing the ded1-ΔCT mutant contained almost no SGs, similar to the control cells. However, both the ded1-Δ591-604 and Δ190-497 mutants both displayed rates of SG induction similar to the wild-type DED1-expressing cells (Fig. 3B). Interestingly, we noted that the Pab1-GFP foci induced by the ded1-Δ190-497 mutant had a somewhat different qualitative appearance with individual foci more elongated and extended compared to the largely round SGs in the wild-type DED1-expressing cells (Fig. 3C). This phenotype is also distinct from that with the tagged version of DED1, which were often very large but were still more rounded than the ded1-Δ190-497 mutant. Importantly, Pab1-GFP levels were similar between the different strains, ruling out an effect from differing expression of the SG marker (Fig. 3D). Overall, these results indicate that the Ded1 C-terminal region plays an important role in inducing SGs, while, surprisingly, the helicase domain does not. Furthermore, the SG results closely correlate with the growth inhibition shown in Fig. 1, suggesting a functional relationship between growth inhibition and SG formation.

FIG 3.

The Ded1 C terminus is required for formation of GAL-DED1-induced granules. (A) Live-cell microscopy showing granules in ded1-ΔCT cells containing a PAB1-GFP plasmid and expressing galactose-inducible wild-type DED1 from a single plasmid construct (GAL-DED1), two inducible constructs (2X GAL-DED1), or the indicated ded1 deletion mutant constructs. Cells were grown in liquid medium containing galactose for 7 h before imaging. Scale bar = 2 μm. (B) Quantitation of the presence of Pab1-GFP granules as the percentage of cells that contained GFP-positive foci. Mean and SEM from 3 to 11 replicates are shown. Statistical significance was determined using Student's t test (unpaired; *, P < 0.05; **, P < 0.01). (C) Close-up of microscopy showing Pab1-GFP granules in ded1-ΔCT cells expressing galactose-inducible wild-type DED1, ded1-Δ190-497, or DED1-HHA (encoding tagged Ded1). Cells were grown in liquid medium containing galactose for 7 h before imaging. Note the elongated morphology of the granules in ded1-Δ190-497 and large size of granules in DED1-HHA cells. Scale bar = 2 μm. (D) Western blot analysis of protein extracts from ded1-ΔCT cells expressing galactose-inducible DED1 or ded1 mutant constructs and PAB1-GFP, grown in liquid medium containing galactose for 7 h. Samples were probed with antibodies specific for Pab1 or Pgk1. Pab1-GFP levels were quantified via band densitometry, normalized to Pgk1 levels in each sample. The mean of 3 biological replicates is shown.

We also examined the localization of Ded1 itself in the overexpression model. To do so, we first coexpressed DED1-GFP on a low-copy-number plasmid along with GAL-DED1 or GAL-ded1-ΔCT overexpression plasmids (Fig. 4A). In control samples lacking the galactose-inducible constructs, Ded1-GFP was largely diffuse throughout the cytoplasm, although Ded1 foci were also observed in 8% of cells, likely due to cell-to-cell differences in plasmid copy number and expression (Fig. 4C). The percentage of cells with Ded1-GFP foci greatly increased (to 49%) upon overexpression of wild-type DED1 (Fig. 4A and C). On the other hand, overexpression of ded1-ΔCT resulted in only background levels of Ded1-GFP foci, consistent with the lack of Pab1-containing SGs under these conditions. We also examined localization of ded1-ΔCT-GFP (Fig. 4B). Interestingly, no foci containing the mutant protein were observed in control cells, while ded1-ΔCT-GFP foci were still increased in cells overexpressing wild-type DED1, albeit reduced compared with wild-type Ded1-GFP (Fig. 4D). Overall, these results are consistent with previous results that Ded1 localizes to the SG-like foci induced by overexpression (19). In addition, while the ded1-ΔCT is defective in inducing the SGs, it is still able to localize to them, perhaps through interaction with other translation factors.

FIG 4.

GAL-DED1-induced granules do not form in the absence of the C-terminus, but Ded1-ΔCT can still localize to granules. (A) Live-cell microscopy using ded1-ΔCT cells expressing DED1-GFP along with GAL-DED1, GAL-ded1-ΔCT, or vector control. Cells were grown in liquid medium containing galactose for 7 h before imaging. (B) Images of cells expressing either DED1-GFP or ded1-ΔCT-GFP together with GAL-DED1 or vector control. Cells were grown in liquid medium containing galactose for 7 h before imaging. (C, D) Percentages of cells containing Ded1-GFP foci from the conditions in panels A and B were quantified. Mean and SEM of 3 replicates are shown.

Ded1 interaction with eIF4G1 has moderate effects on DED1 overexpression phenotypes.

The C-terminal domain of Ded1 has been shown to both interact with eIF4G1 and mediate Ded1 self-oligomerization, but how these individual interactions affect Ded1 function in vivo remain unclear (19–21). To begin to distinguish between these interactions during cellular stress, we constructed eIF4G1-null (tif4631Δ) mutants (in a wild-type DED1 background to avoid synthetic effects with ded1-ΔCT), overexpressed DED1 and the ded1 mutants in these cells, and examined them for defects in growth and SG formation. DED1 overexpression inhibited growth to a similar extent in the eIF4G1-null mutant compared to in wild-type cells, indicating that the Ded1-eIF4G1 interaction is not critical for this inhibition (Fig. 5A). However, rescue of the inhibition with expression of the ded1-ΔCT mutant was greatly reduced in the eIF4G1-null cells. A double mutant of tif4631Δ and ded1-ΔCT (expressed from the endogenous promoter) showed moderate synthetic growth defects even in ideal growth conditions (data not shown), so the effect of overexpressing the ded1 mutant in eIF4G1-null cells may be due to displacing the endogenous wild-type Ded1 rather than a stress-related defect. Western blotting showed roughly similar levels of Ded1, ded1-ΔCT, and ded1-Δ591-604 in wild-type and tif4631Δ cells, arguing against effects due to protein levels for these mutants (Fig. 5B). The induction of ded1-Δ190-497 protein was reduced roughly 2-fold in eIF4G1-null cells, perhaps indicating an effect of eIF4G1 on protein expression or stability, although the difference was not statistically significant.

FIG 5.

Deletion of eIF4G1 has moderate effects on GAL-DED1-induced growth inhibition. (A) fivefold serial dilutions of wild-type TIF4631 (eIF4G1) or tif4631Δ (eIF4G1-null) cells with galactose-inducible wild-type DED1, ded1-ΔCT, or ded1-Δ591-604 spotted on selective medium containing either galactose or glucose. Galactose plates were imaged after 2 days of growth. (B) Western blot analysis of protein extracts from TIF4631 or tif4631Δ cells expressing galactose-inducible DED1 or ded1 mutant constructs and PAB1-GFP, grown in liquid medium containing galactose for 7 h. Samples were probed with antibodies specific for Ded1, eIF4G1, Pab1, or Pgk1. Ded1 and Pab1-GFP levels were quantified via band densitometry, normalized to Pgk1 levels in each sample. The mean from 3 to 5 biological replicates is shown. None of the values were significantly different by Student's t test.

Examining these cells for SG formation revealed no significant difference between wild-type and eIF4G1-null cells when wild-type DED1 is overexpressed (Fig. 6A and B). Likewise, very few Pab1-GFP SGs were formed in eIF4G1-null cells when ded1-ΔCT was expressed, similar to wild-type cells. These results indicate that eIF4G1 is not strictly required for formation of the Ded1-induced SGs. However, expression of the ded1-Δ591-604 and ded1-Δ190-497 mutants had reduced numbers of SGs in eIF4G1-null compared to wild-type cells (Fig. 6A and B). Pab1-GFP levels did not significantly differ between the different mutants (Fig. 5B). Overall, these results suggest that the mutants represent sensitized backgrounds that show that eIF4G1 indeed has an effect on SG formation, though it is moderate. There are two eIF4G paralogs in yeast, eIF4G1 and eIF4G2 (TIF4632), so it is possible that eIF4G2 is compensating for the loss of eIF4G1 in these experiments. However, Ded1 has not been reported to bind eIF4G2, and deletion of both paralogs is lethal. Rather, given these moderate effects and the difference in SGs between ded1-ΔCT and ded1-Δ591-604, it is likely that Ded1 oligomerization, the only other known C-terminal-mediated interaction, is a major contributor to SG formation.

FIG 6.

Deletion of eIF4G1 has moderate effects on GAL-DED1-induced granule formation. (A) Live-cell microscopy of Pab1-GFP granules in TIF4631 or tif4631Δ cells expressing the indicated galactose-inducible DED1 or ded1 deletion mutant constructs. Cells were grown in liquid medium containing galactose for 7 h before imaging. Scale bar = 2 μm. (B) Quantitation of the presence of Pab1-GFP granules as the percentage of cells that contained Pab1-GFP foci. Mean and SEM from 3 to 7 replicates are shown. Statistical significance was determined using Student's t test (unpaired; *, P < 0.05; ***, P < 0.001).

Ded1 promotes cell survival and growth during oxidative stress.

The above approaches provided insight into Ded1-dependent mechanisms for SG formation and growth inhibition. However, overexpression of DED1 does not recapitulate the complexity of physiological stress responses; therefore, we next examined the role of Ded1 in responding to oxidative stress through hydrogen peroxide treatment of cells expressing normal DED1 levels. We first tested whether Ded1 affects cell growth by measuring culture density over time in the absence and presence of peroxide (Fig. 7A). In untreated cells, wild-type, ded1-ΔCT, and ded1-Δ591-604 cells had similar growth rates (left panel). In peroxide-treated cultures (right panel), wild-type cells had a growth lag (λ) of about 11 h, as calculated using the Gompertz growth equation (Fig. 7B) (34). This indicated that growth was inhibited as part of the stress response, followed by an adaptation/recovery phase during which growth resumed. In contrast, the lag in ded1-ΔCT cells was significantly longer (22 h) before growth resumed. Interestingly, the ded1-Δ591-604 mutant also showed an increased growth lag compared to wild-type cells (Fig. 7A and B), unlike the very minor effect in overexpression, although the defect was not as strong as in the ded1-ΔCT mutant. These results suggest that Ded1 and the C-terminal region have roles in cell growth during oxidative stress.

FIG 7.

Ded1 promotes cell survival and growth during oxidative stress. (A) Growth curve analysis of DED1, ded1-Δ591-604 and ded1-ΔCT strains in rich media, untreated (left) and treated with 0.8 mM H₂O₂ (right). Time points were fitted to the Gompertz growth equation. Each time point shows the mean and SEM of 8 biological replicates performed in parallel. (B) The fitted Gompertz parameters, λ (lag-phase length) and μ_max (maximum growth rate), are shown with 95% confidence intervals (**, P < 0.01; ****; P < 0.0001). (C) Growth recovery of DED1, ded1-Δ591-604, and ded1-ΔCT strains treated for 6 h with 0.8 mM H₂O₂. A 2-fold serial dilution series was performed, and cells were plated on plates lacking H₂O₂ and incubated for 2 or 3 days as shown. Untreated cells were diluted and plated in parallel. (D) Cell survival following 6 h of treatment with 0.8 mM H₂O₂. CFU were calculated from dilutions of untreated and H₂O₂-treated DED1, ded1-Δ591-604, and ded1-ΔCT cells after 3 days. CFU in untreated cells are shown normalized to DED1 to show that plating efficiency does not significantly differ between strains (left). Cell survival after treatment is shown relative to untreated CFU for each strain (right). Data represent the mean and SEM from 4 biological replicates. Statistical significance was determined using one-way ANOVA (*, P < 0.05).

To further investigate these effects, cells were treated with peroxide, and then equal numbers of cells were plated on rich media to examine recovery of growth. As expected, peroxide treatment inhibited the growth of wild-type cells compared to untreated cells (Fig. 7C, 2 days). Strikingly, peroxide-treated ded1-ΔCT cells displayed a severe growth inhibition compared to wild-type cells. Consistent with the growth curves, the ded1-Δ591-604 mutant also showed moderate growth defects in this assay. These growth defects could be due to either a delay in growth or an increase in cell death following oxidative stress. To examine the role of cell death, relative cell survival was calculated by counting CFU from wild-type and mutant strains in the presence and absence of peroxide treatment. All strains showed a significant decrease in survival after stress, but the ded1-ΔCT mutant had a significantly larger decrease than wild-type cells, indicating a loss of viability (Fig. 7D). However, delayed growth was eventually observed after extended incubation of the treated ded1-ΔCT cells (Fig. 7C, 3 days), suggesting that there is also a delayed recovery of growth in these cells. Further supporting this idea, the viability of peroxide-treated ded1-Δ591-604 cells was not significantly different from wild-type cells (Fig. 7D), so the reduced growth in this mutant after stress may be largely the result of delayed recovery rather than reduced survival. Overall, these results indicate that Ded1 plays a critical role in both cell survival upon oxidative stress induction and cell growth during stress recovery.

The role of Ded1 in SG dynamics during oxidative stress.

Next, we examined the formation of SGs during hydrogen peroxide treatment. Peroxide treatment of wild-type cells caused an induction of SGs, defined as Pab1-GFP-positive foci (Fig. 8A, arrows), over a time course of several hours. The percentage of cells containing SGs peaked at about 12 h after treatment at 28%, then began to decrease, falling to near pretreatment levels by 20 h (Fig. 8B). Thus, the SG time course correlates with the observed growth curve in Fig. 4, with SGs increasing during the lag in growth and then decreasing as growth resumes. On the other hand, in ded1-ΔCT cells, SGs were more sharply induced, with over 2-fold more cells (61%) containing Pab1-GFP foci at 12 h compared to wild-type (Fig. 8A and B). Interestingly, many ded1-ΔCT cells contained numerous small foci (in addition to larger ones), which may be reflective of SGs still in the process of forming (Fig. 8A, arrowheads). Similar to wild-type cells, SGs began to diminish after 12 h in ded1-ΔCT cells, but mutant cells continued to show a higher percentage of SGs even at later time points. Given the increased growth lag of ded1-ΔCT cells, this result suggests that resumption of growth correlates with eventual SG clearance. As previously, Pab1-GFP levels did not differ significantly in DED1 vs. ded1-ΔCT cells, either untreated or after 12 h of peroxide treatment (Fig. 8C). The ded1-Δ591-604 mutant had a slight trend toward increased SGs compared to wild-type, although the difference was not statistically significant (Fig. 8D and data not shown).

FIG 8.

Ded1 regulates stress granule dynamics in oxidative stress. (A) Representative images of DED1 and ded1-ΔCT strains expressing a Pab1-GFP reporter, after 0, 4, 8, 12, 16, and 20 h of treatment with 0.75 mM H₂O₂. Arrows mark larger Pab1-GFP foci, while arrowheads mark smaller foci/speckles predominantly observed in ded1-ΔCT cells. (B) Quantitation of Pab1-GFP foci during H₂O₂ time course in panel A. (C) Western blot analysis of protein extracts from DED1 and ded1-ΔCT cells expressing PAB1-GFP, untreated or after 12 h of 0.75 mM H₂O₂. Samples were probed with antibodies specific for Pab1 or Pgk1. Pab1-GFP levels were quantified via band densitometry and normalized to Pgk1 levels in each sample. The mean of 4 biological replicates is shown. (D) Pab1-GFP focus quantitation in DED1, ded1-Δ591-604 and ded1-ΔCT strains after 12 h of treatment with 0.75 mM H₂O₂. Mean and SEM from 3 or 4 biological replicates are shown. Statistical significance was determined using one-way ANOVA (**, P < 0.01).

To determine the localization of Ded1 in oxidative stress, we expressed GFP-tagged DED1 and ded1-ΔCT on low-copy-number plasmids along with PAB1 tagged with mCherry in wild-type cells and treated them with peroxide. Similar to Fig. 4, Ded1-GFP formed some foci even in the absence of peroxide (data not shown), but Ded1-positive foci increased after peroxide treatment, consistent with SG induction (Fig. 9A). The Pab1-mCherry signal was considerably dimmer than that of Ded1- or Pab1-GFP; however, Ded1 foci colocalized with visible Pab1 foci (arrows), indicating that Ded1 is found in SGs in these conditions. After peroxide treatment, ded1-ΔCT-GFP formed both large and small foci, similar to the Pab1-GFP foci in ded1-ΔCT cells (Fig. 9A, arrows and arrowheads, respectively). The Pab1-mCherry signal was often too weak to visualize these smaller foci but colocalized with ded1-ΔCT-GFP foci when present. Overall, these results suggest that both wild-type Ded1 and the ded1-ΔCT mutant associate with Pab1 and SGs during oxidative stress, consistent with their localization to overexpression-induced foci as observed in Fig. 4.

FIG 9.

Deletion of the Ded1 C terminus affects granule formation and translation differently in overexpression and oxidative stress. (A) Live-cell microscopy of wild-type cells expressing DED1-GFP (top two rows) or ded1-ΔCT-GFP (bottom two rows) and PAB1-mCherry either untreated or after 12 h of treatment with 0.75 mM H2O2. Single channel and merged images (GFP in green, mCherry in red) are shown. Arrows mark larger GFP/mCherry foci, while arrowheads mark smaller ded1-ΔCT-GFP foci/speckles observed. (B, C) Western blot analysis of protein extracts from DED1 and ded1-ΔCT cells after 0, 2, or 10 h of treatment with 0.75 mM H₂O₂ (B) or from ded1-ΔCT cells expressing galactose-inducible DED1, ded1-ΔCT, or vector control after 7 h of galactose induction (C). Samples were probed with antibodies specific for phosphorylated eIF2α (Sui2), total eIF2α, and Pgk1 as loading control. Relative levels of phosphorylated eIF2α from 3 independent experiments are shown, normalized to total eIF2α.

SG formation in ded1-ΔCT cells treated with peroxide is different from our results with galactose-induced ded1-ΔCT overexpression (Fig. 3). We suggest that this is a result of different means of SG induction. In contrast to the overexpression model, in oxidative stress, multiple different pathways and factors are participating in SG formation, and translation regulation can also affect SG induction (9, 35). To begin to examine these possibilities, we blotted cell extracts treated with peroxide for 0, 2, or 10 h for eIF2α, a translation factor that is inhibited by phosphorylation under multiple stress conditions and is often associated with SG formation (3, 9). In both DED1 and ded1-ΔCT cells, phosphorylated eIF2α levels were substantially induced by peroxide treatment (Fig. 9B), consistent with the idea that translation is repressed and leading to SG formation in these conditions. By contrast, eIF2α phosphorylation was not induced by overexpression of DED1 or ded1-ΔCT (Fig. 9C), perhaps suggesting a more direct induction of SG formation. In any case, these results indicate that the mechanism of SG formation (and translation repression) in overexpression differs from that in peroxide treatment, which may explain the differences in SG formation mediated by ded1-ΔCT.

Ded1 interaction with eIF4G1 affects its role in oxidative stress.

To investigate the role of Ded1’s interaction with eIF4G1 during oxidative stress, we examined null mutants of eIF4G1 (tif4631Δ) and tif4631Δ ded1-ΔCT double mutants for growth and SG defects. Following peroxide treatment, tif4631Δ cells showed a delay in growth that was intermediate (λ = 16 h) between ded1-ΔCT and wild-type control cells (Fig. 10A and B), indicating that eIF4G1 also plays a role in this stress response but that the effects in the ded1-ΔCT mutant cannot be explained solely through Ded1 interaction with eIF4G1. Supporting this hypothesis, the tif4631Δ ded1-ΔCT double mutant had a similar growth delay to the ded1-ΔCT mutant alone (λ = 18 h for both mutants in this set). These results suggest an epistatic relationship between Ded1 and eIF4G1 wherein the effect of eIF4G1 on growth in peroxide is mediated through its interaction with the Ded1 C-terminal region, but the C-terminal region plays an additional role in this process, perhaps by promoting Ded1 oligomerization. The tif4631Δ ded1-ΔCT double mutant also showed a slight defect in the rate of growth during recovery, unlike the other mutants tested (Fig. 10B). However, a similar reduced growth rate was observed in untreated mutant cells, suggesting that it is due to an effect on translation in steady-state conditions rather than stress-related (data not shown). We also examined the formation of SGs in the tif4631Δ mutant cells. Both tif4631Δ and tif4631Δ ded1-ΔCT mutants showed increased SGs compared with wild-type controls after 12 h of treatment with peroxide, similar to the increase observed with ded1-ΔCT (Fig. 10C and D). Taken together, these results suggest that the interaction of Ded1 with eIF4G1 mediates at least part of Ded1 function during oxidative stress.

FIG 10.

Ded1 interaction with eIF4G1 affects its role in oxidative stress. (A) Growth curve analysis of DED1, ded1-ΔCT, tif4631Δ and tif4631Δ ded1-ΔCT strains treated with 0.8 mM H₂O₂. Each time point shows the mean and SEM of 5 biological replicates performed in parallel. (B) The fitted Gompertz parameters, λ (lag-phase length) and μ_max (maximum growth rate), for the indicated strains are shown with 95% confidence intervals (*, P < 0.05; ****, P < 0.0001). (C) Representative images of DED1, ded1-ΔCT, tif4631Δ and tif4631Δ ded1-ΔCT strains expressing a Pab1-GFP reporter after 12 h of treatment with 0.75 mM H₂O₂. (D) Quantitation of Pab1-GFP foci in panel C. Mean from 3 or 4 biological replicates shown with SEM. Statistical significance was determined using a one-way ANOVA (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Ded1 promotes resumption of translation during adaptation to oxidative stress.

Finally, as a translation factor, Ded1 is likely to have effects on translation during oxidative stress. To examine this possibility, we treated wild-type DED1 and ded1-ΔCT mutant cells with peroxide, generated extracts after 0, 2, and 10 h of treatment, and then performed sucrose density centrifugation to generate polyribosomal profiles (Fig. 11A). From the profiles, we calculated the relative polysome to 80S monosome ratio (P/M ratio), where a higher number indicates greater bulk translation (Fig. 11B). In untreated cells, both the wild-type and ded1-ΔCT had similar levels of polysomes. Consistent with prior studies (35), after 2 h of peroxide treatment, the P/M ratio decreased substantially in both strains with the greatly decreased polysomes and increased monosome peak typical of translation repression (see e.g., reference 32). After 10 h, translation had recovered in wild-type cells with a P/M ratio approaching wild-type levels (Fig. 11A, right panels and B). This resumption of translation largely coincides with the end of the lag phase in growth (Fig. 7A and B), although it may precede it somewhat. In ded1-ΔCT cells, however, the P/M ratio remained at a low level at 10 h, indicating that translation did not recover in this time frame. This result suggests that the growth delays in the mutant may be the result of defects in resuming translation and that Ded1 plays an important role in promoting translation during later stages of the stress response.

FIG 11.

Ded1 plays multiple roles in translational regulation during oxidative stress. (A) Polyribosomal profiles of DED1 and ded1-ΔCT cells after 0, 2, and 10 h of H₂O₂ treatment were generated by subjecting cell lysates to 7–47% sucrose density centrifugation and optical density analysis at 254 nm. Representative profiles are shown. (B) Polysome/monosome (P/M) ratios were determined by comparing the sum of the areas of the polysome peaks to the area of the monosome peak, and the ratio for each sample were normalized to the P/M in untreated wild-type DED1 cells. The mean and SEM from 4 to 8 biological replicates are shown. Statistical significance was determined using the Mann-Whitney nonparametric test (unpaired; *, P < 0.05; ***, P < 0.001 treated versus untreated ded1-ΔCT; §, P < 0.05 ded1-ΔCT versus DED1). (C) Diagram of the unstructured and structured 5’UTR firefly luciferase reporter mRNAs. The 5’UTRs are modified versions of the yeast RPL41A 5’UTR; a stem-loop forming sequence is inserted in the structured reporter. (D) Time course of luciferase activity in H₂O₂-treated cultures of DED1 or ded1-ΔCT cells containing either the unstructured or structured luciferase reporter constructs. Luciferase units obtained from each culture at each time point were normalized to the luciferase units obtained from untreated DED1 cells containing the unstructured reporter. Mean and SEM from 3 to 7 biological replicates are shown. Statistical significance was determined using Student's t test (unpaired; *, P < 0.05; **, P; < 0.01; ***, P < 0.001 treated versus untreated samples from the same strain; §, P < 0.05; §§ P < 0.01 ded1-ΔCT versus DED1 sample).

To further explore this idea, we examined Ded1-dependent changes on translation of mRNA reporters with different amounts of secondary structure in the 5’UTR. We used a set of previously published luciferase reporters with 5’UTRs derived from RPL41A that contain either a stem-loop (ΔG_free = −3.7 kcal/mol) or are unstructured (Fig. 11C) (24). The stem-loop containing reporter is translated less well than the unstructured one in wild-type cells, and this difference is often exacerbated in ded1 mutants (23, 24, 36). Here, we transformed these reporters into wild-type and ded1-ΔCT mutant cells, treated the cells with peroxide, and performed luciferase assays over a time course. In wild-type cells, consistent with the polysome results, we observed significant drops in luciferase activity (to 60% of the untreated activity) with both reporters after 2 h of treatment (Fig. 11D). The difference in the magnitude of the repression compared with polysome analysis may be due to the exogenous nature of the reporters. The reduction in translation in wild-type cells was maintained through 5.5 h of treatment, but activity then increased to pretreatment levels or above by 8 h and was maintained through 12 h (Fig. 11D). As above, this resumption of translation precedes the end of the lag phase in growth, suggesting that cells may require this time to reshape their proteome for resumed growth. Consistent with prior studies, the structured reporter showed reduced luciferase activity compared to the unstructured one (approximately 30%) in untreated conditions (Fig. 11D), but the ratio of structured to unstructured activity did not vary significantly during the time course.

In ded1-ΔCT mutant cells, activity from the unstructured reporter was similar to wild-type cells before treatment (Fig. 11D). However, consistent with the polysome analysis, translation in ded1-ΔCT cells progressively decreased over the time course through 8 h, when translation in wild-type cells had begun to recover, and the mutant cells only partially recovered by 12 h. The structured reporter in the ded1-ΔCT mutant showed a similar trend, although its activity did not recover to the same extent as the unstructured reporter at 12 h (Fig. 11D). Furthermore, structured reporter activity in ded1-ΔCT cells was significantly lower at earlier time points as well, including nearly 2-fold lower in untreated cells and after 2 h of peroxide treatment. This early reduction in translation of mRNAs with structured 5’UTRs may underlie the decreased cell survival observed in ded1-ΔCT mutant cells (Fig. 7D). Overall, these results suggest that Ded1 plays critical roles in regulating translation both during the initial response to stress and during the transition to resumed growth in recovering cells.

DISCUSSION

In this study, we examined the role of the DEAD box RNA helicase Ded1 in stress responses using both a DED1 overexpression model and an oxidative stress model. We analyzed ded1 (and tif4631) mutants for defects in growth, SG formation, and translation in these models. To integrate the diverse results obtained, we propose a biphasic framework for cellular stress and Ded1 function (Fig. 12). The first phase is the initial response to stress conditions wherein cells shift away from progrowth homeostasis and redirect resources to counter the stress, resulting in inhibition of growth and proliferation (3–5). Multiple lines of evidence indicate that Ded1 functions to inhibit growth in this phase (Fig. 12, #1), including the inhibition caused by DED1 overexpression (and its suppression by the ded1-ΔCT mutation), as well as Ded1-dependent growth phenotypes in cells treated with rapamycin (Fig. 1) (19, 26, 32, 37). We suggest that a failure to properly halt growth during stress is a misallocation of cellular resources that results in a loss of viability, as we have observed in ded1-ΔCT mutants during stress (Fig. 7D) ( 32). By contrast, the second phase of the stress response consists of a gradual resumption of prestress conditions through either removal of the stress or sufficient adaptation to it (38, 39). Our results for ded1 mutants during oxidative stress suggest that Ded1 also plays a role in this recovery phase by affecting the length of the lag and the transition to resumed growth (Fig. 12, #2). Although ded1-ΔCT cells have reduced viability compared to wild-type, this mutant also has an increased lag phase and a delay in resuming growth, as observed in the growth curves and serial dilutions (Fig. 7). Further supporting this role, the ded1-Δ591-604 mutant has nearly the same viability as wild-type cells but is still delayed in recovery of growth. Overall, our results suggest that Ded1 and its C-terminal region have critical roles in both growth inhibition during the initial stress response and in resumption of growth during recovery.

FIG 12.

Ded1 has multiple effects during cellular stress responses. A model for Ded1 function during both the initial stress response and during adaptation/recovery. Ded1 plays roles in both translation regulation and formation of SGs during the stress response, leading to growth inhibition (no. 1, 3, and 5). Likewise, Ded1 function is important for translation upregulation in the recovery phase, leading to resumption of growth (no. 2 and 4). A role in SG disassembly has not been identified to date (no. 6).

Changes in translation have a critical contribution to cellular stress responses, including repression of bulk translation and upregulation of stress-specific proteins (5, 6). Because Ded1 is known to regulate translation, we propose that during the initial phase, Ded1 inhibits growth through repression of general translation (Fig. 12, #3). This hypothesis has been proposed by several groups and is supported by prior results in rapamycin-treated cells, DED1 overexpressing cells, and in vitro translation assays (19, 26, 32, 37). Consistent with those results, here we observed defects in translation in the structured 5’UTR reporter in ded1-ΔCT mutant cells at all time points during oxidative stress (Fig. 11D). Likewise, the severe growth inhibition of DED1 overexpression and the substantial reduction in cell survival in ded1-ΔCT cells after peroxide treatment (Figs. 1B and 7) suggest that Ded1 affects translation during the early response phase. We note that reductions in the polysomes and in translation of the unstructured reporter in ded1-ΔCT mutant cells at early time points were similar to wild-type cells (Fig. 11A and D). These results may be due to using a 2-h time point during treatment or possibly different effects of Ded1 on different subsets of mRNAs. Consistent with this, RNA-seq analysis has shown that Ded1 has such differential effects in progrowth conditions (24, 25), and this may also be the case during stress. As a second role during stress, here we also present evidence linking Ded1 to translation during the recovery phase when general translation resumes (Fig. 12, #4). In wild-type cells, both polysomes and translation of the reporters recovered in 8 to 10 h, slightly preceding the end of the lag phase (Fig. 7 and 11). In ded1-ΔCT cells, however, translation recovered more slowly, consistent with the delay in growth in this mutant. Therefore, we suggest that Ded1 (and the C-terminal region) play a role in promoting translation as cells transition from a stress-induced lag phase to a resumption of growth. As in the initial response (and during steady-state conditions), it is likely that Ded1 also has mRNA-specific effects during this phase. Future studies may be able to further investigate this possibility.

By definition, SGs are formed in response to stress (9, 40). Ded1 has previously been extensively linked to SGs, and it appears to play a role in their formation (Fig. 12, #5). This hypothesis is most directly supported by the formation of SG-like foci upon DED1 overexpression, observed here and previously (Fig. 3) (19), as well as a reduction in SGs following knockdown or inhibition of DDX3X (28, 41). Furthermore, we showed that Ded1 itself localizes to these foci (Fig. 4 and 9), consistent with the identification of Ded1 and its homologs as SG components, including in proteomic analysis of SGs (9, 19, 28, 29, 42). Complementing these results, ded1 mutations also altered SG dynamics in both the overexpression model and oxidative stress (Fig. 3 and 8). Interestingly, mutants such as ded1-ΔCT had different effects on SG formation in the two models, decreasing SGs when overexpressed but increasing SGs upon peroxide treatment. In overexpression, SG formation is likely driven by Ded1 directly; therefore, defects in Ded1’s interactions with itself and other SG components, such as in the ded1-ΔCT mutant, would lead to a reduction in SGs. By contrast, during oxidative stress, many pathways and factors contribute to SG formation, particularly changes in translation (9, 40), and the differences in eIF2α phosphorylation between the two models are consistent with distinct mechanisms in the two models. Ded1’s specific function in SGs remains unknown and is complicated by limitations in the understanding of SGs themselves (43). As an RNA helicase, Ded1 may affect the sorting of mRNAs in SGs, as recently proposed by Hondele et al., and we note that this study also found that IDRs in DEAD box proteins promote phase separation in vitro and in vivo (30). Further work is needed to dissect this hypothesis, however. Others have suggested that Ded1 may also have a role in SG disassembly (19), but there is little supporting evidence to date (Fig. 12, #6).

The mechanism of Ded1 function in the stress response has yet to be fully defined. Here, we used different mutations in DED1 and eIF4G1 to examine the molecular requirements for the various stress-responsive roles of Ded1. First, it is clear that the C-terminal region of Ded1 is critical to the stress response given the many defects observed in the ded1-ΔCT mutant. To begin to distinguish between the effects of eIF4G1 binding and Ded1 oligomerization, we used both deletions of eIF4G1 (tif4631Δ) and the ded1-Δ591-604 mutant, which severely reduces eIF4G1 binding in vitro while having only a minor effect on oligomerization (32). We observed somewhat variable defects in both growth and SG formation with these mutants (Fig. 1, 5, 6, 7 and 10) that were more moderate than with the full C-terminal deletion mutant. Taken together, these results indicate that eIF4G1 binding by Ded1 does contribute to its function in stress, which is consistent with prior studies implicating eIF4G1 in stress responses and in SGs (33, 44, 45). However, the more severe effects in the ded1-ΔCT mutant suggest that Ded1 oligomerization plays a major role, consistent with the idea that promiscuous protein-protein interactions promote SG assembly (30, 46). It is also possible that another unknown interaction with the C-terminal region is critical; notably, we did not examine mutations in eIF4G2 (TIF4632), the closely-related paralog of eIF4G1. However, eIF4G2 has not been shown to bind Ded1, and in fact the RNA2 and RNA3 regions of eIF4G1, which have been shown to be required for Ded1 binding, are not present in the paralog (19, 20, 47). Interestingly, in the overexpression model, deleting most of the central helicase domain in the ded1-Δ190-497 mutant had similar effects to wild-type DED1 in both growth inhibition and SG formation (Fig. 1, 3, and 6). Since this mutant is predicted to lack significant enzymatic activity and RNA binding affinity, these results suggest that neither is absolutely required for these effects, at least in the overexpression model. However, this is unlikely to be the case for Ded1 translation regulation, and future studies will be required to further tease apart these mechanisms.

Several studies have now shown that Ded1 plays a role in regulating translational responses to multiple different stresses, and the importance of the C-terminal region seems to be preserved in all of these (Fig. 7) (30–32). However, differences in the mechanism of Ded1 involvement in these stresses may also be present. For example, it has been proposed that during heat shock, Ded1 condensation into SG-like structures makes it unable to facilitate translation of housekeeping mRNAs (31). Although this model is straightforward, it does not fully account for results in other models of stress. As another example, our previous study showed that Ded1 plays a role in the response to rapamycin treatment, which does not induce SGs (32). Furthermore, the precise subsets of mRNAs dependent on Ded1 are likely to be different in different stresses, given the need to respond to specific cellular conditions (oxidative imbalance, nutrient deprivation, heat shock, etc.). Figure 12 is therefore intended as a general framework, and individual stresses might include or exclude various specific roles. Future studies will be needed to further delineate these mechanisms, particularly as the complexity of Ded1 function will likely have important implications for pathologies such as cancer and aging, where dysregulation of stress responses contributes to disease progression.

MATERIALS AND METHODS

Yeast strains and plasmids.

Yeast strains and plasmids used are listed in Tables 1 and 2. Strains containing different ded1 mutants under the control of the endogenous DED1 promoter were created by plasmid shuffle starting with strain SWY4093 (ded1::KAN + pCEN/URA3/DED1) or strain TBY134 (ded1::KAN tif4631::HYG + pCEN/URA3/DED1). Strains that conditionally overexpressed wild-type DED1 or ded1 deletion mutants were created by transforming strain TBY121 (ded1::KAN + pCEN/LEU2/ded1-ΔCT), SWY4093, or TBY134 with plasmids that expressed DED1 or the indicated ded1 mutant under the control of the GAL1/10 promoter, plus a plasmid that constitutively expressed PAB1-GFP where indicated. Plasmids for galactose-inducible overexpression of untagged DED1, ded1-ΔCT, or ded1-Δ14aa proteins (ded1-Δ535-548, -Δ549-562, -Δ563-576, -Δ577-590, and -Δ591-604) were constructed as follows: pTB137 was constructed by inversion of the XhoI fragment containing the GAL-DED1-HHA sequence in pRP2086 (encoding the Ded1-His-HA-protein A fusion protein) relative to the vector backbone. Then, untagged GAL-DED1 (pTB138), GAL-ded1-ΔCT (pTB148) and GAL-ded1-Δ14aa (pTB139 through pTB143) plasmids were constructed by replacing pTB137 sequence downstream of the internal BamHI site at nt 33 of the DED1 coding sequence with the analogous sequence from plasmids pSW3619, pTB136, or pTB111 through pTB115, respectively (32, 48). Plasmid pTB144, for galactose-inducible expression of ded1-Δ190-497, was constructed as follows: pTB138 was digested with AgeI (sites at nt 564 and nt 1488 of the DED1 coding sequence) and the plasmid backbone plus the DED1 N- and C termini sequences was isolated and re-ligated, resulting in an in-frame deletion. Plasmid pTB197, for galactose-inducible expression of ded1-Δ546-557, was constructed by divergent PCR amplification of pTB138, omitting the nucleotides between nt 1636 and nt 1671 of the DED1 coding sequence, followed by re-ligation of the linear product and sequencing of DED1 and the GAL1/10 promoter to ensure that no point mutations were introduced. Plasmid pTB198 was constructed by PCR amplification of the ded1-W603/604A sequence in plasmid pTB124, extending from the BamHI site at nt333 to the SalI site at nt 2723 (downstream of the DED1 3′ UTR). The BamHI plus SalI-digested PCR product was then ligated to BamHI plus SalI-digested pTB138, replacing the wild-type sequence.

TABLE 1.

Yeast strains used in this study

Strain name	Genotype	Background	Source or reference
SWY4093	MATα ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/URA3/DED1	W303	48
TBY121	MATα ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/LEU2/ded1-ΔCT	W303	32
TBY134	MATα tif4631::HYG ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/URA3/DED1	W303	32
TBY152	MATα ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/LEU2/DED1	W303	This study
TBY157	MATα ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/LEU2/ded1-Δ591-604	W303	This study
TBY159	MATα tif4631::HYG ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/LEU2/DED1	W303	This study
TBY165	MATα tif4631::HYG ded1::KAN ade2-1 ura3-1 his3-11,15 leu2-3,112 trp1-1 can1-100 +pCEN/LEU2/ded1-Δ536-604	W303	This study

TABLE 2.

Plasmids used in this study

Plasmid name	Description	Plasmid backbone	Source or reference
pSW3619 DED1/LEU2	CEN/LEU2/DED1	pRS315	48
pTB136	CEN/LEU2/ded1-ΔCT (Δ536-604)	pRS315	32
pTB111	CEN/LEU2/ded1-Δ535-548	pRS315	32
pTB112	CEN/LEU2/ded1-Δ549-562	pRS315	32
pTB113	CEN/LEU2/ded1-Δ563-576	pRS315	32
pTB114	CEN/LEU2/ded1-Δ577-590	pRS315	32
pTB115	CEN/LEU2/ded1-Δ591-604	pRS315	32
pTB124	CEN/HIS3/ded1-W603/604A	pRS413	32
pSW3739	2μ/TRP1/GAL1/10-DED1	pRS424	48
pRP2086	2μ/HIS3/GAL-DED1-His-HA-protein A (DED1-HHA)	pRS423	19
pTB137	2μ/HIS3/GAL-DED-His-HA-protein A (orientation inverted relative to pRP2086)	pRS423	This study
pTB138	2μ/HIS3/GAL-DED1	pRS423	This study
pTB139	2μ/HIS3/GAL-ded1-Δ535-548	pRS423	This study
pTB140	2μ/HIS3/GAL-ded1-Δ549-562	pRS423	This study
pTB141	2μ/HIS3/GAL-ded1-Δ563-576	pRS423	This study
pTB142	2μ/HIS3/GAL-ded1-Δ577-590	pRS423	This study
pTB143	2μ/HIS3/GAL-ded1-Δ591-604	pRS423	This study
pTB144	2μ/HIS3/GAL-ded1-Δ190-497	pRS423	This study
pTB148	2μ/HIS3/GAL-ded1-ΔCT (Δ536-604)	pRS423	This study
pTB197	2μ/HIS3/GAL-ded1-Δ546-557	pRS423	This study
pTB198	2μ/HIS3/GAL-ded1-W603/604A	pRS423	This study
pRP1657	CEN/URA3/PAB1-GFP, EDC3-mCh	YCpLAC33	33
pRB16	CEN/LEU2/PAB1-GFP	YCpLAC33	51
pRP1556	CEN/HIS3/DED1-GFP	pRS413	19
pRP2080	CEN/HIS3/ded1-ΔCT-GFP	pRS413	19
pSW3622	CEN/TRP1/DED1-GFP	pRS314	48
pRP435	CEN/LEU2/PAB1-mCherry	YCpLAC33	51
pFJZ342	CEN/URA3/RPL41A 5′UTR with 23 CAA repeats (91 nt) upstream of firefly luciferase coding sequence	YCpLAC33	24
pFJZ623	CEN/URA3/RPL41A 5′UTR with cap-distal stem loop (ΔG = −3.7 kcal/mol) upstream of firefly luciferase coding sequence	YCpLAC33	24

Growth assays.

All yeast cultures were incubated at 30°C. Serial dilution growth assays were performed as previously described (26). To analyze growth under conditions of DED1 or ded1 mutant overexpression, cells were pregrown in selective SD liquid medium containing 2% sucrose, serially diluted 5-fold in SD medium, and spotted on selective SD solid medium containing 2% galactose or 2% dextrose (as control). Microwell growth curves were generated by growing yeast cultures in yeast extract-peptone-dextrose (YPD) overnight in 96-well plates (Costar 96-well flat bottom). For each strain, five to eight biological replicates were grown in parallel. Cells were back-diluted to an optical density at 600 nm (OD₆₀₀) of 0.1 and allowed to grow to mid-log phase. 20 μl of mid-log-phase culture was added to 180 μl of fresh YPD with or without 0.8 mM H₂O₂ (Beantown Chemical, Hudson, NH) in 96-well plates. Plates were incubated at 30°C with shaking at 400 rpm to reduce cell settling. OD₆₀₀ measurements were obtained at various time points with a VERSAmax microplate reader using SOFTmax Pro 3.1 software. For determination of growth parameters, data points were curve fitted using GraphPad Prism to a reparameterized form of the Gompertz growth equation (34):

$$y=A{e}^{-e^{\frac{{\mu }_{\max }\cdot e}{A}\left(\lambda -t\right)+1}}$$

This yielded lag phase duration (λ) and maximal growth rate (μ_max) values for each strain and condition, and statistical significance was determined by the extra sum-of-squares F-test.

Western blotting.

For analysis of protein levels, cells were grown as described, and crude cell lysates were prepared by lysis in 1.85 M NaOH and 7.4% β-mercaptoethanol followed by trichloroacetic acid precipitation (26). Proteins were separated by SDS-PAGE, blotted, and probed with antibodies specific for Ded1 (VU318, described in reference 48), eIF4G1/Tif4631 (gift from R. Parker, University of Colorado-Boulder) (49), Pgk1 (Life Technologies), Pab1 (Santa Cruz), phosphorylated-Ser51 eIF2α (Cell Signaling), and total eIF2α/Sui2 (gift from T. Dever, National Institutes of Health) (50). HRP-conjugated secondary antibodies were used to visualize chemiluminescent bands on a Sapphire biomolecular imager (Azure Biosystems). For densitometry, band intensity for Ded1, Pab1-GFP, eIF2α, or eIF2α phosphorylation was measured using ImageJ/Fiji software, Ded1 and Pab1-GFP band intensity was normalized to Pgk1 band intensity in the same sample while phospho-eIF2α band intensity was normalized to that of total eIF2α in the same sample. Statistical significance was determined via Student's t test.

Granule analysis by fluorescence microscopy.

For SG analysis, the indicated strains were transformed with plasmids that expressed PAB1-GFP alone (pRB16), both PAB1-GFP and EDC3-mCherry (pRP1657), PAB1-mCherry (pRB435), or the indicated GFP-tagged Ded1 construct (pRP1556 or pRP2080). For analysis of Pab1-GFP or Ded1-GFP granules induced by DED1 or ded1 mutant overexpression, cells were cultured in selective SD media containing 2% sucrose at 30°C, back-diluted to an OD₆₀₀ of 0.10 to 0.15 diluted to OD_600nm = 0.10–0.15 in selective SD media containing 1.75% galactose plus 0.25% sucrose, and cultured at 30°C for an additional 6 (TBY121-based strains) or 7 h (SWY4093 and TBY134-based strains) prior to imaging. For analysis of Pab1 or Ded1 granules induced by H₂O₂, strains were grown to an OD₆₀₀ of 0.2 to 0.4 in SD-Leu and then treated with 0.75 mM H₂O₂ for the indicated times. Images were captured using a DeltaVision Elite inverted microscope (Applied Precision/GE Healthcare) with an Olympus 100× Plan Apo 1.4-numerical-aperture objective and appropriate filter sets. Z-series data sets were collected with a pco.edge sCMOS camera at a step size of 0.4 μm. Postacquisition deconvolution was performed using SoftWorx software (Applied Precision). Z-series processing, quantitation and cropping were completed in ImageJ/Fiji; sizing and brightness adjustment were completed in Adobe Photoshop. To determine the percentage of cells with Pab1 or Ded1 foci, the number of cells containing at least one GFP focus in merged Z-series, deconvolved images were manually counted compared to the total number of cells in the image. The ImageJ/Fiji “Cell Counter” (K. De Vos, University of Sheffield), a plug-in that marks user-designated cells/features, was used to track cell numbers. Reported numbers for each strain undergoing either galactose induction or H₂O₂ treatment for the noted time represent a mean of ≥3 biological replicates. An average of >250 cells (minimum of 100) were scored for each replicate. Statistical significance was determined by Student's t test or ANOVA as appropriate.

H₂O₂ growth/viability assays.

All yeast cultures were grown to mid-log phase at 30°C in SD-Leu medium. Cultures were untreated or treated with 0.75 mM H₂O₂ and incubated at 30°C with shaking at 190 rpm for 6 h. Cells were spun down and washed in SD-Leu medium and resuspended to a concentration of 1 × 10⁶ cells/ml. Twofold serial dilutions were plated on 15 cm SD -Leu agar plates. Plates were scanned after 2 and 3 days of incubation at 30°C. Cell viability was measured by counting yeast colonies in 2-fold serial dilution spots after 3 days of recovery. Fiji software was used to count colonies in serial dilution spots where individual colonies were clearly distinguishable, and CFU were calculated. The strain viability was determined by averaging normalized CFU from a minimum of three spots in the same dilution series for each experiment. Strain viability shown was averaged from four independent serial dilution experiments, and statistical significance was determined by one-way ANOVA.

Polyribosome profiles.

Polyribosome profiles were generated as described in reference 26. DED1 and ded1-ΔCT cells were grown to mid-log phase, 0.75 mM H₂O₂ was added (or not for untreated samples), and cells were grown at 30°C for 2 or 10 h in rich media. Cells were lysed and subjected to 7 to 47% sucrose density centrifugation for 2.5 h at ∼178,000 × g, then samples were fractionated and RNA curves were generated by monitoring absorbance at 254 nm. Polysome to monosome ratios were determined by comparing the area under the curve for the sum of the polyribosome peaks to the 80S peak in Fiji/ImageJ. Ratios were then normalized to that for DED1 untreated, set at 1, and significance was determined via Mann-Whitney nonparametric tests in GraphPad Prism.

Translation scanning assays.

Scanning assays for structured 5’UTR sequences were carried out similarly to refrence 36. Briefly, cells transformed with either the control unstructured 5’UTR-firefly luciferase reporter (pFJZ342) or the stem-loop-containing 5’UTR-firefly luciferase reporter (pFJZ623) were cultured in triplicate in selective SD media at 30°C. H₂O₂ was added to a final concentration of 0.75 mM. Cell lysates were generated at various time points via bead beater in luciferase lysis buffer (25 mM Tris phosphate [pH 7.8], 2 mM EGTA, 2 mM DTT, 0.5% Triton X-100, 10% glycerol). Luciferase assays were performed using a standard luciferin reagent (Promega) on a Glomax 20/20 luminometer (Promega). For each biological replicate, values obtained from the triplicate cultures were normalized to cell concentration and averaged. Statistical significance was determined by Student's t test.