Reducing translation through eIF4G/IFG‐1 improves survival under ER stress that depends on heat shock factor HSF‐1 in Caenorhabditis elegans

Restricting translation by lowering ifg‐1 improved survival under ER stress

Previous analysis showed that increased lifespan resulting from reduced ifg‐1 expression in C. elegans was at least partially dependent on several translationally upregulated genes (Rogers et al., 2011). Those most important for increased lifespan are also known to be important for maintaining ER homeostasis and include the ER calcium regulator gene sca‐1 and a gene encoding the transcriptional coactivator mediator complex subunit mdt‐15. Attenuated expression of either is able to induce the UPR^ER (Yan et al., 2006; Hou et al., 2014b). A novel cellular component‐based gene ontological (GO) analysis of genes that were differentially translated when ifg‐1 is reduced via RNAi (Rogers et al., 2011) revealed that cellular processes involving vesicle transport and the endomembrane system were significantly overrepresented (Fig. 1A; Table S1, Supporting information). Furthermore, 75% of the differentially translated genes in these GO categories had increased translation according to changes in the association of mRNA with ribosomes. Given the role of the ER in protein synthesis and maintenance of proteostasis, we investigated the effect of reduced ifg‐1 expression under conditions that induce the UPR^ER.

Animals were fed double‐stranded (ds)RNA corresponding to ifg‐1 or a control vector starting at adulthood followed by exposure to the glycosylation‐inhibiting drug tunicamycin. Survival was significantly improved in animals with ifg‐1 RNAi, with protective effects also observed in mutants bearing the ifg‐1 loss‐of‐function allele cxTi9279 (Fig. 1B; Table S2, Supporting information). Baseline expressions of ER stress sensor genes ire‐1, atf‐6, and pek‐1 were unchanged in ifg‐1 RNAi‐treated animals (Fig. 1C) and ifg‐1(cxTi9279) mutants in the absence of stress (Fig. S1A, Supporting information). However, in animals subjected to ifg‐1 RNAi, a small but significant increase in ire‐1 transcript levels was observed in response to 24 h of exposure to tunicamycin (Fig. 1C), whereas the ifg‐1 mutant showed only a small increase in pek‐1 and decrease in atf‐6 under the same conditions (Fig. S1A, Supporting information). Thus, lowering ifg‐1 increased survival in response to ER stress without major changes in ER stress sensor transcript levels compared with RNAi control animals in a 24‐h window after treatment with tunicamycin.

To see whether knockdown of ifg‐1 changed expression of ER stress sensors with time, measurements were made at day 10 of adulthood, which is closer to the time that a difference in survival was observed between test and control animals. Interestingly, a small but significant increase was observed for each sensor in ifg‐1 RNAi‐treated animals regardless of whether they were exposed to tunicamycin, with a greater than twofold change in pek‐1 by this time (Fig. 1D). These changes were consistent whether a 2‐day (Fig. S1B, Supporting information) or 4‐day (Fig. 1D) RNAi induction phase was used prior to tunicamycin exposure. These results indicated that lowering translation through ifg‐1 resulted in long‐term enhanced expression of ER stress sensors even in the absence of external stress.

As ER stress sensor activity can induce both transcriptional and translational effects, we used polysome profiling to assess changes in global and differential translation. Upon reaching adulthood, animals were fed dsRNA for 4 days prior to profiling. ifg‐1 RNAi dramatically decreased total translation and increased free ribosomal subunits compared with controls (Fig. 2A, left panel). An increase in free ribosomal subunits was observed in both control and ifg‐1 RNAi animals 6 h after treatment with tunicamycin that continued to increase by 24 h (Fig. 2A, middle and right panels). This indicated that global translation was reduced throughout this period in response to tunicamycin for animals under control or ifg‐1 RNAi. Indeed, attenuation of translation is a conserved response to induction of the UPR^ER (Ron & Walter, 2007).

To see whether translation suppression in response to tunicamycin was related to the PEK‐1/PERK arm of the UPR^ER, we examined the inhibitory phosphorylation status of its protein target, eIF2α, which is part of the mRNA translation ternary complex and represents a separate (ifg‐1‐independent) rate‐limiting regulator of translation initiation (Holcik & Sonenberg, 2005). eIF2α phosphorylation was lower in animals treated with ifg‐1 RNAi than control RNAi prior to tunicamycin exposure (Fig. S2A, Supporting information). Subsequently, both control and ifg‐1 RNAi‐treated animals showed an upward tendency of eIF2α phosphorylation after 6 h of exposure to tunicamycin that disappeared by 24 h (Fig. S2A, Supporting information). No significant differences in phosphorylation status between ifg‐1 and control RNAi animals were observed at six and 24 h. Although a tendency for increased eIF2α phosphorylation after 6 h of tunicamycin treatment are in line with the modest translational reduction in ifg‐1 and control RNAi‐treated animals observed in polysome profiles, the continued abatement of translation after 24 h in the absence of increased eIF2α phosphorylation suggested that translation may be attenuated by additional factors.

To determine the response of the IRE‐1 branch of the UPR^ER, we measured transcription and translation of its target, xbp‐1 mRNA. In its active state, IRE‐1 splices immature xbp‐1u (unspliced) transcripts, which allows translation of the mature xbp‐1s (spliced) transcript and synthesis of the transcription factor that it encodes (Ron & Walter, 2007). Prior to tunicamycin treatment, total transcript expression of xbp‐1u and xbp‐1s were comparable between control and ifg‐1 RNAi‐treated animals (Fig. 2B, left panel). Interestingly, recruitment of xbp‐1s and even the nontranslationally productive xbp‐1u to polysomes was constitutively higher in ifg‐1 RNAi‐treated animals (Fig. 2B, right panel). The ratio of spliced to unspliced was similar prior to stress. Upon exposure to tunicamycin, there was a downward trend in total and translated transcripts of xbp‐1u and corresponding increase in xbp‐1s for all conditions (Fig. 2B), indicating activation of IRE‐1. The bias in expression of xbp‐1s abated somewhat by 24 h of tunicamycin treatment, but both total and translated forms of xbp‐1s increased in ifg‐1 RNAi animals by this time. Despite starting at a higher constitutive (no tunicamycin) level, polysome‐associated xbp‐1s in ifg‐1 RNAi animals increased relative to starting levels by 24 h of tunicamycin exposure (P = 0.0092).

Gene expression of several chaperones and other factors are dependent on the IRE‐1/XBP‐1 arm of the UPR^ER. Expression of hsp‐4, which is orthologous to the mammalian binding immunoglobulin protein (BiP), is known to be highly dependent on activity of XBP‐1. We measured hsp‐4 in ifg‐1 RNAi‐treated animals and observed that its transcripts were more highly associated with polysomes after 24 h of tunicamycin treatment than in control RNAi animals (Fig. 2C, right panel). We also tested the DnaJ domain 7 gene, dnj‐7, which encodes the C. elegans ortholog of p58^IPK, a peripheral ER protein that functions in ER associated degradation (ERAD) of misfolded proteins (Oyadomari et al., 2006). Analysis showed that tunicamycin induced dnj‐7 transcription after 24 h of tunicamycin exposure under both conditions. However, animals with reduced ifg‐1 showed a higher increase in polysome‐associated dnj‐7 (Fig. 2D), suggesting that ERAD may be induced to a greater extent by the 24‐h time point. Recently, the protein disulfide isomerase family A member 6 (PDI‐6) was shown to control a negative feedback to the ER stress response by binding and inactivating IRE‐1 in C. elegans (Eletto et al., 2014). Expression analysis here indicated that, although total transcript levels of pdi‐6 rose equally in response to control and ifg‐1 RNAi after 24 h of tunicamycin exposure, reducing ifg‐1 significantly increased pdi‐6 polysome association (Fig. S2B, Supporting information). Such differences in UPR^ER activation could change the kinetics of the ER stress response and lead to faster resolution and rebalancing of organellar proteostasis. Resolution of the response would be indicated by a return of IRE‐1 activity to basal levels, and a tendency for this was observed between the six‐ and 24‐h time points in Fig. 2B by partial reversal of spliced vs. unspliced xbp‐1 for animals under ifg‐1 RNAi (especially among polysomes; Fig. 2B, right panel).

Genes encoding the sarco‐ER calcium ATPase SCA‐1 and the transcriptional coactivator subunit MDT‐15 are translationally upregulated and important for increased longevity when ifg‐1 is attenuated (Rogers et al., 2011). Their requirement for maintaining ER homeostasis (Yan et al., 2006; Hou et al., 2014b) formed part of the rationale for investigating ER responses when translation is attenuated. However, changes in their translational status during the UPR^ER were unknown. Polysome association of sca‐1 and mdt‐15 transcripts increased when ifg‐1 was reduced via RNAi (Fig. S2C,D, Supporting information). When exposed to ER stress from tunicamycin, attenuating ifg‐1 promoted their translation but not their transcription, with no transcriptional nor translational effects in tunicamycin‐treated animals under control RNAi conditions (Fig. S2C,D, Supporting information). Results indicated that these genes were preferentially upregulated to the translation apparatus during the UPR^ER when translation was reduced through attenuation of ifg‐1. Taken together with their importance for increased lifespan when translation is attenuated, findings suggested that they may be important for increased survival under ER stress.

Increased survival under ER stress was dependent on known regulators of proteostasis

Differences in response of ER homeostasis regulators upon exposure to tunicamycin when ifg‐1 was suppressed suggested that one or more may be required for improved survival. We tested the requirement for expression of sca‐1 and mdt‐15 for improved survival under ER stress induced by tunicamycin in the ifg‐1(cxTi9279) mutant. sca‐1 RNAi abrogated enhanced survival in the ifg‐1 mutant compared to sca‐1 RNAi‐treated N2 wild‐type animals (Fig. 3A). Although this result indicated a requirement for sca‐1, it is interesting to note that sca‐1 RNAi in the wild‐type background actually had a small but significant protective effect (Fig. 3A; Table S3, Supporting information). On the other hand, RNAi knockdown of mdt‐15 did not abrogate improved survival in the ifg‐1 mutant (Fig. 3B; Table S3, Supporting information). This result is consistent with recent evidence indicating that, while mdt‐15 is essential for ER homeostasis through changes in the lipid synthesis profile of the ER, it does not alter proteostasis (Hou et al., 2014b). The fact that enhanced longevity through reduced ifg‐1 expression is dependent on both sca‐1 and mdt‐15, but here only required sca‐1 for increased survival under ER stress, demonstrated that the dependence for lifespan and stress resistance phenotypes can be uncoupled.

Next, we tested the importance of the three ER stress sensor genes atf‐6, pek‐1, and ire‐1 for enhanced survival upon exposure to tunicamycin in the ifg‐1 mutant. Lowering atf‐6 or pek‐1 expression did not mitigate enhanced survival in the ifg‐1 mutant (Fig. 3C,D; Table S3, Supporting information). Efficacy of RNAi for both genes was confirmed by testing for developmental effects in larval animals (not shown). In two of the three experiments performed, reducing ire‐1 collapsed survival of the ifg‐1 mutant to that of wild‐type animals on ire‐1 RNAi (Fig. 3E; Table S3, Supporting information). Further testing showed that attenuation of the IRE‐1 substrate xbp‐1 abolished enhanced ER stress resistance of the ifg‐1 mutant in all tests (Fig. 3F; Table S3, Supporting information). Together, these data support an essential role for the IRE‐1 arm of the UPR^ER in mediating the protective effects exhibited in the ifg‐1 mutant.

While unfolded proteins in the ER are known to activate the UPR^ER, unfolded proteins in the cytoplasm trigger the HSR. To determine whether enhanced resistance to ER stress in ifg‐1 animals was dependent on changes in the HSR governing cytoplasmic proteostasis, we tested the importance of the hsf‐1 gene. Despite little or no dependence on survival under ER stress in wild‐type animals, results showed that RNAi targeting hsf‐1 abolished the increased survival under ER stress in the ifg‐1 mutant (Fig. 3G; Table S3, Supporting information). Analysis of total and translated transcript levels revealed that ifg‐1 RNAi‐treated animals did not have significantly different expression of hsf‐1 than control animals in the absence of stress (Fig. S3, Supporting information). However, ifg‐1 RNAi led to preferential translation of hsf‐1 subsequent to treatment with tunicamycin without increasing its transcription (Fig. S3, Supporting information). These results show that ifg‐1 animals required hsf‐1 for improved survival and had increased translation of hsf‐1 under ER stress.

HSF‐1 and the FOXO transcription factor DAF‐16 are key regulators of proteostasis that are required for mediating longevity and stress resistance via the ILS pathway, as previously reviewed (Hartl et al., 2011). Based on results here that identified a requirement for hsf‐1 for enhanced survival under ER stress when translation is attenuated, we analyzed the importance of daf‐16 under the same conditions. Testing showed improved survival under tunicamycin stress in ifg‐1 mutants compared to wild‐type animals when daf‐16 was targeted via RNAi, although the protective effects were dampened (Fig. 3H; Table S3, Supporting information). This incomplete dependence on daf‐16 was similar to its effect on lifespan in this background (Rogers et al., 2011).

We also tested the effects of genetically inhibiting translation at other points in initiation known to increase lifespan. RNAi targeting ife‐2 (cap‐binding subunit of eIF4E), rps‐15 (small ribosomal subunit S15), or iftb‐1 (beta subunit of eIF2, a component of the ternary complex representing another regulatory control point for translation initiation) increased survival in the presence of tunicamycin (Fig. S4; Table S4, Supporting information). These results suggested that enhanced survival under ER stress might be a general response of translation attenuation. We then treated animals with the chemical translation elongation inhibitor, cycloheximide, prior to exposure to tunicamycin, using a concentration of cycloheximide we found slowed growth in developing larvae and suppressed global translation in polysome profiling, phenotypes we observed with RNAi of the translation factor genes above (data not shown). Cycloheximide did not significantly impact median survival under ER stress conditions in either wild‐type or the ifg‐1 mutant strain (Fig. 3I; Table S3, Supporting information), which suggested that inhibiting translation, per se, is not sufficient to increase survival under ER stress. Thus, targeting translation initiation through conserved regulatory nodes (i.e. the cap‐dependent/independent complex and ternary complex) might have promoted survival under ER stress in ways that extend beyond global reduction of translation.

Reducing ifg‐1 enhanced thermotolerance and resulted in higher constitutive and induced expression of genes encoding heat‐shock factors

Given the dependence on hsf‐1 for the protective role of reducing ifg‐1 under conditions that elicit an ER stress response, we analyzed differences in survival under heat stress that activates the cytoplasmic HSR. In two previous studies that tested thermotolerance in response to ifg‐1 RNAi in adult C. elegans, one study found no protection from thermal stress after 2 days on ifg‐1 RNAi (Pan et al., 2007), while the other showed significant protection 4 days after exposure to RNAi (Hansen et al., 2007). We re‐examined survival under continuous thermal stress (35 °C) after 2, 5, or 7 days of ifg‐1 RNAi. Day one adults on ifg‐1 RNAi for 2 days showed no enhanced protection from thermal stress (Fig. 4A; Table S5, Supporting information). However, longer exposure to ifg‐1 RNAi resulted in increased thermotolerance (Fig. 4A,D; Table S5, Supporting information). Results here support both studies based on differences in the duration of dsRNA feeding prior to heat stress. To see whether the delay in enhanced thermotolerance was an artifact of the kinetics of RNAi and/or slow turnover of IFG‐1 protein, Western blot analysis was performed after 2 days on RNAi and showed that IFG‐1 was reduced by 80% (P < 0.05, Wilcoxon test; Fig. 4B). This suggested that slow turnover of IFG‐1 did not account for lack of a protective effect observed by that time. We also tested stress tolerance in the ifg‐1(cxTi9279) mutant, in which longevity and developmental phenotypes are less severe than ifg‐1 RNAi‐treated animals. Results indicated markedly enhanced thermotolerance compared to wild‐type at both 2 and 7 days of adulthood (Fig. 4C,D; Table S5, Supporting information). Thus, although we cannot completely rule out that the delay may be due to a requirement for even lower levels of IFG‐1, the evidence collected does not support this interpretation.

To understand more about how lower IFG‐1 led to enhanced thermotolerance and the role of HSF‐1, we looked at induction of transcript levels for HSR genes after inhibiting expression of ifg‐1 via RNAi for 2 or 7 days. We tested expression of hsp‐16.1/hsp‐16.11, hsp‐16.2, hsp‐16.41, and F44E5.4/F44E5.5 and found that after 2 days of ifg‐1 RNAi, the induction of HSR genes under heat stress (35 °C) was comparable to control RNAi (Fig. 4E). However, after 7 days, animals on ifg‐1 RNAi showed higher constitutive HSR gene expression compared with animals under control RNAi both before and after heating (Fig. 4E). We used the same time points to test expression of F44E5.4/F44E5.5 in animals treated with cycloheximide or RNAi for ife‐2, rps‐15, or iftb‐1. Both rps‐15 and iftb‐1 resulted in upregulation of this chaperone by day 7, similar to ifg‐1 RNAi (Fig. S5, Supporting information). Interestingly, rps‐15 and iftb‐1 were also similar to ifg‐1 in that they each are essential for translation and RNAi for these genes resulted in larval arrest in developing nematodes (data not shown). In contrast, ife‐2, which is required for cap‐dependent translation only, did not result in upregulation of heat‐shock gene expression (Fig. S5, Supporting information). It is also noteworthy that, as only one of five eIF4E orthologs in C. elegans, ife‐2 suppression did not result in any developmental effects, even after multiple generations on RNAi (data not shown) and despite its protective effect in ER stress assays (Fig. S4A, Supporting information). To see whether the ability to inhibit development and reduce translation is required for the effect on heat‐shock gene expression, we tested the response to cycloheximide and found that this treatment, similar to ife‐2, failed to increase heat‐shock gene expression (Fig. S5, Supporting information). Thus, the kinetics of heat‐shock factor gene expression, while it potentially helped explain the delayed effect in enhancing thermotolerance under ifg‐1 RNAi, also suggested potential differences in response to cap‐mediated vs. non‐cap‐mediated translational targeting.

Genes essential for increased lifespan when ifg‐1 is reduced were not required for enhanced thermotolerance

As demonstrated above, hsf‐1 was required for enhanced survival in the ifg‐1 mutant exposed to tunicamycin (Fig. 3G). To test whether ifg‐1 required hsf‐1 for enhanced thermotolerance, we treated wild‐type and ifg‐1 mutant animals with hsf‐1 RNAi for 7 days prior to heat stress. Although reducing hsf‐1 severely decreased heat tolerance, the ifg‐1 mutant demonstrated enhanced survival compared to N2 wild‐type animals (Fig. 5A; Table S6, Supporting information), indicating that there is a HSF‐1‐independent protective effect in the ifg‐1 translation mutant. RNAi of neither sca‐1, mdt‐15, nor ire‐1 had an effect on the ability of the ifg‐1 mutant to enhance thermotolerance (Fig. 5B–D and Table S6, Supporting information). This further uncoupled survival phenotypes, demonstrating that enhanced thermotolerance does not require the same genes that are required for enhanced resistance to ER stress or lifespan. Once again, we tested the effects of cycloheximide as a chemical inhibitor of global translation. For this form of acute stress, thermotolerance was enhanced for wild‐type animals treated with cycloheximide for 7 days, unlike its lack of protective effect under tunicamycin stress. Interestingly, cycloheximide did not further enhance survival under thermal stress in the ifg‐1 mutant (Fig. 5E; Table S6, Supporting information). This indicated that lowering translation, itself, had a protective effect to protein unfolding stress from heat.

Enhanced HSF‐1 activity indicated in Fig. 4E corresponded with enhanced survival when the UPR^ER was induced with tunicamycin. Although null mutants are not viable for ifg‐1 or hsf‐1 in C. elegans, we assessed the effect of ifg‐1 RNAi on lifespan in a hsf‐1(sy411) reduced‐function background. Results indicate that ifg‐1 RNAi was able to extend lifespan in this background (Fig. 5F and Table S7, Supporting information), in contrast to a different method of translation inhibition in another study that found RNAi knockdown of genes encoding ribosomal subunits, including rps‐15, failed to extend lifespan in the hsf‐1(sy411) background (Seo et al., 2013). Thus, although results in the current study indicated that hsf‐1 was required for enhanced proteostasis elicited through attenuation of ifg‐1 under ER stress, it was not required for increased lifespan. Conversely, regulators of ER homeostasis were not required for the enhanced cytoplasmic protective effects associated with lowering translation through ifg‐1 prior to heat stress.

Nematode culture and strains

Caenorhabditis elegans strains were cultured at 20 °C and maintained on normal growth medium plates seeded with OP50 unless otherwise noted. Strains used included wild‐type N2, SJ4005 zcIs4 [Phsp‐4::GFP; lin‐15(n765)] V, hsf‐1(sy441), and ifg‐1(cxTi9279). All strains were permitted to grow at least three generations under normal, unstressed conditions prior to use in experiments.

RNAi experiments

RNAi bacteria strains were cultured and utilized as previously described (Kamath et al., 2001). Induction was performed using 1 mm isopropyl‐β‐D‐thiogalactopyranoside (IPTG) in all cases. NGM agar plates prepared this way are referred to as ‘RNAi plates’. Bacteria expressing dsRNA included empty vector L4440 (Addgene, Cambridge, MA, U.S.A.), ifg‐1 (M110.4), sca‐1 (K11D9.2), mdt‐15 (R12B2.5), ire‐1(C41C4.4), atf‐6 (F45E6.2), pek‐1 (F46C3.1), xbp‐1 (R74.3), hsf‐1 (Y53C10A.12), daf‐16 (R13H8.1), ife‐2 (R04A9.4), rps‐15 (F36A2.6), and iftb‐1 (K04G2.1) from the Ahringer library (Source BioScience, Nottingham, U.K.).

Lifespan

Synchronized populations of nematodes were obtained from gravid adults treated with bleach to remove eggs. Eggs were permitted to hatch overnight in 10 mL of S Basal (0.1 m NaCl, 5.74 mm K₂HPO₄, 44.09 mm KH₂PO₄, 1 mL cholesterol (5 mg mL⁻¹ in ethanol), in H₂O to 1 L) before placing on plates spotted with OP50. We have also tested synchronization methods with timed egg‐lay or bleaching followed immediately by placing eggs on bacteria (a slightly less stringent form of synchronization) and obtained similar results. Upon the first day of adulthood, worms were transferred to RNAi plates that, in addition to 1 mM IPTG, contained 25 mg mL⁻¹ carbenicillin and 50 μg mL⁻¹ 5‐fluoro‐2′‐deoxyuridine (FUdR) to inhibit egg production. Lifespans were scored as day 0 from the time of first RNAi exposure. Nematodes were assessed for life every 2 days by gently tapping the worm with a platinum wire. Worms were scored as dead and removed from the plate upon failure to respond to several taps. Worms were censored if they died because of vulval rupture, internal hatching of progeny, or if they crawled off the plate. Rates of censorship were low in all experiments except those involving the ifg‐1(cxTi9279) mutant, in which a developmental defect results in frequent ruptures during adulthood (> 30%).

Survival in response to tunicamycin treatment

Survival under tunicamycin stress was performed similar to a method previously described (Taylor & Dillin, 2013). Strains were synchronized by bleaching and eggs were allowed to grow on OP50. On the first day of adulthood worms were transferred to RNAi plates spotted with HT115 bacteria containing dsRNA as indicated and treated with FUdR as described above. After 2 days, worms were transferred to new RNAi plates containing a final concentration of 25 μg mL⁻¹ tunicamycin dissolved in DMSO. These plates were spotted with bacteria 1 day prior to addition of tunicamycin, which was added 1 day prior to transferring worms. Control plates lacking tunicamycin contained an equivalent amount of DMSO. Survival was scored beginning at the time of tunicamycin exposure.

Thermotolerance

Synchronized populations of worms were grown until adulthood on OP50 and maintained at 20 °C. Upon reaching day 1 of adulthood, nematodes were transferred to RNAi plates in the absence of FUdR and maintained at 20 °C. To prevent offspring contamination adults were moved to fresh RNAi plates every other day. Worms were shifted to 35 °C after 2, 5, or 7 days of exposure to RNAi and survival was scored by means of touch provocation. Worms not responding to touch were scored as dead.

Chemical inhibition of translation

Cycloheximide (0.5 mm; Sigma, St. Louis, MO, U.S.A.) was added to RNAi plates containing bacteria bearing L4440 control vector or as otherwise indicated. Animals were exposed to cycloheximide 2 days prior to 25 μg mL⁻¹ tunicamycin exposure or 7 days prior to shifting to 35 °C.

Western blotting

Frozen worm pellets were ground in 8 m Urea, 2% SDS, 50 mm DTT, 50 mm Tris pH 7.4, with 1× Halt Protease Inhibitor Cocktail (Thermo Scientific, Waltham, MA, U.S.A.), allowed to solubilize on ice for 10 min, and debris was pelleted by centrifugation at 12 000 g for 3 min. Protein concentrations were determined using Qubit 2.0 protein assay (Life Sciences, Thermo Scientific, Waltham, MA, U.S.A.). Proteins were separated using 4–12% mini Protean TGX gels (Bio‐Rad) and transferred to PVDF membrane. Detection of IFG‐1 or phosphorylated eIF2α was performed using anti‐IFG‐1 polyclonal antibody (a kind gift from Brett Keiper; Contreras et al., 2008) or anti‐phospho‐eIF2α (Ser51; #9721; Cell Signaling Technology) along with goat anti‐rabbit Ig HRP‐conjugated secondary antibody (Thermo Scientific, Waltham, MA, U.S.A.). For IFG‐1 quantification, antibody against beta‐tubulin (Developmental Studies Hybridoma Bank) was used as a loading control with HRP‐conjugated goat anti‐mouse IgG (Thermo Scientific) used as a secondary antibody. Chemiluminescence was detected using the G: Box imager (Syngene, Cambridge, U.K.). For quantifying phosphorylated eIF2α, data were normalized using total protein quantified by loading into TGX stain‐free gels exposed to UV light for five min in a stratalinker prior to blotting. After transfer, the bands were visualized on a UV transilluminator with a fixed exposure time of 1 s. While all bands per lane were used for estimation of total protein loaded, a single representative band is shown in Fig. S1C (Supporting information). Background corrected band intensity was quantified using ImageJ.

RNA isolation, cDNA synthesis, and qRT–PCR

RNA was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's protocol for chloroform extraction. Samples were additionally processed with SurePrep RNA Cleanup and Concentration kit (Fisher BioReagents, Fair Lawn, NJ, USA). The polysome profile RNA isolation included 10 μg glycogen in each sample. A total of 200 ng RNA was reverse‐transcribed using QuantiTect Reverse Transcription kit (Qiagen, Valencia, CA, USA). qRT–PCR was performed in technical duplicate using KAPA SYBR FAST qPCR Master Mix on a LightCycler 480 (Roche Applied Science, Indianapolis, IN, USA). Target gene mRNA was normalized to the housekeeping gene cdc‐42, except for instances of polysome analysis, in which compared samples were normalized to the housekeeping gene act‐1, which we have found does not change among total and translated pools between N2 and ifg‐1, nor before or after exposure to tunicamycin. Changes in gene expression levels were analyzed by using the $2^{-\mathrm{\Delta }\mathrm{\Delta }{C}_{\mathrm{T}}}$ method. Primer sequences are provided in Table S8 (Supporting information).

Polysome profiling

Approximately 200 μL of whole worms were lysed on ice by grinding in solubilization buffer (ground in aliquots of 100 μL gently pelleted worms per 350 μL buffer; 300 mm NaCl, 50 mm Tris–HCL pH 8, 10 mm MgCl₂, 1 mm EGTA, 200 μg mL⁻¹ heparin, 400 U mL⁻¹ RNAsin, 1 mm PMSF, 0.2 mg mL⁻¹ cycloheximide, 1% Triton X‐100, 0.1% sodium deoxycholate). After grinding, an additional 350 μL solubilization buffer was added (1050 μL total) and the lysates were incubated on ice for 30 min. Lysates were centrifuged at 12 000 × g at 4 °C for 10 min to collect debris. Protein content of the lysate supernatants was estimated using Qubit protein assay (Thermo Scientific, Waltham, MA, U.S.A.) and loaded equally onto 10–50% sucrose gradients in high salt resolving buffer (140 mm NaCl, 25 mm Tris–HCL pH 8, 10 mm MgCl₂). Gradients were resolved by ultracentrifugation in a Beckman SW41Ti rotor at 38 000 × g at 4 °C for 2 h. Fractions of the gradients were continuously monitored at absorbance of 254 nm using a Teledyne density gradient fractionator. The polysome fraction was collected and used for qRT–PCR as described above.

GO enrichment analysis

Enrichment of biological GO terms among differentially translated genes was conducted using DAVID. A list of all genes in the original analysis was used for background correction. Broad GO functions were filtered out using the GO FAT option.

Statistical analysis

All statistics were performed using GraphPad Prism 6 software (La Jolla, CA, U.S.A.). Two‐tailed t‐tests were performed on qRT–PCR samples to assess significance. Significance in instances of multiple comparisons was verified via ANOVA followed by post hoc Tukey test. Kaplan–Meier survival curves were plotted and compared using the Mantel–Cox log‐rank test. Western blots were analyzed using a Wilcoxon test.