All strains used are listed in Supplementary Table S1. The worms were cultured at 20°C and maintained on nematode growth medium (NGM) plates seeded with Escherichia coli OP50 unless indicated otherwise. All worms grew at least three generations under normal, unstressed conditions prior to use in experiments. Mutant strains were obtained from the CGC and backcrossed with the wild type Bristol N2 strain at least three times.

Caenorhabditis elegans were developmentally synchronized from a 4 h egg lay and transferred to 10 cm peptone-free nematode growth media (NMG) plates spotted with 10¹¹ CFU/mL (AL) or 10⁹ CFU/mL (DR) OP50 upon reaching day 1 of adulthood. Plates had 25 mg/mL of carbenicillin to prevent bacterial growth. The worms were transferred daily to fresh plates until they reached the post-reproductive stage and then transferred every 4 days. The animals were scored for survival every other day by monitoring the movement after tapping the worm gently with a platinum wire.

Synchronized worms were transferred to fresh plates daily for 1 week and then shifted to 35°C for 4 h. After the heat shock, the nematodes were monitored daily for survival.

NGM plates were spotted with OP50. After 1 day, either DMSO or 25 μg/mL tunicamycin dissolved in DMSO was added to each plate. Another day was allowed to pass before animals were exposed to treated agar plates. Due to the large number of animals required for experiments involving polysome profiling, animals were synchronized by bleaching gravid adults. In translation and eIF-2α phosphorylation experiments, synchronized adults were transferred to tunicamycin or DMSO plates for 3 h and washed 3 times in M9 prior to analysis. To minimize vulval ruptures in survival assays, adult worms were treated with 50 μg/mL 5-fluoro-2-deoxyuridine (FUDR) for 2 days prior to exposure to tunicamycin or DMSO.

150 µL of pelleted day 1 adult worms were lysed on ice in 350 µL of solubilization buffer (300 mM NaCl, 50 mM Tris–HCl [pH 8.0], 10mM MgCl₂, 1 mM EGTA, 400 U RNasin/mL, 3.55 µM cycloheximide, 1% Triton X-100, 1 mM PMSF, 1 mini tablet of protease inhibitor [Thermo Sicentific] and 0.1% sodium deoxycholate). An additional 200 µL of solubilization buffer was added after homogenization and the samples were incubated on ice for 1 h. Lysates were centrifuged at 14,000 × g at 4°C for 5 min and the supernatants were collected. 300 μL of each sample was loaded to the top of a 5%–50% sucrose gradient in high salt resolving buffer (140 mM NaCl, 25 mM Tris–HCl [pH 8.0], and 10 mM MgCl₂) and centrifuged in a Beckman SW41Ti rotor at 38,000 × g at 4°C for 2 h. Gradient fractionation was conducted with a Piston Gradient Station equipped with a Triax flow cell (BioComp Instruments) and the absorbance was monitored at 260 nm continuously. Polysome profile quantification was carried out using R with codes adapted from Shaffer and Rollins (2020).

Bacteria were spotted into 12-well plates containing 2 mL of peptone-free NGM agar. Day 1 adults were assigned to AL (ad libitum, i.e., full fed), fasting (no bacteria, FT), or AL supplemented with 10 mM cycloheximide and incubated for 3 h. 300 μL of 1 mM puromycin was added to each well and incubated for 1 h. The worms were washed three times with M9 and 50 µL of worm slurry was frozen immediately in liquid nitrogen. To extract protein, worms were lysed on ice in 100 µL of solubilization buffer (300 mM NaCl, 50 mM Tris–HCl [pH 8.0], 10mM MgCl₂, 1 mM EGTA, 1% Triton X-100, 1 mini tablet of protease inhibitor, and 1 mM PMSF) for 1 h and centrifuged at 14,000 × g at 4°C for 5 min. The Western blotting was conducted as described in the next section.

Animals used in experiments were Day 1 adults. Worm collection, lysis, and protein extraction were performed as described in the “SUnSET puromycin assay” section. For SDS-PAGE, 20 µg of total protein as determined by Bradford protein assay was loaded on 4%–20% mini-Protean TGX stain-free gels (Bio-Rad) and separated at 100 V for 5 min followed by 150 V for 1 h. The SDS-PAGE gels were exposed to UV for 2.5 min for total protein quantity detection and then transferred to PVDF membranes. The membranes were incubated with primary antibody in EveryBlot blocking buffer (Bio-Rad) at 4°C overnight and further incubated with secondary antibody for 1 h after washing. The protein density was quantified using ImageJ 1.47V and normalized by the total quantity of protein. For puromycin assays, anti-puromycin Ab (clone 12D10, Sigma; 1:5000) was used followed by anti-mouse secondary Ab. For eIF2α S49 phosphorylation detection, the antibodies were anti-phospho-eIF2α rabbit polyclonal Ab (1:3000 dilution; Cell Signaling) and anti-rabbit secondary Ab. Three replicates were performed.

The NMD reporter strain PTCXi was described previously (Longman et al., 2007). Larval 4 stage worms were added to AL, DR or smg-2 RNAi (Y48G8AL.6; Vidal library) (Rual et al., 2004) plates and transferred daily in the same conditions for 3 days. Fluorescence microscopy was performed with individual C. elegans using a Leica M165FC stereo microscope in the GFP channel (narrow band filter set, excitation ET470/40 nm, emission ET510/10 nm). Each worm was imaged on a 1% agarose pad on a glass slide and immobilized with a drop of 20 mM levamisole. Quantification of fluorescence was conducted using ImageJ for mean pixel intensity after correcting for background fluorescence. At least 15 worms were included in each replicate. Three replicates were performed.

Adult worms were treated as described and then collected in TRIzol reagent (Invitrogen). RNA extraction and purification were performed using RNA Clean and Concentrator kit (Zymo Research) according to the manufacturer’s protocol followed by cDNA synthesis (QuantiTect Reverse Transcription kit [Qiagen]). qPCR analysis was conducted in technical triplicate using KAPA SYBR FAST qPCR Master Mix on a LightCycler 480 (Roche Applied Science, Indianapolis, IN, USA). Primers used are listed in Supplementary Table S2. The housekeeping gene cdc-42 was used for target gene mRNA normalization. Gene expression changes were analyzed by the 2^−ΔΔC _T method. Three replicates were conducted.

Worms were synchronized by a 1-h timed egg lay. Time to reproductive development was measured from the time an egg was laid until it became an egg-laying adult. The fecundity of nematodes was determined by counting hatched progeny. Both development and fecundity assays were conducted at 20°C. At least 12 worms were included in each replicate. Three replicates were performed.

In order to understand the role of ISR in response to nutrient deprivation, we started by assessing changes in translation. Polysome profiling provides an assessment of translation based on ribosome association with mRNA. A redistribution of ribosomes from high translation polysomes (i.e., mRNA bound by 2 or more ribosomes) to low translation monosomes was observed on young adult N2 worms after fasting (FT) for 3 h compared to full fed (ad libitum, AL) (Figure 1A). Quantitatively, the relative abundance of polysomes decreased by 20.3% ± 1.7% in this time (p = 0.0009). However, more signal was lost in the highest translation polysomes (mRNA bound by 4 or more ribosomes), indicating that absolute levels of translation were likely decreased by more than 20%. As a separate measure of translation rate, surface sensing of translation (SUnSET) was optimized for use under DR conditions in C. elegans. Previous application of SUnSET in C. elegans relied on ingestion of bacteria mixed with puromycin (Arnold et al., 2014; Heissenberger et al., 2020). To allow puromycin incorporation in worms without food (i.e., from fasted animals), the mutant strain bus-5(br19) was used due to its enhanced cuticle permeability (Xiong et al., 2017), allowing this strain to incorporate puromycin at a higher rate compared to N2 (Somers et al., 2022). In this assay, bus-5(br19) animals were incubated with puromycin for 1 h followed by Western blot to detect puromycin signal. N2 animals were included for comparison. As expected, since the 1 h incubation was much shorter than the traditional puromycin assay, almost no puromycin signal could be detected in N2 (Figure 1B). In contrast, the bus-5(br19) mutant showed clear puromycin incorporation in both AL and fasting conditions. However, almost no signal was observed in worms treated with cycloheximide, an inhibitor of mRNA translation. A reduction in translation by 37.3% was detected in worms fasted for 1 h and 48.9% after 3 h (Figure 1B, lower panel). Taken together, both types of translation assay demonstrate translation reduction under short-term FT. Furthermore, the modified SUnSET method utilizing bus-5(br19) mutants provides an effective approach for assessing translational changes in C. elegans that is not affected by differences in food availability or pharyngeal pumping rate of worms.

It is not clear what role ISR plays in the attenuated translation and extended longevity under DR conditions. We performed Western blot to detect the activation of the ISR by eIF-2α phosphorylation. We included the knockout mutants gcn-2(ok871) and pek-1(ok275), which were crossed to generate the double mutant gcn-2(ok871);pek-1(ok275). We also included a mutant eif-2α(rog3), hitherto referred to simply as eif-2α, with serine 49 replaced by alanine, which cannot be phosphorylated. All these mutants and N2 were treated with AL or FT conditions for 3 h (Figure 1C). In N2, fasted animals showed higher eIF-2α phosphorylation compared to AL, indicating that the ISR was activated under the fasting condition. In the mutants, only pek-1(ok275) showed a similar pattern to wild type, whereas the signal could not be detected in other mutants. Results indicate that GCN-2, but not PEK-1, is responsible for eIF-2α phosphorylation under FT conditions as expected.

Next, we wanted to see how translation is affected in phosphorylation and kinase mutants under short-term fasting conditions. Therefore, we used polysome profiling and SUnSET to determine how translation is influenced in ISR mutants after 3 h FT. Polysome profiling showed that, regardless of the mutations, translational repression was observed in the eif-2α phospho-null mutant as well as gcn-2(ok871) and pek-1(ok275) (Figure 2A). In each instance, polysomes were reduced about 20% (Figure 2B), indicating that phosphorylation of eIF-2α is not required for these early changes in translation when food is removed.

In order to compare total changes in translation between control and the eif-2α phosphorylation mutant using SUnSET, bus-5(br19) was crossed with eif-2α mutants and animals were fasted for 3 h (Figure 2C). Both bus-5(br19) and bus-5(br19);eif-2α mutants showed reduced translation under FT conditions. Interestingly, translation is constitutively low in eif-2α phospho-null mutants compared to WT animals in both AL and fasting conditions (Figure 2C). Constitutively low translation is interesting from the point of view of NMD, which requires translation for its activity and which is, itself required for longevity associated with DR in C. elegans (Rollins et al., 2019; Kim et al., 2020). Importantly, results indicate that translation is downregulated in eif-2α phospho-mutants under FT compared to AL conditions (Figure 2C). Thus, the ISR is not required for translation attenuation under DR conditions.

The activation of NMD relies on translation, thus we sought to confirm whether the ISR downregulates NMD when food is removed. To test its activity, we used the NMD reporter strain PTCXi harboring a PTC that normally abrogates expression of a GFP reporter (Longman et al., 2007). In environments without stress where NMD is highly active, the PTC containing transcript is degraded by NMD, precluding expression of GFP. We tested wild type and the eif-2α phospho-deficient mutant in response to DR. smg-2 RNAi was used as a control, since this gene is required for NMD. As expected, the worms on smg-2 RNAi showed the highest GFP signal. The animals under DR also increased their fluorescence significantly compared to AL, but not as strong as the one in smg-2 RNAi, indicating NMD decreased under DR, but did not turn off completely (Figures 3A, B). The fluorescence in the NMD reporter crossed with eif-2α presented a very similar pattern. The worms under DR increased GFP signal significantly, suggesting that NMD was also downregulated in the phospho-null mutant and that the ISR was not required for suppressed NMD activity under DR.

To better understand the relationship between ISR and NMD, we measured expression of certain transcripts with intron retention (IR), which causes out-of-frame translation and generates at least one PTC. We tested the isoform expression relative to productively spliced transcripts in N2 and eif-2α phospho-mutants under AL and FT conditions for 3 days. The first set of genes tested were the alternatively spliced ribosomal protein mRNA rpl-1, rpl-3, rpl-7A, and rpl-12 that are known NMD targets (Mitrovich and Anderson, 2000). Except rpl-7A that did not show expression change, the IR containing isoforms of rpl-1, rpl-3, and rpl-12 decreased in FT compared to AL significantly (Supplementary Figure S1). Results suggest that alternative splicing decreased and/or NMD increased for these factors. We also analyzed the expression of atf-4, which is the main ISR effector (Supplementary Figure S1E) but did not observe any statistical change in the expression level (p = 0.09 for N2 AL-DR comparison). Furthermore, the expression pattern was very consistent in wild type and eif-2α in all these genes, indicating that these large ribosomal subunit transcripts are not dependent on the ISR for changes in NMD. We further tested the expression of the same IR transcripts in gcn-2(ok871);pek-1(ok275) to confirm the relationship between ISR and NMD. Similarly, under FT only uaf-1 and ifbp-1 increased their expression, but not any other transcripts tested (Supplementary Figure S2 and Supplementary Figure S3), The same pattern was observed in the wild type, confirming again that ISR does not regulate NMD under FT.

We then analyzed the expression of IR containing mRNA with increased abundance under conditions of a food dilution form of DR according to RNAseq data from a previous study (Rollins et al., 2019). We selected the isoforms with IR of uaf-1, ifbp-1, cyp-35A2, ddo-2, and rpl-22. UAF-1 is a large subunit of splicing factor U2AF (U2 auxiliary factor) involved in the recognition of 3’ splice sites (Zorio and Blumenthal, 1999; Ma and Horvitz, 2009). IFBP-1 is an ortholog of human IRF2BP2 and IRF2BPL, which are members of the IRF2BP (interferon regulatory factor 2 binding protein) family of transcriptional repressors (Childs, 2003), while CYP-35A2 is a protein in Cytochrome P450 family (Lim et al., 2022). DDO-2 encodes D-aspartate oxidase that degrades D-amino acids (Saitoh et al., 2012). RPL-22 is a large subunit of ribosomal protein. In these mRNAs, uaf-1 showed the highest increase of expression from 100% to 934.6% in N2 and 134.3% of the N2 AL value to 963.9% in eif-2α respectively (Figure 3C). The fact that UAF-1 is a splicing factor highlights the self-regulation of splicing factors through Alternative splicing and NMD (AS-NMD) (Lejeune, 2022) The alternatively spliced isoform of ifbp-1 also went up 81.9% in N2 and 72.55% in eif-2α respectively (Figure 3D). The higher levels of uaf-1 and ifbp-1 with IR under FT conditions is suggestive of downregulation of NMD for these genes. Interestingly, cyp-35A2, ddo-2, and rpl-22 that increased IR events under DR as previously reported did not increase their abundance in fasted animals (Figures 3E–G), which could be due to experimental condition variation (i.e., food dilution versus FT). However, the gene expression pattern in all these genes were very similar between N2 and eif-2α, suggesting again that the ISR is not a key player to regulate NMD when nutrients are limited.

To address whether inactivation of the ISR displays detrimental effects on longevity under DR, we compared lifespan of eif-2α phosphorylation and ISR kinase mutants under AL and DR. Similar to N2, DR promoted lifespan in eif-2α, gcn-2(ok871), and pek-1(ok275), indicating that neither phosphorylation of eIF-2α, nor the individual kinases targeting this translation factor, are needed for lifespan extension under DR (Figure 4; Supplementary Table S3). Interestingly, the lifespan of double mutant gcn-2(ok871);pek-1(ok275) under DR was not significantly different from AL (Figure 4E). This observation suggests more than one phosphorylation target is regulated by these kinases. As with eIF-2α, the target may be shared between GCN-2 and PEK-1. Taken together with earlier results, the ISR is not required for changes in translation, NMD, or lifespan associated with DR. Furthermore, eif-2α lived longer than wild type under AL feeding (p = 0.0053; Supplementary Figure S4 and Supplementary Table S3), indicating that the inability to activate the ISR positively influences longevity under the conditions tested. This finding is consistent with previously reported results (Derisbourg et al., 2021b).

Considering that the ISR is dispensable under DR involving food dilution, we wanted to see if it is required for other forms of stress. To induce ER stress, C. elegans were treated continuously with tunicamycin starting from young adulthood. Compared to N2, the survival of eif-2α was reduced significantly under ER stress (Figure 5A; Supplementary Table S6). Thus, the ISR is required for a wild-type level of survival under conditions of ER stress.

Given the lack of requirement for the ISR in lifespan extension under DR (Figure 4), we wondered if imposing a stronger form of DR requires the ISR for enhanced survival, since this may lead to a greater level of uncharged tRNA associated with its activation. Under fasting conditions starting from young adulthood, both N2 and eif-2α exhibited an extension in lifespan characteristic of this form of DR (Figure 5B; Supplementary Table S5). Notably, under both AL and fasting conditions, the mutant demonstrated a longer lifespan compared to wild-type animals. We also tested the lifespan of gcn-2(ok871);pek-1(ok275) under fasting due to its non-extended lifespan under DR (Supplementary Figure S5). The results were similar to the one under DR. The survival rate of the double mutant did not differ from AL, also suggesting an unidentified shared target exists between GCN-2 and PEK-1 essential for lifespan extension in nutrient deprivation.

Although heat stress is not typically associated with the ISR, we included it since increased resistance to heat stress is frequently associated with gene changes that increase lifespan as observed for the eif-2α mutant under food dilution DR (Figure 4B and Supplementary Table S3) and fasting (Figure 5B). To investigate whether the ISR is involved in the response to heat stress, day 7 adult worms were incubated at 35°C for 4 h and their survival was scored daily thereafter. The inhibition of EIF-2α phosphorylation did not decrease the survival rate after heat stress (Figure 5C; Supplementary Table S4). As observed with survival under standard conditions, the eif-2α phospho-null mutant showed increased survival compared to N2.

Since eif-2α lived longer than N2 animals, we investigated if ISR inhibition affects development and reproduction of animals. We measured the development time of eif-2α and found that it was not significantly different between the mutant and N2, with both taking between 78 and 80 h to go from egg to egg-laying adult (Figure 6A). However, the total number of progeny was significantly lower in eif-2α compared to N2 (Figure 6B). Therefore, inhibition of the ISR leads to an extension of lifespan in the absence of stress, but it also showed diminished reproductive ability. This is consistent with trade-offs frequently observed in long-lived mutants.

To gain further insights into the response of the ISR to ER stress, we tested the eIF-2α phosphorylation status of kinase mutants upon exposure to tunicamycin. After 3 h of tunicamycin exposure, only N2 and the gcn-2(ok871) mutant were able to phosphorylate eIF2α, as expected. Conversely, pek-1(ok275) and gcn-2(ok871);pek-1(ok275) double mutants could not (Figure 7A), showing that only PEK-1 is responsible for ISR activation under this condition.

Since PEK-1 is required for normal survival and eIF-2α phosphorylation in response to tunicamycin, we wondered whether translational downregulation associated with ER stress and ISR activation was impaired in pek-1 mutants. Polysome profiling was conducted after the worms were challenged with tunicamycin for 3 h. In N2, the polysome fraction was reduced by 19.5% (Figures 7B, D). In the pek-1 mutant, translation remained near DMSO control levels in the presence of tunicamycin (Figures 7C, D), which is consistent with the fact that pek-1 is essential to activate ISR under ER stress. The double mutant gcn-2(ok871);pek-1(ok275) also exhibited a muted response compared to N2 (Figure 7D and Supplementary Figure S6C). There was no statistical difference between pek-1(ok275) and the double mutant (p = 0.093). Surprisingly, the eif-2α phospho-mutant showed similar reduction in translation to N2 (Figure 7D and Supplementary Figure S6A), despite the importance of this residue for survival under ER stress (Figure 5A). To confirm the importance of PEK-1 for survival under ER stress, we challenged the ISR kinase mutants with long-term tunicamycin exposure. As expected, the survival rate of pek-1(ok275) and gcn-2(ok871);pek-1(ok275) dropped dramatically (Figure 7E) while the DMSO control worms all showed similar lifespans (Supplementary Figure S6D). Taken together, PEK-1 activates the ISR under ER stress and the ISR is necessary for the ER stress response.