HSF-1 is required for larval development as the null mutant exhibits larval arrest and lethality (Li et al., 2016; Morton and Lamitina, 2013). To examine HSF-1’s contribution to lifespan and uncouple it from its impacts on development, we took advantage of the AID system to deplete HSF-1 from the somatic tissues post-larval development. Our recent work has shown that expression of the TIR1 E3 ligase and degron tagging at the endogenous HSF-1 do not alter larval development while enabling efficient depletion of HSF-1 within 2 h of auxin treatment in adult animals (Edwards et al., 2021). Depletion of HSF-1 from Day 1 in young adults shortened the median lifespan by more than one third (Figure 1A and Table 1). Time-course analyses revealed that HSF-1 functions throughout the self-reproductive period in hermaphrodites to support lifespan (Table 1). On Day 5 of adulthood, when >98% of progenies were already produced (Supplementary Figure S1A), depletion of HSF-1 still resulted in a small yet significant lifespan shortening. It has been reported that auxin treatment itself extends lifespan in a concentration-dependent manner (Loose and Ghazi, 2021), and in one of our lifespan trials, we observed a modest lifespan extension when applying the standard 1 mM auxin to the control strain that expresses TIR1 but has no degron insertion at HSF-1 (Table 1, Trial #2). We, therefore, repeated the lifespan analyses using 0.5 mM auxin (Figure 1B). This experiment confirmed the results as using 1 mM auxin while it had much smaller effects on the control strain (Table 1, Trial #2). It is known that signals from the reproductive system could impact C. elegans lifespan (Hsin and Kenyon, 1999). Though depletion of HSF-1 from the adult germline dramatically reduced fecundity, depletion of HSF-1 from the soma had a modest impact on the brood size (Supplementary Figure S1B). Thus, our results suggest that HSF-1 functions directly through somatic maintenance to support lifespan. Collectively, our data determined the temporal requirement for HSF-1 in a normal lifespan, which overlaps with the self-reproductive period in early adulthood.

HSF-1 is required for lifespan extension in multiple longevity pathways including those mediated by arrested germline stem cells (GSC) and reduced insulin/IGF-1 signaling (IIS). To understand the roles of HSF-1 in longevity pathways post-larval development, we performed lifespan analyses with HSF-1 depleted at different time points of adulthood in the long-lived glp-1(e2141) <arrested GSC> and daf-2(e1370) <reduced IIS > mutants. As a control for glp-1(e2141), we included another temperature-sensitive mutant fem-3(q20) that is sterile as glp-1(e2141) at the restricted temperature of 25°C < only producing sperm but not oocytes > but has a normal lifespan in the analyses. Similar to wild-type, HSF-1 is required in early adulthood (to at least Day 4 at 25°C) for the lifespan of fem-3(q20), and pan-somatic depletion of HSF-1 since Day 1 of adulthood shortened lifespan by about one third (Figure 2A; Table 2). HSF-1 makes a bigger contribution to the glp-1(e2141) lifespan, and its functional period is extended in glp-1(e2141) (Figure 2B; Table 2 Trial #1). These effects are more obvious when the median lifespan was further increased as animals were fed with antibiotic-treated bacteria (Figure 2C; Table 2 Trial #2 and #3). The additional lifespan extension was likely due to reduced bacterial infection since glp-1(e2141) delays later deaths with an atrophied pharynx but not earlier deaths from pharyngeal pathology by a bacterial infection (Zhao et al., 2017). In this growth condition, HSF-1 supported lifespan till at least Day 8, and depletion of HSF-1 in Day 1 adults resulted in >60% of lifespan reduction. It is noteworthy that depletion of HSF-1 was still effective in both the fem-3 (q20) and glp-1(e2141) animals on Day 4 (Supplementary Figures S2A, B), therefore, the relatively modest lifespan shortening effects in fem-3(q20) are not simply an artifact of defective AID during aging but rather suggest that HSF-1 is less active or its function is minimally required for the lifespan of fem-3(q20) beyond Day 4 at 25°C. While the signal of GSC arrest in glp-1(e2141) more than doubled the lifespan in the presence of HSF-1 (when fed with antibiotic-treated bacteria) (Figure 2D; Table 2, Trial #2), it had marginal effects if HSF-1 was depleted since Day 1 of young adults. These results indicate lifespan extension by arrested GSC is completely dependent on HSF-1 in adulthood, and provide an example that HSF-1’s roles in larval development and longevity are temporally uncoupled.

We also tested the roles of HSF-1 in longevity by reducing IIS using the daf-2(e1370) mutant. Due to the high incidence of internal hatching upon HSF-1 depletion on Day 1 (54%, 189 out of 350) and Day 3 (39%, 136 out of 350) of daf-2(e1370), it is difficult to estimate the total contribution of HSF-1 to lifespan throughout adulthood. However, similar to the glp-1(e2141) mutant, daf-2(e1370) also seems to extend the functional period of HSF-1 as depletion of HSF-1 since Day 5 and Day 7 resulted in ∼35% and ∼20% of lifespan reduction in daf-2(e1370) (Supplementary Figure S2C, Table 2 Trial #4) while depletion of HSF-1 since Day 5 in the wild-type only led to modest (4–8%) lifespan reduction (Table 1). This becomes apparent when daf-2(e1370) animals were fed with antibiotic-treated bacteria, in which HSF-1 contributes to longevity up to Day 19 of adulthood (Supplementary Figure S2D, Table 2 Trial #5). Collectively, we found longevity by GSC arrest and reduced IIS correlates with extended functional period of HSF-1.

The prevailing view is that HSF-1 promotes longevity through its ability to activate the HSR. HSF-1 also drives a transcriptional program that is different from the HSR in C. elegans larval development (Li et al., 2016). In addition, enhancing HSF-1 activities by transgenic over-expression of HSF-1, mild mitochondrial perturbation, or ablation of its negative regulator HSB-1 extends lifespan through functions beyond inducing the canonical HSR (Baird et al., 2014; Kumsta et al., 2017; Higuchi-Sanabria et al., 2018; Egge et al., 2019; Williams et al., 2020; Sural et al., 2020). To better understand the molecular mechanism underlying the physiological roles of HSF-1 in lifespan, we set out to determine the transcriptional program of endogenous HSF-1 in somatic cells on Day 1 of young adults. Recently, we have determined HSF-1 binding sites specifically in the soma or the germline and binding sites used in both tissue types through whole animal ChIP-seq analyses following an acute depletion of HSF-1 in the soma or germline (Edwards et al., 2021). Among those sites, 79 promoter-associated HSF-1 binding peaks are either enriched in the soma (e.g., Y94H6A.10) or shared by the soma and germline (e.g. hsp-1/hsc-70) (Figure 3A). This result suggests that albeit the number of binding peaks is smaller compared to that in larval development (Li et al., 2016), HSF-1 can bind to promoters without thermal stress in adult somatic cells that are postmitotic.

We then examined the transcriptional impact of HSF-1 binding via RNA-seq analyses following HSF-1 depletion from somatic cells by AID. We performed RNA-seq in the control strains that express TIR1 but have no degron insertion in a time course of auxin treatment (Supplementary Figures S3A–D). This set of experiments, when compared to those using the experimental strains with degron tagging at HSF-1, identified the small groups of differentially expressed (DE) genes caused by auxin treatment (Supplementary Figure S3A, mock vs. auxin treatment in the control strain) or by degron insertion at HSF-1 (Supplementary Figure S3B, compare the control and experimental strains with mock treatment). As auxin treatment resulted in almost identical transcriptional changes in the two control strains that express TIR1 in the soma and germline, respectively (Supplementary Figure S3C), we conclude that these changes were due to auxin rather than off-target effects of TIR1. By filtering out HSF-1-independent changes (Supplementary Figures S3A, B, Supplementary Table S1), our methods specifically determined DE genes caused by HSF-1 depletion (Supplementary Figure S3D). Among the 80 DE genes resulting from 24 h of HSF-1 depletion in the soma, 16 out of 50 down-regulated genes have HSF-1 binding at the promoter (Supplementary Figure S3E, Supplementary Table S2) while only 1 out of 34 up-regulated genes are bound by HSF-1, suggesting that HSF-1 functions as a transcriptional activator in somatic cells of young adults. The group of 16 HSF-1-bound down-regulated genes exhibit a decrease of RNA Polymerase II (Pol II) occupancy both at the promoter and at the end of genes (Figure 3B) upon 2 h depletion of HSF-1 suggesting that they are likely the direct targets of HSF-1 and that HSF-1 functions at the step of Pol II recruitment. For examples, a decrease of Pol II occurs across the genes of Y94H6A.10 and hsp-1/hsc-70 upon HSF-1 depletion in the soma (Figure 3A). On the contrary, down-regulated genes without HSF-1 binding and up-regulated genes (including the only HSF-1 bound gene, bath-36 in this group) lack a correlation between mRNA (24 h post HSF-1 depletion) and Pol II occupancy changes (2 h of HSF-1 depletion), suggesting they are indirectly impacted by HSF-1 (Figure 3B).

Among the direct targets of HSF-1 in somatic tissues of young adults (Supplementary Figure S3F), most have substantial HSF-1 binding in the germline as well (Y94H6A.10 is the only exception). To control for the potential interference by mRNA from the germline and increase the specificity and sensitivity of RNA-seq analyses, we also performed the experiments in the germline deficient glp-1(e2141) background. Indeed, we have uncovered a bigger group of HSF-1 direct targets (28 genes), each of which is associated with one of the 79 HSF-1 binding peaks at the promoter in the soma and decreases mRNA upon 8 h and/or 24 h of HSF-1 depletion (Figure 3C, Supplementary Table S2). Similar to the list of genes identified in the wild-type background, the majority of these genes (22 out of 28) (Figure 3C, gene names in black) encode either molecular chaperones or co-chaperones, which function together in protein folding and protein conformation maintenance. These chaperone/co-chaperone genes are grouped into two clusters with one including canonical HSR genes encoding members of HSP70 and small heat shock proteins (HSP-16) whose expression is highly inducible by stress (Figure 3C, cluster #1: red nodes), and the other cluster enriched with constitutively expressed chaperones and co-chaperones (Figure 3C, cluster #2: green nodes). The latter group contains the essential HSP-90 and HSC-70 chaperones as well as their co-chaperones (DNJs, FKB-6, STI-1, AHSA-1, HSP-110, and the HSPBP1 orthologue Y38E10A.22). It also contains subunits of the chaperonin complexes in the cytosol (CCT-3, CCT-7, and CCT-8) and mitochondria (HSP-60). These HSF-1-dependent, constitutively expressed chaperone and co-chaperone genes span most the ATP-dependent chaperone systems in metazoan (Clare and Saibil, 2013) and dictate protein folding capacity in the cytosol, nuclei, and mitochondria. It is noteworthy that the majority of genes in cluster #2 (14 out of 16) belong to the developmental HSF-1 transcriptional program (Li et al., 2016), which are regulated by HSF-1 differently from the HSR. As genetic perturbation of the HSP-90, HSC-70 and cytosolic chaperonin systems accelerates age-dependent proteostatic and physiological declines (Brehme et al., 2014), our data suggest a role of HSF-1 in longevity by activating the expression of these chaperone systems in fully-developed somatic cells.

A decrease of this selective ‘sub-chaperome’ precedes massive transcriptomic changes upon HSF-1 depletion in glp-1(e2141), which also supports that they are the primary targets of HSF-1 in adult somatic cells. 19 out of 22 HSF-1 directly regulated chaperone/co-chaperone genes already decrease expression at 8 h of HSF-1 depletion (Supplementary Table S2). They remain the most prominent functional group (‘protein folding’) among ∼400 DE genes at 24 h of HSF-1 depletion as shown by Gene Ontology (GO) analyses (note the enrichment of genes with ‘ATPase activity’ results from ATP-dependent chaperones, Figure 3D). 48 h of HSF-1 depletion, however, led to much bigger changes in the transcriptome (>1200 DE genes, Supplementary Table S2). Genes in the ubiquitin–proteasome system (UPS) are up-regulated (Figure 3E), which is a typical response to protein misfolding, implicating imbalanced proteome over time caused by loss of HSF-1 and consequent decline in protein folding. This idea is supported by the increased levels of ubiquitylated proteins in somatic tissues upon 48 h of HSF-1 depletion (Figure 3F, Supplementary Figures S3G, H), and is consistent with published results that HSF-1 RNAi leads to increased protein misfolding and aggregation (Nollen et al., 2004; Ben-Zvi et al., 2009).

We also identified a few nonchaperone genes as HSF-1 direct targets (Figure 3C, gene names in red), although their roles in lifespan are not clear. Two of the encoded proteins, MBF-1 (Multiprotein Bridging Factor 1) and W03F9.1/ZPR1 in cluster #2 are proposed to be transcriptional coactivators and interact with chaperones/co-chaperones based on the studies of their orthologues in other model systems (Arce et al., 2006; Kannan et al., 2020). Especially, the mbf-1 gene is a direct target of HSF-1 in C. elegans larval development (Li et al., 2016), and expression of its orthologues in yeast and Drosophila is also controlled by HSF1 (Birch-Machin et al., 2005; Pincus et al., 2018). Similar to mbf-1, expression of Y94H6A.10 (a gene with unknown function) and F19B2.5/HLTF (encodes an SWI/SNF chromatin remodeler family protein) is activated by HSF-1 in larval development, suggesting that at least part of HSF-1 developmental transcription program sustains in adulthood.

To test whether the transcriptional program of HSF-1 identified in glp-1(e2141) is unique to this longevity model and how much it is linked to the higher growth temperature (25 vs. 20°C for wild-type), we performed RNA-seq analyses in the fem-3(q20) control at 25°C. The fem-3(q20) shows a similar decrease of HSF-1 direct targets including constitutively expressed chaperones, stress-inducible chaperones as well as nonchaperone genes upon 24-h HSF-1 depletion (Supplementary Figures S3I, J). Although there is a trend of a bigger difference in glp-1(e2141) compared to that in fem-3(q20) (Supplementary Figure S3I), not all HSF-1 direct targets are more highly expressed in glp-1(e2141) as one may expect if HSF-1 is hyperactivated by GSC-mediated longevity signal (Supplementary Figure S3J). The mRNA of stress-inducible hsp70 and hsp-16.2 are at higher levels in glp-1(e2141) compared to fem-3(q20), suggesting expression levels of these canonical HSR genes are not solely dictated by the growth temperature. On the contrary, higher mRNA levels of constitutively expressed chaperones are observed in fem-3(q20). This expression pattern is likely linked to the different cell compositions in these two types of animals. Though fem-3(q20) is sterile at 25°C, it still has a germline that makes sperm. The smaller changes upon depletion of HSF-1 from the soma in fem-3(q20) could be well explained by the ‘masking effect’ from unchanged mRNAs in the germline. Consistent with this idea, our recent paper shows that hsc-70 and hsp-90 mRNAs are enriched in the germline while HSR is more robust in the soma (Edwards et al., 2021), therefore fem-3(q20) with more germ cells shows higher levels of constitutive chaperones but lower levels of inducible chaperones in whole animal analyses. Collectively, the HSF-1 transcriptional program that we identified in the germline deficient glp-1(e2141) likely applies to somatic cells of animals with normal lifespan as seen in fem-3(q20). Future studies are needed to determine whether HSF-1 activities are enhanced by longevity signals in the absence of external stress.

While most of the HSF-1-dependent, constitutively-expressed chaperones and co-chaperones have roles in all cell types, over-expression of HSF-1 in specific tissues (e.g., neural cells) is sufficient to extend lifespan (Morley and Morimoto, 2004; Douglas et al., 2015). To understand the tissue requirement for endogenous HSF-1 in lifespan, we have made transgenic models that express TIR1 E3 ligase specifically in one of the major somatic tissue types (neural system, intestine, muscle, and hypodermis) to enable tissue-specific depletion of HSF-1 by AID. All of our AID models successfully depleted HSF-1 from the nuclei in the target tissues within 2 h of auxin treatment (Supplementary Figures S4A–D). We also checked HSF-1:degron:GFP in the neighboring tissues and were able to confirm the specificity of AID except for our neural model, in which the nuclear HSF-1 seemed also depleted in intestinal cells near the head but not in the center or toward the tail (Supplementary Figure S4D). The same neural TIR1 transgene did not deplete degron:GFP (not fused with HSF-1) in intestinal cells near the head as it did for HSF-1:degron:GFP upon auxin treatment (Supplementary Figure S4E). This result implies depletion of HSF-1 from a subset of intestinal nuclei in our neural AID model is unlikely due to leaky expression of TIR1 but instead dependent on neural HSF-1. However, due to high levels of auto-florescence in the intestine, future studies with more sensitive and quantitative methods to measure HSF-1 protein levels and its localization, and using alternative neural AID models are needed to confirm the results.

We then tested the tissue requirements of HSF-1 for larval development. Loss of HSF-1 in any of these tissues since egg lay led to larval developmental arrest or delay (Figure 4A, Supplementary Figures S5A–C). HSF-1 depletion from hypodermis caused larval arrest at L3-L4, and the animals were associated with molting defects (Supplementary Figure S5D). Animals with HSF-1 depleted in the intestine, neural cells or muscle managed to develop into adults. Intestinal depletion of HSF-1 led to a huge larval delay, more than doubling the time needed for developing into adults (Supplementary Figure S5A). Loss of HSF-1 in neural cells and muscle had a relatively milder larval delay (Supplementary Figures S5B, C) but greatly reduced mobility eventually making the animals paralyzed at the young adult stage (Figure 4B, Supplementary Figure S5E). Depletion of HSF-1 in the muscle also led to egg-laying defects, resulting in 100% internal hatching (20/20 animals, Supplementary Figure S5F). It is noteworthy that none of the tissue-specific HSF-1 depletion phenocopied the L1-L2 arrest by pan-somatic depletion of HSF-1 (Edwards et al., 2021), suggesting that HSF-1 functions cooperatively in all the tissue types tested to support larval development.

We then performed lifespan analysis with tissue-specific depletion of HSF-1 on Day 1 of adulthood. We chose glp-1(e2141) as the model since pan-somatic depletion of HSF-1 in this background exhibited the biggest lifespan shortening effect, therefore, providing sufficient dynamic range to examine potentially smaller effects from HSF-1 depletion in a single tissue type. Depletion of HSF-1 in the intestine and neural cells both resulted in ∼55% lifespan shortening (Figure 4C, D, Table 3) which is only slightly smaller than that upon pan-somatic depletion (Figure 2C; Table 2). Depletion of HSF-1 from muscle led to a fairly small but still significant decrease (∼9%) in median lifespan (Figure 4E; Table 3). Despite the essentiality of HSF-1 in hypodermis during larval development, loss of HSF-1 from hypodermis post-larval development did not significantly alter lifespan (Figure 4F; Table 3). Collectively, our results indicate the different tissue requirements for HSF-1 in development and lifespan, suggesting that the roles of HSF-1 in these two processes are uncoupled.

Unless stated, C. elegans strains were maintained at 20°C on NGM plates seeded with OP50 bacteria and were handled using standard techniques (Brenner, 1974). The temperature-sensitive glp-1 (e2141) and fem-3(q20) animals were maintained at 15°C and grown at 25°C (since L1) for experiments.

The HSF-1 AID models were made by CRISPR knock-in of aid:gfp to the C-terminus of endogenous hsf-1 gene as detailed in our previous publication (Edwards et al., 2021). New tissue-specific TIR1 models were made by modification of CA1200 (eft-3p:tir1:mRuby) and swapping the eft-3 promoter with tissue-specific promoters. We first removed the eft-3 promoter through microinjection of two chemically modified synthetic sgRNA (Synthego) in CA1200, which are against the upstream and downstream regions of the eft-3 promoter, along with Cas9 Nuclease (Integrated DNA Technologies, IDT) following the previously published protocol (Prior et al., 2017). The DNA sequences targeted by the sgRNAs are GCT​CTG​GTA​CCC​TCT​AGT​CA (upstream) and AGT​TAC​GGT​CCT​TGT​CGA​GT (downstream) respectively. The resulting promoter-less allele contains a short insertion (‘GGCATCCA’) between the two cutting sites. We then inserted the tissue-specific promoters by microinjection of an sgRNA against that short insertion in the promoter-less allele (corresponding DNA sequence: GGT​CCT​TGT​TGG​ATG​CCT​CA) with Cas9 Nuclease and PCR fragments of tissue-specific promoters as the repair templates. The rab-3b (1.2 kb), dpy-7 (350 bp), and myo-3 (2.5 kb) promoters were used for the expression of TIR1 in neural cells, hypodermis, and muscle. The published allele in CA1209 (ges-1p:tir1:mRuby) was used for intestinal expression of TIR1. All the transgenic models made by CRISPR were outcrossed 6 times before use.

Auxin treatment was performed by transferring worms to bacteria-seeded NGM plates containing 1 mM (if not specified) or 0.5 mM auxin (Auxin: indole-3-acetic acid, Sigma). The preparation of auxin stock solution (400 mM in ethanol) and auxin-containing NGM plates was performed as previously described (Zhang et al., 2015). In all experiments, worms were also transferred to NGM plates containing 0.25% or 0.125% of ethanol (EtOH) to serve as the mock-treated control for 1 and 0.5 mM auxin respectively.

The HSF-1 AID animals and the corresponding control animals that only express TIR1 were age-synchronized by egg lay for 1 h on EtOH or Auxin plates. Larvae were grown for the indicated time (Figure 4 and Supplementary Figure S4) and crawling animals were recorded using a Leica M205 FA microscope. Videos were imported into ImageJ and analyzed for the size of animals (body length, mm) and mobility (body length per second, BLPS) using the wrMTrck plugin.

For brood size analyses (Supplementary Figure S1), animals were synchronized by egg lay and singled at L4/young adult stage onto plates containing ethanol (control) or auxin (HSF-1 depl) to lay eggs for 24 h. Worms were then transferred to new plates every day and eggs were allowed to hatch and grow to the L3 stage, at which point the number of progeny was counted.

Age-synchronized worms were scored as dead and removed in the absence of touch response or pharyngeal pumping 6 days per week. For fertile animals, worms were transferred to fresh plates every day through the reproductive period to remove progeny. Bagged, desiccated, or missing animals were censored from analysis.

In a subset of lifespan experiments, carbenicillin treatment of OP50 was performed as previously described to prevent bacterial growth (Lenaerts et al., 2008). Briefly, freshly grown E. coli OP50 cells were spun at 3000 × g for 20 min and resuspended in the same volume of M9 buffer supplemented with 0.5 mM of carbenicillin. The bacteria were then incubated in the shaker at 37°C for 3 h. The bacteria were concentrated by 5 fold (pelleting the cells, removing 80% M9 buffer, and resuspending) and seeded onto 10 cm NGM plates using 400 ul of concentrated bacteria per plate.

Imaging of live animals was done by immobilizing age-synchronized young adult animals in a drop of M9 buffer containing 6 mM levamisole on a 2% agarose pad. Fluorescent images (Supplementary Figure S2) were acquired immediately using a Zeiss LSM880 Confocal Microscope with a ×40 water objective. Zen software was used to obtain z-stacks and subsequent processing. Brightfield images (Supplementary Figure S4) were taken on a Zeiss axio observer inverted microscope with a ×20 air objective.

The HSF-1 AID animals and the corresponding control animals that only express TIR1 were synchronized by treatment of alkaline hypochlorite solution (bleach). The experiments in JTL611, JTL621 (HSF-1 AID models), and CA1200, CA1199 (control) were done as described in our recent paper (Edwards et al., 2021). For the experiments using the set of HSF-1 AID strains in glp-1 (e2141) and fem-3(q20) backgrounds, synchronized L1 larvae were grown for 48 h at 25°C on 10 cm normal NGM plates (∼500 worms per plate) to develop into young adults. Approximately 120 young adult worms were picked onto 10 cm NGM plates containing either EtOH or auxin and kept for an indicated time before collection (8, 24, and 48 h). For each condition, RNA was extracted using a 300 μL Trizol reagent. Worms were vortexed continuously for 20 min at 4°C and then went through one cycle of freeze-thaw to help release RNA. Following this, RNA was purified using the Direct-zol RNA MiniPrep kit (Zymo Research) as per the manufacturer’s instructions using on column DNase I digestion to remove genomic DNA.

Total RNAs were polyA enriched, and directional RNA-seq libraries were prepared using the NEBNext Ultra II RNA library prep Kit. Paired-end sequencing was done at a NovaSeq 6,000 sequencer at OMRF clinical genomics core. The majority of samples were sequenced by 50 bp. A subset of samples that were sequenced with longer reads was trimmed to 50 bp to make all downstream mapping and analyses consistent.

RNA-seq analyses were conducted as previously described (Edwards et al., 2021). Briefly, RNA-seq reads were mapped to the Ensembl WBcel235 genome using RNA STAR (Dobin et al., 2013). The mapped reads were then subject to FeatureCounts in Rsubread (Liao et al., 2019) for quantification. Differential expression (DE) analyses were then done using edgeR (Robinson et al., 2010) with default settings except for using Likelihood Ratio Test and filtering out those lowly expressed genes with CPM (counts per million) value less than 1 in more than 75% samples. Gene ontology analysis (GO) of DE genes was conducted using the program DAVID (http://david.abcc.ncifcrf.gov/) with functional annotation clustering to collapse redundant GO terms. The enrichment score for each cluster was shown.

ChIP-seq reads mapping, peak calling, and generation of genomic occupancy files (bedgraph files) were detailed in our recent publication (Edwards et al., 2021). To assign HSF-1 ChIP-seq peaks to promoters, transcription start sites (TSSs) determined by GRO-cap (Kruesi et al., 2013) were used where available. To visualize and compare the ChIP-seq data in genome browser views, the bedgraph files were normalized to reads per million using MACS2 call peak -B –SPMR and visualized using Integrative Genomics Viewer (IGV) (Robinson et al., 2011) with WS235 genome. Quantification of genomic occupancy was done by mapping the center of ChIP fragments to a reference point (e.g. HSF-1 peak summits) using windowBed in bedtools (Quinlan and Hall, 2010) and Matrix in R. For quantitative comparison of Pol II occupancy between conditions, all Pol II ChIP-seq data was normalized to 8 million reads, corresponding to the lowest coverage after duplicate filtering among all conditions. Heatmaps were generated with the Java TreeView package (Saldanha, 2004).

Worms were prepared by freeze-cracking and fixed in 4% paraformaldehyde, as described (Charlie et al., 2006) with the following changes. JTL667 (glp-1; eft-3p:tir1; hsf-1:degron) were synchronized by bleach synchronization, and L1 larvae were grown on NGM plates containing ethanol at 25°C for 48 h. Young adult worms were rinsed off in M9 buffer and plated on either ethanol or auxin plates for 48 h before rinsing for freeze-cracking. A methanol incubation for 15 min at 4°C after fixation was added to eliminate the TIR1:RFP signal in the worms. The samples were co-stained with Anti-Ubiquitinylated proteins (clone FK2, mouse, Sigma Cat# 04-263) at 1:200 and anti-REC-8 (rabbit, Novus Cat# 49230002) at 1:100 at 15°C overnight. We then labeled FK2 with anti-mouse-Alexa 647 (Invitrogen Cat# A32728) and anti-REC-8 with goat anti-rabbit-Alexa 488 (Invitrogen Cat# A32731) by incubating for 2 h at room temperature.