Age-dependent heat shock hormesis to HSF-1 deficiency suggests a compensatory mechanism mediated by the unfolded protein response and innate immunity in young Caenorhabditis elegans

Abstract

The transcription factor HSF-1 (heat shock factor 1) acts as a master regulator of heat shock response in eukaryotic cells to maintain cellular proteostasis. The protein has a protective role in preventing cells from undergoing ageing, and neurodegeneration, and also mediates tumorigenesis. Thus, modulating HSF-1 activity in humans has a promising therapeutic potential for treating these pathologies. Loss of HSF-1 function is usually associated with impaired stress tolerance. Contrary to this conventional knowledge, we show here that inactivation of HSF-1 in the nematode Caenorhabditis elegans results in increased thermotolerance at young adult stages, whereas HSF-1 deficiency in animals passing early adult stages indeed leads to decreased thermotolerance, as compared to wild-type. Furthermore, a gene expression analysis supports that in young adults, distinct cellular stress response and immunity-related signaling pathways become induced upon HSF-1 deficiency. We also demonstrate that increased tolerance to proteotoxic stress in HSF-1-depleted young worms requires the activity of the unfolded protein response of the endoplasmic reticulum and the SKN-1/Nrf2-mediated oxidative stress response pathway, as well as an innate immunity-related pathway, suggesting a mutual compensatory interaction between HSF-1 and these conserved stress response systems. A similar compensatory molecular network is likely to also operate in higher animal taxa, raising the possibility of an unexpected outcome when HSF-1 activity is manipulated in humans.

Keywords: C. Elegans; skn‐1; autophagy; cellular stress response; heat shock factor 1; heat shock proteins; heat shock response; hormesis; innate immunity; insulin‐like signaling pathway; intracellular pathogen response; proteostasis; thermotolerance; unfolded protein response.

FIGURE 1

Inactivation of HSF‐1 results in increased thermotolerance in young adult C. elegans. (a) Domain organization of full length (FL) and C‐terminally truncated (ΔTAD) CeHSF‐1 proteins. DBD: DNA binding domain, binds the consensus HSE sequence; HR‐A/B oligomerization domain: heptad repeat A (HR‐A) and heptad repeat B (HR‐B) leucine‐zipper domains mediate the trimerization of HSF1. HR‐C heptad repeat domain keeps HSF1 in a monomeric form, due to its interaction with the HR‐A/B domain. TAD: transactivation domain, responsible for transcriptional activation. (b) hsf‐1(sy441) [PS3551] mutant animals have an increased thermotolerance compared to wild‐type at the 1 day old adult stage (animals were shifted from 20°C to 35°C for 5 h). (c) The increase in thermotolerance can be observed in animals treated with a type of hsf‐1 RNAi and shifted from 20°C to 35°C for 5 h. (EV = empty vector, control for the RNAi treatment) (d) At the 1st day of adulthood hsf‐1(sy441) mutant animals tolerate heat stress better than wild‐type (animals were shifted from 20°C to 35°C for 5 h). (e) 2 days old adult hsf‐1(sy441) mutants still had an increased thermotolerance compared to wild‐type. Preconditioning (heat shock for 30 min. at 35°C, 18 h before the thermotolerance assay) increased the thermotolerance of wild‐type but had no effect on the hsf‐1(sy441) mutants. (f) By the 4th day of adulthood, the increased thermotolerance of hsf‐1(sy441) animals disappeared. At this stage the pre‐shock did not increase the survival of either wild‐type or hsf‐1(sy441) mutant animals. (g) Downregulating hsf‐1 by using RNA interference also increased the survival rate of 1‐day old C. elegans adults under heat‐shock conditions (animals were shifted from 20°C to 35°C for 6 h). (h) 2 day old adult animals treated with hsf‐1 RNAi had lower thermotolerance compared to control. In this case preconditioning has no effect on either pre‐shocked or naive animals. (i) At the 4 day old adult stage the control group is clearly more tolerant to heat stress than hsf‐1(RNAi) animals. Pre‐shock has no effect on the animals treated with hsf‐1 dsRNA, while it slightly lowers the survival rate of untreated C. elegans. (b–i) Thermotolerance following a 5 h‐long (6 h long in case of RNAi treatment) heat shock at 35°C. n > 3 replicates of 50 animals per strain. Individual data points represent independent trials (the different biological replicates are indicated by different shapes), lines represent means. Significance compared to wild type control was determined using Cochran–Mantel–Haenszel test; * = p < 0.05 ** = p < 0.01 *** = p < 0.001; error bars represent ± SEM. Source data underlying Figure 1b–i are provided in Table S1. EV = empty vector; hsf‐1 = heat shock factor 1; RNAi = RNA interference; SEM = standard error of the mean.

FIGURE 2

Genes involved in distinct stress and immunity related pathways are upregulated in hsf‐1(sy441) mutant animals. (a) RNA‐Seq volcano plot showing log₂‐fold change in expression levels of genes (FDR <0.05) differentially regulated (1116 up‐regulated; 72 down‐regulated) in 1 day old hsf‐1(sy441) mutant adults compared to wild‐type (N2) at 20°C. (b) GO enrichment analysis shows that genes involved in PERK‐mediated UPR and related to innate immune response are enriched among differentially expressed genes. Horizontal axis shows the fold enrichment. (c) Genes transcriptionally upregulated in hsf‐1(sy441) mutant and hsf‐1(RNAi) genetic backgrounds correlate with genes that are up‐ or downregulated in response to pathogens, abiotic stresses or conserved regulators of cellular stress response pathways. Analysis was performed using the GSEA 4.3.2 software package (see the Materials and Methods). Correlation was quantified as a Normalized Enrichment Score (NES). NES is positive (red) when a gene set show correlation with upregulated genes, while NES is negative (green) when a gene set show correlation with downregulated genes. Cells are white if no significant correlation was detected (FDR >0.25, or nominal p‐value >0.05) or NES values are not available (N/D). Gene sets used are available in Table S5. For more details, see Table S6. (d–f) Venn diagrams showing that genes upregulated in hsf‐1(sy441) mutants compared to wild‐type animals have a significant overlap with genes upregulated in pals‐22(jy3) mutants compared to wild‐type (RF = 3.7; p < 5.218e‐202) (d), with genes upregulated in skn‐1(RNAi) compared to EV control (RF = 4.2; p < 4.418e‐06) (E) or with genes induced upon ER stress that are dependent on IRE‐1, PEK‐1 or ATF‐6 signaling (RF = 2.5; p < 1.638e‐04) (f). (g) Diagram showing the relative levels of genes significantly upregulated in both, hsf‐1(sy441) mutant and skn‐1(RNAi) backgrounds. FC and FDR were calculated by edgeR (see Materials and Methods and Table S4). (h) Diagram showing the relative levels of genes significantly upregulated in both, hsf‐1(sy441) mutant and upon ER stress. FC and FDR was calculated by edgeR (see Materials and Methods and Table S5). Source data underlying Figure 2a,b are provided in Table S4. Source data underlying Figure 2c are provided in Table S7. Source data underlying Figure 2d–f are provided in Tables S5 and S8; FDR = false discovery rate; hsf‐1 = heat shock factor 1; RNAi = RNA interference; GO = Gene Ontology.

FIGURE 3

Heat shock genes are upregulated in hsf‐1(sy441) young adults upon robust heat shock. (a) Heat map showing the relative gene expression levels of differently regulated genes in wild type (N2) and hsf‐1(sy441) mutant 1‐day‐old adult nematodes following heat shock at 35°C for 30 min compared to the untreated control (b) GO enrichment analysis shows that upon heat‐shock genes related to stress responses are up‐regulated in 1 day old hsf‐1(sy441) mutant adults. Horizontal axis shows the fold enrichment. (c) Table showing heat shock protein coding genes upregulated in both, wild‐type, and hsf‐1(sy441) mutant animals. (d–g) Results of qPCR experiment showing that mRNA levels of heat shock protein coding genes hsp‐70a, hsp‐70b, hsp‐16.2 and hsp‐16.11 are induced in hsf‐1(sy441) mutants upon heat shock. The diagrams show the result of four replicates. p values were determined using Welch's t‐test * = p < 0.05 ** = p < 0.01 *** = p < 0.001; error bars represent + SEM (h) Representative fluorescent images showing expression of hsp‐16.2p::GFP reporter quantified in (i). Note that hsp‐16.2p::GFP is expressed exclusively in the intestinal cells in hsf‐1(sy441) mutants upon heat shock. (i) Plot showing that induction of hsp‐16.2p::GFP upon heat shock in hsf‐1(sy441) requires HSF‐1 activity. Three replicates of at least 55 animals per strain / trial were analyzed. Date points represent mean of independent trials, lines represent mean of means, p values were determined using Welch's t‐test * = p < 0.05 ** = p < 0.01 *** = p < 0.001; error bars represent ± SEM. Source data underlying Figure 4a–c are provided in Table S8. Source data underlying Figure 3d–i are provided in Table S9. EV = empty vector; FDR = false discovery rate; hsf‐1 = heat shock factor 1; RNAi = RNA interference; GO = Gene Ontology; SEM = standard error of the mean.

FIGURE 4

Induction of stress and immunity related genes in hsf‐1(sy441) mutant nematodes is age dependent. (a) Expression of several genes that are involved in innate immunity is overactivated in hsf‐1(sy441) mutants compared to the wild type following heat shock. (b) A set of these genes is overlaps with genes induced in pals‐22(jy3) mutants a negative regulator of intracellular pathogen response suggesting that HSF‐1 inhibits IPR related gene expression. (c) Genes that are upregulated in hsf‐1(sy441) mutants following heat shock include genes involved in innate immunity, UPR^ER and genes that are downregulated in a skn‐1(−) mutant background. (d–f) mRNA levels of the UPRER related genes lips‐11, calu‐1 and warf‐1 are higher in hsf‐1(sy441) mutants compared to the wild type. (g–h) The expression levels of genes involved in immunity (pals‐6 and lys‐7) are also increased in 1‐day old hsf‐1(sy441) mutants, but this difference is diminished by 4th day of adulthood. (i–l) Genes related to various stress responses such as proteasomal degradation (skr‐5), the IlS pathway sod‐3, asp‐8, and oxidative stress (osg‐1) showed a similar expression pattern during ageing in hsf‐1(sy441) mutants compared to the wild type. (d–l) All these data indicate that the induction of stress and immunity related genes in hsf‐1(sy441) mutant nematodes is age‐dependent. Bars represent mean of at least three trials. Significance was determined using Mann–Whitney test; * = p < 0.05 ** = p < 0.01 *** = p < 0.001; error bars represent + SEM. Source data underlying Figure 2g–o are provided in Table S10.

FIGURE 5

Inactivation of the UPR^ER and SKN‐1 suppresses the increased thermotolerance of hsf‐1 mutant animals. (a) The expression of the hsp‐4p::gfp marker is increased when hsf‐1 is silenced in 1 day old adult C. elegans (representative figure). (b) Quantified expression of hsp‐4p::gfp reporter in wild‐type and hsf‐1(RNAi) genetic backgrounds. Paired t‐test, * = p < 0.05 ** = p < 0.01 *** = p < 0.001; error bars represent ± SEM. (c) Inactivating two (IRE‐1‐ and PEK‐1‐mediated) of the three signaling arms of the UPR^ER suppressed the increased thermotolerance of the 1 day old adult hsf‐1(sy441) C. elegans (animals were shifted from 20°C to 35°C for 6 h). (d) Increased XBP‐1 activity significantly elevated heat stress tolerance of both wild‐type and hsf‐1(sy441) mutant animals (animals were shifted from 20°C to 35°C for 6 h). (e) The increased thermotolerance of the hsf‐1(sy441) animals is suppressed in the skn‐1(RNAi) background (animals were shifted from 20°C to 35°C for 6 h). Cochran–Mantel–Haenszel test; * = p < 0.05 ** = p < 0.01 *** = p < 0.001; Data points represent mean of independent trials (the different biological replicates are indicated by different shapes), lines represent mean of means, error bars represent ± SEM. Source data underlying Figure 5b–e are provided in Table S13.

FIGURE 6

Immunity related signaling contributes to the increased thermotolerance of hsf‐1 mutant animals. (a) Silencing pals‐22 increases heat tolerance to the same extent in wild type and hsf‐1(sy441) genetic backgrounds (animals were shifted from 20°C to 35°C for 6 h). (b) Silencing elt‐2 suppresses the increased thermotolerance of hsf‐1(sy441) mutants (animals were shifted from 20°C to 35°C for 6 h). Cochran–Mantel–Haenszel test; * = p < 0.05 ** = p < 0.01 *** = p < 0.001; Data points represent mean of independent trials (the different biological replicates are indicated by different shapes), lines represent mean of means, error bars represent ± SEM. (c) An ELT‐2 binding site related motif is common in genes upregulated in hsf‐1(sy441) mutant and hsf‐1(RNAi) background. Enrichment ratio: 2.94 compared to motif found in randomly generated sequences; FRD:3.17E‐08 (d) Our current model: HSF‐1 deficiency induces a mild proteotoxic stress, as a result, activity of other cellular stress response pathways – mainly the unfolded protein response (UPR^ER)and immunity related signaling – is elevated in young animals; paradoxically this hormetic effect contributes to a heat resistant phenotype of 1 day old HSF1 deficient C. elegans. Source data underlying Figure 6a,b are provided in Table S15.