Gene bookmarking by the heat shock transcription factor programs the insulin-like signaling pathway

Abstract

Maternal stress can have long-lasting epigenetic effects on offspring. To examine how epigenetic changes are triggered by stress, we examined the effects of activating the universal stress-responsive heat shock transcription factor HSF-1 in the germline of Caenorhabditis elegans. We show that, when activated in germ cells, HSF-1 recruits MET-2, the putative histone 3 lysine 9 (H3K9) methyltransferase responsible for repressive H3K9me2 (H3K9 dimethyl) marks in chromatin, and negatively bookmarks the insulin receptor daf-2 and other HSF-1 target genes. Increased H3K9me2 at these genes persists in adult progeny and shifts their stress response strategy away from inducible chaperone expression as a mechanism to survive stress and instead rely on decreased insulin/insulin growth factor (IGF-1)-like signaling (IIS). Depending on the duration of maternal heat stress exposure, this epigenetic memory is inherited by the next generation. Thus, paradoxically, HSF-1 recruits the germline machinery normally responsible for erasing transcriptional memory but, instead, establishes a heritable epigenetic memory of prior stress exposure.

Keywords: C. elegans; H3K9 methyltransferase; HSF1; MET-2; daf-2; epigenetic; heat shock; stress; transcriptional memory; transgenerational.

Figure 1:. HSF-1 germline activity alters the stress resilience and proteostasis mechanisms of progeny that develop from stressed germ cells.

(A) Experimental procedure (also see STAR Methods). (B) Stress resilience of F1 progeny of C. elegans mothers subjected to a short (5-minute) or long (30-minute or 60-minute) heat-shock (HS).. X-axis: the time interval at 20°C after the maternal heat-shock at which progeny were collected. Non-heat shocked control: NHS(Ctrl). Y axis: Percent F1 progeny surviving severe heat-stress. Legend: duration of maternal heat-shock. n=3-6 experiments/time point. Each experiment represents 3-20 P0 mothers/condition/time interval and an average of 24.3±2.3 F1 progeny/time interval/experiment for the short heat-shock experiments and 50.4±7.8 F1 progeny/time interval/experiment for long heat-shock experiments. (C) Stress resilience of F1 progeny of NHS(Ctrl), and 5-minute heat-shocked P0 mothers, following hsf-1 germline RNAi. Control (Ctrl) RNAi: L4440 empty vector. n=4-8 experiments, 27.4±0.8 and 23.4±2.7 F1 progeny/experiment on L4440 and hsf-1 RNAi respectively. (D) Stress resilience of F2 and F3 progeny of NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minute, 30-minute and 60-minute. Y axis: Percent NHS(Ctrl), F2 and F3 progeny that surive severe heat-stress. Legend indicates duration of maternal heat-shock. Each experiment represents 3-20 P0 mothers/condition and F2 and F3s collected from 5-40 F1 or F2 progeny/condition. (E) Number of polyglutamine aggregates in day-one to day-four adult F1 progeny of NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minute, 30-minute or 60-minute. n=3 experiments, 25 F1 progeny/experiment from 3-20 P0 mothers/condition. B-E: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; Unpaired Student’s t-test. See also Figures S1 and S2

Figure 2:. HSF-1 recruits MET-2 to HSF-1 target genes upon heat-shock and increases H3K9me2 at their 5’-UTR regions.

(A) H3K9me2 occupancy [relative to non-heat shocked control values [NHS(Ctrl)] of each respective strain; for data normalized to wild-type NHS(Ctrl), see Figure S4C] at the promoter proximal 5’-UTR regions of hsp-70(C12C8.1) I, F44E5.4/.5 II, hsp-16.11 V, hsp-16.2 V, hsp-16.41 V, unc-23 V, nurf-1 II, K10D3.6 I, C32H11.4 IV and pha-4 V in P0 wild-type and hsf-1 (sy441) I under NHS(Ctrl) conditions, immediately upon heat-shock (HS), and 2 hours after heat-shock (HS+rec). Heat-shock: 5-minutes. n=4-7 experiments. See STAR Methods for 5’-UTR regions assayed. (B) H3K9me2 occupancy [relative to respective RNAi NHS(Ctrl); also see Figure S4D] at 5’-UTR of hsp-70(C12C8.1) I, F44E5.4/.5 II and hsp-16.11 V. H3K9me2 occupancy was assessed in P0 animals subjected to control and met-2 germline RNAi, under NHS(Ctrl) conditions and at 2 hours after a 5-minute heat-shock (HS+rec). n=4-8 experiments. Control RNAi: L4440. (C) Representative Western blot from co-immunoprecipitation experiment using anti-HA antibody and lysates from NHS(Ctrl) and heat-shocked (30-minute), wild-type and hsf-1(sy441)I P0 animals expressing FLAG::HSF-1 and MET-2::HA. Left: antibodies used. 10% of Input is included. n=3 experiments. For antibody validation see Figure S1D and S4B. (D) Representative micrographs showing projections of confocal z-sections through nuclei of diakinesis oocytes in wild-type and hsf-1(sy441) I animals, in NHS(Ctrl) and upon heat-shock (30-minute; also see Figure S6). Arrows: HSF-1 and MET-2 colocalization in nSBs at condensed chromosomes (DAPI) upon heat-shock. Dotted lines: nuclei. * somatic gonad sheath cell where HSF-1 constitutively forms nSBs colocalized with MET-2. Panels from left to right: overlap of HSF-1 immunostaining (red) with DAPI (blue), MET-2 immunostaining alone (green), and overlap of HSF-1 (red) and MET-2 (green) immunostaining. A, B: Data: Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; ANOVA with Tukey’s correction. D: Scale bar:5μm. See also Figures S1, S3, S4, S5 and S6

Figure 3:. H3K9me2 levels remain elevated in adult F1 progeny subjected to in utero stress.

(A) Representative micrographs showing H3K9me2 staining in projections of confocal z-sections through the germline of P0 wild-type mothers under non-heat shocked control conditions [NHS(Ctrl)] and upon heat-shock. Heat-shock: 5-minute (5min HS) and 60-minute (60min HS). Insets: magnified images of diakinesis oocytes. (B) Percent H3K9me2 positive diakinesis oocytes per gonad in P0 NHS(Ctrl) and upon heat-shock. Heat-shock: 5-minute and 60-minute. Percent H3K9me2 positive oocytes present in the P0 germline 2 hours after the 5-minute heat-shock (5min HS+rec) are also shown (corresponds to data in Figure 2A). n=2-6 experiments; oocytes in 3-15 gonads quantified/experiment. (C) H3K9me2 occupancy [relative to NHS(Ctrl) from each respective condition] at the 5’-UTR regions of hsp-70(C12C8.1) I, F44E5.4/5 II, hsp-16.11 V, C32H11.4 IV and pha-4 V in day-one adult F1 progeny of NHS(Ctrl), and 5-minute heat-shocked, P0 mothers subjected to germline RNAi. RNAi: hsf-1 and control RNAi (L4440 empty vector). n=5 experiments. (D) Top: Representative Western blot of lysates from day-one-adult F1 progeny of NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minutes and 30-minutes. Antibodies: anti-H3K9me2 (above) and anti-Histone H3 (below). Lysate from met-2(n4256) III animals show specificity of antibody. Also see Figure S7A. Bottom: H3K9me2 levels quantified relative to total H3, and normalized to NHS(Ctrl). n=6 experiments. (E) Representative micrographs showing H3K9me2 staining in F1 embryos from NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minutes and 60-minutes. (F) Fluorescent intensity (arbitrary units) in H3K9me2-stained nuclei of 2-4 cell stage, and <100 cell stage F1 embryos from NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minutes and 60-minutes. n=2-4 experiments, 17-47 nuclei/condition for 2-4 cell stage embryos, and 103-105 nuclei/condition for <100 cell stage embryo. B-D, F: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; ANOVA with Tukey’s correction, and Unpaired Student’s t-test. A: Scale bar:30μm. E: Scale bar: 15μm. See also Figure S7

Figure 4:. Germline MET-2 is required for enhanced stress resilience in progeny.

(A) Stress resilience of F1 progeny from control non-heat shocked [NHS(Ctrl)] and 5-minute heat-shocked P0 mothers subject to germline RNAi. RNAi indicated on the X-axis. Control RNAi: L4440 empty vector. n=5-14 experiments, 22.7±2.7 to 30.0±2.8 F1/ experiment. (B) Stress resilience of F1 progeny of wild-type and met-2 (n4256) III animals. F1 progeny are from NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minutes and 60-minutes. n=3-9 experiments. (C) Number of polyglutamine aggregates scored in day-one to day-four adult F1 progeny of NHS(Ctrl) and 5-minute heat-shocked P0 mothers. P0 mothers were subjected to met-2 and control (L4440 empty vector) RNAi. n=3 experiments, 15 F1 animals/experiment obtained from 3-20 P0 mothers/condition/experiment. A-C: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; Unpaired Student’s t-test

Figure 5:. DAF-16 activity in the F1 progeny of heat-shocked mothers is required for their enhanced stress resilience.

(A) Top: Experimental protocol to test the dependency of stress resilience in F1 progeny. Note:F1 progeny and not the P0 maternal germline were subjected to daf-16 or hsf-1 RNAi. Bottom: Stress resilience of F1 progeny from control non-heat shocked [NHS(Ctrl)], and heat-shocked, P0 mothers. Heat-shock: 5-minutes and 60-minutes. Control RNAi: L4440. n=3-10 experiments, 19.3±4.3 to 28.4±4.8 F1 progeny/experiment/condition. (B) Stress resilience of F1 progeny from NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minute and 60-minute. P0 mothers were subjected to daf-16 germline RNAi. n=5-8 experiments, 24.4±2.7 to 35.2±2.0 F1 progeny/experiment/condition. (C) Schematic of daf-2 gene regions analyzed for HSF-1 binding (5’ Far; −9071 to −8954), and H3K9me2 occupancy: a (−172 to −70), b (Intron 1: +3179 to +3279), c (Intron 2: +8046 to +8177) and d (Exon 9: +15719 to +15834). (D) HSF-1 occupancy at the 5’-far region of daf-2 gene in NHS(Ctrl)] and heat-shocked wild-type and hsf-1 (sy441) P0 mothers. Heat-shock: 5-minute and 60-minute. n=3 experiments. Data normalized to NHS(Ctrl) wild-type and hsf-1(sy441) values respectively (also see Figure S7D). (E) H3K9me2 occupancy at regions a, b, c, and d, depicted in C, in wild-type and hsf-1 (sy441) P0 mothers. Conditions: NHS(Ctrl), 2 hours after a 5-minute heat-shock (5min HS+rec), and immediately after a 60-minute heat-shock. n=3-4 experiments. H3K9me2 occupancy shown relative to NHS(Ctrl) wild-type and NHS(Ctrl) hsf-1 (sy441) values respectively. (also see Figure S7F). A, B, D, E: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; ANOVA with Tukey’s correction, and Unpaired Student’s t-test. See also Figure S7

Figure 6:. HSF-1 and MET-2 in the maternal germline heritably silence the daf-2 gene and activate DAF-16 in progeny.

(A) Schematic of daf-2 gene regions assayed for H3K9me2 occupancy (as in Figure 5C). (B) H3K9me2 occupancy at regions a, b, c and d, depicted in A, in day-one adult F1 progeny from non-heat shocked control [NHS(Ctrl)], and heat-shocked P0 mothers. Heat-shock: 5-minutes and 30-minutes. n=5-7 experiments. H3K9me2 occupancy shown relative to NHS(Ctrl) wild-type values. (C) H3K9me2 occupancy at regions a , b, c and d in day-one adult F1 progeny from NHS(Ctrl), and heat-shocked P0 mothers subjected to hsf-1 or met-2 germline RNAi. Heat-shock: 5-minute. n=5-7 experiments. H3K9me2 occupancy shown relative to NHS(Ctrl) from each respective treatment. Control RNAi : L4440. (D) Average daf-2 mRNA levels in F1 progeny of NHS(Ctrl), and heat-shocked P0 mothers. Heat-shock: 5-minutes and 60-minutes. mRNA levels are relative to pmp-3 and normalized to that in F1 progeny from NHS(Ctrl) P0 mothers. n=5-6 experiments. (E) Representative micrographs showing DAF-16::GFP localization in day-one adult F1 progeny from NHS(Ctrl), and heat-shocked P0 mothers. Heat-shock: 5-minute and 60-minute. (F) Percent F1 progeny that display DAF-16::GFP nuclear localization as day one adults. n=3-4 experiments, 20-35 progeny/experiment. (G) Average sod-3, mtl-1 and lys-1 mRNA levels in day one adult F1 progeny from NHS(Ctrl) and 60-minute heat-shocked P0 mothers. mRNA levels are relative to pmp-3, and normalized to that in F1 progeny from NHS(Ctrl) P0 mothers. n=3-5 experiments, 30 F1s from 15 mothers/experiment. B-D, F, G: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; ANOVA with Tukey’s correction, and Unpaired Student’s t-test. E: Scale bar:30μm. See also Figure S7

Figure 7:. Maternal stress-induced increase in H3K9me2 at daf-2 in reversed over generations.

(A) Top: Number of eggs laid every hour within the time interval on the X-axis, by non-heat shocked control [NHS(Ctrl)], and heat-shocked P0 mothers. Bottom: Percent hatched embryos from NHS(Ctrl) and heat-shocked P0 mothers. Heat-shock: 5-minute and 60-minute. (B) Representative micrographs showing germline and uterus of F1 progeny from NHS(Ctrl), and heat-shocked P0 mothers. Heat-shock: 5-minute and 60-minute. Note all F1 from NHS(Ctrl) and 5min HS P0 mothers, and some F1 from 60min HS P0 mothers are fertile (marked by dotted lines), but 52.5±8.4% F1 from 60min HS P0 mothers are sterile (right). n=3 experiments, 17-101 F1s scored/experiment (C) H3K9me2 occupancy at regions a , b, c and d, of the daf-2 locus (top), in day one adult F2 progeny from NHS(Ctrl), and heat-shocked P0 mothers. Heat-shock: 5-minute and 60-minute. n=2-3 experiments. Compare to H3K9me2 levels in Figure 6B. (D) Model: germline activity of HSF-1 induces a MET-2-dependent increase in H3K9me2 at the daf-2 gene and consequently DAF-16 is activated to confer thermotolerance; this repression of daf-2 is inherited by subsequent generations depending on the severtity of maternal stress. A, C: Data show Mean ± Standard Error of the Mean. *: p<0.05, **: p<0.01, ***: p<0.001, ns: non-significant; ANOVA with Tukey’s correction, and Unpaired Student’s t-test. B: Scale bar:50μm. See also Figure S7