Abstract Transient events during development can exert long-lasting effects on organismal lifespan. Here we demonstrate that exposure of Caenorhabditis elegans to reactive oxygen species during development protects against amyloid-induced proteotoxicity later in life. We show that this protection is initiated by the inactivation of the redox-sensitive H3K4me3-depositing COMPASS complex and conferred by a substantial increase in the heat-shock-independent activity of heat shock factor 1 (HSF-1), a longevity factor known to act predominantly during C. elegans development. We show that depletion of HSF-1 leads to marked rearrangements of the organismal lipid landscape and a significant decrease in mitochondrial β-oxidation and that both lipid and metabolic changes contribute to the protective effects of HSF-1 against amyloid toxicity. Together, these findings link developmental changes in the histone landscape, HSF-1 activity and lipid metabolism to protection against age-associated amyloid toxicities later in life. __________________________________________________________________ Many biological processes contain seemingly stochastic components that cannot be explained by genetic or environmental variations^[47]1. This noise in the system becomes particularly obvious in synchronized populations of isogenic organisms, such as Caenorhabditis elegans, which show marked inter-individual differences in their lifespans as well as their susceptibility toward amyloid-mediated proteotoxicity^[48]1,[49]2. Recent studies in C. elegans strains equipped with chromosomally encoded ratiometric in vivo redox sensors revealed that naturally occurring variations in the levels of reactive oxygen species (ROS) during early development are responsible for some of the observed lifespan variations^[50]3,[51]4. Analysis of worms sorted according to their early-life redox states showed that the most oxidized animals in the L2 larval stage are substantially more stress resistant and develop into more reduced and longer-lived animals than their oxidized counterparts^[52]3,[53]4. These results suggest that naturally occurring early-in-life events exist that can considerably improve health and lifespan. Similar beneficial early-in-life events were also recently reported for mice, where food reduction in pre-weaning mice or rapamycin treatment for the first 4 weeks after birth were found to be sufficient to extend lifespan up to 15%^[54]5,[55]6. Mechanistic follow-up studies in C. elegans revealed that the highly conserved methyltransferase SET-2, the central component of the H3K4me3-depositing COMPASS complex^[56]7, is redox sensitive, and its reversible inactivation is responsible for a substantial reduction in global H3K4me3 levels in the oxidized subpopulation of C. elegans^[57]4. Genetic downregulation of members of the COMPASS complex, including SET-2 and ASH-2, increased heat shock and oxidative stress resistance and substantially extended lifespan, essentially phenocopying the oxidized subpopulation^[58]4,[59]8. Earlier studies in C. elegans, which showed that H3K4me3 is deposited on transcriptional start sites before the L3 larval stage^[60]9, helped to explain how a transient ROS-mediated event during development can exert these long-lasting effects. Age is the greatest non-genetic risk factor for neurodegenerative diseases, such as Alzheimer’s disease or Huntington’s disease^[61]10. These diseases manifest through the deposition of amyloidogenic proteins, including the Aβ[1–42] peptide, Tau, or polyglutamine (polyQ) expansion proteins, as insoluble fibrils and plaques^[62]10,[63]11. Lifespan-extending interventions were recently shown to protect against amyloid-associated proteotoxicity in a variety of model organisms^[64]12–[65]14, suggesting that aging pathways may be suitable targets for preventing or treating age-associated neurodegenerative diseases. In the present study, we explored if and how early-in-life events that extend lifespan protect organisms against amyloid-associated proteotoxicity later in life. We discovered that the transient exposure of C. elegans to ROS in early life significantly delays the onset of amyloid toxicity in a COMPASS-complex-dependent manner, and we demonstrated that the genetic or pharmacological depletion of members of the COMPASS complex (that is, ash-2 and set-2) protects a range of different C. elegans strains against Aβ[1–42]-mediated and polyQ-mediated toxicities. Mechanistic follow-up studies revealed that the depletion of members of the COMPASS complex causes a substantial functional upregulation of heat shock factor 1 (HSF-1), a conserved longevity factor previously shown to act predominantly during early life stages of C. elegans^[66]13,[67]15. Unexpectedly, we found that HSF-1ʼs protective effect is largely unrelated to its known role in regulating the expression of members of the cellular proteostasis network. Instead, we discovered that increasing the activity of HSF-1 in early life substantially alters several aspects of C. elegansʼ lipid homeostasis, previously associated with lifespan extension^[68]16. These changes include an increase of short-chain fatty acids, an increase in mono-unsaturated fatty acids (MUFAs) and an upregulation of mitochondrial β-oxidation. Our studies thus reveal a previously unknown connection between early-in-life changes in the histone landscape, heat-shock-independent HSF-1 activity and altered lipid metabolism and the protection against age-associated amyloid toxicities later in life. Results Early-in-life ROS protects against amyloid toxicity Our previous work in C. elegans demonstrated that transiently elevated levels of ROS during early development, either naturally produced or exogenously applied in the form of paraquat (PQ), extend lifespan through the oxidative inactivation of the COMPASS complex^[69]4. To test the intriguing idea that early-life oxidative stress might protect organisms against amyloid-induced toxicity later in life, we exposed a C. elegans mutant strain, which expresses the human Aβ[1–42] peptide in its body wall muscle cells (that is, strain CL4176), to a transient 10-h PQ treatment during early development, followed by a shift in temperature to induce Aβ[1–42] expression ([70]Fig. 1a). As shown previously, these worms become paralyzed as the levels of amyloids and amyloidogenic aggregates increase ([71]Fig. 1b, black trace)^[72]17. Short-term exposure to the ROS-generating compound PQ, however, significantly delayed this onset of paralysis ([73]Fig. 1b, black dashed trace), providing evidence that transient, early-in-life events might be able to protect organisms against proteotoxicity elicited by amyloidogenic proteins at later stage in life. To test whether this protective effect is mediated by the oxidative downregulation of the COMPASS complex, we crossed the CL4176 strain with a C. elegans mutant strain lacking set-2, the redox-sensitive component of the COMPASS complex^[74]4. Movement analysis revealed that the set-2 deletion strain showed a delay in the onset of paralysis that was similar to PQ-treated CL4176 and no longer responsive to PQ treatment ([75]Fig. 1b, compare green solid and dashed traces). These results suggested that the protective effects of the transient exposure to ROS in early life are likely mediated by the oxidative stress-induced inactivation of members of the COMPASS complex. Fig. 1 |. Disruption of H3K4me3 modifiers protects C. elegans against amyloid toxicity. Fig. 1 | [76]Open in a new tab a, Schematic for PQ treatment of Aβ[1–42]-expressing strain CL4176. b, Paralysis of wild-type or set-2 (ok952) CL4176 animals treated as shown in a. c, CL4176 animals were maintained from egg either on control NGM plates or on plates containing the histone methyltransferase inhibitor MM-401 (1 mM). Paralysis was assessed 40 h after temperature upshift. d, Paralysis of CL4176 C. elegans maintained on bacteria expressing control (L4440), ash-2 or set-2 RNAi. e, Paralysis of Aβ[1–42]-expressing CL2006 maintained on control (L4440), ash-2 or set-2 RNAi. f, Survival of GRU102 animals (neuronal Aβ[1–42]) maintained on control (L4440), ash-2 or set-2 RNAi. g, Paralysis of Q40::YFP (AM141) C. elegans maintained on control (L4440), ash-2 or set-2 RNAi. d,e,g (insets), H3K4me3 levels assessed by western blot analysis using day 1 adults. Total H3 was used as control. Results are representative (b,d,e–g) or the average ± s.e.m. (c, insets in d,e,g) of two (f), three (b–e) or four (g) independent experiments. Statistically significant differences in c were determined by unpaired t-test (one-tailed); *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. Detailed information on the number of worms per condition and statistical analysis for biological replicates of paralysis curves is shown in [77]Supplementary Table 1. NS, not significant. H3K4me3 modifier depletion protects against amyloid toxicity Recent studies in mouse models of Alzheimer’s disease revealed that the pharmacological inhibition of the mammalian COMPASS complex substantially improved memory deficits in mouse models of Alzheimer’s disease^[78]18. In the present study, we found that the addition of the SET-2 inhibitor MM-401 (ref. [79]19) protects Aβ[1–42]-expressing CL4176 worms against premature paralysis ([80]Fig. 1c). The observed delay in paralysis was similar to C. elegans strains either transiently treated with PQ during development ([81]Fig. 1b) or grown on RNA interference (RNAi) bacteria targeting ash-2 or set-2 ([82]Fig. 1d). These results support our conclusion that inactivation of the COMPASS complex, through oxidative stress, genetic depletion or pharmacological intervention, protects organisms against amyloid-mediated toxicity. To investigate whether a reduction in the functional activity of the COMPASS complex protects C. elegans more generally against amyloid-induced proteotoxicity, we tested a number of additional, previously established C. elegans models for amyloid toxicity: CL2006, which chronically expresses the human Aβ[1–42] peptide in the body wall muscle and experiences toxicity and displays premature paralysis later in life^[83]20; GRU102, which expresses the human Aβ[1–42] peptide pan-neuronally and dies prematurely^[84]21; and AM141, a strain that expresses a highly amyloidogenic 40-residue glutamine tract (Q40) fused to yellow fluorescent protein (YFP)^[85]22. This strain accumulates visible foci at the onset of adulthood and displays signs of paralysis about 3–5 d later. We prepared synchronized populations for each of these strains and maintained them on the control bacteria RNAi strain L4440 or on RNAi strains that target either ash-2 or set-2, which reproducibly reduce global H3K4me3 levels by about 50% (inserts in [86]Fig. 1d,[87]e,[88]g). Paralysis and lifespan analysis revealed that the animals deficient in either ash-2 or set-2 are significantly protected from proteotoxicity compared to the isogenic strains grown on control RNAi, independent of the timing of expression—that is, acute versus chronic Aβ[1–42] toxicity ([89]Fig. 1d,[90]e); the tissues in which the amyloidogenic proteins are expressed—that is, muscle versus neurons ([91]Fig. 1e,[92]f); or the types of amyloidogenic proteins expressed—that is, Aβ[1–42] versus Q40::YFP ([93]Fig. 1e,[94]g and [95]Supplementary Table 1). The protective effects of the ash-2 depletion were overall more robust and reproducible than the effects of the set-2 depletion, an observation that was in line with their respective effects on C. elegans lifespan^[96]8. It is noteworthy that the knockdown of members of the COMPASS complex altered neither the ratio of monomeric to higher oligomeric Aβ[1–42] species nor the steadystate levels of either amyloidogenic protein ([97]Extended Data Fig. 1a–[98]d), indicating that the protective effects associated with the reduction in COMPASS complex activity are not linked to the expression levels of amyloidogenic proteins. In fact, analysis of the number of visible foci in 1-day-old Q40::YFP-expressing adults revealed an increase in aggregate formation in worms either maintained on ash-2 or set-2 RNAi ([99]Extended Data Fig. 1e,[100]f) or crossed with a set-2 (ok952) deletion mutant ([101]Extended Data Fig. 1g,[102]h) relative to the control animals. Although the latter results are initially counterintuitive, they are consistent with several more recent studies, which showed that not the quantity but the quality and composition of protein aggregates correlate with their in vivo toxicity^[103]23. Based on these data, we, thus, concluded that the reduction in the functional activity of the COMPASS complex provides C. elegans with a widespread protection against amyloid-related toxicities. Protection by H3K4me3 modifier depletion requires HSF-1 Our previous studies revealed that the oxidative inactivation of H3K4me3 modifiers during development strongly augments heat-stress-induced gene expression in wild-type C. elegans^[104]4. These results suggested that the depletion of the COMPASS complex might increase the activity of HSF-1, the master regulator of the eukaryotic proteostasis network^[105]24. We confirmed these results by monitoring the induction of the hsp-16.2p::GFP reporter before and after heat shock treatment ([106]Extended Data Fig. 2a,[107]b), which showed significantly higher GFP expression in worms depleted for either ash-2 or set-2 compared to worms grown on control RNAi. This result was intriguing because presence of HSF-1 was not only reported to be crucial for lifespan regulation and protection against amyloid-mediated toxicity in C. elegans^[108]15,[109]24 but was also found to be most relevant for these processes when present during development^[110]13,[111]25. These results made us now wonder whether the protective effect of knocking down members of the COMPASS complex works directly through increasing HSF-1 activity. To test this idea, we depleted hsf-1 in Q40::YFP animals, which were maintained on either control or ash-2 RNAi bacteria, and monitored their onset of paralysis. Consistent with previous results^[112]13,[113]15, knockdown of hsf-1 sensitizes the Q40::YFP animals to amyloid toxicity and leads to significantly accelerated paralysis ([114]Fig. 2a, black dashed trace), whereas depletion of ash-2 delays the onset of paralysis ([115]Fig. 2a, red solid trace). In combination, however, knockdown of ash-2 failed to elicit any protective effect on hsf-1-depleted worms ([116]Fig. 2a, red dashed trace) despite similar knockdown efficiency in the strains ([117]Extended Data Fig. 2c). Worms grown on both hsf-1 and set-2 RNAi were even slightly more susceptible to Q40::YFP-mediated paralysis than hsf-1-deficient animals alone ([118]Extended Data Fig. 2d). Based on these results, we concluded that the mechanism by which depletion of H3K4me3 modifiers confers resistance to proteotoxicity requires the presence of functional HSF-1. Fig. 2 |. Protective effect of H3K4me3 depletion is due to HSF-1-mediated FAT-7 upregulation. Fig. 2 | [119]Open in a new tab a, Paralysis of Q40::YFP C. elegans maintained on control (L4440) or ash-2 RNAi in the presence or absence of hsf-1 RNAi. Simultaneous knockdown of ash-2 with hsf-1 was achieved by mixing equal amounts of each RNAi bacteria (knockdown efficiency shown in [120]Extended Data Fig. 2c). b, Accumulation of fat-7 mRNA in Q40::YFP animals maintained on control (L4440), ash-2 or set-2 RNAi in the presence or absence of hsf-1 RNAi was assessed by qPCR in day 1 adults and normalized to the housekeeping gene pmp-3. c,d, Fluorescence of fat-7p::fat-7::GFP-expressing C. elegans maintained on control (L4440), ash-2, set-2, hsf-1, ash-2+hsf-1 or set-2+hsf-1 RNAi was measured in day 1 adults (c) and quantified as arbitrary fluorescence units (a.f.u.) (d). At least 20 animals were assessed per condition. Scale bars in c, 0.1 mm. e, Paralysis of Q40::YFP animals maintained on control (L4440), ash-2, set-2, fat-7, ash-2+fat-7 or set-2+fat-7 RNAi. f, Quantification of fluorescence of adult fat-7p::fat-7::GFP-expressing C. elegans maintained on OP50 bacteria and exposed to PQ (1 mM) for 10 h during the L2 larval stage. FAT-7::GFP fluorescence was imaged in day 1 adults. g, Q40::YFP animals were cultivated on control (L4440) or fat-7 RNAi starting at the time of egg lay. Worms were then either maintained on fat-7 RNAi (dev. + adult) or shifted from L4440 to fat-7 RNAi 72 h after egg lay (adult). Inset in g shows mRNA expression of fat-7 in animals collected at 96 h after egg, showing a similar decrease in fat-7 expression between conditions. Results are representative (a,c–g), average ± s.e.m. (b, inset in g) of two (f), three (b–e,g) or four (a) independent experiments. Statistical significance was assessed by log-rank (a,e,g), one-way ANOVA (b, inset in g), Kruskal–Wallis test (d) or two-tailed Mann–Whitney test (f); *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. Detailed information and statistics for paralysis assays can be found in [121]Supplementary Table 1. adult, adulthood; dev., development; NS, not significant. Effects of HSF-1 are mediated by enhanced FAT-7 expression Neither analysis of the basal hsp-16-2p::GFP expression levels in non-amyloid-expressing C. elegans ([122]Extended Data Fig. 2a,[123]b) nor of HSP-16.2::mCherry fluorescence or mRNA levels of select HSF-1-controlled heat shock genes in Q40::YFP worms ([124]Extended Data Fig. 2e,[125]f) revealed any significant differences between control and ash-2-depleted worms and only modest differences, if any, upon depletion of set-2. These results suggested that the observed increase in HSF-1 activity does not cause a constitutive activation of the heat shock response. This conclusion was fully consistent with global gene expression studies previously conducted in ash-2-depleted or set-2-depleted non-amyloid-expressing C. elegans strains^[126]4,[127]8,[128]9 or performed in ash-2-depleted Q40::YFP-expressing worms for this study ([129]Supplementary Table 3). Neither dataset showed any global enrichment of stress-induced HSF-1 targets upon ash-2 and/or set-2 depletion. These results strongly suggested that the protective effect of HSF-1 is unlikely due to the constitutive upregulation of members of the proteostasis network but more likely due to changes in the expression levels of HSF-1-controlled non-heat shock genes. In fact, a recently conducted RNA sequencing (RNA-seq) analysis in hsf-1-depleted C. elegans revealed that HSF-1 controls the expression of more than 1,200 genes in a heat-shock-independent manner, many of which specifically during C. elegans development^[130]26. Considering these findings, we, therefore, explored which HSF-1-controlled non-heat shock genes might be involved in the protective function of ash-2-depleted or set-2-depleted C. elegans. To our surprise, we noted stearoyl-CoA desaturase FAT-7, an enzyme that converts stearic acid into the MUFA oleic acid^[131]27,[132]28, among the 15 most upregulated genes by HSF-1 in a heat-shock-independent manner^[133]26. Previous studies demonstrated that ash-2-depleted and set-2-depleted C. elegans strains show increased levels of fat-7 mRNA, which seemed responsible for the lifespan extension observed in H3K4me3-depleted worms^[134]16. Our results now suggested that increased activity of HSF-1 in ash-2-depleted or set-2-depleted worms might constitute the missing link between a decrease in global H3K4me3 levels and an increase in fat-7 mRNA levels. To test this idea, we determined fat-7 mRNA expression levels in ash-2-depleted or set-2-depleted Q40::YFP C. elegans using RT–PCR ([135]Fig. 2b) and monitored GFP fluorescence in an ash-2-depleted or set-2-depleted fat-7p::fat-7::GFP reporter strain ([136]Fig. 2c,[137]d). Both types of experiments not only confirmed the substantial increase in fat-7 expression upon depletion of members of the COMPASS complex but also demonstrated that the fat-7 expression levels are tightly controlled by HSF-1. Knockdown of hsf-1 by RNAi significantly decreased the fat-7 expression levels in L4440 control, ash-2 and set-2 RNAi strains, essentially eliminating any increase in fat-7 levels elicited by either ash-2 or set-2 depletion ([138]Fig. 2b–[139]d). These results strongly suggested that the increase in FAT-7 observed upon ash-2 or set-2 depletion is indeed the result of an increase in heat-shock-independent HSF-1 activity. To test whether this increase in the levels of FAT-7 contributes to the protection of H3K4me3-depleted worms against amyloid toxicity, we maintained Q40::YFP-expressing worms on control, ash-2 or set-2 RNAi in the absence or presence of fat-7 RNAi and examined their paralysis. Indeed, we found that the knockdown of fat-7 significantly accelerates paralysis in Q40::YFP worms, and depletion of ash-2 or set-2 no longer conferred any protection ([140]Fig. 2e). Based on these results, we concluded that the protective effect elicited by the depletion of H3K4me3 modifiers involves the upregulation of HSF-1 activity, which leads to a substantial increase in protective FAT-7 levels. Early-in-life FAT-7 protects against proteotoxicity Earlier studies revealed that, whereas depletion of hsf-1 during the entire lifespan of C. elegans (that is, development and adulthood) causes significantly shortened lifespan and accelerated paralysis, depletion of hsf-1 during adulthood alone had only mild effects^[141]13,[142]25. We now reasoned that, if our model was indeed correct and HSF-1-mediated upregulation of FAT-7 was responsible for our observed phenotypes, we should expect to find (1) an increase in fat-7 levels in worms transiently exposed to ROS during development and (2) a primarily developmental requirement of fat-7 expression for the protective effect later in life. To address the first prediction, we exposed C. elegans expressing the FAT-7::GFP reporter protein to the previously established 10-h bolus of PQ at the L2 larval stage and monitored FAT-7::GFP expression ([143]Fig. 2f). To address the second aspect, we maintained Q40::YFP worms on fat-7 RNAi either during their entire life or during adulthood only and monitored their onset of paralysis ([144]Fig. 2g). The results of both studies fully supported our model. We found that exposure of worms to early-in-life oxidative stress triggered a significant increase in FAT-7 levels in adult worms ([145]Fig. 2f). Moreover, depletion of fat-7 during adulthood alone had no major effects, whereas depletion of fat-7 during both development and adulthood significantly increased their sensitivity toward Q40-mediated proteotoxicity ([146]Fig. 2g). Although these results do not explain how FAT-7 expression in early life protects against proteotoxicity later in life, they do represent a further step toward finding the long-sought-after mechanistic explanation for the temporal requirements of HSF-1 expression during C. elegans development^[147]13,[148]25. Oleic acid protection against proteotoxicity requires HSF-1 Recent studies revealed that exogenous addition of the FAT-7 product oleic acid to the growth media extends worm lifespan and protects against amyloid-induced proteotoxicity^[149]16,[150]29,[151]30. Indeed, comparison between the onset of paralysis in Q40::YFP worms grown on ash-2 RNAi bacteria or on media supplemented with oleic acid revealed a striking resemblance, with both interventions causing a similar delay in the onset of paralysis resemblance ([152]Fig. 3a, black dashed and red solid traces). No further delay in paralysis was observed when we grew ash-2 RNAi worms on media supplemented with oleic acid ([153]Fig. 3a, red trace), suggesting that the depletion of ash-2 leads to the production of optimally protective amounts of oleic acid. It is noteworthy that we made similar observations in additional models of amyloid toxicity, including GRU102 worms expressing neuronal Aβ ([154]Fig. 3b). Consistent with our model, we found that neutral lipid levels, as determined by Oil Red O (ORO) staining, significantly increased upon ash-2 depletion, reaching levels similar to control RNAi worms cultivated in the presence of oleic acid ([155]Fig. 3c,[156]d, compare gray and dark pink symbols). However, contrary to our model, we found that oleic acid supplementation did not rescue the premature paralysis phenotype of the hsf-1-deficient animals in either Q40:YFP-expressing or Aβ-expressing worms ([157]Fig. 3a,[158]b). These results were entirely unexpected given that oleic acid supplementation should be sufficient to compensate for the decreased fat-7 expression in hsf-1-depleted worms. We obtained similar results in CB1402 worms, which express a temperature-sensitive variant of paramyosin in their body wall muscle cells and serve as a model to assess general proteotoxicity^[159]31 ([160]Extended Data Fig. 3a). These results could not be explained by issues in the uptake of oleic acid; in fact, we noted that hsf-1-depleted worms have an even higher ORO signal compared to wild-type or ash-2-depleted worms when grown on oleic acid ([161]Fig. 3c,[162]d). Moreover, we observed that, when cultured on oleic acid–supplemented plates, hsf-1-depleted Q40::YFP worms no longer display the developmental growth defect that we found associated with decreased hsf-1 levels ([163]Fig. 3e,[164]f). These results indicated that hsf-1-depleted worms are able to take up and utilize supplemented oleic acid to overcome their developmental defects. However, they seem unable to utilize the fatty acid in a way that provides protection against proteotoxicity later in life. We determined that the ability of free fatty acids to protect against proteotoxicity is selective toward oleic acid, as supplementation of the growth media with linoleic acid—the first derivative of oleic acid, for instance—did not protect worms against Q40::YFP toxicity ([165]Extended Data Fig. 3b) and excluded the possibility that oleic acid exerts its beneficial effects through activating the heat shock response ([166]Extended Data Fig. 3c–[167]f). These findings indicated that HSF-1 either directly or indirectly controls one or more additional steps downstream of oleic acid production, which are necessary to protect C. elegans against amyloid toxicity later in life. Fig. 3 |. Oleic acid protects against proteotoxicity in an HSF-1-dependent manner. Fig. 3 | [168]Open in a new tab a,b, Q40::YFP (a) or GRU102 (neuronal Aβ) (b) animals were synchronized by timed egg lay on L4440, ash-2, hsf-1 or ash-2+hsf-1 seeded NGM plates with or without supplementation with oleic acid (0.8 mM), and paralysis (a) or survival (b) was examined. c,d, ORO staining (c) and quantification (d) of fat content under conditions in a. e,f, Representative images of day 1 adult Q40::YFP animals maintained on L4440 or hsf-1 RNAi ± oleic acid (a) and quantification of animal size (f). A minimum of 30 animals were assessed per condition. Scale bars, 0.1 mm. Results are representative (a–f) of two (a,b) or three (c–f) independent experiments. Error bars represent s.e.m. Statistical significance was assessed by log-rank (a,b) or Kruskal–Wallis (d,f) test; *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. Additional experimental information and data for biological replicates not shown can be found in [169]Supplementary Table 1. NS, not significant. HSF-1 regulates oleic acid utilization Our data strongly suggest that the protective effect of downregulating members of the COMPASS complex during development is caused by HSF-1-mediated changes in lipid composition and/or utilization. To assess how depletion of ash-2 affects the lipid distribution of Q40::YFP-expressing C. elegans and to determine the role that HSF-1 plays in this process, we cultivated worms on L4440 control, ash-2, hsf-1 or ash-2+hsf-1 RNAi in the absence or presence of oleic acid and conducted lipidomic analysis at day 1 of adulthood. We analyzed the distribution of over 700 different lipid species using non-targeted liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis ([170]Fig. 4 and [171]Supplementary Table 2). As illustrated in the heat map, we found that the most profound changes upon depletion of ash-2 involved alterations in triacylglycerides (TGs), the main energy reservoirs in animals, with many TG species increasing in their relative abundance compared to worms grown on control RNAi ([172]Fig. 4a). Unexpectedly, most of these changes were lost upon the additional depletion of hsf-1. In fact, knockdown of hsf-1 caused a major general shift in the lipid landscape, with lipids abundantly present in control or ash-2-depleted Q40::YFP strains being substantially less abundant in worms grown on hsf-1 RNAi and vice versa ([173]Fig. 4a). Principal component analysis agreed with these conclusions and showed that the significant differences in lipid composition between ash-2-depleted and control RNAi worms are primarily mediated by HSF-1 ([174]Fig. 4b); only five lipids were found to be significantly different in abundance when we compared the lipid composition of ash-2+hsf-1-depleted Q40::YFP worms with that of hsf-1-depleted worms, whereas almost 18% of total lipids were different when we compared them with ash-2-depleted worms ([175]Fig. 4c). Closer analysis of the compositional changes in TGs associated with the protective ash-2 depletion revealed an enrichment for short and mono-unsaturated fatty acids (C < 16) over longer and polyunsaturated longer fatty acids (PUFAs) in TGs ([176]Fig. 4d,[177]e and [178]Extended Data Fig. 4a). Depletion of hsf-1, on the other hand, had precisely the opposite effect: a pronounced increase in the ratio of PUFAs over saturated fatty acids and MUFAs stored in TGs and the accumulation of very-long-chain fatty acids ([179]Fig. 4d,[180]e). These results were fully consistent with lipid composition changes in other long-lived mutants of C. elegans^[181]32, providing further evidence for the emerging model that changes in TG composition might be a central unifying feature of mutations that lead to extended health and lifespan^[182]33. Yet our biggest surprise came when we realized that C18:1 species (that is, oleic acid and cis-vaccenic acid) did not accumulate upon ash-2 depletion but were actually even lower than the C18:1 amounts measured in hsf-1 depleted worms, which showed C18:1 levels similar to wild-type or ash-2-depleted worms grown on oleic acid supplemented plates for several days ([183]Fig. 4e, last panel). These results were puzzling given that fat-7 expression levels are increased upon ash-2 depletion and, hence, should yield higher C18:1 levels compared to hsf-1-depleted worms, which show significantly decreased fat-7 expression levels ([184]Fig. 2b). However, they fit to the idea that not the storage of oleic acid but its subsequent utilization make this fatty acid beneficial for the health and lifespan in C. elegans. Based on the finding that hsf-1-depleted worms have significantly lower fat-7 expression levels, we must also assume that much of the detected C18:1 species in hsf-1-depleted worms is not oleic acid but cis-vaccenic acid (C18:1 Δ11), whose synthesis is catalyzed by the desaturase fat-5. Supplementing hsf-1-depleted worms with additional oleic acid caused a further re-distribution of the TGs, leading to the further accumulation of very-long-chain PUFAs at the expense of MUFAs ([185]Fig. 4e). These results further supported our conclusion that hsf-1-depleted worms lack the capacity to effectively utilize oleic acid and explained why supplementation with oleic acid does not protect hsf-1-depleted worms against proteotoxicity ([186]Fig. 3a,[187]b). To fully exclude that HSF-1ʼs protective effects on amyloid proteotoxicity are mediated by changes in lipid utilization rather than lipid storage, we used worms expressing the lipid droplet marker dhs-3p::dhs-3::GFP and monitored their lipid droplet formation upon growth on control, ash-2, hsf-1 and ash-2+hsf-1 RNAi in the absence or presence of oleic acid. As previously shown^[188]34, depletion of hsf-1 leads to a decrease in DHS-3::GFP fluorescence signal, which is unaffected by oleic acid supplementation ([189]Fig. 4f and [190]Extended Data Fig. 4c–[191]e) and suggestive of defects in lipid droplet formation. However, depletion of ash-2 as well as supplementing control RNAi worms with oleic acid also caused a significant reduction in DHS-3::GFP fluorescence signal, which was particularly noticeable in day 1 adult ([192]Fig. 4f). By day 5 of adulthood, however, both control and ash-2 RNAi strains had similar DHS-3::GFP signals, which were not only significantly higher compared to hsf-1-depleted worms but also further increased upon oleic acid supplementation ([193]Extended Data Fig. 4c–[194]e). We did not observe any increase in hsf-1 mRNA or lipid droplet levels in ash-2-deficient worms later in life (day 10 of adulthood; [195]Extended Data Fig. 4f–[196]h). These results ruled out a direct correlation between lipid droplet levels and proteotoxicity and further supported the conclusion that increased lipid utilization in early adulthood might protect against proteotoxicity later in life. Fig. 4 |. Disruption of HSF-1 remodels the lipid landscape. Fig. 4 | [197]Open in a new tab a, Heat map of lipid species detected by mass spectrometry in day 1 adult Q40::YFP animals maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi bacteria. Values represent the average z-score from 4–5 biological replicates. CL, cardiolipin; DAG, diacylglyceride; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PS, phosphatidylserine; SM, sphingomyelin. b, Partial least squares discriminant analysis (PLS-DA) plots of individual samples from conditions in a. c, Volcano plots of differentially expressed lipids under indicated RNAi conditions and comparisons. d,e, Heat map (d) and z-score plot (e) of triglyceride composition features in Q40::YFP animals maintained under conditions in a with and without oleic acid supplementation. f, Quantification of lipid droplet fluorescence in day 1 adult dhs-3p::dhs-3::GFP C. elegans maintained on conditions in a. Results are the average (a,d,e) or representative (f) of at least four (a–e) or three biological (f) replicates. Statistically significant differences were determined by one-way ANOVA. Additional information on lipidomics data can be found in [198]Supplementary Table 2. FC, fold change; OA, oleic acid; SFA, saturated fatty acid. HSF-1 controls the committed step of fatty acid oxidation Given the observed differences in lipid composition and potential utilization in ash-2-depleted and hsf-1-depleted animals, we next assessed the changes in the expression of genes associated with lipid metabolic processes in Q40::YFP worms grown on control, ash-2, hsf-1 or ash-2+hsf-1 RNAi. Not surprisingly, we found many of the gene expression changes associated with the ash-2 depletion to be reversed upon depletion of hsf-1 ([199]Fig. 5a and [200]Supplementary Table 3). Metabolic pathway enrichment analysis further supported our lipidomics results and revealed an enrichment of genes involved in fatty acid biosynthesis and degradation pathways, particularly upon depletion of hsf-1 ([201]Fig. 5b,[202]c and [203]Extended Data Fig. 5a). One of the genes most positively affected in ash-2-depleted versus ash-2+hsf-1-depleted Q40::YFP worms encoded for acyl-CoA dehydrogenase 1 (ACDH-1) ([204]Fig. 5d), the enzyme that catalyzes the committed step of mitochondrial fatty acid oxidation^[205]35. In fact, acdh-1 was previously shown to be one of the top genes whose expression is dependent upon HSF-1 in wild-type C. elegans under non-stress conditions^[206]26. We further confirmed this result by RT–PCR and found that Q40::YFP animals deficient in hsf-1 show an over 90% reduction in acdh-1 mRNA ([207]Extended Data Fig. 5b). In addition to acdh-1 expression, depletion of hsf-1 alone ([208]Extended Data Fig. 5b) or in the background of ash-2-depleted worms ([209]Fig. 5d) also negatively affected the expression of several other proteins involved in mitochondrial β-oxidation, including hacd-1 as well as ech-7, the enzyme catalyzing the second step in mitochondrial β-oxidation^[210]27. Based on these results and given that fatty acids such as oleic acid are thought to be preferentially targeted toward the mitochondria for degradation^[211]27, we now wondered whether the beneficial effects of ash-2 depletion or oleic acid supplementation are caused by an HSF-1-mediated increase in β-oxidation, a lipid metabolic process previously also associated with lifespan extension^[212]36,[213]37. To test whether this is indeed the case, we maintained Q40::YFP worms on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi in the presence or absence of perhexiline (PHX). This compound disrupts fatty acyl transport into the mitochondria through the inhibition of the carnitine palmitoyl transporters and has successfully been used to interrogate the effects of β-oxidation on lifespan in C. elegans^[214]38. Whereas PHX treatment alone showed little effect on the onset of paralysis in Q40::YFP worms grown on control RNAi, it completely abolished the protective effects triggered by ash-2 depletion ([215]Fig. 5e). No additional detrimental effects were observed when we treated hsf-1-depleted or ash-2+hsf-1-depleted worms with PHX ([216]Fig. 5e). These results provided evidence that the transport of fatty acyls into mitochondria might be critical for the protective effects of ash-2 depletion. Once transported into the mitochondria, fatty acids are broken down via β-oxidation to yield NADH, FADH[2] and acetyl-CoA for mitochondrial respiration. To determine whether there are any associated changes in mitochondrial respiration upon ash-2 depletion, we measured the oxygen consumption rate (OCR) in Q40::YFP worms maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi in the presence or absence of oleic acid using a Seahorse XFe96 analyzer ([217]Fig. 5f). Intriguingly, the obtained OCR data effectively mirrored our earlier paralysis results ([218]Fig. 3a). Whereas Q40::YFP worms maintained on either ash-2 RNAi or control RNAi worms cultivated on oleic acid had significantly higher OCRs compared to the respective controls, hsf-1-depleted worms showed significantly decreased OCRs. This defect became particularly obvious when we compared the OCRs of ash-2+hsf-1-depleted with ash-2-depleted worms ([219]Fig. 5f). Furthermore, supplementation of hsf-1-depleted worms with oleic acid had either no or only very modest stimulatory effects on hsf-1-depleted worms, suggesting that hsf-1-depleted worms are unable to utilize supplemented oleic acid as energy source. Notably, the differences in observed OCR were not due to physical impairments caused by hsf-1 RNAi, as, at this stage, worm motility was unaffected by hsf-1 knockdown ([220]Extended Data Fig. 5c). Additionally, the increase in OCR in both ash-2 knockdown and oleic acid supplementation is not transient and is still present in older day 5 adult animals ([221]Extended Data Fig. 5d). Conducting the same experiments in the presence of PHX further confirmed these conclusions ([222]Fig. 5g) and showed that treatment with PHX significantly reduced the OCR of Q40::YFP worms grown on either control or ash-2 RNAi yet failed to significantly affect the OCR of hsf-1-depleted worms. These results not only demonstrated that HSF-1 plays a significant role in stimulating mitochondrial β-oxidation but also provided further proof for the model that disruption of H3K4me3 modifiers, such as ash-2, protect organisms against proteotoxicity through a combination of oleic acid production and utilization early in life ([223]Extended Data Fig. 6). Fig. 5 |. Differential regulation of lipid metabolism by HSF-1 governs resistance to proteotoxicity in the absence of ash-2. Fig. 5 | [224]Open in a new tab a, Heat map of expression changes in genes involved in lipid metabolic processes (GO:0006629) in day 1 adult Q40::YFP worms maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi. Values are the average z-score from 4–5 biological replicates. b,c, Metabolic pathway enrichment analysis (WormFlux) of differentially expressed genes between ash-2 versus L4440 (b) or ash-2+hsf-1 versus ash-2 (c). Categories with P < 0.05 were considered enriched and are colored according to the scale, and categories with P > 0.05 are colored gray. d, Expression changes of genes involved in fatty acid synthesis and degradation in day 1 adult Q40::YFP worms maintained on ash-2+hsf-1 RNAi versus ash-2 RNAi, with statistically significant changes in bold. e, Paralysis of Q40::YFP worms maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi bacteria in the presence or absence of PHX. f, OCRs of day 1 adult Q40::YFP worms maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi ± oleic acid (0.8 mM) measured using a Seahorse XFe96 analyzer. Five baseline OCR measurements were taken per experiment, with each point representing the average baseline OCR reading of 3–5 technical replicate wells across four independent experiments (20 data points in total). Wells contained an average of 15 worms per replicate, with wells containing fewer than five worms excluded from the analysis. g, OCR in day 1 Q40::YFP worms maintained on L4440, ash-2, hsf-1 or ash-2+hsf-1 RNAi in the presence or absence of PHX (2.5 mM), with each point representing the average baseline OCR reading of five technical replicates across two independent experiments (10 data points in total). Results are the average (a,d), average ± s.e.m. (f,g) of two (g) or at least three (a–f) independent experiments. Statistically significant differences were assessed by log-rank test (e) or one-way ANOVA with Fisher’s LSD test (f,g); *P < 0.05 and ****P < 0.0001. Data for biological replicates not shown can be found in [225]Supplementary Table 1, and detailed experimental information for RNA-seq analysis can be found in [226]Supplementary Table 3. FC, fold change; OCR, oxygen consumption rate; FFA, free fatty acid; NS, not significant. Discussion Uncovering the H3K4me3–HSF-1–lipid homeostasis axis H3K4me3 modifiers of the COMPASS complex are known to affect organismal lifespan and stress resistance^[227]4,[228]8,[229]16. Particularly intriguing, however, was the recent discovery that the transient oxidative inactivation of components of the COMPASS complex during C. elegans development was sufficient to extend lifespan^[230]4. These results added to a growing body of evidence that longevity effects can be achieved by appropriate early-in-life interventions^[231]4–[232]6,[233]39. Here we demonstrated that transient exposure to ROS in early life also delays the onset of amyloid toxicity in C. elegans mutants expressing disease-related amyloids. These results support the newly emerging geroscience hypothesis, which states that interventions that extend organismal lifespan might also delay the onset of age-associated diseases^[234]40. However, they also re-emphasized the need to mechanistically interrogate how early-in-life events protect against age-related processes later in life. Here we show that depletion of H3K4me3 modifiers triggers the functional upregulation of HSF-1 activity ([235]Extended Data Fig. 6), the master regulator of the eukaryotic proteostasis network. Based on the prevailing idea that a decline in proteostasis contributes to aging and aging-associated diseases, HSF-1 has long been considered to serve as a longevity factor through its central role in controlling the expression of chaperones and other folding factors^[236]24. However, other studies suggested the involvement of proteostasis-independent effects of HSF-1 in organismal lifespan and healthspan. In worms, for instance, HSF-1 activity was found to be most critical during the developmental phase, whereas depletion of hsf-1 during adulthood had only minor detrimental effects^[237]13,[238]25. These results agreed with the finding that HSF-1 function substantially decreases upon transition into C. elegans adulthood^[239]41. However, they left the intriguing question open as to how HSF-1 maintains protein homeostasis until late in life. Finally, RNA-seq results indicated that HSF-1 controls the expression of more non-heat shock genes than heat shock genes particularly during development, suggesting additional roles of this highly conserved transcription factor beyond its involvement in cellular proteostasis^[240]24,[241]26. Our study potentially reconciles all these results by demonstrating that HSF-1 plays a critical role in combating proteotoxicity in C. elegans through its effects on lipid homeostasis. By directly controlling the levels of desaturases (for example, fat-7), lipases (for example, lipl-2) and enzymes involved in fatty acid oxidation (for example, acdh-1 and ech-7), HSF-1 greatly influences the lipid homeostasis in C. elegans. Notably, by using epistasis experiments and inhibitor studies, we found that this HSF-1-mediated combination of increased MUFA synthesis and utilization plays a critical role in combating proteotoxicity. These results add to an expanding list of studies that demonstrated the beneficial effects of increasing the ratio of MUFAs over PUFAs^[242]42 and supplementing growth media with oleic acid^[243]16,[244]30,[245]42 or reported on the direct link between increased β-oxidation and lifespan extension^[246]37,[247]38,[248]43,[249]44. It is intriguing to note that both measures, increasing MUFA/PUFA ratios and enhancing β-oxidation, have been associated with decreasing the levels of ROS and mitigating oxidative damage^[250]42,[251]43. Given, however, that all three events—H3K4me3 depletion, HSF-1 upregulation and increased MUFA production—need to happen during C. elegans development, further investigations are clearly needed to determine when in life and how increases in mitochondrial β-oxidation promote stress resistance and longevity. HSF-1-depleted worms show signs of lipid starvation Recent work from Watterson et al.^[252]34 demonstrated that depletion of hsf-1 causes a substantial decrease in lipid droplet formation, which was proposed to be triggered by the activation of the nuclear hormone receptor NHR-49, a transcription factor known to upregulate a number of genes involved in β-oxidation^[253]44. However, our OCR measurements did not provide any evidence that hsf-1-depleted worms show any increase in mitochondrial fatty acid oxidation ([254]Fig. 5f). In contrast, OCRs in hsf-1-depleted worms were extremely reduced, non-responsive to oleic acid supplementation and unaffected by PHX, an established inhibitor of fatty acid oxidation. These results strongly suggest that hsf-1-depleted worms lack the capacity to utilize oleic acid and potentially other fatty acids as energy source, resulting in a starvation-like state^[255]34. Indeed, the observed transcriptional upregulation of many lipases in hsf-1-depleted animals ([256]Extended Data Fig. 5c) may be reflective of cellular attempts to access lipid stores and increase lipid mobilization under these conditions. We now argue that it is this inability of hsf-1-depleted worms to appropriately utilize TGs that leads to the activation of nhr-49, which further depletes lipid stores, and causes a major redistribution of the cellular lipid composition. It is also noteworthy that, whereas the number of DHS-3-positive lipid droplets observed in hsf-1-depleted worms does not significantly increase upon supplementation with oleic acid, the ORO signal, which stains for neutral lipids in fixed worms, was significantly higher in hsf-1-depleted strains compared to the control strains upon oleic acid supplementation. These results confirm that hsf-1-depleted worms are unable to either utilize or appropriately store the lipids in DHS-3-positive compartments. Subsequent modification of these incorrectly stored lipids might contribute to the massively altered lipid profile found in hsf-1-depleted worms. Upregulation of HSF-1 activity via COMPASS complex depletion Our studies revealed a previously unknown connection between decreased levels of H3K4me3 modifiers and increased activity of HSF-1. These results were consistent with previous reports, which showed that depleting members of the COMPASS complex augments the heat shock response in both C. elegans and cell culture models^[257]4 and causes increased steady-state expression of a large number of HSF-1-controlled non-heat shock genes in both wild-type and Q40::YFP-expressing worms (this study) without affecting hsf-1 mRNA levels^[258]4. Depletion of the H3K4me3 machinery might either directly or indirectly affect the promoter accessibility of transcriptional targets or otherwise alter the stability or activity of HSF-1, for instance through alterations in its post-translational makeup. Current studies are underway to precisely determine how disruption of the H3K4me machinery affects HSF-1 activity. These results, however, also raised the obvious question as to how depletion of H3K4me3, a mark considered to be activating and reflective of transcriptional memory^[259]8, leads to increased transcriptional activity of HSF-1. Previous work showed that the HSF-1 function in worms is regulated by the activity of the H3K27me3 demethylase JMJD-3.1, which removes the repressive H3K27me3 mark from heat shock promoter regions, thus enhancing HSF-1 activity during development^[260]41. Upon transition into adulthood, however, a germline-dependent downregulation of jmjd-3.1 causes the accumulation of the repressive mark on HSF-1 promoters and, hence, explains the previously observed decrease in HSF-1 activity in adult worms^[261]41. It now remains to be seen whether the oxidative downregulation of H3K4me3 levels either directly or indirectly affects other marks, such as H3K27me3 on HSF-1-controlled heat shock and non-heat shock promoters. Intriguingly, the contribution of ash-2 in COMPASS complex activity (that is, trimethylation and demethylation of H3K4) has recently been shown to also vary depending on the tissue and developmental stage of the worms^[262]9. Given the phenotypical differences that we and others observed between the set-2 and ash-2 depletion strains, with the latter showing reproducibly more protective phenotypes (this study) and longer lifespan^[263]8, it is also possible that changes in H3K4me2 levels rather than H3K4me3 levels affect the promoter occupancy and transcriptional activity of HSF-1. Finally, it is conceivable that depletion of members of the COMPASS complex affect one or more post-translational modifications on HSF-1, which are known to regulate HSF-1 stability, activity and specificity. Future studies are needed to mechanistically dissect this previously unknown link between early-in-life histone modifications and the activity of one of the best known longevity factors in biology. Methods C. elegans strains, maintenance and synchronization The following strains of C. elegans were used in this study: CL4176 (dvIs27 [myo-3p::Aβ (1–42)::let-851 3′ UTR) + rol-6(su1006)]); CL2006 (dvIs2 [pCL12(unc-54/human Aβ peptide 1–42 minigene) + rol-6(su1006)]); AM141 (rmIs133 [unc-54p::Q40::YFP]); UJ002 (dvIs27; set-2(ok952)); UJ001 (rmIs133; set-2(ok952)); CL2070 (dvIs70 [hsp-16.2p::GFP + rol-6(su1006)]); DMS303 (nIs590 [fat-7p::fat-7::GFP + lin15(+)]); PHX4983 (syb4983 [hsp-16.2p::hsp-16.2::mCherry]); UJ003 (rmIs133; syb4983); LIU1 (ldrIs1 [dhs-3p::dhs-3::GFP + unc-76(+)]); GRU102 (gnaIs2 [myo-2p::YFP + unc-119p::Aβ1–42]); and CB1402 (unc-15(e1402)). Transgenic PHX4983 animals were generated by SunyBiotech using CRISPR–Cas9-mediated knock-in of hsp-16.2::mCherry at the endogenous locus. Worms were maintained per standard protocols on NGM plates containing streptomycin and nystatin seeded with streptomycin-resistant Escherichia coli OP50. Strains expressing Aβ were maintained at 15 °C, and all other strains were maintained at 20 °C. Synchronized populations of worms were obtained by transferring gravid adults to new plates, allowing them to lay eggs for 2–3 h and then removing the adults. Paralysis and lifespan assays CL2006 and Q40::YFP animals were synchronized by timed egg lay, transferred to fresh plates when they reached day 1 of adulthood (defined as the first day that eggs appear on the plates, t = 0 on graphs) and incubated at 20 °C. Animals were moved to fresh plates as necessary to avoid contamination with their progeny. Paralysis assays in CL4176 animals were conducted using the temperature shifting regimen previously described^[264]17, with slight modifications. In brief, animals were synchronized by timed egg lay for 2 h and maintained at 15 °C. Fifty hours after the egg lay, animals were shifted to 25 °C. Paralysis was assessed beginning 32 h after temperature upshift by gentle prodding with a platinum worm pick. Animals that failed to propagate a full-body wave in response to stimulation were counted as paralyzed. Lifespan assessment in GRU102 animals was conducted at 20 °C, with animals failing to respond to gentle prodding with a platinum worm pick considered dead. Animals with vulval protrusions or internally hatched embryos were censored. RNAi in C. elegans RNAi was achieved through feeding worms E. coli HT115 (DE3) strains expressing dsRNA to genes of interest from the Ahringer RNAi library (a kind gift from Gyorgyi Csankovszki). Bacteria were seeded onto NGM plates containing ampicillin (50 μg ml^−1), tetracyclin (12.5 μg ml^−1) and IPTG (1 mM) at an OD[600] of 0.8–1.0. Bacterial lawns were allowed to form for at least 48 h before worm placement. In the instances where multiple genes were targeted by RNAi, equal mixtures of the respective HT115 RNAi clones were seeded on plates. HT115 bacteria containing the empty L4440 vector used for dsRNA expression were used as a control. In experiments using multiple RNAi bacteria, dilution of RNAi bacteria under these conditions was accounted for in the other conditions by combining the target gene RNAi with equal amounts of control RNAi encoding L4440 bacteria. In all experiments using ash-2 or set-2 RNAi, worms were maintained on RNAi plates for one generation before the experimental population, as this has been found to increase the RNAi effect of ash-2 or set-2 knockdown. All experiments using other RNAi clones were conducted by hatching synchronized populations of eggs on RNAi plates and conducting experiments in that population. Western blotting For western blotting of C. elegans, 150–200 adult worms were collected in 40 μl of M9 buffer, to which 10 μl of 5× Laemmli loading buffer was added. Samples were boiled for 5 min and vortexed thoroughly until worms had completely solubilized. Electrophoresis, transfer and blocking were conducted per standard protocols^[265]4. Primary and secondary antibodies were diluted in 1% w/v milk/Tris-buffered saline and 0.1% Tween 20 and incubated at 4 °C overnight or at room temperature for 45 min, respectively. Bands were visualized using enhanced chemiluminescence (SuperSignal West Pico PLUS, Thermo Fisher Scientific, 34580). The following primary and secondary antibodies were used in this study: anti-H3 (Abcam, ab1791, 1:2,000); anti-H3K4me3 (Abcam, ab8580, 1:1,000); anti-GFP (Novus Biologicals, NB600-303, 1:2,000); anti-amyloid-β 1–16 (clone 6E10, BioLegend, 803001, 1:1,000); anti-rabbit horseradish peroxidase (Thermo Fisher Scientific, 31460, 1:10,000); and anti-mouse horseradish peroxidase (Thermo Fisher Scientific, 31430, 1:10,000). PQ and MM-401 treatment PQ (methyl viologen hydrate, ACROS Organics) or MM-401 (a kind gift from Yali Dou) was added to NGM plates at the indicated concentrations, allowed to solidify and seeded with OP50. CL4176 animals maintained at 15 °C were transferred to PQ-containing plates during the L2 stage (38 h after egg lay) for 10 h and then moved back to NGM plates without PQ before temperature upshift to 25 °C to stimulate Aβ production^[266]17. RNA isolation, cDNA synthesis and qPCR RNA isolation from C. elegans was performed per standard protocols with slight modifications^[267]4. In brief, 20–25 adult animals were picked into a 5-μl drop of lysis buffer (5 mM Tris, pH 8.0, 0.5% v/v Triton X-100, 0.5% v/v Tween 20, 0.25 mM EDTA) containing 1 mg ml^−1 proteinase K (New England BioLabs) in the lid of a PCR tube. Animals were centrifuged to the tube bottom, subjected to three rounds of freeze–thaw cycles with a dry ice/ethanol bath and lysed in a thermocycler (65 °C for 10 min and 85 °C for 1 min). cDNA synthesis was performed using the iScript cDNA Synthesis Kit (Bio-Rad), and qPCR was performed using Radiant Green Lo-ROX qPCR Kit (Alkali Scientific) per the manufacturer’s instructions. Reactions were performed in technical triplicate using pmp-3 as a housekeeping gene, and relative gene expression was determined using the 2^−ΔΔCT method^[268]45. The following primers were used: fat-7 (for) 5′-GCGCTGCTCACTATTTTGGT-3′ and fat-7 (rev) 5′-GTGGGAATGTGTGGTGGAAA-3′; pmp-3 (for) 5′-TTTGTGTCAATTGGTCATCG-3′ and pmp-3 (rev) 5′-CTGTGTCAATGTC GTGAAGG-3′; hsp-16.11 (for) 5′-TGGCTCAGATGGAACTGCAA-3′ and hsp-16.11 (rev) 5′-TGG CTTGAACTGCGAGACAT-3′; hsp-16.2 (for) 5′-CTGTGAGACGTTGAGATTGATG-3′ and hsp-16.2 (rev) 5′-CTTTACCACTATTTCCGTCCAG-3′; hsp-16.48 (for) 5′-GCTCATGCTCCGTTCT CCAT-3′ and hsp-16.48 (rev) 5′-GAGTTGTGATCAGCATTTCTCCA-3′; acdh-1 (for) 5′-CGAA ATGCAGATCCTAGCC-3′ and acdh-1 (rev) 5′-GTTTGTCTTCCTCCTTATCTACAG-3′; ash-2 (for) 5′-CGCCTATTACCCGTCGATT-3′ and ash-2 (rev) 5′-GTTTGTTCGTGTTGCTGCTC-3′; and hsf-1 (for) 5′-CTGGAGCAGCACGTCGTTAT-3′ and hsf-1 (rev) 5′-CCGGATTTGTTCAAGGTCTCC-3′. Fluorescence microscopy of C. elegans Aggregation of Q40::YFP was visualized by mounting AM141 C. elegans on a 2% agarose pad placed on a microscope slide immobilized with 10 μl of levamisole (25 mM, MP Biomedicals). Fluorescent reporter strains (hsp-16.2p::GFP, fat-7p::fat-7::GFP and dhs-3p::dhs-3::GFP) were imaged by picking worms into a small drop of levamisole on an unseeded NGM plate. Imaging was conducted immediately after levamisole drying on a Shimadzu Olympus (MetaMorph software) or ECHO Revolution microscope using a ×4 objective. Quantification of Q40::YFP foci was performed in ImageJ (National Institutes of Health) by manual counting in images with defined thresholds that were kept uniform across all conditions. Quantification of GFP fluorescence in fluorescent reporter strains (fat-7p::fat-7::GFP, hsp-16.2p::GFP and dhs-3p::dhs-3::GFP) was performed in ImageJ by measuring mean fluorescence intensity per animal. Assessment of HSP-16.2::mCherry levels in Q40::YFP;hsp-16.2p::hsp-16.2::mCherry animals was conducted on levamisole-immobilized animals on microscope slides containing 2% agarose pads. Animals were then imaged with a ×20 objective on a Leica THUNDER imager using an excitation/emission of 475/535 nm for YFP and 575/590 nm for mCherry. All images within an experiment were acquired with uniform acquisition settings. Quantification of HSP-16.2::mCherry was performed in ImageJ by quantifying mean fluorescence intensity in the heads of animals. Images and figures were prepared with Adobe Illustrator 2020. Lipidomics Total lipid extracts from C. elegans pellets were prepared using a modified MTBE lipid extraction protocol^[269]46. In brief, 400 μl of cold methanol and 10 μl of internal standard mixture (EquiSPLASH LIPIDOMIX) were added to each sample, followed by incubation at 4 °C with 650 r.p.m. shaking for 15 min. Next, 500 μl of cold MTBE was added, followed by incubation at 4 °C for 1 h with 650 r.p.m. shaking. Then, 500 μl of cold water was added slowly, and the resulting extract was maintained 4 °C with 650 r.p.m. shaking for 15 min. Phase separation was completed by centrifugation at 8,000 r.p.m. for 8 min at 4 °C. The upper, organic phase was removed and set aside on ice. The bottom, aqueous phase was re-extracted with 200 μl of MTBE, followed by 15 min of incubation at 4 °C with 650 r.p.m. shaking. Phase separation was completed by centrifugation at 8,000 r.p.m. for 8 min at 4 °C. The upper, organic phase was removed and combined with previous organic extract. The organic extract was dried under a steady stream of nitrogen at 30 °C. The recovered lipids were reconstituted in 200 μl of chloroform:methanol (1:1, v/v) containing 200 μM of butylated hydroxytoluene. Before analysis, samples were further diluted 10-fold with acetonitrile:isopropanol:water (1:2:1, v/v/v). The lower, aqueous phase was used to determine the protein content via a bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific). Lipid analysis Total lipid extracts were analyzed by LC–MS/MS. The LC–MS/MS analyses were performed on an Agilent 1290 Infinity LC coupled to an Agilent 6560 Quadrupole Time-of-Flight (Q-TOF) mass spectrometer. The separation was achieved using an C18 CSH (1.7 μm; 2.1 × 100 mm) column (Waters). Mobile phase A was 10 mM ammonium formate with 0.1% formic acid in water:acetonitrile (40:60, v/v), and mobile phase B was 10 mM ammonium formate with 0.1% formic acid in acetonitrile:isopropanol (10:90, v/v). The gradient was ramped from 40% to 43% B in 1 min, ramped to 50% in 0.1 min, ramped to 54% B in 4.9 min, ramped to 70% in 0.1 min and ramped to 99% B in 2.9 min. The gradient was returned to initial conditions in 0.5 min and held for 1.6 min for column equilibration. The flow rate was 0.4 ml min^−1. The column was maintained at 55 °C, and the auto-sampler was kept at 5 °C. A 2-μl injection was used for all samples. Mass spectrometry analysis was separated into two workflows: (1) lipid identification of a pooled sample using an iterative MS/MS acquisition and (2) lipid semi-quantitation of all samples using high-resolution, accurate mass MS^[270]1 acquisition. The MS parameters for the iterative workflow were as follows: extended dynamic range, 2 GHz; gas temperature, 200 °C; gas flow, 10 L min^−1; nebulizer, 50 psi; sheath gas temperature, 300 °C; sheath gas flow, 12 L min^−1; VCap, 3.5 kV (+), 3.0 kV (−); nozzle voltage, 250 V; reference mass m/z 121.0509, m/z 1,221.9906 (+), m/z 119.0363, m/z 980.0164 (−); MS and MS/MS range m/z 100–1,700; acquisition rate, 3 spectra per second; isolation, narrow (~1.3 m/z); collision energy 20 eV (+), 25 eV (−); max precursors per cycle, 3; precursor abundance-based scan speed, 25,000 counts per spectrum; MS/MS threshold, 5,000 counts and 0.001%; active exclusion enabled, yes; purity, stringency 70%, cutoff 0%; isotope model, common organic molecules; and static exclusion ranges, m/z 40 to 151 (+,−). The MS parameters for the MS^[271]1 workflow were the same for source and reference mass parameters and differed only for acquisition. LC–MS/MS data from the iterative MS/MS workflow were analyzed for lipid identification via Agilent’s Lipid Annotator (version 1.0). The default settings for feature finding and identification parameters were used. Positive and negative ion mode adducts included [M+H]^+, [M+Na]^+, [M+NH[4]]^+, [M−H]^− and [M+CH[3]CO[2]]^−, respectively. The results of the Lipid Annotator were saved as a PCDL file. The LC–MS data from the MS^[272]1 workflow were processed using Agilent’s MassHunter Profinder (version 10.0). Batch targeted feature extraction using default parameters and the PCDL file created from Lipid Annotator were used for feature extraction. The processed data generated from Profinder, which included peak area and lipid identification, were exported into MetaboAnalyst 4.0 (ref. [273]47) for multivariate analysis. Univariate analysis was done using Prism 6 (GraphPad Software). Lipids considered to be differentially expressed were those with P < 0.05 by t-test. RNA extraction, sequencing and pathway enrichment analysis Day 1 adult C. elegans was collected from NGM plates by washing with M9. Animals were pelleted by centrifugation (1,000g, 1.5 min); supernatant was aspirated; and 10 ml of fresh M9 was added. This wash step was performed a total of three times. After washing, supernatant was aspirated away, leaving as little M9 as possible. A glass pipette was used to collect the pelleted worms and immediately dropped into liquid nitrogen to form frozen pellets. Frozen pellets were homogenized in 250 μl of TRIzol with 100 μl of zirconia beads, using a bead mill homogenizer (Omni International). RNA was extracted using chloroform and concentrated using an RNeasy column (Qiagen), following the manufacturer’s instructions. DNase I digestion (Thermo Fisher Scientific) was then performed on eluted RNA per the manufacturer’s instructions, and samples were stored at −80 °C until further processing. RNA-seq was conducted by LC Sciences, where RNA integrity was first assessed using an Agilent 2100 Bioanalyzer. Poly(A) RNA-seq libraries were generated using the TruSeq Stranded mRNA protocol (Illumina). Oligo-(dT) magnetic beads were used to isolate mRNAs containing poly(A) tails, with two rounds of purification. Poly(A) RNA was then fragmented using divalent cation buffer in elevated temperature. Paired-end sequencing was performed on the NovaSeq 6000 sequencing system (Illumina). Cleaning of the reads was performed using Cutadapt^[274]48 and in-house perl scripts (LC Sciences), and sequence quality was assessed using FastQC ([275]http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were mapped to the C. elegans genome using HISAT2 (ref. [276]49), and mapped reads were assembled into transcripts using StringTie^[277]50. Transcriptomes were compiled using perl scripts and gffcompare, and StringTie and ballgown ([278]http://www.bioconductor.org/packages/release/bioc/html/ballgown.h tml) were used to determine mRNA expression levels by calculating fragments per kilobase of transcript per million mapped reads (FPKM). Differential mRNA expression analysis was performed using DESeq2 (ref. [279]51) between two different groups and by the R package edgeR^[280]52 between two samples, with mRNAs with a false discovery rate (FDR) < 0.05 and absolute fold change ≥ 2 being considered as differentially expressed. Differentially expressed gene sets identified in RNA-seq analysis were assessed for enrichment in metabolic pathways using the WormFlux Pathway Enrichment Analysis tool^[281]53, with enrichment determined by hypergeometric distribution and P < 0.05. Fatty acid supplementation and ORO staining Supplementation of NGM plates with oleic acid or linoleic acid was performed as previously described^[282]16 with slight modifications. In brief, 0.01% v/v NP-40 was added in all plates before autoclaving to aid in even distribution of the fatty acids. Oleic acid sodium salt (Sigma-Aldrich, O7501) or linoleic acid sodium salt (Sigma-Aldrich, L8134) was added to molten NGM to a final concentration of 0.8 mM. Staining of C. elegans lipids was conducted using ORO as previously described^[283]54. ORO-stained worms were imaged on a stereoscope equipped with an SC50 color camera (Olympus). Lipid accumulation in whole animals was quantified in ImageJ by converting color images to grayscale, inverting images and quantifying mean intensity in outlined worms. PHX treatment PHX (Cayman Chemicals) was administered to C. elegans as previously described^[284]38. Immediately before animals being placed on plates, a 100 mM PHX stock solution was diluted to 2.5 mM PHX solution with ddH[2]O, and 100 μl was pipetted onto the bacterial lawn of seeded NGM plates. For control plates without PHX, 100 μl of ddH[2]O was added. Plates were allowed to dry, and, after drying, animals were placed onto the plates. PHX was administered from egg lay, and new plates were prepared as needed when worm transfer was necessary. Assessment of OCRs in C. elegans OCRs were measured using a Seahorse Xfe96 Analyzer (Agilent) as previously described^[285]55. In brief, day 1 adult Q40::YFP worms (96 h after egg) were collected by washing from plates in M9 buffer, and worms were added to the wells of an Xfe96 well plate in 200 μl of volume. The baseline OCR was assessed five times per condition using a program of 2 min mixing, 0.5 min waiting and 2 min measuring. Measurements were performed at room temperature (~27–30 °C with heat from the instrument). Five technical replicates were used per experimental condition, with wells containing an average of 15 worms per well. Wells containing fewer than five worms were excluded from analysis. Statistically significant changes in OCR were determined by one-way ANOVA with Fisher’s least significant difference (LSD) test. Thrashing assay for motility analysis Worm motility was assessed by recording the rate of body bending of thrashing worms suspended in liquid. In brief, day 1 adult Q40::YFP worms maintained on indicated conditions were picked into a drop of M9 buffer on a glass microscope slide. Animals in the drop were gently stirred with an eyelash pick to disrupt clumps and ensure that worms were uniformly suspended in liquid. A 30-s video of the worms thrashing was recorded, and the number of body bends performed by each animal was counted. One body bend was defined as the movement that an animal made to bring its head and tail to the opposite side of a midline from the middle of its body, forming a concave shape. The conditions were blinded to the counter, and the videos were slowed down to 0.25× speed to ensure accuracy of counting. Thrashing assessment of CB1402 paramyosin(ts) animals was then assessed using the assay conditions listed above, although these animals were first maintained on indicated experimental conditions at until day 1 of adulthood at 15 °C and then shifted to 25 °C for 24 h before thrashing assessment. Motility of these animals was then binned and categorized by degree of motility, and data were analyzed by χ^2 test. Statistics and reproducibility Sample sizes were similar to those in previous publications and protocols within our laboratory or based on literature and established protocols in the C. elegans research field (see [286]Reporting Summary). Statistical information, including n number, error bars, P values and statistical test used, can be found in the figures, figure legends or [287]Supplementary Table 1. Data from paralysis assays were plotted with GraphPad Prism as survival curves and analyzed for statistical significance with the log-rank (Mantel–Cox) test. Statistics for all paralysis and lifespan experiments can be found in [288]Supplementary Table 1. Animals used for experiments were randomly chosen from age-synchronized populations. No data in this study were excluded, with the exception of two biological replicates from RNA-seq analysis, which were excluded due to low Pearson’s correlation with other condition replicates. Most findings are results of at least three independent experiments, and, at minimum, two independent experiments, with exact repetitions specified in the figure legends. In most experiments, data collection and analysis were not performed blinded to the conditions of the experiments. Reproducible observations were obtained across independent repetitions by multiple researchers. Information on which experiments were independently repeated, by how many users and which were blinded can be found in [289]Supplementary Table 4. Extended Data Extended Data Fig. 1 |. Amyloid levels and polyQ foci formation in H3K4me3 modifier-deficient C. elegans (related to [290]Fig. 1). Extended Data Fig. 1 | [291]Open in a new tab (a, b) Levels of Aβ[1–42] in CL4176 worms maintained on L4440, ash-2 or set-2 RNAi were assessed by western blot (a) and quantified by densitometry (b). (c, d) Levels of Q40::YFP in worms maintained on control (L4440), ash-2 or set-2 RNAi were assessed by western blot (c) and quantified by densitometry (d). (e, f) Q40::YFP expressing C. elegans strain AM141 was maintained on control (L4440), ash-2 or set-2 RNAi and Q40::YFP aggregates were examined 96 h after egg lay (that is, day 1 of adulthood) by fluorescence microscopy (e) and quantified (f). Each symbol represents an individual worm and at least 30 animals were assessed per condition. (g, h) Wild-type or set-2 (ok592) Q40::YFP C. elegans were examined 96 h after egg lay (that is, day 1 of adulthood) by fluorescence microscopy (g) and aggregates were quantified (h). Differences in relative aggregate number between (e, f) and (g, h) is likely due to the differences in the bacterial strains (RNAi strain L4440 versus OP50) used as food source. Each symbol represents an individual worm. At least 30 animals were assessed per condition. Results are representative (a, c, e, g) or the average ± SEM (graphs in b, d) of four (e, f), six (c, d), three (a, b L4440 and ash-2) or two (a, b; set-2) independent experiments. Statistical significance was assessed by one-way ANOVA (b, d), Kruskal-Wallis (f), or two-tailed Mann-Whitney (h); ****, p < 0.0001. Scale bar = 0.1 mm. Extended Data Fig. 2 |. H3K4me3 modifier depletion does not enhance heat shock gene expression under non-stress conditions (related to [292]Fig. 2). Extended Data Fig. 2 | [293]Open in a new tab (a) Fluorescence images of day 1 adult hsp-16.2p::GFP expressing worms maintained on control (L4440), ash-2 or set-2 RNAi before and following heat shock (HS). Scale bar is 0.1 mm. (b) Quantification of the fluorescence in individual worms (symbols) shown in (a). At least 20 animals assessed per condition. (c) Knockdown efficiency of ash-2 and hsf-1 in day 1 adult Q40::YFP animals maintained on L4440, ash-2, hsf- 1, or ash-2 + hsf-1 RNAi. (d) Paralysis of Q40::YFP animals maintained on L4440, set-2, hsf-1, or set-2 + hsf-1 RNAi. (e) Quantification of HSP-16.2::mCherry fluorescence in the heads of hsp-16.2p::hsp-16.2::mCherry-expressing Q40::YFP animals aged to day 5 of adult- hood at 20 °C on control (L4440), ash-2 or set-2 RNAi. At least 10 animals assessed per condition. (f) Steady state expression of heat shock genes hsp-16.11, hsp-16.2, hsp-16.48 in Q40::YFP animals maintained on control (L4440), ash-2 or set-2 RNAi as determined by RT-PCR. Results are representative (a, b, d, e) or the average (c, f) of two (d, e) or three or more (a-c, f) independent experiments. Error bars represent SEM. Statistical significance was determined by Kruskal-Wallis (b, f), one-way ANOVA (c, e) or log-rank (d); p < 0.05, *; p < 0.0001, ****. Additional experimental information, and data for biological replicates not shown can be found in [294]Supplementary Table 1. Extended Data Fig. 3 |. Oleic acid does not activate or potentiate the heat shock response (related to [295]Fig. 3). Extended Data Fig. 3 | [296]Open in a new tab (a) CB1402 (unc-15(e1402)) worms expressing paramyosin(ts) were maintained on L4440 or hsf- 1 RNAi ± oleic acid (0.8 mM) at 15 °C. On day 1 of adulthood, animals were shifted to 25 °C, and motility was assessed 24 h later by thrashing assay. (b) Q40::YFP animals were maintained on NGM plates in the absence of supplements or in the presence of oleic acid (0.8 mM) or linoleic acid (0.8 mM). Paralysis was measured as described before. (c, d) L4 or (e, f) day 1 adults expressing the hsp-16.2p::GFP reporter were maintained on control RNAi (L4440) in the absence or presence of 0.8 mM oleic acid. Worms were either left untreated or subjected to heat shock (HS) treatment (35 °C for 2 h, 20° for 2 h). Fluorescence images were taken (c, e) and quantified (d, f). At least 25 animals were assessed per condition. Scale bar is 0.1 mm. Results are representative of four (a) or three (b-f) independent experiments. Statistical significance was assessed by χ^2 test (a), Kruskal-Wallis test (c, e), and log-rank (f); *, p < 0.05; **, p < 0.01; ****, p < 0.0001. Additional experimental information, and data for biological replicates not shown, can be found in [297]Supplementary Table 1. Extended Data Fig. 4 |. Effects of H3K4me3 modifier or hsf-1 depletion and oleic acid supplementation on lipid ontology enrichment and lipid droplet formation (related to [298]Fig. 4). Extended Data Fig. 4 | [299]Open in a new tab (a) Mass spectrometry analysis of lipid levels in Q40::YFP animals maintained on L4440, ash-2, hsf-1, or ash-2 + hsf-1 RNAi was conducted, and significantly altered lipids between indicated conditions were analyzed for ontology enrichment using LION/web analysis. Enriched categories in lipids that increased in each comparison are shown in red, and lipids that decreased in each comparison shown in blue, with statistically significant categories being color-coded by −log(FDR q-value). (b) Changes in saturated, monounsaturated, and polyunsaturated fatty acids in phospholipids in day 1 adult Q40::YFP worms maintained on L4440, ash-2, hsf-1, ash-2 + hsf-1 in the absence or presence of oleic acid (0.8 mM). (c, d) LIU1 (dhs-3p::dhs-3::GFP) lipid droplet reporter animals were maintained on control (L4440), ash-2, hsf-1, or ash-2 + hsf-1 RNAi, and fluorescence was imaged at day 1 (c) and day 5 (d) of adulthood. Representative images are shown. Scale bar is 0.1 mm. (e) Quantification of the dhs-3p::dhs-3::GFP fluorescence in (d). (f) mRNA levels of hsp-16.2 and hsf-1 in day 10 adult Q40::YFP worms maintained on L4440 or ash-2 RNAi. (g) Fluorescence of day 10 adult LIU1 (dhs-3p::dhs-3::GFP) animals maintained on L4440 or ash-2 RNAi. Fluorescence is quantified in (h). At least 20 animals assessed per condition. Results are representative of at least three independent experiments. Error bars represent SEM. Statistical significance was assessed by one-way ANOVA (e), one sample, two-tailed t-test (f) or Kruskal-Wallis test (h); ***, p < 0.001; ****, p < 0.0001. Extended Data Fig. 5 |. hsf-1 knockdown alters genes involved in lipid synthesis and breakdown (related to [300]Fig. 5). Extended Data Fig. 5 | [301]Open in a new tab (a) Differentially expressed genes in day 1 adult Q40::YFP worms maintained on hsf-1 RNAi vs L4440 control RNAi were analyzed for metabolic pathway enrichment (see [302]methods). Significant categories are color-coded by p-value. (b) Q40::YFP worms maintained on L4440 or hsf-1 RNAi, and accumulation of acdh-1 was assessed by qPCR. Levels of the housekeeping gene pmp-3 were used as a control for normalization. (c) Q40::YFP animals were maintained on L4440 or hsf-1 RNAi in the presence or absence of oleic acid (0.8 mM) until day 1 of adulthood. Worm motility was determined by assessing animal thrashing rate following placement in a drop of M9. (d) OCR was assessed in day 5 adult Q40::YFP worms supplemented with oleic acid (0.8 mM) or maintained on ash-2 RNAi. (e) Changes in genes involved in fatty acid synthesis and breakdown in Q40::YFP worms deficient in hsf-1. Genes with statistically significant differences vs L4440 control are listed in bold. Results are the average (a, e), representative (c) or average ± SEM (b, d) of two (c), three (d), or four (a, b, e) independent experiments. Statistical significance was assessed by one-way ANOVA (B-D); *, p < 0.05; ****, p < 0.0001. Additional experimental information on RNAseq analysis can be found in [303]Supplementary Table 3. Extended Data Fig. 6 |. Working model. Extended Data Fig. 6 | [304]Open in a new tab Proposed mechanism by which early-in-life inactivation of H3K4me3-modifiers protects against proteotoxicity later in life. Depletion of H3K4me3 leads to the upregulation of HSF-1 activity, which is essential for the protective effect displayed by H3K4me3-deficient worms. HSF-1 activity is required for the upregulation of FAT-7, an enzyme which converts stearic acid into oleic acid, a protective mono-unsaturated fatty acid (MUFA). Although production of oleic acid is necessary for the observed protection, it’s presence is insufficient to delay the onset of amyloid toxicity. HSF-1 mediated stimulation of mitochondrial β-oxidation is required to fully protect against amyloid-mediated paralysis. Previous studies and our work showed that the activities of H3K4me3-modifiers, HSF-1 and FAT-7 in early life are directly linked to their protective effects later in life. It remains to be tested precisely how changes in lipid homeostasis protect against proteotoxicity. Supplementary Material supplementary Table 1 [305]NIHMS2026991-supplement-supplementary_Table_1.xlsx^ (25.7KB, xlsx) supplementary Table 2 [306]NIHMS2026991-supplement-supplementary_Table_2.xlsx^ (711.1KB, xlsx) supplement reporting summary [307]NIHMS2026991-supplement-supplement_reporting_summary.pdf^ (2.3MB, pdf) supplementary Table 3 [308]NIHMS2026991-supplement-supplementary_Table_3.xlsx^ (521.1KB, xlsx) Source Data Extended Data Fig. 1a [309]NIHMS2026991-supplement-Source_Data_Extended_Data_Fig__1a.xlsx^ (9.7KB, xlsx) Source Data Extended Data Fig. 1b [310]NIHMS2026991-supplement-Source_Data_Extended_Data_Fig__1b.tif^ (5.8MB, tif) supplementary Table 4 [311]NIHMS2026991-supplement-supplementary_Table_4.tif^ (13.3MB, tif) Source Data Extended Data Fig. 1c [312]NIHMS2026991-supplement-Source_Data_Extended_Data_Fig__1c.tif^ (1.6MB, tif) Acknowledgements