Volatile diacetyl exposure induces rapid expression of nutrient-modulated DHAP-Glycerol shunt pathway.

a, Experimental setup for measuring gene expression effects of diacetyl exposure in food-deprived worms, comparing control food deprivation (no volatile) to food deprivation in the presence of volatile diacetyl (1% solution in ethanol, pipetted on plate lid). Three biological replicates collected for RNA-seq over a ninety-minute time course. b, Scatter plot showing differential gene expression between control and diacetyl groups at 30 minutes of exposure. c, Heatmap showing differential gene expression between control and diacetyl groups at 30 minutes of exposure, highlighting genes involved in glycerol production. d, Schematic depicting DHAP-glycerol shunt pathway, connecting glycolysis to synthesis of glycerol and glycerolipids. Highlighted in green are enzymes upregulated by diacetyl. e, Representative images of gpdh-1 transcriptional reporter worms raised on NGM, sorbitol or glucose supplemented plates. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 1 - Source data e. f, Quantification of reporter induction in experiment from (e). Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, one-way ANOVA followed by Tukey’s post hoc test with ***P = 0.0005 NGM vs. sorbitol, ****P < 0.0001 sorbitol vs. glucose. g, Expression levels of shunt genes over four hours of food deprivation followed by one hour of refeeding. Measured by qPCR, normalised to 0h time point for each gene. Error bars represent means ± SD, n = 3. h, DHAP levels in fed and food deprived worms (2h and 4h starvation), normalised to fed condition (n = 10). Error bars represent means ± SD, Kruskal-Wallis test followed by Dunn’s post hoc test with **P = 0.0046, ns: not significant. i, gpdh-1 expression levels at control osmolarity (NGM) or 200 mM NaCl (3 hours), in the presence or absence of food (n = 3). Measured by qPCR, normalised to control treatment for each food condition. Error bars represent means ± SD, ordinary two-way ANOVA followed by Fisher’s LSD test with ****P < 0.0001, **P = 0.0027. j, gpdh-1 expression levels following 30 minutes of diacetyl exposure, in the presence or absence of food (n = 6). Measured by qPCR, normalised to control treatment for each food condition. Error bars represent means ± SD, ordinary two-way ANOVA followed by Sidak’s post hoc test; *P = 0.0447 control vs diacetyl (+food), ****P < 0.0001 control vs diacetyl (-food), *P = 0.0447 +food vs -food (diacetyl).

Diacetyl triggers metabolic shift overlapping with hyperosmotic stress response.

a, Volcano plot comparing metabolomic effects of diacetyl exposure (one hour) to control food deprivation. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as phosphatidylglycerols (PGs). b, Schematic depicting synthesis of PGs from phosphatidic acid (PA) precursor lipids and Gro3P. c, Volcano plot comparing metabolomic effects of hyperosmotic stress (three hours, 200 mM NaCl plates with food) to control NGM plates. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as PGs. d, Heatmap comparing levels of all phosphatidylglycerols (PGs) in fed condition, hyperosmotic stress (three hours, 200 mM NaCl), one hour food deprivation (control) and one hour diacetyl exposure during food deprivation (n = 10). e, Overlap between metabolites significantly altered (FDR <0.05, Welch’s t-test) by hyperosmotic stress (three hours, 200 mM NaCl plates with food) and diacetyl exposure (one hour, in the absence of food). Significance determined by hypergeometric test; increased metabolites x = 31, N = 472, p < 0.008; decreased metabolites x = 30, N = 472, p < 2.73e-06. f, Worm glycerol levels following two hours of diacetyl exposure, normalised by protein content. Error bars represent means ± SD, Welch’s t-test with **P = 0.0065; n =8. g, Effects of diacetyl pre-exposure on acute hyperosmotic stress resistance. Worms were pre-exposed to diacetyl for one hour (in the absence of food), before thrashing was recorded on hyperosmotic stress plates (500 mM NaCl). Data is pooled from three biological replicates; n = 71 worms in the control group; n = 78 in the diacetyl group. Error bars represent means ± SD, Mann-Whitney test with ****P < 0.0001.

Prolonged diacetyl exposure in the absence of food boosts FMO-2 dependent responses.

a, Experimental setup for measuring gene expression effects of diacetyl exposure in food-deprived worms, comparing control food deprivation (no volatile) to food deprivation in the presence of volatile diacetyl (1% solution in ethanol, pipetted on plate lid). Three biological replicates collected for RNA-seq over a ninety-minute time course. b, Scatter plot showing differential gene expression between control and diacetyl groups at 90 minutes of exposure. c, Heatmap showing differential gene expression between control and diacetyl groups at 90 minutes of exposure. d, Representative images of fmo-2 transcriptional reporter induction over six hours of food deprivation, in the presence or absence of diacetyl. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 3 - Source data d. e, Representative images of fmo-2 reporter worms treated with ivermectin for one hour, followed by diacetyl exposure in the presence or absence of food for three hours. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 3 - Source data e. f, Representative images of fmo-2 reporter worms exposed to no volatile (control), ethanol or diacetyl dilutions in ethanol, over three hours of food deprivation. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 3 - Source data f. g, Quantification of reporter induction in experiment from (f). Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA test followed by Dunnett’s T3 post hoc test, ****P < 0.0001 vs no volatile, **P = 0.0035 vs 0.1% diacetyl, ****P < 0.0001 vs 1% diacetyl, ns: not significant. h - i, Individual metabolite levels in fed and food deprived worms, starved either in control conditions or during diacetyl exposure for one hour (n = 10). Log2 normalised levels on vertical axes. h, Methionine levels, Welch’s t-test with ***P = 0.0001. i, Tryptophan levels, Welch’s t-test with ***P = 0.0001. j, Survival (day 3 of adulthood) of worms exposed for three hours to diacetyl in the presence or absence of food (day 1) and heat-shocked on day 2. Error bars represent means ± SD (n = 3), ordinary one-way ANOVA followed by Sidak’s post hoc test with ****P < 0.0001 fed vs starved (ctrl), **P = 0.0015 ctrl vs diacetyl (starved), ns: not significant. k, Representative images of worms fed with GFP expressing bacteria following a three-hour exposure to diacetyl in the presence or absence of food. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 3 - Source data k. l, Quantification of GFP intensity in experiment from (k). Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA test followed by Dunnett’s T3 post hoc test, ****P < 0.0001, ns: not significant.

Diacetyl and hyperosmotic stress induce fmo-2 through DHAP-Glycerol shunt induction.

a, Schematic of gene expression responses to diacetyl exposure, detected from transcriptomics time course. b, Representative images of fmo-2 transcriptional reporter induction by diacetyl exposure in the absence of food (three hours), in worms raised on non-targeting Luciferase (LUC), combined RNAi targeting DHAP shunt components gpdh-1 and gpdh-2 or pgph-2 and pgph-3. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 4 - Source data b. c, Quantification of reporter induction in experiment from (b). Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, Ordinary one-way ANOVA followed by Dunnett’s T3 post hoc test with ****P < 0.0001, ns: not significant. d, Representative images of fmo-2 transcriptional reporter induction by three hours of hyperosmotic stress (Fed, 200 mM NaCl), food deprivation (Starved, NGM), hyperosmotic stress combined with food deprivation (Starved, 200 mM NaCl) and hyperosmotic stress pretreatment in the presence of food followed by food deprivation at control osmolarity (Pretreated 200 mM NaCl + Starved NGM). Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 4 - Source data d. e, Quantification of reporter induction in experiment from (d). Data is pooled from all three replicates (n = 45 worms per condition). Log 10 y-axis. Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA tests followed by Dunnett’s T3 post hoc test with ****P < 0.0001, ns: not significant. f, Representative images of fmo-2 transcriptional reporter induction by hyperosmotic stress pretreatment (200 mM NaCl) in the presence of food followed by food deprivation at control osmolarity, in worms raised on non-targeting Luciferase (LUC), combined RNAi targeting DHAP shunt components gpdh-1 and gpdh-2 or pgph-2 and pgph-3. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 4 - Source data f. g, Quantification of reporter induction in experiment from (f). Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA tests followed by Dunnett’s T3 post hoc test with ****P < 0.0001. h, Survival (day 3 of adulthood) of worms exposed on day 1 to 200 mM NaCl in the presence of food, followed by three hours at control osmolarity in the presence (fed + NaCl pretreated) or absence of food (starved + NaCl pretreated). Worms heat-shocked on day two. Error bars represent means ± SD (n = 3), ordinary one-way ANOVA followed by Sidak’s post hoc test with **P = 0.0077, ns: not significant. i, Heatmap comparing levels of ribonucleoside triphosphates in fed conditions and food deprivation (Starve, one or three hours) in the presence or absence (Control) of diacetyl exposure (n = 10). j, Pyruvic acid levels across samples in fed worms, starved worms in control conditions (ctrl), starved worms exposed to diacetyl (1 and 3 hour timepoints) (n = 10). Log2 normalised levels on vertical axes. Welch’s t-test with ***P = 0.0003 (1h) and ***P = 0.0009 (3h). k, ATP to ADP ratios of LC-MS peak area measurements (Arbitrary Units) detected across samples in fed worms, starved worms in control conditions, starved worms exposed to diacetyl (1 and 3 hour timepoints) (n = 10). Brown-Forsythe and Welch test followed by Dunnett’s T3 post hoc tests with ***P = 0.0001. l, Representative images of fmo-2 transcriptional reporter induction by diacetyl exposure in the absence of food (three hours), in worms raised on non-targeting Luciferase (LUC) or RNAi targeting hlh-30. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 4 - Source data l. m, gpdh-1 expression levels following 30 minutes of diacetyl exposure, in worms raised on non-targeting Luciferase (LUC) or RNAi targeting hlh-30. Measured by qPCR, normalised to LUC control. Error bars represent means ± SD (n = 3), ordinary two-way ANOVA followed by Sidak’s post hoc test; **P = 0.0013 (LUC), ****P < 0.0001 (hlh-30).

gpdh-1 induction and metabolic effects of diacetyl depend on MDT-15.

a, gpdh-1 expression levels following 30 minutes of diacetyl exposure, in worms raised on non-targeting Luciferase (LUC) RNAi targeting nhr-49 or mdt-15. Measured by qPCR, normalised to LUC control. Error bars represent means ± SD, ordinary two-way ANOVA followed by Sidak’s post hoc test with ***P = 0.0003, ns: not significant. n = 4. b, Effects of diacetyl pre-exposure on acute hyperosmotic stress resistance. Worms raised on non-targeting Luciferase (LUC) RNAi targeting nhr-49 or mdt-15 were pre-exposed to diacetyl for one hour (in the absence of food), before thrashing was recorded on hyperosmotic stress plates (500 mM NaCl). Data is pooled from three biological replicates; n = 64 worms (LUC, ctrl), n = 51 (LUC, diacetyl), n = 55 (nhr-49, ctrl), n = 50 (nhr-49, diacetyl), n = 60 (mdt-15, ctrl), n = 55 (mdt-15, diacetyl). Error bars represent means ± SD, ordinary one-way ANOVA followed by Sidak’s post hoc test with ****P < 0.0001 (LUC), *P = 0.03 (nhr-49), ns: not significant. c - d, Volcano plots comparing metabolomic effects of diacetyl exposure (one hour) to control food deprivation, for worms raised on Luciferase (LUC) RNAi (c) or mdt-15 RNAi (d). Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as phosphatidylglycerols (PGs). e, Representative images of gpdh-1 transcriptional reporter worms raised on Luciferase (LUC), nhr-49 and mdt-15 targeting RNAi and exposed to 200 mM NaCl for three hours. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 5 - Source data e. f, Representative images of worms raised on Luciferase (LUC), nhr-49 and mdt-15 targeting RNAi, either at control osmolarity or in the presence of 200 mM NaCl. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 5 - Source data f. g, Representative survival of worms raised on Luciferase (LUC), nhr-49 and mdt-15 targeting RNAi and transferred on day 1 of adulthood to RNAi plates supplemented with 400 mM NaCl. Results shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 5 - Source data g. h, Model summarising acute effects of volatile diacetyl exposure across the levels of gene expression, metabolism and physiological stress responses. Food deprived worms exposed to diacetyl induce expression of gpdh-1 (through MDT-15) and other enzymes in the DHAP-glycerol shunt. This leads to accumulation of glycerol, PGs and protection from acute hyperosmotic stress. The metabolic shift induced by the shunt, in combination with food deprivation (possibly sensed through HLH-30), then triggers a state of enhanced starvation characterised by depleted NTP levels, upregulated fmo-2 expression and starvation-related phenotypes.

a, PCA plot of diacetyl transcriptomic time course from Figure 1a. b - c, Scatter plots comparing gene expression between control and diacetyl groups at 5 (b) and 15 (c) minutes of exposure. d, Predicted tissue expression scores for genes upregulated by diacetyl, in the top ten enriched tissues. e, gpdh-1 expression levels over twenty-four hours of food deprivation. Measured by qPCR, normalised to 0h time point. Error bars represent means ± SD, n = 3. f, Heatmap comparing levels of glycolytic intermediates detected from metabolomics in fed condition and after two or four hours of food deprivation (n = 10). g, Representative images of gpdh-1 transcriptional reporter worms exposed to hyperosmotic stress (200 mM NaCl), in the presence or absence of food. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 1 - Source data S1g. h, gpdh-1 expression levels at control osmolarity (NGM) or 200 mM NaCl (3 hours), in the presence or absence of food (n = 3). Measured by qPCR, normalised to NGM fed levels.

a, Volcano plot comparing metabolomic effects of diacetyl exposure (one hour) to control food deprivation. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as phosphatidic acids (PAs). b, Volcano plot comparing metabolomic effects of hyperosmotic stress (three hours, 200 mM NaCl plates with food) to control NGM plates. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as PAs. c, Effects of diacetyl pre-exposure on thrashing rates. Worms were pre-exposed to diacetyl for one hour (in the absence of food), before thrashing was recorded on unseeded NGM plates. Data is pooled from two biological replicates; n = 30 worms in the control group; n = 27 in the diacetyl group. Error bars represent means ± SD, unpaired t-test with ns: not significant.

a, Quantification of fmo-2 reporter induction in experiment from Figure 3d. Data is pooled from all three replicates in Figure 3 - Source data d (n = 45 worms per condition). Error bars represent means ± SD, Kruskal-Wallis test followed by Dunn’s post hoc test with *P = 0.0347 (30 min), ****P < 0.0001, ***P = 0.0002 (6h). b, Quantification from (a) with diacetyl fmo-2 reporter induction normalised to the control starvation for each timepoint. Error bars represent means ± SD, RM-one way ANOVA followed by Tukey’s post hoc test with **P = 0.0074, ns: not significant. c, Representative images of fmo-2 reporter worms treated with ivermectin for one hour, followed by diacetyl exposure in the presence or absence of food for three hours. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 3 - Source data e. d - g, fmo-2 expression levels in worms starved either in control conditions or in the presence of diacetyl for three hours. Detected by qPCR. Normalised to control for each genotype. d, fmo-2 expression levels in WT, odr-10 (ky225) and sri-14 (ok2865). Error bars represent means ± SD (n = 4), two-way ANOVA followed by Tukey’s post hoc test with **P = 0.0059 ctrl vs diacetyl (WT), **P = 0.0021 ctrl vs diacetyl (odr-10), ****P < 0.0001 ctrl vs diacetyl (sri-14). e, fmo-2 expression levels in WT, tph-1 (mg280) and cat-2 (n4547). Error bars represent means ± SD (n = 3), two-way ANOVA followed by Tukey’s post hoc test with ***P = 0.0003 ctrl vs diacetyl (tph-1), **P = 0.0017 ctrl vs diacetyl (cat-2), *P = 0.0223 WT vs tph-1 (diacetyl). f, fmo-2 expression levels in WT, unc-13 (e450) and unc-31 (e928). Error bars represent means ± SD (n = 4), two-way ANOVA Tukey’s post hoc test with *P = 0.0211 ctrl vs diacetyl (WT), *P = 0.0244 ctrl vs diacetyl (unc-13), ****P < 0.0001 ctrl vs diacetyl (unc-31), *P = 0.0375 WT vs unc-31 (diacetyl). g, fmo-2 expression levels in WT, daf-12 (sa204) and daf-12; daf-19 (sa204; m86). Error bars represent means ± SD (n = 3), two-way ANOVA Tukey’s post hoc test with *P = 0.041 ctrl vs diacetyl (WT), ***P = 0.0005 ctrl vs diacetyl (daf-12), ****P < 0.0001 ctrl vs diacetyl (daf-12; daf-19), ****P < 0.0001 daf-12 vs daf-12; daf-19 (diacetyl). h, Survival (day 3 of adulthood) of worms exposed for three hours to diacetyl in the absence of food (day 1) and heat-shocked on day 2. Error bars represent means ± SD (n = 3), ordinary one-way ANOVA followed by Sidak’s post hoc test with ***P = 0.004. i, Quantification of GFP intensity in worms fed with GFP expressing bacteria following a three-hour exposure to diacetyl in the absence of food. Data is pooled from three replicates in Figure 3 - Source data S3i (n = 45 worms per condition). Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA test followed by Dunnett’s T3 post hoc test, ****P < 0.0001, **P = 0.0011.

a, Overlap between metabolites significantly altered (FDR <0.05, Welch’s t-test) by diacetyl, after one or three hours exposure in the absence of food. Significance determined by hypergeometric test; increased metabolites x = 27, N = 472, p < 2.15e-16; decreased metabolites x = 37, N = 472, p < 3.8e-26. b, Volcano plot comparing metabolomic effects of diacetyl exposure (three hours) to control food deprivation. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as phosphatidylglycerols (PGs). c, Heatmap comparing levels of all phosphatidylglycerols (PGs) in fed condition, control food deprivation (control, one and three hours) and diacetyl exposure during food deprivation (diacetyl, one and three hours) (n = 10). d - f, Individual metabolite levels in fed and food deprived worms, starved either in control (ctrl) conditions or with diacetyl (dia) exposure. One hour and three hour time points. Log2 normalised levels on vertical axes (n = 10). d, CTP levels, Welch’s t-test with *P = 0.03. e, UTP levels, Welch’s t-test with **P = 0.006. f, ATP levels, Welch’s t-test with *P = 0.02. g, Effects of diacetyl pre-exposure on acute hyperosmotic stress resistance. Worms raised on non-targeting Luciferase (LUC) RNAi targeting hlh-30 were pre-exposed to diacetyl for one hour (in the absence of food), before thrashing was recorded on hyperosmotic stress plates (500 mM NaCl). Data is pooled from three biological replicates; n = 64 worms (LUC, ctrl), n = 72 (LUC, diacetyl), n = 61 (hlh-30, ctrl), n = 64 (hlh-30, diacetyl). Error bars represent means ± SD, ordinary one-way ANOVA followed by Sidak’s post hoc test with ***P = 0.0001 (LUC), ****P < 0.0001 (nhr-49).

a, Representative images of fmo-2 transcriptional reporter induction by diacetyl exposure in the absence of food (three hours), in worms raised on non-targeting Luciferase (LUC) or RNAi targeting nhr-49 or mdt-15. Images shown are from one experiment out of three biological replicates. Additional replicates are available in Figure 5 - Source data S5a. b, Volcano plot comparing metabolomic effects of diacetyl exposure (one hour) to control food deprivation, in worms raised on non-targeting Luciferase (LUC) RNAi. Metabolites significantly altered in red (FDR <0.05, Welch’s t-test). Highlighted are metabolites classified as phosphatidic acids (PAs). c, Quantification of gpdh-1 reporter induction in experiment from Figure 5e. Data is pooled from all three replicates (n = 45 worms per condition). Error bars represent means ± SD, Brown-Forsythe and Welch ANOVA tests followed by Dunnett’s T3 post hoc test with ****P < 0.0001 NGM vs NaCl (LUC), ****P < 0.0001 NGM vs NaCl (nhr-49), ***P = 0.0002 LUC vs nhr-49 (NaCl), ****P < 0.0001 LUC vs mdt-15 (NaCl), ns: not significant. d, Quantification worm size in experiment from Figure 5f. Normalised to size of worms in control osmolarity for each RNAi condition. Ordinary one-way ANOVA followed by Dunnett’s post hoc test with ****P < 0.0001.