Metabolic Signatures of Life Span Regulated by Mating, Sex Peptide, and Mifepristone/RU486 in Female Drosophila melanogaster

Abstract

Mating and transfer of male sex peptide (SP), or transgenic expression of SP, causes inflammation and decreased life span in female Drosophila. Mifepristone rescues these effects, yielding dramatic increases in life span. Here targeted metabolomics data were integrated with further analysis of extant transcriptomic data. Each of 7 genes positively correlated with life span were expressed in the brain or eye and involved regulation of gene expression and signaling. Genes negatively correlated with life span were preferentially expressed in midgut and involved protein degradation, amino acid metabolism, and immune response. Across all conditions, life span was positively correlated with muscle breakdown product 1/3-methylhistidine and purine breakdown product urate, and negatively correlated with tryptophan breakdown product kynurenic acid, suggesting a SP-induced shift from somatic maintenance/turnover pathways to the costly production of energy and lipids from dietary amino acids. Some limited overlap was observed between genes regulated by mifepristone and genes known to be regulated by ecdysone; however, mifepristone was unable to compete with ecdysone for activation of an ecdysone-responsive transgenic reporter. In contrast, genes regulated by mifepristone were highly enriched for genes regulated by juvenile hormone (JH), and mifepristone rescued the negative effect of JH analog methoprene on life span in adult virgin females. The data indicate that mifepristone increases life span and decreases inflammation in mated females by antagonizing JH signaling downstream of male SP. Finally, mifepristone increased life span of mated, but not unmated, Caenorhabditis elegans, in 2 of 3 trials, suggesting possible evolutionary conservation of mifepristone mechanisms.

Female Drosophila exhibit a variety of physiological and behavioral changes upon mating. These changes include increased egg production, decreased receptivity to re-mating, increased expression of innate immune response genes, and decreased life span (1). In addition, mating shifts the female’s food preference from carbohydrate-rich food to protein-rich food (2, 3). Many of these changes have been shown to be caused wholly or in part by the male seminal fluid hormone sex peptide (SP), including decreased receptivity to re-mating, increased oogenesis, and increased immune gene expression (4–8). The role of the male SP gene in regulating female life span has been relatively less explored. Wigby and Chapman used RNAi to knock down the SP gene in males and reported females mated to SP knock-down males lived +9% longer relative to females mated to control males, in one assay, and no difference in a second assay (9). We have recently used a heteroallelic SP mutation combination to create SP-null males and reported replicable increases in the median life span of mated females of >+100%, relative to females mated to control males (10). Moreover, transgenic expression of SP in otherwise wild-type virgin females also decreased median life span. Strikingly, feeding female flies the synthetic steroid mifepristone at relatively high concentrations (160 or 200 μg/mL, or ~400 μM), after mating or transgenic SP expression, blocks the effects of mating and SP on median life span, progeny production, innate immune gene expression, and endogenous microbial load (10, 11), resulting in dramatic increases in median life span (typically > +50%). Notably, life span increases were also observed with mifepristone concentrations as low as 10 μg/mL (~23 μM) (11).

Juvenile hormone (JH) is an acyclic sesquiterpenoid hormone synthesized by the corpus allatum (CA) gland, located adjacent to the brain (12), as well as by the midgut (13). The genes germ cell-expressed (Gce) and methoprene-tolerant (Met) encode paralogous bHLH-PAS proteins that function as JH receptors. Upon binding to JH, these receptor proteins act with transcription factor Taiman (Tai) to activate expression of JH-responsive genes. Mating and SP cause increased production of JH, which in turn induces a structural and metabolic remodeling of the midgut. Midgut remodeling includes increased cell proliferation and increased organ size, as well as sterol regulatory element-binding protein (SREBP)-regulated increases in lipid metabolism (14). Gut remodeling is thought to enable increased nutrient absorption and lipid biosynthesis to support increased egg production. Additional hormones have been shown to act downstream of mating to regulate increased egg production, including octopamine (15) and ecdysone (16, 17), and ecdysone signaling has been reported to have sex-specific effects on adult life span (18, 19). Previous studies utilizing sterile flies have reported that egg production is not required for the negative effect of mating on life span (20), or for the effects of JH signaling on life span (21). Our previous studies show that the negative effect of mating is reproducibly small for certain strains and genotypes, and reproducibly large for other strains and genotypes, and that the positive effect of mifepristone is proportional to the negative effect of mating, consistent with a rescue or block of the mating effect (10, 11). Ablation of the microbiome in adult females using high concentrations of the antibiotic doxycycline reduced the negative effect of mating, suggesting that the variability in response to mating across strains might be due to variations in the microbiome (10). In addition, these studies showed that the gene-switch transcription factor is not required. Here, genotypes with reproducibly large response were used to enable investigation of mechanisms.

When life span in Drosophila is increased by feeding a drug, one important consideration is whether the drug might reduce food intake and thereby create a dietary restriction (DR) effect. Food intake was previously assayed at day 12 of drug treatment, which is when the survival curves for mated females and mated females plus mifepristone typically begin to diverge. Food intake was measured using the dye-ingestion assay (11), and the more sensitive capillary-feeding (CAFE) assay (10), and no decrease in food intake due to mifepristone was observed. Indeed, with both assays, food intake was often significantly increased in the mifepristone-treated females, an observation we attribute to the generally greater health of the mifepristone-treated females relative to controls; this result is consistent with the idea that food intake and progeny production can sometimes be de-coupled. Yamada et al. (22) analyzed food intake using 3 different assays, proboscis extension response, CAFE, and radioisotope uptake assay and found that 86 μg/mL mifepristone had little or no effect in adult males. Taken together, these studies indicate that mifepristone does not increase life span through a DR effect involving decreased food intake. Yamada et al. also reported that in male Drosophila, mifepristone had an aversive effect and decreased life span under extremely low nutrient conditions, and we also observe some negative effects of mifepristone with Caenorhabditis elegans, as described below.

In this study, we took advantage of the dramatic life span effects of mating and mifepristone to identify the genes and metabolites most closely correlated with life span. In addition, drug competition experiments were used to investigate whether mifepristone might inhibit either ecdysone signaling or JH signaling pathways. Because mifepristone acts as an antagonist of progesterone receptor and glucocorticoid type II receptor in humans (23), one possibility was that mifepristone might act to antagonize activity of the endogenous steroid hormone ecdysone found in Drosophila. However, the results do not provide strong support for that hypothesis. Instead, the data indicate that mifepristone directly or indirectly antagonizes JH signaling downstream of SP. Integrating the transcriptomic and metabolomics data suggests that mating and SP induce a metabolic shift from glycolysis, fatty acid (FA) oxidation, and somatic maintenance/turnover pathways to costly amino acid (AA) metabolism and lipid production. This metabolic shift sensitizes the female to bacterial toxicity, resulting in decreased life span.

Material and Methods

Drosophila melanogaster were cultured on a standard agar/dextrose/corn meal/yeast media (56) at 25 °C, and adult flies were passaged to fresh media every other day. Drosophila strains are as previously described (10, 11), and additional strains were obtained from Bloomington Drosophila Stock Center. Developmental toxicity experiments used the w[1118] control strain. To generate female flies for life-span experiments, males of the w[1118] strain were crossed to virgin females of the y; Elav-GS strain and the hybrid female progeny were collected as previously described (10), referred to here as the “standard cross.” Transgenic expression of SP strain and the control strain were as previously described; the control was w[1118]; UAS-SP/+, and the over-expression strain was w[1118]; UAS-SP/dsx-GAL4 (10). Drugs were administered as previously described, by applying 100 μL of 10× stock solution in water, or 50 μl of 20× stock solution in ethanol, to the vial and calculating absorption into the top ~1 mL of media, as determined by dye-absorption controls (56, 57); detailed protocols for preparation of drug vials are provided at https://dornsife.usc.edu/towerlab/protocols/. The same drug administration procedure was adapted to the EX-Q assay for food consumption (31), to best match conditions of the life-span assay. Targeted metabolomics assays (~120 metabolites) were conducted essentially as described (50). Intestinal barrier integrity (“SMURF”) assay was conducted essentially as described (33); with modification that for assay of individual flies, SMURF phenotype was scored every other day, as well as in the recently deceased flies. Caenorhabditis elegans mating and life-span assays were performed using wild-type N2 Bristol strain, essentially as previously described (58). Gene expression data are as previously described (GEO accession GSE64474), generated from the progeny of a cross of 2 unrelated transgenic strains w[1118]; p53[B6] × w[1118]; rtTA(3)E2 (11). All statistical analyses include a Bonferroni correction for multiple testing, and the p value for significance is presented in figure legends. Additional details of strains and procedures are presented in Supplementary Materials.

Results

Gene Expression Changes in Females in Response to Mating and Mifepristone

Gene expression changes were previously analyzed using whole-body mRNA from virgin females, mated females, and mated females fed mifepristone. The genotype used was one that was highly responsive to mating and did not contain the gene-switch transcription factor (“Material and Methods”) (11). Females were mated for 2 days, the males were removed, and then half the mated females were moved to food supplemented with mifepristone. Whole-body RNA was isolated at day 12 post-mating (14 days of age); contemporaneous assay showed a +68% increase in the median life span of mated females treated with mifepristone (11). Gene expression changes caused by mating were determined by comparing virgin females to mated females. Gene expression changes caused by mifepristone were determined by comparing mated females to mated females plus mifepristone. Here, the gene lists were analyzed for possible metabolic pathway enrichments and tissue distribution enrichments.

Mifepristone Gene Expression Changes Compared to Previous Studies of Ecdysone and JH

Previous studies have implicated both ecdysone and JH in regulating the physiological changes in the female caused by mating, as discussed above. Here, the 3645 genes regulated by mating and the 118 genes regulated by mifepristone were compared to previous reports of genes regulated by ecdysone and JH (Supplementary Table S1; details of analysis presented in Supplementary Materials and Methods). The 3645 genes regulated by mating were not enriched for genes regulated by ecdysone, but were enriched by 2.2-fold (p = 3.1E-53) for genes regulated by JH. The 118 genes regulated by mifepristone were enriched 1.7-fold (p = .0056, not significant after correction for multiple testing) for genes regulated by ecdysone and enriched 7-fold (p = 2.8E-19) for genes regulated by JH. The direction of change in expression caused by mifepristone was opposite the direction of change caused by JH for the majority of the common genes. Taken together, these data support the idea that mifepristone may antagonize JH signaling.

Identification and Tissue-Distribution of Genes Negatively and Positively Correlated With Life Span

Mating and SP cause decreased life span and inflammation in the adult female fly, and feeding these females mifepristone reverses these effects. Genes upregulated by mating and downregulated by mifepristone are, therefore, negatively correlated with life span. A total of 50 genes were negatively correlated with life span (Supplementary Table S2; Figure 1B). This list contained 9 proteases, including 5 Jonah-family proteases, as well as 2 AA transporters, including the glycine transporter type II/GABA transporter gene CG1698. Also present were 5 genes involved in innate immune response and 11 genes involved in eggshell production. Finally, the list also included the prolyl hydroxylase gene PH4alphaPV, the pheromone response gene Obp99b, and gene Rh5 which encodes a retinal photopigment that absorbs blue light. Pathway analysis showed no significant enrichments. However, analysis of tissue distributions of gene expression showed a significant enrichment (2.4-fold) for genes expressed in the midgut (Table 1); 24 out of the 50 negatively correlated genes are expressed in midgut (Figure 1B). These 50 negatively correlated genes are also significantly under-represented for genes expressed in the eye and the brain (Table 1).

Figure 1.

Transcriptional and metabolic changes caused by mating, sex peptide (SP), and mifepristone. Genes are indicated in italic font. Genes in blue italic font are downregulated by mating and upregulated by mifepristone, and are therefore positively correlated with life span. Genes indicated in red italic font are upregulated by mating and downregulated by mifepristone, and are therefore negatively correlated with life span. Metabolites are indicated in roman font. Metabolites in blue roman font are downregulated by mating and/or transgenic SP, and upregulated by mifepristone, and are therefore positively correlated with life span. Metabolites indicated in red roman font are upregulated by mating and/or transgenic SP and downregulated by mifepristone, and are therefore negatively correlated with life span. (A) Mating outline. During mating, the male introduces SP into the female reproductive tract. The SP activates JH signaling, and feeding the mated females mifepristone antagonizes the effects of SP and JH. (B) Gene expression changes. All 7 positively correlated genes are expressed in brain and/or eye, as well as several negatively correlated genes. Numerous negatively correlated genes are expressed in gut, as indicated. (C) Metabolite changes. Metabolite changes fall into several distinct categories, as indicated by rounded boxes. Positively correlated categories include glycolysis, signaling and memory, purine turnover, muscle turnover, fatty acid (FA) oxidation, and the TCA-ETC inhibitor malonate. Negatively correlated categories include AA metabolism, lipid metabolism, RNA methylation, microbial metabolism, the FA-shuttle inhibitor deoxycarnitine, and the TCA cycle intermediate oxaloacetate. *The analysis does not distinguish between glucose-6P, glucose-1P, fructose-1P, and fructose-6P. (D) Immune function changes. Mating and SP cause decreased resistance, as indicated by increased microbial load and increased immune response gene expression (genes indicated with italic red font), and these changes are reversed by mifepristone. Mating and SP also cause decreased tolerance, as indicated by decreased median life span, and this is also reversed by mifepristone. (E) Life span changes. The typical changes in mortality curves caused by mating, SP, and mifepristone are diagrammed. Mating (or transgenic expression of SP) causes decreased female median life span, and feeding the females mifepristone partly or completely reverses this effect. Lesser or no change in maximum life span is observed, meaning that mating, SP, and mifepristone are affecting the “rectangularization” of the curve, and that mifepristone favors compression of mortality to late ages. The arrow indicates the day when the gene expression and metabolomics assays were conducted.

Table 1.

Tissue Distribution of Genes Negatively and Positively Correlated With Life Span as Regulated by Mating and Mifepristone

Tissue	Total Genes Drosophila	# Genes in Tissue	# Genes Our List	Overlap	Hyper-Geometric p	Result
Mating up, drug down
Midgut	13 961	2776	50	24	.0000	>2.41
Ovary	13 961	5011	50	12	.0508	<1.50
Head	13 961	3523	50	7	.0416	<1.80
Eye	13 961	3356	50	2	.0001	<6.01
Brain	13 961	4066	50	1	.0000	<14.56
Mating down, drug up
Midgut	13 961	2776	7	1	.5799	<1.39
Ovary	13 961	5011	7	4	.2148	>1.59
Head	13 961	3523	7	4	.0727	>2.26
Eye	13 961	3356	7	2	.5308	>1.19
Brain	13 961	4066	7	5	.0253	>2.45

Note: The p value for significance with 5 comparisons is .05/5 = .01.

A total of 7 genes were positively correlated with life span, in that they were downregulated by mating and upregulated by mifepristone (Supplementary Table S3; Figure 1B). These included genes encoding the chromatin-binding nuclear pore subunit Elys, involved in regulation of chromatin condensation, gene expression, and nuclear size (24); the synaptic vesicle endocytosis factor tweek, involved in PI(4,5)P₂ signaling and regulating of NMJ synaptic bouton size (25); the translation and dosage-compensation regulator Unr; the Dopamine 2-like receptor Dop2R; the regulator of histone gene expression mxc; the predicted transmembrane receptor and sterol sensor Ptr; and the stress-inducible humoral factor TotC (11, 26). Pathway analysis showed no significant enrichments. Analysis of known tissue distributions of expression showed an enrichment (2.4-fold) for genes expressed in the brain (Table 1). However, this enrichment was not significant after accounting for multiple comparisons.

In summary, the 50 candidate negative regulators are preferentially expressed in the midgut and are underrepresented in the brain and eye. In contrast, the 7 candidate positive regulators show a trend for expression in the nervous system: 4 out of the 7 genes are most highly expressed in brain (tweek, Unr, Dop2R, and Ptr), and the remaining 3 (Elys, mxc, and TotC) show moderate expression in brain and/or eye (Figure 1B). Because the midgut is a known target for the action of JH, these results support the idea that mifepristone may antagonize JH signaling.

Metabolites Negatively Correlated With Life Span Are Enriched for AA Metabolism Pathways

The gene expression analysis indicates that mating is associated with downregulation of signaling genes in the brain and upregulation of proteases and AA metabolism genes in the midgut. Targeted metabolomics assays were conducted to ask what metabolic changes might correspond with these gene expression changes, and 120 metabolites were quantified. In the first experiment, the effect of mating on the metabolome profile was analyzed, using progeny of the standard cross (w[1118] × y; Elav-GS). Whole-body extracts were generated from virgin females, mated females, and mated females plus mifepristone, at day 12 post-mating (14 days of age), to match the transcriptomic data; contemporaneous assay of life span showed a +71% increase due to mifepristone (10). In the second experiment, the effect of transgenic expression of SP on the metabolome profile was analyzed (10). Extracts were generated from virgin female controls (w[1118]; UAS-SP/+), virgin females with transgenic SP expression (w[1118]; UAS-SP/dsx-GAL4), and virgin females with transgenic SP expression plus mifepristone, at day 12 of age. Contemporaneous assay of life span showed a +128% increase due to mifepristone (10). Because the mating experiment and the transgenic SP expression experiments were separately conducted and analyzed, the absolute levels of metabolites cannot be directly compared between the 2 experiments; however, comparing the patterns of metabolite increase and decrease is potentially informative.

Two metabolites, kynurenic acid and hydroxyproline/5-aminolevulinic acid, were upregulated by mating and downregulated by mifepristone and were therefore negatively correlated with life span (Supplementary Table S4; Figures 1C and 2E). Kynurenic acid is a breakdown product of tryptophan, and 5-aminolevulinic acid is a product of glycine metabolism. Hydroxyproline is a post-translational modification of proline catalyzed by prolyl hydroxylase enzyme, and this may be related to the prolyl hydroxylase gene PH4alphaPV mentioned above. Twenty-seven metabolites were upregulated by transgenic SP expression and downregulated by mifepristone, and therefore negatively correlated with life span (Supplementary Table S5; Figure 1C). These included 11 AAs, both non-essential and essential, as well as several AA metabolites, including kynurenic acid (Figure 2E) and GABA. The increase in GABA may be related to the increased expression of the gut GABA transporter CG1698 mentioned above. Also present was the TCA cycle intermediate oxaloacetate, 2 saturated FAs (margaric acid and myristic acid), and the phospholipid precursors 13-hydroxyoctadecadienoic acid (a polyunsaturated FA) and inositol. Increased RNA methylation (27) was indicated by the presence of 1-methyladenosine, 1-methylguanosine and N2,N2-dimethylguanosine. The inhibitor of the carnitine FA shuttle, deoxycarnitine (28), was increased, as was the likely microbial metabolite 3-hydroxybenzoic acid (29; Figure 1C).

Figure 2.

Effects of mifepristone on transgene expression, life span, and metabolite levels. (A) Effect of mifepristone on expression of the ecdysone-responsive reporter. Beta-galactosidase was quantified in extracts of young (6 d old) virgin female flies, after 6-d treatment with indicated drugs. Change in optical density (OD) for the chromogenic substrate (CPRG) per hour per micro gram of protein is plotted as the mean ± SD for 3 replicate extracts. Analysis of variance (ANOVA) is presented below the graph. All post hoc comparisons are unpaired, 2-sided t tests, and p values are presented above the bars, for comparison of ecdysone alone to ecdysone plus additional drug. A = alcohol vehicle; E 300 = 300 μg/mL ecdysone; R 200 = 200 μg/mL mifepristone (RU486); R 400 = 400 μg/mL mifepristone (RU486); P 200 = 200 μg/mL progesterone; P 400 = 400 μg/mL progesterone. The p value for significance with 4 comparisons is .05/4 = .0125. (B) Effect of methoprene and mifepristone on life span. Flies are virgin females. (−) = no drug; R = 200 μg/mL mifepristone (RU486); M 100 = 100 μg/mL methoprene; M 200 = 200 μg/mL methoprene. Data are plotted as percent survival versus time in days. Statistical comparisons are in the order: R to (−), M 100 to (−), M 100 + R to M 100, M 200 to (−), M 200 + R to M 200. The p value for significance with 5 comparisons is .05/5 = .01. COX proportional hazards analysis is presented in Supplementary Table S10. (C) EX-Q assay of food consumption. The concentration of blue dye in fly excrement is plotted as mean ± SD for 4 replicate samples of 10 flies each. A = alcohol vehicle; R = 200 μg/mL mifepristone (RU486); M = 200 μg/mL methoprene; and M+R = 200 μg/mL methoprene plus 200 μg/mL mifepristone (RU486). ANOVA is presented below the graph. The p value for significance with 4 comparisons is .05/4 = .0125. (D–F) Metabolite levels. Metabolite levels (log₂ abundance) are presented in box plots. V = virgins; M = mated, M+ = mated plus mifepristone. Co = control genotype for transgenic SP expression in virgins; SP = transgenic SP expression in virgins; SP+ = transgenic SP expression in virgins plus mifepristone. (D) 1/3-Methylhistidine. (E) Kynurenic acid. (F) Urate. (G) Survival of flies subjected to continuous SMURF assay. Flies were maintained at one fly per vial, and scored every other day for SMURF phenotype, including deceased flies. The survival curves are presented for the groups virgin, mated, and mated plus mifepristone (200 μg/mL). Live flies from the cohort were imaged (see inset), non-SMURF (left) and SMURF (right). The p value for significance with 2 comparisons is .05/2 = 0.025. COX proportional hazards analysis is presented in Supplementary Table S12.

Because the tryptophan breakdown product kynurenic acid was upregulated by both mating and transgenic SP expression and downregulated by mifepristone in each case, it is negatively correlated with life span under all conditions analyzed here (Figure 2E). In total, 27 metabolites were upregulated by mating and/or transgenic SP and downregulated by mifepristone, and were therefore negatively correlated with life span (Supplementary Tables S4 and S5; Figure 1C, indicated in red font). This group was significantly enriched for several AA metabolism pathways, and in particular, the aminoacyl-tRNA biosynthesis pathway (Supplementary Table S6). Because JH is known to induce gut remodeling and increase gut nutrient uptake and lipid metabolism, these results support the idea that mifepristone may antagonize JH signaling.

Metabolites Positively Correlated With Life Span Are Enriched for Purine Metabolism Pathway

Five metabolites were downregulated by mating and upregulated by mifepristone, and therefore positively correlated with life span (Supplementary Table S7). These included urate, a breakdown product of purines (Figure 2F), and 1/3-methylhistidine, a breakdown product of muscle protein (Figure 2D) (30). Also among these 5 metabolites was adipic acid, a product of omega oxidation of FAs, as well as carnitine and acetylcarnitine, which function in shuttling FAs into mitochondria (Figure 1C, indicated in blue font). Seventeen metabolites were downregulated by transgenic SP expression and upregulated by mifepristone, and therefore positively correlated with life span (Supplementary Table S8). This list included 6 purine pathway metabolites, including the purine breakdown product urate, 3 metabolites of the glycolysis pathway, and the TCA/ETC inhibitor malonate (Figure 1C, indicated in blue font). Two additional metabolites were positively correlated with life span, and are not shown in the summary figure. These are geranyl-PP, a precursor to sesquiterpenes, and the essential nutrient pantothenic acid (vitamin B5), which is required for synthesis of CoA, and therefore required for shuttling of FAs into mitochondria. Four metabolites involved in signaling and implicated in memory were identified, including UDP-GlcNAc, which is involved in modification of nuclear pores, and might be related to the Elys gene mentioned above (Figure 1C). Also identified was histamine, which regulates photoreception in the eye and might be related to the Rh5 gene mentioned above. Finally, the muscle breakdown product 1/3-methylhistidine was present. Because 1/3-methylhistidine and urate were downregulated by both mating and transgenic SP expression, and upregulated by mifepristone in each case, they were positively correlated with life span under all conditions analyzed here (Figure 2D and F). In total, 20 metabolites were downregulated by mating and/or transgenic SP and upregulated by mifepristone, and were therefore positively correlated with life span (Supplementary Tables S7 and S8; Figure 1C, indicated in blue font). This group was most significantly enriched for the purine metabolism pathway (Supplementary Table S9).

Mifepristone Does Not Compete With Ecdysone for Activation of an Ecdysone-Responsive Reporter

In mammals, mifepristone antagonizes the activity of progesterone and glucocorticoids by binding to the progesterone receptor and to the type II glucocorticoid receptor. It was therefore of interest to ask whether mifepristone might act in Drosophila to antagonize ecdysone signaling by competing for binding to the ecdysone receptor. To test this, an ecdysone-responsive transgenic reporter was used, where expression of beta-galactosidase is driven by a promoter containing the binding site for the ecdysone receptor (EcRE-lacZ). Feeding ecdysone to female flies containing the reporter produced a dose-dependent induction of beta-galactosidase activity, as expected, using 2 independent reporter strains, EcRE-lacZ[SS3] and EcRE-lacZ[SS4] (Supplementary Figure S1A and B). Mifepristone alone did not affect the activity of the reporters, nor did the structurally related steroid progesterone (Supplementary Figure S1C and D). Simultaneous feeding of mifepristone with ecdysone showed no inhibitory effect of mifepristone (Figure 2A; Supplementary Figure S1E), indicating that mifepristone does not directly compete with ecdysone for binding to the ecdysone receptor.

Feeding virgin female flies 200 μg/mL ecdysone throughout adulthood did not have a statistically significant effect on life span, in replicate experiments (Supplementary Figure S2). Simultaneous feeding of mifepristone with ecdysone increased life span relative to ecdysone alone by 15% in the first experiment and did not have a significant effect in the second experiment (Supplementary Figure S2). The +15% increase in life span is similar in magnitude to the effect of mifepristone alone on virgin life span, which was 0% in the first experiment, +11% in the second experiment, and has ranged from +0% to +20% in past experiments (10, 11). Taken together, these experiments suggest little if any interaction between dietary ecdysone and dietary mifepristone in affecting adult life span, at least at the one concentration of ecdysone tested.

The hormone octopamine is reported to act downstream of SP in the mated female to regulate increased egg laying and decreased receptivity to re-mating (15). Feeding virgin female flies with 7.5 mg/mL octopamine throughout adulthood caused a decrease in life span of −15% in the first experiment, and −3.8% (not statistically significant) in the second experiment (Supplementary Figure S2). Interestingly, feeding mifepristone plus octopamine caused a significant additional decrease in life span, by −9% in the first experiment and by −17% in the second experiment. These results indicate that when combined with dietary octopamine, dietary mifepristone negatively affects adult female life span.

Mifepristone Rescues the Negative Effect of Methoprene on Adult Female Life Span

Feeding virgin female Drosophila with the JH analog methoprene has been reported to recapitulate several effects of mating, including gut remodeling and decreased life span (14, 21). Here, adult virgin females were fed methoprene throughout adulthood, in attempt to recapitulate the effects of mating and mifepristone on life span (Figure 2B). Feeding methoprene at 100 μg/mL reduced median life span by −66%. Co-feeding 200 μg/mL mifepristone partially rescued this effect, and increased life span by +70%. Similarly, feeding methoprene at 200μg/mL reduced life span by −62%, and co-feeding mifepristone partially rescued this effect, increasing life span by +91%.

To rule out the possibility that mifepristone might rescue methoprene toxicity by decreasing food intake, and thereby decreasing the amount of methoprene ingested, the recently described excreta quantification (EX-Q) assay was employed (31). In the EX-Q assay, flies are fed non-absorbable blue dye in the media, and the total amount of blue dye excreted over 24 hours is quantified (Supplementary Figure S3). A test of various numbers of control flies confirmed the quantitative response of the assay (Supplementary Figure S3E). Virgin flies (progeny of the standard cross) were maintained on either control food, 200 μg/mL mifepristone, 200 μg/mL methoprene, or 200 μg/mL methoprene plus 200 μg/mL mifepristone, and assayed by EX-Q at day 12 and at day 16 (Supplementary Figure S3F). At day 12, neither mifepristone nor methoprene had a statistically significant effect on food intake (Figure 2C). Notably, co-feeding mifepristone with methoprene did not decrease food intake, but instead, significantly increased food intake (Figure 2C), and the same result was obtained at day 16 (Supplementary Figure S3G). These results confirm that the rescue of methoprene toxicity by mifepristone is not due to a decrease in food intake.

The negative effect of methoprene on median life span (approximately −64%) and partial rescue by mifepristone (ranging from +30% to +50%) was also observed in additional experiments, using methoprene concentrations of 100 and 50 μg/mL, in replicate assays (Supplementary Figure S4). In contrast, less consistent rescue of life span was observed when methoprene was administered at 500 μg/mL (Supplementary Figure S5). Feeding methoprene at 500 μg/mL caused a life span decrease of −72% in the first experiment, and −62% in the second experiment, and this was only partially rescued by mifepristone in the first experiment (+32%), and not at all in the second experiment. In summary, mifepristone partially rescues the dramatic negative effect of methoprene on life span across a wide range of methoprene concentrations (50–200 μg/mL), except for the highest and most toxic methoprene concentration (500 μg/mL), where the ability of mifepristone to rescue appears to be reduced.

It has previously been reported that feeding methoprene during larval development causes a dose-dependent lethality (32), and that result was also observed here (Supplementary Figure S6). Interestingly, co-feeding mifepristone with methoprene during development did not rescue the lethality, but instead made it more severe. Mifepristone administered at 200 μg/mL had no effect on larval survival, but in the presence of 10 μg/mL methoprene, the same concentration of mifepristone caused a −72% decrease in survival (Supplementary Figure S6). In summary, mifepristone rescues the negative effect of methoprene on survival of adult females, but promotes the negative effect of methoprene on survival of developing larvae. These data are consistent with the expectation that the toxic effect of methoprene in developing larvae proceeds through a different mechanism than the toxic effect of methoprene in adult females and provide further support for the conclusion that mifepristone antagonizes JH signaling in the adult female.

Mating and Mifepristone Have Limited Effect on the Frequency of SMURFs

Because the transcriptomics and metabolomics data point to the gut as a significant target of mifepristone, it was of interest to assay maintenance of gut integrity. Intestinal barrier integrity was assayed by feeding the flies media supplemented with 2.5% wt/vol blue dye #1, and scoring for leakage of the dye out of the gut and into peripheral tissues (the “SMURF” assay) (33). In the first set of experiments, virgin females and mated females from 6 different wild-type and laboratory genotypes were assayed at multiple time points during the death phase of the survival curve, in replicate cohorts. At time points where 10 or more of the flies remained alive, the frequency of SMURF flies ranged from 0% to 12% for virgins and from 0% to 27% for mated, and in some cohorts, few or no SMURFs were observed (Supplementary Figure S7). These results suggested that the frequency of SMURFs is generally low, and not dramatically affected by mating. To confirm this result, a cohort consisting of 98 virgin females, 99 mated females, and 87 mated females plus mifepristone was maintained at one fly per vial, throughout adult life, on media containing blue dye (Table 2). Previous studies report that the blue dye does not affect life span (33). Analysis of the survival curves for these flies confirmed the negative effect of mating (−27.3%); however, the increase due to mifepristone (+11.7%) was not significant (Figure 2G); one possibility is that the single fly-culture conditions might tend to moderate the effects of mating and mifepristone by moderating bacterial proliferation (see “Discussion”). Each fly was scored every other day for SMURF status until the death of the fly, as well as in the recently deceased fly. For each fly scored as SMURF while alive, the fly was dead within the next 48 hours, and was again scored as SMURF in the deceased fly, consistent with previous reports that the SMURF phenotype is a robust biomarker of impending death (Supplementary Figure S8) (34). The overall frequency of SMURFs was relatively low (Supplementary Figure S9), and the great majority of the flies (86%) died without ever showing the SMURF phenotype (Table 2; Supplementary Figure S8). The proportion of virgin flies that exhibited SMURF was 9.2%, and this increased to 19.2% in mated flies and decreased to 13.8% in mated flies fed mifepristone; however, these differences were not significant (Table 2). These data suggest that the −27.3% decrease in life span caused by mating cannot be attributed to the (possible) +10% increase in the frequency of SMURFs; however, it remains possible that a change in frequency of SMURFs might make a partial contribution.

Table 2.

Frequency of SMURF Phenotype With Continuous Assay

	Non-SMURF Deaths	SMURF Deaths	Total Deaths	% SMURF	p
Virgin	89	9	98	9.18
Mated	80	19	99	19.2	.0443
Mated + mifepristone	75	12	87	13.8	.3242

Note: Mated is compared to virgin, and mated + mifepristone is compared to mated, using chi-squared test. The p value for significance with 2 comparisons is .05/2 = .025.

Mifepristone Can Extend the Life Span of Mated C elegans

To ask if mifepristone might affect life span in other species, the hermaphrodite nematode C elegans was tested. Life span was assayed in unmated and mated C elegans, in the presence and absence of mifepristone, in 3 separate experiments (Supplementary Figure S10). In the first experiment, using 400 μg/mL mifepristone, mating decreased life span relative to unmated by −32% and also increased the number of animals censored due to internal hatching of offspring (bagging) and escape from the plate (Supplementary Figure S10A). Mifepristone administered at 400 μg/mL decreased life span of the unmated by −14%; however, this change was not significant. In contrast, 400 μg/mL mifepristone increased the life span of the mated animals by +7%. In the second experiments, mifepristone was administered at 800 μg/mL, and mifepristone decreased the life span of unmated by −9% and increased the life span of mated animals by +13% (Supplementary Figure S10B). In the third experiment, using 400 μg/mL mifepristone, there was no effect on unmated life span, and mated life span was increased by +25%, but this increase was not significant (log-rank p = .055; Supplementary Figure S10C). Finally, the data from the 3 experiments were combined to compare mifepristone-treated versus untreated animals (Figure 3). In the combined data, mating decreased life span by −41%. In unmated animals, mifepristone decreased life span by −9%, but this change was not statistically significant. In contrast, in mated animals, mifepristone increased life span by +15% (Figure 3). In each experiment, mifepristone increased the fraction of animals that were censored due to internal hatching (bagging) and escape from the plate surface, consistent with some toxic and aversive effects of mifepristone. Taken together, these data suggest that mifepristone can increase the life span of mated, but not unmated, C elegans, but also has variable toxic effects that may mask the benefit for life span.

Figure 3.

Effect of mifepristone on C elegans life span. M = mated; U = unmated. (−) = no drug; (+) = mifepristone. Data are presented as Kaplan–Meier plot of survivorship versus days. N = number animals; D = number deaths; C = number censored; percentage censored is in parentheses. The data are combined from 3 experiments. The separate experiments and analysis of variance (ANOVA) summaries are presented in Supplementary Figure S10. Unmated (+) is compared to unmated (−), mated (−) is compared to unmated (−), and mated (+) is compared to mated (−). The p value for significance with 3 comparisons is .05/3 = .0167.

Discussion

The dramatic decrease in life span caused by mating and SP combined with the dramatic increase in life span caused by mifepristone enables identification of genes and metabolites closely correlated with life span. Genes and metabolites that are upregulated by mating and SP, and downregulated by mifepristone, are negatively correlated with life span, and are therefore candidate negative regulators of life span (Figure 1, red font). Genes and metabolites that are downregulated by mating and SP and upregulated by mifepristone are positively correlated with life span and are therefore candidate positive regulators of life span (Figure 1, blue font). Mifepristone could not compete with ecdysone for activation of an ecdysone-responsive reporter. However, mifepristone was able to compete with the JH-analog methoprene for regulation of virgin life span, indicating that mifepristone either directly or indirectly antagonizes JH signaling downstream of SP. Surprisingly, mifepristone doubled the food intake in methoprene-treated virgin females, leading to greater levels than controls fed neither drug. One possibility is that by rescuing methoprene toxicity, mifepristone reveals a stimulatory effect of methoprene on food intake.

The negatively correlated genes were enriched for ones expressed in the midgut. This points to the gut as a potential negative regulator of life span, consistent with the fact that JH regulates remodeling of gut metabolism downstream of SP. The negatively correlated genes expressed in gut include digestive proteases and AA transporters, consistent with increased nutrient uptake and increased AA metabolism. Certain negatively correlated genes were expressed in the nervous system. These included female-specific independent of transformer (fit) which encodes a peptide hormone reported to regulate feeding in response to dietary protein (35). Also present was Obp99B which encodes an odorant receptor that is often correlated with Drosophila life span (21). Finally, photopigment gene Rh5 was negatively correlated with life span. Rh5 preferentially absorbs blue light, and blue light has been shown to decrease fly life span (36), suggesting the possibility that Rh5 might contribute to decreased life span by increasing blue light absorption. AMP gene expression has long been known to be negatively correlated with life span, and AMP gene transgenic reporters were among the first predictive biomarkers of remaining life span identified (37, 38). Our previous studies revealed that increased AMP gene expression correlates with increased microbial load (10, 39). Notably, neuronal AMP gene expression has been implicated as a negative regulator of life span (40), and each of the 4 AMP genes identified here are expressed in eye and/or brain.

A total of 7 genes were positively correlated with life span, and several of these (Elys, tweek, Unr, Dop2R) are involved in regulation of gene expression and signaling and are implicated in learning and memory. Dopamine signaling is of particular interest, as it is implicated in regulation of immune function and interactions with gut microbiota in humans (41). Robles-Murguia et al. (42) examined the effect of 43 μg/mL mifepristone on gene expression in adult thorax of male w[1118] flies and reported 1069 altered genes. Only 2 of those genes were in common with the 118 genes we find regulated by mifepristone in mated females, which is a 4.5-fold under-representation (hypergeometric p = .0047); this difference might be due to use of different sexes, tissues, and/or concentration of mifepristone.

Several of the negative pathways and metabolites identified here are also negatively correlated with life span in humans and other species. Amino acid metabolism in general, and particularly the metabolism of tryptophan, is negatively correlated with life span across species (43); disruptions of the tryptophan/kynurenine pathway are implicated in human cardiovascular disease, neurodegeneration, and cancer. Notably, the fluorescent tryptophan metabolite, anthranilic acid, is generated by the kynurenine pathway and is a marker of death in C elegans (44), and chemical inhibitors of the kynurenine pathway are reported to increase life span in Drosophila (45). Tryptophan metabolites, including kynurenic acid, interact with the host aryl hydrocarbon receptor (AhR) to regulate intestinal immune response and microbiota maintenance in mammals (46). The mating-induced AA metabolism observed here correlates with increased abundance of the TCA cycle intermediate oxaloacetate and increased lipid levels, which may be used to provision eggs. Because the targeted metabolome profile focused on aqueous metabolites, it is possible that there might be additional lipidome changes that were not identified, and this may be a useful area for future experiments. Three markers of RNA methylation were negatively correlated with life span (Figure 1C). These markers are associated with increased translation and are sometimes biomarkers of cancer in humans (27). Most bacterial toxins are translational inhibitors, and one possibility is that a mating-induced increase in protein synthesis makes the fly more susceptible to the toxic effects of bacterial inhibitors of translation. The negative correlation with life span of the bacterial metabolite 3-hydroxybenzoic acid supports the hypothesis that the microbiome may play a causative role.

Several of the positive metabolites and pathways identified here are also positively correlated with life span in humans and other mammals. Glycolysis is particularly important in the brain for the generation of energy, and for generating reducing equivalents to control oxidative stress (47). The positive correlation of carnitine, acetylcarnitine, and adipic acid with life span suggests a positive role for FA oxidation, and in particular, omega oxidation. Malonate is a breakdown product of Acetyl-CoA and a classical inhibitor of succinate dehydrogenase, thereby potentially inhibiting both the TCA and ETC. Purine metabolism and the abundance of the purine breakdown product urate have been positively correlated with longevity across mammalian species (48). However, increased urate levels are also a marker of disease in certain situations, in both flies and mammals (49). 1/3-Methylhistidine is a classical marker of muscle protein turnover (30), and therefore the increase in 1/3-methylhistidine and urate could be interpreted as markers for a beneficial turnover of muscle and other somatic tissues. Several positively correlated metabolites are implicated in signaling and memory. Epinephrine/normetanephrine protects humans from death during sepsis, and while it is not known to be produced by Drosophila, it can be produced by the microbiome (41). Both UDP-GlcNAC and 8-hydroxy-deoxyguanosine are implicated in memory formation and may be related to the gene expression changes discussed above. Finally, histamine is involved in normal photoreception, and the downregulation of histamine and upregulation of Rh5 expression by SP indicate a mating-induced change in the female’s vision, which conceivably could be related to decreased receptivity to re-mating.

It is worth noting these studies were conducted using standard laboratory strain genetic backgrounds (w[1118] background, or a hybrid of the w[1118] background and the w*, y[1] genetic background), and some patterns observed could be genotype-specific. One of us (D.E.L.P.) and coworkers previously showed that glycolysis metabolites and carnitines decrease with age in Drosophila, and that dopamine and GABA increase with age (50). Moreover, that study showed that AA levels, including tryptophan and tryptophan metabolites, showed interactions between age and genotype.

Important questions include if and how the identified gene expression and metabolite changes might causally relate to the observed changes in life span. The data confirm that the SMURF phenotype is a robust biomarker of impending death, however, the great majority of flies (~80%) died without ever showing a SMURF phenotype. Recently, Bitner et al. also reported that ~72% of flies die without showing SMURF phenotype (51). In contrast, previous studies suggested that all flies die showing SMURF (34); the reason for this difference is not clear, but might be related to differences in media composition, microbiome composition, and/or genotype. We also observed that SMURF frequency was reduced in individually housed flies (maximum value ~2.7%, Supplementary Figure S9) relative to co-housed flies (maximum value ~27%, Figure Supplementary S7), in contrast to a previous report (52), and we speculate this could be due to reduced bacterial load in the individually housed flies. In the future, it may be of interest to assay additional markers of gut function, such as stem cell proliferation and differentiation.

The fact that in our hands most flies died without showing SMURF phenotype suggests that some other cause of death predominates under these conditions. Ablating the adult fly microbiome with high-concentration doxycycline reduced the life-span effects of both mating and mifepristone, and transgenic SP expression and mifepristone (10), indicating the potential action of a toxic bacterial metabolite(s). The cause of death might be nervous system malfunction, as suggested by our recent study showing that episodes of erratic movement are a marker of impending death for the majority of flies (53). Taken together, these results are consistent with the hypothesis that the metabolic shift associated with mating and SP sensitizes the female to the toxic effects of bacterial metabolites, and this will be an important area for future research.

Our previous studies indicate that mating, SP, and mifepristone alter median life span primarily by altering the initial mortality rate, with relatively little change in maximum life span (10, 54). In this way, mifepristone increases life span by compressing mortality to late ages and producing a “rectangularization” of the mortality curve. One possibility is that mifepristone acts preferentially to reduce mortality in the first half of adult life because this is when the gut remodeling and reproduction pathways are most active. Another possibility is that death due to bacterial toxicity removes low-vitality animals from the population that would otherwise undergo rapid aging, such that older individuals represent those that are of higher than average quality, a process known as demographic selection (55). Consistent with this idea, our previous analyses of the life span changes caused by mifepristone using Gompertz-Makeham modeling indicates that rectangularization results from decreased initial mortality rate (parameter a) and a proportional increase in mortality rate acceleration (parameter b), which is traditionally referred to as the Strehler–Mildvan relationship, or sometimes as the compensation law of mortality (54). Finally, the experiments with C elegans suggest that mifepristone can increase the life span of mated, but not unmated, C elegans. Juvenile hormone is not reported to exist in C elegans, but conceivably some other direct or indirect target of mifepristone is conserved, and this will be an interesting area for future research.

In conclusion, the data indicate that mifepristone acts downstream of Drosophila SP to directly or indirectly antagonize JH signaling, thereby preventing a metabolic shift from somatic maintenance/turnover pathways towards increased AA metabolism, decreased immune function, and decreased survival.

Funding

This work was supported by a grant from the National Institute on Aging at the National Institutes of Health to J.T. (R01 AG057741). D.E.L.P. was supported by NIH grants P30 AG013280 in support of metabolomic work, and R01 AG049494 and R01 AG063371. J.S. was supported by a grant from National Natural Science Foundation of China (31500970).

Conflict of Interest

None declared.

Author Contributions

J.T. designed and supervised the overall study and wrote the paper. J.T. and G.N.L. designed and supervised Drosophila experiments and analyzed the data. J.T., G.N.L., D.V.D., Y.F., I.W., J.V., T.W., J.W., P.P., S.L., and M.A. conducted Drosophila experiments, and J.S. contributed statistical analyses. S.P.C. and C.Y. designed C elegans experiments. C.Y. conducted C elegans experiments. D.E.L.P. supervised metabolomics assays. D.P. and L.W. conducted statistical analysis of metabolomics assays.