Basis for study: evidence that Tk responds to microbiota.

A. Microbiota increase total number of enteroendocrine cells (EEs) in fly midgut. Number of GFP+ cells were scored in 5-day old UAS-mCD8::GFP; Voila-Gal4 females. Wilcoxon test, W=25, P=0.008. B. Microbiota alter expression of peptide hormones in the Drosophila intestine. Transcriptome data from (Bost et al., 2018) were reanalysed. Expression of neuropeptide hormone gene transcripts are plotted, dendrograms show hierarchical clustering on Euclidian distance. Differentially expressed transcripts (DESeq2, FDR<0.1) are marked with an asterisk to right. Presence of microbiota increased expression of Tachykinin (top row). Software did not quantify Burs expression, so complementary RT-qPCR data were collected, showing no impact of microbiota (Figure S1). C. Confirmation that microbiota promote Tk expression. Expression was quantified by RT-qPCR (using cDNA generated from midgut to complement RNA-seq data). Wilcoxon test W=1, p=0.04. D. Confirmation of lifelong axenic culture without antibiotics. Photographs show bacterial growth (or lack of) in homogenates of pools of 3 females, reared to day 56 of adulthood either with bacteria or axenic, serial dilutions indicated at top.

Ubiquitous Tk is required for microbiota to shorten lifespan and reduce TAG storage.

A. Kaplan-Meier plots showing survival of axenic and conventionally-reared (CR) DaGS;UAS-TkRNAi flies, faceted by absence/presence (top/bottom) of transgene activation by RU486 (RU). Cox Proportional Hazards analysis revealed an RU:microbiota interaction (p=0.05). Axenic RU- n=135, axenic RU+ n=136, CR RU- n=135, CR RU+ n=135. B. Post-hoc analysis of survival data shown in A. Facets show estimated marginal means of Cox Proportional Hazards model (EMMs±95% confidence intervals (CI95)), multiplied −1*n to make higher values correspond to longer lifespan. Pairwise tests (text on facets) revealed impact of microbiota only in absence of RU. C-D. TkRNAi (RU+) does not block effect of microbiota on feeding rate (C) or fecundity (D). Feeding quantified as average proboscis extension rate per vial (n=5 mated females/vial, 10 days old after 7 days feeding on RU/vehicle.). Fecundity quantified as n. eggs laid per fly over 24h (n=5 females/vial). E. TkRNAi (RU+) blocks reduction of TAG by microbiota. Points give TAG µg mg−1 in pools of five flies. The result was replicated with an independent RNAi construct (Figure S3). Panels A-E generated with RNAi construct V330743. Panels A-B, n=135-136 flies per condition. Panels C-E, Statistics on brackets are from post-hoc EMM analyses of linear models (C, logit-transformed data), when significant differences were detected.

Phenotypic impact of a complete microbiota and interaction with ubiquitous TkRNAi is recapitulated fully by Acetobacter pomorum, but only partially by Levilactobacillus brevis.

A. 16s rRNA amplicon sequencing reveals microbiota dominated by family Acetobacteraceae in the Dahomey fly background. Plot shows proportion ASVs per sample assigned to given families. “low abundance taxa” denotes a bin of all ASVs not assigned to one of the given families. ∼170k reads per sample, n=10 samples. B. Kaplan-Meier plots showing survival of DaGS;UAS-TkRNAi flies (RNAi construct V330743), cultured either under axenic or gnotobiotic conditions. Gnotobiotic flies were cultured in monoassociation with either Acetobacter pomorum (Ap-flies) or Levilactobacillus brevis (Lb-flies). Plots are faceted by microbial condition. Cox Proportional Hazards analysis revealed an RU:microbiota interaction (p=0.02). Ap RU- n=105, Ap RU+ n=105, Ax RU- n=105, Ax RU+ n=104, Lb RU- n=105, Lb RU+ n=105. C. Post-hoc analysis of survival curves shown in B. Plots show a survival coefficient, calculated from estimated marginal means of Cox Proportional Hazards model (EMMs±95% confidence intervals (CI95)), multiplied −1*n such that higher values correspond to longer lifespan. Pairwise tests (text on facets) revealed effect of TkRNAi induction (+RU) was largest in Ap-flies, intermediate in Lb-flies, and no effect was observed in axenics. D. A. pomorum, but not L. brevis, recapitulates effects of a complete microbiota on TAG. Points give TAG µg mg−1 in pools of five flies (n=9). Significant interaction of microbiota and TkRNAi (RU) was detected (ANOVA F=6.96, df=2,48, p=0.002). Statistics on brackets are from post-hoc EMM analyses of linear models, when significant differences were detected.

A. pomorum modulates intestinal Tk expression.

A. Tk expression is enriched in nervous system and midgut tissues (data from FlyAtlas2). B. RT-qPCR of Tk in dissected midguts and heads of axenic and gnotobiotic flies indicates that expression is increased specifically in the midgut by A. pomorum, but not in the head; while L. brevis affects expression in neither tissue. White point and error bars show mean±SD. Statistics on brackets are from post-hoc EMM analyses of linear models, when significant differences were detected. C. Quantifying lysine acetylation and Tk promoter activity in Tk+ cells. Cartoon shows regions quantified in Tk-Gal4/UAS-mCD8::GFP flies with acetyl-lysine staining. Scale bars=50µm. D. Lysine acetylation is increased specifically by A. pomorum in Tk+ enteroendocrine cells. Each point represents average fluorescence intensity for acetyl-lysine staining of ≥3 cells (marked by Tk>GFP) per gut, in anterior and posterior regions. The HDAC inhibitor Trichostatin-A was fed to axenic flies to confirm that increased histone acetylation is detectable by total acetyl-lysine staining. Statistics from one-way ANOVA, showing comparisons relative to axenic control condition. E. A. pomorum increases Tk promoter activity (Tk-T2A-Gal4>UAS-GFP) in anterior but not posterior midgut. Each point represents average GFP intensity of ≥3 cells per gut, in anterior and posterior regions. Statistics from one-way ANOVA, showing comparisons relative to axenic control condition.

Gut-directed TkRNAi abrogates lifespan shortening by A. pomorum.

A. Kaplan-Meier survival plots of the indicated genotypes in presence/absence of A. pomorum, with knockdown activated by Voila-Gal4, in presence/absence of ChAT-Gal80. Statistics from post-hoc tests of CoxPH model. Genotypes indicated above panels. See Figure S8 for tissue-specificity of transgene combinations. All conditions n=105. B. Post-hoc analysis of survival model reveals attenuated lifespan response to A. pomorum with midgut-directed TkRNAi (Genotypes as in A). Text indicates differences between A. pomorum and axenic flies in the given genotype. CoxPH analyses detected significant interactions both for microbiota*genotype (p<2.2e-16) and for Microbiota:Voila-Gal4:ChAT-Gal80:UAS-TkRNAi (p=0.0005). C. Kaplan-Meier survival plots of the indicated genotypes in presence/absence of A. pomorum, with knockdown activated by Tk-T2A-Gal4, and neuronal activity suppressed by R57C10-Gal80. D. Post-hoc analysis of survival model reveals attenuated lifespan response to A. pomorum with TkRNAi directed to midgut by Tk-T2A -Gal4 and R57C10-Gal80. Statistics from post-hoc tests of CoxPH model, indicating differences between Ap-flies and axenic flies in the given genotype. CoxPH detected overall genotype-by- A. pomorum interaction (p=2.46e-5).

Pan-neuronal TkR99DRNAi abrogates lifespan impact of A. pomorum.

A. FlyAtlas data showing expression of the two cognate Drosophila Tk receptors, TkR86C and TkR99D, among adult female tissues. B. Kaplan-Meier survival plots of Elav-GS; UAS-TkR86CRNAi flies in presence/absence of A. pomorum, with knockdown activated by RU486 feeding (bottom facet). CoxPH detected significant microbiota*RU interaction (p=0.0005). All conditions n=105. C. Post-hoc analysis of data from B showing that TkR86CRNAi induction locks axenic flies into a shortened lifespan. Text indicates differences between A. pomorum and axenic flies in each given RU condition. D. Kaplan-Meier survival plots of Elav-GeneSwitch; UAS-TkR99DRNAi flies in presence/absence of A. pomorum, with knockdown activated by RU486 feeding (bottom facet). CoxPH detected significant microbiota*RU interaction (p=1.362e-6). Ap RU- n=163, Ap RU+ n=161, Axenic RU- n=165, Axenic RU+ n=164 (pool of two replicate experiments). E. Post-hoc analysis of data from B showing that TkR99DRNAi induction blocks shortening of lifespan by A. pomorum. Text indicates differences between A. pomorum and axenic flies in each given RU condition. F. Neuronal TkR99DRNAi blocks onset of gut permeability (“smurf” phenotype) in aged flies. Bars show relative frequencies of flies with blue dye appearing to permeate through all tissues (“full smurf”) and through abdomen/thorax (“partial smurf”). χ2=16.126, DF=6, p=0.013. P-values on figure from chi-square tests of each given pair of columns, shown when differences were observed (otherwise omitted).

Insulin signalling is involved in lifespan response to A. pomorum and TkR99D knockdown, but not obligately required.

A. Hypothetical model of role for insulin signalling. In presence of microbiota (left), Tk peptides are released from gut and contact TkR99D in IPCs/brain. IPCs are released into circulation, activating insulin signalling in peripheral tissues and leading to nuclear exclusion of dFOXO. In absence of microbiota (right), extracellular signalling is diminished, reducing downstream insulin signalling and consequent activation of lifespan-extending gene expression program by dFOXO. B. dFOXO connects lifespan responses to A. pomorum and ubiquitous TkRNAi. Facets show Kaplan-Meier plots in axenic and A. pomorum conditions, with/without RNAi induction (RU), and with/without dFoxOΔ94 null mutation. Overall microbiota*RU*Foxo interaction p=5.384e-05 (CoxPH). All conditions n=150 except W.T. A. pomorum -RU (n=135) and W.T. axenic +RU (n=148). C. Post-hoc comparisons (EMmeans), showing impact of A. pomorum under specified conditions of TkRNAi induction and dFoxO deletion. A. pomorum shortens wild-type lifespan, but this effect is diminished by dFoxOΔ94. TkRNAi induction blocks lifespan shortening by A. pomorum in wild-type background, but not in dFoxOΔ94 background; altogether suggesting that dFoxO is required for A. pomorum to influence longevity via Tk modulation. D-E. Insulin-producing cells (IPCs) are required for A. pomorum to influence lifespan (CoxPH p=0.003). All conditions n=120. Post-hoc tests (E) confirm that lifespan effect of A. pomorum (-RU) is absent following IPC ablation (Rpr expression, +RU). F. A. pomorum increases expression of Dilp3 and Dilp5, but not Dilp2, in a TkR99D -dependent manner. Panels show expression (RT-qPCR) in heads of axenic and Ap-flies ±TkR99DRNAi in IPCs (Dilp2-GS/UAS-TkR99DRNAi). by Microbiota*RU interactions were tested for with linear models: Dilp2 F1,24=0.424, p=0.5, Dilp3 F1,24=23.404, p=0.0001, Dilp5 F1,24=29.261, p<0.0001. G-H. TkR99DRNAi in IPCs worsens lifespan impact of A. pomorum. Ap -RU n=160, Ap +RU n=162, Ax -RU n=160, Ax +RU n=163. (I) Post-hoc tests indicate accentuated shortening of lifespan by A. pomorum following TkR99DRNAi induction (+RU). I. TkR99DRNAi in IPCs does not modulate metabolic impact of A. pomorum on fly TAG. The two experimental factors each had significant effects on TAG (ANOVA, A. pomorum F=14.95, Df=1,37, p=0.0004; RU F=4.32, Df=1,37, p=0.04), but no interaction was detected.

dFoxO is partially required for microbiota*Tk interaction for lifespan: post-hoc pairwise comparisons (EMMeans)

TkR99DRNAi in insulin-producing cells blocks increased expression of dilp3 and dilp5, but not dilp2, by A. pomorum: post-hoc tests.

A. pomorum modulation of host lifespan and TAG is independent of TkR99D in Akh-producing cells.

A. Antibody staining of Akh in dissected Corpora Cardiaca (CC) is increased in Ap-flies (Ap) relative to axenics (Ax). (t-test: t = −2.36, p-value = 0.04). B. Microbial shortening of lifespan is independent of TkR99D signalling in Akh+ cells. Facets show Kaplan-Meier survival curves for axenic and Ap-flies when TkR99D is knocked down (middle facet) and in each control genotype. No significant interaction of genotype and A. pomorum was detected (CoxPH p=0.58). Akh-Gal4/+ A. pomorum n=103, Akh-Gal4/+ axenic n=108, Akh-Gal4/UAS-TkR99D-RNAi A. pomorum n=105, Akh-Gal4/UAS-TkR99D-RNAi axenic n=107, +/UAS-TkR99D-RNAi A. pomorum n=103, +/UAS-TkR99D-RNAi axenic n=101. C. Post-hoc comparisons (EMmeans) confirm equivalent lifespan shortening by A. pomorum, independent of TkR99DRNAi. Combination of transgenes indicated above facets. D. A. pomorum modulation of host TAG is independent of TkR99DRNAi in Akh+ cells. Boxplots show TAG levels are reduced independent of TkR99DRNAi induction. No significant interaction of genotype and A. pomorum detected.

Drosophila stocks used in this study.

Primers for RT-qPCR

Microbiota do not alter intestinal Burs expression.

Expression was quantified by RT-qPCR (using cDNA generated from midgut to complement RNA-seq data) because transcripts were not quantified from RNA-seq data (Figure 1). No significant difference in expression was detected (t=-1.65, p=0.15). AX=axenic, CV=conventionally-reared. Expression was quantified relative to Adh, as a housekeeping gene that RNAseq data showed to not be influenced by microbiota (Figure S2)

Microbiota do not alter intestinal Adh expression.

Expression was quantified from public RNAseq data (Bost et al., 2018). Statistics show log2 fold-change and p-value (unadjusted) from DESeq2 analysis of reads quantified with Salmon.

Lifespan extension in conventional flies by induction of an independent TkRNAi.

To validate result show in Figure 2A a distinct construct (Vienna Drosophila Stock Center #25800) was expressed under control of DaGS. Log-rank test p<0.01. -RU condition 108 deaths, 43 censors; +RU condition 124 deaths, 26 censors.

Lifespan extension by feeding RU to CR DaGS;UAS-TkRNAi flies is not attributable to off-target effects of RU or GeneSwitch activation.

Kaplan-Meier plot shows impact of feeding RU to DaGS/+ flies reared with complete microbiota. No effect of RU was detected (Cox Proportional Hazards p=0.5).

CFU. TkRNAi does not affect bacterial load in CR flies.

Violin plots show CFUs of conventionally-reared female adults of the indicated genotypes (all with DaGS), after one week of feeding on RU or vehicle control. All conditions n=10. No effects were detected of presence/absence of UAS-Tk-RNAi (ANOVA of log CFU counts, p=0.32), RU (p=0.23), nor genotype:RU interaction (p=0.25).

Interactive effect of TkRNAi and microbiota on TAG levels confirmed with independent RNAi construct.

The experiment presented in Figure 2E (which used construct V330743) was repeated with an independent construct (V103668). The same result was replicated. Violin plots show density of datapoints, mean±CI95 given in white. ANOVA (Type 3 tests), microbiota:RU F1,26=37.86, p=1.66e-06. Statistics on brackets are from post-hoc EMM analyses of linear model, when significant differences were detected.

Modulation of Tk expression by microbiota correlates lifespan outcome.

A. RT-qPCR indicates elevated Tk expression in Ap-flies above levels in axenics and Lb-flies, and above levels in all conditions expressing TkRNAi. Expression in Lb-flies was significantly greater than in conditions expressing TkRNAi, but not significantly greater than axenics without TkRNAi expression. Statistics on brackets are from post-hoc EMM analyses of linear model, when significant differences were detected. B. Tk expression differences across microbiota and TkRNAi conditions correlate lifespan differences. Tk expression from panel A is plotting against survival coefficients of lifespan experiment shown in Figure 3c-d.

Intersectional Gal80 strategy to direct Gal4 activity to enteroendocrine cells, sparing CNS.

As per Medina et al (Medina et al., 2024), we tested the capacity of Gal80 expressed under control of the ChAT promoter to repress Gal4 expressed under control of the Voila promoter, assayed by expression of UAS-mCD8::GFP. A. Representative confocal micrographs of CNS, and B. Quantification of Voila>GFP intensity in brain and CNS, in presence/absence of Voila Gal80. C. Representative confocal micrographs of midgut (region 4). D. Number of GFP+ cells/ROI in presence/absence of Voila Gal80.

TkR99DRNAi induction in enterocytes does not block impact of A. pomorum on lifespan.

Expression was driven with Mex1-GeneSwitch. Statistics in panel from Cox proportional hazards model, reporting post-hoc (EMM) analysis for effect of RU per microbiota condition.

AkhRRNAi induction in fat body (and gut) does not block impact of A. pomorum on lifespan.

Expression was driven with S106-GeneSwitch. Statistics in panel from Cox proportional hazards model, reporting post-hoc (EMM) analysis for effect of RU per microbiota condition.