Chronic ethanol ingestion impairs Drosophila melanogaster health in a microbiome-dependent manner

Ethanol is one of the worlds most abused drugs yet the impacts of chronic ethanol consumption are debated. Ethanol is a prevalent component in the diets of diverse animals and can act as a nutritional source, behavior modulator, and a toxin. The source of ethanol is microbes, which can both produce and degrade ethanol, and the gut microbiome has been associated with differential health outcomes in chronic alcoholism. To disentangle the various and potentially interacting roles of bacteria and ethanol on host health, we developed a model for chronic ethanol ingestion in the adult fruit fly, Drosophila melanogaster, which naturally consumes a diet between 0 and 5% ethanol. We took advantage of the tractability of the fly microbiome, which can be experimentally removed to separate the direct and indirect effects of commensal microbes. We found that moderate to heavy ethanol ingestion decreased lifespan and reproduction, without causing inebriation. These effects were more pronounced in flies lacking a microbiome, but could not be explained by simple bacterial degradation of ethanol. However, moderate ethanol ingestion increased reproduction in bacterially-colonized flies, relative to bacteria-free flies. Ethanol decreased intestinal stem cell turnover in bacterially-colonized flies and decreased intestinal barrier failure and increased fat content in all flies, regardless of microbiome status. Analysis of host gene expression finds that ethanol triggers the innate immune response, but only in flies colonized with bacteria. Taken together we show that, chronic ethanol ingestion negatively impacts fly health in a microbiome-dependent manner.


Ethanol concentrations of fly diets 93
Evaporation and bacterial metabolism may decrease the effective ethanol concentration of the fly 94 diets. Using a clinical grade breathalyzer, we developed a method to measure ethanol vapor 95 within the headspace of a vial and use this a proxy for dietary ethanol concentration [following 96 (Morton et al. 2014)]. Briefly, a 14-gauge blunt needle attached to 50 mL syringe is used to 97 sample the headspace of vial. The sampled air is then pushed through the mouthpiece of an 98 Intoximeters Alco-Sensor® III breathalyzer. Using 2.5%, 5%, and 10% ethanol media, with 99 either 20 bacterially-colonized or bacteria-free flies, we checked ethanol concentration once per 100 day for four days. Four (2.5% and 5%) or five (10%) replicate vials of each of the ethanol 101 treatments were used. Preliminary experiments show that ethanol vapor concentration in the 102 headspace stabilizes within two hours of opening a vial or taking a measurement (data not 103 shown). 104 105 Inebriation Assay 106 Inebriation was measured using an established method (Sandhu et al. 2015). Briefly, vials were 107 gently tapped and the number of individuals that were able to stand up 30 seconds later was 108 recorded. Inebriation was measured on bacterially-colonized and bacteria-free flies on diets 109 containing 5%, 10%, 12.5% and 15% ethanol, with four independent replicates per ethanol and 110 bacterial treatment. As a positive control, one mL of 85% ethanol was added to a cellulose 111 acetate plug that was pushed into the middle of a vial, and this vial was capped tightly with a 112 rubber stopper. Within 30 minutes, this method leads to inebriation in approximately 50% of 113 flies under a variety of experimental conditions (Sandhu et al. 2015). To measure ethanol vapor 114 in the positive control, a valve was attached to the rubber stopper and the headspace was sampled 115 at 30 minutes using the breathalyzer method described above.

117
Internal ethanol concentration of flies 118 To quantify the ethanol concentration to which fly tissues are exposed, we measured the internal 119 ethanol concentration using a colorimetric enzymatic assay (Sigma-Aldrich MAK076). This 120 approach measures the combined effects of ethanol uptake and internal metabolism. We 121 measured ethanol concentration in individual flies fed 0% or 10% ethanol diets for 15 days. As a 122 positive control, a group of flies not previously exposed to ethanol were enclosed in a rubber-123 stoppered vial with a cotton ball soaked with two ml of 35% ethanol [similar to (Fry 2014)]. A 124 dry cotton ball was added above the ethanol soaked one so that flies were unable to ingest 125 ethanol, while still being exposed to ethanol vapor. After 60 minutes, individuals that could not 126 stand, but still showed leg movements were selected. To calculate final internal concentration per 127 fly, ethanol was considered to be primarily located in the hemolymph and the hemolymph 128 volume was assumed to be 85 µL per fly (Troutwine et al. 2016; Cowmeadow et al. 2005).

130
Measurement of fecundity and lifespan 131 Lifespan and fecundity were measured simultaneously during the same experiment. Four 132 replicate vials of 20 females each were created for the two bacterial treatments (bacterially-133 colonized and bacteria-free) and the seven ethanol treatments (0% to 15%, in 2.5% increments) 134 resulting in 56 total vials for the 14 treatments. Survival was checked each day and dead flies 135 were removed with each transfer. Fecundity was calculated by the number of adults that emerge 136 per transfer to new diet, divided by the number of females alive at the start of that transfer, 137 summed over the entire experiment. Approximately 90% of all pupae that formed survived to 138 adulthood with no differences in eclosion rate between ethanol or microbial treatments ( Figure  139 S6) and thus only adult emergence data is shown. Development rate was measured as the day the 140 first pupae formed following a transfer to a new vial. In a follow-up fecundity experiment that 141 controlled for ethanol evaporation, flies were transferred to new freshly-inoculated media every 142 day. This experiment also included a diet that was isocaloric with the 2.5% ethanol diet. The 143 isocaloric diet was created by the addition of 4.4% glucose (to the 10% glucose added to all 144 diets) and assumes ethanol is 7 kcal/g and glucose is 4 kcal/g (Ja et al. 2007 We used NanoStrings profiling to quantify D. melanogaster gene expression changes due to 211 ethanol ingestion and bacterial colonization (NanoStrings Technologies, Inc. Seattle, WA, USA).

212
A custom NanoStrings probeset was designed to target genes related to ethanol metabolism, 213 innate immunity and inflammation, ethanol-mediated behavior, among others (A full list of 214 genes, raw counts, normalized counts, and P-Values are found in Supplementary Dataset S1). 215 Additionally, probes were designed to the bacterial 16S ribosomal RNA and alcohol 216 dehydrogenase A and B genes. Bacterially-colonized or bacteria-free flies were reared on 0% 217 and 10% ethanol diets for 11 days. Total RNA was obtained from individual whole flies using a 218 Trizol/Chloroform extraction [following (Elya et al. 2016)]. 50 to 75 ng of purified RNA per 219 sample was hybridized to the NanoString reporter and capture probesets following 220 manufacturing instructions, and profiled on an nCounter SPRINT machine (Laboratory of Greg 221 Barton, UC Berkeley). Raw counts were normalized to internal NanoStrings positive and 222 negative control probes and three housekeeping genes (Actin 5C, Gadph, and Ribosomal Protein 223 L32). The correlation between each of the three housekeeping genes and the final normalization 224 factor was always greater than 0.89. Treatment effects were determined with a two-way ANOVA 225 using ethanol and bacterial colonization as independent variables and normalized counts (i.e. 226 expression level) as the dependent variable (Supplementary Dataset S1). Significance was 227 determined using a 5% Benjamini-Hochberg False Discovery Rate. For all bacterial16S genes, 228 the normalized counts were greater than 10-fold higher in the bacterially-colonized treatments 229 compared to the bacteria-free treatments (Supplementary Dataset S1). For all bacterial ADH 230 genes, the normalized counts in the bacteria-free treatment were within three standard deviations 231 of the negative control probes and were greater in the bacterially-colonized treatments 232 (Supplementary Dataset S1 Additionally, we used two methods to expose flies to ethanol vapor. In the first, we soaked a 257 cotton ball with 2 mL of 35% ethanol and covered with a dry cotton ball so flies could not ingest 258 the ethanol [similar to (Fry 2014)]. In the second, we added 1 mL of 85% ethanol to a cellulose 259 acetate plug (Sandhu et al 2015). We found that headspace vapor accurately measures dietary 260 ethanol ( Figure 1A and S1). We also found that these methods lead to ethanol vapor levels many 261 times greater than our dietary ethanol method ( Figure 1A). Therefore, our chronic ingestion 262 model exposes flies to a much lower headspace vapor than previously established acute 263 inebriation models, and suggests the main source of ethanol uptake is ingestion.

265
To confirm that flies effectively uptake ethanol when it is mixed directly in the media, we 266 measured the internal ethanol concentrations of flies fed ethanol diets. We found that flies fed 267 10% ethanol diets contain higher internal concentrations of ethanol than flies fed 0% ethanol 268 diets, which shows that in our treatment conditions flies successfully ingest dietary ethanol 269 ( Figure 1B). We were also interested in how internal ethanol concentrations in flies fed ethanol 270 compare to flies exposed to ethanol vapor. We found that inebriated flies (exposed to 35% 271 ethanol vapor, which causes most flies to become immobile within an hour) have even higher 272 internal ethanol levels. This suggests that our dietary regime exposes flies to sub-inebriating 273 levels of ethanol.

275
We next asked if flies show behavioral signs of intoxication using the inebriation assay of 276 Sandhu et al. 2015, which measures inebriation as the inability for flies to stand after gently 277 tapping the vials. We found that, after 30 minutes, less than 5% of flies show signs of 278 inebriation, even on the highest ethanol diets, while half of flies exposed to 85% ethanol vapor 279 are inebriated ( Figure 1C). Importantly, there was no effect of bacterial treatment on inebriation, 280 consistent with results from a previous study that used therefore not shown. Note that flies are transferred to fresh vials on day 3 or 4 ( Figure 2A and B, 323 3, 4, and 5) or day 1 ( Figure 2C). of D. melanogaster, fermenting fruit, often contains 1-5% ethanol and the unnatural but common 340 habitat of vineyards can contain up to 10% ethanol (Gibson et al. 1981). We therefore tested 341 dietary ethanol concentrations from 0% to 15%, which spans from ecologically relevant 342 concentrations to concentrations above those to which flies are normally exposed. 343 344 We measured lifespan, fecundity, and microbiome composition (see next section) in the same 345 experiment. Four replicate vials of 20 flies were used for each ethanol and bacterial treatment. In 346 our first experiment, we transferred flies to fresh food every three to four days to balance 347 between maintaining dietary ethanol concentration, which decreases over time, (see Figure 1D-348 F) and maintaining bacterial colonization, which requires less frequent transfers (Blum et al. 349 2013)].

351
Bacterially-colonized flies consistently showed a shorter lifespan than bacteria-free flies, in 352 agreement with previous studies (Figure 2A; Table S1; Data for individual flies is shown in 353 Figure S2; Lifespan curves are shown in Figure S3 That bacterially-colonized flies have a reduced lifespan compared to bacteria-free flies has been 365 reported before though never with the same magnitude we found here, suggesting that the flies in 366 our lab may be colonized with a particularly lifespan-shortening consortium of bacteria. 367 However, an equally plausible explanation would be that our media, which lacks the commonly- bacterially-colonized, third order, R 2 =0.832). Data for individual flies is shown in Figure S2 and 380 Table S1. shown. An independent replication of this experiment (with 0% and 2.5% ethanol) is shown in 387 Figure S4. 388 C: The effect of 2.5% ethanol on fecundity is not due to ethanol evaporation or the caloric 389 contribution of ethanol. Flies were transferred to fresh diets each day to reduce the effect of 390 ethanol evaporation ( Figure 1D). The isocaloric diets have added glucose so they contain 391 identical calories as the 2.5% ethanol diets. Adults per female was calculated as in Figure 2B, We found a strong effect of ethanol on fly fecundity that is mediated by bacterial treatment 399 ( Figure 2B). Without ethanol, there was no difference in fecundity between bacterially-colonized 400 and bacteria-free flies, which is consistent with previous work (Ridley et al. 2012). For both 401 bacterial treatments, ethanol reduced fecundity, but bacteria-free flies were more sensitive: at 402 2.5% ethanol, bacterially-colonized flies had significantly higher fecundity (P=4x10 -7 , Figure  403 2B; P=0.03, Figure S4). The same trend was observed on both 5% and 7.5% ethanol diets 404 (though not statistically significant at P<0.05 after correction for multiple comparisons). 405 Interestingly we found that ethanol decreases both components of fitness (Figures 2A and 2B, Figure S5). 408 This suggests that ethanol, even at the low concentrations used in this study, is acting more like a 409 toxin than a source of calories. 410 411 On 2.5% ethanol media, ethanol content is significantly reduced by day two in the bacterially-412 colonized treatment ( Figure 1D). Therefore, the difference in fecundity observed in the 2.5% 413 ethanol treatment ( Figure 2B) could simply be due to less dietary ethanol in the bacterially-414 colonized treatment. Although the fecundity difference between bacterially-colonized and 415 bacteria-free flies remains even when accounting for differential ethanol loss (Supplementary 416 Dataset 2), we nonetheless repeated the fecundity experiment but transferred the flies to fresh 417 diets every day. To ensure the persistence of the intestinal bacterial communities, we seeded each 418 daily batch of media with the frass of bacterially-colonized flies. Also, to test whether the 419 calories added by the ethanol in the 2.5% treatment affect the flies, we added a 0% ethanol 420 treatment that is isocaloric with the 2.5% ethanol treatment.

422
The daily transfer experiments showed comparable results. In concordance with the 3-4 day 423 transfers, bacterially-colonized flies had greater fecundity in the 2.5% ethanol treatment relative 424 to bacteria-free flies (P=1x10 -5 , Figure 2C). This strongly suggests that bacterial metabolism of 425 ethanol on the food does not cause the difference in fecundity between the bacterial treatments. 426 We also found no effect of bacterial treatment in the isocaloric diets suggesting that the 427 differences in fecundity cannot be attributed to ethanol's caloric contribution.

429
The observed fecundity differences could be due either to maternal egg production or larval 430 survival. To differentiate between these causes, we measured larval development time as a proxy 431 for larval survival because we could not directly count egg laying (and thus could not calculate 432 survival from egg to adulthood). Although development time increased on the highest ethanol 433 diets ( Figure S6), we found no effect on development time between bacterially-colonized and 434 bacteria-free treatments at the 0%, and 2.5% ethanol treatments (all pairwise t-tests, P>0. bacterially-colonized flies. Thus, maternal egg production, rather than larval survival, accounts 441 for fecundity effects seen in Figure 2B. 442

443
Taken together, these results suggest significant ecological and evolutionary impacts of microbes 444 in mediating the negative effects of ethanol toxicity on fly fecundity. While the exact doses of 445 ethanol that flies consume in the wild remains obscure, the concentration in naturally fermenting 446 fruit is typically 1-5% and can be as high as 10% in wineries (Gibson et al. 1981). In all cases for 447 flies fed 2.5% to 7.5% ethanol, we found that bacterially-colonized flies had higher fecundity 448 than bacteria-free flies and this effect persists even when controlling for ethanol metabolism by 449 daily transfers to fresh media. Thus, at ecologically relevant concentrations of ethanol, bacterial 450 colonization mitigates the negative effects on fecundity.

452
Ethanol shifts the composition of bacteria associated with D. melanogaster 453 454 Diet is a strong determinant of microbiome composition in flies and other animals. In particular, 455 fruit feeding flies, which are exposed to naturally produced dietary ethanol, have significantly 456 different bacterial and yeast communities than flies collected from other substrates (Chandler et 457 al. 2011;Chandler et al. 2012). We hypothesized that the bacterial communities associated with 458 flies would shift in response to ethanol ingestion. In particular, we expected that ethanol would 459 strongly decrease the total abundance of bacteria in high ethanol treatments and these shifts 460 would favor the abundance of bacteria with low sensitivity to ethanol. Thus, in a parallel 461 replicate of the lifespan-fecundity experiment (Figures 2A and 2B), we determined fly bacterial 462 load and composition by homogenizing individual flies and plating onto selective media. The 463 different bacterial strains were identified by colony morphology.

465
We found that total bacterial load per fly was between 9x10 3 and 3x10 6 colony forming units 466 (CFUs) for the 0% ethanol containing diets (mean=7x10 5 ). This is comparable to previous 467 studies of D. melanogaster (Blum et al. 2013;Obadia et al. 2017). Contrary to our expectations, 468 we found that total bacterial load was relatively constant up to the highest ethanol treatment 469 ( Figure 3A). We next asked how the bacterial composition changes in response to ethanol. In decreased 10-fold from 0% to 2.5% ethanol and remained constant until 12.5% ethanol where 474 they dropped to essentially 0 ( Figure 3B). Conversely, we found that the response of the 475 Lactobacilli to ethanol was remarkably different than A. pasteurianus. The abundance of L. 476 brevis increased with dietary ethanol and this was the only species that was present in all flies at 477 15% ethanol ( Figure 3C). L. plantarum was most abundant at intermediate concentrations of 478 ethanol, but like L. brevis, it did not appear as sensitive to high levels of ethanol as A. 479 pasteurianus ( Figure 3D).

481
To confirm the direct effect of ethanol on the bacterial growth, we measured the in vitro growth 482 response to ethanol of A. pasteurianus, L. plantarum and L. brevis strains isolated during the 483 experiment in Figure 3A-3D. These experiments confirmed that A. pasteurianus is more 484 sensitive to ethanol than L. brevis and L. plantarum ( Figure 3E). These results indicate that the 485 bacterial composition of flies varies, at least in part, according the ethanol sensitivities of the 486 bacterial strains. However, because these in vitro experiments show that these bacteria are more 487 sensitive to ethanol than is suggested by their in vivo abundances, we hypothesized that the fly 488 intestine protects bacteria from ethanol toxicity. In support of this, we found high abundance of 489 L. brevis and L. plantarum within flies fed a 15% ethanol diet despite these bacteria being 490 undetectable on this media ( Figure S8). Similarly, A. pasteurianus is present within flies fed 491 12.5% ethanol despite this bacterium being absent on this media ( Figure S8). This confirms that 492 the host shields the effect of ethanol on the bacteria. Number of individual flies per treatment: 0% N=14; 2.5% N=11; 5% N=16; 7.5%N=10; 501 10%N=11; 12.5% N=11; 15%N=8. We found no effect of fly age [multivariate ANOVA 502 (Adonis, package vegan in R; P = 0.159)] and therefore all four timepoints are pooled (see 503 methods). 504 E. A. pasteurianus is more sensitive to ethanol than L. plantarum or L. brevis in vitro. 505 Strains were isolated in the in vivo bacterial abundance experiment (Figure 4). Growth was 506 measured using MRS or MYPL liquid media containing 0% to 15% ethanol in either a 96-well 507 plate (A. pasteurianus and L. plantarum) or cell culture tubes (L. brevis) for 24 hours at 30C. 508 Datapoints are the final normalized OD of two independent replicates. A two-parameter Weibull 509 function was fit to the normalized ODs from the aggregate data for each strain (R package drc: 510 Analysis of Dose-Response Curves). The inhibitory concentration for 50% growth (IC50) was 511 calculated as the ethanol percentage that reduced normalized maximum OD by half. 512 values are calculated from a pairwise t test between treatments and are Holm-Bonferroni 548 corrected for multiple comparisons within an experiment. All data shown is for females. Male 549 data for IBF is shown in Figure S9 and for triglyceride content is shown in Figure S11. Results from an independent experiment as shown in Figure S10. experiments, we found that in the absence of ethanol, ISC division was significantly greater in 570 bacterially-colonized flies (P=2x10 -6 , Figure 4B; P=2x10 -6 , Figure S10), consistent with previous 571 work showing that ISC hyper-proliferation caused by commensal bacteria shortens fly lifespan 572 [ (Figure 2A), (Buchon et al. 2009;Guo et al. 2014)]. 573 574 In the presence of ethanol, we found that ISC division is significantly decreased in bacterially-575 colonized flies, but unchanged in bacteria-free flies (P=2x10 -4 , Figure 4B; P=3x10 -6 , Figure  576 S10). These results are in accord with the IBF data presented in Figure 4A for bacterially-577 colonized flies (specifically, that both IBF and ISC division, two processes linked to intestinal 578 homeostasis, are reduced), but suggest an additional mechanism reduces IBF in bacteria-free 579 flies fed ethanol. Why these two phenotypes are uncoupled in bacteria-free flies, in which 580 ethanol decreases IBF with no change in ISC turnover, remains unknown and suggests different 581 mechanisms of ethanol-induced pathology in bacterially-colonized and bacteria-free flies. 582 583 The decrease in dividing ISCs in bacterially-colonized flies led us to hypothesize that ethanol 584 might inhibit the ability of ISCs to regenerate following a biological or chemical challenge. To 585 test this hypothesis, we infected flies with Erwinia carotovora carotovora 15 (Ecc15), a non-586 lethal pathogen of Drosophila, which reliably induces ISC division following oral ingestion 587 (Buchon, Broderick, Poidevin, et al. 2009). In both bacteria-free and bacterially-colonized flies 588 ingesting ethanol, infection with Ecc15 increases ISC division (P=0.021 and P=0.013, 589 respectively, Figure 4C). Thus, ethanol does not inhibit the ability of ISCs to regenerate despite 590 the observed decrease in ISC division in bacterially-colonized and ethanol-fed flies ( Figures 4B  591 and S10).

593
Ethanol ingestion increases stored triglycerides in flies 594 595 The maintenance of intestinal homeostasis with ethanol treatment may be due to a change in 596 overall fly metabolism. In flies, poor quality diets are linked to both ISC turnover and obesity 597 (Skorupa et al. 2008;Regan et al. 2016). In humans, increased fat deposits in the liver are a 598 hallmark of alcoholic liver disease (Diehl 2002). We hypothesized that ethanol ingestion is 599 leading to greater accumulation of stored triglycerides in flies. Triglycerides are a primary 600 molecule for fat storage in flies and are mainly found in adipocytes within the fat body, an organ 601 analogous to the mammalian liver that is responsible for the majority of energy reserves in adult 602 fly (Arrese & Soulages 2010). We therefore measured stored triglycerides in bacteria-free and 603 bacterially-colonized flies on 0%, 5% or 10% ethanol diets. Consistent with our hypothesis, we 604 found that dietary ethanol increases triglycerides regardless of bacterial colonization, with no 605 effect on either total fly mass or free glycerides ( Figures 4D and S11, Table S2). Because dietary 606 sugars increase triglyceride content in flies (Skorupa et al. 2008), our finding is consistent with 607 ethanol acting as an energy source with regards to fat storage, despite the lack of tradeoff 608 between lifespan and fecundity due to ethanol ingestion ( Figure S5). The finding that there is no 609 difference in triglyceride content between bacterially-colonized and bacteria-free flies is 610 consistent with the minimal role of bacterial metabolism on 5% and 10% ethanol diets ( Figures  611  1E and 1F) and suggests that fat accumulation does not directly explain either the lifespan 612 (Figure 2A) or intestinal homeostasis ( Figures 4A and 4B) results.

614
Ethanol ingestion and bacterial colonization affect expression of innate immunity genes 615 616 To understand the molecular mechanisms underpinning the differences in lifespan between 617 bacteria-free and bacteria-colonized flies ingesting ethanol (Figure 2A), we surveyed fly gene 618 expression using a custom NanoStrings probeset and selected candidate genes likely to be 619 influenced by ethanol or microbiome status. In concordance with previous work (Broderick et al. 620 2014), we found many immune system (e.g. lysozyme X and the PGRPs), stress related (e.g. 621 GstD5 and HSP23), and cell differentiation (e.g. upd3) genes to be upregulated in response to 622 bacterial colonization (Table 1). Likewise, and in agreement with Elya et al 2016, we found that 623 anti-microbial peptides (AMPs) as a group show increased expression in bacterially-colonized 624 treatments (Table S3). 625 626 We examined many genes and molecular pathways known to mediate the effects of ethanol 627 intoxication in flies, but we found that ethanol only subtle changes the expression of 628 neuropeptideF (Table 1). This is consistent with ethanol ingestion not leading to inebriation 629 ( Figure 1C) and flies fed ethanol having a lower internal concentration of ethanol than inebriated 630 flies ( Figure 1B).

632
There are two potential mechanisms that may contribute to the ethanol-induced lifespan 633 reduction in bacteria-free, but not bacterially-colonized, flies. First, host metabolism of ethanol 634 may be more efficient in bacterially-colonized flies, leading to a faster clearance of ingested 635 ethanol. Contrary to this hypothesis, we did not find that genes in the ethanol metabolism 636 pathway [alcohol dehydrogenase (Adh) and acetaldehyde dehydrogenase (Aldh)] were more 637 strongly induced in bacterially-colonized flies (Table 1), which is consistent with the equivalent 638 internal ethanol concentrations we observed for bacteria-free and bacterially-colonized flies fed 639 10% ethanol ( Figure 1B). This result suggests that ethanol metabolism does not directly underpin 640 the differences in lifespan between bacteria-free and bacteria-colonized flies.

642
In the second mechanism, ethanol may be eliciting the innate immune response in bacteria-free, 643 but Imd decreased with ethanol ingestion in bacterially-colonized flies. However, both of these genes 654 show greater changes with ethanol ingestion in the bacterially-colonized treatment, with little or 655 no change in the bacteria-free treatment (Figure 5), and therefore do not directly explain the 656 lifespan differences identified in Figure 2A. We next examined whether antimicrobial peptide 657 (AMP) gene expression, which is a downstream target of Imd, was affected. We found no 658 ethanol dependence of AMP stimulation, (Table S3). This finding is consistent with literature 659 showing that the AMP expression can be muted when Imd is triggered in the absence of a 660 pathogen (Lhocine et al. 2008 ingestion or bacterial colonization. P-values are adjusted using a 5% Benjamini-Hochberg false 666 discovery rate correction. Genes with less than a 1.5 fold-change or with average normalized 667 counts within two standard deviations of the negative control probes are excluded. In the absence of ethanol or at low ethanol concentrations, the negative effects of bacteria are 710 dominant and reduce lifespan by disrupting intestinal homeostasis and inducing innate immunity. 711 At high ethanol concentrations, the microbiome changes composition (primarily through the 712 reduction of Acetobacter abundance). This eliminates the bacteria-dependent lifespan reduction, 713 but is offset via an unknown ethanol-dependent mechanism which is damaging to fly health (so 714 the net result is no change in lifespan). This mechanism is primarily independent of intestinal 715 homeostasis and the innate immune response. It is also likely different from known intoxication 716 pathways in flies because our method of ethanol administration does not lead to overt 717 inebriation. Firestone. We thank members of the Ludington lab, including Anjali Jain, Alison Gould, Tüzün 726 Güvener, Benjamin Obadia, Vivian Zhang, and Carlos Zuazo, for helpful contributions to this 727 project. We thank the laboratory of Matt Welch for use of their plate reader, April Price and the 728 laboratory of Greg Barton for assistance with NanoStrings data collection, Henrik Jasper for an 729 aliquot of pH3 antibody, and James Fry for help with interpretation of the internal ethanol 730 concentration and NanoStrings results. We thank Alison Gould and Benjamin Obadia for 731 providing helpful comments to the manuscript. 732