Abstract
Animals host multi-species microbial communities (microbiomes) whose properties may result from inter-species interactions; however current understanding of host-microbiome interactions is derived mostly from studies in which is it is difficult to elucidate microbe-microbe interactions. In exploring how Drosophila melanogaster acquires its microbiome, we found that a microbial community influences Drosophila olfactory and egg-laying behaviors differently than individual members. Drosophila prefers a Saccharomyces-Acetobacter co-culture to the same microorganisms grown individually and then mixed, a response mainly due to the conserved olfactory receptor, Or42b. Acetobacter metabolism of Saccharomyces-derived ethanol was necessary, and acetate and its metabolic derivatives were sufficient, for co-culture preference. Preference correlated with three emergent co-culture properties: ethanol catabolism, a distinct volatile emission profile, and yeast population decline. We describe a molecular mechanism by which a microbial community affects animal behavior. Our results support a model whereby emergent metabolites signal Drosophila to acquire its preferred multispecies microbiome.
Introduction
Multispecies microbial communities (microbiomes) influence animal biology in diverse ways (1): microbiomes modulate disease (2), metabolize nutrients (3), synthesize vitamins (4), and modify behavior (5). A central goal in host-microbiome studies is to understand the molecular mechanisms underpinning these diverse microbiome functions.
Some aspects of microbial community function are the product of interspecies interactions (6-9). For example, microorganisms modulate the metabolomes of neighboring species (10) and microbial metabolites (e.g., antibiotics) alter bacterial transcriptional responses (11). Despite current understanding of microbial inter-species interactions in vitro, some of which has been elucidated in exquisite detail, the consequences of microbial interspecies interactions within host-associated microbiomes are just beginning to be explored experimentally.
Insight into host-associated microbiome function has stemmed mostly from whole-microbiome (e.g., re-association of germ-free hosts with whole microbiomes (12)) and modeling microbiome function based on gene annotation (13) or single-microorganism (e.g., re-association of germ-free hosts with a single microorganism (14)) studies. However, these approaches tend to reveal only limited insight into interspecies microbial interactions, which can provide hosts with essential services. For example, termite symbionts carry genes necessary for metabolism of different parts of complex carbohydrates (15), yet their function has not been demonstrated in vivo; co-occurring human gut symbionts share polysaccharide breakdown products cooperatively (16, 17), but the consequences of such interactions for the host are unknown; inter-species bacterial interactions protect Hydra from fungal infection (18), but the mechanism of host protection is unclear. The need to understand the effects of inter-species microbiome interactions motivated our current work.
Attractive model systems in which to study the outcomes of inter-species microbial interactions for host biology would include a tractable host that harbors a simple multispecies microbiome. Here, we report the use of Drosophila melanogaster to study interactions in a simple microbiome and their consequences for host behavior.
The Drosophila microbiome consists largely of yeasts, acetic acid bacteria, and lactic acid bacteria (19-22). Drosophila ingests microbiome members from the environment (e.g., fermenting fruit, (22-26)), a behavior posited as a mechanism for Drosophila to select, acquire, and maintain its microbiome (21, 25). Drosophila behavior toward environmental microorganisms has focused on yeasts (27-31). While yeasts attract Drosophila via ester production (28, 29), lactic and acetic acid bacteria produce metabolites (e.g., acids) that may repel Drosophila, based on olfactory avoidance of high acid concentrations (32). One motivation of our study was to analyze Drosophila behavior toward the yeast and bacteria that dominate the Drosophila microbiome.
Yeast and bacteria are largely studied within separate Drosophila sub-disciplines, despite their shared habitat (21). Yeasts serve as food, providing Drosophila vitamins, sterols, and amino acids (21). Lactic and acetic acid bacteria are gut microbiome members (33) promoting larval development (34, 35), increasing resistance to pathogens (25), inducing intestinal stem cell proliferation (36), and reducing adult sugar and lipid levels (37, 38). Since microorganisms that are traditionally considered ‘food’ co-exist with those considered “microbiome’ in fruit fermentations and the two groups provide Drosophila with different services, we hypothesized that Drosophila might detect a beneficial community via metabolites that are produced cooperatively by the desirable symbionts. Alternatively, Drosophila might detect a different metabolite as the signal for each symbiont.
Fruit undergoes a well-characterized ripening process in which cell-wall degrading enzymes and amylases convert the firm, starchy tissue into soft, sugar-rich fruit (39-41). The high sugar content supports microbial colonization and fermentation by Drosophila-associated microorganisms, including yeasts, lactic acid bacteria, and acetic acid bacteria (23, 42). Drosophila avoids ‘green’ fruit and is attracted to ‘overripe’ fruit (43), yet it is unclear whether Drosophila behavior is influenced by the fruit microbiome and its metabolic properties. To this end, we developed a model fruit fermentation system that afforded measurement of microbial populations, microbial metabolites, and Drosophila behavior.
Here we demonstrate the importance of emergent microbiome metabolism—quantitatively different or unique metabolites produced by the microbiome, but not by any of its members in isolation—on behavior, suggesting that Drosophila modulates its chemical intake by selecting a beneficial microbiome.
Results
To determine whether Drosophila responds to emergent microbial community metabolites, we used the T-maze olfactory assay to analyze Drosophila behavioral responses to several Drosophila microbiome members grown individually or in communities (Figure 1A, Table S1, Figure 1-figure supplement 1). When the strains were grown individually, Drosophila was strongly attracted to yeasts, moderately attracted to acetic acid bacteria, and neutral or slightly repelled by lactic acid bacteria (Figure 1B, C). Because strains within a microbial group attracted Drosophila similarly, a representative yeast, acetic acid bacterium, and lactic acid bacterium were used to test the effect of interactions between microbiome members on Drosophila behavior. Drosophila preferred microbial communities grown together to microorganisms grown individually and then mixed prior to analysis (defined throughout as a separate-culture mixture, Figure 1D). Focusing on a model Saccharomyces cerevisiae and Acetobacter malorum community, we found that when tested against apple juice medium (AJM), Drosophila preferred the community to the separate-culture mixture or individual members (Figure 1E). In sum, Drosophila detects, and prefers, microorganisms growing together to a mixture of the same strains combined after they had completed growth.
We next measured the attractiveness and other properties of the co-culture over time. When grown alone, the microorganisms had similar growth profiles (Figure 2A). However, when grown with A. malorum, S. cerevisiae populations first increased, then decreased between 60 and 72 h, and were undetectable by 96 h (Figure 2A). The decrease in S. cerevisiae viable counts mirrored a decrease in pH (Figure 2-figure supplement 1A). Drosophila did not prefer the co-culture relative to the separate-culture mixture at 34 h, slightly preferred the co-culture at 48 and 54 h, and strongly preferred the co-culture at 60 through 96 h (Figure 2B). Drosophila preference for the co-culture correlated with lower pH and S. cerevisiae populations, despite Drosophila olfactory avoidance of acid (32) and reliance on yeast for nutrition (44) (Figure 2C, D). Drosophila preference did not correlate with viable A. malorum populations (Figure 2-figure supplement 1B). Drosophila preference for the co-culture increased relative to sterile media during 34-96 h of growth, which is consistent with the increase in Drosophila attraction being due to a property of the co-culture rather than to a decrease in attraction to the separate-culture mixture (Figure 2-figure supplement 1C). In sum, several properties of the microbial community (e.g. S. cerevisiae density, pH) parallel Drosophila detection of, and preference for, the co-culture.
Mutants in broadly and narrowly tuned ionotropic and olfactory receptors (Irs and Ors, respectively, (45, 46)) were used to evaluate the role of Drosophila olfactory reception in discriminating the co-culture from the separate-culture mixture during and immediately following peak attraction (Figure 3-figure supplement 1). During the most attractive phase of the co-culture (e.g. 67-115 h), Or83b and Or42b showed significant attraction to the co-culture, whereas no role was detected for Irs or Or35a (Figure 3). As the co-culture proceeded (e.g. 139-163 h), attraction decreased and the role of Or42b and Or83b waned (Figure 3). Or83b is a required co-receptor for all other Or gene products (47) and Or42b, one of the most conserved olfactory receptors, detects esters and 1,1-diethoxyethane (48-50). These results suggest that Or42b enables Drosophila to distinguish the co-culture from the separate-culture mixture. Our results suggest that a non-Or83b factor explains ∼40% of Drosophila co-culture preference (Figure 3). Previous work found that Or83b is fully responsible for Drosophila attraction to apple cider vinegar (51), suggesting that the behavioral circuit activated by inter-species interactions of the co-culture is distinct from the circuit activated by apple cider vinegar.
We speculated that the emergent property of co-culture attractiveness might arise from a distinct metabolic profile of the co-culture. Using gas chromatography-mass spectrometry (GC-MS), we identified volatiles unique to or differentially produced in the co-culture compared to the separate-culture mixture. Five co-culture volatiles (ethanol, isobutanol, isoamyl alcohol, acetic acid, isoamyl acetate) were confirmed with standards (Table 1-figure supplements 1-2) and quantified with standard curves (Table 1-figure supplement 3). The alcohol concentrations were lower, and acetic acid and isoamyl acetate were unique in the co-culture relative to the other experimental groups (Table 1). The molecular profile was reminiscent of ethanol catabolism as the unique co-culture metabolic process. We therefore hypothesized that ethanol catabolism was the emergent metabolic process.
We next measured ethanol and acetic acid levels over time (24-156 h) and compared the chemical dynamics to Drosophila preference. Consistent with a relationship between ethanol catabolism, acetic acid anabolism, and Drosophila attraction, ethanol levels declined and acetic acid accumulated in the co-culture at a similar time of Drosophila co-culture preference (Figure 4A). Furthermore, as ethanol catabolism and acetic acid anabolism proceeded (36-96 h), Drosophila attraction toward the co-culture increased until 96 h, at which point it decreased, consistent with lower turnover of ethanol at the end of ethanol catabolism (Figure 4A, black line).
We hypothesized that Drosophila preferred the community during peak ethanol turnover (e.g. co-cultures at ∼72 h of growth) compared with the community pre-ethanol catabolism (e.g. co-culture at ∼36 h of growth) or during late-stage ethanol catabolism, in which ethanol turnover is low (e.g. co-culture at ∼144 h of growth; Figure 4A). Consistent with our hypothesis, Drosophila preferred the co-culture in the middle stage of ethanol catabolism to earlier or later stages of ethanol catabolism (Figure 4B).
We hypothesized that Drosophila preference for communities during peak ethanol turnover reflected fitness benefits derived from ingesting metabolites associated with different staged communities. To test this hypothesis, we measured Drosophila survival when given ethanol and acetic acid concentrations characteristic of microbial cultures at different stages of ethanol catabolism. Drosophila survival was highest when given metabolites associated with middle-staged ethanol catabolism compared with pre- or end-stage ethanol catabolism (Figure 4C). In sum, our results are consistent with Drosophila preference providing benefits associated with consumption of microbial community-generated metabolites.
To test directly whether ethanol catabolism underpinned Drosophila co-culture preference, we evaluated Drosophila preference for the co-culture harboring a mutant in adhA, which encodes pyrroloquinoline quinone-dependent alcohol dehydrogenase (PQQ-ADH-I), the enzyme that converts yeast-derived ethanol into acetaldehyde on path to acetic acid (34). Co-cultures using either A. malorum or A. pomorum wild-type (WT) along with S. cerevisiae were equally attractive to Drosophila (Figure 5-figure supplement 1). Drosophila preferred the co-culture containing A. pomorum WT versus a separate-culture mixture; however, Drosophila did not prefer the co-culture containing A. pomorum adhA versus a separate-culture mixture (Figure 5A), suggesting that ethanol catabolism is necessary for Drosophila to discriminate between the two conditions. Drosophila preferred the co-culture containing A. pomorum WT to the co-culture containing A. pomorum adhA (Fig. 5A). Moreover, Drosophila preferred to lay eggs in the co-culture containing A. pomorum WT to the co-culture containing A. pomorum adhA, further substantiating the importance of ethanol catabolism in Drosophila biology, as egg-laying decisions are crucial to reproductive success (Figure 5B).
We next identified additional metabolites unique to the co-culture using solid-phase microextraction gas chromatography-mass spectrometry (SPME GC-MS). Acetic acid, six acetate esters, an acetaldehyde metabolic derivative (acetoin), a putative acetaldehyde metabolic derivative (2,4,5-trimethyl-1,3-dioxolane), and two unknown metabolites were more abundant in the co-culture relative to the separate-culture mixture or co-culture with A. pomorum lacking adhA (Table 2, Table 2-figure supplements 1-2). To determine the molecular basis for Drosophila co-culture preference, select metabolites were added to the co-culture containing A. pomorum adhA. Esters and acetic acid, but not esters alone, were sufficient to fully restore A. pomorum adhA attractiveness to A. pomorum WT levels (Figure 5A).
Although acetate and its metabolic derivatives were sufficient for Drosophila co-culture preference, acetaldehyde is a reactive intermediate during ethanol catabolism whose metabolic derivatives might be increased in microbial communities compared with individual microbial cultures. Consistent with this idea, acetoin was moderately increased in the co-culture compared with the separate-culture mixture (Table 2-figure supplement 1); strikingly, acetoin was increased ∼27-fold in the tri-culture (S. cerevisiae, A. malorum, and L. plantarum) compared to the co-culture (Figure 6A, B, Figure 6-figure supplement 1) and was attractive to Drosophila (Figure 6-figure supplement 2). In sum, emergent metabolites from 2- and 3-membered communities, including acetaldehyde metabolic derivatives, attract Drosophila.
To investigate further the potential role of acetaldehyde and its metabolic derivatives in Drosophila behavior, we performed a dose response in which acetaldehyde was added to the separate-culture mixture (Figure 6-figure supplement 3A) to evaluate its ability to induce attractiveness to co-culture levels. Even at the lowest tested levels, acetaldehyde supplementation stimulated the separate-culture mixture to attractiveness levels equal to the co-culture (Figure 6-figure supplement 3A); moreover, acetaldehyde and its metabolic derivatives ‘over-restored’ preference of the separate-culture mixture beyond WT co-culture levels at its most attractive dose (Figure 6-figure supplement 3A). Finally, three acetaldehyde metabolic derivatives—acetoin, 1,1-diethoxyethane (an acetal), and 2,3-butanedione—were sufficient to induce the attractiveness of the separate-culture mixture to WT co-culture levels using concentrations of each metabolite at or below the physiological concentration of acetoin found in the tri-culture (Figure 6-figure supplement 3B). Creating a pure metabolite mixture comprised of key metabolic groups produced by microbial communities and identified in this study (esters, acetaldehyde metabolic derivatives, alcohols, acid) attracted Drosophila similarly to the co-culture (Figure 6-figure supplement 4A, B). Interestingly, the acetaldehyde metabolic derivatives alone were sufficient and the most attractive (Figure 6-figure supplement 4C, D), supporting acetaldehyde metabolic derivatives as potent Drosophila attractants.
Overall, our results suggest that both esters and acetaldehyde metabolic derivatives are keystone microbial community metabolites that attract Drosophila. We next created a simple 9-metabolite mixture in water (containing only 1 acid, 4 esters, and 4 acetaldehyde metabolic derivatives) and measured Drosophila preference toward this mixture in relation to the yeast-acetic acid bacteria co-culture, the yeast-acetic acid bacteria-lactic acid bacteria microbial community, or apple cider vinegar (ACV). The defined mixture used concentrations for each acetaldehyde metabolic derivative similar to the concentration of acetoin in the tri-culture and ester and acid concentrations that were in the range of detected in the co-culture. The defined 9-metabolite mixture was more attractive than all other conditions (Figure 6-figure supplement 5). In sum, acetaldehyde metabolic derivatives and esters are potent Drosophila attractants whose detection may signal the presence of actively metabolizing, multispecies microbial communities.
Discussion
Here, we have demonstrated how emergent properties of a microbial community—volatile profile, population dynamics, and pH—influence Drosophila attraction, survival, and egg-laying behaviors. Our study is the first to identify the consequences of microbe-microbe metabolic exchange on animal behavior and discovers additional microbial interactions for further mechanistic study (Figure 1D).
Microbe-microbe metabolic exchange generates unique and quantitatively different volatiles from those resulting from individual microbial metabolism (Tables 1 & 2, Figure 7). Acetobacter-generated acetate coupled to Saccharomyces-derived alcohols spawn diverse acetate esters (Table 1 & Table 2). We hypothesize that more complex and diverse communities, comprising alcohol-producing yeasts, acetate-producing Acetobacter, and lactate-producing Lactobacillus, will generate a wider array of attractive esters (Figure 7). The community of S. cerevisiae, A. malorum, and L. plantarum emitted higher levels of acetoin and attracted Drosophila more strongly than the co-culture of S. cerevisiae and A. malorum (Figure 1D, Figure 6). Acetoin and 2,3-butanedione are formed by an α-acetolactate intermediate in bacteria and directly from acetaldehyde in yeast (52). We therefore hypothesize that communities of yeasts and bacteria may emit high levels of attractive acetaldehyde metabolic derivatives (Figure 7).
Drosophila behavioral studies have mostly focused on yeasts, even though Drosophila evolved in a bacterial-rich environment. Our results suggest that non-yeast microorganisms, especially when grown in microbial communities, affect Drosophila behaviors. We reason that additional investigations that couple chemical microbial ecology with Drosophila behavior will herald the discovery of microbe-influenced behaviors and microbial community-generated metabolites.
This study demonstrates the coordination of ethanol synthesis and catabolism by S. cerevisiae and Acetobacter, respectively, and the role of ethanol in Drosophila behavior and survival. Non-Saccharomyces Drosophila microbiome members also make ethanol (53) and diverse acetic acid bacteria catabolize ethanol, generalizing our findings to other microbial community combinations. Ethanol can have deleterious or beneficial fitness consequences for Drosophila depending on concentration (54, 55) and ecological context (56). Our results show that Drosophila uses products of inter-species microbiome metabolism to detect a community that titrates ethanol concentration optimally for the host. Work that further dissects the consequences of acetic acid and ethanol concentrations on Drosophila biology and investigates other community-level metabolic profiles will be of interest to enrich the chemical and ecological portrait of the Drosophila microbiome.
Our work raises questions about the consequences of the observed behavior on microbiome assembly and stability in the Drosophila intestine. Drosophila possesses specific and regionalized gut immune responses to the microbiome (57-60) implying a tolerant environment in which privileged microbiome members are maintained and reproduce in the Drosophila intestine. Other work suggests that Drosophila acquires its adult microbiome from exogenous sources, that adult microbiome abundance drops without continuous ingestion of exogenous microorganisms, and that the microbiome can be shaped by diet (19, 25, 26). As such, a combination of internal mechanisms, exogenous factors, and host behavior likely sculpt the microbiome; determining the relative contribution of each will be important moving forward. Complicating our understanding of the contribution of these factors is the opaque distinction between ‘microbiome’ and ‘food’, since both are ingested from the environment (61). To dissect the formation and stability of the Drosophila microbiome, the fate of ingested microorganisms needs to be monitored and microbial intestinal reproduction needs to be surveyed as a function of Drosophila behavior, age, immune status, microbiome membership, and nutritional state (e.g. using synthetic diets without yeast; (34, 62)).
In sum, our results support a model in which the Drosophila olfactory system is tuned to fruity (e.g., esters) and buttery (several acetaldehyde metabolic derivatives, such as 2,3-butanedione) smelling metabolites promoted by microbe-microbe interactions. We anticipate that accounting for microbial interactions in diverse host-microbe studies will lead to new insights into diverse aspects of microbial-animal symbioses.
MATERIALS AND METHODS
Fly maintenance
Fly stocks, genotypes, and sources are listed in Methods Table 1. Drosophila melanogaster was reared at 25°C on a 12h:12h light: dark cycle on autoclaved food (5% yeast, 10% dextrose, 7% cornmeal, 0.6% propionic acid, 0.7% agar).
Microbial strains
Microorganisms used in this study are listed and described in table S1. Microorganisms were streaked onto yeast-peptone dextrose (YPD; 1% yeast extract (Becton Dickinson, and Company, Franklin Lakes, NJ, USA), 2% peptone (Becton Dickinson, and Company, Franklin Lakes, NJ, USA), and 2% dextrose (Avantor Performance Materials, Center Valley, PA, USA)) or Man, de Rosa, Sharpe (MRS, Fisher Scientific, Waltham, MA, USA) plates from a freezer stock.
T-maze olfactory attraction assays
The T-maze apparatus was a kind gift of the Carlson Laboratory. Flies were wet-starved for 15-26 hours prior to T-maze olfactory experiments by placing flies into vials containing Kimwipes (Kimberly Clark, Dallas, TX, USA) soaked with 2 mL of milliQ water. Flies were collected within 4 days (<65 flies per vial) of emergence and matured on autoclaved food. Flies between 3 and 10 days-old were used in experiments.
Single microbial colonies were picked from rich media (MRS and YPD) plates and grown overnight. Cultures were washed 1X in PBS, diluted 100-fold, and 10µl was aliquoted into 3mL of apple juice media (AJM, apple juice (Martinelli’s Gold Medal, Watsonville, CA, USA), pH adjusted to 5.3 with 5M NaOH, with 0.5% yeast extract). Media was filtered with a 0.22 µM-size pore attached to a 250-mL polystyrene bottle (Corning, NY, USA). For co-culture experiments, 1e3-1e5 CFU of each microorganism was placed simultaneously into AJM. Microorganisms were grown in 14-mL round bottom polypropylene tubes (Corning Science, Tamaulipas, Mexico) at 28°C, 200rpm for the time noted in individual experiments. The microbial culture was diluted 1:1 with sterile milliQ water (0.22 µM filter (Millipore, Billerica, MA, USA)) and placed directly onto autoclaved 10 mM round Whatman filter paper (GE Healthcare Life Science, Pittsburgh, PA, USA) placed near the bottom of 15mL CentriStar centrifuge tubes (Corning, NY). A total volume of 10 µl was used for all experiments.
Tubes containing microorganisms and Drosophila were placed into the behavioral room (20-25°C, 50-70% humidity maintained by a humidifier (Sunbeam Tower Humidifier, Boca Raton, FL, USA) and equilibrated for 10 minutes prior to the beginning of the experiment. Flies (∼40-130) were knocked into the T-maze apparatus, rested for ∼1 min, and then allowed to choose from the test and control arms for 2 min in the dark. The test arm was alternated from one side of the apparatus to the other every experimental replicate. A Response Index (RI) was computed to analyze preference for the test arm (flies in test arm-flies in control arm)/(total flies).
Chemicals
Chemicals can be found in Methods table 2.
Microbial populations and pH
Selective plates were used to distinguish S. cerevisiae from A. malorum. MRS containing 50 µg/mL cycloheximide selected for A. malorum while MRS containing 10 µg/mL chloramphenicol and 20 µg/mL tetracycline selected for S. cerevisiae. pH of filtered cultures (0.22 µM) was measured using a Beckman Coulter pH meter (Model Phi510, Fullerton, CA, USA).
Gas chromatography-Mass spectrometry
Microbial samples were grown in AJM for a specified amount of time in 14-mL round bottom tubes fitted with an autoclaved tissue strainer (250 µM nylon mesh (Thermo Scientific Pierce, Grand Island, NY) holding between 0.03 and 0.05 grams of autoclaved Amberlite XAD-4 resin (Sigma-Aldrich, St Louis, MO, USA) prewashed in water and methanol. After microbial growth, XAD-4 from two cultures was dumped into an autoclaved glass vial. XAD-4 was swirled with 900µl methanol for 30 seconds. 500-750µl of methanol was removed for GC-MS analysis. Quantification for table S2 was derived from beads suspended above the cultures from 84-96 h of growth. Quantification for Fig. 2D was derived from beads suspended above the culture every 12 h; time points on the graph refer to the end point of the 12-h span (e.g. 84 h corresponds to beads suspended from 72-84 h of growth).
Samples (5 µl of methanol-extracted samples) were injected into the GC-MS (Agilent 7890A/5975C) at 250 °C using helium as the carrier gas at a flow rate of 1.1 mL per minute (column head pressure 13 psi). The following chromatography temperature program was used for experiments to initially identify metabolites in the co-culture and individually grown microorganisms: 40°C for 3 min ramped at 1.7°C per minute to 200 °C (held for 3 min) then to 220°C at 3 °C per min and held for a further 5 min. The total run time was 111.78 min. For experiments focused on the five major metabolites, a shorter program was used that maintained the same first 10 min of the previous method (all 5 volatiles eluted within 9 min). The chromatography temperature program was 40°C for 3 min, ramped at 1.7°C per min to 46.8°C and held for 3 min, then ramped at 60°C per min until 220°C and held for 5 min. The total run time was 17.9 min.
The mass spectrometer was run in electron-impact (EI) mode at 70 eV. The temperatures of the transfer line, quadrupole, and ionization source were 220°C, 180°C, and 230°C respectively. The ionization was off during the first 4 min to avoid solvent overloading with a source temperature of 230°C. Mass spectra were recorded in the range of 35-300 m/z. Full scan mode was used at a scan rate of 6 scans/sec. The electron multiplier voltage was set in the relative mode to autotune procedure.
In the initial experiments peaks were manually picked using Agilent Chemstation Software. Volatiles associated with peaks were searched against the National Institute of Standards (NIST) 11 database. Subsequent experiments focused on the 5 major volatiles identified in the initial experiments by performing extracted ion chromatograms using an ion that successfully identified a standard at a specific retention time. Quantification was performed by tabulating the maximum abundance of the ion at a characteristic retention time and using a linear regression equation from a dose-response of the standards (fig. S6).
Headspace solid phase microextraction (SPME) Gas chromatography-Mass spectrometry
A Waters GCT Premier gas chromatography time of flight mass spectrometer (Milford, MA) with a DB-5MS column (30m x 0.25 mm ID x 0.25 μm film thickness; Agilent) was used. Live cultures were transferred to autoclaved glass vials (20 mL, 23 X 75 mm, Supelco, Bellefonte, PA, USA) with screw caps (18mm, 35 Shore A, Supelco, Bellefonte, PA, USA) after growing for 72 h.
The glass vials containing live microbial cultures were analyzed via a 50/30μm carboxen/divinylbenzene/polydimethylsiloxane Stableflex solid-phase micro-extraction (SPME) fiber. The extraction methodology was based on previous studies using SPME to extract volatiles form vinegars (65, 66). The syringe was inserted through the membrane of the caps and sampled the volatiles for 30 min at 45°C; subsequently, metabolites were desorbed for 30 sec at 240C and baked for an additional 4.5 min in the injection port. The gas chromatograph was fitted with a microchannel plate (MCP) detector. The temperature program of the column was as follows: 40°C for 5 min, 2°C a min for 17.5 min followed by 25°C a min for 10 min. A final hold time of 5 min at 325°C was used. The carrier gas was helium. A split ratio of 250 was used based on better peak resolution. An internal standard of cineole (Sigma-Aldrich, St. Louis, MO, USA) was run with each sample and used to compute relative abundances. The mass detector was in the range of 40 to 650 m/z.
To analyze the data, MassLynx software was used. The response threshold was set to an absolute area of 10.00. The software automatically picked out peaks and computed peak areas. To obtain a relative quantification, peaks were compared across samples and normalized to the internal standard. Peaks were first searched against the NIST5 database to identify potential hits. Most potential metabolites were confirmed by a standard mixture in 50% AJM. The standard mixtures are in Methods Table 3.
Chemical complementation of co-culture containing A. pomorum PQQ-ADH-I
A co-culture containing S. cerevisiae and A. pomorum WT or A. pomorum PQQ-ADH-I was grown for 72 h before use in the T-maze. For the physiological concentrations of acetate-derived metabolites, concentrations were added as in table S3 and then mixed 1:1 with water prior to behavioral analysis. For the acetaldehyde metabolic derivatives chemical complementation group, a 1:1 mixture of the mutant co-culture: water was supplemented with 1,1-diethoxyethane, 2,3-butanedione, and acetoin at final concentrations of 0.01%, 0.15% and 0.15%, respectively and used immediately in the T-maze. Acetic acid and/or acetaldehyde were added to the culture, allowed to sit at RT for 35 min, mixed 1:1 with water and then placed into the T-maze vials.
Standard curves were used to calculate the concentrations of individual metabolites (fig. S10). The standard curves were generated on 2 separate experiments in which 3 concentrations of each standard was used. The concentrations of the metabolites were independently calculated from the standard curve equations generated on the two separate days. Estimated concentrations from each standard curve equation were averaged (table S3). The experimental data is based on the peak areas of the S.cerevisiae-A. malorum co-culture.
Ester, acid, and acetaldehyde metabolic derivative mixture
The 9-metabolite mixture contains 1.5% acetic acid, 0.0003% isoamyl acetate, 0.0003% 2-phenethyl acetate, 0.01% ethyl acetate, 0.002% ethyl lactate, 0.3% 1,1-diethoxyethane, 0.3% 2,3-butanediol, 0.3% 2,3-butanedione, and 0.3% acetoin in filtered milliQ water.
Drosophila survival in the presence of ethanol and acetic acid
Adult male flies (0-3 d-old) were collected and matured for 1 day on fly food. Flies were then placed into vials containing kimwipes with 5 mL of either Shields & Sang Insect Medium (Sigma, St. Louis, MO; positive control), MilliQ water (negative control), or MilliQ water with ethanol (9.4%), acetic acid (3.42%), or ethanol and acetic acid (1.4 and 2.8% respectively). Survival was assessed every 12 h for 7 d. For each condition 5 mL was given at 0 and 12 h and every 24 h thereafter. Experimental replicates were considered separate vials (5-6 per group). Each replicate contained 8-31 flies.
Egg-laying preference assay
Egg-preference assay was adapted from Joseph et al 2009 (67). Microbial cultures grown for 96 h were heated to 65°C for 10 min, mixed 1:1 with 1.6% agarose and poured into a 35 mm petri dish separated in two by a straight-edge razor blade. Flies were starved for ∼18 h prior to the experiment. The 35 mm petri dish was placed within clear flat top boxes with dimensions 2 5/16” X 2 5/16” X 5 1/16” (TAP plastics, San Leandro, CA, USA). The test and control sides were alternated for each replicate. Drosophila aged 4-10 days (n=50-100) was allowed to lay eggs for 8 h. After the assay, the number of eggs on deposited on each choice was tabulated and an egg-laying index was computed analogously to the olfactory response index.
Data analysis
Data analysis was performed in Prismv6.0b. Specific statistical tests are noted for individual experiments.
Acknowledgements
We thank Dr. Fabian Staubach (Stanford University, USA), Dr. Angus Chandler (University of California, Berkeley, USA), Dr. Matthew Goddard (The University of Auckland, New Zealand), Dr. Dan Jarosz (Stanford University, USA) and the Phaff Yeast Culture Collection (UC Davis, USA) for microbial strains. We thank Dr. Ryan Joseph and Dr. Karen Menuz (Yale University/University of Connecticut) for help with the Drosophila behavioral experiments; the Yale West Campus Analytical Core for use of the GC-MS; Dr. Terence Wu and Dr. Eric Patridge (Yale University, USA) for advice and experimental help with the GC-MS; Dr. Scott Strobel and Michelle Legaspi for use of and help with SPME GC-MS; Dr. John Carlson (Yale University, USA) for the T-maze and several Drosophila lines. We also thank Dr. Won Jae Lee (Seoul National University, S. Korea) for the A. pomorum WT and adhA strains, and John Carlson, Craig Crews, and Andrew Goodman (Yale University, USA) for helpful discussions. This work was supported by National Institutes of Health Grants NIDDKRC12T32GM007499-36, 5T32HG003198-10, 1R01GM099563, 7RC1DK086831. The raw data for the metabolomics experiments can be found at https://figshare.com/account/home#/projects/4735.