Activation of Vibrio cholerae quorum sensing promotes survival of an arthropod host

Abstract

Vibrio cholerae colonizes the human terminal ileum to cause cholera and the arthropod intestine and exoskeleton to persist in the aquatic environment. Attachment to these surfaces is regulated by the bacterial quorum sensing signal transduction cascade, which allows bacteria to assess the density of microbial neighbors. Intestinal colonization with V. cholerae results in expenditure of host lipid stores in the model arthropod Drosophila melanogaster. Here we report that activation of quorum sensing in the Drosophila intestine retards this process by repressing V. cholerae succinate uptake. Increased host access to intestinal succinate mitigates infection-induced lipid wasting to extend survival of V. cholerae-infected flies. Therefore, quorum sensing promotes a more favorable interaction between V. cholerae and an arthropod host by reducing the nutritional burden of intestinal colonization.

Introduction

In a process known as quorum sensing, bacteria produce small molecules termed autoinducers that accumulate in the environment as a function of cell density. Upon reaching a particular concentration threshold, these autoinducers alter collective bacterial gene transcription in a fashion that is hypothesized to be adaptive. Quorum sensing may thus allow a bacterium to respond to environmental conditions as a multicellular rather than unicellular organism. Behaviors that benefit from coordinated multicellular action include utilization of available nutrients ^{1, 2}, surface attachment or biofilm formation ^3–5, natural competence ^6–8, and regulation of virulence and symbiosis strategies ^{9, 10}. Quorum sensing by the human diarrheal pathogen Vibrio cholerae modulates all of these ^{1, 5, 11, 12}.

V. cholerae is adapted to the human intestine and the aquatic environment, two very different milieus. Colonization of the human small intestine, a region of the gut with a relatively low burden of commensal bacteria ¹³, is dependent on the toxin co-regulated pilus (TCP) ¹⁴. Cholera toxin (CTX), which is synthesized in the small intestine, then induces the osmotic diarrhea of cholera ^{15, 16}. V. cholerae also lives in close proximity to arthropods in terrestrial, estuarine, and fresh water environments ^17–21. The V. cholerae biofilm exopolysaccharide VPS enables this pathogen to accumulate on abiotic surfaces and colonize the arthropod exoskeleton and intestine ^22–24. TCP, CTX, and VPS are all repressed at high cell density by the quorum sensing circuitry ^{5, 11, 12, 25}. The quorum sensing signaling cascade also promotes proteolytic degradation of environmental phages, resistance against grazing by amoebae and flagellates, and persistence within the amoebal contractile vacuole ^26–28. Here we show that V. cholerae quorum sensing is activated in the Drosophila intestine and attenuates pathogen virulence as a result of decreased succinate uptake. This quorum sensing-regulated reduction in bacterial succinate transport delays host expenditure of triglycerides, reproductive organ atrophy, and death. Taken together, our studies suggest that a functional V. cholerae quorum sensing system benefits both host and pathogen by minimizing the nutritional impact of infection.

Results

The V. cholerae quorum sensing master regulator HapR represses biofilm matrix synthesis in the Drosophila intestine at high pathogen burden.

The V. cholerae quorum sensing master regulator HapR decreases V. cholerae biofilm formation at high cell density due to transcriptional repression of the vps genes, which encode the VPS synthesis machinery ^{5, 11, 22}. To confirm this phenotype in the quorum sensing-competent strain C6706 originally isolated from Peru in 1991, we measured biofilm formation in a C6706 ΔhapR mutant ²⁹ (Fig. 1a). V. cholerae VPS-dependent biofilm formation is required for colonization of the Drosophila gut by the quorum sensing-deficient clinical strain MO10 ^{24, 30}. We questioned whether wild-type C6706, which forms very little biofilm in a test tube, could colonize the Drosophila intestine. To measure this, we performed a colonization transfer experiment in which the fly is given access to V. cholerae resuspended in PBS for 48 hours, transferred to sterile PBS for 24 hours to allow for expulsion of unattached V. cholerae, surface sterilized, and homogenized. After the wash-out period, Drosophila exposed to wild-type C6706 as well as the ∆hapR mutant remained colonized, while those exposed to the ∆hapR ∆vpsA mutant did not (Fig. 1b). We then gave flies access to wild-type and mutant V. cholerae expressing the green fluorescent protein (GFP) from a chromosomal location, removed the intestines, and assessed colonization by fluorescence microscopy. As shown in Figure 1c, wild-type C6706 and the ∆hapR mutant were abundant in both the midgut and rectum of the fly, while the C6706∆hapR∆vpsA mutant colonized these regions only minimally. This demonstrates that the quorum-sensing competent V. cholerae strain C6706 forms a VPS-dependent biofilm in the Drosophila intestine. To determine whether high cell density, HapR-dependent repression of biofilm formation was ultimately activated in this setting, we measured vpsL gene expression in the intestines of flies colonized with wild-type C6706 or a ΔhapR mutant. In fact, transcription of vpsL was much higher in flies infected with a C6706∆hapR mutant, demonstrating that HapR represses vps gene transcription, not only on abiotic surfaces but also in the colonized fly intestine (Fig. 1d). These data are consistent with a model in which quorum sensing-competent V. cholerae initially accumulate in the Drosophila intestine. When high cell density is reached, HapR represses transcription of genes that promote colonization.

Figure 1: V. cholerae high cell density quorum sensing-dependent signaling is activated in the Drosophila intestine and promotes host survival of infection.

(a) Comparison of biofilm formation by V. cholerae strain C6706 isolated from a Peruvian patient in 1991 and the corresponding ∆hapR mutant. Three independent biological replicates were performed. Bars represent the mean measurement, and error bars indicate the standard deviation. A Welch’s t-test was used to calculate significance. (b) Average CFU/intestine for Drosophila infected for 48 hours with C6706, the corresponding quorum sensing mutant C6706∆hapR, and the double quorum sensing and biofilm mutant C6706∆hapR∆vpsA. Horizontal bars represent the average of six biological replicates. A students t-test was used to calculate significance. (c) Representative micrographs showing the midgut and rectum of flies infected for 48 hours with the indicated V. cholerae strains expressing GFP and then washed out for 24 hours. Ten midguts/rectums, each representing an independent biological replicate, were examined. Nuclei were stained with DAPI. Scale bars, 20µm. (d) qRT-PCR measurements of vpsL transcription by the indicated V. cholerae strains in the Drosophila intestine. Three independent biological replicates were performed. Bars indicate the mean measurement, and error bars represent the standard deviation. A student’s t test was used to calculate significance. (e) Survival over time of flies infected with the indicated strains. Three cohorts of ten co-housed flies representing three independent biological replicates were used. Statistical significance was calculated by log rank analysis. (f) Intestinal bacterial burden and (g) PH3⁺ staining of the indicated V. cholerae strains after infection for 48 hours. Horizontal bars represent the average of six to seven biological replicates in (f) and (g). A Mann-Whitney test was used to calculate significance.

Quorum sensing attenuates V. cholerae virulence.

We questioned whether quorum sensing might also limit pathogenesis. In fact, deletion of hapR accelerated host death without significantly increasing intestinal colonization (Fig. 1e-f). We previously showed that diminution of intestinal stem cell division accompanies infection with MO10 ³¹. To determine the effect of C6706 and a ∆hapR mutant on host intestinal stem cell proliferation, we quantified cells with phosphorylated histone 3 (PH3), a cell division marker (Fig. 1g). Stem cell division was preserved in the intestines of flies infected with wild-type C6706 but not the ∆hapR mutant. These results suggest that V. cholerae quorum sensing prevents infection-induced arrest of epithelial renewal.

A functional quorum sensing cascade is retained in Haitian strains.

An intact quorum sensing pathway is variably found in V. cholerae species isolated from the clinic and the environment ³⁰. Whether these spontaneous quorum sensing mutations accumulate in the field or laboratory is unknown. We hypothesized that quorum sensing should be preserved over time if it provided a selection advantage in natural environments. In October of 2010, a quorum sensing-competent strain from Southeast Asia was introduced into the Haitian population and environment as the result of an index case of cholera contracted in Nepal and disseminated in the Artibonite valley ^32–34. This provided the unique opportunity to track inactivation of V. cholerae quorum sensing after introduction and dispersal in a new environment ³⁵. To study this, we deleted hapR in Haitian strains isolated between 2010 and 2013 and assessed biofilm formation. As shown in Figure 2a, biofilm formation was greatly increased in ∆hapR mutants. As additional strains became available over the course of our studies, they were tested, and these also exhibited repression of biofilm formation by HapR (Supplementary Fig.1). Similar to strain C6706, we observed minimal mortality in Drosophila infected with Haitian strains but equivalent or higher bacterial burdens as compared with the quorum sensing-deficient clinical strain MO10 (Fig. 2b-c). Furthermore, we did not detect a significant difference in the total burden of commensal bacteria in the intestines of infected and uninfected flies (Fig. 2d). We conclude that decreased virulence is not the result of a V. cholerae colonization defect or a change in the number of commensal microbiota. As observed for C6706, deletion of hapR in the Haitian strains accelerated the mortality of infected flies and suppressed intestinal stem cell division (Fig. 2e-f). Taken together, our findings suggest that quorum sensing-regulated repression of V. cholerae virulence in this host is retained over years and possibly decades.

Figure 2: V. cholerae quorum sensing promotes host survival independent of biofilm formation.

(a) Comparison of biofilm formation by wild-type V. cholerae and the corresponding ∆hapR mutants collected from Haiti in the indicated year. Quantification is shown below, and representative biofilm images are shown above. The mean of three independent biological replicates is shown. Error bars represent the standard deviation. A Welch’s t-test was used to calculate significance. (b) Survival of flies infected with V. cholerae O139 strain MO10 collected during the 1992 epidemic in Southeast Asia and three clinical strains collected in Haiti during 2010, 2012, and 2013. Three cohorts of ten co-housed flies representing independent biological replicates were used. Statistical significance was calculated by log rank analysis. (c) V. cholerae burden and (d) commensal burden of uninfected Drosophila (uninf) or those infected with the Haitian strains used in (b). Each symbol represents a biological replicate. Horizontal bars represent the mean measurement. A one-way ANOVA using MO10 as a control with Dunnett’s multiple comparison test was used to calculate statistical significance. Adjusted p values are reported. (e) Survival of flies infected with the indicated ∆hapR mutant strains. Survival of flies infected with the parental 2012 strain is shown for comparison. Three cohorts of ten co-housed flies representing independent biological replicates were used. Statistical significance was calculated by log rank analysis. (f) Quantification of PH3⁺ cells in the intestines of flies infected with the indicated strains. Each symbol represents a biological replicate, and horizontal bars represent the mean measurement. A Mann-Whitney test was used to calculate statistical significance. (g-i) Survival of flies infected with either ∆hapR or ∆hapR∆vpsA mutant V. cholerae strains isolated from Haiti in the indicated year. Three cohorts of ten co-housed flies representing three independent biological replicates were used. Statistical significance was calculated by log rank analysis.

Biofilm formation is not responsible for the increased virulence of V. cholerae ∆hapR mutants.

We previously showed that V. cholerae biofilm formation contributes to virulence in the fly model ²⁴. To determine whether the increased virulence of the ∆hapR strains was due to activation of biofilm formation, we created ∆hapR∆vpsA double mutants, which are unable to form a biofilm. The virulence of the ∆hapR∆vpsA double mutants was similar to that of the ∆hapR mutants (Fig. 2g-i). This suggests that an aspect of HapR regulation other than repression of biofilm formation is responsible for the increased virulence of the ∆hapR mutants.

Quorum sensing impacts pathogen nutrient uptake and metabolite secretion.

Mutation of both V. cholerae acetyl-CoA synthase (Acs1) and the glycine cleavage system (Gcv) reduces virulence in the MO10 quorum sensing-defective background due to decreased bacterial uptake of the survival-prolonging metabolites acetate and methionine sulfoxide, respectively ^36–38. However, when Drosophila are simultaneously infected with equal numbers of MO10Δacs1 and ΔgcvT mutants, rapid mortality comparable to that of the parental strain ensues because each mutant consumes the life-extending metabolite that the other cannot ³⁶. We hypothesized that quorum sensing might similarly prolong host survival by sparing a life-extending metabolite not spared by a quorum sensing-defective strain. To test this, we first co-infected flies with MO10 and each of the Haitian strains. Consistent with our hypothesis, co-infected flies expired at rates similar to that of MO10-infected flies (Supplementary Fig. 2a). To determine whether the metabolite spared by quorum sensing strains was related to acetate or methionine sulfoxide, we performed co-infections with MO10∆acs1 or ∆gcvT mutants. We found that co-infection with any of the Haitian strains and the MO10∆gcvT mutant led to rapid death, comparable to that of the wild-type MO10 infection (Fig. 3a-d). This was not the result of one strain outcompeting its partner (Supplementary Fig. 2b-d). This suggests that the Haitian strains may consume or reduce extracellular methionine sulfoxide, the metabolite spared by the MO10∆gcvT mutant, in the fly intestine. In contrast, co-infection of Haitian strains with the MO10∆acs1 mutant resulted in a survival phenotype similar to solo infection with either of the infecting strains. These data suggest that co-infection with any Haitian strain and the MO10∆acs1 mutant results in accumulation of a survival-prolonging metabolite. To determine whether this metabolite was acetate, we quantified acetate uptake by the Haitian V. cholerae strains and corresponding ΔhapR mutants cultured in LB broth. Both parental and mutant strains consumed acetate quite rapidly as compared with the MO10Δacs1 mutant (Fig. 3e-h). This suggested that acetate was not the survival-prolonging metabolite associated with Haitian strains, that increased consumption of acetate does not underlie the augmented virulence of ΔhapR mutants, and, finally, that the Δacs1 mutant is unable to consume the survival-prolonging metabolite associated with Haitian strains.

Figure 3: V. cholerae quorum sensing and succinate supplementation prevent host lipolysis.

(a) Survival of flies infected with the wild-type V. cholerae strain MO10, and the corresponding ∆gcvT and ∆acs1 mutants. (b-d) Survival of flies infected with co-cultures of the indicated Haitian strain and MO10∆gcvT or ∆acs1 strain. (e-h) Quantification of acetate in the supernatants of the indicated wild-type V. cholerae strains and the corresponding ∆hapR mutant cultured for the indicated lengths of time. A student’s t test was used to calculate significance. (i) Significantly different metabolites present in the spent supernatants of the 2013 Haitian V. cholerae strain that are differentially taken up by the ∆hapR mutant as compared with the non-biofilm forming wild-type and ∆hapR∆vpsA mutant strains. parahydroxybenzoate (p-hb). (j) Metabolites that are differentially consumed by both the ∆hapR and ∆hapR∆vpsA mutant strains. A lower peak intensity threshold of 5 was used to eliminate less abundant metabolites. Data was analyzed as described in the text. (k) Survival of flies infected with the ∆hapR mutant strain in LB alone or supplemented with 5mM concentrations of succinate, glutamate, and threonine. (l) Survival of flies infected with the ∆hapR mutant strain in LB alone (CTL) or supplemented with succinate (5mM) at 0 or 72 hours. (m) Representative micrographs showing Bodipy staining of lipid droplets in the fat bodies of flies exposed for 4 days to LB broth alone or inoculated with V. cholerae from 2013, the corresponding ∆hapR mutant, or the corresponding ∆hapR mutant supplemented with succinate. Ten fat bodies, each representing an independent biological replicate, were examined. Scale bars, 50µm. Quantification of (n) total triglycerides, (o) total glucose, and (p) total protein in flies treated as described in a-d. Three cohorts of ten co-housed flies representing independent biological replicates were used. A one-way ANOVA using 2013 as a reference with Dunnett’s multiple comparison test was used to calculate statistical significance. For all bar graphs, bars represent the mean of three independent biological replicates, and error bars represent the standard deviation. For all survival experiments, three cohorts of ten co-housed flies representing independent biological replicates were used. Statistical significance was calculated by log rank analysis.

To identify other differentially consumed metabolites, we performed metabolomic analysis on the spent supernatants of the Haitian strain isolated in 2013 and the corresponding ΔhapR and ΔhapR ΔvpsA mutants (Supplementary Table 1 and 2). Although biofilm formation did not significantly alter the virulence of the ΔhapR mutants, we hypothesized that it might alter their metabolism. We first identified several small molecules present in significant quantities that were differentially consumed by the ΔhapR mutant as compared with the wild-type strain and the ΔhapR ΔvpsA mutant (Fig. 3i). These findings demonstrate that biofilm formation is associated with consumption of a unique set of metabolites that do not alter virulence. These metabolites were eliminated from further study. We then identified small molecules that were consumed by the ΔhapR and ΔhapR ΔvpsA mutant strains but not the parental strain (Fig. 3j). We hypothesized that uptake of one or more of these metabolites could be responsible for the increased virulence of the ΔhapR and ΔhapR ΔvpsA mutants.

Quorum sensing represses V. cholerae succinate uptake to prolong host survival.

Glutamate, succinate, and threonine, metabolites that were nearly completely consumed by the ΔhapR mutants, were investigated. To identify metabolites whose uptake by the ∆hapR mutant were responsible for increased virulence, we prepared LB medium supplemented with glutamate, succinate, and threonine in 5 mM concentrations and infected flies with the ΔhapR mutant in this medium. As shown in Figure 3k, supplementation significantly prolonged survival of flies infected with a ΔhapR mutant. We then exposed flies to LB inoculated with the ΔhapR mutant and supplemented with additional mixtures of metabolites and found that, in each case, increased survival was correlated with the presence of succinate in the mixture (Supplementary Fig. 2e). To test the efficacy of succinate in prolonging survival, we performed infections with the ΔhapR mutant in the presence of succinate alone, which did not impact V. cholerae growth (Supplementary Fig. 2f). Increased survival was observed, which was greater when succinate was added later in infection (Fig. 3l).

Because succinate uptake was increased in the ∆hapR mutant, we hypothesized that the V. cholerae succinate uptake transporter Vc INDY might be repressed by HapR ³⁹. Indeed, we found that transcription of the gene encoding Vc INDY was increased 3-fold in the ∆hapR mutant (Supplementary Fig. 2g). We propose a model in which HapR represses V. cholerae succinate uptake at high cell density, resulting in an increase in dietary succinate available to the host.

Our data suggest that the MO10Δacs1 mutant does not consume succinate. Interestingly, acetyl-CoA, the product of Acs1, and succinate are key substrates of the tricarboxylic acid (TCA) cycle. One possibility is that intracellular levels of these two carboxylic acids are linked.

Quorum sensing and succinate preserve adipose tissue lipid stores during V. cholerae infection.

Infection-induced depletion of lipids from the Drosophila adipose tissue known as the fat body along with accumulation of large lipid droplets in the intestine is correlated with host mortality ^{36, 38}. To determine whether lipid re-distribution also occurred during infection with the Haitian V. cholerae strain, we assessed triglyceride accumulation in the Drosophila fat body after four days of access to LB alone or inoculated with V. cholerae. As compared with flies fed sterile LB, lipid stores in the fat bodies of flies infected with wild-type V. cholerae decreased less than those in the fat bodies of flies colonized with a ΔhapR mutant (Fig. 3m). Furthermore, supplementation with succinate increased the lipid stores of flies infected with the ΔhapR mutant (Fig. 3m). To determine if this was accompanied by the accumulation of large lipid droplets in the gut, we examined the intestines of flies infected with V. cholerae 2013 and the ∆hapR mutant strain alone or supplemented with succinate on the fourth day of infection. No accumulation of lipid droplets was observed in the intestines of these flies (Supplementary Fig. 3). This suggests that accumulation of lipid droplets in the gut does not drive depletion of fat body stores during Drosophila infection with the ΔhapR mutant.

We hypothesized that quorum sensing might promote conservation of host triglyceride stores. To test this, we performed quantitative assays of total triglycerides (Fig. 3n). This confirmed our hypothesis. Furthermore, while there was a trend towards rescue of triglyceride stores with succinate supplementation, this did not reach statistical significance after adjustment for multiple comparisons. Total glucose and protein were not significantly different among the conditions tested (Fig. 3o-p), and succinate supplementation in the absence of infection had no effect on lipid accumulation (Supplementary Fig. 4a-b) and did not alter total glucose or triglyceride stores (Supplementary Fig. 4c-d). This demonstrates that V. cholerae quorum sensing and succinate ingestion affect lipid stores in host adipose tissue specifically in the setting of infection.

Succinate ingestion does not impact host appetite.

We tested the possibility that quorum sensing and succinate might augment lipid stores by increasing Drosophila food intake during infection. The flies differentially avoided consumption of the ΔhapR mutant. However, by day 4, this difference in consumption was minimal (Supplementary Fig. 5a-b). Furthermore, succinate supplementation had no effect on food intake. These data do not support the hypothesis that quorum sensing and succinate modulate lipid stores through an effect on appetite.

A transcriptional response indicative of starvation is not observed during infection with a quorum sensing mutant.

Succinate is a key intermediate in the TCA cycle ^{40, 41}. We hypothesized that enterocyte succinate depletion due to pathogen ingestion of this metabolite might inhibit the TCA cycle, thus activating a starvation response. In an MO10 infection, we found that intestinal transcription of ImpL2, lnR, and dilp6 was activated and insulin signaling was extinguished, a pattern indicative of enterocyte starvation ^{38, 42–44}. This was reversed by mutation of acs1 or gcvT (Supplementary Fig. 5c-d) ³⁸. We questioned whether quorum sensing might also redirect bacterial metabolism to prevent enterocyte and host starvation. In fact, transcription of these genes either in the gut (Supplementary Fig. 5e-g) or in whole flies (Supplementary Fig. 5i-k) was not significantly altered on the fourth day after exposure to the Haitian V. cholerae strain, the corresponding ΔhapR mutant, or the ΔhapR mutant supplemented with succinate, and insulin signaling, as evidenced by phosphorylation of the serine/threonine kinase AKT, was not decreased (Supplementary Fig. 5m). Succinate supplementation similarly did not alter dilp6 transcription in the setting of starvation (Supplementary Fig. 5h-I). Taken together, these data suggest that quorum sensing and succinate supplementation do not impact enterocyte or host starvation in the setting of intestinal V. cholerae colonization.

Quorum sensing and succinate supplementation minimally impact host anabolic functions.

Succinate is also a substrate for the anabolic functions of gluconeogenesis and lipogenesis ^{40, 41}. We hypothesized that quorum sensing and succinate supplementation might promote one of these processes. As the first committed step in gluconeogenesis and glyceroneogenesis, phosphoenolpyruvate carboxykinase (PEPCK) decarboxylates and phosphorylates oxaloacetic acid in the cytoplasm. In addition, acetyl-CoA carboxylase (ACC) catalyzes the first committed step in fatty acid synthesis, the carboxylation of acetyl-CoA to form malonyl-CoA. PEPCK and ACC activity is regulated at the transcriptional level ^{45, 46}. Transcription of PEPCK was decreased by approximately a factor of 2 in the whole bodies but not intestines of flies infected with the ∆hapR mutant as compared with V. cholerae 2013 regardless of succinate supplementation (Supplementary Fig. 5n-p). This likely contributes to the decrease in lipid accumulation in the fat bodies of flies infected with the ∆hapR mutant but does not reflect a response to succinate. Transcription of Acc was unchanged in the bodies and intestines of flies infected with wild-type V. cholerae, a ∆hapR mutant alone, or ∆hapR mutant in the presence of succinate (Supplementary Fig. 5o-q). This suggests that quorum sensing and succinate do not rescue fly survival through transcriptional activation of gluconeogenesis or lipogenesis.

V. cholerae quorum sensing and succinate do not activate the Drosophila HIF1-α homolog Sima.

Succinate is a signaling molecule through its interaction with specific transcription factors and G-protein coupled receptors. It has been demonstrated to stabilize the α subunit of the hypoxia-inducible factor (HIF1-α) ⁴⁷. In Drosophila, the HIF1-α homolog Sima represses signaling through the IMD pathway. We predicted that activation of the IMD pathway in the setting of infection with a quorum sensing-competent V. cholerae strain might be decreased as compared with a quorum sensing-defective strain due to succinate stabilization of Sima. Because the IMD pathway activates transcription of antimicrobial peptides including Diptericin ⁴⁸, we measured diptericin transcription to assess output through the IMD pathway. However, HapR and succinate supplementation had no effect on diptericin transcription (Supplementary Fig. 5r-s).

Evidence that intestinal succinate extends host survival by inhibiting lipolysis.

We then hypothesized that infection with the V. cholerae 2013ΔhapR mutant might specifically activate lipolysis, and, furthermore, that quorum sensing and succinate might prolong host survival of infection by preventing this. To test this, we blocked lipolysis by two independent mechanisms. Drosophila insulin-like peptide 6 (Dilp6) and Brummer (Bmm), an essential lipase, are required for triglyceride mobilization from the fat body ^{42, 49}. To determine whether reduction of dilp6 and bmm expression in the fat body might increase host resistance to infection, we obtained a bmm¹ heterozygote, two independent dilp6 RNAi constructs, and a bmm RNAi construct. We used these tools to decrease dilp6 and bmm transcription in the fat body (Supplementary Fig. 6a-c) and then measured resistance to infection. As shown in Figure 4a-b and Supplementary Figure 6d, knockdown of dilp6 in the fat body increased host survival of infection but did not alter intestinal colonization. In addition, total triglyceride stores of flies infected with a 2013 ∆hapR mutant were similarly increased by knockdown of dilp6 and succinate supplementation (Fig. 4c). In this genetic background, we also observed a small but significant decrease in total glucose concentration as a result of infection, which was more pronounced in the ΔhapR mutant infection, but dilp6 knockdown in the fat body and succinate supplementation did not rescue glucose levels (Fig. 4d). Bodipy staining revealed triglyceride retention in the fat body without accumulation in the intestine (Fig. 4e-f). Flies with bmm knocked down in the fat body as well as bmm¹ heterozygotes displayed a phenotype very similar to that of dilp6^RNAi flies. They were resistant to infection with the 2013∆hapR strain (Supplementary Fig. 6d-g), and lipids were observed in the fat body but not the gut during infection (Supplementary Fig. 7). These results support our contention that quorum sensing prolongs the V. cholerae-arthropod interaction by reducing infection-activated catabolism of host lipid stores.

Figure 4: dilp6 knockdown inhibits host infection-induced lipolysis and prolongs host survival.

(a) Survival of flies with the indicated genotypes infected with the indicated V. cholerae strains. Dilp6^RNAi construct 111727 was used. Three cohorts of ten co-housed flies representing independent biological replicates were used. P values, corresponding to the relevant comparisons, are indicated on the graph. Log rank analysis was used to assess significance. (b) V. cholerae burden per fly for the indicated bacterial and Drosophila strains. Horizontal bars represent the mean. Seven cohorts of ten co-housed flies representing independent biological replicates were used. Statistical significance was calculated using a one way ANOVA with Dunnett’s multiple comparison test. (c)Total triglycerides and (d) total glucose in the indicated bacterial and Drosophila strains. Three biological replicates were performed. Bars represent the mean, and error bars represent the standard deviation. A one-way ANOVA using ∆hapR as a reference with Dunnett’s multiple comparison test was used to calculate statistical significance. An adjusted p value is given. Representative micrographs showing Bodipy staining of lipid droplets in (e) the fat bodies and (f) midguts of flies carrying yolk-Gal4 (yk) with or without UAS-dilp6 ^RNAi (VDRC strain 111727) and infected with V. cholerae (2013) or V. cholerae∆hapR (∆hapR). Nuclei are stained with DAPI. Ten fat bodies/midguts, each representing an independent biological replicate were examined. Scale bars, 50µm.

Quorum sensing and succinate supplementation preserve host reproductive organ size and triglyceride accumulation.

We noted that infection with a ∆hapR mutant caused the Drosophila ovaries to shrink (Supplementary Fig. 8a). This was reminiscent of Drosophila ovaries that shrink as a result of increased programmed cell death and decreased cell proliferation ⁵⁰. Consistent with this, while the structure of ovarioles from flies infected with the parental V. cholerae strain were normal in appearance, the ovariole nuclei of those infected with the ∆hapR mutant strain were shrunken and disorganized (Supplementary Fig. 8b). Succinate supplementation increased ovary size and restored normal morphology (Supplementary Fig. 8). We hypothesized that disruption of normal oogenesis might be due to modulation of lipid accumulation in the developing oocytes. Indeed, using Bodipy staining, we noted that the developing oocytes of ∆hapR-infected flies were dramatically depleted of lipid stores. This was also reversed by succinate supplementation. We conclude that V. cholerae high cell density HapR-dependent signaling represses pathogen succinate uptake in the intestine, thus increasing host access to this dicarboxylic acid. This, in turn, prevents expenditure of host lipid stores, maintains reproductive organ size, and protects the host against infection-induced mortality.

Discussion

Quorum sensing is a means of interbacterial communication effected by secretion of small molecules that coordinate bacterial behavior under conditions of high cell density ⁵¹. It has been shown to afford biofilm-associated V. cholerae resistance to consumption by amoebae and flagellates and may aid the persistence of V. cholerae within the amoeba contractile vacuole ^{26, 27}. Here we show that the V. cholerae master regulator HapR is activated in the intestine of infected Drosophila and reduces pathogen virulence by repressing transcription of the bacterial succinate transporter. Decreased V. cholerae succinate uptake increases host access to this metabolite, which prevents mobilization of host lipid stores and death.

V. cholerae quorum sensing pathways rapidly accumulate spontaneous inactivating mutations ^{52, 53}. Here we show that after the introduction of a single pathogenic V. cholerae strain into Haiti in 2010, quorum sensing competence was retained in a random selection of isolates over a period of six years in spite of evidence for diversification of other parts of the genome under positive selection ³⁵. This suggests that quorum sensing may provide a selection advantage to the bacterium.

While V. cholerae quorum sensing is not a proven component of a host-microbe symbiosis, the concept of quorum sensing as a symbiosis factor in the interaction of Vibrio species with environmental hosts is well-established ^{54, 55}. V. fischeri colonization of the squid Euprymna scolopes light organ is essential for light organ development ^56–58. Quorum sensing-regulated V. fischeri luminescence within this organ then protects the squid from predators ⁵⁹. Aquatic and terrestrial arthropods are thought to play an important role in survival of V. cholerae in the environment ^{17–19, 60, 61}. We show here that, at high cell density, V. cholerae quorum sensing represses the intestinal colonization factor VPS and mitigates the metabolic demands placed by this pathogen on its arthropod host.

Quorum sensing-mediated repression of V. cholerae succinate uptake has the potential to impact metabolism through multiple avenues in both mammals and arthropods. Similar to other TCA cycle intermediates, succinate participates in post-translational modification of lysine residues in proteins ⁶². Second, succinate stabilizes the hypoxia-induced transcription factor HIF1-α, which promotes weight gain and obesity in mice ⁶³ and suppresses the innate immune response ^{48, 64}. Last, the mammalian G-protein coupled receptor GPR91 or SUCNR1, which inhibits lipolysis in response to extracellular succinate, is abundant in cells of the intestinal brush border ^65–67. SUCNR1 mutant mice have less white adipose tissue and are resistant to succinate-induced inhibition of lipolysis, a host phenotype similar to the one reported here to be activated by V. cholerae quorum sensing ⁶⁸. We conclude that V. cholerae quorum sensing is not only a means of interbacterial communication but also an avenue through which a non-invasive intestinal bacterium may exert control over host metabolic pathways to extend the host-pathogen interaction and prolong environmental dispersal.

Methods

Bacterial strain and fly stocks.

Bacterial strains and fly stocks used are listed in Supplementary Table 3. V. cholerae strain MO10 and the corresponding mutants with in-frame deletions in the genes encoding acetyl-CoA synthase (acs1) and glycine cleavage system protein T (gcvT), were used for co-infection and gene transcription experiments ³⁸. With the exception of UAS-dilp6-RNAi stocks (constructs 3987 and 111727), which were obtained from the Vienna Drosophila Resource Center, fly lines were obtained from the Bloomington Drosophila stock center.

Drosophila husbandry and bacterial culture.

Fly stocks were maintained on standard fly food containing 16.5 g/L yeast, 9.5 g/L soy flour, 71 g/L cornmeal, 5.5 g/L agar, 5.5 g/L malt, 7.5% corn syrup and 0.4% propionic acid in a 12:12 light:dark cycle incubator at 25°C. Vibrio cholerae strains were cultured in Luria-Bertani (LB) broth or on LB agar supplemented with streptomycin (100 μg/ml) at 27°C. Where noted, supplements were added in a 5mM concentration (Sigma).

Generation of V. cholerae deletion mutants.

Suicide plasmids carrying in-frame deletions in hapR (VC0583) and vpsA (VC0917) were constructed using gene splicing by overlap extension and ligation into pWM91 ^{69, 70}. The primers used to construct the vpsA and hapR deletions are listed in Supplementary Table 4. In-frame deletions of hapR and vpsA were engineered into V. cholerae by homologous recombination using the suicide plasmids designed for this purpose as previously described ⁷¹.

Drosophila infections.

Because of their larger size, female flies were used to maximize material for assays. An overnight culture of V. cholerae was diluted in a 1:10 ratio into fresh LB broth. In an arthropod containment level 2 facility, thirty 5 to 7-day-old adult, female flies with genetic background as stated and listed in Supplementary Table 3 were randomly distributed into three fly vials containing a cellulose acetate plug infiltrated with 3 ml of this bacterial suspension. For succinate treatment, 5 mM succinate-containing LB was fed to adult female flies with or without V. cholerae infection as noted for each experimental condition. Flies were scored for survival twice daily, harvested for bacterial counts, or dissected for microscopic analysis.

In vitro, in vivo, and microscopic assessments of bacterial growth and colonization.

To assess in vitro growth, overnight V. cholerae cultures were diluted to an OD₆₅₅ of 0.01 and distributed into microtiter wells in triplicate. The OD₆₅₅ was recorded over time using an Infinite 200 spectrophotometer (Tecan). To assess host bacterial burden, replicates of ten flies were washed three times in ethanol followed by three washes in phosphate buffered saline (PBS) and homogenized in 500 μl of LB broth at 48h post-infection. The resulting homogenates were diluted, plated on LB agar, and incubated at 37 °C for enumeration of V. cholerae or plated on de Man, Rogosa, and Sharp (MRS) agar and incubated at 30 °C for enumeration of gut commensals. Colony forming units/fly were assessed after 24 and 48 hours. In graphs of bacterial burden, each point represents one cohort of ten flies. Colonization transfer experiments, including both enumeration and visualization of bacterial burden, were performed as previously described ²⁴.

Biofilm Assays.

Biofilm formation was quantified as previously described ^{72, 73}. Briefly, V. cholerae cultures were adjusted to an OD₆₅₅ of 0.005 in LB broth and aliquoted into borosilicate tubes in triplicate. The tubes were incubated for 48 hours at 27°C. Planktonic cells were removed and quantified by OD₆₅₅. The remaining surface attached cells were resuspended in PBS by vortexing with 1-mm-diameter glass beads (BioSpec Products, Inc.) for approximately 1 min [30% (vol/vol)]. The density of the resulting suspension was quantified by OD₆₅₅.

Immunofluorescence to detect cells containing phosphorylated histone 3 (PH3).

Numbers of PH3⁺ cells were quantified following 72 hours of exposure to V. cholerae. Intestines were removed and fixed with 4% formaldehyde in phosphate-buffered saline + 0.1% Tween 20 (PBST) overnight at 4°C. After washing in PBST three times, tissues were blocked in 2% BSA in PBST for 1 hour at RT and incubated overnight at 4°C with a 1:500 dilution of a rabbit anti-PH3 antibody (Millipore, CA 92590). After two PBST washes, a 1:200 dilution of a goat anti-rabbit antibody conjugated with Alexa Fluor® 594 (Invitrogen) and 4,6-diamidino-2-phenylindole (DAPI, 1 μg/ml) were added to the tissues and incubated for 2 hours at room temperature. After the final PBST wash, tissues were placed in mounting medium (Vector Laboratories) and examined by fluorescence microscopy using a Leica TCS SP2 AOBS system to quantify PH3⁺ cells per gut.

Metabolomic analyses.

Acetate levels in V. cholerae culture supernatants were measured as previously described using an acetate kinase-based kit (Megazyme) following the protocol provided by the manufacturer ³⁸. To prepare supernatants for polar metabolite analysis, bacterial strains were cultured in triplicate overnight in LB broth at 25°C. Cultures were centrifuged, and the supernatant was removed and filtered through 0.22 µm filter disk (Fisher), dessicated in 80% methanol at ambient temperature in a SpeedVac concentrator (Savant), and subjected to metabolomics analysis as previously described ⁷⁴. MetaboAnalyst 2.0 free online software was used for normalization and statistical analysis of data ⁷⁵. Metabolomics data were normalized to sum, and a baseline of 0.01 was established. Data were log-transformed prior to calculation of statistical significance using a one-way ANOVA with a false discovery rate of 0.05. A Fisher’s Least Significant Difference post-hoc analysis was performed. Normalized data are provided in Supplementary Table 1. Significantly different metabolites are listed in Supplementary Table 2. Metabolomics experiments were performed once.

Lipid Staining.

The fat bodies, intestines, and ovaries of adult flies were dissected at four days post-infection. Tissues were fixed with 4% formaldehyde in PBST, rinsed three times in PBST, and incubated with BODIPY 493/503 (1mg/mL, Invitrogen) and DAPI (1:1000, Invitrogen) for 1 h at room temperature. Confocal images were taken using a Zeiss LSM 780 confocal microscope. These experiments were performed twice with similar results.

Nutritional assays.

Flies were exposed to V. cholerae for four days in three separate vials considered biological replicates, harvested, and homogenized in 100 μl Tris-EDTA buffer supplemented with 0.1% Triton X-100. 10 μl of the supernatant was collected immediately and stored at −80°C for total protein quantification, while the remaining suspension was incubated at 90°C for 20 min to inactivate fly enzymes before proceeding with glucose and triglyceride quantification. Total protein, triglycerides, and glucose were quantified using commercial kits and reagents according to the manufacturers’ protocols [DC Protein Assay kit, Bio-Rad, Triglyceride Assay kit (Sigma), and Glucose (GO) Assay kit (Sigma)]. These experiments were performed twice with similar results.

Intestinal contents:

The volume of food in the Drosophila intestine was quantified as previously described using a fluorescein-based assay ³⁸.

Quantification of gene-specific mRNA levels in bacteria.

For C6706 measurements, triplicate bacterial cultures were grown from separate single colonies. Overnight cultures were harvested and lysed with lysozyme. RNA was extracted using TRIzol (Invitrogen). 500 ng RNA was used for the cDNA synthesis (Invitrogen SuperScript III), and an equivalent amount of RNA was used for genomic DNA control. For measurements of transcription in the Haitian strain 2013, triplicate bacterial cultures were adjusted to an OD₆₅₅ of 0.01 in 2ml LB broth and cultured for 24 hours at 27°C. RNA was extracted using a High Pure RNA isolation kit (Roche Life Science). cDNA synthesis was performed using a Superscript III reverse transcription kit according to the manufacturer’s instructions (Invitrogen). For all experiments, q-PCR was performed using iTaq SYBR Green (Bio-Rad) on a StepOnePlus real-time PCR system (Applied Biosystems). The clpX gene was used as an internal control. Primers are listed in Supplementary Table 4.

Quantification of gene-specific mRNA levels in Drosophila.

For Drosophila, thirty to forty-five female flies divided equally into three vials were used for mRNA quantification. These were considered three biological replicates. For validation of RNAi knockdown, whole flies were used. For intestine-specific transcription, intestines were dissected and removed 42 hours after exposure to V. cholerae. RNA was extracted using a High Pure RNA isolation kit (Roche Life Science) and treated with TURBO DNase treatment (Ambion). Quantification of total RNA was done with a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific), and quality was monitored by agarose gel electrophoresis. 500 ng of the resulting RNA was used for cDNA synthesis using a Quantitech Reverse transcription kit (Qiagen). Real time q-PCR was performed on the StepOnePlus real-time PCR system (Applied Biosystems) using iTaq Universal SYBR Green supermix (Bio-Rad). Relative expression was calculated using the 2-ΔΔCq method. RP49 (CG7939) gene transcription level was used for normalization. Primers used are listed in Supplementary Table 4.

Western Blot analysis.

Ten whole female flies (5–7 days old) were homogenized in PBS (Teknova P0191). Protein samples were reduced and denatured in SDS-PAGE sample buffer (BioRad 161–0747) for 20 min 95°C, separated on a 4–20% polyacrylamide gel (BioRad 456–1096) and transferred to a PVDF membrane (BioRad 1704272). Extracts were immunoblotted with rabbit anti-AKT (1:1000, Cell Signaling Technologies 9272), rabbit anti-Ser505-phospho-dAKT (1:1000, Cell Signaling Technologies 4054), mouse anti-β-Tubulin (1:2500, Developmental Studies Hybridoma Bank AB_579794), anti-mouse and anti-rabbit IgG horseradish peroxidase-conjugated secondary antibodies (1:5000, Cell Signaling Technologies, 7074 and 7072, respectively).

Statistical analysis:

In determining sample size, we relied on extensive experience, while taking into account availability of reagents, feasibility, and expense. Investigators were not blinded to group allocation. In all experiments, the investigators were aware of the experimental condition under investigation. Except where noted, experiments were repeated at least twice with similar results. Measurements represent the mean of at least three biological replicates in all graphs, and error bars represent the standard deviation. As noted and appropriate, a two-tailed students t test, a Welch’s t test (unequal variance), a Mann-Whitney test (non-parametric), or a one way ANOVA with Dunnett’s test as a post-hoc analysis were used to calculate significance. P values are shown above the relevant measurements. Log-rank analysis was used to assess differences in survival curves. For all tests, a p<0.05 was considered significant. Where an indication of significance is not noted, differences between measurements were not statistically significant. Statistical analysis of metabolomics data is detailed in the metabolomics section.