Report

Antagonism toward the intestinal microbiota and its effect on Vibrio cholerae virulence

See allHide authors and affiliations

Science  12 Jan 2018:
Vol. 359, Issue 6372, pp. 210-213
DOI: 10.1126/science.aap8775

Cholera pathogen zaps competition

Many bacterial pathogens inject their hosts with virulence effectors delivered by specialist secretion machines. Vibrio cholerae has a type VI secretion system (T6SS) that can be loaded with protein toxins that target eukaryote host cells or kill competing bacteria. Zhao et al. discovered that mutant V. cholerae lacking a T6SS could not compete against Escherichia coli strains in the mouse gut. In contrast, intact V. cholerae readily gained a foothold in the gut of young mice, pumping up inflammatory immune responses and prompting more violent symptoms.

Science, this issue p. 210

Abstract

The bacterial type VI secretion system (T6SS) is a nanomachine that delivers toxic effector proteins into target cells, killing them. In mice, we found that the Vibrio cholerae T6SS attacks members of the host commensal microbiota in vivo, facilitating the pathogen’s colonization of the gut. This microbial antagonistic interaction drives measurable changes in the pathogenicity of V. cholerae through enhanced intestinal colonization, expression of bacterial virulence genes, and activation of host innate immune genes. Because ablation of mouse commensals by this enteric pathogen correlated with more severe diarrheal symptoms, we conclude that antagonism toward the gut microbiota could improve the fitness of V. cholerae as a pathogen by elevating its transmission to new susceptible hosts.

Bacterial pathogens deploy numerous virulence factors that directly influence their pathobiology. However, little is known about the impact of pathogen interactions with the commensal microbiome on virulence and pathogen fitness. Vibrio cholerae, the causative agent of cholera (1, 2), assembles a dynamic organelle called the type VI secretion system (T6SS) that delivers toxic effector proteins to eukaryotic and prokaryotic cells (3, 4). The T6SS of V. cholerae can kill heterologous bacterial species in vitro, such as Escherichia coli (5), but it is also activated in vivo (within infected experimental animals) to kill V. cholerae cells that lack immunity to its toxic effectors (6, 7). Hence, we asked whether the antibacterial activity of the T6SS is directed against gut commensals that share a niche with V. cholerae (6, 8). Here, we show that V. cholerae T6SS is used against commensal bacterial species and drives measurable changes in the pathogenicity of V. cholerae.

To test whether V. cholerae exerts antagonistic effects on the gut microbiota, we isolated several Gram-negative commensal bacteria from the small intestines of 5-day-old suckling (infant) mice. All isolates were susceptible to T6SS killing by the V. cholerae 2740-80 strain, which expresses an active T6SS in vitro (9) (table S1). Of the commensal strains isolated, we selected two E. coli strains that readily colonized the mouse small intestine—the same niche that V. cholerae also predominantly colonizes during infection. Although each strain was capable of colonizing the majority of animals within a mono-challenged group, two strains (WZ1-1 and WZ2-1), when used as a mixture, were found to reproducibly colonize all challenged animals (fig. S1). These two strains typically achieved a level of 104 to 106 colony-forming units (CFU) per mouse small intestine after 24 to 48 hours of infection and were therefore used as a defined commensal microbiota in subsequent studies. To establish whether the V. cholerae T6SS could affect the gut colonization of these E. coli commensal strains, we inoculated groups of neonatal mice with the two E. coli strains, returned them to their dams for 24 hours, and then arbitrarily selected three groups of pups for challenge with 105 CFU of wild-type V. cholerae [strain C6706 (8)], 105 CFU of an isogenic T6SS-defective vipA mutant, or buffer control. At 12 hours after wild-type V. cholerae challenge, the E. coli intestinal load was lower by a factor of ~300 than the load in mice challenged with the T6SS mutant at 24 hours (Fig. 1A; 5 ± 2 versus 1200 ± 400 CFU per mouse, P < 0.05, wild-type and T6SS-mutant treatment, respectively) or buffer (3840 ± 435 CFU per mouse). Surviving E. coli that were recovered from wild-type V. cholerae–treated mice did not show a reproducible skew toward either WZ1-1 or WZ2-1, indicating that both strains were equally targeted for elimination in vivo. Because the V. cholerae T6SS depends on the vipA gene, and because V. cholerae C6706 is known to express T6SS genes in vivo during mouse infections (8), we conclude that the antibacterial activity directed against the E. coli commensal isolates is dependent on expression of V. cholerae T6SS within the mouse intestine.

Fig. 1 T6SS facilitates V. cholerae colonization of the mouse gut by eliminating T6SS-sensitive commensals.

(A and B) E. coli (A) and V. cholerae (B) CFU counts per small intestine (SI) homogenate collected from mice (n = 2 to 8) colonized first with commensal E. coli strains and later challenged with 105 CFU of wild-type V. cholerae C6706 (Vc WT), T6SS-defective mutant (Vc T6SS–), or buffer control (No Vc). Each data point represents one mouse; error bars represent mean ± SEM of recovered CFUs. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001 (unpaired t test).

Surprisingly, we found that V. cholerae C6706, relative to its T6SS-defective vipA mutant, exhibited enhanced colonization in mice precolonized with E. coli commensals. At 8 hours after challenge, the wild-type V. cholerae bacterial load exceeded that of the T6SS mutant by a factor of ~10 (Fig. 1B; 6 × 104 versus 5 × 103 CFU per mouse, P < 0.05). By 24 hours after infection, this increased to a factor of ~15 difference (Fig. 1B; 1.2 × 107 versus 8.6 × 105 CFU per mouse, P < 0.001). To determine whether T6SS was allowing colonization of V. cholerae strains that could not express a functional T6SS, we simultaneously introduced the wild-type strain (C6706) and the vipA T6SS mutant into mice precolonized with E. coli commensal strains. We observed no significant difference between wild-type and T6SS-mutant colonization levels in these mice (fig. S2; competition index ~1.0), indicating that the wild-type strain could stimulate the colonization of the T6SS mutant, presumably by deploying its functional T6SS to eliminate the E. coli commensals that were antagonizing the vipA mutant.

T6SS-dependent killing of E. coli commensals also caused changes in the transcription of pro-inflammatory factors known to be induced in response to V. cholerae infection (10). Examination of intestinal extracts derived from mice 6 hours after infection with wild-type V. cholerae or with the T6SS-mutant strain showed that interleukin 6 (IL-6) transcriptional levels were higher by a factor of 3.3 (10.7 ± 1.8 versus 3.2 ± 0.7, P < 0.05), and that levels of KC [chemokine (C-X-C motif) ligand 1, also known as CXCL1] were higher by a factor of 5.1 (10.8 ± 2.4 versus 2.2 ± 0.3, P < 0.05), in the mice infected with wild-type V. cholerae (Fig. 2, A and B). These results indicate that during the early stage of infection, the induction of innate immune factors IL-6 and KC is dependent on V. cholerae (C6706) possessing an active T6SS. If this early T6SS-dependent IL-6 and KC cytokine response depends on ablation of gut commensals, we reasoned that antibiotic treatment might suppress this response. We therefore eliminated much of the mouse gut microbiota by antibiotic treatment for 2 days, and then tested whether mice challenged with either wild-type V. cholerae (C6706) or the isogenic T6SS mutant elicited detectable differences in cytokine expression levels. In the absence of streptomycin-sensitive gut commensals, no significant differences in IL-6 or KC expression levels were observed between mice infected with wild-type V. cholerae and mice infected with the T6SS-mutant strain at 6 hours (Fig. 2C). Thus, T6SS-mediated killing of commensal organisms correlates with the up-regulation of IL-6 and KC during early stages of infection, and this cytokine response is suppressed by antibiotic treatment.

Fig. 2 Wild-type V. cholerae induces a more acute host innate immune response than the T6SS-defectivemutant during early stages of infection.

(A and B) Transcript levels of inflammation markers IL-6 (A) and KC (B) were measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR) from wild type–infected or T6SS mutant–infected mice (n = 3 to 5) and are expressed as relative (fold) change with respect to No Vc (buffer) control. (C) IL-6 and KC transcript levels were measured in streptomycin (Sm)–treated or untreated mice. Error bars represent mean ± SEM of fold change. *P < 0.05 (unpaired t test). (D) Heat map of host innate immune response transcript levels measured by RNA-seq and confirmed by qRT-PCR in wild type–infected or T6SS mutant–infected mice (each group n = 5).

To gain a more comprehensive understanding of the T6SS-dependent modulation of the host immune response during early stages of infection, we harvested the small intestine of mice 6 hours after wild-type V. cholerae (C6706) or T6SS-mutant infection to extract host RNA for RNA sequencing (RNA-seq). We found 135 differentially expressed host genes; of the 19 genes involved in host immune function, 14 were up-regulated in the wild type–infected mice relative to the T6SS mutant–infected mice (table S2), including several known or predicted targets of the transcription factor NF-κB pathway (cd14, nfkbia, fcgrt, egr1, and cebpd) and regenerating islet-derived protein 3 gamma and beta (reg3g and reg3b), which were among the most highly induced genes (by factors of 16.8 and 13.1, respectively) (Fig. 2D and table S3). Thus, wild-type V. cholerae triggered a greater innate immune transcriptional response than did the T6SS mutant. We wondered whether the innate immune responses observed in these experiments were driven by release of microbe-associated molecular patterns (MAMPs), bacterial molecules recognized by the host innate immune system (11). Consistent with this hypothesis, we found that V. cholerae T6SS-dependent killing of E. coli releases MAMPs in vitro (fig. S3, A and B).

To test whether disease symptoms might be modulated by the T6SS-mediated killing of commensal species, we gavaged mice with 108 CFU of wild-type V. cholerae (C6706), its vipA T6SS mutant, or buffer and tested for fluid accumulation (FA ratio, indicative of diarrhea) after 22 hours. The FA ratio was highest for the wild type–challenged group (0.0875 ± 0.0025) and was significantly lower (0.0749 ± 0.0026) for the T6SS-mutant group (Fig. 3A, P < 0.01). We also monitored the bacterial load in the small intestine and colon of the infected mice. In both organs, the wild-type CFU exceeded that of the T6SS mutant by a factor of 3 to 9 (Fig. 3B, P < 0.01). Furthermore, elimination of host gut commensals by antibiotic treatment abolished the observed T6SS-dependent enhancement of the FA ratio and bacterial load for animals infected with wild-type V. cholerae relative to the T6SS mutant (Fig. 3, C and D).

Fig. 3 V. cholerae T6SS enhances the development of diarrhea disease symptoms.

(A and B) Fluid accumulation (FA) ratio (A) and V. cholerae CFU count per SI (B) of wild type–treated and T6SS mutant–treated mice. (C and D) FA ratio (C) and V. cholerae CFU (D) measured in mice pretreated with streptomycin. (E and F) tcpA and ctxA transcript levels were measured in mice not treated with streptomycin (No Sm) (E) or Sm-treated (F) at 6 hours after infection. **P < 0.01, ***P < 0.005, ****P < 0.001 (unpaired t test); ns, not significant.

The V. cholerae ToxT regulatory cascade is known to enhance diarrheal responses and bacterial loads in vivo through its control of the ctx and tcp virulence operons (1215). Thus, we asked whether T6SS-mediated killing of commensal organisms could release a commensal-derived signal that activates the ToxT regulatory cascade. We found that tcpA and ctxA transcript levels in bacteria recovered from infected mice were much higher in the wild type than in its T6SS mutant during the early stages of infection (fig. S4). This difference significantly dropped when mice where pretreated with streptomycin (Fig. 3, E and F), indicating that T6SS-dependent killing of streptomycin-sensitive commensal bacteria can activate V. cholerae virulence gene expression during early stages of infection. Because control experiments showed that T6SS-mediated killing of E. coli on laboratory media did activate tcpA or ctxA transcription in vitro (fig. S5), we conclude that the host is driving these T6SS-dependent changes in virulence gene expression.

Genome analysis has recently revealed that variant strains of the seventh pandemic El Tor V. cholerae clade are now responsible for the vast majority of cholera cases in South Asia, Africa, and Haiti (12). Because these variant strains cause more severe diarrhea in cholera victims, we questioned whether they showed higher levels of T6SS expression. Transcriptome analysis revealed that the variant strains H1 (isolated early in the Haitian cholera epidemic) and MDC126 [isolated in Bangladesh in 2008 (12)] showed greater T6SS gene transcript levels, by factors of 4 to 24, than two earlier seventh-pandemic strains [N16961 isolated in Bangladesh in 1971 (13) and C6706 isolated in Peru in 1991] (Fig. 4 and data S1). Variant strains also displayed more T6SS-dependent killing activity in vitro (fig. S6). Thus, the enhanced T6SS expression in V. cholerae variant strains may improve their fitness through T6SS-mediated killing of commensal Gram-negative microbiota or even enteric pathogens, such as pathogenic E. coli that share an upper intestinal niche with V. cholerae and a similar epidemiological distribution (14).

Fig. 4 Constitutive transcriptional up-regulation of T6SS genes in recent El Tor variant strains.

The transcriptomes of past seventh pandemic El Tor strains N16961 (1971) and C6706 (1991) are compared to those of recent variant strains MDC126 (2008) and H1 (2010) within T6SS-related operons in both chromosomes I and II. The relative expression levels of T6SS core/accessory, hcp, vgrG, and effector/immunity genes are normalized as RPKM (reads per kilobase of transcript per million mapped reads) units.

T6SS-mediated antagonistic behavior was previously observed between enteric pathogens (such as Salmonella typhimurium and Shigella sonnei) and gut commensal organisms (15, 16). Although secreted molecules such as autoinducers have been shown to activate or repress the expression of T6SS and virulence factors (1719), it is unclear whether microbial antagonism affects interspecies communication in vivo and pathogen fitness. Our results suggest that microbial antagonism may also change the interaction of an enteric pathogen with its host through altering its expression of virulence determinants as well as driving host innate immune responses. In sum, our data support a working model (fig. S7) that provides a framework for how V. cholerae might use an antibacterial mechanism to clear its target niche of inhibitory competitors and simultaneously enhance disease symptom–associated transmission.

Supplementary Materials

www.sciencemag.org/content/359/6372/210/suppl/DC1

Materials and Methods

Figs. S1 to S7

Tables S1 to S5

References (2025)

Data S1

References and Notes

Acknowledgments: We thank M. Gack and J. Chiang for assistance with luciferase assays and for a kind gift of reporter. We thank all Mekalanos lab members for helpful comments. Supported by National Institute of Allergy and Infectious Diseases grant AI-01845 (J.J.M.). The authors declare no competing financial interests. All authors helped to design and analyze experiments; W.Z., F.C., and W.R. performed experiments; and W.Z., F.C., W.R., and J.J.M. wrote the paper. All data generated or analyzed during this study are included in this published article and its supplementary materials.
View Abstract

Navigate This Article