Reduced levels of B. vulgatus in the fecal microbiota of the V. vulnificus-infected mice
Fecal samples were collected from mice, which had been subjected to gavage with V. vulnificus and died within 20 h post-infection. Two independent sets of infection experiments were performed, and a total of 5 fecal samples were retrieved (three and two animals died in experiments 1 and 2, respectively). As a negative control, fecal samples from healthy mice treated with V. vulnificus-free PBS were also collected at each experimental set (two animals in each set). Analysis of 16S rDNA sequences amplified from the prokaryotic cells of the domain Bacteria in fecal samples showed that Bacteroidetes and Firmicutes were the major phyla occupying approximately 53% and 40% of total fecal microbiomes in the control mice, respectively (Fig. 1A). However, altered compositions of the two phyla were found in the V. vulnificus-fed mice: Although these two phyla were still the major phyla, Bacteroidetes and Firmicutes occupied approximately 33% and 61% of total fecal microbiomes, respectively (Fig. 1A). A principal coordinate analysis (PCoA) showed that the microbial compositions in most fecal samples of the V. vulnificus-fed mice were distinct from those of the control mice (permutational multivariate analysis of variance [PERMANOVA], P = 0.024) (Fig. 1B).
When the percentages of Bacteroidetes in the fecal samples were analyzed, their median values were 53.1% and 32.1% in the control mice and V. vulnificus-fed mice, respectively (Fig. 1C), and the difference between the two groups was significant (Mann-Whitney U test, P = 0.014). Among the bacteria belonging to Bacteroidetes, B. vulgatus, which has been recently reclassified as Phocaeicola vulgatus [23], was shown to be the most abundant species in mouse feces in both sets of experiments (Additional file: Table S1). The predominant abundance of B. vulgatus was consistent with the data previously reported [20]. Median values of the percentages of B. vulgatus in control mice and V. vulnificus-fed mice were 41.6% and 29.0%, respectively (Fig. 1D). Reduced levels of B. vulgatus were observed in all of the V. vulnificus-fed mice in each set of experiments (Mann-Whitney U test, P = 0.050): 52.5% ± 9.5% in the control vs. 33.2% ± 6.5% in the infected mice in experiment 1; and 36.4% ± 1.5% in the control vs. 27.8% ± 1.8% in the infected mice in experiment 2 (Additional file: Table S1). In addition to B. vulgatus, another bacterial species belonged to Bacteroidetes such as Parabacteroides goldsteinii also decreased in the V. vulnificus-fed mice (Additional file: Table S1). In contrast, some Lactobacillus species, which were the main bacteria belonging to Firmicutes in the mouse guts, did not show a decrease in their levels in V. vulnificus-fed fecal samples, but these increased in some animals.
Decreased viability of B. vulgatus in the presence of V. vulnificus
To examine whether the decrease in B. vulgatus levels in the V. vulnificus-fed mice was caused by an interaction between these two species of bacteria, a strain of B. vulgatus (MGM001; Additional file: Table S2) was isolated as described in “Methods.” In addition to MGM001, two strains of B. vulgatus were purchased from the American Type Culture Collection (ATCC) and Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). Three strains of B. vulgatus were anaerobically incubated in the M9-minimal medium with V. vulnificus. After 6 h of coincubation of B. vulgatus with V. vulnificus at ratios of 1:1, 1:2, and 1:5, serially diluted cocultures were spot-inoculated on RCM agar plates and incubated in an anaerobic chamber for the growth of B. vulgatus (Fig. 2A). The same aliquots were spot-inoculated on LBS agar plates for V. vulnificus growth (Fig. 2B). For comparisons of the viable cells of each species in the monocultures, these were also prepared to contain either B. vulgatus only (designated as 1:0) or V. vulnificus only (designated as 0:1, 0:2, and 0:5). The viability of all B. vulgatus strains tested in this study decreased in the presence of V. vulnificus, of which reduction appeared to be dependent on the abundance of V. vulnificus in cocultures. On the other hand, the viability of V. vulnificus was not affected by B. vulgatus.
To examine the dose-dependent reduction in B. vulgatus viability, the same amount of B. vulgatus was incubated under the conditions of various amounts of V. vulnificus, at ratios of B. vulgatus to V. vulnificus ranging from 1:1 to 1:10 (Fig. 2C). After 6 h, viable members of B. vulgatus were enumerated by spreading aliquots of serially-diluted cultures on the RCM agar plates. The results showed that the effect of V. vulnificus on B. vulgatus viability was antagonistic in a dose-dependent manner. The responses of the three strains of B. vulgatus to V. vulnificus were evaluated by comparing the ratios of B. vulgatus to V. vulnificus, which resulted in 50% reduction of viable cells of B. vulgatus: Calculated ratios of MGM001, ATCC8482, and DSM28735 to V. vulnificus were 1:2.0, 1:2.1, and 1:1.0, respectively. The strain MGM001, which has been isolated in this study, was used for the subsequent investigations.
Identification of an antagonistic compound in the V. vulnificus-spent medium
To examine how V. vulnificus affected the viability of B. vulgatus, both bacteria were spotted on a RCMS (salt-enriched RCM) agar plate, on which two bacterial colonies grew and were in contact with each other. Resultant interaction between the two colonies showed the presence of a growth inhibition zone of the B. vulgatus colony confronting the V. vulnificus colony (Fig. 3A). This suggested that the inhibition of B. vulgatus growth may have been mediated by a direct interaction with V. vulnificus cells and/or an indirect interaction via secreted compounds from V. vulnificus. To test these hypotheses, the involvement of the type VI secretion system (T6SS) of V. vulnificus in limiting the growth of B. vulgatus via direct contact, was examined. A mutant (ΔicmF) deficient in one of the T6SS components of V. vulnificus, i.e., IcmF (the intracellular multiplication factor F [25];), was constructed and mixed with B. vulgatus cells. However, it showed the same degrees of B. vulgatus survival shown by the wild-type V. vulnificus (Additional file: Figure S1A), which suggested that B. vulgatus death was not caused by direct cell-to-cell interaction via T6SS of V. vulnificus.
Next, the cell-free supernatant derived from the LBS broth grown by V. vulnificus to the stationary phase (spent medium, SM) was prepared and various volumes of the filter-sterilized SM were added to B. vulgatus (Fig. 3B). Compared to the control, 0% of SM (the fresh LBS medium only), the addition of SM resulted in a decrease in B. vulgatus viability in a dose-dependent manner. This suggested the possible role of V. vulnificus-originating and V. vulnificus-secreted substance(s) in the death of B. vulgatus. To screen the active components in SM, the major metabolites produced by V. vulnificus under anaerobic conditions, including pyruvate, formate, lactate, and acetate [26], were added to a suspension of B. vulgatus cells (Additional file: Figure S1B). None were shown to be effective in decreasing B. vulgatus survival to a concentration of at least 20 mM. In addition, the heated SM, which had been treated at 95 °C for 5 min to denature its proteineous substances, was mixed with B. vulgatus (Additional file: Figure S1C). However, the heat-treated SM showed a similar effect to that of the original SM.
To circumvent the difficulty in finding the active compound(s), B. vulgatus was challenged by various bacterial species and its survival was examined (Fig. 3C). All the Vibrio species tested in this study showed antagonistic effects on B. vulgatus. Among them, V. cholerae was the most effective in decreasing the survival of B. vulgatus, since the incubations containing bacterial cells composed of a 1:1 ratio showed similar levels to those shown by the incubations containing 10 times more V. parahaemolyticus. In contrast, E. coli strain S17-1 spir did not exhibit this effect but instead slightly increased the B. vulgatus levels in the 1:10 incubation. Three Vibrio species tested in this study have been shown to produce a cyclic dipeptide, cFP, in their SM, and the concentration of cFP in the V. vulnificus SM has been estimated to be up to 0.9 mM [27]. In addition, the incubation of B. vulgatus with a mutant V. vulnificus (ΔllcA), of which production of cFP was estimated to be less than ~ 30% of the wild type [24], resulted in lowered effectiveness in decreasing B. vulgatus survival, compared to that with the same amount of wild-type V. vulnificus (Fig. 3D). These results suggested that the death of B. vulgatus might be mediated by cFP secreted to SM.
To verify this, various concentrations of cFP, ranging from 0.125 to 1.0 mM, were added to the suspensions of three strains of B. vulgatus and incubated for 6 h (Fig. 4A–C). The survival of all strains was significantly affected by cFP in a concentration-dependent manner. However, when a single amino acid, F or P, and the linear dipeptide F-P were added to B. vulgatus cells at the same concentrations used for cFP, they did not show any effect on B. vulgatus survival (Fig. 4D). Furthermore, other cyclic dipeptides, such as cyclo(Phe-Val) (cPV) and cyclo(Pro-Thr) (cPT), did not show any antagonistic result up to a concentration of 4 mM (Fig. 4E). These results suggested that the V. vulnificus-originated cFP determined the specificity of the interaction between B. vulgatus and V. vulnificus.
Induction of B. vulgatus cell death by cFP
To directly observe the effect of cFP on B. vulgatus cells, the strain MGM001 was treated for 6 h with various concentrations of cFP ranging from 0 to 4.0 mM, and then stained using a Live/Dead cell double staining kit (Sigma-Aldrich) (Fig. 5A). Through fluorescence microscopy, the number of red-stained dead cells were found to increase with cFP concentration, while the number of green-stained live cells gradually decreased (Fig. 5B).
The increased death of B. vulgatus observed via staining with fluorescence dyes was confirmed using electron microscopes. Scanning electron microscopy (SEM) of the MGM001 cells (Fig. 5C), which were treated with 4.0 mM cFP for 6 h, showed the collapsed cellular morphology with undulated cell surfaces. In addition, transmission electron microscopy (TEM) of three B. vulgatus strains presented dramatic alteration in cellular morphology upon treatment with cFP, with enlarged periplasmic space, lyzed membranes, and partially ruptured cells (Fig. 5D–F). It indicated the occurrence of membrane disruption in the cFP-treated cells.
Effects of cFP on Parabacteroides and Lactobacillus
From an effort to obtain pure cultures of microorganisms from mouse guts, another bacterial species, P. goldsteinii belonging to Bacteroidetes and several Lactobacillus species belonging to Firmicutes have been isolated (mouse gut microorganism (MGM) isolates; Additional file: Table S2). For the addition of V. vulnificus cells (2.0 × 108 cells; Fig. 6A) or cFP (1 and 4 mM; Fig. 6B) to the resuspensions of the isolated gut commensals, only P. goldsteinii experienced decreased survival, while the survival of Lactobacillus species (i.e., L. johnsonii, L. reuteri, L. murinus, and L. intestinalis) was not decreased. Observation of cFP-treated P. goldsteinii cells under TEM showed apparent membrane disruption (Fig. 6C), as shown in the cFP-treated B. vulgatus.
cFP-induced cell death of B. vulgatus via membrane disruption
To examine the changes in membrane integrity, a fluorescent dye, 3,3′-dipropylthiadicarbocyanine iodide (DiSC3(5)), was used to incorporate into bacterial membranes [28]. Then, the amount of DiSC3(5) released from bacterial cells was measured upon cell lysis by exposure to a membrane permeabilizing agent, such as Triton X-100. Total amounts of the membrane-associated DiSC3(5) were estimated by measuring the fluorescence released from the cells treated with 0.1% Triton X-100, for which almost all B. vulgatus cells were lysed. To determine the effective concentration of Triton X-100 causing the lysis of 50% B. vulgatus cells (EC50), various concentrations of Triton X-100 were exposed to DiSC3(5)-associated B. vulgatus. By plotting the percentages of dyes normalized with the estimate of released DiSC3(5) at 0.1% Triton X-100, as described in “Methods,” EC50 was calculated to be approximately 0.002% (Fig. 7A). The released fluorescence upon exposure to 0.002% Triton X-100 from B. vulgatus cells treated with 4 mM cFP (termed by Δ4 mM cFP) and the control B. vulgatus cells (termed by Δ0 mM cFP) was estimated (Fig. 7B). The values of Δ0 mM cFP and Δ4 mM cFP were estimated at 605 (± 64) and 727 (± 80) RFUs, which was equivalent to 29% (± 2.3%) and 43% (± 3.7%) of the total fluorescence initially incorporated into the cells, respectively. This difference was significant with a P value less than 0.008 (Student’s t-test). These results evidencing the increased susceptibility of cFP-treated cells to Triton X-100 suggested that cFP caused the membranes of B. vulgatus to be permeable, thus easily disruptive.
To differentiate whether this increased susceptibility to a membrane permeabilizing stress was specifically mediated by cFP or caused by any dipeptide composed of hydrophobic amino acids, the same experiments were performed using 4 mM of FP, cPT, and cPV (Additional file: Figure S2). B. vulgatus cells treated with FP, cPT, or cPV showed a slightly lowered basal RFU (averaged RFUs for 10 min before Triton X-100 treatment) compared to the control and much lower maximal RFU (averaged RFUs for 10 min after Triton X-100 treatment) than the cFP-treated cells. Upon the addition of 0.002% Triton X-100, B. vulgatus cells treated with these dipeptides produced Δ4χmM dipeptide values (562, 572, and 561 RFUs in treatment with FP, cPT, and cPV, respectively) with similar ranges to that of the control cells (608 RFUs).
Identification of a B. vulgatus factor involved in cFP-mediated membrane disruption
Some factors, such as RecA, CidA, and ObgE, have been found to regulate the integrity of bacterial membranes under specific conditions [29,30,31,32]. The genomes of B. vulgatus have the open reading frames (ORFs) homologous to these genes, though their functions have not yet been studied. To screen the factor(s) involving the cFP-mediated membrane disruption, the cellular abundance of obgE, cidA, and recA transcripts and proteins were monitored in B. vulgatus cells incubated under the conditions in the absence or presence of cFP. Transcript levels of the obgE, of which contents were normalized with the gap transcript levels in the same cells, were increased as the exposure time was extended (about three-fold increase in the 4 h incubation compared to 0 h incubation), while those of cidA and recA were not altered up to 4 h incubation period (Fig. 8A). Protein levels in cells were examined using specific polyclonal antibodies against recombinant ObgE and RecA of B. vulgatus (Fig. 8B). Due to no apparent induction of rCidA from the overexpression vector carrying the cidA gene, an experiment to compare the cellular CidA levels was not successful. In addition to the increase in the obgE transcript, the protein content of ObgE increased in the B. vulgatus cells exposed to cFP. These results lead us to speculate the possible involvement of ObgE in the observed membrane disruption in B. vulgatus exposed to cFP.
To elucidate the specific interaction of ObgE with cFP, the isothermal titration calorimetry (ITC) experiments were performed using nucleotide-free rObgE, as described in “Methods.” In total, 40 μM of rObgE was titrated at 25 °C with a cFP stock solution of 400 μM (Fig. 8C). The equilibrium dissociation constant (KD) and binding stoichiometry (N) for the interaction between rObgE and cFP were 1.8 ± 0.35 μM and 0.78 ± 0.11, respectively. In contrast, the titration of rObgE with 400 μM of another cyclic dipeptide, cPT did not show any signature for the inter-molecular interaction.
Since the B. vulgatus ObgE specifically interacted with cFP and its cellular contents increased in cFP-treated cells, the role of ObgE in membrane disruption was examined. It has been previously reported that a trial of obgE gene deletion was not successful in many bacterial species [33]. Therefore, we have constructed a cell enabling to overexpress ObgE in response to the concentration of arabinose, instead to attempt to construct an obgE deletion mutant of B. vulgatus. Differential expression of ObgE was achieved using B. vulgatus MGM001 harboring the obgEBv gene in a vector, pBAD/Myc-His B (pBAD-obgEBv), which was added with L-arabinose at concentrations ranging from 0 to 1.0%. For quantitative analysis, each cell lysate was subjected to western blotting along with the known concentrations of rObgE (Fig. 8D). Cellular contents of ObgE were estimated by extrapolating the densitometric readings of the ObgE bands derived from cell lysates to the regression line derived from those of the known concentrations of rObgE (from 0.2 to 10 ng). The estimated cellular contents of ObgE were increased from 1.3 to 2.5 ng ObgE per μg of cell lysate by carrying pBAD-obgEBv, and those of B. vulgatus cells harboring pBAD-obgEBv were further increased to 3.1, 4.9, and 9.3 ng ObgE per μg of cell lysate if cells were incubated with 0.01%, 0.1%, and 1.0% arabinose, respectively.
Then, we measured the membrane integrity in B. vulgatus cells having different levels of ObgE. The same cells prepared for the experiments shown in Fig. 8D were exposed to 4 or 0 mM cFP for 6 h and then mixed with 0.4 μΜ DiSC3(5) for 60 min. As described above, the values of released DiSC3(5) from 0.002% Triton X-100-treated cells (Δ0.002%) were estimated and normalized with the total DiSC3(5) initially incorporated into cells (Δ0.1%). The ratios ([Δ0.002%/Δ0.1%] × 100) shown in Fig. 8E revealed that the increase in cellular ObgE resulted in the increased release of the dyes upon treatment of 0.002% Triton X-100 in an ObgE concentration-dependent manner (closed circles). Furthermore, the exogenous addition of 4 mM cFP resulted in 1.7~2.0 times greater release of DiSC3(5) (open circles) than the each control cells without cFP treatment. These differences were significant with P-values less than 0.01 (Student’s t-test).
Alteration of mouse lethality by V. vulnificus via changes in B. vulgatus abundance in mouse guts
The introduction of V. vulnificus resulted in the reduction of B. vulgatus levels in the fecal samples and this reduced composition was postulated to be mediated by cFP secreted by infecting V. vulnificus. Therefore, the effect of the exogenous addition of cFP on the abundance of B. vulgatus in mouse fecal samples was examined. The fecal samples were collected from 4-week-old female mice orally injected with cFP at a concentration of 110 μg cFP per gram of mouse. The total DNA extracted from each fecal sample was subjected to q-PCR using a primer set specific to the 16S rDNA of B. vulgatus (Bv-F and Bv-R [34];) or the universal primer set for the eubacterial 16S rDNA (785F and 907R [35];). The relative abundance of B. vulgatus 16S rDNA was normalized by the abundance of the total 16S rDNAs derived from PCR using the universal primer set. The medians of estimated values of 2−[CT(B.vulgatus) − CT(Total Bacteria)] were 0.1369 (± 0.0748) and 0.0619 (± 0.0766) in the PBS-treated control (n = 20 mice) and cFP-treated mice (n = 20 mice), respectively (Fig. 9A). Therefore, the levels of B. vulgatus decreased approximately 2.2-fold in mice injected with cFP (P = 0.006, two-sided Student’s t-test).
Next, we investigated whether the mice containing decreased levels of B. vulgatus showed a different susceptibility to infection by V. vulnificus. Mice were orally injected with cFP and they were infected at 12 h post-injection of cFP with various doses of V. vulnificus, ranging from 104 to 108 cells (n = 7 mice per each treatment) (open symbol). The same experiments were performed using the control mice that had been injected with cFP-free DMSO cells (n = 7 mice per each treatment) (closed symbol) (Fig. 9B). When infected with the highest dose of V. vulnificus (i.e., 108 cells), mice under both conditions died equally and quickly within 24 h. Similarly, the mice infected with the lowest dose of V. vulnificus (i.e., 104 cells) showed similar patterns of survival, regardless of the exogenous addition of cFP. However, the number of dead mice was higher in the cFP-treated sets of mice when they were preinjected with cFP then infected with 105–107 cells of V. vulnificus. In cases of infection with 106 cells of V. vulnificus, the difference in survival was significant, with a P-value of 0.001 (log-rank test).