Impact of a human gut microbe on Vibrio cholerae host colonization through biofilm enhancement

Essential revisions:

1) Provide data to show that Pa increases biofilm production through transcriptional or protein-level data. Indeed, as Midani et al. (2018) showed that conditioned medium from P. aminovorans significantly increased growth of V. cholerae, it seems essential for your study to distinguish between increased growth (and hence more biofilm in the pellicles) versus biofilm enhancement, as claimed in your manuscript.

We have added Figure 5C which shows vpsL expression results from Vc grown in coculture with Pa and Vc alone. qRT-PCR analyses demonstrated that the relative transcript levels of vpsL (the first and a representative gene in the vpsII operon) are higher in Vc cocultured with Pa than in Vc monoculture. In addition, in the new Figures 6C and D, we have used a P_vpsL-mNeonGreen reporter to show the activation of the vpsII operon in the co-culture pellicle, compared to the Vc monoculture pellicle.

We also want to clarify the results in the previous study (Midani 2018). In that study, Vc yield was higher in the conditioned medium (CM) harvested from Pa than Vc yield in CM harvested from Vc. However, the growth yield of Vc in the CM harvested from Pa was never as high as Vc in fresh medium. Indeed, Vc growth is higher in CM from Pa harvested at low cell-density than in CM from Pa harvested at high cell-density (our unpublished results), suggesting that the growth enhancement observed in the previous study is due to growth-limiting nutrients present in the CM harvested from Pa which were exhausted in CM harvested from Vc. This is the underlying reason that our prior results seem to contrast with our current findings. This was confirmed in our measurement of the planktonic growth yield of Vc monoculture which is comparable to Vc growth yield in the Vc-Pa coculture (Figure 3A). Importantly, biofilm formation is not enhanced when Vc is grown in CM harvested from Pa. We have added this discussion to the revised manuscript to clarify.

2) Please evaluate the impact of Pa on the production of V. cholerae's main virulence factors.

We have added new Figure 7—figure supplement 1 showing the relative transcript levels of ctxA and tcpA in Vc monoculture and Vc-Pa co-culture.

We measured virulence gene expression in regular LB medium (identical to the conditions used in the biofilm coculture experiments). Here, we noted that under these growth conditions the relative transcript levels of these virulence genes were low (e.g., compared to the relative levels of vpsL shown in Figure 5C). These results suggest that these are not ideal conditions for virulence gene expression measurement, based on prior knowledge, it is known that virulence-inducing conditions typically requires specific growth conditions (AKI) and with a shaking phase. However, shaking in AKI medium prevents proper interaction between the two species. Nonetheless, we found that ctxA and tcpA transcript levels are slightly higher in coculture (although only the ctxA comparison is statistically significant). We have edited the manuscript to include this information.

3) Please provide a reanalysis of the data shown in Figure 1 to show the abundance/normalized abundance of P.a. and V.c.

We have clarified in the manuscript that Figure 1 shows normalized abundance data, and we have added new Supplementary file 1 (containing two excel tables of raw data), which include the raw and normalized abundance for Pa, and for Vc in Vc-infected persons.

Reviewer #1 (Recommendations for the authors):

– In Figure 1A, the authors could determine the relative abundance of V. cholerae in the 22 infected individuals to demonstrate that there a positive correlation between the abundance of P. aminovorans and the abundance of V. cholerae in the 6 individuals with a significantly higher proportion of P. aminovorans. The authors could also determine the "Normalized Abundance" of P. aminovorans and V. cholerae in these samples.

We have clarified that the data shown in Figure 1 is normalized relative abundance, and we have shown the full raw and normalized data in the new Supplementary file 1. Please see above for detailed explanation of the results.

– The authors could also determine CFU of P. aminovorans in their suckling mouse model of infection to determine whether there is a correlation between the abundance of P. aminovorans and the abundance of V. cholerae during infection of these mice. This would also allow them to determine whether the ratio of abundance in mice is also 1:1 as observed in human samples.

We have determined the temporal dynamics of Pa in the small intestine of infant mouse (Figure 2—figure supplement 2). Pa can be detected 12 hours after the last Pa inoculation (right before Vc infection at 0 hr). The Pa counts decreased over the next 24 hours and the rate of decrease is independent of the presence of Vc. At the early onset of the infection when Pa was still detectable (6 hrs after Vc infection), the ratio between Vc and Pa is variable but close to 10:1. We have included these new data in the revised Figure 2—figure supplement 2. Although we did not observe the 1:1 ratio, which is likely due to the difference in intestinal physiology between human and mouse, we believe our model is still sufficient to show the effect of the presence of Pa inside the small intestine on Vc host colonization.

– In Supplemental Figure 1, the authors could analyze their microbiota sequencing data to determine whether members of the Proteobacteria Phylum are being displaced by P. aminovorans at the Family level.

We have added Panel C in Figure 2—figure supplement 1 ((original Supplemental Figure 1) which shows the order-level taxonomic impact of Pa colonization) within the Proteobacteria phylum. Differences in the two groups were not found to be significant with FDR-adjusted significance testing at multiple taxonomic levels.

– The authors could measure the expression of a V. cholerae virulence factor (such as tcpA) by qPCR to determine whether there is an increase in virulence factor expression in V. cholerae in the presence of P. aminovorans (Figure 2B and Figure 2C).

We have added this assay to our results as described above. These results are now shown in Figure 7—figure supplement 1. Please see above for detailed explanation.

– In Lines 166-168, the authors state, "Our findings also illustrate that our approach to microbiome studies in humans (6, 7) can be used as a predictive tool to identify gut microbes that alter Vc virulence." The authors should replace the word "virulence" with "colonization" which seems more appropriate for the data shown.

This correction has been made in the revised manuscript.

Reviewer #2 (Recommendations for the authors):

1. Based on the results shown in Figure xx the authors are proposing that P. aminovorans increases the ability of V. cholerae to produce VPL (Vibrio polysaccharides). I think this is an interesting point towards the mechanistic understanding by which P. aminovorans can enhances biofilm formation of this pathogens. I think it would be nice to further obtain support for this hypothesis using a transcription reporter fusion to a vpl-associated promoter to obtain further support for this proposal.

In response to the reviewer’s request, we have obtained a strain harboring P_vpsl-mNeonGreen and we have repeated the pellicle imaging in the updated Figure 6C-D. Interestingly, subpopulations of Vc cells have elevated vpsL level when co-cultured with Pa. These data are consistent with our new qRT-PCR results showing higher vpsL expression (Figure 5C).

Reviewer #3 (Recommendations for the authors):

1) The introduction would be greatly improved by being specific about the model in which previous studies were done, where V. cholerae is found within the intestine in each model, and how the association between V. cholerae and the microbe of interest was established in the study (i.e. stool cultures vs intestinal tissue). When discussing concordant vs discordant observations, these details are very important (eg line 141).

We have revised the manuscript to include the recommended additional information on the models discussed in the introduction.

2) Figure 2: Figure 2A: Here the authors show the results of a colonization assay in which Pa is inoculated into mice every 12 hours for 48 hours. Were the measurements taken at 24 and 48 hrs performed before or after the 24- and 48-hour inoculations, respectively. This should be clarified. If taken immediately after inoculation, this may not demonstrate colonization. Figure 2B: Here is Vc inoculated directly after the 48h inoculation of Pa? If so, have the authors tested whether the earlier inoculations are relevant or necessary?

We apologize for the ambiguity. To clarify, the animals were inoculated with four doses of Pa at 0, 12, 24, and 36-hour time points. In some animals, after two doses of Pa inoculation, the Pa count in the small intestine was determined at the 24-hr (i.e., 12 hrs after the second Pa inoculation). The Pa count was also determined after four doses of Pa inoculation at 48-hr (i.e., 12 hrs after the last Pa inoculation). Vc infection was performed 12 hrs after the last Pa inoculation, and we have shown that Pa was detectable in the small intestine at that time point prior to Vc infection. We have revised the text and figure legends to clarify.

3) Line 227-245 and Figure 6: Because WGA is not specific for VPS and Pa are not specifically labeled, conclusions from these studies must be tempered.

In response to the reviewer’s question, we have added negative controls (new Figure 4—figure supplement 1) showing that Pa cells alone or Vc ΔvpsL mutant does not show WGA staining under the current conditions. While it is true that WGA stains GlcNAC residues, for Gram-negative bacterial cells including Vc and Pa, the WGA lectin molecules do not pass through the outer membrane. Only when the outer membrane is significantly impaired will one see strong WGA signal. Indeed, we always observe dead cells with strong WGA signal but those are easily distinguished from the VPS signal.

Regarding the labeling of Pa cells, currently we do not have good genetic tools to express fluorescent proteins in Pa; therefore, we relied on the universal membrane labeling of FM 4-64 to detect their presence in the co-culture biofilm. Importantly, Pa cells also have a characteristic cocci shape that is distinct from Vc’s curved-rod shape. In the updated Figure 6A-B, we have provided clearer zoom-in image, in which Pa cells can be clearly distinguished from Vc cells by both the absence of SCFP3A signal and the characteristic cocci shape.

4) Line 356-259: The conclusion that Pa activates production of VPS would be greatly strengthened if transcriptional activation of the vps genes could be shown. If transcriptional activation is not observed, there is a concern that increased WGA staining in biofilms is not sufficient to support this claim.

We have added new Figure 5C showing qRT-PCR results and Figure 6C-D showing microscopy analysis on vpsL expression from Vc grown in monoculture or in coculture with Pa. Please see above for detailed explanation.

5) Figure 2B: This strain of V. cholerae is well-known to undergo spontaneous mutations that result in QS-incompetence and increased biofilm production. Is it possible that Pa is promoting a change at the chromosomal level? If not, V. cholerae isolated from the Vc + Pa pellicle should behave like WT Vc in monoculture.

Thank you for pointing out this important issue. We have isolated 15 individual Vc colonies from pellicles harvested in Vc monocultures (5 isolates) and Vc-Pa cocultures (10 isolates). The luxO locus of these isolates was sequenced; all carry the WT luxO allele, suggesting Pa does not promote additional mutational change in Vc in this allele.

6) Figure 7: Based on the authors' model, intestinal titers of Pa should be higher during WT infection as compared with that during ΔvpsL infection. Intestinal titers of Pa increase in response to VPS production by Vc would help to support the in vitro results, which suggest that Vc VPS recruits Pa to biofilm aggregates.

We thank the reviewer for the great suggestion.

As discussed in the response to reviewer #1, we have determined the temporal dynamics of Pa in the small intestine of infant mouse. Pa can be detected 12 hours after the last Pa inoculation (right before Vc infection at 0 hr). The Pa counts decreased over the next 24 hours and the rate of decrease is independent of the presence of Vc. At the early onset of the infection when Pa was still detectable (6 hrs after Vc infection), the ratio between Vc and Pa is variable but close to 10:1. We have included these new data in the revised Figure 2—figure supplement 2. Although we did not observe the 1:1 ratio, which is likely due to the difference in intestinal physiology between human and mouse, we believe our model is still sufficient to show the effect of the presence of Pa inside the small intestine on Vc host colonization. Further studies are needed to fully understand the dynamic of the interaction in vivo.

Reviewing Editor comments:

1) The authors should provide complementation data for the vps / rbmA / rbmC+bap1 mutant. If this isn't possible, they should at least Sanger sequence the luxO gene of all their strains (WT and the mutants) to ensure all strains are QS-proficient and not derived from a QS-impaired circulating mutant of the "WT" C6706 isolate (which is known to produced enhanced biofilms).

We have now provided new complementation data for the vpsL mutant in CV assay and pellicle assay (revised Figure 3—figure supplement 1), and in vivo (new Figure 7—figure supplement 2). All the other strains have been sequenced and they carry the WT allele of luxO. We also tested all our strains with a P_qrr4-lux assay and they all show a Qrr4 expression pattern identical to the WT, confirming they carry a WT luxO allele.

2) Do the authors have colonization data beyond the 24h time point? If these data are not available, the topic should be at least discussed (e.g., whether colonization is expected to be enhanced beyond the 24h time point).

Our current IACUC protocol only allows us to keep the infant animals away from their dams for no more than 24 hours. Therefore, we could not obtain data points beyond 24 hours after infection. As predicted from our time course study (revised Figure 2—figure supplement 2), the difference between the Vc counts in mice with and without Pa steadily increases, therefore, we predict the difference could be bigger if we allow the Vc infection to continue beyond 24 hours.

3) The authors wrote "Unlike previous studies (9, 10), pre-treatment with antibiotics did not change the outcome of Pa colonization (not shown)." => this is an important and somewhat surprising finding for which the data should be provided.

Thank you for pointing out this ambiguity. We intended to point out that unlike previous studies in which antibiotic treatment was required to achieve bacterial colonization, in our study antibiotic treatment was not required for Pa colonization. We have shown Pa can colonize in the small intestines of untreated mice (Figure 2A). We have rewritten this statement to clarify.