Negative feedback of cyclic di-GMP levels optimizes switching between sessile and motile lifestyles in Vibrio cholerae

  1. Aathmaja Anandhi Rangarajan
  2. Jeremy W. Schroeder
  3. Rebecca L. Hurto
  4. Geoffrey B. Severin
  5. Macy E. Pell
  6. Meng-Lun Hsieh
  7. Christopher M. Waters
  8. Lydia Freddolino

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

The signaling molecule cyclic di-GMP (cdG) controls the switch between bacterial motility and biofilm production, and fluctuations in cellular levels of cdG have been implicated in Vibrio cholerae pathogenesis. Intracellular concentrations of cdG are controlled by the interplay of diguanylate cyclase (DGC) enzymes, which synthesize cdG to promote biofilms, and phosphodiesterase (PDE) enzymes, which hydrolyse cdG to drive motility. To track the complete regulatory logic of how V. cholerae responds to changing cdG levels, we followed a timecourse of overexpression of either the V. harveyi diguanylate cyclase QrgB or a variant of QrgB lacking catalytic activity (QrgB*). We find that QrgB increases cdG levels relative to QrgB* for 30 minutes after overexpression, but the effect of QrgB on cdG levels plateaus at 30 minutes, indicating tight adaptive control of cdG levels. In contrast, loss of VpsR, a master regulator activating biofilm formation upon binding to cdG, leads to higher baseline levels of cdG and continuously increasing cdG through 60 minutes after QrgB induction, revealing the existence of a negative feedback loop on cdG levels operating through VpsR. Through a combination of RNA polymerase ChIP-seq, RNA-seq, and genetic approaches, we show that transcription of a gene encoding a PDE, cdgC , is activated by VpsR at high cdG concentrations, mediating this negative feedback on cdG levels. We further identify a transcript encoded within, and antisense to, the cdgC open reading frame which we name s RNA n egative r egulator of C dgC (SnrC). RNA polymerase ChIP-seq and RNA-seq demonstrate SnrC to be expressed specifically under conditions of high cdG in the absence of VpsR. Ectopic SnrC expression increases cdG levels in a manner dependent on CdgC, demonstrating that its effect on cdG levels is likely through interference with CdgC production. Further, although cells lacking cdgC exhibit enhanced biofilm formation, these mutants are outcompeted by wild type V. cholerae in colonization assays that reward a combination of attachment, dispersal, and motility behaviors. These results underscore the importance of negative feedback regulation of cdG to maintain appropriate homeostatic levels for efficient transitioning between biofilm formation and motility, both of which are necessary over the course of the V. cholerae infection cycle.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    The authors do not wish to provide a response at this time

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    This study investigated c-di-GMP (cdG)-dependent gene expression in Vibrio cholerae, an area that has been thoroughly researched in earlier studies. The transcriptomics results of the first sections largely confirm what has been previously described, but identify an unexpected, sudden c-di-GMP drop at OD600=1, which may be either the result of a feedback loop, as the authors suggest, or an experimental artifact (as explained in Major concerns). The most intriguing result of the study, in my opinion, is the identification of a potential antisense RNA, designated SnrC, that may counteract expression of the main cdG phosphodiesterase gene, cdgC. Unfortunately, the data are insufficient to conclude that SnrC is real. The follow-up competitive analysis of the cdgC mutants and growth analysis of the mutants with super-high cdG levels are not insightful because of low biological relevance. Overall, while some intriguing leads are described in this work, major experimental concerns exist and some conclusions may have to be revised.

    Major Concerns:

    1. The experimental setup may have had a major flaw. A sudden drastic drop in c-di-GMP levels at OD600=1 may be due to sampling/inducer addition. Until this is solved, it impossible to evaluate the adequacy of conclusions based on transcriptomics analysis.

    Fig. 1B shows an approx. 3-fold drop in c-di-GMP levels within 15 min upon induction of the QrgB-mutant protein expression. This drop cannot be explained by "transitioning to the high-cell density quorum state", as suggested by the authors, because such a transition likely takes more than 15 min, and it does not start so abruptly at precisely OD600=1 (when IPTG was added). A different, more benign explanation is more likely. Below are a couple of ideas to investigate. - Can the cdG decrease be due to the sudden drop in dissolved oxygen levels during the sample withdrawal for c-di-GMP measurements and IPTG addition? If so, the cultures were experiencing first hypoxia, then (upon restoration of shaking) oxidative stress, which affect the c-di-GMP transcriptome. I suspect that an approx. 3-fold drop in cdG levels does not take place in the undisturbed cultures that are allowed to grow for 15 min after reaching OD600=1. - Does the mutant QrgB* protein contain a cdG-binding site? If so, QrgB* may soak up intracellular cdG, whereas the intact QrgB may counteract the soaking effect by synthesizing cdG.

    1. Evidence of the potential existence of the antisense RNA, SnrC, is weak.

    It is not convincing that snrC is real despite the notion that bioinformatics data lack statistical significance (l. 197-203). A functional test, where an overexpressed antisense RNA lowers expression of the sense RNA (Fig. 3D) is obviously insufficient because this would work for any gene, whether antisense RNA is real or not. To verify that SnrC is real, one needs to show, at least, a snrC RNA band on a Northern blot and/or results of the snrC transcriptional fusion.

    Additional concerns:

    l. 95: More information about QrgB is needed, eg, is QrgB a cytoplasmic or membrane protein; does its DGC activity depend on interactions with small molecules or other proteins; does V. cholerae contain a QrgB homolog; does it have a cdG-binding site? l. 106: It is erroneous to suggest that in the wild type, c-di-GMP levels increased at 15 & 30 min when they were largely unchanged (Fig. 1A). l.143-52: That cdgC may be part of the negative VspR-mediated feedback on cdG levels is evident from refs. 11,13,14 and from the confirmation data in Fig. S1. I'm afraid it is not at all clear from Fig. 2A that cdgC is "the sole clear candidate for negative feedback on cdG levels". Please provide more substantive reasoning.

    Methods section: For major methods, provide a reference that served as the basis for experiments (as done for c-di-GMP measurements, l. 422), or method validation if the method is newly developed.

    Referees cross-commenting

    Serious concerns were raised by me and other reviewers about the quality of the data and the validity of conclusions. The major concerns raised by all reviewers need to be resolved (which, in my estimate, will likely take several months).

    Significance

    Significance is dificult to assess because of the major concerns about data validity.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The work from Rangarajan, Schroeder et al. examines dynamics of c-di-GMP (cdG) gene regulation and homeostasis in Vibrio cholerae following cdG overexpression. Relative increase in cdG levels are stimulated by the heterologous QrgB DGC enzyme, and the resulting changes in gene expression (biofilm up, motility down) are largely due to the cdG-dependent transcriptional activator. VpsR induction leads to activation of GGDEF/EAL protein CdgC; only its EAL (cdG-degrading) domain is active, and this activation leads to a feedback loop that results in a reduction in cdG levels. ChIP-Seq with RNAP led the authors to an unannotated sRNA that is transcribed within the cdgC ORF in the antisense direction that they name sncR, and whose expression is positively regulated by cdG. Expression of sncR negatively regulates cdG levels through CdgG.

    The authors could thus break the negative feedback loop they discovered and generate sustained high levels of cdG in the cell using ΔvpsR or ΔcdgC single or double mutants, and in those cases they observe growth defects.

    Overall, the experiments presented are high quality and the results offer significant novelty. The RNA-seq studies in Fig. 1 provide higher resolution than previous studies, including following the early dynamics upon cdG induction and resolving the VpsR/VpsT-dependence of individual genes and classes of genes. The figures are exceptionally clear, and the text is accurate and well cited/referenced. Multiple instances of the text do require additional clarification as noted below.

    Major comments:

    1. High levels of sncR are identified only in a ΔvpsR strain (Fig. 3B), yet the authors' model is that VpsR is not directly regulating sncR expression (Fig. 5). Instead, they suggest that the VpsR-dependence of the phenotype is due to only observing the highest levels of cdG in the absence of VpsR. If there is a way to clarify the connection of VpsR and sncR that would strengthen the manuscript.
    2. In Fig. 4BC, are these strains similarly vpsL- as most of the paper? If so, is it expected that they can form biofilms? I would recommend explaining this assay further, or if they are vpsL+ then clarifying in the text.
    3. L106-116: In Fig. 1A, why does QrgB overexpression not lead to elevated cdG at 15 m except upon normalization to QrgB* (Fig. 1C)? Is this connected to the feedback loop characterized in the manuscript (does it act that quickly?) or due to another effect?

    Related, the authors attribute the decrease in the denominator (QrgB* levels; Fig. 1B) to the cells transitioning to the high cell density state. Is it clear that this would occur in the time frame of 60 min?

    Minor comments:

    4a. In Fig. 1D, what does the null symbol represent?

    4b. L200-203, more explanation would benefit the assertion of the Type II error proposed. Are the authors claiming this is noted as not significant due to the high number of multiple comparisons?

    Referees cross-commenting

    The skepticism from the other reviewers on the bona fide structure and role for sncR are warranted, and they offer experiments that could test the ideas proposed in the manuscript.

    The other reviewers also highlight the issue I raised with the QrgB induction data. I was more generous to the authors given the Waters Lab's long history in the field, in using this particular construct to induce cdG, and in sharing the construct with other labs to enable its use for this purpose. Nonetheless I think these consensus concerns require additional scrutiny for the time frame and under the media conditions examined.

    Significance

    The work represents a substantial conceptual advance in understanding c-di-GMP signaling. The feedback loop described is likely to be a complicating factor in many studies that analyze c-di-GMP in V. cholerae specifically, and comparable regulation may be at play across bacteria.

    The authors do not address the antisense mechanism in this work. They note this limitation.

    In the Discussion, the authors identify a potential conflict with Ref. 35 on the role of VpsT. I found the authors' data convincing with this regard and it remains possible that the different results are due to distinct laboratory conditions.

    I expect that following audiences will be interested in the work: gene expression, gene regulatory networks, transcriptional regulation, bacterial small RNAs/antisense RNAs, small molecule signaling, c-di-GMP signaling, Vibrio cholerae biofilm formation/dispersal, and Vibrio cholerae colonization/transmission.

    My field: bacterial genetics, signal transduction, Vibrio bacteria.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    The study by Rangarajan, Schroeder et al. uncovers a negative feedback loop that controls cdG levels in Vibrio cholerae. They demonstrate that this feedback loop requires VpsR, one of the master regulators of cdG-dependent responses in this organism. Furthermore, they show that mutating the phosphodiesterase CdgC eliminates this negative feedback loop. This research is timely and addresses an important question about the regulation of this important secondary messenger molecule in bacterial species. The manuscript is well written and clearly presented. However, some of the results are fairly preliminary and there are alternative models that could explain the results obtained. Also, I disagree with the authors interpretations of some of their data. Specific points for the authors to consider are discussed below.

    Line 113-115: This reduction in cdG levels observed with expression of QrgB* is quite striking. In fact, the ratiometric "increase" in cdG levels observed for QrgB/QrgB* is largely driven by the decrease in cdG levels in the QrgB* condition as opposed to an increase in cdG in QrgB. Are the authors suggesting that the decrease observed with QrgB* is due to the natural reduction of cdG levels due to quorum sensing and the transition of cells to the high cell density state? What is the change in OD600 during this time course? Alternatively, isn't it possible that the expression of QrgB* is causing a response that results in a decrease in cdG? If so, this would change a lot of the conclusions in the present study. Perhaps one way to address this is if the authors have data on the cdG levels in wildtype cells during a similar time course without any induction of QrgB or QrgB*.

    Fig 2B: The authors indicate that the results with the PcdgC reporter confirm their RNA-seq results. I don't agree with this based on my interpretation of the presented data. While statistical comparisons are not provided, the relative change in PcdgC reporter activity when comparing 0 min to 60 min is very modest for all samples. This is especially relevant for the WT+QrgB condition, which should see a marked increase in reporter activity based on the RNA-seq results presented in 2A. The potential post-transcriptional regulation of CdgC discussed in the subsequent figures further complicates this. The data in 2C suggest that cdgC is associated with a negative feedback circuit that keeps cdG levels low in the wildtype as shown in 1A. But the mechanism underlying how this actually occurs remains unclear. Defined reporters or direct analyses of cdgC transcription and translation would be valuable to elucidate this.

    The data in 2C show that mutating cdgC now allows for similarly high levels of cdG in parent vs ∆vpsR. But how specific is this response to cdgC? An alternative possibility is that elevating the baseline cdG level is sufficient to bypass this feedback loop. For example, CdgJ is a phosphodiesterase that has previously been characterized in Vibrio cholerae and cdgJ mutants exhibit phenotypes consistent with elevated cdG. Performing similar experiments in the cdgJ mutant to show that the results seen in 2C do not occur when another phosphodiesterase is mutated would strengthen the conclusion that CdgC is specifically required for the negative feedback circuit that controls cdG levels.

    The only data that link CdgC to the negative feedback circuit are the assays performed in 2C that directly measure intracellular cdG levels. But the authors make a compelling case for the negative feedback circuit in generating a pulsatile response to QgrB induction in the wildtype in 1F. if CdgC is required for this negative feedback response, then a CdgC mutant should lose this pulsatile response. This should be tested to further support the model presented.

    Fig. 3B: It would be valuable to also plot the sense transcripts in this same region so that the expression of cdgC can be directly compared to the expression of SnrC. Also, can the authors speculate on why the RNAP enrichment at PsnrC in 3A does not reflect the expression levels of SnrC observed in 3B?

    Line 173-174: I am not sure how the authors are concluding that VpsR induces SnrC. I believe their data in 3B indicates the exact opposite, which is that SnrC expression is negatively regulated by VpsR. This finding is also not reflected in Fig.5.

    The data presented in Fig. 3 are compelling, but they only suggest that SnrC may be an anti-sense small RNA. More rigorous analyses are needed to make any firm conclusions about whether this is a bona fide sRNA and to define its potential mechanism of action. Northern blots should be performed to directly assess SnrC levels. Also, can the PsnrC promoter be mutated to prevent production of this small RNA while leaving cdgC function intact? This would allow for more rigorous genetic analysis to assess the impact of SnrC on CdgC transcription and translation. At the very least, claims of SnrC as a small RNA need to be toned down.

    Fig. 4C: this model system and data are very interesting. But they do not necessarily indicate that it is the negative feedback circuit that CdgC participates in that is important for these transitions. They may simply reflect that elevated cdG results in a defect in these assays. As mentioned above, cdgJ mutants have previously been shown to exhibit phenotypes consistent with elevated cdG. So, what is the phenotype of a cdgJ mutant in these assays? Since the cdgJ mutant should still have CdgC-dependent negative feedback control of cdG levels, this mutant may not have any defect. This result would be particularly compelling and support the authors conclusions that CdgC-dependent feedback control of cdG is critical for balancing biofilm formation and dispersion.

    Line 299-302: This is an interesting hypothesis, and could be tested with a cdgC catalytic site mutant. Or by otherwise inactivating CdgC phosphodiesterase activity while leaving the cdgC gene largely intact so that PsnrC expression can still be monitored. Again, these types of experiments are needed to make any firm conclusions about SnrC.

    Referees cross-commenting

    It appears as though all of the major points I raised in my review were also reflected in one or both of the other reviewers comments. So, I do not have anything to add.

    Significance

    The data presented in the study are strong. But some of the results are fairly preliminary and there are alternative explanations for the results that would make many of the conclusions incorrect or less compelling.

    The study helps fill a gap in knowledge about cdG regulation in bacteria. While the study focuses on Vibrio cholerae, the principles underlying this work are likely relevant to many bacterial species. So this work would be broadly interesting to the field of microbiology.

  5. Version published to 10.1101/2024.09.01.610008 on bioRxiv