Sugar-mediated regulation of a c-di-GMP phosphodiesterase in Vibrio cholerae

Biofilm formation protects bacteria from stresses including antibiotics and host immune responses. Carbon sources can modulate biofilm formation and host colonization in Vibrio cholerae, but the underlying mechanisms remain unclear. Here, we show that EIIAGlc, a component of the phosphoenolpyruvate (PEP):carbohydrate phosphotransferase system (PTS), regulates the intracellular concentration of the cyclic dinucleotide c-di-GMP, and thus biofilm formation. The availability of preferred sugars such as glucose affects EIIAGlc phosphorylation state, which in turn modulates the interaction of EIIAGlc with a c-di-GMP phosphodiesterase (hereafter referred to as PdeS). In a Drosophila model of V. cholerae infection, sugars in the host diet regulate gut colonization in a manner dependent on the PdeS-EIIAGlc interaction. Our results shed light into the mechanisms by which some nutrients regulate biofilm formation and host colonization.

The PEP:glycose phosphotransferase system (PTS) transports and phosphorylates a wide range of carbohydrates. In addition, it has been shown to regulate numerous important cellular functions in response to the availability of efficiently metabolizable carbon sources. The PTS was also suggested to affect the virulence of certain pathogens. Regulation by the PTS is usually mediated via the phosphorylation state of its protein components. The authors of manuscript NCOMMS-19-17087-T add a highly interesting example to the long list of PTS-regulated cellular functions: The glucose-specific EIIA (EIIAGlc) of Vibrio cholerae was found to interact with the c-di-GMP phosphodiesterase PdeS. EIIAGlc inhibits PdeS activity when it is unphosphorylated and stimulates PdeS when it is phosphorylated (P~EIIAGlc). In the presence of glucose, V. cholerae contains mainly EIIAGlc, which inhibits the hydrolysis of c-di-GMP by PdeS and consequently leads to elevated biofilm formation compared to V. cholerae cells grown in LB medium without glucose (contain mainly P~EIIAGlc). Utilization of different carbon sources was ultimately found to affect intestinal colonization of the host in a Drosophila melanogaster infection model. Colonization was more efficient by glucose-grown V. cholerae compared to mannitol-grown cells, in which only about 50% of EIIAGlc is present in dephospho form (elevated PdeS activity, lower biofilm formation).
In conclusion, the authors provide sound evidence for an entire V. cholerae regulatory cascade leading from carbon source utilization (environmental signal) via alterations of the phosphorylation state of PTS proteins, changes of the activity of the c-di-GMP phosphodiesterase PdeS and altered biofilm formation ultimately to differences in intestinal V. cholerae colonization of D. melanogaster. Although the PTS had previously been suggested to regulate biofilm formation, the underlying mechanism via EIIAGlc-mediated regulation of PdeS activity was not known before. Similar mechanisms of biofilm formation are expected to be operative in other organisms and the work described here is therefore of great general interest. The presented results are supported by careful statistical analyses.
A few minor points which should be addressed by the authors: l. 126: Include the references for the interaction of EIIAGlc with IDE and FapA l. 168: Maybe better write: …. VC1710 exhibited the predicted c-di-GMP hydrolytic activity l. 181: What are the "c-di-GMP-bound effectors"? l. 236: "What are "environmental sugars" compared to non-environmental sugars? Do the authors mean sugars encountered by V. cholerae in their environmental niches? See also legend of Fig. 4., l. 556. Also note, succinate is not a sugar. The abbreviation Suc could be easily assumed to stand for the sugar sucrose. l. 237: …..presence of various sugars l. 345 : What do the authors mean with "transmission to nature"? l. 387: "To test the effect of EIIAGlc on the PDE activity, EIIAGlc and a trace amount of EI and HPr were added together with either glucose to dephosphorylate or PEP to phosphorylate EIIAGlc." Under the described conditions, glucose will not dephosphorylate P~EIIAGlc. Dephosphorylation of P~EIIAGlc requires in addition the membrane protein PtsG. Did the authors include PtsGcontaining membrane fragments in the assay mixture? In fact, is glucose-mediated dephosphorylation necessary? Purified EIIAGlc is probably present at 99% in dephosphorylated form. As long as no PEP is added, phosphorylation will not occur! The same point applies to the legend of Fig. 2. In addition, what was the purpose of adding glucose in Supplem. Fig. 3? Glucose should not affect PdeS activity; it arrives as Glc-6-P in the cell. l. 440 and 459: What does " " stand for? Fig. 5: How often was the confocal laser scanning microscopy experiment of the infected D. melanogaster intestine repeated and were the results always similar to those presented in Fig. 5? Suppl. Fig. 2b: The authors present 3 symbols for VC1710 alone, EIIAGlc, and P~EIIAGlc. However, in the figure I see only black and grey squares, but no white squares for unphosphorylated EIIAGlc.

Josef Deutscher
Reviewer #2 (Remarks to the Author): In the manuscript entitled "The sugar-mediated regulation of c-di-GMP phosphodiesterase in Vibrio cholerae" Heo et al describe a novel EllAGlc binding partner encoded by the locus VC1710, which has an EAL domain and represses V. cholerae biofilm formation by degrading c-di-GMP. They name this protein PdeS. The results supporting the conclusion that PdeS is a phosphodiesterase that is activated by phosphorylated EllAGlc and inhibited by dephosphorylated EllAGlc are convincing. EllAGlc is known to regulate cellular physiology through its interactions with multiple partners. Here the authors have identified a novel partner of EllAGlc. However, there are discrepancies between these results and published results. These may simply be the result of strain differences as V. cholerae is known to undergo genetic drift in laboratory culture, and N16961 has been in the laboratory for decades [1]. This should be discussed. Furthermore, the design of the biofilm and Drosophila experiments raises some questions. Specific comments follow: 1) Line 57: Transport of sugars by the PTS has been comprehensively studied by the Dalia lab [2]. The authors should consider referencing this manuscript.
2) Lines 68-70: This statement suggests that only one component of the PTS regulates biofilm formation and that this component has not been identified. This statement does not faithfully represent the literature. From published work, there is strong evidence that phosphorylated Enzyme I represses biofilm formation [3,4]. Furthermore, at least two partners of EllAGlc have previously been shown to repress biofilm formation. These are MshH, the E. coli CsrD homolog that also contains GGDEF and EAL motifs, and adenylate cyclase [4][5][6][7][8]. It would be more appropriate to state that regulation of biofilm formation by the PTS has not been fully elucidated.
3) Lines 85-87: The authors suggest that no direct link between the PTS and biofilm formation has been established. Certainly, a regulatory link between these two processes has been established in the case of EllAGlc partners adenylate cyclase and MshH/CsrD. The transcription factor CRP, which binds cAMP, has been shown to regulate transcription of diguanylate cyclases and phosphodiesterases [8]. 4) Lines 103-106 and Supplementary Figure 1. A) these data set up the entire manuscript. Therefore, I suggest they be moved to the main Figures. B) These data do not seem to be consistent with previously published data showing that EllAGlc activates biofilm formation [9]. However, the authors do not directly compare biofilm formation by wild-type V. cholerae strain and the crr mutant, and this is curious. Upon digging deeper, the authors report biofilm formation as biofilm quantification divided by planktonic growth, so the question arises as to whether these measurements cannot be compared because planktonic growth is different for the wild-type and mutant. If so, a growth curve should be shown to make clear what the differences are. C) The authors divide biofilm formation by planktonic growth, presumably to account for growth differences. This is not the appropriate way to normalize for growth differences. If a strain makes a bigger biofilm, there will be less planktonic cells in the medium. Therefore, the planktonic growth measurements may reflect either a difference in growth or a difference in biofilm formation. Total growth (biofilm+planktonic) is a better method for quantifying growth differences. 5) Lines 106-108: The authors note that cAMP has no effect on biofilm formation. This is contrary to what has been published in the literature multiple times with different V. cholerae strains. What concentration of cAMP was tested? I was unable to find this easily in the Figure Legends  At what point in the procedure were PEP and glucose added? Before the lysate was made, after the lysate was made, or after the complexes had been pulled down on the affinity beads? Also, would one expect EllAGlc protein partners to already be complexed to native EllAGlc, since wild-type V. cholerae was used to prepare the lysates? How much His-EllAGlc was added to the mixture and was this in excess compared to the native amounts of EllAGlc in wild-type cells? 8) Although there have been screens of all the c-di-GMP modulatory enzymes in the V. cholerae genome [10][11][12], VC1710 has not appeared in any of these as a biofilm-regulatory protein. This likely also represents a strain difference. In fact, the sum total of what was published about VC1710 prior to this manuscript was that it bound c-di-GMP [13]. This should be discussed, and all these works should be referenced. 9) Line 249: The use of the Drosophila model to test the function of PdeS is innovative, but there are some questions about the experimental model. A) The methods state that the flies were administered "a medium" containing either 5% glucose or mannitol, but the nature of this medium is not elucidated anywhere in the text. This should be clarified. B) In previous publications, a concentration of approximately 10e8 V. cholerae/ml in LB has been used to infect Drosophila. Here, the authors use "medium containing 10e11 V. cholerae." Because no denominator is given, volume or otherwise, it is impossible to compare this infection protocol to the more commonly used one. A denominator should be given. C) In more conventional models, the infection has been allowed to proceed over at least 24 hours. Here, the authors give the flies access to medium containing V. cholerae for just two hours, and then "wash-out" the bacteria for 2 hours with sterile medium. This is an extremely accelerated experiment. It is not clear to me why the authors chose such a short time. Is it possible that the difference diminishes over time? This should be discussed. D) The question arises of how consistent infection is being obtained in such a short period. Intake varies, especially in the short term, depending on whether the flies "like" the medium/bacteria or not. Therefore, the authors should measure total intake over this two hour period under each condition. E) The authors observe a difference in V. cholerae colonization in the anterior midgut. This is in contrast to most infection models in which pathogens colonize the posterior midgut. Therefore, the authors could be observing ingested bacteria that are passing through. A good way to prove this is not the case is to normalize to passage of a fluorescent or colored food additive. For instance, showing that the additive is excreted, while V. cholerae remains. F) It is possible that inappropriate biofilm formation in the anterior midgut is responsible for the differences observed by the authors. Good proof of this would be to test vps and vps pdeS mutants as well. c-di-GMP controls many things besides biofilm formation. While previous studies demonstrated that PTS systems regulate biofilm formation in V. cholerae, interplay between PTS systems and c-di-GMP signaling systems has not been studied. Therefore the work presented in this study is significant.
Comments: 1) It was shown previously that "regulation of biofilm formation by EIIAGlc does not require phosphorylation of the conserved histidine at position 91" DOI:10.1128/JB.00213-10. It is important to generate an EIIAGlc dephosphomimetic mutant on the chromosome and analyze PdeS interaction and biofilm formation. This also serves as a genetic control for the studies performed.
3) Please provide an explanation for the differences in biofilm formation by N16961in Fig 3a and 3b; indicate the amount of inducer used and include no inducer control. 4) Please provide concentrations of EI and HPr used in the experiments. The description provided ("trace amounts") is not sufficient for reproducibility studies.

Dear Reviewers,
We really appreciate your constructive and invaluable suggestions to improve our manuscript. We tried to go over the manuscript carefully and make it clearer and more accessible to readers. In this version, we added more experimental data and modified the manuscript as recommended. All changes were highlighted in red in the manuscript. 6. We addressed all other issues raised by the three reviewers and made corrections for trivial mistakes.

Point-to-point response to the reviewers' comments
Reviewer #1 (Remarks to the Author): The PEP:glycose phosphotransferase system (PTS) transports and phosphorylates a wide range of carbohydrates. In addition, it has been shown to regulate numerous important cellular functions in response to the availability of efficiently metabolizable carbon sources.
The PTS was also suggested to affect the virulence of certain pathogens. Regulation by the PTS is usually mediated via the phosphorylation state of its protein components.
We changed the sentence to "… and transmission to new hosts" (line 395).

387: "To test the effect of EIIAGlc on the PDE activity, EIIAGlc and a trace amount of EI
and HPr were added together with either glucose to dephosphorylate or PEP to phosphorylate EIIAGlc." Under the described conditions, glucose will not dephosphorylate P~EIIAGlc.
Dephosphorylation of P~EIIAGlc requires in addition the membrane protein PtsG. Did the authors include PtsG-containing membrane fragments in the assay mixture? In fact, is glucose-mediated dephosphorylation necessary? Purified EIIAGlc is probably present at 99% in dephosphorylated form. As long as no PEP is added, phosphorylation will not occur! The same point applies to the legend of Fig. 2. In addition, what was the purpose of adding glucose in Supplem. Fig. 3? Glucose should not affect PdeS activity; it arrives as Glc-6-P in the cell.  We presented symbols for experimental data in each panel.

Reviewer #2 (Remarks to the Author):
In the manuscript entitled "The sugar-mediated regulation of c-di-GMP phosphodiesterase in Vibrio cholerae" Heo et al describe a novel EllAGlc binding partner encoded by the locus VC1710, which has an EAL domain and represses V. cholerae biofilm formation by degrading c-di-GMP. They name this protein PdeS. The results supporting the conclusion that PdeS is a phosphodiesterase that is activated by phosphorylated EllAGlc and inhibited by dephosphorylated EllAGlc are convincing. EllAGlc is known to regulate cellular physiology through its interactions with multiple partners. Here the authors have identified a novel partner of EllAGlc. However, there are discrepancies between these results and published results. These may simply be the result of strain differences as V. cholerae is known to undergo genetic drift in laboratory culture, and N16961 has been in the laboratory for decades [1]. This should be discussed. Furthermore, the design of the biofilm and Drosophila experiments raises some questions. Specific comments follow:

Thank you for the invaluable comments and suggestions.
1) Line 57: Transport of sugars by the PTS has been comprehensively studied by the Dalia lab [2]. The authors should consider referencing this manuscript.

The reference was cited in the manuscript (line 57).
2) Lines 68-70: This statement suggests that only one component of the PTS regulates biofilm formation and that this component has not been identified. This statement does not faithfully represent the literature. From published work, there is strong evidence that phosphorylated Enzyme I represses biofilm formation [3,4]. Furthermore, at least two partners of EllAGlc have previously been shown to repress biofilm formation. These are MshH, the E. coli CsrD homolog that also contains GGDEF and EAL motifs, and adenylate cyclase [4][5][6][7][8]. It would be more appropriate to state that regulation of biofilm formation by the PTS has not been fully elucidated.
We revised this part as suggested, referring to the previously reported binding partners of EIIA Glc (lines 69-70, 86-97).

3) Lines 85-87: The authors suggest that no direct link between the PTS and biofilm
formation has been established. Certainly, a regulatory link between these two processes has been established in the case of EllAGlc partners adenylate cyclase and MshH/CsrD. The transcription factor CRP, which binds cAMP, has been shown to regulate transcription of diguanylate cyclases and phosphodiesterases [8]. A) these data set up the entire manuscript. Therefore, I suggest they be moved to the main

Figures.
We moved Supplementary Figure 1 to the main Figure 1a, as suggested. B) These data do not seem to be consistent with previously published data showing that EllAGlc activates biofilm formation [9]. However, the authors do not directly compare biofilm formation by wild-type V. cholerae strain and the crr mutant, and this is curious.
Upon digging deeper, the authors report biofilm formation as biofilm quantification divided by planktonic growth, so the question arises as to whether these measurements cannot be compared because planktonic growth is different for the wild-type and mutant. If so, a growth curve should be shown to make clear what the differences are.
We compared biofilm formation by wild-type V. cholerae and its crr mutant in LB medium, as shown in Figure 1a, and we added the following sentence: "Interestingly, while the crr mutant had a similar growth (Supplementary Fig. 1), this mutant exhibited a higher level of biofilm formation compared to the wild-type strain in LB medium (Fig. 1a), which is contrary to a previous study 5 ." (lines 116-119) C) The authors divide biofilm formation by planktonic growth, presumably to account for growth differences. This is not the appropriate way to normalize for growth differences. If a strain makes a bigger biofilm, there will be less planktonic cells in the medium. Therefore, the planktonic growth measurements may reflect either a difference in growth or a difference in biofilm formation. Total growth (biofilm+planktonic) is a better method for quantifying growth differences.
To avoid any possible complication due to different growth rates among strains, we measured the total growth and the amount of biofilm formation separately (See figures 1a, 3a, 3c and 4 and Supplementary Figures 1, 6a and 7).

5)
Lines 106-108: The authors note that cAMP has no effect on biofilm formation. This is contrary to what has been published in the literature multiple times with different V. cholerae strains. What concentration of cAMP was tested? I was unable to find this easily in the Figure   Legends or the Experimental Procedures. If an adequate amount of cAMP was used, then the absence of an effect of cAMP on biofilm formation could be the result of strain variation or genetic drift in the laboratory. This discrepancy should be discussed.
We added 5 mM cAMP into the medium, which was previously shown to be an effective concentration 6 . In previous studies, different effects of cAMP-CRP on biofilm formation have been reported among various V. cholerae strains. While biofilm formation was negatively regulated by the cAMP-CRP complex in the C6706 and C6728 strains 2,7 , the supplementation of the growth medium with various concentrations of cAMP had no effect on the total growth and biofilm accumulation by a crr mutant of the MO10 strain 3 . Herein we report that the sugar-dependent regulation of biofilm formation is hardly affected by cAMP-CRP in the N16961 strain. This may simply be the result of strain differences, as V. cholerae is known to undergo genetic drift in laboratory culture 8 . We discussed about these differences in the revised manuscript, as the reviewer suggested (lines 382-390). cholerae genome [10][11][12], VC1710 has not appeared in any of these as a biofilm-regulatory protein. This likely also represents a strain difference. In fact, the sum total of what was published about VC1710 prior to this manuscript was that it bound c-di-GMP [13]. This should be discussed, and all these works should be referenced.

We added the information as suggested (lines 176-177).
9) Line 249: The use of the Drosophila model to test the function of PdeS is innovative, but there are some questions about the experimental model.
A) The methods state that the flies were administered "a medium" containing either 5% glucose or mannitol, but the nature of this medium is not elucidated anywhere in the text.
This should be clarified.
We made clarifications as suggested (lines 295-296, 314, 316, 524, and 533 and the legend to Figure 5). B) In previous publications, a concentration of approximately 10e8 V. cholerae/ml in LB has been used to infect Drosophila. Here, the authors use "medium containing 10e11 V. cholerae." Because no denominator is given, volume or otherwise, it is impossible to compare this infection protocol to the more commonly used one. A denominator should be given. (lines 296, 314-315 and 524).

We made corrections as suggested
C) In more conventional models, the infection has been allowed to proceed over at least 24 hours. Here, the authors give the flies access to medium containing V. cholerae for just two hours, and then "wash-out" the bacteria for 2 hours with sterile medium. This is an extremely accelerated experiment. It is not clear to me why the authors chose such a short time. Is it possible that the difference diminishes over time? This should be discussed. Intake varies, especially in the short term, depending on whether the flies "like" the medium/bacteria or not. Therefore, the authors should measure total intake over this two hour period under each condition.

It is generally accepted that colonization of the
We quantified food consumption using FD&C Blue #1 dye during bacterial infection as suggested, and we added this result in the revised manuscript (Figure 5d).
E) The authors observe a difference in V. cholerae colonization in the anterior midgut. This is in contrast to most infection models in which pathogens colonize the posterior midgut.
Therefore, the authors could be observing ingested bacteria that are passing through. A good way to prove this is not the case is to normalize to passage of a fluorescent or colored food additive. For instance, showing that the additive is excreted, while V. cholerae remains. F) It is possible that inappropriate biofilm formation in the anterior midgut is responsible for the differences observed by the authors. Good proof of this would be to test vps and vps pdeS mutants as well. c-di-GMP controls many things besides biofilm formation.  pdeS mutants (Fig. 5a). As reported previously 16 While previous studies demonstrated that PTS systems regulate biofilm formation in V.
cholerae, interplay between PTS systems and c-di-GMP signaling systems has not been studied. Therefore the work presented in this study is significant.
Please incorporate this work into your discussion.

We revised the manuscript as suggested (line 86-97).
3) Please provide an explanation for the differences in biofilm formation by N16961in Fig 3a   and 3b; indicate the amount of inducer used and include no inducer control.
We added sentences explaining for the differences in biofilm in Figure 3a