Interplay between Cyclic AMP-Cyclic AMP Receptor Protein and Cyclic di-GMP Signaling in Vibrio cholerae Biofilm Formation

Vibrio cholerae is a facultative human pathogen. The abilityof V. cholerae to form biofilms is crucial for its survivalin aquatic habitats between epidemics and is advantageous forhost-to-host transmission during epidemics. Formation of maturebiofilms requires the production of extracellular matrix components,including Vibrio polysaccharide (VPS) and matrix proteins. Biofilmformation is positively controlled by the transcriptional regulatorsVpsR and VpsT and is negatively regulated by the quorum-sensingtranscriptional regulator HapR, as well as the cyclic AMP (cAMP)-cAMPreceptor protein (CRP) regulatory complex. Transcriptome analysisof cyaA (encoding adenylate cyclase) and crp (encoding cAMPreceptor protein) deletion mutants revealed that cAMP-CRP negativelyregulates transcription of both VPS biosynthesis genes and genesencoding biofilm matrix proteins. Further mutational and expressionanalysis revealed that cAMP-CRP negatively regulates transcriptionof vps genes indirectly through its action on vpsR transcription.However, negative regulation of the genes encoding biofilm matrixproteins by cAMP-CRP can also occur independent of VpsR. Transcriptomeanalysis also revealed that cAMP-CRP regulates the expressionof a set of genes encoding diguanylate cyclases (DGCs) and phosphodiesterases.Mutational and phenotypic analysis of the differentially regulatedDGCs revealed that a DGC, CdgA, is responsible for the increasein biofilm formation in the ${Delta}$ crp mutant, showing the connectionbetween of cyclic di-GMP and cAMP signaling in V. cholerae.

INTRODUCTION

Top

INTRODUCTION

Vibrio cholerae, the causative agent of cholera (26), is a naturalinhabitant of aquatic environments (14). Seasonal cholera outbreaksoccur where the disease is endemic and can spread worldwide(14, 34). The ability of V. cholerae to cause epidemics is linkedto its ability to survive in natural aquatic ecosystems. Oneimportant factor for environmental survival and transmissionof V. cholerae is its ability to form biofilms (14, 60, 63).Biofilms are surface-attached microbial communities composedof microorganisms and the extrapolymeric substances that theyproduce (9). The biofilm mode of growth is the preferred lifestylein the microbial world, as it enhances survival in natural settings.In addition, biofilms protect the constituent microbes frompredators, such as protozoa and viruses, and from toxic compounds,such as antimicrobial agents (4, 10, 13, 36, 41, 63).

The process of biofilm development in V. cholerae can be dividedinto distinct stages: transport and attachment of bacteria tothe surface, colonization of the attached surface, formationof a monolayer of cells, and synthesis of the extracellularmatrix, leading to formation of a mature biofilm with a characteristicthree-dimensional (3D) architecture. The Vibrio polysaccharide(VPS), encoded by the vps genes, is essential for the developmentof 3D biofilm structures (63). The vps genes are clustered intwo regions on the large chromosome of V. cholerae O1 El Tor;the vps-I cluster consists of vpsU (VC0916) and vpsA to vpsK(VC0917 to VC0927), and the vps-II cluster consists of vpsLto vpsQ (VC0934 to VC0939). Recently, we identified proteincomponents of the biofilm matrix of V. cholerae and showed thatthe RbmA, RbmC, and Bap1 proteins are also required for theformation of a wild-type biofilm. Mutants that are not ableto produce these matrix proteins form biofilms that are structurallyunstable (15, 16).

The regulation of biofilm formation in V. cholerae is complexand involves several transcriptional regulators. Two proteinsthat positively regulate VPS production and biofilm formationhave been identified, VpsR and VpsT. Disruption of vpsR preventsexpression of the vps genes and production of VPS, and it eliminatesformation of typical 3D biofilm structures (61). A vpsT mutantexhibits reduced vps gene expression and biofilm-forming capacity(7). A population-density-dependent regulatory system, knownas the quorum-sensing system, negatively regulates biofilm formationin V. cholerae. HapR is the master regulator of the quorum-sensingregulatory system, and a hapR mutant has increased biofilm-formingcapacity (19, 62, 64). Consistent with this observation, expressionof the vps genes, including vpsR and vpsT, is increased in thehapR mutant (62). In addition, a second messenger, cyclic di-GMP(c-di-GMP), which is produced by diguanylate cyclases (DGCs)containing a GGDEF amino acid motif and is degraded by phosphodiesterases(PDEs) that have EAL or HD-GYP domains (8, 46, 49), positivelyregulates biofilm formation in V. cholerae (3, 31, 56). We recentlydetermined that cdgA (encoding a DGC), whose transcription ispositively regulated by VpsR and negatively regulated by HapR,positively regulates biofilm formation in V. cholerae (2).

The cyclic AMP (cAMP)-cAMP receptor protein (CRP) regulatorycomplex was recently identified as a negative regulator of biofilmformation in V. cholerae (29, 30). These studies showed thatcAMP-CRP negatively regulates vpsL and vpsT expression and positivelyregulates vpsR expression. Additional study showed that growthin the presence of glucose, which leads to a decrease in cellularcAMP levels, induces biofilm formation, while addition of cAMPto a growth medium leads to a decrease in biofilm formationin wild-type V. cholerae (23). To further evaluate the mechanismby which cAMP-CRP negatively regulates biofilm formation inV. cholerae, we determined whole-genome expression profilesof cyaA (encoding adenylate cyclase) and crp (encoding CRP)deletion mutants. Our analysis revealed that cAMP-CRP negativelyregulates transcription of VPS biosynthesis genes and genesencoding biofilm matrix proteins. cAMP-CRP negatively regulatestranscription of both vps genes and the genes encoding biofilmmatrix proteins indirectly, through its action on vpsR transcription.In addition, cAMP-CRP can also negatively regulate transcriptionof the genes encoding biofilm matrix proteins independent ofVpsR. We also determined that cAMP-CRP regulates the expressionof a set of genes encoding DGCs and PDEs. Through mutationaland phenotypic analysis, we showed that CdgA is largely responsiblefor the increased transcription of vps and biofilm matrix proteingenes, as well as enhanced biofilm formation in a ${Delta}$ crp mutant,revealing the connection between c-di-GMP and cAMP-CRP signalingin V. cholerae.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions. Bacterial strains and plasmids used in this study are listedin Table 1. All V. cholerae and Escherichia coli strains wereroutinely grown aerobically in Luria-Bertani (LB) medium (1%tryptone, 0.5% yeast extract, 1% NaCl; pH 7.5) at 30 and 37°C,respectively, unless otherwise noted. The E. coli DH10B andCC118 ${lambda}$ pir strains were used for DNA manipulation, while the E.coli S17-1 ${lambda}$ pir strain was used for conjugation with V. choleraeA1552. Conjugation with other V. cholerae strains (C6706, N16961,and MO10) was carried out using the E. coli SM10 ${lambda}$ pir strain.Agar medium contained 1.5% granulated agar (Difco), unless otherwisenoted. Ampicillin, rifampin, and streptomycin were used at aconcentration of 100 µg/ml, while gentamicin was usedat a concentration of 50 µg/ml. We observed that ${Delta}$ cyaAand ${Delta}$ crp mutants had increased doubling times in LB medium at30°C compared to the wild type (data not shown). This findingis similar to the growth defect reported for a crp mutant grownin LB medium at 37°C (52). To minimize the effect of reducedgrowth rates in our experiments, expression profiling and β-galactosidaseassays were carried out with cultures grown to stationary phase.

View this table:
[in this window]
[in a new window]

TABLE 1. Bacterial strains and plasmids used in this study

Recombinant DNA techniques. DNA manipulations were carried out by using standard moleculartechniques (47). Restriction and DNA modification enzymes werepurchased from New England Biolabs. PCRs were carried out usingprimers purchased from Operon Technologies (Table 2) and a high-fidelityPCR kit (Roche). DNA sequencing was carried out by the UC BerkeleyDNA Sequencing Facility.

View this table:
[in this window]
[in a new window]

TABLE 2. Sequences of oligonucleotides used in this study

Generation of in-frame deletion mutants. Deletion mutants of V. cholerae strains were generated by usinga previously described protocol (16). The DNA sequences of theconstructed deletion plasmids were verified by DNA sequencing.Primers used in the construction of the deletion plasmids areshown in Table 2.

Pellicle formation and motility assays. Pellicle formation experiments were carried out using glassculture tubes (18 by 150 mm) containing 5 ml of medium. Themedium was inoculated with overnight cultures, resulting in200-fold dilution. The tubes were incubated at 30°C withoutshaking for 2 days. LB soft agar plates (0.3% agar) were usedto determine the motility of the bacterial strains. The diameterof each migration zone (in cm) was measured after 18 h of incubationat 30°C. Assays were repeated with at least two biologicalreplicates.

Generation of lacZ transcriptional fusion constructs. lacZ transcriptional fusions with promoters of rbmC and bap1were constructed by cloning the PCR-amplified $~$ 300-bp promoterregions immediately upstream of the start codons of rbmC andbap1 into pRS415 (51) as described previously (15). The resultingtranscriptional fusion plasmids were sequenced. The plasmidswere electroporated into V. cholerae strains containing a lacZin-frame deletion. The primers used for amplification of thepromoter regions are shown in Table 2.

β-Galactosidase assays. β-Galactosidase assays were carried out by using a protocolsimilar to that described by Miller (38). Briefly, overnightcultures were diluted 200-fold in LB medium supplemented withampicillin and incubated at 30°C for 10 h with shaking (200rpm). The optical densities at 600 nm (OD₆₀₀) of the stationary-phasecultures were determined, and 1-ml portions of the cultureswere harvested and washed with 1 ml of buffer Z (16.1 g/literNa₂HPO₄·7H₂O, 5.5 g/liter NaH₂PO₄·H₂O, 0.75 g/literKCl, 0.246 g/liter MgSO₄·7H₂O; pH 7.0). Cells were lysedby resuspending a cell pellet in 1 ml of buffer Z containing0.69% β-mercaptoethanol, 0.02% cetyltrimethylammonium bromide,and 0.01% deoxycholic acid (sodium salt), followed by incubationat room temperature for 5 min. Cell lysates (100 µl) ofdifferent dilutions were pipetted into flat-bottom 96-well microtiterplates, and 20-µl portions of an o-nitrophenyl-β-D-galactopyranosidesolution (4 mg/ml) were added, followed by incubation at 30°Cuntil sufficient color development was observed. The reactionswere stopped by adding 50 µl of 1 M Na₂CO₃, and the colorintensities were measured at OD₄₂₀ and OD₅₅₀. The duration ofcolor development was noted, and the β-galactosidase activity(expressed in Miller units) was calculated as previously described(15, 38). The assays were repeated with at least two differentbiological replicates and eight technical replicates.

Generation of gfp-tagged strains and confocal laser scanning microscopy (CLSM). V. cholerae strains were chromosomally tagged with the geneencoding green fluorescent protein (gfp), using a previouslydescribed procedure (2, 15). Non-flow-cell experiments werecarried out using Lab-Tek II chambered coverglass systems (NalgeNunc) and a previously described protocol (2). Briefly, overnightcultures were diluted to obtain an OD₆₀₀ of 0.2, and 3-ml portionsof the diluted cultures were placed into chambers and incubatedat 30°C for 8 h. The chambers were then washed twice with1 ml of LB medium. Biofilms formed by non-gfp-tagged strainswere stained for 15 min at room temperature in the dark with1 ml of 5 µM SYTO9 (Molecular Probes). Images of the biofilmsformed in the chambers were acquired using a Zeiss Axiovert200 M laser scanning microscope. 3D images of the biofilms werereconstructed using IMARIS software (Bitplane) and were quantifiedusing the COMSTAT program (22). Non-flow-cell experiments werecarried out with at least two different biological replicates.

Biofilm formation assays. Biofilms were formed in 96-well polyvinyl chloride microtiterplates by using 100-µl portions of overnight culturesdiluted to obtain an OD₆₀₀ of 0.2. The microtiter plates wereincubated at 30°C for 8 h. Crystal violet staining and ethanolsolubilization were carried out as previously described (15,63). The assays were repeated with two different biologicalreplicates and eight technical replicates.

RNA isolation. Total RNA was isolated from V. cholerae strains in stationarygrowth phase by using a previously described protocol (62).Briefly, overnight cultures of V. cholerae grown in LB mediumat 30°C with shaking (200 rpm) were diluted 200-fold inLB medium and incubated at 30°C for 10 h. Aliquots (2 ml)of the cultures were collected and centrifuged for 2 min atroom temperature. The cell pellets were immediately resuspendedin 1 ml of TRIzol (Invitrogen) and stored at –80°C.Total RNA was isolated according to the manufacturer's instructions.To remove contaminating DNA, total RNA was incubated with RNase-freeDNase I (Ambion), and an RNeasy mini kit (Qiagen) was used toclean up RNA after DNase digestion.

Whole-genome expression profiling. Whole-genome expression profiling was performed by using a previouslydescribed procedure (3). A common reference RNA was used, whichcontained equal amounts of total RNA isolated from V. choleraecells grown to stationary phase in LB medium. Normalized signalratios were obtained with LOWESS print tip normalization usingthe Bioconductor packages (http://www.bioconductor.org) in theR environment (18). Differentially regulated genes were determined(with three biological and two technical replicates for eachdata point) using the Significance Analysis of Microarrays (SAM)software (58) with a $≥$ 2-fold difference in gene expression anda false discovery rate (FDR) of $≤$ 1% as cutoff values, unlessotherwise noted.

qPCR. Quantitative PCR (qPCR) was carried out by first synthesizingcDNA from 1 µg of a total RNA sample using an iScriptcDNA synthesis kit (Bio-Rad). The cDNA product was then diluted1:4 with water, and 4 µl was used as a template with 12pmol of each qPCR primer (Table 2) in a PCR performed with theExpand high-fidelity PCR system (Roche). The PCR conditionswere as follows: 94°C for 2 min and then 25 cycles of 94°Cfor 30 s, 55°C for 30 s, and 72°C for 30 s and a finalincubation at 72°C for 2 min. The amplified products wereanalyzed on a 2% agarose gel and were quantified using the ImageQuant5.2 software (Molecular Dynamics). The intensity of each DNAband was normalized to that of the corresponding recA band amplifiedwith primers RecA578 and RecA863 (29). The data presented beloware from three biological replicates, and reaction mixturescontaining no template or reverse transcriptase were used asnegative controls.

RESULTS

Top

RESULTS

Identification of genes differentially regulated in ${Delta}$ cyaA and ${Delta}$ crp mutants. cAMP-CRP negatively regulates biofilm formation in V. cholerae.To further understand how cAMP-CRP regulates biofilm formation,we generated in-frame cyaA (VC0122) and crp (VC2614) deletionmutants of our prototype V. cholerae O1 El Tor A1552 strainand performed whole-genome expression profiling of these mutants.The gene expression data were analyzed by using the SAM softwareand the following criteria to define significantly regulatedgenes, unless otherwise indicated: an FDR of $≤$ 1% and a $≥$ 2-foldtranscript abundance difference between samples. This analysisrevealed that cAMP-CRP differentially regulates transcriptionof a large set of genes and that the overall gene expressionprofiles of the ${Delta}$ cyaA and ${Delta}$ crp mutants are similar (see Fig. S1in the supplemental material). Altogether, 889 genes (22.9%of the genome) and 822 genes (21.2% of the genome) are differentiallyregulated in the ${Delta}$ cyaA and ${Delta}$ crp mutants compared to the wild type,respectively. Of the 889 differentially regulated genes in the ${Delta}$ cyaA mutant, 431 are upregulated and 458 are downregulated,whereas of the 822 differentially regulated genes in the ${Delta}$ crpmutant, 386 are upregulated and 436 are downregulated. All thedifferentially regulated genes are shown in Tables S1 and S2in the supplemental material.

In this study, our main objective was to determine how cAMP-CRPnegatively regulates biofilm formation. Thus, for the genesthat are differentially regulated by cAMP-CRP, we focused ontwo sets of genes: the genes required for biofilm matrix productionand its regulation and the genes predicted to be involved inthe production and degradation of c-di-GMP, as this second messengerregulates biofilm formation in V. cholerae.

cAMP-CRP negatively regulates transcription of the vps, rbmA, rbmC, and bap1 genes. As expected for a negative regulator of biofilm formation, thelevels of expression of vps genes and vpsT were higher in boththe ${Delta}$ cyaA and ${Delta}$ crp mutants than in the wild type (Table 3). Geneexpression profiling also revealed that the expression of hapRwas decreased in both the ${Delta}$ cyaA and ${Delta}$ crp mutants. This findingis consistent with previously described results (29, 50). Inaddition to these genes, we observed that the levels of expressionof the genes encoding the biofilm matrix proteins, rbmA, rbmC,and bap1, were higher in both the ${Delta}$ cyaA and ${Delta}$ crp mutants thanin the wild type. To verify this finding, we monitored transcriptionof the genes encoding biofilm matrix proteins by using rbmA-lacZ,rbmC-lacZ, and bap1-lacZ fusion constructs and measuring β-galactosidaseactivities. In parallel, we also monitored transcription ofthe genes involved in VPS biosynthesis using a vpsL-lacZ fusionconstruct. As expected, vpsL transcription was increased inboth the ${Delta}$ cyaA mutant (205-fold) and the ${Delta}$ crp mutant (142-fold)compared to the wild type (Fig. 1A). Transcription of the genesencoding biofilm matrix proteins was also increased in the ${Delta}$ cyaAmutant (15-fold for rbmA, 36-fold for rbmC, and 32-fold forbap1) and the ${Delta}$ crp mutant (13-fold for rbmA, 27-fold for rbmC,and 15-fold for bap1) compared to the wild type (Fig. 1B to D).This indicates that cAMP-CRP negatively regulates transcriptionof genes required for the production of both VPS and biofilmmatrix proteins.

View this table:
[in this window]
[in a new window]

TABLE 3. Differentially expressed genes involved in biofilm matrix production and c-di-GMP signaling in ${Delta}$ cyaA and ${Delta}$ crp mutants compared to the wild type^a

View larger version (18K):
[in this window]
[in a new window]

FIG. 1. cAMP-CRP negatively regulates expression of genes involved in VPS biosynthesis and biofilm matrix protein production. β-Galactosidase assays of wild-type, ${Delta}$ cyaA, and ${Delta}$ crp strains harboring (A) vpsL-lacZ, (B) rbmA-lacZ, (C) rbmC-lacZ, and (D) bap1-lacZ fusion constructs were performed. The data are representative of at least two independent experiments. The error bars indicate standard deviations.

cAMP-CRP negatively regulates transcription of vpsT and vpsR in V. cholerae O1 El Tor A1552. To understand how cAMP-CRP negatively regulates transcriptionof vps genes and biofilm matrix protein genes, we analyzed howtranscription of vpsR and vpsT is altered in ${Delta}$ cyaA and ${Delta}$ crp mutants.Using vpsT-lacZ and vpsR-lacZ fusion constructs, we determinedthat vpsT and vpsR transcription was increased in the ${Delta}$ cyaA mutant(178-fold for vpsT and 4-fold for vpsR) and the ${Delta}$ crp mutant (144-foldfor vpsT and 4-fold for vpsR) compared to the wild type, indicatingthat cAMP-CRP also negatively regulates transcription of vpsTand vpsR (Fig. 2A and B). Interestingly, transcription of vpsRwas shown to be positively regulated by cAMP-CRP in the V. choleraeO1 El Tor C7258 strain (30). Differences in the regulation ofvpsR by cAMP-CRP in our prototype strain, V. cholerae O1 ElTor A1552, and V. cholerae O1 El Tor C7258 prompted us to lookat vpsR regulation in other commonly used V. cholerae strains.To this end, we generated in-frame crp deletion mutants of V.cholerae strains N16961, C6706, and MO10 and introduced a reporterplasmid harboring a vpsR-lacZ fusion construct into these strains.We then monitored transcription of vpsR by measuring β-galactosidaseactivity (Fig. 2C). In our prototype A1552 ${Delta}$ crp strain, vpsR-lacZtranscription was increased 3.9-fold compared to the wild-typetranscription. N16961 ${Delta}$ crp also exhibited a similar increase intranscription of vpsR-lacZ (3.4-fold), while MO10 ${Delta}$ crp exhibiteda slight increase (1.3-fold) in vpsR-lacZ transcription comparedto the corresponding wild-type strains. On the other hand, inC6706 and transcription of vpsR-lacZ in C6706 ${Delta}$ crp did not differsignificantly. These results indicate that while cAMP-CRP negativelyregulates vpsR transcription in the A1552, N16961, and MO10(albeit slightly) strains of V. cholerae, there is no such regulationin strain C6706. Hence, the results are consistent with theidea that vpsR regulation by cAMP-CRP varies in strains of V.cholerae. Together, our results indicate that in our prototypestrain, cAMP-CRP represses biofilm formation in V. choleraeby negatively regulating transcription of the genes requiredfor VPS biosynthesis and matrix protein production, as wellas the genes encoding the positive transcriptional regulatorsVpsT and VpsR.

View larger version (15K):
[in this window]
[in a new window]

FIG. 2. cAMP-CRP negatively regulates vpsT and vpsR expression. (A and B) β-Galactosidase assays of wild-type strain A1552 and ${Delta}$ cyaA and ${Delta}$ crp mutants harboring (A) vpsT-lacZ and (B) vpsR-lacZ fusion constructs. (C) β-Galactosidase assays of different V. cholerae strains (A1552, N16961, C6706, and MO10) and ${Delta}$ crp deletion strains harboring the vpsR-lacZ fusion construct. (D and E) β-Galactosidase assays of wild-type, ${Delta}$ crp, ${Delta}$ hapR, and ${Delta}$ crp ${Delta}$ hapR strains harboring (D) vpsT-lacZ and (E) vpsR-lacZ fusion constructs. The data are representative of at least two independent experiments. The error bars indicate standard deviations.

crp is epistatic to hapR in the regulation of vpsT and vpsR transcription. Biofilm formation in V. cholerae is negatively regulated byboth HapR (19, 62, 64) and cAMP-CRP (29, 30). Since cAMP-CRPpositively regulates hapR transcription, repression of biofilmformation by cAMP-CRP could be mediated through HapR. Thus,to better evaluate the mechanism by which cAMP-CRP and HapRnegatively regulate biofilm formation, we analyzed vpsT andvpsR transcription in wild-type, ${Delta}$ crp, ${Delta}$ hapR, and ${Delta}$ crp ${Delta}$ hapR strainsharboring vpsT-lacZ and vpsR-lacZ fusion plasmids (Fig. 2D and E).We observed 27- and 3.2-fold increases in β-galactosidaseactivities in the ${Delta}$ hapR mutants harboring vpsT-lacZ and vpsR-lacZfusion plasmids, respectively, compared to the wild type. Theseresults are congruent with the results of our previous studiesshowing that HapR negatively regulates the expression of vpsTand vpsR in strain A1552 (2, 62). It is noteworthy that in V.cholerae strains C6706 and C7258 negative regulation of vpsRby HapR has not been observed (19, 30, 59), indicating that,similar to vpsR repression by cAMP-CRP, vpsR repression by HapRalso varies in strains of V. cholerae. Interestingly, similarincreases in the transcriptional levels of vpsT (210-fold inthe ${Delta}$ crp mutant and 231-fold in the ${Delta}$ crp ${Delta}$ hapR mutant) and vpsR(6.2-fold in the ${Delta}$ crp and mutant 6.0-fold in the ${Delta}$ crp ${Delta}$ hapR mutant)were observed for the ${Delta}$ crp and ${Delta}$ crp ${Delta}$ hapR mutants, indicating thatcrp is epistatic to hapR in regulating vpsT and vpsR transcription.

cAMP-CRP regulates rbmC and bap1 expression both through and independent of VpsR. Formation of mature biofilms in V. cholerae requires productionof VPS and the matrix proteins RbmA, RbmC, and Bap1. As discussedabove, cAMP-CRP negatively regulates the expression of bothvps genes and genes encoding matrix proteins. However, we havea limited understanding of the mechanism by which cAMP-CRP regulatestranscription of these genes and do not know whether vps andmatrix protein genes are regulated differently by cAMP-CRP.

We previously reported that VpsR is the most downstream regulatorof vps gene transcription and genes encoding matrix proteinsin the VpsT, VpsR, and HapR regulatory circuitry (2). We wantedto determine how cAMP-CRP contributes to this regulatory circuitry.To this end, we monitored transcription of vpsL, rbmC, and bap1using lacZ transcriptional fusion constructs in the wild-typestrain and ${Delta}$ crp, ${Delta}$ vpsT, ${Delta}$ vpsR, ${Delta}$ crp ${Delta}$ vpsT, and ${Delta}$ crp ${Delta}$ vpsR mutants (Fig.3). As discussed above, in the ${Delta}$ crp strain harboring the fusionconstruct vpsL-lacZ, transcription of vpsL was markedly increasedcompared to that in the wild type. Transcription of vpsL-lacZwas 1.7- and 4.6-fold lower in the ${Delta}$ vpsT and ${Delta}$ vpsR mutants, respectively,than in the wild type (Fig. 3A). This finding is consistentwith our previous report that the magnitude of regulation ofvps gene expression by VpsR is greater than that by VpsT (2).In the ${Delta}$ crp ${Delta}$ vpsT double-deletion mutant, a 2.1-fold increasein vpsL transcription was observed compared to the wild type.This slight increase in vpsL transcription was likely due todecreased expression of hapR in the ${Delta}$ crp genetic background.cAMP-CRP positively regulates hapR expression, and a decreasein hapR message abundance, in turn, leads to an increase invpsL and vpsR expression (2, 62). Unlike the findings for the ${Delta}$ crp ${Delta}$ vpsT mutant, the vpsL transcription in the ${Delta}$ crp ${Delta}$ vpsR mutantwas 3.3-fold lower than that in the wild type (similar to the ${Delta}$ vpsR mutant), indicating that VpsR acts downstream of cAMP-CRP.

View larger version (43K):
[in this window]
[in a new window]

FIG. 3. Analysis of the cAMP-CRP contribution to VpsR regulation of rbmC and bap1 expression. (A to C) β-Galactosidase assays of wild-type, ${Delta}$ crp, ${Delta}$ vpsT, ${Delta}$ crp ${Delta}$ vpsT, ${Delta}$ vpsR, and ${Delta}$ crp ${Delta}$ vpsR strains harboring (A) vpsL-lacZ, (B) rbmC-lacZ, and (C) bap1-lacZ fusion constructs. The data are representative of at least two independent experiments. The error bars indicate standard deviations. (D) Quantitative comparison of biofilm formation by wild-type, ${Delta}$ crp, ${Delta}$ vps-I, ${Delta}$ crp ${Delta}$ vps-I, ${Delta}$ vps-I ${Delta}$ vps-II, ${Delta}$ crp ${Delta}$ vps-I ${Delta}$ vps-II, ${Delta}$ rbmA, ${Delta}$ crp ${Delta}$ rbmA, ${Delta}$ bap1, and ${Delta}$ crp ${Delta}$ bap1 strains. The data are representative of two independent experiments. The error bars indicate standard deviations. (E) Biofilms formed after 8 h of incubation at 30°C in a non-flow-cell system by the wild-type, ${Delta}$ crp, ${Delta}$ vps-I, ${Delta}$ crp ${Delta}$ vps-I, ${Delta}$ vps-I ${Delta}$ vps-II, ${Delta}$ crp ${Delta}$ vps-I ${Delta}$ vps-II, ${Delta}$ rbmA, ${Delta}$ crp ${Delta}$ rbmA, ${Delta}$ bap1, and ${Delta}$ crp ${Delta}$ bap1 strains. Biofilms were stained with SYTO9, and images were acquired by CLSM. The large images are images of the upper surfaces of biofilms, and the images below and to right of the large images are orthogonal views. Bars = 40 µm.

Transcription of rbmC and bap1 was increased in the ${Delta}$ crp mutantcompared to the wild type. Although deletion of vpsR resultedin decreases in transcription of rbmC (7.6-fold) and bap1 (6.8-fold),deletion of vpsT did not significantly alter rbmC and bap1 transcriptioncompared to the wild type (Fig. 3B and C). This result suggeststhat, like the findings for vpsL transcription, the magnitudeof transcriptional regulation by VpsR is greater than the magnitudeof transcriptional regulation by VpsT for rbmC and bap1. Intriguingly,unlike the results for vpsL transcription, deletion of crp inboth the ${Delta}$ vpsT and ${Delta}$ vpsR genetic backgrounds resulted in increasedtranscription of rbmC (16.4-fold in the ${Delta}$ crp ${Delta}$ vpsT mutant and3.0-fold in the ${Delta}$ crp ${Delta}$ vpsR mutant) and bap1 (8.1-fold in the ${Delta}$ crp ${Delta}$ vpsT mutant and 2.8-fold in the ${Delta}$ crp ${Delta}$ vpsR mutant) compared tothe wild type. Together, these findings indicate that cAMP-CRPnegatively regulates the transcription of vps genes and genesencoding matrix proteins in a different manner. While VpsR isthe most downstream positive transcriptional regulator of vpsgene expression, cAMP-CRP negatively regulates transcriptionof the genes encoding matrix proteins both through and independentof VpsR. Whether the action of cAMP-CRP is mediated by directbinding to the rbmC and bap1 promoter regions or indirectlythrough another regulatory protein(s) remains unknown.

Since we observed that VPS biosynthesis genes and genes encodingmatrix proteins are regulated differently by cAMP-CRP, we wantedto determine the contribution of VPS and biofilm matrix proteinsto biofilm formation in the ${Delta}$ crp genetic background. To thisend, we generated ${Delta}$ crp ${Delta}$ vps-I (vps-I cluster deletion), ${Delta}$ crp ${Delta}$ vps-I ${Delta}$ vps-II (vps-I and vps-II cluster deletion), ${Delta}$ crp ${Delta}$ rbmA, and ${Delta}$ crp ${Delta}$ bap1 mutants and compared their biofilm-forming capacities tothose of the wild type and ${Delta}$ crp, ${Delta}$ vps-I, ${Delta}$ vps-I ${Delta}$ vps-II, ${Delta}$ rbmA, and ${Delta}$ bap1 single mutants. Deletion of the vps-I cluster or the vps-Iand vps-II clusters in the ${Delta}$ crp genetic background eliminatedformation of a pellicle (biofilm formed at the air-liquid interface)(data not shown) and drastically reduced biofilm formation inboth 96-well microtiter plate and non-flow-cell systems (Fig.3D and E). Deletion of rbmA or bap1 in the ${Delta}$ crp genetic backgrounddecreased, but did not eliminate, biofilm formation. Comparedto the ${Delta}$ crp mutant biofilm, the biofilms formed by the ${Delta}$ crp ${Delta}$ rbmAand ${Delta}$ crp ${Delta}$ bap1 mutants were less structured, and there were fewermature pillars in the biofilms formed by the double-deletionmutants (Fig. 3E). Although the total biomasses and averagebiofilm thicknesses were not significantly different for the ${Delta}$ crp, ${Delta}$ crp ${Delta}$ rbmA, and ${Delta}$ crp ${Delta}$ bap1 mutants, the maximum biofilm thicknesswas greater for the ${Delta}$ crp mutant than for the ${Delta}$ crp ${Delta}$ rbmA and ${Delta}$ crp ${Delta}$ bap1 mutants (data not shown). Together, these results indicatethat under the experimental conditions that we utilized, VPSproduction is essential for biofilm formation in a ${Delta}$ crp geneticbackground, while biofilm matrix proteins have an accessoryrole. We do not know yet how biofilm matrix proteins function,whether they bind VPS carbohydrates or mediate cell-cell orcell-surface interactions. It is possible that the relativecontributions of VPS and biofilm matrix proteins to biofilmformation are different when the organisms are tested underdifferent environmental conditions using different surfaces.

cAMP-CRP differentially regulates the expression of a set of genes encoding proteins harboring GGDEF, EAL, and HD-GYP domains. Biofilm formation in V. cholerae is positively regulated byc-di-GMP (3, 31, 32, 56). Since cAMP-CRP negatively regulatesbiofilm formation in V. cholerae, we wanted to investigate ifthere is a connection between cAMP-CRP regulatory circuitryand c-di-GMP signaling in regulation of biofilm formation. Wehypothesized that c-di-GMP levels may be increased in ${Delta}$ cyaA and ${Delta}$ crp mutants due to increased expression of a gene(s) encodinga key DGC or due to decreased expression of a gene(s) encodinga key PDE. Indeed, whole-genome expression profiling of ${Delta}$ cyaAand ${Delta}$ crp mutants revealed that genes encoding proteins predictedto exhibit DGC and PDE activities were differentially regulatedin ${Delta}$ cyaA and/or ${Delta}$ crp mutants compared to the wild type (Table3).

In this set of differentially regulated genes, we focused on10 genes that encode proteins with a conserved GGDEF domainwhich is predicted to act as a DGC: VC0072, VC0653 (rocS), VC0658,VC1029 (cdgB), VC1067 (cdgH), VC1376, VC2750, VCA0074 (cdgA),VCA0217, and VCA0956 (cdgF) (referred to below as the GGDEFgenes). To examine the involvement of the GGDEF genes in theregulation of biofilm formation in V. cholerae, we generatedmutants with in-frame deletion mutations in these genes in thewild-type background, as well as in the ${Delta}$ crp genetic background.We then analyzed the biofilm-forming capacity of each of thesemutants using pellicle formation in glass culture tubes andbiofilm formation in 96-well microtiter plates with a crystalviolet staining assay. As shown in Fig. 4A and B, deletion ofGGDEF genes in the wild type did not alter the pellicle andbiofilm formation phenotypes. However, when the genes were deletedin the enhanced biofilm-forming ${Delta}$ crp genetic background, onlythe ${Delta}$ crp ${Delta}$ cdgA double-deletion mutant failed to form pelliclesand exhibited a reduced biofilm-forming capacity. This findingindicates that the increased pellicle- and biofilm-forming capacitiesof a ${Delta}$ crp mutant are due to increased expression of cdgA. Interestingly,we recently reported that cdgA positively regulates colony rugosityand biofilm formation in V. cholerae (2, 31). The crystal violetstaining assays also revealed that deletion of rocS and VC0658(designated cdgI [cyclic di-guanylate I]) in the ${Delta}$ crp geneticbackground further enhanced biofilm formation. Both RocS andCdgI contain conserved GGDEF-EAL domains. Such proteins canfunction as either DGCs or PDEs, and the enzymatic functionsare commonly regulated by environmental stimuli. Although theenzymatic activities of these proteins have not been determinedyet, based on biofilm-forming phenotypes, RocS and CdgI appearto function as a PDE under the experimental conditions thatwe utilized.

View larger version (44K):
[in this window]
[in a new window]

FIG. 4. Phenotypic characterization of GGDEF deletion mutants and GGDEF crp double-deletion mutants. (A) Pellicle formation, (B) quantitative comparison of biofilm formation, and (C) motility assays for the wild type, for ${Delta}$ crp and GGDEF single-deletion mutants, and for mutants with GGDEF deletions generated in the ${Delta}$ crp genetic background. The data are representative of two independent experiments. The error bars indicate standard deviations.

Flagellar motility is negatively regulated by c-di-GMP in V.cholerae. Therefore, we sought to understand the effect of deletionof GGDEF genes on motility. To this end, we measured the migrationzone formed by each single-deletion mutant, as well as the migrationzones formed by mutants generated in the ${Delta}$ crp genetic background,when the organisms were grown on LB soft agar plates (Fig. 4C).We also used ${Delta}$ flaA as a control. Of the GGDEF single-deletionmutants tested, the ${Delta}$ rocS mutant exhibited a decrease in motilityand the ${Delta}$ cdgH mutant exhibited an increase in motility comparedto the wild type. These results are consistent with our previousreport (31) and recent findings of Beyhan et al. (S. Beyhanet al., submitted for publication). Deletion of crp led to adecrease in motility, consistent with the expression profilingdata showing downregulation of genes involved in flagellar biosynthesisand chemotaxis in ${Delta}$ cyaA and ${Delta}$ crp mutants compared to the wildtype (see Fig. S1 in the supplemental material). Deletion ofcrp in individual GGDEF deletion mutants also led to furtherdecreases in motility compared to the corresponding GGDEF single-deletionmutants. One of the mutants tested, the ${Delta}$ crp ${Delta}$ rocS double-deletionmutant, exhibited a further reduction in motility compared toboth the ${Delta}$ crp and ${Delta}$ rocS single-deletion mutants, suggesting thatcAMP-CRP and RocS have additive effects on motility regulation.

In summary, for the 10 GGDEF genes tested, deletion of cdgA,rocS, and cdgI in the ${Delta}$ crp genetic background altered biofilm-formingphenotypes. Since we chose these genes based on their increasedexpression in the ${Delta}$ cyaA and/or ${Delta}$ crp mutant in whole-genome expressionprofiling experiments, we further confirmed their message abundanceusing qPCR. In agreement with microarray data, the message levelsof cdgA, rocS, and cdgI were greater in the ${Delta}$ crp mutant thanin the wild type (Fig. 5).

View larger version (13K):
[in this window]
[in a new window]

FIG. 5. qPCR analysis of cdgA, cdgI, and rocS message levels in the ${Delta}$ crp mutant: quantification of relative repression of (A) cdgA, (B) cdgI, and (C) rocS in the wild type and the ${Delta}$ crp mutant, normalized using recA. The results are from three independent biological replicates. The error bars indicate standard deviations. P values (two-tailed t test) are indicated at the top.

Expression of VC1934, which encodes a protein containing anEAL domain and a nonconserved GGDEF domain, expression of VC1086and VC1710, which encode proteins containing an EAL domain,and expression of VC1087, VC1348, VC2497, VCA0895, and VCA0931,which encode proteins containing a HD-GYP domain, were downregulated,with $≥$ 1.5-fold changes and FDR of $≤$ 3% in the ${Delta}$ cyaA and/or ${Delta}$ crp mutant(Table 3).

It is possible that a decrease in the expression of these genesencoding proteins predicted to have PDE activity (leading toa decrease in the cellular c-di-GMP level) (45, 49) may alsocontribute to an overall increase in the c-di-GMP level, thusgiving rise to the increased biofilm-forming capacities observedfor the ${Delta}$ cyaA and ${Delta}$ crp mutants. Regulation of the expression ofthese genes by cAMP-CRP in V. cholerae is currently under investigation.

CdgA is required for the enhanced biofilm-forming phenotype in the ${Delta}$ crp mutant and negative regulation of vpsL and rbmC expression by cAMP-CRP. In the experiments described above we showed that while the ${Delta}$ crp single-deletion mutant exhibits enhanced pellicle- and biofilm-formingcapacities (as determined by the crystal violet staining assay),the crp and cdgA double-deletion mutant ( ${Delta}$ crp ${Delta}$ cdgA mutant) exhibiteda decrease in biofilm formation. We confirmed this finding bycarrying out a CLSM analysis of biofilms formed by the wildtype and the ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA mutants in a non-flow-cellsystem (Fig. 6A). After 8 h of biofilm development, the ${Delta}$ crpmutant formed a thicker and more-structured biofilm, and thetotal biomass, average and maximum biofilm thicknesses, andsubstratum coverage were greater for the ${Delta}$ crp mutant than forthe wild type (Table 4). The ${Delta}$ cdgA mutant formed biofilms withreduced total biomass, average biofilm thickness, and substratumcoverage compared to the wild-type biofilms. Although deletionof cdgA in the ${Delta}$ crp mutant significantly reduced biofilm formation,the ${Delta}$ crp ${Delta}$ cdgA double mutant formed biofilms whose total biomass,average biofilm thickness, and substratum coverage were greaterthan those of the biofilms formed by the ${Delta}$ cdgA mutant (Fig. 6Aand Table 4). These results indicate that the negative regulationof biofilm formation by cAMP-CRP is largely mediated by CdgA,but additional factors may also contribute to an increase inbiofilm formation in the ${Delta}$ crp mutant.

View larger version (28K):
[in this window]
[in a new window]

FIG. 6. Phenotypic characterization of ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA mutants. (A) Biofilms of wild-type, ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA strains formed after 8 h of incubation at 30°C in a non-flow-cell system. Images were acquired by CLSM. The large images are images of the upper surfaces of biofilms, and the images below and to right of the large images are orthogonal views. Bars = 40 µm. (B and C) β-Galactosidase assays for the wild-type, ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA strains harboring (B) vpsL-lacZ and (C) rbmC-lacZ fusion constructs. The data are representative of two independent experiments. The error bars indicate standard deviations. (D) Pellicle formation in the wild-type, ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA strains harboring the vector or the cdgA complementation plasmid.

View this table:
[in this window]
[in a new window]

TABLE 4. COMSTAT analysis for biofilms of the wild-type, ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA strains^a

To determine if the altered pellicle- and biofilm-forming phenotypesof the ${Delta}$ crp ${Delta}$ cdgA double-deletion mutant were due to a reductionin the transcription of genes involved in biofilm matrix production,we carried out β-galactosidase assays with the wild-type, ${Delta}$ crp, ${Delta}$ cdgA, and ${Delta}$ crp ${Delta}$ cdgA strains harboring vpsL-lacZ and rbmC-lacZtranscriptional fusion constructs (Fig. 6B and C). While the ${Delta}$ crp mutant exhibited increased transcription of vpsL-lacZ (154-fold)and rbmC-lacZ (37-fold), the ${Delta}$ cdgA mutant exhibited decreasesin the transcription of vpsL-lacZ (1.8-fold) and rbmC-lacZ (1.3-fold)compared to the wild type, consistent with our previous finding(obtained using gene expression profiling) that CdgA positivelyregulates vps and rbm gene expression (2). The ${Delta}$ crp ${Delta}$ cdgA double-deletionmutant exhibited a decrease in the transcription of vpsL-lacZ(79-fold) and rbmC-lacZ (7.1-fold) compared to the ${Delta}$ crp single-deletionmutant. These results indicate that the decreased biofilm-formingcapacities of the ${Delta}$ crp ${Delta}$ cdgA double mutant compared to the ${Delta}$ crpmutant were due to decreased expression of genes involved inVPS biosynthesis and matrix protein production. Interestingly,in the ${Delta}$ crp ${Delta}$ cdgA double-deletion mutant the levels of transcriptionof vpsL-lacZ and rbmC-lacZ were higher than those in the wildtype (1.9-fold higher for vpsL-lacZ and 5.1-fold higher forrbmC-lacZ) and in the ${Delta}$ cdgA mutant (3.5-fold higher for vpsL-lacZand 6.5-fold higher for rbmC-lacZ). These findings suggest thatbesides CdgA, other factors and processes also contribute toan increase in biofilm formation in the ${Delta}$ crp mutant.

We also carried out a complementation assay by introducing cdgA,in which transcription is driven from its own promoter, in amulticopy number plasmid into the wild type and into the ${Delta}$ cdgA, ${Delta}$ crp, and ${Delta}$ crp ${Delta}$ cdgA mutants. The ${Delta}$ cdgA and ${Delta}$ crp ${Delta}$ cdgA mutants harboringthe complementation plasmid exhibited increased pellicle-formingcapacities compared to the same strains carrying only the vector(Fig. 6D), further confirming that CdgA is a key DGC that positivelyregulates biofilm formation. Although we now know many of themolecular players, it is still unclear how c-di-GMP regulatestranscription of the genes involved in biofilm matrix production.

DISCUSSION

Top

DISCUSSION

Environmental cues, such as nutrient availability, affect biofilmformation and/or dispersal (6, 9), linking carbon metabolismand biofilm formation. In Pseudomonas aeruginosa, the cataboliterepression control protein (Crc), which regulates carbon metabolism,positively regulates biofilm formation (40), and a sudden increasein carbon substrate availability (including glucose availability)induces biofilm dispersion (48). In addition, glucose repressesbiofilm formation in Bacillus subtilis (53), as well as in E.coli, Citrobacter freundii, Klebsiella pneumoniae, and Salmonellaenterica serovar Typhimurium (24). In contrast, V. choleraeexhibits an increase in biofilm-forming capacity when it isgrown in minimal medium supplemented with 0.5% (wt/vol) glucose(23). The cAMP-CRP global transcriptional regulatory complex,which regulates carbon metabolism, positively regulates biofilmformation in E. coli, in which a crp mutation results in decreasedbiofilm formation (24), and in Shewanella oneidensis, in whicha crp mutation results in a defect in biofilm detachment inthe stop-of-flow-induced detachment response compared to thewild type (55). In contrast, a V. cholerae ${Delta}$ crp deletion mutantexhibits increased biofilm formation (29), indicating that cAMP-CRPcan act as either a positive regulator or a negative regulatorof biofilm formation. We were, therefore, interested in furtherelucidating the molecular mechanism by which cAMP-CRP regulatesbiofilm formation in V. cholerae.

Using whole-genome transcriptional profiles of ${Delta}$ cyaA and ${Delta}$ crpmutants, we showed that cAMP-CRP regulates the expression ofa large set of genes in V. cholerae (see Fig. S1 and TablesS1 and S2 in the supplemental material), consistent with itsrole as a global transcriptional regulatory complex (5, 12).It is also noteworthy that the overall transcriptional profilesof the ${Delta}$ cyaA and ${Delta}$ crp mutants were similar (see Fig. S1 in thesupplemental material), consistent with the requirement forboth cAMP and CRP in transcriptional regulation (5). More than20% of the predicted genes in the genome of V. cholerae (asannotated by The Institute for Genomic Research) are differentiallyregulated in both ${Delta}$ cyaA and ${Delta}$ crp mutants ( $≥$ 2-fold change and FDRof $≤$ 1%) compared to the wild type, and approximately equal numbersof genes are positively and negatively regulated. Consistentwith the negative role of cAMP-CRP in biofilm formation, manyof the genes involved in biofilm matrix production are upregulatedin ${Delta}$ cyaA and ${Delta}$ crp mutants (Table 3). Several genes involved inpathogenesis are also upregulated, while genes involved in flagellumbiosynthesis and chemotaxis are downregulated in the deletionmutants (see Fig. S1 and Tables S1 and S2 in the supplementalmaterial).

V. cholerae wild-type biofilm formation requires the productionof both VPS and matrix proteins, including RbmA, RbmC, and Bap1(15, 16). These matrix proteins play a crucial role in maintainingthe structural integrity of the biofilm. They may function asan agglutinin/adhesion in binding cells together and/or bindingto VPS. They may also act as anchors for biofilm and/or cellsto attach to surfaces that contain carbohydrates. As a consequenceof environmental signals, V. cholerae may modulate the ratioof matrix proteins to VPS within a biofilm, giving rise to differentdegrees of biofilm rigidity and stability, which may promotedetachment of single cells or cell aggregates for dispersal.It is therefore intriguing but not surprising that, althoughVpsR is the most downstream regulator of vps gene expression,cAMP-CRP can also regulate rbmC and bap1 expression independentof VpsR regulation (Fig. 3).

Putative cAMP-CRP binding sites were predicted upstream of therbmA, rbmC, and bap1 coding regions. We are currently investigatingif cAMP-CRP binds directly to these predicted binding sites.It is very intriguing that a predicted VpsR binding site (GTCTCATTACTGAGGCGT)overlaps by 11 bp the putative cAMP-CRP binding site (AACTTTGAGATGTCTCATTACT)upstream of rbmC. A very plausible hypothesis is that cAMP-CRPnegatively regulates rbmC expression by directly binding toits promoter region and, in doing so, interferes with the bindingof VpsR to the upstream regulatory region of rbmC and thus preventsVpsR from positively regulating rbmC expression. Similar antagonisticfunctions have been reported for cAMP-CRP and AphA/AphB fortcpPH expression (28), as well as for HapR and VpsR for aphAexpression (33). We are currently examining the possible antagonisticrole of cAMP-CRP and VpsR in binding the rbmC promoter region.Using Virtual Footprint software (http://www.prodoric.de/vfp/index.php),a putative cAMP-CRP binding site was also predicted upstreamof the coding region of vpsR, but not upstream of the codingregion of vpsT. It is therefore possible that cAMP-CRP negativelyregulates vpsR expression by direct binding to the upstreampromoter region of vpsR. Since no cAMP-CRP binding sites upstreamof vpsT have been predicted, the cAMP-CRP negative regulationof vpsT may be via VpsR (in addition to HapR) or by a so-far-unidentifiedrepressor protein(s). We are currently investigating if cAMP-CRPbinds directly upstream of the vpsR promoter region.

A glucose effect or catabolite repression has been observedin many microorganisms (5) and is a likely environmental nutritionalcue for regulating biofilm formation. It has been reported thatvpsL expression is regulated by one of the components of thephosphoenolpyruvate phosphotransferase (PTS) system, such thatthe accumulation of the phosphorylated form of enzyme I (EI $~$ P)due to the absence of a PTS sugar, such as glucose, repressesvps gene expression and leads to a reduction in biofilm formation(23). Furthermore, addition of cAMP to the growth medium, mimickingactivation of adenylate cyclase (CyaA) by EIIA^Glu $~$ P (12), repressesbiofilm accumulation in V. cholerae (23). It is noteworthy thataddition of exogenous cAMP eliminated the enhanced pellicle-formingcapacity of the ${Delta}$ cyaA mutant (see Fig. S2 in the supplementalmaterial). Consistent with this model, ${Delta}$ cyaA and ${Delta}$ crp mutantsboth exhibit increased biofilm-forming capacities. We also observedincreased biofilm formation in wild-type V. cholerae grown inLB medium supplemented with 0.2 and 0.5% (wt/vol) glucose (datanot shown). Interestingly, mannose (another PTS sugar) has alsobeen reported to enhance biofilm formation in V. cholerae, likelythrough the PTS system (23, 27, 39). Similarly, growth on N-acetylglucosamine(also a PTS sugar) has been shown to promote attachment of V.cholerae and subsequent colonization of chitinous surfaces (37,42). V. cholerae can colonize surfaces of phytoplankton andzooplankton (14, 34). Nutrients provided by mucous secretionsof phytoplankton and zooplankton, as well as chitinous surfacesof zooplankton, could then facilitate or enhance biofilm formationby V. cholerae by providing environmental stimuli that leadto enhanced biofilm formation, thereby facilitating environmentalpersistence and growth of the pathogen.

The V. cholerae genome contains 62 genes coding for proteinscontaining GGDEF, EAL, and HD-GYP domains (17, 20). These proteinsmodulate cellular c-di-GMP levels (8, 45, 46, 49), which inturn regulate several phenotypic characteristics, includingcolony morphology, pellicle- and biofilm-forming capacities,and motility (3, 31, 32, 56, 57). c-di-GMP also regulates similarbiological processes in several other microorganisms, includingP. aeruginosa, E. coli, and S. enterica serovar Typhimurium(25, 43, 44). Using whole-genome expression profiles of ${Delta}$ cyaAand ${Delta}$ crp mutants, we identified 10 genes encoding proteins thatcontain a conserved GGDEF domain and are upregulated in the ${Delta}$ cyaA and/or ${Delta}$ crp mutant compared to the wild type (Table 3).For these genes, only deletion of cdgA in the ${Delta}$ crp genetic backgroundeliminated pellicle formation and reduced the biofilm-formingcapacity (Fig. 4 and 6), indicating that the increased biofilm-formingcapacity of the ${Delta}$ crp mutant was due in part to increased expressionof cdgA, which in turn modulated the expression of vps genesand genes encoding matrix proteins (Fig. 6B and C). Interestingly,we also observed increases in biofilm formation for the ${Delta}$ crp ${Delta}$ rocS and ${Delta}$ crp ${Delta}$ cdgI mutants compared to the ${Delta}$ crp, ${Delta}$ rocS, and ${Delta}$ cdgIsingle-deletion mutants (Fig. 4B). Although both RocS and CdgIcontain conserved GGDEF-EAL domains, the results of biofilmformation assays suggest that, like RocS, CdgI functions mainlyas a PDE under the conditions that we utilized. The contributionsof other PDEs (Table 3) to the negative regulation of biofilmformation by cAMP-CRP remain to be investigated.

Regulation of biofilm formation in V. cholerae is complex. Wepreviously described the regulatory network controlling vpsgene expression, where vps genes are positively regulated byVpsR, VpsT, and CdgA and negatively regulated by HapR (2). HapRalso negatively regulates the expression of vpsT, vpsR, andcdgA. Interestingly, a recent study reported that HapR bindsdirectly to the promoter regions of vpsT and cdgA (59). We alsopreviously described the positive regulation of cdgA by VpsRand VpsT. Using Virtual Footprint software, two putative cAMP-CRPbinding sites were predicted upstream of the cdgA coding region,suggesting that cAMP-CRP can regulate cdgA expression both directlyby binding to the cdgA promoter region and indirectly throughHapR.

Data obtained in this study, together with results describedin other reports, indicate that cAMP-CRP regulates biofilm formationat multiple levels (Fig. 7). First, cAMP-CRP negatively regulatesbiofilm formation in V. cholerae through positive regulationof hapR expression, which in turn negatively regulates expressionof vps and rbm (and bap1) genes, as well as cdgA, vpsT, and,in our strain, vpsR. VpsR, VpsT, and CdgA, in turn, positivelyregulate expression of vps and rbm (as well as bap1) genes.Second, cAMP-CRP may also negatively regulate expression ofvpsR, cdgA, and rbmC expression by directly binding to theirpromoter regions. The connection between cAMP-CRP regulatorycircuitry and c-di-GMP signaling reported here is particularlyintriguing. Further characterization of this connection shouldprovide additional insight into the wiring of the regulatorycircuitry controlling biofilm formation in V. cholerae.

View larger version (17K):
[in this window]
[in a new window]

FIG. 7. Model of cAMP-CRP regulation of biofilm formation in V. cholerae. cAMP-CRP regulates biofilm formation at multiple levels. Expression of vps and rbm genes is negatively regulated by cAMP-CRP through positive regulation of hapR expression and through negative regulation of vpsR and vpsT expression. In turn, VpsR and VpsT positively regulate vps and rbm gene expression, while HapR negatively regulates vps and rbm gene expression. VpsR, VpsT, and HapR regulation of vps and rbm gene expression also involves c-di-GMP signaling, where cdgA expression is negatively regulated by HapR and positively regulated by VpsR and VpsT. An increase in cdgA transcription leads to an increase in the c-di-GMP level, which in turn could interact with an effector protein(s) to positively regulate biofilm formation. We have a very limited understanding of the link between c-di-GMP pools and the signaling that leads to biofilm formation in V. cholerae. In addition, cAMP-CRP may also directly regulate vpsR, cdgA, and rbmC expression and indirectly regulate vpsT expression.

ACKNOWLEDGMENTS

We thank Chad Saltikov, Manel Camps, and members of the Yildizlaboratory for their valuable comments on the manuscript, aswell as Sinem Beyhan for generating FY_Vc_956, Nicholas Shikumafor generating FY_Vc_3411, and James Meir for generating FY_Vc_152,FY_Vc_869, FY_Vc_154, and FY_Vc_158.

This work was supported by grant AI055987 from NIH.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Environmental Toxicology, University of California, Santa Cruz, Santa Cruz, CA 95064. Phone: (831) 459-1588. Fax: (831) 459-3524. E-mail: yildiz{at}etox.ucsc.edu

Published ahead of print on 15 August 2008.

Supplemental material for this article may be found at http://jb.asm.org/.

REFERENCES

Top