JOURNAL OF BACTERIOLOGY, Nov. 2009, p. 6555–6570 Vol. 191, No. 210021-9193/09/$12.00 doi:10.1128/JB.00949-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

The Vibrio cholerae Flagellar Regulatory Hierarchy ControlsExpression of Virulence Factors�†

Khalid Ali Syed,1 Sinem Beyhan,2 Nidia Correa,1 Jessica Queen,3 Jirong Liu,1 Fen Peng,1Karla J. F. Satchell,3 Fitnat Yildiz,2 and Karl E. Klose1*

South Texas Center for Emerging Infectious Diseases and Department of Biology, The University of Texas at San Antonio,San Antonio, Texas1; Department of Microbiology and Environmental Toxicology, University of California, Santa Cruz,

Santa Cruz, California2; and Department of Microbiology-Immunology, Northwestern University Medical School,Evanston, Illinois3

Received 20 July 2009/Accepted 24 August 2009

Vibrio cholerae is a motile bacterium responsible for the disease cholera, and motility has been hypothesizedto be inversely regulated with virulence. We examined the transcription profiles of V. cholerae strains contain-ing mutations in flagellar regulatory genes (rpoN, flrA, flrC, and fliA) by utilizing whole-genome microarrays.Results revealed that flagellar transcription is organized into a four-tiered hierarchy. Additionally, genes withproven or putative roles in virulence (e.g., ctx, tcp, hemolysin, and type VI secretion genes) were upregulatedin flagellar regulatory mutants, which was confirmed by quantitative reverse transcription-PCR. Flagellarregulatory mutants exhibit increased hemolysis of human erythrocytes, which was due to increased transcrip-tion of the thermolabile hemolysin (tlh). The flagellar regulatory system positively regulates transcription of adiguanylate cyclase, CdgD, which in turn regulates transcription of a novel hemagglutinin (frhA) that mediatesadherence to chitin and epithelial cells and enhances biofilm formation and intestinal colonization in infantmice. Our results demonstrate that the flagellar regulatory system modulates the expression of nonflagellargenes, with induction of an adhesin that facilitates colonization within the intestine and repression of virulencefactors maximally induced following colonization. These results suggest that the flagellar regulatory hierarchyfacilitates correct spatiotemporal expression patterns for optimal V. cholerae colonization and diseaseprogression.

Vibrio cholerae causes the human diarrheal disease cholera.The bacteria are natural inhabitants of aquatic environmentsand are introduced into the human population through theingestion of contaminated food or water. Within the humanhost, V. cholerae expresses virulence factors that facilitate col-onization of the intestine (e.g., toxin-coregulated pilus [TCP])and that stimulate dramatic fluid loss from host tissues (chol-era toxin [CT]) (5, 61). A regulatory cascade consisting of anumber of different proteins, including ToxR, TcpP, and ToxT,induces the coordinated expression of CT and TCP maximallywithin the intestine and under specific in vitro growth condi-tions (for a review, see reference 7).

V. cholerae is a highly motile organism by virtue of its singlepolar flagellum. Flagellar genes are transcribed in a four-tieredtranscriptional hierarchy (51). The single class I gene productFlrA activates �54-dependent transcription of class II genes,which encode components of the MS ring-switch-export appa-ratus as well as the two-component system FlrBC (31). Phos-phorylated FlrC activates �54-dependent transcription of classIII genes, which encode the basal body-hook and the flagellinFlaA (10, 11). Finally, the antisigma factor FlgM is secretedthrough the basal body-hook to allow �28-dependent transcrip-

tion of class IV genes, which encode four additional flagellinsand some of the motor components (9, 30).

Motility has been linked to the virulence of V. cholerae.Spontaneous nonmotile mutants are defective for fluid accu-mulation and adherence in the rabbit ileal loop model (53, 58)as well as defective for adherence to isolated rabbit brushborders (15). O1 El Tor biotype nonmotile mutant strainsgenerally show colonization defects in the infant mouse model(35), while nonmotile O1 classical mutants can generally col-onize similarly to the motile wild-type strain (16). Nonmotilemutants of live V. cholerae vaccine strains induce reduced dis-ease symptoms, or reactogenicity, in human volunteers whilestill being able to colonize the intestine (12, 27). Nonmotilemutants have also been shown to express increased levels ofvirulence factors (CT and TCP), while “hypermotile” mutants(defined by greater swarm diameter in soft agar assays) pro-duce less CT and TCP than wild-type V. cholerae (16). Theauthors proposed that virulence and motility are inversely reg-ulated in V. cholerae, with motility being downregulated andvirulence factor expression simultaneously upregulated whenthe organisms are colonized on the intestinal cell surface. Sincethe majority of these studies were performed with undefinedmotility mutants, the exact connection between flagellar syn-thesis and cholera pathogenesis has remained unclear.

Recently, Silva et al. provided evidence for this model whenthey found that transcription of toxT, ctxA, and tcpA is upregu-lated in a V. cholerae nonmotile (motY) strain (58). Furtherevidence for this model comes from Ghosh et al., who foundthat the histone-like nucleoid structuring protein H-NS stim-

* Corresponding author. Mailing address: Dept. of Biology, Univer-sity of Texas at San Antonio, One UTSA Circle, San Antonio, TX78249. Phone: (210) 458-6140. Fax: (210) 458-4468. E-mail: Karl.Klose@utsa.edu.

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

� Published ahead of print on 28 August 2009.

6555

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ulates motility by stimulating flrA expression while repressingctxAB and tcpA transcription (18). Though a number of studieshave implicated motility as being important for V. choleraevirulence, the contribution of the flagellar regulatory hierarchyto the virulence of V. cholerae has remained obscure. In thecurrent study, we have characterized the transcriptome offlagellar regulatory mutants and found that many nonflagellargenes are also regulated by this hierarchy. Moreover, transcrip-tion of a large number of proven and putative virulence genesis upregulated in the flagellar regulatory mutants, revealing apreviously unappreciated coordinate regulation of disparatefactors that enhance V. cholerae pathogenesis. These studiesfacilitated the identification of a novel RTX-like hemaggluti-nin that is positively regulated by the flagellar regulatory hier-archy through a diguanylate cyclase (DGC) and that mediatesbinding to epithelial cells and enhances intestinal colonization.

MATERIALS AND METHODS

Medium. Luria broth was used for both liquid growth medium and agar plates,supplemented with ampicillin (50 �g/ml), streptomycin (100 �g/ml), chloram-phenicol (2 �g/ml) or kanamycin (35 �g/ml) when appropriate. Medium wassupplemented with 2 mM glutamine for growth of rpoN strains. Motility platesconsisted of LB with 0.3% agar. LB agar without NaCl was supplemented with10% sucrose for counterselection when plasmids containing the sacB gene wereutilized. LB agar with 2 mg/ml of streptomycin was used for counterselectionwhen plasmids containing the rpsL gene were utilized. Medium was supple-mented with 80 �g/ml DL-�,ε-diaminopimelic acid (Sigma) for the growth of theEscherichia coli dapA strain.

Plasmid construction. A complete list of plasmids and oligonucleotide primersused in this study can be found in Table S2 in the supplemental material. Oligonu-cleotide primer pairs for the promoter regions of the frhA, frhC, cdgD, cdgE, fliK,VC1384, VC1252, and sdhC genes are listed in Table S2 in the supplementalmaterial. Primers were designed to amplify a fragment of approximately 300 bp,based upon the genome sequence of V. cholerae N16961 (24), with restriction sitesadded to facilitate cloning. Promoter fragments were PCR amplified from 0395genomic DNA, digested with appropriate restriction enzymes, and ligated into plas-mid pRS551 (59), which had been digested similarly to form pKEK790 (fliKp),pKEK791 (cdgEp), pKEK797 (cdgDp), pKEK823 (VC1384p), pKEK837(VC1252p), pKEK838 (frhCp), pKEK849 (frhAp), and pKEK1105 (sdhCp).

Splicing-by-overlap-extension (SOE) PCR (26) was used to generate the de-letion mutations in tlh, cdgD, and frhC. Upstream and downstream gene frag-ments were PCR amplified in the first reaction by using primers 1 and 2 andprimers 3 and 4, respectively, as listed in Table S2 in the supplemental material.The second PCR was performed with primers 1 and 4, with the two fragmentsgenerated in the first PCRs used as a template. The resulting SOE fragmentswere digested with the appropriate restriction enzymes and ligated into pKAS32(60) to form pKEK844 (�tlh), pKEK806 (�cdgD), and pKEK863 (�frhC). De-letion/insertion mutations of the frhA, cdgE, VC1384, and cdgD genes wereconstructed by SOE PCR with three fragments: an upstream and a downstreamfragment generated with primers 1 and 2 and primers 3 and 4, respectively, asdescribed above, and a fragment containing the chloremphenicol resistance gene(cat). This technique has been described previously (38). The resulting SOEproducts were digested with the appropriate restriction enzymes and ligated intopKAS32 (60) to form pKEK928 (�frhA::cat), pKEK959 (�cdgE::cat), pKEK929(�VC1384::cat), and pKEK836 (�cdgD::cat). Complementation plasmids forcdgD and VC1384 were constructed by PCR amplification of these genes byutilizing the primers listed and then cloning into pBAD24 (20) and pWSK30(67), respectively, to form pKEK822 (cdgD) and pKEK1233 (VC1384).

Bacterial strains. E. coli strain DH5� (21) was used for cloning manipulations.E. coli strains SM10�pir (42) and WM3046 (dapA) (a gift from William Metcalf,University of Illinois) were used to transfer plasmids to V. cholerae by conjuga-tion. The V. cholerae strains used in this study are listed in Table S2 in thesupplemental material. All V. cholerae strains are isogenic with O1 classical strainO395, with the exception of those used in the supplemental biofilm assay, whichare isogenic to O1 El Tor strain A1552, and those used in the adult mousecolonization model, which are isogenic to O1 El Tor strain P27459.

V. cholerae mutant strains were constructed by conjugating plasmids contain-ing the gene deletion and/or insertion (described above) into V. cholerae strain

KKV598 (O395 �lacZ), followed by counterselection on streptomycin as de-scribed previously (60). The gbpA V. cholerae strain was constructed usingp-28VCA0811 (a kind gift from G. Schoolnik) (40). The correct construction ofall strains was verified by PCR and sequencing. The �frhA::cat and �cdgD::catmutations were subsequently moved by CPT1ts transduction (23) into other O1V. cholerae strains.

Whole-genome transcriptome profiling. V. cholerae cultures of wild-type andisogenic rpoN, flrA, flrC, and fliA mutant strains were grown overnight in LBsupplemented with 2 mM glutamine at 37°C with aeration. Cultures were diluted1:50 in fresh medium and grown to an optical density at 600 nm (OD600) of 0.3to 0.4. Two-milliliter aliquots of cells were harvested by centrifugation for 2 minat room temperature. The cell pellets were immediately resuspended in 1 ml ofTRIzol reagent (Invitrogen) and stored at �80°C. Total cellular RNA wassubsequently isolated in accordance with the manufacturer’s instructions (In-vitrogen) and as described previously (72). The microarrays used in this study arecomposed of 70-mer oligonucleotides representing the open reading framespresent in the V. cholerae genome and were printed at the University of Cali-fornia, Santa Cruz. Whole-genome expression analysis was performed using acommon reference RNA, which was isolated from wild-type cells grown in LBsupplemented with 2 mM glutamine as described above. Total RNA from testand reference samples was used in cDNA synthesis and microarray hybridization,and scanning was performed as described previously (3). Normalized signalratios were obtained with LOWESS print-tip normalization using the Biocon-ductor packages (17) in the R environment. Differentially regulated genes weredetermined using three biological and two technical replicates with the signifi-cance-analysis-of-microarrays program (66), using �1.5-fold differences in geneexpression and a �3% false discovery rate as a cutoff value.

ß-Galactosidase assay. V. cholerae strains were transformed with the promot-er-lacZ fusion plasmids described above, grown in LB plus ampicillin to anOD600 of �0.2 to 0.4. Bacterial cells were permeabilized with chloroform andsodium dodecyl sulfate and assayed for ß-galactosidase activity in accordancewith the method of Miller (41). All experiments were performed at least threeseparate times.

Quantitative reverse transcription-PCR. Total RNA was isolated from V.cholerae strains by using Charge Switch total RNA (Invitrogen), and the isolatedRNA was then treated with Turbo DNase (Ambion). Reverse transcriptionreactions were performed using a high-capacity cDNA reverse transcription kit(Applied Biosystems); reverse transcriptase-negative controls were included forevery RNA sample. cDNA was amplified with the Applied Biosystems ABI 7500real-time PCR system, using SYBR green PCR master mix (Applied Biosys-tems), and specific primers designed by Primer Express and listed in Table S2 inthe supplemental material. The relative expression values (R) were calculated byusing the formula R � 2 �(�Ct target � Ct endogenous control), where Ct is thefractional threshold cycle. The constitutively expressed gyrA cDNA was used asthe endogenous control (37).

Hemagglutination and hemolysis assays. The hemagglutination and hemolysisassays were performed as described by Gardel and Mekalanos (16), with thefollowing alterations. V. cholerae strains were grown to an OD600 of �0.6 to 0.8at 37°C, and then bacterial cells were pelleted, washed twice, and resuspended inKRT buffer at a concentration of 1010 CFU/ml. Bacterial cells were seriallydiluted in a round-bottomed 96-well microtiter plate. Human type “O” red bloodcells were harvested by centrifugation, washed twice, and resuspended in KRTbuffer at a 2% concentration. Red blood cells (0.2 ml) were added to 0.1 mlbacteria, the plate was incubated at room temperature for 30 to 60 min, and thehemagglutination titer was recorded. The titer is the reciprocal of the greatestdilution at which hemagglutination occurred. The hemolysis assay was performedin exactly the same manner as the hemagglutination assay; the plates wereincubated for 20 h, at which time the OD540 in the supernatants was measured.

HEp-2 cell binding assay. The HEp-2 cell binding assay was performed asdescribed previously (16), with minor changes. Briefly, HEp-2 cells were grownin a 24-well plate on glass coverslips in Dulbecco’s modified Eagle’s medium(Gibco) with 10%FBS. V. cholerae strains were grown to an OD600 of �0.2 to 0.4at 37°C. Bacterial cells were then pelleted, washed twice, resuspended in phos-phate-buffered saline (PBS), and added to a monolayer of HEp-2 cells at amultiplicity of infection of 50:1. After incubation for 45 min at 30°C, the mediumwas aspirated, and the wells were washed four times with PBS. The cells werethen fixed by methanol for 5 min and stained with Giemsa (1:12.5) for 25 min.Cells were mounted on a glass slide in FluorSave (Calbiochem) and imaged witha Zeiss Axiovert 200 fluorescence microscope. Twenty HEp-2 cells were countedfor each bacterial strain, and experiment was performed three separate times,with similar results.

Chitin bead binding assay. V. cholerae strains were grown to an OD600 of �0.6to 0.8 at 37°C. Bacterial cells were pelleted, washed twice, and resuspended in

6556 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

KRT buffer. A MagnaRack (Bio-Rad) was used for magnetic bead manipula-tions. Magnetic chitin beads (Bio-Rad) were washed three times with KRTbuffer, and then 0.1 ml of chitin bead suspension was mixed with 0.1 ml ofbacteria (109 CFU/ml). After incubation for 30 min at room temperature, su-pernatant was removed and beads were washed three times with KRT buffer.Bound bacterial cells were eluted by vortexing the chitin beads with glass beads(Sigma). Serial dilutions of the input bacteria and the adherent bacteria wereplated to quantitate the input and output, respectively. The adhesion factor wasdetermined by dividing the number of output bacteria by the number of inputbacteria.

Biofilm assay. The biofilm assays were performed as described previously (34),with the following modifications. V. cholerae strains were grown overnight at37°C in LB and then normalized to identical densities based on OD600. Sevenmicroliters of bacterial culture was inoculated into 0.75 ml of LB in 10 mlunscarred borosilicate glass tubes and then incubated statically at 37°C for 21 h.Tubes were rinsed with distilled water and incubated with 1 ml 0.1% crystal violetfor 30 min and then rinsed with distilled water and air dried. The stained biofilmwas dissolved in 1.5 ml of dimethyl sulfoxide by vortexing and then quantitatedby measuring OD570.

Phenotype microarrays. Phenotype microarrays were performed using PM1and PM2 carbon source phenotype microarray 96-well plates (Biolog) in accor-dance with the manufacturer’s protocol, with some modifications. V. choleraestrains were grown in LB at 37°C to an OD600 of �0.6 to 0.8. Cells wereharvested by centrifugation and washed twice with IF-Oa (Biolog) and thenresuspended in 20 ml of IF-Oa (105CFU/ml), redox dye (Biolog) was added at1/10 of the recommended concentration (1 �l/ml), and then the mixture wasinoculated into PM plates (0.1 ml/well) and incubated statically at 30°C. Sampleswere quantitated by measuring OD510.

Mouse intestinal colonization assays. All animals were handled in strict ac-cordance with good animal practice as defined by the relevant national and/orlocal animal welfare bodies, and all animal work was approved by the UTSAIACUC.

The competition assay for intestinal colonization in 5-day-old CD-1 sucklingmice was performed as described previously (16). The inocula consisted of �105

CFU for both wild-type and mutant strains. The competitive indices were cal-culated by dividing the ratio of mutant to wild-type bacteria in the output by theratio of mutant to wild-type bacteria in the input.

For adult mouse colonization studies, 5- to 6-week-old female C57BL/6 mice(Harlan, Indianapolis, IN) were anesthetized, and then 50 �l of 8.5% (wt/vol)NaHCO3 was administered intragastrically, followed immediately by treatmentwith 50 �l of bacterial suspension in PBS, using a 22-gauge feeding needle. At 24or 48 h postinoculation, mice were euthanized, and small intestines were re-moved, weighed, and homogenized in PBS. The homogenates were seriallydiluted in PBS and plated on LB streptomycin agar plates. The detection limitwas 100 CFU in the small intestine.

RESULTS

Transcriptome profiling of flagellar regulatory mutants. Todetermine which V. cholerae genes are controlled by the flagellartranscription hierarchy, we compared the gene expression profileof the wild-type strain to those of the isogenic rpoN, flrA, flrC, andfliA mutant strains by using a V. cholerae whole-genome microar-ray. Total RNA was prepared from exponentially grown (OD600,0.3 to 0.4) V. cholerae rpoN, flrA, flrC, and fliA mutant strains aswell as the wild-type parent strain; growth was performed at 37°Cin LB supplemented with 2 mM glutamine to overcome theknown glutamine requirement of the rpoN strain (31). TheseRNAs were labeled with Cy3 or Cy5 and applied to a DNAglass-spotted microarray of the V. cholerae genome, and differen-tially expressed genes were identified by statistical analysis offluorescence intensity values. Data were analyzed by the signifi-cance-analysis-of-microarrays program (66), utilizing a �3%false-positive discovery rate and �1.5-fold transcript abundancedifferences. This analysis revealed a surprising number of genesregulated by the flagellar hierarchy.

The alternate sigma factor �54 is required for V. choleraeflagellar transcription (31), but on the basis of the presence of

multiple �54-dependent transcriptional activators in V. cholerae(32), it controls additional phenotypes as well, including nitro-gen assimilation, dicarboxylic acid transport, and quorum sens-ing (31, 36). The rpoN mutant showed decreased expression of228 genes and increased expression of 271 genes (Fig. 1; seealso Table S1 in the supplemental material). Among the genespositively regulated by �54 were several expected genes, such asthose encoding 50 flagellar/chemotaxis components, glu-tamine synthetase, phage shock proteins, and dicarboxylic acidtransport proteins, as well as genes encoding, e.g., formatedehydrogenase and purine biosynthesis components, alongwith 77 genes annotated as hypothetical. Among the genesexpressed at higher levels in the rpoN mutant than in thewild-type strain were genes involved in biosynthesis of aminoacids, cofactors, polysaccharides, the cell wall, fatty acids, pro-tein degradation, and secretion and 116 genes annotated ashypothetical. Of particular note were a number of upregulatedgenes with proven or putative roles in pathogenesis, includingthe ctx and tcp genes (discussed below).

FlrA is the �54-dependent activator of class II flagellargenes, and since it lies at the top of the flagellar hierarchy, it isconsidered the “master regulator” of flagellar gene transcrip-tion (51). The flrA mutant showed decreased expression of 152genes and increased expression of 124 genes (Fig. 1; see alsoTable S1 in the supplemental material). Among the genesdownregulated in the flrA mutant were 50 genes annotated asflagellar/chemotaxis genes, consistent with the known functionof FlrA, and 47 genes annotated as hypothetical. FlrC is the�54-dependent activator of class III flagellar genes, and itsactivity is regulated by phosphorylation, which presumably oc-curs at the class II-class III checkpoint (11). The flrC mutantshowed decreased expression of 115 genes and increased ex-pression of 92 genes (Fig. 1; see also Table S1 in the supple-mental material). Among the genes downregulated in the flrCmutant were 50 genes annotated as flagellar/chemotaxis genes,consistent with the known function of FlrC, and 46 genesannotated as hypothetical. FliA (�28) is the sigma factor re-sponsible for transcription of class IV flagellar genes, and itsactivity is regulated by the antisigma factor FlgM, whose se-cretion presumably occurs at the class III-class IV checkpoint(9). The fliA mutant showed decreased expression of 140 genesand increased expression of 195 genes (Fig. 1; see also Table S1

FIG. 1. Genes showing reduced (A) or increased (B) transcription inV. cholerae flrA, flrC, and fliA mutant strains. Venn diagrams show thenumber of genes exhibiting transcriptional repression or induction foreach respective mutant relative to the wild-type level. Numbers in theoverlapping regions represent genes that were coregulated, and the pat-tern of class II, class III, and class IV flagellar genes is shown in panel A,while the virulence genes discussed in text are shown in panel B.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6557

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

in the supplemental material). Among the genes downregu-lated in the fliA mutant were 17 genes annotated as flagellar/chemotaxis genes, consistent with the known function of FliA,and 37 genes annotated as hypothetical. Overlap of expressionpatterns for both downregulated genes as well as upregulatedgenes in the flrA, flrC, and fliA mutants can be seen in Fig. 1;the complete list of differentially expressed genes in the fourflagellar regulatory mutant strains (including rpoN) is providedin Table S1 in the supplemental material.

Refinement of the flagellar transcription hierarchy throughtranscriptome profiling. Cluster analysis of flagellar gene ex-pression in the rpoN, flrA, flrC, and fliA strains revealed thatthe V. cholerae flagellar transcription hierarchy is organizedsimilarly to the four-tiered hierarchy previously postulated(51). The requirement of flagellar regulatory genes for classII, class III, and class IV flagellar genes is illustrated in Fig.1A, and selected genes downregulated in the flagellar reg-ulatory mutants are shown in Fig. 2A. For example, in re-gion I, flgBCDEFGHIJ (VC2200 to VC2192), flgKL (VC2191and VC2190), flaA (VC2189 and VC2188), flgOP (VC2207 andVC2206), and flgT (VC2208) have the RpoN-, FlrA-, and FlrC-dependent expression patterns predicted for class III flagellargenes; flgMN (VC2204 and VC2205) and flaC (VC2187) havethe RpoN-, FlrA-, FlrC-, and FliA-dependent expression pat-terns predicted for class IV flagellar genes; and flgA (VC2203)expression is independent of the flagellar regulatory genes, aspreviously predicted (51). Previous analysis of the cheV-3 andcheR-2 (VC2202 and VC2201) genes in this region suggestedthat their expression was also independent of the flagellarregulatory genes, as determined by transcription analysis of thecheV-3 promoter (51). However, the current microarray ex-pression data suggest a class IV expression pattern for cheV-3and a class III pattern for cheR-2.

For region II, flrBC (VC2136 and VC2135) and fliEFGHIJ(VC2134 to VC2129) have the RpoN- and FlrA-dependentexpression patterns predicted for class II flagellar genes, whileflaE (VC2144), flaD (VC2143), and flaB (VC2142) have theRpoN-, FlrA-, FlrC-, and FliA-dependent expression patternspredicted for class IV flagellar genes, and flrA (VC2137) ex-pression was independent of the flagellar regulatory genes (itshowed some dependence on RpoN), as predicted previously(51). This microarray analysis also revealed an expression pat-tern for fliKLMNOPQ (VC2128 to VC2122) predicted for classIII flagellar genes (i.e., dependence on FlrC) rather than classII genes as had been predicted previously, suggesting the pres-ence of an FlrC-dependent promoter upstream of fliK. The fliKpromoter region was fused to lacZ and assayed for transcrip-tion in the flagellar regulatory mutants (Table 1), which con-firmed the presence of a class III promoter (RpoN, FlrA, andFlrC dependent) upstream of fliK. Thus, fliKLMNOPQ shouldbe considered class III rather than class II genes. The expres-sion of fliR (VC2121) appeared independent of the flagellarregulatory genes, while the expression of flhB (VC2120) wassimilar to that of class III genes; further studies of the expres-sion of these genes should reveal their correct placementwithin the flagellar transcription hierarchy. Finally, flaG-fliD-flaI-fliS (VC2141 to VC2138) were previously classified as classIII genes on the basis of the presence of a �54-dependentpromoter upstream of flaG that is dependent on rpoN, flrA,and flrC, but not fliA, for transcription. However, the microar-

ray analysis revealed FliA dependence for flaG expression andFlrC independence for fliD-flaI-fliS expression, suggesting thatexpression of this region is complicated and that these genesmay not in fact be class III genes.

In region III, the flhA operon (VC2069 to VC2058), whichcontains fliA, along with a number of important chemotaxisgenes, has the RpoN- and FlrA-dependent expression patternspredicted for class II flagellar genes, as predicted previously(51). Region IV, which contains only the motAB operon(VC0892 and VC0893), exhibited the RpoN-, FlrA-, FlrC-, andFliA-dependent expression predicted for class IV genes. Mi-croarray analysis of the motY gene (VC1008) revealed theRpoN-, FlrA-, and FlrC-dependent typical of a class III gene,while motX (VC2601) expression was dependent on RpoN,FlrA, FlrC, and FliA, as seen with other class IV genes. Ourprevious analyses of these genes had characterized motY as aclass IV gene and motX as a class III gene on the basis ofpromoter expression studies. Because the microarray studieshere contradicted these earlier studies, we reanalyzed the pre-vious promoter constructs and discovered that the promoterconstruct annotated as motXp-lacZ actually contained motYp-lacZ and that the promoter construct annotated as motYp-lacZactually contained motXp-lacZ. Due to this mistake, the motXand motY genes were incorrectly classified, and the currentstudy correctly classifies motY as a class III gene and motX asa class IV gene.

Seven methyl-accepting chemoreceptor protein genes that dis-played expression patterns similar to those of class IV genes (i.e.,RpoN, FlrA, FlrC, and FliA dependent) were identified: VC1898,VC2439, VCA0176, VCA0658, VCA0773, VCA0923, andVCA0974. Also, an additional cheV gene (VCA0954) that dis-played a class IV pattern of expression, similar to cheV-3(VC2202) in region I, was identified; we suggest that this gene beannotated as cheV-4. Finally, VC1384, which is annotated as ahypothetical gene that has homology to OmpA and thus likelyencodes an outer membrane protein, displayed a class III patternof expression (i.e., dependent on RpoN, FlrA, and FlrC but notFliA). The VC1384 promoter was further analyzed by promoter-lacZ fusion assays with the V. cholerae flagellar regulatory mu-tants (Table 1), which confirmed this promoter to be a class IIIpromoter. We constructed a �VC1384 V. cholerae mutant andanalyzed this strain for motility (see Fig. S1 in the supplementalmaterial). The �VC1384 strain demonstrated reduced motilitycompared to that of the wild-type strain, and expression ofVC1384 from a plasmid in this strain restored wild-type motility.Moreover, expression of VC1384 from a plasmid in the wild-typestrain led to a greater-than-wild-type level of motility, suggestingthat this gene plays some role in motility, so we propose reclas-sifying this gene as a flagellar/motility gene.

We examined the expression patterns of several additionalgenes downregulated in the flagellar mutants by constructingpromoter-lacZ fusions and measuring -galactosidase activityin the V. cholerae flagellar regulatory mutants (Table 1). Thepromoter-lacZ results indicated a class IV pattern of expres-sion (i.e., dependent on RpoN, FlrA, FlrC, and FliA) forVC1252, a homologue of competence/damage-inducible CinAprotein (PF02464), which contains a probable molybdopterinbinding domain (PF00994). Promoter-lacZ transcription mea-surements also indicated a class IV pattern of expression forthe sdhC promoter, which drives transcription of the succinate

6558 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

dehydrogenase operon (sdhCDAB). These results suggest thatthe flagellar regulatory hierarchy controls some aspects ofcompetence and the tricarboxylic acid cycle/electron transportchain. Promoter-lacZ fusion analysis also indicated that the

flagellar hierarchy controls transcription of the frhA, frhC,cdgD, and cdgE promoters (discussed below).

The microarray data have allowed us to refine the flagellartranscription hierarchy of V. cholerae by categorizing individual

FIG. 2. Cluster analysis of select genes with reduced or increased transcription in flagellar regulatory mutants. (A) Patterns of expression ofselect genes with reduced expression in flagellar regulatory mutants. The intensity of yellow or blue represents the extent of gene induction orrepression, respectively. Shown is the relative expression level of each given gene in the (left to right) flrA, rpoN, flrC, and fliA mutants. The flagellargene clusters I, II, III, and IV are as described previously (51), and the names/functions of individual genes are denoted. wt, wild type. (B) Clusteranalysis of genes with increased expression in all flagellar regulatory mutants. The intensity of yellow represents the extent of gene induction.Shown is the relative expression level of each given gene in the (left to right) flrA, rpoN, flrC, and fliA mutants. The names/functions of individualgenes are denoted. (C) Refined flagellar transcription hierarchy (described in the text).

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6559

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

genes to specific flagellar expression classes based on the ex-pression profiles described above (Fig. 2C).

Flagellar regulatory mutants have increased transcriptionof diverse virulence genes. As mentioned above, the rpoN, flrA,flrC, and fliA mutant strains showed increased transcription ofa large number of genes, including a number of genes withproven or putative roles in pathogenesis (Fig. 1). Cluster anal-ysis of selected upregulated genes is shown in Fig. 2B. Withinthe Vibrio pathogenicity island, tagD (VC0824), tcpQCRD-STEF (VC0830 to VC0837), and toxT (VC0838) were upregu-lated in the flagellar mutants. The tcpA gene, which encodesthe TCP structural subunit, lies at the beginning of the tcpoperon but differs sufficiently from the probe on the microar-ray such that no data were available for this gene; upregulatedtranscription of the tcpA gene in flagellar mutants was subse-quently confirmed by quantitative PCR (see below). Within theCTX prophage, ctxA and ctxB (VC1456 and VC1457) were alsoupregulated in all flagellar mutants. Among the additionalputative virulence genes that were upregulated in the flagellarmutants were vieA and vieB (VC1651 and VC1652), responseregulators involved in modulation of cyclic di-GMP (c-di-GMP) levels and regulation of CT expression (64, 65); epsL(VC2725), required for CT secretion (1); hlyA and hlyB(VCA0219 and VCA0220), encoding the hemolysins that causevacuolation and pore formation in many cell types (2); a geneencoding a thermolabile hemolysin (tlh; VCA0218) that hasphospholipase activity (14); an additional hemolysin gene (hlx;VCA0594) (45); the gene encoding the chitin-binding proteinGbpA (VCA0811), which mediates binding to epithelial cellsas well as chitinous surfaces (28); a putative lipase gene(VC1418); and three genes within the type VI secretion system(T6SS) (VCA0107, VCA0108, and VCA0124) involved intranslocating proteins associated with virulence across thegram-negative membranes (52).

Additional virulence genes were identified by microarrayanalysis as upregulated in some but not all of the flagellarmutants, including CT secretion genes (epsF in rpoN, flrA, andflrC; epsG in rpoN and flrA; epsM in flrC and fliA; and epsN inflrC), additional genes in the Vibrio pathogenicity island (orf3in fliA and rpoN, tcpJ in rpoN and fliA, acfC in fliA, acfA in rpoNand fliA, and acfD in rpoN and flrC); additional genes in theT6SS cluster (VCA0109 in rpoN, fliA, and flrA; vasA in fliA andflrA; VCA0112 in rpoN, fliA, and flrA; VCA0113 in rpoN andflrC; and vasH in rpoN); and genes encoding the RTX toxinsubunit RtxA (VC1451 in fliA), a hemagglutinin (VCA0446 inflrA and flrC), and VieS (VC1653 in rpoN, flrA, and flrC).

The results revealed a general trend of widespread upregu-lation of proven and putative virulence factors. The upregula-tion of several virulence factors in nonmotile V. cholerae had

previously been noted by utilizing phenotypic assays (16), andthe current study reveals that the extent of this upregulationoccurs at the transcriptional level and includes many genes thathave not previously been shown to be linked by common sig-naling pathways. To confirm the upregulation of selected vir-ulence genes, we performed quantitative real-time PCR onRNA isolated from wild-type and fliA V. cholerae strains, quan-titating the message for ctxA, tcpA, hlyA, tagD, gbpA, tlh, andthe T6SS gene VCA0108 (Fig. 3A). The results for three bio-logical replicates, with three technical replicates each, revealedincreased transcription of all seven genes in the fliA mutantversus those in the wild-type strain, ranging from approxi-mately twofold for tcpA to approximately eightfold for hlyA.These results confirmed those obtained with the microarray

TABLE 1. Transcription of promoter-lacZ fusions in V. cholerae flagellar regulatory mutant strains

Promoter fusedto lacZ

Mean activity � SD (Miller units)

Wild type rpoN flrA flrC fliA

fliKp 1,081 � 2 190 � 1 257 � 2 310 � 3 1,264 � 6VC1384p 11,861 � 810 119 � 2 704 � 21 342 � 20 10,051 � 343cdgEp 3,069 � 92 714 � 9 890 � 5 894 � 36 779 � 3VC1252p 15,030 � 883 855 � 12 490 � 8 1,793 � 122 533 � 28sdhCp 14,092 � 10 886 � 46 648 � 6 1,697 � 77 447 � 7

FIG. 3. (A) Increased transcription of select virulence genes in thefliA V. cholerae strain. Strains O395 (WT) and KKV1113 (fliA) weregrown in LB, and mRNA abundance for the specific genes noted wasdetermined by quantitative reverse transcription-PCR (see Materialsand Methods). Results shown represent three biological replicatesconsisting of three technical replicates for each strain. (B) Utilizationof select carbon sources by the rpoN and fliA V. cholerae strains. StrainsO395 (WT), KKV56 (rpoN), and KKV1113 (fliA) were assayed forgrowth on various carbon sources by Biolog PM1 and PM2 phenotypemicroarrays (see Materials and Methods). Results shown representtwo separate experiments for each strain.

6560 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

analysis and revealed the upregulated transcription of multiplevirulence genes in flagellar regulatory mutants.

Phenotype microarrays indicate that several metabolicpathways are regulated by the flagellar hierarchy. Phenotypemicroarrays (Biolog) assaying for growth on 190 different car-bon sources were performed with the wild-type and rpoN andfliA mutant V. cholerae strains to determine if the metaboliccapabilities of flagellar regulatory mutants differed from thoseof the wild-type strain. Although phenotype microarray resultsare often reported as �/�, we have reported them here by ODreading (OD510), which correlates with metabolic activity onthe specific carbon sources. RpoN (�54) is a global regulatorthat controls the expression of a large number of genes outsidethe flagellar hierarchy (Fig. 1; see also Table S1 in the supple-mental material) (31), and it has been shown previously that aV. cholerae rpoN mutant is defective for growth on succinate(32). This is likely due to �54-dependent activation of genesinvolved in dicarboxylic acid transport (54), and thus, it was notsurprising that the rpoN mutant exhibited defects in growth onvarious forms of succinate, malate, and fumarate and that thefliA mutant grew similarly to the wild type (Fig. 3B). The rpoNmutant also grew better on propionate than either the wildtype or the fliA mutant, indicating that this metabolic capabilityis regulated by �54 but not associated with the flagellar hier-archy. In contrast, both the rpoN and the fliA mutants grewbetter than the wild type on L-glutamine, -methyl-D-glu-coside, L-histidine, gelatin, laminarin, and mannan as carbonsources (Fig. 3B), indicating that the flagellar regulatory hier-archy normally represses these metabolic capabilities.

Growth of flagellar regulatory mutants on gelatin agar re-vealed enhanced expression of a gelatinase activity in thesemutants in comparison to that in the wild-type strain (see Fig.S2 in the supplemental material), which may explain the en-hanced growth of these mutants with gelatin utilized as a car-bon source. Phenotype microarrays also demonstrated that therpoN and fliA mutants grew worse than the wild type withTween 20 utilized as a carbon source. These results confirmthat the flagellar regulatory hierarchy controls metabolic capa-bilities of the cell in addition to flagellar synthesis.

Increased hemolysis by flagellar regulatory mutants is dueto increased expression of the thermolabile hemolysin. It hadbeen documented previously that spontaneous nonmotile V.cholerae mutants exhibit increased hemolysis of human type Oerythrocytes (16), but the exact reason for this behavior has notbeen characterized. The rpoN, flrA, flrC, and fliA flagellar reg-ulatory mutants also demonstrate increased hemolysis of hu-man type O erythrocytes (11) (Fig. 4 and data not shown). Wereasoned that transcription of the hemolysin responsible forthis phenotype was likely upregulated in the flagellar regula-tory mutants. Three genes annotated as hemolysins are up-regulated in the flagellar mutants: VCA0218, VCA0219, andVCA0594 (Fig. 2B; see also Table S1 in the supplementalmaterial). VCA0219 encodes the “El Tor” hemolysin HlyA,and the O395 strain used in these studies contains a deletion inhlyA that renders this strain HlyA� (19); thus, HlyA cannot bethe flagellum-regulated hemolysin. VCA0594 encodes a hemo-lysin, Hlx, which was determined to lyse sheep but not humanerythrocytes (45), suggesting that this was also not the flagel-lum-regulated hemolysin. VCA0218 encodes a thermolabile

hemolysin, Tlh, which has phospholipase and lecithinase activ-ity (14).

We constructed a V. cholerae strain with a deletion in the tlhgene (�tlh) (see Materials and Methods) and measured humanO erythrocyte hemolysis by V. cholerae wild-type and fliA, tlh,and fliA tlh mutant strains (Fig. 4). The fliA mutant exhibitedhigh levels of hemolysis in comparison to the wild-type strain,confirming that the hemolysin is negatively regulated by theflagellar regulatory hierarchy. Importantly, the hemolytic ac-tivity of the fliA strain is abrogated when tlh is inactivated,demonstrating that the flagellum-regulated hemolysin is thethermolabile hemolysin Tlh. Because the transcription of tlh isincreased in the flagellar regulatory mutants (Fig. 3A), we canconclude that the flagellar regulatory hierarchy represses thetranscription of tlh. In an infant mouse small intestine coloni-zation assay (see Fig. 6), the tlh V. cholerae strain colonizedonly slightly worse than the wild-type parent (competitive in-dex, 0.67; P � 0.029). This indicates that thermolabile hemo-lysin does not play a critical role in colonization of the intes-tine, at least in this particular animal model.

The decrease in hemagglutination associated with flagellarregulatory mutants is due to decreased expression of a novelRTX-like hemagglutinin, flagellum-regulated hemagglutinin.It had been documented previously that spontaneous nonmo-tile V. cholerae mutants also exhibit decreased hemagglutina-tion of human type O erythrocytes (16), but the exact reasonfor this behavior has not been characterized. We have ob-served agglutinated erythrocytes under the microscope, and V.cholerae bacteria can be seen bound to the surfaces of theerythrocytes, indicating that an adhesin on the bacterial sur-face acts as the hemagglutinin (data not shown). We reasonedthat transcription of the hemagglutinin responsible for thisphenotype was likely downregulated in the flagellar regulatorymutants. A gene annotated as encoding an “agglutination pro-tein,” VC1621, is downregulated in the flagellar mutants, as isa hypothetical gene, VC1620, immediately adjacent and diver-gently transcribed (Fig. 5A; see also Table S1 in the supple-mental material). The VC1620 and VC1621 promoters weremeasured for transcription by fusing the VC1620-VC1621 in-tergenic region in both orientations into a lacZ transcriptionalfusion vector and then assaying for beta-galactosidase activityin the flagellar regulatory mutants (Fig. 5B). Transcription ofboth the VC1620 and the VC1621 promoters is downregulated

FIG. 4. Thermolabile hemolysin is the flagellum-regulated hemo-lysin. V. cholerae strains O395 (WT), KKV1113 (fliA), KKV2077 (tlh),and KKV2078 (fliA tlh) were assayed for hemolytic activity with humantype O erythrocytes as described in Materials and Methods.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6561

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

FIG. 5. The flagellum-regulated hemagglutinin FrhA mediates binding to erythrocytes, epithelial cells, and chitin and enhances biofilmformation. (A) V. cholerae frhA locus. (B) Flagellar regulatory proteins and CdgD positively regulate transcription of the frhA and frhC promoters.V. cholerae strains O395 (WT), KKV56 (rpoN), KKV98 (flrC), KKV1113 (fliA), and KKV1966 (cdgD) carrying plasmids pKEK849 (frhAp-lacZ) andpKEK838 (frhCp-lacZ) were measured for -galactosidase activity. (C) The cdgD promoter requires the flagellar regulatory proteins for tran-scription. V. cholerae strains O395 (WT), KKV56 (rpoN), KKV59 (flrA), KKV98 (flrC), and KKV1113 (fliA) carrying plasmid pKEK797(cdgDp-lacZ) were measured for -galactosidase activity. (D) FrhA and CdgD enhance biofilm formation. (E) FrhA and CdgD are required forflagellum-regulated hemagglutination of human O erythrocytes. (F) FrhA and CdgD enhance adhesion to Hep-2 cells. (G) FrhA and CdgDenhance V. cholerae chitin binding. For panels D to G, biofilm formation, hemagglutination, Hep-2 binding, and chitin binding, respectively, for

6562 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

in the flagellar mutants, confirming that the flagellar regulatoryhierarchy controls transcription of these genes.

VC1620 is predicted to encode a large, 2,251-amino-acid(aa) protein that contains an RTX-like repeat region at the Cterminus with a signature type I secretion motif (Fig. 5A). Dueto the functional properties of this protein (see below), wehave renamed it the flagellum-regulated hemagglutinin A(frhA) protein. The region between aa 484 and 995 showshomology to cadherin domains (cd00031), including two iden-tical 114-aa sequences from aa 768 to 881 and aa 882 to 995.This 114-aa sequence is repeated as many as six times in thehomologous protein in other clinical O1/O139 V. choleraestrains (e.g., 2740-80, MAK757, and MO10) (see alignment inFig. S3 in the supplemental material), suggesting an importantfunction for this motif within the human host. The 114-aasequence shares 36% identity and 46% similarity with humancadherin 18 and similar levels of identity/similarity with othereukaryotic cadherins (e.g., insect, fish, and rodent cadherins).Cadherins are normally glycoproteins involved in calcium-de-pendent cell-cell adhesion of eukaryotic cells; thus, a role forthis protein in adhesion to eukaryotic cells might be antici-pated. The cadherin domains are typically repeated multipletimes within the protein, providing a “zipper” mechanism forstrong interactions between cadherin proteins on adjoiningcells (57). FrhA also contains a region at the C terminus (aa1514 to 2251) that shares homology with the C-terminal por-tions of RTX toxins and related Ca2�-binding proteins(COG2931); this includes the type I secretion signal that di-rects the protein through this secretion pathway.

Type I secretion substrates require additional proteins forsecretion (for a review, see reference 25). The best-studiedtype I secretion substrate, HlyA of E. coli, requires the innermembrane protein HlyB, the periplasm-spanning proteinHlyD, and the outer membrane protein TolC for secretionacross the inner and outer membranes (62). Type I secretionproteins are typically encoded by the same operon as the se-cretion substrate. FrhA is encoded by a single gene operon, butthe divergently transcribed operon, carrying VC1621 andVC1622, is also regulated by the flagellar regulatory hierarchy(Table 1), as mentioned above. VC1621 is predicted to encodea type I secretion component that shares homology with TolCand facilitates movement of the substrate across the outermembrane. Because we have shown that VC1621 also contrib-utes to flagellum-regulated hemagglutination (see below), wehave renamed this gene frhC. VC1622 shares homology withOmpA family proteins (pfam00691) and is predicted to encodean outer membrane porin. Downstream of frhA lies an operoncontaining VC1618 and VC1619 (Fig. 5A). VC1618 is pre-dicted to encode a major facilitator superfamily transmem-brane protein that facilitates the movement of a substrate(s)across the cytoplasmic membrane; it shares some homology to

HlyB (see Fig. S4 in the supplemental material). We suspectthat this gene may represent the type I secretion componentinvolved in translocating FrhA across the inner membrane buthave not yet analyzed its transcription or its function. However,the E. coli cytoplasmic membrane component HlyB also con-tains a nucleotide binding domain at its C terminus withATPase activity important for energizing the secretion process(56), but the homology between VC1618 and HlyB is weak inthis region, and VC1618 lacks the Walker A and B motifs andC and D loops important for nucleotide binding and ATPaseactivity (see Fig. S4 in the supplemental material). Thus,VC1618 likely has no ATPase activity, and no other obviouscandidates with homology to nucleotide binding proteins/ATPases could be found near frhA. VC1619 does not showobvious homology to any known type I secretion componentsbut encodes a member of pfam07209, a family of proteins ofunknown function.

We constructed V. cholerae strains containing mutations infrhA and frhC and analyzed these strains for their ability toagglutinate human type O erythrocytes. The frhA V. choleraestrain failed to agglutinate erythrocytes, similar to a fliA strain(Fig. 5E). Attempts to clone frhA into a plasmid for comple-mentation purposes were unsuccessful, due to the large andrepetitive nature of this gene; however, polar effects in thissingle gene operon are not anticipated, and we have recon-structed this strain several times with an identical resultantphenotype; thus, we are confident that frhA encodes the fla-gellum-regulated hemagglutinin. The frhC V. cholerae strainalso failed to agglutinate erythrocytes, similar to a fliA strain(see Fig. S5 in the supplemental material), demonstrating theinvolvement of FrhC in flagellum-regulated hemagglutinationas well. The frhA and frhC strains exhibited motility patternssimilar to those of the wild-type strain in soft agar, suggestingthat these genes are not involved in motility/chemotaxis (datanot shown).

Flagellum-regulated hemagglutinin mediates binding to ep-ithelial cells and chitin and increased biofilm formation. SinceFrhA mediates binding to and agglutination of erythrocytes,we wished to determine if FrhA could also mediate binding toepithelial cells. Gardel and Mekalanos (16) observed thatspontaneous nonmotile V. cholerae bacteria were defective atbinding to the human epithelial Hep-2 cell line, but the exactreason for this behavior has not been characterized. The wild-type V. cholerae strain binds to Hep-2 cells, but the frhA mutantstrain shows a 10-fold defect in Hep-2 cell binding, similar toa fliA strain (Fig. 5F). The frhC mutant strain was also defec-tive for binding to Hep-2 cells (data not shown), consistentwith the idea that flagellum-regulated hemagglutinin mediatesadhesion to human epithelial cells. To ensure that the involve-ment of FrhA in epithelial cell binding is not restricted to theparent V. cholerae strain used in these studies, we also intro-

the V. cholerae strains O395 (WT), KKV1113 (fliA), KKV2092 (cdgD), and KKV2093 (frhA) carrying plasmid pBAD24 (“�”) or pKEK822(“pcdgD”) were measured as described in Materials and Methods. All assays were performed in triplicate, and error bars are shown for panels B,C, D, F, and G; panel E shows representative results for three separate experiments. For panels B, C, F, and G, P was 0.001, and for panel D,P was 0.01 for comparison of the mutant to the wild type or for comparison of the mutant strain complemented with cdgDp to the parent mutantstrain by Student’s two-tailed t test, with the exception of the frhA strain complemented with cdgDp, which was not significantly different from thefrhA strain without cdgDp.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6563

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

duced the frhA mutation into an O1 El Tor V. cholerae strain,P27459, and found that this frhA mutant also showed a greater-than-sevenfold defect for epithelial cell binding in comparisonto its parent wild-type strain (see Fig. S6 in the supplementalmaterial).

GlcNAc residues on the surfaces of epithelial cells and chiti-nous surfaces are bound by a V. cholerae chitin-binding pro-tein, GbpA (28). We wondered whether FrhA could also me-diate binding to chitin. Utilizing chitin beads, we found that thefrhA mutant showed a greater-than-sixfold defect in binding tochitin in comparison to the wild-type strain, and a similarbinding defect was seen in the fliA mutant (Fig. 5G). Thepreviously characterized chitin-binding protein GbpA is en-coded by VCA0811, which is upregulated in the flagellar reg-ulatory mutants, including the fliA mutant (Fig. 2B and 3A; seealso Table S1 in the supplemental material). The fliA mutantwas defective for binding chitin despite upregulated GbpAexpression, which suggested that FrhA plays a more importantrole in chitin binding in this assay. To confirm this, we con-structed gbpA and gbpA fliA V. cholerae strains and measuredthese for chitin binding in the same assay (see Fig. S7 in thesupplemental material). In our hands, the gbpA and gbpA fliAmutants did not behave differently from their respective iso-genic wild-type and fliA parent strains in this assay, suggestingthat, at least under the conditions tested, FrhA plays a moredominant role in chitin binding than GbpA.

Biofilm development by V. cholerae is a complex processinvolving multiple genes, and flagellar synthesis and motilitycontribute to the ability of V. cholerae to form biofilms (68, 72).Nonmotile V. cholerae strains are known to be defective duringthe early stages of biofilm formation involving attachment toabiotic surfaces (69). We determined whether FrhA contrib-utes to V. cholerae biofilm formation by measuring biofilmsformed at the air-liquid interface in tubes of liquid cultures.The parent wild-type strain, O395, is known to form only weakbiofilms under these conditions, yet the frhA mutant showed agreater-than-twofold defect in biofilm formation in compari-son to the wild-type strain, and this defect was similar to thatseen in the fliA mutant (Fig. 5D). We also introduced the frhAmutation into an O1 El Tor V. cholerae strain, A1552, which isknown to make robust biofilms (72), and in this background,the frhA mutant also showed an approximate threefold defectin biofilm formation in comparison to the A1552 parent strain(see Fig. S8 in the supplemental material). In both strain back-grounds, prolonged incubation led to similar levels of biofilmformation, suggesting that the defect was in the initial stages ofbiofilm formation. These results demonstrate that FrhA con-tributes to biofilm formation of V. cholerae.

The flagellar regulatory hierarchy controls expression of acyclic diguanylate synthase, CdgD, which regulates transcrip-tion of flagellum-regulated hemagglutinin. The second mes-senger c-di-GMP modulates complex behaviors of V. cholerae,and its presence positively regulates biofilm formation andnegatively regulates virulence (37, 63, 64). A large number ofproteins with DGC (also known as GGDEF) and c-di-GMPphosphodiesterase (also known as EAL) domains are presentin V. cholerae; thus, the roles of specific c-di-GMP-modulatoryproteins are often obscured by functional redundancy. Thetranscriptome profiling results (Fig. 2A) indicated that tran-scription of two DGC genes is downregulated in the flagellar

regulatory mutants VCA0697 (cdgD) and VC1367 (cdgE).Transcription of both of these DGC genes has previously beenobserved to be downregulated in rugose variants of V. cholerae(37). In this study, Lim et al. found that a cdgD mutant exhib-ited enhanced motility and diminished early biofilm formationbut that a cdgE mutant had no obvious phenotype.

We measured transcription of cdgDp-lacZ and cdgEp-lacZfusions in the wild-type and rpoN, flrA, flrC, and fliA mutantstrains (Fig. 5C and Table 1). These results confirmed thatcdgD and cdgE transcription is downregulated in all the flagel-lar regulatory mutants. We constructed cdgD and cdgE mutantV. cholerae strains and found that, unlike the fliA mutant strain(Fig. 3A), the cdgD and cdgE strains were unaffected for tran-scription of ctx, tcp, gbpA, hlyA, tlh, or T6SS genes (data notshown). Consistent with the unaffected tlh transcription, thecdgD and cdgE mutants showed low hemolytic activity againsthuman erythrocytes, similar to the wild-type strain (data notshown). Moreover, the cdgD strain did not have the enhancedgelatinase activity characteristic of the flagellar regulatory mu-tants (see Fig. S2 in the supplemental material). These dataindicate that suppression of the transcription of these variousvirulence genes by the flagellar regulatory hierarchy is notmediated by flagellar regulation of these DGC proteins.

However, the cdgD mutant was defective for hemagglutina-tion of human erythrocytes, similar to the fliA and frhA mu-tants (Fig. 5E). When CdgD was expressed from a plasmid inthe cdgD strain, hemagglutination returned to wild-type levels.Importantly, when CdgD was expressed from a plasmid in thefliA strain, hemagglutination also returned to wild-type levels,indicating that the failure of the flagellar regulatory mutant toagglutinate red blood cells is due to its reduced expression ofCdgD. Finally, when CdgD was expressed from a plasmid inthe frhA strain, there was no restoration of hemagglutination,confirming that FrhA is the hemagglutinin responsible for theFliA- and CdgD-dependent hemagglutination. Likewise, whenCdgD was expressed from a plasmid in the frhC strain, therewas no restoration of hemagglutination (see Fig. S5 in thesupplemental material). Interestingly, when CdgD was ex-pressed from a plasmid in the wild-type strain, it increasedhemagglutination activity almost threefold. In contrast, thecdgE mutant showed no alteration in hemagglutination of hu-man erythrocytes, and expression of CdgE from a plasmid hadno effect on hemagglutination by any of these strains (data notshown). These results suggested that the flagellar regulatoryhierarchy controlled FrhA expression by controlling CdgD ex-pression. In order to confirm this, we measured frhAp-lacZ andfrhCp-lacZ transcription in the cdgD V. cholerae strain (Fig.5B). As suspected, the cdgD mutant strain showed the low levelof frhA and frhC transcription present in the flagellar regula-tory mutant strains, demonstrating that the flagellar transcrip-tion hierarchy controls flagellum-regulated hemagglutinin ex-pression via CdgD regulation of frh transcription.

We measured the cdgD mutant for the other phenotypesassociated with FrhA, including epithelial cell binding, chitinbinding, and biofilm formation. The cdgD mutant was defectivefor epithelial cell binding (Fig. 5F) and chitin binding (Fig.5G), similar to the fliA and frhA mutant strains. Expression ofCdgD in the cdgD and fliA mutant strains restored Hep-2 cellbinding and chitin binding, demonstrating that the flagellarregulatory hierarchy controls these activities by controlling ex-

6564 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

pression of CdgD. As expected, expression of CdgD in the frhAmutant was unable to restore epithelial cell binding and chitinbinding, consistent with FrhA mediating these functions. ThecdgD mutant was also defective for biofilm formation (Fig.5D), although it formed biofilms more poorly than the fliA andfrhA mutants. Expression of CdgD in the cdgD and fliA mutantstrains restored wild-type levels of biofilm formation, but ex-pression of CdgD in the frhA mutant did not enhance biofilmformation. These results are consistent with the flagellar reg-ulatory hierarchy controlling flagellum-regulated hemaggluti-nin-dependent enhanced biofilm formation via the expressionof CdgD.

FrhA enhances V. cholerae intestinal colonization. SinceFrhA enhances binding of V. cholerae to epithelial cells, wedetermined whether FrhA also enhances colonization of theintestine. We examined the ability of the frhA mutant to col-onize the infant mouse intestine, the most widely used animalmodel for V. cholerae (29). In this model, the ability of themutant to colonize the intestine in the presence of the isogenicwild-type strain is examined at 24 h postinoculation. We founda significant colonization defect of the frhA mutant under theseconditions, approximately twofold (competitive index, 0.44;P � 0.003) (Fig. 6). The cdgD mutant showed a slightly greatercolonization defect in this assay, approximately threefold(competitive index, 0.31; P � 0.0001) (Fig. 6). These resultsdemonstrate that FrhA and CdgD contribute to intestinal col-onization of the infant mouse by V. cholerae.

We also measured intestinal colonization in the adult mouseintestinal colonization model, a recently developed model forlonger-term colonization studies (46, 47). The O395 back-ground V. cholerae strain used for the studies mentioned abovedoes not colonize the intestines of adult mice. We thereforeconstructed frhA and cdgD mutant strains in the P27459 V.cholerae background; the colonization behavior of this strain inthe adult mouse intestine has been extensively characterized(47). Colonization was monitored at 24 and 48 h postinocula-tion. In this model, the frhA mutant showed a trend toward adefect in colonization, although the differences in colonizationbetween this mutant and the wild-type strain did not reach

statistical significance (see Fig. S9 in the supplemental mate-rial). For example, at 24 h, 5 of 15 mice inoculated with thefrhA strain had no detectable CFU within the intestine, incontrast to the observation that only 1 of 16 mice inoculatedwith the wild-type strain had no detectable CFU within theintestine at this time point. At 48 h, the average number ofintestinal CFU recovered for the frhA strain was more than10-fold lower than the average number of intestinal CFU re-covered for the wild-type strain. These results suggest thatFrhA may contribute to intestinal colonization in the adultmouse model, although more animals would be needed toobtain statistical significance. In contrast, the cdgD mutantcolonized similarly to the wild-type strain at 24 h and 48 h,suggesting that CdgD does not noticeably affect intestinal col-onization in this model and/or this V. cholerae strain back-ground. Our results provide the first demonstration that anadhesin under the control of the flagellar regulatory hierarchyis required for enhanced intestinal colonization.

DISCUSSION

The effect of motility on V. cholerae pathogenesis has beennoted by several different laboratories, but the exact connec-tion between flagellar synthesis and virulence has been elusive.This has been partly due to an incomplete understanding offlagellar synthesis, since many of the earlier studies were per-formed with spontaneous nonmotile V. cholerae strains prior tothe elucidation of the flagellar regulatory hierarchy. Usingdefined flagellar regulatory mutants, we have reexamined theeffect of the flagellar regulatory cascade upon V. cholerae vir-ulence, utilizing whole-genome transcriptome analyses. Ourresults are consistent with the previous report by Gardel andMekalanos (16) that suggested that motility and the expressionof specific virulence factors are inversely regulated.

Transcriptome studies demonstrate that a large number ofV. cholerae genes are both positively and negatively regulatedby the flagellar regulatory hierarchy. This was not unexpectedfor the alternate sigma factor �54, which regulates flagellarsynthesis, but which also controls the expression of a number ofother genes through the action of multiple �54-dependent reg-ulatory proteins (31, 32). However, previously, the only knownfunction of the �54-dependent regulators FlrA and FlrC was inthe stimulation of flagellar and chemotaxis gene transcription.The alternate sigma factor �28, the activator of class IV flagel-lar genes, was recently shown to modulate virulence factorexpression via repression of hapR expression, a component ofthe quorum-sensing circuit (39); however, this circuit is notoperative in the O395 strain used in the present studies, due toinactivation of the hapR gene. Our transcriptome analyses re-vealed that many genes with no known roles in flagellar syn-thesis or chemotaxis are downregulated in the absence of theflagellar regulators. Even more striking is the upregulation ofa variety of proven and putative virulence genes in the absenceof the flagellar regulators, demonstrating a previously un-known coordination in the transcriptional regulation of variousunlinked factors that enhance V. cholerae virulence (discussedbelow).

The flagellar regulatory hierarchy regulates flagellar as wellas other genes. Our laboratory had previously described thefour-tiered flagellar transcription hierarchy, based upon anal-

FIG. 6. V. cholerae frhA and cdgD strains are defective for coloni-zation of the infant mouse intestine. V. cholerae strains KKV2093(frhA), KKV1966 (cdgD), and KKV2077 (tlh) were coinoculated withthe wild-type strain O395 perorally into infant mice at a ratio of �1:1,intestinal homogenates were recovered at 24 h postinoculation, andnumbers of CFU of wild-type and mutant strains were determined.The competitive index is given as the ratio of mutant/wild-type bacteriain the output divided by the ratio of mutant/wild-type bacteria in theinput; each value shown is from an individual mouse.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6565

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

yses of flagellar promoter expression in the regulatory mutants(51). The current transcriptome analysis of the flagellar regu-latory mutants confirms the previous hierarchical model andhas also allowed us to refine and expand the flagellar hierarchy(Fig. 2C). For example, we had previously designated the fliEto -Q genes as a single operon under the control of the class IIfliE promoter. However, the microarray analysis revealed thatthe fliK to -Q genes showed a class III pattern of expression,suggesting the presence of a class III (FlrC-dependent) pro-moter upstream of fliK, which was confirmed by lacZ fusionassays (Table 1). The transcriptome analyses also allowed us toidentify new flagellar genes, such as the class III flgOP genes.FlgP is a lipoprotein required for motility and enhanced intes-tinal colonization, which we described previously (44). In thepresent study, the transcriptome analyses identified VC1384 asa potential class III gene, which was confirmed by a promoter-lacZ assay (Table 1). Inactivation of VC1384 leads to reducedV. cholerae motility, while expression of VC1384 in a wild-typestrain leads to increased motility (see Fig. S1 in the supple-mental material), indicating that this gene should be con-sidered a motility gene. Finally, there are 43 genes anno-tated as methyl-accepting chemotaxis protein genes in the V.cholerae genome (24), and the transcriptome analyses indi-cate that seven methyl-accepting chemotaxis protein genes(VC1898, VC2439, VCA0176, VCA0658, VCA0773, VCA0923,and VCA0974) display the expression pattern of class IV flagel-lar genes.

Among the additional genes downregulated in the flagellarregulatory mutants were a number of genes involved in aminoacid and cofactor biosynthesis, cell division, energy metab-olism, protein and nucleic acid synthesis, transport, andregulation. The change in transcription patterns suggestsmajor alterations in cell physiology in nonflagellated cells.For example, we confirmed that the succinate dehydroge-nase operon is downregulated in the flagellar regulatory mu-tants. Because succinate dehydrogenase participates in boththe tricarboxylic acid cycle and oxidative phosphorylation, itsexpression has been shown to be downregulated under anaer-obic conditions in other bacteria (48). The cell division geneftsZ is also downregulated in the flagellar regulatory mutants,suggesting that cell division is decreased in the absence offlagella. Genes encoding a novel adhesin and c-di-GMP syn-thases are also downregulated in the flagellar regulatory mu-tants; these are discussed below.

However, in addition to the physiologic capabilities down-regulated in the flagellar regulatory mutants observed in thetranscriptome studies, phenotype microarrays revealed en-hanced catabolic capabilities in these mutants for certaincarbon sources. For example, the flagellar mutants showedenhanced growth on polymers of glucose and mannose (lam-inarin and mannan) and on gelatin, glutamine, and histidine ascarbon sources. The flagellar regulatory mutants show en-hanced gelatinase activity in comparison with that of the wild-type strain (see Fig. S2 in the supplemental material), whichmay explain their enhanced growth with gelatin utilized as acarbon source. The results indicate that nonflagellated V. chol-erae bacteria are physiologically programmed for enhancedutilization of certain nutrients, as might be expected underconditions where flagellar synthesis would be downregulated,such as during intestinal colonization or biofilm formation.

The flagellar regulatory hierarchy positively regulates FrhA,a novel adhesin that enhances intestinal colonization. It hadpreviously been shown that spontaneous nonmotile V. choleraestrains were decreased for hemagglutination of human eryth-rocytes (16), but the identity of the flagellum-regulated hem-agglutinin was unknown. The transcriptome analyses facili-tated the identification of FrhA, a novel adhesin that facilitatesV. cholerae binding not only to human erythrocytes but also toepithelial cells, chitin, and abiotic surfaces during biofilm de-velopment. FrhA is in the RTX family of toxin and Ca2�-binding proteins based on homology within the C terminus(COG2931). This large, 234-kDa protein also contains fourcadherin repeats typically found in the cadherin family of pro-teins. Different sequenced pathogenic V. cholerae strains haveup to four additional cadherin repeats in FrhA (see Fig. S3 inthe supplemental material), suggesting an important functionfor these repeats within the human host. A cadherin repeat isan independently folding sequence of �110 aa with a Ca2�-binding pocket, and most cadherin proteins contain multipletandem cadherin repeats (49). Eukaryotic cadherins are sur-face-localized proteins that mediate Ca2�-dependent cell-celladhesion, and we anticipate that the cadherin repeats in FrhAlikely mediate the adherence of V. cholerae to erythrocytes andepithelial cells, perhaps to cadherins on these cells’ surfaces.During the course of these studies, a report by Chatterjee et al.described the presence of two tandem RTX-like genes in V.cholerae (6); upon closer inspection of the genome of the strainutilized in their studies, N16961, we found that a stop codonwithin the coding sequence of frhA is present in strain N16961,thus making it appear as if two tandem RTX-like genes arepresent, rather than a single pseudogene. Sequencing of frhAfrom strain N16961 confirmed the presence of this stop codon(data not shown).

The C terminus of FrhA contains a type I secretion signalmotif, suggesting that it is secreted through a type I-dependentmechanism, similar to other RTX family members (4). Weshowed that the divergently transcribed FrhC protein, whichhas homology with the type I secretion component TolC, isalso regulated by the flagellar hierarchy and is also required forhemagglutination, strongly suggesting that FrhC is required forFrhA secretion. However, the other components normally in-volved in type I secretion (i.e., HlyB and HlyD homologs) werenot obviously encoded nearby, although VC1618 shares somehomology with HlyB.

The flagellar regulatory hierarchy controls the expression ofa DGC, CdgD, and this protein in turn regulates the transcrip-tion of frhA and frhC. CdgD is predicted to contain two trans-membrane segments (aa 27 to 41 and aa 327 to 339) by theDAS algorithm (13) as well as a PAS sensory box domain(which normally binds prosthetic groups) and the GGDEFDGC domain at its C terminus. CdgD was previously shown tonegatively regulate flagellum-mediated motility (37). Since wehave shown here that the flagellar regulatory hierarchy con-trols CdgD expression, this demonstrates a regulatory loopwhereby the flagellum-regulated DGC in turn downregulatesflagellum-mediated motility. CdgD was also previously shownto enhance early biofilm development in El Tor V. cholerae(37). We have confirmed this observation here in the classicalV. cholerae background and, moreover, demonstrated thatCdgD-dependent enhancement of biofilm formation is due to

6566 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CdgD-dependent transcription of the flagellum-regulatedhemagglutinin.

The mechanism(s) by which DGCs modulate gene expres-sion is still not entirely clear, especially in V. cholerae. Theexpression of high levels of c-di-GMP causes global transcrip-tion patterns that generally favor exopolysaccharide expressionand biofilm formation, while low levels of c-di-GMP lead togene expression characteristic of flagellated and virulent V.cholerae (3, 64). Proteins containing PilZ domains bind c-di-GMP and contribute to some of these phenotypes (50). How-ever, a large number of proteins containing DGC (40) andc-di-GMP phosphodiesterase domains (20) exist in V. chol-erae, a number of these domains reside within the same pro-tein, and many are expressed simultaneously under any givencondition. Thus, it is not clear exactly how presumably smallfluctuations in the intracellular c-di-GMP pool stimulated bydiverse GGDEF- and EAL-containing proteins can lead tospecific gene expression changes under a certain set of envi-ronmental conditions. For example, both CdgD and CdgE arepositively regulated by the flagellar regulatory hierarchy, andboth are predicted to be membrane-associated DGCs, yet onlyCdgD is involved in regulating flagellum-regulated hemagglu-tinin and mediating binding to epithelial cells and chitin. Pre-viously, CdgD, and not CdgE, was found to influence biofilmformation and motility, despite both being expressed underbiofilm-inducing conditions (37). In fact, we have evidencesuggesting that both have DGC activity; specifically, overex-pression of both CdgD and CdgE in the O395 V. cholerae strainleads to inhibition of CT expression (data not shown), similarto the effect of overexpression of the DGC VCA0956 (64).

Given these observations, it may be that DGCs influencegene expression changes in a more localized manner, ratherthan nonspecifically contributing to a common intracellularc-di-GMP pool. In other words, a specific Cdg, such as CdgD,may interact with a specific cognate c-di-GMP binding proteinand thus influence gene expression in a specific manner. How-ever, when intracellular c-di-GMP levels rise significantly, ashas been observed with the overexpression of Cdgs, this wouldinfluence multiple c-di-GMP-responsive proteins, which wouldotherwise only respond to their cognate Cdg. The future iden-tification of the c-di-GMP-responsive protein that directly reg-ulates flagellum-regulated hemagglutinin should help in thedissection of flagellar regulation of this adhesin.

FrhA enhances V. cholerae intestinal colonization, and thiswas observed in the infant mouse model and suggested tooccur in the adult mouse model. In infant mice, the ability ofthe mutant to compete with the wild-type strain is observed at24 h postinoculation, and the frhA mutant had an approxi-mately twofold defect in colonization. In the adult mouse, thenumber of successfully colonized mice at 24 h and the averageintestinal burden at 48 h were lower in mice inoculated withthe frhA strain than in mice inoculated with the wild-type strainat these same time points. These results demonstrate thatFrhA enhances intestinal colonization, and given its role as anadhesin to epithelial cells, we predict that FrhA facilitatesbinding to the intestinal epithelial surface.

The most important intestinal colonization factor, TCP, fa-cilitates bacterium-bacterium interactions and microcolonyformation within the intestine (61), but no direct role inbinding to intestinal cells has been identified for TCP. The

GlcNAc-binding protein GbpA was identified as a proteinthat enhances intestinal colonization (28), and we showed herethat GbpA expression is upregulated in the flagellar regulatorymutants. However, the nonmotile mutants with enhancedGbpA expression showed reduced binding to erythrocytes, ep-ithelial cells, and even chitin, which was due to reduced FrhAexpression, suggesting that FrhA may play a more dominantrole in binding to these surfaces, at least in the laboratory.However, it appears likely that multiple adhesins contribute tointestinal colonization as well as biofilm formation, and thus,the individual contributions of select adhesins with redundantfunctions may seem modest in the absence of a single essentialadhesin.

Our results demonstrate that the flagellar regulatory hierar-chy controls transcription of CdgD, which in turn regulatesflagellum-regulated hemagglutinin transcription, presumablythrough DGC activity, and FrhA then mediates binding tosurfaces, which enhances intestinal colonization as well as bio-film formation. We suggest that this adhesin is utilized toinitiate colonization of the intestine or development of a bio-film, when the bacteria are flagellated and motile upon ap-proach to the epithelial cell surface or the environmental sur-face, respectively (Fig. 7).

The flagellar regulatory hierarchy coordinately regulatesdiverse virulence genes. The ToxR/TcpP/ToxT regulon hasbeen extensively studied, because this regulatory circuit con-trols the expression of the two most important virulence factorsfor V. cholerae, TCP and CT (reviewed in reference 7). In theflagellar regulatory mutants, transcription of the ctx, tcp, andacf genes was upregulated, and this is probably due to theupregulation of transcription of toxT, the regulator that di-rectly binds and activates the ctx, tcp, and acf genes. Interest-ingly, transcription of tcpP or toxR was not upregulated in theflagellar mutants. TcpP and ToxR together activate the tran-scription of toxT, and expression of TcpP is correlated with CTand TCP expression, because ToxR is constitutively expressed(22, 33). Thus, the upregulation of toxT, ctx, tcp, and acf genesin the absence of tcpP upregulation suggests that the flagellarregulatory cascade influences the transcription of these genesin a TcpP- and ToxR-independent manner. The genes encod-ing VieSAB were also upregulated in the flagellar regulatorymutants; VieA is a c-di-GMP phosphodiesterase that stimu-lates CT expression and downregulates biofilm formation (63,64). VieA stimulates CT expression by decreasing c-di-GMP,and this effect is TcpP and ToxR independent (64), suggestingthat the upregulation of ctx transcription in the flagellar regu-

FIG. 7. Proposed influence of the flagellar regulatory hierarchy onV. cholerae behavior within the intestine.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6567

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

latory mutants may be mediated by the upregulation of vieAtranscription.

Among the additional virulence genes upregulated in theflagellar regulatory mutants was the gene encoding the “ElTor” hemolysin HlyA (also called cytolysin). This is a pore-forming exotoxin that causes cell vacuolation (8, 43), and al-though classical V. cholerae strains (such as that used in thisstudy) contain the hlyA gene, an 11-bp deletion within thecoding sequence renders the resulting product nonhemolytic(2). HlyA expression is under the control of HlyU (70), buttranscription of the hlyU gene was not upregulated in theflagellar regulatory mutants.

Spontaneous nonmotile V. cholerae strains were previouslyshown to have increased hemolytic activity toward human typeO erythrocytes (16), but the identity of the flagellum-regulatedhemolysin was unknown. We have shown here that transcrip-tion of thermolabile hemolysin (tlh) is increased in the flagellarregulatory mutants and that inactivation of tlh in these mutantsrenders them nonhemolytic, thus identifying thermolabile he-molysin as the flagellum-regulated hemolysin. Thermolabilehemolysin has phospholipase and lecithinase activity (14), andthe tlh gene is adjacent to and divergently transcribed from thehlyA gene. Because hlyA and tlh are similarly upregulated inthe flagellar regulatory mutants, the divergent tlh and hlyApromoters are likely coordinately regulated by the same induc-ing factor(s), although this regulation does not involve in-creased expression of HlyU. The tlh mutant colonized infantmice similarly to the wild-type strain (see Fig. S9 in the sup-plemental material), consistent with a previous study with rab-bits (14) indicating the lack of a dominant role for this proteinin virulence. However, a number of virulence factors with pos-sibly redundant roles are expressed simultaneously within theintestine, which may obscure the individual contributions ofthese various factors, as has been suggested by the studies ofOlivier et al. (47).

Interestingly, HlyU also controls expression of Hcp (hemo-lysin-coregulated protein), one of the secreted factors found inthe type VI secretion cluster (52, 71). Although transcriptionof the hcp gene was not altered in the flagellar regulatorymutants, there was an upregulation of several type VI secretiongenes that lie in a cluster (VCA0107, VCA0108, VCA0109,VCA0112, VCA0113, VCA0117, and VCA0124). The T6SS isrequired for virulence of V. cholerae in Dictyostelium discoi-deum, and previous studies suggested that this cluster of geneswas positively regulated by �54 and the �54-dependent activa-tor encoded by VCA0117 (52). We observed upregulation ofthese genes in the flagellar regulatory mutants, including therpoN strain, which indicates an additional �54-independent,flagellum-dependent mechanism responsible for increasedtranscription.

The eps genes were also upregulated in the flagellar regula-tory mutants; these genes encode the type II secretion systemresponsible for CT secretion as well as secretion of CTX�,hemagglutinin/protease, chitinase, neuraminidase, and lipase(55). The eps genes have previously been shown to be inducedin response to increased intracellular c-di-GMP (3), but in thecase of the flagellar regulatory mutants, intracellular c-di-GMPlevels would likely be lower than those in the wild-type strain,suggesting an alternate (flagellum-dependent) mechanismfor regulating extracellular protein secretion. The chitin-bind-

ing protein GbpA was also upregulated in the flagellar regu-latory mutants, and GbpA expression has previously beenshown to be stimulated by the presence of GlcNAc (40). Theupregulation of gbpA in the flagellar regulatory mutants sug-gests an alternative GlcNAc-independent, flagellum-depen-dent mechanism of control over its expression.

Our results demonstrate coordination of the transcriptionsof diverse virulence genes via the flagellar regulatory hierarchy(Fig. 7). The upregulation of CT, TCP, extracellular proteinsecretion, HlyA, thermolabile hemolysin, GbpA, and the T6SSas a result of the downregulation of flagellar synthesis would bepredicted to enhance prolonged colonization and pathogenesisat the epithelial cell surface. The flagellar regulatory systemthus acts as an important signaling component of the patho-genic process, positively regulating factors that assist in arrivalat the colonization site (motility and chemotaxis) and initialadherence upon arrival (flagellum-regulated hemagglutinin),while negatively regulating diverse virulence genes that facili-tate persistence of colonized, less (non)motile bacteria; aftercolonization we predict that the flagellar regulatory systemdownregulates flagellar, chemotaxis, and adherence genes, andthus the repression of the TCP, CT, and other virulence factorsis relieved, which facilitates a productive colonization phase ofthe bacteria.

ACKNOWLEDGMENT

This study was supported by NIH grant AI43486 to K.E.K.

REFERENCES

1. Abendroth, J., M. Bagdasarian, M. Sandkvist, and W. G. Hol. 2004. Thestructure of the cytoplasmic domain of EpsL, an inner membrane compo-nent of the type II secretion system of Vibrio cholerae: an unusual memberof the actin-like ATPase superfamily. J. Mol. Biol. 344:619–633.

2. Alm, R. A., U. H. Stroeher, and P. A. Manning. 1988. Extracellular proteinsof Vibrio cholerae: nucleotide sequence of the structural gene (hlyA) for thehaemolysin of the haemolytic El Tor strain O17 and characterization of thehlyA mutation in the non-haemolytic classical strain 569B. Mol. Microbiol.2:481–488.

3. Beyhan, S., A. D. Tischler, A. Camilli, and F. H. Yildiz. 2006. Transcriptomeand phenotypic responses of Vibrio cholerae to increased cyclic di-GMP level.J. Bacteriol. 188:3600–3613.

4. Boardman, B. K., and K. J. Satchell. 2004. Vibrio cholerae strains withmutations in an atypical type I secretion system accumulate RTX toxinintracellularly. J. Bacteriol. 186:8137–8143.

5. Cassel, D., and Z. Selinger. 1977. Mechanism of adenylate cyclase activationby cholera toxin: inhibition of GTP hydrolysis at the regulatory site. Proc.Natl. Acad. Sci. USA 74:3307–3311.

6. Chatterjee, R., S. Nag, and K. Chaudhuri. 2008. Identification of a new RTX-like gene cluster in Vibrio cholerae. FEMS Microbiol. Lett. 284:165–171.

7. Childers, B. M., and K. E. Klose. 2007. Regulation of virulence in Vibriocholerae: the ToxR regulon. Future Microbiol. 2:335–344.

8. Coelho, A., J. R. Andrade, A. C. Vicente, and V. J. DiRita. 2000. Cytotoxiccell vacuolating activity from Vibrio cholerae hemolysin. Infect. Immun. 68:1700–1705.

9. Correa, N. E., J. R. Barker, and K. E. Klose. 2004. The Vibrio cholerae FlgMhomologue is an anti-�28 factor that is secreted through the polar sheathedflagellum. J. Bacteriol. 186:4613–4619.

10. Correa, N. E., and K. E. Klose. 2005. Characterization of enhancer bindingby the Vibrio cholerae flagellar regulatory protein FlrC. J. Bacteriol. 187:3158–3170.

11. Correa, N. E., C. M. Lauriano, R. McGee, and K. E. Klose. 2000. Phosphor-ylation of the flagellar regulatory protein FlrC is necessary for Vibrio choleraemotility and enhanced colonization. Mol. Microbiol. 35:743–755.

12. Coster, T. S., K. P. Killeen, M. K. Waldor, D. Beattie, D. Spriggs, J. R.Kenner, A. Trofa, J. Sadoff, J. J. Mekalanos, and D. N. Taylor. 1995. Safety,immunogenicity and efficacy of a live attenuated Vibrio cholerae 0139 vaccineprototype, Bengal-15. Lancet 345:949–952.

13. Cserzo, M., E. Wallin, I. Simon, G. von Heijne, and A. Elofsson. 1997.Prediction of transmembrane alpha-helices in prokaryotic membrane pro-teins: the dense alignment surface method. Protein Eng. 10:673–676.

14. Fiore, A. E., J. M. Michalski, R. G. Russell, C. L. Sears, and J. B. Kaper. 1997.

6568 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

Cloning, characterization, and chromosomal mapping of a phospholipase (lec-ithinase) produced by Vibrio cholerae. Infect. Immun. 65:3112–3117.

15. Freter, R., P. C. M. O’Brien, and M. S. Macsai. 1981. Role of chemotaxis inthe association of motile bacteria with intestinal mucosa: in vivo studies.Infect. Immun. 34:234–240.

16. Gardel, C. L., and J. J. Mekalanos. 1996. Alterations in Vibrio choleraemotility phenotypes correlate with changes in virulence factor expression.Infect. Immun. 64:2246–2255.

17. Gentleman, R. C., V. J. Carey, D. M. Bates, B. Bolstad, M. Dettling, S.Dudoit, B. Ellis, L. Gautier, Y. Ge, J. Gentry, K. Hornik, T. Hothorn, W.Huber, S. Iacus, R. Irizarry, F. Leisch, C. Li, M. Maechler, A. J. Rossini, G.Sawitzki, C. Smith, G. Smyth, L. Tierney, J. Y. Yang, and J. Zhang. 2004.Bioconductor: open software development for computational biology andbioinformatics. Genome Biol. 5:R80.

18. Ghosh, A., K. Paul, and R. Chowdhury. 2006. Role of the histone-likenucleoid structuring protein in colonization, motility, and bile-dependentrepression of virulence gene expression in Vibrio cholerae. Infect. Immun.74:3060–3064.

19. Goldberg, S. L., and J. R. Murphy. 1985. Cloning and characterization of thehemolysin determinants from Vibrio cholerae RV79 (Hly�), RV79 (Hly�),and 569B. J. Bacteriol. 162:35–41.

20. Guzman, L.-M., D. Belin, M. J. Carson, and J. Beckwith. 1995. Tight regu-lation, modulation, and high-level expression by vectors containing the ar-abinose PBAD promoter. J. Bacteriol. 177:4121–4130.

21. Hanahan, D. 1983. Studies on transformation of Escherichia coli with plas-mids. J. Mol. Biol. 166:577–580.

22. Hase, C. C., and J. J. Mekalanos. 1998. TcpP protein is a positive regulatorof virulence gene expression in Vibrio cholerae. Proc. Natl. Acad. Sci. USA95:730–734.

23. Hava, D. L., and A. Camilli. 2001. Isolation and characterization of a tem-perature-sensitive generalized transducing bacteriophage for Vibrio cholerae.J. Microbiol. Methods 46:217–225.

24. Heidelberg, J. F., J. A. Eisen, W. C. Nelson, R. A. Clayton, M. L. Gwinn, R. J.Dodson, D. H. Haft, E. K. Hickey, J. D. Peterson, L. Umayam, S. R. Gill,K. E. Nelson, T. D. Read, H. Tettelin, D. Richardson, M. D. Ermolaeva, J.Vamathevan, S. Bass, H. Qin, I. Dragoi, P. Sellers, L. McDonald, T. Utter-back, R. D. Fleishmann, W. C. Nierman, O. White, S. L. Salzberg, H. O.Smith, R. R. Colwell, J. J. Mekalanos, J. C. Venter, and C. M. Fraser. 2000.DNA Sequence of both chromosomes of the cholera pathogen Vibrio chol-erae. Nature 406:477–483.

25. Holland, I. B., L. Schmitt, and J. Young. 2005. Type I protein secretion inbacteria, the ABC-transporter dependent pathway. Mol. Membr. Biol. 22:29–39.

26. Horton, R. M., H. D. Hunt, S. N. Ho, J. K. Pullen, and L. R. Pease. 1989.Engineering hybrid genes without the use of restriction enzymes: gene splic-ing by overlap extension. Gene 77:61–68.

27. Kenner, J. R., T. S. Coster, D. N. Taylor, A. F. Trofa, M. Barrera-Oro, T.Hyman, J. M. Adams, D. T. Beattie, K. P. Killeen, D. R. Spriggs, J. J.Mekalanos, and J. C. Sadoff. 1995. Peru-15, an improved live attenuatedvaccine candidate for Vibrio cholerae 01. J. Infect. Dis. 172:1126–1129.

28. Kirn, T. J., B. A. Jude, and R. K. Taylor. 2005. A colonization factor linksVibrio cholerae environmental survival and human infection. Nature 438:863–866.

29. Klose, K. E. 2000. The suckling mouse model of cholera. Trends Microbiol.8:189–191.

30. Klose, K. E., and J. J. Mekalanos. 1998. Differential regulation of multipleflagellins in V. cholerae. J. Bacteriol. 180:303–316.

31. Klose, K. E., and J. J. Mekalanos. 1998. Distinct roles of an alternative sigmafactor during both free-swimming and colonizing phases of the Vibrio chol-erae pathogenic cycle. Mol. Microbiol. 28:501–520.

32. Klose, K. E., V. Novick, and J. J. Mekalanos. 1998. Identification of multiple�54-dependent transcriptional activators in Vibrio cholerae. J. Bacteriol. 180:5256–5259.

33. Krukonis, E. S., R. R. Yu, and V. J. DiRita. 2000. The Vibrio choleraeToxR/TcpP/ToxT virulence cascade: distinct roles for two membrane-localized transcriptional activators on a single promoter. Mol. Microbiol.38:67–84.

34. Lauriano, C. M., C. Ghosh, N. E. Correa, and K. E. Klose. 2004. Thesodium-driven flagellar motor controls exopolysaccharide expression inVibrio cholerae. J. Bacteriol. 186:4864–4874.

35. Lee, S. H., S. M. Butler, and A. Camilli. 2001. Selection for in vivo regulatorsof bacterial virulence. Proc. Natl. Acad. Sci. USA 98:6889–6894.

36. Lenz, D. H., M. B. Miller, J. Zhu, R. V. Kulkarni, and B. L. Bassler. 2005.CsrA and three redundant small RNAs regulate quorum sensing in Vibriocholerae. Mol. Microbiol. 58:1186–1202.

37. Lim, B., S. Beyhan, J. Meir, and F. H. Yildiz. 2006. Cyclic-diGMP signaltransduction systems in Vibrio cholerae: modulation of rugosity and biofilmformation. Mol. Microbiol. 60:331–348.

38. Liu, J., X. Zogaj, J. R. Barker, and K. E. Klose. 2007. Construction oftargeted insertion mutations in Francisella tularensis subsp. novicida. Bio-Techniques 43:487–490.

39. Liu, Z., T. Miyashiro, A. Tsou, A. Hsiao, M. Goulian, and J. Zhu. 2008.

Mucosal penetration primes Vibrio cholerae for host colonization by re-pressing quorum sensing. Proc. Natl. Acad. Sci. USA 105:9769–9774.

40. Meibom, K. L., X. B. Li, A. T. Nielsen, C. Y. Wu, S. Roseman, and G. K.Schoolnik. 2004. The Vibrio cholerae chitin utilization program. Proc. Natl.Acad. Sci. USA 101:2524–2529.

41. Miller, J. H. 1992. A short course in bacterial genetics, 2nd ed. Cold SpringHarbor Laboratory Press, Plainview, NY.

42. Miller, V. L., and J. J. Mekalanos. 1988. A novel suicide vector and its usein construction of insertion mutations: osmoregulation of outer membraneproteins and virulence determinants in Vibrio cholerae requires toxR. J.Bacteriol. 170:2575–2583.

43. Mitra, R., P. Figueroa, A. K. Mukhopadhyay, T. Shimada, Y. Takeda, D. E.Berg, and G. B. Nair. 2000. Cell vacuolation, a manifestation of the El Torhemolysin of Vibrio cholerae. Infect. Immun. 68:1928–1933.

44. Morris, D. C., F. Peng, J. R. Barker, and K. E. Klose. 2008. Lipidation of anFlrC-dependent protein is required for enhanced intestinal colonization byVibrio cholerae. J. Bacteriol. 190:231–239.

45. Nagamune, K., K. Yamamoto, and T. Honda. 1995. Cloning and sequencingof a novel hemolysis gene of Vibrio cholerae. FEMS Microbiol. Lett. 128:265–269.

46. Olivier, V., G. K. Haines, Y. Tan, and K. J. Satchell. 2007. Hemolysin and themultifunctional autoprocessing RTX toxin are virulence factors during in-testinal infection of mice with Vibrio cholerae El Tor O1 strains. Infect.Immun. 75:5035–5042.

47. Olivier, V., N. H. Salzman, and K. J. Satchell. 2007. Prolonged colonizationof mice by Vibrio cholerae El Tor O1 depends on accessory toxins. Infect.Immun. 75:5043–5051.

48. Park, S. J., C. P. Tseng, and R. P. Gunsalus. 1995. Regulation of succinatedehydrogenase (sdhCDAB) operon expression in Escherichia coli in re-sponse to carbon supply and anaerobiosis: role of ArcA and Fnr. Mol.Microbiol. 15:473–482.

49. Pokutta, S., and W. I. Weis. 2007. Structure and mechanism of cadherins andcatenins in cell-cell contacts. Annu. Rev. Cell Dev. Biol. 23:237–261.

50. Pratt, J. T., R. Tamayo, A. D. Tischler, and A. Camilli. 2007. PilZ domainproteins bind cyclic diguanylate and regulate diverse processes in Vibriocholerae. J. Biol. Chem. 282:12860–12870.

51. Prouty, M. G., N. E. Correa, and K. E. Klose. 2001. The novel �54- and�28-dependent flagellar gene transcription hierarchy of Vibrio cholerae. Mol.Microbiol. 39:1595–1609.

52. Pukatzki, S., A. T. Ma, D. Sturtevant, B. Krastins, D. Sarracino, W. C.Nelson, J. F. Heidelberg, and J. J. Mekalanos. 2006. Identification of aconserved bacterial protein secretion system in Vibrio cholerae using theDictyostelium host model system. Proc. Natl. Acad. Sci. USA 103:1528–1533.

53. Richardson, K. 1991. Roles of motility and flagellar structure in pathoge-nicity of Vibrio cholerae: analysis of motility mutants in three animal models.Infect. Immun. 59:2727–2736.

54. Ronson, C. W., P. M. Astwood, B. T. Nixon, and F. M. Ausubel. 1987.Deduced products of C4-dicarboxylate transport regulatory genes of Rhizo-bium leguminosarum are homologous to nitrogen regulatory gene products.Nucleic Acids Res. 15:7921–7934.

55. Sandkvist, M., L. O. Michel, L. P. Hough, V. M. Morales, M. Bagdasarian,M. Koomey, V. J. DiRita, and M. Bagdasarian. 1997. General secretionpathway (eps) genes required for toxin secretion and outer membrane bio-genesis in Vibrio cholerae. J. Bacteriol. 179:6994–7003.

56. Schmitt, L., H. Benabdelhak, M. A. Blight, I. B. Holland, and M. T. Stubbs.2003. Crystal Structure of the nucleotide-binding domain of the ABC-trans-porter haemolysin B: identification of a variable region within ABC helicaldomains. J. Mol. Biol. 330:333–342.

57. Shapiro, L., A. M. Fannon, P. D. Kwong, A. Thompson, M. S. Lehmann, G.Gruebel, J. F. Legrand, J. Als-Nielsen, D. R. Colman, and W. A. Hendrick-son. 1995. Structural basis of cell-cell adhesion by cadherins. Nature 374:327–337.

58. Silva, A. J., G. J. Leitch, A. Camilli, and J. A. Benitez. 2006. Contribution ofhemagglutinin/protease and motility to the pathogenesis of El Tor biotypecholera. Infect. Immun. 74:2072–2079.

59. Simons, R. W., F. Houman, and N. Kleckner. 1987. Improved single andmulticopy lac-based cloning vectors for protein and operon fusions. Gene53:85–96.

60. Skorupski, K., and R. K. Taylor. 1996. Positive selection vectors for allelicexchange. Gene 169:47–52.

61. Taylor, R. K., V. L. Miller, D. B. Furlong, and J. J. Mekalanos. 1987. Use ofphoA gene fusions to identify a pilus colonization factor coordinately regu-lated with cholera toxin. Proc. Natl. Acad. Sci. USA 84:2833–2837.

62. Thanabalu, T., E. Koronakis, C. Hughes, and V. Koronakis. 1998. Substrate-induced assembly of a contiguous channel for protein export from E. coli:Reversible bridging of an inner-membrane translocase to an outer mem-brane exit pore. EMBO J. 17:6487–6496.

63. Tischler, A. D., and A. Camilli. 2004. Cyclic diguanylate (c-di-GMP) regu-lates Vibrio cholerae biofilm formation. Mol. Microbiol. 53:857–869.

64. Tischler, A. D., and A. Camilli. 2005. Cyclic diguanylate regulates Vibriocholerae virulence gene expression. Infect. Immun. 73:5873–5882.

VOL. 191, 2009 V. CHOLERAE FLAGELLAR REGULATORY HIERARCHY 6569

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

65. Tischler, A. D., S. H. Lee, and A. Camilli. 2002. The Vibrio cholerae vieSABlocus encodes a pathway contributing to cholera toxin production. J. Bacte-riol. 184:4104–4113.

66. Tusher, V. G., R. Tibshirani, and G. Chu. 2001. Significance analysis ofmicroarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci.USA 98:5116–5121.

67. Wang, R. F., and S. Kushner. 1991. Construction of versatile low-copy-number vectors for cloning, sequencing and gene expression in Escherichiacoli. Gene 100:195–199.

68. Watnick, P. I., and R. Kolter. 1999. Steps in the Development of a Vibriocholerae biofilm. Mol. Microbiol. 34:586–595.

69. Watnick, P. I., C. M. Lauriano, K. E. Klose, L. Croal, and R. Kolter. 2001.

The absence of a flagellum leads to altered colony morphology, biofilmdevelopment, and virulence in Vibrio cholerae O139. Mol. Microbiol. 39:223–235.

70. Williams, S. G., S. R. Attridge, and P. A. Manning. 1993. The transcriptionalactivator HlyU of Vibrio cholerae: nucleotide sequence and role in virulencegene expression. Mol. Microbiol. 9:751–760.

71. Williams, S. G., L. T. Varcoe, S. R. Attridge, and P. A. Manning. 1996. Vibriocholerae Hcp, a secreted protein coregulated with HlyA. Infect. Immun.64:283–289.

72. Yildiz, F. H., X. S. Liu, A. Heydorn, and G. K. Schoolnik. 2004. Molecularanalysis of rugosity in a Vibrio cholerae O1 El Tor phase variant. Mol.Microbiol. 53:497–515.

6570 SYED ET AL. J. BACTERIOL.

on Decem

ber 27, 2020 by guesthttp://jb.asm

.org/D

ownloaded from