Ins and Outs of Vibrio choleraeVibrio cholerae transitions between the human gut and the aquaticenvironment are aided by specific shifts in gene expression

Stefan Schild, Anne L. Bishop, and Andrew Camilli

The gram-negative, motile, curved-rod bacterium Vibrio cholerae cancause cholera—an acute, explosivediarrheal disease. Epidemic cholerawas described in the 1500s, but re-

ports of cholera-like symptoms date back muchearlier, to the times of Hippocrates and theBuddha. In 1849 John Snow, a London physi-cian, determined that cholera is transmitted bywater, and in 1883 Robert Koch successfullyisolated the cholera vibrio. Facilitated by theexpansion of global trade routes, cholera spreadworldwide in 1817 and remains a global healthchallenge.

Cholera patients typically lose large volumesof fluid. If left untreated, even previously healthyadult patients may become severely dehydratedand can die. Although rehydration therapymakes cholera survivable, it does not cure thediarrhea or prevent infections. Cholera remainsa social and economic burden, especially in thedeveloping world. In 2006 the World Health

Organization reported a total of 236,896 chol-era cases with 6,311 deaths worldwide. Thesefigures likely greatly underestimate the toll ofcholera. Many cases go unreported during sea-sonal outbreaks, and a substantial percentage ofpeople die without diagnosis.

Two Serogroups Give Rise

to Cholera Disease

Despite there being more than 200 known sero-groups of V. cholerae in the aquatic environ-ment, only strains of two serogroups, O1 andO139, are known to cause epidemic disease.O-antigen and core oligosaccharide of the O1and O139 lipopolysaccharides, as well as thecapsule of O139, contribute in colonization ofthe small intestine and adhesion to epithelialcells. Strains within serogroup O1 can be furtherdivided into two biotypes, El Tor and classical.These two biotypes are based upon phenotypictraits, including phage and polymyxin B sensi-

tivity or resistance. Classical and El Torexhibit many differences, often at the levelof gene regulation.

The genome sequence of the V. choleraeclinical isolate N16961 is of similar size tothat of other �-proteobacteria. In this case,N16961 contains 4.033 Mbp, which is sim-ilar in size to that of Escherichia coli K12with 4.639 Mbp. However, the V. choleraegenome is split into two chromosomes, withone containing 2.961 Mbp and the other1.072 Mbp, a feature that is seen in otherVibrio species. Having two smaller chromo-somes, which replicate in a coordinated butindependent fashion, might speed replica-tion.

V. cholerae is closely related to otheraquatic vibrios, some of which live closely

Summary

• The facultative pathogen Vibrio cholerae hasadapted to survive and colonize two very differ-ent environments—the aquatic environmentand the human gastrointestinal tract.

• The intracellular signal molecule c-di-GMP caninversely regulate genes specific for the aquaticenvironment or the host.

• Quorum sensing is one of the signals that helpsto regulate c-di-GMP levels.

• While genes essential for colonization are in-duced early during infection, late in vivo in-duced genes can increase fitness for transitioninto the aquatic environment.

Stefan Schild andAnne L. Bishop areresearch associatesand Andrew Camilliis a professor in theHoward HughesMedical Instituteand the Depart-ment of MolecularBiology and Micro-biology, Tufts Uni-versity School ofMedicine, Boston,Mass.

Volume 3, Number 3, 2008 / Microbe Y 131

with marine organisms either as symbionts,commensals, or pathogens. V. cholerae O1likely arose from an aquatic ancestor that wasnonpathogenic to humans. V. cholerae is notonly transmitted via contaminated water, likeother fecal-orally transmitted pathogens such asSalmonella or E. coli, but also spends consider-able time as an aquatic organism in both freshand salt water.

Since epidemiological record-keeping beganin the early 19th century, eight cholera pandem-ics have been recorded. O1 classical caused thefifth and sixth pandemics and possibly the fourearlier pandemics. In contrast, O1 El Tor isresponsible for the seventh pandemic, whichbegan in the 1960s. Although strains of theclassical biotype are probably extinct in nature,both serotypes are still extensively studied in thelaboratory, and differences that may explain theemergence of El Tor remain of great interest toresearchers in the field.

Since its recent emergence in 1992, cholerainfections caused by O139 have spread world-wide. Although O139 has not replaced O1 ElTor, it is responsible for an eighth cholera pan-demic that is running in parallel with the sev-enth. Genetic analysis reveals O139 to be closelyrelated to O1 El Tor, suggesting that O139evolved from an El Tor O1 strain that acquirednew LPS and capsule biosynthesis genes.

Life Cycle of Vibrio cholerae

In our laboratory we are striving to gain a betterunderstanding of the full life cycle (all of the “insand outs”) of the facultative pathogen V. chol-erae. As a facultative pathogen, V. cholerae re-sides in two worlds (Fig. 1): as a natural inhab-itant of aquatic ecosystems and as the causativeagent of a severe diarrhea in the gastrointestinaltract of its human host. In the environment, V.cholerae is found in association with zooplank-ton, crustaceans, and egg masses of chirono-mids, commonly called “midges.” A variety offactors that promote adherence to chitin sur-faces of such organisms have been reported,including N-acetylglucosamine binding proteinA (GbpA) as well as mannose-sensitive hemag-glutinin (MSHA) and PilA pili. This associationmay provide an environmental advantage, sinceV. cholerae is capable of degrading chitin andusing it as a carbon and nitrogen source. Fur-thermore, V. cholerae O1 and O139 form bio-

films. In static cultures biofilm formation isdependent on biosynthesis of Vibrio exopo-lysaccharide (VPS). In contrast, in a flow cellsystem a vps-independent biofilm has recentlybeen described by Alfred Spormann and col-leagues at Stanford University, Calif. It is likelythat survival in aquatic environments is aided bythe formation of such multicellular structures.

When a human ingests V. cholerae, the bacte-ria face a sudden shift in temperature and osmo-larity as well as the extreme acidity of the stom-ach. Later, in the human gut, the bacteriaencounter alkaline pH, digestive enzymes, bile,and components of the innate immune system.In the small intestine, the polar flagellum propelsindividual V. cholerae cells into the mucus layerwhere they can attach to host epithelial cells, themajor sites of bacterial proliferation.

In the mouse as well as in humans, V. choleraeinduces several virulence factors, including chol-era toxin (CTX) and the toxin-coregulated pilus(TCP), the major colonization factor requiredfor microcolony formation. In mice this geneinduction is rapid, occurring within the first fivehours. The activity of CTX causes a massivesecretory diarrhea that, if untreated, can resultin hypotensive shock, organ failure, and deathwithin a day. The ability of V. cholerae to repli-cate with a generation time of 20 minutes maybe advantageous during the gastrointestinal (GI)tract colonization stage of its life cycle, when thepotent effects of CTX provide only a short timein which to proliferate. To induce these viru-lence factors, V. cholerae relies on signal trans-duction and transcriptional regulatory cascadesthat include the two transmembrane regulatorsToxR and TcpP, as well as the cytoplasmicregulator ToxT.

Cholera researchers are making great ad-vances in understanding changes in gene expres-sion at different stages in the V. cholerae lifecycle. Our lab’s research has focused on identi-fying in vivo-induced genes at early stages ofinfection, using recombination-based in vivo ex-pression technology (RIVET) in combinationwith the infant mouse model. Briefly, themethod relies on a library of strains carryingtranscriptional fusions to tnpR (encoding a re-solvase) that are not expressed in vitro. When asubset of these strains expresses the resolvase invivo, they excise a cassette harboring selectableand counter-selectable markers that are used toidentify the strains. Using microarray technol-

132 Y Microbe / Volume 3, Number 3, 2008

F I G U R E 1

Model for the life cycle of V. cholerae. When humans ingest food or water containing V. cholerae, several bacterial genes are induced,including those encoding cholera toxin and toxin coregulated pilus (bottom right), while c-di-GMP levels drop (right). A scanning electronmicrograph (bottom left) shows V. cholerae associated with a microvillus in the infant mouse small intestine (magnification, �750; taken byMichael J. Angelichio). Inducing late genes (bottom left) may help to maintain an infection and increase fitness prior to the transition backinto aquatic environments. Genes induced late during infection include DGCs, which are predicted to increase c-di-GMP level prior to releaseinto the aquatic environment (left). After release into the environment hyperinfectious V. cholerae can infect a new human host (short-termpersistence). Alternatively, V. cholerae may form associations with and biofilms upon chitinous material, facilitating long-term persistencein the aquatic environment. Pictures of the pond in Bangladesh and of a copepod (top) were taken by Lori Bourassa.

Volume 3, Number 3, 2008 / Microbe Y 133

ogy, we and other research groups have ana-lyzed the transcriptome of V. cholerae in stoolfrom cholera patients to better understand whathappens to the bacteria during human infec-tions. A green fluorescent protein-based screenallowed Jun Zhu and co-workers at Universityof Pennsylvania in Philadelphia to identify genesdownregulated in vivo. They demonstrated thatthe repression of the MSHA-operon is necessaryto evade the host immune system. Gary K.Schoolnik and coworkers at Stanford Univer-sity, Stanford, Calif., reported that once V. chol-erae colonize the rabbit ileal loop, they producea stationary-phase sigma factor (RpoS) that en-ables detachment from the epithelial surface.Whether this sigma factor is required for detach-ment during infection in an open intestinal tractis unclear. However, at some stage during theinfection, bacteria detach from the epitheliallayer and are shed in “rice-water” stool into theenvironment.

The “Ready” State of V. cholerae

V. cholerae shed by cholera patients are moreinfectious than laboratory-cultured bacteriawhen tested in mice. Mathematical modeling ofhuman epidemiological data supports the exis-tence of a hyperinfectivity phenomenon, as re-ported by David M. Hartley at the Universityof Maryland School of Medicine in Baltimoreand his collaborators. Hyperinfectivity is main-tained for at least 5 h in pond water, suggestingthat expression of a subset of V. cholerae genesresponsible for hyperinfectivity persists tempo-rarily and could facilitate reinfection.

Siouxsie Wiles, Gordon Dougan, and GadFrankel at Imperial College London in Englandhave observed similar hyperinfectivity of natu-rally transmitted compared with laboratory-grown bacteria for Citrobacter rodentium afterit passes through the mouse GI tract. Whetherthe mechanisms of hyperinfectivity for V. chol-erae and C. rodentium differ or are conservedremains to be determined. However, Frankeland his collaborators also report that well-char-acterized virulence genes are strongly expressedafter shedding of C. rodentium by mice, whereasin human stool the ToxR and ToxT regulons ofV. cholerae are not.

Meanwhile, our analysis indicates that repres-sion of chemotaxis in fresh stool-borne V. chol-erae enhances virulence in a new host using the

infant mouse model and can be observed pheno-typically using chemotaxis assays. The nature ofV. cholerae in stool samples and its effects oncholera transmission are just beginning to beinvestigated. For instance rice-water stool some-times contains V. cholerae aggregates, some ofwhich are associated with mucus, according toStephen B. Calderwood and colleagues at Mas-sachusetts General Hospital and Harvard Med-ical School in Boston, Mass., and Shah M. Fa-ruque and colleagues at the International Centrefor Diarrhoeal Disease Research in Bangladesh.Such aggregates could enhance environmentalsurvival and affect rates of transmission.

Recently we identified V. cholerae genes spe-cifically induced late in infection using a modi-fied RIVET screen and the infant mouse model.Mutant analysis revealed that most of the lategenes are necessary neither for survival in thehost nor for maintenance of the infection. Byestablishing a novel transition assay, we coulddemonstrate that some of those late gene mu-tants, after in vivo passage, exhibit a fitnessdefect if incubated in pond water and/or rice-water stool. In vivo passage is essential to ob-serve these phenotypes. When human volun-teers ingested a RIVET library made in anattenuated vaccine-candidate strain of V. chol-erae, we and James Kaper and colleagues at theUniversity of Maryland in Baltimore identified asimilar set of late-induced genes, indicating thatthis type of gene regulation occurs when thispathogen passes through human hosts.

The transition from host to environment is asharsh as from the environment to the host. Inleaving the mammalian GI tract into fresh wa-ter, V. cholerae faces a 50-fold drop in osmolar-ity, a shift to lower temperature, and a massiveshift in environmental salts and nutrients. Ex-tensive changes in the transcriptome and pro-teome to adapt to the new conditions are ener-getically costly, but V. cholerae has theadvantage that, through evolution, it can “pre-pare” for this transition. That is to say, a facul-tative pathogen like V. cholerae may inducegenes that are beneficial (and repress genes thatare detrimental) for environmental survivalwhile still in the host, due to a selective advan-tage once in the aquatic environment. This ideais supported by the identification of operonsinvolved in chitin utilization and the biosynthe-sis of MSHA as being induced late in infections.

A significant percentage of the late in vivo

134 Y Microbe / Volume 3, Number 3, 2008

induced genes is associated with uptake or trans-port systems. One explanation for this observa-tion is that V. cholerae may need these systemsready to use and fully functional when it facesthe aquatic environment. Expression of theseuptake systems in vivo may allow V. cholerae tohoard factors that will be limited later in theenvironment, such as carbon sources.

V. cholerae Transitions Are Mediated

by c-di-GMP Signaling

V. cholerae induces diguanylate cyclases (DGCs)at the late stage of infection, which would leadto an increase in cyclic-di-guanylate (c-di-GMP).c-di-GMP is an intracellular bacterial secondmessenger commonly found in gram-negativeand some gram-positive organisms. DGCs syn-thesize c-di-GMP from GTP, while specificphosphodiesterases (PDEs) degrade c-di-GMP.V. cholerae encodes about 60 proteins contain-ing a DGC, a PDE, or both domains, only afraction of which have been demonstrated to beactive.

In V. cholerae high levels of c-di-GMP en-hance biofilm formation by increasing expres-sion of vps genes, while inhibiting transcriptionof virulence and motility genes. In the classicalbiotype, mutation of vieA, which is a responseregulator and PDE, elevates c-di-GMP resultingin enhanced biofilm and dramatically reducedmotility and virulence. Furthermore, transcrip-tional changes in a vieA mutant are similar tothose seen with ectopic DGC overexpression.

These data show that vieA is essential for c-di-GMP down-regulation in the classical bio-type, yet in the El Tor biotype, vieA deletionmutants show no detectable phenotypes andrelatively minor changes in gene transcription.This is one of many intriguing differences be-tween O1 El Tor and classical. We suspect thatone or more of the other more than 20 PDEs inV. cholerae are substituting for vieA in El Tor.

Based on such studies, we hypothesize thatlevels of c-di-GMP in V. cholerae are high in atleast some aquatic environments, facilitatingbiofilm formation, but are reduced early duringinfection to allow proper expression of the vir-ulence genes. Our finding that a mutant strain,which lacks three DGCs expressed late in infec-tion, is less fit for transitioning from stool topond water further suggests that elevated c-

di-GMP plays an important role in this transi-tion.

The molecules that link changes in c-di-GMPlevel with observed changes in gene transcrip-tion have proved difficult to identify. Usingbioinformatic analysis, Dorit Amikam and Mi-chael Galperin at the Tel-Hai Academic Collegein Jerusalem, Israel, and National Institutes ofHealth in Bethesda, Md., respectively, identifiedthe PilZ domain as a putative c-di-GMP bindingmotif. The PilZ-domains of cellulose synthase inGluconacetobacter xylinus and other PilZ-proteins in E. coli and Caulobacter crescentusmediate binding to c-di-GMP and control pro-tein activity.

The V. cholerae genome encodes five PilZ-domain-containing proteins, designated PlzA-PlzE; among them, PlzB shows only weak ho-mology to the PilZ-domain consensus. We findthat both PlzD and PlzC bind c-di-GMP in vitro,while a plzD PilZ-domain point mutation abro-gates binding. A double plzCD mutant is re-duced in virulence in the murine model and thiscan be complemented by plzD on a plasmid, butnot by the plzD PilZ-domain point mutant. Wehave yet to assign roles for the other predictedPlz-proteins in the life cycle of V. cholerae, andwe do not understand the molecular mechanismby which Plz proteins transduce c-di-GMP sig-nals. Further, because DGCs are found in organ-isms that lack any predicted PilZ-domain pro-teins, other classes of proteins likely bind andrespond to c-di-GMP.

Recently, an exciting link has been identifiedbetween c-di-GMP signaling and bacterial celldensity. Quorum sensing allows bacteria tosense their density and/or the density of otherbacteria in their vicinity. In many other patho-genic bacteria at high density, quorum sensingactivates virulence gene expression (enterohae-morrhagic E. coli) and/or biofilm formation(Pseudomonas). V. cholerae is surprisingly dif-ferent in that at high density a key protein in itsquorum sensing cascade (HapR) inhibits bothvirulence gene expression and biofilm forma-tion. Intriguingly, hapR is often mutated in bothenvironmental and clinical isolates such as theclassical biotype strain O395. In addition, thebinding site for HapR in the promoter of aphA,which is a key virulence gene regulator that isinhibited by HapR, is mutated in classical strainO395. These observations suggest that HapR isnot essential to the V. cholerae life cycle. Low

Volume 3, Number 3, 2008 / Microbe Y 135

cell density signaling through phospho-LuxOturns off hapR, enabling virulence and/or bio-film genes to be expressed. Recently, Bonnie L.Bassler at Princeton University, Princeton, N.J.,discovered a new target for quorum sensingwhich is independent of hapR, the predictedDGC VCA0939. VCA0939 translation is acti-vated by the same sRNAs that inhibit transla-tion of hapR and destabilize the hapR mRNA.Whereas hapR mutations are common in clini-cal isolates, the quorum regulator RNA (qrr)sRNAs are highly conserved, suggesting thatHapR-independent, sRNA-dependent branchesof quorum sensing may be essential to V. chol-erae. This opens up exciting new fields of inves-tigation into HapR-independent quorum sens-ing and coordination between quorum sensingand c-di-GMP signaling in the V. cholerae lifecycle.

Conclusions

In the last century improved sanitation and im-provements in oral rehydration therapy havebeen the most important breakthroughs in chol-era prevention and treatment. On the pathogen-esis research front, the powerful combination ofstudies using transcriptional profiling or in vivoexpression technologies, such as RIVET, is be-ginning to create a more comprehensive picture

of the “ins and outs” of V. cholerae as it passesthrough very different environments betweenhost and aquatic reservoirs.

Research into the regulation and character-ization of MSHA is a good example of howinformation acquired by different approaches indifferent laboratories comes together to form amore coherent model. Initially thought to be anadherence factor for binding to epithelial cells,MSHA was later shown to be dispensable forcolonization in vivo but important for biofilmformation and attachment to chitinous surfacesby Ronald K. Taylor and coworkers at Dart-mouth Medical School, Hanover, N.H. Jun Zhuand colleagues at the University of Pennsylvaniamade it evident that the expression of MSHAhas to be down-regulated in vivo to evade thehost immune response. Now, we have identifiedMSHA biosynthesis genes as being induced inthe late stage of infection, which leads to theidea that V. cholerae might, in this way, prepareitself for the transition into the aquatic environ-ment at the end of the infection.

In summary, we hope that an improved un-derstanding of the V. cholerae organism at allstages of its life cycle, in both host and aquaticenvironments, will help us to prevent diseasewith novel ways to block key steps in pathogen-esis or transmission and to develop effective andinexpensive vaccines.

SUGGESTED READING

Butler, S. M., and A. Camilli. 2005. Going against the grain: chemotaxis and infection in Vibrio cholerae. Nature Rev.Microbiol. 3:611–620.Hammer, B. K., and B. L. Bassler. 2007. Regulatory small RNAs circumvent the conventional quorum sensing pathway inpandemic Vibrio cholerae. Proc. Natl. Acad. Sci. USA 104:11145–1149.Hsiao, A., Z. Liu, A. Joelsson, and J. Zhu. 2006. Vibrio cholerae virulence regulator-coordinated evasion of host immunity.Proc. Natl. Acad. Sci. USA 103:14542–14547.Lombardo, M., J. Michalski, H. Martinez-Wilson, C. Morin, T. Hilton, C. G. Osorio, J. P. Nataro, C. O. Tacket, A. Camilli,and J. Kaper. 2007. An in vivo expression technology screen for Vibrio cholerae genes expressed in human volunteers. Proc.Natl. Acad. Sci. USA 104:18229–18234.Nelson, E. J., A. Chowdhury, J. B. Harris, Y. A. Begum, F. Chowdhury, A. I. Khan, R. C. LaRocque, A. L. Bishop, E. T. Ryan,A. Camilli, F. Qadri, and S. B. Calderwood. 2007. Complexity of rice-water stool from patients with Vibrio cholerae plays arole in the transmission of infectious diarrhea. Proc. Natl. Acad. Sci. USA 104:19091–19096.Nielsen, A. T., N. A. Dolganov, G. Otto, M. C. Miller, C. Y. Wu, and G. K. Schoolnik. 2006. RpoS controls the Vibriocholerae mucosal escape response. PLoS Pathog 2:e109.Pratt, J. T., R. Tamayo, A. D. Tischler, and A. Camilli. 2007. PilZ domain proteins bind cyclic diguanylate and regulatediverse processes in Vibrio cholerae. J. Biol. Chem. 282:12860–12870.Schild, S., R. Tamayo, E. J. Nelson, F. Qadri, S. B. Calderwood, and A. Camilli. 2007. Genes induced late in infection increasefitness of Vibrio cholerae after release into the environment. Cell Host Microbe 2:264–277.Tamayo, R., J. T. Pratt, and A. Camilli. 2007. Roles of cyclic diguanylate in the regulation of bacterial pathogenesis. Annu.Rev. Microbiol. 61:131–148.

136 Y Microbe / Volume 3, Number 3, 2008