Characterization of Toxigenic Vibrio cholerae from Haiti, 2010–2011

A virulent clone from Africa or southern Asia was likely introduced at a single time point.

V ibrio cholerae has caused epidemics around the world for centuries. Cholera has long been a companion of devastation and poverty, and epidemics occur in areas without clean water, sanitation, or accessible health care. The collapse of Haiti's infrastructure after the January 2010 earthquake created conditions suitable for cholera to affect the country's vulnerable population.
The fi rst clinical cases of V. cholerae infection in Haiti in >100 years were seen on October 17, 2010. Cholera, primarily a waterborne disease, quickly spread from its origin along a main river in the Artibonite Department north of Port-au-Prince to all 10 departments in Haiti and to the Dominican Republic. Earlier that year, in anticipation of outbreaks after the devastation of the January 2010 earthquake, the Centers for Disease Control and Prevention (CDC; Atlanta, GA, USA) had collaborated with local scientists to tactically position rapid diagnostic tests and had provided training in their use. These tests enabled early recognition of the 2010 cholera outbreak in Haiti. By October 21, 2010, scientists at the Haiti National Public Laboratory (Laboratoire National de Santé Publique) cultured V. cholerae and sent isolates to CDC (1).
Our objectives were to confi rm the identifi cation of V. cholerae, characterize the isolates by using multiple genetic and phenotypic methods, evaluate the clonality of the isolates from Haiti, and attempt to explore the genetic origin of the strain. Isolates from Haiti were compared with isolates from western and eastern Africa, southern Asia, Latin America, the Middle East, and the Gulf Coast of the United States.

Bacterial Strains
A total of 122 V. cholerae isolates from all 10 departments in Haiti were characterized. We also included 25 V. cholerae isolates that showed hemolytic and nonhemolytic phenotypes on sheep blood agar and 2

Confi rmation and Characterization of V. cholerae
All isolates were confi rmed positive by using standard methods (2). Serogroup and serotype were determined by using specifi c antisera (Lee Laboratories, Franklin Lakes, NJ, USA). Saline controls were included to detect autoagglutination. Biotypes were determined by PCR of specifi c regions of the biotyping (tcpA) gene (3).
Broth microdilution was run on 122 isolates from Haiti and on all isolates from non-Haiti locations. Broth microdilution methods were performed according to the manufacturer's instructions by using CAMPY and CMV1AGNF Sensititer Plates (both from Trek Diagnostics, Cleveland, OH, USA) with 2 modifi cations: we transferred 50-μL (CAMPY plates) and 20-μL (Sensititer plates) volumes from the suspension to the broth and used Mueller-Hinton broth without blood for the CAMPY panel. Antimicrobial drug sensitivity results from Sensititer plate testing were available for amoxicillin/ clavulanate, ampicillin, azithromycin, chloramphenicol, ciprofl oxacin, nalidixic acid, streptomycin, sulfi soxazole, tetracycline, and trimethoprim/sulfamethoxazole. Both panels were inoculated at concentrations of 5 × 10 4 and 5 × 10 5 CFU/mL. Internal quality controls included those used for disk-diffusion testing plus ATCC 29212 (Enterococcus faecalis). Where available, specifi c interpretive criteria for V. cholerae were used (5). For drugs with no criteria, interpretation was guided by using Clinical and Laboratory Standards Institute criteria for Enterobacteriaceae or consensus breakpoints used by the National Antimicrobial Resistance Monitoring System (6).

Pulsed-Field Gel Electrophoresis
Isolates were analyzed by using a PulseNet standardized pulsed-fi eld gel electrophoresis (PFGE) protocol for V. cholerae (7) with Sfi I and NotI restriction enzymes (Roche Molecular Biochemicals, Indianapolis, IN, USA). Images of restriction patterns were analyzed by using BioNumerics software (Applied Maths, Inc., Austin, TX, USA). Gel patterns were compared with others in the National PulseNet V. cholerae database (www.cdc. gov/pulsenet/whatis.htm) and the PulseNet International V. cholerae database (www.pulsenetinternational.org/ protocols/Pages/vcholeraedatabase.aspx).

Detection of Virulence and Species-specifi c Genes, PCR, and Sequencing
We amplifi ed DNA from boiled lysates for 30 cycles in a multiplex PCR to detect cholera toxin gene subunit A (ctxA) (8) sequences, biotyping genes (tcpA) (3), and species-specifi c genes ompW (9), and toxR (10) by using the primers and methods described. Primer pair smp-F and smp-R (11) was used to amplify the seventh pandemicspecifi c gene VC2346 at cycling conditions of 93°C for 15 min; 35 cycles of 92°C for 40 s, 52°C for 1 min, and 72°C for 1.5 min; followed by 72°C for 7 min. The primers and cycling conditions for PCR amplifi cation of biotype-specifi c repeat sequence transcriptional regulator (rstR) alleles were used as described (12). We sequenced 13 isolates from Haiti and from the strain from the US Gulf Coast by using the same primer set to verify results.
The complete coding region of the cholera toxin gene ctxAB was amplifi ed with fl anking primer pair primers S86 (ctxAB 1 ) and S87 (ctxAB 2 ) (13,14). A step-down PCR to avoid nonspecifi c amplifi cation was run as follows: 15 min at 93°C; 11 cycles of 92°C for 40 s, decrementing by 1°C from 60°C to 50°C for 1 min, 72°C for 1.5 min; followed by 30 cycles of 92°C for 40 s, 50°C for 1 min, 72°C for 1.5 min; with a fi nal extension at 72°C for 7 min. Sequences were determined with amplifi cation primers and 2 internal primers, CTX93-F and CTX618-R (15). The complete tcpA gene was amplifi ed and sequenced with primer pair tcpH1 and tcpA4 (16). PCR cycling conditions were 93°C for 15 min followed by 35 cycles of 92°C for 40 s, 52°C for 1 min, and 72°C for 1.5 min, with a fi nal extension at 72°C for 7 min.
We purifi ed ctxAB, tcpA, and rstR amplicons by using the QIAquick PCR Purifi cation Kit (QIAGEN, Inc., Valencia, CA, USA). Sequencing was performed on the Applied Biosystems 3730 DNA analyzer with POP-7 polymer and a 50-cm capillary array (all from Life Technologies, Carlsbad, CA, USA) following the manufacturer's instructions. Chromatograms were assembled by using Lasergene SeqMan Pro version 8.0.2 (www.dnastar.com). Sequences were aligned with other V. cholerae sequences by using ClustalW (www.clustal. org) in MEGA4 software (17) and trimmed in-frame for analyses. The full genome of 1 V. cholerae strain from Haiti (2010EL-1786 [ATCC BAA-2163]) and partial genomic regions (integrated conjugative elements and cholera toxin phage) of other isolates were sequenced as described (18).

Isolates
We identifi ed all 149 isolates from Haiti as V. cholerae, serogroup O1, serotype Ogawa, biotype El Tor, containing species-specifi c genes ompW and toxR. All contained the cholera toxin gene ctxAB. All isolates from countries other than Haiti selected for comparison were confi rmed as V. cholerae, serogroup O1, biotype El Tor, ctxAB positive. Serotypes varied among geographic regions ( Table 2).

Antimicrobial Drug-Susceptibility
Disk-diffusion testing demonstrated that all strains from Haiti were resistant to furazolidone, nalidixic acid, streptomycin, sulfi soxazole, and trimethoprim/ sulfamethoxazole. Strains were susceptible to tetracycline and either susceptible or intermediately susceptible to ampicillin and chloramphenicol. Broth microdilution testing of the isolates from Haiti showed similar results (furazolidone was not tested by broth microdilution) plus decreased susceptibility to ciprofl oxacin (MIC range 0.25-1.0 μg/mL). These isolates were susceptible to ampicillin, azithromycin, and chloramphenicol, except for 2 isolates that gave intermediate results for chloramphenicol. Antimicrobial drug susceptibility among strains tested from countries other than Haiti varied; 19 showed the same antimicrobial drug resistance as the Haiti outbreak strains, including isolates from Nepal, Cameroon, South Africa, Oman, and India. Two isolates were similar to the strains from Haiti, but they did not display decreased susceptibility to ciprofl oxacin. Twenty-seven isolates from Africa and the Middle East showed the outbreak resistance pattern, but they were susceptible to nalidixic acid and fully susceptible to ciprofl oxacin. Two isolates from Peru and the isolate from the US Gulf Coast were susceptible to all antimicrobial drugs tested.

PFGE Genotypes
Two-enzyme PFGE analyses with Sfi I and NotI identifi ed a predominant Haiti outbreak pattern combination, KZGS12.0088/KZGN11.0092, in 123 (82.6%) of 149 isolates tested (Table 3). This primary pattern combination was detected in isolates from all 10 departments in Haiti, in isolates from the Dominican Republic, and in isolates from travelers returning to the United States from Hispaniola (data not shown). There were 10 PFGE pattern variations, defi ned as >1 band difference in number or size from the    combination pattern seen in isolates from Haiti. One colony from inside the chloramphenicol zone of inhibition also displayed a variant pattern (data not shown).

Virulence and Species-specifi c Genes, PCR Results, and Sequences
The sequencing results for ctxB and tcpA and PCR results for rstR and the VC2346 gene are shown in Table 5. We sequenced the complete 1,148 bp of the ctxAB operon for 107 strains. The ctxB gene sequence of all 56 isolates from Haiti that we tested matched the B-7 allele fi rst seen in an outbreak in 2008 in Orissa, India (20) ( Table 5). The B-7 allele was also contained in 1 isolate from Cameroon, 3 of 6 from India, and 1 from Nepal. The B-7 allele has a single-nucleotide polymorphism (SNP) at nt 58 relative to the classical B-1 allele ("genotype 1" [21]), resulting in substitution of asparagine for histidine at aa 20, which is adjacent to the B subunit proteolytic cleavage site between aa 21 and aa 22. The remaining isolates from non-Haiti locations carried the B-1 allele, with the exception of the strains from Peru, which carried the B-3 allele (genotype 3, reported by Olsvik et al. [21]) that historically has been associated with the El Tor biotype.
A 1,234-bp sequence, which included 675 bp of the entire tcpA gene, matched sequences from CIRS 101, a V. cholerae serogroup O1, biotype El Tor strain (22) that has an integrated El Tor type phage (CTXΦ) El Tor but expresses the classical ctxB-1 allele. This combination is considered to be an altered or atypical El Tor. The tcpA sequence of CIRS 101 has a novel SNP at nt position 266 (aa 89), which differentiates it from typical El Tor strains (22). All 56 tested isolates from Haiti had this novel SNP in the tcpA sequence, which we have designated the tcpET CIRS allele. Isolates from Africa and southern Asia also had the tcpET CIRS allele ( Table 5).
The rstR El Tor allele was found in all isolates tested except the strain from the Gulf Coast of the United States. All isolates tested, except the strain isolate from the US Gulf Coast, contained the VC2346 gene and are therefore confi rmed as seventh pandemic V. cholerae (11). As shown by whole genome sequencing, isolates with the PFGE pattern KGZS12.0088/KGZN11.0092 (containing the ctxB-7/tcpET CIRS alleles from Cameroon, India, and Nepal) are most closely related to the strain from Haiti. Details of whole genome sequencing are provided elsewhere (18).

Discussion
We characterized the strains in this study by using basic phenotypic markers and genotypic tests that in other studies have been shown to be useful for characterizing V. cholerae outbreak strains. Epidemic cholera strains cause human illness by expression of specifi c genes that enable V. cholerae to exist in the environment long enough to be ingested, overcome host immunity, colonize the intestinal tract, and produce cholera toxin in the host (13,(23)(24)(25). The horizontal acquisition of genes and expression of the transcriptional co-regulated pilus (located on Vibrio pathogenicity island 1) and CTXAB (carried on CTXΦ)  determine primary virulence (9,23,24). CTXΦ has 2 regions, a core that includes ctxAB and the RS2 region that carries phage replication genes, such as rstR. Allelic variations in ctxB, tcpA, and rstR can be useful markers to characterize CTXФ types and track cholera strains. Whereas nucleotide sequence analysis of the acquired virulence genes ctxAB, rstR, and tcpA differentiates among V. cholerae O1, the VC2346 gene is part of the nontransferable genomic backbone that identifi es seventh pandemic V. cholerae O1 El Tor strains currently in circulation. The adaptability of V. cholerae as a pathogen is facilitated by extensive genetic diversity driven by acquisition and recombination of various genetic elements. Community defense mechanisms, mediated by cellular signaling, such as quorum sensing in biofi lms, gives the species additional resiliency against changing environmental conditions (26). All V. cholerae have these attributes, but only O1 and O139 serogroups currently have pandemic potential. The El Tor biotype was fi rst identifi ed 100 years ago, and 50 years ago it emerged from Indonesia to begin the ongoing seventh pandemic, displacing the classical biotype of the fi fth and sixth pandemics because of its superior ability to survive in the environment, increased frequency of asymptomatic carriers, and more effi cient transmissibility (27).
Genes encoding hemolysis, such as hlyA, may be a virulence factor in some Vibrio spp., and in the past, hemolysis patterns were used to distinguish biotypes (28). Classical biotype strains carry deletion mutations in the hlyA locus, and many El Tor strains produce no detectable hemolysis on blood agar plates. Our fi ndings confi rm observations that hemolysis is not a reliable marker for strain discrimination. Of interest, hemolytic colonies were often observed within a streak predominated by nonhemolytic colonies.
The predominant KZGS12.0088/KZGN11.0092 Sfi I/ NotI PFGE pattern in the strains from Haiti is a relatively new subtype that was fi rst seen in 2005 in the PulseNet USA database in isolates from travelers returning from India. Although the isolates from Haiti show diversity in their PFGE subtyping patterns, the constancy of the main pattern coupled with virulence genotyping results indicate a high clonality of the outbreak strains, which is consistent with a point-source introduction. Such PFGE diversity, similar to what we observed, has been noticed during outbreaks in the Bengal region (T. Ramamurthy, pers. comm.). This primary KZGS12.0088/KZGN11.0092 pattern and its close variant, KZGS12.0089/KZGN11.0092, were found in strains from Afghanistan, Cameroon, India, Nepal, Oman, Pakistan, and South Africa. Among the strains from Haiti, we identifi ed 4 new NotI restriction patterns, 5 new Sfi I patterns, and 9 new Sfi I/NotI combinations, a fi nding suggestive of continuing evolution of the outbreak strain. The PFGE pattern combination KZGS12.0019/KZGN.0092 was commonly seen in serotype Inaba strains originating from East Africa and the Middle East and was recently seen in isolates from Togo; in this study, the pattern was associated with the ctxB-1 allele. Although the strains from Haiti are genetically an El Tor biotype, they contain the classical ctxB-7 allele. This allele was fi rst identifi ed in 2007 in strains from an outbreak in Orissa, India (20). The appearance of a classical ctxB gene in El Tor strains is not unprecedented. In the early 2000s, hybrid El Tor strains emerged carrying CTXΦ Classical with the ctxB gene of the classical biotype, which is thought to cause more severe clinical disease (29). These hybrid strains most likely arose through horizontal transmission of CTXΦ Classical , and the resulting genotypes with the classical ctxB1 allele have spread to Asia and Africa. In this study, we show that V. cholerae carrying the ctxB-7 allele are also disseminating globally.
The transcriptional co-regulated pilus serves a dual role as the major intestinal colonization factor and CTXΦ receptor. All tested isolates from Haiti had tcpA sequences with an SNP at nt 266 (tcpET CIRS ), an allele previously reported in strain CIRS 101 from Bangladesh (22), which is an El Tor biotype that produces a classical toxin yet carries CTXΦ El Tor ; this SNP produces a distinct allele that distinguishes it from classical and typical El Tor. In the present study, isolates from Haiti, Africa, and southern Asia carried tcpET CIRS (Table 5). Our results show that the tcpET CIRS allele is also spreading globally, although not in tandem with the ctxB-7 allele because the tcpET CIRS allele was also found in isolates with the ctxB-1 allele from Afghanistan, India, Pakistan, South Africa, and Sri Lanka (Table 5). Our fi ndings agree with those from a recent study suggesting a close relationship between 2 isolates from Haiti and isolates from Southeast Asia (30). However, we also observed a relationship between isolates from Haiti and Africa.
The fi nding of these ctxB-7/tcpET CIRS isolates in Cameroon, India, Nepal, and now Hispaniola is not surprising, given the ease of international travel; we are unable to identify the origin of the Haiti strains because of geographic and temporal limitations in our culture collection. In particular, our fi ndings do not rule out the presence of ctxB-7/tcpET CIRS isolates in countries not represented in our collection. Whole-genome sequencing results confi rmed the genetic relationship of these isolates from Haiti, Cameroon, India, and Nepal (18). More extensive whole-genome sequencing studies and other subtyping methods, such as multiple-locus variable-number tandem repeat analysis, hold promise for providing a better understanding of the relationships between isolates.
The strain from Haiti is distinct from the isolate from the US Gulf Coast. The isolate from the Gulf Coast was characterized as KZGS12.0055/KZGN11.0029, ctxB-1, tcpET, rstR Classical and negative for VC2346. The isolate from Haiti is also not related to the isolates from Peru from the 1991 Latin America outbreaks, which were characterized as KZGS12.0114/KZGN11.0033, ctxB-3, tcpET, and rstR El Tor .
The strain from Haiti has the core characteristics of the seventh pandemic El Tor clone. Our fi ndings support the widespread observation that the typical El Tor strain, which started the seventh pandemic, is gradually being replaced by El Tor isolates with classical cholera toxin subunits.
The isolates from Haiti and those from other regions displayed a consistent resistance phenotype, with resistance to the clinically relevant antimicrobial drugs trimethoprim/sulfamethoxazole and sulfi soxazole but susceptibility to other primary antimicrobial drug options, including doxycycline and azithromycin. A discussion of integrating conjugative and other mobile genetic elements that can potentially mediate transfer of antimicrobial drug resistance is provided elsewhere in this issue (31). Development of additional antimicrobial drug resistance, particularly to doxycycline and macrolides, remains a serious clinical threat, and Laboratoire National de Santé Publique and CDC continue to monitor for the emergence of such resistance.
The strains from Haiti are fully virulent and contain all the genes necessary for orchestrating the expression of Vibrio spp. virulence factors. These strain characteristics, coupled with the sudden and explosive course of the 2010 outbreak, are consistent with an introduction of this strain into a vulnerable population at a single point in time.
The adaptive immunity of the local population as well as climate variations will further drive bacterial evolution; for example, it will not be surprising to observe a gradual switch over time from the Ogawa to the Inaba serotype as population immunity to Ogawa rises, as has been reported during several outbreaks (23). The primary PFGE patterns continue to diverge as the infections continue, likely refl ecting interactions with the host immune system and between environmental and epidemic populations of bacteria networking in complex ways. Regardless, rapid diagnosis and continuing public health control of the current outbreak in Haiti as well as future outbreaks is paramount for limiting sickness and death, and intensive studies using a variety of basic science, diagnostic, and epidemiologic tools will remain useful for reducing the overall global impact of cholera.
Dr Talkington is a team lead in the Enteric Diseases Laboratory Branch, Division of Foodborne, Waterborne, and Environmental Diseases, National Center for Emerging and Zoonotic Infectious Diseases, Centers for Disease Control and Prevention. Her research interests include outbreak detection and diagnosis of infectious diseases. This manuscript is dedicated to all health professionals in Haiti.