ICDDR,B maintains a 2% surveillance system at its Dhaka Hospital, in which data from every 50th patient treated at the hospital is collected; these data include clinical information and biologic specimens. We used these data to extrapolate the overall numbers of patients with cholera; specimens from these patients were used in the bacteriologic studies described.

A total of 63 V. cholerae O139 isolates obtained from the recent cholera epidemic were analyzed. Seven strains of O139 vibrios isolated in India between 1992 and 1996, 17 strains of V. cholerae O139 isolated in Bangladesh between 1993 and 1997, and 2 strains isolated in Thailand in 1998 were also included in the study for comparison with the recent epidemic strains. Strains of the recent epidemic were isolated from stools of cholera patients who attended the treatment center of ICDDR,B located in Dhaka during March and April 2002. Stool samples were processed in the laboratory within 2 h of collection for the isolation of V. cholerae. Stools were initially streaked on thiosulphate-citrate-bile-sucrose (Becton, Dickinson and Co., Sparks, MD) agar plates for selection and presumptive identification of V. cholerae. All strains were subsequently examined by biochemical and serologic tests using standard methods (9). Strains were stored in sealed deep nutrient agar at room temperature until used for this study. Details of the strains are shown in Table.

Presence of tcpA genes specific for the classical and El Tor biotypes was determined by using a multiplex PCR assay, as described previously (10). PCR assays for the tcpI and acfB genes have been described previously (6). Presence of classical, El Tor, and Calcutta type rstR genes of CTX phage were also determined with PCR by using specific primers derived from the published sequence of the respective genes. Three different forward primers for rstR^class, rstR^ET, and rstR^Calc with sequences 5′-CTTCTCATCAGCAAAGCCTCCATC, 5′-GCACCATGATTTAAGATGCTC, and 5′-CTGTAAATCTCTTCAATCCTAGG, respectively, were used with a common reverse primer (5′-TCGAGTTGTAATTCATCAAGAGTG) to amplify the respective rstR genes. Presence of the rstC gene was also determined by a PCR assay described previously (11). All primers were synthesized commercially by Oswel DNA Service (University of Edinburgh, Edinburgh, UK). The expected sizes of the amplicons were ascertained by electrophoresis in agarose gels, and the identity of each PCR product was further verified by Southern blot hybridization.

The gene probes used in this study included a 0.5 kb EcoRI fragment of pCVD27 (12) containing part of the ctxA gene and a 2.1 kb SphI-XbaI fragment of pCTX-Km containing the entire zot and ace genes and part of orfU (13). The toxR gene probe was a 2.4-kb BamHI fragment of pVM7 (14). The rstR^ET probe was a SacI-XbaI fragment of pHK1 (15). The rRNA gene probe was a 7.5-kb BamHI fragment of the Escherichia coli rRNA clone pKK3535 described previously (16). The O139-specific DNA probe 2R3 was a 1.3-kb EcoRI fragment of pCRII-A3 (17,18), and the SXT probe was a NotI fragment of pSXT1 (19). PCR-generated amplicons of the rstR genes of classical, El Tor, or Calcutta type CTX prophage were also used as probes whenever appropriate.

For preparation of Southern blots, total cellular DNA was isolated from overnight cultures as described previously (20). Five-microgram aliquots of the DNA were digested with appropriate restriction enzymes (Bethesda Research Laboratories, Gaithersburg, MD), electrophoresed in 0.8% agarose gels, blotted onto nylon membranes (Hybond, Amersham Biosciences, Uppsala, Sweden), and processed by using standard methods (21,22). The probes were labeled by random priming (23) using a DNA labeling kit (Bethesda Research Laboratories) and α-³²P‑deoxycytidine triphosphate (3,000 Ci/mmol, Amersham Biosciences). Southern blots were hybridized with the labeled probes at 68°C and washed under stringent conditions as described previously (6,8). Autoradiographs were developed from the hybridized filters with Kodak X‑Omat AR x‑ray film (Eastman Kodak Co., Rochester, NY) at –70°C.

All V. cholerae isolates were tested for resistance to antimicrobial drugs by using the method of Bauer et al. (24) with standard antibiotic disks (Oxoid Ltd., Basingstoke, Hampshire, UK) at the following antibiotic concentrations (μg/disc): ampicillin, 10; chloramphenicol, 30; streptomycin, 10; tetracycline, 30; trimethoprim‑sulfamethoxazole, 1.25 and 23.75, respectively; kanamycin, 30; gentamicin, 10; ciprofloxcin 5; norfloxacin 10, and nalidixic acid, 30.

We noted a marked increase in cholera cases associated with V. cholerae O139 from March to May 2002 (Figure 1). The highest number of cholera patients admitted to the hospital was in March; 69.8% of these cases were attributed to V. cholerae O139, compared to 30.2% of cases caused by the El Tor biotype of V. cholerae O1. Cholera attributable to V. cholerae O139 occurred with similar frequencies in men and women, similar to those infected with O1 strains. From January 2001 to June 2002, a total of 91 (32%) of 282 of case-patients infected with O1 cholera were <5 years of age, but 15 (15%) of 115 of those infected with O139 were <5 years of age (p<0.001). During the same period, 48% of those infected with V. cholerae O1 were >15 years of age, while 76% of those infected with O139 were >15 years of age (p<0.001).

The rRNA gene restriction patterns using BglI consisted of 10 to 14 bands between 11 kb and 1.6 kb in size (Figure 2). The 89 analyzed strains belonged to four different ribotypes (B-I to B-IV). All 63 recently isolated O139 strains produced identical restriction patterns of their rRNA genes and belonged to ribotype B-II. Analysis of the rstR gene showed that O139 strains isolated from 1992 to 1998 carried El Tor type CTX^ET prophage, whereas the recent epidemic strains carry the Calcutta type CTX^Calc prophage in addition to the CTX^ET prophage (Figure 3). All strains were positive for tcpA, tcpI, acfB, toxT, ctxA, zot, and ToxR genes, as well as for the O139-specific genomic DNA in DNA probe or PCR assays.

All strains isolated from the recent epidemic were resistant to nalidixic acid and were susceptible to ampicillin, tetracycline, gentamicin, chloramphenicol, ciprofloxacin, norfloxacin, streptomycin, trimethoprim, and sulfamethoxazole. In these strains, the SXT element, which encodes resistance to streptomycin, sulfamethoxazole, and trimethoprim, carried a deletion of an approximately 3.6-kb region.

The emergence of the O139 serogroup has provided a unique opportunity to witness the epidemiologic changes associated with the displacement of an existing serogroup by a new emerging one and thus provides new insights into the epidemiology of the disease. All 63 recently isolated O139 strains produced the identical restriction pattern of their rRNA genes. This restriction pattern has been previously designated as ribotype pattern B-II (6–8) and was first detected among epidemic V. cholerae O139 strains that emerged in 1992 and 1993. Cholera epidemics during 1992 to 1993 in India and Bangladesh that were associated with the first appearance of V. cholerae O139 were caused by strains belonging to two different ribotypes, designated as B-I and B-II. Since then, several new ribotypes of O139 vibrios have been detected which were associated with localized outbreaks during 1995 to 1996 or sporadic cases (8). The results suggest that strains of the recent epidemic were clonal and were derived from one of the initial clones of V. cholerae O139. We therefore investigated possible genetic changes sustained by this strain during the nearly 9 years since major epidemics were caused by strains of this ribotype.

In V. cholerae, the genes encoding cholera toxin (ctxAB) are part of the CTX prophage (25). A typical CTXΦ genome has two regions: core and the RS2. The 4.5-kb core region comprises several open reading frames including ctxAB, zot, ace, orfU, and encodes CT as well as the functions that are required for the virion morphogenesis; by contrast, the 2.5-kb RS2 region encodes the regulation, replication, and integration functions of the CTXΦ genome (26). Previous studies have described the existence of at least three widely diverse repressor genes (rstR genes) carried by different CTX phages (i.e., CTX^ETΦ, CTX^classΦ, and CTX^CalcΦ) (27,28). This diversity of rstR constitutes the molecular basis for heteroimmunity among CTX phages, which are otherwise genetically similar. We examined the CTX prophage in the recent and previously isolated O139 strains with specific probes. Analysis of the rstR gene carried by the recent epidemic strains showed that, unlike the O139 strains of 1993, which carried multiple copies of an El Tor type CTX_ET prophage, the new O139 strains carry at least one copy of the Calcutta type CTX^Calc prophage in addition to the CTX^ET prophage. As a result of heteroimmunity, toxigenic classical strains of V. cholerae O1 are known to be infected by CTXΦ isolated from El Tor biotype strains; toxigenic El Tor strains are resistant to further infection by the same phage. Similarly, strains carrying an El Tor type CTX prophage can be superinfected by the Calcutta type CTX phage (29). Therefore, the new epidemic strains appear to have arisen by acquisition of a Calcutta type CTX phage by strains that originally harbored only El Tor type CTX prophage, since the new strains carry both prophages (Figure 3). What determines the reemergence of particular epidemic strains is not clear, but this study clearly shows changes in the CTX genotype attributed to the acquisition of a new CTX phage by the O139 strains associated with the recent epidemic.

V. cholerae O139, which emerged during 1992 and 1993, was sensitive to tetracycline and showed a trend of increased resistance to trimethoprim-sulfamethoxazole (SXT) and streptomycin. This resistance was mediated by a ~99-kb self-transmissible transposon-like element (SXT constin) encoding resistance to sulfamethoxazole, trimethoprim, and streptomycin, the resistance genes being clustered together in a 9.4-kb region (19). In the present study, all strains isolated from the recent epidemic were found to be susceptible to SXT and streptomycin (Table). To identify the genetic changes associated with the observed SXT sensitivity, we used a cloned SXT gene probe to study restriction fragment length polymorphism in the SXT transposon. Three different BglI restriction patterns (patterns A–C) of the SXT element were observed among the O139 strains tested (Figure 4). Strains producing pattern A and B were resistant to SXT and streptomycin and included strains isolated between 1992 and 1996, whereas all strains from the recent epidemic produced pattern C and were susceptible to all the three antibiotics. Further analysis of the restriction patterns suggests that the restriction site heterogeneity possibly occurred as a result of a deletion of approximately a 3.6-kb region of the SXT element in strains that were sensitive to SXT and streptomycin. The deletion in the SXT element associated with sensitivity to SXT and streptomycin was first detected in strains of ribotype B-III isolated from an outbreak in Bangladesh in 1997 (6). In keeping with the observation in Bangladesh, comparison of the antibiotic resistance patterns between the O139 strains isolated during 1992 and 1993 and those isolated in 1996 and 1997 in India also showed that the later strains were susceptible to SXT, unlike the O139 strains from 1992 and 1993 (30). However, in contrast to the previously isolated O139 strains, all O139 strains isolated from the recent epidemic were resistant to nalidixic acid.

Several previous studies have shown that the O139 serogroup of V. cholerae has been undergoing rapid genetic changes (6–8) since its first emergence. We speculate that the observed changes may have provided increased fitness to strains of this serogroup in some unexplained way to survive in competition with the existing seventh pandemic strain of V. cholerae O1 and establish itself as the etiologic agent of a possible eighth pandemic. The transient disappearance of the O139 serogroup in Bangladesh and repeated reemergence associated with somewhat altered genetic or phenotypic properties seem to support this speculation. Our study demonstrated the reemergence of V. cholerae O139 strains belonging to a previously described ribotype which has sustained at least three major genetic and phenotypic changes. These changes include the acquisition of a new CTX prophage, deletion in the SXT element associated with reversion of drug resistance phenotype against SXT and streptomycin, and development of nalidixic acid resistance.

The recent epidemic strains were otherwise similar to previously described O139 strains, including possession of the TCP pathogenicity island, as evidenced by the presence of tcpA, tcpI, and acfB genes; the virulence regulatory genes, toxT and toxR; and the O139–serotype-specific DNA. The role of environmental and host factors that contribute to the emergence of new strains associated with epidemic outbreaks is not clearly known. In the present study, all strains isolated from the recent cholera outbreak belonged to the same ribotype and were genetically and phenotypically identical, suggesting that the recent outbreak in Bangladesh probably started from a point source and may have coincided with the acquisition of one or more critical new properties by a previously existing V. cholerae O139 strain. Clearly these properties included the acquisition of the Calcutta Type CTX prophage. Previous studies showed that O139 strains prevailing in Calcutta during 1996 carried this prophage (29,31,32), which might have contributed to the dissimilar incidence of O139 cholera in Calcutta and Dhaka during that period (33). How the initial enrichment of V. cholerae occurred before the initiation of an epidemic is not clear. We speculate that a critical factor for the recent reemergence of O139 vibrios might have been the development of nalidixc acid resistance. Identifying the first index case of the present cholera epidemic is not possible. A spontaneous nalidixic acid–resistant V. cholerae O139 strain may have been enriched in a patient undergoing nalidixc acid therapy, leading to the eventual spread of the organism. This is certainly possible in view of the widespread use of nalidixic acid in Bangladesh as a drug to treat other gastroenteritis, including shigellosis. The emergence of V. cholerae O139 has received global attention not only as the first non–O1 V. cholerae capable of causing epidemic outbreaks but also because of the rapid genetic re-assortment undergone by strains of this new serogroup. Our study shows yet another set of genetic and phenotypic changes in O139 vibrios and their association with an epidemic of cholera in Bangladesh. These results emphasize the need for continuing molecular epidemiologic surveillance of V. cholerae in Bangladesh and adjoining areas.