INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Cholera is a severe, infectious diarrheal disease caused by the gram-negative bacterium Vibrio cholerae. The principal virulence factor of V. cholerae is cholera toxin (CT), a potent, A-B-type exotoxin that ADP-ribosylates proteins within intoxicated intestinal epithelial cells (19). The CT produced by V. cholerae during the organism's colonization of its host's small intestine accounts for a majority of the symptoms that characterize the disease process (11). In 1996, Waldor and Mekalanos discovered that the genes encoding CT (the operon ctxAB) are not integral components of the V. cholerae genome, but instead are elements of the genome of a filamentous bacteriophage, CTX, that specifically infects V. cholerae (22). Infection of V. cholerae by CTX is frequently followed by integration of the phage genome into the V. cholerae genome, yielding a stable lysogen. Like the filamentous phages of Escherichia coli, CTX can also replicate as a plasmid, and it does so in bacterial strains lacking appropriate integration sites; however, most if not all natural isolates of V. cholerae containing ctxAB contain integrated phage DNA (13).

Integration of the CTX genome is site specific, but the integration sites and the prophage arrays they contain differ between the two biotypes of V. cholerae O1. Within El Tor biotype strains, which have been used for most analyses of phage genes, CTX prophages are found at a chromosomal site known as attRS (16). Integration of CTX DNA into attRS occurs via recombination between an 18-bp sequence (originally designated the end repeat [ER]) in the phage genome and a nearly identical sequence in attRS (16). Some El Tor strains contain a single CTX prophage, while many others contain several in tandem (13). The length of this prophage array can fluctuate (generally expanding) both during the course of an infection and within laboratory cultures, in response to the bacterium's environment (6, 13). We have found that the CTX prophages in El Tor strains generally give rise to infectious phage particles (10).

V. cholerae strains of the classical biotype, which were the dominant cause of epidemic cholera until 1961 when they were replaced by El Tor strains, contain a more complex arrangement of CTX genes. Classical strains have two integration sites, each of which contains a single CTX prophage (13). One site is identical to the attRS integration site found in El Tor strains. The second site has not been well characterized, but it has been localized to a different chromosome than attRS (21). Surprisingly, neither prophage within classical strains apparently gives rise to phage particles (unpublished data). In addition, the DNA of CTX derived from El Tor strains does not integrate following CTX infection of classical strains. Instead, phage DNA replicates as a plasmid in classical strains, rather than recombining into either of the two classical integration sites (22).

The CTX genome is composed of two regions (Fig. 1) (6, 16). The core region contains the genes encoding CT and genes required for phage morphogenesis, including genes that are thought to encode major and minor phage coat proteins and a protein that aids in phage assembly and secretion (24). Some of these morphogenesis genes are similar to genes of E. coli filamentous phages, such as M13 and fd (22). In contrast, the three genes of the other CTX region, RS2, are not similar to those of E. coli filamentous phages. Their products control phage replication and site-specific integration (16, 23). RstA is required for phage DNA replication, RstB is required for site-specific integration, and RstR is a repressor of rstA expression (9, 23). RS2 also contains two intergenic regions: ig-1 and ig-2. Ig-2 appears to encompass the rstA promoter and the RstR operator; no role has yet been established for ig-1. These three genes and the intergenic regions are also components of a related genetic element, RS1, which is found adjacent to CTX prophages in many V. cholerae strains (23).

View larger version (65K): [in this window] [in a new window]   FIG. 1.   Structure and sequence of CTX prophages within AS207, an O139 Calcutta strain of V. cholerae. (A) Within the AS207 chromosome, an RS1 element and a CTXET prophage are followed by two CTXcalc prophages (shown in light grey). The Calcutta prophage is structurally similar to the previously described CTX prophages within El Tor and classical strains, which contain two major domains known as RS2 and core. RS2 contains three genes (rstA, rstB, and rstR) whose transcriptional orientation is indicated by the arrows. The black triangles represent attRS-ER sequences. (B) Alignment of the nucleotide sequences of rstR, ig-2, and parts of rstA and ig-1 from Calcutta, El Tor, and classical prophages. The arrows underneath the sequences depict the ORFs that encode the three variants of RstR.

We recently performed a detailed comparison of the RS2 regions from classical and El Tor CTX prophages (9). We found that rstB and the coding sequence of rstA are highly conserved between the biotypes (94% nucleotide identity), but that rstR and ig-2 (the rstA promoter) sequences diverge considerably (44 and 61% nucleotide identity, respectively). Due to the variations in the sequences and binding sites of both the repressor proteins, each RstR is a biotype-specific repressor of its cognate rstA (9). That is, expression of the classical rstA (rstA^class) reporter construct rstA^class-lacZ is repressed by classical, but not El Tor, RstR, and similarly, expression of the El Tor reporter construct rstA^ET-lacZ is repressed by El Tor, but not classical, RstR. This repression allows integrated phages to inhibit replication of newly infecting phages of the same biotype, thereby conferring immunity to secondary infection. However, the production of RstR^class by the prophages within classical strains of V. cholerae does not prevent infection of these strains by a Kn-marked El Tor CTX, suggesting that classical CTX lysogens are heteroimmune to the El Tor CTX (9).

Until 1997, only these two forms, El Tor and classical, of CTX prophages had been identified. However, analyses in 1997 of novel O139 strains responsible for severe outbreaks of cholera in Calcutta, India, revealed that they contained prophages with atypical restriction endonuclease sites (3, 20). We reported previously that these Calcutta strains contain sequences within RS2 quite dissimilar to both the classical and El Tor RS2s (8). In this study, we present further analyses of the Calcutta CTX prophage, especially of its RS2 domain. We show that RstR^calc, despite a size and structure dramatically different from previously described repressors, also functions as an allele-specific repressor, capable of repressing rstA^calc expression. In addition, we demonstrate that, unlike the classical prophages, the Calcutta prophage generates infectious phage particles. Thus, CTX^calc, as well as the El Tor CTX (denoted here as CTX^ET but in earlier works simply as CTX), can transmit ctxAB to nonpathogenic strains. Lysogens of CTX^calc have immunity to superinfection with CTX^calc, similar to the immunity provoked by CTX^ET. Finally, we have used CTX^calc and CTX^ET to investigate the potential genesis of multiply lysogenized V. cholerae strains, such as those identified in Calcutta.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Nucleotide sequence of the CTX^calc prophage RS2 region. RS2 from the CTX^calc prophage was amplified from strain AS207 using PCR. In this PCR, a forward primer within ctxB (5' GCGATTGAAAGGATGAAGGATAC 3') and a reverse primer within cep (5' AACCCCGAGTGAAAGCGTG 3') allows amplification of the CTX^calc prophage RS2 region from strains such as AS207, which contain Calcutta prophages downstream of a single El Tor prophage (Fig. 1) (8). This PCR product was cloned into pCR2.1 (Invitrogen, Carlsbad, Calif.) to generate pHK268, which was used as a template for dye terminator cycle sequencing, using an Applied Biosystems 373A DNA sequencer. The BLAST programs (1) were used to compare the Calcutta RS2 nucleotide sequence to sequences in the GenBank databases. Potential repressor and helix-turn-helix (hth) DNA binding domains were evaluated by using the matrix of Dodd and Egan (4) and the motif analysis program fingerPRINTScan (5).

Bacterial strains and culture conditions. Bacterial strains and plasmids used in this study are described within Table 1. All bacteria were cultured in Luria-Bertani broth (14) at 37°C unless otherwise noted. Antibiotics were used at the following concentrations: ampicillin (AMP), 50 µg/ml (V. cholerae); AMP, 100 µg/ml (E. coli); KAN, 50 µg/ml; streptomycin (STR), 200 µg/ml; chloramphenicol (CMP), 15 µg/ml) (E. coli). Arabinose (ARA)-induced cultures contained 0.05% ARA, and sucrose-resistant (Sc^r) clones were selected on 10% sucrose.

Plasmid and strain construction. pBD40, which contains an rstA^calc-lacZ fusion, was constructed by first amplifying the rstA^calc promoter and part of the rstA^calc coding sequence with primers rstA^calc proF (5' GATGTTTGTTTTGGTCTCGATTACCG 3') and rstA proR (5' TGAAGCATAAGGAACCGACC 3'). Next, the PCR product was cloned into the TA cloning vector pCRII-TOPO (Invitrogen). An XbaI/HindIII fragment containing the insert was then ligated to XbaI/HindIII-digested pCB192 (18) to generate pBD40. To construct pBD87, a PCR product containing rstR^calc was first amplified with primers rstR-11 (5' AATAGGGCTTTACGGAATC 3') and rstR-10 (5' TGTTTGGAAATCAAGAGAGG 3'). Following subcloning of this product into pCRII-TOPO, a KpnI/XbaI fragment containing the insert was ligated into KpnI/XbaI-digested pBAD33 (7).

Phage transduction assays. To transfer CTX^calc-Kn from AS207 CTX^calc::Kn to O395, 50 µl of agglutinated O395 (grown at 30°C to induce expression of the CTX receptor, TCP [22]) was mixed with 50 µl of filtered supernatant from a log-phase culture of AS207 CTX^calc::Kn. The mixture was shaken gently at room temperature for 20 min, and transductants were selected on Luria-Bertani plates containing KAN. In order to transfer phages to O395 in the absence of antibiotic selection, 1 µl of agglutinated O395 was mixed with 1.5 ml of filtered log-phase-donor (e.g., AS207) supernatant. The mixture was grown overnight at 30°C, and phage transfer was subsequently detected by Southern blotting of plasmid DNA prepared from the culture.

Molecular biology methods. Southern hybridization was carried out using horseradish peroxidase-labelled DNA probes, which were prepared and hybridized using the ECL direct nucleic acid labelling and detection system (Amersham Pharmacia, Little Chalfont, Buckinghamshire, England) according to the manufacturer's instructions. The rstR^calc probe was a PCR product amplified with the primers rstR-10 and rstR-11; the rstR^ET probe was a PCR product amplified with the primers rstR-3 and rstR-8 (9). The PCR primers used for analysis of pCTX integration sites were TLCF1 (5' TGTCGGAGCTGCTTGGATTAAG 3') and RstR Rev (5' CGACCAAGCAAGATAATCGAC 3'). Other techniques were performed using standard protocols (2).

Nucleotide sequence accession number. The sequence of the CTX^calc RS2 region has been assigned GenBank accession no. AF110029.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Structure and sequence of CTX prophages within Calcutta strains of V. cholerae. The O139 strains of V. cholerae that emerged as a cause of widespread disease in Calcutta in 1996 were found by restriction mapping to contain two tandemly arranged copies of a novel CTX prophage (3, 8, 20). These prophages were integrated into the chromosome immediately downstream of an El Tor RS1 element and an El Tor CTX prophage (Fig. 1A). In order to sequence the RS2 region of the Calcutta prophage, we amplified and cloned a PCR product spanning this region from AS207, a representative O139 Calcutta strain (8). Comparison of the putative rstR and ig-2 from the Calcutta prophage with sequences from El Tor and classical prophages revealed striking differences (Fig. 1B). In contrast to rstA and subregions of ig-1, which are highly conserved among the three prophages, the putative Calcutta rstR and ig-2 share no extended sequence identity with the other prophages. In addition, the longest open reading frame (ORF) found within the rstR region of the Calcutta prophage is predicted to encode a protein that is significantly shorter (59 amino acids) than the El Tor and classical RstRs (113 and 112 amino acids, respectively). BLAST searches revealed no significant homology between the Calcutta rstR region and the sequences within the GenBank database, either at the nucleotide or amino acid level. In contrast, both RstR^ET and RstR^class show sequence similarity to a number of bacteriophage repressors (9), and both are predicted by the Dodd and Egan matrix to contain hth DNA binding motifs that start near their amino termini (4). We could not identify a similar hth domain within RstR^calc with the Dodd and Egan matrix. However, the protein motif analysis program fingerPRINTScan did identify RstR^calc as a repressor containing an hth DNA binding domain (hthrepressr fingerprint), but only if the stringency of analysis was reduced from the default value of 15% to 12%. Unlike in RstR^ET and RstR^class, the hth motif in RstR^calc is found near the carboxyl terminus of the protein.

Allele-specific repression of rstA^calc by RstR^calc. We have previously shown that the different RstRs expressed within classical and El Tor CTX lysogens repress rstA-lacZ reporters in a biotype-specific manner (9). To assess whether Calcutta strains similarly encode a repressor specific for the novel rstA^calc promoter sequence, we coexpressed the putative RstR^calc with a panel of rstA-lacZ reporters in an E. coli K-12 strain, CC118 (12). In addition, each reporter, rstA^ET-lacZ, rstA^class-lacZ, and rstA^calc-lacZ, was coexpressed in CC118 with RstR^ET and RstR^class. Production of the RstRs was controlled by an ARA-inducible promoter (pBAD) (7) as described previously (9). We found that -galactosidase activity produced from rstA^calc-lacZ was consistently high when this reporter was maintained alone, maintained with a pBAD33 vector control, or maintained with RstR^ET- and RstR^class-producing plasmids, both under inducing and noninducing conditions (Table 2 and data not shown). However, rstA^calc-lacZ expression decreased 50-fold when production of RstR^calc was induced with ARA (Table 2). Thus, the 59-amino-acid polypeptide encoded by the ORF underlined in Fig. 1B is sufficient to repress rstA^calc expression. No additional repression was observed with a larger repressor construct containing an additional ORF found downstream of rstR^calc, nor was rstA^calc-lacZ expression repressed by the downstream ORF alone (data not shown). The RstR^calc repressor activity was specific; it did not reduce the -galactosidase activity produced either from rstA^class-lacZ or rstR^ET-lacZ (Table 2). Specific repression of the reporter constructs was also observed when the reporter plasmids were transformed into V. cholerae strains producing various repressors from their endogenous prophages (data not shown). Thus, endogenous levels of RstR are sufficient to repress expression of RstA; repression is not an artifact of overexpressing the repressors in E. coli.

The CTX^calc prophage encodes an infectious bacteriophage. We next ascertained whether the CTX^calc prophage gives rise to transmissible bacteriophage particles, or if, like the classical CTX prophage, it lacks the capacity for independent replication. We hypothesized that CTX^calc transmission to the classical strain O395 would result in production of pCTX^calc (the plasmid, or replicative form [RF], of CTX^calc) as occurs following the infection of O395 with CTX^ET (22). Therefore, we incubated cell-free supernatants from AS207 with O395, then prepared plasmid DNA from potentially infected cells and used Southern hybridization analysis to assay for transmission of CTX^calc from AS207 supernatants to O395. Southern blots were probed sequentially with rstR^calc and then with rstR^ET. Control experiments revealed that no rstR^calc-hybridizing species could be detected in plasmid DNA prepared from O395 cultures (Fig. 2). However, an rstR^calc-hybridizing plasmid species was detected within plasmid DNA isolated from O395 cultured at 30°C in the presence of filtered supernatants from AS207. An equally sized rstR^calc-hybridizing species was detected in plasmid DNA prepared from AS207. Restriction digests revealed these plasmids to be the RF of CTX^calc, which probably forms in AS207 as a replication intermediate during CTX^calc production. Treatment of the AS207 cell-free culture supernatant with DNase I did not prevent the transfer of pCTX^calc to O395 from AS207 supernatants (data not shown), thereby suggesting that AS207 gives rise to a bacteriophage, CTX^calc, that is competent to infect and replicate within O395. Rehybridization of these Southern blots with an rstR^ET probe revealed that AS207 also is capable of transfer of CTX^ET, at a level matching or exceeding that of an El Tor strain with two tandemly arranged KAN-marked El Tor CTX prophages (2740-80 [CTX^ET-Kn]) (Fig. 2). Thus, the Calcutta strain AS207 of V. cholerae can give rise to two distinct infectious bacteriophages: CTX^calc and CTX^ET. Consequently, CTX^ET is not the sole phage capable of conveying the genes encoding CT to nonpathogenic V. cholerae.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   Detection of transfer of CTXcalc and CTXET from supernatants of AS207 to O395 using Southern blot analysis of plasmid DNA. Undigested plasmid DNAs were run on agarose gels, transferred to a nylon membrane, and sequentially hybridized with probes for rstRcalc (left panel) and rstRET (right panel). Plasmid DNA was prepared from O395 (lanes 1), AS207 (lanes 2), 2740-80 (CTXET-Kn) (lanes 3), O395 cultured with AS207 supernatant (lanes 4), and O395 cultured with 2740-80 (CTXET-Kn) supernatant (lanes 5). The Knr cassette is slightly larger than the ctxAB genes it replaces in CTXET-Kn, so pCTXET-Kn migrates more slowly than pCTXET.

Immunity and heteroimmunity of a CTX^calc-Kn lysogen. We previously found that strains harboring a CTX^ET prophage are significantly resistant to further infection with CTX^ET-Kn. This immunity results from the repression of rstA^ET expression by RstR^ET, as RstA is essential for phage replication. The data presented above indicate that RstR^ET does not repress RstA^calc production, and similarly, RstR^calc does not repress RstA^ET production (Table 2). Therefore we tested, with marked CTX^calc derivatives, whether a lysogen harboring a CTX^calc prophage is immune to CTX^calc superinfection and heteroimmune to CTX^ET infection. Since immunity results from inhibition of phage replication following infection rather than from inhibition of the initial steps of infection, and since El Tor strains cannot be efficiently infected with CTX^calc in vitro (due to lack of expression of TCP, the phage receptor), we developed a transformation assay to study the immunity properties of the CTX^calc prophage. For these assays, 2740-80, an El Tor, CTX, attRS⁺ strain, and 2740-80 lysogens of marked CTX^calc and CTX^ET were electroporated with differentially marked CTX^calc or CTX^ET plasmid DNA.

View larger version (23K): [in this window] [in a new window]   FIG. 3.   Relative efficiency of transformation of pCTXcalc-Ap versus pCTXET-Ap into 2740-80, 2740-80 (CTXET-Kn), and 2740-80 (CTXcalc-Kn). Percentage of total Apr transformants = (pCTXcalcAp or pCTXETAp transformants)/(pCTXcalcAp + pCTXETAp transformants).

Integration site preference and evolution of V. cholerae O139 Calcutta. We used 2740-80 harboring either marked El Tor or Calcutta CTX prophages to explore potential steps in the evolution of Calcutta strains such as AS207. In these experiments, we determined where the DNA of CTX^ET and CTX^calc integrates on the chromosome in strains harboring either the CTX^calc prophage or the CTX^ET prophage, respectively. We have found that CTX phage DNA reliably integrates into attRS following infection of attRS⁺, CTX El Tor strains, such as 2740-80. However, integration regenerates the 18-bp core of attRS or the very similar ER sequence on both ends of the prophage (16). Thus, a subsequently infecting phage has two or more potential integration targets; the precise number of integration targets is determined by the length of the prophage array. To determine the site of CTX integration following infection of a CTX lysogen, we developed a PCR and restriction-digest-based assay. For the PCR reaction, one primer was complementary to DNA 5' of attRS on the 2740-80 chromosome and the other primer was complementary to a conserved sequence within rstA (Fig. 4). PCR products were digested with enzymes that cleave either the El Tor or the Calcutta rstR, thereby allowing us to identify the furthest 5' prophage within an array of prophages. We found that the initial phage integrated on the 2740-80 chromosome always retained the most 5' position on the chromosome (Fig. 4); thus, CTX^calc-Kn, CTX^ET-Kn, CTX^calc-Ap, and CTX^ET-Ap were each maintained as the most 5' prophage if their DNA was the first in a series to be electroporated into 2740-80. Subsequent integrations, regardless of which particular phage's DNA was tested, occurred 3' of an integrated prophage (Fig. 4 and data not shown). In other words, DNA from the second antibiotic-marked phage always integrated either between tandem prophages or between an integrated prophage and 3' chromosomal DNA. This unambiguous insertion site preference in sequentially transformed strains (and presumably also in sequentially infected strains) strongly suggests that Calcutta strains such as AS207 arose by infection of a V. cholerae CTX^ET lysogen by CTX^calc.

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Integration site selection by sequentially integrated plasmids. (A) PCR followed by restriction digestion was used to determine the order of prophages within the chromosome following sequential integration of CTXET and CTXcalc antibiotic-marked plasmids. When the CTXcalc prophage is the furthest 5', the TLCF1-RstR Rev PCR product contains a HindIII site but not a BglII site. Conversely, when CTXET is upstream, the PCR product contains a BglII site but not a HindIII site. (B) A representative agarose gel containing PCR products digested with HindIII. Template DNA is as follows: 2740-80(CTXET-Kn) transformed with pCTXcalc-Ap (lanes 1 to 6), 2740-80 (CTXcalc-Kn) transformed with pCTXET-Ap (lanes 7 to 12), and 2740-80 (CTXcalc-Kn) (lane 13). (C) Summary and model of integration site selection following sequential CTX integration. Phage DNA does not integrate into the 5' ER (black triangle) if the chromosome already contains a CTX prophage. Instead, the new phage DNA inserts into the 3' ER.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have investigated the repressive and replicative capabilities of a novel variant of CTX, which was found as a prophage within epidemic-linked strains of O139 V. cholerae isolated from Calcutta. This prophage gives rise to CTX^calc virions that infect and/or lysogenize both relatively nonpathogenic (lacking ctxAB) V. cholerae strains, such as 2740-80, as well as classical and El Tor strains that already contain CTX prophages. These findings demonstrate that CTX^ET is not the sole viable, CT-encoding, filamentous phage capable of transmitting ctxAB within V. cholerae populations. CTX^calc lysogens are immune to infection by CTX^calc due to the production by these lysogens of a new repressor, RstR^calc, that inhibits expression from the adjacent rstA^calc promoter. Like RstR^ET and RstR^class, the transcriptional repressor activity of RstR^calc is sequence specific; RstR^calc does not inhibit the activity of either the rstA^class or the rstA^ET promoter. Similarly, the rstA^calc promoter is not repressed by RstR^class or RstR^ET.

Despite the functional similarity of the three CTX repressors, the RstR^calc amino acid sequence is unrelated to the sequences of RstR^class, RstR^ET, and the known repressor proteins of other lysogenic phages. In fact, BLAST searches using the nucleotide and predicted amino acid sequences of this repressor revealed no significant homology to any sequence within the GenBank databases. It is therefore not surprising that most protein analysis algorithms that we tested did not predict any repressor or DNA binding activity for RstR^calc. However, the motif-detecting program fingerPRINTScan does detect a low level of similarity between RstR^calc and the hth domain found in lambdoid repressors and in a subset of homeotic proteins (hthrepressr fingerprint). Interestingly, fingerPRINTScan assigns RstR^class and RstR^ET to different hth categories (homeobox and hthLysR fingerprints, respectively), suggesting that the three V. cholerae repressors may be more closely related to proteins from other species than they are to each other. The genetic mechanism(s) by which the unrelated RstR-RstR operator pairs became associated with otherwise very similar CTXs has not been explored. However, recombination between a CTX containing a particular repressor-operator pair and repressor-operator sequences within other temperate bacteriophages or even within other genetic elements is clearly one possibility.

Production of RstR^calc enables the CTX^calc prophage both to control its own replication and to inhibit RstA-mediated replication of any newly introduced DNA that relies upon the rstA^calc promoter. This confers upon a lysogen a degree of immunity to secondary infections by identical phages, similar to the immunity produced by  prophages within E. coli. Also like lambdoid prophages, CTX^calc lysogens are susceptible to infection (heteroimmune) by CTX with different immunity regions (rstR/RstR operator sequences). In addition, results from our transformation efficiency assay suggest that CTX^ET lysogens are heteroimmune to CTX^calc infection. Integrated CTX^calc-Kn does not completely inhibit replication of the related replicon, CTX^calc-Ap; however, it does significantly diminish CTX^calc-Ap replication relative to that of CTX^ET-Ap. In the transformation efficiency assay system, CTX^calc-Kn is less effective at providing immunity than is CTX^ET-Kn. This finding could reflect the relative weakness of RstR^calc as a repressor; alternatively, it may indicate the strength of the rstA^calc promoter. When assayed in CC118, an rstA^calc-lacZ reporter fusion has higher activity than either rstA^class-lacZ or rstA^ET-lacZ reporters, so even if repressed to the same degree as the other promoters, rstA^calc may maintain higher residual activity. Residual production of RstA, either from the prophage or from the RF, presumably enables some plasmids to replicate and subsequently integrate. Although our assay measures immunity indirectly, by monitoring transformation of lysogens with the RF of the CTX genomes, these experiments yielded results similar to those obtained when CTX^ET immunity was assayed directly, using an intraintestinal transduction assay (9).

Our results support the hypothesis that AS207 and related Calcutta strains arose via infection of an O139 CTX^ET lysogen by the previously unknown CTX^calc. Epidemic O139 V. cholerae strains isolated prior to 1996 contain only El Tor CTX prophages but otherwise are very similar to Calcutta O139 strains (3, 20) and thus are likely AS207 progenitors. Our experiments indicate that CTX^calc infection and lysogenization of such a progenitor strain would result in the same arrangement of CTX prophages as is seen in Calcutta O139 strains such as AS207. A recent survey of environmental V. cholerae strains in Calcutta detected a prophage encoding RstR^calc (presumably integrated CTX^calc) in a non-O1, non-O139 strain of V. cholerae (3a). This may indicate that CTX^calc is transmitted within the estuarine environment as well as within the laboratory; it also reveals a potential source of CTX^calc. Thus, it seems probable that AS207 and other Calcutta strains arose via infection of an earlier, epidemic-linked O139 strain with CTX^calc. It is also possible that recombination between a CTX^ET prophage within an O139 strain and a repressor-operator from an unrelated genetic element gave rise to O139 Calcutta strains, but such a mechanism seems less likely to account for the origin of these strains.

When we proposed using rstR^ET to protect classical and El Tor live-attenuated V. cholerae vaccine strains from reversion to toxigenicity mediated by CTX infection (9), the only described infectious CTX was CTX^ET. Nucleotide sequence analysis of the rstR/ig-2 immunity regions in multiple El Tor clinical isolates from around the world had revealed that they are identical (9), lending credence to this approach. However, the current description of the infectious CTX^calc, which encodes the novel RstR^calc, suggests that rstR-mediated immunity to CTX infection may not constitute a useful method of enhancing the biosafety of live-attenuated V. cholerae vaccines. Besides rstR^calc, two additional putative rstRs have been identified recently in environmental, non-O1/O139 V. cholerae isolates (3a, 3b), and more alleles probably remain to be detected. Thus, introduction of a comprehensive immunizing library of rstRs into vaccine strains may not be practical. Ongoing studies are exploring alternative mechanisms for preventing CTX-mediated transfer of ctxA and -B to vaccine strains.