Vibrio cholerae is the causative agent of cholera, a severe diarrheal disease (Kaper et al., 1995; Sack et al., 2004). The bacterium-produced cholera toxin (CT) is mainly responsible for the clinical manifestations of cholera. The toxigenic V. cholerae strains are generated in the course of a lysogenic conversion of a 6.9-kb single-stranded DNA filamentous phage, the cholera toxin phage (CTX phage), which carries the cholera toxin gene (Waldor and Mekalanos, 1996). Two biotypes of V. cholerae O1 strains, the classical and El Tor biotypes, have been dominant sequentially in the history of the disease; the classical biotype strains were predominant before the 1960s, whereas the El Tor biotype strains emerged in 1961 and have globally replaced the classical biotype strains (Sack et al., 2004).

Two different CTX phages, classified by rstR type and sequence variations in ctxB, are associated with the host bacterial biotype: classical biotype strains harbor the CTX-cla phage, while prototype El Tor strains harbor CTX-1. However, atypical El Tor strains that harbor mosaic CTX phages of CTX-1 and CTX-cla, namely CTX-2 containing the classical type rstR and ctxB or CTX-3~CTX-6 containing the El Tor type rstR and the classical type ctxB, have been prevalent since the early 1990s (Safa et al., 2010; Kim et al., 2015).

The mosaic CTX phages in atypical El Tor strains are generated by recombination between two different types of CTX prophages in the host V. cholerae cell (Kim et al., 2014, 2017). Such inter-phage recombination is possible because V. cholerae strains contain two chromosomes and the CTX phages can be integrated into either one (Das et al., 2013).

CTX phages are generally transmitted by transduction of V. cholerae strains, but chitin-mediated uptake of CTX phage by non-toxigenic V. cholerae strains was also reported (Waldor and Mekalanos, 1996; Udden et al., 2008). The replication and transduction of CTX phages have been verified both in vivo and under laboratory conditions (Waldor and Mekalanos, 1996; Kim et al., 2017). The transmission of the CTX-1 phage to the classical biotype strains, and more recently, the transmission of CTX-2/CTX-cla phages to the El Tor biotype strains, indicates that the replication and maintenance of a CTX phage is not host biotype-specific (Waldor and Mekalanos, 1996; Kim et al., 2017). The replication of CTX-cla and CTX-2 is as efficient as that of CTX-1 under laboratory conditions, suggesting that it is highly likely that the CTX-2 or CTX-cla phages are widespread in nature. Moreover, the identification and the mechanism of generation of the atypical El Tor strains suggest that more strains that harbor various CTX phages and their arrays exist in the ocean, the natural habitat of V. cholerae (Kim et al., 2014).

Since our knowledge of the replication and transmission of CTX-2 and CTX-cla has been established under laboratory conditions, we proposed that V. cholerae strains that contain various CTX array could be designed and constructed under laboratory conditions. In the current study, a panel of El Tor strains harboring CTX-1 and/or CTX-2 were constructed by CTX phage transduction, chromosomal integration, and elimination of the replicative form of the CTX phage, pCTX. In addition to the El Tor biotype strains harboring various CTX array, a classical biotype strain harboring CTX-1 was also generated, indicating that atypical classical biotype strains might exist in nature. Further, we provide evidence for recombination between pCTX and a CTX prophage, which could explain the generation of pCTX-2. Taken together, we show that more diverse V. cholerae strains likely exist in nature, and that the design and construction of V. cholerae strains harboring various CTX phage arrays is possible under laboratory conditions.

The non-coding sequence between the rstR and ctxB genes in pCTX is different in pCTX-1 and CTX-cla/pCTX-2 (Kim et al., 2016). The pCTX integration into either chromosome 1 or 2 is determined by the attP site located within the non-coding sequence of pCTX, and the dif sequences on the bacterial chromosomes (Das et al., 2010). When the pCTX constructs generated in this study harbored the non-coding sequence of pCTX-1 (pCTX-1-kan-N1, pCTX-cm-N1, and pCTX-2-kan-N1), the integration of CTX phages proceeded only on chromosome 1. When the pCTX constructs harbored the non-coding sequence of pCTX-cla (pCTX-1-kan-N2 and pCTX-2-kan-N2), the integration proceeded into chromosome 1 or 2, or both. Therefore, some strains with CTX-1 and CTX-2 on chromosome 1 and/or chromosome 2 are expected to exist in nature; a number of such strains were constructed in this study (Choi et al., 2010).

The non-coding sequence of pCTX is mostly derived from a downstream prophage or an RS1 element during the replication of a CTX prophage (Davis and Waldor, 2000). Most El Tor strains harbor the CTX-1:CTX-1 or CTX-1:RS1 array for the replication of CTX-1; hence only pCTX-1, with the non-coding sequence of the authentic CTX-1, can be generated from such arrays (Kaper et al., 1995). However, if a strain that harbors CTX-1:CTX-2 is generated by a sequential integration of CTX-1 and CTX-2, the resultant pCTX-1 replicating from this array will have the non-coding sequence of CTX-2. pCTX-1-N2 generated in this manner can then be transmitted and integrated into either chromosome of other strains. Thus, pCTX-1 phages with the non-coding sequence of pCTX-2 (pCTX-1-N2) can be expected in nature in addition to those with pCTX-1-N1. Indeed, in the current study, pCTX-1-cm-N2 was produced from PM48, which harbors the TLC:CTX-1-cm-N1:CTX-1-kan-N2 array.

Although the replication of CTX-2 has been confirmed under laboratory conditions, the initial generation of CTX-2 remains to be demonstrated. CTX-2 could have originated from a hypothetical strain harboring CTX-1 and CTX-cla on each chromosome or from a classical strain harboring a tandemly repeated CTX-cla prophage and pCTX. In the current study, we demonstrated rstR exchange between the pCTX-1 and CTX-2 prophage, which could be considered to happen in a more innate manner.

A Wave 2 strain, V212-1, harbors TLC:RS1:CTX-1:RS1 on chromosome 1 and CTX-2:CTX-2 on chromosome 2; this is one of the most complicated CTX arrays harbored by El Tor strains and perhaps the only strain with two different types of CTX phages (CTX-1 and CTX-2) (Kim et al., 2014). V212-1 is considered to be an intermediary strain between Wave 2 and Wave 3 strains, since generation of mosaic CTX phages CTX-3 ~ CTX-6 has been demonstrated in V212-1 (Kim et al., 2014). It is possible that V212-1 was generated upon the integration of two CTX-2 phages into chromosome 2, such as in a strain resembling MG116025.

El Tor strains harboring two different types of CTX phages (CTX-1 and CTX-2) were generated in the current study. A classical biotype strain harboring CTX-1 was also constructed. These results demonstrate that V. cholerae strains harboring various CTX phage arrays may be generated under laboratory conditions. Ultimately, a strain harboring all the different kinds of CTX phage, including CTX-O139 and an environmental-type CTX phage, in a single host cell could be constructed (Nusrin et al., 2004). These results imply that potentially diverse, as yet unidentified, V. cholerae strains may exist in nature.

Strains that contain various CTX arrays were constructed from Wave 2 strains and a pre-seventh pandemic strain in the current study, but the design and construction of new strains from any El Tor strain (Wave 1, 2, or 3 strain) and a classical strain, regardless of whether they harbor CTX prophage/RS1 or not should also be possible.

PM30 is of particular interest because it harbors both CTX-1 and CTX-2 on chromosome 1. A strain harboring TLC:RS1:CTX-3, a typical array in Wave 3 El Tor strains, could be directly generated from PM30, by excision of CTX and the second RS1 via recombination between CTX-1 and CTX-2. This may explain the multiple origins of Wave 3 El Tor strains.

It remains to be determined whether the co-existence of more than two CTX phage (the same or different types) in a single bacterial cell is cooperative or leads to interference. The virulence and toxigenicity of V. cholerae strains containing diverse CTX phage arrays may vary depending on the array. This could help understand the prevalence of strains with a particular CTX array among different strains over a given period of time.

Some strains constructed in the current study could be developed as cholera vaccine strains. Although ctxAB was substituted by an antibiotic-resistance cassette in the V. cholerae strains described in this study, the antibiotic-resistance cassette can be readily replaced with the authentic ctxAB gene. Replacement of the antibiotic-cassette with the ctxAB is under way. The presence of the toxT-139F allele results in high-level expression of toxin co-regulated pilus (TCP) and CT. Elevated expression of CT and TcpA in the parental strains of the constructed V. cholerae strains has been previously reported (Kim et al., 2017). Phenotypic characterization (especially, expression of TcpA and CT) will be performed and the strains will be further evaluated as a potential cholera vaccine. Some strains harbor different types of CTX phage, indicating that the production of different CTs (especially CTB via the elimination of ctxA) in a single cell is also feasible. (Levine et al., 2017).

In conclusion, we have shown that it is feasible to construct V. cholerae strains harboring different CTX arrays. These strains may be selectively designed to produce specific or multiple antigens. The strains generated in the current study and the established protocol for CTX array generation may be further exploited for the development of next-generation cholera vaccines, thereby addressing one of the most pressing concerns of current global health.

DK contributed the conception of this study; DK, GN, and EK designed the experiments; HY, DC, D-HS, and EK performed the experiments; HY, DC, and EK analyzed data, GN, EK, and DK drafted the manuscript; and HY, DC, and D-HS approved the manuscript.