Recent Vibrio cholerae O1 Epidemic Strains Are Unable To Replicate CTXΦ Prophage Genome

ABSTRACT Cholera, an acute diarrheal disease, is caused by pathogenic strains of Vibrio cholerae generated by the lysogenization of the filamentous cholera toxin phage CTXΦ. Although CTXΦ phage in the classical biotype are usually integrated solitarily or with a truncated copy, those in El Tor biotypes are generally found in tandem and/or with related genetic elements. Due to this structural difference in the CTXΦ prophage array, the prophage in the classical biotype strains does not yield extrachromosomal CTXΦ DNA and does not produce virions, whereas the El Tor biotype strains can replicate the CTXΦ genome and secrete infectious CTXΦ phage particles. However, information on the CTXΦ prophage array structure of pathogenic V. cholerae is limited. Therefore, we investigated the complete genomic sequences of five clinical V. cholerae isolates obtained in Kolkata (India) during 2007 to 2011. The analysis revealed that recent isolates possessed an altered CTXΦ prophage array of the prototype El Tor strain. These strains were defective in replicating the CTXΦ genome. All recent isolates possessed identical rstA and intergenic sequence 1 (Ig-1) sequences and comparable rstA expression in the prototype El Tor strain, suggesting that the altered CTXΦ array was responsible for the defective replication of the prophage. Therefore, CTXΦ structures available in the database and literatures can be classified as replicative and nonreplicative. Furthermore, V. cholerae epidemic strains became capable of producing CTXΦ phage particles since the 1970s. However, V. cholerae epidemic strains again lost the capacity for CTXΦ production around the year 2010, suggesting that a significant change in the dissemination pattern of the current cholera pandemic occurred. IMPORTANCE Cholera is an acute diarrheal disease caused by pathogenic strains of V. cholerae generated by lysogenization of the filamentous cholera toxin phage CTXΦ. The analysis revealed that recent isolates possessed altered CTXΦ prophage array of prototype El Tor strain and were defective in replicating the CTXΦ genome. Classification of CTXΦ structures in isolated years suggested that V. cholerae epidemic strains became capable of producing CTXΦ phage particles since the 1970s. However, V. cholerae epidemic strains again lost the capacity for CTXΦ production around the year 2010, suggesting that a critical change had occurred in the dissemination pattern of the current cholera pandemic.

Recently we reported that two distinct lineages of pathogenic V. cholerae strains were concurrently prevalent between 2007 and 2009, while one lineage became predominant in 2010 and later in Kolkata, India (15). These investigations were performed by phylogenetic analyses, based on the single nucleotide polymorphisms of core genomic sequences, using short reads obtained with an Illumina next-generation sequencer. However, these analyses do not provide insights with respect to the copy number and array of repeated genomic sequences, which play important roles in the virulence of V. cholerae (21). To investigate the variations in the chromosomal structures of V. cholerae epidemic strains, including copy numbers and the arrangement of repeated sequences, we developed complete genomic sequences of five representative V. cholerae strains isolated from Kolkata, India, in this study. The analysis demonstrated that the recent epidemic strains of V. cholerae El Tor biotype possess both intact CTXU prophage and RS1, although in the altered array of prototype El Tor strains. The strains with an altered CTXU array were incapable of replicating the CTXU genome, suggesting that a critical change may have occurred in the dissemination and continuation route of the current cholera pandemic.

RESULTS
Structural variations in the genomes of recent V. cholerae epidemic strains. It was reported that two distinct lineages (1 and 2) of pathogenic V. cholerae strains were concurrently prevalent between 2007 and 2009, while lineage 2, sublineage III, appeared in 2010, followed by the predominance of lineage 2, sublineage IV, in 2011 and later in Kolkata, India (15). Two isolates, IDH-00113 (referred to here as strain 13) and IDH-02387 FIG 1 Model for the rolling-circle replication of the CTXU prophage genome. Genes in the CTXU prophage and its array in the classical biotype O395 and El Tor biotype N16961 are shown (top and panel A). Open arrows indicate the genes necessary for DNA replication and the integration of the phage. Striped arrows indicate the genes required for phage packaging and secretion (21). Solid arrows indicate the genes responsible for encoding the cholera toxin. Ig-1 was located adjacent to rstR in CTXU and RS1, and the CTXU prophage genome was located between the two Ig-1 sequences in the prototype El Tor biotype strain (A). RstA nicked Ig-1 in the plus-strand DNA of CTXU (B). Host replication machinery synthesized a new plus strand while displacing the old plus strand (C). RstA nicked Ig-1 at the downstream end of the CTXU prophage genome, resulting in a closed circular single-strand DNA (D). RS1 in recent isolates used in this study was located at the upstream end of the CTXU genome, compared with the prototype El Tor the biotype strains (E). This type of CTXU prophage lacked the Ig-1 at the downstream end of the CTXU genome.
(referred to here as strain 87), isolated in 2007 and 2009, respectively, represented lineage 1 and were predominant until 2009 (15). IDH-03329 (referred to here as strain 29), isolated in 2010, and IDH-03506 (referred to here as strain 06) and BCH-01536 (referred to here as strain 36), isolated in 2011, belong to sublineages III and IV, respectively, of lineage 2. Lineage 2 sublineage III strains were transient in Kolkata and observed only in 2010. Thereafter, sublineage IV strains of lineage 2 became predominant.
Long-read genomic sequences of these strains were obtained using Oxford Nanopore MinION, and the nucleotide sequences were polished by short reads obtained by Illumina sequencing (15). Then, the complete genomic sequences were assembled and their chromosomal structures were compared with that of V. cholerae N16961, a prototype El Tor strain. The detected structural variations were verified through PCR amplification (data not shown). The confirmed differences of .2,000 bp are summarized in Fig. 2 and Table 1. The previously identified variations in the VSP-II genetic island, in which 3,343-bp and 14,376-bp regions were replaced by transposase genes in lineages 1 and 2, respectively, were confirmed in this study (15) (Fig. 2A, boxes a and b). Various integral and conjugative elements FIG 2 Structural variations in the chromosomes of V. cholerae. Replacements, insertions, and translocations of .2,000 bp compared with the N16961 genomic sequences, suggested on the basis of complete genomic sequences obtained in this study and confirmed by PCR, are illustrated. Circular chromosomal maps having genes on the plus strand, those with genes on the minus strand, and GC percent (from outside to inside) were generated using CiVi (Circular Visualization for Microbial Genomes [60]). A, larger chromosome; B, smaller chromosome.  Fig. 2A, boxes d, e, and g). The DNA fragment was inserted in strain 36 into zot, a gene in the CTXU prophage genome, and consequently, the gene was divided ( Fig. 2A, box e). The 14-kbp (approximately) sequences inserted in strains 13 and 36 were nearly identical to the origin-proximal regions, including VC0175 to VC0185, suggesting duplication and insertion of the fragment into the distal region ( Fig. 2A, boxes e and g). In addition to these insertions and replacements, translocation within the CTXU region in all recent isolates compared with the N16961 strain was detected ( Fig. 2A, box f). The analysis of this structural difference is described in detail below. In the smaller chromosome, three replacements by a transposase-encoding gene (Fig. 2B, boxes a, b and c) and an insertion of approximately 14-Kbp fragment in strains 13, 29, 06, and 36 were detected (Fig. 2B, box d). This 14-kbp fragment was also almost identical to the region VC0175 to VC0185 in the chromosome 1 ( Fig. 2A, boxes e and g), indicating duplication and insertion from larger to smaller chromosomes. In summary, the results indicated that the chromosomes of pathogenic V. cholerae strains were frequently replaced with mobile genetic elements and were highly diverse even in spatiotemporally close clinical isolates.
Alteration of CTXU prophage arrays. We next focused on the alteration of the CTXU prophage array among the identified structural differences ( Fig. 2; Table 1). The CTXU prophage genome in the El Tor biotype strains is usually found in tandem and/ or followed by the related genetic element known as RS1, and this tandem array is essential for the replication and induction of CTXU (17) (Fig. 1A to D). In brief, the CTXU prophage genome was replicated by nicking two Ig-1 sites of CTXU and the following element by either tandem CTXU or RS1, i.e., Ig-1 upstream and downstream of the CTXU genome (Fig. 1A). The complete genomic sequences obtained in this study and after PCR confirmation indicated that all recent isolates possessed intact gene sets of CTXU prophage and RS1. However, RS1 was located at the upstream end of the CTXU genome, in contrast to the prototype El Tor strains ( Fig. 1A and E). These arrays of the CTXU region possessed Ig-1, a nicking site for rolling-circle replication, only at the upstream end of the prophage genome. This observation suggests that the recent V. cholerae epidemic strains have lost the ability to replicate the CTXU genome and have led to the subsequent production of infectious phage particles. To confirm this possibility, a circular replication product of CTXU phage was specifically detected using PCR (Fig. 3A). The prototype El Tor type strain N16961 produced the replication product with and without mitomycin C induction (Fig. 3B). However, no such replication was detected in all recent isolates, even in the mitomycin C-induced condition (Fig. 3B). These results confirmed that recent V. cholerae strains are incapable of replicating the CTXU prophage genome.
Factors required for replication. It was reported that rstA is the only CTXU gene required for its replication in V. cholerae (19). We next investigated whether rstA is sufficient for the replication in the absence of CTXU and specific elements of V. cholerae. rstA was cloned into the plasmid pET-21a, and Ig-1 sequences were inserted upstream and downstream of the gene (Fig. 4A). rstA expression was induced in E. coli BL21 using IPTG (isopropyl-b-D-thiogalactopyranoside), and circular rolling-circle replication products were detected using PCR with the primers in the inverse direction ( Fig. 4A and B). The V. cholerae genomic sequence was inserted instead of Ig-1 as a negative control. The replication was detected in Escherichia coli only in the presence of both rstA expression and Ig-1 (Fig. 4C). These results indicated that the two Ig-1 nicking sites as well as rstA expression were necessary and sufficient in V. cholerae and CTXU-specific elements for replication. Because all strains used in this study possessed Ig-1, rstA, and its upstream sequence identical to those of V. cholerae N16961, rstA expression levels in these strains were compared. rstA was mainly expressed in the exponential growth phase, and the expression decreased in the stationary phase in all tested V. cholerae strains (data not shown). Although the relative expression levels of recent isolates in the growth phase compared with N16961 showed some variation (0.5-to 1.3-fold), comparable rstA expression levels in all strains were confirmed (Fig. 5). These results indicated that all strains used in the present study possessed the necessary genetic elements  for CTXU replication, and thus, the altered prophage array structure was supposedly responsible for the inability to replicate.
Impact of CTXU replication. We estimated the number of CTXU phage produced from a single V. cholerae bacterium. Because CTXU is not a plaque-forming phage and phage particles are not detected as PFU, its genomic DNA was quantified instead of phage particles. All strains used in this study possessed a single copy of ctxA in the genome ( Fig. 1A and E). Therefore, the DNA fragment of ctxA from an equal amount of extracted DNA was compared using quantitative PCR (qPCR) for the CTXU replication of positive and negative strains (Fig. 6). In the CTXU replication-positive N16961 strain, the amount of ctxA fragment increased slightly during stationary phase compared to that in the exponential growth phase. All the negative isolates from CTXU replication exhibited approximately half of the amount of the ctxA fragment in N16961 (Fig. 6). It is not clear whether a small cell population replicated the CTXU prophage genome many times or most cells replicated it only a few times. Nonetheless, these results indicated that prototype  El Tor strain produces CTXU phage at number comparable to that of the cell population and that its impact on the dissemination of cholera cannot be ignored.
Variation in CTXU arrays of known V. cholerae strains. The CTXU array patterns detected in 52 V. cholerae strains are depicted in Table 2 and Fig. 7. Data regarding the year of isolation and the complete genomic sequence or CTXU structures are available in the literature (23). Fourteen CTXU array patterns were identified from strains isolated between 1956 and 2015 from Asia, Africa, Latin America, and North America ( Fig. 7; Table 2). Among the 14 CTXU array variations, 11 possessed Ig-1 at both the upstream and downstream ends of CTXU prophage genome (indicated with an "A" in Fig. 7), suggesting that these strains can produce infectious virions, whereas 3 CTXU arrays demonstrated defective replication when the structure was observed (indicated with a "B" in Fig. 7). Of note, non-CTXU-producing strains were isolated in 1965 and earlier, but, later on, CTXU-producing strains were reported to be more dominant worldwide for over 3 decades ( Fig. 8; Table 2). The strain used in this study, i.e., those with the non-CTXU-producing CTXU structure type B1, appeared in 2007 in India, and this type of strain became predominant after 2009 in Asia, Africa, and Latin America ( Fig. 8; Table 2). In summary, these data suggest that V. cholerae epidemic strains did not produce CTXU phage during the sixth and early seventh pandemics, but the pattern shifted to the CTXU-producing strains in the 1970s. This probably generated new pathogenic V. cholerae strains in the environment through CTXU phage infection during these periods, which lasted for over 3 decades. However, the pandemic strains were again found to have lost the ability to produce CTXU phage particles around 2010, indicating that the observed dissemination patterns are at a critical stage during the ongoing cholera pandemic.

DISCUSSION
V. cholerae is generally found in an aquatic environment, where it acquires unique characteristic features that makes it better adapted to a particular environment through the uptake of genetic molecules from natural resources, either through transformation or via interaction with other inhabitants (4). In this manner, the bacteria communicate with  Table 2 are shown. Arrows labeled "CTXU" and "RS1" indicate the CTXU genome in the rstR-to-ctxB direction and the RS1 element in the rstR-to-rstC direction, respectively (Fig. 1A). The right side of the displayed array is followed by the rtxA gene. *, truncated CTXU genome lacking sequence downstream from the internal region of cep, as shown in Fig. 1 (classical). Triangles, Ig-1 region (167 bp), which is required for rolling-circle replication (20). A and B indicate the CTXU arrays, which are expected to be capable and incapable of replication, respectively.  Table 2). The biotypes and waves of each strain were determined on the basis of ctxB, rstR, and rstA genotypes (14). the toxigenic CTXU phage and integrate the genomic constituents of the phage irreversibly into their genomes so as to gain the toxic components of the phage genome, mainly ctxA and ctxB; therefore, the risk of this disease progression in humans has invariably increased (2).
The two naturally occurring V. cholerae biotypes possess CTXU phage-integrated genomes but with different arrangements and cellular functions (17). In recent years, novel variants of V. cholerae O1 have been found to emerge with altered ctxB genotypes and with higher pathogenic potency (5,24). Past reports suggested that the classical V. cholerae strains could not produce virions, as they would not yield extrachromosomal CTXU DNA (17). Comparative genomics revealed that, despite having functional genes for the replication and production of phage particles, classical strains were unable to replicate the CTXU phage genome only because of the structural deficiencies in the CTXU prophage array (17). However, on the emergence of the seventh pandemic, the El Tor strains were found to possess functional CTXU phage genomes that could produce transducible virions (25). With time, different El Tor strains (atypical El Tor) were found to emerge from different places worldwide with modified genetic makeup of the CTXU prophage (5). These atypical El Tor strains were found to arise on the prototype El Tor genomic background only by replacing CTXU phages of different types (21). Thus, the wave 2 El Tor strains possessed tandem repeats of classical-CTXUlike prophages on their second chromosome (14). Other CTXU prophage-containing pathogenic variants of V. cholerae included V. cholerae O139, which harbored an extra copy of a different CTXU prophage located at the downstream end of the preexisting El Tor type CTXU prophage on the first chromosome (26). Faruque et al. demonstrated the presence of a different type of CTXU prophage array in the isolates of the Mozambique variant El Tor strains (27). These strains contained 2 copies of the classical CTXU prophages in the second chromosome, but they were unable to produce virions.
In this study, we demonstrated that recent V. cholerae clinical strains isolated in Kolkata were incapable of replicating the CTXU prophage genome and hence were not responsible for the production of infectious virions. Because CTXU is not a plaqueforming phage, genetic engineering of pathogenic V. cholerae to introduce an antibiotic-resistant gene into the CTXU genome is required to assay phage particle productivity (6). These analyses are expected to further confirm the conclusions of the present study. Nevertheless, the data in the present study strongly suggest that the recent V. cholerae epidemic strains do not produce infectious virions. In strain 36, a gene in the CTXU genome, zot, was disrupted by the insertion of a 14,271-bp fragment that included a transposase-encoding gene ( Fig. 2A, box e). Zot is required for packaging and secretion of the phage as well as possessing enterotoxin activity (21). It may be suggested that the disruption of zot was allowed because phage secretion was no longer required in the strain incapable of replicating the CTXU genome.
The infection of V. cholerae O1 cells by CTXU requires toxin-coregulated pilus (TCP) as the receptor (6). Biogenesis of TCP is dependent on the tcp operon in Vibrio pathogenicity island 1 (VPI-1) on larger chromosomes. The first gene of the operon, tcpA encodes the major pilin subunit (28). Thus, the tcp-positive V. cholerae O1 strains are potential hosts for the CTXU infection to generate epidemic strains. We attempted to isolate V. cholerae O1 strains from lakes, ponds, and rivers in Kolkata, India, between 2014 and 2016 several times and characterized the strains. During this analysis, only one ctxA-positive strain was isolated, whereas 181 tcpA-positive strains were identified (data not shown). These observations suggest that many more potential host cells for the CTXU infection exist than the pathogenic V. cholerae O1 strains in environmental water. Therefore, the cholera pandemic caused by the CTXU-producing strains disseminated CTXU phage particles into the environment and possibly generated new pathogenic V. cholerae O1 strains. Thus, the results of the present study suggest that, during the seventh pandemic, the spread of cholera acquired a secondary disseminating route in which the secreted CTXU phage particles generated new pathogenic V. cholerae O1 strains in the environment (Fig. 8). It was reported that the El Tor type strains were less virulent than the classical type strains, but the recent variant exhibited increased production of toxins (29). It can be concluded that non-CTXU-producing strains with higher virulence were disseminated by the fecal-oral route by causing severe diarrhea and that CTXU-producing strains with lower virulence were spread both by the fecaloral route and by the generation of new pathogenic strains in the environment because of CTXU infection. Thus, the prototype El Tor strains required the spread of CTXU productivity, while the classical type and the recent variant did not. On the other hand, CTXU productivity may be one of the reasons why the seventh pandemic continued for a longer period, i.e., for over half a century, than other pandemics. If this is the case, the appearance of the non-CTXU-producing El Tor strain brought a significant change that can be of great help to restrict the dissemination of V. cholerae and cholera worldwide. Epidemiological studies shall further confirm and reveal the effects of the change in CTXU productivity of V. cholerae found in the present study.
In addition, V. cholerae El Tor strains, which were more stable in the environment but less pathogenic than classical strains, acquired virulence properties of the hypervirulent classical strains in recent years. The first report of V. cholerae El Tor strains mentioned the acquisition of classical ctxB in the El Tor genomic background in the strains of Kolkata during the 1990s (10). Thereafter, the appearance of a new variant hyperpathogenic ctxB genotype (ctxB7) was first observed in the isolates of Kolkata during 2006, which attracted the attention of scientists after the Haitian cholera outbreak of 2010 (13). It was recently discovered that one of the major phenotypic characteristics of the classical biotype strain, i.e., polymyxin B sensitivity, was also transmitted to the El Tor biotype strains circulating in Kolkata (30,31). A similar finding was made in this study with regard to the characteristic of the classical biotype strain, i.e., the inability to produce CTXU virions like the classical strains, although it involved a different mechanism. Thus, the recent trend of gaining classical biotype traits by El Tor biotype strains indicates that the new variant V. cholerae El Tor strains with hyperpathogenic characteristics adapt slowly to the environment and also evolve slowly; such strains can prove fatal to human beings and may lead to a more severe cholera outbreak situation in the near future worldwide.

MATERIALS AND METHODS
Strains used in the study. The strains were isolated from patients with cholera in Kolkata, India, between 2007 and 2011 and were phylogenetically analyzed as described previously (15). Strains IDH-00113 (referred to here as strain 13) and IDH-02387 (strain 87) isolated in 2007 and 2009, respectively, belong to lineage 1, which was predominant in Kolkata until 2009. Strain IDH-03329 (strain 29), isolated in 2010, as well as IDH-03506 (strain 06) and BCH-01536 (strain 36), isolated in 2011, were classified into lineage 2, sublineages III and IV, respectively. It was revealed that lineages 1 and 2 were concurrently prevalent between 2007 and 2009, while lineage 2-III appeared in 2010, followed by the predominance of lineage 2-IV in 2011 and later (15).
Plasmid construction. The oligonucleotide primers used in the study are listed in Table 3. To construct the rstA expression plasmid with Ig-1 or control genomic sequence, the primer pairs P1/P2, P3/P4, P5/P6, and P7/P8 were used to amplify the Ig-1 of CTXU (Ig-1 up), Ig-1 of RS1 (Ig-1 down), an N16961 genomic region of approximately 1.5 Mbp (1.5 genome), and a 1.1-Mbp (1.1 genome) region of the larger chromosome, respectively. Inverse PCR was performed using the primer pair P9/P10 and pET-21a as a template, followed by ligation using a seamless ligation-independent cell lysate (32) with the Ig-1 up or 1.5 genome fragment. Inverse PCR was performed again using the resulting plasmids as a template, with the primer pair P11/P12 ligated with the Ig-1 down or 1.1 genomic fragment. The resulting plasmids were digested using NdeI/XhoI and ligated with an NdeI/XhoI-digested rstA fragment, which was amplified with primer pair P13/P14. Constructed plasmids were verified via sequencing. The plasmids were propagated in Escherichia coli DH5a, and rstA expression was induced in E. coli BL21.
Genome sequencing. The genomic DNA of V. cholerae strains was extracted using the DNeasy blood and tissue kit (Qiagen) as per the manufacturer's instructions. Nanopore-based DNA sequencing was performed using the Native Barcoding Expansion 1-12 (EXP-NBD104; Oxford Nanopore Technologies [ONT, Oxford, UK]) and a ligation sequencing kit (SQK-LSK109; ONT), and DNA was loaded onto the MinION sequencing apparatus flow cell (R9.4.1; FLO-MIN106D; ONT) as per the manufacturer's instructions. The raw reads were base called (i.e., electronic signals were converted to the corresponding base sequence of the DNA strand) using Albacore software (ONT). The DNA sequence reads obtained were separated on the basis of the barcode sequence of each strain, and the adapters were trimmed off by using the Porechop software (33). Circular chromosomal sequences were assembled from the obtained reads with at least 30-fold coverage of the V. cholerae genome using flye (34) or unicycler (35), and the sequences were then polished by short reads (15) using Pilon software (36).