Journal of Bacteriology, December 2003, p . 7231-7240, Vol . 185, No . 24

Classical strains of V . cholerae contain nonfunctional CTX prophages, whereas El Tor and O139 strains contain fully active prophages that produce infective CTX viral particles (10) . CTX site specifically integrates into the host chromosome by a process dependent on the host recombinases XerC and XerD, which ordinarily catalyze the resolution of chromosome dimers at the dif recombination site (25) . Other filamentous phages such as f237 of Vibrio parahaemolyticus, Lf of Xanthomonas campestris, Xff1 of Xilella fastidiosa, CUS-2 of Yersinia pestis, and VGJ of V . cholerae seem to exploit the XerCD recombination system to integrate into the chromosome of their hosts, suggesting that lysogenic filamentous phages are more common than initially thought (5, 25, 26) .

CTX infects V . cholerae through toxin-coregulated pilus (TCP) (49), a type IV pilus essential for intestinal colonization (47) that is encoded by a gene cluster contained in the V . cholerae pathogenicity island (VPI) (33) . Although VPI seems to move horizontally between bacterial populations of V. cholerae, the transfer mechanism is still controversial . Karaolis et al . presented data suggesting that VPI is the prophage state of another filamentous phage, VPI, which they thought to use TcpA as major capsid protein (34); however, this hypothesis has raised several unanswered questions (11), and the results from that study could not be reproduced by other researchers (16) . Perhaps VPI transmission is mediated by several mechanisms that follow different pathways; for example, O'Shea and Boyd have found that VPI can be mobilized by the generalized transducing phage CP-T1 (43); however, the main mechanism accounting for VPI transmission probably has not been discovered yet . Whatever the mechanism, VPI, carrying the CTX receptor, also has the ability to move horizontally between bacterial populations, providing CTX with the advantage of amplifying its host range when VPI moves toward new TCP-negative strains .

An evolutionary model for the origin of pathogenic V . cholerae has been proposed in which this bacterium first acquired the tcp operon and then TCP-producing strains were infected and lysogenized by CTX (2, 11, 17, 18, 49) . However, this model has been impugned by the isolation of several strains of V. cholerae (both O1 and non-O1) that lack TCP genes but contain the CTX prophage (20, 45) . It has been suggested that such strains arose by the mentioned model with a subsequent loss of the VPI (17, 18) . Another possibility is that such strains have alternative CTX receptors (2, 3), but filamentous phages that use more than one receptor have not been described . However, CTX can infect V . cholerae in a TCP-independent fashion that requires the TolQ, TolR, and TolA proteins, but the efficiency of infection by this mechanism is quite low and needs direct cell-cell contact (23). Boyd and Waldor demonstrated that a specialized receptor such as TCP is not always essential for acquisition of CTX, since these authors showed that V . cholerae generalized transducing bacteriophage CP-T1 can transfer the whole CTX genome toward TCP-negative strains of V . cholerae (3) . However, the presence of TCP-devoid, non-O1, non-O139 strains of V . cholerae that contain CTX prophages cannot be explained by this mechanism, since CP-T1 only infects strains belonging to the O1 serogroup (21) .

We recently described a V . cholerae-specific filamentous phage, VGJ, which infects host cells through the mannose-sensitive hemagglutinin (MSHA) pilus and that is able to integrate its genome at the same attB chromosomal site as CTX (5) . We describe here a new type of specialized transduction mediated by VGJ through which CTX and its satellite phage RS1 are transmitted to new hosts in a TCP-independent fashion . We also discuss the potential implications of this alternative mechanism for the emergence of new toxigenic serotypes of V . cholerae and for the development of an environmentally safer live attenuated cholera vaccine .

 
Phage methods. Phage particles of VGJ, CTX-Kn, RS1-Kn, HybP-Kn, or HybRS-Kn were purified and concentrated as described in (5) . For infection assays, phage donor strains were grown until they reached an optical density (OD) at 600 nm of 1.5 . One-milliliter portions of the cultures were filtered through a 0.22-µm-pore-size filter (Sartorius), and 50-µl portions of the filtrates were plated on solid LB medium to check for sterility . Then, 100 µl of pure and diluted cell-free culture supernatants of the donor strains were mixed with 20 µl (10⁸ cells) of a fresh culture of the receptor strains grown under optimal conditions for TCP expression . The mixture was incubated 20 min at room temperature to allow infection and plated on solid LB medium supplemented with kanamycin to count kanamycin-resistant (Kn^r) transductants . Alternatively, infected cells were grown in LB broth to purify phage particles or single-stranded DNA (ssDNA) according to Sambrook et al. (46) .

DNA methods. Total DNA was prepared according to the method of Ausubel et al. (1) . Plasmid DNA was prepared by using the WizardPlus SV System (Promega) . DNA restriction and modification enzymes were used according to the manufacturer's instructions (Promega) . Southern blot analyses were performed with the following digoxigenin (DIG)-labeled probes: a 643-bp fragment containing part of ctxAB genes amplified by PCR with the primer pairs 5'-ATGATCATGCAAGAGGAACTC-3' and 5'-AGGTGTTCCATGTGCATATGC-3' was used as the ctxAB-specific probe; the 954-bp SacI-EcoRI fragment of VGJ RF, containing open reading frame 81 (ORF81), ORF44, ORF29, and part of ORF493, was used as the VGJ-specific probe (see the genome sequence of VGJ; GenBank no . AY242528); the 564-bp SacI-SphI fragment containing the rstC gene from plasmid pURS1 (4) was used as the rstC-specific probe; and, finally, the 2.9-kb EcoRI-PstI fragment containing the RS1 element from pURS1 (4) was used as the RS-specific probe .

Strand-specific probes were generated by asymmetric PCR with HybP-Kn RF as the template . Primer CTB1 (5'-GCGATTGAAAGGATGAAGG-3'), hybridizing with the negative strand of CTX and inside ctxB, was used to generate strand-specific HybP-A probe and primer NJ2 (5'-TAGAACGTGTCATTGCATCG-3'), hybridizing with the negative strand of VGJ and inside ORF136 (AY242528), was used to generate the complementary strand HybP-B probe . Nucleotides were added at the following final concentrations: 100 µM dATP, 100 µM dCTP, 100 µM dGTP, 65 µM dTTP, and 35 µM DIG-dUTP. Amplification reactions were performed with the Taq bead hot-start polymerase system (Promega) .

Sequences of the novel junctions between CTX and VGJ in HybP-Kn were obtained by sequencing its replicative form (RF) with primers CTB1 (see above) for the junction att-CTX/VGJ and NJ4 (5'-CCTGTAGAAATTCCGTCTCC-3'), hybridizing inside the intergenic region I of CTX for the junction att-VGJ/CTX . Similarly, the sequence of the novel junctions between VGJ and RS1 in HybRS-Kn was obtained by sequencing its RF with the primers NJ5 (5'-CGCTCATCAGGTTCAAAACC-3') for the att-RS1/VGJ junction and NJ4 for the att-VGJ/RS1 junction . Sequencing reactions were performed with the Thermo Sequenase CyS dye terminator kit and an ALFexpress DNA sequencer (Amersham Pharmacia Biotech) .

Stability of HybP-Kn in the bacterial host. To evaluate the in vitro stability of HybP-Kn, V . cholerae 1333 infected with this phage [1333(HybP-Kn)] was inoculated into LB medium and grown without kanamycin selection until late stationary phase . The ratio of infected cells (Kn^r) to the total number of bacteria was determined every 2 h for 24 h . At this time, the integrity of HybP-Kn RF was studied by restriction analysis of plasmid DNA preparations from 12 independent Kn^r clones .

The in vivo stability of replicating or integrated HybP-Kn was evaluated in BALB/c suckling mice . Ten mice were orogastrically inoculated with 10⁶ cells of 1333(HybP-Kn) or 1333-HybI2 (an integrant of HybP-Kn in 1333) and then incubated 24 h at 30°C without their mothers . The mice were then sacrificed; the small intestines were homogenized to recover colonizing vibrios, and the ratio of Kn^r cells to the total number of bacteria was determined .

Quantification of CT production. The ability of 1333(HybP-Kn) and 1333-HybI2 to produce CT was determined in AKI (29) and LB media by the G_M1 ganglioside-dependent enzyme linked immunosorbent assay (G_M1-ELISA) (24) by using a standard curve of purified CT and the anti-CTA monoclonal antibody (MAb) 1G10G5 .

Virulence evaluation of the HybP-Kn lysogen 1333-HybI2. Toxicity was evaluated by orogastric inoculation of groups of 15 BALB/c suckling mice with 10⁶ cells of the strains 1333-HybI2, 1333 (negative control), and C6706 (positive control) diluted in 50 µl of phosphate-buffered saline . Mice were fasted 4 h before and after inoculation and incubated for 6 days with their mothers . The survival was monitored daily during this lapse .

Protein methods. To determine the protein composition of the hybrid phage capsids, samples of HybP-Kn (10⁸ particles), HybRS-Kn (10⁸ particles), or VGJ as control (10⁹ particles) were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting (1) with the MAb 2H1H9 specific for the major capsid protein of VGJ phage .

Expression of MshA and TcpA protein subunits was studied by growing the strains of interest in LB medium (pH 6.5, 30°C), LB medium (pH 7.0, 37°C), TSB (37°C), or TSB_G (37°C) with overnight shaking (240 rpm) and by the AKI procedure (29) . For growth in DMEM or PFHM, starting suspensions of 10⁶ bacteria/ml in each medium were incubated for 8 h static at 37°C in an atmosphere of 5% CO₂ . Whole-cell lysates containing 2 x 10⁷ bacteria grown in each condition were fractionated by SDS-PAGE and analyzed by Western blot with the anti-TcpA MAb 10E10E1 and the anti-MshA MAb 2F12F1 (15) .

The nucleotide sequences of the novel junctions attCTX/VGJ, attVGJ/CTX, attRS1/VGJ, and attVGJ/RS1 have been deposited in the GenBank database under the accession numbers AY268047, AY268048, AY349613, and AY349614, respectively .

To test this hypothesis, 569B was infected with VGJ, CTX-Kn, or coinfected with both phages . The coinfected cells were grown under kanamycin selection to warrant the permanency of CTX-Kn, whereas VGJ was self-maintained because it is a very prolific phage and its receptor, MSHA, is constitutively expressed in 569B (see below) . Infection assays were done with cell-free culture supernatants of the classical biotype strain 569B infected with VGJ or CTX-Kn or coinfected with both phages and by using as receptor strains the TCP mutant KHT52 (48), the isogenic MSHA mutant KHT46 (48), their parental strain C6706, the vaccine strain Peru-15, and the classical biotype strain 569B; the last three strains express both TCP and MSHA (Table 2) . As expected, 569B infected with CTX-Kn transmitted this phage only to TCP-expressing strains not restricted by phage immunity (only to 569B and Peru-15 receptors in Table 2) . Although KHT46 and C6706 strains expressed TCP, they were resistant to CTX because they contain CTX and RS1 prophages expressing RstR^{El Tor} repressor, which confers immunity to CTX^{El Tor} infection (36) . 569B infected with VGJ was able to transmit this phage to all strains, except the MSHA mutant KHT46, but this event of transfer never provided Kn^r to the transductants (Table 2) . However, 569B coinfected with CTX-Kn and VGJ transduced the Kn^r gene to all MSHA-expressing strains (KHT52, C6706, Peru-15, and 569B in Table 2) . Thus, the Kn^r gene originally carried by CTX-Kn was transduced to the TCP mutant KHT52 with great efficiency, acquiring the pattern of transmission of VGJ . Since KHT52 cannot produce TCP (the receptor of CTX-Kn), but normally assembles MSHA pilus (the receptor of VGJ), the appearance of KHT52 Kn^r transductants could only be explained by some kind of interaction between VGJ and CTX-Kn .

 
Plasmid DNA preparations from 24 Kn^r transductants of KHT52 from three independent experiments digested with EcoRI, which cuts once into VGJ RF but does not cut into CTX-Kn RF, produced two bands of ca . 15.7 and 7.5 kb, respectively (not shown) . Analysis by Southern blot with two DIG-labeled probes specific for VGJ and CTX showed that the smaller 7.5-kb band hybridized only with the VGJ-specific probe while the larger 15.7-kb band hybridized with both the VGJ- and the CTX-specific probes (not shown) . These results indicated the simultaneous presence of two plasmids: the VGJ RF of 7.5 kb and a larger plasmid whose size (15.7 kb) was compatible with a cointegrate structure comprising both VGJ and CTX-Kn RFs fused into a single recombinant molecule, thus supporting our initial hypothesis . We designated the larger new plasmid pHybP-Kn (from hybrid phage) . Electroporation of these plasmid DNA preparations (containing both plasmids) into KHT46 allowed us to propagate and purify clones containing pHybP-Kn alone, giving final confirmation of its existence (VGJ was washed out due to lack of its receptor, MSHA, in this strain) .

Restriction analysis of pHybP-Kn produced bands characteristic of CTX or VGJ and new bands corresponding to the novel junctions between both phage DNAs (Fig . 1AI) . In addition, pHybP-Kn hybridized with VGJ- and CTX-specific probes in Southern blots (Fig. 1AII and III), indicating that this molecule contained sequences from both phages . These results were absolutely compatible with the theoretic cointegrate structure of pHybP-Kn derived from our initial hypothesis in which the RFs of VGJ and CTX-Kn recombined site specifically through their respective attP sites (Fig. 1B) . Finally, DNA sequencing of the novel junctions of pHybP-Kn confirmed our hypothesis and showed that the genome of both phages were opposite in the hybrid RF (Fig. 1C) .

 
KHT46 carrying pHybP-Kn transmitted this hybrid molecule to several MSHA-expressing V . cholerae strains, indicating that pHybP-Kn was generating an ssDNA genome that was being packaged and exported in a new phage particle designated HybP-Kn . KHT46 produced ca . 10⁷ HybP-Kn particles/ml/OD unit of culture when the titers were measured by using 569B as receptor strain . In general, the titers of HybP-Kn ranged from 10⁶ to 10⁸ phage particles/ml/OD unit of cultures, depending on the donor and receptor strain used to measure phage titer (not shown) . This value was intermediate between titers of CTX (10⁵ to 10⁶) and VGJ (10¹⁰ to 10¹¹) when measured under the same conditions .

RS1 transmission mediated by VGJ. The RF of the satellite phage RS1 contains an attP site nearly identical to that of CTX; we therefore evaluated whether VGJ could also transmit the satellite phage RS1 according to the same procedure described above to assess CTX transmission . Thus, RS1-Kn was used in place of CTX-Kn to coinfect strain 569B, together with VGJ (Table 2) . As occurred with CTX-Kn, RS1-Kn was efficiently transmitted to the TCP mutant KHT52 (Table 2), suggesting that a hybrid phage was also formed by recombination between RS1 and VGJ RFs . Here again, the analysis of several KHT52 transductants by restriction and Southern blotting (not shown) confirmed that the RFs of RS1-Kn and VGJ had recombined site specifically to form a hybrid RF designated pHybRS-Kn, which generated an ssDNA hybrid genome that was packaged and exported in a phage particle designated HybRS-Kn. The analysis of one representative clone of pHybRS-Kn purified from KHT46 (as described above for pHybP-Kn) is shown in Fig. 1D . Sequencing of the novel junctions between RS1 and VGJ DNAs (Fig. 1F) confirmed the structure of HybRS-Kn RF and that RS1 and VGJ DNAs were opposite in the hybrid molecule (Fig. 1E) .

VGJ-dependent transmission of CTX and RS1 from their lysogenic state. Since the initial source of HybP-Kn (and HybRS-Kn) was 569B coinfected with both CTX-Kn (or RS1-Kn) and VGJ, where both phages were replicating, we studied HybP-Kn and HybRS-Kn production from El Tor strains with resident Kn^r-labeled CTX or RS1 prophages after infection with VGJ . Thus, we used strains C72K7 and N16K38, which normally produce CTX-Kn and RS1-Kn particles, respectively, from the resident prophages (Fig. 2) . When these strains were infected with VGJ, they detectably produced HybP-Kn or HybRS-Kn, albeit with significantly reduced titers (Table 2) . However, once the few HybP-Kn or HybRS-Kn particles produced by these strains infected another receptor like KHT52, they could replicate and produce large amounts of hybrid phage particles (10⁷ particles/ml/OD unit) from the new host .

 
DNA and protein composition of HybP-Kn and HybRS-Kn particles. Genomic ssDNA preparations of HybP-Kn particles obtained from KHT46(HybP-Kn) could only be sequenced with oligonucleotides hybridizing with the positive DNA strand of VGJ or the negative DNA strand of CTX; also, this genomic DNA hybridized only with the HybP-A probe, which is specific for the positive strand of VGJ and the negative strand of CTX but not with the reverse probe HybP-B (Fig . 3A to C) . These results were identical for HybRS-Kn (not shown), clearly indicating that the replication machinery of VGJ prevailed by far over that of CTX during replication of the hybrid phages .

 
HybP-Kn and HybRS-Kn particles profiled by SDS-PAGE to study their protein contents revealed a protein profile identical to that of VGJ (5) . The major capsid protein of VGJ was evident in silver-stained gels (not shown) when HybP-Kn or HybRS-Kn particles (10⁸) were electrophoresed, and its identity was confirmed by reaction with a MAb specific for this protein (Fig. 3D), indicating that the hybrid phage capsid is mostly (if not totally) composed by the same proteins of VGJ capsid .

Stability of HybP-Kn. The time course stability and structural integrity of HybP-Kn studied during in vitro growth of 1333(HybP-Kn), in the absence of kanamycin selection, demonstrated that HybP-Kn preserved the same recombinant structure and remained present in about one-third of the bacterial population until the late stationary phase (24 h) (not shown) .

Colonization of 1333(HybP-Kn) in the suckling mice intestines also supported the permanency of HybP-Kn in a similar fraction of the bacterial population at 24 h of colonization but, in contrast to the in vitro condition, the in vivo conditions promoted site-specific integration of pHybP-Kn into an as-yet-unidentified chromosomal site different from the integration site of VGJ, as revealed by the unique Southern blot banding pattern in a significant number of clones tested (see the analysis of one representative clone in Fig. 4) . One of these in vivo isolated Kn^r clones, 1333-HybI2, extensively cultured in vitro and in vivo showed that all vibrios in the bacterial population retained the Kn^r trait . Southern blot analysis of 10 clones isolated from both culture conditions revealed the same original prophage structure of 1333-HybI2 (not shown) .

 
Therefore, the replication and packaging processes used by HybP-Kn (which are the same used by VGJ) preserved the structure of the hybrid recombinant genome during cell growth and sustained the inheritance of the hybrid phage by a significant fraction of the cell population in vitro . In vivo, the inheritance of the recombinant phage was guaranteed by the site-specific integration of HybP-Kn into the bacterial chromosome also in a significant fraction (about one-third) of the bacterial population . Once HybP-Kn integrated it was stably maintained in the bacterial chromosome .

Expression of CT genes transduced by HybP-Kn. 1333(HybP-Kn) and 1333-HybI2 cultured in AKI medium produced 80 to 90 ng of CT/ml as measured by G_M1-ELISA . This amount is similar to the CT levels produced by its toxigenic progenitor C6706 in similar condition but contrasted with the null production of CT by the attenuated vaccine strain 1333. Thus, ctxAB genes are actively expressing CT from the replicating and lysogenic state of HybP-Kn in the AKI condition . Curiously, when CT production was measured in LB medium, 1333(HybP-Kn) produced 60 to 70 ng of CT/ml; however, 1333-HybI2, as well as the control parental strain C6706, did not produce CT under this condition . Thus, CT expression from HybP-Kn behaved in a different manner when HybP-Kn DNA was in the replicative or integrated state showing the same pattern of CT expression from CTX (38) . These results suggest that ctxAB genes are not under the control of the ToxR regulon when HybP-Kn is in the replicative state but that they are regulated by this regulatory system when HybP-Kn is integrated .

HybP-Kn mediates virulence conversion. As expected, attenuated strain 1333 was not lethal at all to suckling mice; however, strain 1333-HybI2 inoculated orogastrically into suckling mice at a dose of 10⁶ CFU reproduced the toxigenic behavior of the pathogenic parental C6706 in the kinetics of mouse killing (Fig. 5) . At day 6 no mice survived inoculation with 1333-HybI2 or C6706 (Fig. 5), indicating that infection and lysogenization with HybP-Kn have full potential for virulence reversion of attenuated vaccine strains such as 1333 .

 
HybP-Kn host range. The host range of HybP-Kn was the same of VGJ (5) and was unaffected by the expression or not of TCP, indicating that HybP-Kn entry into host cells depended only on the MSHA (Table 1) . In addition, HybP-Kn infected most strains tested more efficiently than CTX, even under optimal conditions for TCP expression (Table 1) .

MshA versus TcpA expression. To compare the in vitro relative expression of HybP-Kn and CTX receptors, we analyzed the expression of their major pilus subunits, MshA and TcpA, in classical, El Tor, and O139 strains under different culture conditions by Western blotting . Although in vitro TcpA expression was detected only in restricted conditions and not in all strains, MshA was expressed at similar levels in all strains and conditions tested (Table 3) . These results indicated that MSHA is very likely a more ubiquitously expressed receptor than TCP .

 
The present study describes a novel type of specialized transduction for transmission of CTX or its satellite phage RS1 between V . cholerae strains, which is mediated by the filamentous phage VGJ . Our results strongly suggest that this mechanism consists in the site-specific recombination of the RF of VGJ and CTX (or RS1) to form a single-hybrid molecule in a donor V. cholerae cell . This hybrid RF generates an ssDNA genome that is packaged and exported in hybrid phage particles designated HybP (for VGJ/CTX hybrid) or HybRS (for VGJ/RS1 hybrid) . HybP and HybRS are recombinant and stable phages whose genomes consist of the positive strand of VGJ and the negative strand of CTX or RS1, respectively, linked by their respective attP sites . Although it is possible that the phages first integrate in tandem in the bacterial chromosome and then excise to form the hybrid phage, this mechanism is very unlikely because (i) integration of VGJ has not been detected in vitro, a condition in which the hybrid phages are obtained, and (ii) strains in which VGJ and CTX/RS1 are integrated in tandem (5) do not produce hybrid phage particles (data not shown) . In addition, the hybrid phages are efficiently produced by classical strains in which VGJ and CTX (or RS1) are replicating in the same host cell (Table 2) . However, the hybrid phages are produced at very low levels by El Tor strains of V. cholerae (in which CTX and RS1 are integrated) when these strains are infected with VGJ (Table 2), a finding which suggests that the extrachromosomal RFs of VGJ and CTX (or RS1) are needed for production of the hybrid phages by site-specific recombination .

This mechanism of specialized transduction constitutes a novel TCP-independent pathway for CTX or RS1 transmission and dissemination . The processes of replication and assembly of HybP and HybRS, as well as their capsid, are essentially those of VGJ; thus, HybP and HybRS also use MSHA as receptor . The MSHA is probably a more available phage receptor than TCP since expression of MshA protein is not restricted to specific growth conditions, as TcpA is, and the MSHA has been found also in non-O1 non-O139 strains of V . cholerae (14) . Thus, a hybrid phage such as HybP could have advantage over CTX for transmitting the CT genes under certain conditions, such as those found in the environment . Interestingly, HybP infected all El Tor and O139 strains tested in the present study more efficiently than CTX, even under optimal conditions for TCP expression (Table 1); therefore, phage HybP (containing a CTX^{El Tor} genome) circumvents phage immunity (36), supporting the idea that the replication machinery of VGJ is leading replication of HybP . Although the differences between the efficiency of infection of classical strains with HybP and CTX are less pronounced because these strains are not immune to CTX^{El Tor} (36), these strains were also infected more efficiently by HybP than by CTX-Kn (Table 1), which likely indicates that HybP tranduces CT genes more efficiently than CTX, the ordinary vehicle of these genes, in all conditions tested .

CTX transmission mediated by a VGJ-like phage is a plausible explanation for the occasional emergence of non-O1, non-O139 TCP-devoid toxigenic strains of V . cholerae. It is highly probable that other environmentally circulating V. cholerae filamentous phages can potentially transduce CTX or RS1 by a mechanism similar to the one described here . As a supporting example, phage fs-2 contains a 715-bp fragment that is highly homologous (97% identity) to part of RS1 satellite phage, flanked at one side by an attRS1 (27), strongly suggesting that fs-2 and RS1 recombined once upon their evolutionary history . In addition, it has been recently found that RS1 is transduced by the filamentous phage KSF by an unknown TCP-independent mechanism (19) that perhaps is the same mechanism described in the present study . More recently, we isolated another related filamentous phage, designated VEJ, from the V . cholerae O139 strain MO45, which is also able to transmit CTX by the mechanism described here mediated by VGJ (results not shown) . Phages such as VGJ and VEJ seem to be relatively abundant in strains of the O139 serogroup of V . cholerae, since other related filamentous phages have been isolated from strains of this serogroup, such as VSK, fs1, and 493 (30, 31, 32, 42) .

Consequently, in Fig. 6 we propose an evolutionary model that could explain the origin of new virulent strains from nontoxigenic environmental strains of V. cholerae or from another related bacterial species . Acquisition of CTX phage by a nontoxigenic V . cholerae strain does not automatically create a new pathogen since other factors besides CT are needed for manifestation of the full virulence phenotype (12, 13, 37) . Among these factors are TCP, which allows virulent strains to colonize the human small intestine (47), and the lipopolysaccharide, since only two serogroups of V . cholerae (O1 and O139) are known to cause epidemic cholera (7, 28) . Therefore, for a phage such as HybP to convert a nontoxigenic V. cholerae strain into a full pathogen it is necessary that the phage infect a host cell with adequate virulence factors . One possibility is that these hybrid phages infect V . cholerae strains (non-O1, non-O139) which do not contain CT encoding genes but which contain VPI and thereby the TCP gene cluster and toxT gene, whose product is needed to activate TCP and CT expression. Although, they are not the rule, such strains have been isolated, including some with new variant alleles of toxT and tcpA (6, 20, 41) . If such strains additionally have the MSHA gene cluster, they could express this pilus in a broader number of conditions than TCP, giving advantage to a phage such as HybP over CTX to infect them . Also, the possibility that a phage such as HybP infects V. cholerae strains with other unknown virulence attributes, such as a colonization factor different from TCP cannot be absolutely excluded. Furthermore, another possibility for the emergence of new pathogens is that HybP-like phages infect other bacterial species expressing MSHA homologous pili where the hybrid phage could integrate and express CT . The homology between the putative adsorption proteins of phages VGJ and Vf33, which infects the enteropathogen Vibrio parahaemolyticus, suggests that this bacterium species expresses a pilus homologous to MSHA; in fact, VGJ seems to be more related to Vf33 than to other V . cholerae phages such as CTX or fs2 (5) .

 
More tangible is the fact that live attenuated vaccine strains, which contain MSHA and TCP gene clusters, can potentially reacquire ctxAB genes in the environment by infection with a phage such as HybP. Infection with HybP mediates virulence conversion of the V . cholerae vaccine attenuated strain 1333, in which this phage is able to integrate into the chromosome during in vivo growth in the small intestines of suckling mice, stabilizing the inheritance of CT genes . In contrast to VGJ and CTX, which integrate preferentially at the right flanking end repeat of RS1 (attRS1) (5, 8), HybP integrates at an as-yet-unidentified chromosomal site of 1333 (Fig. 4), indicating that the recombinant nature of HybP shifts integration specificity. Therefore, previously conceived mutations, such as the deletion of the attRS1 site (44), directed to avoid stable reversion to virulence of live attenuated vaccine strains, could be ineffective to prevent integration of HybP-like phages. Thus, it would be desirable that live attenuated vaccine strains do not express MSHA to diminish the risk of virulence reversion in the environment after the release of these strains during massive vaccination campaigns . In addition, it would be desirable that vaccine strains lack any VGJ-related phage (replicating or lysogenized) to avoid dissemination of CTX in case it is reacquired .

The present study emphasizes the importance and versatility of horizontal gene transfer mechanisms for the evolution of bacterial pathogens . To our knowledge, this is the first report of the transmission of one filamentous phage genome by another filamentous phage, which constitutes an example of how mobile genetic elements can subtly interact to expand the pathways for gene transfer . This finding should lead to future studies about the interaction between filamentous phages in particular and between mobile genetic elements in general for horizontal gene transfer in bacteria .

1. Ausubel, F . M., R . Brent, R . E . Kingston, D . D. Moore, J . G . Seidman, J . A . Smith., and K. Struhl. 1995 . Short protocols in molecular biology. John Wiley & Sons, Inc., New York, N.Y.
2. Boyd, E . F., B . M . Davis, and B . Hochhut.2001 . Bacteriophage-bacteriophage interactions in the evolution of pathogenic bacteria . Trends Microbiol. 9:137-144.
3. Boyd, E . F., and M . K . Waldor. 1999. Alternative mechanism of cholera toxin acquisition by Vibrio cholerae: generalized transduction of CTX by bacteriophage CP-T1 . Infect . Immun. 67:5898-5905.
4. Campos, J., R . Fando, A . Silva, B . L . Rodriguez, and J . A. Benitez. 1998 . Replicating function of the RS1 element associated with Vibrio cholerae CTX prophage.FEMS Microbiol . Lett. 164:141-147.
5. Campos, J., E . Martínez, E . Suzarte, B . L . Rodríguez, K . Marrero, T . Ledón, R . Del Sol, and R . Fando.2003 . VGJ, a novel filamentous phage of Vibrio cholerae, integrates into the same chromosomal site as CTX . J . Bacteriol. 185:5685-5686.
6. Chakraborty, S., A . K . Mukhopadhyay, R . K . Bhadra, A. N . Ghosh, R . Mitra, T . Shimada, S . Yamasaki, S . M . Faruque, Y . Takeda, R . R . Colwell, and G . B . Nair.2000 . Virulence genes in environmental strains of Vibrio cholerae. Appl . Environ . Microbiol. 66:4022-4028.
7. Chitnis, D . S., K . D . Sharma, and R . S . Kamat.1982 . Role of somatic antigen of Vibrio cholerae in adhesion to intestinal mucosa . J . Med. Microbiol. 15:53-61.
8. Davis, B . M., H . H . Kimsey, W . Chang, and M . K. Waldor. 1999 . The Vibrio cholerae O139 Calcutta bacteriophage CTX is infectious and encodes a novel repressor . J . Bacteriol. 181:6779-6787.
9. Davis, B . M., H . H . Kimsey, A . V . Kane, and M . K . Waldor. 2002 . A satellite phage-encoded antirepressor induces repressor aggregation and cholera toxin gene transfer . EMBO J. 21:4240-4249.
10. Davis, B . M., K . E . Moyer, E . F . Boyd, and M . K . Waldor. 2000 . CTX prophages in classical biotype Vibrio cholerae: functional phage genes but dysfunctional phage genomes . J . Bacteriol. 182:6992-6998.
11. Davis, B . M., and M . K . Waldor. 2003. Filamentous phages linked to virulence of Vibrio cholerae.Curr . Opin . Microbiol. 6:35-42.
12. DiRita, V . J., N . C . Engleberg, A . Heath, A . Miller, J . A . Crawford, and R . Yu. 2000 . Virulence gene regulation inside and outside . Philos . Trans . R . Soc . Lond. B Biol . Sci. 355:657-665.
13. DiRita, V . J., C . Parsot, G . Jander, and J . J. Mekalanos. 1991 . Regulatory cascade controls virulence in Vibrio cholerae. Proc . Natl . Acad . Sci . USA 88:5403-5407.
14. Ehara, M., S . Shimodori, F . Kojima, Y . Ichinose, T . Hirayama, M . J. Albert, K . Supawat, Y . Honma, M . Iwanaga, and K . Amako.1997 . Characterization of filamentous phages of Vibrio cholerae O139 and O1 . FEMS Microbiol . Lett. 154:293-301.
15. Falero, G., B . L . Rodriguez, T . Valmaseda, M . E . Perez, J . L . Perez, R . Fando, A . Robert, J . Campos, A . Silva, G. Sierra, and J . A . Benitez. 1998 . Production and characterization of a monoclonal antibody against mannose-sensitive hemagglutinin of Vibrio cholerae. Hybridoma 17:63-67.
16. Faruque, A . S., J . Zhu, Asadulghani, M . Kamruzzaman, and J. J . Mekalanos. 2003 . Examination of diverse toxin-coregulated pilus-positive Vibrio cholerae strains fails to demonstrate evidence for Vibrio pathogenicity island phage.Infect . Immun. 71:2993-2999.
17. Faruque, S . M., M . J . Albert, and J . J. Mekalanos. 1998 . Epidemiology, genetics, and ecology of toxigenic Vibrio cholerae. Microbiol . Mol . Biol. Rev. 62:1301-1314.
18. Faruque, S . M., Asadulghani, M . N . Saha, A . R. Alim, M . J . Albert, K . M . Islam, and J. J . Mekalanos. 1998 . Analysis of clinical and environmental strains of nontoxigenic Vibrio cholerae for susceptibility to CTX: molecular basis for origination of new strains with epidemic potential . Infect . Immun. 66:5819-5825.
19. Faruque, S . M., M . Kamruzzaman, Asadulghani, D . A . Sack, J . J . Mekalanos, and G . B . Nair.2003 . CTX-independent production of the RS1 satellite phage by Vibrio cholerae. Proc . Natl . Acad. Sci . USA 100:1280-1285.
20. Ghosh, C., R . K . Nandy, S . K . Dasgupta, G . B. Nair, R . H . Hall, and A . C . Ghose.1997 . A search for cholera toxin (CT), toxin coregulated pilus (TCP), the regulatory element ToxR and other virulence factors in non-O1/non-O139 Vibrio cholerae. Microb. Pathog. 22:199-208.
21. Guidolin, A., and P . A . Manning. 1985 . Bacteriophage CP-T1 of Vibrio cholerae: identification of the cell surface receptor . Eur . J . Biochem. 153:89-94.
22. Hanne, L . F., and R . A . Finkelstein.1982 . Characterization and distribution of the hemagglutinins produced by Vibrio cholerae. Infect. Immun. 36:209-214.
23. Heilpern, A . J., and M . K . Waldor. 2000. CTX infection of Vibrio cholerae requires the tolQRA gene products . J . Bacteriol. 182:1739-1747.
24. Holmgren, J. 1973 . Comparison of the tissue receptors for Vibrio cholerae and Escherichia coli enterotoxins by means of gangliosides and natural cholera toxoid . Infect. Immun. 8:851-859.
25. Huber, K . E., and M . K . Waldor. 2002. Filamentous phage integration requires the host recombinases XerC and XerD . Nature 417:656-659.
26. Iida, T., K . Makino, H . Nasu, K . Yokoyama, K . Tagomori, A . Hattori, T . Okuno, H . Shinagawa, and T . Honda. 2002 . Filamentous bacteriophages of vibrios are integrated into the dif-like site of the host chromosome . J . Bacteriol. 184:4933-4935.
27. Ikema, M., and Y . Honma. 1998 . A novel filamentous phage, fs-2, of Vibrio cholerae O139 . Microbiology 144:1901-1906.
28. Iredell, J . R., U . H . Stroeher, H . M . Ward, and P . A . Manning. 1998 . Lipopolysaccharide O-antigen expression and the effect of its absence on virulence in rfb mutants of Vibrio cholerae O1 . FEMS Immunol . Med. Microbiol. 20:45-54.
29. Iwanaga, M., and K . Yamamoto. 1985 . New medium for the production of cholera toxin by Vibrio cholerae O1 biotype El Tor . J . Clin . Microbiol. 22:405-408.
30. Jouravleva, E . A., G . A . McDonald, C . F . Garon, M. Boesman-Finkelstein, and R . A . Finkelstein.1998 . Characterization and possible functions of a new filamentous bacteriophage from Vibrio cholerae O139.Microbiology 144:315-324.
31. Jouravleva, E . A., G . A . McDonald, J . W . Marsh, R . K . Taylor, M . Boesman-Finkelstein, and R . A. Finkelstein. 1998 . The Vibrio cholerae mannose-sensitive hemagglutinin is the receptor for a filamentous bacteriophage from V . cholerae O139 . Infect. Immun. 66:2535-2539.
32. Kar, S., R . K . Ghosh, A . N . Ghosh, and A . Ghosh.1996 . Integration of the DNA of a novel filamentous bacteriophage VSK from Vibrio cholerae 0139 into the host chromosomal DNA . FEMS Microbiol . Lett. 145:17-22.
33. Karaolis, D . K., J . A . Johnson, C . C . Bailey, E . C . Boedeker, J . B . Kaper, and P . R. Reeves. 1998 . A Vibrio cholerae pathogenicity island associated with epidemic and pandemic strains . Proc. Natl . Acad . Sci . USA 95:3134-3139.
34. Karaolis, D . K., S . Somara, D . R . Maneval, Jr., J. A . Johnson, and J . B . Kaper. 1999 . A bacteriophage encoding a pathogenicity island, a type-IV pilus and a phage receptor in cholera bacteria . Nature 399:375-379.
35. Kenner, J . R., T . S . Coster, D . N . Taylor, A . F . Trofa, M . Barrera-Oro, T . Hyman, J . M . Adams, D . T . Beattie, K . P . Killeen, and D . R. Spriggs. 1995 . Peru-15, an improved live attenuated oral vaccine candidate for Vibrio cholerae O1.J . Infect . Dis. 172:1126-1129.
36. Kimsey, H . H., and M . K . Waldor. 1998. CTX immunity: application in the development of cholera vaccines . Proc . Natl . Acad . Sci . USA 95:7035-7039.
37. Klose, K . E. 2001 . Regulation of virulence in Vibrio cholerae. Int . J . Med. Microbiol. 291:81-88.
38. Lazar, S., and M . K . Waldor. 1998 . ToxR-independent expression of cholera toxin from the replicative form of CTX.Infect . Immun. 66:394-397.
39. Mekalanos, J . J. 1983 . Duplication and amplification of toxin genes in Vibrio cholerae. Cell 35:253-263.
40. Moyer, K . E., H . H . Kimsey, and M . K. Waldor. 2001 . Evidence for a rolling-circle mechanism of phage DNA synthesis from both replicative and integrated forms of CTX . Mol . Microbiol. 41:311-323.
41. Mukhopadhyay, A . K., S . Chakraborty, Y . Takeda, G . B . Nair, and D . E . Berg. 2001 . Characterization of VPI pathogenicity island and CTX prophage in environmental strains of Vibrio cholerae. J . Bacteriol. 183:4737-4746.
42. Nakasone, N., Y . Honma, C . Toma, T . Yamashiro, and M . Iwanaga.1998 . Filamentous phage fs1 of Vibrio cholerae O139 . Microbiol . Immunol. 42:237-239.
43. O'Shea, Y . A., and E . F . Boyd. 2002. Mobilization of the Vibrio pathogenicity island between Vibrio cholerae isolates mediated by CP-T1 generalized transduction . FEMS Microbiol . Lett. 214:153-157.
44. Pearson, G . D., A . Woods, S . L . Chiang, and J . J. Mekalanos. 1993 . CTX genetic element encodes a site-specific recombination system and an intestinal colonization factor . Proc . Natl . Acad . Sci . USA 90:3750-3754.
45. Said, B., H . R . Smith, S . M . Scotland, and B . Rowe.1995 . Detection and differentiation of the gene for toxin co-regulated pili (tcpA) in Vibrio cholerae non-O1 using the polymerase chain reaction . FEMS Microbiol. Lett. 125:205-209.
46. Sambrook, J., E . F . Fritsch, and T . Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed . Cold Spring Harbor Laboratory Press, Cold Spring Harbor Laboratory, N.Y.
47. Taylor, R . K., V . L . Miller, D . B . Furlong, and J . J . Mekalanos. 1987 . Use of phoA gene fusions to identify a pilus colonization factor coordinately regulated with cholera toxin . Proc . Natl . Acad . Sci . USA 84:2833-2837.
48. Thelin, K . H., and R . K . Taylor. 1996. Toxin-coregulated pilus, but not mannose-sensitive hemagglutinin, is required for colonization by Vibrio cholerae O1 El Tor biotype and O139 strains . Infect . Immun. 64:2853-2856.
49. Waldor, M . K., and J . J . Mekalanos. 1996. Lysogenic conversion by a filamentous phag