Molecular mechanism of acquisition of the cholera toxin genes.

One of the major pathogenic determinants of Vibrio cholerae, the cholera toxin, is encoded in the genome of a filamentous phage, CTXφ. CTXφ makes use of the chromosome dimer resolution system of V. cholerae to integrate its single stranded genome into one, the other, or both V. cholerae chromosomes. Here, we review current knowledge about this smart integration process.

Introduction

Most bacteriophages are detrimental to their host metabolism. However, phages also participate in the horizontal transfer of genes among bacteria because their genome can harbour other genes than those strictly required for their life cycle. This can be highly beneficial to the bacterial host. Indeed, many bacterial virulence factors are associated with phage-like DNA sequences. More strikingly, the exotoxins produced by many pathogenic bacteria are encoded in the genome of lysogenic phages. This is notably the case in Bordetella avium8. The integrated prophages harboured by these bacteria profit from the multiplication of their host in the environment, which is in turn favoured by the virulence factors they bring to their host.

The study of Vibrio cholerae, the agent of the deadly diarrhoeal disease cholera, provides a fascinating case of such a bacterium-phage co-evolution. V. cholerae is the host for a variety of phages, commonly known as vibriophages, which can be lytic, non-lytic, virulent or temperate9. On the one hand, phage predation of V. cholerae has been reported to be a factor that influences seasonal epidemics of cholera10. On the other hand, one of the major virulence factors of V. cholerae, cholera toxin, is encoded in the genome of an integrated prophage CTXΦ1112. Furthermore, different variants of the phage CTXΦ exist, which participate in the genetic diversity of epidemic causing cholera strains13–15. Two different attachment sites were found for this family of phages on the V. cholerae genome. They correspond to the dimer resolution sites of the two V. cholerae chromosomes, dif1 and dif216. Indeed, in contrast to most other lysogenic phages, such as bacteriophage λ17, CTXΦ does not encode its integrase, but makes use of XerC and XerD, the two host-encoded tyrosine recombinases that normally function to resolve chromosome dimmers18. This mode of integration is all the more intriguing since CTXΦ phages belong to the filamentous phage family, which are generally not lysogenic and which harbour a single stranded circular genome. Nevertheless, CTXΦ-like prophages were found integrated in the genome of several bacterial species, notably in pathogenic E. coli strains19 and in Yersinia pestis20. Finally, it is remarkable to observe that many filamentous phages and/or genetic elements other than CTXΦ seem to have hijacked the chromosome dimer resolution system of V. cholerae for integration. Thus, TLC21, VEJ22, VGJ23, VSK24, VSKK (AF452449), KSF-1F24, fs125, fs226, f23714, were all found to be integrated at dif1 and/or dif2. Such a diversity of elements has not been observed in any other genera than the vibrios. Together, these elements participate in the dissemination of virulence factors among V. cholerae strains112829 and in the emergence of new genetic variants of epidemic strains of V. cholera13. We review current knowledge on the integration mechanism of filamentous vibriophages that hijack the XerCD recombinases, with a special focus on CTXΦ.

CTXΦintegration mechanism: exception or new paradigm?

CTXΦ has a ~7-kb ss(+)DNA genome arranged in two modular structures, the “RS” and“core”. The core region harbours seven genes, which are psh, cep, gIIICTX, ace, zot, ctxA and ctxB. While the psh, cep, gIIICTX, ace and zot encoded proteins are needed for phage morphogenesis, the products of the ctxAB genes are not strictly required for the life cycle of the phage but are responsible for the severe diarrhoea associated with cholera11. Three proteins, designated as RstR, RstA and RstB, are encoded in RS. Genetic analyses indicated that RstA is essential for phage replication and that RstB plays a crucial role in integration30. RstR acts as a transcriptional repressor by inhibiting the activity of PrstA, the only phage promoter required for CTXΦ replication and integration30. Several CTXΦ have been reported. These can be classified into four families based on the sequence of their rstR gene. These categories were designated as CTXΦET, CTXΦCl, CTXΦClc and CTXΦEnv according to the host cells in which they were originally isolated31–33.

As mentioned earlier, the integration of CTXΦ into the V. cholerae genome depends on two host encoded tyrosine recombinases, XerC and XerD18. XerC and XerD normally serve to resolve circular bacterial chromosome dimers generated by RecA mediated homologous recombination by adding a crossover at a specific 28 bp site dif on the chromosome16. The dif sites consist of specific 11-bp binding sites for each of the two Xer recombinases, separated by a 6-bp central region34. These are generally located opposite to the origin of replication of bacterial chromosomes16. Two dif sites are present on the genome of V. cholerae, one for each of the two circular chromosomes of the bacterium35. Three different chromosome dimer resolution sites (dif1, dif2 and difG) have been identified among the different V. cholerae strains characterized to date36 (Table I).

Table I

Sequences of the chromosome dimer resolution sites found in V. cholerae strains

Site	Sequence
dif1	AGTGCGTATTA TGTATG TTATGTTAAAT
dif2	AATGCGTATTA CGTGCG TTATGTTAAAT
difG	AGTGCGTATTA GGTATA TTATGTTAAAT
Source: Ref. 36

The ssDNA (+) genome of CTXΦ harbours two dif like sites (attP1 and attP2). These are arranged in opposite orientation and are separated by ~90-bp DNA segment in the phage genome37. Integration of CTXΦ at the dif loci of V. cholerae depends on the formation of a forked hairpin structure of 150 bp in the region encompassing attP1 and attP2 in the (+) ssDNA genome38 (Fig. 1). The hybridization of attP1 and attP2 at the stem of this hairpin unmasks the phage attachment site, attP(+). Integration occurs, XerC and XerD recombine this site with one of the two dimer resolution sites harboured by the host cell. This process only requires the catalytic activity of XerC: a single pair of strands is exchanged, which results in the formation of a pseudo-Holliday junction.

Fig. 1

Schematic representation of the XerCD mediated site-specific recombination reaction between the single stranded (+) DNA genome of CTXΦ and V. cholerae dif1. Blue and green bases indicate XerC and XerD binding sites. Bases of the central region of these sites are shown in red. The recombination reaction stops after the exchange of a single pair of strands, which is catalyzed by XerC. Integration is completed when the resulting pseudo-Holliday junction needs to be processed by the host DNA replication and/or DNA repair machineries. Integration of the phage generates one new functional dif site and two non-functional dif like sequences, attP2 and attP1, on the host chromosome38.

A proof of principle for this mechanism of integration was originally obtained for the El Tor variant of CTXΦ and dif1 based on in vivo work performed in Escherichia coli and in vitro work performed with the E. coli Xer recombinases38. Later on, a sensitive and quantitative assay was developed to confirm the ssDNA(+) integration model of CTXΦET into the dif1 site of a V. cholerae El Tor strain36. This system was also used to define rules of compatibilities between the phage attachment sites harboured by the different CTXΦ variants characterized to date and their host dimer resolution sites36 : integration is solely determined by possibility to form Watson-Crick or w0 obble base pair interactions to stabilize the exchange of strands promoted by XerC-catalysis between the phage attachment site and its target dimer resolution site (Table II and Fig. 1). These rules explain how integration of CTXΦET is restricted to dif1, how CTXΦCl can target both dif1 and dif2, and how a third CTXΦ variant targets difG (Table II). This single stranded integration model is not restricted to CTXΦ. Analysis of the att P sites of CUS-1F and Ypf-F phages revealed features for direct ssDNA integration into the chromosome dimer resolution site harboured by their respective host cells38. Another family of mobile genetic element, the integrons, also integrates in the bacterial chromosome via a single stranded intermediate39.

Table II

Sequences of the dif-like sites harboured by CTXΦ variant

CTXΦ variant	attP sequence	Integration site	Accession number
El Tor	AGTGCGTATTA TGTGGCGCGGCA TTATGTTGAGG (attP1) AATGCGTATTA TACGCCA TTATGTTACGG (attP2)	dif1	VCU83796
Classical	AGTGCGTATTA TGTGGCGCGGCA TTATGTTGAGG (attP1) AATGCGTATTA CTCGCCA TTATGTTACGG (attP2)	dif1 dif2	AY349175
Calcutta	AGTGCGTATTA TGTGGCGCGGCA TTATGTTGAGG (attP1) AATGCGTATTA TACGCCA TTATGTTACGG (attP2)	dif1	AF110029
G	AGTGCGTATTA GGTGGTGCGGCA TTATGTTGAGG (attP1) AATGCGTATTA GGGGCA TTATGTTACGG (attP2)	difG	AF416590
Source: Ref. 40

Integration mechanism of CTXΦ-associated genetics elements

Several filamentous phages other than CTXΦ are found to be integrated at the dif loci of V. cholera132223. To date, there is no report about their particular integration mechanism. Like CTXΦ, they do not encode a dedicated recombinase. In addition, a 29-bp dif like sequence can be identified in many of them (Table III). It is, therefore, very likely that these phages take control of the host XerC and XerD recombinases to integrate into the genome of their host. However, the presence of a single putative XerCD binding site on their genome makes it unlikely that the ssDNA form of their genome is directly used as a substrate for integration. We rather favour a model in which the double stranded replicative form of these phages is used for integration (Fig. 2). We are currently investigating this model using the tools we have developed for the study of CTXΦ40.

Table III

Sequences of the dif-ured by other vibriophages

Phage	Genome size (kb)	attP sequence	Host	Integration site	Accession number
VEJ	6.8	ACTTCGCATTA TGTCGGC TTATGGTAAAA	V. cholerae	dif1	NC012757
VGJ	7.5	ACTTCGCATTA TGTCGGC TTATGGTAAAA	V. cholerae	dif1	AY242528.1
VSK	6.9	ACTTCGCAGTA TGTCGGC TTATGGTAAAA	V. cholerae	dif1	NC003327
VSKK	6.8	ACTTCGCATTA TGTCGGC TTATGGTAAAA	V. cholerae	dif1	AF452449
KSF1	7.1	UK	V. cholerae	UK	AY714348
fs1	6.3	UK	V. cholerae	UK	NC004306.1
fs2	8.6	AGTGCGTATTA TGTCGGC TTATGGTAAAA	V. cholerae	dif1	AB002632
f237	8.7	AGTGCGCATTA TGGGCGC TTATGTTGAAT	V. cholerae V. parahemolyticus	dif1	NC002362
UK, unknow; Source: Ref. 40

Fig. 2

Putative mechanism of lysogenic conversion by the second type of filamentous phages that are found integrated into the chromosomal dimer resolution sites of V. cholerae40.

Interestingly, the two TLC elements integrated in strain N16961 are flanked by the half of the dif sequence (TGTGCGCATTA TGTATG for one and AGTGCATATTA TGTATG for the other). It is, therefore, reasonable to argue that their integration might be linked to the activity of the Xer recombinases.

Future prospects

The particular mode of integration of CTXΦ raises several questions. First, the efficiency of integration of a circular single stranded DNA molecule harbouring the sole attachment site of CTXΦ is very low38. However, it becomes extremely efficient when the RS region of the phage is included36. One likely explanation is that constant production and/or stabilization of the phage single stranded circular genome compensate for the instability of single stranded DNA in bacterial cells. RstB, which has been shown to be a single stranded DNA binding protein41, could play a role in the stabilization of the integration substrate. Accordingly, its biochemical properties and sequence differ from those of the single stranded DNA binding protein encoded in the genome of VGJF, a phage that seems to rely on double stranded DNA integration40. Second, only one pair of strands is exchanged between the single stranded DNA genome of CTXΦand the double stranded DNA genome of its host, which leaves open the question of how the resulting pseudo-Holliday junction intermediate is processed. Is it stably maintained until the next round of bacterial DNA replication or processed by some host DNA repair machinery? What occurs when the replication fork collides against this unusual structure? Finally, it is intriguing that so many phages take advantage of the Xer recombination system of vibrios as compared to other bacterial species. We wonder if it could be related to the particular life style and environment of the vibrios and/or their particular genome structure and management.