Fuse or die: how to survive the loss of Dam in Vibrio cholerae



Dam methylates GATC sequences in γ-proteobacteria genomes, regulating several cellular functions including replication. In Vibrio cholerae, which has two chromosomes, Dam is essential for viability, owing to its role in chr2 replication initiation. In this study, we isolated spontaneous mutants of V. cholerae that were able to survive the deletion of dam. In these mutants, homologous recombination and chromosome dimer resolution are essential, unless DNA mismatch repair is inactivated. Furthermore, the initiator of chr2 replication, RctB, is no longer required. We show that, instead, replication of chr2 is insured by spontaneous fusion with chr1 and piggybacking its replication machinery. We report that natural fusion of chr1 and chr2 occurred by two distinct recombination pathways: homologous recombination between repeated IS elements and site-specific recombination between dif sites. Lastly, we observed a preferential fusion of the two chromosomes in their terminus of replication.


Most known epigenetic signals in bacteria are generated by DNA methylation and serve to regulate specific DNA-protein interactions (Casadesus and Low, 2006). In bacteria, DNA methylases are usually part of restriction-modification couples where methylation serves to distinguish between self and non-self DNA (Casadesus and Low, 2006). Nevertheless, some adenine methylases, with no cognate restriction enzymes, were shown to play major roles in the regulation of many cellular processes (Wion and Casadesus, 2006). Such an orphan methylase, Dam, is involved in the regulation of chromosome replication, DNA repair, and gene expression (Lobner-Olesen et al., 2005; Low and Casadesus, 2008).

In Escherichia coli, Dam specifically methylates the adenine of palindromic GATC sequences. These sequences are overrepresented at the replication origin, oriC (Casadesus and Low, 2006). The hemimethylated origin is immediately sequestered after replication by the hemimethylation-specific DNA binding protein SeqA (Boye et al., 2000). Sequestration by SeqA prevents reforming of the initiation complex and re-methylation by Dam for up to one-third of the cell cycle (Boye et al., 2000). Sequestration also contributes to initiation synchrony where, during rapid growth, the multiple origins of a duplicating chromosome will fire simultaneously (Lobner-Olesen et al., 2005). In dam mutants, newly replicated origins are devoid of hemimethylation marks and are therefore indistinguishable from the non-replicated origins. By consequence, the once-per-cell-cycle initiation from each origin is no longer guaranteed and initiation becomes asynchronous (Lobner-Olesen et al., 2005).

Dam hemimethylated DNA is also the marker that allows the Mismatch Repair (MMR) apparatus to discriminate between template DNA (methylated) and nascent DNA (non-methylated) strands. In E. coli, three proteins MutS, MutL and MutH of the MMR system are essential in detecting mismatches and directing the repair machinery to them. MutH marks which strand contains the mismatch by nicking the unmethylated strand of hemimethylated DNA (Barras and Marinus, 1989). In E. coli, dam mutants have a mutator phenotype mediated by the inability of the MMR system to discriminate between the two DNA strands (Marinus, 2010). The nascent error-containing strand can thereby be used as template resulting in the fixation of mutations in the genome. Lack of methylation also leads to the formation of single-strand breaks on both DNA strands which can be converted to double-strand breaks (DSBs) either by replication or by a second incision at the same unmethylated GATC site on the complementary strand (Nowosielska and Marinus, 2008). Since DNA DSBs are lethal, RecA, as well as other factors involved in DSBs repair (such as RecB, RecC, RuvA, RuvB, RuvC and PriA) are required for viability of dam mutants (Marinus, 2000; Nowosielska and Marinus, 2005). As expected from such observations, the inactivation of MMR in dam mutants allows them to survive without these additional DSBs repair factors.

Dam is involved in a number of cellular processes and its loss causes pleiotropic defects including cell death in certain species such as Vibrio cholerae (Julio et al., 2001). V. cholerae is a Gram-negative enteric pathogen and the causative agent of the diarrheal disease cholera. Its genome is distributed on two circular chromosomes of distinct size, genetic content and replication requirements (Heidelberg et al., 2000). Chr1, the larger 3 Mbp chromosome, harbors the majority of housekeeping genes (Heidelberg et al., 2000). Chr1 replication is initiated at ori1 by DnaA, the canonical initiator protein that promotes chromosome DNA replication in most bacteria (Duigou et al., 2006). Ori1 is similar to the E. coli chromosome origin of replication, oriC, and can functionally replace it (Demarre and Chattoraj, 2010). The smaller 1 Mbp chromosome chr2 contains a higher abundance of species-specific and unknown ORFs with commensurately fewer essential genes (Heidelberg et al., 2000). Chr2 is nevertheless a bona-fide chromosome and its loss is lethal (Yamaichi et al., 2007). The origin of replication of chr2 (ori2) resembles that of iteron-plasmids, and initiation is triggered by the binding of a vibrio-specific initiator, RctB, to iterons sites (Venkova-Canova and Chattoraj, 2011). Both ori1 and ori2 display an overrepresentation of GATC methylation sites which remain hemimethylated for an extended period of time (Demarre and Chattoraj, 2010). Moreover, loss of SeqA provokes a deregulation of replication timing for chr1 and chr2 (Demarre and Chattoraj, 2010) suggesting that both origins are subjected to post-replicative sequestration by SeqA similar to what has been described in E. coli (Lobner-Olesen et al., 2005). Dam exerts an additional essential function in the control of chr2 replication in that RctB binds only to fully Dam-methylated iterons (Demarre and Chattoraj, 2010). In absence of Dam methylation, RctB does not bind to iterons and replication initiation of chr2 at ori2 is blocked and is lethal for the cell (Demarre and Chattoraj, 2010; Val et al., 2012).

In this study, we isolated spontaneous mutants of V. cholerae that were able to survive in the absence of otherwise-essential Dam. Strikingly, these mutants were able to survive through the spontaneous fusion of chr2 with chr1. Fusion occurred by two distinct pathways: homologous recombination and site-specific recombination. Fusion of the two chromosomes preferentially occurred in their termini which may be explained by underlying cellular processes in this region of the chromosomes. Furthermore, these viable mutants provided us with a model to elucidate a role of Dam in DNA mismatch repair analogous to that of E. coli. Interestingly, chromosome dimer resolution appeared to be essential in dam mutants and is linked to mismatch repair.


Isolation of viable Vibrio cholerae dam mutants

As Dam is essential in V. cholerae, we deleted dam in the presence of a dam-complementing plasmid, pTs-PBADdam (pGD93, Table S2). pTs-PBADdam is a replication-temperature-sensitive plasmid expressing dam under the control of an arabinose-inducible/glucose-repressible promoter (Demarre and Chattoraj, 2010). Δdam-[pTs-PBADdam] grew normally under permissive conditions on LB plates while under Dam-depleting conditions, Δdam colonies appeared only after an extended 24-hour incubation time and were pinpoint and less numerous (Fig. 1A). These colonies could be successfully re-cultured under restrictive conditions on both solid and liquid media. The absence of Dam methylation activity in these mutants was confirmed by restriction of genomic DNA using Sau3AI, MboI and DpnI which cleave differentially at GATC sites depending on methylation state (Fig. 1B) (Palmer and Marinus, 1994). We measured the rate of appearance of viable Δdam mutants and obtained an unexpectedly high survival rate of 1.5 × 10−2 (Fig. 2A). We estimated the rate of point mutations by monitoring the frequency of rifampicin-resistant mutants arising in each population to be in the range of 10−8 for WT V. cholerae and 10−7 for a dam mutant, dam4 (Fig. 2B). This rate is still significantly lower than our observed rate of survival indicating that survival in the absence of Dam cannot be solely the result of point mutations. To characterize the additional mechanism(s) responsible for the survival of V. cholerae in the absence of Dam, we proceeded to further scrutinize one representative mutant, dam4.

Figure 1.

Isolation of V. cholerae mutants able to survive the loss of Dam.

A. Depletion of Dam in V. cholerae. Growth of Δdam[pTs-PBADdam] under permissive conditions (0.2% arabinose at 30°C, allowing dam expression and pGD93 replication) or restrictive conditions (1% glucose at 42°C, repressing dam and preventing pGD93 replication).

B. Absence of Dam activity in V. choleraedam mutants. Genomic DNA of WT (+) and dam4 (Δ) (a representative dam mutant) was digested with restriction enzymes that cleave differentially at GATC sites depending on their methylation states. Uncut DNA of WT and dam4 show a band above 10 kbp. Undigested dam4 shows smearing above 3 kbp indicative of DNA DSBs. Sau3AI, which cleaves all GATC sites, cuts WT and dam4 genomic DNA equally. DpnI, which only cleaves methylated GATC sites, cut only WT genomic DNA. MboI, which only cleaves unmethylated GATC sites, cut only dam4 genomic DNA. The molecular marker used is the GeneRuler 1 kb DNA ladder from Fermentas. Digested DNA was separated by agarose gel electrophoresis and visualized after Ethidium Bromide staining.

Figure 2.

Role of V. cholerae Dam in DNA mismatch repair.

A. Survival rate of various V. choleraedam derivative mutants after Dam depletion. Various dam derivative mutants carrying pTs-PBADdam were grown overnight in liquid culture under permissive conditions. The number of viable dam mutants was determined by plating the cells under restrictive conditions (1% glucose at 42°C) and the total number of cells expressing Dam from pTs-PBADdam was determined by plating the cells under permissive conditions (0.2% arabinose at 30°C). The bars indicate the survival rate which is the ratio of viable dam mutants with the total number of cells. The error bars indicate the standard deviations from at least 4 independent experiments.

B. V. choleraedam mutants have a mutator phenotype. The point mutation rate was determined from the rate of Rifampicin resistant (RifR) mutants in wild-type (WT), dam mutant (dam4) and dam allelic complemented mutant (dam4::CdamΔrctB). The error bars indicate standard deviations from three independent experiments.

C. recA and xerC are essential in V. choleraedam mutants. Growth of ΔrecA[pTs-PBADdam] and damrecA[pTs-PBADdam] (top) or ΔxerC[pTs-PBADdam] and damxerC[pTs-PBADdam] (bottom) under permissive conditions (0.2% arabinose at 30°C, allowing dam expression and pGD93 replication) or restrictive conditions (1% glucose at 42°C, repressing dam and preventing pGD93 replication).

recA deletion is lethal in V. cholerae dam spontaneous mutants

In E. coli dam mutants, MMR acts randomly if neither strand is methylated. This causes an increase in the rate of point mutations and can generate DNA DSBs (Glickman and Radman, 1980; Robbins-Manke et al., 2005; Nowosielska and Marinus, 2008). As a result, homologous recombination (HR) is required to repair these DSBs, thus rendering inactivation of RecA synthetically lethal with a dam deletion (Marinus, 2000). By comparison of the point mutation rates of WT, dam4 and a Dam complemented mutant dam4::CdamΔrctB, we found an order of magnitude increase in the mutation rate of dam4 (Fig. 2B). This suggests that Dam plays an important role in V. cholerae MMR. Pulsed-field gel electrophoresis (PFGE) analysis of dam4 showed smearing indicative of DNA degradation (Fig. 3C, lane 3). Deletion of recA in dam4 was not possible unless the strain carried the Dam-expressing plasmid, pTs-PBADdam. Additionally, no plasmid segregates could be recovered by propagation of dam4ΔrecA-[pTs-PBADdam] under restrictive conditions (Fig. 2C, top), indicating that recA is essential in dam mutants of V. cholerae. We deduce that in V. cholerae dam mutants, HR is also required for viability, likely for the repair of DNA DSBs.

Figure 3.

Fusion of chr1 and chr2 in V. choleraedam mutants. Orange indicates chr1 sequence and green, chr2.

A. Chromosome structure of the WT V. cholerae (N16961). Ovals and triangles respectively indicate the origins of replication and dif sites at which XerCD site-specific recombination occurs. The genes encoding the initiators of chr1 (dnaA) and chr2 (rctB) are indicated with black arrows.

B. rctB is epistatic to dam deletion in V. cholerae. Growth curves of wild-type (WT), dam mutant (dam4), rctB mutant of dam4 (damrctB), Dam allelic complementation of damrctB (dam4::CdamΔrctB). OD was measured at 450 nm every 10 min during exponential phase.

C. Fusion of chr1 and chr2 in V. choleraedam mutants. Ethidium bromide stained PFGE of native genomic DNA: Lane 1, H. wingei marker Bio-Rad (DNA size: 1.05,1.37, 1.66, 1.81, 2.35, 2.70 and 3.13 Mbp); Lane 2, WT (chr1: 2.96 Mbp; chr2: 1.07 Mbp); Lane 3, dam4; Lane 4, dam4ΔrctB; Lane 5, dam4::CdamΔrctB; Lane 6, S. pombe marker Bio-Rad (DNA sizes: 3.5, 4.6 and 5.7 Mbp); Lane 7, MCH1 (fused chromosomes: 4.03 Mbp).

D. Chromosome structure of the dam4. The genes flanking the sites of fusion are indicated.

E. Chromosome structure of the dam1. The location of fusion occurred between dif1 and dif2, generating two new chimeric dif sites. The genes flanking the dif sites are indicated.

Inactivation of mutH restores the viability of V. choleraedam recA mutants

We tested the effect of MMR inactivation in V. cholerae dam recA mutants. In E. coli, inactivation of MMR restores the viability of dam recA mutants (Wang and Smith, 1986; Nowosielska and Marinus, 2005). Both E. coli and V. cholerae are members of the mono-phyletic clade of the γ-proteobacteria defined by the acquisition of the dam-seqA-mutH genes suggesting that analogous Dam-regulated cellular processes occur in the two species (Lobner-Olesen et al., 2005). V. cholerae possesses all three E. coli orthologues: mutL (VC0345), mutS (VC0535) and mutH (VC0668). We abolished MMR activity in V. cholerae by deleting mutH, since V. cholerae mutH gene is able to fully complement a mutH defect in E. coli (Friedhoff et al., 2002). We constructed a ΔdamΔmutHΔrecA-[pTs-PBADdam] strain, growing it to stationary phase in liquid culture under Dam-expressing conditions. Cells were then platted under Dam-depleting restrictive conditions and yielded viable colonies. The genotypes of 10 clones were checked by PCR, and confirmed all to be ΔdamΔmutHΔrecA. We conclude, therefore, that inactivation of mutH restores the viability of dam recA mutants. We further propose that MMR is also responsible for the formation of DNA DSBs in V. cholerae dam mutants, which renders HR essential for repair and survival.

RecA is involved in the survival process of V. cholerae dam mutants

We assessed the contribution of RecA in the process of survival in ΔdamΔmutHΔrecA strains. Upon Dam depletion in ΔdamΔmutHΔrecA-[pTs-PBADdam], mutants arose at a rate of 4 × 10−5 (Fig. 2A). The three-order decrease in the rate of survival compared with Δdam is not due to deletion of mutH since ΔdamΔmutH mutants arose at similar rate (2 × 10−2) as Δdam mutants (Fig. 2A). This indicates that MutH is not measurably involved, leaving a predominate role to RecA for the survival process in Δdam. We conclude that HR largely contributes to the survival process of V. cholerae dam mutants. However, in light of the remaining difference compared with WT, there remains at least one significant HR-independent pathway allowing ΔdamΔmutHΔrecA spontaneous mutants to survive.

In V. cholerae dam mutants, chr2 is replicated independently of ori2-dependent initiation

RctB is the initiator for chr2 replication (Duigou et al., 2006), and its binding at ori2 requires full methylation of iterons GATC sites by Dam (Demarre and Chattoraj, 2010). In the absence of Dam, it is likely that ori2 is non-functional, thereby rendering RctB dispensable for the viability of V. cholerae dam mutants. Accordingly, we successfully generated viable ΔrctB derivatives of dam4. Additionally, dam4ΔrctB showed the same fitness defect as dam4 (Fig. 3B) indicating that rctB is epistatic to dam in this mutant. Conversely, dam allelic complementation in dam4ΔrctB strongly restored fitness (Fig. 3B, dam4::CdamΔrctB). This result delineates a further role for Dam, exclusive of RctB replication initiation of chr2, in other cellular processes. Thus, dam mutants of V. cholerae are capable of survival owing to another cellular process that allows bypass of replication initiation of chr2.

The two chromosomes of V. cholerae dam mutants are fused

We have previously demonstrated that in a synthetic monochromosomal mutant of V. cholerae (MCH1), where chr2 is physically linked to chr1, Dam and RctB are dispensable because chr2 is able to piggyback chr1 replication initiation machinery (Val et al., 2012). We reasoned that chr1 and chr2 fusion could naturally occur by HR between homologous sequences present on the two chromosomes. We used PFGE to see if such a fusion occurred in V. cholerae dam mutants. PFGE analysis of dam4ΔmutH unexpectedly indicated that DNA degradation was not completely eliminated (data not shown). Even if only two DSBs occur in dam4ΔmutH, located at random loci, they will greatly hinder the visualization of the genomic organization. To prevent the formation of DSBs, we used the dam4::CdamΔrctB mutant complemented for Dam but deleted for rctB so that ori2 remains inactive and the hypothetical chromosome fusion remains stable. PFGE of dam4::CdamΔrctB revealed a single chromosome of approximately 4 Mbp, confirming fusion between chr1 and chr2 (Fig. 3C, lane 5).

Homologous recombination between two IS elements mediates the fusion of chr1 and chr2

We delved the nature of the mechanisms involved in the fusion between chr1 and chr2 by identifying the regions of genome rearrangement. We sequenced a 3 kb mate-paired library of dam4 strain using high-throughput sequencing. Mate-paired reads were aligned to the V. cholerae N16961 reference genome (Heidelberg et al., 2000), and structural rearrangements were recognized by identifying mate-pair sequences whose two extremities did not map to the same chromosome. These ‘broken mate-pairs’ allowed us to re-confirm the chromosome fusion and to pinpoint the regions of genome rearrangement. In the mutant dam4, we identified that HR occurred between two duplicated 1279 bp long ISVch4 elements sharing 99% identity, [VC1789-VC1790] on chr1 and [VCA0791-VCA0792] on chr2 (Fig. 3D). This result confirmed our previous data showing the contribution of HR in the survival of dam deletion.

Site-specific recombination promotes the fusion of chr1 and chr2

We sequenced a 3 kb mate-paired library of another spontaneous dam mutant of V. cholerae, dam1, which was isolated and characterized in parallel to dam4. Surprisingly in dam1, the mechanism for chr1 and chr2 recombination did not involve HR, but rather site-specific recombination. In this mutant, chromosome fusion occurred between dif1 and dif2 sites, located at the termini of replication of chr1 and chr2 respectively (Fig. 3E). Recombination between two identical dif sites is typically employed by the cell for chromosome dimer resolution (CDR). Chromosome dimers are formed when an uneven number of crossovers occurs between replicating sister chromatids (Lesterlin et al., 2004). Recombination between dif sites is catalysed by two site-specific tyrosine recombinases, XerC and XerD, via a transient Holliday Junction (Lesterlin et al., 2004). A dif site consists of a central region (CR) of 6 bp flanked by the binding sites for XerC and XerD (Lesterlin et al., 2004). In V. cholerae, the XerC binding sites and the CR of dif1 and dif2 are divergent, while their XerD binding sites are identical. Despite these differences, the same XerC/D pair acts at dif1 and dif2 (Val et al., 2008). Recombination of two dif1 sites resolves chr1 dimers and recombination of two dif2 sites resolves chr2 dimers (Val et al., 2008). Heterologous recombination between dif1 and dif2 was never observed. Indeed the incompatibility of their CR should greatly impair the chances for strand exchange between dif1 and dif2 (Grindley et al., 2006). We selected a mutant where recombination occurred between dif1 and dif2, proving for the first time that recombination is possible between these two sites.

xerC deletion is lethal in dam mutants

Inactivation of CDR by deleting xerC in V. cholerae results in 9% cell death per generation, which corresponds to the rate of chromosome dimers formed at each cell generation (Val et al., 2008). This defect is suppressed by the inactivation of RecA, consistent with Xer recombination functioning in the resolution of chromosome dimers formed by HR (Val et al., 2008). In E. coli, dam deletion confers a hyper-recombination phenotype presumably because of the formation of recombinogenic lesions (Marinus and Konrad, 1976; Glickman and Radman, 1980; Robbins-Manke et al., 2005). Higher levels of HR are expected to give rise to higher rates of chromosome dimer formation. If dimers are formed in dam strains above a critical threshold, CDR should become essential. We tested the effects of CDR inactivation in the dam1 mutant by constructing dam1ΔxerC-[pTs-PBADdam] strains. No plasmid segregants could be recovered by propagation of the cells under restrictive conditions in the absence of the Dam expressing plasmid (Fig. 2C, bottom). We conclude that CDR, generally, and xerC, specifically, is essential for the viability of V. cholerae dam mutants.

Inactivation of mutH restores the viability of V. choleraedam xerC mutants

We reasoned that MMR, responsible for the hyper-recombination phenotype of dam mutants (Nowosielska and Marinus, 2008), must be indirectly responsible for the lethal rate of chromosome dimer formation in V. cholerae dam1ΔxerC. To confirm our hypothesis, we tested the effect of MMR inactivation in xerC mutants of ΔdamΔmutH-[pTs-PBADdam] strains. Dam depletion in ΔdamΔmutHΔxerC-[pTs-PBADdam] yielded viable colonies whose genotypes were confirmed by PCR. The viability of ΔdamΔmutHΔxerC segregants demonstrates that inactivation of mutH restores the viability of V. cholerae dam xerC mutants. Our results suggest that the lethal rate of chromosome dimer formation is due to hyper-recombination in dam mutants. This is further evidence that in absence of Dam methylation the lack of DNA strand discrimination causes MMR to produce recombinogenic DNA lesions.

XerC is involved in the survival process of V. cholerae dam mutants

We assessed the contribution of CDR in the process of survival of the loss of dam. Upon Dam depletion, the rate of survival of ΔdamΔmutHΔxerC is of 4 × 10−6 compared with ΔdamΔmutH mutants that arose at a rate of 2 × 10−2, indicating an important role of XerC (Fig. 2A). This decrease in survival rate was one order of magnitude greater than ΔdamΔmutHΔrecA, which suggests that CDR contributes more than HR into the survival process of V. cholerae dam mutants.

Rates of RecA- and XerC-mediated recombination in chromosome fusion

We studied 12 additional V. cholerae dam mutants, isolated over independent experiments, to establish an accurate estimation of the contribution of HR and CDR in chromosome fusion. To distinguish HR-mediated chromosomal fusions, we used the Artemis Comparison Tool (Carver et al., 2005) to identify homologous sequences shared by chr1 and chr2. We defined homologous regions as having ≥ 95% nucleotide similarity over ≥ 500 bp. 13 pairs of repeated DNA sequences between chr1 and chr2 fit these parameters (Table 1). Twelve of these pairs correspond to duplicated IS elements (ISVch4, IS1004) and one, to a duplicated gene encoding a transketolase. We subsequently performed a series of three Southern blots using probes specific to ISVch4, IS1004 and transketolase sequences respectively (Table S1). Approximately 36% of Δdam mutants (dam2, dam4, dam10, dam11, dam12) exhibited recombination between ISVch4 elements (Fig. 4A, top). No Δdam mutants showed recombination between IS1004 elements or transketolase genes (data not shown).

Figure 4.

Relative contribution of each recombination pathway in chromosome fusion.

A. Top panel: Southern hybridization with ISVch4 specific probe. Digested genomic DNA of V. cholerae wild-type (WT) and of the two sequenced dam mutants (dam1 and dam4) were used as controls. Twelve additional dam mutants (#2, #3 and #5 to #14) were tested for recombination between ISVch4 repeats. Six copies of ISVch4 are in the genome showing six bands on the gel which in the wild-type are at 3771 bp (VC1789/VC1790), 3994 bp (VCA0371/VCA0372), 4593 bp (VCA0507/VCA0508), 5381 bp (VC0256/VC0257), 10 077 bp (VCA0791/VCA0792) and 15 768 bp (VC1477/VC1478). As expected, WT and dam1 showed no recombination between ISVch4 while dam4 showed recombination between (VC1789/VC1790) and (VCA0791/VCA0792) observable by the loss bands at 3771 bp and 10 077 bp and appearance of bands at 8468 bp and 5381 bp. The dam12 mutant showed the same pattern as dam4, indicating that recombination occurred between the same elements. In dam2 and dam13, recombination occurred between (VC1477/VC1478) and (VCA0371/VCA0372) observable by the loss of bands at 15 768 bp and 3994 bp and the emergence of bands at 15 685 bp and 4082 bp. In dam11, recombination occurred between (VC1789/VC1790) and (VCA0507/VCA0508) observable by the loss of bands at 3771 bp and 4593 bp and the appearance of bands at 2827 bp and 5536 bp. Bottom panel: PCR amplification with primer flanking dif fusion site. The presence of a PCR product signifies that chromosome fusion has occurred between dif1 and dif2 sites. Same results were observed using primers flanking the other related point of fusion (data not shown).

B. Native and chimeric dif sites link the two chromosomes. One strand of each dif sites are depicted. Diverging bases between dif1 and dif2 are shown in red. The binding region of XerC and XerD and the central region (CR) is indicated. The position of strand cleavage by XerC and XerD are indicated by a black triangle and white triangle respectively. The sequences of the chimeric dif sites of the dam1 and dam13 mutants are represented. Black and red asterisks indicate the dif site (chimeric or native) found in the mutants which chromosomes fused at dif, observed after gel purification and sequencing of the PCR products.

C. Chromosome fusion occurs preferentially between the termini of chr1 and chr2. Left panel: Theoretical sites of chromosome fusion between chr1 and chr2. The analysis was performed using the Artemis Comparison Tool (ACT). The two circular chromosomes of the genome of V. cholerae N16961 are represented linearly as horizontal bars. The origin is located at the extremities (black zone) and the termini is located in the middle (white zone) with two gradients from black to white indicating the two arms of the circular chromosome. Linking lines indicate potential points for fusion. Repeats capable of fusion by HR (> 95% sequence similarity over at least 500 bp) are connected by a black line (ISVch4), a blue line (IS1004) or a green line (transketolase). The red line connects dif1 and dif2. Right panel: Chromosome fusion sites experimentally observed. The width of a line is proportional to the number of recombination events observed at its location.

Table 1. List of the pairs of the repeated elements between chr1 and chr2Thumbnail image of

To recognize CDR-mediated chromosomal fusions, we PCR amplified the genomic DNA of all Δdam mutants using primers flanking dif1 and dif2 (Table S1). ∼64% of Δdam mutants (dam1, dam3, dam5, dam6, dam7, dam8, dam9, dam13, dam14) were shown to have recombined between dif1 and dif2 (Fig. 4A, bottom). All mutants that had recombined at dif, did so to the exclusion of ISVch4, and vice-versa. Sequencing of all PCR products further confirmed that chr1 was joined to chr2 by a dif site. We were surprised, however, to find two kinds of dif sites linking the two chromosomes. In one case, a chimeric dif2/1 site connected the two chromosomes (Fig. 4B, dam1, dam13) which is concordant with XerC cleavage and strand exchange occurring at the outer boundary of the CR and the XerC binding site (Val et al., 2008). In the second case, a native dif2 site connected chr1 with chr2 (Fig. 4B, dam3, dam5, dam6, dam7, dam8, dam9, dam14) concordant with XerD-cleavage and strand exchange occurring at the outer boundary of the CR and the XerD binding site (Val et al., 2008). Interestingly, we observed a weak PCR amplification in the WT strain suggesting that recombination of the two chromosomes between dif sites occurs sporadically within the population. No amplification was detected in the mutants recombined at ISVch4. This is logical as recombination at dif in these mutants would result in resolution of the chromosome fusion and the loss of the non-replicating chr2.

Preferential fusion location in the region of the terminus of replication of the two chromosomes

We compared the location of chr1 and chr2 sites that can theoretically recombine by HR and CDR to the experimentally observed sites (Fig. 4C). While putative recombination sites are widespread along the two chromosomes, it appears from all 14 tested mutants that fusion preferentially occurred in the terminus of replication of the two chromosomes (Fig. 4C). Moreover, 64% of the mutants were fused at dif, the replication terminus landmark (Hendrickson and Lawrence, 2007).


Fusion of its two chromosomes allows V. cholerae to survive the loss of Dam

In this study, we demonstrate that the main way to survive the loss of Dam methylation in V. cholerae is through fusion of chr1 and chr2, by any means. In V. cholerae dam mutants, the origin of replication of chr2 is no longer functional and the replication of chr2 is insured through the piggybacking of the replication machinery of chr1 through spontaneous fusion with this replicon. We presume that in V. cholerae dam mutants the fusion of chr1 and chr2 is irreversible as it contains only one functional origin and it is therefore the only genomic architecture under which the bacteria can replicate its genome. In Sinorhizobium meliloti, which carries one chromosome and two megaplasmids, fusion was observed between the three molecules (Guo et al., 2003). However, the co-integrates could not be isolated since the three fused molecules all retained a functional origin of replication rendering the co-integrates reversible and resulting in a mixed population of bacteria with one to three replicons (Guo et al., 2003). As the formation of fused chromosomes occurred naturally, it is expected that wild-type cultures of V. cholerae also potentially harbour the full variety of rearrangements that the genome can generate through the various mechanisms of recombination characterized in this study. Indeed, we were able to show by PCR analysis that recombination of the two chromosomes can occur between dif sites within a population of WT V. cholerae. This indicates that chromosome fusion naturally occurs even when no selective pressure is applied. However, we assume that such rearrangements cannot be isolated as they are quickly reversed or outcompeted under normal conditions.

Viable mutants to study the role of Dam in V. cholerae

DNA adenine methylation has many functions in bacteria which differ between various genera. In α-proteobacteria, CcrM-mediated DNA adenine methylation regulates the bacterial cell cycle (Collier, 2009; Gonzalez and Collier, 2013). In γ-proteobacteria, Dam methylation regulates chromosome replication, DNA mismatch repair and expression of specific genes (Wion and Casadesus, 2006). The isolation of viable dam mutants in V. cholerae provides us with a model to study Dam-regulated cellular processes. For example, the transcriptome analysis of dam mutants will allow us to study the impact of Dam methylation on global gene expression profiles in V. cholerae. Dam is also implicated as a virulence factor in bacterial pathogenesis (Low et al., 2001), and overproduction of Dam attenuates the virulence of V. cholerae (Julio et al., 2001). The knowledge gained from V. cholerae dam mutants could potentially lead to the design of novel live attenuated V. cholerae vaccines.

Dam methylation is involved in DNA mismatch repair in V. cholerae

The isolation of viable dam mutants allowed us to correlate the role of Dam to MMR in V. cholerae. We demonstrated that dam mutants are mutators and generate DNA DSBs, which correlates with a Dam methylation normally acting to direct MMR. We furthermore correlated the essential role of RecA to the activity of MMR in dam mutants as inactivation of mutH restores the viability of ΔdamΔrecA mutants. This is consistent with the requirement of RecA to repair DSBs generated by undirected MMR in absence of methylation marks (Nowosielska and Marinus, 2008). In E. coli, dam mutants have an hyper-recombination phenotype due to the presence of recombinogenic DNA lesions (Marinus and Konrad, 1976). Interestingly, we observed that ΔdamΔmutH mutants arose at similar rate as Δdam mutants suggesting that the presence of recombinogenic lesions does not impact the chances for the two chromosomes to recombine by HR.

Chromosome dimer resolution is essential in dam mutants

We hypothesized that recombinational repair of MMR-mediated DSBs in dam mutants should increase the rate of formation of chromosome dimers up to a critical threshold of 50% where CDR would become essential. Indeed, we demonstrated that xerC is essential in dam mutants and that inactivation of mutH restores the viability of ΔdamΔxerC mutants. This is novel evidence confirming that dam mutants have a hyper-recombination phenotype correlated to the MMR activity. This is also the first evidence of the essential role of CDR in bacterial dam mutants.

Homologous recombination mediated chromosome fusion

In V. cholerae, we identified two recombination pathways that can promote the fusion of its two chromosomes: HR and site-specific recombination. We showed that chromosome fusion can occur by HR between repeated ISVch4 elements. There are three ISVch4 elements on each chromosome yielding 9 different potential combinations for fusion of chr1 and chr2 (Table 1, Fig. 4C). Reciprocally, the chromosome resulting from the fusion between chr1 and chr2 contains many pairs of repeated ISVch4 that might be used for excision. If excision occurs at a site other than the site used for chromosome fusion, this will promote the transfer of DNA segments from one chromosome to the other. As this potential clearly exists, we might, consequently, observe a large heterogeneity within the V. cholerae population. As this is not the case (Feng et al., 2008; Hasan et al., 2010), there must be selective pressure applied to this particular genomic architecture which is presumably optimized for the growth of this bacterium.

Chromosome fusion at heterologous dif sites

We showed that chromosome fusion can occur by site-specific recombination between dif sites. During site-specific recombination, homology in the CR region is ordinarily required for re-annealing to the complementary partner strand, but is not an absolute prerequisite for strand exchange (Rajeev et al., 2009). Several examples show that Xer recombination is permissive and the order and prerequisite of strand exchanges depends on the sites involved in the reaction (Val et al., 2005; Das et al., 2010; 2013). In most Proteobacteria with multipartite genomes, the CR of dif sites from different chromosomes are divergent, as is the case for dif1 and dif2 of V. cholerae N16961 (Fig. 4B) (Val et al., 2008). This heterogeneity was theorized to prevent formation of aberrant products such as fused chromosomes. However, in this work we show that such a fusion involving site-specific recombination between dif1 and dif2 does occur naturally. We further remarked that such fusions result in two kinds of dif sites: chimeric dif sites containing a fragment of dif1 and a fragment of dif2 or native dif sites similar to those observed in the WT. Our data show that the chimeric dif2/1sites have the XerC binding site of dif2 and the CR of dif1 (Fig. 4B). This is consistent with a XerC-mediated cleavage and strand exchange occurring at the outer boundary of the CR and the XerC binding site (Val et al., 2008). A reaction of this type would produce an atypical Holliday Junction (aHJ) due to the heterology of the exchanged sequences. This aHJ is unlikely to be resolved by a second strand exchange mediated by XerD because of the lack of homology in the exchanged region. Instead the aHJ is likely to be resolved by replication similar to the resolution of integron cassette insertion (Loot et al., 2012). To explain the existence of native dif sites connecting chr1 with chr2, we hypothesize that strand exchange must be mediated by XerD with strand cleavage occurring at the boundary of the CR and the XerD binding site. Indeed, the XerD binding sites of dif1 and dif2 are identical and thus strand exchange would be undetectable (Val et al., 2008). However in this case, resolution of the aHJ cannot be mediated by XerC which would produce chimerical dif1/2 sites. The aHJ would at that point be resolved by replication (Loot et al., 2012). Alternatively, we suppose that HR can occur in the vicinity of dif since chr1 and chr2 share a stretch of 131 bp perfect homology outward from the XerD binding sites of dif1 and dif2 (Fig. S1). This recombination pathway would explain the observation of a native dif site linking chr1 to chr2, as well. Finally, an alternative pathway can combine HR and CDR; If RecA mediates strand exchange between the two homologous sequences near dif1 and dif2, a Holliday junction (HJ) will be formed. HJs are normally recognized by RuvAB and resolved by RuvC but an alternative helicase such as RecG could act on the HJ and catalyse branch migration (Michel et al., 2007). If branch migration reaches dif, resolution of the HJ could be catalysed either by XerD, leading to the formation of native dif sites or by XerC leading to the formation of chimeric dif sites.

Preferential fusion of the termini of V. cholerae two chromosomes

From 14 independent dam mutants, we observed that chromosome fusion occurred exclusively between the termini of the two chromosomes. Several biological reasons could explain this observation. Gene acquisition often occurs in the chromosome termini which are among the most divergent segments in closely related bacteria (Suyama and Bork, 2001). Along these same lines, fusion in the termini could be less deleterious for growth than in other regions in V. cholerae. However, the synthetic fusion of the terminus of chr1 with the origin of chr2 had no strong impact on the fitness of the bacteria (Val et al., 2012). Secondly, chr1 and chr2 termini colocalize in the middle of the cell (Srivastava et al., 2006). Hence, preferential chromosome fusion of the termini of chr1 and chr2 could reflect their spatial colocalization. Finally, it is known that RecA-dependent HR events occur at a much higher frequency in the terminus region of the E. coli chromosome (Louarn et al., 1991; 1994). This could also be the case in V. cholerae and would increase the number of recombination events between the termini of chr1 and chr2. Termini fusion of the two chromosomes can also inform us on their termination processes since it is unlikely that a replication fork trap system supports replication termination of chr2 otherwise chr2 could not be replicated when fused to chr1 (Duggin et al., 2008).

In summary, we have studied the Dam system in V. cholerae, at once placing it in context with previous studies in E. coli and shedding light on the dynamics of the genome at work for surviving dam deletion in this important pathogen. It is expected that this work will serve as the basis for continued studies such as a deeper examination of regulation in Δdam mutants or the further quantification of recombination between various regions of the two chromosomes.

Experimental procedures

Bacterial strains and growth conditions

Bacterial strains and plasmids used in this study are listed in Table S2. Cells were grown at 37°C in Luria broth. Antibiotics were used at the following concentrations: ampicillin, 75 μg ml−1; chloramphenicol 5 μg ml−1; kanamycin 25 μg ml−1; rifampicin 1 μg ml−1; spectinomycin 100 μg ml−1; zeocin 25 μg ml−1. Diaminopimelic acid was used at 0.3 mM; arabinose (0.2%), sucrose (15%) and glucose (1%).

General cloning procedures

DNA cassettes to insert into the genome of V. cholerae were cloned into either pDS132 or pSW7848, both R6K γ-ori-based suicide vectors (Philippe et al., 2004; Val et al., 2012). To provide homology for integration, two regions spanning the point of insertion of > 500 bp were PCR amplified from N16961 genomic DNA and cloned up and downstream to the DNA cassettes. For cloning, Π3813 was used as a plasmid host (Le Roux et al., 2007). For the conjugal transfer of plasmids to V. cholerae strains, E. coli β3914 was used as the donor (Le Roux et al., 2007). Selection of the plasmid-borne drug marker resulted in integration of the entire plasmid in the chromosome by a single crossover. Elimination of the plasmid backbone resulting from a second recombination step was selected for by arabinose induction of the ccdB toxin gene when using pSW7848 or on sucrose-containing medium when using pDS132.

Dam depletion

V. cholerae strains deleted for dam in the presence of pTs-PBADdam (pGD93) were then depleted for Dam as previously described (Demarre and Chattoraj, 2010). The rate of viable WTΔdam mutants in an overnight culture of Δdam[pTs-PBADdam] grown in permissive conditions was assessed by platting serial dilutions under permissive and restrictive conditions.

Determination of the methylation state of V. cholerae genomic DNA

Genomic DNA from V. cholerae was digested with three restriction endonucleases cleaving at GATC sites differentially depending on their methylation state (Palmer and Marinus, 1994). Four micrograms of genomic DNA was digested for 1 h with 2 FDU (FastDigest unit, Thermo Scientific) of Sau3AI, DpnI and MboI. Sau3AI cleaves GATC sites regardless of their methylation state, DpnI only cleaves methylated GATC sites and MboI only cleaves unmethylated GATC sites. Digested DNA was separated by agarose gel electrophoresis and stained with Ethidium Bromide.

Determination of point mutation rates

Point mutation rates were measured using a rifampicin resistance assay. Cells were grown until stationary phase in rich LB medium and platted on medium with or without rifampicin (1 μg ml−1). Mutation rates are the ratios between the numbers of rifampicin resistant (RifR) colonies by the total number of colonies.

Pulsed field gel electrophoresis

The preparation of genomic DNA embedded in agarose gels and the protocol for PFGE was performed as previously described (Iida et al., 1997). DNA DSBs of circular chromosomes can be detected by PFGE since only linear molecules can enter the gel (Michel et al., 1997). Random DSBs give rise to a wide range of DNA fragment sizes, resulting in a smeared profile.


Chromosomal DNA was prepared from 50 ml cultures of V. cholerae dam1 and dam4 harvested after overnight growth at 37°C as described (Wilson, 2001). Whole genome shotgun sequencing was performed on an Illumina HiSeq2000. 3 kb mate-pair libraries were prepared and 2 × 100 nt reads generated. A minimum of 160-fold coverage per genome was obtained and reads were mapped on the reference genome Vibrio cholerae O1 biovar El Tor str. N16961. The Burrows-Wheeler Aligner (bwa 0.6.1, http://bio-bwa.sourceforge.net/) was used for the comparison and the alignment BAM files were parsed with the Picard command-line tools (http://picard.sourceforge.net/command-line-overview.shtml).

Southern blot

Genomic DNA of V. cholerae was extracted from 1 ml of stationary-phase bacterial cultures using the GeneJET Genomic DNA purification Kit (Thermo scientific). A total of 20 μg of genomic DNA was digested with NcoI, PstI, SacI. Samples were resolved on 0.5% agarose gel in 0.5× TAE buffer. Gel was dried for 1 h at 60°C. DNA was denatured by incubating the dried gel with a denaturation solution (NaOH 0.5 M, NaCl 150 mM) for 20 min at room temperature. The gel was equilibrated in neutralization solution (Tris-HCl 0.5 M pH = 8, NaCl 150 mM). The neutralized gel was directly pre-hybridized for 30 min at 50°C with 10 ml of RapidHyb Buffer (GE). PCR products of ISVch4, IS1004 and transketolase genes were amplified using primers listed in Table S1. DNA Probes were generated from the PCR products that were radioactively labelled with [α-32P]dCTP using the Prime-a-Gene labelling kit (Promega) and purified using a G-25 column (GE). Hybridization with each probe was performed overnight at 50°C. Gels were then washed in 2× SSPE, 0.1% SDS at 45°C for 15 min, in 2× SSPE, 0.1% SDS at 50°C for 15 min, in 0.1× SSPE, 0.1% SDS for 30 min at 55°C and a final wash was performed at 60°C for 1 h. Radioactive bands were detected with a phosphor screen.


We thank S. Mangenot and B. Quinquis for their contribution to sequencing and D. K. Chattoraj and G. Demarre for kind gifts of plasmids. We thank the anonymous reviewers for insightful comments. This work was supported by a grant from the French National Research Agency (ANR-10-BLAN-131301). Research was funded by the Institut National de la Santé et de la Recherche Médicale (INSERM), the Institut Pasteur, the Centre National de la Recherche Scientifique (CNRS). A.S.B. is funded by EMBO (ALTF-1473–2010) and Marie Curie Actions (BMC FP7-PEOPLE-2011-IIF).

Conflict of interest

The authors declare that they have no conflict of interest