The Two Cis-Acting Sites, parS1 and oriC1, Contribute to the Longitudinal Organisation of Vibrio cholerae Chromosome I

The segregation of bacterial chromosomes follows a precise choreography of spatial organisation. It is initiated by the bipolar migration of the sister copies of the replication origin (ori). Most bacterial chromosomes contain a partition system (Par) with parS sites in close proximity to ori that contribute to the active mobilisation of the ori region towards the old pole. This is thought to result in a longitudinal chromosomal arrangement within the cell. In this study, we followed the duplication frequency and the cellular position of 19 Vibrio cholerae genome loci as a function of cell length. The genome of V. cholerae is divided between two chromosomes, chromosome I and II, which both contain a Par system. The ori region of chromosome I (oriI) is tethered to the old pole, whereas the ori region of chromosome II is found at midcell. Nevertheless, we found that both chromosomes adopted a longitudinal organisation. Chromosome I extended over the entire cell while chromosome II extended over the younger cell half. We further demonstrate that displacing parS sites away from the oriI region rotates the bulk of chromosome I. The only exception was the region where replication terminates, which still localised to the septum. However, the longitudinal arrangement of chromosome I persisted in Par mutants and, as was reported earlier, the ori region still localised towards the old pole. Finally, we show that the Par-independent longitudinal organisation and oriI polarity were perturbed by the introduction of a second origin. Taken together, these results suggest that the Par system is the major contributor to the longitudinal organisation of chromosome I but that the replication program also influences the arrangement of bacterial chromosomes.


Introduction
Bacterial chromosome replication is initiated from a unique origin (oriC) and progresses bidirectionally. The replication of circular chromosomes terminates opposite the oriC in a region termed the terminus (Ter). Within the Ter region is a site-specific recombination site, termed dif, dedicated to the resolution of chromosome dimers [1]. These factors define two replication arms, Left and Right, mirrored by the ,oriC-dif. axis. Detailed investigations of the choreography of chromosomal movements during the cell cycle of several monochromosomic bacteria suggest that segregation is concurrent with replication and starts with the precise positioning of newly replicated sister copies of the oriC region into opposite cell halves [2,3,4,5]. Segregation of other sister chromosomal loci to their positions in daughter cells follows shortly after their replication with sister copies of Ter being segregated last [2,3,4,5]. Less is known about the choreography of chromosome segregation in bacteria with multipartite genomes. However, the analysis of a single locus in the oriC region and a single locus in the putative Ter region of the two Vibrio cholerae chromosomes suggests a model of replication and segregation that is consistent with monochromosomic bacteria [6,7,8]. Taken together, these observations suggest that the active positioning of the oriC region sets the pace for chromosome segregation, raising questions regarding the underlying mechanism.
Similar to most other bacteria, a specific partition system is encoded on each of the Vibrio cholerae chromosomes [9]. Bacterial chromosome partition machineries are related to the Type I partitioning systems of plasmids. They consist of two genes, parA, which codes for an ATPase, and parB, which codes for a sequencespecific DNA binding protein that is able to spread around its binding site, parS [10]. Several parS sites are usually found proximal to and in some cases encompassing the oriC region of bacterial chromosomes [9]. The role of Par systems in DNA segregation is well established for low-copy number plasmids [11,12,13]. However, their role in bacterial chromosome segregation remains controversial, notably because the disruption of Par systems in different bacterial species produces different phenotypes. The Par systems of Caulobacter crescentus, Myxococcus xanthus and V. cholerae chromosome II are essential for chromosome segregation [4,14,15]. The impairment of the Pseudomonas aeruginosa Par system caused the formation of ,20% anucleate cells [16].
However, the disruption of the Par machinery affects few cells (,2%) in Bacillus subtilis [17] and yields no segregation defect for V. cholerae chromosome I [18]. Moreover, several bacterial species, notably Escherichia coli and Streptococcus pneumonia, lack a functional Par system. Partition systems have also been implicated in other cellular processes including replication initiation [19,20], cell cycle coordination [4,14] and chromosome compaction [21,22,23]. Finally, ParB-binding to oriC-proximal parS sites recruits SMC proteins to the origin region in B. subtilis and S. pneumoniae, contributing to chromosome segregation even in the absence of ParA [21,22,23]. Nevertheless, Par systems are directly involved in the polar positioning and the active bipolar migration of the oriC region of the C. crescentus chromosome and V. cholerae chromosome I [8,24,25,26,27]. The polar anchoring mechanisms have been described for these two systems [27,28].
The characterization of chromosomal organisation in different bacterial species suggests a common mechanism of longitudinal organisation in which the oriC region is positioned towards the old pole, Ter is positioned towards the new pole, and the two chromosome replichores extend over the long axis of the cell [3,4,29,30]. The polar localisation of the oriC regions of V. cholerae chromosome I and of the multiple chromosomes of S. meliloti and A. tumefaciens suggests a similar longitudinal organisation [31,32]. In contrast, the E. coli chromosome, which seems devoid of a Par system, adopts a transversal organisation where the oriC region is positioned at midcell and the left and right chromosomal arms extended toward the opposite cell halves [33,34]. Taken together, these observations suggest that it is the active positioning of the oriC region towards the old cell pole by Par systems that is responsible for the longitudinal chromosomal organisation observed in most bacterial species.
However, two characteristics of V. cholerae make it an ideal bacterial model to test this hypothesis. Firstly, the Par system of V. cholerae chromosome II positions the origin region (oriC2) at midcell rather than at the cell pole, which would be expected to drive a transversal arrangement if the hypothesis was correct. Secondly, disruption of the Par system of V. cholerae chromosome I does not affect any step of the cell cycle, allowing for the direct investigation of its contribution to chromosomal organisation. In this current study, the analysis of the intracellular location of 12 chromosome I loci and 7 chromosome II loci in exponentially growing cells indicated that both V. cholerae chromosomes adopted a longitudinal organisation. Chromosome I extends from the old pole to the new pole and chromosome II extends from midcell to the new pole, i.e. in the younger cell half. By displacing parS1 sites away from oriC the Par system was shown to contribute to the longitudinal organisation of chromosome I. However, the parS1deleted chromosome I remained longitudinal, with the oriC locus remaining located close to the old pole. The insertion of an ectopic origin of replication was sufficient to disrupt the longitudinal organisation of parS1 deleted chromosome I. Interestingly, the ectopic origin of replication was often positioned closer to the old pole than the original oriCI. Taken together, these observations suggest that the replication program contributes to the longitudinal organisation of V. cholerae chromosome I.

Fluorescent labelling systems and analysis of chromosome choreographies
Chromosome choreography involves successive steps of chromosome organisation as a function of cell cycle progression. To avoid any complications linked to multiple concurrent rounds of replication, cells were cultivated at 37uC in slow growing conditions (M9 Fructose supplemented with thiamine) characterised by a 55 min generation time divided into three successive periods: a period of 11 min before replication initiation (B period), a 32 min-long replication period (C period) and a 12 min period after replication and before division (D period) [35]. Snapshot images of cells in steady state exponential growth were acquired. By correlating the length of cells with their progression through the cell cycle we studied chromosomal organisation as a function of cell elongation. Cells were classified according to their length in 0.1 mm intervals. A sliding window of 0.3 mm was used to determine the median position and frequency of duplication of any specific chromosomal locus. In our growth conditions the majority of cells had a length between 1.9 mm and 4.3 mm. We will refer to the cells of the first point plotted, corresponding to the 2 mm interval, as newborn cells and the cells of the last interval plotted, corresponding to the 4.2 mm interval, as dividing cells.
The spatial organisation of chromosome I and of chromosome II was deduced from the positioning of 12 and 7 loci respectively. They were broadly distributed over the entire genome, particularly the regions of special interest, such as the origin of replication and the chromosomal dimer resolution site of each of the two chromosomes (ori I , ori II , ter I and ter II ). Five loci were tagged on the right (R1 I to R5 I ) and left (L1 I to L5 I ) replichores of chromosome I. On chromosome II, three loci were tagged on the right replichore (R1 II to R3 II ) and two on the left (L1 II and L2 II ). The loci were visualised in pairwise combinations using two compatible fluorescent labelling systems, a lacO array was inserted at one of the loci and a parS pMT1 site at the other locus. LacI-mCherry and yGFP-D30ParB pMT1 [34] protein fusions were produced from an operon in place of the V. cholerae lacZ gene. In some cases, a third locus was tagged with a tetO array and visualized by the production of a TetR-Cerulean fusion from a plasmid. We observed the same pattern of localisation of the L3 I or terI loci during the cell cycle whether they were tagged with a lacO or tetO array or with a parS pMT1 site, suggesting that these three systems did not affect the positioning of the chromosomal loci under our experimental conditions.

Author summary
Proper chromosome organisation within the cell is crucial for cellular proliferation. However, the mechanisms driving bacterial chromosome segregation are still strongly debated, partly due to their redundancy. Two patterns of chromosomal organisation can be distinguished in bacteria: a transversal chromosomal arrangement, such as in E. coli, where the origin of replication (ori) is positioned at midcell and flanked by the two halves of the chromosome (replichores), and a longitudinal arrangement, such as in C. crescentus, where ori is recruited to the pole and the replichores extend side by side along the long axis of the cell. Here, we present the first detailed characterization of the arrangement of the genetic material in a multipartite genome bacterium. To this end, we visualised the position of 19 loci scattered along the two V. cholerae chromosomes. We demonstrate that the two chromosomes, which both harbour a Par system, are longitudinally organised. However, the smaller one only extended over the younger cell half. In addition, we found that disruption of the Par system of chromosome I released its origin from the pole but preserved its longitudinal arrangement. Finally, we show that the addition of an ectopic ori perturbed this arrangement, suggesting that the replication program contributes to chromosomal organisation.
Dual labelling has three advantages. Firstly, the precise orientation of the long axis of the cell from the new pole to the old pole could be determined. The new pole is defined as the pole resulting from the previous cell division event. Assuming no gross rearrangements of chromosomal DNA after cell division, a locus closer to the septum than to the poles in dividing cells was defined as being closer to the new pole than to the old pole in newborn cells. Time-lapse microscopy observations validated this method ( Figure S1). Secondly, dual labelling allowed for the collection of data in two different strains even if they presented slightly different cell size distributions ( Figure S2). The cell size distribution of the two strains was re-aligned using the frequency of duplication of a common locus as a reference. In the rest of the manuscript we will refer primarily to compiled figures. However, each of our conclusions could be drawn based on individual strain data. The data of each individual strain are presented in Figures S3 to S39. Thirdly, dual labelling allowed for the comparison of the timing of duplication of the two tagged loci of each strain as a function of cell elongation and the direct measure of the distance separating these loci as a function of cell elongation. These comparisons could be made independently of any orientation procedure or cell size re-alignment.

Sequential order of duplication of chromosomal DNA in V. cholerae
The tagged loci were always observed as a single focus in newborn cells ( Figure 1B and 2B). This is consistent with our growth conditions in which newborn cells contain a unique nonreplicating copy of each of the two chromosomes. In dividing cells, two separate foci were observed for most of the analysed loci ( Figures 1B and 2B). The only exceptions were L5 I , R5 I and ter I , whose foci were duplicated in about 50%, 40% and 10% of the dividing cells, respectively ( Figure 1B). This is consistent with a previous report where sister copies of a locus situated at 40 kbp from dif1 remained colocalised until the very end of septation [6].
The order of duplication of sister copies of each locus as a function of cell elongation followed the genetic map from ori to ter along the two arms of chromosome I and of chromosome II ( Figures 1B and 2B). In addition, with the exception of the dif1 proximal loci, the rate of focus duplication of any given locus was similar. The proportion of cells with two foci increased abruptly from 10% to above 80% within #0.6 mm of cell elongation. However, the difference in the timing of duplication of two consecutive loci was not strictly proportional to the genetic distance that separated them. For instance, sister L1 I separation immediately followed sister ori I separation and occurred much earlier than sister L2 I separation, despite L1 I being closer to L2 I than to ori I . This observation is consistent with a previous finding in E. coli suggesting that the time sister loci remain together after replication is variable [36]. In addition, it was sometimes not possible with the resolution of our experiments to differentiate the timing of duplication of loci that were too close on the genetic map, such as ori II , R1 II and R2 II or R3 I and R4 I .
Finally, using a strain with tagged ori II and L3 I , it was observed that the separation of sister L3 I occurred at smaller cell length than for sister ori II ( Figure S9, ADV26). These results suggest that the separation of sister copies of the loci most proximal to the origin of replication on chromosome II occurred later than sister ori I separation ( Figures 1B and 2B). This observation is in agreement with the delayed firing of the origin of replication of chromosome II compared to chromosome I replication [35].

Longitudinal organisation of Chromosome I and II
For most loci and cell length categories, we observed cells with either two separated foci or a single focus. The proportion between these two types of cells varied as a function of cell elongation ( Figures 1B and 2B). To simplify the representation of the data, we plotted the position of the foci that correspond to the dominant cell type in each cell length interval ( Figures 1C and 2C). In addition, we plotted the median positions of the observed foci (filled circles), along with the 25 th -75 th percentiles (error bars).
In newborn cells, the relative cell position of ori I and ter I was approximately 0.8 and 0.1 respectively, reflecting that ori I is positioned near the old pole and ter I close to the new pole. In agreement with previous reports the position of ori I in the overall cell population, i.e. irrespective of cell length, was closer to 0.9 [8,20]. The distance between the position of ori I and the position of any locus on chromosome I correlated with the genetic distance, suggesting a longitudinal arrangement of the replichores within the cell ( Figure 1C). To confirm this longitudinal arrangement we directly calculated the distance between L3 I , which is located approximately in the middle of the left replichore, to additional chromosome I loci in cells that displayed a single focus for each of these two loci ( Figure 1D). Consistent with a longitudinal arrangement L3 I was observed as being closer to the right replichore loci than to ori I and ter I ( Figure 1D). Furthermore, we directly calculated the distance between separated sister loci in individual cells ( Figure 1E). The distance between sister copies of each locus linearly decreased from ori I to ter I , further confirming the longitudinal organisation of chromosome I ( Figure 1E).
Two modes of segregation could be distinguished among chromosome I loci. The first mode of segregation was observed for loci located from ori I to L4 I or R4 I . In this mode, loci remained close to the position they occupied at cell birth until their duplication. As duplication occurred the greater the genetic distance between a locus and ori I , the longer it remained static at its home position. After duplication sister foci segregated to their new home positions in the next cell length interval, suggesting that segregation might be a transient event within the cell cycle. The distances travelled by two sister foci were not identical. The most unbalanced situation was observed for ori I where one copy remained nearly immobile whilst the second copy crossed the whole cell length. In contrast, the two L3 I sister loci exhibited a symmetrical separation on either side of the midcell/future new pole. The second mode of segregation was characterized by the mobilization of the loci towards midcell before duplication. This mode of segregation applied to loci located in the terminus region of chromosome I, i.e. R5 I , L5 I and ter I . These loci migrated towards midcell (within 4 intervals) earlier than the duplication of loci located several hundreds of kb upstream ( Figure 1B).
The ori II of chromosome II positioned near midcell (with a relative position of 0.55) and dif2 near to the new pole (with a relative position of 0.28) in newborn cells. This is consistent with previous reports [6,8]. The positioning and mobilisation of the ori II region depends on a partition system like ori I [15]. The positions of R1 II , R2 II , R3 II and L1 II , L2 II were intermediate between those of ori II and ter II , suggesting that chromosome II occupied only the younger half of the cell. Despite the short distance separating ori II proximal loci ( Figure 2B), the sequential ori II to ter II positioning of the seven chromosome II loci was observed, suggesting a longitudinal organisation ( Figure 2C). Direct measurement of the distance between ter II and R1 I , R2I, R3 I and L2 II ( Figure 2D) and of the distance between sister copies of all of the chromosomal loci ( Figure 2E) confirmed the longitudinal organisation of chromosome II. Furthermore, the duplication of most loci occurred close to midcell ( Figure 2C). This suggests that un-replicated chromosome II loci are pulled towards midcell by replication, which in turn pushes sister copies of replicated loci away from midcell and each other. Impact of parS1 displacement on chromosome I organisation We speculated that the displacement of parS1 away from oriCI would modify the longitudinal arrangement of chromosome I. To test this hypothesis, tandem parS1 sites were introduced at 300 kb, 490 kb and 650 kb on the left arm of chromosome I in cells devoid of their three natural parS1 sites. Displacing the location of the Parmediated anchoring zone on chromosome I did not affect the fitness or the morphology of the cells. The functionality of the displaced parS1 sites was not affected as judged by the visualisation of at least one polar focus of Ypet-ParB1 in the majority of different cell populations ( Figure S40).
To assess the effect of parS1 displacements on the global organization of chromosome I, the cellular positions of ori I , R2 I , L3 I and ter I were visualized. Displacing parS1 sites on chromosome I had a dramatic effect on the positioning and segregation of ori I , L3 I and R2 I ( Figure 3B). In contrast, the positioning and segregation of ter I was not significantly affected ( Figure 3B). However, the displacement of parS1 to the L3 I location affected the polar positioning of ter I prior to its final recruitment to midcell.
The displacement of parS1 sites to a location equidistant to ori I and L3 I , caused the single ori I focus of small cells to shift towards midcell and the ori I sisters to be positioned closer to the quarter positions in long cells ( Figure 3B, parS1@300kbp). This was confirmed by a decrease in the distance between ori I and ter I before ori I duplication ( Figure 3D) and a decrease in the distance between ori I sisters after their separation ( Figure 3C and 3E). The single L3 I focus of small cells was shifted towards the old pole and L3 I sisters were positioned further away from quarter positions in long cells ( Figure 3B, parS1@300kbp). This was confirmed by the direct measurement of the distance between L3 I sisters after their separation ( Figure 3C). In addition, R2 I positioning was shifted toward the new pole, as observed by a reduced distance between sister R2 I loci ( Figure 3C). As a consequence, L3 I and R2 I no longer co-localised ( Figure 3B).
When the parS1 sites were displaced to the L3 I locus, the home position of ori I and L2 I were switched. The ori I locus, which was now situated 700 kb away from the parS1 sites, adopted the positioning of L3 I locus in the wild-type context ( Figure 3B, parS1@650kbp). This was confirmed by a decrease in the distance between ori I and ter I before ori I duplication ( Figure 3D) and a decrease in the distance between ori I sisters after their separation ( Figures 3C and 3E). Reciprocally, L2 I , which was now located 170 kb from the parS1 sites, exhibited a positioning similar to ori I in the wild-type cell ( Figure 3B, parS1@650kbp). The distance between sister L2 I also became larger than the ori I sister distance For each locus, a minimum number of 800 cells were analysed. The first cell size interval where $50% of cells contained duplicated L3 I was used to realigned ADV26 to the ADV24 reference strain. CP708, ADV131, ADV30 and ADV131 cell size distributions were aligned with the GDV552 cell sizes based on the cell size interval where $25% of cells contained two ter II loci. GDV552 was aligned to ADV42 using the cell size interval where ter I is recruited to midcell. ADV123 was aligned with ADV24 using the first cell size interval where $50% of cells contained two ori I . (C) Reconstitution of the segregation choreographies of the 7 chromosome II loci, as in Figure 1C. (D) Relative distance between any of the chromosome II loci to the ter II locus, measured in the cells containing only one focus of each locus. (E) Relative distance between any of the chromosome sister loci, as in Figure 1E. doi:10.1371/journal.pgen.1004448.g002 ( Figure 3C). Finally, the shift of R2 I positioning toward the new pole was exacerbated ( Figure 3B and 3C).
To further demonstrate the extent of the chromosomal reorganisations, we took advantage of our double labelling systems to directly monitor the respective positions of ori I and L3 I within long cells once they were duplicated and had reached their new home position. We determined the proportion of cells in which ori I was more polar than L3 I when the parS1 sites were at their normal position (parS1 65 kb , wild-type) or had been displaced by 300 kb (parS1 300 kb ) or 490 kb (parS1 490 kb ). This method allowed for chromosome rearrangements to be monitored at the single cell level. ori I was more polar than L3 I in almost 100% of  Figure 1C. 2 mm of cell elongation were shown centred on the cell interval where at least 50% of the cells had duplicated ori I . For the WT choreography, ADV114 and ADV78 cell size distributions were aligned with the ADV24 cell sizes using the first cell size interval where 50% or more of cells contained duplicated ori I . For the parS1 300 kb choreography, CP604 and CP582 cell size distributions were aligned with the CP591 cell sizes using the first cell size interval where $50% of cells contained 2 foci of ori I and L3 I , respectively. For the parS1 650 kb choreography, the cell sizes correspond to strain CP605 (ori I and ter I ). CP583 and CP586 cell size distributions could not be realigned. The corresponding data was plotted with dashed lines (L2 I and R2I). (C) Relative distances between any of the sister loci, as in Figure 1E. (D) Relative distance between any loci to ter I locus, as in Figure 1D. (E) Relative distance between ori I sister loci, in with displaced parS1 sites, as in Figure 1E. doi:10.1371/journal.pgen.1004448.g003 the cells in the wild type context ( Figure 4A, parS1 65 kb ). The proportion of cells in which ori I was more polar than L3 I decreased to 60% when parS1 sites were equidistant between ori I and L3 I ( Figure 4A, parS1 300 kb ) and down to 30% when parS1 sites were closer to L3 I than to ori I ( Figure 4A, parS1 490 kb ). The median distances between separated ori I sisters were very similar when parS1 sites had been displaced to 300 kb or 490 kb ( Figure 3E).
In conclusion, the displacement of parS1 sites along the left arm of chromosome I led to the global rearrangement of chromosome I within the cell. These results suggest that the Par system of chromosome I not only mobilises and anchors the parS1 sites at the poles but also directly contributes to the arrangement of the entire DNA molecule. Only the positioning of the terminus region escaped the influence of the Par system.

Impact of parS1 deletion on chromosome I organisation
Previous studies indicated that deletion of parA1 resulted in the release of the parS1 sites from the old pole with the relative position of ParB1 changing from 0.023 in a wild-type context to 0.2 [6,8]. Despite this dramatic change, no effects on cell fitness or cell cycle parameters were reported, suggesting that chromosome I segregation was not affected by the disruption of its Par system. However, we speculated that the organisation and segregation choreography of chromosome I would be globally modified when Par-mediated mobilization and anchoring of ori I were lost. To this end, we followed the localisation of ori I , R2 I , L3 I and ter I loci in DparS1 cells. The longitudinal arrangement of chromosome I was maintained during the whole cell cycle. ori I remained the closest loci to the old pole with a relative position of 0.7 in small cells and a relative ori Iter I distance of 0.68 (Figures 5B and 5C). R2 I and L3 I loci remained co-localised during the whole cell cycle and were positioned around midcell until their duplication ( Figure 5B). After duplication, ori I sisters remained more polar than L3 I and R2 I sisters, which still co-localised ( Figure 5B). However, the relative distance between separated ori I sisters was smaller than in wild-type cells ( Figure 5D). In addition, the increase in the 25 th -75 th percentiles (error bars, Figure 5B) and the increase in the variability of the ori I -ter I distance ( Figure 5C) suggest that the home positions of ori I , L3 I and R2 I before and after duplication were less stringently controlled. Time-lapse experiments confirmed the global behaviour of these three loci (data not shown). In contrast, the choreography and late duplication of ter I was not affected by the absence of the parS1 sites. The positioning of ori II , R2 II and ter II loci suggest that the chromosome II behaviour was not dramatically modified in DparS1 cells ( Figure S41). In conclusion, the impairment of the Par system of chromosome I slightly disorganized the positioning and the segregation of the bulk of chromosome I. However, chromosome I remained longitudinally organised and ori I remained more polar than the rest of chromosome I, suggesting the existence of additional chromosome I organizing and positioning mechanisms.
Impact of an extra oriC1 at 651 kb on chromosome I organisation We hypothesised that the replication program might be, at least in part, responsible for the maintenance of the longitudinal arrangement of chromosome I in the absence of parS1 sites. To test this hypothesis, we introduced an extra origin of replication 651 kb from its normal position on the left replichore of chromosome I in cells lacking parS1 sites. The fitness of cells harbouring the two origins was not affected. Replication profiling demonstrated that the ectopic origin was as efficient as the wild-type origin and that the two origins were used at each cell cycle ( Figure S42). Correspondingly, the separation of ori I and L3 I sisters became synchronous ( Figure S42). The perturbation of the replication program further decreased the polarity of the ori I region in small cells ( Figure 5B), which decreased the distance between single ori I and ter I foci ( Figure 5C). The effect was more evident in long cells ( Figure 5B) where the relative distance between separated ori I sisters decreased to half the distance of wild-type cells with a high level of variability ( Figure 5D). The segregation of ori I sisters also became asymmetric with one of the two copies remaining relatively polar while the other adopted a more central position ( Figure 5D). Finally, the L3 I locus became slightly more polar and no longer co-localised with R2 I (Figure 5B). In contrast, earlier replication of the ter I locus Figure 4. Increase of L3 I polarity over ori I upon parS1 or oriC1 proximity to L3 I locus. For each pole of dividing cells, the most polar locus between ori I and L3 I was determined (A) in WT (ADV24), parS1 300 kb (CP591) and parS1 490 kb (CP634) and (B) in DparS1 (CP568), DparS1 + oriC1 651 kb (CP626), parS1 65 kb + oriC1 651 kb (CP633). This corresponds to approximately 100 cells. The red part of the stacked histogram represents the proportion of case with ori I more polar than L3 I , whereas the green part represents the opposite. doi:10.1371/journal.pgen.1004448.g004 ( Figure S42) did not modify its home position ( Figure 5C). The position of the R2 I locus was not significantly affected by the ectopic origin ( Figure 5C). Then, we decided to directly monitor in each cells which of the two competing locus, ori I and L3 I , was closer to the old pole. If the polar location of a given locus was only dictated by its proximity to the initiation of replication the proportion of cells in which L3 I was closer to the old pole than ori I would reach 50%. Correspondingly, the polarity of ori I compared to L3 I decreased from 80% in the parental cells (DparS1) to 55% in the cells harbouring an ectopic origin on the Left arm at 651 kb from ori I ( Figure 4B). The polar location of ori I compared to R2 I was unchanged by the addition of the ectopic origin, suggesting that the phenotype was not linked to a global disorganisation of the DNA within the cell ( Figure 4B). Thus, the replication program itself contributes to the polar positioning of loci that are close to the origin of replication. However, this contribution is masked by the presence of a Par system ( Figure 5B). The contribution of the replication program may be more evident in cells containing a unique but displaced oriC1 site. However, the presence of an oriC1 site within the origin region is essential for cell viability ( Figure S43).

Discussion
In this manuscript the use of fluorescent microscopy to follow the position of 19 loci within the genome of V. cholerae provides the first detailed characterization of the organization and dynamics of a multipartite bacterial genome.

The two V. cholerae chromosomes are longitudinally arranged
After division, the two replichores of chromosome I were arranged side by side from the old pole to the new pole (Figures 1  and 6A, WT). This longitudinal organization, with the origin positioned near the old pole and the terminus near the new pole in newborn cells, was expected for chromosome I due to the similarities in the positioning and dynamics of the origin and terminus regions with C. crescentus [30]. Previous reports suggested that the origin region of V. cholerae chromosome II was positioned at midcell in newborn cells and that, after replication, sister copies of it migrated to J and L positions [8]. This is similar to the positioning and dynamics of the E. coli chromosome, which suggested a transversal mode of organization for chromosome II.  Figure 1C. For the DparS1 choreography, the CP639 cell size distribution was aligned with the CP655 cell sizes using the cell size interval where $50% of cells contained two R2 I foci; For the DparS1 + oriC1 651 kb choreography, the cell sizes of CP626 and CP656 were aligned to the cell sizes of CP659 using the cell size interval where $50% of cells contained two ori I foci; For parS1 65 kb + oriC1 651 kb choreography, the cell sizes of ADV115 were realigned to the cell sizes of CP633 using the cell size interval where $50% of cells contained two ori I foci. (C) Relative distance between ori I and ter I loci, as in Figure 1D. (D) Relative distance between the ori I sister loci, as in Figure 1E. doi:10.1371/journal.pgen.1004448.g005 However, we demonstrate that the two replichores of chromosome II were arranged side by side with chromosome II only occupying the new half of the younger cells and the two sister chromatids the central part of the older cells (Figures 2 and 6B, WT). Thus, both V. cholerae chromosomes are longitudinally organized within the cell.

Positioning and segregation dynamics of the terminus region
The terminus region of chromosome I behaved differently from the bulk of the chromosome and was recruited to midcell long before the time of sister foci duplication (Figure 1). In addition, terminus sister copies remained together until the very end of the cell cycle ( Figure 1). Finally, whereas displacing parS sites led to the rotation of the whole chromosome in C. crescentus [30], the position of the terminus region of chromosome I was not affected by parS1 displacements (Figure 3). These observations are likely due to the presence of a MatP/matS system in V. cholerae that is absent from C. crescentus [37,38]. Our results also suggest that sister copies of the terminus region of chromosome II separated earlier than sister copies of the terminus region of chromosome I even though both regions harbour matS sites (Figure 2). The differential contribution of the MatP/matS macrodomain organization system to the segregation of the terminus regions of chromosome I and II is the subject of a another study [39].

Longitudinal organisation is not entirely driven by partition systems
As for all other studied bacterial chromosomes with a longitudinal organization the position of the origin of replication region of each of the two V. cholerae chromosomes was driven by a ParABS partition system [40]. We speculate that the longitudinal organization of each chromosome is linked to the action of their partition machineries. Consistent with this, the displacement of parS1 sites rearranged the bulk of chromosome I within the cell with the locus most proximal to the displaced parS1 sites now occupying the polar edge of the chromosome (Figures 3, 4A and  6A, parS1@650kb). This is similar to the impact of parS sites on chromosome organisation in C. crescentus [30]. However, chromosome I remained longitudinally arranged in the absence of parS1 sites suggesting the existence of additional factors involved in chromosomal organisation (Figures 5 and 6A, DparS1).

Replication contributes to chromosome segregation in V. cholerae
In C. crescentus sister loci segregate independently of their timing of replication in an order roughly corresponding to their distance from parS sites [29,41]. The results of this current study suggest that the segregation of chromosome I loci was no longer correlated to their genetic distance from parS1 sites when these sites had been displaced (Figures 3 and S44), but instead followed the replication program. The efficient segregation of V. cholerae chromosome I lacking parS1 sites suggests that parS1-polar mobilisation is not essential for the segregation process. The influence of the replication program on the organization and segregation of chromosome I was further assessed by inserting an ectopic origin of replication in the left replichore. The segregation and the organisation of the bulk of chromosome I was not altered if the partition machinery remained intact ( Figure 5). However, when combined with the deletion of the parS1 sites, the addition of an ectopic origin of replication affected the positioning of loci proximal to the two functional origins (Figures 5 and 4B). Taken together, these results suggest that the longitudinal organization of chromosome I was preserved by the replication program when its partition system was disrupted (Figures 5 and 6A, DparS1).
In E. coli and B. subtilis, modification of the replication program by the addition or the displacement of the replication origin does not influence the organisation of the bacterial chromosome [40,42]. In particular, replication initiation from an origin located in the middle of the right replichore did not disrupt the transversal organisation of the E. coli chromosome [40]. However, it increased the proportion of cells in which the origin sisters were displaced to the outer edges of the nucleoid, i.e. towards the cell poles, suggesting that replication program could contribute to polar mobilisation in E. coli. Moreover, the E. coli chromosome lost its transversal organisation and adopted a longitudinal organisation in mukB mutant cells [43,44]. Preliminary studies suggest that deletion of the mukBEF genes in V. cholerae, which does not confer any loss of cell viability or any aberrant cell morphology (data not shown), does not alter the positioning and segregation of ori I and L3 I ( Figure S24). Nevertheless, future work is needed to investigate the contribution of the MukBEF system to the organization and segregation of the two V. cholerae chromosomes.
In conclusion, this study demonstrates that the partition machinery and the replication program contribute to the polar mobilization of the origin regions in V. cholerae. We hypothesise that it is the genomic proximity of parS sites to the oriC site, which is a conserved feature of most bacterial chromosomes that serves to ensure the convergence of their polar positioning activities.

Plasmids and strains
Bacterial strains and plasmids used in this study are listed in Tables S1 and S2 respectively. All V. cholerae mutants were constructed by integration-excision or natural transformation (Protocols and details on the construction of each strain in Text S1). To this end, a derivative of the El Tor V. cholerae N16961 was rendered competent by the insertion of hapR by specific transposition [45]. Engineered strains were confirmed by PCR.

Fluorescence microscopy
Cells were observed in Minimal Media to have only a single copy of each chromosome after division. Protocols for Microscopy are detailed in Text S1. The snapshot images were analysed using the Matlab-based software MicrobeTracker [45,46], see details for the analysis in Supplementary Text S1.  Figure 1C. For the WT choreography, strain cell sizes were realigned as in Figure 2. For the DparS1 choreography, the cell sizes of ADV128 were aligned to the cell size of ADV27 using the first cell size interval where $50% of cells contained 2 foci of ori II and R1 II , respectively. (C) Relative distance between R1 II and ter II loci measured in the cells containing only one focus of each, as in Figure 1D. The positions of these 4 loci follow the order of the genetic map, from midcell (oriII) to the new pole (terII). Note, that the variability of the terII locus positioning was increased in DparS1 cells.  Figure S43 Presence of oriC1 site is essential within the origin region. After introducing an ectopic oriC1 using natural transformation with pPOS228, we attempted to delete the original Origin of replication of the chromosome I. Two strategies were performed: a direct deletion by replacing the oriC1 with a Rif resistance gene (pAD43), or an indirect deletion, by first replacing oriC1 by an oriC1 flanked by FRT sites (pAD44) and secondly deleting it by inducing a Flipase protein (pFLP2). When the second oriC1 was introduced in the middle of the replichore (near L3 I locus), we were unable to delete the original oriC1, even thought it was possible to replace it by a FRT-oriC1-FRT. It was however possible to delete this Origin when the second oriC1 was at only 50 kb from it. (EPS) Figure S44 Proportion of the different cell types in WT (ADV24) and in parS1 490 kb (CP634). (EPS)