FtsK actively segregates sister chromosomes in <em>Escherichia coli</em>

Improving Loci Visualization Systems.

To visualize chromosome loci, we used previously described systems (6, 23). We inserted parS sites from plasmids P1 and pST1 (parS_P1 and parS_pMT1, respectively) or an array of tet operators (tetO) at a set of chromosome loci (Tables S1–S3). Strains carrying these constructs were observed in slow-growing conditions to facilitate the observation of segregation events at nearby loci (Material and Methods). In these conditions, our reference strain grew with a 210-min doubling time with no overlapping replication cycle and an extended postreplicative period (Fig. S1).

ParB_P1 and TetR-derived fluorescent protein were produced from constructs inserted at the lac and ara chromosome loci, respectively, and the ParB_pMT1-derived protein from the low copy number plasmid pMS11 (Fig. 1A and Tables S1–S3). Fluorescent foci were readily detected without inducer using pMS11, but a moderate concentration of reducers was required using the chromosome-borne constructs (Fig. 1B). Most interestingly, the two ParB-derived systems yielded equivalent numbers of foci per cell, slightly more than the TetR-derived system (Fig. 1C). This result contrasted with previously described systems for the production of ParB derivatives (23, 24). In our strains, XFP-ParB_P1 proteins produced from previously described plasmids provoked a significant delay in loci segregation, even when no inducer was added (Fig. S2B). This effect was significantly enhanced in faster-growing cells (Fig. S2C).

Fig. 1.

Systems used for visualization of chromosome loci. (A) Relevant map of constructs for fluorescent protein production. Plasmid pMS11 is a pSC101 derivative carrying a lacZp-mCherry-Δ30ParB_P1 construct. GFP-Δ23ParB_pMT1 and TetR-GFP were produced from chromosome-borne constructs at the lac and ara loci, respectively. (B) Examples of micrographs showing foci of the indicated fluorescent proteins in strains containing loci tagged with the relevant binding sites (ydgJ::tetO or parS_P1; ydcB::parS_P1, and ydeU::parS_pMT1). (C) The percentage of cells with a single focus or two foci of the ydgJ (ter) or ydhJ (right) loci when visualized using fluorescent proteins produced from constructs shown in A.

We next observed the effects of localization systems on the cell cycle by measuring the number of replication origins per cell using flow cytometry (Fig. S3). The ParB_pMT1 system using plasmid pMS11 did not modify this ratio compared with the wild-type strain. The two other systems provoked the appearance of cells with four replication origins, indicating that cell division was delayed in some cells, undoubtedly because of replication and/or segregation defects. These effects were most pronounced with the TerR system, whereas the ParB_P1 system provoked less than 10% four-foci cells at any locus. We thus used the two ParB-derived systems for the simultaneous localization of two loci.

The Chromosome Segregation Pattern Is Highly Accurate.

Although the ori-to-ter segregation pattern is well established, it is unclear whether this pattern pertains to all cells or if it is subject to important cell-to-cell variation. To measure cell-to-cell variation, we compared the segregation of a locus close to the dif site (ydeU, 2 kb clockwise from dif) with that of other loci. To minimize the risk of artificial segregation delay at the ydeU locus, we used the ParB_pMT1 system to localize it and the ParB_P1 system to localize the other loci: an oriC-proximal locus (ilvA); two midreplichore loci (yffS and ybhJ), and a ter locus (ycgY, 345 kb from dif). When the two systems were used in the same cells, both loci were readily detected without the addition of an inducer because of the titration of the LacI repressor by pMS11-borne lacZp (Fig. 1B). Less than 1.5% cells lacked foci of either visualization system. Cells were classed into categories according to their number of foci at each locus. In cells with two foci, the ratio of the interfocal distance to cell size did not increase with the cell size (Fig. S4). The spatial resolution of sister foci thus is followed quickly by their rapid displacement to opposite halves of the cell so that this movement is not detected in snapshot analysis. This result indicates that the time when sister loci lose their colocalization reflects the kinetics of the whole segregation process in our analysis. As expected, the ratio of cells with one focus for each locus increased with decreasing distance from ydeU (Fig. 2, open bars), and the ratio of cells with two foci at each locus was constant because it depended on ydeU segregation (gray bars). Loci thus segregated in their order of replication. The mean times of postreplicative colocalization, calculated from the inferred replication and measured segregation mean times (see Fig. S1 and legend), ranged from 35 min for ybhJ and ycgY to 50 min for ilvA.

Fig. 2.

The ori-to-ter segregation of the chromosome. The ydeU locus was tagged using the ParB_pMT1 system (red arrowhead),and other loci, indicated inside the chromosome maps, were tagged with the ParB_P1 system (green arrowheads). Positions of loci on the chromosome are indicated by lines representing ori (top) and dif (bottom). The number of cells analyzed is given (n). Cells were classified by the number of foci of each locus (shown in cartoons on the x-axis; the empty cell indicates cells that fall in none of the first four categories). Bars show the mean percentage of each category in the population (y-axis) with individual measured ranges. Data reflect at least two independent experiments.

Cells with three foci (two foci at one locus and one focus at the other) would fall in two classes depending on which focus was duplicated (Fig. 2). The proportion of cells with two ParB_P1 foci and one ParB_pMT1 focus decreased with the distance from ydeU (green bars), consistent with decreasing time between the segregation of the two loci. Strikingly, less than 1% of cells with three foci had duplicated ydeU foci (red bars). Thus the ydeU locus is almost always segregated later than the other loci, including the ycgY locus located in the ter region. The segregation pattern of the chromosome thus is highly accurate, the ter region being the last segregated in almost all cells. These data also pointed out that segregation of the ter region occurs in an ordered sequence, the ydeU locus, closest to dif, being segregated later than ycgY, which is more remote from dif. This result prompted us to analyze ter segregation in more detail.

Chromosome Segregation Ends at dif.

To obtain a detailed view of ter segregation, we constructed insertions of parS sites and tetO arrays at additional ter loci (Fig. 3A). We first analyzed the positioning of single loci depending on cell size (Fig. 3B and Fig. S4). All loci tended to localize close to a cell pole in smaller cells and then migrate to midcell before foci duplication (Fig. 3C). Loci closer to the dif site tended to segregate later, as shown by their lower foci number per cell. We also noticed that loci closer to the dif site tended to stay closer to each other after segregation, as shown by the lower average interfocal distance in cells with two foci (Fig. 3C). We then compared the ycgY and ydgJ loci (342 and 114 kb from dif, respectively) in more detail. We introduced an ssb-mCherry fusion gene in strains tagged at these loci. Cells longer than 3 μm and harboring no SSB-mCherry focus were considered to be in their D period (see also Fig. S1). These cells were categorized into two classes, depending on whether or not they harbored a midcell constriction (38 and 62%, respectively). Segregation of ydgJ was clearly delayed compared with that of ycgY in all cell categories (Fig. 3D). From these data, we inferred that the mean segregation time of ydgJ is 35 min before cell separation, concomitant with the onset of constriction, whereas ycgY segregates about 10 min earlier (see Fig. S1 for calculations).

Fig. 3.

ter segregation is sequential and ends at dif. (A) Map of the loci used, with coordinates indicated. The black and white box represents the dif site. Red loci were tagged with both ParB-derived systems, black loci were tagged with the TetR-derived system, and the ydgJ and ycgY loci were tagged with all three systems. (B) Intracellular position of foci tagged with parS_P1 from their farthest pole (x-axis) as a function of cell length (y-axis). (Left) Cells with a single focus (black dots). (Right) Cells with two foci (red and green dots). Loci are indicated, as well as the number of cells analyzed and the mean interfocal distance (IFd). (C) Plot of the interfocal distance in cells with two foci (black curve and dots, left axis) and the mean number of foci per cell (blue curve and squares, right axis) as a function of the distance from dif (kb, x-axis). (D) Number of foci of the indicated loci in different cell categories. Strains carried an SSB-mCherry fusion. Postreplicative cells are >3 μm long and harbor no SSB focus and no constricting septum. Dividing cells have a constricting septum. (E) Tagged loci are indicated with their position relative to dif. Red arrowheads indicate a parS_pMT1 tag, and green arrowheads indicate a parS_P1 tag. Cells were classified by the number of foci of each locus (shown in cartoons on the x-axis; the empty cell indicate cells that fall in none of the first four categories). Bars show the mean percentage of each category in the population (y-axis) with individual measured ranges. Data reflect at least two independent experiments and more than 600 cells.

We next constructed strains with chosen pairs of loci tagged with parS_P1 and parS_pMT1 sites. The localization systems were swapped between loci to detect any system-induced artifact. Fig. 3D shows the segregation of the ydeU locus compared with ydcB and ydgJ (100 kb counterclockwise and 100 kb clockwise from dif, respectively). As expected, all three loci segregated late; hence most cells had one focus at each locus. Between 7 and 10% of the cells had three foci. Among these, less than 6% had duplicated ydeU foci, whereas more than 94% had duplicated ydcB or ydgJ foci. We concluded that the ydeU locus segregates later than the ydcB and ydgJ loci in the vast majority of cells. Equivalent results were obtained when comparing segregation of the ydeP locus (3 kb counterclockwise from dif) with that of ydcB and ydgJ (Fig. S5 A–D): ydeP segregated later than the two other loci in most cells. In all cases, swapping the localization systems between loci modified neither their timing nor their order of segregation (Fig. 3D and Fig. S5 A–D).

Consistent with an ori-to-dif pattern of segregation, the arpB locus (214 kb from dif) segregated earlier than the dif site (ydeU) and than the ydgJ locus, located closer to dif on the same side (Fig. S5 E and F). However, no segregation order was observed between loci located on opposite sides of dif (Fig. S5 G and H). These data show that the whole chromosome, including the dif-proximal part of the ter region, segregates in an accurately ordered manner, the dif site being the very last to be segregated in the great majority of cells.

Monomeric Chromosomes Are Segregated Orderly.

The orderly segregation of the dif region is unlikely to be restricted to dimeric chromosomes; otherwise the loci in this region would have to segregate simultaneously in monomeric chromosomes, thus minimizing the number of three-foci cells with two foci of dif-proximal loci. To reject this hypothesis unambiguously, we inactivated RecA in cells carrying pairs of tagged loci, thereby preventing dimer formation (25, 26). RecA inactivation resulted in the appearance of about 3% dead cells (blue bars in Fig. 4 and Fig. S6H). Most living cells were of normal size and had two to four foci. The order of segregation between dif-proximal loci (ydeU and ydeP) and loci located about 100 kb (ydbC and ydgJ) or further away from dif (ycgY) was conserved compared with wild-type strains (green and red bars in Fig. 4; compare Fig. 4 with Figs. 2 and ​and3D3D and Fig. S5B). This result shows that ordered segregation is not restricted to chromosome dimers.

Fig. 4.

Segregation of ter in recA ^{strains. (}A–C) Tagged loci are indicated in their position relative to dif (the black and white box). Red arrowheads indicate a parS_pMT1 tag, and green arrowheads indicate a parS_P1 tag. Cells were classified by the number of foci of each locus (shown in cartoons on the x-axis; the empty cell indicate cells that fall in none of the first four categories; blue bars indicate dead cells, see also Fig. S6). Bars show the mean percentage of each category in the population (y-axis) with individual measured ranges. Data reflect at least two independent experiments and 600 cells.

We then asked if the XerCD/dif complex is required for ordered segregation. As expected, inactivation of XerC resulted in the appearance of 10–15% dead cells and cells of abnormal shape (blue bars in Fig. S6 C and D). We then inactivated RecA in xerC ^{strains, thereby lowering the number of abnormal cells to few percent of the population (}Fig. S6 E and F). In both xerC ^and xerC^{recA strains, the order of segregation of the} dif region was conserved in normal cells as compared with wild-type strains (Fig. S6 C–F). We conclude that the XerCD/dif system is not required for orderly segregation of the dif region.

FtsK Segregates the dif-Surrounding Region.

FtsK is an obvious candidate to play a direct role in an ordered segregation process ending at the dif site because it translocates chromosomal DNA toward dif using polarization of the KOPS motifs. We thus attempted to analyze ter segregation in ftsK mutants. We first inactivated translocation using the ftsK_ATP and Δ(ftsK_C) alleles. Both mutations resulted in the appearance of 15–20% dead cells reflecting inactivation of chromosome dimer resolution (Fig. 5 A and B and Fig. S6 A and B). In addition, 50–60% of living cells were classified as abnormal (salmon bars in Fig. 5 and Fig. S6). These include cells of abnormal shape or length or that showed no fluorescent signal or an abnormal number of foci. These effects may be partially linked to a role of the translocase domain of FtsK in the control of cell division (21). It renders the analysis of foci number per cell in normal cells hazardous, because we cannot determine if it impoverished some cell categories more than others. Nevertheless, we noticed that ordered segregation of the dif region seemed to be lost because of the inactivation of translocation (red and blue bars in Fig. 5 A and B and Fig. S6 A and B). We then used the ftsK_KOPSblind mutant, which shows a milder phenotype than translocation-deficient mutants and supports resolution of the largest part of chromosome dimers (20). The FtsK_KOPSblind protein activates XerCD/dif recombination but translocates in a nonoriented manner in vivo and in vitro because of its defect in KOPS recognition (20, 27). As expected, ftsK_KOPSblind strains showed very few abnormal or dead cells. The ratio of cells with duplicated foci of ter loci tended to decrease, suggesting that they segregated later because of the ftsK_KOPSblind mutation (compare Fig. 5 and Fig. 3E). Strikingly, the order of the segregation of the ydcB and ydgJ loci compared with the dif site (ydeU and ydeP) was lost or was significantly altered as compared with wild-type strains (red and green bars in Fig. 5 A–D; compare with Fig. 3E and Fig. S5). However, the order of segregation between ycgY, which is outside the region of high FtsK activity (15), and the dif site (ydeU) was unaffected by the ftsK_KOPSblind mutation (Fig. 5E). We conclude that the KOPS-oriented translocation activity of FtsK is required for the ordered segregation of a large (at least 210 kb) but restricted region around dif.

Fig. 5.

Segregation of ter in ftsK mutant strains. Tagged loci are indicated in their position relative to dif (the black and white box). Red arrowheads indicate a parS_pMT1 tag, and green arrowheads indicate a parS_P1 tag. (A) ftsK_ATP- strain. (B–F) ftsK_KOPSblind strains. Cells were classified by the number of foci of each locus (shown in cartoons on the x-axis; the empty cell indicates cells that fall in none of the first four categories; blue bars indicate dead cells; see also Fig. S6). Bars show the mean percentage of each category in the population (y-axis) with individual measured ranges. Data reflect at least two independent experiments and 600 cells.

MatP Controls ter Segregation.

Because MatP delays ter segregation, we reasoned that its activity might be required for FtsK-dependent ordered segregation of ter. We inactivated MatP in cells carrying pairs of tagged loci and monitored their number of foci per cell. A small number of cells longer than wild-type cells and harboring more than four foci appeared in matP ^{strains, consistent with a mild defect in cell division caused by MatP inactivation (}11). Among cells of normal size, the ratio of cells with four foci (i.e., with both loci segregated) increased in matP ^{strains compared with wild type (compare gray bars in} Fig. 5 A and B with Fig. 3D), consistent with earlier segregation of ter loci as previously observed (11). Cells with three foci (red and green bars in Fig. 5 A and B) were underrepresented because of matP inactivation, indicating that the mean time separating segregation of the two loci (i.e., ydcB and ydeU) was shorter. Among these cells, no consistent order of segregation was detected between the ydcB and ydgJ foci and the dif site (ydeU and ydeP) (Fig. 5 A and B). The order of segregation in the dif-surrounding region thus was lost with MatP inactivation, showing that MatP is required for the ordered segregation of this region. Interestingly, a preferred order of segregation still was apparent between the ycgY locus and dif, although it was less pronounced than in wild-type cells (compare Fig. 5C and Fig. 2). The ycgY locus lies in the region covered by matS sites but not in the region processed by FtsK (see above). This result strongly suggests that MatP is required for ordered segregation only in the region processed by FtsK.

Strains and Plasmids.

Strains used were derived from E. coli K12 strain LN2666 (W1485 Fleu thyA thi deoB or C supE rpsL (StR) (38) and are listed in Table S1. Plasmids used are listed in Table S2. Tagged chromosome loci are listed in Table S3.

Cell Imaging.

Cells were grown in M9 medium containing alanine as a carbon source (0.2%); thiamine (1 μg/mL), thymine (2 μg/mL), and leucine (2 μg/mL). After 48 h cultures were diluted 250 times in fresh medium and were grown to OD₆₀₀ = 0.2 before being mounted on microscope slides. Arabinose (0.2%) or isopropylthio-β-galactoside (50 µM) were added 45 min before mounting, and FM4-64 membrane dye was added 5 min before mounting. Cells were applied to a 1% agarose slide and placed in the microscope temperature-controlled box at 30 °C for at least 30 min before observation on a Ti-E inverted microscope (Nikon) carrying a 100× oil-immersion lens (N.A. 1.36) and a Hamamatsu ORCA-R2 camera. Images were acquired using NIS Elements AR 3.2 software (Nikon) and were converted for manual analysis with MetaMorph v. 7.0 software (Molecular Devices) using a pencil pad.

Supplementary Material