The SMC Complex MukBEF Recruits Topoisomerase IV to the Origin of Replication Region in Live Escherichia coli

ABSTRACT The Escherichia coli structural maintenance of chromosome (SMC) complex, MukBEF, and topoisomerase IV (TopoIV) interact in vitro through a direct contact between the MukB dimerization hinge and the C-terminal domain of ParC, the catalytic subunit of TopoIV. The interaction stimulates catalysis by TopoIV in vitro. Using live-cell quantitative imaging, we show that MukBEF directs TopoIV to ori, with fluorescent fusions of ParC and ParE both forming cellular foci that colocalize with those formed by MukBEF throughout the cell cycle and in cells unable to initiate DNA replication. Removal of MukBEF leads to loss of fluorescent ParC/ParE foci. In the absence of functional TopoIV, MukBEF forms multiple foci that are distributed uniformly throughout the nucleoid, whereas multiple catenated oris cluster at midcell. Once functional TopoIV is restored, the decatenated oris segregate to positions that are largely coincident with the MukBEF foci, thereby providing support for a mechanism by which MukBEF acts in chromosome segregation by positioning newly replicated and decatenated oris. Additional evidence for such a mechanism comes from the observation that in TopoIV-positive (TopoIV+) cells, newly replicated oris segregate rapidly to the positions of MukBEF foci. Taken together, the data implicate MukBEF as a key component of the DNA segregation process by acting in concert with TopoIV to promote decatenation and positioning of newly replicated oris.

The Escherichia coli SMC complex, MukBEF, was identified through a genetic screen, which initially identified a Muk Ϫ mutant that generated anucleate cells and displayed temperature-sensitive growth in rich medium (10,11). MukB is a 170-kDa protein of distinctive SMC architecture that dimerizes through a dimerization hinge located at one end of an~50-nm long intramolecular coiled-coil and has an ATPase formed by the N-and C-terminal portions of the protein at the other end of the coiled-coil (9,12,13). A kleisin-like protein, MukF, bridges the two ATPase heads in a MukB dimer, while a third protein, MukE, binds to MukF. Loss of function of any of these three proteins leads to the same Muk Ϫ phenotype. Functional fluorescent fusions of MukB, MukE, or MukF all associate with the replication origin region (ori) of the E. coli chromosome in live-cell imaging assays that observe fluorescent-focus formation (14,15) in reactions that require ATP binding and hydrolysis by MukBEF (16, 17; our unpublished data). In the absence of functional E. coli MukBEF, oris are relocated from midcell to an old pole in newborn cells, with the whole chromosome undergoing an apparent 90°rotation within a cell (14).
Biochemical experiments in vitro have shown that the MukB hinge interacts with the C-terminal domain of ParC, the catalytic subunit of TopoIV (18)(19)(20), leading to the proposal that these proteins collaborate in chromosome disentanglement (20). This interaction stimulates TopoIV-mediated relaxation of negatively supercoiled DNA, but not positively supercoiled DNA in vitro. Because negative supercoils have the same right-handed chirality as replicative precatenanes (see Fig. S1 in the supplemental material), we predict that MukBEF should also stimulate decatenation by TopoIV, the prelude to chromosome segregation, raising the possibility that MukBEF and TopoIV act in sequential steps during chromosome segregation. Nevertheless, an in vitro study failed to demonstrate MukBEF-stimulated decatenation of multiply linked plasmid replicative catenanes, leading to the proposal that the MukBEF stimulation of TopoIV activity is involved in intramolecular chromosomal events rather than in decatenation (21). The failure of MukB to stimulate decatenation in in vitro assays may reflect the fact that the in vitro assay conditions failed to recapitulate in vivo conditions somehow, such as the absence of MukEF proteins or the fact that in vivo proper loading of a complete MukBEF complex onto DNA is required for this stimulation.
The role of TopoIV in decatenation in vivo has been widely documented, as has its ability to relax supercoils in vitro and in vivo (20,(22)(23)(24)(25). After TopoIV impairment, DNA replication and reinitiation along with cell growth appear to continue normally, although decatenation of newly replicated loci is blocked as measured by failure to segregate newly replicated sisters (3). Therefore, any action of TopoIV in supercoil removal ahead of a replication fork can be compensated for by the action of other topoisomerases. These observations support the view that the major role of TopoIV in vivo is in decatenation rather than supercoil relaxation (3). Indeed, any MukBEF-stimulated negative supercoil relaxation would compound the in vivo unlinking problem because it would act to increase the overall linkage between duplex strands in the chromosome. Other studies in both prokaryotes and eukaryotes have implicated functional interactions between SMC complexes and type II topoisomerases and have suggested that these are important for decatenation, although it is not always clear whether these interactions also influence the activity of the topoisomerase in regulating supercoiling, which is important for chromosome organization (5,(26)(27)(28).
To address whether MukBEF and TopoIV act in a coordinated manner in vivo, we analyzed the functional relationship between MukBEF and TopoIV in live E. coli in which a single round of replication was initiated and completed in the same cell cycle. Functional fusions of the fluorescent protein, mYPet, to the ParC and ParE subunits of TopoIV formed foci that associated specifically with fluorescent foci of functional MukBEF, which frequently localized with the E. coli ori region. The TopoIV foci were dynamic and did not always form discrete spots, probably because of a high dissociation rate of ParC from MukB, making it necessary for us to use a quantitative cumulative distribution method that assessed the distances between the centroid of Gaussian-fitted MukBEF fluorescent foci and the highest pixel intensity exhibited by fluorescent ParC or ParE, which likely marks the cellular site where the complex exhibits the highest residence time. Using this method, we demonstrated that the two subunits of TopoIV colocalize preferentially with MukBEF foci throughout the cell cycle and also in the absence of replication. Specific depletion experiments additionally showed that MukBEF in foci directs TopoIV to the ori region. Additionally, we provide evidence that MukBEF in foci positions ori rather than vice versa. Taken together, the data implicate MukBEF as a key component of the DNA segregation process by acting in concert with TopoIV to promote decatenation and positioning of newly replicated oris.

RESULTS
Topoisomerase IV associates with MukBEF in live E. coli. To address whether we could obtain evidence for an interaction of the ParC and ParE subunits of TopoIV with MukBEF in vivo, we replaced the endogenous ParC and ParE genes with functional fluorescent mYPet fusions to ParC and ParE expressed from their endogenous promoters on the E. coli chromosome and analyzed mYPet fluorescence in live cells using wide-field epifluorescence imaging. The cells also expressed MukB-mCherry and had their ori region marked with fluorescent tetracycline repressor (TetR-CFP) bound to an array of tet operators 15 kb counterclockwise (CCW) of oriC (ori1). Cells with the fluorescent fusions showed near-normal flow cytometry profiles and doubling times (see Fig. S2A and S2C in the supplemental material). Fluorescent MukBEF forms foci that are frequently associated with ori, irrespective of whether cells are growing exponentially, or nonreplicating because of a block in replication initiation (Fig. 1A) (14,15,17).
Although we observed fluorescent ParC-mYPet (ParC fused to mYPet fluorophore) and ParE-mYPet foci in both steady-state and nonreplicating dnaC(Ts) cells (Fig. 1A) (29), they were typically not as well defined as the ori1 and MukBEF foci, and a high background of ParC/ParE outside foci was evident. This is not surprising, given that in vitro analysis has shown a rather weak interaction of ParC with MukB (K d [dissociation constant] of 0.5 M) (18). Initial analysis by simple observation of all three fluorescence channels showed that the ParC-mYPet and ParE-mYPet foci were frequently associated with ori1 and MukBEF, indicative of an in vivo association with MukBEF and/or ori1 (Fig. 1A, images). Analysis of the three fluorescence profiles showed overlapping peaks for MukBEF, ori1, and ParC/E (Fig. 1A, line scans).
In order to establish a quantitative and more objective assessment of the relative localizations of TopoIV and MukBEF, we used automated image analysis that determined the centroid of Muk-BEF foci by fitting elliptical Gaussian functions. We examined the cumulative distributions of the distances between the brightest ParC/E pixels, which identify the population of ParC/E molecules with the highest residence time and the centroid of the nearest MukBEF focus ( Fig. 1B and C; Materials and Methods). For a negative control and to evaluate the level of random coincidence of foci, we performed the same analysis on a simulated random distribution of TopoIV foci within the same cells. The association of TopoIV with MukBEF was now unequivocal: 65% to 78% of the brightest ParC or ParE pixels were located within a distance of 4 pixels (516 nm) from a MukBEF focus, whereas the random distribution yielded 33% and 38%, respectively, for ParE and ParC versus MukBEF. Furthermore, a ParC mutant (ParC with an R-to-E change at position 705 and an R-to-A change at position 729 [ParC R705E/R729A ]), which showed an impaired interaction with MukB in vitro (19), retained almost the same level of colocalization with MukBEF, indicative of at least residual binding to MukBEF in vivo ( Fig. 1; 72% for mutant ParC compared to 78% for wild-type ParC). Comparable analysis showed similar high levels of colocalization of ParC or ParE with ori1 (see Fig. S2E in the supplemental material; 78% for ParC, 71% for ParE, with reduced colocalization of ParC R705E/R729A with ori1 [61%]), as well as MukBEF with ori1 (Fig. S2D). In our experience, it is not unusual for amino acid substitutions in proteins to lead to loss of in vitro activity, but the proteins retain at least some in vivo activity.
In nonreplicating dnaC(Ts) cells, MukBEF often formed foci close to and on both sides of the single ori1 focus [ Fig. 1A, images in the MukB-mCherry ParC-mYPet (no replication) panel;~66% of single ori1 foci were associated with two such MukBEF foci after 120 min at the restrictive temperature]. In this case, ParC was associated with MukBEF rather than with ori1, demonstrating that the primary association of TopoIV is with MukBEF rather than ori1. The observation that ParE colocalization with MukBEF or ori1 was always lower than that of ParC with MukBEF/ori1, when measured by the cumulative distribution of distances, likely reflects the fact that the ParE association with MukB is via ParC and that there is a significant dissociation rate in vivo between the two TopoIV subunits.
We conclude that the robust TopoIV-MukBEF colocalization demonstrated here likely reflects an interaction of MukBEF present in foci with TopoIV in vivo, thereby recapitulating the characterized in vitro reactions. We are confident that the observed association between ParC/ParE and MukBEF/ori1 is physiologically ParC/E-mYPet proteins within E. coli cells. The origin of replication region was identified by the binding of a TetR-CFP protein on a tetO array (ori1) located 15 kb CCW of the replication origin, oriC. The fluorescence profiles of the individual cells show the distribution of fluorescence intensities. Exponential-phase cells were grown in M9-glycerol medium at 30°C prior to imaging. Cells without replication are dnaC(Ts) strains grown at the restrictive temperature for 120 min. Bars, 2 m. (B) In order to localize the fluorescent foci corresponding to MukB and ParC/E proteins or the ori1 DNA locus within cells, microscopy images were first analyzed using the MicrobeTracker Suite (http://microbetracker.org/) to detect and outline bacterial cells. Cumulative distributions present the distances between the centroids of MukB foci and the brightest ParC/E pixels (Materials and Methods). Colocalization (gray shaded rectangle) is defined as when the MukBEF focus centroid is 4 or less pixels (516 nm) from the brightest ParC/E pixel. (C) The percentages of colocalization between MukBEF and ParC/E were plotted in a histogram. Error bars represent the 95% confidence interval. Asterisks indicate that the measured values (meas.) are statistically significantly different from the random calculated values (rand.) (Materials and Methods). relevant and not an artifact for the following reasons. ParC-mYPet and ParE-mYPet fluorescent foci form independently at the same cellular positions where they each colocalize with MukBEF-mCherry in cells that retain MukBEF and TopoIV function. ParC-mYPet and ParE-mYPet fluorescent foci form and colocalize with ori1 in the absence of labeled MukBEF (but dependent on the presence of functional MukBEF; see Fig. S2F in the supplemental material). Furthermore, the slightly reduced colocalization of the ParC double point mutant that has impaired interaction with the MukB hinge in vitro likely relates to the physiological relevance of the association between MukBEF and TopoIV. Finally, extensive analysis of a range of MukB and MukE fusions to mYPet, green fluorescent protein (GFP), and mCherry fluorophores has not only failed to show a fluorophore-dependent interaction but demonstrated in vivo localization and function dependent on ATP binding and hydrolysis (16, 17; our unpublished data). Similarly, experiments with the same combinations of fluorophores fused to replisome proteins did not reveal any fluorophore-interaction artifacts (30).
Because of the high background of TopoIV fluorescence, we cannot eliminate the possibility that TopoIV is additionally associated with other regions of the chromosome or with proteins elsewhere; indeed, since many of the MukBEF molecules are not in foci (17), interaction of these molecules with TopoIV could be responsible for at least some of the observed background. Colocalization of ParC foci with the replisome marker, DnaN, was close to random and no higher than that of the association of DnaN with ori1 (data not shown), as expected from our observation that ParC forms foci in dnaC(Ts) cells that are not undergoing replication (Fig. 1A). We are therefore unable to provide evidence that supports the earlier observations that ParC and/or SMC complexes, interact with the bacterial replisome (31)(32)(33).
MukBEF within foci positions TopoIV. In order to establish whether the observed colocalization between MukBEF and To-poIV is directed by MukBEF, TopoIV, or ori1, we used a MukE degron to deplete functional MukBEF (15). Representative cells in which MukE had been depleted are shown in Fig. 2A (images). The MukB-mYPet signal was now dispersed, whereas the ori1 foci remained intact. All evidence of clear ParC-mYPet focus colocalization with ori1 had also disappeared, with no clear foci elsewhere. Quantitative analysis showed colocalization between Muk-BEF, or ParC-mYPet, and ori1 had almost completely disappeared after 30 min of MukE depletion ( Fig. 2A, left). The cumulative distributions showed that the distance between an ori1 focus and a MukBEF focus or the brightest ParC pixel had become similar to that of the random distribution of localizations after 1 h of MukE depletion and the same as in a ⌬mukB strain ( Fig. 2B and C; see Fig. S3A in the supplemental material). The loss of ParC foci occurred also after MukE degradation in nonreplicating cells, conditions in which cell growth continued (Fig. S3B). In the absence of functional MukBEF, a low level of ParC localization with ori1 remained (Fig. 2C, bar graph; Fig. S3). This may reflect the fact that action of TopoIV at ori1 is required for timely ori1 segregation in the absence of functional MukBEF.
TopoIV depletion does not abrogate MukBEF foci. When the complementary experiment was undertaken, in which a degron derivative of ParC was depleted efficiently (see Fig. S4A in the supplemental material), MukBEF foci persisted. By 1 h of depletion, most cells had a single ori1 focus or two closely spaced ori1 foci, consistent with the expected impairment in decatenation of the ori region (Fig. 3, compare left and right panels) (3). In most cells, several MukBEF foci were evident, evenly spaced throughout the nucleoid and placed on both sides of ori1, sometimes with a high background throughout the nucleoid (e.g., see Fig. 3, 120 min with L-Ara). Two to four MukBEF foci were present in most cells, a number similar to the number of ori1s expected in such cells given that replication continues despite the inhibition of chromosome segregation and cell division after TopoIV impairment (3). The apparent regular position of the MukBEF foci within the nucleoid corresponded approximately to where ori1s would have been if their decatenation and segregation had been possible.
When a similar ParC depletion experiment was repeated in the absence of DNA replication, using a dnaC(Ts) allele at the restrictive temperature, the MukBEF foci persisted, with the majority of cells containing MukBEF foci close to and on either side of the single ori1 focus, as in ParC ϩ cells (see Fig. S4B in the supplemental material). This confirms that MukBEF foci persist in the absence of TopoIV and that the dispersed multiple MukBEF foci observed in replicating cells are dependent on ongoing DNA replication and/or the consequent accumulation of chromosomal DNA.
When ParE activity was impaired using a thermosensitive mutation at the restrictive temperature, a similar behavior was observed after 120 min at the restrictive temperature: cells contained a single ori1 focus or closely spaced ori1 foci, with~4 evenly spaced MukBEF foci throughout the nucleoid (see Fig. S4C in the supplemental material). On return to the permissive temperature, a proportion (~30%) of cells showed a similar distribution of MukBEF foci, but now with oris colocalized. The remaining fraction of cells seemed unable to decatenate and segregate their sister ori1s. To characterize the changes in distribution of MukBEF and ori1 foci during the transition from TopoIV-impaired to functional To-poIV, we undertook time-lapse experiments using parE(Ts) cells in which ori1 and MukBEF were differentially fluorescently labeled ( Fig. 4 and Fig. S5). As expected, the temperature shift to 42°C resulted in cells containing either a single ori1 focus close to midcell, or two separated ori1 foci at approximately cell quarter positions. Given that cells had been maintained for just over 1 cell generation at 42°C prior to the shift back to the permissive temperature, we expected each ori1 focus to contain 2 to 4 catenated ori1 regions at shift-down. Again we observed~30% of cells went on to segregate their ori1 regions and to eventually divide during the 5-h time-lapse experiment at the permissive temperature; in general, these were the shorter cells. In order to simplify the analysis, we focused on cells that started with a single ori1 focus, although the cells starting with two separated ori1 foci showed comparable behavior. We also analyzed only cells that went on to divide into normal looking and growing cells. Typical behavior is shown in the schematic in Fig. 4A, with the primary data for the same three cells shown in Fig. S5. During the 300-min time-lapse experiment, the initial single centrally placed ori1 focus segregated to give 2 to 4 well-separated ori1 foci, with the initial segregation event occurring anywhere between 30 min and 120 min after the shift to the permissive temperature. By the end of the experiment, the segregated ori1 foci were now frequently positioned coincident with or close to a MukBEF focus. We noted that the positions of MukBEF foci before the shift to the permissive temperature were similar to the positions they take up at times immediately prior to cell division. It is tempting to speculate that the MukBEF foci had taken up their final positions before the origins were able to segregate and that this then led to ori positioning; however, this observation is complicated by the fact that the MukBEF foci themselves are mobile during the time-lapse.
MukBEF foci position sister ori regions when the TopoIV-MukBEF association is retained. In the experiments that exploited TopoIV impairment and in those using TopoIV-positive (TopoIV ϩ ) dnaC(Ts) cells, we observed that the association of MukBEF foci with ori1 was reduced, with MukBEF foci taking up regular positions on both sides of ori1. These observations led us to consider that MukBEF focus formation is independent of ori although dependent on the nucleoid, because MukBEF foci always formed on nucleoids rather than in the nucleoid-free space (14). In order to explore the relationship between oris and MukBEF foci, we performed time-lapse experiments with a synchronized TopoIV ϩ dnaC(Ts) strain released after 105 min at the restrictive temperature, conditions under which we know that the majority of cells have a single unreplicated ori and no associated replisome (29). The cells expressed MukE-mYPet and had their ori region marked with fluorescent Lac repressor (LacI-mCherry) bound to an array of lac operators 15 kb CCW of oriC (ori1). The strain was released into permissive conditions to allow initiation of replication, and ori1 and MukBEF foci were monitored using time-lapse microscopy. We focused on two time points 5 min apart that span the interval in which the replication origin foci segregate (Fig. 4B, top portion). We observed that ori1 is generally colocalized with a MukBEF focus both before and after ori1 segregation; however, there are multiple MukBEF foci at times when there is only a single ori1 focus and not all MukBEF foci are colocalized with ori1. We observed that once the ori1 foci have segregated, the colocalization of MukBEF foci with ori1 foci was increased; before ori1 segregation, 46% of the MukBEF foci colocalized with ori1, whereas after ori1 segregation, 92% of MukBEF foci colocalized with ori1 (colocalization was defined as when the foci centers were less than 2 pixels [258 nm] apart). We also measured the total distances these foci moved along the long axis of the cell during the 5-min interval spanning ori1 segregation. As shown in Fig. 4B, we found that ori1 foci moved a greater total distance (median, 0.65 m) than the distance moved by the MukBEF foci (median, 0.38 m). This greater movement of the ori1 foci and the increased localization of MukBEF foci with ori after segregation support the idea that the ori region moves to the position of the MukBEF foci, and therefore, that MukBEF foci play a role in positioning segregated origin regions.

DISCUSSION
We have demonstrated that MukBEF and TopoIV associate in vivo and that this association occurs in the absence of DNA replication and at all stages of the cell cycle. We think it is highly likely that this association reflects the ParC-MukB interaction demonstrated in vitro (18,19,21). A key unanswered question regards the functional significance of the in vivo TopoIV-MukBEF association demonstrated here; specifically, does the MukBEF-directed association of TopoIV with ori modulate sister cohesion in the ori region by stimulating decatenation? Consistent with this hypothesis, earlier work and the work here have demonstrated that in Muk Ϫ cells, newly replicated ori1s can exhibit delayed segregation (14) (see Fig. S6 in the supplemental material), suggesting the possibility that the recruitment by Muk-BEF of TopoIV to the ori region facilitates decatenation of newly replicated sister oris. A quantitative demonstration of this effect was shown when we used a degron to deplete MukE, thereby leading to an abrogation of MukBEF function; the fraction of steadystate cells containing a single ori1 focus increased to 64% from 25% after MukBEF impairment (Fig. S6A). In order to test the hypothesis directly, we analyzed ori1 segregation in the ParC R705E/ R729A mutant that is impaired in its interaction with MukBEF in vitro, although any in vivo impairment appears to be modest. Snapshot analysis of steady-state cells showed an increase in the fraction of single ori1 focus cells from 26% to 37% in the mutant compared to the wild type, while time-lapse analysis showed an increase in cohesion time from 18 min to 23 min in the mutant compared to the wild type, when the time of ori1 segregation after replisome appearance was measured ( Fig. S6A and S6B). These apparent modest defects in decatenation are consistent with, but do not prove, a model in which the TopoIV-MukBEF interaction directs TopoIV-mediated decatenation to ori. A ParC mutant that is catenation proficient but is totally defective in the in vivo interaction with MukBEF (or the complementary mutant for MukB) would be needed to test the hypothesis directly.
The demonstration in vitro that the relaxation activity of To-poIV on right-handed (negative) supercoiled DNA, but not on left-handed (positive) supercoiled DNA, is stimulated by the interaction with MukBEF (21) suggests that the interaction should enhance decatenation because negative supercoils have the same right-handed chirality as replicative catenanes (see Fig. S1 in the supplemental material). Taken together, the in vitro stimulation of TopoIV activity on a substrate that mimics replicative catenanes and the observation of an in vivo association of MukBEF and TopoIV with ori makes us confident that TopoIV will be directed to ori to remove catenanes from newly replicated sisters. TopoIV may also be associated with the majority of MukBEF molecules that are not present as foci, therefore allowing TopoIV to act globally on the chromosome.
Our observations also lead us to conclude that MukBEF com- plexes within foci act to position the ori region and to additionally direct a fraction of TopoIV molecules to ori. Ablation of MukBEF leads both to ori mispositioning (14) and to the loss of ParC foci (this paper). Our observation that TopoIV impairment did not prevent MukBEF focus formation makes us confident that it is the MukBEF within foci that positions TopoIV, a conclusion supported by our observation that in dnaC(Ts) cells grown at 37°C, MukBEF foci on either side of ori1 colocalize with ParC rather than ori itself. We do not know what is positioning the MukBEF foci, but the regular patterning is reminiscent of that which occurs when ParAB-parS systems position low-copy plasmids, chemosensory apparatus, or carboxysomes on the nucleoid (1, 34). Since ori positioning and chromosome segregation are disrupted by Muk impairment, the data provide strong support for a model in which the coordinated and sequential action of TopoIV and Muk-BEF in complexes visualized as foci plays a central role in initiating segregation of newly replicated oris. An earlier study of TopoIV in E. coli found that the catalytic subunit, ParC, was associated with the replisome, while the other subunit, ParE, was located in nucleoid-free regions of the cell (31). Based on this and other observations, it was proposed that TopoIV activity is regulated temporally, with activity occurring at late stages of the cell cycle when ParC could be released from the replisome and associate with the FtsK translocase (35). We do not know how to reconcile this information with the results obtained in this work and in our earlier work (3), which together show that TopoIV can act at ori through its association with MukBEF. It is difficult to imagine how a single replisome, or FtsK complex, could interact with a substantial proportion of cellular TopoIV. Biochemical characterization has also shown that ParCE complexes are stable during gel filtration and ultracentrifugation (22), and therefore, it seems likely that a substantial proportion of ParC and ParE in cells can form functional topoisomerase molecules. We note that in the cell biology analysis of reference 31, immunocytochemistry was used in fixed rapidly growing cells, a situation where resolution is not at its highest. Finally, the roles of TopoIV in decatenating newly replicated regions of the chromosome and in replicating plasmids, which are positioned in the extranuclear space, have been documented (3,7,22,36). A genetic study of Bacillus subtilis indicated an interaction between the SMC complex and TopoIV (26), while studies of the budding yeast Saccharomyces cerevisiae have implicated functional interactions of both cohesin and condensin with topoisomerase II that are important in chromosome decatenation (27,28,37,38). It therefore seems possible that functional interactions between SMC complexes and topoisomerases are ubiquitous.

MATERIALS AND METHODS
Bacterial strains and growth. The bacterial strains used in this study are listed in Table 1. The plasmids and oligonucleotides used in this study are shown in Table S1 in the supplemental material. All strains are derivatives of the Escherichia coli K-12 AB1157 strain (39). Fusion of genes with fluorescent or degron tags was performed using the Red method (40). Fused genes were transferred to generate the final strains through P1 phage transduction (41). For multiple insertions of modified genes, the Kan r gene was removed using site-specific recombination through expression of the Flp recombinase from the pCP20 plasmid (40). All strains were constructed and analyzed several times independently in order to avoid any effect of potential suppressor mutations. dnaC(Ts) cells were grown at 37°C for 120 min to generate a population of cells that could not reinitiate replication, as described previously (15,29). Depletion of proteins carrying a degron tag necessitates the presence of the SspB protein which is expressed chromosomally under the arabinose-controlled promoter (pAra) in order to control the timing of the depletion (15). tetO and lacO arrays (240 copies) are inserted at 15 kb counterclockwise (CCW) of oriC (ori1).
Microscopy. Image captures were performed as described in reference 15. Cells were grown at 30°C in M9 medium supplemented with appropriate amino acids and 0.2% glycerol (M9-gly). For experiments in the absence of replication, strains carrying the dnaC(Ts) allele were shifted 2 h at 37°C (to synchronize cells by allowing completion of existing rounds of replication but preventing any further replication initiation), and the microscope chamber was maintained at 37°C in order to prevent any reinitiation of replication during image acquisition. Induction of SspB protein during depletion experiments was done by addition of 0.5% L-arabinose. For microscopy, cells in exponential phase (A 600 Ϸ 0.1) were concentrated and laid on a 1% M9-gly agarose pad on a slide. During depletion experiments, the L-arabinose was also maintained in the slides when required.
For the parE(Ts) time-lapse experiments, cells were grown at 30°C in M9-gly to an A 600 of~0.05. The cells were shifted to 42°C for 2 h, and then a sample was spun down and spotted onto a prewarmed 1% M9-gly agarose pad on a slide. The cells were imaged at 30°C for 5 h; the initial images were taken 10 and 30 min after release and then every 30 min up until 300 min.
For the dnaC(Ts) time-lapse experiments (Fig. 4B), cells were grown at 30°C in M9-gly to an A 600 of~0.05. The cells were shifted to 37°C for 105 min (the restrictive temperature for dnaC2). A sample was spotted onto a prewarmed 1% M9-gly agarose pad on a slide. The cells were released to the permissive temperature of 30°C to allow replication initiation and imaged every 5 min for 2 h. Image analysis. Images were taken and processed by Metamorph 6.2, and image analysis was done using ImageJ or specific programs run in Matlab.
(i) Semiquantitative analysis of fluorescence distributions. Fluorescence distributions within cells were plotted as line scans using the Plot Profile command of ImageJ or MicrobeTracker software run in Matlab (42). Maximum intensity values were normalized between 0 and 100% for each channel before plotting. For the parE(Ts) experiments (Fig. 4A), peaks in intensity were defined as peaks when the height of the peak was 5% or more of an increase in intensity above the neighboring points of inflection.
(ii) Quantitative analysis of colocalization. Cell outlines were first delineated from a phase image using the MicrobeTracker software run in Matlab (see Fig. S7 in the supplemental material). This segmenting analysis created a "mesh" for each cell, within which each pixel is characterized by a specific x,y coordinate (Fig. S7Bi). This step is critical to determine pairwise distances between two independent pixels of different channels within a specific cell (see further steps).
The second step of the analysis was to find the population of fluorescent molecules with the highest residence time within a cell. These populations of molecules assembled as local maxima of fluorescence intensity. As both TetR-CFP (cyan fluorescent protein) bound to the ori1 locus and MukB-mCherry formed well-defined foci, we adapted the automated localization analysis method of Holden et al. (43) that first identified candidate foci above an intensity threshold and subsequently determined their centroids by fitting an elliptical Gaussian function (see Fig. S7Biii and S7Biv in the supplemental material). This kind of analysis was not possible for ParC/E-mYPet signal, as ParC/E did not always assemble discrete foci. In order to localize the population of ParC/E molecules with the highest residence time, we determined the brightest pixel within each cell (Fig. S7Bii). It should be noted that the Gaussian localization analysis for MukBEF and ori1 can identify multiple fluorescent foci within one cell or none at all, but the brightest pixel analysis finds exactly one pixel with the highest intensity for ParC/E. The third step of the analysis measured the pairwise distances between the brightest ParC/E pixel and the nearest MukBEF or ori1 localization (see Fig. S7C in the supplemental material). To determine the distribution of distances expected from an entirely random localization of ParC/E, we also calculated distances between a pixel randomly positioned within the cell and the nearest MukBEF or ori1 focus (Fig. S7C). For MukBEF-ori1 colocalization analysis with multiple MukBEF foci within each cell, an equal number of random MukBEF localizations was generated per cell, and the smallest pairwise distance was calculated.
In the fourth step, the distances were plotted as cumulative distributions (see Fig. S7D in the supplemental material). A threshold of 4 pixels (516 nm) was chosen to define colocalization. The fraction of cells with colocalizing foci was thus determined from the cumulative distributions at 4-pixel distance with 95% statistical confidence bounds (Fig. S7D). Note that the cumulative distribution curves do not reach 100% of cells even for large distances, because MukBEF or ori1 foci can be absent in some cells; the asymptotic maximum values hence give the fraction of cells that showed MukBEF or ori1 foci. For example,~90% of wild-type cells displayed MukBEF foci (Fig. 1B), which reduced to~30% MukBEF foci under MukE degron conditions (Fig. 2B).
(iii) MukB foci distribution during ParC/E depletion. Because the automated analysis was not sufficiently robust to identify and localize the larger number of closely spaced MukBEF foci during the course of ParC/E depletion, the distances in pixels between centroids of MukB-assembled foci and the ori1 locus were manually measured using ImageJ software and plotted as histograms.

ACKNOWLEDGMENTS
The work was supported by a Wellcome Trust program grant to D.S. (WT083469MA) and by a Wellcome Trust strategic award (091911) supporting advanced microscopy at Micron Oxford (Department of Biochemistry, Dunn School of Pathology, University of Oxford, Oxford, United Kingdom). We thank Ian Dobbie for his generous support and advice on imaging.