The chromosome partitioning proteins Soj (ParA) and Spo0J (ParB) contribute to accurate chromosome partitioning, separation of replicated sister origins, and regulation of replication initiation in Bacillus subtilis
Soj (ParA) and Spo0J (ParB) of Bacillus subtilis belong to a conserved family of proteins required for efficient plasmid and chromosome partitioning in many bacterial species. Unlike most Par systems, for which intact copies of both parA and parB are required for the Par system to function, inactivating soj does not cause a detectable chromosome partitioning phenotype whereas inactivating spo0J leads to a 100-fold increase in the production of anucleate cells. This suggested either that Soj does not function like other ParA homologues, or that a cellular factor might compensate for the absence of soj. We found that inactivating smc, the gene encoding the structural maintenance of chromosomes (SMC) protein, unmasked a role for Soj in chromosome partitioning. A soj null mutation dramatically enhanced production of anucleate cells in an smc null mutant. To look for effects of a soj null on other phenotypes perturbed in a spo0J null mutant, we analysed replication initiation and origin positioning in (soj-spo0J)+, Δsoj, Δspo0J and Δ(soj-spo0J) cells. All of the mutations caused increased initiation of replication and, to varying extents, affected origin positioning. Using a new assay to measure separation of the chromosomal origins, we found that inactivating soj, spo0J or both led to a significant defect in separating replicated sister origins, such that the origins remain too close to be spatially resolved. Separation of a region outside the origin was not affected. These results indicate that there are probably factors helping to pair sister origin regions for part of the replication cycle, and that Soj and Spo0J may antagonize this pairing to contribute to timely separation of replicated origins. The effects of Δsoj, Δspo0J and Δ(soj-spo0J) mutations on origin positioning, chromosome partitioning and replication initiation may be a secondary consequence of a defect in separating replicated origins.
Bacillus subtilis possesses chromosomally encoded homologues of ParA and ParB, called Soj and Spo0J respectively. Spo0J binds to at least eight parS sites spread over nearly 800 kb in the origin-proximal 20% of the chromosome (Lin and Grossman, 1998) and probably brings the parS sites together, forming a nucleoprotein complex that can be visualized as a focus using immunofluorescence microscopy or a green fluorescent protein (GFP) fusion to Spo0J (Lin et al., 1997; Sharpe and Errington, 1998; Teleman et al., 1998). The subcellular location of Spo0J foci reflects the position of chromosomal origins (Lin et al., 1997; Sharpe and Errington, 1998; Teleman et al., 1998). In the Δspo0J mutant, chromosome partitioning occurs with reduced accuracy, and replicated origins are mispositioned such that they are often closer together than the cell quarters (Ireton et al., 1994; Lee et al., 2003; Ogura et al., 2003), leading to the hypothesis that Spo0J may contribute to separation of replicated origins and/or maintenance of replicated origins at the cell quarters. Soj and Spo0J probably do not function to recruit parS sites to the cell quarters, because they are not sufficient to recruit an array of parS sites inserted elsewhere in the chromosome to the cell quarters (Lee et al., 2003). The Δspo0J mutant is pleiotropic, leading also to asynchronous and early initiation of replication (Lee et al., 2003; Ogura et al., 2003), and a defect in spore formation. Spo0J is needed for efficient sporulation because it relieves Soj-mediated transcriptional repression of sporulation genes (Ireton et al., 1994; Cervin et al., 1998; Quisel et al., 1999; Quisel and Grossman, 2000). The mechanism by which Spo0J regulates initiation of replication is not known.
This study presents comprehensive phenotypic analyses of a Δsoj mutant in comparison with Δspo0J and Δ(soj-spo0J) mutants. The results show that Soj and Spo0J contribute to accurate chromosome partitioning, origin separation and regulation of replication initiation. Although a Δsoj null mutant does not have a detectable chromosome partitioning defect on its own (Ireton et al., 1994), inactivating soj in a background lacking smc, the gene encoding structural maintenance of chromosomes (SMC) protein, led to a nearly 10-fold increase in anucleate formation, a degree of enhancement similar to Δspo0J and Δ(soj-spo0J) mutants. Thus, the presence of SMC, a chromosome compaction protein, normally masks the contribution of Soj to chromosome partitioning. The Δsoj, Δspo0J and Δ(soj-spo0J) mutants all showed overinitiation of replication and, to varying extents, defects in positioning the origin of replication. Using an assay to measure separation of chromosomal regions, we found that inactivating soj and/or spo0J leads to defects in separating replicated origins, such that a significant proportion of cells contain replicated origins that have not separated enough to be spatially resolved. Separation of a region outside the origin was not affected. The finding that efficient origin separation requires a committed mechanism suggests that there may be origin region-specific processes or factors that hold sister origins together for part of the replication cycle, and that Soj and Spo0J function to overcome these factors.
Results and discussion
Inactivating soj enhances the chromosome partitioning defect of an smc null mutant
As a spo0J null mutation strongly enhances the chromosome partitioning defect of a null mutation in the smc gene (Britton et al., 1998; Britton and Grossman, 1999), we tested whether a soj null mutation caused a similar enhancement. SMC is a DNA-binding protein that contributes to chromosome compaction and organization (Hirano, 1998; 2002; Lindow et al., 2002a). A soj null mutation significantly enhanced the chromosome partitioning defect of an smc null mutant (Table 1). While most of the nucleoids in the Δsmc single mutant had regular size and spacing, the Δsoj Δsmc double mutant had dramatically perturbed nucleoid morphology (Fig. 1), with irregular nucleoid size, anucleate and cut cells (in which the division septum bisects the nucleoid). The Δsoj single mutant appeared similar to wild-type cells (data not shown). A Δsoj mutation enhanced the partitioning defect of an Δsmc mutant just as dramatically as a Δspo0J mutation did. A Δsoj mutant did not have a detectable chromosome partitioning defect on its own, while a Δspo0J mutant had 0.3% anucleate and 0.2% cut cells (Table 1). An Δsmc single mutant had 2.1% anucleate and 2.5% cut cells, and combining this mutation with Δsoj, Δspo0J or Δ(soj-spo0J) led to similar, strong levels of enhancement with production of 18–19% anucleates and 8–12% cut cells (Table 1). The chromosome partitioning defect of the Δsmc mutant was lower than previously reported (Britton et al., 1998; Moriya et al., 1998). This was due to differences in the growth media and method of culture inoculation (see Experimental procedures).
Table 1. Chromosome partitioning defects in various mutants.
Indicated strains were inoculated from light lawns on minimal plates into defined minimal glucose medium at 30°C. Samples were taken for microscopy during exponential growth. The cell membrane was stained with FM4-64 and the nucleoid was stained with DAPI.
Cells devoid of any visible DAPI staining were scored as anucleate.
Cells containing a nucleoid bisected by the division septum were scored as cut.
The total number of cells analysed.
In B. subtilis, SMC can form a complex with two other proteins, ScpA and ScpB (Lindow et al., 2002b; Soppa et al., 2002; Volkov et al., 2003). Inactivating soj enhanced anucleate production in the ΔscpA or ΔscpB mutants by nearly 10-fold, similar to the enhancement in the Δsmc mutant (Table 1). Our results indicate that the synthetic chromosome partitioning defects are due mainly to loss of the SMC–ScpA–ScpB complex.
The Δsmc mutant background uncovers a role for Soj in chromosome partitioning. The finding that Δspo0J, Δsoj and Δ(soj-spo0J) mutations enhanced the chromosome partitioning defect of an smc null mutation to a similar extent indicates that Soj and Spo0J probably perform chromosome partitioning functions in the same pathway, analogous to ParA and ParB from other organisms (Austin and Abeles, 1983a,b; Ogura and Hiraga, 1983; Mohl and Gober, 1997; Easter and Gober, 2002; Figge et al., 2003). We suspect that Soj may help Spo0J bring the parS sites together, forming an organized nucleoprotein complex that compacts the origin region. Consistent with this model, inactivating soj can cause foci of Spo0J–GFP to mislocalize as many smaller, fragmented foci (Marston and Errington, 1999). ParA from other organisms appears to modulate the size of ParB nucleoprotein complexes on parS-containing DNA (Mohl and Gober, 1997; Bouet and Funnell, 1999; Lemonnier et al., 2000; Figge et al., 2003), indicating that ParA proteins may generally regulate ParB binding and/or higher-order interactions between ParB molecules bound at the partition complex. To account for why a Δsoj mutant does not have a detectable partitioning defect on its own, we propose that SMC-mediated chromosome compaction could also help to bring parS sites closer together. This model was motivated in part by the finding that nucleated cells of an Δsmc null mutant appear to have a defect in assembling Spo0J–GFP foci (Britton et al., 1998).
Inactivating soj, spo0J or both leads to overinitiation of replication and production of elongated cells
After uncovering a role for Soj in chromosome partitioning, we decided to explore whether a soj null mutant has additional phenotypes that are also perturbed in a spo0J null mutant. As a Δspo0J mutant overinitiates replication and produces elongated cells (Lee et al., 2003; Ogura et al., 2003), we examined the effects of soj on these phenotypes. Overreplication was monitored by measuring the DNA to protein ratio and cell lengths of wild-type, Δsoj, Δspo0J and Δ(soj-spo0J) cells. The DNA to protein ratio is a standard assay for perturbations in DNA replication, and overinitiation of replication leads to an increase in the DNA to protein ratio. All of the mutant strains had a DNA to protein ratio 40–50% higher than that of wild-type cells, consistent with overinitiation of replication (Table 2).
Table 2. DNA to protein ratios in wild-type and mutant cells.
Indicated strains were inoculated from light lawns on minimal glucose plates into defined minimal glucose medium and grown at 30°C. Samples were collected during exponential growth at an OD600 of 0.4–0.6. Nucleic acid and protein were extracted and assayed for DNA and protein content as described in Experimental procedures.
Experimentally determined ratios from four experiments, normalized to the ratio from wild type done in parallel. Ratios are followed by the 95% confidence interval for the mean. The average ratio for wild-type cells was 0.030 ± 0.004.
1.4 ± 0.2
1.5 ± 0.2
1.4 ± 0.2
We propose that Soj and Spo0J affect initiation of replication by regulating accessibility of the origin to initiation factors. Formation of a nucleoprotein structure that compacts the origin region could exclude initiation factors from binding until a putative signal opens the structure. As a Δsoj mutation causes overinitiation of replication, but not a significant defect in chromosome partitioning, perhaps cellular factors such as SMC can compensate for Soj’s role in partitioning but not in replication initiation. Furthermore, the similar overreplication phenotype caused by spo0J mutations is not likely to be the cause of its partitioning defect.
We measured cell length in wild-type, soj and Δ(soj-spo0J) cells. The average cell length was ∼20% longer in Δ(soj-spo0J) cells, similar to Δspo0J mutants (Lee et al., 2003), although the Δsoj mutant alone had no detectable defect. Average cell lengths in wild-type, Δsoj and Δ(soj-spo0J) cells were 2.78 ± 0.10 µm, 2.82 ± 0.09 µm and 3.29 ± 0.14 µm respectively (results ± 95% confidence intervals, > 200 cells scored). The cell division delay of Δspo0J and Δ(soj-spo0J) cells cannot explain the increased DNA to protein ratio, because cell growth, DNA replication and segregation appear to proceed normally even under conditions where cell division is blocked (Dai and Lutkenhaus, 1991; Huls et al., 1999; Kawai et al., 2003). Although a Δsoj null mutation did not affect cell length on its own, the Δsoj Δsmc double mutant had significantly elongated cells compared with the Δsmc null mutant alone (data not shown). However, because the double mutant had many cells with long, unsegregated nucleoids, nucleoid occlusion would have inevitably led to longer cell lengths and it would be difficult to conclude whether there were additional effects on the division apparatus.
The proper subcellular positioning of Spo0J and Soj depends on cell division proteins like DivIB, FtsZ, PBP and MinD (Marston and Errington, 1999; Autret and Errington, 2003; Real et al., 2005). Perhaps the cell division delay could reflect a stimulatory effect of Spo0J on the division apparatus. Alternatively, defects in chromosome partitioning may elicit a corresponding delay in cell division through an unidentified regulatory mechanism.
Our observations that the Δ(soj-spo0J) double mutant has a higher DNA to protein ratio and elongated cells similar to a Δspo0J single mutant differ from a report that inactivating soj partially suppresses the overinitiation and cell division phenotypes of a spo0J null mutant (Ogura et al., 2003). From these results, the authors concluded that Soj may have an activity opposite to Spo0J. This discrepancy could reflect differences in the strain backgrounds or growth media, and warrants further investigation, because the results lead to different predictions about the function of Soj.
Inactivating soj, spo0J or both perturbs the number of foci per cell of the origin, 270° and terminus regions of the chromosome
Previously, it was shown that inactivating spo0J perturbs the number of foci per cell of several chromosomal regions: the origin-proximal 359° region, the 270° region, and the terminus-proximal 181° region (Lee et al., 2003). We visualized these regions in Δsoj and Δ(soj-spo0J) mutants using LacI–GFP bound to an array of lac operators inserted in the chromosome (Gordon et al., 1997; Webb et al., 1997). The majority of wild-type cells (80.9%) had two foci of the origin region, and 15.3% had more than two foci (Table 3) (Lee et al., 2003). Inactivating soj, spo0J or both led to a higher proportion of cells with more than two foci of the origin region (30.1%, 35.4% and 27.0% respectively) (Table 3). These results are consistent with overinitiation of replication, demonstrated in the previous section, which would increase the number of origins per cell. Surprisingly, the mutants also had a higher proportion of cells with one focus of the origin: 3.1% of wild-type cells had a single focus, compared with 6.3%, 15.3% and 8.7% in Δsoj, Δspo0J and Δ(soj-spo0J) mutants respectively (Table 3).
Table 3. Number of foci per cell of several chromosomal regions during exponential growth in wild-type and mutant cells.
An array of lac operators was inserted in the indicated region of the chromosome and visualized with LacI–GFP or LacI–CFP.
The percentage of cells with the indicated number of foci of LacI–GFP (or LacI–CFP) was determined for cells growing exponentially in defined minimal glucose medium at 30°C. Cells with no foci were mostly anucleates.
The total number of cells analysed for each strain.
Combined data from newly repeated and previously published results (Lee et al., 2003).
Average number of foci per cell (a) was calculated for each strain using the data showing percentage of cells with indicated number of foci, and the formula: a = 0 *(% cells with 0 foci) + 1 *(% cells with 1 focus) + 2 *(% cells with 2 foci) + 3 *(% cells with 3 foci) + 4 *(% cells with 4 foci) + 5 *(% cells with > 4 foci).
Average number of foci per cell, normalized to the wild-type strain for that chromosomal region (b) was calculated for each strain using the formula: b = awild type or mutant/awild type.
One caveat of using this method to count the number of chromosomal regions is that duplicated regions of the chromosome must be separated enough to be spatially resolved in order to be counted as two copies. If there were a defect in separating a particular chromosomal region, then this would lower the number of resolvable foci, leading to an underestimate of the copy number. The increased proportion of cells with one focus of the origin may be due to a defect in separating replicated sister origins. Because the mutants do not lead to an increased proportion of cells with a single focus of the 270° and 181° regions, the separation defect may be specific to the origin region. We present data supporting this hypothesis below.
To determine whether the net effect of the mutations was to increase or decrease the number of origins, we calculated the average number of origin foci per cell: the wild-type strain had 2.21 foci per cell on average, while the Δsoj, Δspo0J and Δ(soj-spo0J) mutants had 2.42, 2.31 and 2.30 foci per cell respectively (Table 3). Thus, the mutants had 4–9% more foci of the origin than wild-type cells (Table 3). This effect was more pronounced for the 270° and 181° regions, for which the mutants had 8–31% more foci of the origin than wild-type cells (Table 3). These data indicate that the mutants, on average, have increased chromosome copy number relative to wild-type cells. These data also indicate that the mutations increase the apparent copy numbers of the 270° and 181° to a greater extent than the copy number of the origin region in the mutant strains, consistent with a defect in separating replicated origins but not other regions of the chromosome (see below).
Effects of soj and spo0J on positioning separated sister origin regions
Previously, we showed that duplicated foci of the origin are closer together in a spo0J null. To determine whether Soj affects positioning of separated sister origins, we measured origin positioning in wild-type, Δsoj, Δspo0J and Δ(soj-spo0J) mutant cells with two spatially resolved foci of the origin. We found that duplicated foci of the origin were positioned normally at the cell quarters in the Δsoj mutant, but were closer together in Δ(soj-spo0J) cells, similar to the Δspo0J mutant (Lee et al., 2003) (Fig. 2, Table 4). Origins were visualized using LacI–GFP bound to an array of lac operators inserted at 359° (near oriC). Average distance from a focus to the nearest pole was 26.2 ± 1.3% of cell length in the Δsoj mutant, statistically indistinguishable from positioning at 26.1 ± 1.0% of cell length in wild-type cells (± 95% confidence interval for the mean) (Table 4). The Δ(soj-spo0J) mutant had foci positioned at 31.2 ± 1.8% of cell length, similar to 30.1 ± 1.5% in the Δspo0J mutant (Table 4).
Table 4. Subcellular positioning of replicated sister origins in cells with two foci of the origin region.a
Indicated strains were grown at 30°C in defined minimal glucose medium and samples were taken for microscopy during exponential growth. LacI–GFP was used to visualize lac operator arrays integrated at the 359° region of the chromosome.
The distance from each focus to the nearest cell pole was measured in cells with two foci, and is presented as an average percentage of cell length ± the 95% confidence interval for the mean.
The distance between each focus was measured in cells with two foci, and is presented as an average percentage of cell length ± the 95% confidence interval.
The total number of cells with two foci that were analysed.
In Δspo0J and Δ(soj-spo0J) cells, the foci of the origins were positioned closer together than in wild-type cells (Fig. 3, Table 4): the average interfocal distance was 39.7 ± 1.8% of cell length in Δspo0J (Lee et al., 2003) and 37.6 ± 1.9% of cell length in Δ(soj-spo0J), compared with 47.8 ± 1.2% in wild-type cells (Table 4). The average interfocal distance in the Δsoj mutant was 47.5 ± 1.3%, indistinguishable from wild-type cells (Table 4). Our results indicate that the detection of defects in positioning duplicated, spatially resolved origins, correlates with defects in chromosome partitioning. The effects of Δspo0J and Δ(soj-spo0J) mutants on positioning of duplicated foci of the origin could be due to a defect in separating replicated origins and/or a defect in maintaining the position of replicated origins at the cell quarters. Below, we present results indicating that the Δspo0J and Δ(soj-spo0J) mutants, and, to a lesser extent, the Δsoj mutant, have defects in separating replicated origins.
Effects of soj and spo0J on positioning a single focus of the origin region
The Δsoj, Δspo0J and Δ(soj-spo0J) mutations perturbed focus positioning in cells with a single visible focus of the origin region (Fig. 4). Cultures were grown with succinate instead of glucose as a carbon source, resulting in a slower doubling time and more cells with a single focus of the origin. The majority of wild-type cells had a focus positioned between 35% and 65% of cell length, and only 12.6% of cells had a focus outside this central region (Fig. 4). In the Δsoj, Δspo0J and Δ(soj-spo0J) mutants, 41.9%, 26.6% and 25.0% of cells had a focus outside this region respectively (Fig. 4). The mispositioning of the origin region in these mutants could be indicative of a role for Soj and Spo0J in positioning a single origin at or near midcell. Alternatively, some of the mutant cells that appear to have a single origin might actually contain two replicated sister origins that remained close together (paired) and moved away from midcell. The experiments in the next section were designed to distinguish between these possibilities.
Inactivating soj and/or spo0J impaired separation of replicated DNA in the origin region
To determine whether the mutant cells had replicated origins that were too close together to be spatially resolved, we visualized the origin and a nearby distal region simultaneously using two markers: LacI–CFP (cyan fluorescent protein) bound to an array of lac operators inserted at the 359° region of the chromosome, and TetR–YFP (yellow fluorescent protein) bound to an array of tet operators inserted at the 345° region of the chromosome (located ∼164 kb distal to the 359° region) (Fig. 5A). Cells that have two foci of the distal region must have initiated replication whether or not the replicated origins are spatially resolvable. In general, cells with fewer foci of the origin region than the distal region must contain replicated origins that have not separated.
We found that inactivating soj, spo0J or both caused a defect in origin separation: 3.2% of wild-type cells had fewer foci of the origin region than the distal region, compared with 13.7% in Δsoj, 20.3% in Δspo0J and 21.6% in Δ(soj-spo0J) mutants respectively (Table 5). In addition, each of the mutants had an increase in the proportion of cells with more foci of the origin region than the distal region (Table 5). Overinitiation of replication in these mutants could potentially increase the relative copy number of the origin region to the 345° region. Alternatively, there could also be a defect in separating the 345° region of the chromosome, which lies inside the range of the origin-proximal parS sites (Fig. 5A).
Table 5. Comparison of relative numbers of foci of an origin-proximal (359°) and origin-distal (345°) region of the chromosome.a
The number of foci of the origin-proximal marker P (359°) and origin-distal marker D (345°) was determined in individual cells and the percentage of cells with the indicated relationship is shown.
Indicated strains were grown at 30°C in defined minimal succinate medium and samples were taken during exponential growth for microscopic analysis.
Total number of cells scored.
Although Δsoj, Δspo0J and Δ(soj-spo0J) mutants have defects in separating replicated origins, most of the origins must separate eventually, because the proportion of cells with an origin separation defect far outweighs the proportion of anucleate cells produced. This delay in origin separation could be due to several possible primary defects, including: (1) a defect that slows migration of replicated origins, and/or (2) a defect in releasing origins from the constraints of putative factors that could pair them or anchor them in close proximity. The Δspo0J and double mutants had more severe origin separation defects than the Δsoj single mutant, correlated with measurable chromosome partitioning defects and elongated cells that were not observed in the Δsoj single mutant.
Inactivating soj and/or spo0J has little effect on separation of replicated DNA outside the origin region
To determine whether Δsoj, Δspo0J or Δ(soj-spo0J) mutations affected separation of chromosomal regions outside the origin region, we simultaneously visualized the 316° region, using LacI–CFP bound to an array of lac operators, and the 300° region (located ∼187 kb distal), using TetR–YFP bound to an array of tet operators (Fig. 5B). Both sites are located outside the origin-proximal region spanned by the known parS sites. The 316° region is replicated first. Cells that failed to separate the 316° region before the 300° region would have fewer foci of the 316° region than the 300° region.
Inactivating soj, spo0J or both did not substantially affect separation of the 316° region (Table 6). In wild-type cells, 1.4% had fewer foci of the 316° region than the 300° region, compared with 0.7% in Δsoj, 3.1% in Δspo0J and 2.8% in Δ(soj-spo0J) mutants (Table 6). The lack of an appreciable separation defect was not due to the fact that the 316° and 300° regions are slightly further apart than 359° and 345°, because a significant defect in origin separation was also seen when 359° and 329° were visualized simultaneously (data not shown). Taken together, our results demonstrate that soj and spo0J are required for efficient separation of replicated DNA in the origin region, and that these effects are specific to the origin region.
Table 6. Comparison of relative numbers of foci of chromosomal regions (316° and 300°) away from the origin.a
The number of foci of the origin-proximal marker P (316°) and origin-distal marker D (300°) was determined in individual cells and the percentage of cells with the indicated relationship is shown.
Indicated strains were grown at 30°C in defined minimal succinate medium and samples were taken during exponential growth for microscopic analysis.
Total number of cells scored.
Sister origins may be paired prior to separation
The finding that replicated origins, but not other regions of the chromosome, remain closely associated in the absence of soj and/or spo0J indicates that there may be specific barriers to separating sister origin regions. Replicated origins might be paired and/or anchored at nearby sites by an unidentified factor, and Soj and Spo0J function could overcome this barrier to separation. The processes or factors that might promote pairing are not known, but could be related to the SMC complex, similar to MukBEF that appear to promote chromosome pairing in E. coli (Sunako et al., 2001). Alternatively, it is formally possible that Spo0J and Soj pair the origins, as proposed for plasmid Par homologues, although the simple prediction from this model is that deletion of soj or spo0J would lead to increased separation of replicated origins, rather than the observed decrease.
To our knowledge, only one other factor has been identified so far that contributes specifically to separation of replicated origins. Function of the bacterial actin homologue MreB is required during a period early in the cell cycle for separation of replicated origins in C. crescentus (Gitai et al., 2005).
Soj and spo0J mutations cause pleiotropic phenotypes, but most of these effects may be explained by a single function
Inactivating soj and/or spo0J perturbs many cellular processes (Table 7). It seems unlikely that Soj and Spo0J would have distinct functions for each of these processes. We speculate that a single function may be sufficient to account for most of the phenotypes we examined. A common feature was that the Δ(soj-spo0J) double mutant followed the behaviour of the Δspo0J single mutant for phenotypes in chromosome partitioning (formation of anucleate cells), replication initiation (overinitiation), cell division (elongated cell lengths), origin separation and origin positioning. In other words, spo0J was epistatic to soj. Given the epistasis relationship, it is likely that if Soj affects these processes, it probably does so by exerting effects on Spo0J, i.e. Soj → Spo0J → phenotype (Fig. 6A). The soj null either did not have a phenotype or it was not as severe, with the exception of replication initiation, for which all of the mutants exhibited overinitiation to a similar extent. This could be due to partial compensation by another factor, such as SMC. We propose that a primary defect in separating replicated origins could lead to defects in origin positioning, delay cell division until origins separate, and produce anucleates in cases where the origins failed to separate. The Par homologues could plausibly affect these phenotypes by performing a single molecular function: Spo0J could bind parS sites and bring them together through higher-order interactions between Spo0J dimers, and Soj could facilitate these higher-order interactions. Assembly of Spo0J into a nucleoprotein complex that brings distant parS sites into close proximity could presumably organize the origin region into a compacted structure. Origin compaction could facilitate origin separation by preventing entanglement of sister origins, and/or by minimizing resistance against them as they move through the cytoplasm. Compaction of chromosomes prior to segregation is an essential step in eukaryotic chromosome partitioning (Nasmyth and Haering, 2005).
Table 7. Genetic interactions between soj and spo0J mutants for several phenotypes.
‘+’ indicates a phenotype similar to the soj+spo0J+ strain; ‘-’ indicates a defect in the indicated phenotype and ‘–‘indicates a more severe defect.
Epistasis refers to the genetic interaction between two genes, when one allele masks the phenotype of another (e.g. spo0J is epistatic for cell division: the Δ spo0J null mutation leads to a cell division delay whether or not soj is present).
Severity of phenotypes of the Δ soj Δ smc, Δ spo0J Δ smc and Δ (soj-spo0J) Δ smc mutants is considered relative to the Δ smc single mutant, which is normalized to ‘+’.
spo0J is epistatic
Position of sister origin regions
Chromosome partitioning in an Δsmc null backgroundc
All mutants similar
Controlled replication initiation
soj is epistatic
We also found that Soj and Spo0J affect replication initiation. In this case, all of the mutants exhibit overinitiation of replication to a similar degree, so it is not clear whether Soj acts through Spo0J or vice versa. It may be possible that Soj and Spo0J regulate replication initiation by performing some function that is specific to initiation. Alternatively, it is also possible (and simpler) that a primary effect on origin compaction could regulate accessibility of initiation factors, as discussed in the initiation section. Whether other chromosomal or plasmid-encoded Par homologues affect replication initiation remains a provocative question. Where tested, plasmid-encoded Par systems do not appear to affect plasmid copy number substantially, and plasmid loss in the absence of the par system is not due to insufficient replication (Austin et al., 1982; Ogura and Hiraga, 1983). However, the assays used would not have been sensitive enough to detect the degree of overinitiation we observed for Soj and Spo0J.
Soj and Spo0J perform at least one additional function: they are involved in transcriptional regulation during the initiation of sporulation (Table 7). Soj is a transcriptional repressor of early sporulation genes, and Spo0J relieves Soj-mediated repression by preventing Soj from binding to its target promoters (Quisel and Grossman, 2000). This epistasis interaction differs from the previously discussed phenotypes in that soj is epistatic: the spo0J null is sporulation-defective (Spo–) due to unopposed Soj activity, and a soj null is sporulation-proficient (Spo+), as is the double mutant. Spo0J affects initiation of sporulation via its effects on Soj, which is a different order of interactions from what we proposed for the phenotypes discussed earlier (Fig. 6B).
In addition to regulating Soj’s function as a transcriptional repressor, we postulate that nucleotide binding probably also affects Soj’s participation in the phenotypes examined here. Upon binding ATP, Soj from Thermus thermophilus forms a DNA-binding dimer that interacts with Spo0J and parS sites (Leonard et al., 2005). Perhaps ATP binding activates B. subtilis Soj to interact with Spo0J bound to parS sites. We speculate that energy generated from Spo0J-stimulated ATP hydrolysis could be coupled to remodelling of nearby DNA–protein complexes that hold sister origins together, enabling paired origins to separate.
Soj and Spo0J also contribute to origin positioning in the forespore, a compartment formed in sporulating cells when the division septum forms very close to one of the cell poles (Sharpe and Errington, 1996; Wu and Errington, 2002; Lee et al., 2003). A soj null enhances the sporulation defect of a null mutation in racA, the gene encoding a DNA-binding protein that contributes to origin positioning in the forespore (Wu and Errington, 2003; Ben-Yehuda et al., 2005). RacA and Soj/Spo0J both likely contribute to assembling the origin region into a highly condensed structure during sporulation. Thus, the Soj-Spo0J system appears to have partial redundancy with RacA. Ironically, it appears that Soj has both inhibitory and stimulatory effects on sporulation.
Effects of the Par system on plasmids and chromosomes
The Par system has different phenotypic effects on plasmid and chromosome positioning. Spo0J is not sufficient to recruit parS sites inserted at ectopic positions in the chromosome to the cell quarters (Lee et al., 2003), arguing against the model that Spo0J’s primary function is to facilitate attachment of replicated origins to putative anchors at the cell quarters. In contrast, plasmid-encoded Par systems appear to be sufficient to recruit parS plasmids to the cell quarters, consistent with the anchoring model (Niki and Hiraga, 1997; 1999; Erdmann et al., 1999). Chromosomally encoded Par systems can stabilize plasmids containing the cognate parS site (Lin and Grossman, 1998; Godfrin-Estevenon et al., 2002), and Soj and Spo0J can recruit parS plasmids to the cell quarters (Yamaichi and Niki, 2000). One interpretation of these results is that the same Par system can perform different functions on plasmids and chromosomes. However, a more likely interpretation is that the Par system performs the same biochemical function on plasmids and chromosomes, and that plasmid recruitment to the cell quarters is a secondary effect of a Par-dependent function other than anchoring. The difference in phenotypic effects could reflect innate differences between plasmids and chromosomes, such as size.
What similar partitioning function could the Par system perform on plasmids and chromosomes? In both cases, ParB could bind and bring together parS sites, forming a nucleoprotein structure that facilitates partitioning. Plasmid-encoded ParB appears to mediate plasmid pairing by bringing together parS sites on sister molecules (Funnell, 1988; Youngren and Austin, 1997; Edgar et al., 2001). Recent work provides structural insight into how a plasmid-encoded ParB homologue can bring different parS sites together (Schumacher and Funnell, 2005). Plasmid pairing could potentially facilitate co-ordinated plasmid separation in opposite directions. In the case of chromosomes, similar types of higher-order interactions might bring together parS sites that are located far apart on the same chromosome, organizing the origin region into a compacted structure that facilitates separation of replicated origins.
The Par system is widely conserved and contributes to faithful chromosome and plasmid partitioning in diverse bacterial species (Yamaichi, 2000). Par systems confer a significant survival advantage and affect the spread of plasmid-borne antibiotic resistance and virulence genes (Yamaichi, 2000; Youngren et al., 2000). Although Par systems have been intensely studied for over two decades, the molecular mechanism by which they affect segregation remains ill-defined. In addition, how Par homologues regulate other cellular processes like replication initiation, cell division and development, and whether other homologues have similar functions, remain the subject of intense investigation.
Media and growth conditions
For all experiments, cultures were grown at 30°C in S7 defined minimal medium with MOPS (morpholinepropanesulphonic acid) buffer at 50 mM rather than 100 mM supplemented with 0.1% glutamate, required amino acids, and either 1% glucose or 1% succinate (Vasantha and Freese, 1980; Jaacks et al., 1989). Cells were inoculated into liquid cultures from light lawns on Luria–Bertani (LB) or minimal plates. Strains were constructed and maintained on LB plates at 30°C, with the exception of strains containing Δsmc, ΔscpA and ΔscpB mutations (see below). Antibiotics were used at standard concentrations (Harwood and Cutting, 1990).
Δsmc,ΔscpA and ΔscpB strains. The chromosome partitioning defect of Δsmc, ΔscpA and ΔscpB mutants is more severe in LB than in minimal medium (Britton et al., 1998; Soppa et al., 2002). We found that inoculating cultures from light lawns on LB plates caused a more severe chromosome partitioning defect than inoculating cultures from light lawns on minimal plates, even after three generations of growth in minimal medium: 13.3% of cells were anucleate from cultures started from LB plates, similar to the previously reported partitioning defect of an smc null (Britton et al., 1998; Moriya et al., 1998), compared with 2–3% from minimal plates. For all experiments involving Δsmc, ΔscpA and ΔscpB strains, we used Spizizen’s minimal plates (Harwood and Cutting, 1990) supplemented with 1% glucose, 0.1% glutamate and required amino acids for preparing light lawns, strain construction and routine maintenance. 1 mM IPTG was added to induce expression of downstream genes in ΔscpA and ΔscpB strains (Soppa et al., 2002).
Contains Ppen–tetR–yfp flanked by upstream and downstream regions of cgeD (181°), used to make pPSL38
To integrate Ppen*(TATG→TAGG)–tetR–yfp at cgeD (181°)
Used to create non-polar Δsoj132 mutation in the chromosome
Non-polar soj null mutation. pIK208 contains an in frame non-polar soj deletion (Δsoj132) removing codons 6–235 of 253, inclusive, from the soj coding region, followed by the full-length spo0J+ gene (Ireton et al., 1994). A HindIII fragment was removed from pIK208, removing the C-terminal 278 nucleotides from spo0J, creating pRV11. pRV11 was integrated into the B. subtilis chromosome by single crossover, selecting for chloramphenicol-resistant sporulation-defective transformants. SV132, a spontaneous derivative that retained the non-polar soj mutation but that had lost the plasmid-derived sequences by homologous recombination, was obtained by isolating sporulation-competent, chloramphenicol-sensitive derivatives. The Δsoj mutation was confirmed by polymerase chain reaction (PCR) and restriction digests. Western blotting confirmed that Spo0J protein was expressed at a level indistinguishable from wild type (data not shown).
ScpA and scpB null mutants. scpA or scpB were disrupted while permitting expression of downstream genes from a Pspac promoter as described (Soppa et al., 2002). The ΔscpA and ΔscpB mutations were introduced into the JH642 background using chromosomal DNA from NIS6031 and NIS6042 (Lindow et al., 2002b), selecting for erythromycin–lincomycin resistance and sensitivity to spectinomycin. This created PSL640 and PSL642 respectively.
Lac operator system. LacI–GFP and LacI–CFP were used to visualize regions of the chromosome marked with arrays of lac operators. The lacI–gfp and lacI–cfp fusions have been described extensively (Lemon and Grossman, 2000; Lemon et al., 2001).
The lacO cassette was inserted into several regions of the chromosome (181°, 270°, 316° and 359°) by homologous recombination. The lacO cassettes at the 181° and 270° regions, inserted in nonessential genes cgeD and cotS, were introduced using chromosomal DNA from AT54 and AT52 respectively (Teleman et al., 1998). The lacO cassette at the 359° region was introduced using chromosomal DNA from a strain that had been transformed with the plasmid pDL175 (Lee et al., 2003). To insert the lacO cassette at the 316° region of the chromosome, oligonucleotides LEE-106 and LEE-82 were used to amplify a 3′ fragment extending past the stop codon of alsD, so that integration into the chromosome would not disrupt the coding region. The PCR product was digested with BamHI and KpnI and subcloned into the lacO cassette containing plasmid pAT12 (Teleman et al., 1998), creating pPSL27. The lacO cassettes were amplified by selecting for resistance to chloramphenicol (25 µg ml−1) as described previously (Webb et al., 1997; Teleman et al., 1998).
Tet operator system. The tetO/TetR–YFP system was used in conjunction with the lacO/LacI–CFP system so that two different regions of the chromosome could be visualized simultaneously. pPSL38 contains the tetR–yfp fusion under control of Ppen*, a mutagenized version of the Ppen promoter that contains a TATG→TAGG mutation in the extended −10 region, flanked by upstream and downstream regions of the cgeD gene for integration into the chromosome by double crossover. To construct this vector, tetR–gfp from pBW3 (Dworkin and Losick, 2002) was converted to tetR–yfp using site-directed mutagenesis to convert threonine 203 to tyrosine (Ormo et al., 1996). pPSL36 contains tetR–yfp under control of the Ppen promoter, flanked by upstream and downstream regions of cgeD. Mutagenic oligopeptides with the sequence TANN in the extended −10 region (N = random nucleotide) were used to mutagenize pPSL36 to decrease expression of TetR–YFP. One of the mutagenized products, named pPSL38, was selected that had a bright signal and low background in the presence of tet operators inserted at the 359° region of the chromosome, but did not detectably perturb cellular physiology. Sequencing revealed a TATG→TAGG mutation in the extended −10 region.
To visualize the 300° or 345° regions of the chromosome, a tetO cassette was inserted there by homologous recombination. pLAU44 contains an ∼9 kb array of 240 tetO2 sites with randomized sequences between the sites to minimize recombination (Lau et al., 2003). The tetO2 cassette was subcloned NheI/SalI into the kanamycin-resistant vector pDG792 (Guerout-Fleury et al., 1995), to create pPSL6. tetO integration vectors were designed to integrate downstream of chromosomal genes without disrupting them. To insert the tetO cassette at the 300° region of the chromosome, oligonucleotides LEE-85 and LEE-87 were used to amplify a 3′ fragment extending past the stop codon of the yvfI gene. The PCR product was digested with PstI and KpnI and subcloned into pPSL6, creating pPSL18. pPSL18 was transformed into cells of the wild-type strain AG174, and chromosomal DNA from this strain was used to introduce the tetO cassette at the 300° region. To insert the tetO cassette at the 345° region of the chromosome, oligonucleotides LEE-58 and LEE-59 were used to amplify a 3′ fragment extending past the stop codon of the hutM gene. The PCR product was digested with PstI and KpnI and subcloned into pPSL6, creating pPSL10. pPSL10 was transformed into cells of the wild-type strain AG174, and chromosomal DNA from this strain was used to introduce the tetO cassette at the 345° region.
Measurement of DNA to protein ratios
Cultures were inoculated from light lawns on Spizizen minimal 1% glucose plates with required amino acids, and grown at 30°C in defined minimal glucose medium. Approximately five ODs of cells in mid-exponential growth (OD600 = 0.4–0.6) were collected by centrifugation. Nucleic acid and protein fractions were extracted (Kadoya et al., 2002). The cell pellet was resuspended in 0.5 ml of ice-cold 0.25 N perchloric acid and incubated for 30 min on ice. Insoluble materials were collected by centrifugation and the supernatant was discarded. To extract nucleic acids, the pellet was resuspended in 0.2 ml of 0.5 N perchloric acid and incubated at 70°C for 15 min. The soluble fraction was collected and the extraction step was repeated. To extract proteins, the remaining pellet was resuspended in 0.25 ml of 1 M NaOH and incubated at 95°C for 10 min, and the soluble fraction was collected. DNA and protein concentrations were quantified using the diphenylamine reaction and the Lowry Bio-Rad DC Protein Assay Kit, respectively, with appropriate standards.
Live cells were stained with 4′,6′-diamidino-2-phenylindole (DAPI) (40–80 ng ml−1) to visualize nucleoids, and the vital dye FM4-64 (10–200 ng ml−1; Molecular Probes) to visualize cell membranes. The lower concentration of FM4-64 was used for strains containing tetR–yfp fusions to minimize the signal from the dye under the YFP filter. Microscopy was performed as described previously (Lemon and Grossman, 1998; 2000). Briefly, cells were immobilized on pads of 1% agarose in 1× T’base-1 mM MgSO4 (Harwood and Cutting, 1990), and images were captured with a Nikon E800 microscope equipped with a Hamamatsu digital camera. The filters used were: tetramethylrhodamine isothiocyanate for FM4-64, chroma filter set 31044 for CFP, 41012 for GFP and 41029 for YFP. Improvision OpenLabs software was used to process images.
Measurement of focus position in cells with two foci of the origin region
Focus position in cells with two foci was measured as described previously, using a strategy designed to eliminate any unintentional visual bias in scoring (Lee et al., 2003). Briefly, for each cell, the focus closest to a pole was designated as focus A, and the other focus was designated as focus B. Three measurements were made: (1) a, the distance from focus A to the closest pole, (2) b, the distance from that same pole to focus B, and (3) l, the cell length. This created a systematic bias in scoring such that in every cell, focus A was closer to a pole than focus B. In order to remove this bias, a random number generator was used to assign each focus an approximately 50% chance of being counted as closest to a pole. The corresponding distance of each focus from the same cell pole (the a and b measurements) was then recalculated. To construct graphs of focus positions in cells with two foci, the a and b measurements were plotted against cell length for each cell scored.
Focus position as a percentage of cell length was determined by dividing the distance of each focus from the nearest cell pole (a for focus A, l minus b for focus B) by the cell length and multiplying by 100. These numbers were averaged for all cells scored, and were reported followed by the 95% confidence interval for the mean. Interfocal distance as a proportion of cell length was calculated as (b − a) divided by l.
Measurement of focus position in cells with one focus of the origin region
For each cell, two measurements were made: (1) a, the distance from a focus to the nearest pole, and (2) l, the cell length. This created a systematic bias in scoring such that in every cell, the distance a was smaller than half of the cell length. In order to remove this bias, a random number generator was used to assign each focus an ∼50% chance of remaining as scored, and an ∼50% chance of recalculating a as the distance from the focus to the far pole. To construct graphs of focus positions in cells with one focus, the recalculated a measurements were plotted against cell length for each cell scored. Also, focus position as a percentage of cell length (a/l) was calculated for each cell.
We thank Shivkumar Venkatasubramanyam for constructing the non-polar Δsoj null mutation. We gratefully acknowledge Melanie Berkmen, Cathy Lee and Adam Breier for providing invaluable comments on the manuscript. The tetO cassette and tetR–gfp constructs were generous gifts from David Sherratt and Jonathan Dworkin. This work was supported in part by NIH Grant GM41934 to ADG.