RacA and the Soj-Spo0J system combine to effect polar chromosome segregation in sporulating Bacillus subtilis



Sporulating cells of Bacillus subtilis undergo a highly polarized cell division and possess a specialized mechanism to move the oriC region of the chromosome close to the cell pole before septation. DivIVA protein, which localizes to the cell pole, and the Soj and Spo0J proteins, which associate with the chromosome, are part of the mechanism that delivers the chromosome to the cell pole. A sporulation-specific protein, RacA, encodes a third DNA-binding protein, which acts in conjunction with Soj and Spo0J to effect efficient polar chromosome segregation. divIVA mutants and soj racA double mutants have an unexpected phenotype in which specific markers to the left and right of oriC can be captured in the prespore compartment but the central oriC region is efficiently excluded. This ‘residual’ trapping requires Spo0J protein. We suggest that the Soj RacA DivIVA system is required to extract the oriC region from its position determined by the vegetative chromosome segregation machinery and anchor it to the cell pole.


The mechanisms of chromosome segregation in bacterial cells remain poorly understood, although it is now clear that this is a highly organized process (reviewed by Gordon and Wright, 2000; Hiraga, 2000). Sporulating cells of Bacillus subtilis offer a particularly tractable system for studying segregation in bacteria (reviewed by Errington, 2001). Early in sporulation the chromosome destined to enter the spore needs to move to the extreme pole of the cell in order to be captured in the prespore cell, which is formed by asymmetric cell division very close to the pole of the pre-divisional cell. Early electron microscopic studies revealed that this movement is accompanied by a change in the conformation of the nucleoid from its normal compact shape to an extended form called the axial filament (Ryter, 1965; Bylund et al., 1993; Hauser and Errington, 1995; Pogliano et al., 2002). Several proteins involved in prespore chromosome segregation have been identified by mutations that perturb sporulation. Mutations in the spoIIIE gene revealed that the prespore chromosome is initially bisected by the polar septum, leaving the major part of the chromosome incorrectly located in the large (mother cell) compartment (Wu and Errington, 1994). The spoIIIE gene encodes a protein that forms a pore in the sporulation septum through which it transports the mislocated segment of chromosome (Wu et al., 1995; Wu and Errington, 1997; Bath et al., 2000). Studies of spoIIIE mutants also revealed that the segment of DNA initially trapped in the prespore compartment is a specific one, centred roughly on oriC, the origin of bidirectional chromosome replication (Wu and Errington, 1994; Wu and Errington, 1998). This inferred the existence of a mechanism responsible for determining the orientation and positioning of the chromosome early in sporulation.

The soj-spo0J system of B. subtilis was the first locus to be implicated in this process. Ireton et al. (1994) had previously shown that mutations in spo0J affected chromosome segregation in vegetative cells of B. subtilis. This was in accordance with the similarity of spo0J and an immediately adjacent upstream gene soj, to a family of proteins involved in segregational stability of plasmids from a wide range of bacteria (reviewed by Gerdes et al., 2000). Ireton et al. (1994) also showed that spo0J controls gene expression early in sporulation by counteracting the negative effects of Soj protein, which is a repressor of transcription of several key early sporulation genes (Cervin et al., 1998; Quisel and Grossman, 2000). Spo0J is a DNA-binding protein with about eight preferred binding sites scattered around the oriC region of the B. subtilis chromosome (Lin and Grossman, 1998). In the presence of Soj protein, Spo0J molecules are condensed to form discrete foci, tightly associated with the oriC region of the chromosome (Glaser et al., 1997; Lewis and Errington, 1997; Lin et al., 1997; Marston and Errington, 1999; Lee et al., 2003). During sporulation, these foci migrate to the extreme poles of the cell, in preparation for capture of the oriC region by the prespore septum (Glaser et al., 1997; Graumann and Losick, 2001). soj mutants apparently have little effect on sporulation, other than the transcriptional block that is dependent on spo0J (Ireton et al., 1994). However, Soj does have a remarkable dynamic behaviour in which it can jump from nucleoid to nucleoid, via the cell pole (Marston and Errington, 1999; Quisel et al., 1999; Autret and Errington, 2003). This movement is apparently required for proper condensation of spo0J foci (though not for chromosome segregation) (Ireton et al., 1994; Marston and Errington, 1999). Also, Spo0J somehow controls the jumping behaviour of Soj, which may underlie its ability to counteract the transcriptional repression (Marston and Errington, 1999; Quisel et al., 1999; Autret and Errington, 2003).

Recently, DivIVA was identified as another protein involved in prespore chromosome capture (Thomaides et al., 2001). DivIVA is a predominantly coiled–coil protein found throughout the Gram-positive bacteria (Edwards et al., 2000). It is targeted to the cell poles of B. subtilis by mechanisms that are not yet clear (Edwards and Errington, 1997; Hamoen and Errington, 2003; Harry and Lewis, 2003). In vegetative cells, DivIVA is responsible for recruiting the MinCD complex to the poles, and this complex acts to prevent aberrant polar division and the formation of minicells (Cha and Stewart, 1997; Edwards and Errington, 1997; Marston et al., 1998). In sporulation, DivIVA has a second, quite separate function, in recruiting the oriC region of the chromosome to the cell poles. A point mutation in divIVA (divIVA13) was isolated that allows normal targeting of DivIVA protein to the poles, and normal control of MinCD, but sporulation is reduced to about 5% of the wild-type value because the axial filament forms incorrectly and most cells capture little or no DNA in their prespore compartment (Thomaides et al., 2001). DivIVA protein does not appear to have the capability of binding to DNA, so it seemed likely that it would interact with at least one DNA-binding protein, possibly including Soj and or Spo0J, to effect polar chromosome migration.

Although soj spo0J mutants are deficient in prespore chromosome segregation, most cells capture about the normal complement of DNA. However, the precision with which the oriC region is captured is less than for the wild type (Sharpe and Errington, 1996; Wu and Errington, 2002; Lee et al., 2003). To shed light on the nature of the mechanism underlying the soj-spo0J-independent chromosome capture, we recently examined the effects of defined chromosome rearrangements on the segment of DNA captured in a soj-spo0J background. These experiments identified a region of chromosome, called the PLR (polar localization region), which can determine the orientation of the chromosome in the absence of soj-spo0J. The region appeared to lie about 250 kbp to the left of oriC, and sequences within about 150 kbp of oriC were not required for this activity (Wu and Errington, 2002). This strongly suggested the existence of at least one more DNA-binding protein involved in orientation of the chromosome. Here we describe the identification and characterization of such a protein, RacA (which was independently identified and characterized by Losick's group: Ben-Yehuda et al., 2003). We show that this protein is expressed specifically in early sporulating cells, that it probably binds directly to DNA and that it is targeted to the cell poles in a DivIVA-dependent manner. RacA disruption produces a twofold decrease in trapping of DNA in the prespore compartment, but the overall orientation of the chromosome is unchanged because RacA function is partly redundant with that of Soj protein. The absence of RacA and Soj leads to a DivIVA-like defect in prespore chromosome segregation, showing that these are probably the only major DNA-binding proteins responsible for chromosome recruitment to the pole by DivIVA. Finally, we show that in the absence of RacA Soj and Spo0J, the oriC-specificity of orientation of the prespore chromosome is effectively eliminated, defining these proteins, together with DivIVA, as the key players in prespore chromosome segregation in B. subtilis.


A spo0H-dependent gene required for efficient prespore chromosome segregation

Previous studies in this laboratory had suggested that formation of the axial filament is impaired by mutations in the spo0H gene (M. E. Sharp and J.E., unpubl. result). This gene encodes an important regulator of the early stages of sporulation, σH (Piggot and Losick, 2001). Microarray experiments (V. Vagner and J.E., unpubl.) revealed some candidates for a σH-dependent gene that might be involved in PLR function. ywkC was one of the strongest induced genes dependent on σH, was located in the vicinity of the PLR (− 416 relative to oriC), and encoded a protein with a likely helix-turn-helix DNA-binding motif close to its N-terminus (residues 3–22, based on the algorithm of Dodd and Egan, 1987). The protein was significantly conserved only in related spore-forming bacteria (e.g. B. halodurans and Clostridium acetobutylicum). (Ben-Yehuda et al., 2003; have independently identified and characterized the ywkC gene and named it racA, so we have followed this nomenclature.) To test the possible role of RacA in PLR function, the racA gene was disrupted. Preliminary examination of the mutant suggested that it had a weak sporulation defect (about twofold reduction in spores). Microscopic examination of a population of cells induced to sporulate confirmed that many of the cells were able to proceed more or less normally through to spore maturation. However, a significant proportion of the cells did not form spores and most of these cells appeared to be defective in prespore chromosome segregation because they produced prespore-like compartments containing little or no DNA (Fig. 1). Table 1 shows a comparison of the morphological phenotypes of wild-type and racA mutants 98 min after the induction of sporulation. At this time, just less than half of the wild-type cells had formed an asymmetric sporulation septum, in accordance with the normal timing of events (Partridge and Errington, 1993) (Fig. 1A–C). A tiny proportion (2%) of the sporulating cells had formed a second polar septum (to give an aberrant ‘disporic’ cell), but no cells were detected with prespores that were significantly deficient in DNA. In the racA mutant culture (Fig. 1D–F), a similar proportion of cells had formed an asymmetric septum, but about half of these polar compartments appeared to be devoid of DNA (e.g. cell labelled ‘1’ in Fig. 1F). About 14% of the sporulating racA cells (i.e. cells with at least one polar septum; classes III, IV and V) had bipolar septa, and in almost all of these, DNA appeared to be absent from at least one polar compartment. Examples of ‘disporic’ cells with (cell ‘2’) or without (‘3’) DNA in one of their prespore compartments are shown in Fig. 1F. About 20 min later, the number of disporic cells had increased, apparently at the expense of cells that had failed to capture DNA in the first polar compartment (Table 1). Half of the disporic cells had captured DNA in at least one of their polar compartments, and many of these cells were presumably able to go on to sporulate successfully. It thus appeared that the racA product plays an important role in prespore chromosome capture, though in its absence, a large proportion of cells are still able to achieve segregation of their prespore chromosome.

Figure 1.

Effects of racA disruption on sporulation. Wild-type (168; A–C) and racA mutant (1273; D–F) cells were induced to sporulate and samples taken at t110 were stained and examined by fluorescence microscopy. DNA (A, D), membrane (B, E) and merged (C, F) images of the same fields of cells are shown. Examples of mutant cells exhibiting a single empty prespore (1), or disporic cells with one (2) or both (3) polar compartments are numbered in panels E and F. Scale bar 5 µm.

Table 1. Effects of racA disruption on sporulation.
StrainTimeNo. of sporangioles(% total cellsexamined)Sporangiole morphology (%)
I inline imageII inline imageIII inline imageIV inline imageV inline image
Wild type (168)t98315 (34%)98 0 1 01
ΔracA (1273)t98337 (34%)4837 8 51
ΔracA (1273)t115447 (41%)462713112

racA encodes a sporulation-specific DNA-binding protein that targets to the prespore poles in a DivIVA-dependent manner

To examine the localization of RacA protein and, in particular, test whether it was associated with DNA, GFP fusions were constructed at both the N- and C-termini of the protein. Neither protein gave a completely normal phenotype when present as the only copy of the protein in the cell. However, both fusions gave potentially useful information on localization. A GFP fusion to the C-terminus of RacA (expressed specifically in sporulating cells; see below) showed a localization pattern very reminiscent of the nucleoid (Fig. 2A). Indeed, co-staining of GFP and DNA supported a general DNA-associated state for the protein (Fig. 2B). However, there were also some concentrations of GFP into foci. There was generally one focus per chromosome, and these foci tended to be located at the outer edges of the nucleoid (arrows in Fig. 2A).

Figure 2.

Localization and time of synthesis of RacA protein.
A–F. Strains containing a GFP fusion to the C-terminus (168::pSG4915) (A, B) or N-terminus (pSG4916) (C–F) of RacA were grown vegetatively (E, F) or induced to sporulate and sampled at t155 (A, B), t120 (C) or t170 (D). All of the panels show GFP fluorescence except B which is DAPI (DNA) fluorescence. The strain in panel F contains mutations in minD and divIVA. All cultures were grown at 30°C.
G, H. Western blot of RacA-GFP (upper panels) and Spo0J (lower panels) of strain 168::pSG4915 (racA-gfp) during sporulation (G; times of samples given above the lanes) and the effects of mutations in various sporulation mutants (H) at t90: wild type (1), spo0A (2), spo0H (3), spoIIAA (4).

The N-terminal GFP fusion showed a quite different pattern in which the most prominent fluorescence was associated with division sites and the cell poles (Fig. 2C–E). This staining pattern was reminiscent of that of DivIVA protein (Edwards and Errington, 1997) and indeed, when the fusion was introduced into cells with a disruption of divIVA, or bearing the divIVA13 mutation that specifically impairs prespore-chromosome segregation (Thomaides et al., 2001), the polar and septal localization was lost (Fig. 2F). These results were consistent with RacA having at least two targeting functions: a DNA-binding activity, disrupted by fusion of GFP to the N-terminus of the protein (note that the putative helix-turn-helix motif is very close to the N-terminal of the protein); and a DivIVA-targeting activity, probably lost with the C-terminal fusion.

Western blot analysis showed that the racA gene is turned on specifically during sporulation, beginning after about 1 h, just before the first sporulation septa begin to be formed (Fig. 2G). As expected, RacA synthesis was blocked by a spo0H mutation eliminating σH activity (lane 2). It was also blocked by spo0A mutation (lane 3) but not by mutation acting later in sporulation, spoIIA (lane 4). As a control, the Spo0J protein was detected throughout vegetative growth and sporulation, and its accumulation was not affected significantly by any of the sporulation mutations (Fig. 2G and H; lower panels).

racA single mutants are not significantly affected in prespore chromosome orientation

A prespore marker trapping assay was used to characterize the segment of DNA trapped in the prespores of racA mutants, as described previously (Wu and Errington, 1994; Wu and Errington, 1998; Wu and Errington, 2002). The assay depends on use of a spoIIIE mutation that prevents movement of the chromosomal DNA once the prespore septum has been formed. The mutation also allows correct compartmentalized activation of the sF transcription factor in the prespore compartment, allowing a reporter gene dependent on sF activation to indicate whether or not the reporter has been trapped in the prespore. In otherwise wild-type cells, reporters placed anywhere in the region from about −450 to +450, relative to oriC, are efficiently expressed but those at ± 600 (or beyond) are not (Fig. 3A). The pattern of reporter trapping of the racA mutant was almost indistinguishable from that of the wild type (i.e. the spoIIIE36 single mutant) (Fig. 3B). This indicates that even though about half of the sporulating cells fail to capture a significant amount of prespore DNA, the ones that do, capture essentially the same segment of DNA as in wild-type cells. Thus, the overall orientation of the chromosome as the cells enter sporulation is probably similar in racA mutants to that of wild-type cells, with the oriC region closest to the cell pole.

Figure 3.

Reporter trapping assay characterizing the segments of DNA trapped at the cell pole in various mutants. All strains are in the trpC2 spoIIIE36 background and contain a gpr-lacZ reporter gene at the chromosomal location shown at the top. Grey ovals indicate Spo0J-binding sites (Lin and Grossman, 1998). A: wild type; B: racA; C: divIVA13; D: racA soj; E: racA soj spo0J; F: soj spo0J; G: soj.

Spo0J/oriC complexes are excluded from the prespore in the divIVA13 mutant

Because divIVA was the likely target for RacA association with the cell pole, and divIVA mutants also have a DNA-trapping defect (Thomaides et al., 2001), we looked to see whether the pattern of residual DNA trapping in the divIVA13 mutant would also reflect an oriC orientation towards the cell pole. As expected, given that divIVA13 mutants are more severely affected in prespore chromosome segregation than racA (Table 2; and compare Fig. 1 with Fig. 4B), the divIVA13 mutant showed a virtually complete loss of DNA trapping across most of the oriC region (Fig. 3C). Surprisingly, however, two bands of relatively efficient trapping were detected about 400 kbp to the left and the right of oriC. Quantification of the β-galactosidase activities in strains induced to sporulate in liquid medium revealed that gpr–lacZ fusions at −335 and +327 were expressed to about 30–40% of the corresponding divIVA+ strains 225 min after the onset of sporulation (data not shown). Careful examination of images of the prespore compartments of the divIVA13 mutant cells revealed that many of them were not completely devoid of DNA. If so the trapping results suggested that these small amounts of trapped DNA might be enriched for sequences around +400 and −400. To test this we examined the synthesis and localization of GFP expressed from the prespore-specific spoIIQ promoter (at −455). As shown in Fig. 4A, GFP was detected in many of the sporulating cells and it was invariably confined to prespore compartments (Fig. 4B; arrows). Most of these prespore compartments contained only barely detectable amounts of DNA. This confirmed that the small amounts of DNA that are sometimes trapped in the prespore compartments of divIVA mutants frequently contain DNA from around −400 on the chromosome. On the basis of the pattern of trapping in Fig. 3C, we assume that many of the prespores will contain DNA from the other side of oriC, though this was not tested directly.

Table 2. Effects of prespore chromosome segregation mutants on sporulation frequency.
StrainRelevantgenotypeSporulation frequency(% wild type)a
  • a . >500 cells from sporulating cultures taken at t 7 were examined for refractile spores by phase contrast microscopy. Frequencies are expressed as percentage of the wild-type value.

  • b

    . Overestimation because significant cell lysis occurred in the mutant population before the time of counting.

 168wild type100
1273 racA  53
1276 soj  90
1275 soj spo0J 120
1274 divIVA   6
1278 soj racA  13b
1277 racA soj spo0J   8b
Figure 4.

Effect of the divIVA13 mutation on prespore chromosome trapping and Spo0J localization.
A, B. Strain 1280, containing divIVA13 spoIIIE36 mutations and a σF-dependent gfp fusion placed at −455 (PspoIIQ-gfp) was induced to sporulate and at t120 was examined by fluorescence microscopy. A typical field of cells is shown simultaneously visualized for GFP (green) and DNA (DAPI, red) (A) or for DNA only (B). Arrows point to prespores containing very little DNA and expressing the fusion.
C, D. Strain 1281 expressing a Spo0J-GFP fusion was induced to sporulate as for A, B. The field of cells shown were imaged for GFP (C) and for the membrane dye FM5-95 (D). Note that in panel C there is slight bleed through of the membrane dye signal into the GFP channel. Arrows point to pairs of foci showing a vegetative-like separation in cells with clear polar septa.
E. Control illustrating the extreme polar position of Spo0J-GFP foci in divIVA+ cells. Pairs of arrows highlight the extreme separation of pairs of foci towards opposite cell poles.

We considered two possible explanations for the residual trapping of sequences to the left and right of oriC in the divIVA13 mutant. First, there could be an as yet unknown polar targeting system acting to bring specific DNA regions to the left and right of oriC to the cell pole independent of divIVA. Alternatively, loss of the DivIVA polar anchor could result in a rearrangement of the chromosome that happens to leave these particular parts of the chromosome closer to the cell pole than the rest of the oriC region, so that they are more likely to be trapped by the septum. To examine the localization of the oriC region in the divIVA mutant we used a GFP fusion to the Spo0J protein, which associates with several preferred DNA sites around oriC (Lin and Grossman, 1998). As shown previously, Spo0J-GFP forms condensed foci that migrate to highly polar positions in sporulating cells of the wild type (Fig. 4E; paired arrows) (Glaser et al., 1997; Lin and Grossman, 1998; Graumann and Losick, 2001). Consistent with the loss of oriC region trapping in the divIVA13 mutant, prespore compartments of this mutant almost never contained Spo0J-GFP foci (Fig. 4C). Although the cells tended to contain the expected pair of Spo0J foci (one for each chromosomal oriC copy), these foci were usually located far from the cell poles, at positions more reminiscent of the foci in vegetative cells (Fig. 4C, pairs of arrows). This result was consistent with the absence of oriC region trapping revealed by the trapping assay depicted in Fig. 3, and suggested that in the divIVA13 mutant, Spo0J foci mainly fail to be recruited to the cell poles and instead remain much closer to mid-cell, possibly at their ‘default’ vegetative positions (see below).

Soj and RacA have partially redundant functions in prespore capture of the oriC region

A similar Spo0J-GFP experiment with the racA mutant revealed a mixture of prespores containing DNA that usually contained a Spo0J focus, and ones with little or no DNA and no Spo0J focus (data not shown; and Ben-Yehuda et al., 2003). Assuming that Spo0J elicits a condensed organization on the oriC region (Glaser et al., 1997; Lin et al., 1997; Lin and Grossman, 1998), the RacA phenotype could be due to a reduction in the efficiency of capture of the Spo0J-oriC complex. This reduction is clearly much less than for the divIVA mutant, suggesting the existence of another protein(s) that acts in parallel with RacA to bring the Spo0J-oriC complex to the cell pole.

Soj protein is encoded by the gene immediately unpstream of spo0J, near oriC, and is closely related to the parA family of genes required for stable maintenance of many low-copy-number plasmids (Gerdes et al., 2000). soj mutants are deficient in the condensation of Spo0J-oriC foci (Marston and Errington, 1999) but they have no appreciable chromosome segregation phenotype, nor do they have a significantly reduced sporulation frequency (Ireton et al., 1994; Sharpe and Errington, 1996). Soj is a dynamic protein that can cooperatively jump from nucleoid to nucleoid and we recently showed that jumping onto a nucleoid is strongly promoted by the vicinity of a cell pole, probably through an interaction with the polar MinD protein (Marston and Errington, 1999; Autret and Errington, 2003). This possible link between Soj function and the cell pole (Autret and Errington, 2003) suggested this protein as a candidate for the factor acting together with RacA to bring the oriC region to the cell pole. We confirmed that a soj single mutation had no significant effect on sporulation frequency (Table 2), nor did it significantly affect the prespore chromosome trapping pattern (Fig. 3G). However, when soj was combined with racA, a substantial sporulation defect emerged, much greater than for either single mutant (Table 2). The reduction in sporulation frequency was also associated with extensive cell lysis after 4 or 5 h of sporulation, consistent with many cells failing to correctly segregate their prespore chromosome. The trapping assay was used to determine what effect the double mutation had on chromosome orientation. Strikingly, the pattern resembled that of divIVA, with reporters located around oriC being essentially silent but with significant expression occurring at around ± 400 (Fig. 3D). Neither racA nor soj had a significant effect on trapping in the presence of the divIVA13 mutation – the double mutants were indistinguishable from the divIVA13 single mutant (data not shown). These results showed that Soj and RacA have partially redundant roles required to bring the oriC region to DivIVA at the cell pole. They also add strength to the notion that Soj has a role in chromosome dynamics (Lin and Grossman, 1998).

Relaxed chromosome orientation in the combined absence of racA and the soj spo0J system

According to the above interpretation, in divIVA mutants, or soj racA double mutants, Spo0J molecules bind around the oriC region but these complexes are not recruited efficiently to the cell pole. Indeed the complexes remain localized well inside the mother cell compartment in divIVA mutants (Fig. 4C). Nevertheless, Spo0J, perhaps through its role in facilitating chromosome segregation in vegetative growth (which is independent of Soj) (Ireton et al., 1994) leaves the chromosome in an orientation in which markers about 400 kbp to the left and right of oriC are closest to the cell poles and these markers are then the ones most likely to be trapped in the prespore compartment. If this interpretation was correct, combining a spo0J mutation with that of soj and racA or divIVA, should result in the loss of oriC organization and positioning, as well as of RacA binding to the cell pole. As shown in Table 2, the racA soj spo0J triple mutant had a further reduction in sporulation frequency, compared with the various single and double mutants. Furthermore, the specificity of trapping at about ± 400 exhibited by the racA soj mutant (Fig. 3E) was not evident and no reporter showed significant activity at any of the sites tested. A similar loss of trapping at ± 400 was obtained when soj spo0J disruption was combined with the divIVA mutant (data not shown).

Interestingly, microscopic examination of DAPI-stained cells of the racA soj spo0J triple mutant revealed that about half of the prespores captured significant amounts of DNA, similar to the racA single mutant (Table 1; Fig. 5B and D; arrows and arrowheads). The capture of some DNA but failure to exhibit significant activity with the DNA-trapping assay (Fig. 3H) could be due to a non-specific reduction in gene expression or to lack of specificity in the DNA segments that are trapped. To distinguish between these possibilities we again examined the expression of spoIIQ-gfp. As shown in Fig. 5A (arrows) a strong GFP signal was detected in some cells, invariably prespore-like compartments containing small amounts of DNA (Fig. 5B). Thus, at least some prespores were able to express a σF-dependent reporter at high levels. We then looked at the expression and localization of a σF-dependent gfp reporter placed in an oriC-distal part of the chromosome (dacF-gfp at −1770). Figure 5C and D shows that the pattern was similar, despite the difference in chromosome localization of the reporter: again a significant proportion of prespores expressing dacF-gfp could be detected in the triple mutant. These results probably rule out non-specific effects, such as some kind of stochastic defect in σF-dependent gene expression or cell viability, and strongly support the idea that the segments of DNA occasionally trapped in the prespores of the triple mutant are taken randomly or at least from a range of chromosomal locations.

Figure 5.

Relaxation of prespore chromosome trapping specificity in a racA soj spo0J triple mutant (also carrying spoIIIE36). Strains containing a σF-dependent gfp fusion placed at −455 (PspoIIQ-gfp; A, B) or −1770 (PdacF-gfp; C, D) were induced to sporulate and at t120 were examined by fluorescence microscopy. Typical fields of cells are shown simultaneously visualized for GFP (green) and DNA (DAPI, red) (A, C) or for DNA only (B, D). Arrows point to prespores expressing GFP, and arrowheads to prespores with DNA but no detectable GFP signal. Many prespore compartments, located between the major nucleoid masses, contain no detectable DNA.


RacA is a sporulation-specific effector of prespore chromosome segregation

We identified racA on the basis that it was strongly dependent on spo0H, encoded a likely DNA-binding protein, and the gene was located close to the PLR we previously reported as a cis-acting region capable of directing chromosome trapping in the prespore independent of the soj-spo0J system (Wu and Errington, 2002). The mutant phenotype arising from disruption of racA confirmed that it had a role in prespore chromosome segregation (Fig. 1). The partially functional GFP fusions made to RacA showed that the protein has two separable activities involved in nucleoid association and polar targeting (Fig. 2). Furthermore, the polar targeting was dependent on DivIVA, a polar anchor protein implicated both in division site selection and prespore chromosome capture (Marston et al., 1998; Thomaides et al., 2001). Ben-Yehuda et al. (2003) have recently reported a similar analysis of racA function, and our results closely accord with theirs. Ben-Yehuda et al. (2003) also provided more direct evidence for DNA binding, via chromatin immunoprecipitation experiments. Although the protein could be detected binding relatively non-specifically around the chromosome, preferred binding sites for the protein were detected mainly just to the left of oriC. Importantly, Ben-Yehuda et al. (2003) also showed that expression of racA in vegetative cells results in movement of the chromosome towards the cell poles, partially mimicking axial filament formation. On this basis, they proposed a model in which RacA provides an adhesive patch that anchors the oriC region to the cell pole via an interaction (direct or indirect) with divIVA. Our results are in complete accordance with this view.

Complementary roles for RacA and the Soj-Spo0J system in prespore chromosome segregation

Although the phenotype of the racA disruption was modest (sporulation was reduced about twofold), when combined with disruption of the soj-spo0J locus, a much stronger (20-fold) effect was seen (Table 2). This strong synergy suggests that the two systems have partially redundant roles in prespore chromosome segregation. This synergy was also seen with the chromosome trapping results. Thus, when either of these systems was disrupted on its own, the majority of cells succeeded in trapping the oriC region of the chromosome in the prespore compartment (Fig. 3B and F). However, when both systems were disrupted, specific trapping was virtually eliminated (Fig. 3E). These results strongly suggest that RacA is the putative DNA-binding protein that we previously proposed to be responsible for the near correct orientation of the chromosome in soj-spo0J mutants (Wu and Errington, 2002). A series of defined chromosome inversions and translocations established that a region of the chromosome stretching from about −150 to −350 (the PLR), was sufficient to direct adjacent chromosomal sequences into the prespore (Wu and Errington, 2002). We were surprised to find that only one of the preferred RacA-binding sites so far detected by Ben-Yehuda et al. (2003) lies in the PLR (at about −200); the other sites are located closer to oriC, with a particular concentration at approximately −70. These sites behaved as if they were not important for chromosome orientation in the inversion strains described by Wu and Errington (2002). There are several possible explanations for this. The one we currently favour would be that RacA finds its preferred binding sites by scanning along the DNA from the gene sequence at −416, through the PLR (about −350 to −150) to the preferred sites at −70, and this path is disrupted in the inversion strains.

A specific role for Soj in recruitment of the oriC region to DivIVA at the cell pole?

The previous discovery of a role for DivIVA in recruitment of the oriC region to the cell pole in sporulating cells suggested that one or more DNA-binding proteins, probably with some specificity for the oriC region, would interact with DivIVA. Examination of the divIVA13 mutant using the DNA-trapping assay revealed, surprisingly, significant trapping of two short zones of DNA about 400 kbp to the left and right of oriC. Other markers, both outside the normal trapping region and in the central area around oriC, were not trapped significantly. One possible explanation for these results would be the operation of a DivIVA-independent polar recruitment mechanism, acting at approximately ± 400. However, we noticed that in the mutant, Spo0J foci, which generally co-localise with oriC, were located far from the cell poles, at positions reminiscent of that taken up in vegetative cells (Fig. 4C). This led to an alternative model, in which the DivIVA/DNA binding protein system is required to extract the oriC chromosome region from its (hypothetical) vegetative chromosome segregation apparatus and deliver it to the pole. With Spo0J/oriC complexes fixed at their vegetative positions, the nucleoid might take up a configuration in which the ± 400 regions, either side of the Spo0J/oriC complexes, are closest to the cell pole and therefore most likely to be trapped in the prespore compartment (see below).

Irrespective of which of the above explanations is correct, we were surprised to discover that combination of racA and soj mutations led to a very similar pattern of trapping to that of the divIVA mutant (Fig. 3C and D). Furthermore, although neither soj nor racA single mutations reduced sporulation by as much as twofold, the double mutant showed a reduction of nearly 10-fold, near to that of the divIVA13 mutant (Table 2). The simplest explanation for these data would be that Soj and RacA are partially redundant DNA-interacting proteins, either of which can deliver the oriC region to DivIVA at the cell pole during sporulation.

Although at first sight, Soj does not obviously have the properties needed for the putative oriC/pole interaction, there are hints that both interactions might exist. First, Soj clearly interacts with Spo0J protein, which, as mentioned above, is intimately associated with the oriC region. This interaction is evident both in the requirement of soj for proper condensation of Spo0J foci (Marston and Errington, 1999), and in the dependence of Soj ‘jumping’ on spo0J (Marston and Errington, 1999; Quisel et al., 1999). Second, the jumping behaviour of Soj is influenced by the cell pole (Marston and Errington, 1999; Quisel et al., 1999), and we recently showed that this probably involves a transient interaction with MinD protein (Autret and Errington, 2003), which in turn depends on DivIVA protein for association with the cell pole (Marston et al., 1998). Interestingly, Soj jumping is enhanced at the onset of sporulation (Quisel et al., 1999). It is attractive to suppose that Soj shuttling from the nucleoid to the cell pole facilitates delivery of Spo0J/oriC complexes to the cell pole. In support of this idea, detailed studies of soj mutant cells has revealed that they form minicells at low frequency (∼1%) and that they have an associated slight increase in average cell length (A. L. Marston and J.E., unpubl.). These observations could be explained by a deficiency in polar movement of the nucleoid: in such circumstances the ‘nucleoid occlusion’ effect on division (see Harry, 2001) would tend to delay division at mid-cell but allow an increased likelihood of polar division.

On the basis of these ideas, we suggest that RacA and Soj both facilitate movement of the oriC region of the chromosome to the cell pole and that both proteins do so by direct or indirect interaction with DivIVA (Fig. 6A). This provides a simple explanation for the similarity of the phenotypes of divIVA versus racA/soj mutants (Fig. 6B). However, Soj and RacA clearly have specialized roles, as indicated by their distinct sporulation phenotypes. RacA, which is expressed specifically in sporulating cells and is presumably specialized for the extreme chromosome segregation into the prespore, may act primarily as an anchor. Soj, in contrast, is expressed vegetatively and clearly has a number of more general functions (see Introduction).

Figure 6.

Schematic model for prespore chromosome segregation and the possible roles of DivIVA, RacA, Soj and Spo0J.
A. The wild-type state. Spo0J molecules or complexes (green circles) are shown binding to their preferred sites in and around the oriC (black circle) region [(Lin and Grossman, 1998); see Fig. 3 for details]. Soj protein ‘jumping’ somehow facilitates movement of the Spo0J/oriC complex towards the cell pole as indicated by the green arrow (see text). RacA protein (blue circles) binds to multiple sites, generally to the left of oriC and attaches the oriC chromosome region to the pole by interaction (direct or indirect) with DivIVA, which is targeted to the cell pole (red arc). When the prespore septum forms (dotted grey line) the oriC region is efficiently trapped in the small prespore (P) compartment. The bulk of the chromosome (thin black line) remains in the mother cell (MC) compartment at this time and would normally be transported into the prespore by the action of SpoIIIE protein (Wu et al., 1995).
B, C. Effect of mutation of either divIVA (B) or racA and soj (C). In both cases, neither system needed to attach the oriC region to the cell pole is functional and the region, with associated Spo0J molecules, remains localized in the mother cell, positioned by the hypothetical vegetative chromosome segregation machinery (large black arrow). In this configuration, the chromosome is organized such that sequences about 400 kbp to the left or right of oriC are most likely to be trapped in the prespore compartment.
D. With spo0J also mutated, organization of the oriC region is perturbed and the specificity of the sequences trapped in the prespore compartment is relaxed.

Further evidence for a chromosome organization function for Spo0J

In either divIVA13 or soj/racA mutants significant trapping of markers located about 400 kbp to either side of oriC was detected (Fig. 3C and E). Experiments with a spoIIQ–gfp fusion provided direct support for this trapping, at least to the left of oriC (Fig. 4A and B). As mentioned above, the central part of the oriC region was localized far away from the poles in these cells. According to the model shown in Fig. 6B, in a divIVA mutant, or a soj racA double mutant, Spo0J/oriC complexes remain in the mother cell compartment, distant from the cell poles, at a position determined by the hypothetical vegetative chromosome segregation mechanism (black arrow in Fig. 6B). The organization of the chromosome around the oriC region is postulated to be such that the regions at about ± 400 are closest to the cell pole and therefore most likely to be trapped in the prespore compartment. When spo0J was added to the racA soj double mutant, residual trapping of the marker at ± 400 was more or less eliminated (Fig. 3E). We suggest that this specific trapping is lost because Spo0J is required for proper organization and positioning of the oriC region (Fig. 6C). Consistent with this idea, the triple mutant exhibited increased trapping of parts of the chromosome that are normally excluded from the prespore (Fig. 5C), as reported previously for the soj spo0J mutant (Sharpe and Errington, 1996), and recently confirmed by Lee et al. (2003). The precise function of Spo0J in chromosome dynamics of B. subtilis is somewhat controversial and other groups have reported effects on the control of chromosome replication (Webb et al., 1997; Lee et al. 2003). Although our new results do not solve this problem, they lend further support to an effect of Spo0J on organization of the oriC region of the chromosome. Furthermore, the remarkable specific trapping of markers at ± 400 may provide an experimental tool with which to probe for novel factors involved in the still poorly understood vegetative chromosome segregation system.

Experimental procedures

Bacterial strains and plasmids

The B. subtilis strains used in this study are listed in Table 3, together with the plasmids used and their construction.

Table 3. B. subtilis strains and plasmids.
Strain/plasmidRelevant genotypeaConstruction, source, or referenceb
  • a


  •  ′

     ′X or X

  • ′, the 5

  • ′end or the 3

  • end of the gene X has been truncated. Resistance gene abbreviations as follows: cat: chloramphenicol; tet: tetracycline; neo: neomycin; spc: spectinomycin; erm: erythromycin; bla: ampicillin.

  • b

    . For strains constructed by transformation, the source of the DNA used in the transformation is given first, with restriction enzymes, where used. The recipient strain is indicated after the arrow, with selected marker in parentheses.

B. subtilis  Laboratory stock
168 trpC2 Laboratory stock
BH1 trpC2 pheA1 spo0HΔ(HindIII-EcoRI)::cat Healy et al. (1991)
JH21063 trpC2 pheA1 Δ (soj-spo0J)::Kan M. Perego
KI1944 trpC2 pheA1 Δ(soj-spo0J)::spc thr::(Δsoj spo0J + erm) Ireton et al. (1994)
SWV215 trpC2 pheA1 spo0A::kan Xu and Strauch (1996)
36.3 trpC2 spoIIIE36 Piggot (1973)
42.3 trpC2 spoIIAA42 Errington and Mandelstam (1986)
1273 trpC2 ΔracA::spc see Experimental procedures
1274 trpC2 divIVA13-tet see Experimental procedures
1275 trpC2 Δ(soj-spo0J)::kan JH21063 (chr) → 168 (kan)
1276 trpC2 Δ(soj-spo0J)::kan thr::(Δsoj spo0J + erm) KI1944 (chr) → 1275 (erm)
1277 trpC2 Δ(soj-spo0J)::kan ΔracA::spc 1273 (chr) → 1275 (spc)
1278 trpC2 Δ(soj-spo0J)::kan thr::(Δsoj spo0J + erm) ΔracA::spc 1273 (chr) → 1276 (spc)
1279 trpC2 spoIIIE36 divIVA13-tet 1274 (chr) → 36.3 (tet)
1280 trpC2 spoIIIE36 divIVA13-tet spoIIQ::pSG4917 (PspoIIQ-gfp cat)pSG4917 → 1279 (cat)
1281 trpC2 spoIIIE36 divIVA13-tet spo0J-gfpmut1 neo 2641 (chr) → 1279 (neo)
1404 trpC2 spoIIIE36 dacF′-GFP cat Sharpe and Errington (1996)
1901 trpC2 (minD::erm) 1901 Edwards and Errington (1997)
2641 trpC2 spo0J-gfpmut1 neo Autret et al. (2001)
pIC156 bla spc Steinmetz and Richter (1994)
pSG250 bla erm Errington et al. (1992)
pSG1151 bla gfp cat Lewis and Marston (1999)
pSG4902 bla P xyl -gfp cat This study
pSG4915 bla ′racA-gfp cat racA (PCR, KpnI + SalI) intopSG1151 (KpnI + SalI)
PSG4916 bla gfp-racA′ cat racA′ (PCR, SalI + EcoRI) intopSG4902 (SalI + EcoRI)
PSG4917 bla P spoIIQ -gfp cat P spoIIQ (PCR, HindIII + EcoRI) intopSG1151 (HindIII + EcoRI)

General methods

Bacillus subtilis cells were made competent for transformation with DNA either by the method of Kunst and Rapoport, 1995) or by the method of Anagnostopoulos and Spizizen (1961) as modified by Jenkinson (1983) . DNA manipulations and E. coli transformations were carried out using standard methods (Sambrook et al., 1989). Solid medium used for growing B. subtilis was nutrient agar (Oxoid) and liquid medium was PAB (Oxoid antibiotic medium no. 3). Chlor-amphenicol (5 µg ml−1), kanamycin (5 µg ml−1), tetracycline (12 µg ml−1), erythromycin (1 µg ml−1) and lincomycin (25 µg ml−1), or 0.012% Xgal, were added as required. Media used for growing E. coli were 2× TY (Sambrook et al., 1989) and nutrient agar supplemented with ampicillin (100 µg ml−1) as required.

Induction of sporulation

Bacillus subtilis cells grown in hydrolysed casein growth media at 37°C were induced to sporulate according to the re-suspension method of Sterlini and Mandelstam (1969), as modified by Partridge and Errington (1993). Times (min) after resuspension of cells in the starvation medium were denoted t0, t30, etc.

Construction of racA deletion strains

To delete the racA gene from the B. subtilis chromosome, DNA fragments (about 1.4 kb long) from upstream and downstream (including a few bases into the coding region) of the racA gene were amplified from the wild-type strain SG38 by PCR, digested with SstI and SalI, then ligated to the spectinomycin resistance gene excised from pIC156 using the same enzymes. The ligation mixture was transformed into the wild-type B. subtilis strain 168 directly, with selection for spectinomycin resistance. Several transformants each were examined by PCR to confirm the deletion and one of the strain was designated 1273.

Introduction of the divIVA13 mutation into the wild-type (168) background

The original divIVA13 (N99D) mutation was linked to an intergrational plasmid as well as to two antibiotic resistance markers (Thomaides et al., 2001), making it complicated to manipulate the strain. To introduce the point mutation into the background of the wild-type strain 168 with just one selectable marker without the complication of a linked plasmid, the DNA fragment upstream of the divIVA gene (containing the whole of ylmH and part of ylmG), and the divIVA gene containing the mutation were amplified by PCR. The PCR products were digested with KpnI and SstI, respectively, ligated to the tetracycline resistance gene excised from pBEST309 using the same enzymes, and the ligation mixture was transformed directly into the wild-type strain 168. Several Spo- transformants were isolated and examined by PCR and DNA sequencing to confirm the structure and the presence of the divIVA13 mutation. One of the strains, containing the mutation and the tetracycline resistance gene inserted between ylmH and divIVA, were designated 1274.

SDS–PAGE and Western blotting

Cell pellet from 1 ml of sporulating culture was resuspended in 104 ml of lysis buffer containing protease inhibitor cocktail (Roche), incubated at 37°C for 5 min then left to stand on ice for 30 min before adding 36 ml of 4× SDS loading dye. Next, 25 ml of each sample was loaded and the proteins were separated by SDS–PAGE, then analysed by Western immnuoblotting using polyclonal anti-GFP antibodies (Clontech).

Fluorescence microscopy

Cells containing gfp fusion were grown at 30°C after induction of sporulation, and membrane dye FM5-95 (Molecular Probes) was added to the culture at 60 min at a final concentration of 90 µg ml−1. Cells were then viewed on agarose slides and images were obtained as described by Marston and Errington (1999). When nucleoid staining was required 6 µl of cells was mixed with 2 µl 4,6-diamidino-2 phenylindole (DAPI, Sigma) solution (1 µg ml−1 in 50% glycerol) in an Eppendorf for 1 min before viewing.


This work was supported by grants from the Biotechnology and Biological Sciences Research Council. J.E. was supported by a BBSRC Senior Research Fellowship. We thank R. Losick for communicating results ahead of publication.