The earliest event in bacterial cell division is the assembly of a tubulin-like protein, FtsZ, at mid-cell to form a ring. In rod-shaped bacteria, the Min system plays an important role in division site placement by inhibiting FtsZ ring formation specifically at the polar regions of the cell. The Min system comprises MinD and MinC, which form an inhibitor complex and, in Bacillus subtilis, DivIVA, which ensures that division is inhibited only in the polar regions. All three proteins localize to the division site at mid-cell and to cell poles. Their recruitment to the division site is dependent on localization of both ‘early’ and ‘late’ division proteins. We have examined the temporal and spatial localization of DivIVA relative to that of FtsZ during the first and second cell division after germination and outgrowth of B. subtilis spores. We show that, although the FtsZ ring assembles at mid-cell about halfway through the cell cycle, DivIVA assembles at this site immediately before cell division and persists there during Z-ring constriction and completion of division. We also show that both DivIVA and MinD localize to the cell poles immediately upon spore germination, well before a Z ring forms at mid-cell. Furthermore, these proteins were found to be present in mature, dormant spores. These results suggest that targeting of Min proteins to division sites does not depend directly on the assembly of the division apparatus, as suggested previously, and that potential polar division sites are blocked at the earliest possible stage in the cell cycle in germinated spores as a mechanism to ensure that equal-sized daughter cells are produced upon cell division.
An intriguing aspect of cell division in bacteria concerns the mechanism by which the division site is positioned. The earliest forming structure that marks the division site in bacteria is the Z ring, which comprises polymerized FtsZ molecules (Lutkenhaus and Addinall, 1997; Harry, 2001). Although there are possibly many factors that contribute to Z-ring positioning (Margolin, 2000; 2001; Harry, 2001), the Min system is the best characterized and has been studied extensively in both Escherichia coli and Bacillus subtilis. In both organisms, the Min system comprises three proteins. Two of these, MinD and MinC, interact to form a division inhibitor complex, MinCD, which directly prevents Z-ring formation. The third protein is different in the two organisms, DivIVA in B. subtilis and MinE in E. coli but, in both species, it is required for topological specificity of MinCD inhibition (Rothfield et al., 1999; Errington and Daniel, 2001).
During vegetative growth, E. coli and B. subtilis cells divide at the cell centre. The Min proteins are not essential for viability but, in their absence, cells divide at the poles, as well as at mid-cell, to form DNA-less minicells (Adler et al., 1967; Reeve et al., 1973). These polar divisions result from the formation of Z rings at polar positions (Bi and Lutkenhaus, 1993). Clearly, one of the roles of the Min system is to prevent Z-ring assembly in the polar regions of the cell. In E. coli, the MinCD inhibitor achieves this by a MinE-dependent pole-to-pole oscillation such that the time-average concentration of MinCD is highest at the cell poles and lowest at the cell centre (Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a,b; Fu et al., 2001; Hale et al., 2001).
In contrast, the B. subtilis Min proteins are not known to oscillate (Edwards and Errington, 1997; Marston et al., 1998; Marston and Errington, 1999; Thomaides et al., 2001). Fluorescence microscopy experiments have shown that, during vegetative growth, DivIVA concentrates at the cell poles, forming a polar cap (Edwards and Errington, 1997; Marston et al., 1998). It also localizes to the division site, and this requires the prior recruitment of both early (FtsZ) and late (PbpB) division proteins to this site (Marston et al., 1998; Marston and Errington, 1999). MinD and MinC also localize to the polar regions and the division site (Marston et al., 1998; Marston and Errington, 1999). It appears that, although MinC and MinD can be recruited to the division site independently of DivIVA, this protein is required to retain them at the nascent poles after division is complete (Marston and Errington, 1999). Therefore, during vegetative growth in B. subtilis, once the division apparatus is assembled at mid-cell, DivIVA and MinCD localize to this site and persist at the two mature poles formed by cell division. The three proteins are already localized at the old poles of the daughter cells from the previous division event. Thus, MinCD is piloted towards the poles by DivIVA, specifically to inhibit Z-ring formation in this region while allowing them to form within the central region of the cell.
Although previous studies have suggested that, during vegetative growth, the Min proteins localize during the latter stages of division (Marston et al., 1998; Marston and Errington, 1999), it is not known when these proteins localize to mid-cell relative to Z-ring formation and to the progress of septum formation. In this study, we have simultaneously examined the temporal and spatial localization of DivIVA relative to that of FtsZ, leading up to and during the first and second cell division after spore germination and outgrowth. This system is convenient for this purpose as the first cell division in this system occurs relatively synchronously within the population. There are also no complications from previous divisions. We show that DivIVA does indeed localize to the division site significantly later than FtsZ and persists at this site during Z-ring constriction and completion of division.
It is not currently known how the Min proteins are targeted to the division site in B. subtilis. This could occur via a direct interaction with the division apparatus itself or via another signal or structure at this site. The former possibility is consistent with the dependency of Min protein recruitment on the localization of the division apparatus, and it has been suggested that DivIVA may interact with the division protein FtsA (Edwards et al., 2000). DivIVA shares some similarity with tropomyosins (Edwards and Errington, 1997) and localizes to the poles of non-related organisms such as E. coli and Schizosaccharomyces pombe (Edwards et al., 2000). FtsA is also related to actin, which is involved in cytokinesis in eukaryotes and may explain this unusual observation. An interesting aspect of this issue concerns division site selection leading up to the first division after spore germination and outgrowth in B. subtilis. The dormant B. subtilis spore must germinate and allow a new rod-shaped cell to grow out before entry into the vegetative state of growth. After outgrowth, cell division occurs at mid-cell to produce two identically sized daughter cells. The first division appears to be morphologically and genetically the same as that observed during vegetative growth (Harry et al., 1999). However, unlike vegetative cells, the first cell poles generated in the outgrown cell do not result from cell division and therefore do not contain a division apparatus. How then are DivIVA, MinC and MinD targeted to the poles during germination and outgrowth? Is division site selection occurring via a different mechanism? Consistent with Min proteins functioning in this system, minicells have previously been observed after germination and outgrowth of B. subtilis spores (Coyne and Mendelson, 1974), although the majority were produced subsequent to the first division.
We have examined the localization of DivIVA and MinD during the first division after spore germination and outgrowth. We found that both proteins are present in significant levels in the dormant spore and localize to the cell poles of the germinated spore very early in the outgrowth process, well before assembly of the first Z ring at the division site. These results suggest that the division apparatus is not required for localization of the Min proteins to the nascent cell poles and that another signal, perhaps related to pole formation that is conserved in a wide range of organisms, is responsible for this localization.
Differential localization of FtsZ and DivIVA to the division site in germinated and outgrowing spores
The objective of these experiments was to determine the relative timing and pattern of assembly of Min proteins and the division protein FtsZ to the cell centre during the first cell cycle after germination. Spores of strain SU46 (Table 1) were germinated and grown out as detailed in Experimental procedures. This particular strain was used as it had been well characterized previously in spore outgrowth experiments. These spores are very synchronous and behave the same as wild type with respect to growth, DNA replication and Z-ring formation at the permissive temperature of 34°C (see Wu et al., 1995; Harry et al., 1999). One hundred minutes after placing spores in germination media (t100), outgrowth was under way, and samples were taken at 20 min periods from this time to cover approximately two cell cycles. Outgrowth and division were monitored by microscopy and cell length measure-ments. Average cell length increased from 2.6 ± 0.91 µm at t100 to 3.8 ± 0.97 µm at t180, with the first cell division occurring around t140. The doubling time after outgrowth in germination medium was calculated to be ≈ 40 min, indicating that the experiment would have covered an average of two cell cycles.
Table 1. . Strains and plasmids used in this study.
Representative fields of cells immunostained for both FtsZ and DivIVA over the time course are shown in Fig. 1. A number of different mid-cell staining patterns are apparent for both proteins. These staining patterns were broken down into three classes for both proteins and are depicted as cartoons in Table 2 and labelled in Fig. 1. For FtsZ, the categories of staining pattern are similar to those described for the division protein DivIB by Harry and Wake (1997): class 0 represents no signal (Fig. 1A); class I, either two dots on opposite sides of mid-cell representing the early formation of a ring structure (Fig. 1A) or a medial band representing a more mature Z-ring structure (Fig. 1D); class II, a central dot, representing a constricted Z ring at or near the completion of septum formation (Fig. 1G and J). To compare DivIVA mid-cell immunostaining with that of FtsZ, cells were scored for the assembly of DivIVA structures forming at the division site during the first (and second) cell division. Class 0 and 1 patterns were similar to the corresponding FtsZ patterns, 0 and I (Fig. 1B, E and H respectively). DivIVA class 2 staining patterns represent two closely juxtaposed centrally located bands (Fig. 1K). Unlike FtsZ, DivIVA remains at the nascent poles produced by cell division. These two bands therefore represent two populations of DivIVA that are present at the poles of each newborn cell. This type of pattern is observed when immunostaining for DivIVA, but not when using green fluorescent protein (GFP) fusions to detect this protein (see later and also Edwards and Errington, 1997; Marston et al., 1998), and is probably caused by the lysozyme treatment used in immunofluorescence degrading peptidoglycan in the division septum and increasing the level of separation of the two nascent cell poles.
Table 2. . FtsZ/DivIVA distribution in germinating cells.
The proportion of cells classified as FtsZ 0/DivIVA 0 remained quite high throughout the experiment (Table 2). The majority (>90%) of these cells represent permeabilized cells, in which a background level of fluorescence was present, but no central signal for either protein could be detected (Fig. 1A–C). There would also have been a degree of asynchrony of germination, and so some of the FtsZ 0/DivIVA 0 cells at later time points may have represented cells that had only just germinated and not yet reached the stage of the cell cycle at which FtsZ and DivIVA are recruited. Nevertheless, the scoring of these cells did not affect our overall conclusions on these experiments.
Immunostained cells were then scored using the criteria detailed above to determine the relative timing of assembly of FtsZ and DivIVA during the first and second divisions after germination, and the results are summarized in Table 2. At t100, most of the cells were class 0 for both FtsZ and DivIVA, representing cells still in the early stages of the first cell cycle (Fig. 1A and B). However, in a significant proportion of cells, a Z ring was observed (Table 2; Fig. 1A and C). By t120, a considerable proportion of cells contained a Z ring. At both these time points, there was little observable localization of DivIVA at the division site, indicating that FtsZ assembles at this site well before DivIVA (Table 2; Fig. 1D–F). At t140, when most outgrown cells were undergoing the first cell division, the proportion of cells containing both FtsZ and DivIVA signals increased significantly (from 12% to 42%), with most cells containing a Z ring (class I and II) and class 1 DivIVA localization staining patterns (Table 2; Fig. 1G–I). At t160, a significant proportion of cells contained class 2 DivIVA signals, but no FtsZ signal, indicating that Z rings had disassembled after the first cell division and had not yet reassembled at new division sites. This correlation of patterns was even more apparent at t180 (Table 2; Fig. 1O, white arrow). Another major class of cells at these later time points was FtsZ I/DivIVA 0, which represented cells undergoing their second cell division cycle, in which DivIVA had not yet assembled at the second division site (Table 2; Fig. 1O, grey arrow), clearly confirming the order in which these two classes of protein assemble at the sites of cell division. We estimate from our results that DivIVA assembles at the division site ≈ 20 min after FtsZ under these conditions. This corresponds to the shift in FtsZ I/DivIVA 0 cells to FtsZ I/DivIVA 1 between t120 and t140 in Table 2. In germination medium, this represents about half a cell cycle and also suggests that, although FtsZ rings assemble well before cell division, DivIVA rings assemble very close to the actual time of cell division during outgrowth. One further observation concerns the localization patterns of FtsZ and DivIVA during the late stages of septation. At the later time points (after t120), constricted Z rings were frequently observed between two DivIVA bands (e.g. Fig. 1L, arrow; Table 2, FtsZ II/DivIVA 2 cells), consistent with the fact that DivIVA remains at the cell poles and is maintained there while Z rings constrict to complete the septation process.
Image analysis using deconvolution algorithms has confirmed that FtsZ forms a ring-shaped structure (Sun and Margolin, 1998; Sievers and Errington, 2000). We took a similar approach to determine the structure of DivIVA at division sites using a DivIVA–GFP fusion protein. Figure 2 shows a pair of exponentially growing cells containing such a fusion (strain 1803; Table 1). There is a class 1 and a class 2 DivIVA structure (numbered in Fig. 2A). As these experiments were performed with a DivIVA–GFP fusion, only a single band is visible for a class 2 staining pattern. This pattern appears after the formation of the division septum and appears as two closely juxtaposed bands in immunofluorescence be–cause of the degradation of peptidoglycan between the two bands on lysozyme treatment (Marston et al., 1998). Rotation of the reconstructed three-dimensional image shows clearly that the class 1 structure is a ring, whereas the class 2 structure is of a similar diameter and contains no visible annulus (Fig. 2B and C). Therefore, DivIVA probably coats the whole of the invaginating division septum, resulting in a complete DivIVA polar cap upon conclusion of cell division. Thus, there is a clear difference between the contractile ring-shaped structures formed by FtsZ and the polar caps formed by DivIVA during the later stages of cell division. Once cell separation has occurred, polar DivIVA caps are much less intense than those present at newly formed division septa (Fig. 2A, arrows), indicating that some disassembly and/or degradation of DivIVA may occur.
Polar localization of DivIVA in germinating spores
Interestingly, in virtually all cells immunostained for DivIVA, it appeared that this protein was already assembled at the cell poles even at the earliest time points, before Z-ring formation at the mid-cell division site (Fig. 1C, arrows). In contrast, FtsZ did not appear to localize to the poles at all. However, the high levels of whole-cell fluorescence in these immunofluorescence experiments made it difficult to observe clear polar DivIVA staining at the early time points. In order to investigate this further, we examined DivIVA localization in strain 101 (Table 1). This strain contains a divIVA–gfp fusion in addition to an IPTG-inducible copy of ftsZ. In this strain, the background levels of fluorescence are extremely low (see below). Spores of strain 101 were germinated in the presence of IPTG as detailed in Experimental procedures. Germination and outgrowth were relatively synchronous within the population, and the rate of growth after germination was very similar to that in strain SU46. However, strain 101 (and 1979; see below) germinated ≈ 60–80 min later than SU46, and the times in Table 2 (for SU46) and those discussed below are not numerically equivalent. This is a frequently observed phenomenon between different spore preparations (E. J. Harry, unpublished observations). Comparison of germination rates with strain SU46 indicated that strains 101 and 1979 germinate ≈ 60–80 min later than SU46. However, once outgrowth began, the growth rate in germination medium was the same for strains 101 and 1979 as in SU46 (≈ 40 min; see above), and cell division appeared to be normal.
Significant outgrowth in strain 101 occurred at about t180 (Fig. 3; this would be approximately equivalent to t100 in SU46 cells). In this strain, spore coat autofluorescence was much higher than in immunostained cells (see Fig. 1) and is probably a result of the many treatments and washing steps involved in immunofluorescence removing the autofluorescent material from the spore coats. The level of autofluorescence made it difficult to observe DivIVA–GFP clearly at the poles of cells that were still located in ruptured spore coats (most of the cells at this time point; Fig. 3). Nevertheless, there did appear to be small polar DivIVA–GFP caps in many of these cells (Fig. 3A, arrow). This was further supported by the observation of chromosomal DNA in the DAPI-stained images of the same cell (Fig. 3B). Spore coat autofluorescence was also visible in this channel and could be used to confirm the absolute alignment of the images. In this image, however, the arrowed polar cap was not visible, implying that these caps were indeed caused by DivIVA–GFP and not spore coat fragments that remained attached to the cell poles. An overlay of the GFP and DAPI images confirms this (Fig. 3C). The overlaid autofluorescent spore coats appear yellow, whereas a small green polar DivIVA–GFP cap can still be observed at the end(s) of the germinated cell (Fig. 3C, arrow). Polar DivIVA caps were visible in all cells, except those in which the poles were obscured by autofluorescent spore coats that were still attached to both ends of the cell (not shown).
The above results suggest that DivIVA is recruited to the poles before and independently of Z-ring assembly at the division site of the outgrown cell. To confirm this, strain 101 was germinated in the presence and absence of IPTG. Strain 101 also contains an IPTG-inducible copy of FtsZ (see above; Table 1), and so it is possible to prevent Z-ring assembly and cell division by withholding IPTG from the culture medium. At t180, when spore outgrowth was under way, cells with and without IPTG were indistinguishable from each other (Fig. 4A and B respectively). By t210, cells in the presence of IPTG were undergoing cell division, and central DivIVA–GFP bands were clearly visible (Fig. 4C). In addition, polar DivIVA–GFP caps were obvious (Fig. 4C). In stark contrast, and as expected, no DivIVA was visible at the mid-cell division site in cells in the absence of IPTG (Fig. 4D). Nevertheless, polar DivIVA–GFP caps were clearly visible in the same cells (Fig. 4D), indicating that DivIVA assembles at these sites independently of Z-ring formation at mid-cell. By t240, the polar DivIVA–GFP caps appeared to become stronger in intensity, but a considerable amount of delocalized punctate signal was also visible around the cytoplasmic membrane, indicating that DivIVA was still being produced but was unable to localize properly because of the lack of new division septa (Fig. 4F). Therefore, these results clearly show that DivIVA assembles at polar sites very early in the germination and outgrowth process and that this localization occurs before and independently of Z-ring assembly at mid-cell.
GFP–MinD localization in germinating spores
A second function for DivIVA, in chromosome segregation, has been demonstrated for sporulating cells (Thomaides et al., 2001). To determine whether the polar localization of DivIVA immediately after spore germination was directly related to its Min function in the first division event in outgrown cells, we examined the localization of another Min protein, MinD, during spore outgrowth. It has been demonstrated previously that MinD localizes in a similar fashion to DivIVA and is indeed dependent on this protein for its localization (Marston et al., 1998). Spores were prepared from strain 1979 (Table 1), which contains a xylose-inducible gfp–minD fusion and an IPTG-inducible ftsZ gene (Marston et al., 1998). Although this strain used the same IPTG-inducible ftsZ construct as strain 101 (see Table 1), for unknown reasons cell division could not be completely blocked in the absence of inducer. Therefore, data are only presented for cells at t180 and t210 in the presence of 1 mM IPTG and 0.05% (w/v) xylose. Under these conditions, the GFP–MinD localization patterns observed were consistent with those reported previously for this strain in vegetatively growing cultures (Marston et al., 1998).
At t180, the GFP–MinD signal was clearly visible in germinating cells (Fig. 5A and B). In very small cells, where FtsZ rings would not be expected to have assembled (see Fig. 1), the GFP–MinD signal appeared to be localized right around the cytoplasmic membrane, although it did appear to be significantly stronger at the poles (Fig. 5A and B, left-hand cell). Owing to the relatively strong GFP signal, MinD localization (green) was clearly distinguishable from spore coat autofluorescence (Fig. 5A, yellow). As the cells increased in length, the GFP–MinD signal appeared to become relatively weaker around the cytoplasmic membrane and even stronger towards the cell poles (Fig. 5A and B, middle and right-hand cells). By t210, both polar caps and central bands of GFP–MinD were clearly visible (Fig. 5C and D) with a distribution similar to that reported in cells from a vegetatively growing culture (Marston et al., 1998). Thus, polar GFP–MinD caps are clearly visible in germinating cells before the first cell division (Fig. 5C and D, insert), indicating that MinD assembles along with DivIVA at polar caps very early in the germination and outgrowth process, and probably before Z-ring assembly at mid-cell. Furthermore, these results strongly suggest that the Min system does indeed play a role in positioning the first Z ring to form after spore germination and outgrowth.
Min proteins, but not FtsZ, are present in mature spores before outgrowth
Rowland et al. (1997) have shown previously that the division proteins FtsZ and DivIB are probably completely absent in dormant spores, as they cannot be detected in protein extracts of spore preparations. These proteins need to localize to the mid-cell site of the outgrown cell before the first division, and it appears that they are synthesized de novo (Rowland et al., 1997). As DivIVA and MinD could be readily detected in germinating cells at a very early stage, it seemed possible that they might be present within the mature spore. It is known that a σH-dependent promoter drives increased minCD expression during late exponential/early sporulation stages, and that DivIVA is present at the cell poles during the early stages of sporulation, which is consistent with the maintenance of both MinD and DivIVA in mature spores (Lee and Price, 1993; Thomaides et al., 2001). This would ensure recruitment of DivIVA and MinD to the poles during the early stages of germination and outgrowth, before any significant accumulation of cell division proteins, and would prevent polar Z-ring formation during this first cell cycle. De novo synthesis of division proteins would occur during outgrowth to allow their assembly at the mid-cell division site.
To estimate the amount of DivIVA and MinD, spores of strain 1979 were ruptured as described in Experimental procedures. After SDS-PAGE, DivIVA and GFP–MinD were immunodetected by Western blot. Protein stains of gels revealed different protein profiles between dormant spores and germinated cells as would be expected (not shown). Coomassie staining after SDS-PAGE showed no evidence of proteolysis during spore breakage and protein extraction (not shown). Both DivIVA and MinD could be detected in the mature spore samples (Fig. 6A and B, lane 1 respectively), indicating that they are both stored in the mature spore. Nevertheless, the levels of protein detectable in dormant spore samples were considerably less than those in germinated samples at t160 and t210 (Fig. 6A and B, lanes 2 and 3 respectively), indicating that these proteins are also synthesized de novo during germination. We estimate that the level of MinD and DivIVA is ≈10% and 20%, respectively, that of the proteins in germinated cells. This is significantly higher than the calculated highest level for FtsZ and DivIB in mature spores: 1.7% and 1.1%, respectively, of the amounts present before cell division (Rowland et al., 1997).
It should be noted that, although the level of DivIVA was determined from expression of the gene from the wild-type promoter, levels of GFP–MinD were determined from induced expression from the Pxyl promoter. Although not an absolute measurement of MinD levels in spores in germinated cells, we feel that these results still faithfully reflect the relative level of MinD under these conditions, as the amount of DivIVA in both spores and germinating cells closely reflected that of GFP–MinD, suggesting that Min proteins are stored in spores and also produced upon germination, and that the activities of PdivIVA and Pxyl are similar under the conditions used during germination (Fig. 6; see Discussion). As an additional control, spores of strain SU8, which represent wild-type cells, were also processed for immunodetection of MinD and DivIVA. Both proteins could be easily detected in dormant spore preparations at levels similar to those in strain 1979, indicating that both proteins are also naturally present in wild-type spores (not shown).
The Min system plays an important role in division site selection in bacteria by preventing Z-ring formation at the polar regions of cells. In B. subtilis, this is achieved by topological control, exerted by DivIVA, on the MinCD division inhibitor. Previous work has suggested that DivIVA pilots MinCD to the poles of the cell, thereby blocking Z-ring formation in these regions, while allowing it at the future division site at the cell centre. During vegetative growth, the Min proteins are localized to the division site and persist at this site, which becomes two poles once septum formation is complete. Localization of the Min proteins to the division site has been shown to require both early (FtsZ) and late (PBP 2B) division proteins (Marston and Errington, 1999).
In this study, we have examined the temporal and spatial aspects of this localization in more detail by performing co-localization of DivIVA and FtsZ proteins to the division site during the first two cell divisions after spore germination and outgrowth. This system is ideal for such studies as the first cell division occurs at mid-cell and relatively synchronously within the population. There are also no complications from previous cell divisions. We have shown that FtsZ, the first division protein to assemble at mid-cell, localizes to this site approximately halfway through the cell cycle, well in advance of septum formation (Fig. 1; Table 2, t120). This is in agreement with previous observations in both E. coli (Sun and Margolin, 1998) and B. subtilis (Marston et al., 1998). In contrast, DivIVA did not localize to this site until about 20 min later, which is approximately half a cell cycle under these conditions (Fig. 1; Table 2, t140). Therefore, DivIVA localizes to the division site very close to the time of division, and this is probably the same for MinD and MinC. This result is consistent with previous observations (Marston et al., 1998), which suggest that MinD and DivIVA localize at about the same time. Our results also show that DivIVA is recruited to the division site before the final stages of Z-ring constriction.
A further observation from our co-localization analysis is that constricted FtsZ rings are observed between two closely juxtaposed DivIVA bands (class II/2; Table 2), indicating that DivIVA does not constrict with the leading edge of the division septum. Rather, it appears to coat the cytoplasmic sides of the newly synthesized septum, ultimately producing polar caps at the new cell poles upon completion of cell division. This was further supported by image deconvolution analysis, which suggests that, although DivIVA initially assembles as a ring structure, similar to FtsZ (Sun and Margolin, 1998), it does not constrict. Instead, it forms a polar cap with no visible annulus during the later stages of septum formation (Fig. 2). It is likely that the other Min proteins, MinC and MinD, localize in a similar fashion, as their retention at the poles is dependent on DivIVA (Errington and Daniel, 2001).
The spore germination/outgrowth system also allowed us to address the question of how DivIVA and the other Min proteins are targeted to the division site. The co-localization data and previous work (Edwards and Errington, 1997; Marston et al., 1998; Marston and Errington, 1999) show that targeting of the Min proteins to the cell poles relies on their initial localization to the division site. Indeed, the dependence of Min protein localization on the assembly of all the components of the division apparatus has led to the suggestion that one or more of the division proteins might be directly involved in this targeting (Edwards et al., 2000). However, in a cell that has grown out from a spore, the poles have not resulted from completion of cell division, and FtsZ only localizes to mid-cell after spore outgrowth (Harry et al., 1999). This raises the question of whether the Min system actually functions during the first cell cycle in this system and, if so, how is localization of the Min proteins to the cell poles achieved? About 30 years ago, minicells were observed in a divIVB-1 mutant (identified later as an allele of minD; Varley and Stewart, 1992) after spore outgrowth, but most were produced after the first cell division (Coyne and Mendelson, 1974). We have now confirmed that the Min system does indeed play a role in positioning of the first Z ring that forms at mid-cell in the outgrown cell. Interestingly, immunofluorescence and GFP fusion experiments showed that DivIVA localized to virtually all cell poles of the outgrown cell. Furthermore, this localization occurred very early in the outgrowth process and included many cells that would not yet have formed a Z ring at the division site. Localization of a DivIVA–GFP fusion under conditions in which ftsZ expression was repressed (Fig. 4) confirmed our suggestion that DivIVA polar localization does not require FtsZ, at least at levels required for Z-ring formation.
Although the localization of MinC and MinD to the division site does not appear to require DivIVA, this protein is required to maintain them at these sites (Marston and Errington, 1999). To determine whether recruitment of DivIVA to the poles during germination was related to its Min function, or to another function possibly related to another role for DivIVA (chromosome segregation during sporulation; Thomaides et al., 2001), we investigated the localization of MinD under these conditions. MinD was also observed to localize to the cell poles at a very early stage after spore germination (Fig. 5). These results strongly suggest that the Min system does indeed play a role in division site selection during the first cell cycle after spore germination, and are consistent with the observation that polar Z rings form during the first cell cycle in outgrown cells of B. subtilis in the absence of MinCD (M. Migocki and E. J. Harry, unpublished).
As the Min proteins assembled at cell poles so early in the cell cycle, we determined whether they were also stored in the mature, dormant spore. Unlike in a newborn vegetative cell, which already contains accumulated levels of FtsZ and Min proteins to ensure that polar Z-ring formation does not occur, a spore is not likely to have the same level of proteins. It has been determined previously that FtsZ is not detectable in spores, and its level increases at least 60-fold by the time it forms a ring at mid-cell (Rowland et al., 1997). Our observation that the Min proteins localize to the cell poles very early after germination raises the possibility that Min proteins are stored in the dormant spore to ensure that, upon germination, division site selection is regulated immediately, and polar divisions do not occur. We have used Western blotting to show that, in contrast to FtsZ, DivIVA is indeed present in dormant spores at significant levels: 10–20% of the level in germinating cells (Fig. 6). It is very likely that MinD (and MinC) are also present in similar levels in the mature spore. This initial level of Min proteins is presumably just sufficient to ‘preprogramme’ the Min system in advance of FtsZ synthesis to ensure that division does not occur at the poles.
It is not yet known what is responsible for the targeting of Min proteins to division sites and, ultimately, to the cell poles in B. subtilis. Our finding that DivIVA localizes to poles of germinated and outgrown cells without the prior assembly of the division apparatus at this site suggests that its localization does not occur via a direct interaction with one or more components of the division apparatus, as proposed previously (Edwards et al., 2000). Remarkably, DivIVA is promiscuously targeted to the division sites of E. coli and fission yeast (Schizosaccharomyces pombe; Edwards et al., 2000). These organisms do not carry an equivalent known DivIVA homologue. Furthermore, division proteins in prokaryotes and eukaryotes are generally not homologous, although members of the actin superfamily (FtsA and actin) are involved in cell division in all these organisms (see Edwards et al., 2000). Nevertheless, it is tempting to speculate that DivIVA is attracted to a physical property unique to cell poles (see also Edwards et al., 2000). Alternatively, another unidentified protein could be responsible for attracting DivIVA to these sites. Whatever the case, as we observed DivIVA at the division site before completion of Z-ring constriction, the attracting signal would need to be present before this time. We propose that, soon after the initial stages of septal synthesis, perhaps requiring PBP 2B, DivIVA binds to this site in the form of a ring. Subsequently, as the Z ring constricts, DivIVA is laid down alongside the ingrowing envelope layers and forms a polar cap as the septal wall is hydrolysed and the two newborn cells separate. Lastly, although not likely, we cannot rule out the possibility that polar localization of the Min proteins immediately after spore germination occurs via a different signal from that responsible for division site targeting during vegetative growth, or that spores somehow retain polarity after asymmetric septation such that the Min proteins are ‘preassembled’ at cell poles upon germination. We feel the latter possibility is unlikely because there is no evidence that Min proteins become localized in a manner that imparts polarity to the developing spore. In particular, DivIVA does not localize to the asymmetric septum during sporulation (Thomaides et al., 2001). Furthermore, upon engulfment, the prespore is spherical and is not known to contain any peptidoglycan that could mark old cell poles. Peptidoglycan that is synthesized during sporulation is different in composition from that produced in vegetative cells (Popham, 2002). It will be of considerable interest to determine the localization of DivIVA homologues and the mechanism of division site selection in spherical cells such as Staphylococcus, Streptococcus and Enterococcus, and whether the promiscuity of DivIVA localization extends to organisms that do not contain FtsA/actin paralogues such as the Mycoplasma. Recent studies with spherical rodA mutants of E. coli suggest that Min proteins may indeed recognize a geometrical feature of cell poles rather than any specific component (Corbin et al., 2002).
In summary, our results have outlined the timing of assembly of DivIVA and FtsZ proteins to the division site relative to the cell cycle in B. subtilis using a synchronous model system: spore germination and outgrowth. We have shown that FtsZ forms a ring at the division site about halfway through the cell cycle, well in advance of cell division, and that DivIVA only assembles at the site of division just before or, more likely, immediately after the onset of division. DivIVA proteins do not form contractile rings like division proteins, but assemble to coat the newly synthesized sides of the division septum, resulting in new polar caps upon cell division. Furthermore, Min proteins are present at cell poles independently of FtsZ and may assemble at polar sites because of an affinity for curved surfaces rather than through interaction with specific division proteins. The use of a spore germination system will permit the detailed characterization of which factors are required for Min protein localization.
Strains and media
The strains and plasmids used in this study are listed in Table 1.
Where necessary, IPTG was added to either solid or liquid media to a final concentration of 1 mM and xylose to a final concentration of 0.1% (w/v) unless otherwise stated. B. subtilis transformations were carried out by the method of Anagnostopoulos and Spizizen (1961) as modified by Jenkinson (1983).
Spore preparation and germination
Spores were prepared as detailed by McGinness and Wake (1979) and stored in Milli-Q water in the dark at 4°C. A sample of 10 ml of germination medium (GMD; Harry et al., 1999) in a 125 ml flask was inoculated with spores to an A600 of 1.0. Germination of all spores was carried out at 34°C.
Preparation of proteins from dormant and germinated spores
Proteins were prepared from dormant spores of strain 1979 as follows. Approximately 9 × 109 spores were pelleted in a 1.5 ml microfuge tube, and 0.15 g of 75–150 µm acid-washed glass beads (Sigma Chemical) were added. The spores and glass beads were then resuspended in 0.3 ml of 10 mM Tris-HCl, pH 9.8, 5 mM EDTA, 1 mM phenylmethylsulphonyl fluoride (PMSF) (TEP). The spores were then broken open in a cold room using a tight-fitting PTFE pestle (Sigma). Additional PMSF was added to the paste at 5 min intervals. Samples were checked microscopically to ensure ≥ 80% cell/spore breakage (usually after about 30 min grinding). Aliquots were snap frozen in N2 (l) and stored at −80°C until required.
Samples (10 ml) of germinating spores of strain 1979 inoculated to A600 of 1.0 at t0 were harvested by centrifugation at t160 and t210. The pellets were resuspended in 1.0 ml of TEP, and lysozyme was added to a final concentration of 2 mg ml−1. The tubes were then incubated at 37°C for 5 min, and 30 µl of 30% (w/v) Sarkosyl was added. The lysate was aliquoted into tubes, snap frozen in N2 (l) and stored at −80°C until required.
All protein samples were resuspended in SDS loading buffer to an A600 equivalent to 10. Samples (10–30 µl) were loaded into lanes on 1.5-mm-thick duplicate 10% (w/v) polyacrylamide gels and electrophoresed. One of the gels was then stained with Coomassie brilliant blue to obtain protein profiles, whereas the other one was placed in a Mini-transblot electrotransfer apparatus (Bio-Rad), and the proteins were transferred to nitrocellulose membranes (Amersham). The GFP–MinD fusion was detected using rabbit anti-GFP peptide antibodies (1:50 dilution; Clontech), and DivIVA was detected using rabbit polyclonal anti-DivIVA antiserum (1:1000 dilution; Marston et al., 1998). All proteins were detected using horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (1:5000 dilution; Bio-Rad) with the Opti-4CN system (Bio-Rad).
Immunofluorescence microscopy was carried out as detailed by Harry et al. (1999) except that glutaraldehyde was added to 0.008%. Rabbit polyclonal anti-DivIVA antisera followed by fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit secondary antibodies were used to detect DivIVA (Marston et al., 1998), and affinity-purified FtsZ antibodies, purified from sheep FtsZ antiserum (see Harry et al., 1999), followed by Cy3-conjugated donkey anti-goat secondary antibodies (Jackson ImmunoResearch) to detect FtsZ. Samples were also counterstained with DAPI (0.20 µg ml−1) to visualize DNA. Immunostained samples were viewed and processed as described by Harry and Wake (1997).
GFP-labelled samples were mounted on agarose slides as detailed by Glaser et al. (1997) with the following modifications. An aliquot of 1 ml of 1.2% (w/v) molten agarose in dH2O was pipetted onto a clean slide, and a coverslip was immediately laid on top to provide a flat mounting surface. When the agarose had set, the excess was trimmed away from the edges of the coverslip, and the coverslip was gently removed with a scalpel. On addition of the sample to the agarose, the coverslip was replaced and the slide viewed. Images were acquired using a Photometrics Quantix cooled CCD camera driven by metamorph version 4.5.6 software (UIC). GFP images were acquired using an Endow filter set (Chroma set 41018; Chroma Technologies) and DAPI using a triple bandpass filter (Chroma set 6001v2SBX using a GFPuv exciter filter). Image stacks were acquired, and deconvolution and three-dimensional image reconstruction were performed as described by Lewis et al. (2000). Images were processed and prepared for publication as detailed by Feucht and Lewis (2001).
P.L. appreciates the assistance of the Royal Society for the initial stages of this project, and Jeff Errington for supplying anti-MinD antibodies. The majority of this work was funded by an Australian Research Council Large Grant Award to P.L. (A00105185) and E.J.H. (A10009161). E.J.H. is the recipient of an ARC QEII Research Fellowship.