Lateral FtsZ association and the assembly of the cytokinetic Z ring in bacteria

Authors

  • Leigh G. Monahan,

    Corresponding author
    1. Institute for the Biotechnology of Infectious Diseases, University of Technology, Sydney, NSW 2007, Australia.
      E-mail leigh.monahan@uts.edu.au; Tel. (+61) 2 9514 4066; Fax (+61) 2 9514 4201.
    Search for more papers by this author
  • Andrew Robinson,

    1. School of Chemistry, University of Wollongong, Wollongong, NSW 2522, Australia.
    Search for more papers by this author
  • Elizabeth J. Harry

    1. Institute for the Biotechnology of Infectious Diseases, University of Technology, Sydney, NSW 2007, Australia.
    Search for more papers by this author

E-mail leigh.monahan@uts.edu.au; Tel. (+61) 2 9514 4066; Fax (+61) 2 9514 4201.

Summary

Cell division in bacteria is facilitated by a polymeric ring structure, the Z ring, composed of tubulin-like FtsZ protofilaments. Recently it has been shown that in Bacillus subtilis, the Z ring forms through the cell cycle-mediated remodelling of a helical FtsZ polymer. To investigate how this occurs in vivo, we have exploited a unique temperature-sensitive strain of B. subtilis expressing the mutant protein FtsZ(Ts1). FtsZ(Ts1) is unable to complete Z ring assembly at 49°C, becoming trapped at an intermediate stage in the helix-to-ring progression. To determine why this is the case, we used a combination of methods to identify the specific defect of the FtsZ(Ts1) protein in vivo. Our results indicate that while FtsZ(Ts1) is able to polymerize normally into protofilaments, it is defective in the ability to support lateral associations between these filaments at high temperatures. This strongly suggests that lateral FtsZ association plays a crucial role in the polymer transitions that lead to the formation of the Z ring in the cell. In addition, we show that the FtsZ-binding protein ZapA, when overproduced, can rescue the FtsZ(Ts1) defect in vivo. This suggests that ZapA functions to promote the helix-to-ring transition of FtsZ by stimulating lateral FtsZ association.

Introduction

FtsZ, an ancestral homologue of tubulin, is fundamental to the process of cell division in bacteria. FtsZ polymerizes into a ring structure (the Z ring) on the inner surface of the cytoplasmic membrane, which establishes the location of the future division site and acts as a scaffold for the assembly of the division apparatus (Harry et al., 2006). During cytokinesis, the Z ring constricts at the leading edge of the developing septum (Bi and Lutkenhaus, 1991) and actively generates an inward pinching force on the cell membrane (Osawa et al., 2008). The Z ring is constantly turned over in vivo, both before and during constriction, through FtsZ exchange between the ring and the intracellular FtsZ pool (Stricker et al., 2002; Anderson et al., 2004). This dynamic behaviour depends on the ability of FtsZ to bind and hydrolyse GTP (de Boer et al., 1992; RayChaudhuri and Park, 1992; Mukherjee et al., 1993; Stricker et al., 2002).

In the presence of GTP, FtsZ monomers reversibly assemble head-to-tail to form linear protofilaments in vitro (Mukherjee and Lutkenhaus, 1994). Depending on experimental conditions, these protofilaments can exist in isolation (Romberg et al., 2001) or associate together via lateral interactions between subunits in separate protofilaments to form filament bundles and other higher-order polymers (Bramhill and Thompson, 1994; Erickson et al., 1996; Yu and Margolin, 1997; Lowe and Amos, 1999; 2000; Gonzalez et al., 2003; Oliva et al., 2003). In vivo, the Z ring almost certainly consists of some arrangement of head-to-tail FtsZ filaments (Li et al., 2007). However, the extent and physiological significance of lateral FtsZ association is at present unclear. While several studies strongly suggest that lateral interactions between FtsZ protofilaments are important for FtsZ function (Lu et al., 2001; Stricker and Erickson, 2003; Dajkovic et al., 2008; Lan et al., 2009), there is currently little direct evidence to suggest that these interactions play any role in the actual assembly of the Z ring in vivo.

In rod-shaped bacteria, Z ring formation occurs precisely at midcell and is co-ordinated temporally with the progress of the cell cycle. Ultimately, this ensures that the division septum forms between replicated and segregated chromosomes to produce viable daughter cells. Z ring formation is controlled by a large number of proteins that directly bind FtsZ and modulate its assembly properties in vivo. In Bacillus subtilis for example, these proteins include FtsA (Feucht et al., 2001), SepF (YlmF; Hamoen et al., 2006; Ishikawa et al., 2006) and ZapA (Gueiros-Filho and Losick, 2002), all of which positively regulate FtsZ assembly, as well as MinC (Gregory et al., 2008), Noc (Wu and Errington, 2004), EzrA (Levin et al., 1999), ClpX (Weart et al., 2005) and UgtP (Weart et al., 2007), which function as inhibitors. Critically, it is at present unclear how most of these proteins exert their effects on FtsZ function in the cell.

Unlike many of the FtsZ accessory factors identified to date, the ZapA protein is widely conserved among bacteria, suggesting that its role is of general importance (Gueiros-Filho and Losick, 2002). Although ZapA is dispensable under normal conditions in B. subtilis, it colocalizes with FtsZ and becomes essential for cell division when FtsZ function is perturbed by a reduction in cellular FtsZ levels or by the loss of another FtsZ regulatory protein (EzrA; Gueiros-Filho and Losick, 2002). In vitro, ZapA directly binds FtsZ with a 1:1 molar stoichiometry, and stimulates the lateral association of FtsZ protofilaments into extensive bundles (Gueiros-Filho and Losick, 2002; Low et al., 2004; Small et al., 2007). However, it is currently unknown how ZapA affects FtsZ assembly in the cell, and whether it acts to promote the actual formation of the Z ring, enhance the stability of the ring, or otherwise contribute to the dynamic function of FtsZ.

As well as forming ring structures, FtsZ has been shown to assemble into mobile helices in a wide variety of bacterial species (Ben-Yehuda and Losick, 2002; Thanedar and Margolin, 2004; Grantcharova et al., 2005; Chauhan et al., 2006; Thanbichler and Shapiro, 2006). Significantly, a recent study in our laboratory showed that in B. subtilis, FtsZ exists in a dynamic pole-to-pole helix at all stages of growth, and that this helical structure plays a pivotal role in the assembly of the midcell Z ring (Peters et al., 2007). In newborn B. subtilis cells, FtsZ moves randomly within the extended helical pattern. However, at a later stage of the cell cycle, the helix is remodelled so that the majority of FtsZ molecules, while still localized in a helical pattern, are restricted to a small central region of the cell. This central intermediate helix persists for a considerable period, before finally reorganizing into the Z ring at midcell (Peters et al., 2007). To fully understand Z ring formation, it will be crucial to determine how this helix-to-ring transition occurs in vivo, and what role FtsZ accessory proteins play in this process.

We previously showed that in a unique ftsZ mutant of B. subtilis, known as ts1 (Nukushina and Ikeda, 1969), the mutant FtsZ protein is unable to undergo the helix-to-ring transition. This strain is temperature sensitive for cell division, and forms long, septum-less filaments when grown at a non-permissive temperature of 49°C (Callister and Wake, 1981). The mutant protein, FtsZ(Ts1), contains two amino acid substitutions residing in the interior of the C-terminal globular domain of FtsZ, one of which (A240V) is responsible for thermosensitivity (Michie et al., 2006). The second substitution (A278V) partially counteracts the effects of the A240V mutation, but is required to maintain the stability of the FtsZ(Ts1) protein at low temperatures (Michie et al., 2006). At 49°C, FtsZ(Ts1) is unable to form Z rings in vivo, and instead assembles into short helical structures at potential division sites along the filamentous cell (Michie et al., 2006). These helices reorganize into functional Z rings, however, when ts1 cells are shifted from 49°C down to the permissive temperature (Michie et al., 2006). This dynamic reorganization closely mirrors the final stages of wild-type Z ring formation from the short helical intermediate (Peters et al., 2007), and suggests that the FtsZ(Ts1) helix is itself a trapped intermediate in the Z ring assembly pathway (Michie et al., 2006). Elucidating exactly why the FtsZ(Ts1) helix is unable to reorganize into a ring at 49°C is likely to provide valuable information regarding the mechanism of Z ring formation.

For this reason, we have performed molecular, cellular and biochemical experiments to identify the specific defect of the FtsZ(Ts1) protein in vivo. Our results indicate that while FtsZ(Ts1) is able to polymerize normally into head-to-tail protofilaments, it is defective in the ability to support lateral associations between these filaments at high temperatures. This strongly suggests that lateral FtsZ association is required for Z ring formation, or more specifically, for the remodelling of the helical FtsZ intermediate into a ring at midcell. In addition, we show that the ZapA protein, when overproduced, can restore normal Z ring formation and cell division in the ts1 mutant at 49°C, suggesting that ZapA functions to promote the helix-to-ring transition of FtsZ in vivo by stimulating lateral association.

Results

FtsZ(Ts1) is defective in lateral protofilament association at high temperatures in vitro

It is intriguing that the mutant FtsZ(Ts1) protein, while unable to complete the process of Z ring formation at 49°C, maintains the ability to form well-defined, dynamic helical localizations in the cell at this temperature, and is present at a similar intracellular concentration to wild-type FtsZ (Michie et al., 2006). This indicates that FtsZ(Ts1) is soluble and at least partially folded in vivo at 49°C, but suggests that the mutant protein may harbour a specific defect in self-assembly that prevents it from undergoing the normal helix-to-ring polymer transition. As a first step towards identifying this defect, we used an in vitro electron microscopy approach to examine the polymerization properties of purified FtsZ(Ts1). Full-length, untagged FtsZ(Ts1) was purified along with an equivalent wild-type FtsZ control as described in Experimental procedures. Each protein was then polymerized in solution by the addition of GTP, and negative stain electron microscopy was used to examine the polymers formed. Initially, polymerization reactions were carried out at room temperature (23°C). Under these conditions, the wild-type protein readily assembled into large ring-like structures, which were composed of a bundled arrangement of FtsZ protofilaments (Fig. 1A). Individual, isolated protofilaments were also observed (arrows in Fig. 1B). Under the same conditions, the mutant FtsZ(Ts1) protein assembled into bundled ring structures identical to those seen for wild-type FtsZ (Fig. 1C). In addition, FtsZ(Ts1) formed some curved, linear protofilament bundles (Fig. 1D), which occasionally showed branching (arrow in Fig. 1D). These results indicate that at room temperature, purified FtsZ(Ts1) is able to polymerize normally into head-to-tail protofilaments, which can undergo lateral association into higher-order structures that are essentially similar to those of wild-type FtsZ.

Figure 1.

FtsZ(Ts1) is defective in lateral protofilament association at high temperatures in vitro. FtsZ and FtsZ(Ts1), at 0.6 mg ml−1, were incubated at various temperatures in the presence of 2 mM GTP for 2 min. Polymers were visualized using negative stain transmission electron microscopy.
A and B. Wild-type FtsZ at room temperature. Arrows in B point to isolated protofilaments.
C and D. FtsZ(Ts1) at room temperature. Arrow in D illustrates branching within a protofilament bundle.
E. FtsZ(Ts1) at 35°C.
F. Cartoon representation of FtsZ(Ts1) protofilaments in E.
G. (i and ii) Wild-type FtsZ polymerized in the presence of 20 mM CaCl2 at 35°C.
H. (i and ii) FtsZ(Ts1) in 20 mM CaCl2 at 35°C. Scale bars, 50 nm.

To examine the thermosensitive defect of FtsZ(Ts1), it was necessary to repeat the polymerization reactions at higher temperatures. However, preliminary experiments revealed that the purified FtsZ(Ts1) protein becomes insoluble when placed at temperatures above 40°C, rapidly precipitating out of solution. Interestingly, given that FtsZ(Ts1) remains soluble in the cell up to at least 49°C, this indicates that the thermal stability of FtsZ(Ts1) is lower in vitro than in vivo, and suggests that any specific effects of temperature on FtsZ(Ts1) self-assembly in vitro should be visible at temperatures of 40°C or even lower under our experimental conditions.

Accordingly, we examined FtsZ(Ts1) polymerization at 30°C, 35°C and 40°C. At all three temperatures, the mutant protein was found to readily assemble into single protofilament strands, which were very similar in their conformation and dimensions to the isolated protofilaments sometimes seen for wild-type FtsZ (see above). However, in stark contrast to the results obtained at room temperature, FtsZ(Ts1) was not observed to form any bundled protofilament structures under these conditions. A representative electron micrograph of FtsZ(Ts1) polymerized at 35°C is shown in Fig. 1E, while the cartoon in Fig. 1F illustrates the location of FtsZ(Ts1) protofilaments within the micrograph, as these can be difficult to discern among the background staining. Importantly, no ordered structures were detected when FtsZ(Ts1) was assayed in the absence of GTP (data not shown), indicating that the protofilaments observed following GTP addition are genuine polymers arising by GTP-dependent FtsZ(Ts1) self-assembly. As expected, the polymerization of wild-type FtsZ was not affected by temperature. At 35°C, the wild-type protein formed an identical mixture of bundled ring structures and isolated protofilaments to that observed at room temperature (data not shown). Taken together, these results indicate that at high temperatures, FtsZ(Ts1) retains the ability to assemble into head-to-tail protofilaments in vitro, but becomes defective in lateral protofilament association.

Interestingly, a thermosensitive defect in lateral association could explain the tendency of FtsZ(Ts1) to form highly bundled polymer structures rather than isolated protofilaments at low temperatures (compare Fig. 1D and B). Under these conditions, in which the temperature-sensitive A240V mutation is presumably silent, it may be that the counteractive A278V mutation actually promotes lateral protofilament association to a level above that of wild-type FtsZ. This would be consistent with previous work showing that B. subtilis cells with a single A278V mutation in FtsZ divide slightly more efficiently than wild type (Michie et al., 2006).

If the FtsZ(Ts1) defect is specific to lateral association, we hypothesized that it may be possible to restore FtsZ(Ts1) self-assembly to that of the wild-type protein at high temperatures by the addition of Ca2+. Ca2+ is known to induce lateral FtsZ protofilament interaction in vitro (Yu and Margolin, 1997; Lowe and Amos, 1999), although the mechanism by which it does so is unclear. To test this idea, polymerization reactions were repeated for the mutant and wild-type FtsZ proteins at room temperature and 35°C in the presence of 20 mM CaCl2. As predicted, the two proteins formed identical polymer structures under these conditions. At both room temperature (data not shown) and 35°C (Fig. 1G and H), FtsZ(Ts1) and wild-type FtsZ assembled into thick, linear protofilament bundles. No isolated protofilaments were observed for either protein. This confirms that the mutant FtsZ(Ts1) is defective in lateral association in vitro at high temperatures, and that the defect can be rescued by Ca2+.

Ca2+ suppresses ts1 thermosensitivity in vivo

If the lateral interaction defect observed for FtsZ(Ts1) in vitro also holds within the cell, it could explain why the mutant protein is able to polymerize into a helical structure at non-permissive temperatures, but is blocked from reorganizing into a stable Z ring. However, given that the assembly properties of FtsZ depend intimately on solution conditions (Popp et al., 2009), and that the intracellular environment differs remarkably from that encountered in in vitro experiments, it was necessary to confirm the nature of the FtsZ(Ts1) defect directly within ts1 cells. As Ca2+ rescues FtsZ(Ts1) for lateral association in vitro, we wondered whether the addition of Ca2+ to the culture medium would restore Z ring formation and cell division in ts1 cells growing at the non-permissive temperature (49°C). To this end, the ts1 strain (SU111) was cultured at 49°C in media supplemented with a range of concentrations of CaCl2. After growth for 1 h at 49°C, cells were examined by phase-contrast microscopy. As shown in Fig. 2A–C, ts1 cells were extremely long (filamentous) in the absence of added Ca2+, but became progressively shorter with increasing CaCl2 concentrations. In fact, at the highest concentration tested (300 mM), ts1 cells were only slightly longer (1.5-fold) than an isogenic wild-type strain (SU110) grown under the same conditions (Fig. 2A). This reduction in ts1 cell length is indicative of a Ca2+-dependent rescue of cell division at 49°C. It was not simply caused by a reduction in the rate of cell growth (extension), because the ts1 strain exhibited an equivalent mass doubling time in the presence and absence of CaCl2 (∼30 min). Moreover, clear separations, indicative of recently formed division septa, could be seen between cells in the presence of CaCl2 (arrows in Fig. 2C). In contrast, the long ts1 filaments formed in the absence of added CaCl2 were completely devoid of visible septa (Fig. 2B).

Figure 2.

Ca2+ rescues cell division and Z ring formation in vivo in the ts1 mutant. ts1 cells (strain SU111 [ftsZ(ts1)] or SU489 [ftsZ(ts1), amyE::Pxyl-ftsZ(ts1)-yfp]) were grown to mid-exponential phase at 34°C in the presence of various concentrations of CaCl2, shifted to 49°C for 1 h and analysed for cell length and morphology using phase-contrast microscopy (SU111), or for FtsZ(Ts1) localization by fluorescence (SU489).
A. Average cell length values versus CaCl2 concentration for SU111 and an isogenic wild-type control strain (SU110). Error bars denote the standard error of the mean. One hundred cells were scored at each CaCl2 concentration for SU111, while 200 were scored for SU110.
B and C. Phase-contrast images of SU111 cells grown at 49°C in the absence (B) and presence (C) of 200 mM CaCl2. Arrows in C point to separations between divided cells.
D and E. FtsZ(Ts1)–YFP localization in SU489 cells at 34°C in the absence (D) and presence (E) of 200 mM CaCl2.
F. FtsZ(Ts1)–YFP localization at 49°C without added CaCl2. The filled arrow points to an example of a helix-like localization, while the open arrow illustrates a fluorescent dot pattern consistent with helix formation.
G and H. Enlarged images of a helix (G) and a helical dot pattern (H) under the same conditions as in F.
I. FtsZ(Ts1)–YFP localization at 49°C in the presence of 200 mM CaCl2. Arrows point to normal-looking Z rings. Scale bars, 5 μm.

Presumably, Ca2+ rescues cell division in ts1 at 49°C by restoring the ability of the FtsZ(Ts1) protein to assemble into functional Z rings. To verify this, we examined the in vivo localization of FtsZ(Ts1) in the presence and absence of Ca2+ using a yellow fluorescent protein (YFP) tag. Given that fluorescent protein fusions to FtsZ are not fully functional in bacteria (Ma et al., 1996; Levin et al., 1999), a strain harbouring both the native copy of the ftsZ(ts1) allele and a separate ftsZ(ts1)-yfp gene fusion under control of the xylose-inducible Pxyl promoter at the amyE locus (SU489) was used. We have previously shown that when induced using a low level of 0.1% xylose, the FtsZ(Ts1)–YFP fusion in SU489 provides an accurate marker of FtsZ(Ts1) localization without affecting the ts1 phenotype (Michie et al., 2006). Initially, the SU489 [FtsZ(Ts1)–YFP] strain was grown in the presence of 0.1% xylose, with or without 200 mM CaCl2, at 34°C (the permissive temperature for ts1). Cells were examined by fluorescence microscopy at mid-exponential phase, and the FtsZ(Ts1)–YFP fusion was found to localize predominantly into normal-looking, midcell Z rings in both the presence and absence of CaCl2 (Fig. 2D–E). Thus the addition of 200 mM CaCl2 has no discernable effect on Z ring formation in ts1 under permissive conditions.

FtsZ(Ts1) localization was then examined after shifting SU489 [FtsZ(Ts1)–YFP] cells in mid-exponential phase from 34°C to 49°C for 1 h. In the absence of CaCl2, as reported previously (Michie et al., 2006), the FtsZ(Ts1)–YFP fusion localized at regular intervals along the filamentous cell (Fig. 2F), at potential division sites between segregated nucleoids (see Fig. S2). These localizations predominantly comprised of short helix-like structures [36% of 200 localizations scored; Fig. 2F (filled arrow) and G] and dot patterns consistent with helix formation [53%; Fig. 2F (open arrow) and H]. Only 1% of localizations resembled normal Z rings, while the remaining 10% included tilted bands and other irregular patterns. In the presence of 200 mM CaCl2, however, a vast increase in the number of normal-looking Z rings was observed [Fig. 2I (white arrows)], with these making up 34% of 200 localizations scored. The remaining localizations consisted of short helical structures (22%), helix-like dot patterns (36.5%) and other patterns similar to those seen in the absence of added CaCl2. Again the localizations occurred between nucleoids (data not shown). These results confirm that the addition of Ca2+ to the growth medium restores the ability of FtsZ(Ts1) to form functional Z rings at 49°C, thereby enabling cell division to occur.

Taking into account both the in vitro and in vivo data presented thus far, it seems likely that Ca2+ rescues Z ring formation in ts1 cells by promoting lateral connections between FtsZ(Ts1) protofilaments, either directly or indirectly, to compensate for an inherent lateral interaction defect. In support of this, we performed a series of experiments which confirmed that the addition of Ca2+ to the culture medium does indeed stimulate lateral FtsZ association in B. subtilis cells. For example, we found that Ca2+ rescues the filamentous phenotype of a strain overproducing the MinCD protein complex, an inhibitor of lateral FtsZ association (Dajkovic et al., 2008; Scheffers, 2008). These data, which are presented in Supporting information (Fig. S1), confirm that the mutant FtsZ(Ts1) protein is indeed defective in lateral association in vivo.

ZapA overproduction rescues the FtsZ(Ts1) defect

Our observation of a lateral interaction defect for the FtsZ(Ts1) protein has important implications regarding the mechanism of Z ring formation and the role of lateral FtsZ association in wild-type bacterial cells. If the defect blocks FtsZ(Ts1) from completing the normal helix-to-ring FtsZ assembly pathway in vivo, then lateral association must play an essential role in the FtsZ polymer transitions that lead to the formation of the Z ring. It is likely that in wild-type cells, these polymer transitions are controlled at least to some degree by the various proteins known to interact with FtsZ and modulate its assembly properties. As the ZapA protein has been shown to induce lateral FtsZ association in vitro, we wondered whether an increase in ZapA concentration in the cell would enable FtsZ(Ts1) to complete the helix-to-Z ring transition, thereby restoring division at the non-permissive temperature.

To test this, we introduced a second copy of the zapA gene onto the ts1 chromosome by transforming ts1 cells (strain SU111) with genomic DNA from strain FG356 (Gueiros-Filho and Losick, 2002). FG356 harbours a xylose-inducible zapA gene at amyE in addition to the native copy of zapA. Transformants carrying the Pxyl-zapA gene were selected at 30°C or 49°C on media containing either glucose, a known repressor of the B. subtilis Pxyl promoter (Kraus et al., 1994), or xylose. Transformants arose readily in the presence of either glucose or xylose at 30°C, but only in the presence of xylose at 49°C, suggesting that ZapA overproduction from Pxyl restores viability to ts1 cells at the restrictive temperature. We confirmed the ftsZ(ts1) sequence for one of the transformants and designated this strain SU497. We then verified that ZapA overproduction rescues viability at 49°C by measuring the plating efficiency of SU497 (Pxyl-zapA), which we defined as the ratio of colonies forming at 49°C versus 34°C after serially diluting the same mid-exponential culture onto separate plates (see Experimental procedures). A plating efficiency of 0.007 ± 0.001% was observed on glucose-containing plates, while in the presence of xylose this value rose to 90 ± 11%. In comparison, a wild-type control strain (SU110) and the temperature-sensitive ts1 strain (SU111) exhibited plating efficiencies of 106 ± 11% and 0.0008 ± 0.0006%, respectively, both assayed in the absence of added glucose or xylose. These results indicate that ZapA overproduction suppresses the thermosensitivity of ts1.

In culture, the SU497 (Pxyl-zapA) strain and its parent (ts1, strain SU111) grew with similar mass doubling times in the presence of either glucose or xylose. In media containing 1% glucose, as expected SU497 (Pxyl-zapA) cells were blocked for division at 49°C and formed long septum-less filaments (Fig. 3A) equivalent in length to the ts1 strain (SU111) grown under the same conditions. In the presence of xylose, however, SU497 (Pxyl-zapA) cells were much shorter than the ts1 mutant (9.4 ± 0.8 μm versus 47 ± 2 μm after 1 h at 49°C; Fig. 3B) and exhibited clear division septa (see later), confirming that ZapA overproduction rescues cell division in ts1 at the restrictive temperature. The rescued cells still remained around threefold longer than an isogenic wild-type strain (SU110) grown under the same conditions (3.4 ± 0.8 μm; Fig. 3C), indicating that cell division is not fully restored to a wild-type efficiency. Nonetheless, these results demonstrate that ZapA overproduction from the Pxyl promoter relieves the total division block imparted by the ts1 mutation at 49°C and markedly reduces cell length.

Figure 3.

ZapA overproduction rescues cell division and Z ring formation in the ts1 strain. SU497 [ftsZ(ts1), amyE::Pxyl-zapA] cells were grown in the presence of 1% glucose or 1% xylose to mid-exponential phase at 34°C, shifted to 49°C for 1 h and visualized by phase-contrast microscopy (A–C) or immunostained for FtsZ localization (D–I).
A and D. SU497 cells in 1% glucose.
B and E. SU497 cells in 1% xylose. Arrows point to intense, well-defined FtsZ(Ts1) structures, with the open arrow indicating a Z ring and filled arrows pointing to ‘non-ring’ structures.
C. Wild-type control cells (strain SU110) prepared in an identical fashion to those in B.
F–H. Enlarged images of a ring (F), a helix (G) and an arc (H) observed in the presence of xylose at 49°C.
I. ts1 control cells (SU111) grown at the permissive temperature (34°C), showing normal midcell Z rings. Scale bars, 5 μm.

To investigate the effect of ZapA on FtsZ(Ts1) assembly in vivo, we used immunofluorescence microscopy (IFM) to examine the localization of FtsZ(Ts1) in cells overproducing ZapA. IFM was performed on the SU497 (Pxyl-zapA) and SU111 (ts1 parent) strains after growth for 1 h at 49°C. Cultures were supplemented with 1% glucose or 1% xylose. As expected, SU111 (ts1) cells, irrespective of the presence of glucose or xylose, were indistinguishable from cells of the SU497 (Pxyl-zapA) strain grown in glucose-containing media to repress Pxyl. FtsZ(Ts1) localized at fairly regular intervals along these filamentous cells (Fig. 3D), in the spaces between segregated nucleoids (data not shown). As has been reported previously for ts1 (Michie et al., 2006), these localizations appeared as diffuse bands by IFM, distinct from the more highly resolved helical structures visible in live cells using an FtsZ(Ts1)–YFP fusion (see Fig. 2F). Notably, Z rings were never observed under these conditions.

When ZapA overproduction was induced in the SU497 (Pxyl-zapA) strain at 49°C, a distinctly different FtsZ(Ts1) localization pattern was detected. Although a number of faint, diffuse localizations, similar to those in the uninduced cells, could still be seen (Fig. 3E), a second population of much more brightly stained, well-defined FtsZ(Ts1) structures were also observed [Fig. 3E (arrows), F–H]. Importantly, a significant proportion of this latter group (30% of 405 localizations scored) comprised normal-looking Z rings [Fig. 3E (open arrow) and F], identical to those seen in isogenic wild-type cells and in the ts1 strain after growth under permissive conditions (Fig. 3I). This indicates that the mutant FtsZ(Ts1) protein regains the ability to form Z rings at 49°C following an increase in the intracellular level of ZapA, which in turn suggests that ZapA is able to stimulate lateral FtsZ association in vivo to promote the assembly of the Z ring. This result also provides yet further evidence for a lateral interaction FtsZ(Ts1) defect, under conditions which are more physiologically relevant than the presence of high Ca2+ concentrations.

Some non-ring FtsZ(Ts1) assemblies can facilitate cell division at 49°C upon ZapA overproduction

In addition to normal-looking rings, a large number of ‘non-ring’ FtsZ(Ts1) localizations were observed in ZapA overproducing ts1 cells that could be readily distinguished from the diffuse patterns normally seen in the ts1 strain. These accounted for 70% of the intense, well-defined localizations detected at 49°C (see above) and included helix-like structures (Fig. 3G), arcs (Fig. 3H) and other non-ring FtsZ(Ts1) assemblies. Intriguingly, our results suggest that some of these non-ring FtsZ(Ts1) structures are capable of directly supporting cell division at 49°C, complementing division rescue from the FtsZ(Ts1) rings. First, during a simple morphological analysis of SU497 (Pxyl-zapA) cells, we noticed that after induction with 1% xylose many division septa had an abnormal appearance [Fig. 4A (filled arrow), B–E]. These abnormal septa accounted for almost half of all division septa observed at 49°C (48% of 407 scored). Abnormal septa were also detected after ZapA overproduction at 34°C, albeit at a much lower frequency [5% of 870 scored; Fig. 4G (white box)]. Interestingly, no abnormal FtsZ localizations or septa were observed when ZapA was overproduced in isogenic wild-type cells (strain SU498; see Table 1), either at 34°C or 49°C, indicating that these phenotypic effects are not simply a general consequence of elevated ZapA levels (data not shown). Careful examination of the abnormal septa, by imaging a series of transverse sections throughout the cell, revealed a twisted, possibly helical morphology. This is evident in the representative images in Fig. 4B–E, which show a single focal plane. Importantly, similar division septa have been reported in previous studies (Bi and Lutkenhaus, 1992; Addinall and Lutkenhaus, 1996; Stricker and Erickson, 2003; Feucht and Errington, 2005), and are thought to form through the constriction of helical or non-ring FtsZ assemblies, which direct invagination of the cell envelope in an abnormal pattern. Indeed, the non-ring FtsZ(Ts1) structures observed in SU497 (Pxyl-zapA) cells at 49°C were often found to localize at the sites of abnormal septal invagination (Fig. 4F). In contrast, the Z rings and diffuse FtsZ(Ts1) localizations never occurred at such sites. We suggest therefore that non-ring FtsZ(Ts1) assemblies are responsible for generating the 48% of abnormal division septa observed in ZapA overproducing ts1 cells at 49°C. The remaining 52% of septa, which on the other hand exhibit an apparently wild-type morphology, would be likely to form through normal FtsZ(Ts1) ring structures. This was subsequently confirmed using time-lapse microscopy (see below).

Figure 4.

Septal morphology in ZapA overproducing ts1 cells. Strain SU497 [ftsZ(ts1), amyE::Pxyl-zapA] was grown in the presence of 1% xylose and visualized by phase-contrast microscopy or immunostained for FtsZ localization.
A. Phase-contrast image of SU497 cells after 1 h at 49°C. The open arrow points to the remnants of a normal septum, while the filled arrow indicates a septum with an abnormal morphology.
B–E. Enlarged images of abnormal septal invaginations observed by phase-contrast at 49°C. In C and E, the division process is nearing completion as the newborn cells have almost separated.
F. IFM analysis of FtsZ(Ts1) localization at the site of abnormal invagination, again at 49°C. Panel i shows a phase-contrast image of an immunostained cell, with panel ii displaying the FtsZ(Ts1) staining.
G. Phase-contrast image of SU497 cells grown at 34°C. A single abnormal septum is shown within the white box and magnified inset. Scale bars, 5 μm for A, F and G.

Table 1.  B. subtilis strains.
StrainGenotypeaReference/source/constructionb
  • a. 

    Antibiotic resistance genes are expressed as follows: cat, chloramphenicol; erm, erythromycin; spec, spectinomycin; tet, tetracycline.

  • b. 

    Arrows indicate construction by transformation of chromosomal DNA (donor strain → recipient strain).

SU5168 trpC2E. Nester
SU110168 thyA thyB trpC2 Callister et al. (1983)
SU111SU110 ftsZ(ts1) Callister et al. (1983)
SU489SU110 amyE::Pxyl-ftsZ(ts1)-yfp (spec) ftsZ(ts1) Michie et al. (2006)
SU497SU110 amyE::Pxyl-zapA (cat) ftsZ(ts1)FG356 → SU111
SU498SU110 amyE::Pxyl-zapA (cat)FG356 → SU110
SU514SU110 zapA-yshBΔ::tet ftsZ(ts1)FG356 → SU111
SU517SU110 zapA-yshBΔ::tet amyE::Pxyl-gfp-zapA (cat) ftsZ(ts1)FG347 → SU514
SU518SU110 amyE::Pxyl-gfp-zapA (cat)FG346 → SU110
SU519SU110 amyE::Pxyl-gfp-zapA (cat) ftsZ(ts1)FG347 → SU111
SU573SU5 amyE::Pspachy-minCD (cat)PL1138 → SU5
FG347PY79 amyE::Pxyl-gfp-zapA (cat) Gueiros-Filho and Losick (2002)
FG356PY79 Δ(zapA-yshB)::tet amyE::Pxyl-zapA (cat) thrC::Pspac-yshB (erm) Gueiros-Filho and Losick (2002)
PL1138PY79 amyE::Pspachy-minCD (cat) Levin et al. (2001)

Time-lapse analysis of Z ring formation and division in ZapA overproducing ts1 cells

Our static images of FtsZ(Ts1) localization, while clearly showing that ZapA overproduction rescues Z ring formation in the ts1 strain at 49°C, fail to definitively resolve several important questions: (i) Do these Z rings form through the normal helix-to-ring assembly pathway? (ii) Are they functional for cell division? (iii) Are ‘non-ring’ FtsZ(Ts1) structures also able to facilitate division under these conditions? To address these issues, a time-lapse analysis of live, growing cells was required.

Importantly, prior to commencing these time-lapse experiments, we observed that a functional GFP-ZapA fusion protein, expressed under xylose-inducible control in ts1 cells, localized in an identical fashion to FtsZ(Ts1), and rescued both Z ring formation and cell division at 49°C when overproduced using 1% xylose (see Supporting information and Fig. S2). The GFP-ZapA fusion had previously been shown to be functional in wild-type B. subtilis cells, and to localize to the Z ring in a manner specifically dependent on its interaction with FtsZ (Gueiros-Filho and Losick, 2002). With this knowledge in hand, we decided that for our time-lapse experiments we would rescue the ts1 strain through overproduction of GFP-ZapA, and use the fusion as a marker for FtsZ(Ts1) localization. We preferred to label the ZapA protein rather than directly tagging FtsZ(Ts1) to minimize the problem of photobleaching, which we have encountered in previous time-lapse studies with fluorescent protein fusions to FtsZ. Unlike FtsZ fusions, which are non-functional and must be expressed at a low intracellular concentration to prevent aberrant effects on the assembly of native FtsZ (Ma et al., 1996), a functional GFP-ZapA fusion could theoretically be induced at much higher levels, thereby increasing the number of available fluorophores in the cell.

Bacillus subtilis strain SU519 contains the Pxyl-inducible gfp-zapA construct at the amyE locus, providing a second copy of zapA in the ts1 genetic background. For time-lapse experiments, SU519 (GFP-ZapA) cells were grown to mid-exponential phase at 34°C in the presence of 1% xylose, then shifted to 49°C for 30 min. At this time, cells were mounted onto pre-warmed agarose pads and images of the GFP-ZapA fusion were collected at 90 s intervals over a 45 min period using a heated microscope stage set to 49°C. Ten separate time-lapse experiments were performed, in which significant growth (cell length extension) and division could be observed for ∼100 cells. Initially, a total of 15 clear ring structures were detected in these cells. Over the course of the time-lapse analysis, almost all of these were found to constrict, gradually reducing in diameter to produce a single focus before disappearing altogether from the cell (Fig. 5 and Fig. S3). Constriction was usually followed by the appearance of an obvious division septum, which could be detected under phase-contrast (arrow in Fig. 5, image 45ii). This confirms the ability of FtsZ(Ts1) rings to support cell division at 49°C upon ZapA overproduction.

Figure 5.

Time-lapse images depicting normal Z ring formation and constriction in ts1 cells overproducing GFP-ZapA at 49°C. SU519 [ftsZ(ts1), amyE::Pxyl-gfp-zapA] cells were collected after 30 min growth in the presence of 1% xylose at 49°C and visualized by microscopy using a heated stage set to the same temperature. Images were collected at 90 s intervals over a 45 min period. The first and last frames shown in the sequence were obtained by phase-contrast, while all other frames represent GFP-ZapA localization, used as a marker for FtsZ(Ts1). Numbers correspond to time (min) during time-lapse image acquisition. Arrows mark the progress of a midcell Z ring (images 0ii and 15) that undergoes constriction (image 24), followed by the release of fluorescent signal from the ring (image 27) and the formation of two new rings (image 45i) via dynamic helical intermediates (image 33). In the final frame, an arrow highlights the appearance of a normal-looking division septum at the site of constriction of the initial Z ring. Scale bar, 5 μm. The movie can be viewed in Supporting information (Fig. S3).

Importantly, during the process of constriction, the fluorescent signal from the ring localizations was observed to emanate out on both sides of the ring as a series of dots. These dot patterns were highly mobile, with the distribution of foci changing in each successive image, and were suggestive of a dynamic helical assembly. Eventually, many of these helix-like patterns appeared to give rise to new ring structures, first becoming restricted to a small region of the cell and then rapidly reorganizing into a tight band (ring). This can be seen in Fig. 5 (and Fig. S3), which shows the constriction of a ring localization (Fig. 5, images 0ii, 15 and 24) followed by the formation of new rings (arrows in image 45i) on each side through short-lived helical intermediates (arrows in image 33). In fact, all new ring formations observed during our time-lapse experiments were immediately preceded by dynamic helix-like dot patterns. These findings are fully consistent with the recent work of Peters et al. (2007), in which a similar helix-to-ring progression was detected for FtsZ during the wild-type B. subtilis cell cycle (see Introduction). In addition, we have observed equivalent results using wild-type cells overproducing the GFP-ZapA fusion (strain SU518; data not shown). Together, these findings suggest not only that ZapA rescues FtsZ(Ts1) ring formation at 49°C, but that it does this by enabling FtsZ(Ts1) to complete the normal helix-to-ring assembly pathway that occurs in wild-type B. subtilis.

Our time-lapse experiments also allowed us to examine the process of abnormal septum formation that occurs following ZapA overproduction in ts1 at 49°C. In 33 cells, we observed what appeared to represent a twisted or abnormal invagination of the cell envelope (Fig. 6 and Fig. S4). In each case, an intense, non-ring GFP-ZapA localization could be detected at the site of invagination (Fig. 6 and Fig. S4). These localizations were quite irregular in appearance, and seemed to alter their conformation according to the changes in cell morphology (Fig. 6). This further supports the idea that when overproduced in the ts1 strain, ZapA can facilitate cell division through non-ring FtsZ(Ts1) structures.

Figure 6.

Time-lapse images depicting abnormal septal invagination in ts1 cells overproducing GFP-ZapA at 49°C. SU519 [ftsZ(ts1), amyE::Pxyl-gfp-zapA] cells were collected after 30 min growth in the presence of 1% xylose at 49°C and visualized by microscopy using a heated stage. Images were collected at 90 s intervals over a 45 min period. Panels labelled i depict phase-contrast images while those labelled ii show the corresponding GFP-ZapA localization patterns. Arrows provide a point of reference for comparing phase-contrast and fluorescence images. Numbers correspond to time (min) during time-lapse image acquisition. Scale bar, 2 μm. The movie can be viewed in Supporting information (Fig. S4).

Discussion

In this study, we address the problem of how Z rings form inside bacterial cells by exploiting a unique ftsZ mutant of B. subtilis, ts1. At high temperatures, the mutant FtsZ(Ts1) protein appears to become trapped as a short helical intermediate of Z ring formation in vivo, being unable to progress into a functional ring (Michie et al., 2006). To determine why this is the case, and in doing so gain new insight into the assembly mechanism of the Z ring in wild-type cells, we used a combination of biochemical, cytological and genetic methods to identify the molecular defect of the FtsZ(Ts1) protein. Our results indicate that FtsZ(Ts1), both in vivo and as a purified protein, is primarily defective in its ability to establish or maintain lateral interactions between protofilament strands at high temperatures.

Significantly, this suggests for the first time that lateral FtsZ association is directly involved in the transition between the FtsZ helix and the midcell Z ring in wild-type bacterial cells. Based on this finding, we present a simple model for the final phase of Z ring formation (remodelling of the short helical intermediate into a ring at midcell), in which lateral FtsZ interactions would help to drive the remodelling process and stabilize the final ring structure (Fig. 7). In developing this model, we have taken into account the results of recent high-resolution microscopy studies, including electron cryotomography work (Li et al., 2007), which suggests that both the Z ring and the FtsZ helix in vivo are composed of a discontinuous and somewhat erratic arrangement of short FtsZ filaments. In the case of the helix, evidence suggests that these filaments may be arranged onto a helical lipid scaffold within the cell membrane, composed of an ordered arrangement of specific phospholipids (Barák et al., 2008). We propose that at a particular stage of the cell cycle, perhaps in response to an as yet unidentified regulatory signal at midcell, the formation of lateral contacts between protofilaments in the helix simply compresses the helical structure at the cell centre, effectively producing a ‘ring’ composed of a large number of short, laterally associated FtsZ filaments (see Fig. 7). Presumably, this process would be accompanied by a remodelling of the helical lipid scaffold in the vicinity of midcell (Fig. 7), which could itself be powered by lateral FtsZ association. In ts1 cells, the lateral interaction defect of FtsZ(Ts1) would effectively block this final stage in the Z ring assembly pathway.

Figure 7.

Model for the final phase of Z ring assembly.
A. Immediately prior to the appearance of the Z ring, FtsZ filaments (black lines) are concentrated predominantly in the vicinity of midcell, where they are arranged onto a helical lipid scaffold within the cell membrane (grey and white zig-zag) to form a ‘short’ FtsZ helix. It is important to note that the lipid helix itself extends throughout the entire cell, and a small number of FtsZ filaments remain associated with the helix across its length. In response to some regulatory signal at midcell, lateral connections are induced between FtsZ filaments in the short helix, which ultimately compresses the short helical structure into a ring at the cell centre (B). These lateral interactions could occur between FtsZ filaments in close proximity to one another within the short helix, triggering a reduction in the overall pitch of the helical structure at midcell. Alternatively, lateral association could occur between FtsZ filaments in different regions of the helix, essentially collapsing the helical structure into a tight ring at the cell centre.

Interestingly, we found that the ts1 defect could be rescued in vivo through an increase in the intracellular concentration of ZapA, enabling the FtsZ(Ts1) protein to complete the normal Z ring assembly pathway at 49°C and restoring cell division. These results provide the first evidence that ZapA is able to stimulate lateral FtsZ association directly within the bacterial cell. By bringing FtsZ filaments together, our findings suggest that ZapA functions to promote the helix-to-ring transition of FtsZ in vivo. Intriguingly, we have found that overproduction of the other known positive modulators of FtsZ assembly in B. subtilis (FtsA and SepF), under the Pxyl promoter, does not suppress ts1 thermosensitivity. Similarly, deletion of the FtsZ inhibitors EzrA, ClpX and MinC fails to rescue the ts1 strain (data not shown). This suggests that the cellular role of ZapA is unique among the known FtsZ accessory proteins, which is particularly intriguing given that SepF, like ZapA, has recently been shown to exhibit FtsZ bundling activity in vitro (Singh et al., 2008). It is important to remember that ZapA is dispensable in wild-type cells under standard laboratory conditions (Gueiros-Filho and Losick, 2002). This means that additional (perhaps as yet unidentified) proteins may also be able to fulfil the role of ZapA in the cell, or that ZapA may only be required under conditions of stress in which FtsZ assembly and the helix-to-ring transition are perturbed.

As well as ZapA, we found that the lateral interaction defect of FtsZ(Ts1) could be rescued in vivo by the addition of Ca2+ to the growth medium. While this is a very interesting finding, it is important to note that our results do not demonstrate conclusively the mechanism by which Ca2+ actually promotes lateral FtsZ association in the B. subtilis cell. The cation may act directly on the FtsZ protein, or stimulate protofilament bundling indirectly via an effect on the lipid membrane for example. However, given that Ca2+ can induce lateral FtsZ association and rescue the FtsZ(Ts1) defect directly in vitro, we favour a model in which it acts specifically on FtsZ within the cell. If true, this could explain why very high Ca2+ concentrations (much higher than would normally be encountered in the environment) were required in the growth media to rescue the ts1 strain. Presumably, such high concentrations would be needed to overcome the tight homeostatic mechanisms that normally maintain a low basal concentration of Ca2+ in the cytoplasm, even in the presence of much higher external Ca2+ levels (Gangola and Rosen, 1987; Herbaud et al., 1998; Naseem et al., 2009). A limited influx of Ca2+ into the cytoplasm could also account for the fact that the cation did not restore the ts1 phenotype completely back to wild type. Importantly, regardless of the mechanism, our results demonstrate that the addition of Ca2+ to cell growth media does enhance lateral FtsZ association in vivo (see also Supporting information). This could prove to be a valuable tool in future studies.

How might the mutations in FtsZ(Ts1) cause a defect in lateral association at high temperatures? The FtsZ(Ts1) protein harbours two alanine-to-valine substitutions within its C-terminal globular domain: an A240V mutation responsible for ts1 thermosensitivity, and a secondary A278V mutation that has been found to partially compensate for the effects of A240V (Michie et al., 2006). These sites lie close to one another within the FtsZ structure, on the interior face of two adjacent helices, designated H9 and H10. Interestingly, helix H9 (which harbours the A240V substitution) is located on the lateral surface of the FtsZ protofilament, remote from sites known to be associated with GTP hydrolysis and the head-to-tail polymerization of FtsZ units (see Fig. S5 and Oliva et al., 2004). The A240V mutation presumably alters the packing of helix H9 against the rest of the structure, thus altering the conformation of the FtsZ surface in this region. It is possible that this disrupts an important interaction site in FtsZ(Ts1), resulting in diminished lateral associations between protofilaments. The alanine at position 240 is extremely well conserved even among distantly related organisms (Michie et al., 2006), consistent with a role for this residue in forming a crucial lateral interaction site. Further analysis of the FtsZ(Ts1) fold could provide important structural insights into the mechanism of lateral FtsZ association, which at present is very poorly understood.

Intriguingly, Lan and colleagues have recently proposed, on the basis of computational modelling data, that the role of lateral FtsZ association may extend to the constriction of the Z ring and the generation of mechanical force during cell division (Lan et al., 2009). Significantly, the results of our ZapA overproduction study provide some experimental support for this idea. Following an increase in ZapA concentration in ts1 cells, we observed that a large number of FtsZ(Ts1) structures were able to facilitate division without adopting the typical ring conformation. These structures always appeared thicker and stained much more brightly than the non-functional FtsZ(Ts1) assemblies normally observed in the ts1 strain at 49°C, suggesting a ZapA-mediated increase in lateral FtsZ(Ts1) association. While it is now clear that lateral association is required for normal Z ring assembly in vivo, this raises the possibility that it is not the formation of a ring per se, but the lateral interactions of FtsZ filaments within the ring that determines the ability of FtsZ to facilitate bacterial cell division.

Experimental procedures

General methods

DNA manipulation was carried out using standard techniques (Sambrook and Russel, 2001). Platinum Pfx (Invitrogen) or Taq (New England Biolabs) DNA polymerases were used for polymerase chain reactions (PCRs). B. subtilis chromosomal DNA was purified as described previously (Errington, 1984). DNA sequencing was performed by the Australian Genome Research Facility (Brisbane, Australia).

Overproduction and purification of FtsZ and FtsZ(Ts1)

The B. subtilis ftsZ and ftsZ(ts1) alleles were amplified by PCR from chromosomal DNA of strains SU110 and SU111, respectively, using primers 5′-GCACATGTTGGAGTTCGAAACAAAC-3′ (PciI restriction site shown in bold) and 5′-CGGGATCCTTAGCCGCGTTTATTACG GT-3′ (BamHI site in bold). PCR products were digested with PciI and BamHI, then ligated into vector pET-15b (Novagen) at the BamHI and NcoI sites (removing the His tag sequence) to produce pET-FtsZ and pET-FtsZ(Ts1). Ligated plasmids were transformed into Escherichia coli DH5α and correct clones were identified by sequencing. Purified pET-FtsZ and pET-FtsZ(Ts1) were then transformed into E. coli BL21 (harbouring plasmid pLysS; Studier et al., 1990) to give strains that produce untagged B. subtilis FtsZ or FtsZ(Ts1) upon induction with IPTG.

Cultures for production of FtsZ and FtsZ(Ts1) were grown at 30°C in L broth supplemented with ampicillin (100 μg ml−1), chloramphenicol (25 μg ml−1) and 1% glucose. Cultures were induced with 0.5 M IPTG when A600 reached ∼0.8. Four hours after induction, cells were harvested by centrifugation and pellets were stored at −80°C. Cell pellets collected from a total of 5 l of culture were resuspended in 100 ml of lysis buffer [50 mM Tris-HCl (pH 8.0), 50 mM KCl, 10 mM MgCl2, 1 mM EDTA, 10% glycerol]. Phenylmethylsulfonyl fluoride (1 mM), 2 μg ml−1 DNase I and 100 μg ml−1 lysozyme were then added, followed by three freeze-thaw cycles. Insoluble material was removed from cell lysates by centrifugation at 48 000 g for 20 min at 4°C. FtsZ was then selectively precipitated from the supernatant by adding ammonium sulphate to 40% saturation. Precipitated protein was collected by centrifugation at 20 000 g for 20 min at 4°C and resuspended in buffer [50 mM morpholineethanesulfonic acid (MES; pH 6.5, adjusted with NaOH), 10% glycerol, 5 mM MgCl2, 150 mM KCl]. After dialysis against the same buffer, nucleic acids were removed by passing the samples through a Toyopearl DEAE-650 column (55 ml bed volume; Tososh Bioscience) and collecting the protein in the flow through. After diluting the samples 1:1 with buffer containing 50 mM MES (pH 6.5, adjusted with NaOH), 5 mM MgCl2 and 10% glycerol to reduce the salt concentration, ion-exchange chromatography was performed using a MonoQ 10/100 column (8 ml bed volume; GE Healthcare), eluting with a 0–500 mM KCl gradient delivered over 20 column volumes. FtsZ and FtsZ(Ts1) eluted between 200 and 300 mM KCl. The samples were judged to be > 95% pure using SDS-PAGE. FtsZ protein concentration was determined by the BCA protein assay (Pierce) using bovine serum albumin as standard. Aliquots of protein (5–10 mg ml−1; total yield 30–50 mg) were snap frozen and stored at −80°C.

Negative stain electron microscopy

FtsZ and FtsZ(Ts1) were suspended at 0.6 mg ml−1 in 200 μl of 50 mM MES (pH 6.5, adjusted with NaOH), 5 mM MgCl2, 10% glycerol, with or without 20 mM CaCl2, and warmed to the desired reaction temperature. Note that the final suspension also contained KCl (from the protein preparation), at a concentration of ∼10 mM. Following the addition of GTP to 2 mM, samples were incubated for 2 min. Aliquots (20 μl) of the polymer solution were then applied to formvar/carbon-coated copper 300 mesh electron microscope grids (ProSciTech) for 1 min. Excess liquid was blotted off and grids were stained by floating on a drop of 2% uranyl acetate for 1 min. Images were collected with a Philips CM120 Biofilter transmission electron microscope.

Bacillus subtilis growth conditions

Bacillus subtilis strains were grown on tryptose blood agar plates or in L broth, except for time-lapse microscopy experiments in which Spizizen minimal medium (see Peters et al., 2007) was used. All media were supplemented with thymine (20 μg ml−1). CaCl2, MgCl2 or NaCl was added when required, at concentrations specified in the text. Glucose or xylose was also added when required, at a concentration of 1% unless otherwise noted. Antibiotics were used at the following concentrations: chloramphenicol, 5 μg ml−1; spectinomycin, 80 μg ml−1; tetracycline, 20 μg ml−1. Cells were grown at 34°C or 49°C, except for the selection of transformants which was performed at 30°C unless otherwise specified.

Strain construction

Bacillus subtilis strains used in this study are listed in Table 1 along with a summary of their construction where relevant. All double crossover integrations into the amyE locus were verified as described by Lewis and Marston (1999).

Viability assays

To assess viability at the non-permissive temperature for ts1 (49°C), strains were first grown to mid-exponential phase (A600∼0.4) at 34°C and serially diluted from 10−1 to 10−6. Equal volumes were then plated at 34°C and 49°C. Plating efficiency was calculated as the ratio of numbers of colonies forming at 49°C versus 34°C.

Microscopy and image analysis

For examination of cellular morphology and length, cells were fixed in 70% ethanol as described previously (Hauser and Errington, 1995) and visualized by phase-contrast microscopy. IFM was performed according to the method of Harry et al. (1999) but without the addition of glutaraldehyde. Rabbit polyclonal anti-FtsZ serum was used at a dilution of 1:10 000 to bind FtsZ, while goat-anti-rabbit IgG conjugated to Alexa 488 (Molecular Probes) was used at a 1:10 000 dilution for detection. Live cell microscopy involved placing cells on a 2% agarose pad made up in growth medium and prepared within a 65 μl Gene Frame (AB Genes). For cells grown at 49°C, slides were pre-warmed to 49°C in a humidified chamber and images were collected immediately. To visualize DNA, 4′6-diamidino-2-phenylindole (DAPI) was added directly to the fixation solution for IFM or to the culture medium (2 min prior to collection) for live cell microscopy, to a final concentration of 0.2 μg ml−1.

Cells were viewed using a Zeiss Axioplan 2 fluorescence microscope equipped with a 100× Plan ApoChromat phase-contrast objective (Zeiss), an AxioCam MRm cooled CCD camera and the following filter blocks: Filter Set 09 (Zeiss), for visualizing Alexa 488; Filter Set 41018 (Chroma Technology), for GFP; Filter Set 41029 (Chroma Technology), for YFP; Filter Set 02 (Zeiss), for DAPI. Images were collected using AxioVision software, versions 4.4 and 4.5 (Zeiss), and prepared for publication with Adobe Photoshop CS version 8.0 (Adobe Systems). Cell length measurements were recorded using Axiovision (Zeiss), while statistical analysis was performed in Excel (Microsoft). Time-lapse studies were performed using a heated microscope stage equipped with an objective heater, heatable universal mounting frame and an achromatic condenser (Zeiss), set at 49°C; 2 × 2 binning and a gain factor of 3 were applied for the acquisition of all time-lapse images.

Acknowledgements

We thank Kate Michie for advice regarding the expression and purification of FtsZ(Ts1) as well as insightful discussion on the structural consequences of the ts1 mutations. We are also grateful to Nick Dixon for assistance with protein purification, Shigeki Moriya and Naotake Ogasawara for the provision of FtsZ antibodies, and Frederico Guieros-Filho for strains FG347 and FG356. Finally, we acknowledge the facilities as well as technical assistance from staff at the Electron Microscopy Unit, University of Sydney. This work was supported by an Australian Research Council Discovery Project Grant (DP0450770) to E.J.H.

Ancillary