Strains and plasmids used in this study are listed in Table 1.

B. subtilis strains were grown at 37°C in competence medium (Hamoen et al., 2002). When required, xylose and IPTG were used as inducers at concentrations of 0.2–0.5% and 1 mM respectively. Selection of transformants and short term maintenance of B. subtilis strains were performed on nutrient agar (Oxoid), supplemented when required with 10 μg/ml tetracycline, 5 μg/ml chloramphenicol, 50 μg/ml spectinomycin, 5 μg/ml kanamycin, 1 μg/ml phleomycin, or 0.5 μg/ml erythromycin with 25 μg/ml lincomycin.

Plasmid pRD158 was constructed by inserting a PCR derived fragment of ftsW (also named ylaO) into plasmid pSG441, and carries a P_spac promoter located immediately in front of the RBS of the FtsW coding sequence. The ftsW fragment was amplified with the oligonucleotides W1 (5′-ATATCCTTCCCCTGTACAC-3′) and W2 (5′-ATATCCTTCCCCTGTACAC-3′), digested with XbaI and EcoRV and ligated to plasmid pSG441 previously digested with XbaI and SphI and blunted using Klenow treatment.

Strain RD158 was then generated by Campbell integration of pRD158 into strain 168, selecting for kanamycin resistance in the presence of IPTG. The resulting transfomants were screened to confirm that they were dependent upon IPTG for viability and the location of the plasmid integration confirmed using PCR across the site of integration. One such transformant that exhibited the correct phenotype and had the correct genetic construction was then designated RD158.

Strain 4371 was generated by the integration into strain 168 of plasmid pSG1151-FtsW-gfp carrying a 3′-fragment of ftsW fused to the gfp gene. Campbell integration of the plasmid resulted in the expression of GFP fused to the C-terminus of FtsW under the native promoter. The correct integration of this plasmid was confirmed by PCR amplification of the site of insertion and sequencing across the fusion to ensure that the coding sequence of FtsW and GFP were joined as expected.

Overnight cultures grown at 37°C in competence medium were diluted 80-fold in the same medium supplemented with the appropriate inducers and grown at 37°C until they reached an O.D.₆₀₀ of ~0.3. Cells were then diluted 80 times in pre-warmed competence medium, in the presence or in the absence of the inducer. Incubation was continued at 37°C until cultures reached an O.D.₆₀₀ of ~0.3 (about 4 h), at which point cultures were re-diluted (40-fold) in the same pre-warmed medium.

B. subtilis spores were prepared and germinated as previously described (Gamba et al., 2009). Briefly, strains were grown at 37°C in liquid sporulation medium until they reached an O.D.₆₀₀ of ~0.6 and then plated on sporulation agar plates after four serial five-fold dilutions. Plates were incubated at 30°C for 5–7 days. Spores were then scraped off the plates, washed in sterile water, and suspended in TE buffer (10 mM Tris-Cl pH 8.0, 1 mM EDTA) supplemented with 1.5 mg/ml lysozyme and incubated at 37°C for 1 h with shaking. Sodium dodecyl sulfate (SDS) was added to a final concentration of 5%, and the suspensions were incubated at 37°C for 30 min. Spores were then collected by centrifugation and washed four times with sterile water prior to being stored in water at 4°C. Spore preparations were subsequently washed twice a day for a week to avoid spontaneous germination.

For germination experiments, spores were diluted in germination medium (Hamoen and Errington, 2003), heat shocked for 30 min at 70°C, and quickly cooled on ice. The spore mixtures were then diluted in germination medium to an O.D.₆₀₀ of 0.3 in the presence or absence of the appropriate inducers, and incubated at 37°C with continuous shaking. Samples were collected for microscopy 3, 4, or 5 h after the beginning of the heat shock treatment (t = 0).

Samples were mounted on microscope slides coated with a thin layer of 1.2% agarose. Images were acquired with a Zeiss Axiovert 200M or a Zeiss Axiovert 135 microscope coupled to a Sony Cool-Snap HQ2 cooled CCD camera (Roper Scientific) and using Metamorph imaging software (Universal Imaging), or with a Deltavision microscope (Applied Precision) using a CoolSNAP HQ camera (Princeton Instruments) and softWorx imaging software (Applied Precision).

Spores of strain PG87 were germinated in the presence of 0.5% xylose, and in the presence or absence of 1 mM IPTG. Optical density was measured at regular intervals and samples were collected, spun down and flash frozen in liquid nitrogen 3, 4, or 5 h after the beginning of the heat shock (8, 4, and 4 ml volumes respectively). Cell pellets were suspended in 100 μl of 1 × NuPAGE LDS Sample Buffer (Invitrogen) supplemented with Complete mini protease inhibitor (Roche), and broken by sonication (3 × 10 s pulses at 20% amplitude). Samples were denatured at 70°C for 10 min and loaded onto an SDS-PAGE gel (NuPAGE Bis-Tris 4–12% gradient gels, run in MOPS buffer, ThermoFisher Scientific), such that the amount of sample loaded was normalized relative to the optical density (O.D_.600) measured at the time of collection. This allowed the comparison, at each time point, of the uninduced and induced sample, but not of samples taken at different time points, due to the different contribution that hydrated spores and spore coats gave to the total optical density. The resolved proteins were transferred onto a PVDF membrane (GE Healthcare) by using a wet procedure and Western blotting was performed according to standard methods. For the detection of PBP 2B, a 1:10,000 dilution of rabbit polyclonal anti-PBP2B serum was used. Anti-rabbit horseradish peroxidase-linked antiserum (Sigma) was used as secondary antibody at a dilution of 1:10,000. Protein bands were detected using an ImageQuant LAS 4000 mini digital imaging system (GE Healthcare). PVDF membranes were Coomassie stained after imaging to check for even protein transfer between induced and uninduced samples.

We constructed a strain in which FtsW could be depleted by placing the gene under the control of the IPTG inducible promoter P_spac (strain RD158). Microscopic analysis of depleted cells showed very long filaments without normal division septa (Figure 1), confirming that the gene is essential and is involved in cell division, as previously shown (Kobayashi et al., 2003; Suel et al., 2007).

Depletion of FtsW has been shown to have a strong effect on Z-ring stability in the Gram-positive organisms S. coelicolor and Mycobacterium tuberculosis (Boyle et al., 1997; Daniel et al., 1998; Henriques et al., 1998), and to test whether this was also the case in B. subtilis, the P_spac-ftsW construct was introduced into a strain expressing gfp fused to ftsZ (PG80, P_spac-ftsW kan amyE::P_xyl-gfp-ftsZ spc). This strain was grown in a minimal medium supplemented with 0.2% xylose to induce expression and allow Z-rings to be seen. Growth of the strain in the absence of IPTG was found to cause a cell division defect detectable after about 90–120 min (about 3–4 generations) of incubation (although a significant degree of heterogeneity was observed in the cell population). After prolonged depletion of FtsW (4–6 h, 7–10 generations), long filaments were clearly visible while DAPI staining still showed a normal localization pattern of the nucleoids (not shown). Under these conditions, some bright Z-rings were visible, but only one or two rings per filament (Figure 1A). As the GFP-FtsZ fusion is not active, we also tested the effect of FtsW depletion using a strain containing an EzrA-GFP fusion that is known to be biologically active (Daniel, unpublished). Again filamentous cells with only one or sometimes two fluorescent bands were observed (Figure 1C). Thus, it seems likely that the Z-rings/EzrA observed in these filamentous cells are the remains of the last divisome that was active prior to FtsW becoming limiting. To confirm this, we made use of germinating spores. The dehydrated kernel of B. subtilis spore is devoid of any cell division structure and therefore the morphogenesis of spores into vegetative cells is not influenced by a previous cell division event. When spores were germinated in the absence of IPTG, and thus without ftsW expression, one or two GFP-FtsZ or EzrA-GFP rings became visible (Figures 1E,G), but after prolonged growth in the absence of IPTG, the rings disappeared (Figures 1E,G). Control experiments carried out in the presence of IPTG permitted the completion of at least 2 to 3 rounds of division, and Z-ring were clearly localized in all cells at the end of the experiment (Figures 1F,H). To compare the localization pattern of FtsZ upon depletion of other late proteins, spores of strain PG80 (P_spac-ftsW kan amyE::P_xyl-gfp-ftsZ spc) where germinated in parallel with spores of strain PG114 (amyE::P_xyl-gfp-ftsZ, P_spac-yllB-ylxA-ftsL-pbpB), which allowed the depletion of the FtsL and PBP 2B. Here again Z-ring formation could be observed at the early stages of depletion, as was seen with ftsW, but 5 h after germination (sufficient time for ~3 cell cycles under normal conditions) all rings were disassembled (Figure S1). This shows that the stability of the Z-rings is similarly affected by the absence of any of the late assembling division proteins.

In germinating spores Z-rings disassembled after prolonged depletion of FtsW or PBP2B/FtsL. In these experiments this would correspond to the time needed for germination (~3 h) and then 2–3 cell divisions (~35 min per cell cycle), in all about 5 h incubation. However, when ftsW expression was re-induced by adding 1 mM IPTG to the cultures, FtsZ rings re-formed along the filaments. Multiple Z-rings per filament were visible 30 min after re-induction, although the spacing between them was rather irregular (Figures 2B,C). These rings eventually constricted and sites of septation could be detected 1 h after the addition of IPTG (Figure 2A). Similar results were also obtained using EzrA-GFP as a marker of the Z-ring formation (not shown).

In E. coli, recruitment of FtsI depends on FtsW. To test whether this is also the case in B. subtilis, the P_spac-ftsW construct was integrated into strains carrying gfp fusions to ftsL and pbpB, generating strains PG79 and PG87 respectively. Depletion of FtsW was carried out in vegetative cells and germinating spores. In vegetative cells, depletion of FtsW was accompanied by disappearance of GFP signal for both FtsL and PBP 2B (Figures 3A,C). However, this could also be caused by the reduced frequency of Z-ring localization that we have shown before. We therefore repeated the experiments in germinating spores and confirmed that the absence of IPTG did not allow the formation of GFP bands, indicating that, when FtsW was depleted, PBP 2B and FtsL localization was lost (Figures 3B,D). As FtsL is known to be unstable when division is perturbed (Daniel and Errington, 2000) we focused on the stability of PBP 2B and used Western blotting to show that depletion of FtsW did not result in the degradation of PBP 2B (Figure S2) and hence the loss of localisation was not due to degradation of the GFP fusion.

While the assembly of late division proteins follows multiple successive steps in E. coli, the recruitment of late proteins in B. subtilis seems to be a strongly cooperative process, with the late proteins PBP 2B, FtsL, DivIC, and DivIB being interdependent for localization (Errington et al., 2003). To study the hierarchy of recruitment of FtsW to the division site, a C-terminal GFP fusion to FtsW in its native locus (4731, ftsW-gfp) was introduced into a strain which allows the depletion of either PBP 2B or FtsL (Daniel et al., 1998), generating strain PG92. Strain PG92 was subsequently depleted for PBP 2B, by transferring it to a medium lacking IPTG. Cells became filamentous and the clear localization of FtsW-GFP at cell division sites disappeared (Figure 4A). More convincingly, no clear localization could be observed in germinating spores (Figure 4B). Strain PG92 allows also for the depletion of FtsL because ftsL is under the control of a xylose inducible promoter. While long filaments with no fluorescent signal could be detected in the absence of xylose (Figures 4C,D), this experiment could not be confirmed in germinating spores since the P_xyl promoter is leaky during germination (not shown), and therefore efficient depletion is not possible. However, since the localization of FtsW depends on PBP 2B, it is expected that it will also depend on the presence of FtsL. We conclude that the localization of PBP 2B, FtsL and FtsW are interdependent in B. subtilis.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.