A missense mutation in ftsZ differentially affects vegetative and developmentally controlled cell division in Streptomyces coelicolor A3(2)

Authors


Summary

Streptomyces coelicolor A3(2) undergoes at least two kinds of cell division: vegetative septation leading to cross-walls in the substrate mycelium; and developmentally regulated sporulation septation in aerial hyphae. By isolation and characterization of a non-sporulating ftsZ mutant, we demonstrate a difference between the two types of septation. The ftsZ17 (Spo) allele gave rise to a classical white phenotype. The mutant grew as well as the parent on plates, and formed apparently normal hyphal cross-walls, although with a small reduction in frequency. In contrast, sporulation septation was almost completely abolished, resulting in a phenotype reminiscent of whiH and ftsZΔ2p mutants. The ftsZ17 (Spo) allele was partially dominant and had no detectable effect on the cellular FtsZ content. As judged from both immunofluorescence microscopy of FtsZ and translational fusion of ftsZ to egfp , the mutation prevented correct temporal and spatial assembly of Z rings in sporulating hyphae. Homology modelling of S. coelicolor FtsZ indicated that the mutation, an A249T change in the C-terminal domain, would be expected to alter the protein on the lateral face of FtsZ protofilaments. The results suggest that cytokinesis may be developmentally controlled at the level of Z-ring assembly during sporulation of S. coelicolor A3(2).

Introduction

Cell division in prokaryotes occurs by a widely conserved mechanism that is shared by Bacteria, one branch of the Archaea (the Euryarchaeota), plant chloroplasts and the mitochondria of some unicellular Eukarya (Margolin, 2000; Osteryoung, 2001). In these organisms, the FtsZ protein polymerizes on the inner surface of the cytoplasmic membrane to form a ring-like structure – the Z ring – at the future division site (reviewed by Lutkenhaus and Addinall, 1997; Rothfield et al., 1999; Margolin, 2000). FtsZ is a tubulin homologue, and most of its three-dimensional structure and biochemical properties are similar to those of α- and β-tubulins (Erickson, 1998; Löwe and Amos, 1998; 1999; Nogales et al., 1998). The contractile Z ring determines the division plane and recruits other division proteins to this site. Eight additional proteins, FtsA, FtsI, FtsK, FtsL, FtsN, FtsQ, FtsW and ZipA, have been directly implicated in cell division in Escherichia coli, and all are known to localize to the division site in an FtsZ-dependent manner (Margolin, 2000). Their degree of conservation is, however, variable among bacteria, and sequenced microbial genomes contain different subsets of these division proteins (Margolin, 2000; Erickson, 2001). FtsZ has been shown to interact directly with a number of proteins that may have roles in nucleating or stabilizing Z rings, anchoring Z rings to the membrane, controlling Z-ring formation, linking the cytoplasmic Z ring to peptidoglycan-assembling enzymes with activity on the periplasmic side of the membrane or sensing and signalling the completion of septa to developmental pathways (Higashitani et al., 1995; Hale and de Boer, 1997; Wang et al., 1997; Din et al., 1998; Hu et al., 1999; Lucet et al., 2000).

The precision of the cell division process in rod-shaped bacteria such as E. coli and Bacillus subtilis is remarkable, but the underlying mechanisms that direct Z-ring formation at the right time to the precise mid-point of the cell are not fully understood. In E. coli, at least two negatively acting mechanisms contribute to the selection of a site for FtsZ assembly and subsequent division (Margolin, 2000; RayChaudhuri et al., 2000; Harry, 2001). Z rings are not normally formed at sites occupied by the nucleoid (a phenomenon referred to as ‘nucleoid veto’ or ‘nucleoid occlusion’), leaving only sites in the middle and at the poles of the cell available for division. In addition, the minC, minD and minE gene products prevent cell division at most positions outside the middle of the cell. MinCD forms an inhibitor of FtsZ assembly, which with the help of MinE oscillates from cell pole to cell pole in a way that leaves only the central part of the cell competent for Z-ring formation (Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999a,b; Fu et al., 2001; Hale et al., 2001). Although MinC and MinD are widespread, they are not ubiquitous, and additional systems are likely to exist among bacteria for control of cytokinesis (Margolin, 2001).

The Gram-positive bacterium Streptomyces coelicolor A3(2) and its relatives exhibit an interesting developmental regulation of cell division that does not appear to invoke any of the above-mentioned mechanisms. These organisms grow as branching hyphae, containing few and widely spaced cell divisions – vegetative cross-walls – that separate the syncytial hyphal cells (Chater and Losick, 1997). To produce spores that can spread in the environment, streptomycetes first form an aerial mycelium on the surface of colonies (Kelemen and Buttner, 1998; Chater, 2001); syncytial aerial hyphae are then converted into chains of spores by another kind of cell division – sporulation septation. One aerial hyphal cell may simultaneously lay down up to 50 or more sporulation septa, which give rise to a chain of immature spores, each containing a single copy of the genome (Chater, 1999). The ftsZ gene of S. coelicolor A3(2) is required for both vegetative cross-walls and sporulation septation, and Z rings are formed in both cases (McCormick et al., 1994; Schwedock et al., 1997). Thus, sporulation involves the synchronous assembly of multiple Z rings in a single hyphal cell. This massive and synchronous cell division requires a developmentally controlled induction of ftsZ transcription in sporogenic hyphae (Flärdh et al., 2000; Kwak et al., 2001). During sporulation, the formation of Z rings, as well as the initial closure of septa, occurs before detectable partitioning of nucleoids (Schwedock et al., 1997; Miguélez et al., 1998), indicating that an overt nucleoid veto is not acting in these conditions. Furthermore, the recently completed genome sequence of S. coelicolor A3(2) contains no obvious homologues of minC or minE (Bentley et al., 2002). In B. subtilis, the gene divIVA has an analogous role to that of minE in E. coli (Cha and Stewart, 1997; Edwards and Errington, 1997), but we have found that the S. coelicolor homologue of divIVA does not primarily affect cell division and is rather involved in hyphal tip growth (K. Flärdh, manuscript in preparation). This suggests that the placement of Z rings during Streptomyces sporulation is controlled by a mechanism(s) other than those recognized so far in E. coli and B. subtilis. Assuming that such mechanisms may act directly on FtsZ and be specific for sporulation, we speculated that it should be possible to isolate ftsZ mutations interfering with sporulation septation, without affecting vegetative growth. In this paper, we describe one such non-sporulating ftsZ mutant.

The ftsZ17(Spo) mutation interfered strongly with the assembly of Z rings in sporogenic aerial hyphae and thereby blocked sporulation. Also, the numbers of vegetative cross-walls were reduced, but to a lesser extent than the drastic effect on sporulation septa and with few consequences for growth of the strain. The findings suggest that the requirements for Z-ring formation are different during sporulation compared with vegetative growth, and that some kind of post-translational regulation may be involved in the developmental control of cell division. In addition, they demonstrate that unique properties of S. coelicolor, such as the dispensability of septum formation for growth and survival, the availability of a viable ftsZ-null mutant and the distinctive phenotypes of different cell division mutants, make this organism a useful genetic model for the study of bacterial cytokinesis.

Results

Isolation of a non-sporulating ftsZ mutant

Streptomyces coelicolor ftsZ is expressed from at least two promoters in the ftsQZ intergenic region; ftsZ3p appears to be constitutive, and ftsZ2p is strongly induced during sporulation. A mutant with a defective ftsZ2p promoter failed in both sporulation septation and accumulation of the spore pigment, indicating that synthesis of the grey pigment normally associated with mature spores of S. coelicolor is linked to the formation of sporulation septa ( Flärdh et al., 2000 ). Taking advantage of this finding, we have set up a screening strategy to identify ftsZ mutations that specifically affected sporulation. After the introduction of mutagenized pKF32 (an integrating plasmid that carries ftsZ ) into the ftsZ -null mutant HU133, most exconjugants exhibited growth and development like the wild-type strain M145 and formed colonies with an abundant grey aerial mycelium ( Fig. 1A ). The ftsZ -null mutant HU133 formed smaller colonies with a large blue halo and a sparse aerial mycelium ( McCormick et al., 1994 ) ( Fig. 1A ). However, we found that HU133 ( ΔftsZ ) was sensitive to the high MgCl 2 concentration normally used for the interspecies conjugation procedure and barely formed colonies at all under these conditions (data not shown). Thus, severely defective ftsZ alleles were counterselected and could not be isolated in this screening procedure. On the other hand, strains with restored vegetative growth but specific sporulation defects could be identified based on their white aerial mycelium. After isolating candidate mutants, their aerial mycelia were examined microscopically to detect those with decreased or defective spore forma-tion. Screening of about 5000 exconjugants yielded 10 potential mutants with variable degrees of defects in sporulation.

Figure 1.

Sporulation phenotype of ftsZ17 (Spo) mutants.

A. Colony appearance after 4 days of growth on MS agar. The S. coelicolor strains are M145 ( ftsZ + ); HU133 ( ΔftsZ::aphI ); J2417 (HU133 with plasmid pKF32 [ ftsZ + ]); O17 (HU133 with plasmid pO17 [ ftsZ17 (Spo)]); J1915 ( ΔglkA-esp119 ftsZ + ); and K102 ( ΔglkA-esp119 ftsZ17 (Spo)).

B. Phase-contrast micrographs of spore chains and aerial hyphal fragments from the surface of J1915 and K102 colonies. Growth conditions were the same as in (A). Size bar = 10 µm.

To test the linkage of the white phenotypes to ftsZ, the integrated pKF32 derivative was recovered from the potential mutants and transferred back to HU133. In only one of the 10 tested strains was the sporulation phenotype linked to the plasmid, and all new HU133 exconjugants with this plasmid showed the same phenotype as the original mutant, designated O17. The plasmid recovered from this mutant was designated as pO17, and the ftsZ allele carried on it was called ftsZ17(Spo) as it gave a sporulation defect, but allowed apparently normal vegetative growth. DNA sequencing revealed one basepair change in ftsZ borne on pO17 compared with pKF32 (the ftsZ sequence on pKF32 did not deviate from the published ftsZ sequence). This was a transition that changed ftsZ codon 249 from GCG (alanine) to ACG (threonine). Conveniently, the mutation created a new AatII restriction site, which allowed easy detection and confirmation of the ftsZ17(Spo) mutation.

FtsZ17(Spo) differentially affects vegetative and sporulation septation

The mutant strain O17 resembled a classical white developmental mutant (Hopwood et al., 1970). As it was a non-sporulating strain, it gave fewer colonies than wild type upon restreaking but, as judged from the colony size, it grew as well as the isogenic J2417 strain on MS agar and had lost the blue halo phenotype characteristic of HU133 (Fig. 1A). O17 formed an aerial mycelium, but this remained white (although a grey shade developed after prolonged incubation). Examination of cells from the aerial mycelium by phase-contrast microscopy showed no spores and mostly non-septate aerial hyphae (data not shown, but see below).

The ftsZ17(Spo) allele was introduced into the chromosomal ftsZ locus by allelic exchange to produce strain K102. As shown in Fig. 1A, this strain retained the normal colony size and white aerial mycelium of strain O17. The aerial mycelium of K102 contained mostly long non-septated aerial hyphae (Fig. 1B). Many hyphal fragments appeared dark in phase-contrast illumination, and some had been separated from aerial hyphae, indicating that occasional sporulation septa could form (Fig. 1B) (vegetative cross-walls do not normally lead to the separation of daughter cells). The aerial hyphal fragments in K102 were similar to those produced by strain O17, and highly reminiscent of the whiH and ftsZΔ2p phenotypes (Flärdh et al., 1999; 2000).

Although introduction of the ftsZ17(Spo) allele restored apparently normal vegetative growth to the ftsZ-null mutant HU133 (based on colony size), this did not rule out the possibility that ftsZ17(Spo) had any effects on vegetative septation. Therefore, we used the membrane stain FM4-64 (Pogliano et al., 1999) to visualize septation in liquid-grown mycelia. This showed that vegetative septa formed in both K102 and its ftsZ+ parent J1915 (data not shown). Using fluorescently labelled vancomycin (R. A. Daniel and J. Errington, personal communication), which binds to the d-alanyl-d-alanine part of non-cross-linked peptidoglycan, it was confirmed that these structures were hyphal cross-walls (Fig. 2). However, the frequency of septation was reduced in the ftsZ17(Spo) mutant compared with J1915. This reduction was reflected in an at least two- to threefold increase in the average distances from growing tips to the first cross-walls. The interpretation could be complicated by the fact that, although spores were used as inoculum for J1915, aerial hyphal fragments were used for K102 in these experiments. Therefore, the physiological states of the mycelia that were examined may not have been exactly comparable, and this may affect the frequency of septation. It should be noted that the deletion of glkA in J1915 is known to cause some ectopic sporulation in the substrate mycelium (Kelemen et al., 1995). Although spore chains were not seen in the young mycelia used in our analyses, the ΔglkA allele may have affected the observed septation patterns. Despite these limitations, the results showed that K102 formed vegetative cross-walls, although with a reduced frequency compared with the parent. Importantly, this reduction was much smaller than the nearly complete abolishment of sporulation septation in the mutant, and not large enough to have a strong impact on growth. As a comparison, an ftsQ null mutant of S. coelicolor had an approximately 10-fold reduction in the numbers of vegetative septa, and this led to distinguishable defects in colony growth (McCormick and Losick, 1996).

Figure 2.

Hyphal cross-walls in vegetative mycelium of strain J1915 ( ftsZ + ) (A) and its isogenic ftsZ17 (Spo) mutant K102 (B). Mycelia were grown in liquid TSB medium. Cross-walls were visualized by staining the peptidoglycan cell wall with a BODIPY FL vancomycin conjugate. Arrowheads indicate some clear examples of hyphal cross-walls. Size bar = 10 µm.

The ftsZ17(Spo) allele is partially dominant and expressed at normal levels

Partial diploids were constructed carrying ftsZ + in its normal locus and ftsZ17(Spo) on an integrated plasmid, either in the ΦC31 attB site (pO17) or integrated by homologous recombination at the ftsZ locus (pKF44) in strain M145. In both cases, ftsZ17(Spo) showed partial dominance over ftsZ +. Spore pigmentation was reduced, and the frequency and spacing of sporulation septa were affected, as indicated by the heterogeneous spore size and the production of many long aerial hyphal fragments (Fig. 3). Plasmid pO17 showed a stronger effect on pigmentation than pKF44 (Fig. 3), which could be related to them being at different positions on the chromosome, or to the fact that plasmids with the ΦC31 integration system are often present at several copies per genome (M. C. M. Smith, University of Nottingham, personal communication; K. Flärdh, unpublished observations). Also, when ftsZ+ was placed on an integrating plasmid (pKF32 or pKF43) in an ftsZ17(Spo) background, sporulation was disturbed and strains were oligosporogenous and lacked many sporulation septa, leading to highly variable spore sizes (data not shown). Thus, the FtsZ17(Spo) protein interfered with sporulation septation when co-expressed with normal FtsZ.

Figure 3.

Partial dominance of ftsZ17 (Spo) over ftsZ + .

A. Colony appearance after 4 days of growth on MS agar. The strains are derivatives of S. coelicolor strain M145 containing different plasmids integrated in their genomes: pKF43 [ftsZ+] and pKF44 [ftsZ17(Spo)] integrate by homologous recombination at the ftsZ locus, whereas pKF32 [ftsZ+] and pO17 [ftsZ17(Spo)] integrate at the ΦC31 attB site.

B. Phase-contrast micrographs of typical aerial hyphal fragments and spore chains from the surface of colonies of M145 containing the indicated plasmids. Growth conditions were the same as in (A). Examples of putative sporulation septation in strains with pKF44 and pO17 are indicated by white tick marks. Size bar = 10 µm.

The dominance of the mutant allele suggested that it was expressed at levels comparable to the wild-type allele. To confirm this, Western blotting was used to compare FtsZ content in surface-grown cells between mutant and wild-type strains. This showed that K102 accumulated similar amounts of FtsZ as the parent ftsZ + strain J1915 during sporulation, whereas the isogenic strain K101, which is ftsZ + and lacks the sporulation-induced ftsZ2p promoter, produced a lower total amount of FtsZ (reduced by ≈ 30–50%) under these conditions (data not shown).

The mutation affects the assembly of Z rings during sporulation

We have constructed an ftsZ–egfp translational fusion that allows visualization of Z rings in living hyphae of S. coelicolor (N. Grantcharova, U. Lustig and K. Flärdh, manuscript in preparation). To investigate the sporulation defect further and determine how ftsZ17(Spo) affected the assembly of FtsZ in vivo, derivatives of the ftsZ17(Spo) mutant (K201) and an isogenic ftsZ+ strain (K200) were constructed that carried the ftsZ–egfp fusion on pKF41. When K200 was sampled after 30–45 h of growth on MS agar plates, many aerial hyphal cells were observed with markedly increased levels of fluorescence, and several of these had formed regularly spaced multiple Z rings in ladder-like patterns (Fig. 4B). This is in agreement with previous observations of developmental control of ftsZ expression and Z-ring formation (Schwedock et al., 1997; Flärdh et al., 2000). Despite having wild-type FtsZ fused to EGFP, pKF41 did not restore sporulation septation in K201. Although this is consistent with the dominance of ftsZ17(Spo), it should be noted that, as in many other organisms, the EGFP tag disturbs the function of FtsZ, and the ftsZ–egfp fusion on pKF41 can only partially complement the ΔftsZ::aphI mutation in strain HU133 (N. Grantcharova, U. Lustig and K. Flärdh, manuscript in preparation). Many aerial hyphae of K201 showed a marked upregulation of expression of the ftsZ–egfp fusion, but no ladders of Z rings were observed. Aerial hyphae were either evenly fluorescent (Fig. 4D) or showed various types of internal fluorescent filaments or patches, probably reflecting incorrect assembly of FtsZ (Fig. 4F).

Figure 4.

Visualization of FtsZ assembly in the ftsZ + strain K200 (A and B) and the isogenic ftsZ17 (Spo) mutant K201 (C–F). Both strains harboured plasmid pKF41 with an ftsZ–egfp translational fusion. Cultures were incubated on MS agar plates for 43 h, before aerial mycelia were sampled and examined in a fluorescence microscope.

A, C and E. Phase-contrast micrographs.

B, D and F. Fluorescence images showing representative examples of FtsZ–EGFP localization in sporogenic aerial hyphae.

B. A typical ladder of Z rings that are seen at an early stage of sporulation in K200.

D and F. Examples of two types of patterns found in K201 aerial hyphae: dispersed fluorescence throughout the cell (D) or irregular patches and filaments of varying lengths (F). No ordered ladders of rings as in (B) were seen with K201. Size bar = 10 µm.

These experiments indicated that ftsZ17(Spo) disturbed the assembly of Z rings in sporogenic hyphae. However, as they were performed with a wild-type allele of ftsZ fused to EGFP in a mutant background, they may not accurately have reflected the behaviour of the mutant protein. Therefore, immunofluorescence microscopy was also used to visualize the FtsZ17(Spo) protein directly in strain K102. Frequent aerial hyphae containing multiple Z rings were seen in the parent strain J1915 (Fig. 5). In contrast, no regular rings could be found in aerial hyphae of K102, despite several experiments and sampling at different times of development, during which Z rings are normally seen in ftsZ + strains. As in the parent, most hyphae in both substrate and aerial mycelium stained weakly, whereas several aerial hyphae displayed a strong fluorescence (Fig. 5D), indicating that expression of the mutant allele ftsZ17(Spo) was strongly upregulated in sporogenic hyphae, similar to that of the wild-type allele (Flärdh et al., 2000). The darker hyphae seen at the bottom of Fig. 5C appear to be at a more advanced stage of development, when spore-like cell wall characteristics have developed, making them more resistant to lysozyme treatment (Flärdh et al., 1999) and therefore harder to permeabilize. Importantly, such alterations in the cell wall are seen after sporulation septation in wild-type strains (Flärdh et al., 1999), and therefore occur after the time of maximal FtsZ expression. In summary, the developmental upregulation of FtsZ synthesis appeared to be normal in the ftsZ17(Spo) mutant, but the mutant was deficient in the assembly of Z rings in sporulating aerial hyphae, which probably explains why it was unable to sporulate.

Figure 5.

FtsZ immunofluorescence microscopy of strain J1915 ( ftsZ + ) (A and B) and the isogenic ftsZ17 (Spo) mutant K102 (C and D). Aerial mycelia were grown and sampled as described in the legend to Fig. 4 , and subjected to immunofluorescence microscopy with an anti-FtsZ antiserum.

A and C. Phase-contrast micrographs.

B and D. Fluorescence images showing the localization of FtsZ in sporogenic aerial hyphae. A typical example of staining from each strain is shown. Size bar = 10 µm.

Position of the mutation in a structural model of S. coelicolor FtsZ

To evaluate the possible effects of this and other mutations, we constructed a homology model of the structure of S. coelicolor FtsZ, using the known three-dimensional structure of Methanococcus jannaschii FtsZ1 (Löwe and Amos, 1998) as a template (Fig. 6). The amino acid sequence identity is ≈ 50% throughout most of the two proteins, which provides an excellent basis for such modelling. With the exception of one conservative change (Ala-48 in the Methanococcus jannaschii protein is found as Gly-19 in the S. coelicolor FtsZ), the residues lining the GTP binding site are the same, and most other structural features appear to be conserved in both domains. The mutation under investigation here, Ala-249Thr, affects a residue that is equivalent to Val-277 of the M. jannaschii protein. This side-chain points into the hydrophobic core of the protein in the crystal structure, and the placement and local conservation of non-polar residues suggest that it should do so in the S. coelicolor protein as well. It seems reasonable to suggest that the introduction of the slightly larger, and polar, threonine side-chain at this position will have consequences for the FtsZ surface locally. The residue corresponding to Ala-249 of the S. coelicolor protein is not itself generally conserved in other FtsZs, but lies near some highly conserved amino acids. In the immediate vicinity of the residue altered by the mutation, in the three-dimensional structure, the C-terminus of the M. jannaschii protein forms an extended loop on the surface of the structure (black in Fig. 6). The primary structure of these portions of the two sequences is very dissimilar, and varies extremely in length and sequence in other family members. Therefore, the M. jannaschii structure provides no guidance for the situation in S. coelicolor FtsZ after residue 315.

Figure 6.

Homology model for S. coelicolor FtsZ and position of the ftsZ17 (Spo) mutation. The N-terminal domain, which is expected to have GTPase activity, is shown in green; the residues of the GTP binding site are orange, and a modelled molecule of GDP is yellow. One of the residues lining the GTP binding site (Gly-19) is not identical between S. coelicolor FtsZ and M. jannaschii FtsZ1 and is coloured blue. The C-terminal domain, where the Ala-249Thr mutation is located, is shown in blue. The residue altered by mutation is indicated in red, and nearby non-polar residues (Leu-222, Leu-247, Ile-251, Ala-309 and Phe-312) are shown in yellow. The regions from the C-terminus that lie nearby in the M. jannaschii structure are shown in black; these portions almost certainly have a different structure in the two proteins.

Discussion

The ftsZ17(Spo) mutation abolished sporulation, although not overtly affecting vegetative growth. Closer examination revealed that it interfered with both sporulation septation and vegetative cross-wall formation, but had a stronger effect on the former. This preferential effect of the ftsZ(Spo) mutation on sporulation demonstrated a difference in the requirements for FtsZ assembly between the two modes of cell division used by S. coelicolor. The two types of septa are known to differ in several ways. Sporulation septa are thick and lead to the separation of individual spores, whereas vegetative cross-walls are thinner and do not lead to the detachment of daughter cells after the completion of cytokinesis (Wildermuth and Hopwood, 1970). Vegetative hyphae make one septum per cell when they divide and, in Streptomyces granaticolor, this is initiated close to the middle of an apical hyphal cell, which contains multiple copies of the chromosome (Kretschmer, 1982). Sporogenic aerial hyphae, on the other hand, synchronously produce from around 10 up to 100 sporulation septa, one between each of the chromosomes in such cells (Wildermuth and Hopwood, 1970; Schwedock et al., 1997; Flärdh et al., 1999). One penicillin-binding protein (PBPA) in Streptomyces griseus was found to localize specifically to sporulation septa, but it was not required for making spores (Hao and Kendrick, 1998; Jiang and Kendrick, 2000). Otherwise, the molecular differences between the septal types are unknown. Against this background, the ftsZ(Spo) mutant should be useful for studies of cell division and its developmental regulation in Streptomyces.

The formation of sporulation septa depends on at least six developmental regulatory genes: whiA, whiB, whiG, whiH, whiI and whiJ (Chater, 1999; 2001). These early whi genes are needed for the large induction of ftsZ transcription from the ftsZ2p promoter that occurs in aerial hyphae and is required for sporulation (Flärdh et al., 2000). However, transcriptional regulation of ftsZ may not be the only way in which whi genes affect cell division. The ftsZ17(Spo) mutant that is described in this paper is unable to sporulate because of a defect in FtsZ function rather than expression. Interestingly, the sporulation phenotype was the same, independent of whether ftsZ was not working properly or not expressed at a sufficient level in aerial hyphae. This opens the possibility that a developmental regulator may directly or indirectly alter the control of Z-ring assembly. Possible candidates, in addition to whi genes, are ssgA and samR, genes of unknown function that have been implicated in the stimulation of septum formation during development of S. coelicolor A3(2) (van Wezel et al., 2000; Tan et al., 2002).

Developmental control of cytokinesis is well known from other differentiating bacteria. In Caulobacter crescentus, cell division is only initiated in stalked cells, and expression of division genes is intimately connected to the intertwined processes of differentiation and cell cycle progression (reviewed by Martin and Brun, 2000). During B. subtilis sporulation, a shift from medial to asymmetric septation is crucial for establishment of the different fates of the mother cell and the prespore (reviewed by Levin and Losick, 1999). Shortly after initiation of sporulation, the medial Z ring is switched to form two polar Z rings, but only one of them is used for septation (Levin and Losick, 1996). The shift in Z-ring position is mediated by both the SpoIIE protein and an increase in ftsZ transcription, and it involves dynamic spiral-shaped FtsZ structures that move from the mid-cell to polar division sites (Ben-Yehuda and Losick, 2002). It is not yet clear whether such intermediate structures are formed during sporulation in S. coelicolor, and whether the various irregularly shaped filaments seen with the ftsZ–egfp fusion in the ftsZ17(Spo) mutant (Fig. 4F) could be related to this.

Homology modelling of the S. coelicolor protein using the structure of FtsZ1 from M. jannaschii places the mutation (Ala-249Thr) in a surface loop on one side of the C-terminal domain (Fig. 6). This region of the protein is distant from the GTP binding site and is not involved in monomer–monomer interactions within a single protofilament (Löwe and Amos, 1998; 1999). Three mutations of nearby residues in other organisms have been described previously in site-directed mutagenesis approaches targeting charged amino acids predicted to be surface exposed. The Glu-250Ala mutation in E. coli FtsZ alters a conserved glutamate/aspartate corresponding to S. coelicolor position Glu-248. This mutant protein retained both GTPase activity and the ability to polymerize in vitro. However, the function of the mutant allele could not be tested in vivo, apparently because it had a dominant-negative effect, and expression constructs were rapidly inactivated or lost (Lu et al., 2001). A double mutant Glu-250Lys/Asp-253Lys behaved in a similar way (Lu et al., 2001). In C. crescentus FtsZ, an AspGlu254AlaAla mutation changed residues corresponding to S. coelicolor position Glu-248. This was a recessive-lethal allele and gave filamentous cells with constrictions, at which rings or extended bands of FtsZ were seen after immunostaining (Wang et al., 2001). In contrast, this paper describes an in vivo strategy finding an important residue that could not have been anticipated by model predictions. If the ftsZ17(Spo) allele behaves in a way similar to the mutations described above, it is likely that it does not directly affect GTPase activity or polymerization to form protofilaments. However, this has yet to be tested directly. Combined, these results provide strong evidence that the affected region of the C-terminal domain of FtsZ is critical to its function in vivo, and that the property affected is distinct from its GTPase activity.

What is the molecular basis for the effect of ftsZ17(Spo) on sporulation? The predicted effect of the Ala-249Thr mutation on the surface of the FtsZ protofilament would be consistent with an altered interaction between FtsZ and another protein (or other factor) that is specifically required for Z-ring assembly during sporulation. The consequence of the mutation could, for example, be failure to recognize nucleation sites, instability of FtsZ polymers or premature (or uncontrolled) assembly of FtsZ at locations outside the division sites. Alternatively, and also consistent with the observed effect on vegetative cross-walls, the mutation could alter a general property of FtsZ required for both types of cell division. This could, for example, disturb polymer stability, lateral contacts between protofilaments or interaction with another component of the septation machinery. The much stronger effect on sporulation septa would then imply that vegetative cross-wall formation is more resistant to such interference than sporulation septation. These alternative models are not mutually exclusive. Both are consistent with a hypothesis published recently for B. subtilis, which suggested that positioning of FtsZ rings is determined by an interplay of counteracting factors stabilizing or destabilizing FtsZ polymers (Levin et al., 2001). With the currently available data, none of our models can be excluded. Analysis of intra- or extragenic suppressors of ftsZ17(Spo) and biochemical comparison of FtsZ17(Spo) with wild-type FtsZ should help in distinguishing between them.

Experimental procedures

Bacterial strains and media

The S. coelicolor A3(2) strains that were used are listed in Table 1. E. coli strain DH5α (Hanahan, 1983) was used as a host for recombinant plasmids, and strain ET12567/pUZ8002 was used to drive conjugative transfer of non-methylated DNA from E. coli to S. coelicolor as described previously (Kieser et al., 2000). Culture conditions, antibiotic concentrations, genetic manipulations and recombinant DNA work generally followed previously described procedures for E. coli (Sambrook et al., 1989) and Streptomyces (Kieser et al., 2000). S. coelicolor strains were cultivated on MS agar plates or in YEME liquid medium, unless stated otherwise (Kieser et al., 2000).

Table 1. .Streptomyces coelicolor A3(2) strains used in this study.
StrainGenotypeReference
HU133M145 ΔftsZ::aphI McCormick et al. (1994 )
J1915M145 ΔglkA-esp119 Kelemen et al. (1995 )
J2417HU133 attBφC31::pKF32[ftsZ+] Flärdh et al. (2000 )
K101J1915 ftsZΔ2pThis work
K102J1915 ftsZ17(Spo)This work
K200J1915 attBφC31::pKF41[Φ(ftsZ–egfp)]This work
K201K102 attBφC31::pKF41[Φ(ftsZ–egfp)]This work
M145Prototrophic, SCP1 SCP2 Pgl+ Kieser et al. (2000 )
O17HU133 attBφC31::pO17[ftsZ17(Spo)]This work

Strategy for isolation of non-sporulating ftsZ mutants

Plasmid pKF32 contains ftsZ, its promoters and most of the upstream ftsQ gene in a vector that is mobilizable from E. coli to Streptomyces by conjugation. It integrates into the ΦC31 attB site on the S. coelicolor chromosome and fully complements the ΔftsZ::aphI mutation (Flärdh et al., 2000). Plasmid DNA of pKF32 was mutagenized in vitro with hydroxylamine (Miller, 1992) and introduced into ET12567/pUZ8002 by electroporation. Transformants from hydroxylamine treatments that decreased the transformation frequency by around 90% were pooled and used as donors for conjugation into the S. coelicolor strain HU133 (ΔftsZ). Apramycin-resistant exconjugants that gave pale grey or white, normal-sized colonies with no blue halo were isolated. Aerial mycelia from white or pale grey strains were examined by phase-contrast microscopy to identify candidates with defects in sporulation septation. Total genomic DNA was extracted from such strains using the CTAB procedure (Kieser et al., 2000) and used to transform ET12567/pUZ8002 to apramycin resistance by electroporation. Two transformants from each potential mutant were used for conjugative transfer of the plasmid to S. coelicolor strain HU133. The plasmid deriving from one of the potential mutants (O17) conferred a sporulation phenotype when reintroduced into HU133, and the ftsZ allele on this plasmid was sequenced using the Taq Dye terminator procedure on an ABI 877 robotic workstation and an ABI 377 DNA sequencer.

Introduction of ftsZ17(Spo) and ftsZΔ2p alleles into the chromosomal ftsZ locus

The ftsZ17(Spo) allele was excised as a HindIII–EcoRV fragment from plasmid pO17 and ligated between the HindIII and EcoRV sites of plasmid pIJ6650, which is a derivative of pKC1132 carrying the counterselectable glkA gene (Kieser et al., 2000). The resulting plasmid, pKF47, was introduced into ET12567/pUZ8002 and transferred by conjugation to J1915, a ΔglkA derivative of S. coelicolor strain M145 (Kelemen et al., 1995). Integration of the plasmid by homologous recombination at the ftsZ locus was obtained by selection for the apramycin resistance marker (ApraR) on the plasmid. Plasmid co-integrants were isolated and allowed to sporulate on MS agar without selection. Spore preparations were then spread on mannitol minimal medium containing 200 mM 2-deoxyglucose (Dog) to counterselect strains retaining the plasmid (Kieser et al., 2000). Among Dog-resistant strains (DogR), two colony types emerged when plated on MS agar, white and dark grey colonies. Chromosomal DNA was prepared from some representative strains of each type, as well as from DogS ApraR strains that still contained the integrated pKF47. Southern blotting and hybridization with a digoxygenin (DIG)-labelled probe against the ftsZ locus was carried out using the DIG system according to the manufacturer's instructions (Roche Diagnostics). This confirmed that DogR ApraS strains contained a single ftsZ allele, and that the strains with a white phenotype contained a new AatII site, which is characteristic of the ftsZ17(Spo) allele. One such ftsZ17(Spo) mutant was selected and named K102.

The ftsZΔ2p allele was excised as a HindIII–EcoRV fragment from plasmid pKF33 (Flärdh et al., 2000) and ligated between the HindIII and EcoRV sites of plasmid pIJ6650 to generate plasmid pKF46. This was used for allelic exchange of ftsZΔ2p onto the chromosome of S. coelicolor strain J1915 exactly as described above, except that the loss of a DdeI site was used to confirm the presence of ftsZΔ2p in Southern blot hybridizations. The new ftsZΔ2p mutant was named K101.

Tests for dominance of ftsZ17(Spo)

Plasmids pKF43 and pKF44 were constructed by deleting an SphI fragment including the ΦC31 attP site and integrase gene from pKF32 and pO17 respectively. Plasmids pKF32, pO17, pKF43 and pKF44 were first introduced into S. coelicolor strain M145 by conjugation from E. coli, and representative apramycin-resistant exconjugants were purified, plated on MS agar and examined for spore pigmentation and aerial hyphal morphology after about 4 days of development. In a second experiment, the same four plasmids were introduced into the ftsZ17(Spo) mutant K102 and its isogenic parent J1915. In this case, some samples for microscopic analysis of aerial hyphal morphologies were taken directly on the conjugation plates after 4–5 days of growth in the presence of apramycin selection. This allowed evaluation of multiple exconjugant strains without subculturing, thereby reducing the risk of drawing incorrect conclusions from strains that may have suffered gene conversion or homogenotization. Purified exconjugant strains showed the same phenotypes as those seen on the conjugation plates.

Microscopic methods

Impression preparations from the surface of mature colonies (4 days on MS agar) were taken as described by Chater (1972) and examined by phase-contrast microscopy using a Nikon Optiphot-2 microscope equipped with a digital imaging system as described elsewhere (Åkerlund et al., 1995).

For visualization of hyphal cross-walls, vegetative mycelium was grown overnight in liquid YEME medium, fixed by mixing with an equal volume of 60 mM sodium phosphate, pH 7.4, containing 0.04% glutaraldehyde and 5% paraformaldehyde, incubating for 15 min at room temperature and washing twice in phosphate-buffered saline (PBS). The cells were resuspended in PBS with 1 µg ml−1 FM4-64 (Molecular Probes), spread and allowed to air dry on poly l-lysine-coated slides, mounted under coverslips in PBS containing 50% glycerol and 1 µg ml−1 FM4-64 and examined as described below. For staining of the cell wall, fixed cells were resuspended in PBS containing 0.5 µg ml−1 BODIPY® FL vancomycin (Molecular Probes), incubated for 15 min at room temperature, allowed to air dry on poly l-lysine-coated slides and mounted under coverslips in PBS containing 50% glycerol and 1 µg ml−1 BODIPY® FL vancomycin.

The construction of plasmid pKF41, and the behaviour of the translational ftsZ–egfp fusion that it carries, will be described elsewhere (N. Grantcharova, U. Lustig and K. Flärdh, manuscript in preparation). Briefly, the plasmid is a pSET152 derivative containing the ΦC31 site-specific integration system, an apramycin resistance marker, S. coelicolor ftsZ translationally fused to the gene for EGFP and the natural promoters in the ftsQ–ftsZ intergenic region. The localization of FtsZ–EGFP hybrid proteins in aerial hyphae was observed in impression preparations from MS agar-grown cultures. Coverslips with attached aerial mycelium were placed on microscope slides coated with 1% agarose in PBS. Immunofluorescence microscopy of FtsZ was carried out essentially as described previously (Schwedock et al., 1997), except that 7-aminoactinomycin D was used for DNA staining, and the secondary antibodies were conjugated to Alexa Fluor 488 (Molecular Probes).

All fluorescence microscopy was carried out using a Zeiss Axioplan II Imaging fluorescence microscope equipped with appropriate filter sets, an Axiocam CCD camera and axiovision software (Carl Zeiss Light Microscopy). Digital images for the figures were assembled using adobe photoshop software.

Western blotting

The relative cellular contents of FtsZ were estimated by Western blotting. Cultures of strains J1915, K101 and K102 were grown on cellophane-covered MS agar as described previously (Flärdh et al., 2000). Samples were scraped off at appropriate times of development, and protein extracts containing equal amounts of mycelial wet weight were prepared as described by Schwedock et al. (1997), with the addition of a sonication step to aid in breaking up the cells. Proteins were separated on SDS-polyacrylamide gels and transferred electrophoretically to Immobilon-P membranes (Millipore) using a Mini Trans-Blot cell according to the manufacturer's instructions (Bio-Rad Laboratories). FtsZ was detected using a 1:10 000 dilution of an antiserum directed against S. coelicolor FtsZ (Schwedock et al., 1997), a peroxidase-linked donkey Ig anti-rabbit secondary antibody (Amersham Pharmacia Biotech) and the ECL Western blotting detection reagents (Amersham Pharmacia Biotech).

Homology modelling

Similar sequences from GenBank (Benson, 2000) were located and aligned using hidden Markov models (Karplus et al., 1997) and multalign (Corpet, 1988). The structure of M. jannaschii FtsZ1 (Löwe and Amos, 1998) was obtained from the PDB (Berman et al., 2000). A pairwise alignment of this sequence with that of the S. coelicolor FtsZ was the basis of creating a homology model, using the M. jannaschii protein as a template in the program sod (Kleywegt et al., 2001). The model was modified somewhat in the graphics program O (Jones et al., 1991), using rotamers that would improve packing in the interior of the protein and to account for insertions and deletions in some loop regions. Structural comparisons were carried out using O. Figure 6 was prepared using the programs molscript (Kraulis, 1991) and molray (Harris and Jones, 2001).

Acknowledgements

We thank Santanu Dasgupta for comments on the manuscript, Ulrika Lustig and Julie Ferguson for excellent technical assistance, and Mark Paget for gifts of strains. This work was supported by a grant from the Swedish Research Council to K.F. and from the National Institutes of Health (GM56915) to J.R.M. S.L.M and W.U are grateful for the support of the Swedish Foundation for Strategic Research via the Glycoconjugates in Biological Systems network, GLIBS. The Zeiss Axioplan II Imaging fluorescence microscope was purchased with support from the Swedish Natural Sciences Research Council to Kurt Nordström.

Ancillary