Generation of a non-sporulating strain of Streptomyces coelicolor A3(2) by the manipulation of a developmentally controlled ftsZ promoter

Authors


Abstract

The differentiation of Streptomyces aerial hyphae into chains of unigenomic spores occurs through the synchronous formation of multiple FtsZ rings, leading to sporulation septa. We show here that developmental control of ftsZ transcription is required for sporulation in Streptomyces coelicolor A3(2). Three putative ftsZ promoters were detected in the ftsQ–ftsZ intergenic region. In addition, some readthrough from upstream promoter(s) contributed to ftsZ transcription. S1 nuclease protection assays and transcriptional fusions of the ftsZ promoter region to the egfp gene (for green fluorescent protein) provided evidence that ftsZ2p is a developmentally controlled promoter that is specifically upregulated in sporulating aerial hyphae. This upregulation required all the six early regulatory sporulation genes that were tested: whiA, B, G, H, I and J. The DNA sequence of the promoter indicated that it is not part of the developmental regulon that is controlled by the RNA polymerase sigma factor σWhiG. A strain in which the ftsZ2p promoter was inactivated grew normally during vegetative growth and formed aerial mycelium, but was deficient in sporulation septation. Thus, ftsZ2p was dispensable for vegetative growth, but was required for the strain to make sufficient FtsZ to support developmentally controlled multiple cell divisions in aerial hyphae.

Introduction

The subject of this study is the developmental control of cell division that is required for the conversion of Streptomyces aerial hyphae into chains of spores (Chater, 1999). During growth of the branching vegetative mycelium of streptomycetes, cell division is infrequent and gives rise to widely spaced cross-walls (vegetative septa) that delimit the long syncytial hyphal cells, some containing tens of copies of the genome (Prosser and Tough, 1991). Then, as colonies develop, specialized hyphae break the air–water interface at the surface of the colony and form an aerial mycelium. Initially, occasional cross-walls similar to the vegetative septa may be formed but, subsequently, up to 50 or more specialized septa (sporulation septa) are synchronously laid down in sporogenic aerial hyphae. These sporulation septa are regularly spaced and divide the aerial hyphae into unigenomic prespore units, which mature and are eventually released as free spores. In addition to their distinctive placement and temporal regulation, sporulation septa differ morphologically from vegetative septa and lead eventually to full separation and detachment of daughter cells, whereas the vegetative septa simply lead to hyphal cross-walls (Wildermuth and Hopwood, 1970; McVittie, 1974; Hardisson and Manzanal, 1976). An important issue in understanding morphological differentiation in Streptomyces is how this switch in the mode of cell division is controlled.

In Streptomyces coelicolor A3(2), the organism studied in this paper, one class of mutants, whi mutants, has been isolated that allows the formation of aerial mycelium, but is mostly unable to complete sporulation and show characteristic terminal aerial hyphal phenotypes (Hopwood et al., 1970; Chater, 1972; Chater and Merrick, 1976; Ryding et al., 1999). Of the genes identified in this way, whiG encodes an RNA polymerase σ factor that is needed to initiate sporulation (Méndez and Chater, 1987; Chater et al., 1989). At least five additional genes, whiA, B, H, I and J, are required for the formation of sporulation septa (Chater, 1972; Chater and Merrick, 1976; Davis and Chater, 1992; Ryding et al., 1998; Aínsa et al., 1999; 2000; Flärdh et al., 1999), yet the mechanisms by which these developmental genes alter the cell division process are not known.

Both types of septation in S. coelicolor require the major cell division gene ftsZ and may thus involve the same basic cell division machinery (McCormick et al., 1994), a notion that is supported by the involvement of another conserved division gene, ftsQ, in both processes (McCormick and Losick, 1996). FtsZ is present in virtually all bacteria examined and has an essential role in their cell division (Lutkenhaus and Addinall, 1997; Rothfield et al., 1999). FtsZ and eukaryotic tubulins are structurally and biochemically very similar and form a distinct family of GTPases (Löwe and Amos, 1998; 1999; Nogales et al., 1998). In Escherichia coli and other bacteria, the first stage of cell division is the assembly of FtsZ into a ring-like structure, the Z-ring, at the inner surface of the cytoplasmic membrane. This Z-ring is required to recruit other proteins implicated in septum formation to the division site (Lutkenhaus and Addinall, 1997; Rothfield et al., 1999). The only protein known to localize to a position close to the septation site independently of FtsZ, MinE, is not essential for formation of the septum, but helps to determine septal positioning (Raskin and de Boer, 1997). It is still not clear in any organism how the exact site for FtsZ assembly is determined, or what triggers the formation and contraction of the Z-ring at the correct times during the cell cycle.

The ftsZ genes of three streptomycetes have been characterized and found to be almost identical except for a variable C-terminal region (Dharmatilake and Kendrick, 1994; McCormick et al., 1994; Zhulanova and Mikulik, 1998). As with many other bacteria, ftsZ is found in a cluster of cell division and cell wall synthesis genes, which in the streptomycetes includes homologues of the division genes ftsQ, ftsW and ftsI (McCormick and Losick, 1996; Bentley, 2000; Mikulik et al., 2000). In all other bacteria examined, ftsZ is essential, but an ftsZ null mutant of S. coelicolor was viable and grew vegetatively, albeit poorly (McCormick et al., 1994). It overproduced a blue pigment (actinorhodin), was cold sensitive and made some aerial mycelium, but it failed to sporulate. The mycelium of this mutant was devoid of detectable septation, yet it could be successfully restreaked for subculture. (It remains unclear how the fragmentation into viable units needed for such propagation is possible.)

Using immunofluorescence microscopy, it has been shown that sporulation septation in S. coelicolor is preceded by the synchronous assembly of a large number of regularly spaced Z-rings along the sporogenic hypha (Schwedock et al., 1997). Although the accumulation of FtsZ protein was particularly obvious in sporulating hyphae (Schwedock et al., 1997), this did not correlate directly with the results of immunoblot analyses of overall FtsZ content in mycelium during development (Dharmatilake and Kendrick, 1994; Schwedock et al., 1997). In this paper, we address the question of whether ftsZ expression is developmentally controlled in S. coelicolor. Previous studies in Streptomyces griseus (A. J. Dharmatilake and K. E. Kendrick, personal communication) revealed at least two promoters for ftsZ in the ftsQ–ftsZ intergenic region, one of which was markedly upregulated upon induction of sporulation in liquid cultures, whereas transcription from the other, which was the stronger of the two during vegetative growth, decreased during sporulation. We extend those studies by showing that the same promoters are present in S. coelicolor A3(2) and are similarly regulated during development of surface-grown mycelium. The ftsZ2p promoter was specifically activated in those aerial hyphae that showed signs of sporulation, but the shift to preferential use of this sporulation-associated promoter was abolished in developmental mutants disrupted for any of six early whi genes. The ftsZ2p promoter was dispensable for vegetative septation and growth, but essential for sporulation.

Results

Mapping of ftsZ mRNA 5′ ends during a developmental time course

In order to determine whether transcription of the ftsZ gene was altered during differentiation of S. coelicolor A3(2), S1 nuclease protection assays were performed on total RNA prepared from cultures grown for different times on solid medium. Under these conditions, the formation of the vegetative mycelium was followed sequentially by aerial growth and sporulation during the time over which samples were harvested. A probe was designed to cover most of the intergenic region between ftsQ and ftsZ. The probe had a non-homologous tail on the 3′ end, which made it possible to distinguish between probe–probe re-annealing artifacts and transcriptional readthrough from promoters upstream of the BclI site in the ftsQ–ftsZ intergenic region (Fig. 1). Such upstream promoters have been observed for S. griseus ftsZ (A. J. Dharmatilake and K. E. Kendrick, personal communication). Three major mRNA 5′ ends were localized at nucleotide positions 4210–4211, 4161–4163 and 4104–4105 (numbered as in database submission U10879) (Fig. 1). They were tentatively designated ftsZ1p, ftsZ2p and ftsZ3p, respectively, although it was not clear at this stage whether they were produced by transcription initiation or by processing or degradation of mRNA.

Figure 1.

A. S1 nuclease protection analysis of ftsZ transcripts. Total RNA was extracted from cultures grown on MS agar at the times of mycelium development indicated. An equal amount of RNA was added to each S1 mapping reaction, and a control reaction with an equal amount of yeast tRNA was included in the lane labelled ‘tRNA’. The lines at the bottom indicate at which time points aerial mycelium and spore chains were detected by microscopic examination. Lanes labelled G, A, T and C denote a dideoxy sequencing ladder generated using the same oligonucleotide that was used to generate the S1 mapping probe. The lane labelled ‘marker’ contains a set of DNA size markers that facilitate comparison with the data in Fig. 6. Three ftsZ mRNA 5′ ends, defining ftsZ1p, ftsZ2p and ftsZ3p, as well as the specific fragment protected by transcripts originating at upstream promoters (‘readthrough’) are indicated.

B. DNA sequence of the ftsQ–ftsZ intergenic region. The sequence shown is from the database submission U10879. The TGA stop codon of ftsQ and the GTG start codon of ftsZ are bold and double underlined. The mRNA 5′ ends that were mapped in this region are indicated above the sequence (#). Putative −10 and −35 promoter motifs are marked and underlined. The sequences of the corresponding regions in S. griseus and Streptomyces collinus (database submissions U07344 and AF081213 respectively) are highly similar, and only differences from the S. coelicolor sequence are shown in the lines below that sequence. Gaps introduced in the alignment of the sequences are indicated (–). The Δ symbol indicates the BclI site that constituted the 3′ end of the ftsZ homologous sequence in the probe used for S1 mapping.

There are sequence motifs resembling canonical −10 and −35 promoter sequences at appropriate distances upstream from both ftsZ2p and ftsZ3p mRNA start sites (Fig. 1B). The DNA sequence of the ftsQ–ftsZ intergenic region is almost completely conserved between the three Streptomyces species in which it has been sequenced (Fig. 1B). Accordingly, the ftsZ2p and ftsZ3p signals corresponded to the putative promoters that had previously been reported in the ftsQ–ftsZ region of S. griseus (A. J. Dharmatilake and K. E. Kendrick, personal communication). For the bands corresponding to ftsZ1p (position 4210–11), a possible −10 motif is located upstream of this mRNA 5′ end, but no obvious −35 motif was detected (Fig. 1B).

At the earliest time point of growth of the surface cultures (16 h), the relative contributions to the total signal were 13% for ftsZ1p, 27% for ftsZ2p, 43% for ftsZ3p and 17% for transcriptional readthrough from upstream promoters. Similar relative signal strengths were observed during exponential growth in liquid culture (data not shown). Thus, the ftsZ3p signal was the most intense during early vegetative growth. However, its activity decreased gradually during development (Fig. 1A). The readthrough from upstream promoters decreased from the first time point to the second and then remained more or less constitutive throughout development. The ftsZ1p signal remained at a constant and low level throughout development. On the other hand, the intensity of the ftsZ2p signal increased markedly between 28 h and 36 h, which coincided with the time at which abundant aerial mycelium and sporulation septa appeared (some faint aerial mycelium was already observed at 28 h). Thus, ftsZ2p was the most abundant signal at all time points when sporulation septation was detected in the aerial mycelium (Fig. 1A).

Upregulation of the ftsZ2p promoter occurs exclusively in sporulating aerial hyphae

The RNA samples for the S1 assays were prepared from total colony material. Initially, the colonies consisted entirely of vegetative mycelium but, at later times, they were mixtures of vegetative mycelium of different physiological states, aerial mycelium and spores. Furthermore, only a subset of the aerial hyphae at each time point was probably at the early stage of sporulation when septa were laid down and FtsZ was likely to be most abundant. The upregulation of the putative ftsZ2p promoter might have occurred either to a moderate degree throughout most of the colony or to a much higher degree only in sporulating aerial hyphae. To distinguish between these possibilities, we fused a DNA fragment containing the ftsZ promoter region to the egfp reporter gene, encoding an enhanced version of the green fluorescent protein (GFP), and integrated the Φ(ftsZ–egfp) fusion (plasmid pKF12) in the φC31 attB site of the chromosome of strain M145 (Fig. 2). Using confocal laser-scanning microscopy, a weak fluorescence was detected in the vegetative mycelium of this strain (Fig. 3I and J). This signal was only slightly stronger than, and hardly distinguishable from, the autofluorescence that was produced by strains carrying only the vector or a plasmid with a promoterless egfp gene (Fig. 3C and D; data not shown). Therefore, it was not meaningful to compare the intensity of the signal produced in the vegetative mycelium between different strains containing different Φ(ftsZ–egfp) constructions. However, examination of the aerial mycelium revealed intense fluorescence in hyphae that contained sporulation septa (Fig. 3E–H), whereas no strongly fluorescent aerial hyphae were detected in strains carrying the promoterless egfp gene or only the vector (Fig. 3A and B; data not shown). Significantly, some not overtly septated aerial hyphae were also strongly fluorescent in strains carrying pKF12, and these were assumed to be at the early stages of sporulation (without specific staining, one can only infer septation once the spores begin to round off). The bases of this assumption were that such strongly fluorescent hyphae were always in the surface layer of the colony and never in the deeper substrate mycelium, and that they had the smooth, non-branched and slightly thicker appearance characteristic of young sporulating aerial hyphae. In the fluorescent hyphae, fluorescence extended right up to the tip, often from an abrupt border, below which the signal was much weaker (see arrows in Fig. 3H). Thus, the signal showed the same type of distribution as ladders of sporulation septa or FtsZ-rings in aerial hyphae (Schwedock et al., 1997; Flärdh et al., 1999).

Figure 2.

Physical map of the ftsQ–ftsZ genes (top line) and structure of plasmids used in this work. The protein coding regions are indicated by open bars. Putative promoters identified in this work are indicated by arrows. The Δ symbol denotes a deletion of the −10 motif of the ftsZ2p promoter (nucleotides 4150–4155 as numbered in Fig. 1A). The first nucleotide at the 5′ ends of inserts in plasmids pKF20/23 and pKF21/24 are positions 4101 and 4160 respectively. The egfp gene to which the ftsZ promoter region was fused is represented by the striped box. For the Φ(ftsZ–egfp) fusions, it is shown whether strong fluorescence in sporulating hyphae was detected when the respective fusion was placed in the chromosome of S. coelicolor A3(2) strain M145, and surface-grown cultures were analysed by confocal laser-scanning microscopy (see Fig. 3). Restriction sites for relevant enzymes that were used in plasmid construction are shown: S, SmaI; W, BsiWI; B, BclI; and St, StuI. Further details are given in Experimental procedures.

Figure 3.

Confocal microscope images of strains containing the ftsZ promoter region fused to the egfp reporter gene. Fluorescence images are at the right or bottom in each pair, whereas images obtained with transmitted light are at the left or top. The strains shown are S. coelicolor A3(2) strain M145 carrying only the vector pIJ8600 (A–D); pKF12, which has an Φ(ftsZ3p2p1p–egfp) fusion (E–J); pKF23, which has an Φ(ftsZ2p1p–egfp) fusion (K and L); pKF24, which has an Φ(ftsZ1p–egfp) fusion (M and N); and pKF25, which has an Φ(ftsZ3p1p–egfp) fusion (O and P). Images show aerial mycelium in the process of sporulation (A, B, E–H, K–P) or vegetative mycelium (veg: C, D, I and J). Small arrows indicate examples of the frequently observed abrupt border, below which fluorescence was much weaker. Large arrowheads indicate examples of aerial hyphae that are fluorescent, although not showing obvious sporulation septation in transmission illumination. Bar = 10 µm.

A strain carrying pKF23, in which the upstream ftsZ3p promoter had been deleted, so that only ftsZ2p and ftsZ1p could promote egfp expression (Fig. 2), had a similar pattern of intense fluorescence exclusively in overtly sporulating aerial hyphae and some young aerial hyphae (Fig. 3K and L). Arrowheads in Fig. 3K and L indicate two examples of aerial hyphae that are fluorescent, although not showing obvious sporulation septation in transmission illumination. The pKF23 result showed that the developmental induction of ftsZ transcription did not require any upstream promoters and, therefore, that the increase in ftsZ2p signal strength that was detected by S1 mapping was not caused by processing of transcripts originating at other promoters. In pKF24, ftsZ2p was removed by deleting all sequence upstream of its putative transcriptional start site and, in pKF25, ftsZ2p was inactivated by deletion of the 6 bp that constitute the −10 motif (Fig. 2). No aerial hyphae with fluorescence significantly higher than the background level were found in strains carrying pKF24 or pKF25 (Fig. 3M–P). These results, together with the S1 mapping, demonstrated that ftsZ2p is a promoter and that it is strongly upregulated preferentially in aerial hyphae that go into sporulation. The observation of a number of not overtly septated aerial hyphae with strong fluorescence was consistent with this induction occurring at an early stage of sporulation.

The ftsZ2p promoter is required for sporulation septation

As the ftsZ2p promoter was activated specifically in sporogenic aerial hyphae, it seemed possible that its action was important for sporulation septation. To test this possible role of ftsZ2p, we wished to construct a strain in which it was inactivated by a mutation. For this purpose, plasmids were constructed carrying ftsZ either with all of its promoters intact or with ftsZ2p inactivated by a deletion of the −10 sequence motif (Fig. 2, pKF32 and pKF33). Plasmid pKF32 containing an intact ftsZ gene was integrated into the chromosomal φC31 attB site of the ftsZ null mutant HU133 to create strain J2417. All phenotypic effects of the ΔftsZ::aphI mutation, such as defects in vegetative growth and the overproduction of blue pigment, were reversed, and the resulting strain grew and sporulated as well as the wild-type strain M145 (Fig. 4). When plasmid pKF33, lacking ftsZ2p, was introduced into HU133 to yield strain J2418, no mRNA originating at the ftsZ2p was detected by S1 nuclease protection assays (data not shown). Nevertheless, vegetative growth was normal, as judged from the fact that colonies developed equally fast and to the same size as those of M145 or J2417 (Fig. 4). The restoration of vegetative septation in this strain was confirmed (data not shown) by staining of cell walls with a fluorescent wheat germ agglutinin conjugate as described previously (Flärdh et al., 1999). Also, the overproduction of a diffusible blue pigment (γ-actinorhodin) that is observed in HU133 was suppressed by both pKF32 and pKF33 (Fig. 4). Aerial mycelium was formed to a similar extent in J2417 and J2418, but it was significantly less grey in J2418, suggesting a defect in sporulation (Fig. 4). Microscopic examination of aerial mycelium revealed that J2417 formed normal spore chains, whereas J2418 formed long, often slightly coiled or wavy aerial hyphae instead of spore chains (Fig. 5). Just as in wild-type spore chains, the apical parts of many of these aerial hyphae were markedly darker in phase-contrast illumination than normal aerial or vegetative hyphae (Fig. 5C and E). There were, however, almost no overt signs of sporulation septa in such phase-dark parts of aerial hyphae. Only occasionally were phase-dark fragments found that could have been divided off from an aerial hypha or that appeared to contain a sporulation septum (see arrow in Fig. 5C). Staining of DNA in aerial hyphae using DAPI showed that many of the phase-dark aerial hyphae contained condensed DNA irregularly separated by DNA-free spaces into nucleoids of different sizes (Fig. 5D and F). The variation in the degree of DNA condensation in Fig. 5D and F probably reflects the age or stage of development of the hyphae. The DAPI-stained spore chains of J2417 (Fig. 5B) were indistinguishable from spore chains of the parent strain M145. In summary, inactivation of ftsZ2p prevented the formation of ladders of sporulation septa, although many aerial hyphae acquired sporulation-related characteristics such as becoming dark in phase-contrast illumination and having condensed DNA.

Figure 4.

Effects on colony appearance caused by inactivation of the ftsZ2p promoter. Colonies developed for 4 days on MS agar before being photographed. M145 is a morphologically wild-type strain of S. coelicolor A3(2). HU 133 is a ΔftsZ::aphI derivative of M145 (McCormick et al., 1994). J2417 is HU133 with pKF32 (ftsZp+) integrated into the φC31 attB site, and J2418 is HU133 with integrated pKF33 (ftsZΔ2p).

Figure 5.

Effects on morphology of aerial hyphae caused by inactivation of the ftsZ2p promoter. An S. coelicolor A3(2) ftsZ null mutant (HU133) containing either plasmid pKF32 with ftsZ and its promoter region (strain J2417; A and B) or plasmid pKF33 with an ftsZ allele lacking the ftsZ2p promoter (strain J2418; C–F). Phase-contrast microscopy images are on the left (A, C and E), and fluorescence images of DAPI-stained hyphae are on the right (B, D and F). Arrows indicate an example of an apparent sporulation septum in J2418. Bar = 10 µm.

Upregulation of the ftsZ2p promoter during sporulation depends on six known regulatory whi genes

Six genes, whiA, B, G, H, I and J, are required for the normal formation of sporulation septa (Chater, 1999) and for the expression of some late sporulation genes (Kelemen et al., 1996; 1998). To determine whether these whi genes affected ftsZ transcription during development, total RNA was prepared from whi mutants at different times of development on MS agar medium. An isogenic series of constructed null mutants, all derived from the morphologically wild-type strain M145, was used for this purpose (Table 1). The transcription of ftsZ was examined by S1 assays in the same way as was done for the analysis of M145 shown in Fig. 1. The main conclusion was that the ftsZ2p promoter was not strongly upregulated during development in any of the six mutants (Fig. 6).

Table 1. Streptomyces coelicolor A3(2) strains used in this study.
StrainGenotypeReference
HU133M145 ΔftsZ::aphI McCormick et al. (1994)
J2400M145 whiG::hyg Flärdh et al. (1999)
J2401M145 whiA::hyg Flärdh et al. (1999)
J2402M145 whiB::hyg Flärdh et al. (1999)
J2408M145 ΔwhiH::ermE Flärdh et al. (1999)
J2417HU133 attBφC31::pKF32[ftsZ+]This paper
J2418HU133 attBφC31::pKF33[ftsZΔ2p]This paper
J2421M145 attBφC31::pKF11[promoterless egfp]This paper
J2422M145 attBφC31::pKF12[Φ(ftsZ3p2p1p–egfp)]This paper
J2423M145 attBφC31::pKF23[Φ(ftsZ2p1p–egfp)]This paper
J2424M145 attBφC31::pKF24[Φ(ftsZ3p1p–egfp)]This paper
J2425M145 attBφC31::pKF25[Φ(ftsZ1p–egfp)]This paper
J2450M145 whiI::hyg Aínsa et al. (1999)
J2452M145 whiJ::hygJ. A. Aínsa
M145Prototrophic, SCP1 SCP2 Pgl+ Hopwood et al. (1985)
Figure 6.

S1 nuclease protection analysis of ftsZ transcription in S. coelicolor mutants with disruptions of different whi genes. Assay conditions and annotations in the figure are exactly as in Fig. 1, except that the gel was run for a much shorter time.

In the whiG mutant J2400, the pattern of protected fragments did not change significantly over the time period tested and resembled that of the vegetatively growing parent strain (Fig. 6; data for ftsZ1p not shown). In J2401 (whiA) and J2402 (whiB), the ftsZ2p activity was much reduced (although a weak band was detected upon overexposure of the gel) and did not change during aerial mycelium formation (Fig. 6). Also, in the whiH mutant J2408, the ftsZ2p signal did not increase during development (Fig. 6). At 28 h, J2450 (whiI) had relative activities of ftsZ2p and ftsZ3p similar to those seen during vegetative growth of the parent strain M145 (Figs 1 and 6). Coinciding with the appearance of abundant aerial mycelium, an increase in ftsZ2p activity was observed, but this was much smaller than that observed at the onset of sporulation in the wild-type parent and, at 64 h, ftsZ3p and ftsZ2p were of similar intensity again. The whiJ mutant J2452 develops more slowly than the other strains and forms mostly undifferentiated aerial hyphae, although occasional spore chains eventually appear (N. J. Ryding and J. A. Aínsa, personal communication). The relative ftsZ2p activity in J2452 was comparable with that of vegetatively growing M145 and was not upregulated detectably during development (Fig. 6), even after prolonged incubation up to 90 h (data not shown). We conclude from these experiments that all six early whi genes are required for the marked upregulation of ftsZ2p that occurs during sporulation of S. coelicolor A3(2). Although ftsZ2p activity was much reduced in J2401 and J2402, it was not absolutely dependent on any of the whi genes.

Discussion

In streptomycetes, there are at least two kinds of septa. Vegetative septa divide growing substrate hyphae into multigenomic compartments and would seem to require a lower FtsZ to genome ratio than would be typical of unicellular bacteria. On the other hand, the synchronous subdivision of long aerial hyphae into chains of several tens of prespore compartments would probably require a high FtsZ–genome ratio. The data that we report show that these different requirements for FtsZ are (at least largely) met by the use of differentially active promoters for ftsZ.

Transcription of the ftsZ locus was complex. Three mRNA 5′ ends were mapped to the ftsQ–ftsZ intergenic region (designated ftsZ3p2p1p), and there was readthrough from within or upstream of ftsQ. FtsZ3p has putative −10 and −35 sequences, and ftsZ1p has a possible −10 motif, but further studies are required to clarify whether these are promoters or represent mRNA processing sites. Unlike the situation in E. coli (Flärdh et al., 1998), no extensive readthrough from upstream promoters in the dcw gene cluster was required for vegetative cell division, and the growth-associated requirements for FtsZ could be met by ftsZ3p, ftsZ1p and other putative promoters within or downstream of ftsQ. This is consistent with the previous observation that plasmid pJRM11 containing only ftsZ and the ftsQ–ftsZ intergenic region could fully complement an ftsZ null mutant (McCormick and Losick, 1996).

The focal point of this paper is that, coinciding with the onset of sporulation, ftsZ2p was strongly upregulated and that this occurred specifically in sporogenic aerial hyphae. Upstream of the ftsZ2p mRNA 5′ end are −10 and −35 motifs closely resembling those of the σHrdB-like class of promoters (Brown et al., 1992; Strohl, 1992; Bourn and Babb, 1995; Kang et al., 1997). A promoter element required for induction of Φ(ftsZ–egfp) in aerial hyphae was pinpointed to within positions −61 to −2 in relation to the ftsZ2p start site (these are the different 5′ ends of fragments fused to egfp in plasmids pKF23 and pKF24; Fig. 2). Furthermore, specific deletion of the −10 motif led to loss of the ftsZ2p signal in S1 nuclease protection assays and prevented detectable expression of Φ(ftsZ–egfp) in aerial hyphae (pKF25; Fig. 2). These data demonstrate that ftsZ2p is a developmentally controlled promoter. The near absence of any sporulation septa from a strain lacking ftsZ2p implies that transcription from this promoter is crucially and specifically important for sporulation septation. In agreement with this role for ftsZ2p, its upregulation coincided approximately with the onset of sporulation septation (Fig. 1). In addition, confocal microscopy of strains carrying appropriate egfp transcriptional fusions to the ftsZ promoter region (Fig. 3) confirmed that the temporal increase in ftsZ2p activity was spatially localized to aerial hyphae undergoing sporulation. Probably, the upregulation of ftsZ2p in any one aerial hypha is transient, but two technical limitations prevent the easy demonstration of this. First, in the direct detection of ftsZ2p-generated transcripts by S1 mapping, it is necessary to use large populations of developmentally unsynchronized hyphae, masking the fine details of regulation within any one hypha. Secondly, the high stability of EGFP protein probably accounts for the high fluorescence observed in even quite mature spore chains of strains carrying ftsZ2p–egfp fusions.

In E. coli, the level of ftsZ expression has to be tightly controlled for cell division to proceed normally. Decreased expression appears to delay division (Palacios et al., 1996), whereas moderate increases are sufficient to overcome the inhibitory effect of the MinCDE system so that septa can be laid down at cell poles as well as at the cell centre (Bi and Lutkenhaus, 1990; Begg et al., 1998). The pronounced upregulation of ftsZ2p in aerial hyphae is required for sporulation, but it is not clear whether this is sufficient to trigger the multiple sporulation septation. Other levels of control could also be involved. In E. coli, the stoichiometry between FtsA and FtsZ is critical for division (Begg et al., 1998) and, although FtsA has not yet been identified in streptomycetes, other division genes may have to be under similar developmental control to ftsZ for spores to be formed. In addition, post-transcriptional mechanisms (e.g. proteins nucleating or stabilizing FtsZ polymers) may also be involved in formation of the Z-ring at the correct time and positions in aerial hyphae. In this respect, it is noteworthy that, although one ftsZ promoter is developmentally regulated in Bacillus subtilis (Gholamhoseinian et al., 1992; Gonzy-Tréboul et al., 1992), this is not essential for sporulation. Other mechanisms, dependent on the Spo0A regulator, are required to shift from mid-cell Z-ring formation to the polar rings that lead to establishment of the prespore (Levin and Losick, 1996).

Null mutations in any of six developmental regulatory genes (whiA, B, G, H, I and J) greatly reduced or eliminated the developmental increase in ftsZ2p transcripts, consistent with the deficiency in FtsZ levels detected by immunofluorescence in some of the mutants (Schwedock et al., 1997; K. Flärdh and K. F. Chater, unpublished observations). How might these early whi genes control the expression of ftsZ2p? Although the developmental induction of ftsZ2p required whiG, there was still a basal level of ftsZ2p mRNA even in the absence of the whiG-specified σ factor, and the promoter motifs do not resemble the σWhiG consensus sequence (Tan and Chater, 1993; Ryding et al., 1998; Tan et al., 1998; Aínsa et al., 1999). It is therefore likely that ftsZ2p is recognized by σHrdB, rather than by the sporulation-specific σWhiG. A whiG-dependent gene could then be required for upregulation of ftsZ2p. Two candidates are whiH and whiI, which both prevented marked induction of this promoter. In our previous work (Ryding et al., 1998; Aínsa et al., 1999; 2000; Flärdh et al., 1999), we have proposed that WhiA and WhiB are involved in shutting down continued growth of aerial hyphae before sporulation septation, and that the consequent growth cessation generates signals that are recognized by, and change the behaviour of, WhiH and WhiI proteins. As a result, WhiH and WhiI can activate genes responsible for sporulation septation. In this model, WhiH and WhiI (both of which belong to families of known transcription factors) are candidates as activators of ftsZ2p. Moreover, it is striking that the microscopic phenotype of aerial hyphal tips in J2418 (lacking ftsZ2p) is almost the same as that of whiH mutants (Ryding et al., 1998; Flärdh et al., 1999), including the condensation of nucleoids into discretely staining areas separated by almost equal areas that are DNA free. None of the other early whi mutants show any DNA condensation. We therefore suggest that WhiH may play a significant part in the sporulation-associated upregulation of ftsZ2p. In addition, it is possible that whiA and whiB have effects not mediated by whiH or whiI, as knock-outs of whiA or whiB had more pronounced effects on ftsZ2p than whiH and whiI and reduced its activity to a barely detectable level.

The availability of a strain (J2418) with a specific non-regulatory block in sporulation septation provides a tool for investigating whether these septa provide a morphological checkpoint for later stages of sporulation, as is the case in endospore formation in B. subtilis (Levin and Losick, 1999). The reduced grey pigmentation of the aerial mycelium of J2418 is a first indication that late sporulation processes may depend on the extent to which sporulation septa are formed. Future molecular analysis of the effects of sporulation septation deficiency on the expression of late regulatory genes such as sigF (Potúckováet al., 1995; Kelemen et al., 1996) and whiD (Molle et al., 2000) will be interesting in this respect.

The ftsZ gene is dispensable in S. coelicolor, and sporulation septation is separated from vegetative growth in time and space. These advantages over other bacterial systems can be exploited for general studies of FtsZ function, using the knowledge about transcriptional regulation that is presented here. For example, vegetative growth can be supported by expression of a normal allele of ftsZ from ftsZ3p and ftsZ1p, whereas any mutagenized ftsZ allele may be specifically expressed in sporulating aerial hyphae from ftsZ2p or another sporulation promoter on a chromosomally integrated plasmid vector.

Experimental procedures

Bacterial strains and growth conditions

The strains of S. coelicolor A3(2) used in this study are listed in Table 1. E. coli strain DH5α (Hanahan, 1983) was used for plasmid construction, and strain ET12567 (MacNeil et al., 1992) containing pUZ8002, an RK2 derivative that is not self-transmissible but can mobilize other plasmids efficiently (M. S. B. Paget, personal communication), was used for transferring non-methylated plasmid DNA into Streptomyces by intergeneric conjugation (Flett et al., 1997). Media and general techniques for growth and transformation of bacterial strains and manipulation of DNA were as described for Streptomyces (Hopwood et al., 1985) and E. coli (Sambrook et al., 1989). Surface cultures for studies of sporulation-related phenotypes were grown on either agar minimal medium MM (Hopwood et al., 1985) with 0.7% mannitol as carbon source or mannitol soymeal (MS) agar (Hobbs et al., 1989).

RNA preparation and S1 nuclease protection assays

Cultivation of mycelium on cellophane membranes and preparation of total RNA at different time points during development were performed as described previously (Aínsa et al., 1999), except that cultures were grown on MS agar. S1 nuclease protection assays were performed as described by Kelemen et al. (1996), except that the probe was created by polymerase chain reaction (PCR) using oligonucleotide KF52 (CGACACCGATGACTTTGATGA) radiolabelled using T4 polynucleotide kinase, oligonucleotide FOR20 (CGCCAGGGTTTTCCCAGTCA) and DNA template pZS1 (Fig. 2), a plasmid created by inserting an 850 bp BclI fragment from pRJ92 (McCormick et al., 1994) into the BamHI site of pBluescript-II SK+. The probe derived from pZS1 had 291 nucleotides complementary to ftsZ transcripts and a 147 nucleotide tail of vector sequence. Dried gels with the S1 assays were exposed and analysed using a PhosphorImager (Molecular Dynamics), and resulting densitograms were processed for publication using adobe photoshop software.

Construction and analysis of egfp fusions

A HindIII–StuI fragment from pRJ92 (McCormick et al., 1994) was ligated into pEGFP-1 (Clontech Laboratories), which had been digested with PinAI, treated with Mung bean nuclease and digested with HindIII. The resulting plasmid pKF10 (Fig. 2) had an Φ(ftsZ–egfp) fusion, in which the ftsZ Shine–Dalgarno sequence GAGG was followed by 8 nucleotides derived from the pEGFP-1 polylinker and the ATG start codon of egfp, instead of the normal 8 nucleotide spacer between GAGG and the ftsZ GTG start codon. The sequence of the fusion joint was confirmed by DNA sequencing. Deletion derivatives of pKF10, which lacked one or both of the ftsZ3p2p promoters (Fig. 2), were constructed by PCR. Using 5′-phosphorylated oligonucleotides diverging from the region to be deleted, the remaining part of pKF10 was amplified and religated to produce plasmids pKF20, pKF21 and pKF22 (Fig. 2). The sequences of the oligonucleotides used for this purpose are available from the authors on request. The promoterless egfp of pEGFP-1 and the Φ(ftsZ–egfp) fusions in pKF10, pKF20, pKF21 and pKF22 were transferred to pIJ8600 (Sun et al., 1999) as BglII–NotI fragments to yield plasmids pKF10, pKF11, pKF22, pKF24 and pKF25 respectively. The structure of the ftsZ promoter region in these plasmids was confirmed by DNA sequencing. Upon introduction into the BglII–NotI sites of pIJ8600, the tipA promoter was deleted from the vector, and the Φ(ftsZ–egfp) fusions were placed immediately downstream of the fd terminator to prevent possible contributions from plasmid promoters to the expression of egfp. The plasmids pKF11, pKF12, pKF23, pKF24 and pKF25 were introduced into strain ET12567/pUZ8002 by transformation and then to S. coelicolor A3(2) strain M145 by conjugation.

Inactivation of ftsZ2p promoter

Plasmid pKF29 (Fig. 2) was constructed by ligating a BsiWI fragment from the ftsZ-containing cosmid C69 (Redenbach et al., 1996) into the Asp718I site of pIJ2925 (Janssen and Bibb, 1993). Plasmid pKF22 (see above) contained a deletion of the 6 nucleotides that comprise the putative −10 region of the ftsZ2p promoter. This mutant ftsZΔ2p promoter was placed in front of ftsZ by removing an EcoRI–StyI fragment from pRJ92 and replacing it with an EcoRI–StyI fragment from pKF22 to yield pKF26. The promoter mutation was then exchanged for the wild-type ftsZ2p promoter in pKF29 by exchanging an 850 bp BclI fragment between pKF29 and pKF26. From the resulting plasmid pKF31 and from the parent pKF29, ftsZ and most of the upstream ftsQ gene were cut out using the BglII sites that flank the polylinker region of the vector pIJ2925 and ligated into the BamHI site of pSET152 to yield pKF33 and pKF32 respectively (Fig. 2). The presence of the ftsZΔ2p mutation on pKF33 and the ftsZ+ allele on pKF32 was verified by DNA sequencing, as well as by restriction analysis (ftsZΔ2p destroys one DdeI site). The plasmids were transferred to the S. coelicolor ftsZ null mutant HU133 by conjugation as described above. Southern blotting and hybridization with an ftsZ DNA probe verified the integration of pKF32 and pKF33 into the chromosomes of exconjugant strains. Representative exconjugants for each plasmid were selected and named J2417 and J2418 respectively.

Microscopy

Samples for light and fluorescence microscopy were prepared and stained with 4′,6-diamidino-2-phenylindole (DAPI) as described previously (Flärdh et al., 1999). Samples were studied and photographed using a Zeiss Axiophot epifluorescence microscope and Kodak Ektachrome E100S film assuming a film speed of 400 ASA. Colour slides were scanned and processed for publication using adobe photoshop software.

Confocal laser-scanning microscopy was carried out as described previously (Sun et al., 1999) with cultures that had developed on the surface of mannitol MM agar for 42–46 h.

Acknowledgements

We thank Helen Kieser, Joe McCormick and Mark Paget for gifts of strains, and David Hopwood and Joe McCormick for critical reading of the manuscript. K.F.C. and M.J.B. were supported by a grant from the Biotechnological and Biological Research Council (BBSRC) to the John Innes Centre. K.F. was supported by a fellowship from the EU (no. BIO 4CT-965134) and a research grant from the Swedish Natural Sciences Research Council. E.L. was supported by a BBSRC ROPA grant.

Ancillary