A sporulation-specific, sigF-dependent protein, SspA, affects septum positioning in Streptomyces coelicolor

The RNA polymerase sigma factor SigF controls late development during sporulation in the filamentous bacterium Streptomyces coelicolor. The only known SigF-dependent gene identified so far, SCO5321, is found in the biosynthetic cluster encoding spore pigment synthesis. Here we identify the first direct target for SigF, the gene sspA, encoding a sporulation-specific protein. Bioinformatic analysis suggests that SspA is a secreted lipoprotein with two PepSY signature domains. The sspA deletion mutant exhibits irregular sporulation septation and altered spore shape, suggesting that SspA plays a role in septum formation and spore maturation. The fluorescent translational fusion protein SspA–mCherry localized first to septum sites, then subsequently around the surface of the spores. Both SspA protein and sspA transcription are absent from the sigF null mutant. Moreover, in vitro transcription assay confirmed that RNA polymerase holoenzyme containing SigF is sufficient for initiation of transcription from a single sspA promoter. In addition, in vivo and in vitro experiments showed that sspA is a direct target of BldD, which functions to repress sporulation genes, including whiG, ftsZ and ssgB, during vegetative growth, co-ordinating their expression during sporulation septation.


Introduction
Streptomyces coelicolor is a multicellular, filamentous bacterium that is a model organism for both bacterial development and antibiotic production. Unlike in most bacteria, cell division in S. coelicolor is incomplete during vegetative growth because of the lack of regular divisional septa formation followed by cell-cell separation, which results in long, branched, multigenomic filaments that grow across and into a solid medium. In response to yet unknown signals, aerial filaments break the surface tension and branch away from the network of substrate mycelium (Elliot et al., 2008;McCormick and Flardh, 2012;Swiercz and Elliot, 2012). Formation of aerial hyphae depends on a set of regulatory genes, called the bld genes, encoding an RNA polymerase sigma factor BldN, transcription factors, such as BldC and BldD or a rare tRNA, bldA (Elliot et al., 2008;Swiercz and Elliot, 2012). In addition to the regulatory genes, the hyphal emergence into the air is facilitated by hydrophobic, cell surface-associated proteins, such as SapB and the 'chaplins', all exhibiting surfactant properties (Claessen et al., 2003;Elliot et al., 2003a;Kodani et al., 2004;Willey et al., 2006).
Initiation of sporulation begins with the cessation of hyphal growth exclusively in the aerial hyphae followed by the separation of the tip-proximal sporogenic compartment and the subapical stem compartment (Kwak et al., 2001;Dalton et al., 2007;Flardh and Buttner, 2009). In the sporogenic hyphae, 50-100 sporulation septa are laid down synchronously at ∼ 1 μm intervals to give rise to the unigenomic pre-spore compartments, while the fate of the subapical stem is cell lysis. Sporulation septation depends on the cell division protein FtsZ, which, unlike in most bacteria, is not essential during vegetative growth in Streptomyces (McCormick et al., 1994). The increased expression of FtsZ in the aerial hyphae is the result of transcription from a second promoter, which is specific to the aerial hyphae and is dependent on the early whi genes (Flardh et al., 2000). Increased levels of FtsZ result in the formation of the initially helical FtsZ filaments followed by FtsZ rings that are spaced evenly along the sporogenic hyphae (Schwedock et al., 1997;Grantcharova et al., 2005). Sporulation septation coincides with chromosome segregation that is governed by the ParA/ParB segregation protein pair (Jakimowicz et al., 2005; together with the proteins, Smc, ScpA and ScpB that are involved in chromosome organization (Dedrick et al., 2009;Kois et al., 2009).
While numerous studies focus on the highly coordinated event of sporulation septation, much less is known about the process of spore maturation. Transformation of the pre-spore compartments into mature spores includes changes in the cell wall structure by the generation of a thick spore wall, changes in the cell shape from cylindrical to ovoid and finally, changes in chromosome organization producing highly compact DNA in mature spores. Cell shape and cell wall synthesis in the spores rely on the bacterial cytoskeletal proteins, MreB and Mbl (Mazza et al., 2006;Heichlinger et al., 2011). In rod-shaped bacteria, such as Escherichia coli or Bacillus subtilis, cell wall synthesis of the lateral cell wall is dependent on members of the actin-like MreB family, including MreB, Mbl and MreBH (Cabeen and Jacobs-Wagner, 2010;White and Gober, 2012). In Streptomyces the MreB-like proteins are not involved in hyphal growth and their role is exclusive to spore development. Lack of both MreB and Mbl results in deformations in spore shape and the absence of the thick spore wall characteristic of the wild-type, mature spores. This suggests that similar to unicellular rod-shaped bacteria, MreB-like proteins are controlling cell wall synthesis during the transformation of pre-spore compartments to mature spores in Streptomyces (Mazza et al., 2006;Heichlinger et al., 2011). We know very little about the enzymes that are involved in the turnover of the spore wall, although as the major component of bacterial cell wall is the three dimensional polymer, peptidoglycan, they are expected to include both peptidoglycan synthases and hydrolases. Recently, a group of cell wall hydrolytic enzymes, SwlA, SwlB and SwlC, have been implicated in spore development (Haiser et al., 2009). Chromosome compaction in the spores relies on nucleoid-associated proteins, such as HupS (Salerno et al., 2009) and the Dps proteins (Facey et al., 2009;2013;. Spore maturation in the late stages of differentiation is governed by an RNA polymerase sigma factor, σ F . Spores of the sigF mutant have thinner walls and smaller sizes and the spore chains of the sigF mutant do not fragment as easily as those of the wild-type strain, suggesting a deficiency in spore-spore separation (Potuckova et al., 1995;Rezuchova et al., 1997). Within the spores, the chromosome of the sigF mutant is not as tightly packed as that of the wild-type strain suggesting that SigF targets will include genes that are involved in spore maturation. Surprisingly, the only SigF target that has been identified to date is the promoter of ORF8 (SCO5321) in the whiE biosynthetic cluster for the dark grey spore pigment production Novakova et al., 2004). While tran-scription of SCO5321 was dependent on sigF, in vitro run-off transcription failed to confirm that SigF was directly involved in initiating transcription from this promoter.
Here we identify a novel spore-specific protein, SspA and we demonstrate that the single promoter of sspA is not only dependent on sigF, but its transcription is directed by RNA polymerase holoenzyme containing SigF, which establishes sspA as the first direct target identified for the sigma factor, SigF.

2D-PAGE of spore extracts identifies SspA as a potential SigF target
In order to characterize the role of SigF in spore maturation in S. coelicolor we compared protein extracts from spores of wild-type, M145 and sigF mutant strains using twodimensional polyacrylamide gel-electrophoresis (2D-PAGE). Due to the thick spore wall of Streptomyces spores, producing protein extracts from spores is not a trivial matter. Sonication, a method that is often reported when presenting proteomics data on spores, only extracted a minority of proteins from the spore wall in our hands and did not release proteins from inside the spores efficiently and reliably. Instead, a mechanical disruption method in which spores were ground in the presence of glass beads under liquid nitrogen, was used to efficiently extract proteins from spores for further analysis. Comparison of protein extracts from spores of the wild-type and sigF mutant strains following separation using a narrow pH-range IPG strip of pH 4.5-5.5 and colloidal Coomassie staining revealed a highly abundant protein in the wild-type extract that was clearly absent in the sigF mutant (Fig. 1). MALDI-TOF analysis identified this spot as the protein SCO7434, with 9 matched peptides covering 33% of the database protein sequence and producing a probabilitybased MOWSE score of 94 (expect value 2.2e-05). We therefore designated this protein as SspA (sporulationspecific protein).

SspA is a lipoprotein with two PepSY domains
The sspA gene (SCO7434) of S. coelicolor has been annotated as a gene encoding a putative lipoprotein with a possible N-terminal signal sequence and prokaryotic membrane lipoprotein lipid attachment site. Indeed, SignalP4.1 (Petersen et al., 2011) identified the first 32 amino acids (aa) as a signal sequence. Moreover, the conserved lipobox motif of L−3-[A/S/T]−2-[G/A]−1-C+1 (LipoP 1.0; Juncker et al., 2003) is present in SspA (LTAC), predicting a lipoprotein signal peptidase cleavage site between amino acids 28 and 29. This suggests that SspA is secreted through the cell membrane and it is attached to the mem-brane, possibly between the lipid bilayer and the peptidoglycan wall. It is consistent with the appearance of SCO7434 in 2D-PAGE gels (Fig. 1) where this protein formed a 'smear' characteristic of lipoproteins, which, during cell lysis, maintain covalent attachment to certain lipids of the cell membrane. In order to gain more information about SspA we performed an iterative BLAST search (Psi-Blast at NCBI; http://www.ncbi.nlm.nih.gov/BLAST .cgi) together with protein domain identification searches using SMART or Pfam (at http://smart.embl-heidelberg.de and at http://www.sanger.ac.uk). According to the SMART and Pfam searches, the 253 aa long SspA contains two PepSY domains between 81 and 167 aa and 188-249 aa. BLAST searches identified homologous domains that are widespread among bacteria, often as the N-terminal propeptide domain of the M4 family of peptidases, which contains the thermostable thermolysins and related thermolabile neutral proteases (Yeats et al., 2004). Bioinformatics analysis of the distribution of PepSY domains among bacteria and fungi has indicated three main groups for PepSY containing proteins (Yeats et al., 2004). The first group contains M4 peptidases; members of the second group have normally one to five copies of the PepSY domain and no other domains apart from a signal peptide signature and the third group contains one or two PepSY domains surrounded by transmembrane helices. S. coelicolor possesses eight PepSY domain-containing proteins, none of which has been characterized up to date. Four proteins, SCO1226, SCO2474, SO5446 and SCO5447, belong to group 1, carrying M4 peptidase domains. Two proteins, SspA and SCO5402 belong to group 2 with two PepSY domains. Finally, SCO0863 and SCO0987 are group 3 proteins with additional transmembrane helices.

The sspA null mutant has irregular spore size
In order to identify the function of sspA in spores, we generated an sspA null mutant by gene replacement using PCR targetting (Gust et al., 2003). We replaced the sspA gene within the cosmid St6D11 with an apramycin resistance cassette and oriT function using lambda Red recombinase (Datsenko and Wanner, 2000;Gust et al., 2003). The mutant cosmid was then introduced into S. coelicolor M145 using conjugation, and the apramycinresistant and kanamycin-sensitive colonies were selected and screened for, in order to generate the sspA null mutant, designated K55. The constructed sspA null mutant was confirmed using Southern hybridization (data not shown).
The macroscopic phenotype of the sspA mutant was indistinguishable from the wild-type when grown both as confluent patches and as single colonies on a wide range of media including SFM medium, minimal medium with mannitol as carbon source or R5 medium. We also tested the sensitivity of the sspA mutant to osmotic, heat, cold or oxidative stress. No significant difference was detected when the wild-type or the sspA mutant was grown in the presence of 0.4 M KCl, 0.4 M NaCl or 0.8 M sucrose (osmotic stress), at 37°C (heat shock) or 16°C (cold shock) or using H 2O2 disks. Testing the spores of the sspA mutant by treatment with 0.1% (v/v) Triton X-100 established that, unlike the sigF mutant (Potuckova et al., 1995), spores of the sspA mutant were not sensitive to detergents (data not shown). However, microscopic observation identified subtle, but significant differences between spores of the sspA mutant and the wild-type. The length of 3-to 6-day-old mature spores of the sspA mutant was uneven, generating many spores longer than those of the wild-type, when grown in minimal medium supplemented with mannitol ( Fig. 2). In addition, there were differences not only in spore size but also in spore shape between the two strains. While wild-type spores became ovoid following sporulation septation, spores of the sspA mutant were less rounded and rectangular-shaped, suggesting that spore maturation was affected in the sspA mutant ( Fig. 2A). Interestingly, the length defects were detectable on several different media, including minimal medium and SFM medium. However, the 'rectangular' spore shape of the sspA mutant was most pronounced when the strain was grown in minimal medium ( Fig. 2A), but less so in SFM medium (Fig. 2B). Statistical analysis of the length and width of mature spores from 3-day-old cultures grown in SFM medium confirmed that spore length of the sspA mutant varied much more than that of the wild-type (Fig. 2C). The mean length of mature spores was 1.17 μm for the wild-type and 1.18 μm for the sspA mutant with standard deviation (SD) of 0.14 and 0.31 respectively. The spore length varied between 0.69 μm and 2.47 μm for the sspA mutant, compared to the smaller range, between 0.87 μm and 1.75 μm, observed for the A. Wild-type M145 and the sspA mutant strains were grown in minimal medium for 72 h and the spores were viewed using phase-contrast microscopy. B. S. coelicolor wild-type M145 and the sspA mutant K55 were grown in SFM medium for 72 h and the spore morphology was assessed using scanning electron microscopy. Arrows mark spores longer than the wild-type average. The scale bar represents 2 μm. C. Histogram of the length and width of mature spores from S. coelicolor M145 (black) and the sspA mutant, K55 (white) grown in SFM medium for 72 h and viewed using phase-contrast microscopy. Measurements of ∼ 500 spores were analysed using the excel package. D. Transmission electron microscopy of the wild-type and sspA mutant spores, which were cultivated as in (B). Scale bars represent 1 μm.
wild-type strain. The higher variance was specific to the length of the sspA spores as the spore-width of both strains was similar with a mean of 0.83 μm, SD = 0.08 (Fig. 2C). Consistent with these measurements, the wildtype spores typically exhibited ovoid shape in SFM medium. In contrast, the wider size range of spore-length of the sspA mutant resulted in a variety of shapes including spherical, ovoid and elongated rods ( Fig. 2B and D). Interestingly, the sigF mutant produces smaller, spherical spores with thinner cell wall and less condensed chromosomes than that of the wild-type (Potuckova et al., 1995). However, the spore wall and the condensed chromosomes of the sspA mutant were very similar to those of the wild-type, when assessed using transmission electron microscopy ( Fig. 2D). This suggests that apart from the spherical spore shape, the characteristic sigF phenotype is not due to the lack of SspA.
To test whether the differences in spore length of the mature spores were the result of inaccurate septum placement or altered spore maturation, we tested spores at earlier stages, after 48 h growth, when sporulation septa were formed. Septa were stained with Alexa488 conjugate of Wheat Germ Agglutinin (WGA-Alexa488) and propidium iodide staining was used to visualize chromosomal DNA (Fig. 3). Measuring ∼ 500 pre-spore compartments for each strain, the mean length of pre-spores was slightly shorter, 1.26 μm, for the sspA mutant compared to 1.28 μm for the wild-type strain. While the mode was the same for both strains, the distance between septa of the sspA mutant had a much higher variance (0.08 versus 0.02) and higher standard deviation (0.29 versus 0.16). Pre-spore length varied over a wider range between 0.34 μm and 2.38 μm for the sspA mutant and between 0.81 μm and 1.99 μm for the wild-type strain, A. Early spore chains of the wild-type M145 and the sspA mutant, K55, were viewed using laser-scanning confocal microscopy after staining the cell wall with WGA-Alexa488 (green; top panel) and the chromosomes with propidium iodide (PI; red; bottom panel) after growth for 48 h in SFM medium alongside microscope coverslips to allow viewing of aerial development and sporulation. Yellow arrows mark aberrant septum formation in the sspA mutant. B. Histogram of the length and width of pre-spores within spore chains of the wild-type (black) and sspA mutant (white) strains. Samples were generated and viewed as in (A). Measurements of ∼ 550 spore compartments were analysed using the excel package.
suggesting that lack of SspA affected both early septum placement and later spore maturation. There was no difference in the width of the hyphae at the time of septation ( Fig. 3B) with a mean of 0.63 μm and SD = 0.05. The sspA mutant phenotype was fully complemented by the introduction of the sspA gene in trans using the single copy, integrative plasmid pMS82 (Gregory et al., 2003), which confirms that the observed phenotype was due to lack of SspA.
The sspA gene is transcribed from a single, sigF-dependent promoter SspA was identified from the 2D-PAGE analysis as a protein missing from the sigF mutant. To confirm that expression of sspA was indeed sigF dependent, we monitored sspA transcription using S1 mapping. mRNA was extracted from both wild-type, M145 and sigF mutant strains and samples were taken at various time points during morphogenesis including vegetative growth, aerial growth and sporulation. Using S1 mapping we were able to show that sspA was transcribed from a single, developmentally regulated promoter. The onset of sspA transcription coincides with that of sigF transcription. We also showed that sspA transcription was dependent upon sigF since sspA transcription was absent from the sigF mutant (Fig. 4).
The single transcription start point for sspA was determined using high-resolution S1 mapping (Fig. 4). Interestingly, the transcription start point for sspA is identical to its translational start point. Initiation of translation in bacteria involves binding of the 16S ribosome to the target mRNA, at a sequence of up to 6 ribonucleotides, which precedes the translational start codon. This sequence, called the Shine-Dalgarno sequence, base pairs with a region of the 16S RNA of the small ribosomal subunit. However, leaderless translational initiation, where both transcription and translation are initiated at the same position, is widely represented among bacteria and archaea (Zheng et al., 2011). Leaderless mRNA has been documented for 13 S. coelicolor genes (Zheng et al., 2011), including absA (Anderson et al., 2001), whiH (Ryding et al., 1998) or vanH, vanK and vanR (Hong et al., 2004), just to name a few. It is not clear what the significance of the leaderless transcripts is in bacteria in general but the faithful translation of leaderless mRNAs in heterologous systems suggests that translation of leaderless mRNA is conserved among all kingdom of life (Moll et al., 2002).

SigF directs transcription from the sspA promoter of S. coelicolor in vitro
Beside sspA, the only other sigF-dependent promoter identified is whiEP2, which drives transcription of SCO5321, encoding an aromatic hydrolase involved in the biosynthesis of the spore-specific, grey pigment of S. coelicolor (Yu and Hopwood, 1995;Kelemen et al., 1998). The sigF-dependent transcription from whiEP2 was proposed to require a transcriptional activator, because RNA polymerase holoenzyme containing SigF alone failed to initiate transcription from whiEP2, in vitro . In order to test whether SigF was sufficient for sspA transcription, an in vitro transcription assay was performed. Previously we have used non-tagged SigF that was overexpressed and purified from E. coli .
Here, for easy purification, we overexpressed SigF with an N-terminal His extension in E. coli using the vector pET28a (Novagen). During the purification trials His-SigF was found in the insoluble fraction and was purified using denaturing conditions of the Ni-NTA kit (Qiagen). In B. subtilis ctc transcription is initiated by SigB (Igo and Losick, 1986) and the fact that ctc promoter activity does not require any additional transcription factors has made this promoter a commonly used template when testing the activities of SigB-like sigma factors, such as SigF  or SigH (Viollier et al., 2003) of S. coelicolor. In an in vitro run-off transcription assay, core RNA polymerase from E. coli together with His-SigF was sufficient to produce a specific transcript from a ctc template and also from an sspA template (Fig. 4C). Thus sspA is the first confirmed direct target for the RNA polymerase sigma factor SigF.

SspA is expressed in the spores
The S1 nuclease assays confirmed that sspA transcription temporally coincided with sporulation and SspA was purified from spore extract. To confirm the spatial location of sspA transcription and to monitor SspA production we generated both transcriptional and translational fluorescent protein fusions. The transcriptional fusion was generated using the plasmid pIJ8660 carrying the egfp gene downstream of a DNA fragment containing the sspA promoter, sspAP, but not the translated DNA region. This sspAP-egfp fusion was then introduced into S. coelicolor M145 and the green fluorescence was monitored throughout development. The sspAP-dependent fluorescence was detected in the spore chains, but there was no detectable fluorescence in the vegetative hyphae or the nonsporogenic part of the aerial hyphae, the subapical stem compartment (Fig. 5A).
To establish localization of the produced protein, we introduced an SspA-mCherry fusion into S. coelicolor M145 using pK39, a derivative of pIJ8668 (Sun et al., 1999). Integration of pK39 into the chromosomal sspA site via homologous recombination replaced the native sspA gene with an sspA-mCherry allele in such a way that no wild-type SspA was produced and the fusion protein was A. S1 nuclease analysis was performed using RNA samples collected at distinct developmental stages of S. coelicolor wild-type M145 (left) or sigF mutant (right) strains. Three different probes specific to sspA, sigF and sigNP2 were used, this latter as a control to assess RNA integrity in the samples. sp indicated sporulating samples as was assessed by microscopy. B. Forty micrograms of RNA from 72-h-old wild-type (M145) samples, grown on SFM medium, were used together with a radioactively labelled probe containing the presumed transcriptional start for sspA together with a sequencing ladder to identify the exact transcriptional start point. C. In vitro run-off transcription was performed by RNA polymerase core enzymes (lanes 1 and 3) and RNA polymerase holoenzyme containing His-SigF (lanes 2 and 4) using the templates ctc (lanes 1 and 2) or sspA (lanes 3 and 4). The 155 nt ctc-specific transcript and the 54 nt long sspA transcripts are marked with arrows. The sizes of the marker DNA fragments generated using pUC19 digested with Sau3AI (M) are shown in basepairs. D. Sequence of the probe used for S1 nuclease analysis is shown. The starts of horizontal single arrows mark the 5′ ends of the oligonucleotides used. An asterisk marks the transcription initiation site that coincides with the translational start site. Translated sequences are italicized. The double headed arrows mark the regions protected by cell extracts (below the sequence; see also Fig. 6) or by purified BldD (above the sequence; see also Fig. 7) in DNase I footprinting assays. Grey highlighting indicates the most highly conserved nucleotides of the consensus BldD target sequence (den Hengst et al., 2010). E. Putative promoter sequence of sspA. The promoter of sspA is compared to the SigF-dependent promoter, whiEP2 together with the predicted consensus for SigB of S. coelicolor (Lee et al., 2004) and the established consensus for SigB of B. subtilis (Petersohn et al., 1999). Asterisks mark the transcription start points. The −10 and −35 promoter sequences are underlined and the BldD box (den Hengst et al., 2010) is highlighted in grey. expressed from the native sspA promoter. The spore morphology of S. coelicolor M145 carrying pK39 resembled that of the wild-type strain (data not shown), confirming that SspA-mCherry was functional. Monitoring fluorescence from the SspA-mCherry translational fusion confirmed that SspA localized to the spore walls (Fig. 5B). Hence the fluorescence SspA-mCherry signal is ring-like (Fig. 5B) in contrast to the fluorescence from the sspAP-egfp fusion that fills the cytoplasm (Fig. 5A). More interestingly, we could detect SspA-mCherry at an earlier stage, at the developing sporulation septa, which was apparent by the regularly spaced invagination of the hyphal wall (Fig. 5C). To colocalize SspA-mCherry and the developing septa we used WGA-Alexa488 staining (Fig. 4D). Interestingly, at the time when fluorescent WGA-Alexa488 was detectable in sporulation septa, SspA-mCherry was already found around the spores (Fig. 5D). This suggests that the early, septal SspA-mCherry localization precedes the stage of septation that is detectable by wheat germ agglutinin.

The promoter of sspA is targeted by BldD
To test for any potential transcriptional regulators targeting the sspA promoter, we performed electrophoretic mobility shift assays. Cell lysate prepared from the wildtype strain grown for 28 h on SFM medium shifted a radiolabelled DNA fragment that contained the sspA promoter (Fig. 6A), suggesting the presence of a regulator of the sspA promoter. The interaction between the sspA DNA fragment and the putative regulator appeared specific, since the gel shift was detected in the presence of non-specific DNA [poly(dI-dC)]. Moreover, excess cold probe DNA abolished the shift (Fig. 6A). Since the DNAbinding activity was detected in cell extracts made from samples of vegetative mycelium, where sspA is not expressed, it seemed likely that this putative regulator might function as a repressor of sspA transcription.
To determine the binding site of this putative regulator, we carried out DNase I footprinting, using the same sspA DNA fragment used for the S1 mapping and EMSA experiments. The crude protein extract from vegetative mycelium of the wild-type strain weakly protected a region of the sspA promoter (Fig. 6B). A sequence in the protected region resembled the 'BldD box' [TnACnnnnnGTnA], the palindromic consensus sequence targeted by the BldD repressor protein (den Hengst et al., 2010), raising the possibility that the sspA promoter was a direct target of BldD. Further, an equivalent cell extract from a bldD null mutant failed to shift the sspA probe in a gel-shift assay (Fig. 6A). Chromatin immunoprecipitation microarray (ChIP-chip) analysis showed that sspA is a direct target of BldD in vivo (Fig. 7A) (den Hengst et al., 2010). To confirm and extend this analysis, the BldD binding site in the sspA promoter was mapped precisely using DNase I footprinting (Fig. 7B). Incubation with purified histidine-tagged BldD protected a ∼ 30 bp region including the sequences that were protected by the cell extracts from vegetative mycelium of the wild-type strain (Fig. 4D). The overlapping protected region spans the putative BldD box and the −10 promoter site, a location consistent with BldD functioning as a repressor of the sspA promoter (Fig. 4D).

SspA controls spore development by affecting septum placement
The SspA protein has been identified as an abundant protein in wild-type spores and is absent from the spores A. Electrophoretic mobility shift assay. A radiolabelled DNA fragment containing the sspA promoter (lane 1) was incubated with increasing amounts (5 and 10 μg of protein) of cell extract from samples collected at an early developmental stage comprising of vegetative mycelium of wild-type S. coelicolor (lanes 2-3), or of the bldD mutant (lanes 5-6). The specificity of the protein-DNA interaction was confirmed by using 10 μg of protein extracts from the wild-type strain (as in lane 3) with the addition of a 50-fold excess of cold probe (lane 4). B. DNase I footprinting of the sspA promoter. The probe (as in A) was treated with increasing amount of DNase I either in the absence (lanes 1-2) or in the presence of 20 μg of protein extracts from wild-type S. coelicolor extracts (lanes 2-4). The generated fragments were analysed next to a sequencing ladder. The protected region is marked by a bar. of a sigF null mutant. Bioinformatic analysis identified both an N-terminal signal sequence and a lipoprotein signature suggesting that SspA is exported through the cell membrane and attached to it, presumably from the outside. The rest of the SspA protein sequence comprises of two PepSY domains that are widespread in bacteria and fungi and found often in the propeptide domains of M4 proteases (Yeats et al., 2004). Where examined, the PepSY motif of the propeptide was shown to function as an intramolecular inhibitor preventing premature activation of the protease (Braun et al., 2000;Tang et al., 2003;Gao et al., 2010). While the M4 peptidases are well characterized, no clear biological role was established for those PepSYfamily proteins that do not possess additional domains with enzymatic activity. The presence of the PepSY domain in this diverse family of secreted and cell wallassociated proteins suggested a regulatory role for protease activity in the local environment of the cell (Yeats et al., 2004). However, recent studies of PepSY-domain transmembrane proteins suggested that these domains A. BldD ChIP-chip data for the 8 kb region spanning the sspA locus in wild-type S. coelicolor (black circles) and the bldD mutant (grey squares). DNA obtained from immunoprecipitation of BldD was labelled with Cy3 and hybridized to DNA microarrays together with a total DNA control that was labelled with Cy5 (den Hengst et al., 2010). Data generated by den Hengst were retrieved from the Gene Expression Omnibus (GSE23401) and were plotted as Cy3/Cy5 ratios (y-axis), as a function of chromosome location around sspA (x-axis). can regulate the activity of proteins other than proteases. For example, the PepSY proteins BqsP and BqsQ of Pseudomonas aeruginosa are proposed to regulate the activity of a two-component system that senses extracellular Fe(II) (Kreamer et al., 2012). Similarly, the PepSY protein YycI regulates the YycFG two-component system in B. subtilis (Szurmant et al., 2007). Structural analysis proposed a common structural fold for the PepSY domain, the beta-lactamase inhibitor protein fold (BLIP) and two other protein folds, suggesting that all these protein folds function as inhibitors by binding a partner domain that is located either within the same protein or on a separate protein (Das et al., 2010).
Interestingly, one of the cortex-lytic enzymes in B. subtilis, SleB, is associated with YpeB, a protein with two copies of the PepSY domain (Boland et al., 2000). YpeB has been implicated in the control of a peptidoglycan amidase, SleB, important during spore germination (Atrih and Foster, 2001). YpeB was required for SleB localization (Chirakkal et al., 2002) and it was suggested that YpeB might recruit SleB to the spore cortex. However, recently the inhibition of SleB by YpeB was demonstrated, suggesting that the PepSY protein YpeB controls SleB activity (Li et al., 2013) perhaps by inhibiting the premature activation of the SleB lytic activity in the spores of B. subtilis.
SspA possesses two PepSY domains but lacks any catalytic peptidase domain, so its domain organization is very similar to that of YpeB. However, germination of the sspA mutant spores was not affected (data not shown). Instead, lack of SspA altered septum formation on all media tested and spore maturation when grown on minimal medium. Therefore, we hypothesize that SspA might control the activity of a specific peptidase or peptidoglycan hydrolase involved in either septum formation or spore maturation. Septum positioning depends on the key cell division protein FtsZ that assembles into 50-100 regularly spaced FtsZ rings marking future septum sites (Schwedock et al., 1997). In E. coli or B. subtilis, FtsZ positioning is regulated by FtsZ antagonist proteins. These include the MinCD complex located at cell poles promoting FtsZ ring formation at mid-cell and the chromosome associated SmlA or Noc proteins, which prevent FtsZ ring formation and therefore septum formation over the chromosomes (Adams and Errington, 2009;Lutkenhaus, 2012). In Streptomyces no apparent homologues of either the Min system or nucleoid occlusion proteins have thus far been identified. Instead, formation of the FtsZ rings is under a positive control by SsgB in Streptomyces (Willemse et al., 2011;Jakimowicz and van Wezel, 2012) while FtsZ protein levels are negatively affected by CrgA (Del Sol et al., 2006, Del Sol et al., 2003. Following on from the formation of the FtsZ rings, very little is known about the recruitment of specific enzymes for septum synthesis in Streptomyces. Interestingly, the FtsW-FtsI protein pair, which is established in septum formation in E. coli, was proposed for FtsZ ring stabilization in Streptomyces (Mistry et al., 2008;Bennett et al., 2009;McCormick, 2009). In addition, the SsgA-like proteins (SALPs) have been implicated in both septum positioning (SsgA and B) and in septum formation (SsgC-G) (Noens et al., 2005;Willemse et al., 2011;Jakimowicz and van Wezel, 2012).
The late sigma factor, SigF is associated with the control of spore maturation, which is a post-septation event. However, the sigF null mutant produces smaller spores than the wild-type (Potuckova et al., 1995). This could be the result of incorrect septum positioning and possibly altered FtsZ placement suggesting that SigF is active when septa are formed. Similarly, the uneven septum positioning in the sspA mutant could arise from altered FtsZ positioning. Alternatively, SspA might influence the recruitment of cell wall lytic or synthetic enzymes after FtsZ ring formation. Interestingly, monitoring SspA localization using an SspA-mCherry fusion confirmed that in addition to its presence in mature spores, SspA accumulated at sporulation septa (Fig. 5B). This suggests that SspA functions as early as septum formation. It will be important to establish the potential link between SspA and the Fts proteins (FtsZ, FtsW and FtsI), CrgA and the SALPs.
After the completion of septation, spore wall synthesis is governed by the cytoskeletal proteins MreB and Mbl (Mazza et al., 2006;Heichlinger et al., 2011). MreB polymers first assemble at sporulation septa, followed by spreading to the entire spore wall (Mazza et al., 2006). This pattern is reminiscent of SspA localization, raising the possibility that SspA positioning might be MreBdependent. On the other hand, lack of SspA affects septation while MreB only controls post-septational events (Mazza et al., 2006) suggesting that SspA might function prior to MreB assembly during spore development. A recent search for MreB partner proteins established a complex interaction pattern among members of the proposed 'Streptomyces spore wall synthesizing complex', SSSC (Kleinschnitz et al., 2011). A knockout mutant of SCO2097, a putative membrane protein identified among the SSSC proteins, produced elongated spores with sensitivity to heat and cell wall-damaging agents (Kleinschnitz et al., 2011). Interestingly, these elongated spores resemble those of the sspA mutant; however, the latter did not exhibit sensitivity to lysozyme or heat (data not shown). In addition, spore wall hydrolytic enzymes have also been shown to affect spore shape (Haiser et al., 2009). It will be of interest to test whether any of the penicillin-binding proteins of the SSSC, including SCO3901, SCO3580 and FtsI (Kleinschnitz et al., 2011) or any of the hydrolytic enzymes, RpfA, SwlA, SwlB and SwlC (Haiser et al., 2009) are targets of or partnered by SspA.

Transcription of sspA is under the control of the sigma factor, SigF and the principal developmental regulator, BldD
Both the production of SspA protein and transcription of sspA are dependent on sigF in vivo, in S. coelicolor (Figs 1 and 4). Moreover, in vitro transcription assays (Fig. 4) confirmed that SigF is sufficient to initiate transcription from the sspA promoter. Hence, in this report we have presented the first example of a SigF target promoter, sspAP where transcription is initiated by RNA polymerase holoenzyme containing the sigma factor, SigF in the absence of any activator. SigF belongs to a group of nine RNA polymerase sigma factors (SigB, F, G, H, I, K, L, M and N) that control response to environmental stresses (SigB, H, I, L and M) or morphological differentiation (SigH, F and N) or in some cases both (SigH and SigB) in S. coelicolor (Potuckova et al., 1995;Cho et al., 2001;Kelemen et al., 2001;Sevcikova et al., 2001;Viollier et al., 2003;Lee et al., 2005). Members of this, so called, SigB-family of S. coelicolor resemble the general stress response sigma factor SigB of B. subtilis (see review, Price, 2000). Predictably, the putative promoter sequence of sspA (GTGT-16N-GGTTAC) resembles the consensus target sequence of SigB both in S. coelicolor and in B. subtilis ( Fig. 4E; Petersohn et al., 1999;Price, 2000;Lee et al., 2004). Interestingly, the weak similarity between the two sigF-dependent promoters, sspA and whiEP2, together with the unusually long spacer between the −10 and −35 sequences of whiEP2 might explain why SigF was not sufficient to initiate transcription from whiEP2 in vitro . Identification of further SigF target promoters is paramount in order to establish a consensus SigF target sequence and to address the fundamental question of how members of the SigB-family with potentially overlapping promoter specificity can control distinct sets of genes in vivo. One such mechanism could include specific activators, as it was proposed for the promoters whiEP2 and nepA, which are putative targets for SigF and SigN respectively Dalton et al., 2007).
Alternatively, transcriptional repressors could restrict expression of target genes both in time and in space, allowing limited access to promoter sites by cognate sigma factors. Interestingly, gel shift assays together with DNase I footprinting demonstrated a DNA binding activity from wild-type cell extracts collected at early stages of development, when sspA was not expressed, suggesting a putative repressor targeting sspA transcription. Both in vivo and in vitro experiments confirmed that this repressor is BldD, a key developmental regulator of Streptomyces morphogenesis. bldD was among the first developmental genes identified in the study of morphological differentiation in Streptomyces (Merrick, 1976). The bldD mutant fails to progress to aerial development and is also blocked in the production of several secondary metabolites (Elliot et al., 1998). BldD, a small DNA-binding protein, has been shown to target, and mainly repress, the transcription of developmental genes, such as the sigma factor genes, whiG, bldN and sigH, during early development while co-ordinating the timing and, in some cases, the location of their expression at later stages of differentiation (Elliot et al., 2001;Kelemen et al., 2001). Recently, the genomewide BldD regulon has been extensively defined by chromatin immunoprecipitation-microarray analysis identifying ∼ 167 transcription units targeted by BldD (den Hengst et al., 2010). The location of the BldD binding site at the sspA promoter is consistent with BldD functioning as a repressor of sspA, perhaps by blocking access of other SigB-like sigma factors to the sspA promoter during early developmental stages. Surprisingly, microarray analysis showed that sspA transcription was not upregulated but absent in the bldD null mutant (den Hengst et al., 2010). This suggests that either the sspA promoter is recognized exclusively by SigF, which is naturally absent from the bldD mutant, or, more likely, no other SigB-like sigma factors capable of initiating sspA transcription were active under the conditions of the microarray analysis. Interestingly, one of the BldD targets identified in this analysis is the gene encoding SCO4677 (den Hengst et al., 2010), which has been demonstrated to bind SigF, potentially functioning as an anti-sigma factor (Kim et al., 2008). Thus BldD is linked to the sigF regulon via sspA, and perhaps also via SCO4677. Moreover, BldD has also been shown to target cell division genes (den Hengst et al., 2010), such as ftsZ, ssgA and ssgB or the smeA-ssfA operon encoding a DNA translocase required for correct chromosome segregation during sporulation (Ausmees et al., 2007). Hence, BldD appears to coordinate the expression of proteins that are required at the time of septum formation, including the expression of SspA.

Experimental procedures
Bacterial strains and growth conditions Escherichia coli DH5α (Hanahan, 1983) was used for routine cloning. E. coli BW25113 carrying pIJ790 (Datsenko and Wanner, 2000) was the host for recombination between the extended apramycin resistance cassette and the target gene to generate knockout mutants. E. coli ET12567 (MacNeil et al., 1992) containing pUZ8002 (Kieser et al., 2000) aided the transfer of plasmids or cosmids from E. coli into S. coelicolor by conjugation. The strains of S. coelicolor used in this work are listed in Table 1. Streptomyces strains were grown at 30°C on SFM (soya flour medium containing 1% mannitol), MM (minimal medium) supplemented with mannitol (0.5%) solid media. pIJ82 is a hygromycin-resistant derivative of pSET152 (Bierman et al., 1992) in which the apramycin resistance gene is replaced with the hygromycin resistance gene (Dalton et al., 2007). All plasmids and oligonucleotides used in this work are shown in Tables 2 and 3 respectively.

Preparation of protein extracts from spores
Spores were harvested from cultures grown on SFM medium according to Kieser et al. (2000) and were washed with 40 mM Tris pH 9.0; 1 mM EDTA; 1 mM EGTA buffer. Washed spores were transferred to a pestle and mortar submerged in liquid nitrogen (typically a 0.5 ml volume of spore pellet was used), and disrupted by thoroughly grinding in the presence of an equal volume of fine glass beads (106 microns; Sigma G8893). The frozen, ground spore powder was then transferred on ice into the above buffer containing 100 mM DTT and 4 mM Pefabloc SC protease inhibitor and protector solution (Roche). On warming to approximately 4°C, SDS was Knockout generation Datsenko and Wanner (2000)  This plasmid integrates as a single copy at the ΦC31 attB attachment site on the chromosome of S. coelicolor Sun et al. (1999) pIJ8668 This plasmid cannot replicate autonomously in S. coelicolor but it can integrate into the S. coelicolor chromosome via homologous recombination through the appropriate inserted sequences. Sun et al. (1999) pIJ82 This plasmid integrates as a single copy at the at the ΦC31 attB attachment site on the chromosome of S. coelicolor Dalton et al. (2007)

2D gel-electrophoresis and protein identification
Protein extracts were separated by 2D gel-electrophoresis as previously described (Hesketh et al., 2002). Briefly, proteins were separated in the first dimension for 120 000 volt-hours using 18 cm IPG strips pH 4.5-5.5 (Amersham Biosciences) using a Phaser isoelectric focusing unit (Genomic Solutions). Focused strips were separated in the second dimension using in-house fabricated 12.5% SDS-PAGE gels and the Investigator 5000 vertical format system from Genomic Solutions. Protein spots of interest were excised manually from colloidal Coomassie-stained gels, and identified by tryptic digestion and MALDI-TOF mass spectrometry as previously described (Hesketh et al., 2002).

Generation of the sspA knockout mutant
The sspA knockout mutant was generated using PCR targeting (Gust et al., 2003) by replacing the entire sspA gene, excluding the translational start and stop codon, with the apramycin resistance cassette. The apramycin resistance gene and oriT was PCR-amplified using the primers 6D11.301 and 6D11.302 and was introduced into cosmid St6D11 in BW25113 (Datsenko and Wanner, 2000) carrying pIJ790 to replace sspA with the apramycin resistance cassette. The cosmid carrying the mutant allele was passaged through the dcm-dam-ET12567 strain (MacNeil et al., 1992) containing pUZ8002 and was introduced into S. coelicolor M145 by conjugation. The double cross-over exconjugants were screened for resistance to apramycin and sensitivity to kanamycin. One of the sspA mutants, which was confirmed by Southern blot hybridization was designated K55.

Complementation of the sspA mutation
A ∼ 7.7 kb BamHI DNA fragment containing the entire sspA gene was first moved from cosmid 6D11 into pUC18 and from this subclone, pAT77, a ∼ 1.3 kb BamHI-MluI fragment carrying exclusively the sspA gene was introduced, via several steps, into pIJ2925 (Janssen and Bibb, 1993) to generate pAT0. The sspA fragment was liberated from pAT0 using BglII, and introduced into the BamHI site of pIJ82 (Dalton et al., 2007), to create pAT1. pAT1 was introduced into S. coelicolor M145 and K55 by conjugation and hygromycinresistant exconjugants, carrying pAT1 integrated to the ΦC31 attachment site on the chromosome, were examined for their phenotype.

S1 nuclease mapping
Mycelium grown on solid medium covered with cellophane discs was collected, and initial extracts were generated by grinding in liquid nitrogen. From these extracts RNA was prepared and S1 nuclease protection assays were performed according to Kieser et al. (2000) using 40 μg RNA in each assay. Probes labelled with 32 P at their single 5′ end were generated using PCR with two oligonucleotides, one labelled at its 5′ end with [γ-32 P]-ATP using T4 polynucleotide kinase. Probes were generated using the following oligonucleotide pairs: sspA probes using 6D11.30.1* and 6D11.30.2; sigF probes using SIGF15 and SIGF16*; sigN probes using SIGN4 and SIGN7* (asterisks mark the labelled oligonucleotides). DNA fragments protected by RNA were separated on 6% sequencing gels. The transcriptional start point of sspA was identified against a dideoxy-sequencing ladder (Amersham Pharmacia biotech; T7 sequencing TM kit) produced using the same labelled oligonucleotide used to generate the probes for S1 analysis.

Generation of transcriptional and translational fusions
Two appropriate primers, sspA-Bam and sspA-Kpn, were used to generate a PCR product carrying a 100 bp sequence upstream of the sspA transcriptional start. This PCR product was introduced into pIJ8660 (Sun et al., 1999) as a BamHI-KpnI fragment generating pK37, which was introduced into S. coelicolor M145 by conjugation. One of the representative apramycin-resistant exconjugants was used to monitor sspA transcription. S. coelicolor M145 carrying pK37 was propagated on medium containing apramycin to maintain selection for the presence of plasmid integrated in a single copy into the ΦC31 attB site in the S. coelicolor chromosome.
To monitor the localization of SspA-mCherry we constructed pK39. First a flexible linker encoding Gly4MetAla was introduced upstream of the egfp in pIJ8668 (Sun et al., 1999) by cloning the annealed Egfplinker1 and Egfplinker2 primers into the BamHI-NdeI sites of pIJ8668. Then the egfp fragment was replaced by an mCherry fragment using NdeI and BsrGI restriction sites. The sspA coding sequences were generated by PCR using the primers sspA-Bam2 and sspA-Nde, which create BamHI and NdeI sites at the respective ends. The BamHI-NdeI sspA fragment was introduced to the linkered pIJ8668 derivative. This generated a translational sspA-mCherry fusion, which, when introduced to wild-type S. coelicolor M145, integrates via homologous recombination at the sspA site in the chromosome.
Fluorescent protein fusions were monitored using a Leica TCS SP2 laser-scanning confocal microscope with a 63×, 1.4NA oil-immersion objective.

In vitro run-off transcription using His-SigF
In order to produce His-SigF the sigF gene was introduced into pET28a (Novagen) as an NdeI-EcoRI fragment from pIJ5894  that was generated previously to overexpress non-tagged SigF. The resulting clone, pK34, was moved into E. coli BL21 (DE3)pLysS (Novagen).
Cultures of BL21(DE3)pLysS/pK34 were grown at 37°C to an OD600 0.4 and, after induction with 1 mM IPTG, for a further 4 h. After harvesting the mycelium, His-SigF was purified under denaturing conditions (8 M urea) using the Ni-NTA Spin Kit (Qiagen) according to the manufacturer's instructions. Fractions eluted from the Ni-NTA column with a buffer of low pH (4.5) were dialysed against 50 mM NaH2PO4 and 50 mM NaCl, pH 8.0 buffer at 4°C. The His-SigF protein was analysed on an 8% SDS-PAGE and stored at −20°C in the presence of 20% glycerol.
In vitro run-off transcription was performed as described by Buttner et al. (1987) using 0.5 μg His-SigF, core RNA polymerase from E. coli (Cambio), [α-32 P]-CTP (3000 Ci mmol −1 ; Amersham Biosciences) and an appropriate DNA template. The 340 bp EcoRI-BamHI fragment of pMI340 (Igo and Losick, 1986) was used to test transcription from the B. subtilis ctc promoter and was expected to produce a 155nucleotide (nt) ctc transcript. The template carrying the sspA promoter was generated by PCR using the 6D11.30.1, DIR oligonucleotide pair and pAT6 as template. pAT6 carries a 7.7 kb BamHI fragment from cosmid St6D11 in the vector pUC18 and the DIR oligonucleotide anneals to pUC18 sequences. The 312 bp PCR product carried 229 bp Streptomyces sequence that was expected to generate an 54 nt sspA transcript in vitro.

Electrophoretic mobility shift assays and DNase I footprinting
The labelled sspA probe from the S1 nuclease assays were used together with either cell extracts from S. coelicolor strains or BldD protein overexpressed and purified from E. coli (den Hengst et al., 2010). S. coelicolor strains were grown on the surface of cellophane discs positioned onto SFM medium for 28 h producing a lawn of vegetative mycelium. The collected cells were lysed by sonication and the cell debris was removed by centrifugation at 15000 g for 20 min. Samples of the supernatant were incubated with the sspA probe at 30°C for 20 min in the presence of 1 μg Poly(dI-dC)•Poly(dI-dC) (SIGMA) in 25 mM HEPES pH 7.5, 4 mM DTT, 1 mM ATP, 10 mM Mg-acetate, 4% glycerol buffer and the protein:DNA complexes were analysed on a 4% acrylamide gel, followed by autoradiography. For DNase I footprinting using S. coelicolor M145 cell extracts, the binding conditions were identical to that of the gel mobility shift assays, and the DNase I digestion was performed as described previously (Kelemen et al., 2001). The dideoxy-sequencing ladder (Amersham Pharmacia biotech; T7 sequencing TM kit) was produced using the same labelled oligonucleotide used to generate the sspA probe. DNase I footprinting in the presence of BldD purified from E. coli was carried out using single end-labelled sspA probes generated by PCR with the primers 7434_F1 and 7434_R1, as described in den Hengst et al. (2010).