Transcriptional analysis of the cell division-related ssg genes in Streptomyces coelicolor reveals direct control of ssgR by AtrA

SsgA-like proteins are a family of actinomycete-specific regulatory proteins that control cell division and spore maturation in streptomycetes. SsgA and SsgB together activate sporulation-specific cell division by controlling the localization of FtsZ. Here we report the identification of novel regulators that control the transcription of the ssgA-like genes. Transcriptional regulators controlling ssg gene expression were identified using a DNA-affinity capture assay. Supporting transcriptional and DNA binding studies showed that the ssgA activator gene ssgR is controlled by the TetR-family regulator AtrA, while the γ-butyrolactone-responsive AdpA (SCO2792) and SlbR (SCO0608) and the metabolic regulator Rok7B7 (SCO6008) were identified as candidate regulators for the cell division genes ssgA, ssgB and ssgG. Transcription of the cell division gene ssgB depended on the sporulation genes whiA and whiH, while ssgR, ssgA and ssgD were transcribed independently of the whi genes. Our work sheds new light on the mechanisms by which sporulation-specific cell division is controlled in Streptomyces. Electronic supplementary material The online version of this article (doi:10.1007/s10482-015-0479-2) contains supplementary material, which is available to authorized users.


Introduction
The mycelial streptomycetes are a model organism for bacterial multicellularity Elliot Songhee H. Kim, Bjørn A. Traag, and Ayad H. Hasan have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s10482-015-0479-2) contains supplementary material, which is available to authorized users. et al. 2008). In the presence of sufficient nutrients, the soil-bound streptomycetes grow by tip extension and branching, producing an intricate network of vegetative hyphae to benefit optimally from the available nutrients. When the conditions become less favourable, e.g. nutrient deprivation ), activation of a morphological differentiation program (termed aerial development) resulting in the production of stress-resistant spores, is essential for survival and dissemination. The decision to enter aerial development is a critical and irreversible one, and is therefore tightly controlled (Chater 2001). The formation of aerial hyphae and spores is an energyconsuming process, whereby programmed cell death results in the dismantling of the substrate mycelium to provide nutrients to the new mycelium (Manteca et al. 2005(Manteca et al. , 2006. Streptomycetes are highly adapted to survive in diverse and complex ecosystems. This is highlighted by the presence in their genomes of more than 20 gene clusters specifying secondary metabolites, and genes encoding around 65 sigma factors and an unprecedented number of sugar transporters and polysugar hydrolases (Bentley et al. 2002;Cruz-Morales et al. 2013;Ohnishi et al. 2008). On solidgrown cultures, streptomycetes undergo a cycle of morphological development, whereby upon nutrient starvation aerial hyphae are formed on top of the vegetative mycelium. These aerial hyphae in turn undergo an extensive cell division event to produce chains of unigenomic spores (Chater 1972). Most of the developmental genes that control aerial development (the so-called whi genes) encode transcription factors (TFs) (Chater 1972;Ryding et al. 1999;Flärdh et al. 1999). More recently, it was shown that many genes involved in nutrient sensing and transport, such as dasR, dasABC and the pts genes, also control development and antibiotic production (Rigali et al. 2006;van Wezel et al. 2009).
The SsgA-like proteins are a family of proteins that control sporulation (Jakimowicz and van Wezel 2012;Traag and van Wezel 2008). Several Streptomyces spp. are capable of producing spores in liquid cultures (recently reviewed in (van Dissel et al. 2014)). SsgA was originally identified as a suppressor of a hypersporulating S. griseus mutant (designated SY1) and shown to be essential for submerged sporulation by this organism (Kawamoto and Ensign 1995;Kawamoto et al. 1997). Overexpression of SsgA in liquid-grown mycelium of S. coelicolor induces mycelial fragmentation and submerged sporulation . The ability of SsgA to enhance fragmentation and protein secretion was applied in industrial fermentations, revealing a significant improvement in yield and fermentation characteristics . The ssgA, ssgB and ssgG genes control the selection of septation sites in S. coelicolor, with ssgA and ssgB essential for sporulation (Keijser et al. 2003;Sevcikova and Kormanec 2003;van Wezel et al. 2000). SsgA dynamically controls the localization of its paralogue SsgB, which in turn recruits FtsZ to septum sites to initiate sporulation-specific cell division (Willemse et al. 2011). In mutants lacking ssgG septa are frequently skipped, resulting in many large spores containing multiple chromosomes that are well segregated (Noens et al. 2005).
Relatively little is known about how ssg gene expression is controlled. Transcription of ssgA has been studied in the model species S. coelicolor and S. griseus. A major difference between these two species, is that S. griseus sporulates in both surface-and liquidgrown cultures, while S. coelicolor only sporulates on solid media. The transcriptional control of early development differs significantly between these two species (Chater and Horinouchi 2003). The same is true for the transcriptional control of ssgRA; in S. coelicolor, transcription is activated by and dependent on SsgR (Traag et al. 2004), while in S. griseus transcription depends on the A-factor pathway-controlled AdpA, and only one of the two promoters is controlled by SsgR (Yamazaki et al. 2003). In submerged cultures of S. coelicolor, ssgA is poorly expressed (Romero et al. 2014). A recent study showed that the transcription of ssgRA and ssgB is controlled by the pleiotropic developmental regulatory protein BldD, providing an important connection between BldD and the control of sporulation-specific cell division (den Hengst et al. 2010). In this study, we further investigated the transcriptional control of ssg genes, including transcriptional dependency on the early whi genes whiA, whiB, whiG, whiH, whiI and whiJ as well as on ssgB. Transcriptional regulators that might directly control ssg transcription were identified using a DNA affinity capture assay. Of these, AtrA was shown to activate the transcription of ssgR, which in turn is required for the transcription of ssgA. This is the first developmental gene that has been identified as a target of AtrA.

Materials and methods
Bacterial strains and culturing conditions E. coli K-12 strains JM109 (Sambrook et al. 1989) and ET12567 (MacNeil et al. 1992) were used for propagating plasmids, and were grown and transformed using standard procedures (Sambrook et al. 1989). E. coli BL21 (DE3) was used as host for protein production. Transformants were selected in Luria broth containing 1 % (w/v) glucose and the appropriate antibiotics.
The Streptomyces strains used in this work and their corresponding references are listed in Table S1. S. coelicolor A3(2) M145 is the parent of the mutants described in this work. The atrA (SCO4118), rok7b7 (SCO6008), slbR (SCO0608) and ssgB (SCO1541) null mutants were described previously by the authors. Sporulation mutants S. coelicolor J2401 (DwhiA), J2402 (DwhiB), J2400 (DwhiG), J2210 (DwhiH), J2450 (DwhiI) and C77 (whiJ point mutant C77) were obtained from the John Innes Centre strain collection. Preparation of media for growth, protoplast preparation and transformation of Streptomyces were done according to standard procedures (Kieser et al. 2000). As solid media we used SFM (soya flour mannitol) to make spore suspensions; R2YE (regeneration media with yeast extract) for regenerating protoplasts and, after addition of the appropriate antibiotic, for selecting recombinants; and minimal medium (MM) to prepare total RNA samples (Kieser et al. 2000). For standard cultivation of Streptomyces in liquid cultures we used YEME (yeast extract malt extract) containing 30 % (w/v) sucrose, TSBS (tryptone soy broth; Difco) containing 10 % (w/ v) sucrose or NMMP (normal minimal medium buffered with phosphate) (Kieser et al. 2000). Microscopy was performed as described previously (Colson et al. 2008). Cultures were checked at regular intervals by phasecontrast microscopy using a Zeiss Standard 25 microscope and colony morphology was studied using a Zeiss Lumar V-12 stereo microscope.

Plasmids and constructs
For routine subcloning, pIJ2925, a pUC19-derived plasmid, was used (Janssen and Bibb 1993). Plasmid DNA was isolated from ET12567 prior to transformation to Streptomyces. For selection of pIJ2925 in E. coli, ampicillin (100 lg/ml) was used; chloramphenicol (25 lg/ml) was added to select for growth of ET12567. The entire coding regions of the genes SCO0608 and SCO6008 were PCR-amplified from genomic DNA of S. coelicolor M145 using oligonucleotides described in Table S2; subsequently, they were cloned into expression vector pET24 ma (Novagen) using EcoRI and HindIII restriction sites, to allow the production of C-terminally hexahistidine-tagged versions of the corresponding proteins. The construct for the production of AtrA (encoded by SCO4118), which was N-terminally hexahistidine tagged, has been described previously (Uguru et al. 2005).
RNA isolation and semi-quantitative and quantitative RT-PCR analyses All oligonucleotides used for RT-PCR reactions are described in Table S2. For transcriptional analysis of ssgA-ssgG in surface-grown developmental (whi) mutants, mycelium grown on solid MM with mannitol (0.5 % w/v) on cellophane discs was harvested as indicated at three time points corresponding to vegetative growth, early aerial growth, and late aerial growth or, in the case of M145, spore formation. Phase-contrast light microscopy was used to assess the developmental stage of the surface-grown cultures prior to harvesting mycelium and isolating RNA. Semi-quantitative reverse-transcriptase PCR (RT-PCR) analysis was carried out using SuperScript III one-step RT-PCR System (Invitrogen) as described previously ). For each RT-PCR reaction 200 ng of RNA was used together with 0.5 lM (final concentration) of each oligonucleotide. Samples were then analysed by electrophoresis using a 2 % agarose gel in TAE buffer. RT-PCR experiments without prior reverse transcription were performed on all RNA samples to assure exclusion of DNA contamination. As controls, mock reactions were carried out in the absence of the reverse transcriptase. Quantification of the RT-PCR results was done by scanning the gels using the GS-800 imaging densitometer followed by analysis using Quantity One software (Bio-Rad). 16S rRNA levels were analysed and quantified as a control, and values obtained for the ssg genes were corrected for slight differences in the 16S rRNA levels in the corresponding RNA extracts.
Quantitative PCR analysis was carried out following reverse transcription that used SuperScript II, as instructed by the vendor (Invitrogen), in combination with 500 ng of DNase I-treated total RNA as template and 50 ng of random hexamers (Applied Biosystems) as primer in 20 ll reactions. Mock reactions without the reverse transcriptase were also conducted. At the end of each of the reactions, 80 ll of yeast tRNA (10 ng/ml; Ambion) was added, and 2 ll aliquots analysed by PCR using a SensiMix TM SYBR No-ROX kit, as instructed by the vendor (Bioline), in combination with 0.5 lM primers and 3.5 mM MgCl 2 . Reactions were assembled in 0.1 ml strip tubes and analysed using a real-time cycler (Corbett Rotor-Gene 6000). The PCR conditions were 10 min at 95°C, followed by 40 cycles of 10 s at 95°C, 15 s at 60°C and 15 s at 72°C. The products were analysed by determining melting curves and confirming the sizes of the amplicons by gel electrophoresis. A threshold common to the exponential phases of all the reactions in a single run was selected manually and the corresponding number of cycles for each reaction recorded. These CT values were then expressed as the difference relative to rpsL (SCO4735), which was used as the internal control. In turn, the DCT values were used to calculate the difference in abundance using the Equation 2 DCT .

DACA assay
The DACA procedure was described previously (Park et al. 2009). In brief, streptavidin Dynabeads (Dynal Biotech, Oslo, Norway) were washed and incubated with annealed oligonucleotides (100 pmol DNA/mg of beads) for 30 min at room temperature, and biotin (100 lg/ml) was subsequently added and further incubated for 15 min. 1 ml of a pre-mixed cell extract solution (500 lg/ml) containing sheared salmon sperm DNA (0.1 mg/ml) was incubated for 15 min on ice and then added to 100 ll of bead solution (0.5 mg beads and 0.1 mg cellular protein). Following 40 min incubation at room temperature, the beads were washed with standard buffer and then resuspended in 100 mM NH 4 HCO 3 (pH 8). The captured protein mixtures were then digested with trypsin, and tryptic peptides were analysed by liquid chromatography-tandem mass spectrometric analysis (LC-MS/ MS) using an LTQ-orbitrap mass spectrometer (Thermo Finnigan/Thermo Fisher, Waltham, MA, USA). Peptide ions were detected in full scan mode from 400 to 1700 m/z followed by three data-dependent MS/MS scans (isolation width 1.5 m/z, 35 % normalized collision energy, dynamic exclusion for 5 min) in a completely automated fashion.
Proteins were identified by searching the MS/MS spectra against a S. coelicolor protein database using SEQUEST (Eng et al. 1994) with Sequest Sorcerer software (Thermo Scientific). An XCorrelation score was calculated based on the cross-correlation peptide score following a SEQUEST database search (Eng et al. 1994). The following criteria were used to sort out proteins from MS spectra. The cut-off for the cross-correlation score (Xcorr) was [1.7 for ?1 charged tryptic peptides, [2.5 for ?2 charged peptides, or [3.0 for ?3 charged peptides. The delta correlation value (DCn) was at least 0.15, regardless of the charge state (Ducret et al. 1998). Finally, only those hits were considered for which at least two peptides were identified with a significant Xcorrelation score.

Protein production and purification
To produce recombinant hexahistidine-tagged proteins, cultures of E. coli BL21 (DE3) harbouring pET24-SCO0608 or pET24-SCO6008 were grown in 1 l of LB broth at 37°C until an OD 600 of 0.6-0.8, and induced by the addition of IPTG to 0.5 mM. Extracts were prepared and proteins purified using a column packed with Nickel-NTA agarose resin (BioRad) essentially as described (Mahr et al. 2000). Columns were washed with 10 column volumes of binding buffer (50 mM sodium phosphate buffer; 300 mM NaCl; 0.01 % Tween20) and proteins eluted with three column volumes of elution buffer (binding buffer containing 100 mM imidazole). The eluted proteins were desalted and concentrated using 15-ml centrifugal filter unit (Millipore, molecular weight cut-off of 10 kDa). The concentrated protein extracts of *300 ll volume were immediately used or flashfrozen with liquid N 2 for storage at -80°C.

Electrophoretic mobility shift assay (EMSA)
To obtain probes for EMSAs, DNA fragments containing the upstream regions of the target genes were amplified by PCR and purified, following electrophoresis, from agarose gels. For EMSAs with SlbR, PCR products were end-labelled with [c-32P] dATP using T4 polynucleotide kinase and purified using Probe-Quant TM G-50 microcolumns (GE Healthcare, USA). The labeled probe and purified recombinant protein of different concentrations were added to a solution containing 0.5 9 TBE (pH 8.0), 0.0375 % (w/v) glycerol, 5 mM MgCl 2 , 1 mM DTT, and 7.5 lg sheared salmon sperm DNA (20 ll final volume). After 30 min incubation at room temperature, 1 ll of 10 % glycerol was added to each mixture and samples were immediately loaded onto a 4 % non-denaturing polyacrylamide gel, which was subjected to electrophoresis in 0.5 9 TBE. Gels were dried under vacuum and exposed to phosphor-screen and images acquired using a Typhoon 8600 scanner (GE Healthcare). EMSAs with AtrA were done as described previously (Uguru et al. 2005), except the probes were fluorescently labelled via the incorporation of a FAM group at the 5 0 end of one of the PCR primers during its synthesis (Eurofins). The 224-bp ssgR probe (-321 to -97 relative to the start of the gene) was amplified using primers ssgRp1 and ssgRp2, the 100-bp actII-ORF4 (site 2) probe (?115 to ?15) was amplified using actII-4(s2)p1 and actII-4(s2)p2, and the 179-bp SCO4119 probe (-305 to -126) was amplified using SCO4119p1 and SCO4119p2.

Transcription of ssg genes in sporulation mutants of S. coelicolor
We have previously shown that ssgRA are transcribed independently of the sporulation genes whiA, whiB, whiG, whiH, whiI and whiJ in S. coelicolor (Traag et al. 2004). Here we expanded this survey so as to include the transcriptional analysis of ssgB-G in the genetic background of these 'classical' whi mutants as well as in ssgB mutants (Keijser et al. 2003). For this, semi-quantitative RT-PCR was done on RNA isolated from mycelia grown for 24, 48 or 72 h on MM agar plates with mannitol as the sole carbon source. After 24 h, S. coelicolor M145 produced a vegetative mycelium, aerial hyphae were formed after 48 h and abundant sporulation was seen after 72 h. 16S rRNA was used as the control. The results are shown in Fig. 1. RT-PCR data were quantified and corrected for loading differences using 16S rRNA as an internal reference (see Materials and Methods section). In the parental strain S. coelicolor M145 transcription of ssgA, ssgB, ssgC, ssgE, ssgF and ssgG was life cycle-dependent, with increased transcript levels at stages in development, while ssgD transcript levels were equally high at all stages (Fig. 1). As reported earlier, ssgA transcription was not significantly (less than two-fold) altered in any of the whi mutant backgrounds (although ssgA transcript levels appeared somewhat reduced in the whiH mutant), and the same was true for ssgD. Transcription of the cell division gene ssgB was primarily seen during late development (72 h), which corresponds to sporulation in wild-type cells. Transcription of ssgB was more than three-fold reduced in the whiA mutant, and nearly absent in the whiH mutant. Transcription of ssgG, a functional homologue of ssgB, was also strongly development-dependent, and reduced in the whiH mutant, as well as in the whiJ mutant. Surprisingly, transcript levels of ssgG appeared constitutive in an ssgB mutant. Differential expression of other ssg genes was seen in several sporulation mutants: transcription of ssgC was enhanced in one or more time points of most of the whi mutants and in the ssgB mutant, while the sporematuration genes ssgE and ssgF were upregulated moderately in whiI, whiJ and ssgB mutants.
Screening for novel transcription factors that control ssgR, ssgA, ssgB and ssgG To identify TFs that may directly control the transcription of the sporulation-specific ssg genes, we carried out DNA affinity-capture assays (DACA; (Park et al. 2009)) using biotinylated DNA baits corresponding to the promoter regions of ssgA, ssgB, ssgG and ssgR. For details on method and significance of the hits, see Materials and Methods section. S. coelicolor M145 was grown in YEME media, samples were collected after 48, 72 and 100 h of growth, and total cell extracts prepared. Probes were washed extensively and bound proteins identified by high sensitivity mass spectrometry. The power of the DACA method is that DNA-bound proteins are identified at low concentrations, but the sensitivity of mass spectrometry also means that subsequent validation by EMSAs with purified candidate proteins is required. DACA assays identified ten different proteins in the various fractions containing proteins that bound to at least one of the ssg promoter regions, all of which are known or predicted DNA binding proteins (Table 1 and Table S3); this indicates significant specificity for the method used, considering that the S. Fig. 1 Transcription of ssg genes in whi mutants of S. coelicolor. Semiquantitative RT-PCR data of ssgA-G and 16S rRNA (vertical axis) performed on RNA purified from the parental strain M145 and its whiA, whiB, whiG, whiH, whiI, whiJ and ssgB mutants (horizontal axis). Grey shading means significant binding, with a minimum of two identified peptides per hit and significant Xcorrelation scores in SEQUEST (Eng et al. 1994). For details see Table S3 and ''Materials and methods'' section coelicolor genome encodes nearly 1000 proteins with predicted regulatory function (Bentley et al. 2002). These 10 proteins were SCO0608 (SlbR), SCO1839, SCO2792 (AdpA), SCO2950 (HupA), SCO3198 (FruR), SCO3375, SCO3606, SCO3859, SCO4118 (AtrA) and SCO5803. Of these, SCO1839, SCO3375 and AtrA were specifically identified for only one of the ssg promoter fragments. Two proteins were found for all probes, namely HupA, one of the three nucleoid-associated HU-homologous proteins produced in vegetative hyphae (Salerno et al. 2009) and a general DNA-binding protein, and FruR, which most likely controls the fructose metabolic operon fruKA encoding the fructose kinase (SCO3197) and fructose permease (SCO3196) (Nothaft et al. 2003).

AtrA directly controls ssgR transcription
The DACA technology identified AtrA as possible regulator for ssgR. AtrA is a TetR-family transcriptional regulator that in S. coelicolor directly activates actII-ORF4 (Uguru et al. 2005) and nagE2 (Nothaft et al. 2010), which encode the pathway-specific activator for actinorhodin biosynthetic gene cluster and the membrane component of the N-acetylglucosamine transporter, respectively. Analysis using the PREDetector algorithm (Hiard et al. 2007) identified a likely AtrA responsive element (are), namely the inverted repeat sequence GGAACCACCGGTTCC (complementary sequences underlined), corresponding to nt positions -166/-152 relative to the ssgR translational start (see Discussion). This upstream element conforms very well to the consensus sequence published previously (Nothaft et al. 2010). To test whether this putative cis-acting element is indeed recognized specifically by AtrA, we performed electrophoretic mobility shift assays (EMSAs) using purified recombinant AtrA (Uguru et al. 2005). Comparison with previous results was facilitated by including in the analysis fragments containing the AtrA-binding site within the coding region of actII-ORF4 (Uguru et al. 2005) and the recently reported AtrA-controlled promoter of SCO4119 (encoding a putative NADH dehydrogenase) (Ahn et al. 2012), which is divergent from the atrA promoter (SCO4118). As reported previously, AtrA bound within the coding region of actII-ORF4 with an apparent equilibrium dissociation constant of K d ' of 150 nM (Uguru et al. 2005). AtrA also bound to the ssgR promoter (Fig. 2a), albeit that the affinity of this interaction was 8-fold weaker (K d ' of 1.2 lM). It was however stronger than the binding of AtrA to the promoter of SCO4119. The effect of AtrA on the expression of ssgR was investigated using quantitative RT-PCR (qPCR) analysis of transcript abundance in M145 and L645, the congenic atrA knockout (Uguru et al. 2005). To permit several transcripts with a range of abundances to be included in the study, we adopted a quantitative approach (for details, see Materials and Methods). After preliminary investigations, rpsL, which encodes ribosomal protein S11, was adopted as the internal control because it is constitutively expressed and the abundance of its mRNA is in the middle of the range for transcripts of interest (Romero et al. 2014). Furthermore, the analysis focused on mycelia grown on MM agar plates with mannitol as the sole carbon source to the point at which aerial hyphae began to emerge. This offered excellent discrimination of the effects of atrA on actII-ORF4 and members of its regulon (Fig. 2b). The average fold decreases in the abundance of the transcripts of actII-ORF4 and actI in L645 were 20.1 (±1.9) and 7.5 (±2.9), respectively. These data are in line with our previous data showing that transcription of the act biosynthetic gene cluster depends on AtrA (Uguru et al. 2005). The level of the ssgD transcript was unchanged (1.0 ± 0.1). In contrast, the level of the transcript of ssgR and the transcript of ssgA, the target of SsgR activation, decreased on average 2.9 (±0.5) and 2.8 (±0.7) fold, respectively. These qRT-PCR results provide in vivo evidence consistent with the direct transcriptional activation of ssgR by AtrA. Moreover, the overall analysis described in this section provides strong validation for the applicability of DACA. It also should be noted that AtrA binding to the actII-ORF4 promoter was detected by the DACA assay (data not shown). The considerably higher abundance in M145 of the ssgD transcript relative to those of ssgA and ssgR (Fig. 2b) was validated by independent promoter probing results (unpublished data). The level of the ssgD transcript, as well as those of actII-ORF4 and actI, was similar to that of rpsL, which encodes one of the most abundant proteins in bacteria.
The c-butyrolactone receptor protein SlbR binds upstream of ssgA, ssgB and ssgR The DACA assays identified two c-butyrolactoneresponsive proteins, AdpA and SlbR (for ScbR-like cbutyrolactone binding regulator; SCO0608), as potential regulators of multiple ssg genes. This connects to earlier data showing that AdpA is required for transcription of ssgA in S. griseus (Yamazaki et al. 2003). However, deletion of adpA has little effect on ssgA transcription in S. coelicolor (Traag et al. 2004). SlbR is a receptor for the c-butyrolactone Scb1 of S. coelicolor and binds to the scbRA and adpA promoter regions, but does not share significant sequence homology to the canonical c-butyrolactone receptor protein ScbR in S. coelicolor (Yang et al. 2012). To investigate if indeed SlbR binds to the promoter regions of the ssg genes, EMSAs were performed, whereby all four ssg promoter regions were tested as probes (Fig. 3). This revealed direct binding of recombinant SlbR to the upstream regions of ssgA, ssgB, ssgG and ssgR, albeit at higher protein concentrations (lM range), suggesting that the affinity for the promoters in vitro is relatively weak. The binding site for SlbR is yet unknown (Yang et al. 2012), and similarity searches using the promoter regions for the ssg genes and the known SblR target scbR failed to identify a common regulatory element (not shown). Preliminary transcript analysis on surface-grown cultures by qRT-PCR did not reveal major differences in expression of ssgA, ssgB, ssgG or ssgR in the slbR null mutant (unpublished data). It should be noted that the DACA assays were done with extracts obtained from liquid-grown cultures, which may explain this apparent discrepancy.

Discussion
Streptomycetes are multicellular bacteria that undergo an unusually complex life cycle among microorganisms, and while gradually more regulatory networks are being uncovered, the number of key developmental genes whose function has been elucidated in detail is still rather limited (Chater et al. 2010;Flärdh and Buttner 2009). Well-studied developmental genes are the whi genes, which encode different types of regulatory proteins (Chater 2001), and the ssg genes, for members of the family of SsgA-like proteins (SALPs) that do not control transcription, but instead likely act as chaperone-like proteins that mediate processes related to peptidoglycan synthesis and remodeling (Noens et al. 2005). To establish possible connections between the whi and ssg networks, we have looked into ssg gene expression in mutants lacking one of the early whi sporulation genes whiA, whiB, whiG, whiH, whiI, and whiJ, and identified regulatory proteins that control transcription of the cell division regulatory genes ssgA, ssgB, ssgG and ssgR.
Transcription of the key sporulation regulatory gene ssgB was strongly down-regulated in whiA mutants and nearly abolished in whiH mutants. These data conform well to recent microarray analyses, which showed that transcription of both ssgB and ssgG depends on whiA and whiH (Salerno et al. 2013). The transcription of ssgG, which likely is a functional homologue of ssgB , was also reduced significantly in whiH mutants, but under the chosen growth conditions we did not see a decrease in ssgG transcription in whiA mutants. Interesting parallels exist between the developmental phenotypes of S. coelicolor colonies lacking either ssgB or whiA. Mutants of whiA and ssgB have white (non-sporulating) phenotypes, producing aseptate aerial hyphae and no spores (Ryding et al. 1999;Flärdh et al. 1999;Keijser et al. 2003). Furthermore, whiA and ssgB mutants both appear to lack the signal for aerial growth arrest that precedes the onset of sporulation (Chater 2001), with whiA mutants producing very long aerial hyphae, while colonies of the ssgB mutant have a large colony phenotype (Flärdh et al. 1999;Keijser et al. 2003). The precise connection between WhiA and SsgB requires further investigation.
Similar to ssgRA (Traag et al. 2004), transcription of ssgD was not significantly affected in the six early whi mutants or in an ssgB mutant. This is perhaps not surprising, as SsgD is the only SALP that plays a role during vegetative growth: the ssgD gene is strongly transcribed during vegetative growth, and ssgD null mutants have pleiotropic defects in the lateral wall of the vegetative hyphae (Noens et al. 2005). A logical explanation for the whi-independent expression of ssgA is that it also plays a role in germination and branching, which occur during vegetative growth (Noens et al. 2007), and similar arguments can be used for ssgD. Finally, transcription of ssgC, whose function is less well understood, and of ssgE and ssgF, which are involved in spore maturation (Noens et al. 2005), was deregulated to some extent in most of the whi mutants as well as in the ssgB mutant.
Besides the control of ssgB and ssgG by WhiA and WhiH, the roles of the Whi and SALP proteins in development appear primarily parallel. Indeed, the Whi regulatory proteins ensure the correct timing of the transcription of ftsZ during sporulation, whereby accumulation of FtsZ is a major prerequisite for the onset of sporulation-specific cell division (Flärdh et al. 2000;Willemse et al. 2012). Conversely, the SALPs primarily play a role in the control of peptidoglycan remodeling, such as during cell division and sporewall synthesis. More detailed information on the Whi regulatory networks is required to provide further insights into their precise regulatory networks.
Novel regulators of ssg gene expression DACA assays combining affinity chromatography with MALDI-ToF mass spectrometry and using the promoter regions for ssgA, ssgB, ssgG or ssgR as baits, identified in total ten candidate regulatory proteins that bound to one or more of the ssg promoters (Table 1). The fact that all proteins were indeed annotated DNA binding proteins, and that over 1000 regulatory proteins are encoded by the S. coelicolor genome, provides significant validation for the assay. Of the ten proteins, HupA is a 'general' DNA-binding protein belonging to the family of nucleoid-associated HUproteins (Salerno et al. 2009).
Two regulators involved in A-factor mediated quorum sensing, AdpA (SCO2792) and SlbR (SCO0608), and an activator for antibiotic production, AtrA (SCO4118), were also identified. Of these, AdpA is known to control ssgA transcription in S. griseus Fig. 3 EMSAs showing binding of SlbR to the ssg target sequences in vitro. Left, binding of purified SlbR-His 6 (3.5 lg) to 32 P-radio-labeled probes of ssgA, ssgB, and ssgG upstream sequences. Right, Binding of SlbR-His 6 to the upstream sequence of ssgR and ssgA. DNA was incubated with (from left to right) no SlbR protein (control), 0.1, 0.4, 3.4, and 6.8 lg protein Antonie van Leeuwenhoek (2015) 108:201-213 209 (Yamazaki et al. 2003), suggesting that AdpA is indeed a bona fide hit in the DACA assay. However, unlike for the model system S. griseus, the AdpA regulon of S. coelicolor is not well known, and direct developmentrelated targets were identified only recently (Wolanski et al. 2011), namely sti, encoding a protease inhibitor (Kim et al. 2008), and ramR, for an atypical response regulator that activates the expression of the small modified peptide SapB required for the onset of aerial growth (Willey and Gaskell 2011;Nguyen et al. 2002;Keijser et al. 2002). SlbR was recently identified as a novel c-butyrolactone receptor-like protein in S. coelicolor, with apparently similar behaviour as ArpA in S. griseus (Yang et al. 2012). Although both regulators affect antibiotic production and spore formation, none of the studies previously related the sporulation-specific SALPs to these autoregulator-controlled regulators. While EMSAs showed that SlbR binds to the promoter region of all ssg genes analysed (ssgA, ssgB, ssgG and ssgR), preliminary transcript analysis did not reveal significant changes in ssg transcript levels in the slbR null mutants grown on SFM agar plates. However, the DACA assays were done using extracts obtained from liquid-grown cultures, and it is difficult to compare the two growth conditions. While in the function of SsgB has only been studied in surface-grown cultures (i.e. in the aerial hyphae), its expression is also strongly enhanced during late exponential growth (Huang et al. 2001). Besides recruiting FtsZ during sporulationspecific cell division, SsgB also plays a role in growth cessation, with ssgB null mutants displaying a large colony phenotype (Keijser et al. 2003). Furthermore, many streptomycetes sporulate in submerged cultures , a process that is mechanistically very similar to aerial sporulation and requires SsgA (Kawamoto et al. 1997) and most likely also SsgB. The effect of SlbR on the control of growth and sporulation of streptomycetes remains to be analyzed in more detail. A conspicuous DACA hit was the TetR-family transcriptional regulator AtrA. AtrA directly activates actII-ORF4, encoding the pathway-specific transcriptional activator for the actinorhodin biosynthetic gene cluster in S. coelicolor (Uguru et al. 2005), and modulates the expression of strR, the pathway-specific transcriptional activator of streptomycin biosynthesis in S. griseus (Hirano et al. 2008;Hong et al. 2007). Mutants deleted for atrA fail to develop, and the reason for this developmental blockage has so far remained unexplained. We here show that AtrA activates the transcription of the sporulation regulatory gene ssgR. AtrA was shown to bind specifically to sequences in the upstream region of ssgR, and analysis using the PREDetector algorithm (Hiard et al. 2007) identified a likely AtrA-responsive element (are), namely the inverted repeat sequence GGAAC-CACCGGTTCC (-289 to -275 relative to the predicted start of ssgR; complementary sequences underlined). It should be noted that the predicted translational start of ssgR is not correct and is most likely the TTG triplet 123 nt upstream (start at chromosomal position 4318125), since proteomics experiments have unequivocally shown that the SsgR protein is at least 19 aa longer than predicted. The SsgR protein is thus extended by the aa sequence: LATADTAFAQLHALPVQPCPTSRTPPASP-SADPPSATLIGSV, with the underlined residues identified by mass spectrometry (our unpublished data). In addition to identifying AtrA binding to the promoter region of ssgR, disruption of atrA in vivo strongly reduced transcription of ssgR and ssgA.
Taken together, our data strongly suggest that AtrA activates the transcription of ssgR, the gene product of which is in turn required for the transcriptional activation of ssgA. It has always been a bit of a mystery as to why ssgRA should not be influenced by any of the Whi sporulation genes and what the alternative mechanism is. The likely explanation for the former is that SsgA is not only needed for sporulation-specific cell division, but also is active at a time when the Whi proteins are not yet active, namely during germination and tip growth and branching of the vegetative hyphae (Noens et al. 2007). We now show that AtrA is a major regulator of ssgRA gene expression. Also, this is the first developmental gene that was shown to be controlled by AtrA and offers an explanation as to why sporulation is impaired in the atrA null mutant. However, since ssgRA are only required for sporulation, this does not explain why AtrA is required for aerial hyphae formation.

Concluding remarks
This study expands our insights into how in particular the cell division-related ssg genes are controlled during development. Multiple TFs control the sporulation genes ssgA and ssgB, which ultimately control the onset of sporulation-specific cell division. DACA combined with transcript analysis showed that ssgR transcription is directly activated by the developmental regulator AtrA, which also activates actinorhodin biosynthesis. This provides a novel and direct link between development and antibiotic production, and also highlights the first developmental target for AtrA itself. In S. griseus, AdpA controls ssgA transcription, and the DACA suggested that AdpA may indeed control ssgA. Binding of SlbR was seen for the ssgA, ssgB and ssgG upstream sequences and was validated by EMSAs, but under the conditions chosen, we did not see a major effect on ssg gene expression in the slbR null mutant. Finally, several other TFs were identified as possible regulators of the ssg genes, and the precise role of these TFs in the control of ssg gene expression and in the control of cell division and development of streptomycetes should be analysed further.