Control of Morphological Differentiation of Streptomyces coelicolor A3(2) by Phosphorylation of MreC and PBP2

During morphological differentiation of Streptomyces coelicolor A3(2), the sporogenic aerial hyphae are transformed into a chain of more than fifty spores in a highly coordinated manner. Synthesis of the thickened spore envelope is directed by the Streptomyces spore wall synthesizing complex SSSC which resembles the elongasome of rod-shaped bacteria. The SSSC includes the eukaryotic type serine/threonine protein kinase (eSTPK) PkaI, encoded within a cluster of five independently transcribed eSTPK genes (SCO4775-4779). To understand the role of PkaI in spore wall synthesis, we screened a S. coelicolor genomic library for PkaI interaction partners by bacterial two-hybrid analyses and identified several proteins with a documented role in sporulation. We inactivated pkaI and deleted the complete SCO4775-4779 cluster. Deletion of pkaI alone delayed sporulation and produced some aberrant spores. The five-fold mutant NLΔ4775-4779 had a more severe defect and produced 18% aberrant spores affected in the integrity of the spore envelope. Moreover, overbalancing phosphorylation activity by expressing a second copy of any of these kinases caused a similar defect. Following co-expression of pkaI with either mreC or pbp2 in E. coli, phosphorylation of MreC and PBP2 was demonstrated and multiple phosphosites were identified by LC-MS/MS. Our data suggest that elaborate protein phosphorylation controls activity of the SSSC to ensure proper sporulation by suppressing premature cross-wall synthesis.


Introduction
The peptidoglycan (PG) sacculus, an elastic mesh of linear carbohydrate chains cross-linked by short peptides surrounds the bacterial cell [1,2]. This exoskeleton is the cell shape-maintaining element and allows the cell to withstand alterations in internal turgor due to changing environments. Only few bacterial species developed a life style devoid of a PG sacculus [3]. Whereas Gram-negative bacteria have a PG-monolayer, Gram-positive species have a thick multi-layered cell wall that in addition contains large amounts of covalently bound anionic glycopolymers, called wall teichoic acids (WTA) [4,5]. Despite high conservation of the basic PG structure, various species specific modification including different amino acids in the stem peptide or variations in the interpeptide bridge or O-acetylation occur [6].
A major factor in cell shape determination is the site of PG-synthesis [7]. Cocci incorporate PG mainly at the division plane [8], generating a spherical morphology. Incorporation of PG at the lateral wall in a process called elongation growth [9,10], produces a rod-shaped morphology. A key determinant of elongation growth is the actin-like MreB protein [11], which has been originally suggested to form a dynamic helical filament underneath the surface to direct PG synthesis [12][13][14]. More recent studies do not support helical filaments and describe MreB patches that passively move along with the putative cell wall extension machinery [15,16]. Whereas Gram-negative bacteria have a single MreB protein, the Gram-positive Bacillus subtilis requires three MreB homologues, MreB, Mbl and MreBH, for growth [17]. Depletion of MreB, or its inhibition by the antibiotic A22, results in the loss of rod-shape and eventually cell death [17,18]. MreB forms a lateral wall synthesizing complex with MreC, MreD, the PBP2, RodA, and RodZ [19]. In B. subtilis also interactions with WTA-synthesizing proteins have been shown [20], suggesting that the lateral wall synthesizing complex might also be involved in the synthesis of WTAs.
Actinobacteria are a group of Gram-positive bacteria with different shapes, including spheres, rods and branched mycelium. Their mode of growth does not depend on Mre-proteins, as demonstrated by the absence of mre genes in most genera [21]. Instead they incorporate new PG at the cell poles, dependent on the DivIVA protein [22,23]. In mycelial Streptomyces, the hyphal growth at the tip is mediated by the polarisome which consists of the coiled-coil proteins DivIVA, Scy and FilP [24,25]. Although tip growth does not depend on Mre proteins, Streptomyces and related actinomycetes contain an mre gene cluster consisting of mreBCD, pbp2, and the rodA homolog sfr [21,26]. However these genes are not essential but affect synthesis of the thickened spore wall which makes Streptomyces spores resistant to detrimental environmental conditions [27,28].
Since the protein-protein interaction pattern of the S. coelicolor Mre proteins highly resembled the interaction pattern of Mre proteins forming the lateral wall synthesizing complex of rod-shaped bacteria, the term Streptomyces spore wall synthesizing complex SSSC was coined [28,29]. Screening of a genomic library for additional components of the SSSC identified SCO4778 (PkaI), encoding a eukaryotic type serine/threonine kinase (eSTPK) as an interaction partner of MreC, MreD, PBP2, and Sfr [28].
For a long time it was thought that signal transduction in bacteria relies only on histidine/ aspartate phosphorylation via two-component systems, while protein phosphorylation by eSTPKs is limited to eukaryotes. But with the increasing availability of genomic data, more and more eSTPKs were identified in bacteria. For example, S. coelicolor contains more than 30 eSTPKs [30,31] and more than 50 eukaryotic-type protein phosphatases [32]. The catalytic domains of many bacterial eSTPKs are linked to additional domains. These domains can be extracellular and often mediate ligand binding or protein-protein interaction. The M. tuberculosis kinase PknB was shown to become activated by dimerization upon binding of its extracellular PASTA-domain to muropeptides [33,34]. Regulation of proteins involved in cell division and cell wall synthesis by phosphorylation is well documented in different bacteria [35]. Phosphorylation can result in the activation of an enzyme or its inhibition. Whereas activity of the muropeptide ligase MurC is inhibited in Corynebacterium glutamicum, when phosphorylated by PknA [36], GlmM of E. coli, involved in peptidoglycan precursor synthesis is activated by phosphorylation [37].
Phosphorylation of one of the Mre proteins has not been reported yet in any bacterium. Here we show that PkaI is able to multiply phosphorylate coexpressed MreC and PBP2 proteins, providing evidence that sporulation septation and the synthesis of the thickened Streptomyces spores wall by the SSSC is controlled by protein phosphorylation.

Results
Expression of pkaI is down-regulated during morphological differentiation Interaction of PkaI with several SSSC proteins suggested a role of PkaI in sporulation [28]. PkaI is encoded in a cluster of five putative eSTPK genes, SCO4775 (pkaH), SCO4776, SCO4777 (pkaD), SCO4778 (pkaI), SCO4779 (pkaJ) in S. coelicolor M145 [31]; S1A Fig). The genomic organization of the eSTPK genes, with intergenic sequences between 93 bp and 173 bp, indicates that each gene is transcribed separately. This was confirmed by RT-PCR. Primer pairs designed to amplify the intergenic regions of each of these genes did not produce the respective fragments, when cDNA of S. coelicolor was used as a template (S1B Fig).
To elucidate a possible role in morphological differentiation, we characterized the expression profiles of pkaI and the other eSTPK genes at distinct stages of the S. coelicolor life cyle by semi-quantitative RT-PCR with gene-specific primer pairs (S1C Fig). Whereas expression of pkaH could not be detected in any growth phase, SCO4776 was constitutively expressed throughout the whole life cycle. In contrast, pkaD, pkaI, and pkaJ showed highest transcription after 24 hours when mainly substrate mycelium was formed. During aerial mycelium formation, transcription of all three genes was down-regulated. This indicates that PkaD, PkaI, and PkaJ accomplish their function in hyphal growth before or at the onset of morphological differentiation.

The ΔpkaI mutant NLΔPkaI is affected in proper sporulation
To assess the putative role of the eSTPK PkaI in the SSSC, we generated an in frame deletion mutant of pkaI that had no polar effects on the downstream pkaJ (S2 Fig). The NLΔPkaI mutant showed normal growth in liquid culture. Only sporulation was slightly retarded on solid agar (S3A Fig). Whereas the parental M145 started to differentiate on SM agar after two days, white aerial mycelium of NLΔPkaI was not visible before the third day of incubation. After seven days, the sporulation of NLΔPkaI was nearly indistinguishable from that of M145. This delay in sporulation corresponded also to the generation of viable spore particles. After three days of incubation, 50 fold less spores could be isolated from a NLΔPkaI culture, compared to M145. After five days of incubation, the spore titers obtained from NLΔPkaI and M145 were similar (S3B Fig). The effect of pkaI inactivation on spore morphology was studied by seeding cover slides in SM-agar with NLΔPkaI spores and inspecting the slides by phase contrast microscopy after 3-6 days of incubation. Whereas M145 mostly produced very regular spore chains, NLΔPkaI spore chains frequently contained some aberrant spores (Fig 1, arrows) which were enlarged and more spherical. Genetic complementation of NLΔPkaI by integrating an intact copy of pkaI together with its native promoter region into the PhiC31 attachment site (NLΔPkaI:: pSET-pkaI restored the morphological defect. Both NLΔPkaI phenotypes, the delay in sporulation, as well as the formation of aberrant spores were complemented (Fig 1, S3A and S3B Fig). Whereas integration of pSET-pkaI in the NLΔPkaI chromosome complemented aberrant spore formation, integration of pSET-pkaI into the chromosome of the parental strain M145 caused a similar sporulation defect as the deletion of pkaI in NLΔPkaI (Fig 1). Obviously, overexpression of pkaI interfered with the control of proper sporulation.
Overbalancing kinase activity of S. coelicolor affects integrity of the spore envelope Since pkaI is located in a genomic region consisting of five consecutive eSTPK genes and since clustering often indicates a functional relationship, we also deleted the complete cluster (S2 Fig). The five-fold eSTPK mutant NLΔ4775-4779 was viable and growth of its substrate mycelium was not affected. Microscopic analyses of NLΔ4775-4779 spore chains revealed beside normally ovoid spores also aberrant rod-shaped spores (Fig 1, arrows). The sporulation defect was even more obvious when testing viability of the spores by SYTO9/propidium iodide staining (Fig 2A, S3 Table). Whereas the wildtype M145 contained almost only (99.28%) viable spores (green), the spore chains of the deletion mutants NLΔPkaI and NLΔ4775-4779, respectively, also included dead (red) spores (4.29% and 16.81%) or sometimes spores even devoid of DNA (black). Moreover, germinating NLΔ4775-4779 spores showed an increased sensitivity towards vancomycin ( Fig 3B and S4 Fig), suggesting an impaired spore wall. When vancomycin disks were applied at a later stage, when substrate mycelium has already been formed, NLΔ4775-4779 was as resistant as the NLΔPkaI mutant and M145 (S3C Fig), demonstrating that only the spore wall of NLΔ4775-4779 was sensitive but not the cell wall of the substrate mycelium.
To identify the kinase responsible for the vancomycin sensitivity of the NLΔ4775-4779 spores, we inserted each kinase gene under control of its native promoter region into the FC31 attachment site of NLΔ4775-4779, NLΔPkaI and M145. Vancomycin resistance of spores of the five-fold mutant NLΔ4775-4779 as well as the aberrant morphology or the presence of non-viable spores could not be restored by any kinase gene (Figs 1 and 2, S3 Table), suggesting that activities of multiple kinases are required for proper sporulation. Consistent with this observation, introduction of a second copy of any of the kinase genes, with the notable exception of pkaI, into M145 or NLΔPkaI caused aberrant spores (Figs 1, 2 and 3) with increased sensitivity to vancomycin (Fig 3B and S4 Fig). This phenotype was most severe, when pkaJ was expressed in M145. Here, 59.5% of the spores were non-viable ( Fig 3A, S3 Table). Presence of dead spores and vancomycin sensitivity of germinating spores clearly show a defect in the integrity of the spore envelope. Interestingly, vancomycin resistance was restored in all strains, when the spores were plated onto LB supplemented with 3 mM MgCl 2 (Fig 3B), known to rescue B. subtilis mutants impaired in cell wall synthesis [38].
These data demonstrate that the elaborated kinase activity of S. coelicolor is crucial for proper sporulation and any imbalance in the kinase activity by either deletion or over-expression of specific kinases affects sporulation.

The eSTPK PkaI phosphorylates MreC
Interaction of the Ser/Thr kinase PkaI with MreC [28] suggested that PkaI phosphorylates MreC, thereby controlling its activity. To demonstrate phosphorylation of MreC by PkaI, we simultaneously expressed both proteins in E. coli BL21. The respective genes were cloned under control of the P T7 promoter in pCDF-Duet1, generating translational fusions either to an Nterminal His-tag encoding sequence (pkaI), or a C-terminal S-tag (mreC). As controls, MreC_S-tag and His_PkaI were also expressed separately. Following induction of gene expression, MreC_S-tag was purified by S-tag-affinity chromatography and His_PkaI by Ni-NTA chromatography ( Fig 4A). After staining the gels with the phosphoserine/phosphothreonine specific dye ProQ Diamond strong bands were visible that corresponded to His_PkaI and

Phosphorylation of MreC and PBP2
MreC_S-tag, respectively, as demonstrated by immunoblotting. Interestingly, the S-tag affinity chromatography always led to two bands ( Fig 4A, arrows), both of which were phosphorylated. Mass spectrometry analysis revealed that MreC_S-tag and His_PkaI were present in both bands; the upper band likely represents His_PkaI, which obviously has been co-purified with MreC_S-tag, forming the lower band.
To exclude that phosphorylation of MreC_S-tag occurred accidentally due to the overexpression of His_PkaI, we introduced an additional expression vector (pYT9-Crp) into BL21 carrying pCDF-PkaI-MreC. pYT9-CRP encodes the S. tsukubaensis Crp protein with a Cterminal Strep-tagII sequence (Crp_strep) under control of the rhamnose inducible P rham promoter (S. Kocadinc, pers. communication). Following induction of gene expression by IPTG and rhamnose, Crp_strep was purified by Streptactin affinity chromatography. ProQ Diamond staining revealed that, in contrast to MreC_S-tag, purified Crp_strep was not phosphorylated (S5 Fig). This demonstrates that phosphorylation of MreC_S-tag by His_PkaI is specific and not caused by the non-physiological expression conditions in the heterologous host. and spores without DNA (black) is given for each strain. Spores of the different strains were plated onto LB agar and filter discs containing 5 μg vancomycin were applied. Whereas, M145 and the pkaI mutant NLΔPkaI were resistant, NLΔ4775-4779 spores showed vancomycin sensitivity, suggesting an impaired spore wall. Vancomycin sensitivity of M145 or NLΔPkaI was also caused by expressing a second copy of each kinase gene, with the exception of pkaI. Supplementation of the agar plates with 3 mM MgCl 2, known to rescue mutants impaired in cell wall synthesis restored vancomycin resistance to all strains. A, no plasmid integrated; B, :: pSET152-pkaH; C, :: pSET152-SCO4776; D, :: pSET152-pkaD; E, :: pSET152-pkaI; F, :: pSET152-pkaJ.
The phosphosites of MreC were identified using purified MreC_S-tag from E. coli carrying pCDF-MreC-PkaI. The purified protein was run on a 12.5% SDS polyacrylamide gel, to separate it from contaminating proteins. Following staining of the gel with Coomassie blue, the band corresponding to MreC_S-tag was cut out from the gel and subjected to proteolytic digestion with either trypsin or endoproteinase GluC. Resulting peptide masses were analysed by LC-MS/MS (S4 Table). To discriminate specific phosphorylation sites from artificially phosphorylated residues, the intensities of the phosphorylated peptides were compared to the intensities of the corresponding non-modified peptides; only singly-phosphorylated peptides with at least 10-fold higher intensity than unmodified counterparts were considered. A single MreCderived peptide was found to be more abundant in the phosphorylated form. The tryptic peptide LVTFGSQADKPFVPGVPVGp[TIT]RVDPNGGDLTR (bold residues indicate probable phosphorylation sites) comprising MreC 219-251 was detected 100-fold more abundant compared to the non-phosphorylated one ( Table 1). The MreC phosphorylation site (Fig 4B) is localized within the Pfam_MreC domain (MreC 121-273 ). Analysis of the fragmentation spectrum did not lead to unambiguous localisation of the phosphorylation to one of the two threonine (T 238 , T 240 ) residues. When MreC_S-tag was purified in the absence of PkaI, it was non-phosphorylated ( Fig 4A). Also LC-MS analysis detected only non-phosphorylated peptides (data not shown). This demonstrated that MreC was specifically phosphorylated by PkaI and not by an endogenous E. coli kinase.

Detection of phosphorylation sites in PkaI
Purified MreC_S-tag also contained considerable amounts of His_PkaI, since S-tag purification obviously did not disrupt the His_PkaI-MreC_S-tag interaction (Fig 4A). Due to their similar size, MreC_S-tag and His_PkaI are not well separated by SDS-PAGE. Therefore, the MreC_Stag band, which was eluted from the gel to determine the MreC-phosphosites also contained His_PkaI and among the MreC_S-tag peptides also His_PkaI peptides were identified. This further supports the strong interaction of PkaI and MreC, reported by Kleinschnitz et al. [28] and allowed an assignment of the phosphorylation sites not only to MreC but also to PkaI (Table 1 and S4 Table). Two phosphorylated peptides (PkaI 111-124 , VLpTRGPVDAVEAAR, PkaI 164-177 , FGVAQVAGAp[TT]LTE) were identified for PkaI. Both phosphorylation sites are located in the highly conserved S_TKc kinase domain. PkaI 164-177 corresponds to the so called activation loop of eSTPKs, involved in determining substrate specificity [39,40] and contains the conserved T residue, which becomes phosphorylated to activate the kinase.

PkaI phosphorylates also PBP2
In order to study whether PkaI also phosphorylates other PkaI interaction partners, in addition to MreC, we expressed the monofunctional penicillin binding protein PBP2 together with PkaI. Interestingly, expression of PBP2_S-tag could not be achieved, when pbp2 alone was inserted in pCDF-Duet1 (S6 Fig). However, in the presence of his-pkaI, expression of PBP2_Stag was obtained in reasonable amounts. Obviously, PkaI had a stabilizing effect on PBP2 expression. Since purification of PBP2_S-tag was not satisfactorily under native conditions, PBP2_S-tag was denatured by the addition of 6 M urea and purified under denaturing conditions (Fig 5A). ProQ diamond staining and immunoblotting with anti-S-tag specific antibodies demonstrated that also purified PBP2-S-tag was phosphorylated. The PBP2-S-tag band was cut from a Coomassie stained polyacrylamide gel, digested with trypsin, and analysed by LC-MS/ MS (S4 Table) to identify the phosphorylated peptides (Table 1). PBP2 72-81 (GVALAD-NEpTR), PBP2 82-93 (LVVSApSRTDLLK), PBP2 210-220 (SDQVGRpSGLER), and PBP2 550-560 (AVVSPDGKpTVR) were the most prevalent phosphorylated PBP2 peptides (bold letters indicate the phosphosites). Whereas peptide PBP2 550-560 is located in the transpeptidase domain of PBP2 (Fig 5B), all other PBP2 phosphorylation sites are found in the N-terminal dimerization domain (PBP2 60-250 ), suggesting that PBP2 phosphorylation affects its dimerization. LC-MS/ MS analyses of the phosphorylated peptides identified T80, S87, S216 and T558 as the residues phosphorylated by PkaI (Table 1).

PkaI interacts with further proteins involved in differentiation
Previous screening of a S. coelicolor genomic library for interaction partners of MreC and PBP2 identified PkaI 257-357 as the interacting domain [28]. To find additional target proteins of PkaI, we screened genomic libraries of S. coelicolor by bacterial two-hybrid analyses for interaction partners of PkaI. Full length pkaI, as well as the fragment encoding the putative interaction domain PkaI 257-357 were translationally fused to the T18 domain of the Bordetella pertussis adenylate cyclase gene cya. The resulting plasmids were introduced into the E. coli cya mutant BTH101. Subsequently, these strains were transformed with genomic libraries of M145 made in the Cya-T25 domain encoding vector pKT25. In total, about 1.7 x 10 6 colonies were obtained with the enzymatically generated library [28] and about 3 x 10 5 colonies with a library containing S. coelicolor DNA fragments generated by nebulization. Transformants encoding interacting fusion proteins were selected on minimal agar supplemented with lactose. Whereas, not a single colony developed on minimal agar when full length PkaI was fused, numerous colonies grew on minimal agar, when only the PkaI 257-380 interaction domain was fused to the Cya-T18 domain. Sequence analyses of the pKT25 derivatives confirmed the in frame fusion of most inserts and identified the respective S. coelicolor proteins which interacted with PkaI (S5 Table).
About 30% of the interacting clones contained a fragment (SCO2097 21-120 ) of SCO2097. The 135 aa actinomycete signature protein SCO2097 [41] was previously identified as an interaction partner of MreC, MreD, PBP2, Sfr, FtsI and RodZ and was shown to have a role in proper sporulation [28]. Interestingly, several of the PkaI interaction partners, like the putative membrane protein SCO1403 which was isolated five times, or the FtsX-like proteins SCO3110 and SCO3754 have already been identified in the previous screens of SSSC proteins [28]. PkaI also showed self-interaction and interacted with the other eSTPKs PkaA (SCO2974) and PkaD (SCO4777). Moreover, PkaI interacted with several proteins that have a documented role in morphological differentiation, like CrgA (SCO3854), AfsQ1 (SCO4907), BldB (SCO5723) and the FtsH homologue SCO5587. This indicates that PkaI not only phosphorylates specific SSSC proteins but also has a more global role in the regulation of morphological differentiation.

Discussion
A crucial step in morphological differentiation of mycelial Streptomyces is the simultaneous formation of more than 50 septal cross walls in the unbranched aerial hyphae. Sporulation septation and the synthesis of the thickened spore wall involves, besides SsgA-like proteins and FtsZ [42][43][44], the SSSC, a multi-protein complex for spore envelope synthesis that highly resembles the elongasome of rod-shaped bacteria [28]. But whereas, rod-shaped bacteria build only a single septum during cell division, Streptomyces faces the problem, how to build dozens of septal cross walls at the same time [42]. For this process the membrane-and PG-synthesizing machineries have to be provided in sufficient quantities and positioned properly. Moreover, the activities of the complexes have to be controlled to prevent aberrant sporulation by sporadic formation of single cross walls in a non-coordinated manner.
In this study we are describing a route, how the control of sporulation septation and the synthesis of the spore wall could occur by protein phosphorylation.
The crucial role of the eSTPKs PkaH, SCO4776, PkaD, PkaI and PkaJ in sporulation was apparent in gene inactivation and overexpression studies. Deletion of pkaI or all five eSTPK genes, as well as expression of a second copy of any of these kinases produced aberrant spore chains, containing irregular-shaped spores. SYTO9/ propidium iodide staining revealed the presence of spores, which were considerably affected in the integrity of the spore envelope. Moreover the germinating spores of the five-fold deletion mutant and the over-expressing strains were highly sensitive to vancomycin, whereas the spores of the wildtype and the pkaI mutant NLΔPkaI were resistant, due to the presence of the S. coelicolor vanHAX resistance determinants. Although the rationale for the increased sensitivity of germinating spores of the SSSC mutants to different kinds of stress, including cell wall damaging agents, like lysozyme or vancomycin [27,28,29] is not understood, it suggests an impaired spore wall. Since the transpeptidase PBP2 is a key protein of the SSSC, the SSSC mutants are probably affected in the crosslinking of the spore wall. Reduced crosslinking of the PG-layer would correlate with an increase in the vancomycin binding sites (D-ala-D-ala). The increased amount of bound vancomycin then probably interferes with spore germination.
Phosphorylation of proteins as a regulatory mechanism to control their activity is well documented in many bacteria and various proteins of the divisome were shown to be phosphorylated [45]. For example, the mycobacterial kinase PknA was shown to affect septum formation by inhibiting the GTP-dependent polymerization of FtsZ [46]. PknA also phosphorylates the PGsynthesis ligase MurD [47] and the mycobacterial DivIVA homolog Wag31, crucial for polar growth [48]. Activity of DivIVA of S. coelicolor, which directs growth by apical tip extension and hyphal branching, is controlled by the protein kinase AfsK [22].
Here we show for the first time that MreC and PBP2, two important proteins in spore wall synthesis of S. coelicolor are specifically phosphorylated. In contrast, MreC and PBP2 of rodshaped bacteria have not been reported to be phosphorylated. Neither recently published phosphoproteome studies [49,50], nor inspection of phospho-proteome databases (PhosSite and Phosida) of B. subtilis and E. coli indicated phosphorylation of MreC or any other Mre protein of E. coli or B. subtilis.
In vivo phosphorylation of MreC and PBP2 by PkaI was demonstrated in E. coli. Authenticity of MreC_S-tag phosphorylation is confirmed by the control experiments, which showed non-phosphorylation of Strep-tag_Crp and lack of MreC phosphorylation in the absence of pkaI. Direct detection of phosphorylated MreC or PBP2 in S. coelicolor was not possible, since over-expression of each of these proteins is toxic (unpublished results), preventing the production of sufficient amounts for the detection of phosphorylated peptides. In published datasets of S. coelicolor [51,52] MreC or PBP2 are not included, indicative of their low expression level. Expression of PBP2 in E. coli in this study was also problematic and could only be achieved in the presence of PkaI (S6 Fig), but not in the absence of pkaI. The stabilizing effect of pkaI on PBP2 could be due to toxicity of pbp2 expression, probably caused by interference of PBP2 with the E. coli PG-synthesizing machinery. Since three of the four PBP2 phosphosites are located within the dimerization domain, it is tempting to speculate that phosphorylation of PBP2 prevents its dimerization, thus inactivating PBP2 and allowing its expression in E. coli.
On the first view, phosphorylation of extracellular MreC or PBP2 domains by the cytosolic kinase domain of PkaI seems to be implausible. But if the purpose of phosphorylation is to keep the synthesized proteins inactive, the topology of the phosphosites is of minor relevance. Moreover, the phosphosites of the penicillin binding protein PBPA of Mycobacterium tuberculosis, phosphorylated by the eSTPK PknB are also in the (extracellular) transpeptidase domain [31]. Here it was shown that phosphorylation affected positioning of PBPA at the septum, thereby regulating septal peptidoglycan biosynthesis.
Expression of PkaI (and PkaD, PkaJ) was found to be highest in the early growth phase and down regulated in later stages of differentiation. This expression profile is in agreement with microarray data of S. coelicolor [53], which confirm our semi-quantitative RT-PCR data. The only discrepancy concerns pkaH, which was highly expressed in the Yagüe-study [53], while its expression was hardly detectable under our growth conditions. However, the pkaI expression profile differs from the expression of mreC and pbp2, which are induced during morphological differentiation. The different expression profiles of the kinase PkaI and their substrates MreC and PBP2 do not seem to be consistent with the observed phosphorylation of MreC and PBP2 by PkaI. However, this apparent contradiction might be the key for understanding how sporulation septation is controlled by phosphorylation of SSSC proteins. As indicated by the inability to produce PBP2 in the absence of pkaI, phosphorylation of PBP2 might interfere with its activities, probably by affecting subcellular positioning, dimerization, or interaction with other proteins to assemble a functional SSSC. As consequence, the PkaI-mediated phosphorylation would prevent the premature formation of sporadic SSSC complexes unless outgrowth of aerial hyphae is finished and enough SSSC proteins have been synthesized to allow coordinated assembly of more than 50 SSSC complexes. Also, any imbalance in kinase activity, caused by over-expression of single kinase genes interferes with proper sporulation by affecting the coordinated assembly of the SSSC. Control of morphological differentiation most probably also includes one of the 55 S. coelicolor phosphatases [32], able to activate phosphorylated SSSC proteins by dephosphorylation.
The role of PkaI in inhibition of sporadic SSSC complexes is strengthened by the observed interaction of PkaI with CrgA, a septation inhibitor protein that coordinates growth and cell division in aerial hyphae [54]. Interaction of PkaI with other sporulation proteins, like SCO2097, AfsQ, FtsH, FtsX-like proteins, and BldB suggests a more global role of PkaI in differentiation, besides controlling the SSSC. Overexpression of bldB blocked sporulation in aerial hyphae and evidence for the interaction of BldB with an unknown cellular constituent involved in differentiation was reported [55].
The proposed prominent role of PkaI in differentiation of S. coelicolor is in contrast to the quite mild phenotype of pkaI inactivation. However, single eSTPK mutants often only have a mild phenotype [56,57], probably due to the well documented cross reactivity of eSTPK proteins [45]. In S. coelicolor, other eSTPKs probably can substitute the missing PkaI activity. This is supported by the more severe phenotype of the five-fold mutant NLΔ4775-4779 and by our protein-protein interaction data which revealed interaction of PkaI with the other eSTPKs PkaA and PkaD. Moreover, deletion mutants of various SSSC proteins do not completely block sporulation but cause only spores with impaired envelopes [21,27,28], suggesting that the defects can be suppressed by redundant enzymatic activities. In agreement with this, at least four PBPs have been identified in the SSSC [28,29].
Clearly, further experimental work is required to support the proposed differentiation model. The phosphorylation of SSSC proteins should be confirmed in S. coelicolor to reflect their native environment and the biological effect of SSSC phosphorylation has to be demonstrated, e.g. by introducing phosphomimetic mutations. Unfortunately, no enzymatic assays are available for any SSSC protein to directly quantify the effects of protein phosphorylation on the activity of the respective protein.

Bacterial strains and media
Cultivation of strains and procedures for DNA manipulation were performed as previously described [58,59]. Proteins were purified from BL21 (DE3) (Invitrogen). Plasmids and oligonucleotides are listed in S1 and S2 Tables.

RNA isolation and cDNA synthesis
M145 was grown on cellophane disks placed on SM agar plates. Plates were incubated for either 24,48,72 or 96 hours at 30°C. The mycelium was harvested and lysed in Kirby-Mix [59] with a Precellys homogenizer (Peqlab; 5 times program: 6500rpm 2 x 20sec). RNA isolation was carried out as described by Kieser et al. [59] with minor modifications. RevertAid RT Kit (Thermo) was used to synthesize cDNA from isolated RNA according to manufacturer´s protocol.

Construction of mutants
To delete pkaI (SCO4778), a 1.6 kb upstream fragment (primer pair Up4778fwE/ Up4778revB) including the start codon of sco4778 and a 1.6 kb downstream fragment (primer pair lo4778fwB/ lo4778revH) including the sco4778 stop codon was amplified by PCR and cloned into pKT18, yielding pKO4778. After transformation of M145 and integration of the deletion vector pKO4778 by a single crossover (kanamycin-resistant), a ΔpkaI mutant was isolated by selecting for the second crossover (kanamycin-sensitive). Correct gene replacement was confirmed by PCR-analyses and Southern-blotting. The hybridization probe was amplified with primers Up4778fwE/ Up4778revB.
To delete all five serine/threonine kinase genes, a 1.6 kb upstream fragment (primer pair Up4775Ef/Up2775Br) including the start codon of SCO4775 and a 1.6 kb downstream fragment (primer pair Lo4779Bf/Lo4779Hr) including the SCO4779 stop codon were amplified and cloned into pK18. Subsequently the knockout cassette was cut out with EcoRI/HindIII and inserted in pGus21, yielding pGusKO4775-4779.
pGusKO4775-4779 was introduced into S. coelicolor M145 by intergeneric conjugation [60] and apramycin resistant transconjugants were selected that carried pGusKO4775-4779 integrated via a single crossover (M145::pGusKO4775-4779). To screen for the second cross over, resulting in SCO4775-SCO4779 deletion, M145::pGusKO4775-4779 was plated onto soja mannit (SM) agar without antibiotic and incubated for five days. Spores were harvested and appropriate dilutions were plated onto LB agar to obtain single colonies. After two days incubation at 30°C, plates were overlayed with 1 ml H 2 O containing 2.5 mg X-gluc. Colonies that still carried pGusKO4775-4779 were surrounded by a blue halo due to the 4-Cl-3-Br-indigo production by the β-glucuronidase (GusA). Colonies that had lost pGusKO4775-4779 by the second cross over were identified by the lack of the blue halo. Deletion of SCO4775-SCO4779 was confirmed by PCR analyses using primers Intern4778f/Intern4778r, Up4778K416/Lp4778K2240, and c4775-79fwEX/c4775-79revEX.

Spore production assay
Approximately 3x10 7 spores of the respective strain were plated on soya mannit (SM) agar. The plates were incubated at 30°C for 3 and 7 days, respectively, before harvesting the newly formed spores. Dilutions of the spore samples were plated in duplicate to determine the spore titer. 6 biological replicates were made from M145 and NLΔPkaI; 8 replicates from NLΔ4775-4779 and NLΔPkaI::pSET-pkaI.
Heterologous co-expression pkaI (SCO4778), mreC (SCO2610) and pbp2 (SCO2608) were amplified using the primers listed in S2 Table, digested with the respective enzymes, and cloned into the vector pCDFDuet1 (Merck/Novagen). The resulting plasmids (S1 Table) were transferred into E. coli BL21 (DE3) for protein production. Expression of the proteins was performed as described by [61]. Briefly, cells were grown in LB medium containing 100μg spectinomycin/ml at 37°C until an OD600 of 0.6-0.8. Then expression was induced with IPTG at a final concentration of 0.5mM for 3h at 37°C. Purification of S-tagged MreC protein was carried out according to the manufacturers protocol (Novagen) with minor modifications. His-tagged PkaI was purified via Ni-NTA sepharose gravity flow columns using the manufacturers protocol (IBA). As lysis, washing and elution buffer we used 20mM Tris/HCl buffer, pH 7.5 with 150mM NaCl and 0.1% Triton X100. PBP2_S-tag was purified under denaturating conditions by adding urea to the buffer to a final concentration of 6M.
Expression of the Strep-tagII-Crp protein of Streptomyces tsukubaensis (STSU_15619 gene), used as a negative control, was induced with rhamnose at a final concentration of 0.02%. Strep-tagII-CRP was purified using an Äkta purifier (GE Healthcare) with a StrepTrap HP columns following the manufacturers protocol (GE Healthcare). Identity of the purified proteins was confirmed by immunoblotting. Following protein transfer (45 minutes at 400mA) to nitrocellulose membranes, the membranes were incubated, washed and developed according to the protocol for the respective antibody (Anti-S-tag-HRP, Novagen; Anti-Strep-tag-HRP, Biorad; Anti-His6 HRP, Bethyl Laboratories, Inc)

Phospho-serine and -threonine detection
Purified proteins were run on 12.5% SDS-polyacrylamide gels. To detect phosphorylation on serine or threonine residues the gels were stained with ProQ Diamond (Molecular Probes) according to the manufacturers recommendations.

Microscopy
About 10 6 spores were plated onto MS agar and sterile coverslips were inserted in a certain angle of 45°. After five to seven days of incubation at 30°C the coverslips were removed and mounted on slides coated with 1% agarose in PBS.
To detect dead spores, SYTO9 and propidium iodide stains of the LIVE/DEAD BacLight Bacterial Viability Kit (Molecular Probes) were used. The staining solution was prepared by mixing 1.5μl of component A and B in 1ml of water. Spores were incubated on the agar plate with 20μl of staining solution for 15 minutes, then coverslips were removed and mounted on slides coated with 1% agarose in PBS. Images were taken with an Olympus System Microscope BX60 equipped with a F-view II camera (Olympus), using TxRed and eGFP filtersets for detection of the fluorescent markers. Fiji version v.149b was used for image processing and the Cell Counter Plugin for spore counting. Live-dead percentage was calculated from the analyses of7 00-1800 spores from each strain (S3 Table).

Nano LC-MS/MS analysis
Samples were loaded on a NuPAGE Bis-Tris 4%-12% gradient gel (Invitrogen) and the Coomassie stained bands were pooled and in gel digested with trypsin and GluC, respectively, as described elsewhere [62]. LC-MS analyses of the peptides were done on an EasyLC nano-HPLC (Proxeon Biosystems) coupled to an LTQ Orbitrap Elite mass spectrometer (Thermo Scientific) as described elsewhere [63]. MS data were processed using the software suite Max-Quant, version 1.2.2.9 [64] and searched using Andromeda search engine [65] against a targetdecoy E. coli database containing 4,311 forward protein sequences, the sequences of the tagged and overexpressed proteins and 248 frequently observed protein contaminants. Trypsin or GluC, were set as proteases in which two missed cleavage sites were allowed. Carbamidomethylation of cysteine was set as fixed modification; N-terminal acetylation, methionine oxidation and serine/threonine/tyrosine phosphorylation were set as variable modifications. Initial precursor mass tolerance was set to 6 parts per million (ppm) at the precursor ion and 20 ppm at the fragment ion level. False discovery rates were set to 1% at peptide, phosphorylation site, and protein group level.
Bacterial two-hybrid interaction assays pkaI and a pkaI fragment encoding only the putative PkaI interaction domain (PkaI 257-380 ) were amplified with primers (listed in S2 Table) containing XbaI and KpnI sites, respectively. Subsequently, PCR fragments were cloned with XbaI/KpnI into plasmid pUT18c to generate translational fusions with the catalytic T18 domain of the B. pertussis adenylate cyclase [66] and introduced into the E. coli cya mutant BTH101. These strains were electroporated with the S. coelicolor genomic libraries, either constructed by cloning 250 bp-2000 bp DNA fragments obtained by a partial BfuCI digest [28], or 250-1000 bp fragments generated by shearing the chromosomal DNA in a nebulizer device (Invitrogen) (Tesfazgi & Muth, unpublished), into pKT25. Transformation mixes were plated onto M63 minimal agar containing kanamycin and ampicillin to select for the presence of pUT18c-4778 and a library derived pKT25 derivative. The ability of co-transformants to use lactose resulting in growth on minimal agar is based on a functional adenylate cyclase due to the interaction of the fusion proteins. From growing colonies plasmid DNA was isolated and used in retransformation experiments of BTH101 (pUT18c-4778) to confirm the interaction. From still positive clones, pKT25 was isolated and sequenced to identify the encoded protein and to confirm in frame fusion to the T25 domain. Spores of M145 and the mutants NLΔPkaI and NLΔ4775-4779 were plated on SM-agar. After 2, 3 and 7 days of incubation, photos were taken (A). While NLΔPkaI is delayed in sporulation, differentiation of NLΔ4775-4779 does not visibly differ from that of M145. Genetic complementation of NLΔPkaI restored sporulation proficiency. 1: S. coelicolor M145, 2: NLΔPkaI, 3: NLΔPkaI:: pSET-pkaI, 4: NLΔ4775-4779, 5: NLΔ4775-4779::pSET-pkaI. The delay in sporulation is consistent with the amount of spores that could be isolated after 3 and 7 days of incubation, respectively (B). Whereas resistance of germinating spores to lysozyme and vancomycin was not affected in NLΔPkaI, NLΔ4775-4779 spores showed an increased sensitivity (D), suggesting an impaired spore wall. In contrast, vancomycin resistance of vegetative mycelium of NLΔ4775-4779 was indistinguishable from that of M145 (right panel). (EPS) S4 Fig. Effects of unbalancing phosphorylation activity on vancomycin and lysozyme resistance. About 10 5 spores were streaked on LB agar and serial dilutions of vancomycin and lysozyme spotted. Plates were incubated for three days at 30°C. Whereas resistance of germinating spores to lysozyme was not dramatically affected, spores of the five-fold mutant NLΔ4775-4779, as well as the kinase overexpressing strains (with the exception of pkaI) were highly sensitive to vancomycin. (EPS)