During developmental cell division in sporulation-committed aerial hyphae of streptomycetes, up to a hundred septa are simultaneously produced, in close harmony with synchromous chromosome condensation and segregation. Several unique protein families are involved in the control of this process in actinomycetes, including that of the SsgA-like proteins (SALPs). Mutants for each of the individual SALP genes were obtained, and high-resolution and fluorescence imaging revealed that each plays an important and highly specific role in the control of the sporulation process, and their function relates to the build-up and degradation of septal and spore-wall peptidoglycan. While SsgA and SsgB are essential for sporulation-specific cell division in Streptomyces coelicolor, SsgC–G are responsible for correct DNA segregation/condensation (SsgC), spore wall synthesis (SsgD), autolytic spore separation (SsgE, SsgF) or exact septum localization (SsgG). Our experiments paint a picture of a novel protein family that acts through timing and localization of the activity of penicillin-binding proteins and autolysins, thus controlling important steps during the initiation and the completion of sporulation in actinomycetes.
Cell division is a highly dynamic process of cell wall synthesis and breakdown, and its correct timing and localization is one of the most studied topics in modern microbial cell biology. In Escherichia coli, such control mechanisms include localization of the septal ring exactly at the mid-cell position, mediated through the minCDE and sulA SOS systems (Justice et al., 2000; Autret and Errington, 2001), co-ordinated peptidoglycan synthesis by penicillin-binding proteins (PBPs; reviewed in Holtje, 1998; Errington et al., 2003), and timely DNA duplication and segregation (Sharpe and Errington, 1999; Errington, 2001). While in most bacteria a single septum is formed that forms the cleavage furrow dividing the mother cell into the daughter cells, during sporulation of the Gram-positive mycelial bacterium Streptomyces many septa are simultaneously produced to form long chains of spores. In fact, Streptomyces undergoes two apparently different cell division events (McCormick et al., 1994; Flärdh et al., 2000). Streptomyces growth on solid media starts with the germination of a single spore that develops into a complex vegetative mycelium of branching hyphae (Chater and Losick, 1997). These vegetative hyphae are divided into connected multinucleoid compartments by vegetative septa or cross-walls. Environmental signals such as nutrient depletion result in the development of initially aseptate aerial hyphae while part of the vegetative mycelium lyses. Eventually, the aerial hyphae are dissected into spores by specialized sporulation septa, producing chains of connected uninucleoid spores (Chater, 2001), which are subsequently separated by a poorly understood process of autolytic cleavage, to produce mature spores. In contrast to cross-walls, sporulation septa are produced simultaneously and in a highly co-ordinated way. Another important difference with the generally accepted view of microbial cell division is the apparent lack of relatives of FtsA and ZipA, which anchor the spiral ring to the membrane (Errington et al., 2003), and of homologues of minC, minE and sulA. Therefore, the hunt is on for the discovery of novel proteins that facilitate the complex mechanism of cell division in Streptomyces.
The most well-known group of developmental genes is that of the whi genes, which were discovered by their ability to complement the White phenotype of non-sporulating and/or non-pigmented aerial hyphae (Chater, 1972; 1998). In more recent years, several important new families of developmental proteins were discovered, including the rodlins (Claessen et al., 2002) and chaplins (Claessen et al., 2003; Elliot et al., 2003), hydrophobins and hydrophobin-like proteins, respectively, providing a water-repellant sheath around the aerial hyphae, and the WhiB-like (Wbl) proteins, a family of small regulatory proteins with diverse targets that are also found in non-sporulating actinomycetes such as Mycobacterium (Soliveri et al., 2000).
An emerging group of novel developmental regulators is that of the SsgA-like proteins (SALPs), which occur exclusively in sporulating actinomycetes (reviewed in van Wezel and Vijgenboom, 2004). Seven homologues occur in Streptomyces coelicolor (designated SsgA–G), and six in Streptomyces avermitilis (van Wezel and Vijgenboom, 2004). All SALPs are relatively small (125–142 aa) proteins (Keijser et al., 2003), sharing an average amino acid similarity of 30–40%. The two members that have been studied so far, SsgA and SsgB, are essential for sporulation. The best-studied example is SsgA, which was originally identified as an effecter of cell division in Streptomyces griseus (Kawamoto and Ensign, 1995), and specifically stimulates sporulation-specific cell division (van Wezel et al., 2000). ssgA null mutants show strongly reduced sporulation, although some viable spores are produced on mannitol-containing media (Jiang and Kendrick, 2000; van Wezel et al., 2000). The different timing and expression of ssgA explains some of the major differences in developmental control between the phylogenetically divergent species S. coelicolor and S. griseus, such as the ability of the latter strain to produce submerged spores, and the more dominant developmental role of γ-butyrolactones in this organism (Yamazaki et al., 2003; Traag et al., 2004). Mutation of ssgB resulted in a non-sporulating phenotype under all conditions, producing very large white colonies (Kormanec and Sevcikova, 2002; Keijser et al., 2003)
The non-sporulating phenotypes of the ssgA and ssgB null mutants, and the fact that they occur exclusively in sporulating actinomycetes, suggested a role for the SALPs specifically during sporulation. We addressed this issue by the creation and intensive study of knockout mutants of all ssgA-like genes of S. coelicolor, showing that each SALP is involved in a specific step in the sporulation process, from septum-site selection to the ultimate stages of spore maturation. A working model as to when and how these proteins carry out their function is proposed.
Phylogenetic analysis of SALPs
The ssgA-like genes have so far only been identified in actinomycetes that undergo extensive morphological development, and predominantly in streptomycetes. We identified a homologue of ssgB in the partial genome sequences of the actinomycetes Thermobifido fusca and Kineococcus radiotolerans, but it is absent from Mycobacterium, Corynebacterium, Nocardia and Rhodococcus. The streptomycetes whose genome sequence has been completed (S. coelicolor and S. avermitilis) or almost completed (Streptomyces scabies) contain six or seven ssgA-like genes, and on the basis of hybridization and genomics data we ascertained that Streptomyces lividans contains the same seven SALPs as S. coelicolor (i.e. SsgA–G). Phylogenetic analysis of all known SALP protein sequences revealed three main branches, namely the SsgA branch, the SsgBG branch and the SsgDE branch (Fig. 1). SsgC and SsgF of S. coelicolor do not cluster with any of the other proteins, thus forming separate branches, and we have so far only identified homologues of these proteins in S. lividans. Analysis of a zoo blot of various streptomycetes (S. albus, S. fradiae, S. lividans 1326, S. coelicolor, S. griseus, S. ramocissimus, S. mobarense, S. cinnamoneus) revealed homologues of SsgA, SsgB, SsgD, SsgE in all streptomyces species analysed, and typically also SsgG (though absent from S. avermitilis), suggesting they constitute the core set of SALPs, which is supplemented by two to three additional proteins in probably all streptomycetes.
Construction of mutants
The genomic organization around ssgC (SCO7289), ssgD (SCO6722), ssgE (SCO3158), ssgF (SCO7175) and ssgG (SCO2924) is shown in Fig. S1. Considering that the genes located downstream of ssgC–ssgG are divergently transcribed, polar effects due to insertional gene replacements were not anticipated. The individual mutants for each of these genes were created using targeted gene replacement strategies, which are detailed in the supplementary experimental procedures in Appendix S1. For all individual mutants multiple independent recombinants were obtained, and the correct recombination event was verified by polymerase chain reaction (PCR) and Southern hybridization (data not shown). The exact sections of the genes that were replaced by the apramycin resistance cassette aacC4 were (relative to the translational start): −5/+102 of ssgC, +57/+270 of ssgD, and precisely the coding regions of ssgE, ssgF and ssgG genes. As ssgE partially overlaps the divergently transcribed open reading frame (ORF) SCO3157, which encodes a putative PBP, the latter ORF was restored by integration of plasmid pGWS124, resulting in a wild-type SCO3157, but still mutant ssgE. This mutant, called GSE1*, was studied alongside GSE1 to ensure that the ssgE mutant phenotype was solely due to the replacement of ssgE itself and not (entirely or partly) to the disruption of SCO3157. All mutants were restored to the phenotype of the parental strain by the introduction of a low-copy-number vector harbouring the relevant ssgA-like gene and its promoter sequences. In all experiments described in this article, we simultaneously analysed the mutants and their genetically complemented derivatives, to verify that all observed abnormalities were due solely to the gene replacement and not to a second-site mutation.
Phenotypes of the mutants
All seven individual SALP mutants of S. coelicolor were plated together with the parental strain M145. Sporulation was then assessed visually (spores are grey-pigmented), and verified by phase contrast microscopy and viable counts. The ssgA mutant GSA3 had a non-sporulating (White) phenotype on glucose-containing media (not shown), while it produced a small amount of spores on the mannitol-containing soy flour mannitol (SFM) medium (Fig. 2), in accordance with our earlier observations (van Wezel et al., 2000). The ssgB mutant GSB1 displayed the typical unconditional non-sporulating phenotype (Fig. 2) (Keijser et al., 2003).
The new SALP mutants ΔssgC–G all produced grey-pigmented spores after 6 days, although ssgG produced significantly fewer spores, highlighted by its light grey appearance (Fig. 2). SsgG belongs to the SsgBG branch of the phylogenetic tree, and both proteins are important for the early stages of septum formation (Keijser et al., 2003 and below). Both the ssgE mutant and ssgF mutant hypersporulated on mannitol-containing media, while there was no significant difference on glucose-containing media (data not shown). Analysis of impression prints by phase contrast microscopy showed that both mutants produced short spore chains, consisting of mostly three or four spores in the ssgF mutant, while predominantly single spores where observed for the ssgE mutant.
Light microscopy indicated that significant differences in the degree of sporulation also occurred between the ssgC–ssgF mutants, verified by repeated spore analyses. Spores were prepared, and diluted to the same spectrophotometric density. Total counts were then performed using a haemocytometer and viable counts by plating dilutions (1–104 spores per plate) on SFM agar plates. There was a good correlation between microscopic and viable counts, with increased sporulation of GSE1 and GSF1, and reduced sporulation of GSG1. However, spore preparations from the ssgD mutant GSD1 consistently contained at least an order of magnitude fewer spores than expected on the basis of its pigmentation, as judged by viable counts and microscopy, suggesting that production of the WhiE spore pigment (Kelemen et al., 1996) was enhanced in this mutant.
Analysis of the mutants by cryo-scanning electron microscopy (cryo-SEM)
Surface-grown colonies of S. coelicolor ssgC–G mutants and the parental strain M145 were analysed in detail by cryo-scanning electron microscopy (cryo-SEM) (Fig. 3). S. coelicolor M145 produced long spore chains with spore septa at regular intervals, producing approximately 1.0 µm long spores (Fig. 3A and B). Aerial hyphae of the ssgC mutant produced extraordinarily long spore chains (Fig. 3C), and contained spores of highly variable lengths (Fig. 3D). In some of the large spores, indentations were visible at the mid-cell position, suggesting initiated but unfinished septation (Fig. 3D, insert). The ssgD mutant produced many straight and undifferentiated aerial hyphae, while spores appeared relatively normal (Fig. 3E and F). Consistent with light microscopy (above), GSE1 and GSF1 sporulated abundantly. The ssgE mutant GSE1 showed large areas with only free spores, suggesting premature autolysis-driven spore separation. Perhaps as a result of this spore maturation defect, many of the spores had aberrant sizes in length and occasionally in width (Fig. 3G and H). GSF1 also produced spores of variable sizes (Fig. 3I and J). Cryo-SEM failed to visualize the short spore chains typical of the ssgF mutant and confirmed by phase contrast microscopy and confocal fluorescence microscopy (FM) (below); this is most likely a combination of the cryo-fixation procedure, where short spore chains are easily lost, and the fact that longer hyphae obscure the shorter ones. ssgG mutant GSG1 produced clearly fewer spore chains with normal overall lengths (Fig. 3K), but always containing many irregularly sized spores, most of which were exactly two, three or four times the normal size (detailed in Fig. 3L). As discussed below, septa seem to be occasionally ‘skipped’ in the ssgG mutant.
Confocal fluorescence microscopy
Confocal fluorescence microscopy was used to visualize nucleoid distribution and active peptidoglycan synthesis and breakdown during sporulation-specific cell division. Nucleic acids were stained with the fluorescent dye propidium iodide (PI), and confirmed using the alternative fluorescent dyes DAPI and 7-aminoactinomycin D (7-AAD). Accumulation of peptidoglycan precursors (as a result of active peptidoglycan (PG) synthesis or breakdown) was visualized by staining with the fluorescein-labelled lectin wheat germ agglutinin (WGA). WGA binds to short oligomers of N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM), which are alternately coupled to form the peptidoglycan strands (reviewed in Holtje, 1998). Fully polymerized peptidoglycan, such as in the mature spore wall, is not recognized by WGA.
The wild-type sporulation process on surface-grown cultures is visualized in Fig. 4A, showing fully developed 6-day-old S. coelicolor M145. Mature spore chains (spo) are not stained with fluo-WGA (Fig. 4A, right), but DNA staining (PI) visualizes the completed segregation and condensation of chromosomes (Fig. 4A, middle). In contrast, actively dividing aerial hyphae (aer) are stained very well with fluo-WGA, clearly visualizing the build-up of individual septa in a ladder-like fashion, highly similar to FtsZ ladders (Schwedock et al., 1997). Perhaps surprisingly, the DNA still appeared evenly distributed in the aerial hypha, a sign that (completion of) DNA segregation follows septal peptidoglycan synthesis. Earlier stages of sporulation were studied using younger plates (2–4 days old; light image in Fig. 4B, left). WGA binding to the outer side of in particular the spore poles revealed NAM-NAG oligomers, most likely due to peptidoglycan degradation; this shows that these spores undergo the final step of spore maturation, namely the autolytic separation of the cell wall separating the spores (Fig. 4B, right + insert). Expectedly, when wild-type spores had fully matured, no WGA staining was observed. DNA had already fully segregated at this stage (Fig. 4B, middle).
The ssgA and ssgB mutants have a phenotype typical of early sporulation mutants, and fail to produce septa in the aerial hyphae, while vegetative septation was not significantly affected (van Wezel et al., 2000; Keijser et al., 2003). In contrast, the ssgC mutant showed extensive de novo septum formation in younger aerial hyphae (WGA staining). Surprisingly, co-ordination of septum formation and DNA segregation was lost in over half of all aerial hyphae seen in this mutant; many empty pre-spore compartments were observed (Fig. 4D, small arrows), while the intense PI-stained foci indicated the presence of multiple copies of the genome in others (Fig. 4D, arrowheads). This is clearly visualized in the DNA/fluo-WGA overlay (Fig. 4D, right). Occasionally, spore chains with normal DNA distribution were observed, but these invariably contained mature spores with aberrant sizes (and typically double-sized; Fig. 4C). Spores of the ssgD mutant occasionally showed irregular sizes, although DNA distribution appeared normal (Fig. 4E). As described above, mutation of ssgE or ssgF resulted in strains that sporulated very well, but produced predominantly single spores (ssgE; Fig. 4F) or short spore chains (ssgF; Fig. 4G). DNA had segregated normally, as shown the regular pattern of PI staining (Fig. 4, middle panels). In contrast to M145, after 7 days of incubation both late sporulation mutants still had many immature spores, and NAG-NAM oligomers were readily identified by WGA staining. In the case of the ssgE mutant, only chains of spores were visualized by WGA, and at intersections between adjacent spores. Intriguingly, a large proportion of the ssgF spores contained WGA-stained foci exclusively at the poles, and always at the outside of the spore walls, similar to M145 in an early stage of the sporulation process (around 4 days old; Fig. 4G). In contrast to the parental M145, where such polar staining disappeared when spores matured, the spores of the ssgF mutant always showed this typical polar WGA staining, suggesting a distinct defect in the final stages of autolytic spores separation. Furthermore, many spore chains contained spores that were rotated by 90° (Fig. 4G, squares). While such ‘rotated’ spores were never observed in M145, they constituted approximately 14% (28 out of 201) of all ssgF mutant spores. In fact, many spore chains showed a clear transition from normally oriented to rotated spores (Fig. 4H + insert).
As was already apparent from the SEM micrographs, mutation of ssgG resulted in many larger compartments with twice, three times or sometimes four times the regular spore size (Fig. 4I + insert). DNA staining clearly identified multiple copies of well-segregated chromosomes in these larger compartments (Fig. 4I), with the DNA content proportional to the increase in spore size (two genomes in a double-sized spore etcetera). Thus, DNA segregation could be completed in the absence of septum synthesis (see Discussion).
Transmission electron microscopy (TEM)
Transmission electron microscopy (TEM) was used to closely examine hyphae, septa and spores (Fig. 5). S. coelicolor M145 showed typical vegetative cross-walls (Fig. 5A); aerial hyphae produced typical chains of immature (Fig. 5B) and fully developed mature spores (Fig. 5C). The spore chains were surrounded by a sheath (small arrow in Fig. 5B and C) that most likely represents the remnants of the parental hyphal wall containing the rodlet layer (D. Hopwood, pers. comm.).
The ssgA and ssgB mutants were studied previously; on mannitol-containing media, the ssgA mutant produced occasional regular spore chains (van Wezel et al., 2000), while the ssgB mutant was completely devoid of any septa in the sporulation stage (Keijser et al., 2003). Both strains had normal vegetative septa. Interestingly, the ssgC mutant strongly resembled the ssgA-overexpressing strain GSA3. Around 60% of all cross-walls were irregular (insert in Fig. 5D), as counted from many TEM micrographs of biologically independent samples. In often occurring aberrant vegetative hyphae of the ssgC mutant, unfinished cross-walls were observed, which appear like spots of PG in these TEM cross-sections, but are most likely small septal rings in three dimensions. The distance between them was on average 2 µm (Fig. 5D). The spores of GSC1 were also heteromorphous, with varying lengths and spore walls (Fig. 5E). Interestingly, the ssgC mutant strongly fragments in submerged cultures, and overexpression of ssgC inhibits sporulation, again in line with an opposite role of ssgC as compared with ssgA (not shown).
The ssgD mutant showed pleiotropic defects in the integrity of hyphae and spores, perhaps reflecting its high and life cycle-independent expression (Traag et al., 2004 and microarray data). Many of the spores had a cell wall similar to that of the lateral wall of non-sporulating aerial hyphae, lacking the typical thick peptidoglycan layer (Fig. 5F). Furthermore, immature spore chains (Fig. 5G) had a highly irregular appearance, with a squashed appearance and with variable spore sizes. Mature spore chains often showed a mix of relatively normal and aberrant spores (Fig. 5H). Analysis of spore preps of GSD1 revealed no difference in heat or lysozyme tolerance to spores from the parental M145 (not shown).
Like most SALP mutants, the ssgE mutant produced spores with variable spore sizes (Fig. 5I; see also Fig. 3H). Occasionally, mature spore chains (surrounded by a sheath) were observed but there consistently was a large void between the individual spores (Fig. 3J).
The ssgF mutant produces chains consisting of regular-looking spores containing a typical thick spore wall. However, TEM also identified spore chains with spores rotated by 90°, such as also seen by confocal FM (see Fig. 4G and H); here, the individual spores were not attached to each other in the normal head-to-tail fashion, i.e. with their spore poles adjacent to each other, but rather in a parallel fashion, resulting in a wider spore chain (800 nm instead of the regular 600 nm; Fig. 5K and L).
As SEM and FM analysis of the ssgG mutant, also TEM revealed many large spores, with exactly twice, three times or even four times the normal length (Fig. 5M). TEM clearly confirms the presence of multiple genomes in the larger spores, while there is no evidence of any cell wall material between the almost completely segregated chromosomes. This shows that septal peptidoglycan synthesis and DNA segregation are uncoupled in this mutant. Interestingly, while the chromosomes are almost completely separated, they are still attached to each other through a thin filamentous structure (Fig. 5N).
DNA microarray analysis of cell division and developmental genes
The data above established that the SALPs play a role in the control of peptidoglycan synthesis and autolysis. To judge which PBPs and autolysins present on the S. coelicolor genome are most obvious candidates for peptidoglycan synthesis during development, and relate functionally to SALPs, we performed microarray analyses on S. coelicolor (Fig. S2). Transcription of SALP genes was compared with that of genes involved in the maintenance of the peptidoglycan (PBPs and autolysins). As for the SALP genes, microarray analysis of ssgA and ssgC produced data below the confidence threshold. Detailed analysis performed previously showed that transcription of ssgA and its activator ssgR is induced towards the onset of sporulation, and ssgD is strongly transcribed throughout development (Traag et al., 2004). Microarray data (Fig. 6B) support this for ssgD and ssgR, and show that ssgB is strongly activated towards sporulation. Of the two spore maturation genes, ssgE is expressed at a low level, while ssgF is strongly expressed, both with a life cycle-independent expression profile. Finally, ssgG is expressed relatively strongly during vegetative growth and again towards sporulation, suggesting a possible role in both developmental stages. The expression profiles were confirmed by promoter probing using the redD promoter-probe system. Particularly ssgB, ssgD and ssgF were expressed in the aerial hyphae (not shown). The transcription profiles were all confirmed by reverse transcription polymerase chain reaction (RT-PCR) analysis and by promoter probing, except for ssgF; while microarrays predicted high and almost growth phase-independent expression, detailed analysis showed that ssgF is in fact expressed during sporulation, and at a low level.
Several pbp genes are developmentally controlled (Fig. 6B). SCO3847 and SCO4013, both encoding homologues of Bacillus PBP3, were upregulated during both the onset of both vegetative and aerial growth; SCO5487 and SCO5039 were switched on primarily during early vegetative growth, while the ftsI-like genes SCO3156 and SCO3771 were induced at a time related to aerial growth, although transcription of SCO3156 was switched off again during sporulation.
Autolysins are responsible for the breakdown of peptidoglycan, e.g. during spore separation. Several autolysins are developmentally controlled; these are either switched on during both vegetative and aerial growth (SCO2116, for one of several AmpD-type autolysins), or relate to a specific developmental stage, namely at a time related to aerial growth (SCO5466, encoding a lysozyme-like autolysin) or to sporulation (SCO4132, encoding the secreted lytic transglycosylase (SLT) autolysin, the only lytic transglycosylase in S. coelicolor). Interestingly, in collaboration with Dr Rigali (Liege, Belgium) we recently discovered that SCO5466 is strongly downregulated in a crp mutant, which fails to produce the cAMP receptor protein CRP (Piette et al., 2005). This mutant is strongly disturbed in autolysis of the pre-spore chain, confirming the likely involvement of SCO5466 in spore maturation.
This article discusses the role of the SALPs in the control of specific aspects of the sporulation process, from initiation of septal peptidoglycan synthesis to the separation of spores from the maturing pre-spore chain. To better discuss their individual functions, we discriminate between the following landmark events during the sporulation process in aerial hyphae of Streptomyces (see Fig. 6): (I) pre-spore formation, consisting of (IA) septum site selection; (IB) septum initiation; (IC) septum growth; (ID) DNA segregation and condensation; (IE) septum closure; and (II) spore maturation, consisting of (IIA) growth (thickening) of the spore wall peptidoglycan; (IIB) spore separation by PG autolysis; (IIC) spore release. Our data show that each of the individual SALP genes affects one of these landmark events, and relate to the control of peptidoglycan synthesis or breakdown. This includes the first examples of proteins that control the exact sites where septa are initiated (stage IA, by SsgG) and autolytic spore separation (stage IIBC, by SsgE and SsgF). The phenotypes of the mutants and the suggested function of the genes during sporulation are summarized in Table 1.
Table 1. Phenotype of the SALP mutants and the timing of activity during development.
Strictly non-sporulating phenotype, very large (‘immortal’) colonies
Correlates temporally to growth cessation of aerial hyphae before onset of sporulation
Irregular spores, imperfect segregation of DNA, very long ladders of spore septa
Controls septum site initiation and DNA segregation (IB)
Aberrant spore wall
Lateral cell wall synthesis (IIA)
Predominantly single spores due to accelerated autolysis
Correct timing of spore dissociation (IIB–C)
Short spore chains; old spores were stained with WGA at the outside of the spore poles; rotated spores
Final stages of spore separation (IIB–C)
Septa are regularly skipped, without affecting DNA segregation. Many spores exactly two, three or four times the normal size
Controls septum site localization (IA)
Control of peptidoglycan synthesis during pre-spore formation
Microarray data (Fig. S2) and previously published transcriptional analyses revealed that ssgA and ssgB and to a lesser extent ssgG are induced during late aerial growth and early sporulation. Deletion of these genes results in complete (ssgA, ssgB) or partial (ssgG) loss of the ability to sporulate. The aerial hyphae of the ssgA and ssgB mutants are essentially aseptate. Overexpression of ssgA resulted in hyperseptation of both vegetative and aerial hyphae, and TEM micrographs revealed many extremely thick and unfinished cross-walls in the vegetative hyphae, leaving a distinct gap in the middle (van Wezel et al., 2000). This strongly suggests that SsgA activates the initial steps in the polymerization of septal peptidoglycan. ssgB is the most highly conserved cell division-related gene in actinomycetes, and a rare example of a whi gene that is not a transcriptional regulator. ssgB mutants produce very long aseptate aerial hyphae, forming ‘immortal’ colonies (Keijser et al., 2003). The sporulation process is initiated as soon as aerial growth ceases (Chater, 2001), requiring a block of PBP2 enzyme activity, and this signal most likely depends on SsgB. Interestingly, many granules accumulate in the hyphae of the ssgB mutant (Fig. S3 in Supplementary material), which due to their electron-dense nature are likely to contain unincorporated cell wall precursor material. The same granules were also observed in the PBP2 mutant (a kind gift from Dr G. Hobbs), providing evidence of a functional relationship between SsgB and PBP2.
The function of SsgC is the reverse of that of SsgA. ssgC mutants produce the spiral-like distribution of peptidoglycan along the wall of vegetative hyphae, typical of SsgA-overexpressing strains. Also, extraordinary long ladders of septa are often observed in this mutant, with disturbed DNA segregation, similar to strains overexpressing SsgA. ssgC is one of the two strain-specific SALPs of S. coelicolor. So far, we have only found homologues of ssgC in strains with low expression of ssgA (such as S. coelicolor and S. lividans; E.E.E. Noens and G.P. van Wezel, unpubl. data). On the basis of our observations, we propose that SsgC functions as an antagonist of SsgA.
The ssgG gene is a rare cell division mutant, and an opportunity to improve our understanding of how multiple septum sites are selected and chromosomes are segregated in streptomycetes. Typically, highly regular ladders of septa are produced in sporulation-committed aerial hyphae (Schwedock et al., 1997; Grantcharova et al., 2005). In the ssgG mutant, at irregular intervals septa were simply ‘missing’, resulting in spores exactly twice, three times or even four times the size of normal spores. Excitingly, the chromosomes were well segregated in these multiple-sized spores, resulting in long spores with up to four chromosomes joined only by a thread-like filamentous structure. TEM showed the complete absence of any cell wall material (peptidoglycan) between the segregated chromosomes. So while in wild-type cells septal peptidoglycan synthesis is always initiated before DNA segregation (e.g. Fig. 4A), the ssgG mutant reveals that in fact the initiation of septal peptidoglycan synthesis is not a requirement sine qua non for correct DNA segregation. This sheds new light on this intriguing process.
ssgD mutants produce many pre-spores and spores with aberrant hyphal walls, including spores with a wall the width of regular hyphae, and irregular pre-spore chains, suggestive of weakened lateral peptidoglycan. The synthesis of lateral PG is carried out by PBP2 (Den Blaauwen et al., 2003; Errington et al., 2003) and the highly similar protein encoded by SCO2897. Interestingly, mutational analysis showed that both PBP2 and SCO2897 – revealed by microarray analysis to be strongly expressed throughout the S. coelicolor life cycle – are indeed important for lateral wall PG synthesis (G. Hobbs, proceedings of ISBA XIII AND Melbourne, 2003), suggesting that SsgD allows the correct functioning of at least one of these two PBPs.
Spore maturation: SsgEF control autolytic spore separation
While the insight in the sporulation process in streptomycetes is rapidly improving, little is known of the final stage, the autolytic cleavage of the peptidoglycan connecting the pre-spores, resulting in mature and separated spores. Our data show that SsgE and SsgF play a specific role in the control of this process. Strains lacking ssgE sporulated very well, with normal septation and DNA segregation, but produced predominantly single spores. As pre-spores were invariably organized in spore chains, the high frequency of single mature spores must have arisen from an accelerated maturation process followed by a non-co-ordinated separation into single spores. This implicates SsgE as a checkpoint for the correct timing of spore dissociation. Interestingly, all streptomycetes whose genome has been sequenced have one SALP that is predicted by the TMPred algorithm to have a transmembrane helix, namely SsgF (aa 14–32) in S. coelicolor, SsgW (aa 51–75) in S. scabies and SAV580 (aa 85–103) in S. avermitilis, with a strong preference for the N-terminus facing the cytoplasm. ssgF plays a distinct role during the final stage of spore separation. Deletion of the gene resulted in short spore chains, with spores that were readily stained by WGA, invariably at the poles and at the outside of the spore wall. As de novo synthesis occurs from inside to outside, these peptidoglycan subunits accumulated at the spore poles must have resulted from autolysis of the spore-linking peptidoglycan. Hence, in the absence of SsgF spores fail to complete autolytic detachment due to incomplete breakdown of PG subunits, resulting in more loosely attached pre-spores and allowing almost free rotation. This is substantiated by our observation of a ‘flip movement’ in several spore chains, with a visible transition from normally oriented spores (poles connected head-to-tail) to 90° rotated spores (poles parallel), with an intermediate spore (45° rotated) in between (Fig. 4H + insert). The likely driving force is the stronger hydrophobic interaction between the long sides of the spores as compared with the poles. Thus, SsgF ensures the final step in autolysis of pre-spores.
SALPs and the fate of peptidoglycan
Scientists have been searching for an answer to the important question as to how streptomycetes co-ordinate the synchronous formation of multiple septa and the concomitant segregation of the same number of chromosomes. The cell division machinery is generally very similar to that of other eubacteria, although especially aspects of regulation and localization are different. Which then are the proteins that make sure that septa are laid down simultaneously, and with such striking regularity? It is likely that this involves Streptomyces-specific proteins. The SALPs are obvious candidates; they occur exclusively in actinomycetes that undergo extensive morphological differentiation. Four of the SALPs are present in all streptomycetes, namely ssgA, ssgB, ssgD and ssgE, while ssgG-type genes are also relatively common. Sequence conservation of the predicted gene products varies from 55% (SsgE) to 100% (SsgB), SsgB being the most highly conserved cell division protein in Streptomyces (Flärdh and van Wezel, 2003). Our data now show that all SALPs relate specifically to sporulation-specific cell division, as ssg mutants have various defects in the build-up or degradation of peptidoglycan during sporulation, but show normal vegetative growth and regular cross-walls. The only exception is the ssgC mutant; as discussed, its phenotype is highly similar to a strain overproducing SsgA, showing not only developmental defects, but also the typical enhanced and irregular septation in vegetative hyphae.
The main question that is not so easily answered is what exactly is the mechanism by which the SALPs control the different steps of the sporulation process? At this point, it is important to note that while the classical early whi genes such as whiA, whiB, whiG, whiH and whiN (bldN) encode (predicted) DNA-binding proteins such as σ-factors and GntR-family regulators, and hence most likely function at the level of transcriptional control (and are therefore less obvious candidates for an interaction with PBPs and autolysins), ssgA and ssgB are whi genes that likely encode structural proteins. Interestingly, we recently obtained convincing evidence that at least SsgA and SsgF form multimers with itself in vitro even under denaturing conditions – with up to five-membered multimers confirmed by mass spectrometry of purified SsgF (G. van Wezel, unpubl. data). This indicates that SALP multimerization probably is an essential part of their function. SALPs have no known protein motif (!), and lack DNA-binding activity, which suggests that SALP multimers function by recruiting other proteins to their relevant sites, so as to manage enzymes responsible for the synthesis and autolysis of peptidoglycan.
On the basis of our data we predict the following functional relationships. SsgA – antagonized by SsgC – directly stimulates cell division and the build-up of septal PG, and hence it functionally relates to the developmental PBPs involved in septal PG synthesis, i.e. FtsI, SCO3171 and/or SCO3156. On the basis of the highly similar phenotypes of the respective mutants, PBP2 relates functionally to SsgB, and SCO2897 and/or PBP2 – which are both important for lateral wall PG synthesis – to SsgD. SsgE and SsgF control autolytic spore cleavage, with SsgE preventing premature spore release, a function most likely carried out by SCO5466, and SsgF promoting the cleavage of peptidoglycan strands between the spores in almost mature pre-spore chains, in prokaryotes carried out by SLT. In accordance with this, SCO5466 and SLT are transcribed in a highly developmental way, suggesting their involvement primarily in spore separation. Finally, SsgG is the first example of a protein specifically involved in the co-ordination of multiple septation, a unique feature of Streptomyces cell division. Solving the question as to how SsgG determines the exact sites where cell division is initiated will greatly improve our understanding of this intriguing phenomenon.
To further characterize the nature of all functional interactions, we are currently conducting extensive expression analyses in all SALP mutants using genomics approaches, and concentrate on the identification of the exact interaction partners for the SALPs.
Bacterial strains and media
The bacterial strains described in this article are listed in Table 2. E. coli K-12 strains JM109 (Sambrook et al., 1989) and ET12567 (MacNeil et al., 1992) were used for plasmid propagation, and were grown and transformed by standard procedures (Sambrook et al., 1989). E. coli BW25113 (Datsenko and Wanner, 2000) was used to create and propagate the S. coelicolor cosmids used for the creation of knockout mutants of S. coelicolor M145. E. coli ET12567 containing pUZ8002 was used for conjugation to S. coelicolor (Kieser et al., 2000). Transformants were selected in l-broth containing the appropriate antibiotics.
Table 2. Bacterial strains described in this article.
Streptomyces coelicolor A3(2) M145 was obtained from the John Innes Centre strain collection, and was the parent for the previously created ssgA mutant GSA3 (van Wezel et al., 2000) and ssgB mutant GSB1 (Keijser et al., 2003), and for the ssgC–G mutants described in this article. All media and routine Streptomyces techniques are described in the Streptomyces manual (Kieser et al., 2000). SFM medium was used for making spore suspensions and R2YE agar plates for regeneration of protoplasts and, after the addition of the appropriate antibiotic, for selecting recombinants. Phenotypic characterization of mutants was performed on SFM, on R2YE (not shown) and on minimal medium agar plates (not shown) with glucose (MMgluc) or mannitol (MMman) as the sole carbon source (Kieser et al., 2000).
Plasmids, constructs and oligonucleotides
All plasmids and constructs are summarized in Table 3. Required PCRs were performed with Pfu polymerase (Stratagene), in the presence of 10% (v/v) DMSO, with an annealing temperature of 58°C. The oligonucleotides are listed in Table S1 in Supplementary material.
Table 3. Plasmids and constructs.
Streptomyces/E. coli shuttle vector (5–10 and around 100 copies per genome respectively)
pHJL401 with 1.8 kb fragment harbouring ssgC (–1075/+894, relative to ssgC)
pHJL401 with 900 bp fragment harbouring ssgD (–291/+608, relative to ssgD)
pHJL401 with 1.8 kb fragment harbouring ssgE (–732/+1027, relative to ssgE)
pHJL401 with 1 kb fragment harbouring ssgF (–404/+595, relative to ssgF)
pHJL401 with 1.2 kb fragment harbouring ssgG (–459/+679, relative to ssgG)
pBR-KO with 1.7 kb fragment harbouring the last 228 bp of ssgE and the last 1127 bp of SCO3157
Construct for disruption of S. coelicolor ssgC
Construct for disruption of S. coelicolor ssgD
Mutant cosmid with coding region of S. coelicolor ssgE replaced by aacC4, for gene replacement of ssgE
As E87/ΔssgE, but for S. coelicolor ssgF
As E87/ΔssgE, but for S. coelicolor ssgG
General cloning vectors. pIJ2925 (Janssen and Bibb, 1993) is a pUC19-derived plasmid used for routine subcloning. pBR-KO is a PBR322-based vector for gene disruption in streptomycetes, which has the EcoRI–HindIII section replaced by the multiple cloning site of pUC18, and tsr (ThioR) inserted into the BamHI site (Keijser et al., 2003). For cloning in Streptomyces we used the shuttle vector pHJL401 (Larson and Hershberger, 1986), which has the pUC19 ori for maintenance in E. coli and SCP2*ori (around five copies per chromosome) for maintenance in S. coelicolor.
Constructs for the deletion of ssgA-like genes. For the creation of vectors for the gene replacement of ssgC, ssgD, ssgE, ssgF and ssgG, different strategies were deployed, all resulting in gene replacement constructs where (part of) the coding regions were replaced by the apramycin resistance cassette aacC4 (Blondelet-Rouault et al., 1997). The coding sequences of the respective genes that were replaced by aacC4 were: −5/+102 for ssgC, +57/+270 for ssgD, and the entire coding regions of ssgE, ssgF and ssgG. For details on the disruption constructs, for constructs for the complementation of the SALP mutants and for the creation of the SALP mutants, see Supplementary material.
Electron microscopy. Morphological studies of surface-grown aerial hyphae and spores of S. coelicolor M145 and mutant derivatives by cryo-SEM was performed as described previously, using a JEOL JSM6700F scanning electron microscope (Keijser et al., 2003). TEM for the analysis of cross-sections of hyphae and spores was performed with a Philips EM410 transmission electron microscope as described previously (van Wezel et al., 2000),
Confocal fluorescence microscopy. Impression preparations from the surface of 6-day-old colonies on SFM plates were taken and fixed with methanol. For staining of the DNA, the coverslips were incubated with propidium iodide (1 µg ml−1) (Sigma) in the dark for 15 min at room temperature, allowed to air dry and positioned on a microscope slide containing a drop of 20% glycerol, mycelium facing downwards. Staining of the cell wall was performed in 25 mM borate buffer pH 8, 0.9% sodium chloride + 2 mg ml−1 BSA, 5 µl ml−1 FITC-WGA (Biomedica) for 15 min in the dark at room temperature, allowed to air dry and positioned in a drop of 20% glycerol on a microscope slide. Samples were analysed with a Leica TCS-SP2 confocal fluorescence microscope, equipped with an oil-immersed 100× Planapo objective and a CCD camera. Digital images were assembled using ADOBE Photoshop software.
We are very grateful to Dr Glyn Hobbs (Liverpool, UK) for the PBP2 mutant and for discussions, to Hans van der Meulen for excellent technical assistance with TEM, and to Dr Erik Vijgenboom and Dr Barend Kraal (Leiden, the Netherlands) and Professor Sir David Hopwood for discussions. This work was supported by grants from the Netherlands Academy for Arts and Sciences (KNAW) to G.P.v.W. and from the BBSRC (UK) to C.P.S.