Sporulation in aerial hyphae of Streptomyces coelicolor involves profound changes in regulation of fundamental morphogenetic and cell cycle processes to convert the filamentous and multinucleoid cells to small unigenomic spores. Here, a novel sporulation locus consisting of smeA (encoding a small putative membrane protein) and sffA (encoding a SpoIIIE/FtsK-family protein) is characterized. Deletion of smeA-sffA gave rise to pleiotropic effects on spore maturation, and influenced the segregation of chromosomes and placement of septa during sporulation. Both smeA and sffA were expressed specifically in apical cells of sporogenic aerial hyphae simultaneously with or slightly after Z-ring assembly. The presence of smeA-like genes in streptomycete chromosomes, plasmids and transposons, often paired with a gene for a SpoIIIE/FtsK- or Tra-like protein, indicates that SmeA and SffA functions might be related to DNA transfer. During spore development SffA accumulated specifically at sporulation septa where it colocalized with FtsK. However, sffA did not show redundancy with ftsK, and SffA function appeared distinct from the DNA translocase activity displayed by FtsK during closure of sporulation septa. The septal localization of SffA was dependent on SmeA, suggesting that SmeA may act as an assembly factor for SffA and possibly other proteins required during spore maturation.
The life cycle of streptomycetes is one of the most complex developmental processes among prokaryotes. A germinating spore grows out to form a vegetative mycelium of branched hyphae. Nutrient limitation and possibly other stimuli then trigger a complex multicellular development, which includes the emergence of a specialized aerial mycelium on the surface of the colony and the production of antibiotics and other secondary metabolites (Chater and Losick, 1997; Elliot et al., 2007). Subsequently, a profound reprogramming of regulation and modification of fundamental morphogenetic and cell cycle processes take place in the aerial hyphae, leading to the differentiation of these filamentous, multinucleoid cells into chains of unigenomic spores that have thick cell walls and low metabolic activity, permitting dispersal and long-term survival.
A key morphogenetic event in the sporulation of aerial hyphae is the switch in the mode of cell division from the widely spaced thin crosswalls, or vegetative septa, that delimit the multinucleoid hyphal cells in the substrate mycelium to sporulation septation, which partitions the sporogenic aerial hyphal cell into prespores (Chater and Losick, 1997; Chater, 2000). Sporulation septa are thicker, regularly spaced between every nucleoid, and lead to deep cell wall constriction and eventually separation of daughter cells (Wildermuth and Hopwood, 1970; Flärdh and Van Wezel, 2003). Sporulation requires a developmentally induced upregulation of ftsZ (Flärdh et al., 2000), encoding the tubulin homologue that directs bacterial cell division (Margolin, 2005). FtsZ then polymerizes into helical filaments that are remodelled to form a series of regularly spaced FtsZ rings along the sporulating hyphal cell, and subsequently organize the formation of sporulation septa (Schwedock et al., 1997; Grantcharova et al., 2005). Simultaneously, mechanisms for segregation of chromosomes have to be activated in order to ensure that each spore inherits one copy of the genome. The ParA and ParB proteins, which are similar to proteins involved in chromosome and plasmid partitioning in many other bacteria, have a role in developmentally controlled nucleoid partitioning in Streptomyces coelicolor (Kim et al., 2000; Jakimowicz et al., 2006). However, the relatively mild phenotype caused by the lack of ParA and ParB (only about 13% of the spores contained aberrant amounts of DNA) implies that additional proteins are involved (Kim et al., 2000). It has been previously reported that the constriction of sporulation septa in streptomycetes initiates before nucleoid partitioning is complete (Schwedock et al., 1997; Miguelez et al., 1998; Flärdh, 2003; Noens et al., 2005), and thus, some mechanism for transport or movement of DNA through the closing septa must exist. One candidate likely to be involved in this process is the S. coelicolor FtsK protein (SCO5750), which is homologous to the SpoIIIE/FtsK DNA translocases (Errington et al., 2001). Indeed, FtsK was recently reported to localize at sporulation septa and to affect the integrity of chromosomes after sporulation in S. coelicolor, suggesting that it may be involved in sporulation-specific genome segregation (Wang et al., 2007). However, as the majority of spores of the ftsK null mutant contained intact chromosomes, other mechanisms are likely to be involved as well.
Some of the key regulators of the development of aerial hyphae into spores in S. coelicolor are known, but our overall understanding of the pathways and mechanisms is still fragmentary. Commitment to sporulation requires the RNA polymerase sigma factor WhiG (Chater et al., 1989). In the absence of whiG, the aerial hyphae remain straight and form widely spaced vegetative-type hyphal crosswalls (Flärdh et al., 1999). Two target genes under the control of whiG have been characterized, whiH encoding a GntR-family repressor and whiI encoding a response regulator (Ryding et al., 1998; Ainsa et al., 1999). whiH and whiI are both required for efficient sporulation septation, but the respective mutants have different phenotypes (Ainsa et al., 1999; Flärdh et al., 1999; Tian et al., 2007). The genes whiA and whiB constitute a second converging pathway in the control of sporulation in aerial hyphae and are both required for sporulation septation in the aerial hyphae (Chater, 1972; Flärdh et al., 1999; Ainsa et al., 2000). Orthologues of WhiA are ubiquitous among Gram-positive bacterial genomes, and WhiB is the founding member of a large family of Actinobacteria-specific proteins (Soliveri et al., 2000; Gao et al., 2006). Both whiA and whiB are upregulated during sporulation in a whiG-independent fashion (Soliveri et al., 1992; Ainsa et al., 1999), but their exact functions are unclear. These five early whi genes are all required for full sporulation septation, most spore maturation processes, and the developmental upregulation of a number of genes including ftsZ, parAB and sigF (encoding a σ factor that affects spore maturation), as well as ssgB (which encodes a member of the SsgA-like family of proteins that appear to modify peptidoglycan/cell wall during sporulation) and the genes of the whiE cluster (which encode the biosynthetic genes for the grey spore pigment) (Kelemen et al., 1996; 1998; Flärdh et al., 2000; Chater, 2001; Sevcikova and Kormanec, 2003; Noens et al., 2005; Jakimowicz et al., 2006). Apart from some σWhiG regulon members, no direct targets for any of the sporulation regulatory proteins have been identified, and it is still an open question as to how these early regulatory genes can trigger the profound morphogenetic changes taking place during the differentiation of aerial hyphae into spores.
In this article, we characterize a novel locus whose expression depends on both whiA and the whiG/whiI/whiH pathway, and is constrained to the sporogenic aerial hyphae at a time coinciding with the assembly of multiple FtsZ rings. The locus consists of two genes with new roles in sporulation: one encodes a small putative membrane protein of unknown function (SmeA) and the other a member of the SpoIIIE/FtsK family (SffA). We show that deletion of these genes has a pleiotropic effect on spore development, and that smeA is required for the specific targeting of SffA to the sporulation septa. SffA influences the accuracy of nucleoid partitioning, but has a different function from that of FtsK.
smeA-sffA (SCO1415-16) constitutes a novel sporulation locus
In order to identify new genes involved in spore development, and to further elucidate the regulatory networks controlling spore formation in S. coelicolor, we have analysed gene expression profiles of the wild-type strain M145 and sporulation-deficient mutant strains throughout development using a whole-genome microarray approach (comprehensive results will be published elsewhere, P. Salerno et al., in preparation). In a pilot microarray experiment we compared the developmental gene expression profile of the wild-type strain with those of its congenic mutants lacking whiG, whiA, whiH and whiI to identify genes that were upregulated during development in the parent strain, but failed to be upregulated in at least one of the mutants. This initial analysis detected activation of several known sporulation genes (Fig. S1), including open reading frames (ORFs) in the whiE cluster and ssgB (Kelemen et al., 1998; Kormanec and Sevcikova, 2002). In addition, it revealed a number of previously unrecognized genes which appeared to have sporulation-specific expression patterns (to be described elsewhere), including a putative operon consisting of genes SCO1415 and SCO1416, whose expression was upregulated in the later stages of development in the wild-type strain but not in any of the whi mutant strains examined (Fig. S1). SCO1415 encoded a small protein (63 amino acids) and was designated smeA (small membrane protein). In silico analysis of SmeA revealed a centrally located putative transmembrane domain and indicated a putative N-terminal signal peptide, but did not reveal any motifs to infer a possible cellular function (see Discussion and Fig. 6) (Cserzo et al., 1997; Nielsen et al., 1997). Nevertheless, blast and psi-blast searches revealed that similar proteins are widespread among Streptomyces and Frankia (see Discussion). SCO1416 encoded a protein with an N-terminal transmembrane region and a C-terminal domain with similarity to ATPase domains of SpoIIIE/FtsK-family proteins, and as such, was termed sffA (Figs S2 and S3).
To confirm the expression pattern revealed by the microarray experiment, transcription of smeA was investigated using S1 nuclease protection assays on independent RNA preparations. The largest protected fragments corresponded to a transcription start site ∼40–50 bp upstream of the smeA start codon, and transcripts were seen only in samples from time points at which visible signs of sporulation were detected (Fig. 1A), confirming that transcription of smeA is activated during sporulation in S. coelicolor. No smeA transcripts were detected in samples from the whiG, whiA and whiI mutants, and only very faint bands were seen in the whiH mutant (Fig. 1A and data not shown), indicating that smeA expression was completely dependent on the products of whiG, whiA and whiI, and strongly affected by the whiH gene product. The smeA mRNA 5′ ends were also amplified and cloned using the rapid amplification of 5′ cDNA ends (5′RACE) method, from RNA prepared independently of a sporulating wild-type culture. Of 10 sequenced inserts, six showed an mRNA start point at 28 or 29 bp upstream of the predicted start codon for smeA, two started at 38 bp and two at 46 bp from the start codon (Fig. 1A, bottom). This corresponded very well with the sizes of the fragments detected in the S1 nuclease protection assay. However, it is unclear whether the multiple 5′ ends represent multiple transcription start sites or could be generated by mRNA processing.
To further study the putative sporulation-related roles of smeA and sffA, we deleted both genes from the genome of the wild-type strain M145, which normally produces dark grey spores on MS agar plates. A ΔsmeA-sffA strain (K151, Table 1) produced uniformly light grey colonies, indicating that indeed, smeA and sffA were important for normal sporulation (Fig. 1B and C). Introduction of smeA-sffA (pNA276, Table 1) in trans restored the spore pigmentation of the mutant to the wild-type level, confirming that the observed developmental defect was indeed caused by smeA-sffA deletion (Fig. 1B and C). In order to clarify the individual contributions of smeA and sffA to the defective sporulation phenotype, we constructed an in-frame ΔsmeA deletion mutant (K150, Table 1; Fig. 1B) and a ΔsffA mutant (K152, Table 1; Fig. 1B). The smeA deletion strain had a light grey spore phenotype, similar to that of the ΔsmeA-sffA strain, while deletion of sffA alone resulted in a very slight reduction in spore pigmentation, suggesting that the lack of SmeA was primarily responsible for the light grey phenotypes of both the ΔsmeA and ΔsmeA-sffA mutants. However, smeA alone in trans was only able to partially complement the ΔsmeA strain (where sffA was still present) and not the ΔsmeA-sffA strain, indicating that both SmeA and SffA are needed for optimal sporulation (Fig. 1C). Full complementation was only obtained when smeA and sffA were expressed in cis, suggesting that coexpression was necessary for the optimal function of both SmeA and SffA. Together, these results strongly suggest that smeA-sffA constitutes a novel sporulation locus in S. coelicolor.
SmeA and SffA influence multiple developmental processes during later stages of sporulation
Because reduced spore pigmentation is often indicative of defects in underlying developmental processes leading to spore maturation, we studied the ΔsmeA-sffA strain by phase-contrast, fluorescence and electron microscopy to gain better understanding of the developmental roles of SmeA and SffA. Growth of the vegetative mycelium was not affected by deletion of smeA-sffA; however, several differences between the wild-type and the ΔsmeA-sffA strains were observed in sporulating aerial hyphae. First, fluorescence microscopy of fixed and DAPI-stained sporulating hyphae revealed that the chromosomes in ΔsmeA-sff spores displayed a more diffuse staining pattern than chromosomes in wild-type spores (Fig. 2A). The lower panel in Fig. 2A shows the distribution of fluorescence signal intensity along a line through the middle of five consecutive spore compartments. Peaks representing wild-type chromosomes were sharp and declined almost to a background level at positions corresponding to constrictions between spore compartments, whereas peaks of the mutant strain were wider and flatter, indicating the inability of the ΔsmeA-sffA strain to condense its chromosomal DNA to the wild-type level during sporulation. Furthermore, DAPI staining also revealed that smeA-sffA deletion caused a slight defect in chromosome segregation. Anucleate spore compartments within spore chains occurred with a frequency of 0.6% in the ΔsmeA-sffA strain, compared with the wild-type frequency of < 0.1% (> 10 000 spore compartments were counted for both wild-type and ΔsmeA-sffA strains).
Transmission electron microscopy (TEM) revealed architectural defects in the spore envelope of the ΔsmeA-sffA strain (Fig. 2B). The cell walls of the mutant spores were thinner and showed less defined layers than those of the wild-type spores, which had clearly discernible layers of light and dense material, as described previously (McVittie, 1974). In particular, the pronounced electron-dense outer layer of the wild-type spores appeared considerably thinner and ‘paler’ in the mutant (Fig. 2B). Reduced spore wall thickness is often correlated with decreased spore resistance to detergents and high temperatures (Potuckova et al., 1995; Molle et al., 2000; Mazza et al., 2006), and correspondingly, we observed that mutant spores were more susceptible to heat than wild-type spores. For example, 10.3% of the wild-type spores survived treatment at 65°C for 20 min, whereas the corresponding figure for mutant spores was 2.4% (after 30 min the corresponding values were 3.3% and 0.4% respectively). However, mutant and wild-type spores appeared to be equally resistant to detergents (up to 5% SDS, data not shown). Intriguingly, SmeA and Sff also appeared to be required for spore separation, as scanning electron microscopy (SEM) images of 8-day-old colonies grown on MS agar revealed that, while free spores were readily apparent on the surface of the wild-type strain, spores from the ΔsmeA-sff strain appeared almost exclusively in chains (Fig. 2C). Furthermore, the uneven compartment size of developing spore chains of the ΔsmeA-sffA strain, seen clearly in Fig. 2C, was also observed by phase-contrast microscopy and by TEM (data not shown).
The phenotype of the ΔsmeA strain was indistinguishable from that of the ΔsmeA-sffA double mutant strain at the level of microscopic observation. In contrast, the sffA deletion strain showed apparently normal spore maturation and spore wall structure, except that phase-contrast and electron microscopy revealed an uneven septation defect similar to what is characteristic of the ΔsmeA-sffA strain (example in a young ΔsffA spore chain is shown in Fig. 2B). Taken together, these observations show that the products of the smeA-sffA operon influence several processes of normal formation of prespore compartments and spore maturation, including division septum placement, chromosome segregation and condensation, spore separation, spore pigmentation, maturation of the spore envelope and development of heat resistance.
The smeAp promoter is induced specifically in sporogenic aerial hyphal cells independently of the sporulation sigma factor gene sigF
Interestingly, most features of the pleiotropic smeA-sffA phenotype overlap with the phenotype associated with the deletion of the RNA polymerase sigma factor gene sigF, which is also known to affect late stages of sporulation (Potuckova et al., 1995). sigF is specifically induced in sporulating hyphae (Sun et al., 1999) and like smeA-sffA, its transcription depends on the early whi genes (whiG, whiA, whiB, whiH, whiI and whiJ) (Kelemen et al., 1996). Thus it was possible that smeA-sffA transcription was σF-dependent. To test this hypothesis, we fused the promoter region, ribosome binding site and start codon of smeA to the gene for the red fluorescent protein mCherry (smeAp–mCherry, pNA473, Table 1), and used this translational fusion to monitor smeA expression. In the wild-type, a strong fluorescence signal was observed in sporulating aerial hyphae with constrictions or sporulation septa, and a weaker signal was seen in some unconstricted aerial hyphae, but no signal was detected in the vegetative mycelium (Fig. 3A). This showed that there was a specific compartmentalized activation of the smeA promoter in the apical cell of aerial hyphae undergoing sporulation. A similar pattern of expression of smeAp–mCherry was also seen in the ΔsmeA-sffA strain, indicating that SmeA and SffA are not involved in regulating their own expression (data not shown). A wild-type developmental pattern of smeAp–mCherry expression was also observed in the sigF deletion strain (J1984, Table 1), showing that σF is not required for the sporulation-specific expression of smeA-sffA (Fig. 3A). Conversely, we also tested whether sigF expression depended on smeA-sffA. A sigFp–mCherry translational reporter fusion (pNA540, Table 1) was introduced into the wild-type (M145), ΔsigF and ΔsmeA-sffA strains. Strong sporulation-specific expression of mCherry fluorescence was observed in all three strains, showing that sigF expression did not depend on SmeA-SffA or σF itself (Fig. 3A and data not shown). Thus, sigF and smeA-sffA are expressed independently of each other in sporulating aerial hyphae.
Activation of smeA-sffA expression occurs after Z-ring formation, but is not dependent on regular septation
Expression of both sigF and smeA-sffA depends on the early whi genes. As the early whi genes are also needed for the formation of sporulation septa, and ftsZ mutants that are specifically defective in the formation of sporulation septa have pleiotropic sporulation defects including a strong reduction in the production of the grey spore pigment (Flärdh et al., 2000; Grantcharova et al., 2003), it has been speculated that septum formation might represent a morphological checkpoint that triggers the expression of late sporulation genes whose products are needed after compartmentalization of the aerial hyphae (Flärdh et al., 2000). To test whether smeA-sffA was subject to this putative cell-division checkpoint, we first needed to determine the timing of smeA-sffA expression in relation to Z-ring formation. For this purpose, the smeAp–mCherry reporter was introduced into strain K127, which produces both wild-type FtsZ and FtsZ–EGFP. Most sporulating hyphae displayed both FtsZ–EGFP and smeAp–mCherry fluorescence. However, some unconstricted aerial hyphae had clearly visible Z-rings but no detectable mCherry fluorescence (Fig. 3B), but we never detected smeAp–mCherry expression without an FtsZ–EGFP signal. These results suggested that smeA expression occurred no earlier than, and possibly after, FtsZ-ring formation. The same experiment was conducted with the sigFp–mCherry reporter, which allowed comparison with this previously known late sporulation gene (Kelemen et al., 1996). The timing of sigF expression was indistinguishable from that of smeA-sffA (Fig. 3C), indicating that both smeA-sffA and sigF gene expression could in principle be subject to septum checkpoint regulation. We next introduced the smeAp–mCherry promoter probe into a strain carrying the mutant allele ftsZ17(Spo) (K102, Table 1), which is severely impaired in the formation of regular Z-rings and sporulation septa in the aerial mycelium (Grantcharova et al., 2003). If expression of smeA-sffA required normal FtsZ-ring formation and septation, we would expect to see reduced smeAp–mCherry fluorescence in the aerial hyphae in this mutant background. Interestingly, many aerial hyphae showed smeAp–mCherry fluorescence with intensity comparable to the wild-type situation (Fig. 3D). Furthermore, it has been shown that sporulation septa form with low frequency in the ftsZ17(Spo) strain, causing separation of some aerial hyphal fragments (Grantcharova et al., 2003). Such separated fragments (inset in Fig. 3D) showed not only intense mCherry fluorescence, but were also thick and phase-dark, reminiscent of the spore-like aerial hyphal fragments formed by whiH and ftsZΔ2p mutants (Flärdh et al., 1999; 2000; Grantcharova et al., 2003). Thus, there does not seem to be a quantitative dependence of smeA expression on the formation of multiple Z-rings and sporulation septa in aerial hyphae. However, it is still possible that the dependence was qualitative, and formation of one, or a few, sporulation septa per aerial hypha suffices to trigger smeA-sffA expression.
SffA localizes to late sporulation septa in a SmeA-dependent manner
The homology of SffA to FtsK-like DNA translocases and the increased frequency of anucleate spores in the ΔsmeA-sffA mutant suggested a possible involvement of this locus in segregation of chromosomes. It was therefore important to determine whether SmeA and SffA were targeted to the sporulation septa or to other specific sites. Unfortunately, all attempts to create a fully functional tagged derivative of SmeA were unsuccessful. Both N-terminal and C-terminal fusions to mCherry or EGFP, respectively, were non-functional, as was SmeA carrying a short C-terminal FLAG tag. However, insertion of mCherry internally, immediately following a putative signal sequence cleavage site, produced a partially functional fusion protein as indicated by a slight restoration of spore pigmentation on MS agar plates (pNA547, Table 1). This fusion protein (mCh–SmeA) displayed a developmental expression pattern similar to the smeAp–mCherry reporter (Fig. 3A) and the SffA–EGFP protein (see below and Fig. 4) and was evenly distributed in the cell periphery (Fig. 4A), thus supporting the predicted membrane localization of SmeA. However, as the fusion protein was only weakly active, we cannot exclude that the native SmeA protein might have a more distinct cellular localization pattern, such as, for example, colocalization with SffA to the sporulation septa.
Replacement of sffA by sffA–egfp in the chromosome of the wild-type strain M145 (K158, Table 1) did not cause the characteristic uneven septation phenotype of the ΔsffA strain, indicating that the SffA–EGFP fusion protein was at least partially functional. In agreement with the results seen with the smeAp–mCherry reporter, no SffA–EGFP fluorescence was detected during vegetative growth. Weak SffA–EGFP fluorescence appeared in some early non-constricted sporogenic hyphae (Fig. 4B). In later stages, when visible constrictions had appeared between spore compartments, SffA–EGFP had re-localized and accumulated into strongly fluorescent foci in the middle of the constricting septa (Fig. 4C and D). Thus, SffA has a cell type-specific expression and septal localization in S. coelicolor, which would be consistent with a hypothetical role as a translocase protein suggested by its homology to FtsK/SpoIIIE proteins. In addition, SffA–EGFP was also seen at some vegetative-type septa in the subapical segment of aerial hyphae (data not shown), which otherwise showed no SffA–GFP upregulation.
To investigate the cellular role of SmeA and the putative functional relationship with SffA, we studied the localization of SffA–EGFP in a ΔsmeA background. sffA was replaced by sffA–egfp in the chromosome of the ΔsmeA strain (K159, Table 1), and as expected, the ΔsmeA sffA–egfp strain was phenotypically similar to the ΔsmeA strain and showed normal activation of sffA–egfp expression in sporogenic hyphae (Fig. 4E). Strikingly, however, SffA–EGFP did not relocalize to the constricting sporulation septa of the ΔsmeA deletion strain (Fig. 4F and G). In contrast, SffA–EGFP localization to the vegetative-type septa (arrows in Fig. 4E and F) in the non-sporulating part of the hyphae occurred both in the smeA and smeA+ strains. Introduction of smeA-sffA in trans fully restored the dynamic localization pattern of SffA–EGFP (Fig. 4H) and the dark grey spore colour to the smeA sffA–egfp strain. Thus, SmeA was specifically required for SffA accumulation at the constricting sporulation septa, but the localization of SffA–EGFP to vegetative-type septa in basal parts of sporulating aerial hyphae was SmeA-independent. It is important to note that while both the ΔsmeA (SffA mislocalized) and the ΔsmeA-sffA (SffA missing) strains had very similar phenotypes, the ΔsffA strain had a less severe sporulation defect, suggesting that the mislocalization of SffA was not solely responsible for the developmental phenotype of the smeA strains. Consequently, SmeA must fulfil other functions in addition to its role as a localization determinant for the DNA-translocase-domain protein SffA.
Sporulation septa contain two SpoIIIE/FtsK-family proteins with distinct functions
The septal localization of SffA resembled that of FtsK and SpoIIIE proteins from other bacteria, which localize to the middle of cell division or sporulation septa and are involved in active translocation of DNA across the closing septum as well as in resolution of chromosomes (Bath et al., 2000; Aussel et al., 2002). A recent report established that the SCO5750 protein has an FtsK-like function in S. coelicolor, affecting segregation and/or stability of chromosomes (Wang et al., 2007). Therefore, it was important to establish whether SffA and FtsK have overlapping or perhaps redundant functions. To this end, we constructed deletion strains where ftsK was replaced by an apramycin resistance cassette both in the wild-type (M145) and in the ΔsmeA-sffA backgrounds, yielding ΔftsK and ΔsmeA-sffAΔftsK mutants (K161 and K169, Table 1). In contrast to Escherichia coli and Bacillus subtilis, in which FtsK and SpoIIIE have essential functions for viability and sporulation, respectively, the S. coelicolorΔftsK strain grew as well as the wild-type parent and sporulated abundantly, but showed an increased occurrence of aberrant colonies upon restreaking, confirming previous observations (Wang et al., 2007). However, fluorescence microscopy of DAPI-stained spore chains showed a subtle but revealing difference between strains with and without ftsK. In the wild-type strain, 88% (14 out of 16) of the occasional anucleate spore compartments were flanked by a spore compartment containing approximately twice as much DNA as an average spore. Similarly, in the ΔsmeA-sffA strain, which has a higher occurrence of anucleate spores, 87% of those (27 out of 31) were flanked by a spore with two chromosomes (example shown in Fig. 5A). Only situations where an anucleate spore was flanked by another spore on both sides were considered. In the ΔftsK and ΔsmeA-sffAΔftsK strains, however, spores with severely reduced DNA content were readily detected, but spores that were completely devoid of DNA were not observed, even though more than 10 000 spores were examined for each strain. In clear contrast to the situation in the ftsK+ counterparts, in either the ΔftsK strain or the ΔsmeA-sffAΔftsK strain none of 43 observed spores with strongly reduced DNA content were flanked by spores with double DNA content (Fig. 5A). Our interpretation of these results is that the cross-septum DNA translocase activity of FtsK normally contributes to the formation of unigenomic spores by removing chromosomal DNA that becomes trapped under a closing septum to the proper adjacent spore compartment. However, on occasions when septum placement or chromosome positioning is impaired, the trapped chromosome may be pumped in the ‘wrong’ direction resulting in an empty spore flanked by a spore with two chromosomes. These occasions are rare in the wild type, but become more frequent when chromosome positioning is disturbed. In support of this hypothesis, the ΔsmeA-sffA strain, which showed chromosome segregation and septum placement defects, showed proportionally more occasions of zero/double chromosome pairs. This resembles the function of B. subtilis SpoIIIE in moving trapped chromosomes out of closing cell division septa (Sharpe and Errington, 1995). As most empty spores of the ΔsmeA-sffA strain were also flanked by spores with double chromosomes, it seems likely that the clearing of DNA from closing septa is dependent on FtsK activity and not that of SffA.
We also replaced ftsK by ftsK–mCherry in the wild-type and ΔsmeA-sffA background. Like SffA–EGFP, FtsK–mCherry localized to the sporulation septa in the wild-type background (Fig. 5B), confirming the FtsK localization reported by Wang et al. (2007). The same localization pattern was observed in the ΔsmeA-sffA background (Fig. 5B), indicating that, in contrast to SffA, the localization of FtsK was independent of SmeA. Thus, we have shown that the SffA and FtsK putative translocase proteins both localize to sporulation septa, but their localization mechanisms and their cellular functions are different. FtsK appears to be involved in DNA translocation between spore compartments, but is not directly involved in regulating sporulation. SffA, on the other hand, appears to be functionally related to SmeA, which has a pleiotropic role in spore maturation.
In this article we have identified and characterized two dedicated sporulation genes whose expression is specifically activated in sporogenic aerial hyphae. In contrast to several previously characterized sporulation-specific genes (whi genes) which encode regulators of gene expression, the gene products of smeA and sffA are not likely to be directly involved in transcriptional control. Nevertheless, the small membrane protein SmeA was shown to influence several processes during spore development, including septum placement, DNA segregation and condensation, spore wall thickening and spore separation. Although the deletion of sffA caused a subtle phenotype, its specific activation in sporulating aerial hyphae, as well as its genetic and functional coupling to SmeA, suggested a role for SffA in sporulation as well. Understanding the functions of SmeA and SffA in more detail will help to shed new light on the molecular mechanisms underlying sporulation.
smeA-sffA organization is conserved
blast and psi-blast searches revealed that the sequenced Streptomyces species S. coelicolor, S. avermitilis and S. scabies contain four, seven and six smeA-like genes, respectively, and several self-transmissible plasmids from different Streptomyces species harboured single copies of smeA homologues (Fig. 6). Furthermore, two sequenced Frankia strains also encoded one SmeA-like protein each, but we did not find homologoues in other currently available actinomycete genomes. In total, we have identified 22 members of the Sme family, but we suspect that more distantly related proteins may not have been detected using blast, due to the short length of the query sequence (Fig. 6). All members of the SmeA family contain a transmembrane domain predicted with high probability by most algorithms. Furthermore, the SignalP algorithm recognized a putative N-terminal secretion signal in several, but not in all SmeA homologues (Nielsen et al., 1997). For example, the probabilities for the four S. coelicolor SmeA proteins to possess an N-terminal signal sequence vary between 0.684 and 0.995 according to the hidden Markov model prediction (Nielsen and Krogh, 1998). Each of the three sequenced Streptomyces species encodes one canonical SmeA-SffA pair in a conserved genetic environment. Strikingly, 12 of the 22 members of the sme family are adjacent to genes encoding SpoIIIE/FtsK-family proteins, including the single sme family genes of both Frankia species and all sme genes harboured on plasmids (indicated by asterisks in Fig. 6). Given that conservation of genetic linkage often indicates a functional relationship, and as the plasmid-encoded SpoIIIE/FtsK-family proteins (called Tra proteins) function as DNA translocases during conjugal plasmid transfer (Grohmann et al., 2003), the plasmid-encoded SmeA-like proteins may have a function associated with DNA transfer. Several of the chromosomally encoded Sme-Sff-like pairs appear to be part of integrated plasmids, based on neighbouring genes encoding typical plasmid functions; however, the chromosomal neighbourhood of smeA-sffA in S. coelicolor does not contain plasmid-related genes. Thus, an intriguing possibility is that SmeA and SffA function in some aspects of moving or organizing chromosomal DNA during sporulation.
What could be the cellular role of SmeA?
One consequence of SmeA deletion is the inability to correctly localize SffA to the sporulation septa. However, the complex phenotype of smeA deletion could not be explained solely by mislocalization of SffA. Perhaps SmeA is needed for proper localization or membrane assembly of sporulation-specific proteins in addition to SffA. Such a function would be reminiscent of the roles of several small membrane proteins in other bacteria. In E. coli, for example, CcmD (69 aa) has been proposed to function as an assembly factor for the formation of a membrane-bound multiprotein complex functioning in the haem delivery process (Ahuja and Thony-Meyer, 2005), while in B. subtilis, the membrane-associated protein SpoVM (29 aa) tethers the spore coat to the outer spore membrane (Ramamurthi et al., 2006).
Are SmeA and SffA involved in DNA translocation?
In addition to the effects of ΔsmeA-sffA and ΔsffA mutations on the accuracy of chromosome segregation in developing spore chains, several observations support a role for SffA as a DNA translocase. SffA shares the typical domain architecture of other FtsK-like proteins, with an N-terminal membrane-spanning region, and a C-terminal P-loop ATPase domain that is clearly, although distantly, related to the translocase domains of canonical FtsK proteins (Figs S2 and S3) (Rost et al., 2004). When the sequence of SffA was compared with the consensus sequence of the large FtsK-HerA superfamily of pumping ATPases, SffA contained all conserved, distinguishing motifs of this family (Iyer et al., 2004). Most of the characterized proteins of this family are involved in pumping or packaging DNA (Iyer et al., 2004), with the only known exception being the YukA-like proteins containing three tandem ATPase domains, which are implicated in the secretion of extracellular peptides (Pallen, 2002). Another feature shared between SffA and the FtsK proteins is its specific localization to cell division septa. While we have shown that both SffA and FtsK localize to sporulation septa in S. coelicolor, experimental evidence only supported the role of FtsK as a DNA pump. Furthermore, the SffA amino acid sequence showed some curious deviations from the FtsK-HerA superfamily. For example, the P-loop sequence of the Walker A motif of SffA (consensus Gx4GK[TS]) contains a threonine in place of the conserved lysine residue, and a conserved glutamic acid residue in the Walker B motif is replaced by an arginine in SffA. The family of SpoIIIE/FtsK-domain proteins is large and its members exist in most bacteria and archaea, and yet the exact function is known only for a handful of members. Therefore, the similarities and differences in the sequence and function of the distantly related SffA and FtsK proteins make SffA an attractive target for further functional and structural studies.
Compartmentalized expression of smeA-sffA does not require multiple sporulation septa and occurs also in poorly septated whiH and ftsZ mutants
A central but so far unresolved question is what mechanisms are involved in triggering developmentally induced gene expression specifically in the sporulating aerial hyphal cells. In analogy to the role of the asymmetric septum in B. subtilis sporulation (Barak and Wilkinson, 2005), it has been suggested that sporulation septation may act as a morphological checkpoint to which the progression of the developmental programme would be linked (Chater, 2000; Flärdh et al., 2000). In support of such a model, developmental activation of several late sporulation promoters (sigFp, whiE1p, whiE2p and parAB1p) is abolished in mutants that completely lack sporulation septa (such as whiG, whiA and whiB), but can be detected, although at significantly reduced levels, in whiH mutants that are still capable of forming some sporulation septa (in addition, parAB was upregulated in an ftsZΔ2p mutant that also forms only occasional sporulation septa) (Kelemen et al., 1996; 1998; Flärdh et al., 1999; Jakimowicz et al., 2006). We found that smeA-sffA expression also follows the same pattern, suggesting a quantitative dependence of its expression on the number of sporulation septa. Using smeAp–mCherry and sigFp–mCherry constructs in a strain also containing FtsZ–EGFP, we showed that the developmental activation of both late sporulation genes was restricted to the apical sporogenic cell of aerial hyphae, and most probably happened after the assembly of Z-rings (Fig. 3B and C). Furthermore, by monitoring the smeAp–mCherry fluorescence in the ftsZ17(Spo) mutant (which produces only a few sporulation septa and is similar to the whiH and ftsZΔ2p mutants), we were able to show that smeA upregulation in fact did not require the formation of multiple Z-rings/septa per aerial hyphal cell. Based on these results, we propose that in the ftsZ17(Spo) and whiH mutants expression of late sporulation genes is activated in these mostly unseptated hyphal fragments, triggering spore maturation processes that give rise to their already documented spore-like appearance (Flärdh et al., 1999; 2000; Grantcharova et al., 2003). The strongly reduced abundance of transcripts corresponding to smeA and possibly other late sporulation genes in the whiH mutant may then be explained by the lower abundance of such spore-like hyphal compartments in this mutant compared with the number of developing spore chains in wild-type strains (K. Flärdh, unpubl. obs.).
It is obvious from these and other recent studies that the improvement of cytological tools for analysis of cell type-specific gene expression will be essential for further progress in understanding the control of morphological differentiation in S. coelicolor. The use of the red fluorescent protein mCherry (Shaner et al., 2004) enabled us to avoid the significant problem of the strong autofluorescence of Streptomyces hyphae in the spectral region used to study GFP. This often appears as relatively bright fluorescent spots within the hyphae and has limited the use of GFP fusions to cases with a strong signal or very specific pattern of subcellular localization. Thanks to the low and uniform autofluorescence in the red spectral region, we were able to reliably detect even relatively weak mCherry expression. The use of such cytological reporters and markers provides a productive avenue ahead for detailed dissection of the development and sporulation of the aerial mycelium in streptomycetes.
Bacterial strains and media
The S. coelicolor A3(2) and E. coli strains used in this work are listed in Table 1. E. coli strains DH5α (Hanahan, 1983) and TOP10 (Invitrogen) were used for cloning, and strain ET12567/pUZ8002 was used to drive conjugative transfer of non-methylated DNA from E. coli to S. coelicolor as described previously (Kieser et al., 2000). Strain GM2929 was used to prepare non-methylated plasmid DNA for direct transformation of S. coelicolor protoplasts. Strain EKF104 was constructed by integrating the cre expression cassette from p705-Cre (Zhang et al., 1998) into E. coli strain LE392. Cultivation of E. coli strains was performed as described in Sambrook et al. (1989). S. coelicolor strains were grown on mannitol soy flour agar plates (MS agar), in yeast extract-malt extract medium (YEME), in tryptone soy broth (TSB) or on R2YE agar (Kieser et al., 2000).
Construction of S. coelicolor mutants
The PCR-targeting procedure was used for generation of Streptomyces gene knockout mutants essentially as described in Gust et al. (2003). The primers used are listed in Table S1. The apramycin resistance cassette was obtained from plasmid pIJ773 (for replacement of sffA and ftsK) and the chloramphenicol resistance cassette from plasmid pLoxCat2 (for the in-frame deletions of smeA and smeA-sffA) (Palmeros et al., 2000). Target genes were first replaced by resistance cassettes on cosmids 6D7 (smeA, sffA) and 7C7 (ftsK) (Redenbach et al., 1996) in the E. coli strain DY380, which contains an inducible λ RED system (Yu et al., 2000). The LoxCat2 cassette that replaced smeA or smeA-sffA was subsequently excised by inducing the Cre recombinase in strain EKF104, leaving only a short sequence consisting of the loxP sequence and flanking restriction sites. Mutated cosmids were introduced into S. coelicolor wild-type or mutant strains by conjugation or protoplast transformation according to established protocols (Kieser et al., 2000) and screened for clones where a double recombination event had replaced the target gene by a respective mutant allele present on the cosmid. Mutants were verified by diagnostic PCR.
The plasmids used are listed in Table 1. DNA manipulation and cloning were carried out according to standard protocols (Sambrook et al., 1989). All primers used for cloning are listed in Table S1. Plasmid constructs were verified by DNA sequencing. Inserts for construction of plasmids pNA276 (pSmeA-SffA), pNA485 (pSmeA) and pNA277 (pSffA) were obtained by PCR using chromosomal DNA of strain M145, or mutated derivatives of cosmid 6D7 as templates, as shown in Fig. 1. Four hundred and thirty-five base pires of smeA upstream sequence and 317 bp of sffA downstream sequence were included in the clones. For EGFP and mCherry fusion constructs 996 bp of sffA and 1042 bp of ftsK encoding the C-terminal halves of the proteins were fused in frame to egfp or mCherry, respectively, on plasmid pEGFP-N2, creating plasmids pNA303 and pNA585, which were then used to transform S. coelicolor protoplasts. Transformants were selected where homologous recombination via a single crossing-over event had created a full-length recombinant fusion allele in the wild-type locus under control of the native promoter, as well as a second disrupted copy of gene. For insertion of mCherry (Shaner et al., 2004) internally into smeA, a KpnI site was introduced into smeA between regions encoding the putative signal sequence and the putative transmembrane helix by PCR-based mutagenesis of plasmid pNA276 (pSmeA-SffA). This site was then used for in-frame insertion of PCR-amplified mCherry lacking its start and stop codons, resulting in plasmid pNA547. Red fluorescent promoter probe constructs were constructed by cloning PCR-amplified promoter regions (660 bp for smeA and 494 bp for sigF) and including respective ribosome binding sites and start codons of smeA and sigF in frame with the mCherry ORF in plasmid pSET152, resulting plasmids pNA473 and pNA540.
Samples for light and fluorescence microscopy were obtained either by pressing a coverslip on developing colonies on MS agar or by growing the strains in the angle between an inserted coverslip and the MS agar surface (Kieser et al., 2000). GFP and mCherry fluorescence was observed directly after mounting these coverslips on a glass slide with 50% glycerol in phosphate-buffered saline. For visualization of nucleoids, samples were first methanol-fixed and stained with 4′,6-diamidino-2-phenylindole (DAPI) as described previously (Flärdh et al., 1999). All fluorescence and phase-contrast microscopy was performed using an Axioplan II imaging fluorescence microscope equipped with appropriate filter sets, an Axiocam charge-coupled device camera and Axiovision software (Carl Zeiss Light Microscopy). Digital images were processed using Adobe Photoshop CS version 8.0 software.
For SEM the samples were pre-fixed with 2.5% glutaraldehyde, washed three times in phosphate-buffered saline and fixed in 1% osmium tetraoxide. After dehydration in ethanol the samples were injected through nucleopore filters with 0.2 μm pore size, critical point-dried, mounted on Cambridge alloy stubs, silver sputtered and examined in a Zeiss Supra 35-VP field emission SEM equipped with a STEM detector, EDAX Genesis 4000 EDS.
For TEM the samples were pre-fixed with 2.5% glutaraldehyde, washed in phosphate-buffered saline and fixed in 1% osmium tetraoxide. After dehydration in ethanol (20–100%) and infiltration in acetone-resin 1:1, the samples were infiltrated in pure resin and then embedded in silicon plates. Sections (60 nm thick) were made on a LKB-ultramicrotome with a diamond knife from Dupont, stained with lead citrate and uranyl acetate and examined in a Zeiss Supra 35-VP field emission SEM equipped with a STEM detector, EDAX Genesis 4000 EDS.
RNA preparation, S1 nuclease protection assay and 5′RACE
S1 nuclease protection assays were carried out on the same set of RNA preparations and with the same procedure as described previously (Flärdh et al., 2000). The probe was prepared by PCR using primers KF129 (aggcggcgagcatgacgagg) and 5′ radiolabelled KF130 (cgagcaggaccacgcctgag).
5′RACE was carried out using the 5′RACE kit, version 2.0 (Invitrogen), according to instructions from the manufacturer. The template RNA for this analysis was prepared from S. coelicolor strain M145 grown on a cellophane membrane on MS agar for 48 h, using RNeasy Protect Bacteria kit (Qiagen) according the protocol from The S. coelicolor Microarray Resource, University of Surrey (http://www.surrey.ac.uk/SBMS/Fgenomics/Microarrays/html_code/ExperimentalDesign.html). The amplified 5′RACE products were cloned using the TOPO TA cloning kit (Invitrogen) and 10 randomly chosen plasmids were investigated by DNA sequencing to determine the 5′ ends of the smeA transcript.
We are grateful to Francisco Bolivar, Helen Kieser, Francis Stewart and Roger Tsien for gifts of strains. Jakob Engman is acknowledged for assistance with the 5′RACE. This work was supported by Carl Trygger Foundation, a Swedish-UK joint research project from the Royal Swedish Academy of Science and grants from the Swedish Research Council (Grant No. 621-2004-4023 to N.A. and Grant No. 621-2004-4454 to K.F.).