Genetic analysis of the Bacillus subtilis sigG promoter, which controls the sporulation-specific transcription factor {sigma}G

Starvation induces Bacillus subtilis to initiate a simple, two-cell developmental process that results in the formation of dormant spores. Sporulation begins with an asymmetric cell division that gives rise to two distinct cells that have different developmental fates. The smaller cell (or prespore) is destined to become the spore, whilst the larger cell (the mother cell) engulfs and contributes to the development of the spore and eventually lyses to release it (reviewed by Errington, 2003; Stragier & Losick, 1996). Underlying the morphological changes associated with sporulation is a programme of differential gene expression in the two cell types, which is ultimately directed by the activation of four sporulation-specific σ-factors. The activity of each of these σ-factors is compartmentalized both spatially (i.e. between the two cell types) and temporally. The σ-factors are activated sequentially, beginning with σ^F in the prespore, then σ^E in the mother cell, followed by σ^G and σ^K in the prespore and mother cell, respectively. To ensure the correct sequence of morphological events each sigma factor is tightly regulated at both the transcriptional and post-translational levels. The regulatory mechanisms involve signalling pathways between the two cells (reviewed by Kroos & Yu, 2000; Rudner & Losick, 2001).

The late prespore-specific σ-factor, σ^G, is regulated at at least three levels. Firstly its gene sigG (spoIIIG) is transcribed by Eσ^F (and later by Eσ^G itself), thus restricting its localization to the prespore compartment (Fig. 1; Sun et al., 1991). The sigG gene is the distal element of the three-gene spoIIG operon, which comprises spoIIGA (encoding the pro-σ^E processing enzyme), sigE (spoIIGB) and sigG (Karmazyn-Campelli et al., 1989). Transcription of the whole operon is under the control of the housekeeping σ-factor, σ^A, and the sporulation-specific transcription factor Spo0A (Masuda et al., 1988), and begins before asymmetric septation. However, σ^G is not translated from this polycistronic transcript due to the presence of a stemloop structure that blocks its ribosome-binding site (Masuda et al., 1988). A promoter that is recognized by σ^F and σ^G itself is located immediately upstream of the sigG coding region, and it is from transcripts originating at this promoter that the σ^G protein is translated (Sun et al., 1991). Secondly, once translated, the protein apparently does not become active until after the completion of engulfment of the prespore. In the presence of mutations in several different genes, including spoIIB, spoIID, spoIIM, spoIIP, spoIIIA and spoIIIJ, σ^G is synthesized but it does not become active (Errington et al., 1992; Frandsen & Stragier, 1995; Kellner et al., 1996; Partridge & Errington, 1993; Smith et al., 1993). Four of the proteins, SpoIIB, SpoIID, SpoIIM and SpoIIP, are required for prespore engulfment (Frandsen & Stragier, 1995; Smith et al., 1993), suggesting that σ^G activity is coupled to this morphological event. Recently it has been shown that SpoIIIJ and the spoIIIA-encoded products are part of the signalling pathway that results in the activation of σ^G after completion of engulfment (Kellner et al., 1996; Serrano et al., 2003).

(11K): Fig. 1. Regulation of expression of the sigG gene. The sigG gene is only transcribed in the prespore compartment both before and after engulfment (only post-engulfment is shown). Transcription is directly regulated by σF and σG (indicated by a solid arrow) and indirectly regulated by σE (broken arrow).

A third level of regulation is suggested by the fact that transcription from the sigG promoter does not occur in the presence of mutations affecting synthesis or activation of σ^E (the early mother cell σ-factor) and/or an intact spoIIQ locus, encoding a prespore-specific membrane protein required for engulfment under certain sporulation conditions (Partridge & Errington, 1993; Sun et al., 2000). This suggests the existence of a signalling pathway, possibly involving SpoIIQ, that causes the synthesis of σ^G in the prespore to be dependent on the proper occurrence of events in the mother cell (represented by the broken arrow in Fig. 1).

Here we have investigated the σ^E-dependence of sigG expression; we have identified a regulatory site within the sigG promoter by analysing mutant sigG promoters generated by site-directed mutagenesis and sigG promoters from other species. Replacement of the wild-type promoter with σ^E-independent promoters resulted in impairment of sporulation. Our data support the idea that σ^E activity is required for the transcription of sigG, probably by relieving the action of a repressor.

Bacterial strains and plasmids.
Bacterial strains and plasmids used in this study are shown in Table 1. All B. subtilis strains are isogenic with SG38.

Table 1. Strains and plasmids used in this study

General methods.
DNA manipulations and Escherichia coli transformations were carried out using standard methods (Sambrook et al., 1989).

B. subtilis strains were transformed as described previously (Anagnostopoulos & Spizizen, 1961; Jenkinson, 1983). Transformants were selected on Oxoid nutrient agar plates containing chloramphenicol (5 µg ml¹) or kanamycin (5 µg ml¹) as appropriate.

Resuspension and β-galactosidase assay.
B. subtilis cells were induced to sporulate by the resuspension method (Sterlini & Mandelstam, 1969; Partridge & Errington, 1993). Time zero (t₀) was defined as the point at which the cells were resuspended in a starvation medium (SM). Samples (0·5 ml) were removed at intervals, pelleted and frozen in liquid nitrogen to be assayed for β-galactosidase activity.

β-Galactosidase activity was assayed using a method described by Errington & Mandelstam (1986). One unit of β-galactosidase catalyses the production of 1 nmol 4-methylumbelliferone min¹.

Construction of lacZ fusions to different lengths of the sigG promoter.
Plasmids pSG4742, pSG4731 and pSG4743 were constructed by amplifying sigG promoter fragments of, respectively, 338 bp, 143 bp and 82 bp by PCR from chromosomal DNA of wild-type strain SG38. The following primers were used for PCR: 4742 (5'-CCGGAATTCAAAAGCGCTTGA-3'), 4731 (5'-CCGGAATTCATGGTTAGAACC-3') or 4743 (5'-CCGGAATTCGCAGTGCATATT-3'), each of which introduced an EcoRI site, and H1 (5'-CGCCAAGCTTATTTCTCGACAC-3'), which introduced a HindIII site. The EcoRIHindIII-digested PCR products were subcloned into EcoRIHindIII-digested and gel-purified ptrpBGI (Shimotsu & Henner, 1986), thereby replacing the trp promoter with the sigG promoter and generating translational sigGlacZ fusions.

Site-directed mutagenesis.
Mutations were introduced into the sigG promoter by PCR amplification of plasmid pSG4732. Forward and reverse primers were designed to overlap completely and had the mutated base(s) in the middle. Pfu DNA polymerase was used to amplify the whole plasmid. Template DNA was degraded by DpnI digestion; the nicked circular products were then transformed into E. coli and the promoter then replaced into pSG4731. Each mutant promoter was sequenced before introduction into the B. subtilis chromosome at the amyE locus.

Construction of strains with sigG under the control of σ^E-independent promoters.
The P_Bt and P_sigG-14 mutant promoters were cloned in place of the wild-type promoter at the sigG locus. The resulting arrangement of genes is such that sigE and sigG are still in tandem with the aphA-3 gene inserted between them, in the opposite orientation to prevent read-through.

pSG4739.
pSG4738 carries the 3' coding region of sigE up to 10 bp downstream of the stop codon. A 500 bp segment of the sigG gene from 11 bp downstream of the sigE stop codon was amplified by PCR [using primers Eco(IIIG) (5'-CGGAATTCTGGTTAGAACCCCTTGATTTTAC-3') and SigGSphIrev (5'-AAACATGCATGCGTAAGCGATGTCCCGG-3'] from SG38 chromosomal DNA and inserted into pSG4738 along with the aphA-3 cassette generated by BamHI/EcoRI digestion of vector pAM1.

pSG4740 and pSG4741.
The Bacillus thuringiensis promoter was amplified from B. thuringiensis chromosomal DNA by PCR using primers 5'-CGAATTCGTAGGCTGGTCTTATTC-3' and 5'-GGACTAGTTTCCCTCCTATCGGGAGTTGC-3' and digested with EcoRI and SpeI. The coding region of sigG was amplified by PCR of SG38 chromosomal DNA using primers 5'-GACTAGTGTCGAGAAATAAAGTCGAAATC-3' and 5'-AAACATGCATGCGTAAGCGATGTCCCGG-3' and digested with SpeI and SphI. The two PCR products and the aphA-3 fragment from EcoRI and BamHI digestion of pAM1 were ligated all at once into pSG4738 cut with SphI and BamHI, resulting in plasmid pSG4740. The P_sigG-14 mutation was introduced into pSG4739 using the same method for mutagenesis of pSG4731 described above.

Each plasmid was sequenced, then B. subtilis SG38 was transformed with the resulting plasmids, giving strains 2806 (wild-type promoter), 2807 (P_Bt promoter) and 2808 (P_sigG-14 promoter).

Expression of σ^F in vegetative growth.
An IPTG-inducible copy of spoIIAC, encoding σ^F, carried on plasmid pSG635 was integrated by single crossover into the spoIIA operon of strains 2803-P_sigG, 2803-P_sigG-14, 2803-P_Bco-sigG and 2803-P_Bt-sigG. The resulting strains (2821, 2829, 2822 and 2823 respectively) express spoIIAA and spoIIAB from their wild-type promoter whereas spoIIAC is controlled by the IPTG-inducible P_spac promoter. Cells were grown at 37 °C to OD₆₀₀ 0·25 in CH medium (Sterlini & Mandelstam, 1969); at this point the culture was split into two and σ^F expression induced in one half by the addition of 1 mM IPTG. Samples (0·5 ml) were taken over 3 h, pelleted, frozen in liquid nitrogen and assayed for β-galactosidase activity as described above.

Definition of the minimal sigG promoter required for dependence on σ^E
Partridge & Errington (1993) reported that mutations in the spoIIG operon, encoding the pro-σ^E processing enzyme and pro-σ^E, blocked transcription of sigG but not that of another σ^F-dependent gene gpr, suggesting the existence of an unknown regulatory protein. In E. coli the classical position for repressor binding sites lies between the 35 and 10 elements, whereas activator-binding sites are generally located further upstream (Collado-Vides et al., 1991). To determine whether the spoIIG effect was contained within or upstream of the Eσ^F recognition site, DNA sequences extending different distances upstream of the sigG promoter were fused to lacZ (Fig. 2a) and ectopically expressed from the amyE locus. The β-galactosidase activity of these fusions was measured during sporulation in the presence or absence of σ^E activity. As shown in Fig. 2(b), all three fusions still retained strong dependence on σ^E, suggesting that the regulatory site for σ^E control lies very close to the beginning of the sigG gene and within the 82 bp insert of pSG4743. The dependency on σ^E activity has also been observed in a strain containing a deletion of sigG (see below), although the overall values are reduced because σ^G contributes to the expression of its own gene (Sun et al., 1991).

(20K): Fig. 2. Characterization of the minimal sigG promoter still dependent on σE activity. (a) Construction of various translational sigGlacZ fusions. sigG promoter fragments of different lengths (open boxes) were subcloned into plasmid ptrpBGI, thereby fusing in-frame the first four amino acids of the sigG coding sequence (black box) to lacZ (dotted box). The 10 and 35 regions are indicated as hatched boxes. The restriction sites used for plasmid constructions were EcoRI (E) and HindIII (H) and are shown above the figure. The position of each EcoRI site is indicated in parentheses. (b) Plasmids pSG4742 (circles), pSG4731 (triangles) and pSG4743 (squares) were integrated into the amyE locus of strains SG38 (filled symbols) and 901 (open symbols). The strains were induced to sporulate and assayed at intervals for β-galactosidase activity.

Mutations between 10 and 35 of the sigG promoter can remove the dependence on σ^E
Attempts to demonstrate the titration of a putative regulatory protein by cloning the minimal sigG promoter region into a high-copy-number plasmid were unsuccessful. Instead, we used site-directed mutagenesis to search for base pairs within the 82 bp region that are required for σ^E-dependence. The 14 mutations constructed were designed to make the sequence more like that of the σ^E-independent gpr promoter (Partridge & Errington, 1993), but avoiding changes to the 10, 35 and RBS elements (Fig. 3). The mutated promoters were subcloned upstream of a promoterless lacZ gene at the amyE locus in isogenic strains containing wild-type or mutant alleles of sigE. The strains used were sigG, as σ^G contributes to the expression of its own structural gene (Sun et al., 1991) and also lacA, to avoid expression of the B. subtilis endogenous β-galactosidase (Daniel et al., 1997). The generated mutations should not alter the stability of the stemloop structure that overlaps the sigG promoter region (Masuda et al., 1988), as transcripts originating from the truncated promoter do not contain the entire sequence of the stemloop. As shown in Fig. 4(a) and by Partridge & Errington (1993), in the wild-type strain the main phase of transcription of a gprlacZ fusion begins around 150 min after induction of sporulation (dashed arrow). In the absence of σ^E, expression begins earlier (at around 100 min after initiation of sporulation; solid arrow) and is initially higher. This is a characteristic of σ^E-independent σ^F promoters, and partly a result of the disporic phenotype of σ^E mutants, in which two prespore compartments each with active σ^F are formed (Lewis et al., 1994).

(17K): Fig. 3. Nucleotide sequence of the minimal sigG promoter showing dependence on σE. The 35 and 10 promoter elements, transcriptional start site (+1), and RBS are indicated above the sequence. Arrows below the sequence indicate inverted repeats involved in formation of the stemloop structure. The first four amino acids of σG are shown below the sequence; the lacZ gene is fused in-frame with the fourth codon. The site-directed mutations created are numbered and shown above or below the sequence; + represents insertion and Δ deletion of a nucleotide. Mutations that had no effect ($) or exhibited a σE-independent (bold), increased (*) or decreased (#) promoter activity are indicated.

(11K): Fig. 4. Effects of sigG promoter mutations on expression. Strains containing lacZ fusions to gpr (a), the mutant promoters PsigG-3 (b) and PsigG-14 (c) were induced to sporulate by the resuspension method and β-galactosidase activity was measured. Squares indicate the wild-type and circles the mutant promoters. The sigE+ background (strain 2803) is represented by filled shapes, and the sigE background (strain 2804) by open shapes. Arrows point to the beginning of sigG transcription in the sigE+ (dashed) and sigE (solid) strain background. (d) An IPTG-inducible copy of σF was used to induce expression of lacZ fusions to PsigG (squares) and PsigG-14 (triangles). After induction with IPTG, samples were taken for β-galactosidase assay. Filled shapes indicate uninduced cultures and open shapes induced cultures.

The β-galactosidase activity was measured for all of the promoter mutations during sporulation. Several of the mutants showed increased or decreased activity but retained the dependence on spoIIG (Fig. 3 and data not shown). However, P_sigG-3, -11, -12, -13 and -14 appeared to have become σ^E-independent. Two typical experiments are shown in Fig. 4(b, c); transcription started earlier (solid arrow), and was initially higher in the sigE background, compared with the sigE⁺ background (dashed arrow). The promoter activity of these mutants did not reach wild-type levels in the sigE background, but the overall patterns were similar to those of the gprlacZ fusion (Fig. 4a; Partridge & Errington, 1993). Thus, transcription began earlier in the sigE background but the final levels of transcription were lower than in the sigE⁺ background.

It appeared that the mutant promoters were of similar overall strength to the wild-type. Nevertheless, to exclude the possibility that the mutant promoters were simply stronger than the wild-type an IPTG-inducible copy of σ^F was introduced into strains 2803-P_sigG and 2803-P_sigG-14. Expression of σ^F was induced during vegetative growth and β-galactosidase activity from the sigG promoters was measured in the absence of any sporulation-specific regulation. As shown in Fig. 4(d), both promoters were expressed similarly, indicating that the increased promoter activity seen in the sigE background is not simply due to increased promoter strength.

sigG promoters from most other spore-forming species are not dependent on σ^E in B. subtilis
As an alternative means of further analysing the sigG promoter and to examine how conserved this level of regulation is, we isolated and examined sigG promoters from other Bacillus species and from Clostridium acetobutylicum. By using forward and reverse primers specific to the 3' of sigE and the 5' of the sigG coding regions from B. subtilis respectively, the intergenic region was amplified by PCR from B. thuringiensis, B. licheniformis, B. coagulans and B. polymixa chromosomal DNA. The PCR products were sequenced, and new primers were designed to clone the promoters into pSG4731 in-frame with lacZ. The C. acetobutylicum sigEsigG intergenic region was amplified by PCR from plasmid pSE1 (Sauer et al., 1994; kindly provided by P. Dürre, University of Ulm) and inserted in place of the B. subtilis promoter between the EcoRI and HindIII sites of pSG4731. Each plasmid was then integrated into the amyE locus of strains 2803 and 2804.

Expression from the five foreign sigG promoters was measured during sporulation (Fig. 5). The timings of β-galactosidase production showed that all five promoters are indeed recognized by B. subtilis Eσ^F, but only the B. licheniformis promoter (Fig. 5b) was dependent on σ^E activity; for the other species, expression appeared to be independent of σ^E.

(23K): Fig. 5. Activity and σE-dependence of sigG promoters from other spore-forming bacteria. Isogenic strains containing a lacZ fusion to the sigG promoter from B. subtilis (squares) or from other bacteria (circles) were induced to sporulate and samples were taken for β-galactosidase activity. The wild-type (sigE+; filled symbols) and the sigE (open symbols) background were compared.

A sequence alignment of these and several other sigG promoters is shown in Fig. 6. The most striking observation is the presence of TTT immediately downstream from the 35 site in σ^E-dependent promoters (B. subtilis and B. licheniformis), compared with AAA or AAT in σ^E-independent promoters, suggesting that this site might be involved in the σ^E-dependent regulation of the sigG promoter. This idea is supported by the phenotype of P_sigG-14, in which conversion of TTT to AAA leads to earlier expression from the promoter in the absence of σ^E activity.

(34K): Fig. 6. Sequence alignment of the sigEsigG intergenic region from spore-forming species. The stop and start codons of sigE and sigG respectively, the 35 and 10 promoter elements and the RBS are labelled, and indicated in bold in the sequences. The positions of the σE-independent mutations in B. subtilis are indicated by arrows. Bsu, B. subtilis; Bl, B. licheniformis; Ban, B. anthracis (); Bt, B. thuringiensis; Bco, B. coagulans; Bpo, B. polymixa; Bha, B. halodurans (); Ca, C. acetobutylicum; Bst, B. stearothermophilus (). *This sequence is as yet incomplete.

Heterologous σ^E-independent sigG promoters are expressed to higher levels than σ^E-dependent ones
As shown in Fig. 5, the levels of expression from the σ^E-independent promoters from other species were much higher than those from the σ^E-dependent ones. One possible explanation for this would be that the σ^E-independent promoters are stronger. A second possibility would be that the effect of a specific negative regulator is never completely lifted from the σ^E-dependent promoters. To try to distinguish between these two possibilities, an IPTG-inducible copy of σ^F was introduced into the strains carrying the lacZ fusions to the B. subtilis, B. coagulans and B. thuringiensis sigG promoters. σ^F expression could therefore be induced in vegetative growth and β-galactosidase activity from the sigG promoters could be measured in the absence of any sporulation-specific regulation. As shown in Fig. 7 the relative levels of β-galactosidase activity were essentially proportional to those observed from the promoters during sporulation; i.e. the B. subtilis promoter was expressed much less than the other two. The data suggest that the differences in promoter activity observed during sporulation are due to different intrinsic promoter strengths.

(16K): Fig. 7. Expression of various sigG promoters during vegetative growth. An IPTG-inducible copy of σF was used to induce expression of sigGlacZ fusions of the B. subtilis (squares), B. coagulans (triangles), and B. thuringiensis (circles) promoters. After induction with IPTG, samples were taken for β-galactosidase assay. Filled shapes indicate uninduced cultures and open shapes induced cultures.

Uncoupling sigG expression from σ^E activity results in a sporulation defect
To determine the importance of σ^E-dependent expression of sigG during sporulation, strains were constructed in which sigG expression was driven either by the P_sigG-14 mutant or the B. thuringiensis P_Bt promoter, both of which are σ^E-independent promoters. The strain construction resulted in the insertion of an aphA-3 (Kan^R) cassette between sigE and sigG. Strain 2806 has sigG under the control of the B. subtilis wild-type promoter, and in strains 2807 and 2808 transcription is driven by the P_Bt and the P_sigG-14 promoter, respectively. The sporulation frequencies (mean percentage±SE of three independent experiments) of these strains were determined by phase-contrast microscopy, counting phase-bright spores in samples taken 7 h after induction of sporulation. Insertion of the resistance cassette in strain 2806 slightly reduced the sporulation frequency (59±2·5 %) compared with the wild-type strain, SG38 (76±2·1 %), for reasons that are not clear. However, in strains 2807 and 2808, where sigG transcription begins earlier, mean sporulation frequencies (43±6 % and 40±5 %, respectively) were reduced still further.

The reduction in sporulation frequency in strain 2808 compared to strain 2806 could not be due to different expression levels from the two promoters, because the level of transcription from the P_sigG-14 promoter was almost identical to that of the wild-type (Fig. 4d). Moreover, replacement of the wild-type promoter with either weak (P_sigG-14) or strong (P_Bt) σ^E-independent promoters decreased the sporulation frequency similarly.

Surprisingly, the reduction in sporulation frequency was not as severe as that observed when σ^F was transcribed prematurely (Arigoni et al., 1996; Feucht et al., 1999), suggesting multiple levels of control to ensure the timely synthesis and activation of σ^G. Therefore, we examined how the uncoupling of sigG transcription from σ^E activity affected the timing of σ^G activation. σ^G activity was measured in these strains by the introduction of the σ^G-dependent spoVAlacZ fusion. As shown in Fig. 8, strain 2806, where insertion of Kan^R separated spoIIG from spoIIIG, showed similar β-galactosidase activity to SG38. However, with the P_Bt (strain 2807) and P_sigG-14 promoter (strain 2808), expression of β-galactosidase was not earlier, as might be expected, but slightly delayed. A similar pattern was observed using an sspAlacZ fusion as an alternative σ^G-dependent reporter (data not shown). This suggests that when transcription from the sigG promoter is uncoupled from the activity of σ^E, the σ^G protein is held inactive until a later stage than normal.

(19K): Fig. 8. σG activity in strains 2806, 2807 and 2808. β-Galactosidase activity of a spoVAlacZ fusion introduced into strains SG38 (wild-type; squares), 2806 (wild-type with aphA-3 insertion; crosses), 2807 (PBt; circles) and 2808 (PsigG-14; triangles) was measured during sporulation induced by resuspension.

It has been shown previously that transcription of sigG in the prespore depends on an as-yet-unidentified signal transduction pathway requiring the action of σ^E in the mother cell (Partridge & Errington, 1993). Here, we have identified a cis-acting sequence in the sigG promoter region, which makes the promoter responsive to a signal coming from the mother cell, thereby delaying the transcription of the sigG promoter towards the end of the engulfment process (Figs 2 and 4). The location of the mutations between the 10 and 35 elements, which corresponds to a classical position for a repressor-binding site (Fig. 3; Collado-Vides et al., 1991), and the finding that Eσ^F is able to transcribe a sigG template in vitro without the requirement for other proteins (Sun et al., 1991), would favour negative regulation.

Preliminary data suggest that SpoVT (Bagyan et al., 1996), a transcriptional regulator of prespore-specific genes, is not required for σ^E-dependence of sigG transcription. In addition, attempts to identify the putative regulator by random mutagenesis of the B. subtilis chromosome have so far proved unsuccessful (L. Evans, unpublished).

Expressing σ^G early (i.e. from a σ^E-independent promoter) led to a slight but reproducible decrease in sporulation efficiency, emphasizing the importance of co-ordinating the developmental programmes of the two cells. The reduction in sporulation frequency was not as drastic as that observed when σ^F was activated prematurely (Arigoni et al., 1996; Feucht et al., 1999), suggesting the existence of other levels of regulation for σ^G. Indeed, when the expression of two σ^G-dependent lacZ fusions was measured we found that when sigG was expressed from a σ^E-independent promoter, activation of σ^G was only slightly delayed (Fig. 8). Stragier & Losick (1996) reported a similar finding: premature expression of sigG from a strong σ^F-dependent promoter did not affect the timing of σ^G activity. Taken together, the results are in agreement with the notion that multiple levels of control act upon the synthesis and activation of σ^G (see Introduction).

Endospore formation by some Gram-positive bacteria belonging to the genera Bacillus and Clostridium seems to be a highly conserved process despite the triggers for the induction of sporulation being different. Comparison of the genomes of B. subtilis, B. anthracis, B. stearothermophilus, C. acetobutylicum and Clostridium difficile showed that not only the sporulation-specific σ-factors but also the so far known regulatory pathways leading to activation and coordination of their activity in the two compartments have been conserved (Stragier, 2002). This is supported by the finding that sigE, sigG and sigK of C. acetobutylicum have been found to be expressed in the same order as in B. subtilis (Santangelo et al., 1998). Also, the comparison of the promoters from several Bacillus species and C. acetobutylicum (Fig. 6) shows that there is a high degree of conservation in the 10 and 35 regions; thus they are all recognized and transcribed by B. subtilis Eσ^F (Fig. 5). The finding of such a high degree of similarity raises the interesting question of whether the genes are subject to the same σ^E-dependent regulation. However, only the B. licheniformis promoter was expressed in a σ^E-dependent manner (Fig. 5b). B. licheniformis is one of the species more closely related to B. subtilis, so it is possible that in more distant species promoter sequences may have diverged sufficiently for the B. subtilis regulatory protein no longer to recognize them. It is also possible that in these other species sigG expression is not dependent on σ^E, although this would seem less likely given the degree of conservation of the sporulation process across these species (Stragier, 2002).

It is unclear why the σ^E-independent promoters are expressed to such high levels compared with the B. subtilis promoter. The experiment where σ^F was induced in vegetative growth (Fig. 7) suggests that the heterologous promoters are intrinsically stronger. However, they do not appear to be any closer to the σ^F consensus sequence than the σ^E-dependent ones are.

A major challenge now is to identify the proteins that consitute the signal transduction pathway that couples the activation of σ^E in the mother cell with the transcription of sigG in the prespore.

This work was supported by grants from the Biotechnology and Biological Sciences Research Council (BBSRC) and Medical Research Council (MRC). L. E. was the recipient of a BBSRC postgraduate studentship.