Control of the Expression and Compartmentalization of σ^G Activity during Sporulation of Bacillus subtilis by Regulators of σ^F and σ^E

Abstract

During formation of spores by Bacillus subtilis the RNA polymerase factor σ^G ordinarily becomes active during spore formation exclusively in the prespore upon completion of engulfment of the prespore by the mother cell. Formation and activation of σ^G ordinarily requires prior activity of σ^F in the prespore and σ^E in the mother cell. Here we report that in spoIIA mutants lacking both σ^F and the anti-sigma factor SpoIIAB and in which σ^E is not active, σ^G nevertheless becomes active. Further, its activity is largely confined to the mother cell. Thus, there is a switch in the location of σ^G activity from prespore to mother cell. Factors contributing to the mother cell location are inferred to be read-through of spoIIIG, the structural gene for σ^G, from the upstream spoIIG locus and the absence of SpoIIAB, which can act in the mother cell as an anti-sigma factor to σ^G. When the spoIIIG locus was moved away from spoIIG to the distal amyE locus, σ^G became active earlier in sporulation in spoIIA deletion mutants, and the sporulation septum was not formed, suggesting that premature σ^G activation can block septum formation. We report a previously unrecognized control in which SpoIIGA can prevent the appearance of σ^G activity, and pro-σ^E (but not σ^E) can counteract this effect of SpoIIGA. We find that in strains lacking σ^F and SpoIIAB and engineered to produce active σ^E in the mother cell without the need for SpoIIGA, σ^G also becomes active in the mother cell.

Central to cell differentiation is the establishment of distinct programs of gene expression in the different cell types involved. These programs determine the subsequent path of differentiation. Among prokaryotes, formation of spores by Bacillus subtilis has become a paradigm for the analysis of cell differentiation. Soon after the start of spore formation, bacteria divide asymmetrically to give the smaller prespore (also called the forespore) and the larger mother cell. The prespore is then engulfed by the mother cell. The prespore develops into the mature spore, whereas the mother cell ultimately lyses. The process of spore formation is characterized by the cell-specific activation of four RNA polymerase σ factors. Immediately after the completion of the spore division septum, σ^F is activated in the prespore. Its activation leads rapidly to activation of σ^E in the mother cell. Upon completion of engulfment, σ^G becomes active in the prespore; its activation, in turn, leads to activation of σ^K in the mother cell (Fig. 1) (reviewed in reference 10). The activation of the successive σ factors is tightly coordinated within and between the two cell types, a process that has been termed crisscross regulation (19). We explore here the activation of σ^G in circumstances in which its normal tight coupling to the prior activation of σ^F and σ^E has broken down.

FIG. 1.

Schematic representation of stages of spore formation showing the normal location of activity of sporulation-specific sigma factors.

Both σ^F and σ^E are formed soon after the start of spore formation and before the sporulation division. When first formed they are inactive: σ^F because of interaction with the anti-sigma factor SpoIIAB and σ^E because it is formed as an inactive precursor, pro-σ^E. A complex regulatory system centered on SpoIIAB controls the activation of σ^F, which occurs in the prespore shortly after completion of the sporulation division. Activation of σ^E in the mother cell by processing of pro-σ^E depends on SpoIIGA, which is the putative processing enzyme, and on a σ^F-directed signal from the prespore. The appearance of σ^G activity depends on the activities of both σ^F and σ^E and on morphological signals (10).

The spoIIIG locus, which encodes σ^G, is first transcribed early in sporulation by read-through from the upstream spoIIG locus (Fig. 2). However, there is little, if any, translation of this transcript, probably because the spoIIIG ribosome-binding site is sequestered in a stem-loop structure; further, the transcript is not necessary for spore formation (20, 35). Following septation, the spoIIIG locus is transcribed productively from its own σ^F-directed promoter (6, 13), which is active exclusively in the prespore (reviewed in reference 25). Transcription from that promoter, which also depends on a σ^E-directed signal from the mother cell (13, 21), leads to the formation of σ^G (35). When first formed, σ^G is inactive; additional signals are required to activate it. Activation of σ^G requires expression of spoIIIJ in the prespore and of spoIIIA in the mother cell, and these are thought to act via a direct regulator of σ^G that has yet to be identified (30). Activation also requires completion of engulfment of the prespore by the mother cell (33). SpoIIAB can act as an anti-σ for σ^G as well as for σ^F (5, 15, 30) but is now thought not to be a regulator of σ^G activity in the prespore (32).

FIG. 2.

Schematic representation of the spoIIG-spoIIIG region of the chromosome. The promoter for spoIIG requires σ^A and activated Spo0A for expression. The promoter specific to spoIIIG requires σ^F or σ^G for expression. The different transcripts of spoIIIG are indicated. The inverted repeat between spoIIGB and spoIIIG is indicated by opposed arrows.

Here we explore determinants of σ^G regulation. We find that in the absence of both σ^F and the anti-sigma factor SpoIIAB, σ^G becomes active in the mother cell instead of the prespore. Further, activation follows completion of septation rather than completion of engulfment. We describe a previously unrecognized control, in which SpoIIGA can prevent the appearance of σ^G activity, and pro-σ^E can counteract this effect of SpoIIGA. We also find that premature activation of σ^G can prevent septum formation.

MATERIALS AND METHODS

Media.

B. subtilis was grown in modified Schaeffer's sporulation medium (MSSM) or on Schaeffer's sporulation agar (23, 28). When required, the medium contained chloramphenicol at 5 μg/ml, erythromycin at 1.5 μg/ml, neomycin at 3.5 μg/ml, spectinomycin at 100 μg/ml, or tetracycline at 10 μg/ml. Escherichia coli was grown on LB (Luria-Bertani lysogeny broth) agar containing ampicillin at 100 μg/ml when required.

Strains.

B. subtilis 168 strain BR151 (trpC2 metB10 lys-3) was used as the parent strain. B. subtilis strains used are listed in Table 1. The spoIIAΔ4 mutation was described previously (24); the deletion encompassed the entire spoIIA operon, but the ends of the deletion have not been defined. In the spoIIAΔ::neo and spoIIAΔ::spc mutations, the entire spoIIA operon, from 48 bp upstream of the first open reading frame (ORF) to 7 bp downstream of the last ORF, was replaced with the antibiotic resistance cassette. In the mutation designated spoIIAB-ACΔ::neo, the entirety of spoIIAC and all but the 5′ 163 bp of spoIIAB were replaced with a neo cassette. In the spoIIACΔ::neo mutation, 543 bp from the 3′ end of the spoIIAC ORF were replaced with neo. The spoIIGBΔ::spc mutation was derived from EU8701 of Kenny and Moran (16). The spoIIGΔ::cat mutation has 388 bp from the 3′ end of spoIIGA and 338 bp from the 5′ end of spoIIGB replaced with the cat cassette. The spoIIG(P)Δ::cat mutation has the region from 142 bp upstream of spoIIGA (including its promoter) to 388 bp into spoIIGB replaced with the cat cassette. The spoIIGA::cat mutation has the cat cassette inserted at the StuI site located 389 bp from the 3′ end of the ORF. The gene for the pro-less form of σ^E, sigE, was inserted at thrC under the control of the spoIIG promoter; in the encoded protein, N-terminal MH residues are joined to residue 28 (Y) (pro-σ^E numbering). Strain AH2487, containing a translational σ^G-green fluorescent protein (GFP) fusion, was kindly provided by Adriano Henriques. DNA from that strain was used to introduce the fusion into BR151 (spo⁺) to yield SL12673 and into a spoIIAΔ::neo derivative of BR151 to yield SL12674. The spoIIIG::neo mutation has the resistance cassette inserted in the Pst1 site within spoIIIG. DNA containing the amyE::spoIIIG construct (35) was kindly provided by Peter Setlow and DNA with the lonA disruption by Adriano Henriques. The P_spoIIE-spoIIR construct was described by Zhang et al. (38). The P_spac(hy) vector of Quisel et al. (27) was used to place the entire spoIIGB or spoIIGA ORF, with its ribosome-binding site, at thrC under IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible control. E. coli DH5α (Gibco-BRL) was used to maintain plasmids. Details of strain construction are available on request.

TABLE 1.

B. subtilis strains used

Straina	Relevant phenotype	Source
SL10969	sspA-gfp@sspAb	Lab stock
SL10034	spoIIAΔ4 sspA-gfp@sspA	Lab stock
SL10153	spoIIAΔ::spc sspA-gfp@sspA	This study
SL10162	spoIIAΔ::spc amyE::spoIIIG-gfp	This study
SL10215	spoIIAΔ::spc amyE::PspoIIE-spoIIR sspA-gfp@sspA	This study
SL10369	sspA-lacZ@sspA	Simon Cutting
SL11671	spoIIAΔ::neo spoIIGBΔ::spc sspA-lacZ@sspA	This study
SL11727	spoIIAΔ::spc spoIIIG::neo sspA-lacZ@sspA	This study
SL11758	spoIIAΔ::neo spoIIGBΔ::spc thrC::Pspac(hy)-spoIIGB sspA-lacZ@sspA	This study
SL11763	spoIIAΔ::spc spoIIIG::neo amyE::spoIIIG sspA-lacZ@sspA	This study
SL11767	spoIIAΔ::neo spoIIGBΔ::spc sspA-lacZ-cat@sspA lonA::erm	This study
SL11809	spoIIAΔ::spc sspA-lacZ-cat@sspA	This study
SL11813	spoIIAΔ::neo spoIIGBΔ::spc thrC::Pspac(hy)-spoIIGB sspA-gfp@sspA	This study
SL11815	spoIIAΔ::spc spoIIIG::neo amyE::spoIIIG sspA-gfp@sspA	This study
SL11958	spoIIAΔ::spc sspA-lacZ@sspA thrC::PspoIIG-sigE	This study
SL12042	spoIIAΔ::neo spoIIGBΔ::spc thrC::PspoIIG-sigE sspA-lacZ@sspA	This study
SL12137	spoIIAΔ::neo spoIIGΔ::cat sspA-lacZ-tet@sspA	This study
SL12306	spoIIAΔ::neo spoIIGΔ::cat sspA-gfp-spc@sspA	This study
SL12348	spoIIAΔ::neo spoIIGAΔ::cat sspA-lacZ-tet@sspA	This study
SL12359	spoIIAΔ::neo spoIIGΔ::cat sspA-lacZ-tet@sspA thrC::Pspac(hy)-spoIIGA	This study
SL12426	spoIIAΔ::spc spoIIG(P)Δ::catcsspA-lacZ-tet@sspA thrC::sigE	This study
SL12432	spoIIAB, ACΔ::neo sspA-lacZ-cat@sspA	This study
SL12434	spoIIACΔ::neo sspA-lacZ-cat@sspA	This study
SL12436	spoIIAΔ::neo sspA-lacZ-cat@sspA	This study
SL12518	spoIIAΔ::spc thrC::spoIID-gfp amyE::PspoIIE-spoIIR	This study
SL12538	spoIIAΔ::spc spoIIG(P)Δ::catc sspA-gfp-neo@sspA	This study
SL12673	spo+ spoIIIG′-′gfp translational fusion at spoIIIG	This study
SL12674	spoIIAΔ::neo spoIIIG′-′gfp translational fusion at spoIIIG	This study

^aAll strains are in the genetic background of B. subtilis 168 strain BR151 (trpC2 lys-3 metB10). They have all its auxotrophic markers, except for SL10034, which is lys⁺.

^b@ indicates that the fusion has been introduced by single-crossover (Campbell-like) recombination.

^cThe promoter and structural genes of spoIIG are deleted.

Fluorescence microscopy.

Cultures were grown in MSSM at 37°C. A 200-μl volume of culture was mixed with 2 μl of FM4-64 (Molecular Probes) that had been previously diluted to 1 mg/ml in phosphate-buffered saline (Gibco-BRL). Samples were incubated at 37°C for 5 min, and 1 μl of unfixed sample was transferred to a slide and visualized essentially as described by Pogliano et al. (26). Images were captured using a Leica DM IRE2 microscope with a TCS SL confocal system, using a 100× oil immersion objective and Leica imaging software. GFP emission was captured between 500 and 550 nm and FM4-64 emission between 600 and 730 nm; excitation for both fluorophors was at 488 nm. Fluorographs shown are projection images generated from a single stack in the Z plane, with four-point line averaging.

Western blot analysis.

Procedures for Western blotting were performed essentially as described by Serrano et al. (30). The anti-σ^G polyclonal antiserum was incubated with membranes at a dilution of 1:1,000 in TBS-T (20 mM Tris-HCl [pH 7.6], 136 mM NaCl, 0.1% [vol/vol] Tween 20), containing 5.0% nonfat dry milk. Incubation with an anti-rabbit secondary antibody conjugated to horseradish peroxidase was for 30 min at a 1:5,000 dilution, and detection was with an ECL Plus kit (Amersham). Protein samples of 300 μg were used in each lane.

Other methods.

β-Galactosidase was assayed essentially as described previously (23). Specific activity is expressed as nanomoles of ONPG (o-nitrophenyl-β-d-galactopyranoside) hydrolyzed per minute per milligram of bacterial dry weight; results of typical experiments are shown in the figures. B. subtilis transformation, sporulation by exhaustion in MSSM, and other methods were essentially as described previously (2, 38).

RESULTS

Deletion of the genes encoding σ^F and the anti-sigma factor SpoIIAB causes a breakdown of the tight progression of the activation of sporulation-specific σ factors.

When spoIIAC, the structural gene for σ^F, is inactivated by point mutation, no activity is detected for the later-expressed σ factors σ^E, σ^G, and σ^K (4, 13, 21). In contrast to that result, we have found that deletion of the entire spoIIA operon (designated spoIIAΔ) permits activation of σ^G (strain SL12436; Fig. 3), although not σ^E or σ^K (not shown). Thus, σ^G becomes active in the absence of the activities of σ^F and σ^E, effectively disrupting the normal ordered activation of the sporulation-specific sigma factors. Activity was first apparent about 3 h after the onset of spore formation; no activity was detected in a strain with spoIIIG, the structural gene for σ^G, disrupted (SL11727; Fig. 3). Similar results were obtained with a different σ^G-directed promoter (not shown). In the spoIIA deletion strain, activity was detected earlier during spore formation than for the corresponding spo⁺ strain (SL10369). However, although the normal tight regulation of σ^G had been disrupted in the spoIIAΔ mutant, no activity was detected during vegetative growth.

FIG. 3.

Activity of σ^G in a strain with the spoIIA locus deleted. The activity of σ^G is assessed as β-galactosidase expressed from an sspA-lacZ transcriptional fusion in the following strains: filled squares, SL10369, spo⁺; open circles, SL12436, spoIIAΔ; open squares, SL11727, spoIIAΔ spoIIIG::neo.

The spoIIA operon encodes SpoIIAA and SpoIIAB, as well as σ^F; SpoIIAB is an anti-sigma factor for σ^F, and SpoIIAA is the anti-anti-sigma factor that interacts with SpoIIAB (reviewed in references 10 and 37). We tested to see whether deletion of spoIIAB and/or spoIIAA was necessary to obtain σ^G activity in the absence of σ^F. The σ^F-independent activation of σ^G was found to require deletion of spoIIAB (compare SL12434 with SL12432; Fig. 4). SpoIIAB can act as an anti-sigma factor for σ^G, as well as for σ^F (5, 15, 32), so that its loss presumably permitted the establishment of a positive-feedback loop of σ^G-directed transcription of spoIIIG. Consistent with this interpretation, mutations in either spoIIIA or spoIIIJ, which ordinarily block σ^G activation in the mother cell through interaction with SpoIIAB (32), did not block σ^G activation in the spoIIA deletion background (data not shown).

FIG. 4.

The presence of SpoIIAB blocks σ^G activity in a strain that lacks σ^F. The activity of σ^G is assessed as β-galactosidase expressed from an sspA-lacZ transcriptional fusion in the following strains: filled squares, SL12434, spoIIAC (encoding σ^F) deleted; open triangles, SL12432, spoIIAB and spoIIAC deleted; open squares, SL12436, spoIIAA, spoIIAB, and spoIIAC deleted (spoIIAΔ::neo). For each strain, the extent of the deletion in the spoIIA operon is indicated on the right side, with Δ indicating a deleted gene.

The presence of spoIIAA reduced σ^G activity in the strain with spoIIAB and spoIIAC deleted (compare SL12432 with SL12436; Fig. 4) but did not abolish it. SpoIIAA is known to inhibit activation of Spo0A (1), which is a central regulator of early sporulation gene expression (reviewed in reference 25). We think it plausible that inhibition of Spo0A activity by SpoIIAA accounts for the effects of SpoIIAA illustrated in Fig. 4, for example, by reducing expression of the spoIIG operon (see below). However, we did not explore the role of SpoIIAA further.

The location of σ^G activation is switched from the prespore to the mother cell in spoIIA deletion mutants.

During normal spore formation, σ^G activity is confined to the prespore. The prespore specificity is established by the σ^F-directed transcription of spoIIIG, which is itself confined to the prespore. Once σ^G becomes active, a positive-feedback loop is then established in which σ^G directs spoIIIG transcription from the same promoter, which is recognized by both σ^F and σ^G (35, 36). The question arises, what happens in the absence of σ^F? To answer this, we monitored the expression of σ^G-directed sspA-gfp and spoIIIG-gfp transcriptional fusions. Consistent with extensive published results (reviewed in reference 10), their expression was largely confined to the prespore in a spo⁺ background (SL10969; Fig. 5 and Table 2). However, we have found that σ^G activity in spoIIAΔ strains was, within the limits of detection, confined to the mother cell in the majority of GFP-expressing organisms (Fig. 5, strains SL10034 and SL10153; Table 2, SL10034, SL10153, and SL10162); back-crosses of the sspA-gfp fusion into a spo⁺ strain gave recombinants displaying prespore-specific expression, confirming that the fusion was unaltered. The mother cell location of σ^G activity in spoIIAΔ strains was surprising. However, it is consistent with recent results of Serrano et al. (32), who have found that SpoIIAB primarily regulates σ^G by preventing its activation in the mother cell, while having at most a redundant role in blocking σ^G activity in the prespore. That σ^G activity is detected after septation rather than completion of engulfment is probably also the result of loss of SpoIIAB control in the mother cell. The few cells that displayed whole-cell fluorescence (Table 2) did not contain a sporulation division septum.

FIG. 5.

Examples of GFP-expressing cells illustrating the patterns of localization of green fluorescence obtained with different strains containing a σ^G-directed sspA-gfp fusion. Bacteria were stained with FM4-64 to visualize membranes (red). A, SL10969 (spo⁺ sspA-gfp); B, SL10034 (spoIIAΔ4 sspA-gfp); C, SL10153 (spoIIAΔ::spc sspA-gfp); D, SL11813 (spoIIAΔ::neo spoIIGBΔ::spc thrC::P_spac(hy)-spoIIGB sspA-gfp); E, SL11815 (spoIIAΔ::spc spoIIIG::neo amyE::spoIIIG sspA-gfp). An arrow is used to indicate a prespore and an arrowhead a mother cell. A 3-μm scale bar is shown in panel E; all images are on the same scale.

TABLE 2.

Location of GFP expression from different gfp fusions^a

Strain	Relevant genotype	Fusion	No. of cells displaying GFP fluorescenceb
			PS	MC	WC
SL10969	spo+	sspA-gfp	52	0	0
SL10034	spoIIAΔ4	sspA-gfp	1	36	4
SL10153	spoIIAΔ::spc	sspA-gfp	0	21	1
SL10162	spoIIAΔ::spc	amyE::spoIIIG-gfp	0	19	4
SL11813	spoIIAΔ::neo spoIIGBΔ::spc thrC::Pspac(hy)-spoIIGB	sspA-gfp	0	4	46
SL11815	spoIIAΔ::spc spoIIIG::neo amyE::spoIIIG	sspA-gfp	0	3	45
SL10215	spoIIAΔ::spc PspoIIE-spoIIR	sspA-gfp	1	20	38
SL12518	spoIIAΔ::spc PspoIIE-spoIIR	thrC::spoIID-gfp	0	15	18
SL12306	spoIIAΔ::neo spoIIG::cat	sspA-gfp	0	58	32
SL12538	spoIIAΔ::spc spoIIG(P)::cat	sspA-gfp	0	38	0c
SL12673	spo+	spoIIIG′-′gfpd	45	1	0
SL12674	spoIIAΔ::neo	spoIIIG′-′gfpd	0	42	2

^aThe pattern of fluorescence was determined for cells expressing GFP 6 h after the end of exponential growth in MSSM. At this time at least 40% of cells displayed an easily scored GFP signal, with the exception of SL12358, where only about 20% were readily scored. Transcription of sspA-gfp is directed by σ^G and of spoIID-gfp by σ^E; spoIIIG is the structural gene for σ^G. Membranes were visualized by staining with FM4-64; with the exception of SL12306, those cells displaying whole-cell GFP fluorescence exhibited no septa. Fusions to gfp were located at their own locus unless otherwise indicated.

^bPS, prespore specific; MC, mother cell specific; WC, whole-cell fluorescence; almost all cells scored as WC had no visible sporulation septum.

^cA number of other SL12538 cells displayed very weak GFP expression that could not be scored with confidence; they are not included in the table.

^dTranslational fusion; other fusions were transcriptional fusions.

Read-through from the spoIIG locus is important for the mother cell location of σ^G expression.

A factor contributing to the mother cell location of σ^G activity in spoIIA deletion strains might be read through from the spoIIG locus, which is upstream of spoIIIG (Fig. 2). In spo⁺ strains spoIIIG is transcribed by read-through from the spoIIG locus, but there is no detectable translation of spoIIIG from this read-through transcript (20, 35). Indeed, relocating spoIIIG to the distal amyE locus does not impair σ^G activity, and gives efficient spore formation, so that read-through from spoIIG is not ordinarily required for spore formation (35). However, it may be that read-through is important for σ^G activation in the spoIIA deletion strains. Transcription from spoIIG through spoIIIG has been inferred primarily from results with transcriptional lacZ fusions (20, 35) and has proved difficult to detect reproducibly by Northern analysis or S1 mapping (references 16 and 20 and our unpublished observations). We have confirmed by reverse transcription-PCR that under sporulation conditions there was read-through of spoIIIG from spoIIGA in spoIIAΔ as well as in spo⁺ strains (data not shown).

To explore the role of this read-through, we tested the effect of relocating spoIIIG to amyE and found that spoIIIG was actively expressed in a spoIIAΔ strain (Fig. 6; SL11763). Thus, read-through from spoIIG was not necessary for expression of σ^G activity. However, the relocation changed the pattern of σ^G activity, as σ^G became active earlier in spore formation and became more active than when spoIIIG was at its natural locus (Fig. 6; SL12436). The reason for the earlier initiation of transcription of spoIIIG at the ectopic locus is not known. Importantly, in the great majority of organisms expressing the σ^G-directed sspA-gfp fusion, with spoIIIG located at amyE, the fluorescence was uncompartmentalized and the sporulation septum was not formed (Fig. 5 and Table 2; SL11815). Thus, read-through of spoIIIG from spoIIG may be important for obtaining the mother cell specificity of σ^G activity observed in spoIIA deletion strains, even though it was not required to obtain the activity. It should be noted that Fujita and Losick (7) have reported greatly increased activity of the spoIIG promoter in the mother cell following septation in spo⁺ strains.

FIG. 6.

Effect on σ^G activity of relocating spoIIIG to an ectopic locus in a strain with the spoIIA locus deleted. The activity of σ^G is assessed as β-galactosidase activity expressed from an sspA-lacZ transcriptional fusion in the following strains with the spoIIA locus deleted: open triangles, SL12436; open circles, SL11727, spoIIIG::neo; filled circles, SL11763, spoIIIG::neo amyE::spoIIIG.

A second set of experiments reinforced the idea that read-through from spoIIG was indeed important for the mother cell specificity of σ^G expression and that the spoIIG promoter contributed to the strength of σ^G expression in the mother cell of spoIIAΔ strains. In these experiments, two spoIIA deletion strains were compared in which the spoIIG locus was also deleted but not spoIIIG. In one strain, the spoIIG promoter was retained so that it could potentially drive spoIIIG transcription, whereas in the other strain the promoter was not retained. The spoIIG region was replaced with the same cat cassette in the same orientation (away from spoIIIG) in both strains so that the insert should not cause a difference between the strains. There was substantial σ^G activity in the strain that retained the promoter (SL12137; Fig. 7) and much-reduced activity in the strain that did not (SL12426; Fig. 7); both strains displayed similar, abortively disporic phenotypes. In a strain that retained the spoIIG promoter, but not the spoIIG structural genes, σ^G activity was primarily confined to the mother cell (SL12306; Table 2). Assessing the location of σ^G activity in a strain that lacked the promoter was problematic, as the activity was weak; in those cells that expressed sufficient GFP for an unambiguous determination, the activity was confined to the mother cell (SL12538; Table 2). However, in other cells very weak GFP fluorescence was detectable at a level too low to permit determination of its location. Together, the results indicate that the spoIIG promoter contributed to strong mother-cell-specific σ^G activity in spoIIA deletion strains but that some mother-cell-specific activity could be obtained without that promoter.

FIG. 7.

Effect of deletions of the spoIIG operon on σ^G activity in strains with the spoIIA locus deleted. The activity of σ^G is assessed as β-galactosidase expressed from an sspA-lacZ transcriptional fusion in the following strains with the spoIIA locus deleted: open squares, SL12436; closed circles, SL12137 (the spoIIGA and spoIIGB structural genes are deleted, but the spoIIG promoter is retained); open circles, SL12426 (the spoIIGA and spoIIGB structural genes and the spoIIG promoter are deleted).

Pro-σ^E and SpoIIGA control σ^G activity.

We detected no σ^E activity in the spoIIAΔ strains, consistent with previous results (14, 38) and indicating that σ^E was not needed for σ^G activity. However, inactivation of spoIIGB, which is the structural gene for pro-σ^E (12), blocked the appearance of σ^G activity in spoIIAΔ strains (Fig. 8; SL11758 without IPTG). Further, activity of σ^G was restored by expression of spoIIGB in trans from the IPTG-inducible P_spac(hy) promoter (Fig. 8; SL11758 with IPTG) so that the effect of spoIIGB inactivation on the appearance of σ^G activity cannot be explained by polarity on spoIIIG. Rather, the results suggest that either σ^E or pro-σ^E has a role in σ^G activation. Because no σ^E transcriptional activity was detected in spoIIAΔ strains, it seemed likely that pro-σ^E is required and not σ^E. Indeed, expression of a pro-less form of σ^E in a spoIIAΔ spoIIGBΔ mutant did not restore σ^G activity, although the strain did display σ^E activity (data not shown). These results suggest a previously unsuspected role for pro-σ^E that cannot be played by σ^E.

FIG. 8.

Effect of ectopic expression of spoIIGB on σ^G activity in a strain with the spoIIA locus deleted. The activity of σ^G is assessed as β-galactosidase expressed from an sspA-lacZ transcriptional fusion in the following strains with the spoIIA locus deleted: open triangles, SL12436; open circles, SL11758 (spoIIGBΔ::spc thrC::P_spac(hy)-spoIIGB) in the absence of IPTG; filled circles, SL11758 in the presence of IPTG.

Expression of spoIIGB in trans resulted in much stronger and earlier σ^G activity in a spoIIAΔ strain (Fig. 8; SL11758 with IPTG) than when it was expressed in its natural position as part of the spoIIG locus (Fig. 8; SL12436). When spoIIGB was expressed in trans, the σ^G activity was uncompartmentalized, and no sporulation septa were formed (SL11813; Table 2 and Fig. 5). The lack of septa was consistent with the conclusion presented in the previous section that early activation of σ^G prevented spore septum formation. It remains to be established why σ^G became active earlier in SL11758. The spoIIGB gene was expressed earlier than when it was at its natural locus, as the inducer was present throughout growth and sporulation with strains SL11758 and SL12436; presumably, the early appearance of pro-σ^E somehow resulted in the early σ^G activity. Speculatively, pro-σ^E might interact with LonA or some other protease and so protect σ^G from proteolysis.

Pro-σ^E appears to be required only when SpoIIGA is produced. This conclusion is suggested by two sets of experiments. First, when both spoIIGA and spoIIGB were deleted, there was σ^G activity in a spoIIAΔ strain (SL12137; Fig. 7), whereas when spoIIGB and not spoIIGA was deleted, no activity was detected (SL11758; Fig. 8). Second, when both spoIIGA and spoIIGB were deleted, induction of spoIIGA in trans substantially reduced σ^G activity (compare SL12359 in the presence and absence of IPTG; Fig. 9). The expression of σ^G activity in strain SL12359 even in the absence of IPTG was lower than in the corresponding strain, SL12137 (Fig. 7), that did not contain the P_spac(hy)-spoIIGA construct; we think that the reduced expression is a consequence of the leakiness of the inducible promoter. A clue to the possible role of pro-σ^E is provided by the observation that inactivation of lonA, which encodes an ATP-dependent protease (29), partly restored σ^G activity in a spoIIGB mutant strain (data not shown). LonA can degrade σ^G (29), and it may be that SpoIIGA sensitizes σ^G to proteolysis by LonA (or some other protease) and that somehow pro-σ^E but not σ^E can protect σ^G from the proteolysis. Our result is consistent with a role for SpoIIGA in facilitating LonA-directed proteolysis of σ^G, but it does not prove such a role.

FIG. 9.

Effect of ectopic expression of spoIIGA on σ^G activity in strains with the spoIIG and spoIIA loci deleted. The activity of σ^G is assessed as β-galactosidase expressed from an sspA-lacZ transcriptional fusion in the following strains with the spoIIA locus deleted: filled circles, SL12359 (spoIIGΔ::cat thrC::P_spac(hy)-spoIIGA) in the absence of IPTG; open circles, SL12359 in the presence of 1 mM IPTG.

The loss of σ^G-directed transcriptional activity correlates with loss of the σ^G protein in an spoIIA deletion strain in which spoIIGB is also disrupted.

The loss of σ^G activity in spoIIAΔ strains with spoIIGB inactivated could result from absence of the σ^G protein or from the σ^G protein being held inactive. To distinguish between these possibilities, we used two approaches: first, immunoblotting with antibody directed against σ^G; second, fluorescence from a transcriptionally active σ^G-GFP fusion protein. The σ^G protein was first detected by immunoblotting 4 h after the end of exponential growth in spo⁺ and spoIIAΔ strains, and substantially more was detected by 6 h (strains SL10369 and SL12436, respectively; Fig. 10). The presence of the σ^G protein correlated with σ^G activity as detected with an sspA-lacZ fusion (not shown). No σ^G protein was detected in an spoIIIG knockout mutant (strain SL11727). In contrast to the strong band observed for the spoIIAΔ mutant SL12436, the protein was barely detectable in a spoIIAΔ mutant with spoIIGB also inactivated (strain SL11671; Fig. 10). This result indicated that inactivation of spoIIGB resulted in the almost total absence of the σ^G protein, not simply its inhibition, in the spoIIAΔ background. The presence of σ^G was not restored by expression in trans of a constitutively active form of σ^E in a strain with spoIIG deleted (strain SL12042; Fig. 10).

FIG. 10.

Effect of deletion of spoIIA and spoIIGB on the accumulation of σ^G during sporulation. Protein samples (300 μg) were obtained at the indicated time (h) after the start of spore formation in MSSM and fractionated by electrophoresis. They were analyzed for σ^G by Western blotting using a polyclonal antiserum to σ^G. The strains used were SL10369 (spo⁺), lanes 1 to 4; SL12436 (spoIIAΔ), lanes 9 to 12; SL11671 (spoIIAΔ spoIIGBΔ), lanes 5 to 8; SL12042 (spoIIAΔ spoIIGBΔ thrC::sigE), lanes 13 to 16; SL11727 (spoIIAΔ spoIIIG::neo), lane 17. Samples were taken at the end of exponential growth (lanes 1, 5, 9, and 13) and 2 h (lanes 2, 6, 10, and 14), 4 h (lanes 3, 7, 11, and 15), and 6 h after the end of exponential growth (lanes 4, 8, 12, 16, and 17). Lanes 1 to 8 and 9 to 17 are from two separate gels.

We also utilized strains in which spoIIIG was replaced by a translational spoIIIG-gfp fusion via single-crossover (Campbell-like) plasmid integration. The fusion protein retained σ^G activity and did not block spore formation in a strain in which it was the sole copy of σ^G; the location of GFP is inferred to be a good indicator of the location of σ^G protein. In a spo⁺ strain, GFP fluorescence was located in the prespore (SL12673; Table 2). When introduced into a spoIIAΔ strain, however, GFP fluorescence was confined to the mother cell (SL12674; Table 2), correlating with the location of σ^G activity in spoIIAΔ strains. The result is consistent with mother-cell-specific spoIIIG transcription. No GFP was detected in a spoIIAΔ strain in which spoIIGB was also inactivated (data not shown), so there was no indication of σ^G being present in an inactive form in that strain.

σ^E activity in the mother cell does not block σ^G activity.

In strains deleted for spoIIA and with spoIIG intact, pro-σ^E is ordinarily not processed, and so σ^E is not active (38). We tested in two ways the effect on σ^G activity of having active σ^E in spoIIA deletion strains. (i) We introduced spoIIR under the control of the spoIIE promoter. spoIIR is the only σ^F-directed gene required for processing of pro-σ^E to its active form, and this construct results in σ^E activity in the absence of σ^F (38). Transcriptional activity of both σ^E and σ^G was detected with the construct (strains SL10215 and SL12518; Table 2). As reported previously for σ^E (38), about half the GFP-expressing bacteria showed mother cell specificity; the rest showed whole-cell activity and had no sporulation septum, probably because the slightly earlier σ^E activation in that part of the population had prevented septum formation. Similar localization was observed for σ^G activity (SL10215; Table 2). (ii) We inserted at thrC the gene for a constitutively active, pro-less form of σ^E. This construct resulted in lower σ^E activity than the P_spoIIEspoIIR construct but a similar distribution of both σ^E activity and σ^G activity (not shown). Thus, as tested in two ways, σ^E did not have an antagonistic role towards σ^G. That many bacteria displayed mother-cell-specific σ^E activity reinforces the previous view that σ^F has at best a redundant role in directing σ^E activity to be confined to the mother cell (7, 38). The result with the pro-less form of σ^E suggests that processing of the pro sequence is not essential for compartmentalization of σ^E activity.

Both σ^E and σ^G were active before the completion of engulfment in strains SL10215 and SL12518. The σ^E activity in these strains enabled bacteria to complete engulfment (not shown); the corresponding strains, differing only by the lack of active σ^E, did not develop beyond septum formation. We infer that early activation of σ^G in the mother cell does not prevent engulfment. Both σ^E and σ^G activities were detected in the mother cell, suggesting no incompatibility between the two sigma factors, although we did not directly test whether they were active in the same mother cell. Presumably, both activities survive any competition with each other and with σ^A (18) for core RNA polymerase.

DISCUSSION

We report here that in the absence of both σ^F and the anti-sigma factor SpoIIAB, σ^G becomes active in the mother cell and not in the prespore during sporulation of B. subtilis. This is the first report, to our knowledge, of an efficient switch between prespore and mother cell of the location of activity of a sporulation-specific σ factor. The switch to mother cell location of the σ^G activity says that, at least in strains with spoIIAB and spoIIAC deleted, there is no “prespore-only” tag on σ^G and, likewise, no signal in the mother cell saying “no σ^G activity allowed.” SpoIIAB acts as an anti-sigma factor for σ^G as well as for σ^F (15, 17) and is thought to act against σ^G in vivo primarily to prevent inappropriate activation in the mother cell (32). Our results are consistent with this interpretation.

The other factor thought to contribute to the mother cell location of σ^G activity in strains with the spoIIA locus deleted is transcription of spoIIIG, the structural gene for σ^G, from upstream promoters, most notably the spoIIG promoter (Fig. 2). In support of this statement, relocating spoIIIG away from its normal location, which is downstream of the spoIIG locus, abolished the mother cell specificity. Also, mother-cell-specific σ^G activity was detected in a spoIIA deletion strain in which the spoIIG structural genes were deleted while leaving in place the spoIIG promoter upstream of spoIIIG (SL12306; Table 2). Extending the deletion to include the spoIIG promoter substantially reduced σ^G activity (Fig. 7), indicating the importance of that promoter. However, residual mother-cell-specific σ^G activity remained even in the absence of the spoIIG promoter (SL12538; Table 2), suggesting that some other promoter also played a role.

In Spo⁺ strains spoIIIG is transcribed productively (i.e., resulting in σ^G, which becomes active) from its own promoter. This transcription is primed by σ^F and so occurs only in the prespore. A positive-feedback loop is then established in which transcription is directed from the same promoter by σ^G (reviewed in reference 10). However, the spoIIIG locus is also transcribed by read-through from the spoIIG locus. The read-through transcript is normally translated poorly, if at all, probably because it forms a hairpin structure that sequesters the presumed ribosome binding site for spoIIIG (20, 35). Moving spoIIIG to an ectopic locus away from spoIIG does not impair spore formation in an otherwise spo⁺ strain, so any read-through transcript is clearly unnecessary for spore formation under the conditions used (35). Nevertheless, spoIIIG is located immediately downstream of spoIIG in all of the sequenced spore-forming bacteria (34). Such a juxtaposition suggests that in some circumstances the read-through may be important. Presumably in those circumstances the inhibitory effects of mRNA secondary structure can be overcome, as happens for the expression of rpoH and rpoS in E. coli (8), so as to produce some σ^G.

In strains with spoIIAB and spoIIAC deleted there is no σ^F priming and no SpoIIAB to block the activity of any σ^G formed in the mother cell as a result of read-through from spoIIG. In these circumstances, a small amount of active σ^G formed after the burst of spoIIG transcription that follows septation (7) may be sufficient to prime a positive-feedback loop of σ^G-directed transcription of spoIIIG. But now, the feedback loop is established in the mother cell, so that σ^G activity is confined to the mother cell. With respect to the prespore and the predivisional cell, expression from the spoIIG promoter is much reduced compared to that in the mother cell (7). Further, SpoIIAB has at most a redundant role in regulating σ^G in the prespore and also before septum formation, when other unidentified controls are thought to prevent activation (32). The net result is σ^G activity confined to the mother cell in strains with spoIIAB and spoIIAC deleted strains. Consistent with this interpretation, an spoIIIG′-′gfp translational fusion inserted at the spoIIIG locus is expressed only in the mother cell in a spoIIAΔ strain and only in the prespore in an spo⁺ strain (Table 2).

When spoIIIG was moved to an ectopic locus, amyE, away from the spoIIG promoter in spoIIAΔ strains, σ^G became active earlier during spore formation and was more active than when at its natural locus. It is not known why there was this earlier and stronger activity. Whatever the explanation, σ^G activity was uncompartmentalized and no sporulation septum was formed. As neither σ^F nor σ^E was active, the result suggests that σ^G activation can, like that of σ^E (11) and σ^F (3, 9), prevent subsequent septum formation. The function of such an inhibitory role for σ^G in a wild-type genetic background is not clear, but it may relate to the phenomenon of commitment, namely, the ability of an organism to continue to form a spore despite the addition of nutrients that might otherwise trigger an inappropriate restoration of growth and division (22). Thus, σ^G would prevent division of the prespore at later stages of spore formation when σ^F activity is thought to be curtailed (18).

We report a previously unrecognized control of σ^G activity involving pro-σ^E and SpoIIGA, which became apparent in spoIIAΔ strains. We found that in the presence of SpoIIGA, σ^G activity is only detected when pro-σ^E is also present. Two lines of evidence suggest that it is pro-σ^E and not σ^E that is required for this effect. First, no σ^E activity was detected in the spoIIA deletion strains that displayed σ^G activity; second, σ^G activity was not detected when a pro-less form of σ^E, and not pro-σ^E, was expressed from an ectopic locus, although σ^E activity was now detected. We think that pro-σ^E is needed for σ^G activation only when SpoIIGA is present, because σ^G activity was detected in strains with both spoIIGA and spoIIGB deleted. Presumably, pro-σ^E works to protect σ^G from protease action or from some other inhibitory mechanism that is stimulated by SpoIIGA. The amount of σ^G protein was dramatically reduced in the strain with spoIIGB deleted (Fig. 10), so we think it likely that the effect of pro-σ^E is to stabilize σ^G rather than to activate a preexisting inactive form.

The protease LonA has previously been shown to degrade σ^G (29, 31), and inactivation of lonA partly restored σ^G activity to a spoIIGBΔ spoIIAΔ mutant strain. It may be that LonA and SpoIIGA/pro-σ^E represent separate regulators of σ^G activity and that loss of LonA leads to a large σ^G increase that disrupts the other system. Alternatively, or additionally, pro-σ^E may protect σ^G from SpoIIGA acting to stimulate proteolysis of σ^G by LonA. The mechanism of SpoIIGA/pro-σ^E regulation remains unknown. Nevertheless, our results suggest that several partly overlapping mechanisms ordinarily act to prevent σ^G activation in the mother cell. They indicate that regulators of σ^E and σ^F can also regulate σ^G.