Journal of Bacteriology, June 2004, p . 4000-4013, Vol . 186, No . 12

Role of the Anti-Sigma Factor SpoIIAB in Regulation of ^G during Bacillus subtilis Sporulation

Mónica Serrano,¹ Alexandre Neves,^1, Cláudio M . Soares,¹ Charles P . Moran Jr.,² and Adriano O . Henriques^1*

Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, 2781-901 Oeiras Codex, Portugal,¹ Department of Microbiology, Emory University School of Medicine, Atlanta, Georgia 30322²

Received 27 October 2003/ Accepted 16 February 2004

 
RNA polymerase sigma factor ^F initiates the prespore-specific program of gene expression during Bacillus subtilis sporulation . ^F governs transcription of spoIIIG, encoding the late prespore-specific regulator ^G . However, transcription of spoIIIG is delayed relative to other genes under the control of ^F, and after synthesis, ^G is initially kept in an inactive form . Activation of ^G requires the complete engulfment of the prespore by the mother cell and expression of the spoIIIA and spoIIIJ loci . We screened for random mutations in spoIIIG that bypassed the requirement for spoIIIA for the activation of ^G . We found a mutation (spoIIIGE156K) that resulted in an amino acid substitution at position 156, which is adjacent to the position of a mutation (E155K) previously shown to prevent interaction of SpoIIAB with ^G . Comparative modelling techniques and in vivo studies suggested that the spoIIIGE156K mutation interferes with the interaction of SpoIIAB with ^G . The ^GE156K isoform restored ^G-directed gene expression to spoIIIA mutant cells . However, expression of sspE-lacZ in the spoIIIA spoIIIGE156K double mutant was delayed relative to completion of the engulfment process and was not confined to the prespore . Rather, ß-galactosidase accumulated throughout the entire cell at late times in development . This suggests that the activity of ^GE156K is still regulated in the prespore of a spoIIIA mutant, but not by SpoIIAB . In agreement with this suggestion, we also found that expression of spoIIIGE156K from the promoter for the early prespore-specific gene spoIIQ still resulted in sspE-lacZ induction at the normal time during sporulation, coincidently with completion of the engulfment process . In contrast, transcription of spoIIIGE156K, but not of the wild-type spoIIIG gene, from the mother cell-specific spoIID promoter permitted the rapid induction of sspE-lacZ expression . Together, the results suggest that SpoIIAB is either redundant or has no role in the regulation of ^G in the prespore .

 
Gene expression in the prespore and mother cell chambers of sporulating Bacillus subtilis is controlled by RNA polymerase sigma subunits whose activity is restricted to a specific cell type (22, 31, 37, 46) . The activation of the sporulation-specific sigma factors is tightly coupled to the completion of key morphological intermediates in the process and also relies on signaling pathways that operate between the two cell types and that keep the prespore and mother cell lines of gene expression in close register (22, 31, 37, 46) . Soon after the asymmetric division of the sporangial cell, an event that creates the prespore and the much larger mother cell, the first compartment-specific sigma factor ^F becomes active in the prespore (22, 31, 37, 46). ^F triggers the activation of ^E in the mother cell, which together with ^F drive the migration of the septal membranes around the prespore . This process is termed engulfment and results in the formation of a protoplast isolated from the external medium, fully encircled by the mother cell cytoplasm (22, 31, 37, 46) . After engulfment, ^F is replaced by ^G, which controls late stages of development in this compartment and which also triggers the activation of the late mother cell-specific regulator ^K (22, 31, 37, 46) . The activities of both ^G and ^K are required for the assembly of the protective layers that encase the mature spore (22, 31, 37, 46) .

Synthesis of ^F occurs in the predivisional cell, but its activation is restricted to the prespore by the action of three regulatory proteins, SpoIIAA, SpoIIAB, and SpoIIE . SpoIIAB is an anti-sigma factor that binds to ^F as a dimer, preventing its association with RNA polymerase, whereas SpoIIAA is an anti-anti-sigma factor that in an unphosphorylated state interacts with SpoIIAB and releases ^F from the SpoIIAB-^F complex (1, 2; reviewed in references 31 and 37) . SpoIIE is a septum-bound phosphatase that is also produced in the predivisional cell that promotes the preferential dephosphorylation of SpoIIAA-P in the prespore (reviewed in references 31 and 37) .

The transcriptional activity of ^F can be divided into an early phase and a late phase . Transcription of the spoIIIG gene (encoding ^G) is induced as part of the late phase, towards the end of the engulfment process (29) . After synthesis, ^G does not become active until the engulfment process is complete (29) . Once activated and since ^G efficiently recognizes its own promoter, its cellular levels increase rapidly, allowing for the deployment of the ^G regulon (17, 47) . Because of this autoregulation, both the late transcription of spoIIIG and the negative regulation of ^G appear to ensure that its transcriptional activity is effectively coupled to completion of the engulfment process and does not occur prematurely or ectopically (31, 37, 45) . The tight coupling of ^G activation to the conclusion of the engulfment sequence may serve to ensure that biogenesis of the spore integuments is not initiated during movement of the engulfment membranes (31, 37, 45, 46) .

Conclusion of the engulfment process is not sufficient for the activation of ^G, which further requires expression of several genes, including the eight cistrons of the spoIIIA operon and the spoIIIJ gene (6, 19, 34). ^G accumulates in spoIIIA or spoIIIJ mutant cells but is unable to activate transcription from its target promoters (19, 41) . The spoIIIA operon encodes several putative membrane proteins and is expressed in the mother cell under the direction of ^E (15) . The spoIIIJ gene is expressed during vegetative growth and encodes a membrane protein translocase of the YidC/Oxap1 family (6, 27, 48) . Despite the fact that its product may accumulate in both the prespore and the mother cell (6, 27), expression of spoIIIJ in the prespore is sufficient for the activation of ^G and sporulation (41) .

Two negative regulators of ^G are known, the anti-sigma factor SpoIIAB and the LonA protease (3, 8, 19, 21, 35, 40) . Expression of spoIIIG prior to the asymmetric division of the sporangial cell blocks sporulation, a phenotype that can be suppressed by a multicopy allele of spoIIAB (21), and certain point mutations in spoIIAB result in expression of ^G-dependent genes under conditions that do not support efficient sporulation (8, 35) . Moreover, SpoIIAB binds to ^G in vitro under conditions that also promote binding of SpoIIAB to ^F (19), and the structure of a dimer of Bacillus stearothermophilus SpoIIAB in complex with ^F shows that most of the residues involved in the interaction are conserved in ^G, but not in other sigma factors (2) .

While it seems clear that SpoIIAB can regulate ^G under nonsporulation conditions or in the predivisional cell at the onset of sporulation, the evidence for a role in the control of ^G in the prespore is less clear (3, 8, 19, 21, 35) . On the one hand, SpoIIAB seems to disappear from the prespore coincidently with the first manifestations of ^G activity, but it persists in the prespore of a spoIIIA mutant (21) . In addition, production of a form of ^G (^GE155K) that is not efficiently bound by SpoIIAB in vitro allows expression of the ^G-controlled sspE gene in spoIIIA or spoIIIJ mutants, suggesting that the expression of both loci is required to antagonize the inhibitory action of SpoIIAB upon ^G (19, 41) . However, expression of sspE in spoIIIGE155K cells bearing mutations in either spoIIIA or spoIIIJ does not occur prematurely, suggesting that the activity of ^GE155K is still regulated in the double mutants (19, 41) . Also, there seems to be very little, if any, free SpoIIAB in the prespore (28), and the anti-sigma factor would have to be able to negatively regulate ^G at a time when SpoIIAB itself is antagonized by the anti-anti-sigma SpoIIAA in order to release active ^F (reviewed in references 31 and 37) . Since the interaction of SpoIIAB with ^G appears to be very similar to the interaction of SpoIIAB with ^F (7), it seems unlikely that at least prior to completion of the engulfment process, SpoIIAB decisively contributes to the regulation of ^G . Mutations in lonA, coding for the ATP-dependent LonA protease, also result in ^G activity under nonsporulation conditions, and result in some expression of sspE-lacZ in cells of a spoIIIA mutant during sporulation (40) .

Here we have analyzed the role SpoIIAB plays in the regulation of ^G in sporulating cells . We screened for mutations in spoIIIG that allowed expression of the ^G-controlled sspE-lacZ fusion in a spoIIIA background and found a single mutation that converted a glutamate at position 156 of ^G to a lysine . However, we found that expression of sspE-lacZ in a spoIIIGE156K spoIIIA double mutant was delayed relative to the completion of the engulfment process and was not confined to the prespore . Rather, ß-galactosidase accumulated throughout the whole cell at late times of sporulation . We also forced the early expression of spoIIIGE156K in the prespore from the spoIIQ promoter and found no premature induction of sspE-lacZ expression . In contrast, expression of spoIIIGE156K in the mother cell readily results in sspE-lacZ expression . The results suggest that the activity of ^G is regulated in the prespore compartment by a SpoIIAB-independent mechanism and that SpoIIAB is either redundant or plays only a minor role .

 
Bacterial strains and general methods. The B . subtilis strains used in this work (listed in Table 1) are congenic derivatives of the Spo⁺ strain MB24 (trpC2 metC3) (14) . Luria-Bertani (LB) medium was used for the maintenance of Escherichia coli DH5 (Bethesda Research Laboratories) and B . subtilis . Sporulation was induced in Difco sporulation medium (DSM) and assessed as described previously (13) . All other general methods were performed as described previously (13) .

 
Structure of a complex between B . subtilis SpoIIAB and ^G by comparative modelling techniques. The structure of the SpoIIAB-^F complex from B . stearothermophilus (Protein Data Bank code 1LO0) (2) was used here to derive, by comparative modelling techniques, the SpoIIAB-^F and SpoIIAB-^G complexes from B . subtilis (25, 39) . The structure of the SpoIIAB-^F complex from B . stearothermophilus (Protein Data Bank code 1LO0) (2) was used here to derive the SpoIIAB-^F and SpoIIAB-^G complexes from B . subtilis . The structure of SpoIIAB from B . subtilis can be modelled on the basis of SpoIIAB from B . stearothermophilus, because the two sequences present 75% identity and 90% similarity and only five residues at the C terminus cannot be aligned (25, 39) . The structure of the SpoIIAB-^F complex from B . stearothermophilus contains information for only part of ^F (from residues 104 to 160), and modelling of the sigma factors was restricted to the residues that are homologous to this segment . For B . subtilis ^F, this segment presents 84% identity and 88% similarity to ^F from B . stearothermophilus, which also suggests that a very good model will be obtained . In contrast, B . subtilis ^G shows 30% identity and 63% similarity with ^F from B . stearothermophilus . The SpoIIAB-^F from B . stearothermophilus has a bound ADP molecule that was not modelled, because no contacts are made between this region and the sigma factor . Modeller (38) version 6.1 was used for all comparative modelling tasks . Sequences for both proteins in the complex were simultaneously aligned against the X-ray structure of SpoIIAB-^F from B . stearothermophilus, and 20 models were generated using these alignments . The model showing the lowest value of the objective function was chosen and analyzed using PROCHECK (23) . In the case of SpoIIAB-^F from B . subtilis, the Ramachandran plot showed 87.4% residues in most favored regions, 9.5% residues in additional allowed regions, 2.5% in generously allowed regions, and 0.6% residues in disallowed regions . The residues in disallowed regions, Leu 103 and Arg 105, are homologous to residues in SpoIIAB-^F from B . stearothermophilus that are also in disallowed regions . This region corresponds to the ADP binding site, and the conformation of these two residues is the most probable one . In the case of SpoIIAB-^G from B . subtilis, we obtained 88.1% residues in most favored regions, 8.5% residues in additional allowed regions, 3.1% in generously allowed regions, and 0.3% is disallowed regions . In this case, only Arg 105 is in a disallowed zone of the Ramachandran plot, and for the reasons stated above, its conformation was considered the most probable .

Construction of an sspE-lacZ fusion. First, a 500-bp HindIII-to-HincII fragment released from pUC12sspE (10) was inserted between the SmaI and HindIII sites of pBluescript SKII(+) (Stratagene, La Jolla, Calif.) to generate pAH225 . Next, NheI- and EcoRI-digested pAH235 was mixed with the lacZ gene released from pPP207 (49) by digestion with SpeI and MscI, and a neomycin resistance (Nm^r) determinant was released from pBEST502 (16) with EcoRI and SmaI . Linearization of the resulting plasmid, pMS53, with ScaI permitted integration of the sspE-lacZ fusion into the sspE locus of strain MB24, producing strain AH2321 (Table 1) . The sspE::sspE-lacZ allele in strain AH2321 is hereafter abbreviated sspE-lacZ for simplicity (Table 1) .

Construction of spoIIIG mutations. To create an in-frame deletion of the spoIIIG gene, a 1.1-kb DNA fragment was first released from pSP72IIIG (19) by digestion with BglII and SalI and inserted between the BglII and XhoI sites of pLitmus 28 (New England Biolabs, Beverly, Mass.) to yield pAH220, which then served as a PCR template using primers spoIIIG-247R and spoIIIG-559D (Table 2) . The PCR template was first treated with DpnI and then with PstI and last, it was autoligated, yielding pMS124 . Sequencing confirmed the in-frame deletion of codons 13 to 130 of spoIIIG . Competent cells of strain MB24 were cotransformed with pMS124 and chromosomal DNA from strain ZB307 (52), with selection for methionine prototrophy . Spo^– congressants appeared at a frequency of about 3% . One, shown by PCR to carry a deletion of the spoIIIG gene (referred to as spoIIIG) was named AH3795 (Table 1) . Strain AH3795 was transformed with ScaI-linearized pMS53 (sspE-lacZ) to produce AH2452 . Strain AH2452 was then transformed with chromosomal DNA from strain AH62 (spoIIIA::Tn917HU24, hereafter abbreviated to spoIIIA::Tn917) (Table 1) to create strain AH2456 .

 
To create an insertion-deletion spoIIIG mutant, a 922-bp DNA fragment containing the spoIIIG gene was first released from pSP72IIIG (19) by partial digestion with EcoRI and HindIII and inserted between the EcoRI and HindIII sites of pLitmus 29 (New England Biolabs) to create pMS33 . Next, a chloramphenicol resistance (Cm^r) cassette was released from pMS38 (51) by digestion with NsiI and PstI and cloned into PstI-digested pMS33 to yield pMS40 . Strain AH1870 (Cm^r) in which disruption of the spoIIIG gene by a double-crossover event was verified by PCR resulted from the transformation of strain MB24 with pMS40 (Table 1) .

Insertion of an intact copy of the spoIIIG gene at amyE. A copy of the spoIIIG gene was inserted at amyE in two steps . We first isolated a 427-bp HindIII-to-BamHI fragment from pTK4 encompassing the spoIIGB-spoIIIG intergenic region (20), which was introduced between the HindIII and BamHI sites of the amyE integrational vector pMLK83 (18), to create pAH235 . Strain AH1043 (AmyE^–/Nm^r) (Table 1) resulted from the transformation of strain MB24 with XbaI-linearized pAH235 . Then, a spectinomycin resistance (Sp^r) cassette was released from pAH256 (13) by digestion with SpeI and NcoI and cloned between the same sites of pMS33 (see above), yielding pMS37 . Last, a fragment carrying the Sp^r determinant and the spoIIIG gene was released from pMS37 by digestion with SmaI and EcoRI and cloned between the EcoRI and NruI sites of pDG364 (4) . Transformation of AH1043 with the resulting plasmid, pMS45, created AH1842 (Sp^r/Nm^s and Spo⁺) (Table 1), in which the presence of an intact spoIIIG gene at amyE was verified by PCR . Transformation of strain AH1842 with chromosomal DNA from AH1870 (spoIIIG::cat) produced AH1843 (spoIIIG::cat amyE::spoIIIG Sp^r) (Table 1) .

Random mutagenesis of spoIIIG. Strain AH1843 (Table 1) was mutagenized with N-methyl-N'-nitro-N-nitrosoguanidine essentially as described previously (4) . The mutagenesis was effective, as 1% of all colonies obtained from the transformation of strain AH1870 (spoIIIG::cat) (Table 1) with chromosomal DNA from mutagenized AH1843 selecting for Sp^r (which selects for the spoIIIG copy at amyE) failed to complement the null mutation in the spoIIIG gene . In order to select for ^G mutants that would bypass the need of the spoIIIA locus for ^G activity, chromosomal DNA from mutagenized AH1843 was used to transform AH2456 (spoIIIG spoIIIA::Tn917 sspE-lacZ [see above]) to Sp^r . Transformants that showed ß-galactosidase activity on DSM plates supplemented with 5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside (X-Gal) (dark blue colonies) were selected and purified . The linkage between the Lac⁺ phenotype and the Sp^r marker was verified by retransforming the screening strain AH2456 with chromosomal DNA from the Lac⁺ mutants (only mutants showing 100% linkage between the Lac⁺ and Sp^r traits or close to 100% linkage were considered) . One Lac⁺ transformant (AH3791) was selected (Table 1), and the spoIIIG gene present at amyE was sequenced after PCR amplification using primers spoIIIG-392D and spec-R . Strain AH3791 was found to harbor a single-nucleotide change (GAA to AAA) at codon 156 of the spoIIIG gene . A control strain, AH3790, with the wild-type allele of spoIIIG at amyE, was constructed by transformation of AH2456 (spoIIIG spoIIIA::Tn917 sspE-lacZ [see above]) with chromosomal DNA from AH1842 . Strains AH3786 and AH3787 were constructed by transformation of AH2452 (spoIIIG sspE-lacZ [see above]) with chromosomal DNA from AH1842 and AH3791, respectively .

Fusion of the xylA promoter to the spoIIIG gene. Initially, the 5' end of the spoIIIG gene and the xylA promoter region (from positions –256 to –1 relative to the transcriptional start site) were amplified separately, from chromosomal DNA of B . subtilis MB24 . Primers P_xylA-spoIIIG and spoIIIG-2385R were used for the spoIIIG gene, and primers P_xylAD and P_xylAR were used for the xylA gene (Table 2) . The 370-bp spoIIIG fragment was mixed with the 256-bp xylA fragment, and the resulting fragment of 619 bp was amplified using primers P_xylAD and spoIIIG-2385R . The P_xylA-spoIIIG fragment was digested with EcoRI and BamHI and ligated to similarly cut pDG364 (4), to yield pMS237 . Strains AH3786 and AH3787 (Table 1) were transformed with BamHI-linearized pMS237, selecting for Cm^r/Sp^r cells, to yield strains AH2490 and AH2491, which carry a fusion of the xylose-inducible P_xylA promoter to the spoIIIG and spoIIIGE156K alleles at amyE, respectively (abbreviated to P_xylA-spoIIIG and P_xylA-spoIIIGE156K, respectively) (Table 1) .

Fusion of the spac promoter to the spoIIAB gene. The spoIIAB gene was PCR amplified from the chromosomal DNA of strain MB24, using primers spoIIAB-166D and spoIIAB-665R (Table 2) . The 499-bp spoIIAB fragment was digested with BamHI and SpeI and introduced between the BglII and SpeI sites of pDH88 (12), to yield pMS236 . Next, a 2,189-bp EcoRI-to-BamHI fragment released from pMS236 was inserted between the same sites of pDG1664 (9), to generate pMS238 . Strains AH2490 and AH2491 (Table 1) were transformed with XhoI-linearized pMS238 selecting for erythromycin resistance (Er^r), to yield AH2492 and AH2493, respectively, which carry a fusion of the P_spac promoter to spoIIAB inserted at the thrC locus (thrC::P_spac-spoIIAB, abbreviated to P_spac-spoIIAB) (Table 1) .

Fusions of spoIIIG to different sporulation promoters. Fusions of spoIIIG to the spoIIQ and to the spoIID promoters were constructed as follows . Initially, the 5' end of the spoIIIG gene and its promoter region were amplified separately from chromosomal DNA of a wild-type B . subtilis strain . The following primers were used (Table 2): for spoIIIG, spoIIIG-spoIIQ, and spoIIIG-spoIID, primers spoIIIG-761R and spoIIIG-2385R; for spoIIQ, primers spoIIQ-152D and spoIIQ-500R; for spoIID, primers spoIID-1D and spoIID-500R . The 370-bp spoIIIG fragment was mixed with the 350-bp spoIIQ fragment or with the 500-bp spoIID fragment, and the resulting fragments of 720 and 870 bp were amplified using primers spoIIQ-152D and spoIIIG-761R for P_spoIIQ-spoIIIG or primers spoIID-1D and spoIIIG-2385R for P_spoIID-spoIIIG . The P_spoIIQ-spoIIIG fragment was digested with EcoRI and BglII and ligated to pDG364 (4) digested with EcoRI and BamHI to yield pMS134 . The P_spoIID-spoIIIG fragment was digested with EcoRI and BamHI and ligated to similarly cut pDG364 (4) to yield pMS162 . Strain AH3786 (Table 1) was transformed with pMS134 or pMS162 selecting for Cm^r/Sp^r cells, to yield AH3788 and AH2460, respectively . AH3787 (Table 1) was also transformed with pMS134 and pMS162 to yield the Cm^r/Sp^r strains AH3789 and AH2461, respectively . Fusion of the spoIIIG gene to the spoIIQ promoter and to the spoIID promoter by a double-crossover event at amyE was verified by PCR; the fusion alleles are abbreviated to P_spoIIQ-spoIIIG and P_spoIID-spoIIIG . Strain AH2462 was constructed by transformation of AH2452 (spoIIIG sspE-lacZ [see above]) with DNA from AH2464 (lonA::cat) (Table 1) . Last, AH2462 was transformed with chromosomal DNA from AH2460 or AH2461 to produce AH2463 and AH2465, respectively (Table 1) .

Immunoblotting. B . subtilis whole-cell lysates were prepared and Western blot analysis was performed as described previously (41) .

ß-Galactosidase assays. ß-Galactosidase activity was assayed with the substrate o-nitrophenol-ß-D-galactopyranoside (ONPG), and enzyme activity was expressed in Miller units as described previously (13) .

Fluorescence microscopy. Immunofluorescence microscopy was conducted essentially as described previously (11, 33) except that a non-cross-linking fixative, Histochoice (Amresco, Solon, Ohio) was used . Immunolabeling was performed with rabbit polyclonal antibodies against ß-galactosidase (Eppendorf-5 Prime, Inc., Boulder, Colo.) at a 1:1,000 dilution . A secondary antibody conjugated to Alexa Fluor 488 (Molecular Probes, Eugene, Oreg.) was used at a 1:500 dilution . To assess the completion of the engulfment process (morphological stage III of sporulation) (31), samples (0.5 ml) of DSM cultures were collected throughout sporulation and resuspended in the same volume of phosphate-buffered saline (8 mM sodium phosphate [pH 7.5], 150 mM NaCl) supplemented with FM-4-64 (5 µg/ml), 4',6-diaminodino-2-phenylindole (DAPI) (0.2 µg/ml), and Mitotracker green FM (MTG) (15 µg/ml) (44) . Cells were scored as having reached stage III when the membrane-impermeable stain FM4-64 (but not MTG) was excluded from the prespore membrane (44) . In immunofluorescence experiments, the pattern of nucleoid staining with DAPI was used as an indication of the sporulation stage: just after formation of the asymmetric septum, the prespore nucleoid appears highly condensed, whereas soon after conclusion of the engulfment process, the mother cell and prespore nucleoids appear equally condensed (11, 43) . Samples were observed in a Leica fluorescence microscope (DMRA2) using Leica filters A4, L5, and N3 . All samples were observed with a 63x objective lens . Images were acquired with a Cool Snap HQ camera (Roper Scientific, Tucson, Ariz.) and recorded and processed for publication using Adobe Photoshop .

 
Isolation and characterization of the spoIIIGE156K allele. To analyze the mechanism by which ^G is kept inactive in cells of a spoIIIA mutant, we sought random mutations in spoIIIG at amyE that bypassed the requirement for spoIIIA for expression of the ^G-controlled sspE-lacZ fusion (26) (see Materials and Methods) . We isolated one Lac⁺ mutant called AH3791 (Table 1) upon transformation of B . subtilis strain AH2456 (spoIIIG spoIIIA::Tn917 sspE-lacZ) with DNA from strain AH1843 mutagenized by N-methyl-N'-nitro-N-nitrosoguanidine (Table 1) . Strain AH3791 harbored a single-nucleotide change (GAA to AAA) at codon 156 of the spoIIIG gene at amyE causing the replacement of a glutamic acid for a lysine . This allele of spoIIIG was designated spoIIIGE156K . Strains bearing a wild-type spoIIIG gene (AH3786) or the spoIIIGE156K allele (AH3787) at amyE in a spoIIIG mutant background (spoIIIG sspE-lacZ) are Spo⁺ (Table 3) . Therefore, the spoIIIGE156K allele fully restored sporulation to a spoIIIG null mutant .

 
Expression of sspE-lacZ in B . subtilis strains AH3786 and AH3787 was induced around 4 h after the onset of sporulation (Fig . 1), as in a Spo⁺ strain expressing the wild-type spoIIIG allele at its normal position (26; also data not shown), and reached maximum levels around 6 h after sporulation had started . As expected, expression of sspE-lacZ in strain AH3790 (spoIIIG spoIIIA::Tn917 sspE-lacZ amyE::spoIIIG) was severely impaired (19) . However, in the congenic strain AH3791, which bears the spoIIIGE156K allele at amyE, expression of sspE-lacZ was restored (Fig. 1) . Nevertheless, while in the spoIIIA⁺ strain AH3787, expression of sspE-lacZ was strongly induced around 4 h after the onset of sporulation, in the spoIIIA mutant strain AH3791 (spoIIIGE156K), ß-galactosidase accumulated at a reduced rate between 4 and 6 h after the onset of sporulation and reached maximum levels only around 10 h after sporulation had started (Fig . 1) . Also, even though the spoIIIGE156K allele restores ^G activity to spoIIIA cells, strain AH3791 was still unable to sporulate (Table 3) . We also found that the spoIIIGE156K allele restored sspE-lacZ expression (but not sporulation) to a spoIIIJ::km mutant (data not shown) .

 
To investigate whether the increased activity of ^GE156K in Spo⁺ cells or in the spoIIIA mutant relative to wild-type ^G could be attributed to its increased accumulation, we compared the levels of ^G and ^GE156K throughout sporulation by immunoblot analysis using a previously described anti-^G antibody (41) . We found that in agreement with the timing of sspE-lacZ expression, ^G or ^GE156K reached peak levels around 4 h after the onset of sporulation in Spo⁺ cells (Fig . 2A and B) . In a spoIIIA background, the accumulation of the wild-type form of ^G was delayed, reaching maximum levels around 6 h after the start of sporulation (Fig . 2C) . In spoIIIA::Tn917 spoIIIGE156K cells (AH3791), ^GE156K is detected only from 5 h on, and its accumulation reaches a maximum around 8 h after the onset of sporulation (Fig . 2D) . The late accumulation of ^GE156K in the spoIIIA mutant suggests that some ^GE156K (but not wild-type ^G) escapes inhibition late in sporulation and then amplifies its own synthesis . This late accumulation of ^GE156K in the spoIIIA::Tn917 spoIIIGE156K double mutant correlates with the late expression of sspE-lacZ (Fig . 1) . Note that under our electrophoretic conditions, the ^GE156K form migrates slightly faster than wild-type ^G (compare the mobility of ^G and ^GE156K relative to a background band labeled with an asterisk seen in a sample from a spoIIIG deletion mutant [Fig . 2A and B, for example]) . The levels of ^GE156K do not appear to be higher than those of wild-type ^G in either spoIIIA⁺ or spoIIIA mutant cells (Fig . 2, compare panels A and B and panels C and D) . Thus, the difference in expression of sspE-lacZ in spoIIIGE156K strains AH3787 (spoIIIA⁺) and AH3791 (spoIIIA mutant) relative to congenic strains expressing a wild-type spoIIIG gene at the amyE locus cannot be explained by an increase in the synthesis or stability of ^G . Rather, it may reflect increased activity of the sigma factor . In the case of AH3791 (spoIIIA::Tn917 spoIIIGE156K), the increased activity of ^GE156K is manifested only at a late time in development .

 
The E156K mutation is likely to interfere with the interaction between SpoIIAB and ^G. Like spoIIIGE156K, a previously described allele of spoIIIG bearing a glutamate-to-lysine substitution at position 155 (spoIIIGE155K) allows expression of sspE-lacZ in cells with the mutant spoIIIA or spoIIIJ gene (19, 41) . The E155K substitution was introduced in ^G, because a glutamic acid-to-lysine substitution at an equivalent position of ^F (E149K) was found in a genetic screen for ^F mutants with reduced affinity for SpoIIAB (5) . Moreover, under conditions in vitro that promote binding of SpoIIAB to ^F, the anti-sigma factor also binds to ^G, but not to ^GE155K (19) .

The binding of SpoIIAB to ^F or ^G can now be described in molecular terms, using the crystal structure (2) of the SpoIIAB-^F complex from B . stearothermophilus and the comparative models for the SpoIIAB-^F and SpoIIAB-^G complexes from B . subtilis derived here . These structures show that in B . stearothermophilus ^F, the residue (E147) equivalent to E149 in the ^F protein of B . subtilis, as well as three other residues found in genetic screens for mutants resistant to inhibition by SpoIIAB are located within a region that contains 17 amino acids (in B . stearothermophilus) found to interact with SpoIIAB (2, 5) (Fig . 3A to C) . Of the amino acids, 15 are either identical or homologous in ^G (compare Fig . 3, panels C and F) and 3 are uniquely conserved between ^F and ^G (2; also data not shown) .

 
The nature of the E155K and E156K mutations in ^G from B . subtilis can be qualitatively understood by examination of the model for the SpoIIAB-^G complex (Fig . 3D to F) . The model shows that residue E155 of ^G interacts with S17 in one of the SpoIIAB molecules present in the dimer (Fig . 3D to F and insert), a contact that is also conserved in the SpoIIAB-^F complex (E149) (Fig . 3A to C) . The models also predict that residue E156 of ^G contacts residue K41 in the same SpoIIAB molecule contacted by residue E155 of ^G (Fig . 3, insert), and again, this contact is conserved in the SpoIIAB-^F complex (D150) (Fig . 3A to C) . Therefore, both the E155K and E156K substitutions introduce unfavorable interactions expected to destabilize the interaction of SpoIIAB with ^G . We infer that the E156K mutation reduces the binding of SpoIIAB to ^G in a manner similar to that observed for ^GE155K (19) . However, the interaction of E155 with a serine residue (S17) from SpoIIAB is likely to be less strong than the interaction of E156 that forms a salt bridge with K41 from SpoIIAB (Fig . 3, insert) . Therefore, the effect of the E156K mutation, placing two positively charged residues in close proximity would create packing problems, and appears more disadvantageous for complex formation than the E155K mutation . Since the models for the two complexes show that most of the contacts between SpoIIAB and ^F or ^G are conserved, our analysis supports the conclusion of Evans et al . that the interaction of SpoIIAB with either ^F or ^G is very similar (7) .

^GE156K is less sensitive to SpoIIAB in vivo. To determine whether the ^GE156K form was less sensitive to the inhibitory action of SpoIIAB in vivo, we constructed strains engineered to coexpress spoIIAB and either spoIIIG or spoIIIGE156K during vegetative growth in a medium (LB) that does not support efficient sporulation . Strains AH2492 and AH2493 carry a fusion of the xylose-inducible P_xylA promoter to the spoIIIG and spoIIIGE156K genes, respectively, inserted at the amyE locus, as well as an IPTG-inducible P_spac-spoIIAB fusion inserted at the thrC locus; in addition, the two strains carry an sspE-lacZ fusion (Table 1) . Preliminary experiments revealed that expression of P_xylA-spoIIIG and P_xylA-spoIIIGE156K in the absence of xylose resulted in significant expression of sspE-lacZ (data not shown) . Therefore, AH2492 and AH2493 were grown in the absence of xylose and in the absence or presence of IPTG (1 mM) to induce SpoIIAB production .

Expression of spoIIIG or spoIIIGE156K in the presence or absence of SpoIIAB did not result in any detectable growth differences between the strains (Fig . 4A) . The sspE-lacZ-driven production of ß-galactosidase was monitored in the various cultures during the log phase of growth, 1.5, 2, and 2.5 h after inoculation . We found that in the absence of IPTG, the activities of both ^G and ^GE156K increased during the experiment, even though the activity of ^GE156K was always lower than that of the wild-type form (Fig . 4B) . In the presence of IPTG to induce spoIIAB expression, the activity of wild-type ^G was immediately reduced and remained at low levels (Fig. 4B) . In contrast, the activity of ^GE156K was reduced slowly (Fig . 4B) . Moreover, at all the times tested, the fraction of ^GE156K activity remaining after IPTG-induced SpoIIAB synthesis was higher than the fraction of ^G activity remaining after SpoIIAB induction (Fig . 4C) . These results are consistent with the suggestion that the E156K substitution makes ^G less sensitive to the inhibitory action of SpoIIAB .

 
Activity of ^GE156K in a spoIIIA background is delayed relative to completion of the engulfment process. On the basis of an analogy to the E155K mutation and the results discussed above, we expected that the E156K substitution would also relieve the inhibitory action of SpoIIAB on ^G during sporulation . If the interaction of SpoIIAB with ^G were reduced, as suggested by our screen, and because ^G is autoregulatory (17, 47), we would expect premature expression of sspE-lacZ if SpoIIAB were the primary inhibitor of ^G activity in the prespore . In contrast to this expectation, activity of ^GE156K was delayed in spoIIIA cells (Fig . 1), which remained unable to sporulate (AH3791) (Table 3) .

To determine whether expression of sspE-lacZ was still coupled to the completion of the engulfment process in strain AH3791 (spoIIIA spoIIIGE156K), we used the membrane stains FM4-64 and MTG to monitor completion of the engulfment process (44) . We stained samples of the same strains depicted in Fig . 1, and in parallel, we monitored accumulation of ß-galactosidase to control for the onset of ^G activity . We could not use a fusion of the sspE promoter to the gfp gene for this purpose, because even in the absence of ^G, most of the cells showed some prespore decoration, presumably due to the activity of ^F (data not shown) .

As in the experiment documented in Fig . 1, expression of sspE-lacZ in the Spo⁺ strains AH3786 (spoIIIG at amyE) and AH3787 (spoIIIGE156K at amyE) commenced around 4 h after the onset of sporulation, when 42 and 31% of the cells, respectively, had completed the engulfment process (Table 4) . Activity of ß-galactosidase peaked 6 h after the onset of sporulation for strain AH3786, when 48% of the cells showed complete engulfment of the prespore, and 6 to 8 h after the start of sporulation for AH3787, when 51 to 74% of the cells had completed the engulfment process (Table 4) . These observations are in agreement with the results of a previous study, suggesting that the activity of ^G coincides with the completion of engulfment when sporulation is induced by growth and resuspension in a minimal medium (29) . About 20% of AH3791 cells (spoIIIA::Tn917 spoIIIGE156K) showed complete engulfment by 4 h after the onset of sporulation, a fraction that increased to 39% by 6 h (Table 4), and by 8 h after sporulation had started, the number of cells showing clear signs of having completed the engulfment sequence reached a maximum of 54% (Table 4) . However, this did not correspond to peak levels of sspE-lacZ expression (Fig. 1) . Rather, enzyme production reached a maximum 10 h after the onset of sporulation, when the fraction of cells with clear signs of complete engulfment actually decreased to about 35% (Table 4) . The reasons for this decrease may reflect instability of the prespore in cells bearing a spoIIIA mutation (see below) . Consistent with this interpretation, 10 h after the onset of sporulation in strains bearing a mutation in spoIIIA, the pattern of fluorescence decoration resulting from MTG staining tended to change from the ellipsoidal contour of the prespore to a more or less indistinct mass of fluorescence, suggesting coalescence of the prespore membranes (data not shown) . We note that a spoIIIA mutation per se does not significantly interfere with the timing of engulfment (AH3790) (Table 4) . We interpret these observations as indicating that ^GE156K becomes active only about 2 h after the completion of engulfment in a spoIIIA mutant .

 
Activity of ^GE156K in a spoIIIA background is not confined to the prespore. The observation that the activity of ^GE156K was delayed relative to the completion of the engulfment process in a spoIIIA mutant led us to examine the location of ß-galactosidase produced from the sspE-lacZ fusion in this strain . We used immunofluorescence microscopy to examine samples of the same cultures used in the experiment depicted in Fig. 1 6 and 10 h after the onset of sporulation, as this corresponds to peak levels of sspE-lacZ expression in spoIIIA⁺ or spoIIIA mutant cells, respectively (Fig . 1) . We were unable to collect reasonable phase-contrast images or images of cells in which the membrane had been stained with FM4-64 or MTG after fixation and permeabilization of the cells with lysozyme (also see reference 33) . For that reason, the number of sporulating cells was scored on the basis of the analysis of the pattern of nucleoid staining by DAPI (see Materials and Methods), and for each of these cells, the pattern of ß-galactosidase localization was recorded (see Materials and Methods) (Table 5) .

 
In strains AH3786 (spoIIIG sspE-lacZ amyE::spoIIIG) and AH3787 (spoIIIG sspE-lacZ amyE::spoIIIGE156K), production of ß-galactosidase was detected only in cells in which the prespore had been completely engulfed by the mother cell (as defined by DAPI staining, which reveals two equal-size mother cell and prespore chromosomes) . Production of the enzyme was always confined to the prespore compartment of strain AH3786 (Fig . 5a to c and Table 5), whereas for strain AH3787, a small percentage of cells (around 1 or 2%) 6 or 10 h after the onset of sporulation showed fluorescence throughout the entire cell (Fig. 5f and Table 5) . Interestingly, these specimens did not present any distinctive signs of sporulation as judged from the pattern of DAPI staining (Fig . 5d to f) . This effect does not seem to be caused by deficient staining of one of the chromosomes, since the fluorescence signal is distributed by what would be the length of the entire sporulating cell (mother cell plus prespore) . These specimens may represent vegetative cells or cells in which the normal compartmentalization of ^G activity was lost (see below) .

 
In agreement with the results of sspE-lacZ expression shown in Fig . 1, essentially no ß-galactosidase could be detected in cells of AH3790 (spoIIIG spoIIIA::Tn917 sspE-lacZ amyE::spoIIIG) (Fig . 5i shows the only specimen found with signs of fluorescence in the prespore) (Table 5) . About 3% of the cells of AH3791 (spoIIIG spoIIIA::Tn917 sspE-lacZ amyE::spoIIIGE156K) at 6 h after the onset of sporulation showed accumulation of ß-galactosidase, a percentage that increased to 23% at 10 h (Table 5) . Surprisingly, in most of these cells, ß-galactosidase was found to localize throughout the entire cell (Fig . 5j to l and Table 5) . The percentage of sporulating cells of AH3791 (two visible nucleoids of about the same size at this stage) decreased from 75% at 6 h after the onset of sporulation to about 32% at 10 h (Fig . 5j to l [6 h] and m to o [10 h] and Table 5) . Note that the specimens showing whole-cell fluorescence do not show the DAPI staining pattern of a mid to late stage of sporulation and that specimens in which the prespore can be clearly distinguished do not show sspE-lacZ expression (Fig . 5j to o) . The percentage of AH3790 cells (wild-type spoIIIG in a spoIIIA background) with two distinct nucleoids also decreased from 6 h (78%) to 10 h after the onset of sporulation (60%) (Table 5) except that in the latter case, the whole-cell pattern of decoration was never found . No decrease in the percentage of sporulating cells was noticed for the spoIIIA⁺ strains .

The results show that ^G156K does not become active exclusively in the prespore of a spoIIIA mutant . The results suggest that spoIIIA may function to antagonize an as yet unknown negative regulator of ^G following completion of the engulfment process and that spoIIIA may serve an additional function in sporulation related to the maintenance of compartmentalized gene expression in postengulfment cells . Essentially the same observations, i.e., absence of prespore-specific expression of sspE-lacZ, were made for spoIIIA cells harboring the E155K mutation (19; E . M . Kellner and C . P . Moran, Jr., unpublished results), reinforcing the view that the E155K and E156K substitutions affect the activity of ^G similarly .

Early expression of spoIIIGE156K in the prespore does not result in premature activity of ^G. Transcription of the spoIIIG gene by ^G is delayed by an unknown mechanism towards the end of the engulfment process relative to the transcription of the first class of ^F-dependent genes, which includes the spoIIQ gene (24) . We fused the coding region of spoIIIGE156K to the early ^F-dependent spoIIQ promoter, reasoning that the fusion allele would bypass both the mechanism that delays transcription of spoIIIG and a possible negative effect of SpoIIAB prior to engulfment . The promoter fusion was introduced at the amyE locus, producing strain AH3789 (spoIIIG sspE-lacZ P_spoIIQ-spoIIIGE156K) . The spoIIQ promoter was also fused to the coding region of the wild-type spoIIIG gene, and the fusion was inserted at amyE to produce AH3788 (spoIIIG sspE-lacZ P_spoIIQ-spoIIIG) . Both AH3789 and AH3788 sporulate efficiently (Table 3) .

In agreement with unpublished work cited by Stragier and Losick (46), the expression of sspE-lacZ in AH3788 began to increase around 4 h after the onset of sporulation, as in strains bearing the wild-type or spoIIIGE156K allele under the control of its native promoter (AH3786 and AH3787) (Fig . 6) . Moreover, induction of sspE-lacZ also occurred around 4 h of sporulation in strain AH3789 (Fig . 6); the spoIIQ promoter drives expression of spoIIIGE156K in AH3789 .

 
When produced from their own promoter, ^G and ^GE156K were first detected 3 h after the onset of sporulation, and the level of the ^G factor increased until 4 h (Fig . 2A and B), whereas utilization of the spoIIQ promoter permitted the accumulation of ^G or ^GE156K from 2 h on, with maximum levels between 3 and 4 h after the onset of sporulation (Fig . 2E and F) . In all the strains included in the experiment of Fig . 6, induction of sspE-lacZ expression 4 h after the onset of sporulation coincided with completion of the engulfment process (as assayed by FM4-64 or MTG staining) (between 31 and 56% [data not shown]) . Both strains bearing the wild-type or spoIIIGE156K allele under the control of the spoIIQ promoter presented a high background of ^G activity starting at the onset of sporulation . The reason for this behavior is not known, but it could be the result of high levels of expression of spoIIIG or spoIIIGE156K from the strong spoIIQ promoter (24) . In any event, we note that expression of spoIIIGE156K does not result in a higher background relative to expression of the wild-type spoIIIG gene (Fig . 6) . Moreover, we note that in both AH3788 and AH3789, expression of sspE-lacZ remains constant (Fig . 6), while the cellular levels of ^G increase (Fig . 2E and F) . It appears that even though ^GE156K accumulates starting 2 h after the onset of sporulation, it directs induction of sspE-lacZ expression only around 4 h, when the process of engulfment of the prespore by the mother cell is complete . Together, the results suggest that SpoIIAB may not contribute decisively to the inhibition of ^G activity in the prespore . The fact that very little free SpoIIAB seems to accumulate in the prespore and the similarity of the interaction of SpoIIAB with both ^F and ^G (7, 28; this work) are consistent with this interpretation .

Activity of ^G in the mother cell is antagonized by SpoIIAB and LonA. Since ^G is autoregulatory and SpoIIAB is present in both the prespore and mother cell compartments, we wanted to determine whether SpoIIAB could have a role in the negative regulation of ^G in the mother cell, as previously suggested (3, 8, 19, 21, 35) . To investigate this possibility, we fused the coding regions of the wild-type and spoIIIGE156K alleles to the mother cell-specific, ^E-dependent spoIID promoter (36) . The fusions were transferred to the amyE locus of a strain bearing an in-frame deletion of the spoIIIG gene and an sspE-lacZ fusion (AH2452), yielding strains AH2460 (spoIIIG sspE-lacZ P_spoIID-spoIIIG) and AH2461 (spoIIIG sspE-lacZ P_spoIID-spoIIIGE156K) (Table 1) . No ^G activity was detected by monitoring sspE-lacZ-driven ß-galactosidase production, when the wild-type spoIIIG allele was expressed in the mother cell (Fig . 7) . In contrast, expression of the spoIIIGE156K allele promptly resulted in sspE-lacZ expression, which occurred prior to normal expression of sspE in the prespore, in agreement with the timing of utilization of the spoIID and sspE promoters during sporulation (26, 36) (Fig . 7) . This observation supports a role for SpoIIAB in the regulation of ^G activity in the mother cell .

 
To determine whether the elevated levels of ^G activity observed in strain AH2461 (P_spoIID-spoIIIGE156K) relative to strain AH2460 (P_spoIID-spoIIIG) correlated with increased accumulation of ^G, we conducted immunoblot experiments . We found that the ^GE156K protein accumulated starting 2 h after the onset of sporulation, reaching maximum levels around 3 h after sporulation had begun (Fig . 8B), which is in accordance with the temporal pattern of expression of a spoIID-lacZ fusion (36) . In contrast, the wild-type form of ^G was detected only in trace amounts (Fig . 8A), suggesting that ^G is subjected to proteolysis in the mother cell .