A threshold mechanism governing activation of the developmental regulatory protein sigma F in Bacillus subtilis.

The RNA polymerase sigma factor sigma(F) is a developmental regulatory protein that is activated in a cell-specific manner following the formation of the polar septum during the process of spore formation in the bacterium Bacillus subtilis. Activation of sigma(F) depends on the membrane-bound phosphatase SpoIIE, which localizes to the septum, and on the formation of the polar septum itself. SpoIIE is responsible for dephosphorylating and thereby activating the phosphoprotein SpoIIAA, which, in turn, triggers the release of sigma(F) from the anti-sigma(F) factor SpoIIAB. Paradoxically, however, the presence of unphosphorylated SpoIIAA is insufficient to cause sigma(F) activation as SpoIIAA reaches substantial levels in mutants blocked in polar septation. We now describe mutants of SpoIIE, SpoIIAA, and SpoIIAB that break the dependence of sigma(F) activation on polar division. Analysis of these mutants indicates that unphosphorylated SpoIIAA must reach a threshold concentration in order to trigger the release of sigma(F) from SpoIIAB. Evidence is presented that this threshold is created by the action of SpoIIAB, which can form an alternative, long lived complex with SpoIIAA. We propose that formation of the SpoIIAA-SpoIIAB complex serves as a sink that traps SpoIIAA in an inactive state and that only when unphosphorylated SpoIIAA is in excess to the sink does activation of sigma(F) take place.

Cellular differentiation often depends on the establishment of differential gene expression between two progeny cells following asymmetric cell division. An attractive developmental system for studying the establishment of differential gene expression is endospore formation in the bacterium Bacillus subtilis (1). Endospore formation involves the formation of an asymmetrically positioned septum that divides the developing cell (the sporangium) into unequally sized progeny called the forespore (the smaller cell) and the mother cell. The earliest event in establishing differential gene expression is the activation in the forespore of the transcription factor F , a member of the RNA polymerase sigma factor family of prokaryotic transcription factors (2)(3)(4)(5). A key to the rapid activation of F following polar division is that the transcription factor is syn-thesized in the predivisional sporangium but is held in an inactive complex by the anti-F factor SpoIIAB (6 -9). Once the polar septum is complete, F is released from the SpoIIAB-F complex by the action of the anti-anti-F factor SpoIIAA. SpoIIAA reacts with the SpoIIAB-F complex to cause the discharge of the free and active F (8 -13). SpoIIAA is a phosphoprotein that exists in an inactive, phosphorylated state (SpoIIAA-P) and an active, unphosphorylated state (10,11,14). The conversion of SpoIIAA to SpoIIAA-P is mediated by SpoIIAB, which in addition to being an anti-F factor is a protein kinase that phosphorylates SpoIIAA on serine residue 58 (9,15). Dephosphorylation is accomplished by SpoIIE, a membrane-bound protein phosphatase (16 -18). For simplicity, we henceforth refer to SpoIIAA and SpoIIAA-P as AA and AA-P, respectively, and SpoIIAB as AB.
SpoIIE is an 827-residue-long integral membrane protein with 10 membrane-spanning segments in its N-terminal region (region I) (19), a region (region III) that is homologous to the PP2C family of eukaryotic phosphatases near its C terminus (20), and a 250-amino acid-long central region (region II) that exhibits little or no similarity to other nonorthologous proteins in the data bases. Evidence indicates that region II mediates interactions between SpoIIE molecules and with the cytokinetic protein FtsZ (21) and also that region II contributes significantly to the activity of F (22)(23)(24)(25). The SpoIIE phosphatase localizes to the polar septum (26), and evidence indicates that this localization contributes to, but is not exclusively responsible for, the forespore-specific activation of F (19). Also contributing to the cell-specific activation of F is the delayed entry of the gene (spoIIAB) for AB into the forespore, which is believed to result in the partial depletion of the anti-F factor in the forespore due to the proteolytic instability of the protein (27,28).
The activation of F is known to be almost entirely dependent upon the formation of the polar septum (29 -31). However, dephosphorylation of AA-P is not strongly dependent upon septum formation. Mutants blocked in asymmetric division accumulate substantial levels of unphosphorylated AA yet fail to activate F (24,32,33). These observations led to the view that unphosphorylated AA must reach a threshold concentration before it can activate F (25,32), but the molecular basis for such a threshold has been mysterious.
Here we report on the characterization of amino acid substitutions of SpoIIE that block F activation. We show that certain substitutions lying in region II and hence outside the phosphatase domain of SpoIIE nevertheless strongly impair dephosphorylation of AA-P in vivo. We show that the effects of these substitutions can be reversed by other amino acid changes in SpoIIE, an amino acid change in the catalytic center of the AB kinase, or an amino acid substitution in AA. We show that all of these suppressors act by increasing the level of unphosphorylated AA in the cell. Importantly, amino acid substitutions that greatly increase the level of unphosphorylated AA result in the activation of F in a manner that does not depend upon septum formation, a finding consistent with the idea that a threshold level of unphosphorylated AA must be reached in order for F to be activated. Finally, we present evidence consistent with the idea that this threshold is established by the formation of an alternative complex between AA and AB that traps the anti-anti-F factor in an inactive state. Formation of the AA-AB complex could act as a buffer to prevent activation of F prior to the formation of the polar septum and in the mother cell after asymmetric division.
A spoIIE-linked kanamycin resistance marker was introduced into RL2226 via Campbell integration of pSA56 (gift from Scott Alper). pSA56 contains the 3Ј end of spoIIE (nucleotides 2146 -2492 from the start of the coding sequence) in pUK19 (a derivative of the cloning vector pUC19 into which a kanamycin resistance cassette was introduced). The wild type spoIIE gene and mutants of spoIIE were introduced at the amyE locus via double crossover recombination of pKC2, and its derivatives were linearized by XhoI digestion. The entire spoIIE coding sequence and upstream regulatory region were obtained by HincII/BamHI digestion of pNK1 (gift from Nicole King). (pNK1 was constructed by insertion of spoIIE DNA extending from 400 bp upstream of the start codon (to the chromosomal HincII site) to 11 bp downstream from the stop codon into pUK19.) The spoIIE-containing, HincII-BamHI fragment was cloned into pLD30 (36) that had been cut with EcoRV and BamHI to create pKC2. spoIIE mutations were introduced into pKC2 with appropriately designed oligonucleotides and the QuikChange TM site-directed mutagenesis kit (Stratagene): pPE27 (Q483A), pKC44 (V697A), and pKC45 (Q342P).
A wild type spoIIE-gfp fusion was constructed via Campbell integra- Site-directed mutagenesis of pLD28 was used to introduce the R105C mutation into the coding sequence of spoIIAB, creating pKC59. Wild type spoIIAA and spoIIAB-R105C were introduced together at the amyE locus via double crossover recombination of pKC67 linearized with XhoI digestion. The promoter region and entire coding sequence of spoIIAA and spoIIAB-R105C was PCR-amplified from KC332 genomic DNA with oligonucleotides that added a HindIII site to the 5Ј end of the PCR fragment and a BamHI site the 3Ј end. The PCR fragment was digested with HindIII and BamHI and ligated into pDG364 (37) cut with the same enzymes to create pKC67. amyE::P Xyl -spoIIAA-S58A cat was integrated into the chromosome via double crossover recombination of pKC24. pKC24 was created by amplifying the entire coding sequence for spoIIAA-S58A from pLD23 (14) with a forward primer (5Ј-GGCAAGCTTGGGACATAAGGAGGA-ACTACTAYGAGCCTTGGAATTGAC) that added a HindIII restriction site and an ideal ribosome binding site (38) to the 5Ј end of the gene and a reverse primer that added a BamHI site just 3Ј to the stop codon. The PCR fragment was digested by HindIII and BamHI and cloned into pOR277⌬Sal (39) cut with the same enzymes.
The spectinomycin resistance marker of strains containing amyE::pPE27 (spoIIE-Q483A spc) was changed to tetracycline resistance via double crossover recombination with pPE19. pPE19 is a derivative of pDG1514 (40) containing the 5Ј end of spoIIE (400 nucleotides upstream to 1068 nucleotides downstream from the start of the coding sequence) on the 5Ј side of the tetracycline resistance gene, in the opposite orientation. The tetracycline resistance gene in pPE19 is flanked on the 3Ј end by the 3Ј end of the amyE gene ("amy back") derived from the fragment released by EcoRV/BglII digestion of pDG1662 (41).
An in-frame deletion of spoIIAA linked to the spectinomycin resistance marker was obtained by transformation of linearized pNK7 (gift from Nicole King) into PY79 to create RL2218. pNK7 was created by ligating the ScaI/BspDI fragment containing an in-frame deletion of spoIIAA from pLD28 (36) into pLD19 (14) cut with the same enzymes.
Selection for Suppressors of spoIIE-S361F and spoIIE-Q483A-The oligosporogenous strains KC308 or KC29 were inoculated into 100 ml of DS medium (42), and cultures were incubated with shaking in a 37°C water bath for 30 h to induce sporulation by nutrient exhaustion. Thirty ml of the culture were incubated in an 80°C water bath for 20 min to kill cells that failed to develop into mature spores. The heat-treated culture was diluted and plated to determine the number of heat-resistant spores/ml, and ϳ100 of the colonies that arose from surviving spores were patched to DS agar plates containing 80 g/ml X-gal 1 to assay for F -dependent lacZ expression. The heat-treated culture was diluted into 300 ml of fresh DS medium and incubated in a 37°C shaking water bath for 30 h. The heat kill, plating, patching, and dilution procedure was repeated for several rounds until the colonies from the surviving spores exhibited F -dependent lacZ activity on DS agar plates containing X-gal and bred true as sporulation-proficient strains. This took two rounds of enrichment for KC308, which produces ϳ10 4 spores/ml (versus ϳ5 ϫ 10 8 spores/ml for the wild type), and five rounds of enrichment for KC29, which produces ϳ2 ϫ 10 7 spores/ml. Genomic DNA was prepared from several Spo ϩ suppressor mutants, and the linkage of the mutation to the antibiotic resistance gene associated with either spoIIE or the spoIIA operon were determined by backcrossing into RL2220 for mutations in KC308 and RL2114 for mutations in KC29. The mutated genes were then amplified from the genomic DNA by PCR and sequenced to identify the nucleotide changes. Two independent selections were carried out for each strain. Four mutants were sequenced in the first selection of KC308 suppressors, and 12 were sequenced from the second. The same three mutations in spoIIE (V697A, Q342P, and R502L) were the only mutants identified in both selections. Four mutants were sequenced from the first selection of KC29 suppressors, and 11 were sequenced from the second. Each mutation sequenced was unique with the exception of spoIIE K649T , which was present in two independent colonies, both isolated in the same selection. It is unknown if these two colonies were siblings or mutated independently.
Sporulation Conditions Used for Monitoring ␤-Galactosidase Activity and Carrying Out Immunoblotting Experiments-Cells were induced to sporulate by the resuspension method (43). All cultures were grown and sporulated at 37°C except in experiments with strains harboring the div355 allele. For these experiments, all strains were grown in culture medium at the permissive temperature (28°C) to an A 600 of ϳ0.6 and shifted to the restrictive temperature (37°C) upon resuspension in sporulation salts. One-ml aliquots of each culture were harvested by centrifugation at different time points after the start of sporulation, and the cell pellets were stored at Ϫ70°C until further analysis. ␤-Galactosidase activity was assayed as described by Harwood and Cutting (35).

Region II of SpoIIE Influences Phosphatase Activity-SpoIIE
consists of three domains, an N-terminal region (I) that is composed of multiple membrane-spanning segments (19), a C-terminal region (III) that corresponds to the catalytic center (20,21,44), and a central region (II) that affects F activity by an unknown mechanism (22-25) (see Fig. 2). A classic mutation of spoIIE, spoIIE48, which blocks F activation, causes a phenylalanine substitution at serine residue 361 in region II (22). To investigate the role of region II and dephosphorylated AA in the activation of F , we employed an improved isoelectric focusing protocol for resolving AA and AA-P in extracts of sporulating cells and for detecting the proteins by immunoblotting with anti-AA antibodies (see "Experimental Procedures"). We determined that the strength of the signal obtained by immunoblotting was directly proportional to the amount of protein applied to the isoelectric focusing gel and that AA and AA-P reacted with approximately equal efficiencies to the antibodies (data not shown). We have also found that AA-P is subject to SpoIIE-dependent dephosphorylation during protein extraction and that this can be prevented by the use of the general phosphatase inhibitor sodium fluoride (NaF) (data not shown).
Taking advantage of our improved isoelectric focusing system, we reinvestigated the effect of the S361F substitution on the levels of AA and AA-P. The results show that the mutation had a pronounced effect in inhibiting dephosphorylation of AA-P (see Fig. 4, B and C, lanes 3 and 4), which had not been detected in earlier work in which cells extracts were prepared in the absence of NaF. Evidently, then, region II contributes importantly to the level of phosphatase activity although it lies outside of the catalytic domain. This effect is not due to an impaired capacity of the mutant protein to properly localize, since a fusion of SpoIIE-S361F to GFP is known to exhibit a normal pattern of subcellular localization to the polar septum (24).
In light of the results with S361F, we sought to identify other residues in region II that were important for F activation and to investigate whether they were important for the phosphatase activity of SpoIIE. Our strategy was to create alanine substitutions at residues in region II that were highly conserved as judged by comparison with the amino acid sequence of SpoIIE orthologs from other endospore-forming species (Fig.  1). We focused on three of the four conserved cysteine residues (Cys 402 , Cys 408 , and Cys 446 ), the conserved arginine at position 477, and the conserved glutamine at position 483. Substituting alanine or serine for cysteine at residue 402, 408, or 446 did not impair sporulation or F activity. Substituting arginine 477 with alanine or glutamate also did not affect sporulation or F activity (data not shown). However, substituting glutamine 483 with alanine severely impaired sporulation and F activation. Sporulating cells of a spoIIE-Q483A mutant exhibited little or no F activity (see Fig. 3A) and produced heat-resistant spores at an efficiency of 0.005% that of the wild type strain.
Next, we investigated the capacity of cells producing the SpoIIE Q483A mutant phosphatase to dephosphorylate AA-P.
The results show that, as in the case of SpoIIE S361F , the ratio of AA to AA-P was markedly lower in the SpoIIE Q483A mutant than in the wild type (see Fig. 4A, lanes 3 and 4). Again, like SpoIIE S361F -GFP, SpoIIE Q483A -GFP exhibited a normal pattern of subcellular localization to the polar septum (data not shown). The simplest interpretation of these results is that region II contributes significantly to the phosphatase activity of SpoIIE and that both the S361F and Q483A amino acid substitution mutants are impaired in dephosphorylating AA-P.
Additional support for the idea that region II contributes to the phosphatase activity of SpoIIE came from experiments in which we engineered cells to produce a truncated form of SpoIIE lacking region II (an in-frame deletion of codons 324 -584) or consisting entirely of region III (the catalytic domain) alone (residues 585-827). The results showed that sporulating cells producing the truncated proteins exhibited a markedly lower ratio of AA to AA-P than did cells producing wild type SpoIIE (data not shown). Western blot analysis indicated that the truncated proteins accumulated to only modestly lower levels than the full-length protein (data not shown).
Suppressors of Mutations in the Coding Sequence for Region II-To investigate the role of region II further, we sought to identify suppressor mutations that would restore the capacity of cells producing SpoIIE S361F or SpoIIE Q483A to sporulate. Because cells producing the mutant proteins were impaired in sporulation, suppressor mutations could be conveniently obtained by selection for the production of heat-resistant spores (see "Experimental Procedures"). In total, 15 mutants were The asterisks indicate residues that did not cause a sporulation phenotype when changed to alanine. Residue 361 is the site of a Ser to Phe substitution caused by the classic sporulation mutation spoIIE48, and residue 483 is the site of a Gln to Ala switch that caused a strong sporulation defect (arrowheads obtained that proved to contain distinct and unique amino acid substitutions. (Two additional mutants that were derived from SpoIIE S361F turned out to contain a simple revertant that restored serine to position 361 and a pseudorevertant that introduced a cysteine at position 361.) All 15 mutants had regained the capacity to sporulate with efficiencies similar to that of the wild type. Furthermore, in all cases, F activity, as judged by use of lacZ fused to a gene (spoIIQ) under F control, was restored to approximately wild type levels (see Fig. 3, A-C). Genetic analysis revealed that 13 of the mutants harbored intragenic suppressor mutations in spoIIE itself (Fig. 2) and that the remaining two mutants harbored extragenic suppressors. The extragenic suppressors were located in the genes for AB (spoIIAB-R105C) and AA (spoIIAA-Q73R).
We rebuilt seven of the 13 intragenic suppressor mutations in an otherwise wild type copy of spoIIE and found that five of the seven exhibited no phenotype on their own. In other words, in these five cases, an effect of the amino acid substitution could only be detected in the presence of the original S361F or Q483A substitution. The remaining two mutations, spoIIE-V697A and spoIIE-Q342P, caused conspicuous phenotypes on their own, resulting in premature and high level activation of F and impaired sporulation (Fig. 3D). Likewise, we examined the effect of the two extragenic suppressors, spoIIAB-R105C and spoIIAA-Q73R, in cells that were wild type for spoIIE. Both of these mutations caused F to be activated prematurely and at a high level (Fig. 3, E and F). Hereafter, we restrict our discussion to spoIIE-V697A and spoIIE-Q342P and the two extragenic suppressor mutations, spoIIAB-R105C and spoIIAA-Q73R.
The spoIIE-Q342P and spoIIE-V697A mutations caused amino acid substitutions in regions II (Q342P) and III (V697A) of SpoIIE, respectively. The V697A substitution was located near the catalytic center of the enzyme as judged by projecting region III onto the crystal structure of the homologous PP2C protein from humans (45) (not shown). The spoIIAB-R105C mutation caused a substitution in a residue near the nucleotide binding pocket of the AB kinase that is positioned close to the ␥-phosphate of ATP (46). Finally, the spoIIAA-Q73R mutation replaced a surface-exposed residue located at the opposite end of the ␣-helix (and at the opposite end of the molecule) from the site (serine 58) at which phosphorylation of AA takes place (47).
The Suppressor Mutations Increase the Cellular Ratio of AA to AA-P-What is the basis for the premature and excessive activation of F seen in the four mutants? Using a fusion to GFP and staining with the vital membrane dye FM4-64, we determined that SpoIIE V697A and SpoIIE Q342P localized to the polar septum and supported polar septation in a normal manner (data not shown). We also determined that AB R105C and AA Q73R activated F in a SpoIIE-dependent manner (i.e. activation of F in cells producing AB R105C or AA Q73R was almost completely blocked by a null mutation of spoIIE (Fig. 3, B and C). These findings established that the unregulated F activity seen in the AB R105C and AA Q73R mutants was not a consequence of an inability of the mutant AB protein to bind to and inhibit F or an inability of the mutant AA protein to be phosphorylated and inactivated by the AB kinase.
Next, we investigated the relative levels of AA and AA-P in mutant cells at 60 and 90 min after the start of sporulation. The results show that all four mutants exhibited a higher ratio of AA to AA-P than seen in the wild type. Indeed, in the case of the SpoIIE V697A , SpoIIE Q342P , and AB R105C mutants, almost all of the AA was in the unphosphorylated form (Fig. 4, A-C).
The simplest interpretation of the foregoing results is as follows. The original region II mutants, SpoIIE S361F and SpoIIE Q483A , are unable to activate F because they are impaired in phosphatase activity. Conversely, the SpoIIE V697A and SpoIIE Q342P mutants cause premature and excessive F activation because they have greater than wild type levels of phosphatase activity. In the case of the doubly mutant SpoIIE Q483A,V697A and SpoIIE Q483A,Q342P proteins, the inhibitory effect of the Q483A substitution is balanced by the enhancing effect of the V697A or Q342P substitutions, resulting in approximately normal phosphatase activity (Fig. 4A, lane 9) and normal sporulation. Likewise, we interpret the effect of AB R105C as being due to impaired kinase activity. Indeed, in other work we have shown that the K m for the kinase activity of AB R105C was 2 orders of magnitude higher than that for wild type AB 2 (13). Also, as mentioned above, residue 105 lies in the catalytic center of the kinase. Thus, in cells producing SpoIIE S361F and AB R105C , the impaired phosphatase activity of the former is approximately balanced by the decreased capacity of the latter to phosphorylate AA. Finally, we suppose that AA Q73R is either more efficient in its phosphorylated form than wild type AA-P as a substrate for the SpoIIE phosphatase or a less efficient substrate in its unphosphorylated form than the wild type for the kinase activity of AB. If our interpretations of the behavior of the mutant proteins are correct, then the timing and level of F activity are evidently highly sensitive to the extent of dephosphorylation of AA-P.
Suppressor Mutations Bypass the Septation Requirement for F Activation-Activation of F depends on the formation of the polar division septum. Mutations in several genes (e.g. ftsZ, divIC, ftsL, and ftsA) that block polar septation also preclude F activation (29 -31, 33). We hypothesized that the four mutants discussed above bypassed the septation requirement for F activation based on the observation that in strains producing the mutant SpoIIE, AA, or AB proteins, F -directed synthesis of ␤-galactosidase commenced before the time that most cells had formed a polar septum (Fig. 3, D-F). To test this hypothesis directly, we assayed for F -directed lacZ expression in strains producing SpoIIE V697A , SpoIIE Q342P , AB R105C , or AA Q73R and harboring a temperature-sensitive allele (div355) of the division gene divIC. When shifted to a restrictive temperature (37°C), sporulating cells of a div355 mutant allow for the formation of a cytokinetic ring composed of the FtsZ protein and an associated ring of SpoIIE but fail to complete polar septation or to activate F (30). As expected, the SpoIIE V697A -, SpoIIE Q342P -, AB R105C -, or AA Q73R -producing derivatives of a div355 mutant remained aseptate but were unimpaired in F activation, allowing ␤-galactosidase to accumulate to high levels (Fig. 3, G-I). Analysis by isoelectric focusing confirmed that almost all of the AA protein in the spoIIE-V697A, div355 double mutant was in its unphosphorylated form (Fig. 5A, lanes  7-9).
We also investigated the ability of our suppressor mutations to cause F activation in a mutant of the cell division protein FtsA, FtsA D265G . The FtsA D265G mutant protein has little effect on medial division during growth but impairs polar division and F activation during sporulation (33). The results showed that F activity and accumulation of dephosphorylated AA reached high levels in double mutants harboring ftsA-D265G and the suppressor mutations spoIIE-V697A or spoIIE-Q342P (data not shown). Independent evidence that SpoIIE V697A bypassed the septation requirement for the activation of F came from the use of a fusion of the gene for GFP to a promoter under the control of F . Whereas F -directed gene expression is normally confined to the forespore compartment of the postdivisional sporangium, a strong fluorescence signal from GFP was observed in predivisional sporangia of cells producing the SpoIIE V697A mutant protein (data not shown). Indeed, cells producing SpoIIE V697A were blocked in polar septation, a phenotype characteristic of abnormally high levels of F activity. (Deletion of the gene encoding F restored polar septation to cells producing SpoIIE V697A , a finding that confirms that the block in polar septation was an indirect effect due to heightened level of F activity.) We considered two models for how the SpoIIE V697A , SpoIIE Q342P , AB R105C , and AA Q73R mutant proteins could break the normal dependence of F activation on polar division. In one model, the mutant proteins act simply by raising the level of unphosphorylated AA, which is known to induce the release of F from its inhibitor AB. In the other model, the mutant proteins exert their effect both by raising AA levels and by lowering the level of AA-P. In this second model, we assume that AA-P also contributes to the regulation of F and does so by acting as a previously unrecognized inhibitor of the transcription factor. To investigate this possibility, we took advantage of the observation that a mutant form of AA, in which the residue at which phosphorylation takes place (serine 58) is replaced with threonine, is readily subject to phosphorylation by AB, but the resulting phosphorylated form of the mutant protein, AA S58T -P, is refractory to dephosphorylation by SpoIIE (16,17). Accordingly, by building a merodiploid strain that produced both wild type AA and AA S58T , we were able to maintain a high level of the phosphorylated protein (in the form of AA S58T -P) while still allowing unphosphorylated AA to accumulate to normal levels (data not shown). If phosphorylated AA acts as an inhibitor of F , then a strain that produces both AA and AA S58T should exhibit less F activity than the wild type. The results show that the levels of F activity were similar in the presence or in the absence of the mutant protein (Fig. 6). The simplest interpretation of our results is that AA-P is not an inhibitor of F . (Alternatively, AA-P is an inhibitor of F , but this inhibitory effect is eliminated by the S58T substitution.) If, in fact, AA-P is not an inhibitor of F , then it would appear that SpoIIE V697A , SpoIIE Q342P , AB R105C , and AA Q73R break the dependence of F activation on septation by raising the level of unphosphorylated AA. Activation of F Requires a Threshold Level of Unphosphorylated AA-As reported previously and confirmed here, the overall level of dephosphorylation of AA-P in the sporangium is not appreciably dependent upon the formation of the polar septum (24,32,33). For example, as shown in Fig. 5, the ratio of AA to AA-P was no lower in a div355 or a ftsA-D265G mutant than in the wild type. However, both mutants were dramatically impaired in F activation as compared with the wild type (Fig. 5B). A key finding of the present investigation is that F activity is restored to division mutants such as div355 and ftsA-D265G by mutations that appreciably increase the ratio of AA to AA-P. In toto, these observations lead us to the view that the mere presence of unphosphorylated AA, even at AA/AA-P ratios up to 1.5:1, are insufficient to turn on F -directed gene expression. Rather, activation of F requires a high, threshold level of unphosphorylated AA, and this threshold is normally reached only in the forespore. If we are correct in positing the existence of a threshold level of unphosphorylated AA, then what is the molecular mechanism that is responsible for preventing low levels of AA from triggering the activation of F ?
Excess AB R105C Blocks F Activation-An attractive mechanism for creating a threshold is complex formation with AB. It is known that AA is capable of binding to AB in a manner that depends on the adenosine nucleotide ADP to form a long lived AA-AB(ADP) complex, trapping AA in such a way that it is unable to react with, and thereby displace F from, the AB(ATP)-F complex (10,11,48). According to this model, AB(ADP) is a sink for sequestering unphosphorylated AA. Such a sink would buffer the effect of the SpoIIE phosphatase, preventing activation of F until and unless enough unphosphorylated AA accumulates (normally only in the forespore) to exceed the capacity of the sink. Only at levels of AA in excess to AB is the unphosphorylated protein free and available to cause the activation of F .
The model that AB serves as a sink for unphosphorylated AA explains striking and counterintuitive observations concerning the behavior of the AB kinase mutant AB R105C as we now explain. As we have seen, cells producing AB R105C generate high levels of unphosphorylated AA and, as expected, exhibit high levels of F -directed gene expression (about 20-fold greater than the wild type; Figs. 3E and 7A). We discovered, however, that when the gene for AB R105C was present in two copies, the level of F activity was dramatically diminished. The introduction of a second copy of the gene for AB R105C lowered F activity ϳ400-fold. Indeed, rather than exhibiting higher than wild-type levels of F activity, cells harboring two copies of the mutant gene had about 20-fold lower F activity than did a strain harboring two copies of the wild type gene (Fig. 7A). Isoelectric focusing analysis confirmed that AA was almost entirely in its unphosphorylated form in cells bearing two copies of the gene for AB R105C (Fig. 7B). A merodiploid strain that harbored the wild type gene for AB as well as the gene for AB R105C was also strongly impaired in F activity despite the presence of high levels of unphosphorylated AA (Fig. 7). Western blot analysis confirmed that the mutant and wild type proteins accumulated to similar levels (Fig. S1).
These results are unexpected, because high levels of unphosphorylated AA normally lead to high levels of F activation. The results can be explained, however, by supposing that AB R105C preferentially binds ADP rather than ATP as inferred from the crystal structure of the protein, which implicates Arg-105 in the stabilization of the ␥-phosphate of ATP (46), and from the fact that AB R105C is strongly impaired in kinase activity 2 (13). Thus, rather than phosphorylating AA, the mutant protein preferentially binds to it, trapping it in an inactive complex (AA-AB R105C (ADP)). Further, when present in excess, AB R105C would sequester much or all of the unphosphorylated AA in complexes, thereby blocking F activation.
Increasing the Level of AA Reverses the Inhibition of F Caused by AB R105C -If our interpretation of the above results is correct, then the block in F activation caused by excess AB R105C should be reversed by increasing the cellular concentration of AA. To test this prediction, we compared strains harboring two copies of the gene for AB R105C and one or two copies of the gene for AA. Whereas two copies of the gene for AB R105C severely impaired the activation of F in the presence of one copy of spoIIAA (as seen in the results presented above), the addition of a second copy of spoIIAA reversed this inhibition, causing F to be activated at high levels (Fig. 8). These results are consistent with the idea that excess AB R105C (and by extension wild type AB in its ADP-containing form) creates a sink that traps AA in an inactive complex and that the sink can be overcome by raising the cellular concentration of AA (thereby restoring F activation).
An AA Mutant Defective in Binding to ADP-containing AB Activates F at Subthreshold Concentrations-As a further test of our model, we asked whether a mutant of AA that is impaired in its ability to bind to AB(ADP) is capable of activating F at subthreshold levels. We used for this purpose a mutant of AA, AA S58A , that has a 50-fold higher K d value for binding to AB(ADP) than does wild type AA (48). AA S58A is a mutant for the residue at which phosphorylation takes place and hence causes F to be active constitutively (11). Previous work had shown that when present at normal cellular levels, AA S58A breaks the dependence of F activation on polar septation (24). To investigate whether AA S58A would activate F in a div355 mutant when present at low concentrations, we drove expression of the gene encoding AA S58A from an inducible promoter so that the level of the mutant protein could be modulated (Fig.  9A). We found that in a div355 mutant, AA S58A activated F at much lower concentrations than did wild type AA (Fig 9B). These results are consistent with the idea that wild type AA is effective in activating F only at high concentrations due to complex formation with AB. DISCUSSION The developmental transcription factor F is subject to tight temporal and spatial control by a pathway involving the anti-F factor/kinase AB, the phosphoprotein AA, and the phosphatase SpoIIE. Here we have presented results that reinforce the view that this activation depends critically on the appearance of a threshold level of unphosphorylated AA and reveal an important new feature of the regulatory system, namely that the threshold level of AA may be set by a "sink" for AA created by complex formation between the unphosphorylated form of the protein and the ADP-containing form of the anti-sigma factor/kinase AB (AB(ADP)).
The starting point for this investigation was two amino acid substitution mutants of the SpoIIE phosphatase that block the activation of F . Both the S361F and Q483A substitutions lie in a region (II) of SpoIIE that is distinct from the phosphatase domain (region III) of the protein (20 -23, 44), and hence the basis for their effect on F activation was uncertain. Here we have shown that both substitutions markedly impaired the accumulation of unphosphorylated AA, suggesting that region II plays a previously undemonstrated role in the in vivo phosphatase activity of SpoIIE. Moreover, second site mutations were obtained that efficiently suppressed the effects of S361F and Q483A on F activation and spore formation. These suppressor mutations caused amino acid substitutions in SpoIIE itself (Q342P and V697A) or in AA (Q73R) or AB (R105C). In all cases, suppression was associated with the restoration of unphosphorylated AA to levels comparable with that seen in the wild type. Indeed, one of the amino acid substitutions (AB R105C ) was located in the catalytic center of AB and impaired its kinase activity. Evidently, the AB R105C substitution compensated for the deficient phosphatase activity of SpoIIE S361F simply by reducing the phosphorylation activity of the kinase. In toto, these findings and those of previous work (11,14,16,49) reinforce the view that the activation of F is strongly correlated with the appearance of unphosphorylated AA.
Importantly, however, the simple appearance of unphosphorylated AA is not sufficient to activate F . This is seen most clearly in experiments with mutants impaired in polar division. Mutants that fail to form a polar septum are blocked in F activation. Nonetheless, such mutants accumulate significant levels of unphosphorylated AA (24,32,33). This dependence of F activation on septation can be broken, however, by use of the suppressors described in the present work. One of these, spoIIE-V697A, was isolated independently by Hilbert and Piggott (50), and a mutation similar to spoIIE-Q342P, spoIIE-Q344P, was isolated by Feucht et al. (25). When separated from the original S361F or Q483A substitutions that impaired F activation, all four suppressors caused F to be activated more rapidly and to higher levels than that seen in the wild type. Isoelectric focusing analysis demonstrated that the suppressor mutations exerted their effect by greatly increasing the cellular level of dephospho-AA, rendering almost all of the AA protein in its unphosphorylated form. Feucht et al. have also described a hyperactive phosphatase mutant, spoIIE1301, that breaks the dependence of F activation on septation (18,32). Taken together, these results suggest that unphosphorylated AA must reach a high, threshold level in order to activate F . In keeping with this interpretation, previous work has shown that the dependence of F activation on septation can also be broken by overexpression of the gene for SpoIIE (17,19) and by the production of an unphosphorylatable mutant form of AA (AA S58A ) (11,14,24).
It might be argued that AA does not need to reach a threshold concentration per se. Rather, it might be necessary to reduce AA-P to very low levels. Perhaps AA-P is an inhibitor of F , and extensive dephosphorylation is required to deplete the cell of this inhibitor rather than to raise AA to very high levels. However, an experiment based on a mutant of AA (AA S58T ) that is subject to phosphorylation but not dephosphorylation demonstrated that F could be efficiently activated even in the presence of significant levels of AA-P. We therefore favor the view that the triggering event for F activation is the accumulation of unphosphorylated AA to levels that exceed a critical, threshold concentration.
What is the basis for this threshold? It is known that in addition to its role as an anti-F factor and a protein kinase, AB, when in its ADP-containing form, is capable of forming a long lived complex with AA (10,11,48). We propose that AB(ADP) serves as a sink for unphosphorylated AA, and only when the cellular concentration of AA exceeds the capacity of the sink is excess (uncomplexed) AA available in the cell to react with the AB(ATP)-F complex and thereby trigger the activation of F . Experiments using a mutant, AB R105C , that is predicted to bind ADP preferentially (as opposed to ATP), furnish evidence for an AB sink. When produced in addition to wild type AB (i.e. in a merodiploid strain), AB R105C severely blocked F activity. This was true both in cells producing wild type SpoIIE and in cells producing the hyperactive phospha- Shown are F -dependent ␤-galactosidase activity of a strain with two copies of spoIIAB-R105C and one copy of spoIIAA(KC471) (OE) and activity of a strain with two copies of spoIIAB-R105C and two copies of spoIIAA(KC533) (f). The experiment was done as described in the legend to Fig. 3. tase mutant, SpoIIE V697A (data not shown). In these experiments, almost all of the AA was in its unphosphorylated form (a state that is normally associated with exceedingly high levels of F activity). However, the presence of additional AB R105C molecules almost completely abolished F -directed gene expression. We conclude that AB R105C blocks F activation by sequestering unphosphorylated AA in an inert complex with the anti-sigma factor/kinase. Consistent with this interpretation, the inhibitory effect of excess AB R105C could be reversed by increasing the cellular concentration of AA. In toto, results from this and previous studies are consistent with a model in which F is not activated when the concentration of unphosphorylated AA is lower than the concentration of AB(ADP) but is activated efficiently once the level of dephospho-AA exceeds the level of available AB(ADP).
Other evidence consistent with our threshold model comes from experiments with a nonphosphorylatable mutant of AA, AA S58A , which has a 50-fold higher K d value for binding to AB(ADP) than does wild type AA (48). Using an inducible promoter to control the level of production of the mutant protein, we have found that AA S58A is a potent inducer of F activity even at levels well below those at which the dephospho form of wild type AA fails to activate F . Our explanation for the high potency of AA S58A is that the mutant protein is less subject to being trapped in inert complexes with AB(ADP) and hence does not need to reach a threshold concentration in order to trigger the activation of F . If our threshold model for the activation of F is correct, then how is AA maintained at subthreshold levels until and unless the polar septum is formed? Feucht et al. (25) have proposed that completion of the polar septum triggers an increase in the phosphatase activity of SpoIIE. However, our data revealed no measurable effect of blocking polar septation on the cellular ratio of AA to AA-P. A simple explanation for how the level of AA reaches a threshold concentration in the forespore is that the SpoIIE phosphatase is more concentrated in the forespore than it is in the predivisional sporangium or in sporangia in which polar septation has been prevented (16,51). This high concentration of SpoIIE is achieved by virtue of the localization of SpoIIE to the polar septum and the small size of the forespore (ϳ8-fold smaller than the mother cell). It is uncertain whether SpoIIE is sequestered to the forespore face of the septum (24,51), but whether it is displayed on both sides of the septum or on the forespore side only, its concentration in the forespore would be significantly higher than in the mother cell. Also contributing to the postseptation activation of F is the delayed entry of the gene for AB into the forespore by translocation of the chromosome across the polar septum. (The gene for AB is located near the terminus of the chromosome, which is the last region to enter the forespore) (27,52). AB is proteolytically unstable, and hence the anti-sigma factor/kinase may be replenished less efficiently in the forespore than in the mother cell (28). This combination of a reduced relative concentration of AB in the forespore and an increased relative FIG. 9. Low levels of AA S58A activates F more potently than does wild type AA. A, SDS-PAGE and immunoblotting of lysates from KC301 (div355 spoIIAA⌬1-AB ϩ -AC ϩ dacF::spc amyE:: spoIIAA cat thrC::spoIIQ-lacZ erm) (lane 1) sporulated in the presence of 25 mM xylose, and lysates of KC254 (div355 spoIIAA⌬1-AB ϩ -AC ϩ dacF::spc amyE:: P XylA -spoIIAA-S58A cat thrC::spoIIQ-lacZ erm) (lanes 2-6) sporulated in the presence of 25, 8.3, 2.8, 0.93, and 0 mM xylose, respectively. Xylose was added to the cultures at commencement of sporulation by resuspension, and cells were harvested for analysis 90 min after resuspension. B, F activity of each culture described in A versus the normalized level of AA in each culture 90 min after resuspension. q, KC254; f, KC301. F activity was determined by ␤-galactosidase assay, and the AA level was determined by phosphor imager analysis (Fuji) of the immunoblot shown in A. Since the anti-SpoIIAA immunoblot signal for KC301 in A represents both AA and AA-P, the level of AA was estimated based on the ratio of AA/ AA-P seen by IEF (data not shown).
concentration of SpoIIE could contribute to raising the ratio of unphosphorylated AA to AB in the forespore as compared with the mother cell or the predivisional sporangium (27). We propose that only in the forespore does unphosphorylated AA reach the critical concentration that enables it to exceed the sink of AB. On the other hand, mutants such as those described here that artificially raise the cellular concentration of AA by enhancing dephosphorylation or impeding phosphorylation allow AA to reach a threshold level in the absence of polar division.
In summary, we propose that AB serves as a buffer that soaks up the AA effector molecule until and unless it reaches a threshold concentration. This buffer ensures that F is not activated prematurely in the predivisional sporangium, although SpoIIE (and unphosphorylated AA) is present in the sporangium prior to the time that the polar septum is formed. The buffer also ensures that in the postdivisional sporangium F is not activated inappropriately in the mother cell, which also contains SpoIIE. Only in the forespore, where, as we postulate, dephosphorylated AA reaches high levels, is AA in excess to the AB buffer and hence free to trigger the activation of F . According to this view, the central event in the cell-specific activation of F is the accumulation of unphosphorylated AA to a critical concentration in the forespore.

Karen Carniol, Patrick Eichenberger and Richard Losick
Bacillus subtilis in F σ Protein