Evidence in support of a docking model for the release of the transcription factor sigma F from the antisigma factor SpoIIAB in Bacillus subtilis.

Cell-specific activation of the transcription factor sigmaF during the process of sporulation in Bacillus subtilis is governed by an antisigma factor SpoIIAB and an anti-antisigma factor SpoIIAA. SpoIIAB, which exists as a dimer, binds to sigmaF in a complex of stoichiometry sigmaF.SpoIIAB2. Escape from the complex is mediated by SpoIIAA, which reacts with the complex to cause the release of free sigmaF. Previous evidence indicated that Arg-20 in SpoIIAB is a contact site for both sigmaF and SpoIIAA and that contact with sigmaF is mediated by Arg-20 on only one of the two subunits in the sigmaF.SpoIIAB2 complex. Here we report the construction of heterodimers of SpoIIAB in which one subunit is wild type and the other subunit is a mutant for Arg-20. We show that the dissociation constant for the binding of sigmaF to the heterodimer was similar to that for the wild type, a finding consistent with the idea that sigmaF contacts Arg-20 on only one of the two subunits. Although SpoIIAA was highly effective in causing the release of sigmaF from the wild type homodimer, the anti-antisigma factor had little effect on the release of sigmaF from the heterodimer. This finding is consistent with a model in which SpoIIAA docks on the sigmaF.SpoIIAB2 complex, making contact with the subunit in which Arg-20 is not in contact with sigmaF. SpoIIAB is both an anti-sigmaF factor and a protein kinase that phosphorylates and thereby inactivates SpoIIAA. We show that SpoIIAA effectively displaces sigmaF from a complex of sigmaF with a mutant (SpoIIABR105A) that is impaired in the kinase function of SpoIIAB. This result shows that SpoIIAA-mediated displacement of sigmaF from SpoIIAB does not require concomitant phosphorylation of SpoIIAA.

Sporulation in the bacterium Bacillus subtilis is an attractive system in which to study the problem of cell-specific gene transcription (1,2). Sporulation involves the formation of an asymmetrically positioned septum that partitions the developing cell or sporangium into a forespore (the smaller cell) and a mother cell. The transcription factor F is activated in a cellspecific manner that limits its activity to the forespore compartment of the sporangium (3). The activity of F is governed by a pathway involving the proteins SpoIIAB, SpoIIAA, and SpoIIE (3)(4)(5)(6)(7)(8)(9)(10). SpoIIAB (henceforth abbreviated AB) is an anti-F factor that binds to F , trapping it in an inactive complex. AB is responsible for holding F inactive prior to the formation of the polar septum and in the mother cell after polar division (9 -11). The F factor escapes from the complex with AB in the forespore in a process that is mediated by the antiantisigma factor SpoIIAA (henceforth abbreviated AA; see Refs. 12 and 13). The activity of AA is, in turn, controlled by phosphorylation at Ser-58 through the opposing activities of SpoIIE, a phosphatase, and AB itself, which is both a serine protein kinase and an antisigma factor (6,10,14).
Here we are concerned with the mechanism by which AA mediates the release of F from the complex with AB. AB is a dimer and contains an adenosine nucleotide binding pocket that can hold either ATP or ADP (9). The binding of AB to F is dependent on the presence of nucleotide in the pocket, with ATP being more effective than ADP (15). The complex is asymmetric, having a stoichiometry of one molecule of F per dimer of AB ( F 1 ⅐AB 2 ) (16). Previous work indicated that AA reacts with the F ⅐AB⅐AB complex to induce the release of F from its antagonist (12,17). An attractive model for the mechanism of this displacement reaction arose from the x-ray crystallographic studies, which revealed a potential docking site for AA on one of the two AB subunits (18). According to this model, AA would dock on the complex, making contact with one molecule of AB and sterically displacing F from the other molecule. Previous work (13) had identified Arg-20 of AB as a contact site for both F and AA. A key feature of the docking model is that Arg-20 on one AB subunit is in contact with F , whereas Arg-20 on the other subunit is exposed to solvent and is free to make contact with AA. A second residue of importance in the docking model is Glu-104 of AB, which is a contact site for AA but not for F . Substitutions at Glu-104 impaired the capacity of free AB to phosphorylate AA and prevent AA from causing the release of F from the F ⅐AB⅐AB complex (17). Here we describe the use of heterodimeric mutant forms of the AB to investigate further the docking model and other aspects of the reaction of AA with the F ⅐AB⅐AB complex.

EXPERIMENTAL PROCEDURES
Construction of spoIIAB Expression Plasmids-Wild type and mutant forms of spoIIAB were expressed in Escherichia coli strain BL21(DE3)/pLysS (Novagen) using an isopropyl-1-thio-␤-D-galactopyranoside-inducible phage T7 RNA polymerase system. Three kinds of expression plasmids were constructed for the purification of AB proteins as follows.
An expression plasmid for the purification of untagged AB in the experiment of Fig. 1 was made by amplifying the wild type gene from pDAG14 (13,16) by PCR using oligonucleotides MH07 and MH08, which contained the restriction sites EcoRI and HindIII, respectively. The PCR product was digested with these enzymes and ligated to the expression plasmid pET-29a (Novagen), which had been similarly enzyme-treated, to create pMH13. Insertion into pET-29a joined spoIIAB to the coding sequence for an S-tag. pMH13 was introduced into E. coli strain BL21(DE3)/pLysS (Novagen) to create MHE31.
The co-expression plasmid for producing the AB⅐AB R20E heterodimer was made by amplifying the wild type spoIIAB gene from pDAG14 using oligonucleotides MH04 and MH17, which contained BamHI and HindIII/EcoRI sites, respectively. (The oligonucleotide MH17 contains a HindIII site 4 bases away from the end of the EcoRI site, which was subsequently used for cloning the second mutant copy of spoIIAB gene between restriction sites HindIII and EcoRI.) The PCR product was digested with BamHI and EcoRI and ligated to the expression plasmid pGEX-2TK (Amersham Biosciences) that had also been treated with BamHI and EcoRI to create pMH21. This fused spoIIAB in-frame to the coding sequence for GST 1 in the vector. By using pMH51 (the expression plasmid for producing His 6 -AB R20E ) as a template, a DNA sequence containing the mutant spoIIAB, its ribosome-binding site, and the upstream histidine codons was amplified by PCR using oligonucleotides MH16 and MH11, which contained the restriction sites HindIII and EcoRI, respectively. The PCR product was digested with these enzymes and ligated to HindIII/EcoRI-digested pMH21 to create pMH46. pMH46 was transformed into BL21(DE3)/pLysS to create MHE21.
Construction of spoIIAC Expression Plasmids-Strain LDE7 used for the production of F in the affinity chromatography experiment of Fig.  3 was described previously (9). The spoIIAC expression plasmid used for the experiment of Table I was constructed by amplifying the spoIIAC gene from chromosomal DNA from B. subtilis PY79 using the primers, 5Ј-atggatccatggatgtggaggttaag-3Ј and 5Ј-atgaattccatccgtatgatccat-3Ј. The PCR fragment was digested by BamHI and EcoRI and ligated into the BamHI/EcoRI-digested vector pGEX-2T (Amersham Biosciences) to generate pMF14 in which the F -coding sequence was fused to the coding sequence for GST. pMF14 was then transformed into DH5␣ to create MHE02.
Protein Purification-E. coli strains used for the production of AB and F were grown at 37°C in 1 liter of LB containing 75 g/ml ampicillin and 25 g/ml chloramphenicol until the culture reached an A 600 Ͼ0.4 (In the case of strains used for producing AB heterodimer and F , only ampicillin was added to the culture.) T7 RNA polymerase synthesis was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM. The cells were harvested 3 h later by centrifugation. Cell pellets were suspended in the appropriate binding buffer and disrupted by sonication. After sonication, the insoluble debris and the unlysed cells were removed by centrifugation, and the supernatant fluid was collected.
Four kinds of purification procedures were used for purifying AB and F proteins. For purification of the untagged AB, cell pellets from strain MHE31, which produces S-tagged AB, were suspended in buffer A (20 mM Tris-HCl, pH 7.5, 1.5 M NaCl, 1% Triton X-100). The supernatant fluid was loaded onto S-protein-agarose (Novagen) for 1 h at 4°C. After washing with 6 column volumes of buffer A, the S-protein-agarose bound with S-tagged AB was suspended in buffer B (20 mM Tris-HCl, pH 8.4, 150 mM NaCl, 2.5 mM CaCl 2 ). 25 units of biotinylated thrombin were added to the agarose mixture for up to 2 h at room temperature. Untagged AB was released from the agarose during this cleavage reaction and stored in 15% glycerol at Ϫ80°C.
For purification of the His 6 -AB (wild type and mutant proteins), cell pellets from strains MHE08, MHE53, MHE55, and MHE62, which produce His 6 -AB, His 6 -AB E104K , His 6 -AB R20E , and His 6 -AB R105A respectively, were suspended in buffer C (0.05 M Tris-HCl, pH 8, 0.5 NaCl, 5% (v/v) glycerol, 0.5 mM ␤-mercaptoethanol) plus 5 mM imidazole. The supernatant fluid was loaded onto Ni 2ϩ -NTA-agarose (Qiagen) for 1 h at 4°C. After washing with 6 column volumes of buffer C plus 20 mM imidazole, proteins bound to the column were eluted with buffer C plus 200 mM imidazole. Elution fractions containing the purified proteins were dialyzed into storage buffer and stored in 15% glycerol at Ϫ80°C.
Cell pellets from strain MHE21, which produced AB⅐AB R20E heterodimer, were suspended in buffer C plus 5 mM imidazole. The supernatant fluid was loaded onto Ni 2ϩ -NTA-agarose for 1 h at 4°C. After washing with 6 column volumes with buffer C plus 20 mM imidazole, proteins bound to the column were then eluted with buffer C plus 200 mM imidazole. Elution from Ni 2ϩ -NTA-agarose was diluted and mixed directly with glutathione-Sepharose 4B resin (Amersham Biosciences) with 0.5% (w/v) bovine serum albumin (Sigma) for 1 h at 4°C. The column was washed with 5 column volumes of 1ϫ PBS and eluted with buffer D (10 mM reduced glutathione in 50 mM Tris-HCl, pH 8). The eluate was dialyzed into buffer E (0.02 M Tris-HCl, pH 8, 0.15 M NaCl, 5% (v/v) glycerol, 0.5 mM ␤-mercaptoethanol, 5 mM CaCl 2 ) and treated with 50 units of thrombin. SpoIIAB heterodimers were separated from the GST tag by remixing the fraction with glutathione-Sepharose 4B resin for 1 h at 4°C. The flow-through from the resin contained the pure SpoIIAB heterodimer. It was stored in 15% glycerol.
Cell pellets from strain MHE02-produced GST-tagged F were resuspended in 1ϫ PBS. The supernatant fluid was loaded onto glutathione-Sepharose 4B resin. The column underwent extensive washes with 1ϫ PBS, and F protein was released from the GST tag by cleaving the column with 50 units of thrombin in 1ϫ PBS.
Site-directed Mutagenesis-A mutant spoIIAB encoding AB R105A was created by site-directed mutagenesis using appropriately designed oligonucleotides MH85 and MH86, pDAG14, and the Quick-Change Sitedirected mutagenesis kit (Stratagene).
Testing Subunit Exchange-Equimolar amounts (10 M each) of AB and His 6 -AB were mixed and incubated at 4°C for 3 h. The mixture was diluted in buffer C plus 5 mM imidazole from 50 to 500 l. 25-l resin (bed volume) was added to the 500-l mixture, and the sample was rotated on a rotator at 4°C for another hour. The resin was collected by centrifugation (4000 rpm), washed with 5 column volumes of buffer C plus 20 mM imidazole, and suspended in 25 l of 2ϫ SDS sample buffer (100 mM Tris-HCl, pH 6.8, 200 mM dithiothreitol, 4% SDS, 0.2% bromphenol blue, 20% glycerol). Eluates from the resin were analyzed by SDS-PAGE.
Surface Plasmon Resonance-Kinetic and equilibrium constants of the protein-protein interactions were measured by surface plasmon resonance with a BIAcore 3000 instrument (Amersham Biosciences). To prepare the sensor chip, 80 l of anti-His antibodies (concentration 50 g/ml) diluted in 10 mM acetate, pH 4.5, were immobilized on the chip through the amino coupling procedure with HBS buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 0.0005% surfactant P-20; Amersham Biosciences). The sensor surface was activated by injection of 1:1 mixture of N-hydroxysuccinimide/1-ethyl-3-(dimethylaminopropyl)carbodiimide for 8 min at 10 l/min. Anti-His antibodies were then injected at a flow rate of 5 l/min for 8 min. 1 M ethanolamine, pH 8.5, was injected at a flow rate of 10 l/min for 8 min to block any unreacted esters. Unbound material near the chip surface was removed by injecting 10 mM HCl at a flow rate of 10 l/min for 2 min.
A sample of solution containing His 6 -AB, His 6 -AB R20E , or His 6 -AB⅐AB R20E heterodimer diluted in HBS buffer was applied to one flow cell of the CM5 sensor chip already coupled with anti-His antibodies. The analyte F was dialyzed into binding buffer (100 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 0.5 mM dithiothreitol, 1 mM ATP) and injected over the sensor surface at a flow rate of 30 l/min for a total of 60 l. Concentrations of the analyte F applied ranged from 20 to 200 nM. After the interaction, 50 l of 20 mM HCl was injected to regenerate the surface. Moreover, no specific interactions were observed when analyte F was injected over a flow cell with immobilized anti-His antibody only. Apparent kinetic constants (k on and k off ) were obtained by use of the BIAevaluation software (Amersham Biosciences).
Production of [ 35 S]Methionine-labeled F -Radiolabeling of F using strain MHE5 was as described by Alper et al. (15) with the following changes. B 1 vitamin was added to the culture (1 g/ml). The cells were grown in the presence of rifampicin for 2.5 h before labeling. Unlabeled cells were harvested before the addition of [ 35 S]methionine, and 1 ml of the cultures was pelleted and lysed in 500 l of buffer that contained 50 mM Tris-HCl, pH 8, 100 mM NaCl, 0.1 mM dithiothreitol, and 0.5% Triton X-100.
Affinity Chromatography-Mutant and wild type forms of AB were mixed with Ni 2ϩ -NTA-agarose (30 l of resin) for 1 h at room temperature in buffer C. The resin was collected by centrifugation (4000 rpm). 100 l of the radiolabeled F was added to the resin, and the mixture (original volume of 150 l) was diluted to a volume of 500 l with buffer C. The column support bed was formed with ϳ5 l of glass beads (200 -300 M; Sigma, catalog number G-1277), and the column was filled with buffer C. The column bed was formed by adding the resin mixture and allowing it to settle while the column flowed by gravity. The column was then washed 5-10 times with 5 column volumes of buffer C. The column was eluted with 50 l of 1% SDS, buffer C, 8 M non-radioactive F , or 8 M AA purified as described previously (13).
Kinetic Measurements of Kinase Activity-Phosphorylation reactions were carried out at 21°C in 50-l reaction volumes containing 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM ATP plus 5 Ci of [␥-32 P]ATP (6000 Ci/mmol), 25 nM AB. The AA concentration was varied from 62.5 nM to 2 M. Reactions were stopped by the addition of 850 l of 10% (v/v) trichloroacetic acid before the rate of phosphorylation reached a maximum. The protein was precipitated by the addition of bovine serum albumin to a final concentration of 1 mg/ml. The pellets were washed three times with 1 ml of 10% trichloroacetic acid and dissolved in 1 ml of Tris base. The final sample was counted in liquid scintillation counter after adding 7 ml of scintillation fluor (Opti-Fluor, Packard Instrument Co.). The time 0 blank was subtracted from each value.
F -Directed ␤-Galactosidase Synthesis-pMH68 containing the spoIIAB-R105A mutation was linearized with restriction enzymes StuI and XcmI, and introduced by double recombination into the spoIIA operon of competent cells of MHB26 (thr::spoIIQ-lacZ::erm) by transformation and selection for spectinomycin resistance, creating strain MHB10. Cell pellets from MHB10 were collected at each time point from the beginning of sporulation (time 0) in resuspension medium and treated with 10 l of 20 mg/ml lysozyme at 30°C for 15 min. ␤-Galactosidase activity of the sample was determined according to Harwood and Cutting (25).

AB Dimer Subunits Do Not
Exchange-Our strategy for studying the contribution of each of the two subunits of AB⅐AB to its interactions with F and AA was to construct heterodimers of the antisigma factor in which one of its two subunits was a mutant for a contact site. The feasibility of such a strategy rested on the premise that AB⅐AB is stable and hence that heteromeric dimers would not undergo subunit exchange with each other. Subunit exchange would defeat the purpose of constructing heterodimers as it would generate a mixture of homodimers and heterodimers. To investigate whether AB⅐AB would undergo subunit exchange, we separately purified homodimers of AB⅐AB that had been tagged with an extension of six histidine residues at its NH 2 terminus (His 6 -AB⅐His 6 -AB) and homodimers of the unmodified, wild type protein. Next, we prepared an equimolar mixture of His 6 -AB⅐His 6 -AB and AB⅐AB and incubated the mixture at 4°C for 3 h. Finally, we recovered the histidine-tagged protein from the mixture using Ni 2ϩ -NTAagarose. When analyzed by SDS-PAGE, material that had adhered to the resin was found to contain only His 6 -tagged AB ( Fig. 1, lane 11; which could be distinguished from unmodified AB by its larger size). Control experiments carried out in parallel showed that His 6 -tagged AB bound to the resin, whereas unmodified AB did not (Fig. 1, lanes 4 and 8). Conversely, little or no His 6 -tagged AB was present in the flow-through when either His 6 -AB alone (lane 3) or a mixture of His 6 -AB and unmodified AB (lane 10) was applied to the resin. On the other hand, unmodified AB was present in the flow-through when either the untagged protein alone (lane 7) or the mixture of tagged and untagged proteins (lane 10) were applied to the resin. We conclude that little or no His 6 -AB⅐AB heterodimers had formed during the incubation period and hence that the dimer does not undergo a significant level of subunit exchange.
Purification of an AB Heterodimer-Arg-20 of AB was identified previously as a probable contact site both for F and AA in that a loss of side chain substitution (R20A) or a replacement with a negatively charged side chain (R20E) impaired binding to either partner protein in vitro and caused high levels of F activity in vivo (13). The assignment of Arg-20 as a contact site for F was confirmed by the recent determination of the crystal structure of the F ⅐AB⅐AB complex (18). The crystal structure revealed that Arg-20 on one of the two AB subunits in the complex is in contact with F , whereas Arg-20 on the other subunit is exposed to solvent and hence would potentially be available as a docking site for AA. This interpretation predicts that heterodimeric AB in which one subunit is wild type and the other harbors the R20E substitution would be unimpaired in binding to F . (We used R20E rather than R20A because the effect of the former substitution is more severe than that of the latter; see Refs. 13 and 17.) To create the heterodimer, we co-expressed the genes for the mutant and wild type protein in E. coli. The genes were arranged in tandem in a construct in which their transcription was under the control of a promoter recognized by the phage T7 RNA polymerase. The gene for the wild type protein in the construct was preceded by the coding sequence for GST as well as the coding sequence for the cleavage site for the protease thrombin, whereas the gene for the mutant protein was preceded by six histidine codons. Thus, E. coli cells harboring the construct were expected to produce GST-tagged AB and His 6 -tagged AB R20E .
Expression of the construct was expected to generate three kinds of dimers: GST-AB homodimers, His 6 -AB R20E homodimers, and GST-AB⅐His 6 -AB R20E heterodimers. A lysate was prepared from cells of E. coli harboring the construct, and the following procedure was used to separate the heterodimers from the homodimers and other proteins in the lysates. First, we used Ni 2ϩ -NTA-agarose to purify His 6 -AB homodimers and GST-AB⅐His 6 -AB R20E heterodimers from the lysate. Lane 1 of Fig. 2 shows that both GST-and His 6 -tagged proteins had adhered to the column. Second, a glutathione-Sepharose (GST) resin was used to separate GST-AB⅐His 6 -AB R20E heterodimers from the His 6 -AB homodimers. The purified heterodimers (Fig.  2, lane 2) were treated with thrombin to remove the GST tag, resulting in purified AB⅐His 6 -AB R20E heterodimers. Finally, glutathione-Sepharose was used to remove the GST tag that had been released by thrombin treatment and residual GST-AB⅐His 6 -AB R20E heterodimers that had not undergone proteolytic cleavage from the AB⅐His 6 -AB R20E heterodimers. The purified AB⅐His 6 -AB R20E heterodimers were recovered in the flowthrough fraction (Fig. 2, lane 3).
By using similar procedures we also constructed and purified wild type homodimers of His 6 -tagged AB and mutant homodimers of His 6 -tagged AB R20E . For simplicity the two His 6 -tagged homodimers are henceforth referred to as AB⅐AB and AB R20E ⅐AB R20E , and the His 6 -tagged heterodimer is called AB⅐AB R20E .
Kinetic and Equilibrium Constants for the Interaction of F with SpoIIAB Homodimers and Heterodimers-We used surface plasmon resonance to determine the equilibrium dissociation constants for the interaction of F with His 6 -tagged AB⅐AB, AB⅐AB R20E , and AB R20E ⅐AB R20E . The purified homoand heterodimers (the ligands) were separately immobilized on a layer of dextran that had been coupled with anti-His 6 antibodies and attached to a thin film of gold. Next, purified F (the analyte) was applied to the surface and allowed to bind with immobilized ligand during the association phase of the analysis. After the analyte had been applied, F was allowed to dissociate from the ligand during the dissociation phase of the analysis. Association and dissociation of F were detected optically. Software provided with the instrument was used to calculate association rate constants from the association phase of the interactions and dissociation rate constants from the dissociation phase of the interactions. All interactions were monitored in the presence of 1 mM ATP.
First we measured association and dissociation rate constants for the interaction between AB⅐AB (wild type homodimer) and F , from which we derived an equilibrium dissociation constant (K d ) of 12 nM (Table I). This value was similar to that (14 nM) reported previously by Magnin et al. (19). Second, we investigated the interaction of F with AB⅐AB R20E . Importantly, we obtained a K d value that was only modestly higher (28 nM) than that observed with the wild type homodimer (see also Table I). Finally, we attempted to measure the interaction of F with AB R20E ⅐AB R20E . The association rate constant observed using the mutant homodimer was so low that an equilibrium dissociation constant could not be derived, a finding that underscores the importance of the Arg-20 side chain in the interaction of F with AB. In toto, these results indicate that complex formation requires the presence of one and only one Arg-20 side chain, a finding consistent with the idea that the F -Arg-20 interaction occurs on only one of the two subunits of the antisigma factor.
Rate constants obtained by surface plasmon resonance can differ substantially from those obtained by other procedures and can be strongly influenced by such factors as the flow rate of the analyte over the surface of the sensor (20,21). In this regard, we note that the apparent rate constant for dissociation (k off ) for the F ⅐AB⅐AB complex would appear to indicate a relatively short half-life (ϳ3 min). Yet other evidence from affinity chromatography indicates that the complex is longlived (see Refs. 12 and 13; see below). With this caveat in mind, the absolute values of the rate constants and equilibrium dissociation constants presented in Table I should be treated with caution. Nonetheless, we believe that we are safe in the general conclusion that F ⅐AB⅐AB and F ⅐AB⅐AB R20E have similar dissociation rate constants and much lower constants than that for F ⅐AB R20E ⅐AB R20E .

A Complex of F with the AB Mutant Heterodimer Is Resistant to AA-mediated Dissociation-Previous work indicated that
Arg-20 on AB is responsible for contacting AA as well as F . If, as our present results and those of Campbell et al. (18) indicate, F contacts Arg-20 on only one of the two subunits of the antisigma factor, then Arg-20 on the other subunit might serve as a contact site for AA during the AA-mediated dissociation of the F ⅐AB⅐AB complex. If so, then AA could cause the release of F from the F ⅐AB⅐AB complex by a simple displacement mechanism in which the anti-antisigma factor docks on the subunit with a free Arg-20 side chain. To investigate this hypothesis, we separately immobilized AB⅐AB and AB⅐AB R20E on solid matrices. Next, we applied radioactive F to the matrices, thereby creating immobilized complexes of F ⅐AB⅐AB and F ⅐AB⅐ AB R20E . Finally, we measured the release of radioactive F from the complexes following the application of buffer, buffer containing non-radioactive F , and buffer containing purified AA. Fig. 3a shows the results of a representative experiment, and Fig. 3b summarizes the results of four independent experiments. As observed previously, only a small proportion of the radioactive F was released from the F ⅐AB⅐AB complex by buffer (5%; column 1 of Fig. 3b) or buffer containing unlabeled F (13%; column 2), whereas a relatively high proportion (57%; column 4) was discharged by the anti-antisigma factor. In contrast, AA was no more effective than unlabeled F or buffer alone in releasing radioactive F from the F ⅐AB⅐AB R20E complex (7%; column 3). These results are consistent with a model in which AA docks on the subunit with a free Arg-20 side chain to effect the release of F from the other subunit.
The results of Fig. 3b also reinforce the view (12, 13) that the F ⅐AB⅐AB complex is relatively long lived and argue against an alternative model for the release of F based on the idea the FIG. 2. Purification of AB⅐AB R20E . Heterodimeric AB was prepared as described under "Experimental Procedures." The figure shows SDS-PAGE using 18% polyacrylamide of protein from various steps of the procedure. Protein from cells producing GST-AB and His 6 -AB R20E was applied to Ni 2ϩ -NTA-agarose. Lane 1 shows protein eluted from the column with 0.2 M imidazole. The eluate was applied to a glutathione-Sepharose 4B resin. Lane 2 shows protein eluted from the column with reduced glutathione. This material was treated with thrombin and reapplied to glutathione-Sepharose 4B resin. Lane 3 shows the material that followed through the resin. Numbers on the left indicate the size and position of molecular weight markers. complex is highly dynamic. In the dynamic model, the complex undergoes rapid dissociation and re-association, with AA binding to and trapping free AB⅐AB, thereby blocking the re-association of AB⅐AB with F . If the complex were dynamic, then excess unlabeled F should have been able to cause the displacement of radioactive F from the column of immobilized F ⅐AB⅐AB. Also, AA should have been able to cause the release of F from the F ⅐AB⅐AB R20E complex as effectively as from the F ⅐AB⅐AB complex.
Efforts to Create a Catalytic Mutant Defective in the Phosphorylation of AA-AB is both an antisigma factor that binds to F and a serine protein kinase that is capable of phosphorylating AA (9,10). Indeed, evidence indicates that AA becomes phosphorylated when the anti-antisigma factor reacts with the F ⅐AB⅐AB complex (17). We therefore wondered whether phosphorylation is required in the AA-mediated dissociation of F ⅐AB⅐AB or is simply a consequence of it. To address this question, we sought to create a mutant of AB that was defective in the kinase reaction but was unaltered in its ability to bind to F and AA. Earlier work had shown that an amino acid substitution at Glu-104 (E104K) impaired phosphorylation of AA, but kinetic analysis revealed that the defect was due to impaired binding to the AA substrate (K m ) rather than impaired catalysis (k cat ) (17).
The crystal structure of the F ⅐AB⅐AB complex reveals two candidates for amino acids that could be directly involved in the kinase reaction. One candidate is Glu-46, which is in the catalytic center of the kinase in a position from which it could promote the nucleophilicity of the attacking water molecule in the ATPase reaction. Accordingly, we built a loss of side chain substitution mutant in which Glu-46 was replaced with alanine. Unexpectedly, however, AB E46A proved to be impaired in its ability to bind F (data not shown). It is known that the binding of AB to F is dependent upon adenosine nucleotide, and perhaps the Glu-46 side chain is needed to retain ATP in the nucleotide binding pocket of the antisigma factor/kinase. In any event, the E46A substitution is not simply impaired in catalysis and hence could not be used to address the question of the role of phosphorylation in the AA-mediated dissociation of the F ⅐AB⅐AB complex.
A second candidate for a residue functioning in catalysis was Arg-105, which as inferred from the crystal structure could be involved in stabilizing the transition state of the phosphotransfer reaction as well as in contacting Ser-59 in AA, which is immediately adjacent to the site of phosphorylation (Ser-58). We built an R105A substitution mutant and found that it was unimpaired in its ability to bind to F . Next, we carried out a kinetic experiment to measure the rate of incorporation of ␥-32 P into the substrate AA as catalyzed by AB R105A and as compared with wild type AB. The Michaelis-Menten curve of Fig. 4 shows that the mutant enzyme was little altered in its catalytic activity (k cat ) but was markedly impaired in its K m ; we obtained a value of 557 nM for the mutant enzyme as compared with 4.4 nM for the wild type kinase (17). In this regard, the R105A mutant resembles E104K, which as discussed above is also impaired in its binding to AA. Evidently, the side chains of the adjacent residues Arg-105 and Glu-104 are both needed for substrate binding but not for catalysis.
AB R105A Is Defective in Phosphorylation of AA during Sporulation-Next, we carried out experiments monitoring phosphorylation of AA in vivo. Strains producing wild type AB or AB R105A were induced to sporulate, and samples were collected at 1.5 and 2 h after the onset of sporulation. Cell lysates were then prepared and subjected to IEF to separate unphosphorylated and phosphorylated (AA-P) forms of AA. AA and AA-P FIG. 3. AA-mediated release of F from complexes of the transcription factor with wild type and mutant forms of AB. Affinity chromatography was carried out using Ni 2ϩ -NTA-agarose and Histagged AB proteins as described under "Experimental Procedures." Radiolabeled F was applied to the column, and then after washing was eluted with the solutions indicated below. Radioactive F in the eluates was subjected to SDS-PAGE and visualized using a PhosphorImager (Fuji). a shows the results with F bound to the wild type dimer AB⅐AB (left-hand panel) and heterodimer mutant dimer AB⅐AB R20E (righthand panel). Radioactive F was eluted from the column as indicated with 1% SDS, buffer C (described under "Experimental Procedures"), 8 M non-radioactive F , or 8 M AA. b shows the percentages (the average of four experiments) of radioactive F that were recovered in the eluates relative to elution with SDS for buffer C (column 1) and for non-radioactive F (column 2) from columns with the wild type homodimer. (Very similar values were obtained with the mutant heterodimer.) Also shown are the percentages relative to the SDS elution of radioactive F recovered in the eluates with AA from the column with F ⅐AB⅐AB R20E (column 3) and the column with F ⅐AB⅐AB (column 4). The results are averages from four experiments for columns 3 and 4. c is the same as a except that F was in a complex with AB E104K ⅐AB E104K (left-hand panel) or with AB R105A ⅐AB R105A . d is the same as b except that column 3 represents the percent of radioactive F eluted with AA from F ⅐AB E104K ⅐AB E104K and column 4 radioactive F eluted from F ⅐AB R105A ⅐AB R105A . were visualized by immunoblotting with polyclonal anti-AA antibodies (Fig. 5). In a strain producing wild type AB, AA was phosphorylated with normal efficiency, as observed previously (13). Lanes 1 and 2 (1.5 and 2 h of sporulation, respectively) of Fig. 5 show both AA and AA-P, with the lower band representing AA-P. In striking contrast, we detected little or no AA-P in samples from cells producing AB R105A (lanes 3 and 4), suggesting that AB R105A is defective in phosphorylating AA. Note that the lower band present in lanes 3 and 4 is shifted slightly higher than the bands representing AA-P in the other lanes and represents one of two background bands that are also detected in a lysate of cells of a null mutant lacking AA (lane 9). To assess further the in vivo kinase activity of AB R105A , we monitored phosphorylation of AA in the absence of SpoIIE, the phosphatase responsible for dephosphorylating AA-P. As shown previously (6,22) in cells lacking SpoIIE but producing wild type AB, AA was almost entirely in the phosphorylated form ( lanes 5 and 6). Importantly, in cells lacking SpoIIE but producing AB R105A , a significant level of AA-P was observed, although the proportion of AA that was in the phosphorylated form was substantially lower than that observed with wild type AB (compare lanes 7 and 8 with lanes 5 and 6). We interpret these results to indicate that AB R105A is impaired but not completely blocked in kinase activity. Its residual kinase activity is masked by the action of SpoIIE but in the absence of the competing phosphatase kinase activity can be detected.
AB E104K and AB R105A Have Opposite Effects on F Activation in Vivo and on the Dissociation of the F ⅐AB⅐AB Complex in Vitro-As reported previously (17) and confirmed here, sporulating cells producing AB E104K are blocked in the induction of a lacZ fusion to a gene under the control of F . Also, and again as confirmed here, the complex of F with AB E104K ( F ⅐AB E104K ⅐AB E104K ) is known to be immune to attack by AA (9% release of radioactive F ; column 3, Fig. 3d). We interpret these results as indicating that F does not become activated in mutant cells simply because it is unable to escape from the F ⅐AB E104K ⅐AB E104K complex.
Remarkably, however, the R105A substitution had the opposite effect on F -directed ␤-galactosidase synthesis in vivo and on AA-mediated dissociation of the F ⅐AB⅐AB complex in vitro. The results of Fig. 6 show that in cells producing AB R105A , F was activated earlier and to a much greater extent than in cells producing the wild type protein. Also, the F ⅐AB R105A ⅐AB R105A complex was susceptible to AA, efficiently releasing F in response to the anti-antisigma factor (59% release of radioactive F , column 4 of Fig. 3d). We interpret this result to indicate that AA-mediated release of F is not dependent upon the kinase activity of AB. In further support of this interpretation, we observed that approximately half of the AA that eluted from the column with immobilized F ⅐AB⅐AB was phosphorylated, as detected by isoelectric focusing, whereas almost all the AA that eluted from the column with F ⅐AB R105A ⅐AB R105A was in the unphosphorylated form (data not shown). Approximately equal amounts of F were released from both columns but with little phosphorylation of AA in the case of the complex with the mutant AB.
In summary, the E104K substitution impairs the capacity of AB to phosphorylate AA and the ability of the F ⅐AB⅐AB complex to be dissociated by AA. The R105A substitution, on the other hand, impairs the ability of AB to phosphorylate AA but not the susceptibility of the F ⅐AB⅐AB complex to AA-mediated dissociation. The simplest interpretation of these observations is that Arg-105 is only needed in the kinase reaction and that AA-mediated dissociation of the F ⅐AB⅐AB complex does not require concomitant phosphorylation of AA. In contrast, we presume that Glu-104 is an important contact site between AB and AA both during the kinase reaction and during the displacement of F from AB. DISCUSSION We have presented biochemical evidence in support of a model for the release of F from F ⅐AB⅐AB complex in which the anti-antisigma factor AA docks with the complex and thereby displaces the transcription factor from its inhibitor (Fig. 7). A key feature of the model is residue Arg-20, which previous genetic, biochemical, and structural analyses identified as a contact site both for F and for AA (13,16,18). According to the crystal structure of the F ⅐AB⅐AB complex, Arg-20 is in contact with F on one of the two subunits of the AB dimer but is exposed to solvent on the other subunit where it could dock with AA (18). The principal findings of the present investigation were as follows. First, we have shown that the dissociation constant of the F ⅐AB⅐AB R20E complex is similar to that of the corresponding wild type complex. This finding sup- ports the conclusion that the stability of the F ⅐AB⅐AB complex depends on the presence of Arg at position 20 on one and only one of the two AB subunits in the complex. Second, F was efficiently displaced from the wild type complex by AA but poorly from the F ⅐AB⅐AB R20E complex. This finding together with previous findings indicating that Arg-20 is a contact site for AA support the view that AA docks with the F ⅐AB⅐AB complex and does so by contacting Arg-20 on the subunit that is not in contact with F . Thus, Arg-20 serves two distinct functions. Reflecting the asymmetry of the F ⅐AB⅐AB complex, Arg-20 on one subunit interacts with F and on the other subunit is exposed in a manner that allows it to contact AA.
AB is both an anti-F factor and a protein kinase that is responsible for phosphorylating and thereby inactivating AA (9 -12). A point of uncertainty has been whether AA becomes phosphorylated during the displacement reaction and, if so, whether phosphorylation is required in order for AA to liberate F from the complex. Campbell et al. (18) demonstrated that a non-phosphorylatable mutant of AA (AA S58A ) in which the phosphorylated residue, Ser-58, was replaced with alanine, was able to cause the displacement of F from the complex with AB. This finding is consistent with the idea that displacement does not require concomitant phosphorylation. We have been able to obtain complementary evidence on this point through the creation of an amino acid substitution mutant of AB at residue Arg-105 (R105A). The AB R105A mutant was defective in its capacity to phosphorylate AA, as shown biochemically as well as in vivo. Nevertheless, the complex of the mutant protein with F ( F ⅐AB R105A ⅐AB R105A ) was fully susceptible to undergoing dissociation in response to wild type AA. Moreover, cells producing AB R105A exhibited abnormally high levels of F activity during sporulation, a finding consistent with the idea that the mutant is defective in phosphorylating AA but not in releasing F . The crystallographic structure of the AA-AB complex reveals that Arg-105 contacts the residue (Ser-59) that is immediately adjacent to the side chain (that of Ser-58) that undergoes phosphorylation in AA. Taken together, these findings are consistent with the idea that AB R105 A is involved in the kinase reaction but is not needed for the interaction of AA with AB when F is displaced from the F ⅐AB⅐AB complex. Thus, two complementary lines of evidence, one based on a kinase-defective mutant of AB and the other on a non-phosphorylatable mutant of the substrate AA, indicate that the displacement reaction does not require concomitant phosphorylation of AA. Nevertheless, it is entirely possible that AA does indeed become phosphorylated in reacting with the complex, and previous evidence is consistent with this idea. The conclusion we draw is that AA need not be phosphorylated concomitantly in order for dissociation of the complex to take place. Rather, it seems likely 2 that displacement and phosphorylation represent successive steps in the reaction of AA with the F ⅐AB⅐AB complex.
Finally, we comment on one additional feature of the interplay between AA and AB. The antisigma factor and the antiantisigma factor are mutually antagonistic proteins. On the one hand, and as we have seen, AB that contains ATP in its nucleotide binding pocket is capable of phosphorylating and thereby inactivating AA. On the other hand, AA is capable of binding to an ADP-containing form of AB (13,15,23). As a result of the kinase reaction, AB is left with ADP in the nucleotide binding pocket. A fresh molecule of unphosphorylated AA can bind tightly to the ADP-containing form of AB to form a long lived complex. AB in the resulting AA⅐AB(ADP) complex is inert both as a kinase and as an anti-F factor. Formation of the AA⅐AB(ADP) complex is believed to contribute importantly to the activation of F by sequestering AB in a form in which it is unable to phosphorylate AA or to inhibit F (15,24). Recent work by Masuda et al. 3 reveals that AA undergoes a conformational distortion upon binding to ATP-containing AB. The authors hypothesize that this energetically unfavorable distortion does not occur when AA binds to the ADP-containing form of AB, thereby explaining the high stability of the AA⅐AB(ADP) complex.
These findings also shed new light on the role of residue Glu-104 in AB. As confirmed here and shown previously, the Glu-104 side chain is required for AB(ATP)-mediated phosphorylation of AA and in the AA-mediated displacement of F from the F ⅐AB(ATP) 2 complex. It is not, however, required in order for AA to form the AA⅐AB(ADP) complex. Masuda et al. 3 propose that the favorable electrostatic interaction of Glu-104 with AA compensates for the energetic cost of the conformation distortion required in order for AA to interact with the ATPcontaining form of AB. Because no such distortion is expected to occur when AA interacts with the ADP-containing form of AB, Glu-104 is dispensable in AA⅐AB(ADP) complex formation. The findings underscore the intricacy of the interplay between the proteins that govern the activation of cell-specific transcription.