The Bacillus subtilis response regulator Spo0A stimulates sigmaA-dependent transcription prior to the major energetic barrier.

At the spoIIG promoter phosphorylated Spo0A (Spo0A approximately P) binds 0A boxes overlapping the -35 element, interacting with RNA polymerase to facilitate open complex formation. We have compared in vitro transcription from a series of heteroduplex templates containing denatured regions within the promoters. Transcription from heteroduplex templates with 12, 8, or 6 base pairs denatured was independent of Spo0A approximately P, but heteroduplexes with 4 or 2 base pairs denatured required Spo0A approximately P for maximal levels of transcription. Investigation of the thermal dependence of transcription suggested that strand separation was the primary thermodynamic barrier to transcription initiation but indicated that Spo0A approximately P does not reduce this energetic barrier. Kinetic assays revealed that Spo0A approximately P stimulated both the rate of formation of initiated complexes as well as increasing the number of complexes capable of initiating transcription. These results imply that Spo0A approximately P stimulates transcription at least in part by stabilizing the RNA polymerase-spoIIG complex until contacts between RNA polymerase and the -10 element induce strand separation.

At the spoIIG promoter phosphorylated Spo0A (Spo0AϳP) binds 0A boxes overlapping the ؊35 element, interacting with RNA polymerase to facilitate open complex formation. We have compared in vitro transcription from a series of heteroduplex templates containing denatured regions within the promoters. Transcription from heteroduplex templates with 12, 8, or 6 base pairs denatured was independent of Spo0AϳP, but heteroduplexes with 4 or 2 base pairs denatured required Spo0AϳP for maximal levels of transcription. Investigation of the thermal dependence of transcription suggested that strand separation was the primary thermodynamic barrier to transcription initiation but indicated that Spo0AϳP does not reduce this energetic barrier. Kinetic assays revealed that Spo0AϳP stimulated both the rate of formation of initiated complexes as well as increasing the number of complexes capable of initiating transcription. These results imply that Spo0AϳP stimulates transcription at least in part by stabilizing the RNA polymerase-spoIIG complex until contacts between RNA polymerase and the ؊10 element induce strand separation.
Transcription initiation is an intricate process accomplished by the interplay between promoter DNA and a multisubunit RNA polymerase core enzyme, ␣ 2 ␤␤Ј, complexed with one of several alternative sigma factors, . In Bacillus subtilis, the majority of transcripts during logarithmic growth are produced by RNA polymerase complexed with A , a homolog of the Escherichia coli sigma factor, 70 . Promoters used by such RNA polymerase (RNAP) 1 holoenzymes contain two conserved sequence elements centered at Ϫ35 and Ϫ10 (relative to the start site of transcription) separated by an optimal distance of 17 bp (1). Additional nearby sequence elements may also contribute to the kinetics of transcription initiation (2,3). RNAP-promoter complexes pass through several intermediates during initiation including closed complexes in which the DNA is fully double-stranded and open complexes in which the DNA strands are partially separated as a prelude to RNA polymerization (4). The physical basis for DNA strand separation remains poorly understood.
The exact sequence of promoter elements and their architecture dictates RNAP occupancy at each promoter and the attendant isomerization rates between each of the intermediates and so sets the intrinsic level of transcription initiation. Transcription factors are thought to compensate for suboptimal sequence or spacing of promoter elements whose regulatory task it is to allow specific promoters to respond under appropriate conditions at appropriate rates. Some transcription factors, best exemplified by catabolite activator protein at the lac promoter in E. coli, interact with the ␣ subunit of RNAP to recruit the holoenzyme, increasing the number of RNAP-promoter complexes formed and therefore the amount of transcripts produced (Ref. 5, but see Ref. 6). Others, the canonical example being cI at P RM , facilitate open complex formation through contacts with the subunit of RNAP without stimulating holoenzyme binding (7)(8)(9)(10). It has been proposed that cI might facilitate initiation by counteracting the tendency of to disengage from the Ϫ35 element during isomerization (11). We have studied the effects of Spo0AϳP at the A -dependent promoter of the spoIIG operon, which encodes E , one of the first of the sporulation-specific sigma factors in B. subtilis, as representative of the class of transcription factors that facilitate open complex formation without recruiting RNAP.
Spo0A is a response regulator controlling a genetic network that governs commitment to sporulation in B. subtilis (12). It lies at the terminus of a phosphorelay, an extended two-component signal transduction system whose complexity is indicative of the diversity of signals that must be integrated prior to the activation of this protein (13)(14)(15)(16)(17). Like most other response regulators, Spo0A is a transcription factor whose activity is modulated by reversible phosphorylation of an aspartate residue within the highly conserved N-terminal receiver domain (18,19). Although not well understood mechanistically, phosphorylation of this module enhances the affinity of the protein for its DNA-binding site, the 0A box, 5Ј-TGTCGAA-3Ј (20,21). Improved DNA binding permits transcription modulation by the C-terminal domain, which may either activate or repress transcription depending on the orientation and position of the 0A boxes (18). At the spoIIG promoter, Spo0AϳP stimulates transcription (20,22).
The spacing of the spoIIG promoter elements is unusual; the transcription start site is located 2 bp further downstream from the Ϫ10 element than at an ideal promoter, and 22 bp instead of the optimal 17 separate the Ϫ35 and Ϫ10 elements. The latter characteristic forms the basis of the requirement for activated Spo0A to stimulate transcription (23). In vitro RNAP binds readily, albeit weakly, to this promoter but on linear templates requires Spo0AϳP to initiate efficiently (22,24). Normal regulation appears to be a consequence of Spo0AϳP binding to a pair of 0A boxes located between Ϫ53 and Ϫ37 relative to the start site of transcription (ϩ1). The promoter proximal of these boxes completely overlaps the Ϫ35 element used by RNAP, and the contacts between Spo0A and the A subunit of RNAP necessary to stimulate transcription from this promoter have been defined (25)(26)(27)(28).
Kinetic analysis of in vitro transcription initiation has shown that Spo0AϳP increases the overall transcription rate by increasing the rate of isomerization of RNAP-promoter complexes to the initiated state but has no apparent effect in recruiting RNAP to the promoter (24). Moreover, structural studies have found that the addition of Spo0AϳP and RNAP induces DNA strand separation between Ϫ14 and Ϫ3 at the wild type spoIIG promoter and that the activator requirement can be bypassed by artificial strand separation using heteroduplex templates implying that Spo0AϳP cooperates with RNAP to denature the DNA (29). In this communication we have used an in vitro transcription assay to compare the effects of Spo0AϳP on the temperature dependence and the rate of initiated complex formation using an extended series of heteroduplex templates with decreasing numbers of denatured bases. The data indicate that the thermodynamic properties of initiation are driven by transitions involving only RNAP and the DNA and that Spo0AϳP stimulates the rate at which RNAP establishes interactions with the nontemplate strand of the Ϫ10 element.

EXPERIMENTAL PROCEDURES
Heteroduplex Construction-Heteroduplex templates were created as described earlier (29). A PCR product was generated using Taq polymerase (Invitrogen), the template pUCIIGtrpA (22), a mutagenic primer complementary to the nontemplate strand, SS3 (5Ј-CAGAGCT-TGCTTATATCTTATGAAGCAAGAAGGGG-3Ј; purchased from the Nucleic Acid and Protein Service Unit, University of British Columbia), and a downstream primer (IIG2), which anneals to the coding strand adjacent to the BamHI site. The primer SS3 contained nucleotides that were identical to the template strand from Ϫ14 to Ϫ11. The PCR product was cloned into the pGEM-T vector (Promega) to create pSS3, and the inserts were verified by sequencing. The HindIII-AluI fragment was isolated from pUCIIGtrpA and ligated to the AluI-BamHI fragments from pSS3, and the ligation mixture was used as a template for PCR by using IIG2 and an upstream primer, IIGA, which anneals to the noncoding strand adjacent to the HindIII site. The resulting product was ligated into pGEM-T to create pSS3IIG. The E. coli strain DH5␣ was used for all transformations, and plasmid preparations were performed as described by Sambrook et al. (30).
The plasmid pSS3IIG was digested with HindIII and BamHI, and the promoter-bearing fragment ligated into pBluescript SK ϩ vector (Stratagene) was digested with the same enzyme to create pSK3IIG ϩ . The wild type spoIIG promoter had previously been cloned into pBluescript SK ϩ and pBluescript SK Ϫ to create pSKIIG ϩ and pSKIIG Ϫ (29).
The pSKIIG ϩ plasmid was used as the template in a PCR to create pSK6IIG ϩ and pSK7IIG ϩ . The reactions composed of primer pairs, IIG6NT (5Ј-CCCACAGAGCTTGCTTATATGATATGAAGCAAGAAGG-G-3Ј) and IIG6T (5Ј-CCCTTCTTGCTTCATATCATATAAGCAAGCTCT-GTGGG-3Ј) or IIG7NT (5Ј-CCCACAGAGCTTGCTTATTACTTATGAA-GCAAGAAGGG-3Ј) and IIG7T (5Ј-CCCTTCTTGCTTCATAAGTAATA-AGCAAGCTCTGTGGG-3Ј), along with 2.5 mM dNTPs, 1ϫ Pfu Turbo buffer, and 1 unit of Pfu (Stratagene) were subjected to 18 cycles of 30 s at 95°C, 1 min at 55°C, and 6 min at 68°C. The PCR products were treated with two successive additions of 10 units of DpnI and ethanolprecipitated after the addition of 0.5 g of salmon sperm DNA. The DNA was resuspended in water and used to transform E. coli DH5␣. Mutations in transformants were verified by sequencing. Singlestranded DNA was produced and purified, and the heteroduplex templates were generated as described earlier (29). The templates are named to reflect the fact that they contain the DNA sequence from the nontemplate strand on both stands in the single-stranded region. The name also indicates the boundaries of the single-stranded region (see Fig. 1). For consistency, templates MB12NT and MB8NT that were described earlier were renamed NT14/3 and NT14/7, respectively.
Template DNA-The spoIIG and IIG17 templates were isolated from the plasmids pUCIIGtrpA and pUCIIG17trpA, respectively, as 415-bp fragments generated by digestion with PvuII and BamHI. These fragments contained the promoter and both upstream Spo0A-binding sites but lacked the trpA terminator resulting in run-off transcripts of ϳ130 bp. The fragments were isolated by electrophoresis, recovered using a QIAquick gel extraction kit (Qiagen), and stored in 10 mM HEPES (pH 7.9), 20 mM potassium acetate, and 0.1 mM EDTA at 4°C. The concentration of template fragments was determined by measuring absorbance at 260 nm. After annealing the single-stranded DNA, heteroduplex templates were treated in an identical fashion. The plasmid pUCIIG17trpA contains a modified promoter in which 5 bp have been deleted between the Ϫ35 and Ϫ10 consensus elements (see Fig. 1). 2 In Vitro Transcription Assays-RNAP was prepared from mid-log phase B. subtilis cells as described by Dobinson and Spiegelman (31). Spo0A and phosphorelay proteins were prepared as previously described (32)(33)(34), as was the activation of Spo0A (22).
Transcription assays were carried out by composing 8 l of an initiation mix that contains: 1 l of 10ϫ transcription buffer (24), 2 l of 20 nM template DNA, 2 l of 4 mM Spo0AϳP, 600 nM ATP, 50 M GTP, and 3 Ci of [␣ 32 P]GTP (800 Ci/mmol; PerkinElmer Life Sciences). The reaction tubes were incubated at the indicated temperature for 90 s before the addition of 1 l of RNAP. After the indicated preincubation period, the complexes formed were challenged with the addition of 1 l of a mixture containing 100 g/ml heparin, 600 nM UTP, and 600 nM CTP to allow RNA elongation. The reaction was stopped after 5 min with 5 l of loading buffer, and the transcripts were separated by denaturing electrophoresis. The transcripts were detected by autoradiography using Kodak XAR film overnight at Ϫ20°C, and promoter activity was quantified on a PhosphorImager SI (Molecular Dynamics; Amersham Biosciences) using ImageQuant 5.2 software.
The percentage of templates transcribed was calculated by dividing the moles of 131-bp spoIIG transcript produced by the moles of template added to the reaction. The transcript produced was calculated from Cerenkov radiation in an excised gel slice containing the transcript, the number of G residues/transcript, and the specific activity of the [␣ 32 P]GTP in the reaction.
The apparent van't Hoff enthalpy was calculated by taking advantage of the thermodynamic equivalence of the Gibb's free energy func- , and plotting lnK eq as a function of T Ϫ1 yields a curve with a slope equivalent to Ϫ(⌬H o /R). The tangent to the resulting curve at T m Ϫ1 permits the calculation of ⌬H o (35)(36)(37)(38).

RESULTS
We have examined the influence of Spo0AϳP on the events underlying transcription initiation by using an extended series of templates that contain the nontemplate strand sequence on both strands over defined regions within the spoIIG promoter. In addition to those templates described earlier (29), the series included templates that were single-stranded at positions Ϫ14 to Ϫ9 inclusive relative to the transcription start site (NT14/9), Ϫ14 to Ϫ11 (NT14/11), or at positions Ϫ14 and Ϫ13 (NT14/13). These data were compared with those obtained using a pair of fully double-stranded promoters, the wild type spoIIG promoter, and a variant, IIG17, in which the spacer length has been reduced to a consensus 17 bp instead of the 22 bp observed in the wild type. The spoIIG promoter sequence and a diagram of the seven templates used in this study are shown in Fig. 1. Several transcripts were observed by denaturing polyacrylamide gel electrophoresis of the products from the in vitro transcription reactions using some of these templates. In the following analysis we discuss only the major transcript that had been shown to represent initiation at the position seen on wild type spoIIG in the presence of Spo0AϳP (29). Figs. 2A and 3A show autoradiograms of the section of the polyacrylamide gels containing the major transcript.
Levels of in Vitro Transcription from Heteroduplex Templates-RNAP alone directed only a low level of transcription from a template created by annealing fully wild type spoIIG DNA strands (Table I), whereas transcription was increased 6 -50-fold for all the heteroduplex templates tested. In general, the amount of transcription from the predenatured templates in the absence of Spo0AϳP increased with the amount of single-stranded DNA. The exception, NT14/7, also gave rise to a transcript that utilized the opposite strand of the heteroduplex, which would be predicted to obstruct RNAP binding and reduce the amount of the appropriate transcript generated. The relative affinity of RNAP for the heteroduplex templates as determined by a gel mobility shift paralleled the relationship between the templates described above (data not shown). In addition, the fact that transcription initiation from the IIG17 promoter was independent of Spo0AϳP confirms the long held notion that Spo0AϳP is required to overcome the overlong spacer of the wild type promoter.
Effects of Spo0AϳP on in Vitro Transcription from Heteroduplex Templates-Because the influence of Spo0A is obscured by the variation in promoter strength, we calculated the fold stimulation of transcription by Spo0AϳP as the most meaningful comparison between the heteroduplexes. Spo0AϳP stimulated the level of transcription from the wild type spoIIG promoter ϳ20-fold (Table I). By comparison, transcription from IIG17 and from the templates with the three largest predenatured regions, NT14/3, NT14/7, and NT14/9, was independent of Spo0AϳP. Spo0AϳP stimulated transcription from templates with the two smallest predenatured regions, NT14/11 and NT14/13, by 2.8-and 3.5-fold, respectively. This sharp transition to Spo0A dependence as the single-stranded region shrank from 6 to 4 bp (NT14/9 to NT14/11) indicated that the effect of Spo0A was limited to stages preceding the complete exposure of the nontemplate strand of the Ϫ10 element.
We tested a variety of other templates. Templates containing mismatches at Ϫ28 and Ϫ27 supported only low levels of transcription by RNAP alone and were indistinguishable from the wild type promoter in their properties (data not shown). This suggested that the single-stranded regions do not simply increase template flexibility that might help to align the Ϫ35 and Ϫ10 sequences on the spoIIG promoter with the appropriate regions in the subunit. Furthermore, heteroduplexes containing the template strand sequence on both strands exhibited reduced levels of transcription that correlated with the extent of nontemplate strand Ϫ10 element consensus sequence retained (data not shown). These results are consistent with the interpretation that interactions between A and singlestranded DNA from the nontemplate strand of the Ϫ10 element were essential for holoenzyme-directed transcription. Temperature Dependence of Transcription from Heteroduplex Templates-Because Spo0AϳP is required to form an open complex at spoIIG (29) and because separation of the DNA strands is widely believed to be an energy-intensive process (39,40), we reasoned that Spo0A might facilitate open complex formation by reducing an energetic barrier to transcription. It seemed possible then that lower reaction temperatures might enhance the dependence of transcription on Spo0AϳP even on templates with denatured regions. Denaturing additional base pairs could also require additional energetic input so that transcription from templates containing smaller predenatured regions might be more sensitive to reduced temperature. Consequently, we investigated the effect of Spo0AϳP on the temperature dependence of transcription from each of these templates.
Without Spo0AϳP transcription from the fully duplexed wild type spoIIG promoter was low at all temperatures, although a slight increase in transcription with increasing temperature could be detected (Fig. 2B). Even in the presence of Spo0AϳP, below 22°C the level of transcription from the wild type spoIIG promoter was less than 4% of that observed at 37°C. The amount of transcription increased dramatically between 27 and 37°C, and the transition temperature (T m ) at which point the transcription was half-maximal was ϳ30°C (Table I). This value was higher than expected from studies with E. coli RNAP where the T m is typically 25°C under similar reaction conditions (41).
We had anticipated that reducing the overlong spacing at the spoIIG promoter would both obviate the requirement for Spo0AϳP and, because of the appropriate phasing of the conserved promoter elements, would be less energetically demanding. Although transcription from IIG17 was independent of Spo0A ( Fig. 2C), the thermal profile of transcription from IIG17 was indistinguishable from spoIIG, both in terms of its response to temperature (the shape of the curve) and in its T m (also 30°C; Table I). Spo0AϳP had no effect on the temperature dependence of transcription from IIG17.
In contrast to the low levels of transcription from the fully double-stranded promoters at reduced temperatures, appreciable amounts of transcription from NT14/3 were observed at temperatures as low as 12°C (Fig. 2H). The amount of transcription increased comparatively gradually about 20% every 5 degrees between 12 and 22°C before leveling off and increasing only about 10% every 5 degrees between 27 and 37°C. Most notably, artificially denaturing the promoter between Ϫ14 and Ϫ3 reduced the T m for transcription from this template to ϳ20°C. Spo0AϳP appeared to have no effect; the temperature dependence of transcription in the presence or absence of Spo0AϳP appeared identical. The same thermal profile of transcription and T m were observed from a NT10/3 template (data not shown).
The remaining templates fell into two classes on the basis of FIG. 1. spoIIG promoter sequence and templates. A, nontemplate strand sequence of the spoIIG promoter from Ϫ60 to ϩ20. 0A boxes are underlined, and the Ϫ35 and Ϫ10 elements are denoted by bold type. The transcription start site (ϩ1) is marked by an arrow. B, representations of the structures of templates used in this study. The sequence removed from spoIIG during construction of IIG17 is shown above the template. Heteroduplexes contain the nontemplate sequence (NT) on both strands and are named to reflect the limits of the noncomplementary region. The sequence of the nontemplate (top strand) strand and the template (bottom strand) are shown beside the singlestranded regions. The Ϫ10 element and transcription start site are marked as above.

Spo0A Stimulation of Transcription
shared transition temperatures and Spo0AϳP dependence. Although there were modest differences in the relative amounts of transcription at any given temperature, within experimental uncertainty, NT14/7 and NT14/9 exhibited a common T m of 23-24°C and were Spo0AϳP independent (Fig. 2, G and F). In comparison, the T m values of NT14/11 and NT14/13 were sig- Transcription from the spoIIG template was plotted relative to transcription in the presence of Spo0AϳP at 37°C. Transcription from the other templates was plotted relative to transcription in the absence of Spo0A at 37°C. The reactions containing either 800 nM Spo0AϳP (closed circles) or an equivalent buffer lacking Spo0A (open circles), 4 nM DNA template, and the initiating nucleotides ATP and GTP in 1ϫ transcription buffer were prepared on ice. The reactions were incubated at the indicated temperature for 2 min, and then transcription was initiated with the addition of 1 l (400 fmol) of RNAP. After 2 min, a single round of elongation was permitted by the addition of heparin, UTP, and CTP. The transcription products were separated by electrophoresis, and the relative amounts of transcription were quantified as described under "Experimental Procedures." The graphs represent averages from four independent experiments. The errors are less than 10% of the plotted values. The transition temperatures for each of the templates are listed in Table I. DNA template, and the initiating nucleotides ATP and GTP in 1ϫ transcription buffer were prepared on ice. The reactions were incubated at 37°C for 2 min, and then transcription was initiated with the addition of 1 l (400 fmol) of RNAP. At the indicated times, samples were withdrawn and added to a 1-l mixture containing heparin, UTP, and CTP. Elongation was allowed to proceed for 5 min before the reactions were terminated, and the transcription products separated by electrophoresis. Relative amounts of transcription were quantified as described under "Experimental Procedures." The graphs represent averages from three independent experiments. The errors are less than 8% of the plotted values. The time required to achieve maximal formation of initiated complexes and the estimated effect of Spo0AϳP in stimulating the rate of this process are listed in Table I. nificantly higher (28°C), slightly less than that observed at the fully double-stranded promoters (Fig. 2, E and D). Critically, inclusion of Spo0AϳP did not shift the T m , indicating that the activator had no effect on the energetics of the overall transcription reaction from these templates. Spo0AϳP appeared to affect the amount of transcript formed but not the thermal energy required to initiate transcription.
Rates of Initiated Complex Formation on Heteroduplex Templates-Heparin resistance has been used to discriminate between stable and unstable RNAP-promoter complexes. At the spoIIG promoter, heparin resistance requires the synthesis of an ApApG trimer and thus represents a relatively late stage in the initiation reaction (24). In practice, inclusion of Spo0AϳP, ATP, GTP, and the spoIIG promoter results in the formation of an 11-mer RNA. As a means of investigating the effect of denatured base pairs on the formation of initiated complexes at the spoIIG promoter, we examined the rates of formation of heparin-resistant complexes on the series of predenatured templates.
In the absence of Spo0AϳP, the formation of heparin-resistant initiated complexes at spoIIG was negligible regardless of the incubation time (Fig. 3B). When Spo0AϳP was present, formation of elongation-competent complexes increased in an approximately linear fashion for the first 15 s. By 30 s ϳ60% of the maximal number of initiated RNAP-spoIIG-Spo0AϳP complexes had formed, and the proportion of resistant complexes increased to just less than 90% of the maximal level within the first 60 s. The formation of initiated complexes at the IIG17 promoter occurred with an initial rate only slightly faster than that at wild type spoIIG in the presence of Spo0AϳP (Fig. 3C and Table I). Although inclusion of Spo0AϳP resulted in a small increase in the amount of initiated complexes formed, this effect was negligible.
The heteroduplex templates tested fell into two distinct categories. Initiated complexes formed extremely rapidly on NT14/3 (Fig. 3G) and NT14/9 (Fig. 3F). Maximal levels of initiated complexes were formed in less than 10 s, and this process occurred independently of Spo0AϳP. By comparison, heparin-resistant initiated complexes formed more slowly on NT14/11 (Fig. 3E) and NT14/13 (Fig. 3D) than on the templates with larger denatured regions (about 30s) but more rapidly than on the fully double-stranded promoters. The effects of Spo0AϳP at NT14/11 and NT14/13 were 2-fold. First, although RNAP alone was able to isomerize to a heparin-resistant form on NT14/11 and NT14/13, Spo0AϳP stimulated the initial rate of this transition about 4-fold in each case. Second, Spo0AϳP increased the total number of complexes that acquired heparin resistance on the heteroduplex templates containing the smallest denatured regions.

DISCUSSION
In this work, a series of heteroduplex templates of the spoIIG promoter was used in an in vitro transcription assay to investigate the stage(s) during transcription initiation affected by Spo0AϳP. Spo0AϳP did not affect transcription from the 12-, 8-, or 6-bp heteroduplex templates (NT14/3, NT14/7, and NT14/ 9); however, it did stimulate transcription from templates with 4 or 2 bp denatured (NT14/11 and NT14/13). This transition from Spo0AϳP dependence to independence occurred as the denatured region extended to include the entire nontemplate sequence of the Ϫ10 element. In addition, the rate of formation of initiated complexes increased as the denatured region increased (Table I). Finally, we found that templates with increasingly large denatured regions were transcribed at lower temperatures, characterized by the mid-point of the temperature dependence of transcription (the T m ).
Although Spo0AϳP enhanced the amount of transcription from spoIIG, NT14/13, and NT14/11 templates, it had no effect on the T m of any of the templates that generated enough transcripts to be reliably quantified. Thus Spo0AϳP did not reduce the overall energetic barrier to initiation of transcription at the spoIIG promoter. In addition, transcription from the Spo0AϳP independent variant, IIG17, had a T m of 30°C, identical to that of wild type spoIIG in the presence of Spo0AϳP. These findings eliminate the possibility that Spo0AϳP stimulates transcription from the wild type spoIIG promoter by lowering the activation energy required to denature the promoter. Although DNA melting is the major energetic barrier to transcription initiation at this promoter, the energetics of promoter denaturation are apparently a consequence of transitions required of the RNAP-spoIIG complex alone. Because the ratelimiting step correlates with the exposure of the six bases comprising the Ϫ10 element, we suggest that the role of Spo0AϳP is to facilitate the interaction of A with the nontemplate strand of the Ϫ10 element.
The crystal structure of the activation domain of Spo0A has been determined both for free protein (42) and for the activation domain complexed with DNA (43). The activation domain contains a helix-turn-helix DNA-binding motif, a central bundle of three ␣-helices, and a helix protruding from one side (␣E), connected by flexible regions to adjacent helices. When a dimer of Spo0A sits on the DNA, the central axis of the ␣E helix is nearly parallel to the axis of the DNA helix (43). In addition, the sequence conservation between the 4.2 regions of B. subtilis

Spo0A Stimulation of Transcription
A , E. coli 70 , and the fragment of Thermus aquaticus A that was crystallized with a Ϫ35 element (44) is high, so that a reasonable picture of how the two proteins might interact can be modeled. Inspection of the structures reveals that Spo0AϳP and the 4.2 region of A are expected to make contact with overlapping and in some cases identical moieties on the DNA (Fig. 4). We have shown that RNAP rapidly binds to the spoIIG promoter in vitro to protect the Ϫ35 sequence but not the Ϫ10 sequence (24), and we presume it does so in vivo. Thus to allow Spo0AϳP binding to the 0A boxes to stimulate transcription, RNAP must be displaced from the Ϫ35 consensus sequence.
The laboratory of Moran and co-workers (45) recently used molecular modeling to examine co-occupancy of the spoIIG promoter by Spo0AϳP bound at the 0A boxes and the 4.2 region of A . They suggest that when a dimer of Spo0A is bound to the 0A boxes in the region of the Ϫ35 sequence, RNAP can be accommodated with the 4.2 region of A binding to a DNA sequence centered at Ϫ33. This sequence could potentially be a low consensus version of a Ϫ35 sequence. Under the modeling conditions, a glutamate (Glu 221 ) in the ␣D helix (the recognition helix of the helix-turn-helix motif) is in close contact with Arg 355 in the 4.2 region of A . The in vivo effects of site-directed mutations lend strong credence to the model that Glu 221 contacts Arg 355 in A during stimulation of transcription activation.
Analyses of spo0A mutations that specifically eliminate transcription activation of A found that the majority of the mutants were found either in ␣E or in the flexible regions adjacent to ␣E (25)(26)(27)(28). ␣E has been termed the A activation region (42), although it is not where Glu 221 described above lies. The mutations found in ␣E were of two types. One type, represented by five mutations, would be predicted to reduce side chain interactions (I229A, E230A, I232A, F236S, and V240A). The second type, represented by one mutation, would be predicted to alter the structure of the helix (S233P). A mutation in ␣E (S231F) suppresses H359A, H359R, and K356E mutations in A (27), but the mechanism is unclear, and the region of A contacted by ␣E is unknown. Potentially there may be two regions of interaction between Spo0A and A , as in the case of PhoB-70 interaction in E. coli (44).
Because Spo0AϳP and A appear to contact identical DNA residues and because RNAP initially binds at the Ϫ35 sequence and must translocate to contact the Ϫ10 sequences, transcription initiation at this promoter must involve remodeling of the initial RNAP-promoter complex. We can envisage several mechanisms whereby the initial binding of RNAP to the Ϫ35 region is reversed and translocation downstream is stimulated.
In a "simple competition" model, the binding of Spo0AϳP dimers and of RNAP are in rapid reversible equilibrium with unbound protein. As Spo0AϳP dimers accumulate, they outcompete RNAP binding to the Ϫ35 sequences, with the result that RNAP binds to an alternate site, possibly downstream at the Ϫ33 region indicated by Moran's group. When both Spo0AϳP and RNAP bind to the DNA interactions between them would stabilize RNAP situated over the Ϫ10 element so that strand separation can be initiated.
Alternatively, an "active displacement" model can be imagined where repulsive interactions between Spo0AϳP dimers and RNAP force the polymerase to release the Ϫ35 region. This model recognizes that RNAP binds readily on its own to the Ϫ35 region of the promoter, but at present, the mutations affecting Spo0AϳP-stimulated A -dependent transcription suggest most of the interactions would be stabilizing, rather than repulsive. A third "release and tether" model proposes that Spo0AϳP promotes release of RNAP contacts with DNA sequences at or upstream of the Ϫ35 element and also prevents RNAP from diffusing away from the spoIIG promoter when it releases the Ϫ35 region. In this mechanism Spo0A stabilizes a transition by RNAP that would not be expected to be energydependent. This event could also be described by a "ping-pong" type mechanism in which as A releases the Ϫ35 region Spo0AϳP binds to both A and the 0A boxes, tethering RNAP to the promoter and thereby increasing the rate at which RNAP binds to the Ϫ10 region.
The differences between the transition temperatures of transcription for the heteroduplex templates allow an estimate of the energetic requirements for denaturing short segments of the spoIIG promoter by calculating the apparent van't Hoff enthalpy (here termed ⌬H app, txn ; see "Experimental Procedures") required to separate the strands. This determination of enthalpy assumes that RNAP-DNA binding equilibrates in the transcription reaction (24) and that the differences between the temperature dependence profiles reflect the thermodynamic  Table I and used to calculate the energy required during stepwise melting of the promoter (see text for details).
FIG. 4. Overlapping Spo0AϳP and A contacts at the spoIIG ؊35 element. A schematic representation of the spoIIG promoter from Ϫ46 to Ϫ34 is shown including contacts predicted to be made by A alone (light gray), Spo0AϳP alone (dark gray), or A and Spo0AϳP (black). Bases to which all specific contacts are made in the two structures are identical at the spoIIG promoter despite differences in the sequences of oligonucleotides used in each of the crystal structures. Predictions of A contacts were based on the structure from Campbell et al. (44), and predictions of Spo0AϳP contacts were based on Mol A-Box 1 interactions from Zhao et al. (43).
properties of the DNA melting. Using this approximation, the differences in ⌬H app, txn between initiation at spoIIG and NT14/13 or NT14/11 indicated that denaturation of the bases between Ϫ14 and Ϫ11 required 9 kcal mol Ϫ1 (Table I and Fig.  5). The differences in ⌬H app, txn between initiation at NT14/11 and NT14/9 or NT14/7 suggest that denaturation of the 4 bp between Ϫ10 and Ϫ7 required an additional 9 kcal mol Ϫ1 . Separating the DNA strands 4 bp further to denature Ϫ6 to Ϫ3 imposed an energetic cost of 11 kcal mol Ϫ1 because of the difference in ⌬H app, txn between NT14/7 and NT14/3. Thus formation of an open complex in which the DNA strands were denatured between Ϫ14 and Ϫ3 required 29 of the 42 kcal mol Ϫ1 required for the overall transcription initiation reaction at the spoIIG promoter. The latter value is similar to the values of 25 and 41 kcal mol Ϫ1 estimated for transcription from the T7 A1 and lacUV5 promoters, respectively (39,40), and with the estimate of 29 kcal mol Ϫ1 measured for open complex formation at P R between 25 and 37°C (41).
The apparent energy input for spoIIG promoter melting was strikingly discontinuous, and there does not appear to be a simple relationship between the type (A:T versus G:C) or number of base pairs melted and the energetic cost. This would seem to imply stepwise melting, as found for the B. subtilis flagellin promoter (46), and may reflect a series of conformational rearrangements in RNAP domains that enable precise control over DNA strand separation. Cross-links between ␤Ј and ␤ specific to the open complex and a mutant RNAP that forms open complexes that separate only the upstream 4 -6 bp independently support such a model (47)(48)(49).
Our results suggest that the thermodynamics of Spo0AϳP stimulation of transcription initiation at the spoIIG promoter are dominated by the interactions between the A subunit and the downstream promoter elements. We suggest that the role of Spo0AϳP is to promote alignment of A with these sequences by two possible mechanisms: 1) stimulating release of upstream contacts and 2) holding RNAP near the DNA after it releases Ϫ35 element contacts. Neither of these steps would be expected to be energy-dependent. Further investigation is required to understand the relative contributions, if any, of these two actions in the stimulation of initiation by Spo0AϳP.