Nucleotide-dependent triggering of RNA polymerase-DNA interactions by an AAA regulator of transcription.

Enhancer-dependent activator proteins, which act upon the bacterial RNA polymerase containing the sigma54 promoter specificity factor, belong to the AAA superfamily of ATPases. Activator-sigma54 contact is required for the sigma54-RNAP to isomerize and engage the DNA template for transcription. How ATP hydrolysis is used to trigger changes in sigma54-RNA polymerase and promoter DNA that lead to DNA opening is poorly understood. Here, band shift and footprinting assays were used to investigate the DNA binding activities of sigma54 and sigma54-RNA polymerase in the presence of the activator protein PspF bound to poorly hydrolysable analogues of ATP and the ATP hydrolysis transition-state analogue ADP.AlFx. Results show that different nucleotide-bound forms of PspF can change the interactions between sigma54, sigma54-RNA polymerase, and a DNA fork junction structure present within closed promoter complexes. This provides evidence that in the activation transduction pathway, several functional states of the activator, prior to ATP hydrolysis, can serve to alter the fork junction binding activity of sigma54 and sigma54-RNA polymerase that precede full DNA opening. A sequential set of nucleotide-dependent transitions in sigma54-RNA polymerase promoter complexes needed for productive open complex formation may therefore depend upon different nucleotide-bound forms of the activator.

Transcription initiation is a major point at which gene expression is regulated. Within the three kingdoms of life, structurally conserved multisubunit DNA-dependent RNAPs 1 catalyze transcription and require auxiliary factor(s) to facilitate and regulate transcription. In bacteria, a core RNAP (␣ 2 ␤␤Ј; E) associates with one of several sigma () factors to form a functional holoenzyme. The subunit confers promoter specificity to the bacterial RNAP (reviewed in Refs. 1 and 2). In Escherichia coli and many other bacteria, two types of RNAP holoenzymes exist: one that utilizes the vegetative factor 70 , or one of the five alternative 70 -like factors, and an enhancerdependent one that uses the 54 factor (reviewed in Refs. 3 and 4). The key difference between the 70 and the 54 types is the conversion step from closed promoter-RNAP complexes to productive open complexes. For the 70 family, this conversion can occur spontaneously without the involvement of activators. For the enhancer-dependent 54 -RNAP, the transition from closed complex to open complex strictly relies upon the mechanochemical activity of enhancer-binding activators that belong to the ATPases associated with various cellular activities (AAA) family (reviewed in Refs. 4 and 5).
factors bind double-stranded promoter DNA and promoter DNA structures with single-stranded and doublestranded DNA juxtaposed (fork junctions) that mimic the conformational state of the promoter DNA as it exists within the transcription bubble. In the case of enhancer-dependent closed complexes, two base pairs at positions Ϫ12 and Ϫ11 (where DNA melting originates), which are adjacent to the transcription start site proximal promoter element (GC region; Table Ia), are transiently melted (6). Interactions between 54 and this transient fork junction structure are repressive, keeping the 54 -RNAP silent for transcription by inhibiting its ability to melt DNA and isomerize to form open complexes (7,8). 54 -RNAP binds tightly to the template strand of this transient fork junction structure and as a consequence fails to make interactions with the non-template strand adjacent to the Ϫ12 fork junction. The latter interaction is a critical feature of open complex formation by enhancerdependent and -independent RNAPs (8,9).
The amino-terminal region 1 of 54 makes a major contribution to binding the fork junction structure at Ϫ12 (7, 8, 10 -12). This nucleoprotein organization constitutes the direct binding target for AAA activators and is known as the regulatory center (12,13). The AAA activators of the 54 -RNAP couple the energy derived from ATP hydrolysis to remodel the regulatory center, relieving the inhibitory interactions with the template strand of the fork junction structure at Ϫ12 and promoting new interactions with the adjacent non-template strand required for forming open complexes (5,8,11,14).
Open complex formation by the 70 -RNAP and 54 -RNAP proceeds via several intermediate states involving large conformational changes in the RNAP (15). However, very little information exists on the mechanisms that trigger changes in RNAP and promoter DNA that lead to transcription. Analysis of the mechanisms of transcription initiation by the 54 -RNAP will extend our understanding further on the functioning of the RNAP as a complex molecular machine. Recently, we demonstrated that it is possible to "trap" 54 or the 54 -RNAP bound to promoter DNA with the E. coli AAA activator phage shock protein F (PspF) in the presence of ADP aluminum fluoride (ADP⅐A1F x ), an analogue of ATP at the point of hydrolysis (13). Using heteroduplex forms of the Sinorhizobium meliloti nifH promoter probes containing fork junction structures at Ϫ12 and, for experimental simplicity, only the AAA domain of PspF (PspF 1-275 ) ( Table Ib), we now report that nucleotide-dependent action of AAA activators on 54 or 54 -RNAP significantly changes the interactions made with fork junction DNA at Ϫ12. Altered activator-dependent promoter DNA binding activities of 54 and 54 -RNAP were observed in the presence of poorly or non-hydrolyzed forms of ATP (adenosine 5Ј-(␥-thio)triphosphate (ATP␥S) and ADP⅐AlF x , respectively), suggesting that several nucleotide bound states of the AAA activator, each capable of remodeling 54 -RNAP prior to completion of the ATP hydrolysis cycle, exist and that ATP hydrolysis per se is not needed for the partial remodeling of the regulatory center. Evidence that a mixed nucleotide-bound (ATP and ATP␥S) state of the activator resulted in altered remodeling activity was also obtained. Overall the results provide clear evidence that discrete functionalities are associated with different nucleotide bound states of PspF, which orchestrate open complex formation by the 54 -RNAP during the ATP hydrolysis cycle by remodeling 54 -RNAP-fork junction interactions. Hence by inference we suggest that other AAA proteins will have several distinct nucleotide-dependent functional states.
Native Gel Mobility Shift Assays-Binding reactions were conducted at 30°C in STA buffer (25 mM Tris acetate, pH 8.0, 8 mM Mg-acetate, 100 mM KCl, 1 mM dithiothreitol and 3.5%(w/v) PEG-6000). Where indicated, 54 , ⌬I 54 , or mutant 54 were present at 1 M and the 54 -RNAP holoenzyme at 100 nM (formed with 1:4 ratio of core RNAP to 54 ). 54 proteins or 54 -RNAP were incubated for 5 min with 16 nM promoter probe and nucleotides as required: ADP⅐AlF x (formed in situ by the addition of 0.2 mM AlCl 3 to a mixture containing 0.2 mM ADP and 5.0 mM NaF; Ref. 13), ATP (2 mM), ATP␥S (2 mM, unless otherwise stated), or ATP␥S/ATP mixture (0.2/2 mM, unless otherwise stated). After addition of 10 M PspF 1-275 (unless otherwise stated) the reactions were incubated for a further 10 min prior to resolving on a 4.5% native polyacrylamide gel. Native gels were run in 25 mM Tris, 200 mM glycine buffer (pH 8.6). Complexes were detected by PhosphorImager analysis. The error range for the amount of DNA band shifted was Ϯ5%, depending upon the 54 and ATP␥S preparations used. For comparative purposes when using variants in PspF or DNA the same 54 and ATP␥S preparations were used.
DNA Footprinting Assays-Binding reactions (10 l) were conducted as described above but with 100 nM promoter DNA probe and in STA buffer without dithiothreitol. Footprinting reagents were added as described (11,14), reactions were terminated, and bound and unbound DNAs were separated on native polyacrylamide gels. DNA was then excised, processed, and analyzed on a denaturing 10% polyacrylamide gel. For ortho-copper phenanthroline (OP-Cu) footprinting, 0.5 l of a solution of 4 mM ortho-phenanthroline, 0.92 mM CuSO 4 , and 0.5 l of 0.116 M mercaptoproprionic acid were added to the reactions and incubated for 2 min. The reactions were terminated by the addition of 1 l of 28 mM 2,9-dimethyl-1,10-phenanthroline. For DNase I footprints, 1.75 ϫ 10 Ϫ3 units of enzyme (Amersham Biosciences) was added to the reactions for 1 min, followed by addition of 10 mM EDTA to stop cutting. For KMnO 4 footprinting, 4 mM fresh KMnO 4 was added for 30 s, followed by 50 mM ␤-mercaptoethanol to quench DNA oxidation. Gelisolated DNA was eluted into 0.1 mM EDTA (pH 8) (for DNase I footprinting) or H 2 O (for OP-Cu and KMnO 4 footprinting) overnight at 37°C. KMnO 4 -oxidized DNA was cleaved with 10% (v/v) piperidine at 90°C for 20 min. Recoveries of isolated DNA were determined by dry Cerenkov counting and equal numbers of counts were loaded onto 10% sequencing gels. Dried gels were visualized and quantified using a PhosphorImager.

ADP⅐AlF x Form of PspF 1-275 Changes the DNA Binding
Activities of 54 and 54 -RNAP DNA footprinting data suggest that the stable binding between the ADP⅐AlF x associated form of PspF 1-275 (PspF 1-275 -ADP⅐AlF x ) and promoter-bound 54 or 54 -RNAP changes the DNA interactions made by 54 or 54 -RNAP (13). 2 To further explore these alterations in DNA interactions we analyzed the ability of 54 and 54 -RNAP to bind a range of S. melitloti nifH promoter probes (that mimicked the conformation of the DNA within closed promoter complexes) by native-PAGE in the absence and presence of PspF 1-275 -ADP⅐AlF x . The promoter probes differed in the template strand sequence in the region of heteroduplex at the Ϫ12/Ϫ11 fork junction (Fig. 1a).
As shown in Fig. 1a, 54 does not detectably bind to promoter probes with mutant or missing template strand sequences at position Ϫ12 (probes 2-9; Ref. 11). In contrast, strong binding of 54 and 54 -RNAP was evident on probe 1, which contains wild-type template strand sequences within the heteroduplex segment (Table Ia and Fig. 1, a and b, respectively). Strikingly, in the presence of PspF 1-275 -ADP⅐AlF x , the binding activity of 54 to probes 2, 7, and 8 was significantly increased, suggesting that the stable complex formation between 54 and PspF 1-275 -ADP⅐AlF x (hereafter called the trapped 54 ) has altered the DNA-binding activity of 54 (Fig. 1a). Similarly, the promoter DNA-binding activity of the 54 -RNAP was altered in the presence of PspF 1-275 -ADP⅐AlF x . As shown in Fig. 1b, we found that the PspF 1-275 -ADP⅐AlF x -bound form of 54 -RNAP (hereafter called the trapped 54 -RNAP) bound to probes 2, 4, 7, and 8 more efficiently than the untrapped 54 -RNAP. Overall, the results clearly demonstrate an apparent activator-dependent change in the DNA-binding activity of 54 and 54 -RNAP that critically involves PspF 1-275 -ADP⅐AlF x . Furthermore, as shown in Fig. 1, a and b, the binding of 54 and 54 -RNAP to the promoter probe with wild-type template sequences at positions Ϫ12 and Ϫ11 (probe 1) was unchanged in the presence of PspF 1-275 -ADP⅐AlF x , suggesting that the activator does not interact to produce an overall general increase in DNA binding. Rather, a change in binding activity is evident only on certain heteroduplex promoter probes. Controls where ADP⅐AlF x or PspF 1Ϫ275 was omitted showed that the change in DNA-binding activity was specific to conditions that led to formation of the stable association of PspF 1-275 -ADP⅐AlF x with 54 or 54 -RNAP (data not shown). A hallmark of 54 -RNAP open complexes is their stability in the presence of heparin. The trapped 54 -RNAP complex bound to promoter probes 1-9 were no more heparin-stable than 54 -RNAP promoter complexes, indicating that the complexes had not fully isomerized to acquire heparin stability, but rather might represent an intermediate state (data not shown). DNA footprinting data on promoter probe 1 have established that compared with the binary 54 -DNA complexes, ADP⅐AlF xdependent trapped 54 -DNA complexes have extended DNase I footprints toward the transcription start site, as do ATP hydrolysis-dependent isomerized binary 54 -DNA complexes (11,13). This indicates that the extended 54 -DNA interactions occur in response to interactions with certain nucleotide-bound forms of the activator. We used the promoter probe with the mutant template strand sequence (GT at positions Ϫ12 and Ϫ11 instead of TG; probe 2), to which trapped 54 and 54 -RNAP bound most efficiently ( Fig. 1) to further examine the increased DNA binding functionality of trapped 54 and 54 -RNAP. By using a series of shortened variants of promoter probe 2 (probes 2A-2F; Table II), we explored whether the changed DNA-binding activity seen above ( Fig. 1) relied upon the transcription start site proximal DNA sequences. As shown in Table II, increased binding of the trapped 54 and trapped 54 -RNAP complexes to promoter probes that extend downstream of (and including) position Ϫ1 occurred (probes 2A-2D). However, no increased binding activity of trapped 54 and trapped 54 -RNAP to promoter probes lacking sequences downstream of the Ϫ1 position was observed (Table II). This suggests that sequences upstream of Ϫ1 are required for the increased binding activity of the trapped 54 and 54 -RNAP complex to promoter probe 2. Furthermore, binding assays with promoter probes 2G and 2H demonstrated that the DNA between positions Ϫ3 and Ϫ1 had to be in duplex form.

Binding of Trapped 54 and 54 -RNAP to Promoter Probe 2 Leads to Altered Interactions with Fork Junction DNA at Ϫ12
OP-Cu is a minor groove-specific DNA cleaving reagent that has been widely used to study 54 -RNAP interactions with the Ϫ12 promoter region (6,19). In closed promoter complexes formed with the 54 -RNAP on homoduplex promoter probe (Table Ia), the minor groove at the Ϫ12 position is distorted and thus is susceptible to cleavage by OP-Cu (6). OP-Cu-mediated DNA cleavage at this position is not evident in open promoter complexes suggesting that the distortion distinguishes closed and open promoter complexes (6). We used OP-Cu footprinting to probe whether binding of trapped 54 and trapped 54 -RNAP to promoter probe 2 changes 54 interactions with the fork junction structure at Ϫ12. To do so, we formed promoter complexes with 54 and 54 -RNAP on probe 2 in the absence and presence of PspF 1-275 -ADP⅐AlF x , treated them with OP-Cu, and separated the probe 2 bound and unbound complexes by native-PAGE. Promoter complexes were excised from native gels, the DNA was isolated and analyzed by denaturing-PAGE.
As shown in Fig. 2, OP-Cu treatment of naked promoter probe 2 revealed a hypersensitive site around position Ϫ12 (lane 2), as expected because of the heteroduplex segment at this position (Table Ia). This hypersensitive site was also evident in the weak complexes formed between 54 or 54 -RNAP and probe 2 in the absence of PspF 1-275 -ADP⅐AlF x (Fig. 2, lanes 5 and 6). This is striking because the OP-Cu-mediated DNA cleavage at Ϫ12 is significantly reduced in complexes formed between 54 or 54 -RNAP and probe 1 (which contains wildtype template strand sequences in the heteroduplex segment; Table Ia) because of the tight binding of 54 or 54 -RNAP to this heteroduplex segment (19). Interestingly, analysis of the complexes formed between trapped 54 or trapped 54 -RNAP and probe 2 showed that the DNA cleavage around position Ϫ12 is

Interaction of ATP␥S Bound PspF 1-275 with 54 Results in a Novel Promoter Complex
To increase our understanding of how new functional states of PspF 1Ϫ275 are created by nucleotide interactions, we explored whether changes in DNA-binding activity of 54 and 54 -RNAP in the presence of PspF 1-275 -ADP⅐AlF x are seen with other nucleotides. The poorly or non-hydrolyzable forms of ATP, ATP␥S, and adenosine 5Ј-(␤,␥-imido)triphosphate (AMP-PNP) do not support productive open complex formation by 54 -RNAP (20), but could influence prior steps. We compared the functional states of PspF 1Ϫ275 bound to ADP⅐AlF x , ATP␥S, or AMP-PNP in assays using promoter probes 1 and 2 (Fig. 3). Strikingly, the presence of the ATP␥S-bound form of PspF 1Ϫ275 (PspF 1-275 -ATP␥S) resulted in the formation of a novel complex on promoter probe 2 only (Fig. 3, lane 13). Interestingly, the novel complex migrated considerably faster than the trapped 54 -DNA complex and thus appeared to differ significantly from the latter (Fig. 3, compare lanes 12 and 13). The addition of AMP-PNP, ATP, or GTP did not substitute for ATP␥S in formation of the ATP␥S-dependent complex on promoter probe 2 ( Fig. 3 and data not shown). The complex (hereafter called the ATP␥S complex), like the trapped 54 -DNA complex relied upon promoter sequences between positions Ϫ3 and Ϫ1 (Table  II). Thus, these initial results demonstrate that a new 54 -DNA complex can be obtained that is dependent upon a certain fork junction structure at position Ϫ12 and PspF 1-275 -ATP␥S.

The ATP␥S-Complex Is a Novel Binary 54 -DNA Complex
At high concentrations of 54 and promoter probe 2, we observed a weak 54 -DNA complex that formed independently of PspF 1-275 -ADP⅐AlF x or PspF 1-275 -ATP␥S. This binary 54 -DNA complex migrated faster than the trapped 54 -DNA complex and the ATP␥S-dependent complex (data not shown). Us-ing activators of different molecular weights (full-length PspF 1-325 , PspF 1-292 , PspF 1-275 , and heart muscle kinasetagged PspF 1-292 ), we obtained no evidence that the ATP␥Scomplex was a ternary complex containing the activator (data not shown). We were also unable to demonstrate the presence of 32 P-PspF 1Ϫ292 in the ATP␥S-complex in reactions where sensitivity was not limiting (data not shown). From these results we conclude that PspF 1-275 -ATP␥S enables 54 to bind promoter probe 2 and form a stable binary complex that does not depend on the stable association of PspF  . Because the formation of the ATP␥S-complex and the trapped 54 -DNA complex relied upon promoter probe 2 and was independent of the order of addition of 54 , DNA, or PspF 1-275 (data not shown), we suggest that the PspF 1-275 -ATP␥S interacts transiently with 54 to change the DNA binding functionality of 54 .
The ADP⅐AlF x and ATP␥S forms of PspF 1Ϫ275 are therefore distinct in that only the ADP⅐AlF x -bound PspF 1-275 forms a stable complex with 54 (13), but both share the property of altering the DNA-binding activity of 54 on promoter probe 2.

Specificity of the ATP␥S-Complex
We investigated the requirements for ATP␥S-complex formation. Initial experiments in which increasing amounts of ATP, ADP, AMP, (NH 4 ) 2 SO 4 (to mimic the PO 4 at the point of ATP hydrolysis), AMP-PNP, guanosine 5Ј-(␥-thio)triphosphate (GTP␥S), or ATP␥S were present demonstrated that ATP␥Scomplex formation was specific to the ␥S forms of the trinucleotide (data not shown). As shown in Fig. 4a, the amount of ATP␥S-complex formation increased with ATP␥S, and saturated at 1 mM ATP␥S (20% ATP␥S-complex). Increased 54 -DNA binding was also observed when ATP␥S was substituted with GTP␥S (data not shown), although ATP␥S was the most effective in promoting binding of 54 to promoter probe 2 (ϳ5fold better than GTP␥S). The amount of ATP␥S-complex formation also titrated with PspF 1Ϫ275 and 54 (Fig. 4, b and c, respectively). Increasing amounts of DNA did not increase the fraction of DNA in the ATP␥S-dependent complex, suggesting that PspF 1-275 -ATP␥S acts on free 54 and not on its promoter DNA-bound form (data not shown). Furthermore, ATP␥S-complex formation was dependent upon the integrity of the GAFTGA motif of PspF 1-275 (residues 83-88), a signature motif in the AAA domain of 54 activators, which is involved in direct binding interactions with 54 (11,13,17). No ATP␥Scomplexes were detected when PspF 1-275 harboring the T86A, T86S, F85A, or F85L mutations within the GAFTGA motif were used (data not shown). Because these PspF 1-275 mutants were not defective for nucleotide binding, we conclude that the failure to form the ATP␥S-complex arises from a defective interaction with 54 and not defects in nucleotide binding per se. As expected, attempts to form the ATP␥S-complex with PspF 1-275 harboring mutations within motifs involved in nucleotide interactions or in ATP hydrolysis (Walker A and B motifs and the putative "arginine finger" residue) failed (data not shown).

Interaction of 54 with a Mixed Nucleotide-bound Form of PspF 1-275 Results in a Novel 54 -DNA Complex
Activators of the 54 -RNAP must form oligomeric structures to hydrolyze ATP (5,21,22). The state of the bound nucleotide within each protomer of the oligomer during active ATP hydrolysis is not known. However, we considered that differences could exist between protomers, and attempted to form the ATP␥S-complex in the presence of increasing amounts of ATP␥S plus either 4 mM ATP or AMP. As shown in Fig. 5a, we observed that the amount of ATP␥S-dependent complex formation was significantly increased in the presence of ATP, espe- cially at low concentrations of ATP␥S (e.g. 0.2 mM). This increase was not seen with AMP (recall ATP alone does not cause increased binding of 54 to promoter probe 2). This suggests protomers of PspF 1-275 differentially loaded with nucleotide were functioning together to either produce a more active oligomer, or produce differentially loaded oligomers that work together to remodel the DNA-binding activity of 54 . Strikingly, we noticed a slight but distinctive difference in migration between the ATP␥S-complex and the complex formed in the presence of ATP␥S and ATP (Fig. 5b). Thus, it appears that in the presence of ATP and ATP␥S, PspF 1-275 adopts a functional state that is different from the ATP␥S-bound form, the action of which on 54 leads to the formation of a novel 54 -promoter DNA complex (hereafter called the ATP␥S/ATP-complex). In all the experiments involving the use of the ATP␥S/ATP mixture, we combined 2 mM ATP with 0.2 mM ATP␥S to give the maximum amount of ATP␥S/ATP-complex at low concentration of ATP␥S (Fig. 5c).

ATP␥S-and ATP␥S/ATP-Complex Formation Requires the Same DNA and Protein Determinants
Assays with 32 P-labeled PspF 1-292 confirmed that the ATP␥S/ATP-complex is a binary 54 complex (data not shown). Unsuccessful attempts to form the ATP␥S/ATP-complex with PspF 1-275 mutants in the GAFGTA motif, in the Walker A motif or Walker B motif suggest that, like the ATP␥S-complex, formation of the ATP␥S/ATP-complex requires a form of PspF capable of hydrolyzing ATP and interacting with 54 (data not shown). Interestingly, the mixed nucleotide bound state of PspF 1-275 (hereafter called PspF 1-275 -ATP␥S/ATP) was still able to hydrolyze ATP as directly observed in ATPase assays (data not shown). AMP did not substitute for ATP in the ATP␥S ϩ ATP reactions (Fig. 5a). Furthermore, ADP was almost as efficient as ATP in working in combination with ATP␥S to increase the number of 54 -probe 2 complexes (data not shown). These results imply that the hydrolysis of ATP could be required for the ATP␥S/ATP-complex formation. Overall, these results suggest that (i) the in situ formation of ADP ϩ P i and (ii) the presence of ATP␥S in the nucleotide binding pockets of PspF 1-275 are important for the formation of the ATP␥S/ATP-complex.

Action of PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP on 54 and 54 -RNAP Leads to Altered Interactions with the Fork Junction at Ϫ12
We first attempted to characterize the interactions between the fork junction promoter structure and 54 or 54 -RNAP in the PspF 1-275 -ATP␥S and PspF 1-275 -ATP/ATP␥S complexes with probe 2 by OP-Cu and potassium permanganate (KMnO 4 ) footprinting techniques (Fig. 6, a and b). KMnO 4 is a singlestranded thymine-reactive DNA oxidizing agent that is widely used to detect local DNA melting. Treatment of naked probe 2 by OP-Cu or KMnO 4 revealed a hypersensitive region around position Ϫ12 (Fig. 6, a, lane 2, and b, lane 5). This is because of the heteroduplex segment at Ϫ12/Ϫ11 (for OP-Cu-mediated DNA cleavage) and to the unpaired thymine at position Ϫ11 (for KMnO 4 -mediated DNA cleavage) ( Table Ia). 54 Interactions-The hypersensitive site at position Ϫ12 is evident in reactions containing only 54 and probe 2 ( Fig. 6, a,  lane 3, and b, lane 1). This is consistent with 54 being weakly bound to probe 2 in the absence of any nucleotide-bound forms of PspF (see above). Strikingly, in the presence of PspF 1-275 -ATP␥S or PspF 1-275 -ATP/ATP␥S the hypersensitivity site at Ϫ12 almost completely disappeared in the OP-Cu reactions, demonstrating altered interactions with the Ϫ12 promoter region (Fig. 6a, lanes 4 and 5). In contrast, the hypersensitive site at position Ϫ12 was still evident within the ATP␥S and ATP␥S/ ATP complexes when probed with KMnO 4 (Fig. 6b, lanes 2 and  3). Yet, the intensity of the cleavage at the Ϫ11 thymine is reduced in the ATP␥S-complex when compared with the ATP␥S/ATP-complex (Fig. 6b, compare lanes 2 and 3). This difference in cleavage intensity was also observed on the probe 2 version of the 54 -dependent E. coli glnHp2 promoter (data not shown), and again indicates changed interactions with the Ϫ12 promoter region. 54 -RNAP Interactions-As before (Fig. 2), the binding of 54 -RNAP to probe 2 did not affect the inherent hypersensitive site at the Ϫ12 position to cleavage by OP-Cu (Fig. 6a, lane 6). However, in the presence of PspF 1-275 -ATP␥S or PspF 1-275 -ATP␥S/ATP, 54 -RNAP interactions with probe 2, particularly around the Ϫ12 position, are changed as demonstrated by the reduction of OP-Cu-mediated cleavage (Fig. 6a, compare lanes  3 and 6 with 4 and 5 and 7 and 8, respectively). The progressive increase in the cleavage intensity at Ϫ12 in the presence of PspF 1-275 -ATP␥S, PspF 1-275 -ATP␥S/ATP, and PspF 1-275 -ATP, respectively (Fig. 6a, compare lanes 7, 8, and 9), supports the view that 54 -RNAP interactions with the Ϫ12 promoter region are altered in response to interactions with certain nucleotidebound forms of PspF.
To determine whether the DNA has locally melted in response to PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP, we probed complexes formed between the 54 -RNAP and probe 2 in the presence of PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP, respectively, by KMnO 4 . No local DNA melting toward the transcription start site was detected (data not shown). However, consistent with the OP-Cu footprinting data, binding of 54 -RNAP to probe 2 in the absence of nucleotide-bound forms of PspF did not protect the thymine at Ϫ11 from oxidation by KMnO 4 (data not shown). Furthermore, differences in the cleavage pattern at the Ϫ11 thymine were observed in the complexes formed between 54 -RNAP and probe 2 in response to PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP on the probe 2 version of the S. meliloti nifH and E. coli glnHp2 promoters (data not shown). These differences were very similar to those seen with 54 (see above and Fig. 6b). Overall, the OP-Cu and KMnO 4 footprinting data strongly suggest that different nucleotide-bound forms of PspF have different effects on the Ϫ12 promoter region interactions made by the 54 -RNAP, but that local DNA melting does not occur.

Action of PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP on 54 and 54 -RNAP Leads to Altered Interactions with Promoter DNA
Next, we attempted to differentiate the interactions created between the promoter DNA and 54 or 54 -RNAP in the PspF 1-275 -ATP␥S and PspF 1-275 -ATP/ATP␥S complexes with probe 2 by DNase I footprinting (Fig. 6c). Binding of 54 or 54 -RNAP to the S. meliloti nifH homoduplex probe or probe 1 (Table Ia)  response to ATP hydrolysis by the activator (11,14,19). 54 -RNAP Interactions-Complexes were formed between the 54 -RNAP and probe 2 in the presence and absence of various nucleotide-bound forms of PspF. As expected, in the absence of PspF the DNA is protected between positions Ϫ34 and Ϫ1 by the 54 -RNAP (Fig. 6c, lane 3). The presence of PspF 1-275 -ATP␥S leads to an extended footprint in the downstream direction (Fig. 6c, lane 4); indicating isomerization of the 54 -RNAP-probe 2 complex in response to PspF 1-275 -ATP␥S. Interestingly, such an extension is not evident in the presence of PspF 1-275 -ATP␥S/ATP (Fig. 6c, compare lanes 4  and 5). In striking contrast to the extended footprint seen on the homoduplex probe, the 54 -RNAP footprint on probe 2 appears to shorten in the upstream direction in response to the action of PspF 1-275 -ATP (Fig. 6c, lane 6). It seems that ATP␥S, ATP␥S/ATP, and ATP each cause changes in 54 -RNAP-promoter interactions. 54 Interactions-Consistent with the gel-mobility shift assays (Fig. 1), no strong interaction between 54 and probe 2 was detected in the absence of PspF  . However, at higher concentrations of probe 2, a weak 54 -probe 2 complex could be detected. Analysis of this complex by DNase I footprinting revealed that the 54 footprint was short (between Ϫ34 and Ϫ10; data not shown), in contrast to that seen with probe 1 (between Ϫ34 and Ϫ5; Refs. 11, 14, and 19). In the presence of PspF 1-275 -ATP␥S, the 54 footprint on probe 2 is extended in the downstream direction (beyond position Ϫ1) as seen with the 54 -RNAP (Fig. 6c, compare lanes 7 and 4). Interestingly, in the presence of PspF 1-275 -ATP␥S/ATP, 54 footprints the DNA between Ϫ34 and Ϫ5, as seen in complexes formed between 54 and probe 1 or the homoduplex probe (Fig. 6c, lane 8 and data not shown).
Overall, these footprinting results suggest that different sets of activator-dependent DNA interactions (within and outside the Ϫ12 promoter region) are made by 54 and 54 -RNAP depending on the nucleotide bound states of the activator and involves promoter sequences downstream of Ϫ12 to about Ϫ1 (the promoter region where DNA melting is seen in normal open complexes).

Role of 54 Region 1 in PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP-mediated Complex Formation
Previous data have indicated that region 1 of 54 and the promoter sequences at Ϫ12 are intimately involved in the activator responsiveness of the 54 -RNAP-closed complex (7,8,11). The interaction 54 region 1 makes with the Ϫ12 promoter sequence generates a nucleoprotein target for the activator (12,13). We investigated the role of 54 region 1 in PspF 1-275 -ATP␥S and PspF 1-275 -ATP␥S/ATP mediated binding of 54 to probe 2.
⌬1 54 -The binding of wild-type 54 and a mutant form of 54 lacking its amino-terminal residues 1-56 (⌬1 54 ) to probe 2 were compared. In contrast to the wild-type 54 , ⌬1 54 bound the promoter probe 2 well (Fig. 7a, compare lane 2 and 6), suggesting that region 1 inhibits the initial binding activity of 54 to promoter probe 2. To examine whether formation of the ATP␥S-and/or the ATP␥S/ATP-complex requires 54 region 1, we added PspF 1-275 -ATP␥S or PspF 1-275 -ATP␥S/ATP to the ⌬1 54 -promoter probe 2 complex. As shown in Fig. 7a, no ATP␥S-complex or ATP␥S/ATP-complex was detected (com- pare lanes 3 and 7 and 4 and 8, respectively). Similarly, the addition of PspF 1-275 -ADP⅐AlF x to the ⌬1 54 -probe 2 complex did not result in trapped complex formation (data not shown). Therefore, we conclude that the increased binding of 54 to probe 2 is likely to be because of PspF 1-275 -ATP␥S, PspF 1-275 -ATP␥S/ATP, or PspF 1-275 -ADP⅐AlF x -mediated conformational change of 54 region 1. Overall, it appears that in the absence of activation or in the presence of sequences within the fork junction that prevent tight binding (as in probe 2), region 1 of 54 acts negatively so as to limit DNA binding. DNase I Footprints of the ⌬1 54 -Probe 2 Complex-Within the ATP␥S-complex, the 54 -specific footprint is extended in the downstream direction (beyond position Ϫ1), contrasting conventional 54 footprints on the homoduplex probe or probe 1 where protection occurs between positions Ϫ34 and Ϫ5 (Fig. 6c,  lane 7, and data not shown). To determine whether this extension is dependent on region 1 of 54 , we conducted DNase I footprinting on the ⌬1 54 -probe 2 complex. The DNase I foot-print of the ⌬1 54 -probe 2 complex was similar to the wildtype 54 footprint (protection between positions Ϫ34 and Ϫ8). However, this footprint was unchanged in the presence of PspF 1-275 -ATP␥S contrasting that seen with full-length 54 (data not shown). This indicates that 54 region 1 contributes to the extended protection of DNA in the ATP␥S-complex.
Region 1 Ala Mutants-By using a series of triple alanine substitution mutants (Ala mutants) in 54 region 1 (10), we attempted to identify region 1 residues that (i) prevented binding of 54 to probe 2 and (ii) allowed ATP␥S-and ATP␥S/ATPcomplex formation. Initially, we tested the binding activity of the Ala mutants to probe 2. In contrast to ⌬1 54 , none of the alanine mutations resulted in significantly increased binding of 54 to probe 2 (Fig. 7b). This suggests that several sequences in region 1 prevent initial binding of 54 to probe 2. Based on the ability of the Ala mutants to form the ATP␥S-complex, two classes of mutants can be distinguished: Class 1 includes the Ala mutants 6 -8, 9 -11, 12-14, 15-17, 21-23, 24 -26, 30 -32,  33-35, 36 -38, 42-44, 45-47, and 48 -50 that did not show increased probe 2-binding activity in response to PspF 1-275 -ATP␥S (Fig. 7b). Class 2 essentially includes the wild-type and mutants 27-29 and 39 -41 that show an increased probe 2-binding activity in response to PspF 1-275 -ATP␥S (Fig. 7b). Thus, the data suggest that only a small group of residues in region 1 of 54 are unimportant for productive interaction between 54 and PspF 1-275 -ATP␥S.

DISCUSSION
Many cellular processes involving DNA manipulation like replication, transposition, transcription, and restriction may all be regulated through the control of interactions made with fork junction structures. A conserved location for a DNA fork junction, where DNA opening originates, is evident within transcription complexes formed with the enhancer-dependent 54 -RNAP and enhancer-independent 70 -RNAP (8). We have shown that the ADP⅐AlF x -bound form of the AAA domain of 54 -dependent activator PspF is able to change fork junction DNA interactions made by 54 and its RNAP. This property of altering DNA-binding activity is also shared by PspF bound by ATP␥S, and by mixed nucleotide bound states (ATP␥S and ATP). We propose that an early nucleotide-dependent step in activator-dependent 54 open complex formation occurs prior to ADP or P i release, and is associated with changing the 54 -fork junction interactions that are needed for subsequent interactions made by the RNAP to form the productive open promoter complex.
Mechanochemical Functions of PspF-The ways in which PspF uses ATP binding and hydrolysis to promote open complex formation by 54 -RNAP are not well understood. The effects of ADP⅐AlF x , ATP␥S, and ATP␥S/ATP-bound forms of PspF have on 54 -DNA interactions begin to address this issue. Several lines of evidence have shown that the 54 -RNAP makes use of a fork junction DNA structure to limit DNA opening prior to activation (7,8,11). An extension of this view is that changed interactions at the fork junction are required to allow open complexes to form. Variations in promoter sequences suggest that a range of natural DNA fork junctions will exist, and that although we have used artificial fork junctions in our work, closely related structures will exist in many natural 54 -dependent promoter complexes. The DNA band shift results establish that the functional state of the activator required for stable binding to the 54 -RNAP was created by interaction with ADP⅐AlF x , an analogue of ATP at the transition state for hydrolysis. This stable binding of the activator to the 54 -RNAP leads to altered interactions between 54 and the fork junction structure at Ϫ12. The amino-terminal region 1 of 54 is required for creating the fork junction structure and activator binds directly to region 1. This suggests that the activator is able to couple events in ATP hydrolysis to changes in interactions between 54 and promoter DNA through re- structuring the 54 determinant, region 1, that creates and maintains the fork junction structure. This view is consistent with the chaperone-like property of AAA proteins and predicted changes in activator protomer structure during ATP binding and hydrolysis (5,23). Our data indicate functional significance for different nucleotide bound states of PspF.
New Binary 54 and 54 -RNAP Promoter Complexes-Results with ATP␥S and ATP␥S/ATP suggest that a set of interactions between the activator and 54 or 54 -RNAP can occur in response to changes in the DNA binding properties of 54 . Again, the critical property of the DNA probe used to demonstrate changes in DNA binding was the presence of a fork junction structure. In contrast to the results with ADP⅐AlF x , no stable ternary 54 -DNA complex was seen with ATP␥S or ATP␥S/ATP and activator, suggesting that more than one nucleotide-bound form of activator can transiently interact with 54 and 54 -RNAP to change its DNA binding properties. This ATP␥S-dependent transient interaction is presumably not dependent upon rapid hydrolysis of the triphosphate, but could involve sensing of the ␥-phosphate prior to hydrolysis or the formation of some intermediate in a reaction related to normal ATP hydrolysis. Because the ADP⅐AlF x -bound form of the activator also changed DNA binding properties of the 54 and 54 -RNAP at the same fork junction probe, it seems that activator bound by ATP␥S, ATP␥S/ATP, or ADP⅐AlF x have some shared functionality. It seems that ATP␥S can cause the activator to overcome an energetically unfavorable interaction between 54 and the DNA fork junction that is otherwise inhibitory for 54 -DNA binding interactions. The inhibition appears to be caused by region 1 of 54 and region 1 also seems to be needed for the formation of the ATP␥S-and ATP␥S/ATP-dependent binary complexes.
Nucleotide Hydrolysis-independent Isomerization of 54 -There are remarkable contrasts to previous results in which region 1 of 54 was required to bind to a set of heteroduplex probes carrying certain Ϫ12 fork junction structures and where ATP hydrolysis was necessary to remodel and isomerize the complex (probe 1 and variants thereof: Refs. 11 and 14, and Table Ia). In this work 54 region 1 inhibits binding to the heteroduplexes with alternative fork junctions and poorly hydrolyzed NTPs are used to remodel the 54 without stable DNA melting (probe 2; Table Ia). DNA heteroduplexes with opposing requirements for ATP versus ATP␥S for altered binding of 54 may reflect two alternate states of a natural promoter: one in which unfavorable interactions caused by region 1 of 54 need to be overcome for binding (e.g. on probe 2; Table Ia and this work), and one in which a strong initial set of interactions that rely upon region 1 need to be changed (e.g. on probe 1; Table Ia, and Refs. 11 and 14). The extent to which the proposed alternate states of the promoter contribute to activation is likely to be DNA sequence-specific, and therefore promoter-specific. Several lines of evidence indicate the ATP␥S-dependent reaction relates closely to the normal activation of 54 -dependent promoters. These are the common reliance upon (i) region 1 of 54 , (ii) the integrity of the GAFTGA sequences in PspF, (iii) the "arginine" finger in PspF, and (iv) similar concentration dependences upon 54 , PspF, and nucleotide.
Nucleotide-dependent Activation of Transcription-Because a common core RNAP is used by the enhancer-independent class of factors, it would appear that the special features of 54 -dependent transcription relate closely to the activator targeting an unusually stable 54 -RNAP-fork junction complex. Some of the 54 -RNAP-DNA interactions that activator change seem to have a modest energetic cost, as demonstrated by the action of the ADP⅐AlF x -and ATP␥S-bound forms of the activator. Others, notably the DNA opening per se seems to correlate to ATP hydrolysis and appear to have a higher energetic cost. A feature of AAA proteins is the use of certain amino acid sequences that function to "sense" the state and the presence of the ␥-phosphate of ATP, and potentially relay this information to cause conformational changes required for their biological output. The shared property of ADP⅐AlF x -and ATP␥S-bound forms of the activator in causing changes in DNA binding by 54 and its RNAP may well be related to a ␥-phosphate sensing process upon binding to these ATP. The failure of ADP, ATP, and AMP-PNP to behave like ADP⅐AlF x and ATP␥S can be explained by differences or absences of ␥-phosphate interactions among the various nucleotides tested. For ATP␥S or ADP⅐AlF x the sensing of the ␥-phosphate of the ATP is implicated as critically changing the functionality of the activator, a common theme for AAA proteins where nucleotide binding and hydrolysis control the binding interactions needed for substrate remodeling. It seems that the nucleotide-dependent changes in 54 and DNA structure lead to the delivery of the promoter DNA into the DNA binding cleft of the RNAP where stable melting occurs.
Overview-Our results can be viewed as providing snapshots of potential intermediates in the activation process. The use of DNA probes that mimic the proposed states of promoter DNA, and the ATP analogues provide the useful tools. However, proof of the mechanism will require detailed time resolved analyses of ATP-dependent conformational changes in the native protein and DNA components. The effects of ATP␥S and ATP together upon the activity of PspF is interesting and suggests new or increased functionality through different bound nucleotides and creation of a mixed nucleotide bound state. Structures of the AAA proteins HslU and p97 have shown that binding of different nucleotides leads to conformational changes in HslU and p97 (24,25). Here the content of the nucleotide binding pocket determines HslU conformation by bringing the ␣/␤ and ␣-helical domains of HslU together. In so doing, the I-domain of HslU is moved. The 54 binding site in PspF is believed to be equivalent to the I-domain of Hs1U (5,17), and so the combined effects of ATP␥S and ATP or ADP might be explained by their effects upon the 54 binding site in PspF. In relation to ATP hydrolysis, structural differences between the ATP and ADP bound states may be a key element in how PspF acts on 54 and its RNAP. Large conformational changes associated with ATP binding as opposed to hydrolysis are common in ATP-dependent molecular machines (25). It seems a range of nucleotidebound protomers of PspF will contribute to full activation, some contributing more toward particular steps than others, and some preceding hydrolysis and product release. The precise nucleotide dependence seems to be DNA-dependent, indicating that energy coupling in 54 -dependent systems critically involves key promoter sequences that communicate via the factor to the activator. The interaction of 54 with the fork junction DNA is clearly subject to regulation and given the similarity between bacterial and eukaryotic multisubunit RNAPs, a range of control proteins may act to regulate transcription through targeting the fork junction structure. Interestingly TATA-binding protein DNA binding is regulated in an ATP-dependent fashion by Mot1 (26).