Introduction

Transcription is the first step in gene expression and is a major point of regulation. The DNA-dependent RNA polymerase (RNAP) is the central enzyme of gene expression whose core machinery is conserved throughout evolution (Ebright, 2000). In bacteria, the five-subunit core RNAP (subunit composition ₂'; E) has to collaborate with a sixth subunit, the sigma () factor, to form the RNAP holoenzyme (subunit composition ₂'; E) for promoter-specific transcription initiation. The common form of bacterial RNAP contains the ⁷⁰-type factors (so-called after the prototypical housekeeping ⁷⁰ factor of Escherichia coli) and utilises promoters characterised by conserved sequences centred at positions -35 and -10 from the transcription start point at +1 (Paget and Helmann, 2003). The variant RNAP form contains the ⁵⁴ factor. E⁵⁴ binds to promoters characterised by conserved sequences at positions -24 (the GG-region) and -12 (the GC-region) from the transcription start point (Buck et al, 2000).

In contrast to E⁷⁰, when E⁵⁴ binds promoters it forms transcriptionally inactive closed promoter complexes. Within the E⁵⁴ closed promoter complex, the bases immediately downstream of the consensus -12 GC-region are transiently distorted, thereby creating a fork junction DNA structure (Buck et al, 2000). This fork junction structure is used by E⁵⁴ to set up a network of interactions that prevent DNA melting in the absence of activation. DNA melting and formation of a transcription-competent E⁵⁴ open complex requires a specialised activator of the AAA (ATPases Associated with various cellular Activities) family (Buck et al, 2000; Zhang et al, 2002). Activators of E⁵⁴ bind to enhancer-like sequences located 150 bp upstream of the transcription start point and use ATP hydrolysis to interact (via a DNA looping event) with a nucleoprotein interface at the promoter GC-region of the E⁵⁴ closed promoter complex, referred to as the E⁵⁴ regulatory centre. The E⁵⁴ regulatory centre constitutes a tripartite interface involving regulatory ⁵⁴ residues (notably the amino-terminal Region I and residues within Region III; Figure 1A), DNA sequences immediately downstream of the consensus GC-region and surfaces of the core RNAP upstream of the RNAP catalytic centre (Wigneshweraraj et al, 2001; 2002; 2003). Activators of E⁵⁴ bring about conformational changes in the E⁵⁴ regulatory centre, altering the interaction between regulatory ⁵⁴ residues and the fork junction DNA structure that results in DNA melting and open complex formation (Buck et al, 2000). In contrast, open complex formation by E⁷⁰ occurs at most promoters spontaneously without the requirement of activators or energy source (Paget and Helmann, 2003).

Region I and regulatory residues within Region III of ⁵⁴ constitute a major target for the AAA activator within the E⁵⁴ regulatory centre. We have previously demonstrated that the relationship between ⁵⁴ Region I and Region III regulatory residues and the promoter GC-region is different within closed and open promoter complexes, whereas the relationship between other ⁵⁴ residues and the promoter GG-region remains unchanged (Casaz and Buck, 1999; Wigneshweraraj et al, 2001; Burrows et al, 2003). We reasoned that the change in the relationship between regulatory ⁵⁴ parts and promoter GC-region reflects a step in the 'unlocking' of the silencing mechanism within the E⁵⁴ regulatory centre by the AAA activator (Casaz and Buck, 1999; Guo et al, 2000; Wigneshweraraj et al, 2001; Burrows et al, 2003). However, information about the AAA activator-driven reorganisation of the E⁵⁴ regulatory centre in the E⁵⁴ transcription initiation pathway that is associated with the change in the relationship between ⁵⁴ regulatory parts and the promoter GC-region is sparse. To address this issue, we have converted ⁵⁴ to a proximity-based DNA cleavage reagent through the conjugation with ((p-bromoacetamidobenzyl)-EDTA Fe) (FeBABE) (Ishihama, 2000) at four defined positions (Figure 1A). We now show that the reorganisation of the E⁵⁴ regulatory centre during open complex formation, measured as a change in the relationship between ⁵⁴ regulatory Region I, a regulatory residue in Region III and the promoter GC-region, is an early event that occurs upon interaction of the E⁵⁴ closed complex with the AAA activator at the transition point of ATP hydrolysis. The reorganisation of the E⁵⁴ regulatory centre is dependent on core RNAP subunits, but does not require and ' parts (-subunit residues 186–433 and R454 and '-subunit residues 215–220 and 1149–1190) normally associated with later stages of stable open promoter complex formation. Using photoreactive promoter DNA probes, we provide evidence for an interaction between the AAA activator and promoter DNA positions within and downstream of the E⁵⁴ regulatory centre when the AAA activator reorganises the E⁵⁴ regulatory centre at the transition point of ATP hydrolysis. Importantly, we describe for the first time how an initial bacterial RNAP–DNA complex is reorganised into a putative intermediate transcription initiation complex by an AAA ATPase where full DNA melting has not yet occurred.

Results

Interaction with AAA activator changes the relationship between regulatory ⁵⁴ Region I residue L46 and the promoter GC-region

We have recently shown using E⁵⁴ reconstituted with FeBABE-conjugated (via a sulphydryl group at position 46) ⁵⁴ (^Cys46*E⁵⁴) that the relationship between regulatory ⁵⁴ Region I residue L46 and the promoter GC-region is different within closed and open promoter complexes (Burrows et al, 2003; Supplementary Figure 1). Since the regulatory Region I of ⁵⁴ is necessary for the E⁵⁴ silencing mechanism and constitutes a major interaction target for the AAA activator (Chaney et al, 2001), we wanted to determine at which point during the activation process the change in the relationship between ⁵⁴ Region I and the promoter GC-region takes place. As FeBABE is a proximity-based cleavage reagent (12–15 Å cleavage range from its attachment site; for review, see Ishihama, 2000), we rationalised that information about the point at which the change in the relationship between Region I and the promoter GC-region occurs can be obtained by monitoring changes in the DNA cleavage pattern within intermediate promoter complexes formed by ^Cys46*E⁵⁴. Previously, we showed that intermediate E⁵⁴ promoter complexes can be studied by using the poorly hydrolysable ATP analogue ATPS or the ATP hydrolysis transition state analogue ADPAlF_x-bound AAA activator PspF (Phage shock protein F) (Cannon et al, 2003). For experimental simplicity, we used a truncated form of PspF, PspF_1–275, that lacks the enhancer DNA-binding helix–turn–helix motif and can activate E⁵⁴ from solution (Cannon et al, 2003). The assay uses E⁵⁴ bound to a heteroduplex promoter probe, which mimics the conformation of the promoter DNA in the closed complex (probe 2 in Figure 1B; Cannon et al, 2003). Exposure of the E⁵⁴–probe 2 complex to either PspF_1–275 and ATPS (PspF_1–275:ATPS) or PspF_1–275 and ADPAlF_x (PspF_1–275:ADPAlF_x) induces conformational changes at the promoter GC-region, which can be monitored with ortho-copper phenanthroline, a probe for detecting minor groove specific conformational changes (Cannon et al, 2003). Using ATPS and ADPAlF_x in combination with ^Cys46*E⁵⁴, we investigated whether PspF_1–275:ATPS, PspF_1–275ADPAlF_x or both can induce conformational changes in the ^Cys46*E⁵⁴–probe 2 complex that lead to changes in the DNA cleavage pattern by the FeBABE reagent at position L46 in regulatory ⁵⁴ Region I.

First, we confirmed by native-PAGE analysis and ortho-copper phenanthroline footprinting that ^Cys46*E⁵⁴ can (1) bind to probe 2 (Figure 2A) and (2) undergo PspF_1–275:ATPS- and PspF_1–275:ADPAlF_x-dependent conformational changes in a manner similar to the wild-type E⁵⁴ (Cannon et al, 2003 and data not shown). Next, we tested whether ^Cys46*E⁵⁴ can cleave the DNA within E⁵⁴–probe 2 complexes upon the addition of ascorbate and hydrogen peroxide, which leads to the generation of hydroxyl radicals at the FeBABE moiety. To do so, promoter complexes were purified from a native-PAGE gel and the DNA was analysed by denaturing PAGE (see Materials and methods). As shown in Figure 2B, strong cleavage of the nontemplate DNA strand between positions -10 and -6 is seen within the ^Cys46*E⁵⁴–probe 2 complex (purified from Figure 2A, lane 4) (compare lanes 1 and 2). Similarly, template strand positions -9 to -5 were cleaved within the ^Cys46*E⁵⁴–probe 2 complex in the presence of ascorbate and hydrogen peroxide (see later). Control reactions containing ^Cys46E⁵⁴ (i.e. in the absence of FeBABE reagent) or the use of E⁵⁴ reconstituted with a mock-conjugated cysteine-free ⁵⁴ variant (^Cys-E⁵⁴) (Burrows et al, 2003; Supplementary Figure 1) did not result in discernible DNA cleavage on either DNA strand (data not shown). This result extends previous studies and indicates that in the closed promoter complex, ⁵⁴ Region I position L46 is near the GC-region and the E⁵⁴ regulatory centre. To investigate the effect of PspF_1–275:ATPS and PspF_1–275:ADPAlF_x on the proximity relationship between ⁵⁴ Region I and the promoter GC-region, we exposed the ^Cys46*E⁵⁴–probe 2 complex to PspF_1–275 in the presence of ATPS (Figure 2A, lane 6) or ADPAlF_x (Figure 2A, lane 5). As shown in Figure 2B, no detectable changes in the nontemplate strand (compare lanes 2 and 4) or template strand (data not shown) DNA cleavage pattern or intensity were observed in the presence of PspF_1–275:ATPS. In contrast, no DNA cleavage was detected on either the nontemplate (Figure 2B, compare lanes 2 and 6) or the template strand (see later) within the ternary E⁵⁴–probe 2–PspF_1–275ADPAlF_x complex. Analysis of the binary E⁵⁴–probe 2 complex from Figure 2A, lane 5, revealed the expected DNA cleavage pattern (Figure 2B, lanes 7 and 8), suggesting that the absence of DNA cleavage within the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x is not due to the absence of hydroxyl radicals generated by the FeBABE under the conditions used. Since PspF_1–275:ATPS represents the state of the AAA activator when it is bound by ATP and PspF_1–275:ADPAlF_x represents a state of the activator at the point of ATP hydrolysis when the AAA activator can form a stable ternary complex with the E⁵⁴–probe 2 complex (Figure 2A, compare lanes 2 and 5 with 3 and 6; Chaney et al, 2001; Cannon et al, 2003), the results strongly suggest that a change in relationship between regulatory ⁵⁴ Region I residue L46 and the promoter GC-region, and thus a step in the reorganisation of the E⁵⁴ regulatory centre en route to open complex formation, occurs at the transition point of ATP hydrolysis. In agreement with this conclusion, the DNA cleavage patterns on both strands were unchanged in control reactions containing PspF_1–275 and ADP, AMPPNP or AMP (data not shown).

ADPAlF_x form of the AAA activator is required for reorganisation of the E⁵⁴ regulatory centre

In the next set of assays, we conducted a series of reactions to establish that reorganisation of the E⁵⁴ regulatory centre indeed occurs at the transition point of ATP hydrolysis, that is, in the presence of PspF_1–275:ADPAlF_x. As expected, we did not detect any changes in the DNA cleavage pattern (between positions -9 and -5 on the template DNA strand and positions -10 and -6 on the nontemplate DNA strand) in reactions containing (1) ADPAlF_x in the absence of PspF_1–275 (Figure 3A, compare lanes 3, 4, 13 and 14 with 7, 8, 17 and 18) and (2) PspF_1–275 in the absence of ADPAlF_x (Figure 3A, compare lanes 3, 4, 13 and 14 with 9, 10, 19 and 20). A change in the DNA cleavage pattern was only evident in reactions containing both PspF_1–275 and ADPAlF_x (Figure 3A, compare lanes 3, 4, 13 and 14 with 5, 6, 15 and 16). These data confirm that a conformational change involving ⁵⁴ Region I and the promoter GC-region requires the presence of PspF_1–275 and ADPAlF_x. Similarly, in assays using homoduplex promoter probe (probe 1; Figure 1B), a change in the relationship between ⁵⁴ Region I residue L46 and the promoter GC-region only occurred in the presence of PspF_1–275:ADPAlF_x (Figure 3B and data not shown). Overall, our new result extends previous observations (Burrows et al, 2003) and firmly establishes that activator interaction(s) induces an organisational change in the E⁵⁴ closed promoter complex that occurs at the transition point of ATP hydrolysis. By inference, we suggest that this organisational change reflects a step in the 'unlocking' of the E⁵⁴ silencing mechanism.

The organisational change induced by the ADPAlF_x form of the AAA activator is restricted to the E⁵⁴ regulatory centre

To further ascertain that the ADPAlF_x-bound form of the AAA activator induces an organisational change within the E⁵⁴ regulatory centre, we conducted assays with E⁵⁴ reconstituted with ⁵⁴ containing FeBABE at positions Q20 (^Cys20*E⁵⁴) and K388 (^Cys388*E⁵⁴). To investigate whether the organisational change is restricted to the E⁵⁴ regulatory centre, we conducted assays with E⁵⁴ reconstituted with ⁵⁴ containing FeBABE at position E463 (^Cys463*E⁵⁴). ⁵⁴ Q20 is located within the regulatory Region I; K388 was previously reported to be involved in regulation of E⁵⁴ open complex formation (Wang and Gralla, 2001) and E463 is part of the highly conserved RpoN-box motif (Figure 1A) and has been shown to be proximal to the transcription start-site distal conserved promoter element, the GG-region (Figure 1B) (Burrows et al, 2003). E⁵⁴ reconstituted with Cys20^*, Cys46^*, Cys388^* and Cys463^{* 54} mutants show wild-type levels of transcription activity from the Sinorhizobium meliloti nifH promoter (Supplementary Figure 1). Within the ^Cys20*E⁵⁴–probe 2 complex, only the DNA nontemplate strand is cleaved between positions -15 and -11 (Figure 4A, lanes 2 and 5). Within the ^Cys388*E⁵⁴–probe 2 complex, cleavage of the nontemplate strand DNA between positions -18 and -15 and the template strand DNA between positions -11 and -4 was observed (Figure 4A, lanes 8 and 9, respectively). Within the ^Cys463*E⁵⁴–probe 2 complex, cleavage of both DNA strands at positions -24 and -25 was observed (Figure 4B, lane 2 and data not shown). All three enzymes, ^Cys20*E⁵⁴, ^Cys388*E⁵⁴ and ^Cys463*E⁵⁴, formed ternary complexes with PspF_1–275:ADPAlF_x and probe 2 as judged by native-PAGE (data not shown). We tested whether the FeBABE-mediated DNA cleavage pattern by ^Cys20*E⁵⁴ and ^Cys388*E⁵⁴ at the promoter GC-region has changed within ternary complexes with PspF_1–275:ADPAlF_x when compared to the binary E⁵⁴–probe 2 complexes. As shown in Figure 4A, DNA cleavage was not detected or was greatly reduced in ternary complexes formed by ^Cys20*E⁵⁴ or ^Cys388*E⁵⁴ (compare lanes 4, 5 and 7–10). In contrast, no change in the DNA cleavage pattern between the binary ^Cys463*E⁵⁴–probe 2 and ternary ^Cys463*E⁵⁴–probe2–PspF_1–275:ADPAlF_x complexes was observed (Figure 4B, lanes 2 and 3). Similar assays using PspF_1–275:ATPS revealed no changes in the DNA cleavage pattern with either of the three enzymes (data not shown). Overall, the results obtained using E⁵⁴ containing FeBABE at three different regulatory positions (L46, Q20 and K388) firmly establish that an organisational change within the E⁵⁴ closed complex that occurs upon the interaction with the AAA activator at the transition point of ATP hydrolysis is restricted to the E⁵⁴ regulatory centre. Further, the results with^Cys463*E⁵⁴ also confirm that the presence of PspF_1–275 and/or ADPAlF_x does not per se inhibit DNA cleavage.

The organisational change induced by the ADPAlF_x form of the AAA activator at the promoter GC-region is dependent on core RNAP

In the next set of assays, we wanted to investigate whether the change in the relationship between regulatory ⁵⁴ parts and the promoter GC-region is dependent on the core RNAP. Since the wild-type ⁵⁴ alone does not bind either promoter probe 1 or 2 efficiently, we used promoter probe 3 for these assays (Cannon et al, 2003). Promoter probe 3 is identical in conformation to probe 2 and contains a heteroduplex segment at positions -12 and -11 that mimics the conformation of DNA within closed promoter complexes (Figure 1B) (Cannon et al, 2000). In contrast to probe 2, probe 3 contains wild-type template strand DNA sequence within the heteroduplex segment that allows tight binding of ⁵⁴ in the absence of core RNAP (see Figure 1B). Initially, we conducted experiments with ^Cys46*E⁵⁴ to confirm that FeBABE-mediated DNA cleavage near the promoter GC-region (between positions -10 and -6 on the nontemplate strand and positions -9 and -6 on the template strand) was seen within ^Cys46*E⁵⁴–probe 3 complex (Figure 5, lane 3 and data not shown). As expected from experiments using probe 1 (Figure 3B) and probe 2 (Figure 3A), no DNA cleavage was detected within the ^Cys46*E⁵⁴–probe 3–PspF_1–275:ADPAlF_x complex (Figure 5, lane 5), revealing that the conformational change between ⁵⁴ Region I residue L46 and the promoter GC-region of promoter probe 3 occurs in the presence of PspF_1–275:ADPAlF_x. Strikingly, this conformational change is not detected in the absence of core RNAP, since no changes in the DNA cleavage pattern are detected when ^Cys46*54–probe 3 and ^Cys46*54–probe 3–PspF_1–275:ADPAlF_x complexes are compared (Figure 5, lanes 8 and 10). Similar results were obtained with ^Cys20*54 on probe 3 in the absence of core RNAP (data not shown). Thus, the AAA activator- and ADPAlF_x dependent reorganisation of the E⁵⁴ regulatory centre at the promoter GC-region requires core RNAP, implying a role of core RNAP surfaces in controlling the organisation of the E⁵⁴ regulatory centre.

Core RNAP surfaces associated with open complex formation do not contribute to the reorganisation of the E⁵⁴ regulatory centre by the AAA activator

In order to identify core RNAP surfaces involved in reorganisation of the E⁵⁴ regulatory centre upon interaction with AAA activator at the point of ATP hydrolysis, we reconstituted ^Cys46*E⁵⁴ using core RNAP mutants harbouring deletions/substitutions in parts involved in open promoter complex formation in the context of E⁷⁰ and/or E⁵⁴. These were (i) core RNAP lacking the -subunit downstream lobe domain (^186-433E; Nechaev et al, 2000; Wigneshweraraj et al, 2002), (ii) impaired for -subunit downstream lobe functionality (^RH454E; Nechaev et al, 2000; Wigneshweraraj et al, 2002), (iii) lacking the '-subunit jaw domain (^'1149–1190E; Ederth et al, 2002; Wigneshweraraj et al, in preparation) and (iv) defective for intermediate promoter complex formation (^'215–220E; Bartlett et al, 1998) (Figure 6A). First, we confirmed by native-PAGE that mutant ^Cys46*E⁵⁴s can form the binary ^Cys46*E⁵⁴–probe 2 and ternary ^Cys46*E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complexes. The ^Cys46*E⁵⁴ containing the RH454 and '1149–1190 mutations bound probe 2 with reduced activity when compared to the wild-type and the other two mutant ^Cys46*E⁵⁴s (Figure 6B, compare lanes 1, 3 and 9 with 5 and 7). However, all mutant ^Cys46*E⁵⁴s formed relatively comparable amounts of ^Cys46*E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex (Figure 6B, lanes 2, 4, 6, 8 and 10). DNA cleavage between positions -9 and -5 on the template strand and -10 and -6 on the nontemplate strand was detected within the mutant E⁵⁴–probe 2 complexes (Figure 6C, lanes 1, 3, 5 and 8 and data not shown). Within mutant ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complexes, DNA cleavage was not detected as seen within the wild-type ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex (Figure 6C, lanes 2, 4, 6 and 7). Therefore, we suggest that the -subunit downstream lobe, '-subunit jaw domain and the '-subunit residues 215–220 are not involved in the initial organisation or AAA activator-driven reorganisation of the E⁵⁴ regulatory centre. Overall, the results suggest that core RNAP surfaces that are normally associated with later stages of open complex formation do not contribute to the reorganisation of the E⁵⁴ regulatory centre, and by inference, do not contribute to the E⁵⁴ silencing mechanism. It therefore appears that AAA activator- and ADPAlF_x-dependent reorganisation of the E⁵⁴ regulatory centre is an early event in the E⁵⁴ activation pathway prior to the open complex formation event.

Promoter GC-region is protected from cleavage by hydroxyl radicals within the E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex

We studied the binary E⁵⁴–probe 2 complex and ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex by solution hydroxyl radical footprinting in order to distinguish whether the basis for the change in the relationship, implied from the loss of DNA cleavage by tethered FeBABE forms of ⁵⁴, between regulatory ⁵⁴ parts and the promoter GC-region is due to (1) a change in proximity between ⁵⁴ and the promoter DNA and/or (2) protection of promoter GC-region by other protein part(s). As shown in Figure 7A, binding of E⁵⁴ to probe 2 detectably protects promoter DNA between positions -32 and -11 from hydroxyl radical cleavage (lanes 4 and 10). Two hyper-reactive sites centred at positions -20 (on the template strand, lane 4) and -20 and -6 (on the nontemplate strand, lane 10) are also evident. Within the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex, a strong protection of the DNA is seen between positions -11 and -6 (box II) on the template strand and between positions -13 and -4 (box III) on the nontemplate strand (Figure 7A, box II, lanes 3 and 4 and box III, lanes 10 and 11, respectively). Interestingly, the protected positions (boxes II and III) overlap with positions that are subject to localised FeBABE-mediated cleavage within the binary E⁵⁴–probe 2 complex and that are absent in the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex (see Figure 3). Therefore, we suggest that the failure to see FeBABE-mediated DNA cleavage within the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex is due to the promoter GC-region being protected by protein sequences—possibly other ⁵⁴, AAA activator and/or core RNAP parts. However, we cannot exclude the possibility that the proximity relationship between regulatory ⁵⁴ parts and the promoter GC-region has also changed within the reorganised E⁵⁴ regulatory centre to cause loss of DNA cleavage. Further, we note increased cleavage of both DNA strands downstream of position -20 within the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex (Figure 7A, box I, lanes 3 and 4 and box IV, lanes 10 and 11). This observation further supports the view that the binary E⁵⁴–probe 2 and the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complexes are conformationally different.

AAA activator can be crosslinked to the promoter GC-region within the reorganised E⁵⁴ regulatory centre

We used a proximity-based protein–DNA crosslinking approach to further study the relationship between regulatory ⁵⁴ residues and the promoter GC-region upon ADPAlF_x-dependent interaction with activator. Our aim was to investigate which protein component of the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complex (i.e. , , ', , ⁵⁴ and/or PspF_1–275) contributes to the protection of the promoter GC-region from FeBABE and solution hydroxyl radical-mediated cleavage within ternary promoter complexes. Synthetic photoactive ³²P-tagged probe 2 was made by derivatising a single, specifically placed phosphorothioate with p-azidophenacyl bromide (APB) (see Materials and methods). Upon UV irradiation, the aryl azide group of APB reacts with protein residues within 12 Å range from its attachment site on the DNA, which results in the formation of covalently crosslinked protein–DNA complexes (Mayer and Barany, 1995).

Since FeBABE-mediated cleavage by ^Cys46*E⁵⁴ occurs between positions -10 and -6 on the template DNA strand and between positions -9 and -5 on the nontemplate DNA strand; Figure 3A), we initially APB-derivatised phosphorothioates incorporated between nucleotide positions -6 and -7 on both DNA strands. First, we used the wild-type E⁵⁴ to establish that APB derivatisation did not prevent formation of the binary E⁵⁴–probe 2 and the ternary E⁵⁴–probe 2–PspF_1–275:ADPAlF_x complexes (data not shown). Next, we used ^Cys46*E⁵⁴ to confirm that the change in relationship between ⁵⁴ Region I residue L46 and the promoter GC-region in the presence of PspF_1–275:ADPAlF_x occurs in the APB-derivatised probe 2 as seen with the unmodified probe 2 (data not shown). To identify which protein component(s) of the binary E⁵⁴–probe 2 and ternary E⁵⁴–probe 2–PspF_1–275:ADPALF_x complexes is proximal to DNA position -7/-6, we formed promoter complexes in the presence and absence of PspF_1–275:ADPAlF_x on ³²P-tagged APB-derivatised probe 2 (in each case, the APB-containing DNA strand was tagged with ³²P). Following UV irradiation, the promoter complexes were purified from a native-PAGE gel (Figure 7B, left panel) and analysed by SDS–PAGE (see Material and methods). As shown in Figure 7B (right panel), SDS–PAGE analysis of the crosslinked binary E⁵⁴–probe 2 complex (from Figure 7B, lane 1) revealed a single radioactive band (band A, lane 3). SDS–PAGE analysis of the crosslinked ternary E⁵⁴–probe 2–PspF:ADPAlF_x complex (from Figure 7B, lane 2) revealed two bands: one that migrated at the identical position as band A (Figure 7B, compare lanes 3 and 4) and a second faster migrating band (band B, Figure 7B, lane 4). Bands A and B were not detected within control reactions, which were not subjected to UV irradiation or unmodified probe 2 was used (data not shown). Analysis of the SDS–PAGE gel shown in Figure 7B using antibodies against , ', ⁵⁴ and anti-His-tag antibodies (to detect PspF) identified band A as ⁵⁴ and band B as PspF_1–275 (data not shown).

Since ⁵⁴ alone, in the absence of core RNAP, does not bind probe 2 well but can bind tightly to probe 3, we repeated the reactions on ³²P-tagged APB-derivatised probe 3 in the absence of core RNAP subunits in order to further confirm that band A is indeed a crosslinked ⁵⁴–DNA complex. As shown in Figure 7C, bands A and B were evident in the presence or in the absence of core RNAP on ³²P-tagged APB-derivatised probes 2 and 3, respectively (compare lanes 1 and 2 with 3 and 4, respectively). Complementary results were obtained in reactions where APB was placed on the nontemplate strand DNA of probes 2 and 3 (data not shown).

To further ascertain that band B is a crosslinked PspF_1–275–DNA complex, we used three different molecular weight forms of PspF (PspF_1–275, PspF_HMK-1–275 and PspF_1–325) to form the ternary E⁵⁴–probe 2–PspF:ADPAlF_x complex. SDS–PAGE analysis of ternary E⁵⁴–probe 2–PspF:ADPAlF_x complexes formed using the three different forms of PspF confirmed that band B is indeed a crosslinked PspF–DNA complex (Figure 7D, lanes 2–4). Assays with the E⁵⁴-dependent AAA activator DctD_141–390 (C₄-dicarboxylic acid transport protein D) from S. meliloti (Xu et al, 2004) further confirmed that band B is a bona fide crosslinked AAA activator–DNA complex (Figure 7E, lane 2). Additional assays using probe 2 with APB at positions -15/-14, -14/-13, -13/-12, -12/-11, -11/-10 and -1/+1 gave similar results in that ⁵⁴ and PspF_1–275 became crosslinked in an ADPAlF_x-dependent manner (data not shown). Further, additional control reactions in which only PspF_1–275 and ADPAlF_x were present did not result in the formation of a crosslinked PspF_1–275–DNA complex (data not shown).

Overall, the crosslinking data complement the solution hydroxyl radical footprinting results and suggest that both ⁵⁴ and the AAA activator contribute to the protection of the promoter GC-region from FeBABE and solution hydroxyl radical-mediated cleavage within ternary promoter complexes with the AAA activator. Interestingly, we failed to detect crosslinks between core RNAP subunits (notably and ' subunits) and the DNA within the binary E⁵⁴–DNA or ternary E⁵⁴–DNA–activator:ADPAlF_x complexes (see Discussion). Importantly, however, the crosslinking data report that the AAA activator is proximal to the promoter GC-region at an intermediate step in the E⁵⁴ transcription initiation pathway.

Discussion

DNA melting and transcription-competent open complex formation by ⁵⁴ RNAP only occur upon an ATP hydrolysis-dependent interaction with an AAA activator. Tight regulation of transcription initiation by E⁵⁴ requires a complex network of protein–DNA interactions between regulatory residues of ⁵⁴ and a fork junction DNA structure at the promoter GC-region within initial (closed) E⁵⁴ promoter complexes referred to as the E⁵⁴ regulatory centre. The regulatory centre is reorganised by the AAA activator in an ATP hydrolysis-dependent manner during open complex formation. This study was prompted by our previous observation that the relationship between regulatory ⁵⁴ residues and the promoter GC region is changed upon the transition from closed promoter complex to open promoter complex (Wigneshweraraj et al, 2001; Burrows et al, 2003). We have used an ATP analogue, ADPAlF_x, to capture an intermediate E⁵⁴ promoter complex and establish that at least one step in reorganisation of the E⁵⁴ regulatory centre occurs upon interaction with the activator at the transition point of ATP hydrolysis. Guo et al (2000) suggested that during open complex formation by E⁵⁴, the AAA activator causes tight interactions between regulatory ⁵⁴ residues and the template strand of the fork junction DNA structure at the promoter GC-region to be broken and new interactions with the nontemplate DNA strand to be established. This so-called 'strand switching' event is an integral part of the DNA melting process by the bacterial RNAP and is expected to be an early event in the bacterial transcription initiation pathway (Guo et al, 2000; Fenton and Gralla, 2003). We suggest that reorganisation of the E⁵⁴ regulatory centre by the ADPAlF_x form of the AAA activator reflects at least one aspect of the 'strand switching' event and by inference is an early event in the E⁵⁴ transcription initiation pathway.

We have demonstrated a role for core RNAP surfaces in reorganisation of the E⁵⁴ regulatory centre. Since and ' parts known to be associated with stable DNA melting and open complex formation do not appear to contribute to the reorganisation event, this observation further emphasises that the step at which the E⁵⁴ regulatory centre is reorganised by the AAA activator is an early event in the E⁵⁴ transcription initiation pathway. In agreement with this view, DNA melting is not detected within E⁵⁴ promoter complexes formed on probes 1–3 or a supercoiled plasmid harbouring the S. meliloti nifH promoter in the presence of the AAA activator and ADPAlF_x (Cannon et al, 2003; our unpublished observations). An interaction between the AAA domain of E⁵⁴ activators and the 'to be melted' DNA sequences downstream of the promoter GC-region could be significant for causing changes in the DNA structure important for making open promoter complexes. In support of this view, two motifs within the AAA domain of the E. coli helicase RuvB interact with, and modulate, DNA in an ATP hydrolysis-dependent manner (Iwasaki et al, 2000).

The most striking observation is that within the intermediate E⁵⁴ promoter complex, the AAA activator is proximal to DNA sequences that are melted within normal open promoter complexes. Therefore, at least one intermediate E⁵⁴ promoter complex is organised in such a way that the AAA activator is within 12 Å (recall that the maximum distance between the phosphorothioate and the photoreactive site on APB is 12 Å) of promoter DNA positions -15 to -1. In support of a putative interaction between the AAA activator and the promoter GC-region DNA, the ATP hydrolysis activity of the AAA activator is stimulated in the presence of promoter DNA and ⁵⁴ promoter complexes (Cannon et al, 2004).

The proximity-based photo-crosslinking also indicates that DNA across the core promoter GC-region (from positions -15 to -10) and promoter DNA sequences that are melted within open complexes (from positions -12 to +1) are contacted exclusively by the ⁵⁴ subunit within the closed and at least one intermediate E⁵⁴ promoter complex. Consistent with this observation, the crystal structure of the E^A promoter complex revealed that major promoter DNA contacts were exclusively made by ^A (Murakami et al, 2002). An analogous situation exists in eukaryotic RNAP II transcription, where no or very little contact of promoter DNA and RNAP II exists prior to DNA melting (Bushnell et al, 2004). Since closed promoter complex formation by RNAP occurs in a conformation referred to as the 'closed clamp state' where access to the catalytic cleft of the RNAP is restricted by the ' clamp and ' jaw domain and thus is not wide enough for interaction with double-stranded DNA, we envisage that DNA contacts with and ' subunits are only established once DNA melting occurs or has occurred. In support of this view, we can photo-crosslink and ' subunits to several positions (from -15 to +1) on the S. meliloti nifH promoter within bona fide open promoter complexes, that is, when the 'open clamped state' conformation is established (Wigneshweraraj et al, in preparation). Thus, it appears that within bacterial transcription complexes, the path of the promoter DNA is defined by interactions with the subunit and AAA activator (in the case of E⁵⁴) prior to open complex formation. Similarly, within the RNAP II transcription initiation complex, the path of the promoter DNA is defined by accessory transcription factors Tfg2 subunit of TFIIF (the homologue of the bacterial factor), TFIIE and TFIIH (Bushnell et al, 2004). Interestingly, TFIIH is an ATPase that is required to trigger DNA melting and thus is conceptually analogous to the AAA activator within intermediate E⁵⁴ transcription complexes.

The use of ⁵⁴-containing RNAP in combination with the ADPAlF_x-bound AAA activator has enabled us to capture and study one intermediate bacterial RNAP–DNA complex in which full DNA melting has not yet occurred. This intermediate complex is clearly different from the initial E⁵⁴ closed promoter complex and likely reflects the 'unlocked' conformation where full 'strand-switching' and 'clamp opening' events have not yet occurred and where E⁵⁴ has not acquired full heparin stability—the latter being a hallmark feature of bona fide open promoter complexes. Since the ADPAlF_x-bound AAA activator is locked in the ATP hydrolysis transition state conformation, we envisage that release of ADP and/or P_i by the AAA activator and DNA melting are linked events. Thus, we postulate that product(s) release (ADP and/or P_i) by the AAA activator following ATP hydrolysis is accompanied by full 'strand-switching' and 'clamp opening' events. In agreement with the latter, we recently demonstrated that ATP hydrolysis-dependent interaction of the AAA activator with the E⁵⁴ regulatory centre (notably ⁵⁴ Region I) induces conformational changes in the ' jaw domain (see above), which precedes open complex formation (Wigneshweraraj et al, in preparation). It appears that modulation of accessory factors, like the subunit, which bind to the so-called 'upstream face' of the RNAP (in this case, modulation of the ⁵⁴ subunit by the AAA activator) enables RNAP (and factor) to adopt several different conformations needed for transcription initiation.

Materials and methods

Promoter DNA probes and proteins

S. meliloti nifH promoter probes 1–3 were constructed as described by Cannon et al (2003) and Wigneshweraraj et al (2003. K. pneumoniae single-cysteine variants of ⁵⁴, Q20C, L46C, K388C and E463C (Figure 1A) were constructed, purified and modified with FeBABE as described by Burrows et al (2003). The AAA domain (residues 1–275) of PspF (PspF_1–275) and the heart-muscle-kinase tagged version thereof (PspF_HMK-1–292) were constructed and purified as detailed by Chaney et al (2001) and Wigneshweraraj et al (2003). Wild-type E. coli core RNAP was purchased from Epicentre Technologies. Full-length PspF (PspF_1–325) was purified as described by Wigneshweraraj et al (2003). The AAA domain of the S. meliloti DctD protein (residues 141–390) was purified as described by Xu et al (2004). The E. coli core RNAPs harbouring -subunit mutations 186–433 and RH454 were purified as described by Wigneshweraraj et al (2003). The E. coli core RNAPs contain