Exploitation of a Chemical Nuclease to Investigate the Location and Orientation of the Escherichia coli RNA Polymerase α Subunit C-terminal Domains at Simple Promoters That Are Activated by Cyclic AMP Receptor Protein*

The C-terminal domain of the α subunit (αCTD) of bacterial RNA polymerase plays an important role in promoter recognition. It is known that αCTD binds to the DNA minor groove at different locations at different promoters via a surface-exposed determinant, the 265 determinant. Here we describe experiments that permit us to determine the location and orientation of binding of αCTD at any promoter. In these experiments, a DNA cleavage reagent is attached to specific locations on opposite faces of the RNA polymerase α subunit. After incorporation of the tagged α subunits into holo-RNA polymerase, patterns of DNA cleavage due to the reagent are determined in open complexes. The locations of DNA cleavage due to the reagent attached at different positions allow the position and orientation of αCTD to be deduced. Here we present data from experiments with simple Escherichia coli promoters that are activated by the cyclic AMP receptor protein.

RNA synthesis in bacteria is due to holo-RNA polymerase (RNAP), 1 which is a multisubunit complex with subunit composition ␣ 2 ␤␤Ј. It is well known that the major factor ensuring "correct" gene expression in bacteria is the efficiency with which RNAP recognizes the promoters of different genes. Although the principal determinants for promoter recognition reside in the RNAP subunit, at many promoters, the ␣ subunits also play a key role (1)(2)(3). Each RNAP ␣ subunit consists of two independently folding domains connected by a flexible linker. The N-terminal domain (␣NTD; residues 8 -235 in Escherichia coli ␣) carries determinants for the interaction of the ␣ subunits with the rest of RNAP, whereas the C-terminal domain (␣CTD; residues 249 -329 in E. coli ␣) carries determinants for interaction with promoter DNA elements and with certain transcription factors (4,5). At some promoters, the interaction of ␣CTD with specific DNA sequence elements (known as UP elements) is essential for optimal transcription initiation. At other promoters, optimal expression depends on activator proteins that function by making a direct contact with ␣CTD, thereby recruiting RNAP to the promoter (3).
High resolution structures for E. coli ␣CTD, either free or bound to the UP element DNA, have been obtained (6 -8). When bound to DNA, ␣CTD contacts the minor groove (9,10). Genetic analyses, together with the structural studies, have identified residues in two helix-loop-helix motifs that are responsible for DNA binding. Thus, Arg 265 , Asn 268 , Gly 296 , Lys 298 , and Ser 299 appear to be the crucial ␣ subunit residues that make contact with the DNA minor groove (reviewed in Ref. 3 and see Ref. 8). These residues are part of a small segment of the surface of ␣CTD known as the 265 determinant. The linker joining ␣NTD and ␣CTD contains at least 13 amino acids and appears to be very flexible and unstructured (11). A consequence of this is that ␣CTD can bind at different locations at different promoters. At promoters where RNAP initiates transcription without the need for an activator protein, one ␣CTD contacts the DNA minor groove near position Ϫ41, upstream of the promoter Ϫ35 element (3,12). In some cases, a distinct segment of the surface of this ␣CTD, known as the 261 determinant (including ␣ residue Glu 261 ), has been shown to contact domain 4 of the RNAP subunit, which docks with the promoter Ϫ35 element (13,14). At activator-independent promoters, the second ␣CTD may contact UP element DNA or may be free to contact upstream sequences at different locations (3,12). At activator-dependent promoters, the location of ␣CTD is dependent on the organization of the DNA site(s) for the activator, and there are great variations in the positioning of ␣CTD at different promoters (15). For any activator-dependent promoter, a key challenge is to understand how the different determinants of bound ␣CTD are oriented with respect to the rest of RNAP and other bound proteins.
The role of ␣CTD has been most intensively studied at E. coli promoters that are dependent on the cyclic AMP receptor protein (CRP; also known as catabolite activator protein). When triggered by cyclic AMP, CRP binds to target sequences at dozens of different promoters and activates transcription by recruiting RNAP to these promoters (reviewed in Ref. 16). CRP is a homodimer that binds to 22-bp sequences at target promoters. Simple CRP-dependent promoters (i.e. promoters that are dependent on binding of a single CRP dimer) fall into two classes (reviewed in Ref. 17). At class I promoters, CRP binds upstream of the RNAP binding determinants (usually centered near position Ϫ61, Ϫ71, or Ϫ81), and the downstream subunit of the CRP dimer makes a direct interaction with ␣CTD to recruit it, and thereby the rest of RNAP, to the promoter. At class II promoters, CRP binds to a site centered near position Ϫ41 and overlaps the promoter Ϫ35 region. The upstream * This work was supported by project grants from the UK Biotechnology and Biological Sciences Research Council and the Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. subunit of the CRP dimer makes a direct interaction with ␣CTD that recruits it to the DNA segment immediately upstream of bound CRP. Although two ␣CTDs of RNAP could potentially interact with the two subunits of the CRP dimer, a single CRP-␣CTD interaction suffices for CRP-dependent transcription activation at both classes of CRP-dependent promoter (18).
The same surface of CRP interacts with ␣CTD at both class I and class II promoters (17). This has been identified as a surface-exposed loop adjacent to the helix-turn-helix DNA binding motif (activating region 1, AR1; residues 156 -164). Analysis of the high resolution crystal structure of a CRP⅐␣CTD⅐DNA ternary complex shows that AR1 contacts a segment of the surface of ␣CTD that is located on the opposite face from the 261 determinant (8). This surface, known as the 287 determinant, includes residues 285-289 that had been identified in previous genetic analyses as being the most likely residues in ␣CTD to make direct contact with CRP (19,20).
The high resolution crystal structure of a CRP⅐␣CTD⅐DNA ternary complex, together with genetic and footprint analyses, provides a framework for understanding the arrangement of CRP and RNAP subunits during open complex formation at any CRP-dependent promoter. Notwithstanding this, there is a need for direct methods to find the location and, more importantly, the orientation of the two RNAP ␣CTDs at different promoters with different organizations. A novel approach to this problem was taken by Murakami et al. (21,22) who exploited the DNA cleavage reagent, iron [S]-1-[p-bromoacetamidobenzyl]ethylenediaminetetraacetate (FeBABE). RNAP was reconstituted with purified ␣ subunits that had been covalently modified with FeBABE, open complexes were formed at different promoters, and the DNA cleaving ability of Fe-BABE was triggered. In these experiments, the FeBABE reagent was tethered to Cys 269 , immediately adjacent to the DNA binding 265 determinant, and thus the positions of DNA cleavage at the target promoter reveal the locations of ␣CTD binding. In this study, we have extended this analysis by attaching the FeBABE reagent either to position 273 or to position 302 of the RNAP ␣ subunit; these positions are located on opposite faces of ␣CTD (Fig. 1). Our aim was to chose two well separated positions that were removed from the DNA binding surface of ␣CTD, in order to investigate not only the location of the two RNAP ␣CTDs but also their orientation when bound at different CRP-dependent promoters. Thus, we present data for complexes at simple class I and class II CRP-dependent promoters, as well as at a CRP-independent derivative.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Promoter Derivatives-Bacterial strains, plasmids, and promoter derivatives used in this study are listed in Table I. EcoRI-HindIII fragments carrying promoter derivatives were cloned in vector plasmid pSR. Primers used in the mutagenesis of rpoA are listed in Table II. The CC(Ϫ61.5)p12T promoter was generated by PCR from the CC(Ϫ61.5) promoter (28) using the oligonucleotide 5Ј-TAAGATCTCCCCTCACTCCTGCTATAATTCT-3Ј that carries the C to T change at position Ϫ12 in the promoter.
Construction of Plasmids Encoding ␣-To overexpress and purify ␣ protein, plasmid pHTT7f1NH␣ was used. This plasmid carries the rpoA sequence with cysteine codons at positions 54, 131, 176, and 269. Thus, using PCR, these were changed to alanine codons, prior to the insertion of single cysteine codons at desired locations. For the PCR, mutagenic oligos were used together with flanking oligos, and the resulting PCR fragments were cloned into pHTT7f1NH␣ vector DNA, cut with the appropriate restriction enzymes. Oligos D25330 and D25335 were first used to replace codon Cys 176 , with the resulting PCR fragment being cloned into pHTT7f1NH␣ using EcoRI and HindIII sites. Next, oligos D25331 and D25332 were used to replace codon Cys 131 with the resulting PCR fragment being cloned into pHTT7f1NH␣ Ala 176 using PmlI and EcoRI sites. In the next step, megaprimer PCR was used to remove codon Cys 54 . The megaprimer fragment was generated using oligos D25333 and D25387, and then the megaprimer and oligo D25334 were used to add codon Ala 54 . The resulting PCR product containing Ala 54 was cloned into the pHTT7f1NH␣ Ala 131 Ala 176 plasmid using XbaI and PmlI sites. Finally, a derivative of pHTT7f1NH␣, encoding rpoA with no cysteine codons, was made by cloning a DNA fragment from plasmid pHTf1␣ carrying an Ala 269 codon (27) using oligos D10885 and D21316 and HindIII and BamHI sites. This derivative was then used as the template for the construction of mutants carrying single cysteine codons at positions 273 or 302 in the rpoA gene, using megaprimer PCR. Mutagenic oligos D25817 or D26543, respectively, were used together with oligo D10885 to generate the megaprimer fragments, and then the megaprimers and oligos D21316 were used to add the single cysteine codons. Resulting PCR fragments were cloned into pHTT7f1NH␣ Ala 54 -Ala 131 -Ala 176 -Ala 269 by using HindIII and BamHI sites. The sequences of all plasmids were checked using automated sequencing with the sequencing primers, D25612 and D25613 (University of Birmingham Genomics Facility).
Purification of ␣ Mutants-Strain BL21 DE3 carrying pHTT7f1NH␣ derivatives was used to overexpress His-tagged ␣ protein with single cysteines at position 273 or position 302. 500-ml cultures in Lennox broth containing ampicillin (200 g/ml) were grown at 37°C to an A 650 of ϳ0.6 -0.8, induced by the addition of 0.2 mM isopropyl-1-thio-␤-Dgalactopyranoside (and an additional 200 g/ml ampicillin), and grown for a further 3 h. Cells were then harvested and resuspended in lysis buffer (20 mM Tris/HCl, pH 8.0, 500 mM NaCl, 10% glycerol). Following sonication, the lysate was cleared by centrifugation and subjected to Ni 2ϩ -nitrilotriacetic acid-agarose affinity chromatography (Amersham Biosciences). The column was pre-equilibrated in lysis buffer, and the bound ␣ subunits were eluted with a gradient of 0 -200 mM imidazole. Fractions containing ␣ were pooled, concentrated, and exchanged into buffer for conjugation with FeBABE (10 mM MOPS, pH 8.0, 0.5 mM EDTA, 100 mM NaCl, 5% glycerol) using spin columns (Vivascience).
Conjugation of ␣ Derivatives with FeBABE-FeBABE was purchased from NBS Biologicals Ltd., dissolved in Me 2 SO, and stored at Ϫ70°C. Conjugation with FeBABE was initiated by mixing 100 M ␣ (final concentration) in 700 l of conjugation buffer with 1 mM FeBABE (final concentration). Following incubation at 37°C for 1 h, samples were quenched by the addition of 700 l of lysis buffer and concentrated using spin columns (Vivascience). The efficiency of conjugation was determined by quantifying free sulfhydryl groups on both conjugated and unconjugated protein with the fluorescent reagent, 7-diethylamino-3-(4Јmaleimidylphenyl)-4-methylcoumarin, as described by Greiner et al. (30). The ability of the conjugated ␣ subunits (compared with wild type ␣ and unconjugated single cysteine ␣ protein) to bind to DNA was confirmed using electromobility shift assays at the ␣(Ϫ63)CC(Ϫ41.5) promoter, as in Savery et al. (19). Samples were stored at Ϫ20°C after the addition of 50% (v/v) glycerol.
Reconstitution of RNAP Derivatives-Preparation of inclusion bodies containing ␤, ␤Ј, or 70 subunits from strains XL1-Blue [pLHN12␤], BL21 DE3 [pT7␤Ј], and BL21 DE3 [pMKSe2], respectively, and reconstitution of RNAP using these inclusion bodies was performed as described in Tang et al. (26). The reconstituted holoenzyme was then purified based on the methods of Tang et al. (26) and Igarashi and Ishihama (31). Briefly, the reconstituted holoenzyme was diluted 2.5fold with dilution buffer (10 mM Tris/HCl, pH 7.9, 100 M EDTA, 5% (v/v) glycerol) and subjected to DEAE-Sepharose chromatography and Ni 2ϩ -nitrilotriacetic acid-agarose affinity chromatography (Amersham Biosciences). The DEAE-Sepharose column was pre-equilibrated with DEAE buffer (10 mM Tris/HCl, pH 7.9, 100 mM NaCl, 5% (v/v) glycerol), and the bound holoenzyme was eluted with a linear gradient of 100 -700 mM NaCl in DEAE buffer. Pooled fractions were then applied to a Ni 2ϩ -nitrilotriacetic acid-agarose affinity column. The column was preequilibrated with storage buffer (50 mM Tris/HCl, pH 7.9, 200 mM NaCl, 5% (v/v) glycerol), and the holoenzyme was eluted with a linear gradient of 0 -250 mM imidazole in storage buffer. Fractions containing holoenzyme were pooled, and buffer was exchanged into storage buffer and concentrated using spin columns (Vivascience). Samples were stored at Ϫ20°C after the addition of 50% (v/v) glycerol. In Vitro Transcription Experiments-To show that reconstituted RNAP derivatives were transcriptionally active, transcription assays were carried out according to Savery et al. (19). Reactions were initiated by the addition of the RNAP derivative and were terminated after 15 min at 30°C by the addition of 25 l of stop solution (7 M urea, 1% SDS, 10 mM EDTA, 0.05% bromphenol blue, 0.05% xylene cyanol). Products were analyzed on 6% acrylamide gels containing 7 M urea. Gels were processed and then scanned using a PhosphorImager (Amersham Biosciences) and Quantity One software (Bio-Rad).
DNA Cleavage by FeBABE-DNA templates for footprinting were obtained using cesium chloride preparations of plasmids carrying the desired promoters cloned in pSR. AatII-HindIII fragments were purified and labeled at the HindIII end with either [␥-32 P]ATP and T4 polynucleotide kinase (for the template strand) or [␣-32 P]ATP and E. coli DNA polymerase Klenow fragment (for the non-template strand). RNAP holoenzyme was mixed with promoter DNA in a reaction volume of 35 l (20 mM HEPES, pH 8.0, 50 mM potassium glutamate, 5 mM MgCl 2 , 1 mM dithiothreitol, 0.5 mg/ml bovine serum albumin), and incubated for 10 min at 37°C. After 10 min, complexes were treated with heparin (50 g/ml final concentration) for 1 min at 37°C. DNA cleavage was then initiated by the addition of 3 mM sodium ascorbate and 3 mM hydrogen peroxide. Samples were incubated for at least 2 min at 37°C before cleavage was stopped by the addition of EDTA and thiourea to final concentrations of 45 and 7 mM, respectively. Modified DNA was extracted with phenol/chloroform and precipitated with ethanol and then analyzed on a 6% polyacrylamide gel containing 6 M urea. Gels were calibrated with Maxam-Gilbert G ϩ A sequence ladders. Gels were processed and then scanned using a PhosphorImager (Amersham Biosciences) and Quantity One software (Bio-Rad).

Construction of RNA Polymerase Conjugated with
Fe-BABE-Models of ␣CTD bound to DNA are shown in Fig. 1. Contact with the minor groove of the DNA target is due to residues of the 265 determinant (pink), and the 261 determinant (yellow) and 287 determinant (green) are displayed on opposite faces of ␣CTD so that they could interact with neighboring proteins bound to adjacent sites on the DNA. The aim of our work was to devise a method to allow experimental determination of the orientation of binding of RNAP ␣CTD in tran- scriptionally competent complexes at any promoter. Thus, for any case, we could investigate whether the 287 determinant was pointing upstream or downstream (and vice versa for the 261 determinant). To do this, we exploited the DNA-cleaving agent, FeBABE, which can be tethered to specific residues in the RNAP ␣ subunit by engineering derivatives carrying single cysteine residues.
Starting with plasmid pHTT7f1NH␣, a T7 vector for overexpression of RNAP ␣ carrying a hexahistidine tag, a derivative was first engineered that encoded cysteine-free ␣, in which all the naturally occurring cysteine residues were replaced with alanine residues. By using this construct, further derivatives, pHTT7f1NH␣ 273C and pHTT7f1NH␣ 302C, encoding ␣ with single cysteines, either at position 273 or position 302, were then constructed by replacing the glutamic acid residues at positions 273 or 302 with cysteines. These two positions in ␣CTD were chosen because they are located on opposite surfaces of ␣CTD (Fig. 1) and because alanine substitution at these locations is known to have little or no detrimental effect (19,20,32). Plasmids pHTT7f1NH␣ 273C and pHTT7f1NH␣ 302C were used to overexpress the single cysteine ␣ derivatives, and the overexpressed proteins were purified and conjugated with FeBABE reagent. Titrations with the fluorescent reagent, 7-diethylamino-3-(4Јmaleimidylphenyl)-4-methylcoumarin, showed that the conjugation efficiency was 50 -70% with Cys 273 ␣ and 70 -85% with Cys 302 ␣.
In preparatory experiments, we used electromobility shift assays to check the binding of conjugated and non-conjugated ␣ protein to DNA fragments containing an UP element and to DNA fragments to which CRP was already bound (details of the assays are shown in Ref. 19). The assays showed that the interactions of ␣ with DNA and with CRP are unaffected by conjugation with FeBABE (data not shown). The FeBABElabeled ␣ subunits were then combined with the RNAP ␤ and ␤Ј subunits to make core RNAP, before being mixed with to form holoenzyme. The ability of the preparations of RNAP, reconstituted with FeBABE-labeled Cys 273 or Cys 302 ␣ subunits, to initiate transcription was confirmed by multiround in vitro transcription assays at the lacUV5 promoter. Results illustrated in Fig. 2 show that the preparations of RNAP labeled with FeBABE are transcriptionally active, although we cannot be certain that they are fully active compared with the starting RNAP. This uncertainty is principally due to variations in reconstitution efficiencies and in the specific activity of the reconstituted RNAP preparations. However, the clarity of the FeBABE cleavage patterns, discussed below, argues that any variations in transcriptional activities between different preparations are unlikely to affect substantially our conclusions.
Binding of ␣CTD at a Class II Promoter Activated by CRP-The different promoters that we have investigated are illustrated in Fig. 3. The first promoter that we studied is a simple class II CRP-dependent promoter, CC(Ϫ41.5), that carries a consensus DNA site for CRP centered at position Ϫ41.5. This promoter was chosen because previous footprinting studies had clearly shown that both ␣CTDs bind upstream of bound CRP (33). One ␣CTD protects the DNA minor groove near position Ϫ60, and its 287 determinant interacts with AR1 of CRP (19). The second ␣CTD protects the DNA minor groove near positions Ϫ70 and Ϫ80. By using purified CRP and purified RNAP, reconstituted with ␣ subunits carrying FeBABE at residue 273 or residue 302, open complexes were formed at the CC(Ϫ41.5) promoter using end-labeled DNA fragments. Fig. 4A shows the patterns of DNA cleavage revealed by gel and PhosphorImager analysis after the nuclease activity of FeBABE had been trig- gered by the addition of H 2 O 2 and sodium ascorbate. A single cluster of DNA cleavages near position Ϫ60 was observed with both the template and the non-template promoter strand. We interpret these signals as due to one ␣CTD binding to the minor groove just upstream of bound CRP, in agreement with previous footprinting results (33). For both DNA strands, cleavage due to FeBABE located at residue 273 occurs 4 -5 bp downstream of the sites of cleavage due to FeBABE located at residue 302. The locations of the different cleavage sites on the two strands are illustrated in Fig. 4B. The simplest interpretation of these data, illustrated in Fig. 4C, is that ␣CTD binds to the minor groove near position Ϫ60 and is oriented such that the 287 determinant points downstream and the 261 determinant points upstream. The 287 determinant would thus be well placed to interact with AR1 in the upstream subunit of the CRP dimer, in agreement with previous footprinting data (33), genetic studies (19), and the CRP⅐␣CTD⅐DNA co-crystal structure (8).
In control experiments, we checked that the observed signals were dependent on the addition of H 2 O 2 and sodium ascorbate, and that no signals were seen in the absence of CRP. Surprisingly, we found no clear strong signal from FeBABE attached to the second ␣CTD, presumed by conventional footprinting studies to bind further upstream (33). Previous studies have shown that this second ␣CTD is not required for CRP-dependent activation at CC(Ϫ41.5) (18), and it is possible that its location is not sufficiently fixed to generate a strong signal.
Binding of ␣CTD at a Class I Promoter Activated by CRP, Site for CRP at Position Ϫ61.5-Next, we turned our attention to the simple class I CRP-dependent promoter, CC(Ϫ61.5), a derivative of CC(Ϫ41.5) with the consensus DNA site for CRP moved to position Ϫ61.5 (28). Previous studies with CC(Ϫ61.5), and the related lac promoter, had shown that one ␣CTD protects the DNA minor groove near position Ϫ41 (34), and its 287 determinant interacts with AR1 of the downstream subunit of the CRP dimer (20), whereas the second ␣CTD binds somewhere upstream (12). In this study we used the p12T derivative of the CC(Ϫ61.5) promoter, in which the Ϫ10 hexamer has been changed from 5Ј-CATAAT-3Ј to 5Ј-TATAAT-3Ј (Fig. 3). We found that this change to a consensus Ϫ10 hexamer improves the occupancy of promoter open complexes in vitro. In addition, the p12T substitution allows open complex formation in the absence of CRP, and thus it is possible to compare the positioning of ␣CTD in the absence and in the presence of CRP. Fig. 5A shows the patterns of DNA cleavage by FeBABEtagged RNAP, revealed by gel and PhosphorImager analysis, in open complexes at the CC(Ϫ61.5)p12T promoter in the absence of CRP. The DNA cleavage pattern shows a series of groups of bands, separated by 10 -11 bp, suggesting that ␣CTD binds successively to the minor groove along one face of the promoter DNA. The strongest DNA cleavage is found near position Ϫ41, and we interpret this as due to the binding of one of the two ␣CTDs. For both DNA strands, cleavage due to FeBABE conjugated to residue 302 occurs 4 -5 bp downstream of the sites of cleavage due to FeBABE conjugated to residue 273. The locations of the different cleavage sites on the two strands are illustrated in Fig. 5B. The simplest interpretation of these data, illustrated in Fig. 5C, is that this ␣CTD binds to the minor groove near position Ϫ41 and is oriented such that the 261 determinant points downstream and the 287 determinant points upstream. The 261 determinant would thus be well placed to interact with domain 4 of the RNAP subunit in agreement with previous data (13,14).
The results in Fig. 5A argue that, at the CC(Ϫ61.5)p12T promoter in the absence of CRP, ␣CTD can also bind to promoter DNA near positions Ϫ52, Ϫ62, and Ϫ72. The signals at these positions are weaker than those near position Ϫ41, and we suggest that they are due to the second ␣CTD being able to visit different sites at different times. The signals due to RNAP reconstituted with ␣ subunits tagged with FeBABE at residue 273 and 302 are similar. This suggests that the orientation of this ␣CTD on the DNA is not fixed.
In the next set of experiments, we investigated the patterns of DNA cleavage by FeBABE-tagged RNAP in open complexes at the CC(Ϫ61.5)p12T promoter in the presence of CRP. In preparatory experiments, we found that CRP activated this promoter by 3-4-fold in vivo and that both CRP-dependent and CRP-independent activity were reduced to background levels if the Ϫ10 hexamer was changed from 5Ј-TATAAT-3Ј to 5Ј-TG-TAAT-3Ј (p11G). Similarly, in in vitro experiments, only background levels of DNA cleavage were observed with the CC(Ϫ61.5)p12T promoter carrying the p11G mutant Ϫ10 hexamer (data not shown). Fig. 6 shows the patterns of DNA cleavage revealed by gel and PhosphorImager analysis after the nuclease activity of FeBABE had been triggered. The patterns of cleavage observed in the presence of CRP are clearly different to those in the absence of CRP, although DNA cleavage is still seen on both strands near position Ϫ41. We ascribe this to cleavage by FeBABE attached to one ␣CTD that is Our results support these conclusions: for both DNA strands, sites of cleavage due to FeBABE conjugated to residue 302 occurs 4 -5 bp downstream of the sites of cleavage due to FeBABE conjugated to residue 273. Thus, as with ␣CTD bound near this location at the CC(Ϫ61.5)p12T promoter in the absence of CRP, the simplest interpretation of the data is that ␣CTD binds to the minor groove and is oriented such that the 261 determinant points downstream and the 287 determinant points upstream (Fig. 5C). However, our results indicate that this ␣CTD is shifted 1-2 bp upstream by the binding of CRP. This is shown in Fig. 7, where patterns of DNA cleavage with RNAP reconstituted with ␣ subunits tagged with FeBABE at position 302 were analyzed with and without CRP.
Comparison of results in Fig. 5A and Fig. 6 shows that the binding of CRP to the CC(Ϫ61.5)p12T promoter suppresses DNA cleavage by the FeBABE reagent near positions Ϫ52, Ϫ62, and Ϫ72. However, cleavage appears near position Ϫ81, which corresponds to the first free minor groove on the same face of the DNA upstream of bound CRP. We suppose that this results from binding of the second RNAP ␣ subunit at this site. The signals due to RNAP reconstituted with ␣ subunits tagged with FeBABE at residue 273 and 302 are similar, suggesting that the orientation of this ␣CTD on the DNA is not fixed.
Binding of ␣CTD at a Class I Promoter Activated by CRP, Site for CRP at Position Ϫ69.5-In the final part of our study, we wanted to investigate the location of ␣CTD at a promoter with CRP bound further upstream. In previous work, we studied several class I CRP-dependent promoters with the DNA site for CRP located upstream of position Ϫ61.5, and we noted that open complex formation in vitro was inefficient. Thus, we have exploited the CC(Ϫ69.5)␣(Ϫ44) promoter, which is a derivative of CC(Ϫ41.5) carrying an UP element upstream of the Ϫ35 region and the DNA site for CRP centered at position Ϫ69.5 (Fig. 3). We found previously that expression from this promoter is dependent on CRP and that transcription activation can be reproduced in vitro (29).  2 and 6). Lane 4 shows a Maxam-Gilbert sequence ladder that was used for the calibrations shown alongside the gel.
By using purified CRP and purified RNAP, reconstituted with ␣ subunits carrying FeBABE at positions 273 or 302, open complexes were formed at the CC(Ϫ69.5)␣(Ϫ44) promoter using end-labeled DNA fragments. Fig. 8 shows the patterns of DNA cleavage revealed by gel and PhosphorImager analysis after the nuclease activity of FeBABE had been triggered. Two clusters of DNA cleavages are observed with both the template and the non-template promoter strands. We interpret these signals as due to the two ␣CTDs binding to the minor groove just downstream, near position Ϫ52, and just upstream, near position Ϫ90, of bound CRP. For the binding near position Ϫ52, cleavage due to FeBABE located at residue 273 occurs 4 -5 bp upstream of the sites of cleavage due to FeBABE located at residue 302. The simplest interpretation, illustrated in Fig. 9C, is that this ␣CTD is oriented such that the 287 determinant points upstream and the 261 determinant points downstream. For the binding near position Ϫ90, cleavage due to FeBABE located at residue 273 occurs 4 -5 bp downstream of the sites of cleavage due to FeBABE located at residue 302. The simplest interpretation is that this ␣CTD is oriented such that the 287 determinant points downstream and the 261 determinant points upstream. DISCUSSION The location of the ␣CTD of RNAP in transcriptionally competent complexes at bacterial promoters has been studied by many different methods (3). Perhaps the most direct method is that described by Naryshkin et al. (12), who incorporated a photo-activable cross-linking reagent at different locations in the lacUV5 promoter. After formation of transcriptionally competent complexes and activation of the reagent, they determined which RNAP subunits were cross-linked by the reagent located at different positions. The alternative approach, taken by Ishihama and co-workers (21,22,35,36), exploits RNAP carrying a DNA cleavage reagent specifically located in ␣CTD. In this work, we have developed Ishihama's method so that the orientation of ␣CTD, as well as its location, can be determined experimentally. We focused on 3 simple CRP-dependent promoters, so that we could cross-check our results with data from other studies. Our conclusions are summarized in Fig. 9.
Our experiments with the class I CC(Ϫ61.5) promoter were performed with the p12T derivative that permits CRP-independent promoter activity. Thus, our study is directly comparable with that of Naryshkin et al. (12) who used the lacUV5 promoter. Recall that the lac promoter has a single DNA site for CRP centered at position Ϫ61.5 and that the UV5 mutation permits CRP-independent activity. Our conclusion that ␣CTD binds near positions Ϫ41, Ϫ52, Ϫ62, and Ϫ72 in open complexes in the absence of CRP is in perfect agreement with the data from Naryshkin et al. (12). Our results suggest that one ␣CTD is fixed near position Ϫ41, whereas the other ␣CTD can "dance" on the DNA between the other three sites. Our data show that the orientation of the promoter-proximal ␣CTD is such that the 261 determinant is directed toward the promoter Ϫ35 element. This is in perfect agreement with recent results from Chen et al. (13) and Ross et al. (14), which suggest that the 261 determinant of the promoter-proximal ␣CTD can interact with domain 4 of the RNAP subunit bound to a promoter Ϫ35 element.
Our data with the CC(Ϫ61.5)p12T promoter show that, in the presence of CRP, the promoter-proximal ␣CTD remains near position Ϫ41 and is sandwiched between bound CRP and , whereas occupation of the sites near positions Ϫ52, Ϫ62, and Ϫ72 is lost, and ␣CTD binds upstream of bound CRP. Again, this result is identical to that found by Naryshkin et al. (12). Our data show experimentally that the promoter-proximal ␣CTD remains oriented with its 261 determinant pointing downstream and its 287 determinant pointing upstream. Thus, it can make the simultaneous interactions with AR1 of CRP and domain 4 of that have been described (8,13,20). Interestingly, the promoter-distal ␣CTD, which binds upstream of bound CRP, does not appear to adopt a fixed orientation. Note that this ␣CTD is not essential for CRP-dependent activation (18). Some previous studies (37,38) have presented evidence that RNAP can make contact with the DNA upstream of bound CRP at class I promoters. Results presented here and by Naryshkin et al. (12) argue that these contacts are most likely due to the "spare" ␣CTD.
Our experiments with a second class I promoter, the CRPdependent CC(Ϫ69.5)␣(Ϫ44) promoter, yielded the unexpected result that the two ␣CTDs bind on each side of the CRP dimer, just downstream near position Ϫ52, and just upstream near position Ϫ90. Law et al. (29) had interpreted their footprint data with this promoter to suggest that both ␣CTDs bind downstream of CRP. However, their data did show upstream protection, and our present result argues that this was likely due to ␣CTD. Interestingly, our experiments suggest that both ␣CTDs adopt a defined orientation with their 287 determinants pointing toward CRP, such that they could interact with AR1. Further experiments will be needed to establish whether interactions involving both CRP subunits contribute to transcription activation, and whether this arrangement is found at other class I promoters where CRP binds further upstream than position Ϫ61.5. A potential problem with such experiments is that CRP-dependent transcription activation cannot be reproduced in vitro, and thus it will probably be essential to use promoters with improved Ϫ10, Ϫ35, or UP element sequences. It will be particularly interesting to find other cases where interactions with an upstream activator are sufficiently strong to displace ␣CTD from its location near position Ϫ41 and disrupt its interaction with the RNAP subunit.
The third promoter that we studied was the class II CRP-dependent promoter, CC(Ϫ41.5), where it was already known that both ␣CTDs are located upstream of bound CRP (33). Our results show that one ␣CTD binds near position Ϫ60, but as we obtained no clear signal from the second ␣CTD, we conclude that it is mobile and is not anchored at any particular site (at least in our conditions). The ␣CTD at position Ϫ60 is bound in a fixed orientation, such that its 287 determinant points toward CRP. This provides experimental support for the interaction between the 287 determinant of ␣CTD and AR1 of the upstream subunit of the CRP dimer proposed by Savery et al. (19).
In conclusion, although the studies described here were restricted to simple promoters that are activated by CRP, the methodology we have described could be applied to any promoter where open complexes can be formed efficiently in vitro. Thus, Boucher et al. (39) recently used our constructs to study the position of ␣CTD at a promoter that is activated by the Bordetella pertussis BvgA protein. One advantage of this approach is that it can give signals due to ␣CTD binding that makes little or no contribution to promoter activity, and thus cannot be detected by genetic means. Because it is unlikely that structural information can be obtained for every activator-RNAP interaction, we suggest that, in many systems, this methodology will provide a useful complement to genetic studies.