Few membrane proteins with a role in transcriptional regulation have been studied, and none are able to perceive their respective stimuli and activate transcription of their regulons without the aid of auxiliary proteins. The bacitracin resistance regulator, BcrR, of Enterococcus faecalis is a membrane-bound DNA binding protein and is required for bacitracin-dependent expression of the bacitracin resistance genes, bcrABD. Here, we show that BcrR interacts directly with Zn2+ bacitracin (Kd = 2–5 μM), but not metal-free bacitracin. A solution-based DNA binding assay demonstrated that the affinity of BcrR for its target DNA is much higher (Kd = 40 nM) than previously found for transmembrane regulators and is comparable to that of soluble DNA binding proteins. A construct of BcrR that lacked the transmembrane domain was unable to bind to DNA, indicating that membrane localization was important for DNA binding. Bacitracin did not cause a change in the DNaseI footprint of BcrR on the bcrA promoter, but in vitro transcription assays with BcrR proteoliposomes showed bacitracin-dependent activation of transcription. These findings demonstrate that BcrR is a bona fide one-component transmembrane signal transduction system, which perceives an extracellular stimulus (presence of bacitracin) and relays it to an intracellular transcriptional response independent of any auxiliary proteins.
Signal transduction across the cell envelope in bacteria occurs predominantly via the use of two-component regulatory systems, where a membrane-bound sensor protein detects the extracellular stimulus and relays the signal to an intracellular effector protein, usually a transcriptional regulator (Stock et al., 2000). One-component systems, which contain both a sensor and an effector domain within the same protein, are thought to be the predominant form of intracellular signal transduction (Ulrich et al., 2005). Comprehensive analysis of 145 genome sequences available in 2005 showed that the vast majority (97%) of the c. 14 300 one-component DNA binding proteins were predicted to be soluble (Ulrich et al., 2005). However, this still leaves > 400 of these systems with predicted transmembrane domains. At present, very few membrane-bound DNA binding proteins have been studied, and most of these reports have focused on members of the ToxR family. ToxR-like proteins contain an N-terminal DNA binding domain, a single transmembrane helix and a C-terminal periplasmic domain (Miller et al., 1987), and include the virulence gene regulator ToxR of Vibrio cholerae (Miller et al., 1987; Crawford et al., 2003) and the lysine and pH sensor CadC of Escherichia coli (Küper and Jung, 2005; Tetsch et al., 2008). It has been shown for both of these proteins that they require interaction with other membrane proteins for induction of at least a subset of their regulon (DiRita and Mekalanos, 1991; Crawford et al., 2003; Tetsch et al., 2008). The primary function of the transmembrane segment of ToxR-like regulators appears to be as an interaction interface with the respective auxiliary proteins (Crawford et al., 2003; Tetsch et al., 2008).
We previously identified a novel membrane-bound transcriptional regulator, BcrR of Enterococcus faecalis, which activates transcription of the bacitracin resistance operon, bcrABD, in response to Zn2+ bacitracin in vivo (Manson et al., 2004; Gauntlett et al., 2008). BcrR contains four predicted transmembrane helices and an N-terminal DNA binding domain (Manson et al., 2004). Contrary to the situation in Bacillus subtilis, where the bacitracin-responsive two-component system requires the bacitracin ABC transporter for stimulus perception (Bernard et al., 2007; Rietkötter et al., 2008), BcrR is functional in the absence of the transporter, and we proposed that BcrR is able to sense bacitracin directly (Gauntlett et al., 2008). BcrR binds to the bcrA promoter (PbcrA) DNA in vitro, but there was no apparent effect of bacitracin on its affinity for its target DNA (Gauntlett et al., 2008). Three possible explanations can be envisaged for this observation: First, BcrR may not in fact perceive its stimulus directly, but is dependent on signal transduction via a second protein. Second, it is conceivable that not free Zn2+ bacitracin, but a membrane-associated undecaprenyl pyrophosphate (UPP) bacitracin complex is the true stimulus for BcrR, as has been suggested in context of the bacitracin response of B. subtilis (Bernard et al., 2007). Third, BcrR may always be bound to its target DNA, with bacitracin affecting a further downstream step of transcriptional activation.
Previous studies of the DNA binding properties of membrane-bound regulators have been performed using electrophoretic mobility shift assays (EMSA), and a common problem of these studies has been the very high concentration (μM range) of protein required to achieve retardation of the DNA (Miller et al., 1987; Küper and Jung, 2005; Gauntlett et al., 2008). This has largely been attributed to the characteristics of proteoliposomes, namely, their low protein content, tendency to form aggregates and the potential orientation of the protein with the DNA binding domain facing towards the inside of the liposome (Küper and Jung, 2005; Gauntlett et al., 2008). However, an alternative explanation might be that a gel-based approach such as EMSA is not suited for the study of membrane proteins, because the proteoliposome : DNA complex may not be stable during electrophoresis.
The aim of the current study was to determine if membrane-bound BcrR senses bacitracin directly or indirectly, elucidate the interaction of BcrR with the bcrABD promoter, and determine if BcrR is able to activate RNA polymerase (RNAP) in a bacitracin-dependent manner. The results herein demonstrate that BcrR constitutes a bona fide one-component transmembrane signal transduction system.
Results and discussion
DNA binding properties of BcrR in the absence of its transmembrane domain
A commonly used approach in the study of DNA binding properties of transcriptional regulators is the expression of the truncated forms of the regulator, in which the regulatory domain is removed. Such constructs often bind their target DNA irrespective of any modifications (e.g. phosphorylation) or ligands required by the full-length protein (Sola-Landa et al., 2005; Wickstrum et al., 2007). Expression of the soluble DNA binding domain alone would be of particular advantage with a normally membrane-bound protein, such as BcrR. The N-terminal domain (amino acid residues 5–61) of BcrR has homology to the Xre family of helix–turn–helix DNA binding proteins and, based on topological modelling, the C-terminal domain is predicted to contain four membrane-spanning α-helices (Fig. 1A). An alignment of the Xre family of helix–turn–helix DNA binding proteins with BcrR reveals considerable conservation within the DNA binding domain of other Xre family members (Fig. 1B). The location of DNA binding and dimerization domains in the alignment was based on the crystal structures of the five homologues (see Fig. 1B). To test if a soluble form of BcrR was functional, we created a construct that was truncated at the threonine residue at position 73 and contained all four predicted α-helices of the N-terminal cytoplasmic domain (Fig. 1A). The resulting protein, carrying a C-terminal hexa-histidine tag, was expressed in E. coli C41(DE3) and purified to apparent homogeneity with high yields via a single Ni2+-affinity chromatography step (Fig. 1C). EMSA were carried out under the same conditions used previously for full-length BcrR (Gauntlett et al., 2008), but no retardation of PbcrA DNA was observed even at protein concentrations as high as 2 μM (Fig. 1D). These findings suggest that the DNA binding domain of BcrR alone is unable to bind to DNA and that the protein requires its transmembrane domain for activity. To confirm these results in vivo, the same truncated form of BcrR (T73) was cloned into the enterococcal shuttle vector pAM401 (Wirth et al., 1986) under control of the native bcrR promoter and transformed into E. faecalis JH2-2 carrying the PbcrA -lacZ reporter construct pTCVA (Gauntlett et al., 2008). Very low β-galactosidase activities were detected both in the presence (1 μg ml−1) and absence of bacitracin (2.0 ± 1.7 Miller Units and 2.3 ± 1.5 Miller Units respectively), compared with the activities obtained previously for full-length BcrR (248 ± 52 Miller Units) (Gauntlett et al., 2008). While we cannot exclude the possibility that the truncated, soluble form of BcrR is not correctly expressed in E. faecalis, these data again appear to support our findings that membrane localization of BcrR is required for its DNA binding and transcriptional regulation activities. Ottemann and Mekalanos (1995) have demonstrated that ToxR from V. cholerae requires membrane localization for full activity. However, a truncated, soluble form of ToxR still retained the ability to bind to DNA and could regulate expression of some of its target genes (Crawford et al., 2003). It should be noted that ToxR has only one membrane-spanning segment compared with the four predicted for BcrR, suggesting a more essential and specialized role for membrane localization of this protein (Fig. 1A).
Membrane-bound BcrR has a high affinity for its target DNA
The EMSA experiments with membrane proteins reconstituted into proteoliposomes are problematic and require high concentrations of protein to achieve retardation of the target DNA (Miller et al., 1987; Küper and Jung, 2005; Gauntlett et al., 2008). We therefore chose an approach where DNA binding is assayed solely in solution, based on protection of a restriction site (Gaballa and Helmann, 1998). A recognition sequence for AluI endonuclease is located immediately downstream of the promoter-distal BcrR binding site identified previously (Gauntlett et al., 2008) (Fig. 2A). Binding of BcrR to this region should therefore lead to protection of the restriction site from digestion. After removal of bound protein with SDS, the DNA fragments are analysed by electrophoresis, and the proportion of uncut (i.e. bound) DNA can be determined. As shown in Fig. 2B, digestion of a 240 bp PbcrA fragment with AluI resulted in two bands of 137 and 103 bp respectively. Pre-incubation with increasing concentrations of BcrR reconstituted into liposomes (BcrR liposomes) led to a decrease in the amount of cut DNA, with full protection achieved at 60 nM BcrR. The binding constant, Kd, was determined as approximately 40 nM BcrR, nearly two orders of magnitude lower than the concentration required to cause a band shift in EMSA (Gauntlett et al., 2008). When BcrR liposomes were prepared in the presence of 25 μg ml−1 Zn2+ bacitracin, no significant difference was observed compared with no Zn2+ bacitracin (compare top and bottom panels, Fig. 2B). Pre-incubation of the target DNA with liposomes containing no BcrR activity (i.e. inactivated by freeze–thawing) or detergent-solubilized BcrR did not cause protection from AluI digestion (data not shown). These data are consistent with the observation that only membrane-localized BcrR was able to bind DNA in EMSA experiments (Gauntlett et al., 2008). As a further control, a SacI restriction site was introduced between the −10 and −35 elements of the PbcrA fragment, and while this site was protected by B. subtilis RNAP (S. Gebhard and G.M. Cook, unpubl. results), it was not protected by BcrR liposomes (data not shown), thus demonstrating that protection of the AluI site is specifically due to BcrR binding.
BcrR interacts directly with Zn2+ bacitracin
As discussed above, it was not clear from our earlier studies on BcrR, whether it has the ability to bind Zn2+ bacitracin directly or requires an auxiliary protein or a UPP–bacitracin complex for stimulus perception. To address this question, we performed bacitracin binding studies with purified BcrR based on changes in the intrinsic fluorescence of tryptophan residues. BcrR contains four tryptophan residues, one in the N-terminal cytoplasmic domain, and one each in transmembrane helices 2–4. A conformational change due to binding of bacitracin was therefore likely to affect at least one of these residues, leading to a change in total fluorescence. As shown in Fig. 3, addition of Zn2+ bacitracin to detergent-solubilized BcrR caused quenching of tryptophan fluorescence in a concentration-dependent manner, following single-site binding kinetics (r2 = 0.997). A Hill plot of the same data (Fig. 3, inset) gave a Hill coefficient of h = 1.08 (r2 = 0.992). The binding constant of BcrR for Zn2+ bacitracin was between 2.2 and 5.3 μM in three independent experiments, with maximum quenching reached between 7% and 14%. The Kd value for Zn2+ bacitracin is approximately 10-fold higher than the concentration required for activation of PbcrA in vivo (c. 0.4 μM) (Gauntlett et al., 2008). Similar experiments carried out with Zn2+-free bacitracin showed no quenching of fluorescence (Fig. 3, open symbols), thus BcrR binds only the biologically active Zn2+ complex of the antibiotic. These data show that BcrR is indeed capable of stimulus perception independent of auxiliary proteins, and that it detects free Zn2+ bacitracin, as no UPP was present in the described experiments.
BcrR protects its proposed binding sites from DNaseI digestion
Because DNA binding by BcrR was not dependent on the presence of bacitracin, we hypothesized that BcrR is always bound to its target DNA and undergoes a conformational change in response to bacitracin, which leads to activation of bcrABD transcription. Mutagenesis of the two inverted repeats in PbcrA followed by in vivo (LacZ-reporter assays) and in vitro (gel retardation assays) characterization had led us to the proposal that these two regions constitute the BcrR binding sites (Gauntlett et al., 2008). To confirm that BcrR does indeed bind to these sites, we performed DNaseI footprinting analysis on PbcrA. The same experiments were also carried out in the presence of Zn2+ bacitracin, to see if the location or extent of the region protected by BcrR changed under these conditions. Addition of increasing amounts of BcrR proteoliposomes led to increasing protection of two regions, covering the two proposed binding sites (Fig. 4). Addition of empty liposomes (i.e. containing no BcrR) did not cause any protection from DNaseI (Fig. 4, 3rd lane), confirming that the observed effect with BcrR liposomes was specifically due to BcrR binding to the DNA. The protection pattern observed for BcrR liposomes prepared in the presence of bacitracin was the same as that in the absence of bacitracin (Fig. 4, last lane), suggesting that BcrR does not change its DNA binding sites in response to the conformational change induced by bacitracin binding.
BcrR stimulates transcription from PbcrA in vitro
We showed above that BcrR is able to bind Zn2+ bacitracin, but we did not observe any effect of bacitracin on the affinity or location of DNA binding by BcrR. We therefore postulated that binding of bacitracin to BcrR might lead to a conformational change that allows productive interaction with RNAP and thus transcription activation. To test this hypothesis, we conducted multiple-round in vitro transcription assays (Fig. 5). Because no in vitro transcription has previously been reported for a membrane-bound transcriptional regulator, we decided to use the well-established system using B. subtilis RNAP reconstituted with the primary sigma factor, SigA, of B. subtilis. E. faecalis PbcrA shares high similarity with SigA-dependent promoters of B. subtilis (Sonenshein et al., 2002), and it was therefore likely that it would be recognized by B. subtilis RNAP. As shown in Fig. 5, in the presence of RNAP alone, a small amount of RNA transcript was synthesized (1st lane), confirming that the enzyme can recognize PbcrA.
Addition of BcrR liposomes led to an increase in transcription (2nd−4th lanes), and using BcrR liposomes reconstituted in the presence of Zn2+ bacitracin led to a further induction of transcription (Fig. 5, 5th−7th lanes). The inducing effect of bacitracin over transcription activation by BcrR alone was independent of the protein concentration, thus showing that it was a true effect of bacitracin and not due to differences in BcrR activity between the two preparations of proteoliposomes. The relatively high levels of transcription in the absence of bacitracin can be attributed to the assay used involving multiple rounds of transcription and this effect has been observed previously with other regulatory proteins (Gaballa and Helmann, 2007; Lindemann et al., 2007). Densitometric analysis of the bands shown in Fig. 5A confirmed the increase in BcrR-dependent transcription in the presence of Zn2+ bacitracin (Fig. 5B). Empty liposomes containing no BcrR protein had a similar level of transcription to that of the control with RNAP alone (Fig. 5, compare first and last lanes).
To our knowledge, this is the first report of successful in vitro transcription with a membrane-bound transcription factor. Taken together, our findings lead us to propose the following model of BcrR function (Fig. 6). Based on homology to other XRE family proteins and the inverted-repeat structure of its recognition sequence, BcrR is most likely present as a dimer. BcrR is inserted into the cytoplasmic membrane, and membrane localization is required for its function. BcrR is always bound to its target DNA, independent of the presence of bacitracin and this might represent an important feature for membrane-bound DNA binding proteins. Binding occurs at two inverted repeat sequences upstream of PbcrA, and it can be assumed that each site is bound by a BcrR dimer. It is possible that these two dimers interact with each other, but we have so far been unable to detect such a tetramer of BcrR. BcrR binds Zn2+ bacitracin, and the kinetics suggest that it contains a single binding site per monomer and that binding is not cooperative in the BcrR dimer. Alternatively, a single binding site could be formed between the two monomers. The location of the binding site is not known yet. Topology predictions show only small extracellular loops of 14 and 3 amino acids respectively, which appear too small to form a binding pocket. However, for the histidine kinase ApsS of Staphylococcus epidermidis, an extracellular loop of nine amino acids has been proposed to act in the binding of antimicrobial peptides (Li et al., 2007), and it is conceivable that a similarly small region of BcrR is responsible for bacitracin binding. Bacitracin is known to interact with and affect the structure of cell membranes (Ming and Epperson, 2002), and it is therefore also possible that BcrR detects membrane-associated or even membrane-inserted bacitracin, eliminating the need for a pronounced extracellular binding site. It is unlikely that binding occurs via an intracellular domain of BcrR, because of the extremely low concentration (i.e. 0.05 μg ml−1) of Zn2+ bacitracin required to elicit a response in BcrR activity in vivo (Gauntlett et al., 2008). In light of the relatively high intrinsic resistance of the host strain (MIC = 32 μg ml−1), it can be assumed that no significant amounts of Zn2+ bacitracin would penetrate the cell envelope during the short incubation time (i.e. 1 h) used in these experiments. While BcrR is clearly able to bind Zn2+ bacitracin, the presence of the antibiotic does not affect the affinity of BcrR for DNA, nor does it affect the location of sites occupied by BcrR. However, addition of bacitracin causes stimulation of in vitro transcription from PbcrA, supporting our hypothesis that the antibiotic induces a conformational change to allow productive interaction of BcrR with RNAP. Based on the location of the promoter-proximal BcrR binding site at −64 bp relative to the transcriptional start site, it is likely that the mechanism of transcription activation involves an interaction between BcrR and the RNAP α-subunit (Hochschild and Dove, 1998), where bacitracin could induce contact formation, or alter existing contacts. It is also possible that bacitracin changes the dimerization interface between BcrR monomers or affects interaction between the two dimers of BcrR bound to PbcrA. Initial results from in vitro chemical cross-linking (S. Gebhard and G.M. Cook, unpubl. results) suggest that bacitracin does not affect the oligomerization state of BcrR, but bacitracin-induced changes may well be too subtle to be detected by such methods. The precise mechanism underlying transcription activation by BcrR requires further investigation.
The data presented here show unequivocally that BcrR is a bona fide transmembrane one-component system, recognizing extracellular bacitracin and linking it to induction of bacitracin resistance genes. It functions independently of any auxiliary proteins and is thus different from the previously characterized membrane-bound transcriptional regulators of the ToxR family. The use of a single protein for both functions provides a simple, yet efficient means of linking stimulus perception to transcription activation. Data available from genome sequencing projects suggest that such proteins are a common feature of bacterial signal transduction.
The E. faecalis BcrR carrying a C-terminal hexa-histidine tag was expressed from plasmid pTrcBcrRHis in E. coli C41(DE3) and purified by Ni2+-affinity chromatograpy as described previously (Gauntlett et al., 2008) (Table 1). Purified BcrR was reconstituted into liposomes (proteoliposomes) at a ratio of lipid to protein of 50:1, using TritonX-100 solubilized phosphatidylcholine and removal of detergent by Bio-Beads as described previously (Gauntlett et al., 2008). The proteoliposome buffer was modified to contain 10 mM Tris/HCl (pH 8), 5 mM MgCl2 and 5 mM DTT. In some instances, Zn2+ bacitracin was added at a final concentration of 25 μg ml−1 to the proteoliposome buffer. Empty liposomes were created by the same method but omitting the addition of protein. Proteoliposomes were used within 24 h of reconstitution, as snap freezing frequently resulted in complete loss of BcrR activity.
Table 1. Bacterial strains and plasmids used in this study.
pTrc99a harbouring a fragment of bcrR encoding the N-terminal first 73 amino acids
pAM401 harbouring the bcrR promoter region and coding region for the first 73 amino acids of BcrR
The RNAP holoenzyme and FLAG-tagged B. subtilis SigA were supplied by S.R. MacLellan and had been purified as described previously (MacLellan et al., 2008). Prior to use, RNAP was reconstituted with SigA by incubation with a twofold molar excess of SigA in 1× NEB4 buffer [20 mM Tris-acetate (pH 7.9), 10 mM Mg-acetate, 50 mM K-acetate, 1 mM DTT] on ice for 10 min.
Truncation of BcrR
To create an expression construct for the DNA binding domain of BcrR carrying a C-terminal hexa-histidine tag, a fragment of bcrR encoding the first 73 amino acids of the protein was PCR-amplified with primers BcrRFwd (Gauntlett et al., 2008) and BcrRT73Rev (5′-AAATTTGTCGACTCAATGATGATGATGATGATGAGTTTCGGCAAGTGTAATCAG-3′) and cloned into the NcoI and SalI sites of the expression vector pTrc99a (Amann et al., 1988). The resulting construct, pTrcBcrRDBD, was transformed into E. coli C41(DE3) (Miroux and Walker, 1996) and expression induced with 1 mM IPTG at an optical density (OD600) of 0.5. A band migrating at approximately 6 kDa (theoretical size 9 kDa) was overexpressed in the cytoplasmic fraction and purified to apparent homogeneity using a 1 ml Profinity IMAC column (Bio-Rad). Protein was loaded onto the column in buffer [20 mM potassium phosphate (pH 7.3), 250 mM NaCl, 1 mM DTT] containing 10 mM imidazole and eluted as a single peak with a 10 ml linear gradient from 10 to 500 mM imidazole. EMSAs of a PCR product of the bcrA promoter region were carried out with purified truncated BcrR (1 nM to 2 μM protein) as described previously for the full-length protein (Gauntlett et al., 2008), except that DNA was labelled with digoxigenin (Roche) and detected according to the manufacturer's instructions.
For in vivo analysis of the same construct, a fragment containing the bcrR promoter region and the coding region for the first 73 amino acids of BcrR was PCR-amplified using primers BcrRpAMf (Gauntlett et al., 2008) and BcrRT73EcoRev (5′-AAATTTGAATTCTCAAGTTTCGGCAAGTGTAATCAG-3′) and cloned into the NcoI and EcoRI sites of the E. coli/E. faecalis shuttle vector pAM401 (Wirth et al., 1986). The resulting construct, pAMBcrRDBD, was transformed into E. faecalis JH2-2 harbouring the PbcrA -lacZ reporter construct pTCVA (Gauntlett et al., 2008), and β-galactosidase activities were determined in three independent experiments as described previously (Gauntlett et al., 2008).
Restriction protection assays
Target DNA containing the bcrA promoter (PbcrA) was PCR-amplified as a 240 bp fragment using primers EfbcrAP2F (Gauntlett et al., 2008) and FootpR (5′-CACATATTCCATAATCATCAA-3′). The product was end-labelled with [γ-32P]-ATP using T4-polynucleotide kinase (Roche) according to the manufacturer's instructions. Digests (20 μl) were performed in 1× NEB2 buffer [10 mM Tris/HCl (pH 7.9), 50 mM NaCl, 10 mM MgCl2, 1 mM DTT] and contained radiolabelled DNA (c. 1000 cpm) and various concentrations of BcrR in proteoliposomes. After pre-incubation at room temperature for 20 min, 5 U AluI enzyme was added, followed by incubation at 37°C for 45 min. Reactions were stopped by the addition of 1% final concentration of SDS in DNA loading dye and heated to 95°C for 20 min. Insoluble material was pelleted by centrifugation at 16 000 g for 2 min, and supernatants were loaded onto a 6% acrylamide (19:1 acrylamide : bis-acrylamide) gel. Bands were detected using a phosphoimager screen and quantified with ImageQuant software. Per cent digested DNA was calculated from the band intensities of the cut fragments relative to those in a sample containing no BcrR (i.e. 100% cut). Binding constants (Kd) were calculated as half-maximal protection from a hyperbolic fit of per cent protected DNA over concentration of BcrR present in the reaction.
Tryptophan fluorescence assays
The intrinsic fluorescence of tryptophan residues of BcrR was measured using a Cary Eclipse fluorescence spectrophotometer. The excitation wavelength was 280 nm, with a slit width of 10 nm, and emission was scanned from 305 to 455 nm with a slit width of 10 nm. The peak of the spectrum was between 330 and 335 nm and the maximum fluorescence read-out was recorded. A minimum of three readings were taken and averaged for each sample. Reactions contained 0.4 μM freshly purified BcrR in 20 mM potassium phosphate buffer (pH 7.3), 100 mM NaCl, 5 mM DTT and 0.05% (w/v) n-dodecyl-β-D-maltoside. Protein was dialysed against reaction buffer prior to measurements. Zn2+ bacitracin was added at various concentrations and the mixtures were incubated for 5 min before measurement. To remove the Zn2+ ions from bacitracin, a stock solution of Zn2+ bacitracin was treated with a fourfold molar excess of EDTA for at least 30 min prior to addition to the reaction. No steps were taken to remove EDTA prior to tryptophan fluorescence measurements. However, EDTA bacitracin itself has a low level of intrinsic fluorescence under the assay conditions and therefore the change in fluorescence of BcrR after addition of EDTA bacitracin was corrected for this intrinsic fluorescence of EDTA bacitracin. Control experiments were carried out with addition of 60 μM EDTA or 60 μM ZnSO4 and no change in fluorescence of BcrR was observed. Total fluorescence of BcrR varied between preparations of protein and was in the range of 470–520 a.u. Data are expressed as per cent quenching relative to the total fluorescence of BcrR before addition of any ligands, corrected for dilution, and were analysed by a hyperbolic fit of per cent quenching over concentration of bacitracin present in the reaction. Binding constants (Kd) were calculated from Scatchard plots.
DNaseI footprint analysis
The 240 bp PbcrA fragment was amplified by PCR using [γ-32P]-labelled primer EfbcrAP2F (Gauntlett et al., 2008) and unlabelled primer FootpR. Reactions (50 μl) contained 1× binding buffer [40 mM Tris/HCl (pH 8), 50 mM KCl, 10 mM MgCl2, 5 mM DTT, 50 μg ml−1 BSA, 5 μg ml−1 salmon sperm DNA], labelled DNA fragment (c. 60 000 cpm) and BcrR liposomes as indicated and were incubated at room temperature for 20 min. CaCl2 was added to a final concentration of 3 mM, followed by digest with 0.06 unit DNaseI for 7 min. Reactions were stopped by adding 700 μl ice-cold stop solution (645 μl ethanol, 50 μl of 3 M sodium acetate and 5 μl of 1 mg ml−1 yeast RNA) and nucleic acids collected by centrifugation for 20 min at 16 000 g followed by two washes in 500 μl cold 70% ethanol. Pellets were dissolved in 3 μl 0.5% SDS at 55°C for 10 min, followed by addition of 7 μl formamide loading buffer and denaturation at 95°C for 5 min. Samples were then loaded onto a 6% sequencing gel. An AG ladder generated from the same labelled fragment was used to calibrate positions on the gel.
In vitro transcription assays
The DNA template for multiple-round in vitro transcription assays was a 389 bp PCR product of the bcrA promoter [amplified using primers EfbcrAP2F and EfbcrAP2R (Gauntlett et al., 2008)], which extends 200 bp past the transcriptional start site of bcrA. Reaction mixtures (20 μl) contained 100 ng PCR product in transcription buffer [40 mM Tris/HCl (pH 8), 50 mM KCl, 10 mM MgCl2, 10 mM DTT, 50 μg ml−1 acetylated BSA, 0.5 μl RNase-inhibitor]. BcrR liposomes, prepared in the presence or absence of bacitracin, were added to the reaction as indicated and incubated at room temperature for 10 min, followed by 2 min at 37°C. RNAP was added to a final concentration of 120 nM and the mixture incubated for a further 2 min at 37°C. The reaction was started by adding an NTP mixture containing 20 nmol each of ATP, CTP and GTP, 2 nmol UTP and 18 nmol [α-32P]-UTP (800 Ci mol-1). Incubation was carried out at 37°C for 10 min, and the reaction was chased by addition of 20 nmol of each unlabelled nucleotide, followed by further incubation at 37°C for 15 min. Reactions were stopped and RNA products precipitated by addition of ice-cold 550 μl stop mix (500 μl of 100% Ethanol, 50 μl of 3 M sodium acetate) and 5 μg yeast RNA. RNA was collected by centrifugation for 20 min at 16 000 g, washed once in 500 μl cold 70% ethanol, dried and dissolved in formamide loading dye. Samples were denatured at 95°C for 5 min and separated on a 6% sequencing gel.
We thank S.R. MacLellan for the kind gift of B. subtilis RNAP and SigA. We would also like to thank S.M. Wilbanks for expert advice with fluorescence assays. This work was funded by the Marsden Fund of the Royal Society of New Zealand and the International Science and Technology Linkages Fund of the New Zealand Ministry of Research, Science and Technology, and by NIH grant GM047446 (to J.D.H.).