|Home | About | Journals | Submit | Contact Us | Français|
XylS protein, a member of the AraC family of transcriptional regulators, comprises a C-terminal domain (CTD) involved in DNA binding and an N-terminal domain required for effector binding and protein dimerization. In the absence of benzoate effectors, the N-terminal domain behaves as an intramolecular repressor of the DNA binding domain. To date, the poor solubility properties of the full-length protein have restricted XylS analysis to genetic approaches in vivo. To characterize the molecular consequences of XylS binding to its operator, we used a recombinant XylS-CTD variant devoid of the N-terminal domain. The resulting protein was soluble and monomeric in solution and activated transcription from its cognate promoter in an effector-independent manner. XylS binding sites in the Pm promoter present an intrinsic curvature of 35° centered at position −42 within the proximal site. Gel retardation and DNase footprint analysis showed XylS-CTD binding to Pm occurred sequentially: first a XylS-CTD monomer binds to the proximal site overlapping the RNA polymerase binding sequence to form complex I. This first event increased Pm bending to 50° and was followed by the binding of the second monomer, which further increased the observed global curvature to 98°. This generated a concomitant shift in the bending center to a region centered at position −51 when the two sites were occupied (complex II). We propose a model in which DNA structure and binding sequences strongly influence XylS binding events previous to transcription activation.
XylS, the transcriptional regulator of the meta-cleavage pathway for alkylbenzoate catabolism in Pseudomonas putida TOL plasmid, senses the presence of 3-methylbenzoate (3MB), and activates transcription from the meta-cleavage operon Pm promoter (15), a process that requires σ32 in the exponential phase and σ38 in the stationary phase (11). XylS overproduction through a toluene-induced regulatory cascade also leads to transcription activation in a 3MB-independent manner (15, 20). XylS protein is a member of the AraC family of transcriptional regulators (17, 36, 62). The proteins of this family share significant homology over a 100-amino-acid segment that constitutes the DNA binding domain. Most family members contain an additional domain involved in effector recognition and oligomerization which modulates their transcriptional activity (6, 37, 55).
The AraC family members studied in detail so far use different mechanisms to regulate transcription. Most AraC class proteins related to stress response (e.g., MarA and SoxS) are active as monomers. These two proteins are composed of a single domain, the C-terminal DNA binding domain (22, 35), and they activate transcription from both class I and class II promoters in a process exclusively depending upon their concentrations in the cell (22, 34). In contrast, AraC family members involved in carbon metabolism (e.g., AraC, RhaS, and MelR) are usually active as dimers and appear to mainly activate class II promoters (17). Activation by this second group of AraC family proteins responds to the presence of their specific effectors. Effector recognition by the N-terminal domain (NTD) triggers a conformational change in the regulator that leads to its correct positioning at its binding sites (21, 32, 55, 56).
XylS is composed of two separate and independently functional domains. Genetic analysis located its DNA binding domain in the C-terminal end of the protein, connected by a short linker domain (amino acids 205 to 213) to the 204-amino-acid N-terminal end involved in 3MB recognition and dimerization (12, 28, 49, 50). We have shown in electrophoresis mobility shift assays (EMSA) and in vivo that the XylS N-terminal domain repressed XylS-C-terminal domain (CTD) DNA binding and activation capacity. In both cases, 3MB had a releasing effect on this repression (12). In fact, early work showed a tagged XylS C-terminal domain was able to activate transcription without the contribution of any additional determinant present in the N-terminal domain and in a 3MB-independent manner (28).
X-ray crystallography of MarA-DNA and Rob-DNA complexes showed that the conserved DNA binding domain was composed of seven α-helices folding into two helix-turn-helix (HTH) motifs. A similar arrangement is observed in the recent high-resolution structure of ToxT, the regulator of the two primary virulence factors of Vibrio cholerae (33). The structure of the two domains of this protein is most similar to the structure of the N-terminal and C-terminal domains of AraC (48). Nuclear magnetic resonance (NMR) techniques revealed an inherent flexibility in MarA that accounted for its ability to bind different marbox sequences present in E. coli chromosome. In MarA interaction with its binding sites, the recognition helix of each HTH motif was inserted into adjacent major grooves of the DNA (10, 47). This is in contrast to the structure observed in Rob interaction with the micF promoter, which occurs through only one of the HTH motifs, the second helix making contacts with the phosphate backbone (31). In XylS, a thorough genetic analysis allowed us to determine XylS binding geometry. This protein recognizes two 15-bp direct repeats (TGCA-N6-GGNTA) in the Pm promoter, each consisting of two highly conserved sequences, the 5′-box A (TGCA) and the 3′-box B (GGNTA) separated by 6 bp, spanning one DNA helical turn (19, 27) (Fig. (Fig.1).1). This organization parallels the promoter organization of the AraC-controlled PBAD promoter (40). Previous results showed that XylS binds the direct repeats in a head-to-tail organization, with α-helix 3 of the first HTH motif (residues Arg242 and Asn246) recognizing specific bases of the A boxes and α-helix 6 of the second HTH motif interacting with the B boxes (residues Arg296 and Glu299) (13).
Like most members of the AraC family, XylS is poorly soluble, hampering the possibility to obtain active protein preparations and making it unamenable to biochemical analysis (1, 26, 37, 39, 61). In the work reported here, we have integrated different approaches to explore XylS DNA binding properties. We overexpressed and purified a His-tagged XylS C-terminal domain fusion protein and obtained active protein preparations. We have found that XylS-CTD monomers bind to tandem sites and that their sequential binding provokes a gradual bending of DNA. Our analysis suggests that XylS-CTD DNA recognition not only depended on specific amino acid base contacts (13) but also was influenced by intrinsic DNA conformation features, especially by binding-induced DNA bending. In the end, this repositioning of the regulator-DNA complex would enable to establish the contacts with RNA polymerase (RNAP) required for transcription activation.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. Bacterial strains were grown at 30°C in Luria-Bertani medium supplemented, when required, with 100 μg/ml ampicillin, 25 μg/ml kanamycin, or 20 μg/ml rifampin. Escherichia coli BL21(DE3) was grown on 2×YT medium (1.6% [wt/vol] Bacto tryptone, 1% [wt/vol] yeast extract, and 0.5% [wt/vol] sodium chloride) for protein production.
The 968-bp DNA fragment covering wild-type XylS gene was PCR amplified from the TOL plasmid DNA, digested with the NdeI and XhoI enzymes, and cloned into the pET16b vector (Novagen). The resulting clones were sequenced to confirm XylS sequence. The 400-bp DNA fragment covering the XylS C-terminal domain (XylS residues 196 to 321) was PCR amplified from the TOL plasmid DNA, digested with NdeI and XhoI enzymes, and cloned into the pET16b vector (Novagen). The resulting clones were sequenced to confirm XylS-CTD sequence. To purify the truncated derivative XylS-CTD protein (XylS C-terminal domain), freshly transformed BL21(DE3) cells harboring pET-XylS-C plasmid were grown in 500 ml 2× YT medium (53) at 30°C until turbidity at 660 nm reached 0.5. Culture was then transferred to 16°C and incubated with 0.1 mM isopropyl-beta-d-thiogalactopyranoside (IPTG) for 3 to 4 h, and cells were pelleted and frozen. The cell pellet was resuspended in 50 ml of lysis buffer (20 mM sodium phosphate, pH 7.5, 300 mM NaCl, 0.1 mM EDTA, 2.5 mM 2-mercaptoethanol, 10% [vol/vol] glycerol, 10 mM imidazole, and 1 mM Complete [Roche] protease inhibitor cocktail) and disrupted in a French pressure cell. The crude extract was centrifuged at 30,000 × g for 1 h, filtered through a 0.45-μm pore-size filter and loaded onto a 5 ml Ni-agarose column (Amersham Biosciences) preequilibrated with buffer A-XylS-C (lysis buffer without the Complete cocktail). The column was washed with buffer A-XylS-C until nonspecifically bound material had been removed. A 40-ml imidazole gradient in buffer A-XylS-C (from 0 to 1 M) was then applied. XylS-CTD eluted at about 500 mM imidazole. Eluted fractions containing XylS-CTD protein were dialyzed against storage buffer (20 mM sodium phosphate, pH 7.5, 300 mM NaCl, 2 mM dithiothreitol [DTT], 10% glycerol) and stored at −70°C until use. The 643-bp DNA fragment covering the XylS N-terminal domain (XylS residues 1 to 207) was amplified by PCR from pERD103 plasmid, and the PCR product was digested with the NdeI and XhoI enzymes and subsequently cloned into the pET16b vector (Novagen). The resulting clones were sequenced to confirm XylS-NTD sequence.
XylS-CTD molecular weight was estimated by gel filtration chromatography on a Superdex 75 HR 10/30 column (Amersham Biosciences) equilibrated with a mixture of 30 mM Tris-HCl (pH 8), 200 mM KCl, 2 mM DTT, 10% (vol/vol) glycerol at room temperature. A 300-μl solution containing 60 to 100 μM protein was loaded per run. Bovine serum albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (29 kDa), trypsin inhibitor (20 kDa), and cytochrome c (12.4 kDa) were used as molecular mass standards. Proteins were detected at 280 nm. Eluted fractions of XylS-CTD were pooled and analyzed by SDS-PAGE (Fig. (Fig.22).
Complementary oligonucleotides containing one (AB) or two (ABAB) intact XylS binding sites (Fig. (Fig.1)1) were annealed and blunt-end ligated into HincII-linearized pUC19. DNA probes (130- or 150-bp DNA fragments, respectively) were obtained by PCR amplification of these constructs with M13 universal and reverse primers. The PCR products were isolated from agarose gels using the QIAquick gel extraction kit (Qiagen) and end labeled with [γ-32P]ATP using T4 polynucleotide kinase (59). Labeled probes (1 nM) were incubated with increasing amounts of purified XylS-CTD domain at 30°C for 15 min in 10 μl of binding buffer [5 mM Tris, 24 mM HEPES, pH 8, 50 mM potassium glutamate, 20 mM NaCl, 1.4 mM EDTA, 0.4 mg/ml bovine serum albumin (BSA), 9% [vol/vol] glycerol, 0.5 μg of poly(dI-dC), and 1 mM DTT]. Samples were loaded onto 8% (wt/vol) nondenaturing polyacrylamide gels in Tris-glycine buffer (0.2 M glycine, 0.025 M Tris-HCl [pH 8.6]). The gels were run for 3 h at 7 V/cm and 4°C, vacuum dried, and exposed to PhosphorImager plates. The results were analyzed using Molecular Imager FX equipment and QuantityOne software (Bio-Rad, Madrid, Spain).
The molecular weight of XylS-CTD-DNA complexes was determined electrophoretically using the procedure of Ferguson (14), as adapted by Orchard and May for protein-DNA complexes (42). Briefly, binding reaction mixtures containing 500 nM to 5 μM XylS-CTD and 50 bp of target DNA or molecular weight markers were analyzed on a series of polyacrylamide gels (7 to 10% [wt/vol] polyacrylamide at 37.5:1) electrophoresed in Tris-glycine buffer (described above) until the bromophenol blue front in flanking sample lanes reached the bottom of the gels. The gels were silver stained. The distance from the loading well to each complex was measured and standardized with the bromophenol blue migration distance to determine the relative mobility (Rf). The logarithm of the Rf was plotted against the gel concentration for each complex and protein standards, and best-fit lines were obtained. The negative slopes of these lines were then plotted against the molecular weights of the protein standards on a double-logarithmic scale, and best-line fit was obtained. The slopes for free DNA and protein-DNA complexes (the CI and CII fast- and slow-migrating complexes, respectively) were depicted, and their molecular weight was extrapolated according to the equation y = 0.1039x + 2.1505. The marker proteins used were carbonic anhydrase, α-lactalbumin, bovine serum albumin dimer, bovine serum albumin monomer, and ovalbumin.
A 158-bp DNA fragment covering XylS binding sites (positions from −113 to +35 of the Pm promoter) was specifically radiolabeled at either strand 5′ end using [γ-32P]ATP and T4 polynucleotide kinase. End-labeled DNA probes (10 nM) were preincubated for 15 min at 30°C in binding buffer with different amounts of purified XylS-CTD. DNase I was added to a final concentration of 0.5 mU/μl and incubated for 5 min at 30°C, and the reaction was stopped by addition of EDTA. Footprint patterns were analyzed on polyacrylamide sequencing gels calibrated with Maxam-Gilbert sequence ladders.
Gel retardation assays of a DNA fragment containing the two XylS binding sites were carried out as described above. The results were quantified using Molecular Imager FX equipment and QuantityOne software (Bio-Rad). The fraction of free DNA (E0), complex I (E1), or complex II (E2) is given by equations 1 to 3. To estimate KI and KII, data from gel mobility shift assays were fit to equations 1 to 3 described by Chen et al. (9) using a nonlinear curve-fitting algorithm (Dynamic Curve Fitting from Sigma Plot; Systat Software, Inc.), where P is the total protein concentration and KI and KII are the macroscopic association constants for XylS-CTD binding to one or two sites, respectively, of the A1B1A2B2 DNA fragment.
For complex CI, DNA + mP DNA-Pm, where KI = [DNA-Pm]/[DNA][P]m.
For complex CII, DNA-Pm + nP DNA-Pm + n, where KII = [DNA-Pm + n]/KI [DNA][P]m + n.
where m and n represent the number of protein monomers per site in each case, and equal to 1 in both cases, and DNA-Pm and DNA-Pm + n represent CI and CII, respectively.
The pBend2 plasmid derivative pBend2-Pm80 (18) was digested with different enzymes to yield a series of DNA fragments with permuted positions of XylS binding sites. These fragments were end labeled, and gel retardation assays were performed as described above. To estimate the apparent bend angle (α) and to locate the center of the bend, changes in the mobility of the protein-bound DNA fragments (relative to the well position) were fitted to a cosine function (E) (60), (E) cos α/2 = μM/μE, where μM represents the relative migration of the complex with the lowest mobility and μE represents the relative migration of the complex with the highest mobility.
E. coli CC118λPm::lacZ cells bearing pGP1-2 plasmid were transformed with pET-XylS-C, pET-XylS, pET-XylS-N, or the control plasmid, pET16b (Table (Table1).1). Transformants were grown overnight on LB medium containing the appropriate antibiotics. Three independent clones of each strain were used. Quadruplicate cultures were prepared by diluting cells from overnight cultures to 1/100. When cultures had reached an optical density at 600 nm (OD600) of 0.4, one of them was maintained as a noninduced control, while the other three were supplemented with 3MB (1 mM), IPTG (0.5 mM), or both, respectively. Samples for β-galactosidase activity determination were taken 1 h after induction. β-Galactosidase activity was determined in permeabilized whole cells according to Miller (38).
As for many members of the AraC family, previous attempts to purify the complete XylS protein were unsuccessful and produced protein aggregates (3, 7, 49, 61). To investigate the basis for this insolubility, we analyzed the two protein domains independently. We found that a His10 fusion protein of the C-terminal domain could be overproduced and directly purified in a soluble form. In contrast, XylS N-terminal domain with a His10 fusion (XylSN) aggregated and formed inclusion bodies under every expression condition tested (12), showing that the NTD was the main contributor to the low solubility of full-length XylS. Since the HTH DNA binding elements of XylS are located at the C-terminal end, we tested if both His-tagged XylS wild-type protein and XylS C-terminal domain were able to activate in vivo expression from the Pm promoter (Table (Table2).2). E. coli CC118 (Pm::lacZ, pGP1-2) bears a Pm::lacZ fusion in the chromosome as well as the T7 RNAP gene under the control of the phage λ PL promoter. This strain was transformed with either pET-XylS or pET-XylS-C, which carry the wild-type or truncated XylS version, respectively, under the control of T7 promoter or with the control plasmid pET16b (Table (Table1).1). The values of β-galactosidase activity determined in the presence and absence of 1 mM 3MB showed that in the absence of any inducer, activity was 1,490 ± 132 Miller units (MU) for XylS-CTD protein and 2,290 ± 234 MU in the case of full-length XylS, while in the control strain bearing pET16b the activity was negligible (Table (Table2).2). These high effector-independent levels correspond to high basal T7 polymerase-dependent XylS expression. It is well established that the high level of XylS protein obtained from a high-copy-number vector induces Pm transcription in the absence of effector (45, 49). Activity in the E. coli CC118 (Pm::lacZ, pGP1-2) strain transformed with pET-XylSN was negligible under all conditions (6 MU) and similar to values obtained with the control plasmid, pET16b. In E. coli strain CC118 (Pm::lacZ, pGP1-2), carrying the XylS-CTD fusion protein gene, no significant increase in Pm activity was detected in the presence of either 3MB or IPTG, while in the case of full-length XylS, the addition of 3MB promoted a 2-fold increase in Pm expression level (Table (Table2).2). These results confirm that both the DNA-binding domain devoid of the N-terminal domain and the full-length XylS are capable, when overexpressed, to activate transcription even in the absence of effector (19, 28). However, our results demonstrate that the 3MB response relies solely on XylS N terminal.
Most AraC family members appear to form dimers in solution, and dimerization determinants have been proposed to reside in the N-terminal domain (6, 8, 43, 44, 49). If this were the case for XylS, a truncated protein devoid of its N-terminal domain would behave as a monomer in solution. We determined the XylS-CTD oligomeric state using gel filtration through a Superdex G-75 column as described in Materials and Methods. A single peak was detected at an elution volume corresponding to a molecular mass of 16 kDa, in good correlation with the molecular mass of XylS-CTD (16.81 kDa), thus implying that the protein was a monomer under our experimental conditions (Fig. (Fig.22).
We next examined XylS-CTD binding to its operator, defined as two binding sites organized as direct repeats, one overlapping the RNA polymerase binding site (proximal site) and the second one located 6 nucleotides upstream (distal site). Each binding site is composed of two different boxes, A1/A2 and B1/B2, conserved in both sites and separated by a sequence of six nonconserved nucleotides (13, 19) (Fig. (Fig.1).1). To test if a single binding site was sufficient for XylS-CTD binding, a DNA fragment containing the distal XylS binding site composed of A1 and B1 half-sites was titrated with increasing XylS-CTD concentrations. A single shifted band was observed, corresponding to complex CI (Fig. (Fig.3a).3a). As expected (12, 28), when an A1B1A2B2 DNA fragment containing the complete two-site wild-type binding region was titrated, a second, slower-migrating shifted band was formed (complex CII) (Fig. (Fig.3b).3b). The figure shows that as the XylS-CTD concentration increased, the CI complex was progressively replaced by the lower-mobility CII complex.
To determine the DNA/protein stoichiometry of CI and CII complexes, we measured the molecular weight of the two complexes using the electrophoretic method adapted by Orchard and May (42), based on changes in gel mobility at different acrylamide concentrations. Briefly, gel mobility of CI and CII complexes and of a series of molecular weight markers was first plotted against polyacrylamide concentration, and slopes were calculated in each case (Fig. (Fig.4a).4a). The linear regression slopes obtained for each molecular weight marker were plotted against the logarithm of the molecular weight (Fig. (Fig.4b).4b). When slope values of protein-DNA complexes CI and CII were fitted to the plot, an apparent molecular mass of 46 kDa was estimated for the CI XylS-CTD-DNA complex. The molecular mass of the protein component of the CI complex was extrapolated as 17 kDa after subtracting 29 kDa corresponding to the apparent molecular mass of the 50-bp A1B1A2B2 DNA fragment, consistent with CI resulting from a single XylS-CTD monomer (16.81 kDa) bound to the promoter probe. Likewise, an apparent molecular mass of 57 kDa was obtained for CII complex (28 kDa corresponding to the protein component), showing a slight deviation from the expected mass of 33.62 kDa for two XylS-CTD monomers.
Previous work showed that Pm promoter DNA region was intrinsically curved, with an apparent bent angle of 35° ± 5° (16, 18, 19). To explore if binding of XylS-CTD induced further distortion in the DNA, we applied the circular permutation assay developed by Kim and coworkers (30, 64) (see Materials and Methods). A series of 195-bp DNA fragments with permuted positions of the XylS binding sites were incubated with XylS-CTD, and the mobility of the two complexes (CI and CII) was analyzed by native polyacrylamide gel electrophoresis (64) (Fig. (Fig.5).5). The intrinsic bend angles (α) and the location of the center of the bend were estimated according to Thompson and Landy (60). Although this method does not resolve the degree of curvature at each single position in the DNA, we could observe that the Pm intrinsic curvature centered within the A track spanning positions −41 to −46 increased to a global value of 50° ± 5° after binding of one XylS-CTD monomer and to 98° ± 2° after binding of two monomers (Fig. (Fig.5).5). The new center of the bent region was located between positions −49 and −52 in both CI and CII complexes, evidencing a shift from the previously determined free DNA bending center within the XylS binding site toward the DNA region between the two binding sites (Fig. (Fig.11 and and55).
In mobility shift assays, we observed that at a low XylS-CTD concentration, CI was preferentially formed, followed by a progressive accumulation of CII concomitant with a decrease of CI at a higher protein concentration (Fig. (Fig.3).3). To analyze if this could result from the first monomer at the proximal site facilitating binding of the second one in a cooperative manner, we performed a series of quantitative analyses of the binding interactions. To this end, five repetitions of titration assays of a DNA fragment containing one or two binding sites with XylS-CTD were performed and the amounts of CI and CII complexes of XylS-CTD bound to an A1B1A2B2 fragment were plotted as a function of XylS-CTD concentration (Fig. (Fig.6;6; see equations 1 to 3 in Materials and Methods) (9). XylS binding sites are not identical in sequence; thus, separate estimates of intrinsic binding affinity for each site could not be derived from two-site gel shift experiments. We used equations 1 to 3 as derived by Chen et al. (9) to estimate the macroscopic association constants, KI and KII. The average data were fit to equations 1 to 3, and KI and KII constants were derived using nonlinear curve fitting (see Materials and Methods). The data obtained (KI = 1.56 × 106 M−1 and KII = 7.3 × 105 M−1) are not supportive of significant cooperative interaction in the sequential binding of two XylS-CTD monomers to their recognition sites.
To analyze if XylS-CTD monomer preferentially bound any of its binding sites, we carried out DNase I footprinting assays of XylS-CTD bound to a Pm fragment. Figure Figure77 shows that XylS-CTD protected two regions centered at positions −60 and −40, corresponding to the two 15-bp direct repeats (distal and proximal sites). We observed that at a low protein concentration, the proximal site was fully protected, whereas the distal site was only progressively protected by increasing XylS-CTD concentration (see especially the bottom strand footprinting in Fig. Fig.7b).7b). Interestingly, a hyperreactive band corresponding to a T residue at position −42, visible in free DNA, almost disappeared at low XylS-CTD concentrations (Fig. (Fig.7a).7a). Concomitant with this, a band corresponding to a T residue at position −51 became hyperreactive, suggesting a shift in the bending center at the Pm promoter (Fig. (Fig.7a).7a). These results support sequential and directional binding of the two XylS-CTD monomers at Pm, generating a progressive curvature in the DNA between the two binding sites. Surprisingly, the hyperreactive band had not been previously observed except as a very weak signal in footprint assays using an antigen-tagged XylS C-terminal domain derivative (28).
In this work, we have analyzed XylS DNA binding properties based on two facts: (i) the XylS DNA binding domain was soluble when overproduced, unlike most proteins of the family (4); and (ii) this domain could reproduce all XylS features except for 3MB responsiveness. Gel filtration column chromatography of purified XylS-CTD showed that it behaved as a monomer in solution, and the protein was able to activate transcription in vivo (Table (Table2)2) and in vitro (12, 28). This is analogous to the situation in RhaS and ExsA, two members of the AraC family in which the C-terminal domain contains all of the determinants essential for activation (4, 63). The situation in XylS, ExsA, and RhaS contrasts with that of AraC and MelR. In AraC, the DNA binding domain strictly required a leucine zipper or an artificial covalent linkage-mediated dimerization to activate transcription from the PBAD promoter (61), and the MelR C-terminal domain did not activate transcription despite the fact it bound DNA and provoked a similar bend to full-length MelR (2, 25). Thus, XylS-CTD resembles those AraC family proteins related to stress response such as MarA, Sox, and Rob, which activate transcription as monomers critically and exclusively depending upon their intracellular concentrations.
XylS-CTD was able to bind a DNA probe containing only one binding site, rendering a single shift product (Fig. (Fig.3a).3a). When the entire binding site including the two direct repeats was probed, XylS-CTD binding gave rise to two distinct complexes that were formed sequentially (Fig. (Fig.3b).3b). Gel shift analysis at different acrylamide concentrations showed that CI represents one molecule of XylS-CTD bound to the promoter (Fig. (Fig.4),4), which is consistent with our observation that the protein is a monomer in solution (Fig. (Fig.2).2). The data also supported that CII represents two XylS-CTD molecules bound to the promoter. The DNase footprint protection pattern showed that XylS-CTD bound the two direct repeats at the Pm promoter in two consecutive steps, where the first XylS monomer occupied the proximal binding site, facilitating the binding of the second monomer to the distal site (Fig. (Fig.7).7). Previous results indicated that interactions between XylS HTH α6 and base pairs at B boxes were the most important contacts for XylS binding to its target sequence and subsequent transcription activation (13). However, it is well established that noncontacted bases can also have important effects on protein affinity by affecting DNA flexibility or intrinsic curvature (54). In fact, the Pm 35° ± 5° intrinsic curvature centered in the A track in the proximal binding site (19) increased progressively to 50° ± 5° and 98° ± 2° after binding of the first and second XylS-CTD monomers, respectively (Fig. (Fig.5).5). Concomitantly, the bending center moved toward the region between XylS binding sites (Fig. (Fig.11 and and5),5), in good correlation with the band of enhanced DNase sensitivity located at −51 (Fig. (Fig.7).7). AT base pairs have been referred to as “natural hinges” for protein-induced DNA bending (41). The A track present at XylS proximal binding site between boxes A2 and B2 might act as a flexible motif, increasing XylS-CTD affinity. Analysis of the MarA-DNA cocrystals showed that the two recognition helices are arranged in a parallel manner and separated by 27 Å, whereas the distance between consecutive major grooves of canonical DNA B form is 34 Å. Thus, MarA binding to its target DNA induces a 35° bend in that region. In Pm, the 35° bend is centered in the proximal site and probably explains the observation in DNase footprint assays that at low concentrations XylS-CTD preferentially occupies the proximal binding site. The movement of the bending center toward the distal site, together with the increase of the bending angle to 50°, seems to be required to allow binding of the second monomer, which finally entails an increase of the overall curvature to 98°. The solution structures of AraC-CTD and especially of ToxT containing a buried cis-palmitoleic acid ligand show that the two recognition helices are not parallel but divergently oriented, thus suggesting that not only the DNA but also the protein is distorted in the protein-DNA complex (33, 48). In ToxT, this conformational change is accomplished by the release of the fatty acid from the N-terminal domain, allowing the regulator to bind DNA (33). If protein affinity and complex stability could be enhanced by flexibility in the DNA binding site, it is tempting to speculate that in XylS more extensive protein-DNA contacts and interactions with RNA polymerase would be allowed as the Pm promoter gradually bends, despite the absence of a specific interprotein contact domain involved in dimerization. It should be noted that from our results we cannot distinguish if stability of the complex increased exclusively through the protein-DNA contacts favored by bent DNA or if additional effects caused by unexpected protein-protein interactions in DNA-bound XylS-CTD monomers also play a role (61).
Accumulated evidence suggests XylS native protein functions as a dimer (49). The absence of cooperativity observed in the binding of the two XylS-CTD monomers is probably due to the absence of the N-terminal dimerization domain present in the wild-type protein. This is the case for the Pseudomonas aeruginosa ExsA regulator, where binding of the entire protein to its PexsC target promoter showed a cooperativity that disappeared when only the C-terminal domain of the protein was assayed, suggesting cooperative binding was dependent upon the N-terminal domain (4). In XylS full-length protein, using cell extracts of the wild-type protein we have shown that only CII was formed (12). Complexes with a single monomer are only detectable when the proteins are unable to dimerize in solution, as in the dimerization mutant XylS3L (12) or in XylS-CTD (Fig. (Fig.3).3). Unfortunately, experiments with pure preparations of full-length XylS retaining the dimerization domain cannot be carried out because of the protein's poor solubility.
The Pm promoter can be classified as a class II promoter (5), in which multiple interactions between XylS and RNA polymerase would be established. In fact, XylS establishes interactions with the α-CTD subunit of the RNA polymerase (51, 52). The above-discussed XylS-induced bending process is likely to facilitate and stabilize XylS-RNAP contacts in vivo to form an active transcription initiation complex. In fact, besides protein-protein interactions induced by DNA curvature, some activators stimulate contacts of DNA sequences with the RNA polymerase, favoring the transition from a closed to open complex (24, 57). We have previously shown in chromatin immunoprecipitation assays that XylS recruits RNA polymerase to the Pm promoter in response to 3MB and also promotes open complex formation in permanganate footprinting assays (12). Preliminary results using XylS T283A and E284A mutations in the α-5 helix conserved patch involved in transcription activation (21) rendered XylS derivatives severely defective in their ability to activate transcription from the Pm promoter (P. Domínguez-Cuevas and Marqués, unpublished data). Further studies to define specific amino acid contacts between XylS and the sigma RNA polymerase subunit are required to determine the influence of XylS-induced DNA bending on transcription activation.
Figure Figure88 illustrates our current model of the molecular events underlying XylS-CTD Pm binding and transcriptional activation. Pm promoter intrinsic 35° bending centered in the proximal XylS binding site overlapping the RNA polymerase −35 hexamer (Fig. (Fig.8,8, step 1) facilitates interaction at the proximal binding site such that a XylS-CTD monomer first recognizes and binds this sequence (step 2). This first event would induce a shift in DNA bending angle toward the intersite region, together with a significant increase in the bending angle (50°), triggering a fast binding of the second monomer and recruitment of the RNA polymerase (step 3) (12). Contacts with RNA polymerase through α-CTD and probably also through the sigma factor would then favor open complex formation and transcription initiation (step 4). The set of results from this and previous works could point to wild-type XylS being a dimer in solution and binding Pm sequentially, with monomer 1 first contacting the proximal site, rapidly followed by DNA bending and setting of the second monomer to obtain a stable dimer-DNA complex.
This work was supported by the Ministerio de Educación y Ciencias grants BCM 2001-0515 to S.M. and BIO2006-05668 and VEM2004-08560 to J.L.R. and grant CIV344 from the Junta de Andalucía. P.D.-C. was financed by a Junta de Andalucía predoctoral fellowship.
We thank Patricia Marín for excellent technical assistance and Chechu Rodríguez-Herva for artwork.
Published ahead of print on 2 April 2010.