Activation of Gene Expression by a Ligand-induced Conformational Change of a Protein-DNA Complex*

IlvY protein binds cooperatively to tandem operator sites in the divergent, overlapping, promoter-regulatory region of the ilvYC operon of Escherichia coli. IlvY positively regulates the expression of the ilvC gene in an inducer-dependent manner and negatively regulates the transcription of its own divergently transcribed structural gene in an inducer-independent manner. Although binding of IlvY protein to the tandem operators is sufficient to repress ilvYpromoter-specific transcription, it is not sufficient to activate transcription from the ilvC promoter. Activation ofilvC promoter-specific transcription requires the additional binding of a small molecule inducer to the IlvY protein-DNA complex. The binding of inducer to IlvY protein does not affect the affinity of IlvY protein for the tandem operator sites. It does, however, cause a conformational change of the IlvY protein-DNA complex, which is correlated with the partial relief of an IlvY protein-induced bend of the DNA helix in the ilvC promoter region. This structural change in the IlvY protein-DNA complex results in a 100-fold increase in the affinity of RNA polymerase binding at theilvC promoter site. The ability of a protein to regulate gene expression by ligand-responsive modulation of a protein-DNA structure is an emerging theme in gene regulation.

IlvY protein binds cooperatively to tandem operator sites in the divergent, overlapping, promoter-regulatory region of the ilvYC operon of Escherichia coli. IlvY positively regulates the expression of the ilvC gene in an inducer-dependent manner and negatively regulates the transcription of its own divergently transcribed structural gene in an inducer-independent manner. Although binding of IlvY protein to the tandem operators is sufficient to repress ilvY promoter-specific transcription, it is not sufficient to activate transcription from the ilvC promoter. Activation of ilvC promoter-specific transcription requires the additional binding of a small molecule inducer to the IlvY protein-DNA complex. The binding of inducer to IlvY protein does not affect the affinity of IlvY protein for the tandem operator sites. It does, however, cause a conformational change of the IlvY protein-DNA complex, which is correlated with the partial relief of an IlvY protein-induced bend of the DNA helix in the ilvC promoter region. This structural change in the IlvY protein-DNA complex results in a 100-fold increase in the affinity of RNA polymerase binding at the ilvC promoter site. The ability of a protein to regulate gene expression by ligand-responsive modulation of a protein-DNA structure is an emerging theme in gene regulation.
LysR-type proteins are the most common class of transcriptional regulatory proteins in prokaryotes (1). Regulation by LysR-type proteins is distinguished by several highly conserved unique properties. LysR-type proteins typically activate transcription of a target gene(s) and autoregulate their own synthesis from a single regulatory locus. In most cases, the target gene is divergently transcribed from the structural gene encoding the LysR-type protein. Activation of the target gene requires the binding of a metabolically important small molecule inducer, whereas autoregulation of the LysR gene is inducer-independent. Inducer binding, however, does not significantly alter the DNA binding affinity of these proteins (1). Thus, unlike other well characterized inducer-responsive activator proteins, LysR-type proteins are unusual in that the binding of inducer activates transcription by affecting an activator property of the protein distinct from its DNA binding activity. However, the molecular basis of this regulation has not been determined for any member of this important class of transcriptional regulatory proteins.
IlvY protein is a prototypic member of the LysR family of transcriptional regulatory proteins and is required for the regulation of ilvC gene expression in Escherichia coli (2)(3)(4)(5). The ilvC gene encodes acetohydroxy acid isomeroreductase (EC 1.1.1.86), an enzyme involved in the biosynthesis of the branched chain amino acids, L-isoleucine, L-valine, and L-leucine (5). Together with IlvY protein, either substrate of this enzyme, ␣-acetolactate or ␣-acetohydroxybutyrate, induces the activation of ilvC gene expression. Like other LysRtype proteins, IlvY protein also represses transcription of its own divergently transcribed structural gene in an inducerindependent manner (3,4,6).
In this report, we describe the first molecular mechanism for inducer-mediated activation by a LysR-type protein. Specifically, we show that IlvY protein binds cooperatively to the tandem operator sites in the divergent-overlapping promoter region of the ilvYC operon ( Fig. 1) in an inducer-independent manner and autoregulates its own gene expression by an RNA polymerase occlusion mechanism. We further show that the binding of IlvY protein to the tandem operators is necessary but not sufficient to activate transcription from the ilvC promoter. Activation of transcription from the divergent ilvC promoter requires the additional binding of an inducer molecule to an IlvY protein-operator DNA complex. Inducer binding effects the partial relief of an IlvY protein-induced DNA bend centered around the Ϫ35 hexanucleotide element of the ilvC promoter. This ligand-induced structural change in the DNA helix enhances the binding affinity of RNA polymerase at the ilvC promoter. Thus, IlvY protein activates transcription from the ilvC promoter by the binding of an inducer molecule that directs a conformational change in the structure of an IlvY protein-operator DNA complex and enhances the recruitment of RNA polymerase to the ilvC promoter without affecting the occupancy of IlvY protein at the tandem operator sites.

MATERIALS AND METHODS
Chemicals and Reagents-Polyethyleneimine was purchased from Miles Laboratories. Restriction endonucleases, T4 DNA ligase, and T4 polynucleotide kinase were purchased from New England Biolabs. E. coli RNA polymerase, pancreatic RNasin, and DNase I were purchased from Boehringer Mannheim. Shrimp alkaline phosphatase was purchased from United States Biochemicals. Radiolabeled nucleotides were purchased from NEN Life Science Products. DNA oligonucleotides were synthesized by Operon Technologies. Site-directed mutagenesis was performed using the oligonucleotide directed in vitro mutagenesis kit version 2.0 from Amersham Pharmacia Biotech. DNA sequencing was performed using the Sequenase TM kit from United States Biochemicals.
Plasmids-Plasmid DNA isolation and all recombinant DNA manipulations were carried out using standard methods (7). Plasmid pET3cY, used for the overexpression of the ilvY gene product, was created by site-directed mutagenesis of the ilvY gene, contained on a 1400-bp BamHI-PvuII DNA fragment (derived from plasmid pRW1Y) (3). An NdeI restriction site was created at the beginning of the ilvY protein coding sequence. This site replaced the natural GTG translation initiation codon with the more common ATG codon. The 1225-bp NdeI-BamHI restriction fragment from the resulting plasmid, containing the ilvY gene, was ligated into the compatible NdeI-BamHI sites of the Studier vector pET3c (8) to yield the plasmid pET3cY. Plasmid pRWSR-2, used to generate the DNA fragments (containing ilvYC sequence from bp position Ϫ111 to ϩ1 cloned into the BamHI and EcoRI sites of pUC19) described in the binding and chemical footprinting experiments, has been previously described (6). Plasmid pBENDY, used to create the isometric O 1 O 2 -containing DNA fragments for circular permutation assays, was created by ligating an end-filled 132-bp EcoRI-HindIII DNA fragment containing the tandem O 1 O 2 operator sites into the unique SalI restriction site of plasmid pBEND2 (9) Plasmid pDD3Y, used for the in vitro transcription reactions, was constructed by ligating an end-filled 220-bp BglII-EcoRI DNA restriction fragment (containing ilvYC sequence from bp Ϫ220 to ϩ1 relative to the transcription start site of ilvC) into the BamHI site (end-filled) of the plasmid pDD3 (10). The BamHI site in pDD3 is flanked on both sides by tandem rrnBT 1 T 2 transcriptional terminator sequences (10).
Purification and Characterization of IlvY Protein-A 4-liter culture of strain BL21DE3 (8) containing plasmid pET3cY was grown to an A 600 ϭ 0.8 at 20°C, induced with the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.75 mM, and allowed to continue growing for an additional 6 -8 h. Cells were harvested by centrifugation and resuspended in 40 ml of an ice-cold solution containing 50 mM Tris-HCl (pH 7.6), 100 mM NaCl, 5% glycerol, and 0.5 mM dithiothreitol (Buffer A). Cells were disrupted by sonication in a Rosett cell using a Tekmar Sonic Disruptor set at 100 W. Sonication of cells was performed on ice with 12 ϫ 20 s bursts interrupted by 2-min cooling intervals. To avoid partial proteolysis owing to the high arginine/lysine ratio of the IlvY protein (11), phenylmethylsulfonyl fluoride was added to the cell suspension at a final concentration of 1 mM immediately before and after sonication. All subsequent steps were performed at 4°C or on ice. The crude cell lysate was cleared by centrifugation at 12,000 rpm in a Beckman JA20 rotor for 40 min at 4°C. The resulting supernatant solution was transferred to an Erlenmeyer flask, and a 10% (v/v) solution of polyethyleneimine was slowly added dropwise to a final concentration of 0.6% (v/v) with gentle stirring and incubated for an additional 20 min. This mixture was centrifuged for 30 min at 10,000 rpm in a Beckman JA20 rotor. The resulting precipitate was resuspended in 10 ml of Buffer A containing 0.3 M NaCl and gently stirred for 15 min. This suspension was centrifuged for 20 min at 12,000 rpm in a Beckman JA20 rotor. The resulting supernatant solution was slowly loaded onto a 4 ml phosphocellulose column pre-equilibrated in Buffer A. After loading the sample, the column was washed with one column volume of Buffer A. IlvY protein was eluted from the column with an 80-ml linear KCl gradient (0.1-0.5 M) in Buffer A. Typically, the protein eluted at a [KCl] ϭ 0.4 M. 1-ml fractions were collected and analyzed by SDS-polyacrylamide gel electrophoresis. 2 l of each fraction were assayed in a 20-l reaction volume using the mobility shift binding protocol described below. Appropriate fractions were pooled and concentrated in an Amicon Microcon-10 ultrafiltration device. Protein concentrations were measured according to the method of Bradford (12) using the Bio-Rad DC Protein Assay Kit. A final yield of about 0.5 mg of greater than 95% purified IlvY protein is routinely obtained from 1 liter of cells.
The specific DNA binding activity of the purified IlvY protein was estimated from a titration of its binding to a molar excess of DNA fragments containing the tandem operator sites. The concentration of O 1 O 2 -containing DNA fragments used yielded an operator site concentration of 4.4 ϫ 10 Ϫ8 M, which was titrated with substoichiometric concentrations of IlvY protein dimer between 1.9 ϫ 10 Ϫ9 M and 1.5 ϫ 10 Ϫ8 M. Bound and free DNA were separated by nondenaturing gel electrophoresis. The fraction of bound complex was proportional to the IlvY dimer concentration over the entire concentration range. As expected, this result was consistent with stoichiometric formation of a protein-DNA complex since the operator site concentration exceeds the effective binding dissociation constant for the operator sites (2.2 ϫ 10 Ϫ9 M; see below) by 20-fold. The fraction of active IlvY protein dimer estimated from the slope of the plot of operator sites bound per IlvY protein dimer was never less than 50%. The percent activity was used to calculate active dimer concentrations in subsequent experiments.
The quarternary structure of the IlvY protein was determined by gel filtration chromatography. A 0.05-ml sample containing 0.2 mg of purified IlvY protein in Buffer A was applied to a 0.6 ϫ 30-cm column of Sephacryl S200HR equilibrated in Buffer A. The IlvY protein was eluted from the column in Buffer A at a flow rate of 1.5 ml/hr. 110-l fractions were collected and assayed for protein content and for DNA binding activity using the mobility shift binding protocol described below. The elution profiles of a mixture of molecular weight standards (100 mg of alcohol dehydrogenase, 150 kDa; 100 mg of hexokinase, 102 kDa; 100 mg of ovalbumin, 46 kDa) were determined in a separate experiment. The molecular weight of an IlvY monomer based on its deduced amino acid sequence is 33 kDa. The DNA binding activity of the IlvY protein eluted from the column at the position of a globular protein with a molecular weight of 65-70 kDa. No evidence of lower or higher molecular weight species was apparent. Thus, at concentrations as high as 200 g/ml (6 M protein monomer), IlvY protein exists in solution as a homodimer.
Mobility Shift Binding Assays-Mobility shift binding assays were performed according to the methods of Fried and Crothers (13) as described by Wek and Hatfield (6) with the exception that gels were run in 1ϫ TAE buffer (40 mM Tris acetate (pH 8.5), 2 mM EDTA⅐2H 2 O). Radiolabeled DNA used in these experiments was present at final concentrations not exceeding 2 ϫ 10 Ϫ11 M to ensure that the free concentration of IlvY protein was equivalent to the total IlvY protein concentration. Binding reactions were performed in a final volume of 20 l containing 10 mM Tris-HCl (pH 7.6), 50 mM KCl, 10 mM MgCl 2 , 0.1 mM EDTA, 6 mM ␤-mercaptoethanol, 100 g of bovine serum albumin/ ml, and 2.5 g of sonicated herring sperm DNA/ml. Free and bound DNA fragments were visualized by autoradiography following the exposure of Kodak XAR-5 film to the dried gels overnight at 25°C. The autoradiograms were digitized with a Hewlett-Packard ScanJet IIcx digital scanner. Band intensities were measured using the public domain NIH IMAGE gel quantitation software. The intensity of each band was corrected for background differences and normalized to the total intensity of all bands in each lane. The equilibrium binding data were analyzed by nonlinear least squares estimations. The algorithm for this analysis (14) uses a variation of the Gauss-Newton procedure to determine the best fit model-dependent parameter values corresponding to a minimum in the variance of each data point (15). The confidence levels for the curve fits reported correspond to approximately one standard deviation (65% confidence).
DNase I Footprinting Experiments-DNase I footprinting reactions were performed according to the methods of Galas and Schmitz (16) as described by Wek and Hatfield (6) using uniquely 5Ј-32 P-end-labeled DNA fragments containing ilvYC DNA from bp positions Ϫ111 to ϩ1, derived from plasmid pRWSR-2. Binding reactions contained 10 mM Tris-HCl (pH 7.6), 50 mM KCl, 10 mM MgCl 2 , 0.1 mM EDTA, 6 mM ␤-mercaptoethanol, 100 mg of bovine serum albumin/ml, and 2.5 mg of sonicated herring sperm DNA/ml in a final volume of 40 l. Operator DNA was present at a final concentration of less than 10 Ϫ10 M. Purified IlvY protein was present at a final protein dimer concentration of 5 ϫ 10 Ϫ8 M. ␣-Acetohydroxybutyrate was present as specified in the figure legends. DNase I exposure was for 1 min at 25°C. This yielded less than 30% nicked DNA fragments, determined by Cerenkov counting of the band corresponding to uncleaved DNA. This low nicking frequency is not expected to affect the intrinsic protein-DNA binding (17). Samples were resolved by electrophoresis on an 8% denaturing polyacrylamide gel (7.6% acrylamide, 0.4% N,NЈ-methylenebisacrylamide) containing 8 M urea in TBE buffer (7) and visualized by autoradiography following the exposure of Kodak XAR-5 film to the gels at Ϫ70°C in the presence of a Cronex Quanta III intensifying screen (DuPont). The autoradiograms were digitized with a Hewlett-Packard ScanJet IIcx digital scanner. The band intensity in each lane corresponding to the DNase I-hypersensitive site located at bp Ϫ37 in the ilvC promoter region was measured using the public domain NIH IMAGE gel quantitation software. The intensity of each band was corrected for background differences and normalized to the intensity of the unprotected band in the same lane at bp Ϫ90.
Hydroxyl Radical Footprinting Experiments-Hydroxyl radical footprinting was performed as described by Tullius and Dombroski (18) using the same DNA fragments (Ͻ1 ϫ 10 Ϫ10 M) used in the DNase I footprinting reactions. Binding reactions were performed under equilibrium binding conditions as described above with the exception that glycerol was omitted from the reaction mixture. Samples were treated with hydroxyl radical for exactly 2 min at 25°C and quenched with the addition of 20 l of 0.2 M thiourea to ensure conditions of single-hit kinetics as described above. Reaction products were resolved by electrophoresis on a 10% denaturing polyacrylamide gel (9.5% acrylamide, 0.5% N,NЈ-methylenebisacrylamide) containing 8 M urea in TBE buffer (7) and visualized by autoradiography as described above. Data were analyzed by utilizing the public domain NIH IMAGE gel quantitation software. Regions of protection were determined by comparing the relative peak heights of each integrated band obtained from an integration of the bands of a digital image of each lane.
In Vitro Transcription Reactions-In vitro transcription reactions were conducted using the closed circular supercoiled (Ͼ80%) plasmid pDD3Y (described above), in the absence and presence of purified IlvY protein and/or inducer, according to the procedures of Hauser et al. (19). Reactions were terminated after 4 min with the addition of 20 l of stop solution (95% formamide, 0.025% bromphenol blue, 0.025% xylene cyanol). Transcription under these conditions was determined to be linear with respect to time for at least 6 min and was proportional to the amount of plasmid DNA template used. To correct for differences in the total amount of transcription between reactions, the amount of ilv-specific transcription was determined by normalizing the relative signal of ilv-specific transcription to the relative signal of transcription arising from the RNA-I promoter present on each plasmid DNA template in the transcription reaction. Reaction products were separated by electrophoresis on an 8% denaturing polyacrylamide gel (7.6% acrylamide, 0.4% N,NЈ-methylenebisacrylamide) containing 8 M urea in TBE buffer (7) and visualized by autoradiography as described above.
Circular Permutation Assays-Binding reactions were performed as described above using isometric DNA fragments containing the tandem O 1 O 2 operators derived from the plasmid pBENDY. Products were resolved by electrophoresis through an 8% nondenaturing polyacrylamide gel (7.92% acrylamide, 0.079% N,NЈ-methylenebisacrylamide) in 1ϫ TAE (20) buffer (7). In reactions performed with inducer, 1 mM ␣-acetolactate was included both in the gel and in the running buffer. Analysis of bend mapping was performed as described by Wu and Crothers (21). Estimates of bend angles were obtained as described by Thompson and Landy (22) using the empirical relationship cos(␣/2) ϭ R f (slowest)/R f (fastest).
Abortive The inclusion of the dinucleotide ApA in the reaction mixture restricts transcription initiation to the ilvC promoter of the plasmid DNA template, pDD3Y, yielding the abortive product ApApUpU. Abortive transcription products were isolated by nondenaturing electrophoresis in a 25% polyacrylamide gel (23.75% acrylamide, 1.25% N,NЈ-methylenebisacrylamide) in 1ϫ TBE buffer (7), and quantitated by Phosphor-Imager analysis. is a measure of the lag time required for open complex formation observed in a product versus time plot. obs values were determined by a nonlinear least-squares analysis of the kinetic data using the program LAGPLOT described by Goodrich (25). The apparent binding affinity of RNA polymerase (K B ) and the first-order isomerization rate constant (k 2 ) were determined by plotting obs as a function of 1/(RNA polymerase) on the basis of a least-squares fit to the equation obs ϭ (1/k 2 ) ϩ (1/k 2 K B (RNA polymerase)) using the TAUPLOT program of Goodrich (25).

Operator Binding by Purified IlvY Protein-Previous analyses with IlvY protein-enriched cell-free extracts indicated that
IlvY protein binds to tandem operator sites (O 1 O 2 ) in the divergent-overlapping ilvYC promoter region and that this binding is cooperative (6). To explore the quantitative nature of this cooperativity, binding titrations of purified IlvY protein to DNA fragments containing the tandem operators (O 1 O 2 ) and to DNA fragments with either O 1 or O 2 deleted were performed (Fig. 2). These assays were conducted using concentrations of operator sites well below the effective binding dissociation constant for the tandem operator sites to evaluate the microscopic binding constants governing the assembly of IlvY protein-operator DNA complexes. The gel mobility shift of the tandem operatorcontaining DNA fragment showed only a single liganded complex, which we interpret to be O 1 O 2 with both sites bound by IlvY protein dimer. The half-saturation point for IlvY binding to the DNA fragment containing O 1 and O 2 corresponds to an effective affinity of 2.2 nM, consistent with a previous estimate (2 nM) obtained with IlvY protein enriched cell free extracts (6). The lack of an intermediate band corresponding to IlvY protein bound to either O 1 or O 2 alone in the gel mobility shift experiment performed to generate the data in Fig. 2 is indicative of highly cooperative binding (17). In fact, the IlvY protein concentration required to half-saturate O 1 alone is 8-fold greater than the concentration required to half-saturate the tandem operator sites (Fig. 2). We found previously that it was not possible to saturate the O 2 site alone; at 8 ϫ 10 Ϫ8 M IlvY dimer, the highest concentration used in the binding experiments, less than 15% of the DNA migrated as bound in gel shift assays (6). Based on this measurement, the affinity of IlvY protein for the O 2 site alone is calculated to be K 2 Ͻ 0.45 M. When this experiment was repeated with the highly purified IlvY protein reported here, the same result was obtained. Therefore, the affinity for O 2 alone is at least 200-fold weaker than for sites O 1 O 2 together (6). The observation that the apparent binding affinity for either operator site is reduced when the other site is removed indicates cooperative binding to the adjacent sites.
To determine the affinity of IlvY protein for site O 1 and to more precisely evaluate the cooperativity, the IlvY protein ti-trations of O 1 O 2 and O 1 alone (Fig. 2) were analyzed according to a general cooperative binding model (17). The adjustable parameters are the Gibbs free energy changes corresponding to the microscopic equilibrium constants for IlvY protein binding to O 1 (⌬G 1 ϭ ϪRT ln K 1 ) and to O 2 (⌬G 2 ϭ ϪRT ln K 2 ), and for cooperativity (⌬G c ϭ ϪRT ln K coop ). When a complete titration of site O 2 alone is absent, the analysis does not provide independent estimates of ⌬G 2 and ⌬G c . Instead, we found such a high numerical correlation that increasing values of one compensated for decreasing values of the other. This correlation did not affect ⌬G 1 for which a precise estimate, ⌬G 1 ϭ Ϫ10.4 Ϯ 0.1 kcal/mol, was obtained. Thus, it was possible to define a lower limit to the cooperativity and an upper limit to the affinity for O 2 alone by conducting a series of analyses of the O 1 alone and O 1 O 2 data in which different values of ⌬G c were entered as fixed input parameters. For each fit, estimates of ⌬G 1 , ⌬G 2 , and the variance of the fit were obtained. A minimum in the variance was obtained for ⌬G c ϭ Ϫ3.0 kcal/mol. For ⌬G c Ն Ϫ2.6 kcal/mol the variance increased substantially reflecting the fact that it is not possible to account for the offset between the curves for IlvY protein binding to O 1 alone versus to O 1 O 2 without significant cooperativity. By comparing ratios of variances obtained for fits with different values of ⌬G c against an F-statistic, we conclude that ⌬G c Յ Ϫ2.6 kcal/mol, corresponding to at least 120-fold cooperativity, is required to obtain an adequate fit. This also sets a maximum affinity for O 2 of 35 nM (⌬G 2 Ն Ϫ10.0 kcal/mol). Setting ⌬G 2 ϭ Ϫ8.5 kcal/mol corresponding to our previous estimate of K 2 (0.45 M) yields ⌬G c ϭ Ϫ4.3 kcal/mol or 1500-fold cooperativity. Whether 1500-fold or only 120-fold, the cooperativity is sufficient that the O 1 O 2 operator exists as liganded at either O 1 alone or at O 2 alone less than 1% of the time at any IlvY dimer concentration, a conclusion that is consistent with the absence of any intermediate band in the gel shift assay. The effect is that IlvY protein fills the tandem operator sites in a single concerted step.
DNA Binding Site Specificity of Purified IlvY Protein-DNase I footprinting experiments were conducted to demonstrate that the purified IlvY protein exhibits site-specific binding to the tandem operator sites in the divergent promoter region of the ilvYC operon (Fig. 1). The results in Fig. 3A show that purified IlvY protein protects 2 adjacent 27-bp regions of DNA (ilvC bp Ϫ76 to Ϫ50 and Ϫ44 to Ϫ18) separated by 4 intervening DNase I-sensitive nucleotides on a 240-bp EcoRI-HindIII DNA fragment containing the tandem operator sites. These observed regions of protection on the nontranscribed strand of the ilvC gene correspond to those previously defined as O 1 and O 2 with partially purified IlvY protein (6). The results in Fig. 3B demonstrate that the purified protein protects three regions of DNA on the transcribed strand of the ilvC gene (ilvC bp Ϫ78 to Ϫ55, Ϫ51 to Ϫ43, and Ϫ41 to Ϫ19). These sequences are also contained in the DNA regions of dyad symmetry previously defined as O 1 and O 2 .
Effect of IlvY Protein Binding to O 1 O 2 on Transcription from the ilvY Promoter-To investigate how IlvY protein binding to the tandem operators, O 1 O 2 , affects transcription from the ilvY promoter, transcription from the ilvY promoter was assayed in vitro using a negatively supercoiled plasmid DNA template, pDD3Y. This plasmid contains the divergent ilvC and ilvY promoters flanked by Rho-independent ribosomal, rnnB T 1 T 2 , terminators. Transcription from the ilvY and ilvC promoters generates 369 nucleotide and 154 nucleotide products, respectively. The results in Fig. 4  The simplest model consistent with these repression data and with the observation that O 1 O 2 and the ilvY promoter occupy the same region of DNA on the same face of the DNA helix (6), is that repression is mediated by competition between IlvY protein and RNA polymerase binding. To assess whether such a model can account for the repression curve in Fig. 5, these data were analyzed using a simple competitive binding model in which the rule is that RNA polymerase binding to the ilvY promoter and IlvY protein dimer binding to either operator are mutually exclusive. The ⌬G values for IlvY protein binding to O 1 O 2 were set as equal to the best estimates obtained from analysis of the data in Fig. 2. The fitted parameter was the free energy change corresponding to the equilibrium dissociation constant for RNA polymerase binding to the ilvY promoter. The results of this analysis, shown as the solid curves in Fig. 5, match the offset between the two curves with a predicted K d for RNA polymerase binding to the ilvY promoter in the absence of IlvY protein of 2 Ϯ 0.6 ϫ 10 Ϫ9 M. We conclude, therefore, that IlvY protein represses transcription from the ilvY promoter by an RNA polymerase occlusion mechanism in an inducer-independent manner.
Effects of Inducer on IlvY Protein-ilvC Promoter Interactions-The binding of inducer to IlvY protein does not significantly increase the affinity of IlvY protein for the tandem operators. Yet, it results in the appearance of a DNase I hypersensitive band in the region of the O 2 operator sequence that overlaps the Ϫ35 region of the ilvC promoter (6). These results suggested that the binding of inducer might direct a conformational change of the IlvY protein-operator DNA complex. In addition, the binding of RNA polymerase to the ilvC promoter is detectable by DNase I footprinting only when both IlvY protein and inducer are present (6). We proposed, there- The 108-nucleotide transcript designated ori originates from the RNA-I promoter.

FIG. 5. Comparison of IlvY protein binding (‚) to the tandem operators (O 1 O 2 ) and repression of transcription originating from the ilvY promoter (Ⅺ).
IlvY protein DNA binding was determined by mobility shift experiments of the type described under "Materials and Methods." Transcription levels were determined from autoradiographic band intensities of ilvY promoter-specific transcripts obtained from experiments of the type illustrated in Fig. 4. All measurements were obtained from autoradiogram exposures where band intensities were linear with respect to time. The solid curves show the results of the analysis of these data as described in the text. For this analysis, the free energy changes used for binding of IvY protein to O 1 O 2 were: ⌬G 1 ϭ Ϫ10.4 kcal/mol; ⌬G 2 ϭ Ϫ9.8 kcal/mol; ⌬G c ϭ Ϫ3.0 kcal/mol. fore, that the role of inducer is to facilitate a conformational change in the IlvY protein-O 1 O 2 complex, which recruits RNA polymerase to the ilvC promoter (6). To test this hypothesis, in vitro transcription and DNase I footprinting assays were conducted to quantitatively compare the effects of inducer on the DNase I hypersensitivity in the ilvC Ϫ35 hexanucleotide region and on transcription from the ilvC promoter.
The autoradiogram in Fig. 6 shows the effect of increasing inducer concentrations on the DNase I footprinting pattern produced by the binding of IlvY protein to O 1 O 2 . At all inducer concentrations, IlvY protein remains bound to both operator sites. However, the DNase I hypersensitive site at bp Ϫ37 on the nontranscribed strand (relative to the ilvC transcription start site; Fig. 6A) increases in intensity as inducer concentration is increased. Inducer binding also results in the appearance of several DNase I hypersensitive sites on the transcribed strand (Fig. 6B). The inducer concentration dependence of the hypersensitive site on the nontranscribed strand is plotted in Fig. 7 (open squares).
In vitro transcription assays were performed with a negatively supercoiled plasmid DNA template, pDD3Y. In the absence of IlvY protein, basal level transcription from both the ilvY and ilvC promoters is observed (Fig. 8, lane 1). Addition of 4 ϫ 10 Ϫ8 M IlvY protein dimer, which yields greater than 99% saturation of O 1 O 2 (Fig. 2), represses production of the 369nucleotide ilvY transcript but does not significantly affect production of the 154-nucleotide transcript originating from the ilvC promoter (Fig. 8, lane 2). The addition of inducer, however, activates transcription from the ilvC promoter 10 -15-fold (Fig.  8, lanes 3-11). This inducer-mediated activation is plotted in Fig. 7 (open triangles).
The data derived from these ilvC transcription and DNase I hypersensitivity titrations (Fig. 7) were analyzed to estimate the affinity of the inducer, ␣-acetohydroxybutyrate, for the IlvY protein-O 1 O 2 complex. Using a simple, noncooperative binding model, a K d value of 0.31 mM (⌬G4.7 Ϯ 0.3 kcal/mol) was obtained from the transcription data and a K d value of 0.55 mM (⌬G ϭ Ϫ4.4 Ϯ 0.2 kcal/mol), was obtained from the DNase I data. Since these experimental values are indistinguishable from one another, both transcriptional induction and DNase I hypersensitivity appear to be the result of a single inducer binding event. Neither binding transition shows any indication of either positive or negative cooperativity.
These results demonstrate that the binding of IlvY protein to the tandem operators is necessary, but not sufficient, for activation of transcription from the ilvC promoter. Activation requires the additional binding of inducer to an IlvY proteinoperator DNA complex.
Effects of Inducer on the Conformation of the IlvY Protein-DNA Complex-The appearance of DNase I hypersensitive sites in and around the protected regions of DNase I footprints (Fig. 6) are often indicative of protein-induced DNA bends. In fact, other LysR type activator-repressor proteins have been shown to induce DNA bends at their target DNA binding sites; and, in some cases, these bends are modulated by the binding of an effector ligand (26)(27)(28). To examine the possibility that IlvY protein might incite a bend in the DNA helix, circular permutation assays were conducted according to the methods of Wu and Crothers (21). A 240-bp EcoRI-HindIII DNA fragment containing the tandem operator sites of the ilvYC operator-promoter region, was ligated into the unique SalI site of the plasmid pBEND2 (9), and isometric (circularly permuted) DNA fragments containing the tandem operators were generated by cleavage at six tandemly repeated restriction endonuclease sites flanking the SalI site. The relative base pair positions of the ilv specific DNA fragment within each of these circularly permuted fragments is identified in Fig. 9B. These DNA fragments were incubated with IlvY protein in the presence and absence of inducer, and the products of these binding reactions were resolved by nondenaturing polyacrylamide gel electrophoresis.
The binding of IlvY protein to these circularly permuted DNA fragments results in the formation of protein-DNA complexes with differing position-dependent electrophoretic mobil- Transcription levels were determined from autoradiographic band intensities of ilvC promoter-specific transcripts obtained from experiments of the type illustrated in Fig. 8 and normalized to the transcription level obtained at the highest inducer (AHB) level (1.25 mM). The band intensity in each lane corresponding to the DNase I hypersensitive site located at bp Ϫ37 in the ilvC promoter region was corrected for background differences and normalized to the intensity of an unprotected band in the same lane. All measurements were obtained from autoradiogram exposures where band intensities were linear with respect to time. ities (Fig. 9A), indicating a protein-induced DNA bend. The center of the IlvY protein-induced DNA bend was estimated by plotting the relative mobility of each of the IlvY protein-DNA complexes against the relative bp position of the center of each DNA fragment and extrapolating this curve to the relative bp position of the IlvY protein-DNA fragment with the slowest mobility (21). This localized the center of the IlvY proteininduced DNA bend to within the O 2 site near the DNase I-hypersensitive site at bp Ϫ37. In the presence of inducer, the center of the IlvY protein-induced bend of the DNA helix did not change but the difference between the mobilities of the fastest and slowest migrating complexes decreased (Fig. 9B). This result suggests that the binding of inducer to the IlvY protein-DNA complex relaxes the angle of the IlvY proteininduced bend of the DNA helix. In the absence of IlvY protein, the free DNA fragments also exhibited differing position-dependent electrophoretic mobilities. Using the empirical relationship cos ␣/2 ϭ R f (slowest)/R f (fastest) (22), the angles of IlvY protein-induced DNA bending were estimated to be 60°in the absence of inducer and 50°in the presence of 1 mM inducer. In the absence of IlvY protein, a sequence-specific DNA bend of 36°, also centered in the O 2 region, was observed (Fig. 9B).
Although the difference between these bend angles is small and may not be significant, additional evidence for this inducer-mediated change of IlvY protein-induced bending of the DNA helix was obtained from the results of DNase I footprinting experiments of the transcribed strand of the ilvC gene. In the absence of substrate inducer, IlvY protein protects three contiguous regions of DNA (ilvC bp Ϫ78 to Ϫ55, Ϫ51 to Ϫ43, and Ϫ41 to Ϫ19) in the operator sequences defined as O 1 and O 2 (Fig. 3B). The addition of inducer to an IlvY protein-DNA complex, however, results in the appearance of several DNase I-hypersensitive sites (ilvC bp Ϫ76, Ϫ70, Ϫ61, Ϫ57, Ϫ49, Ϫ47, and Ϫ30) aligned along one face of the DNA helix (Fig. 6B). The addition of inducer also results in the protection of an additional DNase I reactive site at ilvC bp Ϫ55 on the opposite face of the DNA helix. A schematic summary of these DNase I footprinting protection patterns in the presence and absence of inducer on both the transcribed and nontranscribed strands of the ilvC gene is shown in Fig. 10. Given the pattern and periodicity of DNase I reactive sites induced and protected, these results and the DNA bending assays suggest that the binding of inducer facilitates the conversion of a sharp IlvY protein-induced DNA bend to a smoother bend of a smaller angle.
The results of hydroxyl radical footprinting experiments also support an inducer-mediated conformational change in an IlvY protein-operator DNA complex. The binding of IlvY protein to the tandem operators results in the protection of nucleotide residues predominantly aligned along one face of the DNA helix, and alters the relative chemical reactivity of several residues within the protected regions. These results are shown in Fig. 11 and schematically summarized in Fig. 10. In addition, it is important to note that in both DNase I and hydroxyl radical footprinting experiments, IlvY protein remains bound to both operator sites even in the presence of inducer.
Effects of Substrate Inducer on the Kinetics of the Transcription Initiation Reaction at the ilvC Promoter-To determine the effects of inducer binding to a preformed IlvY protein-DNA complex on the kinetics of the transcription initiation reaction at the ilvC promoter, abortive transcription assays of the ilvC promoter region complexed with IlvY protein were performed in the presence and absence of inducer (23). An IlvY protein dimer concentration of 4 ϫ 10 Ϫ8 M, which yields greater than 99% saturation of O 1 O 2 (Fig. 2) was used in each transcription reaction. The results of these assays showed that the addition of 1 mM inducer to an IlvY protein-operator DNA complex increases the binding affinity of the ilvC promoter for RNA polymerase (K B ) nearly 100-fold and decreases the isomerization rate for open complex formation (k 2 ) approximately 7-fold (Table I). Thus, the principal effect of substrate-inducer binding to an IlvY protein-DNA complex on the transcription initiation reaction at the ilvC promoter is to increase the affinity of ilvC promoter for RNA polymerase. DISCUSSION We previously demonstrated that, in vivo, IlvY protein autoregulates expression of its own structural gene in the diver-gent-overlapping ilvYC operon in an inducer-independent manner and that both inducer and IlvY protein are required for the activation of ilvC gene expression (6). We further demonstrated that both activation and repression appear to be mediated from the same regulatory locus. Thus, we suggested that the binding of inducer to IlvY protein targets an activator property distinct from its DNA binding activity and that the inducer-directed activation of transcription from the ilvC promoter might functionally be correlated with a conformational change of a preformed IlvY protein-operator DNA complex (6). Previous work also demonstrated that the binding of inducer to IlvY protein does not significantly affect the occupancy of IlvY protein at the tandem operators (6). The results of DNase I and hydroxyl radical footprinting experiments reported here demonstrate that, in the absence or presence of inducer, IlvY protein remains bound to both operator sites. Therefore, these results definitively establish that the role of inducer in the IlvY protein-mediated activation of transcription from the ilvC promoter is not to recruit IlvY protein to the ilvC promoter region.
The binding of inducer to an IlvY protein-DNA complex evokes the induction of a hypersensitive DNase I site in the Ϫ35 hexanucleotide region of the ilvC promoter (Fig. 6) and other changes in the DNase I and hydroxyl radical footprinting patterns in the tandem operator region (Figs. 6, 10, and 11). These results suggest a ligand-induced conformational change in the IlvY protein-operator DNA complex. Quantitative analyses of these ligand-induced DNA structural changes further establish that this conformational change is functionally cor- related with transcriptional activation. For example, the inducer concentration that results in half-maximal induction of the DNase I hypersensitive site at bp Ϫ37 in the Ϫ35 hexanucleotide region of the ilvC promoter is the same concentration that half-maximally activates transcription from this promoter (Fig. 7).
The nature of the inducer-directed conformational change in the IlvY protein-operator DNA complex is suggested by the results of circular permutation assays (Fig. 9). These experiments indicate that the binding of IlvY protein to DNA fragments containing the tandem operator sites increases a sequenced-directed DNA bend of approximately 36°to a bend angle of approximately 60°. The mean geometric center of both the sequence-directed and IlvY protein-induced bends is positioned in the Ϫ35 hexanucleotide region of the ilvC promoter. The addition of inducer to the IlvY protein-operator DNA complex causes the partial relief of this bend to about 50°. This conclusion is supported by the distribution of hydroxyl radical and DNase I reactive sites on both strands of an IlvY proteinoperator DNA complex in the presence and absence of inducer. In fact, these results suggest that the binding of inducer facilitates the conversion of a sharp IlvY protein-induced DNA bend to a smoother bend of a smaller angle. We conclude, therefore, that alterations in the local geometry of the DNA helix around the ilvC promoter region play an important role in the IlvY protein-mediated activation of transcription from the ilvC promoter.
Protein-induced DNA bending is known to influence the rate of transcription initiation at many promoters. In some cases, protein-induced bending of the DNA helix enhances the binding of RNA polymerase to the promoter region (29). In other cases, it enhances the rate of open complex formation (23,24). The results of abortive transcription assays (Table I) indicate that the principal effect of the inducer-mediated conformational change in the IlvY protein-operator DNA complex is to increase the binding affinity of RNA polymerase for the ilvC promoter nearly 100-fold. This suggests that the functional effect of the inducer-mediated conformational change in the IlvY protein-DNA complex and the associated alterations in DNA bending is to remodel the structure of the poor Ϫ35 hexanucleotide region of the ilvC promoter region to make it a better template for RNA polymerase recognition. The results of these abortive transcription assays are also in general agreement with the proposal that the Ϫ35 region of a promoter is involved in the initial recognition of a promoter by RNA polymerase (29). The nucleotide sequence of the Ϫ35 hexanucleotide region of the ilvC promoter (TTTCCG) exhibits only a 3/6 match to the consensus sequence (TTGACA).
The results reported here also demonstrate that IlvY protein represses the expression of its own structural gene by a simple RNA polymerase occlusion mechanism. The results of in vitro transcription reactions conducted in the presence of increasing concentrations of IlvY protein show that IlvY protein binding to the tandem operators represses transcription from the ilvY promoter and that the fractional repression of ilvY expression and the fractional saturation of the tandem operator sites exhibit the same dependences on IlvY protein concentration (Fig. 5). The simplest interpretation consistent with this set of data is that IlvY protein competitively inhibits the binding of RNA polymerase at the ilvY promoter site. This conclusion is consistent with the results of chemical and enzymatic structural probing experiments, which demonstrate that both IlvY protein and RNA polymerase bind to overlapping DNA se-  quences on the same face of the DNA helix (6). It is interesting that the inducer titration curves of ilvC transcription shown in Fig. 7 correspond to a simple noncooperative binding model. This result shows that the activation of transcription from the ilvC promoter is not dependent on cooperative binding of inducer to multiple sites on the IlvY protein at the O 1 O 2 sites. These experiments also show that the measured inducer concentration required to half-maximally activate in vitro transcription from the ilvC promoter is nearly the same as that required to half-maximally saturate the enzyme product of the ilvC gene (5). These concentrations are in good agreement with the inducer concentrations required for in vivo activation of ilvC expression (6,30,31).
It is interesting to consider the activation mechanism described here in relation to that of the MerR protein (32), a mechanistically similar ligand-responsive activator protein which is not a member of the LysR family of proteins. In this case, the binding of the inducer, mercuric ion, to a MerRoperator DNA complex in the divergent-overlapping merR-merTPCAD operator-promoter region also facilitates the partial relief of a MerR-induced DNA bend. However, in this case the primary effect of the binding of inducer, and relief of protein-induced DNA bending, is the partial unwinding of the spacer region between the Ϫ 35 and Ϫ10 hexanucleotide regions of the merTPCAD promoter. This unwinding facilitates open complex formation and/or promoter clearance by the transcription apparatus rather than the recruitment of RNA polymerase to the promoter site as is the case for IlvY protein.
Thus, whereas both of these cases illustrate an important role for ligand-induced conformational changes of preformed protein-DNA complexes in the regulation of gene expression, they also highlight the functionally distinct roles these conformational changes may play. However, both of these examples emphasize the importance of protein-induced conformational changes on the structure of the DNA helix for the regulation of gene expression.
The kinetic effect of the ligand-induced, IlvY protein-directed, structural change in the IlvY protein-ilvC promoter DNA complex reported here is to recruit RNA polymerase to the promoter site. This might be accomplished by remodeling the DNA helix to facilitate improved RNA polymerase-DNA interactions. Alternatively, the primary effect of the inducermediated IlvY protein conformational change might be to facilitate protein-protein interactions between IlvY and RNA polymerase. In this case, the relaxation of the bend angle in the ilvC promoter region might be the fortuitous consequence of the inducer-mediated IlvY protein conformational change, and the primary role of IlvY protein might be to enhance RNA polymerase recruitment via facilitated protein-protein interactions with RNA polymerase across opposite faces of the DNA helix. The relative contributions of each of these components awaits further analysis.
Finally, it is interesting to note that inducer-responsive alterations in the angle of protein-induced bending of the DNA helix have also been reported for other LysR-type proteins such as OccR (33), CysB (28), and CatR (27). Thus, ligand-mediated alterations in protein-induced bending of the DNA helix appears to be a general feature among members of the LysR family. However, the functional relationship between ligandinduced conformational changes and the activation mechanism in these cases is uncertain. Thus, the initial elucidation of the activation mechanism employed by the IlvY protein reported here should facilitate our understanding of the regulatory mechanisms of other LysR-type regulatory systems.