BzdR, a Repressor That Controls the Anaerobic Catabolism of Benzoate in Azoarcus sp. CIB, Is the First Member of a New Subfamily of Transcriptional Regulators*

In this work, we have studied the transcriptional regulation of the bzd operon involved in the anaerobic catabolism of benzoate in the denitrifying Azoarcus sp. strain CIB. The transcription start site of the PN promoter running the expression of the bzd catabolic genes was identified. Gel retardation assays and PN::lacZ translational fusion experiments performed both in Azoarcus sp. CIB and Escherichia coli cells have shown that bzdR encodes a specific repressor that controls the inducible expression of the adjacent bzd catabolic operon, being the first intermediate of the catabolic pathway (i.e. benzoyl-CoA, the actual inducer molecule). This is the first report of a transcriptional repressor and a CoA-derived aromatic inducer controlling gene expression in the anaerobic catabolism of aromatic compounds. DNase I footprinting experiments revealed that BzdR protected three regions (operators) at the PN promoter. The three operators contain direct repetitions of a TGCA sequence that forms part of longer palindromic structures. In agreement with the repressor role of BzdR, operator region I spans the transcription initiation site as well as the -10 sequence for recognition of the RNA polymerase. Primary sequence analyses of BzdR showed an unusual modular organization with an N-terminal region homologous to members of the HTH-XRE family of transcriptional regulators and a C-terminal region similar to shikimate kinases. A three-dimensional model of the N-terminal and C-terminal regions of BzdR, generated by comparison with the crystal structures of the SinR regulator from Bacillus subtilis and the shikimate kinase I protein from E. coli, strongly suggests that they contain the helix-turn-helix DNA-binding motif and the benzoyl-CoA binding groove, respectively. The BzdR protein constitutes, therefore, the prototype of a new subfamily of transcriptional regulators.

In the last 3 decades, some insights have been added to the knowledge of the anaerobic catabolism of aromatic acids. Most of the studies have focused on the biochemical steps that lead to the mineralization of the aromatic ring in a few microorganisms such as certain strains of Rhodopseudomonas (photosynthetic ␣-Proteobacteria), Thauera, and Azoarcus (denitrifying ␤-Proteobacteria), in some Fe(III) and sulfate reducers, and in fermentative bacteria (1)(2)(3)(4).
Benzoate is a good model compound by which to study the anaerobic catabolism of aromatic compounds (1). A general strategy used by microbes to degrade anaerobically benzoate is its activation to benzoyl-CoA that is then degraded to central biosynthetic intermediates through a series of reactions that involve aromatic ring reduction, ␤-oxidation-like reactions, and ring cleavage as main steps. Biochemical and genetic insights on benzoate degradation have been reported in Rhodopseudomonas palustris, Magnetospirillum magnetotacticum MS-1, Thauera aromatica, Azoarcus evansii, and Azoarcus sp. CIB (1,2,5). Despite some progress on the enzymology of anaerobic catabolism of benzoate by these bacteria in recent years, very little is known about the regulation of the genes involved in this particular catabolic process. Although several genes have been suggested to be involved in the regulation of the anaerobic benzoate degradation, so far only badR and aadR have been shown to control the expression of a benzoate catabolic cluster in R. palustris (6,7).
The bzd cluster of Azoarcus sp. CIB codes for the set of proteins involved in the anaerobic catabolism of benzoate (2) (Fig. 1A). The bzd catabolic genes are organized in a single operon (bzdNOPQMSTUVWXYZA) encoding the enzymes that activate, dearomatize, and cleave the aromatic ring of benzoate yielding CoA derivative molecules that are channeled into the central metabolism of the cell (2). The bzd catabolic operon is driven by the P N promoter (Fig. 1A). Upstream of the bzd operon is located the bzdR gene that is expressed from the P R promoter (Fig. 1A) (2). It has been suggested previously that the bzdR gene might encode the specific transcriptional regulator (BzdR) of the bzd cluster (2).
By using different genetic and biochemical approaches, we show in this work that bzdR codes for a specific regulator of the bzd catabolic genes, being the first example of a benzoyl-CoA-dependent transcriptional repressor that is described in anaerobic catabolism of aromatic compounds. Moreover, primary sequence analyses and three-dimensional modeling of BzdR reveal an unusual modular structure that allows us to consider BzdR as the first member of a new subfamily of transcriptional regulators.
Molecular Biology Techniques-Recombinant DNA techniques were carried out by published methods (8). Plasmid DNA was prepared with a High Pure plasmid isolation kit (Roche Applied Science). DNA fragments were purified with Gene-Clean Turbo (Q-BIOgene). Oligonucleotides were synthesized on an Oligo-1000 M nucleotide synthesizer (Beckman Instruments, Inc.). All cloned inserts and DNA fragments were confirmed by DNA sequencing through an ABI Prism 377 automated DNA sequencer (Applied Biosystems Inc.). Transformation of E. coli cells was carried out by using the RbCl method or by electroporation (Gene Pulser; Bio-Rad) (8). Plasmids were transferred from E. coli S17-1pir (donor strain) into Azoarcus sp. recipient strains by biparental filter mating as described previously (2). Proteins were analyzed by SDS-PAGE as described previously (8). The protein concentration in cell extracts was determined by the method of Bradford (16) by using bovine serum albumin as the standard.
N-Terminal Amino Acid Sequence Determination-The N-terminal sequence of BzdR was determined by Edman degradation with a 477A automated protein sequencer (Applied Biosystems Inc.). A crude extract of E. coli DH5␣ (pBzdR4) cells was loaded in a 12.5% SDS-polyacrylamide gel, and the BzdR (298 amino acids) protein encoded by plasmid pBzdR4 was directly electroblotted from the gel onto a polyvinylidene difluoride membrane as described previously (17).
␤-Galactosidase and Benzoate-CoA Ligase Assays-␤-Galactosidase activities were measured with permeabilized cells as described by Miller (15). Benzoate-CoA ligase activities were measured through a direct spectrophotometric assay as described previously (2).
Sequence Data Analyses-The amino acid sequences of open reading frames were compared with those present in finished and unfinished microbial genome data bases using the TBLAST algorithm (18) at the National Center for Biotechnology Information server (available on the World Wide Web at www.ncbi.nlm.nih.gov/blast/blast.cgi). Multiple protein sequence alignments were made with the ClustalW (19) program at the INFOBIOGEN server (available on the World Wide Web at www.infobiogen.fr/services/menuserv.html). Phylogenetic analysis of the BzdR-like proteins was carried out according to the neighbor-joining method of the PHYLIP program (20,21) at the TreeTop-GeneBee server (available on the World Wide Web at www.genebee.msu.su/service/ phtree_reduced.html).
Construction of E. coli Strains Harboring Chromosomal P N ::lacZ Translational Fusions-Two different ЈlacZ translational fusions were accomplished. Thus, a 598-bp DNA fragment spanning from the 3Ј-end of the bzdR gene to the translation start codon of the bzdN gene was PCR-amplified using plasmid pECOR7 (Table I) as template and oligonucleotides 3REG (5Ј-GGGGTACCCGTGCATCAATGATCCGGCAAG-3Ј; an engineered KpnI site is underlined) and 5BZN (5Ј-GCTCTAGAC-CCATCGAACTATCTCCTCTGATG-3Ј; the bzdN start codon is indicated in boldface type, and an engineered XbaI site is underlined) (2). On the other hand, a 2070-bp DNA fragment spanning from the 5Ј-end of the bzdR gene to the translation start codon of bzdN was PCR-amplified using plasmid pECOR7 as template and oligonucleotides 5REG (5Ј-GGGGTACCGGTTTCGTCGCAGGTGCTGTCTGGC-3Ј; an engineered KpnI site is underlined) and 5BZN. The PCR-amplified fragments were digested with KpnI and XbaI restriction endonucleases and ligated to the KpnI/XbaI double-digested pSJ3 promoter probe vector, giving rise to plasmids pSJ3P N (P N ::lacZ) and pSJ3RP N (Pr-bzdR/P N ::lacZ) (Table I, Fig. 4). The correct lacZ translational fusions were confirmed by nucleotide sequence analysis. Plasmids pUTminiTn5P N and pUTminiTn5RP N were constructed by subcloning the NotI cassettes from pSJ3P N and pSJ3RP N into the mini-Tn5 delivery plasmid pUTminiTn5Km2 (Table I, Fig. 4), and they were transferred from E. coli S17-1pir to E. coli AFMC cells (Table I) by biparental filter mating (22). Exconjugants containing the P N ::lacZ and P R -bzdR/P N ::lacZ translational fusions inserted into their chromosomes were selected for the transposon marker, kanamycin, on rifampicincontaining LB medium, generating the E. coli AFMCP N and AFMCRP N strains, respectively (Table I).
Construction of Azoarcus sp. Strain CIBdbzdR-For disruption of the bzdR gene through single homologous recombination, a 499-bp internal fragment of bzdR was PCR-amplified by using primers BzdRInt5Ј (5Ј-GCGAGCGCAGCGGGGAATG-3Ј) and N-INV-III (5Ј-GCTTGCGAC-CACCGACTCC-3Ј), and it was cloned into the SmaI-digested pK18mob (a mobilizable plasmid that does not replicate in Azoarcus) ( Table I). The resulting construct, pK18mobbzdR (Table I), was transferred from E. coli S17-1pir (donor strain) into Azoarcus sp. strain CIB (recipient strain) by biparental filter mating (2). Exconjugants harboring the disrupted bzdR gene by insertion of the suicide plasmid, namely Azoarcus sp. strain CIBdbzdR, were isolated aerobically on kanamycin-containing MC medium lacking nitrate and containing 0.4% citrate as the sole carbon source for counterselection of donor cells. The mutant strain was analyzed by PCR to confirm the disruption of the target gene.
Overproduction and Purification of His 6 -BzdR-The recombinant plasmid pQE32-His 6 BzdR (Table I) carries the bzdR gene without the ATG start codon and with a His 6 tag coding sequence at its 5Ј-end, under control of the T5 promoter and two lac operator boxes. The His tail added 13 amino acids (MCGSHHHHHHGIL) to the N-terminal end of His 6 -BzdR-protein (34.842 Da). The His-tagged protein was overproduced in E. coli M15 strain harboring plasmid pREP4 ( Table I) that produces the LacI repressor to strictly control gene expression from pQE32 derivatives in the presence of isopropyl-1-thio-␤-D-galactopyranoside. E. coli M15 (pREP4, pQE32-His 6 BzdR) cells were grown at 37°C in 50 ml of ampicillin-and kanamycin-containing LB medium until the cultures reached an absorbance at 600 nm of 0.5. Overexpression of the His-tagged protein was then induced during 3 h by the addition of 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside. Cells were harvested at 4°C, resuspended in 5 ml of lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, pH 8.0, 20 mM imidazole), and disrupted by passage through a French press (Aminco Corp.) operated at a pressure of 20,000 p.s.i. The cell lysate was centrifuged at 26,000 ϫ g for 20 min at 4°C. The clear supernatant fluid was carefully decanted and applied to nickel-nitrilotriacetic acid-agarose columns (Qiagen). Columns were then washed at 4°C with 50 volumes of wash buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 40 mM imidazole, pH 8.0), and the His-tagged BzdR protein was eluted by using elution buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 750 mM imidazole, pH 8.0). The purified protein was dialyzed at 4°C in FP buffer (20 mM Tris-HCl, pH 7.5, 10% glycerol, 2 mM ␤-mercaptoethanol, and 50 mM KCl) and stored at Ϫ20°C.
Gel Retardation Assays-A 268-bp DNA fragment spanning from position Ϫ174 to ϩ79 with respect to the transcription start site of P N promoter ( Fig. 1B) was PCR-amplified from plasmid pSJ3P N (Table I) by using oligonucleotides 5BZN (see above) and RetIII (5Ј-GGGAATTC-CGAGCCTCGCGTTTTACTGC-3Ј; an engineered EcoRI site is underlined). This fragment was digested with XbaI and EcoRI restriction enzymes, and it was cloned in XbaI/EcoRI double-digested pUC18 vector, giving rise to plasmid pUC18FP4 ( Table I). The DNA fragment was then excised from pUC18FP4 by digestion with EcoRI and HincII restriction enzymes, and it was labeled by filling in the overhanging EcoRI-digested end with [␣-32 P]dATP (6000 Ci/mmol; Amersham Biosciences) and the Klenow fragment of E. coli DNA polymerase I as described by Sambrook et al. (8). The labeled fragment (P N probe) was purified by using Gene-Clean Turbo (Q-BIOgene). The retardation reaction mixtures contained 20 mM Tris-HCl, pH 7.5, 10% glycerol, 2 mM ␤-mercaptoethanol, 50 mM KCl, 0.05 nM DNA probe, 250 g/ml bovine serum albumin, and purified His 6 -BzdR protein in a 9-l final volume. After incubation of the retardation mixtures for 20 min at 30°C, mixtures were fractionated by electrophoresis in 5% polyacrylamide gels buffered with 0.5ϫ TBE (45 mM Tris borate, 1 mM EDTA). The gels were dried onto Whatman 3MM paper and exposed to Hyperfilm MP (Amersham Biosciences).
DNase I Footprinting Assays-The DNA fragment used for DNase I footprinting assays was PCR-amplified from plasmid pECOR7 by using oligonucleotides 5IVTPN (5Ј-CGGAATTCCGTGCATCAATGATCCG-GCAAG-3Ј), which hybridizes with the 3Ј-end of bzdR, and 3IVTPN (5Ј-CGGAATTCCATCGAACTATCTCCTCTGATG-3Ј; an engineered EcoRI site is underlined, and the bzdN start codon is indicated in boldface type). The amplified DNA fragment was then digested with PvuII and EcoRI restriction enzymes, and the resulting 376-bp substitution was singly 3Ј-end-labeled by filling in the overhanging EcoRIdigested end with [␣-32 P]dATP and the Klenow fragment as indicated above. The labeled fragment was purified by using Gene-Clean Turbo (Q-BIOgene). For DNase I footprinting assays, the reaction mixture contained 2 nM DNA probe, 500 g/ml bovine serum albumin, and purified protein in 15 l of FP buffer (see above). This mixture was incubated for 20 min at 30°C, after which 3 l (0.05 units) of DNase I (Amersham Biosciences) (prepared in 10 mM CaCl 2 , 10 mM MgCl 2 , 125 mM KCl, and 10 mM Tris-HCl pH 7.5) was added, and the incubation was continued at 37°C for 20 s. The reaction was stopped by the addition of 180 l of a solution containing 0.4 M sodium acetate, 2.5 mM EDTA, 50 g/ml calf thymus DNA, and 0.3 l/ml glycogen. After phenol/ chloroform extraction, DNA fragments were precipitated with absolute ethanol, washed with 70% ethanol, dried, and directly resuspended in 5 l of 90% (v/v) formamide-loading gel buffer (10 mM Tris-HCl, pH 8.0, 20 mM EDTA, pH 8.0, 0.05% (w/v) bromphenol blue, 0.05% (w/v) xylene cyanol). Samples were then denatured at 95°C for 2 min and fractionated in a 6% polyacrylamide-urea gel. A ϩ G Maxam and Gilbert reactions (23) were carried out with the same fragments and loaded in the gels along with the footprinting samples. The gels were dried onto Whatman 3MM paper and exposed to Hyperfilm MP. Primer Extension Analysis-Azoarcus sp. CIB cells harboring plasmid pBBR2P N (P N ::lacZ) were grown anaerobically on benzoate-containing MC medium until the culture reached an A 600 of 0.3. Total RNA was isolated by using the RNeasy Mini kit (Qiagen) according to the instructions of the supplier. Primer extension reactions were carried out with the avian myeloblastosis virus reverse transcriptase (Promega) and 10 g of total RNA as described previously (24), using oligonucleotide 5BZN (which hybridizes with the coding strand of the bzdN gene; see above) labeled at its 5Ј-end with phage T4 polynucleotide kinase and [␥-32 P]ATP (3000 Ci/mmol; Amersham Biosciences). To determine the length of the primer extension products, sequencing reactions of pBBR2P N were carried out with oligonucleotide 5BZN by using the T7 sequencing kit and [␣-32 P]dCTP (Amersham Biosciences) as indicated by the supplier. Products were analyzed on 6% polyacrylamide-urea gels. The gels were dried onto Whatman 3MM paper and exposed to Hyperfilm MP (Amersham Biosciences).
Modeling of BzdR-The threading methods used to obtain the threedimensional models for the N-and C-terminal regions of BzdR were BIOINBGU (25), FFAS (26), FUGUE (27), HMM (28), and 3D-PSSM (29). These programs provide a list of hits with known three-dimensional structures, which presumably are similar to that of the query protein, and thus can be used as modeling templates (Table II). Sequence alignment results between the BzdR sequences and those of the templates were analyzed with ClustalX (19). Once a modeling template was chosen, a full-atom three-dimensional model was obtained by using the SWISS-MODEL Protein Modeling Server (30,31). The models were energy-minimized with CNS (32), and their structures were validated using PROCHECK (33) and the ERRAT Protein Verification Server (34). The final models after energy minimization with CNS showed a correct stereochemistry, with 100% of the predicted residues in allowed regions in the Ramachandran plot (not shown). Additionally, the results obtained with ERRAT (34) indicate that the percentage below the 95% confidence limit is 77.3 and 86.3 for N-terminal and C-terminal BzdR regions, respectively. Manual docking of the interaction between BzdR and benzoyl-CoA was accomplished by using the program O (35).

BzdR Is the Prototype of a New Subfamily of Transcriptional
Regulators-Determination of the N-terminal amino acid sequence (MSNDENSSRL) of the bzdR gene product (see "Experimental Procedures"), suggests that BzdR is a protein of 298 amino acids (33,488 Da). Primary sequence analyses and amino acid sequence comparisons revealed that the BzdR protein exhibits two distinct regions. The N-terminal region (residues 1-130) shows a significant similarity with members of the helix-turn-helix (HTH) 1 -XRE family of transcriptional reg-FIG. 1. Scheme of the bzd cluster and P N promoter from Azoarcus sp. CIB. A, the bzdR regulatory gene and the bzdNOPQMSTUVWXYZA catabolic operon are shown as black and white boxes, respectively. The P R and P N promoters are indicated as thick white and gray arrows, respectively. The BzdR protein is represented as a black ellipse; benzoyl-CoA is represented as a white rectangle. The minus sign indicates repression of P N by the BzdR protein. The ellipse within the rectangle represents a BzdR protein that is not able to repress the P N promoter. In the anaerobic catabolism of benzoate, this molecule is first activated to benzoyl-CoA by a benzoate-CoA ligase (BzdA) and is then channeled to the central metabolism of the cell through a central pathway that involves the products of the genes bzdNOPQMSTUVWXYZ. B, expanded view of the P N promoter. The nucleotide sequence from positions Ϫ174 to ϩ79 is indicated. The transcription start site (ϩ1) and the inferred Ϫ10 and Ϫ35 boxes of the P N promoter are indicated. The ribosome binding site (RBS) and the ATG start codon of the bzdN gene are also shown in italic and boldface type, respectively. The BzdR-binding region I, II, and III (operators) are boxed. Inverted repeats within the operator regions are marked with convergent arrows below the sequence. ulators (SMART release; available on the World Wide Web at smart.embl-heidelberg.de/smart) that includes more than 1300 proteins from eukaryota, archaea, bacteriophages, and bacteria with an HTH DNA-binding motif similar to that of the well characterized Cro protein of phage (36,37). The highest amino acid sequence identity (34%) was found with SinR ( Fig.  2), a 14-kDa multifunctional protein that, besides repressing sporulation, regulates genes for motility, alkaline protease expression (aprE), and competence development in Bacillus subtilis (38). The C-terminal region of BzdR (residues 131-298) is homologous to shikimate kinases (PFAM accession number PF01202), enzymes that catalyze the conversion of shikimate to shikimate 3-phosphate using ATP as co-substrate, and it shows a sequence identity of 23% with the E. coli shikimate kinase I (aroK gene product) (Fig. 2) (39).
All of these data strongly suggest that BzdR is the specific regulator that controls expression of the bzd catabolic operon in Azoarcus sp. CIB. Since the modular organization of BzdR has not been reported so far in other regulators of the HTH-XRE family, BzdR might constitute the prototype of a new subfamily of transcriptional regulators.
BzdR Is the Specific Repressor of the bzd Catabolic Operon-To determine whether bzdR is indeed involved in the regulation of the anaerobic benzoate degradation, a bzdR mutant of Azoarcus sp. strain CIB was constructed by gene disruption as described under "Experimental Procedures." The mutant, Azoarcus sp. CIBdbzdR (Table I), grew as the wild-type Azoarcus sp. CIB strain in benzoate-containing minimal medium (data not shown), which suggests that the BzdR protein, if playing any regulatory role, is not an activator but a repressor of the bzd catabolic operon. When we measured expression of the last gene of the bzd operon (i.e. the bzdA gene encoding benzoate-CoA ligase) (Fig. 1A), we observed that whereas At the bottom is shown a suggested consensus sequence for BzdR-like proteins. The amino acid residues of each sequence are numbered on the right. Sequences were aligned using the multiple sequence alignment program Clustal (19). Amino acids are indicated by their standard one-letter code. Dark gray shows identical residues in all of the BzdR-like sequences. Light gray indicates identical residues in more than 80% of the BzdR-like sequences. The Walker A motif is boxed. ␣ and ␤ refer to ␣-Proteobacteria and ␤-Proteobacteria subgroups, respectively. Secondary structure elements predicted from the BzdR three-dimensional model (Fig. 10) are drawn at the bottom of the alignment.
Azoarcus sp. CIB cells showed a benzoate-inducible expression of bzdA, a high level of benzoate-CoA ligase activity was detected in Azoarcus sp. CIBdbzdR cells growing either in the presence of benzoate (inducing conditions) or pyruvate (noninducing conditions) (Fig. 3A). This result indicates a constitutive expression of bzdA in Azoarcus CIB sp. cells that lack the bzdR gene, which is in agreement with BzdR being a putative repressor of the bzd catabolic operon. To confirm that BzdR controls the expression of the bzd catabolic operon at the level of the P N promoter, plasmid pBBR5P N , which harbors a P N ::lacZ translational fusion (Table I), was introduced both in wild-type Azoarcus sp. CIB cells and in Azoarcus sp. CIBdbzdR cells. Whereas a significant ␤-galactosidase activity was observed when Azoarcus sp. CIB (pBBR5P N ) cells grew in benzoate-containing minimal medium but not when the cells used pyruvate as the sole carbon source (Fig. 3B); high levels of ␤-galactosidase were observed when Azoarcus sp. CIBdbzdR (pBBR5P N ) cells grew either in benzoate or pyruvate as the sole carbon sources (Fig. 3B). These data indicate a benzoateinducible expression of the P N ::lacZ reporter fusion when the cells contain the BzdR protein and a constitutive expression of the reporter fusion in the absence of BzdR. All of these results taken together suggest that BzdR negatively regulates the expression of the bzd catabolic operon by repressing the P N promoter when Azoarcus sp. CIB cells do not use benzoate as a carbon source.
To further investigate the regulatory role of the BzdR protein on the expression of the P N promoter, E. coli AFMC cells were transformed with plasmids pSJ3P N (P N ::lacZ), pSJ3RP N (P R -bzdR/P N ::lacZ), which harbors the bzdR gene under control of its own promoter located adjacent to the P N ::lacZ reporter fusion (Fig. 4), and pCK01BzdR (Plac::bzdR), which encodes  (Table I). Whereas ␤-galactosidase assays of permeabilized E. coli AFMC (pSJ3P N ) cells grown anaerobically in glycerolcontaining mineral medium revealed expression of P N (Fig.  5A), the presence of the BzdR protein both in cis (E. coli AFMC (pSJ3RP N ) cells) or in trans (E. coli AFMC (pSJ3P N , pCK01BzdR) cells) repressed the expression of the P N ::lacZ translational fusion (Fig. 5A). As expected, ␤-galactosidase levels in E. coli AFMC cells harboring plasmid pSJ3P N and the control plasmid pCK01 were similar to those observed in E. coli AFMC (pSJ3P N ) cells (Fig. 5A). To analyze faithfully the bzd regulatory system at monocopy dosage, the P N ::lacZ and P R -bzdR/P N ::lacZ reporter fusions were subcloned as NotI-DNA cassettes within mini-Tn5 vectors, rendering plasmids   (oriColE1 and oriR6K) and RP4-mediated mobilization functions (oriTRP4) are also indicated. Ap r and K m r show the genes encoding ampicillin and kanamycin resistance, respectively. tnp*, gene devoid of NotI sites encoding Tn5 transposase. 1, 2, and 3 indicate the oligonucleotides 5REG, 3REG, and 5BZN (see "Experimental Procedures"), respectively, used for PCR amplification of the P N promoter (oligonucleotides 2 and 3) and the bzdR-P N fragment (oligonucleotides 1 and 3). Relevant restriction sites shown are EcoRI (E), HindIII (H), KpnI (K), NotI (N), and XbaI (X). pUTminiTn5P N and pUTminiTn5RP N , respectively (Fig. 4), which were then used to deliver by transposition the corresponding translational fusions into the chromosome of E. coli AFMC. The presence of a strong T7 phage transcriptional terminator downstream of the lacZ fusion and the orientation of such fusions within the mini-Tn5 elements (Fig. 4) prevented read-through transcription from nearby promoters after insertion into the chromosome of the resulting E. coli AFMCP N and AFMCRP N strains (Table I). Despite the fact that ␤-galactosidase levels of E. coli AFMCP N cells harboring the control plasmid pCK01 were about 1 order of magnitude lower than those observed with E. coli AFMC (pSJ3P N ) cells (Fig. 5B) due to the decrease in the genetic dosage, the repressor role of BzdR on the P N promoter was confirmed by monitoring a significant reduction of ␤-galactosidase activity both in E. coli AFMCRP N and E. coli AFMCP N (pCK01BzdR) cells (Fig. 5B).
It should be noted that all transcriptional regulators so far described in the anaerobic catabolism of aromatic compounds (i.e. BadR, AadR, and HbaR controlling degradation of benzoate and p-hydroxybenzoate in R. palustris (6,7) and TutBC and TdiRS controlling degradation of toluene in Thauera sp. T1 and T. aromatica K172, respectively (41,42)) were shown (BadR, AadR, and HbaR) or were suggested to be (TutBC and TdiRS) activators of their corresponding promoters. The BzdR protein from Azoarcus sp. CIB constitutes, therefore, the first transcriptional repressor that is shown to control the expression of genes involved in the anaerobic catabolism of aromatic compounds.
Benzoyl-CoA Is the Actual Inducer of the Benzoate Degradation Pathway-To elucidate whether the substrate of the bzd pathway from Azoarcus sp. CIB (i.e. benzoate) was the inducer of the bzd catabolic operon, E. coli AFMCRP N cells were grown anaerobically on minimal medium supplemented with glycerol in the presence of benzoate and the expression of the P N ::lacZ fusion was monitored through ␤-galactosidase assays. As shown in Fig. 5C, the addition of benzoate did not alleviate the BzdR-mediated repression of the P N ::lacZ fusion, which indicates that benzoate is not able to act as an inducer of the P N promoter. To check whether the first intermediate of the an-aerobic benzoate degradation pathway (i.e. benzoyl-CoA) could be the real inducer of the bzd catabolic genes, it was necessary to engineer an E. coli strain able to synthesize benzoyl-CoA. To do that, the bzdA gene that codes for the anaerobic benzoate-CoA ligase from Azoarcus sp. strain CIB (Fig. 1A) (2) was cloned into the low copy number pCK01 vector rendering plasmid pCK01BzdA (Table I). As expected, cell extracts from E. coli AFMCRP N (pCK01BzdA) cells showed benzoate-CoA ligase activity (0.98 mol min Ϫ1 mg Ϫ1 ). Interestingly, E. coli AFMCRP N (pCK01BzdA) cells growing anaerobically in glycerol-containing minimal medium presented ␤-galactosidase levels similar to those observed with E. coli AFMCP N cells when they grew in the presence of benzoate but not when benzoate was omitted from the culture (Fig. 5C). These results strongly suggest that the activation of the P N promoter is triggered by the benzoyl-CoA produced from the benzoate added to the culture medium in E. coli cells that express the bzdA gene. To check the specificity of the induction, we analyzed whether a benzoyl-CoA analogue such as phenylacetyl-CoA could behave as an inducer of the P N ::lacZ fusion. To accomplish this, we transformed plasmid pAFK3, which expresses the paaK gene encoding the phenylacetate-CoA ligase from E. coli (Table I) (12), in E. coli AFMCRP N cells. As shown in Fig. 5C, phenylacetate was not able to alleviate the repression of the P N promoter in E. coli AFMCRP N (pAFK3) cells growing in glycerolcontaining minimal medium, suggesting that BzdR is not able to recognize the phenylacetyl-CoA formed by the PaaK ligase. To further investigate the range of aromatic CoA derivatives able to act as inducers of the P N promoter, we added 2-fluorobenzoate, 3-fluorobenzoate, 2-chlorobenzoate, 3-chlorobenzoate, and isonicotinate, all of them substrates of the BzdA enzyme (2), to E. coli AFMCRP N (pCK01BzdA) cells growing anaerobically in glycerol-containing minimal medium. As shown in Fig. 5C, only 2-fluorobenzoate was able to avoid significantly the repression of the P N ::lacZ fusion by the BzdR protein. Since Azoarcus sp. strain CIB is able to grow on benzoate and 2-fluorobenzoate as the sole carbon sources but, however, is unable to catabolize any of the other aromatic compounds cited above, it is tempting to speculate that BzdR is only able to efficiently recognize aromatic CoA derivatives that can be catabolized through the bzd-encoded pathway, suggesting that this regulator has evolved to become efficiently adapted to such a particular catabolic system.
Although previous reports have suggested that the inducer of the genes involved in anaerobic catabolism of benzoate in R. palustris and T. aromatica might be benzoyl-CoA rather than benzoate (1,43), no direct experimental evidence supporting this assumption has been provided. Here we present for the first time experimental evidence that benzoyl-CoA is the actual inducer of the pathway in Azoacus sp. CIB. It should be noted that another aromatic CoA derivative, phenylacetyl-CoA, was shown to be the inducer of an aerobic hybrid pathway, the phenylacetic acid degradation pathway, whose intermediates are also CoA-derived compounds (11). Whether transcriptional regulators of both anaerobic and aerobic hybrid pathways for catabolism of aromatic acids have evolved to recognize CoAderived aromatic compounds rather than the free acids is a hypothesis that requires further confirmation when additional regulatory systems of such type of degradative routes become characterized.
In Vitro Binding of BzdR to the P N Promoter-To study in vitro the interaction of the BzdR protein with the P N promoter, we first mapped the transcription start site of this promoter and overproduced the regulatory protein in recombinant E. coli cells. Primer extension analysis was performed with total RNA isolated from Azoarcus sp. strain CIB cells containing plasmid pBBR2P N (Table I). The transcription initiation site in the P N promoter was mapped 75 nucleotides upstream of the ATG translation initiation codon of the bzdN gene (Fig. 6), showing putative Ϫ10 (TAACAT) and Ϫ35 (TCAACA) boxes typical of 70 -dependent promoters (Fig. 1B).
To overproduce BzdR, we have engineered plasmid pQE32-His 6 BzdR ( Table I) that expresses a His-tagged BzdR protein (see "Experimental Procedures"). As expected, the His 6 -BzdR protein was a repressor of the P N promoter in E. coli AFMCP N (pQE32-His 6 BzdR) cells growing anaerobically in glycerol-containing minimal medium (Fig. 7A). Induction by benzoyl-CoA of the His 6 -BzdR mediated repression of P N was confirmed in E. coli AFMCP N (pQE32-His 6 BzdR, pCK01BzdA) cells growing anaerobically in glycerol-containing minimal medium in the presence (induction) or absence of benzoate (Fig. 7A). These results indicate that the functionality of His-tagged BzdR is similar to that of the wild-type protein, and, thus, His 6 -BzdR can be used to study the interaction of this regulator with the cognate P N promoter. The purification of the His 6 -BzdR protein was carried out through affinity chromatography as described under "Experimental Procedures," and it was checked by SDS-PAGE (Fig. 7B). To demonstrate the interaction of the BzdR regulatory protein with the P N promoter, the purified His 6 -BzdR protein was subjected to gel retardation assays, using as probe a 268-bp DNA fragment that carries the P N promoter region from position Ϫ174 to ϩ79 (P N probe) (Fig. 1B). The His 6 -BzdR protein was able to retard the migration of the P N probe in a protein concentration-dependent manner (Fig. 8A). Moreover, binding of BzdR to the P N promoter was highly specific, because whereas it was inhibited by adding an unlabeled P N probe to the retardation assays, it was not affected by adding an unlabeled heterologous probe (data not shown). Depending on the amount of His 6 -BzdR protein used in the retardation assay, two distinct complexes, complex 1 (higher mobility) and complex 2 (lower mobility), were observed (Fig. 8A). Thus, whereas complex I appeared preferentially at low amounts (2.5-10 nM) of His 6 -BzdR protein, 5-fold higher amounts of His 6 -BzdR gave rise to complex II (Fig. 8A). These results might suggest that BzdR binds to different regions at the P N promoter (see below). Gel retardation assays with His 6 -BzdR were also carried out in the presence of different concentrations of the benzoyl-CoA (Sigma) inducer molecule. As shown in Fig. 8B, increasing concentrations of benzoyl-CoA ranging from 250 M to 2 mM significantly reduced the amount of bound P N probe, suggesting that benzoyl-CoA inhibits binding of His 6 -BzdR to the P N promoter. Although a percentage of P N probe (about 20%) remained free when 2 mM benzoate was added to the retardation assay, this percentage increased to 95% when 2 mM benzoyl-CoA was used (Fig. 8B). All of these data taken together are in agreement with the lacZ-reporter fusion experiments reported above, and they confirm BzdR as the repressor of the P N promoter and benzoyl-CoA as the actual inducer molecule. Nevertheless, the molecular mechanism of the benzoyl-CoA-mediated induction of the P N promoter is still unknown, and it will be subject of future research.
Identification of the BzdR Binding Site(s)-The purified His 6 -BzdR protein was also used to determine the BzdR-binding site(s) (operator) in the P N promoter region by DNase I footprinting. As shown in Fig. 9, the His 6 -BzdR protein protected three different regions: region I (63 bp) spanning from position Ϫ32 to ϩ31, region II (21 bp) spanning from position Ϫ83 to Ϫ63, and region III (21 bp) spanning from position Ϫ146 to Ϫ126 (Fig. 1B). The three protected regions contain direct repetitions of a sequence, TGCA, that forms a part of longer palindromic structures (Fig. 1B). Whereas the TGCA sequences are separated by 6 nucleotides in regions II and III, the longer region I presents a pair of TGCA sequences separated by 1 nucleotide and another pair of TGCA sequences separated by 15 nucleotides (Fig. 1B). Other members of the HTH-XRE family of transcriptional regulators (see above) such as the SinR regulator from B. subtilis and the Cro and 434 repressors from phages and 434, respectively, also bind to short repeated sequences that, in most cases, are located within palindromic regions that span the promoters (38, 44 -46). Binding of such regulators to the multiple operators of their cognate promoters generates different protein-DNA complexes in gel retardation assays (38,47), a pattern of binding that was also observed with BzdR at the P N promoter (Fig. 8). Moreover, binding of BzdR induces changes in the DNA structure of P N as revealed by several phosphodiester bonds that become hypersensitive to DNase I cleavage (Fig. 9). The fact that BzdR-binding region I spans the transcription initiation site as well as the Ϫ10 sequence for recognition of the 70 -RNA polymerase (Fig. 1B) is in agreement with the observed repressor role of BzdR at the P N promoter. Nevertheless, to precisely understand the molecular basis of the BzdR-mediated repression at the P N promoter, more work needs to be done.

Modeling of BzdR and Docking Simulations of Benzoyl-CoA
Binding-With the aim of building three-dimensional models of BzdR, we carried out a -fold recognition approach for the N-terminal region, from residue 25 to 87 (N-BzdR), and the C-terminal region, residues 131-298 (C-BzdR), of the protein.
The five threading methods used in this work found potential templates with high scores, always above the 95% certainty level (see "Experimental Procedures"). The threading results for N-BzdR revealed that all of the templates found belong to the same SCOP superfamily, the repressor-like DNA-binding domains. Pairwise comparisons of the C␣ atoms of the crystal structures of all of the templates found, as calculated with the program O (35), yield a maximum root mean square deviation of 1.3 and correspond to a five-helix closed bundle folding (Fig.  10A). The template selected for building a three-dimensional model of N-BzdR was SinR of B. subtilis (1b0n) (Table II) it shares the highest identity. The presence within this fold of an HTH (residues 38 -58 of BzdR) (Figs. 2 and 10A), a motif found in numerous DNA-binding proteins (51), provides a reasonable structural basis for suggesting N-BzdR as the region directly interacting with the target DNA.
The threading results obtained for C-BzdR also provided consistent results, since all of the hits found share the same SCOP fold, the P-loop-containing nucleoside triphosphate hydrolase fold, almost all of them belonging to the shikimate kinase family (see "Experimental Procedures"). The template used to build a full-atom three-dimensional model of C-BzdR was the shikimate kinase I of E. coli (52, 53) (Fig. 2). Interestingly, the amino acid sequence identity shared between C-BzdR and the selected template (23%) is similar to that observed between the two different isoenzymes of E. coli, shikimate kinases I and II (30% identity) (39). This observation indicates that the protein fold of shikimate kinases is highly versatile, since it can accommodate a wide array of sequences without significant structural departures. As shown in Fig.  10B, C-BzdR presents a canonical mononucleotide-binding fold (54) found in a number of structurally diverse proteins (55), which is constituted by a five-stranded parallel ␤-sheet, strand order 23145, flanked by eight ␣-helices. Within this highly conserved fold, a phosphate-binding loop (P-loop) or Walker-A motif (56,57) can be observed between ␤ 1 and ␣ 6 (Fig. 10B). The Walker-A motif (consensus sequence: GLRGAGK(T/S)) is highly conserved in all BzdR orthologues (Fig. 2), and, in fact, the suggested three-dimensional structure of C-BzdR reveals that the environment around the P-loop is also essentially conserved in all of these proteins. Moreover, the strictly conserved Gly present in the Walker B-motif of purine nucleotidebinding proteins (56), usually located around the ␤ 3 strand (55), is also present in BzdR (Gly-209) and its orthologues (Fig. 2).
The three-dimensional model of C-BzdR permits us to envisage a putative three-dimensional model for the complex between C-BzdR and the effector molecule benzoyl-CoA. The feasibility of this model is accentuated by the striking structural similarity between both ends of the benzoyl-CoA molecule (namely the ADP and the benzoyl moieties) and the two substrates of shikimate kinases (ATP and shikimate, respectively). Manual docking of the ADP moiety of benzoyl-CoA into the C-BzdR structure is immediate, since the C-BzdR fold has a conserved nucleotide-binding site (Fig. 10C). On the other hand, the pantothenate and ␤-mercapthoethylamine units fit a deep groove (Fig. 10C) whose walls are formed by the loops between residues 163-158 and 231-271 (52,55). The groove ends in a cavity, equivalent to the shikimate-binding site described for shikimate kinases (52,55), that perfectly hosts the benzoyl moiety of the effector molecule (Fig. 10D). It has been proposed that the enzymatic activity of shikimate kinases proceeds through an induced fit mechanism, since the loops that constitute the walls of the groove suffer important conformational changes upon substrate binding (52). In this sense, binding of benzoyl-CoA to BzdR might involve a similar conformational change at C-BzdR that would then trigger structural modifications at the DNA-binding region. The crystallization of BzdR and the BzdR-benzoyl-CoA complex is currently under way, since the resolution of their three-dimensional structures will undoubtedly permit the elucidation of the structural basis that determines the biological action of this prototype of a new subfamily of regulatory proteins.
Acknowledgments-The technical work of I. Alonso is greatly appreciated. We thank E. Aporta and J. Varela for help with oligonucleotide synthesis and N-terminal protein sequencing, respectively. The help of FIG. 10. Three-dimensional model of N-BzdR and C-BzdR regions and that of the C-BzdR-benzoyl-CoA complex. A, ribbon diagram of N-BzdR showing the five-helix bundle. The helices and loop forming the classical HTH motif are drawn in dark green. B, ribbon diagram of C-BzdR. The overall fold contains five ␤-strands (red arrows) and eight ␣-helices (blue ribbons). The phosphate-binding site (P-loop) is drawn in light green. The corresponding N and C termini are labeled as NH 2 and COOH, respectively. The figures have been prepared with MOLSCRIPT (48) and RASTER3D (49). C, adenine-, pantothenate-, and ␤-mercapthoethylamine-binding sites. D, benzoyl-binding site. The benzoyl-CoA molecule is represented with rods. The surface cavities have been drawn with GRASP (50). a In these cases, the E-score is indicated, since the -fold recognition step was not activated because BIOINBGU found good hits in the preliminary search with PSI-BLAST.
b In this case, E values are indicated (E values Ͻ 3e-5 are typical of close homologues (28).