INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

An early documentation of the physiological plasticity of bacteria was demonstration of specific patterns of induction in response to the chemically similar growth substrates p-hydroxybenzoate and protocatechuate (44). More recent investigations have elucidated molecular mechanisms underlying the specific inducible response of Acinetobacter to these compounds. As summarized in Fig. 1A, the transcriptional activator PobR elicits transcription of pobA in response to p-hydroxybenzoate (8). Action of PobA upon this compound produces protocatechuate, which triggers the action of PcaU, the transcriptional activator of the complete set of genes required for catabolism of protocatechuate (17). In light of the tight specificity of their controls, it is remarkable that PobR and PcaU are similar in many respects. The amino acid sequences of the proteins are 54% identical (17), and as shown in Fig. 1B, the operators to which the proteins bind are markedly similar (17).

View larger version (35K): [in this window] [in a new window]   FIG. 1.   Similar mechanisms govern transcriptional activation of p-hydroxybenzoate and protocatechuate catabolism in Acinetobacter. (A) Rectangles represent enzymes, and ovals represent transcriptional activators. Proteins are shaded, and genes are not. Genes and proteins associated with PobR are circumscribed by solid lines, and those associated with PcaU are surrounded by dashed lines. The curved arrows at the top represent the enzymatic conversions of p-hydroxybenzoate to protocatechuate and of protocatechuate to carboxymuconate. Carboxymuconate is a toxic metabolite, and strains blocked in its metabolism can be used to select secondary mutations blocking catabolism of either p-hydroxybenzoate (8, 15, 23) or protocatechuate (16). As indicated by the curved dashed arrows, p-hydroxybenzoate and protocatechuate act upon the respective activators PobR and PcaU to exert specific control over gene transcription (9, 17). PobA, the enzyme that acts upon p-hydroxybenzoate, is induced in response to interaction of the compound to PobR bound to an operator upstream from pobA (9). By an analogous mechanism, protocatechuate triggers expression of genes encoding its catabolism by interaction with PcaU bound to an operator upstream from the pcaIJFBDKCHG operon, which encodes genes for protocatechuate catabolism (17). Dashed horizontal arrows indicate directions of transcription. Curved solid arrows pointing from gene to protein indicate formation of the translated gene product. (B) Similar nucleotide sequences in the PobR and PcaU operators (17). Horizontal arrows indicate inverted repetitions in nucleotide sequences, and shaded boxes mark nucleotide residues that are identical in the aligned operator sequences.

Investigation of pobR and pcaU has been facilitated by the remarkable competence of Acinetobacter strain ADP1 for natural transformation (22), and this trait facilitated isolation of 89 mutants in which pobR function had been impaired by nucleotide substitutions caused by errors during PCR amplification of the gene (23). The mutations were distributed widely throughout pobR, and the mutant strains exhibited a range of phenotypes: some were null, some were leaky, some were heat sensitive, and others were cold sensitive. In this report, we describe how some of these mutations influence the ability of PobR to bind to its DNA target, the PobR operator. As might be expected, these mutations are clustered in and around nucleotide sequences encoding an apparent helix-turn-helix region, a DNA-binding motif found in many regulatory proteins.

An alternative approach to understanding regulatory proteins is investigation of amino acid substitutions that allow gain rather than loss of function. This was made possible by selection of mutant strains in which PCR mutagenesis of pobR produced a protein altered so that it could activate pca gene expression. As indicated in Fig. 2A, proteins emerging from such a selection required p-hydroxybenzoate for their activity. This requirement was relieved by a limited subset of PCR-generated mutations within the central region of pobR. It also proved possible to select strains in which PCR-generated nucleotide substitutions had modified pcaU so that it formed a protein that mimicked PobR function (Fig. 2B). As described here, the selection invariably produced strains containing a single amino acid substitution in the C-terminal region of PcaU, remote from the helix-turn-helix motif segment that would be predicted to be most directly associated with DNA binding.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Mutations altering the specificity of transcriptional activation by PobR and PcaU. (A) A knockout mutation in pcaU impairs expression of the pca operon including PcaHG. PCR mutagenesis can alter PobR so that it assumes the function of PcaU. The altered PobR still requires p-hydroxybenzoate for activity because this compound is required for growth of the cells with either quinate or protocatechuate. Secondary mutations relieve PobR of the demand for p-hydroxybenzoate and allow the activator to elicit pca gene expression. (B) PCR mutagenesis of pcaU creates mutations allowing PcaU to replace the function of an inactivated PobR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strain and culture conditions. Mineral medium (23) supplemented with 10 mM succinate was routinely used for growth of Acinetobacter strains in tubes on a gyratory shaker, or on plates (solidified with 1.8% [wt/vol] agar), at 37 or 22°C. Mutations causing defects in the naturally transformable Acinetobacter strain ADP1 were prepared and sequenced in an earlier investigation. Where indicated, p-hydroxybenzoate (5 mM), quinate (5 mM), or protocatechuate (3 mM) was used as the carbon and energy source. Luria-Bertani (LB) broth was used for growth of Escherichia coli strains. Ampicillin was added to a concentration of 80 µg/ml for selection of resistance in E. coli; kanamycin was used at a concentration of 50 µg/ml for both E. coli and Acinetobacter.

Recombinant DNA techniques. All recombinant DNA techniques were performed as described previously (24, 25) and according to Sambrook et al. (40). Isolation of bacterial chromosomal DNA, to be used as template DNA in PCRs with Taq polymerase, was done with Instagene DNA purification matrix (Bio-Rad) as recommended by the supplier. A scaled-down version of the protocol described before (16) was used for isolation of chromosomal template DNA to be used with Pfu polymerase. Restriction enzymes were obtained from New England Biolabs, Inc. PCR primers were custom synthesized (Keck Biotechnology Resource Laboratory, Yale University).

Transformation-facilitated mutagenesis. PCR for transformation-facilitated mutagenesis was performed as described above except that the number of cycles was increased to 35. Mutagenesis of pobR was performed as in a previous study (23) with primers pob1 (5'-GCAGTTGACCGAGTAGTAATCCCG-3') and pob2 (5'-GAAAACTGTCCACTCCGATTCC-3'), which generated a 1,434-bp product. For mutagenesis of pcaU, primers pcaU1 (5'GATAACTCCAATGTGCATCTAGC-3') and pcaU4 (5'-GATGAATCAGATCGATATGGCAA-3') were used to generate a 1,460-bp product. Strains of Acinetobacter were transformed with PCR DNA as described before (23) or with slight modification. Routinely, 10 µl of an unpurified PCR product was added to 500 µl of an early-exponential-phase culture. After growth for an additional 3 to 16 h, the transformation mixture was plated directly onto selective medium.

Sequence analysis of mutations. Sequence analysis was performed as described before (23), with Taq-amplified chromosomal DNA as the template in cycle sequence reactions, using an ABI PRISM dye terminator cycle sequencing kit with Amplitaq DNA polymerase (-FS) as recommended by the supplier (Perkin-Elmer). DNA fragments were denatured at 95°C for 2 min prior to electrophoresis on a denaturing 6% polyacrylamide gel in an ABI 373 automated sequencer (Perkin-Elmer ABI), linked to an Apple PowerMac, equipped with appropriate sequencing software (Perkin-Elmer ABI). Sequences were analyzed using the program DNASTAR (Lasergene).

Overproduction of wild-type and mutant PobR in E. coli. Expression vector pBCP367 (47) was used for generation of PobR overexpression plasmids. This vector contains an NdeI site (CATATG) that allows cloning of a gene with its own ATG translation initiation codon inserted directly downstream of both the isopropyl--D-thiogalactopyranoside (IPTG)-inducible trc promoter and a plasmid-derived ribosome-binding site. The availability of a unique BamHI site downstream of NdeI permitted forced cloning of PCR-amplified pobR alleles for overproduction of PobR in E. coli (Fig. 3A). Wild-type and mutant alleles of Acinetobacter pobR were amplified by PCR with the high-fidelity polymerase Pfu (Stratagene), using chromosomal DNA as the template and with primers pobR-Nde and pobR-Bam (Fig. 3A). Primer pobR-Nde (5'-GGATTTGAATCATATGGAACAACATCACCAATAC-3') overlaps with the first seven codons of the 816-bp pobR open reading frame (in italics) and carries an engineered (two nucleotide substitutions; doubly underlined) NdeI restriction site (underlined). The second primer, pobR-Bam (5'-TCTAGGATCCAAATTATACCAAATTACGCAG-3'), carries an engineered (two nucleotide substitutions) BamHI recognition site and anneals at the end of pobR (in italics). The 837-bp fragment formed after PCR amplification was separated from Pfu polymerase, unincorporated primers, and dNTPs and digested with NdeI and BamHI; the resulting fragments were gel purified and ligated into NdeI/BamHI-digested pBCP367. Transformants of E. coli DH5 (19) were selected for vector-encoded ampicillin resistance, and the nucleotide sequence of each cloned pobR gene was verified. Plasmid pZR85 contains the cloned wild-type pobR, and plasmids pZR110 through pZR130 carry mutant pobR alleles.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   (A) Construction of PobR overproduction plasmids. (B) Generation of PCR fragments, including the PobR operator in the pobA-pobR intergenic region, for use as DNA-binding probes in BIAcore experiments. The biotin label coupled at the 5' end to primer pobR-O2bio is depicted as an asterisk.

Surface plasmon resonance detection of PobR operator binding. Wild-type Acinetobacter PobR binds to the intergenic region between pobR and the divergently transcribed pobA in a 35-bp region containing inverted sequence repetitions (9). This region was independently amplified by PCR, using Taq polymerase and chromosomal DNA of Acinetobacter as the template, with two sets of primers. (i) A 114-bp DNA fragment was generated with a standard primer annealing in the pobA-pobR intergenic region (pobR-O1 [5'-AAAGACATCTATAATCAAGGCTC-3']) and a 5'-biotinylated primer that anneals near the start of pobR (pobR-O2bio [5'-biotin-GCAAGGTATTGGTGATGTTGTTC-3']) (Fig. 3B). The resulting biotinylated PCR fragment was used for immobilization of the operator onto a sensor chip (see below), to be used as the probe in determining DNA-binding characteristics of PobR proteins. (ii) A second operator fragment was generated via PCR amplification with primers pobR-O1 and pobR-04 (5'-CTTCACTTGAATGGGGATGTGC-3'). The resulting 134-bp fragment (Fig. 3B) is not labeled with biotin and was used as free operator DNA in competition experiments to determine the specificity of the association signal. The PCR fragments were purified on a 12% polyacrylamide gel, eluted with a buffer containing 10 mM Tris HCl (pH 7.5), 0.5 mM EDTA, and 200 mM NaCl, after precipitated, and then taken up in 10 mM Tris HCl (pH 8.0)-1 mM EDTA-300 mM NaCl. The concentration of the PCR-amplified operator DNA fragments in these solutions was estimated by relative UV-induced ethidium bromide fluorescence of samples run on a 12% acrylamide gel. A BIAcore (Pharmacia Biosensor) apparatus was used as recommended by the supplier for semiquantitative surface plasmon resonance detection of binding of wild-type PobR and mutant PobR proteins in cell extracts of E. coli (see above for overproduction of PobR in E. coli) to the 114-bp operator fragment immobilized on a dextran-streptavidin-coated sensor chip (sensor chip SA; Pharmacia Biosensor). Routinely, between 40 and 80 ng of biotinylated operator DNA (1 to 2 pmol) was injected (in HBS with 0.5 M NaCl) at a flow rate of 5 µl/min (for 9 min) into the flow cell of a sensor chip for binding of the operator DNA to the chip. This resulted in a net increase in relative response units of between 1,100 and 1,700.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Distribution of PCR-generated mutations in Acinetobacter pobR. In a previous study (23), the transcriptional regulator gene pobR was used to illustrate the advantage of combining PCR mutagenesis with natural transformation to obtain Acinetobacter mutants impaired in growth with p-hydroxybenzoate as the sole carbon and energy source. In this investigation, 57 such regulatory mutations were analyzed in detail by phenotypic characterization and by sequencing the corresponding mutant pobR allele. Most of the mutations caused single amino acid substitutions in PobR; three mutations altered the length of the PobR C terminus (Table 1). The mutations impeded growth with p-hydroxybenzoate in one of four ways: 23 of the mutants exhibited a null phenotype, 9 were heat sensitive, 16 were cold sensitive, and 9 were leaky (Table 1).

View larger version (40K): [in this window] [in a new window]   FIG. 4.   Alignment of the amino acid sequence of Acinetobacter PobR (Acin PobR) with those of Acinetobacter PcaU (Acin PcaU) and P. putida PcaR (Pseu PcaR). Only identical residues (as asterisks) and similar residues (dots) are represented. A shaded rectangle encompasses the potential DNA-binding helix-turn-helix motif (8). The effects in PobR of the nucleotide substitutions in the pobR mutants listed in Tables 1 and Table 3 are shown; the numbers in parentheses correspond to the pobR alleles (Tables 1 and 3). The phenotypes of the strains containing these mutations, with respect to growth on 4-hydroxybenzoate as the sole carbon source, are indicated as follows: null (no growth at all; not boxed), heat sensitive (little or no growth at 37°C; boxed), cold sensitive (little or no growth at 22°C; black-shaded box), and leaky (slow growth at both 22 and 37°C; boxed with a dashed line). Downward-pointing grey arrows mark the mutant PobR proteins that have been overproduced in E. coli (Table 2; Fig. 5) and tested with respect to operator binding. Six consecutive arrows represent the six independent mutations giving rise to pobR1350 (boxed with dark outline). The six different mutations leading to relaxation of inducer specificity are marked by dashed arrows, and the boxes surrounding the mutations are dashed.

Identification of PobR residues involved in DNA binding. Acinetobacter PobR binds in the 134-bp pobA-pobR intergenic region to a 35-bp interval containing inverted DNA repeats (9) (Fig. 1). PobR negatively regulates its own expression and activates transcription of pobA, the structural gene for p-hydroxybenzoate hydroxylase (8, 9). In this study, PobR amino acid residues involved in DNA binding were identified by qualitative surface plasmon resonance detection in a BIAcore apparatus. Ten of the mutant pobR alleles generated by PCR mutagenesis (23) were cloned into expression vector pBCP367 (47) (Fig. 3), and the encoded PobR mutant proteins were overproduced in E. coli (Table 2). In cell extracts of these E. coli hosts, PobR appeared as a major protein band of the expected size (29 kDa) on Coomassie blue-stained SDS-polyacrylamide gels (Fig. 5), indicating its presence as soluble protein. The mutant alleles were chosen to include a range of phenotypes. The 10 mutants were also chosen so as to sample a range of locations in the pobR gene; included were two strains with mutations in the N-terminal helix-turn-helix motif and three strains with mutations in the conserved region centered on A158 (Fig. 4). Originally, an additional eight pobR alleles (pobR1410, pobR1411, pobR1413, pobR1414, pobR1421, pobR1423, pobR1427, and pobR1428 [Table 1; Fig. 4]) were cloned in pBCP367. Unexpectedly, however, cell extracts of the eight respective E. coli strains lacked detectable soluble PobR protein. Since these mutant PobR proteins were detectable in whole-cell extracts (on SDS-PAGE) and were produced in amounts comparable to the first 10 mutant proteins, their absence in cell extracts suggests that the encoded proteins form inclusion bodies in E. coli. This is striking since they all differ from the wild-type protein by just a single amino acid. Possibly, the latter eight amino acid substitutions individually lead to a destabilization of the PobR secondary structure, resulting in aggregation of the mutant proteins in the E. coli cytosol.

View larger version (58K): [in this window] [in a new window]   FIG. 5.   Coomassie blue-stained SDS-polyacrylamide gel of cell extract of E. coli DH5 strains carrying plasmid pBCP367 (vector; control without pobR), pZR85 (overexpressing wild-type pobR), or plasmids pzR112 through -129 (overexpressing mutant pobR genes). Overproduction of PobR was induced for 1 h with IPTG (see Materials and Methods), and cell extract corresponding to 15 µg of total protein was loaded for each sample. The band corresponding to PobR is indicated. MW, molecular weight markers (Pharmacia).

View larger version (28K): [in this window] [in a new window]   FIG. 6.   BIAcore sensorgram (relative response versus time) traces derived from the PobR-operator binding experiments. The response is given in relative response units (RU). Black arrows mark the time of injection of the cell extract into the flowcell (onset of the association phase), grey arrows indicate the time of replacement of the cell extract by HBS (onset of the dissociation phase). The reference trace shows the response observed after injection of cell extract of the E. coli strain with the vector (pBCP367) alone (no PobR protein [Fig. 5]). PobR numerals next to the traces correspond to the mutant allele designations given in Table 2. (A) Competition experiment showing how the addition of free specific competitor DNA (operator) together with wild-type PobR protein (PobR-wt) affects binding of PobR to the immobilized operator on the chip surface. The amount of injected PobR protein and free operator DNA is next to each trace. (B) Response observed after injection of cell extract of E. coli strains with overproduced PobR proteins that appear to exhibit wild-type binding properties. (C) Responses observed after injection of cell extract of E. coli strains with overproduced PobR proteins that exhibit abnormal binding properties.

Importance of the PobR C-terminal region. Amino acid conservation between PobR and its homologs in the -ketoadipate pathway extends almost to the last residue (Fig. 4). Two mutations that result in 15 amino acids lost (PobR1410) or 2 amino acids gained (PobR1433) at the C terminus confer, respectively, null and cold-sensitive phenotypes (Fig. 4). The frameshift mutation altering the C-terminal 38 residues (PobR1440) abolishes growth with p-hydroxybenzoate (Fig. 4; Table 1). Implying that the conserved PobR C-terminal region may (directly or indirectly) play a role in DNA binding, the null mutation in PobR1424 (M241K [Fig. 4]) produces a protein that appears to bind relatively tightly to its operator in vitro (Fig. 6B).

Importance of the PobR N-terminal helix-turn-helix region. A potential helix-turn-helix motif consisting of amino acid residues 43 to 64 was identified near the N terminus of PobR (8) (Fig. 4). In the second helix of the PobR helix-turn-helix, there are three arginyl residues conserved with its two closest homologs, PcaU and PcaR, that appear to be essential for PobR function (Fig. 4). The PCR-generated PobR1442 (R60Q [Fig. 4]) and the previously identified (8) spontaneous mutation R61H both completely prevent growth with p-hydroxybenzoate. The PCR-generated mutation R56S, although it produces a leaky phenotype (PobR1412 [Fig. 4]), abolishes DNA binding in vitro (Fig. 6C). The helix-turn-helix motif also includes the other PobR mutation assayed in this study that eliminates DNA binding, K64M (PobR1429), and is flanked by the three remaining mutations that severely affect binding, most significantly W80R in the tightly binding PobR1425 (Fig. 4; Table 2). Immediately adjacent to the helix-turn-helix region, PobR R40 is intolerant of amino acid substitutions (23), consistent with its conservation in the PobR family (Fig. 4).

The amino acid substitution T57A in the PobR helix-turn-helix appears to broaden operator-binding specificity. Acinetobacter PobR and PcaU are homologous transcriptional regulators that govern sequential steps in the -ketoadipate pathway (8, 17) (Fig. 1). The two proteins are 54% identical at the amino acid level (Fig. 4), recognize similar effector molecules (Fig. 1), and bind to operator sites containing nearly identical inverted DNA repeats (9, 17) (Fig. 1). A potential helix-turn-helix motif can be identified near the N terminus in both proteins (9, 17). Despite these similarities, the functions of pobR and pcaU do not appear to overlap. Knockout mutations in one gene are not complemented by the wild-type copy of the other gene. The similarity of PobR and PcaU tempted us to use PCR mutagenesis to isolate gain-of-function mutations that would confer on one regulatory protein some of the activity of its homolog (Fig. 2). We anticipated that such mutations could highlight the specificity determinants in each protein.

Mutations in PobR amino acid residues 118 to 135 define a domain where single substitutions relax dependence of PobR activity on p-hydroxybenzoate. To take advantage of the selective growth conditions still available with the PobR1350 mutant, a further round of PCR mutagenesis was performed. Strain ADP1349 was again transformed, this time with PCR-amplified DNA containing the pobR1350 mutation, and recombinants were selected on plates containing quinate as the sole carbon source. Quinate, a metabolite upstream of protocatechuate in the -ketoadipate pathway (10, 11) (Fig. 2), was used rather than protocatechuate because of the latter's relative toxicity and instability. Recombinants which could grow with quinate or protocatechuate were readily obtained, and pobR from 10 independently generated mutant strains was sequenced.

A single amino acid substitution in PcaU (G222V) confers PobR activity. To complete the symmetric strategy depicted in Fig. 2, we used transformation-facilitated PCR mutagenesis to obtain gain-of-function mutations allowing PcaU to replace PobR. The choice of starting strain ADP607 (10) for this selection was straightforward: the deletion removing the 59 C-terminal amino acids of PobR in this strain completely abolishes growth with p-hydroxybenzoate (9). Transformation of ADP607 with PCR-amplified pcaU DNA readily produced recombinants that grew with p-hydroxybenzoate. Sequencing of pcaU from six independently generated mutant strains revealed the same mutation in each case (pcaU1300 [Table 4]). This mutation caused an amino acid substitution, G222V, located not in the helix-turn-helix domain of PcaU but near the C terminus in a residue equivalent in position to PobR G219 (Fig. 4).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Selection in the laboratory and selection during evolutionary history. Nucleotide substitutions introduce variations in protein structure that are played upon by selection. Throughout evolutionary history, those variations that impair protein function are likely to have been lost. Therefore amino acid residues conserved within the aligned primary sequences of divergent proteins may be regarded as essential or at least valuable contributors to function. In this investigation, direct selection for loss of PobR function was imposed, and it might have been anticipated that most PCR-generated mutations would alter amino acids that had been conserved during the evolutionary divergence of PobR and PcaU. As shown in Fig. 7A, this prediction is not particularly accurate. About half (29 of 54) of the amino acid substitutions selected on the basis of loss of PobR function occurred at positions where the PobR and PcaU sequences resisted change during their evolutionary divergence. The other mutations leading to loss of function are located at positions where change, presumably associated with either maintenance or acquisition of function, was accepted during evolution of the proteins.

View larger version (28K): [in this window] [in a new window]   FIG. 7.   Locations of amino acid substitutions within the primary sequences of PobR and PcaU. The height of each rectangle indicates the frequency with which a selected mutation was observed at that position. (A) Selection for loss of function after PCR mutagenesis compared with selection for retention of function during evolution. Null mutations are indicated by dark rectangles, and mutations allowing some residual activity are indicated by shaded rectangles. Vertical lines connect the sites of PCR-generated mutations with loci that accepted change during evolutionary divergence of PobR and PcaU. (B) Selection for PcaU transcriptional activator function yielded the same mutation six times after PCR mutagenesis of pobR. (C) Demand for PcaU function in the absence of p-hydroxybenzoate produced PCR-generated mutants carrying amino acid substitutions at four different loci in the middle of the PobR primary sequence. (D) PCR mutagenesis of pcaU followed by selection for PobR function invariably yielded strains carrying the same amino acid substitution in the carboxy-terminal region of PcaU.

Linear organization of PobR functional domains. Based on amino acid sequence alignment, PobR is a member of a small family of bacterial transcriptional regulators, all of which contain an N-terminal helix-turn-helix motif (8, 17, 21, 39, 43, 45). The distribution of mutations affecting DNA binding detected in this study emphasizes the functional importance of the helix-turn-helix region and suggests that the first 100 of the 271 amino acid residues of PobR constitute the DNA-binding domain (Fig. 4 and 8). The PobR C-terminal region, particularly in light of comparisons to members of the LysR family discussed below, may function indirectly in DNA binding by mediating PobR multerimization. Mutations that do not significantly affect DNA binding are found toward the middle of the PobR primary sequence, both in the conserved region centered on A158 (Fig. 4 and 8) and in the region where mutation can alleviate dependence of PobR activity on the coinducer p-hydroxybenzoate (Fig. 4 and 8). Further work is needed to more precisely define the coinducer response domain and to characterize the mutations that alter coinducer response, either directly, by allowing a new effector (most likely protocatechuate) to bind, or indirectly, by producing a constitutively active protein.

View larger version (20K): [in this window] [in a new window]   FIG. 8.   Properties of mutations within the linear amino acid sequence of PobR. Mutations that have been tested with respect to their effect on DNA binding are indicated along with those that convert PobR into the regulator of both pobA and the pca operon. The N-terminal helix-turn-helix motif is shaded grey. Numerals in parentheses correspond to the PobR alleles (Fig. 4). The amino acid substitutions L98F and M241K have relatively subtle effects on DNA binding in vitro (Fig. 6; Table 2).

Comparison of the organization of functional domains in PobR with that in LysR family regulators. It has been noted (9) that the proposed model for regulation by PobR has characteristics typical of the large LysR family of bacterial transcriptional regulators (20, 41). Not unexpectedly then, the linear domain organization of PobR presented here is also similar to that found in the LysR family (41), particularly the well-characterized member NahR (5, 42), regulating two operons for the catabolism of the aromatic compound naphthalene and its metabolic intermediate salicylate. NahR mutations eliminating DNA binding cluster in and around the N-terminal helix-turn-helix motif, but as in PobR, the C terminus of the protein is also essential: a nonsense mutation resulting in loss of nine amino acid residues from the NahR C terminus abolishes DNA binding (42). Most nearby amino acid substitutions that eliminate DNA binding generate NahR proteins that are not trans dominant, suggesting that the C terminus may be indirectly involved in DNA binding by mediating multerimization (41, 42).

Distribution in the -ketoadipate pathway of regulators from the PobR and LysR families. In Acinetobacter strain ADP1, the two branches of the -ketoadipate pathway are encoded by genes in two supraoperonic clusters, separated on the chromosome by approximately 280 kb (18) and each governed by regulatory proteins from distinct families: conversion of p-hydroxybenzoate to citric acid cycle intermediates is governed by the PobR family regulators PobR (8) and PcaU (17), whereas the equivalent enzymatic steps in the benzoate branch of the pathway are regulated by LysR family members BenM (6) and CatM (38). This same branch specificity has been noted (39) for the P. putida regulators PcaR and CatR. Given the similarity of PobR to LysR family members discussed above, this strict branch specificity in Acinetobacter may reflect more than historical accident.

Implications of PcaU and PobR gain-of-function mutations. A single amino acid substitution in PcaU allows it to functionally replace PobR and activate PobA expression. This mutation near the PcaU C terminus is far from the domain containing the helix-turn-helix that is expected to mediate DNA sequence-specific binding, implying that wild-type PcaU can already bind the PobR operator, albeit not productively. Supporting this possibility is the fact that substitution at one position of the PobR helix-turn-helix with the equivalent PcaU residue appears to allow the PobR protein to bind both the PobR and PcaU operators.