FIG. 1 The protocatechuate and catechol branches of the 3-oxoadipate pathway. Designations as gene products are given in parentheses. For the reactions common to both branches, some bacteria possess only a single set of genes, while others harbor separate pca (more ...)

Rhodococcus opacus (erythropolis) 1CP has previously been isolated by virtue of its ability to grow with 4-chlorophenol and 2,4-dichlorophenol (23). The characterization of its chlorocatechol catabolic enzymes revealed unusual properties (37, 38, 64). Interestingly, the muconate and the chloromuconate cycloisomerase of strain 1CP in their product formation from 2-chloro-cis,cis-muconate and in their inability to convert (+)-2-chloromuconolactone were more similar to each other than to muconate and chloromuconate cycloisomerases of proteobacteria (64). If this phenotypic similarity of the Rhodococcus enzymes indicates a higher degree of relatedness compared to proteobacterial enzymes, then the improved turnover of 3-chloro- and 2,4-dichloro-cis,cis-muconate by the chloromuconate cycloisomerases should have evolved independently among gram-positive bacteria and proteobacteria by functionally convergent evolution. Despite the structural similarity of 3-oxoadipate enol-lactone and dienelactones (4-carboxymethylenebut-2-en-4-olides), the enol-lactone hydrolase was apparently not adapted to convert dienelactones during evolution of the proteobacterial chlorocatechol pathway (60). Rather, some other lactone-hydrolyzing enzyme was recruited from an unknown pathway. An independent evolution of the chlorocatechol pathway among the gram-positive bacteria would imply that a different solution to the chemical problem of dienelactone hydrolysis may have been found. And especially since the dienelactone hydrolase of R. opacus 1CP has an unusual substrate specificity (38), we considered it appropriate to include the lactone hydrolases in sequence comparisons of enzymes for catechol and chlorocatechol catabolism. We have previously reported on the sequences of the catechol 1,2-dioxygenase, the muconate cycloisomerase, and the muconolactone isomerase (16). In the accompanying report (18), we characterize a gene cluster for chlorocatechol degradation of R. opacus 1CP. Here we describe the cloning and sequence analysis of a DNA fragment comprising the gene for the 3-oxoadipate enol-lactone-hydrolyzing enzyme of strain 1CP. We found the gene to be part of a protocatechuate catabolic gene cluster and to be merged to the pcaC gene, which in proteobacteria codes for a separate 4-carboxymuconolactone decarboxylase.

(Some of the results reported in this study have previously been mentioned in a preliminary communication [17].)

Strains, plasmids, and cultivation conditions. Strain 1CP was previously isolated with 2,4-dichlorophenol as the carbon source and, based on phenotypic properties, assigned to the species Rhodococcus erythropolis (23). Very recently, 16S rDNA and fatty acid analyses (68) called for a reclassification of the strain to the species R. opacus, which also belongs to the Rhodococcus rDNA group IV (55).

For the purpose of this study, R. opacus 1CP was cultivated in a 10-liter glass fermentor at 30°C in a mineral medium (13) with 50 mM phosphate buffer (pH 7) and with 25 mM sodium benzoate (fed in 5-mM portions) as the only carbon and energy source. To provide biomass for the purification of protocatechuate dioxygenase, sodium 4-hydroxybenzoate (5 portions of 5 mM) was used as the only carbon and energy source. Cells were harvested by centrifugation, frozen in liquid nitrogen, and stored at −20°C until used.

Escherichia coli DH5α (4), obtained from GIBCO BRL, was cultivated aerobically with constant shaking at 37°C in Luria-Bertani (LB) medium (58) supplied, if appropriate, with 100 μg of ampicillin per ml. For induction of E. coli DH5α harboring pARO523 (51), 5 ml of overnight precultures was used to inoculate 200 ml of LB medium (incubation at 37°C). After induction with 1 mM IPTG (isopropyl-β-d-thiogalactoside) at an optical density at 600 nm of 0.6, incubation was continued for 3 h at 30°C.

Properties of plasmids used in this study are summarized in Table 1.

TABLE 1 Plasmids used in thisstudy

Preparation of cell extracts. Extracts of R. opacus 1CP were prepared by resuspension in 50 mM Tris-HCl (pH 7.5), disintegration by grinding with glass beads, and clarification of the preparation as described previously (16).

E. coli cells were harvested by centrifugation, washed with 0.25 volume 50 mM Tris-HCl (pH 7), and resuspended in 0.01 volume of the same buffer. The cells were disintegrated by passage through a precooled Aminco French pressure cell (115 MPa), and cell debris was removed by centrifugation as described above.

Enzyme assays and estimation of protein concentration. The activity of 3-oxoadipate enol-lactone hydrolase was measured spectrophotometrically at 230 nm as described previously (61). The assay mixtures contained 0.67 mM racemic muconolactone (prepared by the method of Elvidge et al. [15]) plus partially purified preparations of muconolactone isomerase (47, 48). The activity of 4-carboxymuconolactone decarboxylase was also measured at 230 nm (47). The assay mixture (1 ml) initially contained 50 mM Tris-Cl (pH 7.5), 0.167 mM protocatechuate, and 6 μl of cell extract from IPTG-induced E. coli DH5α(pARO523) (corresponding to 0.21 mg of protein) with the protocatechuate 3,4-dioxygenase and the 3-carboxymuconate cycloisomerase from A. tumefaciens. The mixture was incubated for 40 min at room temperature to generate 4-carboxymuconolactone before addition of preparations of partially purified 3-oxoadipate enol-lactone hydrolase and of the fraction that was to be measured. The activities of protocatechuate 3,4-dioxygenase were measured at 290 nm as described previously (65). The following extinction coefficients were used: for the conversion of 3-oxoadipate enol-lactone to 3-oxoadipate, 1,430 M⁻¹ · cm⁻¹ (48); for the conversion of 4-carboxymuconolactone to 3-oxoadipate, 4,200 M⁻¹ · cm⁻¹ (47); and for the conversion of protocatechuate to 3-carboxy-cis,cis-muconate, 2,300 M⁻¹ · cm⁻¹ (65). Protein concentrations were measured by the method of Bradford (6), with bovine serum albumin as the standard.

Protein purifications. For the initial purification of 3-oxoadipate enol-lactone hydrolase (for sequencing purposes), extract from benzoate-grown cells (volume, 160 ml; protein, 370 mg; total activity, 157 U; specific activity, 0.42 U/mg) was chromatographed on a Pharmacia Q Sepharose High Performance column (HR 16/10; bed volume, 20 ml) with 50 mM Tris-HCl (pH 7.5)–0.1 mM dithiothreitol and, for elution, an NaCl gradient (0 to 0.15 M over 60 ml and 0.15 to 0.5 M over 350 ml). The most active fractions (eluting at ca. 0.4 M NaCl) were pooled (volume, 25 ml; protein, 17.9 mg; total activity, 135 U; specific activity, 7.5 U/mg). After addition of (NH₄)₂SO₄ to 1 M (final concentration), the preparation was filtered and subjected to Pharmacia Phenyl Superose HR 10/10 chromatography. Elution was achieved with a decreasing (NH₄)₂SO₄ gradient of 1 to 0.9 M over 5 ml, 0.9 to 0.4 M over 120 ml, and 0.4 to 0 M over 10 ml. The fractions with the highest activities, eluting at about 0.8 to 0.75 M (NH₄)₂SO₄ (volume, 4.5 ml; protein, 0.42 mg; total activity, 38 U; specific activity, 90 U/mg), were pooled and desalted by using Pharmacia PD-10 columns. The pool was then subjected to chromatography on Pharmacia MonoQ (HR 5/5) by using an NaCl gradient (0 to 0.2 M over 5 ml and 0.2 to 0.5 M over 70 ml) for elution, which resulted in a preparation (volume, 2 ml; protein, 0.14 mg; total activity, 21.9 U) with a specific activity of 153 U/mg, equivalent to a 364-fold purification. The details for a second enrichment of 3-oxoadipate enol-lactone hydrolase, performed to track the presumed 4-carboxymuconolactone decarboxylase activity of the enzyme, are given in Table 2.

TABLE 2 Copurification of the 3-oxoadipate enol-lactone-hydrolyzing and 4-carboxymuconolactone-decarboxylating activities of R. opacus1CP

Protocatechuate 3,4-dioxygenase was purified by subjecting an extract from 4-hydroxybenzoate-grown cells (volume, 120 ml; protein, 492 mg; total activity, 153 U, specific activity, 0.32 U/mg) to Q Sepharose High Performance (HR 16/10) chromatography with 50 mM Tris-HCl (pH 7.6) and an increasing NaCl gradient (0 to 1 M over 400 ml) for elution. Dioxygenase-containing fractions (eluting at ca. 0.35 M NaCl) were pooled (volume, 10 ml; protein, 29.5 mg; total activity, 77 U; specific activity, 2.6 U/mg), mixed with (NH₄)₂SO₄ to 1.5 M (final concentration), and centrifuged. The preparation was then chromatographed on a Phenyl Superose HR 10/10 column with a gradient decreasing from 1.5 to 0 M (NH₄)₂SO₄ in buffer over 120 ml [activity peak at ca. 0.56 M (NH₄)₂SO₄]. The combined fractions (volume, 3 ml; protein, 0.6 mg; total activity, 29.5 U; specific activity, 49 U/mg) were then subjected to gel filtration (Pharmacia Superdex 200 prep-grade HiLoad 16/60; bed length, 60 ml). Elution with 50 mM Tris-HCl (pH 7.6)–0.1 M NaCl resulted in a preparation (volume, 1 ml; protein, 62 μg; total activity, 3.3 U) with a specific activity of 53 U/mg, equivalent to a 166-fold purification.

SDS-polyacrylamide gel electrophoresis and Western blotting. For purity checks of enzyme preparations, discontinuous sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was carried out and gels were stained as described previously (16). For N-terminal sequencing of the protocatechuate 3,4-dioxygenase subunits, ca. 7 μg of the purified enzyme was loaded onto an SDS-polyacrylamide gel for separation of the α and β subunits. Electrotransfer of the subunits onto an Immobilon-P membrane (Millipore) was performed for 90 min at 80 V and 4°C in a Bio-Rad Trans-Blot cell as described by Moos (43). After staining with Coomassie brilliant blue R-250 (42), the protein bands were cut out, rinsed, air dried, and subjected to N-terminal sequencing.

Protein cleavage, isolation of peptides, and sequencing of peptides and N termini. Trypsin digestion of 3-oxoadipate enol-lactone hydrolase and subsequent separation of tryptic peptides by reversed-phase high-performance liquid chromatography (HPLC) were performed by the procedure of Stone et al. (67) as described previously (16). The amount of enol-lactone hydrolase after trichloroacetate precipitation, washing, and drying was 40 μg.

Purified 3-oxoadipate enol-lactone hydrolase prior to sequencing was partially desalted by changing the buffer to 5 mM sodium phosphate buffer (pH 7) by ultrafiltration and repeated dilution, while selected peptide-containing fractions from reversed-phase HPLC as well as the blotted subunits of the protocatechuate 3,4-dioxygenase were sequenced directly in automated sequencers (Applied Biosystems model 473A or model 491, respectively).

Purification and general in vitro manipulations of DNA. Genomic DNA of R. opacus 1CP was obtained as described previously (16). Plasmid DNA from E. coli DH5α was isolated with Pharmacia Flexiprep kits. Vector DNA digested with only one enzyme was dephosphorylated prior to ligation. Insert DNA for ligation or digoxigenin labeling was isolated from gels by use of a Bio 101 Gene Clean II or a Biozym Easy Pure kit. Transformation of E. coli DH5α was achieved by the method of Chung et al. (9) or Inoue et al. (33). Loss of LacZ complementation by insertion of DNA into vectors was detected by using LB plates with 0.13 mM IPTG and 32 μg of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactoside) per ml.

Amplification of segments of the 3-oxoadipate enol-lactone hydrolase gene. Oligonucleotides were custom synthesized according to the sequences of the N terminus and of two internal peptides (Table 3). The codon usage was assumed to correspond to that of catA, catB, and catC of strain 1CP (16). PCR mixtures (50 μl) for the amplification of genomic DNA contained 50 pmol of each primer, 0.1 to 0.2 μg of genomic template DNA, 0.1 mM each deoxynucleotide triphosphate, Goldstar polymerase reaction buffer [75 mM Tris-HCl (pH 9.0), 20 mM (NH₄)₂SO₄, 0.01% (wt/vol) Tween 20], 1.5 mM MgCl₂, and 0.5 U of Eurogentech Goldstar DNA polymerase. As a denaturing agent (70), formamide was used in concentrations between 5 and 7% (vol/vol). The PCR was performed with a touchdown thermocycle program (12): initial denaturation (94°C; 3 min) before addition of the polymerase; 6 cycles with decreasing annealing temperature (66 to 56°C; 1 min), polymerization (72°C; initially 1 min, plus 1 additional s for each cycle), and denaturation (94°C; 30 s); 28 more cycles with 54°C as the annealing temperature; an additional 14 min of polymerization during the last cycle. Of the two products initially obtained, the larger one (ca. 440 bp as determined by agarose gel electrophoresis) required reamplification by the same program to increase the amounts of DNA for cloning.

TABLE 3 Sequences of N termini, tryptic peptides, and deducedprimers

Cloning strategy and hybridization procedures. The PCR products were ligated into a T-tailed vector (39) prepared from pBluescript II KS (+), yielding pRERE1 and pRERE2, respectively. The ca. 170-bp insert of pRERE1 was digoxigenin labeled by using a Boehringer Mannheim DIG DNA Labeling and Detection Kit Nonradioactive. A Southern blot of 3 μg of EcoRI-digested R. opacus 1CP DNA (run on a 1% agarose gel with TAE buffer [58]) was then hybridized with the digoxigenin-labeled probe to detect the corresponding fragment (procedure as described in the Boehringer manual). An area that corresponded in size to the hybridization signal (about 2.5 to 4 kb) was excised from a second gel, and DNA eluted from the gel slice was ligated into pBluescript II KS (+). This ligation mixture was used to transform E. coli DH5α. By colony hybridization with the labeled insert of pRERE1 (performed in a manner analogous to the procedure described in reference 16), a clone carrying pRER3, which contains a ca. 3.6-kb insert of R. opacus DNA, was identified. After sequencing had shown that the gene coding for 3-oxoadipate enol-lactone hydrolase was not complete on the insert, a clone with an overlapping, ca. 2.5-kb PstI insert (pRER4) was isolated in an analogous way by using a digoxigenin-labeled, ca. 650-bp BamHI fragment of pRER3 as a probe.

DNA sequence analysis. The nucleotide sequence of the overlapping inserts of pRER3 and pRER4 was determined by sequencing subclones in pBluescript II KS (+), by sequencing of nested deletions generated with a double-stranded nested deletion kit (Pharmacia), and by sequencing with custom-synthesized oligonucleotides as primers. For sequencing reactions, an Applied Biosystems Prizm kit was used, with subsequent electrophoresis and analysis in an Applied Biosystems A373 sequencer. Computer-based sequence analyses were performed mainly with the PC/GENE program package (Intelligenetics Inc., Mountain View, Calif.). The PC/GENE version of FASTA (53) as well as the BLAST programs (2, 22) were used to screen the EBI and NCBI databases. Multiple sequence alignments were performed by the PC/GENE version of CLUSTAL (28), with open and unit gap costs set to 10.

Nucleotide sequence accession number. The 5,475-bp sequence of the overlapping inserts of pRER3 and pRER4 shown in Fig. 4 will appear in the EMBL, GenBank, and DDBJ nucleotide sequence databases under accession no. AF003947.

FIG. 4 Sequence of the 5,475 bp carried on pRER3 and pRER4. The predicted amino acid sequences are shown below the DNA sequence. Amino acid sequences derived from direct sequencing of N termini or tryptic fragments are underlined. For pcal′, presumably (more ...)

Cloning of pca genes based on sequences from the purified 3-oxoadipate enol-lactone-hydrolyzing enzyme. As a basis for cloning of the corresponding gene, the 3-oxoadipate enol-lactone-hydrolyzing enzyme was purified to homogeneity from an extract of benzoate-grown cells of R. opacus 1CP. Surprisingly, the subunit molecular mass of the enzyme turned out to be ca. 42 kDa (Fig. 2), considerably greater than reported for the 3-oxoadipate enol-lactone hydrolases of P. putida and A. calcoaceticus. On SDS-gels, the latter enzymes appeared to have masses of between 30 and 33 kDa (40, 72); for the Acinetobacter catD and pcaD gene products, sizes of 29.3 and 29.1 kDa, respectively, were calculated (25, 63). Tryptic peptides were prepared from the Rhodococcus 3-oxoadipate enol-lactone hydrolase, and several of them as well as the N terminus were sequenced (Table 3). Degenerate oligonucleotides derived from these sequences allowed the amplification of ca. 170- and ca. 440-bp fragments with genomic DNA of strain 1CP as the template. After labeling with digoxigenin, the ca. 170-bp fragment first served to identify hybridizing bands on a Southern blot of EcoRI-digested Rhodococcus DNA and later to screen colonies of transformed E. coli DH5α for the presence of plasmids with the desired insert. A clone containing plasmid pRER3 with a 3,628-bp EcoRI insert was isolated (Fig. 3). A labeled ca. 650-bp BamHI fragment of pRER3 served to clone the overlapping 2,449-bp PstI insert present on pRER4.

FIG. 2 Silver-stained SDS-polyacrylamide gel showing purified 4-carboxymuconolactone decarboxylase:3-oxoadipate enol-lactone hydrolase from R. opacus 1CP in lane 2. Lane 1, Pharmacia low-molecular-weight markers: bovine α-lactalbumin (14.4 kDa), soybean (more ...)

FIG. 3 Restriction map of the inserts of pRER3 and pRER4, together carrying 5,475 bp of R. opacus 1CP DNA in the multiple cloning site of pBluescript II KS (+). Arrows indicate the lengths and orientations of open reading frames. Overlapping open reading (more ...)

Characterization of the inserts of pRER3 and pRER4. The sequence of the overlapping inserts of pRER3 and pRER4 comprises five complete and two incomplete open reading frames (Fig. 3 and 4). The first two complete open reading frames start at positions 408 and 1121, respectively. They are transcribed in the same direction and overlap by 1 bp. Their predicted N-terminal sequences exactly match those determined for the two subunits of purified protocatechuate 3,4-dioxygenase (Table 3). The predicted sequences of the gene products were found to be homologous to PcaH and PcaG of A. calcoaceticus, Burkholderia cepacia, and P. putida (Fig. 5). Thus, the two open reading frames, coding for polypeptides with molecular masses of 26,837 and 22,936 Da, were designated pcaH and pcaG.

FIG. 5 Sequence alignment of the subunits of protocatechuate 3,4-dioxygenases. Positions identical in both α and β subunits of all four enzymes are highlighted by black boxes; those identical in only one subunit of all four enzymes are shaded. (more ...)

A third complete open reading frame extends from positions 1791 to 3143 (Fig. 4). The predicted gene product is identical to the sequences predicted for 3-carboxymuconate cycloisomerases of A. calcoaceticus, Bradyrhizobium japonicum, and P. putida (L05770 [36], Y10223 [3], and L17082 [71]) in 36.9, 42.0, and 40.1%, respectively, of the positions. Thus, the third complete open reading frame was considered to represent the pcaB gene coding for 3-carboxymuconate cycloisomerase.

Overlapping pcaB by 4 bp, a fourth complete open reading frame, extending from positions 3140 to 4342, was detected (Fig. 4). All sequences of tryptic peptides and of the N terminus of the 3-oxoadipate enol-lactone-hydrolyzing enzyme occur in the protein sequence predicted from the open reading frame (Table 3; Fig. 4). In addition, the calculated molecular mass of the gene product (42,230 Da) corresponds to that of the purified protein (Fig. 2). Most interestingly, comparisons of the sequence to databases revealed similarities to both 3-oxoadipate enol-lactone hydrolases and 4-carboxymuconolactone decarboxylases. Further analyses showed that the N-terminal two-thirds of the protein are homologous to the hydrolases whereas the C-terminal third is homologous to the decarboxylases (Fig. 6). These results suggested that the protein encoded by the fourth open reading frame has 4-carboxymuconolactone-decarboxylating activity in addition to 3-oxoadipate enol-lactone-hydrolyzing activity. To test this hypothesis, we attempted to purify the enzymes responsible for these activities from an extract of benzoate-grown cells of R. opacus 1CP. As shown in Fig. 7 and Table 2, the 4-carboxymuconolactone decarboxylase and 3-oxoadipate enol-lactone hydrolase activities could be separated from each other by neither anion-exchange chromatography nor hydrophobic interaction chromatography. The fused gene coding for both PcaC and PcaD activities was designated pcaL, to differentiate it from the two precursor genes.

FIG. 6 Sequence alignment of the 4-carboxymuconolactone decarboxylase:3-oxoadipate enol-lactone hydrolase PcaL of R. opacus 1CP to the 4-carboxymuconolactone decarboxylases PcaC of A. calcoaceticus (M33798 [24]) and B. japonicum (Y10223 [ (more ...)

FIG. 7 Elution profiles of 3-oxoadipate enol-lactone-hydrolyzing and 4-carboxymuconolactone-decarboxylating activities during chromatography on Q-Sepharose High Performance (A) and Phenyl Superose (B). Experimental details are given in Table 3. Closed and open (more ...)

The fifth complete open reading frame, between positions 4339 and 5136, overlaps pcaL by 4 bp. The predicted gene product shares 43% identical positions with PcaR of P. putida (L33795 [57]), 34.7% with PcaU of A. calcoaceticus (U04359 [20]), and 37.4% with PobR of A. calcoaceticus (L13114 [11]). All three proteins belong to the IclR family of regulators and are activators of pca and pob genes, respectively, in the two species. Thus, the fifth open reading frame was assumed to represent the regulatory gene pcaR.

Downstream of the presumed pcaR (start at position 5158) and transcribed in the same direction, an incomplete open reading frame was detected. The predicted gene product appeared to be homologous to 3-oxoadipyl-CoA thiolases, as shown by 38.8% identical positions to PcaF of A. calcoaceticus (L05770 [36]) and P. putida (U10895 [27]). Therefore, this incomplete open reading frame was designated pcaF′.

Transcribed divergently from pcaH, another incomplete open reading frame was identified (start at position 255). It was termed pcaI′ due to the similarity of the predicted gene product to PcaI, a subunit of succinyl-CoA:3-oxoadipate CoA-transferases, of A. calcoaceticus (L05770 [36]) and P. putida (M88763 [49]) (37.7 and 41.2%, respectively, identical positions).

Characterization of the gene coding for the 3-oxoadipate enol-lactone-hydrolyzing enzyme induced during growth of R. opacus 1CP with benzoate showed it to be part of a protocatechuate catabolic gene cluster. This was unexpected, since Rann and Cain (56) as well as Cain (8) had argued that in their Rhodococcus strains, two different 3-oxoadipate enol-lactone hydrolases are induced for the catechol and protocatechuate branches of the 3-oxoadipate pathway. The isoenzymes could be distinguished by their denaturation kinetics at 41°C, by their K_m values, and by their gel filtration behavior. Whether the reason for the discrepancy between our results and those of Cain’s group is the investigation of different strains or whether other factors are responsible is not known.

The 3-oxoadipate enol-lactone-hydrolyzing enzyme of R. opacus 1CP did not reveal any unusual similarities to the dienelactone hydrolases of chlorocatechol catabolic pathways. The enzyme was, however, found to be quite unique in being able to catalyze two subsequent reactions, 4-carboxymuconolactone decarboxylation and 3-oxoadipate enol-lactone hydrolysis. This conclusion was inferred from the occurrence of a pcaD-like segment as well as a pcaC-like segment within one open reading frame, from the molecular mass being consistent with such a merged protein, and from the inability to separate the two activities chromatographically. The proteobacterial 4-carboxymuconolactone decarboxylases, in contrast, are separate enzymes (47), i.e., not part of a larger enzyme with additional hydrolytic activity (and vice versa). Since the general trend of protein evolution goes from simple to more complex enzymes and since gene fusions play a major role in this process (5, 34, 62), the enzymatic situation in proteobacteria may be considered the more ancient one. Consequently, during evolution of the protocatechuate pathway found in Rhodococcus, the original pcaD and pcaC genes were fused to code for a protein with both decarboxylase and hydrolase activities. This merger of two proteins catalyzing sequential reactions might be beneficial to the cells for two reasons: (i) it might contribute to minimizing the distance which a metabolic intermediate has to diffuse to the next active site; (ii) alternatively, if the distances are already optimized by the enzymes of the 3-oxoadipate pathway forming a multienzyme complex, as suggested by Meagher et al. (41), this complex might be somewhat stabilized by two of the components being fused into one protein. The fusion could apparently occur without other major rearrangements: in the γ-proteobacterium P. putida as well as in the α-proteobacteria A. tumefaciens and B. japonicum, pcaC is located directly downstream of pcaD within the same operon (Fig. 8) (3, 32, 52). In A. calcoaceticus, this order is also present, but the genes are interrupted by pcaK, a gene for a transporter (36). The order whereby the PcaC segment follows the PcaD segment is retained in the fused protein, which we designated PcaL. The PcaD-homologous region of PcaL extends to position 268 of the alignment, and the PcaC-homologous segment starts at position 271 (Fig. 6). Thus, it was apparently not necessary to introduce any major loops for the new protein to accommodate both domains.

FIG. 8 Structures of gene clusters for protocatechuate catabolism. Homologous genes are shaded in the same way. Arrows with dotted borderlines represent Rhodococcus genes for which no or only partial sequence information is available. The information was compiled (more ...)

Other features of the protocatechuate gene clusters have also been conserved. Thus, pcaD directly follows pcaB, the gene for 3-carboxymuconate cycloisomerase, also in A. calcoaceticus, B. japonicum, and P. putida (3, 32, 36), whereas in A. tumefaciens a different order occurs (51). Moreover, pcaG, the gene for the β subunit of protocatechuate 3,4-dioxygenase, has been reported to be located downstream of pcaH, the gene for the α subunit, for A. tumefaciens, B. cepacia, A. calcoaceticus, and P. putida, representatives of the α-, β-, and γ-proteobacteria (19, 24, 51, 74).

From the order of the Rhodococcus pca genes, it is clear that pcaI and probably pcaJ, the genes coding for the two sub- unitsofsuccinyl-CoA:3-oxoadipateCoA-transferase,are transcribed independently from the other genes for protocatechuate catabolism. This corresponds to the finding of Cain (8) that the transferase activity showed a different behavior in the presence of glucose as a catabolite repressor than the other enzymes of the protocatechuate or catechol branch. It is unclear, however, whether pcaH, -G, -B, -L, -R, and -F are transcribed together as one operon or in separate units which are possibly governed by the same regulator and coinducer. An unusual feature of this organization is that pcaR, the presumed regulatory gene, overlaps pcaL by 4 bp, which suggests that these genes are transcribed together and may even be translationally coupled (44, 69). In contrast, pcaU of A. calcoaceticus and pcaR of A. tumefaciens are transcribed divergently from other pca genes (21, 52), whereas in P. putida pcaR is transcribed in the same direction as the pca structural genes but is separated from them by more than 400 bp (27, 57). Activators of catabolic pathways, like many LysR-type proteins, often repress their own formation at the transcriptional level and thus tend to generate a relatively constant level of the regulator (59). Transcriptional repression of their own gene has also been found for the IclR-type regulators PobR (10), PcaR (29), and PcaU (54). Although genes forming one operon may in addition be transcribed separately, as in the case of P. putida catA (31), the organization of the pca genes of R. opacus 1CP suggests that the presumed pcaR gene may be cotranscribed with the structural genes, presumably resulting in vaying levels of the activator. It would be interesting to study this regulation and the possible influence of other regulatory elements in more detail.

We thank H.-J. Knackmuss for providing the environment in which we could do this work. We are indebted to R. Getzlaff and V. Noedinger for protein sequencing, to R. Schmid for the opportunity to use an automated DNA sequencer, and to S. Bürger for performing the DNA sequencing. Thanks are due to D. Parke for providing pARO523 and for helpful advice on setting up the decarboxylase assay.

The work was supported by a fellowship of the Landesgraduiertenförderung to D.E. and by grants from the Deutsche Forschungsgemeinschaft.