FIG. 1 Proposed pathway of anaerobic toluene oxidation to benzoyl-CoA. The enzymes involved are indicated by their gene names: (1) benzylsuccinate synthase, BssABC; (2) succinyl-CoA:benzylsuccinate CoA-transferase, BbsEF; (3) benzylsuccinyl-CoA dehydrogenase, (more ...)

Further oxidation of benzylsuccinate to benzoyl-CoA was proposed to proceed via a modified β-oxidation pathway (5). Biochemical evidence for the existence of specific enzymes for benzylsuccinate oxidation was recently obtained: a substrate-induced succinyl-CoA-dependent CoA-transferase for benzylsuccinate and enzymes catalyzing benzylsuccinyl-CoA oxidation to benzoyl-CoA and succinyl-CoA have been identified in toluene-grown cells of Thauera aromatica (22). In this report, we present the cloning and molecular characterization of a toluene-induced operon of nine genes which appears to code for all of the enzymes required for β oxidation of benzylsuccinate.

Materials. Chemicals were obtained from Fluka, Merck, Roth, Sigma, and Bio-Rad. Synthesis of ferricenium-PF₆ was performed as described previously (19); benzylsuccinyl-CoA was prepared as described previously (22). Enzymes used for recombinant DNA techniques were obtained from MBI Fermentas or Pharmacia. Nitrocellulose membranes and nylon membranes were from Schleicher & Schuell or Amersham. ³²P- and ³⁵S-labeled compounds were from Amersham.

Strains and culture conditions. T. aromatica K172 (DSMZ6984; 2) was grown at 30°C under denitrifying conditions in mineral salt medium with toluene or benzoate as the sole substrate (5, 29). Escherichia coli DH5α [Δ(argF-lac)U169 (Φ80dlacΔM15)], K38 (hfrC ompF267 phoA4 pit-10 relA1), and P2392 (hsdR514 supE44 supF58 lacY1 galK2 galT22 metB1 trpR55 mcrA, P2 lysogen) were grown at 37°C in Luria-Bertani (LB) medium (23). The T. aromatica gene library used was prepared in λEMBL3 and has been described previously (7). Antibiotics were added at the following concentrations: ampicillin, 50 μg ml⁻¹; kanamycin, 40 μg ml⁻¹; and rifampin, 200 μg ml⁻¹. For λ phage propagation, liquid cultures of E. coli were infected by adding 5% (vol/vol) lysate and incubated for 8.5 h at 37°C.

Recombinant DNA and DNA sequencing techniques. Isolation of plasmid and phage DNA, PCR, and restriction, modification, and ligation of DNA fragments were performed by standard techniques (3, 25). E. coli DH5α was used as the host strain for plasmid manipulation; P2392 was used for phage growth. Transformation of E. coli strains was performed as described in reference 8 or 15. DNA sequencing was performed by cycle sequencing using fluorescein- or Cy5-labeled primers, fragment detection was performed by ALF or ALFexpress sequencers (Pharmacia). The complete DNA sequence of both strands was determined. Sequenced regions were extended by primer walking.

Protein electrophoresis. Discontinuous sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed in gels containing 10.5% acrylamide-bisacrylamide as described elsewhere (18). Two-dimensional (2D) gel electrophoresis was performed as described in reference 24 with 100,000 × g cell extracts prepared from toluene- and benzoate-grown cells of T. aromatica. For each experiment, 400 μg of protein was applied to the gels. The second dimension was performed with gels containing 11% acrylamide-bisacrylamide. The separated proteins were blotted on polyvinylidene membranes. The toluene-induced proteins were cut out and sequenced by gas/liquid-phase sequencing with an Applied Biosystems 473A sequencer.

Mapping of the 5′ ends of mRNA. Total cellular RNA was prepared from exponentially growing toluene- and benzoate-grown cells of T. aromatica, using the hot phenol method (1). The 5′ ends of mRNAs encoded by the bbs operon were mapped by a primer extension method with a Cy5-labeled oligonucleotide. For each experiment, 40 μg of RNA and 0.2 pmol of primer were used. The experiments were otherwise performed as described elsewhere (6). Extension products were purified by phenol extraction and analyzed with an ALFexpress sequencer.

T7 promoter-polymerase labeling of gene products. Fragments of the bbs operon were subcloned in the vector pCRScriptAmp SK (+). The resulting plasmids carry the corresponding genes under the control of the phage T7 10 promoter and were transformed into E. coli K38 carrying plasmid pGP1-2. Gene products were specifically labeled with an l-[³⁵S]methionine–l-[³⁵S]cysteine mixture as described elsewhere (28). Labeled proteins were separated by SDS-PAGE and detected by autoradiography.

Heterologous expression of bbsG and test of the activity of the gene product. The bbsG gene was cloned into the expression vector pTrc99A (Pharmacia) and overexpressed in E. coli DH5α. A 200-ml culture of transformed cells was grown to an optical density at 578 nm of 0.5 in LB medium at 37°C; then 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added, and incubation was continued for 4 h. Cells were harvested by centrifugation; the cell yield was 0.5 g (wet mass). Crude extracts were prepared by resuspending the cells in 2 volumes of basal buffer (100 mM Tris-HCl, 1 mM MgCl₂ [pH 7.6]), passing the suspension through a French press cell (137 MPa), and centrifuging the lysate 1 h at 100,000 × g. The supernatant was passed over a PD10 gel permeation column to remove low-molecular-mass molecules, some of which interfere with the enzyme assay. Benzylsuccinyl-CoA dehydrogenase was assayed in basal buffer, containing 0.1 mM ferricenium-PF₆ as an artificial electron acceptor and 10 μl of PD10-treated extract (equivalent to 0.13 mg of protein). The reaction was started by adding 0.3 to 0.6 mM benzylsuccinyl-CoA. Enzyme activity was monitored as decrease of absorbance of the ferricenium ion at 300 nm ( = 4,300 M⁻¹ cm⁻¹).

Nucleotide sequence accession number. The sequence data reported here were submitted to the GenBank database under accession no. AF173961.

Analysis of toluene-induced proteins in T. aromatica. The enzymes catalyzing benzylsuccinate oxidation to benzoyl-CoA play an important role in anaerobic toluene oxidation. They were expected to be specifically induced in toluene-grown cells. Analysis of extracts of toluene- and benzoate-grown cells of T. aromatica by one- and two-dimensional PAGE showed the occurrence of at least nine toluene-induced proteins, in addition to the previously known subunits of the first enzyme of the pathway, benzylsuccinate synthase (10, 20). Eight of these proteins were resolved on 2D gels, whereas an induced protein of 15.6 kDa was visible only on one-dimensional gels (Fig. 2). The N-terminal sequences of seven of these proteins were determined from blotted gels (Table 1). One of the determined sequences (Table 1, BbsB) showed similarity to the β-ketothiolase domain of eukaryotic sterol binding proteins (27) and some archaeal homologues. This indicated that the protein may be involved in the last step of the proposed degradation pathway of benzylsuccinate (Fig. 1, enzyme 6).

FIG. 2 Analysis of toluene-induced proteins. (A) SDS-PAGE of extracts of benzoate-grown (lane 2) and toluene-grown (lane 3) cells of T. aromatica. Molecular masses of marker proteins (lane 1) are given in the left margin; the 2D gels are aligned relative to (more ...)

TABLE 1 Features of bbs geneproducts

Cloning and sequencing of the bbs operon. To obtain a hybridization probe for the genes coding for toluene-induced proteins, PCR assays were performed with degenerate primers derived from the N-terminal sequence of the toluene-induced thiolase (BbsB) and from a C-terminal consensus sequence derived from an alignment of a number of known thiolases (equivalent to Prosite entry PS00737). A specifically amplified DNA fragment of 1 kb was obtained, as predicted from the size of the protein. Direct DNA sequencing of the fragment confirmed that it coded for the determined amino acid sequence (data not shown). This PCR product was then labeled with [³²P]dCTP and used as a hybridization probe to screen a λEMBL3 gene bank of T. aromatica (7). Two overlapping positive clones containing inserts of 10 to 12 kb were obtained from 1,500 tested plaques. The insert DNAs were subcloned in pCRScript and used for sequencing. The nucleotide sequence of a 9.4-kb segment was determined in both orientations from several subclones. The subclone borders were checked by PCR amplification of overlapping fragments and sequencing of the overlaps. Small segments of DNA missing at subclone borders were sequenced directly from the primary amplicons. The predicted reading frames and the operon structure of the sequenced DNA segment are shown in Fig. 3. The deduced gene products of the bbs operon were confirmed by specific labeling experiments using the T7 promoter-polymerase system in E. coli. Plasmids containing different parts of the bbs operon behind a T7 promoter (Fig. 3) were constructed and used for labeling of the encoded gene products with [³⁵S]methionine–[³⁵S]cysteine. An SDS-gel analysis of the labeled proteins showed that most of the expected gene products were synthesized (Fig. 4). Expression of a plasmid containing the bbsF and bbsG genes yielded only a single labeled product, even when analyzed under conditions that resolved the different induced proteins of that size range in cell extracts (data not shown). Since the bbsG gene was shown to be expressed in labeling experiments with a shortened plasmid lacking bbsF (data not shown), we assume that the bbsF gene product was not synthesized from the T7 promoter plasmid used.

FIG. 3 Organization of the bbs operon. The sequenced fragment is indicated by the locations of open reading frames. The bbs genes are shown as hatched boxes; flanking open reading frames are shown as open boxes. Restriction sites for EcoRI (E) and HindIII (H) (more ...)

FIG. 4 Expression of the bbs genes by the T7 promoter-polymerase system. The autoradiogram is of a 12.5% polyacrylamide gel in which SDS lysates from equal amounts of cells were separated. Migration positions of nonlabeled standard proteins are indicated. (more ...)

Organization of the genes and features of the derived gene products. Nine closely spaced open reading frames were identified in the sequence. They code for enzymes of a β-oxidation pathway and were therefore termed bbsA to bbsI (for beta-oxidation of benzylsuccinate) in Fig. 3. A rather long intergenic spacer of 118 bp is found between bbsF and bbsG, whereas the bbsE and bbsF genes overlap by 14 bp. The distance of the bbs genes to flanking genes is 340 bp at the 5′ flank (same orientation) and 146 bp at the 3′ flank (opposite orientation). With the exception of the bbsC gene product, the predicted N-terminal sequences of all bbs gene products were verified from our initial analysis of toluene-induced proteins. Table 1 shows the predicted and experimentally obtained sizes, pI values, and N-terminal sequences of the proteins. The N-terminal methionines are retained for BbsB, BbsG, and BbsI; those of the other bbs gene products are cleaved off. In the following paragraphs, the derived gene products of the bbs genes and their predicted functions are discussed in the order of their positions in the β-oxidation pathway.

(i) BbsEF. The bbsE and bbsF genes code for two toluene-induced proteins of very similar masses but different pI values (Fig. 2), as revealed by analysis of 2D gels. The gene products exhibit 26% identity with each other. Database searches yielded a number of proteins related to each of the two gene products (best matches: 29 and 31% identity with hypothetical proteins from Sphingomonas aromaticivorans [accession no. AF079317] and Streptomyces coelicolor [accession no. CAB52867], respectively. The best-characterized of the similar proteins was an oxalyl-CoA:formate CoA-transferase of Oxalobacter formigenes, one of the enzymes involved in oxalate fermentation (27). Since a specific succinyl-CoA:(R)-(+)-benzylsuccinate CoA-transferase has been shown to activate benzylsuccinate to the CoA thioester as the first step of benzylsuccinate oxidation (22), we propose that the bbsE and bbsF genes code for this CoA-transferase. The presence of two genes and the observed equal staining intensity of the two derived proteins in 2D gels suggest that the enzyme may consist of two subunits. This assumption is confirmed by preliminary data from the purified CoA-transferase, which indeed consists of two subunits of the predicted sizes (C. Leutwein and J. Heider, unpublished data). The translation efficiency of the bbsF gene, which is preceded by a poor ribosome binding site, is probably enhanced by translational coupling to bbsE. This assumption may also explain why no apparent expression of bbsF could be observed from T7 promoter plasmid pE3 (Fig. 3), which does not contain the translational start region of the preceding the bbsE gene (Fig. 4).

(ii) BbsG. The bbsG gene codes for the second enzyme of the postulated pathway, a benzylsuccinyl-CoA dehydrogenase. This is clear from the functional overexpression of the gene in E. coli (see below) and from the strong similarity of the predicted gene product to a number of acyl-CoA dehydrogenases (best match, 36% identity to a hypothetical acyl-CoA dehydrogenase from Mycobacterium tuberculosis [accession no. P96808]). The consensus patterns listed in the Prosite database (Prosite PS00072 and PS00073) are not perfectly conserved, but both can be found in the BbsG sequence. A conserved glutamate close to the C terminus (E₃₇₉ in BbsG) is known to be an active-site residue in acyl-CoA dehydrogenases. Residues that participate in the binding of the cofactor flavin adenin dinucleotide are also conserved in BbsG (16). As with other identified acyl-CoA dehydrogenases, the bbsG gene product appears to be a soluble protein and probably transfers the redox equivalents received from the substrate to an electron-transferring flavoprotein. The bbsG gene is separated from the previous gene of the operon by a rather long spacer of 118 bp and is preceded by a weak ribosome binding site. Nevertheless, the derived gene product was detected as a major toluene-induced protein on 2D gels, exhibiting intensity equal to that of the other gene products of the operon; its mass and pI values fit well with the predictions (Table 1).

(iii) BbsH. The bbsH gene product exhibits strong similarity to enoyl-CoA hydratases (41% identity with a putative enoyl-CoA hydratase from Archaeoglobus fulgidus [accession no. O29814]). The amino acid sequence of BbsH contains the Prosite pattern of this enzyme family (PS00166), and two glutamate residues of the active site of enoyl-CoA hydratases (E₁₀₉ and E₁₂₉ in BbsH) (14) are conserved. Therefore, we conclude that BbsH is the third enzyme of the β-oxidation pathway of benzylsuccinate and catalyzes the hydration of phenylitaconyl-CoA to 2-carboxymethyl-3-hydroxy-3-phenylpropionyl-CoA.

(iv) BbsCD. The gene products of bbsCD are 34% identical with each other. Both are highly similar to a number of known short-chain alcohol dehydrogenases (best matches, 35 and 45% identity with 3-oxoacyl-acyl carrier protein reductases from Pseudomonas aeruginosa [accession no. O54438] and Thermotoga maritima [accession no. AAD36790], respectively). Thus, we attribute one or both of these genes to the fourth enzyme of the proposed pathway, the 3-hydroxyacyl-CoA dehydrogenase. Only BbsD was identified as a toluene-induced protein in our 2D gel analysis. Although the N terminus of BbsC has not been obtained from 2D gels, the bbsC gene product may correspond to a 27.5-kDa toluene-induced protein whose N terminus was blocked for N-terminal sequencing (Fig. 2). We have no indication yet of whether the bbsCD gene products are subunits of a single enzyme or represent two different isoenzymes.

(v) BbsB. The bbsB gene product most probably codes for the last enzyme required for the proposed metabolic pathway of benzylsuccinate oxidation, benzoylsuccinyl-CoA thiolase. This can be derived from the results of database searches, which showed strong similarity of BbsB to thiolases (e.g., 37% identity with the thiolase domain of a putative lipid transfer protein of Methanobacterium thermoautotrophicum [accession no. O26884]). Only two of three Prosite patterns for thiolases (PS0098 and PS00737) can be located in the sequence. Cys₈₄ of BbsB is conserved in all thiolases and corresponds to the site of formation of a covalent acyl-S enzyme intermediate in the reaction mechanism. A second active-site cysteine close to the C terminus of well-analyzed thiolases is missing in BbsB. Significantly, this cysteine is also absent in the thiolases in the database that exhibit the highest sequence similarity to BbsB. Some of these enzymes, such as the thiolase domain of eukaryotic sterol binding protein, were recently shown to preferentially catalyze the thiolytic cleavage of α-methyl-branched 3-oxoacyl-CoA substrates (26). This correlates well with the expected specificity of BbsB for an α-carboxymethyl-branched substrate (Fig. 1).

BbsA and BbsI. Surprisingly, the bbs operon contains two additional genes, which code for hypothetical proteins of unknown functions and are expressed in toluene-grown cells at approximately the same levels as the other genes of the operon. The first reading frame, bbsA, codes for a 16.2-kDa protein which has been identified in cell extracts separated by one-dimensional SDS-PAGE but not in 2D gels. This is probably due to the very high theoretical pI value of BbsA (Table 1), which was outside the pH range used in our study. Database searches revealed that BbsA belongs to a family of proteins of no known function (e.g., 34% identity to hypothetical protein MJ1552 from Methanococcus jannaschii [accession no. Q58947]). The last reading frame of the operon, bbsI, codes for a toluene-induced protein of 209 amino acids and a mass of 23.2 kDa (Table 1). The sequence of BbsI is similar to those of several proteins in the database (e.g., 29% identical with a hypothetical protein from Aquifex aeolicus [accession no. O67293]). However, the functions of these proteins are unknown; therefore, we are unable to determine the role of BbsI in the catabolism of benzylsuccinate.

Transcript mapping of the bbs operon. The transcription start site of the bbs operon was determined by primer extension analysis with a fluorescence-labeled primer. Experiments were performed with RNA prepared from cells growing anaerobically on either toluene or benzoate, which were harvested during the exponential growth phase. A single start signal was obtained 35 bp upstream of the start codon of bbsA in the experiments with toluene-grown cells, whereas no signal above background was obtained with RNA from benzoate-grown cells (Fig. 5). This finding confirms that substrate induction of the toluene-catabolic enzymes occurs at the transcriptional level, as previously reported for the bss operon (20). Closer inspection of the sequence suggests remote similarity to −10 and −35 boxes of E. coli promoters upstream of the determined start site (Fig. 5), but since this is the first mapped transcription start in any Thauera species, we cannot yet compare this promoter site to a consensus sequence.

FIG. 5 Promoter mapping of the bbs operon. (A) Fluorograms of primer extension assays with RNA prepared from cells grown on toluene (Tol) and benzoate (Bz). (B) Fluorogram of a DNA sequencing assay with the same primer. (C) Indication of the transcription start (more ...)

Functional overexpression of the bbsG gene in E. coli. The bbsG gene, which is predicted to encode benzylsuccinyl-CoA dehydrogenase (enzyme 3 in Fig. 1), was overexpressed in E. coli by an IPTG-inducible trc promoter system. Significant amounts of soluble BbsG protein, corresponding to ca. 5% of the total protein, were obtained in cell extracts of overproducing cells (data not shown). These extracts were used to test for the activity of benzylsuccinyl-CoA dehydrogenase, using chemically synthesized (R/S)-benzylsuccinyl-CoA (22) to initiate the reaction. As shown in Table 2, benzylsuccinyl-CoA was oxidized by crude extracts of E. coli cells containing overproduced bbsG gene product. About sevenfold-lower activity was recorded in nonoverexpressing control cells of E. coli, probably caused by some cross-reactivity of other acyl-CoA dehydrogenases from the host cell. Substrate analogs such as free benzylsuccinate or succinyl-CoA were not accepted by the enzyme (Table 2). BbsG thus catalyzes the oxidation of benzylsuccinyl-CoA, consistent with its suggested role as a benzylsuccinyl-CoA dehydrogenase in the β oxidation of benzylsuccinate.

TABLE 2 Benzylsuccinyl-CoA dehydrogenase activity of recombinant BbsGproteina

Identification of the bbs operon increases the number of genes required for the anaerobic toluene catabolic pathway of T. aromatica to 13, located in the bss and bbs operons. The operons are not closely linked, as evident from the amount of analyzed flanking DNA of the obtained clones. Whereas the bss operon codes for the first enzyme of the pathway, benzylsuccinate synthase, plus an additional activating enzyme (20), the bbs operon apparently contains the remaining genes required for toluene oxidation to benzoyl-CoA. Several arguments support the notion that the encoded enzymes catalyze the β oxidation of benzylsuccinate to benzoyl-CoA. (i) Eight of the nine predicted gene products of the bbs operon were strongly induced by toluene, together with the bssA gene product, which is a subunit of benzylsuccinate synthase, the first enzyme of toluene degradation (20). (ii) A succinyl-CoA:benzylsuccinate CoA-transferase rather than a CoA ligase has been identified biochemically as the enzyme activating benzylsuccinate for β oxidation (22). This is matched by the similarity of the bbsE and bbsF gene products with CoA-transferases and the absence of a gene coding for a CoA ligase in the operon. (iii) The detected in vitro activity of heterologously produced BbsG is consistent with that of benzylsuccinyl-CoA dehydrogenase, one of the enzymes proposed to participate in the β oxidation of benzylsuccinate. (iv) The sequences of the proteins encoded by the bbs operon are consistent with all enzyme activities that have been proposed to constitute the benzylsuccinate β-oxidation pathway.

In addition to the genes coding for the putative enzymes required for the pathway of activation and β oxidation of benzylsuccinate, we found two extra genes in the bbs operon (bbsA and bbsI). These are expressed in toluene-grown cells at levels comparable to those of other bbs genes, and their possible roles cannot be predicted from sequence comparisons. Still, some clues as to the role of the bbsA gene product might be derived from the database searches. We observed that the bbsA gene product belongs to a family of hypothetical proteins from several sequenced genomes whose genes are always located immediately adjacent to thiolase genes. For example, 12 of the 15 genes coding for thiolases in the genome sequence of Archaeoglobus fulgidus are accompanied by genes coding for members of the BbsA family. It is therefore possible that BbsA is connected to the function of the benzoylsuccinyl-CoA thiolase, BbsB.

Expression of the bbs and bss operons is induced only in cells grown on toluene. This induction is mediated at the transcriptional level for both operons, as evident from primer extension analysis of the bbs operon and Northern blotting of the bss operon, as reported previously (20). Toluene induction appears to be mediated by the gene products of the tdiSR operon, which codes for a regulatory two-component system and is located immediately upstream of the bss operon (21). Taken together, the three operons containing the genes of the toluene pathway occupy ca. 20 kb of the genome of T. aromatica.

This work was supported by grants from the Deutsche Forschungsgemeinschaft.

We thank G. Fuchs (University of Freiburg) for his constant support and encouragement. We acknowledge H. Schägger (University of Frankfurt) for N-terminal sequencing of toluene-induced proteins, C. Leutwein for the synthesis of benzylsuccinyl-CoA, and J. Alt-Mörbe (Freiburg) for help with DNA sequencing.