A Novel Pathway of Aerobic Benzoate Catabolism in the BacteriaAzoarcus evansii and Bacillus stearothermophilus*

The aerobic catabolism of benzoate was studied in the Gram-negative proteobacterium Azoarcus evansii and in the Gram-positive bacterium Bacillus stearothermophilus. In contrast to earlier proposals, benzoate was not converted into hydroxybenzoate or gentisate. Rather, benzoyl-CoA was a product of benzoate catabolism in both microbial species under aerobic conditionsin vivo. Benzoyl-CoA was converted into various CoA thioesters by cell extracts of both species in oxygen- and NADPH-dependent reactions. Using [ring-13C6]benzoyl-CoA as substrate,cis-3,4-[2,3,4,5,6-13C5]dehydroadipyl-CoA,trans-2,3-[2,3,4,5,6-13C5]dehydroadipyl-CoA, the 3,6-lactone of 3-[2,3,4,5,6-13C5]hydroxyadipyl-CoA, and 3-[2,3,4,5,6-13C5]hydroxyadipyl-CoA were identified as products by NMR spectroscopy. A protein mixture ofA. evansii transformed [ring-13C6]benzoyl-CoA in an NADPH- and oxygen-dependent reaction into 6-[2,3,4,5,6-13C5]hydroxy-3-hexenoyl-CoA. The data suggest a novel aerobic pathway of benzoate catabolism via CoA intermediates leading to β-ketoadipyl-CoA, an intermediate of the known β-ketoadipate pathway.

This report describes the conversion of [ 13 C]benzoyl-CoA by cell extracts of A. evansii and B. stearothermophilus under aerobic conditions. Intermediates identified by NMR spectroscopy point to a gentisate-independent and hitherto unknown mechanism of aerobic benzoate degradation.
Bacterial Cultures-A. evansii KB740 (DSM6869) (6) (formerly designated Pseudomonas sp. KB740) (7) was grown aerobically at 37°C with benzoate or 3-hydroxybenzoate as the sole source of cell carbon and energy (8,14) in a 200-liter fermentor (air flow, 100 liters/min; 200 rpm). Benzoate (10 mM) was added continuously when the initially added substrate (5 mM) was almost consumed. Cells were harvested in the exponential growth phase at an optical density (578 nm) of 2.3, corresponding to ϳ0.6 g of cells (dry weight)/liter. The culture was cooled to 15°C, and cells were harvested by continuous flow centrifugation. The yield was ϳ200 g of cells (wet weight)/mol of benzoate. Alternatively, cells were grown in shaker flasks containing 5 mM 3-hydroxybenzoate.
B. stearothermophilus PK1 (5) (courtesy of F. Lingens and J. Eberspä cher, Universitä t Hohenheim) was grown at 56°C in mineral salt medium (5) with benzoate (14 mM) as the sole source of cell carbon and energy in a 200-liter fermentor (air flow, 100 liters/min; 300 rpm). At an optical density (578 nm) of 1.2, the culture was cooled to 15°C, and then cells were harvested by continuous flow centrifugation. Alternatively, cells were grown in 1-liter shaker flasks containing 400 ml of medium (150 rpm).
Preparation of Cell Extracts-All steps were performed at 4°C. Frozen cells were suspended in an equal volume of water (A. evansii) or 200 mM Tris-HCl (pH 8) (B. stearothermophilus) containing 0.1 mg of DNase I/ml. The suspensions were passed through a French pressure cell at 123 megapascals and then centrifuged (100,000 ϫ g).
Synthesis of CoA Thioesters-Benzoyl-CoA and 3-hydroxybenzoyl-CoA were prepared by published procedures (15,18). The yield was 70 -80%. [ring-13 C 6 ]Benzoyl-CoA was prepared by a slight modification of a published procedure (15). A reaction mixture containing 5 mmol of [ring-13 C 6 ]benzoate, 5 mmol of N-hydroxysuccinimide, and 5 mmol of dicyclohexylcarbodiimide in 30 ml of dioxane was filtered, and the solvent was evaporated under reduced pressure. An aliquot of the succinimidyl ester (400 mol) of the residue was dissolved in 1 ml of dioxane. Aliquots of 100 l were added to 20 ml of 0.1 M sodium bicarbonate (pH 8) containing 200 mol of coenzyme A. The reaction mixture was stirred at room temperature. Aliquots were retrieved at intervals, and thioester formation was monitored by the nitroprusside assay (16). After 1 h, the mixture was acidified to pH 3.5 by addition of 4 ml of 2 M formic acid and was then extracted with diethyl ether (3 ϫ 50 ml). The aqueous phase was lyophilized. The residue was dissolved in 5 ml of 20 mM ammonium formate (pH 3.5) containing 2% (v/v) methanol (solvent 1). The solution was applied to a solid-phase extraction column (end-capped C 18 material, 10 g; reservoir volume, 60 ml; flow rate, 0.5 ml/min; ICT, Bad Homburg, Germany) that had been equilibrated with the same solvent. The column was washed with 120 ml of solvent 1 and then eluted with 80% (v/v) aqueous methanol. The eluate was concentrated under reduced pressure and lyophilized. The yield was 80%.
In an attempt to inhibit the aerobic benzoate catabolism, 5 mM trifluoroacetate or 5 mM 2-, 3-, or 4-fluorobenzoate was added to the assay and incubated for 20 min at 37°C prior to addition of 1 mM [ 14 C]benzoate (0.11 GBq/mmol); 3-hydroxybenzoate and gentisate were not added. The subsequent steps were performed as described above.
In Vitro Assays with Cell Extracts of A. evansii-Assay mixtures (0.25 ml, pH 6.3) containing 0.6 mM NADPH, 0.2 mM [ring-14 C]benzoyl-CoA (66.6 MBq/mmol), and 25 l of cell extracts of benzoate-or 3-hydroxybenzoate-grown A. evansii (52 mg of protein/ml) (20) were stirred at 22°C. Aliquots (50 l) were retrieved at intervals. The enzyme reaction was stopped by adding 200 l of ethanol (Ϫ80°C). Denatured protein was removed by centrifugation. The supernatant was evaporated under reduced pressure at 45°C, and the residue was dissolved in 100 l of water. Aliquots of 30 l were applied to a column of Lichrospher RP18 (5 m, 125 ϫ 4 mm; Merck, Darmstadt, Germany) that had been equilibrated with solvent 3 containing 2% (v/v) acetonitrile. The column was eluted at a flow rate of 1 ml/min with 10 ml of solvent 3 containing 2% (v/v) acetonitrile and then with 25 ml of a linear gradient of 2-10% (v/v) acetonitrile in solvent 3. Residual benzoyl-CoA was eluted with 10 ml of a linear gradient of 10 -40% (v/v) acetonitrile in solvent 3. The effluent was monitored with the radioactivity detector and the UV-diode array detector. Retention times were as follows: polar products, 1.5 min; benzoate, 6 min; product 1, 17 min; product 2, 23 min; product 3, 28 min; benzoyl-CoA, 41 min; and hydrolytic degradation product of benzoyl-CoA, 42.5 min.
Control assays were performed as follows. (i) NADPH was replaced by NADH (0.6 mM); (ii) FAD (0.2 mM) was added; (iii) NADPH was omitted; (iv) oxygen was removed by flushing the assay with nitrogen; and (v) [ring- 14   30 ml of solvent 3 containing 2% (v/v) acetonitrile, then with 75 ml of a linear gradient of 2-10% (v/v) acetonitrile in solvent 3, and finally with 30 ml of a linear gradient of 10 -40% (v/v) acetonitrile in solvent 3. Retention times were as follows: polar products, 7 min; benzoate, 33 min; product 1, 62 min; product 2, 82 min; product 3, 96 min; benzoyl-CoA, 124 min; and hydrolytic degradation product of benzoyl-CoA, 127 min. Products 1-3 were collected separately. The volume of the three fractions was reduced to 0.1 liter by flash-evaporation at 45°C and 40 millibars. The residual solution was lyophilized.
The residue containing product 1 was dissolved in water containing 0.1% (v/v) trifluoroacetic acid and 1% (v/v) methanol (solvent 4) and applied to a solid-phase extraction column (ENV ϩ , 500 mg; reservoir volume, 6 ml; flow rate, 0.5 ml/min; ICT) that had been equilibrated with solvent 4. The column was washed with 12 ml of solvent 4 and then developed with 10 ml of 80% (v/v) aqueous methanol. The eluent was concentrated under reduced pressure and lyophilized.
The residues containing product 2 or 3 were dissolved in water and applied to a solid-phase extraction column (ODS-AQ, 500 mg; reservoir volume, 2 ml; flow rate, 0.1 ml/min; YMC) that had been equilibrated with water. The column was washed with 12 ml of water and then developed with 10 ml of 80% (v/v) aqueous methanol. The solution was concentrated under reduced pressure and lyophilized.
Experiments with Protein Fractions of A. evansii-A protein fraction transforming benzoyl-CoA (as determined by HPLC analysis) was prepared from extracts from 17.5 g of cells (wet weight). 35 ml of a cell extract of A. evansii (from 17.5 g of cells (wet weight)) were applied to a DEAE-Sepharose Fast Flow column (diameter, 4 cm; volume, 80 ml; Amersham Pharmacia Biotech, Freiburg, Germany) that had been equilibrated with 10 mM Tris-HCl (pH 8.0) (solvent 5) at a flow rate of 3.5 ml/min. The column was washed with 80 ml of solvent 5 and then with 680 ml of solvent 5 containing 90 mM KCl. The column was developed with 200 ml of solvent 5 containing 150 mM KCl. The fraction was collected, and glycerol was added to a final concentration of 10% (v/v). The solution (1.3 mg of protein/ml) was kept at Ϫ20°C for 15 h. Aliquots (50 ml) of this protein fraction were applied to a column of hydroxylapatite (40 m; bed volume, 10 ml; Macro-Prep cerami hydroxylapatite, Bio-Rad, Mü nchen, Germany) that had been equilibrated with solvent 5 at a flow rate of 1 ml/min. The column was washed with 30 ml of solvent 5 and 40 ml of solvent 5 containing 2 mM sodium phosphate (pH 8). Subsequently, the column was developed with 10 ml of solvent 5 containing 100 mM sodium phosphate (pH 8). The fraction was collected; glycerol was added to a final concentration of 10% (v/v); and the solution (4.6 mg of protein/ml) was stored at Ϫ20°C.
Assay mixtures containing 10.6 ml of this protein fraction, 0.6 mM NADPH, 0.4 mM [ring-13 C 6 ]benzoyl-CoA (99.9% 13 C), and 2 Ci of [ 14 C[benzoyl-CoA were incubated at 22°C for 4.5 min. The reaction was stopped by addition of 40 ml of ethanol (Ϫ80°C). The mixture was centrifuged (20,000 ϫ g, 4°C), and the supernatant was concentrated to 10 ml under reduced pressure. The solution was acidified to pH 5 by addition of 20 l of 10% (v/v) formic acid, and aliquots of 100 l were analyzed by analytical HPLC as described above. Retention times of 14 C-labeled compounds were as follows: benzoyl-CoA, 42.5 min; polar products, 1.5 min; benzoate, 6 min; and product 6, 32.5 min.
Prior to product analysis by NMR spectroscopy, the acidified sample was desalted using a solid-phase extraction column (isolute end-capped C 18 ; volume, 6 ml; flow rate, 0.1 ml/min; ICT) that was equilibrated with 20 mM ammonium formate (pH 5) containing 1% (v/v) ethanol (solvent 6). The acidified sample was applied to the column. The column was washed with 6 ml of solvent 6 and then developed with 100% (v/v) ethanol. The eluent was concentrated to dryness under reduced pressure (35°C).
The fractions were desalted as described above.
NMR Spectroscopy-The lyophilized samples were dissolved in 0.5 ml of D 2 O (pH 6) (uncorrected glass electrode reading). 1 H and 13 C NMR spectra were measured at 20°C using a four-channel Bruker DRX 500 spectrometer. One-dimensional experiments and two-dimensional HMQC, HMQC-TOCSY, and INADEQUATE experiments were performed according to standard Bruker software (XWINNMR). The duration of the 1 H spin lock was 60 ms in the HMQC-TOCSY experiment. Chemical shift predictions were performed with SpecInfo (Chemical Concepts, Darmstadt).

Experiments with Suspensions of Whole Cells of A. evansii-
Suspensions of whole cells of benzoate-grown A. evansii were incubated with [ 14 C]benzoate (0.27 mM) at room temperature. The mixture contained 3-hydroxybenzoate (1 mM) and gentisate (1 mM) to trap intermediately formed 14 C-labeled 3-hydroxybenzoate or gentisate. At intervals, aliquots of the incubation mixture were retrieved, extracted, and analyzed by HPLC as described under "Experimental Procedures." 3-Hydroxybenzoate and [ 14 C]benzoate were consumed simultaneously. The analysis of gentisate consumption was not possible since gentisate is coeluted with other ultraviolet lightabsorbing compounds. After 15 s, a radiolabeled intermediate (5% of the initially added radioactivity) coeluting with an authentic sample of benzoyl-CoA was detected. After 10 min, the putative benzoyl-CoA intermediate disappeared, and polar products appeared, which did not bind to the column. None of the radiolabeled intermediates coeluted with 3-hydroxybenzoate, gentisate, or 3-hydroxybenzoyl-CoA, even if unlabeled 3-hydroxybenzoate was still present (data not shown). These results indicate that 3-hydroxybenzoate, 3-hydroxybenzoyl-CoA, and gentisate do not serve as intermediates in the aerobic catabolism of benzoate in A. evansii.
In an attempt to accumulate unknown intermediates, whole cells of A. evansii were incubated with [ 14 C]benzoate (0.27 mM) A. evansii (A) and B. stearothermophilus (B): HPLC profiles after different periods of incubation. 14 C elution profiles are shown. The recorder sensitivity for 2-and 5-min samples was increased by a factor of 2. For further details, see "Experimental Procedures." p1-p5, products 1-5. and fluoroacetate or fluorobenzoate specimens (5 mM) in the absence of 3-hydroxybenzoate and gentisate. HPLC analysis revealed that radiolabeled 3-hydroxybenzoate, 3-hydroxybenzoyl-CoA, and gentisate did not accumulate. However, the rates of product formation were significantly decreased.  (Fig. 3A). More specifically, a minor fraction (ϳ5%) of [ 14 C]benzoyl-CoA was found to be hydrolyzed to labeled benzoate and coenzyme A probably due to nonspecific thioesterases present in the crude cell extracts. The major part of benzoyl-CoA was converted into products (products 1-3) (Fig.  3A and Table I) that eluted from the reversed-phase column at retention times typical for CoA adducts. The product pattern shown in Fig. 3A was detected over a pH range of 6 -8. The products were collected, lyophilized, and analyzed by HPLC as described above. The rechromatography of product 1 reflected the original compound (retention time, 17 min; 70% of radioactivity) and a new compound with a retention time of 28 min (30% of radioactivity). Obviously, product 1 was partially converted into a novel compound during isolation. As shown below, NMR analysis of the fraction containing product 1 confirmed the presence of two compounds. Rechromatography of product 2 afforded one peak at a retention time of the original peak. Rechromatography of product 3 showed a major peak (75% of radioactivity) at the retention time of product 2 (23 min). NMR analysis confirmed that product 3 was converted into product 2 during isolation (see below). Fig. 4A shows the kinetics of substrate consumption and product formation in more detail. After 5 min, 80% of the initially added benzoyl-CoA was consumed. This process was accompanied by the formation of the major product 1 (maximum after 5 min; peak intensity reflected 40% of the initially added radioactive benzoyl-CoA) and the formation of the minor products 2 and 3. Products 1 and 3 were subsequently consumed in the course of the reaction, whereas product 2 seemed not to be converted any further. The amount of polar products increased steadily from the beginning on up to 40% of the initially added radioactivity. The total radioactivity decreased slightly after 7 min presumably due to the formation of labeled volatile 14 CO 2 .

FIG. 3. Conversion of [ 14 C]benzoyl-CoA into labeled products by cell extracts of
Addition of FAD (0.2 mM) inhibited the catabolism of benzo-yl-CoA. After 5 min, 38% of benzoyl-CoA was consumed compared with 80% in the control assay (without FAD). The product pattern was virtually identical to that in the control assay at all time points. Obviously, FAD inhibits one of the initial steps of benzoyl-CoA conversion. No labeled products were formed when NADPH was substituted by NADH (0.6 mM) or when NADPH was omitted. Moreover, product formation was strictly dependent on oxygen (Table I).
Experiments with Protein Fractions of A. evansii-Two chromatographic steps carried out with extracts of A. evansii afforded a protein mixture transforming [ring-13 C 6 ]benzoyl-CoA into a new product (product 6) by an NADPH-and oxygen-dependent reaction. Product 6 was isolated from the assay mixture by preparative HPLC and analyzed by NMR spectroscopy. It should be noted that product 6 was not detected in assays with crude cell extracts of A. evansii.
Product Analysis by NMR Spectroscopy-To determine the structures of the novel intermediates, [ring-13 C 6 ]benzoyl-CoA (0.2 mM) was incubated in large-scale experiments with cell extracts of A. evansii in the presence of NADPH (0.6 mM) and oxygen. A small amount of [ 14 C]benzoyl-CoA was added to facilitate product analysis and purification by HPLC. After 5 min, the incubation was stopped when the concentrations of products 1-3 were at their maximum. The products were purified by preparative HPLC and solid-phase extraction as described under "Experimental Procedures." Approximately 0.2 mg of products 1-3 each were obtained in virtually pure form and subjected to NMR spectroscopy.
The presence of CoA residues could be gleaned from the 1 H NMR signature for each product (data not shown). The 13 C NMR spectra were dominated by 13 C-coupled signals reflecting carbon atoms that were derived from the 13 C-labeled phenyl moiety of [ring-13 C 6 ]benzoyl-CoA. More specifically, the 13 C NMR spectrum of the fraction containing product 1 displayed

FIG. 4. Kinetics of product formation from [ 14 C]benzoyl-CoA by cell extracts of A. evansii (A) and B. stearothermophilus (B).
Radioactivity of individual products is referenced to the radioactivity of [ 14 C]benzoyl-CoA initially added to the assay (100%). 10 signals with one-bond 13 C-13 C coupling constants of 50 -30 Hz (Table II). Additionally, some signals showed long-range couplings with coupling constants of 2-5 Hz. On the basis of two-dimensional INADEQUATE experiments (Fig. 5) and analysis of carbon-carbon coupling constants (Table II), the 13 C NMR signals were attributed to two 13 C 5 -labeled compounds with similar coupling signatures (compounds 22 and 21) (Table  II). Two signals of compound 22 (181.1 and 47.9 ppm) and two signals of compound 21 (177.9 and 50.3 ppm) showed 13 C couplings to one adjacent 13 C atom, whereas three signals of each compound were characterized by simultaneous couplings to two directly adjacent 13 C-labeled atoms. This coupling signature provides firm evidence that the connectivity of the ring-13 C 6 -labeled substrate has been broken, leading to linear 13 C 5 spin systems, which indicated the loss of one 13 C atom in the degradation process, most probably as 13 CO 2 . The observed chemical shifts (Table II) were typical for aliphatic monohydroxylated carboxylates or derivatives thereof. The structures could be assigned from the spin networks gleaned from INAD-EQUATE experiments (carbon-carbon connectivities) and HMQC experiments (carbon-hydrogen connectivities) (Table  II). In more detail, the 13 C-labeled carboxylic carbon atom of compound 21 was shown to be connected to a 13 CH 2 -13 CH 2 -13 CH(OH)-13 CH 2 moiety, and compound 21 was established as ␤-[2,3,4,5,6-13 C 5 ]hydroxyadipyl-CoA. It should be noted that C-1 of compound 21 is biogenetically equivalent to the unlabeled carbonyl C-1 of benzoyl-CoA and could therefore not be directly detected by 13 (Table II) had chemical shift values that were consistent with alkene motifs. One doublet signal of product 2 was found at a chemical shift value of 181.5 ppm, typical for a carboxylic acid moiety; and two signals were detected at 35.1 and 28.3 ppm, reflecting two carbon atoms with sp 3 hybridization. The doublet signal of product 6 was found at 60.6 ppm, suggesting a 13 CH 2 OH moiety. The observed chemical shift pattern, as well as the carbon-carbon connectivities obtained from INADEQUATE experiments (Table II) and the carbonhydrogen connectivities obtained from HMQC and HMQC-TOCSY experiments (Fig. 7 and Table II) Product 3 gave NMR data identical to that described above for product 2 (compound 20). This might be explained by decay or isomerization, yielding trans-2,3-dehydroadipyl-CoA (product 2) during isolation (see also above).
Experiments with B. stearothermophilus-The conversion of [ring-13 C 6 ]benzoyl-CoA and [ 14 C]benzoyl-CoA by cell extracts of B. stearothermophilus was studied as described for A. evansii, but in the presence of 1 mM iodoacetamide. It was found that iodoacetamide increased the accumulation of early prod- ucts, most likely by inhibiting a late step in the overall transformation. HPLC (Fig. 3B) and NMR analysis identified the same products 1-3 as described above. Moreover, two additional products (products 4 and 5) (Fig. 3B) could be detected. Fig. 4B shows the kinetics of substrate consumption and product formation in more detail. Product 4 was the earliest of the detected products and was consumed in the course of the reaction. Products 2 and 3 were detected later. More polar products (non-CoA thioesters) accumulated steadily. The 13 C NMR spectrum of the fraction containing products 2 and 4 displayed five major signals that were identical to the NMR signature of trans-2,3-[2,3,4,5,6-13 C 5 ]dehydroadipyl-CoA (product 2, compound 20). Five novel signals were attributed to product 4 (Table II). The 13 C NMR chemical shifts again indicated a dehydroadipyl derivative. In contrast to compound 20, both signals of the alkene moiety displayed simultaneous coupling to two adjacent 13 C atoms. The carbon connectivity was further analyzed by INADEQUATE experiments, establishing a 13 CH 2 -13 CHϭ 13 CH-13 CH 2 motif (Table II). Thus, product 4 was assigned as 3,4-[2,3,4,5,6-13 C 5 ]dehydroadipyl-CoA (compound 18). A cis-configuration is suggested from the analysis of the 1 H NMR signature of H-3 and H-4 (7-Hz couplings) ( Table  II). The amount of product 5 was too low for NMR analysis. The identified compounds cannot be explained by conventional pathways (Fig. 1) or by previously suggested mechanisms via gentisate (compound 11) (Fig. 2), but point at a gentisate-independent novel pathway operative in the microbial species under study. A hypothetical pathway of aerobic benzoate metabolism in A. evansii and B. stearothermophilus integrating the identified intermediates is shown in Fig. 8. A. evansii and B. stearothermophilus appear to metabolize benzoate (compound 1) via coenzyme A thioesters. The first step of the pathway is the formation of benzoyl-CoA (compound 8) from benzoate (compound 1) and coenzyme A by a benzoate-CoA ligase.
Obviously, compound 19 or a compound closely related to compound 19 is an early intermediate of the new aerobic benzoate pathway. The compound was detected only with the protein mixture, but not with crude cell extracts. Probably, compound 19 could be oxidized to compounds 18 and 20 in assays containing crude cell extracts and a large excess of NADP ϩ (Fig. 8).
The conversion of benzoyl-CoA into 6-hydroxy-3-hexenoyl-CoA (compound 19) is a multistep reaction involving oxygenation, ring cleavage, and decarboxylation. A hypothetical mechanism of benzoyl-CoA (compound 8) conversion into compound 19 is shown in Fig. 8 A more conventional 2,3-dioxygenation of benzoyl-CoA (compound 8) followed by oxygenolytical intradiol ring cleavage would result in compound 18 via compounds 12 and 14 (Fig. 8). However, the reduction of compound 18 to compound 19 is hardly conceivable under the assay conditions. Therefore, a pathway via compound 16 appears more probable.
The postulated benzoyl-CoA oxygenase is currently under study. Remarkably, an iron-sulfur flavoprotein (BoxA protein) catalyzing the FAD-and benzoyl-CoA-dependent oxidation of NADPH (H 2 O 2 -forming) has been purified from A. evansii. 2 The boxA gene shows similarity to reductase domains of various hydroxylases. Moreover, boxA is adjacent to boxB in a hypothetical operon that is induced by benzoate. boxB has similarity to domains of putative multicomponent oxygenases. We suggest that a BoxA-BoxB complex is involved in benzoyl-CoA oxygenation in A. evansii. However, up to now, we were not able to reconstitute the native BoxA-BoxB complex.
It must be emphasized that the proposed reaction sequence remains speculative at the present level of experimental evidence. Moreover, it remains to be established whether the novel pathway is operative in other Bacillus strains and in the halophilic archaebacterium Haloferax sp., which have been claimed to metabolize benzoate via gentisate (2)(3)(4)22).
We can only speculate about the selective advantage of the proposed pathway. (i) One advantage may be that more complex aromatic substrates such as phenyl propionate and cinnamate can be converted by ␤-oxidation to benzoyl-CoA rather than benzoate. Recently, it was found that phenyl derivatives containing an odd or even number of carbon atoms will be catabolized through ␤-oxidation to either benzoyl-CoA (odd) or phenylacetyl-CoA (even), which could be further catabolized to the intermediates of general metabolism (23). (ii) Another advantage for facultative anaerobic bacteria could be that benzoyl-CoA, which is a characteristic intermediate of the anaerobic aromatic metabolism, may be immediately used as substrate of the aerobic metabolism when oxygen becomes available. This would allow a rapid shift from the anoxic to oxic mode of growth and vice versa. (iii) Also, thioester formation allows an efficient trapping of aromatic acids, very much the same as in the case of aliphatic fatty acids, e.g. in Escherichia coli. Fatty acid-CoA ligase is found in the soluble fraction after cell disruption, but is thought to act as a membrane-associated protein in whole cells, trapping incoming fatty acids (24). The energy-rich thioester bond is retained in the course of the pathway; therefore, the energy initially spent is useful not only for the active transport, but at the same time, for the activation of the acid. One may therefore speculate that benzoate-CoA ligase could behave similarly to fatty acid-CoA ligase. It was found that Pseudomonas putida U lost the ability to take up phenyl acetate when the gene coding for phenylacetyl-CoA ligase, the first enzyme of a new pathway of catabolism of phenylacetic acid, was mutated (23).
It is remarkable that besides benzoate, also 2-aminobenzoate (19,25) and phenyl acetate (26 -28) are aerobically metabolized in A. evansii (19,26) and possibly in some other bacteria via coenzyme A thioesters (10 -13). An inducible CoA ligase was also shown to be involved in the aerobic catabolic 4-chlorobenzoate metabolism (29,30). CoA thioesters are essential for the anaerobic metabolism of those compounds. This requires an additional whole set of specific coenzyme A ligases that are induced under aerobic conditions only and that differ, to some extent, from the respective "anaerobic" isoenzymes. The thio-esterified carboxy group may facilitate enzymatic steps that otherwise would not be possible with free acids. One example is the hydroxylation and subsequent reduction of 2-aminobenzoyl-CoA by the flavoenzyme 2-aminobenzoyl-CoA monooxygenase/reductase (31)(32)(33). Another example is the hydrolytic chloride elimination in 4-chlorobenzoyl-CoA (34,35). It will be interesting to see which kind of initial reactions benzoyl-CoA and phenylacetyl-CoA will undergo and what kind of role the coenzyme A carboxy thioester group plays in these unprecedented reactions.