Functional genomics by NMR spectroscopy Phenylacetate catabolism in Escherichia coli Wael Ismail 1 , Magdy El-Said Mohamed 1 , Barry L. Wanner 2 , Kirill A. Datsenko 2 , Wolfgang Eisenreich 3 , Felix Rohdich 3 , Adelbert Bacher 3 and Georg Fuchs 1 1 Mikrobiologie, Institut fu ¨ r Biologie II, Universita ¨ t Freiburg, Germany; 2 Department of Biological Sciences, Purdue University, West Lafayette, IN, USA; 3 Lehrstuhl fu ¨ r Organische Chemie und Biochemie, Technische Universita ¨ tMu ¨ nchen, Germany Aerobic metabolism of phenylalanine in most bacteria proceeds via oxidation to phenylacetate. Surprisingly, the further metabolism of phenylacetate has not been elucida- ted, even in well studied bacteria such as Escherichia coli. The only committed step is the conversion of phenylacetate into phenylacetyl-CoA. The paa operon of E. coli encodes 14 polypeptides involved in the catabolism of phenylacetate. We have found that E. coli K12 mutants with a deletion of the paaF, paaG, paaH, paaJ or paaZ gene are unable to grow with phenylacetate as carbon source. Incubation of a paaG mutant with [U- 13 C 8 ]phenylacetate yielded ring-1,2- dihydroxy-1,2-dihydrophenylacetyl lactone as shown by NMR spectroscopy. Incubation of the paaF and paaH mutants with phenylacetate yielded D3-dehydroadipate and 3-hydroxyadipate, respectively. The origin of the carbon atoms of these C 6 compounds from the aromatic ring was shown using [ring- 13 C 6 ]phenylacetate. The paaG and paaZ mutants also converted phenylacetate into ortho-hydroxy- phenylacetate, which was previously identified as a dead end product of phenylacetate catabolism. These data, in conjunction with protein sequence data, suggest a novel catabolic pathway via CoA thioesters. According to this, phenylacetyl-CoA is attacked by a ring-oxygenase/reductase (PaaABCDE proteins), generating a hydroxylated and reduced derivative of phenylacetyl-CoA, which is not re-oxidized to a dihydroxylated aromatic intermediate, as in other known aromatic pathways. Rather, it is proposed that this nonaromatic intermediate CoA ester is further metabo- lized in a complex reaction sequence comprising enoyl-CoA isomerization/hydration, nonoxygenolytic ring opening, and dehydrogenation catalyzed by the PaaG and PaaZ proteins. The subsequent b-oxidation-type degradation of the resulting CoA dicarboxylate via b-ketoadipyl-CoA to succinyl-CoA and acetyl-CoA appears to be catalyzed by the PaaJ, PaaF and PaaH proteins. Keywords: aromatic metabolism; phenylacetate; phenyl- acetyl-CoA oxygenase; phenylalanine metabolism. The aerobic catabolism of aromatic compounds in micro- organisms has been studied in some detail. Hayaishi [1] was the first to show the formation of hydroxylated products by mono-oxygenases and dioxygenases. The resulting aromatic vicinal dihydroxy derivatives can be cleaved by dioxygen- ases between the hydroxy groups (ortho cleavage) or adjacent to one hydroxy group (meta cleavage). The bacterial metabolism of phenylalanine generally proceeds via phenylacetate. Surprisingly, phenylacetate metabolism is still largely unknown, even in Escherichia coli, despite many efforts [2–10]. The genomes of several proteobacteria, including E. coli, Pseudomonas putida and Azoarcus evansii, contain clusters of 11–16 genes believed to be involved in the catabolism of phenylacetate [2,5,6,9,10] (Fig. 1). The paaK gene of E. coli and orthologous genes in other bacteria specify a CoA ligase catalysing the conversion of phenylacetate into phenylacetyl-CoA, which is the first and only committed intermediate in the catabolic pathway [3,4,7]. The use of substrate CoA thioesters is unprecedented in aerobic aromatic metabolism, which may explain why this pathway has been overlooked. Recombinant expression of the paaABCDEK genes allows an E. coli Wmutant lacking the paa genes to convert phenylacetate into 2-hydroxyphenylacetate but not to catabolize phenyl- acetate. Because 2-hydroxyphenylacetate cannot be meta- bolized by E. coli, it is believed to be a dead end product of phenylacetate metabolism [5,6]. This paper describes catabolic studies on mutants of E. coli K12 using NMR and multiply 13 C-labeled phenyl- acetate samples. This has led to the proposal of a new phenylacetate catabolic pathway, allowing putative func- tions to be assigned to most of the paa catabolic genes. Materials and methods Materials [U- 14 C]Phenylalanine was from Amersham-Pharmacia Bio- tech (Freiburg, Germany), and [1- 14 C]phenylacetic acid was from American Radiolabeled Chemicals (Ko ¨ ln, Germany). L -[U- 13 C 9 ]phenylalanine and L -[ring- 13 C 6 ]phenylalanine were purchased from Cambridge Isotope Laboratories Correspondence to G. Fuchs, Mikrobiologie, Institut Biologie II, Scha ¨ nzlestrasse 1, D-79104 Freiburg, Germany. Fax: + 49 761 203 2626, Tel.: + 49 761 203 2649, E-mail: Georg.Fuchs@biologie.uni-freiburg.de, Web site: http://www.biologie.uni-freiburg.de Abbreviation: HMQC, heteronuclear multiple-quantum correlation. (Received 14 March 2003, revised 29 April 2003, accepted 22 May 2003) Eur. J. Biochem. 270, 3047–3054 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03683.x (Andover, MA, USA). Oligonucleotides were from IDT (Coralville, IA, USA). Qiaprep Spin Miniprep Kit (for plasmid DNA isolation) and Qiaquick Gel Extraction Kit (for gel purification of DNA fragments and PCR products) were supplied by Qiagen (Hilden, Germany). Synthesis of 13 C-labeled phenylacetate Reaction mixtures containing 0.1 M phosphate, pH 6.5, 40 mg L -[U- 13 C 9 ]phenylalanine or L -[ring- 13 C 6 ]phenyl- alanine, 740 kBq [U- 14 C 9 ]phenylalanine, and 2 U L -amino acid oxidase (Fluka, Neu-Ulm, Germany) in a total volume of 60 mL were incubated at 37 °C for 3 h. A solution (12 mL) containing 1% H 2 O 2 in 6 M NaOH was added. After 5 min at room temperature, the pH was adjusted to 3.0 by the addition of HCl, and the mixture was extracted three times with equal volumes of ethyl acetate. The solvent was evaporated under reduced pressure. The yield was 50%. Bacteria, media and growth conditions Wild-type E. coli K12 (DSM 498) and E. coli BW25113 [11] were grown aerobically with phenylacetate (5 m M )or glycerol (10 m M ) as carbon and energy source in a phosphate-buffered mineral salt medium supplemented with vitamins, as described previously [7,10]. Cultures were incubated at 37 °C with shaking (180 r.p.m.). Cells were harvested in the exponential growth phase (A 578 0.4–0.6) by centrifugation at 10 000 g for 10 min. Growth of E. coli K12 with 2-hydroxyphenylacetate (at 1 and 5 m M concen- trations) was tested in the same medium. The ability of different mutants to grow on phenylacetate was checked in the same medium containing 5 m M phenylacetate and 1 m M isopropyl thio-b- D -galactoside (to induce expression of the genes located downstream of the deleted gene). Construction of mutants Genes paaF, paaH, paaI, paaJ and paaZ were targeted in E. coli K12 as previously described [11]. All gene mutations were verified by PCR and sequencing (Microbiology and Molecular Genetics Core Facility at Harvard Medical School, Boston, MA, USA). Deletion of the paaG gene required promoter fusion in E. coli BW25113 [11] and then transfer of the mutation into E. coli K12byP1using published procedures [12]. Complementation assays The paaH gene of wild-type E. coli K12 was amplified and cloned into the expression vector pLA35 [13] using standard protocols [14,15]. Integration of the recombinant vector into the chromosome of the paaH mutant was carried out as described [13]. Metabolic transformation of 13 C-labeled phenylacetate Mutants were grown in phosphate-buffered mineral salt medium supplemented with vitamins [10] containing 10 m M glycerol, 1 m M isopropyl thio-b- D -galactoside, and 5 m M phenylacetate. Cells were harvested by centrifugation, washed with 30 m M NH 4 HCO 3 , pH 7.3, and resuspended in the same buffer (1.5 g cells per 15 mL). [U- 13 C 8 ]Pheny- lactate or [ring- 13 C 6 ]phenylacetate (6 mg), 111 kBq [1- 14 C]phenylacetate or [U- 14 C]phenylacetate, and glycerol (final concentration, 0.3 m M ) were added to 15 mL of cell suspension. The mixtures were incubated at 30 °C under shaking (180 r.p.m.). Samples were retrieved at intervals and centrifuged. The supernatants were analyzed by HPLC and lyophilized. HPLC A reversed-phase C 18 column (RP-C 18 ,Grom-Siloctadecyl silane-4 hydrophilic; end capped; particle size, 5 lm; 120 · 4 mm; Grom, Herrenberg, Germany) was equili- brated with 50 m M potassium phosphate, pH 4, containing 8% (v/v) acetonitrile for 15 min and then developed with a linear gradient of 8–40% acetonitrile in the same buffer for 5min.The flowratewas 1mLÆmin )1 . The effluent was monitored photometrically (260 nm) and by online liquid- scintillation counting. The retention times for phenylacetate, phenylacetyl-CoA and 2-hydroxyphenylacetate were 20, 21 and 12 min, respectively. NMR spectroscopy The lyophilized samples were dissolved in 0.5 mL of a 1 : 1 (v/v) mixture of D 2 O and methanol-d 4 .1D 13 C-NMR spectra and 2D INADEQUATE 1 spectra were measured at 125.6 MHz using a Bruker DRX 500 spectrometer equipped with a dual 13 C/ 1 H probe head. 2D gradient-enhanced HMQC and HMQC-COSY experi- ments were performed with an AV 500 spectrometer equipped with a triple-resonance inverse cryo probe head. Acquisition and processing parameters were according to standard Bruker software ( XWINNMR ). Spectral simula- tions were performed with NMRSIM software (Bruker, Karlsruhe, Germany). Results Mutagenesis and mutant phenotype Mutants with deletions of the genes paaF, paaH, paaI, paaJ and paaZ (Fig. 1) were constructed using PCR fragments as described elsewhere [11]. For unknown reasons, it was not possible to delete the paaG gene directly in E. coli K12. Therefore, we constructed the Fig. 1. Organization of paa gene cluster in E. coli which codes for aerobic phenylacetate metabolism. The only proven function of a catabolic gene product is that of PaaK, phenylacetate-CoA ligase. The following functions are putative: PaaABCD, phenylacetyl-CoA oxygenase; PaaE, oxygenase reductase; PaaF, enoyl-CoA hydratase/ isomerase; PaaG, enoyl-CoA hydratase/isomerase; PaaH, 3-hydroxy- acyl-CoA dehydrogenase; PaaI, unknown, low similarity to thioest- erase; PaaJ, b-ketothiolase; PaaZ, unknown, putative ring-cleavage enzyme, aldehyde dehydrogenase domain (N-terminus), MaoC-like protein domain (C-terminus). PaaX and PaaY may be involved in regulation. 3048 W. Ismail et al.(Eur. J. Biochem. 270) Ó FEBS 2003 deletion mutant with promoter fusion in E. coli BW25113 and transferred it into E. coli K12byP1[12].Toavoid polar effects, deletions were either in-frame or the lacUV5 promoter was introduced to drive expression of the respective downstream gene(s). Chromosomal regions modified were verified by PCR amplification and sequencing. Except for the paaI mutant, all mutants were unable to grow on phenylacetate as sole carbon and energy source in mineral salt medium within 48 h. A plasmid carrying an intact paaH gene restored the growth of the DpaaH mutant on phenylacetate as sole carbon and energy source. All wild-type and mutant strains grew on succinate, but none grew on adipate. Transformation of labeled phenylacetate into labeled products by the mutants Mutant cell suspensions were incubated in buffer contain- ing traces of [U- 14 C]phenylacetate, 1–3 m M [U- 13 C 8 ]phenyl- acetate or [ring- 13 C 6 ]phenylacetate, and 0.3 m M glycerol. Isopropyl thio-b- D -galactoside was added for expression of downstream genes, as appropriate. The expected con- sumption of phenylacetate was confirmed by HPLC of the culture fluid using online liquid-scintillation counting for detection (data not shown). Metabolites in the culture supernatants derived from [ 14 C/ 13 C]phenylacetate were detected by HPLC. They were subsequently analyzed by 13 C-NMR spectroscopy with high selectivity and sensitivity because of the multiple 13 C-labeling. In 1D 13 C-NMR spectra, the signals of these metabolites appeared as complex multiplets because of multiple 13 C 13 C coupling. Carbon–carbon connectivities were gleaned from 2D spectra of the totally 13 C-labeled metabolites. Metabolites from paaG and paaZ mutants 13 C-NMR spectra of supernatants from paaG and paaZ mutants were dominated by eight 13 C-coupled signals. Six signals in the range 117–157 p.p.m. (Table 1) suggested a benzenoid ring with one downfield-shifted signal (157.2 p.p.m., Fig. 2A), indicating a phenolic hydroxy group. Two signals (40.9 and 179.1 p.p.m.) suggested the presence of a CH 2 COOH side chain. The carbon connectivities were gleaned from correlation patterns detected by HMQC-COSY experiments (Table 1). Numerical simulation was used for an in depth study of the complex coupling patterns (for an example, see Fig. 2B). On this basis, the structure of the major metabolite was assigned as 2-hydroxy[U- 13 C 8 ]phenylace- tate (5,R¼ H, Fig. 3). Table 1. NMR data of 13 C-labeled products from [U- 13 C 8 ]phenylacetate in paa gene-deficient mutants of E. coli. nd, Not determined. Position Chemical shifts (p.p.m.) Coupling constants, J CC (Hz) Correlation patterns d 13 C d 1 H HMQC HMQC-COSY 2-Hydroxy[U- 13 C 8 ]phenylacetate (5) 1 179.07 52.1(2) 2 40.86 3.52 52.4(1), 43.3(3) 2 2 3 124.54 43.6(2), 65 (4, 8) 4 157.22 65.0, 66.8(3,5), 9.0(7), 1.6(8,6) 5 117.15 6.76 5 5, 6 6 128.91 7.04 55.8, 57.5(5, 7), 6.8 6 6, 5, 7 7 120.57 6.74 56.2(6, 8), 8.7(4) 7 7, 6, 8 8 131.69 7.05 57(3, 7) 8 8, 7 ring-1,2-Dihydroxy-1,2-dihydro[U- 13 C 8 ]phenylacetate or its c-lactone (1) 1 177.94 50.2(2) 2 41.62 2.91, 2.83 50.2(1), 37.8(3) 2, 2¢ 2, 2¢ 3 73.46 nd 6 126.25 6.11 67.7(7), 50.2(5), 5.1 6 6, 5, 7 7 122.14 5.86(d) 67.3(6), 45.1(8), 6.9 7 7, 8, 6 4 129.33 5.91(d) 67.0(5), 46.6(3), 6.9 4 4, 5 5 123.10 5.99 nd 5 5, 4, 6 8 85.54 5.23 44.9(7), 36.5(3), 4.2(6) 8 8, 7 cis-D3-Dehydro[U- 13 C 6 ]adipate (3) 1/6 nd 2/5 34.76 2.95 51.6(1/6), 42.5(3/4) 2/5 2/5, 3/4 3/4 125.31 5.60 3/4 3/4, 2/5 3-Hydroxy[U- 13 C 6 ]adipate (4) or its lactone 1 176.82 50.2(2) 2 42.36 2.52 52.0(1), 39.9(3), 1.2 (6,4) 2 2, 3 3 79.81 4.86 39.8(2), 32.1(4), 4.2(6) 3 3, 2, 4 4 27.01 2.34, 1.88 32.1(3.5), 3.2(1) 4, 4¢ 4, 5, 3 5 28.53 2.59, 2.54 48.0(6), 32.7(4) 5, 5¢ 5, 4 6 181.17 48.0(5), 4.2(3), 1.2 (4,2) Ó FEBS 2003 Bacterial phenylacetate and phenylalanine metabolism (Eur. J. Biochem. 270) 3049 Besides the signal set described above, an additional set of eight minor signals was detected in the sample from the paaG-deficient mutant. Their relative intensities were  2% compared with the signals of 2-hydroxy- phenylacetate (Table 2). Four signals had chemical shifts typical of olefinic carbon atoms (122.1, 123.1, 126.5 and 129.3 p.p.m.; Table 1); they were all correlated to directly attached protons as shown by 2D HMQC spectroscopy (Fig. 4). Carbon connectivities established on the basis of the coupling constants and the correlation pattern in HMQC-COSY experiments (Table 1) identified the com- pound as a conjugated cyclohexadiene derivative. The carbon atoms resonating at 85.5 and 73.5 p.p.m. are likely to carry heteroatom substituents. The carbon atom resonating at 85.5 p.p.m. has an attached proton ( 1 H- NMR signal at 5.23 p.p.m.), whereas the signal at 73.46 p.p.m. represents a quaternary carbon atom. On this basis, the structure can be assigned as ring-1,2- dihydroxy-1,2-dihydrophenylacetyl lactone (1, Fig. 3; note the different carbon numbering of 1 in the figure). A structurally similar lactone (2, Fig. 3) has been found in the Caribbean sponge, Aplysina cauliformis [16]. Metabolites of paaF and paaH mutants A set of six intense 13 C-NMR signals was detected in the supernatants of the paaF and paaH mutants which had been incubated with [U- 13 C 8 ]phenylacetate. Two carbon atoms resonated at chemical shifts typical of carboxylic groups (181.2 and 176.8 p.p.m.). Four signals were detected in the region for aliphatic carbons; one of these had a chemical shift (79.8 p.p.m.) characteristic of a carbon atom carrying an OH or OR residue. The detailed analysis of the 13 C spin system by simulation (Table 1) identified the metabolite as 3-hydroxy[U- 13 C 6 ]adipate (4) or the cognate lactone (Fig. 3). To identify the precursor carbon atoms that are lost in the formation of 4, we incubated cells of the paaH-deficient mutant with [ring- 13 C 6 ]phenylacetate. The same set of six coupled signals as in the experiment with [U- 13 C 8 ]phenyl- acetate was observed. As an example, the signals of C6 and C1 from the two different experiments are displayed in Fig. 2. 13 C-NMR signal of ring-C2 of 2-hydroxy[U- 13 C 8 ]phenylacetate. (A) Detected in the experiment with the paaG-deficient mutant of E. coli; (B) simulated signals on the basis of the chemical shifts and coupling constants summarized in Table 1. Fig. 3. Compounds observed in supernatants of E. coli mutants given [U- 13 C 8 ]phenylacetate (cf. Table 2). Bold lines connect 13 C-labeled carbon atoms. Table 2. Product patterns observed in supernatants of paa gene-deficient mutants of E. coli given [U- 13 C 8 ]phenylacetate. Relative amounts of product are shown estimated from 13 C-NMR signal intensities referred to an arbitrary value of 100 for the major product. Mutant Product 134 5 paaZ – – – 100 paaG 2 – – 100 paaH – – 100 20 paaF – 3 100 3 3050 W. Ismail et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Fig. 5. The data show that the formation of 3-hydroxy- adipate (4) proceeds by the loss of the acetyl side-chain carbon atoms of phenylacetate. A set of two relatively weak signals ( 3% signal intensity compared with the signals of 3-hydroxyadipate) was observed in the experiments with the paaF-deficient mutant. One signal had a chemical shift typical of olefinic carbon atoms (125.3 p.p.m.), and the other one (34.76 p.p.m.) was found in the frequency range typical of aliphatic carbon atoms. The coupling constants of the latter signal indicated an attached 13 C-labeled olefinic atom (42.5Hz),aswellasa 13 C-labeled carboxylate group (51.6 Hz); the expected signal for the carboxylate carbon was not observed directly because of signal overlap. The detailed analysis of the coupling pattern by spectral simulation suggests an inherently symmetrical metabolite (Fig. 6). All spectroscopic data, as well as a comparison with published NMR data for adipate derivatives [17], support the assignment of the minor metabolite as cis-D3- dehydro[U- 13 C 6 ]adipate (3,Fig.3). The supernatant of the paaF mutant also showed the signal set of 2-hydroxyphenylacetate at a relative intensity of  3% compared with that of the major metabolite 4 (Table 2). Discussion We have shown in this work that E. coli K12 mutants with deletion of the paaF, paaG, paaH, paaJ,orpaaZ gene are unable to use phenylacetate as carbon source. In contrast, deletion of the paaI gene did not impair growth on phenylacetate and phenylacetate consumption. This sug- gests that the paaI gene product is not essential for phenylacetate metabolism or can be substituted by the translation product of another similar gene that is consti- tutively expressed. The function of the PaaI protein is not known. Incubation of the paaF, paaG, paaH or paaZ mutant with multiply 13 C-labeled phenylacetate yielded several uni- formly 13 C-labeled metabolites (Fig. 3, Table 2), which were identified by 13 C-NMR spectroscopy without prior purification. This approach was possible by the sensitivity and selectivity enhancement due to multiple 13 C-labeling of the precursor. The deconvolution of the 13 C spin systems by heteronuclear correlation spectroscopy was greatly facilita- ted by the contiguous 13 C labeling of the metabolites. Moreover, this experimental approach minimizes the risk of artefact formation by decomposition of chemically unstable metabolites. The mutant phenotypes and the observed products (Table 2) led us to propose a working hypothesis for further studies (Fig. 7). The conversion of phenylacetate [6] into phenylacetyl-CoA [7] is established (see the Intro- duction). Although details of the pathway remain unknown, the available data cannot be reconciled with conventional Fig. 4. Part of an HMQC spectrum of ring-1,2-dihydroxy-1,2- dihydro[U- 13 C 8 ]phenylacetate (1) or its c-lactone. Fig. 5. 13 C-NMR signals of 3-hydroxy- [U- 13 C 6 ]adipate (4) or its lactone observed in the supernatant of a paaH-deficient mutant of E. coli given (A) [U- 13 C 8 ]phenylacetate or (B) [rin g- 13 C 6 ]phenylacetate. 13 C coupling patterns are indicated. Ó FEBS 2003 Bacterial phenylacetate and phenylalanine metabolism (Eur. J. Biochem. 270) 3051 principles of aromatic metabolism. Therefore we postulate a new strategy in which CoA thioesters are used throughout the pathway and the ring is cleaved nonoxygenolytically at the stage of a nonaromatic CoA ester intermediate. Moreover, we postulate the presence of an unprecedented phenylacetyl-CoA oxygenase/reductase. One major finding of this study was that both paaG and paaZ mutants accumulated C 8 compounds rather than C 6 compounds. This suggests a role for the PaaG and PaaZ proteins early in the pathway. The other major finding is that C 6 dicarboxylic acids formed later in the pathway are derived from the aromatic ring carbons of phenylacetyl- CoA. This indicates that removal of the original C 2 side chain yields a C 6 intermediate via an open-chain C 8 intermediate. The formation of 1 by the paaG mutant may suggest that phenylacetyl-CoA is attacked by phenyl- acetyl-CoA (di)oxygenase/reductase (PaaABCDE), adding molecular oxygen and reducing the intermediate, possibly to a cis-dihydrodiol derivative of phenylacetyl-CoA [8]. Fur- ther metabolism of this intermediate appears to be blocked in the paaG mutant, suggesting that the PaaG protein uses this nonaromatic product of the oxygenase/reductase as substrate. Artificial formation of the lactone 1 from 8 may be facilitated by the CoA thioesterification of the carboxy group. Likewise, formation of 5 from 8 by various mutants (Table 2) may be explained by water elimination from accumulated 8 resulting in re-aromatization and subsequent enzyme-catalyzed or spontaneous hydrolysis of the thioester bond. Wild-type and mutant E. coli strains are unable to use 2-hydroxyphenylacetate as carbon source, probably because the compound and its CoA derivative are dead end products rather than intermediates of the pathway under study. Sequence similarity (40% identity and 11% of conserva- tive exchange by comparison with ChcB of Streptomyces collinus) of the PaaG protein with members of the enoyl- CoA hydratase/isomerase family [18,19] suggests a similar function. ChcB (D 3 ,D 2 -enoyl-CoA isomerase) catalyzes the isomerization of cyclohex-1-ene-1-carbonyl-CoA and cyclo- hex-2-ene-1-carbonyl-CoA. Thus, ring opening may be preceded by a reversible PaaG-catalyzed D 3 ,D 2 isomeriza- tion of double bonds in 8 and/or addition of water. The enzyme may even play a role in C–C cleavage as Fig. 6. 13 C-NMR signals of cis-D3- dehydro[U- 13 C 6 ]adipate (3). (A) Detected in the experiment with the paaF-deficient mutant of E. coli; (B) simulated for the AA¢MM¢XX¢ spin system using the chemical shifts and coupling constants summarized in Table 1. Fig. 7. Proposed outline of the pathway of aerobic metabolism of phenylacetate in E. coli. For details see text. 3052 W. Ismail et al.(Eur. J. Biochem. 270) Ó FEBS 2003 C–C-cleaving enoyl-CoA hydratases are known [20]. Re-aromatization of the product of phenylacetyl-CoA oxygenase/reductase 8 by a cis-diol dehydrogenase, as is common in aromatic pathways, is unlikely because no such gene could be found in the paa gene cluster. Formation of the (di)hydroxylated and reduced deriva- tive 8 from 7 is proposed to be catalysed by a protein complex specified by the paaABCDE genes. Based on sequence similarities, the paaABCDE genes may jointly specify a five-subunit oxygenase/reductase enzyme com- plex using phenylacetyl-CoA as substrate [2,5,6,9,10]. paaABCDE mutants cannot grow with phenylacetate but convert it into phenylacetyl-CoA by the catalytic action of the PaaK protein [5,6]. Putative orthologs of paaABCD genes are found in numerous proteobacteria believed to catalyse the degradation of phenylacetate, but only the PaaE protein shows similarity to enzymes of micro-organisms outside that group. PaaABC may func- tion as terminal oxygenase, and the small protein PaaD may be required as an additional component, as is found in some oxygenases. The similarity of the PaaE protein to various oxidoreductases (2Fe-2S ferredoxin flavo- proteins) indicates that it functions as a reductase which delivers electrons from NAD(P)H to the oxygenase components. Tentatively, we suggest that the ring opening affords an aldehyde which is converted into a carboxylic acid by the PaaZ protein. The N-terminal part of PaaZ is similar to various aldehyde dehydrogenases, e.g. succinate semialde- hyde dehydrogenase GabD of E. coli (23.7% identity and 12.7% conserved exchange). The C-terminal part shows similarity to MaoC-like proteins, the function of which are unknown. It was recently shown that a mutant of Azoarcus evansii,inwhichthepaaZ ortholog pacL was disrupted by integration of a resistance cassette, excreted 2,4,6-cyclo- heptatriene-1-one (9); the pacL gene in this organism contains only the aldehyde dehydrogenase domain [21]. One can envisage elimination of water from 8 yielding the corresponding conjugated C 8 triene, which becomes rearranged to 2,4,6-cycloheptatriene-1-one with the release of CoASH and a C 1 unit. paaF and paaH mutants transform phenylacetate into the open-chain dicarboxylic acid derivatives 3 and 4, respect- ively. Consequently, the PaaF and PaaH proteins appear to catalyse reactions in the downstream part of phenylacetate degradation. In the absence of the PaaF and PaaH proteins, catabolism of phenylacetate is proposed to be terminated at the level of 10 or its dehydration product, which are converted into 4 and 3, respectively, by spontaneous or enzyme-catalyzed hydrolysis. The PaaH protein has sequence similarity to 3-hydroxy- acyl-CoA dehydrogenases. This suggests that it catalyzes the dehydrogenation of 3-hydroxyadipyl-CoA [10], affording 3-ketoadipyl CoA [11] which could then be thiolytically cleaved by the PaaJ protein with formation of acetyl-CoA and succinyl-CoA [12]. The PaaJ protein is similar to b-ketoadipyl-CoA thiolases. No orthologs of the PaaH or PaaJ protein appear to exist in the E. coli genome. The PaaF protein has sequence similarity (44% identity and 11% conserved exchange) to BadK of Rhodopseudo- monas palustris and to proteins of the enoyl-CoA hydratase (isomerase) family (crotonase family) [18,19]. BadK catalyzes the reversible addition of water to cyclohex-1- ene-1-carbonyl-CoA forming 2-hydroxycyclohexane-1-car- bonyl-CoA. Many enoyl-CoA hydratases of this type have cis-D 3 -trans-D 2 -enoyl-CoA isomerase activity, in addition to the enoyl-CoA hydratase activity [18,19]. The accumulated products 3, 4 and 5 are devoid of CoA moieties, which they have probably lost by enzyme- catalysed or nonenzymatic hydrolysis of catabolic inter- mediates. Enzyme-catalysed hydrolysis of CoA derivatives may serve as a salvage reaction to avoid the breakdown of intermediary metabolism due to depletion of the CoA pool in situations where CoA derivatives cannot be metabolized further. Interestingly, the PaaI protein shows low similarity to thioesterases, and cell extracts are notorious for CoA thioesterase activity, which greatly impairs enzyme studies using CoA thioesters. paaI mutants could still grow with phenylacetate, indicating that the PaaI protein does not serve an essential function in the pathway itself. The postulated phenylacetate pathway has considerable similarity to the catabolism of benzoate in Azoarcus evansii, Geobacillus sp., and probably other bacteria [20,22]. In both cases, CoA thioesters are used throughout the pathway. The aromatic substrate is first transformed into the CoA thio- ester, followed by ring oxygenation, isomerization, nonoxy- genolytical ring cleavage, and subsequent b-oxidation to b-ketoadipyl-CoA. This intermediate is finally cleaved into acetyl-CoA and succinyl-CoA, as in the conventional b-ketoadipate pathway. A variant of this principle exists in the catabolism of 2-aminobenzoate in Azoarcus evansii and related bacteria. A mono-oxygenase/reductase rather than a dioxygenase/reductase transforms 2-aminobenzoyl-CoA into a nonaromatic monohydroxylated product [20,23,24]. Acknowledgements We thank W. Buckel, University of Marburg, for suggesting the enzymatic conversion of phenylalanine into phenylacetate, and R. Bru ¨ ckner, University of Freiburg, for initial synthesis of [ 13 C]phenyl- acetate. This work was financially supported by the Deutsche Forschungsgemeinschaft (Bonn), the Fonds der Chemischen Industrie (Frankfurt), the Graduiertenkolleg ÔBiochemie der EnzymeÕ (University of Freiburg) to G.F., and the U. S. National Science Foundation to B.L.W. References 1. Hayaishi, O. (1994) Tryptophan, oxygen, and sleep. Annu. Rev. Biochem. 63, 1–24. 2.Diaz,E.,Ferrandez,A.,Prieto,M.A.&Garcia,J.L.(2001) Biodegradation of aromatic compounds by Escherichia coli. Microbiol. Mol. Biol. Rev. 65, 523–568. 3. Martinez-Blanco, H., Reglero, A., Rodriguez-Aporicio, L.B. & Luengo, J.M. (1990) Purification and biochemical characterization of phenylacetyl-CoA ligase from Pseudomonas putida.Aspecific enzyme for the catabolism of phenylacetic acid. J. Biol. Chem. 265, 7084–7090. 4. Vitovski, S. (1993) Phenylacetate-CoA ligase is induced during growth on phenylacetic acid in different bacteria of several genera. FEMS Microbiol. Lett. 108, 1–6. 5. Ferrandez, A., Minamberes, B., Garcia, B., Olivera, E.R., Luengo, J.M., Garcia, J.L. & Diaz, E. (1998) Catabolism of phenylacetic acid in Escherichia coli. J. Biol. Chem. 273, 25974–25986. Ó FEBS 2003 Bacterial phenylacetate and phenylalanine metabolism (Eur. J. Biochem. 270) 3053 6. Olivera, E.R., Minamberes, B., Garcia, B., Muniz, C., Moreno, J.M., Ferrandez, A., Diaz, E., Garcia, J.L. & Luengo, J.M. (1998) Molecular characterization of the phenylacetic acid catabolic pathway in Pseudomonas putida U: the phenylacetyl-CoA cata- bolon. Proc.Natl.Acad.Sci.USA95, 6419–6424. 7. Mohamed, M.E. (2000) Biochemical and molecular characteriza- tion of phenylacetate-coenzyme A ligase, an enzyme catalyzing the first step in aerobic metabolism of phenylacetic acid in Azoarcus evansii. J. Bacteriol. 182, 286–294. 8. Garcia, B., Olivera, E.R., Minamberes, B., Carnicero, D., Muniz, C., Naharro, G. & Luengo, J.M. (2000) Phenylacetyl-CoA is the true inducer of the phenylacetic acid catabolism pathway in Pseudomonas putida U. Appl. Environ. Microbiol. 66, 4575–4578. 9. Luengo, J.M., Garcia, J.L. & Olivera, E.R. (2001) The phenyl- acetyl-CoA catabolon: a complex catabolic unit with broad biotechnological applications. Mol. Microbiol. 39, 1434–1442. 10. Mohamed, M.E., Ismail, W., Heider, J. & Fuchs, G. (2002) Aerobic metabolism of phenylacetic acids in Azoarcus evansii. Arch. Microbiol. 178, 180–192. 11. Datsenko, K.A. & Wanner, B.L. (2000) One-step inactivation of chromosomal genes in Escherichia coli K12 using PCR products. Proc.NatlAcad.Sci.USA97, 6640–6645. 12. Wanner, B.L. (1994) Gene expression in bacteria using TnphoA and TnphoA¢ elements to make and switch phoA gene, lacZ. (op), and lacZ. (pr) fusions. In Methods in Molecular Genetics (Adolph, K.W., ed.), Vol. 3, pp. 291–310. Academic Press, Orlando. 13. Haldimann, A. & Wanner, B.L. (2001) Conditional-replication, integration, excision, and retrieval plasmid-host systems for gene structure-function studies of bacteria. J. Bacteriol. 183, 6384–6393. 14. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Plainview, NY. 15. Ausubel, F.M., Bent, R., Klingston, E., Moore, D.D., Seidman, J.G.,Smith,J.A.&Struhl,K.(1987)Current Protocols in Mole- cular Biology. John Wiley & Sons, New York. 16. Makarieva, T.N., Stonik, V.A., Alcolado, P. & Elyakov, Y.B. (1981) Comparative study of the halogenated tyrosine derivatives from Demospongiae (Porifera). Comp. Biochem. Physiol. 68B, 481–484. 17. Perlman, N. & Albeck, A. (2000) Efficient and stereospecific synthesis of (Z)-hex-3-enedioic acid, a key intermediate of Gly-Gly cis olefin isostere. Syn. Commun. 30, 4443–4449. 18. Clark, D.P. & Cronan, J.E. (1996) In Escherichia coli and Sal- monella (Neidhardt, F.C., eds), 2nd edn, pp. 343–357. ASM Press, Washington, DC. 19. Haller,T.,Buckel,T.,Retey,J.&Gerlt,J.A.(2000)Discovering new enzymes and metabolic pathways: conversion of succinate to propionate by Escherichia coli. Biochemistry 39, 4622–4629. 20. Schu ¨ hle, K., Jahn, M., Ghisla, S. & Fuchs, G. (2001) Two similar gene clusters coding for the enzymes of a new type of aerobic 2-aminobenzoate (anthranilate) metabolism in the bacterium Azoarcus evansii. J. Bacteriol. 183, 5268–5278. 21. Rost, R., Haas, S., Hammer, E., Herrmann, H. & Burchhardt, G. (2002) Molecular analysis of aerobic phenylacetate degradation in Azoarcus evansii. Mol. Genet. Genomics 267, 656–663. 22. Zaar, A., Eisenreich, W., Bacher, A. & Fuchs, G. (2001) A novel pathway of aerobic benzoate catabolism in the bacteria Azoarcus evansii and Bacillus stearothermophilus. J. Biol. Chem. 276, 24997–25004. 23. Hartmann, S., Hultschig, C., Eisenreich, W., Fuchs, G., Bacher, A. & Ghisla, S. (1999) NIH shift in flavine dependent mono- oxygenation: mechanistic studies with 2-aminobenzoyl-CoA monooxygenase reductase. Proc. Natl Acad. Sci. USA 96, 7831–7836. 24. Torres, R.A. & Bruice, T.C. (1999) Sigmatropic hydrogen migration in the mechanism of oxidation of 2-aminobenzoyl-CoA by 2-aminobenzoyl-CoA monooxygenase/reductase. Proc. Natl. Acad.Sci.USA96, 14748–14752. 3054 W. Ismail et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Functional genomics by NMR spectroscopy Phenylacetate catabolism in Escherichia coli Wael Ismail 1 , Magdy El-Said Mohamed 1 ,. several proteobacteria, including E. coli, Pseudomonas putida and Azoarcus evansii, contain clusters of 11–16 genes believed to be involved in the catabolism of phenylacetate