Catabolism of Phenylacetic Acid in Escherichia coli

The paa cluster of Escherichia coli W involved in the aerobic catabolism of phenylacetic acid (PA) has been cloned and sequenced. It was shown to map at min 31.0 of the chromosome at the right end of the mao region responsible for the transformation of 2-phenylethylamine into PA. The 14 paa genes are organized in three transcription units:paaZ and paaABCDEFGHIJK, encoding catabolic genes; and paaXY, containing thepaaX regulatory gene. The paaK gene codes for a phenylacetyl-CoA ligase that catalyzes the activation of PA to phenylacetyl-CoA (PA-CoA). The paaABCDE gene products, which may constitute a multicomponent oxygenase, are involved in PA-CoA hydroxylation. The PaaZ protein appears to catalyze the third enzymatic step, with the paaFGHIJ gene products, which show significant similarity to fatty acid β-oxidation enzymes, likely involved in further mineralization to Krebs cycle intermediates. Three promoters, Pz, Pa, and Px, driven the expression of genes paaZ, paaABCDEFGHIJK, and paaX, respectively, have been identified. ThePa promoter is negatively controlled by thepaaX gene product. As PA-CoA is the true inducer, PaaX becomes the first regulator of an aromatic catabolic pathway that responds to a CoA derivative. The aerobic catabolism of PA in E. coli represents a novel hybrid pathway that could be a widespread way of PA catabolism in bacteria.

phenyl substituent, and some of their metabolites (1)(2)(3). The aerobic catabolism of these aromatic compounds by E. coli could occur close to the epithelial cells in the guts of warmblooded animals, as well as in soil, sediment, and water once E. coli is excreted from its intestinal residence (4). The ability of E. coli to mineralize 3-and 4-hydroxyphenylacetic acids (5), 3-phenylpropionic, 3-(3-hydroxyphenylpropionic), and 3-hydroxycinnamic acids (2,6), and phenylacetic acid (2, 7) has been reported previously. Recently, the molecular characterization of these catabolic pathways, with the only exception of that for PA degradation, has been carried out (1, 8 -11), demonstrating that E. coli is endowed with its own set of genes and enzymes for the catabolism of aromatic compounds, and that they are similar to those of other microorganisms more relevant in the environment such as bacteria of the genus Pseudomonas.
Although PA is a common source of carbon and energy for a wide variety of microorganisms, the bacterial catabolism of this natural aromatic compound is still poorly understood (12,13). Earlier reports suggested that aerobic PA catabolism implicated the typical initial attack by hydroxylation of the aromatic ring with the formation of the corresponding 2,5-or 3,4-dihydroxyphenylacetate as intermediates (13). However, much of this evidence was circumstantial, and none of the typical aerobic routes that could explain PA degradation were responsible of this catabolism in different PA-degrading bacteria (13,14). According to these data, it has been recently shown that Pseudomonas putida U mineralizes PA aerobically through a novel catabolic pathway, which does not follow the conventional routes for the aerobic catabolism of aromatic compounds and whose first step is the activation of PA to phenylacetylcoenzyme A (PA-CoA) by the action of a PA-CoA ligase (12,15). In this sense, the participation of a PA-CoA ligase in the aerobic catabolism of PA has been also inferred from its specific induction during growth on PA of different bacterial strains (13,16).
Here we present the cloning, genetic characterization, mechanism of regulation, and a partial biochemical characterization of the PA biodegradation pathway from E. coli W. This work reveals that the PA degradation in E. coli follows an unusual route for the aerobic catabolism of aromatic compounds, which involves CoA derivatives. With the molecular characterization of the paa-encoded pathway, all aromatic catabolic routes so far reported in E. coli are now described at the molecular level.
Tn1000 Transposition Mutagenesis of paa Genes-Transposition with Tn1000 was carried out according to the previously described method (30). The pAAD plasmid was mutagenized with Tn1000 by selecting for conjugative transfer of this plasmid from the recA F ϩ donor strain E. coli MG1063 to the F Ϫ recipient strain E. coli CC118. Before mating, both parents were grown at 37°C, without shaking, to an optical density at 600 nm of about 0.5. Two milliliters of donor cells and one milliliter of recipient cells were mixed and incubated at 37°C without shaking for 2 h. After addition of 12 ml of LB medium and further incubation with vigorous shaking for 3 h, exconjugants were selected on LB medium containing rifampicin and chloramphenicol. The pAAD derivatives bearing Tn1000 insertions within the paa genes were further analyzed. The primer Tn5900 (5Ј-AAAAGGGGAACT-GAGAGCTC-3Ј) that hybridized with the ␦ terminus of transposon Tn1000 was used to sequence the insertion sites.
Construction of E. coli AF1411 and AF1412 Strains-By means of RP-4 mediated mobilization, the plasmids pAFPA1T and pAFPA2T, which contain mini-Tn5 hybrid transposons expressing Pa-lacZ fusions (Fig. 5), were transferred from E. coli S17-1pir into E. coli AF141. Filter matings were performed as described previously (22). Exconjugants containing the lacZ translational fusions stably inserted into the chromosome, E. coli AF1411 and E. coli AF1412, were selected for the transposon marker, kanamycin, on rifampicin-containing LB medium.
Determination of the Transcription Start Sites by Primer Extension-E. coli CC118 cells containing plasmids pAFPA1, pAFPA2, pAFPZ, pAFPX, or pSJ3 (Fig. 5), were grown in minimal medium containing glycerol and PA until the cultures reached an optical density at 600 nm of about 1.0. Total RNA was isolated using the Qiagen RNA/DNA Midi Kit (Qiagen) according to the instructions of the supplier. Primer extension reactions were carried out with the avian myeloblastosis virus reverse transcriptase as described previously (31), using primers LAC-57 (5Ј-CGATTAAGTTGGGTAACGCCAGGG-3Ј, which hybridized at 57 nucleotides downstream of the lacZ start codon) and PA5-4 (5Ј-CGGGCATCCAGTCCTGTGGCTCG-3Ј, which hybridized at 55 nucleotides downstream of the paaA start codon). Products were analyzed on 6% polyacrylamide-urea gels.
Resting Cell Reactions-E. coli W14 cells harboring different pAAD::Tn1000 derivatives were grown in minimal medium containing glycerol and PA to an optical density at 600 nm of about 0.8. Cell cultures were then centrifuged (3,000 ϫ g, 10 min at 20°C), and cells were washed and resuspended in a 0.05 volume of minimal medium. The resting cell reactions were performed in a final volume of 5 ml containing 4.5 ml of M63 minimal medium and 0.5 ml of the cell suspension. The reaction was started by the addition of 1 mM PA, and the tubes were incubated on a rotary shaking platform at a temperature of 30°C. Samples of 0.5 ml were taken at different times and centrifuged (10,000 ϫ g, 5 min) to remove the cells.
To detect the accumulation of intracellular intermediates in PA catabolism, E. coli W14 (pAFK5) and E. coli W14 (pAAD::Tn1000 derivatives) cells were grown in LB medium containing 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside and minimal medium containing glycerol and PA, respectively. When the cultures reached an optical density at 600 nm of about 1.0, cells were centrifuged, washed as described above, and resuspended in 0.005 volume of minimal medium supplemented with 10 mM glucose. The reaction was started by the addition of 50 M [1-14 C]PA (4 Ci/ml) (Sigma) and incubated for 5 min at 30°C. The following treatments were performed at 4°C. Cells were centrifuged, washed with minimal medium containing 10 mM glucose, and resuspended in 0.01 volume of 0.5 M HClO 4 . After vortexing vigorously for 1 min, the sample was frozen and thawed once and then centrifuged (10,000 ϫ g, 5 min). While the supernatant was collected, the sediment was treated again with 0.01 volume of 0.5 M HClO 4 . Finally, the two supernatants were mixed, filtered through cellulose-triacetate membrane filters (10,000 M r ) (Alltech Associates, Inc.), vacuum-dried, and resuspended in 200 l of H 2 O.
Phenylacetyl-Coenzyme A Ligase Assay-E. coli W14 cells containing different pAAD derivatives were grown overnight in minimal medium containing glycerol and vitamin B 12 (500 ng/ml) in the presence or absence of PA, and then diluted into fresh medium to an optical density of about 0.1. Growth was resumed at 30°C until the cultures reached an optical density at 600 nm of about 1.0. Cells were harvested by centrifugation, washed, and resuspended in 0.05 volume of 0.5 M potassium phosphate buffer, pH 8.2, prior to disruption by passage through a French press (Aminco Corp.) operated at a pressure of 20,000 p.s.i. The cell debris was removed by centrifugation at 26,000 ϫ g for 30 min. The clear supernatant fluid was carefully decanted and used as crude extract. Protein concentration was determined by the method of Bradford (32) using bovine serum albumin as standard. Phenylacetyl-coenzyme A ligase was assayed as described previously (33). One unit of enzyme activity is defined as the catalytic activity leading to the formation of 1 nmol of phenylacetylhydroxamate in 1 min at 37°C.
␤-Galactosidase Assay-␤-Galactosidase activities were measured with permeabilized cells from cultures grown to mid-log phase, as described by Miller (23).
Analytical Methods-The N-terminal sequence of PaaK was determined by Edman degradation with a model 477A automated protein sequencer (Applied Biosystems Inc.). The protein was directly electroblotted from a SDS-polyacrylamide gel onto a polyvinylidene difluoride membrane as described previously (18).
The metabolites accumulated in the supernatants of resting cells and in culture supernatants, were analyzed with a Gilson HPLC equipment using a Lichrosphere 5 RP-8 column (150 ϫ 4.6 mm) and an isocratic flow of a 40% methanol-H 2 O mobile phase pumped at a flow rate of 1 ml/min. Peaks with retention times of 5.6 and 10.6 min, corresponding to those of authentic standard 2-HPA and PA, respectively, were monitored at 220 nm.
The intracellular 14 C-labeled metabolites accumulated in resting cell assays were analyzed with the HPLC equipment described above but using an isocratic flow of 0.2 M KH 2 PO 4 (pH 4.2), isopropyl alcohol (92:8, v/v) as mobile phase pumped at a flow rate of 1 ml/min. Samples (400 l) were collected, and 14 C-labeled products were detected in aliquots of the fractions by liquid scintillation counting. Peaks with retention times of 11.2 and 20.2 min, corresponding to those of authentic standard PA and PA-CoA, respectively, were monitored at 220 nm.
The 2-HPA was extracted from culture supernatants with an equal volume of ethyl acetate and dried with sodium sulfate. Samples were derivatized with N, O-bis(trimethylsilyl)trifluoroacetamide and subjected to gas chromatography-mass spectrometry analysis as described elsewhere (34).

Identification of the paa Genes for the Catabolism of PA-We
had recently reported the isolation of an E. coli W mutant, strain W14, unable to grow on PA and 2-phenylethylamine as the sole carbon and energy source (18). A recombinant plasmid, pFA2, which contains a 33.3-kb BamHI DNA insert from the chromosome of the wild-type E. coli W strain (Fig. 1A), had been also constructed and was able to confer to E. coli W14 the ability to grow on either PA or phenylethylamine as the sole carbon source (18). When the HindIII-digested total DNA from E. coli W14 and E. coli C, a strain also unable to grow on PA as the sole carbon source (2), was analyzed by Southern blot using the 33.3-kb BamHI fragment of pFA2 as probe, no hybridization bands were observed (data not shown). These results indicated that E. coli strains W14 and C have a deletion encompassing at least the 33.3-kb DNA fragment cloned in pFA2, and therefore it is likely that they lack the genes involved in PA catabolism (paa genes). Moreover, we have observed that, whereas the E. coli K-12 strains MV1190, C600, TG1, ET8000, W3110, and MG1655 grew on PA, the K-12 strains HB101, DH1, CC118, DH5␣, and JM109 did not grow on this aromatic compound. A Southern blot analysis of the HindIII-digested total DNA from E. coli K-12 PA ϩ (W3110, MG1655, and C600) and PA Ϫ (DH5␣, HB101, and DH1) strains revealed the same pattern of hybridization bands, thus suggesting that the PA Ϫ phenotype in E. coli K-12, in contrast to that in E. coli W14 and E. coli C, could be produced by point mutations or small deletions or insertions. Interestingly, all E. coli PA Ϫ strains were able to grow on this aromatic compound when harboring plasmid pFA2, and the loss of this plasmid after several rounds of cultivation of the recombinant strains in the absence of selective pressure was also accompanied by the loss of the PA ϩ phenotype. As several aromatic catabolic pathways are encoded by plasmids and E. coli W was shown to host cryptic plasmids (35), it was checked whether the paa genes were also located in a plasmid. When a pulse field electrophoresis of unrestricted total DNA from E. coli W was analyzed by Southern blot using the 33.3-kb DNA fragment as probe, hybridization signals were only found in the sample wells (data not shown), thus indicating a chromosomal location (36) of the paa genes.
To localize the paa genes within the cloned 33.3-kb DNA fragment, its physical map was determined and different subclones were constructed (Fig. 1A) and checked for their ability to restore the growth of E. coli W14 on PA-containing minimal medium. Interestingly, although plasmids pFCP142 and pFCH112 (Fig. 1A) did not confer to E. coli W14 the ability to grow on PA, plasmid pAAD restored the growth of strain W14 on this aromatic compound, thus locating the paa genes in a 15.4-kb DNA fragment at the right end of the previously identified mao region (Fig. 1A) responsible of the transformation of 2-phenylethylamine into PA (18,37,38).
Sequencing and Gene Arrangement of the paa Cluster-To genetically characterize the PA catabolic pathway of E. coli W, the 15.4-kb insert of plasmid pAAD (Fig. 1A) was sequenced. The nucleotide sequence of a 14,328-bp region of this insert is shown in Fig. 2. Computer analysis of this sequence revealed the presence of 14 ORFs, all of which are transcribed in the same direction with the sole exception of paaZ (Figs. 1 and 2). Data bases were searched for similar proteins to the paa gene products, and those showing the highest similarity values were then retrieved and compared (Table I). The putative Shine-Dalgarno sequences of paaFGHIJ and paaY overlap the preceding ORFs (Fig. 2), suggesting that the most common mechanism of translational coupling (39) may occur. Downstream of the paaZ, paaK, and paaY genes we found inverted repeat sequences (Fig. 2) predicted to form hairpin loops with ⌬G values of Ϫ13.4, Ϫ25.3, and Ϫ15.7 kcal/mol, respectively, which could act as transcriptional terminators. The GϩC content of the paa cluster averaged 52.5%, a value close to the mean GϩC content of E. coli genomic DNA (51.5%) (40). At the 3Ј-end of the sequenced fragment, a partial ORF corresponded to the 5Ј-end of ydbC (Fig. 2), a gene of unknown function from E. coli K-12 (29,41).
To define the essential genes required for PA catabolism in E. coli W, transposon mutagenesis of the 15.4-kb NotI-DNA cassette in pAAD was carried out. A collection of Tn1000 insertion derivatives was obtained, and to determine the physical location and orientation of each insertion in pAAD::Tn1000 mutant plasmids, DNA from each derivative was isolated and analyzed by cleavage with different restriction enzymes. In most of the cases, the sites of the Tn1000 insertions were sequenced with a primer that hybridized with the ␦ terminus of transposon Tn1000. Twenty different Tn1000 insertions were identified and checked for their ability to avoid the growth on PA of E. coli W14 harboring the corresponding pAAD::Tn1000 derivatives (Fig. 1B). On the basis of these studies, it was shown that only Tn1000 insertions within genes paaX and paaY did not affect PA catabolic functions, suggesting that the 12 remaining paa genes were essential for the catabolism of this aromatic compound (Fig. 1B).
Analysis of PA Pathway Intermediates-To identify possible intermediates of the PA catabolic pathway in E. coli W, we first checked the capability of this organism to grow on different aromatic compounds that were thought to be produced during PA degradation, e.g. benzoate, phenylpyruvate, p-hydroxyphenylpyruvate, mandelate, and phenylglyoxylate. The strain was patched onto PA-containing mineral agar plates to induce the PA pathway, and then replicated onto mineral agar plates containing the aromatic compounds at a concentration of 5 mM.
As none of the compounds tested allowed the growth of E. coli Inverted repeats in promoter regions are marked with convergent arrows underneath the sequence. A sequence that shows good similarity to the core consensus integration host factor-binding motif (1) is double-underlined. W, they appeared not to be intermediates in PA catabolism.
To check whether the different pAAD::Tn1000 derivatives caused accumulation of PA pathway intermediates, E. coli W14 cells harboring these plasmids were grown in minimal medium containing PA and glycerol. Supernatants of the cultures were then analyzed by HPLC, revealing that only the clones containing plasmids with Tn1000 insertions in genes paaX, paaY, and paaZ, showed PA consumption. However, although disruptions of genes paaX and paaY did not cause the accumulation of any compound, disruption of the paaZ gene caused the conversion of PA into a metabolite whose retention time in HPLC was identical to that of standard 2-HPA. Gas chromatography-mass spectrometry analysis confirmed this metabolite as 2-HPA (data not shown). Similar results were obtained when resting cell assays of the paaZ insertion mutants were performed in the presence of 1 mM PA and then analyzed by HPLC (Fig. 3A). These data were also in agreement with the observation that plasmid pFCP142, which contains the paa cluster with a truncated paaZ gene (Fig. 1A), conferred to E. coli W14 cells the ability to transform PA into 2-HPA.
To analyze whether 2-HPA was an intermediate in PA degradation, E. coli W cells were grown in PA-containing minimal medium and then inoculated into minimal medium containing 2-HPA as the sole carbon source; however, this compound did not support bacterial growth. Moreover, E. coli W cells grown in minimal medium containing both PA and 2-HPA did not attack the latter compound, as observed by HPLC analysis of the culture supernatants. Therefore, these data suggested that 2-HPA could be a product derived from an unstable intermediate rather than a true intermediate of the PA catabolic pathway. As it has been shown that the cis-dihydrodiols formed during the catabolism of different aromatic compounds readily dehydrate nonenzymatically under acidic conditions leading to the corresponding monohydroxy derivatives (2, 42), we checked whether 2-HPA could be also the product of dehydration of a dihydrodiol by monitoring the reported spectral changes associated with such decomposition (42). However, the UV spectrum of the supernatants from E. coli W14 cells containing the pAAD::Tn1000 derivative 3 ( Fig. 1B) did not show any change under acidic conditions, suggesting that a putative dihydrodiol of PA was not present in the culture medium.
As analyses of culture supernatants did not reveal any true intermediate in PA catabolism, intracellular accumulation of possible metabolites was monitored by thin-layer chromatography of resting cell assays of E. coli W14 (pAAD::Tn1000) mutants in the presence of radioactive [ 14 C]PA. It was observed that only mutations in the paaZ gene caused incorporation of radioactivity into the cells. To identify the radioactive products accumulated, extracts of E. coli W14 (pAAD::Tn1000 derivative 3) were analyzed by HPLC and the detected radioactive peaks were shown to cochromatograph with authentic PA and PA-CoA (data not shown). Furthermore, radioactive 2-HPA was detected by HPLC analysis of the supernatants derived from E. coli W14 (pAAD::Tn1000 derivative 3) in resting cell assays. Therefore, all these data taken together suggest that 2-HPA is secreted to the culture medium when the catabolism of PA is blocked, CoA derivatives such as PA-CoA being the true intermediates of the pathway. The paaK Gene Encodes a PA-CoA Ligase-The formation of PA-CoA during the catabolism of PA in E. coli W suggests the existence of a PA-CoA ligase activity involved in the activation of PA to its CoA derivative. As the derived amino acid sequence of the paaK gene product revealed a putative AMP-binding consensus motif (Table I), and showed a high identity to that of the PhaE, former Pcl (65.6%), and PaaK_Y2 (67.3%) proteins responsible for the conversion of PA to PA-CoA in P. putida U (12) and Pseudomonas sp. Y2 (16), respectively, we assumed that PaaK could be the PA-CoA ligase of E. coli W. To experimentally demonstrate that paaK encoded a PA-CoA ligase, this gene was subcloned in plasmid pUC19 under the control of the lac promoter resulting in plasmid pAFK5 (Fig. 4A). Crude extracts of E. coli W14 (pAFK5) cells grown in glycerol-containing minimal medium showed a high level of PA-CoA ligase activity (450 units/mg protein), and this activity was dependent on the presence of ATP, CoA, and PA. SDS-polyacrylamide gel electrophoresis analysis of crude lysates from these cells revealed the presence of an intense band corresponding to a protein with an apparent molecular mass of 49 kDa (data not shown), which is in good agreement with the predicted molecular mass for the PaaK protein (Table I). The N-terminal sequence, MITNTK, of the overproduced protein corresponded with that deduced from the nucleotide sequence of the paaK gene, thus confirming it as the paaK gene product and showing that no processing of its N-terminal end occurs.
To demonstrate that the product of the reaction catalyzed by PaaK was PA-CoA, we performed resting cell assays of E. coli W14 (pAFK5) in the presence of [ 14 C]PA. Two radioactive peaks that cochromatographed with authentic PA and PA-CoA were observed (Fig. 3B). To confirm that the second radioactive peak corresponded to PA-CoA, we performed the hydrolytic removal of the CoA moiety by treatment of the sample with NaOH at 65°C for 30 min, and, as expected, the resulting 14 Clabeled product was shown to elute in HPLC as standard PA.
Analyses of crude extracts of E. coli W14 (pAAD::Tn1000) mutants grown in minimal medium containing PA and glycerol revealed that insertions of transposon Tn1000 in genes paaB, paaC, paaD, paaE, paaF, paaG, and paaJ caused a significant decrease in the PA-CoA ligase activity, such activity being below detection limits when the Tn1000 insertions were located within the paaK gene (Table II). These results may indicate that genes paaABCDEFGHIJK constitute an operon and that insertions of transposon Tn1000 within this operon exert strong polar effects on the genes located downstream of the insertion site. In agreement with this hypothesis, Tn1000 insertions in genes paaZ, paaX, and paaY did not reduce the PA-CoA ligase activity of the corresponding mutants (Table II).
Functional Organization of the paa Genes-To study the functional organization of the paa genes within the PA catabolic pathway, plasmids pAFK3 (paaK), pAFZ1 (paaZ), pFB67 (paaZpaaABCDE), pAFAF1 (paaABCDE), and pAFFK (paaF-GHIJKpaaXY) (Figs. 1 and 4), were constructed. E. coli W14 cells harboring these plasmids were grown in minimal medium containing both glycerol and PA, and the supernatants of the cultures were analyzed by HPLC. We could only observe PA consumption when genes paaABCDE were expressed simultaneously to the paaK gene. Thus, E. coli W14 (pAFK5, pAFAF1) cells removed PA with the concomitant accumulation of 2-HPA in the culture medium (data not shown). However, PA remained unaltered in the culture medium when E. coli W14 cells containing plasmid pAFK5, pAFAF1, pAFFK, pAFZ1, or pFB67 were analyzed. Therefore, these results indicate that all or some of the paaABCDE genes are involved in 2-HPA formation and that this hydroxylation reaction requires the expression of the paaK gene responsible of PA-CoA formation. As 2-HPA is not a true intermediate in the PA catabolic pathway, it can be assumed that after the first catabolic step in the PA degradation, i.e. activation to PA-CoA, a hydroxylation reaction occurs leading to the formation of a hydroxylated derivative of PA-CoA. A blockade in the PA degradation pathway preventing further catabolism of the hydroxylated PA-CoA derivative could cause the secretion of the latter to the culture medium as 2-HPA.
Although the simultaneous expression of the paaABCDE and paaK genes gave rise to 2-HPA, the additional presence of the paaZ gene did not reveal accumulation of 2-HPA in the supernatants of E. coli W14 (pFB67, pAFK3) cultures. Therefore, these data suggest that the paaZ gene product could be responsible of the third enzymatic step in the aerobic catabolism of PA in E. coli W, genes paaFGHIJ being likely involved in further catabolism of PA to Krebs cycle intermediates.
Regulation of the paa Cluster-Analysis of the paaK expression showed that the paa-encoded pathway was inducible. Thus, although crude extracts of E. coli W14 (pAAD) cells grown in glycerol-containing minimal medium in the absence of PA did not reveal PA-CoA ligase activity, a significant level of PaaK activity was observed when the cells were grown in minimal medium containing glycerol and PA (Table III). Interestingly, extracts from E. coli W14 (pAAD::Tn1000 derivatives  (Table III), thus indicating that Tn1000 insertions in gene paaX cause a constitutive expression of the paa-encoded pathway. However, no constitutive expression of the paaK gene was observed when paaY was disrupted by Tn1000 insertion (Table III). Therefore, these data suggest that the paaX gene product behaves as a negative regulator of the paa catabolic genes.
The arrangement of the paa genes and the polar effects derived from the Tn1000 insertions in the paa cluster suggest the existence of three transcription units, two of them, paaZ and paaABCDEFGHIJK, responsible of catabolic functions, and a third one, paaXY, involved in regulation. To study the promoter regions of the paa cluster, DNA fragments containing the potential Pz, Pa, and Px promoters of genes paaZ, paaAB-CDEFGHIJK, and paaXY, respectively, were PCR-isolated and ligated to the lacZ gene of the promoter-probe vector pSJ3. The resulting translational fusion plasmids pAFPA1 (Pa 424 -lacZ, Pa 424 being a 424-bp DNA fragment containing Pa), pAFPA2 (Pa 255 -lacZ, Pa 255 being the Pa 424 fragment lacking 169 bp at its 5Ј-end), pAFPZ (Pz-lacZ), and pAFPX (Px-lacZ) (Fig. 5), conferred to the host strain E. coli CC118 the ability to produce blue colonies on media containing the ␤-galactosidase indicator 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside, thus indicating the presence of a functional promoter in the four cloned fragments.
To determine the transcription initiation sites in the Pa, Pz, and Px promoters, primer extension analyses were performed with total RNA isolated from E. coli CC118 cells containing plasmids pAFPA1, pAFPX, and pAFPZ (Fig. 6). The transcription initiation sites of the paaA gene were mapped utilizing two different primers (Fig. 6, A and B), and potential Ϫ10 (TG-TAAC) and Ϫ35 (TTGTGA) boxes typical of 70-dependent promoters were identified in the Pa promoter region (Fig. 2). The same results were obtained when plasmid pAFPA2 (Pa 255 -lacZ) was used as source of RNA for the primer extension analyses (data not shown). The transcription initiation site in the Pz promoter (Fig. 6C) was located 27 nucleotides upstream of the ATG translation initiation codon of the paaZ gene, showing a putative Ϫ10 box (TTTAAC) but lacking a consensus Ϫ35 sequence (Fig. 2). Analysis of the 194-bp region between the transcription start sites of the paaZ and paaA genes showed a high AϩT content (70%), and revealed several inverted repeat sequences and a putative integration host factor-binding consensus motif (Fig. 2), which might be involved in the control of gene expression. Although these potential regulatory elements are present in Pa 424 , they are absent in Pa 255 . Transcription of paaX (Fig. 6D) starts 29 nucleotides upstream of the ATG translation initiation codon (Fig. 2). The presence in the Px promoter of a putative extended Ϫ10 box (TGCTATGAT) could explain the absence of a consensus Ϫ35 hexamer (Fig. 2). It is worth noting that the putative stem-loop structure that could act as a transcriptional terminator of the paaABCDEFGHIJK operon is located only 15 bp upstream from the extended Ϫ10 box of the Px promoter (Fig. 2).
We have shown above that expression of the paa catabolic operon is controlled by the paaX gene product; therefore, to FIG. 4. Schematic representation of the subcloning and expression of different paa genes. Plasmids are drawn with the relevant elements and restriction sites indicated. DNA containing the paa genes from E. coli W is represented with a black box. Vector-derived sequences are indicated with continuous (pUC derivatives) or discontinuous (pCK01 derivatives) lines. Arrows show the Plac promoter and the direction of transcription of the genes. Plasmid constructs derived from pAAD (A) and pAAD::Tn1000 derivative 10 (B), are shown. ␥ and ␦, the gamma and delta ends of transposon Tn1000, respectively. Ap r and Cm r , the genes that confer ampicillin and chloramphenicol resistance, respectively. further analyze faithfully this regulatory system, we have engineered the reporter Pa-lacZ fusion within a mini-Tn5 vector (Fig. 5). The resulting constructions, pAFPA1T (Pa 424 -lacZ) and pAFPA2T (Pa 255 -lacZ) were used to deliver by transposition the corresponding translational fusions into the chromosome of E. coli AF141, a rifampicin-resistant E. coli W14 (lacZ Ϫ ) mutant strain devoid of ␤-galactosidase activity, giving rise to the reporter strains E. coli AF1411 and AF1412, respectively. The presence of a strong T7 phage transcriptional terminator downstream of the lacZ fusions and their orientation within the mini-Tn5 elements (Fig. 5), prevented read-through transcription from nearby chromosomal promoters after insertion, thus facilitating the regulatory studies. To check the influence of the PaaX protein on the expression of the reporter fusions, paaX was cloned in plasmid pAFX2 (Fig. 7) and expressed into the reporter strains. The ␤-galactosidase assays of permeabilized E. coli AF1411 and AF1412 cells harboring control plasmid pCK01 showed a similar and constitutive expression of the reporter fusions (Table IV). When the gene paaX was expressed in trans, we observed a drastic decrease (more than 2 orders of magnitude) in the ␤-galactosidase levels of E. coli AF1411 (pAFX2) and E. coli AF1412 (pAFX2) cells (Table IV), thus indicating that PaaX behaves as a transcriptional repressor of Pa both in the Pa 424 and Pa 255 promoter fragments.
The repressor effect of PaaX on Pa could not be significantly alleviated by growing the cells in the presence of 5 mM PA (Table IV), which indicates that this aromatic compound is not the true inducer of the pathway. Interestingly, when the paaK and paaX genes were simultaneously expressed in the reporter strains, ␤-galactosidase activities were shown to be inducible by 5 mM PA (Table IV), suggesting that the reaction product of the PaaK enzyme, i.e. PA-CoA, is the inducer of the Pa promoter. Furthermore, it is worth noting that Pa 424 and Pa 255 responded similarly to the PaaX-mediated regulation, thus suggesting that the 169-bp region that is absent in Pa 255 is not directly involved in the promoter-operator sites of the paa catabolic operon. DISCUSSION In this report, we describe the molecular characterization of the PA catabolic pathway of E. coli. Previous work had shown that, whereas E. coli K-12 and E. coli W were able to grow on PA as the sole carbon source, this catabolic ability was lacking in E. coli C (2). The molecular analysis presented here confirm the previous observations, indicating that a 33.3-kb DNA fragment that appears to contain the paa genes responsible of the PA catabolism in E. coli W is lacking in E. coli C as well as in the mutant strain E. coli W14. However, we have shown here that the ability of E. coli K-12 to grow on PA was strain-dependent, with point mutations or small gene rearrangements being the most probable reason for the PA Ϫ phenotype of some K-12 laboratory strains such as DH5␣, HB101, and DH1.
The paa genes from E. coli W were located in a chromosomal 15.4-kb DNA fragment cloned in plasmid pAAD, and they mapped at the right end of the mao region (Fig. 1A), which is involved in the transformation of 2-phenylethylamine into PA (18,37,38). As the equivalent mao genes in E. coli K-12 have been mapped at min 31.0 on the chromosome (43), and two PA Ϫ mutants of E. coli K-12 had been located in this chromosomal region (7), a similar location of the paa genes in the chromosome of E. coli W can be suggested.
The nucleotide sequence of the paa cluster revealed the presence of 14 ORFs, paaZpaaABCDEFGHIJKpaaXY (Figs. 1 and 2), that corresponded with those of unknown function whose Protein Identification Database accession numbers are g1787653-g1787664, g1787666, and g1787667, respectively, and that have been recently sequenced in E. coli K-12 (accession numbers AE000236, AE000237, D90777, and D90778) (41). Although the left end of the paa cluster was near to the maoA gene both in E. coli W and K-12, the right end of the paa cluster was different in the two strains. Thus, although the paaY stop codon was found 231 bp upstream of the ATG start codon of the ydbC gene in E. coli W (Fig. 2), a 9.2-kb sequence encoding a long ORF (ydbA) disrupted by two insertion sequences (IS2 and IS30) was found between paaY (Protein Identification Database accession number g1787667) and ydbC in E. coli K-12 (29). The presence of insertion sequences near the paa cluster and the location of this cluster in a nonessential region of the chromosome (44) provide some clues on the possible mechanisms of gene mobilization of a catabolic cassette that would enhance bacterial adaptability, and could explain the heterogeneity observed among different E. coli strains respect to their ability to mineralize PA. It is also noteworthy that the mao genes for the metabolism of 2-phenylethylamine, an aromatic amine whose degradation gives rise to PA, lie adjacent to the paa cluster responsible for the further catabolism of PA. This association between genes belonging to the same catabolon (15), i.e. genes involved in convergent degradative routes, could be considered as an important evolutionary and adaptive advantage. Another example of such association within a PA catabolon has been recently described in the path-  way for styrene degradation in Pseudomonas sp. Y2, where the sty genes responsible of the oxidation of styrene to PA are in tight association with the genes involved in PA degradation (16). 3 The genetic arrangement of the paa cluster and the mutagenesis of pAAD with transposon Tn1000 revealed that the 14 paa genes are organized in three transcriptional units, two of them, paaZ and paaABCDEFGHIJK, essential for the catabolism of PA, and a third one, paaXY, that contains the paaX regulatory gene. An overall sequence comparison analysis of the paa gene products showed that they were homologous to the recently described pha genes responsible of the catabolism of PA in P. putida U (15) (Fig. 8B). Here, we have presented experimental evidence that the paaK gene product is the PA-CoA ligase of E. coli W (Fig. 3B), an activity that had been FIG. 5. Schematic representation of the construction of lacZ translational fusion cassettes. The PCR-amplified promoter regions (detailed under "Experimental Procedures") cloned in the promoter-probe plasmid pSJ3 are shown. The promoterless lacZ reporter gene is indicated with a thick arrow. The promoters and the direction of transcription of the genes are represented with white and black thin arrows, respectively. ⌬, a truncated gene. The number of amino acid residues fused to the LacZ protein is indicated in parentheses. The white box represents the early T7 transcriptional terminator (19). The I and O termini of the hybrid mini-Tn5 transposons are also indicated. Ap r and Km r , the genes that confer ampicillin and kanamycin resistance, respectively. tnp*, gene devoid of NotI sites encoding the Tn5 transposase. Restriction sites are: B, BamHI; E, EcoRI; H, HindIII; K, KpnI; N, NotI; S, SphI; Sm, SmaI; Ss, SspI; X, XbaI. C, G, and T), using plasmids pAFPA1, pAFPZ, and pAFPX as the templates, respectively. Primer extension and sequencing reactions were performed with primers LAC-57 (A, C, and D), and PA5-4 (B), as described under "Experimental Procedures." To the right of each panel, an expanded view of the nucleotide sequence surrounding the transcription initiation site(s) (*) is shown. Note that the sequence corresponds to the coding strand (Fig. 2). detected in this strain when it was grown in PA-containing medium (13). Analysis of the primary structure of PaaK (Fig. 2) revealed that residues 103 SSGTTGKPTV 112 match the AMPbinding site consensus sequence T(SG)-S(G)-G-(ST)-T(SE)-G(S)-X-P(M)-K-G(LAF) in acyl-adenylate-forming enzymes (residues that predominate at that position are underlined, with alternates given in parentheses; X represents a hypervariable position) (45). It is worth noting that the Lys residue of this signature motif is substituted by Thr in all phenylacetyl-CoA ligases so far sequenced, i.e. PaaK, PhaE (12) and PaaK_Y2 (16), an observation that supports recent studies showing that this residue does not assume a major role in ATP binding (45). The sequences 236 DIYGLSE 242 and 302 YRTRD 306 (underlined are the stringently conserved residues) in PaaK also match the conserved motifs II and III that may contribute to the substrates binding sites in acyl-adenylate-forming enzymes (45).
The detection of radiolabeled PA-CoA inside E. coli W14 (pAAD::Tn1000 derivative 3) cells, indicates that disruption of the paaZ gene causes a blockade of the PA catabolic pathway leading to the accumulation of this CoA derivative, and confirms the physiological role of PaaK in the catabolism of PA in this microorganism. Assuming that the paaK gene product catalyzes the first enzymatic step of the PA catabolic pathway, the polar effects caused by the Tn1000 insertions within the potential paa catabolic operon containing the paaK gene at its 3Ј-end, can explain why pathway intermediates did not accumulate in E. coli W14 cells expressing the corresponding pAAD::Tn1000 derivatives. The degradation of PA in P. putida U also appears to require PA-CoA as the first intermediate of the pathway (12), and a similar situation could be inferred in other bacteria that are able to use aerobically PA as the sole carbon source (13,16). The aerobic catabolism of aromatic compounds via their initial activation to CoA derivatives constitutes an unusual strategy that resembles anaerobic degradation mechanisms (46), and could be a widespread way of PA catabolism in bacteria. The participation CoA ligases in the initial step of the aerobic catabolism of 2-aminobenzoate (47) and benzoate (48) in Azoarcus evansii KB740 (formerly Pseudomonas sp. KB740), ferulate in P. putida (49) and Pseudomonas fluorescens (50), and 2-furoic acid in P. putida Fu1 (51) has been also reported, and the existence of a CoA ligase has been suggested for the aerobic catabolism of salicylate in Rhodococcus sp. strain B4 (52) and thiophen-2-carboxylate (53). Moreover, some dehalogenation mechanisms of aromatic compounds also involve CoA thioester formation in aerobiosis (54). Although the rationale for utilizing such hybrid pathways, i.e. aerobic catabolic pathways endowed with typical features of an anaerobic catabolism, is not known, it has been suggested that they could represent a strategy of facultative microorganisms to cope with the fluctuations of oxygen supply (55). In this sense, the existence of a hybrid pathway for the catabolism of PA in E. coli could reflect the facultative anaerobe character of this bacterium.
All or some of the paaABCDE genes appear to be responsible of the second enzymatic step in the catabolism of PA in E. coli. Thus, the expression of paaK and paaABCDE genes in E. coli W14 caused the consumption of PA and the accumulation of 2-HPA in the culture medium. However, 2-HPA appears not to be a true intermediate in the PA catabolic pathway as it does not support growth of E. coli W and is not consumed even when E. coli W cells are growing also in the presence of PA. Interestingly, a similar lack of growth on 2-HPA and accumulation of this compound after adding PA to some cultures of PA Ϫ mutant strains from E. coli K-12 (7) and P. putida U (15), has been also observed. Although the possibility that exogenous 2-HPA does not enter the cells cannot be ruled out, the fact that 2-HPA formation requires the simultaneous expression of the paaK and paaABCDE genes strongly suggests that 2-HPA is not a true intermediate in PA degradation but derives from the accumulation of a hydroxylated PA-CoA intermediate that cannot be further degraded. The excretion to the culture medium of a hydroxylated aromatic compound as a dead-end product derived from the intracellular accumulation of a hydroxylated CoA derivative has been also reported in the hybrid pathway for the catabolism of 2-aminobenzoate (47), and could be a general strategy of the cells to prevent the possible metabolic risk of depletion of the intracellular pool of CoA (15,56).
The second catabolic step in PA degradation in E. coli seems to be, therefore, the hydroxylation of PA-CoA. Although we could not detect a hydroxylated CoA derivative in E. coli W14 (pAAD::Tn1000 derivative 3) cells, intracellular accumulation of 2-HPA-CoA has been observed during the catabolism of PA by a PA -P. putida U mutant strain (15). Sequence comparison analyses of the paaABCDE gene products revealed that the PaaE protein (356-amino acid length) showed significant similarity with the class IA-like reductases (Table I). These enzymes are members of the ferredoxin-NADP ϩ reductase (FNR) family and they contain a FNR-like domain consisting of a FMN(FAD)-and a NAD(P)-binding region (57). The residues 55 RCYS 58 in PaaE fit the RXYS consensus motif for binding of the isoalloxazine ring of the flavin cofactor, and residues 121 GS-GITP 126 and 216 CGPAAM 221 match the GXG(X) 2-3 P and CG(X) 3-4 M sequences for the binding of the NAD(P) ribose and NAD(P)-pyrophosphate-nicotinamide moieties of the nicotinamide cofactor, respectively (58). At the C terminus of the FNR-like domain, residues 299 -337 in PaaE correspond to the CX 4 CXXCX 24 -34 C conserved motif of the plant-type ferredoxin [2Fe-2S] binding domain (58). Other members of the extended FNR family are the reductase components of the methane, alkene, phenol, and toluene diiron monooxygenases (59 -63), a group of bacterial hydrocarbon oxidation enzymes that com- prises an evolutionarily related protein family (60). These soluble multicomponent monooxygenases contain, in addition to the reductase component, a heteromultimeric (␣␤␥) oxygenase component, a low molecular weight activator protein (61)(62)(63), and, in some cases, a Rieske-type ferredoxin (59,60). Interestingly, the primary structure of the PaaA protein (309-amino acid length) shows the two repeats of residues EX 2 H separated by approximately 100 amino acids (positions 155-158 and 249 -252) that characterize the dinuclear iron binding-site of the large (␣) oxygenase subunit of the methane, phenol, and toluene diiron monooxygenases (60). Moreover, the amino acid sequence of PaaB (95-amino acid length) shows the strictly conserved residues found in the low molecular weight dissociable activator protein that is required for optimal turnover of the oxygenase component in multicomponent diiron monooxygenases (64). Therefore, these sequence comparison analyses suggest that genes paaABCDE may encode the five subunits of a diiron multicomponent oxygenase, with PaaB being the effector protein and PaaE the reductase that mediates electron transfer between NAD(P)H and the PaaACD oxygenase component. It is worth noting that the paaE gene product can constitute the first example of a reductase subunit from a multicomponent oxygenase that shows a reversed domain order, i.e. a FNR-like N-terminal domain and a plant-type ferredoxin C-terminal domain, which supports the previous hypothesis that class IA-like reductases may have been recruited for a variety of aromatic ring oxidation reactions (65). Moreover, the putative PaaABCDE oxygenase, and its counterpart encoded by the phaFGHI operon of P. putida U (Fig. 8B), may represent the first reported multicomponent oxygenase acting on a CoAactivated aromatic acid.
The paaZ gene product appears to be responsible of the third enzymatic step of the PA catabolic pathway. The putative PaaZ protein (681-amino acid length) presents an N-terminal region (residues 1-527) whose primary structure shows similarity with that of aldehyde dehydrogenases (Table I). In this sense, the PaaZ residues 229 FTGSAATG 236 and 291 GQKCTAIR 298 , respectively, match the consensus NAD(P) ϩ -binding site and the active site motif spanning the catalytic cysteine (underlined) of aldehyde dehydrogenases (18,66). Moreover, the sequence 254 MEADSLN 260 in PaaZ encompasses the potential catalytic glutamic acid residue (italicized) of aldehyde dehydrogenases (18,66). The amino acid sequence of the C-terminal region of PaaZ shows similarity to that of the maoC and nodN gene products of unknown function (Table I). As has been suggested for the analogous PhaL protein of P. putida U (15) (Fig. 8B), the paaZ gene product in E. coli might catalyze the aromatic ring cleavage of the hydroxylated CoA derivative formed during PA degradation. Nevertheless, the formation by PaaZ of a nonaromatic CoA cyclic intermediate, similar to that described as the product of the reaction catalyzed by the aminobenzoyl-CoA monooxygenase-reductase during the aerobic catabolism of 2-aminobenzoate (67), cannot be ruled out.
The paaF, paaG, paaH, and paaJ gene products show significant sequence similarities to fatty acid ␤-oxidation enzymes (Table I), and therefore can tentatively constitute a ␤-oxidation-like pathway involved in the successive oxidation reactions of the non-aromatic CoA intermediate. Interestingly, a  Blocks with similar shading or hatching indicate homologous regions encoding potential functional units in both gene clusters. The location and size of the intergenic regions, are also indicated. Bent arrows represent the promoters. B, percentages of amino acid sequence identity between the analogous paa and pha gene products. Note that genes phaJK do not have counterparts in the paa cluster, and that genes paaB and paaI have not been described in the pha cluster.
␤-oxidation-like mechanism is another typical feature of the anaerobic catabolism of aromatic compounds (46). The primary structure of the putative PaaF (255-amino acid length) and PaaG (262-amino acid length) proteins shows similarity with that of members of the enoyl-CoA hydratase/isomerase superfamily (54,68) (Table I). The paaH gene encodes a protein (475-amino acid length) that shares the signature sequence motives of 3-hydroxyacyl-CoA dehydrogenases (69) (Table I), thus suggesting that it could attack the product of the reaction catalyzed by the PaaF and PaaG enzymes. Although the paaI gene product (140-amino acid length) did not show a high level of sequence similarity with other proteins in the data bases, the paaJ gene product (401-amino acid length) presented a significant sequence similarity with the PcaF and CatF ␤-ketoadipyl-CoA thiolases (Table I), residues 90 and 386 in PaaJ being the putative catalytic cysteines. As PcaF and CatF catalyze the last step in the ortho-cleavage pathway for the aerobic degradation of protocatechuate and catechol, respectively (70), it is tempting to speculate that PaaJ and its analogous PhaD protein in P. putida U (Fig. 8) are also responsible for the last enzymatic step in PA degradation.
In the paa cluster, we have identified three promoters, Pz, Pa, and Px, which drive the expression of genes paaZ, paaAB-CDEFGHIJK, and paaXY, respectively (Figs. 6 and 8A). The expresion of the paa-encoded catabolic pathway is inducible, and it has been shown that the Pa promoter is negatively controlled by the paaX gene product (Table III and IV). The PaaX protein (316-amino acid length) contains a stretch of 25 residues at amino acids 39 -64 that shares similarity with the helix-turn-helix motif predicted to be important for DNA recognition and binding in other transcriptional repressors such as GntR (71) and FadR (72). The GntR and FadR binding sites within the respective promoters contain a region of dyad symmetry, which is located very close to the transcription initiation sites (71)(72)(73). Interestingly, a region of dyad symmetry can also be found centered near the transcription initiation sites in the Pa promoter (Fig. 2). As the repression caused by PaaX was only alleviated by PA in the presence of the PaaK protein (Table IV), PA-CoA appears to be the true inducer of the paaencoded pathway. In this sense, gel retardation assays have confirmed PA-CoA as the effector molecule. 4 Therefore, PaaX constitutes the first reported transcription factor regulated by CoA derivatives that controls the catabolism of aromatic compounds. It is worth noting that the FadR transcriptional repressor, which is regulated by acyl-CoA compounds and shows local similarity to PaaX, is also controlling the expression of genes involved in ␤-oxidation mechanisms (73).
Overlapping the 3Ј-end of paaX, we have found the putative ATG translation initiation codon of the paaY gene. A palindromic sequence (⌬G value of Ϫ15.7 kcal/mol) followed by a (T) 7 tract is located 42 bp downstream of the TAA stop codon of paaY (Fig. 2), and may act as a -independent transcription terminator of the putative paaXY operon. Although the primary structure of the paaY gene product (196-amino acid length) and its analogous PhaM protein from P. putida U (Fig.  8B) show several repeats of the hexapeptide (LIV)GX 4 motif that characterizes the members of the bacterial transferases family, e.g. the CaiE protein from the carnitine operon of E. coli and the Fbp ferripyochelin-binding protein of P. aeruginosa (Table I), the physiological role of these proteins in PA catabolism is still unknown.
Comparative studies of the whole structure and organization of the paa and pha clusters from E. coli and P. putida U, respectively (Fig. 8A), revealed interesting functional and evo-lutionary data. Thus, although the pha genes appear to be cotranscribed in four discrete DNA segments or modules encoding the six different functional units for the catabolism of PA, i.e. the ␤-oxidation and activation (phaABCDE), hydroxylation (phaFGHI), transport and dearomatization (phaJKL), and regulation (phaMN) units, the paa cluster showed the transcriptional coupling of the hydroxylation-␤-oxidation-activation functional units into the single operon paaABCDEF-GHIJK (Fig. 8A). As there is good evidence that operons coding for the catabolism of aromatic compounds are assembled in a stepwise manner from existing catabolic genes (74), it is tempting to speculate that the paa cluster from E. coli arose by the fusion of some gene blocks that are contiguous but separately regulated in the pha cluster of P. putida U, and therefore it could be considered as a further step in the evolution toward a single regulon of a common ancestral gene cluster involved in PA catabolism. Moreover, the differences in gene order within some of the DNA modules, and the relative locations of these modules in the paa and pha clusters, suggest that various DNA rearrangements have occurred during their evolution. As the GϩC content of the paa (52.5%) and pha (63.5%) genes averaged a value close to the mean GϩC content of E. coli (51.5%) and P. putida (60%) genomic DNA (40), it could be thought that these two set of genes have been imprisoned within each host over a long period of evolution. Especially remarkable is the observation that the phaJ and phaK genes of P. putida U, encoding a permease and a specific-channel-forming protein for the uptake of PA, respectively (15), are absent in the paa cluster from E. coli W (Fig. 8A). Interestingly, the phaJ gene product shows significant amino acid sequence identity (62.1%) with the product of the yjcG gene that is located at min 92.2 of the E. coli K-12 chromosome (29). Whether a permease, such as the putative YjcG protein, and a channel-forming protein are required for the catabolism of PA in E. coli is still an open question.
The identification and genetic characterization of the hybrid paa-encoded pathway complete our knowledge on the pathways so far described for the aerobic catabolism of aromatic compounds in E. coli. Although, in some Pseudomonas and Acinetobacter species, a supraoperonic clustering of the aromatic catabolic genes has been observed in a limited region of the chromosome, the aromatic catabolic clusters are dispersed throughout the genome in E. coli, with cluster mhp (3-(3-hydroxyphenyl)propionate and 3-hydroxycinnamate) at min 8 (1,10), paa at min 31, hca (3-phenylpropionate) at min 57.5 (11), and hpa (3-or 4-hydroxyphenylacetate) at min 98 (8). These data also indicate that E. coli is not an "empty box" for the catabolism of aromatic compounds; on the contrary, it is endowed with typical aerobic degradation routes as well as with a novel hybrid pathway, which are considered among the most ubiquitous aromatic compound catabolic systems and therefore are thought to be closer to the central catabolism than those involved in the degradation of xenobiotic compounds (75).
The results presented in this work provide a framework for additional studies to determine the role and properties of the enzymes involved in PA catabolism through a hybrid aerobic pathway that is likely to be a widespread route for the metabolism of this aromatic compound. In this sense, the cloned paa genes should be useful as probes to identify homologous genes from distinct groups of bacteria. Moreover, we anticipate that the unique features of the aerobic paa-encoded pathway will reveal novel catabolic activities that can be of great biotechnological interest to improve some microorganisms for the degradation of PA-related aromatic environmental pollutants (e.g., styrene), and for the synthesis of pathway intermediates that can be useful for the production of new or modified antibiotics and plastics (15).