Novel Biodegradable Aromatic Plastics from a Bacterial Source

Novel biodegradable bacterial plastics, made up of units of 3-hydroxy-n-phenylalkanoic acids, are accumulated intracellularly by Pseudomonas putida U due to the existence in this bacterium of (i) an acyl-CoA synthetase (encoded by the fadD gene) that activates the aryl-precursors; (ii) a β-oxidation pathway that affords 3-OH-aryl-CoAs, and (iii) a polymerization-depolymerization system (encoded in the phalocus) integrated by two polymerases (PhaC1 and PhaC2) and a depolymerase (PhaZ). The complete assimilation of these compounds requires two additional routes that specifically catabolize the phenylacetyl-CoA or the benzoyl-CoA generated from these polyesters through β-oxidation. Genetic studies have allowed the cloning, sequencing, and disruption of the genes included in the phalocus (phaC1, phaC2, and phaZ) as well as those related to the biosynthesis of precursors (fadD) or to the catabolism of their derivatives (acuA, fadA, and paa genes). Additional experiments showed that the blockade of eitherfadD or phaC1 hindered the synthesis and accumulation of plastic polymers. Disruption of phaC2 reduced the quantity of stored polymers by two-thirds. The blockade ofphaZ hampered the mobilization of the polymer and decreased its production. Mutations in the paa genes, encoding the phenylacetic acid catabolic enzymes, did not affect the synthesis or catabolism of polymers containing either 3-hydroxyaliphatic acids or 3-hydroxy-n-phenylalkanoic acids with an odd number of carbon atoms as monomers, whereas the production of polyesters containing units of 3-hydroxy-n-phenylalkanoic acids with an even number of carbon atoms was greatly reduced in these bacteria. Yield-improving studies revealed that mutants defective in the glyoxylic acid cycle (isocitrate lyase−) or in the β-oxidation pathway (fadA), stored a higher amount of plastic polymers (1.4- and 2-fold, respectively), suggesting that genetic manipulation of these pathways could be useful for isolating overproducer strains. The analysis of the organization and function of the pha locus and its relationship with thecore of the phenylacetyl-CoA catabolon is reported and discussed.

2-fold, respectively), suggesting that genetic manipulation of these pathways could be useful for isolating overproducer strains. The analysis of the organization and function of the pha locus and its relationship with the core of the phenylacetyl-CoA catabolon is reported and discussed.
The restricted capacity of plastic-producing bacteria to store polyesters bearing a phenyl group suggests that the synthesis of these unusual compounds requires (i) a specific uptake system for the transport of the aromatic precursors, (ii) a specific acyl-CoA synthetase (ACS), (iii) a specific polymerase or a mutated enzyme with broader substrate specificity (14), and/or (iv) the existence of an additional catabolic route, linked to the ␤-oxidation pathway, to ensure complete assimilation of the ␤-oxidation products (benzoyl-CoA or phenylacetyl-CoA) generated from the monomers (3-hydroxyphenylalkanoyl-CoA derivatives) once released from the stored polymer ( Fig. 1) (9). This complete assimilation could reduce the possibility that intermediate metabolites might inhibit the biosynthetic process.
In the present work, we describe the existence of novel natural aromatic plastics and analyze their structure for the first time. Genetic and biochemical studies were also performed in order to establish the sequence and characteristics of the genes and enzymes specifically involved in the synthesis and degradation of these polyesters (PHPhAs). The inclusion of this pathway in the phenylacetyl-CoA catabolon (53) and its influence in the evolution of the catabolic potential of P. putida U are also discussed.

EXPERIMENTAL PROCEDURES
Materials-n-Phenylalkanoic acids, n-alkanoic acids, and [1-14 C]phenylacetic acid were supplied by Lancaster Synthesis or by Sigma. [1-14 C]Octanoic acid was from American Radiolabeled Chemicals. All other products were of analytical quality or HPLC grade.
Microorganisms and Culture Conditions-The strain of P. putida (U) (Colección Españ ola de Cultivos Tipo 4848) used in all of the experiments was from our collection. It was maintained on Trypticase Soy Agar (Difco), and growth slants (8 h at 30°C) were used to inoculate liquid medium. Each 2000-ml Erlenmeyer flask containing 500 ml of the required medium was inoculated with 10 ml of a bacterial suspension (10 10 bacteria). Incubations were carried out in a rotary shaker (250 rpm) at 30°C for the time required in each set of experiments. The medium (MM) used for the growth of P. putida U and its mutants was a chemically defined one (54). When required, the carbon source (phenylacetic acid, PhAc) was replaced by a different one (several aromatic analogues or fatty acids). The concentration of the molecule used as carbon source was indicated in each set of experiments. The synthesis of plastic polymers was studied in bacteria grown for different times in a MM (54) in which PhAc had been replaced by several n-alkanoic (As) or PhAs at the required concentrations (between 5 and 15 mM). In some experiments, the carbon source was 4-hydroxyphenylacetic acid . This compound, which is efficiently assimilated by P. putida U, cannot be used as plastic precursor by this bacterium.
Escherichia coli HB101 containing the plasmid pGS9, which includes the transposon Tn5, was kindly supplied by J. L. Ramos (Estación Experimental del Zaidín, Consejo Superior de Investigaciones Científicas, Granada, Españ a). E. coli XL1-Blue (Stratagene) was supplied by the commercial firm, and it was used for overexpressing different proteins. E. coli fadR (a strain defective in the ␤-oxidation transcriptional repressor) and fadB (mutated in the gene encoding the enoyl-CoA hydratase and the 3-hydroxyacyl-CoA dehydrogenase) mutants, which were used to study the functional expression of phaC1, phaC2, or the complete pha locus (phaC1-phaZ-phaC2) from P. putida U, were kindly supplied by C. C. DiRusso (Department of Biochemistry and Molecular Biology, Albany Medical College, Albany, NY). A different culture of the fadB E. coli mutant was also supplied by A. Steinbü chel and B. Rehm (Institut fü r Mikrobiologie, Westfä lische Wilhems-Universität Mü nster, Mü nster, Germany). When required, a double mutant E. coli JMU 193 (fadR, fadB) supplied by Q. Ren, Institut Fü r Biotechnologie, ETH, Zü rich, Switzerland) was used. Micrococcus luteus (ATCC 9341) was used for the evaluation of acyl-CoA synthetase by bioassay (54) Characterization of PHPhAs-PHPhAs were isolated following the procedure reported by Lageveen et al. (9). For these experiments, cells grown in a chemically defined medium (MM) (54) containing different concentrations of n-phenylalkanoic acids as the sole carbon sources (see above) were used. The contents and composition of PHPhAs were determined by gas chromatography as previously reported (49).
NMR spectral analyses were recorded at 18°C on a Varian Unity 300 NMR spectrometer at 300 ( 1 H) and 75 ( 13 C) MHz, using tetramethylsilane as internal standard. Spectra were measured in CD 3 OD.
Isolation of Mutants-Mutants of P. putida unable to degrade octanoic acid or phenylacetic acid or those affected in the production of poly-(3-hydroxyphenylalkanoic) acids were selected by mutagenesis with the transposon Tn5 as reported (56). In some cases, mutants were obtained by disruption of the required gene (see below).
Mutants lacking a functional glyoxylic acid cycle or a ␤-oxidation pathway were characterized by enzyme assay (57), metabolically (studying their ability to grow in MM containing acetate or octanoate as the sole carbon source), and by location of the insertion. Mutants handicapped in the biosynthesis of plastic polymers were identified by the different contrast of the colonies (translucent) when properly cultured (8,13).
The strains unable to assimilate phenylacetic acid (indicated as PhAc Ϫ ) were classified according to the intermediate accumulated in the culture broth (see below) or as a function of the presence or absence of phenylacetyl-CoA ligase activity in their cell-free extracts (58).
HPLC Equipment and Chromatographic Procedure-To determine the rate of utilization of the carbon sources and to identify the catabolic intermediates accumulated by certain mutants, samples of culture broth (50 l) were taken at different times, centrifuged, and filtered through a Millipore filter (pore size, 0.45 m). Aliquots were analyzed on a HPLC apparatus (SP8800; Spectra Physics) equipped with a variable wavelength UV-visible detector (Waters 486), Millenium software (Waters 2010), and a microparticulate (particle size 10 m; pore size 100 nm) reverse-phase column (Nucleosil C-18, 4.6 (inner diameter) by 250 nm; Phenomenex Laboratories). The mobile phase was as follows: A, 0.2 M KH 2 PO 4 (pH 4.5); B, CH 3 CN in a linear gradient ranging from 95% A:5% B to 50% A:50% B over 1 h. Flow rate was 1 ml min Ϫ1 , and the eluate was monitored at 254 nm. Column temperature was 30°C. Under these conditions, the retention times of 4-HPhAc, 2-hydroxyphenylacetic acid, phenylacetic acid (PhAc), 6-phenylhexanoic acid, and 8-phenyloctanoic acid were 19, 25, 30, 45, and 56 min, respectively. HPLC analysis of other products was carried out as reported elsewhere (56, 59 -60).
DNA Manipulations and Sequencing-DNA manipulations, gel electrophoresis, DNA sequence analysis, primer synthesis, promoter analysis, and polymerase chain reaction amplification were performed as reported (53). To analyze the function of the genes encoding the polymerases and the depolymerase involved in the biosynthesis and mobilization of the plastic polymers, each particular gene was disrupted by homologous recombination. Thus, an internal fragment (400 -900 base pairs) belonging to the gene to be mutated was cloned in pK18:mob, and the resulting construct was introduced into P. putida U by triparental mating as previously reported (53). Correct insertion of the pK18:mob in the desired gene was established by polymerase chain reaction amplification (53). When required, overexpression of the different genes and proteins was carried out using the plasmid pQE-32 (Qiagen Inc.) according to the manufacturer's instructions.
[ 14 C]Octanoic Acid Transport Studies-The uptake of labeled octanoic acid was carried out in bacterial cultures as previously reported (60). When required, different phenylalkanoates were added to the transport system in order to analyze their effect on the uptake of octanoic acid. Unlabeled octanoic acid was analyzed by gas liquid chromatography as described elsewhere (61).
The ACS from P. putida U was assayed as indicated for other analogous enzymes (62)(63). When the molecules tested as substrates were longer than C 5 , a different assay method (bioassay against M. luteus) was used (64).
PhaC1 and PhaC2 activities were assayed in vivo by analyzing the monomeric composition of the plastics stored by mutants of P. putida U specifically blocked in one or both enzymes. The substrate specificities of these enzymes were also studied in recombinant E. coli JMU193 containing either the phaC1 or the phaC2 genes (see "Results and Discussion").
Isocitrate lyase was measured spectrophotometrically in cell-free extracts (54) of P. putida U as indicated previously (65).

Strategy Followed for Producing Novel PHAs
The collection of unusual bacterial plastics (those containing units of PHPhAs) was approached under the assumption that bacteria able to accumulate aliphatic plastics would also be able to synthesize aromatic polyesters whenever the substrate specificity of the polymerases is broad enough and when the effective degradation of the final ␤-oxidation product (phenylacetyl-CoA or benzoyl-CoA) is produced by a catabolic route. Accordingly, we first selected bacterial species able to (i) produce aliphatic plastics and (ii) grow in a chemically defined medium (MM) containing either benzoic acid or phenylacetic acid as the sole carbon source (54). These microbes would have (i) the biosynthetic pathway necessary for aliphatic polymer production and (ii) the catabolic enzymes required for the degradation of the final ␤-oxidation products: benzoyl-CoA (or benzoic acid) and phenylacetyl-CoA (or phenylacetic acid). This second requirement seems to be essential, since if n-phenylalkanoic acids are catabolized, as expected, through the ␤-oxidation pathway ( Fig. 1) (66), the final product would be benzoyl-CoA or phenylacetyl-CoA if they have an odd or an even number of carbon atoms, respectively. Accordingly, only those microbes able to assimilate phenylacetic acid (PhAc) or benzoic acid via a CoA intermediate would be able to grow efficiently in MM containing n-phenylalkanoic acids as the sole carbon source.
Taking into account all the above considerations, we selected a bacterial strain, P. putida U, that meets the following metabolic conditions: (i) it is able to catabolize PhAc and benzoic acid aerobically (54 -56), and (ii) it synthesizes different polyesters when cultured in MM containing C 6 -C 12 alkanoic acids (60).
Ultrastructural studies revealed that P. putida U is able to synthesize biodegradable plastics when cultured in a chemically defined medium containing several PhAs whose aliphatic moiety ranges between 6 and 10 carbon atoms as carbon sources (see Table I). These compounds, which appeared as electron-transparent granules, were stored intracellularly as reserve materials (Fig. 2).
Although certain bacterial species only accumulate PHAs when an essential nutrient (e.g. nitrogen, phosphorus, sulfur, magnesium) becomes limiting (8,10), P. putida U synthesizes plastic polyesters from the very early logarithmic phase of growth (Fig. 2), suggesting that the regulatory mechanisms that control the biosynthesis of these polymers in both types of bacteria are quite different.
Bacterial growth and polymer contents were higher when PhAs containing acyl chains longer than six carbon atoms were used as carbon sources ( Fig. 3 and Table I), the highest values being detected when 10-phenyldecanoic acid was supplied to the medium. When other PhAs (containing an acyl chain of less than six carbon atoms) were tested, no polyesters were produced, since P. putida cannot assimilate these compounds or does so very poorly (Fig. 3A).

Structural Analysis of the PHPhAs
The general structure of these novel polyesters, established by NMR studies, 2 is shown in Fig. 4. A 1 H-13 C correlation 2 NMR data for the polyester 3-hydroxy-6-phenylhexanoate were as follows: 1  spectrum allowed the specific 13 C assignments reported, whereas the multiplicities and connectivities in the 1 H NMR spectra were established by distortionless enhancement by polarization transfer and 1 H-1 H COSY experiments, respectively (67). These analyses allowed us to conclude that the structures of the monomers are 3-hydroxy-5-phenylvaleric acid, 3-hydroxy-6-phenylhexanoic acid, 3-hydroxy-8-phenyloctanoic acid, and 3-hydroxy-10-phenyldecanoic acid (see Table I). This is the first description of the existence of natural plastics containing units of 3-hydroxyphenyl derivatives with an even number of carbon atoms. Thus, when P. putida U was cultured in MM containing 6-phenylhexanoic acid as carbon source, a homopolymer of 3-hydroxy-6-phenylhexanoate was produced. However, when the carbon source was 8-phenyloctanoic or 10phenyldecanoic acid, copolymers were formed. In the first case, study of the monomer composition revealed that the polymer contained two different molecules: 3-hydroxy-6-phenylhexanoate and 3-hydroxy-8-phenyloctanoate at relative proportions of 62.5 and 37.5%, respectively. However, when the carbon source was 10-phenyldecanoate, the polymer contained units of 3-hydroxy-6-phenylhexanoate (48.2%), 3-hydroxy-8-phenyloctanoic acid (31.2%), and 3-hydroxy-10-phenyldecanoate (20.6%), respectively (see Table I). Interestingly, when the molecules used as carbon sources were PhAs containing an odd number of carbon atoms, the only plastic produced was a homopolymer of 3-hydroxyphenylvaleric acid (see Table I). These results allowed us to conclude that the polymerases (see below) responsible for the condensation of aromatic monomers into plastic polymers recognize CoA derivatives of 3-hydroxyalkanoic acids containing acyl chains longer than four carbon atoms as substrates.
Analysis of the polymers stored by P. putida U when this microbe was cultured in MM containing different As (from butyric acid to dodecanoic acid) as carbon sources revealed that the minimum length of the monomers was six carbon atoms and that, depending on the carbon source used for growth, different polymers (or copolymers) were produced (see Table  II). In sum, when C 2nϩ2 PhAs or C 2nϩ2 As are used as polymer precursors, the plastics synthesized contain monomers that have the same length in the acyl moiety, regardless of whether or not they are linked to a benzene ring. However, when the precursors are molecules with an odd number of carbon atoms (aromatics or aliphatics) the monomer length is not the same. Thus, when 6-phenylhexanoic, 8-phenyloctanoic, or 10-phenyldecanoic acids were used as carbon sources, the monomeric units of these polymers were the same as those obtained when hexanoic, octanoic, and decanoic acid were employed but linked to a benzene ring (see Tables I and II). However, when 7-phenylheptanoic or 9-phenylnonanoic acids were supplied to the medium, polymers containing 3-hydroxy-7-phenylheptanoic acid or 3-hydroxy-9-phenylnonanoic acids were not found, a homopolymer of poly-3-hydroxy-5-phenylvaleric acid being the only polyester synthesized. Although the explanation for the absence of polymers containing units of 3-hydroxyphenylheptanoic or 3-hydroxyphenylnonanoic acids is unknown, it could be speculated that the physicochemical differences between structurally related substrates (containing an odd or an even number of carbon atoms) might be the cause of this structural difference. Moreover, we have never observed monomers with a size longer than the fatty acid precursor, indicating that in P. putida U the precursors of these polyesters are not obtained through fatty acid synthesis.

Analysis of PHPhAs Biosynthetic and Catabolic Enzymes
Although certain bacteria such as Alcaligenes eutrophus or P. oleovorans accumulated high amounts of plastic polymers containing 3-hydroxy-n-alkanoic acids as monomers (2, 8, 50 -51), to the best of our knowledge they do not produce polyesters when n-phenylalkanoic acids containing an even number of carbon atoms are used as plastic precursors. It could therefore be speculated that the ability of P. putida U to synthesize these aromatic polyesters could be due to the existence in this bacterium of (i) a transport system with broader substrate specificity that could recognize both kinds of polymer precursors (or two different ones, specifically involved in the uptake of phenylalkanoic acids and alkanoic acids, respectively), (ii) two (or more) acyl-CoA synthetases that activate PhAs and As separately or a single enzyme with broader substrate specificity that would activate both kinds of precursors, (iii) a different polymerizing-depolymerizing system, (iv) two different ␤-oxidation pathways (or only one but including some duplicate, less specific or channeling steps) required for the accumulation and partial catabolism of both kinds of plastic polymers, and/or (v) specific routes that ensure the complete catabolism of the final ␤-oxidation products (acetyl-CoA, propionyl-CoA, phenylacetyl-CoA, or benzoyl-CoA). To establish which of these possibilities were true, the following experiments were performed. 128.40 (C-2Ј, C-3Ј, C-5Ј, C-6Ј), 142.50 (C-1Ј), 169.41 (C-1). NMR data for the polyester 3-hydroxy-10-phenyldecanoate were as follows: 13 (15 mM) as the sole carbon source The percentage of PHPhA is indicated as percentage of bacterial dry weight. PHPhA relative composition is indicated as the molar fractions of the total of 3-hydroxyphenylalkanoates in the polymer. 3HPhV, 3-hydroxyphenylvalerate; 3HPhH, 3-hydroxyphenylhexanoate; 3HPhh, 3-hydroxyphenylheptanoate; 3HPhO, 3-hydroxyphenyloctanoate; 3HPhN, 3-hydroxyphenylnonanoate; 3HPhD, 3-hydroxyphenyldecanoate. Similar results were obtained when the isocitrate lyase Ϫ mutant was cultured in MM plus 8-phenyloctanoic acid (15 mM). ACS Ϫ mutant did not accumulate plastic polymers when cultured in MM plus 8-phenyloctanoic acid (15 mM) plus 4-HPhAc (5 mM) and neither when 8-phenyloctanoic acid was replaced by octanoic acid (15 mM) (see Table 2). The results indicated in this table and in the following ones are the means of the data obtained in three different experiments (bacteria harvested from 2.5 L of the required MM). Transport System-When the uptake of [ 14 C]octanoic acid was studied (60), we observed that 5-phenylpentanoic acid, 6-phenylhexanoic acid, 7-phenylheptanoic acid, and 8-phenyloctanoic acid strongly inhibited the uptake of octanoic acid (70, 85, 98, and 97%, respectively) when added at the same concentration (154 M), whereas other related compounds with a lower molecular weight such as 3-phenylpropionic acid and 4-phenylbutyric acid did not cause any significant effect (zero inhibition and 10%, respectively). Furthermore, an efficient transport of octanoic acid (98%) was observed when P. putida U was cultured in MM containing phenyloctanoic acid. We therefore suggest that the uptake of n-phenylalkanoic and n-al-FIG. 2. Electron micrographs. 1, P. putida U (wild type) cultured in minimal medium containing 8-phenyloctanoic acid (15 mM) as the sole carbon source. a, early exponential phase (ϫ 10,000); b, early stationary phase (ϫ 15,000); c, late stationary phase (ϫ 10,000). 2, P. putida Oct 1 Ϫ (ACS Ϫ mutant) (handicapped in the degradation of n-phenylalkanoic acids or n-alkanoic acids, containing an even or an odd number of carbon atoms) cultured until stationary phase in MM supplemented with 5 mM 4-hydroxyphenylacetic acid (which cannot serve as a plastic precursor) with 15 mM 8-phenyloctanoic acid (a) or 15 mM 7-phenylheptanoic acid (b) as carbon source (ϫ 12,000). When this mutant was cultured in the same MM in which 8-phenyloctanoic acid had been replaced by other n-phenylalkanoic acids, by C 6 -C 10 n-alkanoic acids, by citrate, or by citrate and acetate, no synthesis or accumulation of plastic polyesters were observed. 3, P. putida PhaC1 Ϫ mutant cultured until stationary phase in MM containing 15 mM 8-phenyloctanoic acid (ϫ 20,000). Identical results were obtained when 8-phenyloctanoic acid was replaced by different n-phenyloctanoic acids or n-alkanoic acids. 4, P. putida PhaZ Ϫ mutant grown under the same conditions as those described for PhaC1 Ϫ . a, early exponential phase of growth (ϫ 15,000); b, early stationary phase (ϫ 12,000); c, late stationary phase (ϫ 15,000). kanoic acids are at least tightly linked and very probably brought about through the same mechanism. Although a definitive conclusion could only be obtained after analyzing the uptake of radioactive n-phenylalkanoic acids, the lack of commercially available labeled compounds and the difficulty involved in synthesizing them did not allow further and more accurate analyses of this particular point. However, taking into account that a blockade in acyl-CoA synthetase (which activates medium chain-length fatty acids; see below) also handicapped the cellular incorporation of octanoic acid and other polyester precursors (as shown after HPLC analysis of the culture broths), it seems reasonable to propose that the same transport system could be involved in the uptake of all of these compounds (Fig. 1). In fact, it has been suggested (68 -70) that the acyl-CoA synthetase from E. coli activates fatty acids concomitant with transport, and it has been proposed that this enzyme would either be a component of the transport apparatus involved in the uptake of fatty acid or that it would interact, inside the inner membrane, with a hypothetical fatty acid cotransporter. More recently, several authors (71-75) have provided evidence indicating that the uptake of long chain fatty acid in E. coli requires the direct participation of an outer membrane protein and acyl-CoA synthetase.
Acyl-CoA-activating Enzyme-The analysis of the acyl-CoAactivating enzyme(s) was carried out. It could be speculated that two different enzymes (or more) would be involved in the activation of alkanoic and phenylalkanoic acids and that, therefore, the absence of this enzyme that activates phenylalkanoic acids would explain why certain bacteria do not synthesize aromatic polyesters. To clarify this point, we analyzed different mutants unable to grow in chemically defined media (MM) containing 6-, 7-, 8-, or 9-phenylalkanoic acids (or the equivalent alkanoic acids) as carbon sources. These mutants, which were selected by random insertion of the transposon Tn5 (76), were separated into two different groups: those unable to catabolize octanoate and those that do not grow in MM containing acetate as the sole carbon source. The first class of mutants corresponded to bacteria in which the Tn5 had been inserted into some of the genes (or their regulatory elements) encoding a general acyl-CoA-activating enzyme or some of the proteins belonging to the ␤-oxidation pathway (66). In the second group, we included bacteria in which the transposon has    strong homology (accession number AF150669) with the long chain fatty acid ACSs from E. coli and Pseudomonas fragi (77)(78)(79). This gene (fadD) was cloned, sequenced, and expressed in E. coli DH5␣Ј.
The ACS Ϫ mutant efficiently catabolized acetate and butyrate but was unable to assimilate n-alkanoic or n-phenylalkanoic acids with an acyl moiety ranging between C 5 and C 10 , undoubtedly indicating that a single enzyme is responsible for the activation of all of these compounds and that, as expected, this reaction is not catalyzed by acetyl-CoA synthetase or by the acetate kinase (EC 6.2.1.1 and EC 2.7.2.1, respectively). Furthermore, study of the polyesters stored revealed that this mutant does not accumulate plastic polymers when cultured in the appropriate conditions (MM plus 4-HPhAc plus PhAs or As as carbon sources) ( Fig. 2 and Table III). Additionally, we observed that the ACS Ϫ mutant was unable to transport [ 14 C]octanoic acid (0.3% of the final amount transported by the wild type), suggesting that, as mentioned above, the mechanism involved in the uptake of this acid and the activation to its CoA thioester comprises a single event or at least two closely related events (Fig. 1). Similar results were obtained when the transport of [ 14 C]phenylacetic acid was analyzed in mutants in which the gene encoding phenylacetyl-CoA ligase (EC 6.2.1.30), the first enzyme of the specific route involved in the catabolism of phenylacetic acid (56), had been disrupted. We observed that the P. putida U mutants lacking this enzymatic activity were unable to take up phenylacetic acid from the broths. 3 Thus, it can be concluded that in the absence of this single acyl-CoA synthetase activity the uptake of all of the substrates recognized by this enzyme, which are the precursors of the mono-mers used for the synthesis of polyesters, will be blocked, hence avoiding an unnecessary loss of energy, since once transported these compounds cannot be catabolized. As a consequence of this blockade, polyesters cannot be synthesized. In light of the above results, it seems that this single acyl-CoA synthetase is responsible for activation of alkanoates and phenylalkanoates.
On the other hand, it is interesting to draw attention to the lack of plastic storage in the ACS Ϫ mutant when grown in MM plus acetate plus citrate (Fig. 2), since this result indicated that P. putida U, unlike other related bacteria (48), does not synthesize polyesters (or does so very poorly) via fatty acid biosynthesis. This assumption is supported by the fact that when this mutant or the wild type was cultured in minimal media containing either glucose or gluconate as carbon source, synthesis of plastic polymer was not detected (data not shown).
The Polymerizing-Depolymerizing Enzymatic System-We have analyzed the sequence and organization of the genes and proteins involved in the polymerization and depolymerization of these polyesters in P. putida U. For these experiments, mutants blocked in the genes involved in the synthesis of poly-3-hydroxyphenylalkanoates were isolated (13). Sequence analysis of the DNA fragment in which the transposon Tn5 had been inserted revealed that (i) in P. putida U, three genes (phaC1, phaZ, and phaC2 included in the pha locus; see Fig. 5) are involved in the synthesis and mobilization of these polymers, (ii) the sequences of these proteins are similar to the enzymes reported in P. oleovorans and in Pseudomonas aeruginosa (13,21), and (iii) they also have a similar organization (PhaC1-PhaZ-PhaC2) to that reported in such bacteria (see Fig.  6). The contribution of each particular enzyme in the biosynthesis and catabolism of these polymers was analyzed by studying different mutants in which specific disruptions (by homol-ogous recombination; Ref. 53) of the genes encoding PhaC1, PhaC2, and PhaZ (see Fig. 5) had been done. We revealed for the first time in Pseudomonads of rRNA homology group I that a blockade on the phaC1 prevented accumulation of plastic polymers, whereas a mutation in the phaC2 reduced the intracellular amount of polymers by two-thirds. Furthermore, disruption of the gene encoding PhaZ prevented the mobilization of the polymer accumulated intracellularly and decreased the total PHAs content, suggesting that PhaC2 is not synthesized, since it must be expressed from a promoter located upstream the gene encoding the depolymerase (see Table IV).
An analysis based on the amounts of polymer stored by the different mutants allowed us to obtain interesting conclusions about the genetic organization of the pha locus. Thus, when P. putida U (wild type) was cultured in MM plus octanoic acid until the stationary phase of growth, the amount of PHAs extracted was 365 mg/g of dry weight cells (dwc). This quantity represents a balance of the biosynthetic activities of PhaC1 and PhaC2 (additive) and the mobilizing activity corresponding to PhaZ (subtractive). This balance could be represented as follows: PhaC1 ϩ PhaC2 Ϫ PhaZ ϭ 365. When PhaC2 was mutated, the quantity of plastic synthesized (PhaC1 Ϫ PhaZ) was reduced to 115 mg/dwc; therefore, we conclude that the contribution of PhaC2 to the synthesis of plastic should be about 250 mg/dwc. However, when a PhaZ mutant was cultured in the same conditions, we obtained only 250 mg/dcw. This later quantity suggests that in this mutant only the PhaC1 is working, since otherwise total synthesis should have been greater. Although these data suggest that there is not an active promoter downstream from the phaZ gene, we cannot rule out the possibility that a promoter could exist between the phaC1 and the phaZ genes. Moreover, we observed that mutants in which the phaC1 gene was disrupted did not synthesize plastics at all (Table IV), which should not have happened if there had been a promoter downstream from the phaC1 gene. It is worth noting that although the depolymerase activity was lacking in the PhaZ mutant (and therefore mobilization of the polymer did not occur; see Table IV), the quantity of polymer accumulated (250 mg/dwc) was considerably lower than that found in the control (365 mg/dwc), suggesting that the polymerases and probably all three enzymes (PhaC1, PhaC2, and PhaZ) must be organized in a complex to function efficiently.
Genetic studies revealed that whereas the sequences of the genes included in the pha locus are well conserved, further differences were found when the promoter and the two intergenic regions reported in the pha loci of P. oleovorans, P. aeruginosa, and P. putida U were compared (Fig. 6). Thus, in P. putida U the lengths of the intergenic regions are shorter than in other species and lack the inverted repeat sequences found in P. oleovorans and in P. aeruginosa (13,21). This genetic organization suggested that in P. putida U the three pha genes are organized in an operon and that translational coupling may exist.
On the other hand, the study of the putative promoter regions in the three bacterial species revealed that the sequence of P. aeruginosa is quite different from the other two species (Fig. 6A). This variation could account for a different regulation of the three pha loci. When genetic studies, using a plasmid specifically designed for testing promoters (53), were performed, we did not find any functional promoter sequences within the fragment of DNA containing the structural sequences (PhaC1, PhaZ, PhaC2) from the ATG of phaC1 to the TGA of phaC2, whereas a promoter sequence was found in a DNA fragment of 200 base pairs located upstream from the ATG of phaC1.
All of these results allow us to conclude the following: (i) only one promoter (located 5Ј-upstream from phaC1) exists in the pha locus; (ii) the three genes are expressed under the control of this promoter; (iii) the two polymerases (PhaC1 and PhaC2) have fairly similar polymerizing rates; (iv) the activity of PhaZ released 26% of the total polymer synthesized; (v) the same enzymatic system (PhaC1-PhaZ-PhaC2) is involved in the synthesis and mobilization of all of the plastic polymers (aliphatic and aromatic) found in P. putida U; and (vii) the lack of depolymerase does not imply a greater accumulation of polymers but rather the impossibility of mobilizing them (see Table IV and Fig. 2).
Next, the general specificity of both polymerases was analyzed independently by studying the structure of the polymers synthesized and stored in the different mutants reported above. We observed that the polyesters accumulated by the wild-type strain and by the PhaZ and by the PhaC2 mutants were similar in monomer composition and in the percentage of monomers present in the polymer, indicating that the two polymerases recognize the same substrates with similar efficiency (see Tables IV and V). We attempted to confirm the substrate specificity of these enzymes by cloning phaC1 and phaC2 separately in a E. coli fadB mutant handicapped in fatty acid ␤-oxidation (75), as reported for the PHA synthetases from P. aeruginosa (48,80). Unfortunately, we failed to observe the synthesis of plastics in this strain even when the complete pha locus was cloned. It could be that the production of plastic polymers would require the existence of a ␤-oxidation constitutive pathway, so we cloned each independent gene (phaC1 or phaC2) as well as the entire pha locus in the E. coli fadR mutant (a strain defective in the ␤-oxidation transcriptional repressor but which is sensitive to carbon catabolite repression) (74,75), which were then cultured in minimal media containing different PHA or PHPhA precursors as the sole carbon source. Ultrastructural and chemical analysis revealed that plastic polymers did not accumulate in these recombinant strains. Furthermore, when these bacteria were cultured in media containing different fatty acids, a cellular size lower than the wild types (about one-fifth) was observed, suggesting that, under these conditions, overexpression of the polymerases causes cellular damage in the recipient strain. Accordingly, taking into account that when studies on overexpression of PhaC1 in E. coli DH5␣Ј were performed, an abnormal division, due to the absence of septum formation (see Fig. 7), was observed, we used E. coli JMU193 for cloning the pha biosynthetic genes in a monocopy construction, under the control of a inducible promoter, as reported by other authors (55,(81)(82)(83). With these experiments, we attempted simulate the genetic organization (monocopy of each different gene and strict control of their expression) required for the synthesis of these plastics in the wild type (P. putida U), thus avoiding excess synthesis of the polymerase(s) that could cause harmful effects on the recombinant strain. We observed that when these genetically engineered strains were cultured in minimal medium containing glycerol (20 mM) and palmitic acid (3 mM) as carbon sources, the limited expression of the polymerases led to lower synthesis of aliphatic plastics (10% of bacterial dry weight). However, synthesis of aromatic polymers was not observed when palmitic acid was replaced by 8-phenyloctanoic acid. These data suggest that the low yield of synthesis could be due to the fact that either even the low expression of these genes is harmful for the recipient bacteria or that additional proteins (probably an Rspecific enoyl-CoA hydratase or the granule-associated proteins) would be required for a more efficient synthesis of such polymers (7,45,47,84). Although the quantity of the aliphatic polymers accumulated was very low, their analysis revealed that, as indicated above, PhaC1 and PhaC2 showed identical substrate specificity.

Yield Improvement Studies
Since the overexpression of the pha locus from P. putida U in E. coli does not lead to a higher production of PHAs (or PHPhAs) in the recombinant strains, certain biochemical studies were also performed in order to improve the production of plastic polymers in P. putida U. It is well established that the degradation of n-alkanoic acids and phenylalkanoic acids through ␤-oxidation generates several units of acetyl-CoA (66) that are later catabolized by the glyoxylic cycle (65). It could therefore be expected that the introduction of a mutation in some of the genes encoding enzymes belonging to the glyoxylic acid cycle could slow down the ␤-oxidation route (the source of acetyl-CoA) and hence cause the accumulation of some catabolic intermediates (polyester precursors), thus channeling them to the polymeric system and increasing the amount of stored polymers. To check this hypothesis, we isolated a mutant in which the transposon Tn5 (76) had been inserted into a gene belonging to the glyoxylic acid cycle (in the aceA gene encoding the isocitrate lyase gene, accession number AF150671). This mutant, which shows an Oct Ϫ Ac Ϫ phenotype, was unable to grow in MM containing either acetate or nalkanoic acids whose carbon length was an even number of carbon atoms, whereas it grew well in MM containing either several n-alkanoic acids with an odd number of carbon atoms or n-phenylalkanoic acids with an even or odd number. These results indicate that unlike the catabolism of alkanoic acids containing an even number of carbon atoms, those other compounds do not require the glyoxylic acid cycle to be assimilated (see below). Moreover, when this isocitrate lyase Ϫ mutant was cultured in MM plus citrate plus acetate plus octanoate (or 8-phenyloctanoate), a higher percentage of PHA (or PHPhA) content was observed (see Table VI). To explain this result, we assume that although some octanoic acid is used as the carbon source and catabolized through ␤-oxidation, the presence of citrate and acetate slows down the metabolic flux of this catabolic pathway, thus channeling the intermediates (3-OH-acyl-CoAs) to the synthesis of polymer.
Taking into account that a blockade in ␤-oxidation could lead to the accumulation of 3-OH-acyl-CoA derivatives that could be used for the synthesis of plastic polymers, we analyzed the accumulation of PHAs and PHPhAs when P. putida U was cultured in MM containing 15 mM octanoic acid or 8-phenyloctanoic acid as the source of intermediates, 5 mM 4-OHphenylacetic acid for supporting bacterial growth, and the ketothiolase inhibitor acrylic acid (0, 5 mM) (85,86). We observed that, as Steinbü chel and co-workers reported in a E. coli recombinant strain (86), accumulation of polymer increased 2-fold (60 and 38% of bacterial dry weight, respectively) with respect to a control cultured in the absence of acrylic acid.
When a PhaZ Ϫ mutant (handicapped in the mobilization of the plastic once synthesized, see Table IV) was cultured in MM supplemented with acrylic acid, the quantity of polymer accumulated was double that observed in the control (without acrylic acid) but lower (65%) than the amount of plastic accumulated by the wild-type P. putida U. These results again support the hypothesis that the polyhydroxyalkanoate biosynthetic enzymes work more efficiently as a complex enzymatic system.
To further analyze the overproduction of PHAs, the fadA gene encoding the 3-oxoacyl-CoA thiolase (accession number AF150672) involved in the ␤-oxidation pathway (see below) was disrupted by homologous recombination by using an internal fragment of the gene, obtained by polymerase chain reaction amplification using primers of the homologous gene of P. fragi (87). We observed that the quantity of PHAs accumulated in this mutant, when it was cultured in MM supplemented with 15 mM octanoic acid and 5 mM 4-HPhAc, was higher (68% dwc) than that observed in the parental strain, showing that a blockade in the ␤-oxidation contributes efficiently to improve the synthesis of plastic polymers.

PhA-specific Catabolic Pathways: The Link with the Phenylacetyl-CoA Catabolon
According to the results presented above, we assumed that the biosynthesis and mobilization of PHPhAs and PHAs in P. putida U must be carried out by the same enzymatic system and that it was similar to those reported in P. oleovorans and in P. aeruginosa (13,21). However, the synthesis of aromatic polyesters is not a common event among the bacteria, although many species are able to accumulate high amounts of aliphatic plastics intracellularly (1). This difference cannot be explained by the requirement of a different acyl-CoA synthetase or different polymerases, since, as we have shown above, the same enzymes participate in the activation and in the polymerization of aromatic and aliphatic monomers. However, the degradation of PhAs containing an even number of carbon atoms (C 2nϩ2 PhAs) by P. putida U seems to require two different pathways: the first, a ␤-oxidation route, that catabolizes PhAs to phenylacetyl-CoA, and a route that must transform this thioester into assimilable metabolites. Under this assumption, it could be argued that only bacteria able to metabolize PhAs would be able to synthesize aromatic plastics (PHPhAs). In fact, the accumulation of these polymers would not represent any advantage if the microbe were unable to catabolize them. Thus, phenyl derivatives containing either an odd or an even number of carbon atoms will be catabolized through ␤-oxidation to benzoyl-CoA (odd) or phenylacetyl-CoA (even), which could be further degraded to the intermediates of general metabolism. Bacteria lacking these additional pathways would only be able to partially catabolize these compounds (releasing some units of acetyl-CoA), and this incomplete degradation would afford scant energetic benefit for the microbe. However, it is not clear whether a single ␤-oxidation pathway exists in P. putida U or whether different pathways might be involved in the catabolism of PhAs and alkanoates in this bacterium. To clarify this point, the different mutants unable to catabolize octanoic acids (Oct Ϫ ) were classified into three groups: (i) those lacking an specific acyl-CoA activating activity (ACS Ϫ also called Oct 1 Ϫ ; see above); (ii) those mutants lacking a functional glyoxylic acid cycle (Oct 2 Ϫ ), and (iii) a different group affected in the ␤-oxidation pathway (Oct 3 Ϫ ) obtained by genetic disruption of the ketothiolase gene (the fadA mutant). We observed that all these strains efficiently catabolized phenylacetic acid and benzoic acid, whereas those named Oct 1 Ϫ and Oct 3 Ϫ were unable to grow in MM containing either aliphatic or phenylalkanoic acids with a number of carbon atoms (even or odd) higher than 4 or did so very poorly. These data allow us to conclude that in P. putida U a single acyl-CoA synthetase and single ␤-oxidation pathway are involved in the catabolism of alkanoic and arylalkanoic acids with the above indicated carbon length.
Recently, an additional pathway required for the specific catabolism of phenylacetic acid (via phenylacetyl-CoA) (56) has been discovered (53). We identified a piece of DNA of about 18 kilobase pairs containing 15 open reading frames that are required for the catabolism of PhAc and phenylacetyl-CoA in P. putida U (53). This catabolic pathway, called the phenylacetyl-CoA catabolon core, appears to be organized in three contiguous operons that contain five different functional units and several regulatory elements (Fig. 8). Using mutagenesis with the transposon Tn5, different mutants of P. putida U unable to grow in MM containing PhAc as the carbon source (PhAc Ϫ strains) were selected. We observed that all of the PhAc Ϫ mutants were able to catabolize n-As and n-PhAs containing an odd number of carbon atoms (C 2nϩ1 PhAs), whereas they assimilated C 2nϩ2 PhAs with an even number very poorly, since they were only able to utilize the acetyl-CoA molecules released by ␤-oxidation (Fig. 3C). Analysis of plastic polymer accumulation in these mutants grown in MM containing 4-hydroxyphenylacetic acid for supporting bacterial growth and 8-phenyloctanoic acid as a plastic precursor revealed that whereas mutations in all of the catabolic genes involved in the specific route of phenylacetic acid did not affect the production of aliphatic polymers (PHAs), most of them (with the exception of the mutants lacking phenylacyl-CoA ligase, an enzyme that is not required for the catabolism of phenylalkanoic acids) (53) accumulated lower quantities of aromatic polymers (PHPhAs) (Table III). These results suggest that both pathways (PHPhAs biosynthetic/degradative route and PhAc catabolic pathway) are closely related, since a blockade in the second one causes the accumulation of some intermediate(s) that negatively affect(s) (probably by feedback control) the activities of the enzymes involved in the process of polymerization (Table III).
We also observed that all the PhAc Ϫ mutants were able to grow in MM containing benzoic acid as the sole carbon source (not shown), indicating that the ␤-oxidation final products of both types of PhAs (benzoyl-CoA/phenylpropionyl-CoA and phenylacetyl-CoA) are catabolized through different pathways. Similar data had been reported in E. coli W, since two unrelated pathways are involved in the degradation of phenylacetic acid and phenylpropionic acid in this bacterium (88,89).

Conclusions
In sum, from the present findings the following conclusions can be drawn. (i) P. putida U synthesizes and accumulates novel biodegradable plastic polymers containing as monomers 3-hydroxy-n-phenyl derivatives (10 Յ n Ն 5). (ii) A single acyl-CoA synthetase seems to be involved in the activation of phenylalkanoics and alkanoic acids to their CoA thioesters (which are the precursors of the different polymers), and a similar transport system seems to be required for the uptake of alkanoic and phenylalkanoic acids. The disruption of the fadD gene handicapped the uptake of precursors, their activation, and, therefore, the synthesis of plastics. (iii) Polymerization of the monomers is carried out by two polymerases (PhaC1 and PhaC2), which are also involved in the synthesis of polyhydroxyalkanoates. These enzymes are encoded by two genes (phaC1 and phaC2) organized in a single operon, which also includes a depolymerase gene (phaC1-phaZ-phaC2) and which is very similar to the pha locus reported for P. oleovorans and P. aeruginosa. Both enzymes (PhaC1 and PhaC2) showed a similar substrate specificity and polymerizing rate, whereas the depolymerizing rate shown by PhaZ is about one half of the activity shown by each of these polymerases. Expression of these enzymes in recombinant E. coli fadB or fadR mutants does not permit the synthesis of polymers, whereas the synthesis and accumulation of PHAs occurred at a very low rate when a monocopy construction was used. (iv) A blockade in a ␤-oxidation gene affecting the expression of ketothiolase (fadA mutant) greatly improves the accumulation of these polymers. Similar results were obtained when the ketothiolase was inhibited with acrylic acid as well as when the metabolic flux through the glyoxylic acid was altered (disruption on the gene encoding the isocitrate lyase). (v) Finally, at least two specific catabolic routes linked to a single ␤-oxidation pathway are required for the complete catabolism of the 3-hydroxyphenylalkanoates containing an even number of carbon atoms. Both are integrated in a complex catabolic unit (phenylacetyl-CoA catabolon) responsible for the degradation of several structurally related aromatic compounds (53).
Deeper knowledge of these biodegradable plastics could be important not only to establish the physicochemical properties of novel, nonpolluting compounds but also to obtain derivatives with new or broader applications. Moreover, expression of the genes encoding the plastic polymerases and depolymerase in other microbes, via genetic engineering (incorporation of genes encoding R-specific enoyl-CoA to the pha locus in order to efficiently channel ␤-oxidation intermediates to the polymeric system), could help to enhance both their biosynthetic and their catabolic potential as well as increase the quantity of biodegradable plastics that could be stored (45). Furthermore, a biochemical approach to the regulatory mechanisms controlling the expression of these genes and study of the proteins participating in both the synthesis and in the stabilization of such polymers could help in the design of genetically engineered microbes that could be employed to eliminate certain aromatic contaminants from the biosphere under a broad range of metabolic and environmental conditions. Thus, we have recently shown that the disruption of the paaN gene, encoding the repressor of the PhAc catabolic pathway (Fig. 8) (53), elicits a dual effect, i.e. the constitutive expression of the pathway and the suppression of the carbon catabolite repression of this route. This result shows that genetically engineered strains could be efficiently used to degrade PhAc and certain aromatic precursors even from complex media containing readily metabolizable carbon sources that in the wild-type strain would strongly repress the PhAc catabolic pathway.
Acknowledgments-Thanks are given to Dr. C. C. DiRusso, Dr. A. Steinbü chel, Dr. B. Rehm, and Dr. Q. Ren for the gift of some of the E. coli strains used in this work and to Dr. J. L. Ferná ndez Puentes (Instituto BIOMAR, Onzonilla, León, Españ a) for the NMR analyses.