INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Vanillin (4-hydroxy-3-methoxybenzaldehyde) is an important aromatic flavor compound that is frequently used in flavored foods and as a fragrance for perfumes. Since vanillin occurs as an intermediate in the catabolism of phenolic stilbenes, eugenol (4-allyl-2-methoxyphenol), ferulate (4-hydroxy-3-methoxycinnamate), and lignin (4, 38, 43, 45), there is widespread interest in producing it from natural raw materials by biotransformation (for an overview, see reference 16).

We are currently investigating biotransformation processes based on the catabolism of eugenol by Pseudomonas sp. strain HR199, which proceeds via ferulate, vanillin, and vanillate (4-hydroxy-3-methoxybenzoate) (33). In the context of developing a biotechnological process for the production of vanillin, the investigation of mutants of Pseudomonas sp. strain HR199 which were impaired in the catabolism of vanillin was of particular interest. The genes encoding vanillin dehydrogenase (vdh) and vanillate-O-demethylase (vanA and vanB) of this strain, which complemented one type of vanillin-negative mutants, were characterized in a recent study (29). In the present study, we characterized additional mutants with defects in the catabolism of vanillin and cloned and molecularly characterized genes which are involved in vanillin degradation in Pseudomonas sp. strain HR199.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The Pseudomonas sp. and Escherichia coli strains and the plasmids used in this study are listed in Table 1.

Growth of bacteria. Cells of E. coli were grown at 37°C in Luria-Bertani medium (34). Cells of Pseudomonas sp. were grown at 30°C either in a nutrient broth medium (0.8%, wt/vol) or in mineral salts medium (MM) (35) supplemented with carbon sources as indicated in the text. Vanillin, vanillate, ferulate, and protocatechuate were dissolved in dimethyl sulfoxide and added to the medium at final concentrations of 0.1% (wt/vol). Eugenol was directly added to the medium at a final concentration of 0.1% (vol/vol). Tetracycline and kanamycin were used at final concentrations of 12.5 and 50 µg/ml for E. coli or 25 and 300 µg/ml for Pseudomonas sp., respectively.

Nitrosoguanidine mutagenesis. The nitrosoguanidine mutagenesis of Pseudomonas sp. strain HR199 was performed as described previously (29).

Preparation of the soluble fraction of crude extracts. Cells were washed twice with 10 mM sodium phosphate buffer (pH 7.5), resuspended in twice the volume of the cell pellet, and disrupted by sonication with a Branson sonifier 250 apparatus (amplitude, 16 µm; 1 min per ml of cell suspension; 20-s bursts). The soluble fraction of the crude extract was obtained by a 1-h centrifugation at 100, 000 × g and 4°C.

Enzyme assays. Protocatechuate 3,4-dioxygenase activity was measured spectrophotometrically by monitoring the decrease of absorbancy at 290 nm by the method of Fujisawa and Hayaishi (13). Protocatechuate 4,5-dioxygenase (4,5-PCD) activity was measured spectrophotometrically by monitoring either the decrease of absorbancy at 250 nm or the increase of absorbancy at 410 nm by the methods described by Ono et al. (26). 2,3-PCD activity was measured spectrophotometrically by monitoring the increase of absorbancy at 350 nm by the method described by Wolgel and Lipscomb (51). One unit of enzyme activity was defined as the conversion of 1 µmol of protocatechuate per min. The protein concentrations were determined as described by Lowry et al. (24) with bovine serum albumin as the standard.

Qualitative and quantitative determination of catabolic intermediates. Culture supernatants were analyzed by high-performance liquid chromatography (HPLC) on a Nucleosil-100 C₁₈ column without prior extraction as described previously (29).

Isolation and identification of 3-carboxy muconolactone. From a culture of mutant SK6190 grown in MM with eugenol as the sole carbon source, 80 ml of cell-free supernatant was prepared, which exhibited only one peak with a retention time of 2 min on HPLC analysis. This supernatant was treated by a method described for the isolation of -ketoadipate (5). The pH of the culture supernatant was adjusted to 2.8 by addition of 2 M phosphoric acid, and the supernatant was then given a 20-min centrifugation at 4°C and 2,772 × g. The supernatant was extracted six times with an equal volume of diethyl ether. The ether phases were combined, the solvent was evaporated, and 44.2 mg of brownish crystals was obtained.

Electrophoretic methods. Proteins were separated under denaturating conditions in 11.5% (wt/vol) polyacrylamide gels by the method of Laemmli (23) and stained with Serva Blue R.

Isolation and manipulation of DNA. Plasmid DNA and DNA restriction fragments were isolated and analyzed by standard methods described in references compiled in a previous study (30).

Transfer of DNA. Competent cells of E. coli were prepared and transformed by the CaCl₂ procedure as described by Hanahan (17). Conjugations of E. coli S17-1 (donor) harboring hybrid plasmids and of Pseudomonas sp. (recipient) were performed on solidified nutrient broth medium as described by Friedrich et al. (12) or by a "minicomplementation method" on solidified MM containing 0.5% (wt/vol) gluconate as the carbon source and 25 µg of tetracycline per ml or 300 µg of kanamycin per ml, as described previously (29).

DNA sequence determination and analysis. The nucleotide sequences were determined with a 4000L DNA sequencer (LI-COR Inc., Biotechnology Division, Lincoln, Nebr.) and a Thermo Sequenase fluorescence-labelled primer cycle-sequencing kit with 7-deaza-dGTP (Amersham Life Science, Little Chalfont, United Kingdom) as specified by the manufacturers. The DNA sequence of E58 was determined by using subcloned fragments as templates and universal and reverse primers. Additional sequences were obtained with synthetic fluorescence-labelled oligonucleotides as primers, employing the primer-hopping strategy (41). Nucleotide and amino acid sequences were analyzed with the Genetics Computer Group sequence analysis software package (GCG Package, version 6.2, June 1990) as described by Devereux et al. (6).

Chemicals. Restriction endonucleases, T4 DNA ligase, lambda DNA, and enzymes or substrates used in the enzyme assays were obtained from C. F. Boehringer & Soehne (Mannheim, Germany) or from GIBCO/BRL-Bethesda Research Laboratories GmbH (Eggenstein, Germany). Agarose type NA was purchased from Pharmacia-LKB (Uppsala, Sweden). Radioisotopes were from Amersham/Buchler (Braunschweig, Germany). Synthetic oligonucleotides were purchased from MWG-Biotech (Ebersberg, Germany). All other chemicals were from Haarmann & Reimer (Holzminden, Germany), E. Merck AG (Darmstadt, Germany), Fluka Chemie (Buchs, Switzerland), Serva Feinbiochemica (Heidelberg, Germany), or Sigma Chemie (Deisenhofen, Germany).

Nucleotide sequence accession number. The nucleotide and amino acid sequence data reported in this paper have been submitted to the EMBL, GenBank, and DDBJ nucleotide sequence databases and are listed under accession no. Y18527.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of the genes involved in the vanillin degradation pathway. Pseudomonas sp. strain HR199 is able to grow on vanillin as the sole carbon source. To identify the genes which are essential for vanillin degradation, nitrosoguanidine-induced mutants were isolated and screened on solidified mineral medium containing vanillin or protocatechuate as the sole carbon source. Three independent mutants (SK6169, SK6184, and SK6190) which had lost their ability to grow on vanillin but were still able to grow on protocatechuate were identified. These mutants were not phenotypically complemented after conjugational reception of plasmid pE207 (29), encoding vanillin dehydrogenase and vanillate-O-demethylase. Since vanillin is converted via vanillate to protocatechuate in the wild type, the phenotype of these mutants was obscure.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Organization of the pcaH and pcaG genes of Pseudomonas sp. strain HR199. (a) Physical map of fragment E58. (b) Subfragments relevant for complementation and heterologous expression studies. (c) Identified genes and ORFs of a putative 4-hydroxybenzoate 3-hydroxylase (ORF5), a putative PobR (ORF4), a putative PcaQ (ORF3), and 3,4-PCD (pcaHG).

Subcloning of pcaH and pcaG. Plasmids pSKE58-1 and pSKE58-2 were digested with KpnI, and the resulting 1.9-kbp (EK19), 1.75-kbp (K18), and 2.15-kbp (KE22) KpnI fragments were cloned in pMP92. After conjugative transfer of the resulting plasmids from corresponding E. coli S17-1 strains to the "vanillin-negative" mutant SK6169, complementation was achieved only with fragment KE22. By subcloning of the PstI fragments of KE22 in pMP92 and subsequent conjugative transfer to the mutant SK6169, the complementing region was assigned to a 1.6-kbp KpnI-PstI (KP16) subfragment of KE22 (Fig. 1).

Nucleotide sequence of fragment E58. The nucleotide sequence of the entire E58 fragment was determined (Fig. 2). An open reading frame (ORF) of 720 bp (ORF1), whose putative translational product exhibited significant homology to  subunits of 3,4-PCD from various other bacteria (Fig. 3) and which was therefore referred to as pcaH, was identified on fragment KP16. Thus, mutant SK6169 most probably lacked a functional  subunit of 3,4-PCD. At 12 bp downstream of the translational stop codon of pcaH at position 4789 (Fig. 2), the ATG start codon of a second ORF of 606 bp (ORF2) was identified, which was referred to as pcaG since its putative translational product exhibited significant homology to  subunits of 3,4-PCD (Fig. 3). Typical Shine-Dalgarno sequences, AGGAG or AGGAGG, preceded the ATG-start codons of pcaH and pcaG at distances of 6 or 7 nucleotides, respectively. An inverted repeat, which may represent a factor-dependent transcriptional terminator, was identified 20 bp downstream from the translational stop codon of pcaG (Fig. 2). The free energy of this structure is approximately 70.2 kJ/mol according to Tinoco et al. (44).

View larger version (151K): [in this window] [in a new window]   FIG. 2.   Nucleotide sequence of fragment E58. Amino acids deduced from the nucleotide sequence are specified by standard one-letter abbreviations. The putative ribosome-binding sites are overlined and indicated by S/D. The position of the hairpinlike structure downstream of pcaG is indicated by arrows.

View larger version (30K): [in this window] [in a new window]   FIG. 3.   Relationship of Pseudomonas sp. strain HR199 3,4-PCD and 3,4-PCDs from other sources. The dendrogram was constructed with the CLUSTAL program from pairwise similarity scores generated by the method of Wilbur and Lipman (49) with the following parameters: k-tuple length, 1; gap penalty, 3; number of diagonals, 5; diagonal window size, 5; gap-opening penalty, 10; gap extension penalty, 0.10; protein weight matrix, blosum. Relatedness is represented by the branch length (distances are given as 102 percent divergence in parentheses). The amino acid sequences of the  and  subunits of 3,4-PCD from P. putida (pironly:DAPSAA; pironly:DAPSBA) (11), Acinetobacter calcoaceticus (Swissprot:PCXA_ACICA; Swissprot:PCXB_ACICA) (18), and Burkholderia cepacia (Swissprot:PCXA_BURCE; Swissprot:PCXB_BURCE) (53) were compared with the  and  subunits of 3,4-PCD from Pseudomonas sp. strain HR199 deduced from pcaG and pcaH.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Relationship of the amino acid sequence deduced from ORF3 from Pseudomonas sp. strain HR199 and four representative transcriptional activator proteins of the LysR family from different sources. The program and parameters for the construction of the dendrogram were the same as for Fig. 3. The amino acid sequences of PcaQ from A. tumefaciens (Swissprot:PCAQ_AGRTU) (27), of TdcA from E. coli (Swissprot:TDCA_ECOLI) (15), of CynR from E. coli (Swissprot:CYNR_ECOLI) (42), and of OxyR from Mycobacterium avium (Swissprot:OXYR_MYCAV) (37) were compared with the amino acid sequence of Pseudomonas sp. strain HR199 deduced from ORF3.

Deduced properties of the pcaH and pcaG gene products. The relative molecular masses of the  and  subunits of 3,4-PCD, calculated from the amino acid sequence deduced from pcaG and pcaH, were 22,364 and 26,800 Da, respectively. These values corresponded well to those reported for other 3,4-PCDs, e.g., 22,300 and 26,600 Da for the 3,4-PCD subunits from P. putida (11).

Putative functions of the pcaH and pcaG gene products. The amino acid sequences deduced from pcaG and pcaH were compared with those collected in GenBank. The highest homology was achieved to the 3,4-PCD of P. putida (81 and 88% identical amino acids for the  and  subunits, respectively). As reported for other 3,4-PCDs, there was a high degree of homology between the  and  subunit of this enzyme from Pseudomonas sp. strain HR199 (36.2% identical amino acids), which was also obvious at the DNA level, where the identity of the nucleotide sequences of pcaG and pcaH was 58%. The relationship of the  and  subunit of 3,4-PCD from Pseudomonas sp. strain HR199 to 3,4-PCDs from other sources is shown in Fig. 3.

Expression of pcaHG from Pseudomonas sp. strain HR199 in the "vanillin-negative" mutant SK6169, and heterologous expression in E. coli. The 2.2-kbp KpnI-EcoRI subfragment (KE22) of E58 (Fig. 1) was cloned in the vector pMP92, which was then conjugatively transferred from corresponding E. coli S17-1 strains to the mutant SK6169. This mutant was not able to grow on solidified medium with vanillin as the carbon source and lacked 3,4-PCD activity. Plasmid pMPKE22 restored the ability to grow on vanillin in the corresponding transconjugants. Moreover, these transconjugants exhibited 3,4-PCD activities comparable to the wild type (Table 2).

Identification of ORFs exhibiting homology to pobA and pobR. Upstream of ORF3, two additional ORFs of 882 and 1,185 bp were identified (Fig. 1), which were preceded by putative Shine-Dalgarno sequences (Fig. 2) and showed all the features typical of coding regions in Pseudomonas sp. strain HR199. Comparison of the amino acid sequences deduced from ORF4 and ORF5 revealed extended homologies to PobR from Pseudomonas aeruginosa (49.3% identical amino acids) (Fig. 5) and to 4-hydroxybenzoate 3-hydroxylases from different sources (Fig. 6), respectively.

View larger version (20K): [in this window] [in a new window]   FIG. 5.   Relationship of the amino acid sequence deduced from ORF4 from Pseudomonas sp. strain HR199 and four representative regulatory proteins from different sources. The program and parameters for the construction of the dendrogram were the same as for Fig. 3. The amino acid sequences of PobR from P. aeruginosa (pironly:S41382) (10), of PobR from Rhizobium leguminosarum (U40388) (52), of HpaA from E. coli (pironly:A55349) (32), of AraC from E. coli (Swissprot:ARAC_ECOLI) (48), and of PobR from Acinetobacter calcoaceticus (L13114) (8) were compared with the amino acid sequence of Pseudomonas sp. strain HR199 deduced from ORF4.

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Relationship of the amino acid sequence deduced from ORF5 from Pseudomonas sp. strain HR199 with 4-hydroxybenzoate 3-hydroxylases (PobA) from other sources. The program and parameters for the construction of the dendrogram were the same as for Fig. 3. The amino acid sequences of PobA from P. aeruginosa (Swissprot:PHHY_PSEAE) (9), P. fluorescens (Swissprot:PHHY_PSEFL) (46), and Acinetobacter calcoaceticus (Swissprot:PHHY_ACICA) (7) were compared with the amino acid sequence of Pseudomonas sp. strain HR199 deduced from ORF5.

Analysis of mutant SK6169, which exhibited a defective 3,4-PCD but was still able to grow on protocatechuate. The results of the genetic analysis of mutant SK6169 revealed a defect in the  subunit of 3,4-PCD. In consequence, this mutant lacked 3,4-PCD activity (Table 2). Nevertheless, it was able to grow on eugenol, ferulate, vanillate, or protocatechuate as the sole carbon source. To investigate this obscure phenotype in more detail, the mutant was grown in the presence of vanillin, protocatechuate, or vanillin plus gluconate in liquid MM. These investigations revealed that mutant SK6169 grew as well as the wild type on vanillin plus gluconate and was still able to grow on vanillin without an additional carbon source. However, in comparison to the wild type, the mutant grew only after a long lag phase of about 20 h. In contrast to the analysis on MM agar plates, the growth of the mutant in liquid MM with protocatechuate as the sole carbon source was reduced compared to the wild type. As revealed by HPLC analysis of culture supernatants, vanillin was converted to vanillate and protocatechuate by this mutant, which was further metabolized. In comparison to the wild type, the occurrence of these intermediates was retarded. To exclude the appearance of revertants, aliquots of the cultures were spread on vanillin- or protocatechuate-containing MM agar plates, respectively. No growth of revertants on vanillin agar plates could be observed, and there was no difference in the growth of the wild type and the mutant on protocatechuate-containing agar plates.

Analysis of mutants SK6184 and SK6190, which were also impaired in the catabolism of vanillin. During the nitrosoguanidine mutagenesis, mutants SK6184 and SK6190, which had lost their ability to grow on vanillin, vanillate, and ferulate but were still able to grow on eugenol or protocatechuate, were isolated. These mutants were not complemented by the hybrid cosmids pV372 and pV801 after conjugative reception. When these mutants were cultivated in MM with eugenol, vanillin, or protocatechuate as the sole carbon source, the substrates were completely converted to an unknown substance. To identify this product, we isolated it from the culture supernatant of mutant SK6190 grown in MM with eugenol. The light brownish crystals we obtained exhibited a blurred melting point of 114 to 150°C. The isolated substance was soluble in H₂O, methanol, ethyl acetate, and diethylether. Its mass spectrum showed a molecular ion at m/z 186. From these results along with the data obtained from ¹H NMR (four protons) and ¹³C NMR (seven carbon atoms) spectra and the infrared spectrum (OH group[s], three carbonyl bands), the molecular formula was found to be C₇H₆O₆. The ¹H NMR spectrum of the substance showed an ABXZ spin system (J_AB = 16.6 Hz, J_AX = 8.0 Hz, J_BX = 3.2 Hz) at 2.680, 3.207, and 5.566 ppm, with an allylic coupling (2.1 Hz) between the X proton and the olefinic proton at 6.678 ppm. These findings established the structure fragments shown in Fig. 7. The ¹³C NMR spectrum showed seven signals at 173.1, 172.6, 163.7, 160.1, 127.0 (==CH) 80.6 (CH), and 37.9 (CH₂) ppm. Correlations between ¹H and ¹³C extracted from the HMQC spectrum are shown in Fig. 7. Because the protons at C-3 are diastereotopic, an asymmetrically substituted carbon atom must be in the direct neighborhood. C-1 must be part of a double bond with a quaternary carbon atom (160.1 ppm) as a partner. H-1 must be in the -position to a carbonyl function because of its high shift value. The shift value of 80.6 ppm for C-2 indicates that it must be bonded to an oxygen atom. From these data, the chemical structure of the isolated substance could be assigned to 3-carboxy muconolactone (Fig. 7). Furthermore, the obtained spectral data are in good agreement with the values published by Kirby et al. (21).

View larger version (15K): [in this window] [in a new window]   FIG. 7.   Identification of the substance accumulated by mutants SK6184 and SK6190 during growth on eugenol. (A) Structure fragments established by NMR spectroscopy: 1H-NMR data (CD3OD),  [ppm], J [Hz]. 13C NMR shift values (ppm) are shown in parentheses. (B) Chemical structure of 3-carboxy muconolactone.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A 5.8-kbp EcoRI fragment cloned from genomic DNA of Pseudomonas sp. strain HR199 was found to encode proteins which are essentially involved in the degradation of vanillin. A "vanillin-negative" mutant which lacked 3,4-PCD activity was phenotypically complemented by the structural gene pcaH, encoding the  subunit of 3,4-PCD. The structural gene of the corresponding  subunit, pcaG, was localized downstream of pcaH. The amino acid sequences deduced from pcaH and pcaG showed extended homology to the 3,4-PCD from P. putida (11). pcaHG were expressed in E. coli driven by the lac promoter of pBluescript SK. The 3,4-PCD activity was 13-fold higher than that obtained by the expression of pcaHG of P. putida in E. coli (11). Since the G+C contents and the codon bias of the pcaHG genes of Pseudomonas sp. strain HR199 and P. putida were very similar, it is unlikely that the weak expression of the P. putida genes in E. coli is due to these properties, as concluded by Frazee et al. (11).

Upstream of pcaHG was an ORF which had the same orientation and whose deduced amino acid sequence exhibited strong homology to the transcriptional activator protein PcaQ from Agrobacterium tumefaciens (27). The region of highest homology was located in the N-terminal region as described by Viale et al. (47) for transcriptional activator proteins belonging to the LysR family, comprising the helix-turn-helix motif (20) for DNA binding.

In the immediate neighborhood of the pca genes were found two ORF that exhibited high homologies to the pob genes responsible for p-hydroxybenzoate metabolism. The first ORF exhibited high homology to the PobR regulator protein of A. tumefaciens (28), which belongs to transcriptional regulator proteins of the XylS/AraC type (14). As also observed for the PobR regulator protein of A. tumefaciens (28), no homology was obtained with the transcriptional regulator proteins of the so-called PobR subfamily (8), which refers to PobR of Acinetobacter calcoaceticus. The amino acid sequence of the second ORF exhibited 77% identity to PobA from P. aeruginosa and P. fluorescens and 61% identity to PobA from Acinetobacter calcoaceticus and most probably resembles the typical NADPH-dependent 4-hydroxybenzoate 3-hydroxylase of the P. aeruginosa type (36).

In the wild-type Pseudomonas sp. strain HR199, vanillin is catabolized by oxidation to vanillate, which is subsequently demethylated to yield protocatechuate (29). Since cells grown on these substrates exhibited 3,4-PCD activity, the catabolism most probably proceeds via the ortho-cleavage pathway in this bacterium. The "vanillin-negative" mutant SK6169 exhibited a defect in the  subunit of 3,4-PCD, thus lacking the corresponding enzyme activity. Nevertheless, this mutant was able to grow on protocatechuate. Since cells grown on this substrate exhibited neither 3,4-PCD, 4,5-PCD, nor 2,3-PCD activities, mutant SK6169 must have another mechanism for protocatechuate catabolism, which might be due to activities of other ring-cleaving enzymes which were not covered by the applied spectrophotometric assays. Two other "vanillin-negative" mutants accumulated 3-carboxy muconolactone in the medium during growth on eugenol. The occurrence of this intermediate seems to be due to a lactonization of 3-carboxy-cis,cis-muconate by a syn absolute stereochemical course. This kind of cyclization is observed only in fungi (25), whereas in bacteria the lactonization catalyzed by 3-carboxy-cis,cis-muconate lactonizing enzyme (CMLE) causes an anti addition (3, 50), resulting in 4-carboxy muconolactone as the product. In contrast, the lactonization of cis,cis-muconate, catalysed by the cis,cis-muconate lactonizing enzyme (MLE) proceeds by a syn addition (19). The accumulation of 3-carboxy muconolactone by the aforementioned mutants could be a hint that Pseudomonas sp. strain HR199 possesses an unusual "syn-CMLE" and that the mutants exhibited a defect in a following step of the degradation pathway. Another explanation would be an alteration of the CMLE gene by mutation, changing the "anti-CMLE" to a "syn-CMLE" in the mutants, resulting in the accumulation of the carboxy muconolactone with nonstandard regiochemical properties. Since the "vanillin-negative" phenotypes of the investigated mutants were due to defects in 3,4-PCD or in enzymes of the protocatechuate branch of the ortho-cleavage pathway, vanillin is degraded via this pathway in the wild type.