Biochemical and Molecular Characterization of a Ring Fission Dioxygenase with the Ability to Oxidize (Substituted) Salicylate(s) from Pseudaminobacter salicylatoxidans*

The gene coding for a dioxygenase with the ability to cleave salicylate by a direct ring fission mechanism to 2-oxohepta-3,5-dienedioic acid was cloned from Pseudaminobacter salicylatoxidans strain BN12. The deduced amino acid sequence encoded a protein with a molecular mass of 41,176 Da, which showed 28 and 31% sequence identity, respectively, to a gentisate 1,2-dioxygenase from Pseudomonas alcaligenes NCIMB 9867 and a 1-hydroxy-2-naphthoate 1,2-dioxygenase from Nocardioides sp. KP7. The highest degree of sequence identity (58%) was found to a presumed gentisate 1,2-dioxygenase from Corynebacterium glutamicum. The enzyme from P. salicylatoxidans BN12 was heterologously expressed in Escherichia coli and purified as a His-tagged enzyme variant. The purified enzyme oxidized in addition to salicylate, gentisate, 5-aminosalicylate, and 1-hydroxy-2-naphthoate also 3-amino- and 3- and 4-hydroxysalicylate, 5-fluorosalicylate, 3-, 4-, and 5-chlorosalicylate, 3-, 4-, and 5-bromosalicylate, 3-, 4-, and 5-methylsalicylate, and 3,5-dichlorosalicylate. The reactions were analyzed by high pressure liquid chromatography/mass spectrometry, and the reaction products were tentatively identified. For comparison, the putative gentisate 1,2-dioxygenase from C. glutamicum was functionally expressed in E. coli and shown to convert gentisate but not salicylate or 1-hydroxy-2-naphthoate.

The oxygenolytic cleavage of the aromatic nucleus by bacteria requires in most cases the presence of two hydroxy groups attached to the aromatic ring (1)(2)(3). Only a few examples have been described previously in which monohydroxylated aromatic compounds were cleaved by ring fission dioxygenases, and in most of these examples aminohydroxybenzene derivatives were observed as ring fission substrates. The ability of "traditional" ring fission dioxygenases to oxidize aminohydroxybenzene derivatives is mechanistically easily explained, because the amino group activates the aromatic nucleus in a similar way as the hydroxy group for an electrophilic attack of the ring-cleaving dioxygenases (4 -9). There are also a few examples that describe the ring fission of monohydroxylated aromatic compounds that do not possess a second electron-donating substituent. This has been described for the oxidation of 5-chlorosalicylate by a "Bacillus" sp. and for the conversion of 1-hydroxy-2-naphthoate by several Gram-negative (e.g. Aeromonas sp.) and Gram-positive bacteria (Nocardioides sp.) From these two reactions only the oxidation of 1-hydroxy-2-naphthoate by Nocardioides sp. KP7 has been analyzed on an enzymatic and genetic level, and the ring fission product was isolated and characterized by various spectroscopic techniques (10 -14).
We recently described a new ring fission dioxygenase from the naphthalenesulfonate-degrading strain Pseudaminobacter salicylatoxidans, which oxidized salicylate by a novel ring fission mechanism to 2-oxohepta-3,5-dienedioic acid (Scheme 1). The ring fission dioxygenase resembled gentisate 1,2-dioxygenases or 1-hydroxy-2-naphthoate dioxygenases, because of the ability of the enzyme to convert gentisate and 1-hydroxy-2naphthoate, the size of the subunits, the structure of the holoenzyme, and the dependence of the enzyme on Fe 2ϩ ions (15). In order to allow a more detailed analysis of the ability of ring fission dioxygenases to oxidatively cleave the aromatic ring of monohydroxylated benzene derivatives in the current study, the encoding gene was cloned and the substrate specificity of the enzyme analyzed in greater detail.

MATERIALS AND METHODS
Bacterial Strains and Media-The isolation and characterization of P. salicylatoxidans strain BN12 DSM 6986 T has been described previously (16,17). Corynebacterium glutamicum ATCC 13032 was obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany. Escherichia coli JM109 and E. coli BL21(DE3) were used as host strains for recombinant DNA work.
Mineral media were prepared as described by Dorn et al. (18) and were supplemented with 100 mg/liter of yeast extract. P. salicylatoxidans was routinely grown in this supplemented mineral medium with 5 mM 6-aminonaphthalene-2-sulfonate. For the cultivation of C. glutamicum, a complex medium proposed by the Deutsche Sammlung von Mikroorganismen und Zellkulturen was used. The recombinant E. coli strains were routinely cultured in Luria-Bertani medium supplemented with ampicillin (100 g/ml).
Oxygen Uptake Experiments-P. salicylatoxidans BN12 was grown in a mineral medium with 6-aminonaphthalene-2-sulfonate (5 mM) supplemented with 100 mg/liter yeast extract. The cells were harvested during the late exponential growth phase by centrifugation and resuspended in Tris-HCl buffer (20 mM, pH 8) to an optical density (A 546 nm ) of 14. The resting cells were incubated at 30°C in an oxygen electrode (YSI 5350, YSI Inc., Yellow Springs, OH). The endogenous respiration was determined for 2 min, and then the respective salicylates were added (1 mM each). The oxygen uptake was recorded for 5 min, and the reaction rates were corrected for the endogenous respiration.
Preparation of Cell Extracts-Cell suspensions in 20 mM Tris-HCl buffer (pH 8.0) were disrupted by using a French press (Aminco, Silver Springs, MD) at 80 MPa. Cell debris was removed by centrifugation at * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY323951.
¶ To whom correspondence should be addressed.  (19) with bovine serum albumin as a standard.
Protein Purification, Enzymatic Cleavage of the Protein, Isolation of Peptides, and Sequencing of Peptides and Amino Termini-The dioxygenase was purified from P. salicylatoxidans BN12 by fast protein liquid chromatography as described previously (15). The digestion of the purified dioxygenase by trypsin and the subsequent separation of tryptic digests by reversed-phase HPLC 1 were performed by established procedures (20). The amino acid sequences were determined by automated Edman degradation by using an ABI 476 protein sequencer (Applied Biosystems, Foster City, CA).
PAGE-SDS-PAGE was performed by the method of Laemmli (21), and the gels were routinely stained with Coomassie Blue.
Enzyme Assays with Purified Enzyme Preparations-One unit of enzyme activity was defined as the amount of enzyme that converts 1 mol of substrate/min. The conversion of 1-hydroxy-2-naphthoate, gentisate, 5-aminosalicylate, and salicylate were determined spectrophotometrically as described previously (15).
The conversion of all other salicylates and benzoic acids was analyzed by HPLC. The reaction mixtures contained, in a total volume of 1 ml, 20 mM Tris-HCl (pH 8.0) and the respective substrates (0.1 mM each). The reactions were started by the addition of the purified enzyme (0.04 -0.1 mg of protein). After 10 -30 s, 50 l of the reaction mixtures were removed and the reactions terminated by the addition of 5 l of 1 M HCl or by boiling for 10 min (for the experiments with 3,5-dichloro-, 3,5-dibromo-, and 3,5-diiodosalicylate). The precipitated proteins were removed by centrifugation in an Eppendorf centrifuge (2 min at 14,000 rpm). The concentrations of the substrates were subsequently analyzed by HPLC.
For the determination of the kinetic V max and K m values, the kinetic parameters were determined basically as described above but using a substrate concentration range from 0.01 to 5 mM.
Analytical Methods-The turnover of the substituted salicylates was analyzed by reversed-phase HPLC (HPLC pumps model 510 equipped with a photo-diode array detector model 996 and Millenium Chromatography Manager 2.0, Waters, Milford, MA.). A reversed-phase column (125 ϫ 4.0 mm, internal diameter), packed with 5-m particles of Merck Lichrospher RP8 (end capped) was used. The separated compounds were detected photometrically at 210 nm and at the wavelength indicated in Table I by using a photodiode array detector. Liquid Chromatography-Mass Spectrometry-Product identification was performed by liquid chromatography-mass spectrometry (HP1100, Agilent) coupled to a triple quadrupole mass spectrometer (Quattro LC, Micromass, Manchester, UK) using electrospray ionization in the negative ion mode. Substrate solutions before and after addition of the enzyme and 20 min after addition of the enzyme were injected (20 l) into the HPLC system without any pretreatment. Analytes were separated by ion-pair chromatography on a Luna C18 (2) 3-m column, 15 cm ϫ 3 mm inner diameter at 40°C. Eluent A was H 2 O/MeOH (80:20 v/v), and eluent B was H 2 O/MeOH (5:95 v/v) with 1 mM tributylamine and 1 mM acetic acid each. Gradient elution started with 20% (v/v) eluent B at 0 min, 90% eluent B at 11 min, isocratic to 15 min, and 16 min 20% (v/v) eluent B, 21 min, 20% (v/v) eluent B. A diode array detector and the MS were coupled in a series. The mass spectrometric interface was operated at a cone voltage of 18 V and a capillary voltage of 2.9 kV. Probe temperature was 220°C, and source block temperature was 120°C. Product ion spectra were recorded at collision energies of 10 and 15 eV with a scan rate of 0.5 s.
PCR-Oligonucleotides were custom-synthesized according to the known or deduced sequences of the amino-terminal amino acid sequence and various internal peptides of the ring fission dioxygenase. PCR mixtures (25-50 l) for the amplification of genomic DNA contained 2 mM of each primer, 10 -100 ng of genomic template DNA, 0.1 mM of each deoxynucleotide triphosphates, 1.5 mM MgCl 2 , 0.5 units of Taq (Eppendorf, Hamburg, Germany) or Pwo (Peqlab, Erlangen, Germany) DNA polymerase and the corresponding reaction buffers.
For the amplification reaction with the primers deduced from the amino terminus and the peptide P28, the following PCR program was used: an initial denaturation (94°C, 3 min) was followed by 35 cycles consisting of an annealing temperature of 38.7°C (30 s), a polymerization step (72°C, 2 min), and denaturation (94°C, 30 s). The last polymerization step was extended to 15 min.
For the determination of the complete sequence of the gene encoding the ring fission dioxygenase, a partial inverse PCR was performed (22). The template was prepared by digesting the chromosomal DNA of strain BN12 with PsuI (for the 3Ј-sequence) or NcoI (for the 5Ј-sequence), each of which possessed one restriction site within the 0.7-kb sequence obtained from the initial PCR experiment. The fragments obtained were religated using T4 DNA ligase. Thus intramolecular ligation of these DNA fragments resulted in circular DNA molecules, which were then used as a template for the following PCR. Primers for the PCR experiments were deduced from the sequence of the incomplete gene present on the 0.7-kb PCR product previously obtained and were facing in both directions outward from the known DNA sequence. The following nucleotide primers were used for the amplification of the DNA sequence 3Ј of the known part of the dioxygenase gene: Paul-fwd, 5Ј-ATTGGGGCCTATCGCTGG-3Ј, and Paul-rev, 5Ј-CTGATCGGT-GTCGTTGTGG-3Ј. For the determination of the nucleotide sequence encoding the amino terminus of the protein, the primers NcoI-fwd 5Ј-CGCATGTCTCGTGGC-3Јand NcoI-rev 5Ј-CTTCGGCTGCATTGC-3Ј were used. For the amplification reactions the following PCR programs were used: an initial denaturation (94°C, 3 min) was followed by 35 cycles consisting of an annealing temperature of 55 (for the 5Ј-sequence) or 60°C (for the 3Ј-sequence) (30 s each), a polymerization step (72°C, 2.5 min), and denaturation (94°C, 30 s). The last polymerization step was extended to 15 min. The PCR products were initially cloned into the T-tailed EcoRV-site of pBluescript II KS(ϩ) (23).
Expression of the Dioxygenase in E. coli-For expression in E. coli, the dioxygenase gene was inserted in the plasmid vector pET28a (Novagen, Madison, WI) under the control of the T7 promoter. The DNA segment encompassing the dioxygenase gene was amplified by PCR (using a Pwo DNA-polymerase; Peqlab). The upstream primer (5Ј-GGAGGTCCATATGCAGAACG-3Ј) incorporated an NdeI site (underlined), and the downstream primer (5Ј-CGGGATCCTCACTTCTGC-CCCTCG-3Ј) incorporated a BamHI site (underlined). The amplified product was cloned into the EcoRV site of pBluescript II KS(ϩ). The resulting plasmid was cleaved with NdeI and BamHI and the DNA fragment with the dioxygenase gene ligated into pET28a, which was also previously cut with NdeI and BamHI. The resulting plasmid (pJPH100exN) was subsequently transformed into E. coli BL21(DE3). The expression of the dioxygenase gene was induced by isopropyl-1thio-␤-D-galactopyranoside as suggested by the supplier of the pET system.
Purification of the His-tagged Enzyme-Cell extracts of E. coli JM109(pJPH100exN) were prepared in Tris-HCl buffer (20 mM, pH 8) as described above. The nickel-nitrilotriacetic acid-agarose (Qiagen, Hilden, Germany) was suspended in Tris-HCl (20 mM, pH 8) and transferred (2 ml) to an empty 10-ml polypropylene column. The filled column was equilibrated with a buffer system (pH 8) consisting of Tris-HCl (20 mM), NaCl (300 mM), and imidazole (50 mM). The cell extracts (about 100 mg of protein) were applied to the column, and the protein was eluted with subsequent steps of the Tris-HCl/NaCl buffer (2-4 ml each) with increasing imidazole concentrations (50, 60, 150, and 500 mM). The fractions with dioxygenase activity eluted at an imidazole concentration of 150 mM. The imidazole was removed from the active fractions on a "HiTrap desalting column" (Amersham Biosciences) using Tris-HCl (20 mM, pH 8) plus 100 mM NaCl as eluent buffer.
Cloning of the Presumed Gentisate 1,2-Dioxygenase from C. glutamicum-The gene was amplified from the genomic DNA of C. glutamicum by using the oligonucleotide primers GDO-Cglu-fwd 5Ј-CACATAT-GGGCGCCCCAGG-3Ј and GDO-Cglu-rev 5Ј-CAGGATCCCTAGATTC-CTTCCGGAG-3Ј, which simultaneously introduced NdeI and BamHI cleavage sites (underlined). The following PCR program was used: an initial denaturation (94°C, 3 min) was followed by 30 cycles consisting of annealing at 54.5°C (for 30 s), polymerization at 72°C (for 90 s), and denaturation at 94°C (for 30 s). The last polymerization step was extended to 5 min. The PCR product was initially cloned into the T-tailed EcoRV site of pBluescript II KS(ϩ). Finally, the gene was isolated from positive clones by using the restriction enzymes NdeI and BamHI and cloned into the expression vector pJOE2702 under the control of an L-rhamnose-inducible promoter (24,25).
DNA Sequencing and Nucleotide Sequence Analysis-The DNA sequence was determined by dideoxy chain termination with doublestranded DNA of clones and overlapping subclones in an automated DNA sequencing system (ALFexpress-Sequencer, Amersham Biosciences) with fluorescently labeled primers or nucleotides.
Sequence analysis, data base searches, and comparisons were done using the NCBI facilities. The alignment of the ring fission dioxygenases was obtained with the program Clustal using the default parameters.

Oxidation of Different Salicylates by Resting Cells of P. salicylatoxidans BN12
-It was demonstrated previously that the enzyme from P. salicylatoxidans BN12 oxidized salicylate, gentisate, 5-aminosalicylate, and 1-hydroxy-2-naphthoate by an 1,2-dioxygenolytic cleavage (15). This ring fission mechanism principally allows the cleavage of various substituted salicylates. Therefore, resting cells of P. salicylatoxidans BN12 were incubated with a wide range of substituted salicylates, and the substrate-dependent oxygen uptake rates were determined. These experiments suggested that the bacteria were also able to oxidize a wide range of methyl-, chloro-, or bromo-substituted salicylates. Most surprisingly, the cells oxidized 5-fluorosalicylate with its exceptionally electronegative fluorine substituent with higher rates than salicylate (Fig. 1).
Cloning of the Gene Encoding the Salicylate 1,2-Dioxygenase Activity-The ring fission dioxygenase was purified from cell extracts of P. salicylatoxidans BN12 as described previously (15). The purified dioxygenase was digested with trypsin, and several peptides were isolated by HPLC. The amino acid se-quences of the amino terminus and five internal fragments were determined and used for the design of oligonucleotide primers for PCR experiments (Table II). Using genomic DNA of P. salicylatoxidans BN12 as template and primers derived from the amino-terminal sequence and of the peptide P28 (Table II), a DNA fragment with a size of about 0.7 kb was amplified. The amplified fragment was sequenced, and the sequence obtained was used to complete the sequence of the gene by using partial inverse PCR (see "Material and Methods"). Thus a continuous stretch of DNA of 1249 bp was obtained. The gene for the salicylate dioxygenase was unequivocally identified in this sequence by the presence of the amino-terminal region and the internal peptides determined before by Edman degradation (Fig. 2).
Sequence Analysis-The size of the gene encoding the ring fission dioxygenase was 1107 bp, and it demonstrated a GC content of 62.6% which is in good agreement with the total GC content of the organism (63.9 mol % (16)). The gene encoded a protein consisting of 368 amino acids which corresponded to a molecular mass of 41,176 Da. This value agreed with the molecular mass of the dioxygenase subunits (45 kDa) determined earlier by SDS-gel electrophoresis after purification of the enzyme from P. salicylatoxidans BN12 (15). The deduced complete amino acid sequence showed significant sequence similarities with different gentisate 1,2-dioxygenases and 1-hydroxy-2-naphthoate 1,2-dioxygenases from different bacteria (Fig. 3). The highest degree of sequence identity (31%) to proteins with a known function was observed with a gentisate 1,2-dioxygenase from Pseudomonas alcaligenes NCIMB 9867 (27). Yet a much higher degree of sequence identity (58%) was found to a presumed gentisate 1,2-dioxygenase that had been identified in C. glutamicum in the course of a genome project.
Expression of the Dioxygenase Gene in E. coli-The dioxygenase gene was amplified by PCR from the genomic DNA of strain BN12 by using a set of primers that created new NdeI and BamHI restriction sites. The amplified fragment was then ligated into plasmid pBluescript II SKϩ (previously cut with It was demonstrated previously that the conversion of salicylate, 5-amino-, and 5-hydroxysalicylate and also 1-hydroxy-2-naphthoate could be analyzed spectrophotometrically because of the pronounced UV-visible spectra of the products formed (15). Therefore, the reactions of other salicylates were also analyzed by UV-visible spectroscopy. Thus it was found that cell extracts of the recombinant E. coli strain converted 3-, 4-, and 5-substituted amino-, hydroxy-, chloro-, bromo-, and methylsalicylates (Fig. 3). The conversion of all isomers of the chloro-and bromosalicylates (and also of 5-fluorosalicylate which was the only available fluorosalicylate) resulted in the formation of products with absorption maxima in the range of max ϭ 290 -302 nm. The turnover of the methylsalicylates also resulted in the initial formation of new absorbance maxima in the range from 293 to 300 nm. The changes in the spectra observed during the oxidation of the chloro-, bromo-, and methylsalicylates clearly resembled the changes in the UV-visible spectrum determined previously for the formation of 2-oxohepta-3,5-dienedioic acid ( max ϭ 283 nm), the product formed from salicylate by the same enzyme (15). The observed bathochromic shifts for the methylated and halogenated derivatives of 2-oxohepta-3,5-dienedioic acid corresponded to the established Woodward diene rules (28).
The changes of the UV-visible spectra observed during the conversion of the isomeric amino-and hydroxysalicylates were different from those observed with the methylated and halogenated salicylates (Fig. 3) but also were consistent with a 1,2dioxygenolytic cleavage of these substrates. Thus, the turnover of gentisate and 5-aminosalicylate resulted in the formation of new absorption maxima at max ϭ 335 and 352 nm, respectively. These changes in the UV-visible spectra had been described previously for the 1,2-dioxygenolytic cleavage of these two substrates (5,8,11). The products formed from 4-hydroxyand 4-aminosalicylate demonstrated similar absorption maxima at max ϭ 336 and 344 nm.
The oxidation of 3-hydroxysalicylate resulted in the formation of a pronounced new absorption maximum at max ϭ 393 nm. The extraordinary bathochromic shift observed in this reaction could possibly be explained by the almost symmetrical structure of the expected product (2-oxo-6-hydroxyhepta-3,5dienedioic acid) that allowed keto-enol tautomerizations involving both hydroxy/oxo groups in the 2-and 6-positions. In contrast, with 3-aminosalicylate there was only a decrease in the absorbance at wavelengths Ͼ300 nm, and a new absorbance maximum was formed at max ϭ 270 nm (Fig. 3). These spectral changes were presumably due to the ability of the ring fission product of 3-aminosalicylate to undergo an intramolecular cyclization and rearomatization to a pyridinecarboxylic acid shown previously for the reaction products formed by ring fission dioxygenases from 3-hydroxyanthranilate and 2-aminophenol (29,30). Thus, the results suggested that all substituted salicylates were cleaved by the "salicylate dioxygenase" by the same 1,2-dioxygenolytic mechanism that had been proven previously for the cleavage of salicylate by the same enzyme.
The results confirmed that from all available 5-substituted salicylates those that carried substituents with electron-donating properties (such as the OH or NH 2 groups) were by far the best substrates. Most surprisingly, 5-substituted salicylates with a weak (such as the methyl group) or very strong (fluorine) elec- tron-withdrawing effect were also converted with a higher reaction rate than found with the unsubstituted salicylate as a substrate. In contrast to 5-fluorosalicylate, 5-chloro, 5-bromo-, and 5-iodosalicylate were converted with lower reaction rates than salicylate. This might indicate some steric hindrance with these substrates. A similar effect was also found with the 3,5-disubstituted salicylates tested because the enzyme oxidized 3,5-dichlorosalicylate but not 3,5-dibromo-and 3,5-diiodosalicylate.
A comparison of the reaction rates observed with those salicylates for which the respective 3-, 4-, and 5-substituted isomers were available (amino-, hydroxy-, methyl-, bromo-, and chlorosalicylates) suggested that for the salicylates with an activating effect (amino-and hydroxysalicylates), only a substituent in the 5-position resulted in a pronounced increase (Ͼ100 times) in the oxidation rates compared with the respective 3-or 4-substituted isomers. In contrast, in the case of the chloro-or bromo-substituted salicylates, the 5-substituted salicylates were converted with a lower reaction rate than the respective 3-or 4-substituted substrates.
In a further set of experiments it was attempted to determine the K m values of the enzyme for different salicylates. The comparability of these experiments was hampered by different  FIG. 3. Conversion of various substituted salicylates by cell extracts from E. coli JM109 (pJPH100exN). The reaction mixtures contained in 1 ml of 20 mM Tris-HCl (pH 8.0), 0.1 mM of the respective salicylate and 1-140 g of protein. The spectra were recorded every minute against a reference cuvette, which contained the same amount of protein in Tris-HCl buffer but in which the respective substrates were omitted. degrees of substrate inhibition effects observed with the different substrates but basically confirmed the results obtained with fixed substrate concentrations (Table IV). These experiments demonstrated that the enzyme showed significantly higher K m values with 3-and 5-aminosalicylate compared with other substituted salicylates.
Analysis of the Enzymatic Reactions by LC-MS/MS-The turnover of several substituted salicylates was also analyzed by LC-MS techniques in order to confirm further the proposed reaction mechanism. In these experiments the reaction mixtures were directly injected into the liquid chromatography system (without terminating the reactions by acidification) and were also analyzed in an almost neutral solvent system (at pH 6.5). Based on LC-MS analyses, two groups of salicylates could be distinguished as follows: (a) compounds showing the expected dioxygenolytic cleavage only, and (b) other compounds that underwent consecutive reactions after dioxygenation.
Formation of the substituted 2-oxohepta-3,5-dienedioic acids was proven by the mass of the molecular anions (Table V) and by the product ion spectra that showed consecutive losses of CO 2 (from one of the carboxylate groups), of CO (from the carbonyl group in 2-position), and sometimes decarboxylation of the second carboxylate group. In the case of the halogenated salicylates, the product ion spectra of the dioxygenation products also showed elimination of HF, HCl, or HBr (Table V).

TABLE IV Kinetic constants of the salicylate-1,2-dioxygenase with different substrates
The reaction conditions were basically the same as described in Table II. The amount of substrates added was varied between 0.01 and 0.5 mol, and the reactions were analyzed as described below. The enzymatic activities with gentisate, salicylate, and 5-aminosalicylate were determined spectrophotometrically. The K m values for 4-amino-and 4-bromosalicylate could not be determined, because the observed activities were constant (v max ) under the experimental conditions applied.  The reaction mixtures contained in a total volume of 1 ml, 0.04 -0.1 mg of the purified dioxygenase, 0.1 mol (or 0.5 mol in the case of gentisate as substrate) of the respective substrates, and Tris-HCl buffer (20 mM, pH 8). Several aliquots (50 l) of the reaction mixtures were taken within 3 min, and the reactions were stopped by the addition of 5 l of 1 M HCl. The precipitated protein was removed by centrifugation in an Eppendorf centrifuge (2 min, 14,000 ϫ g), and the supernatants were analyzed by HPLC. The relative activities are presented (in %) in comparison to the specific activity with salicylate (3 units/mg ϭ 100%) as substrate. a n.a., compound not available. tanedienedioic acids) that underwent lactone formation while eliminating HF, HCl, or HBr, as suggested previously (11). Correspondingly, the initial dioxygenation products could not be detected from 5-chloro-and 5-fluorosalicylate by means of LC-MS. Instead, the lactone was the first reaction product that was detected. This lactone was then hydrolyzed to form 4-hydroxy-2-oxo-3,5-heptanedienedioic acid (maleylpyruvate) (Scheme 2, top). The retention time, UV spectrum, and mass spectrometric data of the product generated from both 5-chloroand 5-fluorosalicylate were identical to those obtained from 5-hydroxysalicylate upon dioxygenation. The 4-methyl-2-oxo-3,5-heptadienedioic acid that was generated from 5-methylsalicylate was also unstable and yielded two addition products. One can be ascribed to the addition of water to one of the double bonds (m/z 201), which is consistent with the product ion spectrum in which the loss of water (Ϫ18 atomic mass units) was the most pronounced fragmentation. The major product appeared at m/z 205, which was 22 mass units above the initial dioxygenation products. The identity of this product is yet unknown, although its UV spectrum and MS/MS fragmentations are very similar to those observed for the lactone formed from the dioxygenation products of the 5-halo-salicylates (m/z 167; Table V).
Expression Cloning of the Presumed Gentisate 1,2-Dioxygenase from C. glutamicum-As indicated above, the highest degree of sequence identity (58%) was found between the salicylate dioxygenase activity from P. salicylatoxidans and a presumed gentisate 1,2-dioxygenase from C. glutamicum. It was therefore tested whether the open reading frame encoded by C. glutamicum indeed encoded a gentisate-1,2-dioxygenase and whether this enzyme also converted salicylate. Therefore, the open reading frame was amplified by PCR from the SCHEME 2. Consecutive reactions observed for the dioxygenation products of halogenated salicylates. Top, from 5-halo-salicylates; bottom, from 3-halo-salicylates.  genomic DNA, and the encoded protein was expressed by using the expression vector pJOE2702 (24,25). Cell-free extracts from the construct obtained converted gentisate with rather high specific activities (20.6 units mg of protein) but did not convert salicylate or 1-hydroxy-2-naphthoate, although a competitive inhibition of gentisate oxidation in the presence of salicylate was observed (K i ϭ 2.3 mM).

DISCUSSION
The salicylate dioxygenase activity from P. salicylatoxidans BN12 is rather unique among the currently known ring fission dioxygenases because the enzyme is able to cleave various substituted salicylates that carry only a single hydroxy group and that are not activated by additional electron-donating substituents for a ring fission reaction. A similar reaction has been described previously only for the ring fission of 1-hydroxy-2naphthoate, which is an intermediate in the degradation of phenanthrene by certain bacteria. This reaction is catalyzed by a 1-hydroxy-2-naphthoate dioxygenase that results in the formation of trans-2-carboxybenzalpyruvate (10,(12)(13)(14). Our previous biochemical characterization of the salicylate dioxygenase activity from P. salicylatoxidans BN12 demonstrated that the enzyme converted gentisate, 5-aminosalicylate, and 1-hydroxy-2-naphthoate with much higher catalytic activities compared with salicylate and suggested that the ring fission dioxygenase was also structurally similar to gentisate 1,2-dioxygenase or 1-hydroxy-2naphthoate dioxygenase. This was indicated by the size of the subunits, the structure of the holoenzyme, and the dependence of the enzyme from Fe 2ϩ ions. Nevertheless, it became evident that the ring fission dioxygenase from P. salicylatoxidans was clearly different from the presently known gentisate 1,2-dioxygenases or 1-hydroxy-2-naphthoate dioxygenases because of its unique ability to oxidatively cleave salicylate and also the ability to cleave gentisate and 1-hydroxy-2-naphthoate with high catalytic efficiencies (15). In contrast, the 1-hydroxy-2-naphthoate dioxygenase from Nocardioides sp. KP7 did not oxidize gentisate or salicylate (12), and gentisate 1,2-dioxygenases do not oxidize salicylate (and presumably also not 1-hydroxy-2-naphthoate) (6).
The previously performed enzymatic analysis did not allow a clear distinction if the ring fission dioxygenase from P. salicylatoxidans was more closely related to gentisate 1,2-dioxygenases or to 1-hydroxy-2-naphthoate dioxygenases, although a closer relationship of the enzyme from P. salicylatoxidans with gentisate 1,2-dioxygenases was suggested by the observation that the enzyme was composed of four identical subunits that have also been observed for several bacterial gentisate 1,2dioxygenases (5,27,(31)(32)(33)(34)(35). In contrast, it was suggested that the 1-hydroxy-2-naphthoate dioxygenase from Nocardioides sp. KP7 was composed of six subunits (12).
The deduced amino acid sequence of the enzyme obtained in the present study confirmed the results of our previous enzymatic studies because it was found that the enzyme was distantly related to gentisate 1,2-dioxygenases and also 1-hydroxy-2-naphthoate dioxygenases. Thus 31% sequence identity was found between the "salicylate-1,2-dioxygenase" and the gentisate 1,2-dioxygenase from P. alcaligenes NCIMB 9867 (27) and 28% sequence identity for the 1H2NC dioxygenase from Nocardioides sp. KP7 (12). The sequence comparisons between these distantly related ring fission dioxygenases with the ability to cleave aromatic compounds between adjacent carboxyl and hydroxyl groups (gentisate 1,2-dioxygenases, 1-hydroxy- 2-naphthoate dioxygenase, and "salicylate-1,2-dioxygenase") demonstrated the presence of some highly conserved amino acid residues. Thus, it became evident that two histidine pairs (corresponding to His-128/His-130 and His-170/His-172 in the ring fission dioxygenase from P. salicylatoxidans) were present in all three enzymes. The conservation of these histidine pairs had been identified previously in a sequence comparison performed with different gentisate 1,2-dioxygenases, and it was shown by site-directed mutagenesis that the replacement of each of these four histidine residues resulted in the inactivation of the gentisate 1,2-dioxygenase from P. alcaligenes NCIMB 9867 (36). The conservation of these residues also in the more distantly related enzymes gave further evidence for their importance in catalysis. For the gentisate-1,2-dioxygenase from P. alcaligenes NCIMB 9867, it had been suggested that the two pairs of histidines were involved in the binding of the catalytic active ferrous iron (36). It was shown previously for the extradiol dioxygenases, which form the second group of ring fission dioxygenases containing ferrous iron in their catalytic center, that the Fe(II) ions in the catalytic center are bound by a "2-His-1-carboxylate facial triad" and that these three ligands anchor the iron in the active side (37). Thus it may be possible that two of the four conserved histidine residues are involved in the binding of the ferrous iron. In the sequence alignment of the enzyme from P. salicylatoxidans, the 1-hydroxy-2-naphthoate dioxygenase, and the gentisate-1,2-dioxygenases (Fig. 4), there also appears one glutamate residue to be highly conserved (corresponding to Glu-200 in the salicylate dioxygenase from P. salicylatoxidans), and this residue may be a candidate for the glutamate residue involved in a possible 2-His-1-carboxylate triad.
It was suggested previously that the oxidation of gentisate by gentisate-1,2-dioxygenases (and of homogentisate by the isofunctional homogentisate 1,2-dioxygenases) requires a direct bidentate ligation of the ferrous iron in the catalytic center by the carboxyl and 2-hydroxyl groups of the (homo)gentisate substrate. It was also suggested that the binding of the substrate is followed by a deprotonation of the 5-hydroxyl group of the substrate that presumably involves a histidine residue of the ring fission dioxygenase(s) (6,38). Although this function may be fulfilled during the oxidation of gentisate by one of the two other highly conserved histidine residues that are not involved in the speculative iron binding described above, this mechanism cannot be involved in the oxidation of (substituted) salicylates catalyzed by the ring fission dioxygenase from P. salicylatoxidans, because these substrates are missing a 5-hydroxyl group in their substrates. This suggests that the reaction mechanism for gentisate-1,2-dioxygenases, previously suggested by Harpel and Lipscomb (6), which required the isomerization of a 5-hydroxy (or 5-amino) group to the keto (or imino) resonance form during the catalytic cycle (Scheme 3), will probably need some revision. The observation that in C. glutamicum a gentisate dioxygenase is present, which shows a high degree of sequence identity (58%) with the salicylate dioxygenase activity but which is not able to convert salicylate, may allow the identification of those specific amino acid substitutions that allow the enzyme from P. salicylatoxidans to attack salicylate.
The tests with a wide range of substituted salicylates confirmed our previous observation that gentisate and 5-aminosalicylate were converted with extraordinary high V max values compared with the other substrates, because all other salicylates available were oxidized only with less than 10% of the value found with gentisate. On the other hand, it became evident that 3-and 5-aminosalicylates were only bound with a very low affinity to the enzyme and that the K m values with these substrates were at least 10 times higher than those found with the other substrates. A much higher K m value for 5-aminosalicylate compared with gentisate as substrate has also been observed previously for the gentisate 1,2-dioxygenases from "Pseudomonas" testosteroni and "Pseudomonas" acidovorans as described by Harpel and Lipscomb (6). This may suggest some kind of "repulsion" of an amino group in the 5-position of the respective substrate from the relevant part of the enzyme involved in the binding of the hydroxyl group of the natural substrate gentisate.
A comparison of the V max values observed with the halogenated salicylates carrying the respective substituents in the 5-position demonstrated a clear preference for 5-fluorosalicylate compared with 5-chloro-and 5-bromosalicylate. Furthermore, it was found that the V max value for 5-fluorosalicylate was higher than that observed with salicylate. This suggested that at least for salicylates carrying a substituent in the 5-position, the electronic effect of the highly electronegative fluorine atom did not significantly decrease the reaction rates and that the increasing size of the halogen atoms in the series fluorine, chlorine, and bromine were mainly responsible for the decreasing reaction rates if the substituent was changed from fluorine, via chlorine, to the bromine.
Many of the highly oxidized and highly unsaturated products generated from substituted salicylates by the 1,2-dioxygenase were chemically unstable and underwent elimination and addition reactions. These reactions were governed by the type and the position of the substituents. Thus, the LC/MS analysis clearly demonstrated that the products formed by the 1,2-dioxygenation of 3-halo-salicylates underwent decarboxylation. On the contrary, the conversion of 5-halo-salicylates resulted in a dehalogenation via lactone formation. A similar dehalogenation process has been suggested previously for the degradation of 5-chlorosalicylate by a "Bacillus" that was isolated from the Mississippi River (11). Thus, the dioxygenation of 5-substituted halogenated salicylates ultimately leads to a dehalogenation reaction and may convert xenobiotic substrates into intermediates also formed during the degradation of natural compounds. SCHEME 3. Proposed mechanism of gentisate 1,2-dioxygenase action (6).