Bacterial 2,4-Dioxygenases: New Members of the α/β Hydrolase-Fold Superfamily of Enzymes Functionally Related to Serine Hydrolases

Abstract

1H-3-hydroxy-4-oxoquinoline 2,4-dioxygenase (Qdo) from Pseudomonas putida 33/1 and 1H-3-hydroxy-4-oxoquinaldine 2,4-dioxygenase (Hod) from Arthrobacter ilicis Rü61a catalyze an N-heterocyclic-ring cleavage reaction, generating N-formylanthranilate and N-acetylanthranilate, respectively, and carbon monoxide. Amino acid sequence comparisons between Qdo, Hod, and a number of proteins belonging to the α/β hydrolase-fold superfamily of enzymes and analysis of the similarity between the predicted secondary structures of the 2,4-dioxygenases and the known secondary structure of haloalkane dehalogenase from Xanthobacter autotrophicus GJ10 strongly suggested that Qdo and Hod are structurally related to the α/β hydrolase-fold enzymes. The residues S95 and H244 of Qdo were found to be arranged like the catalytic nucleophilic residue and the catalytic histidine, respectively, of the α/β hydrolase-fold enzymes. Investigation of the potential functional significance of these and other residues of Qdo through site-directed mutagenesis supported the hypothesis that Qdo is structurally as well as functionally related to serine hydrolases, with S95 being a possible catalytic nucleophile and H244 being a possible catalytic base. A hypothetical reaction mechanism for Qdo-catalyzed 2,4-dioxygenolysis, involving formation of an ester bond between the catalytic serine residue and the carbonyl carbon of the substrate and subsequent dioxygenolysis of the covalently bound anionic intermediate, is discussed.

A number of bacterial strains which utilize quinoline or a quinoline derivative as the sole source of carbon and energy have been isolated and characterized (14). Degradation of 1H-4-oxoquinoline by Pseudomonas putida 33/1, for instance, is initiated by hydroxylation at C-3, followed by 2,4-dioxygenolytic cleavage of 1H-3-hydroxy-4-oxoquinoline to carbon monoxide and N-formylanthranilate (5, 8). Correspondingly, quinaldine utilization by Arthrobacter ilicis Rü61a (formerly Arthrobacter sp. strain Rü61a) proceeds via 1H-4-oxoquinaldine and 1H-3-hydroxy-4-oxoquinaldine to N-acetylanthranilate and carbon monoxide (5, 24). In both strains, the N-acylanthranilate is degraded via anthranilate and catechol to intermediates of central metabolic pathways. As proven by ¹⁸O₂/¹⁶O₂ incorporation studies, purified 1H-3-hydroxy-4-oxoquinoline 2,4-dioxygenase (Qdo) from P. putida 33/1 and purified 1H-3-hydroxy-4-oxoquinaldine 2,4-dioxygenase (Hod) from A. ilicis Rü61a catalyze the insertion of a single molecule of oxygen at C-2 and C-4 of the respective substrate with concomitant release of carbon monoxide (6). The mechanism of Qdo- or Hod-catalyzed 2,4-dioxygenolytic ring cleavage is poorly understood. Based on a chemical model reaction, it has tentatively been proposed that the reaction proceeds via base catalysis (6). After removal of a proton from the substrate by a catalytic base, molecular oxygen might attack the C-2 carbanion of the substrate, forming a peroxy anion which in a nucleophilic attack could react with the carbonyl carbon (C-4) of the substrate, forming a five-membered cyclic peroxide intermediate that decomposes to carbon monoxide and the N-acylanthranilate (6).

Biochemical and spectroscopic studies of purified Qdo and Hod have shown that these enzymes do not contain a metal center or organic cofactor (6). This is remarkable, since transition metal ions, such as the mononuclear iron centers present in the well-known nonheme iron dioxygenases catalyzing the cleavage of dihydroxy-substituted aromatic compounds, or organic cofactors, such as flavins, are generally thought to be required for the activation of molecular oxygen and/or the organic substrate (17, 34, 53). Based on the unique feature of performing a dioxygenolytic reaction without the requirement of cofactors or metal ions, the bacterial 2,4-dioxygenases are presumed to belong to a novel type of dioxygenases.

Cloning and sequencing of both qdo and hod confirmed that the gene products apparently do not have any similarity to known oxygenases (35). Instead, Qdo and Hod showed low but significant sequence similarity to P. putida atropinesterase, a serine hydrolase (20).

Serine hydrolases belong to the α/β hydrolase-fold family of enzymes, which comprises a large group of both procaryotic and eucaryotic proteins that share a three-dimensional core structure, even though there is only modest similarity between primary sequences. However, as outlined by Ollis et al. (42), these widely different sequences can code for remarkably similar structures, since structural similarity is preserved much longer than sequence similarity. Despite the marked variance in primary structure, it has been found that all of these enzymes contain a catalytic triad, which is conserved in the primary sequence in the invariant order of nucleophilic residue-acidic residue-histidine (42). These amino acid residues, which are widely separated from each other in the primary structure, are found in similar topological locations in the folded proteins. The nucleophile, for example, which in most cases is a serine residue, is located in a sharp bend, the “nucleophile elbow.” This nucleophile elbow is represented by a conserved motif with the proposed consensus sequence Sm-X-Nu-X-Sm-Sm, where Sm indicates a small amino acid residue, X is any amino acid, and Nu is the nucleophile (42). Although the enzymes of the α/β hydrolase-fold family apparently show structural conservation of a catalytic subsite framework, their catalytic specificities are radically different (42). The enzyme family comprises all known cofactor-free haloperoxidases (19, 22a, 44) and diverse hydrolases catalyzing a wide variety of reactions (ester, amide, epoxide, C-halogen bond, and even C-C bond hydrolysis) (2, 3, 12, 42, 43, 49, 56, 57, 59). However, the enzymes of the α/β hydrolase-fold family are not only structurally but also functionally and mechanistically related (42). Despite the differences in catalytic specificities, they catalyze a hydrolytic reaction (however, the cofactor-free haloperoxidases use H₂O₂ instead of H₂O as the cosubstrate [22a]).

Here we propose that the 2,4-dioxygenases Qdo and Hod, which belong to the class of oxidoreductases (E.C. 1.), are members of the α/β hydrolase-fold superfamily of enzymes, having structural and functional features in common with serine hydrolases (E.C. 3.). As a first step toward the analysis of the structure-function relationship in Qdo and Hod, we have compared the amino acid sequences of Qdo and Hod with known members of the α/β hydrolase-fold family. Based on these comparative sequence analyses, which suggested the possibility of a catalytic triad, we performed in vitro site-directed mutagenesis of Qdo to assess the putative involvement of distinct serine, aspartate, and histidine residues of Qdo in catalysis.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Escherichia coli M15(pREP4) (Qiagen, Hilden, Germany) was used as a host for (recombinant) plasmid pQE30 (Qiagen) and for expression of qdo. Epicurian Coli XL1-Blue supercompetent cells (Stratagene, Heidelberg, Germany) were used as the host for pQE30 derivatives carrying mutations in the qdo gene constructed by site-directed mutagenesis. E. coli clones were grown in Luria-Bertani (LB) medium (51) at 37°C, either in the presence of kanamycin (50 μg/ml) and ampicillin (100 μg/ml) (for cells harboring pQE30 derivatives as well as pREP4) or in the presence of ampicillin (100 μg/ml, for XL-1 Blue cells).

Recombinant DNA techniques.

The qdo gene (EMBL accession no Y14779) has previously been amplified by PCR with primers that created additional sequences for BamHI recognition and inserted into the BamHI restriction site of pQE30, resulting in pQE30-qdo (35). Transformation of E. coli M15(pREP4) with pQE30 derivatives was performed by electroporation. Recombinant plasmid DNA was prepared with the QIAprep Spin Miniprep kit (Qiagen) as specified by the supplier. Agarose gel electrophoresis was carried out by standard techniques (51).

Oligonucleotide site-directed mutagenesis.

Mutagenesis was performed with the QuikChange site-directed mutagenesis kit (Stratagene) by the procedure recommended by the supplier. Plasmid pQE30-qdo was used as the template for Pfu polymerase-mediated amplification of qdo variants with the following oligonucleotides as primers for mutagenesis of the codons: S93A, 5′-TTCCAAATGGTCGCCACCTCCCACGGC-3′; S95A, 5′-ATGGTCTCCACCGCCCACGGCTGTTGG-3′; S95C, 5′-ATGGTCTCCACCTGCCACGGCTGTTGG-3′; D120A, 5′-ACCATCGTCATCGCCTGGCTGCTGCAACCG-3′; S213A, 5′-TGCCACATCTACGCGCAACCCCTTTCC-3′; D219A, 5′-CCCCTTTCCCAGGCCTACCGCCAGCTAC-3′; H244A, 5′CCGGGACGGACCGCCTTCCCTTCCCTG-3′ (altered codons are underlined). Mutations were confirmed by sequence analysis of the DNA fragments encompassing the mutations. DNA sequencing was performed by the dideoxy chain termination method (52). The primers used for sequencing were 5′-end modified with IRD-800 (MWG-Biotech, Ebersberg, Germany). Primers SeqS95 (5′-GCGAAACAGACCGATAGC-3′) (for sequencing of the DNA regions encompassing the tripletts encoding A93, A95, and A120), SeqH244 (5′-CCTGAAATCTGCCACATC-3′), and SeqD219 (5′-CGCGAGATCGAAGCGAAC-3′) were used for sequencing. The sequencing reaction products were analyzed with a LI-COR 4000 automatic sequencer (MWG-Biotech).

Analyses of amino acid sequences.

Database searches and binary sequence comparisons were run with the program BLAST (1). Multiple amino acid sequence alignments were performed with the CLUSTAL W program (55; HUSAR 4.0 program package; European Molecular Biology Laboratory, Heidelberg, Germany). For comparison of the regions of the putative catalytic amino acid residues, the sequences were manually realigned. Secondary-structure predictions were carried out with the program PredictProtein (50).

Enzyme assay.

The activity of Qdo was determined spectrophotometrically by measuring substrate consumption at 337 nm as described previously (6). One unit was defined as the amount of enzyme that converted 1 μmol of substrate per min at 22°C. K_m and V_max were calculated from Lineweaver-Burk plots (32).

Protein determinations.

Protein concentrations were determined by the method of Lowry et al. (33) with bovine serum albumin as the standard.

Purification of recombinant Qdo.

E. coli M15(pREP4) clones harboring recombinant pQE30-qdo or pQE30-qdo variants were grown in 100 ml of Luria-Bertani broth at 37°C to an optical density at 600 nm of 0.6. A 700-ml volume of Luria-Bertani broth was inoculated with this cell suspension, and synthesis of hexahistidine-tagged Qdo protein (His₆Qdo) was induced by adding 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG). After incubation for 16 h at 20°C, the cells were harvested by centrifugation at 4000 × g for 20 min at 4°C, resuspended in 50 mM sodium phosphate buffer containing 0.5 M NaCl (pH 7.8), and broken by ultrasonic treatment (Sonifier model 250; Branson Ultrasonics, Danbury, Conn.). Cell debris was removed by centrifugation at 48,000 × g for 40 min at 4°C. In a polypropylene tube slowly rotated overnight at 4°C, the crude cell extract was mixed with 4 ml of a 50% (vol/vol) slurry of Ni²⁺-nitrilotriacetate resin beads (Ni-NTA Superflow, Qiagen) that had been equilibrated in sodium phosphate buffer containing NaCl (see above). Subsequently, the resin was packed into a Bio-Scale MT2 column (Bio-Rad Laboratories, Munich, Germany). A BioLogic HR liquid chromatography system (Bio-Rad) was used for all chromatographic steps. After two washing steps, first with sodium phosphate buffer containing NaCl and supplemented with 5 mM imidazole and then with sodium phosphate buffer containing NaCl and supplemented with 20 mM imidazole, His₆Qdo was eluted with a linear gradient (25 ml) from 100 to 300 mM imidazole in the same NaCl-containing buffer at a flow rate of 0.5 ml/min. Fractions containing recombinant enzyme were identified by measurement of Qdo activity and/or by polyacrylamide gel electrophoresis. The eluted His₆Qdo protein was washed in an ultrafiltration device (membrane cutoff 10 kDa) with 50 mM Tris-HCl containing 10 mM EDTA (pH 7.8). The purified His₆Qdo was used immediately for determination of its catalytic properties or was stored at −80°C. Storage was not accompanied by any losses in activity (data not shown).

SDS-PAGE.

Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE) was performed with 10% separating gels by the method of Laemmli (31). Proteins were stained with Coomassie blue G-250 (0.2% [wt/vol] in water-methanol-acetic acid [60/30/10, by vol]).

RESULTS AND DISCUSSION

Comparison of the amino acid sequences of Qdo, Hod, and α/β hydrolase-fold enzymes, and predicted secondary structures of Qdo and Hod.

The amino acid sequences of Qdo and Hod showed 37% identity (Fig. 1). The highest similarity between Qdo and any sequence deposited in databases was found with atropinesterase (Tpes) from P. putida (20). Qdo and Tpes had 18% identical amino acids (Fig. 1). Further hydrolases and cofactor-free haloperoxidases that had low but significant sequence similarity to Qdo and Hod are listed in Fig. 1 to 3. The three-dimensional structures of haloalkane halidohydrolase (DhlA) from Xanthobacter autotrophicus GJ10 (15, 56, 57), bromoperoxidase A2 (BpoA2) from Streptomyces aureofaciens ATCC 10762 (19), and epoxide hydrolase (EchA) from Agrobacterium radiobacter AD1 (39) have been elucidated by X-ray diffraction. These enzymes belong to the structural family of α/β hydrolase-fold enzymes, and the other enzymes aligned with Qdo and Hod have also been proposed to show the general topology of the α/β hydrolase fold (2, 3, 12, 19, 29). Low sequence similarities between the members of the α/β hydrolase-fold family are common (42). The NH₂-terminal end and a large central segment of the proteins compared in Fig. 1 indeed exhibit very low (if any) similarity. The central segment of the α/β hydrolase-fold enzymes has been suggested to be involved in substrate specificity (12, 48). The central region of Qdo, which was predicted to be completely α-helical, may correspond to the α-helical “cap” domain of DhlA (helices H₄ to H₈), which would be in agreement with its possible role in contributing to substrate specificity.

FIG. 1.

Multiple amino acid sequence alignment of Qdo, Hod, and some hydrolases and cofactor-free haloperoxidases performed with the program CLUSTAL W (55), and prediction of secondary-structure elements for Qdo and DhlA. Abbreviations: Qdo, P. putida 33/1 1H-3-hydroxy-4-oxoquinoline 2,4-dioxygenase (35; EMBL accession no. Y14779); Hod, A. ilicis Rü61a 1H-3-hydroxy-4-oxoquinaldine 2,4-dioxygenase (35; EMBL accession no. Y14778); Tpes, P. putida atropinesterase (20); CpoL, Streptomyces lividans TK64 chloroperoxidase (4); BpoA2, Streptomyces aureofaciens ATCC 10762 bromoperoxidase (45); XylF, P. putida 2-hydroxymuconic semialdehyde hydrolase (23; the sequence has been corrected in the N-terminal region according to reference 12; see also reference 22); DhlA, Xanthobacter autotrophicus GJ10 haloalkane dehalogenase (26). The numbering does not refer to any of the sequences. Presumed triad residues are marked with a star and printed in boldface. Conserved amino acid motifs surrounding the catalytic nucleophile and the catalytic histidine residue of the α/β hydrolase-fold enzymes are enclosed in a box. S93, D120, and S213 of Qdo are marked by an arrow. The first line at the bottom of the sequences shows the secondary-structure elements of Qdo as calculated by the program PredictProtein (50). The second and third lines indicate the secondary-structure elements of DhlA as predicted (line 2) and as deduced from the X-ray analyses of DhlA (56) (line 3). Hatched and open boxes indicate β-sheets and α-helices, respectively. The α-helices of DhlA are numbered according to the numbering scheme of Verschueren et al. (56). Note that the Qdo sequence (35) has been corrected in the C-terminal region (FLQA instead of FLHGLSTCNHELKR) and at position 98 (C instead of V). Qdo thus consists of 264 amino acids, its calculated molecular mass is 30,347 Da, and its calculated isoelectric point is pH 5.43.

FIG. 3.

Multiple alignment of 2,4-dioxygenases, various hydrolases, and some cofactor-free haloperoxidases in the region of the potential catalytic aspartate and histidine residues of the enzymes. The alignments are based on database searches and binary alignments run with the program BLAST (1) and have been manually realigned. Abbreviations are given in the legends to Fig. 1 and 2. The structures of BpoA2 (19), DhlA (15, 56, 57), and EchA (39) (abbreviations are underlined in the figure) have been solved by X-ray diffraction. The putative triad histidine residues are underlined and marked by a star. The putative acidic residues of the catalytic triad are underlined. S213 and D219 of Qdo are marked by arrows. Amino acid residues conserved in at least half of the aligned sequences are in boldface.

Despite the low sequence similarity, which limits the accuracy of the secondary-structure prediction, most of the β-strands of Qdo were predicted to be at roughly the same position in the alignment as the β-strands of DhlA (Fig. 1). The same holds for the predicted α-helices of Qdo corresponding to helices H₁, H₂, H₃, H₄, H₅, H₁₀, and H₁₁ of DhlA (Fig. 1). The pattern of the secondary-structure elements predicted for Hod (not shown) resembled the pattern calculated for Qdo. Taken together, the comparative sequence analyses as well as the secondary-structure predictions suggest that Qdo and Hod are structurally related to the α/β hydrolase-fold superfamily of enzymes.

Conserved amino acid sequence motifs of Qdo, Hod, and α/β hydrolase-fold enzymes.

The consensus motif Sm-X-Nu-X-Sm-Sm encompassing the catalytic nucleophilic residue of the α/β hydrolase-fold enzymes is well conserved in all aligned sequences (Fig. 1 and 2). In DhlA (15, 46, 56, 57), EchA (39, 49), and rat soluble epoxide hydrolase (rsEH) (3), for instance, the nucleophilic residues are D124, D107, and D333, respectively. In DhlA, the residues Q123, D124, and W125 form an extremely sharp bend, the nucleophile elbow (56). At the position of the putative nucleophile, a serine residue is conserved in many hydrolases and in all cofactor-free haloperoxidases. This serine residue is catalytically essential in BpoA2 (44), 2-hydroxymuconic semialdehyde hydrolase (XylF) (12), and human pancreatic lipase (hPL) (59). Thus, S95 of Qdo and S101 of Hod are likely candidates for a putative catalytic residue. It is interesting that the first small amino acid within the consensus sequence, which usually is a glycine residue, is another serine in Qdo (S93) and Hod (S99). With S93, T94, and S95, Qdo even has three adjacent amino acids with a hydroxyl group (Fig. 2). The side chains of residues Nu−2 and Nu+2 are thought to be close to each other (42). If the two serine residues in the conserved motif S-X-S-H-G-Sm of Qdo and Hod indeed are in close proximity, this might well be of functional significance.

FIG. 2.

Multiple alignment of 2,4-dioxygenases, various hydrolases, and some cofactor-free haloperoxidases in the region of the potential catalytic nucleophile of the enzymes. The alignments are based on database searches and binary alignments run with the program BLAST (1) and have been manually realigned. Abbreviations (see also the legend to Fig. 1): BphD, P. putida KF715 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoate hydrolase (18) (the amino acid sequences of BphD from strain KF715 and from Pseudomonas sp. strain LB400 [22] show 95% identity); TodF, P. putida F1 2-hydroxy-6-oxohepta-2,4-dienoate hydrolase (37); DmpD, Pseudomonas sp. CF600 2-hydroxymuconic semialdehyde hydrolase (41); Lip3, Moraxella sp. strain TA144 lipase 3 (13); CpoP, Pseudomonas pyrrocinia chloroperoxidase (62); LinB, Sphingomonas (formerly Pseudomonas) paucimobilis 1,3,4,6-tetrachloro-1,4-cyclohexadiene halidohydrolase (38); DhaA, Rhodococcus rhodochrous NCIMB 13064 1-chloroalkane dehalogenase (30); DehH1, Moraxella sp. strain B haloacetate dehalogenase (27); EchA, Agrobacterium radiobacter AD1 epoxide hydrolase (49); rsEH, soluble epoxide hydrolase from rat liver (28); hPL, human pancreatic lipase (59). The structures of BpoA2 (19), DhlA (15, 56, 57), EchA (39), and hPL (59) (the abbreviations are underlined in the figure) have been solved by X-ray diffraction. The putative catalytic nucleophilic residues are underlined and marked by a star. S93 and D120 of Qdo and the catalytic triad acid D176 of hPL are marked by arrows. Amino acid residues conserved in at least half of the aligned sequences are in boldface.

The amino acid residue immediately following the putative nucleophile of the α/β hydrolase-fold enzymes has been proposed to be involved in substrate binding (3, 12, 56). Whereas most hydrolases have an aromatic (F or W) or methionine residue in this position, the 2,4-dioxygenases Qdo and Hod are the only enzymes where a histidine residue follows the presumptive nucleophile (Fig. 1 and 2). In DhlA, where the active-site nucleophile is D124, the side chain of the adjacent residue W125 points into the internal active site cavity and is proposed to be involved in stabilization of the negatively charged five-coordinate carbon in the transition state and the released halide ion (56, 57).

Apart from the Sm-X-Nu-X-Sm-Sm motif, which occurs in all aligned sequences shown in Fig. 2, a number of further motifs or residues are conserved in subgroups of the aligned enzymes (Fig. 2 and 3). An example is a P-E-R-V motif, which is conserved in XylF, DmpD, DhaA, and rsEH and is similar in most other enzymes, situated 13 to 15 amino acid residues from the catalytic nucleophile. An amino acid segment conserved in a number of hydrolases and cofactor-free haloperoxidases is the sequence D/E-A-L-G/D(/E)-L(/I), which is located 8 to 9 amino acid residues NH₂-terminal to the nucleophilic residue (Fig. 2). This region, however, is rather different in Qdo, Hod, and Tpes (P. putida atropinesterase). The motif R-V-I-A(/X)-P(/X)-D-X-X-G-X-G-X-S-(X)_2/4-P, which is found 30 to 35 amino acids NH₂-terminal to the nucleophilic residue in many α/β hydrolase-fold enzymes, is also conserved in Hod (except that the conserved aspartate is replaced by asparagine), whereas in Qdo, this motif is hardly recognizable (Fig. 2).

The presumptive histidine residue of the catalytic triad is completely conserved in all hydrolases and cofactor-free haloperoxidases. H244 of Qdo and H251 of Hod clearly align with the triad histidine of the α/β hydrolase-fold enzymes (Fig. 3). In BpoA2 (44), DhlA (47, 56, 57), XylF (12), EchA (39, 49), and rsEH (3), the equivalent histidine residue indeed has been proven to be the general base residue which is functional in the catalytic triad. The sequence encompassing the conserved histidine (underlined) is G-(X)-H-Ar in most hydrolases (where Ar is an aromatic amino acid residue). In Qdo and Hod, an aromatic residue likewise follows the histidine, but the conserved glycine residue preceding the putative catalytic histidine is replaced by threonine. Although the primary sequences of the cofactor-free haloperoxidases diverge in this region, their presumptive catalytic histidine aligns with the histidine of all hydrolases (Fig. 3).

Whereas the nucleophilic residue and the histidine residue of the putative catalytic triads are located within distinct sequence motifs, the amino acid sequences around the catalytic acidic residue appear to be less highly conserved among the α/β-fold hydrolases. However, as shown in Fig. 3, some (but not all) hydrolases show the sequence H/(X)-G-X-X-D-X-X-X-(X)-P 23 to 24 residues NH₂-terminal to the catalytic histidine (the catalytic aspartate is underlined). A number of hydrolases and cofactor-free haloperoxidases possess another acidic residue immediately preceding the catalytic aspartate residue. In Tpes, an asparagine residue aligns with the catalytic acidic residue of the cofactor-free haloperoxidases and many hydrolases, but its functional significance is not known. In Qdo and Hod, the above-mentioned conserved sequence is not evident, and comparison of this region of Qdo and Hod to the α/β hydrolase-fold proteins led to divergent alignments when different algorithms were used (Fig. 1 and 3). All BLAST searches indicated that in Qdo and Hod, a glutamine residue might be in the equivalent position of the catalytic acidic residue of the hydrolases and cofactor-free haloperoxidases (Fig. 3). D219 of Qdo and E226 of Hod are located 4 amino acid residues from these glutamine residues Q214 (Qdo) and Q221 (Hod). The primary sequences of BphD, TodF, Lip3, DhlA, rsEH, and all cofactor-free haloperoxidases also show an acidic residue, which is located 5 or 6 residues from the catalytic aspartate, but this residue is not involved in the catalytic mechanism. In DhlA, this nonfunctional aspartate (D266) is positioned within the α-helix H₁₀, near to its NH₂-terminal end (56) (Fig. 1). The prediction of secondary-structure elements of Qdo likewise assigned D219 near to the NH₂-terminal end of an α-helix (Fig. 1). In contrast, the region around Q214 of Qdo was proposed to form a loop. Due to the secondary-structure predictions and to the poor sequence similarities between Qdo, Hod, and the aligned members of the α/β hydrolase-fold superfamily of enzymes in this region, the possibility of a catalytic significance of D219 in Qdo appeared ambiguous.

The conserved amino acid sequence motifs and presumptive catalytic serine and histidine residues of Qdo show the same sequence order and similar spacings as the respective motifs and triad residues of the α/β hydrolase-fold enzymes. Despite the poor amino acid sequence homology among the proteins belonging to the α/β hydrolase-fold superfamily of enzymes, their catalytic-triad residues are known to have similar topological and three-dimensional positions (42).

Mutagenesis of qdo and detection of the Qdo variants.

To investigate the potential functional role of the conserved S95 and H244 and to assess the hypothesis of a putative catalytic triad of the 2,4-dioxygenase Qdo, these and other residues were replaced by site-directed mutagenesis. Provided that the folding of the recombinant His₆Qdo proteins is correct, loss of activity should be a direct consequence of the replacement of amino acid residues involved in catalysis and/or substrate binding by nonfunctional amino acid side chains.

In the case of the presumptive catalytic nucleophile S95, the serine residue was replaced by the nonionizable amino acid alanine and by cysteine as another possible but stronger nucleophile. H244 of Qdo, which aligns with the catalytic histidine residues of the α/β hydrolase-fold enzymes, also was replaced by alanine. As described above, the amino acid sequence comparisons were difficult in the region of D219 of Qdo, and the secondary-structure predictions of Qdo indicated an α-helical structure around D219 rather than a loop. However, considering the spacing between the active-site residues, D219 could nevertheless be a possible candidate for a catalytic aspartate residue. Its functional significance was also assessed by replacing it by an alanine residue. Furthermore, His₆QdoS93A, His₆QdoD120A, and His₆QdoS213A mutants were constructed. The secondary-structure predictions for the 93A, 95A, 95C, 120A, 213A, 219A, and 244A Qdo proteins, as determined by the program PredictProtein (50), corresponded to those of wild-type Qdo. Thus, the replacements were not supposed to alter the secondary-structure elements.

Recombinant His₆Qdo as well as His₆-tagged QdoS95A, QdoS95C, QdoH244A, QdoD219A, and QdoD120A were purified to electrophoretic homogeneity (Fig. 4a) by metal chelate affinity chromatography on an Ni²⁺-nitrilotriacetate column. Recombinant His₆Qdo as well as the His₆-tagged S95A, S95C, H244A, D219A, and D120A variants eluted from the metal chelate affinity column at an imidazole concentration of 180 mM. In contrast, no protein peaks eluted at this imidazole concentration when crude extracts from the E. coli clones supposed to express His₆QdoS93A and His₆QdoS213A were separated by metal chelate affinity chromatography, indicating that these crude extracts did not contain soluble His₆-tagged Qdo protein. In SDS-PAGE, protein bands corresponding to a molecular mass of about 31 kDa (corresponding to His₆Qdo) were hardly visible when crude extracts from the E. coli expression clones supposed to express His₆QdoS93A and His₆QdoS213A were separated (Fig. 4b). The presumed absence of soluble Qdo protein in these recombinant E. coli expression clones could be caused either by precipitation of the Qdo variant to insoluble aggregates and/or proteolysis or by a lack of Qdo synthesis.

FIG. 4.

SDS-PAGE of recombinant His₆-tagged Qdo proteins. (a) His₆Qdo variants purified by metal chelate affinity chromatography. Lanes: 2, recombinant Qdo; 3, QdoD120A; 4, QdoD219A; 5, QdoH244A; 7, QdoS95A; 8, QdoS95C. (b) Crude extracts of E. coli clones expressing recombinant His₆-tagged Qdo (lane 2) and of clones supposed to express His₆-tagged QdoS93A (lane 3) and QdoS213A (lane 4). The molecular masses of the marker proteins (lanes 1, 6, and 9 in panel a and lanes 1 and 5 in panel b) are 97.2, 66.2, 55, 42.7, 40, and 31 kDa.

The specific activity of recombinant His₆Qdo in crude extracts of E. coli M15(pREP4)pQE30-qdo was 0.25 U mg of protein⁻¹. Crude extracts of the E. coli clones containing mutant qdo showed specific activities as follows: His₆QdoS95A, 0.003 U mg of protein⁻¹; His₆QdoS95C, below detection; His₆QdoH244A, below detection; His₆QdoD219A, 0.24 U mg of protein⁻¹; and His₆QdoD120A, 0.28 U mg of protein⁻¹. Crude extract of the clone supposed to produce His₆QdoS213A, which apparently did not contain soluble Qdo (Fig. 4b), likewise did not show detectable Qdo activity. In contrast, freshly prepared crude extract of the E. coli mutant synthesizing His₆QdoS93A exhibited a specific 2,4-dioxygenase activity of about 0.01 U mg of protein⁻¹, which rapidly vanished, presumably due to precipitation of the protein (Fig. 4b).

The catalytic properties of wild-type Qdo isolated from P. putida 33/1 (7), purified recombinant His₆Qdo, and purified mutant His₆Qdo proteins are summarized in Table 1. The apparent K_m values of wild-type and recombinant Qdo were similar, suggesting that the recombinant Qdo enzyme is not affected by the NH₂-terminal hexahistidine tag.

TABLE 1.

Kinetic parameters for purified wild-type, recombinant, and mutant Qdo enzymes

Enzyme	Km (μM)	kcat (s−1)	kcat/Km (s−1 μM−1)
Qdo (wild-type)	24a	NDb
Qdo (recombinant, His6 tagged)	20	2,214	110.70
QdoS95C (His6 tagged)	61	7	0.11
QdoS95A (His6 tagged)	95	164	1.73
QdoH244A (His6 tagged)	BDc	BD	BD
QdoD120A (His6 tagged)	35	1,298	37.09
QdoD219A (His6 tagged)	50	397	7.94

^aData from reference 7. 

^bND, not determined. 

^cBD, activity was below the limit of detection. 

Assessment of the potential nucleophilic serine residue of Qdo.

Interestingly, replacement of the putative active-site nucleophile S95 by alanine or cysteine did not result in complete loss of enzyme activity. The apparent K_m value of the S95A mutant was increased 4.8-fold compared to recombinant His₆Qdo, and its turnover number was decreased 13.5-fold. Purified His₆QdoS95C showed a relatively low K_m value, but its turnover number was decreased more than 300-fold compared with recombinant His₆Qdo (Table 1). However, if the critical active-site nucleophilic side chain has been removed in His₆QdoS95A and His₆QdoS95C, complete loss of activity would be expected. Possibly, S95 has some backup in Qdo. Thus, the assumption that the nearby S93, which was predicted to immediately follow a β-sheet structure, might to some extent take over the catalytic role of S95 was assessed by construction of an E. coli clone which should produce His₆QdoS93A. Unfortunately, it was not possible to isolate the His₆QdoS93A protein, as described above. On the other hand, it is even possible that S213 of Qdo, which presumably is located in the topological position of the acidic residue of a catalytic triad (Fig. 1) and which is conserved in Qdo and Hod (Fig. 3), is of functional significance. However, as outlined above, the His₆QdoS213A enzyme was not detected in the crude extract of the corresponding E. coli clone. Thus, none of the assumed nearby potential nucleophiles could be tested, because the relevant Qdo proteins were not expressed or folded.

The nucleophile of the hydrolases possessing a catalytic triad can be either a serine, an aspartate, or a cysteine residue. In hydrolases with an active-site cysteine residue, cysteine has been replaced by serine. Pathak et al. (43), for instance, show that about 10% of the activity was retained in a C123S mutant of dienelactone hydrolase whereas the C123A dienelactone hydrolase was inactive. In contrast, conversion of the active-site thiol group of C25 in papain to a hydroxyl group led to S25 papain that was devoid of enzymatic activity (11). Considering the serine hydrolases, Witkowski et al. (60, 61) engineered a S101C mutant of the rat mammary gland thioesterase II which retained up to 90% of catalytic activity. Whereas replacement of the seryl oxygen with the larger cysteinyl sulfur had only a minor effect on the catalytic efficiency of thioesterase II, attempts to engineer efficient thiol active sites from the serine active-site proteases subtilisin (40) and trypsin (21, 36, 63) as well as acetylcholinesterase (16) and XylF (12) produced enzymes with activities reduced by 2 to 6 orders of magnitude. In this study, the activity of His₆QdoS95C was likewise very low (Table 1). In dienelactone hydrolase, the active-site cysteine thiol is converted to the catalytically functional thiolate only in the presence of substrate (10). Additionally, a glutamate residue is required to form the cysteine-thiolate from cysteine thiol (10). Thus, the serine hydrolases unable to utilize cysteine in their active site may not possess a mechanism for activation (i.e., deprotonation) of the cysteine thiol, or steric problems may preclude the use of cysteine as an active-site nucleophile. Analysis of the crystal structure of the S195C trypsin mutant revealed that the sulfur atom was pointing away from the binding pocket, obstructing the oxyanion hole in trypsin (36). Thus, most serine hydrolases appear to show an absolute requirement for serine in their catalytic triad. Thioesterases utilize both types of serine codons, TCN (where N is a purine or pyrimidine base) and AGY (where Y is a pyrimidine base). The latter could be derived from the cysteine codons TGY by a single base change. In contrast, the 2,4-dioxygenases Qdo and Hod use TCN codons for the presumptive catalytic serine residue. Whereas the S101C thioesterase II mutant might support the hypothesis of Brenner (9) that considers the serine active-site hydrolases to be evolved from ancestral cysteine active-site enzymes, other studies indicate that during evolution of the hydrolases, a shift toward a catalytic mechanism of higher stringency with a strong bias for serine has occurred (12). Indeed, in response to Brenner’s hypothesis, Irwin (25) proposed that there are multiple origins of the serine codons of the hydrolase active site, with an accompanying switch from one type of codon to the other.

Catalytic histidine residue of Qdo.

The activity of the His₆QdoH244A protein, which was purified by metal chelate affinity chromatography (Fig. 4a), was below the level of detection, clearly indicating a functional role of this histidine residue in Qdo.

Assessment of possible catalytic aspartate residues of Qdo.

According to the alignment shown in Fig. 1, there was a remote possibility that D219 of Qdo was a candidate for the acidic residue of the catalytic triad. Its replacement by alanine resulted in His₆QdoD219A protein showing a ca. 2.5-fold increase in K_m and a 5.6-fold decrease in k_cat (Table 1). Although this is a marked decrease in activity, the substrate turnover is significant. Based on the relatively high turnover number and the secondary-structure predictions strongly suggesting an α-helix in this region (see above), it seems rather unlikely that D219 is the catalytically essential acidic residue of a putative active-site triad.

It is interesting that atropinesterase possesses an asparagine which aligns with the catalytic aspartate of the cofactor-free haloperoxidases and most hydrolases (Fig. 3). The proteolytic enzyme papain possesses a Cys-His-Asn triad, where the asparagine residue is assumed to be required for the formation of the Cys⁻/His⁺ ion pair. (However, papain constitutes a separate group of the hydrolases and forms a nucleophilic elbow quite different from the sharp bend of the α/β hydrolase-fold enzymes [42].)

The acidic residue corresponding to the triad acid of the classical α/β hydrolase-fold enzymes likewise does not seem to be conserved in the halidohydrolases LinB, DhaA, and DehH1 (Fig. 3). On the other hand, LinB, DhaA, and DehH1, as well as Qdo and Hod, possess an asparate or glutamate residue at the equivalent position of the triad acid D176 of hPL (Fig. 2) (29, 54). This aspartate residue aligns with D120 of Qdo. Although there is no significant sequence similarity between hPL and the hydrolases compared in Fig. 2, the topology of the first domain of hPL is known to be like that of the α/β hydrolase-fold enzymes (42, 54, 59). The catalytic-triad residues of hPL are organized differently from the classical α/β hydrolase-fold enzymes. Whereas the active-site serine and histidine residues in hPL are located after β-strands 5 and 8 and thus correspond to the topological organization of the α/β hydrolase-fold enzymes, the catalytic aspartate (D176) follows β-strand 6 instead of β-strand 7 (42, 54). In DhlA, N148 is the analogue of D176 of hPL (Fig. 2) and is located directly behind β-strand 6, where it forms a hydrogen bond with the nucleophilic D124 (56). Krooshof et al. (29) suggested that the acidic residue of the catalytic triad is present in LinB, DhaA, and DehH1 at a position analogous to D176 of hPL and N148 of DhlA (Fig. 2). Experimental evidence for this hypothesis was provided by replacing N148 of DhlA with aspartate or glutamate and replacing the catalytic D260 of DhlA by asparagine. In the DhlA-D260N+N148E and DhlA-D260N+N148D mutants, the acidic residue at position 148 indeed could to some extent take over the role of the triad acid, interacting with H289 of DhlA (29). Thus, we tentatively assumed that D120 of Qdo, which aligns with N148 of DhlA, the catalytic D176 of hPL, and the presumptive triad acids of LinB, DhaA, and DehH1, might be the triad acid. However, replacement of D120 of His₆Qdo by alanine resulted in a Qdo enzyme whose catalytic activity was of the same order of magnitude as the recombinant His₆Qdo enzyme (Table 1), clearly disproving the hypothesis that D120 is the acidic residue of a potential catalytic triad of Qdo.

Tentative hypothesis on the possible functional role of the catalytically essential amino acids of Qdo.

The hydrogen-bonded triad, consisting of a nucleophilic amino acid, histidine, and an acidic amino acid residue, is essential in the catalysis of known hydrolases and haloperoxidases belonging to the α/β hydrolase-fold family of enzymes. In the catalysis of a hydrolytic reaction, the side chain of the nucleophilic amino acid residue attacks the electropositive carbon atom of the substrate (e.g., a carbonyl carbon in an ester or amide hydrolysis, or, as in DhlA, the chlorine-substituted carbon atom) to form a covalently bonded enzyme-substrate complex (46, 56, 57). The role of the catalytic histidine is very different in different members of the α/β hydrolase-fold family. In the serine proteases, the histidine is needed both for formation and hydrolysis of the acyl-enzyme intermediate, since it activates the nucleophile by base catalysis and also donates a proton to the serine leaving group when the acyl-enzyme intermediate is hydrolyzed. In dienelactone hydrolase, which has a cysteine residue as the nucleophile, the catalytic histidine is involved only in maintenance of the nucleophilicity of the catalytic cysteine residue and not in hydrolysis of the covalently bound intermediate (10). In DhlA, which has an aspartate as catalytic nucleophile, the catalytic histidine residue is abstracting a proton from the water molecule which hydrolyzes the alkyl-enzyme intermediate (47, 56, 57). The breakdown of the covalent intermediate by the activated water molecule results in release of the product and restoration of the active-site residues. The acidic residue of the triad of the hydrolases facilitates the proton abstraction by histidine and stabilizes the positive charge on histidine electronically (58).

By analogy to the serine hydrolases, S95 of Qdo should be a likely candidate for an active-site nucleophile, which in its activated (deprotonated) form might perform a nucleophilic attack on the electropositive carbonyl carbon atom (C-4) of the substrate 1H-3-hydroxy-4-oxoquinoline, likewise leading to the formation of a covalent bond, i.e., an ester intermediate, as shown in Fig. 5. Formation of an ester bond, which involves cleavage of the N-heterocyclic ring of 1H-3-hydroxy-4-oxoquinoline, would thus generate an enzyme-bound substrate anion (Fig. 5). By further analogy to the serine hydrolases, the catalytic base (H244 of Qdo) may activate the nucleophilic serine residue by accepting a proton. The positively charged catalytic histidine residue may be stabilized by a yet unknown acidic residue (the triad acid). Alternatively, the protonated catalytic histidine of Qdo may interact with the enzyme-bound substrate anion, and so the substrate anion and the positively charged histidine might stabilize each other. In the latter case, it is even possible that Qdo and Hod are devoid of the catalytic acid of the canonical triad. Another alternative is that the substrate anion is stabilized by another positively charged residue of the enzyme (e.g., protonated H96 of Qdo, if H96, by analogy to W125 of DhlA, contributes to the active-site cavity).

FIG. 5.

Tentative hypothesis for the mechanism of Qdo-catalyzed 2,4-dioxygenolysis of 1H-3-hydroxy-4-oxoquinoline to carbon monoxide and N-formylanthranilic acid (for a detailed explanation, see the text). [Ser-O⁻], deprotonated serine residue of Qdo; [His-H⁺], protonated histidine residue of Qdo.

Qdo and Hod have been proven to be 2,4-dioxygenases, catalyzing the incorporation of a single molecule of oxygen at C-2 and C-4 of their respective substrate (6). Since dioxygen instead of a water molecule must be involved in the subsequent catalytic steps, the mechanistic role of the presumed catalytic amino acids of the hydrolases and the 2,4-dioxygenases is difficult to compare. As suggested previously, the C-2 carbanion form of the enzyme-bound substrate anion subsequently might be attacked by dioxygen, yielding a peroxy anion intermediate (6). This peroxy anion, via a five-membered cyclic peroxide transition form, could release the catalytic serine and eliminate carbon monoxide, thus generating the product N-formylanthranilic acid (Fig. 5).

Conclusions.

Although the structural relatedness of Qdo, Hod, and enzymes belonging to the α/β hydrolase-fold superfamily of enzymes and the catalytic significance of H244 and, presumably, S95 of Qdo have been established in this study, it is not yet clear whether Qdo actually possesses the canonical catalytic triad of the hydrolases and cofactor-free haloperoxidases. Despite the marked similarity of Qdo to serine hydrolases and other proteins belonging to the α/β hydrolase-fold superfamily, our tentative hypothesis on the mechanism of Qdo-catalyzed 2,4-dioxygenolysis is highly speculative and the structure-function relationship of Qdo is not yet clear. It is obvious that further investigations are needed to validate the roles of putative catalytically essential amino acid residues of Qdo. In particular, it would be most exciting to perform crystallization and X-ray diffraction analyses of the 2,4-dioxygenases in order to describe the active site and to elucidate the catalytic mechanism of 2,4-dioxygenolysis.