Identification of Catalytic Residues in Glyoxal Oxidase by Targeted Mutagenesis*

Glyoxal oxidase is a copper metalloenzyme produced by the wood-rot fungus Phanerochaete chrysosporiumas an essential component of its extracellular lignin degradation pathways. Previous spectroscopic studies on glyoxal oxidase have demonstrated that it contains a free radical-coupled copper active site remarkably similar to that found in another fungal metalloenzyme, galactose oxidase. Alignment of primary structures has allowed four catalytic residues of glyoxal oxidase to be targeted for site-directed mutagenesis in the recombinant protein. Three glyoxal oxidase mutants have been heterologously expressed in both a filamentous fungus (Aspergillus nidulans) and in a methylotrophic yeast (Pichia pastoris), the latter expression system producing as much as 2 g of protein per liter of culture medium under conditions of high density methanol-induced fermentation. Biochemical and spectroscopic characterization of the mutant enzymes supports structural correlations between galactose oxidase and glyoxal oxidase, clearly identifying the catalytically important residues in glyoxal oxidase and demonstrating the functions of each of these residues.

Glyoxal oxidase is a copper metalloenzyme produced by the wood-rot fungus Phanerochaete chrysosporium as an essential component of its extracellular lignin degradation pathways. Previous spectroscopic studies on glyoxal oxidase have demonstrated that it contains a free radical-coupled copper active site remarkably similar to that found in another fungal metalloenzyme, galactose oxidase. Alignment of primary structures has allowed four catalytic residues of glyoxal oxidase to be targeted for site-directed mutagenesis in the recombinant protein. Three glyoxal oxidase mutants have been heterologously expressed in both a filamentous fungus (Aspergillus nidulans) and in a methylotrophic yeast (Pichia pastoris), the latter expression system producing as much as 2 g of protein per liter of culture medium under conditions of high density methanol-induced fermentation. Biochemical and spectroscopic characterization of the mutant enzymes supports structural correlations between galactose oxidase and glyoxal oxidase, clearly identifying the catalytically important residues in glyoxal oxidase and demonstrating the functions of each of these residues.
Glyoxal oxidase (GLOX) 1 from the wood-rot fungus Phanerochaete chrysosporium is a secreted enzyme that functions as an extracellular factory for production of hydrogen peroxide, fueling peroxidases (lignin peroxidase and manganese peroxidase) that are responsible for microbial lignin degradation (1)(2)(3). Glyoxal oxidase catalyzes the oxidization of aldehydes to carboxylic acids, coupled to reduction of dioxygen to hydrogen peroxide, RCHO  The enzyme has fairly broad specificity for the reducing substrate, and a variety of simple dicarbonyl and hydroxycarbonyl compounds have been shown to support turnover. However, there is biological evidence that P. chrysosporium specifically secretes simple dicarbonyls (glyoxal and methylglyoxal) to drive this reaction. Further metabolism of the glyoxylic acid product leads to formation of oxalic acid, which has been identified as a cofactor for manganese peroxidase turnover (4). Previous studies of glyoxal oxidase (5) have demonstrated that it is a copper metalloenzyme containing an unusual free radical-coupled copper active site similar to that found in galactose oxidase (GAOX) (6,7). In these enzymes, an amino acid side chain radical ligates the active site metal ion, forming a catalytic motif characteristic of a class of enzymes known as radical copper oxidases. This radical-copper complex acts as a two-electron redox active site, a distinction from other free radical enzymes (ribonucleotide reductase, etc.) (8 -12) which typically exhibit single-electron reactivity. Glyoxal oxidase is isolated as an inactive, reduced form lacking the free radical, and requires treatment with a strong oxidant (e.g. Ir(IV) or Mo(V)) for activation (5).
The crystal structure of galactose oxidase (13) (Fig. 1) shows the active site metal ion coordinated by four protein ligands, including two histidines (His 496 and His 581 ) and two tyrosines (Tyr 272 and Tyr 495 ), forming a roughly square pyramidal metal complex. Two of the metal ligands (Tyr 495 and His 496 ) occur as consecutive residues in the protein sequence. The tyrosine ligands are distinct both in terms of coordination mode and, more especially, covalent modification of the side chain. In particular, one of the tyrosines (Tyr 272 ) is cross-linked to a cysteinyl residue, forming a Tyr-Cys dimer site (13) that has been identified in spectroscopic (14,15) and modeling (16 -18) studies as the radical redox site in the enzyme. The second tyrosine (Tyr 495 ) is bound axially to the active site metal ion, giving rise to distinctive features in the electronic spectra of the enzyme in the resting (Tyr ON ) state and this residue has been shown to be displaced (producing a Tyr OFF complex) when anions replace water in the inner sphere of the metal complex (19). This displacement is coupled to a protonation event that implies that Tyr 495 can serve as a general base, capable of activating bound substrate by proton abstraction. A tryptophan residue (Trp 290 ) stacked over the thioether side chain (13) is thought to contribute to the unusual stability of the free radical-containing form of this enzyme, which can persist for weeks in the absence of reductants (20,21). Spectroscopic studies on the glyoxal oxidase (5) provide evidence for an active site structure nearly identical to that found in galactose oxidase, despite the distinct catalytic function. EPR spectroscopy has identified two nitrogen donor ligands in the Cu(II) complex, and a combination of optical absorption and resonance Raman spectroscopy have identified tyrosine and tyrosine-cysteine dimer residues in the active site.
Site-directed mutagenesis has been used to investigate the structural and catalytic roles of the active site residues in galactose oxidase (22)(23)(24)(25). Conservative mutagenesis of the "axial" tyrosine to phenylalanine (Y495F) (22,24) permanently converts the enzyme to a Tyr OFF form (19) lacking the spectroscopic signatures of axial tyrosinate coordination and virtually eliminates catalytic activity. Elimination of the cysteinyl residue that is involved in the Tyr-Cys redox cofactor of the mature protein by C228G mutagenesis (23) likewise profoundly alters both spectroscopic and catalytic properties of the copper center. Replacement of the active site tyrosine 272 by phenylalanine appeared to interfere with copper binding and protein stability, and the Y272F mutant of galactose oxidase could not be purified (23).
Crystallographic studies on glyoxal oxidase, a glycoprotein for which carbohydrate comprises approximately 15% of the total molecular mass (26), have been complicated by a twinning disorder in the protein crystals that may be related to the high degree of glycosylation. 2 In the absence of high resolution structure information, theoretical sequence comparison makes it possible to identify putative active site residues (27). Using predictions from a comparative sequence analysis, we have undertaken a mutagenic program targeting the key active site residues in glyoxal oxidase. Characterization of these mutant proteins clearly identifies the residues important for metal binding and catalysis.
Comparative Protein Sequence Analysis-Sequences for galactose oxidase and glyoxal oxidase were aligned by the BLAST local sequence similarity search program using the BLOSUM50 substitution scoring matrix for protein-protein comparison and default values for other constraints (29,30).
Plasmids-The A. nidulans expression vector (pGLAGLOX) utilizes the promotor, secretion signal, and terminator of the A. niger glucoamylase gene (glaA) which is highly expressed and induced by maltose (28). The vector also features A. nidulans sequences (argB; ANS1) which permit efficent transformant selection on minimal medium (32). The pGLAGLOX plasmid was used as a template for the production of mutant plasmids using the Stratagene QuickChange in Vitro sitedirected mutagenesis procedure for 20 cycles of amplification as described previously (31) with the appropriate primer set (C70A, 5Ј-CG-GTCCTGACGGACTCCTTCGCGGCCAGCGGTGCGCTGC-3Ј; Y135F, 5Ј-CTTGAGGAGCGGTGGTTCCCATCGTCCGTGCGCATTTTCG-3Ј; H153W, 5Ј-CATGATCATTGGTGGTTCGTGGGTCCTCACACCGTTC-TAC-3Ј; Y377F, 5Ј-CGATTCCGCGCATGTTCCACTCGACCGTG-3Ј). Sequences of the mutant plasmids were verified by DNA sequence analysis (Molecular Biology Core Facility, Oregon Regional Primate Research Center, Beaverton, OR) of the entire inserts.
Aspergillus Expression-A. nidulans protoplasts were transformed and argB ϩ transformants were cultured and screened immunologically for GLOX expression as described previously (28,32). Large scale culture for purification of the recombinant protein was performed as described previously (2,5,26).
Construction of Pichia Expression Vectors-Expression vectors for yeast fermentation were based on commercial Invitrogen pPICZB Zeocin TM -selection vector containing AOX promotor for methanol-regulated expression. The entire coding sequence including both sequences of the glucoamylase leader peptide (gla) and the sequence coding for the mature protein of glyoxal oxidase (GLOX) was transferred from the corresponding wild type and mutant pGLAGLOX vectors by polymerase chain reaction amplification and ligated with the pPICZB plasmid to form the pPICZglaGLX series of vectors. Ligation products were transformed into Escherichia coli and selected on one-half ϫ LB agar containing 25 mg/liter Zeocin TM . Positive clones were isolated and purified plasmids were sequenced to confirm the fidelity of the insert.
Pichia Electrotransformation-PmeI-linearized plasmid DNA was electrotransformed into electrocompetent P. pastoris and transformants were selected on YPD agar containing 1 M sorbitol and 1 g/liter Zeocin TM . The plates were incubated for 3 days at 30°C allowing selection for clones containing multicopy tandem inserts. Transformants were screened for methanol-induced recombinant protein expression as described by the manufacturer using SDS-polyacrylamide gel electrophoresis detection of the secreted protein.
High Density Fermentation (33)-Large scale Pichia cultivation for protein biomass production was performed in a BioFlo 3000 Bioreactor equipped with a 14-liter fermentation vessel. Overnight cultures of recombinant P. pastoris were used to inoculate 5 liters of FM22 medium with PMT4 trace metals, and glycerol batch fermentation was continued for approximately 20 h with regulated air/oxygen feed to maintain 40% dissolved oxygen level. The medium was maintained at pH 4.5 by automatic addition of 5 N NH 4 OH, L-histidine (40 g/liter) was added continuously during fermentation (10 ml/h), and the temperature regulated at 29°C. When the initial quantity of glycerol was consumed, glycerol feed phase was initiated with delivery of 300 ml of 10% glycerol at 15 ml/min. At the end of glycerol-fed phase, methanol feed was initiated using a Watson-Marlow peristaltic pump for continuous-duty delivery of methanol (containing 4 ml of PMT4 trace metals solution per liter of methanol) to the culture. Methanol delivery was continued at a rate of 24 ml/h for 3 days.
Biochemical Methods-Secreted proteins were isolated from cell-free fermentation broth of high density P. pastoris cultures. Yeast cell biomass was removed by centrifugation and the supernatant was made saturated in ammonium sulfate by addition of ultrapure reagent (ICN Biochemicals) at room temperature. The solution was centrifuged three times at 20,000 ϫ g at 30°C and the precipitate was dialyzed versus 5 mM NaH 2 PO 4 buffer (pH 7). Chromatographically pure glyoxal oxidase was obtained as described previously (2,26). For GLOX(C70A) an additional chromatofocussing chromatography step was included using PBE-94 polybuffer exchanger and Polybuffer-74 ampholyte. Protein concentration was determined using a specific extinction for glyoxal oxidase (E 280 nm 1% ϭ 5.1) based on the method of Lowry et al. (40). Enzyme activity was measured polarographically with an oxygen-sensitive electrode using methylglyoxal (5 mM) as substrate in 100 mM 2,2Ј-dimethylsuccinic acid-KOH buffer (pH 5.0) in the presence of 20 M Na 2 IrCl 6 (5). The response of the Clark electrode was calibrated as described previously (21). Oxidized enzyme was prepared by addition of a slight excess of Na 2 IrCl 6 to the protein solution. Reconstitution of C70A and Y135F apoproteins with CuCl was performed as described previously (14).
Spectroscopic Measurements-Optical absorption spectra were recorded using a Varian Instruments Cary 5 UV-visual-NIR absorption spectometer interfaced with a microcomputer for data acquisition. Circular dichroism (CD) spectra were recorded using an AVIV Associates Model 40DS UV-visual-NIR dichrometer as described previously (34). Electron paramagnetic resonance spectra were recorded on a Varian E-109 X-Band EPR spectrometer equipped with an Air Products helium flow cryostat. g Value calibration was performed using a powdered sample of ␣,␣Ј-diphenyl-␤-picryl hydrazyl (g av ϭ 2.0037 (35)) as standard. Spin quantitation of paramagnetic samples was performed using a Cu(ClO 4 ) 2 spin standard (10 mM, in 2 M NaClO 4 ). EPR spectral simulations were computed using the program sim15 (Quantum Chemistry Program Exchange QCPE265). Metal ion analyses were performed using a Varian Instruments SpectrAA atomic absorption spectrometer equipped with a GTA 96 graphite furnace for high sensitivity metal determinations.

RESULTS
Sequence Comparison-Glyoxal oxidase and galactose oxidase share a modest protein sequence similarity including 28% identity over their primary structures (Fig. 2). The sequence of glyoxal oxidase, the smaller of the two proteins, matches the C-terminal three-quarters of galactose oxidase, implying that glyoxal oxidase lacks a ϳ150-residue N-terminal domain that is present in the larger enzyme. However, alignment of the protein sequences indicates that the critical active site residues characteristic of the radical copper oxidases (highlighted in Fig.  2) are conserved between these structures, consistent with spectroscopic comparisons (5) that demonstrate a close structural similarity of the active sites in these two enzymes.
Both sequences contain a unique pair of consecutive tyrosine and histidine residues (Tyr 495 -His 496 in GAOX, Tyr 377 -His 378 in GLOX). In galactose oxidase (13) both of these residues serve as copper ligands arising within a single turn in the protein fold. This metal binding loop feature occurs within a conserved RXYHS pentapeptide motif in the alignment with glyoxal oxidase, supporting the identification of Tyr 377 and His 378 in that enzyme as metal ligands corresponding to Tyr 495 and His 496 in GAOX. Similarly, a histidine residue that occurs approximately 90 residues later in both sequences may be assigned as the second conserved histidine ligand. The redox active tyro-sine (GAOX Tyr 272 ) occurs in a hexapeptide context RXYXSS that is common to both proteins and leads to the prediction that, in glyoxal oxidase, Tyr 135 is the radical-forming tyrosine metal ligand. The cysteinyl residue that forms the Tyr-Cys linkage in galactose oxidase (Cys 228 ) also has a counterpart in the glyoxal oxidase sequence (Cys 70 ). On the other hand, the tryptophan that overlies the Tyr-Cys thioether in galactose oxidase (Trp 290 ) ( Fig. 1) does not appear to be conserved, being replaced by histidine (His 153 ) in the glyoxal oxidase sequence. However, this replacement is conservative in that it would preserve the hydrogen bonding characteristic of the indole ring in the pyrrole-type bonding of the imidazole ring, and the Trp and His residues both follow a conserved IGGS pattern in their respective sequence. These sequence correlations permit a strategy of targeted mutagenesis to be used to identify catalytic residues in GLOX by conservative substitution.
Heterologous Expression-Expression levels for GLOX C70A, Y135F, and Y377F mutant proteins in P. pastoris high-density fermentation cultures averaged 80-fold higher than the levels for the same protein variant in A. nidulans shake cultures (1-2 g/liter from Pichia cultures compared with 25-40 mg/liter for Aspergillus expression). GLOX(C70A), which had relatively low expression levels in both systems, was produced at 100 mg/liter in Pichia fermentation but was detectable in only trace amounts in Aspergillus culture. One of the mutants (H153W) could not be detected in either case.
Y377F-GLOX(Y377F) was isolated from expression cultures as a copper complex containing slightly substoichiometric amounts of metal ion (0.7 (Pichia) or 0.75 (Aspergillus) copper/ active site) but with 4 orders of magnitude lower catalytic activity than the wild type enzyme (Table I). Optical spectra for the red-colored Y377F complex (Fig. 3, right, (ϩ2.8)). The spectra of the complex reflect almost no sensitivity to pH over the range 5.4 to 8.2 (Fig. 4, top), with only a slight red shift of the absorption at the higher pH range. Treatment of the Y377F mutant with a powerful oxidant (Ir(IV)) leads to dramatic increase in the intensity of the visible absorption, with appearance of a strong band near 450 nm and a weaker, broader feature near 900 nm in the near IR (Fig. 3, left,  The EPR spectrum of the Y377F mutant in the absence of oxidant (Fig. 5A) was associated with ground state parameters characteristic of a tetragonal Cu(II) complex: g x ϭ 2.01, g z ϭ 2.27; a 2 Cu ϭ 174 G. Quantitation of the EPR signal (by double integration calibrated by a spin standard) indicates that it represents 0.6 spins/active site. The envelope of the m I ϭ ϩ 1 ⁄2 copper hyperfine feature (⌫ FWHM ϭ 38 G) is structured, five components being resolved in the derivative of the experimental EPR spectrum in a symmetric pattern with 12 G splitting (data not shown). The g Ќ region of the spectrum is richly structured. The EPR spectrum is perturbed in the presence of NaN 3 (Fig. 5B) (g x ϭ 2.01, g z ϭ 2.25; a 2 Cu ϭ 176 G, representing 0.6 spins/active site) and the envelope of the m I ϭ ϩ 1 ⁄2 copper hyperfine feature is broadened (⌫ FWHM ϭ 45 G). Addition of an excess of Ir(IV) to the sample virtually eliminated the Cu(II) EPR absorption and a new signal appeared near the freeelectron g value (g ϭ 2.0), the entire spectrum quantitating to 0.02 spins/active site (Fig. 5C). Expanding the g ϭ 2 signal (Fig.  6) reveals a structured spectrum that can be simulated (Fig. 6, bottom) in terms of a nearly axial electronic g-tensor and hyperfine coupling to two dissimilar protons.
C70A-GLOX (C70A) was isolated as a metal-free apoenzyme (Ͻ0.01 copper/active site) that could be reconstituted by anaerobic incubation with CuCl to restore the full complement of copper to the enzyme (Table I). Both apo and metallated forms exhibited dramatically reduced catalytic activity, and a 100-fold increase in copper content on reconstitution resulted in less than 2-fold increase in the measured specific activity. The optical spectrum of the reconstituted C70A mutant (Fig. 4, lower, A) exhibited absorption across the visible region with a resolved band near 465 nm and a shoulder near 700 nm (ABS (nm) (⑀ (M Ϫ1 cm Ϫ1 )): 465 (830), 700 (320)). The EPR spectrum of the reconstituted protein (Fig. 5D) lacks structure in the g Ќ region, with g x ϭ 2.05, g z ϭ 2.26; a 2 Cu ϭ 178 G. This spectrum quantitates to 0.95 spins/active site.  Y135F-Like the C70A mutant, GLOX(Y135F) was isolated as metal-free apoenzyme but could be reconstituted to full metal content (Table I). Similarly, the enzyme exhibited less than 0.05% of the wild type glyoxal oxidase catalytic activity in either metallated or apoforms. The copper complex exhibited weak and poorly resolved absorption in the visible spectrum (⑀ 600 ϭ 160 M Ϫ1 cm Ϫ1 ) (Fig. 4, lower, B). The EPR spectrum (Fig. 5E) reflects the full complement of enzyme-bound copper (1.0 spins/active site), and exhibits a rhombic splitting of g values (g x ϭ 2.05, g z ϭ 2.26; a 2 Cu ϭ 178 G) lacking resolved structure in the g Ќ region. DISCUSSION The crystal structure for galactose oxidase (13) shows that its active site, comprising four metal binding residues (Tyr 272 , Tyr 495 , His 496 , and His 581 ) together with a covalently attached cysteine (Cys 228 ), is located in a distinct catalytic domain exhibiting an unusual 7-fold ␤-barrel structure (which has been described as a "super-barrel" or "␤-flower"), a ring of seven protein wedges formed from ␤-strands, each wedge being four strands thick. The core of the ring is filled by a long, looping chain arising in a separate C-terminal domain that contributes one of the metal ligands (His 581 ). Within the compact and tightly organized basket structure of the super-barrel fold there are a large number of turns, often associated with proline and glycine residues in the sequence. This pattern of turns may account for the relatively large number of conserved prolines and glycines identified in the sequence alignment of GAOX and GLOX, representing 25% of all identical residues (Fig. 2), implying a common fold for the two proteins.
Because the residues forming the active site of galactose oxidase arise from remote segments of the linear polypeptide chain, large gaps separate most of the elements of the metalbinding site. This situation is quite different from what is commonly found in other metalloproteins, where short consensus sequences can often be identified as a metal binding motif. The most convincing theoretical evidence that the alignment correctly identifies active site residues is that these groups occur in a conserved context and that the spacings are well preserved between the two sequences, with the modularity of the super-barrel structure (27) determining the sequence distance between non-adjacent metal ligands. Thus, the gap between the two tyrosines that serve as metal ligands in galactose oxidase (Tyr 272 and Tyr 495 ) is 223 residues, representing four wedges containing an average of 55 residues. A similar gap (242 residues) occurs between the (putative) active site tyrosines (Tyr 135 and Tyr 377 ) in the GLOX sequence. The similarity of sequence context for the critical residues further supports the significance of the sequence correlations and their identification as structural motifs.
Mutagenesis of the putative active site residues provides a crucial experimental test of the predictions provided by these sequence comparison studies. While in vitro site-directed mutagenesis of structural genes has become straightforward, systematic expression of the protein products is more problematic.
We have found it difficult to consistently prepare stable transformants for high level production of GLOX protein in the A. nidulans expression system that has been useful for production of wild type protein, and have explored the methylotrophic yeast P. pastoris (36,37) as an alternative heterologous expression host. In Pichia, it is possible to directly select (by antibiotic resistance) for rare multicopy integrants that tend to produce the highest levels of protein through gene dosage effects. In addition, the strong induction provided by the alcohol oxidase (AOX1) promotor in Pichia and the efficient secretory pathways in this organism make it an extremely useful system for high level extracellular expression of recombinant proteins (36,37), allowing us to produce up to 2 g/liter of protein in high density fermentation for three of the four mutants chosen for this study. For the three highly expressed mutants (C70A, Y135F, and Y377F) the mutation involves conservative substitution by a less bulky amino acid, while the increased bulk of the H153W mutant may actually destabilize the protein, accounting for lack of detectable expression in that case. Efficient export of foreign proteins by heterologous expression host is known to depend on the nature of the N-terminal leader peptide signal sequence. We find that the 24-amino acid A. niger glucoamylase secretion signal (gla) served very effectively to direct protein export in Pichia.
All three mutants exhibited a dramatic loss of enzymatic activity that supports their identification as critical catalytic residues. Spectroscopic and biochemical characterization of the mutants confirms the structural analogies to galactose oxidase and demonstrate the roles of these residues in active site function. Y377F mutagenesis leaves the essential metal-binding site intact, allowing this mutant to be purified with nearly full copper complement. The optical absorption and CD spectra of the Cu(II) complex (Fig. 3, right) are characteristic of the Tyr OFF (low pH or ligand-bound) form of wt GLOX (Fig. 3,  right, inset), but, unlike the wt protein, the mutant does not convert at high pH to a distinct Tyr ON form (Fig. 4, top). Instead, only a slight red-shift in the absorption is observed, consistent with formation of a hydroxide complex at high pH. The EPR spectrum of the Cu(II) complex (Fig. 5A) has g z and a z Cu values characteristic of tetragonal Cu(II) site, and is similar to that found for wt GLOX including the presence of partly resolved ligand hyperfine splittings in the m I ϭ ϩ 1 ⁄2 copper hyperfine feature that reflect coordination by two nitrogen donors (e.g. bis-histidine ligation). Exogenous ligands (e.g. azide) are able to bind to the metal ion, and the azide complex is nearly indistinguishable from the same complex of wt enzyme by absorption, CD (Fig. 3) and EPR (Fig. 5B) spectroscopies. Direct azide ligation to Cu(II) produces additional ligand hy- perfine broadening in the EPR spectrum, due to the additional 14 N interaction in the complex. Together, these results confirm the assignment of tyrosine Tyr 377 to the "axial tyrosine" in the GLOX active site corresponding to Y495 in GAOX. The GLOX(Y377F) mutant retains the ability to form a radical-Cu(II) complex (shown by the disappearance of the Cu(II) EPR signal) (5) but this complex is inactive, as expected based on our previous identification of the axial tyrosine requirement for substrate activation in the radical-copper oxidase turnover mechanism. The absence of any significant Cu(II) signal in the oxidized complex contrasts with results reported for the axial tyrosine mutant of GAOX (Y495F) where only ϳ10% of the expected radical-coupled copper complex was formed (24).
The optical spectrum of the oxidized Y377F mutant (Fig. 3, left) exhibits strong absorption features that correspond to the Tyr OFF form of the radical-containing wt GLOX (Fig. 3, left,  inset). The absence of an intense near IR absorption feature of the oxidized wt GLOX, previously assigned to a ligand-to-ligand charge transfer absorption between axial and radical tyrosine ligands in galactose oxidase, provides additional evidence that Tyr 377 is the axial tyrosine ligand in GLOX. The observation that azide binding does not affect the near IR absorption feature strengthens the correlation with the Tyr OFF form of the active wt enzyme. The oxidized Y377F protein exhibits a minority free radical EPR signal (Fig. 5C) (Ͻ0.02 spins/active site) that probably arises within the ϳ30% apoprotein in the sample. A similar signal is observed in wt enzyme under the same conditions (5). The signal is characteristic of a phenoxyl radical (g av ϭ 2.006) and spectral simulation requires a relatively axial g tensor that is the signature of the Tyr-Cys radical species (Fig. 6). The g values are slightly larger than observed for the Tyr-Cys radical in galactose oxidase, indicating distinct radical environments in the two proteins. The observation of a redox active Tyr-Cys group in the protein demonstrates that although Tyr 377 is required for turnover it is not essential for biogenesis of the radical cofactor.
Two mutants that are predicted to affect the equatorial ligation of copper (C70A and Y135F) are both isolated mainly as apoprotein, although reconstitution restores full metal content and permits spectroscopic characterization of the complex. Reconstituted C70A mutant exhibits an optical spectrum characteristic of Cu(II)-phenolate complexes (Fig. 4, bottom, A), with absorption near 465 nm typical of phenolate to Cu(II) ligandto-metal charge transfer, but at longer wavelength and somewhat lower intensity than found for phenolate-Cu(II) spectra of wt GLOX (Fig. 3, right, inset). The corresponding galactose oxidase C228G mutant was also reported to contain substoichiometric copper as isolated and exhibited optical spectra similar to that found here although with slightly lower intensities. In GAOX this Cys residue forms a covalent bond to the redox-active Tyr, and its substitution by an unreactive amino acid prevents formation of the mature redox cofactor. The spectra in these complexes arise from a tetragonal Cu(II) having an unmodified Tyr phenolate coordinated in the equatorial plane. The EPR spectrum of the Cu(II) complex (Fig. 5D) is relatively broad and lacking resolved ligand hyperfine structure, implying a degree of heterogeneity in the metal-binding site that is consistent with the greater mobility afforded an uncross-linked tyrosine residue. Y135F mutagenesis has an even greater effect on the optical spectrum (Fig. 4, bottom, B), dramatically lowering the absorption intensity with loss of the 465 nm absorption assigned to equatorial tyrosinate coordination. These features, and the absence of radical forming cofactor in GLOX(Y135F) mutant allows identification of Tyr 135 as a copper ligand and the radical redox site corresponding to Tyr 272 in GAOX. EPR evidence for a tetragonal Cu(II) complex (Fig. 5E) may imply that a hydroxide ion substitutes for the tyrosinate oxygen in the mutant protein, forming a tetragonal complex together with a second water and the two nitrogenous ligands.
For all three mutants, replacement of residues predicted to be essential for catalysis dramatically reduces catalytic activity (Table I), leaving a small but detectable activity approximately 10-fold greater than the background observed for a nonspecific protein control (bovine serum albumin). This activity might reflect a residual activity of the mutant active sites, as concluded in previous studies of the corresponding mutants of galactose oxidase (23). However, that interpretation cannot account for the 5-fold difference in activity observed for Y377F mutants expressed in different hosts (Aspergillus and Pichia), for proteins that are nearly indistinguishable in terms of spectra and metal content. Furthermore, the residual activity of the mutant proteins does not appear to correlate with metal content. Thus, for the C70A mutant, a 100-fold increase in copper content resulted in only a doubling of the specific activity, and similarly a 30-fold increase in copper content was associated with only 4-fold higher specific activity for reconstituted GLOX(Y135F) ( Table I). The residual activity is, in fact, so minute (0.01-0.04% of wild type) as to be consistent with a minority fraction of wild type enzyme that might result, for example, from unselected spontaneous reversion or, more likely, from missense translation errors (38,39). Missense substitution and processivity errors have been reported to rise as high as 1% during overexpression of heterologous proteins (39).
In conclusion, mutagenesis of GLOX permits identification of three catalytic residues (Cys 70 , Tyr 135 , and Tyr 377 ) and demonstrates their respective functions in the active site. Based on these studies, the active site of glyoxal oxidase can be represented as shown in Fig. 1. As proposed for GAOX, an axial tyrosine (Tyr 377 in GLOX) will function as a general base required for proton abstraction from the coordinated hydroxyl group, activating the substrate for oxidation. In addition to this basic group, radical-redox site derived from a tyrosine residue by covalent coupling to cysteine must also be present in the enzyme to support turnover. In GLOX this site is formed from Cys 70 and Tyr 135 , with the latter side chain coordinating directly to copper in the active site. Elimination of any of these groups removes an essential catalytic function and blocks turnover.