Cu(I)-dependent biogenesis of the galactose oxidase redox cofactor.

Galactose oxidase is a copper metalloenzyme containing a novel protein-derived redox cofactor in its active site, formed by cross-linking two residues, Cys228 and Tyr272. Previous studies have shown that formation of the tyrosyl-cysteine (Tyr-Cys) cofactor is a self-processing step requiring only copper and dioxygen. We have investigated the biogenesis of cofactor-containing galactose oxidase from pregalactose oxidase lacking the Tyr-Cys cross-link but having a fully processed N-terminal sequence, using both Cu(I) and Cu(II). Mature galactose oxidase forms rapidly following exposure of a pregalactose oxidase-Cu(I) complex to dioxygen (t(1/2) = 3.9s at pH7). In contrast, when Cu(II) is used in place of Cu(I) the maturation process requires several hours (t(1/2) = 5.1 h). EDTA prevents reaction of pregalactose oxidase with Cu(II) but does not interfere with the Cu(I)-dependent biogenesis reaction. The yield of cross-link corresponds to the amount of copper added, although a fraction of the pregalactose oxidase protein is unable to undergo this cross-linking reaction. The latter component, which may have an altered conformation, does not interfere with analysis of cofactor biogenesis at low copper loading. The biogenesis product has been quantitatively characterized, and mechanistic studies have been developed for the Cu(I)-dependent reaction, which forms oxidized, mature galactose oxidase and requires two molecules of O2. Transient kinetics studies of the biogenesis reaction have revealed a pH sensitivity that appears to reflect ionization of a protein group (pKa = 7.3) at intermediate pH resulting in a rate acceleration and protonation of an early oxygenated intermediate at lower pH competing with commitment to cofactor formation. These spectroscopic, kinetic, and biochemical results lead to new insights into the biogenesis mechanism.

extracellular space by oxidizing simple alcohols and subsequently reducing dioxygen to H 2 O 2 (1)(2)(3)(4). Together with a closely related enzyme, glyoxal oxidase (5,6), galactose oxidase represents a family of radical-copper oxidases defined by the presence of an unusual free radical-coupled copper active site (comprising a free radical associated with a redox-active metal ion) that functions as a two-electron redox unit in substrate oxidation and O 2 reduction (7,8). These free radical enzymes (9, 10) generate catalytic free radicals by reversible oxidation of a tyrosyl side chain in the protein (11). X-ray crystallography has revealed that the active site of galactose oxidase (Fig. 1) contains a novel post-translational modification, a covalent bond between Cys 228 and Tyr 272 creating a new, thioetherbridged cross-linked amino acid, tyrosyl-cysteine (Tyr-Cys), without addition of exogenous atoms (12, 13) (Scheme 1). Spectroscopic and chemical modeling studies have demonstrated that this Tyr-Cys side chain is, in fact, the free radical site in the protein (15)(16)(17)(18)(19). Characterization of the radical copper catalytic motif in galactose oxidase is complicated by the presence of multiple species in the as-isolated enzyme, including a significant fraction of apoenzyme, distinct forms of the metal-and cofactor-containing holoenzyme differing essentially in the number of electrons in the active site, as well as a portion of holoenzyme that is unable to generate a free radical complex and therefore does not contribute to catalytic activity (2,7,8).
X-ray structural studies are revealing cross-linked amino acid side chains in the active sites of a number of other redox metalloenzymes, including cytochrome c oxidase (20 -22) (tyrosyl-histidine), catechol oxidase, hemocyanin (23-25) (histidyl-cysteine), and methylamine dehydrogenase (26,27) (tryptophanyl-tryptophan). The origins and functions of these specialized elements of protein structure are just beginning to be investigated in detail. Formation of the tryptophanyl-tryptoquinone catalytic cofactor in methylamine dehydrogenase appears to involve processing by ancillary enzymes (28). In contrast, galactose oxidase cofactor biogenesis appears to be a self-processing event requiring only copper and O 2 (29). Galactose oxidase is distinct from the other examples in being a secretory protein, which means that its release in functional form involves at least three processing steps as follows: cleavage of the prepro leader sequence that directs translocation of the nascent polypeptide chain into the secretory pathway, metal binding, and cofactor biogenesis. The structure of a partially processed precursor form of the protein, containing the 17-amino acid N-terminal prosequence leader peptide but lacking both copper and cofactor, has been reported recently (30). The active site region of the unprocessed protein in the crystal exhibits large amplitude displacements of critical active site residues, reflecting a significant degree of conformational flexibility in the incompletely processed protein. Covalent modification of the active site cysteine to a sulfenic acid (Cys 228 -SOH) observed in the crystal may reflect oxidative damage to the protein. The precursor form of galactose oxidase has been shown to form spontaneously the cofactor in the presence of Cu(II) and dioxygen (29). The successful heterologous expression of galactose oxidase in Pichia pastoris using non-native leader peptide to direct secretion indicates that the native prepro signal is not strictly required for either secretion of the protein or biogenesis of functional cofactor (31). Therefore, in this work we have investigated in vitro biogenesis of the galactose oxidase redox cofactor, making use of recombinant protein containing the fully processed N terminus (no signal sequence) and lacking the thioether bond between Cys 228 and Tyr 272 . Extensive and quantitative analysis of the cross-linking reaction and thorough characterization of the biogenesis product (by SDS-PAGE, activity measurements, optical absorption spectroscopy, and metal quantitation) is now leading to new insight into the mechanism of cofactor formation in galactose oxidase.

EXPERIMENTAL PROCEDURES
Biological Materials-P. pastoris X33 (32) was obtained from Invitrogen. Recombinant galactose oxidase was purified from high density methanol fermentation medium (33) of a P. pastoris transformant prepared by multicopy chromosomal integration of an expression cassette comprising the pPICZ Zeocin-selection plasmid (Invitrogen) linearized within the AOX1 promoter by digestion with PmeI. The coding region of the expression cassette contains a 5Ј-nucleotide sequence coding for either the Aspergillus niger glucoamylase leader peptide (Gla) or the Saccharomyces cerevisiae ␣-mating factor leader peptide (␣MF) spliced with the cDNA sequence corresponding to the secreted galactose oxidase protein. The galactose oxidase coding sequence was modified for expression of GAOX mutational variants (C228G, Y272G) using Stratagene (La Jolla, CA) QuikChange TM site-directed mutagenesis kit, and the protein was expressed and purified as described previously. For secretion of the unprocessed, cofactor-free pregalactose oxidase protein, the transformant was grown as described previously (31) except that the copper content of the medium was reduced to 15% of the amount described for the PTM4 trace metals supplement in the earlier work (33). The other modifications to fermentation conditions reported previously (31) were the elimination of casamino acids from the methanol induction phase and the exclusion of copper from the methanol feedstock during expression. Pregalactose oxidase was purified as described previously (31) except that 2 mM EDTA was present in all buffer solutions. For biogenesis studies, the protein was passed through a gel filtration column to remove EDTA. Protocatechuate 3,4-dioxygenase was isolated from Brevibacterium fuscum as described previously (34).
Biochemical Methods-Protein concentrations of purified galactose oxidase and pregalactose oxidase were determined by optical absorption measurements, using the molar extinction coefficient at 280 nm (⑀ 280 ϭ 1.05 ϫ 10 5 M Ϫ1 cm Ϫ1 ) as reported previously (35). Proteins resolved by SDS-PAGE (Bio-Rad ready-gels) were stained with GelCode Blue TM staining solution (Pierce). Gels were digitized using a scanner and analyzed with tnimage Measurement and Analysis program (36). Strip densitometric scan data were further analyzed using the line shape deconvolution routines of the Grams spectral analysis program (Galactic Industries Corp., Salem, NH). N-terminal sequence analysis of purified pregalactose oxidase was performed by Debra A. McMillen at the Biotechnology Laboratory, Institute of Molecular Biology, University of Oregon.
Quantitation of free sulfhydryl groups in pregalactose oxidase, galactose oxidase, and variants was done using the DTNB test on protein in 4 M guanidinium hydrochloride using cysteine as standard (37). Samples were heated in a water bath at 100°C for 1 min and cooled on ice before addition of DTNB. Deblocking of cysteine sulfenic acid groups was performed as described (38 -40). Briefly, protein was incubated with 20 mM L-ascorbate in 50 mM MES buffer, pH 5.3, for 0.5 h, and the reductant was removed by gel filtration.
Biogenesis of the galactose oxidase cofactor from pregalactose oxidase and Cu(I) was conducted under argon purge in 20 mM MOPS, pH 7. Tetrakis(acetonitrile) Cu(I) hexafluorophosphate [Cu(I)-(CH 3 CN) 4 ⅐PF 6 ] was dissolved in anaerobic acetonitrile immediately before addition of an aliquot of this solution to argon-purged pregalactose oxidase. After mixing, the pregalactose oxidase-Cu(I) solution was rapidly purged with pure oxygen. Quantitation of the oxygen stoichiometry for cofactor biogenesis was performed by adding aliquots of air-saturated buffer to a mixture containing pregalactose oxidase and a substoichiometric amount of Cu(I). The pregalactose oxidase-Cu(I) complex was prepared by combining 150 nmol of Cu(I)(CH 3 CN) 4 ⅐PF 6 with 600 nmol of pregalactose oxidase in a 3-ml anaerobic cuvette under argon. The optical absorption of the stirred solution was monitored at 445 nm.
The mutual stability of pre-GAOXCu(I) and fully oxidized mature GAOX in a mixture containing both was determined by sequential preparation of a pre-GAOXCu(I) complex, addition of oxidized GAOX, and O 2 in that order. Copper(I) acetonitrile (60 nmol, 0.6 eq) was added to an anaerobic solution of pre-GAOX (100 M, 1 ml) under argon. An anaerobic solution of fully oxidized mature GAOX (AGAOX) (60 nmol in 50 l) was added, and the optical absorption spectrum was recorded and monitored at 445 nm. The cuvette was then purged with O 2 gas and the optical spectrum recorded.
Copper analyses were performed using a Varian Instruments Spec-trAA atomic absorption spectrometer equipped with a GTA 96 graphite furnace. Galactose oxidase activity was measured by oxygen uptake in a Clark-type oxygen electrode. The assay mixture contained 50 mM O-methyl-␣-D-galactopyranoside, 2 mM K 3 Fe(CN) 6 , and 50 mM KH 2 PO 4 , pH 7. The reaction was monitored at 25°C, and the electrode was calibrated using the protocatechuic acid/protocatechuate dioxygenase reaction (41).
Spectroscopic Methods-Optical absorption spectra were measured using a Varian Instruments Cary 5E UV-visible near infrared absorption spectrophotometer. Electron paramagnetic resonance spectra were recorded on a Bruker E500 X-Band EPR spectrometer with a Super X microwave bridge and SHQ resonator equipped with a nitrogen flow cryostat. EPR signal quantitation was performed using Cu(II) perchlorate spin standard. First derivative solution EPR spectra were simulated using the program sim15 (Quantum Chemistry Program Exchange QCPE265). Solution EPR spectra were recorded for samples in a quartz flat cell (Wilmad Glass, Buena, NJ) at 305 K. Resonance Raman spectra were collected with a custom McPherson 2061/207 spectrograph (0.67-m focal length, 600 groove grating, 7 cm Ϫ1 spectral resolution) using a Coherent Inova 302 krypton laser (413 nm), a Kaiser optical super-notch filter, and a Princeton Instruments (LN-1100PB) liquid N 2 -cooled CCD detector. Spectra were obtained for samples in glass capillaries at 300 K using a 90°scattering geometry and a 20-min accumulation time. Sample integrity was verified by the observation of the same absorption spectrum before and after laser irradiation.
Transient Kinetics-The rate of formation of mature galactose oxidase from pregalactose oxidase and Cu(I) was measured using a Biologic SFM-300 rapid mixing stopped-flow module connected to an OLIS RSM-1000 rapid scanning monochromator via fiber optic light pipes. Prior to the rapid mixing experiment, the stopped-flow system was scrubbed free of oxygen using 50 ml of protocatechuic acid/protocatechuate dioxygenase mixture in 50 mM Tris, pH 8, for at least 24 h (42). All syringes were periodically emptied and refilled with the scrubbing reagents. During this period the syringe chamber was purged with nitrogen gas, and the circulating water bath reservoir was sparged with nitrogen. Collars fitted around the syringe ports on the Biologic SFM-300 head permitted a continuous argon purge of these points. Anaerobic solutions delivered to the syringe block were prepared in gas-tight tonometers. Cu(I)(CH 3 CN) 4 ⅐PF 6 in acetonitrile was added to an anaerobic solution of pregalactose oxidase in 20 mM MOPS, pH 7, that was subsequently transferred anaerobically to an argon-purged tonometer using a double-ended catheter needle. The cofactor biogenesis reaction was initiated by mixing this solution with oxygenated buffer in the stopped-flow system, monitoring the formation of fully oxidized cofactor-containing mature galactose oxidase. The pH sensitivity of the biogenesis reaction was investigated by shooting pregalactose oxidase/ Cu(I) (prepared anaerobically in buffer-free water) against airsaturated 40 mM buffer solutions in the SFM-300 syringe driver. The

RESULTS
Production of Pregalactose Oxidase-P. pastoris transformants containing galactose oxidase cDNA linked to either A. niger glucoamylase leader peptide or S. cerevisiae ␣MF leader peptide coding sequences efficiently secrete mature galactose oxidase under methanol-induced expression in complete medium (31). When copper supplementation was reduced in the glycerol batch growth phase and eliminated completely in the induction phase of high density methanol fermentation, the transformants produced pregalactose oxidase lacking the thioether cross-link between Cys 228 and Tyr 272 ( Fig. 1) but exhibiting the N-terminal sequence (ASAPI) of the authentic mature protein. The yield of purified pregalactose oxidase was ϳ200 mg/liter for a 5-liter fermentation culture harvested 1 day after induction. The background galactose oxidase activity of the purified pregalactose oxidase was about 0.1% (ϳ 0.7 mol of O 2 /mg of protein at 25°C) of the recombinant WT galactose oxidase. Analysis of the sulfhydryl content of pregalactose oxidase using the DTNB assay (Table I) shows that two representative preparations of pre-enzyme contain approximately twice as many free SH groups as mature WT galactose oxidase, confirming the presence of an additional, predominantly unblocked cysteine residue (Cys 228 ) in these preparations. The amount of free sulfhydryl varies somewhat between preparations.
Detection of Cross-linked Product by SDS-PAGE-Mature, cross-linked cofactor-containing galactose oxidase shows an altered mobility on SDS-PAGE gels (Fig. 2, lane 6) and is resolved from unmodified pregalactose oxidase (Fig. 2, lane 1), as reported previously (29). Mature galactose oxidase appeared following addition of Cu(I)(CH 3 CN) 4 ⅐PF 6 to pregalactose oxidase under anaerobic conditions followed by oxygen gas purging (Fig. 2, lanes 2-5). The amount of mature product formed was roughly proportional to Cu(I) added up to about 0.8 eq (Table I, pre-GAOX(A)). Lower conversion was observed in some preparations (Table I, pre-GAOX(B)), in which the amount of cross-linked product was proportional to Cu(I) added only up to about 0.5 eq. Thus, the upper limit to conversion appears to vary somewhat between pre-enzyme preparations. Similar results were obtained by addition of Cu(II)SO 4 to pregalactose oxidase in the presence of oxygen (Fig. 2, lane 7). Varying the sample concentration (from micromolar to millimolar) did not change the yield, and the protein began to precipitate when more than 1 eq of copper was added. Unmodified pre-GAOX appears to be more sensitive to precipitation in excess copper than the mature GAOX protein. Treating pre-GAOX with ascorbic acid at low pH (a method for deblocking cysteine sulfenic acid groups, Cys-SOH) did not increase the FIG. 1. Active site of mature galactose oxidase. The active site includes a novel redox cofactor formed from a covalent cross-link between Cys 228 sulfur S␥ and Tyr 272 ring carbon C⑀ arising from distinct turns of the ␤-propeller fold. This figure is based on crystallographic coordinates (Protein Data Bank code 1GOG) and has been rendered using Rasmol (14).
a Sulfhydryl content determined by DTNB assay as described under "Experimental Procedures." b Determination of free sulfhydryl content for two independent pre-GAOX preparations.

FIG. 2. SDS-PAGE analysis of cross-link formation in the cofactor biogenesis reaction.
Top, GelCode TM -stained 12% SDS-PAGE (loaded with 0.5 g of protein in each lane) resolving uncross-linked pre-GAOX and cross-linked mature GAOX product. Reaction conditions are as described under "Experimental Procedures" and in Table II yield of mature GAOX. Other attempts (varying salt concentration, pH, buffer, addition of superoxide and hydrogen peroxide scavengers, and repeating the Cu(I) treatment on the initially formed biogenesis product) also failed to increase the upper limit of conversion to the mature GAOX product.
Densitometric analysis of the scanned gel (Fig. 2, bottom) provides quantitative information on the yield of mature, crosslinked GAOX in the biogenesis reaction (Table II). Gaussian resolution of the individual protein bands in the stained SDS-PAGE is reported in Table II. The yield of cross-link correlates with the amount of copper added up to about 0.8 eq. The upper band on the gel (a), corresponding to pre-GAOX, is broader than that associated with the mature protein (b), and the broadening increases with the age of the sample.
Monitoring Formation of Fully Oxidized Cofactor in Optical Absorption Spectra-When freshly prepared Cu(I)-(CH 3 CN) 4 ⅐PF 6 was added to pre-GAOX under strictly anaerobic conditions, the spectroscopic signature of fully oxidized galactose oxidase appears immediately following exposure to oxygen (Fig. 3, spectra 2-5). The reaction is very fast, and rapid mixing stopped-flow techniques were required for more detailed kinetic analysis (see below). Quantitation of the amount of fully oxidized GAOX formed in these reactions, based on the published extinction coefficient for the mature protein-free radicalcoupled copper complex, corrected for inactive species (⑀ 445 ϭ 9500 M Ϫ1 cm Ϫ1 ) (41), shows that the yield is proportional to the amount of Cu(I) added up to about 0.8 eq, similar to the results found for cross-link formation in the SDS-PAGE gel analysis of the reaction products ( Fig. 2). No oxidized GAOX was observed prior to addition of O 2 to the pre-GAOX (Fig. 3, spectrum 0) or the pre-GAOXCu(I) mixture (Fig. 3, spectrum 1). The biogenesis products were analyzed for enzymatic activity and copper content before and after gel filtration (Table III). Galactose oxidase activity was essentially unchanged by gel filtration, whereas the copper content decreased slightly, indicating the presence of a fraction of relatively weakly bound metal ions in the biogenesis product that were not associated with functional active sites. The catalytic activity of the biogenesis product, normalized to copper, was similar to that observed for the as-isolated galactose oxidase (Table III). Higher normalized specific activity was observed at lower copper loading, possibly reflecting microheterogeneity within the functional fraction of pre-enzyme, with the fraction that binds Cu(I) most tightly yielding the most efficient processing. Comparison of the fraction of functional active sites formed in the reaction (Table III) with the fraction of sites converted to oxidized mature enzyme based on the absorption spectra ( Fig. 3) indicates that every functional site was represented in the 445 nm absorption of oxidized product. Regression of the absorption intensity data for in vitro cofactor biogenesis (from Fig. 3) against the cofactor concentration determined by densitometric analysis of the SDS-PAGE resolved products from these reactions (Fig. 2) yielded an empirical absorption coefficient for cofactor-containing product (⑀ 445 ϭ 5200 M(cofactor) Ϫ1 cm Ϫ1 ) which was very similar to the observed absorption coefficient for the asisolated enzyme (⑀ 445 ϭ 5100 M(protein) Ϫ1 cm Ϫ1 (7)). The maximum optical absorption at 445 nm and catalytic activity of the biogenesis product reached a plateau at 85% of the as-isolated enzyme ( Fig. 3; Table III).
In order to evaluate the stability of the pre-GAOXCu(I) complex in the presence of fully oxidized mature GAOX, the optical absorption spectra of anaerobic pre-GAOXCu(I) has been measured following addition of anaerobic fully oxidized AGAOX, as described under "Experimental Procedures." The resulting mixture exhibited the full absorption intensity of the added AGAOX, which was stable for at least 10 min. Purging this mixture with oxygen gas resulted in a further increase in the intensity of the AGAOX spectrum, yielding an absorption change ⌬A 445 ϭ 0.25, corresponding to 90% of the change observed for reaction of pre-GAOXCu(I) with O 2 in the absence of AGAOX (see above).
The stoichiometry of the O 2 reaction with pre-GAOXCu(I) was investigated at low copper loading (Fig. 4) and the lower range of conversion (spanning the first third of the complete reaction) to ensure the most efficient processing. Aliquot additions of airsaturated buffer produced a linear increase in absorption at 445 nm, corresponding to formation of the oxidized mature AGAOX product, and regression analysis of the data yielded a reaction stoichiometry of 1.8 O 2 /AGAOX product under these conditions. Cu(II) Reactions-In the presence of oxygen the addition of Cu(II)SO 4 to pre-GAOX led to the slow formation of mature cross-linked protein (Fig. 2, lane 7) exhibiting absorption features resembling those characteristic of the oxidized enzyme (Fig. 5). The slow conversion allowed the progress of the reac-  tion to be monitored by sequential scanning in a conventional UV-visible near infrared absorption spectrophotometer. As observed with Cu(I), the maximum conversion to mature protein was reached on addition of about 0.5-0.8 eq of Cu(II) metal ion (Fig. 2, lane 7). Some precipitation was observed during the slow biogenesis process, a side reaction that was exacerbated at higher Cu(II) concentrations. The onset of precipitation was evident in an increase in apparent optical absorption below 400 nm due to light scattering from protein aggregates. The time course of the optical absorption increase in the visible spectrum obeyed a simple single-phase exponential relaxation (Fig. 5, inset) associated with an effective first-order rate constant k ox ϭ 3.8 ϫ 10 Ϫ5 s Ϫ1 . Under anaerobic conditions, addition of Cu(II)SO 4 to pre-GAOX produces a solution with very weak optical absorption (⑀ 600 ϭ 200 M Ϫ1 cm Ϫ1 ) (Fig. 6A). This initially formed complex slowly transforms into a distinct form exhibiting strong absorption at 406 nm (⑀ 406 ϭ 3790 M Ϫ1 cm Ϫ1 based on copper added) (Fig. 6B). This complex (identified as a 410 nm feature (29)) has been observed previously as a transient species in a study of Cu(II)-dependent galactose oxidase maturation under aerobic conditions. Conversion to the 406 nm species followed a single phase first order exponential relaxation with k 406 ϭ 3.2 ϫ 10 Ϫ4 s Ϫ1 (Fig. 6, inset). The Cu(II) binds to protein immediately after addition to pre-GAOX solution, as shown by the appearance of well developed EPR signal from an axial Cu(II) complex (Fig.  7A). Quantitation of this spectrum by double integration of the derivative EPR spectrum and comparison with a spin standard showed that virtually all of the Cu(II) added (0.19 mM S ϭ 1 ⁄2 compared with 0.20 mM Cu(II) added, 95%) may be accounted for in the complex. In contrast, free Cu(II) in 50 mM MOPS, pH 7, buffer is electronically coupled (through dimer formation) or magnetically interacting in solution, and the EPR spectrum for this sample has relatively low integrated intensity (Ͻ 0.03 mM S ϭ 1 ⁄2, ϳ10%). The 406 nm species has an EPR spectrum   resembling, but distinct from, that observed for the initial complex ( Fig. 7B) with the full complement of copper contributing to the spectrum (0.21 mM S ϭ 1 ⁄2, 105%). Spectral simulations showed that the copper spectra alone were sufficient to account for the observed signals and that there were no other significant EPR features near g ϭ 2. Spectra of the latter complex obtained in solution at room temperature were essentially the same, exhibiting Cu(II) EPR features and lacking detectable free radical EPR signals near g ϭ 2 (Fig. 7C).
Resonance Raman (RR) spectroscopy has been used to characterize further the 406 nm species. The RR spectrum of the pregalactose oxidase Cu(II) complex (in 20 mM MOPS, pH 7, buffer) showed strong resonance enhancement of a fundamental at 343.5 cm Ϫ1 on excitation at 413 nm (20-milliwatt laser power), with smaller bands at 334 and 306 cm Ϫ1 . The RR spectrum is sharp (15 cm Ϫ1 full-width at half-height) corresponding to a well defined structure from a discrete copper complex. No other RR signals were observed at higher frequency. When this pre-formed 406 nm species (Fig. 6B) is exposed to oxygen, the spectra progressively develop the characteristic features of the oxidized mature GAOX (Fig. 8). The kinetics of this transformation are associated with a first order rate constant k 406,ox ϭ 3.9 ϫ 10 Ϫ5 s Ϫ1 (Fig. 8, inset).
Biogenesis of the Tyr-Cys cofactor with Cu(II) was blocked in the presence of EDTA (Fig. 9, spectrum E; inset, lane 3). Weak absorption is observed near 750 nm, consistent with formation of a Cu(II)-EDTA complex under these conditions. This spectrum did not change for up to 6 h, indicating that the EDTA chelate competes very effectively with pre-GAOX for the Cu(II) metal ion. In contrast, the Cu(I)-dependent biogenesis reaction was not affected by the presence of EDTA (Fig. 9, spectrum C; inset, lane 2). There was no evidence for the formation of Cu(II)-EDTA in this sample (Fig. 9, spectrum B), demonstrating that the copper remained in the Cu(I) oxidation state.
Transient Kinetics-The relatively rapid biogenesis reaction supported by Cu(I) and dioxygen, which occurs efficiently on a physiologically relevant time scale, has been developed for a more detailed kinetic and mechanistic analysis. Rapid kinetics analysis of the biogenesis reaction takes advantage of the intense optical spectrum of the oxidized GAOX product to monitor the formation of the protein cross-link and potentially detect other absorbing species involved in the reaction. Because the pre-GAOXCu(I) complex was extremely sensitive to O 2 , special precautions are required in its preparation and handling to exclude air and ensure anaerobiosis without compromising the stability of the oxidized GAOX product. We have found that preparation of the pre-GAOXCu(I) complex in an air-free tonometer allowed storage for the duration of the kinetics experiment and that argon purging of the luer inlet ports for the stopped-flow syringe head substantially improved the anaerobic transfer of the sample into the firing syringe. Nitrogen purging of the syringe chamber and scrubbing dioxygen from internal Teflon surfaces in the flow circuit by preincubation with a protocatechuic acid/protocatechuate dioxygenase mixture (41,42) also contributed to the anaerobic performance of the stopped-flow system.
The reaction of the preformed pre-GAOXCu(I) complex with air-saturated buffer at 20°C is shown in Fig. 10. The raw spectra (Fig. 10, top left) were recorded with an OLIS RSM-1000 rapid scan monochromator, collecting the complete 325-800 nm visible spectrum (3 nm resolution for acquisition) every 1 ms and averaging to generate 31 scans/s. Data were collected for 10 s or as required to reach an end point for the absorption transients. Raw binary data were subsequently analyzed using global fitting routines (Specfit/32, Spectrum Software Associ- FIG. 7. EPR spectra of pre-GAOXCu(II) complexes. EPR spectra (solid lines) of samples prepared by addition of 0.8 eq CuSO 4 to 0.25 mM pre-GAOX in 50 mM MOPS, pH 7, as described in the text. A, pre-GAOX ϩ Cu(II) frozen promptly; B, pre-GAOX ϩ Cu(II) after 4 h of incubation under argon. Instrumental parameters: microwave frequency, 9.398 GHz; modulation amplitude, 10 G; microwave power, 0.1 milliwatt; temperature, 100 K; 10 scan average. Simulations (dashed lines): A, g ʈ ϭ 2.21; g Ќ ϭ 2.05; a (copper) ϭ 170 G; B, g ʈ ϭ 2.225; g Ќ ϭ 2.055; a (copper) ϭ 175 G. The vertical bar indicates the g ʈ value for the two spectra. C, room temperature (305 K) solution EPR spectrum for pre-GAOX ϩ Cu(II) complex after 4 h of incubation under argon. Instrumental parameters: microwave frequency, 9.771 GHz; modulation amplitude, 10 G; microwave power, 2.0 milliwatts. The magnetic field strength was adjusted for the microwave frequency to align the spectra. a , nuclear hyperfine coupling constant.
FIG. 8. Reaction of pre-formed pregalactose oxidase Cu(II) complex with air. Sample described in Fig. 6 was exposed to air, and the progress of the reaction was followed by repetitive scanning (30 min intervals displayed). Inset, analysis of the absorption data in terms of single-exponential relaxation. The theoretical curve (solid line) is calculated for k 406,ox ϭ 3.9 ϫ 10 Ϫ5 s Ϫ1 . ABS, absorbance.
ates, Marlborough, MA). The noise-factored spectra (Fig. 10, left middle) illustrate the quality of spectroscopic information that may be evaluated in the large data sets resulting from rapid scan spectroscopy. The formation of the oxidized GAOX product occurs in a single exponential phase with a unique set of kinetic parameters giving excellent fits to the data over the entire spectral region (Fig. 10, right) and leaving uniform excursions for the residuals over the entire time course of the data. Allowing both starting and final species to contribute to the optical absorption spectrum leads to a resolution of spectral vectors as shown in Fig. 10 (lower left).
Varying the pH of the air-saturated buffer solution used to initiate the biogenesis reaction allowed the pH profile for the reaction to be systematically explored. In the higher range of pH (7 and above), the simple exponential relaxation kinetics described above provided an excellent description of the data, although the rate of the transformation showed a strong pH sensitivity. At lower pH, a very different product spectrum was observed lacking the NIR absorption characteristic of oxidized GAOX and exhibiting a blue-shifted absorption maximum (Fig.  11, top). The kinetics associated with this low pH transformation is clearly multiphasic, and the process has been analyzed in terms of competing first order reactions yielding oxidized GAOX and the 406-nm species of pre-GAOXCu(II) as the respective products (Fig. 11, bottom). The model is justified in terms of the following mechanistic interpretation shown, The two alternative pathways compete for the same reactive pre-GAOXCu(I) starting complex (A), and the relative rates determine the commitment to either cofactor biogenesis and formation of oxidized GAOX (B) (Reaction 1) or an abortive process in which the initially formed oxygenated complex breaks down to a pre-GAOXCu(II) complex (C) without forming the Tyr-Cys cross-link, and subsequently converts to the more strongly absorbing 406-nm species (D) (Reaction 2). In this scheme, only species B and D contribute significantly to the optical absorption spectrum. In order to explore this scheme, the full data set was resolved onto target spectra corresponding to pure oxidized GAOX and the 406 nm-absorbing (relaxed) form of the pre-GAOXCu(II) complex (Fig. 11, middle), providing estimates for the rate constants k 1 and k 2 for the two processes competing for the pre-GAOXCu(I) initial complex. As expected from previous characterization of the pre-GAOX-Cu(II) complex (Fig. 6), the rate constant for emergence of the 406-nm species (k 3 ) is orders of magnitude slower and not well determined by measurements on the seconds time scale. The spectroscopic resolution of the kinetic processes predicts a spectrum associated with C and defined by a weak absorption maximum above 500 nm (⑀ ϭ 800 M Ϫ1 cm Ϫ1 ), similar to that observed for the initial pre-GAOXCu(II) complex (Fig. 6A). The pH dependence of the rates is plotted in Fig. 12, which shows that above pH 7, Reaction 1 dominates, whereas at lower pH Reaction 2 becomes significant. SDS-PAGE analysis of the pH 5 reaction mixture (data not shown) shows a dramatic decrease in the yield of cross-linked product. The sensitivity of the biogenesis reaction to the isotopic composition of the solvent (H 2 O or D 2 O) has also been investigated over the pH range 6.5-8.35 at 20°C (Table IV). At low pH, the k H 2 O /k D 2 O observed for the rate of appearance of fully oxidized mature GAOX product reached a maximum SKIE value of ϳ1.5 and decreased as the pH was raised, approaching 1.0 above pH 8.

DISCUSSION
The active site of galactose oxidase has evolved two distinct but related functions, cofactor assembly and catalysis. The turnover reaction, which has been the focus of the majority of earlier work (2)(3)(4), depends on the presence of a correctly formed redox cofactor in the active site, and a mutational variant lacking the active site Cys 228 that is unable to form the Tyr-Cys cross-link is virtually inactive as a catalyst, even though it binds copper (43). Self-assembly of the Tyr-Cys redox cofactor is a second intrinsic, but noncatalytic, reactivity of the active site (29). Little information is presently available on the in vivo maturation of galactose oxidase, which likely occurs as a post-translational event in the trans-Golgi compartment of the secretory pathway. Copper is delivered to this compartment by a Cu(I)-specific transporter (for example, the CCC2 copper transporter in S. cerevisiae) (44). Previous work has demonstrated that the GAOX biogenesis reaction can occur in vitro when purified prosequence-containing precursor protein is treated with Cu(II) and O 2 (29). However, the observation that chimeric proteins based on a fusion between heterologous leader sequences (Gla or ␣MF) and the mature galactose oxidase protein sequence are efficiently expressed as functional, cofactor-containing enzyme by P. pastoris (31) implies that the native signal sequence is not specifically required for cofactor formation. As demonstrated in the present work, the purified recombinant pre-enzyme free of any signal sequence is also competent for cofactor biogenesis.
P. pastoris secretes pregalactose oxidase (rather than the mature enzyme) under copper-limiting conditions even without stringent exclusion of copper from the culture media, apparently through regulation of the metal ion distribution within the cell. All of the copper present in the medium can be accounted for in the Pichia cell mass following high density fermentation, implying efficient absorption of the trace metal. Intracellular copper may be specifically targeted to the mitochondria and peroxisomes in methylotrophic yeast, particularly under methanol induction (45), effectively restricting the amount of copper reaching the secretory pathway without impairing the robust respiratory proficiency required for methanol fermentation. In contrast, pre-GAOX expression in Aspergillus culture is reported to require strict exclusion of copper from the culture media (29), resulting in copper starvation and pleiotropic stress effects on the expression host that may account for the incomplete leader sequence processing reported under these conditions. The metal oxidation state dependence for cofactor biogenesis has also been investigated, comparing reactions based on Cu(I) FIG. 11. Reaction of pre-galactose oxidase/Cu(I) complex with O 2 at pH 5. Anaerobic preformed pre-GAOXCu(I) complex in unbuffered water was mixed with air-saturated buffer (40 mM MES, pH 5) in the stopped-flow spectrophotometer thermostated at 20°C, and the reaction was monitored by rapid scanning spectroscopy. The concentration of pre-GAOXCu(I) (after mixing) was 15 M. Top, noise-factored rapid scan spectral data recorded with 31-ms averaging. Every 10th scan is displayed. Middle, spectral vectors for complexes B (Ϫ⅐Ϫ), C (O), and D (---) based on the multiphase relaxation model described in the text for the low pH reaction using the spectra for oxidized mature GAOX and the 406 nm species as target spectra for complexes B and D, respectively. Bottom, kinetic time course extracted from the complete data set at 400 nm, together with theoretical fit (based on k 1 ϭ 3.58 ϫ 10 Ϫ3 s Ϫ1 , k 2 ϭ 4.47 ϫ 10 Ϫ2 s Ϫ1 , and k 3 ϭ 1.8 ϫ 10 Ϫ5 s Ϫ1 ), with the residuals shown below. ABS, absorbance.  and Cu(II) treatment. Copper in both oxidation states appears to support cross-linking and cofactor formation (Fig. 2), but the reactions occur on very different time scales. While the Cu(I)dependent reaction is completed in seconds (t1 ⁄2 ϭ 3.9 s at pH 7), cofactor formation following addition of Cu(II) occurs over several hours (t1 ⁄2 ϭ 5.1 h), so slowly, in fact, that we cannot rule out the possible involvement of a Cu(I) species arising from reduction of Cu(II) by minor contaminants in the sample. Cu(II) interactions with pre-GAOX are relatively complicated, and in the absence of O 2 , several distinct complexes are formed on different time scales. An initially formed weakly absorbing complex converts slowly to a distinct form exhibiting strong absorption near 406 nm in the visible spectrum (Fig. 6). The latter absorption feature may be assigned to a Cu(II)-thiolate ligand-to-metal charge transfer excitation based on resonance Raman enhancement of a Cu-S stretch mode (46 -48). The measured vibrational frequencies are characteristic of Cu-S stretch modes from tetragonal type II copper sites (49). The spectrum implies coordination by a single thiolate ligand, because tetragonal copper sites with two thiolate ligands display additional combination bands at higher frequency, which are not observed for this sample. The transition energy and intensity for the 406-nm absorption feature is typical of low-coordinate copper thiolate complexes, as have been characterized previously in mutational variants of Cu/Zn superoxide dismutase (46,47) and azurin (48). The appearance of the 406 nm absorption band reflects a slow isomerization of the active site structure permitting the Cys 228 sulfur to bind directly to the metal ion. The change in active site structure is also reflected in the perturbed EPR parameters for the protein-bound paramagnetic metal ion (Fig. 7). There is no evidence for stabilization of a protein radical within the anaerobic Cu(II) complex in the EPR spectra for any of these complexes at either cryogenic or ambient temperatures (Fig. 7).
The rapid Cu(I)-dependent biogenesis reaction may more closely resemble the physiological process. This reaction is easily monitored by observing the appearance of the optical absorption signature of fully oxidized mature (cofactor-containing) product following exposure of a preformed pre-GAOXCu(I) anaerobic complex (Fig. 3) to dioxygen. Characterization of biogenesis products formed over a range of copper loading indicates that there are two main subpopulations within the pre-GAOX with relative proportions that vary between preparations (Fig. 3). One subpopulation binds Cu(I) relatively tightly (and is therefore the first to titrate with metal when substoichiometric copper is added) and efficiently forms the cross-linked product on subsequent exposure to O 2 . Titration of this fraction with O 2 (Fig. 4) indicates that two molecules of oxygen are required for the appearance of the oxidized mature enzyme product, which is formed even in the presence of excess unreacted pre-GAOXCu(I) complex. The presence of a second component is indicated by the anomalous behavior of the protein above about 0.5-0.8 eq copper loading (Figs. 2 and 3; Table  III) and the persistence of a fraction of pre-GAOX in the reaction mixture (Fig. 2, Table II), implying a limit to conversion to cofactor-containing product. EDTA prevents Cu(II) from binding to pre-GAOX (Fig. 9), yet the Cu(I)-dependent biogenesis reaction proceeds in the presence of the chelator, demonstrating that the Cu(I) in solution is efficiently bound by the preenzyme and supports rapid biogenesis of the Tyr-Cys cofactor. Absorption spectra reported in an earlier study of Cu(II)-dependent cofactor biogenesis imply a substoichiometric conversion to mature galactose oxidase product using propeptidecontaining precursor (29).
We have investigated a number of possible explanations for the substoichiometric conversion of pre-GAOX. One possibility, that a diffusible species generated during the in vitro biogenesis reaction interferes with complete conversion, appears to be ruled out by the observation that the same result is observed at widely different protein concentrations, and is not affected by including scavengers like superoxide dismutase or catalase in the reaction mixture. The sulfhydryl content measured for the precursor protein is less than the theoretical amount (Table I), suggesting that a fraction of the free cysteine residues may be blocked. However, ascorbate treatment (to reduce cysteine sulfenyl modifications, reported to be present in the crystal structure of the GAOX precursor (30)) did not alter the yield of cofactor. Other irreversible modifications of protein side chains may still interfere with the biogenesis reaction, or the incompetent fraction may represent an alternative conformation. If so, this alternate conformation must not equilibrate with the competent population, because repeating the Cu(I) ϩ O 2 treatment on the biogenesis product did not increase the yield of cross-link or the specific activity. Conformational changes and side chain modifications would have less of an effect during the prompt processing of pre-GAOX that is expected to occur in vivo, accounting for the higher level of conversion observed for the in vivo reaction product. By controlling the amount of copper added to the pregalactose oxidase protein, it has been possible to focus on the competent fraction to investigate the in vitro biogenesis reaction of pre-GAOX.
Gel densitometry has allowed us to correlate formation of a cross-linked biogenesis product with the absorption intensity of the oxidized mature enzyme produced in the reaction. The absorption coefficient of the oxidized product, normalized to copper, is nearly identical to that observed for oxidatively activated as-isolated GAOX (⑀ 445 ϭ 5100 M(protein) Ϫ1 cm Ϫ1 (7)). This is substantially less than the theoretical extinction coefficient for the oxidized product predicted by correcting for the apoprotein and unactivable fractions in the as-isolated enzyme (⑀ 445 ϭ 1 ϫ 10 4 M(protein) Ϫ1 cm Ϫ1 (2, 7)). The specific activity of the biogenesis product at lower copper loading (Table III) is also comparable with that found for the as-isolated enzyme. It is clear that in vitro biogenesis leads to a product distribution very similar to that found for the in vivo reaction, with a substantial fraction of unactivable (catalytically inactive) but cross-linked sites, implying that the complexity observed for the as-isolated protein is an intrinsic feature of the biogenesis reaction itself. Whether this component reflects an alternative cross-linking pattern (for example, Cys 228 -Trp 290 ) or irreversible modification of the Tyr-Cys site (for example, oxygenation to form a sulfoxyl or sulfonyl derivative (38 -40)) will require a more detailed structural characterization of these products.
Kinetic studies are providing important clues to the mechanism of cofactor biogenesis in galactose oxidase. The Cu(I)-dependent reaction is very fast, and rapid reaction techniques (i.e. stopped-flow spectrophotometry) are required to analyze the kinetics. Lacking signals from either the initial pre-GAOXCu(I) complex or any reaction intermediate (Fig. 10), cofactor formation has been monitored through the appearance of oxidized GAOX product in these experiments. This product is likely formed by oxidation of an initial, reduced mature enzyme complex (see below) in a reaction corresponding to the O 2 halfreaction of the normal turnover cycle (2). The reaction of fully reduced, mature GAOX (containing Cu(I) and no radical) with dioxygen is known to be extremely rapid (k ox ϭ 8 ϫ 10 6 M Ϫ1 s Ϫ1 ) (41,50), independent of pH (50), shows no significant SKIE (41), and is therefore well suited to serve as a non-interfering monitoring reaction. Although oxidized mature GAOX has been reported to be unstable in the presence of the fully reduced, Cu(I)-containing mature GAOX, undergoing rapid bimolecular redox comproportionation (k comp ϭ 4.4 ϫ 10 3 M Ϫ1 s Ϫ1 ) to form the catalytically inactive one-electron reduced Cu(II) complex in solution (51), we find that fully oxidized mature GAOX and pre-GAOXCu(I) are each stable in the presence of the other. Thus, the kinetics of cofactor biogenesis are not complicated by reaction of the AGAOX product with pre-GAOXCu(I) also present in the mixture.
Broad spectral range rapid scanning data collection spanning the entire visible region (325-800 nm) allows global fitting methods to be applied to the kinetic analysis (Fig. 10). This improves the quality of the data by factor-filtering noise (Fig.  10, left, middle) and also permits evaluation of kinetic models over the entire spectrum, a stringent test for kinetic models involving absorbing species. A single exponential phase cleanly accounts for all of the observed absorption intensity changes over the entire visible spectrum during conversion of pre-GAOXCu(I) to oxidized GAOX product at high pH (Fig. 10,  right). A single set of kinetic parameters provides an excellent fit to the experimental data for all wavelengths with uniform residuals reflecting the quality of the fit (Fig. 10). No additional signals that could be attributed to intermediates of the biogenesis reaction are detected. Specifically, there are no sharp features near 410 nm that would be associated with accumulation of a tyrosyl phenoxyl species during the reaction (52) (see below).
At lower pH, the reaction becomes slower (Fig. 12), and below pH 7 the product absorption spectrum changes, with shift in the absorption maximum to 435 nm (Fig. 11). The reaction at lower pH is also more complex, requiring multiple kinetic phases to adequately fit the data. This behavior may be under-stood as the result of a competition (at lower pH) between commitment to cofactor biogenesis on the one hand, and loss of hydroperoxyl (protonated superoxide) resulting in commitment to formation of the 406 nm pre-GAOXCu(II) complex on the other (Reactions 1 and 2). The spectral data recorded for the low pH reaction (Fig. 11, top) resolve onto a mixture of oxidized GAOX (minority product, rapid formation), 406 nm species (majority product, slow formation), and a weakly absorbing pre-GAOXCu(II) complex (prompt formation) which is an intermediate in conversion to the 406 nm species (Fig. 6). The pH dependence of the biogenesis reaction between pH 6.5 and 8.5 appears to reflect titration of a group in the protein ionizing with a pK a ϭ 7.3, higher reaction rates being associated with the deprotonated group (Fig. 12). Because the Cu(I) complex is formed prior to mixing with buffer in these reactions, it seems unlikely that this ionization is associated with the two active site histidine residues that serve as metal ligands in both oxidation states (53). Although other possibilities cannot be ruled out by these results, it is attractive to assign this ionization to Cys 228 . Cysteine residues in proteins typically exhibit a pK a near 8.3, but stabilization of the thiolate (e.g. by hydrogen bonding with Trp 290 indole or exposure to solvent) would be expected to perturb the side chain acidity. Hydrogen-bonded cysteines have been shown to titrate with pK a values as low as 6.3 in thioredoxin and glutaredoxin (54,55). At lower pH, protonation may destabilize the initial oxygenated complex, formulated as Cu(II)-O 2 . , leading to loss of HO 2 ⅐, blocking further reaction and producing pre-GAOXCu(II) rather than mature GAOX as the product. Our working hypothesis for the biogenesis mechanism (Scheme 2A) is based on involvement of mononuclear Cu(I) and dioxygen, an odd-electron system, implying that free radical intermediates occur in the formation of the Tyr-Cys cross-link. In the model shown here, Cu(I) is bound in a trigonal complex most likely formed by Tyr 272 , His 496 , and His 681 . Dioxygen then reacts with pre-GAOXCu(I) to generate an oxygenated complex, formulated as Cu(II)-superoxide free radical (Cu(II)-O 2 . ) (Scheme 2A, step 1). This oxygenated species may abstract a hydrogen atom from Cys 228 , forming a thiyl free radical, likely representing the rate-limiting step in the reaction sequence (step 2). The thiyl free radical attacks the C⑀ ring carbon of Tyr 272 (electrophilic radical addition to a double bond) breaking the conjugation of the aromatic ring system (step 3) (56). Formation of this initial covalent adduct involves a significant change in geometry around the C⑀ carbon of Tyr 272 that must be accommodated by the protein environment in order for the step to proceed efficiently. This initial adduct can decompose by loss of a proton and electron transfer (ET) reducing the Cu(II) to Cu(I), forming a fully reduced, cofactor-containing active site (step 4). Subsequent reaction of this reduced complex with a second molecule of O 2 would lead to appearance of fully oxi-  dized mature GAOX, as observed (step 5). This is a two-electron biogenesis reaction, requiring two reductants (identified here as Cu(I) and cysteine thiol). Replacing Cu(I) by Cu(II) in this mechanism (for example, in the Cu(II)-dependent biogenesis experiments described above) would require an additional (exogenous) 1e Ϫ equivalent to reduce dioxygen to the stable H 2 O 2 product following one-electron oxidation of a protein side chain.
This model accounts for the pH sensitivity of the biogenesis reaction in terms of ionization of Cys 228 at high pH (Scheme 3). The deprotonated thiolate group has a lower reduction potential (E 0 (RS⅐/RS Ϫ ) ϭ 0.8 V versus normal hydrogen electrode, compared with E 0 (RSH⅐/RSH) ϭ 1.37 V) (57) and would react more readily (via electron transfer, ET) to generate a thiyl free radical and a coordinated peroxide. If thiyl radical formation is rate-limiting, as proposed above, increasing the rate of formation will lead to an increase in the observed biogenesis rate, as is experimentally observed. Exchange of the cysteine sulfhydryl group in D 2 O would also account for the modest SKIE measured for this reaction in the lower pH range, if thiol oxidation involves S-H bond cleavage. The pH sensitivities of rate (Fig. 12) and SKIE (Table IV) are the same, implying that both arise from a unique ionization process in the active site. The SKIE vanishes (the ratio k H 2 O /k D 2 O approaches unity) as the biogenesis reaction rate reaches a maximum above pH 8, consistent with ionization of a mechanistically important proton. In the lowest pH range (below pH 7) protonation of the an initial oxygenated complex appears to result in an abortive pathway forming pre-GAOXCu(II) without cross-linking, perhaps by displacement of a protonated superoxide (hydroperoxyl radical) from this complex.
An alternative description of the initially formed oxygenated mononuclear copper complex as a Cu(III)-peroxo species (58,59) also accounts for its reactivity but suggests that direct inner sphere oxidation of the Tyr 272 ligand might be more favorable. This alternate Cu(I)-dependent pathway (Scheme 2B) would then generate a tyrosyl phenoxyl free radical (step 2) capable of undergoing nucleophilic substitution by the neighboring Cys 228 thiolate (step 3). Both of the Cu(I) pathways proposed here involve proton transfers from the active site Cys 228 , so the observation of pH and isotope effects does not clearly distinguish between thiyl and phenoxyl reactions. However, phenoxyl free radicals (like the tyrosyl free radical in Escherichia coli ribonucleotide reductase) characteristically exhibit sharp, intense absorption features near 410 nm (⑀ 410 ϭ 3200 M Ϫ1 cm Ϫ1 ) (52). Because no such features are detected during cofactor biogenesis (Fig. 10), the phenoxyl pathway is considered to be less likely. A previous proposal for Cu(II)-dependent cofactor biogenesis, involving resonance stabilization of a tyrosine free radical within a pre-enzyme Cu(II) complex (29,30), predicts the appearance of spectroscopic features associated with the tyrosyl phenoxyl species in the anaerobic Cu(II) complex. As shown above, all of the Cu(II) spins can be accounted for in the EPR spectra for these complexes (Fig. 7) with no evidence of any additional signals associated with free radical species (with a sensitivity of ϳ0.1%) at either cryogenic or ambient temperatures. There is also no evidence for stabilization of a phenoxyl species in the optical absorption spectra of the anaerobic pre-GAOXCu(II) complex (Fig. 6). Instead, the 406 nm absorbing species may be assigned to an isomerized active site complex having Cys 228 directly coordinated to copper which is not an intermediate in the biogenesis reaction. CONCLUSIONS Biogenesis of the cross-linked catalytic Tyr-Cys cofactor in galactose oxidase is an intrinsic feature of the active site chemistry. The Cu(I)-dependent reaction, which we believe mimics the in vivo maturation process, is very rapid and does not require any propeptide leader sequence to be present. Cofactor biogenesis utilizes some features of the active site (like the copper coordination environment) that are retained in the mature protein and contribute to the catalytic reaction. However, other features (e.g. the special reactivity of the unmodified Cys 228 thiol and the Tyr 272 side chains) are unique to the pre-enzyme complex. The self-processing in vitro formation of the Tyr-Cys redox cofactor in galactose oxidase shows that the path to a functional enzyme, involving formation of reactive intermediates and structures with special geometric constraints, must be templated within the structure of the active site. We have proposed a mechanism for cofactor biogenesis that accounts for the significant experimental characteristics (copper oxidation state dependence, oxygen stoichiometry, pH sensitivity, and solvent isotope sensitivity) that we find for this reaction. Further work will be required to investigate the nature of reactive intermediates formed during galactose oxidase cofactor biogenesis in more detail.