Evidence That Dioxygen and Substrate Activation Are Tightly Coupled in Dopamine β-Monooxygenase

Oxygen activation occurs at a wide variety of enzyme active sites. Mechanisms previously proposed for the copper monooxygenase, dopamine β-monooxygenase (DβM), involve the accumulation of an activated oxygen intermediate with the properties of a copper-peroxo or copper-oxo species before substrate activation. These are reminiscent of the mechanism of cytochrome P-450, where a heme iron stabilizes the activated O2 species. Herein, we report two experimental probes of the activated oxygen species in DβM. First, we have synthesized the substrate analog, β,β-difluorophenethylamine, and examined its capacity to induce reoxidation of the prereduced copper sites of DβM upon mixing with O2 under rapid freeze-quench conditions. This experiment fails to give rise to an EPR-detectable copper species, in contrast to a substrate with a C–H active bond. This indicates either that the reoxidation of the enzyme-bound copper sites in the presence of O2 is tightly linked to C-H activation or that a diamagnetic species \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{Cu(II)}-\mathrm{O}_{2}^{{\cdot}}\) \end{document} has been formed. In the context of the open and fully solvent-accessible active site for the homologous peptidylglycine-α-hydroxylating monooxygenase and by analogy to cytochrome P-450, the accumulation of a reduced and activated oxygen species in DβM before C-H cleavage would be expected to give some uncoupling of oxygen and substrate consumption. We have, therefore, examined the degree to which O2 and substrate consumption are coupled in DβM using both end point and initial rate experimental protocols. With substrates that differ by more than three orders of magnitude in rate, we fail to detect any uncoupling of O2 uptake from product formation. We conclude that there is no accumulation of an activated form of O2 before C-H abstraction in the DβM and peptidylglycine-α-hydroxylating monooxygenase class of copper monooxygenases, presenting a mechanism in which a diamagnetic Cu(II)-superoxo complex, formed initially at very low levels, abstracts a hydrogen atom from substrate to generate Cu(II)-hydroperoxo and substrate-free radical as intermediates. Subsequent participation of the second copper site per subunit completes the reaction cycle, generating hydroxylated product and water.

Dopamine ␤-monooxygenase (D␤M) 1 along with peptidylglycine-␣-hydroxylating monooxygenase (PHM) comprise a unique class of enzymes that contain only copper as a cofactor and catalyze the cleavage of O 2 to form hydroxylated product and water. D␤M is of central importance in the catecholamine biosynthetic pathway, catalyzing the conversion of dopamine to norepinephrine (Scheme 1, top), where both substrate and product serve as neurotransmitters within the central nervous system (1). Primarily localized within the secretory granules of adrenal chromaffin cells and neurons, D␤M is a large, tetrameric glycoprotein (75 kDa per monomer) consisting of two disulfide-linked dimers.
Although no crystal structure has been reported, extensive structural data exist for D␤M. Extended X-ray absorption fine structure was used to characterize the ligand environment of the two copper atoms per subunit in both oxidized and reduced forms of D␤M; in the absence of any evidence for back scattering between the metal sites, the distance between the two coppers per subunit was concluded to exceed 4 Å (2, 3). EPR spectroscopy also failed to detect any spin coupling between metal sites in oxidized, resting enzyme (4) and in a catalytically generated product complex (5). These findings provided early evidence against a reactive binuclear center and, instead, implicated separate functions for the two copper centers. The Cu A (Cu H in PHM), liganded by three histidines (D␤M: His-255, His-256, His-326; PHM: His-107, His-108, His-172) and water, has historically been assigned to the electron transfer site, and Cu B (Cu M in PHM), liganded by two histidines (D␤M: His-405, His-407; PHM: His-242, His-244) and water together with a long bond to methionine (D␤M: Met-480; PHM: Met-314), has historically been assigned to the substrate binding and hydroxylation site.
Despite their large difference in size, D␤M and PHM share a 28% sequence identity extending through a common catalytic domain of ϳ270 residues, which includes the conserved copper ligands (8). In addition both enzymes require two coppers per subunit for full activity (9 -11) and are believed to utilize ascorbate as the in vivo two-electron donor (12).
The crystal structure of PHM, solved for both oxidized (13) and reduced (14) enzyme, confirmed many of the earlier spectroscopic data. Important features determined from the crystal structure are (i) a two-domain structure in which each domain binds a single copper atom, (ii) a distance of ϳ10.6 Å (oxidized PHMcc with bound peptide substrate) between the two copper sites, (iii) the absence of closure of the copper binding domains in either enzyme form studied, and (iv) the identification of a water-filled cavity that is at the solvent interface and "links" the two copper binding domains. Extensive debate has taken place in the recent literature regarding the pathway for electron transfer between copper sites and the nature of O 2 activation in D␤M and PHM (14,15).
Comparison of the kinetic parameters for D␤M and PHM with substrates of comparable reactivity indicates the same intrinsic hydrogen/deuterium isotope effect (ϳ11) for the C-H activation step (16). Additionally, similar O-18 isotope effects for these two enzymes that decrease with substrate deuteration imply a chemical mechanism for substrate oxidation that is likely to be identical (17). Studies of the kinetic mechanism indicate that both D␤M (in the presence of the dianion activator fumarate) and PHM proceed in a preferred ordered mechanism with substrate binding to enzyme before O 2 (16,18). Thus, all available data imply that D␤M and PHM can be regarded interchangeably with respect to mechanism and active site structure.
In early studies of D␤M with either substrates or substrate analogs it was concluded that functionalization of substrate involved hydrogen atom abstraction to yield a free radical intermediate (19,20). Identification of the oxygen species catalyzing hydrogen atom abstraction from substrate has proven far more elusive. The observation of pH-dependent isotope effects for D␤M provided evidence for the involvement of a single proton in the chemical conversion process, leading to the proposal of a copper hydroperoxide as the reactive oxygen intermediate (III in Scheme 2) (18). However, a detailed analysis of the effects of substrate structure and deuteration on O-18 isotope effects (21) was found to be inconsistent with the earlier proposed Cu(II)-OOH 2 and suggested a reductive cleavage of this intermediate to generate copper-oxo as the hydroxylating agent (IV in Scheme 2). This interpretation assumed classical behavior of hydrogen during transfer from substrate to oxygen, which has now been shown in the case of PHM to be dominated by hydrogen tunneling (22). Additionally, recent site-specific mutagenesis studies with PHM unambiguously eliminates a role for the most plausible active site candidate (PHM: Y318; D␤M: Y484) in reductive activation of Cu(II)-OOH (17). The lack of extensive pH studies for the PHM reaction together with a failure to identify a proton donor from the crystal structure has led to the proposal of mechanism II in Scheme 2 (14). A common feature of all previously proposed mechanisms for PHM and D␤M (cf. II, III, and IV in Scheme 2) is that oxidation of both copper centers occurs before substrate activation and leads to the accumulation of a partially reduced form of dioxygen. Blackburn et al. (23) observe significant changes in the copper coordination structure during oxidation, which would appear to preclude a rapid inter-copper electron transfer. They proposed an alternative possibility based on the finding that CO binds to the electron transfer copper in the presence of peptide substrate; this involves a role for superoxide as the electron carrier between Cu A and Cu B (15).
Many of the paradigms for the copper monooxygenase mechanism have come from the very detailed chemical analyses of cytochrome P-450, which has been shown to release hydrogen peroxide as product either in the absence of a substrate or in the presence of a poor substrate (24). Additionally, activated oxygen at the active site of cytochrome P-450 has been shown to be capable of further reduction to water (25). These side reactions of cytochrome P-450 lead to an uncoupling of substrate hydroxylation from uptake of O 2 and occur despite the sequestration of the active site from bulk solvent (26). For enzymes like D␤M and PHM, whose active sites are fully exposed to solvent, it is expected that extensive uncoupling of substrate hydroxylation and O 2 uptake may occur. In the case of D␤M, the size of the primary isotope effects on k cat and k cat /K m for certain phenethylamine substrates indicates that C-H cleavage is rate-determining, with the implication of a steady state accumulation of an enzyme complex that contains the activated 2 Note that the charge on oxygen species complexed to copper is omitted when a solid line is drawn between metal and oxygen, e.g. representation of Cu(II)--Ϫ OOH (Scheme 2) as Cu(II)-OOH in text. SCHEME 2. Four possible activated oxygen species capable of hydrogen atom abstraction in the mononuclear copper containing monooxygenases D␤M and PHM. The first (I) involves oxidation of only a single copper before substrate activation. The second (II), proposed by Prigge et al (14) for PHM involves the reversible transfer of an electron from the second copper site, Cu A (I) to dioxygen. The third (III) illustrates the mechanism for D␤M in which a proton is involved in the activation of dioxygen to form a copper hydroperoxide; note that both electrons must be transferred (reversibly, see "Discussion") to dioxygen before chemistry on the substrate. The fourth (IV) proposed for D␤M involves reductive activation of the copper hydroperoxide by an active site tyrosine to generate a copper-oxo species. SCHEME 1. Reactions catalyzed by D␤M and PHM. PAL, peptidylglycine-␣-amidating lyase. oxygen intermediate. Thus, if the copper monooxygenase mechanism occurs with a build-up of any type of activated oxygen intermediate, we would expect to observe some degree of uncoupling between oxygen reduction and substrate consumption that would increase significantly as the chemical reactivity of the substrate diminishes.
The stoichiometry of the D␤M (27) and PHM (28) reactions with their respective substrates, dopamine and D-Tyr-Val-Gly, was previously shown to be 1 eq of dioxygen consumed for each equivalent of substrate and reductant. We have now extended these studies with D␤M using substrates of varying reactivity by changing the para substituent of the phenyl ring. Additionally, single turnover experiments have been performed with a ␤,␤-difluoro analog of phenethylamine that cannot be functionalized. From these results it is possible to rule out the presence of any appreciable level of a reduced oxygen species before substrate functionalization. The extremely tight coupling of O 2 and C-H activation in D␤M points toward a new mechanistic pathway for D␤M and PHM while indicating the inappropriateness of inferences derived from the family of heme-iron-dependent monooxygenases.

EXPERIMENTAL PROCEDURES
Materials and General Methods-Soluble D␤M was isolated as previously described (5) from bovine adrenal glands. The protein concentration was estimated from the absorbance at 280 nm (⑀ 280 ϭ 1.24 ml mg Ϫ1 cm Ϫ1 ). A monomer mass of 75 kDa was used in calculations of enzyme concentration. Trace metal analysis of enzyme-bound copper was performed on a PerkinElmer 3000DV Inductively Coupled Plasma-Atomic Emission Spectrophotometer using commercially available metal standard solutions. Catalase (65,000 units/mg) was from Roche Applied Science. All other materials were of reagent grade. The compounds 4-hydroxyphenethylamine (tyramine), 3,4-dihydroxyphenethylamine (dopamine), 4-hydroxyphenethanolamine (octopamine), and phenethylamine were purchased from Sigma and used as the hydrochloride salts. [1-14 C] tyramine hydrochloride, with a specific activity of 55 mCi/mmol, was purchased from American Radiolabeled Chemicals, Inc. 4-(Trifluoromethyl)phenethylamine hydrochloride was synthesized from the commercially available nitrile (purchased from Aldrich) as previously described (19). The elemental analysis gave 47.91% C, 4.91% H, 6.21% N (calculated: 47.98% C, 5.05% H, 6.05% N).
Enzyme Kinetic Assays-Steady-state rates of oxygen consumption were measured with a YSI model 5300 biological oxygen electrode. The assay conditions, similar to those described previously (29), were as follows: 10 mM disodium fumarate, 10 mM sodium ascorbate, 100 g/ml catalase, 2 M CuSO 4 , 50 mM potassium P i (pH 6.0), T ϭ 35°C, 0.218 mM O 2 (air saturation). The concentration of the substrates (tyramine, dopamine, and phenethylamine) ranged from 0.05 to 10 mM, whereas the ionic strength was maintained at a constant value of 0.15 M with the addition of NaCl. The concentration of 4-(trifluoromethyl)phenethylamine ranged from 6 to 100 mM, whereas the ionic strength was maintained at 0.2 M with NaCl. The apparent k cat and k cat /K m values were obtained by fitting the data to the Michaelis-Menten equation using the program Kaleidagraph.
Enzyme End Point Assays-These were initiated on the oxygen electrode by the addition of various amounts of substrate to a reaction mixture (1 ml) containing 50 mM potassium P i (pH 6.0), 10 mM disodium fumarate, 10 mM sodium ascorbate, 100 g/ml catalase, 2 M CuSO 4 , and various amounts of D␤M. The assays were maintained at a constant temperature of 35°C using a Neslab circulating water bath. Sufficient amounts of enzyme were added to consume all of the added substrate within 5 min. The concentration of dissolved oxygen under the conditions used was determined to be 218 M by the protocatechuic acid/protocatechuate dioxygenase assay (30). Briefly, the full amplitude of oxygen uptake was recorded after the addition of protocatechuic acid (22.4 M) to an assay mixture containing 50 mM potassium P i (pH 6.0), 94 mM NaCl (ionic strength of 0.15 M), and excess of protocatechuate dioxygenase, purchased from Sigma. All end point assays of dopamine, tyramine, and phenethylamine were maintained at an ionic strength of 0.15 M with NaCl. The concentrations of dopamine and tyramine were determined using their reported extinction coefficients: dopamine Otherwise, all substrate concentrations were determined by weight.
Enzyme Initial Rates Assays-These were performed with the sub-strates tyramine and 4-(trifluoromethyl)phenethylamine. Assay conditions for tyramine were identical to those described above and were performed on the oxygen electrode. Two tyramine concentrations were used, 5 and 0.2 mM with 0.1 Ci of 14 C-labeled tyramine added to trace the extent of reaction. Assays of 4-(trifluoromethyl)phenethylamine contained 50 mM substrate (ionic strength adjusted to 0.2 M with NaCl) and were performed either in the presence or absence of catalase. All assays were quenched with 0.091 M HClO 4 (100 l of a 1 M solution) after 6.5-8.7 min and were immediately cooled to Ϫ80°C and stored for subsequent analysis by HPLC. Total dioxygen consumption was corrected for the ascorbate background rate. All HPLC separations were performed on an Alltech Adsorbosphere reversed phase C-18 column.
Octopamine, detected at 224 nm, was separated from tyramine with a mobile phase of 5 mM acetic acid (pH 5.8), 600 M heptane sulfonic acid, and 15% methanol at 1 ml/min. Separation of the ␤-hydroxylated product from 4-(trifluoromethyl)phenethylamine was accomplished with an isocratic flow of 17% acetonitrile, 83% water, and 0.1% trifluoroacetic acid at 1 ml/min and was detected at 210 nm. A standard curve based on integrated peak area with mock-quenched reaction mixtures was linear through the range of interest. Blanks, in which no enzyme was added, were used to determine the amount of contaminant eluting with the same retention time as product by HPLC at 0% conversion and were subtracted from each assay. The integrated peak area of this peak corresponded to 5.2-6.0 nmol of product (based on the standard curve constructed from mock quenched assays) in the experiments with catalase present. In those without catalase, the peak integrated to 9.5 nmol of product. The contaminant in the product peak is only 0.01-0.02% of the total amount of 4-(trifluoromethyl)phenethylamine (50 mM) present in the assays.
Synthesis of 4-(Trifluoromethyl)phenethanolamine Hydrochloride-4-(Trifluoromethyl)benzaldehyde was reacted with sodium metabisulfite and sodium cyanide as described previously (31). A saturated solution of sodium metabisulfite (896 mg, 4.7 mmol) was added to 4-(trifluoromethyl)benzaldehyde (1g, 5.7 mmol), and the solution was stirred on ice. A layer of ether (5 ml) was added on top, and an ice-cold saturated solution of sodium cyanide (782 mg, 15.9 mmol) was added. The aqueous layer was diluted with a small amount of water and extracted with ether, and the combined ether solutions were washed with bisulfite solution and then with water. The resulting cyanohydrin was isolated by silica-gel flash chromatography (20% ethyl acetate, 80% hexane) and reduced with LiAlH 4 as described in Poos et al. (32) without further purification. The amine was converted to the hydrochloride salt with HCl in ether. Elemental analysis of the hydrochloride salt was found to contain 44.58% C, 4.29% H, 5.80% N (calculated: 44.74% C, 4.59% H, 5.80% N).
Synthesis of ␤,␤-Difluorophenethylamine Hydrochloride-Styrene was first converted to the azirine (2-phenyl-1-azirine) as described previously (33). To a solution of styrene (52.19 g, 57.5 ml, 0.50 mol) in 400 ml of CCl 4 in an ice bath was added a solution of bromine (80 g, 25.8 ml, 0.50 mol) in 100 ml of CCl 4 through an addition funnel over 30 min. The reaction mixture was stirred at room temperature for 1 h. The solvent was evaporated under vacuum to yield 1,2-dibromo-2-phenylbenzene, a white solid that was immediately dissolved in 750 ml of dry Me 2 SO and placed in a 1-liter 3-necked round-bottomed flask. While on an ice bath NaN 3 was added (49 g, 0.75 mol) with stirring and under nitrogen. After 12 h the reaction mixture was cooled on an ice bath, and 20 g of NaOH (0.5 mol) in 20 ml of H 2 O was added. After 1 h of stirring only one spot was visible by TLC (n-hexane). The mixture was diluted to a volume of 2 liters with 2% NaHCO 3 in H 2 O and extracted with methylene chloride 4 ϫ 200 ml, and the combined extracts were washed with H 2 O, filtered, and concentrated in vacuo to yield a dark brown liquid. This was diluted with petroleum ether (200 ml) and applied to an alumina (basic, 200 g) column with petroleum ether as the eluent. This gave 72.8 g of a light yellow oil, which was dissolved in 1.25 liter of toluene. 1.2 liters of the ␣-azidostyrene in toluene was placed in a round-bottom flask. The mixture was heated under reflux, and progress of the reaction was monitored by TLC (9:1, n-hexanes:ethyl acetate). After 1 h and 20 min the reaction was stopped. The product was distilled to yield 37.02 g of 2-phenyl-1-azirine, a colorless oil.
Next, the azirine was converted to the ␤,␤-difluoroamine by reaction with HF in pyridine (34). 12 g (0.1 mol) of 2-phenyl-1-azirine was added to 120 ml of 70% HF in pyridine at Ϫ35 to Ϫ30°C over 15 min and stirred at Ϫ25 to 0°C for 30 min and for 80 min at room temperature. The reaction was quenched with the addition of 60 ml of water at Ϫ78°C. Finally the amine was converted to the hydrochloride salt by heating in an HCl solution under reflux for 0.5 h. The solvent was evaporated and solid-washed with acetone. The white solid was recrystallized 3 times from a mixture of ethanol and ether to give 1.26 g (7% yield) of the hydrochloride salt. The final product was found to have a melting point of 181-183°C, and the elemental analysis gave 49.50% C, 5.15% H, 7.35% N (calculated: 49.63% C, 5.21% H, 7.23% N).
Freeze-quench Experiments with D␤M-A detailed description of the set-up as well as the experimental procedures can be found in Brenner et al. (5). A dead time of 4 ms was estimated as described previously. The samples used for EPR analysis were quenched at various time points ranging from 9 to 470 ms. The manner in which reaction solutions were mixed for a sample with tyramine was the following; initially, equal volumes of the enzyme solution consisting of 60 M D␤M monomer, 100 mM potassium P i (pH 6.19), 30

RESULTS
Kinetic Studies of ␤,␤-Difluorophenethylamine-The coupling of O 2 and substrate activation can be tested by the use of substrate analogs that are unable to undergo C-H activation. Fluorinated substrates appear to be well suited for this approach, and the analog, ␤,␤-difluorophenethylamine was synthesized as described under "Experimental Procedures." We find that the analog is a weak competitive inhibitor of substrate (K i ϭ 26 Ϯ 3 mM) for D␤M (data not shown). Under steady state conditions, up to 75 mM difluoro analog and 10 M D␤M failed to support any uptake of O 2 using a Clark oxygen electrode. Steady state assays indicated that there was no consumption of oxygen above the ascorbate background in the presence of DFPA. As a control it was shown that preincubation of enzyme with substrate plus DFPA gave a similar stability profile to incubation with substrate alone, ruling out any enzyme inactivation by DFPA.
We, therefore, turned to pre-steady state conditions to see if a stoichiometric amount of pre-reduced enzyme could be reoxidized in the presence of the difluoro analog. These experiments contained 40 mM of the analog and 15 M D␤M, leading to an estimated 60% occupancy of the enzyme sites. This concentration was a balance between achieving significant occupancy of the enzyme active sites and having sufficient analog to measure reaction at many time points. In the course of these experiments, enzyme is first pre-reduced with ascorbate under anaerobic conditions and then mixed with either tyramine or DFPA. As reported earlier (5), rapid mixing of anaerobically reduced enzyme with tyramine and dioxygen led to full reoxidation of both coppers within 500 ms of mixing (Fig. 1). By contrast, mixing of the anaerobically reduced Cu(I) form of enzyme with DFPA and O 2 failed to give rise to a detectable Cu(II) EPR signal within 500 ms (Fig. 1). The time course for DFPA appears similar to that in the absence of substrate or analog, consistent with earlier reports that substrate is required for reoxidation of reduced copper either under steady state (35) or pre-steady state conditions (5). After 20 s 43% of the copper was reoxidized in the DFPA-incubated samples compared with 52% of nonspecific oxidation for enzyme-bound copper in the absence of substrate or substrate analog; these are considered within experimental error. Even after 20 min only 53% of the total copper had been reoxidized to Cu(II) in the air-incubated sample.
The failure of DFPA to support copper reoxidation in a single turnover experiment could be due to two explanations. The first is that copper reoxidation is tightly coupled to C-H activation and that no activated oxygen species is formed at the level of detection of the EPR (limits of detection in these experiments, ϳ10% of total enzyme). The second possibility is that an oxidized copper, spin-coupled species had formed. One candidate is a Cu(II)-OOH substrate intermediate that is anti-ferromagnetically coupled to the second copper at the Cu A site. This seems highly unlikely in view of the absence of any spin coupling in catalytically generated enzyme-product complexes with D␤M where both coppers are in the ϩ2 oxidation state (36). As an alternative, we propose that the addition of O 2 leads to oxidation of only a single copper site (Cu B ) to produce a spin-coupled cupric-superoxide complex, Cu B (II)-O 2 ⅐ . Though not considered in previous investigations of PHM and D␤M mechanisms, this offers an alternate view of O 2 activation that is described under "Discussion.'' Tight Coupling of Substrate and O 2 Activation-Two experimental approaches have been used in this study to quantify the stoichiometry of substrate hydroxylation to O 2 uptake. Using fast substrates, end-point assays were performed to compare the stoichiometry of dioxygen consumption (by Clark O 2 electrode) to that of product formation (by HPLC). In the case of slower substrates, where achievement of reliable end-points becomes extremely difficult, we compared initial rates of O 2 consumption and product formation. Experiments were reproducible over multiple experiments, as shown in Table I, for the slowest substrate, 4-(trifluoromethyl)phenethylamine.
The data in Table I show a background rate of O 2 consumption before the addition of enzyme that was assumed to be constant over the time course of the experiment. This is largely due to the presence of Cu(II) ions and ascorbate, which are necessary to optimize conditions for D␤M turnover. The rate of air oxidation of ascorbate in aqueous solution is very dependent on pH and catalytic metals and produces dehydroascorbate with concomitant production of hydrogen peroxide, superoxide, and hydroxyl radicals (37). These reduced oxygen species, more specifically H 2 O 2 , are known to inactivate D␤M (38) and PHM (39). For this reason catalase was included in the majority of reactions to disproportionate hydrogen peroxide to water and dioxygen and prevent inactivation of D␤M over the time course of the experiments. Although the presence of catalase would underestimate D␤M uncoupling by a factor of two (for H 2 O 2 production) or by a factor of four (for superoxide leakage), the sensitivity and reproducibility of our experiments is very high and indicates a molar ratio of O 2 consumption to product formation of 0.99 Ϯ 0.04. To demonstrate that catalase did not alter the measured stoichiometries with p-(trifluoromethyl)phenethylamine, a limited number of experiments in the absence of catalase were performed. To avoid high backgrounds, no exogenous copper was added to the assays. Samples of D␤M were prepared with 1.9 eq of pre-bound copper, and the ratio of O 2 uptake to product formation was determined. Although the background correction for O 2 consumption was greater, increasing the error in the molar ratio of O 2 to product, the final stoichiometry of 1.01 Ϯ 0.10 was unchanged (Table I).
As anticipated, the relatively fast substrates, tyramine, dopamine, and phenethylamine, indicate an average ratio of 1.02 Ϯ 0.05 for O 2 consumption to product formation (Table II). Importantly, comparison of the end-point assay to the initial rate assay led to a similar value of 0.99 Ϯ 0.04 with tyramine. This was considered a key control for the very slow 4-(trifluoromethyl)phenethylamine substrate, which could only be quantified using the initial rate assay. The lack of uncoupling among the three fastest substrates may not be surprising in light of their relatively large apparent k cat /K m values for substrate under standard assay conditions. Use of the 4-(trifluoromethyl)-phenethylamine substrate markedly decreases the reactivity of D␤M by more than three orders of magnitude in the apparent k cat /K m . Despite the enormous decrease in rate, the ratio for O 2 consumption to product formation remains at unity. This result is especially remarkable in light of an earlier finding of a kinetic isotope effect of ϳ18 using ␤,␤-dideuterated 4-(trifluoromethyl)phenethylamine as the substrate (19), indicating that C-H cleavage is fully rate-determining with this substrate. For a mechanism in which reductive activation of dioxygen precedes C-H cleavage (III in Scheme 2) and occurs at a similar rate for all phenethylamine substrates, the extended lifetime of an oxygen intermediate at the solvent interface with a very slow substrate would be expected to produce some "leakage" of either peroxide (III in Scheme 2) or superoxide (I in Scheme 2) into the solvent. In solution, mononuclear copperreduced oxygen species are generally too unstable to detect and study at room temperature with, for example, rate constants of 10 7 -10 8 s Ϫ1 for dissociation of superoxide anion from a Cu(II)-O 2 ⅐ complex at 25°C (40). We note that for the analogous PHM, Blackburn and co-worker (15) report a slight change in the stretching frequency for copper-bound CO in the presence of peptides of varying structures. In the study of D␤M, where changes in substrate structure occur solely at the ring substituent, little or no impact on O 2 affinity is anticipated. We conclude that our observation of complete coupling with 4-(trifluoromethyl)-phenethylamine is incompatible with the accumulation of a reduced dioxygen intermediate preceding substrate activation. An alternate, heretofore unconsidered, explanation is that C-H bond cleavage and O 2 activation are close to fully coupled. c Measured using end point assay. d Measured using initial rates assay and is the average of all experiments in Table I.  (Tables I  and II), and the second is the failure to observe any Cu(II) formation from Cu(I) in the presence of O 2 with an inert substrate analog (Fig. 1). The aggregate of these experiments point toward an extremely tight coupling between O 2 activation and C-H cleavage from substrate. What mechanisms can be proposed that take into account such tight coupling while also accommodating the extensive mechanistic literature on PHM and D␤M? Any proposed mechanism must involve a net hydrogen atom abstraction from substrate to O 2 in the initial step of substrate oxidation coupled to some type of activation of the ground state, triplet O 2 . The detection of sizeable O-18 isotope effects on k cat /K m (O 2 ), which are perturbed by substrate deuteration (21), means that any O 2 chemistry that occurs before loss of hydrogen from substrate must be a fully reversible process (17). The actual transfer of hydrogen occurs by a tunneling mechanism (22), indicating that the origin of the O-18 isotope effect lies with any preequilibrium reduction of O 2 together with heavy atom motions within O 2 that are necessary for effective tunneling of hydrogen from reactant to the O 2 acceptor. Finally, there are no active site residues other than copper and its ligands, which appear to play a direct role in bond cleavage (13,14,17).
The above properties rule out a copper-oxo species from further consideration (cf. Ref. 17). Similarly, given the twoelectron, one-proton transfer that is necessary for formation of a Cu(II)-OOH intermediate from O 2 (Scheme 2, III), it seems very unlikely that this species would be produced at an undetectable level via a reversible, proton-coupled long range electron transfer process. This leaves open the possibility of a copper-peroxo (Scheme 2, II) or copper-superoxo species (Scheme 2, I). The fundamental difference between these two species lies with the number of electrons that are transferred to O 2 before C-H activation. A copper-peroxo mechanism (Scheme 2, II) involves a twoelectron reduced form of O 2 , requiring a long range electron transfer between copper centers before C-H activation. Because k cat /K m (O 2 ) is at least partially limited by C-H cleavage, this means that electron transfer across a solvent interface must occur, at a minimum, at a rate comparable with C-H cleavage (680 s Ϫ1 for dopamine plus fumarate (19); 1200 s Ϫ1 for dopamine minus fumarate (18), and 810 s Ϫ1 for hippuric acid with PHM (16)), and as noted above, it must be fully reversible. The suggestion that the long range electron transfer is mediated by O 2 itself (as superoxide anion) appears to be ruled out by the observation of complete coupling of O 2 uptake to substrate hydroxylation under a variety of conditions (Table II). Thus, the requirement for a rapid and reversible electron transfer between Cu A and Cu B may also be too great to make copper- peroxo a viable intermediate. A copper-peroxo mechanism would accommodate the size of the measured O-18 isotope effects; a pre-equilibrium formation of Cu(II)-O 2Ϫ from O 2 predicts an O-18 isotope effect of ϳ3%, close to the experimentally measured value of 2%. There would be little tendency for the peroxide to dissociate from the metal ion in the absence of an active site proton donor (although bulk water at the solventexposed active site might function in this manner). In any case, a copper-peroxo would only be formed at very low levels, given the failure to detect paramagnetic Cu(II) species after incubation of D␤M with O 2 and DFPA.
As a second option, a role for copper-superoxo in C-H activation has a great deal of weight. The superoxo species is more electrophilic than peroxo and would be expected to be a reasonable acceptor of the hydrogen atom from substrate. The presence of a diamagnetic, spin-coupled Cu(II)-O 2 ⅐ (41) would be consistent with the failure of freeze-quench EPR experiments ( Fig. 1) to detect any paramagnetic species in the presence of DFPA. Perhaps most appealing is the fact that the second electron transfer from the copper site would not take place until after formation of the substrate-derived free radical and a Cu(II)-OOH. This is expected to be a highly reactive state, capable of driving transfer of an electron from the reduced copper site to the terminal oxygen of the Cu(II)-OOH. This can be modeled to occur through a distance of only 7.5 Å and via a network of three water molecules (Scheme 3). The mechanism of Scheme 2, I and Scheme 3 places the long range electron transfer between copper sites after the initial substrate activation, i.e. in the k cat parameter. Although structure reactivity correlations have implicated a rate-limiting dissociation of copper-bound hydroxylated product in k cat (19), electron transfer may contribute to this parameter as well. This requires that the electron transfer occurs faster than the rate for k cat (12.7 s Ϫ1 for dopamine plus fumarate (19), and 39.1 s Ϫ1 for hippuric acid with PHM (16)), a far less stringent condition than its contribution to k cat /K m (O 2 ). Electron tunneling rates through water over a distance of 8 Å have been predicted to occur at a rate of 10 9 s Ϫ1 (42). However, as pointed out by Blackburn et al. (23) substantial ligand reorganization of the copper sites would decrease this rate, although it should remain significantly faster than the rate for k cat of 12.7 s Ϫ1 (dopamine plus fumarate) (19).
As with the other postulated mechanisms, the copper-superoxo mechanism must accommodate the finding of complete coupling of O 2 uptake to substrate hydroxylation and the size of the O-18 isotope effects. With regard to the former, one possibility is a steady state level of copper-superoxo that is simply so low that any loss of superoxide anion lies in the noise of the experiments. In general, superoxide will act as a reductant toward Cu(II), and the species Cu(II)-O 2 ⅐ may be better written as an equilibrium between Cu(I)-O 2 and Cu(II)-O 2 ⅐ . The mode of complexation may be side-on, by analogy to mononuclear copper-dioxygen species that have been characterized by x-ray crystallography (41) and vibrational spectroscopy (43). The enzyme-bound coppers are reported to have a redox potential of 310 -380 mV (44,45), elevated compared with the potential of free copper (160 mV); this will favor an oxygenated species with considerable Cu(I) character. With regard to the size of the experimental kinetic O-18 isotope effects, the observed values of 2% exceed the equilibrium value of only 1%, expected for formation of copper-superoxo. However, this equilibrium limit for the O-18 isotope effect does not take into account any reorganization of the superoxo species that must occur before transfer of hydrogen by tunneling. It is possible that the bond between the copper and superoxide anion undergoes transient weakening to achieve the requisite degeneracy between reactant and product that allows the tunneling process to pro-ceed. The oxygen species at the time of H-tunneling would under these circumstances have bonding that lies between Cu(II)-O 2 ⅐ and free O 2 . , with the expectation of an O-18 kinetic isotope effect between 1 and 3%. We note that hydrogen atom transfer to superoxo-metal complexes has been demonstrated (46), although with C-H bonds that are weak compared with the benzylic C-H bond of Ca. 85 kcal/mol (19). As originally detected from extended X-ray absorption fine structure studies of D␤M, a change in the ligand environment occurs at the Cu B site upon its reduction such that the coordination number is reduced, and a long bond to methionine becomes significantly shorter (2,3). This ligand movement has remained an enigma within the context of a mechanism in which copper is first reduced by ascorbate and then reoxidized by O 2 before substrate activation. However, if copper reoxidation and substrate functionalization occur in a tightly coupled step at Cu B , it becomes easier to rationalize the presence of a methionine at this site. Specifically, because methionine is a poor electron donor, it would ensure that the reduced copper remains electrophilic, preventing significant reduction of O 2 in the absence of substrate activation. This implies that the ligand reorganization occurring upon Cu B (I) oxidation (in which methionine is displaced) may contribute to the driving force for substrate oxidation. In this manner, changes in bonding at substrate become linked not only to changes in bonding at O 2 but to alterations in copper ligation as well.
In a recent investigation of the impact of selected mutants on the PHM-catalyzed hydroxylation of peptidyl substrates, Bell et al. (47) postulate that the substrate undergoes migration from the Cu B to Cu A site, reversing the roles of the metal sites such that Cu A performs hydroxylation and Cu B electron transfer. The basis for this proposal comes from the finding that elimination of the residue proposed to mediate inter-metal electron transfer (Gln 170 ) has little effect on rate. Additionally, these authors invoke the earlier demonstration that substrate induces binding of CO to Cu A in PHM and that the CO-stretching frequency is dependent to a small extent on the nature of the bound substrate (15). We point out that although our basic mechanism (Scheme 3) could be accommodated by a reversal in roles of the copper sites, it also eliminates the need to use extreme means (such as substrate and superoxide migration) to achieve an activated complex.
In conclusion, the mechanism in Scheme 2, I and Scheme 3 both allows for the convergence of a large body of kinetic and structural data for D␤M and PHM and provides a basis for future theoretical and experimental work on these systems. The data reported herein also argue against invoking cytochrome P-450 as well as other iron-containing monooxygenases as a "universal" paradigm in describing the mechanism of oxygen-activating metallo-enzyme systems. It is evident that the copper monooxygenases use a very different strategy for substrate functionalization than cytochrome P-450. The rules that govern copper and iron reactivity may turn out to be significantly further apart than previously recognized.