Peroxidase-catalyzed N-Demethylation Reactions SUBSTRATE DEUTERIUM ISOTOPE EFFECTS*

Deuterium isotope effects on the kinetic parameters for the hydroperoxide-supported N-demethylation of N,N-dimethylaniline catalyzed by chloroperoxidase and horseradish peroxidase were determined using N,N-di-(trideuteromethy1)aniline. The isotope effect on the V,, for the chloroperoxidase-catalyzed demeth- ylation reaction supported by ethyl hydroperoxide was 1.42 f 0.31. The isotope effects on the V,, for the horseradish peroxidase-catalyzed reaction supported by ethyl hydroperoxide and hydrogen peroxide were 1.99 f 0.39 and 4.09 f 0.27, respectively. Isotope effects ranging from 1.76 to 5.10 were observed on the VmdKm for the hydroperoxide substrate (i.e. the second order rate constant for the reaction of the hydroperoxide with the peroxidase to form compound I) in both enzyme systems when the N-methyl groups of N,N-dimethylaniline were deuterated. These results are not predicted by the simple ping-pong kinetic model for peroxidase-catalyzed N-demethylation re- actions. The data are most simply explained by a mechanism involving the transfer of deuterium (or hydro- gen) from N,N-dimethylaniline to the enzyme during catalysis. The deuterium must subsequently be displaced from the enzyme by the hydroperoxide, causing the observed iostope effects.

tiary). The lack of inhibition of the chloroperoxidase-and horseradish peroxidase-catalyzed N-demethylation of DMA' by reagents which react specifically with the superoxide anion, singlet oxygen, and the hydroxyl radical suggests that these activated oxygen species are not free intermediates in the reaction (4.5). Moreover, the near identity of the V,,./K,,, for the hydroperoxide substrate in the horseradish peroxidasecatalyzed demethylation of DMA (5) with the reported rates of horseradish peroxidase compound I formation (6) suggests that this oxidized enzyme intermediate mediates the demethylation reaction. The results of a detailed steady state kinetic analysis of the chloroperoxidase-catalyzed demethylation of DMA indicate that the reaction proceeds by a ping-pong kinetic mechanism (7), as would be expected for a reaction mediated by peroxidase compound I. Initial velocity studies of the horseradish peroxidase-catalyzed demethylation of DMA are also consistent with a ping-pong kinetic mechanism (5). In this mechanism, the hydroperoxide substrate reacts with the native peroxidase to form the oxidized enzyme intermediate, compound I, with the concomitant release of the alcohol product. DMA then binds to compound I and is oxidized, resulting in the formation of N-methylaniline and formaldehyde and the regeneration of the native peroxidase. However, while these kinetic results describe the interactions of the substrates with the enzyme, they do not provide chemical detail about the interactions. Previous studies (4,5) have demonstrated that the peroxidase-catalyzed N-demethylation of DMA does not proceed via N-oxidation of the amine moiety and support the involvement of a carbinolamine intermediate.
The observation by EPR spectroscopy of free radicals during horseradish peroxidase-catalyzed oxidation of DMA has led to the suggestion that substrate free radicals are free intermediates in the demethylation reaction (8). However, those studies were done under conditions where the oxidation of DMA to N,N,N',N'-tetramethylbenzidine is known to occur ( 9 ) , and a subsequent report (10) has suggested that these radicals are charge-transfer complexes between tetramethylbenzidine radicals and DMA. Although the identity of the intermediate species has not been conclusively determined, it is clear that the formation of formaldehyde from DMA involves the cleavage of a carbon-hydrogen bond at some point in the overall reaction.
For reactions in which a carbon-hydrogen bond is broken at some point during catalysis, an investigation of the effect of substituting a deuterium or tritium for the hydrogen atom that is released during catalysis can provide mechanistic information regarding both the chemical and kinetic details of the overall reaction. As pointed out by , the interpretation of an isotope effect on enzyme-catalyzed reactions as indicating the rate-limiting nature of bond cleavage may be oversimplified in many cases. Unlike simple chemical reactions where the intrinsic or actual magnitude of an isotope effect is often observed directly, enzyme-catalyzed reactions are complex multistep processes where the apparent isotope effects observed are often of lesser magnitude than the intrinsic isotope effect on bond cleavage. The intrinsic isotope effect on bond cleavage in enzyme-catalyzed reactions is often masked by complicating rate factors (11)(12)(13). Nevertheless, apparent isotope effects can provide mechanistic information about enzyme-catalyzed reactions.
Since the peroxidase-catalyzed demethylation of DMA involves the breaking of carbon-hydrogen bonds, we have investigated the effect of deuteration of the methyl groups of DMA on the kinetic parameters of the reaction as an additional diagnostic tool for studying the mechanism of the reaction. The results presented here indicate that the peroxidase-catalyzed N-demethylation of DMA proceeds by a mechanism where hydrogen (or deuterium) is transferred from DMA to the enzyme and must be displaced by the hydroperoxide substrate on the next turnover.

EXPERIMENTAL PROCEDURES
Enzyme Preparation-Chloroperoxidase was isolated and purified from CaMariomyces furnugo as reported previously (14). The preparations used for these studies had specific activities of greater than 2000 units/mg of protein in the standard chlorination assay (14) and exhibitedAlo3/Am ratios greater than 1.40, indicating that the enzyme preparations were at least 95% pure ( Materials-N,N-Dimethylaniline (redistilled before use) and 2,4pentanedione (gold label) were obtained from Aldrich. Ethyl hydroperoxide (10%) and ammonium acetate (ultrapure) were obtained from Polysciences Inc. Hydrogen peroxide (30%) was obtained from Fisher, and aniline (redistilled before use) was obtained from Eastman Kodak Co. The hexanes (high performance liquid chromatography grade) were purchased from Fisher. All other materials were reagent grade and obtained from commercial sources. The hydroperoxide concentrations were determined by iodometric titration (18).
The method of Fones and White (19) was used to synthesize N,Ndi(trideuteromethy1)aniline. A saturated solution of sodium acetate and aniline was placed in a thick walled glass tube and a 2-fold molar excess of trideuteromethyl iodide (Stohler Isotope Chemical CO., >99.5 atom % D) was added. The total reaction volume was 3 ml. The tube was sealed and heated overnight in an oven at 130 "C. After cooling, the contents of the tube were made basic and steam distilled into 1 N HCl. The distillate was made basic and extracted twice with 20 ml of diethyl ether. The ether extracts were combined, dried over sodium sulfate, concentrated by a gentle stream of dry nitrogen, applied to a preparative TLC plate, and chromatographed in hexanes:ethyl acetate (7.5:l) with four solvent passes. The band corresponding to N,N-di(trideuteromethy1)aniline was scraped and the product was eluted from the silica with diethyl ether. The eluate was dried over sodium sulfate and the ether removed by a gentle stream of dry nitrogen. The product was >97% pure as judged by high performance liquid chromatography, and the extent of deuteration was >99.5% as judged by the complete absence of a peak at 2.90 ppm downfield from tetramethylsilane in the proton NMR spectrum which would have been due to the methyl protons.
Enzyme Assays-The N-demethylase activities of chloroperoxidase and horseradish peroxidase were assayed by measuring the amount of formaldehyde formed using a modification of the procedure of Nash (20) as previously described (4). The reactions were initiated by the addition of the peroxidase, incubated at 25 "C for the times indicated, and terminated by the addition of 0.75 ml of 60% trichloroacetic acid. The chloroperoxidase-catalyzed reactions were incu-bated for 10 min. The horseradish peroxidase-catalyzed reactions where the concentration of the amine substrate was varied were incubated for 5 min. A 1-ml aliquot of the terminated reaction mixture was incubated with 0.5 ml of the Nash reagent as previously described (41, and the absorbance of the resulting conjugate was read at 421 nm on a Gilford 2400-S UV-visible spectrophotometer. The horseradish peroxidase-catalyzed reactions where the concentration of the hydroperoxide substrate was varied were incubated for 3 min and formaldehyde formation was quantitated by the fluorescent assay as previously described (5). These reactions were terminated by the addition of 0.45 ml of 60% trichloroacetic acid. A 1.5-ml aliquot of the terminated reaction mixture was incubated with 0.75 ml of Nash reagent as previously described (5) and the fluorescence of the resulting lutidine derivative was read in an Aminco-Bowman spectrophotofluorometer using an excitation wavelength of 410 nm and measuring the emission at 500 nm. Under the conditions used for these studies, the rate of formaldehyde formation was linear with time for the times indicated. Standard curves, in which solutions containing known amounts of formaldehyde were taken through the same procedure, were run with each experiment.
All experiments were done at least twice with each point carried out in duplicate. All lines were determined by linear regression analysis of the data and had correlation coefficients greater than 0.990. The data presented are mean values plus or minus standard errors from at least two determinations of K,,, and at least four determinations of V-. The standard errors on the isotope effects were determined by the propagation of error of the measured values.

RESULTS
The initial rate of the chloroperoxidase-catalyzed demethylation exhibited normal Michaelis-Menten saturation kinetics when N,N-di(trideuteromethy1)aniline (DMA-'Hs) was substituted for DMA. As shown in Fig. 1, the double reciprocal plots of the initial rates uer.su.s the substrate concentrations exhibited good linearity for both DMA and DMA-'Hs when the EtOOH concentration was kept constant (Fig. la) and for EtOOH when the amine concentrations were kept constant (Fig. Ib). Normal Michaelis-Menten saturation kinetics was also observed for the horseradish peroxidase-catalyzed demethylation of DMA and DMA-2H6 supported by hydrogen peroxide or by EtOOH (data not shown). The kinetic parameters for the chloroperoxidase-catalyzed demethylations of DMA and DMA-'Hs with EtOOH as the oxidant are shown in Table I. The V, , for the demethylation reaction with DMA-2Hs was somewhat less than that with DMA, giving an isotope effect of 1.42 -+ 0.31 (Table I). The isotope effect on the V,.,/K,,, for DMA was 2.99 f 0.78. Table I also shows that there was an isotope effect of 1.76 f 0.52 on the VmaJ K,,, for EtOOH when DMA-'Hs replaced DMA as substrate.
The kinetic parameters for the horseradish peroxidasecatalyzed demethylations of DMA and DMA-'Hs are shown in Tables I1 and I11 with EtOOH and hydrogen peroxide as the respective oxidants. The isotope effect on the Vmax for the EtOOH-supported demethylation was 1.99 f 0.39, while the isotope effect on the V,,, for the hydrogen peroxide-supported reaction was 4.09 f 0.27. The differences in the magnitudes of the isotope effects on Vmax with the two hydroperoxides are probably due to differences in the magnitudes of the rate factors suppressing the expression of the intrinsic isotope effect, those factors being smaller with hydrogen peroxide than with EtOOH. This difference is most likely a consequence of the almost 5-fold greater rate of horseradish peroxidase compound I formation with hydrogen peroxide relative to EtOOH (6). The isotope effects on the VrnaJKm for DMA in the horseradish peroxidase-catalyzed demethylation reaction were 3. were initiated by addition of the enzyme and incubated at 25 "C for 10 min. Formaldehyde formation was determined using the Nash assay described under "Experimental Procedures."

TABLE I TABLE 111 Isotope effects on the kinetic parameters for chloroperoxidasecatalyzed demethylation of N,N-di-(trideuteromethyUaniline
The reactions were run in sodium potassium phosphate buffer (0.5 M), pH 6.0, under the conditions described under "Experimental Procedures" and assayed for formaldehyde formation by the Nash assay.

F + B + F B + E Q + E + Q k-, k-6 SCHEME 1
where E is the native peroxidase, A is the hydroperoxide substrate, F is peroxidase compound I, P is the alcohol prodwhen DMA-*Hs was substituted for DMA. The observation uct, is DMA, and Q is the oxidation product of DMA. The of isotope effects on the Vmm/Km for the hydroperoxide subping-pong model for the demethylation reaction shown in strates in both enzyme systems indicates that the rate of Scheme 1 consists of two half-reactions, the reaction of the peroxidase compound I formation is decreased by the deuterhydroperoxide substrate with the peroxidase to form comation of the N-methyl groups of DMA. pound  where [E]T represents the total enzyme concentration. It is usually assumed that only catalysis (k5 in Scheme 1) is affected by substrate deuteration, but even if substrate binding (kl) and product release (4) were also affected, no isotope effect is predicted on the V/KA parameter on the basis of Scheme 1 (11,12). For example, the oxidation of l-deuterioglucose catalyzed by glucose oxidase proceeds without an isotope effect on the V,,,/K,,, for oxygen (22), as expected for a ping-pong mechanism (11,12). We are unaware of any reports in the literature where deuteration of one of the substrates in a reaction which proceeds by a ping-pong mechanism results in an isotope effect on the V,,,/K, for the unlabeled substrate. In contrast to the predictions for a classical ping-pong mechanism (11,12), isotope effects were observed on the V,,./K,,, for the hydroperoxide substrate in the peroxidasecatalyzed demethylation reaction when DMA-'H6 was substituted for DMA (Tables 1-111). Therefore, the simple pingpong kinetic model shown in Scheme 1 must be modified to a model which is consistent with both the steady state kinetics and isotope effect data. The isotope effects are most simply explained by a mechanism in which deuterium is transferred from DMA-'H6 to the enzyme and must subsequently be displaced by the hydroperoxide substrate on the next turnover, resulting in either a primary or secondary isotope effect.
The large magnitude of the isotope effect on the V,.,/K, for hydrogen peroxide (Table 111) suggests a primary isotope effect on compound I formation. However, a mechanism involving bond cleavage of the transferred deuterium which is both chemically reasonable and consistent with current knowledge of compound I formation (23, 24) is not apparent to us.
One possible mechanism which would explain the results by a secondary isotope effect is shown in Fig. 2. The ferric peroxidase is shown with a hydroxide as an axial ligand. This is not necessarily a representation of the native resting state of the enzyme and, in fact, there is considerable controversy as to whether the ferric heme of horseradish peroxidase is penta-or hexacoordinate (25). In any case, the original state of the enzyme is not important beyond the first turnover, where all steady state data are obtained. The ferric enzyme is shown as the hydroxide complex in Fig. 2 to more clearly demonstrate a possible origin of the isotope effect on the V,,,/K, for the hydroperoxide when DMA-'Hs is the substrate. The hydroperoxide substrate (ROOH) binds (with k l ) to the enzyme to form an initial complex. Because the hydroperoxide itself is probably too weak a nucleophile to displace the iron-bound ligand (24), the hydroperoxide may be deprotonated (with kp) by a basic active site residue (B) to produce the hydroperoxy anion, which would be an excellent nucleo-phile because of the (Y effect of the adjacent oxygen atom (26).
The hydroperoxy anion then displaces (with k3) the hydroxide to coordinate to the heme iron. The ferric heme of the enzyme is oxidized as the hydroperoxide oxygen-oxygen bond is broken, resulting in the formation (with k,) of peroxidase compound I and the alkoxy anion. The alkoxy anion recombines with the proton on active site residue B to form the alcohol (ROH), which dissociates (with k5) from compound I. Compound I then binds (with &) the DMA-'H, (R"R'-N-CD3) to form another initial complex and abstracts (with k7) a deuterium atom (or hydrogen atom when DMA is the substrate) to form a protein-bound DMA radical and peroxidase compound 11. The DMA radical then transfers (with k6) an electron to compound 11, reducing it to the ferric deuteroxide (or hydroxide) species and forming an iminium cation. The iminium cation can react (with 129) with water in the active site or can be released from the enzyme to react with water in the medium, forming a carbinolamine. The carbinolamine formed from DMA would be very unstable and would rapidly decompose to form N-methylaniline and formaldehyde.
Thus, in the mechanism shown in Fig. 2, the isotope effect on the V,.,/K,,, for the hydroperoxide substrate with D mwould arise from having to displace (with k3) the deuteroxide on the next turnover as opposed to a hydroxide when DMA is the substrate. Since deuteroxide is about twice as good a nucleophile as hydroxide (27), it would be expected to be a worse leaving group than hydroxide in the displacement reaction and would result in isotope effects of approximately 2 on the V,,/K,.
Consistent with these considerations, the isotope effects on the V,,,/K, for EtOOH were 1.76 k 0.52 in the chloroperoxidase-catalyzed demethylation reaction (Table I) and 2.14 +-0.43 in the horseradish peroxidasecatalyzed reaction (Table 11) when DMA-'H6 was substituted for DMA. The much larger isotope effect (5.10 f 0.53) observed on the V,,,../K, for hydrogen peroxide in the horseradish peroxidase-catalyzed reaction (Table 111) implies that either the model shown in Fig. 2 is not a complete description of the events taking place during peroxidase compound I formation or that deuteroxide may be an even better nucleophile relative to hydroxide when bound to the heme iron in the active site of horseradish peroxidase. The hydrophobic environment of the active site of horseradish peroxidase (28) may make the enzyme-bound deuteroxide a better nucleophile (26) than indicated by the data of Steffa and Thornton (27) for the free anion in aqueous solution.
The results of a recent intramolecular isotope effect study of the hemeprotein-catalyzed N-demethylation of N-methy1,N-trideuteriomethylaniline were interpreted as indicating that the horseradish peroxidase-catalyzed reaction proceeds via a-carbon hydrogen abstraction (k7 in Fig. 2) while the chloroperoxidase-catalyzed reaction involves electron transfer from nitrogen followed by a-carbon deprotonation of the anilinium cation radical (29). The electron transfer-deprotonation mechanism for chloroperoxidase-catalyzed N-demethylation is the stepwise equivalent of hydrogen atom abstraction (k7 in Fig. 2). The initial electron transfer step must be reversible in order to predict the observed isotope effect on the V,,,/K, for DMA in the chloroperoxidasecatalyzed reaction (Table I).
Based on their studies of the horseradish peroxidase-catalyzed N-demethylation of aminopyrine, Griffin and Ting (30) proposed a mechanism involving initial irreversible electron transfer from nitrogen followed by hydrogen atom abstraction to yield the iminium cation. When the expression for the V,,,,/K, for DMA is derived (21) for this mechanism, it does not include the rate constant for the hydrogen atom abstrac-tion step and thus predicts that there will be no isotope effect on the Vma,/K,,, for DMA. Although an isotope effect was not observed on this parameter in the EtOOH-supported reaction catalyzed by horseradish peroxidase (Table 111, isotope effects were observed for the hydrogen peroxide-supported reaction (Table 111) and for the chloroperoxidase-catalyzed reaction (Table I). Therefore, the mechanism proposed by Griffin and Ting (30) is inconsistent with the observed isotope effects.
Since the same compound I species is formed upon reaction of different hydroperoxides with horseradish peroxidase (311, the differences in the magnitudes of the isotope effects observed on the V,,,/K,,, for DMA in the horseradish peroxidase-catalyzed reaction (Tables 11 and 111) must be due to differences in the rate factors suppressing the intrinsic isotope effect (11)(12)(13), those factors being larger in the EtOOHsupported reaction.
The proposed mechanism for the oxidative demethylation of DMA (Fig. 2) is similar to the mechanism proposed by Shannon and Bruice (32) for the iodosobenzene-supported demethylation of DMA catalyzed by chlorotetraphenylporphrinatoiron (111). An alternative mechanism involving the release of DMA radicals from the enzyme and their subsequent reaction with water to form a carbinolamine species can be excluded because it would be inconsistent with the 1:1 stoichiometry of hydroperoxide consumption and formaldehyde formation observed for the horseradish peroxidasecatalyzed demethylation reaction (5). A mechanism involving the recombination of the enzyme-bound DMA radical with the oxygen in compound 11, as has been suggested for the hydroperoxide-supported oxidations catalyzed by cytochome P-450 (33), would be inconsistent with the isotope effects observed on the V,.,/K, for the hydroperoxide when DMA-*Hs is the substrate (Tables 1-111).
In the detailed model for peroxidase-catalyzed demethylation reactions' (Fig. 2), the iminum cation formed during the oxidation of DMA can react with water in the active site (derived from either the oxidant via displacement of the hydroxide or the medium) or it can dissociate from the enzyme and react with water in the medium, forming an unstable carbolinamine of DMA which would rapidly decompose to Nmethylaniline and formaldehyde. Thus, the proposed mechanism would be consistent with either the oxidant or solvent water as the source of the oxygen atom in the formaldehyde formed in peroxidase-catalyzed demethylation reactions. We have recently demonstrated (34) that the oxidant was the source of the oxygen atom incorporated into the product of the rabbit liver cytochrome P-450LM2-catalyzed N-demethylation of N-methylcarbazole when the reaction was supported by peroxy compounds or carried out using a reconstituted enzyme system including NADPH-cytochrome P-450 reductase and NADPH. The stability of the carbinolamine intermediate in the N-demethylation of N-methylcarbazole enabled analysis of its oxygen isotope content directly. The identification of the source of the oxygen atom incorporated into the product of the peroxidase-catalyzed N-demethylation reactions is important in defining the events following the formation of the enzyme-bound DMA radical (Fig. 2), but this question is very difficult to approach because of the instability of the carbinolamine of DMA and the rapid exchange between the carbonyl oxygen of formaldehyde and solvent water oxy-bony1 oxygen of formaldehyde and solvent water oxygen (35).
Substrate deuterium isotope effects on cytochrome P-450catalyzed demethylation reactions have generally been observed to be of smaller magnitude than those reported here for the demethylations catalyzed by peroxidases. Substrate deuterium isotope effects were not observed on the demethylation of DMA (36)  by a reconstituted rat liver microsomal enzyme system containing highly purified cytochrome P-450, NADPH-cytochrome P-450 reductase, NADPH, and molecular oxygen. Foster et al. (38) did not observe any isotope effect on the demethylation of l-(o-carbamoylphenyl)-3,3-dimethyltriazene catalyzed by rat liver microsomes but did observe isotope effects of 1.85 with p-nitroanisole, 1.90 with p-methoxyacetanilide, and 2.10 with p-methoxyanisole. The interpretation of isotope effects on cytochrome P-450-catalyzed demethylation reactions in microsomal suspensions or reconsituted enzyme systems is complicated by the presence of a second enzyme system, NADPH-cytochrome P-450 reductase, for which cytochrome P-450 is a substrate. The presence of the reductase may further add to the suppression of the magnitude of the isotope effects on the overall reaction. The determination of isotope effects on cytochrome P-450-catalyzed demethylation reactions using highly purified enzyme preparations and hydroperoxides as co-substrates would eliminate any suppression of the isotope effects due to the reductase and might simplify their interpretation. Comparison of the isotope effects on hydroperoxide-supported demethylation reactions catalyzed by cytochrome P-450 with those reported here for the peroxidase-catalyzed demethylation of DMA might add to our understanding of the catalytic relationship between cytochrome P-450 and peroxidases.