INTRODUCTION

Microsomal cytochromes P450¹ (also termed heme-thiolate protein P450 by the Enzyme Commission, EC 1.14.14.1) (1) catalyze a variety of mixed-function monooxygenase reactions that often result in detoxication of drugs and other xenobiotics (2-4). Occasionally, oxidation results in the bioactivation of potentially potent carcinogens, particularly with substrates metabolized by P450 2E1 (5, 6). P450 2E1 is active in the oxidation of many low molecular weight organic compounds (e.g. nitrosamines and alkenes) associated with human cancers, and the reactivity of products with DNA has been demonstrated (7, 8). DNA alkylating ability and carcinogenicity were shown to be decreased upon deuterium substitution of N-nitrosodimethylamine, now known to be a substrate of P450 2E1 (9). When rat liver microsomes were examined, deuteration was found to increase K_m for these reactions ~5-fold, but the k_cat (V_max) remained unaffected by deuterium substitution (10-12).

Most P450s are considered to operate according to a general scheme (Fig. 1) (13). Following substrate binding (step 1), ferric P450 receives 1 electron via NADPH-P450 reductase (step 2). Steps 1a and 2a represent a potential pathway by which Fe³⁺ is reduced to Fe²⁺ in the absence of substrate and suggests a possibility for later entry of the substrate into the catalytic cycle. The ferrous form of the heme binds O₂ (step 3) before undergoing a second 1-electron reduction to begin O₂ activation (step 4). Although this second electron originates from NADPH-P450 reductase, the accessory protein cytochrome b₅ (b₅, EC 4.4.2 group) appears to play some role in the delivery of the electron to the P450 (14). Insertion of the activated oxygen into the substrate is believed to occur by way of C-H bond cleavage (step 6) followed by rapid oxygen rebound to form product (step 7) (15). Step 8 is release of the product from the enzyme active site. Within the context of this scheme, the reduction (steps 2 and 4) and chemistry (step 6) are generally considered to be rate-limiting (13). Steps 9 and 10 reflect the potential for the Fe³⁺ROH complex to receive an electron from NADPH-P450 reductase, possibly leading to a second cycle of oxidation.

[View Larger Version of this Image (16K GIF file)]

High intramolecular kinetic hydrogen isotope effects are seen in many P450 reactions involving C-H bond cleavage and are usually interpreted as evidence for a hydrogen atom abstraction mechanism (16), with some caveats (17). There are fewer reports of non-competitive intermolecular hydrogen isotope effects on P450 reactions, and these tend to be on k_cat not K_m (18).

Although the effects of deuterium substitution on N-nitrosodimethylamine N-demethylation were first reported in 1973, a definitive explanation for the intermolecular isotope effect on K_m (and not k_cat) has not been given. The observation has been repeated using several P450 2E1 substrates (9-11, 19-22). Yang et al. (10) compared the competitive inhibition of alternate ²H and ¹H substrates of P450 2E1 and suggested that the observed deuterium effect was due, in part, to a rate difference in the breaking of the C-H bond and its relationship with other rate constants. It was also proposed (10) that the effect on V/K might be interpreted in the context of a generalized scheme (23, 24) in which (i) the isotopically sensitive step (C-H bond breaking) precedes a slower step (possibly product release) and (ii) the enzyme exhibits a low commitment to catalysis. Yang et al. (10) also suggested that rate-limiting product release could explain the results if a relatively high degree of uncoupling of the activated P450-oxygen complex also occurred.

MMO is a mixed-function monooxygenase that is functionally similar to P450 2E1 (25). MMO requires a multi-enzyme complex and NADH as a co-substrate. The non-heme, two-iron catalytic center catalyzes the oxidation of a variety of low molecular weight hydrocarbons. Additionally, MMO has been shown to exhibit apparent isotope effects that are similar to those seen with P450 2E1. Chemical quench studies of the intermediates and interconversion rates in the MMO catalytic cycle revealed that product release is the rate-limiting step in this cycle (26, 27). Therefore, it is plausible that this might be the case for P450 2E1.

To determine the basis for the kinetic hydrogen isotope effects observed with P450 2E1, we have used deuterium substitution and pre-steady-state kinetic techniques to characterize the effect of deuteration on individual steps of the catalytic cycle. The role of b₅ and its ability to enhance the rate of product formation were also examined. We interpret the observed kinetic isotope effects terms of rate-limiting product release and isotopically sensitive C-H bond cleavage.

EXPERIMENTAL PROCEDURES

Rabbit NADPH-P450 reductase (EC 1.6.2.4) and rabbit b₅ were purified as previously reported (28, 29). Recombinant human P450 2E1 was expressed in Escherichia coli and purified essentially as described (30). Recombinant NADPH-cytochrome P450 reductase was expressed in E. coli and purified by a modification of the method of Shen et al. (31). Apo-b₅ was prepared from rabbit liver b₅ by acid/acetone treatment (32).

Reagent grade ethanol was obtained from McCormick Distilling Co. (Weston, MO). [1,1-²H]Ethanol and d₂₃-lauric acid were purchased from Cambridge Isotope Laboratories (Andover, MA). Lauric acid was purchased from Aldrich and recrystallized from CH₃OH:H₂O (50:50, v/v). 11-Hydroxylauric acid was synthesized from 11-dodecenoic acid (Nu-Check-Prep, Elysian, MN) as described by Brown and Geoghegan (33). The identity of 11-hydroxylauric acid was confirmed by NMR and fast atom bombardment-mass spectral analysis. Cumene hydroperoxide (80%, Aldrich) was purified by extraction with alkali (34) and stored under argon Dinitrophenylhydrazine·HCl, purchased from Eastman Kodak Co., was recrystallized from H₂O before use. 18-Crown-6-ether and 4-bromomethyl-6,7-dimethoxycoumarin were purchased from Aldrich. Florox^TM reagent (2.5 mg of O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine·HCl ml¹, in pyridine) was obtained from Pierce. Pesticide grade hexane was purchased from Mallinckrodt-Baker (Paris, KY), and pesticide grade CH₃OH was from Burdick and Jackson (Muskegon, MI).

[1-³H]Ethanol (16.3 mCi/mmol, anhydrous), [¹⁴C]ethanol (55 mCi/mmol), and [1-¹⁴C]lauric acid (55 mCi/mmol) were purchased from American Radiolabeled Chemicals (St. Louis, MO). Before use, [1-³H]ethanol and [1-¹⁴C]ethanol were purified on Bakerbond^TM octadecylsilane 3-ml disposable extraction columns (J. T. Baker Inc.) to remove, respectively, ³H₂O and [¹⁴C]acetaldehyde, by-products of radioactive ethanol synthesis. For purification, extraction columns were first washed with 10 ml of CH₃OH and then equilibrated with 10 ml of H₂O. [1-³H]Ethanol (100 µl diluted with H₂O to 340 mM, 16.3 mCi mmol¹) or 150 µl of [¹⁴C]ethanol (diluted with [¹H]ethanol and H₂O to 170 mM, 10 mCi of mmol¹) was loaded onto the column. Analytes were eluted with <3 ml of H₂O, and ~150-µl fractions were collected. A portion of each fraction was counted by liquid scintillation spectrometry. Radioactivity was plotted against fraction number, and the fractions comprising the major radioactive peak were pooled. The purity of the pooled solutions was confirmed by HPLC interfaced with radioflow detection (described under "Intrinsic Isotope Effect Estimation," see below). The concentrations of ethanol solutions were determined by measuring NADH formation spectrophotometrically at 340 nm in a reduction assay using yeast alcohol dehydrogenase (35). The purified solutions were then counted by liquid scintillation to determine the specific activity.

Unless otherwise specified, P450 2E1 (1.0 µM) was reconstituted with b₅ (2.0 µM), NADPH-P450 reductase (3.0 µM), and DLPC (30 µM) in 100 mM potassium phosphate buffer, pH 7.4. For steady-state measurements, 100-µl reactions were initiated with either an NADPH-generating system (36) or 1.0 mM cumene hydroperoxide and incubated for 10 min at 37 °C in reaction vials sealed with Teflon-lined rubber septa. Reactions supported by cumene hydroperoxide lacked NADPH-P450 reductase and b₅. Reactions were terminated with 20 µl of a mixture of 17% ZnSO₄ (w/v) and 0.55 mM semicarbazide (1:1, v/v) and centrifuged following addition of saturated Ba(OH)₂ (5.0 µl). The acetaldehyde product was rederivatized to form the 2,4-dinitrophenylhydrazone (37) and then analyzed by HPLC using a Zorbax 6.2 × 80-mm octadecylsilane reversed-phase analytical column (3 µm, DuPont Chromatography Products, Wilmington, DE) (H₂O:CH₃CN, 45:55, v:v; 2.0 ml min¹), monitoring A₃₄₀) (38). For reactions with [³H]- and [¹⁴C]ethanol, the HPLC peaks containing the radioactive product were collected and counted by liquid scintillation spectrometry, with calibration of counting efficiency using external [³H]- and [¹⁴C]toluene standards.

P450 2E1 (0.5 µM) was reconstituted with NADPH-P450 reductase and b₅ as described for the ethanol oxidation assay. Reactions (400 µl) with lauric acid or d₂₃-lauric acid (0-200 µM, as sodium salt) were terminated after 10 min at 37 °C by adding 50 µl of 12.5% H₂SO₄ (v:v). The quenched reactions were extracted twice with 6 ml of (C₂H₅)₂O, and the combined extracts were dried over Na₂SO₄ and then evaporated under a stream of N₂. The residue was redissolved in 100 µl of 18-crown-6-ether dissolved in CH₃CN (2.5 mg ml¹), to which was added 100 µl of 4-bromomethyl-6,7-dimethoxycoumarin dissolved in (CH₃)₂CO (10 mg ml¹) (in the presence of 2 mg of anhydrous K₂CO₃). The samples were incubated 60 min at 70 °C to form fluorescent derivatives. Product formation was measured by HPLC using a Zorbax 6.2 × 80 mm octadecylsilane reversed-phase analytical column (3 µm, DuPont, H₂O:CH₃CN, 41:59, v:v; 2.0 ml min¹). Fluorescence was monitored at _excitation 375 nm, _emission 470 nm (39). External standards of 11-hydroxylauric acid were used for quantitation. Experiments using [1-¹⁴C]lauric acid were quenched with H₂SO₄ as described above, and reactions (40 µl) were extracted twice with 3 ml of (C₂H₅)₂O, and the combined extracts were dried under N₂. The residue was redissolved in 100 µl of CH₃OH, and 50-µl aliquots were analyzed by HPLC interfaced with radioflow counting (Zorbax^TM 4.6 × 250-mm octadecylsilane reversed-phase analytical column, 5 µm, DuPont) (H₂O:CH₃CN, 50:50, v:v; 2.0 ml min¹). The [1-¹⁴C]11-hydroxylauric acid peak was collected and re-counted by liquid scintillation spectrometry.

Deuterium isotope effects were determined by two methods (40). In a non-competitive method, P450 2E1 was incubated with d₀-ethanol or d₂-[1,1-²H]ethanol (0-100 mM) or d₀-lauric acid or d₂₃-lauric acid (0-150 µM), and the products were analyzed as described. K_m and k_cat were calculated using a k_cat nonlinear regression program (Bio-Metallics, Princeton, NJ). In a competitive method, P450 2E1 was incubated with a 1:1 mixture (v/v) of d₀-ethanol and d₂-ethanol (20 mM). The 2,4-dinitrophenylhydrazones were extracted into CH₂Cl₂ and analyzed by capillary column GC-PIEIMS (17, 21). ^D(V/K) was determined as the ratio of ¹H to ²H product formed in the competitive reaction.

The approach was that of Miwa et al. (41) and Northrop (42), only modified in regard to the analysis of isotopes in the products. The tritium isotope effect ^T(V/K) on ethanol oxidation was determined by reaction of P450 2E1 with [1-³H]ethanol (20 mM). Previous work by others (21), with rabbit P450s, indicated no stereoselectivity in the removal of hydrogen from ethanol. ³H₂O formation during ethanol oxidation was measured by injecting one-half of the total reaction supernatant onto two 4.6 × 250-mm octadecylsilane reversed-phase analytical columns (Zorbax^TM column from DuPont and Econosphere^TM column from Alltech Associates, Deerfield, IL, both 5 µm) connected in tandem (mobile phase 100% H₂O; 1.0 ml min¹) interfaced with a radioflow counter (IN/US Systems, Inc., Tampa, FL). Retention times and separation of analytes are shown in Figs. 2 and 3, respectively. The ³H₂O peak was collected and re-counted by liquid scintillation spectrometry. ^T(V/K) and ^Dk were calculated according to the method of Northrop (42, 43) (see Equation 1),

(Eq. 1)

where f is the fractional conversion of substrate to product, determined as the ratio of ³H₂O to the initial amount of [³H]ethanol; SA_p is the specific activity of the product, expressed as a ratio of ³H₂O formed/total acetaldehyde formed, and SA_o is the specific activity of the substrate. Finally, ^Dk was determined (Equation 2) from the Swain relationship (44)

(Eq. 2)

[View Larger Version of this Image (10K GIF file)]

[View Larger Version of this Image (11K GIF file)]

Pre-steady-state ethanol oxidation reactions were done in a quench-flow apparatus (model RQF-3, KinTek Corp., State College, PA). P450 2E1 (40 or 200 pmol) was incubated in the presence of either 200 mM [¹H]ethanol, 20 mM [³H]- or [¹⁴C]ethanol, or 100 µM [¹⁴C]lauric acid in a 40-µl reaction volume for a period ranging from 2 ms to 10 min (as indicated) at 37 °C. Reactions with radioactive ethanol were quenched with a ZnSO₄/semicarbazide mixture and analyzed as described for steady-state reactions. Reactions using [¹H]ethanol were quenched by the addition of 30% pesticide grade CH₃OH and derivatized with PFB-hydroxylamine by reacting the 250-µl quenched reaction volume with 20 µl of Florox^TM reagent for 20 min at 60 °C. The oxime derivative was extracted into 1.0 ml of pesticide grade hexane, and the residual reagent was back-extracted following addition of 3 drops of concentrated H₂SO₄. The hexane layer (0.7 ml) was dried under N₂ and then redissolved in 20 µl of hexane. The oxime derivative was analyzed by GC-NICIMS (45-47). Reactions with [¹⁴C]lauric acid were analyzed as described previously.

P450 2E1 was reconstituted with NADPH-P450 reductase with or without b₅ as described for steady-state ethanol oxidation experiments. Reconstituted enzyme (196 µl) was preincubated for 1 min at 37 °C in the presence or absence of substrate (20 mM ethanol or 0.15 mM chlorzoxazone). Reactions were initiated with the addition of 4 µl of 10 mM NADPH, and the decrease in absorbance at 340 nm was monitored spectrophotometrically. UV-visible spectra were recorded using a modified Cary 14/OLIS spectrophotometer (On-Line Instrument Systems, Bogart, GA). Rates of NADPH oxidation were calculated using ₃₄₀ = 6.22 mM¹ cm¹ for NADPH.

Reaction systems were prepared exactly as described above, except that reaction volumes were 0.5 ml. Reactions were initiated by adding the NADPH-generating system and were terminated by adding 0.8 ml of cold CF₃CO₂H (3%, w:v) after 10 min at 37 °C. H₂O₂ was determined spectrophotometrically by reaction with ferroammonium sulfate and potassium thiocyanate as described (48).

Enzyme mixtures and NADPH (600 µM) solutions were prepared separately as described for steady-state reactions. Some enzyme mixtures contained 20 mM ethanol (or 0.15 mM chlorzoxazone). The solutions were made anaerobic by alternating applications of vacuum and argon to a closed system as described previously (49). Reduction kinetics were monitored in a stopped-flow apparatus (Applied Photophysics SX-17MV instrument, Applied Photophysics, Leatherhead, UK) at 37 °C under a CO atmosphere. Rapid reduction of P450 2E1 to a Fe²⁺·CO complex was observed at 450 nm upon mixing of reconstituted enzymes with NADPH. Data were collected using the Applied Photophysics software system and fitted to exponential equations using a Marquardt-Levenberg algorithm for nonlinear regression analysis. Results are reported as three to eight individually monitored reactions averaged using the manufacturer's software.

The hydrazone derivatives were analyzed by GC-PIEIMS. Analytes were separated on a 15-m SPB^TM-1 fused silica capillary column (Supelco, Bellefonte, PA) interfaced to a Finnigan INCOS 50 mass spectrometer (Finnigan, San Jose, CA). GC conditions were as follows: carrier gas (²He) at constant pressure of 10 p.s.i.; injection port 230 °C; transfer line 260 °C. The initial column temperature was 150 °C and was increased to 300 °C at a rate of 20 °C min¹. The [¹H]acetaldehyde hydrazone was detected by selected ion monitoring at m/z 224, and the [²H]acetaldehyde hydrazone was monitored at m/z 225 (Fig. 4). GC-NICIMS of the PFB-oxime derivatives was performed using a 30-m SPB^TM-5 fused silica capillary column (Supelco) coupled to a Hewlett-Packard 5989A mass spectrometer (Hewlett-Packard, Wilmington, DE). GC conditions were as follows: carrier gas (²He) at constant pressure of 0.8 p.s.i.; injection port 265 °C; transfer line 270 °C. The column was initially held at 65 °C for 10 min, increased by 5 °C min¹ to 100 °C, and then raised at 20 °C min¹ to 250 °C (Fig. 5). Acetaldehyde-PFB oxime was quantitated by selected ion monitoring at m/z 181, using propionaldehyde-PFB oxime as an internal standard. The retention times of the analytes were verified by monitoring ions corresponding to M⁺  20 (M⁺-HF) for each of the oximes (Fig. 6).

[View Larger Version of this Image (18K GIF file)]

[View Larger Version of this Image (13K GIF file)]

[View Larger Version of this Image (14K GIF file)]

RESULTS

In rat liver microsomes, deuterium substitution of P450 2E1 substrates usually results in a 3-5-fold increase in K_m without any effect on k_cat (9, 11, 12, 19, 20, 50). Recombinant human cytochrome P450 2E1 was reconstituted with NADPH-P450 reductase and b₅ and incubated with a variety of substrates and their deuterated analogs to determine the effects of deuterium substitution on the steady-state kinetic parameters (Fig. 7 and Table I). Deuteration of ethanol increased the K_m 4.7-fold. k_cat was essentially unaffected so that ^DV  1 (Fig. 7A). In a reaction system supported by the oxygen surrogate cumene hydroperoxide, the effects on K_m and k_cat for ethanol oxidation were the same. K_m was increased 6.0-fold and k_cat was unchanged. For each of these systems, the apparent deuterium isotope effect, ^D(V/K), was ~5.²

Hanes-Woolf plots of P450 2E1-catalyzed oxidation of [¹H]ethanol () and 1,1-[²H]ethanol () (A) or [¹H]lauric acid () and d

-lauric acid () (B).

1

1

[View Larger Version of this Image (12K GIF file)]

Table I. Steady-state kinetic parameters for oxidations catalyzed by recombinant human cytochrome P450 2E1 Rate determinations were performed in duplicate for each substrate concentration. Kinetic values and the stated error estimates were calculated by nonlinear regression analysis using the kcat program. System kcat Km DV D(V/K) nmol min(nmol P450)1 mM Reductase, b5, CH3CH2OH 2.7 (±0.21) 11.2 (±1.6) Reductase, b5, CH3CD2OH 2.4 (±0.14) 52.6 (±6.4) 1.1 5.2 Cumene hydroperoxide, CH3CH2OH 0.81 (±0.07) 5.3 (±1.2) Cumene hydroperoxide, CH3CD2OH 1.02 (±0.05) 32 (±4) 0.79 4.7 Reductase, b5, lauric acid 3.2 (±0.5) 0.066 (±0.024) Reductase, b5, d23-lauric acid 1.5 (±0.3) 0.061 (±0.020) 2.1 2.0

For lauric acid 11-hydroxylation, the isotope effect was on k_cat rather than on K_m (Fig. 7B). Deuteration of lauric acid decreased k_cat from 2.4 to 1.5 min¹ (^DV = 2.1), whereas K_m remained unaffected.

^Dk for ethanol oxidation was determined from the Swain relationship, which considers ^D(V/K) as well as the apparent tritium isotope effect, ^T(V/K) (44). The approach has been applied to another P450-mediated reaction, 7-ethoxycoumarin O-deethylation, by Miwa et al. (41), and an identical strategy was used here. ^D(V/K) was measured in a competitive assay in which P450 2E1 was incubated with a 1:1 mixture of d₀- and d₂-ethanol. ^D(V/K) was calculated as a ratio of ¹H to ²H removal from the substrate and was found to be 3.9 ± 0.3. ^T(V/K) was also determined competitively by incubating the enzyme with a mixture of [¹H]- and [1-³H]ethanol. Removal of ³H from the C1 position results in the formation of a molecule of ³H₂O. Alternatively, removal of ¹H from [1-³H]ethanol results in the formation of [1-³H]acetaldehyde. Using the equation ^T(V/K) = (log(1  f))/log(1  f(SA_p/SA_o)) (defined under "Experimental Procedures"), ^T(V/K) = 6.8. From the relationship (^Dk  1)/(^Dk^1.442  1) = (^D(V/K)  1)/ (^T(V/K)  1), ^Dk = 3.8.

Effect of b₅ on Steady-state Kinetic Parametersb₅ has been shown to enhance the rate of product formation in P450 2E1-catalyzed reactions (38, 51

54). When P450 2E1 was reconstituted with or without b₅, the values for k_cat were 16 and 19 min¹, respectively (data not shown). The K_m for ethanol was increased ~2-fold, from 28 to 52 mM, in the absence of b₅. Apo-b₅ did not have an effect. The effect of deuterium substitution on ethanol oxidation in the reactions system lacking b₅ was the same as observed in the reaction system that included b₅.

To evaluate the role of substrate on P450 2E1 reduction rates, P450 2E1 was reconstituted with NADPH-P450 reductase, and reduction was monitored in the presence or absence of 170 mM ethanol (Fig. 8). P450 2E1 showed biphasic reduction kinetics that could be fit to bi-exponential equations to calculate rate constants. P450 reduction rates were not substantially altered by adding substrate to the reaction mixture. The rapid phase of reduction was 1800 min¹ without ethanol and 1380 min¹ with ethanol. The rates of the slow phase were 120 and 300 min¹ with and without ethanol, respectively.

[View Larger Version of this Image (22K GIF file)]

b₅ decreased the apparent rate of reduction of ferric P450 2E1 from 1800 to 420 min¹ (rapid phase) and from 300 to 12 min¹ (slow phase). Further addition of ethanol to the reaction system already containing b₅ did not significantly change the reduction rate (results not shown).

The effects of b₅ on NADPH oxidation and O₂ utilization were also examined. P450 2E1 reconstituted with b₅ showed a 35% decrease in NADPH oxidation rates (from 55 to 36 min¹) as compared with P450 2E1 systems that lacked b₅. H₂O₂ formation remained relatively unchanged (14 min¹). b₅ yielded a 42% increase in the rate of acetaldehyde formation, indicating more efficient coupling, consonant with results of others with P450 2E1 (55).

Rapid quench kinetic techniques were used to determine pre-steady-state product formation in P450 2E1-catalyzed ethanol oxidation. Initially, ³H- and ¹⁴C-radiolabeled substrates were used to improve sensitivity for detection of product formation (Fig. 9, A and B). Although substrates were purified before use, the remaining background contribution confounded determination of the burst phase amplitude for oxidation of [¹⁴C]ethanol (Fig. 9B). An improved method of quantitation involved formation of the PFB-oxime derivative of acetaldehyde, which could be analyzed by GC-NICIMS (Fig. 9C). This system also has the advantage that stable isotopes may be utilized.

[View Larger Version of this Image (22K GIF file)]

All three methods clearly revealed two distinct phases of product formation for ethanol oxidation. In each case, a rapid pre-steady-state burst of product formation was followed by a much slower phase of product formation that roughly corresponded to the observed steady-state rate. The rates of burst phase product formation were 410, 370, and 410 min¹ for [³H]-, [¹⁴C]-, and [¹H]ethanol, respectively (Fig. 10). The amplitudes of the bursts correlated well with the amounts of P450 used in the reaction.

[View Larger Version of this Image (14K GIF file)]

Pre-steady-state analysis of product formation during lauric acid hydroxylation did not show a kinetic burst. The time course for product formation was linear through the first turnover cycle and into subsequent cycles (Fig. 9D).

Since b₅ has been shown to enhance both the rate of product formation and the extent of O₂ coupling in steady-state product formation, its effect on pre-steady-state product formation was examined. The amplitude of the burst was diminished in comparison to reactions containing b₅ (Fig. 10B). Only ~30% of P450 yielded product during the burst. The rate of the burst, however, was similar to the rates observed with b₅ present.

Although deuterium substitution does not affect steady-state rates of product formation at high substrate concentration, the measurable intrinsic isotope effect suggests that subsets of reaction steps may be affected if they include the isotopically sensitive step. Deuteration of ethanol decreased the rate of the burst phase from 410 to 130 min¹, so that ^DV for burst phase product formation is 3.2 (the experiment was done at saturating substrate concentration, thus the effect is presented as ^DV rather than ^Dv).

DISCUSSION

Isotopic substitution is frequently used as a probe for studying enzyme mechanisms. In reactions that involve a proton transfer step, deuteration of substrates often produces an apparent isotope effect on k_cat (V_max). Oxidation of substrates by P450 2E1 involves a C-H bond breaking step, but deuterium substitution has been shown by several investigators to have little effect on k_cat. Rather, there is an intermolecular isotope effect on K_m for many substrates of P450 2E1 (9, 11, 12, 19, 20, 50).

In liver microsomes, this isotope effect on V/K (but not k_cat) has been reported for N-nitrosodimethylamine (10, 12, 54), CH₂Cl₂ (50), and ethanol (10). We measured ^D(V/K) values of 3-5 for ethanol oxidation catalyzed by purified recombinant human P450 2E1. In a system where NADPH and O₂ were replaced with the oxygen surrogate cumene hydroperoxide, similar isotope effects were observed, indicating that the basis for the effect on K_m is related to steps in the latter portion of the catalytic cycle (as described in Fig. 1). Additionally, early steps, i.e. ferric P450 2E1 reduction (step 2 or 1a) and oxygen activation (steps 3-5), were not affected by substrate. Enzyme reduction prior to the addition of substrate is reminiscent of the mechanism of liver microsomal flavin-containing monooxygenases in which FAD reduction by NADPH is preceded by and is independent of substrate (56-58). Following oxygen activation, substrate undergoes a first-order reaction with activated oxygen (flavin 4a-hydroperoxide) to form product. Andersen et al. (50) proposed a similar mechanism for P450 2E1 that involved rate-limiting O₂ activation preceding substrate binding. We are of the opinion that bacterial MMO (25-27) serves as a better model for the observations with P450 2E1. The catalytic center of MMO can be reduced prior to substrate binding. Additionally, MMO shows kinetic hydrogen isotope effects similar to P450 2E1, and product release has been proven to be rate-limiting in the MMO catalytic cycle (26, 27). Although a mechanism resembling that for flavin-containing monooxygenase cannot be dismissed, there is no strong evidence that this should be preferred to explain the results, and the MMO model appears more reasonable.

b₅, which has been shown to enhance the product formation rates with a number of substrates for P450 2E1 (i.e. chlorzoxazone, N-nitrosodimethylamine, aniline, diethyl ether) (38, 53, 53, 54), did not alter the deuterium effect on V/K for ethanol oxidation but did slow the rates of steps 1a and 2 and decrease steady-state NADPH oxidation, in agreement with previous reports (55). The effect of b₅ on apparent P450 reduction rates may be explained in a model in which electrons from NADPH-P450 reductase are transferred to P450 2E1, which then reduces b₅ and requires reduction again (59).

Rapid quench kinetic experiments clearly indicated a rapid burst of pre-steady-state product formation (k 400 min¹) for ethanol oxidation. This behavior is indicative of a mechanism in which the rate-limiting event occurs after product formation. We assign the rate-limiting step as product release.

The isotope effect on K_m, but not on k_cat, cannot be explained in terms of a model where K_m ~ K_D = (k₁ + k₂)/k₁, where k₂ is a single slow step approximating k_cat and both k₁ and k₁ (substrate binding and dissociation) are rapid. The only isotopically sensitive step in P450 catalysis should be step 6 of Fig. 1. We propose that in the optimal reconstituted P450 2E1 system (with NADPH-P450 reductase and b₅), or in microsomes, product release is effectively rate-limiting. Thus (using the nomenclature of the steps in Fig. 1) k_cat ~ k₈ [E]_T and (see Equation 3)

(Eq. 3)

(where [E]_T is the total enzyme concentration). However, the units for the approximation of K_m are inappropriate, and no dependence on [RH], the concentration of substrate, would be expected. A more extensive treatment of the system is available in a model presented by Kuby (60) and suggested for consideration in our own earlier review of P450 isotope effects (61). In this case, which is still simplified, the reaction is depicted as shown in Equation 4.

(Eq. 4)

where S is the substrate and P the product. Then

(Eq. 5)

(Eq. 6)

so that K_m has units of molarity. Therefore,

(Eq. 7)

and

(Eq. 8)

where ^DV is the kinetic deuterium isotope effect on k_cat, and ^D(V/K) is the kinetic deuterium isotope effect on the ratio of k_cat/K_m, according to the convention of Northrop (40, 42, 43). In this mechanism the isotopically sensitive step is included in k₃. The value ^D(V/K) is a reflection of ^Dk₃ when k₃/k₂ approaches 0, i.e. k₂  k₃. The value of ^D(V/K) is ~3-5 in our own and other studies (10). ^Dk for C-H bond breaking was found to be 3.8, and the isotopic effect on product formation in the pre-steady-state studies was estimated to be 3.2 (Fig. 10), in reasonable agreement. Therefore, k₃/k₂ must be near 0. O₂ activation preceding substrate binding may occur to give a high level of oxygen uncoupling, or reversible substrate binding may be rapid so that k₂  k_3. Since ^DV  1, then ^Dk₃ + k₃/k₅  1 + k₃/k₅. This is valid only if k₃  k₅. In fact, from results presented here, k₃ (the rate of product formation) is much faster than k₅ (the rate of product release). Hence, the above expressions for k_cat and K_m are reduced to Equations 9 and 10.

(Eq. 9)

(Eq. 10)

Thus, this paradigm includes rate-limiting product release in the overall reaction. K_m increases as k₃ decreases, consistent with the observed effect of isotopic substitution (and a decrease in V/K and a value of ^D(V/K) greater than unity). This model may also be used to explain the observations with b₅. With respect to this scheme, b₅ essentially has the effect of increasing k₃, consequently decreasing K_m. The step measured by k₃ is a simplification of several events, including P450 reduction, oxygen binding, further oxygen activation, and insertion of oxygen into the substrate, the step that must be sensitive to isotopic substitution.

Although the pre-steady-state burst of product formation may be interpreted in terms of rate-limiting product release, it is acknowledged that the rate-limiting step may be any first-order event required to renew P450 to its ground state. This step may include some degree of conformational change associated with product release. However, the burst kinetics can only be explained by a rate-limiting step following formation of acetaldehyde. A chemical explanation for rate-limiting product release is not obvious. The possibility was considered that the carbonyl product may bond with residues in the active site. In support of this, lauric acid, which is hydroxylated to form an alcohol product, showed an isotope effect on k_cat rather than K_m and did not show a kinetic burst of product formation (Fig. 9D). Also, no burst was observed in similar studies done on chlorzoxazone 6-hydroxylation (62) (an aromatic hydroxylation, results not shown). Therefore, we conclude that the nature of the carbonyl product may account for the kinetic behavior observed in the reactions.

There is some evidence, at least for ethanol oxidation, that the aldehyde product may be an even better substrate for P450 2E1 than the alcohol (63, 64). Because this P450 is rapidly reduced (k > 1000 min¹) in the presence or absence of substrate, it is possible that the product of alcohol oxidation might remain in the enzyme active site of P450 2E1 and undergo further oxidation to the carboxylic acid (Fig. 1, steps 9 and 10). (We also measured the rate of ferric P450 2E1 reduction in the presence of 500 µM acetaldehyde and found it to be identical to that observed in the absence.) Occupation of the active site by-product may account for the high level of uncoupling seen with this enzyme. Continued production of reaction oxygen species in the absence of renewable substrate may force release of the oxygen in the form of H₂O₂ or H₂O. This is supported by the high level of coupling seen in the pre-steady-state phase of product formation preceding rate-limiting product release.

It is clear that this model is not valid for all substrates for P450 2E1. As with lauric acid, deuterium substitution of the substrate 7-ethoxycoumarin resulted in a decrease in k_cat with no effect on K_m when 7-hydroxycoumarin was measured as the product (results not shown). Whether this is a result of switching to alternate pathways (65-68) or formation of a product that does not show rate-limiting product release remains to be investigated. However, our current view is that the paradigm of rate-limiting product release should explain all P450 2E1-catalyzed reactions in which a kinetic hydrogen isotope effect on K_m has been observed (9-12, 19, 20, 50).