Cytochrome P450 7A1 Cholesterol 7α-Hydroxylation

Cytochrome P450 (P450) 7A1 is well known as the cholesterol 7α-hydroxylase, the first enzyme involved in bile acid synthesis from cholesterol. The human enzyme has been reported to have the highest catalytic activity of any mammalian P450. Analyses of individual steps of cholesterol 7α-hydroxylation reaction revealed several characteristics of this reaction: (i) two-step binding of cholesterol to ferric P450, with an apparent Kd of 0.51 μm, (ii) a rapid reduction rate in the presence of cholesterol (∼10 s−1 for the fast phase), (iii) rapid formation of a ferrous P450-cholesterol-O2 complex (29 s−1), (iv) the lack of a non-competitive kinetic deuterium isotope effect, (v) the lack of a kinetic burst, and (vi) the lack of a deuterium isotope effect when the reaction was initiated with the ferrous P450-cholesterol complex. A minimum kinetic model was developed and is consistent with all of the observed phenomena and the rates of cholesterol 7α-hydroxylation and H2O and H2O2 formation. The results indicate that the first electron transfer step, although rapid, becomes rate-limiting in the overall P450 7A1 reaction. This is a different phenomenon compared with other P450s that have much lower rates of catalysis, attributed to the much more efficient substrate oxidation steps in this reaction.

from Fisher. 5-Deazaflavin was a gift of the late V. Massey (University of Michigan, Ann Arbor, MI).
The expression level of P450 7A1 was ϳ 600 nmol/liter. Purified P450 7A1 (final yield ϳ250 nmol from 1 liter of bacteria) showed a single major band when analyzed for electrophoretic purity as well as typical P450 spectral properties (supplemental Figs. S1 and S2).
Other Enzyme Preparations-Rat NADPH-P450 reductase was expressed in E. coli and purified as described elsewhere (19).
Spectroscopy-NMR spectra were recorded using a Bruker 500-MHz spectrometer in the Vanderbilt facility (Bruker, Billerica, MA).
Mass spectra of synthesized compounds were acquired in the Vanderbilt facility using a Thermo LTQ instrument (ThermoFinnigan, Sunnydale, CA) equipped with a Waters Acquity UPLC system (Waters, Milford, MA). A Thermo-Finnigan TSQ Quantum mass spectrometer (ThermoFinnigan) connected to a Waters Acuity UPLC system was used for quantitation of dansylated compounds. UV-visible spectra were recorded with either an Aminco DW2/OLIS or a Cary14/OLIS spectrophotometer (On-Line Instrument Systems, Bogart, GA).
Stopped-flow UV-visible absorbance experiments were performed using an OLIS RSM-1000 instrument (On-Line Instrument Systems) as described previously (16,17). Stoppedflow experiments were generally reported as the averages of four individual kinetic assays.
Synthesis of 7-[ 2 H 2 ]-Cholesterol-A general procedure (20) was used; Al 2 HCl 2 was prepared by stirring LiAl 2 H 4 (Sigma, 58 mg, 1.4 mmol, Ͼ 98% D atom excess) and AlCl 3 (0.73 g, 5.5 mmol) in dry (C 2 H 5 ) 2 O (freshly opened can) for 10 min at Ϫ20°C under dry N 2 . To this was added (dropwise, in (C 2 H 5 ) 2 O) 184 mg of 7-ketocholesterol (Sigma, 0.46 mmol) over 10 min. The reaction was stirred for an additional 10 min and quenched with the careful addition of H 2 O. The product was extracted into (C 2 H 5 ) 2 O three times, and the combined layers were dried with MgSO 4 and concentrated in vacuo. The product was purified using isocratic reversed-phase HPLC with a Phenomenex octadecylsilane column (10 ϫ 250 mm, 5 m, Phenomenex, Torrance, CA). HPLC conditions were as follows. The solvent contained 45% CH 3 CN, 50% CH 3 OH, and 5% tert-butyl methyl ether (v/v/v), and the flow rate was 1.0 ml/min. The column temperature was maintained at 25°C. The t R of d 2 -cholesterol was 32 min. LC-MS analysis indicated that the purified d 2 -cholesterol had Ͼ98.5% isotopic purity (supplemental Fig. S3, A and B). The disappearance of the ␦ 1.97 1 H NMR peak and the decrease in the number of integrated 1 H peaks located between ␦ 1.   FEBRUARY 11, 2011 • VOLUME 286 • NUMBER 6 terol is also indicative of the incorporation of deuterium atoms at both the 7␣-and 7␤-positions.

Kinetics of P450 7A1
Measurement of Enzyme Activity with Reconstituted Enzyme Systems-Steady-state enzyme assays were generally conducted in a 0.50-ml reaction volume at 37°C for 60 s in 50 mM potassium phosphate buffer (pH 7.4). Enzyme reaction mixtures typically contained 0.02 M P450 7A1, 1.0 M NADPH-P450 reductase, 60 M di-12:0 GPC, 3.1 mM HP␤CD (0.45%, w/v)), and 0.41 mM Tween 20 (0.05%, v/v)). Incubations were initiated by the addition of an NADPH-generating system (23). Reactions were quenched with 2.0 ml of CH 2 Cl 2 including 150 pmol of 17␣-ethynylestradiol (as an internal standard) and mixed with a vortex device. After centrifugation (2000 ϫ g, 5 min), 1.4 ml of the organic layer was transferred and taken to dryness under an N 2 stream. A general method for derivatization with dansyl chloride (24) was used with slight modification. Briefly, the samples were dissolved in 200 l of CH 2 Cl 2 containing 2 mg of dansyl chloride, 0.5 mg of 4,4-dimethylaminopyridine, and 2 l of triethylamine and incubated at 65°C for 1 h. The samples were dried under an N 2 stream and then dissolved in 100 l of CH 3 CN for analysis.
Dansylated products were analyzed with LC-MS using a UPLC system connected to a TSQ Quantum mass spectrometer with a Hypersil GOLD octadecylsilane column (2.1 ϫ 150 mm, 3 m, Thermo Scientific). LC conditions were as follows. Solvent A contained 95% H 2 O, 5% CH 3 CN, and 0.1% HCO 2 H (v/v/v), and solvent B contained 5% H 2 O, 85% CH 3 CN, 10% tert-butyl methyl ether, and 0.1% HCO 2 H (v/v/v/v). The column was maintained at the initial condition of 50% B (v/v) for 0.5 min with a flow rate of 350 l/min followed by a linear gradient increasing to 100% B over 2.0 min. This condition was maintained for 6.5 min and then returned to the initial condition over 0.05 min and maintained until the end of a 12-min run. The column temperature was maintained at 40°C. The injection volume (onto the column) was 15 l. MS analyses were performed in the positive ion electrospray mode. Quantitation was done in a multiple reaction monitoring mode (dansylated 7␣-OH cholesterol, m/z 636 3 367, collision energy 10 V; dansylated 17␣-ethynylestradiol, m/z 530 3 171, collision energy 40 V). The following (optimized) parameters were used for the detection of the analyte and the internal standard; N 2 sheath gas, 27 p.s.i.; N 2 auxiliary gas, 25 p.s.i.; spray voltage, 5.0 kV; capillary temperature, 270°C; capillary offset, 35 V; tube lens voltage, 221 V; argon collision gas, 1.5 mtorr; scan time, 50 ms Q3 scan width 1 m/z; Q1/Q3 peak widths at half-maximum, 0.7 m/z. The data were collected and quantified using ThermoFinnigan XCalibur Version 1.0 software.
Kinetic deuterium isotope effects were determined using a non-competitive method. P450 7A1 was incubated with either d 0 -cholesterol or 7,7-d 2 -cholesterol, varying the concentration from 0.625 to 20 M in the general reconstituted system. The k cat and K m values were calculated using the program Dynafit (25). The product formed from 7,7-d 2 -cholesterol by P450 7A1 was 7␤-d 1 ,7␣-OH cholesterol (26) (see below). Dansylated 7␤-d 1 ,7␣-OH cholesterol was quantified in the multiple reaction monitoring method, monitoring at m/z 637 3 368, assuming that the ion intensity of dansylated 7␤-d 1 ,7␣-OH cholesterol is the same as that of dansylated 7␣-OH cholesterol.
Pre-steady-state rapid quench kinetic experiments were conducted with a quenched-flow apparatus (model RFQ-3, KinTek Corp., Austin, TX). Enzyme reaction mixtures contained 0.5 M P450 7A1, 7.5 M NADPH-P450 reductase, 60 M di-12:0 GPC, 50 M cholesterol, 3.1 mM HP␤CD, and 0.41 mM Tween 20. Incubations were initiated by the rapid addition of 500 M NADPH and quenched by the addition of 2% ZnSO 4 (w/v) after a time period varying from 20 ms to 2 s at 37°C. After extraction, the derivatization and quantitation steps were conducted as described above.
Rates of NADPH Oxidation and H 2 O 2 and H 2 O Formation-NADPH oxidation and H 2 O 2 and H 2 O formation were measured as described previously (18). These experiments were conducted with a reconstituted system described above using 200 M NADPH instead of an NADPH-generating system.
Anaerobic Experiments-The basic system is as described previously (16,17), with a protocatechuate/protocatechuate dioxygenase oxygen-scrubbing system used in the reduction studies and the cholesterol binding study with the ferrous form of P450 7A1 (27). The scrubbing system was not used in the reactions of ferrous P450 7A1 with oxygen.
Kinetic Analyses and Modeling-The program Dynafit (25) was used for fitting of steady-state binding and activity data and the P450 reaction model. KinTek Explorer software (KinTek Corp., Austin, TX) (28) was used for pre-steady-state binding studies. GraphPad Prism (GraphPad, San Diego, CA) was employed for fitting other data.
Kinetic Deuterium Isotope Effect Study with Human Liver Microsomes-The amount of cholesterol in human liver microsomal samples was first quantified. Human liver microsomes were diluted with H 2 O (0.01 mg protein/ml), and 1.0 ml of each diluted sample was extracted with 2.0 ml of CH 2 Cl 2 , including 1.0 nmol of 7␣-OH cholesterol (as the internal standard). After centrifugation, 1.4 ml of the organic layer was transferred, dried under an N 2 stream, and dissolved in 200 l of CH 2 Cl 2 containing 2 mg of dansyl chloride, 0.5 mg of 4,4-dimethylaminopyridine, and 2 l of triethylamine and incubated at 65°C for 1 h (24). The samples were dried under an N 2 stream and then dissolved in 75 l of CH 3 OH for analysis.
Dansylated products were analyzed with a UPLC system connected to an Acquity fluorescence detector using an Acquity UPLC BEH C18 octadecylsilane column (2.1 ϫ 50 mm, 1.7 m, Waters). LC conditions were as follows. Solvent A contained 95% H 2 O, 5% CH 3 CN, and 0.1% HCO 2 H (v/v/v), and solvent B contained 5% H 2 O, 85% CH 3 CN, 10% tert-butyl methyl ether, and 0.1% HCO 2 H (v/v/v/v). The column was maintained at the initial condition of 50% B (v/v) for 0.25 min with a flow rate of 500 l/min followed by a linear gradient increasing to 100% B over 1.0 min. This condition was maintained for 4.25 min then returned to the initial condition over 0.01 min and maintained until the end of a 7-min run. The column temperature was maintained at 40°C. The injection volume was 7.5 l. Fluorescence measurements were made using an excitation wavelength of 340 nm and an emission wavelength of 525 nm. The data collection and quantitative analysis were conducted with Waters MassLynx version 4.1 and QuanLinx Version 4.1 software, respectively.
Aliquots of individual human liver microsomes including exactly 10 nmol of cholesterol were added to a 0.50-ml of reaction mixture contained 3.1 mM HP␤CD in 50 mM potassium phosphate buffer (pH 7.4). The assays were conducted at 37°C for 30 min. Incubations were initiated by the addition of an NADPH-generating system. After extraction, the derivatization and quantitation steps were conducted as described above.
Immunoquantitation of P450 7A1 in Human Liver Microsomes-The proteins in individual human liver microsomes (75 g of protein per well, in triplicate) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (7.5%, w/v) along with standard samples that included 75 g of human liver microsomal protein and varying amounts of purified P450 7A1 (0 -600 fmol) per well. The proteins were electrophoretically transferred to a polyvinylidene fluoride membrane (Bio-Rad). A commercial primary rabbit polyclonal antibody to human P450 7A1 (ab79847, Abcam, Cambridge, UK) was used (1500-fold dilution in 10 mM potassium phosphate buffer (pH 7.4 containing 150 mM NaCl)). The secondary antibody was goat anti-rabbit IRD800CW, which emits infrared light at 800 nm (detected using an Odyssey Li-Cor instrument (Li-Cor, Lincoln, NE)). The slope of the standard curve (based on the spiked samples) was used to quantify the amount of P450 7A1 in individual human liver microsomal samples.

RESULTS
Cholesterol 7␣-Hydroxylation Activity of P450 7A1-The rates of cholesterol 7␣-hydroxylation activity of P450 7A1 measured in the presence of 0.41 mM Tween 20 were more than 6-fold higher than those measured in the absence of Tween 20 ( Fig. 2A). This result confirmed that Tween 20 is essential for the high catalytic activity under these experimental conditions, as reported previously (29,30). Cholesterol binding to Tween 20 and HP␤CD has been reported with K d values of 2.5 and 1.5 mM, respectively (29), which were used here. All models used to fit the data in this report include these two equilibria (i.e. cholesterol ϩ Tween 20^cholesterol-Tween 20 and cholesterol ϩ HP␤CD^cholesterol-HP␤CD), further assuming that k on values of cholesterol for both compounds are ϳ10 8 M Ϫ1 s Ϫ1 (supplemental Fig. S4A).
The rate of cholesterol 7␣-hydroxylation was dependent on the concentration of NADPH-P450 reductase (Fig. 2B), and with 0.20 M P450 7A1 in the assay a 15-fold greater concentration of the reductase was required for saturation (i.e. 3 M). When the experiment was repeated with 20 nM P450 7A1, the concentration of reductase required for saturation was 1000 nM (50-fold excess) (results not shown). The estimated K d value of the NADPH-P450 reductase-P450 7A1 complex was estimated to be 1.2 Ϯ 0.3 M using a Dynafit analysis of three sets of data (P450 concentrations of 20, 200, and 500 nM P450 7A1), i.e. treating the reductase as a substrate and assuming that steps after binding are not limiting (somewhat over-sim- Steady-state and Pre-steady-state Kinetics of Cholesterol Binding to P450 7A1-Binding of cholesterol to P450 7A1 induced a substrate-type (type I) difference spectrum (31) (Fig. 3, inset). The rates of spectral changes were monitored using stopped-flow spectroscopy (Fig. 4). The traces could not be fit well to a simple one-step binding model (E ϩ L^EL), even when including the steps of cholesterol binding to   Fig. S4 regarding fitting.

JOURNAL OF BIOLOGICAL CHEMISTRY 4635
Tween 20 and HP␤CD (see above) using KinTek Explorer software, although the calculated k on value, 2.3 Ϯ 0.2 ϫ 10 6 M Ϫ1 s Ϫ1 , is acceptable as a diffusion-limited rate constant for interaction of an enzyme and ligand (32). Instead of a onestep model, a two-step binding model (E ϩ L^EL^LE) was used, with a spectroscopically silent first step followed by a second conformational change step associated with the change in the spectrum of P450 7A1 (33-35) (supplemental Fig. S2). This two-step binding model provided a good fit to the curve, and the K d value calculated from the rate constants was 0.51 M (Fig. 4). This model, with the calculated rate constants, also provided a good fit to steady-state titration data, indicating that the model is reasonable for cholesterol binding to P450 7A1 (Fig. 3, supplemental Fig. S4B).
Ferric P450 7A1 Reduction Kinetics-The rate of CO binding to ferrous P450 7A1 was measured. One syringe contained 2.0 M P450 7A1, 120 M di-12:0 GPC, 40 M cholesterol, 6.2 mM HP␤CD, and 0.82 mM Tween 20 in 50 mM potassium phosphate buffer (pH 7.4, reduced with excess solid Na 2 S 2 O 4 immediately before loading), and the other syringe contained CO-saturated (nominally 1000 M CO) 50 mM potassium phosphate buffer (pH 7.4). The rate of the binding of CO to P450 7A1 was estimated at 24 Ϯ 1 s Ϫ1 using GraphPad Prism software (single exponential, data not shown). This CO binding step was incorporated into a model to estimate reduction rates of ferric P450 7A1 (see below).
Ferric P450 7A1 reduction rates were measured in an anaerobic CO environment, with ferrous P450 trapped as the CO complex. The rate of reduction of ferric P450 7A1 was slow in the absence of cholesterol, with a fit to a single exponential of 0.18 Ϯ 0.01 s Ϫ1 (Fig. 5D). In the presence of choles-terol, the trace included fast and slow phases (Fig. 5A). Thus, the parts were fit separately, with k fast ϭ 9.9 Ϯ 0.8 s Ϫ1 and k slow ϭ 0.10 Ϯ 0.01 s Ϫ1 , corrected for the rate of CO binding (see above, Fig. 5, B and C, and supplemental Fig. S4C). About 60% of the P450 7A1 was reduced in the fast phase.
Cholesterol Binding to Ferrous P450 7A1-A difference spectrum between ferrous P450 7A1 and a ferrous P450 7A1cholesterol complex was observed, with a trough at 435 nm and a weak peak at 480 nm (Fig. 6A, supplemental Fig. S2), similar to that reported for P450 2A6 (17). A titration of 1.0 M P450 7A1 (reduced with excess Na 2 S 2 O 4 ) was done with increasing concentrations of cholesterol up to 50 M (data not shown). The data were fit to a one-step binding model with a K d value of 0.18 Ϯ 0.03 M using the program Dynafit. The rate of the change of absorbance at 438 nm was monitored using stopped-flow spectroscopy and could be fit to a singleexponential binding model (Fig. 6B). The rate constants for binding to ferrous P450 7A1 were k ϩ3 ϭ 1.9 ϫ 10 6 M Ϫ1 s Ϫ1 and (for dissociation) k Ϫ3 ϭ 0.30 s Ϫ1 (K d ϭ k -3 /k ϩ3 ϭ 0.16 M), consistent with the K d value estimated by steady-state spectral analysis. Thus, the affinity of cholesterol for the ferric and ferrous forms of P450 7A1 appears to be similar.
Formation and Decomposition of Ferrous P450 7A1-O 2 Complex-A photo-reduced P450 7A1-cholesterol complex was loaded into one syringe of the stopped-flow spectrometer under an anaerobic environment (argon) and mixed with aerobic buffer (nominal final concentration 100 M O 2 ). Rapid changes in the absorbance, particularly at 430 and 550 nm, were detected, in line with previous work with P450s 1A2 and 3A4 (16,36,37) (up to a reaction time of 160 ms, followed by a slower decrease (Fig. 7)). The rate for the first change in ab- sorbance was 29 s Ϫ1 (Fig. 7C, single-exponential) and is considered to be associated with the formation of a ferrous P450 7A1-O 2 complex. The rate for the second phase (decreased absorbance) was 1.2 s Ϫ1 (Fig. 7B, single-exponential). This decrease of absorbance is interpreted to be related to the decomposition of ferrous P450 7A1-O 2 complex to ferric P450 7A1 (Fig. 1).
Kinetic Deuterium Isotope Effects with P450 7A1 (Steadystate)-A non-competitive deuterium isotope experiment comparing the rates of 7␣-hydroxylation of  Table 1). 3 The results clearly show the lack of a significant kinetic deuterium isotope effect for C-H bond breaking in steady-state cholesterol 7␣-hydroxylation activity.
Because of the stereospecific nature of the hydroxylation, the intrinsic kinetic deuterium isotope effect (2) could not be estimated in a simple intramolecular experiment. However, we found that P450 3A4 catalyzed slow 7␣and 7␤-hydroxylation of cholesterol with apparent isotope effects of 6.3 and 11, respectively, indicating that the C-H bond breaking step has an inherently high intrinsic kinetic isotope effect that is not expressed in the P450 7A1 reaction. 4 Product Release from P450 7A1-Binding of 7␣-OH cholesterol to P450 7A1 induced a type I difference spectrum, al-though the amplitude of the absorbance change was small. With the assumption that the K d values of 7␣-OH cholesterol with Tween 20 and HP␤CD are similar to those for cholesterol (see above), the titration data fit to a one-step binding model with a K d value of 3.7 Ϯ 0.4 M (Fig. 9).
No kinetic burst was observed in the P450 7A1 cholesterol 7␣-hydroxylation reaction (Fig. 10). Therefore, this result indicates that any rate-limiting steps do not occur after product formation. Furthermore, in this experiment P450 7A1 was mixed with cholesterol in the initial reaction mixture; i.e. the P450 reaction cycle was started not from the free form of P450 7A1 but from a ferric P450 7A1-cholesterol complex (Fig. 1). This result also indicates that the substrate binding step cannot be a rate-limiting step in that linearity was observed beyond a stoichiometric formation of product (0.5 M).
Formation of Product and Kinetic Deuterium Isotope Effects in Limited Turnover Experiments-Non-competitive deuterium isotope effects were examined when the P450 cycle was started from a Fe 2ϩ -cholesterol complex, with modification of the approach used in Fig. 7. P450 7A1 (5 nmol) was either photo-reduced or reduced by NADPH-P450 reductase (5 nmol) under an argon atmosphere. In the cases in which the reductase was included, a limited amount of NADPH (12.5 nmol, i.e. enough to fully reduce the reductase and P450 for a single P450 cycle) was added. The reduction of ferric P450 7A1 was confirmed by monitoring the spectral changes. After reduction, the solutions were mixed with O 2 buffer, and the amount of product formed was measured (LC-MS).
7␣-OH cholesterol was detected in the experiment with only (photo-reduced) P450 present ( Table 2). As reported previously with other P450s (16, 17), the product must result . Product formation was more efficient when electrons were transferred from the reductase to P450, and the yield of d 1 -7␣-OH cholesterol in the system was similar to that for d 0 -7␣-OH cholesterol. A replicate experiment yielded a similar lack of a kinetic deuterium isotope effect. These results suggest that the rate of C-H bond breaking step is faster than the rates of O 2 binding, the second electron transfer, and the O 2 activation steps (Fig. 1). 5 Stoichiometry of NADPH Utilization-Rates of NADPH oxidation, product formation, and H 2 O 2 formation were measured at a single concentration of the substrate cholesterol (20 M) ( Table 3). The H 2 O formation rate was calculated by difference (38). Under these experimental conditions, 65% of the NADPH that was oxidized was used to form the product 7␣-OH cholesterol, which is a relatively high value compared with other microsomal P450s (17,33,36). These data were used in the kinetic modeling (see below).
P450 Spectra under Steady-state Reaction Conditions-One approach to discerning what step(s) in a P450 reaction is ratelimiting is to record UV-visible spectra in the steady-state reaction and observe which electronic state of P450 is domi- 5 The experiments done in the absence of the reductase (photochemical reduction) did show a kinetic isotope effect, but this system was considered less relevant to the normal reaction (and had a much lower yield).    Table 1 for parameters.  (16,17,39). We performed such an experiment in this case (Fig. 11). Because a high P450 concentration was required to observe spectra and the cholesterol 7␣-hydroxyla-tion is very rapid, we used the rapid-scanning mode of the stopped-flow spectrophotometer. Although some changes between Fig. 11 A and B were observed (mainly due to the reduction of the flavins of the NADPH-P450 reductase), the magnitude of the Soret band and the characteristic ␣,␤-band doublet were preserved in the steady-state spectra, as seen in the 1-and 20-s spectra (see supplemental Fig. S2 for comparisons of the Fe 3ϩ and Fe 2ϩ forms; the steady-state spectra are also not consistent with the spectra expected for the FeO 2 2ϩ complex, based on other work; i.e. Fig. 7).
Kinetic Modeling-All steps shown in the general P450 reaction cycle (Fig. 1) were incorporated into a minimal kinetic model for the cholesterol 7␣-hydroxylation reaction (Fig. 12). The model included the following phenomena: (i) the twostep binding of substrate to ferric P450 (Figs. 3, 4), (ii) irreversible decomposition of the ferrous P450-O 2 -substrate      complex, (iii) irreversible loss of reduced oxygen species from the activated complex (O 2 . , H 2 O 2 , H 2 O), (iv) the lack of burst kinetics ( Fig. 10), (v) the lack of a non-competitive kinetic deuterium isotope effect ( Fig. 8 and Table 1), and (vi) the low kinetic deuterium isotope effect when the reaction was started from the ferrous P450-substrate complex ( Table 2). The program Dynafit was used to fit plots of 7␣-OH cholesterol, H 2 O 2 , and H 2 O formation rates (single concentration points were used in the cases of the two reduced oxygen species) (supplemental Fig. S3D). The rate constants for cholesterol binding to ferric P450 (k ϩ3 , k -3 , k ϩ4 , and k Ϫ4 ; Figs. 3 and 4), the first electron transfer (k ϩ5 , Fig. 5), cholesterol binding to ferrous P450 (k ϩ7 and k Ϫ7 , Fig. 6), and the decomposition of the ferrous P450-O 2 -substrate complex (k ϩ8 , Fig. 7) were used directly from the estimated rates obtained with experimental data.
In the modeling, a k Ϫ6 value of 1 s Ϫ1 was assumed to keep the k ϩ6 /k Ϫ6 value in the low micromolar range (40). An estimate of k ϩ6 was calculated by dividing the formation rate constant for the ferrous P450-O 2 -cholesterol complex (Fig.  7B) by the O 2 concentration, 100 M. The values of k ϩ9 (the second electron transfer), k ϩ10 (oxygen activation), k ϩ11 (formation of H 2 O 2 from the activated complex), k ϩ12 (formation of H 2 O from the activated complex), and k ϩ13 (the C-H bond breaking) were estimated by fitting plots of the data. k ϩ10 and k ϩ13 were set as Ն100 s Ϫ1 for fitting. The value of k Ϫ14 (k on for product binding) was optimized at 2.5 ϫ 10 7 M Ϫ1 s Ϫ1 (Table 4) and could not be decreased without compromising the fit to the data. Using a K d of 3.7 M for 7␣-OH cholesterol binding to P450 7A1 (Fig. 9), a value for k ϩ14 was calculated.
The final fits for 7␣-OH cholesterol, H 2 O 2 , and H 2 O formation are shown in Fig. 13 using the rate constants presented in Table 4. The predicted curves of product formation rates fit the experimental data well without changing any of the rate constants estimated using the experimental results. Among the forward rate constants in the P450 reaction cycle producing 7␣-OH cholesterol (i.e. k ϩ3 , k ϩ4 , k ϩ5 , k ϩ6 , k ϩ9 , k ϩ10 , k ϩ13 , and k ϩ14 ), the first electron transfer rate constant (k ϩ5 ) showed the smallest value.
Kinetic Deuterium Isotope Effects on Cholesterol 7␣-Hydroxylation Activity in Human Liver Microsomes-Human liver microsomes contain endogenous cholesterol (12). Cho-lesterol 7␣-hydroxylation activity was measured in individual human liver microsomal samples using the endogenous cholesterol as substrate, adjusting the concentration of cholesterol to 20 M by dilution ( Table 5).
The amount of P450 7A1 in each liver sample was estimated by immunoblotting and varied ϳ3-fold among the 10 liver samples examined. The catalytic activity (per nmol P450 7A1) varied ϳ5-fold, from 0.66 to 3.2 nmol of product formed/min/nmol of P450 7A1. The mean value, 1.4 min Ϫ1 (0.023 s Ϫ1 ), was Ͻ1/100 the rate measured in the reconstituted system (Fig. 2).
Competitive deuterium isotope effects were also examined by changing the percentage of d 2 -cholesterol in the total cholesterol concentration (100 M). The percentage of d 1 -7␣-OH cholesterol formed increased linearly with the increase of the percentage of d 2 -cholesterol used as substrate, with a slope of 1.1 (Fig. 14). This result indicates a lack of a kinetic deuterium isotope effect on 7␣-hydroxylation activity in human liver microsomes.

DISCUSSION
The purpose of this study was to characterize the individual steps involved in cholesterol 7␣-hydroxylation, a reaction catalyzed by P450 7A1 at a much higher rate than those of other mammalian P450s. The catalytic efficiency of the reaction is ϳ2.4 ϫ 10 6 M Ϫ1 s Ϫ1 (Fig. 2), which can be compared with a typical P450 1A2 reaction, phenacetin O-deethylation at 2.3 ϫ  Table 4. See supplemental Fig. S4 regarding fitting.
-hydroxylation reaction at 5.7 ϫ 10 4 M Ϫ1 s Ϫ1 (15,41). Previous reports on the low efficiency reactions of P450s 1A2, 2A6, 2D6, and 3A4 indicate that the C-H bond breaking step of each reaction is mainly or partially rate-limiting, as judged by kinetic deuterium isotope effects (15)(16)(17)(18). Our results revealed that the substrate binding, C-H bond breaking, and product release steps are not rate-limiting and suggest that the first electron transfer step is the rate-limiting step in this high efficiency reaction. Although some of the older literature suggest that rate-limiting first-electron transfer is a general case (42,43), the evidence argues against this view for most mammalian P450s in that high non-competitive kinetic deuterium isotope effects are rather common (15)(16)(17)(18). In both early work with rat liver microsomes (42,43) and later work with human liver microsomes (44), the rate of the fast phase of reduction of (total) P450 was ϳ0.7 s Ϫ1 (41 min Ϫ1 ). This rate is considerably higher than the rates of most P450catalyzed oxidations of xenobiotic chemicals.
In our experimental conditions using reconstituted systems containing purified P450 7A1 and NADPH-P450 reductase, we used Tween 20 and HP␤CD for delivering the substrate cholesterol and obtained high catalytic activities for P450 7A1 ( Fig. 2A). Cholesterol binds to both Tween 20 and HP␤CD (45). Thus, all models used to fit the data in this report included two additional steps (i.e. cholesterol ϩ Tween 20ĉ holesterol-Tween 20 and cholesterol ϩ HP␤CD^cholesterol-HP␤CD). The kinetic traces of spectral changes of P450 7A1 followed by binding of cholesterol could be well fit to a two-step binding model (E ϩ L^EL^LE), with a K d value of 0.51 M, as shown previously with some other P450s (Fig.  4, supplemental Fig. S5) (33-35). Interestingly, when the lipid (di-12:0 GPC) was removed from the reaction mixture, the observed rates (k obs ) were Ͼ3-fold lower than those with di-12:0 GPC (results not shown). A previous report indicated the importance of some amino acid residues in the membrane binding domain of P450 7A1 for cholesterol binding and substrate specificity (9). Our results also suggest that the interaction between P450 7A1 and a lipid membrane is important for rapid binding of cholesterol.
A rapid reduction rate was obtained in the presence of cholesterol, although only ϳ60% of the P450 7A1 was reduced in the fast phase (9.9 s Ϫ1 at 37°C) (Fig. 5, supplemental Fig.   S4C). This reduction rate was comparable with those of several other mammalian P450s (16,17,33,44).
The catalytic efficiency of a reaction can be expressed as the second-order rate constant for binding multiplied by the efficiency of productive catalytic events. The efficiency here is high (2.4 ϫ 10 6 M Ϫ1 s Ϫ1 , Fig. 2A) but the reaction is clearly not diffusion-limited, as shown by the evidence against substrate binding being rate-limiting (Fig. 10).
The work on kinetic deuterium isotope effects was done with 7,7-d 2 -cholesterol. We did not utilize the individually labeled 7-d isomers (7␣, 7␤) in our work. However, in vivo studies with rats by Corey and Gregoriou (46) and Bergstrom et al. (26) have shown that only the 7␣-hydrogen atom is abstracted. We presume that the same course is probably followed in humans, i.e. that hydrogen abstraction and oxygen rebound both occur from the ␣-face. The potential for a contributing geminal secondary kinetic deuterium isotope effect (due to the use of 7,7-d 2 -cholesterol) in our work is not an issue, in that no isotope effects were observed (Table 1, Fig. 8).
The lack of a kinetic deuterium isotope effect (Fig. 8) and kinetic burst when the reaction cycle was initiated from the ferric P450 7A1-cholesterol complex (Fig. 10) clearly shows that the substrate binding, C-H bond breaking, and productrelease steps are not rate-limiting. Therefore, we considered the possibility that the rate-limiting step could be the first electron transfer. The deuterium isotope effect was measured in an experiment in which the reaction cycle was started from the ferrous P450 7A1-cholesterol complex, namely just after the first electron transfer (Table 2). No kinetic isotope effect was observed, indicating that the C-H bond breaking step is faster than the rates of O 2 binding, the second electron transfer, and the O 2 activation steps (Fig. 1). A minimal kinetic model (Fig. 12, supplemental Fig. S4D) for the cholesterol 7␣-hydroxylation reaction was developed, including all steps shown in the general P450 reaction (Fig. 1). The predicted progress curves for 7␣-OH cholesterol, H 2 O 2 , and H 2 O formation fit the experimental data well without changing any of the rate constants estimated using the experimental results (Fig. 13, Table 4). The value of k ϩ13 (C-H bond breaking) was greater than those of k ϩ6 (O 2 binding to ferrous P450-cholesterol complex), k ϩ9 (the second electron transfer), and k ϩ10 (oxygen activation), consistent with the kinetic deuterium isotope effect results in the limited turnover experiment. The ratio of k ϩ13 and k ϩ12 was comparable with the ratio of the 7␣-OH cholesterol and H 2 O formation rates (Tables 3 and 4). Among the forward rate constants in the P450 reaction cycle producing 7␣-OH cholesterol (i.e. k ϩ3 , k ϩ4 , k ϩ5 , k ϩ6 , k ϩ9 , k ϩ10 , k ϩ13 , and k ϩ14 ), the first electron transfer rate constant (k ϩ5 ) showed the smallest value. These results indicate that the first electron transfer step is rate-limiting. However, a more appropriate view of the catalytic cycle is not to conclude that a single step is rate-limiting but that the overall reaction rate reflects the contribution of rate constants of back reactions and side reactions (47).
A typical ratio of NADPH-P450 reductase and P450 in reconstituted P450 systems is ϳ2:1 for maximum catalytic activity (16,17,48,49). However, the ratios of NADPH-P450 reductase and P450 7A1 (50:1 for steady-state enzyme kinetic work and 15:1 for burst kinetic analysis) used in this report were much higher (Fig. 2B), consistent with our conclusion that the rate of reduction limits the overall reaction both in the reconstituted system and in microsomal membranes. For other mammalian P450s the C-H bond breaking step is often rate-limiting in steady-state turnover (16,17), and the "saturation" of a P450 by NADPH-P450 reductase leads to an approach to a limiting steady-state rate (imposed by the C-H bond breaking step). On the other hand, the increase of the reduction capacity in the reaction system enhances the catalytic activity of P450 7A1 because the rate-limiting step of the P450 7A1 reaction may be the first electron transfer step.
Competitive kinetic deuterium isotope effect experiments on cholesterol 7␣-hydroxylation activity were also conducted in human liver microsomes. The ratio of the total amounts of P450 and NADPH-P450 reductase in human liver microsomes has been reported to be ϳ20:1 (50). If the rate-limiting step in the cholesterol 7␣-hydroxylation reaction is the first electron transfer step, then no isotope effect should be observed due to the very limited capacity for reduction in human liver microsomes. The percentage of formation of (7␤d 1 )7␣-OH cholesterol increased proportionately with the increase of the percentage of d 2 -cholesterol added, with a slope of 1.1 (Fig. 13), indicating the lack of an isotope effect on 7␣-hydroxylation activity in human liver microsomes. This result supports the proposal that the rate-limiting step is the first electron transfer step in the microsomal membranes as well as the reconstituted system. The high catalytic efficiency of P450 7A1, once reduced, is consistent with evidence for high binding selectivity for cholesterol in the mode for 7␣hydroxylation (45), leaving the rate of the reaction dependent upon reduction (Fig. 2B). However, it was noted (45) that even this fit is precarious in that the P450 7A1 mutant A358V formed 7␤-OH cholesterol and an unidentified product, and we have recently determined that 7-dehydrocholesterol, the immediate precursor of cholesterol, is also a substrate for P450 7A1. 6 In summary, we evaluated the kinetic mechanism of human P450 7A1-catalyzed cholesterol 7␣-hydroxylation, measuring rate constants of individual steps and kinetic deuterium isotope effects. A minimal kinetic model for the P450 7A1 reaction indicates that the first electron transfer step is rate-limiting and that this is a clearly different phenomenon compared with other P450s that have much lower rates of catalysis.