Introduction

Top Abstract Introduction Experimental Procedures. Results Discussion References

Midazolam (MDZ) is one of the most commonly used drugs for sedation in emergency rooms (see Nordt and Clark, 1997 and references therein). It is also used as a safe and effective drug for the treatment of generalized seizure, status epilepticus, and acute agitation. The biotransformation of MDZ is mediated by at least three different CYP3A enzymes: 3A4, 3A5, and 3A7 (Gorski et al., 1994; Kuehl et al., 2001). Although CYP3A7 is predominantly expressed in fetal tissues, CYP3A4 and CYP3A5 represent the majority of the total hepatic and intestinal P450 content in adults (Guengerich, 1995). These enzymes are of particular clinical significance because of their ability to metabolize a large number of therapeutic agents of very diverse structures (Guengerich, 1995). Moreover, intestinal CYP3A accounts for significant first-pass metabolism of ingested drugs. Because of the large number of therapeutic agents that alter CYP3A expression or activity, a significant potential for drug-drug interactions exists (Fuhr et al., 1996). In particular, CYP3A4 is known to exhibit both homotropic and heterotropic cooperativity, which could influence drug metabolism and excretion or bioactivation (Schwab et al., 1988; Shou et al., 1994, 2001; Harlow and Halpert 1997, 1998; Domanski et al., 1998, 2000; Korzekwa et al., 1998).

In recent years, MDZ has emerged as one of the best in vivo probes for prediction of CYP3A activity (Thummel and Wilkinson, 1998). MDZ can be administrated both orally and intravenously, which can provide a measure of CYP3A activity relative to intestinal and hepatic metabolism, respectively. Furthermore, according to in vitro studies, MDZ is not subject to P-glycoprotein-mediated transport across the intestinal epithelium (Kim et al., 1999). Additionally, a difference in the regioselectivity of MDZ metabolism at lower concentrations can be used to discriminate among individual subjects with or without CYP3A5, because CYP3A5 shows a much higher 1'-OH/4-OH ratio of MDZ metabolism than CYP3A4 (Gorski et al., 1994; Kuehl et al., 2001).

The oxidation of MDZ by CYP3A4 has been the focus of many in vitro investigations (Kronbach et al., 1989; Gorski et al., 1994; Ghosal et al., 1996; Maenpaa et al., 1998; Hosea et al., 2000; Wang et al., 2000). These studies have found very different K_M values for the formation of 1'-OH MDZ compared with 4-OH MDZ by CYP3A4. Production of the two metabolites has also been reported to be stimulated/inhibited differentially by various compounds. Thus, whereas the presence of ANF stimulated 1'-OH MDZ formation, 4-OH MDZ formation was decreased or unaltered. In contrast, testosterone increased 4-OH MDZ and decreased 1'-OH MDZ formation. A recent study by Hosea et al. (2000) also reported two distinct K_i values for inhibition of 1'-OH and 4-OH MDZ formation by a peptide (YPFP-NH₂) shown to interact with CYP3A4. These and other studies have led to the suggestion that MDZ binds at two different locations in the CYP3A4 active site (Ghosal et al., 1996; Hosea et al., 2000).

In recent years, understanding structure-function relationships of CYP3A4 has been a major focus of our laboratory (see Domanski and Halpert 2001 and references therein). A sequence alignment with bacterial P450s of known structure was used to localize putative SRSs (Gotoh, 1992) within CYP3A4 (Szklarz and Halpert, 1997) based on a similar successful approach with P450 family 2 enzymes. Site-directed mutagenesis, homology modeling, and functional analysis using substrates such as progesterone, testosterone, AFB1, RPR 106541, 7-alkoxycoumarins, ANF, and 7-benzyloxy-4-trifluoromethylcoumarin led to the identification of a number of SRS residues that are involved in determining substrate specificity. These studies have also provided strong evidence in support of the hypothesis that there are multiple substrate binding sites in CYP3A4 (Shou et al., 1994; Korzekwa et al., 1998). In the present study, we have investigated MDZ metabolism by CYP3A4 wild-type and 30 mutants generated at 15 different SRS positions to systematically evaluate the role of various SRS residues in substrate oxidation and the possible existence of multiple MDZ binding sites. The study showed that MDZ metabolism causes CYP3A4 inactivation and suggests that such enzyme inactivation is related to the 1'-OH MDZ metabolic pathway. The SRS residues at which amino acid substitution showed the most significant effect on regioselectivity of MDZ metabolism were Phe-108, Ser-119, Ile-120, Leu-210, Ile-301, Phe-304, Ala-305, Tyr-307, Thr-309, Leu-373, and Leu-479. Correlational analysis of 1'-OH formation versus 4-OH formation and analysis of the effect of side chain size on the metabolite profile by all mutants, along with steady-state kinetic analyses and substrate binding studies of selected mutants suggest that the two putative MDZ binding sites in CYP3A4 are near each other.

	Experimental Procedures.

Top Abstract Introduction Experimental Procedures. Results Discussion References

Materials. MDZ, flunitrazepam, NADPH, CHAPS, and DOPC were purchased from Sigma Chemical Co. (St. Louis, MO). HEPES was obtained from CalBiochem Corp. (La Jolla, CA). The Expand PCR Kit and Rapid Ligation kits were obtained from Roche (Indianapolis, IN). The QuickChange site-directed mutagenesis kit and GeneClean kit were from Stratagene (La Jolla, CA) and BIO 101 (Carlsbad, CA), respectively. The bovine serum albumin protein assay kit was purchased from Pierce (Rockford, IL). Thin-layer chromatography plates [silica gel, 250 µm; Si 250F (C19)] were purchased from J. T. Baker, Inc. (Phillipsburg, NJ). 1'-OH and 4-OH MDZ were gifts from Dr J. C. Stevens (Pharmacia, Kalamazoo, MI). Recombinant NADPH-cytochrome P450 reductase and cytochrome b₅ from rat liver were prepared as described earlier (Harlow and Halpert, 1997). All other chemicals were of the highest grade available and were obtained from standard commercial sources.

Mutant Construction. Mutants P107A, P107W, and F108A were generated with the megaprimer method. In the first amplification reaction, the mutagenic primer (see Fig. 1) and a pSE380 3'-specific primer were used with pSE3A4His as the template (Domanski et al., 1998). The product was used in a second amplification as a primer in conjunction with a pSE380 5'-specific primer and pSE3A4His again as the template. To increase the amount of product for further cloning, the product of the second PCR was used in a third reaction as the template with the same primers used in round 2. Both pSE3A4His and the final products were digested with NcoI and BamHI, the desired bands were purified with GeneClean (Bio101, Vista, CA), ligated together, and transformed into E. coli DH5 cells. Because the P107A, P107W, and F108A mutagenic primers contained alterations that removed a StuI site, clones were screened for the absence of this site.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Primers used for generation of CYP3A4 cDNA mutants. The mutated nucleotides are underlined. The generations of unique restriction sites due to these mutations are represented in bold (PstI for I120A and Eco47III for L479A). The deletion of a StuI site due to the mutations (in the case of P107A, P107W, and F108A) is represented in italics.

Expression and Purification of CYP3A4 and Mutants. Wild-type CYP3A4, as well as newly and previously generated mutants, were expressed as His-tagged proteins in Escherichia coli TOPP3 (or DH5) cells and purified using Talon metal affinity resin (CLONTECH, Palo Alto, CA) as described earlier (Harlow and Halpert, 1997, 1998; He et al., 1997; Domanski et al., 1998, 2000, 2001; Wang et al., 1998; Stevens et al., 1999; Khan and Halpert, 2000; Roussel et al., 2000). The P450 content was determined by carbon monoxide difference spectra in the presence of 1% Triton X-100 added to the protein sample before dilution with microsome solubilization buffer containing 100 mM potassium phosphate, pH 7.3, 20% glycerol, 0.5% sodium cholate, 0.4% Renex, and 1.0 mM EDTA. Total protein content in each sample was determined with the bicinchoninic acid protein assay kit.

MDZ Hydroxylase Assay. The reconstituted system for the assay contained 10 pmol of purified P450, 20 pmol of rat liver cytochrome b₅, 40 pmol of recombinant NADPH-cytochrome P450 reductase, 0.04% CHAPS, and 1 mg/ml of DOPC (Harlow and Halpert, 1997). The mixture was preincubated for 10 min at room temperature. MDZ dissolved in methanol was added to the reconstituted protein system in 50 mM HEPES buffer, pH 7.6, and 15 mM MgCl₂. The protein-substrate mixture was further incubated for 5 min at 37°C, and the reaction was initiated by adding NADPH (1 mM final concentration). The total reaction volume of the assay was 100 µl. After 5 min of incubation at 37°C, the reactions were stopped by addition of 100 µl of methanol containing 5 nmol of flunitrazepam as an internal standard. For all mutants as well as wild-type, the assay was carried out at least twice to confirm the validity of the metabolic profiles.

HPLC Analysis. The HPLC system consisted of two Beckman 110 solvent delivery modules, a Beckman 421A system controller (Beckman, Berkeley, CA), a 50-µl injection loop, a spectroflow 757 UV-absorbance detector (Kratos Analytical, Ramsey, NJ), and a Spectra-Physics SP4270 Integrator (Spectra-Physics, Piscataway, NJ). An ultrasphere ODS column (5 µm × 250 mm × 4.6 mm; Beckman Coulter, Fullerton, CA) was used with an ultrasphere C18 guard column (5 µm × 7.5 mm × 4.6 mm; Alltech, Deerfield, IL). The separation of the metabolites of MDZ was achieved isocratically using the mobile phase (methanol/phosphate buffer, pH 7.4/ tetrahydrofuran, 52:46:2). The flow rate was 1.0 ml/min and the UV detector was set at 230 nm. All chromatographic separations were performed at room temperature.

Inactivation Assays. The reconstitution conditions were essentially the same as described above. The preincubation mixture contained 35 pmol of purified P450, 70 pmol of rat liver cytochrome b₅, 140 pmol of recombinant NADPH-cytochrome P450 reductase, 0.04% CHAPS, and 1 mg/ml of DOPC. After incubating for 10 min at room temperature, MDZ was added to the reconstituted protein system in 50 mM HEPES buffer, pH 7.6, and 15 mM MgCl₂. The protein-substrate mixture was further incubated for 5 min at 37°C, and the reaction was initiated by adding NADPH (1 mM final concentration). The total volume of the mixture was 560 µl. Aliquots (80 µl) of this mixture were taken at various time intervals (0-6 min) and added to a secondary reaction mixture (20 µl) for determination of the residual progesterone hydroxylase activity ([progesterone] = 25 µM). The reaction was stopped after 6 min by the addition of 50 µl of tetrahydrofuran, and 50 µl of the reaction mixture was spotted on the preadsorbent loading zone of a thin-layer chromatography plate (Baker silica gel; 250 ml, Si 250F). The plate was developed twice in benzene/ethyl acetate/acetone [10:1:1 (v/v/v)]. Metabolites were visualized by autoradiography and identified by comparison with unlabeled standards. The radioactive areas from the plate were scraped into scintillation vials, and the metabolites were quantified by liquid scintillation counting.

Spectral Binding Studies. Binding spectra were recorded on a Shimadzu-2600 spectrophotometer fitted with a temperature controller (TCC-240A). A solution (0.8 ml) containing 0.5 µM protein in 100 mM phosphate, pH 7.4, was divided into two quartz cuvettes (10-mm path length) and a baseline was recorded between 350 and 500 nm. An aliquot of substrate in methanol was then added to the sample cuvette, and the same amount of methanol was added to the reference cuvette. The difference spectra were obtained after the system reached equilibrium (3 min). All the spectra were recorded at 37°C.

	Results

Top Abstract Introduction Experimental Procedures. Results Discussion References

Inactivation of CYP3A4 by MDZ. The oxidation of MDZ by CYP3A4 generated two hydroxylated products, 1'-OH MDZ and 4-OH MDZ, as reported earlier (Gorski et al., 1994). The reaction was found to be nonlinear with respect to time (Fig. 2A), suggesting that MDZ might be inactivating the enzyme, as reported previously (Podoll et al., 1995; Schrag and Wienkers, 2001). The nonlinear behavior of MDZ metabolism by CYP3A4 could be fitted to a first-order exponential rate equation, r = R_max × exp(k_obs × t), where k_obs represents the rate constant, and R and R_max are the nanomoles of product formed per nanomole P450 at a particular time (t) and at infinity, respectively. At 250 µM MDZ, the values of R_max and k_obs for 4-OH MDZ were 51.4 ± 1.4 nmol/nmol P450 and 0.19 ± 0.01 min¹, respectively, and for 1'-OH MDZ, they were 51.0 ± 1.5 nmol/nmol P450 and 0.19 ± 0.02 min¹, respectively. If the decrease in the rate of metabolism of MDZ over time is caused by the inactivation of the enzyme, the k_obs determined by this fit should be equal to the rate constant of inactivation at the same MDZ concentration. To further investigate this possibility, the kinetics of CYP3A4 inactivation by MDZ was studied by measuring the time-dependent decrease in progesterone 6-hydroxylation. The enzyme was inactivated by MDZ in a time- and concentration-dependent manner (Fig. 2B). The inactivation followed pseudo-first-order kinetics (Fig. 2B), and was saturable with MDZ (Fig. 2C). The kinetic constants were determined from the fit of the rate constants of inactivation versus MDZ concentration to the Michaelis-Menten equation. The concentration of inactivator required for half-maximal inactivation (K_I) was 5.8 ± 0.3 µM, and the maximal rate constant of inactivation at saturation (k_inact) was 0.15 ± 0.00 min¹. Thus, the k_inact determined above showed good agreement with the k_obs obtained from the fit of the time dependence of MDZ metabolism by the enzyme at saturating substrate (see next paragraph).

View larger version (20K): [in this window] [in a new window]   Fig. 2.   A, the time dependence of MDZ (250 µM) hydroxylation by CYP3A4. The details of the reaction are described under Experimental Procedures.  and  represent 4-OH and 1'-OH MDZ, respectively. The solid lines through the experimental data points show the fit to the first-order exponential rate equation, r = Rmax × exp(kobs × t) (see Results for detail). B, time- and concentration-dependent inactivation of the wild-type enzyme by MDZ as measured by the percentage decrease in progesterone 6-hydroxylation. The details of the inactivation reactions are described under Experimental Procedures. The concentrations of MDZ were 0 (), 10 (), 25 (), 100 (), and 250 () µM. The lines shown through experimental points were generated by linear regression analysis of the natural logarithm of the residual activity as a function of time. Rate constants of inactivation (ki) are derived from the negative slope of the lines, whereas the extent of reversible inhibition is reflected by a decrease in the extrapolated activity at zero preincubation time compared with the methanol control. C, double-reciprocal plot of the rate constants of inactivation as a function of MDZ concentration.

MDZ Metabolism by Wild-Type Enzyme. Despite the inactivation of CYP3A4 by MDZ, analysis of the protein concentration dependence of MDZ metabolism (5-min incubation) showed that the reaction was linear between 25 and 125 nM P450 (2.5-12.5 pmol in 100-µl reaction volume; data not shown). Therefore, all subsequent reactions were carried out using 100 nM enzyme and a 5-min incubation time unless stated otherwise. Under these conditions, analysis of steady-state kinetics of MDZ oxidation by the wild-type enzyme gave a hyperbolic V versus S plot for both products (data not shown). Kinetic parameters determined from the fit to the Michaelis-Menten equation showed two very distinct K_M values (3.7 ± 0.9 and 64 ± 4 µM for 1'-OH and 4-OH MDZ, respectively; Table 1), as reported by others (Gorski et al., 1994; Ghosal et al., 1996; Hosea et al., 2000).

Effect of SRS Substitutions. Recent studies from our laboratory involving the use of site-directed mutagenesis based on structure-based sequence alignments have proven informative in understanding the structure-function relationships of CYP3A4 (Harlow and Halpert, 1997, 1998; He et al., 1997; Domanski et al., 1998, 2000, 2001; Wang et al., 1998; Stevens et al., 1999; Khan and Halpert, 2000; Roussel et al., 2000, Xue et al., 2001). These mutagenesis studies were the first to establish that Gotoh's SRS model (Gotoh, 1992) is also applicable to the CYP3A subfamily and have identified counterparts to all the active-site residues inferred from the X-ray crystal structure of rabbit CYP2C5 (Williams et al., 2000; reviewed in Domanski and Halpert, 2001). Here, we systematically evaluated SRS residues with the goals of elucidating the role of these residues in MDZ regioselectivity and of mapping the two proposed MDZ binding sites (Ghosal et al., 1996; Hosea et al., 2000). For this purpose, 30 mutants generated at 15 different SRS positions, which outline almost the entire putative CYP3A4 active site, were analyzed for MDZ oxidation. All 15 SRS residues selected for the study have been either previously shown to play an important role in the oxidation of one or more CYP3A4 substrates (Domanski and Halpert, 2001) or were selected based on analogy with the active-sites in the CYP2C5 or P450eryF crystal structures (Cupp-Vickery et al., 2000; Williams et al., 2000). Furthermore, unlike some of the other P450s, CYP3A4 does not show any special charge requirement in the active site for the various substrates it metabolizes [see Khan and Halpert (2000) and references therein]. Therefore, to maximize the effect of side chain substitutions, for each selected SRS position one larger and one smaller residue was chosen. Because of the large difference in the two K_M values for MDZ metabolism by CYP3A4 and the ensuing concentration dependence of the metabolite profile, metabolism of MDZ by the mutants was studied at two different concentrations, 25 and 250 µM (Fig. 3). MDZ metabolism by CYP3A4 wild-type showed a decrease of more than 3-fold in the ratio of 1'-OH MDZ to 4-OH MDZ between the two different concentrations.

View larger version (49K): [in this window] [in a new window]   Fig. 3.   The metabolite profiles of the wild-type and various CYP3A4 SRS mutants at 25 µM (A) and 250 µM (B) MDZ. Assay conditions are described under Experimental Procedures. The error bars represent average deviation from duplicate incubations. C, the ratio of 1'-OH/4-OH product formation at both concentrations.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Midazolam hydroxylation kinetics by S119A, I120W, F304A, and F304W.  and  represent 4-OH and 1'-OH MDZ, respectively. The solid lines through experimental data points show the fit to the Michaelis-Menten equation, and the dotted lines through the I120W and F304W data points show the fit to a modified Michaelis-Menten kinetics with uncompetitive substrate inhibition (Perloff et al., 2000).

Correlational Analysis. To assess whether the two hydroxylated MDZ products are produced from binding of the substrate in the same or two different pockets, a correlational analysis of 1'-OH and 4-OH MDZ formation by all mutants was performed. It is well known that the mutation of a single residue could selectively increase or decrease the formation rate of either or both metabolites. However, analysis of a large number of residues that span almost the entire putative CYP3A4 active site was expected to show a strong correlation if MDZ were to bind at a single site, because the effect of mutation of a particular residue should be similar on formation of both metabolites in most cases. We also reasoned that if the two metabolites resulted from MDZ binding at two distinct sites, the effect of mutation of various residues would not be the same for both metabolites, and a poor correlation would result. This should be specially valid in the present study, considering that all 15 residues, with the exception of Tyr-307, are thought to directly affect oxidation of a number of CYP3A4 substrates (Khan and Halpert, 2000; Domanski and Halpert, 2001). The correlational analysis of 1'-OH and 4-OH MDZ formation by all mutants showed r² values of 0.015 and 0.21 at 25 and 250 µM MDZ, respectively (Fig. 5). These low r² values are inconsistent with the existence of a single MDZ binding site in the CYP3A4 active site.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   A plot showing the correlational analysis of 4-OH formation versus 1'-OH formation at 25 µM (A) and 250 µM MDZ (B). The r2 values were determined from linear regression of all 30 mutants.

Spectral Binding Studies. A recent study by Hosea et al. (2000) found that the MDZ binding dissociation constant (K_D) is very similar to the K_M value for 1'-OH MDZ formation, indicating that the spectral change caused by enzyme-substrate complex formation is induced mainly by the binding of MDZ in the 1'-OH orientation. To test this, dissociation constants and the maximal absorbance (A_max) change due to MDZ binding were determined for a few selected mutants and wild-type CYP3A4 (Fig. 6 and Table 3). Interestingly, I120W and T309F, which predominantly form 1'-OH MDZ, show a A_max very similar to the wild-type. In contrast, I369W, which mainly forms 4-OH MDZ, shows approximately one-fourth of the A_max of the wild-type. Furthermore, although T309F and I369W have very similar dissociation constants, I369W showed only one-third of the A_max of T309F. In brief, although these binding studies agree with the suggestion that MDZ binding in the 1'-OH orientation is mainly responsible for the spectral changes upon enzyme-substrate complex formation, MDZ binding in the 4-OH orientation also leads to a small spin-state change in some cases.

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Spectral titration of T309F and I369W with increasing concentrations of MDZ (a) and the plots of change in absorbance versus [MDZ] (b). The solid lines through the experimental data points show the fit to the equation A = AmaxS/(KD + S), where KD is the dissociation constant, and A and Amax are the change in absorbance at a particular substrate concentration (S) and at a saturating substrate concentration, respectively. The MDZ concentration as well as absorbance change have been adjusted for the dilution caused by the addition of MDZ. The protein concentration was 0.5 µM in both cases.

View this table: [in this window] [in a new window]   TABLE 3 Midazolam dissociation constant, maximal absorbance change and percent spin-state change on enzyme-substrate complex formation for the wild-type CYP3A4 and selected mutants. An extinction coefficient of 126 mM1 cm1 has been used to calculate the percentage spin change due to MDZ binding with the enzyme (Harlow and Halpert, 1998). The values in parentheses show the average deviation obtained from the fit of the equation A = AmaxS/KD + S, where KD is the dissociation constant and A and Amax are the change in absorbance at a particular substrate concentration (S) and at saturating substrate concentration, respectively.

The Possible Pathway of Inactivation. One of the most likely causes of CYP3A4 inactivation by MDZ is the irreversible binding to the enzyme of one or more reactive intermediates (Podoll et al., 1995; Schrag and Wienkers, 2001). Thus, to investigate whether one or both of the metabolite pathways leads to CYP3A4 inactivation, three mutants that predominantly form only one of the products were chosen for further analysis. Mutants F304W and T309F mainly form 1'-OH MDZ, while I369W produces 4-OH MDZ preferentially (Table 1). The time dependence of MDZ metabolism by F304W and T309F was very similar to wild-type (data not shown). F304W also showed time-dependent inactivation of progesterone 6-hydroxylation by MDZ similar to wild-type with the k_i of 0.15 min¹ at 10 µM MDZ (data not shown). The progesterone 6-hydroxylation activity of I369W was too low to test for inactivation by MDZ (Domanski et al., 2001). However, the time dependence of MDZ hydroxylation by mutant I369W showed linearity for more than 15 min (Fig. 7). This result indicated that the 1'-OH MDZ and not 4-OH MDZ metabolic pathway is involved in enzyme inactivation.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   The time dependence of MDZ hydroxylation by mutant I369W.  and  represent 4-OH MDZ formation at 25 µM and 250 µM MDZ, respectively. No significant formation of 1'-OH MDZ was observed at 25 µM MDZ; at 250 µM, the mutant did show some formation of 1'-OH MDZ (). The solid lines through the data for 4-OH MDZ formation show a fit to a straight line.

	Discussion

Top Abstract Introduction Experimental Procedures. Results Discussion References

Human CYP3A4 is the most abundant P450 present in both liver and small intestine. Because of the number of substrates CYP3A4 metabolizes and its proposed involvement in numerous drug-drug interactions, proper evaluation of hepatic as well as intestinal CYP3A4 activity is of paramount importance. Among the various substrates available, MDZ is considered one of the best to probe in vitro and in vivo CYP3A activity (Thummel and Wilkinson, 1998). In the present investigation, we have studied the metabolism of MDZ by CYP3A4 and various SRS mutants. The study provides evidence to indicate that the 1'-OH MDZ metabolic pathway leads to CYP3A4 inactivation. The study also highlights the crucial role of SRS residues in regioselectivity of MDZ hydroxylation and provides information regarding the localization of the proposed multiple MDZ binding sites in the CYP3A4 active site.

Despite the fact that MDZ has been used extensively as a probe of CYP3A4, the inactivation caused by MDZ has not been studied in great detail. Podoll et al. (1995) have shown that MDZ caused a time- and concentration-dependent loss of CYP3A4 catalyzed testosterone 6-hydroxylation in human liver microsomes. K_I and k_inact were reported to be 16 µM and 0.32 min¹, respectively, and the inactivation was attributed to the modification of the protein, not heme moiety. More recently Schrag and Wienkers (2001) have shown that MDZ metabolism is prerequisite for CYP3A4 inactivation, and that the loss of enzyme activity is not reversible upon dialysis. These authors also suggested that protein modification rather than heme destruction leads to CYP3A4 inactivation. The present study using recombinant CYP3A4 expressed in E. coli showed MDZ causes a time- and concentration-dependent loss of progesterone 6-hydroxylation. The K_I = 5.8 µM and k_inact = 0.15 min¹ determined from these experiments showed good agreement with those reported (Podoll et al., 1995). We were further able to implicate the 1' -OH metabolic pathway in CYP3A4 inactivation based on several observations. First, the time course of MDZ metabolism by I369W, which predominantly forms 4-OH MDZ, was linear (Fig. 7), in contrast to the nonlinear time course of MDZ metabolism by F304W and T309F, which predominantly formed 1'-OH MDZ. MDZ also caused time-dependent inactivation of F304W similar to the wild-type. Second, there is a good agreement between the K_I determined from inactivation studies of the wild-type enzyme and the K_M for 1'-OH MDZ formation. However, the precise details of the underlying mechanism of enzyme inactivation remain to be established. Binding to the protein of a rearrangement product of the initial 1'-methylene radical, such as a methylene imine (Lanza et al., 1999), is a likely possibility.

Determination of in vitro kinetic parameters is essential for the prediction of drug metabolism by a particular P450 in vivo. However, the exact evaluation of these parameters for CYP3A4 has proven far more difficult than for other P450s because of the unusual substrate kinetics, including activation, autoactivation, partial inhibition, and substrate inhibition often observed (Shou et al., 1994, 2001; Ueng et al., 1997; Harlow and Halpert, 1997, 1998; Domanski et al., 1998, 2000, 2001; Korzekwa et al., 1998). Several hypotheses involving two-site or three-site models as well as the existence of functionally distinct conformers have been proposed to explain the unusual CYP3A4 kinetics (Shou et al., 1994, 2001; Ueng et al., 1997; Harlow and Halpert, 1998; Korzekwa et al., 1998; Domanski et al., 2000, 2001; Hosea et al., 2000). Observation of two very distinct K_M values for the two metabolites of MDZ (Ghosal et al., 1996; Hosea et al., 2000), which has been confirmed in our study (Table 1), suggests the existence of two MDZ binding sites in CYP3A4. The differential stimulation/inhibition by ANF and testosterone (Ghosal et al., 1996; Wang et al., 2000) and observation of two distinct K_i values for inhibition of 1'-OH and 4-OH MDZ formation by the peptide YPFP-NH₂ have provided additional evidence to support two binding sites (Hosea et al., 2000).

Two other possibilities to account for the above observations are: 1) two functionally different conformers are responsible for the conversion of MDZ to the different products (Koley et al., 1995), or 2) the substrate binds at a single site but in two different orientations. The fact that an identical nonlinear time course for both metabolic pathways was observed (Fig. 2A), although only the 1'-OH metabolic pathway seems to lead to enzyme inactivation, suggests that the formation of both products is catalyzed by a single conformer of CYP3A4, tending to exclude the first possibility. Correlational analysis provides one way to evaluate the possibility that MDZ binds at only one site but in two orientations. If this possibility were correct, substitution of several SRS residues would be expected to affect the formation of both metabolites in a similar way and a good correlation would be observed. In fact, a reanalysis of the data from previous studies of AFB1 (Wang et al., 1998; Xue et al., 2001) and progesterone (Domanski et al., 2001) oxidation revealed significant positive correlations between the production of two major metabolites by a panel of mutants (r² = 0.51 between 6- versus 16-hydroxylation of progesterone, and r² = 0.62 between 3-hydroxylation versus 8,9-epoxidation of AFB1). The strong correlations and similar S₅₀ values for the two different products for the both substrates (Ueng et al., 1997; Roussel et al., 2000) are consistent with substrate binding at same site but in two orientations. However, the preponderance of evidence (including the very poor correlations (Fig. 5), two distinct K_M values and the differential effect of various effectors/inhibitors on the two metabolites) is inconsistent with a single MDZ binding site in the CYP3A4 active site.

In recent years, the three common approaches used to understand CYP3A4 active-site dynamics include the use of mathematical models based on analysis of large sets of steady-state kinetic data, the effect of inhibitors or effectors, and site-directed mutagenesis (Domanski et al., 2001). Among these, use of site-directed mutagenesis is the only approach that can provide insight into the structural basis of the functional properties of this enzyme in the absence of a CYP3A4 crystal structure. In fact, in recent years, mutagenesis studies from this and other laboratories have implicated a number of specific amino acids in various CYP3A4 functions, and virtually all the residues that we have proposed to comprise the active site have direct counterparts in available crystal structures (Cupp-Vickery et al., 2000; Williams et al., 2000). In the present study, a closer inspection of the relative rates of formation of 1'-OH and 4-OH MDZ by various mutants also provides a possible way to determine the role of a specific amino acid in regioselectivity of MDZ oxidation and to assess the relative location of the two putative sites (Table 2). This comparison showed that the substitution of residues Ser-119, Ile-120, Leu-210, Phe-304, Ala-305, Tyr-307, and Thr-309 with a smaller amino acid caused preferential formation of 4-OH MDZ (Fig. 8). In contrast, the substitution of residues, Phe-108, Ile-120, Ile-301, Phe-304, and Thr-309 with a larger amino acid favors 1'-OH MDZ formation. The close proximity of several of these residues, as well as the fact that some of them (Ile-120, Phe-304, and Thr-309) form part of both subgroups, indicates that the two putative sites of MDZ may be partially overlapping. The steady-state kinetic analyses of mutants S119A, L210A, F304A, and L373F showed that the relative increase in 4-OH versus 1'-OH MDZ formation was caused by a 4- to 10-fold increase in the K_M value for 1'-OH MDZ, and a 2- to 4-fold increase in the V_max for 4-OH MDZ (Fig. 4 and Table 1). Additionally, the increase in the selectivity for 1'-OH MDZ by the mutants I120W and F304W was caused by a 2.5- to 3-fold increase in the V_max for 1'-OH MDZ and a slight decrease in the V_max for 4-OH MDZ and the K_M value for 1'-OH MDZ. The fact that the alterations in metabolite profile reflected changes in kinetic parameters for the formation of not just one but both metabolites in most cases provides additional support for the close proximity of the two MDZ binding sites. The spectral studies of the mutants showing that MDZ binding in either orientation causes at least a small spin-shift also indicates that a part of CYP3A4 active site may be common for both binding orientations.

View larger version (38K): [in this window] [in a new window]   Fig. 8.   Molecular model of the CYP3A4 active site showing some of the SRS residues at which side chain substitution had the most significant effect on the regioselectivity of MDZ oxidation. The substitution of residues shown in green with a smaller amino acid caused a decrease in the ratio of 1'-OH/4-OH MDZ formation, whereas the replacement of the residues shown in pink with a larger amino acid caused an increase in this ratio. The residues shown in gold fit well in both categories, showing a decrease or increase in this ratio depending on whether amino acid substitution was smaller or larger, respectively.

In conclusion, the complex kinetic behavior of CYP3A4 still remains one of the most studied yet controversial subjects in P450 research. Although MDZ hydroxylation by CYP3A4 has not been reported to show cooperativity, a number of studies have suggested the possible existence of two MDZ binding sites within the CYP3A4 active site. The present study with MDZ strongly supports this hypothesis and further indicates that the two MDZ binding sites may be in close proximity. The finding that the 1'-OH MDZ metabolic pathway might be causing CYP3A4 inactivation could be of important clinical significance. Because 1'-OH is the only product formed at clinically approved dosage of the drug, the use of MDZ as in vivo probe may require special consideration of CYP3A4 inactivation and the complications that could arise from interaction with flavones and steroids.