Abstract
In the present study, N-(α-methylbenzyl-)-1-aminobenzotriazole (MBA) and ketoconazole (KET) were identified as the inhibitors with selectivity toward dog CYP2B11 and CYP3A12, respectively. Their selectivity was evaluated using phenacetin O-deethylation (CYP1A), diazepam (DZ) N1-demethylation (CYP2B11), diclofenac 4′-hydrxylation (CYP2C21), bufuralol 1′-hydroxylation (CYP2D11), and DZ C3-hydroxylation (CYP3A12) activities in dog liver microsomes (DLM). MBA exhibited potent mechanism-based inhibition of DZ N1-demethylase activity catalyzed by both baculovirus-expressed CYP2B11 and DLM. In both cases, inhibition was characterized by a low KI (0.35 and 0.46 μM, respectively) and high kinact (1.5 and 0.56 min–1, respectively). Despite complete loss of DZ N1-demethylase activity in the presence of MBA, there was no significant loss of cytochrome P450 (P450) CO-binding spectrum. These data suggest that the inactivation involved covalent modification of P450 apoprotein, instead of the prosthetic heme moiety. A homology model of CYP2B11 was constructed, based on the crystal structure of rabbit CYP2C5, for docking the substrate (DZ) and the inhibitor (MBA), respectively. The model, within the limits of our approximations, helped explain the substrate specificity and inhibitor selectivity of CYP2B11. In contrast to MBA, KET was identified as a potent and selective reversible (competitive) inhibitor of CYP3A12 (KI = 0.13–0.33 μM). In fact, complete inhibition of CYP3A12-dependent DZ C3-hydroxylation was possible at a low KET concentration (1 μM). Therefore, it is concluded that one can attempt to conduct P450 reaction phenotype studies with DLM using MBA and KET as selective inhibitors of CYP2B11 and CYP3A12, respectively.
Dog is an important and commonly used species for the preclinical assessment of metabolism, pharmacokinetics, safety, and efficacy in drug discovery and preclinical development. Cytochromes P450 (P450s) in dog play a pivotal role in drug biotransformation and clearance. To date, nine dog cytochrome P450 genes have been isolated and sequenced: CYP1A1 (Uchida et al., 1990), CYP1A2 (Uchida et al., 1990), CYP2B11 (Graves et al., 1990), CYP2C21 (Uchida et al., 1990), CYP2C41 (Blaisdell et al., 1998), CYP2D15 (Sakamoto et al., 1995), CYP2E1 (Lankford et al., 2000), CYP3A12 (Ciaccio et al., 1991), and CYP3A26 (Fraser et al., 1997). Seven of them have been expressed individually as functional enzymes in baculovirus-insect cells, which have enabled the evaluation of their substrate selectivity and the identification of dog P450 isoform marker substrates (Shou et al., 2003). The advantage of such isoform markers is that they help screen new chemical entities as inhibitors of specific P450s in pooled dog liver microsomes (DLM) and provide useful probes for selecting isoform-selective inhibitors that can be used to assess the contribution of a P450, or subclass of P450, to the metabolism of a drug or drug candidate. Many chemical inhibitors have been known to selectively inhibit human and animal P450s. In the absence of P450 isoform-selective substrates, however, their selectivity toward dog P450s has not been fully characterized.
Mechanism-based inactivators are among the most specific enzyme inhibitors. They are substrates for the target enzyme, which is irreversibly inhibited by conversion of the inhibitor to chemically reactive intermediates or products capable of covalently binding to the enzyme. The selectivity of the inhibitors is high because of selectivity of both the binding and catalytic inactivation, in contrast to reversible inhibition. The N-aralkyl derivatives of 1-aminobenzotriazole, such as N-(α-methylbenzyl-)-1-aminobenzotriazole (MBA), have been reported to be potent mechanism-based inhibitors of cDNA-expressed CYP2B, the phenobarbital (CYP2B)- and β-naphthoflavone-induced enzymes (CYP1A) in pulmonary or liver microsomes of rat (Kent et al., 1997), goat (Huijzer et al., 1989), guinea pig (Sinal and Bend, 1995, 1996; Sinal et al., 1998), and rabbit (Mathews and Bend, 1986; Woodcroft et al., 1990; Grimm et al., 1995). They exhibit pronounced isozyme and tissue (lung) selectivity for P450 inactivation, both in vitro and in vivo (Woodcroft and Bend, 1990; Woodcroft et al., 1990; Knickle and Bend, 1992; Mathews and Bend, 1993). The derivatives can inactivate P450 activities by mechanisms involving suicide inactivation, i.e., covalent modification of the prosthetic heme group (Mathews and Bend, 1986) and covalent modification of the apoprotein moiety (Woodcroft et al., 1997). On the other hand, ketoconazole (KET) has been well documented as an inhibitor of CYP3A in many species (www.druginteractioninfo.org; Baldwin et al., 1995; Sai et al., 2000; Kuroha et al., 2002).
In the present study, the selectivity and potency of MBA and KET as inhibitors of different dog P450 activities were evaluated using both DLM and recombinant P450 proteins. The study was expanded to include a homology model for CYP2B11, based on the structure of rabbit CYP2C5, the only available mammalian cytochrome P450 structure at the time our models were developed, and used for docking of MBA to the enzyme. It is concluded that under the appropriate conditions, MBA and KET are selective inhibitors of CYP2B11 and CYP3A12 in DLM, respectively. Therefore, in combination with recombinant dog P450 proteins, the two inhibitors can be used as chemical probes when conducting reaction phenotyping studies.
Materials and Methods
Materials. Chemicals were obtained from the following sources: KET from Fisher Scientific Co. (Pittsburgh, PA); MBA, diclofenac (DIC), phenacetin (PHE), acetaminophen (ACE), diazepam (DZ), nordazepam (NDZ), temazepam (TMZ), and NADPH from Sigma-Aldrich (St. Louis, MO); bufuralol (BUF), 1′-OH-bufuralol, and 4-OH-diclofenac from Ultrafine Chemicals (Manchester, UK); and DLM from 10 male beagle dog donors were purchased from BD Gentest (Woburn, MA). All other chemicals were purchased from commercial sources and were of the highest purity available. Expression, microsomal preparation, and functional characterization of cDNA-expressed recombinant dog P450s (CYP1A1, 2B11, 2C21, 2C41, 2D15, 3A12, and 3A26) containing coexpression of human P450-oxidoreductase in baculovirus-Sf21 cells in our laboratory were described elsewhere (Shou et al., 2003).
Isoform-Specific Assays. DZ N1-demethylation (CYP2B11) and C3-hydroxylation (CYP3A12), DIC 4′-hydroxylation (CYP2C21), and BUF 1′-hydroxylation (CYP2D15) have been characterized as the P450 isozyme-selective monooxygenase activities, and the assays were performed according to our previous report (Shou et al., 2003). To determine PHE O-deethylation as a marker of CYP1A, the reactions were performed in a final volume of 1 ml (duplicate). The reaction mixture contained 0.1 M potassium phosphate (pH 7.4), MgCl2 (10 μM), and individual recombinant P450s (25 pmol/ml) or DLM (0.1 mg/ml). Reactions were initiated by the addition of NADPH (1 mM), allowed to proceed for 10 min at 37°C, and were terminated with acetonitrile (3 volumes) containing the appropriate internal standard (4-OH-butyranilide). Samples were spun for 10 min at 3000 rpm to precipitate protein, and the supernatant was separated and diluted for high-pressure liquid chromatography or liquid chromatography-tandem mass spectrometry analysis. To obtain estimates of Km and Vmax values, reaction rates were measured under linear conditions using a number of substrate concentrations: PHE (1–300 μM), DIC (0.4–1000 μM), BUF (0.5–500 μM), and DZ (1.5–300 μM). Sample preparation for individual incubations was previously described (Shou et al., 2003). Rates for the formation of metabolite in each assay were quantitated with respect to internal standards and normalized by calibration curves of the metabolite standard versus internal standard. Data were used to fit the Michaelis-Menten equations below (eq. 1) to obtain Km and Vmax. Substrate concentration approaching Km was used for the enzyme inhibition studies (IC50).
Enzyme Inhibition. For initial screening (IC50) with MBA (0.004–3.3 μM) or KET (0.004–3.3 μM), DLM (0.1–0.25 mg/ml), recombinant CYP3A12 (20 pmol/ml), or CYP2B11 (20 pmol/ml) were used. Each inhibitor dissolved in 50% acetonitrile in water was added into the incubation mixture (<1% final concentration of acetonitrile) just before addition of each marker substrate (concentration equivalent to Km). The reaction was initiated with addition of NADPH (1 mM), allowed to proceed for 10 or 15 min, and stopped with addition of acetonitrile (2–3 volumes). The procedure of sampling and metabolite quantitation were described earlier. The concentration-inhibition curves were plotted (Fig. 2) and IC50 values were calculated using eq. 2. Similarly, Ki values of KET for the DLM and recombinant CYP3A12 were determined by incubation combinations (56 data points) of varying DZ (1–200 μM) versus KET (0.004–10 μM). The data were fit to a competitive inhibition kinetic model (eq. 3) to generate Ki. To determine MBA as a mechanism-based inhibitor, pooled DLM and recombinant CYP2B11 were preincubated with MBA (0.001–6 μM) for different time points (0, 1, 2, 4, and 6 min) in the presence of NADPH. Controls consisting of preincubation mixture at each time point in the absence of the inhibitor were used to determine the fraction of enzyme loss. The inactivation of the enzyme by the inhibitor was terminated by the dilution of the reaction mixture with 0.1 M phosphate buffer (10-fold) in ice bath. Substrate (10-fold Km) was added alone or together with the inhibitor (control) for an additional 10 min incubation. The reactions were stopped with acetonitrile (3 volumes) containing internal standard. Enzyme activity was measured by comparing rate of product formation with and without the inhibitor in the preincubation.
Liquid Chromatography-Tandem Mass Spectrometry. Quantitative analysis of parent compound and its metabolite(s) from each reaction (DIC 4′-hydroxylation, BUF 1′-hydroxylation, DZ N1-demethylation, and C3-hydroxylation) was previously described in detail (Shou et al., 2003). ACE (m/z = 151.9) formed in PHE O-deethylation was determined and quantitated with standard curves generated from ratio of authentic metabolite standard to internal standard (4′-OH-butyranilide, m/z = 180.2). The rate of metabolite formation in the absence or presence of the given inhibitors represents the inhibition extent as a percentage of enzyme activity loss.
Kinetic Parameter Estimation. Nonlinear regression was performed using nonlinear least-squares algorithms. Fits were performed using multiple initial parameter estimates to ensure minimum global variance. Values for all kinetic parameters were calculated by SigmaPlot 8.0 (SPSS Inc., Chicago, IL), an iterative procedure based on appropriate initial estimates that best fitted the data. Statistical analysis of the data was performed using both the residual sum of squares (RSS) and regression coefficient (r2). Kinetic equations used in the text for Michaelis-Menten kinetics (eq. 1), IC50 determination (eq. 2), competitive inhibition (eq. 3), and mechanism-based inhibition (eq. 4) are expressed and interpreted as below:
Equation 2 was used to determine IC50, where Emax is the maximum inhibition and n is the slope factor, and I is the inhibitor concentration. Equation 3 was used to determine Ki for completive inhibition, and eq. 4 is used to evaluate the inactivation of a given enzyme (percentage of remaining enzyme activity, [E]/[E]total), which is a function of [I] (inactivator concentration) and t (preincubation time of inactivator with enzyme). kinact and KI represent rate constant for enzyme inactivation and dissociation constant between the inactivator and enzyme, respectively.
CO-Binding Difference Spectra. MBA (1 μM) and rCYP2B11 (200 pmol/ml) in phosphate buffer (100 mM) in a total volume of 1 ml were mixed and preincubated (6 min) at 37°C in the presence or absence of NADPH (1 mM). The carbon monoxide absorbance difference (scanning from 400–500 nm) was measured in the presence of dithionite as a reductant by a Cary 400BIO spectrophotometer (Varian, Inc., Palo Alto, CA).
Homology Modeling (CYP2B11). The crystal structure of CYP2C5 (1DT6; Williams et al., 2000) was used as the template to generate a homology model of dog CYP2B11. The homology model presented here was generated in the following manner: a sequence alignment of dog CYP2B11 and CYP2C5 was generated, the best model of the 10 intermediate models for CYP2B11 was generated based on the alignment to the X-ray coordinates of CYP2C5 structure, and this model was then subsequently energy minimized. Homology modeling was performed using MOE version 2000.02 (Chemical Computing Group, Montreal, CA). The cofactor was modified to add oxygen atom to the heme-iron to simulate the heme-iron-oxo radical abstracting species for substrate dockings. In our substrate docking calculations, the oxygen atom attached to the heme iron-carried –1 negative charge since we cannot represent radical species in our empirical calculations. However, the inhibitor dockings were carried out without the oxygen atom.
Substrate and Inhibitor Docking. An energy minimized structure of DZ was visually docked into the active site of CYP2B11 proximal to the heme-iron. Docked models of the substrates and inhibitors of CYP2B11 in its active site were generated in the following manner: a 12A sphere of the active site of the enzyme was defined around the initial docked position of DZ, substrate and inhibitor structures were energy minimized with mmffs force field (Halgren et al., 1999), approximately 25 diverse conformations of each ligand were generated with Enumerate Torsions methodology (Feuston et al., 2001) and energy minimized with mmffs force field, and all the energy-minimized conformations of each ligand were then docked into the enzyme active site with the FLOG methodology (Miller et al., 1994) using the docked orientation of DZ as a guide. The top-scoring (20%) docked orientations were used to identify substrate binding models that contained complementary protein-ligand interactions and presented metabolic hot spots toward the heme-iron-oxo species for the substrates and for mechanism-based inhibitor. The metabolic hot spots or hydrogen atom abstraction energies for molecules in our study were calculated using the methodology presented recently (Singh et al., 2003).
Results
DLM were incubated with various substrates (PHE, DZ, DIC, and BUF), and the time- and protein concentration-dependent formation of the corresponding metabolites was assessed. Subsequently, metabolite formation was determined at varying concentrations of substrate under linear conditions. The reactions were optimized, and linearity was ensured up to 0.25 mg/ml protein and at least 20 min of incubation at 37°C (data not shown). Under optimal conditions, less than 20% of substrate was consumed, and generation of secondary metabolites was kept to a minimum.
Marker Assay for CYP1A. To explore a marker substrate for CYP1A activity in DLM, phenacetin O-deethylation activity was determined. Among P450 isozymes tested, CYP1A1 was most capable of catalyzing PHE O-deethylation (Fig. 1) and exhibited 6-fold higher specific activity than any of the other isoforms tested. Although CYP1A2 is also expressed in DLM, specific activity of CYP1A2 for PHE was not determined due to a lack of expressed CYP1A2 either in our laboratory or commercial sources.
Km andVmax Determination for the Marker Substrates in DLM. Under linear conditions for product formation, the isoform-specific reactions in DLM, enzyme kinetics for phenacetin O-deethylation (CYP1A), DZ N1-demethylation (CYP2B11), and C3-hydroxylation (CYP3A12), DIC 4′-hydroxylation (CYP2C), and BUF 1′-hydroxylation (CYP2D15) were determined to give apparent Km values of 80, 6.4, 155, 110, and 3.5 μM, and apparent Vmax values of 3.2, 6.3, 6.5, 0.4, and 3.5 nmol/min/mg, respectively (Table 1). DZ N1-demethylation catalyzed by recombinant CYP2B11 was characterized by a similar Km (5.4 μM) to that of DLM (6.4 μM). However, the Km describing DZ C3-hydroxylation in the presence of recombinant CYP3A12 was lower than that for DLM (41 versus 156 μM).
IC50 Determination. To initially evaluate the inhibition potency and selectivity of MBA and KET for the various P450 isoforms, IC50 curves were generated in DLM and expressed P450s using the isoform-selective marker assays. In the incubation conditions, a fixed substrate concentration (equivalent to Km) and varying inhibitor concentrations were chosen, and the inhibition curves were plotted by percentage of control (percentage of metabolite formed between the presence and absence of the inhibitor examined) versus inhibitor concentration. As shown in Fig. 2A, MBA (IC50 = 0.1–0.15 μM) was very potent toward the DZ N1-demethylation in DLM and recombinant CYP2B11 (Fig. 2C) but had relatively small effects on other P450 activities studied. Similarly, KET (IC50 = 0.1–0.12 μM) exhibited selectivity toward DZ C3-hydroxylation in DLM (Fig. 2B) and rCYP3A12 (Fig. 2D) with no marked effect on other marker activities, except for a moderate inhibition (IC50 = 2.5 μM) of BUF 1′-hydroxylation (CYP2D15) in DLM. The cross inhibition of CYP2D15 and CYP3A12 activities in DLM was not observed at lower concentrations of KET (i.e., < 0.33 μM), where microsomal CYP3A12 activity was inhibited by 80%, with no effect on the CYP2D15 activity. The results from the rapid screening demonstrated that MBA and KET were selective and potent inhibitors of CYP2B11 and CYP3A12, respectively.
Competitive Inhibition of CYP3A12. As shown in the previous data, KET selectively inhibited the CYP3A12-mediated DZ C3-hydroxylation in DLM with a low IC50 (∼0.1 μM). To identify the type of inhibition, various concentrations of DZ (1–200 μM) and KET (0.004–10 μM) in the presence of enzyme (DLM = 0.2 mg/ml or CYP3A12 = 20 pmol/ml) were designed to generate a total of 54 experimental data points, which were employed to perform a diagnostic analysis for all types of inhibition kinetics (competitive, non-, uncompetitive, and mixed types). The result showed the best fit for competitive inhibition (Fig. 3, r2 = 0.989 and RSS = 0.25–0.28). Ki values of 0.13 (DLM) and 0.33 (CYP3A12) μM were generated (Table 2). In contrast, Ki values for CYP2B11-mediated N1-demethylation in rCYP2B11 and DLM were 31.5 and 14.4 μM, respectively, which were much larger (>44-fold) than that observed for CYP3A12 activity, suggesting that KET had high binding affinity and potency for CYP3A12 relative to any of other P450s.
Mechanism-Based Inhibition of CYP2B11. DZ N1-demethylation was established in our previous investigation as a highly selective marker of CYP2B11 present in DLM and used to evaluate the kinetics of MBA as a mechanism-based inhibitor. The time- (five time points, up to 6 min) and concentration- (eight concentrations, up to 3.3 μM) dependent inactivation of CYP2B11 for the marker reaction in DLM or CYP2B11 is shown in Fig. 4. Control incubations were conducted in the absence of the inhibitor at each preincubation time point. After the preincubation, the reaction mixture was diluted 10-fold with phosphate buffer in the presence of a high concentration (100 μM) of the substrate (DZ). After additional 20 min of incubation, the remaining CYP2B11 activity was determined. The preincubation time-dependent decrease in CYP2B11 activity was observed at various concentrations of MBA. In all cases, the extent of reversible inhibition (decrease in percentage of activity at zero time) was estimated to achieve up to 30% as MBA (0–3.3 μM) increased. To minimize the contribution of this reversible inhibition, an equal amount of the inhibitor was added to the control incubation after a completion of the preincubation for subtraction. Thus, binding (KI) and rate constants (kinact) were reliably determined. The inactivation (% control) was calculated from the ratio of the rate for metabolite formation (NDZ) in the presence and absence of the inhibitor (MBA) at the various preincubation time and concentration points. The observed data (inactivation versus time [t] and inhibitor concentration [I]) were plotted in a three-dimensional manner. The dot and surface plots represent the observed and predicted results (fitted to eq. 4). Kinetic parameters (KI and kinact) were resolved by nonlinear regression (Table 2). As indicated in Fig. 4, inactivation of the enzyme activity increased as the preincubation time (t), and the inhibitor concentration [I] increased. MBA at 1 μM and 6 min of preincubation inhibited exclusively the formation of NDZ in DLM or CYP2B11 preparations. KI and kinact for MBA in the DZ N1-demethylation were 0.35 μM and 1.56 min–1 for rCYP2B11 and 0.46 μM and 0.56 min–1 for DLM, respectively (Table 2). Overall, MBA exhibited a potent mechanism-based inactivation of DZ N1-demethylase activity in the presence of recombinant CYP2B11 and of DLM.
CO-Binding Spectra of CYP2B11. Carbon monoxide binding study was conducted to understand whether the intermediate metabolite(s) formed in incubation of MBA with CYP2B11 binds to heme-iron moiety, thereby preventing formation of CO-binding spectrum. MBA (1 μM) was preincubated with rCYP2B11 (200 pmol/ml) for 6 min in the presence of NADPH (Fig. 5), at which time the enzyme activity was almost completely inhibited (Fig. 4A). Sodium dithionite was added into the incubations preincubated with the absence and presence of MBA to reduce the CYP2B11. The extent of CO-bound CYP2B11 was measured and found to have no difference between the absence and presence of NADPH.
CYP2B11 Homology Model and Docking Orientations of DZ and MBA. Docked models of DZ and TMZ presented in Fig. 6 show that the top-scoring orientations of these compounds present N1-methyl toward and within 3A of the heme-iron-oxo species consistent with the experimental observations reported above. In the case of DZ the lowest hydrogen atom abstraction energy is for the 3-position followed by the energy for the N1-methyl (see Appendix). According to our hypothesis (Singh et al., 2003), CYP3A enzymes abstract hydrogen atoms with the lowest abstraction energy; hence, it is interesting to note that the major metabolite produced by CYP3A12 is 3-hydroxy-diazepam (TMZ). However, the lowest scoring docked orientation of DZ in CYP2B11 places the N1-methyl with the second highest hydrogen atom abstraction energy closest to the heme-iron-oxo species, consistent with the experimentally observed metabolite. The docked orientation of MBA places the compound in the active site with the α-methyl-benzyl moiety, containing the hydrogen atom with the lowest abstraction energy, in the proximity of the heme-iron-oxo species.
Discussion
Assessment of the individual contribution of different P450s to the metabolism of a test compound during drug development is essential. These types of data allow one to evaluate the enzyme(s) involved in the metabolism of a drug and whether or not the clearance of that drug can be influenced by other drugs that modulate the activity of the enzyme involved. For nearly all animal species, P450 isoform-selective inhibitors (chemical and antibody) have been largely used as diagnostic tools for such a purpose. However, investigation to define the enzyme selectivity of diagnostic inhibitors requires the use of highly isozyme-selective substrates and/or pathways, which are usually studied in liver microsomes or hepatocytes containing a heterogeneous mixture of different P450s. At the same time, individual (cDNA) expressed P450 proteins have proven to be very useful and can serve as a complimentary tool. For example, in a previous study, several isozyme-selective monooxygenase reactions were identified using a panel of recombinant dog P450s that have enabled us to screen for inhibitors of the individual P450s (Shou et al., 2003). DZ N1-demethylation (CYP2B11), BUF 1′-hydroxylation (CYP2D15), DIC 4′-hydroxylation (CYP2C21), and DZ C3-hydroxylation (CYP3A12) reactions were employed for the present study. The known CYP1A marker assay, PHE O-deethylation, was performed in the study with a panel of individual cDNA-expressed P450s to determine substrate specificity of CYP1A in dog. Among the seven recombinant enzymes tested, CYP1A1 was most active, and others showed negligible or no PHE O-deethylation activity (Fig. 1). Although CYP1A2 is expressed in dog liver (Uchida et al., 1990) and selectively catalyzes PHE O-deethylation in many other species (Kobayashi et al., 2002; Yuan et al., 2002), the specific activity of dog CYP1A2 cannot be evaluated without the recombinant enzyme. As a result, the contribution to the PHE O-deethylation in the liver between CYP1A1 and CYP1A2 could not be determined in the present study.
Members of the CYP2B subfamily have been shown to play a role in the biotransformation of various endogenous and xenobiotic compounds in numerous animal species (Nilsen et al., 1981; Schulze et al., 1990; Chang et al., 1993). CYP2Bs are expressed not only in liver but also in lung from many species and thus may contribute to extrahepatic metabolism (Domin et al., 1984; Vanderslice et al., 1987). N-aminobenzotriazole and its derivatives have been extensively reported to inhibit CYP2B and CYP1A enzymes in both lung and liver microsomes of rabbit and guinea pig induced with PB. The inhibition is mechanism-based and is characterized by a low KI and high kinact (Grimm et al., 1995). Inhibition of P450 activity by these compounds is most likely associated with the NADPH-dependent covalent binding to the P450 apoproteins in pulmonary and hepatic microsomes of guinea pigs treated with PB, which were migrated on SDS-polyacrylamide gel electrophoresis gel following incubation with NADPH (Woodcroft et al., 1997). Our studies indicate that MBA was very potent and selective toward the CYP2B11-mediated DZ N1-demethylation (IC50 = 0.1–0.15 μM), among the five maker activities in DLM. MBA at 1 μM almost completely inhibited DZ N1-demethylase activity in DLM but had no significant effect on other marker activities (Fig. 2). Further study demonstrated that the inhibition is preincubation time-dependent (mechanism-based). The preincubation (6 min) of MBA (1 μM) with DLM or rCYP2B11 resulted in a complete inhibition of the DZ N1-demethylation, with no significant effect on P450 CO-binding absorption spectrum. This suggests that inactivation of CYP2B11 in DLM is due to covalent binding of MBA-reactive metabolites to apoproteins rather than the prosthetic heme group. The inactivation kinetics for DZ N1-demethylase activity in DLM (KI = 0.46 μM and kinact = 0.56 min–1) and rCYP2B11 (KI = 0.35 μM and kinact = 1.5 min–1) revealed that MBA was an effective mechanism-based inhibitor. The kinact values are similar to those for the CYP2B-mediated O-dealkylation of 7-pentoxyresorufin in rabbit (kinact = 0.55–0.65 min–1) and guinea pig (kinact = 0.35–0.49 min–1) liver microsomes (Grimm et al., 1995; Sinal et al., 1998). However, the KI (0.46 μM) for CYP2B11 activity in DLM is much lower than for CYP2B activity in rabbit (KI = 6.9 μM) and guinea pig (KI = 2.5 μM) liver microsomes.
Although the metabolic activation of MBA has not been fully characterized, a previous study has shown that N-benzyl-1-aminobenzotriazle, a analog of MBA exhibits mechanism-based inhibition of CYP2B18-dependent 7-pentoxyresorufin O-dealkylase and of CYP1A-dependent 7-ethoxyresorufin O-deethylase activities in liver microsomes from PB-induced guinea pigs without a significant loss of the P450 absorption spectrum (Woodcroft and Bend, 1990), suggesting that the inactivation occurs via the covalent modification of the P450 apoproteins. The metabolism of N-benzyl-1-aminobenzotriazle with the liver microsomes was further characterized, and the inactivation mechanism was elucidated through the formation of at least three reactive metabolite species (Woodcroft et al., 1997). These are nitrogen- or carbon-centered free radicals of intact BBT and a benzyl radical, which are proposed to involve covalent binding to the apoproteins. The intermediates were tentatively identified by the formation of three respective ultimate products observed: 1-aminobenzotriazole, benzaldehyde, and benzotriazole. Based on this assumption, the proposed pathways of MBA bioactivation by CYP2B11 are shown in Fig. 7.
Mechanism-based inhibition of P450s gives a number of advantages because it can afford selective inhibition of individual P450 forms. In contrast to reversible inhibitors, mechanism-based inhibitors require metabolic activation of the target enzyme, which is often P450 form specific and subsequent covalent modification of the apoproteins or extremely tight interactions with the active site. As a result, mechanism-based inhibition is long-lasting, and enzyme activity and expression level are only restored by de novo protein synthesis. Study of covalently modified proteins, after incubation with an inhibitor, is also helpful because it can provide information related to catalytic mechanisms and the nature of the active site of the enzyme (Bend et al., 1985; Halpert et al., 1994; Halpert, 1995).
The selectivity toward the DZ N1-demethylation with a high Vmax (35.9 nmol/min/nmol) and low Km (5.4 μM) indicates that the N1-methyl group is orientated closely to the oxygen atom on the heme-iron in the active site for the demethylation. The docked orientation of DZ in CYP2B11 homology model displays that the N1-methyl group with the second highest hydrogen atom abstraction energy is placed in close proximity to the heme-iron-oxo species and consistent with its observed site of the metabolism (Fig. 6A). MBA was designed to have molecular features that structurally mimic benzphetamine, a substrate for CYP2B-dependent N-demethylation (Serabjit-Singh et al., 1983; Mathews and Bend, 1986). Common structural features include an N-benzyl group, an aromatic region of similar size, an amino group in the desired region of oxidation, and a α-carbon for substitution. Similarly, MBA is characterized by a very low KI (∼0.4 μM) and requires activation to form the reactive metabolite that inactivates CYP2B11 activity. A model describing the binding of MBA to CYP2B11 shows the orientation of the substrate/inhibitor in the active site and presents metabolically labile sites in the proximity of the heme-iron-oxo species, which can possibly lead to mechanism-based inhibition (Fig. 6C). It is important to note that the model described herein is based on the crystal structure of CYP2C5, not CYP2B4, which shares 72.3% homology with CYP2B11 (Scott et al., 2003). It would have been advantageous to use the CYP2B4 crystal structure at the time of the study. The following differences are observed between the CYP2C5 and CYP2B4 crystal structures: the conformation of the B and C helices are significantly different; hence, the conformation of the BC loop presented in our model would be affected; the positions of the F and G helices are quite different, with a helical structure observed in CYP2B4 for the so-called FG loop, thus affecting the interactions with the F helix shown in our structure; the conformation of the loop between helices H and I is also different but probably would have no bearing on the binding interactions in the active site; and finally, the conformation of the c-terminal loop is turned away from the active site and hence may not provide the packing interactions from V477 as observed in our model. Future modeling efforts of CYP2B11 will be based on CYP2B4 and should address these shortcomings and provide a more accurate picture of the binding interactions with its substrates and inhibitors.
KET has been used widely as a potent and selective inhibitor for CYP3A (www.druginteractioninfo.org; Baldwin et al., 1995; Sai et al., 2000). However, previous reports have indicated that KET can also inhibit other P450 activities, particularly CYP1A and CYP2C. The degree of inhibition is dependent on the substrates and KET concentration used. Therefore, KET must be used carefully when attempting to define the role of CYP3A in the metabolism of a drug. The results of the present study have shown that KET is a potent and selective inhibitor of CYP3A12 activity in DLM (IC50 = 0.1–0.12 μM). The results indicate that KET (0–3.3 μM) elicits a minimal inhibitory effect on CYP1A2-, CYP2B11-, and CYP2C21-mediated reactions in DLM. However, KET at 1.2 and 3.3 μM inhibited CYP2D15-mediated bufuralol 1′-hydroxylation by 40 and 56% (IC50 = 2.5 μM), respectively. The CYP3A12 selectivity of KET is limited to the concentration used. Thus, caution should be taken when the KET is used as a CYP3A12 probe at higher concentrations. The inhibition kinetics of CYP3A-mediated DZ C3-hydroxylation in DLM displayed by KET (Ki = 0.13–0.33 μM) are consistent with that of midazolam 1′- and 4-hydroxylations (Ki = 0.024–0.11 μM) (Kuroha et al., 2002).
In summary, the present studies have demonstrated that MBA and KET are selective inhibitors of CYP2B11 and CYP3A12, respectively, using the P450 isoform markers in DLM. MBA is a potent mechanism-based inhibitor, and KET behaves as a potent competitive inhibitor. Because of their selectivity, it is proposed that P450 reaction phenotyping studies can be attempted with DLM in combination with cDNA-expressed dog P450 proteins.
Appendix
The hydrogen abstraction energies presented for the molecules below (Scheme 1; DZ, TMZ, and MBA correspond to Fig. 6, A–C) are calculated using the Trend Vector methodology (Singh et al., 2003). The hydrogen abstraction energy corresponding to the hydrogen is transferred to the heavy atom to which it is attached. For topologically equivalent hydrogens, the energy of the hydrogen with the highest surface area exposure is given. The energies are shown only for the chemically equivalent hydrogens. The atoms with energy of 0.0 indicates that their surface area exposure is less than 8A2 (Singh et al., 2003) and hence are not accessible for abstraction.
Acknowledgments
We thank Matthew Walker for calculation of the hydrogen abstraction energy.
Footnotes
-
doi:10.1124/jpet.104.077651.
-
ABBREVIATIONS: P450, cytochrome P450; DLM, dog liver microsomes; MBA, N-(α-methylbenzyl-)-1-aminobenzotriazole; KET, ketoconazole; DIC, diclofenac; PHE, phenacetin; ACE, acetaminophen; DZ, diazepam; NDZ, nordazepam; TMZ, temazepam; BUF, bufurolol; RSS, residual sum of squares.
-
↵1 Current address: Department of Medicinal Chemistry, Vitae Pharmaceuticals, Fort Washington, PA.
-
↵2 Current address: Pfizer Inc., San Diego, CA.
-
↵3 Current address: Drug Metabolism and Pharmacokinetics, Bristol-Myers Squibb, Pharmaceutical Research Institute, Princeton, NJ.
- Received September 17, 2004.
- Accepted January 25, 2005.
- The American Society for Pharmacology and Experimental Therapeutics