Cytochrome P450 (P450; EC1.14.14.1) enzymes play a key role in the clearance of many drugs, and alteration of the activity of these enzymes is the major cause of drug-drug interactions. P450 enzymes comprise a superfamily of hemoproteins residing on the cytoplasmic face of the endoplasmic reticulum membranes. Together with cytochrome P450 reductase, P450 enzymes catalyze a variety of types of oxidation reactions such as hydroxylation and dealkylation (Nelson et al., 1996). Three families of P450 enzymes (CYP1, CYP2, and CYP3) are involved in the metabolism of most xenobiotics in humans; CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, and 3A4/5 are responsible for the metabolism of the majority of drugs. Glucuronidation represents a major biotransformation pathway that increases the elimination of many lipophilic drugs and /or metabolites. UDP-glucuronosyltransferases (UGTs) are a superfamily of enzymes that use uridine diphosphoglucuronic acid (UDPGA) as a cofactor and reside on the lumen side of the endoplasmic reticulum membranes (Miners and Mackenzie, 1991). There are two subfamilies of UDP-glucuronosyltransferases involved in drug metabolism in humans, UGT1A and UGT2B (Wells et al., 2004). A total of 18 human UGT enzymes have been characterized to date (Wells et al., 2004). UGT enzymes catalyze glucuronidation reactions at a nucleophilic functional group such as a hydroxyl, carboxylic acid, or amine leading to the formation of β-d-glucuronic acid conjugates (Senafi et al., 1994; King et al., 1996).

Muraglitazar (Pargluva, BMS-298585), N-[(4-methoxyphenoxy-)carbonyl]-N-[[4-[2-(5-methyl-2-phenyl-4-oxazolyl)ethoxy] phenyl-]methyl]glycine), is an oxybenzyl-glycine analog (nonthiazolidinedione) and a novel dual α/γ PPAR activator (Devasthale et al., 2005). Muraglitazar has both glucose- and lipid-lowering effects when tested in animal models of diabetes and dyslipidemia and in patients with diabetes (Devasthale et al., 2005; Harrity et al., 2006; Kendall et al., 2006). After oral administration to humans, [¹⁴C]muraglitazar underwent extensive metabolism by both oxidation and glucuronidation pathways (Wang et al., 2006; Zhang et al., 2006). Fecal excretion was the major elimination pathway for [¹⁴C]muraglitazar (>95% of the recovered dose), and metabolites including 12-hydroxy muraglitazar (M10), 17-hydroxy muraglitazar (M11), 9-hydroxy muraglitazar (M14), O-demethyl muraglitazar (M15), and 12-hydroxy O-demethyl muraglitazar (M5) were the predominant drug-related components in the feces, whereas formation of O-dealkyl muraglitazar (M1), and oxazole-ring opening metabolites (M9 and M16) were relatively minor metabolic pathways for muraglitazar (Zhang et al., 2006). The structures of these metabolites are shown in Fig. 1. Human bile collected with an oral-gastro-duodenal tube during a limited collection period of 3 to 8 h postdose contained 40% of the dose-related material, 70% of which were glucuronide conjugates of muraglitazar and its oxidative metabolites. These results suggested that muraglitazar was cleared primarily by hepatic metabolism and excreted into the bile mainly as glucuronide conjugates (Wang et al., 2006).

The objective of this study was to identify the predominant human P450 and UGT enzymes responsible for the in vitro metabolism of muraglitazar. [¹⁴C]Muraglitazar was incubated with human cDNA-expressed P450s and UGTs to determine the catalytic turnover for oxidation and direct glucuronidation of muraglitazar. The individual enzymes involved in muraglitazar biotransformation were further investigated in studies with specific inhibitory antibodies and selective chemical P450 inhibitors in human liver microsomes. Kinetic parameters for metabolite formation were determined in incubations with both human liver microsomes and cDNA-expressed enzymes. The intrinsic clearance determined for each P450 enzyme, together with the relative concentration of each P450 enzyme in the human liver, was used to estimate the contribution of each enzyme to the formation of a particular metabolite and to the overall metabolism of muraglitazar. The results predict that multiple P450 and UGT enzymes will be involved in the metabolic clearance of muraglitazar.

Materials and Methods

Materials. [¹⁴C]Muraglitazar (specific radioactivity 8.4 μCi/mg, radiochemical purity 99.6%) was synthesized at Bristol-Myers Squibb. Its structure and the position of the radiolabel are shown in Fig. 1. Alamethicin, magnesium chloride, β-NADPH, and UDPGA were obtained from Sigma-Aldrich (St. Louis, MO). Sodium phosphate and trifluoroacetic acid (TFA) were obtained from EM Scientific (Gibbstown, NJ). Pooled human liver microsomes (HLMs; prepared from 22 donors), membranes from insect cells transfected with baculovirus containing cDNA of human P450s (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 3A4, and 3A5) and UGTs (UGT1A1, 1A3, 1A4, 1A6, 1A8, 1A9, 1A10, 2B4, 2B7, and 2B15), and P450 inhibitors (furafylline, 8-methoxypsoralen, orphenadrine, sulfaphenazole, tranylcypromine, quinidine, ketoconazole, montelukast, benzylnirvanol, 1-aminobenzotriazole) were purchased from BD Gentest (Woburn, MA). Two batches of pooled human liver microsomes, designated HLM A and HLM B, were used in this study. As indicated on the manufacturer's data sheet, HLM A contained 360, 320, and 570 pmol/mg of total P450, cytochrome P450 reductase, and cytochrome b₅, respectively, and HLM B contained 240, 340, and 500 pmol/mg of total P450, cytochrome P450 reductase, and cytochrome b₅, respectively. P450 enzyme activities with respect to the metabolism of specific probe substrates were also different: 240, 3100, 48, 71, and 6800 pmol/min/mg protein in HLM A versus 500, 1700, 26, 70, and 4200 pmol/min/mg protein in HLM B for CYP2C8, 2C9, 2C19, 2D6, and 3A4, respectively. Further characterization of specific enzyme content was not carried out. Inhibitory monoclonal anti-human P450 antibodies for CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2Cs (2C8, 2C9, 2C18, and 2C19), 2D6, 2E1, and 3A4 were obtained from the National Institutes of Health (NIH) (Gelboin and Krausz, 2006). Cryopreserved human hepatocytes from three donors were purchased from In Vitro Technologies (Baltimore, MD). Acetonitrile was purchased from Allied Signal Inc. (Muskegow, MI). All other organic solvents and water were of HPLC grade. Stock solutions of 5.17 mM [¹⁴C]muraglitazar were prepared by dissolving 2.67 mg of [¹⁴C]muraglitazar in 1 ml of water/acetonitrile (1:1 v/v). P450 inhibitor stock solutions (1–20 mM) were prepared in acetonitrile and all aliquots were kept at –20°C. NADPH and UDPGA stock solutions (10–12 mM) were prepared in water before use.

Incubations with Human cDNA-Expressed P450s and HLMs in the Presence of NADPH, and in Human Hepatocytes. [¹⁴C]Muraglitazar was incubated with human cDNA-expressed P450 enzymes (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 3A4, or 3A5) in duplicate. The reaction mixtures consisted of 200 pmol/ml P450 enzyme or 1 mg/ml human liver microsomes (pooled HLM A), 26.6 μM [¹⁴C]muraglitazar, 1 mM NADPH, and 0.1 M sodium phosphate buffer, pH 7.4. The final volume of the incubation mixture was 0.25 ml. After a 30-min incubation at 37°C with shaking (100 rpm), 0.25 ml of ice-cold acetonitrile was added to each tube to stop the reaction. The samples were centrifuged for 10 min at 2000g. After centrifugation, aliquots of 75 μl of supernatant were analyzed by HPLC.

For substrate concentration-dependent metabolite formation studies, [¹⁴C]muraglitazar at seven substrate concentrations (0, 10, 15, 25, 40, 65, and 150 μM) was incubated for 20 min with 1 mg/ml HLM A or cDNA-expressed P450 enzymes supplemented with 1 mg/ml heat-inactivated HLMs at 37°C. The incubation mixtures were prepared in the same way as described above. The amount of each P450 enzyme (CYP2C8, 2C9, 2C19, 2D6, or 3A4) was 40 pmol/ml in each incubation. Under these conditions, the amount of each metabolite formed was linear with the protein concentrations and the incubation time up to 35 min.

[¹⁴C]Muraglitazar (25 μM) was incubated with 2 × 10⁶ cells/ml human hepatocytes in 2 ml of buffer for 3 h at 37°C. Ice-cold acetonitrile (2 ml) was added to stop the reaction. The samples were centrifuged for 10 min at 2000g. After centrifugation, aliquots of 100 μl of supernatant were analyzed by HPLC.

HLM Incubations in the Presence of Chemical P450 Inhibitors.Effects of P450 inhibitors on total muraglitazar metabolism as measured by disappearance of the parent. Chemical P450 inhibitors were incubated with HLM B and muraglitazar in triplicate. Each reaction mixture (0.2 ml) contained 10 μl of 1 M phosphate buffer to a final concentration of 50 mM, 145 μl of water, 10 μl of 100 mM MgCl₂ to a final concentration of 5 mM, 5 μl HLM B to a final concentration of 0.5 mg/ml, and 2 to 5 μl of inhibitors or water. The final pH was 7.4. The mixture was incubated at 37°C for 10 min before 5 μl of 0.0516 mg/ml muraglitazar to a final concentration of 2.5 μM, and 20 μl of NADPH to a final concentration of 1.2 mM were added to initiate the reactions. Chemical inhibitors were montelukast for CYP2C8 (Walsky et al., 2005), sulfaphenazole for CYP2C9 (Newton et al., 1995; Mancy et al., 1996), benzylnirvanol for CYP2C19 (Suzuki et al., 2002), quinidine for CYP2D6, ketoconazole for CYP3A4, and 1-aminobenzotriazole as a general P450 enzyme inhibitor, and the respective concentrations used were 3, 10, 1, 1, 1, and 1000 μM. At these specific concentrations, the chemical inhibitors are specific to the corresponding enzymes. Although ketoconazole at a 1 μM concentration showed a better inhibition selectivity for in vitro experiments, the clinical C_max in human subjects was 5 to 15 μM following 200-mg twice a day or 400-mg once a day dose regimens (Huang et al., 1986). Some of these inhibitors at a higher concentration were also used to test their inhibition selectivity. After the incubation for 30 min at 37°C with shaking (100 rpm) in a water bath, 0.4 ml of acetonitrile containing 3% acetic acid and 1.5 μg/ml of an internal standard (MS/MS transition was m/z 531–292) was added to quench the reactions. The mechanism-based inhibitor, 1-aminobenzotriazole, was preincubated with microsomes in the presence of NADPH for 10 min before muraglitazar was added. After the samples were centrifuged for 15 min at 2000g, the supernatants were diluted with acetonitrile/water (2:1 v/v) before analysis by LC/MS.

Effect of P450 inhibitors on formation of muraglitazar metabolites. Chemical P450 inhibitors were incubated with HLM A and [¹⁴C]muraglitazar in duplicate. Each 1 ml of incubation mixture contained 100 μl of 1 M phosphate buffer, pH 7.4, 100 μl of NADPH (10 mM), 5 μl of selective inhibitors, 5 μl of [¹⁴C]muraglitazar (5.17 mM), 50 μl of HLMs (20 mg/ml), and 740 μl of water. Mechanism-based inhibitors, furafylline, orphenadrine, and 1-aminobenzotriazole, were preincubated with microsomes in the presence of NADPH for 10 min before the substrate was added. The volume of organic solvent (acetonitrile) in the incubations was kept to a minimum (<0.5% v/v). Each of these samples was incubated for 15 min at 37°C while shaking (100 rpm). After the incubation, 1.5 ml of ice-cold acetonitrile was added to each tube to stop the reaction. After the samples were centrifuged for 15 min at 2000g, the supernatants were dried under N₂, and the residues were reconstituted in 300 μl water/acetonitrile (2:1, v/v). An aliquot of 75 μl of each sample was analyzed by HPLC.

HLM B Incubations in the Presence of Inhibitory Antibodies. Antibodies were incubated with HLM B and muraglitazar in triplicate. A reaction mixture (0.2 ml) contained 10 μl of 1 M phosphate buffer to a final concentration of 50 mM (pH 7.4), 145 μl of water, 10 μl of 100 mM MgCl₂ to a final concentration of 5 mM, 5 μl HLM B to a final concentration of 0.5 mg/ml, and 2to5 μl of antibodies or water. The mixture was incubated at 37°C for 10 min before 5 μl of 0.0516 mg/ml muraglitazar was added to a final concentration of 2.5 μM, and 20 μl of NADPH to a final concentration of 1.2 mM were added to initiate the reactions. After the incubation for 30 min at 37°C with shaking (100 rpm) in a water bath, 0.4 ml of acetonitrile containing 3% acetic acid and 1.5 μg/ml of an internal standard (MS/MS transition was m/z 531 to m/z 292) was added to quench the reactions. After the samples were centrifuged for 15 min at 2000g, the supernatants were diluted with 2:1 (v/v) acetonitrile/water before analysis by LC/MS.

Under similar conditions, [¹⁴C]muraglitazar (25 μM) was incubated with HLM B in the presence of 3 μM montelukast, or 5 μl/ml anti-CYP2C8 antibody. After the samples were centrifuged for 15 min at 2000g, the supernatants were analyzed by HPLC with the effluent split 4:1 to a fraction collector for radioactivity determination in conjunction with LC/MS.

HLM A Incubations with cDNA-Expressed UGTs in the Presence of UDPGA. Incubation mixtures (total volume 0.5 ml) contained 25 mM tris(hydroxymethyl)-aminomethane HCl buffer (pH 7.5), 10 mM MgCl₂, 25 μg/ml alamethicin, HLM A (0.8 mg of protein/ml) or UGT enzyme (UGT1A1, 1A3, 1A4, 1A6, 1A8, 1A9, 1A10, 2B4, 2B7, and 2B15) at a protein concentration of 0.1 mg of protein/ml, and 26.6 μM [¹⁴C]muraglitazar. The mixtures were preincubated at 37°C for 5 min; then, UDPGA (2.5 mM) was added to start the reaction. After a 30-min incubation at 37°C with shaking (100 rpm), 0.5 ml of ice-cold acetonitrile was added to stop the reaction. The samples were centrifuged for 15 min at 2000g. After centrifugation, an aliquot of 100 or 200 μl of supernatant was used for HPLC analysis.

For enzyme kinetic studies, HLM A or human cDNA-expressed UGT enzymes (UGT1A1, UGT1A3, or UGT1A9) were incubated with [¹⁴C]muraglitazar in triplicate. The concentrations of [¹⁴C]muraglitazar were 0, 2, 4, 6, 8, 10, 15, 20, and 30 μM. The protein concentration was 0.1 mg/ml for each UGT enzyme or for HLMs. The formation of [¹⁴C]muraglitazar glucuronide was linear up to 40 min (for UGT1A1, UGT1A3, and UGT1A9) at a 10 μM (UGT1A9) or 30 μM (UGT1A1 or UGT1A3) concentration of [¹⁴C]muraglitazar and up to 0.2 mg/ml cDNA-expressed UGT enzyme.

HPLC Analysis. HPLC was performed on a Shimadzu Class VP system equipped with two pumps (model LC-10AT), an auto-injector (SIL 10AD), and a diode array detector (SPD-M10A; Shimadzu, Kyoto, Japan). YMC ODS AQ C-18 4.6 × 150 mm columns (5 μm; Waters, Milford, MA) were used for radioactivity profiling. The HPLC flow was 1 ml/min. HPLC effluent fractions were collected into 96-well LumaPlates at 0.26-min intervals for 70 min after sample injection with a Gilson model 202 fraction collector (Gilson Medical Electronics, Middleton, WI). The plates were dried in a SpeedVac (Thermo Electron, Waltham, MA) and analyzed with a TopCount analyzer (PerkinElmer Life and Analytical Sciences, Boston, MA). Two solvents, A and B, were used for HPLC elution. Solvent A was water containing 0.06% TFA. Solvent B was acetonitrile containing 0.06% TFA. The gradient started with 5% solvent B. The composition of B increased linearly to 25% (5 min), 40% (15 min), 53% (40 min), 60% (3 min), and 90% (2 min), and was then held at 90% for 7 min.

LC/MS/MS Identification of Muraglitazar Metabolites. Mass spectral analysis was performed on a Finnigan LCQ XP mass spectrometer (Thermo Electron) with an ESI probe. Samples were analyzed in the positive ionization mode. A Waters YMC ODS-AQ S-3 μm 120 Å column (2.0 × 150 mm) was used for HPLC and a heater maintained column temperature at 35°C. The mobile phase flow rate was 0.31 ml/min. The HPLC effluent was directed to the mass spectrometer through a divert-valve set to divert the flow to waste from 0 to 5 min. The capillary temperature was set at 210°C. The nitrogen gas flow rate, spray current, and voltages were adjusted to give maximum sensitivity for muraglitazar.

LC/MS/MS Quantification of Muraglitazar in Incubations with HLM B. The concentration of muraglitazar was determined by an LC/MS/MS method. After microsomal incubation mixtures were centrifuged for 5 min, 50 μl of supernatant was diluted with 950 μl of acetonitrile/water (1:1) and 10 μl was injected onto the LC/MS/MS system. A Shimadzu pump system (Shimadzu Corp., Columbia, MD) integrated with a CTC PAL (Carrboro, NC) autosampler and a YMC ODS-AQ column (2 × 50 mm; Waters) was used for chromatographic separation. A gradient with mobile phase A (0.1% formic acid) and mobile phase B (acetonitrile) was used at a flow rate of 0.3 ml/min with effluent split so that 20% was directed to the MS. B% changed as follows: 20% (0 min), 90% (1.6 min), 90% (3.0 min), 20% (3.7 min), and 20% (4.5 min). Detection was by positive ESI on TSQ quantum (Thermo Electron). The multiple reaction monitoring (MRM) transition was m/z 517 to m/z 186 with collision energy of 22 eV for muraglitazar. The standard curve ranged from 0.05 to 5 μg/ml, and was fitted to the a 1/x weighted quadratic regression model (y = 0.0053 + 0.8923*x, r² = 0.9999).

Data Analysis. For each radiochromatographic profile, the average cpm value of the first eight fractions (column void volume) was subtracted from the cpm value of each subsequent fraction. Radioactivity profiles were prepared by plotting the resulting net cpm values against time after injection. Radioactive peaks in the chromatographic profiles were reported as a percentage of the total radioactivity (cpm) recovered during the HPLC run. The amount of each metabolite was calculated based on its percentage distribution and total amount of the parent compound used in the incubation. The amounts of metabolites were averaged from duplicate or triplicate incubations and HPLC analyses. The K_m and V_max values of each enzyme were estimated by fitting the average data to the Michaelis-Menten equation, using a nonlinear regression analysis in SigmaPlot [eq. 1 as recommended in Bjornsson et al. (2004), although the data showed imperfect fitting because of limited substrate concentrations used]. The relative contribution of each P450 enzyme to a particular metabolic pathway was predicted using eqs. 2 and 3. In the above equations, V_i refers to the individual velocity for the formation of a metabolite by each P450 enzyme; V_max and K_m refer to kinetic parameters; S refers to substrate concentrations; v_i refers to the velocity for each P450 enzyme per mg of microsomal protein; A_i refers to the nominal specific content of individual P450 proteins in HLMs as pmol of P450/mg of microsomal protein. The content of each P450 was taken from literature values determined by immunoquantitation (Rodrigues, 1999); F_i refers to the relative contribution of each P450 enzyme to a particular metabolic pathway. The substrate concentration used in the calculation was 1, 2.5, or 3.8 μM, near steady-state C_max plasma concentrations observed after a clinically effective dose of 5, 10, or 20 mg of muraglitazar (data not shown). The relative contribution of each enzyme (F_CYPi) to overall oxidative metabolism was calculated using eq. 4.

Percentage inhibition was calculated based on disappearance of the parent compound or formation of a metabolite. For disappearance of muraglitazar, % Inhibition = (muraglitazar concentration in the presence of inhibitor–muraglitazar concentration in the absence of inhibitor)/(muraglitazar concentration in control–muraglitazar concentration in the absence of inhibitor). For formation of a metabolite, % Inhibition = (amount of the metabolite in the absence of inhibitor–amount of the metabolite in the presence of inhibitor)/(amount of the metabolite in the absence of inhibitor).

Results

In Vitro Metabolite Profiles of [¹⁴C]Muraglitazar.Figure 2 shows the biotransformation profiles of [¹⁴C]muraglitazar in HLMs in the presence of NADPH or UDPGA or in human hepatocytes. M1, M10, M11, M14, and M15 were the major metabolites of [¹⁴C]muraglitazar in the HLM incubations in the presence of NADPH. Multiple radioactive peaks eluted between 17 and 23 min on the HPLC system used in these experiments, and they represented a significant amount of radioactivity in incubations of HLMs in the presence of NADPH and in human hepatocytes. This region included the O-dealkylated metabolite M1. Since there was a minimal amount of M1 found in vivo in humans after oral administration of [¹⁴C]muraglitazar (Wang et al., 2006; Zhang et al., 2006), M1 and other radioactive peaks eluting in this region were not further characterized. In the presence of UDPGA, M13 was formed as a single metabolite in the HLM incubation. M13 was identified as the acylglucuronide metabolite of muraglitazar based on HPLC and MS analysis comparison with previous studies (Wang et al., 2006). In the 3-h incubation with human hepatocytes, both oxidative metabolites and M13 were prominent metabolites, suggesting that both oxidative and conjugative pathways were active. M13 existed as multiple isomers because of acyl migration in this 3-h incubation. Table 1 shows LC/MS identification of the primary metabolites of muraglitazar in human liver microsomes and hepatocyte incubations. M10, M11, M13, M14, and M15 had the same HPLC retention time, molecular ions, and fragmentation patterns as the previously identified metabolites in humans (Wang et al., 2006; Zhang et al., 2006).

Metabolism of Muraglitazar Catalyzed by P450 Enzymes. [¹⁴C]Muraglitazar was incubated with human cDNA-expressed P450 enzymes to determine the catalytic turnover by oxidation. Table 2 shows the metabolite distribution in incubations of [¹⁴C]muraglitazar with individual cDNA-expressed P450 enzymes. Only enzymes CYP2C8, 2C9, 2C19, 2D6, and 3A4 generated significant amounts of metabolites; all other P450 enzymes barely metabolized muraglitazar, and metabolites M10, M11, M14, and M15 were very minor metabolites (<0.5%) in these incubations. Figure 3 shows the biotransformation profiles of [¹⁴C]muraglitazar in CYP2C8, 2C9, 2C19, 2D6, and 3A4. The parent drug was the largest radioactive peak in all incubations. M10 and M15 were major metabolites of CYP2C8. M10 was the major metabolite of CYP2C9. M10 and M11 were major products of CYP2C19. The significant metabolites of CYP2D6 were M10 and M15. M10, M14, and M15 were major metabolites of CYP3A4. CYP3A5 produced very small amounts of M10 and M15 but not M14 (data not shown). One radioactive peak eluted after the parent compound in some incubations was not characterized. M1 was formed in incubations of CYP3A4, 2D6, 2C8, and 2C9. Oxazole-ring opening metabolites M9 and M16 were observed in incubations of CYP2D6, 3A4, 2C9, and 2C8, suggesting that these enzymes might be involved in generation of M9 and M16. These results demonstrated that the major in vitro oxidative metabolites of muraglitazar are predominantly formed by P450 enzymes CYP2C8, 2C9, 2C19, 2D6, and 3A4 and not by CYP1A2, 2A6, 2B6, 2C18, 2E1, or 3A5.

Effects of Enzyme Inhibitors on Oxidative Metabolism of Muraglitazar.Table 3 shows inhibition of muraglitazar metabolism results by anti-P450 antibodies and chemical inhibitors in human liver microsomes by monitoring the disappearance of muraglitazar at 2.5 μM. Anti-CYP1A2, 2A6, 2B6, and 2E1 antibodies inhibited the disappearance of muraglitazar by <10%, and anti-CYP2C8, 2C9, 2Cs, 2D6, and 3A4 significantly inhibited the disappearance of muraglitazar. Approximately 50% of muraglitazar clearance was inhibited by antibodies or chemical inhibitors of 2C enzymes and the other half was inhibited by antibodies or chemical inhibitors of CYP2D6 and 3A4. CYP2C8 antibody and montelukast inhibited muraglitazar clearance by >35%, whereas the CYP2C19 antibody and benzylnirvanol did not seem to inhibit the disappearance of muraglitazar significantly.

Inhibition of muraglitazar disappearance by montelukast and anti-CYP2C8 antibody was mainly due to inhibition of formation of M15, a major metabolite of muraglitazar incubation in HLM B (Fig. 4). This result is consistent with that from cDNA-expressed enzymes showing that CYP2C8 was responsible for formation of M15. However, M15 was not as significant a metabolite in HLM A as in HLM B (Figs. 2 and 4), which might explain the apparent larger contribution of CYP2C8 from the inhibition study in HLM B. The relatively smaller contribution of CYP2C9 and CYP2C19 from the inhibition study in HLM B than that obtained from the scaling of intrinsic clearance of muraglitazar (to be discussed) may also be due to the reduced concentrations of CYP2C9 and CYP2C19 in HLM B compared with HLM A. A reduced concentration of a P450 enzyme would reduce the rate of its metabolism (V_max = k_E, where E is the P450 concentration and k is an intrinsic reactivity constant), which will reduce the apparent inhibition effect by inhibitors when the disappearance of parent compound is monitored.

Inhibition effects of chemical inhibitors and the anti-CYP2C8 antibody on formation of oxidative metabolites M10, M11, M14, and M15 of muraglitazar were evaluated in HLM incubations, and the results are shown in Table 4 and Fig. 4. 1-Aminobenzotriazole, a nonselective time-dependent P450 inhibitor, reduced the formation of all metabolites to near background levels, suggesting that these metabolites were formed by P450 enzymes. Both the anti-CYP2C8 antibody and montelukast appeared to selectively inhibit M15 formation by >70%. Ketoconazole was a specific inhibitor to CYP3A4 at a low concentration (e.g., 1 μM), but also inhibited other enzymes such as CYP2C8, CYP2C9, CYP2C19, and CYP2D6 at higher concentrations (Ong et al., 2000; Stresser et al., 2004). Ketoconazole significantly inhibited formation of all metabolites at a concentration of 10 μM with a greater potency toward inhibition of M14 formation (>91%). Tranylcypromine reduced the formation of M11 to near background level. Sulfaphenazole inhibited formation of M10, M11, and M14 by 30 to 43%. Quinidine appeared to inhibit formation of M11 and M14. Selective inhibitors of CYP1A2 and 2B6, furafylline and ophenadrine, showed minor inhibition for formation of these metabolites, whereas the CYP2A6 inhibitor, 8-methoxypsoralen (Zhang et al., 2001), seemed to lead to a minor activation with formation of M10 and M15 metabolites.

Kinetic Determination for Muraglitazar Oxidative Metabolism. The enzyme kinetic parameters for the formation of the primary oxidative metabolites of muraglitazar by HLMs and expressed P450 enzymes and the predicted relative contribution of each enzyme to metabolism of muraglitazar at different drug concentrations are shown in Table 5. The formation of M10, M11, M14, and M15 in HLMs appeared to follow apparent Michaelis-Menten kinetics with K_m values of 9 to 40 μM (Fig. 5; Table 5). In human liver microsomal incubations, M10 appeared to be formed by multiple enzymes since data points follow multiple trends; in contrast, M11, M14, and M15 appeared to be formed by individual enzymes. These results are consistent with those obtained from incubations with cDNA-expressed enzymes and with inhibitors in human liver microsomes. Figure 6 shows the substrate concentration-dependent formation of metabolites M10, M11, M14, and M15 in the cDNA-expressed enzymes. Based on the calculated intrinsic clearance (CL_int) and relative contribution (F_i) values, CYP2C19 is predicted to be the major enzyme responsible for the 17-hydroxylation of muraglitazar (M11) and to contribute to 36% of muraglitazar total oxidative metabolism. CYP2C8 is predicted to be the major enzyme responsible for O-demethylation of muraglitazar (M15) and to contribute to 12% of muraglitazar total oxidative metabolism. CYP3A4 is predicted to be the major enzyme responsible for 9-hydroxylation of muraglitazar (M14) and to contribute up to 36% of muraglitazar total oxidative metabolism. CYP2C8, 2C9, 2C19, and 3A4 were involved in 12-hydroxylation of muraglitazar. Apparent K_m values for M14 and M15 formation were somewhat lower in HLMs than in the expressed enzymes. Table 5 also shows that although initial metabolism rates V_i and scaled initial rates v_i depend on the substrate concentrations with a given set of A_i (P450 content in HLMs, pmol of P450/mg of microsomal proteins), relative contribution of each enzyme to formation of individual metabolite did not significantly change with the substrate concentrations. Based on the calculation (F_CYPi), the contribution of each enzyme to the overall oxidative metabolism of muraglitazar followed the order of CYP2C19 ≈ 3A4 > 2C8 ≈ 2C9 > 2D6 (Table 6).

Glucuronidation of Muraglitazar. [¹⁴C]Muraglitazar glucuronide was produced in incubations with HLMs, and UGT1A1, 1A3, and 1A9, but not with UGT1A6, 1A8, 1A10, 2B4, 2B7, and 2B15 (Fig. 7). Figure 8 shows the concentration-dependent formation rate for [¹⁴C]muraglitazar glucuronide in UGT1A1, 1A3, and 1A9. There appeared to be substrate inhibition of muraglitazar glucuronidation at substrate concentrations higher than 30 μM (data not shown). The enzyme kinetic parameters of muraglitazar glucuronidation are shown in Table 7. The apparent K_m values of muraglitazar glucuronidation were similar among UGT1A1, 1A3, 1A9, and human liver microsomes (2.1–3.8 μM).

Discussion

The common approaches for identifying the enzyme(s) responsible for metabolic pathways include: 1) incubations with commercially available human cDNA-expressed enzymes to evaluate the metabolic turnover of a compound, 2) inhibition of the in vitro metabolism of the compound in human liver microsomes by enzyme-specific inhibitors (antibodies or chemicals), and 3) correlation studies comparing the P450 or UGT activity toward metabolism of an unknown to a substrate known to be metabolized by a single enzyme in a panel of human liver microsome samples isolated from individual donors (Tang et al., 2005). However, whenever multiple enzymes are involved in the formation of one metabolite or one metabolic pathway, the inhibitor experiments can yield results that are difficult to interpret because of nonspecific inhibition by chemical inhibitors, submaximal inhibition by chemical inhibitors or antibodies, and substrate concentration-dependent inhibition. Correlation studies with a panel of human liver microsomes isolated from individual donors was also very difficult to interpret when multiple enzymes are involved in the metabolism of a given substrate. Because early studies showed that multiple metabolic pathways were important for muraglitazar clearance, expressed enzymes were used as the primary technique to identify the enzymes for metabolism of muraglitazar. An evaluation of the metabolic turnover of muraglitazar with expressed enzymes was conducted to determine the P450 enzymes that demonstrated catalytic activity. Upon initial screening with cDNA-expressed enzymes, multiple P450 (2C8, 2C9, 2C19, 2D6, and 3A4) and UGT (1A1, 1A3, and 1A9) enzymes were found to be involved in the oxidation and glucuronidation of muraglitazar. Furthermore, more than one enzyme was found to catalyze formation of several of the individual metabolites. Further studies with selective chemical P450 inhibitors showed a complicated inhibition pattern; however, the inhibition of formation of the primary oxidative metabolites in HLMs gave results in general agreement with the results found from the studies with the cDNA-expressed enzymes.

Since muraglitazar was oxidized by multiple P450 enzymes, the disappearance of the parent compound was monitored to evaluate the effects of inhibitors on overall metabolism of muraglitazar at a clinically efficacious concentration of 2.5 μM. The results showed that both chemical inhibitors and antibodies of CYP2C8, 2C9, 2D6, and 3A4 inhibited the clearance of muraglitazar in HLMs. However, the inhibition study with anti-CYP2C19 antibody and benzylnirvanol, a CYP2C19 inhibitor, suggested that CYP2C19 did not significantly contribute to the clearance of muraglitazar in the microsomal incubations. From the study with cDNA-expressed enzymes, CYP2C19 was the only enzyme responsible for formation of M11, and the formation of M11 was strongly inhibited by tranylcypromine, another CYP2C19 inhibitor. The formation of M11 was a significant clearance pathway for muraglitazar in humans (Zhang et al., 2006). The reason for this discrepancy is not clear; however, it may be due to 1) the difficulty in measuring small changes in parent disappearance despite the nonlinear reaction conditions used, leading to significant parent depletion (30–40%), and 2) the liver concentration of muraglitazar, which was not reflected by the plasma concentration and was actually higher than 2.5 μM.

The intrinsic clearance (V_max/K_m) values of individual P450s were used to estimate the contributions of each P450 enzyme to the overall oxidative metabolism of muraglitazar and to formation of particular oxidative metabolites. The intrinsic clearance of each P450 enzyme was determined through enzyme kinetic studies with cDNA-expressed enzymes. To study multiple enzymes involved in the formation of one metabolite, Venkatakrishnan et al. (2001a,b) used the relative activity approach to scale from heterologously expressed enzymes to human liver microsomes. In addition, Rodrigues (1999) has described a method to normalize reaction rates measured with individual expressed enzymes with respect to the nominal specific content of the corresponding P450 enzyme in the native human liver microsome samples. This method assumes that catalytic efficiency for a metabolic reaction in cDNA-expressed enzyme is equivalent to that in HLMs for the corresponding enzyme. Different sets of nominal specific enzyme contents for major P450 enzymes in the native human liver microsome samples estimated by immunoblotting have been published (Rodrigues, 1999; Venkatakrishnan et al., 2001a). The intrinsic clearance values determined in incubations with cDNA-expressed enzymes and published values of the relative concentrations of each P450 enzyme expressed in an average human liver (Rodrigues, 1999) were used to estimate the contribution of each enzyme to the oxidative metabolism of muraglitazar. This estimation procedure does not consider the contribution of individual enzymes to formation of minor in vivo metabolites (M1, M9, M16) and a minor component that eluted after the parent in some incubations. M10 (12-hydroxylation) was predicted to be formed predominantly by CYP2C9 and 3A4. M11 (17-hydroxylation), M14 (9-hydroxylation), and M15 (O-demethylation) were predicted to be formed almost exclusively by CYP2C19, 3A4, and 2C8, respectively. Although the predicted contribution of each P450 enzyme to the overall clearance of muraglitazar, based on the scaling of intrinsic clearance, did not fully match the prediction from inhibition studies measuring the disappearance of the parent compound, combined results from both approaches contributed to a better understanding of the clearance mechanisms of muragliatzar.

It is not possible to estimate the relative contribution of each of the three UGT enzymes demonstrated to form M13 in the expressed systems. This is due to the fact that the relative expression levels of different UGTs in expressed enzyme systems as well as in the human liver have not been determined. There is evidence that all three UGTs (UGT1A1, 1A3, and 1A9) that metabolize muraglitazar are expressed in the liver (Tukey and Strassburg, 2000). Highly expressed kidney UGT1A9 may also be involved in formation of M13, a significant urinary metabolite of muraglitazar. The lack of specific UGT inhibitors limits further estimation of the contribution of each UGT enzyme to the in vivo clearance of muraglitazar. Because of similar apparent K_m values for all three UGTs relative to the K_m value in HLMs, their contributions to muraglitazar clearance are subject to their relative expression in tissues such as liver, gut, and kidney.

In summary, muraglitazar was demonstrated to be metabolized by multiple P450-mediated oxidation and UGT-mediated conjugation pathways. These results suggested that the clearance of muraglitazar should not be dependent on a single enzyme in humans and therefore should not be dramatically affected by particular P450 or UGT inhibitors or inducers, or by polymorphisms controlling the activity or expression of a single enzyme. Indeed, coadministration of muraglitazar with ketoconazole, an inhibitor of CYP3A4, and gemfibrozil, an inhibitor of CYP2C8, CYP2C9, CYP2C19, and UGT1A1 (Wen et al., 2001; Prueksaritanont et al., 2002; Wang et al., 2002; Ogilvie et al., 2006) in human subjects did not produce substantial changes in muraglitazar exposure (data not shown), which further supports the overall conclusion drawn from these in vitro experiments.