Characterization of Cytochrome P450-Mediated Bioactivation of a

Aug 21, 2009 - mediating the activation of Cmpd A to CPIC in human liver microsomes based ... Evidence of the bioactivation to CPIC in ViVo was obtain...
0 downloads 0 Views 793KB Size
Download PDF
Chem. Res. Toxicol. 2009, 22, 1603–1612

1603

Characterization of Cytochrome P450-Mediated Bioactivation of a Compound Containing the Chemical Scaffold, 4,5-Dihydropyrazole-1-carboxylic acid-(4-chlorophenyl amide), to a Chemically Reactive p-Chlorophenyl Isocyanate Intermediate in Human Liver Microsomes Hao Chen,*,†,§ Michael Zientek,†,| Mehran Jalaie,‡,| Yanhua Zhang,†,⊥ Christopher Bigge,‡ and Abdul Mutlib†,§ Departments of Pharmacokinetics, Dynamics and Metabolism, and Chemistry, Pfizer Global Research and DeVelopment, Michigan Laboratories, 2800 Plymouth Road, Ann Arbor, Michigan 48105 ReceiVed May 13, 2009

Compound A (Cmpd A) was previously reported to form p-chlorophenyl isocyanate (CPIC), which was trapped by GSH to yield S- (N- [p-chlorophenyl] carbamoyl) glutathione adduct (SCPG) in the presence of human liver microsomes. In this study, P450 3A4 and 2C9 were demonstrated to be the enzymes mediating the activation of Cmpd A to CPIC in human liver microsomes based on inhibitory and correlation studies. Enzyme kinetics studies indicated that P450 3A4 was the primary enzyme involved in the activation of Cmpd A. In silico P450 3A4 active site docking of Cmpd A exhibited a low energy pose that orientated the pyrazole ring proximate to the heme iron atom, in which the distance between the C-3 and potential activated oxygen species was shown to be 3.4 Å. Quantum molecular calculations showed that the electron density on C-3 was relatively higher than those on C-4 and C-5. These measurements suggested that the C-3 of Cmpd A was the preferred site of oxidation and hence predisposed Cmpd A in forming CPIC as previously proposed. The in silico prediction was corroborated by studies with the C-3 substituted analogue (methyl at C-3), which showed minimal conversion to CPIC in human liver microsomes. These results demonstrated a pivotal role for P450 3A4 in bioactivating Cmpd A by oxidizing at C-3 of the pyrazoline, hence facilitating the CPIC formation. Evidence of the bioactivation to CPIC in ViVo was obtained by liquid chromatography-mass spectrometry (LC/MS) analysis of urine samples from human subjects administered a structural analogue of Cmpd A. The presence of S-(N-[p-chlorophenyl] carbamoyl) N-acetyl L-cysteine (SCPAC) as well as p-chlorophenyl aniline (CPA) was unequivocally demonstrated in the urine samples. The chemical scaffold, 4,5-dihydropyrazole-1-carboxylic acid-[(4-chlorophenyl)-amide], was demonstrated to possess potential metabolic liability in forming a reactive intermediate, CPIC, in humans. Bioactivation to CPIC may cause undesirable side effects through its reactivity and subsequent conversion to CPA, an established rodent carcinogen. Introduction Drug-induced adverse reactions are significant health problems and believed to be one of the leading causes of death in the United States (1). Drugs continue to be withdrawn from the market, or have their use severely restricted because of unexpected toxicities that become apparent only after they are launched. Hepatotoxicity is the primary safety reason for the withdrawal of drugs postmarketing and the termination of clinical drug trials. A recent study showed that more than 600 drugs have been associated with hepatotoxicity, and greater than 50% of cases of acute liver failure were attributed to the use of drugs (2). It has been estimated that approximately 30% of new * Corresponding author. Tel: (484) 865-2385. Fax: (484) 865-9404. E-mail: [email protected]. † Department of Pharmacokinetics, Dynamics and Metabolism. ‡ Department of Chemistry. § Present address: Drug Safety and Metabolism, Wyeth Research, 500 Arcola Rd., Collegeville, PA 19426. | Present address: Pfizer Global Research and Development, Science Center Dr., San Diego, CA 92121. ⊥ Present address: Pfizer Global Research and Development, Eastern Point Rd., Groton, CT 06340.

chemical entities fail due to toxicity (3), resulting in increased costs and delays in successful drug development. Although the biochemical mechanisms of drug-induced toxicities are not fully understood and remain a challenge for the pharmaceutical industry, a large amount of circumstantial evidence suggests that chemically reactive metabolites of a drug, rather than the parent drug, are responsible for toxicities (4-10). It has been recognized that reactive metabolites formed from the parent compound initiate damages to cellular macromolecules that, depending on the nature of the damage (covalent binding, oxidative stress, etc.), cause changes in cellular signaling, regulatory, and other pathways that subsequently lead to cytotoxicity (11-16). Consequently, substantial effort in a number of pharmaceutical companies has been made to mitigate or minimize the potential of metabolic activation of new chemical entities, particularly by P450 enzyme(s), to reactive metabolites at the various stages of drug discovery and development (17-21). In general, the formation of electrophilic reactive metabolites is often examined through appropriate in Vitro trapping experiments, typically human and animal liver microsomal incubations fortified with a nucleophile(s) such as reduced glutathione (GSH), a major scavenger of reactive

10.1021/tx900167y CCC: $40.75  2009 American Chemical Society Published on Web 08/21/2009

1604

Chem. Res. Toxicol., Vol. 22, No. 9, 2009

Chen et al.

Scheme 1. P450-Mediated Bioactivation of Cmpd A to Reactive CPICa

a

B: general base without a known identity.

metabolites in ViVo. Once the metabolic activation of lead compounds is established, the identity of reactive intermediate(s) is often inferred from the structural characterization of the trapped product(s), i.e., the thioether conjugate(s) if GSH is employed, and subsequently, the mechanism can be rationalized. The information is valuable to medicinal chemists who can then seek ways to design out or mitigate the potential for reactive metabolite liability through structural modification/optimization. For potential drug candidates being evaluated in preclinical and early clinical stages, demonstration of reactive metabolite formation, understanding of reactive metabolite(s) and its downstream product(s), and an insight into the enzymes responsible for the bioactivation, particularly P450 enzymes, offer invaluable information to aid the risk assessment and the decision-making on the selection or advancement of such candidates for successful development. As part of a preliminary screening strategy, compound A (Cmpd A)1 (Scheme 1), 4,5-dihydropyrazole-1,5-dicarboxylic acid 1-[(4-chlorophenyl)-amide] 5-[(2-oxo-2H-[1,3′] bipyridinyl6′-yl)-amide, was investigated for potential metabolic activation to chemically reactive intermediate(s). It was found that a chlorophenyl carbamoyl glutathione conjugate was formed during the metabolism of Cmpd A in Vitro in rat and human 1 Abbreviations: Cmpd A, compound A; CPIC, p-chlorophenyl isocyanate; CPA, p-chlorophenyl aniline; SCPG, S-(N-[p-chlorophenyl] carbamoyl)glutathione; SCPAC, S-(N-[p-chlorophenyl] carbamoyl) N-acetyl L-cysteine; SPG, S-(N-[phenyl] carbamoyl)glutathione; FAL, furafylline; COUM, coumarin; SPA, sulfaphenazole; MEPH, S-methylphenytoin; QND, quinidine; DDC, diethyldithiocarbamate; TAO, troleandomycin; KTO, ketoconazole; mAb, monoclonal antibody; pmol, picomole; ng, nanogram; ESI, electrospray ionization; LC/MS, liquid chromatography mass spectrometry; MS/MS, tandem mass spectrometry; LC/MRM, liquid chromatography multiple reaction monitoring; LC/EC/MRM, liquid chromatography-electrochemistry-multiple reaction monitoring.

liver microsomes as well as in ViVo in rats (22). This finding indicated the possible formation of p-chlorophenyl isocyanate (CPIC) as the reactive intermediate produced during the metabolism of Cmpd A both in Vitro and in ViVo. These results also revealed that the chemical scaffold, 4,5-dihydropyrazole1-carboxylic acid-[(4-chlorophenyl)-amide], was metabolically labile and apparently contributed to the formation of the isocyanate intermediate. It was further demonstrated that metabolic intervention at C-3 of pyrazoline was a prerequisite for predisposing the compound toward isocyanate formation in rat liver microsomes. As depicted in Scheme 1 (22), the chemical mechanism leading to CPIC is postulated to involve an initial P450-mediated hydroxylation of the C-3 of pyrazoline, which leads to a 3-oxo derivative. The introduction of an electron-withdrawing amide-type group may destabilize the molecule in a manner that could result in decomposition to CPIC. The resultant CPIC reacts with GSH to form SCPG or with water to afford p-chlorophenyl aniline (CPA). The bioactivation of Cmpd A to CPIC was previously demonstrated in human liver microsomes (22). This prompted us to conduct further studies to delineate the bioactivation process. One of our goals was to characterize human cytochrome P450 enzyme(s) responsible for the activation so that the contribution of individual P450 enzymes and the potential for metabolic drug-drug interactions could be evaluated. The kinetics of activation of Cmpd A to CPIC in human liver microsomes was investigated to gain an insight into the relative contributions of various enzymes to the activation process. Since direct measurement of CPIC formation was not practical because of its reactivity, the kinetics of CPIC formation was indirectly studied by monitoring the formation of the CPIC-derived GSH conjugate from Cmpd A. To further evaluate the mechanism of bioactivation, in silico P450 active site docking and quantum

P450-Mediated BioactiVation to Isocyanate

molecular calculations were performed to further justify the chemistry effort in mitigating the bioactivation process. Finally, the in ViVo relevance of bioactivation to CPIC in clinics was investigated by analyzing human urine samples in an attempt to demonstrate the presence of thioether conjugates and related metabolic end-point products. These urine samples were obtained from human subjects who received an oral dose of a structural analogue of Cmpd A, which consisted of the same chemical scaffold, 4,5-dihydropyrazole-1-carboxylic acid-[(4chlorophenyl)-amide.

Materials and Methods Caution: p-chlorophenyl aniline (CPA), obtained from SigmaAldrich Chemical Co., is a highly toxic and care should be taken in the handling, analysis, and disposal of this substance. Chemicals and Reagents. Compound A and its structural analogues described herein, unless otherwise noted, were synthesized by the Chemistry Department of Pfizer Global Research and Development (Ann Arbor, MI) and fully characterized. The purity of each compound was >95% on the basis of HPLC analysis. Thioether conjugates including S-(N-[p-chlorophenyl] carbamoyl) glutathione (SCPG), S-(N-[p-chlorophenyl] carbamoyl) N-acetyl L-cysteine (SCPAC), and S-(N-[phenyl] carbamoyl) glutathione (SPG) were previously synthesized and characterized (22). Selective P450 inhibitors including coumarin (COUM), sulfaphenazole (SPA), S-methylphenytoin (MEPH), quinidine (QND), diethyldithiocarbamate (DDC), and troleandomycin (TAO) were obtained from Sigma-Aldrich Chemical Co. (Milwaukee, WI). Furafylline (FAL) was purchased from Research Biochemicals International (Natick, MA). Ketoconazole (KTO) was obtained from ICN Biomedicals Inc. (Aurora, OH). Waters Symmetry C18 columns (2.1 × 50 mm, 5 µm) were obtained from Waters Corporation (Milford, MA). All solvents and reagents were of the highest grade commercially available. Mixed human liver microsomes (pooled from 10 individuals), cDNA-expressed human P450 microsomes, and monoclonal antibody (mAb) against human P450 enzymes (1A2, 2A6, 2C9, 2C19, 2D6, 2E1, and 3A4) were purchased from BD Biosciences (Woburn, MA). Each cDNA-expressed P450 microsome contained cDNA-expressed human P450 reductase, human cytochrome b5, and a specific human P450 enzyme such as 1A2, 2A6, 2B6, 2C9*1, 2C18, 2C19, 2D6*1, 2E1, and 3A4. Human P450 oxidoreductase insect control and the preimmune IgG were also obtained from BD Biosciences. Individual human liver microsomes designated HLA through HLT were also obtained from BD Biosciences. The activities of P450 individual enzymes (1A2, 2B6, 2C9, 2C19, 2D6, and 3A4) in each human liver microsome were characterized by the vendor through the use of isoform-selective substrates. Characterization of the P450 Enzymes Responsible for Cmpd A Activation. To investigate the P450 enzymes responsible for the bioactivation of Cmpd A to CPIC, which was trapped as the GSH conjugate, S-(N-[p-chlorophenyl] carbamoyl) glutathione (SCPG), in Vitro studies were conducted using commercially available cDNA-expressed human P450 enzymes, selective P450 inhibitors, mAb against P450 enzymes, and correlation assay with a bank of individual human liver microsomes. The incubation reagents including MgCl2, GSH, and NADPH were freshly prepared in 0.1 M phosphate buffer (pH 7.4). Cmpd A was dissolved in methanol (2 mM) and used for the incubations described below. All microsomal incubations were conducted in microcentrifuge tubes (1.5 mL) at 37 °C in a water bath agitated at a constant speed. The reaction was terminated by the addition of 50 µL of ice-cold acetonitrile containing SPG (the internal standard, 1 µg/mL) and acetic acid (0.1%, v/v). The samples were vortexed and centrifuged at 1600g for 5 min. The supernatant was removed, and aliquots were analyzed by liquid chromatography multiple reaction monitoring (LC/MRM) to quantitate SCPG as described below. The significance of inhibition was tested by one-tailed unpaired t test using GraphPad Prism 3.0 software (GraphPad Software, La Jolla, CA).

Chem. Res. Toxicol., Vol. 22, No. 9, 2009 1605 Incubations with cDNA-Expressed Human P450 Enzymes. To initially screen for human P450 enzymes involved in the formation of SCPG, incubations (in duplicate) were conducted with commercially available cDNA-expressed human P450s including 1A2, 2A6, 2B6, 2C9*1, 2C18, 2C19, 2D6*1, 2E1, and 3A4. The incubation mixture consisting of the P450 (20 pmol), MgCl2 (3 mM), GSH (0.5 mM), and Cmpd A (10 µM) was preincubated at 37 °C for 5 min in the phosphate buffer. The reaction was initiated by adding NADPH (1 mM). The volume of incubation was adjusted to 0.2 mL with the phosphate buffer containing MgCl2. The mixture was incubated for 20 min at 37 °C. The control experiment was performed with human P450 oxidoreductase insect control. Experiments with Selective P450 Inhibitiors. A number of selective P450 inhibitors were employed to assess the contribution of individual P450 enzymes responsible for SCPG formation in human liver microsomes. The incubations, performed in triplicate, comprised pooled human live microsomes (0.4 mg), MgCl2 (3 mM), GSH (0.5 mM), NADPH (1 mM), Cmpd A (10 µM), and various chemical inhibitors. The following inhibitors were examined for their effects: FAL (20 µM), COUM (100 µM), SPA (10 µM), MEPH (100 µM), QND (10 µM), DDC (300 µM), TAO (50 µM), and KTO (1 µM). The stock solutions of the chemical inhibitors were prepared in dimethyl sulfoxide and subsequently diluted with methanol to appropriate concentrations prior to the incubation. The final concentration of methanol in the incubation was below 1.0% (v/v), and the concentration of dimethyl sulfoxide was kept less than 0.1% (v/v). For the mechanism-based inactivators including FAL, DDC, and TAO, each inhibitor was preincubated with the microsomes in the phosphate buffer containing MgCl2 at 37 °C for 5 min prior to the addition of NADPH. The reaction was run at 37 °C for 15 min before adding Cmpd A and GSH and continued at 37 °C for 20 min. The volume of incubation was adjusted to 0.2 mL with the phosphate buffer. In parallel, a control experiment was performed using blank solvent for the preincubation. For nonmechanism-based inactivators including COUM, SPA, MEPH, QND, and KTO, each inhibitor or the blank solvent used as the control was coincubated with the microsomes, GSH, and Cmpd A in the phosphate buffer containing MgCl2 at 37 °C for 5 min before adding NADPH. The volume of incubation was adjusted to 0.2 mL with the phosphate buffer. The incubations were carried out at 37 °C for 20 min. Experiments with Inhibitory P450 Antibodies. To verify the role of individual P450 enzymes in forming SCPG, a study was conducted using inhibitory antibodies against specific P450 enzymes. Incubations were performed in triplicate consisting of the pooled human liver microsomes (0.2 mg), mAb (20 µL, 0.2 mg protein), MgCl2 (3 mM), Cmpd A (10 µM), GSH (0.5 mM), and NADPH (1 mM). Microsomes were preincubated with individual antibodies for 30 min at room temperature in the phosphate buffer containing MgCl2, followed by the addition of Cmpd A, GSH, and NADPH. The final volume was adjusted to 0.2 mL with the phosphate buffer. The incubations were carried out at 37 °C for 20 min. The control was conducted using the preimmune IgG under the same experimental condition. Correlation Study. Twenty individual human liver microsome preparations, previously characterized with respect to phenacetin O-deethylation (P450 1A2), S-mephenytoin O-demethylation (P450 2B6), diclofenac 4′-hydroxylase (P450 2C9), S-mephenytoin 4′hydroxylase (P450 2C19), bufuralol 1′-hydroxylase (P450 2D6), and testosterone 6β-hydroxylase (P450 3A4) activities, were employed for the correlation study. The activities of individual P450 in these human live microsomes were provided by the vendor. The rates of SCPG formation from Cmpd A in human live microsomes were determined from the incubations (in triplicate) of Cmpd A (10 µM) with individual human liver microsomal protein (0.4 mg) in the presence of GSH (0.5 mM). The incubation mixture was preincubated at 37 °C for 5 min in 0.1 M phosphate buffer (pH 7.4) containing MgCl2 before adding NADPH (1 mM). The final volume was adjusted to 0.2 mL with the phosphate buffer. The reaction was run at 37 °C for 10 min. Under these experimental conditions, the formation of SCPG was linear with respect to P450

1606

Chem. Res. Toxicol., Vol. 22, No. 9, 2009

concentration and incubation time. The correlation between the rates of SCPG formation (the mean value of three measurements) and the activity of individual P450 enzymes on the conversion of probe substrates for these P450 enzymes was examined. Linear regressions were performed using GraphPad Prism 3.0 software (GraphPad Software, La Jolla, CA). Enzyme Kinetic Studies of Cmpd A Activation. The kinetic features of Cmpd A activation to CPIC in human liver microsomes were studied by determining the relationship between the rates of SCPG formation and various concentrations of Cmpd A. It was assumed that initial P450-mediated oxidation at C-3 of Cmpd A, which was proposed to be the metabolic reaction ultimately leading to CPIC formation (Scheme 1), was the rate-limiting step during the overall activation process. The subsequent steps involving the decomposition of oxidized metabolite to CPIC and conjugation with GSH to yield SCPG were considered to be spontaneous chemical reactions (Scheme 1). Under the conditions employed for the kinetic studies, these two reactions were presumed not to be rate limiting and hence did not contribute to the overall kinetics of the activation. A similar approach used to study the kinetics of P450-mediated bioactivation to reactive intermediates has been previously described (22, 23). In preliminary experiments, the incubation times and the concentrations of microsomal proteins were studied to establish the linearity of SCPG formation. It was found that the formation of SCPG from Cmpd A in the presence of human liver microsomes increased linearly with microsomal protein concentration up to 4.0 mg/mL and with incubation time up to 20 min. Subsequently, the kinetics of SCPG formation in human liver microsomes was determined with Cmpd A concentrations ranging from 10 to 500 µM. Incubations, conducted in duplicate in microcentrifuge tubes (1.5 mL), were performed using the following protocol: human liver microsomal protein (0.4 mg), Cmpd A (various concentrations), MgCl2 (3 mM), GSH (0.5 mM), and NADPH (1 mM) in a final volume adjusted to 0.2 mL with 0.1 M phosphate buffer (pH 7.4). The individual human liver microsome preparations randomly chosen for the kinetics studies were HLB, HLK, HLO, and HLQ. A series of methanol solutions of Cmpd A at various concentrations were freshly prepared and used for the incubations. The reaction was run at 37 °C for 10 min and terminated by the addition of 50 µL ice-cold acetonitrile containing SPG (1 µg/mL) and acetic acid (0.1%, v/v). Additionally, the kinetics of SCPG formation in cDNAexpressed P450 3A4 was studied with Cmpd A concentrations ranging from 0.05 to 500 µM. Incubations, conducted in duplicate in microcentrifuge tubes (1.5 mL), were conducted using the same protocol as described above, except with 10 pmol (0.2 mg) of P450 3A4 instead of human liver microsomes. The reaction was carried out for 7 min at 37 °C and terminated by the addition of 50 µL of ice-cold acetonitrile containing SPG (1 µg/mL) and acetic acid (0.1%, v/v). After termination, the incubation mixture was vortexed and centrifuged at 1600g for 5 min. The supernatant was removed and aliquots analyzed by LC/MRM for SCPG quantitation as described below. Kinetic parameters of SCPG formation with either individual human liver microsomes or the recombinant P450 3A4 were estimated using nonlinear regression analysis of data by fitting to the Michaelis-Menten equation V ) (Vmax·S)/(Km + S). Eadie-Hofstee plots were used to examine the presence of biphasic kinetics. If the Eadie-Hofstee plot indicated biphasic kinetics, an alternative two-enzyme model V ) (Vmax1 × S)/(Km1 + S) + (Vmax2 × S)/(Km2 + S) was applied to the respective data. Data fitting and analysis were performed using SigmaPlot 7.1 (SPSS, Inc., Chicago IL). P450 3A4 Active Site Docking Studies. Docking studies were conducted using an in-house docking and scoring software (24-26). Fifty restrained docking poses in the active site of APO P450 3A4 (Protein Data Bank code: TQN-1) was performed for Cmpd A. The C-3 of the pyrazoline was anchored near a water molecule located 2.5 Å above the heme iron, which acted as a surrogate for the activated oxygen species during catalysis. The docking protocol was designed to allow for the protein flexibility and minimization of binding energy, and thus, the binding poses with low energy

Chen et al. could be identified. The binding energy was estimated using the Pfizer Pairwise Linear Potential (PLP) scoring function (24). Quantum Molecular Calculations. Spartan modeling package (Wave function, Inc. Irvine, CA 92612) was employed to conduct SCF Hartree-Fock calculations. The 6-311-G* level of theory was selected for predicting the equilibrium energy of the molecular ground state. The electrostatics potential of the molecule was calculated and projected on the Connolly surface. Substitution Effect on the Activation. Microsomal incubations were conducted using the following protocol: pooled human liver microsomes (0.4 mg), MgCl2 (3 mM), GSH (0.5 mM), NADPH (1 mM), Cmpd A, or its methyl-substituted analogues (10 µM), with the final volume adjusted to 0.5 mL with 0.1 M phosphate buffer (pH 7.4). The incubation mixture was mixed in glass tubes and preincubated at 37 °C for 5 min before adding NADPH. The reaction was run at 37 °C for 20 min in a water bath agitated at a constant speed and terminated by the addition of 1 mL of ice-cold acetonitrile containing acetic acid (0.1%, v/v). The samples were vortexed and centrifuged at 1600g for 10 min. The supernatant was separated, dried under N2, and reconstituted in 200 µL of HPLC mobile phase (50% acetonitrile in 10 mM ammonium acetate, v/v). Aliquots of the sample were analyzed by LC/MRM for SCPG formation. Analysis of Human Urine. Urine samples were obtained from subjects administered single oral doses of an analogue of Cmpd A. After an overnight fast (>10 h), three subjects received an oral dose (150 mg) of the compound, and the other three subjects were given the placebo. All subjects remained in the clinical facility for sample collection. Postdose urine samples were collected and pooled from 0-24 h and 24-48 h and stored at -80 °C until assayed. The clinical trial was sponsored by Pfizer and conducted in compliance with the regulations for clinical studies of new chemical entities and good clinical practice. The study protocol was approved by the internal and external clinical study protocol review board. The investigator brochure was provided to the medical staff who conducted the study. Subjects were provided informed written consent prior to admission to the clinical facility. Prior to analysis by LC/MS, urine samples were thawed at room temperature and mixed well by vortexing. Aliquots (5 mL) were taken and centrifuged at 1600g for 5 min. Aliquots (50 µL) of supernatants were mixed with 150 µL of 50% acetonitrile in 10 mM ammonium acetate (v/v), and aliquots (10 µL) were analyzed by LC/MRM to detect the presence of CPIC-related thioether conjugate, for instance, SCPAC. In parallel, aliquots (190 µL) of supernatants were mixed with 10 µL of 50% acetonitrile in 10 mM ammonium acetate (v/v), and aliquots (10 µL) were analyzed by liquid chromatography electrochemistry multiple reaction monitoring (LC/EC/MRM) to detect the presence of CPIC-related metabolic end-point product, CPA (see below). LC/MS. To detect and quantitate CPIC-derived thioether conjugates present in various biological matrices, a quadrupole linear ion-trap mass spectrometer API 4000 Q-Trap (PE-Sciex, Toronto, Ontario) that was coupled to an Agilent 1100 HPLC system was used. The HPLC effluent was introduced into the ion source using a turbo ionspray interface. The mass spectrometer was operated in the positive ion mode for LC/MS analysis. The output signal from the mass spectrometer was interfaced to a computer operating Analyst 1.4 software (PE-Sciex) for data collection, peak integration, and analysis. The chromatography was conducted on a Waters Symmetry C18 column (2.1 × 50 mm, 5 µm) that was eluted with a gradient solvent system consisting of acetonitrile and 10 mM ammonium acetate (pH 5). The percentage of acetonitrile was increased linearly from 10 to 28% over 4 min followed by a linear ramp to 65% in 1 min at a flow rate set at 0.20 mL/min. After 5 min, the column was re-equilibrated with the initial mobile phase for 2 min before the next injection. Detection and quantitation of SCPG was achieved through an optimized and validated LC/MRM method. The mass transitions monitored during the LC/MRM analysis included m/z 461 f 332 for SCPG and m/z 427 f 298 for SPG. Quantitation was conducted using a standard calibration curve of SCPG over a concentration range from 10 to 800 ng/mL.

P450-Mediated BioactiVation to Isocyanate

Chem. Res. Toxicol., Vol. 22, No. 9, 2009 1607 Table 1. Inhibition of SCPG Formation from Cmpd A in Human Liver Microsomes in the Presence of Selective P450 Inhibitors a

inhibitor control FAL COUM SPA MEPH QND DDC TAO KTO

Figure 1. Formation of SCPG from Cmpd A in the presence of various cDNA-expressed human P450 enzymes.

A weighted (1/x2) linear regression of SCPG concentrations and measured peak area ratios of SCPG to SPG was used to construct the calibration curve. Detection of SCPAC present in human urine samples was also accomplished by LC/MRM analysis using the same LC/MS instrument and HPLC conditions as those described above. The mass transition m/z 317 f 164 was monitored for SCPAC during the LC/MRM analysis. LC/EC/MRM. To detect the CPA present in human urine, the LC/MS system described above was coupled to an electrochemical system (ESA Biosciences, Chelmsford, MA). The combination of electrochemistry with LC/MS for detection and quantitation of CPA present in biological samples has been previously described by us (27). Briefly, the electrochemical system consisted of a GuardStat potentiostat and a model 5021 conditioning cell. The cell contained a coulometric working electrode and a Pd reference electrode. To couple LC/MS with electrochemical oxidation, the conditioning cell was placed between the column and the electrospray ionization (ESI) interface of the mass spectrometer. The eluent from the cell was introduced into the turbo ionspray interface. CPA was electrochemically converted to a dimer derivative, which exhibited a parent ion ([M + H]+) at m/z 217 with a characteristic one-chlorine M + 2 isotope cluster. The electrochemical conditions for CPA dimerization were optimized for the maximal signal intensity of the mass transition m/z 217 f 182, in which the ion at m/z 182 was exhibited to be the most abundant fragment ion by the tandem mass spectrometric analysis (MS/MS) of the CPA dimer. Subsequently, the analysis of CPA employing liquid chromatography-electrochemistry-multiple reaction monitoring (LC/EC/ MRM), in which the mass transition m/z 217 f 182 was monitored, was developed, optimized, and validated to detect the CPA present in human urine samples. The chromatographic analysis was achieved on a Waters Symmetry C18 column (2.1 × 50 mm, 5 µm). The mobile phase consisted of acetonitrile and 10 mM ammonium acetate (pH 8), which was delivered through a gradient program. The percentage of acetonitrile was increased from 25 to 75% (v/v) over 4 min. After 4 min, the percentage of acetonitrile was ramped to 85% in 2 min, then back to 25% in 30 s to re-equilibrate for 2 min before the next injection. The flow rate was set at 0.25 mL/ min.

Results P450 Enzymes Responsible for Cmpd A Activation. Studies with nine cDNA-expressed human P450 enzymes demonstrated a marked difference in their ability to form SCPG from Cmpd A. The highest yield of SCPG was obtained with P450 3A4 as compared to that with other enzymes at the same P450 concentration of 100 pmol/mL per enzyme (Figure 1). Other enzymes including P450 2C9*1, P450 2C19, and P450 2D6*1 also produced SCPG, but the yields were substantially lower as compared to that of P450 3A4. Small quantities of SCPG were also produced in the presence of P450 1A2, 2A6, 2B6, 2C18, and 2E1.

SCPG formation

20 µM 100 µM 10 µM 100 µM 10 µM 300 µM 50 µM 1 µM

100 ((6) 109 ((9) 110 ((5) 62 ((4) 76 ((7) 84 ((4) 86 ((5) 14 ((6) 12 ((8)

a Values are the mean ((SD) of triplicates and expressed as percentage relative to the control without the inhibitor.

Figure 2. Inhibition of SCPG formation from Cmpd A in human liver microsomes in the presence of mAb against various human P450 enzymes (values are the mean of triplicates and expressed as percentage relative to the preimmune IgG control).

The effect of selective P450 inhibitors on SCPG formation from Cmpd A in pooled human liver microsomes is summarized in Table 1. Preincubation of human live microsomes with FAL (20 µM) or DDC (300 µM), mechanism-based inactivators of P450 1A2 and P450 2E1, respectively, resulted in little effect on the formation of SCPG as compared to the control. In contrast, preincubation with TAO (50 µM), a mechanism-based inactivator of P450 3A4, led to substantial inhibition of SCPG formation from Cmpd A (approximately 86% reduction as compared to the control). Coincubation of Cmpd A with SPA (10 µM) or MEPH (100 µM) in human live microsomes, competitive inhibitors of P450 2C9 and 2C19, respectively, showed 38% and 24% reduction on SCPG formation, while coincubation with QND (10 µM), a competitive inhibitor of P450 2D6, exhibited a small (16%) but statistically important reduction (p < 0.05) in SCPG formation. However, coincubation of human live microsomes and Cmpd A with COUM (100 µM), a competitive inhibitor of P450 2A6, led to no inhibition of SCPG formation. Consistent with the effect of TAO, coincubation with KTO (1 µM), a competitive inhibitor of P450 3A4, produced a significant inhibition of SCPG formation; approximately 88% reduction was achieved as compared to the control. Further studies with specific antibodies directed against human P450 enzymes were conducted to confirm the isoenzyme(s) responsible for the bioactivation process. The production of SCPG was reduced by approximately 46%, 29%, and 47% (Figure 2), respectively, as compared to the preimmune IgG control, when human liver microsomes were treated separately with the antibodies against human P450 2C9, 2C19,

1608

Chem. Res. Toxicol., Vol. 22, No. 9, 2009

Chen et al.

Figure 3. Correlation between the rates of SCPG formation from Cmpd A in individual human liver microsomes and the catalytic activity of P450 3A4 using testosterone as the probe substrate (6β-hydroxylation).

and 3A4. In contrast, human liver microsomes that were treated with the antibodies against human P450 1A2, 2A6, 2D6, and 2E1 showed little effect on the formation of SCPG. To further substantiate the involvement of P450 3A4 and other isoenzymes in the bioactivation process, the rates of SCPG formation were determined in a P450-characterized bank of 20 individual human live microsomes using 10 µM Cmpd A. The average rate of SCPG formation in these microsomes was 5.2 ( 3.8 pmol/min/mg microsomal protein. A 12-fold variation in the rates of SCPG formation was demonstrated (range, 0.8-10.3 pmol/min/mg). A significant correlation (p < 0.05, r ) 0.927) between the rates of SCPG formation and the catalytic activity of P450 3A4 on the 6β-hydroxylation of testosterone was demonstrated (Figure 3). In addition, a significant correlation (p < 0.05, r ) 0.753) was also shown between the rates of SCPG formation and the catalytic activity of P450 2C9 using diclofenac as the probe substrate (data not shown). However, significant correlation between the rates of SCPG formation and the catalytic activities of P450 enzymes was not demonstrated by other P450 enzymes including 1A2, 2B6, 2C19, and 2D6 (data not shown). Enzyme Kinetic Studies of Cmpd A Activation. The activation kinetics of Cmpd A to CPIC was studied in four individual human liver microsomes as well as with the recombinant P450 3A4. It was found that the relationship between the rates of SCPG formation and concentrations of Cmpd A in four individual human liver microsome incubations was hyperbolic when the data (the rates of SCPG formation vs concentrations of Cmpd A) was initially fitted to the MichaelisMenten equation. A representative substrate-velocity curve is shown in Figure 4. The insert shows the Eadie-Hofstee plot, which displays a biphasic profile (Figure 4), suggesting at least two distinct sets of enzymes with different kinetic properties involved in the activation of Cmpd A in human liver microsomes. As a result, the kinetic parameters of SCPG formation in these human liver microsomes were obtained using a dualenzyme Michaelis-Menten model. The apparent Km and Vmax of four individual human liver microsomes are presented in Table 2. The turnover numbers estimated as the ratio of Vmax/ Km are also listed in Table 2. The average Km value of the enzyme with low Km values in human liver microsomes is 26 µM, which is much lower (17-fold) than the average Km (437 µM) of the other enzyme. The average turnover number associated with the P450 enzyme with low Km values is approximately 2-fold larger than that of the enzyme with high

Figure 4. Representative substrate-velocity Michaelis-Menten curve for SCPG formation from Cmpd A in the presence of human liver microsomes (insert is the Eadie-Hofstee plot).

Table 2. Kinetic Parameters of SCPG Formation from Cmpd A in Individual Human Liver Microsomes and cDNA-Expressed P450 3A4a microsomal sample HLB HLK HLO HLQ mean ( SD P450 3A4

d b

Km1

c

Vmax1

turnover number

d b

Km2

c

Vmax2

turnover number

34.7 58.1 1.67 515 467 0.91 14.6 19.9 1.36 168 93 0.55 22.6 11.8 0.52 663 173 0.26 32.0 3.90 0.12 402 23 0.06 26 ( 8 23 ( 21 0.92 ( 0.62 437 ( 181 189 ( 169 0.45 ( 0.32 36.2 750 21 NA NA NA

a Kinetic parameters were obtained using a dual-enzyme Michaelis-Menten equation for human liver microsomes and a single-enzyme Michaelis-Menten equation for P450 3A4. Km1 and Km2, Michaelis-Menten constants; Vmax1 and Vmax2, maximum velocities; Vmax/Km, turnover number. b In µM. c In pmol/min/mg. d In min-1. NA, not applicable.

Km values, indicating that the former catalyzed the activation more efficiently than the latter. The kinetics of SCPG formation in the recombinant P450 3A4 also exhibited a hyperbolic saturation profile when the data (the rates of SCPG formation vs concentrations of Cmpd A) was fitted to the Michaelis-Menten equation (data not shown). In contrast to human liver microsomes, the Eadie-Hofstee plot was shown to be linear as expected with a single enzyme (data not shown). The apparent Km and Vmax of recombinant P450 3A4 was estimated to be 36 µM and 750 pmol/min/mg, respectively (Table 2). The turnover number of P450 3A4 was estimated to be 21 (min-1) (Table 2). Studies on the Mechanism of Cmpd A Activation. Of the 50 ligand-restrained docking poses, a low energy pose of Cmpd A was achieved within the P450 3A4 active site, which was compatible with both the predicted site of activation and the shape of the active site (Figure 5A). This pose orientated the pyrazole ring of Cmpd A proximate to the heme iron and showed the distance of 5.1 Å between the C-3 and the heme iron atom, and 3.4 Å between the C-3 and the water molecule as a surrogate for the activated oxygen species (Figure 5B), suggesting that the docking orientation is compatible for oxidation at C-3. Quantum molecular calculations of the pyrazole ring indicated that the electron density located on C-3 was relatively high compared to those on the carbons at the 4and 5-positions, respectively, on the basis of both the scalar and vectorial values (Table 3). As a result, these experiments predicted that the C-3 would be the preferred site for P450 3A4mediated oxidation.

P450-Mediated BioactiVation to Isocyanate

Chem. Res. Toxicol., Vol. 22, No. 9, 2009 1609

Figure 5. Cmpd A docks into P450 3A4 active site: (A) the pyrazole ring is orientated proximate to the heme iron, (B) a distance of 5.1 Å between the C-3 and the iron atom, and of 3.4 Å between the C-3 and the surrogate water molecule, suggesting that the docking orientation is probable for oxidation at C-3. The aromatic rings are in yellow. Nitrogen is in blue, oxygen is in orange, and chlorine is in green. The heme iron is in brown, and the water molecule is in light blue.

Table 3. Electron Density of Pyrazole Ring of Cmpd A Estimated by Quantum Molecular Calculations a

atom of pyrazole ring 1

N 2 N 3 C 4 C 5 C a

electron density -0.2955 -0.3796 -0.2807 -0.0703 +0.1647

Numbering is shown in Scheme 1.

Incubations of the structural analogues of Cmpd A with human liver microsomes were conducted under the same experimental conditions as those used for Cmpd A. These analogues were substituted with a methyl group at the 3-, 4-, and 5-positions. LC/MS analysis of the incubation extracts showed minute levels of SCPG produced from the structural analogue with a methyl group placed at C-3 of pyrazoline; there was an approximately 95% reduction in the SCPG formation as compared to those from Cmpd A (data not shown). However, the substitution of a methyl group on the 4- or 5-positions of pyrazoline showed little effect on the formation of SCPG (data not shown). Bioactivation to CPIC in Humans. To demonstrate the in ViVo relevance of bioactivation to CPIC in humans, urine samples of human subjects who were administered an oral dose of a structural analogue of Cmpd A were analyzed by LC/MS to show the formation of CPIC-related thioether conjugates, i.e., SCPAC, the N-acetyl cysteine conjugate of CPIC. The structural analogue of Cmpd A consisted of the same scaffold, 4,5dihydropyrazole-1,5-dicarboxylic acid 1-[(4-chlorophenyl)amide]. Because of the proprietary nature of this compound, the full structure cannot be revealed. Representative chromatograms obtained from the analysis of urine are shown in Figure 6. The upper chromatogram A represents the LC/MRM chromatogram of urine collected from a subject who received the placebo. Chromatograms B and C represent urine collected from a subject who received the compound and control urine spiked with the synthetic standard of SCPAC (10 ng/mL), respectively. These results clearly demonstrate the presence of SCPAC in the urine of subjects dosed with the structural analogue of Cmpd A. To confirm the presence of CPA, a hydrolytic product of CPIC in human urine, a novel hyphenated LC/EC/MS system (26) was used to analyze urine samples. As shown in Figure 7, the upper chromatogram A represents the LC/MRM of urine from a placebo subject. Chromatograms B and C represent urine from a subject who was dosed with the compound and control urine spiked with standard CPA (50 ng/mL), respectively. These chromatograms confirmed the presence of in CPA in urine, as indicated by the peak at a retention time of 3.7 min. This finding

Figure 6. LC/MRM analysis of SCPAC ([M + H]+ at m/z 317) present in the urine of a human subject administered the structural analogue of Cmpd A: (A) urine (0-24 h) of placebo subject, (B) urine (0-24 h) of treated subject, and (C) the placebo urine spiked with the synthetic standard of SCPAC at 10 ng/mL.

confirmed the in ViVo generation of CPA and provided further evidence for the probable formation of CPIC during the metabolism of the structural analogue of Cmpd A in humans.

Discussion The conversion of Cmpd A to CPIC was previously demonstrated to be NADPH and human liver microsomal protein dependent (22), indicating a possible role of cytochrome P450 enzyme(s) in mediating the bioactivation process. As a result, the specificity of human P450 enzymes responsible for the bioactivation of Cmpd A to CPIC was investigated using cDNAexpressed P450 enzymes, selective P450 inhibitors and specific antibodies, and the correlation assay. The results from these various experiments corroborated and provided convincing evidence that P450 3A4 and P450 2C9 were involved in the bioactivation of Cmpd A to CPIC in human liver microsomes (Figures 1-3, Table 1). P450 2C19 also appeared to be involved in the bioactivation of Cmpd A as indicated by cDNA-expressed P450 and inhibitory studies (Figures 1 and 2, Table 1). However, insignificant correlation between the activity of P450 2C19 and the rates of SCPG formation from 20 human liver microsomes was found (data not shown). It is noteworthy to point out that the content of P450 2C19 in human liver microsomes was reported to be significantly low (approximately 1% of total

1610

Chem. Res. Toxicol., Vol. 22, No. 9, 2009

Figure 7. LC/EC/MRM analysis of CPA ([M + H]+ at m/z 217) present in the urine of a human subject administered the structural analogue of Cmpd A: (A) urine (0-24 h) of placebo subject, (B) urine (0-24 h) of treated subject, and (C) the placebo urine spiked with the standard of CPA at 50 ng/mL.

hepatic P450) (28). The greater inhibition than expected seen with the mAb against P45 2C19 may be due to cross-reactivity with other P450 2C enzymes. The cross-reactivity of inhibitory antibodies across P450 families has been documented with monoclonal antibodies (29). The contribution from P450 2C19 in bioactivating Cmpd A to CPIC in human liver microsomes remains unresolved, and further studies are needed to define its role in the bioactivation process. The role of other major P450 enzymes (normally responsible for metabolism of other marketed drugs) including P450 1A2, 2A6, 2D6, and 2E1 was ruled out primarily based on the negative results from inhibitory and correlation studies conducted with human liver microsomes. The involvement of multiple P450 enzymes in the bioactivation of Cmpd A was also supported by kinetics studies. A biphasic profile was obtained using the Eadie-Hofstee equation (Figure 4), suggesting that at least two P450 enzymes were responsible for the SCPG formation. The enzyme with the low Km value led to a relatively high turnover number for the SCPG formation at low concentrations of Cmpd A (Table 2). The activity of this enzyme in human liver microsomes was similar to that of cDNA-expressed P450 3A4 (Table 2), suggesting that P450 3A4 was most likely the enzyme with the low Km value in human liver microsomes. Hence, P450 3A4 is postulated to

Chen et al.

be the principal enzyme responsible for the bioactivation of Cmpd A to CPIC at low concentrations of Cmpd A. The other enzyme (demonstrating high Km value), assumed to be P450 2C9, contributes to the bioactivation process at higher concentrations of Cmpd A. On the basis of the Km and turnover number (Table 2), it is quite apparent that the contribution of P450 3A4 toward the bioactivation process is important at clinically relevant concentrations (