Abstract
Here we report the phase I metabolism of the rationally designed Janus kinase-3 (JAK) inhibitor 4-(4′-hydroxyphenyl)-amino-6,7-dimethoxyquinazoline (WHI-P131; JANEX-1). JANEX-1 was metabolized by the cytochrome P450 enzymes CYP1A1 and CYP1A2 in a regioselective fashion to form the biologically inactive 7-O-demethylation product 4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline (JANEX-1-M). Our molecular modeling studies indicated that the CYP1A family enzymes bind and demethylate JANEX-1 at the C-7 position of the quinazoline ring since the alternative binding conformation with demethylation at the C-6 position would result in a severe steric clash with the binding site residues. The metabolism of JANEX-1 to JANEX-1-M in pooled human liver microsomes followed Michaelis-Menten kinetics withVmax and Kmvalues (mean ± S.D.) of 34.6 ± 9.8 pmol/min/mg and 107.3 ± 66.3 μM, respectively. α-Naphthoflavone and furafylline, which both inhibit CYP1A2, significantly inhibited the formation of JANEX-1-M in human liver microsomes. There was a direct correlation between CYP1A activities and the magnitude of JANEX-1-M formation in the liver microsomes from different animal species. A significantly increased metabolic rate for JANEX-1 was observed in Aroclor 1254-, β-naphthoflavone-, and 3-methylcholanthrene-induced microsomes but not in clofibrate-, dexamethasone-, isoniazid-, and phenobarbital-induced microsomes. The formation of JANEX-1-M in the presence of baculovirus-expressed CYP1A1 and 1A2 was consistent with Michaelis-Menten kinetics. The systemic clearance of JANEX-1-M was much faster than that of JANEX-1 (5525.1 ± 1926.2 ml/h/kg versus 1458.0 ± 258.6 ml/h/kg). Consequently, the area under the curve value for JANEX-1-M was much smaller than that for JANEX-1 (27.5 ± 8.0 versus 94.8 ± 18.4 μM · h; P < 0.001).
Acute allergic reactions, also known as immediate (type I) hypersensitivity reactions, including anaphylaxis with a potentially fatal outcome, are triggered by three major classes of proinflammatory mediators, namely preformed granule-associated bioactive amines (e.g., histamine, serotonin) and acid hydrolases (e.g., β-hexosaminidase), newly synthesized arachidonic acid metabolites (e.g., leukotriene C4, prostaglandin D2, and platelet activating factor), and a number of proinflammatory vasoactive cytokines [e.g., tumor necrosis factor (TNF1)-α and interleukin-6] (Galli, 1993). These proinflammatory mediators are released from sensitized mast cells upon activation through the antigen-mediated cross-linking of their high-affinity cell surface IgE receptors/Fc epsilon (ε) RI (Galli, 1993; Malaviya et al., 1993). IgE receptor/FcεRI is a multimeric receptor with α-, β-, and homodimeric γ-chains. Both β- and γ-subunits of the IgE receptor/FcεRI contain immunoreceptor tyrosine-based activation motifs, which allow interaction with protein tyrosine kinases (PTK) and PTK substrates via their SH2 domains (Scharenberg et al., 1995). The engagement of IgE receptors by antigen triggers a cascade of biochemical signal transduction events, including activation of multiple PTK (Scharenberg et al., 1995). The activation of PTK and subsequent tyrosine phosphorylation of their downstream substrates have been implicated in the pathophysiology of type I hypersensitivity reactions (Scharenberg et al., 1995). The elucidation of the PTK-dependent signal transduction events that lead to FcεRI-mediated mast cell degranulation and mediator release may provide the basis for the rational design of potent mast cell inhibitors for prevention and treatment of allergic reactions.
In a recent study, we found that the IgE/antigen induced degranulation and mediator release are substantially reduced withJak3−/− mast cells from JAK3-null mice that were generated by targeted disruption of the Jak3 gene in embryonic stem cells (Malaviya and Uckun, 1999), indicating that JAK3 plays a pivotal role in IgE receptor/FcεRI-mediated mast cell responses both in vitro and in vivo. We subsequently reported that treatment of rodent and human mast cells with 4-(4′-hydroxyphenyl)-amino-6,7-dimethoxyquinazoline) (WHI-P131; JANEX-1), a rationally designed potent and specific inhibitor of JAK3 that does not affect the enzymatic activity of other protein tyrosine kinases, including the Src family tyrosine kinase LYN (lyn-gene product), the ZAP/SYK family tyrosine kinase SYK (spleenic tyrosine kinase), the TEC family tyrosine kinase BTK (Bruton's tyrosine kinase), the receptor family tyrosine kinase IRK (insuline receptor kinase), or Janus kinases JAK1 and JAK2 (Goodman et al., 1998; Sudbeck et al., 1998, 1999), inhibited degranulation and proinflammatory mediator release after IgE receptor/FcεRI cross-linking. In vivo administration of this potent JAK3 inhibitor prevented mast cell degranulation and development of cutaneous and systemic fatal anaphylaxis in mice. More recently, JANEX-1 was also found to be a potent inhibitor of asthmatic reactions in mice (Malaviya et al., 2001). Thus, targeting JAK3 with a specific inhibitor, such as JANEX-1, may provide the basis for new and effective treatment and prevention programs for mast cell-mediated allergic reactions and asthma.
In addition to its anti-allergic activity, JANEX-1 may also be useful for treatment of JAK3-positive hematologic malignancies (e.g., acute lymphoblastic leukemia) (Sudbeck et al., 1999) and inflammatory disorders (e.g., graft versus host disease after allogeneic bone marrow transplantation) mediated by JAK3-positive lymphocytes (Cetkovic-Cvrlje et al., 2001). Furthermore, recent studies have indicated that JANEX-1 may also be useful for prevention of fatal thromboembolism (Tibbles et al., 2001) and treatment of amyotrophic lateral sclerosis (Trieu et al., 2000). The pharmacokinetic features of JANEX-1 in rodents and cynomolgus monkeys have been studied using a high-performance liquid chromatography (HPLC)-based quantitative detection method (Chen et al., 1999a; Uckun et al., 1999b). JANEX-1 was very well tolerated by mice and monkeys, and therapeutic plasma concentrations of JANEX-1 could be achieved at nontoxic dose levels. The biological activity profile and lack of significant systemic toxicity of JANEX-1 suggest that this JAK3 inhibitor may be useful in the treatment of allergic and inflammatory disorders and hematologic malignancies.
The purpose of the present study was to investigate the phase I metabolism of JANEX-1. Here, we show that JANEX-1 is metabolized by the cytochrome P450 enzymes CYP1A1 and CYP1A2 in a regioselective fashion to form the biologically inactive 7-O-demethylation product 4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline (JANEX-1-M). Our molecular modeling studies indicated that the cytochrome P4501A family enzymes bind and demethylate JANEX-1 at the C-7 position of the quinazoline ring since the alternative binding conformation with demethylation at the C-6 position would result in a severe steric clash with the binding site residues.
Materials and Methods
Chemicals.
Acetonitrile and acetone were purchased from Fisher Chemicals (Fair Lawn, NJ). Furafylline and (S)-(+)-mephenytoin were purchased from Ultrafine Chemicals (Manchester, UK). Ketoconazole was from ICN Biomedicals, Inc. (Aurora, OH), and all other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). The reagents for synthesis were purchased from Aldrich (Milwaukee, WI). The human liver microsomes (individuals: H003, H006, H023, H030, H042, H043, H056, H066, H070, H089, H093, and H112; pooled: H161) were purchased from GENTEST (Woburn, MA). The protein concentrations and specific activities of each P450 isoform in the human microsomes were provided in the data sheets by the manufacturer.
The microsomes containing baculovirus-expressed P450 enzymes, including CYP1A1, CYP1A2, CYP2A6, CYP3A4, CYP4A11, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and recombinant human flavin-containing monooxygenase (FMO)-3, were purchased from GENTEST. All enzymes were coexpressed with oxidoreductase. CYP2A6, 2B6, 2C8, 2C9, 2C19, and 2E1 were also coexpressed with cytochrome b5.
The liver microsomes from ICR/CD-1 mice, Sprague-Dawley rats, Dunkin-Hartley guinea pigs, New Zealand white rabbits, beagle dogs, and cynomolgus monkeys were purchased from In Vitro Technologies, Inc. (Baltimore, MD). Induced rat liver microsomes from treatment of Sprague-Dawley rats with Aroclor 1254 (ARO) (single dose of 100 mg/kg i.p.), β-naphthoflavone (BNF) (100 mg/kg/day i.p. for 4 days), clofibrate (CFB) (200 mg/kg/day i.p. for 4 days), dexamethasone (DEX) (50 mg/kg/day i.p. for 4 days), isoniazid (ISO) (200 mg/kg/day i.p. for 4 days), 3-methylcholanthrene (3-MC) (54 mg/kg/day i.p. for 4 days), and phenobarbital (PHEN) (80 mg/kg/day i.p. for 4 days) were also purchased from In Vitro Technologies, Inc. The protein concentrations and specific activities of each P450 isoform in the animal liver microsomes were provided in the data sheets by the manufacturer.
JANEX-1 was synthesized as previously described (Narla et al., 1998). Its structure (Fig. 3A) and physicochemical properties were previously reported (Narla et al., 1998; Sudbeck et al., 2000).
Synthesis and Characterization of Putative JANEX-1 Metabolites.
Sodium ethanethiolate (1.26 g, 12.0 mmol) was added to a suspension of JANEX-1 · HCl (1.00 g, 3.0 mmol) inN,N-dimethylformamide (14 ml) in a dry round-bottom flask under N2. The reaction was heated to 100°C and stirred 16 h. The mixture was concentrated in vacuo, and the isomers were separated by flash chromatography on silica (MeOH/CH2Cl2, 10:90, v/v). The two products (compound 1 and compound 2) were recrystallized from methanol, yielding compound 1 · 0.5 MeOH (275 mg, 31%) and compound 2 · 0.5 MeOH (80 mg, 9%).
Compound 1 [4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline] · 0.5 MeOH: yellow solid, m.p. 230°C; 1H NMR (d,d6-DMSO) 10.2 (broad s, 1 H), 9.3 (broad s, 1 H), 9.26 (s, 1 H), 8.29 (s, 1 H), 7.79 (s, 1 H), 7.45 (d,J = 8.5 Hz, 2 H), 7.02 (s, 1 H), 6.78 (d,J = 8.5 Hz, 2 H), 3.93 (s, 3 H), 3.16 [s, ∼1.5H (CH3OH)]; 13C NMR (d,d6-DMSO) 157.3, 154.4, 153.5, 153.1, 149.0, 147.2, 131.3, 125.5, 115.6, 110.8, 108.8, 103.0, 56.9; infrared (cm−1) 3425, 1633, 1514, 1423, 1234; MS (M + H): calculated 284.1035, found 284.1051.
Compound 2 [4-(4′-hydroxyphenyl)-amino-7-methoxy-6-hydroxyquinazoline] · 0.5 MeOH: bright yellow solid, m.p. 278°C (decomposed);1H NMR (d,d6-DMSO) 9.54 (broad s, 1 H), 9.23 (broad s, 1 H), 9.16 (s, 1 H), 8.32 (s, 1 H), 7.74 (s, 1 H), 7.51 (d,J = 8.8 Hz, 2 H), 7.14 (s, 1 H), 6.75 (d,J = 8.8 Hz, 2 H), 3.94 (s, 3 H), 3.16 [s, ∼1.5 H (CH3OH)]; 13C NMR (d,d6-DMSO) 157.1, 154.1, 153.1, 146.9, 146.5, 131.7, 124.9, 115.5, 110.1, 107.7, 106.2, 102.8, 56.6, 49.4 (CH3OH); infrared (cm−1) 3342, 1637, 1514, 1470, 1429, 1209, 1011, 837; MS (M + H): calculated 284.1035, found 284.1042.
X-Ray Structure of JANEX-1-M.
Colorless rectangular plates of JANEX-1-M were grown from methanol by slow evaporation at room temperature. A crystal was mounted on a glass fiber using epoxy, and X-ray diffraction data for a 0.6 × 0.5 × 0.1-mm crystal were collected at room temperature using a SMART 1K CCD X-ray detector (Bruker AXS, Madison, WI). Structure solution and refinement were performed using the SHELXTL suite of programs (Bruker AXS), and absorption correction was applied using the SADABS program (Sheldrick, 2001).
Incubation of JANEX-1 with Liver Microsomes.
All incubation mixtures contained the following ingredients (a modified procedure of Chen et al., 1995, 1999b) at the indicated final concentrations (200 μl): 1× PBS, microsomes (1 mg/ml), glucose 6-phosphate (10 mM), glucose-6-phosphate dehydrogenase (2 units/ml), and MgCl2 (5 mM). The mixtures were preincubated at 37°C for 5 min. The metabolism reaction was initiated by the addition of JANEX-1 (various concentrations). After incubation for 10 or 15 min, the reaction was stopped by addition of 800 μl of acetone and thoroughly mixed using a vortex device. The mixtures were extracted and separated through the HPLC system, as described below.
The cytochrome P450 contribution to the metabolism of JANEX-1 was determined by addition of SKF-525A to the microsome. To evaluate the nonenzymatic metabolism of JANEX-1, microsomes were not incubated in the incubation mixture, which eliminated the enzyme source; the NADPH-generating system was eliminated from the incubation mixture, which in turn eliminated the cofactors for the enzymes; and microsomes were inactivated by heating for 5 min at 90°C. The effects of the recombinant FMO (0.25 mg/ml) and human liver cytosol were studied using incubations conducted as described above.
Inhibiton Experiments.
The effects of various cytochrome P450 substrates/inhibitors on the formation of JANEX-1-M in pooled human liver microsomes (catalog no. H161) were studied by measuring JANEX-1-M formation after preincubation with the following compounds: α-naphthoflavone, furafylline, coumarin, diethyldithiocarbamic acid, orphenadrine, sulfaphenazole, (S)-mephenytoin, quinidine, p-nitrophenol, ketoconazole, troleandomycin, and lauric acid. All inhibitors were dissolved in DMSO and added to the incubations at a final DMSO concentration of 0.5% (v/v). Control incubations (without inhibitors) also contained 0.5% DMSO. The final concentration of JANEX-1 was 50 μM, and final concentrations of P450 substrates/inhibitors were those recommended in previous publications (Kajita et al., 2000; Wynalda, 2000).
Incubation of JANEX-1 with Microsomes Containing Baculovirus-Expressed Enzymes.
The metabolism of JANEX-1 by specific cytochrome P450 isoforms was studied using incubations conducted as described above, except that baculovirus-expressed P450 enzymes were used at a concentration of 50 pmol/ml for the incubation. JANEX-1 was used at a final concentration of 50 μM in these experiments.
Identification of Metabolite(s) by LC-MS.
Mass spectrum analysis was carried out using atmospheric pressure ionization-electrospray and a high-energy-dynode electron multiplier (Hewlett Packard, Palo Alto, CA), which is connected to the LC system (Chen et al., 1999b). The conditions for mass spectrum analysis were set at a fragmentor of 75 V, a drying gas flow of 10 l/min, a nebulizer pressure of 25 psig, and a drying gas temperature of 350°C.
Pharmacokinetic Studies in Mice.
Female CD-1 mice (6–8 weeks old) from Charles River Laboratories, Inc. (Wilmington, MA) were housed in a USDA-accredited animal care facility under standard environmental conditions. All rodents were housed in microisolator cages (Lab Products, Inc., Maywood, NJ) containing autoclaved bedding. Mice were allowed free access to autoclaved pellet food and tap water throughout the study. All animal studies were approved by the Parker Hughes Institute Animal Care and Use Committee, and all animal care procedures conformed to the principles outlined in the Guide for the Care and Use of Laboratory Animals(National Research Council, Washington, DC).
A 50-μl solution of JANEX-1 or JANEX-1-M (40 mg/kg) dissolved in DMSO was administered intravenously via the tail vein. This volume of DMSO is well tolerated by mice when administrated by rapid i.v. or i.p. injection. Four mice per time point were used for pharmacokinetic studies. Blood samples (∼500 μl) were obtained from the ocular venous plexus by retro-orbital venipuncture at 0, 2, 5, 10, 15, 30, 45, 60, 120, 240, and 360 min after i.v. injection.
For studying the in vivo metabolism of JANEX-1, 10 mice received a 100 mg/kg i.v. bolus dose of JANEX-1, and then the mice were collectively placed in the Nalgene metabolic cage system (Nalge Company, Rochester, NY). Urine was collected at 0 to 6, 6 to 24, 24 to 30, 30 to 48, and 48 to 72 h.
In Vitro Mast Cell Biology Experiments.
RBL-2H3 cells [a rat origin mucosal mast cell line kindly provided by Dr. R. P. Siraganian (NIH, Bethesda, MD)] in 48-well tissue culture plates were sensitized with a monoclonal anti-dinitrophenyl (DNP)-IgE antibody (0.24 mg/ml) by overnight incubation at 37°C in a humidified 5% CO2 atmosphere. Unbound IgE was removed by washing cells with PIPES-buffered saline containing 1 mM calcium chloride.
RBL-2H3 cells were treated for 1 h at 37°C with the JANEX-1 or JANEX-1-M and then challenged with 20 ng/ml DNP-BSA for 30 min at 37°C. The plates were centrifuged at 200g for 10 min at 4°C; supernatants were removed and saved. The pellets were washed once with PIPES-buffered saline and solubilized in PIPES-buffered saline containing 0.1% Triton X-100. Degranulation of RBL-2H3 cells was monitored by measuring β-hexosaminidase activity in cell free supernatants and Triton X-100-solubilized pellets by the colorimetric assay usingp-nitrophenyl-2-acetamide-deoxy-β-glucopyranoside, as previously described (Ozawa et al., 1993). TNF-α levels in cell free supernatants were estimated as previously described (Malaviya et al., 1999).
Anaphylaxis Models.
To examine the effect of JANEX-1 and JANEX-1-M on passive cutaneous anaphylaxis in mice, dorsal sides of the ears of BALB/c mice were injected intradermally with 20 ng of DNP-IgE (left ears) or PBS (right ears) in a 20-μl volume using a 30-gauge needle, as previously described (Miyajima et al., 1997). After 20 h, mice were treated with JANEX-1 or JANEX-1-M (20 mg/kg i.p.) twice at 1-h intervals before the antigen challenge. Control mice were treated with an equal volume of vehicle. Thirty minutes after the last dose of JANEX-1 or JANEX-1-M or vehicle, mice were challenged with 100 μg of antigen (DNP-BSA) in 200 μl of 2% Evans blue dye intravenously. Mice were sacrificed by cervical dislocation 30 min after the antigen challenge. For quantitation of Evans blue dye extravasation, as a measure of anaphylaxis-associated vascular hyperpermeability, 8-mm skin specimens were removed from the ears of mice, minced in 2 ml of formamide, and incubated at 80°C for 2 h in a water bath to extract the dye. The absorbance was read at 590 nm.
HPLC Determination of JANEX-1 and Its Metabolites.
The levels of JANEX-1 and its metabolite in the microsome systems and plasma were determined using a quantitative HPLC method (Chen et al., 1999a). In brief, 800 μl of acetone was added to the microsome system to terminate the reaction. Following centrifugation (300g, 5 min), the supernatant was transferred to a clean tube and dried under a slow, steady stream of nitrogen gas. The residue was reconstituted in 50 μl of methanol/water (9:1, v/v). A 35-μl aliquot of the reconstituted solution was injected for HPLC analysis using a Hewlett Packard series 1100 instrument equipped with a quaternary pump, an autosampler, an auto electronic degasser, an automatic thermostatic column compartment, a diode array detector, and a computer with Chemstation software for data analysis. A 250 × 4-mm Lichrospher 100 RP-18 (5 μm) analytical column and a 4 × 4-mm Lichrospher 100 RP-18 (5 M) guard column were obtained from Hewlett Packard. The mobile phase contained acetonitrile/water (0.1% of TFA and 0.1% TEA) (28:72, v/v), which was degassed automatically by the electronic degasser system. The column was equilibrated and eluted under isocratic conditions using a flow rate of 0.6 ml/min at ambient temperature. The wavelength of detection was set at 340 nm for JANEX-1 and its metabolite.
The levels of JANEX-1 and its metabolite in the urine of mice treated with JANEX-1 was determined using the above-described procedures for the microsome system and plasma, except that a 250 × 4.6-mm Zorbax SB-phenyl analytical column (5 μm) and a mobile phase containing acetonitrile/water (0.1% of TFA) (28:72, v/v) at flow rate of 0.6 ml/min were used. On-line MS detection was set up with a selected ion model at both 284 and 298 and a fragmentor voltage of 75 V.
Enzyme Kinetic Analysis and Pharmacokinetic Analysis.
Enzymatic kinetic analysis and pharmacokinetic analysis were performed using WinNonlin program V 3.0 (Pharsight, Mountain View, CA) (Chen et al., 1999b, 2001; Uckun et al., 1999a,b). Kinetic parameters such asVmax (maximum reaction velocity) andKm (the substrate concentration that corresponds to 50% Vmax) were estimated by fitting the data to the Michaelis-Menten model (V =Vmax ∗Cr /(Km +Cr ), where C is the drug concentration and r is the Hill coefficient. Metabolic clearance was estimated as (CLmet =Vm/Km), and Eadie-Hofstee plots were constructed to determine whether the kinetics was mono- or biphasic. The relationship between the formation of JANEX-1-M and various cytochrome P450 activities was examined by calculating the nonparametric Spearman rank correlation coefficient using Instat computer software, version 3.0 (GraphPad Software, Inc., San Diego, CA).
For pharmacokinetic modeling, an appropriate model was chosen on the basis of the lowest sum of weighted squared residuals, the lowest Schwartz criterion, the lowest Akaike's information criterion value, the lowest standard errors of the fitted parameters, and the dispersion of the residuals. The elimination half-life was estimated by linear regression analysis of the terminal phase of the plasma concentration-time profile. The systemic clearance (CL) was determined by dividing the dose by the area under the curve.
Homology Model of Human CYP1A1 and Docking of JANEX-1-M.
A homology model of human CYP1A1 was constructed based on the crystal structure of CYP2C5 (PDB accession number: 1dt6) using homology and docking modules with INSIGHT II software (Molecular Simulations, Inc., San Diego, CA). The homology modeling of CYP1A1 was carried out in two steps summarized as follows: 1) the most reasonable sequence alignment between the CYP1A1 and a coordinate template was determined; and 2) new coordinates were assigned to the CYP1A1 residues according to the template coordinates (based on the sequence alignment), followed by the determination of loop coordinates and an energy minimization of the entire structure. The coordinates for several loop regions where sequence insertions occur were chosen from a limited number of possibilities automatically generated by the program and manually adjusted to a more ideal geometry using the program CHAIN (Sack, 1988). Finally, the constructed model was subjected to energy minimization using the X-PLOR program (Brunger, 1992) so that any steric strain introduced during the model-building process could be relieved. The model was screened for unfavorable steric contacts, and if necessary, such side chains were adjusted either by using a rotamer library database or by manually rotating the respective side chains. Initially, JANEX-1 was manually docked into the three-dimensional model of CYP1A1. The initial coordinates of JANEX-1 was modified directly from its X-ray structure, rotated, and translated as a rigid body into the active site with the O-7 group fixed at a distance from the heme iron of 3.0 Å. After the manual maneuver, major steric interactions with nearby residues were avoided. Otherwise, an even sampling of various initial positions was used within the general binding region that is known from crystal structures of P450 enzyme complexes. Subsequently, an automatic docking procedure was followed and evaluated using the Ludi score function (Bohm, 1992) for maximum binding affinity. Finally, the docked position of the molecule was inspected and compared with known compounds in superimposed homologous protein complexes using templates with PDB access codes 1BVV, 1F4T, 1E9X, and1EA1.
Results
Metabolism of JANEX-1 in Mice and Human Microsomes.
We used analytical HPLC to examine mouse urine samples collected from a group of 10 mice between 0 to 6, 6 to 24, 24 to 30, 30 to 48, and 48 to 72 h after i.v. injection of a 100 mg/kg bolus dose of JANEX-1 lactobionate for the presence of JANEX-1 and potential JANEX-1 metabolites. The parent compound JANEX-1 was detected at a retention time (RT) of 10.2 min in urine samples collected between 0 to 6 and 6 to 24 h but not at later time points (Fig.1, A and B). A potential metabolite peak with a retention time of 7.9 min was also detected in the same urine samples (Fig. 1B). The on-line LC-MS yielded m/zvalues of 298 for the parent compound and 284 for the potential metabolite (Fig. 1C), prompting the hypothesis that the putative metabolite is a demethylated form (M-14) of JANEX-1.
To test the hypothesis that JANEX-1 is metabolized by demethylation, we next examined the in vitro metabolism of WHI-P31 in human liver microsomes. The representative metabolite profile of JANEX-1 following incubation with human hepatic microsomes is presented in Fig.2A. The parent compound is shown with a RT of 7.3 min, whereas the major metabolite of JANEX-1 in the microsome system had a RT of 5.9 min (JANEX-1-M). Notably, the on-line LC-MS yielded for JANEX-1-M anm/z value of 284 (Fig. 2B), which is the same as the m/z value of the putative in vivo metabolite of JANEX-1 detected in mouse urine samples (see Fig. 1C).
Notably, the formation of JANEX-1-M was significantly decreased when NADP was omitted from the system (6.1 versus 47.1 pmol/min/mg,P < 0.0001; Fig. 2C). No metabolite was formed in denatured microsomes or in the absence of microsomes (Fig. 2C). Furthermore, the P450 inhibitor proadifen (SKF-525A) abolished the microsomal metabolism of JANEX-1 (Fig. 2C). Human liver cytosol and FMO did not participate in the in vitro metabolism of JANEX-1 (Fig. 2C). Taken together, these results indicate that JANEX-1 is metabolized by cytochrome P450-mediated enzymatic demethylation.
Identification of 4-(4′-Hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline as the Major Metabolite of JANEX-1.
We next sought to determine the structural identity of JANEX-1-M, the major metabolite of JANEX-1. Since demethylation could occur on either one of the two methoxy groups attached to the quinazoline ring of JANEX-1, it was important to determine whether the 6-O-demethylated or the 7-O-demethylated form of JANEX-1 was the actual metabolite (Fig.3A). To this end, we first used sodium ethanethiolate to demethylate the methoxy groups of JANEX-1. The demethylation reaction yielded two structural isomers, which were separated by flash chromatography on silica and then further purified by recrystallization from methanol. The electrospray ionization-MS yielded a molecular mass of 284 for both synthetic isomers.
The retention times of the two synthesized demethylation products (compound 1 and compound 2) were different, but the retention time of compound 1 was identical to the retention time of JANEX-1-M formed in the microsome system (Fig. 3B). Furthermore, compound 1 had the samem/z (i.e., 284) as the urine-derived or microsome-derived JANEX-1-M by LC-MS (Fig. 3C). Therefore, compound 1 was further characterized to elucidate the structural features of the metabolite of JANEX-1.
Compound 1 was crystallized and analyzed using X-ray crystallography to determine the molecular structure of JANEX-1-M. The crystal structure of compound 1 showed that it is the 7-O-demethylation product of JANEX-1. The refined small molecule X-ray crystal structure of JANEX-1-M is shown in Fig. 3D; data statistics are listed in Table1, and atomic coordinates are listed in Table 2. All nonhydrogen atoms were refined using anisotropic displacement parameters. Hydrogen atoms were placed at ideal positions (taking into consideration hydrogen bonding interactions for H-3 and H-4; Fig. 3D) and refined as riding atoms with relative isotropic displacement parameters except for H-1 (Fig. 3D), which was located in the electron density difference map and refined isotropically. Two disordered solvent regions (one near O-1 and the other near O-3 and O-7) were identified in the structure and were represented as a methanol molecule and a disordered group of three carbon atoms for refinement purposes. Taken together, these results demonstrate that JANEX-1-M is 4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline, the 7-O-demethylated form of JANEX-1.
Activity and Pharmacokinetic Features of JANEX-1-M, 4-(4′-Hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline.
Targeting JAK3 in mast cells with JANEX-1 inhibits IgE receptor-mediated mast cell responses (Malaviya et al., 1999). To examine the effect of JANEX-1-M on IgE receptor/FcεRI-mediated mast cell degranulation and cytokine release, IgE-sensitized RBL-2H3 mast cells were preincubated with increasing concentrations of JANEX-1-M, JANEX-1, or vehicle for 1 h before challenge with antigen (DNP-BSA). Notably, JANEX-1 prevented mast cell degranulation and release of preformed granule-associated β-hexosaminidase (Fig.4A) and release of the proinflammatory cytokine TNF-α (Fig. 4B) in a concentration-dependent fashion, with near to complete inhibition at ≥30 μM. Unlike JANEX-1, its 7-O-demethylated metabolite JANEX-1-M did not inhibit mast cell degranulation or TNF-α release after IgE receptor/FcεRI cross-linking (Fig. 4, A and B). These results demonstrate that JANEX-1-M is >10-fold less potent than JANEX-1 in inhibiting IgE receptor-mediated mast cell responses.
Increased vascular permeability induced by mast cell mediators, such as histamine and leukotrienes, is a hallmark of anaphylaxis (Oettgen et al., 1994). Therefore, we next compared the effects of JANEX-1 and JANEX-1-M on vascular permeability in a well characterized murine model of passive cutaneous anaphylaxis (Miyajima et al., 1997). JANEX-1, but not JANEX-1-M, significantly inhibited the IgE/antigen induced plasma exudation, as measured by extravasation of systemically administered Evans blue dye, in mice that had been presensitized with antigen specific IgE (Fig. 4C).
We next examined the pharmacokinetics of JANEX-1-M in mice in side-by-side comparison with the parent compound JANEX-1. Following i.v. injection of JANEX-1-M at a dose level of 40 mg/kg, the plasma concentration of JANEX-1-M as a function of time can best be described by using a two-compartment model (Fig. 4D). Both JANEX-1-M and JANEX-1 had moderate volumes of distribution at steady state (JANEX-1-M:Vss = 811.6 ± 336.0 ml/kg,n = 4; JANEX-1: Vss = 857.7 ± 80.8 ml/kg), which is slightly larger than the volume of water in the body (Davies and Morris, 1993) (Table3). JANEX-1-M had a larger volume of distribution at the central compartment (745.7 ± 40.7 versus 429.4 ± 206.1 ml/kg, P < 0.02) and a shorter elimination half-life (33.0 ± 20.1 versus 77.1 ± 10.2 min,P < 0.01) than the parent compound JANEX-1 (Table 3). The CL of JANEX-1-M was much faster than that of JANEX-1 (5525.1 ± 1926.2 versus 1458.0 ± 258.6 ml/h/kg, P < 0.01), which is higher than the blood flow to either the kidney or the liver (Davies and Morris, 1993) (Fig. 4D; Table 3). Consequently, the area under the curve value for JANEX-1-M was much smaller than that for JANEX-1 (27.5 ± 8.0 versus 94.8 ± 18.4 μM · h,P < 0.001) (Table 3).
Kinetics of JANEX-1 Metabolism in Liver Microsomes and Its Relationship to CYP1A2 Activity.
The metabolism of JANEX-1 to JANEX-1-M in pooled human liver microsomes followed Michaelis-Menten kinetics withVmax and Kmvalues (mean ± S.D.) of 34.6 ± 9.8 pmol/min/mg and 107.3 ± 66.3 μM, respectively (Fig.5A). The corresponding rate of metabolic clearance for JANEX-1 was 0.3 μl/min/mg.
An Eadie-Hofstee plot of the formation of JANEX-1-M (V) versus V/S in human liver microsomes was monophasic (Fig. 5B), suggesting that a single enzymatic pathway was the main contributor to the metabolism of JANEX-1. Notably, at a final concentration of 50 μM, the rate of JANEX-1-M formation from JANEX-1 in human liver microsomes (n = 12) correlated with the microsomal CYP1A2 (phenacetin O-deethylase) activity (r = 0.95, P < 0.0001) (Fig. 5C) but not with total P450 content, cytochrome C reductase activity, cytochrome b5 activity, or the activities of other cytochrome P450 enzymes, including CYP2A6 (coumarin 7-hydroxylase), CYP2B6 [(S)-mephenytoinN-demethylase], CYP2C8 (paclitaxel 6α-hydroxylase), CYP2C9 (diclofenac 4′-hydroxylase), CYP2C19 [(S)-mephenytoin 4′-hydroxylase], CYP2D6 (bufuralol 1′-hydroxylase), CYP2E1 (chlorzoxazone 6-hydroxylase), CYP3A4 (testosterone 6β-hydroxylase), CYP4A (lauric acid 12-hydroxylase), and FMO (methyl p-tolyl sulfide oxidase) (data not shown). These results prompted the hypothesis that JANEX-1 metabolism is mediated primarily by CYP1A2.
The effects of 12 cytochrome P450 substrates/inhibitors on the formation of JANEX-1-M in human liver microsomes are presented in Table4. In all 12 compounds tested, only α-naphthoflavone and furafylline, which both inhibit CYP1A2, significantly inhibited the formation of JANEX-1-M in human liver microsomes, which supports the hypothesis that JANEX-1 metabolism is mediated primarily by CYP1A2.
We next studied the metabolism of JANEX-1 in induced Sprague-Dawley rat liver microsomes. ARO, BNF, 3-MC are the inducers for the CYP1A subfamily (Waxman, 1999) (Fig. 6A). CYP4A is induced by CFB (Waxman, 1999), CPR3A by DEX (Grange et a., 1994;Waxman, 1999), CYP2E by ISO (Grange et al., 1994), and CYP2B by PHEN (Waxman, 1999). A significantly increased metabolic rate for JANEX-1 was observed in ARO-, BNF-, and 3-MC-induced microsomes but not in CFB-, DEX-, ISO-, and PHEN-induced microsomes (Fig. 6B). These results further support the importance of CYP1A in the metabolism of JANEX-1 to form JANEX-1-M. The results also exclude a significant role for CYP4A, CPR3A, CYP2E, and CYP2B in the formation of JANEX-1-M.
We also examined the interspecies differences in JANEX-1 metabolism in relationship to the CYP1A content of the microsomes. There was a direct correlation between CYP1A activity and the magnitude of JANEX-1-M formation in the liver microsomes from different animal species (Fig.6C). The metabolic rate was the highest in rabbit liver microsomes, which had the highest CYP1A activity, and lowest in canine microsomes, which had the lowest CYP1A activity (Fig. 6C).
Metabolism of JANEX-1 by Recombinant Human CYP1A1 and CYP1A2.
We next sought to determine the identity of the cytochrome P450 isoform, which is responsible for the metabolism of JANEX-1 by using microsome preparations containing baculovirus-expressed recombinant cytochrome P450 isoforms (Crespi, 1995). The results presented in Fig.7A indicate that metabolism of JANEX-1 to form JANEX-1-M is mainly mediated by cytochrome P450 1A1 and 1A2.
The metabolite formation of JANEX-1-M in baculovirus-expressed CYP1A1 and 1A2 was consistent with Michaelis-Menten kinetics (Fig. 7B). The estimated Vmax andKm values for JANEX-1-M were 1842.7 ± 268.8 pmol/min/nmol and 1.5 ± 0.6 μM in the presence of CYP1A1 and 9292.4 ± 307.9 pmol/min/nmol and 8.5 ± 0.8 μM in the presence of CYP1A2, respectively. Therefore, the average metabolic clearance of JANEX-1 was 1228.5 μl/min/nmol in the presence of CYP1A1 and 1093.2 μl/min/nmol in the presence of CYP1A2.
Structural Basis for the RegioselectiveO-Demethylation of JANEX-1 by CYP1A1/CYP1A2.
We next set out to elucidate the structural basis for the regioselective demethylation of JANEX-1 by enzymes of the CYP1A family. To this end, we constructed a series of three-dimensional models of human cytochrome P450 enzymes based on their amino acid sequence similarity with several cytochrome P450 proteins with known crystal structures, including rabbit CYP2C5 (see Materials and Methods). A homology model of CYP1A1 was established using the HOMOLOGY module within the INSIGHT II program (Molecular Simulations, Inc.) (see Materials and Methods). The binding mode of JANEX-1 was determined by a docking procedure using the AFFINITY module within INSIGHT II, with a fixed distance between the O-7 atom and the heme iron atom. Next, the homology model of CYP1A1 was superimposed with crystal structures of a number of P450 enzymes and their complexes with small molecules (PDB access codes: 1DT6, 2C17A, 1FAG, and 2BMH) (Hasemann et al., 1994; Hishiki et al., 2000).
The crystal structures of the ligated P450 enzyme complexes revealed a general orientation of ligand binding in which a ligand is typically 2 Å away from the heme iron stacking against the I helix and extending toward the BC loop. The comparison showed that the docked position of JANEX-1 was consistent with the general orientation. In the docked model, the quinazoline group of JANEX-1 is situated on top of the heme group of CYP1A1 (see Fig. 8A), with the O-7 atom close to the iron in the heme center. On one side, the quinazoline is stacked against the I helix, containing residues 317 to 321; the side chain of A317 has VDW contact with the quinazoline ring. On the same side, the 4-hydroxyphenyl group is stacked against the F helix, containing residue G225, and is sandwiched from the other side by residue F123. Mostly perpendicular to the aromatic ring plane of the ligand, T111 from the BC loop forms a hydrogen bond with the 4-hydroxyl group. Residue V228 is near the hydroxyphenyl ring and V382 is in nonbonded contact with the 6-methoxyl group. This model provides an opportunity to examine what residues might be involved in the binding of JANEX-1 and what might be the reasons for regioselective demethylation. Based on the model, it is clear that an alternative binding conformation, with O-6 close to the iron center, is unlikely (explained in Fig. 8B). This model of CYP1A1 and crystal structures of other P450 enzymes reveal that the binding site is asymmetric, with the heme iron at the center of a bottom plateau of a “valley-shaped” region. The alternative conformation (with demethylation at the C-6 position) would result in a severe steric clash with the residues that are associated with the heme (Fig. 8B).
Finally, we examined which residue(s) might be responsible for JANEX-1 being favorably catalyzed by CYP1A enzymes. Our model indicated that T111 in CYP1A1 forms a hydrogen bond with JANEX-1. An ideal hydrogen bond could result in a nearly 8-fold improvement in the binding constant based on a LUDI score function (Bohm, 1992). T111 is conserved in the 1A family but is not found in other P450 enzymes that we examined, except CYP2A6. It is also noted that a small residue at 225 (a glycine in CYP1A1 and a valine in CYP1A2) has close VDW contact with the 4-hydroxyphenyl ring. This would be less feasible when 225 is a larger residue, as is the case for other enzymes. Considering the tight fit in this region, a larger residue may be sufficient to impair the binding of JANEX-1.
Discussion
Our experimental data presented here provides unprecedented evidence that the JAK3 inhibitor JANEX-1 is metabolized by cytochrome P450 enzymes CYP1A1 and CYP1A2 in a regioselective fashion to form the biologically inactive 7-O-demethylation product 4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxylquinazoline. Our molecular modeling studies indicated that the cytochrome 1A family enzymes must bind and demethylate the molecule at the C-7 position of the quinazoline ring since the alternative binding conformation with demethylation at the C-6 position would result in a severe steric clash with the residues that are associated with the heme iron of the enzyme at the center of a bottom plateau of a valley-shaped binding region. Our model further indicated that the combination of the hydrogen bond with T111 and the presence of a small residue at 225 makes the cytochrome 1A family most suitable to bind JANEX-1 and demethylate the molecule at the C-7 position of the quinazoline ring.
O-Demethylation has been reported for other 6,7-dimethoxy-quinazoline compounds, including prazosin (Taylor et al., 1977), doxazosin (Kaye et al., 1986), and DDQ (Yamato et al., 1982). The metabolic pathway of JANEX-1 is similar to that of DDQ, which was biotransformed in vivo to form only the 7-demethylation metabolite, but it is different from the metabolism of prazosin and doxazosin, which were metabolized in vivo to form both 6- and 7-demethylation metabolites. However, the detailed metabolic pathways of prazosin, doxazosin, and DDQ have not been reported.
CYP3A is the most abundant cytochrome P450 enzyme subfamily in human liver, which is abundantly expressed in the intestinal mucosa (Kronbach et al., 1989; Crespi, 1995). CYP3A did not appear to be involved in the demethylation of JANEX-1. This may account for good oral bioavailability of JANEX-1 in animals (Uckun et al., 1999b). In human tissues from nonsmokers, the CYP1A1 gene is expressed at very low levels (Crespi, 1995). Due to the low abundance of CYP1A1 in human tissues, CYP1A1 would be unlikely to control the metabolism of JANEX-1 in patients. CYP1A2 activity varies considerably from individual to individual and appears to be modulated by environmental factors (Crespi, 1995). Therefore, the O-demethylation of JANEX-1 is likely to show a significant patient-to-patient variation in clinical settings. It is also likely that the metabolism of JANEX-1 in patients will be affected by other drugs that are metabolized by or induce CYP1A subfamily members. Finally, further characterization of the other routes of excretion and metabolism pathways is required to better understand the metabolite pharmacokinetics of JANEX-1-M.
Acknowledgments
We thank H. Bergstrom, G. Mitcheltree, Dr. S. Kazi, and Dr. T. K. Venkatachalam for assistance in this study.
Footnotes
- Abbreviations used are::
- TNF
- tumor necrosis factor
- PTK
- protein tyrosine kinases
- JAK3
- Janus kinase-3
- JANEX-1
- 4-(4′-hydroxyphenyl)-amino-6,7-dimethoxyquinazoline
- HPLC
- high-performance liquid chromatography
- JANEX-1-M
- 4-(4′-hydroxyphenyl)-amino-6-methoxy-7-hydroxyquinazoline
- P450
- cytochrome P450
- FMO
- flavin-containing monooxygenase
- ARO
- Aroclor 1254
- BNF
- β-naphthoflavone
- CFB
- clofibrate
- DEX
- dexamethasone
- ISO
- isoniazid
- 3-MC
- 3-methylcholanthrene
- PHEN
- phenobarbital
- DMSO
- dimethyl sulfoxide
- MS
- mass spectrometry
- PBS
- phosphate-buffered saline
- LC
- liquid chromatography
- DNP
- dinitrophenyl
- PIPES
- piperazine-N,N′-bis(2-ethanesulfonic acid)
- BSA
- bovine serum albumin
- TEA
- tetraethylammonium
- CL
- clearance
- RT
- retention time
- VDW
- van der Waals
- DDQ
- 2,3-dichloro-5,6-dicyano-1,4-benzoquinone
- Received July 9, 2001.
- Accepted September 27, 2001.
- The American Society for Pharmacology and Experimental Therapeutics