Abstract
5-Isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)-N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)pyrrolo[2,1-f][1,2,4]triazin-4-amine (BMS-645737) is a potent and selective vascular endothelial growth factor receptor-2 antagonist. In this study, liquid chromatography/tandem mass spectrometry and NMR were used to investigate the biotransformation of BMS-645737 in vitro and in the cynomolgus monkey, dog, mouse, and rat. Metabolic pathways for BMS-645737 included multistep processes involving both oxidation and conjugation reactions. For example, the 2-methyl-1H-pyrrolo moiety underwent cytochrome P450-catalyzed hydroxylation followed by oxidation to a carboxylic acid and then conjugation with taurine. Alternatively, the 5-methyl-1,3,4-oxadiazol-2-yl moiety was metabolized by hydroxylation and then conjugation with sulfate. The pyridin-5-yl group underwent direct glucuronidation in hepatocytes (dog, monkey, human) and conjugation with N-acetylglucosamine in the monkey. Conjugation with glutathione and processing along the mercapturic acid pathway was a minor metabolic pathway in vivo, although BMS-645737 did not form conjugates in the presence of glutathione-supplemented liver microsomes. Other minor biotransformation pathways included oxidative dehydrogenation, dihydroxylation, and hydrolytic opening of the oxadiazole ring followed by either deacetylation or hydrolysis of the resulting diacyl hydrazide. Whereas previous studies have shown the formation of N-acetylglucosamine conjugates of alcohols, arylamines, and other small molecules, this report describes the biotransformation of a heterocyclic aromatic amine via direct conjugation with N-acetylglucosamine.
In recent years, vascular endothelial growth factor (VEGF) receptor tyrosine kinases have garnered interest as potential therapeutic targets in drug discovery. These receptors are imbedded in the plasma membranes of vascular endothelial cells and are activated by a homologous series of VEGF ligands, most notably a dimeric 45-kDa glycoprotein that is referred to as VEGF or VEGF-A. Binding of VEGF causes these receptors to dimerize and undergo phosphorylation of select tyrosine moieties in the catalytic domain, initiating a cascade that leads to angiogenesis and the growth of blood vessels in the vascular endothelium. In healthy subjects, this process seems to be limited to special circumstances including vasculogenesis during embryonic development (Carmeliet et al., 1996; Ferrara et al., 1996), the growth of bone from cartilage (Kronenberg, 2003), ovarian function (Phillips et al., 1990; Goede et al., 1998), and the healing of damaged tissue (Brown et al., 1992). However, VEGF-mediated angiogenesis is a hallmark of certain hyperproliferative diseases including cancer (Carmeliet, 1996; Jain, 2005) and rheumatoid arthritis (Paleolog and Miotla, 1998). Hence, VEGF receptors offer an attractive target for therapeutic intervention in these diseases, and in particular, VEGF receptor-2, which most closely controls angiogenesis (Kadambi et al., 2001) as well as cancer cell migration, proliferation, and differentiation, stands out as a target of choice in the treatment of cancer. The efficacy of the monoclonal, anti-VEGF antibody bevacizumab (Avastin; Genentech, South San Francisco, CA) in treating metastatic renal cell (Yang et al., 2003) and rectal carcinomas (Willett et al., 2004) provided a clinical proof-of-concept for sequestering VEGF in cancer therapy. Small-molecule VEGF receptor antagonists with varying selectivity have also shown promise in the clinic (Bold et al., 2000; Hennequin et al., 2002; Abrams et al., 2003; Beebe et al., 2003; Gingrich et al., 2003; Traxler, 2003; Wedge et al., 2005; Zakarija and Soff, 2005).
BMS-645737 (1) is a VEGF receptor-2 antagonist (IC50 = 25 nM) belonging to a series of pyrrolotriazine derivatives disclosed previously (Borzilleri et al., 2005; Ruel et al., 2008) (Fig. 1). The biotransformation of compounds with similar structural moieties has been studied previously in the context of other small molecule therapeutics. For example, the dopamine D4-selective antagonist L-745,870 was metabolized in rats, rhesus monkeys, and humans by successive oxidations α to azaindole C3, giving 7-azaindole-3-carboxylic acid, and by conjugation to N-acetylcysteine, suggesting a bioactivation pathway proceeding via transient imine-methide and glutathionyl intermediates (Zhang et al., 2000). The biotransformation of structurally related indoles, which are used more commonly as building blocks for drug-like molecules, has also been reviewed (Dalvie et al., 2002). Regarding 5-methyl-1,3,4-oxadiazoles, the antidiarrheal agent nufenoxole showed good stability in plasma, urine, and human gastric fluid and was metabolized in part by hydroxylation of the 5-methyl moiety in rats and rhesus monkeys but not humans (Cook et al., 1990).
Endogenous steroids such as the bile acid ursodeoxycholic acid (Marschall et al., 1989) and certain metabolites of progesterone (Meng et al., 1996) are eliminated as N-acetylglucosamine (GlcNAc) conjugates in human urine. Conjugation with amino sugars also occurs during the metabolism of certain xenobiotics including selenium (Kobayashi et al., 2002), bisphenol A (Zalko et al., 2003), and nucleotide derivatives such as bohemine (Chmela et al., 2001) and 5-fluorouridine (Weckbecker and Keppler, 1984). The public literature contains one report of arylamine biotransformation in the cynomolgus monkey by this pathway (Chang et al., 1997), although no GlcNAc conjugates of heterocyclic aromatic amines seem to have been reported previously.
In the present study, the biotransformation of 1 was investigated in vitro and in laboratory animals to assist in a broad evaluation of 1 as a clinical development candidate for the treatment of cancer. In light of the precedents cited above on the biotransformation of azaindole and oxadiazole functional groups, it was hypothesized that multiple oxidative and conjugative pathways would contribute to the metabolism of 1. This hypothesis was tested by using a variety of synthetic, biosynthetic, and analytical techniques to elucidate the structures of metabolites of 1 and characterize the biotransformation pathways involved in their formation.
Materials and Methods
Substances and Enzymes. 1 and compounds 2, 3, 4, 5, and 6 were synthesized chemically at Bristol-Myers Squibb Company (Princeton, NJ). All the chemical reagents including glutathione (GSH), potassium cyanide (KCN), NADPH, UDP-GlcNAc, and UDP-N-acetylgalactosamine (GalcNAc) were purchased from Sigma-Aldrich (St. Louis, MO) unless noted otherwise. All the solvents [high-performance liquid chromatography (HPLC) grade] were purchased from Mallinckrodt Baker (Phillipsburg, NJ). Liver microsomes and S9 (cynomolgus monkey, dog, mouse, rat, human) and expressed human cytochrome P450 (P450) enzymes (Supersomes CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C18, CYP2C19, CYP2D6, CYP2E1, CYP3A4, and CYP3A5) were obtained from BD Biosciences (San Jose, CA). Aroclor 1254-induced rat liver S9 was acquired from Molecular Toxicology, Inc. (Boone, NC).
Preparation of Samples. Compound 1 was incubated together with 1 mM NADPH and liver microsomes (dog, human, monkey, mouse, rat; 1 mg of protein/ml) or liver S9 (dog, human, monkey, rat, Aroclor 1254-induced rat; 5 mg of protein/ml) using substrate concentrations of 30 and 10 μM, respectively, in 50 mM phosphate buffer, pH 7.4. Incubations (n = 3) of 1 (1 μM) were also carried out in Supersomes expressing the CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C18, CYP2C19, CYP2D6, CYP2E1, CYP3A4, and CYP3A5 (100 pmol of P450 protein/ml). Incubations of the diacyl hydrazide 2, the product of a reaction where the oxadiazole ring of 1 was opened hydrolytically, were also carried out in liver S9 (dog, human, Aroclor-induced rat; 5 mg of protein/ml) to monitor the formation of 3. To scale up the formation of 5 and 6, 2 mg of 1 [predissolved in 200 μl of dimethyl sulfoxide (DMSO)] was mixed with 15 mg of NADPH and 20 ml of 25 mM sodium phosphate buffer, pH 7.4. Depending on the metabolite to be generated, this solution was combined with either 1 ml of CYP2C19 Supersomes (5) or 2 ml of Aroclor-induced rat liver S9 (6). Samples were incubated at 37°C with constant, gentle shaking and were quenched after 1 h (2 h in scale-up experiments) by adding equal volumes of acetonitrile. Supernatants were analyzed using liquid chromatography (LC)/UV/tandem mass spectrometry (MSn).
Simulated gastric fluid (SGF), U.S. Pharmacopeia (USP) was prepared by combining 2.0 g of sodium chloride, 3.2 g of purified porcine pepsin, 7.0 ml of hydrochloric acid, and sufficient water to make 1 liter. In an experiment to determine the stability of 1 in SGF over time, approximately 1.2 mg of 1 (free base) and 1.5 mg of 1 (bisulfate salt) were weighed into each of several 10-ml volumetric flasks. To each flask, 1 ml of prewarmed SGF, USP was added, and the flasks were stored at 37°C. The pH of these mixtures was approximately 1.2. At each time point (0, 0.5, 1, 2, 3, 4, 5, 29 h), one flask each of free base and salt was removed from the oven and quenched with 50% aqueous (pH 6.5 phosphate buffer) acetonitrile. The final pH after quenching was approximately 4.0. Aliquots of these solutions were analyzed by HPLC/UV.
The hydroxyl metabolites 5 and 6 were isolated from in vitro preparations as follows: supernatants were concentrated to 12 ml under a nitrogen stream to remove acetonitrile, and the concentrate was diluted with water to 20 ml. This aqueous solution was applied to a Waters (Milford, MA) Oasis HLB (12 cc, 500 mg) prepacked LP extraction cartridge, washed using a house vacuum with water (50 ml), and then eluted with methanol (50 ml). The methanol eluate was evaporated to dryness and chromatographed using a YMC, Inc. (Wilmington, NC) Pro C18 S-5 (4.6 × 150 mm) column. The mobile phase consisted of 0.1% trifluoroacetic acid in water (solvent A) and acetonitrile (solvent B) at a flow rate of 1.2 ml/min with a linear gradient of 10 to 50% B over 25 min; analytes were detected by UV (254 nm). The hydroxyl metabolite 5 (retention time = 9.8 min) was obtained from the incubation with expressed CYP2C19, and 6 (retention time = 10.6 min) was obtained from the Aroclor-induced rat liver S9 incubation. Solutions containing both metabolites were collected and lyophilized (<100 μg each).
Incubations of 1 were carried out in the presence of liver microsomes (human, rat) supplemented with NADPH, GSH, and KCN to generate surrogates for chemically reactive metabolites that might have been generated. These samples were prepared by combining 10 μM parent compound, 1 mM NADPH, either 5 mM GSH or 1 mM KCN, and microsomes (1 mg/ml protein) in a 100 mM sodium phosphate buffer solution, pH 7.4, and incubating at 37°C for 30 min. Reactions were quenched by adding equal volumes of acetonitrile, and the supernatant liquids were recovered for analysis using LC/UV/MSn.
The biotransformation of 1 was evaluated in suspensions of hepatocytes isolated from the CD-1 mouse, Sprague-Dawley rat, beagle dog, cynomolgus monkey, and human. Fresh suspensions of mouse and rat hepatocytes were prepared in-house using a published method (Berry et al., 1991). These hepatocytes were purified by centrifuging a cell suspension (12.5 ml) with Percoll solution (12.5 ml) at 50g and 4°C for 5 min. Cells were washed with a suspension buffer and resuspended in an incubation buffer. Cryopreserved human hepatocytes (lot IEM, ECM, HRU), dog hepatocytes (lot LPB), and monkey hepatocytes (lot TGR) were obtained from In Vitro Technologies (Baltimore, MD). The gel-entrapped dog, monkey, and human hepatocytes were processed in a stepwise manner according to the vendors' instructions, washed with a Krebs-Henseleit suspension buffer, pH 7.4, and resuspended in a Krebs-Henseleit incubation buffer fortified with glucose. Mouse, rat, dog, and monkey hepatocytes were from a single donor, whereas the cryopreserved human hepatocytes were pooled from three donors. Cell viability, determined via trypan blue exclusion, was >85 and >65% for fresh and cryopreserved hepatocytes, respectively. Compound 1 (30 μM; final organic solvent content 0.15%) was incubated with hepatocytes (n = 3; density = 0.67 × 106 cells/ml) for 2 h at 37°C, 95% humidity and 5% CO2. Reactions were quenched using equal volumes of acetonitrile. Supernatants were stored at -20°C and analyzed using LC/UV/MSn. Compound 2 was also incubated with hepatocytes (human, mouse) as described above to monitor the formation of 3. Samples containing hepatocytes (mouse, rat, dog, monkey, human) and either 7-ethoxycoumarin (20 μM) or 7-hydroxycoumarin (20 μM) were prepared as positive controls, and the rate of disappearance of these compounds was monitored to verify the viability of phase I and phase II enzymes in these systems.
Studies Using Animals. All the methods were in accordance with the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996) as adopted by the U.S. National Institutes of Health. Plasma, bile, urine, and gastrointestinal samples were collected from fasted bile duct-cannulated (BDC) rats (2 rats/group) dosed p.o. with 1 in 0.25% Methocel (Dow Chemical Company, Midland, MI; 30 mg/kg). Blood was collected in tubes containing EDTA at 0.25, 0.5, 1, 3, 6, and 9 h after dosing, and plasma was prepared by centrifugation. Bile was collected in 3-h segments over 9 h, and urine (0–9 h) was pooled into a single sample. At the end of the study, the gastrointestinal tracts and feces were collected and homogenized in 3 volumes of water. Samples were stored at -20°C until processed further. In support of toxicology studies, an additional high-dose biotransformation study was conducted in noncannulated, nonfasted male Sprague-Dawley rats (n = 3) that were administered 1 p.o. via gavage (100 mg/kg). The vehicle was 100% polyethylene glycol (PEG) 400, and the dosing volume was 10 ml/kg. On the final day of a 10-day toxicokinetics study in female athymic nude mice, blood samples were collected at 1, 5, 8, and 24 h after p.o. administration of 1 (15, 30, 45, and 60 mg/kg/day) in PEG 400/water (1:1). These samples were allowed to clot on ice and then were centrifuged to obtain serum. Plasma and urine samples were also collected after single p.o. doses of 1 [5 mg/kg in PEG 400/0.01 N HCl (1:1)] to male cynomolgus monkeys or male beagle dogs (n = 3 for each species). Blood was collected in tubes containing EDTA at 1, 2, 4, 6, 8, and 24 h after dosing, and plasma was obtained by centrifugation. Urine was collected continuously over 24-h intervals. The plasma, urine, and bile samples from each time point were pooled and treated with an equal volume of acetonitrile (2 volumes for plasma and serum), and all the samples were centrifuged for 5 min at 1000g to obtain supernatants. These supernatants were stored at 4°C and analyzed within 24 h using LC/UV/MSn.
Isolation of Selected Metabolites. Urine was also collected over 0 to 8- and 8 to 24-h intervals after single-dose p.o. administration of 1 to male cynomolgus monkeys (30, 100, 300 mg/kg) for use in isolating 7, 10, and 11 in quantities sufficient for NMR analysis. The N-linked sugar conjugates 10 and 11 were hydrolytically labile, reverting to 1 and the corresponding sugar moiety slowly at -20°C or more rapidly under acidic conditions at room temperature; hence, these metabolites were exposed to acid only when necessary and for limited amounts of time. For the isolation, an aliquot from the 8- to 24-h urine sample was treated with an equal volume of acetonitrile, centrifuged to remove insoluble matter, and concentrated under nitrogen. Metabolites of interest were concentrated further via solid-phase extraction. In general, a 100-ml aliquot was applied to a Waters Oasis HLB (Waters) (35 cc, 6 g) prepacked LP extraction cartridge, washed (house vacuum) sequentially with water (250 ml) and methanol/water 1:1 (150 ml), and then eluted with methanol (100 ml). The methanol eluate was dried and subjected to preparative HPLC on a Beckman “System Gold” Semipreparative Gradient HPLC, model LC126 (Columbia, MD), using a YMC Pro C18 S-5 (20 × 150 mm) column. The mobile phase consisted of 0.1% trifluoroacetic acid in water (solvent A) and acetonitrile (solvent B) at a flow rate of 20 ml/min with a linear gradient of 15 to 35% B over 25 min; analytes were detected by UV (254 nm). After an injection of extracted material (16 mg/200 μl in DMSO), the glucuronide 10 (15.6 min), GlcNAc conjugate 11 (16.5 min), and carboxylate 7 (19.8 min) were collected immediately and lyophilized (all <100 μg each).
Analytical Methods and Instrumentation. An LC/UV/MS method for the analysis of drug-related compound in biologic matrices was developed using an authentic standard of 1. Chromatographic separations were carried out using a Waters Alliance 2695 Separations Module and a Phenomenex (Torrance, CA) CuroSil-PFP (2.0 × 150 mm) column that was maintained at a temperature of 30°C. The mobile phase consisted of a 95:5 mixture of 0.1% aqueous acetic acid and acetonitrile (solvent C) and acetonitrile (solvent D) at a flow rate of 300 μl/min with the following linear gradient conditions: 0% D for 1 min, 0 to 60% D over 21 min, 60 to 100% D over 3 min, and 100% D isocratic for 4 min. The eluate from the HPLC column was routed on-line to a Thermo Fisher Scientific (Waltham, MA) Surveyor photodiode-array detector and then to a Thermo Fisher Scientific LCQ Deca XP ion trap mass spectrometer. Samples were analyzed using positive-ion electrospray LC/MSn with data-dependent product-ion scanning. Instrument settings were as follows: m/z range = 100 to 900, capillary temperature = 350°C, nitrogen sheath gas flow rate = 45 (arbitrary units), auxiliary gas = 15, spray voltage = 5.0 kV, capillary voltage = 13 V, and tube lens offset = 35 V. Selected samples were also analyzed using a Thermo LTQ-Orbitrap tandem mass spectrometer for chemical formula determination via accurate mass measurement. Instrument settings were as follows: MS resolution = 60,000, scan rate = 1/s, m/z range = 95 to 1200, capillary temperature = 325°C, sheath gas = 45, auxiliary gas = 10, source voltage = 4.5 kV, capillary voltage = 30 V, and tube lens offset = 85 V. In addition, 1 exhibited a UV-absorption maximum of 303 nm. This wavelength was monitored during the analysis of all the samples, and the absorbance of drug-related components at this wavelength was used to make judgments about the relative levels of metabolites therein. It should be noted that this method of semiquantification was limited by potential differences between the molar absorptivities of 1 and metabolites, and also by the possibility that these components were not resolved chromatographically from endogenous interferences that absorb UV light.
Isolated metabolites were analyzed using a Varian, Inc. (Palo Alto, CA) Inova 600-MHz NMR spectrometer equipped with a 5-mm triple-resonance cold probe. The temperature of the probe was maintained at 25°C during all the experiments. The metabolites were dissolved in 50 μl of DMSO-d6 and transferred to 1.7-mm capillary tubes to maximize NMR detection sensitivity, and small amounts of deuterium chloride were added to enhance peak resolution as needed. The data collected for each metabolite included a 1H NMR spectrum and a two-dimensional (2D) 1H-13C correlation [heteronuclear single quantum correlation (HSQC)] spectrum. Nuclear Overhauser enhancement (NOE) experiments (one-dimensional or 2D) were performed as needed, and 2D long-range 1H-13C correlation (heteronuclear multiple-bond correlation spectroscopy) spectra were acquired when sufficient sample was available. Structures of metabolites were elucidated by comparing their 1H and 13C chemical shifts with those of 1 under similar solution conditions.
Results
Positive-ion electrospray MS/MS product ions of 1 were generated via collision-induced dissociation (CID) and analyzed using a Thermo Fisher Scientific LTQ-Orbitrap high-resolution mass spectrometer. The monoisotopic mass-to-charge ratios and proposed molecular formulas of these product ions were as follows: m/z 306.1105, C15H12N7O+ (error = 2.3 ppm); m/z 347.1371, [C17H14N8O + H]+ (error = 2.3 ppm); m/z 348.1572, C18H18N7O+ (error = 1.4 ppm). These observations, which were consistent with neutral losses of C3H6 (isopropyl group), C2H3N (oxadiazole fragment), and C5H9N (both isopropyl and oxadiazole fragments) to give the product ions of m/z 347, 348, and 306, respectively, were used to assist in elucidating the structures of metabolites of 1. Compound 1 was incubated with Supersomes containing specific recombinant P450 enzymes, liver microsomes (dog, human, monkey, mouse, rat), liver S9 (dog, human, monkey, rat, Aroclor-induced rat), hepatocytes (dog, human, monkey, mouse, rat), and simulated gastric fluid. It was also administered p.o. to dogs, monkeys, mice, rats, and BDC rats. The biotransformation of 2 was studied using liver S9 (dog, human, Aroclor-induced rat) and hepatocytes (human, mouse). The results of these studies are summarized in Table 1, and proposed biotransformation pathways are illustrated in Fig. 2.
The stability of free-base and bisulfate-salt forms of 1 as suspensions in SGF, USP was examined to simulate in vivo conditions after p.o. dosing and to determine the extent of formation of 2 under these conditions. The percentages of 2 that were formed at various time points are listed in Fig. 3. The rate of conversion to 2 was faster for the bisulfate salt than for the free base, most likely because of the higher solubility of the salt form. Compound 2 was not detected after the incubation of 1 in other in vitro systems, such as liver microsomes or S9. However, 3 was detected after incubations of 2 with liver S9 and hepatocytes, and in S9 this reaction occurred in the presence and absence of NADPH (data not shown).
Although the metabolic turnover of 1 (VEGF receptor-2 IC50 = 25 nM) was modest in microsomes from each species, the most abundant metabolites seemed to be the hydroxylation products 5 (IC50 = 27 nM) and 6 (IC50 > 0.4 μM). The regiochemical arrangement of 5 was confirmed by comparison to an authentic standard and via HSQC NMR (Tables 2 and 3) after its isolation from a large-scale incubation of 1 and CYP2C19 Supersomes. The hydroxylation product 6 was the most abundant metabolite generated by CYP3A4 Supersomes and liver S9 from monkey, dog, human, rat, and Aroclor-induced rat. Turnover was comparatively high in Aroclor-induced rat liver S9, where 7 and a dihydroxylation product were also formed. Considering this high extent of turnover, Aroclor-induced rat liver S9 was selected as a bioreactor for the formation of 6 in quantities that would allow further characterization. Analysis of this isolated metabolite by HSQC NMR showed attachment of the hydroxyl group to C28, a regiochemical assignment that was also supported by comparing the MS/MS fragmentation patterns of the isolated metabolite 6 with an authentic standard. Although minor metabolites in microsomes included the dehydrogenation product 12, no adducts were detected after incubations of 1 with microsomes (human, rat) supplemented with either GSH or KCN.
Whereas mouse serum collected 1 h after dosing contained 1 as the major drug-related component, samples collected at 5 and 8 h also contained high levels of 5 and 9. No drug-related compounds were detected in serum collected 24 h after dosing. The MS3 product-ion mass spectrum of 9, after removal of the sulfate moiety via CID, was similar to the MS/MS spectrum of 5. In contrast, plasma samples that were drawn after single p.o. doses of 1 to monkeys (5 mg/kg), dogs, and rats contained considerably lower levels of 5 than 1. Plasma samples from the monkey also contained low levels of 10. Urine collected from dog, monkey, and rat all contained 7 as a major metabolite, although 7 was not detected in circulation in any species.
After p.o. dosing of 1 to BDC rats, bile that was collected during the postdose interval of 6 to 9 h was rich in metabolites compared with samples collected during earlier intervals. Although 6 was not detected in rat bile, 7 through 9 were among the most abundant metabolites in this sample. In addition, the MS3 product-ion mass spectrum of 8, after removal of the putative taurine moiety via CID, was similar to the MS/MS spectrum of 7. The glutathione conjugate 13, which was not formed in GSH-supplemented rat liver microsomes, was a minor component of BDC rat bile. The fragmentation pattern of 13 included neutral losses of 75 Da (glycine moiety), 129 Da (γ-glutamyl moiety), and 307 Da (GSH). Other components of BDC rat bile collected at 6 to 9 h included 1 and 15 through 17.
Direct conjugation of 1 with certain sugars, resulting in the formation of the N-linked conjugates 10 and 11, contributed to metabolic clearance in hepatocytes from nonrodents. It is noteworthy that 11 was the most abundant metabolite in monkey hepatocytes as estimated by LC/UV and was also detected after the incubation of 1 with UDP-GlcNAc-supplemented monkey liver microsomes. The metabolites 7, 10, and 11 were also detected in a pooled urine sample after p.o. administration of 1 (30, 100, 300 mg/kg) to monkeys. Using a high-resolution mass spectrometer, the monoisotopic mass-to-charge ratios and molecular formulas of these metabolites were determined as follows: 7, [M+H]+ = 419.1579, C20H18N8O3 (error = 0.24 ppm); 10, [M+H]+ = 565.2153, C26H34N9O6 (1.1 ppm); and 11,[M+H]+ = 592.2629, C28H34N9O6 (0.51 ppm). During MS/MS with CID, the protonated molecule of 7 lost 18 Da (water), and those of 10 and 11 gave characteristic neutral losses of 176 and 203 Da, respectively, corresponding to the masses of glucuronic acid and GlcNAc. Regiochemical assignments for 7, 10, and 11 were obtained by isolating these metabolites from urine and analyzing them by NMR. In general, biotransformations produced only small changes in the chemical shifts of 13C and 1H atoms remote from the sites of metabolism, allowing metabolic soft spots to be identified by comparing the 1H spectra of parent and metabolites. During NMR analysis of both 10 and 11, an NOE correlation was observed between the anomeric proton of the sugar and H18 of the parent-compound moiety. The glucuronide 10 was further characterized by gradient heteronuclear multiple-bond correlation spectroscopy, where both the anomeric proton and H25 of the pyridinyl moiety exhibited 3-bond coupling to C18. Because NMR signals corresponding to the exchangeable 1H-pyrrolo proton were not observed during the analysis of either conjugate and because chemical shift predictions for C18 were inconsistent with the possibility of a quaternary N19 in proximity, the structures 10 and 11, with the tautomeric forms of the N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl) moiety shown in Fig. 4, were proposed. When dissolved in DMSO-d6, several drug-related compounds that were analyzed using NMR exhibited tautomerism on the NMR timescale, although the relative amounts of the tautomeric forms varied. For example, two tautomers of 1 and two tautomers of 10 were identified, with the major tautomer of 10 and the minor tautomer of 1 exhibiting the proton H11.
A few minor drug-related components were detected in urine after p.o. administration of 1 to monkeys at high doses (30, 100, and 300 mg/kg) but not at a low dose (5 mg/kg). These included 2, 4, and a putative mercapturic acid conjugate 14. The structure of 14 was proposed based on its molecular weight and fragmentation pattern, which included positive-ion electrospray MS/MS product ions of m/z 387 and 421, possibly corresponding to a neutral loss of N-acetylcysteine and the product ion RSH+, respectively. However, this metabolite was not present in a quantity sufficient for isolation and analysis by NMR, precluding a more detailed analysis of its structure.
The arylamine bridge connecting the N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl and pyrrolo[2,1-f][1,2,4] triazin-4-amine moieties was intact in all the drug-related compounds identified during biotransformation studies, i.e., no metabolites with free arylamine moieties were detected.
Discussion
Because the conversion of 1 to 2 was observed in simulated gastric fluid but not in other in vitro systems, the presence of 2 in urine (monkey, rat) was interpreted to follow from acid-catalyzed hydrolytic ring-opening of the oxadiazole moiety of 1 in the stomachs of these animals after p.o. dosing. The diacyl hydrazide group of 2 and the acyl hydrazide group of 3 were recognized as potential liabilities in light of certain adverse side effects that have been associated with isoniazid and other hydrazide-containing drugs. The bioactivation of such groups has been shown to involve the formation of nitrogen-centered or alkyl radicals (Goodwin et al., 1996) and to entail redox cycling in the presence of peroxidases and redox-active metal ions (Yamamoto and Kawanishi, 1991). None of the hydrazide-containing biotransformation products of 1 were detected in the blood of any animals in the present study, and compounds 2 and 4 were detected in monkey urine after high doses. Although it does not seem to be an issue for 1, these results do raise the question of whether acid-catalyzed formation of acyl hydrazides from p.o. delivered xenobiotics containing 1,3,4-oxadiazoles might be a general issue. As mentioned previously, another such compound was found to be stable in a simulated gastric environment (Cook et al., 1990), but the literature contains too few examples for this question to be answered conclusively.
The observation of 5 through 7 and 12 after incubations of 1 in human liver microsomes, S9, or hepatocytes showed the relevance of multiple pathways for the oxidative biotransformation of 1. To identify the enzymes that catalyzed these reactions, 1 was incubated discretely with 13 of the most important recombinant human P450s, and supernatants from these incubations were analyzed for the presence of metabolites. The results suggested that 5 was formed to a large extent by CYP2C19, CYP2C9, and CYP3A4, and the formation of 6 was catalyzed most effectively by CYP3A4, CYP1B1, and CYP2C8. Furthermore, the regiochemical arrangements of 6 and 7, as determined using synthetic standards and NMR, suggest that these metabolites were formed in sequence. A high degree of similarity between the MS/MS product-ion mass spectrum of 7 and the MS3 product-ion spectrum of 8 indicated that these metabolites were also formed along a common multistep pathway, and similar reasoning suggested that 5 and 9 were linked sequentially as well. These observations suggested that P450-catalyzed hydroxylations of C15 and C28 represented “trunk pathways” leading to the formation of downstream metabolites. Such pathways seemed to contribute to metabolic clearance in animals including the BDC rat, where 7 through 9 were abundant components of rat bile. Of these compounds, 7 was anticipated as a metabolite of 1 in humans based on its formation in human hepatocytes and occurrence in the urine of multiple species.
The GSH conjugate 13, which was detected in rat bile, also might have formed via multistep processes involving both oxidative and conjugative components. This pathway is poorly understood because it has not been observed in vitro and seems to contribute marginally to clearance in the BDC rat, complicating efforts to isolate a sufficient amount of 13 for analysis by NMR. In a previous study, the formation of a mercapturic acid conjugate of a 3-substituted 7-azaindole was proposed to proceed via a transient imine intermediate similar to that in the bioactivation of 3-methyl indoles (Zhang et al., 2000). However, it was difficult to locate the conjugation site of GSH in 13 using MS/MS because the N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl and pyrrolo[2,1-f][1,2,4] triazin-4-amine moieties did not separate readily from each other during CID. Nevertheless, it is reasonable to assume that 13 and 14 are linked sequentially via the mercapturic acid pathway because only one isomer of each metabolite was resolved chromatographically.
The formation of 11 in the monkey was an unexpected outcome that apparently represented a novel biotransformation pathway for heterocyclic amines. This metabolite was not detected in biologic fluids from the dog, mouse, or rat, and its formation in the human was not anticipated because it was generated by monkey hepatocytes but not human hepatocytes. Hence, conjugation with GlcNAc seems to be a monkey-specific pathway for the metabolism of 1. The sample-preparation methods used in this study, which did not entail the acidification of samples immediately after collection, enabled the identification of the N-linked sugars 10 and 11. Such acidification techniques are used commonly in biotransformation studies to stabilize acyl glucuronic acid conjugates and normalize the pH of bile and urine, which can vary considerably both between and among species. The acidification of samples can assist in the detection of certain metabolites but may have the unintended consequence of hydrolyzing N-linked sugars. Such an observation was made previously with respect to the 5-N-glucuronic acid conjugate of clozapine (Breyer-Pfaff and Wachsmuth, 2001). Furthermore, our experience in handling 10 and 11 suggests that both compounds may undergo acid-catalyzed hydrolysis to give 1 and the corresponding free sugar, especially in the presence of a strong acid such as deuterium chloride, which was used selectively in this study during the acquisition of NMR spectra.
The involvement of GlcNAc in drug metabolism is interesting because it is tied to biochemical processes that may fortify certain disease states including diabetes. Recently, post-translational modifications of serine and threonine residues by the enzyme uridine diphospho-N-acetylglucosamine/polypeptide β-N-acetylglucosaminyltransferase have emerged as an important mechanism for the functional regulation of certain proteins, including enzymes involved in metabolism. The hexosamine biosynthetic pathway, which provides the activated substrate UDP-GlcNAc for these reactions, consumes approximately 4% of cellular glucose and represents an important energetic commitment for the cell. Increased flux through the hexosamine biosynthetic pathway has been linked to insulin resistance, and the modification of key proteins by O-linked GlcNAc may constitute a negative-feedback loop for insulin signaling (Wells and Hart, 2003). Accordingly, the consumption of GlcNAc stores during the biotransformation of xenobiotics might be expected to influence cellular sugar metabolism. Although GlcNAc conjugation of 1 is known only to occur in monkeys, the dynamics of such processes as biochemical and biotransformation pathways compete for a common store of UDP-GlcNAc may be a topic for additional research as xenobiotic acceptors of GlcNAc are identified in humans.
In conclusion, 1 is a potent and selective VEGF receptor-2 antagonist that was considered for development as an angiogenesis suppressor for the treatment of cancer. This compound has 5-methyl-1,3,4-oxadiazol-2-yl and N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl) moieties that constitute soft spots for biotransformation in multiple species. In this study, 1 was metabolized by P450-catalyzed hydroxylation and other oxidative pathways, conjugation with sulfate, taurine, glucuronic acid, and GlcNAc (monkey only), and multistep pathways involving both oxidative and conjugative components. Formation of the GlcNAc conjugate 11 seems to represent a novel biotransformation pathway for heterocyclic amines, possibly with interesting implications for cellular metabolism.
Acknowledgments
We thank Thermo Fisher Scientific for the use of an LTQ-Orbitrap tandem mass spectrometer; Kevin J. McHale and Thomas Philip for assistance with sample preparation and analysis; Richard A. Westhouse and Ronald G. Lindahl for the collection of animal samples; and Nelly Aranibar, Joseph Fargnoli, Jonathan L. Josephs, David Rodrigues, Punit H. Marathe, Daniel R. Schroeder, and W. Griffith Humphreys for helpful discussions.
Footnotes
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.108.022624.
-
ABBREVIATIONS: VEGF, vascular endothelial growth factor; 1, BMS-645737, 5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)-N-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)pyrrolo[2,1-f][1,2,4]triazin-4-amine; GlcNAc, N-acetylglucosamine; 2, N′-acetyl-5-isopropyl-4-[(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)amino]pyrrolo[2,1-f][1,2,4]triazine-6-carbohydrazide; 3, 5-isopropyl-4-[(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)amino]pyrrolo[2,1-f]-[1,2,4]triazine-6-carbohydrazide; 4, 5-isopropyl-4-[(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)amino]pyrrolo[2,1-f][1,2,4]triazine-6-carboxylic acid; 5, (5-{5-isopropyl-4-[(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-yl)amino]pyrrolo[2,1-f][1,2,4]triazin-6-yl}-1,3,4-oxadiazol-2-yl)methanol; 6, (5-{[5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)pyrrolo[2,1-f][1,2,4]triazin-4-yl]amino}-1H-pyrrolo[2,3-b]pyridin-2-yl)methanol; GSH, glutathione; KCN, potassium cyanide; GalcNAc, N-acetylgalactosamine; HPLC, high-performance liquid chromatography; P450, cytochrome P450; DMSO, dimethyl sulfoxide; LC, liquid chromatography; MSn, MS/MS, tandem mass spectrometry; SGF, simulated gastric fluid; USP, U.S. Pharmacopeia; BDC, bile duct–cannulated; PEG, polyethylene glycol; 7, 5-{5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)pyrrolo[1,2-f][1,2,4]triazin-4-ylamino}-1H-pyrrolo[2,3-b]pyridine-2-carboxylic acid; 10, (Z)-3,4,5-trihydroxy-6-(5-{5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)pyrrolo[1,2-f][1,2,4]triazin-4(3H)-ylideneamino}-2-methyl-7H-pyrrolo-[2,3-b]pyridin-7-yl)tetrahydro-2H-pyran-2-carboxylic acid; 11, N-{4,5-dihydroxy-6-(hydroxymethyl)-2-(5-{5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)pyrrolo[1,2-f][1,2,4]triazin-4-ylamino}-2-methyl-7H-pyrrolo[2,3-b]pyridin-7-yl)tetrahydro-2H-pyran-3-yl}acetamide; 2D, two-dimensional; HSQC, heteronuclear single quantum correlation; NOE, nuclear Overhauser enhancement; CID, collision-induced dissociation; 9, (5-{5-isopropyl-4-(2-methyl-1H-pyrrolo[2,3-b]pyridin-5-ylamino)pyrrolo[1,2-f][1,2,4]triazin-6-yl}-1,3,4-oxadiazol-2-yl)methyl sulfate; 8, 2-(5-{5-isopropyl-6-(5-methyl-1,3,4-oxadiazol-2-yl)pyrrolo[1,2-f][1,2,4]triazin-4-ylamino}-1H-pyrrolo[2,3-b]pyridine-2-carboxamido)ethanesulfonate; L-745,870, 3-{[4-(4-chlorophenyl)piperazin-1-yl]-methyl}-1H-pyrrolo-2,3-β-pyridine; M3, three-stage tandem mass spectrometry.
- Received July 17, 2008.
- Accepted September 10, 2008.
- The American Society for Pharmacology and Experimental Therapeutics