Metabolic Activation of a Pyrazinone-Containing Thrombin Inhibitor

198

Chem. Res. Toxicol. 2003, 16, 198-207

Metabolic Activation of a Pyrazinone-Containing Thrombin Inhibitor. Evidence for Novel Biotransformation Involving Pyrazinone Ring Oxidation, Rearrangement, and Covalent Binding to Proteins Rominder Singh,*,† Maria V. Silva Elipe,‡ Paul G. Pearson,† Byron H. Arison,‡ Bradley K. Wong,† Rebecca White,† Xiao Yu,† Christopher S. Burgey,§ Jiunn H. Lin,† and Thomas A. Baillie† Departments of Drug Metabolism, Merck Research Labs, West Point, Pennsylvania and Rahway, New Jersey, and Department of Medicinal Chemistry, Merck Research Labs, West Point, Pennsylvania Received September 30, 2002

Compound I, (2-{3-[(2,2-difluoro-2(2-pyridyl)ethyl)amino]-6-methyl-2-oxohydropyrazinyl}-N[(3-fluoro(2-pyridyl))methyl]acetamide, is a potent competitive inhibitor of thrombin that reacts stoichiometrically with the protease. Compounds of this class possess therapeutic potential as anticoagulation agents. During the metabolic characterization of compound I, evidence was obtained for extensive metabolic activation of the pyrazinone ring system. Following administration of 14C-labeled I to rats, significant levels of irreversibly bound radioactivity to proteins were detected in rat plasma and liver. LC/MS/MS analysis of metabolites formed in rat and human liver microsomes fortified with glutathione (GSH) revealed the presence of two structurally distinct GSH adducts. It is proposed that the first of these GSH conjugates derives from a two electron oxidation of the 6-methyl-2-oxo-3-aminopyrazinone moiety to afford an electrophilic imine-methide intermediate, while the second is formed by addition of GSH to an epoxide formed by P-450-mediated oxidation of the double bond at the 5-6 position of the pyrazinone ring. The addition of GSH to the proposed epoxide facilitates opening of the pyrazinone ring and a rearrangement to afford a stable, rearranged imidazole-containing metabolite. Elucidation of the metabolic activation pathways of I provides structural guidance for the design of thrombin inhibitors with decreased potential for the generation of chemically reactive intermediates.

Introduction Thrombin plays a key role in hemostasis by mediating conversion of fibrinogen to fibrin and activating platelets. Thrombin inhibitors prevent the formation of intravascular clots and thus are indicated for the treatment of cardiovascular diseases such as myocardial infarction, arterial fibrillation, deep vein thrombosis, and ischemic stroke. The clinical utility of the available thrombin inhibitors (1) is limited by their narrow therapeutic index, slow onset and offset of action, and large inter- and intraindividual variability (2). These shortcomings have stimulated efforts to develop new oral anticoagulants (3). Thrombin inhibitors represent a potential new class of drug that may offer a therapeutic alternative for the treatment and prevention of thromboembolic disorders (4). These inhibitors are potent and specific inhibitors of coagulation that work via the intrinsic or extrinsic pathways, which permit pharmacological treatment and prevention of thrombotic disorders (5). * To whom correspondence should be addressed. Tel: 484-344-2806. Fax: 484-344-7058. E-mail: [email protected]. † Department of Drug Metabolism, Merck Research Labs, West Point, Pennsylvania. ‡ Department of Drug Metabolism, Merck Research Labs, Rahway, New Jersey. § Department of Medicinal Chemistry, Merck Research Labs, West Point, Pennsylvania.

Compound I1 is a potent oral thrombin inhibitor that displays efficacy both in vitro and in vivo in preclinical animal models and belongs to a series of pyrazinone acetamide derivatives (6). As part of a detailed metabolic characterization of compound I, evidence was obtained for extensive metabolic activation of the compound to chemically reactive intermediates. The purpose of this study was to identify the metabolically labile sites on the compound through nucleophilic trapping experiments with GSH (7, 8). This approach revealed the operation of two distinct metabolic activation pathways, both of which involved oxidation of the pyrazinone moiety and, in one case, a novel rearrangement of this heterocycle.

Experimental Procedures Chemicals. All chemicals and solvents were of HPLC or analytical grade. Water was distilled and purified through a Milli-Q reagent system (Millipore Corp., Bedford, MA). Compound I and two of its metabolites (M1 and M4) were prepared 1 Abbreviations: ACN, acetonitrile; ESI, electrospray ionization; compound I, (2-{3-[(2,2-difluoro-2(2-pyridyl)ethyl)amino]-6-methyl-2oxohydropryrazinyl}-N-[(3-fluoro(2-pyridyl))methyl]acetamide; CID, collision-induced dissociation; GSH, glutathione; gHMBC, gradient heteronuclear multiple bond correlation; gHSQC, gradient heteronuclear single quantum correlation; NCE, new chemical entities; NOESY, nuclear Overhauser effect spectroscopy; TFA, trifluoroacetic acid.

10.1021/tx025635l CCC: $25.00 © 2003 American Chemical Society Published on Web 01/24/2003

Activation and Rearrangement of a Pyrazinone Ring at Merck Research Labs (West Point, PA). Radiolabeled I was prepared by incorporation of carbon-14 at the C-9 position (Merck Research Labs, Rahway, NJ); the product exhibited a specific activity of 118.7 µCi/mg and a radiochemical purity of >98% as determined by radio-HPLC. Glucose 6-phosphate, NADP+, NADPH, glucose 6-phosphate dehydrogenase, and GSH (reduced) were obtained from Sigma Chemical Co. (St. Louis, MO). All other chemicals were purchased from standard vendors and were of the highest purity available. Human liver microsomal pools were purchased from Human Biologics, Inc. (Scottsdale, AZ). LC/MS/MS. The HPLC system consisted of a Perkin-Elmer SYS-S200 autosampler (Perkin-Elmer Corp., Norwalk, CT) and a Rheos HPLC pump (Flux Instruments, Switzerland) attached to a BDS hypersil C18 (2.1 mm × 150 mm, 5 µm) column (Keystone Scientific, Bellfonte, PA). The mobile phase was delivered at 0.2 mL/min and began at 95% solvent A (0.1% formic acid) and increased linearly to 90% solvent B (ACN) in 18 min. The effluent was introduced into a Finnigan LCQ ion trap mass spectrometer (Finnigan MAT, San Jose, CA) equipped with an ESI source. The instrument was operated at a 5 kV potential in the positive ionization mode, and the capillary temperature was maintained at approximately 200 °C. NMR. The spectra were recorded in deuterated methanol (CD3OD) at 25 °C in 3 mm NMR tubes. 1H spectra were acquired at either 400 or 500 MHz using Varian Unity 400 and Varian Inova 500 instruments (Varian, Palo Alto, CA), respectively. In a similar manner, 13C spectra were acquired at 100 or 125 MHz using Varian Unity 400 or Varian Inova 500 instruments, respectively. Data were processed using linear prediction for the heteronuclear experiments. Chemical shifts are reported on the δ scale (ppm) by assigning the residual solvent (methanol) peak to 3.30 and 49.0 ppm for 1H and 13C, respectively. The 1H NOE (nuclear Overhauser effect) difference spectra at 500 MHz were acquired using a frequency list of one irradiation point (on-resonance) and one control (off-resonance) with a relaxation delay of 3.0 s. The difference spectrum was obtained by subtracting the control (off-resonances) from the other signals (on-resonances). The two-dimensional (2D) 1H1H NOESY spectrum was acquired with a spectral width of 3998.8 Hz into 1K data points in the f2 dimension and with 394 increments in the f1 dimension. The delay between successive pulses was 3 s, the mixing time was 0.3 s, and the 90° pulse was 5.35 µs. The 2D gHSQC and gHMBC experimental data were acquired with spectral widths of 3399.0 and 22624.4 Hz for 1H and 13C dimensions, respectively, 1K data points in the f2 dimension and with 207 and 247 increments in the f1 dimension for gHSQC and gHMBC, respectively. The 90° pulses were 5.35 µs for 1H and 12.0 µs for13C. The 2D gHSQC experiments used a delay of 3.6 ms (1/2JCH) during acquisition. For the low pass, delays corresponding to a 13C-1H coupling constant of ca. 140 Hz were used, while values of 5-7 Hz were used for the gHMBC experiments (13C-1H long-range coupling constants). Microsomal Incubations. The metabolism of compound I (2 µM) was investigated by incubating the drug in phosphate buffer (pH 7.4) with rat and human liver microsomes (1 mg protein/mL) in the presence of NADPH (1 mM) at 37 °C for 60 min. Incubations in the absence of NADPH, or in heatinactivated (boiled) rat or human liver microsomes, served as negative controls. Selective incubations were supplemented with GSH (5 mM) to trap electrophilic drug metabolites. The incubations were terminated by addition of two volumes of ACN, followed by vortexing and centrifugation to pellet the precipitated protein. The supernatant was evaporated and reconstituted in HPLC mobile phase and analyzed by LC/MS/MS. Metabolites M1-M4 were isolated using preparative HPLC from rat and human microsomal incubates. Covalent Binding. 1. In Vivo Studies. Oral doses (10 and 100 mg/kg) of 14C-labeled I were administered orally to rats (∼10 µCi/rat) as a suspension in 0.5% methyl cellulose. Plasma and liver tissue were harvested at predose, 1, 4, and 24 h postdose.

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 199 For removal of unbound radioactivity, aliquots of liver tissue homogenate and plasma were extracted repeatedly using 0.4 N trichloroacetic acid, methanol/water (80/20, v/v), and diethyl ether/ethanol (1/3, v/v). The extractions were repeated until the radioactivity in the supernatant had fallen to background levels. The protein pellets were solubilized in 1 N sodium hydroxide and neutralized, and the radioactivity remaining in the pellet was determined by liquid scintillation spectrometry. The time zero control samples consisted of predose plasma and liver tissue spiked with radiolabeled I at a concentration corresponding to that in the postdose samples. Protein concentrations were determined by the method of Lowry. All animal studies were approved by MRL Institutional Animal Care and Use Committee (IACUC). 2. In Vitro Studies. To assess the potential for covalent binding in vitro, radiolabeled I (2 µM) was incubated with rat and human liver microsomes (2 mg protein/mL) under the conditions described above for 60 min. The amount of radioactivity irreversibly bound to protein was determined using the same procedure as that for rat plasma and liver tissue (vide infra).

Results Identification of Metabolites of Compound I. LC/ MS/MS analysis of products formed in human and rat liver microsomal incubations fortified with GSH afforded the total ion current chromatograms depicted in Figure 1, which revealed the presence of four metabolites (M1M4), none of which were evident in the control incubations devoid of NADPH. Detailed LC/MS/MS and NMR spectroscopic analyses of the parent compound I were carried out to facilitate the structural characterization of metabolites. The full scan mass spectrum of I showed a prominent protonated molecular ion (MH+) at m/z 433, CID of which afforded an abundant ion at m/z 307 corresponding to cleavage of the acetamide linkage (Figure 2). The aromatic region of the 1H NMR spectrum of I indicated the presence of two aromatic systems corresponding to the two pyridine rings, one of which was 2-monosubstituted (8.62, 7.49, 7.93, and 7.70 ppm) and the other 2,3-disubstituted (8.34, 7.38, and 7.57 ppm) (Table 1). A singlet proton at 6.62 ppm was assigned to the vinyl position at C-8 on the pyrazinone ring. The aliphatic region showed the presence of three methylenes and one methyl group. The signal at 2.14 ppm was assigned to the methyl group on the pyrazinone ring, and one of the signals at 4.24 ppm was assigned as the methylene group next to fluorine, exhibiting the characteristic 1H-19F coupling constant of 14 Hz. The other two methylenes (4.78 and 4.60 ppm) were assigned to the 9 and 10 positions, respectively (Table 1). gHSQC and gHMBC were carried out on I to obtain the carbon assignments and the 1H-13C correlations for the structural elucidation of its metabolites (Table 2). A full scan mass spectrum of M1 yielded a protonated molecular ion at m/z 395, which was 38 Da less than that of the parent. The product ion spectrum of m/z 395 exhibited an abundant fragment ion at m/z 269, consistent with loss of the (3-fluoro-2-pyridyl)methylamine moiety, and an ion at m/z 377 resulting from loss of elements of water (Figure 3). A minor ion at m/z 127 suggested that the (3-fluoro-2-pyridyl)methylamine moiety remained intact. The molecular weight of this metabolite is consistent with loss of elements of the methylpyrazinone moiety, and proton NMR analysis revealed the absence of resonances attributed to protons attached to C-8 and C-12 suggesting that the metabolite had been

200

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Singh et al.

Figure 1. Representative total ion current chromatograms of the products obtained following incubation of compound I with (A) rat liver and (B) human liver microsomes fortified with GSH.

Figure 2. Product ion spectrum obtained by CID of the protonated molecular ion of compound I at m/z 433.

formed via degradation of the pyrazinone ring. This conclusion was supported by the fragment ion at m/z 237 (Figure 3). Confirmation of the proposed structure of M1 was obtained when the synthetic reference material was shown to have identical LC/MS/MS and NMR characteristics to those of the biological isolate. Upon LC/MS analysis, metabolite M2 exhibited a protonated molecular ion at m/z 738 and a product ion spectrum dominated by a fragment at m/z 431 ([MH307]+), suggesting the presence of an aliphatic or benzylic GSH moiety (9). Further fragmentation (LC/MS3) of the ion at m/z 431 resulted in a product ion spectrum similar to that observed for the parent molecule, suggesting that the integrity of the core structure of the parent molecule was retained following addition of GSH (Figure 4). The C-8 proton in the 1H NMR spectrum appeared at 6.75 ppm and was shifted downfield to 2.33 ppm (∆ +0.13 ppm) as compared to the corresponding signal in the spectrum of the parent molecule (Table 1), while the resonance for the methyl group at position 12 was no longer observed suggesting that this functional group was the site of biotransformation. The expected signal corresponding to -CH2-S protons at the proposed site of GSH attachment would appear between 3 and 4.2 ppm but could not be detected due to interfering peaks from

endogenous material in the sample. However, several GSH-related protons were easily identified (b′, c′, and f′), and signals derived from the protons on the pyridine rings remained essentially unchanged from those in I (Table 1), strongly suggesting that the GSH moiety was attached at the C-12 position. The protonated molecular ion of M3 appeared at the same m/z value (738) as that of the GSH adduct M2 and underwent CID to yield a major fragment ion at m/z 609 (neutral loss of the γ-glutamyl moiety [MH-129]+), a characteristic loss of GSH adducts (9) (Figure 5). Further CID (LC/MS3) of the ion at m/z 609 led to a fragment at m/z 591 (-H2O), which in turn fragmented (LC/MS4) to an ion at m/z 465, which was 32 Da higher than the MH+ of the parent molecule (Figure 5). Examination of the 1H NMR spectrum of M3 indicated four notable differences when compared to that of I (Figure 6). First, the proton attached to C-8 (at 6.62 ppm in I) was absent; second, the C-12 methyl protons at 2.33 ppm were shifted downfield (∆ +0.19 ppm) relative to the parent drug. These changes are consistent with biotransformation having occurred on the pyrazinone moiety. In contrast to I (Figure 6; bottom), significant changes in the chemical shifts of protons at positions 9 and 11 were observed in the spectrum of M3 (Figure 6; top). Thus, the protons

Activation and Rearrangement of a Pyrazinone Ring

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 201

Table 1. 1H NMR Chemical Shift (ppm) Assignments for Compound I and Its Major Rat and Human Liver Microsomal Metabolites

a 400 MHz, CD OD. b 500 MHz, CD OD. c Signal splitting patterns: s ) singlet, d ) doublet, t ) triplet, bs ) broad singlet, bd ) broad 3 3 doublet, dd ) doublet of doublets, bdd ) broad doublet of doublets, dt ) doublet of triplets, t ) triplet, bt ) broad triplet, and m ) multiplet. d Overlapping with impurity signals.

in M3 at position 11 (5.42 ppm) were shifted downfield (∆ +1.18 ppm); the protons at position 9 (4.06, 4.00 ppm), probably representing an AB system (J ) 16.9 Hz), were shifted upfield (∆ -0.72 and -0.78 ppm); and the pyridine protons were shifted only slightly as compared to those in parent molecule (Table 1). These observations are consistent with a structure for M3 involving addition of the GSH moiety to the pyrazinone ring at the position defined as C-8 in Table 1 in such a way that the protons at position C-11 are adjacent to a more highly electronwithdrawing group than in compound I. These observations also suggested that the protons at position C-9 might be adjacent to a less electron-withdrawing group than in I, implying that rearrangement and/or ring opening of the pyrazinone moiety may have occurred. A possible rearrangement was inferred from the chemical shift displacements of the C-9 and C-11 protons, which were considered too large to have resulted from a simple substitution at C-8. The structure of M3 was investigated further by the use of gHMBC and 2D NOESY experiments (Figure 7A,B). The absence of the vinyl proton at position C-8 and the C-H correlation between the carbon at position C-8 and the CH2S group suggested a rearranged product. The correlations between the CH2S protons and the C-8 carbon atoms established that the methylene protons were within three bonds of C-8, thus confirming the point of attachment of the GSH moiety. Also, the correlations between the protons attached to C-9 and the f and h carbonyl groups established that this sequence was intact

(see Figure 7 for description of nomenclature). The remaining correlations, however, did not define the nature of the rearrangement. Finally, the NOESY spectrum of M3 yielded a vital clue in revealing a cross-peak that defined a spacial proximity between the methyl group at C-12 and the methylene protons attached to C-11 (Figure 8). Proximity between these protons was independently confirmed by an NOE difference experiment whereby irradiating the methyl gave an NOE signal from the C-11 protons (at 5.42 ppm). These findings led to the proposed structure for M3 in which the pyrazinone ring rearranged to form a GSH adduct attached to an imidazole ring. The proposed structure of M3 was compatible with the H and 13C chemical shift changes relative to compound I. An upfield shift of the protons on C-9 is to be expected since they are no longer influenced by the deshielding effect of an attached unsaturated ring system. An additional upfield displacement would result from the fact that the protons are not restricted to the same plane of the nearby carbonyl group (f) where the latter exerts its maximum deshielding effect. These observations are reinforced by the close chemical shift correspondence with the protons of the central glycine residue (4.0 ppm) in the synthetic standard, Ala-Gly-Gly. In contrast, the downfield displacement of the C-11 protons results from the fact that these protons are now attached to an unsaturated ring system and are more exposed to the deshielding influence of the f carbonyl (Figure 7B). 1

202


Singh et al.

Table 2. 13C NMR Chemical Shift (ppm) Assignments for Compound I and Its Major Rat and Human Liver Microsomal Metabolites

a

Assignments may be interchanged.

Figure 3. Product ion spectrum obtained by CID of the protonated molecular ion of metabolite M1 at m/z 395.

The full scan LC/MS spectrum of M4 exhibited a protonated molecular ion at m/z 449, consistent with monooxidation of the parent compound. The product ion

spectrum of m/z 449 exhibited in two major ions at m/z 431 (loss of water) and 323 (loss of the (3-fluoro-2pyridyl)methylamine moiety) suggested that oxidation





Figure 6.

1H

NMR spectra of compound I (bottom), GSH (middle), and metabolite M3 (top).

had occurred on the difluoro-2-(2-pyridyl)ethyl)amino]6-methylhydropyrazinone moiety (Figure 9). Upon further fragmentation (LC/MS3) of the ion at m/z 431, major ions were observed at m/z 411 and 391, which arise via sequential loss of two molecules of HF (data not shown).

1 H NMR analysis indicated several significant changes in the spectral characteristics of M4 relative to compound I, including the absence of signals representing the intact methyl group, downfield displacement of the vinyl proton at the C-8 position to 6.76 ppm (∆ +0.14 ppm), and the

204


Singh et al. Table 3. Irreversibly-Bound Radioactivity in Rat Plasma and Liver Following Oral Administration of a 10 and 100 mg/kg Dose of 14C-Labeled Ia dose (mg/kg)

time (h)

10

0 2 6 24 0 2 6 24

100

irreversibly-bound radioactivity (pmol/mg protein) plasma liver BLQb 60 ( 15 76 ( 14 42 ( 8 BLQ 335 ( 150 610 ( 28 225 ( 46

BLQ 110 ( 14 105 ( 8 40 ( 6 BLQ 660 ( 220 880 ( 120 400 ( 46

a Data shown as mean ( SD of n ) 3. b BLQ, below the limit of quantification, which was 5.8 pmol/mg protein (set at two times background).

Figure 7. (A) 1H-13C HMBC correlations for compound I around the pyrazinone ring. (B) 1H-13C HMBC correlations for metabolite M3 around the imidazole ring. Note that the NOE cross-peak between the protons at position 11 and 12 can be accommodated satisfactorily by this structure.

presence of a singlet at 4.33 ppm integrating for two protons (Table 1). These observations suggest that the methyl group in I has undergone conversion to a hydroxymethyl moiety. 13C NMR spectra provided further evidence for hydroxylation of the methyl moiety of the pyrazinone ring with the C-12 carbon appearing downfield at 59.8 ppm (∆ +44.3 ppm) (Table 2). Confirmation of the proposed structure of M4 was obtained when the synthetic standard reference material was shown to have

identical LC/MS/MS and NMR characteristics to those of biological isolate. Covalent Binding in Rats In Vivo. The potential for I to undergo metabolic activation to intermediates that bind irreversibly to protein was assessed in rats in vivo. Following oral administration of 10 and 100 mg/kg doses of [14C] I to rats, significant levels of irreversibly bound radioactivity were detected in plasma and liver samples. At the 10 and 100 mg/kg dose, the maximum level of irreversibly bound radioactivity in liver was 105 and 880 pmol/mg protein, respectively (Table 3). The irreversibly bound radioactivity to rat plasma proteins was 76-610 pmol/mg protein for the 10 and 100 mg/kg doses, respectively. The concentration of I in rat plasma at 2 h was 1.93 and 19.6 µM for the 10 and 100 mg/kg dose, respectively. In a separate experiment using bile duct cannulated rats, less than 2% of the dose was excreted in rat bile and urine as unchanged drug over a

Figure 8. Two-dimensional 1H-1H NOESY spectrum of metabolite M3.



Figure 9. Product ion spectrum obtained by CID of the protonated molecular ion at m/z 449 of metabolite M4.

Figure 10. Proposed mechanisms for the formation of metabolites M2 and M4 from compound I. Table 4. Irreversibly-Bound Radioactivity of 14C-Labeled I in Rat and Human Liver Microsomes Fortified with NADPH species rat

human

time (min)

irreversibly-bound radioactivity (pmol/mg protein)

0 30 60 0 30 60

8.90 ( 3.40 68.9 ( 8.10 82.4 ( 3.60 BLQb 65.0 ( 0.70 83.0 ( 4.30

a Data shown as mean ( SD of n ) 3. b BLQ, below the limit of quantification, which was 5.8 pmol/mg protein (set at two times background). In the absence of NADPH, the maximum level of irreversible binding was BLQ.

period of 72 h, suggesting that biotransformation was the major route of elimination of compound I. Covalent Binding Following Incubations with Rat and Human Liver Microsomal Preparations. Following incubation of 14C-labeled I (2 µM) with rat and human liver microsomes, radioactivity was found to be irreversibly bound to proteins. As compared to the values observed in vivo in rats, the maximum level of irreversibly bound radioactivity in human and rat liver microsomes was relatively low at 85 pmol/mg protein (Table 4).

Discussion The four major metabolites of I that were detected in GSH-fortified rat and human liver microsomes also were detected in the rat bile following iv administration of compound I. Most of the metabolism occurred on the pyrazinone moiety resulting in hydroxylation of the methyl group, addition of the elements of GSH, or degradation of the pyrazinone moiety. The metabolite M1 involves loss of three carbon atoms from the 6-methylpyrazinone moiety and is envisioned to arise from the epoxide intermediate (I-A) shown in Figure 12. Upon hydrolysis, the resulting dihydrodiol intermediate (I-D) undergoes spontaneous ring opening to the R-hydroxyacetone moiety, which readily collapses to yield M1 and pyruvaldehyde (which was not detected in this study). A similar metabolite also was observed as an oxidative degradate of an analogue of compound I (10). The GSH adduct M2 could arise either via a direct displacement of the hydroxyl functional group in metabolite M4 or via addition to the proposed methide-imine intermediate (Figure 10). The former proposed mechanism was shown to be plausible when M2 was allowed to react with GSH in phosphate buffer (pH 7.4) and M4 was formed. The methide-imine intermediate (Me-Im), while not observed in these studies, can be envisioned to

206


Singh et al.

Figure 11. Proposed mechanism for the formation of metabolite M3.

Figure 12. Proposed mechanism for the formation of metabolite M1.

arise either via a direct two electron oxidation of I or by loss of the elements of water from the hydroxy metabolite M4. The structure of the second GSH adduct (M3) was much more complex. The proposed mechanism for the formation of M3 involves initial formation of the same epoxide (I-A) discussed above under the origin of M1 (Figure 11). In this case, however, GSH (rather than H2O) attacks the epoxide (at the less sterically hindered position) yielding a carbinolamide intermediate (I-B), which undergoes ring opening to afford a more stable keto analogue. Following syn-anti isomerization of this substituted amidine derivative (I-C), intramolecular cyclization occurs as depicted in Figure 11 to yield a second carbinolamide that loses the elements of water to generate the stable imidazole derivative M3. The use of 1H and 13C NMR spectroscopy proved to be indispensable for identification of these metabolites,

especially the rearranged GSH adduct M3. Comparison of chemical shifts and correlation spectra of compound I with those of the isolated metabolites was extremely useful in assigning structures, particularly in view of small amounts available and the lack of synthetic standards. Likewise, the use of tandem mass spectrometry facilitated metabolite identification, establishing biotransformation of the substituted pyrazinone moiety as the major route of metabolism of compound I. Consistent with the observation of GSH adducts of I, implicating the formation of chemically reactive intermediates, compound I was shown to covalently modify proteins, both in vitro and in vivo. Over the past few years, a number of reports have implicated reactive intermediates in the etiology of drug-induced liver injury (11). While there are examples of reactive intermediates that apparently do not cause liver toxicity (12), it should be recognized that idiosyncratic drug reactions in humans may be mediated


by the covalent binding of reactive metabolites to proteins (13). Given this potential liability of electrophilic drug metabolites, it would appear desirable to minimize the formation of such intermediates from drug candidates by appropriate structural modification early in the drug discovery process (14). The work reported in this manuscript represents an example of such an effort inasmuch as the information derived from these studies was used to guide the design of thrombin inhibitors with favorable metabolic characteristics (15). Finally, it may be noted that the present paper appears to be the first to document the metabolic activation of a substituted pyrazinone ring system and points to the rich metabolic chemistry associated with nitrogen-containing heterocycles.

Acknowledgment. We thank Drs. P. E. J. Sanderson and Y. Wu for helpful discussions, Dr. M. P. Braun and Ms. T. M. Marks for preparation of the radiolabeled compound I, Dr. X. Xu with covalent protein binding studies, and F. A. Deluna with animal studies. We also thank Dr. K. M. Baillie for helpful discussions and for assistance in preparation of this manuscript.

References (1) Weitz, J. I., and Jirsh, K. (2001) New anticoagulant drugs. Chest 120 (Suppl. S), 95S-107S. (2) Hirsh, J., Dalen, J. E., Anderson, D. R., Poller, L., Bussey, H., Ansell, J., Deykin, D., and Brandt, J. T. (1998) Oral anticoagulants. Mechanism of action, clinical effectiveness, and optimal therapeutic range. Chest 114, 445-469. (3) Vacca, J. P. (2000) New advances in the discovery of thrombin and factor Xa inhibitors. Curr. Opin. Chem. Biol. 4, 394-400. (4) Leadley, R. J., Chi, L., Rebello, S. S., and Gagnon, A. (2000) Contribution of in vivo models of thrombosis to the discovery and development of novel antithrombotic agents. Pharmacol. Toxicol. Methods 43, 101-111. (5) Bates, S. M., and Weitz, J. I. (2000) The mechanism of action of thrombin inhibitors. J. Invasive Cardiol. 12F, 27-32.

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 207 (6) Sanderson, P. E. J. (1998) The Design of Orally Active Pyridinone and Pyrazinone Acetamide Thrombin Inhibitors, Presented at the 215th ACS National Meeting, Dallas, TX. (7) Ding, X., Spink, D. C., Bhama, J. K., Sheng, J. J., Vaz, A. D., and Coon, M. J. (1996) Metabolic activation of 2,6-dichlorobenzonitrile, an olfactory-specific toxicant, by rat, rabbit, and human cytochromes P450. Mol. Pharmacol. 49, 1113-1121. (8) Jochheim., C. M., Davis, M. R., Baillie, K. M., Ehlhardt, W. J., and Baillie, T. A. (2002) Glutathione-dependent metabolism of the antitumor agent sulofenur. Evidence for the formation of p-chlorophenyl isocyanate as a reactive intermediate. Chem. Res. Toxicol. 15, 240-248. (9) Baillie, T. A., and Davis, M. R. (1993) Mass spectrometry in the analysis of glutathione conjugates. Biol. Mass Spectrom. 22, 319-325. (10) Wu, Y., Chen, X., Gier, L., Almarsson, O., Ostovic, D., and Loper, A. E. (1999) Identification of oxidative degradates of the thrombin inhibitor, 3-(2-phenethylamino)-6-methyl-1-2-(amino-6-methyl-5methylenecarboxamidomethyl pyridine) pyrazinone using liquid chromatography/tandem mass spectrometry. J. Pharm. Biomed. Anal. 20, 471-485. (11) Chitturi, S., and George, J. (2002) Hepatotoxicity of commonly used drugs: nonsteroidal antiinflammatory drugs, antihypertensives, antidiabetic agents, anticonvulsants, lipid-lowering agents, psychotropic drugs. Semin. Liver Dis. 22, 169-84. (12) Nelson, S. D., and Pearson, P. G. (1990) Covalent and noncovalent interactions in acute lethal cell injury caused by chemicals. Annu. Rev. Pharmacol. Toxicol. 30, 169-195. (13) Uetrecht, J. P. (1999) New concepts in immunology relevant to idiosyncratic drug reactions: the “danger hypothesis” and innate immune system. Chem. Res. Toxicol. 12, 387-395. (14) Baillie, T. A., and Kassahun, K. (2001) Biological reactive intermediates in drug discovery and development: a perspective from the pharmaceutical industry. Adv. Exp. Med. Biol. 500, 45-51. (15) Burgey, C. S., Robinson, K. A., Lyle, T. A., Sanderson, P. E. J., Lewis, S. D., Lucas, B. J., Krueger, J. A., Singh, R., Miller-Stein, C., White, R. B., Wong, B., Lyle, E. A., Williams, P. D., Coburn, C. A., Dorsey, B. D., Stranieri, M. T., Holahan, M. A., Sitko, G. R., Cook , J. J., McMasters, D. R., McDonough, C. M., Sanders, W. M., Wallace, A. A., Clayton, F. C., Bohn, D., Leonard, Y. M., Detwiler, T. J., Jr., Lynch, J. J., Jr., Yan, Y., Chen, Z., Kuo, L., Gardell, S. J., Shafer, J. A., and Vacca, J. P. Metabolism directed optimization of 3-aminopyrazinone acetamide thrombin inhibitors. Development of an orally bioavailable series containing P1 and P3 pyridines. J. Med. Chem., in press.

TX025635L

Metabolic Activation of a Pyrazinone-Containing Thrombin Inhibitor

Recommend Documents