Metabolism and Bioactivation of 3-Methylindole by Human Liver

Zhengyin Yan,*,† LaHoma M. Easterwood,‡ Noureddine Maher,† Rhoda ... and Department of Pharmacology and Toxicology, UniVersity of Utah, Salt Lak...
0 downloads 0 Views 170KB Size
140

Chem. Res. Toxicol. 2007, 20, 140-148

Metabolism and Bioactivation of 3-Methylindole by Human Liver Microsomes Zhengyin Yan,*,† LaHoma M. Easterwood,‡ Noureddine Maher,† Rhoda Torres,† Norman Huebert,† and Garold S. Yost‡ DiVision of Drug DiscoVery, Johnson & Johnson Pharmaceutical Research & DeVelopment, LLC Spring House, PennsylVania 19477, and Department of Pharmacology and Toxicology, UniVersity of Utah, Salt Lake City, Utah 84112 ReceiVed September 21, 2006

Metabolism and bioactivation of 3-methylindole (3MI) were investigated in human liver microsomes. The metabolism of two deuterium-labeled analogues of 3MI permitted a relatively unambiguous identification of multiple metabolites and glutathione (GSH) adducts of reactive intermediates. A total of eight oxidized metabolites were detected, five of which were assigned as previously identified 3-methyloxindole, 3-hydroxy-3-methylindolenine, 3-hydroxy-3-methyloxindole, 5-hydroxy-3-methylindole, and 6-hydroxy-3-methylindole. Among the three new metabolites, one was either 4- or 7-OH-3methylindole, and the other two were derived from additional oxidation on the phenyl ring of 3-methyloxindole. When GSH was added to the microsomal incubations, seven conjugates that had molecular ions corresponding to the incorporation of GSH and an atom of oxygen at m/z 453 (group I) were produced, and two additional conjugates had molecular ions at m/z 437 that corresponded to the incorporation of GSH with no additional oxygen (group II). Two conjugates in group I (m/z 453) were apparently derived by GSH addition to the 5,6-epoxide metabolite of 3-methyloxindole. These two GSH adducts were tentatively identified as 5-(glutathione-S-yl)-3-methyloxindole and 6-(glutathione-S-yl)-3methyloxindole. The most abundant conjugate in group I was identified as 3-(glutathione-S-yl)-3methyloxindole, which substantiated the presence of the putative 2,3-epoxy-3-methylindole intermediate. The remaining four adducts in group I were likely formed by conjugation of GSH at different positions of the phenyl ring, possibly via oxidation of 5-hydroxy-3-methylindole and 6-hydroxy-3-methylindole to two very interesting new electrophilic benzoquinone imine intermediates. For the group II conjugates (m/z 437), two isomers were identified as 2-(glutathione-S-yl)-3-methylindole and 3-(glutathione-S-ylmethyl)-indole. The former adduct was primarily derived from the 2,3-epoxide intermediate by thiol conjugation followed by dehydration. The latter adduct was consistent with our previously published work on the dehydrogenation of 3MI. In those studies, we showed that the reactive intermediate, 3-methylenenindolenine, was formed by hydrogen abstraction at the methyl group and was trapped with GSH. The putative dehydrogenation bioactivation mechanism is also substantiated by the finding that CYP2E1 selectively generated 2-(glutathione-S-yl)-3-methylindole but did not produce 3-(glutathioneS-yl-methyl)-indole. In summary, the results not only confirmed the formation of 2,3-epoxide-3methylindole in human liver microsomes but also suggested that the phenolic metabolites of 3-methylindole were dehydrogenated to previously uncharacterized reactive intermediates. Introduction

Scheme 1. Bioactivation of 3MI via Dehydrogenation and Epoxidation and Thiol Trapping of Reactive Metabolites

3-Methylindole (3MI), a degradation product of tryptophan formed in the rumen of cattle and goats and in the large intestine of humans, is well-known as a highly selective pulmonary toxicant for ruminants (1). It is generally accepted that the pulmonary toxicity of 3MI is at least partially associated with bioactivation mediated by cytochrome P450s, resulting in reactive electrophilic intermediates that potentially modify cellular proteins (2) and nucleic acids (3). This hypothesis is supported by multiple studies showing that both covalent binding and toxicity of 3MI were correlated with P4501 activity (4-6). As shown in Scheme 1, two metabolic pathways have been * To whom correspondence should be addressed. Tel: 215-628-5036. Fax: 215-540-4878. E-mail: [email protected]. † Johnson & Johnson Pharmaceutical Research & Development. ‡ University of Utah. 1 P450, cytochrome P450; HLM, human liver microsomes; GSX, Υ-glutamyl-cystein-glycin-13C2-15N

proposed for the bioactivation of 3MI (7, 8). The predominant route in goat lung microsomes is hydrogen abstraction at the methyl group followed by a one-electron oxidation, producing a highly reactive electrophilic intermediate, 3-methylenenindo-

10.1021/tx060239e CCC: $37.00 © 2007 American Chemical Society Published on Web 12/22/2006

Metabolism and BioactiVation of 3-Methylindole

lenine. The second bioactivation pathway is formation of 2,3epoxide-3-methylindole. The fate of the epoxide may include alkylation reactions with cellular proteins and rearrangement to form the stable oxindole. Alternatively, the epoxide may rearrange to form 3-hydroxy-3-methylindolenine, another reactive intermediate that may participate in alkylation reactions. Detection and characterization of these reactive intermediates would potentially provide useful information for our understanding of 3MI-induced toxicity. Direct detection and characterization of reactive metabolites is difficult, so a common strategy to identify them is to use a thiol-trapping agent such as glutathione (GSH) or N-acetylcysteine (NAC) to capture reactive intermediates; resulting stable adducts can be characterized. Thiol adducts derived from 3MI have been characterized by LC/MS techniques (9). In the presence of N-acetylcysteine, from goat lung microsomal incubations, two isomeric adducts were detected with conjugation sites at the exocyclic methylene group (G1) and the C-2 position of the indole ring (G2) (Scheme 1). However, LC/MS techniques failed to detect potential thiol adducts such as G3 and G4 (Scheme 1), which would have been produced from the addition of N-acetylcysteine to the 3 or 2 positions of 2,3epoxide-3-methylindole, respectively. The G4 adduct presumably rapidly lost water to produce G2 (9). However, thioglycolic acid was used to trap 3-hydroxy-3-methylindolenine to form a lactone through a cyclocondensation mechanism before the thioether had a chance to dehydrate (9). The formation of 3-hydroxy-3-methylindolenine is not necessarily dependent on a precursor epoxide and in fact may proceed by an additionrearrangement mechanism that has been proposed as an operative mechanism in arene oxidation catalyzed by P450 enzymes (10). Although indirect evidence has previously been presented to support the formation of 2,3-epoxy-3-methylindole (11), detection and characterization of the thioether adducts, particularly the G3 adduct, would substantiate the existence of this reactive intermediate. Human exposures to 3MI can also occur via ingestion of dietary tryptophan, followed by anaerobic fermentation in the colon (12), or from cigarette smoke (13). Human susceptibility to the toxicant has not been fully investigated, although previous studies (14) have shown that 3MI is metabolically activated by human cytochrome P450s in the lung. To more thoroughly characterize all oxidative metabolites and reactive intermediates, we profiled the metabolism of 3MI and its deuterated analogues in human liver microsomal incubations. To investigate potential new metabolic pathways leading to the bioactivation of 3MI, we utilized stable isotope trapping in combination with tandem mass spectrometry to detect and characterize glutathione conjugates derived from 3MI in human liver microsomsal incubations. In addition to the thiol conjugates that were previously identified, several new GSH adducts were detected. The results not only provided direct evidence for the production of the 2,3-epoxide but also suggested that additional metabolic pathways may bioactivate 3MI.

Materials and Methods Materials. Reagents and solvents used in the current study were of the highest possible grade available. The following chemicals were purchased from Sigma-Aldrich (St. Louis, MO): 3-methylindole, 3-methyloxindole, 1-aminobenzotriazole (ABT), GSH, β-nicotinamide adenine dinucleotide phosphate (NADP+), glucose-6phosphate, and glucose-6-phosphate dehydrogenase. Stable isotopelabeled glutathione [GSX, Υ-glutamyl-cystein-glycin-13C2-15N] was obtained from Cambridge Isotope Laboratories (Andover, MA), and

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 141 isotopic purity was 90% estimated by the supplier using nuclear magnetic resonance (NMR). Pooled human liver microsomes and Supersomes containing cDNA-expressed P450 enzymes were obtained from BD Gentest Corp. (Woburn, MA). [2-2H]-3-Methyindole (3MI-d1-[2H]) and 3-[2H3-methyl]indole (3MI-d3) were synthesized by following published methods (11). Microsomal Incubations. All incubations were performed at 37 °C in a water bath. 3MI or its isotopic analogs were mixed with human liver microsomal proteins in 50 mM potassium phosphate buffer (pH 7.4). After a 5-min preincubation at 37 °C, reactions were initiated by the addition of an NADPH generating system to give a final volume of 1000 µL. The final reaction mixture contained 100 µM of 3MI, [2-2H]-3-methyindole (3MI-d1-[2H]), or 3-[2H3methyl]indole (3MI-d3), 1 mg/mL microsomal proteins, 1.3 mM NADP+, 3.3 mM glucose-6-phosphate, 0.4 U/mL glucose-6phosphate dehydrogenase, and 3.3 mM magnesium chloride. After a 60-min incubation, reactions were terminated by the addition of 150 µL of trichloroacetic acid (10%). To trap reactive metabolites, separate incubations were performed in the presence of either GSH or a mixture of GSH and GSX (1: 1). The trapping agents were supplemented to the reaction mixtures prior to the addition of an NADPH generating system. The final concentration of GSH and GSX was 1 mM. Incubation samples were centrifuged at 10 000g for 15 min at 4 °C to pellet the precipitated protein, and supernatants were cleaned by C18 solidphase extraction by following the manufacturer’s protocol (Waters Corp., Milford, MA), prior to LC-MS/MS analyses. Incubations with individual recombinant P450 enzymes were performed similarly, except that human liver microsomes were substituted by Supersomes. LC-MS/MS Analyses. MS analyses were performed on a Micromass (Manchester, United Kingdom) Quattro Micro triple quadrupole mass spectrometer interfaced to an Agilent 1100 highperformance liquid chromatography (HPLC) system (Agilent Technologies, Palo Alto, CA). The ESI ion source was operated in the positive ion mode, and experimental parameters were set as follows: capillary voltage 3.2 kV, source temperature 120 °C, desolvation temperature 300 °C, sample cone voltage 20 V. Data were processed using the Masslynx version 4.0 software from Micromass. For rapid screening of GSH adducts, 10-µL aliquots were injected onto an Agilent Zorbax SB C18 column (2.1 × 50 mm). The HPLC mobile phase started with 95% aqueous (0.5% acetic acid) solvent and 5% acetonitrile at a flow rate of 0.3 mL/min and ended at 95% acetonitrile/water over a 5-min linear gradient. The Quattro Micro triple quadrupole mass spectrometer was operated in the neutral loss mode scanning over the range m/z 330-500 in 2 s. For complete profiling of oxidation metabolites and GSH conjugates, an Agilent Zorbax SB C18 column (2.1 × 150 mm) was used for chromatographic separations. The HPLC profile consisted of 95% water (0.5% acetic acid), and the metabolites were eluted using a single linear gradient of 95% water to 85% acetonitrile over 25 min at a flow rate of 0.3 mL/min. At 25 min, the column was flushed with 85% acetonitrile for 2 min before re-equilibration at initial conditions. MS and MS/MS data were collected on the Quattro Micro triple quadrupole mass spectrometer that was operated in the positive mode.

Results Profiling Oxidation Metabolites. Incubations were performed with 3MI to profile oxidation metabolites in human liver microsomes, and resulting samples were analyzed using LCMS techniques. As shown in Figure 1A, five metabolites were detected at m/z 148 (corresponding to the molecular mass of 3MI, m/z 131, plus oxygen, plus a proton) and were designated as M1 (11.2 min), M2 (16.6 min), M3 (16.9 min), M4 (18.2 min), and M5 (19.8 min). It appeared that M1 and M3 are more abundant than M2, M4, and M5, as estimated by their peak areas. Tandem MS spectra of those metabolites were highly

142 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Yan et al.

Figure 1. Ion chromatograms from LC/MS detection of oxidation metabolites of 3MI (A, m/z 148), 3MI-d3 (B, m/z 151; C, m/z 150), and 3MId1-[2H] (D, m/z 148; E, m/z 149) formed from incubations with human liver microsomes.

similar (data not shown) and not helpful for structural elucidation. To further identify these metabolites (M1-5), separate incubations were carried out with deuterated analogues 3MId3 and 3MI-d1-[2H]. In the incubation with 3MI-d3, all five metabolites (M1-5) were detected at the expected m/z 151 (Figure 1B), whereas none of these metabolites showed significant response at m/z 150 (Figure 1C). Retention of all three deuterium atoms clearly indicated that M1-5 were not derived from oxidation of the methyl group. In the incubation of 3MI-d1-[2H], M3 was detected at m/z 148 as the most intense peak (Figure 1D). A loss of deuterium at the C-2 position suggested that M3 was 3-methyloxindole. This was confirmed by coelution with the authentic compound under the same LCMS conditions (data not shown). In the same incubation with 3MI-d1-[2H], the M1, M2, M4, and M5 metabolites retained the C-2 deuterium and were detected at m/z 149 (Figure 1E), suggesting that those metabolites were derived from oxidation of either the phenyl ring or the methyl group, but the 3MI-d3 experiment ruled out methyl oxidation. The M3 metabolite was also detected at m/z 149, but its intensity was significantly lower than that of the other metabolites (Figure 1E). Partial retention of the C-2 deuterium

was consistent with an “NIH shift” mechanism of deuterium migration from the C-2 to the C-3 position via the epoxidation mechanism. Our previous work showed that approximately 81% of the deuterium atoms were retained upon epoxide ring-opening and hydride shift (11). M1 was first eluted from the C18 column as a very broad peak, which is consistent with the anticipated chemistry of 3-hydroxy-3-methylindolenine. 3-Hydroxy-3-methylindolenine is a relatively abundant metabolite from incubations with pig liver microsomes, and it exhibits higher hydrophilicity than other oxidation metabolites (15). Additionally, 3-hydroxy-3-methylindolenine probably exists in equilibrium between its neutral and protonated forms, which would be expected to produce considerably peak broadening. Although we could not definitely determine the structures of M2, M4, and M5 because of a lack of reference compounds, it is certain that these metabolites were derived from oxidation of the phenyl ring. We speculate that these three metabolites are 5-OH-3MI, 6-OH-3MI, and 4- or 7-OH-3-methylindole, since both 5-OH3MI and 6-OH-3MI have been previously identified from incubations with pig liver microsomes (15). As shown in Figure 2A, three metabolites were detected with a molecular ion of m/z 164: M6 (5.5 min), M7 (10.3 min), and

Metabolism and BioactiVation of 3-Methylindole

Figure 2. Ion chromatograms from LC/MS detection of oxidation metabolites derived from 3MI (A, m/z 164), 3-methyloxindole (B, m/z 148), and 3MI-d3 (C, m/z 167) from incubations with human liver microsomes.

M8 (11.6 min). The same three metabolites (M6-8) with identical retention times and molecular ions of m/z 164 were produced when 3-methyloxindole was employed as the starting substrate (Figure 2B). The results suggested that metabolites M6-8 were derived from oxidation of 3-methyloxindole. In an incubation with 3MI-d3, M6-8 were detected at m/z 167 (Figure 2C), which confirmed that the methyl group remained intact. We assumed that 3-hydroxy-3-methyloxindole was one of these three metabolites, because it has previously been identified from incubations of 3MI with pig liver microsomes (15). We concluded that the other two metabolites were derived from additional oxidation of the phenyl ring of 3-methyloxindole. Although we presumed that hydroxylation occurred at the C-5 and C-6 carbons, other positions are also possible, and because we lacked reference compounds, we could not precisely identify the other two metabolites. In summary, a total of eight oxidation metabolites were detected in human liver microsomes. The structures of these metabolites are tentatively proposed and are shown in Scheme 2. Stable Isotope Trapping and MS Detection of GSH Adducts. Stable isotope trapping has recently been described as a highly sensitive and efficient approach for the initial screening of potential reactive metabolites formed in microsomal incubations (16). In this method, natural glutathione (GSH) and stable isotope-labeled glutathione (GSX) were mixed at an equal molar ratio and were added to microsomal incubations to trap reactive metabolites. Because all isotope-labeling sites are located at the glycine residue (two carbons and one nitrogen), both natural (M-SG) and isotope-labeled conjugates (M-SGx) undergo the same neutral loss of pyroglutamate (129 Da) via a collision-induced fragmentation. As a result, two molecular ions appeared in the neutral loss mode as an isotopic doublet that

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 143

Figure 3. Stable isotope trapping and LC-MS/MS analyses of GSH adducts derived from incubations of 3MI with human liver microsomes. A, ion chromatogram from LC-MS/MS scan for neutral loss of 129 Da; B, the MS/MS spectrum of group I adducts obtained by neutral loss scan; C, the MS/MS spectrum of group II adducts obtained by neutral loss scan.

Scheme 2. Oxidation Metabolites of 3MI That Are Formed with Human Liver Microsomesa

a Metabolites M2 and M4-8 were not assigned because of the lack of ref compounds.

differs in mass by 3 Da, which can be readily used as an MS signature for rapid detection of GSH adducts (16). To profile all potential reactive intermediates, an incubation of 3MI was performed in the presence of GSH and GSX, and the resulting sample was cleaned by SPE and was subjected to LC-neutral loss MS/MS analyses. Figure 3A is the total ion chromatogram from neutral loss MS/MS analysis of 3MI incubation sample. Two groups of peaks displayed isotopic doublets of the expected mass difference of 3 Da. Group I exhibited a doublet at m/z 453 and m/z 456 (Figure 3B), whereas group II displayed a different isotope doublet at m/z 437 and

144 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Yan et al.

Figure 4. Ion chromatograms from LC/MS detection of group I GSH conjugates derived from 3MI (A, m/z 453), 3-methyloxindole (B, m/z 453), 3MI-d3 (C, m/z 456), and 3MI-d1-[2H] (D, m/z 453; E, m/z 454) incubations with human liver microsomes. The peak labeled with an * was a contaminant.

m/z 440 (Figure 3C). These molecular ions indicated that group I adducts had incorporated 3MI, GSH, and an atom of oxygen, whereas group II adducts had incorporated 3MI and GSH with no additional oxygen. The isotopic patterns were not observed when either 3MI or the NADPH regenerating solution was omitted from the microsomal incubations. Additionally, formation of both group I and II adducts was abated by preincubation of human liver microsomes (HLM) with 30 µM ABT for 20 min in the presence of NADPH generating system (data not shown). These preliminary results indicated that two groups of GSH adducts were derived from 3MI via P450-mediated bioactivation. Characterization of Group I Adducts. Tandem mass spectrometry was utilized to analyze group I adducts using a long HPLC gradient to completely resolve individual isomeric conjugates. As shown in Figure 4A, seven components were detected at m/z 453 and were designated as A (12.2 min), B (12.3 min), C (13.1 min), D (13.3 min), E (13.7 min), F (14.1 min), and G (14.8 min). Tandem MS spectra of these adducts (A-G) all showed product ions at m/z 324 (data not shown) resulting from the characteristic neutral loss of 129 Da. Additionally, tandem MS analyses of stable isotope-labeled GSH adducts (A-G) at

m/z 456 revealed expected product ions at m/z 327 (data not shown), which additionally confirmed that A-G are GSH adducts. However, tandem MS spectra of those adducts were highly similar and insufficient for structural elucidation. To examine the possibility that some conjugates in group I were derived from bioactivation of 3-methyloxindole, an incubation was performed with 3-methyloxindole in the presence of GSH. As shown in Figure 4B, LC-MS/MS analyses detected two isomeric GSH conjugates, which exhibited retention times identical to that of conjugates A and B derived from 3MI. The results clearly suggested that both A and B were derived from subsequent oxidation of the major metabolite, 3-methyloxindole. Therefore, a putative bioactivation pathway is shown in Scheme 3. Epoxidation of the phenyl ring of 3-methyloxindole, followed by the addition of GSH and a subsequent loss of water, produces two isomeric conjugates. Although it is presumed that epoxidation occurs between the C-5 and C-6 carbons, additional studies are required to determine the exact conjugation sites. The results from this bioactivation study are also consistent with our previous observation that P450-mediated metabolites were produced via oxidation of the phenyl ring in the incubation of 3-methyloxindole.

Metabolism and BioactiVation of 3-Methylindole Scheme 3. Bioactivation Pathways of 3-Methyloxindolea

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 145 Scheme 4. Bioactivation Pathways of 5-Hydroxy-3-methylindole and 6-Hydroxy-3-methylindole in Human Liver Microsomesa

a Metabolites M6 and M8 and conjugates A and B were not assigned because of the lack of ref compounds.

To determine the conjugation sites of the other conjugates (C-G), 3MI-d3 was incubated with HLM in the presence of GSH, and the resulting sample was analyzed for isotopic conjugates. As shown in Figure 4C, all group I conjugates (AG) were detected at m/z 456. However, none of those conjugates appeared at m/z 455 (data not shown). These results suggested that the methyl group remained intact for all group I GSH adducts (A-G). Subsequently, a similar incubation was performed with 3MId1-[2H] in the presence of GSH. LC-MS analyses of the incubation sample of 3MI-d1-[2H] detected the conjugate E at m/z 453 as the most intense peak (Figure 4D). It appears that conjugate E is 3-(glutathione-S-yl)-3-methyloxindole, which is consistent with the epoxidation pathway (Scheme 1, G3). In this case, the epoxide was presumably opened by GSH addition at the C-3 position to produce the conjugate and resulted in a loss of the deuterium at the C-2 position. The ring-opened alcohol may then lose water to form an imine and subsequently be oxidized by cytochrome P450s to the lactone (oxindole). A similar imine oxidation to a lactone has recently been shown for the P450-mediated oxidation of nicotine imine to continine (17). Additionally, in the incubation of 3MI-d1-[2H], seven conjugates were detected at m/z 454 (Figure 4E), which suggested that deuterium at the C-2 position remained for C, D, F, and G. From these results, we concluded that the GSH moiety is attached to the phenyl ring for conjugates C, D, F, and G. As shown in Figure 4E, conjugate E derived from 3MI-d1[2H] also appeared at m/z 454. To clarify our confusion in Figure 4D and 4E, we calculated the ratio of conjugate E peak area at m/z 454 relative to the peak area at m/z 453 and found that the ratio is 26.8%. Also, using the formula of conjugate E (C19H29O7N4S), we calculated the natural isotope abundance (M + 1) relative to M and found that the value is 24.0%. Therefore, we concluded that the response of conjugate E at m/z 454 (Figure E) resulted from the natural isotopic effect and not from the retention of the deuterium at the C-2 position. Several experiments led to postulated structures for conjugates C, D, F, and G. First, they are not formed by oxidation of oxindole, because only conjugates A and B were derived from 3-methyloxindole. Second, they could not have been formed through oxidation of the methyl group, because none of the methyl hydrogens were lost from incubations with 3MI-d3. Third, they could not have been derived, even in part, from the 2,3-epoxide, because none of the deuterium label was lost from

a Conjugates C, D, F, and G were not assigned because of the lack of ref compounds.

incubations with 3MI-d1-[2H]. Even though the NIH shift mechanism of deuterium migration from C-2 to C-3 was efficient (approximately 80%), at least a small portion of the label should have been lost, and we observed no ion intensities at all at m/z 453. Thus, these conjugates were very unlikely to be substituted oxindoles. Therefore, it seemed possible that conjugates C, D, F, and G could be structural isomers that were formed by addition of GSH to two putative quinone imines. These electrophilic intermediates could have been produced by P450-mediated dehydrogenation of 5-hydroxy-3-methylindole and 6-hydroxy-3-methylindole (Scheme 4). Characterization of Group II Adducts. In HLM incubations of 3MI with GSH, two adducts in group II were detected at m/z 437 and were designated as H (14.2 min) and I (15.4 min) (Figure 5A). This molecular mass indicated that the adducts had been formed without incorporation of oxygen or with oxygen addition followed by dehydration. Tandem MS spectra of H and I showed characteristic product ions at m/z 308 resulting from a neutral loss of 129 Da (data not shown). In the incubations with 3MI-d1-[2H], conjugate I was detected at m/z 437, loss of the deuterium, as the predominant peak (Figure 5B), suggesting that I may be 3-methyl-2-(glutathione-S-yl)indole (G2, Scheme 1). In the same incubation, H showed a stronger response than that of I at m/z 438 (Figure 5C), indicating that this conjugate may be 3-(glutathione-S-ylmethyl)-indole (G1, Scheme 1). As shown in Scheme 1, conjugation of GSH to 3-methylenenindolenine would be expected to form a thioether bond at either the exocyclic methylene or the imine C-2 carbon. The latter reaction would produce 3-methyl-2-(glutathione-S-yl)indole. 3-Methyl-2-(glutathione-S-yl)-indole can also be derived from 2,3-epoxy-3MI via thiol conjugation at the C-2 carbon, followed by dehydration (Scheme 1). If 3-methyl-2-(glutathioneS-yl)-indole was formed via the dehydrogenation pathway, one would expect that this conjugate would be detected at m/z 439 in the incubation with 3MI-d3 because of a loss of deuterium at the methyl group. In contrast, if 3-methyl-2-(glutathione-Syl)-indole was derived from the epoxide, it should appear at m/z 440 in the incubation with 3MI-d3. As shown in Figure 5D, I was detected at m/z 440 as the major conjugate but did show a minor response at m/z 439 (Figure 5E). These results clearly suggested that GSH adduction to C-2 of the epoxide is the major pathway leading to formation of 3-methyl-2-(glu-

146 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Yan et al.

Figure 5. Ion chromatograms from LC/MS detection of group II GSH conjugates.

tathione-S-yl)-indole. Conversely, H was detected in the same incubations with 3MI-d3 at m/z 439 as the major conjugate and exhibited very weak response at m/z 440, confirming that H was formed through GSH adduction to the methylene of 3-methyleneindolenine. Formation of GSH Adducts by cDNA-Expressed P450s. The formation of the GSH adducts was investigated in these recombinant individual human P450 enzymes: CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4. Conjugates A, B, C, E, H, and I were detected in GSH-fortified incubations with different P450 enzymes. However, D, F, and G were not detected in any of those incubations, probably because of their low abundance. An alternative explanation for the absence of those three conjugates is that formation of oxidized metabolites and subsequent bioactivation of those metabolites require different P450 enzymes that were not evaluated. As seen in Table 1, 3MI is a nonselective substrate for human P450 enzymes, and individual enzymes showed different activities in the formation of different adducts. For group I adducts, CYP3A4 was the most active enzyme catalyzing formation of C and E, which was followed by CYP2C19, CYP2B6, and CYP1A1. For conjugates A and B, CYP1A2 was the most potent enzyme, which was followed by CYP1A1, CYP3A4, and CYP2C19. For group II conjugates, CYP2A6, CYP1A2, and

Table 1. Major GSH Adducts Formed by Individual P450 Enzymes group I conjugates (%)a

group II conjugates (%)a

P450

A & Bb

C

E

H

I

1A1 1A2 1B1 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4

42 100 6 n.d.c 25 n.d n.d. 30 15