Use of a Trapping Agent for Simultaneous Capturing and High

May 4, 2007 - In the present study, a bifunctional trapping agent (γGSK, ..... 4′-hydroxyme- phenytoin. The MRM transitions monitored were m/z 233 ...
0 downloads 0 Views 258KB Size
Anal. Chem. 2007, 79, 4206-4214

Use of a Trapping Agent for Simultaneous Capturing and High-Throughput Screening of Both “Soft” and “Hard” Reactive Metabolites Zhengyin Yan,* Noureddine Maher, Rhoda Torres, and Norman Huebert

Division of Drug Discovery, Johnson & Johnson Pharmaceutical Research & Development, LLC, Spring House, Pennsylvania 19477

Glutathione (GSH) has been widely used for in vitro trapping and subsequently detecting reactive metabolites using liquid chromatography-mass spectrometry. A major drawback of GSH is its low trapping efficiency for “hard” reactive metabolites such as reactive aldehydes. In the present study, a bifunctional trapping agent (γGSK, γ-glutamylcysteinlysine) is investigated as an alternative of GSH for simultaneous trapping both “hard” and “soft” reactive metabolites. In microsomal incubations, soft and hard reactive metabolites are captured by conjugation to the free thiol and the amine group of γGSK, respectively, resulting in formation of stable peptide adducts. Similar to GSH conjugates, all γGSK adducts derived from both soft and hard reactive metabolites contain a γ-glutamyl moiety and, thus, undergo a neutral loss of 129 Da under collision-induced dissociation. As a result, an NL MS/ MS scan can be utilized as a generic method for rapid detecting of both hard or soft reactive metabolites. As demonstrated by a number of model compounds, this approach, in combination with the isotope trapping technique, is reliable, sensitive, and efficient and can be potentially utilized as a high-throughput method for screening and rapid identification of both soft and hard reactive metabolites. In comparison with other methods, this approach is highly efficient and suitable in drug discovery for screening a wide variety of compounds for different reactive metabolites. Xenobiotics such as drugs and chemical toxicants are metabolized by a variety of oxidative enzymes, such as cytochrome P450s (CYPs) and flavin-containing monooxygenases, expressed predominantly in liver, to stable and more polar metabolites that can be readily eliminated from body. Thus, metabolism is generally considered a detoxicification process. However, it has been recognized that, for some drugs and other xenobiotics, biotransformation can also lead to formation of chemically reactive metabolites that can covalently modify endogenous macromolecules such as proteins and nucleic acids (DNA). Covalent modifications of DNA and cellular proteins resulting from bioactivation are thought to play an important role in drug-induced toxicities such as genotoxicity and idiosyncratic toxicity.1 * To whom correspondence should be addressed. Tel.: (215)-628-5036. Fax: (215)-540-4878. E-mail: [email protected].

4206 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

Generally, genotoxicity can be effectively assessed by using a combination of in vitro assays (i.e., Ames tests) and animal toxicity models. In contrast, drug-induced idiosyncratic toxicity is very difficult to predict even after clinic studies. Being characterized by a high degree of individual susceptibility and lacking simple dose responses, idiosyncratic drug reactions do not significantly occur until a large population has been studied after drugs have been on the market. Severe idiosyncratic drug reactions can be life-threatening and thus lead to restricted use and even withdrawal of drugs from the market.2 Although drug-induced idiosyncratic toxicities have been a serious concern in drug discovery and development, our current understanding of biochemical mechanisms of drug-induced toxicity is still very limited. Current theories, which are largely based on conceptual analyses, all postulate that reactive metabolites derived from metabolism of drugs are responsible for observed drug-induced toxicities. A recent analysis has revealed that, for 21 marketed drugs examined, 5 out of 6 drugs that were withdrawn from the market and 8 out of 15 drugs that had black box warning showed evidence for the formation of reactive metabolites,3 which, along with many other studies, suggests a potential association of reactive intermediates with drug-induced toxicities. There have been several recently published reviews on mechanistic aspects of ADRs.3-5 All of these mechanisms propose that events such as formation of reactive intermediates, inadequate detoxification of reactive intermediates, covalent binding to macromolecules, and events leading to disruption of cellular signaling processes are important in generating IDRs. Because of potential safety concerns, a drug candidate undergoing bioactivation is less favorable for further development. Although it is still controversial, screening of reactive metabolites in drug discovery has increasingly become an integral part of the ADMET-guided lead optimization process in many major pharmaceutical companies.6,7 Therefore, a sensitive and efficient (1) Park, B. K.; Kitteringham, N. R.; Maggs, J. L.; Pirmohamed, M.; Williams, D. P. Annu. Rev. Pharm. Toxicol. 2005, 45, 177-202. (2) Waring, Jeffrey F.; Anderson, Mark G. Cur. Opin. Drug Discovery Dev. 2005, 8, 59-65. (3) Walgren, J. L.; Mitchell, M. D.; Thompson, D. C. Curr. Rev. Toxicol. 2005, 35, 325-361. (4) Park, B. K.; Kitteringham, N. R.; Maggs, J. L.; Pirmohamed, M.; Williams, D. P. Annu. Rev. Pharmacol. Toxicol. 2005, 45, 177-202. (5) Uetrecht, J. Toxicology 2005, 209, 113-118. (6) Evans, D. C.; Baillie, T. A. Curr. Opin. Drug Discovery Dev. 2005, 8, 4450. (7) Doss, George A.; Baillie, T. A. Drug Metab. Rev. 2006, 38, 641-649. 10.1021/ac0701029 CCC: $37.00

© 2007 American Chemical Society Published on Web 05/04/2007

method for detecting reactive metabolites is highly desired for in vitro evaluation of drug candidates, since such information can be very helpful for medicinal chemists to optimize lead compounds at an early stage of drug discovery.6,7 Based on the HSAB concept,8 reactive intermediates can be classified into “soft” and “hard” reactive metabolites. For soft reactive metabolites, functional groups are often characterized by a large radius, and they are easy to polarize, whereas functional groups of hard reactive metabolites usually have a low radius and are weakly polarizable. Soft reactive metabolites comprise a majority of electrophilic metabolites including quinones, quinone imines, iminoquinone methides, epoxides, arene oxides, and nitrenium ions, and those metabolites readily react with soft electrophiles such as the sulfhydryl group of cysteine. In contrast, hard reactive metabolites, most commonly seen as reactive aldehydes, preferentially react to hard electrophiles such as amines of lysine, arginine, and nucleic acids. Because of their instability and low levels, direct detection and structural characterization of reactive metabolites are inherently difficult. A commonly utilized approach is to trap reactive metabolites with a capture molecule, resulting in formation of stable adducts that are subsequently characterized by tandem mass spectrometry. For soft reactive electrophilic metabolites, glutathione (GSH) is the most common trapping agent used in microsomal incubations. Resulting GSH adducts are analyzed and structurally characterized by liquid chromatography-tandem mass spectrometry (LC-MS/MS), and structural information of reactive metabolites can be deduced from tandem MS spectra. Some early studies9,10 have shown that all GSH adducts undergo a common neutral loss of 129 Da (the γ-glutamyl moiety) in the collisioninduced dissociation (CID), and thus, neutral loss scanning has been widely used as a generic method for rapidly detecting GSH adducts formed in microsomal incubations.11,12 This approach combined with stable isotope labeling has proven to be very sensitive and suitable for high-throughput screening,13,14 which can be utilized for analyzing a large volume of compounds in drug discovery. However, the same strategy cannot be applied to the detection of hard electrophiles, largely due to low trapping efficiency of GSH. Alternative trapping agents such as semicarbazide, methoxylamine, and R-acetyllysine have been used to mimic lysine residues of proteins for capturing reactive aldehydes.15 However, structurally different adducts derived from any of those trapping agents lack a common MS characteristic such as a neutral loss of 129 Da observed for GSH conjugates. Therefore, a MS full scan must be first carried out as a survey (8) Pearson, R. G. J. Am. Chem. Soc. 1963, 85, 3533-3539. (9) Baillie, T. A.; Davis, M. R. Biol. Mass Spectrom. 1993, 22, 319-325. (10) Deterding, L. J.; Mahmood, N. A.; Burka, L. T.; Tomer, K. B. Anal. Biochem. 1989, 183, 94-107. (11) Chen, W. G.; Zhang, C.; Avery, M. J.; Fouda, H. G. In Biological Reactive Intermediates VI. Chemical and Biological Mechanisms in Susceptibility to and Preventation of Environmental Diseases; Dansetter P. M., Snyder, R., Delaforge, M., Gibson, G. G., Greim, H., Jollow, D. J., Monks, T. J., Sipes, I. G., Eds.; Kluwer Academic/Plenum Press: New York, 2001,; pp 521524. (12) Castro-Perez, J.; Plumb, R.; Liang, L.; Yang, E. Rapid Commun. Mass Spectrom. 2005, 19, 798-804. (13) Yan, Z.; Caldwell, G. W. Anal. Chem. 2004, 76, 6835-6847. (14) Yan, Z.; Maher, N.; Torres, R.; Caldwell, G. W.; Huebert, N. Rapid Comm. Mass spectrum. 2005, 19, 3322-3330. (15) Evans, D. C.; Watt, A. P.; Nicoll-Griffith, D. A.; Baillie, T. A. Chem. Res. Toxicol. 2004, 17, 3-16.

scan to screen potential adducts, and tandem MS analysis is subsequently performed to confirm the structure of the adduct. This approach is largely effective for identifying hard reactive metabolites, but it is not efficient and cannot be used in a highthroughput screening fashion for several reasons. First, the background is usually high in MS full scans; as a result, responses of adducts are not readily visually identified; second, acquisition and interpretation of tandem MS spectra is time-consuming. Additionally, in order to detect different reactive metabolites, one must be able to predict the type of reactive metabolites (hard or soft) likely derived from a test compound and select an appropriate trapping agent accordingly. This task is challenging and very timeconsuming when a large number of compounds with diverse structural scaffolds need to be analyzed. Furthermore, if both hard and soft reactive metabolites are potentially derived from a single test compound, two separate trapping experiments should be performed in order to cover all potential bioactivation pathways. Comparing to MS full scan, neutral loss MS scan is highly preferred since it is a more specific detection method and provides greater signal-to-noise ratios. The signal-to-noise ratio is very critical for one to use a survey scan such as neutral loss scan to trigger an MS/MS full scan in order to obtain tandem MS spectra in a single run. This technique is becoming increasingly common because of high efficiency for analyzing multiple samples. In order to fully take advantage of the benefits of neutral loss scan, a bifunctional peptide is evaluated for simultaneous trapping and rapid screening of both hard and soft reactive metabolites. In the present method, the glycine of GSH is replaced by a lysine residue. The new trapping peptide (γ-glutamylcysteinlysine, γGSK) can capture both hard and soft reactive metabolites formed in microsomal incubations by conjugation to either the sulfhydryl or the alphatic amine group, and resulting γGSK adducts are subsequently analyzed by LC-MS/MS. Similar to the GSH trapping, all adducts contain a γ-glutamyl moiety and thus undergo the same neutral loss of 129 Da. Therefore, they can be readily detected using MS/MS neutral loss scanning as a generic MS method. Combining with stable isotope trapping, this approach is very sensitive and highly efficient; therefore, it can be potentially used as a high-throughput method for screening and identifying both hard and soft reactive metabolites in drug discovery. EXPERIMENTAL SECTION Materials. All reagents and solvents used in the present study were of the highest possible grade available. The following chemicals were obtained from Sigma-Aldrich (St. Louis, MO), including clozapine, 2-methylfuran, p-cresol, diclofenac, β-estradiol, furan, 2(2-thienyl)furan, glutathione, midazolam, omeprazole, acetaminophan, testosterone, tolbutamide, dextromethorphan, glucose 6-phosphate, β-nicotinamide adenine dinucleotide phosphate (NADP+), and glucose-6-phosphate dehydrogenase. The trapping peptide γGSK and its stable-isotope-labeled analogur (γ-glutamylcystein-13C6-15N2-lysine, γGSK*) were custom synthesized. For γGSK*, the isotopic purity was 92%, estimated by the supplier using NMR. Pooled human liver microsomes were from BD Gentest (Woburn, MA). (S)-Mephenytoin was supplied by Ultrafine Chemicals (Manchester, UK). Microsomal Incubations. All incubations were performed at 37 °C in a water bath, as previously described.13 For trapping reactive metabolites formed in microsomal incubations, test Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4207

Figure 1. Experimental design for stable isotope trapping (A) and NL MS scan detecting both hard and soft reactive metabolites (B). Ms and Mh denote soft and hard reactive metabolites, respectively.

compounds were mixed with human microsomal proteins in 50 mM potassium phosphate buffer (pH 7.4) supplemented with γGSK and γGSK* premixed at an equal molar ratio. Reaction mixtures were prewarmed at 37 °C for 5 min. Reactions were initiated by the addition of a NADPH generating system to give a final volume of 1 mL. Final reaction mixtures contained 10 µM test compounds, 1 mg/mL microsomal proteins, 1 mM γGSK and γGSK*, respectively, 1.3 mM NADP+, 3.3 mM glucose 6-phosphate, 0.4 unit/mL glucose-6-phosphate dehydrogenase, and 3.3 mM magnesium chloride. After a 60-min incubation, reactions 4208

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

were terminated by the addition of 600 µL of acetonitrile. Samples were centrifuged at 10000g for 15 min at 4 °C to pellet the precipitated protein, and supernatants were concentrated to 400300 µL on a SpeedVac dryer to evaporate acetonitrile. Then, samples were subjected to MS analysis for γGSK adducts. To assess protease stability, γGSK was incubated in 50 mM phosphate buffer (pH 7.4) containing either human or rat liver microsomes. After a 2-h incubation, microsomal proteins were precipitated by the addition of acetonitrile, and resulting samples were diluted in water (1:1000) and then analyzed for degradation of γGSK by LC-MS.

temperature 120 °C, desolvation temperature 300 °C, and sample cone voltage 26 V. LC-MS/MS Analyses. For rapid screening of reactive metabolites, samples were first subjected to chromatographic separations with an Agilent 1100 HPLC system integrated with an autosampler (Agilent Technologies, Palo Alto, CA), and eluents were introduced to the Quattro Micro triple quadrupole mass spectrometer operated in the neutral loss scanning mode as described above. An Agilent Zorbax SB C18 column (2.1 × 50 mm) was used for chromatographic separations. A generic LC profile was used as below: a single gradient at a flow rate of 0.3 mL/min from 95% aqueous (0.5% acetic acid) to 95% organic (acetonitrile) over 7 min and holding at 95% organic for 2 min before re-equilibration at the initial condition (95% aqueous). LCMS/MS analyses were carried out on 10-µL aliquots of samples. For neutral loss scan, the collision energy was set at 18 eV. Mass spectra collected in the neutral loss scan mode were obtained by scanning over the range m/z 400-800 in 2.0 s. Data were processed using the Masslynx version 4.0 software from Micromass. After a positive adduct was detected, a CID MS/MS spectrum was subsequently obtained to further confirm the structure of the conjugate. For analysis of metabolites generated from CYP marker substrates, the electrospray ionization source was operated in either the positive ion mode detecting 6β-hydroxytestosterone, dextrorphan, and 1-hydroxymidazolam or the negative ion mode detecting acetaminophan, 4-hydroxytolbutamide, and 4′-hydroxymephenytoin. The MRM transitions monitored were m/z 233 f 190 for 4-hydroxymephenytoin, m/z 342 f 324 for 1-hydroxymidazolam, m/z 285 f 186 for 4-hydroxytolbutamide, m/z 150 f 107 for acetaminophan, m/z 305 f 269 for 6β-hydroxytestosterone, and m/z 258 f 157 for dextrorphan.

Figure 2. Tandem MS spectra of the trapping agent (γGSK, A) and its isotopic analogue (γGSK*, B).

Inhibition of CYP Enzymes by GSK. The effect of γGSK (2 mM) on the metabolism of marker substrates by corresponding CYP enzymes was investigated using pooled human liver microsomes. Drugs used as CYP marker substrates were 80 µM phenacetin (CYP1A2), 32 µM (S)-mephenytoin (CYP2C19), 120 µM tolbutamide (2C9), 5 µM dextromethorphan (CYP2D6), 80 µM testosterone (CYP3A4), and 5 µM midazolam (CYP3A4). Individual marker substrates were preincubated with human liver microsomes in the presence and absence of γGSK for 5 min at 37 °C prior to the addition of a NADPH-generating solution to a final volume of 250 µL. After incubation for 20 min, the reactions were terminated by the addition of 75 µL of acetonitrile, and metabolites generated from individual CYP marker substrates were analyzed by the LC-MS/MS method described below. A comparison was made to the controls (no γGSK) and activity expressed as the percentage of control activity remaining. Mass Spectrometry. MS analyses were performed on a Micromass (Manchester, UK) Quattro Micro triple quadrupole mass spectrometer, as previously described.13 The ESI ion source was operated in the positive ion mode, and experimental parameters were set as follows: capillary voltage 3.2 kV, source

RESULT AND DISCUSSION Experiment Design. In order to trap both soft and hard reactive metabolites, a bifunctional peptide (γGSK) was synthesized (Figure 1A). For the new trapping peptide, a lysine reside was used to replace the glycine of GSH, and thus, γGSK contains both a sulfhydryl and an alphatic amine group that react with soft and hard reactive metabolites, respectively. Because all adducts derived from both conjugation reactions undergo the same neutral loss of the γ-glutamyl moiety (129 Da) under collision-induced dissociation, neutral loss scan can be utilized as a generic method for rapidly screening both hard and soft reactive metabolites. In order to rapidly eliminate false responses to neutral loss scan and facilitate identification of peptide conjugates, an isotopic analogue (γGSK*) was used for stable isotope trapping experiments,13 in which both natural and labeled trapping agents were equally mixed and added to microsomal incubations, resulting in formation of both natural and stable isotope-labeled γGSK conjugates (Figure 1B). Because all isotopic labeling sites locate at the lysine residue, both natural and labeled γGSK adducts undergo the same neutral loss of 129 Da under collision-induced dissociation and thus display a distinct isotopic doublet with a mass difference of 8 Da. The unique isotopic doublet can be readily used for visual identification of reactive metabolites. False responses to NL of 129 Da can be easily eliminated because of lacking the expected isotopic pattern. Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4209

Table 1. Inhibition of CYP Activities by the Trapping Agent GSK (2 mM) CYP enzyme CYP1A2 CYP2C9 CYP2C19 CYP2D6 CYP3A4

marker activity

relative activity (%)a

phenacetin O-deethylation tolbutamide methylhydroxylation (S)-mephenytoin 4’-hydroxylation dextromethorphan O-demethylation testosterone 6β-hydroxylation midazolam 1-hydroxylation

108 94 95 90 103 97

a CYP activity is expressed in relative to the controls (without γGSK). Data are derived from pooled human liver microsomes and are the average of three determinations.

It should be noted that, for the synthesis of γGSK*, isotope labeling positions were chosen solely based on the availability of the starting material. Apparently, isotopic labeling at any positions on either lysine or cystein should be fine. Characterization of γGSK. Figure 2A is the tandem MS spectrum of the molecular ion of the trapping peptide (γGSK). The CID spectrum of m/z 379 generated fragment ions at m/z 361, 250, 233, 215, 147, and 129. The product ion at m/z 361 was formed by the loss of water. Apparently, the most intense fragment ion at m/z 250 was produced via a neutral loss of 129 Da (the γ-glutamyl moiety). Two product ions at m/z 233 and 147 were attributed to a cleavage of the peptide bond between lysine and cysteine, and those two fragment ions subsequently lost a molecule of water to give rise two ions at m/z 215 and 129, respectively. For the isotope-labeled peptide (γGSK*), the CID spectrum of m/z 387 showed fragment ions at m/z 233, 215, and 130 (Figure 2B), which are identical to those generated from γGSK. As expected, a mass shift of 8 Da was observed for other fragment ions, resulting in isotope-labeled product ions at m/z 369 (vs 361), 258 (vs 250), 155 (vs 147), and 137 (vs 129). Stability of the trapping agent (γGSK) was assessed by LCMS/MS under different experimental conditions. No significant degradation was detected after incubation for 2 h at 37 °C with either human or rat liver microsomes. It was also found that γGSK was fairly stable in an aqueous solution for at least two weeks when stored at -20 °C. Apparently, inhibition of CYPs is an undesirable characteristic for a trapping molecule such as γGSK, since formation of reactive metabolites in liver microsomes is primarily catalyzed by this family of enzymes. In order to evaluate the effect on CYPs in human liver microsomes, marker-specific activity of five major isozymes (CYP1A2, 2C9, 2C19, 2D6, and 3A4) was measured in the presence of 2 mM γGSK. As shown in Table 1, no significant inhibition by γGSK was observed for all five major CYP enzymes in human liver microsomes. Trapping and Detecting Soft Reactive Metabolites. Several compounds, such as p-cresol, clozapine, diclofenac, and β-estradiol, known to form different soft reactive metabolites in human liver microsomal incubations, were tested in this study. As expected, in HLM incubation samples, all γGSK adducts derived from those model compounds were identified by the appearance of an isotopic doublet with a mass difference of 8 Da. Figure 3A is a representative NL total ion chromatogram of a microsomal incubation sample of p-cresol. Two major components, 4210 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

Figure 3. Stable isotope trapping and NL MS/MS analyses of the soft reactive metabolite formed by p-cresol. (A) total ion chromatogram of the neutral loss scanning of 129 Da; (B) NL MS/MS spectrum of the γGSK adduct M1; (C) NL MS/MS spectrum of the γGSK adduct M2; (D) tandem MS spectrum of the adduct M1; (E) tandem MS spectrum of the adduct M2.

Scheme 1. Bioactivation of p-Cresol in HLM and Trapping Soft Reactive Metabolites

coeluted at RT 0.6 (M1) and 0.7 min (M2), showed responses to the NL scan of 129 Da. The NL MS/MS spectrum of M1 exhibited an isotopic doublet at m/z 501 and 509 (Figure 3B), whereas component M2 displayed an isotopic doublet at m/z 485 and 493 (Figure 3C). A mass difference of 8 Da clearly suggested that both M1 and M2 are γGSK adducts. The data are in agreement with our previous results obtained from a GSH trapping study, which showed that two metabolic pathways mediated bioactivation of p-cresol in human liver microsomes: dehydrogenation resulting in formation of a reactive quinone methide, and aromatic oxidation leading to production of 4-methyl-o-benzoquinone16 These two soft reactive metabolites were captured by conjugation to the thiol group of γGSK to form stable adducts by losing a molecule of water (Scheme 1). As shown in Figure 3D, the CID spectrum of the molecular ion of M1 (m/z 501) generated fragment ions at m/z 483, 372, 355, 337, 147, and 129. For M2, the CID spectrum of m/z 485 generated fragment ions at m/z 467, 379, 356, 339, 321, 250, 233, 147, and 129 (Figure 3E). The characteristic NL of 129 Da was observed for both M1 (m/z 372) and M2 (m/z 356), which is the key evidence confirming the putative structures of those two γGSK adducts (Scheme 2). For clozapine, an antipsychotic drug often associated with a number of adverse drug reactions in clinic, two γGSK adducts were detected by NL scan. The major adduct appeared at m/z 703 and 711 (Figure 4A), which is consistent with the bioactivation of clozapine to form nitrenium ions17 The nitrenium ions were captured by thiol conjugation to the γGSK. The minor adduct displayed an isotopic doublet at m/z 719 and 727 (Figure 4B), which is likely derived from the N-oxide metabolite of clozapine via a similar bioactivation pathway. The putative structures of those two γGSK adducts derived from clozapine were verified by tandem MS spectra (data not shown). Two γGSK adducts were detected for diclofenac, a nonsteroidal anti-inflammatory drug associated with a rare, but severe, inci(16) Yan, Z.; Zhong, H. M.; Maher, N.; Torres, R.; Leo, G. C.; Caldwell, G. W.; Huebert, N. Drug Metab. Dispos. 2005, 33, 1867-1876. (17) Uetrecht, J. P.; Ma, H. M.; MacKnight, E.; McClelland, R. Chem. Res. Toxicol. 1995, 8, 226-233.

dence of hepatic injury. As shown in Figure 4C, one γGSK adduct exhibited an isotopic doublet at m/z 688 and 696, which was produced by first forming 4′-hydroxydiclofenac that was further oxidized to a benzoquinone imine intermediate.18 The benzoquinone imine intermediate was trapped by thiol conjugation to a stable γGSK adduct. The second γGSK adduct showed an isotopic doublet at m/z 654 and 662 (Figure 4D), which was consistent with our previously published work on the bioactivation of diclofenac.19 In that study, we showed dechlorination of the dichlorophenyl ring in bioactivation of diclofenac followed by GSH conjugation. The putative structures of those two γGSK adducts derived from diclofenac were verified by tandem MS spectra (data not shown). CYP-mediated oxidation of β-estradiol generates a catechol metabolite that can subsequently be oxidized to a reactive quinine metabolite. In the presence of γGSK, the reactive metabolite was trapped by thiol conjugation to form a peptide adduct that appeared in NL scan as an isotopic doublet at m/z 665 and 673 (Figure 4E), which is apparently consistent with what was previously found using GSH as a trapping agent.13 The putative structure of this γGSK adduct derived from β-estradiol was confirmed by the tandem MS spectrum (data not shown). Our results, as expected, suggest that γGSK is as effective as GSH in trapping a variety of soft reactive metabolites. Trapping and Detecting Hard Reactive Metabolites. Furans were selected as our model compounds to evaluate the effectiveness of γGSK for trapping hard reactive metabolites, since it has been well documented that metabolism of furans leads to formation of reactive aldehydes,20,21 and those reactive aldehydes were not effectively captured by GSH. In the presence of γGSK in (18) Tang, W.; Stearns, R. A.; Bandiera, S. M.; Zhang, Y.; Raab, C.; Braun, M. P.; Dean, D.; Pang, J.; Leung, K.; Doss, G.; Strauss, J. R.; Kwei, G. Y.; Rushmore, T. H.; Chiu, S-H, L.; Baillie, T. A. Drug Metab. Dispos. 1999, 27, 365-372. (19) Yan, Z.; Li, J.; Huebert, N.; Caldwell, G. W.; Du, Y.; Zhong, H.: Drug Metab. Dispos. 2005, 33, 706-713. (20) Kobayashi, T.; Sugihara, J.; Harigaya, S. Drug Metab. Dispos. 1987, 15, 877-881. (21) Madyastha, K. M.; Raj, C. P. Drug Metab. Dispos. 1992, 20, 295-301.

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4211

Scheme 2. Trapping of the Hard Reactive Metabolite Derived from Furan

produced fragment ions at m/z 253 via a loss of CO2. A cleavage of the amide bond between γ-glutamyl group and cysteine gave rise to product ions at m/z 130. A cleavage of γ-glutamic acid from the γGSK adduct yielded a product ion at m/z 281 that was subsequently fragmented to form two product ions at m/z 237 (-CO2) and 263 (-H2O), respectively.

Figure 4. Stable isotope trapping and NL MS/MS analyses of soft reactive metabolites derived from diclofenac (A, B), clozapine (C, D), and β-estradiol (E).

microsomal incubations, all those reactive metabolites formed stable conjugates that were detected using NL MS/MS. Figure 5A is a representative NL total ion chromatogram of a microsomal incubation sample of furan. A major component eluted at RT 8.9 min exhibited an isotope doublet at m/z 427 and 435 (Figure 5B), clearly suggesting formation of a γGSK adduct. The isotope pattern seems consistent with the bioactivation pathway: cis-butene-1,4-dial, a hard reactive metabolite derived from the ring scission of furan, first conjugated to the amine group, and subsequently formed a stable peptide adduct by reacting to the sulfhydryl group at the C-3 positions (Scheme 2). The proposed structure was confirmed by the tandem MS spectrum of the γGSK adduct. As shown in Figure 5C, molecular ions of the γGSK adduct (MH+ m/z 427) underwent dehydrogenation to give rise of ions at m/z 409 and produced fragment ions at m/z 364 by a sequential loss of CO2 and NH3. A neutral loss of 129 Da (the γ-glutamyl moiety) yielded product ions at m/z 298 that subsequently 4212 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

Figure 5. NL MS/MS analyses of the hard reactive metabolite formed by furan in human microsomal incubations. (A) total ion chromatogram of the neutral loss scanning of 129 Da; (B) NL MS/ MS spectra of γGSK adducts; (C) tandem MS spectrum of the molecular ions of natural γGSK adduct.

Scheme 3. Bioactivation Pathways of 2-(2-Thienyl)furan and Trapping both Soft and Hard Reactive Metabolites by γGSK

Similar adducts were detected in microsomal incubations with other furan derivatives such as 2-methylfuran and menthofuran. As shown in Figure 6A, a γGSK conjugate derived from 2-methylfuran exhibited an isotopic doublet at m/z 441 and 449. For menthofuran, a γGSK conjugate was identified by the appearance of an isotopic doublet at m/z 527 and 535 (Figure 6B). The proposed γGSK adducts derived from 2-methylfuran and menthofuran were further confirmed by tandem MS spectra (data not shown). These results demonstrated that γGSK is effective in trapping hard reactive metabolites. Simultaneous Trapping and Detecting Both Soft and Hard Reactive Metabolites. 2-(2-Thienyl)furan contains a furan ring and a thiophene moiety. It was expected that 2-(2-thienyl)furan could be metabolized via two bioactivation pathways (Scheme 3): epoxidation of the thiophene ring to produce an epoxide (FM1) known as a soft RM22 and the ring scission of the furan moiety to form a hard RM (FM2) as cis-butene-1,4-dial derivative.20,21 It was also anticipated that, in microsomal incubations, both the epoxide and cis-butene-1,4-dial derivative be captured by γGSK to form two stable adducts (FM1C and FM2C), respectively. Consistent with the predicted metabolic pathways, two γGSK adducts were (22) Bray, H. G.; Carpanini, F. M. B.; Waters, B. D. Xenobiotica 1971, 1, 157168. (23) Dalvie, D. K.; Kalgutkar, A. S.; Khojasteh-Bakht, S. C.; Obach, R. S.; O’Donnell, J. P. Chem. Res. Toxicol. 2002, 15, 269-299. (24) Soglia, J. R.; Contillo, L. G.; Kalgutkar, A. S.; Zhao, S.; Hop, C. E. C. A.; Boyd, J. G.; Cole, M. J. Chem. Res. Toxicol. 2006, 19, 480-490. (25) Soglia, J. R. J. Pharm. Biomed. Anal. 2004, 36, 105-116.

identified by the appearance of two isotope doublets in NL-MS: one at m/z 509 and 517 (FM1C, Figure 7A), and the other at m/z 527 and 535 (FM2C, Figure 7B). Surprisingly, an additional adduct (Figure 7C) was detected in the same incubation at m/z 495 and 503. A likely metabolic pathway is a ring opening of the epoxide, giving rise of the R,β-unsaturated aldehyde (FM3) that was subsequently captured by γGSK to form a stable peptide conjugate (FM3C, Scheme 3). This metabolic pathway has been previously documented,23 although it is not very common for a thiophene epoxide. All three proposed γGSK adducts (FM1C,

Figure 6. NL MS/MS analyses of hard reactive metabolites derived from 2-methylfuran (A) and methofuran (B) in human microsomal incubations.

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4213

Figure 7. Simultaneously trapping and detecting both hard and soft reactive metabolites derived from 2-(2-thienyl)furan in human microsomal incubations. (A) NL MS/MS spectrum of FM1C; (B) NL MS/ MS spectrum of FM2C; (C) NL MS/MS spectrum of FM3C (Scheme 3).

FM2C, FM3C)derived from 2-(2-thienyl)furan were further confirmed by the tandem MS spectra (data not shown). This example clearly demonstrated that γGSK could be used for simultaneous trapping both soft and hard reactive metabolites in a single incubation, which is a great advantage for investigating bioactivation pathways compared to other trapping agents such as GSH, semicarbazide, and methoxylamine. (26) Dieckhaus, C. M.; Fernandez-Metzler, C. L.; King, R.; Krolikowski, P. H.; Baillie, T. A. Chem. Res. Toxicol. 2005, 18, 630-638.

4214

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

CONCLUSION Detection and identification of both hard and soft reactive metabolites are important for a comprehensive evaluation of lead compounds for bioactivation. Presently, several analytical methods have been developed for detecting reactive metabolites12,13,24-26 using the GSH trapping method. Evidently, the success of GSH trapping is mainly limited to the detection of soft reactive metabolites. This study has clearly demonstrated that the bifunctional trapping agent γGSK is capable of trapping both soft and hard reactive metabolites, and thus, two types of reactive metabolites can be simultaneously analyzed by neutral loss MS scan, which is a significant improvement compared to the GSH trapping approach. Additionally, in combination with stable isotope trapping, this method can be utilized for screening of reactive metabolites in a high-throughput fashion, regardless of the type of reactive metabolites. This capacity was not accomplished previously by using any other trapping agents including GSH, semicarbazide, methoxylamine, and R-acetyllysine. This feature is important in a lead optimization process in which a large number of compounds with a wide variety of structural moieties are evaluated for bioactivation. A disadvantage of stable isotope trapping is the high cost associated with the isotope-labeled trapping agent. However, this disadvantage can be overcame by utilizing a more updated triple quadrupole mass spectrometer on which one can use a neutral loss as a survey scan to trigger an MS/MS full scan. Therefore, tandem MS spectra can be simultaneously obtained in a single run and used for structural identification. In drug discovery, throughput is more critical for analyzing high volumes of compounds in a time-critical fashion. Unlike GSH, the bifunctional peptide (γGSK) is not an endogenous molecule. Its cell permeability remains to be tested if one wants to extend this approach to trap reactive metabolites formed in hepatocytes. Received for review March 28, 2007. AC0701029

January

17,

2007.

Accepted