896
Chem. Res. Toxicol. 2005, 18, 896-903
Dansyl Glutathione as a Trapping Agent for the Quantitative Estimation and Identification of Reactive Metabolites Jinping Gan,* Timothy W. Harper, Mei-Mann Hsueh, Qinling Qu, and W. Griffith Humphreys Department of Pharmaceutical Candidate Optimization, Pharmaceutical Research Institute, Bristol-Myers Squibb Company, Princeton, New Jersey 08543 Received November 16, 2004
A sensitive and quantitative method was developed for the estimation of reactive metabolite formation in vitro. The method utilizes reduced glutathione (GSH) labeled with a fluorescence tag as a trapping agent and fluorescent detection for quantitation. The derivatization of GSH was accomplished by reaction of oxidized glutathione (GSSG) with dansyl chloride to form dansylated GSSG. Subsequent reduction of the disulfide bond yielded dansylated GSH (dGSH). Test compounds were incubated with human liver microsomes in the presence of dGSH and NADPH, and the resulting mixtures were analyzed by HPLC coupled with a fluorescence detector and a mass spectrometer for the quantitation and mass determination of the resulting dGSH adducts. The comparative chemical reactivity of dGSH vs GSH was investigated by monitoring the reaction of each with 1-chloro-2,4-dinitrobenzene or R-(+)-pulegone after bioactivation. dGSH was found to be equivalent to GSH in chemical reactivity toward both thiol reactive molecules. dGSH did not serve as a cofactor for glutathione S-transferase (GST)mediated conjugation of 3,4-dichloronitrobenzene in incubations with either human liver S9 fractions or a recombinant GST, GSTM1-1. Reference compounds were tested in this assay, including seven compounds that have been reported to form GSH adducts along with seven drugs that are among the most prescribed in the current U.S. market and have not been reported to form GSH adducts. dGSH adducts were detected and quantitated in incubations with all seven positive reference compounds; however, there were no dGSH adducts observed with any of the widely prescribed drugs. In comparison with existing methods, this method is sensitive, quantitative, cost effective, and easy to implement.
Introduction The preclinical prediction of the complete adverse event profile of a drug candidate is a difficult and elusive task. The prediction is especially difficult in the case of idiosyncratic drug reactions (IDR),1 which have a very low frequency of occurrence, no obvious dose-response relationship, and no universal animal models for evaluation (1-5). At present, there is no commonly accepted experimental approach for the prediction of IDR. However, a number of drugs that are associated with IDR form reactive metabolites, which react with endogenous nucleophiles, including proteins (3). Although there is no proof of a causal relationship between reactive metabolites and IDR and there is no commonly accepted hypothesis for the underlying mechanism of such adverse events (1, 2, 6), the formation of a reactive metabolite is considered a potential liability in the development of new chemical entities (NCE) for the pharmaceutical industry (7). To minimize such liabilities, preclinical screening for * To whom correspondence should be addressed. Tel: 609-818-6529. E-mail:
[email protected]. 1 Abbreviations: dGSH, dansylated reduced glutathione; GSH, reduced glutathione; GSSG, oxidized glutathione; CDNB, 1-chloro-2, 4-dinitrobenzene; DCNB, 3,4-dichloronitrobenzene; HLM, human liver microsome; GST, glutathione S-transferase; ESI, electrospray ionization; DTT, dithiothreitol; NSAID, nonsteroidal antiinflammatory drug; CYP, cytochrome P450; IDR, idiosyncratic drug reactions; NCE, new chemical entities; SAR, structure-activity relationship.
reactive metabolites has been adopted by a number of companies through covalent protein binding studies and/ or the use of chemical trapping agents (7-9). Covalent protein binding studies measure the extent of protein-bound drug-related material after incubation of radiolabeled compounds with an in vitro bioactivation system or through direct in vivo dosing to experimental animals (7, 10). The quantitation of the level of covalent protein binding has been proposed as a means to minimize the risk of reactive metabolite-mediated toxicity. The method can be applied both to provide rank ordering information on a series of compounds or as an absolute measurement of binding vs reference standards. However, the use of this technique is often limited due to the prerequisite need for the synthesis of radiolabeled material for every compound of interest. The formation of covalent adducts between reactive metabolites and small molecule nucleophiles can also be used to gain information on the formation and nature of reactive metabolites. Reduced glutathione (GSH) has been widely used as an in vitro chemical trapping agent for the characterization and mechanistic study of reactive metabolites (11). GSH is an important cellular component that is crucial for the cellular homeostasis of redox potential and a natural defense against oxidative stress (12, 13). Glutathione also serves the role as the primary intracellular nucleophile. It reacts with many known
10.1021/tx0496791 CCC: $30.25 © 2005 American Chemical Society Published on Web 04/20/2005
Dansyl Glutathione as a Trapping Agent
reactive metabolites, and the resulting GSH adducts are usually nontoxic and readily excreted, although in some cases, the formation of GSH adducts has been associated with kidney and bile duct toxicities after further processing of the adducts (14-17). GSH adducts of NCEs can be identified, and the molecular mass of the GSH adduct can be determined by various LC-MS techniques such as neutral loss LC-MS/MS (8, 9, 18). Detailed structural characterization can be accomplished by further LC-MS/MS and NMR studies. It is also important to gain quantitative information regarding the extent of adduct formation to fully characterize the adduct. A quantitative assay allows for the determination of the magnitude of the reactive intermediate trapping as well as enabling measurements such as the effects of various substitution patterns on the amount of adduct formed for structure-activity relationship (SAR) development around a problematic NCE. With this information, it may be possible to formulate a working hypothesis of the bioactivation mechanism that, in turn, would enable chemistry efforts attempting to block the route of metabolism leading to reactive metabolite formation. Tritiated GSH trapping allows adducts to be directly quantified by radioactivity counting of the adduct peaks (19). However, adequate separation of the GSH adducts from the unreacted material is challenging and often results in insufficient sensitivity. Furthermore, the use of radioactivity requires special facilities and creates environmentally hazardous waste material. Finally, the radiolabeled glutathione method is relatively expensive. Thus, there exists a need for a sensitive, quantitative, and cost effective method to trap reactive metabolites in vitro. Here, we describe a method that exploits the fluorescent signal of a derivatized GSH to achieve sensitive and quantitative adduct detection.
Experimental Procedures Materials. Celecoxib, sertraline, troglitazone, and paroxetine were purchased from Toronto Research Chemicals (North York, ON, Canada). All other chemicals were purchased from SigmaAldrich (Milwaukee, WI). Pooled human liver microsomes (HLMs) and pooled human liver S9 fractions were purchased from BD Gentest (Woburn, MA). Recombinant glutathione S-transferase (GST) M1-1 was purchased from Invitrogen (Carlsbad, CA). Instrumentation. A Shimadzu LC-10Avp HPLC system (Shimadzu, Columbia, MD) was used for separation. A 4.6 mm × 150 mm Phenomenex Prodigy ODS-2 column (Torrance, CA) was used to separate dansylated reduced glutathione (dGSH) adducts using a gradient mobile phase system at a flow rate of 1 mL/min. Initial chromatographic conditions were 0.1% formic acid in water:acetonitrile (80:20, v/v), followed by a linear increase of acetonitrile to 50% from 3 to 23 min, followed by a second linear gradient to 90% acetonitrile from 23 to 40 min. The gradient was kept at 90% acetonitrile until 44 min and was then returned to the initial running condition at 45 min. The column was then reequilibrated for 5 min to give a total analysis time of 50 min. A fluorescence detector (Shimadzu RFL-10A) was used for the detection and quantitation of adducts. The excitation and emission wavelengths were set at 340 and 525 nm, respectively. The HPLC eluent from the fluorescence detector was connected in-line with a Finnigan LCQ ion trap mass spectrometer (Thermo Finnigan, San Jose, CA). Full scans with mass ranges from m/z 600-1100 Da were obtained with alternating positive and negative ionization. Quantitation of dGSH trapped reactive metabolites was accomplished by com-
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 897 parison of peak area in the fluorescence chromatogram with a standard curve using dGSH as an external standard. For sample analysis in the experiments involving 1-chloro2,4-dinitrobenzene (CDNB) and 3,4-dichloronitrobenzene (DCNB), a Waters 2690 HPLC system (Waters, Milford, MA) coupled with a Finnigan LCQ ion trap mass spectrometer was used. Samples were separated with a YMC-ODS AQ column (2.0 mm × 150 mm) at a flow rate of 0.2 mL/min. Mobile phases consisted of 20 mM ammonium acetate (pH 5.1) in water (A) and acetonitrile (B). The initial mobile phase composition was at 95% A:5% B. After equilibration for 3 min at 5% B, the concentration of solvent B was increased linearly to 90% in 20 min. Conditions were held at 90% B for an additional 6 min before returning to 5% B. The total analysis time was 35 min. Adduct masses were determined by ion trap mass spectrometry. The reaction rate of dGSH or GSH with CDNB was calculated from CDNB concentration vs time curve using linear regression. A Shimadzu LC-8A preparative HPLC system was used for the purification of dGSH. A Phenomenex Luna 21.2 mm × 250 mm column was used at a flow rate of 20 mL/min. A linear gradient was used from 100% A:0% B to 33% A:67% B (A ) 90% water/10% methanol/0.1% TFA and B ) 90% methanol/ 10% water/0.1% TFA) in 20 min. A Bruker AVANCE 500 MHz NMR instrument was used for the collection of 1H and 13C NMR spectra. Synthesis of Dansylated Oxidized Glutathione (GSSG). The synthesis of dansylated GSSG is modified from literature procedures (20, 21). A solution of dansyl chloride (140 mg; 0.52 mmol) in acetone (2 mL) was added dropwise to a solution of GSSG (150 mg; 0.24 mmol) in aqueous NaOH (1.2 mL of a 1 M solution). The reaction mixture was stirred at room temperature for 30 min. The mixture was then washed with diethyl ether (2 × 10 mL), and the aqueous phase was purified by preparative HPLC. The HPLC effluent was lyophilized to give dansylated GSSG (214 mg; 80% yield) as a white powder. MS [positive electrospray ionization (ESI)] [M + H]+ ) 1079.2. 1H NMR (DMSO-d ; 500 MHz): δ 1.72-1.75 (2 CH, m), 1.826 1.85 (2 CH, m), 2.09-2.16 (2 CH2, m), 2.75-2.78 (2 CH, m), 2.85 (4 CH3, s), 3.05-3.08 (2 CH, m), 3.70-3.75 (2 CH2 and 2 CH, m), 4.50-4.55 (2 CH, m), 7.27-7.29 (2 Ar-H, d, J ) 10 Hz), 7.57-7.62 (4 Ar-H, m), 8.11-8.14(2 Ar-H and 2 N-H, m), 8.24-8.27 (2 N-H, t, J ) 10 Hz), 8.32-8.34 (2 Ar-H, d, J ) 10 Hz), 8.43-9.49 (2 Ar-H and 2 N-H, m). Synthesis of dGSH. Nitrogen gas was bubbled into a solution of dansylated GSSG (285 mg; 0.264 mmol) in 0.1 M Tris buffer (pH 8, 10 mL) for 10 min, after which dithiothreitol (DTT; 77 mg, 0.5 mmol) was added. The reaction mixture was stirred at room temperature for another 30 min under an atmosphere of N2. The solution was adjusted to pH ∼4 with acetic acid and then directly purified by preparative HPLC. The HPLC effluent was lyophilized to give dGSH (265 mg; 93% yield, >99.5% pure as assessed by HPLC/UV and fluorescence detection) as a white powder. MS (positive ESI) [M + H]+ ) 541.1. 1H NMR (DMSO-d ; 500 MHz): δ 1.66-1.75 (CH, m), 1.766 1.89 (CH, m), 2.13-2.16 (CH2, m), 2.26-2.29 (SH, t, J ) 8.25 Hz), 2.585-2.64 (CH, m), 2.71-2.77 (CH, m), 2.85 (2 CH3, s), 3.69-3.79 (CH2 and CH, m), 4.32-4.37 (CH, m), 7.27-7.29 (Ar-H, d, J ) 10 Hz), 7.57-7.62 (2 Ar-H, m), 7.98-8.00 (N-H, d, J ) 7.2 Hz), 8.10-8.12 (Ar-H, d, J ) 7.2 Hz), 8.288.30 (N-H, t, J ) 6 Hz), 8.32-8.34 (Ar-H, d, J ) 8.8 Hz), 8.439.45 (Ar-H, d, J ) 8.2 Hz), 8.46-8.48 (N-H, d, J ) 8.6 Hz). 13C NMR (DMSO-d ; 125 MHz): δ 26.1, 27.9, 31.0, 40.6, 45.0, 6 54.6, 55.2, 115.1, 119.6, 123.4, 127.5, 128.1, 128.7, 129.0, 129.2, 136.3, 143.4, 170.1, 170.9, 171.2, 172.5. In Vitro Incubation Conditions. The incubation of the reference compounds with dGSH and HLM contained the following: 50 µM substrate, 1 mM dGSH, 1 mg/mL HLM protein, and 100 mM potassium phosphate buffer (pH 7.4). This mixture was preincubated for 5 min at 37 °C. The reaction was initiated by the addition of NADPH to reach a final concentration of 1 mM. The final incubation volume was 0.2 mL. Samples without substrate or dGSH were used as blanks or controls,
898
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Gan et al. etry. In the incubations with GSTM1-1, 0.1 mg/mL GSTM1-1 was used instead of human liver S9 fractions.
Results
Figure 1. Structures of test compounds. respectively. The substrates were dissolved in organic solvent (acetonitrile for all compounds except troglitazone; DMSO was used for troglitazone); the concentration of solvent was kept under 1% in all incubations. After 30 min of incubation, the reaction was terminated by the addition of two volumes of icecold methanol containing 5 mM DTT. After vortexing and centrifugation, 60 µL of the resulting supernatant was analyzed by HPLC. The structures of all reference compounds are shown in Figure 1. Reference compounds reported to form GSH adducts included acetaminophen (22), R-(+)-pulegone (22, 23), clozapine (24, 25), diclofenac (26), troglitazone (27-29), bromobenzene (30, 31), and precocene I (32). Marketed compounds that have not been reported to form GSH adducts included omeprazole, lansoprazole, fluoxetine, celecoxib, paroxetine, loratadine, and sertraline. Effect of GSH or dGSH Concentrations on R-(+)-Pulegone Adduct Formation. Fifty micromolar pulegone was incubated with varying concentrations of GSH or dGSH in the presence of 1 mM NADPH and 1 mg/mL HLM. The reaction mixtures were quenched and prepared for analysis in the same fashion as described in the in vitro incubations section. The amount of R-(+)-pulegone adduct formed was determined by the measurement of the peak areas from fluorescence chromatograms for dGSH incubations using dGSH as external standard or from UV chromatograms (λmax at 254 nm) for GSH incubations using R-(+)-pulegone as external standard. Comparison of Reaction Rates of GSH and dGSH with CDNB. Fifty micromolar CDNB was incubated with 1 mM GSH or dGSH in 60 mM (pH 7.4) phosphate buffer at 37 °C in a total volume of 1 mL. Aliquots of 150 µL were taken at 0, 5, 10, 20, and 40 min. Aliquots were immediately frozen on dry ice. Upon thawing, samples were immediately subjected to HPLC analysis. Enzymatic Reaction of DCNB with GSH or dGSH. Fifty micromolar DCNB was incubated with 1 mg/mL human liver S9 fractions and 1 mM GSH or dGSH in 60 mM potassium phosphate buffer (pH 7.4) at 37 °C in a total volume of 1 mL. Aliquots of 150 µL were taken at 0, 15, 30, and 60 min. After they were quenched with an equal volume of acetonitrile and subsequent centrifugation, supernatants were analyzed by HPLC and adduct formation was confirmed by mass spectrom-
Synthesis and Characterization of dGSH. dGSH was synthesized from GSSG with satisfactory purity. MS/ MS fragmentation of dGSH, along with NMR data, confirmed the addition of a dansyl group to the terminal amino group of GSH. Fluorescence scanning of dGSH dissolved in the HPLC mobile phase gave a maximum excitation wavelength at 340 nm and a maximum emission wavelength at 525 nm. These wavelengths were subsequently utilized for the detection of all adducts formed in the incubations. Analytical Method Development. An analytical method that employed a C18 reverse phase HPLC column gradient elution was developed to ensure the separation of dGSH from potential dGSH trapped reactive metabolites. This method yielded excellent peak shape and resolution for most of the adducts described in this study except for the acetaminophen adduct, which is only modestly separated from dGSH. Incubations with Reference Compounds. Seven compounds that have been reported to form glutathione adducts upon incubation with HLM supplemented with NADPH and GSH were tested in this study. These compounds are troglitazone, acetaminophen, R-(+)-pulegone, diclofenac, clozapine, bromobenzene, and precocene I. When incubated with HLM, NADPH, and dGSH, all seven compounds showed evidence of adduct formation. Table 1 lists the LC-MS characteristics and quantitation of these adducts. The highest level of adduct formation was observed in incubations with troglitazone (12.5% of initial substrate concentration), while the lowest level of adduct formed with the set of positive reference compounds was with bromobenzene. Representative chromatograms and mass spectra of troglitazone and acetaminophen are shown in Figures 2 and 3, respectively. As illustrated in Figures 2B and 3B, the MS/MS fragmentation patterns confirmed dGSH adduction. The retention times for the adducts ranged from 11.9 to 29.3 min and were in general well-resolved from the dGSH peak. As a comparison, seven widely prescribed drugs that have not been reported to form GSH adducts were also tested in this assay. These drugs are omeprazole, lansoprazole, fluoxetine, celecoxib, paroxetine, loratadine, and sertraline. None of these reference compounds formed detectable adducts. Substrate turnover was not monitored in the incubations, so the extent of metabolism of these compounds is unknown. Incubations with CDNB and DCNB. Incubations were performed with the well-characterized probes for GSH conjugation, CDNB and DCNB, to compare the chemical and enzyme catalyzed reactivity of GSH and dGSH. A linear decline of CDNB was observed when it was incubated with GSH or dGSH as shown in Figure 4. CDNB thiol adduct formation was confirmed by mass spectrometry as chlorine substitution by the thiol groups, with masses of 473 and 706 for GSH and dGSH, respectively. The reaction rates are very similar for GSH (466 pmol/min) and dGSH (410 pmol/min). Incubation of DCNB with GSH in both human liver S9 and recombinant GSTM1-1 generated a product consistent with displacement of a chlorine atom with GSH (mass of 462). However, incubation of DCNB with
Dansyl Glutathione as a Trapping Agent
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 899
Table 1. Retention Time, Mass, and Quantitative Characteristics of DGSH Adductsa
compound troglitazone
acetaminophen bromobenzene clozapine
diclofenac precocene I
R-(+)-pulegone
retention time (min) 19.5 22.5 23.0 29.3 total 11.9 18.5 15.1 16.2 19.5 total 22.7 23.8 total 14.7 15.1 16.4 17.7 22.4 total 28.9
mass (Da)
postulated adduct composition
979
M + dGSH - 2H
689 726 864 880
M + dGSH - 2H M + dGSH + 2O - 2H M + dGSH - 2H M + dGSH + O - 2H
815 849
M + dGSH + O - HCl M + dGSH + O - 2H
732 732 762 746
M + dGSH + O - CH2 M + dGSH + O - CH2 M + dGSH + 2O M + dGSH + O
688
M + dGSH - 4H
adduct concn (µM) 0.16 0.05 0.04 5.98 6.23 0.26 0.10 4.39 0.44 0.09 4.92 0.64 0.11 0.75 0.16 0.95 1.44 0.29 0.38 3.22 4.90
% of substrate concn
12.5 0.5 0.2
9.8 1.5
6.4 9.8
a Adduct concentrations are taken as the average of two determinations. Adduct masses were determined by full scan mass spectrometry in the negative ion mode.
dGSH under similar conditions did not generate any detectable adduct. Effect of GSH or dGSH Concentrations on R-(+)Pulegone Adduct Formation. Incubations were performed with R-(+)-pulegone, HLM, NADPH, and GSH or dGSH to examine the effect of thiol concentration on adduct formation. The amount of pulegone adduct formation was dependent on thiol trapping agent concentration. As shown in Figure 5, formation of both GSH and dGSH adduct increased linearly up to 5 mM thiol trapping agents, the highest concentration tested. No difference was observed in the level of adduct formation between GSH and dGSH incubations at all concentrations above 0.1 mM. Because of UV sensitivity limitations, the amount of GSH adduct formation could not be calculated at 0.1 mM GSH.
Discussion The current method utilizes the fluorescent dansyl group to derivatize GSH at the free amino group of the glutamyl moiety. Dansyl chloride has been used since the early days of protein chemistry to derivatize terminal amino acids (33). Because thiol groups react with dansyl chloride as well, GSSG was used as the starting material. After dansyl derivatization, the disulfide bond was reduced with DTT yielding dGSH. Because detection of adducts is based on the appearance of new peaks in the fluorescence chromatograms following incubations of test compounds with dGSH and metabolic activation systems, a slow HPLC gradient system was developed to ensure adequate separation of dGSH from potential dGSHtrapped reactive metabolites. The adducts found in this study eluted at retention times from 12 to 30 min. The dGSH adduct of acetaminophen was the earliest eluting peak and did elute fairly close to the dGSH peak, but adducts from all the other reference compounds were easily resolved. As is usual with unlabeled GSH and tritiated GSH trapping approaches, dGSH was used in a large excess (20 times greater than substrate concentrations), and some background fluorescence peaks were
present in the blank chromatogram (Supporting Information). It was noted that during the incubation and subsequent analysis period, dGSH underwent oxidation resulting in a dansylated GSSG peak that eluted at 17.5 min. DTT was added along with the reaction quenching solution to reduce the dansylated GSSG, thereby minimizing the interference from the dansylated GSSG peak. It is worth pointing out that no difference in the overall quantitation of troglitazone adducts was observed with or without the addition of DTT in the quenching solution, despite literature reports of disulfide adduct formation with troglitazone (Supporting Information). However, the major adduct of troglitazone found in this study and elsewhere (28, 29) was an adduct with nominal addition of GSH that does not involve ring scission and disulfide adduct formation. Nevertheless, it is recommended that DTT be omitted from incubations with sulfur-containing test compounds if detection of disulfide-containing product is of interest. To compare the trapping efficiency of GSH and dGSH toward reactive intermediates, varying concentrations of GSH and dGSH were incubated with R-(+)-pulegone in a HLM incubation system. R-(+)-Pulegone has been shown to form GSH adduct when incubated with HLM supplemented with NADPH and GSH (22, 23). No apparent difference was observed between the two thiol agents. It was also observed that the amount of R-(+)pulegone adduct formed increased linearly with increasing GSH/dGSH concentrations. A dGSH concentration of 1 mM was selected for later studies in order to balance the need for trapping efficacy and practical considerations of background fluorescence interference. To test whether dGSH could inhibit cytochrome P450 (CYP) enzymes and thus not be suitable as a trapping agent in microsomal studies, a series of in-house CYP inhibition assays in HLM were completed. dGSH at concentrations up to 1 mM did not inhibit any of the CYP enzymes tested, including CYP3A4, CYP2C9, CYP2C19, and CYP2D6 (Supporting Information). GST plays a physiological role in the detoxification of electrophilic agents by catalyzing the reaction of the thiol
900
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Gan et al.
Figure 2. Chromatograms and mass spectra of incubations of troglitazone with HLM and dGSH. (A) Overlaid fluorescence chromatograms of troglitazone incubations. The dotted trace represents a blank incubation without troglitazone, the broken trace represents a control incubation without dGSH, and the solid trace represents the sample incubation. (B) MS/MS fragmentation of m/z 978 in the negative ESI mode. The adduct structure is proposed based on the literature reported structure of troglitazone GSH adduct, and it is not determined in this study.
group of GSH with a variety of electrophiles (34, 35). In contrast with GSH, dGSH does not appear to be a cofactor of GST in the enzymatic thiol substitution of chlorine in DCNB. Crystal structures of cytosolic GSTs have revealed that the γ-glutamyl residue of GSH is buried in the interface of the GST dimer (36). The addition of a bulky dansyl group to GSH not only introduces bulk and lypophilicity of the γ-glutamyl group but also converts the positively charged amino group to a neutral sulfonamide. All of these changes are likely to diminish the binding affinity of dGSH to the binding
pocket of GST and suggest that dGSH is likely to be a poorer cofactor than GSH for GST. This appears to be borne out by the observations that dGSH unlike GSH does not displace the chlorine atom in DCNB. On the other hand, CDNB can react with GSH via direct nucleophilic substitution of chlorine. Because this is strictly a chemical reaction involving the thiol group, dGSH reacts with CDNB at a similar rate to that seen for GSH. It therefore appears that dGSH is a trapping agent that readily reacts directly with thiol reactive intermediates but does not serve as a cofactor for GST-mediated adduct
Dansyl Glutathione as a Trapping Agent
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 901
Figure 3. Chromatograms and mass spectra of incubations of acetaminophen with HLM and dGSH. (A) Overlaid fluorescence chromatograms of acetaminophen incubations. The dotted trace represents a blank incubation without troglitazone, the broken trace represents a control incubation without dGSH, and the solid trace represents the sample incubation. (B) MS/MS fragmentation of m/z 688 in the negative ESI mode. The adduct structure is proposed based on the literature reported structure of acetaminophen GSH adduct, and it is not determined in this study.
formation. This is unlikely to be a significant limitation for this reagent, since reactive intermediates are likely to be of greatest concern when they are of sufficient chemical reactivity to not require GSTs for GSH adduct formation. Additionally, this suggests that dGSH may be of use in probing mechanistic questions, i.e., involvement of GST vs direct chemical reaction in the formation of GSH adducts. Troglitazone, diclofenac, clozapine, and acetaminophen are drugs that have been reported to form GSH adducts in in vitro incubations with microsomes in the presence of NADPH and GSH (24, 26-29, 37, 38). Troglitazone, an antidiabetic agent launched in 1997, was associated with idiosyncratic but severe hepatotoxicity and was withdrawn from the U.S. market in March 2000 (39, 40). Diclofenac is a widely used nonsteroidal antiinflammatory drug (NSAID) that is associated with extremely low incidences of hepatic injuries (6-18 cases/100000 person years) (41). Clozapine, a dibenzodiazepine antipsychotic, is associated with idiosyncratic agranulocytosis with a relatively highly incidence (0.8%) (42). Aceta-
minophen, one of the most widely used NSAIDs, is hepatotoxic in experimental animals at high doses, and its hepatotoxicity in man is often associated with overdose (38). The formation of reactive metabolites of all these drugs has been reported with in vitro systems, and at least in the case of acetaminophen, CYP-mediated metabolism leading to reactive metabolite formation is prerequisite for liver toxicity in experimental animals. R-(+)-Pulegone, bromobenzene, and precocene I are all hepatotoxic agents shown to form GSH adducts via CYPmediated bioactivation (22, 23, 30, 32). All seven of these compounds were evaluated for dGSH adduct formation, and dGSH adducts were observed with all compounds in incubations with HLM. The observed adduct of acetaminophen had a mass consistent with dGSH addition to the quinoneimine intermediate. This adduct was not well-separated from dGSH in the HPLC fluorescence chromatogram but was clearly identifed by MS. The major adduct of troglitazone, however, was separated by almost 20 min from the dGSH peak in its respective HPLC chromatogram. Considering the screening nature
902
Chem. Res. Toxicol., Vol. 18, No. 5, 2005
Figure 4. Reaction of CDNB with GSH and dGSH. The CDNB concentration at each time point is the average of two determinations.
Figure 5. Formation of R-(+)-pulegone adducts at varying concentrations of GSH or dGSH when incubated with HLM supplemented with NADPH. R-(+)-Pulegone adduct concentrations are the average of two determinations.
of this assay and the generally larger molecular weight and lipophilicity of today’s pharmaceutical candidates, most adducts in a discovery screen should be adequately separated from the probe by using this generic separation method. To test the level of adduct detected for compounds not reported to form GSH adducts, seven widely prescribed drugs in the current U.S. market were tested in this system. None of these compounds are associated with any significant adverse events thought to be due to reactive metabolites. The lack of adduct formation was expected based on their safety profile, and these studies served as a method of further validation to test the level of false positives in this system. None of these drugs showed any signal for adduct formation. In summary, this manuscript describes a new method for the in vitro trapping of thiol reactive metabolites. The attachment of a fluorescent group to GSH allows detection and quantitation of adduct upon HPLC fluorescence analysis. Although methods exist for detection and quantitation of thiol trappable reactive metabolites, none provide the combination of quantitative capability, cost effectiveness, and ease of implementation provided by the dGSH assay. Also, this method does not involve the handling or disposal of radioactivity. However, this method does have its limitations. Because it is fluorescence-based, it is not applicable for compounds that cause fluorescence interference, unless an adduct standard is
Gan et al.
readily available. Although no difference in the amount of adduct formation was observed in incubations with R-(+)-pulegone when comparing dGSH vs GSH incubations, it is impossible to prove that dGSH will have identical reactivity to GSH toward all other reactive metabolites. dGSH was not a cofactor for GST-mediated conjugation of DCNB and is not expected to serve as a cofactor for other GST-mediated conjugation reactions. These facts should be taken into consideration when comparing results from incubations with GSH. Nevertheless, the sensitive, specific, and quantitative nature of this method makes it attractive for the screening of preclinical compounds for reactive metabolite formation. Because idiosyncratic toxicities generally surface only late in the development of a compound, they pose significant health risk to patients and can lead to late stage clinical trial failures or market withdrawal. While the direct causative links between reactive metabolite formation and idiosyncratic toxicities have not been established, it is plausible to suggest that increased levels of reactive metabolite formation present an increased risk for adverse effects. The dGSH-based assay provides a method to quantitatively evaluate reactive metabolite formation early in the drug discovery process. The method can be used to evaluate the level of thiol adduct formed by compounds from distinct chemotypes or within a given chemotype. Similar to other liability screens employed during drug discovery, selection of acceptance/ rejection criteria for compounds in this assay must be balanced against other compound selection criteria and will be dependent on a number of factors, such as therapeutic area, anticipated dosing regimen, and project human dose.
Acknowledgment. We thank Hao Zhang, Jinsong Xing, Dauh-Rurng Wu, and Bing He for their contributions in dGSH preparation, purification, and characterization. Part of this work was presented at the 7th International ISSX Meeting. Supporting Information Available: Blank chromatogram, comparison of troglitazone adduct profiles with or without the addition of DTT in the quenching solution, and CYP inhibition profiles of dGSH. This material is available free of charge via the Internet at http://pubs.acs.org.
References (1) 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. (2) Park, B. K., Kitteringham, N. R., Powell, H., and Pirmohamed, M. (2000) Advances in molecular toxicology-towards understanding idiosyncratic drug toxicity. Toxicology 153, 39-60. (3) Li, A. P. (2002) A review of the common properties of drugs with idiosyncratic hepatotoxicity and the “multiple determinant hypothesis” for the manifestation of idiosyncratic drug toxicity. Chem.-Biol. Interact. 142, 7-23. (4) Park, B. K., Kitteringham, N. R., Maggs, J. L., Pirmohamed, M., and Williams, D. P. (2005) The role of metabolic activation in drug-induced hepatotoxicity. Annu. Rev. Pharmacol. Toxicol. 45, 177-202. (5) Roth, R. A., Luyendyk, J. P., Maddox, J. F., and Ganey, P. E. (2003) Inflammation and drug idiosyncrasysIs there a connection? J. Pharmacol. Exp. Ther. 307, 1-8. (6) Park, B. K., Naisbitt, D. J., Gordon, S. F., Kitteringham, N. R., and Pirmohamed, M. (2001) Metabolic activation in drug allergies. Toxicology 158, 11-23. (7) Evans, D. C., Watt, A. P., Nicoll-Griffith, D. A., and Baillie, T. A. (2004) Drug-protein adducts: An industry perspective on minimizing the potential for drug bioactivation in drug discovery and development. Chem. Res. Toxicol. 17, 3-16.
Dansyl Glutathione as a Trapping Agent (8) Chen, W. G., Zhang, C., Avery, M. J., and Fouda, H. G. (2001) Reactive metabolite screen for reducing candidate attrition in drug discovery. Adv. Exp. Med. Biol. 500, 521-524. (9) Yan, Z., and Caldwell, G. W. (2004) Stable-isotope trapping and high-throughput screenings of reactive metabolites using the isotope MS signature. Anal. Chem. 76, 6835-6847. (10) Boelsterli, U. A. (1993) Specific targets of covalent drug-protein interactions in hepatocytes and their toxicological significance in drug-induced liver injury. Drug Metab. Rev. 25, 395-451. (11) Samuel, K., Yin, W., Stearns, R. A., Tang, Y. S., Chaudhary, A. G., Jewell, J. P., Lanza, T., Jr., Lin, L. S., Hagmann, W. K., Evans, D. C., and Kumar, S. (2003) Addressing the metabolic activation potential of new leads in drug discovery: A case study using ion trap mass spectrometry and tritium labeling techniques. J. Mass Spectrom. 38, 211-221. (12) Davis, W., Jr., Ronai, Z. E., and Tew, K. D. (2001) Cellular Thiols and Reactive Oxygen Species in Drug-Induced Apoptosis. J. Pharmacol. Exp. Ther. 296, 1-6. (13) Meister, A., and Anderson, M. E. (1983) Glutathione. Annu. Rev. Biochem. 52, 711-760. (14) Jean, P. A., and Roth, R. A. (1995) Naphthylisothiocyanate disposition in bile and its relationship to liver glutathione and toxicity. Biochem. Pharmacol. 50, 1469-1474. (15) Nash, J. A., King, L. J., Lock, E. A., and Green, T. (1984) The metabolism and disposition of hexachloro-1:3-butadiene in the rat and its relevance to nephrotoxicity. Toxicol. Appl. Pharmacol. 73, 124-137. (16) Dekant, W., Metzler, M., and Henschler, D. (1986) Identification of S-1,2,2-trichlorovinyl-N-acetylcysteine as a urinary metabolite of tetrachloroethylene: Bioactivation through glutathione conjugation as a possible explanation of its nephrocarcinogenicity. J. Biochem. Toxicol. 1, 57-72. (17) Mutlib, A. E., Gerson, R. J., Meunier, P. C., Haley, P. J., Chen, H., Gan, L. S., Davies, M. H., Gemzik, B., Christ, D. D., Krahn, D. F., Markwalder, J. A., Seitz, S. P., Robertson, R. T., and Miwa, G. T. (2000) The species-dependent metabolism of efavirenz produces a nephrotoxic glutathione conjugate in rats. Toxicol. Appl. Pharmacol. 169, 102-113. (18) Soglia, J. R., Harriman, S. P., Zhao, S., Barberia, J., Cole, M. J., Boyd, J. G., and Contillo, L. G. (2004) The development of a higher throughput reactive intermediate screening assay incorporating micro-bore liquid chromatography-micro-electrospray ionizationtandem mass spectrometry and glutathione ethyl ester as an in vitro conjugating agent. J. Pharm. Biomed. Anal. 36, 105-116. (19) Thompson, D. C., Perera, K., and London, R. (1995) Quinone methide formation from para isomers of methylphenol (cresol), ethylphenol, and isopropylphenol: Relationship to toxicity. Chem. Res. Toxicol. 8, 55-60. (20) Jones, D. P., Carlson, J. L., Samiec, P. S., Sternberg, P., Jr., Mody, V. C., Jr., Reed, R. L., and Brown, L. A. (1998) Glutathione measurement in human plasma. Evaluation of sample collection, storage and derivatization conditions for analysis of dansyl derivatives by HPLC. Clin. Chim. Acta 275, 175-184. (21) Hammermeister, D. E., Serrano, J., Schmieder, P., and Kuehl, D. W. (2000) Characterization of dansylated glutathione, glutathione disulfide, cysteine and cystine by narrow bore liquid chromatography/electrospray ionization mass spectrometry. Rapid Commun. Mass Spectrom. 14, 503-508. (22) Nelson, S. D. (1995) Mechanisms of the formation and disposition of reactive metabolites that can cause acute liver injury. Drug Metab. Rev. 27, 147-177. (23) Khojasteh-Bakht, S. C., Chen, W., Koenigs, L. L., Peter, R. M., and Nelson, S. D. (1999) Metabolism of (R)-(+)-pulegone and (R)-(+)-menthofuran by human liver cytochrome P-450s: Evidence for formation of a furan epoxide. Drug Metab. Dispos. 27, 574-580. (24) Maggs, J. L., Williams, D., Pirmohamed, M., and Park, B. K. (1995) The metabolic formation of reactive intermediates from clozapine, a drug associated with agranulocytosis in man. J. Pharmacol. Exp. Ther. 275, 1463-1475.
Chem. Res. Toxicol., Vol. 18, No. 5, 2005 903 (25) Liu, Z. C., and Uetrecht, J. P. (1995) Clozapine is oxidized by activated human neutrophils to a reactive nitrenium ion that irreversibly binds to the cells. J. Pharmacol. Exp. Ther. 275, 1476-1483. (26) Tang, W., Stearns, R. A., Bandiera, S. M., Zhang, Y., Raab, C., Braun, M. P., Dean, D. C., Pang, J., Leung, K. H., Doss, G. A., Strauss, J. R., Kwei, G. Y., Rushmore, T. H., Chiu, S.-H. L., and Baillie, T. A. (1999) Studies on cytochrome P-450-mediated bioactivation of diclofenac in rats and in human hepatocytes: Identification of glutathione conjugated metabolites. Drug Metab. Dispos. 27, 365-372. (27) Kassahun, K., Pearson, P. G., Tang, W., McIntosh, I., Leung, K., Elmore, C., Dean, D., Wang, R., Doss, G., and Baillie, T. A. (2001) Studies on the metabolism of troglitazone to reactive intermediates in vitro and in vivo. Evidence for novel biotransformation pathways involving quinone methide formation and thiazolidinedione ring scission. Chem. Res. Toxicol. 14, 62-70. (28) He, K., Talaat, R. E., Pool, W. F., Reily, M. D., Reed, J. E., Bridges, A. J., and Woolf, T. F. (2004) Metabolic activation of troglitazone: Identification of a reactive metabolite and mechanisms involved. Drug Metab. Dispos. 32, 639-646. (29) Tettey, J. N., Maggs, J. L., Rapeport, W. G., Pirmohamed, M., and Park, B. K. (2001) Enzyme-induction dependent bioactivation of troglitazone and troglitazone quinone in vivo. Chem. Res. Toxicol. 14, 965-974. (30) Buben, J. A., Narasimhan, N., and Hanzlik, R. P. (1988) Effects of chemical and enzymic probes on microsomal covalent binding of bromobenzene and derivatives. Evidence for quinones as reactive metabolites. Xenobiotica 18, 501-510. (31) Madhu, C., Klaassen, C. D., Buben, J. A., Narasimhan, N., and Hanzlik, R. P. (1992) Bromobenzene-glutathione excretion into bile reflects toxic activation of bromobenzene in rats. Toxicol. Lett. 60, 227-236. (32) Ravindranath, V., Boyd, M. R., and Jerina, D. M. (1987) Hepatotoxicity of precocene I in rats. Role of metabolic activation in vivo. Biochem. Pharmacol. 36, 441-446. (33) Sanger, F. (1988) Sequences, sequences, and sequences. Ann. Rev. Biochem. 57, 1-29. (34) Habig, W. H., Pabst, M. J., and Jakoby, W. B. (1974) Glutathione S-transferases. The first enzymatic step in mercapturic acid formation. J. Biol. Chem. 249, 7130-7139. (35) Rinaldi, R., Eliasson, E., Swedmark, S., and Morgenstern, R. (2002) Reactive intermediates and the dynamics of glutathione transferases. Drug Metab. Dispos. 30, 1053-1058. (36) Armstrong, R. N. (1997) Structure, catalytic mechanism, and evolution of the glutathione transferases. Chem. Res. Toxicol. 10, 2-18. (37) Gardner, I., Zahid, N., MacCrimmon, D., and Uetrecht, J. P. (1998) A comparison of the oxidation of clozapine and olanzapine to reactive metabolites and the toxicity of these metabolites to human leukocytes. Mol. Pharmacol. 53, 991-998. (38) James, L. P., Mayeux, P. R., and Hinson, J. A. (2003) Acetaminophen-induced hepatotoxicity. Drug Metab. Dispos. 31, 14991506. (39) Smith, M. T. (2003) Mechanisms of troglitazone hepatotoxicity. Chem. Res. Toxicol. 16, 679-687. (40) Graham, D. J., Drinkard, C. R., and Shatin, D. (2003) Incidence of idiopathic acute liver failure and hospitalized liver injury in patients treated with troglitazone. Am. J. Gastroenterol. 98, 175179. (41) Boelsterli, U. A. (2003) Diclofenac-induced liver injury: A paradigm of idiosyncratic drug toxicity. Toxicol. Appl. Pharmacol. 192, 307-322. (42) Guest, I., Sokoluk, B., MacCrimmon, J., and Uetrecht, J. (1998) Examination of possible toxic and immune mechanisms of clozapine-induced agranulocytosis. Toxicology 131, 53-65.
TX0496791