Bioactivation of a Toxic Metabolite of Valproic Acid - American

Chem. Res. Toxicol. 1996, 9, 517-526

517

Bioactivation of a Toxic Metabolite of Valproic Acid, (E)-2-Propyl-2,4-pentadienoic Acid, via Glucuronidation. LC/MS/MS Characterization of the GSH-Glucuronide Diconjugates† Wei Tang and Frank S. Abbott* Division of Pharmaceutical Chemistry, Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z3 Received July 5, 1995X

The hepatotoxicity of the anticonvulsant drug valproic acid may be associated with the formation of potentially reactive metabolites, one of which is (E)-2-propyl-2,4-pentadienoic acid ((E)-2,4-diene VPA). This report describes the characterization of new GSH-related conjugates of this diene. Bile samples collected from male Sprague-Dawley rats dosed ip with (E)-2,4diene VPA (100 mg/kg) were analyzed by LC/MS/MS. Initial Q1 parent ion scanning indicated that the daughter ions m/z 162 and 123 could be derived from the ions at m/z 624 and 480, respectively. Subsequent collision-induced dissociation (CID) of these parent ions revealed a common neutral loss of 176 Da which is diagnostic for glucuronides. A similar neutral loss of 176 Da was observed in daughter ion spectra of the biliary metabolites arising from [2H7]-4ene VPA dosed ip to rats, where the ion fragments containing the VPA portion were 7 amu higher than those derived from the unlabeled drug. CID of the ion at m/z 624 also gave fragments characteristic for GSH conjugates such as the loss of glycine and glutamate moieties. Based on the MS data, the metabolites were assigned the diconjugate structures 1-O-(2-propyl5-(glutathion-S-yl)-3-pentenoyl)-β-D-glucuronide (5-GS-3-ene VPA-glucuronide I, MH+, 624) and the corresponding 5-NAC-3-ene VPA-glucuronide (MH+, 480). Further proof of structural identity was obtained from 1H NMR of HPLC-purified metabolites. The amount of biliary 5-GS-3-ene VPA-glucuronide I was 7-fold greater than the corresponding 5-GS-3-ene VPA, the sum of the two metabolites accounting for 6.6% of the dose. Incubation of 1-O-(2-propyl-2,4pentadienoyl)-β-D-glucuronide (2,4-diene VPA-glucuronide) with GSH in the presence or absence of GST enzyme led to the formation of 5-GS-3-ene VPA-glucuronide I which was readily detected by LC/MS/MS, suggesting that in vivo the diconjugate may arise from the reaction of GSH with 2,4-diene VPA-glucuronide. To our knowledge, this is the first recorded instance in which glucuronide formation activates a drug to further conjugate with GSH via a Michael addition reaction.

Introduction Valproic acid (VPA, 2-propylpentanoic acid)1 is a unique anticonvulsant having a branched-chain fatty acid structure. Despite its simple chemistry, the metabolism of VPA is quite complex. Mitochondrial β-oxidation and microsomal cytochrome P450 (P450)-mediated hydroxy† A preliminary account of this study was presented at the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 1995. * Address correspondence and reprint requests to this author at the Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver, BC, Canada V6T 1Z3. X Abstract published in Advance ACS Abstracts, February 1, 1996. 1 Abbreviations: VPA, valproic acid, 2-propylpentanoic acid; P450, cytochrome P450; 4-ene VPA, 2-propyl-4-pentenoic acid; (E)-2,4-diene VPA, (E)-2-propyl-2,4-pentadienoic acid; (E)-2-ene VPA, (E)-2-propyl2-pentenoic acid; 3-keto-4-ene VPA, 2-propyl-3-oxo-4-pentenoic acid; GST, glutathione S-transferase; NAC, N-acetyl-L-cysteine; TFA, trifluoroacetic acid; CDNB, 1-chloro-2,4-dinitrobenzene; 2,4-diene VPAglucuronide, 1-O-(2-propyl-2,4-pentadienoyl)-β-D-glucuronide; 5-GS-3ene VPA, 2-propyl-5-(glutathion-S-yl)-3-pentenoic acid; 5-GS-2-ene VPA, 2-propyl-5-(glutathion-S-yl)-2-pentenoic acid; 5-GS-3-ene VPAglucuronide I, 1-O-(2-propyl-5-(glutathion-S-yl)-3-pentenoyl)-β-D-glucuronide; 5-GS-3-ene VPA-glucuronide II, 2-O-(2-propyl-5-(glutathionS-yl)-3-pentenoyl)-β-D-glucuronide; 5-GS-2-ene VPA-glucuronide, 1-O(2-propyl-5-(glutathion-S-yl)-2-pentenoyl)-β-D-glucuronide; 5-GS-2fluoro-4-hydroxy VPA lactone, 2-propyl-5-(glutathion-S-yl)-2-fluoro-4hydroxypentanoic acid lactone; LC/MS/MS, combined liquid chromatography/tandem mass spectrometry; MRM, multiple-reaction monitoring; CID, collision-induced dissociation.

0893-228x/96/2709-0517$12.00/0

lation/dehydrogenation of the side chains provide the major phase I biotransformation pathways of the drug (1). Glucuronidation is the principal phase II metabolic pathway of VPA and of many of its phase I metabolites due to the presence of the carboxylate functional group. The excretion of VPA glucuronide typically accounts for ∼50% of the dose in rats (2) and ∼30% in patients receiving chronic VPA therapy (3). Studies of VPA metabolism have been stimulated by the hypothesis that the rare but fatal hepatotoxicity of VPA results from the formation of reactive, toxic metabolite(s) (4). The VPA-associated toxicity in patients is characterized by liver microvesicular steatosis frequently associated with necrosis (5). The pathological feature of steatosis can be reproduced in rats by chronic administration of VPA metabolites, namely, 2-propyl-4pentenoic acid (4-ene VPA) and (E)-2-propyl-2,4-pentadienoic acid ((E)-2,4-diene VPA) (6). When tested in vitro, 4-ene VPA was cytotoxic to rat hepatocytes (7) and was an effective inhibitor of mitochondrial β-oxidation in rat liver preparations (8, 9). Radiolabeled 4-ene VPA was seen to bind covalently to both rat liver proteins (10) and P450 enzymes (11). Evidence for a reactive form of 4-ene VPA was obtained in this laboratory when the GSH and N-acetylcysteine (NAC) conjugates of (E)-2,4-diene VPA were detected in the bile and urine, respectively, of rats © 1996 American Chemical Society

518 Chem. Res. Toxicol., Vol. 9, No. 2, 1996

dosed with either (E)-2,4-diene VPA or 4-ene VPA (12). The NAC conjugate was also detected in the urine of patients on VPA therapy with the levels being 3-4 times higher than normal in two patients who had developed VPA-induced liver failure (12). The reaction of GSH with the 4-ene VPA metabolite, (E)-2,4-diene VPA, in mitochondria could produce a localized depletion of GSH that would result in oxidative stress with accompanying hepatocellular damage. Alternatively (E)-2,4-diene VPA may eventually form 2-propyl-3-oxo-4-pentenoic acid (3keto-4-ene VPA), a putative inhibitor of certain β-oxidation enzymes (4). Further support for the hypothesis that VPA-associated hepatotoxicity is mediated by metabolic bioactivation was obtained by comparing 4-ene VPA with its R-fluorinated analogue (R-fluoro-4-ene VPA) for their capabilities of inducing liver microvesicular steatosis in rats. In sharp contrast to more than 85% of hepatocytes affected by steatosis in 4-ene VPA-treated rats, lipid droplets were hardly detected in the liver of animals administered R-fluoro-4-ene VPA (13). Formation of (E)2,4-diene VPA and the NAC conjugate of (E)-2,4-diene VPA was also selective for 4-ene VPA (13), the latter being an indication for the in vivo reaction of (E)-2,4diene VPA with GSH. Taken together, the evidence suggests that the metabolism of 4-ene VPA to (E)-2,4diene VPA is essential for expression of the 4-ene VPA hepatotoxicity. Recently, a number of GSH conjugates were identified in the bile of rats treated with 4-ene VPA, including 2-propyl-5-(glutathion-S-yl)-4-hydroxypentanoic acid lactone (14), 2-propyl-5-(glutathion-S-yl)-3-pentenoic acid (5GS-3-ene VPA) (12, 14), and 2-propyl-3-oxo-5-(glutathionS-yl)pentanoic acid (14) derived from the reaction of GSH with 2-propyl-4,5-epoxypentanoic acid (14), (E)-2,4-diene VPA (12, 14), and 3-keto-4-ene VPA (14), respectively. The conjugation of GSH with (E)-2,4-diene VPA was thought to be mediated by coenzyme A (CoA); i.e., the double bonds of (E)-2,4-diene VPA are activated via the formation of the corresponding CoA thioester to react with GSH through a Michael addition reaction (12, 14). By analogy, other esters of (E)-2,4-diene VPA could presumably initiate in vivo addition of GSH. Thus, in our continuing effort to explore mechanisms of VPAassociated hepatotoxicity, we report herein the LC/MS/ MS characterization of new GSH-glucuronide and NACglucuronide diconjugates of (E)-2,4-diene VPA in rats treated with (E)-2,4-ene VPA. Data are presented to support the hypothesis that (E)-2,4-diene VPA can also be bioactivated via glucuronidation.

Materials and Methods Chemicals. β-Glucuronidase, GSH, trifluoroacetic acid (TFA), and urethane were purchased from Sigma Chemical Co. (St. Louis, MO). 4-(Dimethylamino)pyridine, 2,2,2-trifluoroethanol, dicyclohexylcarbodiimide, potassium monoperoxy sulfate (Oxone), 1-chlorobutane, and 1-chloro-2,4-dinitrobenzene (CDNB) were the products of Aldrich Chemical Co. (Milwaukee, WI). Oxytocin was obtained from Calbiochem (La Jolla, CA), tris-(2carboxyethyl)phosphine from Pierce Chemical Co. (Rockford, IL), polyethylene tubing PE-10 from Clay Adams (Parsippany, NJ) and Spe octadecyl extraction column from J. T. Baker (Phillipsburg, NJ). (E)-2,4-Diene VPA (12), 2-[2H7]propyl-4-pentenoic acid ([2H7]4-ene VPA) (15) and ethyl 2-propyl-2-fluoro-4-pentenoate (13) were synthesized in this laboratory. Instrumentation and Analytical Methods. NMR spectra were obtained on a Bruker WH-400 spectrometer in the Depart-

Tang and Abbott ment of Chemistry, the University of British Columbia (UBC), and chemical shifts are expressed relevant to tetramethylsilane. LC/MS/MS experiments were carried out on a Fisons VG Quattro tandem mass spectrometer interfaced to a HewlettPackard 1090II liquid chromatograph. Positive electrospray was used as the means of ionization, and collision-induced dissociation (CID) involved argon as the target gas at a pressure of 3.0 × 10-4 mbar. Other parameters were capillary voltage 3.36 kV, core voltage 29 V with skimmer offset by 5 V, and collision energy 50 eV. The multipliers 1 and 2 were generally both set at 650 V, or as otherwise indicated. The low-mass and high-mass resolutions were set at 5.5 for MS1 and 12.5 for MS2. The source temperature was 80 °C. HPLC was performed on a Hewlett-Packard Hypersil ODS column (100 × 2.1 mm, 5 µm), and samples were delivered at a flow rate of 50 µL/min. The mobile phase consisted of methanol/ water (0.05% TFA) and was programmed as follows. Method A: 75% water for 2 min, a gradient decrease to 50% water at 5 min, a hold at 50% water to 6 min, a gradient decrease to 15% water at 8 min, and a hold at 15% water to 30 min. Method B: 75% water for 3 min, a gradient decrease to 50% water at 6 min, a hold at 50% water to 13 min, a gradient decrease to 15% water at 14 min, and a hold at 15% water to 30 min. Method C: a gradient decrease of water from 75% at 0 min to 50% at 15 min, a hold at 50% water to 20 min, a gradient decrease to 15% water at 22 min, and a hold at 15% water to 45 min. Qualitative GC/MS analysis was performed on a HewlettPackard HP5700A gas chromatograph coupled with a MAT-111 mass spectrometer. The GC was fitted with a glass column packed with 3% Dexsil 300 on 100/200 mesh Supelcoport (1.8 m × 2 mm; oven temperature 32 °C/min from 50 to 300 °C). The mass spectrometer was operated using electron impact ionization at an energy of 70 eV and emission current of 300 mA. HPLC purification of synthetic compounds was performed on a Whatman Partisil ODS2 column (250 × 9 mm, 5 µm) and the isolation of biliary metabolites on a Hewlett-Packard Spherisorb ODS2 column (250 × 4 mm, 5 µm) using a Hewlett-Packard 1050 liquid chromatograph with UV detection at 210 nm. The mobile phase consisted of acetonitrile/water (0.05% TFA) as detailed in the following sections. Chemical Synthesis. Synthesis of 2-Propyl-5-(glutathion-S-yl)-3-pentenoic Acid (5-GS-3-ene VPA). To (E)2,4-diene VPA (5.7 mmol, 0.8 g) in dry methylene chloride (5 mL) was added (4-dimethylamino)pyridine (0.64 mmol, 78 mg) and 2,2,2-trifluoroethanol (5.9 mmol, 0.59 g). The mixture was cooled to 0 °C, and dicyclohexylcarbodiimide (7.1 mmol, 1.5 g) was added. The mixture was stirred at room temperature for 24 h followed by consecutive washes with aqueous hydrochloric acid solution (1.0 M), saturated sodium bicarbonate solution, and water. Trifluoroethyl (E)-2-propyl-2,4-pentadienoate was obtained through distillation at 68 °C (2.5 mmHg). GC/MS mass spectrum: m/z (%) 95 (100), 222 (M+, 42), 123 (32), 165 (17), 193 (12), 179 (8), 207 (5). To an aqueous solution of GSH (0.16 mmol or 50 mg/5 mL, pH 10.5) was added trifluoroethyl (E)-2-propyl-2,4-pentadienoate (0.45 mmol, 100 mg) in acetonitrile (2 mL). The mixture was stirred at room temperature for 24 h, acidified to pH 2.5, and washed with ethyl acetate. The volume of the aqueous phase was reduced in vacuo to 200 µL and crude product purified by HPLC: acetonitrile/water, 15/85 (0.05% TFA); flow rate, 2.5 mL/min; tR, 10.5 min. LC/MS/MS mass spectrum: m/z (%) 448 (MH+, 100), 162 (51), 216 (31), 319 (21), 256 (18), 373 (9). 1H NMR (D2O): δ 0.9 (t, 3H, JHH ) 7 Hz, CH2CH3), 1.1-1.4 (m, CH2CH3), 1.4-1.7 (m, 2H, dCCH2), 2.2 (q, 2H, JHH ) 7 Hz, Glu, CHCH2), 2.5 (t, 2H, JHH ) 7 Hz, Glu CH2CO), 2.7-2.9 (m, 2H, Cys CH2), 3.0-3.2 (m, 3H, SCH2CHd, CHCO), 3.9 (t, H, JHH ) 6 Hz, Glu CH), 4.0 (s, 2H, Gly CH2), 4.5 (dd, 1H, JHH ) 6 Hz, Cys CH), 5.5-5.7 (m, 2H, CHdCH). Synthesis of 2-Propyl-2-fluoro-5-(glutathion-S-yl)-4-hydroxypentanoic Acid Lactone (5-GS-2-Fluoro-4-hydroxy VPA Lactone). To sodium bicarbonate (34 mmol, 2.88 g),

Bioactivation of 2,4-Diene VPA via Glucuronidation acetone (12 mL), and water (6 mL) in a distillation apparatus with the receiver immersed in dry ice/acetone bath was added potassium monoperoxy sulfate (10 mmol, 6 g). The reaction mixture was stirred vigorously at room temperature and a vacuum (water aspirator) applied. The effluent was collected into a receiver immersed in dry ice/acetone bath as a yellow solution consisting of dimethyldioxirane and acetone (16). To the solution of dimethyldioxirane prepared above was added ethyl 2-propyl-2-fluoro-4-pentenoate (0.6 mmol, 100 mg). The mixture was stirred at room temperature for 18 h and ethyl 2-propyl-2-fluoro-4,5-epoxypentanoate obtained upon removal of solvent in vacuo. GC/MS mass spectrum: m/z (%) 91 (100), 131 (93), 148 (93), 184 (14), 204 (M+, 14). To an aqueous solution of GSH (0.31 mmol or 50 mg/5 mL, pH 9.5) was added ethyl 2-propyl-2-fluoro-4,5-epoxypentanoate in acetonitrile (2 mL). The mixture was stirred at room temperature for 24 h, washed with diethyl ether, acidified to pH 2.5, and extracted with ethyl acetate three times. The volume of the aqueous phase was reduced in vacuo to 200 µL and crude product purified by HPLC: acetonitrile/water, 13/87 (0.05% TFA); flow rate, 2.5 mL/min; tR, 10 min. LC/MS/MS mass spectrum: m/z (%) 466 (MH+, 100), 337 (28), 234 (26), 391 (9), 320 (9), 177 (9). 1H NMR (D2O): δ 0.9 (t, 3H, JHH ) 7 Hz, CH2CH3), 1.3-1.5 (m, 2H, CH2CH3), 1.7-2.1 (m, 2H, CFCH2), 2.2 (q, 2H, JHH ) 7 Hz, Glu CHCH2), 2.6 (t, 2H, JHH ) 7 Hz, Glu CH2CO), 2.8-3.2 (m, 6H, Cys CH2, SCH2, CH2CFCO), 3.9 (s, 2H, Gly CH2), 4.0 (t, 1H, JHH ) 6 Hz, Glu CH), 4.55 (m, 1H, Cys CH), 4.9 (m, 1H, OCH). Animal Experiments. Male Sprague-Dawley rats (Vancouver, BC) weighing 230-280 g (mean 250 g) were allowed free access to food (Purina Laboratory Chow) and water. They were housed in regular cages and exposed to a controlled 12 h cycle of light and darkness. Three rats were anesthetized with urethane (1 g/kg) and their bile ducts cannulated with PE-10 tubing. Control bile was collected for 15 min. An aqueous solution (pH 7) of (E)-2,4-diene VPA was then administered at 100 mg/kg by ip injection and bile collected for an additional 6 h. One rat was treated the same as above but dosed with [2H7]4-ene VPA in an aqueous solution (pH 7) at 100 mg/kg. Three rats, after control urine was collected, were administered the aqueous solution of (E)-2,4-ene VPA (pH 7) at 100 mg/ kg by ip injection and housed in metabolic cages to collect urine for 24 h. Isolation of 1-O-(2-Propyl-2,4-pentadienoyl)-β-D-glucuronide (2,4-Diene VPA-glucuronide) from Rat Bile. Purification of 2,4-diene VPA-glucuronide was carried out according to the procedure of Willliams et al. for VPA-glucuronide (17) with modifications. Briefly, bile (4 mL) collected from rats treated with (E)-2,4-diene VPA was acidified to pH 2, washed with 1-chlorobutane (3 × 12 mL), and further extracted by diethyl ether (3 × 12 mL). The combined ether extracts were evaporated to dryness under nitrogen. The residue was dissolved in a minimum amount of methanol and subjected to purification by flash chromatography (silica gel 60, 230-400 mesh). The column was 120 × 2.5 mm, and the eluting solvents consisted initially of ether (100%) followed by ether/methanol (60/30, v/v). LC/MS/MS mass spectrum, m/z (%): 141 (100), 317 (76, MH+), 123 (10), 159 (7). 1H NMR (CD3OD): δ 0.9 (t, 3H, JHH ) 7 Hz, CH3), 1.5 (sextet, 2H, JHH ) 7 Hz, CH2CH3), 2.3 (t, 2H, JHH ) 7 Hz, dCCH2), 3.3-3.8 (m, 4H, glucuronic acid 4CH), 5.4-5.8 (m, 3H, CHdCH2; glucuronic acid CH), 6.7-6.9 (m, 1H, CHdCH2), 7.4 (d, 1H, JHH ) 11 Hz, OCCdCH). Detection of the Biliary GSH-Glucuronide, and the Biliary and Urinary NAC-Glucuronide Diconjugates of (E)-2,4-Diene VPA. Rat bile (100 µL) was mixed with an equivalent volume of aqueous TFA (0.05%) and the precipitate removed via centrifugation at 13600g for 15 min. Rat urine (500 µL) was acidified to pH 3 and applied to a C18 extraction cartridge which was prewashed with methanol and water. The column was consecutively washed with water and methanol. The methanol eluate was evaporated in vacuo to dryness and the residue reconstituted in aqueous TFA (0.05%, 100 µL). An aliquot of the bile or urine samples (2-20 µL) was injected onto

Chem. Res. Toxicol., Vol. 9, No. 2, 1996 519 the Hewlett-Packard Hypersil ODS column and eluted using LC method A. To record full daughter ion spectra of the metabolites, MS/ MS dwell times were adjusted to provide a scan rate of ∼1 s/100 amu. When the MS/MS was operated in the multiple-reaction monitoring (MRM) mode, three transitions, m/z 480 f 304, 480 f 286, and 480 f 123 were used as criteria for the detection of the urinary NAC-glucuronide diconjugate. Isolation of the GSH-Glucuronide Diconjugates of (E)2,4-Diene VPA from Rat Bile. The bile (4 mL) collected from rats administered (E)-2,4-diene VPA was acidified to pH 2, washed consecutively with 1-chlorobutane (3 × 12 mL) and diethyl ether (3 × 12 mL), and evaporated in vacuo to dryness. The residue was dissolved in water (200 µL) and subjected to purification by HPLC: acetonitrile/water, 13/87 (0.05% TFA); flow rate, 1.0 mL/min. Two fractions were collected. LC fraction 1 (tR ) 7.7 min): 2-O-(2-propyl-5-(glutathion-Syl)-3-pentenoyl)-β-D-glucuronide (5-GS-3-ene VPA-glucuronide II). LC/MS/MS mass spectrum, m/z (%): 624 (MH+, 100), 123 (17), 299 (14), 141 (9), 162 (9), 495 (8), 549 (5). 1H NMR (D2O): δ 0.85 (t, 3H, JHH ) 7 Hz, CH2CH3), 1.2-1.35 (m, CH2CH3), 1.5-1.8 (2m, 2H, CHCH2), 2.2 (m, 2H, Glu, CHCH2), 2.55 (m, 2H, Glu CH2CO), 2.7-2.95 (2m, 2H, Cys CH2), 3.1-3.3 (m, 3H, SCH2CHd, CHCO), 3.45-3.65 (m, 3H, glucuronic acid 3CH), 3.9 (m, 3H, Glu CH, Gly CH2), 4.1 (d, 1H, glucuronic acid CH), 4.5 (dd, 1H, JHH ) 6 Hz, Cys CH), 5.5-5.7 (m, 3H, glucuronic acid C1H; CHdCH). 1-O-(2-Propyl-5-(glutathion-S-yl)-2-pentenoyl)-β-D-glucuronide (5-GS-2-ene VPA-glucuronide) (present as an impurity in LC fraction 1). LC/MS/MS mass spectrum, m/z (%): 624 (MH+, 100), 448 (28), 226 (13), 123 (12), 319 (8), 373 (5). 1H NMR (D2O): δ 1.4 (sextet, 2H, JHH ) 7 Hz, CH2CH3), 2.3 (t, 2H, JHH ) 7 Hz, SCH2CH2), 6.96 (t, 1H, JHH ) 7 Hz, CdCH). Other signals are likely superimposed with those of 5-GS-3-ene VPA-glucuronide II. LC fraction 2 (tR ) 9.9 min): 1-O-(2-propyl-5-(glutathion-Syl)-3-pentenoyl)-β-D-glucuronide (5-GS-3-ene VPA-glucuronide I). LC/MS/MS mass spectrum, m/z (%): 624 (MH+, 100), 448 (32), 319 (20), 162 (14), 216 (11), 256 (6). 1H NMR (D2O): δ 0.85 (t, 3H, JHH ) 7 Hz, CH2CH3), 1.2-1.35 (m, CH2CH3), 1.51.8 (2m, 2H, CHCH2), 2.2 (m, 2H, Glu, CHCH2), 2.55 (m, 2H, Glu CH2CO), 2.7-2.95 (2m, 2H, Cys CH2), 3.1-3.3 (m, 3H, SCH2CHd, CHCO), 3.45-3.65 (m, 3H, glucuronic acid 3CH), 3.9 (m, 3H, Glu CH, Gly CH2), 4.1 (d, 1H, glucuronic acid CH), 4.5 (dd, 1H, JHH ) 6 Hz, Cys CH), 5.5-5.7 (d + m, 3H, glucuronic acid CH; CHdCH). Quantitation of the Biliary GSH Conjugate and the GSH-Glucuronide Diconjugate of (E)-2,4-Diene VPA. An aliquot of bile sample (100 µL) was mixed with an equivalent volume of aqueous TFA (0.05%) containing 5-GS-2-fluoro-4hydroxy VPA lactone (internal standard, 17 µg/mL) and the precipitate removed by centrifugation at 13600g for 15 min. The supernatant was subjected to LC/MS/MS analysis. The LC method B and MS/MS MRM of transition m/z 448 f 162 were employed for the selective detection of 5-GS-3-ene VPA while the LC method C and MRM of transition m/z 624 f 448 were used for 5-GS-3-ene VPA-glucuronide I. The dwell times were set at 2s. Standard curves (r2 > 0.99) covered ranges of 1 to 15 µg/mL and 3.125 to 50 µg/mL for the mono-conjugate and diconjugate, respectively. In Vitro Formation of the GSH-Glucuronide Diconjugates of (E)-2,4-Diene VPA. Cytosolic fractions isolated from the livers of naive rats were used as the source of GST enzymes. Briefly, after separation of the mitochondrial pellet, the supernatant was centrifuged at 105000g for 70 min and the resulting supernatant collected as the cytosolic fraction (18). The GST enzyme activity in the cytosolic fractions was evaluated using CDNB as substrate (19). To incubations containing phosphate buffer (0.1 M, pH 6.5) and GSH (2.5 mM) in the presence or absence of rat liver cytosol (200 µL) at 25 °C was added 2,4-diene VPA-glucuronide in methanol (16 mM, 20 µL) to initiate the reaction. The final volume was 500 µL. Incubations that contained no GSH or cytosol were employed as controls. The reaction was quenched

520 Chem. Res. Toxicol., Vol. 9, No. 2, 1996 at 30 min by adding 10% aqueous TFA (50 µL) and precipitates were removed via centrifugation at 13600g for 15 min. An aliquot of the resulting supernatant (2 µL) was injected onto the Hewlett-Packard Hypersil ODS column and eluted using LC method A. MS/MS detection of the GSH-glucuronide diconjugates formed in the incubation was achieved via MRM of two transitions: m/z 624 f 448 and 624 f 319. The dwell times were set at 2 s. The β-Glucuronidase-Catalyzed Hydrolysis of the GSHGlucuronides of (E)-2,4-Diene VPA. Bile (50 µL) collected from rats treated with (E)-2,4-diene VPA was incubated with β-glucuronidase (400 units) in phosphate buffer (500 µL, pH 5.7) at 37 °C for 15 h. Incubations containing no β-glucuronidase served as control. The reaction was quenched by adding 10% aqueous TFA (50 µL) and the precipitate removed by centrifugation at 13600g for 15 min. An aliquot of the supernatant (2 µL) was injected onto the Hewlett-Packard Hypersil ODS column and eluted using LC method C. The progress of the hydrolysis was determined by following the disappearance of 5-GS-3-ene VPA I by MS/MS in MRM mode as described in the previous section. Alkylation of Reduced Oxytocin by 2,4-Diene VPAglucuronide. To oxytocin (500 µg, 0.5 µmol) in phosphate buffer (0.05 M, 0.5 mL, pH 7.7) was added tris(2-carboxyethyl)phosphine (4 mM, 500 µL) (20). The mixture was stirred at ambient temperature for 15 min. To the aqueous solution containing the reduced oxytocin was added 2,4-diene VPAglucuronide in phosphate buffer (12.5 µmol/mL, 0.5 mL, pH 7.7), and the mixture was stirred at ambient temperature for 24 h. After acidifying to pH 3.0, the reaction mixture was applied to a C18 extraction cartridge which was prewashed with methanol and water. The column was consecutively washed with water and methanol. The methanol eluate was evaporated in vacuo to dryness and the residue reconstituted in aqueous methanol (1/1, v/v, 0.1% formic acid, 100 µL). An aliquot of the sample (8 µL) was introduced by direct infusion into the ion source using aqueous methanol (1/1, v/v, 0.1% formic acid) as the mobile phase at a flow rate of 50 µL/min. To detect the alkylated oxytocin, MS was operated at a scan rate of ∼1 s/155 amu and the multipliers were set at 700 V. All data were presented as mean ( standard deviation.

Tang and Abbott

Figure 1. MS/MS CID mass spectra of 5-GS-3-ene VPAglucuronide I obtained as the biliary metabolite in (A) rats treated with (E)-2,4-diene VPA and (B) a rat dosed with [2H7]4-ene VPA. The fragmentations are discussed in the text.

Results Characterization of GSH-Glucuronide Diconjugates of (E)-2,4-Diene VPA. Previous LC/MS/MS studies of the biliary GSH conjugates of (E)-2,4-diene VPA had revealed a number of unique fragmentation patterns associated with these compounds, such as the daughter ion m/z 162 related to the GSH moiety ([HSCH2CHCOGly]+) (14, 21) and the daughter ion m/z 123 derived from the 2,4-diene VPA portion (CH2dCHCHdC(CO+)CH2CH2CH3) (21). In a continuing effort to profile the conjugated metabolites of (E)-2,4-diene VPA using LC/ MS/MS, initial Q1 scanning for parents of the ions m/z 162 and 123 directed our attention to the ions at m/z 624 and 480 that were present in the bile samples. The former matched the molecular weight of a protonated GSH-glucuronide diconjugate of (E)-2,4-diene VPA while the latter appeared to fit the corresponding NACglucuronide diconjugate. Subsequent MS/MS CID of the parent ions at either m/z 624 or 480 produced a neutral loss of 176 Da which was indicative of the elimination of the carbohydrate moiety of glucuronides (Figures 1A and 2A) (22). Following the loss of 176 Da, the daughter ion spectrum of the putative GSH-glucuronide diconjugate gave fragments characteristic of GSH conjugates. A further loss of 75 Da (glycine) gave m/z 373, and loss of 129 Da (pyroglutamate) gave m/z 319. These ions plus the daughter ion m/z 162 were evidence for the tripeptide Glu-Cys-Gly being part of the molecule (Figure 1A).

Figure 2. MS/MS CID mass spectra of 5-NAC-3-ene VPAglucuronide obtained as the biliary metabolite in (A) rats treated with (E)-2,4-diene VPA and (B) a rat dosed with [2H7]-4-ene VPA. The fragmentations are discussed in the text.

Additional informative fragments are the ions at m/z 216, which represents RSCH2CHdNH2+ (R ) 3-ene VPA) (14), and m/z 256, which possibly results from the combined neutral loss of glutamine, glucuronic acid, and carbon monoxide (Figure 1A). CID of the parent ion at m/z 480 produced a prominent fragment ion at m/z 123, which is likely to arise from the combined neutral loss of the NAC moiety, via a retroMichael reaction, plus loss of glucuronic acid (Figure 2A). The fragment at m/z 299 could result from the loss of NAC plus a molecule of water while the fragment m/z 286 corresponds to the loss of glucuronic acid (Figure 2A). Other daughter ions are possibly associated with the NAC moiety, including ions at m/z 164 ([NAC + H]+), 146 ([NAC + H - H2O]+), and 130 ([NAC + H - H2S]+). The putative NAC-glucuronide diconjugate of (E)-2,4diene VPA was detected in both the bile and urine samples (supporting information).

Bioactivation of 2,4-Diene VPA via Glucuronidation

Figure 3. MS/MS mass spectra of (A) 5-GS-3-ene VPAglucuronide II and (B) 5-GS-2-ene VPA-glucuronide in the bile of rats treated with (E)-2,4-diene VPA. The fragmentations for these compounds are discussed in the text.

Evidence for the identity of the diconjugates was also obtained from the rat dosed with [2H7]-4-ene VPA, which was expected to be metabolized to [2H7]-(E)-2,4-diene VPA in vivo. A neutral loss of 176 Da was observed in the daughter ion spectra of the biliary metabolites, where the protonated molecular and fragment ions containing the VPA portion were 7 amu higher than that seen in the bile of rats treated with unlabeled drug. Daughter ions associated with GSH and NAC moieties remained unchanged. For example, fragment ions m/z 162 (Figure 1A,B) and 146 (Figure 2A,B) were common for the two sets of parent ions, namely, m/z 624/631 and 480/487, respectively. The fragment m/z 130, which appears in the daughter ion spectrum of the parent at m/z 487 (Figure 2B), may represent a merger of two fragments, namely, the septet-deuterated VPA portion and the ion derived from [NAC + H - H2S]+. On the basis of the MS data and the similarity of the CID fragmentation to that for 5-GS-3-ene VPA (MH+, 448) (14, 21) and 2-propyl-5-(N-acetylcystein-S-yl)-3-pentenoic acid (MH+, 304) (unpublished data), the diconjugates were assigned as 5-GS-3-ene VPA-glucuronide I and 1-O-(2-propyl-5-(Nacetylcystein-S-yl)-3-pentenoyl)-β-D-glucuronide, respectively. Further scrutiny of the parent ions at m/z 624 led to the identification of a second GSH-glucuronide diconjugate of (E)-2,4-diene VPA which had a relatively shorter HPLC retention time than that of the apparent 5-GS-3ene VPA-glucuronide I. The CID mass spectrum of this compound was characterized by a greater number of lowabundance fragment ions (Figure 3B). A potential structure for this compound was an isomeric form of the diconjugate in which an intramolecular migration of the acyl moiety between adjacent hydroxyl groups on the glucuronic acid ring had occurred to give 5-GS-3-ene VPA-glucuronide II. This structural assignment was based on the observation that the neutral loss of 176 Da was no longer a dominant process while loss of the glycine and glutamate moieties, giving the fragments at m/z 549 and 495, respectively (Figure 3A), appeared to be more significant than in the case of 5-GS-3-ene VPA-glucuronide I (Figure 1A). The retro-Michael reaction resulting in the loss of GSH was an important process which,

Chem. Res. Toxicol., Vol. 9, No. 2, 1996 521

in combination with the loss of a water molecule, produced the prominent fragment ion at m/z 299 (Figure 3A). The most abundant daughter ion at m/z 123 was likely formed through the combined neutral loss of GSH and glucuronic acid (Figure 3A). Additional evidence for the structural identities of the GSH-glucuronides was obtained from 1H NMR spectra of the HPLC purified metabolites (Figure 4). For the LC fraction containing the putative 5-GS-3-ene VPA-glucuronide I (LC fraction 2), the NMR signal accounting for two protons at downfield 5.9 ppm is diagnostic of the double bond positioned between the β- and γ-carbons of the VPA portion (Figure 4, upper panel). The saturated propyl side chain can be easily identified by a triplet at 0.85 ppm (CH3), a multiplet between 1.2 and 1.4 ppm (CH2CH3), and a twin multiplet around 1.5-1.8 ppm (CHCH2). Existence of the GSH moiety is marked by a characteristic doublet of doublets at 4.5 ppm (Cys, CH), a twin doublet of doublets at 2.75 and 2.9 ppm (Cys, CH2), a singlet at 4.0 ppm (Gly, CH2), and two multiplets at 2.2 and 2.5 ppm, respectively (Glu, 2CH2). Multiplets at 3.2-3.3 and 3.5-3.6 ppm, a doublet at 4.1 ppm (HOOCCH), and a portion of the multiplets at 5.5-5.7 ppm are considered to be proton signals for the glucuronic acid ring (Figure 4, upper panel). Despite the presence of an impurity, the NMR spectrum of 5-GS-3-ene VPA-glucuronide II (LC fraction 1) (Figure 4, lower panel) is highly similar to that of 5-GS3-ene VPA-glucuronide I. Although the shapes of the peaks related to the glucuronic acid moiety (3.2-3.3 ppm) appear to be different, from the limited NMR information it is not clear at the present time to which hydroxyl group in the glucuronic acid ring the xenobiotic VPA portion is attached. The impurity in LC fraction 1 is possibly associated with the isomeric 1-O-(2-propyl-5-(glutathion-S-yl)-2pentenoyl)-β-D-glucuronide (5-GS-2-ene VPA-glucuronide), which could be identified in the 1H NMR spectrum by a characteristic downfield triplet at 6.95 ppm (CdCH), a recognizable sextet at 1.4 ppm (CH2CH3), and a triplet at 2.3 ppm (SCH2CH2) (Figure 4, lower panel, peaks with an asterisk). Other signals associated with this isomeric diconjugate were apparently superimposed with those of 5-GS-3-ene VPA-glucuronide II. Peak assignments for 5-GS-2-ene VPA-glucuronide were largely based on NMR spectra obtained for 2-propyl-5-(glutathion-S-yl)-2-pentenoic acid (5-GS-2-ene VPA) and the corresponding N-acetylcysteamine thioester of 5-GS-2-ene VPA (21). An MS/MS full daughter ion spectrum was also generated for 5-GS-2-ene VPA-glucuronide following a proper subtraction of the background derived from 5-GS-3-ene VPA-glucuronide II (Figure 3B). The fragmentation pattern of this diconjugate was shown, upon loss of the carbohydrate moiety (176 Da), to closely resemble that of 5-GS-2-ene VPA (MH+, 448) (21). Interpretation of the CID-produced fragmentation, however, was somewhat ambiguous. For example, the abundant daughter ion at m/z 226 could arise from the combined loss of glycine, pyroglutamate, and glucuronic acid and the ion m/z 123 from the loss of GSH and glucuronic acid. On the other hand, it was also possible for these two ions to result from CID of the primary daughter ion m/z 448 following identical fragmentation patterns, as in the case of 5-GS2-ene VPA. Quantitation of Biliary 5-GS-3-ene VPA and 5-GS3-ene VPA-glucuronide I. Upon administration of (E)2,4-diene VPA by ip injection of ∼178 µmol, the biliary


Tang and Abbott

Figure 4. 1H NMR spectra (D2O) of 5-GS-3-ene VPA-glucuronide I (upper panel) and 5-GS-3-ene VPA-glucuronide II (lower panel) isolated from the bile of rats dosed with (E)-2,4-diene VPA. The signals were assigned as indicated in the figures. The asterisks indicate signals associated with the putative 5-GS-2-ene VPA-glucuronide. G represents proton signals from the glucuronide moiety.

excretion of 5-GS-3-ene VPA-glucuronide I and 5-GS-3ene VPA over a 6 h period was determined to be 9.9 ( 1.0 and 1.9 ( 0.21 µmol, respectively. The total biliary GSH-glucuronide diconjugates, however, could be in excess of 10 µmol due to the presence of isomeric forms of the metabolites. Thus, the sum of 5-GS-3-ene VPA and 5-GS-3-ene VPA-glucuronides excreted in bile accounted for ∼6.6% of the dose. Incubation of the Biliary 5-GS-3-ene VPA-glucuronides with β-Glucuronidase. Following incubation of the rat bile with β-glucuronidase for 18 h at pH 5.7, the LC/MS/MS MRM signals recognized as 5-GS-3-ene VPA-glucuronide I were decreased significantly in comparison with the control incubation containing no enzyme, while the peaks corresponding to 5-GS-3-ene VPAglucuronide II remained unchanged (Figure 5). There was no difference between the 0 and 18 h incubations when β-glucuronidase was absent (Figure 5).

In Vitro Formation of the GSH-Glucuronide Diconjugate of (E)-2,4-Diene VPA. The procedure described by Williams et al. (17) was effective in eliminating most of the unreacted parent drug and endogenous compounds excreted in the bile. Further purification through flash chromatography afforded 2,4-diene VPA-glucuronide as a mixture of E and Z isomers. CID of the compound was dominated by the neutral loss of 176 Da characteristic for glucuronides. In preparation for the GSH addition reaction to 2,4diene VPA-glucuronide, the GST activity in the rat liver cytosolic fraction was determined to be 1533 ( 104 nmol min-1 (mg of protein)-1 using CDNB as a substrate, which is in agreement with the literature value (23). Following a 30 min incubation in buffer of GSH with 2,4diene VPA-glucuronide, 5-GS-3-ene VPA-glucuronide I was readily detected by LC/MS/MS (Figure 6B,C). Although no attempt was made to quantitate absolute


Figure 5. On-line LC/MS/MS detection (ion transitions for MRM were m/z 624 f 319 and 624 f 448; LC method C) of 5-GS-3-ene VPA-glucuronide I (tR 29.8 min) and 5-GS-3-ene VPA-glucuronide II (tR 27.7 min) in (A) after 0 min and (B) after 18 h of incubation containing the bile of rats treated with (E)2,4-diene VPA and (C) after 18 h of incubation of the sample with β-glucuronidase.


Figure 7. LC/MS mass spectrum of the mixture of products formed upon reaction of 2,4-diene VPA-glucuronide with reduced oxytocin. The alkylation was presumed to occur at the free cysteine residues. R represents 2,4-diene VPA-glucuronide.

forming a protonated molecular ion at m/z 1325 (Figure 7).

Discussion

Figure 6. On-line LC/MS/MS detection of 5-GS-3-ene VPAglucuronide I (ion transitions for MRM were m/z 624 f 319 and 624 f 448; LC method A) in (A) aqueous solution spiked with the purified metabolites, (B) after 30 min of incubation of 2,4-diene VPA-glucuronide with GSH in the presence of GST enzyme, (C) after 30 min of incubation of 2,4-diene VPAglucuronide with GSH, and (D) after 30 min of incubation of 2,4-diene VPA-glucuronide only.

amounts of the resultant diconjugate, it was estimated, based on the intensity of the MS/MS responses, that the rat liver GST-catalyzed reaction produced 3.5-fold more product than in the uncatalyzed buffer reaction. No ion signal corresponding to the GSH-glucuronide diconjugate appeared for incubations in which GSH was absent (Figure 6D). Alkylation of Reduced Oxytocin by 2,4-Diene VPA-glucuronide. LC/MS evidence indicated the formation of a bis-adduct when reduced oxytocin was allowed to mix with 2,4-diene VPA-glucuronide (Figure 7). The ions at m/z 1641 and 821 correspond to the single charged and double charged protonated dialkylated peptide, respectively. Other peaks associated with the bis-adduct are the ion m/z 832 representing the double charged monosodium adduct of the alkylated oxytocin and the ion at m/z 1465 possibly resulting from the neutral loss of 176 Da. In addition to the dialkylation, the MS spectrum revealed the presence of a mono-adduct,

The biotransformation of VPA to its metabolites entails two major phase I pathways, of which the secondary metabolite (E)-2,4-diene VPA may emerge from either pathway, namely, the microsomal P450-catalyzed dehydrogenation of (E)-2-propyl-2-pentenoic acid ((E)-2-ene VPA) (24) or the mitochondrial β-oxidation of 4-ene VPA (1). The diene metabolite has been shown to be hepatotoxic (6) and is suspected to play an important role in VPA-induced liver injury (13, 14). In this study, LC/MS/ MS and NMR data were used to clearly identify two new GSH-related metabolites of (E)-2,4-diene VPA, namely, the GSH-glucuronide diconjugates of (E)-2,4-diene VPA, in the bile of rats dosed with (E)-2,4-diene VPA. Sufficient on-line LC/MS/MS data were also obtained to indicate the presence of the NAC-glucuronide diconjugate of (E)-2,4-diene VPA in both rat bile and urine. The detection of the NAC-glucuronide diconjugate in bile is in accordance with our observations of formyl amphetamine metabolism in rats, whereby all metabolites of the GSH conjugates along the mercapturic acid pathway could be identified in the bile (25). The detected GSH-glucuronide diconjugates fall mainly into two categories according to whether the conjugate is β-glucuronidase sensitive (GSH-glucuronide I) or β-glucuronidase resistant (GSH-glucuronide II). The β-glucuronidase-resistant form of the diconjugate presumably results from a nonenzymatic intramolecular migration of the primary 1-O-acyl-glucuronide (26). This assignment of structure is based on earlier findings that VPA-glucuronide can undergo pH-dependent rearrangement and the resulting isomeric conjugates appear resistant to cleavage by β-glucuronidase (27). Theoretically, the migration could occur sequentially to form four positional isomers, namely, 1-, 2-, 3-, and 4-O-acylglucuronides either in vivo or in vitro after the bile was collected. It is not clear to which position the migration takes place, and 5-GS-3-ene VPA-glucuronide II is likely a mixture of 2-, 3-, and 4-O-glucuronides.


Tang and Abbott

Scheme 1. Summary of the Metabolic Pathways of (E)-2,4-Diene VPA Leading to the Formation of GSH Conjugatesa

a

Broken arrows indicate pathways lacking direct evidence.

Differences identified between these two types of diconjugates were not limited to their sensitivity toward β-glucuronidase but also to their MS/MS CID properties. The latter was reflected in the full daughter ion spectrum obtained for 5-GS-3-ene VPA-glucuronide II whereby scission of the linkage between the acyl moiety and glucuronic acid was not a favored process, suggesting improved strength of the C-O ester bond after migration. The data obtained in this study indicate that the normally used neutral loss scan of 176 Da to detect glucuronides by LC/MS/MS (22) would not be suitable to search for the isomerized 5-GS-3-ene VPA-glucuronide II. Whether this difference in fragmentation represents a general phenomenon between positional acyl glucuronides remains to be investigated. The discovery of the GSH-glucuronide diconjugates of (E)-2,4-diene VPA was to our mind significant in that the glucuronide ester of this VPA metabolite appears to represent one of the major bioactivated forms of this hepatotoxic diene. It has been proposed that the double bonds of (E)-2,4-diene VPA are activated via the formation of the corresponding CoA thioester in either mitochondria (from 4-ene VPA) or endoplasmic reticulum (from (E)-2-ene VPA) and thereby add to the terminal position of the diene through a Michael addition reaction (12, 14). However, direct experimental evidence has not been reported to pinpoint the role of CoA in the conjugation reaction of (E)-2,4-diene VPA with GSH. Using N-acetylcysteamine as a structural mimic of CoA, a recent in vitro investigation revealed that this thioester of (E)-2,4-diene VPA could react spontaneously with GSH and the conjugation reaction was greatly enhanced by rat hepatic GST enzymes (21). No reaction could be detected when the diene free acid was used as a substrate (21). This result was in agreement with findings obtained from chemical synthesis; i.e., (E)-2,4-diene VPA does not react with GSH unless first being converted to its ester form (12).

In this study, the in vitro incubation of GSH with 2,4diene VPA-glucuronide led to the formation of 5-GS-3ene VPA-glucuronide, and this conjugation reaction was further catalyzed by rat liver GST enzyme. These results, together with the characterization of the GSHglucuronide diconjugate formed in vivo in (E)-2,4-diene VPA-treated rats, were the first evidence that GSH conjugation of (E)-2,4-diene VPA was not dependent on the CoA thioester; i.e., the glucuronide would serve equally to activate the resulting unsaturated ester to react with GSH. Thus, the GS- thiolate arising from the GST-promoted deprotonation of GSH (28) attacks the activated terminal double bond of 2,4-diene VPA-glucuronide to form the GSH-glucuronide diconjugate via either 5,6- or 1,6-addition with the latter being the dominant process (Scheme 1). The conjugation reaction of GSH with the unsaturated glucuronide is probably limited to the cytosolic compartment because of the localization of uridine 5′-diphosphoglucuronosyl transferase, the enzyme catalyzing glucuronidation, in the membrane of the endoplasmic reticulum (29). Following the mercapturic acid biosynthesis pathway, the GSHglucuronides of (E)-2,4-diene VPA are degraded to the corresponding NAC-glucuronides and eventually excreted in the urine. It is less likely that the GSH-glucuronide diconjugate was formed from 5-GS-3-ene VPA, because the diconjugate in bile was determined to be 7-fold that of the corresponding GSH monoconjugate following a dose of (E)-2,4-diene VPA to rats. If the precursor of the diconjugate was indeed 5-GS-3-ene VPA, glucuronidation of the GSH conjugate would need to be very efficient. A high rate of glucuronidation has been known to demand sufficient lipid solubility of the substrate (29), and 5-GS3-ene VPA may not fulfill this requirement. Additional evidence to support the hypothesis that the GSH-glucuronide diconjugate is formed from the glucuronide ester is the fact that 2,4-diene VPA-glucuronide


appears to be the most abundant metabolite following a dose of (E)-2,4-diene VPA to the rat. For example, the 2,4-diene VPA-glucuronide used in this study was readily isolated from the bile of rats treated with the diene. Metabolic profiling in patients receiving VPA therapy revealed that the amount of conjugated (E)-2,4-diene VPA in the urine was 7.1-fold that of the unconjugated diene metabolite, indicating extensive glucuronidation of (E)2,4-diene VPA in vivo in the human (30). Thus, a significant quantity of the unsaturated glucuronide is available for conjugation with GSH. The hepatotoxic metabolite (E)-2,4-diene VPA can also be formed via the P450-catalyzed dehydrogenation of (E)2-ene VPA (24) with levels of the diene being severalfold that produced during the metabolism of VPA in rat as well as in human models (31, 32). A recent quantitative analysis comparing biotransformation of (E)-2-ene VPA and 4-ene VPA in rats found that the GSH conjugate of (E)-2,4-diene VPA formed from (E)-2-ene VPA was less than that from 4-ene VPA (14). It was suggested that metabolic activation of (E)-2-ene VPA through the CoAmediated pathway in endoplasmic reticulum was not as efficient as 4-ene VPA bioactivation in mitochondria (14). However, our detection of the GSH-glucuronide diconjugate of (E)-2,4-diene VPA implies that a major portion of the (E)-2,4-diene VPA derived from (E)-2-ene VPA metabolism could in fact react with hepatic GSH via the glucuronide-dependent process. Further investigations into the toxicological significance of this process with regards to the relative hepatotoxicity of (E)-2-ene VPA and VPA (33) need to be examined. Glucuronidation represents a major phase II metabolic pathway for those compounds that can afford to lose a reactive proton, such as alcohols, amines, and carboxylic acids (34). Conjugation of xenobiotics with D-glucuronic acid may take place directly or following introduction of an appropriate functional group into the parent molecule upon phase I metabolism. Although it is generally accepted that reaction of xenobiotics with D-glucuronic acid leads to conjugates that are rapidly excreted because of their hydrophilic nature, an increasing body of evidence suggests that drugs can be bioactivated through the glucuronidation pathway to bind with GSH (35) or with proteins (36). Among these glucuronides, the acyl glucuronides are by far the most reactive (36). For example, the incubation of VPA-glucuronide with human serum albumin was demonstrated to generate covalent VPA-protein adducts (17). The glucuronide mediated reaction of xenobiotics with GSH or proteins may follow a transacylation mechanism, where the reactive center is the acyl carbonyl carbon and the glucuronic acid moiety is displaced by free thiols or tyrosine or lysine residues in the target proteins (34). An example of such a reaction is the formation of clofibric acid GSH thioester that occurred when 1-O-clofibryl glucuronide was mixed with GSH. Transacylation was suggested to be the mechanism (35). Alternatively, the carbohydrate ring of the glucuronide may open to give a sugar chain having an aldehyde functional group. The aldehyde may then attack a free amino group on proteins to form an imine Schiff base which would ultimately rearrange to a 1-amino-2-keto protein adduct (37). Tolmetin glucuronide was observed to covalently bind to human serum albumin via this imine mechanism (37). To our knowledge, the conjugation of 2,4-diene VPAglucuronide with GSH is the first recorded instance in which a drug is activated by glucuronide ester formation to further react with GSH by Michael addition. The con-


sequences of forming this metabolite with respect to VPAassociated hepatotoxicity are far from clear, although one would suspect that this is one more reaction that would contribute to the observed decrease in hepatic GSH reported for VPA (38, 39). The 2,4-diene VPA is hepatotoxic in the rat (6), and the glucuronide ester is the main form of this metabolite found in the urine of rats (13) and patients on VPA therapy (30). GSH deficiency caused by the challenge of reactive compounds is known to result in cell injury and death due largely to the impaired antioxidant function of the GSH redox system (40). In view of the idiosyncratic nature of VPA hepatotoxicity, any metabolite of the drug that reacts with GSH should be considered suspect, particularly in patients who, because of genetic and/or environmental factors, may be already compromised with respect to GSH. On the other hand, the reactive 2,4-diene VPA-glucuronide may attack directly certain critical proteins to produce toxicity. Irreversible binding to rat liver proteins was reported to occur with both radiolabeled VPA and 4-ene VPA in vivo and in vitro, and the incorporation of radiolabel into proteins was observed to be decreased in rat hepatocytes when inhibitors of glucuronidation were added to the incubation (10). In the present study, the capability of the diene glucuronide to alkylate peptides or proteins was investigated using reduced oxytocin as a model biological substrate. The reaction of 2,4-diene VPA-glucuronide with reduced oxytocin resulted in an adduct in which the bound 2,4-diene VPA-glucuronide was stoichiometric to the available cysteine residues. Reduced oxytocin was shown to bind covalently with the N-acetylcysteamine thioester of (E)-2,4-diene VPA in a similar manner (21). The toxicological significance of 2,4-diene VPA-glucuronide formation has yet to be assessed. Normally the liver would be the logical target organ, being the site for reactive metabolite formation, but other tissues might also be affected, such as the kidney, in which glucuronides are excreted. Although a rare occurrence, VPA is reported to cause renal toxicity characterized by marked accumulation of lipids in the proximal tubular epithelium and alterations in mitochondria (41). The common excretion of the glucuronide of (E)-2,4-diene VPA in patients on VPA therapy, however, is evidence that formation of a reactive metabolite does not necessarily predict toxic consequences. In summary, the GSH-glucuronide diconjugates of (E)2,4-diene VPA were characterized by LC/MS/MS and 1H NMR. The formation of the diconjugates occurs most likely through a GST-catalyzed reaction of GSH with the activated 2,4-diene VPA-glucuronide via Michael addition. This discovery provides direct evidence for a reactive form of (E)-2,4-diene VPA and may be significant to our further understanding of the potential mechanism for VPA hepatotoxicity.

Acknowledgment. We thank Ms. J. Zheng (Faculty of Pharmaceutical Sciences, UBC) for the sample of [2H7]4-ene VPA, and Ms. M. Austria (Department of Chemistry, UBC) for conducting the NMR experiment. Technical assistance from Mr. Roland Burton (Faculty of Pharmaceutical Sciences, UBC) in the use of LC/MS is greatly appreciated. This research was supported by a program grant from the Medical Research Council of Canada. Supporting Information Available: Figure S1: LC/MS/ MS MRM chromatograms of 5-NAC-3-ene VPA-glucuronide

526 Chem. Res. Toxicol., Vol. 9, No. 2, 1996 detected in the bile and urine of rats treated with (E)-2,4-diene VPA (1 page). Ordering information is given on any current masthead page.

References (1) Baillie, T. A., and Rettenmeier, A. W. (1989) Valproate: Biotransformation. In Antiepileptic Drugs (Levy, R. H., Mattson, R. H., Meldrum, B. S., and Penry, J. K., Eds.), pp. 601-619, Raven Press, New York. (2) Dickinson, R. G., Harland, R. C., Ilias, A. M., Rodgers, R. M., Kaufman, S. N., Lynn, R. K., and Gerber, N. (1979) Disposition of valproic acid in the rat: Dose-dependent metabolism, distribution, enterohepatic recirculation and choleretic effect. J. Pharmacol. Exp. Ther. 211, 583-595. (3) Dickinson, R. G., Hooper, W. D., Dunstan, P. R., and Eadie, M. J. (1989) Urinary excretion of valproate and some metabolites in chronically treated patients. Ther. Drug Monit. 11, 127-133. (4) Baillie, T. A. (1988) Metabolic activation of valproic acid and drugmediated hepatotoxicity. Role of the terminal olefin, 2-n-propyl4-pentenoic acid. Chem. Res. Toxicol. 1, 195-199. (5) Zimmerman, H. J., and Ishak, K. G. (1982) Valproate-induced hepatic injury: Analysis of 23 fatal cases. Hepatology 2, 591597. (6) Kesterson, J. W., Granneman, G. R., and Machinist, J. (1984) The hepatotoxicity of valproic acid and its metabolites in rats. I. Toxicologic, biochemical and histopathologic studies. Hepatology 4, 1143-1152. (7) Kingsley, E., Gray, P., Tolman, K. G., and Tweedale, R. (1983) The toxicity of metabolites of sodium valproate in cultured hepatocytes. J. Clin. Pharmacol. 23, 178-185. (8) Bjorge, S. M., and Baillie, T. A. (1985) Inhibition of medium-chain fatty acid β-oxidation in vitro by valproic acid and its unsaturated metabolite, 2-n-propyl-4-pentenoic acid. Biochem. Biophys. Res. Commun. 132, 242-252. (9) Ponchaut, S., van Hoof, F., and Veitch, K. (1992) In vitro effects of valproate and valproate metabolites on mitochondrial oxidations. Relevance of Co A sequestration to the observed inhibitions. Biochem. Pharmacol. 43, 2435-2442. (10) Porubek, D. J., Grillo, M. P., and Baillie, T. A. (1989) The covalent binding to protein of valproic acid and its hepatotoxic metabolite, 2-n-propyl-4-pentenoic acid, in rats and in isolated rat hepatocytes. Drug Metab. Dispos. 17, 123-130. (11) Prickett, K. S., and Baillie, T. A. (1986) Metabolism of unsaturated derivatives of valproic acid in rat liver microsomes and destruction of cytochrome P-450. Drug Metab. Dispos. 14, 221-229. (12) Kassahun, K., Farrell, K., and Abbott, F. (1991) Identification and characterization of the glutathione and N-acetylcysteine conjugates of (E)-2-propyl-2,4-pentadienoic acid, a toxic metabolite of valproic acid, in rats and humans. Drug Metab. Dispos. 19, 525535. (13) Tang, W., Borel, A. G., Fujimiya, T., and Abbott, F. S. (1995) Fluorinated analogues as mechanistic probes in valproic acid hepatotoxicity: Hepatic microvesicular steatosis and glutathione status. Chem. Res. Toxicol. 8, 671-682. (14) Kassahun, K., Hu, P., Grillo, M. P., Davis, M. R., Jin, L., and Baillie, T. A. (1994) Metabolic activation of unsaturated derivatives of valproic acid. Identification of novel glutathione adducts formed through coenzyme A-dependent and -independent processes. Chem.-Biol. Interact. 90, 253-275. (15) Zheng, J. J. (1993) Metabolism and pharmacokinetic studies of valproic acid using stable isotope techniques. M.Sc. Thesis, The University of British Columbia (Vancouver). 147 pp. (16) Curci, R., Fiorentino, M., Troisi, L., Edwards, J. O., and Pater, R. H. (1980) Epoxidation of alkenes by dioxirane intermediates generated in the reaction of potassium caroate with ketones. J. Org. Chem. 45, 4758-4760. (17) Williams, A. M., Worrall, S., Jersey, J. D., and Dickinson, R. G. (1992) Studies on the reactivity of acyl glucuronides. III. Glucuronide-derived adducts of valproic acid and plasma protein and anti-adduct antibodies in humans. Biochem. Pharmacol. 43, 745755. (18) Siekevitz, P. (1962) Preparation of microsomes and submicrosomal fraction: Mammalian. Methods Enzymol. 5, 61-68. (19) Habig, W. H., Pabst, M. J., and Jakoby, W. B. (1974) Glutathion S-transferases: The first enzymatic step in mercapturic acid formation. J. Biol. Chem. 249, 7130-7139. (20) Erve, J. C., Deinzer, M. L., and Reed, D. J. (1995) Alkylation of oxytocin by S-(2-chloroethyl)glutathione and characterization of adducts by tandem mass spectrometry and Edman degradation. Chem. Res. Toxicol. 8, 414-421.

Tang and Abbott (21) Tang, W., Borel, A. G., and Abbott, F. S. (1996) Conjugation of glutathione with a toxic metabolite of valproic acid, (E)-2-propyl2,4-pentadienoic acid, catalyzed by rat hepatic glutathione Stransferases. Drug Metab. Dispos. (in press). (22) Straub, K. M., Rudewicz, P., and Garvie, C. (1987) ‘Metabolic mapping’ of drugs: Rapid screening techniques for xenobiotic metabolites with m.s./m.s. techniques. Xenobiotica 17, 413-422. (23) DePierre, J. W., and Morgenstern, R. (1983) Comparison of the distribution of microsomal and cytosolic glutathione S-transferase activities in different organs of the rat. Biochem. Pharmacol. 32, 721-723. (24) Kassahun, K., and Baillie, T. A. (1993) Cytochrome P-450mediated dehydrogenation of 2-n-propyl-2(E)-pentenoic acid, a pharmacologically-active metabolite of valproic acid, in rat liver microsomal preparations. Drug Metab. Dispos. 21, 242-248. (25) Borel, A. G., and Abbott, F. S. (1995) Characterization of novel isocyanate-derived metabolites of the formamide N-formyl amphetamine with the combined use of electrospray mass spectrometry and stable isotope methodology. Chem. Res. Toxicol. 8, 891899. (26) Spahn-Langguth, H., and Benet, L. Z. (1992) Acyl glucuronides revisited: Is the glucuronidation process a toxification as well as a detoxification mechanism? Drug Metab. Rev. 24, 5-48. (27) Dickinson, R. G., Hooper, W. D., and Eadie, M. J. (1984) pH-Dependent rearrangement of the biosynthetic ester glucuronide of valproic acid to β-glucuronidase-resistant forms. Drug Metab. Dispos. 12, 247-252. (28) Armstrong, R. N. (1991) Glutathione S-transferases: Reaction mechanism, structure, and function. Chem. Res. Toxicol. 4, 131140. (29) Mulder, G. J., Coughtrie, M. W. H., and Burchell, B. (1990) Glucuronidation. In Conjugation reactions in drug metabolism (Mulder, G. J., Ed.) pp 52-105, Taylor & Francis Ltd., London. (30) Kassahun, K., Farrell, K., Zheng, J., and Abbott, F. (1990) Metabolic profiling of valproic acid in patients using negativeion chemical ionization gas chromatography-mass spectrometry. J. Chromatogr. 527, 327-341. (31) Loscher, W., Wahnschaffe, U., Honack, D., Wittfoht, W., and Nau, H. (1992) Effects of valproate and E-2-en-valproate on functional and morphological parameters of rat liver. I. Biochemical, histopathological and pharmacokinetic studies. Epilepsy Res. 13, 187198. (32) Fabre, G., Briot, C., Marti, E., Montseny, J.-P., Bourrie, M., Masse, D., Berger, Y., and Cano, J.-P. (1992) 2-Valproate biotransformation using human liver microsomal fractions. Pharm. Weekbl., Sci. Ed. 14, 146-151. (33) Lo¨scher, W., Nau, H., Wahnschaffe, U., Ho¨nack, D., Rundfeldt, C., Wittfoht, W., and Bojic, U. (1993) Effects of valproate and E-2en valproate on functional and morphological parameters of rat liver. II. Influence of phenobarbital comedication. Epilepsy Res. 15, 113-131. (34) Faed, E. M. (1984) Properties of acyl glucuronides: implications for studies of the pharmacokinetics and metabolism of acidic drugs. Drug Metab. Rev. 15, 1213-1249. (35) Shore, L. J., Fenselau, C., King, A. R., and Dickinson, R. G. (1995) Characterization and formation of the glutathione conjugate of clofibric acid. Drug Metab. Dispos. 23, 119-123. (36) Zia-Amirhosseini, P., Spahn-Langguth, H., and Benet, L. Z. (1994) Bioactivation by glucuronide-conjugate formation. Adv. Pharmacol. 27, 385-397. (37) Ding, A., Ojingwa, J. C., McDonagh, A. F., Burlingame, A. L., and Benet, L. Z. (1993) Evidence for covalent binding of acyl glucuronides to serum albumin via an imine mechanism as revealed by tandem mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 90, 3797-3801. (38) Jezequel, A. M., Bonazzi, P., Novelli, G., Venturini, C., and Orlandi, F. (1984) Early structural and functional changes in liver of rats treated with a single dose of valproic acid. Hepatology 4, 1159-1166. (39) Cotariu, D., Evans, S., Zaidman, J. L., and Marcus, O. (1990) Early changes in hepatic redox homeostasis following treatment with a single dose of valproic acid. Biochem. Pharmacol. 40, 589593. (40) DeLeve, L. D., and Kaplowitz, N. (1991) Glutathione metabolism and its role in hepatotoxicity. Pharmacol. Ther. 52, 287-305. (41) Hawkins, E., and Brewer, E. (1993) Case 4: Renal toxicity induced by valproic acid (Depakene). Pediatr. Pathol. 13, 863-868.

TX950120Y

Bioactivation of a Toxic Metabolite of Valproic Acid - American

Recommend Documents