Abstract
Reactive and hepatotoxic metabolites formed from the biotransformation of valproic acid (VPA) are normally detoxified by conjugating with GSH and followed by mercapturic acid metabolism to produce their respective N-acetylcysteine (NAC) conjugates. Hence, the levels of NAC conjugates of VPA in human urine are an indirect measure of exposure of the liver toward reactive metabolites of the anticonvulsant drug. We report here the synthesis, identification, and characterization of a second NAC conjugate of (E)-2-propyl-2,4-pentadienoic acid in the urine samples (n = 39) of humans on VPA therapy, namely, (E)-5-(N-acetylcystein-S-yl)-2-ene VPA by gas chromatography/mass spectrometry and liquid chromatography with tandem mass spectrometry. In this study, we were able to separate the diastereomers of (E)-5-(N-acetylcystein-S-yl)-3-ene VPA by HPLC. The NAC conjugate of 4,5-epoxy VPA, namely, 5-NAC-4-OH-VPA γ-lactone, previously identified in rats treated with 2-propyl-4-pentenoic acid (4-ene VPA), was not detected in any of the human urine samples studied. This suggests that in humans, the P-450 metabolism of 4-ene VPA to the reactive epoxide is not a significant pathway. The excretion of the NAC conjugate of (E)-2,4-diene VPA glucuronide in the urine of seven patients on VPA was also examined and was not detected. The limit of detection of 5-NAC-3-keto VPA and its decarboxylated product, 1-NAC-3-heptanone, was estimated at 25 ng (signal to noise ratio > 3). Neither 5-NAC-3-keto VPA nor 1-NAC-3-heptanone was detected in the urine of patients on VPA therapy or 4-ene VPA-treated guinea pigs, but 1-NAC-3-heptanone was detected in the urine of 4-ene VPA-treated rats.
Valproic acid (VPA)2is well known for its broad spectrum antiepileptic activity and is gaining popularity as a component in antipsychotic therapy (Costello and Suppes, 1995; Bowden et al., 1994; Swann et al., 1997; Citrome et al., 1998). However, the drug suffers from a rare but potentially fatal hepatotoxicity side effect (Bryant and Dreifuss, 1996). The manifestation of the disease is such that researchers believe that reactive metabolites arising from VPA biotransformation are possibly involved (Gerber et al., 1979; Zimmerman and Ishak, 1982).
The P-450 metabolite, 2-propyl-4-pentenoic acid (4-ene VPA), was first singled out as the offending compound on account of its structural similarity to known hepatotoxins such as 4-pentenoic acid (Gerber et al., 1979). To gain insight into the mechanism by which this monoene induces the side effect, several toxicological and biotransformation studies of 4-ene VPA have been conducted in the rat and the rhesus monkey (Kesterson et al., 1984; Rettenmeier et al., 1985, 1986a,b, 1987). A review of these studies proposed that 4-ene VPA produces reactive intermediates that may be responsible for the liver injury observed (Baillie, 1988).
In principle, these reactive species are trapped as their GSH conjugates in the liver, then excreted in the bile. Subsequently, the GSH conjugates undergo additional metabolism along the mercapturic acid pathway to produce their respectiveN-acetylcysteine (NAC) conjugates, which are normally excreted in urine. Thus, evidence of biliary GSH and urinary NAC adducts of xenobiotics can be regarded as compelling proof of their reactive nature in vivo. Kassahun et al. (1991) reported the identification of 5-GS-2-propyl-3-pentenoic acid (3-ene VPA) (GSH I), the GSH conjugate of (E)-2-propyl-2,4-pentadienoic acid [(E)-2,4-diene VPA], in the bile of rats treated with 4-ene VPA and (E)-2,4-diene VPA. Furthermore, the investigators reported the identification of (E)-5-NAC-3-ene VPA (NAC I) in the urine of rats treated with 4-ene VPA and (E)-2,4-diene VPA and patients on VPA therapy, respectively. The same study reported what appeared to be a second conjugate, (E)-5-NAC-2-ene VPA (NAC II), whose identity could not be characterized at the time.
The detection of GSH I and NAC I marked the first evidence for the reactivity of (E)-2,4-diene VPA in both humans and rats. In a second study conducted to investigate the formation of thiol conjugates of reactive metabolites of VPA in rats, 5-GS-4-OH-VPA γ-lactone (GSH III) and 5-NAC-4-OH-VPA γ- lactone (NAC III) were reported as the most prominent thiol conjugates in the bile and urine of 4-ene VPA-treated rats, respectively (Kassahun et al., 1994). Both conjugates are believed to arise from the conjugation of GSH with 4,5-epoxy VPA, a microsomal metabolite of 4-ene VPA. The investigation also led to the identification of the minor metabolite, 5-GS-3-keto-VPA in bile and its respective NAC conjugate in the decarboxylated form, 1-NAC-3-heptanone (NAC IVb) in the urine of 4-ene VPA-treated rats. 5-GS-3-keto-VPA and NAC IVb are believed to be indirect evidence for the formation of their precursor, 2-propyl-3-oxo-4-pentenoic acid (3-keto-4-ene VPA; Kassahun et al., 1994). The latter is the ultimate electrophilic species, which is perceived to bind covalently to 3-ketoacyl-coenzyme A (CoA) thiolase to impairβ-oxidation of fatty acids (Rettenmeier et al., 1985).
Recently, the identification of biliary 5-GS-3-ene VPA glucuronide (GSH V) and urinary 5-NAC-3-ene VPA glucuronide (NAC V) in rats (Tang and Abbott, 1996a), has suggested that the conjugation of GSH with (E)-2,4-diene VPA can also occur while the diene is in the glucuronide ester form. Neither NAC III, 5-NAC-3-keto VPA (NAC IVa), NAC IVb, nor NAC V has been identified in humans; their detection in the urine of patients on VPA therapy would bear significance in determining the metabolic fate of 4-ene VPA in humans, and their possible relevance to the mechanism of VPA hepatotoxicity.
Because it is perceived that VPA hepatotoxicity is caused, in part, by reactive metabolites and their imposition on the GSH pool (Cotariu et al., 1990; Kassahun et al., 1991), it is crucial in the investigation of the mechanism of this side effect to identify all thiol conjugates of VPA as an indirect means to determine liver exposure toward hepatotoxic metabolites. To this end, this study was designed to examine the formation of NAC conjugates arising from VPA biotransformation in humans. The goals of this study were: 1) to investigate the in vivo isomerization of NAC I, 2) to characterize NAC II, and 3) to investigate the formation of NAC III, NAC IV, and NAC V in patients on VPA therapy, and in rats and guinea pigs treated with 4-ene VPA.
Materials and Methods
Chemicals Purchased.
Pentafluorobenzyl bromide (PFBBr) and diisopropylethylamine were purchased from Aldrich Chemical Co. (Milwaukee, WI). Trifluoroacetic acid (TFA) was purchased from Sigma Chemical Co. (St. Louis, MO).N-methyl-N-trimethylsilyl-trifluoroacetamide andN-tert-butyldimethylsilyl-N-methyltrifluoroacetamide were purchased from Pierce Chemical Co. (Rockford, IL). All other chemicals and solvents were of analytical grade.
Chemicals Synthesized.
Synthesis of NAC I and NAC II
The methyl ethyl ester of NAC I was first synthesized by the conjugation of the methyl ester of NAC with (E)-2,4-diene VPA ethyl ester as described by Kassahun et al. (1991). The resulting methyl ethyl ester of NAC I (3.45 g, 10 mmol) was hydrolyzed with 0.6 N sodium hydroxide (34 ml, 20 mmol) under reflux conditions. The mixture was extracted with ether and purified by gel column chromatography on silica gel (7.5% methanol and 2.5% acetic acid in chloroform) to afford a mixture of NAC I and NAC II (1:2 in favor of NAC II) as determined by 1H NMR, gas chromatography (GC)/mass spectrometry (MS), and liquid chromatography with tandem mass spectrometry (LC/MS/MS). Additional purification of the resulting mixture of NAC I and NAC II by HPLC gave pure isomers of NAC IA, NAC IB, and NAC II, respectively.
Negative ion chemical ionization (NICI) spectra of the pentafluorobenzyl (PFB) derivatives of HPLC isolated NAC IA, NAC IB, NAC II: m/z 482 (100%). Electron impact (EI) spectra of the tert-butyldimethylsilyl (t-BDMS) derivatives of HPLC isolated NAC IA, NAC IB, and NAC II:m/z (%): 474 (100), 415 (40). MS/MS positive electrospray (ES+) spectra of NAC IA and NAC IB: m/z 304 (MH+), MS/MS [collision-induced dissociation (CID) of MH+]:m/z (%): 95 (100), 304 (80), 130 (80), 123 (85). NAC II: m/z 304 (MH+), MS/MS (CID of MH+): m/z(%): 123 (100), 95 (50), 304 (40), 130 (40), 146 (15).1H NMR (300 MHz, D2O) of HPLC-isolated NAC II: δ 0.85 (t, 3H, CH3 -), 1.30 to 1.50 (m, 2H, CH2–CH3), 2.02 (s, 3H, -NHCOCH3), 2.25 (t, 2H, -CH2-CH2-CH3), 2.55 (q, 2H, -CH2-CH2-CH -), 2.75 (t, 2H, CH2-CH2-), 2.85 to 3.0 and 3.08 to 3.20 (m, 2H, -S-CH2 –CH), 4.4 to 4.6 (dd, 1H, S-CH2-CH -), 6.75 (t, 1H, CH2-CH=C-, J = 7.2 Hz). The 1H NMR of NAC IA and NAC IB (300 MHz, D2O) were identical and consistent with the data reported for the dimethyl ester of NAC I (Kassahun et al., 1991). δ 0.90 (t, 3H, -CH2-CH2-CH3), 1.15 to 1.40 (m, 2H, CH2-CH2-CH3), 1.52 and 1.75 (2H, 2m, -CH2-CH2-CH3), 2.02 (s, 3H, -NH-CO-CH3), 2.78 and 3.00 (2H, 2m, S-CH2-CH-), 3.10 (1H,q, =CH-CH-CH2), 3.20 (2H, d, CH2-CH=), 4.45 (1H, dd, S-CH2-CH-), 5.50 to 5.65 (2H, m, CH2-CH=CH).
Synthesis of NAC III.
NAC III was synthesized by the conjugation of NAC with trifluoroethyl 4,5-epoxy VPA as reported for a previous study (Tang and Abbott, 1996b). NICI spectrum of the PFB derivative of NAC III:m/z 302 (100%). MS/MS ES+spectrum of NAC III was consistent with previously reported data (Tang and Abbott, 1996b): m/z 304 (MH+), MS/MS (CID of MH+):m/z (%); 304 (100), 74 (45), 262 (40), 216 (40), 130 (30).
Synthesis of NAC IVa.
Methyl NAC IVa methyl ester was synthesized for a previous study (Kassahun and Abbott, 1993). To obtain the conjugate in the acid form, 10 mg (30 μmol) of the ester was mixed with 60 μl (60 μmol) of 1 N NaOH, and the mixture was allowed to stir at room temperature overnight. The solution was made up to volume with water to a resulting solution of 1 mg/ml, and the pH of the solution was adjusted to 7 with 3 N HCl. MS/MS ES+ spectrum of NAC IVa:m/z 320 (MH+), MS/MS (CID of MH+): m/z (%); 55 (100), 164 (70), 130 (55), 122 (50), 189 (40), 320 (35). MS/MS ES+ spectrum of the decarboxylated product of NAC IVa, NAC IVb: m/z 276 (MH+), MS/MS (CID of MH+):m/z (%); 113 (100), 163 (70), 145 (75), 276 (65), 122 (60).
In Vivo Metabolism Studies in Humans and Animals.
Human studies
Urine samples (n = 39) were obtained from pediatric patients diagnosed with different types of epilepsy and who attended the Seizure Clinic of the British Columbia Children's Hospital (Vancouver, British Columbia, Canada). All patients were on VPA therapy and had reached steady state. Urine samples were collected just before the morning dose of VPA was administered. Control urine samples were obtained from healthy volunteers.
Animal studies.
Urine samples (24-h) obtained from adult male Sprague-Dawley rats (n = 4, 211 ± 6 g) and adult male Dunkin Hartley guinea pigs (n = 4, 324 ± 11 g) dosed with 4-ene VPA at 100 mg/kg (i.p) and collected for a previous study (Tang et al., 1996b) were also studied here for the investigation of NAC IVa and NAC IVb. Control urine samples collected from the same animals before dosing were also studied.
Sample preparation for the identification of NAC IA, NAC IB, NAC II, NAC III, NAC IVa, NAC IVb, and NAC V by LC/MS/MS.
One milliliter of urine sample was adjusted to pH 3 to 4 with 3 N HCL and applied to C2 cartridges (500 mg; Varian, Harbor City, CA), which were preconditioned with methanol and water. The loaded cartridges were washed with 3 ml each of water/methanol (95:5) and water, and the conjugates were eluted with 6 ml of methanol. The methanol layers were dried under N2 and reconstituted in 1 ml of mobile phase.
Sample preparation for the derivatization and identification of NAC I, NAC II, and NAC III by GC/MS.
One milliliter of urine sample from each of the patients studied (n = 39) was adjusted to pH 3 to 4 with 40 μl of concentrated phosphoric acid and saturated with 1 g of NH4Cl. The conjugates were then extracted with 5 ml of ethyl acetate for one-half hour. The organic layer was separated and dried with anhydrous sodium sulfate and reduced in volume to 1 ml under N2. The organic layer was derivatized with 20 μl of PFBBr in the presence of 10 μl of diisopropylethylamine at 50°C for 1 h in a conical reaction vial.
Sample preparation for the derivatization and identification of 3-n-propyl-5-hydroxymethyltetrahydro-2-furanone (4,5-diOH-VPA γ-lactone) by GC/MS.
One milliliter of urine sample from two patients was extracted twice with 5 ml of ethyl acetate at pH 2. The organic layers were separated, combined, and dried with sodium sulfate and reduced in volume to 100 μl under N2 and derivatized with 100 μl ofN-methyl-N-trimethylsilyl-trifluoroacetamide at 60°C for 1 h to form the trimethylsilyl (TMS) derivative of 4,5-diOH-VPA γ-lactone.
Analytical Methods.
All 1HNMR experiments were carried out on a Varian XL-300 spectrometer in the Department of Chemistry, University of British Columbia.
GC/MS conditions for the identification of NAC I, NAC II, and NAC III.
The GC/MS experiments were conducted on a Hewlett-Packard (HP; Avondale, PA) 5890 II GC coupled to a HP 5989A mass spectrometer operating in EI and NICI mode. The column was a J & W Scientific DB1701 (Folsom, CA), 30 m × 0.32 mm (i.d.) with a 0.25-μm film thickness. The oven temperature program was set as follows: 50–150°C at 20°/min, held for 5 min; 150–300°C at 10°/min and held for 5 min. The injection port was at 240°C, the source temperature was at 200°C, and the temperature of the interface was set at 280°C. The pressure for the carrier gas, helium, was set at 10 psi.
For NICI analysis, the ion source temperature was kept at 200°C and the interface temperature was set at 280°C. The reagent gas, methane, was maintained at 1 torr. The emission current was at 300 μA and the ionization energy was 120 eV. For purposes of identification and characterization of the thiol conjugates of VPA, both the scan mode and the selected ion monitoring (SIM) mode were used. In SIM mode, the ions monitored corresponded to the [M-181]−carboxylate anion fragments of the PFB derivatives of NAC I (m/z 482), NAC II (m/z482), and NAC III (m/z 302).
For EI analysis, the ionization energy was 70 eV and the emission current was 300 μA with a source pressure of 10−6 torr. The detection of 4,5-diOH-VPAγ-lactone was performed by the selective monitoring of the [M-15]+ fragment of its TMS derivative at m/z 215.
HPLC conditions for the isolation and purification of NAC IA, NAC IB, and NAC II.
Purification of the synthetic compounds was performed on a HP 1050 HPLC equipped with a variable wavelength UV detector. The column was a Beckman C18 (San Ramon, CA), 25 cm × 4.6 mm × 5 μm. The mobile phase (methanol/water, 43:57, containing 0.05% TFA as modifier) was delivered isocratically at a flow rate of 0.5 ml/min, and the UV detector was set at 254 nm. The fractions of each isomer were collected in 13 × 100-mm disposable culture tubes (Fisher Scientific, Pittsburgh, PA) combined, and each product was then isolated after lyophilization.
HPLC/MS/MS conditions for the identification of NAC IA, NAC IB, NAC II, NAC III, NAC IVa, NAC IVb, and NAC V.
Experiments were carried out on a Micromass Quattro triple quadrupole mass spectrometer (Montreal, Quebec) interfaced to a HP 1090 series II HPLC equipped with a Phenomenex C8 column (100 mm × 2.1 mm, 5 μm; Torrance, CA). HPLC mobile phase A consisted of methanol/water (43:57) with 0.05% of TFA, and mobile phase B consisted of methanol. At time 0, mobile phase A was pumped isocratically for 30 min. Mobile phase B was increased from 0 to 100% at time 30.1 min and held for 5 min, followed by a sharp gradient increase of mobile phase A to 100% at time 35.1 min. All flow rates were 0.1 ml/min.
The MS/MS conditions used ES+ as the ionization energy of choice. The HPLC eluent was introduced into the stainless steel electrospray capillary sprayer held at 3.5 kV. The mass spectrometer was operated for the purpose of precursor ion scanning, product ion scanning, or multiple reaction monitoring (MRM). Multipliers 1 and 2 were set at 650 V for all three experiments, cone voltage was set at 22 V with skimmer offset by 5 V. The low and high mass resolution were set at 12.5 for both precursor ion scanning and product ion scanning, and at 5 for MRM.
In this study, the precursor ion scanning determined the pseudomolecular ion or the precursor ion (MH+) of the metabolite of interest, the product ion scanning determined the product ions of MH+ using argon as the target gas for CID and with collision energy typically set at 40 eV. During MRM, the pseudomolecular ion for each analyte was selected in the first quadrupole and subjected to CID with target gas argon set at a pressure of 3.0 × 10 −4 millibar and collision energy was set at 50 eV to improve sensitivity. Specific fragment ions were then selected by the third quadrupole. For all the experiments, the source temperature was maintained at 140°C.
To identify NAC IA, NAC IB, and NAC II by MRM, the transitionsm/z 304 to m/z 130 andm/z 304 to m/z 123 were monitored in all samples. For the identification of NAC III, the transition m/z 304 to m/z130 was monitored in all samples. To detect NAC IVa, the transitionsm/z 320 to m/z 130 andm/z 164 were monitored for the intact molecule in urine extracts of patients (n = 7), rats (n = 4), and guinea pigs (n = 4). The transitions m/z 276 to m/z113 and m/z 163, the diagnostic product ions of NAC IVb were also monitored. To detect NAC V, the transitionm/z 480 to m/z 304, diagnostic for the loss of a glucuronide moiety, was monitored in seven of the urine extracts of patients.
Results
Characterization of Products Formed from Hydrolysis of Methyl Ethyl Ester of NAC I.
The synthesis of methyl ethyl ester of the NAC conjugate of (E)-2,4-diene VPA was easily reproduced following the procedure described previously (Kassahun et al., 1991). The analysis of the product by 1HNMR confirmed the presence of a single compound and verified the identity and purity of the diester. The hydrolysis of the methyl ethyl ester under strong alkaline and reflux conditions yielded a mixture of NAC I and NAC II.1HNMR of the pure mixture as well as the GC/MS analysis of the PFB- and t-BDMS-derivatized mixture indicated that NAC II was about 2 to 3 times the amount of NAC I. The isolation of NAC II and each diastereomer of NAC I from the synthetic mixture was easily achieved by preparative HPLC. Although the chromatographic characteristics of the PFB and t-BDMS derivatives of NAC I and NAC II were distinct from each other, their mass spectroscopic characteristics did not differentiate them from each other, suggesting that the compounds were isomers of each other. The structure of NAC II was confirmed to be 5-(N-acetylcystein-S-yl)-2-propyl-2-pentenoic acid, a positional isomer of NAC I, by 1H NMR.
The presence of the double bond in NAC II was confirmed by a downfield triplet signal at 6.75 ppm corresponding to the single proton at carbon 3 (Fig. 1). The single vinylic proton was coupled to the adjacent -CH2- and was consistent with a trans double bond based on the chemical shift observed for this substituted alkene (Silverstein et al., 1981). The coupling constant (J) of the vinylic proton was 7.2 Hz, the same as that reported for the vinylic proton of (E)-2-propyl-2-pentenoic acid (2-ene VPA; Acheampong et al., 1983). The absence of a proton signal at carbon 2 additionally confirmed the position of the double bond. Furthermore, the protons at carbon 3′ were coupled to the neighboring -CH2- and were split into a triplet at 2.25 ppm. The quartet at 2.55 ppm corresponded to the protons at carbon 4 and the triplet at 2.75 ppm was assigned to the two protons at carbon 5 adjacent to sulfur. The position of the double bond at carbons 2 and 3 in NAC II eliminated the structural presence of a chiral carbon in the VPA portion of the molecule and, consequently, no diastereomers were observed for NAC II by HPLC.
Both of the HPLC fractions assigned as the diastereomers of NAC I had identical 1H NMR spectra as anticipated byKassahun et al. (1991) for trans NAC I dimethyl ester. The1H NMR results reported here were identical with data published for the acid (Tang and Abbott, 1996b). The downfield signal at 5.50 to 5.65 ppm was a multiplet and consistent with the vinylic protons at carbons 3 and 4 of NAC I as depicted in Fig. 1. The quartet at 3.10 ppm was distinct for both isomers of NAC I and could be assigned to the single proton at carbon 2. The doublet at 3.20 ppm was assigned to the protons at carbon 5.
The separation of the two diastereomers of NAC I under the HPLC conditions used was consistent with previous reports for the separation of the diastereomeric mercapturates of bromoisovalerylurea by HPLC using mobile phases containing an ion pair-forming reagent (Te Koppele et al., 1986, 1988). The addition of TFA to the mobile phase enhanced the baseline separation of both diastereomeric peaks of NAC I, which were sharp.
Tandem mass spectra of NAC I and NAC II were found to be similar to the results reported for NAC I (Tang and Abbott, 1996b). For both compounds, fragments resulting from the cleavage of the thioether bond on the NAC side of the parent molecule gave rise to the diagnosticm/z 130 product ion as shown in Fig.2, the product ion mass spectrum of NAC II. Cleavage of the thioether bond on the VPA side of the molecule followed by the loss of water resulted in the fragment ion atm/z 123. The loss of the NAC and the COOH moieties from the protonated parent molecule led to the formation of the product ion at m/z 95. The spectra of NAC IA and NAC IB did not produce any distinguishing MS/MS features that differentiated them from each other or from NAC II.
The GC/MS NICI spectrum of the corresponding di-PFB derivative of NAC II was characterized by the [M-181]−carboxylate anion fragment at m/z 482. Similarly, in EI mode, the spectrum of the di-t-BDMS derivative of NAC II was characterized by the diagnostic [M-57]+ion at m/z 474. The di-PFB and di-t-BDMS derivatives of both isomers of NAC I produced similar spectra by EI and NICI, respectively, and the derivatives of the isomers coeluted on the GC column. However, the elution of the PFB and t-BDMS derivatives of NAC I ahead of those of NAC II was consistent with the chromatographic characteristics of phase I metabolites whereby the PFB or t-BDMS derivatives of (E)-3-ene VPA elute ahead of the corresponding derivatives of (E)-2-ene VPA (Abbott et al., 1986; Kassahun et al., 1990).
Identification of NAC Conjugates of VPA Metabolites in Patients on VPA Therapy.
NAC conjugates of (E)-2,4-diene VPA (NAC I, NAC II, and NAC V)
The identification and characterization of NAC I and NAC II in urine extracts of patients on VPA were conducted by comparing their retention times and mass spectral data with those of synthetic reference samples analyzed under the same experimental conditions.
Figure 3 shows typical selected ion chromatograms in the NICI mode obtained for the PFB-derivatized urine extracts of: 1) a control sample spiked with NAC I and NAC II and 2) a urine sample from a patient on VPA therapy while monitoring form/z 482. A comparison of the two chromatograms indicated that the di-PFB derivatives of NAC I and NAC II in patient samples had identical retention times (tR= 28.77 and 30.19 min, respectively) to those of the PFB derivatives of the synthetic standards. Furthermore, a full NICI mass spectrum of the PFB derivative of NAC II could be obtained in some of the patients studied. The fragmentation pattern was similar to that of the authentic reference sample. In both cases, the mass spectra were characterized by the most abundant [M-181]− fragment ion atm/z 482.
The detection of NAC I and NAC II was additionally confirmed by LC/MS/MS analysis using MRM, which showed that the diagnostic product ions m/z 123 and m/z 130 were present in all of the extracted urine samples of patients (n = 39) analyzed. At both MRM transitions, the peaks corresponding to NAC I and NAC II eluted at the same retention times and in the same area ratio as observed for the authentic reference samples of both conjugates. The detection of the NAC conjugates of (E)-2,4-diene VPA by LC/MS/MS in one urine extract of a patient sample is shown in Fig. 4. A full product ion mass spectrum of NAC II could also be obtained in some patient samples. The isomers of NAC I were detected in all patients studied in about equal amounts. The NAC conjugate of (E)-2,4-diene VPA glucuronide (NAC V) was not detected in the urine extracts of patients (n = 7) on VPA by monitoring the transition m/z 480 to 304, a diagnostic product ion that is formed as a result of the loss of a glucuronide group from the protonated parent compound (Tang and Abbott, 1996a).
NAC conjugate of 4,5-epoxy VPA (NAC III).
Under the LC/MS/MS conditions used, the MS/MS ES+spectrum of NAC III was consistent with that previously reported (Tang and Abbott, 1996b). The CID of the protonated molecular ion (m/z 304) produced a fragment atm/z 262, corresponding to a loss of the elements of ketene. Similar to NAC I and NAC II, we observed a fragment ion atm/z 130 corresponding to the cleavage of the thioether bond on the NAC moiety. Cleavage of the carbon-sulfur bond on the side of the VPA moiety of NAC III did not occur.
GC/MS NICI analysis indicated that NAC III formed the mono PFB derivative when derivatized by PFBBr. This result indicated that the lactone moiety of the conjugate remained closed during derivatization. The mass spectrum of the PFB derivative of NAC III was characterized by the [M-181]− fragment atm/z 302. Interestingly, the epimers of the conjugates in a ratio of 2:1 were separable on a DB1701 column but not on the relatively nonpolar DB101 column (Fig.5).
In spite of the sensitivity provided by the GC/MS for the mono PFB derivative, NAC III was not detectable in any of the patients samples studied (n = 39). The synthetic reference conjugate could be detected at concentrations as low as 10 ng/ml or 10 pg (on column injection, signal to noise ratio > 3) in control samples spiked with NAC III. Similarly, by MRM of m/z 304 to 130, NAC III was not detected at its expected retention time in any of the samples. During the same analysis, an unknown peak eluting at tR=12 min and ahead of NAC III was also observed in all samples (n = 39) studied. This peak was not found in the control urine samples and appears to be an NAC adduct as well. The peak appeared to be more abundant than NAC I or NAC II in the patient samples analyzed.
Furthermore, a search for the rearrangement end product of 4,5-epoxy VPA, i.e., 4,5-diOH-VPA γ-lactone, by monitoring form/z 215, the [M-CH3]+ fragment of the TMS derivative of the lactone by GC/MS EI as reported by Rettenmeier et al. (1985), was negative in two urine samples studied.
NAC conjugate of 3-keto-4-ene VPA (NAC IVa and the decarboxylated product NAC IVb).
On the alkaline hydrolysis of a portion of the synthetic dimethyl ester of 3-keto-4-ene VPA, part of the diester produced NAC IVa whereas part of the compound was decarboxylated to NAC IVb. By LC/MS/MS analysis, the CID of the protonated molecular ion of NAC IVa,m/z 320, produced the product ion spectrum shown in Fig. 6A. Cleavage of the thioether bond in this molecule produced the protonated NAC fragment atm/z 164 and 3-keto VPA atm/z 157. We also observed the diagnosticm/z 130 ion, which corresponded to the loss of H2S from MH+ ion of NAC. Similarly, the CID of m/z 276, the protonated molecular ion of NAC IVb, produced m/z 113 as the most prominent product ion corresponding to the protonated 3-heptanone portion of the molecule. The positive charge can also be retained by the NAC moiety to afford m/z 163 (Fig. 6B).
NAC IVa could not be detected by LC/MS/MS using MRM ofm/z 320 to m/z 164 in either the urine extracts of patients on VPA or in the urine extracts of rats and guinea pigs dosed with 4-ene VPA. However, NAC IVb, the decarboxylated product of NAC IVa, was clearly identified in the urine extracts of all the rats treated with 4-ene VPA by monitoring the transitions m/z 276 to m/z163 and m/z 276 to m/z 113 (Fig. 7). NAC IVb was not evident by this technique in any of the urine extracts of patients on VPA or in the urine extracts of guinea pigs dosed with 4-ene VPA.
Discussion
In this study, a new thiol conjugate of (E)-2,4-diene VPA, namely, (E)-5-(N-acetylcystein-S-yl)-2-ene VPA (NAC II), has been unequivocally detected using GC/MS and LC/MS/MS in the urine of all patients (n = 39) treated with the antiepileptic drug VPA. The key evidence to elucidate the structure of NAC II was obtained by the 1H NMR analysis of the synthetic product, which showed conclusively that NAC II was a positional isomer of NAC I, the first thiol conjugate of VPA characterized in humans. NAC II was more prominent than NAC I by both methods of detection in all of the patients studied. The diastereomers of NAC I were confirmed in all of the patients studied. That the diastereomers of NAC I were detected by LC/MS/MS in about equal amounts in all samples analyzed suggests an in vivo isomerization of the conjugate.
There was no evidence for the identification of NAC III, the thiol conjugate of 4,5-epoxy VPA, the microsomal P-450 epoxidation product of 4-ene VPA, in any of the samples (n = 39) studied despite the high sensitivity of the detection technique for the compound. In contrast, in rats treated with 4-ene VPA, NAC III was reported to be a prominent metabolite (Kassahun et al., 1994). This suggests that the epoxidation of 4-ene VPA is not a significant pathway in humans, at least those not exhibiting overt toxicity. Additional evidence to support this conclusion was provided by the lack of detection of 4,5-diOH-VPA γ-lactone, the more stable metabolic end product of the epoxide, in two patients samples studied. The lactone has been reported as a metabolite of 4-ene VPA in rats (Rettenmeier et al., 1985), and the findings in this study suggest that a species difference may exist between humans and rats for the metabolic pathway of 4-ene VPA. This should be confirmed with a larger sample size of patients studied.
There was no evidence of NAC IVa or its decarboxylated product, NAC IVb, in humans and guinea pigs by LC/MS/MS. The relatively high limits of detection of both compounds, which were estimated at about 25 ng (on column injection, signal to noise ratio > 3), suggest two possible interpretations. First, the precursor of the conjugates, 3-keto-4-ene VPA, is not formed in detectable quantities in humans or guinea pigs. Alternatively, the metabolite could be formed by an intrahepatic process as proposed by Hinchman and Ballatori (1994), such that it is eliminated in the feces of guinea pigs and humans. However, the detection of NAC IVb in 4-ene VPA-treated rats but not in 4-ene VPA-treated guinea pigs favors the former reasoning. These possibilities need to be investigated more to characterize species differences with respect to the biotransformation of VPA and 4-ene VPA.
The detection of NAC IVb in rats is in support of the findings reported previously (Kassahun et al., 1994). These investigators proposed that the detection of NAC IVb, the decarboxylated product of NAC IVa, is indirect evidence for the formation of the reactive metabolite, 3-keto-4-ene VPA. However, the detection of NAC IVb in rats could also be the β-oxidation product of NAC II or its precursor, GSH II, by the same mechanism that (E)-2-ene VPA produces 3-keto VPA, although there is no prior evidence in the literature for such a biotransformation of a phase II conjugate.
The lack of detection of NAC IVa and the relatively high concentrations of NAC I and NAC II detected in the urine samples of patients on VPA therapy suggest that the in vivo GSH conjugation of (E)-2,4-diene VPA predominates over theβ-oxidation of the diene to form 3-keto-4-ene VPA. This result can be explained by the high reactivity of the diene ester. Whether (E)-2,4-diene VPA may play a significant role in the induction of VPA hepatotoxicity is still to be determined.
Overall, the evidence from this study indicated that thiol conjugates of VPA arise primarily from the biotransformation of (E)-2,4-diene VPA in humans. The detection of an unknown peak that eluted ahead of NAC I and NAC II by LC/MS/MS and bearing similar mass spectral characteristics to NAC I, NAC II, and NAC III suggests that another NAC conjugate could possibly be formed. The use of LC/NMR/MS would serve as an important tool to elucidate the structure of this apparent conjugate.
By synthesis, both NAC I and NAC II were formed in this study as a result of the heat catalyzed and base hydrolysis of NAC I methyl ethyl ester. On attack by base, the acidic proton alpha to the carbonyl group of the conjugate is abstracted by the resulting carboxylate anion to form a tertiary carbanion. The latter can exist in equilibrium with a second compound where a more stable carbanion is in conjugation with an allylic double bond and that of a carbonyl group as shown in SchemeFS1. On acidification, NAC I and NAC II are formed in a ratio of 1:2∼3 favoring the latter conjugate. This is consistent with the ratio of NAC I and NAC II observed in all patients samples studied, and can in part explain the formation of the two conjugates in vivo.
First, the diene is reactive only in its ester form (Kassahun et al., 1991; Kassahun and Baillie, 1993; Tang et al., 1996a). In principle, (E)-2,4-diene VPA can be formed as the CoA ester in mitochondria after the β-oxidation of 4-ene VPA or as the glucuronide ester in the cytosol after the microsomal oxidation of (E)-2-ene VPA. Contrary to reports for 4-ene VPA-treated rats (Tang and Abbott, 1996a), NAC V was not detected in patients on VPA therapy. This was evidence favoring the CoA ester but not the glucuronide ester of (E)-2,4-diene VPA as the reactive intermediate formed in humans.
Second, it has been proposed that the in vivo conjugation of GSH with (E)-2,4-diene VPA CoA is a nucleophilic attack of the GS− anion on the terminal double bond of the ester generating a tertiary carbanion (Kassahun et al., 1991). Thus, the mechanism proposed for the GSH conjugation reaction of (E)-2,4-diene VPA can be modified to include the migration of a negative charge to favor a resonance structure, which leads to the formation of both 5-GS-3-ene VPA CoA and 5-GS-2-ene VPA CoA in vivo as illustrated in Scheme FS2. The formation of the thiol anion (GS−) is believed to be promoted by the action of GSH transferase in vivo (Graminski et al., 1989). Both of the resulting GSH conjugates would undergo additional metabolism along the mercapturic pathway to give NAC I and NAC II. Evidence for the possible involvement of GSH transferase in the formation of both 5-GS-2-ene VPA ester and 5-GS-3-ene VPA ester from the GSH conjugation reaction of (E)-2,4-diene VPA has been demonstrated from in vitro studies using rat liver mitochondrial preparations (Tang et al., 1996a).
Alternatively, the formation of NAC II could be explained by invoking the action of isomerase, the enzyme responsible for converting a 3-ene double bond to a trans 2-ene double bond (Osmundsen and Hovik, 1988), converting NAC I to NAC II or their respective GSH precursors. The action of 2,3-enoyl CoA isomerase has been invoked to explain the origin of 2-ene VPA and 3-ene VPA (Granneman et al., 1984; Rettenmeier et al., 1987), both of which are believed to be formed in mitochondria (Bjorge and Baillie, 1991). This would support the theory that the mitochondria is the site of formation for the GSH conjugates of (E)-2,4-diene VPA, which produces NAC I and NAC II, and the site most susceptible to injury as a result of exposure to reactive metabolites of VPA. A scheme for the metabolic pathway of VPA leading to the formation of thiol conjugates in humans has been proposed in Scheme FS3.
It has been postulated that a high exposure of hepatic mitochondria toward reactive metabolites arising from 4-ene VPA biotransformation could interfere with the β-oxidation of fatty acids, which on accumulation cause microvesicular steatosis (Zimmerman and Ishak, 1982; Kesterson et al., 1984; Rettenmeier et al., 1985) in rats and humans. Furthermore, it has been proposed that the formation of reactive intermediates in mitochondria can selectively deplete the GSH pool to cause oxidative stress (Kassahun et al., 1991). Based on the results from this study, (E)-2,4-diene VPA appears to be the metabolite more likely to be responsible for both events. Thus, the detection of the NAC conjugates of (E)-2,4-diene VPA in urine could provide indirect but valuable information to assess the degree of exposure in patients on VPA therapy to the reactive (E)-2,4-diene VPA (Kassahun et al., 1991).
In brief, this study provided conclusive evidence to indicate that thiol conjugates of VPA in humans are formed primarily through the (E)-2,4-diene VPA route. The quantitative significance of NAC I and NAC II excreted as metabolites of VPA in humans will be addressed in future reports.
Acknowledgments
We thank Dr. A. Mutlib for assistance in the synthesis of the NAC conjugate of (E)-2,4-diene VPA methyl ethyl ester.
Footnotes
-
Send reprint requests to: Frank Abbott, Ph.D., The University of British Columbia, Faculty of Pharmaceutical Sciences, 2146 East Mall, Vancouver, British Columbia, Canada V6T 1Z3. E-mail:fabbott{at}interchange.ubc.ca
-
↵1 A preliminary account of these studies was presented at the 44th American Society for Mass Spectrometry conference, Seattle, Washington, 1996 and at the 7th International Society for the Study of Xenobiotics meeting, San Diego, California, 1996.
-
This work was supported by the Medical Research Council of Canada Research Grant MT-13744 and was part of the doctoral dissertation and postdoctoral fellowship of S.V. Gopaul (1998).
- Abbreviations used are::
- VPA
- valproic acid (2-propylpentanoic acid)
- CoA
- coenzyme A
- 4-ene VPA
- 2-propyl-4-pentenoic acid
- 2-ene VPA
- 2-propyl-2-pentenoic acid
- 3-ene VPA
- 2-propyl-3-pentenoic acid
- (E)-2,4-diene VPA
- (E)-2-propyl-2,4-pentadienoic acid
- 3-keto-4-ene VPA
- 2-propyl-3-oxo-4-pentenoic acid
- 4,5-diOH-VPA γ-lactone
- 3-n-propyl-5-hydroxymethyltetrahydro-2-furanone
- NAC
- N-acetylcysteine
- PFB
- pentafluorobenzyl
- PFBBr
- pentafluorobenzyl bromide
- t-BDMS
- tert-butyldimethylsilyl
- TMS
- trimethylsilyl
- TFA
- trifluoroacetic acid
- GC
- gas chromatography
- MS
- mass spectrometry
- LC/MS/MS
- liquid chromatography with tandem mass spectrometry
- EI
- electron impact
- NICI
- negative ion chemical ionization. NAC I, (E)-5-NAC-3-ene VPA
- NAC II
- (E)-5-NAC-2-ene VPA
- NAC III
- 5-NAC-4-OH-VPA γ- lactone
- NAC IVa
- 5-NAC-3-keto VPA
- NAC IVb
- 1-NAC-3-heptanone
- NAC V
- 5-NAC-3-ene VPA glucuronide
- HP
- Hewlett-Packard
- SIM
- selected ion monitoring
- MRM
- multiple reaction monitoring
- ES+
- positive electrospray
- Received August 24, 1999.
- Accepted April 12, 2000.
- The American Society for Pharmacology and Experimental Therapeutics