Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The use of valproic acid (2-n-propylpentanoic acid, VPA²; Fig. 1) for the treatment of seizures has been associated with an idiosyncratic hepatotoxicity that is usually characterized by hepatic microvesicular steatosis (Zimmerman and Ishak, 1982). It has been proposed that a metabolite (or metabolites) of VPA induce the associated hepatotoxicity (Lewis et al., 1982; Zimmerman and Ishak, 1982; Baillie, 1992). ⁴-VPA, an unsaturated metabolite of VPA, elicits hepatic steatosis in rats to a greater degree than VPA or other metabolites tested (Kesterson et al., 1984). The mechanism by which ⁴-VPA causes its hepatotoxic effects is proposed to involve metabolic activation of ⁴-VPA by the enzymes of fatty acid -oxidation, resulting in the formation of reactive species that bind covalently to important enzymes involved in fatty acid metabolism (Fig. 2). Evidence for the formation of chemically reactive metabolites of VPA and ⁴-VPA comes from covalent binding studies in isolated rat hepatocytes. Covalent binding was abolished in cells pretreated with 4-pentenoic acid (a potent inhibitor of -oxidation) and increased in incubations with hepatocytes from rats pretreated with clofibrate (an inducer of -oxidation) (Porubek et al., 1989). In addition, similar studies in isolated rat liver mitochondria showed that ⁴-VPA covalently binds to proteins by a process that is dependent on the presence of the cofactors of -oxidation (CoA, ATP, L-carnitine, and Mg²⁺) (Kassahun et al., 1994). Other experiments revealed that treatment of rats with ⁴-VPA leads to a depletion of total hepatic and mitochondrial glutathione (GSH), which supports a proposal that the depletion of GSH by reactive metabolites of ⁴-VPA may lead to the associated hepatotoxicity (Kassahun et al., 1991). In agreement with this finding are observations identifying the GSH conjugate of ^2,4-VPA in the bile of ⁴-VPA-dosed rats and the excretion of the corresponding N-acetylcysteine conjugate in the urine of patients treated with VPA (Kassahun et al., 1991). The authors proposed that a reactive diene metabolite coming from the -oxidation of ⁴-VPA-CoA, namely ^2,4-VPA-CoA, represents the reactive intermediate undergoing conjugation with GSH (Fig. 2). Acyl-CoA dehydrogenase enzymes catalyze the first step in mitochondrial fatty acid -oxidation by converting fatty acyl-CoA thioesters to their corresponding trans-2,3-enoyl-CoA derivatives (Schulz, 1990). These enzymes are believed to function by a mechanism where removal of the -proton by an active-site base is followed by transfer of the -hydride to a noncovalently bound flavin adenine nucleotide group (Thorpe, 1990).

View larger version (6K): [in this window] [in a new window]   Fig. 1.   Structures of compounds cited in the text.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Proposed scheme for the metabolic activation of VPA. Questionable effect of -fluoro substitution on the metabolic activation of 4-VPA. The F-4-VPA-CoA intermediate shown in brackets has not been identified.

The present work was prompted by an interesting study by Tang et al. (1995) on -fluorinated analogs as mechanistic probes in VPA-induced hepatotoxicity. The substitution of a fluorine atom at the -position to the carboxylic acid group provides a derivative that contains a chemically and enzymatically inert carbon center. The authors proposed that fluorine substitution at the -carbon would block its metabolism to the reactive ^2,4-VPA-CoA diene metabolite, via acyl-CoA dehydrogenases, and thus markedly reduce hepatotoxicity. Results from their studies showed that the -fluorinated analog of ⁴-VPA (F-⁴-VPA) does not cause hepatic steatosis, nor does it metabolize to the glutathione conjugate of the ^2,4-VPA diene in vivo in rats. These data support the acyl-CoA dehydrogenase-mediated metabolic activation of ⁴-VPA-CoA hypothesis. A critical part of the hypothesis that may explain the lack of hepatotoxicity induced by the fluoro derivative involves the formation of F-⁴-VPA-CoA (Fig. 2), although the ability of F-⁴-VPA to form the corresponding acyl-CoA thioester derivative has never been shown.

There have been reports indicating that -fluorinated carboxylic acid-containing compounds do not form acyl-CoA thioester conjugates. These reports include studies with -fluoropalmitic acid (Soltysiak et al., 1984) and perfluorooctanoic and perfluorodecanoic acids (Kuslikis et al., 1992) showing the lack of formation of their respective acyl-CoA thioester conjugates in vitro.

Hypotheses requiring the formation of the VPA-CoA thioester have been put forward to explain the mechanism of VPA-mediated hepatotoxicity and include VPA-induced CoA depletion (Thurston et al., 1985), carnitine depletion (Coulter, 1984), as well as competitive inhibition of one or several of the enzymes of fatty acid -oxidation by VPA-CoA or VPA-CoA metabolites (Becker and Harris, 1983; Bjorge and Baillie, 1985). Therefore, we propose that an absence of formation of the acyl-CoA derivative for the -fluoro analogs of VPA, or its hepatotoxic metabolites, could account for the associated lack of toxic effects. To test this hypothesis, the -fluoro analog of VPA was used in in vivo metabolic studies in the rat. The studies presented here were performed to determine whether F-VPA is able to be converted to F-VPA-CoA in vivo in rat liver.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. VPA, EDTA, triethylamine, ethyl chloroformate, potassium bicarbonate, diisopropylamine, butyllithium (1.6 M in hexanes), hydroxylamine hydrochloride, BSTFA, ibuprofen, and THF (anhydrous) were purchased from Aldrich Chemical (Milwaukee, WI). F-VPA, VPA-CoA, and F-VPA-CoA were synthesized as described below. CoA was purchased from Sigma (St. Louis, MO). N-Fluorobenzenesulfonimide was a gift from Allied Signal, Inc. (Buffalo, NY). All solvents used for HPLC were of chromatography grade.

Instrumentation and Analytical Methods. HPLC was carried out on a Shimadzu LC-600 isocratic system coupled to a Shimadzu SPD-6AV UV-Vis detector (Shimadzu, Kyoto, Japan). All HPLC analyses were performed on a reverse-phase column (Beckman C₈, 15 cm, 5 µ, 1 ml/min; Beckman Coulter, Inc., Fullerton, CA). Tandem liquid secondary-ion mass spectrometry was performed on a Kratos Concept IIHH four-sector tandem mass spectrometer (Kratos Analytical Instruments, Chestnut Ridge, NY) equipped with a cesium ion source and a continuous flow liquid secondary-ion mass spectrometry probe. The solvent used contained acetonitrile (5%), thioglycerol (2%), and trifluoroacetic acid (0.1%) in water. Samples were delivered at a flow rate of 3 µl/min using a syringe pump (Applied Biosystems, Foster City, CA). ESI/MS/MS analysis was performed on a Finnigan-MAT TSQ 7000 (San Jose, CA). HPLC purified samples were introduced into the ESI source via a Harvard Apparatus syringe pump (Holliston, MA) at a flow rate of 0.3 ml/min. CID of the ¹²C component of the protonated molecules-of-interest was performed in the collision cell. GC/MS analysis was conducted on a HP 5890 chromatograph (Hewlett Packard, Palo Alto, CA) interfaced with a Varian 70/70 magnetic sector mass spectrometer (Varian, Palo Alto, CA). Analyses were performed in the positive ion EI mode with an emission current of 350 µA. The GC column used was a fused silica capillary column (60 m × 0.32-mm i.d., 0.25-mm film thickness) coated with a DB-1 bonded stationary phase (J&W Scientific, Rancho Cordova, CA). Samples were analyzed as their trimethylsilyl esters after derivatization with BSTFA. GC conditions were 60°C for 0.5 min, 20°C/min to a final oven temperature of 290°C. Ibuprofen was used as the internal standard for quantitation of VPA and F-VPA, and all ions monitored corresponded to the [M  CH₃]⁺ fragment. Quantitative measurements were based on the ratios of peak areas (free acid/internal standard) relative to a linear calibration curve.

Synthesis of -F-VPA. F-VPA was synthesized by electrophilic -fluorination of ethyl valproate by an analogous method to that described by Tang et al. (1995) for the synthesis of 2-fluoro-2-propyl-4-pentenoic acid. GC/MS analysis of the synthetic product showed it to be 99% F-VPA with approximately a 1% impurity of VPA. GC/MS mass spectrum of F-VPA-trimethylsilyl ester, m/z (%): 73 (100), 219 (12) (M⁺-15), 117 (8). No molecular ion was detected. ¹H NMR (CDCl₃):  0.90 (t, 6H, J = 7.2 Hz, ---CH₃ groups), 1.2 to 1.6 (m, 4H, CH₃---CH₂---groups), 1.7 to 2.0 (m, 4H, CH₂---CH₂---groups). The 1% impurity of VPA could not be detected by ¹H NMR (complete absence of the  2.39 chemical shift for the -proton of VPA).

Synthesis of S-Acyl-CoA Thioester Derivatives of VPA and F-VPA. Synthesis of CoA thioesters was performed by conventional procedures employing ethyl chloroformate (Stadtman, 1957; Grillo, 1993). Briefly, triethylamine (1.6 mmol) followed by ethyl chloroformate (1.6 mmol) was added to the free acid (1.6 mmol) dissolved in anhydrous THF (25 ml) at room temperature and while stirring. After 30 min of continued stirring, the precipitate that formed (triethylamine hydrochloride) was removed by passing through a glass funnel, fitted with a glass wool plug, and added directly into a solution containing CoA (100 mg) and KHCO₃ (1.6 mmol) in distilled water (10 ml) and THF (15 ml). The solution was stirred continuously under nitrogen gas at room temperature for 2 h, after which the reaction was terminated by the addition of concentrated HCl (8 drops). The THF was then removed by evaporation under reduced pressure, and the remaining aqueous phase was extracted with diethyl ether (4 × 50 ml). Residual diethyl ether was removed by evaporation under reduced pressure at room temperature. The solution was adjusted to pH 7.0 by the addition of NaOH (1 N). Purification of the acyl-CoA thioester standards was achieved by reverse-phase HPLC using isocratic elution with acetonitrile (20%) in 0.2 M ammonium acetate on a reverse-phase column (C₁₈, 25 cm × 4.6 mm, 5 µm, 1 ml/min) and detected by UV absorbance (262 nm). CoA thioester-containing fractions were collected and desalted by passing through a cation exchange solid phase extraction cartridge (J. T. Baker, Phillipsburg, NJ). Desalted solutions containing CoA thioesters were then frozen (80°C) and lyophilized to dryness. MS/MS analysis VPA-CoA (CID of MH⁺ at m/z 894, 100%: m/z 136 ([adenine + H]⁺, 90%), m/z 387 ([M + H  adenosine triphosphate]⁺, 12%), m/z 428 ([adenosine diphosphate + 2H]⁺, 8%), and m/z 285 ([M + H  609]⁺, 6%). F-VPA-CoA (CID of MH⁺ at m/z 912, 100%: m/z 136 ([adenine + H]⁺, 100%), m/z 405 ([M + H  adenosine triphosphate]⁺, 100%), m/z 428 ([adenosine diphosphate + 2H]⁺, 42%), m/z 508 ([adenosine triphosphate + 2H]⁺, 36%), m/z 303 ([M + H  609]⁺, 14%), and m/z 565 ([adenosine monophosphate]⁺, 31%).

Analysis of Acyl-CoA Metabolites. After the administration of VPA or F-VPA (0.7 mmol/kg i.p., in distilled water, pH 7.0) to male Sprague-Dawley rats (200-220 g), and at 0.05-, 0.5-, 1-, 2-, and 5-h time points, rodents were anesthetized with diethyl ether, decapitated, and their livers immediately excised. Livers were then quickly frozen at 80°C and stored until analysis for acyl-CoA derivatives by HPLC and mass spectrometry. The preparation of liver samples for HPLC analysis was performed as follows: frozen livers were broken into small pieces (approximately 0.5 g), and 1.0-g portions (duplicate) were then added to ice-cold 7% perchloric acid (2.0 ml) and homogenized for 3 min on ice using a glass mortar and drill-driven Teflon pestle. The homogenized sample (1.5 ml) was then transferred to a microcentrifuge tube and centrifuged (14,000 rpm, 3 min). The resulting supernatant was transferred to a glass vial (20 ml) containing a solution of potassium phosphate buffer (0.05 M, pH 7.0, 5 ml). The pH of this solution was adjusted to 6.0 with 1 N NaOH. An aliquot of this solution (200 µl) was then immediately injected onto the HPLC (Beckman C₈, 15 cm, 5 µ, 1 ml/min). The acyl-CoA derivatives were eluted by a linear gradient whereby the acetonitrile concentration of the 0.2 M ammonium acetate mobile phase was increased from 0 to 50% over 50 min. All acyl-CoA derivatives were detected by UV analysis at 262 nm. Quantitative measurements of acyl-CoA derivatives detected in liver tissue were made using a standard curve generated from absolute peak areas. Acyl-CoA metabolites formed in vivo were analyzed by tandem mass spectrometry after purification by HPLC and further processing as described above (Grillo, 1993).

Analysis of Free Acids. Liver homogenates (prepared as above, 1.5 ml, duplicate) from VPA- and F-VPA-treated rats were added to 2 N NaOH (1.5 ml), to hydrolyze any acyl-CoA or acyl glucuronide esters, containing internal standard (ibuprofen, 5 µg), and the samples were left to stand at room temperature for 1 h. These solutions were then acidified (2 N HCl, 3 ml) and extracted with ethyl acetate (2 × 5 ml). The combined ethyl acetate extracts were dried (MgSO₄) and evaporated to dryness under nitrogen gas at room temperature. Diethyl ether (1 ml) was added to these dried samples, and the mixture reacted with BSTFA (300 µl) overnight at room temperature. Aliquots (10 µl) of these derivatized extracts were analyzed by GC/MS as described above.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Analysis of Acyl-CoA Metabolites in Liver Tissues of VPA- and F-VPA-Treated Rats. Livers of rats treated with VPA (100 mg/kg) or F-VPA (110 mg/kg) were analyzed for their respective acyl-CoA thioester derivatives at 0.05-, 0.5-, 1-, 2-, and 5-h post i.p. administration. HPLC analysis of liver extract from a VPA-dosed rat at the 0.5-h time point revealed the presence of VPA-CoA eluting at an HPLC retention time of ~22 min (Fig. 3). No substance was detected at this retention time in control liver extracts of rats treated only with water (pH 7.0, 1 ml i.p., data not shown). The identity of this metabolite was confirmed by comparison of the HPLC and tandem mass spectrometric characteristics of the biological and corresponding synthetic standard (Figs. 3 and 4). CID of the parent MH⁺ ion (m/z 894) of the HPLC-purified biological extract provided a mass spectrum characteristic of the VPA-CoA synthetic standard, which included ions at m/z 285, m/z 387, and m/z 428 originating from the proposed cleavages (Grillo, 1993) as shown in Fig. 4. The hepatic concentration-time profile of VPA-CoA in rat liver (Fig. 6) showed that the presence of VPA-CoA was maximal in liver tissue after 0.5 h (185 nmol/g of liver) and was still measurable 5-h postadministration (90 nmol/g of liver). The concentrations of VPA-CoA at the 1- and 5-h time points are approximately equal and are consistent with the relative VPA free acid concentrations in livers 1- and 5-h post-VPA administration (Fig. 6). In addition, the nadir in hepatic VPA-CoA concentration occurring at the 2-h time point is consistent with the nadir in the hepatic VPA free acid concentration. The secondary rise in VPA and VPA-CoA concentrations at the 5-h time point is due to a reabsorption of VPA, which is released by hydrolysis of the VPA-1-O-acyl glucuronide in the large bowel (Dickinson et al., 1979). In agreement with our hypothesis, F-VPA-CoA was not detected in liver extracts from F-VPA-treated rats (Fig. 5). The small amount of VPA-CoA that is detected during the HPLC analysis of the F-VPA rat liver extract (retention time 22 min) is presumably derived from the metabolism of the 1% impurity of VPA in the F-VPA synthetic test compound.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Representative chromatograms from the reverse-phase isocratic HPLC analysis of rat liver extract from a rat dosed with VPA (100 mg/kg, 0.5-h postadministration) (A) and synthetic VPA-CoA standard (B).

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Spectrum of product ions formed by CID of the MH+ ion (m/z 894) of VPA-CoA synthetic standard (A) and VPA-CoA isolated by HPLC purification from liver tissue of a rat dosed with VPA (B).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Representative chromatograms from the reverse-phase isocratic HPLC analysis of rat liver extract from a rat dosed with F-VPA (110 mg/kg, 0.5-h postadministration) (A) and synthetic F-VPA-CoA standard (B).

GC/MS Analysis and Quantitation of VPA and F-VPA in Rat Liver. The hepatic concentration-time profile of VPA and F-VPA in liver tissue at the 0.05-, 0.5-, 1-, 2-, and 5-h time points, as assessed by GC/MS analysis of the liver extracts for the trimethylsilyl derivatives of the test compounds, indicated that the concentrations of the VPA and F-VPA were similar throughout the 5-h experiment (Fig. 6). The concentration of VPA and F-VPA free acids in liver tissue extracts at the 1-h time point were determined to be 150 nmol/g of liver and 141 nmol/g of liver, respectively, although only the acyl-CoA derivative of VPA was detected (Fig. 6) at the same time point.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Hepatic concentration-time profile of VPA, F-VPA, and VPA-CoA in rat liver after i.p. administration of VPA (100 mg/kg) or F-VPA (110 mg/kg) to rats. Values are expressed as the average of duplicate experiments. The small amount of CoA thioester measured in the liver extracts from F-VPA-treated rats is presumably due to VPA-CoA formation from the 1% impurity of VPA in the administered synthetic F-VPA.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

A mechanism that may explain the hepatotoxicity associated with VPA therapy involves the formation of chemically reactive metabolites of the drug (Zimmerman and Ishak, 1982; Baillie, 1992). Most attention has focused on the unsaturated metabolite of VPA, ⁴-VPA, because this metabolite is metabolically activated to reactive electrophilic species that may covalently bind to and injure important cellular proteins (Fig. 2; Baillie, 1992). One hepatotoxic metabolite of ⁴-VPA is ^2,4-VPA, which has been shown to be reactive with nucleophiles such as glutathione (Kassahun et al., 1991). The conversion of ⁴-VPA to ^2,4-VPA first involves the formation of ⁴-VPA-CoA thioester that is catalyzed by acyl-CoA synthetases and requires the cofactors CoA, ATP, and Mg²⁺(Caldwell, 1984; Brass, 1994). ⁴-VPA-CoA then is metabolized by acyl-CoA dehydrogenases, the enzymes involved in the first step of fatty acid -oxidation (Thorpe, 1990), to ^2,4-VPA-CoA thioester. It has been proposed that ^2,4-VPA-CoA reacts with and depletes mitochondrial GSH resulting in hepatocellular injury (Kassahun and Abbott, 1993; Kassahun et al., 1994). Although these data provide evidence for ⁴-VPA-CoA metabolic activation to reactive and potentially toxic metabolites, there are data that do not support this hypothesis. Evidence against reactive unsaturated metabolites of VPA leading to VPA-induced hepatotoxicity comes from studies in rats (Loscher et al., 1993), where ⁴-VPA levels did not correlate with hepatic steatosis in VPA-treated animals, and in humans (Siemes et al., 1993), where increased amounts of ⁴-VPA could be detected only in one of the five patients with fulminant liver failure. To clarify the role of ⁴-VPA, and subsequently ^2,4-VPA, in VPA-induced hepatotoxicity, Tang et al. (1995) synthesized F-⁴-VPA and used it in histopathological, biochemical, and metabolic studies in rats. Because of the chemical and metabolic stability of the carbon-fluorine bond at the -position, the authors propose that fluorine substitution at the -carbon would preclude its metabolism to the reactive diene metabolite, via acyl-CoA dehydrogenases, and thus markedly reduce hepatotoxicity. Their results, all of which strongly support the ⁴-VPA bioactivation hypothesis, showed that F-⁴-VPA when given to rats resulted in no hepatic steatosis, depletion of liver mitochondrial glutathione, or formation of glutathione conjugates of the ^2,4-VPA metabolite (Fig. 2).

The ability of F-⁴-VPA to be converted to F-⁴-VPA-CoA thioester has not been directly determined but has been presumed to occur because the amino acid conjugate of F-⁴-VPA, N²-(-fluoro-2-propyl-pentenoyl)glutamine, has been isolated as a urinary metabolite from F-⁴-VPA-treated rats (Tang et al., 1997). In similar studies with mice, F-VPA was also shown to be converted extensively to the F-VPA-glutamine amide conjugate (Tang et al., 1997). Amino acid metabolites of acidic drugs are believed to occur through the intermediacy of the respective acyl-CoA thioester metabolites (Caldwell et al., 1979; Caldwell, 1984), and therefore F-⁴-VPA is proposed to have formed the intermediate CoA thioester derivative. These results are in contrast to literature reports showing examples that -fluorinated derivatives of carboxylic acid compounds do not form acyl-CoA thioester conjugates. These include studies with -fluoropalmitic acid (Soltysiak et al., 1984), shown not to be incorporated into triglycerides in cultured mammalian cells, an acyl-CoA-mediated process, and in studies with perfluoro-medium chain fatty acids (Kuslikis et al., 1992), shown not to form acyl-CoA thioesters in incubations with liver microsomes and the cofactors of acyl-CoA formation (CoA, ATP, and Mg²⁺). To our knowledge, it is not known whether amino acid conjugation can occur independently of the formation of reactive acyl-CoA intermediates, although results from studies with -fluoropalmitate support involvement of CoA thiol ester-independent steps in the modification of membrane lipids (Soltysiak et al., 1984), which are processes requiring a reactive acylating intermediate. -Fluoropalmitic acid was found to be incorporated without modification into membrane lipids such as phosphatidylcholine, sphingomyelin, neutral glycosphingolipids, and ceramides, independent of the formation of -fluoro-palmitoyl-CoA. It may be that the formation of the glutamine-amide conjugate of F-VPA occurs by a CoA-independent process.

Another potential mechanism for VPA-mediated hepatotoxicity comes from studies showing that acyl-CoA thioester derivatives of carboxylic acid-containing compounds are chemically reactive species that may contribute to the acylation of protein (Tishler and Goldman, 1970; Faed, 1984; Hertz and Bar-Tana, 1988; Bharadwaj and Bizzozero, 1995; Sallustio et al., 2000). In addition to the effects of VPA on CoA depletion (Thurston et al., 1985), carnitine depletion (Coulter, 1984), and competitive inhibition of fatty acid metabolism (Baillie, 1992), the chemical reactivity of VPA-CoA in transacylation-type reactions with protein nucleophiles may contribute to the covalent binding of VPA to protein in vitro and in vivo (Porubek et al.; 1989; Kassahun et al.; 1994; Bailey and Dickinson, 1996) and the associated hepatotoxicity (Nelson and Pearson, 1990; Baillie, 1992). The ability of VPA-CoA to transacylate protein nucleophiles remains to be evaluated.

We propose that the lack of hepatotoxic effects of the -fluoro-substituted derivative of ⁴-VPA may be due to a lack of the formation of the F-⁴-VPA-CoA thioester conjugate, rather than a block in the formation of the reactive diene, ^2,4-VPA-CoA. To test this hypothesis, we synthesized F-VPA and in the present studies tested for its ability to form the F-VPA-acyl-CoA derivative in vivo in rat liver. We assumed that the effect on acyl-CoA formation that occurs for F-VPA would also occur for F-⁴-VPA. Analysis of liver tissue extracts by HPLC from rats treated with VPA showed the presence of VPA-CoA (Fig. 3), but no F-VPA-acyl-CoA was detected in extracts from the F-VPA-treated animals (Fig. 5). Analysis for the presence of VPA and F-VPA free acids in liver tissue showed only small differences in their concentrations, indicating that F-VPA-CoA thioester is not formed in vivo in liver even though the free acid is present (Fig. 6). Furthermore, standard curves of the acyl-CoA synthetic standards isolated from acidified liver tissue were identical for VPA-CoA and F-VPA-CoA, indicating similar extraction efficiencies for both CoA thioesters (data not shown).

The inability of the -fluoro-substituted analogs to form acyl-CoA derivatives may result from the increased acidity of the carboxyl group (Schwartz et al., 1995; Tang et al., 1997), due to the electronegativity effects of the fluorine atom (Welch and Eswarakrishnan, 1991), which may influence the binding of the analog to the acyl-CoA synthetase enzymes (Soltysiak et al., 1984). The F-VPA analog was shown to be more acidic (pK_a = 3.55) than VPA (pK_a = 4.80), which may also help to explain the lack of the formation of F-VPA-acyl glucuronide relative to the extensive glucuronidation of VPA in vivo (Tang et al., 1997).

From these observations showing that the -fluoro analog of VPA does not form the respective acyl-CoA derivative in vivo and that -fluorination precludes hepatic steatosis (Tang et al., 1995), we propose that the metabolism of VPA, and unsaturated metabolites of VPA, by acyl-CoA formation may mediate the hepatotoxicity of the drug. Ongoing studies in our laboratory are designed to evaluate the abilities of VPA and unsaturated hepatotoxic metabolites of VPA to form acyl-CoA thioesters, in vitro and in vivo, in that differences in acyl-CoA formation may be related to differences in the ability to induce hepatotoxicity.