Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Although VPA is an effective anticonvulsant agent, a serious drawback of the drug is a rare but fatal hepatotoxicity characterized by liver microvesicular steatosis frequently accompanied by necrosis (Zimmerman and Ishak, 1982; Dreifuss et al., 1987; Dreifuss et al., 1989). Toxic metabolites have been implicated in the VPA-associated hepatotoxicity (Gerber et al., 1979). Experimentally, 4-ene VPA and (E)-2,4-diene VPA were demonstrated to induce massive lipid accumulation in rat liver (Kesterson et al., 1984). Expression of 4-ene VPA toxicity was suggested to require further biotransformation of 4-ene VPA via mitochondrial -oxidation to (E)-2,4-diene VPA (scheme 1). The reaction of (E)-2,4-diene VPA, possibly in the CoA thioester form, with glutathione in mitochondria could produce a localized depletion of glutathione in susceptible individuals that would result in oxidative stress with accompanying hepatocellular damage (Kassahun et al., 1990; Kassahun and Abbott, 1993). Alternatively, (E)-2,4-diene VPA may eventually be converted to 3-keto-4-ene VPA, a far more reactive species that would inhibit certain -oxidation enzymes (Baillie, 1988).

The toxic intermediate, (E)-2,4-diene VPA, may also arise from the microsomal cytochrome P450 catalyzed dehydrogenation of the -oxidation metabolite (E)-2-ene VPA (Kassahun and Baillie, 1993), and the diene metabolite could react with hepatic glutathione through a glucuronide mediated pathway (Tang and Abbott, 1996) (scheme 1). (E)-2-Ene VPA was shown to be possibly involved in the inhibition of mitochondrial medium-chain acyl-CoA synthase (Ponchaut et al., 1992).

View larger version (16K): [in this window] [in a new window]   Scheme 1.   Bioactivation of VPA leading to hepatotoxic metabolites which are prevented by -fluorination. CoA, coenzyme A; GSH, glutathione; GST, glutathione S-transferase.

Based on this proposed mechanism of VPA hepatotoxicity, it was postulated that preventing the -oxidation of VPA would either eliminate or markedly reduce the toxicity (Tang et al., 1995). In support of this hypothesis, a fluorine substituent introduced at the -position of 4-ene VPA resulted in a non-hepatotoxic analogue (2-fluoro-2-propyl-4-pentenoic acid) as judged by the absence of hepatic microvesicular steatosis in rats chronically administered this compound (Tang et al., 1995). In this study, -fluoro VPA was synthesized for administration to mice in the hope of averting metabolism via -oxidation and therefore preventing this fluorinated VPA analogue from generating toxic metabolite(s) in the liver (scheme 1). The drug was subsequently tested as an anticonvulsant in mice using PTZ-induced seizure model. A time course of -fluoro VPA disposition in mouse brain and serum was also established and correlated with the observed pharmacological activity of the compound.

The anticonvulsant activity of VPA is suggested to be associated with its capability of elevating synaptosomal GABA levels in the brain (Cotariu et al., 1990; Loscher, 1993). It was illustrated that synaptosomal GABA levels were increased 20 to 30% in mice administered VPA at doses of 125 to 290 mg/kg (Loscher, 1981). Because the synthesis, uptake and degradation of GABA occur in both GABA-containing nerve terminals and cells that do not use this compound as an neurotransmitter (Baxter, 1976), only the elevation of GABA induced by VPA in the synaptosomal fraction is considered to be directly associated with its anticonvulsant activity (Loscher and Vetter, 1985). Thus, in addition to a pharmacokinetic study, this report presents a correlation of the suppressive effects of -fluoro VPA against PTZ-promoted seizures with drug-induced elevations in brain synaptosomal GABA levels. VPA was included in this study for the sake of comparison.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. Sucrose, 3-mercaptopropionic acid, Triton X-100 and TFA were purchased from Sigma Chemical Co. (St. Louis, MO). Ethylenediaminetetraacetic acid and trichloroacetic acid were obtained from British Drug Houses, Inc. (Vancouver, BC). PTZ, tert-butyldimethylsilyl chloride, dicyclohexylcarbodiimide, N-hydroxysuccinimide, L-glutamine, butyric acid, valeric acid, ethylbutyric acid, hexanoic acid, ethylhexanoic acid and VPA were the products of Aldrich Chemical Co. (Milwaukee, WI), and the VPA was distilled before use. 4-Aminobutyric acid-2,2,3,3,4,4-²H₆ ([²H₆]GABA) was obtained from MSD Isotopes (Montreal, Canada), N-methyl-N-tert-butyldimethylsilyltrifluoroacetamide from Pierce (Brockville, Ontario, Canada), N, N-bis(trimethylsilyl)trifluoroacetamide-²H₁₈ from Caledon (Edmonton, Alberta, Canada) and N-Fluorobenzenesulfonimide from Allied Signal Inc. (Buffalo, NY).

Instrumentation. NMR spectra were obtained on a Bruker WH-200 spectrometer and high resolution mass spectra were recorded on a Kratos MS 50 mass spectrometer (70 eV, 150°C) in the Department of Chemistry at the University of British Columbia (UBC). Chemical shifts are expressed relative to tetramethylsilane for ¹H and ¹³C NMR and trifluoroacetic acid for ¹⁹F NMR.

4

Synthesis. -Fluoro VPA was synthesized through electrophilic fluorination of ethyl 2-propylpentanoate with N-fluorobenzenesulfonimide (Differding and Ofner, 1991; Tang et al., 1993) followed by alkaline hydrolysis. High resolution mass spectrum of the tBDMS ester (formed by derivatization with N-methyl-N-(tert-butyldimethylsilyl)trifluoroacetamide): m/z (M-tBu)⁺: 219.1212, calc: 219.1217; (FSi(CH₃)₂)⁺: m/z 77.0227, calc: 77.0223; and the [²H₉]-TMS ester (formed by derivatization with N, N-bis(trimethylsilyl)trifluoroacetamide-²H₁₈): (M-C[²H ₃])⁺: m/z 225.1607, calc: 225.1587; (FSi(C[²H₃])₂)⁺: m/z 83.0583, calc: 83.0593. ¹H NMR (C[²H]Cl₃):  0.89 (t, 6H, J_HH = 7.0 Hz, 2 × CH₃), 1.20-1.60 (m, 4H, 2 × CH₂CH₃), 1.78-2.05 (m, 4H, CH₂CH₂). ¹³C NMR (C[²H]Cl₃):  13.95 (s, C-5, 5), 16.52 (d, J_CF = 3.5 Hz, C-4, 4), 39.28 (d, J_CF = 22.1 Hz, C-3, 3), 97.55 (d, J_CF = 185.7 Hz, C-2), 177.72 (d, J_CF = 26.8 Hz, C-1). ¹⁹F NMR (C[²H]Cl₃):  -90.08.

Determination of the lipophilicity and the ionization constants of VPA and -fluoro VPA. The octanol-water partition coefficients (P) of VPA and -fluoro VPA were determined from the HPLC capacity factors (Abbott and Acheampong, 1988). Briefly, a standard curve which correlated the HPLC capacity factors of compounds (butyric acid, 0.998; valeric acid, 1.721; ethylbutyric acid, 2.438; hexanoic acid, 2.919; ethylhexanoic acid, 6.958) with their literature log P values (0.98, 1.51, 1.68, 1.93, 2.64) (Keane et al., 1983) was constructed with a coefficient of determination r² > 0.99 and used to calculate the log P values of VPA and -fluoro VPA. The assay was performed using a Hewlett Packard Spherisorb ODS2 column (250 × 4 mm, 5 µm). The mobile phase consisted of acetonitrile-phosphate buffer (50 mM, pH 2.5, 45/55, v/v) at a flow rate of 1.0 ml/min.

Animals. Mature male mice (CD-1, Charles River, Quebec, Canada) were housed in regular cages except those in the metabolic study where the animals were housed in metabolic cages, and allowed free access to food and water.

Anticonvulsant activity test. Anticonvulsant activity was evaluated using the PTZ seizure threshold test (Swinyard et al., 1989). The sodium salt of VPA or its fluorinated analogue in aqueous solution (pH 7.0) was administered i.p. to mice (eight mice/dose) at different concentrations. PTZ (85 mg/kg) was given s.c. at different time intervals after the anticonvulsant dose and the animals were observed for an additional 30 min for seizures that were rated as: 0, no seizure; 0.5, clonic seizure, animal remained on its feet; 1, tonic-clonic seizure, animal lost its balance (Loscher and Vetter, 1985). The percentage of animals protected from PTZ induced seizures was plotted against the logarithm of dose to obtain the ED₅₀ of the tested drug. Seizures occurred at a mean time of 7.2 ± 4.3 min in control mice following the administration of PTZ.

Isolation of mouse brain synaptosomes and GABA assay. Mouse brain synaptosomes were isolated through the sedimentation of a postnuclear fraction in a discontinuous sucrose gradient (Dodd et al., 1981). Mice were decapitated into ice-cold saline (0.9% sodium chloride) at different time intervals after receiving VPA or -fluoro VPA by i.p. injection. The brain was removed and homogenized in 4 ml of sucrose (0.32 M) containing 3-mercaptopropionic acid (1 mM, pH 7.0) at 0°C. The homogenate was made up to a total volume of 6.5 ml and a 500 µl aliquot stored at 78°C for the assay of brain drug concentration. The remaining homogenate was centrifuged at 1000 × g for 10 min, the resultant supernatant applied to 9.5 ml of sucrose (1.2 M) and centrifuged again at 220,000 × g. The fluffy interface was collected, diluted with sucrose (0.32 M) to 10 ml, layered on top of 9.5 ml of sucrose (0.8 M), and centrifuged as before to pellet the synaptosomes.

GC/MS detection and quantitation of VPA, -fluoro VPA and the metabolite (E)-2-ene VPA in mouse brain, serum and urine. Mouse brain and serum samples were prepared at selected time points over a period of 0 to 400 min, although urine was collected for 24 hr after the i.p. administration of the drugs.

LC/MS/MS detection and quantitation of the glutamine conjugate of -fluoro VPA. For detection and quantitation of the L-glutamine conjugate, mouse urine (50 µl) was mixed with an equivalent volume of aqueous TFA (0.05%) and the precipitates were removed via centrifugation at 13,600 × g for 15 min. An aliquot (2-5 µl) of each sample was injected onto a Hewlett Packard Hypersil ODS column (100 × 2.1 mm, 5 µm) and delivered at 50 µl/min. The HPLC mobile phase consisted of methanol/water (0.05% TFA) and was programed: 75% water for 2 min, a gradient decrease to 50% water during 2-5 min, a hold at 50% water to 6 min, a gradient decrease to 15% water during 6-8 min, and a hold at 15% water to 30 min.

Pharmacokinetic calculations. The pharmacokinetic parameters for -fluoro VPA were calculated using a two compartment model (fig. 1). The mass balance differential equations for this model were: where C₁ is the drug concentration in compartment 1 at time t; A₂ is the amount of the drug in compartment 2 at time t; k₁₀ is the first-order rate constant for drug elimination from compartment 1; k₁₂ and k₂₁ are the first-order rate constants for the transfer of the drug between the two compartments; k_a is the first-order rate constant for drug absorption into compartment 1; and V₁ is the distribution volume of compartment 1. The equations were solved by the MULTI (RUNGE) program (Yamaoka and Nakagawa, 1983) which allowed simultaneous curve-fitting for two sets of experimental data.

View larger version (8K): [in this window] [in a new window]   Fig. 1.   Two-compartment model for absorption, distribution and elimination of -fluoro VPA in mouse serum and brain after an i.p. dose of 0.83 mmol/kg, where k10 is the first-order elimination rate constant, k12 and k21 are the first-order rate constants for the transfer of the drug between the two compartments, and ka is the first-order absorption rate constant.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Ionization constants and lipophilicity. The apparent pK_a value for VPA was determined to be 4.80, which is consistent with the literature value of 4.56 to 4.8 (Kupferberg, 1989). The introduction of a fluorine atom, a strong electron-withdrawing substituent, into the -position of VPA decreased the pK_a by more than 1 pH unit (table 1), reflecting the increased acidity of -fluoro VPA in comparison with VPA. The log P value, a measure of the lipophilicity of a molecule, was found to be 2.65 for VPA at both pH 4.0 and pH 2.5, in agreement with the value of 2.75 obtained by the shake-flask procedure (Keane et al., 1983). Under the conditions used, the "log P" value for -fluoro VPA appeared to be pH-dependent due to its relatively lower pK_a, being 2.07 at pH 4.0 and 2.44 at pH 2.5. Because P is the intrinsic partition coefficient of the unionized form of a compound between an aqueous buffer and an organic phase (Jezequel, 1992), the log P value determined at pH 2.5 would thus be a better choice to reflect the lipophilicity of -fluoro VPA.

View this table: [in this window] [in a new window]   TABLE 1 pKa and log P values for VPA and -fluoro VPA

Structural characterization and quantitation of the glutamine conjugate of -fluoro VPA. An earlier metabolic study conducted in rats revealed that -fluorinated VPA analogues could undergo conjugation with L-glutamine (Tang and Abbott, 1993). In this investigation, LC/MS/MS screening for urinary metabolites in mice treated with -fluoro VPA directed our attention to the ion m/z 291 which appeared to match the molecular weight of protonated -fluoro VPA-Gln. The collision induced dissociation of m/z 291 of the putative glutamine conjugate was characterized by scission of the -amino group on the glutamine moiety with charge retention occurring possibly on either side of the molecule. Thus, the neutral loss of 2-fluoro-2-propylpentanamide produced a prominent daughter ion at m/z 130, whereas the fragment at m/z 162 was indicative of the formation of protonated 2-fluoro-2-propylpentanamide (fig. 2). The neutral loss of N-formyl glutamine would result in the daughter ion at m/z 117 and, following a further loss of hydrogen fluoride, produce the fragment at m/z 97 (fig. 2). An identical fragmentation pattern was observed in the LC/MS/MS mass spectrum of the synthetic reference compound.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   MS/MS mass spectrum of the urinary metabolite N2-(2-fluoro-2-propyl-4-pentenoyl)glutamine in mice treated with -fluoro VPA at 0.83 mmol/kg. Fragmentation was identical to the synthetic reference compound, and is discussed in the text.

View this table: [in this window] [in a new window]   TABLE 2 Metabolites of VPA and -fluoro VPA detected in the urine of micea,b

Detection and pharmacokinetic characterization of VPA and -fluoro VPA in mouse brain, serum and urine. An abundant fragment ion, m/z 77, was found in the GC/MS mass spectra of the TMS and tBDMS derivatives of -fluoro VPA (fig. 3a). High resolution mass spectra of the derivatives, including the [²H₉]TMS derivative, confirmed that the m/z 77 ion possessed the chemical structure [F-Si(CH₃)₂]⁺, presumably resulting from the migration of the fluorine ion of the halo acid tBDMS or TMS derivative to the charged silicon center (fig. 3b) (White and McCloskey, 1970). This ion was used in the GC/MS selective ion monitoring mode for the quantitation of -fluoro VPA and proved to be more sensitive than the typical [M-57]⁺ ion at m/z 219 but equally reliable. For example, at 60 min after a dose of 0.83 mmol/kg, the brain concentration of -fluoro VPA was determined to be 0.37 ± 0.03 µmol/g of tissue using m/z 77 and 0.37 ± 0.02 µmol/g of tissue using m/z 219 for the analysis.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   (a) GC/MS mass spectrum of the tBDMS derivative of -fluoro VPA (upper panel) and (b) mechanism proposed for the formation of the m/z 77 fragment ion from the tBDMS derivative of -fluoro VPA (lower panel).

1

4

1

1

3

1

View larger version (10K): [in this window] [in a new window]   Fig. 4.   (a) Drug concentration-time profiles of -fluoro VPA in mouse brain and serum after an i.p. dose of 0.83 mmol/kg. , -Fluoro VPA in the serum (µg/ml); , -fluoro VPA in the brain (µg/g of tissue). Each value represents an average of four animals. The solid lines represent the pharmacokinetic model curve-fitting. (b) Drug concentration-time profiles of VPA in mouse brain and serum and its metabolite (E)-2-ene VPA in the serum after an i.p. dose of VPA of 0.83 mmol/kg. , VPA in the serum (µg/ml); , VPA in the brain (mg/g of tissue); , (E)-2-ene VPA in the serum (µg/ml). Each value represents an average of four animals.

2

1

2

1

Anticonvulsant activity and elevation of synaptosomal GABA levels. VPA anticonvulsant testing in mice was performed by injecting PTZ s.c. at 10 min after the anticonvulsant dose and was followed by a 30-min observation period (Swinyard et al., 1989). The ED₅₀ of VPA was determined to be 0.83 mmol/kg, in agreement with the literature value of 0.83 to 1.04 mmol/kg (Swinyard et al., 1989; Loscher et al., 1991) (fig. 5). The anticonvulsant activity of -fluoro VPA was evaluated by injecting PTZ at 45 min after the anticonvulsant dose, the time point at which the drug reached its apparent peak concentration in the brain. The ED₅₀ estimated for -fluoro VPA was 1.70 mmol/kg (fig. 5).

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Evaluation of the anticonvulsant activities for VPA (209) and -fluoro VPA (). The sodium salt of VPA or -fluoro VPA in aqueous solution was administered i.p. to mice at different concentrations. PTZ (85 mg/kg) was given s.c. at 15 min (VPA) or 45 min (-fluoro VPA) and the animals were observed for an additional 30 min for seizures that were rated as: 0, no seizure; 0.5, clonic seizure, animal remained on its feet; 1, tonic-clonic seizure, animal lost its balance. Seizures occurred at a mean time of 7.2 ± 4.3 min in control mice after the dose of PTZ.

View this table: [in this window] [in a new window]   TABLE 3 Mouse brain synaptosomal GABA levels after administration of VPA or -fluoro VPA

View this table: [in this window] [in a new window]   TABLE 4 Relationship between protection against PTZ-induced seizures, synaptosomal GABA levels and brain -fluoro VPA concentrations

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The hepatotoxicity of VPA has been suggested to be due to the generation of toxic metabolite(s) via -oxidation (Baillie, 1988; Kassahun et al. 1990). Recent data derived from in vivo metabolic and toxicological studies in rats indicated that -fluorination of 4-ene VPA prevented (E)-2,4-diene VPA formation and eliminated the steatogenic effect of 4-ene VPA (Tang et al. 1995). Our investigation, -fluoro VPA, synthesized for evaluation of anticonvulsant activity, was observed to have a suppressive effect against PTZ-induced seizures in mice. This drug, however, is characterized by a slower onset of the pharmacological action as compared to VPA. Subsequent kinetic studies indicated that the delayed action was due largely to a slower uptake of -fluoro VPA by the brain.

The brain is efficiently separated from the blood by the blood-brain barrier, which is known to have several transport systems regulating the passage of endogenous and exogenous compounds (Jezequel, 1992; Spector, 1990). It was reported that at physiological pH VPA could cross the blood-brain-barrier through the monocarboxylic acid carrier system at the cerebral capillaries (Terasaki et al. 1991). More recently, data were presented to suggest that uptake of VPA into rat brain was mediated by a medium-chain fatty acid transporter (Adkison and Shen, 1996). However, the relatively low brain concentration of VPA, as compared to other anticonvulsant drugs such as phenytoin or phenobarbital, led to investigations that concluded that the transport of VPA was asymmetric such that the brain-to-blood efflux exceeded the blood-to-brain influx (Cornford et al., 1985). Recent data indicate that a probenecid-sensitive anion transporter at brain capillary endothelium may be involved in the clearance of VPA from the central nervous system (Adkison et al. 1994, 1996). In an earlier pharmacokinetic study, VPA was demonstrated to enter the CSF rapidly and leave at a rate parallel to that disappearing from the plasma (Levy, 1980). In accordance with this finding, it was observed in our investigation that the arrival of the peak VPA concentrations in the serum and brain appeared within 15 min of dosing and the apparent elimination rate constants for these compartments were comparable with each other.

The substitution of the -hydrogen in VPA with a fluorine atom should result in little steric effects (Welch, 1990). The lipophilicities of VPA and -fluoro VPA were shown to be comparable. Although VPA has a higher pK_a, both compounds should be present almost exclusively in their ionized forms at physiological pH. If -fluoro VPA is transported into the brain by a carrier-mediated system similar to that responsible for VPA transportation, the slower access of -fluoro VPA into the brain could thus be due to either serum drug-protein binding which would decrease free drug concentration available for the transportation or a low affinity of the drug toward the active transport carriers. The results derived from the pharmacokinetic calculations indicated that k₁₂, a measure of blood-to-brain influx, was smaller than k₂₁, an index of brain-to-blood efflux, implying that -fluoro VPA, as with VPA, might also be asymmetrically transported across the blood-brain barrier. Such an asymmetric transport was further reflected in the lower brain/serum drug concentration ratio of -fluoro VPA than that of VPA.

Unlike VPA, for which the brain/serum concentration ratio increases rapidly after the dose followed by a decrease with time (Hammond et al., 1982), the ratio for -fluoro VPA was observed to reach a plateau. These results give rise to the speculation on whether mechanisms other than carrier-mediated processes are involved in transport of the drug into the brain. Finally, it should be noted that the low brain/serum concentration ratio of -fluoro VPA, as with its slow uptake by the brain, could also be under the influence of serum protein binding. Attempts to characterize the serum protein binding properties of -fluoro VPA was not feasible at this time owing to the limited availability of mouse blood. Further investigations are deemed necessary.

Serum drug-protein binding may also be partially responsible for the slower elimination of -fluoro VPA from the general circulation. Although the apparent elimination of -fluoro VPA from the brain paralleled its disappearance from the serum (fig. 4a), the pharmacokinetic model-generated micro-constant k₂₁ was greater than the serum elimination constant k₁₀, suggesting that clearance of -fluoro VPA from the brain was limited by the clearance from the general circulation.

Previously, the fluorinated compound was observed to undergo little glucuronidation and was not a substrate of the -oxidation enzymes in rats (Tang et al., 1993), whereas VPA is known to be extensively metabolized in human and laboratory animals (Baillie and Rettenmeier, 1989). In this study, the -oxidative metabolite, (E)-2-ene VPA, was detected in the serum and urine of mice administered VPA but not in -fluoro VPA-treated animals. It would be reasonable to assume that this interruption of the metabolism of -fluoro VPA via -oxidative pathway shifted its CoA thioester to conjugate with L-glutamine, because acyl CoA forms are generally believed to be the precursors of acyl amino acid conjugates (Hutt and Caldwell, 1990). Acyl-CoA: L-glutamine N-acyltransferase, the enzyme catalyzing acyl transfer from acyl-CoA to L-glutamine, is located in liver mitochondria (Webster et al. 1976) wherein -fluoro VPA CoA ester, as with VPA CoA ester, could possibly be formed. In contrast to the extensive glucuronidation of VPA, which would facilitate the clearance of the drug, little -fluoro VPA glucuronide conjugate was formed, indicating that the observed distinction between VPA and its -fluorinated analogue in both phase I and phase II metabolism in mice may account for the apparent differences in elimination. The slower clearance of -fluoro VPA from the circulation may be an advantage for the drug in terms of the duration of its pharmacological activity.

It is considered that the elevation of synaptosomal GABA levels is relevant to the anticonvulsant activity of VPA (Loscher, 1993). The synaptosomal GABA was observed to be increased 20 to 30% in mice treated with VPA at doses of 125 to 290 mg/kg (Loscher, 1981; Loscher et al., 1981). Clinical relevance of the GABA hypothesis resides in the observation that CSF GABA levels in 33 epileptic children on VPA therapy were 2-fold higher than those of untreated patients and those treated by anticonvulsant drugs other than VPA (Loscher and Siemes, 1984).However, a direct effect of VPA on neuron membranes has also been postulated. VPA was demonstrated to induce hyperpolarization in the resting membrane potential, possibly due to increased potassium conductance (Slater and Johnston, 1978; Walden et al., 1993). It was also observed that VPA at therapeutic concentrations (6-200 µM) could limit sustained repetitive firing in mouse central (spinal cord and cortical) neurons without affecting postsynaptic GABA responses (MacDonald, 1988; McLean and MacDonald, 1986).

In agreement with previously reported data (Gale and Iadarola, 1980), VPA was observed in this investigation to be able to simultaneously protect mice against PTZ induced seizures and increase mouse brain synaptic GABA content after an ED₅₀ dose (table 3). This result may be viewed as another piece of evidence which is in support of the hypothesis that GABA-elevation is one of the mechanisms of VPA action.

The GABA-elevation mechanism could also be used to interpret the observed anticonvulsant activity of the -fluorinated analogue. Administration of -fluoro VPA increased synaptic GABA levels, and at 2.08 mmol/kg doses the anticonvulsant effect of -fluoro VPA was demonstrated to have an association with the elevation of synaptosomal GABA instead of the peak brain concentration of the drug (table 4). A delay in the elevation of synaptic GABA compared with the brain drug concentration may be due to either a slow distribution of -fluoro VPA within the brain or to unknown regulation factors that either govern the interaction of the drug with as yet undetermined receptors or limit the entry rate of the drug into the target cells. However, an argument can be made against the GABA theory on the basis that, despite an elevation of brain synaptosomal GABA, no protection was provided to mice by -fluoro VPA at the dose of 0.83 mmol/kg. Such paradoxical observations appeared to imply that more than one mechanism of action or other mechanisms of action are components of the detected pharmacological effect of -fluoro VPA. The broad spectrum of anticonvulsant activity of VPA itself has been suggested to be indicative of several mechanisms of action being involved (Loscher, 1993; Wolf and Tscherne, 1994).

In conclusion, -fluoro VPA was demonstrated to have anticonvulsant activity in the PTZ seizure test and to be capable of increasing brain synaptic GABA, but the connection between these two events remains to be clarified. The delayed maximal effect of -fluoro VPA is mainly attributed to its slow access into the brain. The relatively slower elimination of -fluoro VPA compared to VPA is probably due to the absence of -oxidation and the markedly reduced glucuronidation of -fluoro VPA. The resulting persistence of this drug in the circulation and in brain may represent a distinct therapeutic advantage. Because preliminary toxicity studies in rats suggest that -fluorinated VPA analogues are not hepatotoxic (Tang et al., 1995), -fluoro VPA has potential as a new anticonvulsant drug.