Wolfgang Löscher PhD
Professor and Chairman, Department of Pharmacology, Toxicology and Pharmacy, School of Veterinary Medicine, Hannover, Germany
Valproic acid or valproate (VPA) is the common name for 2-propylpentanoic acid (also called n-dipropylacetic acid). As a simple branched-chain carboxylic acid, it differs markedly in structure from all other antiepileptic drugs in clinical use. VPA was first synthesized in 1882 by Burton (1), but there was no known clinical use until its anticonvulsant activity was fortuitously discovered by Pierre Eymard in 1962 in the laboratory of G. Carraz, a discovery published by Meunier et al. (2). The first clinical trials of the sodium salt of VPA were reported in 1964 by Carraz et al. (3). The drug was marketed in France in 1967 and was released subsequently in more than 100 other countries (in the United States in 1978) for the treatment of epilepsy. Since then, VPA has established itself worldwide as a major antiepileptic drug for patients with several types of epileptic seizures, including both partial and generalized seizures. Clinical experience with VPA has continued to grow, including the use of VPA for diseases other than epilepsy, such as in bipolar disorders and migraine.
Although limitations of space do not permit a detailed review of the numerous effects of VPA in diverse experimental preparations, this chapter summarizes the drug's major properties and effects on nervous tissue, with particular emphasis on VPA's actions that appear to be of importance for its anticonvulsant effect. For a more comprehensive survey of the multiple effects of VPA, several previous reviews and monographs are available (4, 5, 6, 7).
MULTIPLE ACTIONS OF VALPROIC ACID AND ITS METABOLITES
Experimentally, VPA exerts anticonvulsant effects in almost all animal models of seizure states examined in this respect, including models of different types of generalized seizures as well as focal seizures (6). The anticonvulsant potency of VPA strongly depends on the species, the type of seizure induction, the seizure type, the route of administration, and the time interval between drug administration and seizure induction. Because of the rapid penetration into the brain but the short half-life of VPA in most species (6), the most marked effects are obtained shortly (i.e., 2 to 15 minutes) after intraparenteral (i.p.) injection. Depending on the preparation, the onset of action after oral administration may be slower. In most laboratory animal species, the duration of anticonvulsant action of VPA is short, so high doses of VPA are needed to suppress long-lasting or recurring seizures in animal models. In general, the anticonvulsant potency of VPA increases in parallel with the size of the animal. In rodents, the highest anticonvulsant potencies are obtained in genetically seizure-susceptible species, such as gerbils, with an anticonvulsant median effective dose (ED50) of 73 mg/kg i.p., and rats with spontaneously occurring spike-wave discharges (ED50 80 mg/kg i.p.), and against seizures induced by the inverse benzodiazepine receptor agonist DMCM in mice (ED50 60 mg/kg i.p.), whereas ED50 doses in other rodent models of generalized or focal seizures are usually substantially higher, in the range of 200 to 400 mg/kg (6).
In addition to animal models of generalized or focal seizures, VPA also has been evaluated in models of status epilepticus. As shown by Hönack and Löscher (8) in a mouse model of generalized convulsive (grand mal) status epilepticus, VPA, by intravenous injection, was as rapid as benzodiazepines in suppressing generalized tonic-clonic seizures; this effect was related to the instantaneous entry of VPA into the brain after this route of administration. In view of the different mechanisms presumably involved in anticonvulsant activity of VPA against different seizure types, the situation may be different for other types of status epilepticus, because not all cellular effects of VPA occur rapidly after administration. This finding is substantiated by accumulating clinical experience with parenteral formulations
of VPA in treatment of different types (e.g., convulsive versus nonconvulsive) of status epilepticus.
Whereas most reports dealing with VPA's anticonvulsant activity in animal models examine the acute short-lasting anticonvulsant effects after single-dose administration, several studies have evaluated the anticonvulsant efficacy of VPA during long-term administration. During the first days of treatment of amygdala-kindled rats, a marked increase in anticonvulsant activity was observed that was not related to alterations in brain or plasma drug or metabolite levels (9,10). Similarly, when anticonvulsant activity was measured by means of timed intravenous infusion of pentylenetetrazol (PTZ), prolonged treatment of mice with VPA resulted in marked increases in anticonvulsant activity on the second day of treatment and thereafter compared with the acute effect of VPA, although plasma levels measured at each seizure threshold determination did not differ significantly (11). This late effect of VPA developed irrespective of the administration protocol (once per day, three times per day, continuous infusion) used for treatment with VPA in the animals. Such an increase in anticonvulsant activity during long-term treatment was also observed in epileptic patients and should be considered when acute anticonvulsant doses or concentrations of VPA in animal models are compared with effective doses or concentrations in epileptic patients during long-term treatment. In other words, doses or plasma levels that are ineffective after acute administration can become effective during long-term administration. The possible mechanisms involved in early (i.e., occurring immediately after first administration of an effective dose) and late (i.e., developing during long-term administration) anticonvulsant effects of VPA are discussed later in this chapter. Early and late effects of VPA have been also observed in in vitro preparations (12,13).
In addition to short- or long-term anticonvulsant effects in animal models of seizures or epilepsy, data from the kindling model indicate that VPA may exert antiepileptogenic effects (14).
The active concentrations of VPA in the brain or plasma strongly depend on the model examined. When a VPA-sensitive model, such as the threshold for clonic seizures determined by intravenous infusion of PTZ in mice, is used, the drug concentrations in brain tissue after administration of effective doses are close to the range of effective concentrations determined in brain biopsies of epileptic patients, which are in the range of 40 to 200 µmol/L (6). However, because of the marked differences in pharmacokinetics of VPA between rodents and humans (rodents eliminate VPA about 10 times more rapidly than humans), the doses that have to be administered to reach these brain concentrations in mice or rats are much higher (~100 to 200 mg/kg) than respective doses in humans (~20 mg/kg). Such determinations of effective brain concentrations are important for interpretation of in vitro data on VPA, because the neurochemical or neurophysiologic effects of VPA found in vitro are only of interest if they occur in concentrations that are reached in vivo at anticonvulsant (nontoxic) doses.
In addition to its anticonvulsant activity, VPA exerts several other pharmacodynamic effects in animal models, including anxiolytic, antiaggressive, anticonflict, antidystonic, antinociceptive, sedative-hypnotic, immunostimulating, and antihypertensive actions (6). Several of these preclinical actions are in line with VPA's therapeutic potential in indications other than epilepsy (7,15).
Because VPA is rapidly metabolized to various pharmacologically active metabolites in vivo (16), these substances have to be considered when mechanisms of action of VPA are discussed. One of the major active metabolites of VPA in plasma and central nervous system (CNS) of different species, including humans, is the trans isomer of 2-en-VPA (E-2-en-VPA). This compound is the most potent and most extensively studied active metabolite of VPA (6,17,18). Trans-2-en-VPA is effective in the same seizure models as VPA, often with higher potency than the parent drug. Accordingly, in most neurochemical and neurophysiologic experiments with trans-2-en-VPA, the compound exerted more potent effects than VPA (6).
The precise mechanism of action of VPA or its active metabolites, like that of many other antiepileptic drugs, is unknown. Much attention has focused on VPA's effects on γ-aminobutyric acid (GABA), one of the principal inhibitory neurotransmitters in the CNS. However, given VPA's various experimental and clinical effects and its numerous effects on neuronal tissue, no single action of VPA can completely account for these effects.
EFFECTS ON EPILEPTIFORM DISCHARGES
Various in vitro preparations were used to study the anticonvulsant action of VPA on epileptiform discharges. In slices prepared from guinea pig brain, VPA was shown to prevent the appearance of penicillin-induced epileptiform spikes (19). In contrast, VPA was either ineffective or caused an increase in both burst frequency and amplitude when epileptiform activity was induced by PTZ in the CA3 region of the in vitro hippocampus, a finding indicating that these chemically induced hippocampal epileptiform activities may be differentially sensitive to antiepileptic drugs (20). Epileptiform bursting induced by bicuculline in rat amygdala slices was decreased by VPA (21). When epileptiform discharges were induced by combined application of bicuculline and 4-aminopyridine in combined entorhinal cortex-hippocampal slices from rats, these discharges were resistant to VPA and other standard anticonvulsants (22), whereas epileptiform discharges induced by 4-aminopyridine alone were potently suppressed by VPA (23). In studies on the age-dependency of VPA's anticonvulsant effect on 4-aminopyridine-induced epileptiform discharges in hippocampal slices, VPA blocked the ictal discharges in slices from both young and adult rats,
whereas interictal epileptiform activity was blocked by VPA only in slices from young rats (24). In young rat hippocampus, extracellular magnesium was shown to influence the effects of VPA on 4-aminopyridine-induced epileptiform events (25).
When epileptiform discharges were induced in the combined entorhinal cortex-hippocampal slice by removing magnesium ions from the perfusion fluid, early clonic-tonic discharges in the entorhinal cortex and the interictal-like activity in area CA3 were effectively suppressed by VPA, whereas the late recurrent tonic discharge state in the entorhinal cortex was unaffected by the drug (26). This late epileptiform activity was, however, still sensitive to an N-methyl-D-aspartate (NMDA) receptor antagonist. Subsequent experiments showed that the late recurrent discharges produced in the entorhinal cortex by prolonged exposure to low magnesium are resistant to all major anticonvulsants, a finding suggesting that they may represent a model of difficult-to-treat status epilepticus (27). In addition to induction of epileptiform events by reducing extracellular magnesium, such events can be produced in the entorhinal cortex and hippocampus by reducing extracellular calcium (Ca2+) or increasing extracellular potassium, but the patterns of epileptiform activity differ between these extracellular ion manipulations (28). VPA and its major active metabolite, trans-2-en-VPA, were shown to block all these forms of epileptiform activity, except the late recurrent discharges in the entorhinal cortex (28). The metabolite appeared to be more effective than VPA in these experiments. Analogous to entorhinal cortex-hippocampal slices, in rodent thalamocortical slices, different types of spontaneous epileptiform activity can be elicited by a medium containing no added magnesium. VPA was found to be effective in this in vitromodel of primary generalized epilepsy (29).
Like phenytoin and carbamazepine, therapeutic concentrations of VPA were shown to limit the ability of cultured mouse CNS (cortical and spinal cord) neurons to fire sodium (Na+)-dependent action potentials at high frequency (30). Such high-frequency firing has, for instance, been detected along subcortical pathways from a penicillin-induced cortical epileptogenic focus (31). Limitation of such firing may be important in preventing the spread of seizures. More recently, the effects of VPA on high-frequency sustained repetitive firing (SRF) in mouse central neurons in cell culture was compared with those of its major active metabolite trans-2-en-VPA (13). Both compounds limited firing in a concentration-, voltage-, rate-, and time-dependent fashion. The concentration dependence of limitation by both drugs markedly shifted to the left with duration of exposure; VPA was slightly more potent than trans-2-en-VPA after prolonged exposure (13). Although the precise biophysical mechanism underlying the ability of VPA (and its metabolite) to reduce SRF has not been elucidated, it was suggested that this effect probably relates to a phenytoinlike use- and voltage-dependent blockade of voltage-dependent Na+ channels (30). Detailed voltage clamp experiments of VPA actions on Na+ currents are described later in this chapter.
With respect to in vivo studies, VPA suppressed electrically induced afterdischarges in the hippocampus of cats (32) and significantly elevated the afterdischarge threshold and decreased the afterdischarge duration in the rat amygdala (33). VPA also raised the threshold for thalamic afterdischarges induced by electrical stimulation of the cat nucleus centralis lateralis and the rat nucleus reticularis without changing the duration of the afterdischarges in both species (34). With respect to focal seizures, VPA suppressed the epileptiform activity generated by a cobalt focus in the cat hippocampus and blocked the spreading of spontaneous as well as electrically induced seizure discharge from the hippocampus to the neocortex (32). Mutani and Fariello (35) noted that VPA suppressed the ictal and interictal seizure discharges in cats with an epileptogenic cobalt focus in the cruciate cortex. After administration of the drug, electrical stimulation of the cobalt focus failed to produce seizure activity. The same authors (36) observed that subcortical injection of aluminium gel into the sensorimotor cortex of cats produced focal cortical seizure discharges and myoclonic jerks of the head; this focal seizure activity could generalize, and VPA prevented this secondary generalization without influencing the epileptogenic focus. Moreover, van Duijn and Beckmann (37) noted that VPA did not decrease the focal discharge in the sensorimotor cortex of the awake cat produced by topical cobalt administration, but it effectively inhibited the spread of seizure activity from the focus. When two epileptogenic foci were formed in homotopic areas of the sensorimotor cortex of rats by application of penicillin, VPA blocked the focal discharges and secondary generalization of these discharges (38). In the amygdala-kindling model in rats, VPA was found to increase the threshold for electrical induction of afterdischarges and to reduce seizure severity, seizure duration, and afterdischarge duration recorded at the elevated threshold, findings indicating that VPA suppresses both the initiation and propagation of focal seizures in this model (39).
When cortical self-sustained afterdischarges were induced by rhythmic electrical stimulation of subcortical structures as a model of primary generalized seizures of the absence type, these epileptiform discharges were almost completely blocked by VPA (40). Similarly, VPA suppressed spontaneous spike-wave discharges in a genetic rat model of absencelike seizures (41). The major metabolite of VPA, trans-2-en-VPA, was more potent than VPA in blocking the spontaneous spike-wave discharges in these rats (42). Taken together, with few exceptions, VPA proved to be effective in suppressing epileptiform discharges in all in vitro and in vivo models tested in this respect, a finding in line with its broad spectrum of anticonvulsant activity against different seizure types in patients.
The well-documented effects of VPA on seizure discharge and the spread of neuronal excitability set the stage for studying the physiologic mechanisms underlying these effects. However, whether VPA acts through postsynaptic effects on neurotransmitter functions or ion channels or by presynaptic biochemical effects is still a matter of debate.
Effects on Excitability or Inhibition
Macdonald and Bergey (43) were the first to describe that VPA potentiates neuronal responses to GABA by a postsynaptic effect. However, VPA was examined after microiontophoretic application, so the local (extracellular) drug concentration was unknown. Subsequent in vitro studies showed that increased postsynaptic GABA responses are only obtained with very high VPA concentrations (6). To my knowledge, only one report demonstrated GABA potentiation at therapeutically relevant concentrations of VPA in vitro (44). The authors, using locus coeruleus neurons for their experiments, suggested that the difference between their data and those of other groups may result from the different brain regions examined in these studies. Indeed, based on neurophysiologic data, a regionally specific action of VPA in the brain was also suggested by Baldino and Geller (45).
In in vivo experiments, VPA was shown to lead to a potentiation of postsynaptic GABA responses at doses of 200 to 400 mg/kg (6). Because brain concentrations of VPA after these doses are much lower than the concentrations that potentiate GABA responses in vitro, the in vivo effect of VPA was likely not related to a direct postsynaptic action but more probably was the result of VPA's presynaptic effects, that is, enhanced GABA turnover (see later).
Experiments on central mouse neurons in culture indicated that neuronal responses to glycine or excitatory amino acids, such as glutamate, are not altered by VPA at relevant concentrations (6). However, one study showed that VPA suppresses glutamate responses and, much more potently, NMDA-evoked transient depolarizations in rat neocortex (46). The authors suggested that attenuation of NMDA receptor-mediated excitation is an essential mode of action for the anticonvulsant effect of VPA. This view is substantiated by numerous reports, using different preparations to study synaptic responses mediated by the NMDA subtype of glutamate receptors (6). In all studies, VPA blocked these responses, a finding indicating that antagonism of NMDA receptor-mediated neuronal excitation may be an important mechanism of VPA. VPA, but not phenobarbital, phenytoin, or ethosuximide, blocked seizures induced by N-methyl-DL-aspartate in rodents (47). In contrast to NMDA receptors, VPA had no effect on membrane responses mediated by kainate or quisqualate (α-amino-3-hydroxy-5-methyl-4-isoxazole propionate; AMPA) receptors (48).
The spontaneous firing of neurons is usually inhibited only by high doses or concentrations of VPA (49). However, in substantia nigra pars reticulata (SNR) of rats, a rapid and sustained reduction in the firing rate of GABAergic neurons was found in vivo after intraparenteral administration of doses as low as 50 to 100 mg/kg (50, 51, 52). This inhibitory effect on SNR neurons may result from the selective increase in GABA turnover induced by VPA in the nigra of rats (53) rather than from the direct effects of VPA on GABAergic neurons in SNR. Reduction of SNR firing as found with VPA has been shown effectively to suppress different types of seizures in diverse animal models of epilepsy, and this effect is explained by the important role of the SNR in seizure propagation (54,55). The inhibitory effect of VPA on SNR firing could therefore be crucially involved in its mechanisms of anticonvulsant action.
Effects on Ion Channels
At much lower concentrations than those depressing normal neuronal cell activity, VPA has been shown to diminish high-frequency repetitive firing of action potentials of central neurons in culture (30). It has been suggested that this effect may be critically involved in the anticonvulsant action of VPA on generalized tonic-clonic seizures (30). The effect of VPA on SRF was similar to limitation of SRF produced by phenytoin and carbamazepine (30). The most likely explanation for this effect of VPA would be a use-dependent reduction of inward Na+ current (30). However, in most studies carried out to date, effects on Na+ channels were inferred indirectly from changes in the maximal rate of increase of Na+-dependent action potentials. In an electrophysiologic study using cultured rat hippocampal neurons, VPA indeed strongly delayed the recovery from inactivation of Na+ channels, a finding that would be consistent with reduction of Na+ conductance (56). Studies using nonvertebrate preparations also indicated that VPA has a direct inhibitory effect on voltage-sensitive Na+ channels (6). However, the concept that VPA mediates its main anticonvulsant effect by slowing the recovery from inactivation of voltage-dependent Na+ channels has been questioned because, in contrast to cultured neurons, VPA has no effects on the refractory period and, consequently, the bursting behavior of neurons when rat hippocampal slice is used for studying the neurophysiologic effects of VPA (57). The latter authors concluded that, at least in the hippocampal slice, the drug's principal anticonvulsant effect cannot be explained by an action on voltage-dependent Na+ channels.
When phenytoin-sensitive Na+ channels in cultured neuroblastoma cells and rat brain synaptosomes were studied, VPA did not affect the Na+ influx (58). Furthermore, VPA had no effect on the phenytoin binding site on voltage-dependent Na+ channels (59). In rat neocortical neurons in culture, VPA (0.2 to 2 mmol/L) was reported to reduce voltage-dependent Na+ currents (60). More recent
whole-voltage clamp measurements of Na+ currents in acutely isolated CA1 neurons from rats and patients with pharmacoresistant temporal lobe epilepsy showed that VPA induced a shift of the voltage dependence of inactivation in a hyperpolarizing direction with an EC50 of 2.5 mmol/L (rats) or 1.6 mmol/L (patients), respectively (61,62). In view of these high concentrations, the fatty acid VPA may modulate the Na+ channel by influencing the biophysical properties of the membrane surrounding the channel, as has been proposed for free polyunsaturated fatty acids (63). Indeed, when using therapeutically relevant VPA concentrations, VPA (≤400 µmol/L) was found to be ineffective on the fast Na+ current in acutely dissociated neocortical rat neurons (64). In contrast, VPA potently reduced the persistent fraction of the Na+ current with an EC50 of 14 µmol/L (64). Whether this highly potent effect of VPA can explain its action on SRF remains to be determined. Apart from interference with Na+ channels, VPA's effect on SRF could be the result of activation of Ca2+-dependent potassium conductance (65).
An activating effect on potassium conductance has been repeatedly discussed as a potential mechanism for the action of VPA (65, 66, 67), although such an effect has been demonstrated only at high drug concentrations. Previous experiments using various potassium channel subtypes from vertebrate brain expressed in oocytes of Xenopus laevissubstantiated that the effects of VPA on potassium currents are too small to be significant in its mechanism of anticonvulsant action (68).
In regard to Ca2+ channels, the antiabsence drugs ethosuximide and dimethadione (the major active metabolite of trimethadione) have been shown to block use-dependent activation of T-type Ca2+ channels in thalamic neurons, which have been implicated in the generation of spike-wave activity associated with absence epilepsy (69). However, VPA did not affect this T current in thalamic neurons, although VPA is as effective as ethosuximide in blocking absence seizures. In contrast to thalamic neurons, VPA was shown to block low-threshold T Ca2+ channels in peripheral ganglion neurons (70). Veratrine-stimulated Ca2+ influx in brain slices was not altered by VPA, whereas the drug reduced NMDA- or quisqualate-stimulated Ca2+ influx at 5 mmol/L (67,71). In such high, millimolar concentrations of VPA, this lipophilic compound interferes with membrane functions by partition into cell membranes (72,73), and this may explain many of the neurochemical or neurophysiologic effects of the drug in studies involving this high concentration range.
As a consequence of early reports establishing that VPA leads to an elevation of cerebral GABA levels in rodents (74), and that the period of GABA elevation coincides with the protection against seizures (75,76), numerous subsequent studies dealt with the effects of VPA on the GABA system (6). However, unlike GABAmimetic drugs, which selectively affect the GABA system, VPA certainly acts through more than one mechanism in providing its broad anticonvulsant activity. Furthermore, despite the clear effects of VPA on GABA metabolism, the role of these effects in VPA's anticonvulsant action is a matter of ongoing controversy.
Effects on the GABA System
It seems generally accepted that impairment of GABAergic inhibitory neurotransmission can lead to convulsions, whereas potentiation of GABAergic transmission results in anticonvulsant effects (77). Several clinically used antiepileptic drugs are GABAmimetic drugs, that is, they act by potentiating GABAergic neurotransmission either by increasing GABA concentrations through inhibition of GABA degradation (vigabatrin) or GABA uptake (tiagabine) or by directly acting at the postsynaptic GABAA receptor complex (e.g., benzodiazepines, barbiturates). It is thus not astonishing that the initial reports on brain GABA increase by VPA (74,78) led to the assumption that enhancement of GABAergic neurotransmission is the mechanism of anticonvulsant action of VPA. Since it was first postulated in 1968, this GABA hypothesis has been the matter of ongoing dispute in the literature. For instance, it has been claimed that increases of GABA levels in the brain of rodents are seen only after high doses of VPA, whereas lower doses, which still exert anticonvulsant effects, do not change GABA levels (65,66). Furthermore, the finding that in some seizure models the rise in brain GABA by VPA lags behind the earlier appearance of the anticonvulsant effect led to questions about the relevance of the GABA increases by VPA (65). However, these apparent discrepancies result because most studies on VPA's effects on brain GABA levels used GABA determination in whole brain or whole tissue of few brain regions, thus ignoring the marked differences in GABA metabolism between brain regions and the cellular compartmentation of GABA within a brain region (6). Regional brain studies in rats showed marked differences in VPA's effects on GABA levels across brain areas, with significant GABA increases in midbrain regions, such as substantia nigra (SN), which are thought to be critically involved in seizure generation and propagation (79,80). In the SNR, the GABA increases induced by VPA occurred predominantly in nerve terminals, that is, in the “neurotransmitter pool” of GABA (80, 81, 82). The onset of VPA's effects on presynaptic (synaptosomal) GABA levels in brain regions was very rapid (significant increases were already observed after 5 minutes), and the time course of anticonvulsant activity correlated with that of the nerve terminal alterations in GABA levels (80). In in vivo experiments in dogs, in which VPA was infused continuously to obtain
plasma levels in the range known from long-term oral treatment in humans, GABA increases were determined in cerebral cortex and cerebrospinal fluid (CSF) at an infusion rate of 25 mg/kg/hr (83). Accordingly, significant increases in CSF GABA levels were also found during treatment with VPA in patients with epilepsy or schizophrenia (6). Furthermore, significant GABA increases were determined in plasma of dogs and humans under VPA treatment (6). In dogs, the plasma GABA increases paralleled those in CSF and brain tissue and thus indicated that plasma GABA determination may be an indirect indicator of alteration in CNS GABA levels in response to VPA (83).
Although there is substantial evidence that VPA increases GABA concentrations at clinically relevant doses, the mechanism and functional meaning of the increase in brain GABA levels are still matters of debate. The increase of presynaptic GABA levels induced by VPA could be explained by three different mechanisms: (a) an inhibitory effect of VPA on GABA degradation; (b) an enhancement of GABA synthesis; or (c) no direct effects of VPA on synthesis or degradation of GABA, but an indirect effect on presynaptic GABA levels by direct potentiation of postsynaptic GABAergic function leading to feedback inhibition of GABA turnover and thereby to increases in nerve terminal GABA.
Shortly after the initial reports on the GABA-elevating effect of VPA, several groups examined the action of this drug on GABA degradation. GABA is synthesized in GABAergic nerve terminals by decarboxylation of glutamate and is degraded in nerve terminals, glia cells, and postsynaptic neurons (after diffusion) by transamination to succinic semialdehyde (SSA). SSA can either be oxidized by SSA dehydrogenase to succinate or it can be reduced by SSA reductase to γ-hydroxybutyrate (GHB). The relative importance of these two degradative pathways in vivo is unclear, although it appears that the reduction to GHB is generally a minor route of metabolism. The GABA-elevating effect of VPA was originally attributed to inhibition of GABA transaminase (GABA-T), which catalyses the degradation of GABA to SSA (74). Yet most studies on in vitro inhibition of GABA-T by VPA found inhibitory effects only at very high (millimolar) concentrations that are not reached in vivo (6). Indeed, when VPA was administered to rodents and GABA-T was determined in brain homogenates ex vivo, no inhibition of the enzyme was found (74,84, 85, 86). However, a significant reduction of GABA-T activity was found in synaptosomes prepared from brain tissue after VPA administration in mice (84,85). Similarly, in rats, VPA treatment induced a significant inhibition of synaptosomal GABA-T activities in several brain regions, including SN, hippocampus, hypothalamus, pons, and cerebellum (87). These data may be explained by assuming that the presynaptic (nerve terminal) GABA-T is different from glial GABA-T (which predominates in whole tissue homogenates) in terms of susceptibility to VPA. Alternatively, the significant reduction in synaptosomal GABA-T observed in vivo is not the result of a direct inhibition of GABA-T by VPA but is a secondary effect caused by alterations in the subsequent steps of GABA metabolism. The first assumption was substantiated by experiments showing that VPA is much more potent to inhibit GABA-T in neurons (mean inhibitory concentration, 630 µmol/L) than in astrocytes or whole-tissue homogenates (88). The second assumption has been extensively discussed, and it has been concluded that it is not possible to raise brain GABA levels by inhibition of SSADH (6). Thus, the GABA-T reduction observed in synaptosomes but not in whole-tissue homogenates of rodents after treatment with VPA is most certainly the result of a higher susceptibility of nerve terminal GABA-T compared with the enzyme outside nerve terminals. Inhibition of nerve terminal GABA-T could explain the increase of presynaptic GABA levels by VPA, although the reduction of synaptosomal GABA-T activity was not marked (6).
Besides effects of VPA on GABA degradation, activation of GABA synthesis could be another likely explanation for the GABA-elevating action of this drug (6). Godin et al. (74) measured the relative incorporation of carbon-14 (14C) into GABA in rat brain after the subcutaneous injection of [14C]glucose. Thirty minutes after administration of VPA 400 mg/kg i.p., the incorporation of 14C into the GABA molecule was increased by 30%, but this was not significant on account of the small number of animals studied. In similar experiments in mice, Taberner et al. (89) found that VPA, 80 mg/kg i.p., produced a significant increase in the rate of production of GABA by 90%. Studies on GABA turnover in various rat brain regions demonstrated that the most marked increase in GABA synthesis by VPA is found in the SN (53); the reason for this finding could be that the SN is one of the regions with the highest rates of GABA synthesis. Indeed, the increase in GABA synthesis by VPA most likely relates to an activation of the GABA-synthesizing enzyme glutamic acid decarboxylase (GAD). Increase of GAD activity by VPA has been demonstrated ex vivo after administration in mice and rats in several independent studies (6). In rats, GAD was not activated in all regions, a finding indicating a regional specificity of VPA's effects (86). Increase of GAD by VPA is rapid in onset, and the time course of GAD activation matches that of the GABA increase and the anticonvulsant effect (90). The rapid activation of GAD by VPA may indicate that VPA converts the inactive apoenzyme to the active holoenzyme (86). However, at high, toxic doses, VPA has been shown to inhibit GAD activity and to reduce GABA synthesis (91).
Activation of GAD by VPA has also been reported in vitro (6). GAD from neonatal rats was more sensitive to activation by VPA than GAD from adult animals (92). In neonatal rat brain slices, VPA significantly increased the activity of the GABA shunt, which was related to an increase of GAD activity (93). However, high (toxic) concentrations of VPA (7 mmol/L) significantly decrease GAD (94).
Neurochemical experiments in beef brain preparations have shown that the coenzyme A ester of VPA, which is rapidly formed from VPA in vivo, is a potent inhibitor of α-ketoglutarate dehydrogenase complex (95). Because decreased activity of this complex would reduce substrate flux through the citric acid cycle and may increase flux into GABA synthesis, this finding adds to the accumulating evidence that VPA increases GABA levels predominantly by enhancing the synthesis of this amino acid (95).
An increase of presynaptic GABA levels by VPA would potentiate GABAergic neurotransmission only if the release of GABA into the synaptic cleft would be also increased. The first direct evidence of enhanced GABA release by VPA came from studies on cortical slices prepared from VPA-treated animals and from studies in neuronal culture (96,97). Thus, in the cortical slice from VPA-treated rats, the potassium-induced release of GABA was increased, and this was potentiated further by the GABAB receptor antagonist phaclofen (97). Similarly, VPA increased the potassium-induced release of GABA from cortical neurons in culture in clinically relevant concentrations (96). Similar to the biphasic effects of VPA on GABA synthesis (i.e., increase at low doses but decrease at high doses), high concentrations of VPA seem to inhibit GABA release. The uptake of GABA from the synaptic cleft is not affected by VPA (6). However, in a study on GABA transporter proteins in rats with spontaneous recurrent seizures induced by amygdala injection of ferric chloride, VPA caused downregulation of GABA transporters (98).
Indirect evidence of enhanced GABA release by VPA comes from in vivo studies in rats in which microdialysis was used to measure extracellular GABA levels in the hippocampus (99,100). Biggs et al. (99) reported biphasic effects of VPA on extracellular GABA levels, which depended on the dose used. At 100 mg/kg, VPA transiently reduced GABA concentrations by 50% when compared with basal levels, and 200 mg/kg VPA had virtually no effect, whereas 400 mg/kg VPA raised extracellular GABA levels to 200% of basal levels. Such biphasic effects of VPA on extracellular GABA levels were also found by Wolf et al. (101), using local application of VPA into the rat preoptic area through push-pull cannulae. Similar to the study of Biggs et al. (99), Rowley et al. (100) reported that VPA, 400 mg/kg, significantly increased extracellular levels of GABA measured by microdialysis in the hippocampus of freely moving rats. Furthermore, VPA prevented decreases in GABA in response to maximal electroshock seizures in these animals (100). Using the push-pull technique to measure extracellular GABA in the SN of rats, Farrant and Webster (51) found no effect of VPA, 200 mg/kg, on the spontaneous release of GABA into the perfusate. However, as pointed out by Timmermann and Westerink (102), none of these techniques of measuring extracellular GABA allow investigators to draw direct conclusions on drug effects on GABA release, because of the marked compartmentalization of this neurotransmitter.
An enhancement of GABA release at clinically relevant concentrations of VPA is indirectly indicated by the increase in CSF GABA levels observed in different species, including humans (6). In view of the different reports demonstrating an increase of GABA turnover and release by VPA (6), the previous hypothesis that the increased brain GABA concentrations induced by VPA treatment are only secondary as a result of feedback inhibition of GABA turnover because of direct postsynaptic effects of the drug has to be rejected.
In contrast to VPA's effects on GABA synthesis and degradation, the drug does not exert direct effects on the major components of the postsynaptic GABAA receptor complex. Thus, in vitro, VPA did not alter GABA binding, benzodiazepine binding, or the binding of the selective chloride ionophore ligand [35S]t-butylbicyclophosphorothionate (TBPS) (6). Therefore, the previous assumption that VPA could potentiate GABAA receptor function by a barbituratelike effect on the picrotoxinin site, which was based on low-potency inhibition by VPA of [3H]α-dihydropicrotoxinin binding (103), could not be substantiated by subsequent experiments with the more suitable ligand [35S]TBPS. However, in vivo VPA has been shown to reduce TPBS binding and to increase benzodiazepine binding, which is most likely secondary, as a result of the increase in GABA levels produced by VPA in vivo (104). The functional meaning of the alteration in benzodiazepine receptor binding is not clear, because benzodiazepine receptor antagonists do not reduce the anticonvulsant potency of VPA (105). Conversely, prolonged pretreatment of mice with benzodiazepines reduced the anticonvulsant potency of VPA and thus demonstrated the development of cross-tolerance between benzodiazepines and VPA (106). Furthermore, the anticonflict action of VPA was reversed by benzodiazepine antagonists (107), a finding indicating that enhanced binding of benzodiazepines to the GABAA receptor complex may be involved in this pharmacodynamic effect of VPA.
Two groups independently reported increase of GABAB receptor binding by long-term treatment of rats with VPA (108,109). In a study by Motohashi (109), short-term treatment with VPA had no effect on [3H]baclofen binding in frontal cortex, hippocampus, and thalamus, whereas longterm treatment enhanced binding in the hippocampus. [3H]muscimol binding to the GABAA receptor did not change after VPA administration in any region (109). Because similar effects were observed with lithium and carbamazepine, Motohashi (109) concluded that one common mechanism of action of mood stabilizers may be mediated by GABAB receptors in the hippocampus.
Taken together, the numerous neurochemical reports of VPA's effects on the GABA system indicate that increases in GABA function may be involved in several pharmacodynamic effects of this drug, including the anticonvulsant, anticonflict, and antimanic actions (6). Furthermore, in
view of the role of GABA in antinociception (110), increased GABAergic function by VPA may be involved in its antinociceptive effects. However, the effects of VPA on GABA alone are not sufficient to explain its broad anticonvulsant activity, and several of VPA's effects on neuronal tissue, such as on acutely dissociated neurons, have been demonstrated not to be related to GABA potentiation (29).
Effects on γ-Hydroxybutyrate, Glutamate, and Aspartate
Compared with the numerous studies on VPA's effects on the GABA system, only relatively few neurochemical studies were done on other transmitter amino acids (6). Several of these studies dealt with VPA's effects on GHB metabolism. VPA was shown to be a potent inhibitor of NADPH (reduced form of nicotinamide-adenine dinucleotide phosphate)-dependent aldehyde reductase (111). Aldehyde reductase is presumably identical to nonspecific SSA reductase (SSAR) (112). Whereas the specific SSAR is thought to reduce SSA to GHB, the nonspecific SSAR is thought to be responsible for the catabolism of GHB to SSA (112). In contrast to the potent effect of VPA on nonspecific SSAR, specific SSAR is not affected by the drug (112). However, Whittle and Turner (113), using rat brain homogenates, demonstrated that VPA inhibited the formation of GHB in vitro, a finding indicating that the specific SSAR is not exclusively responsible for GHB formation but the nonspecific (VPA sensitive) aldehyde reductase may also contribute to a significant extent to this metabolic pathway. Inhibition of GHB formation by VPA could be of considerable interest because this amino acid has been shown to produce epileptogenic (absencelike) effects in several species (114). Administration of VPA to rats has been shown to increase (rather than decrease) the brain level of GHB in vivo (115). This increase in GHB levels by VPA is time and dose dependent and appears to result from a reduction in synaptic release of GHB (112). Because GHB produces absencelike epileptic seizures in animals (114), reduction of GHB release could be an important factor in the antiabsence action of VPA (112).
Glutamate concentrations in regional brain homogenates or in extracellular fluid obtained by microdialysis from hippocampus or SN were not significantly altered by systemic administration of VPA (51,99,100). However, Dixon and Hokin (116) reported that VPA stimulates glutamate release in mouse cerebral cortex slices at therapeutic concentrations. This effect may be involved in the antimanic action of VPA (116). Nilsson et al. (117) reported that VPA inhibits the transport of glutamate and aspartate in astroglial cells in primary cultures from newborn rat cerebral cortex. In adult rats, VPA treatment decreased the expression of glutamate transporter-1 in the hippocampus (98).
With respect to aspartate, VPA was shown to reduce the concentration and release of this excitatory amino acid in rat and mouse brain (6). Furthermore, some reports found that concentrations of glycine and taurine increased in brain tissue (6). However, there is as yet no evidence that these effects on amino acids other than GABA are relevant to the anticonvulsant effect of VPA.
Effects on Serotonin and Dopamine
VPA induces in rats a behavioral syndrome with “wet dog shakes” and other symptoms reminiscent of the “serotonin syndrome” induced by serotonin precursors or receptor agonists in rodents. Indeed, microdialysis studies in rats have demonstrated that VPA enhances the extracellular concentration of serotonin (5-hydroxytryptamine or 5-HT) in hippocampus and striatum of rats (118). However, in contrast to the increase in anticonvulsant efficacy during prolonged treatment, the wet-dog-shake behavior induced by VPA is markedly diminished within some days of treatment, thus indicating that activation of serotonergic transmission is not related to the anticonvulsant action of the drug (9). Accordingly, Horton et al. (119) showed that pretreatment of mice with p-chloro-phenylalanine, which blocked serotonin synthesis and prevented the increase in serotonin metabolism by VPA, did not diminish the anticonvulsant action of VPA.
As with serotonin, microdialysis studies have demonstrated an increased extracellular level of dopamine in response to VPA (118,120). Thus, the initial assumption (119) that VPA does not exert effects on serotonin or dopamine levels but only blocks outward transport of their metabolites from the CNS has to be rejected. As with serotonin, the alterations in dopamine levels seem not associated with VPA's anticonvulsant effect, because pretreatment of mice with α-methyl-p-tyrosine to inhibit dopamine synthesis did not diminish the anticonvulsant action of VPA (119). However, alterations of dopaminergic functions by VPA may be important for the antipsychotic effects of the drug (6). Ichikawa and Meltzer (120) showed that the increase in prefrontal dopamine release by VPA can be blocked by a selective 5-HT1A receptor antagonist, a finding indicating that VPA's effect on dopamine release is mediated by this 5-HT receptor subtype.
Other Biochemical Effects
Guanosine 3′,5′-monophosphate (cGMP) has been implicated as a second messenger in a variety of cellular events (121). For instance, the levels of cGMP in the cerebellum and cortex are known to increase sharply at the onset of experimentally induced seizures, and it has been proposed that an elevated cerebellar cGMP level is involved in initiating or maintaining seizure activity through the regulation of Purkinje cell activity (122,123). VPA was shown to decrease the cerebellar cGMP level during the time of anticonvulsant activity, whereas the cortical cGMP level was elevated
(122,123). In contrast to cGMP, cyclic adenosine monophosphate levels were not altered by VPA. Because levels of cGMP in the CNS are altered by several neurotransmitters, including amino acids (121), the effects of VPA on cGMP may be secondary, resulting from the various alterations of neurotransmitter systems described earlier.
VPA was reported to downregulate the myristoylated alanine-rich C kinase substrate (MARCKS) in an immortalized hippocampal cell line, which is thought to be a property of protein kinase C-mediated mood stabilizers (124). MARCKS is a prominent and preferential substrate in the brain for protein kinase C and has been implicated in cellular processes associated with neuroplasticity (e.g., neurotransmitter release and transmembrane signaling) and cytoskeletal restructuring (125). Long-term exposure of hippocampal cells with VPA (for ≥3 days) at therapeutic concentrations produced a decrease in MARCKS protein levels, whereas short-term exposure produced no significant change (124). This retarded onset of action is in line with the delay of several days in the onset of VPA's antimanic effects seen in patients. MARCKS was also downregulated by lithium, whereas other psychotropic drugs, including carbamazepine, haloperidol, diazepam, and fluoxetine, did not affect levels of the MARCKS protein (124). These data thus indicate that the property of regulating MARCKS is shared by the mood stabilizers lithium and VPA, which may be specific to a class of drugs effective in the treatment of bipolar disorders (124).
VPA, lithium, and carbamazepine were shown to inhibit at therapeutically relevant concentrations the high-affinity myoinositol transport in astrocytelike cells in vitro (126). When rats were treated on a long-term basis with VPA or lithium, the brain concentrations of myoinositol and inositol monophosphates significantly increased (127). It was suggested that both lithium and VPA may share a common mechanism of action in the treatment of bipolar disorder through actions on the phosphoinositol cycle.
PUTATIVE MECHANISMS INVOLVED IN THE EARLY AND LATE ANTICONVULSANT EFFECTS OF VALPROIC ACID
As described earlier in this chapter, depending on the seizure model or seizure type, VPA's anticonvulsant effect may either occur immediately (early) or late, that is, with some lag time after acute administration or only during long-term administration, a finding suggesting that these early and late effects of VPA are mediated by different cellular mechanisms. As outlined in this chapter, VPA has both extracellular (e.g., ion channels) and intracellular (e.g., GABA synthesis) sites of action. The access to these sites will determine how rapidly VPA acts after systemic action. Frey and Löscher were the first to describe that VPA enters and leaves the brain by active, carrier-mediated transport at the blood-brain barrier (128). Whereas it was initially thought that this probenecid-sensitive transport is mediated by the monocarboxylic acid carrier (128), more recent experiments have revealed that the bidirectional movement of VPA across the blood-brain barrier is mediated jointly by passive diffusion and carrier transport, with different transporters responsible for each direction of transport (129). The uptake of VPA from blood into brain is facilitated by a medium- and long-chain fatty acid selective anion exchanger at the brain capillary epithelium, which accounts for two-thirds of the barrier permeability, whereas the mechanism governing the efficient transport of VPA in the reverse direction, that is, from brain to blood, appears to involve a probenecid-sensitive, active transport system (129). Huai-Yun et al. (130) showed that VPA is a potent inhibitor of the adenosine triphosphate-dependent probenecid-sensitive transporters of the multidrug resistance-associated protein (MRP) family in the blood-brain barrier, which raises the possibility that MRPs may serve as the efflux transporters of VPA. This active transport at the blood-brain barrier explains why VPA, despite its physicochemical properties (highly ionized at physiologic pH, highly plasma protein bound), enters the brain so quickly (91,129). Assuming that VPA has to enter neurons by passive diffusion, its rapid anticonvulsant effect in some seizure models after parenteral administration of a single dose is most likely explained by an effect on extracellular sites. The late anticonvulsant effects observed both preclinically and clinically are more likely explained by slow access to intracellular sites of action. This view is corroborated by neurophysiologic experiments. Thus, in the buccal ganglia Helix pomatia preparation, extracellular application of VPA decreased frequency of occurrence of PTZ-induced epileptic depolarizations immediately (early effect) and, with a delay, led to a decay in paroxysmal depolarizations (late effect) (12). This late effect was obtained immediately when VPA was applied intracellularly (E.-J. Speckmann, personal communication), a finding substantiating that the delay in this effect after extracellular application was the result of slow penetration of VPA into the neuron. Slow diffusion into and out of neurons could also be involved in carryover effects observed both preclinically and clinically, because whereas extracellular levels of VPA will rapidly leave brain or CSF by active outward transport, elimination from neurons may be retarded. More recent microdialysis experiments by Shen's group in rabbits suggested that VPA does not solely enter neurons by passive diffusion, but that another set of transporters at the neural cell membrane is involved (129). The putative parenchymal cell transport system is able to concentrate VPA within the cellular compartment, and this finding has important implications in our understanding of the intracellular mechanisms versus membrane actions of VPA (129). The efflux component from the cellular compartment is inhibited by probenecid (129), and this could indicate that MRPs are involved.
No single action of VPA can completely account for the drug's numerous effects on neuronal tissue and its broad clinical activity in epilepsy and other brain diseases. In view of the diverse molecular and cellular events that underlie different seizure types, the combination of several neurochemical and neurophysiologic mechanisms in a single drug molecule may explain the broad antiepileptic efficacy of VPA. Furthermore, by acting on diverse regional targets thought to be involved in the generation and propagation of seizures, VPA may antagonize epileptic activity at several steps of its organization. The finding that VPA exerts not only anticonvulsant but also several other pharmacodynamic and pharmacotherapeutic effects, including antimanic and migraine-prophylactic efficacy, certainly relates to the multiplicity of VPA's effects on neuronal functions (6,110,117,131,132). Because of the different pharmacodynamic effects of VPA, it is difficult to ascertain which specific neurochemical or neurophysiologic actions are related to the anticonvulsant activity of the drug. There is now ample evidence that VPA increases GABA turnover and thereby potentiates GABAergic functions in some specific brain regions thought to be involved in the control of seizure generation and propagation. Furthermore, the effect of VPA on neuronal excitation mediated by the NMDA subtype of glutamate receptors may be important for its anticonvulsant effects. In contrast, the relevance of the often-cited effect of VPA on SRF in cultured neurons remains debatable, because there is no convincing evidence that VPA—at therapeutically relevant concentrations—blocks voltage-dependent Na+ currents and SRF in more conventional preparations (57). Whereas the GABA potentiation and glutamate-NMDA inhibition could be a likely explanation for the anticonvulsant action on focal and generalized motor seizures, they do not explain the effect of VPA on nonconvulsive seizures, such as absences. In this respect, the reduction of GHB release reported for VPA could be of interest.
Despite considerable discussion about the possible mechanisms of action of VPA, no definite answer has gained general acceptance so far, and much remains to be learned at numerous different levels of VPA's actions. In view of the advances in molecular neurobiology and neuroscience, future studies will undoubtedly further our understanding of the mechanisms of action of this major antiepileptic drug.