Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

GHB is a naturally occurring substance that is located in almost all brain regions (Vayer et al., 1988), together with succinic semialdehyde reductase, the enzyme responsible for its synthesis. However, it is thought to play a direct functional role only in some restricted brain areas, a view supported by the heterogeneous distribution of its receptor sites. These are located largely in the cortex, hippocampus and thalamus, together with dopaminergic brain structures including the dorsal and ventral striatum, olfactory tracts, A₉, A₁₀ and A₁₂ (Hechler et al., 1992). The major part of the hypothalamus, pons-medulla and cerebellum are totally devoid of high-affinity binding sites for GHB, as are peripheral tissues such as liver, muscles and kidneys. Specific high-affinity GHB binding sites have also been found in cell membranes prepared from human brain (Snead and Liu, 1984). This binding does not require Na⁺ and is not displaceable by GABA, muscimol, baclofen, isoguvacine, dopamine or picrotoxin, but only by GHB and structurally related analogs (Benavides et al., 1982).

Electrophysiological studies have shown an effect of GHB on about 50% of the cells examined in the nigro-striatal pathway (Harris et al., 1989), in the neocortical region (Olpe and Koella, 1979) and in the parietal cortex (Kozhechkin, 1980). When used at low doses in vivo (5-10 mg/kg), GHB induces a depolarizing effect that is blocked by the GHB receptor antagonist NCS-382 (Godbout et al., 1995). However, when used at higher doses both in vivo and in vitro (in general 100 µM in vitro and 300 mg/kg in vivo), GHB induces a membrane hyperpolarization that is bicuculline-resistant (Olpe and Koella, 1979) but that has been reported to be sometimes inhibited by GABA_B antagonists (CGP 35 348 or CGP 55 845) (Xie and Smart, 1992; Williams et al., 1995; Ito et al., 1995). The number of GHB-responsive neurons appears to be much lower than the number of GABA-responsive neurons in the brain regions investigated. The neuronal hyperpolarization induced by GHB in vivo or after incubation of brain tissue slices with GHB probably explains the decrease in dopaminergic neuronal activity resulting in a decreased dopamine release in the nigro-striatal pathway after administration of GHB (Walters et al., 1973). Baclofen has similar effects on dopaminergic neurons (Da Prada and Keller, 1976).

Thus GHB induces specific physiological responses that are dependent on its interaction with GHB receptors that are distinct from GABA_B receptors in kinetics, pharmacology, distribution and ontogeny (Benavides et al., 1982; Hechler et al., 1992; Snead, 1994). However, a possible GABAergic contribution to the pharmacological effects of GHB must be considered. This contribution can be explained by a direct interaction of GHB with GABA_B sites, because GHB displaced GABA_B binding with an IC₅₀ value of 100-200 µM (Bernasconi et al., 1992), 500 µM (Ito et al., 1995) or 796 µM (Ishige et al., 1996). These values largely exceed endogenous GHB levels in brain, which peaked at maxima of 5 to 6 µM (Vayer et al., 1988).

Several authors have suggested that labeled GABA is formed in vivo after the administration of labeled GHB with no increase in GABA concentration (see, for exemple, DeFeudis and Collier, 1970), although one group has suggested that brain GABA levels are increased (Della Pietra et al., 1966). In our hands, [³H]-GHB is consistently transformed into [³H]-GABA by brain extract (Vayer et al., 1985). This conversion is due to the coupled effect of GHB dehydrogenase and NADP to yield succinic semialdehyde (SSA); then GABA-T activity transaminates SSA into GABA. GHB dehydrogenase is a cytosolic enzyme that is inhibited by a wide range of antiepileptic compounds, including barbiturates, valproate, ethosuximide and trimethadione (Kaufman and Nelson, 1991). Most of these compounds, when administered to rats, induce an accumulation of GHB in the brain (Snead et al., 1980).

The purpose of this study was to demonstrate that, under the conditions used for in vitro GABA_B binding experiments, under in vivo conditions and in experiments carried out with brain slices or cell cultures, GHB is partially degraded by brain extract into GABA, which then displaces GABA_B binding. In our experiments, GHB degradation into GABA was prevented by GHB dehydrogenase inhibition with either valproate or ethosuximide or by GABA-T inhibition with aminooxyacetic acid.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Male Wistar rats weighing 250 to 300 g were killed by a blow on the head; their brains were rapidly extracted and used as starting material. Procedures involving animals and their care were conducted in conformity with national and international regulations (decree n° 87848, October 19, 1987, and EEC council directive 86/609, QJ L 358, December 12, 1987).

GABA_B binding to rat brain membranes. The methods of Hill and Bowery (1981, method 1) and of Bernasconi et al. (1992, method 2) were used to assess the ability of GHB to displace GABA_B binding. Method 1 was used in general, but method 2 was adopted in some experiments because an IC₅₀ value of 150 µM was measured for GHB under these conditions. Crude synaptic membranes (P₂ fraction) were prepared from total brain or from cerebrum or cerebellum. In method 2, the vesicular preparation was further purified by centrifugation on 0.8 M buffered sucrose. After hypoosmotic shock, the membranes were centrifuged and frozen at 20°C overnight (method 1) or for 2 days (method 2). After several incubations and washings at ambient temperature, the pellets were used for GABA_B binding determinations. Incubations were carried out in 600 µl of buffer (50 mM Tris-HCl, 2.5 mM CaCl₂, pH 7.4) at ambient temperature with 25 nM [³H]-GABA (Dupont-NEN, France, 74 Ci/mmol). Isoguvacine (100 µM, final concentration) and GHB (concentrations from 10 µM to 5 mM) were added. In some experiments, media were supplemented with valproate or ethosuximide at a final concentration of 1.5 mM. Nonspecific binding was determined in the presence of 100 µM baclofen.

GABA_A binding in the presence of GHB. The effect of GHB on GABA_A binding was tested using [³H]-muscimol (19 Ci/mmol, Dupont-NEN). Membranes were prepared from a crude synaptosomal/mitochondrial fraction of rat brain according to the method of Olsen et al. (1981). GABA_A receptor binding was measured by a rapid filtration assay at 0-4°C in Na⁺-free buffer. [³H]-muscimol was included at 25 nM (final concentration) with or without 0.1 mM nonradioactive GABA. Samples containing 1 mg of protein in an assay volume of 600 µl were incubated 15 min at 0-4°C with increasing concentrations of GHB (10 µM to 10 mM). The incubation media were rapidly filtered at 4°C under suction and then were rinsed twice with 2 ml incubation buffer (50 mM Tris-citrate, pH 7.1, at 0°C). Radioactive filters were counted by liquid scintillation.

Effects of antiabsence drugs on the conversion of [³H]-GHB to [³H]-GABA by rat brain membranes. Crude synaptic membranes were prepared according to Hill and Bowery (1981). These membranes were incubated at ambient temperature in 50 mM potassium phosphate buffer, pH 7.4, containing 200 µM [³H]-GHB (10 µCi/mmol) and 1.5 mM of either ethosuximide or valproate. The kinetics of the [³H]-GABA formed was monitored after separation from [³H]-GHB on a Dowex 50W-X8 column (0.5 × 3 cm, H⁺ form). Controls were carried out in the absence of antiepileptic drugs. Radioactive GABA eluted from the columns by 0.1 N NaOH was counted by means of a liquid scintillation counter (Vayer et al., 1985).

Measurement of [³H]-aminoacids formed from [³H]-GHB in the presence of rat brain crude synaptosomal membranes. Crude synaptosomal membranes were prepared from a whole rat brain according to the method of Hill and Bowery (1981). These membranes were incubated 20 min at ambient temperature with 1 ml of 50 mM Tris-HCl, pH 7.4, containing CaCl₂ (2.5 mM) and 200 µM [³H]-GHB (100 µCi/200 nmol). Perchloric acid (0.1 M, final concentration) was added to precipitate the proteins, which were removed by centrifugation. The amino acid content of the supernatant was determined by separation of the amino acids' o-phthalaldehyde derivatives by high-performance chromatography/fluorimetric detection, using a modification of the method of Allison et al. (1984). Briefly, all chromatographic separations were performed with a Nucleosil C 18 column (5 µm, 25 × 0.4 cm) with two Waters pumps 590 and a Waters Baseline 810 integrator. Detection was carried out with a Waters fluorimeter 470 (excitation: 345 nm, emission: 455 nm). The mobile phase was a binary gradient of solution A (0.1 M NaH₂PO₄, pH 6.0, containing 2% methanol, pH 6.0) and of solution B (40% 0.1 M NaH₂PO₄, pH 6.0, 30% methanol and 30% acetonitrile). Precolumn autoderivatization (2 min) and injection were achieved with a CMA 200 refrigerated Microsampler (Carnegie Medicine, Sweden) by adding to 20 µl of tissue extract 20 µl of the following derivatization mixture: 5 ml of 0.1 M sodium tetraborate, pH 9.5, containing 10 µl of 3-mercaptopropionic acid (Sigma, Aldrich Chimie, France) and 15 mg of o-phthalaldehyde (Sigma) in 500 µl of methanol. Elution was carried out at a rate of 0.8 ml/min and at a temperature of 35°C with the following steps: 0 min, 90% A/10% B; 15 min, 40% A/60% B (linear gradient); 16 min, 40% A/60% B (isocratic); 19 min, 100% B (isocratic); 24 min, 90% A/10% B (isocratic) until 29 min.

Statistical analysis. Nonlinear regression fitting and IC₅₀ calculations were performed using the Graphpad-Prism program. Comparison between regression curves was analyzed using the two-way ANOVA statistical test.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of GHB on GABA_B binding in the presence and absence of GHB dehydrogenase inhibitors. In a first set of experiments, GABA_B binding was carried out on rat brain crude synaptosomal membranes prepared according to the method of Bernasconi et al. (1992) or to that of Hill and Bowery (1981). The presence of 100 µM GHB in the incubation medium led to different percentages of displacement of radioactive GABA (from zero to a maximum of 37%, table 1). This heterogeneity was probably due to the variation in the amount of GABA formed from GHB in the different incubation media. However, when valproate (5 mM) was present in the medium, GHB was without effect on GABA_B binding no matter what technique was used for membrane preparation (table 1).

View this table: [in this window] [in a new window]   TABLE 1 Effects of GHB on GABAB binding in the presence and in the absence of valproate Crude synaptosomal membranes were prepared according to Bernasconi et al. (1992) or Hill and Bowery (1981). Membranes were incubated in Tris-HCl 50 mM, CaCl2 2.5 mM, pH 7.4, containing 100 µM isoguvacine, [3H]GABA (25 nM, 74 Ci/mmol) and GHB 100 µM. In some experiments, valproate (5 mM) was added in order fully to inhibit GHB dehydrogenase. After a 15-min incubation at room temperature, bound [3H]GABA was separated from free [3H]GABA by rapid centrifugation at 40,000 × g for 30 min.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   GABAB binding was carried out as described by Hill and Bowery (1981). Crude synaptic membranes were prepared from a whole rat brain P2 fraction dispersed in distilled water and centrifuged at 8000 × g for 20 min. The supernatant was then centrifuged at 50,000 g, and the resulting pellet, after a second wash in distilled water, was recentrifuged and stored at 20°C overnight. The pellet was then incubated and washed as indicated in the original protocol. Binding assays were performed in 50 mM Tris-HCl buffer, pH 7.4, containing 2.5 mM CaCl2 at ambient temperature. Incubation media contained [3H]-GABA (25 nM) and 100 µM isoguvacine. Total reversible binding was measured in the presence of 100 µM baclofen. A) Displacement curve of GHB on GABAB binding from rat brain crude synaptosomal membranes. Increasing concentrations of GHB displace [3H]-GABA in the presence of 100 µM isoguvacine with an IC50 value of 23 ± 0.66 µM (nonlinear regression line, Graphpad-Prism program). B) Same experiment as in panel A, but all the incubation media contained 1.5 mM sodium valproate. IC50 is increased to a value of 5.1 ± 0.38 mM. Under the same conditions, the activity of baclofen in displacing [3H]-GABAB binding was not altered (nonlinear regression line, Graphpad Prism program). C) Same experiment as in panel A, but all the incubation media contained 1.5 mM ethosuximide. The potency of GHB in displacing GABAB binding is greatly decreased (IC50 = 0.51 ± 0.012 mM) (nonlinear regression line, Graphpad-Prism program).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Displacement curve of [3H]-GABAB binding according to Hill and Bowery (1981) in the absence () or presence () of 1.5 mM valproate. Binding was carried out in the presence of 100 µM isoguvacine, and nonspecific binding was determined with 100 µM baclofen. The differences between the two curves are not significant (P = .09, two-way ANOVA). Each data point is the mean of three separate determinations.

Effect of GHB on GABA_B binding when GHB degradation was blocked by GABA-T inhibitor. The degradation of GHB to GABA implies the presence in the brain membrane preparation of GABA-T, which is capable of converting SSA to GABA. To demonstrate the role of this GABA-T activity, GABA_B specific binding was measured in the presence of GHB alone (300 µM) or in the presence of GHB (300 µM) and aminooxyacetic acid (500 µM). The results of these experiments are shown in figure 3. GHB alone displaced specific GABA_B binding by about 35%, whereas the presence of aminooxyacetic acid completely blocked this effect of GHB. Compared with those in figure 1A, these results demonstrate that the ability of GHB to displace GABA_B binding is not uniform but depends on the batch of membranes used and their potency to convert GHB into GABA.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Displacement of GABAB binding by GHB in the presence or absence of a GABA-T inhibitor. Incubation conditions and GABAB membranes were identical to those described in the protocol of Hill and Bowery (1987). Column A = control; specific GABAB binding displaceable by 100 µM baclofen. Column B = specific GABAB binding displaceable by 300 µM GHB (significantly different from column A, P < .01). Column C = specific GABAB binding in the presence of 300 µM GHB and 500 µM aminooxyacetic acid. The inhibition of GABA-T from rat brain crude synaptosomal membranes blocks the synthesis of GABA from GHB and inhibits the effect of GHB on GABAB binding. In this set of experiments, 300 µM GHB displaced [3H]-GABAB binding by about 35%. Each data point is the mean of three separate determinations.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Determination of the Ki value for aminooxyacetic acid (339 µM, r = 0.81). Ordinate = 1/radioactive GABA produced from 200 µM GHB after a 20-min incubation, Abscissa = concentration of aminooxyacetic acid. Conditions were those described in the legend for figure 5.

Demonstration that GABA is formed from GHB in a standard incubation medium used for GABA_B binding assays. The formation of [³H]-GABA from [³H]-GHB was directly quantified in the medium incubated with the crude synaptosomal membranes under the conditions required for GABA_B binding. Membranes prepared from rat brain (method 1) were incubated for 20 min at room temperature with radioactive GHB. Chromatographic profiles revealed that all amino acids were present in significant amounts in the brain membrane extract, but only GABA was radioactive. That 0.36% of [³H]-GHB was converted into [³H]-GABA suggests a concentration of about 720 nM GABA in the medium.

Effects of antiabsence drugs on [³H]-GHB transformation into [³H]-GABA by rat brain membranes. On incubation with crude brain synaptosomal membranes under the same conditions as for the GABA_B binding assay, [³H]-GHB was rapidly converted to [³H]-GABA. The kinetics of this conversion were followed for 30 min (fig. 5). Under control conditions, the reaction was linear for about 10 min, and the GABA formation was 18.7 pmol/min/mg protein. During a 20-min incubation, about 0.37% (0.32%-0.37%) of [³H]-GHB was converted. In the presence of 1.5 mM ethosuximide or 1.5 mM valproate, GABA synthesis from GHB was linear for 30 min, and the activity was reduced to 6.6 pmol/min/mg (35% of control activity) or to 1.7 pmol/min/mg (9% of control activity), respectively.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Kinetics of [3H]-GABA formation from 200 µM [3H]-GHB in the presence of rat brain crude synaptosomal membranes prepared according to Hill and Bowery (1981). Incubations were carried out at ambient temperature in 50 mM potassium phosphate buffer, pH 7.4. The [3H]-GABA formed was separated from [3H]-GHB by ion exchange chromatography on a Dowex 50 W-X8 column and elution with 0.1 N NaOH.  Control;  In the presence of 1.5 mM ethosuximide (65% inhibition compared with control, the activity being calculated during the linear phase of the kinetics;  In the presence of 1.5 mM valproate (94% inhibition compared with the linear phase of the control).

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Determination of apparent Ki values for valproate () and ethosuximide (). Ordinate = 1/amount of radioactive GABA produced in a 20-min incubation under the conditions described in the legend for figure 5. Abscissa = concentration of inhibitors, the concentration of GHB being fixed at 200 µM. During this period of time, the conversion of GHB into GABA could be considered linear in the presence or absence of inhibitors. The Ki value measured for valproate is 1 mM (r = 0.93) and for ethosuximide is 2 mM (r = 0.97). Because the mechanism of inhibition is noncompetitive, these Ki values are the real ones measured in GABAB binding conditions.

GABA_A binding in presence of GHB. Under the conditions described by Olsen et al. (1981) for GABA binding, [³H]-muscimol was displaced by GHB with an IC₅₀ value of 4.6 ± 0.4 mM (r = 0.91). However, in the presence of 1.5 mM valproate, no significant [³H]-muscimol displacement was induced by GHB (fig. 7).

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Displacement curve of [3H]-muscimol binding in the presence of increasing concentrations of GHB. The methodology of Olsen et al. (1981) has been used, because under the conditions, the Kd values for GABA are of high affinities. An IC50 value of 4.6 ± 0.4 mM (r = 0.91) was calculated for GHB (). In the presence of valproate (1.5 mM), the displacement of radioactive muscimol disappears (). Each data point is the mean of three separate determinations.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Several authors have described the displacement of [³H]-GABA from GABA_B sites by GHB, but they have reported IC₅₀ values varying from 150 µM (Bernasconi et al., 1992), to 500 µM (Ito et al., 1995) and 796 µM (Ishige et al., 1996). Our own results have ranged from 23 µM (the present results) to about 520 µM (unpublished results) and largely depend on the batch of membranes and the protocol used for GABA_B binding. Using the conditions of Hill and Bowery (1981) or Bernasconi et al. (1992), such large variations suggest the degradation of GHB by the synaptosomal membranes, which can be modified by the methods used for preparing the membranes and/or the incubation conditions (time, temperature, pH and concentrations of GHB). GHB could be converted into GABA in vitro by the sequential action of GHB dehydrogenase, which oxidizes GHB to SSA, and then a GABA-T activity transaminating SSA to GABA. All the free amino acids that could be detected under the present conditions were identified in the extract of the synaptosomal/mitochondrial membranes, in concentrations of about 0.1 to 0.4 µM. This result suggests that the cofactors (glutamate, NADP and so on) necessary for the enzymatic conversion of GHB to GABA are present in significant amounts in the crude synaptosomal membrane preparation used for GABA_b binding experiments.

Two types of enzymes in brain are able to catalyze the oxidation of GHB to SSA (Kaufman and Nelson, 1991). One of these enzymes is a cytosolic NADP⁺-dependent oxidoreductase, whereas the other is present in the mitochondrial fraction and does not require NAD⁺ or NADP⁺. The former enzyme, which has been named GHB dehydrogenase, is more likely to be the main route for GHB degradation in brain because its inhibition by valproate and other antiepileptic drugs (trimethadione, ethosuximide) leads to an accumulation of GHB in brain (Snead et al., 1980). The mitochondrial enzyme is not sensitive to valproate. In the in vitro experiments, the presence of valproate and ethosuximide with synaptosomal/mitochondrial membranes renders GHB ineffective for displacing GABA from GABA_B binding sites. The same result is obtained when GABA-T activity of the membrane preparation is blocked by incubation with aminooxyacetic acid. Thus inhibition of the conversion of GHB to GABA results in a lack of interference with GABA_B binding.

The synthesis of [³H]-GABA from [³H]-GHB has been demonstrated in vitro under the conditions required for GABA_B binding. The concentration of GABA in the medium at the end of the 20-min incubation period in the presence of 200 µM GHB was about 720 nM, a concentration high enough to interfere with GABA_B binding. This result explains the apparent interaction of GHB with GABA_B sites described in vitro. Interference with the GABA_A receptor(s) is probably less evident because the K_d value for GABA_A binding is higher (micromolar range; see Edgar and Schwartz, 1992). Even with the membrane preparation and the binding protocol of Olsen et al. (1981; K_d values for GABA of 13 and 300 nM), GHB displaced [³H]-muscimol with a weak affinity. This result is in agreement with the studies of Serra et al. (1990) and Snead and Liu (1993), which demonstrated no modification of [³H]-muscimol or [³H]-flunitrazepam binding in the presence of 1 mM GHB. The muscimol-stimulated ³⁶Cl uptake by synaptoneurosomes was not altered in these studies, probably because of the low EC₅₀ value (8-11 µM) calculated for muscimol (Edgar and Schwartz, 1992), which should be compared with the low concentration of GABA found in the membrane medium after a 20-min incubation. In addition, it is possible that conditions of GABA_B binding (ambient temperature instead of 0-4°C, the nature of the membranes and the nature of the incubation medium) favor the synthesis of GABA from GHB in vitro.

Nevertheless, [³H]-GHB binding has been described as possessing some of the properties of the GABA_A receptor complex. It has been claimed that picrotoxin, diazepam and pentobarbital enhance [³H]-GHB binding (Snead et al., 1992), and an effect of GHB on chloride conductance has been proposed (Snead and Nichols, 1987). Under the conditions described for the above studies, an effect of GABA synthesized from GHB cannot be ruled out.

Moreover, in some studies, an antagonistic effect of bicuculline to GHB-mediated effects has been noted. Hösli et al. (1983) described a hyperpolarizing effect of GHB that is blocked by bicuculline and is associated with an increase in chloride conductance in cultured spinal, brainstem and cerebellar neurons. No apparent GHB binding sites in these structures have been reported in rat brain (Hechler et al., 1992). Thus it seems that GABA, formed from GHB in these cell cultures, is responsible for the GABA_A receptor(s) stimulation. In other experiments in vivo, GHB possesses properties of its own that were bicuculline-resistant, whereas under the same conditions, the effects of GABA were antagonized by bicuculline (Olpe and Koella, 1979).

In several electrophysiological studies carried out by in vivo administration of GHB or by application of GHB to cerebral tissue slices, GHB behaves like a GABA_B ligand, its effects being blocked by antagonists at GABA_B receptors (see, for example, Xie and Smart, 1992). In vivo, conversion of radioactive GHB into GABA has been described, and furthermore, a down-regulation of GABA receptors in the rat brain was induced by chronic GHB administration (Gianutsos and Suzdak, 1984).

Thus it appears that besides exerting a specific GHBergic effect at GHB receptors, GHB possesses GABAergic properties both in vitro and in vivo. When observed in vitro, this GABA-like effect is due to GHB conversion by the tissue or the tissue extract into GABA, which displaces radioactive GABA from its binding sites (GABA_A and GABA_B). Also in vivo, it seems likely that the GABA_B response induced by GHB is due to local conversion of GHB into GABA. Evidence does point to a regional segregation of GABA_A and GABA_B synapses (Misgeld et al., 1995); perhaps GHB selectively potentiates the GABA_B neurons. This could be realized either by regulating GABA release by a GHB-dependent mechanism at GABA_B synapses or by potentiating GABA_B synapses with GHB acting as precursor of a specific GABA pool. This phenomenon could explain an in vivo effect of GHB at both GABA_B and GHB receptors.

Such a mechanism could also explain the role of GHB in inducing general absence epilepsy in rodents. In this model, the GABA_B agonist baclofen, and GABAergic compounds in general, aggravate the symptomatology and EEG disturbances (Snead, 1992). GHB must be given at doses not less than 375 mg/kg (about 300-400 µM in brain), and the absence seizures appear with a latency of 10 to 15 min (Snead, 1991). Valproate, ethosuximide and trimethadione inhibit GHB conversion into GABA in vitro and induce GHB accumulation in brain after administration in vivo (Snead et al., 1980). Despite this GHB accumulation, these compounds normalize the EEG. Thus it seems likely that the GABA arising from GHB participates largely in the induction and severity of the epileptic syndrome. All the synthetic ligands (agonists or antagonist, Maitre et al., 1990; Hechler et al., 1993) for the GHB receptor that have been tested so far are devoid of epileptic activity, and furthermore, these ligands cannot be converted to GABA in vivo.