Gamma-hydroxybutyrate (GHB) is a drug of abuse which induces sedation and euphoria. However, overdoses can severely depress the level of consciousness or can be fatal especially when combined with other substances. Studies have suggested that the GHB-effects are mediated via actions on thalamocortical pathways and local neocortical circuits, although the effect of GHB at the level of single neocortical neurons is not clear. Using whole-cell patch-clamp recordings, we studied the effects of GHB on neocortical neurons in brain slices from 12- to 33-day-old mice. We found that GHB depressed the frequency and amplitude of GABAergic and glutamatergic spontaneous inhibitory and excitatory post-synaptic currents (IPSCs and EPSCs) driven by presynaptic action potential firing, while the amplitude and frequency of Ca2+ entry-independent miniature IPSCs were not affected. Using minimal stimulation, GHB reduced the probability of release at inhibitory synapses onto neocortical layer 2/3 pyramidal cells. Also, GHB directly hyperpolarized layer 2/3 non-pyramidal cells by up to 11 mV and inhibited action potential firing. All these effects of GHB were mediated via GABAB-receptors. In conclusion, GHB activates both pre- and postsynaptic GABAB-receptors in neocortical neurons participating in fast synaptic transmission, leading to a powerful depression of neocortical network activity. We propose that GABAB-receptor antagonists may be useful in the treatment of acute GHB intoxication.
Gamma-hydroxybutyrate (GHB) is a naturally occurring substance in the brain, where it is synthesized locally from gamma-aminobutyric acid (GABA) (Bernasconi et al., 1999). Due to its profound effects in the central nervous system, GHB has been used pharmacologically as an anesthetic agent (Kleinschmidt et al., 1999), in the treatment of alcoholism (Gallimberti et al., 1989) and opioid dependency (Rosen et al., 1996), and in sleep disorders (Mamelak et al., 1986).
Recently, GHB and its precursor gamma-butyrolactone (GBL) have become popular drugs of abuse due to their ability to induce euphoria, hallucinations, sedation and relaxation (Galloway et al., 1997; Bernasconi et al., 1999). Pharmacodynamically, GHB has a narrow concentration window within which the desired effects are obtained and aggravating side effects are absent. Thus, during abuse, when the GHB concentration is not monitored, toxic effects can easily become manifest as headache, vomiting, agitation, myoclonus and seizures, or as a severe depression of consciousness, and of cardio-respiratory function (Ingels et al., 2000). Although most patients recover quickly with no apparent sequelae, fatalities occasionally occur (Timby et al., 2000), primarily because there are currently no well- established antidotes.
As a research tool, GHB induces epileptic absence seizures in rats and mice (Aizawa et al., 1997; Hu et al., 2000). Studies have suggested that GHB modulates transmitter release and neuronal excitability in thalamic (Liu et al., 1992; Emri et al., 1996), neocortical (Hu et al., 2000) and hippocampal (Xie and Smart, 1992a,b) regions in the brain. It was suggested that GHB acts via GABAB-receptors, although the existence of distinct GHB- receptors have also been proposed (Bernasconi et al., 1999). Nevertheless, in many cases the actions of GHB on GABAergic (Xie and Smart, 1992a; Hu et al., 2000), glutamatergic (Xie and Smart, 1992a,b; Berton et al., 1999) and dopaminergic (Engberg and Nissbrandt, 1993; Madden and Johnson, 1998; Erhardt et al., 1998) neurons are blocked by GABAB-receptor antagonists.
The actions of GHB have not been studied at the level of single neurons in the neocortex. Therefore we used whole-cell patch- clamp recordings to examine its effects on membrane properties and synaptic transmission in neocortical and hippocampal neurons in mouse brain slices. We determined that GHB, by acting on pre- and postsynaptic GABAB-receptors (Mott et al., 1999), profoundly depresses neuronal activity and fast synaptic transmission in the neocortex, especially under conditions with a high level of action potential (AP) firing.
Materials and Methods
Slice Preparation and Electrophysiological Recordings
Twelve- to 33-day-old C57Black6 mice [P12–P33, 22.6 ± 0.6 days (mean ± SEM), 49 mice; for neocortical layer 2/3 cells: P12–P33, 22.3 ± 0.8 days, 37 mice, n = 62 cells; for hippocampal CA1/CA3 pyramidal cells: P20–P29, 23.5 ± 0.8 days, 10 mice, n = 17 cells; for dentate granule cells: P15–P30, 22.9 ± 1.2 days, 10 mice, n = 12 cells] were anesthetized with halothane before decapitation, in accordance with the guidelines of the UCLA Office for Protection of Research Subjects. The brains were removed and placed into an ice-cold artificial cerebrospinal fluid (aCSF) containing (mM): 126 NaCl, 2.5 KCl, 2 CaCl2, 2 MgCl2, 1.25 NaH2PO4, 26 NaHCO3 and 10 d-glucose, pH 7.3, when bubbled with 95% O2 and 5% CO2. In several experiments, l-ascorbic acid (1 mM) and pyruvic acid (1 mM) were added to the extracellular solutions to improve slice viability. The brain was glued to a platform, and 350-μm-thick coronal slices were cut with a Leica VT1000S vibratome. The slices were stored at room temperature in bubbled aCSF until transferred to the recording chamber.
During recordings, the slices were continuously perfused with bubbled aCSF at 30–32°C. Somatic recordings were made from visually identified neurons (Zeiss Axioscope infrared differential interference contrast (IR-DIC) videomicroscopy, 40× water immersion objective) with an Axopatch 200B amplifier (Axon Instruments, Foster City, CA). Patch electrodes were pulled (Narishige PP-83, Tokyo) from borosilicate glass (o.d. 1.5 mm, i.d. 1.10 mm; Garner, Claremont, CA) and were filled with a solution containing (in mM): 140 CsCl, 2 MgCl2, 10 HEPES, and titrated to a pH of 7.2 with CsOH (osmolarity 275–290 mOsm, Wescor 5520 osmometer). Voltage-clamp recordings were made at a holding potential (Vh) of –70 mV, unless otherwise stated. Where inhibitory postsynaptic currents (IPSCs) were recorded without kynurenic acid, the electrode solution contained (in mM): 135 Cs-gluconate, 10 CsCl, 5 TEA, 0.1 EGTA, 15 HEPES, pH 7.2 with CsOH. For current-clamp recordings the electrode contained (in mM): 135 K-methylsulfate, 10 KCl, 2 MgCl2, 0.2 Tris–GTP, 2 MgATP, 10 HEPES, pH 7.2, with KOH. The resistance of the electrodes was between 3 and 6 MΩ when filled with solution. The series resistance and whole-cell capacitance were monitored repeatedly during the experiment, and recordings were discontinued if the series resistance increased by >50%. The series resistance was always compensated by 70–85% using lag values of 7–8 μs. For minimal stimulation (Nusser et al., 1998), bipolar wires were inserted in an aCSF-filled theta glass (o.d. 2 mm) pulled to a tip size of ~2 μm and positioned 15–30 μm from the soma. For these experiments, the Ca2+ and Mg2+ concentrations were altered to 1 and 3 mM, respectively, to reduce the probability of release. Current pulses (20–40 μs duration) of increasing intensity were applied until IPSCs eventually appeared. IPSCs had a typical mean amplitude of ~50 pA (including failures) and did not increase further with larger stimulus intensities.
Recordings were low-pass filtered (8-pole Bessel, Brownlee 210A) at 3 kHz (1 kHz for minimal stimulation) and digitized on-line at 20 kHz using a PCI-MIO 16E-4 data acquisition board (National Instruments, Austin, TX). Spontaneous synaptic events were detected in 30 s epochs with amplitude- (typical threshold 6–8 pA) and kinetics-based criteria using custom-written LabView 5.1 based software (National Instruments) running on a Pentium III IBM/AT compatible computer. Traces were imported into a custom-written analysis program, where currents and voltages were analyzed and averaged, and amplitudes and kinetics were measured. Paired and unpaired t-tests were performed in Microsoft Excel (v. 2000), while Kolmogorov–Smirnov (KS) tests were done using Stastitica (v. 5.1, StatSoft). Data are expressed as means ± SEM, with n indicating the number of cells.
Solutions and Drugs
GHB, picrotoxin, kainic acid, kynurenic acid, l-ascorbic acid and pyruvic acid were purchased from Sigma, while tetrodotoxin was from Calbiochem (La Jolla, CA). As an Na+ salt, GHB (up to 10 mM) had no effect on the osmolarity of the extracellular solutions. The GABAB- receptor antagonists CGP 56999A and CGP 56433 were kindly provided by Dr Wolfgang Froestl (Novartis, Basel, Switzerland). These CGP compounds are highly selective GABAB-receptor antagonists, and have proven to be efficacious in brain slices at the concentrations used here (Pozza et al., 1999).
GHB Depresses the Frequency and Amplitude of Spontaneous GABAergic IPSCs in Neocortical Neurons via GABAB-receptors
Spontaneous IPSCs (sIPSCs) were recorded in the presence of the glutamate receptor antagonist kynurenic acid (3 mM) in Cl– loaded layer 2/3 pyramidal cells of the sensorimotor cortex (Fig. 1A). sIPSCs appeared as rapidly rising inward currents (10–90% rise-time = 685 ± 70 μs) with a frequency ( f ) of 14.4 ± 2.1 Hz and mean amplitudes of 45.9 ± 4.3 pA (n = 15). The sIPSCs were blocked by picrotoxin (50 μM, n = 3, not shown), indicating that they are Cl– currents mediated via GABAA- receptors. Bath perfusion of GHB (10 mM, Fig. 1A) reduced f of sIPSCs by 47.8 ± 5.3% (n = 5, P < 0.001) and IPSC amplitudes by 20.3 ± 8.7% (P = 0.08). Amplitude distributions were significantly different in control and GHB (P < 0.005, KS tests), and showed that the drug preferentially inhibited the largest IPSCs. No changes in the kinetics of the sIPSCs were observed (Fig. 1B, right traces). The GABAB-receptor antagonists CGP 56433 or CGP 56999A (both at 2 μM) completely blocked the effects of GHB on sIPSCs (Fig. 1A). f returned to 102.4 ± 2.3% of the control level (n = 4) and the effects of GHB on amplitude distributions were reversed (Fig. 1B, left panel), causing a slight enhancement of the prevalence of the largest IPSCs in three- quarters of the cells. The time-dependent effect of GHB and CGP 56433 on f are illustrated in Figure 1C, showing that both drugs exerted full effects on f following ~10 min perfusion. CGP 56999A alone had no effects on the IPSCs in three other cells (P > 0.05). The summarized effects of GHB at 1, 3 and 10 mM on f in neocortical cells are shown in Figure 1D (open bars). We also measured the effect of GHB on sIPSC f in Cl– loaded dentate gyrus granule cells (Vh = –80 mV, n = 12) and hippocampal CA1 pyramidal cells (Vh = –70 mV, n = 15), where GHB acted with a similar concentration-profile as in the neocortex (Fig. 1D).
Comparison of the Effect of GHB during High- and Low-frequency Neuronal Activity
In order to increase AP firing of GABAergic interneurons, kynurenic acid was withdrawn from the bath solution and 700–900 nM kainate was added. Layer 2/3 pyramidal cells were recorded with Cs-gluconate pipettes and held at + 5 to + 7 mV. Under these conditions, sIPSCs appear as outward currents with minimal contamination by sEPSCs. The f of sIPSCs was 24.5 ± 3.0 Hz (n = 5, not shown) and GHB (10 mM) depressed f strongly, by 73.3 ± 9.3% (P < 0.005), while the amplitudes changed from 28.0 ± 2.7 to 20.6 ± 2.0 pA (P < 0.01).
In another set of experiments, miniature IPSCs (mIPSCs, not shown) were recorded in Cl– loaded pyramidal cells in the presence of kynurenic acid (3 mM). mIPSCs were isolated using extracellular perfusion of either TTX, a sodium channel blocker, or Cd2+, a Ca2+ channel blocker. In TTX (1 μM), mIPSCs occurred at 5.0 ± 1.2 Hz (n = 4) and GHB (10 mM) reduced the mIPSC frequency by 31.3 ± 5.5% (to 3.6 ± 1 Hz, P < 0.05). mIPSCs amplitudes did not change (35.3 ± 3.5 pA vs. 33.5 ± 1.9 pA, P > 0.05). Amplitude distributions were not different in ¾ cells (P > 0.05, KS tests) while the fourth cell showed a small decrease in amplitudes. 10–90% rise-times (740 ± 144 vs. 726 ± 158 μs) and τdecay were not affected by GHB (P > 0.05). When mIPSCs were recorded in the presence of CdCl2 (50–100 μM) to block presynaptic Ca2+ entry, the effect of GHB on f was abolished ( f = 88.8 ± 15% of control, P > 0.05, n = 6), and amplitudes (109 ± 6% of control, P > 0.05) and kinetics were still unaffected by GHB.
GHB Hyperpolarizes Neocortical Non-pyramidal Cells via GABAB-receptors
Next, we examined the postsynaptic effects of GHB in neocortical layer 2/3 non-pyramidal cells in current-clamp recordings. Cells with a small, round soma and lacking an apical dendrite were chosen under IR-DIC visualization. Using K-methylsulfate pipettes, the membrane potential (Vmem) was –70.2 ± 4.5 mV (range –52 to –91 mV, n = 9). GHB (10 mM) was tested on six of these presumed non-pyramidal cells. Every 8–10 s, a depolarizing current was injected for 500 ms to evoke repetitive AP firing (Fig. 2). Four neurons showed a fast-spiking firing pattern typical for GABAergic neurons (e.g. the cell in Fig. 2A), while two others were regular spiking non-pyramidal cells. GHB caused a hyperpolarization in all cells tested by up to 11 mV (mean 4.8 ± 1.5 mV, P < 0.05), and reduced the number of APs in response to current injection. The effects on Vmem and AP firing reversed slowly upon washing (Fig. 2C, left panel). In the presence of CGP 56999A or CGP 56433 (2 μM), GHB did not significantly change Vmem, where the hyperpolarization was reduced to 0.2 ± 0.3 mV (Fig. 2C, right panel and D, n = 4; two fast-spiking and two regular spiking non-pyramidal cells tested).
GHB Lowers the Probability of GABA-release via Presynaptic GABAB-receptors in Neocortical Neurons
To test the effect of GHB at GABAergic terminals, IPSCs were evoked using perisomatic minimal stimulation in a total of 10 neocortical layer 2/3 pyramidal cells (Fig. 3). Mini-trains consisting of five pulses at 5 Hz evoked five IPSCs, termed eIPSC1–5. eIPSC1 had a mean amplitude of 50.6 ± 8.2 pA, while subsequent eIPSCs in the train displayed tetanic depression, leading to an eIPSC5:eIPSC1 ratio of 0.85 ± 0.1 (n = 5). GHB (10 mM) increased the number of failures on the first pulses in the train (Fig. 3A), thus reducing the mean amplitude of eIPSC1 to 21.6 ± 7 pA (i.e. by 56.6 ± 11%, n = 4, P < 0.01). Later eIPSCs in the train were less affected by GHB which led to an increase in the eIPSC5:eIPSC1 ratio to 1.50 ± 0.2 (P < 0.05). CGP 56999A (2 μM) antagonized the effect of GHB (Fig. 3B,C) such that eIPSC1 was not significantly depressed (from 43.4 ± 15 to 39.9 ± 14 pA, i.e. to 91 ± 5.5% of control, n = 3, P > 0.05). Also, CGP blocked the effects of GHB on short-term plasticity, where eIPSC5:eIPSC1 was 0.73 ± 0.04 (Fig. 3E). In control experiments, baclofen (5 μM) depressed eIPSC1 by 76.0% (from 61.4 to 17.3 pA, mean of two cells) and turned synaptic depression towards facilitation in both cells, leading to a change in eIPSC4:eIPSC1 from to 0.68 to 1.19 and in eIPSC5:eIPSC1 from to 0.83 to 1.00. The effects of GHB, baclofen and CGP on eIPSC1 are summarized in Figure 3D, while the changes in short-term synaptic plasticity are shown in Figure 3E.
Spontaneous EPSCs are Depressed by GHB via GABAB-receptors
We also tested whether fast glutamatergic spontaneous EPSCs (sEPSCs) were affected by GHB. sEPSCs were recorded in neocortical layer 2/3 non-pyramidal cells (Fig. 4A) in the presence of picrotoxin (50 μM). sEPSCs occurred at a frequency of 5.6 ± 3.0 Hz, with mean amplitudes of 18.2 ± 2.4 pA (n = 7). GHB depressed f by 47.6 ± 7.8% (to 3.4 ± 2.0 Hz, P < 0.05) and the amplitude by 14 ± 6.2% (P < 0.05), with no significant effect the sEPSC kinetics (Fig. 4Ab). Again, the effect of GHB was blocked by the GABAB-antagonists, whereupon f returned to 101.5 ± 7.4% of the control level (Fig. 4B). In CA3 pyramidal cells (n = 2), sEPSCs were depressed by GHB (10 mM) similarly as in the neocortex (not shown).
According to our results, GHB reduces neuronal excitability and synaptic activity in neocortical and hippocampal neurons. In short, GHB hyperpolarized layer 2/3 neurons of the neocortex and inhibited AP firing, and GHB depressed afferent synaptic input onto layer 2/3 neocortical pyramidal and non-pyramidal cells, hippocampal CA1 and CA3 pyramidal cells, and dentate gyrus granule cells. These effects of GHB were mediated entirely via GABAB-receptors.
At GABAergic connections onto neocortical layer 2/3 pyramidal cells, GHB depressed the probability of release from proximally located boutons activated by minimal stimulation. This conclusion was based on the fact that GHB increased the apparent proportion of failures in response to stimulation, and turned a slight synaptic depression into facilitation (Thomson, 2000). Furthermore, during spontaneous activity, where the pyramidal cells receive a mixture of AP-driven IPSCs and miniature events, GHB depressed the frequency of synaptic currents by ~50%. When the spontaneous activity was increased using kainate (Mody, 1998) to depolarize other neurons in the slice and increase spontaneous presynaptic firing to shift the mixture of events towards AP-driven IPSCs, GHB's effect was greater. This is probably because GHB depressed AP-initiation at the soma, and eventually increased failure of GABA-release at the synapses. Conversely, Ca2+-entry independent spontaneous release of GABA which occurs at a lower frequency, was much more resistant to GHB.
The blocking effect of GABAB-receptor antagonists points to an activation of GABAB-receptors by GHB (Lorente et al., 2000). GABAB-receptors (Misgeld et al., 1995) are present both pre- and postsynaptically in rat and mouse neocortical neurons (Fukuda et al., 1993; Badran et al., 1997; Deisz, 1999), and GABAB-receptor activation affects synaptic activity similarly to GHB. This effect is most likely to be mediated via an increase in somatodendritic K+ conductances (Takigawa and Alzheimer, 1999) and a depression of transmitter release at the nerve terminals (Fukuda et al., 1993). In support of this, GABAB-agonist and GHB-binding overlap in the neocortical layers I–III, in contrast to the thalamus and cerebellum, where GHB-binding is less prominent or absent (Mathivet et al., 1997). Thus, we can confirm the hypothesis (Hu et al., 2000) that GHB at the level of single neocortical neurons activates pre- and postsynaptic GABAB-receptors to depress neuronal network activity. However, in contrast to previous reports using microdialysis (Hu et al., 2000), we found that neocortical glutamate release was also depressed by GHB, since the frequency of sEPSCs was reduced by ~50% by GHB.
The concentration threshold for the effects of GHB was near 1 mM, which is close to that found earlier in electrophysiological studies in tegmental dopaminergic neurons (Madden and Johnson, 1998), hippocampal pyramidal cells (Xie and Smart, 1992a) and at recombinant GABAB-receptors where GHB acts as a partial agonist (Lingenhoehl et al., 2000). Furthermore, in the rodent GHB absence epilepsy model, electroencephalographic spike-and-wave seizure activity starts when GHB exceeds ~250 μM in the brain (Snead, 1991). GHB is water-soluble and passes freely across the blood–brain barrier. In humans, sedation is achieved at plasma concentrations close to 1 mM (Hoes et al., 1980), while plasma concentrations of 3 mM induces moderate anesthesia (Kleinschmidt et al., 1999). Furthermore, for abuse purposes (Galloway et al., 1997) subjects can consume tens of grams of GHB (10 g Na-GHB is equivalent to 80 mmol), which would similarly give rise to millimolar concentrations in the volume of distribution for GHB. Since the GHB concentrations found to be effective in our study are similar to those that generate behavioral effects in vivo, the effects reported here are probably relevant to understanding conditions such as human abuse or anesthesia, or GHB-induced rodent absence seizures (Banerjee et al., 1993).
In summary, GHB in millimolar concentrations causes a general depression of neuronal firing and fast synaptic transmission mediated entirely via GABAB-receptors. The net effect on neocortical activity will be determined by the resulting dynamic balance between inhibitory and excitatory mechanisms. The resemblances between GHB and GABAB-agonists at the cellular level are likely to reflect the effects in humans, where both substances induce sedation, dizziness, agitation, vomiting, respiratory and cardiovascular depression and coma (Korsgaard, 1976; Bernasconi et al., 1999). Therefore, based on our present findings, GABAB-antagonists may be beneficial in acute GHB-intoxication.
K.J. is a Glaxo/Wellcome scholar. This work was supported by NIH grant NS 30549 to I.M. We thank Dr W. Froestl for kindly providing CGP 56999A and CGP 56433.
Address correspondence to Istvan Mody, Department of Neurology, RNRC 3-131, UCLA School of Medicine, 710 Westwood Plaza, Los Angeles, CA 90095–1769, USA. Email: email@example.com.