Suppression of In Vitro Epileptiform Activity by GABA Transaminase Inhibition Is Dependent on GABA_B Receptors and GABA Transporter Type 1 (GAT1)

Abstract

Many antiepileptic drugs act by modulation of ionotropic GABA_A receptors or effects on extracellular GABA concentrations. Metabotropic GABA_B receptors also affect neuronal excitability but are not recognized as a target of antiepileptic drugs. We investigated effects of GABA_B receptors and antiepileptic drug vigabatrin on epileptiform discharges induced by 4-aminopyridine (4AP) in hippocampal brain slices from male mice. The GABA_B receptor agonist baclofen caused dose-dependent reduction in frequency of 4AP-induced epileptiform discharges but did not affect amplitude or duration of discharges. The inhibitory effects of baclofen were blocked by GABA_B receptor antagonist CGP55845 (CGP) and the K⁺ channel blocker Ba²⁺, indicating that baclofen was acting on GABA_B receptors and activating GIRK channels. Baclofen effects were independent of GABA_A receptors. Vigabatrin inhibits GABA transaminase, thereby increasing GABA concentrations. Pretreatment of brain slices with vigabatrin suppressed 4AP-induced discharges, significantly prolonging the latency to onset of spontaneous activity and reducing frequency of discharges. Similar to baclofen, vigabatrin effects on latency and frequency were reversed by GABA_B antagonist CGP and the GIRK channel blocker Ba²⁺. Vigabatrin had no effect on 4AP-induced activity in GABA transporter type 1 knock-out mice (GAT1KO), suggesting that vigabatrin effects are dependent on GAT1 function. Our results indicate for the first time that a clinically used antiepileptic drug (vigabatrin) suppresses in vitro epileptiform activity via activation of GABA_B receptors and GIRK channels. Data are consistent with a model that depicts altered thermodynamic equilibrium of GAT1 after GABA transaminase inhibition that leads to elevated extracellular GABA concentration and activation of GABA_B receptors.

Significance Statement

GABA receptor subtypes are ubiquitously expressed in central nervous system neurons and are important therapeutic targets for neuropsychiatric disease. Therapeutic mechanisms include modulation of ionotropic GABA_A receptors for treatment of anxiety and seizures and activation of GABA_B receptors for treatment of spasticity. Although GABA_B receptors have potent effects on neuronal excitability, they are not recognized targets of antiepileptic drugs. Our data demonstrate that suppression of in vitro epileptiform activity (4-aminopyridine model) after GABA transaminase inhibition with the antiepileptic drug vigabatrin is dependent on GABA_B receptors and GIRK channels. GABA_B receptor activation is hypothesized to contribute to vigabatrin’s clinical efficacy as an antiepileptic drug.

Introduction

GABA receptors are ubiquitously expressed in CNS neurons and broadly categorized as ionotropic GABA_A receptors (ligand-gated Cl⁻ channels) and metabotropic GABA_B receptors (G-protein coupled receptors, G_i/o). Both GABA_A and GABA_B receptors are important therapeutic targets in treatment of neurologic and psychiatric disease, including mood/anxiety disorders, motor spasticity, narcolepsy, and epileptic seizures (Farrant and Nusser, 2005; Lüscher and Slesinger, 2010). Many antiepileptic drugs affect GABA_A receptors, either through direct modulation of receptors (barbiturates, benzodiazepines, cenobamate) or effects on GABA transporters/GABA metabolism that increase GABA levels in the brain (valproic acid, tiagabine, vigabatrin, gabapentin; Ransom and Barker, 1976; Olpe et al., 1988; Thompson et al., 1996; Whitworth and Quick, 2001; Wu et al., 2003; Cheng et al., 2006; Sharma et al., 2020). GABA_B receptors are not recognized targets of antiepileptic drugs.

GABA_B receptor effects on cellular and network excitability are complex, with presynaptic and postsynaptic actions on both GABAergic and glutamatergic neurons. GABA_B receptors are located extrasynaptically and activation requires robust neural activity or altered function of GABA transporters (Fritschy et al., 1999; Scanziani, 2000; Kulik et al., 2003; Beenhakker and Huguenard, 2010). Presynaptic GABA_B receptors inhibit voltage-gated Ca²⁺ channels and reduce neurotransmitter release. Both pre- and postsynaptic GABA_B receptors activate G-protein coupled inwardly rectifying K⁺ channels (GIRK channels) and affect intracellular signaling pathways (primarily G_i/o; Dittman and Regehr, 1996; Ladera et al., 2008; Lüscher and Slesinger, 2010). Many of these actions serve to inhibit neuronal and network excitability (i.e., K⁺ channel activation, reduction of glutamate release). However, GABA_B receptors also reduce synaptic release of GABA (“disinhibition”) and can have paradoxical effects on network activity, particularly in thalamocortical networks where GABA_B receptors promote network oscillations and seizures (Vergnes et al., 1984; Bernasconi et al., 1992; Snead, 1992; von Krosigk et al., 1993). The effects of GABA_B receptors on seizures may vary between brain regions and seizure types, promoting hyperexcitability of thalamocortical networks but inhibiting seizures involving hippocampus or neocortex (Vergnes et al., 1997; Mares and Slamberová, 2006).

The antiepileptic drug vigabatrin inhibits GABA transaminase, the degradative enzyme for GABA, causing increased extracellular GABA levels and enhanced activation of extrasynaptic GABA_A receptors (increase tonic inhibition; Overstreet and Westbrook, 2001; Wu et al., 2003). Vigabatrin also promotes activation of presynaptic GABA_B receptors, thereby affecting neurotransmitter release (Overstreet and Westbrook, 2001; Sayin et al., 2001). In addition, vigabatrin reduces in vitro ischemic injury, a neuroprotective effect that is dependent on both GABA_A and GABA_B receptors (Costa et al., 2004). However, activation of postsynaptic GABA_B receptors and GIRK channels is not recognized as part of vigabatrin’s antiepileptic mechanism(s) of action.

We investigated effects of GABA_B receptor activation and GABA transaminase inhibition on spontaneous, epileptiform field potentials induced by 4-aminopyridine (4AP) in hippocampal brain slices. The GABA_B receptor agonist baclofen reduced frequency of spontaneous discharges (Siniscalchi and Avoli, 1992). Baclofen effects were occluded by GABA_B receptor antagonist CGP55485 or inhibition of GIRK channels with Ba²⁺ but were insensitive to GABA_A receptor inhibition. These data indicate baclofen suppresses 4AP-induced activity by activating GIRK channels. Pretreatment of slices with vigabatrin profoundly suppressed 4AP-induced activity. Similar to baclofen, vigabatrin effects were reversed by the GABA_B receptor antagonist CGP55845 and the GIRK channel inhibitor Ba²⁺. These data indicate that vigabatrin suppresses in vitro epileptiform activity via activation of GABA_B receptors and GIRK channels. Vigabatrin treatment reduced excitability of Schaffer collateral axons and activation of field excitatory postsynaptic potentials (fEPSPs) but did not alter paired-pulse ratio or presynaptic GABA_B effects on fEPSPs, suggesting vigabatrin effects are in part due to postsynaptic GABA_B receptors. Additionally, the effects of GABA transaminase inhibition with vigabatrin on 4AP-induced activity were strongly dependent on GAT1 function. Our results are consistent with a model depicting elevations of intracellular GABA after GABA transaminase inhibition that affects thermodynamic behavior of GAT1, thereby causing GAT1-dependent increases in extracellular GABA concentration to levels that activate GABA_B receptors.

Materials and Methods

All protocols and animal use were done in accordance with The Guide and with approval of local Institutional Animal Care and Use Committee.

Acute brain slices

Hippocampal brain slices were prepared from 8–14-week-old male C57Bl/6 mice. Mice were anesthetized with 4% isoflurane, decapitated, and the brain dissected free. Transverse slices (300 μm) of hippocampus were made using a vibratome (Leica VT1200). Slices were cut in ice-cold solution containing the following (in mM): 125 NaCl, 3 KCl, 24 NaHCO₃, 1.2 NaH₂PO₄, 0.5 CaCl₂, 4 MgCl₂, 20 dextrose, 1 ascorbic acid, and 1.5 pyruvate. Solutions were continuously gassed with 95% O₂/5% CO₂ to maintain a pH of 7.4. Slices were stored in cutting solution at room temperature for at least 1 h prior to transfer to recording chamber.

Electrophysiology

Extracellular field potentials were recorded from slices kept in an interface chamber (Harvard Instruments) at 34–35°C. Slices were continuously superfused at a rate of ∼3 ml/min with extracellular solution containing the following (mM): 134 NaCl, 3 KCl, 1.4 NaH₂PO₄, 26 NaHCO₃, 10 Dextrose, 2 MgCl₂, and 2 CaCl₂. Osmolarity of extracellular solution was measured with a vapor pressure osmometer (Wescor) and adjusted to 300–305 mOsm with H₂O. The pH was 7.35–7.4 when bubbled with 95% O₂/5% CO₂. This setup allowed recording field potentials in area CA1 from two slices simultaneously using Ag-AgCl wire in glass micropipettes (2–3 MΩ resistance when filled with 150 mM KCl). Recordings were made using Axopatch-1C amplifiers, a Digidata 1200 series A-D converter, and pClamp 10 software (Molecular Devices). Data were acquired at 10 kHz and low-pass filtered at 2.5 kHz. Evoked field EPSPs were measured in stratum radiatum of area CA1 using a DC-powered, constant current stimulator (WPI, A365) and bipolar tungsten electrodes (FHC, MX21AEW) placed to stimulate Schaffer collateral axons.

All chemicals were purchased from Sigma or Tocris. Vigabatrin was used at concentrations of 150–300 µM; slices were incubated in cutting solution containing vigabatrin for 2–4 h prior to transfer to recording chamber. No vigabatrin was present after transfer to recording chamber. Test substances were applied during recordings via superfusing bath solution.

Analysis

Data analysis was performed with Clampfit (pClamp 10, Molecular Devices) and Origin (MicroCal Software) software. The amplitude, half-width, and instantaneous frequency of spontaneous field potentials were determined using threshold detection paradigm provided by Clampfit software.

Data values are presented as mean ± standard error of mean (SEM). Statistical analyses were performed using Microsoft Excel and MicroCal Origin. A two-tailed, paired, or unpaired Student’s t test was used with a p value of 0.05 considered as significant. Normality of data was confirmed using Kolmogorov–Smirnov test with a significance level of 0.05; normality could not be rejected for all data presented. Data in Figure 8 were analyzed with two-way ANOVA test. Where appropriate, the actual p values are reported.

Figure 8.

Vigabatrin reduces excitatory postsynaptic field potential (fEPSP) amplitude in area CA1 (Schaffer collateral→CA1 synapse) but does not affect paired-pulse ratio or the actions of baclofen. A, Extracellular voltage in area CA1 during electrical stimulation (125 µA) of stratum radiatum under control conditions, during application of baclofen, or coapplication of baclofen and CGP. Baclofen reduced fEPSP amplitude and this was reversed by GABA_B receptor antagonist CGP. B, Time course of fEPSP amplitude for experiment in A. C, Summary data (mean and SEM) of fEPSP amplitude under control conditions or with baclofen (connected circles represent data from individual slices). fEPSP amplitude was significantly reduced by baclofen. D, Normalized fEPSPs during paired-pulse stimulation (30 ms interval) at baseline and during baclofen application. E, Paired-pulse ratio (PPR) of fEPSP slope as a function of pulse interval for experiment in D. F, Summary data (mean ± SEM) of PPR under control conditions and in presence of baclofen. Baclofen significantly increased PPR, consistent with a presynaptic effect of baclofen. G, fEPSP slope as function of stimulus intensity for control and vigabatrin (VGB)-treated slices. fEPSPs in VGB-treated slices were significantly reduced compared with control at stimulus intensities of 50–200 µA. H, Extracellular voltage in area CA1 during electrical stimulation (250 µA) of stratum radiatum in a VGB-treated slice at baseline and during application of baclofen. I, Time course of fEPSP amplitude for experiment in H. J, Summary data (mean ± SEM) of fEPSP amplitude in VGB-treated slices at baseline or with baclofen. Baclofen significantly reduced fEPSP amplitude in VGB-treated slices. K, Normalized fEPSPs during paired-pulse stimulation (30 ms interval) in a VGB-treated slice in control solutions and during baclofen application. L, PPR of fEPSP slope as a function of pulse interval for experiment in K. M, Mean ± SEM of PPR in VGB-treated slices under control conditions and in presence of baclofen. Baclofen significantly increased PPR in VGB-treated slices. **p < 0.01, *p < 0.05, n.s., not significant.

Calculations of extracellular GABA as a function of membrane potential were based on equilibrium of GABA transporter function and made assuming a stoichiometry for GABA transporter type 1 that involves translocation of 2 Na⁺ ions, 1 Cl⁻ ion, and one electroneutral GABA molecule per reaction cycle. These calculations were made using experimentally verified equations describing equilibrium membrane potential of GABA transporters, as previously described (Lu and Hilgemann, 1999; Richerson and Wu, 2003; Wu et al., 2007a):

$$\begin{array}{rl}& E_{\mathrm{m}}=[R*T/(2*Z_{\mathrm{Na}}+Z_{\mathrm{Cl}})*F]*\ln ([\mathrm{GAB}{\mathrm{A}}_{\mathrm{out}}]/[\mathrm{GAB}{\mathrm{A}}_{\mathrm{in}}]*[\mathrm{N}{\mathrm{a}}_{\mathrm{out}}^{+}]/\\ & [\mathrm{N}{\mathrm{a}}_{\mathrm{in}}^{+}]*[\mathrm{C}{\mathrm{l}}_{\mathrm{out}}^{-}]/[\mathrm{C}{\mathrm{l}}_{\mathrm{in}}^{-}]),\end{array}$$

where E_m is the membrane potential, R is the ideal gas constant, F is Faraday’s constant, Z is the valence for ionic species, and [X_out] and [X_in] are the extracellular and intracellular concentrations of transported substrate, respectively (i.e., Na⁺, Cl⁻, and GABA). This equation is rearranged to solve for equilibrium concentration of extracellular GABA as a function of membrane potential (Fig. 11).

Figure 11.

Schematic of proposed mechanisms for vigabatrin effects on 4AP-induced activity. A, Schematic of cell illustrating the stoichiometry for GAT1 function and theoretical distribution of substrate concentrations across the membrane (in mM). GAT1 translocates two Na⁺ ions, one Cl⁻ ion, and an electroneutral GABA molecule per reaction cycle. Bottom panel shows the predicted equilibrium concentration of extracellular GABA that will be produced by GAT1 under these conditions across a range of membrane potentials (Vm). B, Under basal conditions illustrated in A, there is sufficient ambient GABA to activate high-affinity extrasynaptic GABA-A receptors but postsynaptic GABA-B receptors and GIRK channels are not engaged. C, Schematic of a cell after GABA transaminase inhibition with vigabatrin, illustrating the theoretical distribution of substrate concentrations across the membrane (in mM), including a twofold increase in intracellular GABA. Bottom panel shows the predicted equilibrium concentration of extracellular GABA produced by GAT1 after treatment with vigabatrin as a function of membrane potential; GABA transaminase inhibition alters GAT1 thermodynamics leading to higher extracellular GABA concentrations (red line) compared with basal conditions (dashed line). D, Vigabatrin treatment alters GAT1 thermodynamic function to cause higher levels of extracellular GABA that enhances activation of high-affinity extrasynaptic GABA-A receptors and is sufficient to activate postsynaptic GABA-B receptors and GIRK channels. E, Schematic of GAT1KO cell during vigabatrin treatment, with hypothesized equilibrium concentration of extracellular GABA produced in absence of GAT1. In GAT1 KO, GABA levels will be elevated but they will not be modulated by vigabatrin or membrane potential. F, GAT1KO has higher levels of extracellular GABA that increase activation of extrasynaptic GABA_A receptors, but GABA_B receptors are not activated.

Results

4AP-induced epileptiform activity in hippocampal brain slices is sensitive to GABA_B receptor activation

The voltage-gated K⁺ channel blocker 4AP at concentrations of 30–100 µM induces spontaneous field potential discharges in brain slices from many brain regions, including hippocampus and cortex. This spontaneous activity occurs as periodic discharges and sporadic seizure-like events. 4AP-induced activity has been viewed as an in vitro model of refractory status epilepticus and a tool to investigate pharmacology of spontaneous, recurrent discharges (D'Antuono et al., 2010). Prior work has shown that the GABA_B receptor agonist baclofen potently suppresses 4AP-induced activity (Siniscalchi and Avoli, 1992).

We began our study by characterizing the effects of baclofen on 4AP-induced activity. Field potentials were recorded from area CA1 of acute hippocampal slices in an interface chamber during application of 4AP. Superfusion of slices with 4AP (50 µM) generated spontaneous periodic discharges, typically within 5–15 min. These 4AP-induced potentials were sensitive to bath application of baclofen; baclofen produced a significant, dose-dependent reduction in frequency of discharges but did not alter the properties of individual discharges (i.e., amplitude or half-width; Fig. 1A,C,D). The effect of baclofen was reversed by coapplication of the GABA_B receptor antagonist CGP55485 (CGP, 10 µM), indicating this was a pharmacologically specific effect (Fig. 1A,B,E; data from 6–13 slices under each condition).

Figure 1.

GABA_B receptor activation reduces the frequency of spontaneous field potentials induced by 4AP in mouse hippocampal slices. A, Field potential recordings induced by 4AP (100 µM) at baseline and during application of increasing concentrations of the GABA_B receptor agonist baclofen followed by addition of the GABA_B receptor antagonist CGP55485 (CGP, 10 µM). Bottom panels show field potentials on expanded time scale. Field potentials are recorded in stratum radiatum of area CA1. B, Time course of amplitude and instantaneous frequency of 4AP-induced field potentials for experiment in A. Solid circles represent measurements of amplitude and instantaneous frequency of individual discharges; the solid lines represent running average over 20 s epochs. Horizontal bars indicate timing of drug application in this and subsequent figures. Baclofen causes a dose-dependent reduction in field potential frequency, but not amplitude, of 4AP-induced field potentials and this effect is reversed by CGP. C, Mean and standard error or mean (SEM) of field potential amplitude in 4AP and after coapplication of 3 or 10 µM baclofen (n = 6–7 slices). D, Mean ± SEM of field potential half-width in 4AP and with either 3 or 10 µM baclofen (n = 8 slices). E, Mean ± SEM of field potential frequency in 4AP and with either 3 or 10 µM baclofen (n = 8 slices). n.s., nonsignificant, **p < 0.01.

Baclofen acts on GABA_B receptors in presynaptic terminals (Ca²⁺ channel inhibition and reduction of neurotransmitter release at both GABAergic and glutamatergic synapses) and on postsynaptic GABA_B receptors (activation of G-protein coupled inwardly rectifying K⁺ channels/GIRK channels, reduction of cAMP; Lüscher and Slesinger, 2010). GIRK channels are also associated with presynaptic GABA_B receptors and contribute to reductions in Ca²⁺ to presynaptic terminal (Ladera et al., 2008). To assess the role of GIRK channels in the acute effects of baclofen, we used the nonspecific inwardly rectifying K⁺ channel blocker Ba²⁺ to inhibit GIRK channels. Tertiapin-Q is a more selective inhibitor of GIRK channels, but this selective inhibitor provides only partial block of GIRK channels and has no effect on GABA_B responses in some cell types, which could lead us to overlook or miss important effects (Breton and Stuart, 2017). For this reason we used Ba²⁺ to provide potent inhibition of GIRK channels. We recorded field potentials from area CA1 during superfusion of 4AP to induce spontaneous discharges and then applied the GIRK channel inhibitor Ba²⁺ (100 µM) prior to and during application of baclofen (Fig. 2A,B). Ba²⁺ alone significantly increased frequency of 4AP-induced discharges, likely due to inhibition of basally active K⁺ channels, but did not affect amplitude or half-width of periodic discharges (Fig. 2C). Most importantly, the effects of baclofen were occluded in slices exposed to Ba²⁺ (Fig. 2A,B,D). In control solution, 1 or 3 µM baclofen reduced the frequency of 4AP-induced charges by 23 ± 3% and 58 ± 7%, respectively. In the presence of Ba²⁺, the effects of baclofen were significantly attenuated compared with control (Fig. 2D; 1 ± 3% and 9 ± 6% reduction in 1 and 3 µM baclofen, respectively; n = 4–6 slices under each condition, p < 0.01). These results indicate that baclofen was acting via GIRK channels to suppress 4AP-induced activity.

Figure 2.

Baclofen inhibits 4AP-induced potentials via postsynaptic GABA_B receptors and GIRK channels. A, Field potentials in the presence of 4AP and during application of the K⁺ channel inhibitor Ba²⁺ (100 µM) and baclofen plus Ba²⁺. B, Time course of frequency from the experiment in A. C, Mean and SEM for effect of Ba²⁺ on amplitude, half-width, and frequency of 4AP-induced field potentials. Solid circles represent individual measurements. D, Mean ± SEM of baclofen effects (normalized) on frequency of 4AP-induced field potentials from control and Ba²⁺-treated slices. Ba²⁺ increased frequency of discharges and occluded baclofen effects. E, Field potentials during coapplication of 4AP and the GABA_A receptor antagonist gabazine (GBZ, 10 µM), followed by application of 1 or 3 µM baclofen. F, Time course of 4AP-induced discharge frequency from the experiment in E. G, Mean and SEM for discharge frequency in the presence of 4AP (50 µM) and 4AP coapplied with either 1 or 10 µM gabazine. Discharge frequency was reduced by coapplication of gabazine compared with 4AP alone. H, Mean and SEM of normalized effects of baclofen (1 or 3 µM) on frequency of 4AP-induced field potentials in slices treated with 4AP alone (control) and 4AP coapplied with either 1 or 10 µM gabazine. Baclofen effects were not prevented by GABA_A receptor inhibition; baclofen effects were increased in GBZ-treated slices. n.s., nonsignificant, *p < 0.05, **p < 0.01.

Both pre- and postsynaptic GABA_B receptors can affect signaling at GABA_A receptors, either by reducing vesicular GABA release from interneurons or by potentiation of tonic inhibition mediated by extrasynaptic GABA_A receptors (Tao et al., 2013; Booker et al., 2020). To assess if baclofen effects were dependent on GABA_A receptors, we measured field potential activity during coapplication of 4AP and the GABA_A receptor antagonist gabazine (1 or 10 µM) followed by baclofen application (Fig. 2E–G). Gabazine concentration of 1 µM is believed to provide some selectivity for synaptic type GABA_A receptors, while 10 µM gabazine is predicted to inhibit both synaptic and extrasynaptic GABA_A receptors (Stell and Mody, 2002). Baclofen inhibited activity induced by coapplication of 4AP and gabazine (1 or 10 µM; Fig. 2E–G). Somewhat surprisingly, the frequency of discharges was significantly lower in slices exposed to coapplied 4AP and gabazine compared with control slices (4AP alone; n = 16 slices for control, n = 8 for 1 µM gabazine, and n = 15 for 10 µM gabazine; p < 0.05 for both 1 and 10 µM gabazine compared with control; Fig. 2G). Additionally, the effects of 1 µM baclofen on frequency of discharges were significantly larger in the presence of either 1 or 10 µM gabazine. Normalized frequency (mean ± SEM) during application of 1 µM baclofen was 0.77 ± 0.03 in control slices, 0.49 ± 0.04 and 0.51 ± 0.06 in 1 or 10 µM gabazine, respectively (p < 0.01 for both 1 and 10 µM gabazine; n = 6–11 slices under each condition). Similar trend was seen with 3 µM baclofen (Fig. 2H). Although GABA_A receptor blockade can promote hyperexcitability in some situations, GABA_A receptor inhibition can suppress 4AP-induced activity in the hippocampus; we speculate that this paradoxical effect of GABA_A receptor inhibition with gabazine contributed to the lower frequencies and enhanced effects of baclofen (Scalmani et al., 2023). Most importantly, baclofen effects on 4AP-induced activity were not inhibited by gabazine, indicating that GABA_B receptor effects in this model are independent of both synaptic and extrasynaptic GABA_A receptors.

The occlusion of baclofen effects by Ba²⁺, and persistent effects in presence of gabazine, support conclusion that baclofen suppresses 4AP-induced activity via GABA_B receptor activation of GIRK channels. GIRK channels are associated with both pre- and postsynaptic GABA_B receptors, so these data do not fully distinguish between pre- and postsynaptic actions of baclofen. However, because baclofen effects were independent of synaptic GABA_A receptors, reduction of GABA release due to activation of presynaptic GABA_B receptors is unlikely a contributing mechanism. Presynaptic GABA_B effects at glutamatergic synapses are investigated below (Fig. 8).

Effects of extrasynaptic GABA_A receptor activation on 4AP-induced activity

We next determined the effect of enhancing tonic inhibition on 4AP-induced activity. Selective potentiation of extrasynaptic GABA_A receptors can be accomplished with the agonist THIP and GABA transporter antagonists (Semyanov et al., 2003; Mortensen et al., 2010). We assessed the effects of enhancing tonic GABAergic inhibition on 4AP-induced field potentials (50 µM) by coapplication of the GABA transporter type 1 (GAT1) inhibitor SKF (Fig. 3A). Acute inhibition of GAT1 did not have any significant effects on amplitude or frequency of 4AP-induced discharges (Fig. 3B), suggesting that 4AP-induced activity is insensitive to enhancement of tonic GABAergic inhibition. However, direct activation of extrasynaptic GABA_A receptors with the delta subunit-selective agonist THIP did produce dose-dependent reduction in 4AP-induced activity (Fig. 3C,D). THIP concentrations of 1 and 3 µM are expected to cause large (two- to fivefold) increases in tonic GABA currents of hippocampal neurons (Parga Becerra et al., 2021), but these concentrations had modest effects on 4AP-induced activity compared with baclofen (Fig. 3C,D; 8 ± 6% and 21 ± 7% reduction in frequency with 1 or 3 µM THIP, respectively; n = 7–11 slices). Higher concentrations of THIP (10 µM) that cause massive (10- to 20-fold) increases in tonic inhibition significantly reduced frequency of 4AP-induced activity by 57 ± 9% (Fig. 3D; n = 8 slices, p < 0.01).

Figure 3.

Effects of enhancing tonic GABAergic inhibition with GAT1 antagonism or the delta subunit-selective GABA_A receptor agonist THIP on 4AP-induced activity are modest compared with effects of baclofen. A, Field potentials induced by 4AP before and during application of the GAT1 antagonist SKF89976a (10 µM). Right panel shows time course for frequency of 4AP-induced discharges for this experiment. B, Summary data for the effects of acute GAT1 inhibition on frequency and amplitude of 4AP-induced discharges (bar graphs are mean ± SEM; circles are individual measurements). GAT1 inhibition with SKF did not affect frequency or amplitude of discharges. C, Field potentials induced by 4AP before and during application of GABA-A receptor agonist THIP. Right panel shows time course for frequency of 4AP-induced discharges during application of 1 or 3 µM THIP. D, Summary data of normalized frequency in presence of 1, 3, or 10 µM THIP.

Inhibition of GABA transaminase with vigabatrin suppresses 4AP-induced activity

The antiepileptic drug vigabatrin irreversibly inhibits GABA transaminase, the degradative enzyme for GABA, and rapidly increases brain GABA concentrations, both in vivo and in slice preparations (French, 1999; Errante et al., 2002). Increased extracellular GABA concentrations produced by vigabatrin is well established to enhance activation of high-affinity, extrasynaptic GABA_A receptors/tonic inhibition and presynaptic of GABA_B receptors (Overstreet and Westbrook, 2001; Sayin et al., 2001; Wu et al., 2003). We hypothesized that vigabatrin treatment will activate postsynaptic GABA_B and GIRK receptors and thereby suppress 4AP-induced activity similar to the effects of baclofen.

We assessed the effects of vigabatrin treatment on 4AP-induced activity in hippocampal slices. Recordings of CA1/stratum radiatum field potentials were made simultaneously from control slices and slices pretreated with vigabatrin (300 µM) for 2–3 h; slices from the same preparation served as control for all experiments with vigabatrin (Fig. 4A). After incubation in control buffer or vigabatrin-containing buffer, slices were transferred to a recording chamber with control ACSF and then superfused with 4AP (50 µM; no vigabatrin was present during recordings). Data in Figure 4B shows field potential recordings from a control and a vigabatrin-treated slice during application of 4AP. Vigabatrin caused a profound suppression of 4AP-induced activity, significantly increasing the latency to onset of 4AP-induced activity and reducing both the amplitude and frequency of discharges (Fig. 4B,C; n = 4 slices under each condition, p < 0.01).

Figure 4.

Treatment of hippocampal slices with the antiepileptic drug vigabatrin suppresses 4AP-induced field potentials. A, Schematic of experimental design. Hippocampal slices were incubated in either control holding solution or solutions containing vigabatrin (300 µM) for 2–3 h prior to transfer to recording chamber. Field potentials are recorded simultaneously from both control, untreated slices and vigabatrin-treated slices. B, Field potentials recorded simultaneously from control and vigabatrin-treated slices during application of 4AP. Right panel shows portion of data enclosed in dashed box on an expanded time scale. C, Mean and SEM for onset latency, peak frequency, and amplitude of 4AP-induced discharges in control and vigabatrin-treated slices. Pretreatment with vigabatrin significantly prolonged latency and suppressed the frequency and amplitude of discharges (n = 4 slices under each condition). **p < 0.01.

Vigabatrin suppresses 4AP-induced activity via GABA_B receptors and GIRK channels

To test the hypothesis that the effects of vigabatrin on 4AP-induced activity involved GABA_B receptors, we made simultaneous measurements of field potentials from control and vigabatrin-treated (150 µM, 2–3 h) hippocampal slices during superfusion with 4AP, followed by coapplication of the GABA_B receptor antagonist CGP55485. The latency to onset of spontaneous, periodic discharges after 4AP application was 13.5 ± 2 min on average in control slices, while no activity was seen in vigabatrin-treated slices after 20 min (Fig. 5A,B; n = 6 slices). After 20 min in 4AP, the GABA_B receptor antagonist CGP (10 µM) was coapplied to slices. Inhibition of GABA_B receptors with CGP increased frequency of discharges in vigabatrin-treated slices, but not in control slices (Fig. 5A,C). The frequency of 4AP-induced discharges in control slices was unaffected by CGP; frequency was 0.38 ± 0.7 Hz after 20 min in 4AP and 0.42 ± 0.06 Hz in CGP (n = 6, p = 0.72). In contrast, vigabatrin-treated slices (from same preparation) showed a significant increase in frequency of 4AP-induced activity after application of CGP; after 20 min in 4AP, the frequency of discharges in vigabatrin-treated slices was 0.02 ± 0.02 Hz, and this increased to 0.21 ± 0.06 Hz after application of CGP (n = 6 slices, p = 0.02). Acute application of CGP to block GABA_B receptors did not completely correct the effects of vigabatrin on frequency of discharges; in the presence of CGP the mean frequency of discharges in control slices was significantly higher than in vigabatrin-treated slices (0.42 vs 0.21 Hz; n = 6, p = 0.03; Fig. 5D). These results indicate that suppression of 4AP-induced activity by vigabatrin is due in part to activation of GABA_B receptors.

Figure 5.

Vigabatrin effects on 4AP-induced field potentials are dependent on GABA_B receptors. A, Field potential recordings from control and vigabatrin-treated (150 µM) slices in the presence of 4AP (50 µM) and during application of the GABA_B receptor antagonist CGP55485 (10 µM). Right panel shows time course of field potential frequency for the experiment in A. B, Latency to onset of 4AP-induced activity in control and vigabatrin-treated slices that were studied with CGP. C, Mean and SEM (solid circles) of field potential frequency in 4AP and during application of CGP55485 in control and vigabatrin-treated slices. Lines represent measurements from individual slices (n = 6 for each condition). GABA_B receptor antagonism with CGP did not affect discharge frequency in control slices but significantly increased frequency in vigabatrin-treated slices. D, Mean and SEM of frequency in the presence of CGP55485 from control and vigabatrin-treated slices plotted for comparison (same data as in panel C). GABA_B receptor antagonism had no effect in control slices but recovered 4AP-induced activity in vigabatrin-treated slices. This recovery was incomplete, however, and in presence of CGP55485 vigabatrin-treated slices had lower frequencies compared with control slices. n.s., nonsignificant, **p < 0.01.

Based on results of experiments with baclofen, we hypothesized that vigabatrin was acting via postsynaptic GABA_B receptors and GIRK channels to inhibit 4AP-induced activity. To test this, we made simultaneous recordings from control and vigabatrin-treated hippocampal slices during superfusion with 4AP for 20–30 min and then applied the GIRK channel inhibitor Ba²⁺ (0.1 mM; Fig. 6A,B). Vigabatrin-treated slices again showed a prolonged latency to onset of 4AP-induced activity compared with control with little to no activity seen in vigabatrin-treated slices. After 20 min in 4AP the mean frequency of discharges was 0.42 ± 0.05 and 0.06 ± 0.04 Hz in control and vigabatrin-treated slices, respectively (Fig. 6B,C; n = 10 slices under each condition, p < 0.01). Similar to prior experiments, Ba²⁺ significantly increased frequency of 4AP-induced discharges in both control and vigabatrin-treated slices (Fig. 6C). However, in the presence of Ba²⁺ there was no difference in the frequency of 4AP-induced discharges between control and vigabatrin-treated slices (1.1 ± 0.2 vs 0.98 ± 0.2 Hz, respectively; n = 10 slices under each condition, p = 0.69). These results suggest that vigabatrin effects are dependent on GIRK channels.

Figure 6.

Vigabatrin effects on 4AP-induced field potentials are dependent on GIRK channels. A, Field potential recordings from control and vigabatrin-treated slices in the presence of 4AP (50 µM) and during application of the GIRK channel inhibitor Ba²⁺ (0.1 mM). Right panel shows time course for this experiment. B, Latency to onset of 4AP-induced activity in control and vigabatrin-treated slices that were studied with Ba²⁺. C, Mean and SEM (solid circles) of field potential frequency in 4AP and during application of Ba²⁺ in control and vigabatrin-treated slices. Lines represent measurements from individual slices (n = 10–11 slices for each condition). D, Mean and SEM of frequency in the presence of Ba²⁺ from control and vigabatrin-treated slices plotted for comparison (same data as in panel C). Inhibition of inwardly rectifying K⁺ channels, including GIRK channels, with Ba²⁺ reversed the effects of vigabatrin on 4AP-induced activity and in presence of Ba²⁺ the frequency of 4AP-induced discharges were similar in both control and vigabatrin-treated slices. n.s., nonsignificant, **p < 0.01.

To insure that GABA_B receptor antagonism with CGP or GIRK channel inhibition with Ba²⁺ were not by themselves inducing spontaneous field potential activity (independent of 4AP), we repeated experiments in control and vigabatrin-treated slices that were pretreated with either CGP (10 µM) or Ba²⁺ (0.1 mM) prior to application of 4AP. In experiments with CGP-pretreatment, slices were incubated with both CGP and vigabatrin and then kept in CGP after transfer to recording chamber. Neither CGP nor Ba²⁺ caused detectable spontaneous field potential activity in the absence of 4AP (Fig. 7A,D). In the presence of CGP or Ba²⁺, there was no difference in the latency to onset of 4AP-induced activity between control and vigabatrin-treated slices (Fig. 7B,E). Additionally, after pretreatment with either CGP or Ba²⁺, there were no significant differences in the frequency of 4AP-induced discharges between control and vigabatrin-treated slices (Fig. 7C,F). In slices pretreated with CGP, the frequency of discharges after 20 min in 4AP was 0.44 ± 0.15 and 0.39 ± 0.11 Hz in control and vigabatrin-treated slices, respectively (n = 6–7 slices under each condition, p = 0.78). In slices pretreated with Ba²⁺, the frequency of discharges after 20 min in 4AP was 0.88 ± 0.15 and 0.74 ± 0.13 Hz in control and vigabatrin-treated slices, respectively (n = 5–8 slices under each condition, p = 0.52). Control slices pretreated with Ba²⁺ showed a trend toward higher frequency of discharges compared with CGP-pretreated slices, but this was not significant (p = 0.06); vigabatrin-treated slices pretreated with Ba²⁺ had higher frequencies compared with those pretreated with CGP (p = 0.04).

Figure 7.

Pretreatment with GABA_B receptor antagonist CGP55485 or GIRK channel inhibitor Ba²⁺ does not induce spontaneous field potential activity but occludes vigabatrin effects on 4AP-induced activity. A, Extracellular recordings from control and vigabatrin-treated hippocampal slices in the presence of the GABA_B receptor antagonist CGP 55485 and after addition of 4AP (50 µM). CGP was included in incubation solution during pretreatment with vigabatrin. Right panel shows time course of field potential frequency for this experiment. CGP alone did not induce spontaneous field potential discharges. B, Mean and SEM of latency for onset of 4AP-induced activity in control and vigabatrin-treated slices pretreated with CGP. C, Mean and SEM of frequency for 4AP-induced activity in control and vigabatrin-treated slices pretreated with CGP. There were no significant differences in latency to onset or frequency of 4AP-induced discharges between control and vigabatrin-treated slices pretreated with CGP. D, Extracellular recordings from control and vigabatrin-treated hippocampal slices in the presence of the GIRK channel inhibitor Ba²⁺ (100 µM) and after addition of 4AP (50 µM). E, F, Mean and SEM of frequency for latency to onset (E) and frequency (F) for 4AP-induced activity in control and vigabatrin-treated slices pretreated with Ba²⁺. There were no significant differences between control and vigabatrin-treated slices pretreated with Ba²⁺. n.s., not significant.

Vigabatrin reduced field excitatory postsynaptic potentials in area CA1 but did not affect paired-pulse ratio or the effects of baclofen

Data presented above indicate that baclofen and vigabatrin suppress 4AP-induced activity via GABA_B receptors and GIRK channels. Presynaptic GABA_B receptors limit neurotransmitter release at both GABAergic and glutamatergic synapses, an effect that is due to Ca²⁺ channel inhibition but can also involve GIRK channels (Dittman and Regehr, 1996; Ladera et al., 2008). Vigabatrin's effect on network excitability could result from presynaptic GABA_B receptors that cause heterosynaptic inhibition of glutamate release at glutamatergic synapses (Dittman and Regehr, 1997).

To assess presynaptic GABA_B receptor effects, we measured fEPSPs in stratum radiatum of area CA1 evoked by electrical stimulation of Schaffer collaterals in control and VGB-treated slices (Fig. 8). Stimulus intensity was adjusted to ∼60–70% of current that produced maximal amplitude of fEPSP to avoid population spikes during paired-pulse facilitation. The amplitude of evoked fEPSPs was inhibited by baclofen (10 µM), and this was reversed by the GABA_B antagonist CGP; fEPSP amplitude was −0.85 ± 0.08 mV at baseline and −0.55 ± 0.04 mV in the presence of baclofen (n = 9 slices, p < 0.01; Fig. 8A–C). Paired-pulse facilitation at this synapse is due to presynaptic mechanisms that regulate Ca²⁺ entry, residual Ca²⁺, and transmitter release (Dittman et al., 2000). Activation of presynaptic GABA_B receptors is predicted to reduce Ca²⁺ entry, reduce fEPSP amplitude, and enhance paired-pulse facilitation (Dittman and Regehr, 1997). In the presence of baclofen, paired-pulse facilitation of fEPSP slope was enhanced in control slices, consistent with a presynaptic action of baclofen. With a paired-pulse interval of 30 ms, the paired-pulse ratio (PPR) of fEPSP slope was 1.32 ± 0.05 at baseline and 1.61 ± 0.04 in baclofen (n = 9 slices, p < 0.01; Fig. 8E,F).

The slope of fEPSPs was significantly reduced in VGB-treated slices compared with control slices across a range of stimulus intensities (50–200 µA), consistent with the reduced excitability seen during 4AP exposure (Fig. 8G). The reduction of fEPSPs after vigabatrin treatment could be due to effects of GABA_A receptors on axonal excitability or activation of presynaptic GABA_B receptors that reduce glutamate release (Ruiz et al., 2003). If the decreased excitability in VGB-treated slices were due to presynaptic GABA_B activation and reduction in glutamate release, then PPR would be expected to be increased. Additionally, if these effects were due to presynaptic GABA_B receptors, the effects of baclofen on fEPSP amplitude and PPR are predicted to be attenuated. We repeated experiments to measure fEPSPs in VGB-treated slices; stimulus intensity was adjusted to achieve fEPSP amplitudes of −0.5 to −1 mV (comparable to control slices). Baclofen suppressed fEPSP amplitude in VGB-treated slices by 27.7 ± 4% (n = 7 slices), similar to the effects of baclofen in control slices (31.8 ± 8%; Fig. 8H–J). Two-way ANOVA testing on effects of baclofen on fEPSP amplitude in control and VGB-treated slices showed highly significant effect of baclofen (p = 0.002), but no significant effects of VGB or interaction between VGB and baclofen (p values of 0.12 and 0.41, respectively). Baclofen increased PPR in VGB-treated slices similar to control slices. In VGB-treated slices, PPR (30 ms pulse interval) of fEPSP slope was 1.37 ± 0.08 at baseline and 1.67 ± 0.06 in baclofen (n = 7 slices, p < 0.05; Fig. 8K–M). Two-way ANOVA testing on effects of baclofen on PPR in control and VGB-treated slices identified highly significant effect of baclofen (p = 0.002), but no effects of VGB (p = 0.56) or interaction between VGB and baclofen (p = 0.76). These results indicate that VGB treatment does not affect presynaptic GABA_B receptors to a level sufficient to alter PPR or occlude the effects of baclofen on fEPSP amplitude and PPR at the CA1→Schaffer collateral synapse.

Vigabatrin effects are dependent on GAT1 function

Vigabatrin is well established to potentiate tonic GABAergic inhibition mediated by extrasynaptic GABA_A receptors, and this effect is believed to depend on altered thermodynamic function of GAT1 due to elevated intracellular GABA that results from GABA transaminase inhibition with vigabatrin (Richerson and Wu, 2003; Wu et al., 2007a). We hypothesized that vigabatrin effects leading to activation of GABA_B receptors are also dependent on GAT1 function.

We assessed the effects of vigabatrin on 4AP-induced activity in slices from GABA transporter type 1 KO mice (GAT1KO). Tonic GABA_A currents are increased in GAT1KO mice, presumably due to increased extracellular [GABA], but these GABA concentrations do not activate postsynaptic GABA_B receptors/GIRK channels or affect postsynaptic GABA_B responses to baclofen (Jensen et al., 2003). Slices from GAT1KO mice were prepared and incubated in control or vigabatrin-containing solutions (150 µM, 2–3 h incubation). In contrast to wild-type slices, the latency to onset of 4AP-induced activity in GAT1KO slices was unaffected by vigabatrin treatment (Fig. 9A,B; n = 6 slices, p = 0.16). As hypothesized, there were no significant differences in frequency of 4AP-induced discharges in control or vigabatrin-treated slices from GAT1KO mice (Fig. 9C,D). GAT1 inhibition and GAT1KO increases extracellular GABA concentrations, which could lead to activation of extrasynaptic GABA_B receptors, but there were no significant effects of the GABA_B receptor antagonist CGP on 4AP-induced activity in control or vigabatrin-treated slices from GAT1KO mice (Fig. 9A,C; n = 6 slices under each condition, p = 0.38 and 0.53 for control or VGB-treated, respectively). These data indicate that the effects of vigabatrin on in vitro epileptiform activity are strongly dependent on GAT1 function.

Figure 9.

Vigabatrin effects on 4AP-induced activity are dependent on GABA transporter type 1 (GAT1). A, Field potential recordings in slices from GAT1KO mouse, incubated in either control or vigabatrin (150 µM) solutions, before and during 4AP application and coapplication of 4AP with CGP. Right panel shows time course of frequency for this experiment. B, Latency to onset of 4AP-induced activity in control and vigabatrin-treated slices from GAT1KO animals. Vigabatrin did not significantly affect latency in slices from GAT1KO mice. C, Frequency of 4AP-induced activity in control (left) and vigabatrin-treated slices (right) before and during application of GABA_B receptor antagonist CGP (10 µM). CGP had no significant effect in GAT1KO slices. D, Comparison of frequency (mean ± SEM) for 4AP-induced discharges in control and vigabatrin-treated slices from GAT1KO mice (measurements taken prior to CGP application). E, Mean ± SEM of normalized frequency as function of baclofen concentration, data for wild-type (open circles) or GAT1KO slices (solid circles). 4AP-induced activity in GAT1KO slices is sensitive to baclofen, although at low concentrations baclofen effects are partially occluded.

Although GAT1KO mice are believed to have normal responsiveness to baclofen (Jensen et al., 2003), altered GABA_B receptor function in GAT1KO slices could contribute to the vigabatrin resistance of GAT1KO slices. We determined the effects of GABA_B receptor activation with baclofen on 4AP-induced activity in GAT1KO mice for comparison to control slices (Fig. 9E). At low concentrations of baclofen (1 µM), suppression of 4AP-induced activity (Hz) was significantly reduced in GAT1KO mice compared with control (24 ± 5% vs 50 ± 6% reduction, respectively; n = 13–20 slices for each condition, p < 0.01). However, higher baclofen concentrations (3–10 µM) suppressed 4AP-induced activity similarly in both GAT1KO and control slices (Fig. 9E). On average, application of 10 µM baclofen reduced frequency of 4AP-induced discharges in control and GAT1KO slice by 81 ± 6% and 76 ± 5%, respectively (n = 7–8 slices, p = 0.58). These results indicate that GAT1 KO slices retain baclofen sensitivity, although baclofen effects are partially occluded at lower concentrations (possibly due to increased extracellular GABA and partial agonism of GABA_B receptors). Furthermore, these data demonstrate that the insensitivity of GAT1KO slices to vigabatrin is not due to dysfunction/downregulation of GABA_B receptors.

To further assess the GAT1 dependence of vigabatrin effects we tested the effects of acute, pharmacologic inhibition of GAT1 using the highly selective GAT1 antagonist SKF98876a. Slices from WT animals were incubated in either control or vigabatrin-containing solutions (150 µM) with the GAT1 antagonist SKF98876a (SKF, 10 µM) and kept in SKF after transfer to recording chamber. Pharmacologic inhibition of GAT1 with SKF reduced the effects of vigabatrin, reproducing the findings in GAT1KO mice (Fig. 10). In the presence of SKF, there was a trend toward longer latency to onset and lower frequencies of 4AP-induced activity in vigabatrin-treated slices compared with control, similar to results in GAT1KO, but these changes were not statistically significant (Fig. 10B,C). Following SKF pretreatment the latency to onset of 4AP-induced activity was 10.3 ± 0.5 min in control slices versus 13.8 ± 2 min in vigabatrin-treated slices (Fig. 9B; n = 8–9 slices under each condition, p = 0.09). Following pretreatment with SKF, the mean frequency of 4AP-induced discharges was 0.33 ± 0.06 Hz and 0.13 ± 0.07 Hz in control and vigabatrin-treated slices, respectively (Fig. 10C; n = 8–9 slices under each condition, p = 0.07). These data indicate that suppression of 4AP-induced epileptiform activity by GABA transaminase inhibition is in part dependent on GAT1.

Figure 10.

Pharmacologic inhibition of GAT1 with SKF89976a attenuates the effects of vigabatrin on 4AP-induced activity. A, Field potential recordings from slices incubated with the GABA transporter inhibitor SKF89976a (SKF, 10 µM) in either control or vigabatrin-containing solutions (150 µM), before and during 4AP application. Right panel shows time course of frequency for this experiment. B, Latency to onset of 4AP-induced activity in slices pretreated with SKF or SKF with vigabatrin (n = 8–10 slices). Vigabatrin did not significantly affect latency in SKF-treated slices. C, Frequency of 4AP-induced activity in slices pretreated with SKF or SKF plus vigabatrin (n = 8–10). There was a trend for lower frequencies in vigabatrin-treated slices but this difference was not significant.

Discussion

We show that the antiepileptic drug vigabatrin suppresses in vitro epileptiform activity via activation of GABA_B receptors and GIRK channels, similar to effects of baclofen. GABA transaminase inhibition leads to elevated extracellular [GABA] and activation of both GABA_B and GABA_A receptors, thereby affecting cellular and network excitability (Overstreet and Westbrook, 2001; Wu et al., 2001; Pavlov et al., 2009; Liddiard et al., 2024). However, baclofen effects were independent of GABA_A receptors, excluding a role of presynaptic GABA_B receptors at GABAergic synapses. Additionally, vigabatrin did not alter presynaptic GABA_B function at glutamatergic synapses. These results suggest that effects of GABA transaminase inhibition are due in part to postsynaptic GABA_B receptors.

GABA transaminase inhibition, extracellular GABA, and GABA_B receptors

Inhibition of GABA transaminase, the degradative enzyme for GABA, rapidly increases brain GABA levels. Brain GABA levels increase by two- to threefold after a single dose of vigabatrin (Mattson et al., 1995; Petroff et al., 1999; Errante et al., 2002). Because intracellular brain volume is much larger than extracellular volume, increased brain GABA levels measured by MRI spectroscopy is presumed to be primarily intracellular. However, GABA must leave cells and enter extracellular space to exert physiological effects on GABA receptors. In vivo microdialysis measurements have confirmed increased extracellular GABA concentrations after vigabatrin treatment (Peng et al., 2009).

Mechanisms regulating extracellular GABA levels, and activation of GABA receptors, are expected to contribute to ambient levels GABA after GABA transaminase inhibition. Synaptic GABA release contributes to ambient GABA levels, but tonic GABA_A currents in hippocampus and cerebellum are independent of vesicular GABA release, indicating that mechanisms other than vesicular release maintain extracellular GABA at levels that activate extrasynaptic GABA_A receptors (Rossi et al., 2003; Glykys and Mody, 2007; Ransom et al., 2013). Anion channels and connexin hemichannels are permeable to neurotransmitters and mediate GABA efflux from neurons and glia (Lee et al., 2010; but see Diaz et al., 2011; Ransom et al., 2017). GABA transporters, GAT1 and GAT3, serve primary roles in maintenance of extracellular GABA levels and receptor activation (Solis and Nicoll, 1992; Keros and Hablitz, 2005; Scimemi et al., 2005; Beenhakker and Huguenard, 2010). GABA transporters are responsible for both uptake and release of GABA and contribute to the effects of vigabatrin on extracellular GABA concentration (Wu et al., 2003; Wu et al., 2007a; Peng et al., 2009).

We propose a model for vigabatrin actions that requires a sequence of events: (1) movement of vigabatrin into GABAergic interneurons and astrocytes; (2) GABA transaminase inhibition; (3) increased intracellular [GABA]; (4) altered thermodynamic operation of GAT1, leading to higher equilibrium values for extracellular [GABA]; and (5) increased activation of GABA receptors (Fig. 11). Central to our model is the well-characterized, thermodynamic function of GAT1 that involves translocation of 2 Na⁺ ions, 1 Cl⁻ ion, and one electroneutral GABA molecule per reaction cycle (Lu and Hilgemann, 1999; Richerson and Wu, 2003). Electrogenic GABA transport dynamically adjusts to changing thermodynamic conditions during neural activity (i.e., substrate concentrations and membrane potential) and operates to maintain extracellular GABA concentrations at levels that activate high-affinity extrasynaptic GABA_A receptors (Wu et al., 2007a; Fig. 11A). At typical membrane potentials with physiological substrate concentrations GAT1 will set extracellular GABA at levels that activate high-affinity GABA_A receptors but not GABA_B receptors (∼100 nM; Fig. 11B). GABA transaminase and GAT1 are present in both astrocytes and neurons, but our model depicts only a GABAergic terminal because that is where GAT1 exists in highest density (Chiu et al., 2002; Park et al., 2025).

Mechanism(s) for vigabatrin entry into cells are not fully established. Vigabatrin is a GABA analog and, like other amino acid neurotransmitters, permeates anion channels to enter cells (Peng et al., 2009; Fig. 11D). Vigabatrin weakly inhibits GABA transporters, but this effect is unlikely to account for our results because pretreated slices were transferred to recording chamber without vigabatrin prior to recording, and acute or chronic GAT1 inhibition did not reproduce vigabatrin effects (Abdul-Ghani et al., 1981; Eckstein-Ludwig et al., 1999). Vigabatrin could enter neurons and astrocytes via GAT1 but because vigabatrin inhibits GABA transport this is believed to be a minor mechanism for vigabatrin entry into cells (Schousboe et al., 1986; Eckstein-Ludwig et al., 1999; Fig. 11D).

We hypothesize that GABA transaminase inhibition alters GAT1 thermodynamics, leading to increased extracellular [GABA] and activation of GABA_B receptors and GIRK channels (Fig. 11C,D). Both pre- and postsynaptic GABA_B receptors may be activated after vigabatrin treatment. Tonic inhibition mediated by extrasynaptic GABA_A receptors will also increase after GABA transaminase inhibition and contribute to effects (Liddiard et al., 2024; Park et al., 2025). However, increased tonic inhibition likely plays a smaller role in 4AP model, because increasing tonic inhibition (acute GAT1 antagonism, GAT1 KO mice, GABA_A agonist THIP) had small and variable effects compared with effects of vigabatrin and baclofen (Jensen et al., 2003; Semyanov et al., 2003; Parga Becerra et al., 2021; Figs. 3, 9). Notably, suppression of in vitro epileptiform activity (4AP, zero Mg²⁺) requires concentrations of THIP (10 −50 µM) that produce large tonic GABA_A currents (100–300 pA; Parga Becerra et al., 2021; Liddiard et al., 2024). In contrast, tonic GABA_A currents in GAT1KO are in the range of 40–50 pA, and 4AP-induced activity was not suppressed in GAT1KO slices (Jensen et al., 2003).

In GAT1KO mice, we hypothesize vigabatrin enters cells via anion channels and inhibits GABA transaminase, but modulation of ambient GABA by GAT1 is absent. After GAT1 inhibition, tonic GABA_A currents (and presumably extracellular [GABA]) are static and are not modulated by membrane depolarization/increased intracellular Na⁺ (Ransom et al., 2013). Similarly, increased intracellular [GABA] after vigabatrin treatment is unable to modulate extracellular [GABA] in the absence of GAT1 function (Fig. 11E). Under these conditions GABA_B receptors and GIRK channels are not activated and in vitro epileptiform activity is not suppressed, despite increased tonic inhibition (Fig. 11F). Resistance of GAT1KO mice to vigabatrin could be due to failure of uptake, but direct measurements of vigabatrin uptake into cells would be needed to prove this.

Data suggest increased extracellular GABA produced by vigabatrin is uniquely positioned to activate GABA_B receptors and suppress in vitro epileptiform activity compared with increases in extracellular GABA produced by GAT1 inhibition because (1) latency and frequency of 4AP-induced discharges were not affected in GAT1KO mice, (2) GABA_B antagonists had no effect in GAT1KO mice, and (3) GAT1KO mice retain baclofen sensitivity. Furthermore, genetic knock-down of GABA transaminase generates tonic GABA_A currents of 30–50 pA, similar to tonic GABA_A currents in GAT1KO mice, yet 4AP-induced activity was unaffected in GAT1 KO slice but robustly inhibited by vigabatrin (Park et al., 2025). We speculate that GABA transaminase inhibition increases GABA in distinct spatiotemporal patterns that promote GABA_B activation, but our methods and data do not address this important question.

GABA_B receptors and seizures

GABA_B receptor agonists (baclofen and GHB) may provoke seizures and exacerbate idiopathic generalized epilepsy syndromes (Schuele et al., 2005; Shorvon and Walker, 2005; Weißhaar et al., 2012; Hleuhel et al., 2024). Basic science data also suggest that GABA_B receptors aggravate generalized epilepsy/thalamocortical seizures (Snead, 1992; Richards et al., 2000; Handforth et al., 2005; Wu et al., 2007b). For these reasons GABA_B agonists are used with caution in epilepsy and GABA_B receptors are not considered a target for antiepileptic drugs. The consequences of GABA_B receptor activation are less certain for focal epilepsy, limited data suggest baclofen may even improve seizures in some individuals (Terrence et al., 1983; Sandyk and Bamford, 1987). In cerebral palsy, a condition with high prevalence of focal epilepsy, baclofen therapy rarely exacerbates existing epilepsy or causes provoked seizures, and up to 13% of epilepsy patients enjoy a reduction in seizures on baclofen therapy (Buonaguro et al., 2005; Malalla and Gorman, 2023; Alyoubi and Abu-Zaid, 2024; Masrour et al., 2024). We speculate that GABA_B receptor/GIRK channel activation observed in hippocampal brain slices will also occur in patients treated with vigabatrin. Furthermore, because vigabatrin is highly effective for infantile spasms and refractory focal epilepsy, we conclude that GABA_B receptor activation contributes to therapeutic effects of vigabatrin (Ben-Menachem and Sander, 2011).

Significance and conclusion

Our data show for the first time that an antiepileptic drug (vigabatrin) acts in part on postsynaptic GABA_B receptors and GIRK channels to suppress epileptiform activity. Although GABA_B receptors have potent effects on excitability of neural networks and are activated during robust neural activity, they are not considered an antiepileptic target (Lüscher and Slesinger, 2010; Cattani et al., 2023). Our study is limited by the use of an in vitro model of epileptiform activity and our focus on interictal, periodic discharges; our data does not provide information on GABA_B effects on seizure-like events or seizures. Our study is also limited by use of pharmacological tools and indirect measurements to characterize vigabatrin's mechanism(s) of action. Future experiments using different methods are needed to further characterize GABA_B receptor activation after vigabatrin treatment. Our results suggest that GABA_B receptor activation contributes to the clinical efficacy of vigabatrin and motivate future work on GABA_B receptors in both focal and generalized epilepsy syndromes.