Our study used male and female wild-type Harlan C57BL/6J-OlaHsd and $\gamma$2R43Q knock-in mice bred into a background of Harlan C57BL/6J-OlaHsd mice. Behavioral and electrographic markers of absence epilepsy in these animals were confirmed by video-EEG monitoring. Surgery and electrode implantation were performed as described by Nelson et al. [37]. Briefly, P24 mice were implanted under isoflurane (#sc-363629Rx; Santa Crus Biotechnology, Dallas, TX, USA) anesthesia (1%–2% in 100% O${}_{\text{2}}$) for chronic EEG recordings with gold plated miniature screw electrodes over the right and left frontal and parietal cortices, and one over the cerebellum as a reference. Two vinyl-coated braided stainless steel wire electrodes were placed in the nuchal muscle for electromyogram (EMG) recording of muscle activity. All electrodes were gathered into a flexible cable and connected to the Multichannel Neurophysiology Recording system (Tucker-Davis Technologies, TDT, Alachua, FL, USA). EEG and EMG signals were collected continuously at a sampling rate of 256 Hz (digitally filtered between 0.1 and 100 Hz). Continuous EEG recordings with occasional video monitoring were made and SWDs were manually scored off-line. Animals were given a 3-day recovery period after surgery before initiating SWD-scoring. A SWD event was defined as a brief (~2 seconds long) ~6 Hz signal synchronized across all EEG leads, with a corresponding lack of signal in the EMG lead. Only SWD events that occurred $>$2 min from slow-wave-sleep periods were used for quantification. SWD event durations were measured from the first synchronized positive peak signal to the last synchronized positive peak within an event. “Seizures” were defined as groups of SWD events separated from other events by 30 seconds. Ictal intervals were defined as the time between the beginnings of consecutive seizures. All animal procedures followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Wisconsin-Madison (No. A3368-01). All facilities were inspected and accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC).

L655,708 (L9787), GANX (G7795) and THIP (T101; 4,5,6,7-tetrahydroisoxazolo[5,4-c]-pyridine-3-ol HCl) were all obtained from Sigma (St. Louis, MO, USA). L655 and GANX were dissolved in a 30% dimethyl sulfoxide (DMSO)-saline solution (v/v), whereas THIP was dissolved in 100% saline. Mice were intraperitoneally (i.p.) injected with 2 mg/kg doses of L655, 2 and 5 mg/kg doses of GANX, or 0.5 and 1.5 mg/kg doses of THIP. 160 µL of solution was injected for each drug. L655 was administered to wild type (RR) mice 2 and 4 hours after lights out for 2 consecutive days beginning 5 days after surgery. These mice were not injected for the subsequent 2 days but were given vehicle injections on day 9. GANX or THIP injections were administered to RQ mice 1 and 4 hours after lights out. Drug injections for RQ mice began on day 5 post surgery and consisted of 2 injections of one drug and dose, with a different drug and dose for days 6, 10, and 11. No injections were given to RQ mice on days 7–9.

Horizontal slices (400 µm) were prepared from the brains of RR and RQ mice of either sex (16–26 days old). All procedures were approved by the University of Wisconsin IACUC. Mice were anesthetized with isoflurane, decapitated, and the brain was removed and placed in an ice-cold cutting solution containing (in mM): 125 NaCl, 25 NaHCO${}_{\text{3}}$, 2.5 KCl, 1.25 NaH${}_{\text{2}}$PO${}_{\text{4}}$, 0.5 CaCl${}_{\text{2}}$, 3.35 MgCl${}_{\text{2}}$, 25 D-Glucose, 13.87 sucrose, and bubbled with 95% O${}_{\text{2}}$ and 5% CO${}_{\text{2}}$. Slices were cut using a vibratome (Leica VT 1000S, Global Medical Imaging; Ramsey, MN, USA) and placed in a bubbled incubation chamber containing standard artificial cerebrospinal fluid (aCSF) (in mM): 125 NaCl, 25 NaHCO${}_{\text{3}}$, 2.5 KCl, 1.25 NaH${}_{\text{2}}$PO${}_{\text{4}}$, 2 CaCl${}_{\text{2}}$, 1 MgCl${}_{\text{2}}$, 25 D-Glucose, at room temperature for 1 hour before being used for recordings. Whole cell patch-clamp recordings were made from somatosensory cortical layer 2/3 pyramidal cells or ventrobasal thalamic relay cells, visualized using an upright differential interference contrast microscope (Axioskop FS2, Zeiss; Oberkochen, Germany). Patch pipettes were pulled from thin-walled borosilicate glass (World Precision Instruments; Sarasota, FL, USA) with a resistance of 3–5 M$\mathrm{\Omega }$ when filled with an intracellular solution containing (in mM): 140 KCl, 10 EGTA, 10 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 20 phosphocreatine, 2 Mg2ATP, 0.3 NaGTP (pH 7.3, 310 mOsm). Voltage clamp (–60 mV) recordings were made in a submerged chamber, perfused with bubbled aCSF (4 mL/min) containing (500 nM) tetrodotoxin, (25 µM) DNQX, and (50 µM) AP5 at room temperature using a MultiClamp 700B amplifier (Axon Instruments; Foster City, CA, USA), filtered at 4 kHz and digitized at 10 kHz using a Digidata 1322A analog-digital interface (Axon Instruments). Data were acquired to a Macintosh G4 (Apple Computer; Cupertino, CA, USA) using Axograph X v1.1.4 (Molecular Devices; Sunnyvale, CA, USA).

Data segments (30 seconds) just prior to and 90 seconds after drug administration were analyzed to quantify inhibitory tonic currents. All-point amplitude histograms were computed for each segment and fit with a Gaussian function only to the outward current portions relative to the peak in order to omit components arising from inward miniature inhibitory post-synaptic currents (mIPSCs) [38]. Tonic current was calculated as the difference between the fitted Gaussian means before and after (100 nM or 1 µM) THIP, (30 nM) ALLO or (10 nM) GANX administration. Current density (pA/pF) was calculated by dividing the current by cell capacitance. Bicuculline (100 µM; bicuculline methiodide #2503, Tocris Bioscience, Minneapolis, MN, USA)) was added at the conclusion of some experiments for each drug tested to verify full current block and, therefore, only GABAergic contribution.

The Kruskal-Wallis test of medians was used to compare multiple groups with a Dunn’s post-hoc evaluation. Tonic current amplitude and density data were normally distributed; therefore, an analysis of variance (ANOVA) was used to compare multiple groups with a Bonferroni post-hoc evaluation. A p-value of 0.05 is considered significant. MATLAB (version 2013b, Mathworks Inc, Natick, MA, USA) and Prism (macOS v10.0.0, GraphPad Software, Boston, MA, USA) software were used.

The $\gamma$2R43Q mutation confers absence seizures and generalized SWDs in humans [32] and knock-in (RQ) mice [35], consistent with results presented in this study. Fig. 1 illustrates bilateral, synchronous (~6 Hz) SWDs in a RQ mouse using continuous EEG and EMG recordings. Quantification was done off-line after recordings were completed. A “seizure” was classified as two or more individual SWD events occurring 30 seconds apart. SWDs were assessed for individual event duration, events per seizure, seizure duration and ictal intervals. EEG and EMG recordings from one RQ mouse during a seizure are presented (Fig. 1a,b), along with quantified SWD assessment for three different RQ mice (Fig. 1c). All RQ mice assessed with EEG and EMG monitoring presented with synchronized SWDs across all EEG leads with coincident cessation of EMG activity. Some spindle activity but no SWDs were observed in drug-naïve wild type (RR) mice.

Fig. 1.

RQ mice express spike-and-wave discharges (SWDs) associated with absence epilepsy. (a) Electroencephalogram (EEG) recording of an RQ mouse (red). Top trace to bottom trace: frontal right cortex (F.R.); frontal left cortex (F.L.); parietal right cortex (P.R.); parietal left cortex (P.L.); electromyogram (EMG). Note the brief yet frequent (~11 times during the 1.5-minute segment) synchronized events that occur across all EEG leads during the absence of signal in the EMG. (b) Expanded F.R. EEG recording from grey bar in (a) (10 seconds). Note the brief ~6 Hz SWD events (black bars) that occur 3 times during the 10 second trace. (c) Cumulative probability distributions from three different RQ mice (solid & dashed black lines and red dashed line represent individual mice) all display similar characteristics for SWD event durations, SWDs per seizure, seizure durations and ictal-intervals. SWDs were not detected in litter-mate control mice (not shown). CDF, cumulative density function; RQ, $\gamma$R43Q mutant.

Although RQ mice express slightly decreased IPSC amplitudes [35], this mutation has also been shown to hinder GABA${}_{\text{A}}$ receptor assembly, trafficking and surface expression [39, 40, 41, 42, 43]. Based on studies in transfected cultured neurons, Eugène et al. [42] reported that this mutation may contribute to absence epilepsy by reducing tonic inhibition. To directly test this hypothesis in RQ animals, we examined tonic inhibition levels in brain slices from RR and RQ knock-in mice. Using whole cell voltage clamp recordings, we found that whereas RR neurons exhibit a substantial inhibitory tonic current, this current was entirely abolished in RQ mice somatosensory cortical layer 2/3 neurons (mean $\pm$ standard error of the mean (SEM) in pA, N) (RR: 6.0 $\pm$ 0.8, 4; RQ: –1.2 $\pm$ 1.6, 4, p 0.05; Fig. 2a,b), as well as in thalamic relay neurons (RR: 6.9 $\pm$ 2.2, 9; RQ: 0.6 $\pm$ 0.7, 6, p 0.05; Fig. 2c,d). One-sample t-test indicated that tonic currents in RQ were indistinguishable from zero (p $>$ 0.45).

Fig. 2.

Tonic currents are abolished in RQ cortex and thalamus. (a) Example voltage-clamp traces for RR (black) and RQ (red) cortical layer 2/3 cell recordings during 100 µM Bicuculline administration (maroon bars). Insets; Corresponding all-points amplitude histograms for data before (black) and after (grey) bicuculline administration. Histograms were fit with a Gaussian function (dark grey and maroon traces) only on the right side of the distribution, thus omitting components due to phasic miniature inhibitory post-synaptic currents (mIPSCs). (b) Tonic current amplitude (pA) (left axis) and tonic current density (pA/pF) (right axis) are abolished in RQ cortical cells (*p 0.05) compared to control. (c,d) Same as a-b, but for ventrobasal thalamic relay neurons. RR, wild type.

The tonic current in RR somatosensory cortical layer 2/3 cells (Fig. 2a) was completely blocked by the $\alpha$5 subunit-selective inverse agonist L655,708 (L655: 30 µM; Fig. 3a), consistent with previous studies showing this subunit is responsible for most or all of the native tonic inhibition in these neurons [44].

Fig. 3.

RQ mice display region- and subunit-specific changes in tonic inhibition. (a) Example voltage-clamp traces for RR cortical layer 2/3 cell recordings during 30 µM L655,708 administration (purple bar). The current density blocked by L655,708 was not significantly different than that blocked by bicuculline (see Fig. 2). In cortical neurons, both THIP. (b) (1 µM; green bar) and allopregnanolone. (c) (ALLO; 30 nM; blue bar) induce indistinguishable current amplitude and density in RQ (red traces) compared to RR (black traces). In thalamic relay neurons, however, THIP (d) and ALLO-induced (e) current densities are significantly reduced in RQ compared to RR (~50%; *p 0.05). ALLO, allopregnanolone; THIP, 4,5,6,7-tetrahydroisoxazolo[5,4-c]-pyridine-3-ol HCl.

In contrast, application of the agonist THIP (1 µM, a concentration previously shown to be selective for $\delta$ subunit-containing receptors [45]) evoked currents of similar magnitude in RR and RQ cortical neurons (mean $\pm$ SEM in pA, N) (RR: 21.4 $\pm$ 5.7, 4; RQ: 23.8 $\pm$ 2.2, 5; p = 0.67; Fig. 3b). A similar profile of effect was observed with allopregnanolone (ALLO: 30 nM; Fig. 3c), a neurosteroid that also selectively activates $\delta$ subunit-containing GABA${}_{\text{A}}$ receptors [46, 47]. Together, these results suggest that receptors containing the $\delta$ subunit are present in cortical neurons and can be recruited by both exogenous drugs and endogenous modulators, providing potential pharmacological pathways to rescue cortical tonic inhibition in cases where it has been genetically compromised.

Distinct from cortical layer 2/3 neurons, thalamic relay neurons rely solely on $\delta$ subunit-containing GABA${}_{\text{A}}$ receptors to produce inhibitory tonic currents [7, 45, 48]. We found that RQ relay neurons in the ventrobasal thalamus responded to THIP (1 µM) with 47% of the current produced in RR thalamic neurons (mean $\pm$ SEM in pA, N) (RR: 131.7 $\pm$ 31.2, 5; RQ: 69.3 $\pm$ 22.4, 4, p 0.05; Fig. 3d). Similarly, in RQ thalamic neurons, ALLO (30 nM) produced 39% of the current observed in RR (RR: 34.7 $\pm$ 6.5, 5; RQ: 13.7 $\pm$ 3.8, 3, p 0.05; Fig. 3e). These results suggest that RQ mouse thalamic relay neurons express reduced levels and/or activation of $\delta$ subunit-containing GABA${}_{\text{A}}$ receptors.

Previous work has demonstrated that a positive correlation between SWDs and thalamic inhibitory tonic current amplitude exists [7], leading to the conclusion that enhanced GABAergic tonic inhibition is “necessary and sufficient” to cause typical absence epilepsy [9, 10]. In this study, we found (by using the $\alpha$5 subunit-selective blocker L655) that altering thalamic tonic inhibition is not “necessary” for SWD generation and, furthermore, that the loss of $\alpha$5 subunit-mediated tonic inhibition (present mainly in cortex and hippocampus) is “sufficient” to produce SWDs (Fig. 4). Inhibitory tonic currents in somatosensory cortical layer 2/3 principal neurons are generated by $\alpha$5 subunit-containing GABA${}_{\text{A}}$ receptors, evident by the total block of this current by the $\alpha$5 subunit-selective inverse agonist L655 (Fig. 3a). This is in contrast to thalamic relay neurons that do not express a5 subunit-containing GABA${}_{\text{A}}$ receptors [49, 50]. Intraperitoneal (i.p.) administration of L655, at a concentration (2 mg/kg) known to bind the majority of $\alpha$5 subunit-containing receptors [51], produced SWDs (~6 Hz) in RR mice (RRL6) that are electrographically similar, yet distinct, to SWDs observed in RQ mice (Fig. 4a,b). L655 administration induced hallmark synchronous and bilateral SWDs accompanied by behavioral arrest, although these L655-induced SWDs (L6-SWDs) display longer event durations (p 0.05), fewer events per seizure (p 0.05) and shortened seizure durations (p 0.01) compared to RQ (Fig. 4c). The appearance of SWDs 3-days post final L655-injection (Fig. 4d, hour 1 vehicle) highlight lingering plasticity and epileptogenesis induced by the earlier acute insults that provoked SWDs.

Fig. 4.

Blocking cortical tonic inhibition produces SWDs in wild-type mice. (a) Electroencephalogram (EEG) recording of a wild-type (RR) mouse i.p. injected with 2 mg/kg of the GABA${}_{\text{A}}$ receptor $\alpha$5 subunit-selective inverse agonist L655,708 (RRL6; purple). Similarly to RQ mice, note the brief yet frequent (~6 times during the 1.5 min. trace) synchronized events that occur across all EEG leads during the absence of signal in the EMG. (b) Expanded F.R. EEG recording from grey bar in (a) (10 seconds) displays prolonged ~6 Hz SWD event (black bar). (c) Cumulative distributions show RRL6 mice (purple line) display significantly longer SWD event durations (*p 0.05), fewer SWDs per seizure (*p 0.05) and shorter seizure durations (*p 0.05) than RQ mice (red line). (d) Quantification of SWDs show that RRL6 mice did not display SWDs (purple boxes) prior to L655 injection (Hour 1: L655, middle purple bars) but do display SWDs for hours following injection (Hour 2: *p 0.05; Hour 4: *p 0.05). Black outlined bars represent median values. SWDs were still present in RRL6 mice 3 days after the final L655 dosing (vehicle: Hour 1, right turquois diamonds; *p 0.05).

Although RQ principal cortical neurons lack inhibitory tonic currents (Fig. 2a), these neurons also display an inhibitory conductance in response to selective $\delta$ subunit-selective GABA${}_{\text{A}}$ receptor agonists THIP (1 µM) and ALLO (30 nm) (Fig. 3b,c). This finding is consistent with the presence of latent $\delta$ subunit-containing GABA${}_{\text{A}}$ receptors in RQ cortical neurons and suggests that the lost tonic inhibition in these neurons can be rescued. We used whole-cell patch-clamp recordings to titrate a concentration of selective-$\delta$ subunit-containing GABA${}_{\text{A}}$ receptor modulators that rescued wild-type tonic inhibition levels in RQ cortical neurons. We found (Fig. 5a) that a low concentration of THIP (100 nM) or the synthetic neurosteroid GANX (10 nM) can activate a latent inhibitory conductance in RQ cortical neurons equal to the inhibitory tonic current observed in RR cortical neurons.

Fig. 5.

Rescuing cortical tonic inhibition alleviates SWDs in RQ mice. (a) Voltage-clamp experiments reveal that tonic current amplitude (left y-axis) and density (right y-axis) levels can be rescued in RQ (red bars) cortical layer 2/3 neurons with 100 nM THIP or 10 nM ganaxolone (GANX) treatment, whereas a 1 µM THIP produces 2–4 times more holding current amplitude (**p 0.01) and density (**p 0.01) than that observed in untreated RR neurons (#). (b) Schematic depicts administration and data collection (Hour 2) times and drug-day schedules investigating treatment conditions for RQ mice. GANX (2 and 5 mg/kg) solutions were i.p. injected into RQ mice twice a day for 4 out of 7 days. (c) RQ-SWD event quantification shows that the 2 mg/kg GANX (*p 0.05) (Dark Blue) treatment decreased SWD expression compared to control hours, while 5 mg/kg GANX (p = 0.12) (Light Blue) treatment trends towards ameliorating SWD expression. (d) Cumulative distributions of RQ-SWD activity shows that SWD event duration (*p 0.05) and seizure duration (*p 0.05) were decreased with the 2 mg/kg GANX treatment.

Using video-EEG monitoring, we investigated if treatment with the $\delta$ subunit-selective GABA${}_{\text{A}}$ receptor agonist (GANX) could ameliorate the SWDs observed in RQ mice. RQ mice were i.p. injected twice a day with GANX for 4 out of 7 days (Fig. 5b). Two concentrations of GANX (2 and 5 mg/kg) were tested for amelioration of SWD expression. We found that only the lower concentration (2 mg/kg) of GANX significantly (p 0.05) decreased RQ-SWD expression (Fig. 5c). This low dose GANX treatment (2 mg/kg) also decreased seizure duration (p 0.05) and event duration (p 0.001) (Fig. 5d). The efficacy of the low GANX dose (2 mg/kg), being half the ED${}_{50}$ dose that protects against partial seizures [30, 51, 52], suggests that the mechanism diminishing SWDs in RQ mice involves activation of latent $\delta$ subunit-containing GABA${}_{\text{A}}$ receptors in cortical neurons.

Studies suggest a dichotomy of effects for neurosteroids in absence epilepsy: lower levels can ameliorate SWDs in RQ mice, whereas higher concentrations can result in SWD exacerbation or generation [20]. Based on the results from this study, we suggest a concentration-dependent relationship of thalamocortical tonic inhibition in regard to SWDs and absence seizure modulation. Evaluation of genetic (Genetic Absence Epilepsy Rat from Strasbourg (GAERS), stargazer, lethargic) and pharmaco-induced (GHB, PTZ) rodent models of absence epilepsy provide ample evidence that excessive thalamic tonic inhibition triggers SWDs [7, 31]. However, the novel RRL6-absence animal model introduced here, and the beneficial effects of low-dose GANX treatment in RQ mice, combine to indicate that a reduction in cortical tonic inhibition also results in SWDs. These findings suggest that an ‘optimal level’ of tonic inhibition in the thalamocortical circuit is a requirement for normal function and that deviation either above or below this optimal range results in aberrant thalamocortical function, SWDs and absence seizures.