Temperature elevation increases GABAA-mediated cortical inhibition in a mouse model of genetic epilepsy


Address correspondence to Steven Petrou, Florey Neuroscience Institutes, The University of Melbourne Parkville 3010, Vic., Australia. E-mail: spetrou@unimelb.edu.au


A missense mutation (R43Q) in the γ2 subunit of the γ-aminobutyric acid (GABA)A receptor is associated with generalized (genetic) epilepsy with febrile seizures plus (GEFS+). Heterozygous GABAAγ2(R43Q) mice displayed a lower temperature threshold for thermal seizures as compared to wild-type littermates. Temperature-dependent internalization of GABAAγ2(R43Q)–containing receptors has been proposed as a mechanism underlying febrile seizure genesis in patients with this mutation. We tested this idea using the GABAAγ2(R43Q) knockin mouse model and analyzed GABAergic miniature postsynaptic inhibitory currents (mIPSCs) in acute brain slices after exposure to varying temperatures. Incubation of slices at an elevated temperature increased mIPSC amplitude in neurons from heterozygous mice, with no change seen in wild-type controls. [3H]Flumazenil binding measured in whole-brain homogenates from mutant and control mice following elevation of body temperature showed no temperature-dependent differences in γ2-containing receptor density. Therefore, in vivo mouse data do not support earlier in vitro observations that proposed temperature-dependent internalization of γ2 R43Q containing GABAA receptors as the cellular mechanism underlying febrile seizure genesis in patients with the GABAAγ2(R43Q) mutation.

Febrile seizures (FS) are a common form of epilepsy that affect 3% of children. Up to 30–50% of patients exhibiting recurrent seizures, including temporal lobe epilepsy, have a history of FS (Cendes & Andermann, 2002). The GABRG2R43/Q43 (R43Q) mutation was identified in a large Australian family presenting with generalized (genetic) epilepsy with febrile seizures plus (GEFS+) (Wallace et al., 2001). The γ-aminobutyric acid (GABA)Aγ2 subunit plays an important role in subcellular receptor trafficking and is essential for postsynaptic clustering of receptors (Jacob et al., 2008).

In vitro data suggest that the R43Q mutation confers temperature sensitivity to the GABAA receptor complex, such that even a brief (30 min) increase in temperature to 40°C results in a marked decrease in cell surface expression and reduced current responses to GABA application (Kang et al., 2006). Based on these findings, the authors proposed that a heat-mediated reduction in GABAA-receptor cell surface expression is responsible for seizure genesis in patients with the R43Q mutation.

We used a genetically engineered knockin mouse model harboring the R43Q mutation (Tan et al., 2007) and characterized the effect of this mutation on receptor trafficking in situ by investigating phasic cortical GABAergic inhibition at elevated temperatures in brain slices. In addition, we conducted a radioligand binding assay to assess for changes in the number of membrane-bound GABAA-receptor complexes in R43Q mice after thermal stress. Thermal seizure thresholds were also assessed in these mice. In contrast to studies in heterologous systems, we observed no heat-dependent decreases in either of these markers of GABAA-receptor cell surface expression. In contrast, preincubating slices from R43Q animals at 38°C increased the amplitude of miniature inhibitory postsynaptic currents (mIPSCs) compared to that of wild-type, suggesting that internalization of R43Q-containing GABAA receptors is unlikely to be responsible for FS in patients harboring this mutation.


All experimental protocols were approved by the Howard Florey Institute Animal Ethics Committee. Thermal Seizure threshold: P14–17 wild-type and heterozygous mice body temperatures were maintained at ∼40–41.5°C for 30 min using a warm air stream (hairdryer sound pressure level of approximately 65 dB) with core temperature monitoring as outlined by Baram et al. (1997). Seizure activity was determined by behavioral analysis, and experiments were conducted blind to genotype.

Brain slice electrophysiology

Following anesthesia with 1–3% isoflurane (inhalation), postnatal day 14–17 mice were decapitated, cortical slices were cut (300-μm thick), and mIPSCs were recorded and analyzed as described in Tan et al. (2007). Drugs and salts were obtained from Sigma (Sigma-Aldrich Pty. Ltd., Castle Hill, NSW, Australia). Miniature IPSCs in layer 2/3 cortical pyramidal neurons were recorded in voltage clamp mode at 34°C from wild-type (Gabrg2R43R; wt) and heterozygous (Gabrg2R43Q; het) slices incubated at 22°C or 38°C for 1 h. Only cells recorded within 40 min of being transferred to 34°C were included for analysis. For each cell recorded, the number of events and event amplitudes were averaged over three, 90-s recording periods.

Radioligand binding assay experiments

FMZ [[3H]Flumazenil (N-methyl-[3H]-Ro 15-1788, 78.6 Ci/mol)] was obtained from Perkin Elmer Life Sciences (Boston, MA, U.S.A.). P14–17 animals were heated as described earlier. Animals were decapitated following anesthesia with isoflurane, and whole brains (excluding cerebellum) were quickly removed and homogenized individually for 30 s in approximately 10 tissue volumes of 0.32 mm sucrose at 4°C. The resulting supernatant was centrifuged (1,240 g for 10 min at 4°C) yielding a pellet that was resuspended in 2 ml Tris-HCl (50 mm, pH7.4). Pellet protein concentration was determined using the BCA assay kit (Sigma-Aldrich) with bovine serum albumin as a standard. The binding assay of Sihver et al. (1997) was modified for 96-well filter plates of the Millipore MultiScreen system (Millipore, Billerica, MA, U.S.A.). Brain homogenates were incubated for 30 min in flumazenil at saturated binding concentrations as a measure of receptor density (0.2 and 30 nm, 78.6 Ci/mmol). Filters were next washed with Tris-HCl buffer. Flumazenil binding on filters was read by a Beckman Ls6500 scintillation counter (Beckman, Brea, CA, U.S.A.) after a 20-h equilibration period with scintillant (5 ml, Ultima Gold; Perkin Elmer, Waltham, MA, U.S.A.). Cold flumazenil was used to measure nonspecific binding. All group data are expressed as mean ± standard error of the mean (SEM), and comparisons were made using a two-way analysis of variance ANOVA test unless otherwise indicated. p < 0.05 was taken as statistical significance.


Upon heating, mice underwent seizures that included hypotonic spells similar to those described by Baram et al. (1997). Control experiments in which the hairdryer fan was on but without heat did not result in hypotonic behavior (n = 3 heterozygotes and n = 3 wild-type mice). Mice heterozygous for the mutation (R43Q) displayed a seizure threshold of 38.00 ± 0.13°C (n = 44), whereas the threshold for wild-type mice was 38.44 + 0.13°C (n = 47; Student t-test, p = 0.013, mean ± SEM). We investigated the effect of temperature on GABAA receptor–mediated mIPSCs in cortical slices from wild-type and R43Q (heterozygous) animals. Cortical brain slices were preheated at either 22°C or 38°C for 1 h prior to recording at 34°C. No overt differences were observed in the quality of the tissue or voltage clamp recordings obtained from either the 22°C or 38°C slices (Fig. 1A–D). The current study confirmed our earlier observations (Tan et al., 2007) that mean mIPSC amplitude is significantly lower in layer 2/3 pyramidal neurons from heterozygous mice compared to wild-type counterparts (Fig. 1E,F). We next compared the effect of preincubation at either 22ºC or 38°C. Curiously, the only temperature-dependent change we observed was in het mice, where mIPSC mean amplitudes were significantly greater following incubation at 38°C for 1 h (Fig. 1E,F). The increase in mean mIPSC amplitude may be explained by either an increase in mutant receptor function or cell surface expression following a brief exposure to increased temperature. Cumulative probability histograms confirmed these observations (Fig. 1G,H) and further suggested that for 38°C preincubated het slices, the distribution curve was right shifted, indicating fewer small amplitude events contributing to the total event population (Fig. 1H). No change in mIPSC frequency following temperature elevation was observed (p > 0.05, Fig. 2A). Frequency histograms showed no significant difference between genotypes or treatment groups (p > 0.05; Kolmogorov-Smirnoff test; Fig. 2B).

Figure 1.

(AD) Representative examples of traces recorded from pyramidal neurons in acute cortical slices taken from wild-type (wt) and heterozygous (het) animals following incubation at 22°C (A, B) or 38°C (C, D) for 1 h. All recordings were performed at 34°C. (EH) Effect of elevated temperature on mIPSC amplitude. (E) Mean IPSCs (averaged traces) following incubation at 22°C or 38°C. (E,F) Amplitudes of mIPSCs from heterozygous slices incubated at 22°C were reduced compared to wild-type (40.76 ± 1.50 pA, 50.30 ± 2.42 pA; p = 0.02, Student’s t-test; black traces and columns). (G,H) Cumulative probability histograms illustrating event amplitudes recorded from wt or het, following incubation at 22°C or 38°C. No change in mean amplitude was observed for events recorded from wild-type slices preincubated at 22°C or 38°C (50.30 ± 2.42 pA, 49.60 ± 7.43 pA; n = 18 and 10, respectively; p = 0.91); G. In contrast, mIPSCs recorded from neurons heterozygous for the R43Q mutation showed a statistically significant increase in mean current amplitude following incubation at 38°C degrees compared with 22°C (61.22 ± 3.83 pA, 40.76 ± 1.50 pA; n = 11 and 13, respectively, p < 0.05); F, asterisk. (G) Cumulative probability histograms showed no difference between the distributions of event amplitudes recorded from wt tissue under both conditions (p > 0.05, Kolmogorov-Smirnoff test). (H) For het slices, the distribution curve was right shifted following preincubation at 38°C (p = 0.001, Kolmogorov-Smirnoff test). (EH) Number of cells compared: 18 (wt 22°C), 10 (wt 38°C), 13 (het 22°C), 11 (het 38°C) cells.

Figure 2.

Mean frequency of GABAergic mIPSCs and receptor density were unchanged with heating. (A) Mean event frequencies for each condition were (Hz): wt 22°C: 3.03 ± 0.37, het 22°C: 2.61 ± 0.43; wt 38°C: 4.42 ± 1.17, het 38°C: 4.10 ± 1.21; total n (cells) = 52; p > 0.05. (B) Cumulative probability histogram for mIPSC inter-event interval showing no significant difference between groups (p > 0.05; Kolmogorov-Smirnoff test); wt (22°C and 38°C) and het (22°C and 38°C). Data from 18 (wt 22°C), 10 (wt 38°C), 13 (het 22°C), and 11 (het 38°C) cells (A, B). (C) Mean [3H]FMZ binding was reduced in het compared with wt homogenates independent of temperature stress. [3H]FMZ binding (pM/mg protein) was reduced in het (31.20 ± 1.22, n = 8) compared to wt samples (50.82 ± 8.60, n = 9; p < 0.001) obtained from animals maintained at room temperature. Similarly, following heating, binding was reduced in heterozygous tissue (34.85 ± 2.45, n = 8) compared to wild-type (47.85 ± 3.44, n = 11; p < 0.05). No temperature-dependent changes were observed for either genotype (p > 0.05). Asterisks indicate statistical difference between wt and het only (p < 0.05). Whole-brain homogenates were obtained from nonheated (RT; n = 9 wt, n = 8 het) or heated (heat; n = 11 wt, n = 8 het) animals.

To investigate the effect of temperature on GABAA-receptor binding, a radioligand binding assay using the competitive GABAA antagonist [3H]Flumazenil (FMZ) was conducted. The density of membrane-bound γ2 subunit containing receptors was determined from whole-brain homogenates collected from heated or nonheated (RT) animals. We observed a genotype effect whereby [3H]FMZ binding was reduced in R43Q samples compared to wild-type. However, this reduction in [3H]FMZ binding was evident in samples from both the heated and RT animals (Fig. 2C) indicating no temperature-dependent change in the total number of membrane-bound γ2 subunit containing receptors.


The current study demonstrates that the GABAAγ2(R43Q) mouse model recapitulates the excitable phenotype seen in patients, providing a unique tool with which to investigate pathophysiologic mechanisms of epilepsy. We further demonstrate a significant increase in mIPSC amplitude measured in het brain slices following heating. In contrast, previous work in heterologous systems has shown a temperature-dependent decrease in GABA-mediated current for R43Q-containing GABAA receptors (Kang et al., 2006). Differences between our mouse model data and those reported by Kang et al. (2006) could be due to limitations imposed by cell culture models.

The increase in mIPSC amplitude following 38°C preincubation seen only in het neurons may be explained by an availability of an extrasynaptic reserve pool of wt GABAA receptors (for review see Jacob et al., 2008) which could preferentially replace synaptic R43Q GABAA receptors at elevated temperatures. In vitro systems may lack the necessary complexity to model these critical aspects of receptor trafficking.

We show a reduction in mean [3H]FMZ binding in heterozygous compared with wild-type homogenates, in agreement with [11C]FMZ binding data in patients heterozygous for the R43Q mutation (Fedi et al., 2006). We and others report a reduction in cell surface expression of the γ2 subunit expressing the R43Q mutation using affinity purification of total and cell surface receptor fractions at RT (Sancar & Czajkowski, 2004; Tan et al., 2007). Historical analysis suggests that these animals are seizure naive prior to this intervention (i.e., spontaneous absence seizures are not evident in these animals prior to P21; Tan et al., 2007), implying that a role for seizure activity in influencing GABAA-receptor membrane expression is unlikely. Although we cannot discount specific effects of anesthesia, the lack of temperature-dependent changes in our FMZ data suggest that the total amounts of γ2-subunit–containing receptors remain unchanged during the higher temperature preincubation and further suggest that mutant receptors are not preferentially degraded.

Increased inhibitory neurotransmission has been reported previously in animal models of epilepsy (Cohen et al., 2003; Cope et al., 2009) as well as after experimental febrile seizures (Chen et al., 1999). Specifically, Cohen et al. (2003) reported an increase in mIPSC amplitude in dentate granule cells with concurrent changes in GABAA -receptor properties. These authors propose a role for disinhibition of neural networks and altered subunit composition of GABAA receptors in seizure susceptibility.

In conclusion, we observed an increase in mIPSC amplitude following a transient increase in temperature in acute slices taken from het mice. These observations do not support a reduction in receptor expression at the cell surface following heating, as observed in heterologous systems (Kang et al., 2006). Contrasting properties of slice and cell culture environments may contribute to differences between our observations and those reported by Kang et al. (2006). Moreover, [3H]FMZ binding data show no temperature dependent change in heterozygote receptor density, further contradicting the idea that temperature-dependent receptor trafficking underlies FS genesis in the R43Q mouse or patients.


This study was supported by a New Investigator project grant from the National Health and Medical Research Council (NHMRC) of Australia (509224; to EH), an NHMRC program grant (400121; to SP), the Caitlin’s Fund for Epilepsy Research (to EH), and the National Institute of Health (NS35439; to TZB and CMD and NS046378 to MVJ).


Steven Petrou served as a paid consultant for Bionomics Limited (Thebarton, SA, Australia), which holds the intellectual property surrounding the R43Q mouse. None of the other authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.