Multiple molecular mechanisms for a single GABA_A mutation in epilepsy

Abstract

Objective:

To understand the molecular basis and differential penetrance of febrile seizures and absence seizures in patients with the γ2(R43Q) GABA_A receptor mutation.

Methods:

Spike-and-wave discharges and thermal seizure susceptibility were measured in heterozygous GABA_A γ2 knock-out and GABA_A γ2(R43Q) knock-in mice models crossed to different mouse strains.

Results:

By comparing the GABA_A γ2 knock-out with the GABA_A γ2(R43Q) knock-in mouse model we show that haploinsufficiency underlies the genesis of absence seizures but cannot account for the thermal seizure susceptibility. Additionally, while the expression of the absence seizure phenotype was very sensitive to mouse background genetics, the thermal seizure phenotype was not.

Conclusions:

Our results show that a single gene mutation can cause distinct seizure phenotypes through independent molecular mechanisms. A lack of effect of genetic background on thermal seizure susceptibility is consistent with the higher penetrance of febrile seizures compared to absence seizures seen in family members with the mutation. These mouse studies help to provide a conceptual framework within which clinical heterogeneity seen in genetic epilepsy can be explained.

Family members harboring the GABA_A γ2(R43Q) mutation display multiple seizure types, incomplete penetrance, and variable seizure severity,^1,2 a common feature of genetic epilepsies.³ To predict clinical outcomes for patients based on their personal genomes, it is important that we develop a conceptual framework explaining the genetic and molecular basis of heterogeneity. Animal models of epilepsy, based on human mutations, provide a means of investigating this. The knock-in mouse model based on the GABA_A γ2(R43Q) mutation recapitulates the 2 major phenotypes seen in family members, including febrile seizures and typical absence seizures.^4,5 However, the molecular mechanisms causing epilepsy in this model are unclear. In vitro studies suggest that the deficit could be through haploinsufficiency (loss-of-function) or a dominant impact of the mutated protein.⁶ To investigate this, we compared the seizure phenotypes of heterozygous Gabrg2 knock-out mice with those of knock-in mice.

Penetrance is relatively low for absence seizures in the GABA_A γ2(R43Q) family.^1,2 By backcrossing the GABA_A γ2(R43Q) knock-in mouse to strains with different seizure susceptibility, we have demonstrated that the spike-wave phenotype requires additional susceptibility alleles for full expression,^5,7 potentially explaining the low penetrance of absence in the family. In contrast, febrile seizures segregate as a more highly penetrant autosomal dominant trait, which suggests that background genetics have less impact.^1,2 Here we investigate the influence of genetic background on thermal seizure and spike-wave susceptibility in the knock-in mouse model.

METHODS

Mice.

All experiments were approved by the Animal Ethics Committee at the Florey Institute of Neuroscience and Mental Health (09-046). Genotyping of the GABA_A γ2(R43Q) knock-in mouse model and the GABA_A γ2^+/− heterozygous knock-out mice was done at P12 using PCR-based methods previously described.^5,8 Mice were used between the ages P35–P40 for electrocorticogram recordings. For thermogenic stress assay mice were used at P16–P17. Animals were crossed to >N9 into a DBA/2J and to >N20 for C57BL/6J background.

Electrocorticogram recordings.

For electrocorticograms, surgeries were performed as previously described.⁵ Mice were anesthetized with 1%–3% isoflurane and 2 epidural silver “ball” electrodes implanted on each hemisphere of the skull. Electrodes were placed 3 mm lateral of the midline and 0.5 mm caudal from bregma. A ground electrode was placed 2.5 mm rostral from bregma and 0.5 mm lateral from the midline. Mice were allowed to recover for at least 24 hours after surgery. Electrocorticograms were continuously recorded in freely moving mice for a 4-hour period during daylight hours following a standard 30-minute habituation period. Total time in seizure was defined as the percent time each animal spent having spike-and-wave discharges (SWD). Data from male and female mice were pooled in this study. Signals were bandpass filtered at 0.1 to 200 Hz and sampled at 1 kHz using Powerlab 16/30 (ADInstruments Pty. Ltd., Sydney, NSW, Australia). In a subset of mice electrocorticogram recordings were simultaneously recorded with video to enable the scoring of behavioral arrest. Statistical comparison was made using an unpaired t test (GraphPad Prism, La Jolla, CA).

Thermogenic seizure testing.

P16–P17 mice were placed in a container heated to constant 42°C and the time to first tonic-clonic seizure recorded based on methods previously described.⁹ Mice were killed immediately after the first observed seizure to comply with our animal ethics approval. Survival curves were analyzed using the Mantel-Cox method (GraphPad Prism). There was no difference in latency to seizure between wild-type littermates of the knock-in and knock-out models (p = 0.4) and these animals were combined for the control.

Drug studies.

Ethosuximide was dissolved in saline to a final concentration of 20 mg/mL delivered via IP injection at 0.01 mL/g. Electrocorticogram recordings were done for 150 minutes prior to injection and recorded for a further 135 minutes following injection. SWDs were counted in a 120-minute epoch prior to injection and compared to a 120-minute epoch following injection. Statistical comparison was made using paired t test (GraphPad Prism).

qPCR.

Total RNA from cortex was extracted with TRizol (Invitrogen, Carlsbad, CA) and purified and concentrated using an RNeasy Kit according to manufacturer's instructions (Qiagen, Hilden, Germany). RNA quality was assessed by spectrophotometry for 260/280 (all were 2.0–2.05) and 260/230 ratios (all above 1.8). A total of 1 µg of total RNA was reverse transcribed to cDNA in a 20-µL reaction using random hexamer primers and a Roche (Branchburg, NJ) High Fidelity Transcription Kit. qPCR of 10 ng cDNA from 4 (knock-out mouse) and 7 (knock-in mouse) littermate pairs was performed using a SYBR Green Master Mix (Fermentas Life Sciences, Vilnius, Lithuania) and primers to exon 8. Exon 8 is deleted in the mutant allele, causing a frameshift that curtails translation after exon 7.⁹ Primers for the exon 8 R allele were as follows: forward 5′ tctctgcccaaggtctccta 3′ and reverse 5′ acaaaataatgcagggtgcc 3', and for GADPH (endogenous control) forward 5′ ggtggtctcctctgacttcaaca 3′ and reverse 5′ gttgctgtagccaaattcgttgt 3′. The R allele in the knock-in mice was assayed using primers forward 5′ gtcatcttaaacaacctgctggaa 3′ and reverse 5′ ccaatgctgttcacatacatatctgt 3′ and a hydrolysis probe 5′ tatgacaacaaacttcgacctgacatagg 3′ to exon 2. The knock-in mutation changes 5 bp in exon 2. The level of R allele transcript in the knock-out heterozygote was calculated relative to that of the wild-type littermate and corrected for amplification efficiencies (Pfaffl, 2001 REST software, Qiagen). Comparison between genotypes was made using the Mann-Whitney nonparametric test (GraphPad Prism).

RESULTS

Haploinsufficiency is the underlying molecular basis of the absence seizure phenotype.

In this study we compare the seizure phenotypes of the GABA_A γ2(R43Q) knock-in mouse model with that of GABA_A γ2^+/− heterozygous knock-out mice,¹⁰ both backcrossed to the same seizure-susceptible DBA/2J strain (DBA).¹¹ The DBA-GABA_A γ2(R43Q) mice display SWD on electrocorticograms associated with behavioral arrest,⁵ consistent with absence seizures seen in patients (figure 1, A and B). Further analyses of electrocorticograms showed that the DBA-GABA_A γ2^+/− knock-out model spent a similar amount of time in SWD seizures as the DBA-GABA_A γ2(R43Q) knock-in mouse (figure 1, A and B). Moreover, SWDs in the GABA_A γ2^+/− were sensitive to the first-line anti–absence seizure drug ethosuximide (200 mg/kg, 2.08 ± 0.4 vs 0.31 ± 0.08, n = 5, p < 0.001), similar to the ∼80% reduction in SWDs reported for the DBA-GABA_A γ2(R43Q) mouse.⁵ These findings suggest that haploinsufficiency is the underlying molecular basis of the absence seizure phenotype in the knock-in as well as the knock-out model.

Figure 1. Two distinct molecular mechanisms underlie the 2 major epilepsy phenotypes caused by the GABA_A γ2(R43Q) mutation.

(A) Electrocorticogram recordings from DBA/J2 GABA_A γ2(R43Q) (DBA-γ2^R/Q) and DBA/J2 GABA_A γ2^+/− (DBA-γ2^+/−) mice illustrating spontaneous spike-and-wave discharges (SWDs). Upward and downward pointing arrows signify onset and offset of behavioral arrest. The asterisk indicates an event on the expanded time scale illustrated below each trace. Scale (x) bars represent 4 s and 1.5 s for top and bottom traces, respectively. (B) Average total time spent in SWD for the wild-type (DBA γ2^R/R [n = 10], DBA-γ2^R/Q [n = 8], and DBA-γ2^+/− [n = 5]) mouse models. *p < 0.005 as compared to the DBA-γ2^R/Q or DBA-γ2^+/− models. (C) Assessment of susceptibility to thermal-induced tonic-clonic seizures (DBA-γ2^R/Q; n = 13, DBA γ2^R/R; n = 20, DBA-γ2^+/−; n = 12). Mantel-Cox method was used for statistical comparison of survival curves (DBA-γ2^R/Q vs DBA γ2^R/R p < 0.05, DBA-γ2^+/− vs WT p = 0.95, DBA-γ2^+/− vs DBA-γ2^R/Q p < 0.05).

Haploinsufficiency cannot account for the febrile seizure phenotype.

Febrile seizures are a feature of the GABA_A γ2(R43Q) family.² To test the thermogenic seizure susceptibility of the mice models, we subjected young mice to a constant 42°C environment and measured the latency to the first tonic-clonic seizure.⁹ In this assay, DBA-GABA_A γ2(R43Q) mice developed seizures with a lower latency than wild-type controls, suggestive of heightened febrile seizure susceptibility as seen in the patients (figure 1C, p < 0.005). Importantly, the time to seizure of the DBA-GABA_A γ2^+/− mouse was indistinguishable from that of wild-type (figure 1C, p = 0.95), demonstrating that although the knock-out model shares the absence seizure phenotype it does not share the thermal seizure phenotype with the DBA-GABA_A γ2(R43Q) knock-in model.

Similar levels of the wild-type allele in the 2 mouse models.

The mRNA levels of the wild-type (R) allele determined by qPCR were the same for both the GABA_A γ2(R43Q) heterozygous knock-in and GABA_A γ2^+/− knock-out models when compared with their wild-type littermates (0.57 ± 0.05 vs 0.59 ± 0.09, n = 7 pairs and 4 pairs, respectively, p = 0.83). This suggests that the major genetic difference between the 2 models is in haploinsufficiency and the missense mutation.

Differential sensitivity of the 2 seizure phenotypes to background genetics.

We have previously demonstrated that SWD expression is sensitive to background genetics by backcrossing the GABA_A γ2(R43Q) knock-in model to 2 mouse strains.^5,7 Here we confirm this by showing that the GABA_A γ2(R43Q) knock-in backcrossed into the C57BL/6J strain (C57) does not have a SWD phenotype (figure 2, A and B). To test the sensitivity of the febrile seizure phenotype to genetic background, we exposed C57-GABA_A γ2(R43Q) and DBA-GABA_A γ2(R43Q) knock-in mice and their wild-type controls to thermogenic stress as described above. Both strains of knock-in mice developed seizures sooner than their wild-type controls (figure 2C, p < 0.005), suggesting that they both have the febrile seizure phenotype.

Figure 2. Spike-and-wave discharges but not thermogenic seizures of the GABA_A γ2(R43Q) mice are sensitive to background genetics consistent with penetrance seen in the family.

(A) Electrocorticogram recordings from GABA_A γ2(R43Q) mice backcrossed into 2 different background strains. Spontaneous spike-and-wave discharges (SWDs) are observed in the DBA-γ2^R/Q knock-in but not the C57BL/6J GABA_A γ2(R43Q) knock-in strain (C57-γ2^R/Q). (B) Summary histogram of SWD events recorded from the DBA-γ2^R/Q (n = 40) and C57-γ2^R/Q (n = 10) strains. (C) Assessment of susceptibility to thermal-induced tonic-clonic seizures between knock-in mice and their wild-type controls (DBA-γ2^R/Q [n = 16], DBA γ2^R/R [n = 10], C57-γ2^R/Q [n = 13], C57-γ2^R/R [n = 12]). Mantel-Cox method was used for statistical comparison of survival curves (DBA-γ2^R/Q vs DBA γ2^R/R p = 0.021, C57-γ2^R/Q vs C57-γ2^R/R p = 0.003, DBA-γ2^R/Q vs C57-γ2^R/Q p = 0.543, DBA γ2^R/R vs C57-γ2^R/R p = 0.641). (D) Table summarizing the seizure phenotype of each mouse model tested.

DISCUSSION

Over the past decade, it has remained an open question as to whether the fundamental molecular deficit of the mutant γ2(R43Q) protein is due to haploinsufficiency or a dominant impact of the mutated protein, or both.⁶ We had assumed that one or another of these molecular mechanisms was the underlying basis of all epilepsy phenotypes seen in this family. In this study, we demonstrate that haploinsufficiency is able to account for the genesis of SWDs but cannot account for the thermal seizure susceptibility. Therefore, thermal seizure susceptibility is governed by a different molecular mechanism, presumably as a consequence of an additional effect of the mutant γ2(R43Q) protein. Importantly, the current study provides a unifying explanation for in vitro data on the γ2(R43Q) mutation.^12–15

There are clues in the literature as to what the downstream cellular mechanisms might be for each phenotype seen in the γ2(R43Q) family. Analysis of the binding of radioactive flumazenil, a competitive antagonist at the benzodiazepine site that is dependent on the GABA_A receptor γ2 subunit, indicates that the knock-in mice, knock-out mice, and the GABA_A γ2(R43Q) family members all demonstrate a similar degree of reduced binding consistent with simple haploinsufficiency in each case.^4,16,17 One possible mechanism of spike-wave genesis in the knock-in and knock-out models may be that single channel conductance of GABA_A receptors is lower. In the GABA_A γ2^+/− mouse there is a significant increase in the proportion of smaller conductance amplitude measured from hippocampal and dorsal root ganglia neurons.¹⁸ Also consistent with this is a reduction in phasic GABA_A receptor-mediated inhibition in the cortex of the GABA_A γ2(R43Q) knock-in mouse model.⁵ In contrast, the febrile seizure phenotype is likely to be caused by an interaction that results in an additional neuronal phenotype unique to the GABA_A γ2(R43Q) knock-in mouse. One interesting example proposed for this mutation is a reduction in the cell surface expression of the GABA_A α5 subunit by the mutated γ2(R43Q) protein.¹³ GABA_A α5–containing subunits are found extrasynaptically with their activation resulting in a tonic GABA-mediated current in the hippocampus.¹⁹ Altered tonic inhibition has been associated with increased neuronal excitability and epilepsy.²⁰ Alternatively, differences in kinetics or altered sensitivity to modulation of any surface receptor containing mutated γ2 subunit may contribute to febrile seizure susceptibility.^12,14 Having a clear idea of the molecular start point for the basis of each seizure type is critical to how we interpret data aimed at understanding cellular mechanisms. Comparing molecular, cellular, and neuronal network deficits of the knock-in with the knock-out model will provide a powerful means of helping to isolate the precise disease mechanism for both seizure phenotypes in future studies.

Our mouse studies show that the absence phenotype can be dependent on genetic background, with knock-in C57BL/6J mice having no SWDs while the knock-in DBA/2J robustly expresses the phenotype⁷ and the febrile seizure phenotype occurs in both background strains. This is consistent with the clinical prediction made in the original family,² where the penetrance of these 2 primary seizure phenotypes is different in family members harboring the mutation, with febrile seizures in ∼65% and absence seizures in ∼20%. The pattern observed in the mouse models suggests that this increased penetrance in patients is, at least in part, due to a differential sensitivity of the 2 seizure phenotypes to genetic background. The genetic basis of SWD susceptibility in animal models is yet to be fully established. Several studies have implicated a region on chromosome 1 termed Szs 1 as the primary basis of the different proconvulsant susceptibility of the C57BL/6(B6) and DBA/2(D2) strains.^21–23 More recently, using congenic mice strains, it was demonstrated that spontaneous SWD seizure susceptibility is sensitive to the Szs 1 locus but that changes at this locus alone cannot account for the phenotype.²⁴ Our study implies that similar genetic principles occur in human populations and that mouse studies may provide the tools to begin to understand the complex heritability of common epilepsies.²⁵

We demonstrate that a single gene mutation can cause distinct seizure types through independent molecular mechanisms and that these seizure types differ in their sensitivity to genetic background. Febrile seizures are not a usual feature of childhood absence epilepsy and the occurrence of febrile seizures in the GABA_A γ2(R43Q) family is likely to be due to molecular mechanisms associated with specific cell biological changes caused by the γ2(R43Q) mutation. These observations now provide a framework to understand the apparently puzzling finding of phenotypic heterogeneity associated with a major dominant gene.

GLOSSARY

SWD

spike-and-wave discharges

AUTHOR CONTRIBUTIONS

Dr. Reid: study concept and design, analysis and interpretation, drafting of manuscript. T. Kim: acquisition of data, analysis and interpretation, figures. Dr. Phillips: acquisition of data. Prof. Berkovic: critical revision of the manuscript for important intellectual content. Prof. Luscher: study concept and design. A/Prof Petrou: study concept and design, analysis and interpretation, drafting of manuscript. All authors were involved in critically reading the manuscript.

STUDY FUNDING

Supported by NHMRC project grant 628520 to C.A.R. and a NHMRC program grant 400121 to S.P. and S.F.B. S.P. acknowledges support from a NMHRC fellowship (1005050) and S.F.B. support from an NHMRC Australian Fellowship (466671). C.A.R. also acknowledges the support of an ARC Future Fellowship (FT0990628). The Angior Foundation supported the purchase of equipment. The Florey Institute of Neuroscience and Mental Health is supported by Victorian State Government infrastructure funds.

DISCLOSURE

The authors report no disclosures relevant to the manuscript. Go to Neurology.org for full disclosures.