Novel GABRA2 variants in epileptic encephalopathy and intellectual disability with seizures

Sir, We read with great interest the article by Butler et al. (2018) reporting patients with de novo variants in genes encoding GABAA receptors. In particular, they described a patient with severe early onset epileptic encephalopathy (EOEE) who had multiple and polymorphic seizures starting from the age of 6 weeks. The course of the disorder was marked by severe developmental delay, hypotonia, microcephaly, cortical visual impairment and profound intellectual disability. This patient had a de novo NM_000807.2: c.875C>A p.(Thr292Lys) missense variant in GABRA2, encoding the GABAA receptor (GABAA-R) alpha-2 subunit. Functional study showed a decrease of the expression of the mutant receptor at the cell surface and an alteration of the gating of the channel that was trapped in an open state (Butler et al., 2018). Previously, a single other patient with de novo c.1003A>C p.(Asn335His) GABRA2 variant had been reported, without functional characterization (Orenstein et al., 2018). This patient also had severe EOEE associated with poor cognitive development. Because only two patients have been reported so far, the clinical spectrum and the functional impact of GABRA2 variants remain to be fully defined. GABA is the major inhibitory neurotransmitter in the vertebrate brain and mediates neuronal inhibition by binding to the GABA receptors. GABRA2 is one of 19 genes encoding GABAA-R subunits. GABAA-Rs belong to the Cys loop ligand-gated ion channel superfamily and are generally composed of five different subunits combined from the selection of six α subunits, three β subunits, three γ subunits, three ρ subunits, and one ɛ, δ, θ and π subunit. GABAA-R including α2 subunits (principally α2βy2, or α2βy1) represent ∼25% of the GABAA-Rs in the rat brain (McKernan and Whiting, 1996; Whiting et al., 1999; Olsen and Sieghart, 2009). Variants responsible for epileptic encephalopathies have previously been identified in eight of the 19 genes encoding GABAA-R: GABRA1, GABRA3, GABRB1, GABRB2, GABRB3, GABRG2, and more recently, GABRA2 and GABRA5 (Baulac et al., 2001; Wallace et al., 2001; Epi4K Consortium et al., 2013; Carvill et al., 2014; Srivastava et al., 2014; Epi4K Consortium, 2016; Lien et al., 2016; Ishii et al., 2017; Møller et al., 2017; Niturad et al., 2017; Butler et al., 2018; Orenstein et al., 2018). In this letter, we describe the genetic and clinical data of five patients, including three sporadic cases and two affected siblings, with four novel de novo GABRA2 missense variants, and the functional characterization of these variants. Whole-exome sequencing was performed for all four families in diagnostic or research laboratories. GABRA2 variants were classified as pathogenic or likely pathogenic according to ACMG criteria (Richards et al., 2015). Data collection of several individuals with GABRA2 variants was enabled using GeneMatcher (Sobreira et al., 2015). The study was approved by the ethics committees of the respective institutions and biological samples were obtained following written informed consent from the legal representatives. Patients 1–3 presented with similarly severe EOEE phenotypes, with seizure onset ranging from 1 day to 1 month of life (Table 1). Seizure types included clonic seizures (n = 3), ocular revulsions (n = 2), absences (n = 1) and spasms (n = 1). All patients had multiple seizures (10 to 20) per day that were resistant to anti-epileptic treatments. The three patients had status epilepticus. All three patients showed hypotonia, poor eye contact and severe developmental delay, illustrated by the inability to walk or to speak at 2.5 years (Patient 2) and 5 years (Patient 3). Patient 1 was still very young (5 months) at the time of the study but she presented with neonatal hypotonia, failure to thrive and pyramidal syndrome. Exome sequencing respectively identified three novel de novo variants NM_000807.2 c.851T>C p.(Val284Ala), c.871C>G p.(Leu291Val), c.788T>C p.(Met263Thr) in GABRA2. Clinical and genetic summary of the patients with GABRA2 variants Antiepileptic treatments (AETs): CBZ = carbamazepine, CLB = clobazam, CLN = clonazepam, GVG = vigabatrin, LTG = lamotrigine, OXC = oxcarbamazepine, PB = phenobarbital, PHT = phenytoin, TPM = topiramate, VPA = sodium valproate. ASD = autism spectrum disorders; CC = corpus callosum; d = day; F = female; M = male; m = months; N = normal; NA = not available; w = weeks; y = years. aPatient 4’s sister. Clinical and genetic summary of the patients with GABRA2 variants Antiepileptic treatments (AETs): CBZ = carbamazepine, CLB = clobazam, CLN = clonazepam, GVG = vigabatrin, LTG = lamotrigine, OXC = oxcarbamazepine, PB = phenobarbital, PHT = phenytoin, TPM = topiramate, VPA = sodium valproate. ASD = autism spectrum disorders; CC = corpus callosum; d = day; F = female; M = male; m = months; N = normal; NA = not available; w = weeks; y = years. aPatient 4’s sister. Patient 4 showed language delay without motor delay and was diagnosed with autism spectrum disorder at 18 months of age and presented a regression of acquisitions at 2.5 years. He had moderate intellectual disability, without other neurological symptoms. At 17 years, he exhibited generalized tonic-clonic seizures that were treated by lamotrigine. His sister also had mild/moderate intellectual disability with seizures beginning at the age of 2 years, but without autistic features. Both patients had the heterozygous GABRA2 c.975C>A p.(Phe325Leu) variant. Analysis of parental samples showed that the variant was also present at a low level in the blood of the father, indicating paternal mosaicism. All four variants are absent from GnomAD, predicted damaging by PolyPhen-2 and have a CADD score between 23.9 and 26.2. They alter amino acids highly conserved during evolution. Two variants [p.(Val284Ala) and p.(Leu291Val)] are located in the second transmembrane domain of the protein aligning the pore of the receptor while the other two are located in the first p.(Met263Thr) and third p.(Phe325Leu) transmembrane domain (Fig. 1A and B). Novel GABRA2 variants in epileptic encephalopathy and intellectual disability with seizures. (A) GABRA2 cDNA structure showing positions of exon boundaries and four newly identified variants p.(Met263Thr), p.(Val284Ala), p.(Leu291Val), and p.(Phe325Leu) (top). Two previously reported variants p.(Thr292Lys) and p.(Asn335His) are also depicted. Protein structure of the α2 subunit including positions of the signal peptide, extracellular N-terminus, transmembrane domains M1-M4 and intracellular M3-M4 domain (bottom). (B) Schematic illustrating the putative structure of the α2 subunit containing all so far identified disease variants, one each in M1 and M3 segment, three in the M2 segment and one in the intracellular M3-M4 loop. A predicted structure of the pentameric GABAA receptor with M2 segments forming the channel pore is shown on the right. (C) Average normalized GABA-induced currents recorded from Xenopus laevis oocytes expressing GABRA2 wild-type (WT) and different variants upon application of 1 mM GABA. All current amplitudes were normalized to the mean amplitude of the wild-type recorded on the same day and pooled from different experiments. Data are presented as box-and-whisker plots, indicating the median, interquartile ranges, and minimum and maximum values. Plus symbol indicates mean values which are as follows: wild-type 1.00 (n = 169), Met263Thr 0.33 (n = 45), Val284Ala 0.66 (n = 27), Leu291Val 0.05 (n = 45), Phe325Leu 0.60 (n = 69); *P < 0.05, ****P < 0.0001. (D) Dose-response curves were obtained from the responses to application of different GABA concentrations (in µM: 1, 3, 10, 30, 100, 300 and 1000). EC50 values in µM: wild-type 18.8 ± 1.4 (n = 9), Met263Thr 7.2 ± 2.6 (n = 8), Val284Ala 9 ± 3 (n = 9), Leu291Val 67 ± 24 (n = 10), Phe325Leu 14.1 ± 2.5 (n = 69); P < 0.05 for wild-type versus Met263Thr. Hill coefficient values: wild-type 1.48 ± 0.14, Met263Thr 0.96 ± 0.17, Val284Ala 1.20 ± 0.10, Leu291Val 1.03 ± 0.06 (n = 45), Phe325Leu 0.90 ± 0.10 (n = 69); P < 0.05 for wild-type versus Met263Thr and wild-type versus Phe325Leu. Significance was tested using Kruskal-Wallis test with Dunn’s multiple comparison test. To investigate their functional impact, we expressed the four variants in Xenopus laevis oocytes and recorded GABA-induced currents using automated and manual two microelectrode voltage clamp technique as described previously (Forster et al., 2001; Li et al., 2017; Niturad et al., 2017). To this end, cRNA encoding wild-type or mutant α2 subunits was co-injected with β3 and γ2 cRNAs in a 1:1:2 ratio. The application of 1 mM GABA for 15 s was used to assess the maximal current response at the holding potential of −70 mV. Each current was normalized to the mean value of the wild-type current recorded on the same day so that the data from different experiments could be pooled. All four variants caused a significant reduction in current amplitudes compared to the wild-type recorded on the same day and under the same conditions (Fig. 1C). Application of increasing GABA concentrations (in µM: 1, 3, 10, 30, 100, 300 and 1000) was used to obtain dose-response curves (Fig. 1D). The current response to each GABA concentration was normalized to the maximum response evoked by 1 mM GABA. The normalized GABA responses of each cell were fitted with the four-parameter logistic equation to estimate the EC50 value, representing the concentration of the agonist at which half of the maximum response is achieved, and the Hill coefficient, which reflects the steepness of the dose response curve. Only p.(Leu291Val) showed a significant change in EC50. Even though three variants appeared to have increased sensitivity to GABA concentrations ≤10 µM compared with the wild-type, after correction for the overall current reduction measured for each variant (Fig. 1C), the difference was predicted to be significant only for the p.(Val284Ala) at concentrations of 3 µM (P < 0.01) and 10 µM (P < 0.05) (data not shown). In addition, for two of the variants, p.(Met263Thr) and p.(Phe325Leu), we found significantly reduced Hill coefficient values. These data show that all four variants caused a loss of function, with the most severe impact seen for the p.(Leu291Val) variant. In addition to diminishing GABA-induced currents, p.(Met263Thr) and p.(Phe325Leu) variants may reduce cooperative binding of the ligand. Lastly, p.(Val284Ala), showed the least prominent impact on the current amplitudes but also an increased response to low GABA concentrations. This suggests that different molecular pathological mechanisms may contribute to the observed clinical outcomes associated with these variants. Altogether, we identified four heterozygous missense variants in GABRA2 and provided a functional screen that yielded results compatible with a loss of function mechanism. Three variants occurred de novo in three unrelated families while the fourth variant, present in two affected siblings, was inherited from their mosaic father. Genes encoding GABAA-R have been associated with a broader spectrum of epileptic syndromes and rare variants in genes encoding GABA-R can either be associated with monogenic epileptic syndromes or constitute risk factors for the genetically complex generalized epilepsies (May et al., 2018). Variants in GABRA3, GABRB1 and GABRA5 have been only associated with severe phenotypes so far (Lien et al., 2016; Niturad et al., 2017; Butler et al., 2018), whereas variants in most other genes including GABRA1, GABRB2, GABRB3 and GABRG2 are associated with a broader clinical spectrum, ranging from EOEE, moderate/mild intellectual disability with epilepsy, to benign familial epilepsy without intellectual disability as illustrated by variants in GABRG2 and GABRA1 (Baulac et al., 2001; Wallace et al., 2001; Boillot et al., 2015; Epi4K Consortium, 2016; Johannesen et al., 2016; Kang and Macdonald, 2016; Hamdan et al., 2017; Zou et al., 2017; Heyne et al., 2018). Variants in GABRA2 have been associated with a severe epileptic encephalopathy phenotype in five of six unrelated patients described to date. However, in this study, Patients 4 and 5 have a less severe phenotype, with pharmacoresponsive epilepsy, autism spectrum disorder and moderate intellectual disability while all other patients show severe to profound intellectual disability. The variant itself could define the phenotype; however, other genetic and/or environmental factors may also contribute to disorder severity. Our study therefore confirms that de novo GABRA2 variants can lead to severe forms of EE and expands the clinical spectrum to patients with mild/moderate intellectual disability, autistic disorders, and late-onset epilepsy, as previously described for genes encoding others GABAA receptor subunits and others genes encoding ion channels, such as SCN1A, SCN2A or HCN1 (Gambardella and Marini, 2009; Wolff et al., 2017; Marini et al., 2018). The data that support the findings of this study are available from the corresponding author, upon reasonable request. We thank the patients and their family for their participation in this study. We thank Claude Estrade, Sabina Karagic, Elodie Lejeune, Corinne Mach and Valérie Olin for the technical assistance. The research generating these results was funded by the Agence National de la Recherche (ANR EUHFAUTISM), the “Investissements d’avenir” program ANR-10-IAIHU-06 (IHU-A-ICM), the Bio-Psy labex, INSERM, the Assistance Publique des hôpitaux de Paris (AP-HP), the National Health and Medical Research Council of Australia (Program grant 10915693), and the Pierre and Marie Curie University.