Ultra-Rare Genetic Variation in the Epilepsies: A Whole-Exome Sequencing Study of 17,606 Individuals

Abstract

Sequencing-based studies have identified novel risk genes associated with severe epilepsies and revealed an excess of rare deleterious variation in less-severe forms of epilepsy. To identify the shared and distinct ultra-rare genetic risk factors for different types of epilepsies, we performed a whole-exome sequencing (WES) analysis of 9,170 epilepsy-affected individuals and 8,436 controls of European ancestry. We focused on three phenotypic groups: severe developmental and epileptic encephalopathies (DEEs), genetic generalized epilepsy (GGE), and non-acquired focal epilepsy (NAFE). We observed that compared to controls, individuals with any type of epilepsy carried an excess of ultra-rare, deleterious variants in constrained genes and in genes previously associated with epilepsy; we saw the strongest enrichment in individuals with DEEs and the least strong in individuals with NAFE. Moreover, we found that inhibitory GABA_A receptor genes were enriched for missense variants across all three classes of epilepsy, whereas no enrichment was seen in excitatory receptor genes. The larger gene groups for the GABAergic pathway or cation channels also showed a significant mutational burden in DEEs and GGE. Although no single gene surpassed exome-wide significance among individuals with GGE or NAFE, highly constrained genes and genes encoding ion channels were among the lead associations; such genes included CACNA1G, EEF1A2, and GABRG2 for GGE and LGI1, TRIM3, and GABRG2 for NAFE. Our study, the largest epilepsy WES study to date, confirms a convergence in the genetics of severe and less-severe epilepsies associated with ultra-rare coding variation, and it highlights a ubiquitous role for GABAergic inhibition in epilepsy etiology.

Keywords: burden analysis; epilepsy; epileptic encephalopathy; exome; seizures; sequencing.

Figure 1

Burden of Ultra-Rare Singletons in LoF-Intolerant Genes (pLI > 0.995) This analysis was restricted to 4,042 non-ExAC controls for comparison with individuals who have epilepsy. We focused on “ultra-rare” variants not observed in the DiscovEHR database. Significance of association was displayed in false discovery rate (FDR)-adjusted p values; errors bars indicated 95% confidence intervals (CIs) of the corresponding odds ratios. Odds ratios and 95% CIs were not multiplicity adjusted. The five functional coding annotations were defined as described in Table S5. PTV denotes protein-truncating variants; the “damaging missense” and “benign missense” categories were predicted by PolyPhen-2 and SIFT, and “damaging missense-MPC” was a group of missense variants with a missense badness, PolyPhen-2, and regional constraint (MPC) score ≥2. From top to bottom are the results based on all epilepsies: DEEs, GGE, and NAFE. Compared to controls (FDR < 0.05), individuals with epilepsy, except for individuals with NAFE, carried a significant excess of ultra-rare PTV and damaging missense (MPC ≥ 2) variants. PTV burden was higher than missense (MPC ≥ 2) burden across epilepsy types.

Figure 2

Burden of Ultra-Rare Singletons Annotated as Protein-Truncating Variants or Damaging Missense (MPC ≥ 2) Variants “Ultra-rare” variants (URVs) were defined as not observed in the DiscovEHR database. Gene sets were defined in Table S6 and the number of genes was specified in the parenthesis. DEE stands for individuals with developmental and epileptic encephalopathies, GGE for genetic generalized epilepsy, NAFE for non-acquired focal epilepsy, and EPI for all epilepsy; NDD-EPI genes are genes with de novo burden in neurodevelopmental disorders with epilepsy. Asterisks indicate significance after FDR control (^∗ for FDR-adjusted p value < 0.05, ^∗∗ for adjusted p value < 1 × 10⁻³, and ^∗∗∗ for adjusted p value < 1 × 10⁻⁵). Effects were displayed in odds ratios with the corresponding 95% confidence intervals (CIs). (A) Protein-truncating variants (PTVs) were enriched in candidate epilepsy-associated genes for individuals with DEEs relative to other epilepsy subgroups but did not show a strong signal in inhibitory, excitatory receptors, or voltage-gated cation channel genes. (B) The burden of damaging missense (MPC ≥ 2) variants, on the other hand, was stronger across these gene sets than was that of PTVs, especially for GABA_A receptor genes and genes involved in GABAergic pathways. Relative to other epilepsy types, individuals with NAFE consistently showed the least burden of deleterious URVs. No enrichment was observed from excitatory receptors.

Figure 3

Gene Burden for Individuals Diagnosed with Developmental and Epileptic Encephalopathies, Genetic Generalized Epilepsy, or Non-Acquired Focal Epilepsy This analysis focused on ultra-rare (non-DiscovEHR) singleton variants annotated as protein-truncating variants (PTVs), damaging missense variants (MPC ≥ 2), or in-frame insertions and deletions and used Fisher’s exact test (FET) to identify genes with a differential carrier rate of these ultra-rare deleterious variants in individuals with epilepsy in comparison to controls. Exome-wide significance was defined by a p value < 6.8 × 10⁻⁷ after Bonferroni correction (see Materials and Methods). Only SCN1A achieved exome-wide significance for individuals with developmental and epileptic encephalopathies (DEEs). Panels refer to individuals diagnosed with (A) developmental and epileptic encephalopathies, (B) genetic generalized epilepsy, and (C) non-acquired focal epilepsy.