The Molecular Biology of Genetic-Based Epilepsies

Epilepsy is one of the most common neurological disorders characterized by abnormal electrical activity in the central nervous system. The clinical features of this disorder are recurrent seizures, difference in age onset, type, and frequency, leading to motor, sensory, cognitive, psychic, or autonomic disturbances. Since the discovery of the first monogenic gene mutation in 1995, it is proposed that genetic factor plays an important role in the mechanism of epilepsy. Genes discovered in idiopathic epilepsies encode for ion channel or neurotransmitter receptor proteins, whereas syndromes with epilepsy as a main feature are caused by genes that are involved in functions such as cortical development, mitochondrial function, and cell metabolism. The identification of these monogenic epilepsy-causing genes provides new insight into the pathogenesis of epilepsies. Although most of the identified gene mutations present a monogenic inheritance, most of idiopathic epilepsies are complex genetic diseases exhibiting a polygenic or oligogenic inheritance. This article reviews recent genetic and molecular progresses in exploring the pathogenesis of epilepsy, with special emphasis on monogenic epilepsy-causing genes, including voltage-gated channels (Na+, K+, Ca2+, Cl−, and HCN), ligand-gated channels (nicotinic acetylcholine and GABAA receptors), non-ion channel genes as well as the mitochondrial DNA genes. These progresses have improved our understanding of the complex neurological disorder.


SCN1A
The sodium channel neuronal type I alpha subunit gene SCN2A The sodium channel voltage-gated type II alpha subunit gene SCN3A The sodium channel voltage-gated type III alpha subunit gene SCN8A The sodium channel voltage-gated type VIII alpha subunit gene SCN9A The sodium channel voltage-gated type IX alpha subunit gene SCN1B The sodium channel voltage-gated type I beta subunit gene KCNQ2 The potassium channel voltage-gated KQT-like subfamily member 2 gene KCNQ3 The potassium channel voltage-gated KQT-like subfamily member 3 gene KCNA1 The potassium channel voltage-gated shakerrelated subfamily member 1 gene KCTD7 The potassium channel tetramerization domaincontaining protein 7 gene CACNA1H The calcium channel voltage-dependent T type alpha-1H subunit gene CLCN2 The chloride channel 2 gene HCN The hyperpolarization-activated cyclic nucleotide-gated potassium channel gene GABRA1 The gamma-aminobutyric acid receptor alpha-1 gene

GABRB3
The gamma-aminobutyric acid receptor beta-3 gene GABRD The gamma-aminobutyric acid receptor delta gene GABRG2 The gamma-aminobutyric acid receptor gamma-2 gene CHRNA4 The cholinergic receptor neuronal nicotinic alpha polypeptide 4 gene CHRNB2 The cholinergic receptor neuronal nicotinic beta polypeptide 2 gene LGI1 The leucine-rich glioma-inactivated 1 gene MASS1 The monogenic audiogenic seizure-susceptible 1 gene SLC2A1 The solute carrier family 2 member 1 gene GLUT1 Glucose transporter 1 EFHC1 The ef-hand domain (C-terminal)-containing protein 1 gene PRRT2 The proline-rich transmembrane protein 2 gene ALDH7A1 The aldehyde dehydrogenase 7 family member a1 gene POLG The polymerase DNA gamma gene MTTL1 The transfer RNA mitochondrial leucine 1 gene MTTK

Introduction
Epilepsy is one of the most common neurological disorders, characterized by abnormal electrical activity in the central nervous system (CNS). Recurrent seizures are a cardinal clinical manifestation. The phenotypic expression of each seizure is determined by the original point of the hyperexcitability and its degree of spread in the brain [1]. This disorder affects up to 3 % of the world's population, and many factors are thought to contribute to its pathogenesis, such as trauma, virus infection, acquired structural brain damage, altered metabolic states, inborn brain malformations, etc. [2]. However, about 1 % of cases have no obvious reasons and without any other neurological abnormalities, named "idiopathic epilepsies," and usually presumed to be genetic, which are often associated with mutations in ion channel subunits. In recent years, almost all the identified epilepsy monogenic disease-causing genes encode ion channel subunits (voltage-gated ion channels and ligandgated ion channels), which led to the concept that the idiopathic epilepsies are a family of channelopathies. However, the majority of common genetic epilepsies, like juvenile myoclonic epilepsy (JME) or childhood and juvenile absence (accounting for almost 30 to 40 % of all epilepsies), have a complex inheritance in which different genes with only a small effect on the risk for epilepsy, possibly in combination with environmental factors [3,4]. In this article, we summarized the known genes responsible for different epilepsy syndromes and the role these genes play in epilepsy (Tables 1 and 2); we also reviewed the recent progress made in the molecular genetics of epilepsy, especially monogenic epilepsy-causing genes including voltage-gated channels (Na + , K + , Ca 2+ , Cl − , and HCN) and ligand-gated channels (nicotinic acetylcholine and GABA A receptors). Through this article, we hope to provide some new insights on better understanding of the pathogenesis of this complicated disorder.

Na + Channel Genes
The principal role of Na + channels is for the initiation and propagation of action potentials in the CNS and peripheral nervous system (PNS). Nine genes encode the pore-forming α subunit and four genes encode the ancillary β subunits. The basic architecture of the Na + channels is a four-domain protein with its 24 transmembrane segments, which is conserved in all family members, and the amino acid sequence identity varies from 50 to 85 % between channels [5]. Both α and β subunit gene mutations have been reported to be the causes of epilepsy.
The Sodium Channel Neuronal Type I Alpha Subunit Gene (SCN1A, MIM 182389) The SCN1A gene encodes the Na v 1.1 subunit, which forms a fast inactivating voltage-dependent Na + channel and plays a critical role in the control of action potential generation and propagation. This subunit always associates with one or two β subunits. Mutations in it are the most common genetic causes of familial and sporadic epilepsy, all of which are dominant inheritance [6]. The first SCN1A mutation was identified in families with epilepsy with febrile seizures plus (GEFS+), followed by the finding in sporadic patients with severe myoclonic epilepsy of infancy (SMEI) [7,8]. Over 650 mutations in the SCN1A gene have been identified in patients with GEFS+ and SMEI [5]. But there has been no mutation "hotspot," and the genotype-phenotype relationships are complex [9,10]. Half of these mutations are truncation mutations, showing haploinsufficiency in the SCN1A-caused epilepsy. The rest are missense mutations, most of which known to date are associated with GEFS+ causing either gain of function or loss of function. Both missense and nonsense mutations were observed in SMEI [5,6]. However, dominant monogenic inheritance makes up only a small part of GEFS+ and most of the disorders appear to have complex inheritance. The presence of marked phenotypic variation in some of these dominant families suggests that modifier genes, which are yet to be identified, also play a significant role. Beyond the natural impacts of the mutations (gain or loss of function), genetic compensation, cell biological effects, and others all contribute to the overall consequence on networks, and seizure genesis may result in the development of varied clinical phenotypes of epilepsy. Despite the identification of so much mutation in SCN1A, there are about 20-30 % SMEI whose etiology remains unknown. Maybe they can be explained by the duplications and deletions in SCN1A [11]. In the Scn1a knockout mice, although the Na + current was unchanged in hippocampal excitatory pyramidal neurons, it was reduced in inhibitory interneurons [6,12,13]. It is possible that a loss and gain of function can result in an excitable phenotype, via impact on different neuron types. It remains to be seen if this is true for more subtle loss-and gain-of-function mutations that result in the mild type of epilepsy-GEFS+ phenotype.  The Sodium Channel Voltage-Gated Type II Alpha Subunit Gene (SCN2A, MIM 182390) The SCN2A gene encodes the Na v 1.2 subunit, which is important for action potential initiation, propagation, and the generation of repetitive firing. Although it is abundant in the brain, Nav1.2-related epilepsy is rare, suggesting that it may be only weakly associated with neuronal hyperexcitability [14]. The first mutation (Arg118Trp) in the SCN2A gene was reported in a patient with GEFS+, followed by the identification of eight different mutations in patients with benign familial neonatalinfantile seizures (BFNISs). BFNIS, firstly described by Kaplan and Lacey in 1983, is an autosomal dominant disorder presenting between day 2 and 7 months with afebrile secondarily generalized partial seizures [15][16][17]. To date, more than 20 mutations in the SCN2A gene have been identified in patients with BFNIS, GEFS+, and SMEI, the majority of which are missense mutations and most frequently associated with BFNIS. While both nonsense (Arg102X) and missense (Glu1211Lys, Ile1473Thr, Arg1312Thr) mutations cause SMEI [5,10,[18][19][20], Arg1319Gln mutation appeared to be a mutation "hotspot" confirmed by haloptype analysis for BFNISs [21]. The electrophysiologic analysis revealed that Glu1211Lys and Ile1473Thr mutations resulted in hyperpolarizing shifts in voltage dependence of activation, consistent with premature channel opening and hyperactivity [20]. However, the Arg1319Gln mutation results in a depolarizing shift in voltage dependence of activity, consistent with reduced activity, indicating that the different biophysical effect severity of the SCN2A mutations is correlated with clinical severity, suggesting a dominant negative effect or loss-of-function mechanism, while the identification of novel SCN2A mutations in benign familial neonatal seizures (BFNS), a rare autosomal dominant disorder characterized with the onset of partial or generalized clonic convulsions occurring around 3 days of age and remitting within 3 months, indicated that SCN2A is not only a specific gene for BFNIS but also for families with a delayed age of onset [22]. The heterozygous Scn2a knockout mice showed no seizures, while double heterozygous mice with Scn2a and Kcnq2 displayed severe myoclonic seizures, suggesting that an interaction between genes might contribute to the variable expressivity [23, 24].
The Sodium Channel Voltage-Gated Type Ш Alpha Subunit Gene (SCN3A, MIM 182391) The SCN3A gene encodes Na v 1.3 subunit, which is widely expressed in adult brain. The voltage dependence and kinetics of activation of SCN3A channel was found to be similar to that of SCN2A, but it was inactivated at more hyperpolarized potentials and was slower to recover from inactivation than SCN2A channel [25]. Unlike SCN1A and SCN2A, only a Lys354Gln coding variant has been reported in the SCN3A gene in a single patient with cryptogenic pediatric partial epilepsy [26]. The mutant residue 354, located in the pore region of domain 1 of Na v 1.3, is an evolutionary conservation in all human sodium channel genes except SCN1A. The Lys354Gln variant enhances the persistent and ramp currents of Nav1.3, reduces current threshold, and produces spontaneous firing and paroxysmal depolarizing shift-like complexes in hippocampal neurons, providing robust support for the concept that Na v 1.3 mutations can lead to epilepsy in postnatal humans [27]. Unlike the expression in human brain, Scn3a expression in rodent is highest in young animals but low in adults [28].
The Sodium Channel Voltage-Gated Type VIII Alpha Subunit Gene (SCN8A, MIM 600702) The SCN8A gene encodes the Na v 1.6 subunit, which is composed of four homologous domains (D1 to D4) concentrating at the axon initial segment and nodes of Ranvier. It is widely expressed in neurons of the CNS and PNS, regulating firing patterns of excitatory and inhibitory neurons in CNS. Recently, a de novo heterozygous SCN8A Asn1768Asp mutation was found in a 15-year-old female patient, which alters an evolutionarily conserved residue in Na v 1.6, leading to a large increase in ramp and persistent currents and incomplete channel inactivation, consistent with a dominant gain-offunction phenotype [29]. An animal study revealed that Scn8a can function as a genetic modifier of SMEI by restoring normal seizure thresholds and improving survival [30]. Heterozygous Scn8a V929F/+ , Scn8a med/+ , and Scn8a med-jo/+ (med: motor endplate disease; jo: jolting) mice displayed spike-wave discharges, a character of absence epilepsy, indicating that the SCN8A gene may be involved in common human absence epilepsy [31].
The Sodium Channel Voltage-Gated type IX Alpha Subunit Gene (SCN9A, MIM 603415) The SCN9A gene encodes the Na v 1.7 subunit, expressing primarily in neurons of the dorsal root ganglia and has been classified as a PNS channel. Mutations in this gene were previously found to be associated with three inherited disorders: autosomal dominant primary erythermalgia (PE), paroxysmal extreme pain disorder (PEPD), and autosomal recessive channelopathy associated insensitivity to pain (CIP). A missense Asn641Tyr mutation in the SCN9A gene was identified in a large Utah family with febrile seizures (FS). Extended analysis showed that the SCN9A gene mutations were detected in 8 % (9/109) patients with SMEI and two thirds (6/9) of SCN9A patients also harbored the SCN1A mutations, though how these two mutations affect phenotype of the patients was not yet clear. Homozygous Scn9a N641Y/N641Y knock-in mice displayed reduced thresholds to electrically stimulated seizure [32].
The Sodium Channel Voltage-Gated Type I Beta Subunit Gene (SCN1B, MIM 600235) The SCN1B gene encodes the β1 ancillary subunit, with high abundance in several brain regions but unclear precise subcellular location, which is a single transmembrane domain glycoprotein composed of a large N-terminal extracellular and a short C-terminal intracellular domain. Na + channel β subunits are multifunctional, modulating channel gating, regulating the level of channel expression, and potentially acting as a cell adhesion molecule [10,33]. In 1998, a heterozygous SCN1B Cys121Trp mutation was reported in an Australian family with GEFS+, and the mutation changes a conserved cysteine residue disrupting a putative disulfide bridge, resulting in damaging a normal extracellular immunoglobulin-like fold [34]. After that, several mutations were identified in the SCN1B gene [35][36][37][38][39]. In 2003, a heterozygous Ile70_Glu74del mutation was identified in a family with FS and early-onset absence epilepsy, resulting in a deletion of five amino acids in the same domain of the Cys121Trp mutation and potential a loss of function. Both Cys121Trp and Ile70_Glu74del mutations may cause a persistent inward Na + current though they reduce the inactivation rate of the voltage-gated Na + channels, which may result in hyperexcitability [38]. A homozygous SCN1B Arg125Cys mutation, which prevents trafficking of the β1 subunit to the cell surface, was identified in a patient with SMEI, inferring that the SCN1B gene is an autosomal recessive cause of SMEI by functional gene inactivation [40]. Homozygous Scn1b knockout mice also exhibit spontaneous seizures [41], supporting the role of β1 subunits in epilepsy.

K + Channel Genes
The human genome encodes about 100 K + channel subunits, which are expressed mostly in the brain, especially ubiquitous in neuronal and glial cell membranes. These channels can be classified into three structural families (2TM, twotransmembrane; two pore, four-transmembrane; voltagegated, six-transmembrane), depending on the number of transmembrane domains in each subunit. The voltage-dependent K + channels contain six transmembrane domains and include several subfamilies such as KCNQ and K v 1 channels in which epilepsy-causing mutations have been reported [10,42]. The KCNQ2 and KCNQ3 genes belong to a subfamily of K + channels genes and significantly expressed in the brain, mainly in the hippocampus, temporal cortex, cerebellar cortex, and medulla oblongata from late fetal life to early infancy, consistent with the time in which BFNS occurs. Their proteins produce the M current, which is a noninactivating voltagegated K + current and mediating the medium after hyperpolarization conductance. The M current is responsible for stabilizing the resting membrane potential and important in regulating neuronal excitability; thus, its loss of function may be expected to lead to an increase of neuronal excitability [47][48][49]. Since the mutations in KCNQ2 and KCNQ3 genes were identified to be associated with BFNS in 1998, more than 50 mutations in KCNQ2 and KCNQ3 have been identified in BFNS, the majority of which are KCNQ2 mutations located mostly in the C-terminus (57 %) of the gene [10,44,[50][51][52]. Mice with mutations in Kcnq2 and Kcnq3 show spontaneous seizures [53] or low seizure thresholds leading to increased seizure susceptibility [54,55], proving a further understanding of how mutations in M current lead to epilepsy.

The Potassium Channel Voltage-Gated Shaker-Related Subfamily Member 1 Gene (KCNA1, MIM 176260)
The KCNA1 gene encodes the K v 1.1 subunit which concentrates on the axonal membrane and presynaptic nerve terminals where it contributes to repolarize and shape action potentials. Most of the KCNA1 gene mutations cause episodic ataxia type 1 (EA1), a neuronal channelopathy, presenting with brief episodes of cerebellar dysfunction and persistent neuromyotonia [56]. In a Scottish family with EA1 harboring a KCNA1 mutation, five patients were found to have the mutation including two with partial epilepsy, indicating that KCNA1 cannot be regarded as a gene of major effect in causing epilepsy, but likely to be a risk factor for seizures [57,58]. Mice with Kcna1-null mutation displayed frequent spontaneous seizures which correlated on the cellular level with alterations in hippocampal excitability and nerve conduction and indicated that a loss of function with K v 1.1 results in an increased excitability that perhaps contributes to an epileptic phenotype [59].
The Potassium Channel Tetramerization Domain-Containing Protein 7 Gene (KCTD7 , MIM 611725) The KCTD7 gene encodes a protein containing an Nterminal domain that is homologous to the T1 domain of voltage-gated potassium channel proteins. The T1 domain is a tetramerization domain on the cytoplasmic side of the subunits. It is involved in the modulation of voltage gating of voltage-gated potassium channels. Van Bogaert et al. reported a homozygous KCTD7 Arg99X mutation in a large consanguineous Moroccan family with progressive myoclonic epilepsy (PME), which presents as a heterogeneous group of disorders with different genetic inheritance, including Univerricht-Lundborg's disease (EPM1), Lafora disease (EPM2), neuronal ceroid lipofuscinosis and storage diseases, dentatorubral-pallidoluysian atrophy (DRPLA), and myoclonic epilepsy with ragged red fibers (MERRF) [60,61]. Despite its broad spectrum of manifestations, PME shares some common clinical findings, such as myoclonic seizures and progressive neurological dysfunction, particularly ataxia and dementia. To date, at least eight KCTD7 mutations have been identified in patients with PME [62,63]. Its molecular function, however, remains unknown. Current research shows that KCTD7 overexpression hyperpolarizes the cell membrance and reduces the excitability of transfected neurons in in vitro patch clamp experiments [64]. However, the precise molecular mechanism of KCTD7 affecting the membrane potential remains to be elucidated.

Ca 2+ Channel Genes
The importance of Ca 2+ channel genes has been demonstrated in several murine models of generalized epilepsy involving absence epilepsy with ataxia. Voltage-gated calcium channels can be classified into high voltage-activated (HVA) and low voltage-activated (LVA) channels. HVA channels can be further subdivided into L-, N-, P-, Q-, and R-type, while LVA channels (also referred to as T-type) are thought to be α1 subunit monomers [65]. HVA channel family members are heteromultimers of a pore-forming α1 subunit that coassembles with ancillary β, α2, δ, or a γ subunit into a functional channel complex. The channel subtype was defined by the α1 subunit, and the ancillary subunits regulate α1 subunit function and surface expression. Patients with mutations in the Ca v 2.1 α1 subunit (CACNA1A) can present with absence seizures, and mutations and loss-of-function mutations in ancillary HVA calcium channel subunits also result in IGE phenotypes in mice. Gain-offunction mutations in Ca v 3.2 (an LVA or T-type calcium channel encoded by the CACNA1H gene) result in congenital forms of IGEs [66].
The Calcium Channel Voltage-Dependent T Type Alpha-1H Subunit Gene (CACNA1H, MIM 607904) In 2004, Chen et al. first reported 12 rare heterozygous missense mutations associated with childhood absence epilepsy (CAE) in Han ethnicity [67]. To date, over 30 mutations in the CACNA1H gene, encoding the Ca v 3.2 subunit, have been found in IGE patients [10]. A functional study in HEK293 cells revealed that Phe161Leu and Glu282Lys mutations mediated an~10-mV hyperpolarizing shift in the half-activation potential, while V831M mutation caused a~50 % slowing of inactivation and shifted half-inactivation potential~10 mV towards more depolarized potentials [68]. Heron et al. discovered three missense mutations (Pro618Leu, Gly755Asp, and Ala480Thr) in patients with idiopathic generalized epilepsies (IGE) or GEFS+, while the Pro618Leu and Gly755Asp mutations exhibited functional changes that were consistent with increased channel function [69,70]. An extended study suggested that the CACNA1H variants may contribute to susceptibility, but not sufficient to epilepsies including childhood absence, juvenile absence, juvenile myoclonic and myoclonic astatic epilepsies, febrile seizures, and temporal lobe epilepsy [66,71].
The Chloride Channel 2 Gene (CLCN2, MIM 600570) The CLCN2 gene encodes a voltage-gated chloride channel protein, which is a ubiquitously expressed chloride channel with 18 transmembrane alpha helical segments. It is expressed widely in the brain and plays an important role in maintaining the low intracellular Cl − concentration, which is essential for GABA-mediated inhibition [72]. In 2000, genomewide search identified a susceptibility locus for common IGE subtypes on chromosome 3q26 [73]. Subsequently, Haug et al. found the CLCN2 mutations responsible for four of the most common IGE subtypes including childhood and juvenile absence epilepsy (CAE and JAE), juvenile myoclonic epilepsy (JME), and epilepsy with grand mal seizures on awakening [74]. M200fsX231 and del74-117 mutation found by Haug et al. causes a loss of function, expecting to lower the transmembrane chloride gradient essential for GABAergic inhibition, while Gly715Glu mutation alters voltage-dependent gating, causing membrane depolarization and hyperexcitability. Heterozygous variants in the CLCN2 gene were reported to be responsible for idiopathic generalized epilepsies (IGE), yet subsequent studies showed that the involvement of CLCN2 in epilepsy remains controversial, and these variants may represent innocuous mutations by functional analysis [74][75][76]. Clc-2 −/+ knockout mice had no overt behavioral or morphological consequence, and Clc-2 −/− mice had no evidence of spontaneous seizures or increased susceptibility to flurothyl-induced seizures [76]. Further studies will help reveal whether this gene could contribute to a greater susceptibility to IGE.
The Hyperpolarization-Activated Cyclic Nucleotide-Gated Potassium Channel Gene (HCN) Hyperpolarization-activated cyclic nucleotide-gated potassium channel genes (HCN1, HCN2, HCN3, and HCN4) encode four different channel isoforms (HCN1-4) mediating hyperpolarization-activated currents (Ih) in the brain. Ih, activating upon relative hyperpolarization of the cell membrane, reduces the input resistance of the membrane (Rin) and plays complex and important roles in the trimming cellular and network activity [10,77,78]. Since the first implication of HCN channels in epilepsy in 2001, more and more studies have linked these channels to the epileptogenic process [79]. c.2156-2164delCGCCGCCGC variant in HCN2 gene, predicted to lead to the deletion of three consecutive proline residues (delProProPro), was found to confer a risk for FS and GEFS+ [80]. A HCN2 homozygous Glu515Lys mutation, causing loss of function, was found in a patient with sporadic idiopathic generalized epilepsy in 2011 [77]. The Hcn1 −/− knockout mice exhibited increased excitability and seizure susceptibility, while the Hcn2 −/− knockout mice showed spontaneous "absence" seizure phenotypes. In vitro studies showed that homomeric mutant, but not heteromeric wildtype/mutant channels, lowered the threshold of action potential firing and strongly increased cell excitability and firing frequency in transfected rat cortical neurons [77]. Evidence for HCN channelopathy in human epilepsy is far limited; however, a significant body of evidences obtained from animal modeling suggests that HCN channelopathy could be causative of genetic epilepsy, more likely as a polygenic or susceptibility trait [81]. Further studies are warranted to establish if mutations in HCN contribute to an epileptic phenotype in human.

GABA Receptors
There are three classes of GABA receptors, the ionotropic GABA A , GABA C receptors, and the metabotropic GABA B receptors. While all identified mutations to date are localized to GABA A receptors, members of the cys-loop family of ligandgated ion channels regulate the majority of inhibitory neurotransmission in the CNS [10,82,83]. Mutations in GABA A receptor subunit genes (GABRA1, GABRB3, GABRG2, and GABRD) have been reported to be associated with genetic epilepsy syndromes including CAE, JME, pure febrile seizures (FS), GEFS+, and SMEI.

The Gamma-aminobutyric Acid Receptor Alpha-1 Gene (GABRA1, MIM 137160)
The GABRA1 gene encodes the α1 subunit of the gammaaminobutyric acid receptor subtype A (GABA A receptor). Cossette et al. reported that GABRA1 Ala322Asp mutation was responsible for JME in a large French Canadian family. This nonconserved mutation introduced a negatively charged aspartate into the middle of the M3 transmembrane helix resulting in an impairment folding of α1 subunit thus causing a heterozygous loss of function of α1 subunit and eliciting a modest dominant negative effect [82,[84][85][86].
Maljevic et al. also described a heterozygous GABRA1 Ser326fs328X mutation in a German boy with CAE. Functional studies in HEK293 cells showed that the mutant protein had no channel current, and the subunit was retained in the cytoplasm and was not integrated into the plasma membrane, resulting in a complete loss of function [87]. The studies show a possibility that the reduction in GABA A receptor-medicated inhibition may result in a neuronal hyperexcitability, leading to epilepsy.
The Gamma-aminobutyric Acid Receptor Beta-3 Gene (GABRB3, MIM 137192) The GABRB3 gene, encoding the β3 subunit of the GABA A receptor, is highly expressed in embryonic brain where repressor-element-1-silencing transcription factor (REST) regulates neuronal genes, while expressed at lower levels in adult brain except in the hippocampus [88]. Urak et al. reported that a GABRB3 promoter haplotype 2 was associated with CAE and it may impair transcriptional activity [89]. The association was further confirmed by the discovery of Pro11Ser, Ser15Phe, and Gly32Arg mutations in CAE families [90]. Functional analysis indicated that mutated β3 subunit proteins could cause CAE via increased glycosylation and reduced GABA-evoked current [88,90]. Further studies are warranted to understand these epigenetic mechanisms.

The Gamma-aminobutyric Acid Receptor Delta Gene (GABRD, MIM 137163)
The GABRD gene encodes the δ subunit of the GABA A receptor, and its heterozygous variants (Glu177Ala and Arg220His) were reported to be associated with GEFS+ by Dibbens et al. [91]. It was reported that the GABRD susceptibility variant E177A is adjacent to one of the two cysteines that form a disulfide bond, while the R220H variant is located between the cys-loop and the beginning of the first transmembrane domain (M1), and both of them significantly reduced GABA A receptor current by impairing channel gating [82,91]. But Lenzen et al. [92] failed to find an association between Arg220His and IGE or JME.

The Gamma-aminobutyric Acid Receptor Gamma-2 Gene (GABRG2, MIM 137164)
The GABRG2 gene encodes the γ2 subunit, which forms the most abundant GABA A receptor subtype with α1 and β2 subunits in the CNS and plays a critical role in the brain function. Wallace et al. reported a heterozygous Arg43Gln mutation in the GABRG2 gene in a large family of patients with CAE and FS [93]. The γ2 subunit is known to be responsible for modulation of benzodiazepine and receptor targeting. Arg43Gln mutation may alter benzodiazepine sensitivity, receptor kinetics, assembly, trafficking, and cell surface expression, which is consistent with a reduction in GABA A receptor-mediated current [93]. Lys289Met and Lys328Met mutations were observed in families with a phenotype closely related to GEFS+. Lys289Met affects a highly conserved residue located in the extracellular loop between transmembrane segments M2 and M3, showing a decrease in the amplitude of GABA-activated currents and an acceleration of deactivation. Lys328Me is located in the short extracellular loop between transmembrane domains M2 and M3, which unchanged the brief GABA-evoked currents but had accelerated deactivation [82,94]. In a recent study, IVS6+2T→3G mutation was identified in an Australian family with CAE and FS, while the effect of this mutation on GABA A receptor function is unknown but was predicted to lead to a nonfunctional protein through exon skipping [2]. Homozygous Gabrg2 Arg43Gln mice showed rarely viable while heterozygous mutation demonstrated behavioral arrest associated with 6 to 7 Hz spike-and-wave discharges, which are blocked by ethosuximide, a firstline treatment for absence epilepsy [95]. A subtle reduction in cortical inhibition may underlie CAE seen in Arg43Gln mutation patients.

Neuroal Nicotinic Acetylcholine (nACh) Receptors
Nicotinic acetylcholine receptors (nAChRs) are ligand-gated ion channels that mediate fast signal transmission at synapses. They are hetero-or homomeric pentamers and are permeable to Na + , K + , and Ca 2+ . A total of 17 subunits (α1-10, β1-4, δ, ε, and γ) were identified, and mutations in the α and β subunit classes have been observed in patients with epilepsy.
Heterozygous Ser252Phe or +Leu264 insertion knock-in mice showed abnormal EEG patterns consistent with seizure activity [106], suggesting that the mechanism of ADNFLE seizures may involve inhibitory synchronization of cortical networks via activation of mutant CHRNA4 located on the presynaptic terminals and somatodendritic compartments of cortical GABAergic interneurons.

The Cholinergic Receptor Neuronal Nicotinic Beta Polypeptide 2 Gene (CHRNB2, MIM 118507)
The CHRNB2 gene encodes the β2 subunits, which is the main functional nAChR in the brain by a pentamer with α4 subunit. Heterozygous missense Val287Leu, within the M2 domain forming the wall of the ion channel, located in an evolutionarily conserved region of the CHRNB2 gene, was identified in a family with ADNFLE [107]. Since then, more mutations were identified in CHRNB2 gene associated with ADNFLE [108][109][110][111]. A 192-kb duplication in 1q21.3, encompassing the CHRNB2 gene, was identified in a boy with early onset absence epilepsy (EOAE) but not found in other independent patients (including 93 EOAE cases) [112].
Mouse model with the Chrnb2 Val287Leu mutation showed a spontaneous epileptic phenotype by electroencephalography with very frequent interictal spikes and seizures, indicating that mutant nicotinic receptors are responsible for abnormal formation of neuronal circuits and/or long-lasting alteration of network assembly in the developing brain, thus leading to epilepsy [113].

Other Nuclear Genes Linked to Epilepsy
The Leucine-Rich Glioma-Inactivated 1 Gene (LGI1, MIM 604619) The LGI1 gene, mapped to 10q23.33, encodes a secreted protein which has an N-terminal leucine-rich repeat (LRR) domain containing four LRRs flanked by two conserved cysteine-rich regions and a C-terminal epitempin (EPTP) domain containing seven EPTP repeats. The function of this protein in the CNS is largely unknown [10,114]. Morante-Redolat et al. reported that c758delC and c1320C>A mutations in this gene caused autosomal dominant lateral temporal epilepsy (ADLTE), a rare form of epilepsy characterized by partial seizures, usually preceded by auditory disturbances [115,116]. The LGI1 gene is the first non-ion channel gene identified in human idiopathic epilepsy. Although its precise mechanism is undefined, it probably differs from the so far known mechanisms of epileptogenesis, a potential mechanism of persistent immaturity of glutamatergic circuits [115,117]. At least 33 unique LGI1 mutations have been reported in ADLTE families and sporadic patients with idiopathic focal epilepsy with auditory symptoms, most of which are missense substitutions in both the N-terminal leucine-rich repeat (LRR) and C-terminal (EPTP) beta-propeller protein domains, and no obvious genotype/phenotype correlations were discovered [114,115]. Frameshift, nonsense, and splice site point mutations could result in protein truncation or internal deletion. Both truncating and missense mutations appear to prevent secretion of mutant proteins, suggesting a loss-of-function mechanism. A novel Arg407Cys mutation in familial temporal lobe epilepsy without any auditory or aphasic phenomena, as the first mutation does not prevent secretion of the mutant Lgi1 protein, is important for diagnostic purposes [118]. Recent research, however, shows no point mutation but a microdeletion about 81 kb in LGI1 in a family with ADLTE, indicating that copy number variant (CNV) analysis may be useful for identifying the pathogenesis [119]. Lgi1 Leu385Arg/Leu385Arg homozygous rats generated early onset spontaneous epileptic seizures from P10 and died prematurely. Heterozygous rats were more susceptible to sound-induced, generalized tonic-clonic seizures than control. A physiopathological loss of function may emerge not only due to a failure of protein secretion but also from a lack of correctly folded neuronal Lgi1 by functional studies [120].
The Monogenic Audiogenic Seizure-Susceptible 1 Gene (MASS1, MIM 107323) A natural mutation in monogenic audiogenic seizuresusceptible 1 gene (MASS1), mapped to 5q14.3, producing a fragment of the very large G protein-coupled receptor (VLGR1), has initially been reported in the Frings mouse strain that is prone to audiogenic seizures [121]. Subsequently, the MASS1 Ser2652X nonsense mutation, suggesting a loss of function, was identified in one family with febrile and afebrile seizures [122]. Deprez confirmed linkage of febrile seizures to MASS1 region highlighting the importance of this region [123], and a negative result in sequencing analysis of the MASS1 gene does not rule out the potential role of MASS1, because the method used for sequencing analysis was unable to detect a large heterozygous deletion in the gene.
The Solute Carrier Family 2 Member 1 Gene (SLC2A1, MIM 138140; Glucose Transporter 1, GLUT1) The SLC2A1 gene, mapped to 1p34.2, encodes glucose transport protein type 1 (GLUT1) which transports glucose from the bloodstream across the blood-brain barrier (BBB) to the central nervous system. Heterozygous mutations in the SLC2A1/GLUT1 gene were reported to cause cerebral energy failure and a clinical condition termed GLUT1-deficiency syndrome (GLUT1-DS), an autosomal dominant disorder, characterized by motor and mental developmental delay, seizures with infantile onset, deceleration of head growth often resulting in acquired microcephaly, and a movement disorder with ataxia, dystonia, and spasticity [116,124,125]. Klepper et al. identified a novel heterozygous mutation (Gly272Ala) in a father and two children from separate marriages affected by GLUT1 deficiency, all of whom developed into different kinds of epilepsy [125]. A co-occurring paroxysmal exertion-induced dyskinesia (PED), which is an additional phenotype of GLUT1-DS, and epilepsy were identified [124,126]. Striano et al. identified a Arg232Cys mutation in SLC2A1 in one Italian family with IGE in 2012, showing that GLUT1 defects were a rare reason for classic IGE. They considered that a chronic metabolic disturbance caused by lowered glucose transport across the blood-brain barrier is responsible for generalized epileptiform activity and epilepsy in patients with GLUT1 deficiency [127]. Importantly, seizures and movement disorders caused by the mutation in SLC2A1/GLUT1 gene may be treatable by ketogenic diet.
The EF-Hand Domain (C-terminal)-Containing Protein 1 Gene (EFHC1, MIM 608815) The EFHC1 gene, mapped to 6p12.2, contains 11 exons and encodes a protein with a Ca 2+ -binding EF-hand motif which is a microtubule-associated protein (MAP) involved in cell division and radial migration during cerebral corticogenesis [128]. It has been described that heterozygous mutations in EFHC1 cause JME, whereas homozygous Phe229Leu mutation was associated with primary intractable epilepsy in infancy [129][130][131][132]. A recent study shows that mutation of EFHC1 impaired mitotic spindle organization and disrupted radial and tangential migration by affecting the morphology of radial glia and migrating neurons, explaining how the EFHC1 mutations disrupt brain development and potentially produce structural brain abnormalities on which epileptogenesis is established [128,132]. Both the Efhc1 −/− and the Efhc1 −/+ mutant mice developed frequent spontaneous myoclonus in adult stages and a reduction of the threshold of seizures induced by pentylenetetrazol, suggesting that decrease or loss of function of myoclonin1, encoded by the mouse Efhc1 gene, may be the molecular basis for epilepsies caused by EFHC1 mutations [133].
The Proline-Rich Transmembrane Protein 2 Gene (PRRT2, MIM 614386) The PRRT2 gene, mapped to 16p11.2, contains four exons and encodes a 340 amino acid, proline-rich transmembrane protein, which is expressed primarily in the brain, especially in the cerebral cortex and basal ganglia [134,135]. Heterozygous mutations in PRRT2 were identified in families affected with benign familial infantile epilepsy (BFIE) and infantile convulsions and choreoathetosis (ICCA) syndrome. BFIE is characterized by an autosomal dominant inheritance self-limited seizure disorder that occurs in infancy, while ICCA is characterized by co-occurrence of infantile seizures and an adolescent-onset movement disorder, paroxysmal kinesigenic choreoathetosis (PKC). Mutations in the PRRT2 gene are the most common causes of BFIE or ICCA, responsible for 82 and 83 % of the cases, respectively [135]. Different clinical phenotypes and severity of PRRT2-mutated patients may due to different mutation types, location in gene, genetic background, epigenetic, environmental factors, etc.
The Polymerase DNA Gamma Gene (POLG, OMIM 174763) The POLG gene, mapped to 15q26.1, encodes the polymerase gamma, which is present in both the nucleus and the mitochondria and involved in the replication of mitochondrial DNA [143]. Its mutations are classically associated with Alpers syndrome, an autosomal recessive disorder characterized by a clinical triad of psychomotor retardation, intractable epilepsy, and liver failure in infants and young children [144]. Van Goethem et al. also reported one patient homozygous for a recessive missense mutation in POLG presenting with myoclonus, seizure, and sensory ataxic neuropathy, which overlapped with the syndrome of MERRF [145].

Mitochondrial DNA Mutations and Epileptic Features
Mutations in mitochondrial DNA can result in mitochondrial diseases which can affect any tissue and can start at any time of life. The CNS and muscles are two common body systems to be affected because of a high demand of energy generated via the mitochondria. These mitochondrial diseases are also known as mitochondrial encephalomyopathy. Epilepsy may be the presenting or late feature of mitochondrial encephalomyopathy. Common mitochondrial epileptic syndromes include the mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS) and MERRF syndromes [146]. MELAS syndrome, characterized by seizures, hemiparesis, hemianopsia, cortical blindness, and episodic vomiting, is a genetically heterogeneous mitochondrial disorder including mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episodes. Goto et al. identified a point mutation (Ala3243Gly) in the transfer RNA mitochondrial leucine 1 gene (MTTL1) in 80 % of patients with MELAS [147], while the rest of the mutations associated with MELAS were observed spreading in other parts of MTTL1 and other mitochondrial genes. MERRF is characterized by myoclonus, epilepsy, ataxia, muscle weakness, hearing loss, and elevated serum lactate and pyruvate levels. The transfer RNA mitochondrial lysine gene (MTTK) Ala8344Gly mutation, which alters the T psi C loop of the tRNA(Lys) gene for 80 to 90 % of MERRF cases, providing a simple molecular diagnostic test for the disease, was identified by Schoffner et al. in MTTK gene [148,149].

Conclusion
In this article, we summarized the recent research and progress in the genetics of epilepsy. Since CHRNA4 has been identified as the disease-causing gene of human ADNFLE, an idiopathic partial epilepsy syndrome in 1995, dramatic advances in epilepsy research have been made in the past two decades with the discovery of a series of genes that are responsible for monogenic epilepsy. However, it is hard to predict how many epilepsy-associated genes will exist in the human genome. We know that ion channel mutations play a central role in the pathological mechanism of epilepsy and the epilepsy syndrome. However, epilepsy can also be caused by dysfunction in neuronal migration, glycogen metabolism, or respiratory chain activity. For example, mutations leading to a deficiency of glucose transporters will lead to failure of transporting glucose from the bloodstream across the BBB to the CNS, which will result in a cerebral energy failure and then leading to a seizure [124]. However, monogenic determined epileptic syndromes may account only for a minority of the idiopathic epilepsies, and consequently, genetic tests should be performed after accurate clinical selection of families and probands. Epilepsy is complex and multifactorial, likely involving a combination of environmental exposures, polygenic inheritance, and gene-environment interactions. Molecular karyotyping has revolutionized the discovery of rare and CNVs-deletion, insertions, duplications-associated with epilepsy including recurrent microdeletions at 15q11.2, 15q13.3, and 16p13.11 as substantial risk factors for epilepsy [150]. Microdeletions or deletions may lead to either haploinsufficiency of specific genes, or unmasking of recessive mutations in the remaining allele. Epigenetic effect or sensitization to environmental influences the genetic background of the individuals and then improves the susceptibility [150]. Usually, mutation in an important domain of a gene may cause a monogenic form of disease, whereas a nucleotide variant in a noncritical region may enhance susceptibility to, or protect against, the disorder [151]. With the continued application of genomewide approaches including genomewide association study, exome sequencing, whole genome sequencing, or whole genome expression strategy in large cohorts of individuals with epilepsy syndromes, more genomic regions and novel genes important to the genetic etiology of this complex and common neurologic disorder will be discovered. Moreover, little attention was paid on nonchannel mechanisms in genetic epilepsies, such as the brain energy metabolism, which possibly would be another fruitful area by a wider exploration. With the development of new molecular biology techniques such as whole exome and whole genome sequencing, the subsequent research on pathogenesis of epilepsy should focus not only on monogenic epilepsy but also on the complex epilepsy. We hope that, in the near future, other epilepsy-causing genes will be discovered and other genetic and nongenetic factors responsible for epileptic phenotypes will be identified, making the therapy on epilepsy to include not only drug or surgery but also gene therapy.