- Top of page
Approximately 40% of epilepsy has a complex genetic basis with an unknown number of susceptibility genes. The effect of each susceptibility gene acting alone is insufficient to account for seizure phenotypes, but certain numbers or combinations of variations in susceptibility genes are predicted to raise the level of neuronal hyperexcitability above a seizure threshold for a given individual in a given environment. Identities of susceptibility genes are beginning to be determined, initially by translation of knowledge gained from gene discovery in the monogenic epilepsies. This entrée into idiopathic epilepsies with complex genetics has led to the experimental validation of susceptibility variants in the first few susceptibility genes. The genetic architecture so far emerging from these results is consistent with what we have designated as a polygenic heterogeneity model for the epilepsies with complex genetics.
Most of the known genetic basis for idiopathic epilepsy relates to those ‘monogenic’ epilepsies with mutations that were relatively easy to map by linkage and to identify because they segregated in large families and localized to chromosomal regions containing neuronal ion channels (Heron et al. 2007b). The genes so far identified for these rare monogenic epilepsies (CHRNA4, CHRNA2, CHRNB2, KCNQ2, KCNQ3, SCN1A, SCN2A, SCN1B, GABRA1, GABRG2, CLCN2, LGI1 and EFHC1) all code for ion channels except for the last two, and these last two are known to be associated with ion channels (Schulte et al. 2006; Suzuki et al. 2004). The variation in expressivity and penetrance associated with mutations of large effect in these ‘monogenic’ epilepsies strongly suggest that variation in additional modifier genes and/or environmental factors act in combination with the mutated primary gene in moulding the final phenotype. These modifier genes may be the same susceptibility genes underlying epilepsy with complex genetics.
Recognition of an epilepsy susceptibility gene is based on detection of allelic variation associated with an experimentally detectable change in functionality, consistent with seizure propensity. In epilepsy with complex genetics, seizure syndromes can be similar or identical to those seen in ‘monogenic’ epilepsy, but are thought to arise instead from the additive or interactive effects of more than one susceptibility gene. We refer to these as epilepsies with complex genetics to avoid confusion in a field where complex epilepsy has traditionally referred to seizure complexity; but epilepsies with complex genetics are what in population genetics would generally be regarded as a complex epilepsy. Both idiopathic generalized epilepsies (IGEs) and temporal lobe epilepsies (TLEs) are common epilepsies with complex genetics.
The γ-aminobutyric acid (GABA)A receptor is the primary mediator of synaptic inhibition in the brain, and the subunits encoded by GABRG2 and GABRA1 are associated with monogenic forms of IGE and generalized epilepsy with febrile seizures plus (GEFS+) in rare families (Heron et al. 2007b). The effects on ion channel characteristics of at least two genetic variants encoding the δ-subunit of the GABAA receptor, GABRD, are consistent with a role as a susceptibility gene for IGE with complex genetics. This is based on the electrophysiological effects on GABAA receptor chloride channels containing the E177A and R220H variants (Dibbens et al. 2004; Feng et al. 2006). Expression of the GABRD variants in human embryonic kidney 293T cells results in reduced receptor current amplitude in comparison with the common GABRD allele. The finding that receptors containing E177A or R220H substitutions show decreased surface receptor expression and a shorter duration of channel opening (Feng et al. 2006) most likely explains the basis for the reduced GABA-induced currents.
The rare GABRD E177A variant was present in one small GEFS+ family and the low-frequency polymorphism R220H was present at a similar frequency in patients and controls (Dibbens et al. 2004), suggesting that GABRD is just one of a large number of susceptibility genes in epilepsy with complex genetics. These variants have not been observed segregating with epilepsy in a large family, consistent with their polygenic role affecting seizure threshold in sporadic cases or small families. The δ-subunit is present in extrasynaptic and perisynaptic receptors that mediate tonic inhibition, suggesting that this mechanism plays a role in epilepsy. Tonic inhibition dampens the size and duration of excitatory synaptic activities, making it less likely to fire action potentials (Farrant & Nusser 2005). A potential reduction of tonic inhibition by the GABRD variants is, therefore, consistent with neuronal hyperexcitability and increased susceptibility to seizures.
Variants in the T-type calcium channel gene, CACNA1H, which is involved in the thalamocortical network, were initially associated with childhood absence epilepsy (CAE) (Chen et al. 2003) but the relationship has since been extended to other epilepsy phenotypes (Heron et al. 2004). In both cases, the variants affected conserved amino acids, suggesting that they would potentially alter protein function, but they did not segregate with affection status in families. Electrophysiological analysis showed that at least some of the variants did affect protein function (Khosravani et al. 2004, 2005; Peloquin et al. 2006; Vitko et al. 2005). Furthermore, analysis of a polymorphism, R788C, showed that it also alters protein function, consistent with it being a susceptibility polymorphism. When this polymorphism was expressed with the CAE-related variant, G755D, protein properties were different from those observed when either variant was expressed alone. This shows that the interaction of common polymorphisms with rarer variants or mutations can also play a role in altering protein function. These studies are consistent with a view that variation in CACNA1H contributes to the pathogenesis of epilepsy with complex genetics but no variants in CACNA1H have been described, which are sufficiently pathogenic to cause epilepsy on their own and segregate with the phenotype in large families.
TLE has one documented rare variant in KCND2, which has functional consequences (Singh et al. 2006) that we interpret as consistent with the recognition as a susceptibility gene. Similarly, for IGE, the promoter polymorphism in GABRB3 must be considered a susceptibility gene on functional evidence (Urak et al. 2006).
The rarity and/or weakness of involvement of genetic variants at CACNA1H, GABRD, KCND2 and GABRB3 as effectors in epilepsy with complex genetics suggests a model in which subsets of causative variants at susceptibility genes are drawn from a much larger population of genes with rare and polymorphic (mostly at low frequency) susceptibility alleles. That is, variants of none of the above genes are necessary or sufficient effectors of epilepsy, and in the absence of proven associations, that may be the general rule for all susceptibility genes in epilepsy. This model is consistent with observations from the ‘monogenic’ epilepsies. SCN1A, SCN1B and GABRG2 mutations of large effect account for few of the GEFS+ cases and CHRNA4 and CHRNB2 account for few of the nocturnal frontal lobe cases (Scheffer & Berkovic 2003). Rare CACNA1H variants, C456S, A480T and D1463N and a rare GABRD variant, R220C showed no detectable abnormality in vitro (Dibbens et al. 2004; Khosravani et al. 2004, 2005). These could truly represent rare neutral ion channel variants (at the other end of the spectrum compared with mutations of large effect in monogenic epilepsies) or perhaps the definitive physiological tests for any as yet cryptic abnormal ion channel property related to susceptibility for these variants has not yet been carried out or devised, or their effects are too small to be detected.
So far, most CACNA1H variants, one GABRD variant and one KCND2 variant linked to susceptibility by functional experimentation are consistent with the multiple rare variant common epilepsy model (Mulley et al. 2005a). The susceptibility polymorphisms of GABRD, CACNA1H and GABRB3 are consistent with the ancestral common variant common epilepsy (ACVCE) model (Mulley et al. 2005a). Thus, the meagre evidence so far available points to a mixture of both models underlying epilepsy with complex genetics, which we call the polygenic heterogeneity model (Table 1).
Table 1. Architectural models for epilepsy with complex genetics
|MRVCE: multiple rare variant complex epilepsy model|
| Rare functional variants: CACNA1H, GABRD and KCND2|
|ACVCE: ancestral common variant common epilepsy model|
| Simplistic form: same functional polymorphism at intermediate frequency underlying sufficient cases to be detectable by association study: GABRB3?|
| Realistic form: low-frequency polymorphism or intermediate frequency polymorphism of very weak effect: CACNA1H and GABRD|
|Polygenic heterogeneity model: combination of all of the above with heterogeneous but pathogenic subsets of susceptibility alleles drawn from a much larger pool of potential susceptibility genes (Mulley et al. 2005a), meaning variation at no individual susceptibility gene is necessary or sufficient for seizures|
The message from these early results from CACNA1H, GABRD and KCND2 is that neither the rare mutations nor the low-frequency single nucleotide polymorphisms (SNPs) (Dibbens et al. 2004; Khosravani et al. 2004, 2005) would have been statistically detectable by an association analysis. As an example, the GABRD polymorphic susceptibility allele R220H was no more prevalent among affected cases than controls, so was recognized for all practical purposes as being undetectable by the association study approach (Dibbens et al. 2004). This was subsequently experimentally confirmed in another population (Lenzen et al. 2005).
However, Heron et al. (2007a) followed a promoter polymorphism of GABRB3, which came to attention initially through an association study, was confirmed by an association study, subsequently shown on functional grounds to have properties consistent with its recognition as a susceptibility gene, and then subsequently the association could not be replicated in a much larger sample from the same population! This led to questioning the process and proposing the argument for going straight to functional studies at least for naturally occurring variation in ion channel genes (Heron et al. 2007a).
Screening and/or resequencing of all candidate genes for a complex genetic disorder (with the cost of such an exercise explored by Dean 2003) followed by functional testing of variations may be the only practical approach currently on the horizon. Functional rare variant or SNP variation in any of the approximately 200 neuronal channels from approximately 400 known channels (Gargus 2003; Gustincich et al. 2003) are the most plausible candidates for epilepsy with complex genetics, based on current knowledge of genes responsible for monogenic epilepsies. Gene families other than ion channels cannot be ruled out; however, their identities are elusive. The worst case scenario is that much of that portion of the genome expressed in neurons and especially at the synapse could also be involved in epileptogenesis, which would render epilepsy an extremely complex genetic disorder, if variants at many of these additional genes were involved in concert with ion channel variation.
Conversely, the degree of genetic heterogeneity for the same ‘monogenic’ syndromes provides insight into the potential number of underlying genetic options also available for each of the syndromes of epilepsy with complex genetics. Several genes known to segregate mutations of large effect can give rise to the same syndrome: CHRNA4, CHRNB2 and CHRNA2 to ADNFLE (Aridon et al. 2006; Phillips et al. 2001; Steinlein et al. 1995), and SCN1B, SCN1A and GABRG2 to GEFS+ (Baulac et al. 2001; Escayg et al. 2000; Wallace et al. 1998, 2001). In both syndromes, these known genes only account for approximately 10–20% of familial cases (Combi et al. 2004; Scheffer & Berkovic 2003), suggesting that the remaining 80–90% may be complex with polygenic contributions from these and/or other genes. If various permutations and combinations of susceptibility loci can underlie the same syndrome, then translated from monogenic epilepsies to the polygenic epilepsies, this multiplies the underlying genetic complexity for particular syndromes relative to what is already known for ‘monogenic’ epilepsy.
The relative contribution of polymorphisms vs. rare variants to the genesis of the epilepsies with complex genetics is not known; however, even in these early days both mechanisms have been experimentally shown (Dibbens et al. 2004; Khosravani et al. 2004, 2005). There is theoretical evidence for a contribution of common variants to disease susceptibility (Lohmueller et al. 2003) and examples that are now applied in clinical practice (apolipoprotein E [APOE] in Alzheimer’s disease: Corder et al. 1993; factor V in deep vein thrombosis: Bertina et al. 1994). Dean (2003) lists many others; however, it is unlikely that all the variation underlying any complex disorder can be totally explained by common variants (Pritchard & Cox 2002; Terwilliger et al. 2002). Intuitively, we know that a plethora of intragenic rare variants must also play a role in complex disease, and in the epilepsies with complex genetics, in the same way that they have been shown to underlie the monogenic diseases (Mulley et al. 2005b), and indeed epilepsies with complex genetics (Mulley et al. 2005a).
Attempts to identify susceptibility genes from multiplex families with clustering of epilepsies with complex genetics have not yet been successful. While Sander et al. (2000) identified a ‘susceptibility’ region in 3q26, which was later examined and found to have CLCN2 mutated in 3 of the 46 families (Haug et al. 2003), it is unlikely that just these three families could have accounted for the linkage result seen on the genome scan. These three families are examples of rare ‘monogenic’ familial epilepsy rather than epilepsy with complex genetics. The mutant alleles of large effect in these families segregate vertically with affected family members, consistent with autosomal dominant monogenic inheritance. The three mutations were shown convincingly to represent true epilepsy mutations; however, as with all other epilepsy genes discovered so far, CLCN2 is not a gene commonly mutated in IGE (D’Agostino et al. 2004). The 3q26 susceptibility region differs from the susceptibility region on chromosome 3 detected by Durner et al. (2001). Both studies were carried out on multiplex families of epilepsy with complex genetics but neither localization has been replicated.
Similarly, Pal et al. (2003) suggested BRD2 (RING3) as the first probable susceptibility gene for juvenile myoclonic epilepsy (JME) in 6p21, based on linkage and linkage disequilibrium studies. However, no associated variation with pathogenic consequences within BRD2 makes its status as a putative susceptibility gene for epilepsy with complex genetics doubtful. Pinto et al. (2005) similarly suggested candidate genes for photosensitive epilepsy at 7q32 and 16p13, but outcomes from testing the named candidate genes from these regions remain to be reported. ME2 has also been proposed as a susceptibility gene (Greenberg et al. 2005) unique in that it would explain recessive epilepsy on metabolic grounds, but consistent with previous association studies in that no supporting functional evidence has been presented. Audenaert (2005) recently mapped a common GEFS+ susceptibility locus to 2p24, which is of interest given that most GEFS+ cases appear not to be monogenic.
None of the above genes proposed on the basis of statistical association studies is proven susceptibility genes by the criteria of detectable alterations in functionality. This suggests that they could represent false-positive signals. The absence of true-positive signals is consistent with a polygenic susceptibility model. However, Tan et al. (2004) remain ‘cautiously optimistic’ that with methodological refinements and vast increases to sample sizes involving large multicentre studies, the molecular genetic basis for epilepsy with complex genetics can be determined by case–control association studies. This requires considerable faith in the existence of the simplistic form of the ACVCE model (Table 1) for at least some epilepsy susceptibility genes. They reviewed numerous genetic association studies in epilepsy that had been carried out without successful replication. There are repeatability and power problems in detecting any associations that really may exist; so Tan et al. (2004) proposed a set of guidelines for future association studies in epilepsy.
Alternatively, the underlying genetic reservoir of alleles at susceptibility genes needs to be reduced in other ways, for example, by population bottleneck or founder event, by targeting ethnically homogeneous or isolated populations. Parametric linkage analysis applied to single large families segregating a mutation of large effect is the extreme example for eliminating genetic heterogeneity. Unlike epilepsy, for other brain disorders such as schizophrenia, autism, depression and bipolar disorder, there do not appear to be highly penetrant pathogenic mutations of large effect in rare large autosomal dominant families, so progress in the identification of their susceptibility genes has been frustratingly slow in the absence of an entrée into their complex genetics. The polygenic heterogeneity model proposed for epilepsy with complex genetics, comprising heterogeneous permutations and combinations of subsets of rare and mainly low-frequency polymorphic variation, perhaps represents a paradigm for other genetically complex brain disorders.