Translational Research in Epilepsy Genetics: Sodium Channels in Man to Interneuronopathy in Mouse

Voltage-gated sodium channels are critical for membrane excitability. Mutations in the genes coding for these proteins cause diseases related to altered excitability of cardiac or skeletal muscle and neurons. Mutations in the central nervous system–specific voltage-gated sodium channel α1 subunit gene (SCN1A) lead not only to seizure syndromes but also to familial hemiplegic migraine. The epilepsies range from benign febrile seizures to the catastrophic epileptic encephalopathy of Dravet syndrome (severe myoclonic epilepsy of infancy). Recently developed animal models of SCN1Amutants recapitulate the human disease. These models exemplify the potential inherent in translational research to debunk preconceived ideas regarding pathogenesis by showing the cellular substrate of Dravet syndrome to be interneurons rather than excitatory cells. This illustrates the key role that basic science plays in the development of targeted novel therapies and, ultimately, in the prevention of devastating genetic disorders.

The key role of sodium channel mutations in disease underscores their importance in membrane excitability whether they are located in brain, heart, nerve, or muscle cells (Table). The sodium channel comprises a pore-forming α subunit with auxiliary β subunits that regulate channel function. The sodium channel α1 subunit gene (SCN1A, OMIM *182389) is now the most clinically significant epilepsy gene with mutational analysis serving as a diagnostic test for specific phenotypes. Mutations in SCN1Awere first associated with the familial epilepsy syndrome genetic (generalized) epilepsy with febrile seizures plus (GEFS+).^1,2From this beginning, the spectrum of diseases associated with SCN1Anow includes severe infantile epileptic encephalopathy, Dravet syndrome, and familial hemiplegic migraine (FHM). Herein, SCN1A-related disorders will be described before discussion of the new SCN1Aanimal models and the ways they expand our understanding of the pathogenesis of human epilepsy.

Unlike most epilepsy syndromes, where the diagnosis is based on a patient's seizure syndrome in isolation, GEFS+ is diagnosed by the presence of appropriate phenotypes in 2 or more family members.^3,4The GEFS+ phenotypes are childhood-onset seizure syndromes with a strong tendency toward fever-induced seizures, although persistence of occasional generalized or partial seizures into adulthood may occur. The most common GEFS+ phenotypes are febrile seizures (FS) and febrile seizures plus (FS+). The syndrome FS+ is defined by febrile seizures that occur outside the usual age range of 3 months to 6 years or that are intermixed with afebrile convulsions. Both FS and FS+ may be accompanied by myoclonic, absence, and astatic seizures. Other phenotypes, including partial seizures, are seen as part of the GEFS+ spectrum, as are, rarely, the severe childhood epilepsies Dravet syndrome, described in the following subsection, and myoclonic-astatic epilepsy. Although the initial description emphasized the generalized nature of the phenotypes that predominate in GEFS+, the more inclusive term genetic epilepsy with febrile seizures plusbetter expresses the core features of the syndrome.

Genetic (generalized) epilepsy with febrile seizures plus was initially identified through the study of families with evidence of a gene of dominant effect. With the aid of these rare, large families, first the sodium channel β1 subunit gene (SCN1B) then SCN1Aand the γ-aminobutyric acid (GABA) receptor γ2 subunit gene (GABRG2) were isolated as epilepsy genes.^5-7SCN1Ais the most frequently implicated gene, although it accounts for only a small proportion of families with GEFS+. Most families with GEFS+ show complex inheritance.

SCN1Ais also not causative in most febrile seizures in the community. There is a large family with a pure febrile seizures phenotype with an SCN1Amutation, but this phenomenon can be conceptualized as fitting within the spectrum of GEFS+.⁸Sporadic febrile seizures cases have not shown mutations.⁹Similarly, most tested sporadic FS+ cases also have not shown mutations.¹⁰

Dravet syndrome is strongly associated with mutations in SCN1A. To comprehend the evolution in the understanding of this syndrome, it is worth reviewing the history of severe myoclonic epilepsy of infancy (SMEI) and SCN1Amutations.

Severe myoclonic epilepsy of infancy was originally described by Charlotte Dravet et al as an epileptic encephalopathy with onset in the first year of life.¹¹The classic syndrome starts with prolonged febrile convulsions at approximately 6 months of age. Even if afebrile initially, sensitivity to temperature-induced seizures is usual. Hemiconvulsion with the affected side varying is characteristic, but generalized convulsions and convulsive status epilepticus are also common. Seizures worsen in the second year of life, with multiple seizure types appearing. Myoclonus was required for diagnosis under the original description and usually was accompanied by a variety of complex partial, absence, atonic, and other seizure types. Spasms were not a feature. Development is initially normal but slows in the second year of life. Unlike in many catastrophic childhood epilepsies, the typical progression of findings via electroencephalogram (EEG) is present. Generalized spike and wave appears at approximately 2 years of age. Paroxysmal fast activity and slow spike and wave suggestive of symptomatic generalized epilepsy are not seen. Seizure-related mortality is high in childhood.

SCN1Awas investigated in SMEI owing to the strongly fever-related seizures and the high frequency of GEFS+ in patients' families.⁴Current estimates are that 80% of patients will harbor a causative SCN1Amutation.^12,13There are isolated cases of SMEI associated with GABRG2mutations.¹⁴

The term Dravet syndromeis becoming preferred to SMEIbecause many patients lack features previously regarded as cardinal. Clinically, the absence of myoclonus, a defining feature of SMEI, does not change the prognosis if onset, temperature sensitivity, and the typical changes on the EEG are present.¹⁵The more important step has been to look at the spectrum of catastrophic infantile-onset epilepsies that carry a similarly high rate of SCN1Amutations.

In a large cohort of SMEI borderland (SMEB) missing 1 or more features of SMEI, 70% were found to still harbor mutations.¹³The key features of seizure onset between the ages of 0 and 1 and developmental slowing were present in all these patients (Figure). Overall, the mutation rate in SMEB is similar to that in classic SMEI, as is the prognosis. The previously described intractable childhood epilepsy with generalized tonic-clonic seizures (ICE-GTC) also shows a high mutation rate.¹⁶It is, in essence, Dravet syndrome in which only convulsive seizures occur. The modern limits of Dravet syndrome are not precisely drawn. These disorders are clinically very similar to SMEI and have a similar mutation rate, suggesting that the term Dravet syndromeshould encompass all 3 entities.

In a fourth syndrome associated with SCN1Amutations, severe infantile multifocal epilepsy (SIMFE), early onset of partial seizures, and a multifocal EEG reading precede developmental slowing (Figure).¹³Generalized spike and wave is not seen. The developmental slowing may have a later onset than in Dravet syndrome. Although SIMFE is a catastrophic form of childhood epilepsy carrying a high rate of SCN1Amutation, the difference in clinical presentation from Dravet syndrome is significant. Despite the electroclinical differences, the group of SCN1A-associated childhood epileptic encephalopathies shares the features of onset before 1 year of age and initial normal development (Figure). Scattered cases of other syndromes harbor SCN1Amutations. A few myoclonic-astatic epilepsy cases and some idiopathic Lennox-Gastaut syndrome cases are accounted for by SCN1Amutations.^9,10,13

Dravet syndrome remains recognizable into adulthood.¹⁷Nocturnal generalized tonic-clonic seizures are seen in all reported cases, and focal, myoclonic, and atonic seizures are seen in some. Unlike the more common Lennox-Gastaut syndrome, tonic seizures are generally not seen. Apart from epilepsy, significant intellectual disability is present in almost all patients with Dravet syndrome, as are, variably, cerebellar ataxia, spasticity, and extrapyramidal dystonia. In addition, Dravet syndrome accounts for so-called vaccine encephalopathy, with a retrospective study¹⁸showing nearly all patients having the clinical features and 11 of 14 having SCN1Amutations.

The evidence for therapy in Dravet syndrome has recently been systematically reviewed.¹⁹Freedom from seizures is rarely achieved, but, as Dravet syndrome is an epileptic encephalopathy, aggressive treatment of seizures has the potential to improve cognitive outcome. Two randomized trials^1,2of the drug stiripentol show a short-term reduction in seizures when combined with sodium valproate and clobazam. Observational evidence³suggests that this reduction is potentially durable. Topiramate, levetiracetam, and the ketogenic diet have observational evidence^4-6to support their use. Lamotrigine therapy may worsen seizures in Dravet syndrome, and there are reports^7,8of phenytoin sodium use precipitating chorea.

Familial hemiplegic migraine (FHM) is a dominant condition in which, in addition to migraine with typical aura, patients experience hemiplegia as an aura symptom, often with stupor. Precipitation of attacks with minor head trauma is not uncommon. Two different mutations in SCN1Ahave been recently described in 4 families with FHM.^20-22SCN1Ais an uncommon cause of FHM, with most families with identified genes having mutations in either the calcium channel gene CACNA1Aor the sodium/potassium pump gene ATP1A2. There is no difference in the headache phenotype between the SCN1A-related cases and those of other patients with FHM. Although there are FS in a few patients with SCN1A-related FHM, overall there does not seem to be significant clinical crossover between GEFS+ and FHM.

SCN1Ais located on the long arm of chromosome 2 in a cluster of voltage-gated sodium channel genes. SCN1Acodes for a brain-specific voltage-gated sodium channel (Na_v) subunit.²³Voltage-gated sodium channels consist of a pore-forming α subunit, of which there are 4 brain-expressed forms, and 2 β subunits. The protein product of SCN1A, Na_v1.1, is an α subunit that forms the transmembrane pore through which sodium ions flow. β Subunits act as regulatory elements, altering the voltage sensitivity and subcellular localization. The gene is large, with 26 exons across a 100-kilobase span of the genome.

Mutations in SCN1A-related disease are spread throughout the gene and are rarely repeated.¹²Rather than a single, ancestral variant, there are multiple “private” mutations. The mutations described in GEFS+ and FHM are single base changes leading to amino acid substitution (missense mutations). Deletions, mutations that grossly alter transcription (nonsense and frameshift mutations), and mutations that radically alter the preparation of RNA for translation (splice site mutations) are seen in Dravet syndrome but not in GEFS+ or FHM.¹⁵Despite this fact, 40% of Dravet syndrome–related mutations are missense rather than the catastrophic sequence changes mentioned previously herein. This means that the discovery of a missense mutation upon testing a child does not guarantee a benign outcome.

Dravet syndrome is largely associated with de novo mutation without detectable abnormality in the parents (95%). When inherited missense mutations occur, they lead to mild GEFS+ phenotypes in the other family members.¹²These inherited mutations are an uncommon cause of Dravet syndrome.

Voltage-gated sodium channels are responsible for the initial upstroke of the action potential in neurons and other excitable cells.²³Depolarization of the membrane produces opening of the channel at the external surface. As the membrane potential rises, a separate inactivation gate on the inner surface of the channel stops ion flow. This inactivation gate then remains closed to produce a refractory period, preventing further voltage-induced opening of the channel. How the channel performs is related to the function of these gates. Voltage sensitivity, current density, and the refractory period may all vary, as may residual current flow after inactivation. When expressed in model cell lines, the Na_v1.1 protein products of the SCN1Amutations associated with GEFS+ show diverse alterations in their physiologic features.

Most reported GEFS+ mutations seem to increase excitability, although the opposite also has been found. The results for even a single mutation vary between model systems. A good example of this is the GEFS+-associated missense mutation R1648H, which has been studied in 3 in vitro models. Using the homologous region of human SCN4A,it slowed inactivation and accelerated recovery from inactivation.²⁴With rat Scn1aand a Xenopusoocyte model, only accelerated recovery was shown.²⁵Human SCN1Aco-expressed with SCN1Bin a human kidney cell line showed a third pattern of increased persistent current despite inactivation.²⁶It has been suggested that this increased persistent current is the most prevalent finding, but its relevance to in vivo channel function is unclear.²³Overall, the alteration in channel kinetics in GEFS+-related mutations is sufficient to support their pathogenicity but not to make conclusions about the mechanism.

The mutations seen in FHM have been modeled using the homologous sequence in the cardiac sodium channel gene SCN5Arather than the SCN1A-derived channel.²⁰Different FHM mutations produce a similar change in kinetics to that seen with the SCN4Amodel of the GEFS+ mutation R1648H. The fact that mutations with similar sequence and electrophysiologic changes can result in such different phenotypes as GEFS+ and FHM is remarkable.

In a significant proportion of Dravet syndrome–related mutations, no functional protein is produced. The epilepsy due to SCN1Adeletion is identical to that caused by transcribed but untranslated messenger RNA.¹⁵This result implies that it is the absence of 1 of the 2 copies of the gene (haploinsufficiency) that causes disease, rather than a dominant negative effect. Haploinsufficiency is, however, insufficient as an explanation for all of Dravet syndrome. Missense mutations account for 40% of Dravet mutations and do not always show complete failure of Na_v1.1 function. For example, the Dravet-associated mutation R1648Chas a different change at the same locus as the GEFS+-associated allele R1648Hmentioned previously herein. R1648Cshows similar persistent current effects in the human SCN1Amodel as those mentioned previously herein for R1648H.²⁷

Why missense mutations should produce Dravet syndrome rather than GEFS+ in a patient is not known. Modifier genes, in addition to SCN1A,have been suggested as reasons. This explanation would be supported by the finding of an excess of GEFS+ in the families of patients with Dravet syndrome.⁴This frequently reported excess of familial seizures has recently been challenged by an Italian group²⁸whose study results showed no difference in seizure risk in the relatives of patients with Dravet syndrome compared with controls.

Dravet syndrome–associated haploinsufficiency has recently been modeled in mutant mice. These mice provide a means to investigate Na_v1.1 function in vivo. This allows the effects of the mutation to be seen at a range of levels, from subcellular localization to cell-type–specific and network behavior in a functioning animal. Any compensatory effects also can be observed.

The mouse model that was published first has the SCN1Ahomologue deleted, whereas the second uses a nonsense mutation found independently in 3 patients with Dravet syndrome.^29,30Heterozygous mice, in which the genotype is analogous to that of patients with Dravet syndrome, had spontaneous seizures and ataxia beginning in the fourth postnatal week. Spontaneous death from a presumed seizure-related cause was frequent. Homozygous mutant mice developed seizures and ataxia earlier, during the second postnatal week, and died during the third postnatal week.

Labelling studies in normal rat brains suggest that Na_v1.1 makes up less than 15% of the total Na_v.³¹Na_v1.1 is present in the axon hillocks and soma of neurons, including interneurons and pyramidal cells. These studies have not been repeated in mice.

Studies in mutant mice suggest that the contribution of Na_v1.1 to the functioning of pyramidal cells is minimal. Instead, in the hippocampus and neocortex, Na_v1.1 is critical for GABAergic interneurons. In the heterozygous mutants, there is a 50% loss in sodium current density in interneurons without any change in the kinetics of the channels. This suggests that not only is most sodium current in interneurons due to Na_v1.1 but also that interneurons are susceptible to haploinsufficiency of Scn1a. At a neuronal level, interneurons initiated but did not maintain bursts of fast-spiking activity.³⁰In situ hybridization and immunolabeling also supported Scn1abeing most prominently expressed in interneurons. Although an increase in the number of interneurons staining for Na_v1.3 suggests that this protein is being upregulated in heterozygotes, the electrophysiologic findings demonstrate that this is inadequate to compensate for the loss of Scn1a.²⁹

A similar specificity to GABAergic neurons is also seen in the cerebellum.³²Purkinje cells are the only output of the cerebellar cortex and they use GABA as a transmitter. Antibodies to Na_v1.1 label Purkinje cells strongly. In heterozygote mice, a loss in Purkinje cell sodium current density similar to that in neocortical interneurons was seen. The GABAergic interneurons in the molecular layer of the cerebellum may also rely on Na_v1.1, but this proposition awaits confirmation.³⁰

The mouse findings suggest that Scn1ahaploinsufficiency causes a loss of cellular excitability in inhibitory cells. The failure of GABAergic interneurons then produces an increase in overall cortical excitability. In humans, Na_v1.1 is present in the cerebral cortex in interneurons and pyramidal cells. Whether a similar Na_v1.1 dependence of interneurons in humans exists awaits quantitative cellular studies.

It is also not yet clear how GEFS+- and FHM-related mutations cause disease. Most mutations in both diseases seem to increase rather than decrease membrane excitability in model systems. Whether this effect reverses under in vivo conditions or whether some mechanism other than the interneuronopathy seen with Scn1ahaploinsufficiency is present awaits further study.

Although the phenotype of homozygous Scn1amutant mice does not vary with the genetic background, the heterozygote phenotype is markedly dependent on mouse strain. In the deletion model,²⁹mice from a C57BL/6background had catastrophic seizures and 80% mortality by 15 weeks of age. Heterozygote mice with the same Scn1agenotype from a 129/SvJbackground had no spontaneous seizures. A cross between the 2 strains showed spontaneous seizures, with 40% mortality. Ogiwara et al³⁰showed similar variability due to genetic background with their nonsense mutation model. The alteration of severity with genetic background has correlates in human disease. Dravet syndrome varies markedly between patients, particularly in developmental outcome. Modifier genes are also thought to be crucial in Dravet mutations inherited from mildly affected parents.^13,14

A double-transgenic mouse with Scn1aand Scn8amutations has been used to look at the question of modifier genes.³³Scn8acodes for Na_v1.6, a sodium channel in the peripheral and central nervous systems. Scn8amutation or deletion in mice affects cerebellar and neuromuscular function in a recessive manner, but also increases seizure threshold to pharmacologically induced seizures. In mice heterozygous for Scn1aand Scn8amutations, there was a reduction in spontaneous seizures and an increase in seizure threshold that partially compensated for the Scn1amutation. The magnitude of this effect was less than that of the altered background strain mentioned previously herein, but SCN8Amay be part of the group of genes that modify the severity and outcome of Dravet syndrome.

In the 7 years since mutations in SCN1Awere first described, much has been learned about this gene, its related phenotypes, and the pathologic abnormalities of the diseases it causes. Particularly, the understanding of Dravet syndrome has been expanded by the clinical genetics of SCN1A. Recognition of the various presentations of SCN1A-related epileptic encephalopathy is important because it will enable earlier, more aggressive treatment that may improve long-term prognosis. From the perspective of basic science, the development of transgenic mice has produced a model that gives novel and important insights into the pathogenesis of epilepsy. Interneuronopathy offers an unexpected and coherent explanation of Dravet syndrome in humans associated with haploinsufficiency and new information for understanding the function of missense mutations.

Dravet syndrome is a catastrophic disease, and recent advances lead us significantly closer to understanding and effectively treating this condition. Mouse models, by recapitulating the seizures and motor disturbance of Dravet syndrome, offer the possibility of major advances in therapeutics to improve seizure control and developmental outcome.

Correspondence:Ingrid E. Scheffer, MBBS, PhD, Epilepsy Research Centre and Department of Medicine, University of Melbourne, Austin Health, Heidelberg, Victoria 3081, Australia (i.scheffer@unimelb.edu.au).

Accepted for Publication:March 28, 2008.

Author Contributions:Both authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mullen and Scheffer. Analysis and interpretation of data: Scheffer. Drafting of the manuscript: Mullen and Scheffer. Critical revision of the manuscript for important intellectual content: Scheffer. Study supervision: Scheffer.

Financial Disclosure:None reported.

Additional Contributions:Samuel Berkovic, MD, provided critical review of the manuscript.