SCN8A encodes one of the major voltage-gated sodium channels that regulate the initiation and propagation of action potentials in the nervous system. The sodium channel transmembrane proteins were first purified 30 years ago (Hartshorne and Catterall, 1981; Tamkun and Catterall, 1981) and cDNA clones were isolated shortly thereafter (Noda et al., 1986). The Scn8a gene, encoding the sodium channel Na_v1.6, was identified in 1995 by positional cloning of the mouse neurological mutant motor endplate disease (med) (Burgess et al., 1995) and by isolation of a novel sodium channel cDNA from rat brain (Schaller et al., 1995). SCN8A is a member of the gene family comprised of nine evolutionarily related sodium channels with specific roles in neurons and in skeletal muscle and cardiac muscle (Lopreato et al., 2001; Meisler and Kearney, 2005; Meisler et al., 2010; Zakon et al., 2011; Zakon, 2012).

Human SCN8A was mapped to chromosome 12q13 in 1998 (Plummer et al., 1998). The role of SCN8A in human disease was initially investigated by screening for mutations in families segregating inherited disorders such as ataxia, dystonia, and tremor (Trudeau et al., 2006; Sharkey et al., 2009a). These analyses identified only one family with an inherited mutation of SCN8A (Trudeau et al., 2006). Recently, the ability to sequence the entire exome or genome from an individual patient has made it possible to identification of de novo mutations in patients who do not have a family history of disease (Bamshad et al., 2011; Doherty and Bamshad, 2012; Need et al., 2012; Rauch et al., 2012). Using this technology, more than ten mutations of SCN8A have been described during the past year, in patients with epileptic encephalopathy and intellectual disability. This rapid progress indicates that mutations of SCN8A are a previously unrecognized cause of these and possibly other neurological disorders. Here we describe the recently discovered patient mutations and review the unique features of Na_v1.6 as a framework for understanding the pathological consequences of human mutations.

The first de novo mutation in SCN8A was discovered in 2012 by whole genome sequencing of a child with an early onset, debilitating epileptic encephalopathy. The clinical picture included developmental delay, features of autism, intellectual disability and ataxia (Veeramah et al., 2012). Afebrile seizures began at 6 months of age, and by 5 years EEG recordings detected short bursts of frontocentrally predominant generalized spike-wave activity, and bifrontal and multifocal spikes. Neither the parents nor an unaffected sibling carried the de novo mutation, p.Asn1768Asp, that was detected in the patient. The biophysical properties of the mutant channel include increase in persistent sodium current, incomplete channel inactivation, and a depolarizing shift in the voltage dependence of steady-state fast-inactivation (Veeramah et al., 2012). Current tracings of cells transfected with mutant channels reveal as much as 20% of maximal current remaining after 100 ms, compared with only 1% in cells transfected with wild-type channel (Figure 1). The elevated persistent current increases the likelihood of premature firing of neurons after subthreshold depolarization. Transfection of mouse hippocampal neurons with the mutant cDNA resulted in increased spontaneous and induced firing characteristic of neuronal hyperexcitability, consistent with the dominant expression of seizures in the heterozygous patient. Increased persistent current is also a common feature of mutations in the channel SCN1A that cause the epileptic encephalopathy Dravet Syndrome (Meisler and Kearney, 2005). Increased activity of Na_v1.6 has also been implicated in the seizure-prone Celf4^−/− mouse mutant (Sun et al., 2013) and suggested in fibroblast-derived neurons from patients with Dravet syndrome (Liu et al., 2013).

A second missense mutation, SCN8A-p.Leu1331Val, was identified by targeted resequencing of 65 candidate genes in 500 individuals with epileptic encephalopathy (Carvill et al., 2013). The proband presented with epileptic encephalopathy at 18 months of age, and the mutation was inherited from a mosaic father. Two additional mutations were identified in this study, p.Arg662Cys and p.Arg1872Gln, but family data regarding inheritance was not available (Carvill et al., 2013). The mutation, SCN8A-p.Arg223Gly, was recently identified in child that presented with epileptic encephalopathy at 6 months of age (Kovel et al., submitted). In a screen for de novo mutations in 264 patients with infantile spasms or Lennox-Gastaut syndrome, the SCN8A mutation p.Leu876Gln was found in a child with Lennox-Gastaut (Epi4K Consortium and Epilepsy Phenome/Genome Project, 2013). The locations of the epilepsy-associated mutations are indicated in Figure 2.

In 2006, we described the heterozygous loss-of-function mutation P1719RfsX1724 that segregated with cognitive deficits in a small family (Trudeau et al., 2006). Heterozygous children in this family were enrolled in special education classes, and heterozygous adults were unable to live independently. In 2012, Rauch and colleagues sequenced the exomes of 51 individuals with severe non-syndromic intellectual disability (Rauch et al., 2012). These patients were offspring of healthy, non-consanguineous parents and presented with intellectual disability, grossly normal motor function, and lack of syndrome-specific abnormality. The de novo missense variant p.Arg1617Gln in the voltage-sensing transmembrane segment of domain 4 of SCN8A was identified in one patient (Figure 2). Four additional de novo missense mutations in SCN8A have been discovered by exome sequencing of patients with intellectual disability (Figure 2). The limited functional data suggest that mutations causing increased channel activity are associated with seizures, while heterozygous loss-of-function of SCN8A predisposes to intellectual disability (Figure 2).

Over the past 18 years, fifteen mutant alleles of mouse Scn8a have been characterized. These include six spontaneous mutants, eight ENU-induced mutations, and one random transgene insertion (Figure 3) (Meisler et al., 2004). Several of these are null mutations with complete loss of Scn8a function. Homozygous null mice exhibit motor defects at 2 weeks of age, including ataxia and tremor, and do not survive beyond 3 weeks (Burgess et al., 1995; Kohrman et al., 1995). Homozygosity for severe hypomorphic alleles such as medJ and nmf58 is viable, but results in ataxia and tremor with progression to muscle weakness and dystonia. Homozygosity for five mildly hypomorphic alleles (medjo, jolting2J, tremorD, clth, 9J) results in tremor, ataxia and reduced body size. These observations suggest that mutations of human SCN8A may be found in the future in patients with movement disorders. Na_v1.6 is expressed at a low level in cardiac myocytes, and null mice have prolonged cardiac action potentials, suggesting a possible role in cardiac arrythmias (Noujaim et al., 2012). Homozygous knockout of Scn8a in Purkinje cells results in impaired learning in Morris Water Maze and eyeblink conditioning tests (Woodruff-Pak et al., 2006).

Mice that are heterozygous for loss-of-function mutations exhibit minor abnormalities such as spike-wave discharges suggestive of absence epilepsy (Papale et al., 2009), disrupted sleep architecture (Papale et al., 2010), and behavioral deficits including anxiety (McKinney et al., 2008). Haploinsufficiency of Na_v1.6 also reduces susceptibility to genetic- and chemically-induced seizures (Martin et al., 2007, 2010). Scn8a^med/+ and Scn8a^med−jo/+ heterozygotes have reduced susceptibility to flurothyl and kainic acid induced seizures, and the combination of one mutant allele of Scn8a with Na_v1.1 heterozygous or homozygous null mice results in extended lifespan and reduced seizure susceptibility. These observations suggest that reduced expression of Scn8a protects against seizures by decreasing neuronal excitability.

Direct evidence for the in vivo role of Na_v1.6 has been advanced by recordings from neurons from several different lines of Scn8a null and conditional null mice developed in our laboratory (Burgess et al., 1995; Levin and Meisler, 2004; Levin et al., 2006) (Table 1). Reduced repetitive firing is consistently observed in cerebellar Purkinje cells, granule neurons, trigeminal mesencephalic neurons, and retinal ganglion cells from Scn8a mutant mice (Raman and Bean, 1997; Raman et al., 1997; Van Wart and Matthews, 2006; Aman and Raman, 2007). Reduced persistent and resurgent current was observed in several types of neurons by multiple investigators (Table 1). In addition to induced firing, spontaneous firing is reduced in Purkinje neurons isolated from null mice (Khaliq et al., 2003). Overall, the work summarized in Table 1 demonstrates that Scn8a is a key determinant of neuronal excitability in vivo.

The role of Scn8a in regulating neuronal excitability may be related to three properties of Na_v1.6: its role in persistent and resurgent current, its voltage dependence of activation, and its subcellular localization at the axon initial segment (AIS), the site of initiation of action potentials. Persistent current is a steady-state sodium current that persists after firing and is involved in action potential initiation at membrane voltages near the threshold of firing (Crill, 1996; Smith et al., 1998; Rush et al., 2005; Osorio et al., 2010). Persistent current is important for generation of repetitive firing in neurons such as cerebellar Purkinje cells. In cerebellar Purkinje cells isolated from Scn8a null mice, persistent current was reduced by 70% compared with wild-type littermates (Raman et al., 1997). In tsA-201 kidney cells, the persistent current generated by Na_v1.6 is five-fold higher than that generated by Na_v1.2 (Chen et al., 2008). The differences in magnitude of persistent current in different types of neurons suggests that this property is modulated by neuron-specific factors (Rush et al., 2005; Chen et al., 2008). Mutations that further increase Na_v1.6 persistent current result in epileptogenesis (e.g., Figure 1) (Veeramah et al., 2012).

Resurgent current is a voltage- and time-dependent property in which depolarization after the initial action potential elicits a small, transient current (Hille, 2001). This rapidly reversible form of inactivation allows neurons to fire quickly and repetitively. Resurgent current is thought to contribute to spontaneous firing and multi-peaked action potentials in cerebellar Purkinje cells that are compromised in mutants lacking Na_v1.6 (Raman and Bean, 1997; Raman et al., 1997). The β4 sodium channel subunit is involved in generating resurgent current in cerebellar Purkinje neurons and cerebellar granule cell neurons, but the blocking factor appears to vary by neuron type (Raman and Bean, 2001; Grieco et al., 2005; Bant and Raman, 2010).

In transfected DRG neurons, there is a 15 mV leftward shift in voltage dependence of fast activation of Na_v1.6 compared to Na_v1.2, meaning that Na_v1.6 is more activated earlier during depolarization (Rush et al., 2005). Na_v1.6 is also less likely to inactivate at higher stimulation frequencies (20–100 Hz) (Rush et al., 2005). In transfected HEK-tsA-201 cells, Na_v1.6 displayed a more positive voltage dependence of slow inactivation, passing ~10% more current in the −35 to −25 mV range than Na_v1.2 (Chen et al., 2008). These features of Na_v1.6 contribute to the positive effect of Na_v1.6 on neuronal excitability.

The AIS is the membrane domain at the proximal end of the axon in which sodium channels are highly concentrated, electrical signals from the soma and dendrites are summed, and the threshold for action potential initiation is lowest (Royeck et al., 2008). The channel composition of the AIS appears to determine the firing threshold for different types of neurons (Lorincz and Nusser, 2008). Na_v1.6 is highly concentrated in the distal half of the AIS in many neurons, including cerebellar granule cells and cerebellar Purkinje cells (Van Wart and Matthews, 2006; Lorincz and Nusser, 2008; Royeck et al., 2008). In the absence of Na_v1.6, there is relocation of Na_v1.1 and Na_v1.2 to occupy the distal AIS (Van Wart and Matthews, 2006; Xiao et al., 2013). Cultured hippocampal CA1 pyramidal cells from Scn8a-null mice exhibit a 5 mV depolarizing (rightward) shift in the voltage dependence of activation, 60% reduction in persistent current, and 75% reduction in resurgent current (Royeck et al., 2008). This combination renders Scn8a null neurons less excitable than their wild type counterparts, as demonstrated by an 8 mV depolarizing shift in the spike threshold (Royeck et al., 2008).

In cortical pyramidal neurons, action potentials initiate at the distal part of the AIS, where sodium channel concentrations are highest (Van Wart et al., 2007; Kole and Stuart, 2008; Kole et al., 2008). The distal AIS in these cells contains predominantly Na_v1.6, while the proximal AIS contains predominantly Na_v1.2 (Hu et al., 2009). Step-depolarizations of patched neurons revealed that the activation threshold in the distal AIS was −55 mV, while the activation threshold in the proximal AIS closest to the soma was −43 mV (Hu et al., 2009), consistent with a role for Na_v1.6 in lowering the threshold of action potential initiation.

Action potentials are primarily directed down the axon, away from the soma, but backpropagation into the soma occurs at low frequency (Hu et al., 2009). Current injection into the distal AIS does not produce backpropagation, while current injection at the proximal AIS leads to detectable action potentials in the soma (Hu et al., 2009). Thus, localization of Na_v1.6 to the distal AIS is associated with a lower threshold for action potential initiation and direction of the action potential away from the soma. Overall, membranes containing Na_v1.6 are more excitable than those containing only Na_v1.1 and Na_v1.2, and loss of Na_v1.6 results in a higher threshold for initiation of action potentials (Van Wart and Matthews, 2006).

The SCN8A gene is located on human chromosome 12q13.13 (Plummer et al., 1998) and mouse distal chromosome 15 (Burgess et al., 1995). The 27 exons of SCN8A span 170 kb and encode a protein of 1980 residues (GenBank AF050736). The location of the amino acid residues within the 4 homologous domains and 24 transmembrane segments of Na_v1.6 is shown in Figure S1. Na_v1.6 protein is concentrated ~1,000-fold in two membrane domains, the AIS and the nodes of Ranvier of myelinated axons (Schaller and Caldwell, 2000; Boiko et al., 2001, 2003; Van Wart and Matthews, 2006; Van Wart et al., 2007; Lorincz and Nusser, 2008, 2010). Na_v1.6 is also present at lower abundance in non-myelinated axons, neuronal soma, and dendrites (Krzemien et al., 2000; Lorincz and Nusser, 2010). The full-length SCN8A transcript is highly expressed throughout the brain, with concentration in the cerebellum and olfactory bulb of the rat (Schaller and Caldwell, 2000).

Transcriptional regulation of sodium channel genes is not well characterized. The transcription start sites for Scn8a are located in noncoding exons 70 kb upstream of the translation initiation site (Drews et al., 2005). Exon 1c is highly conserved through evolution and includes potential binding sites for neuronal transcription factors Pou6f1/Brn5, YY1, and REST/NRSF (Drews et al., 2007). Exon 1c and upstream sequences are sufficient to drive neuron-specific expression of LacZ in transgenic mice (Drews et al., 2007).

SCN8A contains two pairs of mutually exclusive, alternative coding exons whose splicing regulates channel function. Exons 5N/5A and 18N/18A encode the S3–S4 transmembrane segments of domain I and domain III, respectively (Plummer et al., 1997). Exon 18N contains an in-frame stop codon and is only expressed in non-neuronal cells (Plummer et al., 1997) including glia (O'Brien et al., 2012a). The neuronal splice factors RBFOX1 and RBFOX2 can activate inclusion of exon 18A in neurons, resulting in neuron-specific expression of the full length, active channel (Gehman et al., 2012; O'Brien et al., 2012a; Zubovic et al., 2012). Splice enhancers and silencers in exons 18A and 18N also contribute to temporal and spatial regulation (Zubovic et al., 2012). Alternative polyadenylation sites are located 4 and 6.5 kb downstream from the translation termination site of Scn8a, generating full-length coding transcripts of 9 and 12 kb (Drews et al., 2005). Transcripts with the shorter and longer 3′ UTR are equally represented in brain RNA and are not known to be associated with specific functions.

The pharmacology of compounds that target voltage-gated sodium channels has recently been reviewed (Eijkelkamp et al., 2012). The epileptic encephalopathies described in this review could in principle be treated with specific inhibitors of Na_v1.6. However, the extensive sequence conservation among the neuronal and muscle sodium channels has made it difficult to develop drugs with specificity for a single channel. Two compounds with preferential effects on Na_v1.6 have been described. The tetrodotoxin derivative 4,9-anhydrotetrodotoxin inactivates Na_v1.6 expressed in Xenopus oocytes at concentrations that have minimal effects on six of the other channels (Rosker et al., 2007). The beta-scorpion toxin Cn2 also binds Na_v1.6 specifically (Schiavon et al., 2006); this compound enhanced resurgent current inducing a hyperpolarizing shift in voltage dependence of channel activation in Purkinje slices, indicative of channel activation, while in HEK cells the effect was inhibitory. We have generated a mouse model of epileptic encephalopathy carrying the SCN8A-p.Asn1768Asp mutation that may be useful for future evaluation of drug specificity and effectiveness in vivo.

Voltage-gated sodium channels are components of large, multi-protein complexes that vary between neurons and at specific subcellular domains. The known sites of protein interaction with Na_v1.6 are indicated in Figure 4. The N-terminus of Na_v1.6 interacts with the light chain of microtubule-associated protein Map1b (Mtap1b), and co-transfection increases current density in transfected cells via increased trafficking of Na_v1.6 to the cell surface (O'Brien et al., 2012b). Phosphorylation of Na_v1.6 by the stress-activated MAP kinase p38 facilitates binding of E3 ubiquitin ligases and channel degradation (Sudol and Hunter, 2000; Zarrinpar and Lim, 2000; Gasser et al., 2010). Protein kinases PKA and PKC have only a small effect on channel activity (Chen et al., 2008). Ankyrin G binds to the first intracellular loop of Na_v1.6 and other neuronal sodium channels (Srinivasan et al., 1988; Davis et al., 1996; Hill et al., 2008), and is essential for targeting and localization of Na_v1.6 to nodes of Ranvier (Gasser et al., 2012).

The intracellular fibroblast growth factors FGF11-FGF14 interact with Na_v1.6 and other voltage-gated sodium channels (Wittmack et al., 2004; Laezza et al., 2009; Shakkottai et al., 2009; Xiao et al., 2013). FGF13 interacts with the C-terminus in an isoform-dependent manner (Wittmack et al., 2004), which may allow specific sub-populations of neurons to fine-tune firing properties via alternative splicing of FGF13. Fgf14 null mice develop ataxia and ~80% of their cerebellar Purkinje cells lack repetitive firing (Shakkottai et al., 2009). The abundance of Na_v1.6 in the AIS is reduced in cerebellar Purkinje cells from Fgf14 null mice, suggesting that FGF14 plays a key role in the organization of a subunits in the AIS (Xiao et al., 2013).

The sodium channel subunits β1 to β4 are small single-transmembrane cell-adhesion molecule proteins that modulate current and surface expression of the α subunit (Patino and Isom, 2010). Studies of mice null for the β1 subunit (Scn1b^−/−) suggest that interaction between β1 and Na_v1.6 is required for function of Na_v1.6 at the distal AIS (Brackenbury et al., 2010). The β4 subunit has been implicated in the generation of resurgent Na_v1.6 current in Purkinje neurons (Grieco et al., 2005; Aman et al., 2009), but resurgent current was not generated by co-transfection of β4 and Na_v1.6 in HEK cells (Chen et al., 2008; Aman et al., 2009).

The calcium responsive protein calmodulin binds the IQ motif located in the C-terminus of Na_v1.6 (residues 1902–1912). Apo-calmodulin accelerates inactivation and Ca²⁺ increases excitability of Na_v1.6 (Herzog et al., 2003). The E3 ubiquitin ligase Nedd4 also binds to the C-terminus of Scn8a at a PXY motif (residues 1943–1945), and the PXpS/pTP motif in the first cytoplasmic loop (residues 551–554) (Abriel et al., 2000; Sudol and Hunter, 2000; Fotia et al., 2004; Ingham et al., 2004; van Bemmelen et al., 2004; Rougier et al., 2005). Both sites are necessary for Nedd4 binding and internalization of Na_v1.6 (Gasser et al., 2010). Ubiquitination of Na_v1.6 by Nedd4 is thought to target Na_v1.6 for degradation and may be part of the neuronal stress response.

These interactions are relevant to the genetics of neurological and psychiatric disorders, since proteins that bind Na_v1.6 may be considered candidate genes for the same disorders caused by mutations of Na_v1.6. Further, common variants of the interacting proteins may act as modifiers of the severity of SCN8A mutations in patients (Meisler et al., 2010; Meisler and O'Brien, 2012).

Na_v1.6 is a major sodium channel in human brain. The features of Na_v1.6 that influence neuronal excitability include contributions to persistent and resurgent neuronal currents, low threshold for excitation, and concentration in the AIS. Mutations of Scn8a in the mouse result in movement disorders including ataxia, dystonia, and tremor. Within the past year, de novo mutations of human SCN8A detected by exome sequencing have revealed a role for Na_v1.6 in epilepsy and intellectual disability. Hypoactivity and hyperactivity of Na_v1.6 are both pathogenic, but with different outcomes: haploinsufficiency is associated with impaired cognition (Trudeau et al., 2006; McKinney et al., 2008; Rauch et al., 2012) while hyperactivity can result in epilepsy (Veeramah et al., 2012). Analysis of additional mutants in the near future should provide insight into structure-function relationships of Na_v1.6 and the mechanisms of pathogenesis in neurological disease.