Article Text

PDF

Genotype phenotype associations across the voltage-gated sodium channel family
  1. Andreas Brunklaus1,
  2. Rachael Ellis1,2,
  3. Eleanor Reavey1,2,
  4. Christopher Semsarian3,4,
  5. Sameer M Zuberi1,5
  1. 1The Paediatric Neurosciences Research Group, Royal Hospital for Sick Children, Glasgow, UK
  2. 2Molecular Diagnostics, West of Scotland Genetic Services, Southern General Hospital, Glasgow, UK
  3. 3Agnes Ginges Centre for Molecular Cardiology, Centenary Institute, Sydney, Australia
  4. 4Sydney Medical School, University of Sydney, Australia
  5. 5School of Medicine, College of Medical, Veterinary & Life Sciences, University of Glasgow, UK
  1. Correspondence to Dr Sameer M Zuberi, The Paediatric Neurosciences Research Group, Royal Hospital for Sick Children, Glasgow G3 8SJ, UK; sameer.zuberi{at}nhs.net

Abstract

Mutations in genes encoding voltage-gated sodium channels have emerged as the most clinically relevant genes associated with epilepsy, cardiac conduction defects, skeletal muscle channelopathies and peripheral pain disorders. Geneticists in partnership with neurologists and cardiologists are often asked to comment on the clinical significance of specific mutations. We have reviewed the evidence relating to genotype phenotype associations among the best known voltage-gated sodium channel related disorders. Comparing over 1300 sodium channel mutations in central and peripheral nervous system, heart and muscle, we have identified many similarities in the genetic and clinical characteristics across the voltage-gated sodium channel family. There is evidence, that the level of impairment a specific mutation causes can be anticipated by the underlying physico-chemical property change of that mutation. Across missense mutations those with higher Grantham scores are associated with more severe phenotypes and truncating mutations underlie the most severe phenotypes. Missense mutations are clustered in specific areas and are associated with distinct phenotypes according to their position in the protein. Inherited mutations tend to be less severe than de novo mutations which are usually associated with greater physico-chemical difference. These findings should lead to a better understanding of the clinical significance of specific voltage-gated sodium channel mutations, aiding geneticists and physicians in the interpretation of genetic variants and counselling individuals and their families.

Statistics from Altmetric.com

Introduction

Voltage-gated sodium channels (SCN) are vital for the functioning of excitable tissue across the nervous system, heart and muscle, with variants in SCN genes associated with epilepsy, cardiac conduction defects, skeletal muscle channelopathies and peripheral pain disorders. As genetic testing is becoming more readily accessible, geneticists, often in partnership with neurologists and cardiologists, are increasingly asked to comment on the clinical significance of specific SCN gene mutations. Recent advances in the understanding of genotype–phenotype associations highlight the importance of knowing the pathophysiological basis of sodium channel functioning as this can unlock the door to the interpretation of specific mutation findings and aids genetic counselling.

Voltage-gated SCN are responsible for the initiation and propagation of action potentials in neurons and most electrically excitable cells.1 Upon depolarisation of the cell membrane, SCN are activated and inactivated within milliseconds, initiating the rapid influx of sodium ions and regulating the voltage-dependent channel inactivation.

Voltage-gated SCN are transmembrane complexes that consist of an α subunit and one or more auxillary β subunits.2 The α subunits are large proteins of approximately 2000 amino acid residues that fold into four homologous domains (I–IV). These domains are similar to one another, contain six α-helical transmembrane (S1–S6) segments and are highly conserved throughout evolution (see online supplementary figure S1a). The S4 segment harbours the voltage sensor which has positively charged amino-acid residues in every third position. These function as gating charges and start channel activation once the cell membrane has been depolarised. The pore consists of the ion-selectivity filter, a narrow re-entrant loop between segments S5 and S6, and the inner pore made out of four S6 segments. The transmembrane helices are connected by small extracellular and intracellular loops, with the four homologous domains being linked by large intracellular loops. The loop connecting domains 3 and 4 is highly conserved and plays an important role in channel inactivation forming the inactivation gate.2 The four homologous domains form a sodium selective pore, through which sodium ions flow during channel opening along the existing sodium gradient across the membrane, which is perpetually being restored by the sodium/potassium ATPase.

Every sodium channel α subunit is linked to one or more auxiliary β subunits (β 1–4). These consist of a single transmembrane protein with an extracellular immunoglobulin-like loop and an intracellular C terminus. The additional β subunits play an important role in the kinetics and voltage-dependence of channel gating and their presence is necessary for full channel functioning. They facilitate channel localisation, interaction with cell-adhesion molecules, extracellular matrix and intracellular cytoskeleton.2

From an evolutionary point of view, the sodium channel family is the youngest among the voltage-gated ion channels, being very closely related to calcium channels that also contain four homologous domains. Their structure is so similar that a change of four amino acids in the selectivity filter renders SCN selective for calcium.3 Calcium channels are most likely to have originated from gene duplication of potassium channels which are tetramers of identical single-domain subunits joined together to form a central ion-conducting pore.2 ,4

Subsequently, the members of the voltage-gated sodium channel family have diverged into nine isoforms, each encoding a different tissue-specific channel. However, their basic structure comprising of four homologous domains with voltage sensor and pore region remains largely preserved in all nine channels, sharing up to 85% of the amino acid sequence similarity between them. Major expression sites and associated disease mutations have been identified in skeletal muscle, cardiac channels and the neuronal SCN (table 1).

Table 1

The voltage-gated sodium channel family—genotypes and phenotypes

Given the common evolutionary origins and highly conserved nature of voltage-gated SCN, we reviewed the evidence for genotype–phenotype associations across the known voltage-gated sodium channel-related diseases of excitable tissue in central nervous system, peripheral nervous system (PNS), heart and muscle. We reviewed over 1300 sodium channel mutations and evaluated whether clinical presentation is associated with specific genetic changes that may be important in predicting disease outcome. We examined whether the mutation class or nature of amino acid substitution correlated with the epilepsy phenotype, using the Grantham Score (GS) as a measure of physicochemical difference between amino acids. We calculated whether changes occur randomly across the protein or whether they can be predicted and we reviewed the clinical implications of genotype–phenotype associations (e-methods in online supplementary material).

Subsections

SCN1A

Mutations in the gene encoding the α1 subunit of the voltage-gated sodium channel (SCN1A) are associated with several epilepsy syndromes, ranging from relatively mild phenotypes found in families with genetic epilepsy with febrile seizures plus (GEFS+) to severe myoclonic epilepsy of infancy (SMEI) also known as Dravet syndrome.7–10 To date, more than 750 sequence variants of the SCN1A gene have been identified.11

Dravet syndrome typically presents around 6 months of age in children who were previously well, with prolonged, febrile and afebrile, generalised clonic or hemiclonic epileptic seizures. Myoclonic, focal, atypical absence and atonic seizures appear between the ages of 1 and 4 years.12 The epilepsy is usually not responsive to standard antiepileptic medication and affected children develop an epileptic encephalopathy with cognitive, behavioural and motor impairment. Around 70%–80% of children with classical Dravet syndrome have point mutations or gross rearrangements in the SCN1A gene.11 Over half the mutations associated with a Dravet phenotype are truncating in nature and cause loss of function, proving that haploinsufficiency of SCN1A is pathogenic. By contrast, almost all the mutations associated with GEFS+ phenotypes are missense.13 Less severe phenotypes are more common in individuals with a missense mutation, which does not necessarily lead to a complete abolition of protein function.14 Patients with truncating mutations, in contrast, have an earlier onset of seizures indicating worse disease compared with those with a missense mutation.11

Common functional biophysical abnormalities observed in experimental systems of mutant Nav1.1 have been an impaired channel inactivation leading to increased persistent current and a decreased entry into and increased rate of recovery from slow inactivate state.15 ,16

In vivo studies on mouse models of Dravet syndrome have shown that loss-of-function mutations in Nav1.1 cause selective loss of excitability of hippocampal GABAergic interneurons. This, in turn, may result in hyperexcitability of pyramidal neurons leading to epilepsy in patients with Dravet syndrome.17 ,18 The Dravet mouse model further demonstrated that cerebellar Purkinje neurons were affected, resulting in ataxia in Scn1a mice and reduced expression of GABAergic interneurons in the forebrain of mice resulted in abnormal behaviour, suggesting that a widespread loss of sodium currents in GABAergic neurons might underlie the many difficulties patients with Dravet syndrome display.18–20 Recent evidence, however, has demonstrated that abnormalities in excitatory neurons are also important, as iPSC-derived neurons from patients with Dravet syndrome show increased sodium currents in both bipolar-shaped and pyramidal-shaped neurons; and Nav1.1 haploinsufficiency in excitatory neurons improved seizure-associated sudden death in a Dravet syndrome mouse model.21–23

It has been suggested that GEFS+ and SMEI may be caused by a continuum of mutational effects that selectively impair firing of GABAergic inhibitory neurons.19 This leads the way to targeted pharmacological treatment including drugs that increase GABAergic transmission such as benzodiazepines and avoiding sodium channel blocking agents that might exacerbate symptoms.

Truncating mutations have been shown to be distributed evenly across the gene, whereas missense mutations are more likely to be located within the four conserved homologous domains (figure 1), most frequently in segments S4 and S6 (see online supplementary figure S1b). From detailed descriptions of structure and function of voltage-gated SCN, we know that segment 6 forms the inner pore lining the sodium channel.24 Although segments 1–4 form the voltage sensor, it is particularly segment 4 that accounts for most of the gating charge.2 Differences in physico-chemical differences between amino acids measured by the median GS25 are highest in the Dravet group (94; semi-IQR=38), lower in the GEFS+ group (81; semi-IQR=35) and lowest in the polymorphism group (58; semi-IQR=18) and across different species (45; semi-IQR=18) (figure 2). Studies in several inherited disorders (cystic fibrosis, glucose-6-phosphate dehydrogenase deficiency, tuberous sclerosis) have shown a correlation between the degree of change in an amino acid substitution and the likelihood of clinical presentation and average values of non-disease GS rarely reach beyond 60, compared with average disease GS between 80 and 110.26 ,27 Children with a high GS have an earlier onset of seizures and are more likely to have had myoclonic seizures than those with a low GS. Significant polarity changes are more frequently seen in the voltage sensor and pore region of the SCN1α protein. Any significant change in polarity in these areas may alter voltage sensing or lead to an alteration of charge in the pore region.11 ,28 It appears likely that SCN1A phenotypes are not determined by chance, but are, in part, determined by defined physico-chemical changes affecting a specific location in the protein structure.

Figure 1

Frequency of missense and truncating mutations in homologous versus non-homologous domains among 754 SCN1A and 486 SCN5A mutations. The average mutation frequency was defined as 1 for illustration purposes. Homologous domains are highly conserved transmembrane segments that contain the main ion transport sequence. Non-homologous domains consist of large intracellular loops linking the four homologous domains plus N-terminal and C-terminal. χ22 test; df, degrees of freedom; p, significance level, (adapted from Zuberi et al [11]).

Figure 2

Comparison of Grantham scores among Dravet syndrome, genetic epilepsy with febrile seizures plus (GEFS+), polymorphisms and across species. Median Grantham scores of missense mutations associated with Dravet and GEFS+ phenotypes compared with the Grantham Score of missense polymorphisms and across different species/orthologs; p value derived using Mann–Whitney U test; n=number of individuals, (adapted from Zuberi et al [11]).

It has been suggested that the different phenotypes associated with SCN1A mutations result from a spectrum of increasing severity of loss-of-function mutations of Nav1.1 channels, increasing impairment of action potential firing in GABAergic inhibitory neurons and excitatory neurons.19 ,22 The level of impairment might be anticipated by the physico-chemical property changes—the mildest phenotypes being associated with polymorphisms; moderate to severe phenotypes being associated with more significant missense mutations; and truncating mutations being associated with the most severe phenotypes.

The incidence of sudden unexpected death (SUD) in epilepsy (SUDEP) has been reported as significantly higher in Dravet syndrome compared with other epilepsy syndromes.12 However, there is currently no evidence to suggest that the nature of a mutation increases the risk of SUDEP, and it may be that other genetic factors contribute to the rise in SUDEP risk.

The genotype–phenotype relationships in SCN1A have to be interpreted in the context of a gene in which it is known that a particular mutation may be associated with different phenotypes within the same family.29 This variability has also been observed in the mouse-model and might be accounted for by other genes, post-translational processes and protein–protein interaction.18

Recent work has demonstrated that homozygous SCN1B loss-of-function mutations are associated with Dravet syndrome, most likely by altering Nav1.1 expression on the cell surface.30 A possible modifying effect has been shown for Scn8a: mice that were heterozygous for both Scn1a and Scn8a had a higher threshold for drug-induced seizures and lived longer than mice that were heterozygous for Scn1a only.31

SCN2A

The majority of SCN2A mutations have been associated with mild epilepsy phenotypes, such as benign familial neonatal-infantile seizures (BFNIS), and GEFS+,32 however, the phenotypic spectrum of SCN2A mutations also includes more severe phenotypes such as Ohtahara Syndrome, early onset epileptic encephalopathies (EOEE) and Dravet syndrome.33 ,34 All known cases of BFNIS are dominantly inherited SCN2A missense mutations, whereas, severe phenotypes are mainly associated with a de novo missense or truncating mutations.33 ,34

BFNIS is a self-limiting disorder presenting with mainly afebrile secondarily generalised partial seizures often starting in the neonatal period. Seizure frequency can vary from few attacks not requiring treatment up to many clusters per day, however, most cases remit spontaneously by 1 year of age.32

BFNIS is associated with a range of biophysical defects in Nav1.2 function including loss of function, gain of function and decreased channel availability.35 Using whole-cell patch-clamp recording of heterologously expressed Nav1.2, three BFNIS mutations (R1319Q, L1330F and L1563V) exhibited a significant reduction in cell surface expression compared with wild type (WT), suggesting loss of function.35 Insights into the transient nature of this self-limiting seizure disorder have recently been gained from mouse models showing that developmental expression of Nav1.2 in principal neurons of hippocampus and cortex diminishes with time. During maturation, these are replaced by the dominant channel type Nav1.6 rescuing channel function.36

We calculated the physico-chemical difference between amino acid changes for the 11 known BFNIS missense mutations and found a low median GS of 29 (semi-IQR=11).34 This suggests that the amino acid substitutions observed in BFNIS missense mutations are between physico-chemically similar amino acids and are, therefore, less likely to cause severe disease. SCN2A mutations in milder phenotypes, such as BFNIS and GEFS+, were all inherited, whereas those associated with severe phenotypes arose de novo.33 ,34 ,37 We calculated the physico-chemical difference for the 18 known SCN2A mutations associated with Ohtahara Syndrome, EOEEs and Dravet syndrome and found a median GS of 62 (semi-IQR=20).33 The GS of the group with epileptic encephalopathies was significantly higher compared with the GS of the BFNIS group (p=0.031; Mann–Whitney U test). Only one non-sense SCN2A mutation (R102X) was seen in a patient with sporadic intractable childhood epilepsy and severe mental decline, and biophysical studies revealed a truncated channel protein exerting a dominant negative effect on channel function.37

Overall, there is evidence from functional work, mutation class, inheritance pattern as well as physico-chemical properties suggesting a correlation between genotype and phenotype in SCN2A mutations.

SCN3A

Abnormal expression of the Nav1.3 Na+ channel has been known to play a role in pain pathways after spinal cord injury involving dorsal horn and thalamic neurons.38 An epilepsy-associated SCN3A variant (K354Q) was found in a patient who presented with infrequent complex partial seizures from the age of 2 years and normal intellect.39 The SCN3A variant was inherited from the patients’ heterozygous father, with no history of seizures, possibly suggesting a reduced penetrance of the phenotype.

Functional studies have shown that the K354Q variant increases ramp and persistent current of Nav1.3 and induces hyperexcitability in hippocampal neurons in rodent models.40 However, global Nav1.3 KO mice that were investigated for the role Nav1.3 channels play in chronic pain were healthy, grew as well as their WT littermate controls, and were not reported to have spontaneous seizures.41

An additional four novel SCN3A missense variants (R357Q, D766N, E1111K and M1323V) were identified among a cohort of paediatric patients with focal epilepsy of unknown cause. Functional work demonstrated that a common feature shared by all variant channels was an increased current activation in response to depolarising voltage ramps and SCN3A may contribute to neuronal hyperexcitability and epilepsy.42 Four out of these five missense variants (80%) were located in homologous domains, and the median GS was 43 (semi-IQR=16). The clinical significance of SCN3A variants in relationship to epilepsy remains uncertain.

SCN4A

Mutations in the voltage-gated sodium channel Nav1.4 are associated with a range of hereditary skeletal muscle channelopathies including potassium-aggravated myotonia (PAM), paramyotonia congenita (PMC), hyperkalemic periodic paralysis (HyperPP), hypokalemic periodic paralysis (HypoPP) and congenital myasthenic syndrome.43 Over 40 different SCN4A missense mutations have been described, nearly all of which are inherited in an autosomal dominant pattern and there are no reports of truncating mutations. Main symptoms are myotonia caused by uncontrolled repetitive muscle fibre discharges and muscle weakness due to muscle fibre inexcitability. 43

In PAM, patients feel stiff after periods of rest which resolves with exercise (warm-up phenomenon), however, often develop stiffness 10–30 min after strenuous activity that can be painful and disabling.44 Three mutations located in identical positions of the human muscle sodium channel Nav1.4, G1306A, G1306V and G1306E, cause different phenotypes of PAM. The greater the physico-chemical difference from the original amino acid G1306, the more severe are the clinical symptoms. A change from Glycine to Alanine, with a GS of 60, results in a benign, often ‘subclinical’ form of myotonia, whereas a change to valine (GS 109) or glutamic acid (GS 98) causes moderate to severe myotonia.45 Electrophysiological studies in human embryonic kidney cells (HEK293) demonstrated that according to disease severity, G1306A showed the mildest disturbance of channel functioning compared with G1306V and G1306E that showed more severe slowing of fast inactivation and acceleration of recovery for inactivation in the heterologous expression system.45

PMC is characterised by cold-induced muscle stiffness that increases with continued activity and that can evolve to flaccid weakness. The majority of PMC mutations are situated in protein parts relevant for channel inactivation and functional work demonstrated slowed fast inactivation and accelerated recovery from the inactivated state leading to gain of function changes.46 There is considerable heterogeneity of phenotypic expression in humans as, for example, the same A1156T mutation has been observed in different patients with completely different phenotypes, including PMC, HyperPP and pure myotonia. This variability in phenotypes among patients with identical SCN4A mutations indicates that the clinical expression of distinct mutations may be subject to the genetic background and other epigenetic factors.47 Treatment has been available in the form of local anaesthetics and antiarrhythmic drugs which effectively relieve stiffeness in PAM and prevent muscle stiffness and weakness in PMC.48

Patients with HyperPP present with increased serum potassium during episodes of weakness. Attacks can be triggered by potassium-rich food, cold environment, stress, fasting or rest after exercise. Mutations are mainly situated within the intracellular loops between domains and have an effect on structures that regulate fast inactivation.49

By contrast, HypoPP is associated with hypokalaemia provoked by carbohydrate-rich and sodium-rich food, and acute weakness attacks can be treated with potassium. Nearly all mutations neutralise a positively charged amino acid in one of the voltage sensors situated in DI, DII and DIII.43 Recent work has shown that mutations in the voltage sensor of Nav1.4 can impact on ion permeation and blockage of the gating pore of SCN4α channels HypoPP.50

Recently, severe neonatal episodic laryngospasm, a form of life-threatening myotonia, has been associated with SCN4A mutations (G1306E, T1313M and A799S).51 ,52 Functional work via patch-clamp techniques showed that the A799S mutation promoted the channel open state with sustained activity leading to hyperexcitability of laryngeal muscles which could be lethal during infancy.52

We calculated the median GS of 47 SCN4A missense mutations to be 58 (semi-IQR=36) which is just on the threshold of disease-causing levels, demonstrating that SCN4A mutations are not associated with major physico-chemical differences between amino acids.43 Given that all the SCN4A mutations are dominantly inherited this makes a complete loss of function unlikely. Interestingly, in 2010, the first cases of homozygous patients for SCN4A mutations (I1393T, R1132Q) were reported. These individuals showed much more severe clinical features and compound muscle action potentials than heterozygous patients, confirming that presence of 100% defective ion channels in homozygous individuals account for the most severe phenotype.53

Experimental work clearly shows that within the group of skeletal muscle channelopathies, different disease subtypes are associated with distinct mutation loci in the SCN4A gene that lead to specific changes resulting in gain or loss of function. There is evidence that distinct physicochemical changes as well as specific mutation locations affecting protein structure partly determine clinical presentations.

SCN5A

Nav1.5 encoded by the SCN5A gene, is the main cardiac sodium channel and can be found in the sarcolemma of atrial and ventricular myocytes and the cardiac autonomous nervous system, where it is involved in pacemaker activity of the sinoatrial and atrioventricular node. Hence Nav1.5 plays a crucial role in cardiac excitability and the conduction of electrical impulses through the heart.

Mutations in SCN5A are associated with severe arrhythmias some of which can be life-threatening leading to sudden unexplained death.54 Functional work on heterologous expression systems, such as HEK-293 and on mouse models, have illustrated that there are two main disease-causing mechanisms in SCN5A: (1) Gain-of-function mutations cause a rise in sodium influx leading to delayed cardiac repolarisation, longer action potential duration and result in long QT syndrome type 3 (LQT3) and (2) Loss-of-function mutations diminish sodium influx leading to reduced cardiac excitability, slower electrical conduction velocity and result in Brugada syndrome (BS), sick sinus syndrome (SSS), progressive cardiac conduction disease (PCCD) and overlap syndromes.55 ,56 In cases of SUD, where no definitive cause is identified, postmortem evaluation and genetic testing now play key roles in identifying the genetic cause of death (ie, the ‘molecular autopsy’).57

Most SCN5A mutations (∼250) have been found in patients with BS presenting with ventricular tachyarrhythmias, persistent ST segment elevation on ECG and high risk of sudden cardiac death.58 Episodes mainly occur at rest or during sleep and can be triggered by fever suggesting that gating properties of the cardiac sodium channel may be temperature dependent. Treatment options in BS have been limited and consist of avoidance of known triggers (fever) or implantation of a cardioverter defibrillator (ICD).59

Functional work at biophysical level revealed a number of mechanisms leading to sodium channel loss-of-function contributing to BS, including decreased Nav1.5 expression from trafficking defects, expression of non-functional channels or a change in gating properties due to a shift in the voltage dependence of activation and/or inactivation.59

Approximately 20% of patients affected by BS have a SCN5A mutation. Two-thirds of those are missense and the remainder truncating mutations (13% frame shift, 11% nonsense, 7% splice site and 3% in-frame ins/del). Comparison of patients with truncating mutations with those with functionally less severe missense mutations revealed that subjects with truncating mutations had more syncopes and longer PR and QRS intervals after drug provocation testing. Furthermore, the proportion of families in whom sudden cardiac death had occurred in a first-degree relative was almost twice as high in those families with a truncation mutation compared with those with a missense mutation.60 The disease penetration in families with BS has been very variable, and individuals harbouring identical mutations can have entirely different phenotypes. Similarly, affected family members might not carry the familial disease-causing mutation, highlighting the role of the genetic background in the pathophysiology of BS.61 Recently, a number of genetic modifiers have emerged including the common polymorphism H558R that, if it co-occurred in carriers of a SCN5A mutation, was found to improve ECG characteristics and resulted in a milder phenotype.62

There is a considerable overlap between BS and PCCD which leads to gradually worsening fibrosis of the conduction system. Both diseases are caused by loss-of-function SCN5A mutations and the same SCN5A mutation can either cause PCCD alone or PCCD combined with BS as an overlap syndrome.63

LQT syndrome is a cardiac conduction disease characterised by QT prolongation on ECG and ventricular tachyarrhythmias, which carries an increased risk of sudden cardiac death particularly in young people.59 Most of the LQT subtypes are caused by potassium channel disorders. However, 10%–15% of cases are linked to a SCN5A mutation, nearly all of which are missense mutations, and define the subtype LQT type 3. Affected individuals usually present with arrhythmias at rest or during sleep when the heart rate is low and medical treatment includes sodium channel blockers and a prophylactic ICD.64 Up to 10% of sudden infant death syndrome cases have been associated with genetic variants in LQTS genes, half of which were SCN5A variants and mutations.65 SCN5A mutations have also been associated with SUDEP suggesting a link between SUDEP, mutations in ion channel genes and familial LQTS.66 Recently, SCN5A mutations were found to be associated with arrhythmic dilated cardiomyopathy (DCM) suggesting that disruption of voltage-sensing may lead to DCM67; and SCN5A variants have further been associated with familial atrial fibrillation.68

We mapped 486 SCN5A missense and truncating mutations from a publicly available SCN5A database (http://www.fsm.it/cardmoc),69 which are associated with BS, LQT3, PCCD and SSS, across the protein and analysed whether their distribution occurred randomly or whether mutation clusters could be identified; for methods refer to ref.11. Missense mutations occurred more than twice (2.3 times) as often in the four highly conserved homologous domains compared with the non-homologous linker regions. By contrast, truncating mutations were only 1.3 times more likely to occur in the homologous domains compared with the linker regions (χ2=4.81; df = 1; p=0.028; figure 1).

These results confirm that SCN5A missense mutations do not occur randomly but affect specific areas of the Nav1.5 channel that are important for channel function. Truncating mutations occurred with near-similar frequency across transmembrane and linker regions and may lead to abolition of protein function regardless of their position.

We further examined whether the distribution of missense mutations would vary depending on the clinical phenotype (see online supplementary figure S2). We found that BS mutations, that cause enhanced slow inactivation, are mainly clustered in the S1–2 to S2–3 areas and the S5 to S5–6 loop segments that form the ion-selective filter at the extracelluar end of the pore.24

By contrast, mutations associated with LQT3, that cause abnormal sustained non-inactivation, are primarily situated in proximity of the voltage sensor in the S4 segment, the S6 segment which forms the inner lining of the pore and the C terminus. These are regions that promote and stabilise fast inactivation.55 Not a single mutation had been located in the large ion-selectivity filter (S5–6 segment).

It appears that depending on the phenotype, mutations are clustered in specific regions affecting specific channel function and, thereby, contributing to the phenotype. The median GS for 250 missense mutations associated with BS was higher (median = 81; semi-IQR=33) than the one associated with 75 LQT3 syndrome mutations (median = 56; semi-IQR=35), however, this difference did not reach statistical significance. Nevertheless, the question arises whether BS which is associated with truncating loss-of-function mutations and higher GS might represent a more severe phenotype.

Several regulatory proteins have been identified such as Glycerol 3 Phosphate Dehydrogenase 1-Like Protein, which is linked to the intracellular trafficking of Nav1.5 to the sarcolemma, Caveolin-3 whose likely function is exerted via adaptor proteins such as dystrophin or ankyrin and α1-Syntrophin, which plays a vital role in controlling peak and persistent INA via sodium channel nitrosylation.64

Overall, there is good evidence that the phenotype in cardiac sodium channelopathies is related to the mutation type and to a specific location in the protein structure with important modulating effects exerted by associated proteins. However, one cannot predict the exact biophysical impact of a particular mutant based on the physico-chemical properties alone and in vitro/vivo experiments remain an invaluable tool for the detailed characterisation of a specific mutation. These functional expression models have greatly enhanced our understanding of the pathophysiology of cardiac sodium channelopathies and allowed specific approaches in the clinical management of affected patients.59

SCN8A

The Nav1.6 channel, encoded by the SCN8A gene, can be found widespread across the central and PNS and is localised at dendrites, synapses, the axon initial segment and the nodes of Ranvier.70 Mice harbouring cell-specific knockout of Scn8a in cerebellar purkinje neurons display signs of ataxia, tremor and impaired coordination, highlighting its importance on cerebellar function.71

There is evidence emerging from whole exome and genome sequencing studies identifying individuals with epileptic encephalopathy and/or intellectual disability associated with SNC8A mutations.72 A de novo missense mutation (Asn1768Asp) was detected in a patient with severe epileptic encephalopathy, early onset seizures, features of autism, intellectual disability, ataxia and sudden unexplained death in epilepsy. Functional studies confirmed the likely pathogenicity of this mutation and revealed a gain-of-function effect causing an increase in persistent sodium current and incomplete channel inactivation.73 To date, more than 10 de novo SCN8A missense mutations in epileptic encephalopathies have been identified and published.72 ,74 Nearly all of these (12 out of 13, 92%) are located in homologous domains, and the median GS was 89 (semi-IQR=35).

In a mouse model of SMEI, Scn8a has been shown to function as a genetic modifier by restoring normal seizure threshold in Scn1a mutants and prolonging the lifespan of Scn1a mutants in mice.31 The increase in seizure activity in Scn1a mutants due to the malfunction of inhibitory interneurons might have been rescued by the decrease in excitability of pyramidal cells in the hippocampus and cortex that is caused by a Nav1.6 dysfunction. These results provide evidence of how an epilepsy phenotype might not only be determined by a single genetic change, but can be modified by changes in other ion channels.

SCN9A

There are four pain disorders known to be caused by mutations in the SCN9A gene encoding the α subunit of the sodium channel Nav1.7. Loss of function mutations in SCN9A are associated with congenital insensitivity to pain (CIP),75 whereas, gain of function mutations are seen in individuals with erythromelalgia (IEM), paroxysmal extreme pain disorder (PEPD) and painful peripheral neuropathies (PPN).76

Nav1.7 is mainly expressed in dorsal root ganglion neurons and sympathetic ganglion neurons of the PNS and has an important role as ‘gatekeeper’ within the peripheral pain-signalling pathway.77

Individuals with CIP have a complete insensitivity to pain that leads to frequent painless injuries, fractures and burns, ultimately resulting in a shortened life span. All mutations are truncating in nature and randomly distributed across the SCN9α protein, and affected individuals are either homozygous or compound heterozygous for the mutation. When tested in heterologous expression systems, mutations do not produce a functional Nav1.7 channel, highlighting the crucial role it plays in pain signalling pathways.78

IEM is an autosomal dominant syndrome presenting with attacks of burning pain mainly affecting distal extremities that can be triggered by mild warmth, which first manifests in childhood and persists into adult life. Mutations are missense with a median GS of 58 (n = 13; semi-IQR=36), and are mainly located in the S4–S6 segments of the SCN9α protein. The gain-of-function effect is likely due to an activation shift in a hyperpolarising direction leading to hyperexcitability of affected neurons.79

Genotype–phenotype analyses revealed that mutations that produce smaller effects on sodium channel activation are associated with a smaller degree of dorsal root ganglion neuron excitability and later onset of clinical signs.80

PEPD, formerly called familial rectal pain syndrome, is an autosomal dominant disease with onset as early as the neonatal period presenting with mainly autonomic symptoms, such as skin flushing, bradycardia and rectal, ocular, or jaw pain. Attacks can be triggered by defecation, cold, or emotion, and episodes might respond to treatment with carbamazepine.81 PEPD mutations impair fast inactivation of Nav1.7 and thereby produce a prominent increase in persistent current. All mutations are missense with a median GS of 89 (n = 9; semi-IQR=40) and are mainly located in the intracellular linker areas between domains three and four that are essential for fast inactivation.

More recently, gain-of-function mutations in Nav1.7 have been associated with PPN such as idiopathic, painful, small-fibre neuropathy.82

Overall, there is good evidence of genotype–phenotype associations in that loss-of-function mutations in Nav1.7 are distributed randomly and produce insensitivity to pain, whereas gain-of-function mutations cause severe pain and are clustered in specific areas. IEM mutations are mainly located intramembranously leading to hyperpolarisation and PEPD mutations are located in areas responsible for fast inactivation; both being associated with moderate to severe GS.

SCN10A

Nav1.8 is mainly expressed in dorsal root ganglia and trigeminal sensory neurons.83 Mice harbouring Scn10a mutations show diminished responses to inflammatory pain due to a lack of upregulation of Nav1.8 compared with the WT.84 Nav1.8 channels further support action potential conduction at low temperatures and might play a role in IEM. Recently, mutations in Nav1.8 have been associated with idiopathic PPN, and mutations were shown to enhance recovery from fast inactivation resulting in hyperexcitability of dorsal root ganglia.85

SCN11A

Nav1.9 is preferentially expressed in nociceptive neurons of the dorsal root and trigeminal ganglia,86 and mutations in Scn11a can lead to greatly reduced or absent thermal and mechanical inflammation-induced hyperalgesia in mice.87 Gain-of-function mutations in Nav1.9 have been identified as cause for painful peripheral neuropathy resulting in depolarisation of the resting membrane potential of dorsal root ganglia, enhancing spontaneous firing of these neurons.88

Conclusions

Reviewing over 1300 sodium channel mutations, there are many similarities in the genetic and clinical characteristics across the voltage-gated sodium channel family (table 1).

There is evidence that the level of impairment a specific mutation causes can be anticipated by the underlying physico-chemical property change. Polymorphisms that are at the threshold of disease are associated with the mildest phenotypes. Across missense mutations, those with higher GSs are associated with more severe phenotypes and truncating mutations underlie the most severe phenotypes. Missense mutations are clustered in specific areas and are associated with a distinct phenotype according to the position in the protein that is affected. Inherited mutations tend to be less severe than de novo mutations which are usually associated with greater physico-chemical difference.

However, there is great variability in the phenotypical expression of a sodium channel mutation among members of the same human or mouse family, indicating the importance of the genetic background. Several modifier genes have been identified that can influence the ultimate phenotype as shown in SCN1A and SCN5A. Mutations in the β subunits or associated membrane proteins, that are essential for sodium channel function, have equally been shown to affect the phenotype and might play a role as disease modifiers. The discovery of several regulatory proteins in SCN5α channel function might serve as an example for other voltage-gated SCN, and has implications for further research in this field.

Developmental upregulation and downregulation of specific ion channels can cause gain or loss of function effects that influence neuronal excitability. Leading the way to new treatment strategies, depending on the types of nerve cells that are mainly affected. In SCN1A-related epilepsy, for example, increasing GABAergic neurotransmission has been shown to reduce seizures by counteracting the impaired inhibitory response.

Many voltage-gated SCN are temperature-sensitive, including SCN1α, SCN4α, SCN5α and SCN9α and the prevention of trigger factors is of clinical relevance. Similarly, the avoidance of sodium channel blocking agents will prevent an exacerbation of seizures in SCN1A-related epilepsy, however, the same agent may be effective in the treatment of SCN9A-related PEPD—depending on the underlying defect in neurotransmission.

Skeletal muscle sodium channelopathies (SCN4A) are sensitive to alterations in serum potassium concentration, pH, cooling and carbohydrate loading. Local anaesthetics and antiarrhythmic drugs are used to reduce myotonia, and potassium-lowering diuretics help to prevent attacks in hyperPP.43

In SCN5A-related BS sodium channel-blocking drugs may aggravate ECG changes or trigger arrhythmias, as they reduce the inward depolarising current.59

These findings illustrate how knowledge of trigger factors and phenotypical consequences at biophysical level are important for rational therapeutic design and should lead to a better understanding of the clinical significance of voltage-gated SCN mutations. Based upon the similarities across the voltage-gated sodium channel family, findings might be transferable to other family members and facilitate research on treatment approaches and provide information on prognosis for individuals and their families in the future.

References

View Abstract
  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors AB had substantial contribution to conception, analysis and interpretation of data and drafted the article. RE, ER, CS and SMZ had substantial contribution to conception, interpretation of data and revised the article critically for important intellectual content.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.