Ion channels in genetic and acquired forms of epilepsy

Abstract

Genetic mutations causing dysfunction of both voltage- and ligand-gated ion channels make a major contribution to the cause of many different types of familial epilepsy. Key mechanisms comprise defective Na⁺ channels of inhibitory neurons, or GABA_A receptors affecting pre- or postsynaptic GABAergic inhibition, or a dysfunction of different types of channels at axon initial segments. Many of these ion channel mutations have been modelled in mice, which has largely contributed to the understanding of where and how the ion channel defects lead to neuronal hyperexcitability. Animal models of febrile seizures or mesial temporal epilepsy have shown that dendritic K⁺ channels, hyperpolarization-activated cation channels and T-type Ca²⁺ channels play important roles in the generation of seizures. For the latter, it has been shown that suppression of their function by pharmacological mechanisms or in knock-out mice can antagonize epileptogenesis. Defects of ion channel function are also associated with forms of acquired epilepsy. Autoantibodies directed against ion channels or associated proteins, such as K⁺ channels, LGI1 or NMDA receptors, have been identified in epileptic disorders that can largely be included under the term limbic encephalitis which includes limbic seizures, status epilepticus and psychiatric symptoms. We conclude that ion channels and associated proteins are important players in different types of genetic and acquired epilepsies. Nevertheless, the molecular bases for most common forms of epilepsy are not yet clear, and evidence to be discussed indicates just how much more we need to understand about the complex mechanisms that underlie epileptogenesis.

Holger Lerche (left) is Clinical Director and Head of the Department of Neurology and Epileptology at the Hertie Institute of Clinical Brain Research at the University of Tübingen, Germany. His main research interest is to unravel the genetics and pathophysiology of inherited epilepsies and related paroxysmal disorders using a combination of genetic and neurophysiological tools. He is also interested in molecular ion channel function, their specific roles in the brain and their pharmacology. After graduating from the University of Munich (LMU), he worked as a postdoc in neurophysiology and as a resident and consultant in neurology and epileptology at the Institute of Applied Physiology and the Department of Neurology of the University of Ulm. He undertook clinical and research fellowships in Bonn/Germany, London/UK and Melbourne/Australia. Mala Shah (right) did her PhD at University College London (UCL, UK) under the supervision of Dr Dennis Haylett. She then obtained a Wellcome Prize Travel Research Fellowship to work in the laboratories of Professors Daniel Johnston at Baylor College of Medicine (Houston, USA) and David Brown at UCL (UK). She subsequently received a lectureship at UCL School of Pharmacy (UK) where she is currently a Reader in Neuroscience. Her research interests include understanding how voltage-gated ion channels activated at sub-threshold membrane potentials affect hippocampal and cortical cell excitability under physiological as well as epileptogenic conditions.

Introduction

The epilepsies are disorders of neuronal network excitability. They can be divided into two major groups. In the first group, which is called ‘symptomatic’, an acquired or inborn structural or metabolic defect of the brain can be identified as the underlying cause of the disease. These forms of epilepsies have a mainly focal origin meaning that the seizures start from a point around the structural lesion. The clinical presentation of the resulting epileptic seizures depends on the respective brain region in which the seizures start and spread, and can vary from light symptoms such as a strange feeling in the stomach or paresthesia in a certain body area, to loss of consciousness and severe convulsions. Typical examples for epileptogenic lesions are tumours, stroke or hippocampal sclerosis, the latter causing mesial temporal lobe epilepsy, one of the most frequent and often pharmacoresistant forms of focal epilepsy. An example of increasing clinical importance is given by epilepsies with antibodies directed against proteins involved in membrane excitability such as ion channels. The second group, termed ‘idiopathic’, is genetically determined and characterized by the lack of structural or other predisposing causes. Both focal and generalized forms of epilepsy can be caused by genetic defects and the resulting epileptic phenotypes can range from mild seizures occurring only in neonates or infants, to severe epileptic encephalopathies with mental retardation, pharmacoresistant epilepsy and other neurological symptoms. The most common disease entity is ‘idiopathic generalized epilepsy’ (IGE) comprising the well-known absence, myoclonic and primary generalized tonic–clonic seizures. The detection of mutations causing idiopathic forms of epilepsy has dramatically advanced our understanding about the pathophysiology in the last 15 years, which is one major topic covered in this review.

There are three main ways in which ion channels are known to be involved in epilepsy. Firstly, there are specific mutations in familial idiopathic epilepsies; secondly, there are specific antibodies in acquired seizure-related disorders; and thirdly, there are changes of ion channel expression and function associated with modification of seizure activity which may contribute to all forms of epilepsy. Here we review these rapidly developing areas using tables to provide more details. For the sake of brevity we will focus on disorders with the main symptom of epilepsy. Recent other developments, for example, increasing research into epileptic seizures in Alzheimer's disease or connections to autism, are not covered here.

Genetic defects in voltage-gated or ligand-gated ion channels

Table 1 lists gene mutations found in different epilepsies and Figure 1 shows the localization of some of the relevant channel proteins in different neuronal subtypes and compartments. Classical linkage methodology, on large pedigrees with rare monogenic syndromes, identified mutations in familial idiopathic focal epilepsies: for instance, the genes encoding nicotinic acetylcholine receptors in autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) or potassium channels (K_V7) in benign familial neonatal seizures (BFNS). In ‘sporadic’ cases, in which only one member of a family is affected, candidate gene approaches have been successful in identifying de novo pathogenic mutations, the most prominent being SCN1A nonsense mutations in Dravet syndrome A few mutations have also been identified in families with the most common IGEs, childhood and juvenile absence or myoclonic epilepsies, with different GABA_A receptor subunit defects being of particular importance (Table 1). However, the substantial complexity of ion channel gene variant profiles among individuals with epilepsy as well as unaffected controls precludes a simple monogenic channelopathy model in the majority of cases, and particularly IGE is considered to be a prototype of a complex genetic disorder in which many – both rare and common – genetic variations play a pathogenic role. Pathophysiological studies demonstrated that two key defects are: (i) a neuronal dysinhibition that can be caused both by loss-of-function defects in different subunits of the postsynaptic GABA_A receptor and presynaptic loss-of-function defects of the sodium channel Na_V1.1 expressed specifically in inhibitory interneurons, or (ii) dysfunction of axon initial segments, the neuronal structure in which action potentials are generated and in which many of the described channels (for example, Na_V1.2 sodium and K_V7 potassium channel subunits) are mainly localized. In addition, these clinically originated studies identified novel genes, defined their neuronal functions and sometimes established new physiological principles (such as Na_V1.1 as the major sodium channel in GABAergic interneurons). Moreover, K_V7 channels have proven to be a novel therapeutic target (reviewed by Weber & Lerche, 2008; Reid et al. 2009).

Table 1.

Genes and proteins mutated in idiopathic epilepsies and epileptic encephalopathies

	Abbreviation	Gene	Protein	References
Idiopathic focal epilepsies
Benign familial neonatal seizures	BFNS1/EBN1	KCNQ2	KV7.2 (K+ channel)	Biervert et al. 1998; Singh et al. 1998
	BFNS2/EBN2	KCNQ3	KV7.3 (K+ channel)	Charlier et al. 1998
Benign familial neonatal–infantile seizures	BFNIS	SCN2A	NaV1.2 (Na+ channel)	Heron et al. 2002; Berkovic et al. 2004
Autosomal dominant nocturnal frontal lobe epilepsy	ADNFLE	CHRNA4	α4 subunit (nACh) receptor	Steinlein et al. 1995
		CHRNB2	β2 subunit (nACh) receptor	De Fusco et al. 2000
		CHRNA2	α2 subunit (nACh) receptor	Aridon et al. 2006
Idiopathic generalized epilepsies and associated syndromes
Childhood absence epilepsy with febrile seizures	CAE+FS	GABRG2	γ2 subunit (GABAA receptor)	Wallace et al. 2001
Absence epilepsy and episodic ataxia	CAE+EA2	CACNA1A	CaV2.1 (Ca2+ channel)	Jouvenceau et al. 2001; Imbrici et al. 2004
Juvenile myoclonic epilepsy	JME	GABRA1	α1 subunit (GABAA) receptor	Cossette et al. 2002
		EFHC1	EF hand motif protein	Suzuki et al. 2004
Genetic (generalized epilepsy with febrile seizures plus (GEFS+)		SCN1A	NaV1.1 (Na+ channel)	Escayg et al. 2000
	GEFS+	SCN1B	β1 subunit (nACh receptor)	Wallace et al. 1998
		GABRG2	γ2 subunit (GABAA) receptor	Baulac et al. 2001
Generalized epilepsy and paroxysmal dyskinaesia	GEPD	KCNMA1	KCa1.1 (K+ channel)	Du et al. 2005
Epileptic encephalopathies
Dravet syndrome (severe myoclonic epilepsy of infancy)	SMEI	SCN1A	NaV1.1 (Na+ channel)	Claes et al. 2001
Other syndromes
Focal epilepsy and episodic ataxia	EA1+FE	KCNA1	KV1.1 (K+ channel)	Zuberi et al. 1999

The table lists part of the affected ion channel genes in humans illustrating how different clinical syndromes arise from inherited disorders of ion channels. Only unequivocally proven genetic defects that have been described more than once in the literature are included. Full references for Table 1 can be found in Supplementary information. The neuronal localization of some of the channel proteins is shown in Figure 1.

Figure 1. Neuronal localization of some relevant voltage- and ligand-gated ion channels.

Shown is a schematic view of an excitatory pyramidal (purple) and an inhibitory (green) neuron and their synaptic connections. Distinctive intracellular compartments are targeted by different populations of ion channels, examples of which as mentioned in this review are shown here: in the somatodendritic compartment, Ca_V (L- and T-type), HCN and some K_V channels; at axon initial segments (AIS) and nodes of Ranvier in pyramidal neurons, K_V1.1, Na_V1.2, K_V7.2, K_V7.3; at AIS of inhibitory neurons, Na_V1.1; in the presynaptic terminals, Ca_V P/Q type; in the postsynaptic compartment, GABA_A and acetylcholine receptors. Upper insert: Colocalization of K_V7.2 and Na_V1.2 channels at AIS of cortical neurons in an adult mouse brain, as revealed by immunofluorescent staining using an anti-K_V7.2 (green) and an anti-Na_V1.2 (red) antibody of sections obtained from an unfixed brain; DAPI (blue) was used to mark the nuclei. Lower insert: Distribution of GABA_A receptors in a primary cultured hippocampal neuron shown by immunofluorescent staining using an anti-GABAAR alpha-1 subunit antibody (red). An anti-MAP2 antibody (green) was used as a somatodendritic and DAPI (blue) as a nuclear marker (Figure kindly provided by Dr. Snezana Maljevic, modified after Maljevic and Lerche, 2011).

Novel genetic technologies allowing sequencing of the whole coding genomic DNA (whole exome) or even the whole genome now allow identification of new mutations and involvement of loci that could not be efficiently explored before by classical methods. As an example for another ion channel alteration in epilepsy, a de novo mutation was recently identified in the sodium channel gene SCN8A in a patient with a severe epileptic encephalopathy. This mutation leads to a dramatic gain-of-function with increased sodium inward current and membrane hyperexcitability (Veeramah et al. 2012).

Ion channel defects modelled in mice

Transgenic and spontaneously mutant mice have proven crucial for identifying novel molecular pathways for epilepsy, validating the causality of candidate genes isolated from human pedigrees, and for unravelling their pathogenic mechanisms. Beginning with the first spontaneous single gene murine model of epilepsy, the mutant mouse tottering, mutations in at least 17 distinct genes have been reported that produce spontaneous electrographic seizures accompanied by behavioural manifestations (Table 2). Along with these spontaneous models, genetic engineering has allowed the creation of an expanding list of targeted null alleles, transgenic overexpression using selected or endogenous promoters, and knock-in of human mutant point mutations.

Table 2.

Genes for voltage-gated ion channel subunits with spontaneous epilepsy phenotypes in mice

Gene	Current/protein	Seizure type	Mutation Ko, knock-out Ki, knock-in	Reference
Sodium
Scna1a	Nav1.1 α subunit	convulsive	Ko	Yu et al. 2006; Ogiwara et al. 2007
Scna1a			Targeted R1648H human Ki	Martin et al. 2010
Scna1b			Ko	Chen et al. 2004
Scn2a	Nav1.2 α subunit	convulsive	transgenic GAL879-881QQQ	Kearney et al. 2001
Calcium
Cacna1a	Cav2.1 P/Q-type α subunit	absence	Spontaneous alleles (tottering, leaner, rocker roller, tg4J, tg5J)	Fletcher et al. 1996; Zwingman et al. 2001; Mori et al. 2000; Miki et al. 2008
Cacna1a	Cav2.1 P/Q-type	absence	Cerebellar selective PCP2-Cre promoter Ko	Mark et al. 2011
Cacna1a	Cav2.1 P/Q-type	absence	Ko	Llinas et al. 2007
Cacnb4	β4 regulatory subunit	absence	Spontaneous (lethargic)	Burgess et al. 1997
Cacna1g	Cav3.1 T-type α subunit	absence	Bac transgene overexpression	Ernst et al. 2009
Cacna2d2	α2δ regulatory subunit	absence	Spontaneous several alleles	Barclay et al. 200; Brill et al. 2004
Potassium
Kcna1	Kv1.1	convulsive	Ko	Smart et al. 1998
Kcna1	Kv1.1	absence	Overexpression	Sutherland et al. 1999
Kcna2	Kv1.2	convulsive	Ko	Brew et al. 2007
Kcnh3	Kv12.2	non-convulsive	Ko	Shang et al. 2010
Kcnmb4	β4 regulatory subunit calcium-activated	non-convulsive	Ko	Brenner et al. 2005
Kcnc2	Kv3.2	convulsive	Ko	Lau et al. 2000
KcnJ6	Girk2 ATP sensitive	convulsive	Ko	Signorini et al. 1997
Kcnq1	Kv7.1	convulsive	Human Ki T311I A340E	Goldman et al. 2009
Kcnq2	Kv7.2	convulsive	Human Ki A306T	Singh et al. 2008
Kcnq3	Kv7.3	convulsive	Human Ki G311V	Singh et al. 2008
Cyclic nucleotide gated
Hcn2	Ih hyperpolarization-activated cyclic nucleotide-gated	absence	Ko spontaneous apathetic	Ludwig et al. 2003; Chung et al. 2009
Chloride
Clcn3	Ichloride	convulsive	Ko	Dickerson et al. 2002

Full references can be found in Supplementary information. The neuronal localization of some of the channel proteins is shown in Figure 1.

Unexpected pathogenic mechanisms are emerging from comparative analysis of current defects in excitatory and inhibitory networks. For instance, as already mentioned above, different types of sodium channels expressed in both glutamatergic and GABAergic cell types play unequal roles in excitability, providing an explanation for the network disinhibition arising from Scna1 deletion (Yu et al. 2006; Ogiwara et al. 2007). Another example is how mutations of calcium Cacna1a and Cacnb4 channel subunit genes that decrease P/Q-type currents may lead to the ‘acquired’ downstream enhancement of low-voltage T-type calcium currents sufficient to produce thalamocortical spike-wave epilepsy (Zhang et al. 2002, Ernst et al. 2009).

Genotype–phenotype correlations of inherited ion channel disorders are far from being well understood. For example, gain- and loss-of-function mutations in the same gene, or in related members of the same gene family, can give rise to alternative seizure phenotypes. In addition, even when the electrographic seizures themselves appear similar, epilepsies arising from different ion channel genes may be accompanied by remarkably different behaviours and co-morbidities, such as the absence of hippocampal remodelling (Singh et al. 2008), or sudden unexpected death (Goldman et al. 2009; Glasscock et al. 2010) in the K⁺ channel family.

Various specific digenic interactions have been explored in mouse models, illustrating the prominent additive (Kearney et al. 2006; Hawkins et al. 2011), and suppressant (Glasscock et al. 2007), effects of ion channel mutations on epilepsy phenotypes. These pairwise combinations are highly informative, but still too simple to explain the complex inheritance of human epilepsy.

Recent analysis of ion channel mutation profiles in idiopathic epilepsy reveals a marked complexity, with no two individuals showing sharing gene variant pattern, but similar channel variant pattern diversity, including ‘deleterious’ variants in known ‘monogenic’ epilepsy genes, in unaffected individuals (Klassen et al. 2011). In mouse models, it is well known, but often not emphasized, that epilepsy phenotypes, as well as induced seizure thresholds (Frankel et al. 2001), are strongly dependent on the genetic background of inbred mouse strains. Thus, the contribution of specific ion channel defects to brain excitability disorders is complex even when they are associated with a clear mendelian disease phenotype in a particular mouse strain or human pedigree. This complexity poses a major challenge to the clinical genetic assessment of the individual's epilepsy risk.

Ion channel plasticity in acquired epilepsy

Temporal lobe epilepsy (TLE), which is the most common form of acquired epilepsy, can be initiated by an insult such as traumatic head injury, febrile seizures or status epilepticus (Engel, 1996), but inflammatory or autoimmune diseases may also be a cause (Bien et al. 2007). Following these insults, there is often a ‘latent period’ before the onset of chronic TLE. The insult preceding TLE can be mimicked in animals by the administration of chemoconvulsants such as kainic acid or pilocarpine to induce status epilepticus (SE), or kindling in which focal seizures are repeatedly induced using electrical stimulation and febrile seizures. Studies both on the mouse tissue and human tissue from TLE patients have provided valuable information on some of the potential mechanisms underlying TLE. In excitatory glutamatergic neurons, the cell surface expression and biophysical properties of numerous voltage-gated ion channels are altered (Table 3), leading to fundamentally altered integrative properties of these neurons.

Table 3.

Voltage-gated ion channels that undergo plasticity during acquired epilepsies

Channel/current	Form of epilepsy	Nature of change	Impact on cell excitability	Reference
HCN channel/ HCN current Ih	Temporal lobe epilepsy (TLE)	Sustained reduction in current density following status epilepticus (SE) induction	Enhanced pyramidal and interneuron excitability	Dugladze et al. 2007; Jung et al. 2007; Marcelin et al. 2009; Shah et al. 2004; Shah et al. 2012; Shin et al. 2008; Wierschke et al. 2010
HCN channels/ HCN current Ih	Febrile seizure-induced epilepsy	Enhanced HCN channel expression and current	Enhanced rebound activity following inhibitory post-synaptic potentials (IPSPs)	Bender et al. 2003; Chen et al. 2001; Dyhrfjeld-Johnsen et al. 2008
HCN current Ih	Fragile X syndrome	Enhanced HCN channel current	Impaired long-term potentiation (LTP in pyramidal neurons)	Brager et al. 2012
CaV3.2 channels/ T-type Ca2+ current	TLE	Transient elevation of expression and current from SE to chronic epilepsy	Enhanced pyramidal cell bursting	Becker et al. 2008; Su et al. 2002
KV4.2 channels/ A-type K+ current	TLE	Reduction 1 week after SE and persisting during chronic TLE	Enhanced pyramidal cell dendritic excitability	Bernard et al. 2004; Monaghan et al. 2008
BK channels	TLE	Reduction in expression during chronic TLE	??	Pacheco Otalora et al. 2008
Kir2 channels/ inward rectifier current	Chronic TLE	Enhanced expression	Reduced dentate gyrus granule cell excitability	Young et al. 2009
KCNN1 (SK1) KCNN2 (SK2) and KCNN3 (SK3) channels	TLE	Transient reduction during chronic TLE	Increased number of hippocampal population spikes	Oliveira et al. 2010
Persistent sodium current	TLE	Sustained increase following SE	Enhanced neuronal excitability	Agrawal et al. 2003; Chen et al. 2011; Epsztein et al. 2010; Hargus et al. 2011; Vreugdenhil et al. 2004

Full references can be found in Supplementary information. The neuronal localization of some of the channel proteins is shown in Figure 1.

The expression and function of one particular ion channel, the hyperpolarization-activated cyclic nucleotide gated (HCN) channel, is implicated in these changes. HCN channel activity is reduced in the cortex and hippocampus within a few hours of SE induction in animal models (Shah et al. 2004; Jung et al. 2011), and the reduction persists for weeks in the models (Shah et al. 2012). Furthermore, a decrease in HCN mRNA and current function has also been found in cortical and hippocampal tissue obtained from TLE patients (Bender et al. 2003; Wierschke et al. 2010), suggesting that decreased HCN channel current is playing a role in epileptogenesis.

HCN channels are cation channels that are activated at potentials more hyperpolarized to −40 mV (Biel et al. 2009). They are predominantly expressed in cortical and hippocampal pyramidal cell dendrites where, intriguingly, they reduce pyramidal cell excitability by restricting synaptic integration and excitability (Shah et al. 2010). HCN channels, however, are also expressed in certain interneurons (Aponte et al. 2006; Dugladze et al. 2007; Matt et al. 2011) as well as pre-synaptically (Bender et al. 2003; Aponte et al. 2006; Huang et al. 2011).

Thus, does a reduction in HCN channel function per se facilitate the onset of chronic TLE? Four HCN subunits (HCN1–4) have been cloned to date. HCN1 subunits are predominantly expressed in the cortex and hippocampus (Biel et al. 2009) and HCN1 null mice are more susceptible to seizures induced by chemoconvulsants or kindling, suggesting that a loss of HCN channel function is likely to enhance epileptogenesis (Huang et al. 2009; Santoro et al. 2010). Further, transiently restoring HCN channel expression by disrupting the interaction between the neuron-restrictive silence factor (NRSF) and HCN1, delays the onset of spontaneous seizure activity (McClelland et al. 2011), and several anti-convulsants used for TLE, such as lamotrigine, augment HCN channel function (Poolos et al. 2002).

By contrast, in some forms of acquired epilepsy such as that following febrile seizures, HCN channel expression and function is increased in hippocampal pyramidal neurons (Chen et al. 2001; Brewster et al. 2002; Dyhrfjeld-Johnsen et al. 2008). Moreover, dendritic HCN current, I_h, is enhanced in hippocampal pyramidal neurons in mice with targeted deletions in the fragile X FMR1 gene (Brager et al. 2012), although FMR1 knock-out mice do not exhibit spontaneous seizures but are more susceptible to audiogenic seizures (Musumeci et al. 2000). Similarly, only about a third of the rodents subjected to febrile seizures develop chronic epilepsy (Walker & Kullmann, 1999; Dube et al. 2009). Hence, whether I_h upregulation under these conditions is a homeostatic change or epileptogenic remains to be further investigated.

In addition to HCN channels, the mRNA levels and activity of the low-threshold T-type Ca²⁺, Ca_V3.2, channel are transiently elevated in the hippocampus following SE induction (Su et al. 2002; Becker et al. 2008). Inhibition of these channels is also likely to benefit TLE treatment as seizure frequency and incidence is significantly lowered in Ca_V3.2 null mice (Becker et al. 2008). Intriguingly, hippocampal mossy fibre sprouting, a hallmark of TLE, is absent following SE induction in Ca_V3.2 null mice (Becker et al. 2008), suggesting that Ca²⁺ entry through Ca_V3.2 channels may have effects additional to alterations in neuronal excitability.

Although there is considerable evidence that HCN and Ca_V3.2 channels undergo plasticity very early on in the epileptic process, other channels are likely to be involved in seizure induction. Certainly during chronic TLE, the expression and biophysical properties of a substantial number of K⁺ and Na⁺ channels are altered, examples of which are shown in Table 3. A better understanding of how ion channel expression and function contributes to epileptogenesis, is important and may lead to more targeted treatments for TLE and other epilepsies.

Ion channel plasticity contributing to epileptic phenotypes may also occur following inherited lesions in non-ion channel gene mutations. A striking example is the aberrant excitability and electrographic seizures identified in models of amyloid precursor protein beta (Abeta) overexpression (Palop et al. 2007). The seizure activity emanates from the hippocampal and neocortical circuitry and has been described in a wide variety of human and mouse models of Alzheimer's disease (Noebels, 2011). Recent analysis of tissue from Alzheimer's disease patients and a mouse model identified decreased levels of sodium Na_V1.1 currents and SCN1A subunit expression in parvalbumin-containing interneurons (Verret et al. 2012). When these currents were restored in the mouse model by transgenic expression, inhibitory synaptic activity and gamma oscillations were restored, hypersynchrony was reduced and memory deficits were ameliorated, indicating that ion channel-mediated disinhibition makes a strong contribution to epileptogenesis in this model.

Antibodies to voltage-gated or ligand-gated ion channels

Antibodies to AMPAR3 were reported in a few patients with Rasmussen's encephalitis, a devastating unihemispheric condition in childhood (Rogers et al. 1994), but they are not found frequently among cases of this rare epilepsy syndrome (Watson et al. 2004). More recently, antibodies to other neuronal surface proteins have been found in patients with seizures presenting as part of a more widespread inflammatory brain disorder (see Vincent et al. 2011a). The ion channel and related proteins that are the targets for these specific antibodies are listed in Table 4.

Table 4.

Seizure-related syndromes associated with antibodies to ion channels or receptors

Target channel	Antibodies to:	Clinical syndrome	Clinical features	Associated features	Key references
Kv1 complex includes LGI1, CASPR2	LGI1, CASPR2	Limbic encephalitis	Amnesia, change in personality or psychosis, temporal lobe and other seizure types, mainly partial complex seizures	Serum hyponatraemia	Vincent et al. 2004; Irani et al. 2010
Kv1	LGI1	Faciobrachial dystonic seizures (FBDS)	Brief frequent dystonic seizures usually unilateral often involving the arm and ipsilateral face	Can precede limbic encephalitis Often resistant to AEDs Immunotherapy-responsive	Irani et al. 2011
AMPAR1/2	AMPAR1/AMPAR2 mainly	Limbic encephalitis	As above but with more evident psychosis		Lai et al. 2009
AMPAR3	AMPAR3 in a few reports, otherwise none defined	Rasmussen's encephalitis	Intractable unilateral seizures with hemiplegia	Patients develop epilepsia partialis continua (EPC)	Bien et al. 2004
GABAbR	GABAbR	Limbic encephalitis	As above but often dominated by TLE		Lancaster et al. 2010
NMDAR	NR1	Psychiatric features and seizures	Seizures are part of the presentation but are not well defined.	Most patients progress over days or weeks to a complex encephalopathy	Dalmau et al. 2011; Niehusmann et al. 2009

Full references can be found in Supplementary information. The neuronal localization of some of the channel proteins is shown in Figure 1.

Most of the patients have a form of limbic encephalitis – a syndrome that includes amnesia, seizures and psychiatric or behavioural changes, and MRI or cerebrospinal fluid (CSF) evidence of inflammation (Vincent et al. 2004). Antibodies that immunoprecipitate dendrotoxin-labelled K_V1 channels extracted from rodent brain tissue, and bind to LGI1, a secreted protein tightly associated with these channels in situ (see also Table 1), are also found in a recently described seizure type termed faciobrachial dystonic seizures. These involve brief (a few seconds), very frequent (up to 200 per day), usually unilateral dystonic epileptic events. These seizures may precede the onset of limbic encephalitis, and are frequently anti-epileptic drug (AED) resistant, but seizure frequency is markedly reduced following immunotherapies (Irani et al. 2011). Antibodies that immunoprecipitate the K_V1 complexes are also found in a proportion (around 10–15%) of patients with idiopathic forms of epilepsy (McKnight et al. 2005; Vincent et al. 2011b; Quek et al. 2012).

Most antibodies reduce the surface expression of their target by binding divalently to adjacent cell-surface proteins and causing their internalisation – a mechanism that has been shown to apply to NMDAR antibodies binding to hippocampal neurons in culture (Hughes et al. 2010). Direct functional effects on ion channel function have not yet been demonstrated, but purified IgG from one patient with K_V1/LGI1 antibodies increased the release probability at the mossy fibre–CA3 synapse, similar to that of dendrotoxin, and suggesting that the antibodies modify K_V1 channel function (Lalic et al. 2011). It is still early days and these disorders present many challenges including how and where the antibodies access the brain parenchyma, whether they induce permanent damage with possible compensatory changes, or changes that are fully reversible, and the relative roles of serum and CSF antibodies.

Challenges and questions for the future

Although a large amount of the data on ion channel defects shed light on the aetiology of seizures in both genetic and acquired epilepsies, there is still a lack of understanding of the precise mechanisms underlying epileptogenesis leading to a chronically epileptic brain. In addition to adding to the cellular and molecular changes that occur in epilepsy, it will be necessary to ask new qualitatively different questions.

How complex are the genetic epilepsies and what is the role of genetic background in defining the disease phenotype?

Is it justified to call the familial forms of epilepsy with known gene mutations monogenic? There is no doubt that many single genes are responsible for the respective epilepsy syndromes, as summarised in Table 1, but all these syndromes also show considerable phenotypic variability. For example, mutations with complete loss of function of SCN1A are found more frequently in Dravet syndrome than in generalized epilepsy with febrile seizures plus (GEFS+), which is more often caused by missense mutations; however, the opposite relationship has also been described (http://www.molgen.vib-ua.be/SCN1AMutations/). The inconsistant relationship between the phenotype and a single monogenic mutation may be explained in part by other co-expressed ion channel variants in the genomic background, as seen in mouse models. Large sequencing efforts in groups with different phenotypes might be able to detect such variability in humans, but proof-of-concept studies are so far lacking.

Despite the existence of monogenic or major gene effects in some patients (Table 1), in the majority such defects have not been found (Heinzen et al. 2012). Search for individual patterns of variability among ion channels and related proteins has not yet provided clear differences between patients and controls (Klassen et al. 2011). However, more detailed bioinformatic analyses may demonstrate pathways that are affected more commonly in cases than in controls. In addition, the genetic variation in ion channel genes is only one piece of the puzzle. Variations in copy numbers of several chromosomal regions are significantly more frequent in IGE patients than in controls (Sisodiya & Mefford, 2011), suggesting non-ion channel genetic influences. Time will tell whether whole exome or genome sequencing efforts will be able to clarify a significant part of the genetic origin of complex genetic epilepsies.

Are there neuronal compartments that have been insufficiently examined with respect to epilepsy-related changes?

While a large number of beautiful studies have addressed changed properties of somatodendritic compartments, we currently know very little about the local integrative properties of small-calibre dendrites, even though these processes receive the majority of excitatory inputs in most excitatory neurons. It will be necessary to apply techniques suitable for the analysis of small-calibre dendrites such as multiphoton glutamate uncaging and imaging, along with novel techniques for obtaining patch-clamp recording from these processes, to study local integration at these sites. Similar approaches will be required to interrogate other critical excitability compartments, including the axon initial segment and presynaptic terminals. The properties of aberrantly sprouted axons and their collateral terminals involved in seizure-induced neosynaptogenesis also require exploration.

How are changes integrated at the mesoscale (i.e. the level of micronetworks and assemblies of these hundreds of neurons)?

The manifestations of all neurological disorders, such as seizures in the case of epilepsy, rely on a disturbed interplay of different neuron types in the CNS. However, in both genetic and acquired epilepsies, we know surprisingly little about the precise changes in excitability and synaptic integration in different types of principal neurons and interneurons. In the case of some genetic epilepsy models, changes in interneurons appear to be crucial in mediating hyperexcitability, but these studies have so far not dissected the role of specific interneuron types (Yu et al. 2006; Ogiwara et al. 2007). It will undoubtedly be important to understand disease-related modifications in different, well-defined cell types in the CNS. This applies to genetic epilepsy models, the effects of putative disease-related autoantibodies, as well as to acquired epilepsies. Furthermore, it will be necessary to apply and further develop the techniques to record and manipulate the activity of large-scale networks of principal cells and interneurons in normal and diseased brain in order to understand how the different neuron types interact in the epileptic brain to cause aberrant synchronization (Feldt et al. 2011).

What is the in vivo relevance for changes in ion channels observed in vitro?

The manifold changes in ion channels, described in part above, predict powerful changes of synaptic integration and neuronal input–output behaviour. However, it would be highly desirable to identify how neuronal firing behaviour in vivo changes in chronic epilepsy. Studies using juxtacellular multielectrode arrays, single-unit, or direct intracellular and patch-clamp recordings as well as in vivo imaging techniques will be required to probe the in vivo relevance of in vitro membrane excitability findings.

What is the role of homeostasis in defining the disease phenotype?

Neurons, and most probably all neural cells, have an immense inherent capability for homeostasis, manifested as the ability to conserve functional properties in the face of continuous environmental perturbation and turnover of their proteolipid components. The arsenal of neuronal homeostatic mechanisms includes regulation of both synaptic and intrinsic properties to maintain neuronal functions. In view of this fact, it is perhaps surprising that in epilepsy the underlying ion channel defects are not homeostatically compensated. It will be important to understand fully which homeostatic mechanisms are active under normal excitability conditions in order to identify the conditions under which they fail.

Glossary

IGE

idiopathic generalized epilepsy

SE

status epilepticus

TLE

temporal lobe epilepsy

Supplementary material

Supplementary information