doi:10.1093/brain/awq057

Brain 2010: 133; 1403–1414

| 1403

BRAIN A JOURNAL OF NEUROLOGY

Molecular correlates of age-dependent seizures in an inherited neonatal-infantile epilepsy Yunxiang Liao,1,2, Liesbet Deprez,3, Snezana Maljevic,1,2, Julika Pitsch,4 Lieve Claes,3 Dimitrina Hristova,5 Albena Jordanova,3,6 Sirpa Ala-Mello,7,8 Astrid Bellan-Koch,1 Dragica Blazevic,1 Simone Schubert,1 Evan A. Thomas,9 Steven Petrou,9,10 Albert J. Becker,4 Peter De Jonghe3,11 and Holger Lerche1,2 1 2 3 4 5 6 7 8 9 10 11

Neurological Clinic and Institute of Applied Physiology, University of Ulm, Ulm, 89081, Germany Department of Neurology and Epileptology, Hertie-Institute for Clinical Brain Research, University Hospital Tu¨bingen, Tu¨bingen, 72076, Germany Neurogenetics Group, VIB Department of Molecular Genetics, University of Antwerp, Antwerpen, 2610, Belgium Department of Neuropathology, University of Bonn Medical Center, Bonn, 53105, Germany Tokuda Hospital Sofia, Sofia, 1407, Bulgaria Department of Chemistry and Biochemistry, Medical University-Sofia, Sofia, 1407, Bulgaria Department of Clinical Genetics, Helsinki University Central Hospital, Helsinki, 00290, Finland Rinnekoti Foundation, Rinnekodintie 10, Espoo, 02980, Finland Florey Neurosciences Institute, Melbourne, Parkville 3010, Australia The Centre for Neuroscience, The University of Melbourne, Melbourne, Victoria 3010, Australia Division of Neurology, University Hospital Antwerp, Antwerpen, Belgium

*These authors contributed equally to this work. Correspondence to: Professor Dr Holger Lerche, Department of Neurology and Epileptology, Hertie-Institute for Clinical Brain Research, University Hospital Tu¨bingen, D-72076 Tu¨bingen, Germany E-mail: [email protected]

Many idiopathic epilepsy syndromes have a characteristic age dependence, the underlying molecular mechanisms of which are largely unknown. Here we propose a mechanism that can explain that epileptic spells in benign familial neonatal-infantile seizures occur almost exclusively during the first days to months of life. Benign familial neonatal-infantile seizures are caused by mutations in the gene SCN2A encoding the voltage-gated Na+ channel NaV1.2. We identified two novel SCN2A mutations causing benign familial neonatal-infantile seizures and analysed the functional consequences of these mutations in a neonatal and an adult splice variant of the human Na+ channel NaV1.2 expressed heterologously in tsA201 cells together with beta1 and beta2 subunits. We found significant gating changes leading to a gain-of-function, such as an increased persistent Na+ current, accelerated recovery from fast inactivation or altered voltage-dependence of steady-state activation. Those were restricted to the neonatal splice variant for one mutation, but more pronounced for the adult form for the other, suggesting that a differential developmental splicing does not provide a general explanation for seizure remission. We therefore analysed the developmental expression of NaV1.2 and of another voltage-gated Na+ channel, NaV1.6, using immunohistochemistry and real-time reverse transcription–polymerase chain reaction in mouse brain slices. We found that NaV1.2 channels are expressed early in development at axon initial segments of principal neurons in the hippocampus and cortex, but their expression is diminished and they

Received April 30, 2009. Revised February 8, 2010. Accepted February 15, 2010. Advance Access publication April 5, 2010 ß The Author (2010). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

1404

| Brain 2010: 133; 1403–1414

Y. Liao et al.

are gradually replaced as the dominant channel type by NaV1.6 during maturation. This finding provides a plausible explanation for the transient expression of seizures that occur due to a gain-of-function of mutant NaV1.2 channels.

Keywords: epilepsy; seizure; sodium channel; development; axon Abbreviations: EGFP = enhanced green fluorescent protein; IRES = internal ribosomal entry site; mRNA = messenger RNA

Introduction Idiopathic epilepsies comprise a group of clinically well-defined syndromes. They have a primary genetic background, usually no structural brain abnormalities and most of them have a benign course without additional neurological symptoms. Therefore, they constitute interesting models to study the primary pathophysiological mechanisms of epilepsy. A hallmark of many of these syndromes is a syndrome-specific age dependence with a developmental pattern of seizure onset and remission, the mechanisms of which are poorly understood. One group of seizure syndromes with a clear age dependence occurring in neonates and infants can be divided in three subforms: benign familial neonatal seizures, benign familial neonatal-infantile seizures (BFNIS) and benign familial infantile seizures. They are characterized by clusters of partial or secondarily generalized seizures with onset either in the first days of life (benign familial neonatal seizures), between 3 and 12 months of age (benign familial infantile seizure) or more variably between the first days or months after birth (BFNIS) (Berkovic et al., 2004; Specchio et al., 2006). For two of these syndromes, genetic defects have already been identified. Whereas benign familial neonatal seizures are caused by loss-of-function mutations in the two genes KCNQ2 and KCNQ3 encoding the voltage-gated K+ channels KV7.2 and KV7.3 (Maljevic et al., 2008), mutations in the gene SCN2A encoding the voltage-gated Na+ channel NaV1.2 have been detected in BFNIS (Heron et al., 2002; Berkovic et al., 2004). NaV1.2 is one of four voltage-gated Na+ channel alpha-subunits expressed in the mammalian brain (Vacher et al., 2008), which are responsible for the initiation and conduction of action potentials. The first two studies investigating the functional consequences of a few SCN2A mutations using the rat or human isoforms of the channel predicted subtle gain-of-function mechanisms and an increase in neuronal firing (Scalmani et al., 2006; Xu et al., 2007). Another study suggested a loss-of-function by decreased surface expression for some mutations (Misra et al., 2008). Since the mechanisms underlying the striking age dependence of these syndromes are elusive, we set out to investigate both the mechanisms of seizure generation and seizure remission in BFNIS. New mutations were identified in BFNIS families and functionally studied with the patch clamp technique using two different splice variants of the human NaV1.2 channel expressed either early (neonatal) or late (adult) in development (Kasai et al., 2001). While a previous study already suggested for one BFNIS mutation that alternative splicing by itself does not explain a predominant seizure generation in neonates and infants (Xu et al., 2007), a differential developmental expression of distinct Na+ channel subunits could provide an intriguing explanation for an age-dependent

occurrence of seizures. We investigated this hypothesis for BFNIS, studying the developmental expression of two voltagegated Na+ channels, NaV1.2 and NaV1.6 using immunohistochemistry in mouse brain slices. The expression of these two channels had been shown previously to be developmentally regulated in retinal ganglion cells (Boiko et al., 2001, 2003; Kaplan et al., 2001).

Materials and methods Patients and pedigrees In this study, eight unrelated patients with a benign form of epilepsy starting in the first year of life were recruited. The two families in which mutations have been detected are described here in more detail. All patients and relatives or their legal representatives gave written informed consent to participate in this study. Ethical approval was obtained from the responsible local authorities. In Family 1, of Bulgarian origin, the index patient (Fig. 1A, III.1) experienced three seizures in 1 day around the age of 4 months. The seizures were bilateral tonic–clonic and associated with sweating and eye deviation to the left. Acute treatment with diazepam was administered. Eleven days later, seizures recurred in a cluster of 11 episodes over a period of 24 h with the duration of about 1 min each. Therapy with valproate was started, and the patient remained seizure-free. At the age of 2 years, anti-epileptic treatment was discontinued without a relapse (the patient is now 8 years old). The interictal EEG was normal. The proband’s brother (Fig. 1A, III.2), now aged 6 years, experienced three bilateral clonic seizures at Day 6 of life. At the age of 10 weeks, he had recurring clusters of 15 seizures in 24 h for 2 weeks. Seizures lasted 1 min and were associated with eye deviation to the right, sweating and blushing. He was also treated with valproate. Interictal EEG recordings were normal. Seizures disappeared after the age of 3 months and pharmacotherapy was stopped at the age of 2 years without further seizures. Both parents experienced no epileptic seizures. The information about the childhood of the maternal grandfather (Fig. 1A, I.1) was based on history. He was reported to have experienced multiple seizures ‘as a baby’, characterized by clonic jerking of both arms and legs. Several relatives of the maternal grandfather also had seizures early in life, but no DNA samples were available for genetic analysis. They all were reported to have a normal mental status and no seizures later in life. In Family 2 from Finland, the index patient (Fig. 1A, II.1), now aged 7 years, experienced his first seizure 24 h after birth, characterized by unilateral clonic jerks. Later on, other seizure types occurred including bilateral clonic seizures, apnoeas and atypical absence seizures. Seizures were difficult to control but finally responded well to a combination of phenobarbitone and phenytoin. Interictal EEG recordings

Molecular correlates of age-dependent seizures

Brain 2010: 133; 1403–1414

| 1405

Figure 1 Pedigrees and genetic analysis. (A) Pedigrees of the two BFNIS families with a SCN2A mutation. Square = male; circle = female; open symbols = unaffected individual; filled symbol = patient with BFNIS; + /M = individual carrying a heterozygous SCN2A mutation; + / + = individual not carrying a SCN2A mutation. (B) Chromatograms of the SCN2A mutations in comparison with wild-type sequences. (C) Predicted transmembrane topology of NaV1.2 showing the location of the mutations. Green diamond = published missense mutations causing BFNIS or benign familial infantile seizure; blue star = published de novo mutations causing intractable epileptic encephalopathies including infantile spasms and Dravet syndrome; yellow square = missense mutation associated with febrile and afebrile seizures; black triangle = splice variant polymorphism, N209D; red circles = new missense mutations reported in this study.

showed multi-spike bursts and irritative activity without focal findings. Brain MRI was unremarkable. Seizures disappeared at the age of 3 months. At the age of 10 months, the interictal EEG was normal and anti-epileptic treatment was discontinued. The further development of the child was normal. None of the parents or close relatives of the patient had a history of seizures.

Genetic analysis DNA was extracted from peripheral blood of the patients and their family members. Patients had different ethnic backgrounds and for each background, DNA samples of 200 randomly selected control individuals were available. The 27 exons of SCN2A were amplified using the polymerase chain reaction with flanking intronic primers designed with the software tool SNP box (Weckx et al., 2005) (primers available upon request). The primers for exon 6 were designed in order to amplify both alternative exons 6A and 6N (from the adult and neonatal splice variant, respectively) (Kasai et al., 2001). The resulting polymerase chain reaction fragments were sequenced with the BigDye Terminator v3.1 Cycle Sequencing kit from Perkin-Elmer Applied Biosystems (Foster City, CA). Sequences were analysed on an ABI3730 automated sequencer using Sequencing Analysis 5.0 software (Applied Biosystems, Foster City, CA). Pyrosequencing with the PSQTM96 System (Pyrosequencing AB, Uppsala, Sweden) was used to confirm the presence of the mutations in patients and their absence

from control individuals. Mutations were numbered according to the published cDNA sequence (accession number NM_021007) with nucleotide +1 corresponding to the A of the ATG translation initiation codon and the nomenclature followed the MDI/HGVS Mutation Nomenclature Recommendations (http://www.hgvs.org/mutnomen) (den Dunnen and Antonarakis, 2001). To test paternity, we genotyped 23 short tandem repeat markers located on 11 different autosomal chromosomes. The markers were amplified in two multiplex polymerase chain reactions.

Cloning and mutagenesis The neonatal and adult splice variants of the human NaV1.2 channel had been cloned before into the mammalian expression vector pcDNA3.1 (Xu et al., 2007). Site-directed mutagenesis was performed to engineer both mutations into both splice variants of human NaV1.2 using overlap polymerase chain reaction strategy (primers are available upon request). All mutant cDNAs were fully resequenced before used in experiments to confirm the introduced mutations and exclude any additional sequence alterations. The human auxiliary subunits hb1 and hb2 in the pCLH vector were kindly provided by GlaxoSmithKline. We exchanged the hygromycin coding region in the vector with the sequence coding for either enhanced green fluorescent protein (EGFP) or CD8 marker genes to obtain pCLH-hb1-EGFP or pCLH-hb2-CD8, respectively. The presence of the internal ribosomal

1406

| Brain 2010: 133; 1403–1414

entry site (IRES) site in the vector enabled parallel expression of the auxiliary subunit and the marker protein in transfected cells, as has been described previously by Lossin et al. (2002).

Transfection and expression in tsA201 cells Human tsA201 cells were cultured at 37 C, with 5% CO2 humidified atmosphere and grown in 50% Dulbecco’s modified Eagle medium + 50% F-12 HAM (Invitrogen, Carlsbad, CA) + 10% (v/v) foetal bovine serum. A standard calcium phosphate transfection method was performed for transient expression of wild-type or mutant Na+ channel -subunits together with b1- and b2-subunits in tsA201 cells. Approximately 6 mg of total DNA was transfected in a molar ratio 1:1:1. Anti-CD8 antibody-coated microbeads (Dynabeads M450, Dynal, Norway) suspended in phosphate buffered saline were added to the cells and gently shaken. Only the cells positive for both CD8 antigen and green fluorescent protein fluorescence were used for electrophysiological measurements.

Electrophysiology Standard whole-cell recordings were performed using an Axopatch 200B amplifier, a Digidata 1320A digitizer and pCLAMP 8 data acquisition software (Axon Instruments, Union City, CA, USA). Leakage and capacitive currents were automatically subtracted using a pre-pulse protocol (  P/4). Currents were filtered at 5 kHz and digitized at 20 kHz. All measurements were performed at room temperature of 21–23 C. Na+ currents of 1.5–12 nA were recorded from the transfected tsA201 cells, at least 10 min after establishing the whole cell configuration. Borosilicate glass pipettes were fire polished with a final tip resistance of 1–1.3 MV when filled with internal recording solution (see below). We carefully checked that the maximal voltage error due to residual series resistance after up to 90% compensation was always 55 mV. The pipette solution contained (in mM): 105 CsF, 35 NaCl, 10 EGTA, 10 (4-(2-hydroxyethyl)-1-piperazineethanesulphonic acid (HEPES) (pH 7.4). The bath solution contained (in mM) 150 NaCl, 2 KCl, 1.5 CaCl2, 1 MgCl2, 10 HEPES (pH 7.4).

Data recording and analysis The following voltage clamp protocols were used for recordings. The membrane was depolarized to various test potentials from a holding potential of 140 mV. A second-order exponential function was the best fit to the time course of fast inactivation during the first 70 ms after onset of the depolarization, yielding two time constants. The weight of the second slower time constant was relatively small (55%). Only the fast time constant, named  h, was therefore used for data presentation in the ‘Results’ section. Persistent Na+ currents (ISS, for the ‘steady-state’ current) were determined at the end of depolarizing pulses, lasting 70 ms, to different test potentials and are given relative to the initial peak current (IPEAK). Recovery from fast inactivation was recorded from holding potentials of  140 mV. Cells were depolarized to  20 mV for 100 ms to inactivate all Na+ channels and then repolarized to various recovery potentials ( 80, 100 or 120 mV) for increasing duration. A first-order exponential function with an initial delay was the best fit to the time course of recovery from inactivation.  rec, is shown for data evaluation.

Y. Liao et al. The activation curve (conductance–voltage relationship) was derived from the current–voltage relationship that was obtained by measuring the peak current at various step depolarizations from the holding potential of 140 mV. The following Boltzmann function was fit to the obtained data points: g 1 ¼ , gmax ðV Þ f1 þ exp½ðV  V1=2 Þ=kV g with g = I/(V  Vrev) being the conductance, I the recorded current amplitude at test potential V, Vrev the Na+ reversal potential, gmax the maximal conductance, V1/2 the voltage of half-maximal activation and kV a slope factor. Steady-state inactivation was determined using 300 ms conditioning pulses to various potentials followed by the test pulse to  20 mV at which the peak current reflected the percentage of non-inactivated channels. A standard Boltzmann function was fit to the inactivation curves: I 1 ¼ Imax ðV Þ f1 þ exp½ðV  V1=2 Þ=kV g with I being the recorded current amplitude at the conditioning potential V, Imax being the maximal current amplitude, V1/2 the voltage of half-maximal inactivation and kV a slope factor. All data were analysed using a combination of pCLAMP, Microsoft Excel and Origin (OriginLab Inc., Northampton, MA, USA) software. For statistical evaluation, Student’s t-test was applied. All data are shown as mean  SEM.

Preparation of brain slices and immunohistochemistry The use of animals and all experimental procedures were approved by local authorities (Regierungspraesidium Tuebingen, Tuebingen, Germany). C57BL/6J mice at different postnatal days (P1, P3, P5, P8, P10, P15, P20, P30, P40 and P90) were sacrificed by CO2 inhalation followed by decapitation. Brains were removed and frozen immediately in liquid nitrogen vapour and stored at 80 C until sectioning. Coronary sections (5 mm) were serially cut in a cryostat and mounted on Superfrost microscope slides (Menzel GmbH, Braunschweig, Germany). After drying at room temperature overnight, brain slices were frozen and stored at 80 C. For immunohistochemistry, the cryostat sections were air-dried for 30 min at room temperature and then fixed in pre-cooled acetone at 20 C for 10 min. Blocking was performed in 3% normal goat serum in tris buffered saline containing 0.3% Triton-X. The polyclonal rabbit anti-AnkyrinG antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was diluted 1:1000 and combined with either anti-NaV1.2 (clone K69/3, 1:500 dilution) or anti-NaV1.6 (clone K87 A/10, 1:200 dilution), which were both mouse monoclonal antibodies obtained from Neuromab (Davis, CA). Sections were incubated with primary antibodies at 4 C (NaV1.2) or room temperature (NaV1.6) overnight. After extensive washing with Tris buffered saline, detection was performed using fluorescently labelled secondary antibodies–Alexa 488, goat anti-mouse and Alexa 568 goat anti-rabbit (Invitrogen, Carlsbad, CA) at room temperature for 1 h. The slides were then stained with 40 ,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St Louis, MO) to identify the nuclei. After the final washing step sections were air-dried and covered with a mounting medium (Vecta Shield, Vector Laboratories, Burlingame, CA). The stained sections were stored at 4 C prior to examination on an Axiovision2 plus Zeiss microscope (Jena, Germany).

Molecular correlates of age-dependent seizures Freshly frozen hippocampal dissections were obtained from pharmacoresistant epilepsy patients who underwent epilepsy surgery for seizure control at the Bonn University Medical Centre. All patients gave written informed consent. Procedures were in accordance with the Declaration of Helsinki and approved by the local ethics committee. Cryosections, 12 mm thick, were fixed and stained using the same protocol as for the mouse brain sections. The antibodies used were (i) rabbit polyclonal anti-NaV1.2 (1:200; Alomone Labs, Jerusalem, Israel) with compatible anti-Ankyrin G antibodies (mouse monoclonal anti-AnkG, 1:1000, Calbiochem, Merck, Darmstadt, Germany) and (ii) mouse monoclonal anti-Pan NaV (1:1000, Sigma, Deisenhofen, Germany) combined with rabbit polyclonal anti-AnkyrinG antibodies (see above).

Real-time reverse transcription polymerase chain reaction CA1 and DG regions were microdissected from hippocampal slices (400 mm) of C57BL/6 mice at two different time points of postnatal development (P8 and P30). Messenger RNA (mRNA) was isolated using a Dynabeads mRNA Direct Micro kit (Dynal Biotech, Invitrogen, Carlsbad, CA, USA) and the following cDNA-synthesis was performed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, California, USA) according to the manufacturer’s protocols. NaV1.2 and NaV1.6 transcript quantification was performed by real-time reverse transcription–polymerase chain reaction (PRISM 7700; PE Biosystems, California, USA). Primers for NaV1.2 and NaV1.6 subunits and for synaptophysin were designed with Primer Express software (PE Biosystems, California, USA) (Supplementary Table 1). No significant homology of the amplicon sequences with other previously characterized genes was found searching GenBank databases by the BLASTN program. Relative quantification of the starting mRNA copy numbers using multiple replicates for each reaction was performed according to the

Ct method (Fink et al., 1998). The signal threshold was set within the exponential phase of the reaction for determination of the threshold cycle. Relative quantification started from 5–10 ng of mRNA. Real-time reverse transcription polymerase chain reaction was performed in a 6.25 ml reaction volume containing 3.125 ml of SYBR Green PCR Master Mix (Invitrogen, Carlsbad, CA, USA), 0.1875 ml of forward and reverse primers (10 pmol/ml), 1.5 ml of diethylpyrocarbonate– H2O and cDNA dissolved in 1.25 ml of diethylpyrocarbonate–H2O. Reactions were performed in triplets. After preincubation for 10 min at 94 C, we performed 40 polymerase chain reaction cycles (20 s at 94 C followed by 30 s at 59 C and 40 s at 72 C). The SYBR Green fluorescence signal was measured in each cycle. Statistical significance was analysed using Student’s t-test. Outliers were rejected using the Grubb’s test. The values were considered significantly different at P50.05. Results are shown as means  SEM.

Results Genetics We identified two heterozygous SCN2A mutations: c.754A4G predicting M252V in Family 1 and c.781G4A predicting V261M in Family 2 (Fig. 1A, B). The clinical phenotypes of the patients

Brain 2010: 133; 1403–1414

| 1407

carrying these mutations are described in detail in the ‘Materials and methods’ section. The M252V mutation was present in the two affected siblings, their affected maternal grandfather and in their asymptomatic mother. The V261M was only detected in the index case of Family 2. All affected individuals presented with a phenotype compatible with BFNIS, i.e. with a neonatal or infantile onset of benign partial-onset seizures remitting completely with maturation. The unaffected parents of the index case of Family 2 did not carry the mutation. Since non-paternity was ruled out, this strongly suggests that the mutation arose de novo. Both mutations were absent from 200 ethnically matched control individuals. The affected amino acids are highly conserved among the -subunits of mammalian brain sodium channels (Supplementary Fig. 1). They are located very near to each other in segment S5 of domain I (Fig. 1C).

Effects of the mutations on the gating of the neonatal and adult splice variants of NaV1.2 channels There are two common splice variants that are conserved among neuronal voltage-gated Na+ channels, a neonatal and an adult form. The difference is determined by an amino acid substitution at position 209, asparagine by aspartic acid, located in the S3–S4 extracellular loop of domain I. The mutations were introduced into the cDNAs of the SCN2A gene in each of the two splice variants. Wild-type or mutant plasmids together with those encoding human b1- and b2-subunits were co-transfected into tsA201 cells to study their gating properties. The co-expression of both b-subunits was controlled by co-expression of the enhanced green fluorescent protein or CD8 using bicystronic constructs (Lossin et al., 2002). Typical whole cell Na+ currents for both splice variants of wild-type and mutant channels were elicited by various depolarizing voltage steps from a holding potential of 140 mV (Fig. 2A–C). The presence of an increased persistent current compared to the wild-type was observed for mutant M252V channels when engineered into the neonatal splice variant (Fig. 2D and E; Table 1). This non-inactivating current was quantified at the end of a longer test depolarization (70 ms) relative to the peak current (ISS/IPEAK). It could be reversibly and completely blocked by the application of 20 nM tetrodotoxin, a specific blocker of voltage-gated Na+ channels (Fig. 2D, inset). Such an increase in persistent current has been described for many different Na+ channel disorders going along with a hyperexcitability including myotonia, cardiac arrhythmia and epilepsy (George, 2005; Lerche et al., 2005; Cannon, 2006). It increases the Na+ influx and can readily explain a membrane depolarization and increased firing in neurons (Golomb et al., 2006; Vervaeke et al., 2006). In contrast to these observations in the neonatal splice form, the persistent current was only slightly and not significantly increased for this mutation in the adult splice variant (Fig. 2D and E; Table 1). The predicted increase in neuronal firing could thus disappear or be diminished as soon as the neonatal splice variant is no longer expressed during development (Kasai et al., 2001). All other

1408

| Brain 2010: 133; 1403–1414

Y. Liao et al.

Figure 2 Representative wild-type and M252V whole-cell Na+currents and their electrophysiological properties. (A–C) Whole-cell recordings of wild-type (WT) and M252V mutant channels from transfected tsA201 cells: (A) wild-type, neonatal (neo) splice variant, (B) M252V in the neonatal splice variant, (C) M252V in the adult (ad) splice variant. Na+ currents were elicited by step depolarizations ranging from 105 to +67.5 mV from a holding potential of 140 mV. (D) Representative wild-type and M252V persistent, tetrodotoxin-sensitive Na+ currents. Current amplitudes were recorded at the end of a 70 ms depolarization to 0 mV and are normalized to the peak amplitude (ISS/IPEAK). Upon application of 20 nM tetrodotoxin (TTX), the persistent current was completely abolished, as shown in the inset. (E) Voltage dependence of the persistent current (ISS/IPEAK). (F) Voltage dependence of Na+ channel steady-state activation and fast inactivation. (G) Voltage dependence of the fast inactivation time constant,  h. (H) Time course of recovery from fast inactivation recorded at  100 mV. (I) Voltage dependence of the time constant of recovery from fast inactivation,  rec. Voltage clamp protocols are described in the ‘Materials and methods’ section. Numbers of recorded cells and statistical analysis are provided in Table 1. All data are shown as mean  SEM. investigated gating parameters of activation, fast and slow inactivation, as well as the current density, were not significantly different between mutant M252V and wild-type channels for both the neonatal and the adult splice forms, except a slight but significant acceleration of recovery from slow inactivation for the neonatal splice variant of M252V (Table 1; Supplementary Fig. 2 and Table 2). This latter finding may contribute to an increase in neuronal excitability by a shortening of the refractory period after high frequency firing, when slow inactivation may occur in native neurons. Typical whole-cell recordings from cells expressing mutant V261M channels are displayed in Fig. 3A and B. In contrast to the M252V mutant channel, a significantly increased persistent Na+ current was not observed in the background of the neonatal but was seen in the adult splice variant, when compared to the wild-type (Fig. 3C and D; Table 1). The slope of the activation curve was significantly decreased for both splice variants (Fig. 3E; Table 1) while the voltage of half-maximal activation was not significantly changed. This finding predicts an increase in channel availability at subthreshold voltages. Furthermore, we found a subtle slowing of the fast inactivation time course for the adult splice variant of V261M and a marked acceleration of recovery from fast inactivation for both splice variants (Fig. 3F–H; Table 1). In particular the latter finding predicts an increase in neuronal firing via a shortening of the refractory period after an action potential. Another finding for the V261M mutation was an acceleration of recovery from slow inactivation in the background of the neonatal splice variant, similarly as seen for M252V (Supplementary Fig. 2 and Table 2). The current density was not significantly changed for the mutation (Table 1).

In contrast to the M252V mutation, however, the gain-of-function gating changes for V261M were seen in both the neonatal and the adult splice variants, with some changes only seen in the adult, so that a developmentally regulated modifying effect of the polymorphism in the D1/S3-S4 loop cannot provide a general explanation for seizure remission later in development for BFNIS.

Comparison of neonatal and adult splice variants of wild-type channels It has been reported previously that there is a difference in the gating properties between the neonatal and the adult human splice variants of NaV1.2 WT channels expressed without b-subunits (Xu et al., 2007). The adult splice variant was predicted to enhance neuronal excitability compared to the neonatal one by a depolarizing shift in the voltage dependence of steady-state fast inactivation, a slower time course of and a faster recovery from fast inactivation, while a depolarizing shift of the steady-state activation curve (which would decrease excitability) did not reach statistical significance (Xu et al., 2007). We observed the same gating changes between both splice variants when the NaV1.2 -subunit was expressed alone, although these differences did not reach statistical significance (Supplementary Fig. 3; Supplementary Table 3). However, when we co-expressed both the b1- and b2-subunits, these differences could not be reproduced. Hence, the relatively small changes at the border of resolution between the different splice variants may be influenced by the co-expression of auxiliary subunits.

Electrophysiological parameters as determined in whole cell patch clamp recordings of transfected tsA201 cells for activation and fast inactivation of wild-type (WT), M252V and V261M mutants in neonatal (neo) or adult (ad) splice variants of NaV1.2 channels. See ‘Experimental procedures’ section for the voltage clamp protocols used. V1/2 = voltage of half-maximal activation or inactivation; k = slope factor; n = number of recorded cells;  h = time constant of fast inactivation; ISS/IPEAK = persistent (steady-state) Na+ current divided by the peak current;  rec = time constant of recovery from fast inactivation. Mean  SEM values are shown. Significant differences between mutant and wild-type channels are indicated as follows: *P50.05, **P50.01, ***P50.001.

0.8  0.0 1.8  0.1*** 0.9  0.2 0.8  0.1 1.1  0.2 1.4  0.3* 11 13 10 10 10 8 WT neo M252V neo V261M neo WT ad M252V ad V261M ad

9  1.3 42.3  1.2 41.0  0.9 43.3  1.4 42.3  1.8 41.8  1.3

4.8  0.3 5.2  0.2 6.4  0.5** 5.60.4 5.5  0.4 6.7  0.2*

15 15 11 12 11 10

75.5  1.1 75.9  1.1 77.8  0.6 77.9  1.4 77.8  1.7 79.51.1

5.1  0.1 5.3  0.2 5.2  0.2 4.5  0.1 4.8  0.2 5.8  0.2***

11 12 9 9 10 10

0.24  0.01 0.25  0.01 0.25  0.02 0.24  0.01 0.26  0.01 0.33  0.03**

n  h at 0 mV (ms) n k

Steady-state inactivation

V1/2 (mV) N k

Steady-state activation

V1/2 (mV)

Table 1 Main electrophysiological parameters for wild-type and both mutant channels

ISS/IPEAK at 0 mV (%) n

8 12 6 7 9 7

4.8  0.4 6.2  0.6 3.4  0.2** 6.6  0.7 6.1  0.8 4.6  0.3*

 rec at 100 mV (ms) n

10 13 11 9 8 10

810  170 540  160 600  180 800  210 390  100 320  40

Current density (A/F) n

16 21 11 17 12 13

Molecular correlates of age-dependent seizures

Brain 2010: 133; 1403–1414

| 1409

Differential developmental expression of NaV1.2 and NaV1.6 channels at axon initial segments of principal neurons Since we did not find a clear explanation for the transient expression of the clinical phenotype during the neonatal-infantile period in our electrophysiological experiments, we studied the expression of NaV1.2 and NaV1.6 channels at axon initial segments, at which voltage-gated Na+ channels are highly concentrated to generate action potentials. We used unfixed mouse brains in different stages of postnatal development (P1–P90) to study the differential expression of both channels at the axon initial segments of pyramidal neurons in the hippocampus and cortex. The specificity of both monoclonal antibodies was tested by heterologous expression of the respective subunits in tsA201 cells. We did not observe any cross reaction between both channel subtypes in this system (Supplementary Fig. 4). Furthermore, the antigen for the NaV1.2 antibody was located in the C-terminal region of the channels, thus far away from the polymorphism in the D1/S3-S4 loop that differs among the neonatal and adult splice variants, so that both splice variants are predicted to be stained with the same channel-specific antibodies. Both antibodies heavily stained the axon initial segments, as was verified by co-staining with antibodies directed against AnkyrinG (Fig. 4). Anti-NaV1.2 antibodies stained the axon initial segments of pyramidal neurons in the CA1 region of the hippocampus, with the highest intensity between P5 and P15 during development. This staining was diminished at later stages of development, although still visible at adult ages. In contrast, staining of the axon initial segments with anti-NaV1.6 antibodies was not detectable before P15 and then gradually increased to reach its highest levels from P30 onwards (Fig. 4). The same pattern of differential developmental staining was observed for the axon initial segments of principal neurons in the cortex (Supplementary Fig. 5A). Recently, Hu et al. (2009) published a study reporting a distinct localization and function of NaV1.2 and NaV1.6 in the axon initial segments of pyramidal neurons in the adult rat cortex. In some of the cortical neurons we could observe a similar proximal staining of the axon initial segments using the NaV1.2 antibody in adult mouse brain sections (Supplementary Fig. 5B), indicating that the diminishment of NaV1.2 may result from a redistribution of this channel from the whole axon initial segments to the proximal part during maturation. The staining pattern within the dentate gyrus and CA3 region revealed an additional aspect. A similar developmental staining compared to CA1 and cortex with both anti-NaV antibodies was observed at the axon initial segments of granule cells within the dentate gyrus and pyramidal cells in the CA3 region. However, heavy staining with anti-NaV1.2 antibodies of mossy fibres projecting from the dentate gyrus to CA3 increased with age and did not disappear in the adult stage. There was no detectable staining with anti-NaV1.6 antibodies of these unmyelinated fibres (Supplementary Figs 6 and 7). A similar observation has been made with unmyelinated fibre bundles of the optic nerve (Boiko et al., 2001, 2003; Kaplan et al., 2001).

1410

| Brain 2010: 133; 1403–1414

Y. Liao et al.

Figure 3 Representative V261M whole-cell Na+ currents and their electrophysiological properties compared to wild-type channels. (A and B) Whole-cell recordings of V261M mutant channels from transfected tsA201 cells: (A) V261M in the neonatal (neo) splice variant, (B) V261M in the adult (ad) splice variant. Na+ currents were elicited by membrane depolarizations ranging 105 to + 67.5 mV from a holding potential of 140 mV. (C) Representative wild-type and V261M mutant persistent Na+ currents normalized to the peak current (ISS/IPEAK). (D) Voltage dependence of the persistent current (ISS/IPEAK). (E) Voltage dependence of Na+ channel steady-state activation and fast inactivation. (F) Voltage dependence of the fast inactivation time constant,  h. (G) Time course of recovery from fast inactivation recorded at  100 mV. (H) Voltage dependence of the time constant of recovery from fast inactivation,  rec. Voltage clamp protocols are described in the ‘Materials and methods’ section. Numbers of recorded cells and statistical analysis are provided in Table 1. All data are shown as mean  SEM.

The observed immunohistochemical stainings suggest a transiently higher expression of NaV1.2 in early stages of development, and a gradual replacement by NaV1.6 channels as the predominant channel at the axon initial segments of principal neurons in the hippocampus and cortex with further maturation. This may concern mainly the distal axon initial segments, as Hu et al. (2009) described a proximal localization of NaV1.2 in adult rat cortical neurons that we also observed in some of our stainings. These findings can nicely explain how gain-of-function mutations in NaV1.2 led to an age-dependent expression of epileptic seizures, regardless if these mutations affect more the gating properties of neonatal or adult splice variants of this ion channel. To compare our findings in mice with human brain slices, we tested several available anti-NaV antibodies and could stain the axon initial segments in sections from unfixed, frozen hippocampal tissue obtained from epilepsy surgery samples of adult individuals, using a different anti-NaV1.2 and an anti-Pan NaV antibody, the latter recognizing all NaV -subunits (Supplementary Fig. 8). With the Pan NaV antibody, we observed a very strong signal in the axon initial segments, whereas NaV1.2 staining appeared weaker. However, these data have to be interpreted with care, since we had to use two different antibodies against AnkG for co-staining (anti-mouse and anti-rabbit). We could not acquire sections from neonates or young infants, so we were not able to compare the expression at different time points. These two NaV antibodies were also tested in mouse sections and revealed very similar expression patterns compared to both our previous stainings with other NaV antibodies in mice, as well as to the stainings of human samples (data not shown).

Differential developmental expression of Nav1.2 and Nav1.6 on the mRNA level in the mouse hippocampus We used microdissected regions of the mouse hippocampus (CA1 and dentate gyrus) at two different time points of postnatal development (P8 and P30) to confirm our immunohistochemical results and analyse the relative gene expression of NaV1.2 and NaV1.6 with a different method on the mRNA level. Similar to the immunohistochemical studies, the results indicated that the amount of NaV1.2 mRNA decreases significantly from P8 to P30 (relative reduction by 3.3-fold) in the hippocampal CA1 region, whereas the NaV1.6 mRNA level largely increased by 5.5-fold. In the dentate gyrus, the NaV1.2 mRNA amount also decreased with maturation, but the difference did not reach statistical significance, and for NaV1.6 the increase was less pronounced (2.3-fold) than that in the CA1 region (Fig. 5).

Discussion Our combined genetic, electrophysiological, immunohistochemical and reverse transcription polymerase chain reaction studies provide a comprehensive analysis of the molecular mechanisms that may underlie the pathogenesis and the age dependence of an autosomal dominant idiopathic epileptic syndrome, BFNIS. The discussion is divided into two parts dealing with the mechanisms of (i) seizure generation and (ii) seizure remission.

Molecular correlates of age-dependent seizures

Brain 2010: 133; 1403–1414

| 1411

Figure 4 Developmental expression of NaV1.2 and NaV1.6 channels at axon initial segments in principal neurons of mouse brains. Immunohistochemical stainings of mouse brain slices in the CA1 region of the hippocampus in different stages of development (postnatal days P1 until P90). The Na+ channel  subunits NaV1.2 and NaV1.6 are stained with specific monoclonal antibodies (red fluorescence). Axon initial segments are co-stained with specific antibodies against Ankyrin G (green fluorescence); nuclei are stained with DAPI (blue fluorescence). The overlay shows all three stainings together. The anti-NaV1.2 and anti-NaV1.6 immunofluorescence signals suggest that NaV1.2 channels are expressed at the axon initial segments of CA1 pyramidal neurons early in development and that they are partially replaced with increasing maturation by NaV1.6. Scale bar 100 mm.

1412

| Brain 2010: 133; 1403–1414

Figure 5 Messenger RNA expression levels of NaV1.2 and NaV1.6 at two developmental time points. A quantitative analysis of the mRNA of NaV1.2 and NaV1.6 channels (corresponding to the genes SCN2A and SCN8A) in the mouse hippocampus at two different developmental stages was performed using a real-time reverse transcription–polymerase chain reaction approach. (A) In the CA1 region, a significant 3.3-fold reduction of the NaV1.2 (P 5 0.05) mRNA expression level was observed between P8 and P30 (values normalized to P8 were: 1.0  0.2 at P8 and 0.3  0.1 at P30), whereas the NaV1.6 mRNA expression level significantly increased (P50.05) between the two time points (from 1.0  0.1 at P8 to 5.5  2.0 at P30). (B) In the dentate gyrus (DG), the mRNA of NaV1.2 was also reduced from 1  0.3 at P8 to 0.3  0.1 at P30 (although not significantly), and the mRNA of NaV1.6 showed a significant upregulation (2.3-fold, P50.001, values normalized to P8: 1.0  0.2 at P8 and 2.3  0.2 at P30).

Mechanisms of seizure generation Our genetic and electrophysiological investigations strongly suggest that these mutations are the cause of the epilepsy in the affected individuals. Although Family 1 is relatively small, mutation c.754A4G predicting the amino acid exchange p.M252V co-segregated with the disease phenotype including one asymptomatic mutation carrier, whereas mutation c.781G4A predicting p.V261M clearly occurred de novo in the only affected individual of Family 2. The clinical variability of disease onset (neonatal in two and infantile in two of altogether four affected patients) and the finding of the first de novo mutation in this syndrome indicate that it is worthwhile to search for mutations in ‘sporadic’ patients or families when the clinical picture is typical for either benign familial neonatal seizures or BFNIS. Both mutations were not detected in a large number of normal controls and are strongly conserved both in evolution and among other voltage-gated Na+ channels suggesting their pathogenicity already based on genetic circumstantial evidence.

Y. Liao et al. The electrophysiological investigations then clearly indicated subtle but significant changes in channel gating, with several gain-of-function mechanisms that have been previously described in a similar way in other Na+ channel diseases of skeletal muscle, cardiac muscle or brain, including some mutations in NaV1.2 that had been described to be associated with BFNIS in several or with generalized epilepsy with febrile seizures plus in one family (Sugawara et al., 2001; George, 2005; Lerche et al., 2005; Cannon, 2006; Scalmani et al., 2006; Xu et al., 2007). Such changes predict an increase in membrane excitability of the respective cells. In our study, an increase of neuronal excitability by the two mutations can be explained by (i) a membrane depolarization due to an increase of the persistent Na+ inward current relative to the peak current or a slowing of fast inactivation; (ii) a shortening of the refractory period after an action potential by an acceleration of recovery from fast (and maybe slow) inactivation; and (iii) by an increase of the availability of Na+ channels at subthreshold voltages induced by a decreased slope of the activation curve. We tested this hypothesis in a simple one-compartment model based on the Hodgkin–Huxley theory. For both mutations, an increase in neuronal firing was predicted by the model. For M252V channels, which mainly increased the persistent current, we obtained a longer duration of burst firing, whereas mainly the altered slope of the activation curve led to a decrease of the action potential threshold and increased the firing rate for V261M compared to wild-type channels (Supplementary Fig. 9). Altogether, these results can explain the occurrence of epileptic seizures in the mutation carriers. From a biophysical point of view, our results demonstrate the importance of the S5 segment of domain D1 for fast inactivation of the NaV1.2 channel, which had not been described so far for any of the known voltage-gated Na+ channels. The finding is of interest, since domains D1 and D2 have been thought to be mainly responsible for the activation process of the channel, at least in the skeletal muscle Na+ channel (Cha et al., 1999). Another disease-causing mutation in the skeletal muscle Na+ channel with similar functional consequences has been described previously in the S5 segment of domain D4 (Bendahhou et al., 1999).

Mechanisms of seizure remission The mechanism of age dependence in BFNIS and other idiopathic epilepsies with a defined onset, and of remission of seizures during different phases of development had been unclear up to now. We used two different approaches to investigate this open question for BFNIS. When we first studied both mutations in the background of the known neonatal and adult splice variants of human NaV1.2 channels, we did not obtain a general answer applying to all mutations. The M252V mutant only revealed gating changes (increased persistent Na+ current) in the background of the neonatal splice variant, which could indeed be an explanation for the remission of seizures at later stages of development when the neonatal splice variant is no longer expressed. However, this did not apply to the V261M mutant, for which most of the observed gain-of-function changes (decreased slope of the activation curve and acceleration of recovery from fast inactivation) were observed in the background of both splice

Molecular correlates of age-dependent seizures variants, and two of the changes even occurred only for the adult channel (increased persistent current and slowing of fast inactivation). The relative proximity of both mutations in D1/S5 and the structural link via the D1/S4 voltage sensor to the extracellular D1/S3–S4 loop, containing the splice mutation N209D, may explain the distinct effects observed in a differential splice background on a molecular and biophysical level. One other mutation has also been investigated in the background of both splice variants (L1563V), and the results did not reveal an explanation for the remission of seizures (Xu et al., 2007). Therefore, it is likely that another more general mechanism is responsible for this characteristic clinical observation. In our second approach, we studied the developmental expression pattern of NaV1.2 channels in comparison to another voltage-gated Na+ channel, NaV1.6. Our results suggest that both channels, NaV1.2 and NaV1.6, are highly expressed at the axon initial segments of pyramidal cells in the hippocampus and cortex but that their developmental expression patterns differ markedly. Whereas NaV1.2 channels are expressed early in development with a decrease of the staining intensity and a redistribution towards the proximal part of the axon initial segments at least in some neurons during maturation, NaV1.6 channel expression increases with development more or less replacing NaV1.2 channels, in particular at the distal axon initial segments. A similar effect has been described previously in retinal ganglion cells (Boiko et al., 2001, 2003; Kaplan et al., 2001), but here we could show the differential developmental expression of specific voltage-gated Na+ channels at the axon initial segments of cortical and hippocampal excitatory neurons on both protein and mRNA level, and thereby provide a plausible and general explanation for the transient expression of seizures caused by gain-of-function NaV1.2 mutations in BFNIS. Our results thus support an intriguing hypothesis for the age dependence of an idiopathic epilepsy syndrome, which directly involves the mutated channels, here an age-dependent expression matching with the period in which the seizures occur. Of course, we need to assume that the expression pattern of these two investigated Na+ channels is transferable from mouse to man. Examination of human brain slices presents a challenge not only due to the unavailability of high quality sections from different developmental stages but also due to a limited number of antibodies that can specifically stain NaV channels in human tissue. Hence, although the results we obtained with stainings of human adult brain tissue may support our findings in mice, they have to be interpreted with care. Furthermore, the available data about a comparison between developmental stages of mouse and man suggest that the observed pattern in the mouse would match the neonatal and infantile period of man (Bayer et al., 1993; Clancy et al., 2001). An additional observation that also matches the results obtained previously in retinal ganglion cells (Boiko et al., 2001, 2003; Kaplan et al., 2001) was a developmental increase and persistence of NaV1.2 channel expression in adult ages in unmyelinated mossy fibres in which we found no evidence for an expression of NaV1.6. We therefore hypothesize that hyperexcitability of mossy fibres caused by gain-of-function NaV1.2 mutations does not induce epileptic seizures in patients with BFNIS, but that the transient expression in the axon initial segments of principal neurons is

Brain 2010: 133; 1403–1414

| 1413

responsible for the clinical phenotype. In agreement with this hypothesis, it has been described that mossy fibres mainly target inhibitory neurons (Acsady et al., 1998), so that a gain-of-function of Na+ channels expressed along mossy fibre axons could predominantly promote inhibition instead of triggering excitation. This may also explain why a SCN2A mutation with a complete loss of function, for which even a dominant-negative effect on the wild-type channel has been described, is associated with a severe epilepsy of infancy with a later onset, but not with neonatal or infantile seizures (Kamiya et al., 2004). It is noteworthy, however, that three recently reported de novo missense mutations in SCN2A, two of which have been shown to exert strong effects on channel gating with either gain or loss of function, can cause intractable epileptic encephalopathies (Ogiwara et al., 2009; Shi et al., 2009), so that the pathophysiology of severe SCN2A-associated syndromes still has to be elucidated. Knock-out mice of SCN2A did not unravel the possible differential function of this channel so far, since homozygous animals die early after birth and heterozygous ones were reported to develop normally without obvious epileptic seizures (Planells-Cases et al., 2000). Finally, an increasing expression of NaV1.2 in unmyelinated axons (not only mossy fibres) might explain why expression of this channel has been found to increase steadily from neonatal to adult in whole brain preparations (Gong et al., 1999), despite the observed decrease of NaV1.2 expression in principal neurons as observed here on both protein and local mRNA levels.

Conclusions Our studies and previous investigations (Scalmani et al., 2006; Xu et al., 2007) indicate that SCN2A (NaV1.2) mutations cause BFNIS by a gain-of-function mechanism increasing the membrane excitability at the axon initial segments, the site of action potential generation, in principal neurons of the hippocampus and cortex. This would fit with the clinical observation of partial onset seizures with a variable seizure semiology. Only one study reported a loss-of-function by a significantly decreased surface expression of mutated NaV1.2 channels in tsA201 cells (Misra et al., 2008), which was not observed in other studies. Our immunohistochemical and real-time reverse transcription–polymerase chain reaction investigations suggest a transiently higher expression of NaV1.2 channels at the axon initial segments of principal neurons during development and a replacement of NaV1.2 by another voltage-gated Na+ channel, NaV1.6, upon maturation, in particular in the distal part of the axon initial segments. This observation provides a plausible explanation for the age dependence of BFNIS. A differential regulation of the developmental expression of ion channels, modifying or interacting proteins that alter neuronal excitability, might also apply to explain other age-dependent epilepsy syndromes.

Acknowledgements We thank all patients and relatives for participating in the study, Dr Anna-Elina Lehesjoki and Dr Anna-Kaisa Anttonen for

1414

| Brain 2010: 133; 1403–1414

screening Finnish control individuals, GlaxoSmithKline for providing the clones of hb1 and hb2 subunits, and Dr J. Trimmer for providing antibodies before they became available commercially. We acknowledge the contribution of the VIB Genetic Service Facility to the genetic analyses.

Funding German Research Foundation (DFG Le1030/10-1, /8-2; SFB TR3, project C6), the National Genome Network of the Federal Ministry for Education and Research (BMBF: NGFN2/01GS0478 and NGFNplus/01GS08123, 01GS08122), the European Union (Epicure: LSH 037315), the University of Ulm, the Fund for Scientific Research Flanders (FWO-F), Methusalem excellence grant of the Flemish Government and University of Antwerp, and the National Health and Medical Research Council of Australia program grant (NHMRC #400121).

Supplementary material Supplementary material is available at Brain online.

References Acsady L, Kamondi A, Sik A, Freund T, Buzsaki G. GABAergic cells are the major postsynaptic targets of mossy fibers in the rat hippocampus. J Neurosci 1998; 18: 3386–403. Bayer SA, Altman J, Russo RJ, Zhang X. Timetables of neurogenesis in the human brain based on experimentally determined patterns in the rat. Neurotoxicology 1993; 14: 83–144. Bendahhou S, Cummins TR, Tawil R, Waxman SG, Ptacek LJ. Activation and inactivation of the voltage-gated sodium channel: role of segment S5 revealed by a novel hyperkalaemic periodic paralysis mutation. J Neurosci 1999; 19: 4762–71. Berkovic SF, Heron SE, Giordano L, Marini C, Guerrini R, Kaplan RE, et al. Benign familial neonatal-infantile seizures: characterization of a new sodium channelopathy. Ann Neurol 2004; 55: 550–7. Boiko T, Rasband MN, Levinson SR, Caldwell JH, Mandel G, Trimmer JS, et al. Compact myelin dictates the differential targeting of two sodium channel isoforms in the same axon. Neuron 2001; 30: 91–104. Boiko T, Van Wart A, Caldwell JH, Levinson SR, Trimmer JS, Matthews G. Functional specialization of the axon initial segment by isoform-specific sodium channel targeting. J Neurosci 2003; 23: 2306–13. Cannon SC. Pathomechanisms in channelopathies of skeletal muscle and brain. Annu Rev Neurosci 2006; 29: 387–415. Cha A, Ruben PC, George AL Jr, Fujimoto E, Bezanilla F. Voltage sensors in domains III and IV, but not I and II, are immobilized by Na+ channel fast inactivation. Neuron 1999; 22: 73–87. Clancy B, Darlington RB, Finlay BL. Translating developmental time across mammalian species. Neuroscience 2001; 105: 7–17. den Dunnen JT, Antonarakis SE. Nomenclature for the description of human sequence variations. Hum Genet 2001; 109: 121–4. Fink L, Seeger W, Ermert L, Hanze J, Stahl U, Grimminger F, et al. Realtime quantitative RT–PCR after laser-assisted cell picking. Nat Med 1998; 4: 1329–33. George AL Jr. Inherited disorders of voltage-gated sodium channels. J Clin Invest 2005; 115: 1990–9. Golomb D, Yue C, Yaari Y. Contribution of persistent Na+ current and M-type K+ current to somatic bursting in CA1 pyramidal cells:

Y. Liao et al. combined experimental and modeling study. J Neurophysiol 2006; 96: 1912–26. Gong B, Rhodes KJ, Bekele-Arcuri Z, Trimmer JS. Type I and type II Na(+) channel alpha-subunit polypeptides exhibit distinct spatial and temporal patterning, and association with auxiliary subunits in rat brain. J Comp Neurol 1999; 412: 342–52. Heron SE, Crossland KM, Andermann E, Phillips HA, Hall AJ, Bleasel A, et al. Sodium-channel defects in benign familial neonatal-infantile seizures. Lancet 2002; 360: 851–2. Hu W, Tian C, Li T, Yang M, Hou H, Shu Y. Distinct contributions of Na(v)1.6 and Na(v)1.2 in action potential initiation and backpropagation. Nat Neurosci 2009; 12: 996–1002. Kamiya K, Kaneda M, Sugawara T, Mazaki E, Okamura N, Montal M, et al. A nonsense mutation of the sodium channel gene SCN2A in a patient with intractable epilepsy and mental decline. J Neurosci 2004; 24: 2690–8. Kaplan MR, Cho MH, Ullian EM, Isom LL, Levinson SR, Barres BA. Differential control of clustering of the sodium channels Na(v)1.2 and Na(v)1.6 at developing CNS nodes of Ranvier. Neuron 2001; 30: 105–19. Kasai N, Fukushima K, Ueki Y, Prasad S, Nosakowski J, Sugata K, et al. Genomic structures of SCN2A and SCN3A—candidate genes for deafness at the DFNA16 locus. Gene 2001; 264: 113–22. Lerche H, Weber YG, Jurkat-Rott K, Lehmann-Horn F. Ion channel defects in idiopathic epilepsies. Curr Pharm Des 2005; 11: 2737–52. Lossin C, Wang DW, Rhodes TH, Vanoye CG, George AL Jr. Molecular basis of an inherited epilepsy. Neuron 2002; 34: 877–84. Maljevic S, Wuttke TV, Lerche H. Nervous system KV7 disorders: breakdown of a subthreshold brake. J Physiol 2008; 586: 1791–801. Misra SN, Kahlig KM, George AL Jr. Impaired NaV1.2 function and reduced cell surface expression in benign familial neonatal-infantile seizures. Epilepsia 2008; 49: 1535–45. Ogiwara I, Ito K, Sawaishi Y, Osaka H, Mazaki E, Inoue I, et al. De novo mutations of voltage-gated sodium channel alphaII gene SCN2A in intractable epilepsies. Neurology 2009; 73: 1046–53. Planells-Cases R, Caprini M, Zhang J, Rockenstein EM, Rivera RR, Murre C, et al. Neuronal death and perinatal lethality in voltagegated sodium channel alpha(II)-deficient mice. Biophys J 2000; 78: 2878–91. Scalmani P, Rusconi R, Armatura E, Zara F, Avanzini G, Franceschetti S, et al. Effects in neocortical neurons of mutations of the Na(v)1.2 Na+ channel causing benign familial neonatal-infantile seizures. J Neurosci 2006; 26: 10100–9. Shi X, Yasumoto S, Nakagawa E, Fukasawa T, Uchiya S, Hirose S. Missense mutation of the sodium channel gene SCN2A causes Dravet syndrome. Brain Dev 2009; 31: 758–62. Specchio N, Vigevano F. The spectrum of benign infantile seizures. Epilepsy Res 2006; 70(Suppl 1): S156–67. Sugawara T, Tsurubuchi Y, Agarwala KL, Ito M, Fukuma G, MazakiMiyazaki E, et al. A missense mutation of the Na+ channel alpha II subunit gene Na(v)1.2 in a patient with febrile and afebrile seizures causes channel dysfunction. Proc Natl Acad Sci USA 2001; 98: 6384–9. Vacher H, Mohapatra DP, Trimmer JS. Localization and targeting of voltage-dependent ion channels in mammalian central neurons. Physiol Rev 2008; 88: 1407–47. Vervaeke K, Hu H, Graham LJ, Storm JF. Contrasting effects of the persistent Na+ current on neuronal excitability and spike timing. Neuron 2006; 49: 257–70. Weckx S, Del-Favero J, Rademakers R, Claes L, Cruts M, De Jonghe P, et al. novoSNP, a novel computational tool for sequence variation discovery. Genome Res 2005; 15: 436–42. Xu R, Thomas EA, Jenkins M, Gazina EV, Chiu C, Heron SE, et al. A childhood epilepsy mutation reveals a role for developmentally regulated splicing of a sodium channel. Mol Cell Neurosci 2007; 35: 292–301.