Role of the hippocampus in Na_v1.6 (Scn8a) mediated seizure resistance

Abstract

SCN1A mutations are the main cause of the epilepsy disorders Dravet syndrome (DS) and genetic epilepsy with febrile seizures plus (GEFS+). Mutations that reduce the activity of the mouse Scn8a gene, in contrast, are found to confer seizure resistance and extend the lifespan of mouse models of DS and GEFS+. To investigate the mechanism by which reduced Scn8a expression confers seizure resistance, we induced interictal-like burst discharges in hippocampal slices of heterozygous Scn8a null mice (Scn8a^med/+) with elevated extracellular potassium. Scn8a^med/+ mutants exhibited reduced epileptiform burst discharge activity after P20, indicating an age-dependent increased threshold for induction of epileptiform discharges. Scn8a deficiency also reduced the occurrence of burst discharges in a GEFS+ mouse model (Scn1a^R1648H/+). There was no detectable change in the expression levels of Scn1a (Na_v1.1) or Scn2a (Na_v1.2) in the hippocampus of adult Scn8a^med/+ mutants. To determine whether the increased seizure resistance associated with reduced Scn8a expression was due to alterations that occurred during development, we examined the effect of deleting Scn8a in adult mice. Global Cre-mediated deletion of a heterozygous floxed Scn8a allele in adult mice was found to increase thresholds to chemically and electrically induced seizures. Finally, knockdown of Scn8a gene expression in the adult hippocampus via lentiviral Cre injection resulted in a reduction in the number of EEG-confirmed seizures following the administration of picrotoxin. Our results identify the hippocampus as an important structure in the mediation of Scn8a-dependent seizure protection and suggest that selective targeting of Scn8a activity might be efficacious in patients with epilepsy.

Introduction

Epilepsy is characterized by unprovoked, recurrent seizures that are manifestations of abnormal neuronal synchrony and excitability. Several genes that are known to cause monogenic forms of epilepsy encode neuronal ion channels, including voltage-gated sodium channels (VGSCs) (Noebels, 2003; Steinlein, 2004). VGSCs are important integrators of synaptic input and are responsible for the initiation and propagation of action potentials in neurons (Gong et al., 1999; Whitaker et al., 2001; Lorincz and Nusser, 2010). Three pore-forming VGSC α-subunit genes are primarily expressed in the adult mammalian central nervous system (CNS): SCN1A, SCN2A, and SCN8A, which encode the Na_v1.1, Na_v1.2, and Na_v1.6 channels, respectively. SCN1A has emerged as an important epilepsy gene and is responsible for a number of different epilepsy disorders, including the catastrophic, treatment-resistant childhood encephalopathy Dravet syndrome (DS) and genetic epilepsy with febrile seizures plus (GEFS+) (Escayg et al., 2000; Claes et al., 2001).

Approximately 30% of epileptic patients do not achieve adequate seizure control with currently available anti-epilepsy drugs (AEDs). Moreover, while new AEDs are generally better tolerated, the percentage of patients with treatment-resistant seizures has not changed significantly in the last 40 years, indicating a critical need to identify new treatment options for patients with refractory epilepsy. VGSCs are known targets for several AEDs; however, these drugs are not isoform specific and affect all VGSCs, likely contributing to unwanted side effects. Alternatively, selective targeting of a specific VGSC isoform, may provide improved seizure control with fewer unintended consequences.

We previously demonstrated that thresholds for flurothyl- and kainic acid-induced seizures are elevated in Scn8a^med/+ and Scn8a^medjo/+ mutant mice when compared to wild-type (WT) littermates (Martin et al., 2007). Furthermore, the seizure phenotypes of Scn1a mutant mice that model DS and GEFS+ were dramatically improved by the co-expression of an Scn8a mutation (Martin et al., 2007; Hawkins et al., 2011). In addition, Scn8a-deficient mice are resistant to amygdala kindling, while there is increased Scn8a expression in the CA3 region of the hippocampus of amygdala-kindled rats (Blumenfeld et al., 2009). Together, these studies raise the possibility that reduced hippocampal Scn8a expression may contribute to the seizure protection observed in Scn8a^med/+ mutants and that the selective targeting of Scn8a may be efficacious in some forms of epilepsy. However, a caveat of these earlier experiments is that the relationship between Scn8a and seizure resistance was based on the use of Scn8a mutant mice, in which the activity of Scn8a was reduced throughout brain development. Therefore, whether reduced Scn8a activity would confer seizure protection in the adult brain was unknown.

In this study, we examined the effect on seizure susceptibility of reducing Scn8a expression in adult mice. In addition to exploring the mechanism of seizure protection associated with altered Scn8a function, we used electrophysiological analysis to monitor developmental changes in hippocampal excitability in heterozygous Scn8a^med/+ mutants. Finally, using a lentiviral-Cre strategy, we assessed the effect of reducing Scn8a expression in the adult hippocampus on seizures generated following the administration of picrotoxin.

Materials and Methods

Animals

C3HeB/FeJ-Scn8a^med/J mice (Scn8a^med/+, Jackson Laboratory, Bar Harbor, stock number 003798) were maintained on the C3HeB/FeJ background. The med mutation is the result of a spontaneous LINE element insertion in the second exon of the Scn8a gene leading to truncation and loss-of-function of the channel (Kohrman et al., 1996). Scn8a^med/+ male mice were crossed to C3HeB/FeJ females to generate Scn8a^med/+ and wild-type (WT) Scn8a^+/+ littermate controls. Mice expressing the UBC-Cre-ER^T2 transgene (ER-Cre) were obtained from the Jackson Laboratory (stock number, 008085). Scn8a-floxed mice, in which exon 1 of the Scn8a gene is flanked by loxP sites (floxed allele), were a gift from Dr. Miriam Meisler at the University of Michigan, Ann Arbor (Levin and Meisler, 2004). The ER-Cre and Scn8a-floxed lines were maintained on the C57BL6/J background. Male ER-Cre mice were crossed to female Scn8a-floxed mice, producing progeny that express the floxed allele and the ER-Cre transgene (flox/+,ER-Cre). Controls consisted of offspring that express the floxed allele but lack the ERCre transgene (flox/+,no Cre), progeny that lack the floxed allele but carry the Cre transgene (+/+,ER-Cre), and progeny that lack the floxed allele and the ER-Cre transgene (+/+,no Cre). Vehicle-injected mice that carry both the floxed allele and the ER-Cre transgene were also used as controls (flox/+,ER-Cre, Veh.). Homozygous Scn8a-floxed mice that expressed the ER-Cre transgene (flox/flox,ER-Cre) were used to evaluate recombination efficiency. Homozygous floxed mice (flox/flox) were used for all experiments involving intrahippocampal viral Cre and GFP injections. Scn1a^R1648H/+ mice were generated as previously described (Martin et al., 2010). Double heterozygous mutants (Scn1a^R1648H/+,Scn8a^med/+) were generated by crossing Scn1a^R1648H/+ with Scn8a^med/+ mutant mice. All crosses yielded genotypes in expected Mendelian ratios. Littermates were used for all experiments to minimize confounds associated with differences in rearing conditions and genetic background. Experimenters were blinded to subject groups during data collection and analysis. Mice were maintained on a 12 h light/dark cycle. Food and water were available ad libitum. All experimental procedures were performed in accordance with the guidelines of Emory University and the University of California, Irvine Institutional Animal Care and Use Committees.

Genotyping

The presence of the Cre transgene and the Scn8a floxed allele was determined by PCR analysis using the primer pairs CreF/CreR (TGA CCC GGC AAA ACA GGT AGT TA / TTC CCG CAG AAC CTG AAG ATG TT) and FloxF/FloxR (GTG TGT GAT TCT CAA CAG TGG GTT / GTC TGT AAG AAG GCC TGA AAG TGA), respectively (Levin and Meisler, 2004). The Scn8amed mutation was identified as previously described (Martin et al., 2007) using the primer pair medF/medR (TCC AAT GCT TAT ACC AAA AGT CCC / GGA CGT GCA CAC TCA TTC CC). Scn1a^R1648H/+ mice were genotyped as previously described (Martin et al., 2010).

Tamoxifen-induced inactivation of Scn8a

Individually housed 2-month-old mice were intraperitoneally (i.p.) administered tamoxifen (TAM, 100 mg/kg, Sigma-Aldrich) or vehicle (corn oil, Sigma-Aldrich) for 5 consecutive days. TAM was prepared by dissolving in 100% ethanol and then mixing with 300 μl of corn oil. Excess ethanol was evaporated by vacuum centrifugation. Mice were left undisturbed for 15 days after the last injection.

Preparation and incubation of slices

Mice were anesthetized with halothane, decapitated, and their brains were placed in ice-cold sucrose artificial cerebrospinal fluid (ACSF) containing (in mM): 85 NaCl, 65 sucrose, 2.5 KCl, 25 glucose, 1.25 NaH₂PO₄, 4 MgSO₄, 0.5 CaCl₂, and 24 NaHCO₃. Transverse hippocampal slices (350 μm) were cut using a Leica VT1000S vibrating blade microtome (Leica, Germany). Slices were incubated at 33°C for at least 1 h before electrophysiological recordings in oxygenated (95% O₂, 5% CO₂) standard ACSF containing (in mM): 126 NaCl, 2.5 KCl, 1.25 NaHPO₄, 1.2 MgSO₄, 10 glucose, 1.2 CaCl₂, and 24 NaHCO₃.

Electrophysiological recordings

Slices were submerged in a recording chamber and continuously perfused at 2 mL/min with oxygenated ACSF at 33°C during electrophysiological experiments. Cells were visualized with an upright microscope (Zeiss Axioskop Plus) equipped with infrared differential interference contrast optics. Recording pipettes (2-5 MΩ) were pulled from borosilicate glass (Sutter Instruments, Novato, CA) with a P-87 Flaming-Brown puller (Sutter Instruments). Electrophysiological recordings were obtained in current clamp mode using MultiClamp 700B amplifier at 100X AC (Molecular Devices, Union City, CA) and digitized with a Digidata 1322A digitizer (Molecular Devices). Continuous extracellular recordings were acquired at 10 kHz and population spikes were recorded at 50 kHz. All recordings were low-pass filtered at 1 kHz for offline analysis with pClamp 10.2 software (Molecular Devices).

For extracellular recordings, pipettes were filled with 150 mM NaCl and positioned in the CA3 stratum pyramidal layer. ACSF with elevated potassium was prepared by supplementing standard ACSF with 3 M KCl to raise the potassium concentration to 8.5 mM [K⁺]_O. The experimental paradigm consisted of a control recording for 2-5 min in standard ACSF, followed by 20-25 min with 8.5 mM [K⁺]_O, in which burst activity was recorded. To evoke population spikes, constant-current stimuli (100 μsec) were applied via stimulus isolator (A365D; WPI, Sarasota, FL) using a bipolar tungsten electrode (50μm in diameter; A-M Systems, Carlsborg, WA) positioned in the mossy fiber tract. Population spikes were evoked in physiological 2.5 mM [K⁺]_O before and after exposing the slice to 8.5 mM [K⁺]_O. Slices in which population spike amplitudes did not change more than 20% after exposing the slice to 8.5 mM [K⁺]_O were used for analysis.

Protein extraction

Hippocampi were dissected and frozen at −80°C. To determine developmental changes in Scn8a expression, individual hippocampi were pooled from 10-15 mice at each age. Hippocampi from adult, 2-3-month-old, mice were not pooled. Hippocampi were homogenized on ice using a Dounce homogenizer in homogenization buffer (320 mM sucrose, 5 mM Na phosphate, pH 7.4, 100 mM Na fluoride and protease inhibitor (Pierce, Rockford, IL) or 50 mM Tris, pH = 7.5; 10 mM EGTA, and one complete EDTA-free protein inhibitor tablet (Roche, Germany) per 50 mL solution. Homogenates were centrifuged at 750 × g for 10 min at 4°C, and the supernatant was centrifuged at 39,000 × g for 30 or 60 min at 4°C. The pellet was suspended in protease inhibitor buffer and stored at −80°C. BCA (Thermo Fisher Scientific, Waltham, MA) or Bradford (Bio-Rad, Hercules, CA) protein assays were used to normalize protein concentrations across samples.

Western blot analysis

Membrane-enriched whole-brain or hippocampal tissue homogenates (15-30 ug) were subjected to SDS-PAGE electrophoresis. After blocking in 5% milk, blots were incubated overnight at 4°C in either polyclonal rabbit anti-Na_v1.6 primary antibody (1:200, Millipore, Billerica, MA), polyclonal rabbit anti-Na_v1.6 (1:225, Alomone, Israel), polyclonal rabbit anti-Na_v1.1 (1:200, Millipore), or monoclonal mouse anti-Na_v1.2 (1:1000, Neuromab, Davis, CA). Blots were then incubated in either HRP-conjugated donkey anti-rabbit secondary (1:10,000, GE Healthcare, United Kingdom), HRP-conjugated goat anti-mouse secondary (1:10,000, Jackson ImmunoResearch, West Grove, PA), or HRP-conjugated goat anti-rabbit secondary (Sigma, St. Louis, MO, 1:16,000) for 1 h followed by washing in SuperSignal West Pico Chemiluminescent substrate (Pierce) and imaging. Blots were also probed using a monoclonal mouse anti-α-tubulin (1:10,000, Millipore) or monoclonal mouse anti-pan-cadherin (1:100,000, Sigma) antibody followed by HRP-conjugated goat anti-mouse secondary (1:10,000, Jackson ImmunoResearch) or HRP-conjugated goat anti-mouse secondary (Pierce, 1:26,000) for normalization of sample loading. Image quantification was performed using ImageJ software (NIH).

Quantitative real-time PCR

Total RNA was extracted from the hippocampus of WT and Scn8a^med/+ adult mice using the RNeasy Lipid Tissue Mini Kit (Qiagen, Germany). All RNA samples were quantified using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific), and RNA quality was determined by agarose gel electrophoresis. RNA was reverse-transcribed into cDNA using random hexamer primers and SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA). Scn1a (F: TCAGGAGGAAGGGGTTTCGCTTC, R: CCCCACATCCTTGGCTCGCCCTC) and Scn2a (F: CTGCAACGGTGTGGTCTCCCTAG, R: ATGTAGGGTCTTCCAACAAGTCC) primers were designed to span introns to avoid amplification of genomic DNA. Each primer pair generated standard curves with efficiencies of 90-100%. Real-time PCR (qPCR) data were generated from cDNA samples in technical triplicate using the BioRad CFX96 Real-Time PCR Detection System and SYBR Green fluorescent dye (BioRad). Expression levels were normalized to beta-actin (F: CAGCTTCTTTGCAGCTCCTT, R: ACGATGGAGGGGAATACAGC). Expression analysis and statistics were performed using the GenEx 5 software (MultiD).

Lentiviral constructs and virus

Previously described viral vectors expressing green fluorescent protein (LV-GFP) or Crerecombinase (LV-Cre) were obtained from the Emory University Viral Core (Heldt et al., 2007). Briefly, LV-GFP viral vectors were derived from the pLV-CMV-GFP-U3Nhe backbone. LV-Cre viral vectors were created by replacing the GFP coding sequence with Cre-recombinase coding sequence using the BamH1/Sal1 restriction sites in the pLV-CMV-GFP-U3he backbone. Viral production procedures have been described previously (Miyoshi et al., 1998; Pfeifer et al., 2001). Viral titer of the LV-GFP and LV-Cre was assessed in HEK-293T cells and ranged from 1 × 10⁸ to 1 × 10⁹ infectious particles per ml.

Surgical procedures

Surgical implantation of EEG electrodes was performed as previously described (Martin et al., 2007; Dutton et al., 2012; Papale et al., 2013). Mice were anesthetized by isoflurane inhalation, and then fixed in a stereotaxic apparatus (Kopf, Tujunga, CA). Four holes were drilled in the skull above each injection site. Injections were performed using a Hamilton 0.5-μl microsyringe (model #75) and 30-gauge needle. Four injection sites were chosen to target the hippocampus at the following coordinates, from bregma: anteroposterior (AP), -1.9 mm; mediolateral (ML), ±1.1 mm, ±2.1 mm; dorsoventral (DV), -1.9 mm. The syringe was lowered to -2.0 mm, and then retracted to -1.9 mm and left in place for 4 min. Virus solution (1.0 μl) was injected into each site at a rate of 0.12 μl/min. After each injection, the syringe was left in place for 4 minutes before being retracted. Following viral injections, 4 sterile cortical screw electrodes (Vintage Machine Supplies, OH) were implanted subdurally in the skull for EEG recordings at the following coordinates, from bregma: 1.5 mm (AP) and +2.1 mm ML; -1.9 mm AP and ±2.1 mm ML; 0.5 mm AP and -2.1 mm ML. Fine-wire electrodes were implanted into the neck muscle for EMG acquisition. Incisions were closed with tissue adhesive (Vetbond, 3M; St. Paul, NM), and the animal was allowed to recover on a heating pad. Postoperative analgesic (ibuprofen, 0.1 mg/kg) was provided for 3 days in the drinking water.

Histological procedures

Mice were anesthetized using isoflurane and transcardially perfused with 4% paraformaldehyde (PFA) 15-20 d following LV-GFP injection. Brains were post-fixed for 2 h in ice-cold 4% PFA, and then transferred to 30% sucrose for 3 d at 4°C. Coronal sections (45 μm) were sliced using a cryostat (Leica) between bregma -1.0 mm and -4.0 mm. Sections were washed in TBS/Triton-X and mounted. Images were collected using a Leica DM6000 B upright fluorescence microscope and Simple PCI software.

Seizure induction paradigms

Flurothyl seizure induction

Flurothyl seizure induction was performed as previously described (Martin et al., 2007; Martin et al., 2010). Briefly, 3-month-old mice were placed in a clear Plexiglas chamber. Flurothyl (2,2,2-trifluroethyl ether, Sigma-Aldrich) was introduced at a rate of 20 μl/min. The latencies to the first myoclonic jerk (MJ) and generalized tonic-clonic seizure (GTCS) were recorded. The MJ was defined as a jerking movement of the neck and shoulders sometimes associated with tail clonus. The GTCS was defined as complete loss of postural control associated with forelimb and hindlimb tonic-clonic movement lasting longer than 3 s.

Kainic acid (KA) seizure induction

Three-month-old mice were administered KA (i.p., 25 mg/kg, Sigma-Aldrich) and observed for 2 h. The latency to the first occurrence of each event on a modified Racine scale was recorded: 1 = freezing, 2 = head nodding, 3 = tail clonus, 4 = forelimb clonus, 5 = rearing and falling, 6 = GTCS. The GTCS was defined as complete loss of postural control associated with forelimb and hindlimb tonic-clonic movement lasting longer than 3 s.

6-Hz psychomotor seizure induction

A brief electric current (6-Hz, 26 mA, 3-ms pulse width, 3-s duration) was applied to the cornea of 3-month-old mice (Ugo Basile S.R.L, Germany). Seizure severity was recorded using the following modified Racine scale: 1 = staring, 2 = forelimb clonus, 3 = rearing and falling. Mice that did not exhibit seizure activity in response to the first exposure were excluded from the study and did not proceed to receive TAM or vehicle. Those mice that exhibited seizure activity were placed into two groups for TAM or vehicle administration. These groups were balanced based on seizure severity to ensure that seizure susceptibility was comparable between groups before administration of TAM or vehicle. TAM or vehicle injections were initiated one day after the first current exposure. Two weeks after the last TAM injection, the mice underwent 6-Hz seizure induction a second time.

Picrotoxin (PT) seizure induction

. Fifteen days following either LV-Cre or LV-GFP hippocampal injection and surgical EEG implantation, PT seizure induction was performed as previously described (Distler et al., 2013). Video/EEG recordings were collected and amplified using the Stellate Harmony System and software (Natus Medical, Inc., San Carlos, CA). Signals were filtered (EEG 0.3-35 Hz bandpass, EMG 10-70 Hz bandpass) and digitized at a sampling rate of 200 Hz. Twenty min of baseline video/EEG recording was collected before administering PT. Following PT administration (i.p., 10 mg/kg, Sigma-Aldrich), the latency to the first seizure, duration of each seizure, and number of seizures were recorded. Seizures were defined behaviorally by the presence of generalized convulsions and confirmed by the presence of highly synchronous EEG activity that reached at least twice the baseline amplitude in all cortical EEG channels and persisted for at least 3 s. Video/EEG monitoring was performed for at least 2 h following PT administration.

Statistics

All bar graphs indicate the mean and all error bars represent ± standard error of the mean (SEM). Statistical analyses were performed using SigmaPlot 11.0 software (Systat Software Inc., Chicago, IL), Prism 6 (GraphPad Software Inc, La Jolla, CA), and GenEx 5 software (Multi D). The Levene's test was used to assess for differences in the variance between groups, and the Shapiro-Wilk test was used to determine normality of continuous data sets. One-way analysis of variance (ANOVA) was used to evaluate seizure thresholds between ER-Cre groups treated with flurothyl. Two-way analysis of variance (ANOVA) was used to evaluate KA seizure thresholds. Post-hoc comparisons were performed using either the Dunnett's or Bonferroni test. The Student's two-tailed t test was used to analyze the percentage knockdown of Scn8a protein levels by the ER-Cre transgene and by LV-Cre injection, as well as to compare qPCR results between genotypes for each sodium channel. qPCR expression analysis and statistics were performed using the GenEx 5 software. Student's two-tailed t test was also used to determine the significance of burst discharge latency, duration, and frequency, and Fisher's exact test with Bonferroni correction was used to compare the occurrence of burst discharges. Fisher's exact test was used to assess categorical seizure outcomes following 6-Hz seizure induction. The Mann-Whitney-U test was used to assess PT seizure outcomes.

Results

Epileptiform burst discharge activity is reduced in Scn8a^med/+ mutant mice

Since the hippocampus is known to play an important role in seizure generation, we examined temporal changes in the excitability of hippocampal slices from heterozygous Scn8a null mutant mice (Scn8a^med/+). Epileptiform burst discharges were induced in the CA3 pyramidal cell layer of the hippocampus by elevating the extracellular potassium concentration (Fig. 1Aa-c). Burst discharges were not observed during the initial 5 minutes at physiological potassium concentration (2.5 mM [K⁺]_o) in hippocampal slices from either WT (Fig. 1Aa) or Scn8a^med/+ mutant mice (Fig. 1B). However, elevating the extracellular K⁺ concentration to 8.5 mM induced synchronous burst discharge activity in hippocampal slices from most WT mice at all ages (P13-P84) (Fig. 1A,D). Individual burst discharges were composed of a positive extracellular potential with multiple population spikes (Fig. 1Ac). Burst activity continued until the extracellular K⁺ was returned to physiological concentration (2.5 mM). Similarly, hippocampal slices from young P13-P20 Scn8a^med/+ mutant mice displayed burst discharges when exposed to elevated potassium (Fig. 1D). In contrast, burst discharges were not observed in the majority of slices from Scn8a^med/+ mutant mice older than P20 (Fig. 1B,D). The lack of activity was not due to poor health of the cells because synaptic connections and neuronal firing remained intact, as determined by stable population spike amplitudes in the CA3 pyramidal layer before and after exposure to elevated potassium (Fig. 1C). The frequency of observed epileptiform burst discharges was lower in more mature WT mice (P26 and P70-84), but was still comparably much higher than that observed in slices from Scn8a^med/+ mutant mice (Fig. 1D). For example, burst discharges were observed in 86% of slices from adult WT mice, but in only 12% of slices from adult Scn8a^med/+ mice (P70-84).

Figure 1.

Spontaneous seizure-like burst discharges in CA3 pyramidal cell layer induced by elevated potassium in hippocampal slices from Scn8a^med/+ mutant and WT mice.

Aa, CA3 pyramidal cells did not show spontaneous burst activity in physiological 2.5 mM [K⁺]_o; however, burst activity was induced when extracellular K⁺ concentration was elevated to 8.5 mM [K⁺]_o. Ab, Expansion of a segment of spontaneous activity showed synchronous firing of burst activity. Ac, Expansion of a single interictal-like burst discharge. Arrow indicates start of a burst discharge and star shows the end of a single burst. B, Extracellular recording from CA3 pyramidal cell layer of a Scn8a^med/+ mutant mouse (P24). C, Population spike amplitude before and after exposing hippocampal slice to 8.5 mM [K⁺]_o. Solid gray – before perfusing 8.5 mM [K⁺]_o into the bath. Black dash – after exposing slice to 8.5 mM [K⁺]_o and washout with standard 2.5 mM [K⁺]_o ACSF. Population spikes were evoked with 50 αA stimulation for 100 μs. D, Hippocampal slices from Scn8a^med/+ mutant and WT mice from P13-P20 demonstrated comparable levels of spontaneous burst discharge activity in elevated extracellular potassium. Beginning at P21, the propensity for burst activity dramatically decreased in hippocampal slices from Scn8a^med/+ mutant mice. Number of slices = 4-17, number of mice = 2-7. *p < 0.001 when compared to WT; Fisher's exact test with Bonferroni correction.

Latency to burst discharges is increased in Scn8a^med/+ mutant mice

Because Na_v1.6 channels are well suited for repetitive firing of action potentials and hence hyperactivity of the neuronal network, we compared the latency to the initiation of burst discharges in slices from Scn8a^med/+ mutants and WT littermates. Latency to the initial burst discharge was measured from the time that elevated potassium ACSF entered the recording chamber. Scn8a^med/+ mutant mice in all 3 age groups tested displayed a significant increase in the latency to the initiation of epileptiform burst activity (Fig. 2A). However, once burst discharge activity was induced, there were no qualitative differences in the pattern of synchronous activity. To determine whether there were any quantitative differences in burst activity, we analyzed the inter-burst frequency and single burst duration. Because the majority of hippocampal recordings from Scn8a^med/+ mutant mice did not exhibit burst activity (Fig. 1D), only hippocampal slices from the few Scn8a^med/+ mutant mice that produced burst discharges were used for this analysis. Inter-burst frequency was measured as the number of burst discharges during a 4-min period, and burst duration was determined from an average of 50 individual bursts (Fig. 1Ac). There were no significant differences in either the inter-burst frequency or single burst duration between Scn8a^med/+ mutant and WT mice in all 3 age groups examined (Fig. 2B-C).

Figure 2.

Analysis of burst characteristics between Scn8a^med/+ and WT mice.

A, Hippocampal slices from Scn8a^med/+ mutants demonstrate longer latencies to the onset of burst activity compared to slices from WT mice for all age groups (WT vs mutant at P13-P15: 2.7 ± 0.2 min vs 3.8 ± 0.2 min, p < 0.05; WT vs mutant at P24-P26: 6.1 ± 0.8 min vs 11.1 ± 1.2 min, p < 0.01; WT vs mutant at P70-P84: 6.9 ± 0.9 min vs 9.2 ± 1.2 min, p <0.05). Comparison of burst activity showed similar inter-burst frequency B, and burst duration C, between slices from Scn8a^med/+ mutants and WT mice. P13-P15, n = 4-5; P24-P26, n = 6-9; P70-P84, n = 3-5; Student's two-tailed t test, *p < 0.05; **p < 0.001; error bars indicate SEM.

Reduced Scn8a expression decreases the occurrence of burst discharges in Scn1a^R1648H/+ mice

To investigate whether reduced Scn8a expression could alter the occurrence of epileptiform burst discharges in a mouse model of GEFS+ (Scn1a^R1648H/+ mice), we generated double heterozygous mutants (Scn1a^R1648H/+/Scn8a^med/+) and quantitatively examined burst activity in hippocampal slices exposed to elevated potassium. Burst occurrence, defined as the number of slices that exhibited bursting, was significantly lower in slices from Scn8a^med/+ and double heterozygous mice compared to WT (Table 1). Burst occurrence was also significantly reduced in double heterozygous mice compared to Scn1a^R1648H/+ mice (Table 1). While burst occurrence was comparable in slices from Scn1a^R1648H/+ mutants and WT littermates, burst latency was significantly shorter in Scn1a^R1648H/+ mice. An increase in inter-burst frequency was also observed in slices from Scn1a^R1648H/+ mutants compared to WT; however, this difference was not statistically significant. It was not possible to statistically evaluate burst latency or frequency in slices from Scn8a^med/+ mice (with or without the Scn1a^R1648H/+ mutation) because of the low percentage of hippocampal slices that demonstrated burst activity. However, increased latencies to bursting and decreased inter-burst frequencies were observed in Scn8a^med/+ slices (with or without the Scn1a^R1648H/+ mutation) compared to Scn1a^R1648H/+ slices in the few instances in which bursting occurred.

Table 1.

Genotype	Burst Occurrence (# of slices with bursting/total # of slices)	Burst Latency (min)	Inter-burst Frequency (Hz)
WT	9/9	7.3 ± 0.4 (9)	31.8 ± 2.9 (9)
Scn1aRH/+	7/9	4.5 ± 0.3 (7)†	41.9 ± 6.2 (7)
Scn8amed/+	2/14*	9.4, 12.7 (2)	20.2, 17.0 (2)
Scn1aRH/+; Scn8amed/+	1/13*#	6.3 (1)	14.3 (1)

Developmental expression profile of Na_v 1.6 channels in the hippocampus

Because decreased burst activity was seen only in slices from Scn8a^med/+ mutant mice older than P20, we investigated whether this decrease was correlated with changes in developmental expression of Na_v1.6 channels by performing Western blot analysis. Na_v1.6 was detectable in the hippocampus as early as P10 and progressively increased until adult levels were reached between 3-4 weeks after birth (Fig. 3A). A high level of Na_v1.6 expression is observed at P18 prior to the time-point (P21) at which decreased burst activity is observed. To determine whether reduced Scn8a expression in med mutants is associated with altered levels of other VGSC alpha subunits, we compared Scn1a and Scn2a mRNA and protein levels from hippocampi of 3-month-old Scn8a^med/+ and WT littermates. Consistent with previous observations (Kearney et al., 2002), we found no significant differences in expression levels (Fig. 3 B-C).

Figure 3.

Developmental expression of VGSCs in the hippocampus of P10-P30 WT mice.

A, Western blot analysis of hippocampal membrane fractions from WT mice using rabbit anti-Na_v1.6 polyclonal antibody. Pan-cadherin (135 kDa) was used as a loading control. Statistical analysis of Na_v1.6 levels during development was performed using one-way ANOVA; * p < 0.001. Post hoc analysis demonstrated that Na_v1.6 levels in the hippocampus reach adult levels at 3-4 weeks after birth (p<0.05, Holm-Sidak post hoc, n=3 per age group). Hippocampal tissues from WT and Scn8a^med/+ mice were analyzed by Western blotting (WB) and qPCR for alterations in the expression of B, Scn1a and C, Scn2a. Alpha tubulin (50 kDa) was used a loading control. No significant differences in protein or RNA levels were observed. Scn1a and Scn2a protein and RNA expression levels were analyzed by the Student's two-tailed t test, p>0.05. WB, n = 7 per group; qPCR n = 4-5 per group. Error bars indicate SEM.

Cre recombinase induction leads to deletion of the floxed allele and reduced Na_v 1.6 levels

To examine the effect of reducing Scn8a expression on seizure thresholds in adult mice, we crossed a tamoxifen (TAM)-inducible Cre line (ER-Cre) with Scn8a floxed mice to generate progeny in which inactivation of the Scn8a gene could be achieved by TAM injection. PCR analysis performed on hippocampal tissue from flox/+,ER-Cre mice 15 d after the last TAM injection detected a 300-bp PCR band, indicating deletion of exon 1 of the Scn8a gene (Fig. 4A). This PCR product was not observed in mice that were not administered TAM, or in mice that either lacked the floxed Scn8a allele or the Cre transgene (Fig. 4A). Reduced optical density (78% reduction) of the Scn8a band by Western blot analysis demonstrated reduced expression of the Na_v1.6 protein in flox/flox,ER-Cre mice that were administered TAM compared to flox/flox,ER-Cre mice that were administered vehicle (Fig. 4B-C).

Figure 4.

Tamoxifen-induced global deletion of the Scn8a gene.

A, PCR product amplified from dissected neocortical tissue from flox/+,ER-Cre,TAM mice as well as control littermates (flox/+,ER-Cre,Veh) demonstrated that deletion is only detectable following TAM injection in mice that express the ER-Cre transgene and the floxed Scn8a allele (third lane). Additionally, the Scn8a floxed allele was only marginally detectable following TAM-initiated deletion, providing further evidence of efficient deletion. B, Western blotting demonstrated significant (79%) knockdown of Na_v1.6 protein in homozygous floxed mice carrying the ER-Cre transgene following TAM injection (flox/flox,ER-Cre,TAM) compared to flox/flox,ER-Cre mice administered vehicle. n = 3 per group; Student's two-tailed t test, *p < 0.05; Error bars indicate SEM. C. Representative Western blot from vehicle and TAM treated flox/flox,ER-Cre mice demonstrates reduced band intensity in the TAM treated group.

Inactivation of Scn8a leads to increased resistance to chemically and electrically induced seizures

Latencies to flurothyl-induced seizures following TAM-induced inactivation of the Scn8a allele were compared between Scn8a-deficiant animals (flox/+,ER-Cre,TAM) (white bar, Fig. 5A) and control littermates with normal levels of Scn8a (+/+,no Cre,TAM; +/+,ER-Cre,TAM; flox/+,no Cre,TAM; and flox/+,ER-Cre,Veh) (black bars, Fig. 5A). Myoclonic jerk latency was reduced in +/+,ER-Cre,TAM; flox/+,no Cre,TAM; and flox/+,ER-Cre,Veh animals but not +/+,no Cre,TAM mice compared to flox/+,ER-Cre,TAM animals. Significantly increased latencies to the GTCS were observed between Scn8a-deficient mice when compared to all control groups (Fig. 5A).

Figure 5.

Inducible deletion of the Scn8a gene in the adult mouse is sufficient to increase the latency to flurothyl- and KA-induced seizures and to protect against electrically induced 6-Hz psychomotor seizures.

A, Latencies to flurothyl-induced myoclonic jerk (MJ) and generalized tonic-clonic seizures (GTCS) were increased following TAM-induced inactivation of the Scn8a gene (flox/+,ERCre,TAM) (white bar) compared to control littermates (+/+,no Cre,TAM; +/+,ER-Cre,TAM; flox/+,no Cre,TAM; and flox/+,ER-Cre,Veh) (black bars). n = 10-12 per group. MJ and GTCS, one-way ANOVA, p < 0.001; Dunnett's post hoc, p < 0.05. B, Latencies to KA-induced seizure events were increased following Scn8a deletion (flox/+,ER-Cre,TAM) (solid line) when compared to vehicle-injected controls (flox/+,ER-Cre,Veh) (dashed line). Seizure activity was scored using a modified Racine scale: 1 = freezing, 2 = head nodding, 3 = tail clonus, 4 = forelimb clonus, 5 = rearing and falling, 6 = GTCS. n = 12-13 per group. Two-way ANOVA, p < 0.01; Bonferroni post hoc, p < 0.05. n = 12-13. C, 6-Hz psychomotor seizure severity was decreased after TAM-initiated inactivation of Scn8a (flox/+,ER-Cre,TAM) (solid line) when compared to vehicle-injected controls (flox/+,ER-Cre,Veh) (dashed line). n = 7-9 per group. Mann-Whitney-U test, p < 0.05. Seizure activity was scored using the following modified Racine scale: 1 = freezing, 2 = head nodding 3 = rearing and falling. Average Racine scores: flox/+,ERCre,Veh. pre = 2.1, post = 2.0; flox/+,ER-Cre,TAM pre = 2.0, post = 0.4. D, There was a trend toward reduced 6-Hz psychomotor seizure activity following inactivation of Scn8a (flox/+,ERCre,TAM) when compared to vehicle-injected controls (flox/+,ER-Cre,Veh). n = 7-9 per group; Fisher's exact test, n.s. p = 0.07. *p < 0.05; **p < 0.01; error bars indicate SEM

Latencies to KA-induced seizures were also compared between Scn8a-deficient (flox/+,ERCre,TAM) mice and control (flox/+,ER-Cre,Veh) mice. Latency to the first occurrence of each seizure stage on a modified Racine scale was recorded. Significantly increased latencies to stage 5 and stage 6 were observed in the flox/+,ER-Cre,TAM mice when compared to control littermates (Fig. 5B).

The 6-Hz model of psychomotor seizures was used to determine the effect of reducing Scn8a expression on electrically induced seizures. The severity of evoked seizures in Scn8a^flox/+,ER-Cre mice before and after TAM or vehicle administration was recorded using a modified Racine scale. A significant reduction in seizure severity was observed after TAM injection but not after vehicle injection (Fig. 5C). A trend toward a reduction in the percentage of mice exhibiting seizure activity was also observed following TAM administration, with 43% (3/7) of mice exhibiting seizures after TAM administration, compared to 89% (8/9) of mice exhibiting seizures after vehicle administration (Fig. 5D).

Lentiviral Cre-induced inactivation of the Scn8a gene in the hippocampus reduces seizure generation in vivo

To determine whether restricting Scn8a inactivation to the hippocampus would be sufficient to increase seizure resistance in vivo, LV-Cre or LV-GFP were injected into the hippocampus of flox/flox mice. GFP-positive cells were observed in the CA1, CA3, and DG layers of the hippocampus with highest expression in the DG and CA3 (Fig. 6A). Responses to picrotoxin (PT) were measured 15 days after injection of the lentiviral constructs. We observed a statistically significant reduction in the number of seizures in LV-Cre-injected mice compared to LV-GFP-injected controls (Fig. 6B); however, the average latency to the first seizure and the average seizure length were not significantly altered (Fig. 6C-D). Representative EEG recordings of baseline and seizure activity are shown for LV-GFP-injected controls and LV-Cre-injected mice (Fig. 6E).

Figure 6.

Lentiviral Cre-mediated knockdown of Scn8a in the hippocampus reduces seizure activity following PT administration.

Scn8a^flox/flox mice were injected with either LV-GFP or LV-Cre to achieve knockdown of Scn8a in the hippocampus. A, GFP-positive cells were observed in the CA1, CA3, and DG layers of the hippocampus with highest expression in the DG and CA3. Images were captured at 4X and 40X magnification (inset), scale bar = 100 μm. B, Fewer seizures were recorded in LV-Cre- compared to LV-GFP-treated mice following PT administration. No significant difference in the latency to the first seizure C, or the average duration of each seizure event D, was observed in LV-Cre- compared to LV-GFP-injected mice. E, Representative examples of baseline EEG and seizure activity indicating normal baseline EEG activity in both GFP and Cre groups and seizure activity in Cre and GFP groups following PT administration. n = 6-7 per group. Mann-Whitney-U test, *p < 0.05. Error bars indicate SEM.

Discussion

Previous studies examining the relationship between Scn8a and seizure resistance relied on the use of mice that expressed Scn8a mutations throughout development. To determine whether seizure protection would also be observed if Scn8a expression was reduced only in the adult brain, we used a TAM-inducible Cre transgenic line to initiate the inactivation of a floxed Scn8a allele in adult mice. We found that global reduction of Scn8a expression in the adult brain was capable of raising latencies to flurothyl- and KA-induced seizures. In addition, we found that reduced Scn8a expression in adult mice increased resistance to 6-Hz psychomotor seizures, a paradigm used to screen for novel anti-epileptic drugs (Brown, 1953; Barton et al., 2001). These results show that seizure protection can be achieved by reducing Scn8a expression in the adult brain, which is a critical requirement for an anticonvulsant therapy. This finding and the other main results of this study are summarized in Table 2.

Table 2.

Summary of main findings

Scn8a Genotype	Scn1a Genotype	Experimental System	Observation
med/+	+/+	Hippocampal slices with high K+	Reduced bursting from P21
med/+	R1648H/+	Hippocampal slices with high K+	Decreased bursting in a model of GEFS+
flox/+	+/+	Scn8a deletion in adult followed by seizure induction	Increased resistance to flurothyl, kainic acid, and 6Hz-induced seizures
flox/+	+/+	Lentiviral Cre-mediated hippocampal knockdown of Scn8a followed by seizure induction	Reduced number of picrotoxin- induced seizures

To test the hypothesis that reduced Scn8a expression leads to reduced hippocampal excitability and seizure suppression, we studied hippocampal slices from Scn8a^med/+ and WT littermates in a high-potassium model of epileptiform activity throughout postnatal development. We found that Scn8a^med/+ hippocampal slices were resistant to interictal-like burst discharges beginning at P20. There was an increase in the latency to epileptiform burst discharges in Scn8a^med/+ slices, suggesting that reduced Scn8a expression inhibits the initiation of bursting activity; however, there were no changes in inter-burst frequency or single burst duration relative to slices from WT mice. Our data indicate that there is a deficiency in epileptiform burst discharge initiation but not burst discharge propagation since epileptiform activity progresses comparably between WT and Scn8a-deficient slices following initiation. Additionally, the abrupt reduction in burst percentage in Scn8a-deficient slices at P21 is remarkable and cannot fully be explained by developmental changes in Scn8a expression which occurs over the course of several weeks. P21 in the mouse corresponds to a developmental stage at which neuronal excitability is likely to be influenced by a number of processes. For example, Na_v1.6 replaces Na_v1.2 during this time period and myelination is at a maximum at postnatal day 20 (Norton and Poduslo, 1973). We speculate that the effect of these changes on neuronal excitability in combination with reduced Scn8a activity likely contributes to the abrupt reduction in neuronal excitability in Scn8a-deficient animals.

Na_v1.6 channels have characteristic biophysical properties, including persistent current, resurgent current, and resistance to inactivation, which makes this isoform well suited for the sustained high-frequency firing that occurs at the axon initial segment and nodes of Ranvier, but may also allow this channel to facilitate the propagation of high-frequency ictal activity (Raman et al., 1997; Zhou and Goldin, 2004). Despite these unique characteristics, Scn8a^med/+ mutant mice showed no significant changes in inter-burst frequency or single burst duration. It is possible that there are changes in action potential firing that we could not detect because they are below the sensitivity of our assay; or, it may be that the decreased seizure activity is a property of the overall network and therefore does not reflect simple changes in bursting in any one region. Prior studies have shown that mice with altered Scn8a expression have reduced persistent current and depolarized action potential spike thresholds in hippocampal pyramidal neurons that may impede the firing of repetitive action potentials (Royeck et al., 2008; Blumenfeld et al., 2009). These biophysical changes are consistent with the longer latency to the initial burst discharge and higher seizure threshold in Scn8a^med/+ mutants. These results suggest that reduced Scn8a expression leads to reduced hippocampal network excitability and consequently reduced susceptibility to epileptiform activity. We also demonstrated that reduced Scn8a expression suppresses the occurrence of burst discharges in the hippocampus of Scn1a^R1648H/+ mice thereby providing a mechanistic basis for our previous observations of ameliorated seizure phenotypes in Scn1a mutant mice that co-express Scn8a mutations (Martin et al., 2007; Hawkins et al., 2011).

Parallel to our measurement of hippocampal bursting activity throughout early postnatal development, we also examined developmental changes in the expression levels of Na_v1.6 in the hippocampus. Consistent with previous observations, Western blot analysis revealed that Na_v1.6 is detectable by P10 and reaches adult levels by the third postnatal week (Plummer et al., 1997; Liao et al., 2010). Also, consistent with previous reports (Kohrman et al., 1996; Vega et al., 2008), we observed reduced Scn8a expression in Scn8a^med/+ mutants compared to WT littermates (data not shown). The emergence of reduced epileptiform burst discharges in Scn8a^med/+ mice coincided with the beginning of the developmental period in which Scn8a is highly expressed, and therefore likely to significantly influence neuronal excitability. This suggests that the reduced epileptiform bursting in Scn8a^med/+ mice is, in part, a consequence of lower Scn8a expression levels during this period. The expression levels of Scn1a (Na_v1.1) and Scn2a (Na_v1.2) were unaltered in the hippocampi of adult Scn8a^med/+ mutants, which is consistent with previously reported observations on Scn8a^medJ/medJ mice which have a 90% reduction in Na_v1.6 expression (Kearney et al., 2002).

Previous studies examining Scn8a-mediated seizure resistance raised the possibility of a self-reinforcing relationship between Scn8a expression and neuronal activity in which increases in neuronal activity lead to increased Scn8a expression, which in turn further increases neuronal excitability in the CA3 region of the hippocampus and contributes to epilepsy (Blumenfeld et al., 2009). Genetically reducing Scn8a expression might block this positive feedback cycle. An additional possibility is that Scn8a localization at the proximal axon initial segment of hippocampal dentate gyrus neurons places these channels in a prime location to regulate the low-pass filtering capability of these neurons, which is thought to help protect the hippocampus from seizure activity (Hsu, 2007; Kress et al., 2010). Furthermore, dentate gyrus neurons express comparatively lower levels of Scn8a when compared to CA3 (Kress et al., 2010), perhaps making their intrinsic excitability more sensitive to changes in Scn8a expression. These results, together with our findings, indicate that the hippocampus is a key region for Scn8a-mediated seizure resistance.

To directly test the hypothesis that targeting Scn8a in the hippocampus is sufficient to increase seizure resistance, we performed lentiviral-Cre and GFP injections into the hippocampus of conditional Scn8a mice (flox/flox) and measured response to PT administration. This approach resulted in an overall 14% reduction in Na_v1.6 levels in the hippocampus (data not shown). GFP positive cells were detected in the CA3, dentate gyrus and dentate hilus regions of the hippocampus while few targeted cells in CA1 were identified. Hippocampal knockdown of Scn8a led to a significant reduction in the number of electrographic seizures resulting from PT administration. However, the latency to the first electrographic seizure and the average duration of all seizure events were comparable between LV-GFP and LV-Cre groups. This finding indicates that the hippocampus is an important regulator of Scn8a-dependent CNS excitability and supports the possibility that targeting of Scn8a may provide an efficacious treatment for some forms of epilepsy. Furthermore, we speculate that reduction of Scn8a expression in the CA3 and DG hippocampal regions is primarily responsible for the observed seizure protection as reducing sodium channel expression in inhibitory neurons of the dentate hilus is not likely to lead to seizure protection.

Although heterozygous Scn8a null mice (Scn8a^med/+) are visibly normal, they exhibit anxiety-like behavior and have spike-wave discharges that are characteristic of absence seizures (Burgess et al., 1995; McKinney et al., 2008; Papale et al., 2009). In one human family, SCN8A haploinsufficiency was associated with cognitive deficits (Trudeau et al., 2006). Additional studies will therefore be required to evaluate the effects on behavior and cognition of targeting Scn8a in epileptic animals. It is possible that normalizing the excitability of an epileptic region by targeting Scn8a will not result in the behavioral deficits that may occur when Scn8a is reduced in an otherwise normal animal. In addition, restricted targeting of Scn8a may reduce the likelihood of unwanted consequences. In support of this hypothesis, mice with hippocampal knockdown of Scn8a displayed normal visible behavior and no absence seizures during video/EEG recordings.

The SCN8A mutation N1786D was recently identified in a patient with severe epilepsy, cognitive deficits, and sudden unexpected death in epilepsy (SUDEP) (Veeramah et al., 2012). In addition, loss of the RNA-binding protein Celf4 has been shown to lead to an increase in Scn8a expression and to spontaneous seizure generation in mice (Sun et al., 2013). These findings seem to contradict observed improvements in seizure protection in Scn8a mutant mice. However, the available mouse models of Scn8a dysfunction have reduced sodium currents, whereas the N1786D is a gain-of-function mutation with increased persistent current (Veeramah et al., 2012), and Celf4 deficiency is believed to result in increased Na_v1.6 expression at the axon initial segment and increased persistent current (Sun et al., 2013). Thus, reduced Na_v1.6 currents likely lead to reduced neuronal excitability and seizure protection, whereas alterations that increase Na_v1.6 currents, in particular those that increase persistent current, lead to greater neuronal excitability and seizure susceptibility.

Conclusions

We have determined that hippocampal Na_v1.6 is an important regulator of neuronal excitability and seizure susceptibility. These results raise the possibility that targeting Scn8a, either genetically or with Na_v1.6 specific compounds, may be an efficacious treatment strategy for some forms of epilepsy.

Highlights.

• Hippocampal burst latency and occurrence is reduced in Scn8a^med/+ mice after P20.

• The Scn8a-med mutation reduces burst discharges in a GEFS+ model of epilepsy.

• Cre-mediated hippocampal knockdown of Scn8a confers seizure resistance.

• Global inactivation of Scn8a in the adult mouse increases seizure thresholds.

• Selective targeting of SCN8A might be efficacious as a treatment for epilepsy.