Pathogenic in-Frame Variants in SCN8A: Expanding the Genetic Landscape of SCN8A-Associated Disease

Numerous SCN8A mutations have been identified, of which, the majority are de novo missense variants. Most mutations result in epileptic encephalopathy; however, some are associated with less severe phenotypes. Mouse models generated by knock-in of human missense SCN8A mutations exhibit seizures and a range of behavioral abnormalities. To date, there are only a few Scn8a mouse models with in-frame deletions or insertions, and notably, none of these mouse lines exhibit increased seizure susceptibility. In the current study, we report the generation and characterization of two Scn8a mouse models (ΔIRL/+ and ΔVIR/+) carrying overlapping in-frame deletions within the voltage sensor of domain 4 (DIVS4). Both mouse lines show increased seizure susceptibility and infrequent spontaneous seizures. We also describe two unrelated patients with the same in-frame SCN8A deletion in the DIV S5-S6 pore region, highlighting the clinical relevance of this class of mutations.

The first mouse model of SCN8A epileptic encephalopathy was generated by knock-in of the SCN8A p.N1768D mutation . Heterozygous mice expressing the N1768D mutation recapitulated several phenotypes observed in the original patient, including spontaneous seizures and SUDEP (35). A conditional mouse model with the SCN8A R1872W mutation also exhibited spontaneous seizures and early mortality when the mutation was expressed globally in the brain or selectively in excitatory neurons (Bunton-Stasyshyn FIGURE 1 | Patient SCN8A variants and in-frame variants from gnomAD and ClinVar. (A) SCN8A channel with variants denoted by: red star, R1620L mutation; green star, in-frame deletion in Patients 1 and 2; filled blue circle, in-frame duplication in proband from Stringer et al., 2021; filled green circle, in-frame variants from Johannesen et al., 2021; open circles, in-frame deletions from ClinVar (July 2021); filled black circle, in-frame deletions and duplications from the gnomAD database (v2.1.1 and v3.1.1; July 2021); filled red circle, in-frame duplication in both gnomAD and ClinVar. (B) DNA and protein sequence alignment for Patients 1 and 2 and the proband identified in Stringer et al., 2021. Green indicates amino acids deleted in Patients 1 and 2. Dash (-) indicates a deleted nucleotide or amino acid compared to WT. Blue indicates inserted DNA nucleotides or amino acids. Underline shows close proximity of amino acids deleted in ΔVIR and ΔIRL mouse lines to the variant identified in Stringer et al., 2021. Frontiers in Pharmacology | www.frontiersin.org November 2021 | Volume 12 | Article 748415 et al., 2019). Recently, using CRISPR/Cas9 technology, we generated a mouse line expressing the SCN8A R1620L mutation ( Figure 1A), which was identified in a patient with relatively mild epilepsy and behavioral deficits (Wong et al., 2021a). Mice heterozygous for this mutation exhibit increased seizure susceptibility, spontaneous seizures, impaired learning and memory, social deficits, and altered neuronal excitability (Wong et al., 2021a). Prior to the development of the mouse lines expressing human SCN8A epilepsy mutations, most of the published Scn8a mouse models carried loss-of-function missense or truncating Scn8a mutations (Martin et al., 2007;Papale et al., 2009;Hawkins et al., 2011;Makinson et al., 2014). Mice heterozygous for loss-offunction Scn8a mutations display increased resistance to induced seizures (Martin et al., 2007;Hawkins et al., 2011;Makinson et al., 2014), and depending on the genetic background, some lines also exhibit spike-wave discharges (Papale et al., 2009). There are currently only a few Scn8a mouse models with in-frame deletions or insertions (Jones et al., 2016;Inglis et al., 2020), and notably, none of these mouse lines show increased seizure susceptibility. Jones et al., 2016 reported an in-frame deletion (p.I1750del) in the DIVS6 domain, and homozygous mutants with this mutation exhibited motor impairments and early mortality (Jones et al., 2016). We previously generated two Scn8a mouse lines, one with an inframe deletion (p.R848_F850del, Δ9) and the other with an inframe insertion (p.R848_V849insD, ∇3) in the DIIS4 (Inglis et al., 2020). Heterozygous mutants from both of these lines exhibit increased seizure resistance (Inglis et al., 2020). In the current study, we report the generation and characterization of two novel Scn8a mouse models with overlapping in-frame deletions in the DIVS4 that exhibit increased seizure susceptibility and spontaneous seizures. We also describe two unrelated patients with the same in-frame SCN8A deletion in the DIV S5-S6 pore region, highlighting the clinical relevance of in-frame SCN8A mutations.

Animals
Using CRISPR/Cas9, we knocked in the human SCN8A p.R1620L mutation into the mouse Scn8a gene on the C57BL/6J background (corresponding to R1618L in the mouse,   Wong et al., 2021a). As a result of nonhomologous end joining during this process, we generated two additional Scn8a mouse lines (Scn8a 9Δ_IRL and Scn8a 9Δ_VIR ) with overlapping 9 base pair in-frame deletions which include removal of the positively charged R1618 residue ( Figure 2A). Heterozygous mutants from the Scn8a 9Δ_IRL (ΔIRL/+) and Scn8a 9Δ_VIR (ΔVIR/+) lines were backcrossed to C57BL/6J mice (Strain: 000664, Jackson Laboratories) for four generations. We performed Sanger sequencing on mutant mice from each line to confirm that the in-frame deletions were the only changes in Scn8a exon 26 and that the conserved exon 26 of the other brain sodium channels, Scn1a, Scn2a, and Scn3a, were unaltered. The 9 bp deletions were the only alterations observed in Scn8a exon 26 and no off-target CRISPR editing was observed in exon 26 of the other sodium channels. Male and female ΔIRL/+ and ΔVIR/+ mutants and respective wildtype (WT) littermates at the N4 generation were used for all experiments. To examine weight gain and survival of ΔIRL mice, male and female heterozygous mutants were bred to generate homozygous ΔIRL/ΔIRL mutants, heterozygous ΔIRL/+ mutants, and WT littermates. We tried to similarly breed the ΔVIR mouse line, however an insufficient number of litters were generated. Therefore, for the ΔVIR mouse line, heterozygous male mutants were bred with C57BL/6J females to generate heterozygous ΔVIR/+ mutants and WT littermates.
To generate biallelic mice expressing the R1620L and ΔVIR mutations, male ΔVIR/+ mutants were crossed with female RL/+ mutants to generate the following genotypes: ΔVIR/+, RL/+, ΔVIR/RL, and WT. This breeding scheme allowed us to examine littermates expressing the R1620L and ΔVIR mutations. Survival and weights were recorded. Mice were housed on a 12 h light/dark cycle with food and water ad libitum. All experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of Emory University.
Genotyping DNA was isolated from tail biopsies. Mice with the 9Δ_IRL or 9Δ_VIR allele were identified by PCR amplification using primer R1620L_R1: TACGCGAAGTTGGACATCCC and forward primers that span the respective 9 bp deletions: Scn8a_9del_IRL_F1: CCCTATTCCGCGTGGCCC and Scn8a_9del_VIR_F2: TCTCCCCGACCCTATTCCGATT. The 9Δ_IRL and 9Δ_VIR alleles generated 175 bp and 184 bp PCR products, respectively. The WT allele was identified using the primers R1620L_R1 and Scn8a_WT_F: CTATTCCGCGTCATC CGATTGG, which generated a 183 bp product. The RL/+ mutants were genotyped as previously described (Wong et al., 2021a).

qRT-PCR
Whole brains were extracted from P20-P21 WT, heterozygous, and homozygous mutants of both sexes from each mouse line. RNA extraction and cDNA synthesis were performed as previously described (Lamar et al., 2017). The following Scn8a primer pair was used: Scn8a_F: AGATTTAGCGCCACTCCTGC and Scn8a_R: GGACCATTCGGGAGGGTTAC. Analyses were conducted in technical triplicates using the Real-Time PCR Detection System and SYBR Green (BioRad). Expression levels were normalized to beta-actin (F: CAGCTTCTTTGCAGCTCC TT and R: ACGATGGAGGGGAATACAGC). Relative expression between genotypes was compared using the ΔΔct method.

EEG Surgery and Analyses
Cortical electrodes were implanted into adult (2-4 months old) male ΔIRL/+ and male ΔVIR/+ mutants as previously described (Lamar et al., 2017;Wong et al., 2018;Inglis et al., 2020;Shapiro et al., 2019). Four cortical screw electrodes were implanted into the skull at the following coordinates relative to bregma: anteriorposterior(AP) +2.0 mm and medial-lateral(ML) +1.2 mm; AP −1.5 mm and ML +1.2 mm; AP +0.5 mm and ML −2.2 mm; and AP −3.5 mm and ML −2.2 mm. Two fine-wire electrodes were implanted into the neck muscles for EMG recordings. Each mouse was allowed 1 week to recover from the surgical procedure. Stellate Harmonie rodent software was used to obtain and analyze EEG recordings. Seizures were manually identified and characterized by high frequency and amplitude EEG signals that were at least 3 s in duration and twice the background. Simultaneous video recordings were used to confirm behavioral seizures.

Statistical Analyses
Data are presented as mean ± SEM with p ≤ 0.05 considered as statistically significant. All statistical analyses were performed with Prism 9.0 (GraphPad Software, San Diego, CA

mRNA Expression
Quantitative real-time RT-PCR analysis was performed on whole brain samples from P20-P22 mice of each genotype from the ΔIRL and ΔVIR mouse lines. We observed similar levels of Scn8a expression within and between each mouse line ( Figures 2D-G).
Heterozygous ΔIRL/+ and ΔVIR/+ Mutants Exhibit Increased Seizure Susceptibility Susceptibility to 6 Hz-and flurothyl-induced seizures were compared between heterozygous mutants (ΔIRL/+ or ΔVIR/+) and WT littermates from each line. We did not observe any statistically significant differences in susceptibility to 6 Hz-or flurothyl-induced seizures between the sexes; therefore, data from both sexes were combined for analysis. Both ΔIRL/+ and ΔVIR/+ mutants were significantly more susceptible to 6 Hz-induced seizures when compared to their respective WT littermates ( Figures 3A,D). For the ΔIRL/+ mutants, 12/20 mice seized (RS Racine Score; 8RS0, 1 RS1, 11 RS2, 1 RS3) when compared to 6/25 WT littermates (19 RS0, 1 RS1, 5 RS2). Similarly, all of the ΔVIR/+ mutants exhibited a 6 Hz seizure (21 RS2, 2 RS3) whereas most of the WT littermates did not seize (16 RS0, 1 RS1, 3 RS2). When tested with flurothyl, ΔIRL/+ and ΔVIR/+ mutant mice exhibited significantly shorter latencies to the first myoclonic jerk (Figures 3B,E) and the first GTCS ( Figures 3C,F) when compared to their WT littermates. Although we cannot directly compare the two mouse lines, the average latencies to the first MJ and GTCS in the ΔVIR/+ mutants were approximately 40% shorter than the corresponding measurements from the ΔIRL/+ mutants, suggesting that the ΔVIR/+ mutants may be more severely affected.

Heterozygous ΔIRL/+ and ΔVIR/+ Mutants Exhibit Spontaneous Seizures
Continuous EEG recordings were obtained from five male ΔIRL/ + and five male ΔVIR/+ mutants. 2/5 ΔIRL/+ mutants and 3/5 Frontiers in Pharmacology | www.frontiersin.org November 2021 | Volume 12 | Article 748415 ΔVIR/+ mutants exhibited spontaneous seizures. Figure 4A provides an example of a spontaneous seizure observed in a ΔVIR/+ mutant. Spontaneous seizures were between 17-60 s in duration with the exception of one ΔVIR/+ mutant (Mouse #6, Figure 4B) that had a 4 min-long seizure, followed by 2 min of recovery, entry into status epilepticus for approximately 4 h, and ultimately, death. Spontaneous seizure frequency was infrequent, with an average seizure frequency of less than 1 seizure/day in the mice that exhibited seizures ( Figure 4B). Electrographic seizures were accompanied by rearing, paw waving, loss of posture, and in some instances, wild running and bouncing.

Biallelic Mutants Exhibit Premature Mortality and Increased Seizure Susceptibility
Based on our previous characterization of the RL mutants (Wong et al., 2021a), we know that homozygous RL mutants gain weight normally until postnatal day 15 (P15) and have a maximum lifespan of 22 days. Thus, to examine the relative severity of the ΔVIR mutation compared to the pathogenic human SCN8A p.R1620L mutation (Wong et al., 2021a), we crossed heterozygous male ΔVIR/+ mutants with heterozygous female  RL/+ mutants to generate WT, ΔVIR/+, RL/+, and biallelic (ΔVIR/RL) littermates. Beginning at P12, ΔVIR/RL mutants of both sexes were smaller than their same-sex littermates ( Figures  5A,B). Male and female ΔVIR/RL mutants exhibited premature death, and maximum lifespans of 22 and 31 days were observed, respectively ( Figure 5C). We observed no sex differences in the latency to the first MJ or GTCS following flurothyl exposure; therefore, we combined the data from male and female mice of the same genotype. Although not statistically significant, the RL/+ mutants exhibited a lower average latency to the first MJ and GTCS compared to WT littermates ( Figures 5D,E). Average latency to the first MJ was significantly lower in ΔVIR/+ mutants compared to WT littermates ( Figure 5D), and the average latency to the first GTCS was significantly lower in ΔVIR/+ mutants compared to RL/+ mutants and WT littermates ( Figure 5E), suggesting greater severity of the ΔVIR mutation. Due to the premature death of the ΔVIR/RL mutants, we were unable to evaluate their susceptibility to flurothyl-induced seizures.

Patients With in-Frame SCN8A Variants
Patients 1 and 2: c.5077_5091del, p.Asn1693_Cys1697del Gene panel testing of two unrelated patients (Patient 1 and Patient 2) identified the same heterozygous in-frame deletion variant, SCN8A c.5077_5091del (p.N1693_C1697del) ( Figure 1B), located in the DIV S5-S6 pore region ( Figure 1A, green star). The c.5077_5091del variant is currently classified as a "variant of uncertain clinical significance" (Richards et al., 2015). Patient 1 presented with autism, developmental delay, dysmorphic facial features, but no seizures. The deletion was determined to be inherited from the unaffected mother. Patient 2 presented with autism, encephalopathy with developmental delay, tremors, facial asymmetry, low muscle tone, skeletal abnormalities, and femoral torsion, but no history of seizures. The deletion was determined to be de novo for Patient 2. Patient 2 also had another de novo heterozygous variant in ANK3 c.3821C > A (p.S1274Y), which was classified as a "variant of uncertain clinical significance."

DISCUSSION
In the current manuscript, we describe the generation and characterization of two Scn8a mouse lines (ΔIRL and ΔVIR) with overlapping, in-frame deletions. Heterozygous ΔIRL/+ and ΔVIR/+ mutants exhibit increased seizure susceptibility and spontaneous seizures, demonstrating the potential for this class of genetic variation to contribute to the clinical burden associated with SCN8A dysfunction. Consistent with this, we also report the identification of an in-frame SCN8A variant in two unrelated patients with neurodevelopmental phenotypes, but no seizures.
We previously observed increased resistance to induced seizures in heterozygous Scn8a Δ9/+ and Scn8a ∇3/+ mutant mice, expressing an in-frame 9 bp deletion (Δ9) and 3 bp insertion (∇3) in the DIIS4, respectively (Inglis et al., 2020). EEG analyses of the Scn8a Δ9/+ mutants revealed normal electrographic activity and no spontaneous seizures (Inglis et al., 2020). In contrast, ΔIRL/+ and ΔVIR/+ mutants, harboring overlapping 9 bp deletions within the DIVS4, exhibited increased seizure susceptibility and spontaneous seizures. We performed in silico analyses of the DIIS4 and DIVS4 transmembrane domains ( Figure 6) and found that the Δ9 mutation in the DIIS4 causes a loss of a positive charge and shift of polar charges. The ∇3 mutation in DIIS4 causes a shift of the polar charges to one side of the helix. In contrast, the VIR and IRL deletions in DIVS4 do not significantly alter the structure and distribution of charges in the alpha helix. Differences between the phenotypes of mice expressing the Δ9 and ∇3 mutations versus the ΔVIR and ΔIRL mutations may also be due, in part, to differences in the functional properties of the domains, with DI-DIII primarily for activation Yu and Catterall, 2003) and DIV primarily involved in inactivation Yu and Catterall, 2003). These contrasting observations also highlight that the phenotypic consequences of in-frame deletions or insertions may likely be difficult to predict. Additional studies are warranted to characterize the biophysical impact of the ΔIRL and ΔVIR alleles relative to other SCN8A missense and in-frame variants.
To date, most identified human SCN8A-epilepsy associated mutations have been de novo amino acid substitutions, many of which are predicted or shown to have gain-of-function properties (Veeramah et al., 2012;Estacion et al., 2014;Liu et al., 2019;Heyne et al., 2020). A small number of truncating and frameshift mutations have also been described and appear to contribute to neurodevelopmental phenotypes such as autism and intellectual disability Brunklaus and Lal, 2020). Of direct clinical relevance, we report two unrelated individuals with the same in-frame SCN8A variant. Patient 1 and Patient 2 both carry the same in-frame deletion (p.N1693_C1697del) which removes five amino acids ( Figure 1B) from the DIV S5-S6 pore region ( Figure 1A). Previous genotype-phenotype correlations found that variants in the pore region were more likely to be associated with loss-of-function rather than gain-offunction effects and epilepsy (Holland et al., 2018;. Consistent with this, Patients 1 and 2 exhibit developmental delay and autism without seizures. Although this variant was not observed in the gnomAD database, additional functional studies will be necessary to resolve its clinical significance. This is particularly relevant since the p.N1693_C1697del variant occurred de novo in Patient 2, but in Patient 1, was inherited from an unaffected parent. Patient 2 was also found to harbor a de novo substitution (p.S1274Y) in ANK3, which is important for connecting integral proteins with the spectrin-actin cytoskeleton (Kordeli and Bennett, 1991). ANK3 is associated with both autosomal dominant and recessive neurodevelopmental disorders, including intellectual disability and autism spectrum disorder (Bi et al., 2012;Iqbal et al., 2013;Kloth et al., 2017;Hu et al., 2019). In silico algorithms suggest a possible deleterious effect of the p.S1274Y variant (e.g., CADD 26.3); however, this variant is observed twice in the gnomAD database (v3.1.1), and there are no pathogenic variants in the HGMD database that are in close proximity to this variant. Interestingly, ANK3 has been shown to associate with VGSCs via a binding site on the DII-DIII intracellular loop (Jenkins and Bennett, 2001;Lemaillet et al., 2003;Gasser et al., 2012), raising the possibility that it may act as a genetic modifier of SCN8A. Stringer and others recently described a patient with epileptic encephalopathy who harbors a de novo heterozygous in-frame duplication in SCN8A (p.G1625_I1627dup) and an inherited heterozygous missense variant in CACNA1H (p.G318S) (Stringer et al., 2021). The p.G1625_I1627dup variant ( Figure 1A, blue circle), located in the DIVS4 transmembrane domain, is proximal to several pathogenic variants, including p.R1620L (Rossi et al., 2017;Liu et al., 2019) and p.A1622D (Liu et al., 2019). This duplication results in the insertion of three amino acids, including an additional positively charged arginine residue ( Figure 1B). Interestingly, while this duplication does not overlap with the amino acids altered in the ΔIRL and ΔVIR mice, it is immediately adjacent to them ( Figure 1B). The p.G1625_I1627dup variant resulted in a hyperpolarizing shift of the voltage-dependence of activation of Na v 1.6 but had no effect on sodium current density or gating mechanisms (Stringer et al., 2021). This observation is consistent with a gain-of-function effect on the Na v 1.6 channel; however, Stringer et al. also demonstrated that the c.952G > A, p.G318S variant in CACNA1H was associated with loss-of-function effects. Further work will be required to resolve the relative contribution of each variant to the clinical presentation.
Johannesen et al. also recently reported two "likely pathogenic" in-frame SCN8A variants in a large cohort of Danish patients with SCN8A mutations (Johannesen et al., 2021). The maternally inherited p.I888_V892delinsM variant was identified in one patient with moderate intellectual disability without epilepsy (Johannesen et al., 2021). Another patient had a maternally inherited variant (E1774_A1777del) and presented with several types of seizures, including febrile, myoclonic, and atonic seizures, but normal intellect (Johannesen et al., 2021). It is unclear whether the mothers of these patients displayed similar clinical features.
There are at least six in-frame variants in the ClinVar database (Table 1; Figure 1A), including the p.N1693_C1697del variant observed in Patients 1 and 2. One ClinVar variant (p.Q1866_Q1867insRELDILR) introduces seven amino acids in the C-terminus ( Figure 1A) and is classified as "likely pathogenic". This variant is not observed in the gnomAD database (v3.1.1), and it is adjacent to several reported variants associated with epileptic encephalopathy, including p.L1865P (Trump et al., 2016), E1870D (Boerma et al., 2016), and R1872 which is  one of the most frequently mutated SCN8A residues (Ohba et al., 2014;Larsen et al., 2015;Wagnon et al., 2016). Examination of the population gnomAD database identified 11 in-frame variants in SCN8A (Table 2; Figure 1A). Whether these variants alter the biophysical properties of the channel is currently unknown; however, most of these variants are located in the intracellular DI-DII and DII-DIII linkers ( Figure 1A), which are generally more tolerant of variation (Yu and Catterall, 2003;Meisler et al., 2021). In contrast, pathogenic SCN8A variants are typically located in the more conserved parts of the channel, such as the transmembrane segments, inactivation gate, and pore (Wagnon and Meisler, 2015;. In summary, it is likely that additional in-frame SCN8A variants will be identified as more patients undergo whole exome and genome sequencing, further expanding the genetic landscape of SCN8A-associated disease, and potentially posing challenges for genetic counseling and precision therapy. Due to the rare nature of in-frame SCN8A variants, it is currently unclear if penetrance is reduced for this class of variants. Loss-of-function SCN8A variants (nonsense, frameshift) have been previously reported to exhibit incomplete penetrance (Trudeau et al., 2006); therefore, penetrance may depend on the functional consequence of the individual inframe variant. The ΔIRL and ΔVIR Scn8a mouse lines will provide the opportunity to further study genotype-phenotype relationships in SCN8A-related disease and will assist in the identification of appropriate treatments for patients with this class of SCN8A mutation.

DATA AVAILABILITY STATEMENT
The human datasets presented in this article are not readily available because of ethical and privacy restrictions. Requests to access the datasets should be directed to the corresponding author(s).

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Institutional Review Board of Emory University. Written informed consent to participate in this study was provided by the participant's legal guardian/next of kin. The animal study was reviewed and approved by the Institutional Animal Care and Use Committee of Emory University.

AUTHOR CONTRIBUTIONS
JW and AE contributed to the conception and design of the study. JW performed the experiments, statistical analyses, and wrote the first draft of the manuscript. LS and JT performed experiments and statistical analyses. KB, KM, and KG surveyed databases. KG, PG, SK, and BS performed genetic testing and clinical diagnoses. All authors contributed to manuscript revision, read, and approved the submitted version.

FUNDING
This project was supported by the National Institutes of Health (JCW, R21NS114795; AE R03NS114791). The content is solely the author's responsibility and does not necessarily reflect the official view of the National Institutes of Health. The authors have no financial interests related to this work and declare no competing interests.