Deletion of a non-canonical promoter regulatory element causes loss of Scn1a expression and epileptic phenotypes in mice

Genes with multiple co-active promoters appear common in brain, yet little is known about functional requirements for these potentially redundant genomic regulatory elements. SCN1A, which encodes the NaV1.1 sodium channel alpha subunit, is one such gene with two co-active promoters. Mutations in SCN1A are associated with epilepsy, including Dravet Syndrome (DS). The majority of DS patients harbor coding mutations causing SCN1A haploinsufficiency, however putative causal non-coding promoter mutations have been identified. To model the functional role of potentially redundant Scn1a promoters, we focused on the non-coding Scn1a 1b regulatory region, previously described as a non-canonical alternative transcriptional start site. Mice harboring deletion of the extended evolutionarily-conserved 1b non-coding interval exhibited surprisingly severe reduction of Scn1a and NaV1.1 expression in brain with accompanying seizures and behavioral deficits. This identified the 1b region as a critical disease-relevant regulatory element and provides evidence that non-canonical and apparently redundant promoters can have essential function.


Introduction
A large proportion of brain-expressed and indeed all mammalian genes are believed to rely on multiple alternative promoters [1][2][3] . For many genes, the alternative promoters produce distinct 5' untranslated regions that are not translated, leading to identical proteins from distinct transcription start sites (TSSs) 4,5 . Much of the focus on understanding the role of alternative promoters in mammalian transcriptional regulation has been on the potential for discrete function enabling compartmentalized expression in specific cells or tissues [6][7][8][9] . However, TSS activity mapping has found many genes where alternative promoters are active in the same tissue 10,11 . More recent evidence from single cell RNA sequencing and chromosome conformation suggests that annotated alternative promoters are frequently co-active in the same cells and physically interact in the nucleus [12][13][14] . However, in contrast to work on the requirement for alternative promoters with presumed discrete activity, studies investigating the functional requirement for apparently redundant co-active promoters are lacking.
Epilepsy is one of the most common neurological disorders, with both rare highlypenetrant and common variants contributing to genetic etiology. Mutations in SCN1A, which encodes the NaV1.1 sodium channel alpha subunit, result in a range of epilepsy phenotypes from generalized febrile seizures to Dravet Syndrome (DS), a severe childhood-onset disorder [15][16][17] . The majority of DS cases are caused by heterozygous de novo mutations in SCN1A resulting in truncation of the protein, with haploinsufficiency of NaV1.1 presumed to underlie pathology 18,19 . Mouse models with heterozygous coding mutations in Scn1a recapitulate features of DS, including seizures and sudden unexpected death in epilepsy (SUDEP) [20][21][22][23][24][25][26] . DS remains pharmacoresistant, with generalized tonic-clonic seizures beginning in the first year of life and comorbidities developing including cognitive impairment, ataxia and SUDEP [27][28][29] . SCN1A transcripts have a variable 5' untranslated region (UTR) containing one of two TSSs, 1a and 1b, that are conserved between human and mouse 30,31 . The proteins produced from 1a and 1b are expected to be identical. 1a (also referred to as h1u) has been defined as the major SCN1A promoter, however, comparison across brain tissues in human and mouse suggests that 1a and 1b are co-active, with ~35% of transcripts arising from 1b 30 . The apparent functional redundancy of 1a and 1b promoter activity and of 1a-and 1b-associated SCN1A transcripts raises the question of whether there are distinct roles or requirement of the transcribed 1a and 1b UTR and associated regulatory sequences.
In addition to serving as an example in which to dissect the role of multiple co-active promoters, there is high disease relevance for understanding the functional requirements for SCN1A regulatory DNA. SCN1A is one of the most common and welldocumented genes associated with severe medical consequences of haploinsufficiency.
Further, genome-wide association studies (GWAS) have implicated non-coding SCN1A DNA variants as contributing to epilepsy risk 32,33 , presumably via more subtle perturbation to transcriptional regulation, and non-coding promoter deletions have been found in DS patients 34,35 . A recent study of common variation in the promoter regions of SCN1A found that promoter variant haplotypes reduced luciferase in cells and that such non-coding variants in the functional SCN1A allele may modify DS severity 36 . Based on these findings, it is plausible that pathogenic variation in regulatory regions modulates SCN1A transcription, contributing to epilepsy. Functional studies are needed to determine the consequences of perturbations to SCN1A expression caused by mutations in non-coding DNA. Here, we used Scn1a as a model for examining transcriptional and phenotypic consequences associated with loss of a potentially redundant co-active promoter. Combining genomics, neuroanatomy, behavior, and EEG, we show that the Scn1a 1b non-canonical promoter and flanking conserved noncoding DNA sequence is independently essential for expression and brain function via characterizing the impact of 1b knockout in mice. In addition to mapping an essential regulatory region of a critical disease-relevant gene, our findings provide evidence that non-canonical promoters may play essential roles in general transcriptional activation.

Scn1a 1a and 1b chromosomal regions physically interact and share chromatin signatures indicating pan-neuronal transcriptional activator regulatory activity
To define the regulatory landscape of the SCN1A locus, we examined publicly-available chromosome conformation (Hi-C), transcriptomic and epigenomic data obtained from analysis of human and mouse brain tissues (Fig. 1a-d). We generated contact heatmaps from published Hi-C from prefrontal cortex 37 (PFC) at 10-kb resolution (Fig.   1a), and for additional tissues at 40-kb resolution 37 (Fig. S1). In PFC, SCN1A was located at the boundary of two major TADs (Topological Associated Domains), with extensive local interactions within the SCN1A locus. Differential analysis of Hi-C from PFC versus lung showed stronger local interactions in PFC, while there were no major differences between the PFC and hippocampus, suggesting brain-specific local SCN1A chromosomal interactions ( Fig. 1b and Fig. S1).
Previous work using 5' RACE 30 and luciferase assays defined regulatory sequences at SCN1A, including two genomic intervals, non-coding exons 1a and 1b, that are conserved between human and mouse and where the majority of SCN1A transcripts originate 30 (Fig. 1c). 1a (GenBank: DQ993522) was found to be the majority TSS for SCN1A transcripts across human and mouse (54% and 52% RACE transcripts, respectively). 1b (GenBank: DQ993523) was annotated as an alternative TSS, with 25% and 35% of SCN1A transcripts originating at this locus in both human and mouse.
No strong region-specific differences in 1a versus 1b TSS usage across brain regions were identified in previous work 30,38 . DNA sequence at 1a and 1b is highly evolutionarily-conserved across vertebrates. Notably, conservation at 1b extends nearly 3 kb downstream of the defined UTR transcribed sequence. Interaction models from an independent capture Hi-C dataset 37 also suggested physical interaction between 1a and 1b as well as between 1b and the nearby TTC21B promoter (Fig. 1c).
We examined chromatin state at the SCN1A locus across seven histone posttranslational modifications (PTMs) from human mid frontal lobe 39 (Fig. 1c). The strongest chromatin signatures for regulatory elements were at the previously-defined 1a and 1b loci, with H3K27ac, H3K4me3 and H3K9ac, weak H3K4me1, and absence of H3K27me3, H3k9me3, H3K36me3 in these regions. In ATAC-seq and H3K27ac across the majority of non-CNS tissues profiled in the ENCODE or Roadmap projects, 1a and 1b show reduced or absent signal, further indicating primary importance in the nervous system (data not shown). In addition to 1a and 1b, there were several other non-coding regions showing weaker, but still significant enrichment for H3K27ac in brain, representing potential additional SCN1A regulatory elements. ATAC-seq from neuronal and non-neuronal cells from dorsolateral PFC (DLPFC) 40 showed that neuronal cells have increased chromatin accessibility across SCN1A generally (Fig. 1c), with specific enrichment at 1a, 1b, and a third region also within the first intron of SCN1A. Comparing human neuronal data with mouse ATAC-seq and histone PTM data 41 , accessibility and chromatin states appeared largely conserved ( Fig. 1c and 1d). Finally, ATAC-seq data 41 from sorted neuronal subtypes in mouse, including excitatory neurons and parvalbumin (PV) and vasointestinal peptide (VIP) interneurons, indicated no difference in open chromatin signatures for the Scn1a locus and 1a and 1b across neuron types (Fig. 1d).

The evolutionarily-conserved Scn1a 1b non-coding region acts as a Scn1a transcriptional activator and is essential for survival
Taken together, the comparative and functional genomics data indicates evolutionarily conserved brain-specific pan-neuronal regulatory and TSS activity of 1a and 1b and chromosomal physical interaction between the two promoters. The 1b region has been annotated as an alternative TSS, yet the extended region surrounding the annotated transcribed UTR also shows the strongest enrichment across non-coding DNA at the SCN1A locus for evolutionary conservation and for chromatin signatures indicating enhancer and promoter activity (i.e. H3K27ac and H3K4me3). We sought to validate the specific role of 1b DNA in activation of SCN1A expression. We used luciferase assay to functionally test the core human 1b (h1b) region in cell lines. A 941 bp region containing 1b and conserved flanking sequence induced expression in HEK293 and SK-N-SH cells when cloned into a vector with a minimal promoter (Fig. 1e). To further demonstrate the regulatory role of 1b in SCN1A expression, we showed that a pool of 6 sgRNAs targeted to human 1b sequence and delivered along with dCas9-p300, a histone acetyltransferase, was sufficient to induce SCN1A expression 2.5-fold in HEK293 cells compared to non-transfected control (Fig. 1f).
The strength of evolutionary conservation and transcriptional activation-associated epigenomic signatures at the extended 1b interval is paradoxical considering its presumed role as a secondary TSS. Thus, we sought to test whether the extended 1b regulatory region is essential for Scn1a expression, and whether loss of this element is sufficient produce epilepsy and DS-relevant phenotypes in mouse. We used CRISPR/Cas9 targeting of C57BL/6N oocytes to generate mice harboring a 3063 bp deletion of the interval flanking the 1b regulatory element of Scn1a, removing the entire mammalian conserved region (Fig. 2a). We identified an F0 mutation carrier that transmitted the deletion to F1 offspring and confirmed the deletion interval via Sanger sequencing (Fig. 2b). We expanded this Scn1a 1b deletion line and eliminated potential off-target Cas9-induced mutations via breeding to wildtype C57BL/6N (WT) mice.
Previous work found that mice harboring homozygous coding mutations to Scn1a die in the third postnatal week and mice with heterozygous coding mutation exhibit reduced survival 20,26 . In comparison, 46 female WT by male heterozygous 1b harem trios pairings yielded 41 litters (Fig. 2c) and survival rates of WT and heterozygous 1b deletion pups were indistinguishable (Fig. 2d, p>0.9999, Chi squared with Fisher's exact test), indicating reduced severity of 1b deletion in comparison to coding loss-offunction mutation. However, female heterozygous 1b by male heterozygous 1b harem trios required nearly double the number of pairings at 74 and produced only 18 litters (Fig. 2c). Further, survival rates for WT and heterozygous 1b pups from these litters were indistinguishable (p>0.9999), but 42% of homozygous 1b deletion mice died by weaning (Fig. 2d, p=0.0005). We additionally tested heterozygous 1b deletion mice for measures of general health, finding no consistent deficits in growth, reflexes, and limb strength (Table S1). Thus, heterozygous Scn1a 1b deletion mice survive, but female deletion carriers fail to produce litters, indicating behavioral or physiological deficits associated with 1b deletion in females. Due to breeding and survival issues we include limited numbers of homozygous Scn1a 1b deletion mice in further analyses.

Loss of extended 1b interval causes loss of NaV1.1 across postnatal brain regions
We first sought to characterize changes in distribution and amount of NaV1.1 protein caused by loss of 1b in postnatal mouse brain. Reduction in protein expression across the mouse brain was confirmed by Western blot analysis of hippocampus from 3-month-old mice, showing that both heterozygous and homozygous 1b deletion resulted in decreased NaV1.1 protein expression compared to WT (Fig. 2e). NaV1.1 expression was reduced by 36% in heterozygous mice and 62% in homozygous deletion mice in the hippocampus. A similar change in protein expression was measured in the cerebellum with a 41% reduction in heterozygous mice and 63% in deletion mice (Fig.   2e). Raw western blots can be seen in supplementary (Fig. S2).
NaV1.1 immunohistochemistry (IHC) in WT mice showed expression across cerebellum, hippocampus and cortex ( Fig. 2fi-iii), consistent with previous studies of RNA and protein expression. We compared expression of NaV1.1 along with the interneuron marker parvalbumin across 1b homozygous deletion, heterozygous deletion, and WT mice (Fig. 2g). Notably, deletion of 1b appeared to generally ablate NaV1.1 expression, rather than specifically impact certain brain regions, consistent with 5' RACE TSS activity 30 . There was no obvious qualitative change in NaV1.1 IHC between WT and heterozygous 1b deletion mice, while homozygous deletion mice had obvious reduction of expression in the brainstem, cerebellum and hippocampus.

Differential gene expression in in Scn1a 1b deletion mouse hippocampus
First, we tested for expected reduced Scn1a RNA expression via quantitative reversetranscription PCR (qRT-PCR) performed on cortex, hippocampus and cerebellum of 3month-old Scn1a 1b deletion carriers and WT littermates (Fig. 3a). In agreement with GTEx 38 and previous studies 30 , we observed the highest level of WT Scn1a expression in the cortex with expression in cerebellum and hippocampus 34% and 60% lower, respectively. When comparing 1b deletion to WT mice, there was a trend for reduction in Scn1a expression in heterozygous 1b deletion mice, and homozygous 1b deletion carriers had a significant reduction to approximately 40% WT in all regions. These results are consistent with Western blot results indicating that deletion of the extended 1b interval had a larger than expected impact on Scn1a and NaV1.1 expression considering the proportion of transcripts expected to originate at this element.
We next used RNA sequencing (RNA-seq) on P7 forebrain from WT (n=2), heterozygous (n=4) and homozygous (n=2) 1b deletion mice and P32 micro-dissected hippocampus tissue from WT (n=2), heterozygous (n=2) and homozygous (n=2) 1b deletion mice. For both P7 and P32, Scn1a expression showed significant dosage dependent decrease using an additive model (  (Fig. 3b), consistent with qRT-PCR data. To compare transcripts arising from either 1a or 1b at P32, when Scn1a expression in WT brain is high, we measured the number of splice junction reads that linked the 1a and 1b non-coding exons with the first Scn1a coding exon and the number that mapped unambiguously to 1a or 1b (Fig. 3c).
As expected, splice junction and overlapping reads associated with mouse 1b (m1b) were reduced in heterozygous 1b deletion mice and abolished in homozygous 1b deletion mice. While 1a (m1a) splice junction and overlapping reads were not significantly reduced in heterozygous or homozygous 1b deletion carriers, relatively few reads were identified. The change in total Scn1a reads from RNA-seq ( Fig. 3b-c) was consistent with qRT-PCR and NaV1.1 western blot data, indicating much higher than anticipated decrease in RNA and protein levels considering the proportion of Scn1a transcripts originating at 1b. These findings are inconsistent with a model where 1b deletion specifically and solely impacts 1b transcript levels without affecting levels of Scn1a transcripts originating at 1a or other minor TSSs.
We tested for differential expression across 9260 and 8460 genes that were robustly expressed in the P7 and P32 RNA-seq datasets, respectively. As heterozygous 1b deletion carriers were more variable in Scn1a expression and transcriptomic signature ( Fig. 3b, S3a), we focused analysis on comparison of homozygous 1b deletion carriers with WT littermates. At P7, Scn1a was the only differentially expressed (DE) gene specific to homozygous 1b deletion mice using a threshold of FDR < 0.05 ( Table S6).
The minimal effect of 1b deletion on genes other than Scn1a at P7 is consistent with the low expression and non-essential role of Scn1a in early postnatal development 42

Homozygous but not heterozygous 1b deletion causes cognitive deficits in novel objection recognition (NOR) and spontaneous alternation in the Y-maze
To investigate the impact of 1b deletion on behavior, we performed a tailored battery focused on learning and memory and motor abilities. Heterozygous 1b deletion mice were additionally tested in a comprehensive behavioral battery of standard assays of overall physical health across development, sensorimotor reflexes, motor coordination, anxiety-like, and social behavior. A summary of the results from these experiments is reported in Table S2. Both heterozygous and surviving homozygous 1b deletion mice were comparable to WT littermates in developmental milestones.
Cognitive deficits were observed in homozygous 1b deletion but not heterozygous mutant mice by two corroborating assays of learning and memory, NOR and Y-maze.
Following established NOR methods 44,45 manual scoring by a highly-trained observer blinded to genotype indicated WT and heterozygous 1b deletion mice spent more time investigating the novel object versus the familiar object, as expected. In contrast, homozygous 1b deletion mice did not exhibit typical novel object preference ( Fig. 4a: (p < 0.0001) and 16-20 (p = 0.0270). In addition to indicating that homozygous 1b deletion causes hyperactivity, which is linked to DS 46,47 , these results indicate that there were no gross motor abnormalities, inability to rear, or hindlimb weakness that would prevent exploration of objects in NOR and confound that assy.
Heterozygous 1b deletion mice did not exhibit significant consistent phenotypes in learning and memory assays or a comprehensive battery of assays standard for examining mouse models of neurodevelopmental disorders (Table S2). Heterozygous 1b mice spent less time in the dark chamber in the light-dark assay (T (69) = 2.121, p = 0.0375, unpaired two-tailed t-test) and had decreased ultrasonic vocalizations (T (25) = 2.143, p = 0.0420, unpaired two-tailed t-test), but in the absence of corroborating assays for anxiety and socialization these results are only suggestive.

Adult mice harboring heterozygous 1b deletion are susceptible to seizures and exhibit abnormal EEG activity
We next tested for epilepsy-relevant phenotypes in heterozygous 1b deletion mice using a standard pentylenetetrazole (PTZ) chemoconvulsant seizure induction analysis and using EEG 48 [49][50][51] , and the link between 1b deletion and reduced Scn1a produced epileptiform phenotypes in the mouse brain validates disease relevance of this model.

Discussion
The majority of focus and functional studies of alternative promoters has been on genes where the multiple alternative TSSs are predicted to have discrete cell-type or tissuespecific activity [6][7][8][9] . However, recent studies of TSS usage and promoter interactions suggest a model where alternative promoters interact physically and are co-active in the same cells [12][13][14] . In these situations, it is largely unknown what the requirement for individual TSS and associated regulatory DNA may be. Here we focus on one specific putative non-canonical disease-relevant alternative promoter, a 3 kb evolutionarilyconserved DNA region including the previously described Scn1a 1b TSS. We show that deletion of this interval from the mouse genome causes significant decrease in Scn1a expression, NaV1.1 protein, and results in susceptibility to seizures and an epilepsyrelevant neurophysiology phenotype. These results define an essential disease-relevant regulatory region and show that loss of regulatory DNA associated with a non-canonical TSS has an interactive impact on total expression across all start sites.
There are multiple possible explanations for the observed strong impact of loss of the 1b interval on Scn1a expression. First, 1a and 1b may indeed be discretely regulated, but previous measures of 1b-originating transcripts must have significantly underestimated the actual proportion of 1b expression. In this case, our findings would simply reflect that 1b is actually the dominant Scn1a promoter. While we cannot disprove this, there is no evidence that earlier studies were incorrect and our estimates of 1a and 1b RNA-seq read frequency in 1b deletion mice do not support a model where 1b is dominant. Alternatively, the loss of the 1b genomic interval could result in decreased TSS activity at 1a as well via an interactive effect where 1b-associated regulatory DNA activity is required for 1a transcription. This model is plausible based on evidence for physical interaction of 1a and 1b, the correlation between 1b and 1a chromatin state across neuronal cell types, and the severe reduction in overall Scn1a transcript and NaV1.1 protein in 1b deletion mice. Considering the frequency of promoter-promoter interaction and reported common co-expression of alterative TSSs in single neurons, many brain genes could share similar regulatory structure, where regulatory DNA at putative alternative promoters contributes to transcriptional activation across interacting TSSs. While further experiments are needed to resolve the function of the 1a and 1b intervals and similar studies of other genes are needed to show that this phenomenon is widespread, our findings represent initial insights into the potential essential regulatory roles of non-canonical promoter DNA.
Annotation of the genome has led to major gains in understanding transcriptional wiring, yet it has been surprisingly difficult to predict the sufficiency and necessity of specific regulatory elements, even those expected to be critical based on comparative and functional genomics 3,10,52 . Knockout mouse models have been a gold standard for testing the phenotypic consequences of mutations, and recent efforts deleting noncoding DNA have provided critical insights into the role of regulatory DNA [52][53][54][55] . Here, we used CRISPR/Cas9-mediated deletion to assess the role of the evolutionarilyconserved 1b interval on higher order neurological phenotypes. Homozygous 1b deletion caused behavioral deficits and had a strong impact on survival, demonstrating the essential nature of the deleted interval. Our transcriptomics data suggests that homozygous deletion impacts expression of genes relevant to epilepsy, with downregulation on mature synaptic and signaling genes and upregulation of neural maturation and developmental genes. These findings indicate that the extended 1b DNA sequence plays a critical role in Scn1a expression. Further studies are needed to define the minimal and core nucleotides within the 1b interval and to define proteins that bind and participate in regulation. In addition, similar functional studies of other Scn1a regulatory DNA elements, and specifically of the 1a region, are necessary to determine which regulatory DNA regions are necessary and sufficient for expression in the brain.
In the disease-relevant heterozygous 1b deletion state, we identified susceptibility to induced seizures and epileptiform EEG activity. Thus, our studies show that heterozygous 1b deletion produces epilepsy-relevant phenotypes in mice.
Heterozygous loss of the 1b interval appears to have a less severe phenotypic impact compared to heterozygous truncating Scn1a mutations. Such truncating mutations reduce survival and cause behavioral and cognitive deficits relevant to DS in mice 20,26 and are more similar to the homozygous 1b deletion phenotype identified here. It is possible that phenotypes are milder in heterozygous 1b deletion mice in this study compared to previously analyzed DS mouse models due in part to differences in genetic background or environment. Regardless of the specific relevance to DS phenotypes, our results prove that perturbation to 1b function produce cognitive impairments and robust epilepsy-relevant phenotypes, justifying increased focus on non-coding regulatory DNA in genetic screening of DS and epilepsy patients.
While we did not identify corroborated behavioral phenotypes in heterozygous 1b deletion mice, female 1b deletion carriers failed to efficiently reproduce and heterozygous 1b deletion mice exhibited abnormal EEG spectral bandwidths. Thus, it is possible that more subtle neurodevelopmental disorders (NDDs) and DS behavioral and cognitive deficits are caused by heterozygous 1b deletion. Furthermore, the EEG spectral phenotypes in heterozygous 1b deletion mice overlap with other NDD and epilepsy models. Elevated delta spectral power is a biomarker of Angelman Syndrome (AS) 56

Generation of 1b mutant mice
We used Cas9-medaited mutagenesis of C56BL/6N oocytes to generate a mouse line harboring deletion of a conserved portion on the noncoding region of Scn1a containing the previously described 1b 30 regulatory region. Guide RNA was designed and synthesized according to described methods 62 , pooled with Cas9 mRNA and injected into mouse oocytes. We identified the unique guides GGAGATCTGGGTAGTCCTCG and GCTTTTCATACTATAGTGAG. Initial Cas9 targeting was performed at Lawrence Berkeley National Laboratory. F0s (induced on C57BL/6N background) carrying mutations were genotyped and bred to expand lines that harbored a mutation. We identified F0 pups carrying a 3063 bp deletion (mm10 -chr2:66407567-66410630) in 1b.
The colony was rederived and maintained by crossing male 1b deletion carriers with C57BL/6N wild-type females (Charles River). Extensive crossing of heterozygous mutation carriers to wild-type animals vastly reduces the likelihood that any potential offtarget mutations caused by Cas9 targeting would persist in our 1b deletion line.
Genotypes were identified via PCR and sequence-verified for all animals included in analyses, with the primers in Table 1

RNA collection
Cortex, hippocampus and cerebellum were regionally dissected from one hemisphere of P7, P32 and 3-month-old homozygous deletion, heterozygous and wildtype mice. Both male and female mice were used, though there were not equal sex representation across genotypes. Total RNA was isolated using RNAqueous kit (Ambion) and assayed using an Agilent BioAnalyzer instrument.

qRT-PCR
Differential expression of Scn1a was verified by qRT-PCR at 3 months old. Briefly, 500 ng RNA was used for reverse transcription using SuperScript VILO cDNA synthesis kit (Invitrogen). Primers are reported in Table 2 and qPCR was performed with SYBR green PCR master mix (Applied Biosystems). Samples were excluded if technical replicates failed. Cycle counts were normalized to Gapdh. Statistical analysis was performed using unpaired Student's t-tests on normalized relative gene expression between genotypes using ΔΔCT.

RNAseq
RNA from P7 forebrain and P32 hippocampus was collected as described above.
Samples included males and females of each genotype. Total RNA was isolated using Ambion RNAqueous and assayed using an Agilent BioAnalyzer instrument. Stranded mRNA sequencing libraries were prepared using TruSeq Stranded mRNA kit. All eight samples were pooled and sequenced in one lane on the Illumina HiSeq platform using a single-end 100-bp strategy. Each library was quantified and pooled before sequencing at the UC Davis DNA Technologies Core.
The transcriptomic analysis was performed as before 63 . Reads from RNA-seq were aligned to the mouse genome (mm9) using STAR (version 2.7.2) 64 . Aligned reads mapping to genes were counted at the gene level using subreads featureCounts 65 . The mm9 knownGene annotation track and aligned reads were used to generate quality control information using the full RSeQC tool suite 66 . Unaligned reads were quality checked using FastQC.

Differential expression analysis
Raw count data for all samples were used for differential expression analysis using edgeR 67 . Genes with at least 20 reads per million in at least one sample were included for analysis, resulting in a final set of 9260 and 8460 genes for differential testing in P32 and P7 mice, respectively. Multidimensional scaling analysis indicated that Scn1a expression was the strongest driver of variance across samples. Tagwise dispersion estimates were generated and differential expression analysis was performed using a For the enrichment analysis, the test set of differentially expressed genes was compared against the background set of genes expressed in our study based on minimum read-count cutoffs described above.

Immunohistochemistry
All histological experiments were performed at least in triplicate and experimenters were blinded to genotype. Following anesthesia, P28 male and female mice were transcardially perfused with 4% paraformaldehyde (PFA) in HEPES, followed by

Western blot
Flash frozen samples were prepared for Western blot by sonication in 2x Laemmli buffer. After sonication, samples were spun down and the supernatant was used for a BCA Bradford assay using the Spectramax 190 plate reader to assess protein concentration using a standard curve. We ran 20 µg of protein on 8% and 12% gels using the Mini-PROTEAN Tetra Cell western blotting system (Bio-Rad). Anti-NaV1.1 (ASC-001, rabbit, 1:1000, Alomone) and anti-Gapdh (chicken, 1:10,000) primary antibodies were incubated overnight in Odyssey blocking buffer (LI-COR), visualized using a LI-COR Odyssey CLx system and quantified in FIJI (National Institutes of Health). Protein levels assayed via western blot were compared by unpaired Student's t-test.

Design
Both male and female subjects were used in this study. Subjects began the behavioral battery at 6-weeks of age. All behavioral tests were performed between 09:00 and 17:00-h during the light phase of the 12:12 light/dark cycle. Mice were brought to an empty holding room adjacent to the testing area at least 1-h prior to the start of behavioral testing. Mice were tested in the follow order: open field, spontaneous alternation, and novel object recognition. After completing the behavioral battery, a small subset of mice was used for EEG acquisition and the remaining animals were administered a lethal-dose of pentylenetetrazole to evaluate seizure susceptibility.
Order of testing: elevated plus maze, light dark, open field, spontaneous alternation, novel object recognition, self-grooming, beam walking, rotarod, social approach, malefemale social interaction, acoustic startle, pre-pulse inhibition, and fear conditioning.
Body weight, length (nose to edge of tail), and head width were measured using a scale (grams) and a digital caliper (cm). Cliff avoidance was tested by placing each pup near the edge of a cardboard box, gently nudging it towards the edge, and measuring the time for it to turn and back away from the edge. Failures to avoid the cliff was recorded as a maximum score of 30-s. Righting reflex was tested by placing each pup on its back, releasing it, and measuring the time for it to fully flip over onto four paws on each developmental day. Negative geotaxis was tested by placing each pup, facing downwards, on a screen angled at 45° from parallel, and measuring the time for it to completely turn and to climb to the top of the screen. Failures to turn and climb were recorded as a maximum score of 30-s. Circle transverse was tested by placing each pup in the center of a circle with a 5″ (12.5 cm) diameter drawn on a laminated sheet of 8.5″ x 11″ white paper, and measuring the time for it to exit the circle. Failures to exit the circle were recorded as a maximum score of 30-s.

Elevated-plus maze
The assay was performed using a mouse EPM (model ENV-560A) purchased from Med Associates (St. Albans, VT) and performed as previously described 69 . The EPM contained two open arms (35.5 cm x 6 cm) and two closed arms (35.5 cm x 6 cm) radiating from a central area (6 cm x 6 cm). The maze was cleaned with 70% ethanol before the beginning of the first test session and after each subject mouse was tested with sufficient time for the ethanol odor to dissipate before the start of the next test session. Room illumination was ∼30 lx.

Light↔dark conflict
The light↔dark assay was performed in accordance with previously described procedures 69 . The test began by placing the mouse in the light side (∼320 lx; 28 cm x 27.5 cm x 27 cm) of an automated 2-chambered apparatus, in which the enclosed/dark side (∼5 lx; 28 cm x 27.5 cm x 19 cm) was reached by traversing the small opening of the partition between the two chambers. The mouse was allowed to explore freely for 10-min. Time in the dark side chamber and total number of transitions between the light and dark side chambers were automatically recorded during the 10-min session using Labview 8.5.1 software (National Instruments, Austin, TX).

Open Field
General exploratory locomotion in a novel open field arena was evaluated as previously described 44,63,69 . Briefly, each subject was tested in a VersaMax Animal Activity Monitoring System (Accuscan, Columbus, OH, USA) for 30-min in a ~30 lux testing room. Total distance traversed, horizontal activity, vertical activity, and time spent in the center were automatically measured to assess gross motor abilities in mice. Repeatedmeasures ANOVA was used to detect differences in horizontal, vertical, total, and center time activity obtained during the open field assay. Multiple comparisons were corrected for using Sidak post hoc methods and F, degrees of freedom, and p-values are reported.

Spontaneous Alternation in a Y-maze
Spontaneous alternation was assayed using methods modified based from previous studies 63 in mice. The Y-shaped apparatus (SIZE) was made of non-reflective matte white finished acrylic (P95 White, Tap Plastics, Sacramento, CA, USA). Subjects were placed in the middle of the apparatus and transitions between the three arms were scored by an investigator blind to genotype. Mice are placed midway of the start arm, facing the center of the y for an 8 minute test period and the sequence of entries into each arm are recorded via a ceiling mounted camera integrated with behavioral tracking software (Noldus Ethovision). % spontaneous alternation is calculated as the number of triads (entries into each of the 3 different arms of the maze in a sequence of 3 without returning to a previously visited arm) relative to the number of alteration opportunities.
One-way ANOVA was used to detect differences in alternation. Multiple comparisons were corrected for using Sidak post hoc methods and F, degrees of freedom, and pvalues are reported.

Novel Object Recognition
The novel object recognition test was conducted as previously described 44

Balance beam walking
Balance beam walking is a standard measure of motor coordination and balance 71,72 .
We followed a procedure similar to methods previously described using our behavioral core 73 . Balance beam walking is sensitive to genetic mutations that affect neuromuscular, spinal, cerebellar, and other motor systems. It is a non-invasive and non-stressful assay. The apparatus consists of a start platform, an enclosed finish platform, and dowel suspended between the two platforms. The beam is approximately

Rotarod
Motor coordination, balance, and motor learning were assessed using an accelerating rotarod (Ugo Basile, Schwenksville, PA) as previously described 74,75 . Mice were placed on a cylinder which slowly accelerated from4 to 40 revolutions perminute over a 5minute (300-second) test session. The task requires the mice to walk forward in order to remain on top of the rotating cylinder rod. Mice were given 3 trials per day with a 30-60-minute intertrial rest interval. Mice were tested overtwo consecutive days for a total of 6 trials. Latency to fall was recorded with a 300-second maximum latency.
70% ethanol between each subject trial and following the conclusion of all testing.

Repetitive self-grooming
Spontaneous repetitive self-grooming behavior was scored as previously described 63,69,74,76,77 . Each mouse was placed individually into a standard mouse cage (46 cm long × 23.5 cm wide × 20 cm high). Cages were empty to eliminate digging in the bedding, which is a potentially competing behavior. The room was illuminated at ~40 lx. A front-mounted CCTV camera (Security Cameras Direct) was placed ~1 m from the cages to record the sessions. Sessions were videotaped for 20 min. The first 10-min period was habituation and was unscored. Each subject was scored for cumulative time spent grooming all the body regions during the second 10 min of the test session.

Three chambered social approach
Social approach was tested in an automated three-chambered apparatus using methods similar to those previously described 44,63,69,77,78 . Automated Ethovision XT videotracking software (Version 9.0, Noldus Information Technologies, Leesburg, VA) and modified nonreflective materials for the chambers were employed to maximize throughput. The updated apparatus (40 cm × 60 cm × 23 cm) was a rectangular, threechambered box made from matte white finished acrylic (P95 White, Tap Plastics, Sacramento, CA). Opaque retractable doors (12 cm × 33 cm) were designed to create optimal entryways between chambers (5 cm × 10 cm), while providing maximal manual division of compartments. Three zones, defined using the EthoVision XT software, detected time in each chamber for each phase of the assay. Zones were defined as the annulus extending 2 cm from each novel object or novel mouse enclosure (inverted wire cup, Galaxy Cup, Kitchen Plus, https://www.spectrumdiversified.com/whs/products/Galaxy-Pencil-Utility-Cup). Direction of the head, facing toward the cup enclosure, defined sniff time. A top-mounted infraredsensitive camera (Ikegami ICD-49, B&H Photo, New York, NY) was positioned directly above every pair of three-chambered units. Infrared lighting (Nightvisionexperts.com) provided uniform, low-level illumination. The subject mouse was first contained in the center chamber for 10 min, then allowed to explore all three empty chambers during a 10 min habituation session, then allowed to explore the three chambers containing a novel object in one side chamber and a novel mouse in the other side chamber. Lack of innate side preference was confirmed during the initial 10 min of habituation to the entire arena. Novel stimulus mice were 129Sv/ImJ, a relatively inactive strain, aged 10-14 weeks, and matched to the subject mice by sex. Number of entries into the side chambers served as a within-task control for levels of general exploratory locomotion.

Male-female social interaction
The male-female reciprocal social interaction test was conducted as previously described 63,69,74,76,77 . Briefly, each freely moving male subject was paired for 5-min with a freely moving unfamiliar estrous WT female. A closed-circuit television camera (Panasonic, Secaucus, NJ) was positioned at an angle from the Noldus PhenoTyper arena (Noldus, Leesburg, VA) for optimal video quality. An ultrasonic microphone (Avisoft UltraSoundGate condenser microphone capsule CM15; Avisoft Bioacoustics, Berlin, Germany) was mounted 20 cm above the cage. Sampling frequency for the microphone was 250 kHz, and the resolution was 16 bits. The entire apparatus was contained in a sound-attenuating environmental chamber (Lafayette Instruments, Lafayette, IN) under dim LED illumination (~10 lx). Duration of nose-to-nose sniffing, nose-to-anogenital sniffing and following were scored using Noldus Observer 8.0XT event recording software (Noldus, Leesburg, VA) as previously described49. Ultrasonic vocalization spectrograms were displayed using Avisoft software and calls were identified manually by a highly trained investigator blinded to genotype. Overhead lighting was turned off. The cued test consisted of a 3-min acclimation period followed by a 3-min presentation of the tone CS and a 90-s exploration period.
Cumulative time spent freezing in each condition was quantified by VideoFreeze software (Med Associates).

Acoustic Startle and Prepulse inhibition
Subjects were tested in San Diego Instruments startle chambers using standard methods as described previously 74,79 . Test sessions began by placing the mouse in the Plexiglas holding cylinder for a 5 min acclimation period. For the next 8 min, mice were presented with each of six trial types across six discrete blocks of trials, for a total of 36 trials. The intertrial interval was 10-20 s. One trial type measured the response to no stimulus (baseline movement). The other five trial types measured startle responses to 40 ms sound bursts of 80, 90, 100, 110, or 120 dB. The six trial types were presented in pseudorandom order such that each trial type was presented once within a block of six trials. Startle amplitude was measured every 1 ms over a 65ms period beginning at the onset of the startle stimulus. The maximum startle amplitude over this sampling period was taken as the dependent variable. Background noise level of 70 dB was maintained over the duration of the test session. For prepulse inhibition of acoustic startle, mice were presented with each of seven trial types across six discrete blocks of trials for a total of 42 trials, over 10.5 min. The intertrial interval was 10-20 s. One trial type measured the response to no stimulus (baseline movement) and another measured the startle response to a 40 ms, 110 dB sound burst. The other five trial types were acoustic prepulse stimulus plus acoustic startle stimulus trials. The seven trial types were presented in pseudorandom order such that each trial type was presented once within a block of seven trials. Prepulse stimuli were 20 ms tones of 74, 78, 82, 86, and 92 dB intensity, presented 100 ms before the 110 dB startle stimulus. Startle amplitude was measured every 1 ms over a 65 ms period, beginning at the onset of the startle stimulus. The maximum startle amplitude over this sampling period was taken as the dependent variable. A background noise level of 70 dB was maintained over the duration of the test session.

Seizure Susceptibility Following Administration of Pentylenetetrazole
Subjects were weighed then administered pentylenetetrazole (80 mg/kg) intraperitoneally. Dosing was conducted in the morning (9:00 -12:00) in a dim (~30 lux) empty holding room. Directly after administration of the convulsant, subjects were placed in a clean, empty cage and subsequent seizure stages were live-scored for 30min. Seizure stages were scored using latencies to (1) first jerk/Straub's tail, (2) loss of righting, (3) generalized clonic-tonic seizure, and (4) death. Time to each stage was taken in seconds and compared by genotype. One-way ANOVA was used to analyze latencies to first jerk, loss of righting, generalized clonic-tonic seizure, and death. F, degrees of freedom, and p-values are reported. µV-1000 µV that lasted between 0.5 and 100 ms, while spike trains had a minimum duration of 0.5s, a minimum spike interval of 0.05s and a minimum of 4 consecutive spikes. Power spectral densities were determined using a periodogram transformation from amplitude to frequency domains then log transformed for clearer data illustration.
Latency to seizure onset and subsequent death following administration of PTZ was first quantified by observed latencies then confirmed by spectral EEG and amplitude response read-outs. One-way ANOVA was used to analyze bouts of spike train activity and latency to seizure onset and death between genotypes. An overall ANOVA was used to detect a genotype difference across power bands, then genotype differences were analyzed within power bands using multiple comparisons.

Luciferase assay
We constructed luciferase reporter plasmids by cloning an ~900 bp region containing human 1b 30 into the pGL4.24 vector (Promega) upstream of the minP, primers in Table   3. HEK293 cells or SK-N-SH cells (40%-60% confluent) were transfected using Lipofectamine 3000 with each construct (400 ng) and the Renilla luciferase expression vector pRL-TK (40 ng; Promega) in triplicate. After 24 hours, the luciferase activity in the cell lysates was determined using the Dual Luciferase Reporter System (Promega).
Firefly luciferase activities were normalized to that of Renilla luciferase, and expression relative to the activity of an inactive region of noncoding DNA (NEG2) was noted.

CRISPR/dCas9 in HEK293 cells
HEK293 cells were transfected with 500 ng of equimolar pooled SCN1A_h1b sgRNAs (Table 4) and 500 ng dCas9 p300Core (Addgene, plasmid #61357) using Lipofectamine 3000. After 24 hours media was refreshed. 48 hours following transfection RNA was collected using RNAqueous kit (Ambion) and cDNA was generated using Superscript III reverse transcriptase (Invitrogen). Changes in gene expression were quantified via qPCR using SYBR green, primers are listed in Table 2.  (e) Activity of human 1b region in luciferase assay with minimal promoter in HEK293 (****P < 0.0001) and SK-N-SH cells (****P < 0.0001) shown as mean ± SEM. (d) Transcriptional activation of SCN1A using gRNAs targeting 1b co-transfected with dCas9-p300 in HEK293 cells increases SCN1A expression (*P = 0.047) when compared to empty vector (EV) as measured by qPCR and normalized to non-transfected control, shown as mean ± SEM. For panels a through e, see text for data sources.  Figure S1 -Tissue and brain regional differences in chromatin conformation. Hi-C contact map heatmaps at 40-kb resolution for a 5-Mb region around SCN1A gene. In the central rows and columns are the absolute contact maps, while the corner plots represent the differential contact maps between pre-frontal cortex (PFC) and the indicated tissue. For the latter, red color indicates stronger contacts in PFC in relation to the other tissue, and blue color the opposite.