Fine mapping and candidate gene analysis of a dravet syndrome modifier locus on mouse chromosome 11

Pathogenic variants in SCN1A result in a spectrum of phenotypes ranging from mild febrile seizures to Dravet syndrome, a severe infant-onset epileptic encephalopathy. Individuals with Dravet syndrome have developmental delays, elevated risk for sudden unexpected death in epilepsy (SUDEP), and have multiple seizure types that are often refractory to treatment. Although most Dravet syndrome variants arise de novo, there are cases where an SCN1A variant was inherited from mildly affected parents, as well as some individuals with de novo loss-of-function or truncation mutations that presented with milder phenotypes. This suggests that disease severity is influenced by other factors that modify expressivity of the primary mutation, which likely includes genetic modifiers. Consistent with this, the Scn1a+/− mouse model of Dravet syndrome exhibits strain-dependent variable phenotype severity. Scn1a+/− mice on the 129S6/SvEvTac (129) strain have no overt phenotype and a normal lifespan, while [C57BL/6Jx129]F1.Scn1a+/− mice have severe epilepsy with high rates of premature death. Low resolution genetic mapping identified several Dravet syndrome modifier (Dsm) loci responsible for the strain-dependent difference in survival of Scn1a+/− mice. To confirm the Dsm5 locus and refine its position, we generated interval-specific congenic strains carrying 129-derived chromosome 11 alleles on the C57BL/6J strain and localized Dsm5 to a 5.9 Mb minimal region. We then performed candidate gene analysis in the modifier region. Consideration of brain-expressed genes with expression or coding sequence differences between strains along with gene function suggested numerous strong candidates, including several protein coding genes and two miRNAs that may regulate Scn1a transcript.


Introduction
Dravet syndrome is an infant-onset epileptic encephalopathy caused by haploinsufficiency for SCN1A. Seizure onset usually occurs between 6 and 18 months of age with generalized tonic-clonic or hemiclonic seizures, often precipitated by fever. Individuals subsequently develop pleiomorphic afebrile seizure types that often respond poorly to conventional therapies (Dravet 2011;Dravet and Oguni 2013;Wirrell et al. 2017). Development is normal prior to onset, but disease progression is frequently accompanied by stagnation or decline of psychomotor and cognitive development (Battaglia et al. 2021;Dravet 2011). Individuals with Dravet syndrome have a significantly increased risk of mortality, mainly attributed to prolonged status epilepticus in early childhood and sudden unexplained death in epilepsy (SUDEP) in adolescents and adults (Dravet 2011). In most cases, the SCN1A variants arise de novo and result in heterozygous loss-of-function (Li et al. 2021). However, there have been reports of Dravet syndrome caused by SCN1A variants inherited from mildly affected parents, as well as some individuals with de novo loss-of-function or premature truncation variants that presented with milder phenotypes, like generalized epilepsy with febrile seizures plus (GEFS + ) (Depienne et al. 2010(Depienne et al. , 2009Goldberg-Stern et al. 2014;Guerrini et al. 2010;Nabbout et al. 2003;Osaka et al. 2007;Yordanova et al. 2011;Yu et al. 2010). This variable expressivity suggests that disease severity is influenced by additional factors, which may include genetic modifiers (de Lange et al. 2020;Hammer et al. 2017).
In this study, we used interval-specific congenic (ISC) strains to fine map the Dsm5 locus on chromosome 11 and identified a minimal interval that influenced survival, particularly in female Scn1a +/− mice. Within the mapped interval, we performed an initial survey of potential candidate modifier genes and identified several high-priority candidate modifier genes.
Mice were maintained in a Specific Pathogen Free (SPF) barrier facility with a 14:10 light:dark cycle and ad libitum access to food and water. All animal care and experimental procedures were approved by the Northwestern University Animal Care and Use Committee in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The principles outlined in the ARRIVE (Animal Research: Reporting of in vivo Experiments) guideline were considered when planning experiments (Percie du Sert et al. 2020).

Interval specific congenic (ISC) lines
We generated nine ISC lines carrying 129-derived alleles on chromosome 11 on a C57BL/6 J (B6) background, designated as B6.129-Dsm5A through B6.129-Dsm5F. F1 progeny were generated by crossing 129 males with B6 females, and then successively crossed to B6 to generate congenic lines. Genotyping for chromosome 11 markers was performed at each generation and mice retaining 129 alleles were propagated. Whole genome and selective genotyping was performed at generations N2 and N5 to select breeders with low percentages of 129 in the rest of the genome. All lines were crossed to B6 for ≥ N9 generations prior to any experiments.

Genotyping
DNA was prepared from tail biopsies (Gentra Puregene Mouse Tail Kit, Qiagen, Valencia, CA, USA). Scn1a genotype was determined by multiplex PCR using a common primer (5'-AGT CTG TAC CAG GCA GAA CTTG) and two allele specific primers (WT: 5'-CCC TGA GAT GTG GGT GAA TAG; KO: 5'-AGA CTG CCT TGG GAA AAG CG). Amplicons include a 357 bp WT product and a 200 bp KO product. Genotyping of microsatellite markers was performed by analysis of PCR products on 7% denaturing polyacrylamide gels stained with ethidium bromide. ISC breakpoints were refined using the mini Mouse Universal Genotyping Array (miniMUGA) (Neogen, Lincoln, NE, USA). Strain background was surveyed for Trpv1 mice by miniMUGA genotyping (Neogen).

Phenotyping
B6.129-Dsm5 females were bred with heterozygous 129.Scn1a +/− males to generate F1 offspring carrying homozygous 129/129 alleles or heterozygous 129/B6 alleles in Dsm5. Offspring were ear-tagged and genotyped at P12-14. At P19-21, mice were weaned into standard vivarium cages containing 4-5 mice of the same sex and age. Wildtype littermates were included in holding cages. Survival was monitored to 8 weeks of age. Over the 8-week period, mice were monitored daily for general health and any visibly unhealthy mouse (e.g., underweight, dehydrated, poorly groomed, or immobile) was euthanized and excluded from the study; this occurred rarely. The focus of the study was sudden and unexpected death in otherwise healthy appearing Scn1a +/− mice. Survival was compared between groups using Kaplan-Meier analysis with p-values determined by LogRank Mantel-Cox tests. Group sizes were based on data simulations using data from our prior survival studies.

Candidate gene analysis
We defined the Dsm5 gene set using the Mouse GRCm38.p6 reference genome (B6) in Ensembl BioMart, which included classification by gene type. To define a subset of genes with CNS expression, we used the MGI database and EMBL-EBI Expression Atlas. Differential expression of Dsm5 genes between 129 and B6 or F1 was assessed using two RNA-Seq datasets that we previously reported: (1) B6 and 129 forebrain (Hawkins et al. 2016); and (2) F1 and 129 hippocampus (Hawkins et al. 2019) (NCBI GEO GSE112627). Differential expression of miRNAs was assessed using a previously reported dataset (Trontti et al. 2018). To identify consequential coding sequence changes between the strains, we performed whole genome re-sequencing of 129S6/SvEvTac and compared with the C57BL/6J reference sequence and mouse strains in Ensembl (Mouse GRCm38. p6) (NCBI SRA PRJNA817075). The effect of single-nucleotide variants (SNVs) was assessed using the Ensembl Variant Effect Predictor tool (VEP) (McLaren et al. 2016), and SNVs classified as deleterious were retained. TargetScan and LncRRIsearch were used to predict targets of Dsm5 high confidence miRNAs and lncRNAs, respectively (Agarwal et al. 2015;Fukunaga et al. 2019).

Fine mapping of Dsm5 to central chromosome 11
Low resolution mapping of Dsm5 on chromosome 11 localized a 1.5-LOD support interval to 4.7-39.7 cM (6.99-63.9 Mb; GRCm38.p6) (Miller et al. 2014). To refine the map interval, we used ISC lines carrying varying 129-derived chromosome 11 segments on a congenic B6 background ( Fig. 1A; Table 1). Each B6.129-Dsm5 strain was crossed to 129.Scn1a +/− to generate offspring with either homozygous 129/129 or heterozygous 129/B6 alleles in Dsm5. Survival of Scn1a +/− offspring was monitored to 8 weeks of age and compared between those carrying homozygous (129/129) alleles at Dsm5 or F1 controls with heterozygous (B6/129) alleles at Dsm5. First, we compared survival between males and females with control heterozygous alleles and observed a sex-difference (p < 0.01 LogRank Mantel-Cox), with females having worse survival than males (Hazard Ratio (HR): 1.561; 95% confidence interval: 1.088-2.240) (Fig. 1B), consistent with other reports (Gerbatin et al. 2021;Niibori et al. 2020). Therefore, all subsequent ISC comparisons were performed separately by sex. Next, we compared survival between mice that were homozygous 129/129 in each ISC interval with their sex-matched F1 (129/B6) controls. Improvement in survival was observed with some ISC lines (Fig. 1C, D). Improved survival of females was observed with 129/129 alleles from lines ISC-D (65-88 Mb), ISC-E (65-76 Mb) and ISC-F (70-88 Mb) (Fig. 1C), while males only showed improvement in survival with ISC-D. Furthermore, the magnitude of the effect of 129/129 alleles from lines ISC-D was greater in females (HR: 3.183; Table 1) relative to males (HR: 2.851; Table 1) (Fig. 1). This suggests that the effect of 129 alleles in Dsm5 is more pronounced in females. Under a single modifier gene model, the smallest 129-derived region to harbor a modifier gene would be the ~ 7 Mb region of overlap between strains ISC-E and ISC-F.
To further refine the ISC intervals, breakpoints were defined by SNP genotyping using the mini MUGA platform that includes over 11,000 SNP probes. For ISC-D and ISC-E, the proximal breakpoint was localized between rs216826621 (65.018062 Mb) and rs48061329 (65.423779 Mb

Candidate gene analysis
Under a single modifier gene model, the Dsm5 locus was narrowed to a 5.4 Mb interval (70.808812-76.169847 Mb, GRCm38.p6) that contains 126 known genes, including 12 RNA genes, and 114 protein coding genes (including 30 olfactory receptors genes). Of those, 28 protein coding and 10 RNA genes have confirmed brain expression.

Discussion
In the present study, we constructed a set of ISCs to fine map the Dsm5 locus on mouse chromosome 11. Under a single gene modifier model, our results narrow the modifier interval to a 5.7 Mb region of overlap between ISC-E and ISC-F. Within this interval, we identified a number of potential candidate modifier genes with coding sequence variation and/or expression differences between the strains. However, future studies will be needed to validate candidate modifier genes.
Among the Dsm5 potential modifier genes, Trpv1 is an intriguing candidate as a proposed target of cannabidiol, an FDA/EMA-approved Dravet syndrome therapeutic (Gray and Whalley 2020). Recently, we reported differential expression of Trpv1 transcript in cortex, with higher expression in F1.Scn1a +/− mice relative to seizure resistant 129.Scn1a +/− mice (Satpute Janve et al. 2021). However, deletion of Trpv1 had both pro-and anti-convulsant effects when combined with the Scn1a +/− allele. Double mutant F1.Scn1a +/− ;Trpv1 +/− mice had lower temperature thresholds for hyperthermia-induced seizure compared to F1.Scn1a +/− mice, suggesting a pro-convulsant effect. Conversely, Trpv1 deletion resulted in reduced severity of spontaneous GTCS in double mutant F1. Scn1a +/− ;Trpv1 +/− mice, although it did not affect seizure frequency or survival compared to F1.Scn1a +/− mice (Satpute Janve et al. 2021). Although Trpv1 deletion alters seizure phenotypes in Scn1a +/− mice, it is not possible to separate an effect of Trpv1 deletion from residual 129 alleles in the chromosome 11 region remaining from the homologous  (Caterina et al. 2000). Lack of any effect of a Trpv1-selective antagonist on F1.Scn1a +/− phenotypes lends support for an effect of residual 129 alleles rather than Trpv1 deletion itself on Scn1a +/− seizure phenotypes (Satpute Janve et al. 2021). Additional intriguing candidates include the miRNA genes for miR-132-3p and miR-212-3p that are predicted to target Scn1a and are seizure-responsive genes. However, to date, we have no evidence for altered expression of Scn1a transcript between strains or in response to seizures. It is possible that miRNA regulation could be post-transcriptional and alter protein levels, The Dsm5 region overlaps previously identified seizure susceptibility QTL, including Szs10 and Gss4 (Ferraro et al. 2001;Gu et al. 2020). In addition, this region of mouse chromosome 11 is syntenic with recurrent CNVs implicated in epilepsy and neurodevelopmental disorders (Coppola et al. 2019;Kolishovski et al. 2019). This includes synteny with Miller-Dieker lissencephaly 17p13.3 microdeletion syndrome, which includes epilepsy as part of the neurological features. The 17p13.3 microdeletion includes deletion of PAFAH1B1 and YWHAE, as well as deletion of the intervening genes. The lissencephaly phenotype is attributed to deletion of PAFAH1B1, formerly known as LIS1, while disruption of YWHAE, encoding 14-3-3ε, is associated with variable structure brain abnormalities, cognitive impairment and seizures (Cardoso et al. 2000;Noor et al. 2018;Romano et al. 2020). Rare microduplications at 17q12 are associated with epilepsy, including familial FS/GEFS + that is also often caused by SCN1A variants (Hardies et al. 2013;Fig. 3 String protein association network for candidate protein coding modifier genes and Scn1a. Only Trpv1 had a direct association with Scn1a, while Aspa and Trpv3 had indirect associations. The 30 protein coding candidate genes in Table 2 and Scn1a were used as input in String (v. 11.0) (Szklarczyk et al. 2019). At least medium confidence associations are displayed (interaction score ≥ 0.4). Edges represent protein associations, with line color indicating the type of evidence as follows: yellow, text-mining; black, co-expression; pink, experimentally determined interactions; blue, known interactions from curated databases; purple, protein homology Mefford et al. 2016Mefford et al. , 2007. Genes associated with the 17q12 microduplication are AATF, ACACA, C17orf78, DDX52, DHRS11, DUSP14, GGNBP2, HNF1B, LHX1, MRM1, MYO19, PIGW, SYNRG, TADA2A, and ZNHIT3 (Mefford et al. 2016). LHX1, encoding LIM homeobox protein 1, and ACACA , encoding acetyl-CoA carboxylase, have been proposed as potential candidate genes responsible for neurodevelopmental and epilepsy phenotypes, although there is not sufficient evidence to confirm causality to date (Hardies et al. 2013;Mefford et al. 2007). The 17p13.3 deletion overlaps with the region of overlap between ISC-E and ISC-F, while the 17q12 microduplication overlaps with the nonoverlapping segment of ISC-F.
One caveat of our study is that the fine mapping data could also support an alternative interpretation of additive modifier genes from the non-overlapping portions of ISC-E and ISC-F. This seems less likely when examining the Kaplan-Meier survival curves (Fig. 1C, D), where the net effect of ISC-E, ISC-F or ISC-D appear similar and are not significantly different from one another (p > 0.3, LogRank Mantel-Cox). However, examination of the female hazard ratios shows a modest trend toward smaller magnitude of effects for ISC-E (2.258 HR) and ISC-F (2.162 HR) relative to ISC-D (3.183 HR) (Fig. 1C), which could support the possibility of each interval harboring modifiers with additive effects. Additional support for this hypothesis comes from males that showed robust improvement in survival with ISC-D, whereas ISC-E or ISC-F alone had no protective effect. Ideally, these two possibilities could be tested empirically by generation of ISC lines separating the E-F overlapping and non-overlapping regions; however, such recombination events have been elusive in our ISC colony. It may be possible in the future to use genome engineering to further dissect the interval.
Haploinsufficiency for SCN1A is a major cause of Dravet syndrome, and yet there is variable expressivity among patients with this shared genetic basis. This suggests that clinical presentation is influenced by other factors, which may include genetic modifiers. Identification of modifier genes that influence disease course will provide insights into understanding the molecular basis of Dravet syndrome and identify potential pathways for intervention. Data availability Genomic and transcriptomic datasets generated during and/or analyzed during the current study are available in the NCBI GEO repository [GSE112627] and SRA repository [PRJNA817075]. Other datasets are available from the corresponding author on reasonable request.

Author Contributions
Code availability Not applicable.

Declarations
Conflict of interest JAK serves as a consultant to Encoded Genomics, Praxis Precision Medicines and NeuroCycle Therapeutics, and serves on the Scientific Advisory Boards of the Dravet Syndrome Foundation and FamilieSCN2A Foundation. JAK receives research funding from Praxis Precision Medicines and Pfizer. NAH serves as a consultant to NeuroCycle Therapeutics. All other authors have declared that no competing interests exist.
Ethical approval All animal care and experimental procedures were approved by the Northwestern University Animal Care and Use Committee in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.