Identification of a SCN2A variant in Family 1
In Family 1, six members had attacks fulfilling the ICHD-3 diagnostic criteria for FHM, which started between ages 4 and 11 years, and consisted of transient motor deficits with sensory disturbances, dysarthria, aphasia, and/or visual disturbances, preceding migraine headache (Fig. 1A). The aura typically lasted between 3 and 12 hours. Attacks could be triggered by anxiety, heat or fatigue. Two individuals reported prolonged HM auras lasting four days (III-4) to one month (II-1). Neurological examination between attacks was normal. None of the family members had epilepsy, ataxia, abnormal movements or intellectual disability.
A genome-wide linkage scan was conducted in Family 1. Multipoint linkage analyses identified seven genomic regions on chromosomes 1, 2, 3, 7, 10, 14 and 19 that reached the maximum theoretical LOD score (max LOD = 1.4). The linked regions had a cumulated physical size of 190.7 Mb and included 1844 known genes (Supplemental Table S1, Supplemental Figure S1).
TruSight One sequencing was performed in individuals III-1 and IV-4 from Family 1. Only missense, nonsense, insertion/deletion variants or variants potentially affecting splicing were considered for further analysis. Variants present in dbSNP132, as well as those reported with a minor allele frequency > 0.1% in the 1000 Genomes Project database and the NHLBI Exome Variant Server (EVS) were excluded [13]. Three variants with a frequency < 0.1% in the databases and located in one of the candidate regions were identified: SCN2A (NM_001040142) c.4438A > G, p.Lys1480Glu; MLH3 (NM_001040108.) c.3548A > G, p.Lys1183Arg; and SERPINA1 (NM_001002235) c.1177C > T, p.Pro393Ser. No pathogenic or likely pathogenic variants were observed in CACNA1A, ATP12A, SCN1A or PRRT2.
SCN2A was considered a good candidate because it encodes the pore-forming subunit of a voltage-gated sodium channel (NaV1.2), which is highly expressed throughout the brain including the cerebellum and is a known cause of paroxysmal disorders of the nervous system [14]. The SCN2A variant p.Lys1480Glu (K1480E) discovered in Family 1 converts a highly conserved positively charged lysine located in the intracytoplasmic loop between the third and the fourth domains of NaV1.2 (Fig. 1B, Fig. 2) to a negatively charged glutamate. This variant is absent in the Genome Aggregation Database (gnomAD) [15] and is predicted to be pathogenic by SIFT, MutationTaster, and PolyPhen-2 algorithms [16–18]. Sanger sequencing demonstrated the presence of the SCN2A variant in all affected members of Family 1, but none of the unaffected individuals including the two with uncertain phenotype (Fig. 1A).
Identification of other SCN2A variants
Screening of probands from six other FHM families detected another SCN2A variant in Family 2 (c.769T > A, p.Phe257Ile, F257I; Fig. 1C, D). Further sequencing in Family 2 showed that this variant was present in two additional affected members. Phenylalanine 257 is located in the transmembrane S5 segment of the first domain (Fig. 2). This variant was absent in the gnomAD database and predicted to be pathogenic by SIFT, MutationTaster, and PolyPhen-2 algorithms.
In Family 2, four members (II-2, II-3, II-5 and III-5) had typical HM attacks. Individual II-2 reported that with ageing, HM attacks became rare, with milder motor symptoms and increased headache severity. Her two young children (III-1 and III-2) had self-limited infantile seizures during their first year of age that resolved before the onset of other paroxysmal manifestations. Individual III-1 (4 years old) had recurrent paroxysmal episodes of dysarthria, visual and motor deficits lasting from one to seven days, while III-2 (2 years old) had a single episode of loss of walking and cerebellar syndrome triggered by a minor head trauma, which resolved after a few days. No pathogenic or likely pathogenic variants were observed in CACNA1A, ATP12A, SCN1A and PRRT2.
In the series of 594 unrelated probands, a variant of interest was detected in SCN2A in one patient (c.3955C > G; p.Arg1319Gly, R1319G), which was absent in gnomAD, had a CADD score of 25.4, and was predicted to be probably damaging by Polyphen and deleterious by SIFT. This individual started at the age of six to have right or left-sided hemiplegic attacks lasting 20 minutes followed by contralateral severe headaches lasting several hours. During adolescence, he also had migraine with visual aura, manifesting as transient homonymous lateral hemianopia followed by headache, photophobia and phonophobia lasting several days. He had no epilepsy and a normal neurological examination. EEG and cerebral MRI were unremarkable. At the age of inclusion (43 years old), he still had one HM attack per month. Neither parent had HM episodes. They were unavailable for genetic analysis.
Functional analysis of the three SCN2A variants
We employed automated patch clamp recording to analyze the functional properties of the HM-associated SCN2A variants F257I, R1319G, and K1480E. In vitro functional analysis revealed that all three variants produced functional NaV1.2 channels, but with biophysical properties divergent from the wildtype (WT) channel assayed in parallel. While F257I and R1319G exhibited whole-cell sodium current density indistinguishable from WT NaV1.2, K1480E had significantly smaller whole-cell sodium current density (Fig. 3A, B, Supplemental Table S2). Additionally, all variants mixed biophysical defects including depolarized voltage-dependence of activation (R1319G and K1480E), depolarized voltage-dependence of inactivation (F257I and K1480E), and larger window current (all variants) (Figs. 3D-E, Supplemental Table S2). Window current is a property defined by the overlap of the voltage-dependence of activation and inactivation curves, and thus represents a voltage range where channels may be activated but not inactivated. Interestingly, variants also exhibited slower recovery from inactivation, slower onset of inactivation, and greater loss of activity in response to repeated depolarizations (Fig. 4, Supplemental Table S2). While these functional features support pathogenicity of the SCN2A variants, the constellation of functional perturbations cannot be easily classified as simply gain- or loss-of-function.