Widespread genomic influences on phenotype in Dravet syndrome, a 'monogenic' condition

Abstract

Dravet syndrome is an archetypal rare severe epilepsy, considered 'monogenic', typically caused by loss-of-function SCN1A variants. Despite a recognizable core phenotype, its marked phenotypic heterogeneity is incompletely explained by differences in the causal SCN1A variant or clinical factors. In 34 adults with SCN1A-related Dravet syndrome, we show additional genomic variation beyond SCN1A contributes to phenotype and its diversity, with an excess of rare variants in epilepsy-related genes as a set and examples of blended phenotypes, including one individual with an ultra-rare DEPDC5 variant and focal cortical dysplasia. The polygenic risk score for intelligence was lower, and for longevity, higher, in Dravet syndrome than in epilepsy controls. The causal, major-effect, SCN1A variant may need to act against a broadly compromised genomic background to generate the full Dravet syndrome phenotype, whilst genomic resilience may help to ameliorate the risk of premature mortality in adult Dravet syndrome survivors.

Keywords: SCN1A; Dravet syndrome; blended phenotypes; polygenic risk scores; polymorphism.

Figure 1

Method for selection of variants in epilepsy-related genes. Method for selection of variants in epilepsy-related genes with ‘potential clinical relevance’ that may contribute to blended phenotypes. GEL = Genomics England.

Figure 2

Distribution of SCN1A variants found in the Dravet syndrome cohort. A schematic diagram of the SCN1A gene. Exons are indicated by vertical black boxes (1–29) and introns by the horizontal black line (not to scale). Missense (purple), splicing (dark blue), frameshift insertion (light blue), frameshift deletion (green) and stop-gain (red) variants are shown. The whole gene deletion is not shown. Variants are shown according to the NM_001165963.4 reference sequence.

Figure 3

FCD and details of DEPDC5 variant. (A) Brain MRI showing FCD. Coronal T₁-weighted brain MRI from Case 1-102398, with DEPDC5 variant NM_001242896.3:c.G4183A:p.A1395T, showing left temporal lobe FCD (right of patient is on the left of the image, following radiological convention), with blurred grey–white interface and cortical thickening apparent in the left temporal lobe across several consecutive slices. (B) MetaDome map of regional constraint in DEPDC5. Grey bar below the graph represents the protein, pink bars showing Pfam domains: PF12257, Vacuolar membrane-associated protein Iml1 domain; PF00610, Domain found in Dishevelled, Egl-10, and Pleckstrin (DEP); A1395 is marked by a vertical green line, with a reported tolerance score of 0.28 (‘intolerant’). (C) VarSite sequence logo for DEPDC5 residues 1375–1414, based on alignment of structural homologues; below the logo is the sequence of DEPDC5 itself, with A1395 boxed; sequence conservation score for this residue was 0.92 [range 0 (low)–1 (high)]; alanine was observed at this position in 31/33 aligned sequences. (D) Structure of the GATOR1-Rag GTPases complex and context of DEPDC5 Ala1395. PDB 6ces, the structure of the heterotrimeric GATOR1 complex (DEPDC5:NPRL2:NPRL3) bound to RagA and RagC GTPases; protein surfaces shown by colour as indicated (except DEPDC5, shown as a ribbon and coloured by structural domains as annotated by Shen et al.: bright green = N-terminal domain (NTD) (residues 38–165); cyan = SABA domain (166–425); orange = steric hindrance for enhancement of nucleotidase activity (SHEN) domain (721–1010); dark green = DEP domain (1175–1270); violet = C-terminal domain (CTD) (1271–1600); Ala1395 is pink with sidechain atoms shown as spheres. (E and F) Ala1395 lies at an inter-domain interface in DEPDC5. The figure shows selected residues of DEPDC5 from PDB 6ces (chain D); residues of the NTD, SABA domain and CTD are shown as separate surfaces; residues of the SHEN domain and DEP domain are shown as ribbons. F shows the same structure as E with SHEN and DEP domains removed; residues Tyr108 (bright green), Phe326 (blue) and Ala1395 (rose pink) lie in close proximity at a three-way interface between the NTD, SABA and CTD. (G) Enlarged image of the DEPDC5 structure (PDB 6ces, chain D) as in E and F, zoomed to show detail around the three-way interface between the NTD, SABA and CTD; (H) The Ala1395Thr substitution results in reduced space at the inter-domain interface in 6cesD. This figure shows the same structure as G after introduction of the Ala1395Thr variant by in silico mutagenesis. Quantitative results are given in Supplementary material 18. Analysis of DEPDC5 from PDB 6cet is shown in Supplementary Fig. 1

Figure 4

PRS applied across the cohorts. (A) Polygenic risk score (PRS) for intelligence was lower in the Dravet syndrome cohort than in GEL Epilepsy (adjusted P = 0.0024) and GEL control cohorts (adjusted P = 0.003). The difference between GEL Epilepsy and GEL controls was not significant (adjusted P = 0.69). (B) PRS for longevity was significantly higher in the Dravet syndrome cohort than in GEL Epilepsy controls (adjusted P = 0.011), and higher than, but not significant, in GEL controls (adjusted P = 0.024) and not significantly different in GEL Epilepsy controls compared to GEL controls (adjusted P = 0.68). (C) PRS for epilepsy was not significantly different in the Dravet syndrome cohort compared with the GEL controls (adjusted P = 0.89) and GEL Epilepsy controls (adjusted P = 0.11). PRS for epilepsy was significantly higher in the GEL Epilepsy controls than in the GEL controls (adjusted P < 2.22 ×10⁻¹⁶). The per-PRS P-values shown in the graphics are estimated using a post hoc multiple pairwise comparison (Tukey’s test). As multiple PRS analyses were performed, the final adjusted P-value significance threshold was set to α = 0.05/3. GEL = Genomics England.

Figure 5

PRS applied across the GEL SCN1A control and Dravet syndrome cohorts. (A) Polygenic risk score (PRS) for intelligence was lower, but not significant, in the Dravet syndrome cohort than in GEL SCN1A controls (adjusted P = 0.033). (B) PRS for longevity was higher, but not significant, in the Dravet syndrome cohort than in GEL SCN1A controls (adjusted P = 0.049). (C) PRS for epilepsy was not significantly different between the Dravet syndrome cohort and GEL SCN1A controls (adjusted P = 0.28). Black circles = individuals from the GEL SCN1A control cohort with variants previously reported to be associated with disease. The per-PRS P-values shown in the graphics are estimated using a post hoc multiple pairwise comparison (Tukey’s test). As multiple PRS analyses were performed, the adjusted P-value significance threshold was set to α = 0.05/3. GEL = Genomics England.