Intellectual disability and structural defects of the CaV2.1 channel in episodic ataxia type 2: correlation using an AI prediction model

BACKGROUND: Episodic ataxia type 2 (EA2) results from pathogenic variants in CACNA1A that encodes the CaV2.1 P/Q-type calcium channel. The molecular basis of cognitive impairments requires further elucidation in EA2. OBJECTIVE: To correlate AI-predicted structural alterations of the CaV2.1 channel with intellectual function observed in patients with EA2. METHODS: Using AlphaFold3, we modeled the wild-type and variant CACNA1A proteins. Structural similarity between the wild-type and variant proteins was quantified using the Template Modeling (TM) score. To assess degree of truncation, the relative amino acid length ratio (AA%) was also calculated. These protein-level metrics were then compared with the standardized intellectual indices in 13 patients with EA2. RESULTS: The TM scores ranged from 0.624 to 0.838, and showed a strong correlation with most intellectual indices, including the full-scale IQ (FSIQ, r = 0.722, p = 0.005), verbal comprehension index (VCI, r = 0.834, p < 0.001), perceptual reasoning index (PRI, r = 0.624, p = 0.023), and working memory index (WMI, r = 0.700, p = 0.008). The AA% ranged from 50.6% to 100%, and also showed a correlation with VCI (r = 0.566, p = 0.044) and WMI (r = 0.649, p = 0.016), but less consistently when compared to the TM score. CONCLUSIONS: Structural preservation of CaV2.1 correlates more strongly with intellectual function in patients with EA2 than protein length, which suggests that structural disruption of the CaV2.1 channel may contribute to cognitive impairments in EA2. AI-based protein modeling is a valuable tool for linking genotype to phenotype, particularly in channelopathies with diverse clinical presentation.