Analyse in silico de l'impact des variants de RNU4-2 sur la structure et la fonction de l'ARN U4 dans le syndrome ReNU

Abstract

RNU4-2-related neurodevelopmental syndrome (ReNU syndrome) is a frequent monogenic cause of severe intellectual disability, caused by de novo variants affecting the RNU4-2 gene, which encodes the U4 small nuclear RNA, a key component of the major spliceosome. Functional interpretation of these variants remains challenging, as standard prediction tools were designed for protein-coding sequences and perform poorly on non-coding RNAs. This study presents an in silico analysis of 61 RNU4-2 variants retrieved from ClinVar, classified as either pathogenic or likely pathogenic (Patho/LP, n = 30) or of uncertain significance (VUS, n = 31), using three complementary metrics: ΔΔG RNAfold (secondary structure stability), ΔΔG RNAcofold (U4/U6 interaction energy), and RNAdistance (topological reorganization). ΔΔG RNAfold was the only metric significantly differentiating the two groups (p = 0.012; Cliff's delta = −0.37), yet Patho/LP variants showed a median ΔΔG of zero, suggesting that global RNA destabilization is not the predominant pathogenic mechanism. The strongest discriminating signal was variant localization within the critical region (T-loop + Stem III): 18 of 30 Patho/LP variants mapped to this region versus only one VUS (OR = 45.0; p = 9.7 × 10⁻⁷). Variants n.7G>C, n.140G>A, and n.129G>C/A were identified as priority candidates for experimental functional validation. These findings highlight the mechanistic heterogeneity underlying ReNU syndrome and support a multi-criteria approach combining structural analysis and functional localization for the interpretation of variants in non-coding RNA genes.