Analysis of R-loop forming regions discovers RNU2-2P and RNU5B-1 as non-coding neurodevelopmental disorder genes

초록

Background: The contribution of DNA-RNA hybrid R-loops to Mendelian disorders has not been explored. Material and methods: Genomic (rare disease cohorts and general population cohorts), expression (human brain and retina samples) and phenotype data analyses (Genomics England data and clinical data) and functional (yeast growth and splicing assays) studies were performed. Results: De novo R-loop region variants (RRVs) were highly frequent across the genome but RRVs in ribozyme, snoRNA and snRNA genes were enriched only in rare disease individuals. Interrogation of RRVs in these genes across multiple rare disease datasets revealed 36 unsolved individuals with rare RNU2-2P or RNU5B-1 variants, in constrained regions, predicted to disrupt the interactions of the encoded snRNAs with other components of the major spliceosome complex. Both snRNA genes were highly expressed in human brain and retina snRNA-Seq data. As snRNAs are highly conserved between humans and yeast, functional growth and splicing assays in yeast have been used initially to determine the dominate effects of the variants identified in patients. The individuals with RNU2-2P or RNU5B-1 variants were enriched for multiple neurodevelopmental HPO terms such as global developmental delay and seizures. Speech impairment was prominent in RNU2-2P-related, and relative macrocephaly and failure to thrive characterised RNU5B1-related disorder. Conclusion: We report two novel disorders and demonstrate the utility of incorporating non-coding DNA secondary structure information in variant analysis. De novo RRVs in non-coding RNU genes explain an exceptionally large proportion (~1.5%) of individuals with previously unsolved neurodevelopmental disorders.