KMT2A and KMT2B episignatures address diagnostic challenges associated with rare neurodevelopmental disorders.

Abstract

Pathogenic variants in KMT2A and KMT2B, encoding histone H3 lysine 4 methyltransferases, cause distinct neurodevelopmental disorders-Wiedemann-Steiner syndrome (WDSTS) and Dystonia 28 (DYT28), respectively. We generated and validated DNA methylation (DNAm) signatures for both disorders using cohorts with truncating and missense variants. The generated signatures demonstrate high sensitivity and specificity, the KMT2A-WDSTS signature consists of 264 CpG sites, achieving >80% pathogenicity scores in validation, and the KMT2B-DYT28 signature including 752 mostly hypermethylated CpG sites, with >85% pathogenicity scores. The two signatures show limited overlap with each other, this reflects variant- and cohort-specific methylation patterns. We applied these signatures to classify variants of uncertain significance, finding strong agreement with in-silico predictions and the available phenotypic data, particularly for variants in functional domains. Notably, some KMT2A variants in the transactivation domain and KMT2B variants in disordered regions showed control-like methylation, indicating alternative mechanisms of pathogenesis. Gene ontology analysis revealed KMT2A signature genes are enriched for embryonic development and transcriptional regulation, notably HOX genes, while KMT2B signature genes relate to ion transport and enzymatic activity. These gene-specific DNAm signatures improve variant interpretation and diagnosis, especially for uncertain and atypical cases, underscoring the utility of integrating epigenetic, genetic, and phenotypic data in clinical diagnostics.