A chemoinformatics-guided platform for efficient discovery of RNA-binding small molecules: Proof-of-concept for myotonic dystrophy type 1

Structured RNAs cause human diseases but remain challenging to target selectively with small molecules. Here, we report a chemoinformatics-guided discovery framework that integrates fingerprint-based molecular design, experimental validation, and mechanistic profiling to identify small molecules that bind highly structured, disease-associated RNAs. Using an RNA-binder fingerprint derived from known ligands, a Tversky similarity screen of >8 million compounds yielded a 150-member library enriched in chemical space for RNA-active scaffolds. Target engagement and cell-based assays identified multiple selective ligands for the pathogenic expanded triplet repeat, r(CUG)exp, that causes myotonic dystrophy type 1 (DM1) by binding and sequestering the RNA-binding protein muscleblind-like 1 (MBNL1). Biophysical and single-molecule analyses revealed that the small molecules bind the 1x1 nucleotide U/U internal loops formed when r(CUG)exp folds, partially block MBNL1 binding, and modulate RNA folding equilibria. Two optimized scaffolds rescued MBNL1-dependent splicing in patient-derived myotubes with micromolar potency and minimal cytotoxicity. This study establishes a generalizable, data-driven platform for discovering drug-like RNA-binding lead small molecules and demonstrates its application to the toxic repeat expansion RNA underlying DM1.