TMOD-01. Prime Editing of ATRX point mutations in glioblastoma reveals lethal dependencies and targetable vulnerabilities

Abstract Glioblastoma multiforme (GBM) is a highly aggressive and treatment-resistant brain tumour. ATRX mutations are frequently observed in both adult and paediatric GBM, including many single-residue missense mutations. These remain underexplored due to technical limitations in modelling, as conventional knockout models do not fully reflect their biology. This study aimed to establish a Prime Editing-based platform to precisely model clinically recurrent ATRX mutations that have been previously intractable with existing tools and to investigate their functional and therapeutic implications. We analysed the COSMIC database and identified 186 ATRX single-residue variants in glioma. Nine recurrent mutations were shortlisted, and three were selected based on feasibility for Prime Editing—a precise DNA engineering method in which a Cas9-nicked site primes synthesis of replacement DNA via reverse transcription from a guide RNA-encoded template. Editing efficiency was enhanced using our artificial intelligence (AI)-designed MLH1 small binder (PE7-SB system). For each target site, we designed more than eight Prime Editing schemes—combinations of specialized RNA sequences—and evaluated editing outcomes in human cancer cell lines. Editing attempts in multiple GBM cell lines consistently resulted in lethality, suggesting these ATRX mutations may be incompatible with GBM cell survival without additional adaptations. In contrast, sixteen viable mutant clones were generated in HeLa and HCT116 cells. These clones demonstrated increased sensitivity to a compound in clinical trials for ATRX-deficient tumours, enabling functional and therapeutic exploration of recurrent ATRX mutations. Our Prime Editing platform, enhanced by AI-guided design, enables precise and efficient modelling of clinically observed ATRX mutations. This approach uncovers novel biological constraints and lethal dependencies in glioblastoma. By enabling residue-level interrogation of tumour-associated gene function, this toolkit supports the broader neuro-oncology research community in dissecting key molecular alterations underlying glioma biology and lays the groundwork for future functional studies and therapeutic exploration.