A cycling, progenitor-like cell population at the base of atypical teratoid rhabdoid tumor subtype differentiation trajectories

Atypical teratoid rhabdoid tumors (ATRTs) are highly aggressive pediatric central nervous system tumors defined by the inactivation of the SMARCB1 gene. Despite the identification of three distinct molecular subtypes, each defined by unique clinical and molecular characteristics, no subtype-specific therapeutic strategies are currently available. This highlights an urgent need to deepen our understanding of the cellular heterogeneity and developmental origins of ATRTs. We generated a comprehensive single-nucleus transcriptomic atlas of ATRT samples, integrated it with single-nucleus ATAC-seq and spatial transcriptomics data, and validated our findings experimentally using patient-derived ATRT tumoroid models. Our analyses revealed distinct subtype-specific differentiation trajectories, each resembling different brain progenitor lineages. We identified key transcription factors that appear to drive these developmental pathways. Furthermore, a shared cycling, intermediate precursor cell (IPC)-like cell population, interspersed throughout tumors, was consistently present within all ATRT samples. We demonstrate that these subtype-specific differentiation pathways can be pharmacologically manipulated in patient-derived ATRT tumoroids. By directing tumor cells along their respective subtype-specific trajectories, we were able to induce a shift toward more differentiated, non-proliferative states. Collectively, our findings show that ATRTs recapitulate fetal brain signaling programs in a subtype-specific manner. This work provides a framework for understanding ATRT heterogeneity and supports the feasibility of maturation-based therapeutic strategies tailored to the molecular subtype of the tumor.