Atypical teratoid rhabdoid tumors (ATRTs) are among the most lethal pediatric central nervous system (CNS) tumors, which only harbor one single genetic driver mutation, the bi-allelic loss of the BRG1/BRM-associated factors (BAF) complex member SMARCB1. Based on DNA methylation profiles, three molecular subtypes have been further identified: ATRT-SHH, ATRT-MYC, ATRT-TYR. Each subtype has their own molecular and clinical features. Due to the very low occurrence rate (1–2 cases per million children per year), there are currently no subtype-specific treatments available, highlighting the necessity to better understand inter- and intra-subtype heterogeneity. Because paediatric cancers are often caused by block of developmental maturation, this knowledge will also be crucial for development of maturation therapy (inducing cancer cells towards more mature, less aggressive state).

We generated an unique single-nucleus transcriptome atlas of 19x ATRTs covering all three subtypes, validated by single-nucleus ATAC-seq and spatial transcriptomics, to study subtype-specific differentiation trajectories. Patient-derived-tumoroid models were used to validate potential maturation-incuding drugs.

We discovered brain progenitor-like expression profiles within unique subtype-specific differentiation lineages. A shared cycling, progenitor-like cell population, interspersed throughout tumors, was observed across all ATRT samples. Additionally, top hits for maturation therapies, e.g., HDAC inhibitor Entinostat and PKC inhibitor Ro31-8220 were tested in our ATRT tumoroids, pointing that subtype-specific differentiation trajectories can be induced pharmacologically to push tumor cells towards a more mature and less-proliferative state.

Our study decodes ATRT subtype-specificity at single cell level and reveals that each ATRT subtype is enriched for distinct signaling trajectories mirroring normal fetal brain development, enabling the development of maturation therapies tailored towards ATRTs.