A Systems approach identifies MERTK as a therapeutic vulnerability in ZFTA-RELA-driven ependymomas

Abstract

Ependymomas (EPN) are rare central nervous system tumors that account for approximately 10% of intracranial tumors in children and 4% in adults. Despite their clinical and molecular heterogeneity, spanning supratentorial, posterior fossa, and spinal subtypes, treatment remains limited to surgery and radiotherapy, with chemotherapy offering minimal benefit. Here, we performed transcriptomic analysis of 370 human ependymoma samples and identified two distinct molecular subgroups: EPN-E1 and EPN-E2. The EPN-E1 cluster is enriched for supratentorial tumors harboring ZFTA-RELA fusions (ZFTA-RELAfus), which occur in over 70% of cases and are associated with poor prognosis. To identify targeted therapies for this aggressive subtype, we validated a ZFTA-RELAfus mouse model that recapitulates the human EPN-E1 transcriptome and used it for target discovery. Through Kinome Regularization, a machine learning-driven polypharmacology approach, we identified MERTK as a critical regulator of tumor cell viability. Genetic depletion or pharmacologic inhibition of Mertk reduced cell growth ex vivo, and treatment with a clinical-grade MERTK inhibitor significantly suppressed tumor proliferation in vivo. Both human EPN-E1 tumors and ZFTA-RELAfus mouse tumors exhibited elevated expression of MERTK and its ligand GAS6, and MERTK inhibition led to suppression of pro-survival signaling pathways including MEK/ERK and PI3K/AKT. Notably, over 80% of genes upregulated in ZFTA-RELAfus tumors were downregulated following MERTK inhibition, indicating a strong dependency on this pathway for tumor maintenance. These findings define a signaling vulnerability in ZFTA-RELA-driven ependymomas and support the clinical development of MERTK-targeted therapies for patients with the high-risk EPN-E1 subtype.