Single-cell systems network and machine learning dissection of ferroptosis in recurrent glioblastoma highlights astrocytic ZFP36 as a therapeutic hub for network medicine repurposing

BACKGROUND: Recurrent glioblastoma multiforme (rGBM) arises after conventional treatment strategies for primary GBM (pGBM) fail, leading to a more aggressive, therapy-resistant phenotype, with an overall survival of ∼9 months. Current treatment strategies rely on apoptosis and are frequently ineffective. Moreover, ferroptosis has recently been identified as a potential alternative cell death pathway in glioma. Therefore, this study investigated the mechanisms and therapeutic potential of ferroptosis in rGBM at the glial cell type level. METHODS: We analyzed a publicly available single-cell datasets to profile tumor-specific glial cells and assess ferroptosis-related perturbations in rGBM versus pGBM. Network-based machine learning analyses were employed to highlight key ferroptosis drivers in rGBM. Finally, we employed network medicine, molecular docking, and molecular dynamics simulations to identify a potential therapeutic target. RESULTS: Network-based ranking captured a ferroptosis suppressor, ZFP36, in astrocytes, and pathway enrichment analysis of its interaction subnetwork revealed MAPK and metabolic pathways, supporting its role in tumor growth. Network medicine identified drugs targeting ZFP36, with proximity analysis showing that drug-binding MAP kinases are closer to ZFP36, suggesting that these drugs may also modulate ZFP36 activity. Furthermore, similarity analyses with known ferroptosis inducers identified 4-Phenoxy-N-(Pyridin-2-Ylmethyl)Benzamide as a potential drug target for ZFP36. Molecular docking and dynamics simulations showed that the drug binds to the druggable binding site of ZFP36, modulating its structure. CONCLUSION: The findings of this study suggest that modulating ferroptosis suppressor ZFP36 could promote cell death in rGBM via the ferroptosis pathway.