TMIC-24. Single-cell decoding of the genotype-immunophenotype relationship in glioblastoma

Abstract Glioblastoma (GBM) is the most aggressive and lethal adult brain tumor. The cellular heterogeneity within the tumor microenvironment (TME) plays a critical role in the complexity of treatment and poor survival. GBM is typically classified into 3 molecular subtypes—Classical, Mesenchymal, and Proneural—associated with EGFR, NF1, and PDGFRA genetic drivers, respectively. Our team previously generated human GBM-specific driver mutations using RCAS/tv-a gene transfer system to create immunocompetent genetically engineered mouse models (GEMMs) of murine GBM (mGBM) that share the same cell-of-origin. While we and others have shown that these GEMMs closely mimic myeloid cell composition and bulk expression profiles observed in human GBM (hGBM) subtypes, it is critical to identify species-specific similarities and differences in tumor microenvironment (TME) heterogeneity at cellular resolution. Hence, we reasoned that single-cell characterization of mGBM models and comparison to hGBM would highlight the cross-species similarities/differences and enable investigators to identify and validate the most relevant human target candidates. Towards this goal, we performed single-cell RNA sequencing (scRNA-seq), multicolor flow cytometry, and in-depth computational analysis on EGFR-, Nf1-, and PDGFB-driven GEMMs to dissect the effect of common GBM genetic drivers on the TME. To assess the human relevance, we compared our findings to bulk RNA-seq data from human GBM tumors with mutually exclusive EGFRvIII deletion, NF1 loss of function mutation, or PDGFRA amplification on chromosome 4. Our analysis revealed that PDGFB-driven tumors were more proliferative and enriched for Wnt signaling interactions, while EGFRvIII-driven tumors showed an elevated interferon signaling response. Moreover, Nf1-silenced tumors displayed higher myeloid abundance, myeloid immunosuppressive interactions involving Osteopontin, Treg infiltration, and expression of immune checkpoint molecule Ctla4. Overall, we established a human-mouse analytical platform for genotype-aware target discovery and validation, which offers promising new avenues for more effective, personalized treatments in GBM.