Interrogating intra-tumoral heterogeneity and treatment resistance in Glioblastoma patient-derived xenograft models using single cell RNA sequencing

Despite available treatment options for glioblastoma (GBM), GBM has one of the poorest prognosis, resist treatment, and recur aggressively in the majority of cases. Intra-tumoral heterogeneity and phenotypic plasticity are major factors contributing to treatment resistance and underlie tumor escape in GBM. Several potential therapeutic agents showing promising therapeutic effects against GBMs at the preclinical level failed to translate into effective therapies for GBM patients. This is partly attributed to the inadequacy of preclinical models to fully recapitulate the complex biology of human GBMs. This project aimed to characterize the transcriptomic heterogeneity and understand the dynamic GBM ecosystem in patient-derived xenograft (PDOX) models at the single-cell level. To achieve this aim, I established cell purification and cryopreservation protocols that enable the generation of high-quality single-cell RNA seq data from PDOX models including longitudinal and treated PDOXs. Different computational strategies were used to interrogate the transcriptomic features as well as the interactions between GBM cells and the surrounding microenvironment. This work critically analyzed and discussed key components contributing to intra-tumoral heterogeneity and phenotypic plasticity within the GBM ecosystem and their potential contributions to treatment resistance. Here, we provide evidence that PDOX models retain histopathologic and transcriptomic features of parental human GBMs. PDOX models were further shown to recapitulate major tumor microenvironment (TME) components identified in human GBMs. Cells within the GBM ecosystem were shown to display GBM-specific transcriptomic features, indicating active TME crosstalk in PDOX models. Tumor-associated microglia/macrophages were shown to be heterogeneous and display the most prominent transcriptomic adaptations following crosstalk with GBM cells. The myeloid cells in PDOXs and human GBM displayed a microglia-derived TAMs signature. Notably, GBM-educated microglia display immunologic features of migration, phagocytosis, and antigen presentation that indicates the functional role of microglia in the GBM TME. Taking advantage of a cohort of longitudinal PDOXs and treated PDOX models, I demonstrated the utility of PDOX models in elucidating longitudinal changes in GBM. We show that temozolomide treatment leads to transcriptomic adaptation of not only the GBM tumor cells but also adjacent TME components. Overall, this work further highlights the importance and the clinical relevance of PDOX models for the testing of novel therapeutics including immunotherapies targeting certain tumor TME components in GBM.