Glioblastoma (GBM) is the most fatal and aggressive brain tumor in adults, and with an average prognosis of 15 months, new therapeutic avenues are desperately needed. A major contributing factor to GBM malignancy is the maintenance of a GBM stem cell (GBMSC) population within a tumor propagating perivascular microenvironment, though the specific interactions mediating vascular support of GBMSCs are not well understood. We thus developed a comprehensive interactome, outlining all putative vascular endothelial cell (ECs) ligand-GBMSC receptor interactions as identified through whole transcriptome profiling of ECs and GBMSCs isolated from the same freshly resected primary patient GBM samples. We revealed a prominent role for the perivascular niche (PVN) interactions to be in promoting tumor cell migration and invasion. From our interactome, the function of the EC-secreted Integrin Binding Sialoprotein (IBSP) angiocrine in promoting Proneural GBMSC migration, proliferation, and a transition to a more aggressive Mesenchymal phenotype was specifically demonstrated. Our analyses are a first to report on such a detailed interrogation of the mechanisms of GBMSC dependence on the PVN, and provide a large database to fuel further research in this area.

Glioblastoma (GBM) remains the most common and lethal primary brain tumor in adults despite concerted efforts to establish more effective treatments. The oncogenic-associated alterations that make GBM cells metabolically distinct from surrounding tissue also represent prime targets for the development of novel therapies. Due to the interconnectivity of signaling networks and the overall heterogeneity of the disease, identifying key metabolic pathways that drive neoplastic pathogenesis is essential to establishing more effective therapeutic strategies. In this dissertation, we performed expression analysis and unbiased metabolomics to better characterize metabolic differences between a cohort of patient-derived isocitrate dehydrogenase 1 (IDH1) mutant and wildtype gliomasphere cultures. This analysis revealed clear, cell type-specific differences in glucose metabolism, nucleotide synthesis utilization, and DNA repair capacity following radiation that could be exploited for therapy. Furthermore, we investigated the ability of GBM cells to oxidize fatty acids (FAO) and ketone bodies to support tumor growth, while also interrogating the effectiveness of the ketogenic diet (KD) as an adjuvant therapy for GBM. We discovered extensive FAO utilization throughout the GBM metabolome, identified CPT1A as a potential therapeutic target, and determined that the KD can have adverse effects on tumor growth.

Glioma is the most common primary brain malignancy and remains a challenging and intractable disease. Two major barriers inhibiting progress in therapeutic development are the unclear clinical implications of molecular differences across patients and inevitable tumor recurrence likely due to the difficulty in removing the entire malignant compartment during the initial resection surgery. Here, I present a strategy to begin combatting these challenges by leveraging flexible and scalable sequencing analysis pipelines to deconvolve complex glioblastoma cellularity and discover functional drivers of disease progression and recurrence. I first describe an approach that aims to better characterize therapeutically relevant patterns of intertumoral glioblastoma heterogeneity by classifying tumors based on functional pathway enrichment rather than gene expression alone, which identified E2F1 as a key proliferation driver specifically in E2F1-activated tumors. We also found a biological parallel of one of our candidate gene lists, as it correlated significantly with markers of endothelial, mural, and mesenchymal-like glioma cells. These findings suggest that pathway enrichment is a more functionally relevant grouping metric and that we can use this approach to identify therapeutic targets in specific patient subsets. I then describe an approach aimed at addressing the challenge of recurrence by characterizing the cellular and molecular differences between the core and edge regions of recurrent glioma at single cell resolution. Using MRI-guided biopsy target selection and a multifaceted snRNA-seq analysis pipeline, we thoroughly profiled both malignant and non-malignant cell types in the edge region and identified a functional “state switching” event in mesenchymal-like malignant cells, with vascular remodeling consistently enriched in the tumor core and proliferation consistently upregulated in the tumor edge. Robust sequencing analysis pipelines thus represent an effective approach to identifying key functional drivers in multiple contexts and demonstrate the potential to identify targetable candidate cells and pathways in glioma.

There is an increasing recognition of the important role cancer predisposition syndromes play in the development of oncological diseases. Consequently, there's an urgent need to identify and understand these syndromes to uncover underlying disease mechanisms. Among these syndromes are Rhabdoid Tumor Predisposition Syndrome Type 1 (RTPS-1) and Neurofibromatosis Type 1 (NF1). RTPS-1 is marked by mutations in the SMARCB1 gene, whereas NF1 is linked to mutations in the NF1 gene. Given their rarity, there's a critical need for novel therapeutic approaches and preclinical models to accelerate the translational research pipeline. Previously, our team developed genetically engineered animal models (GEMM) for both RTPS-1 and NF1. This dissertation aimed to explore the mechanisms of rhabdoid tumorigenesis by pinpointing specific biomarkers and drug targets within the RTPS-1 GEMM. Additionally, we endeavored to establish a large animal porcine model for RTPS-1 as a proof-of- concept. We then investigated Cold Atmospheric Plasma (CAP) as a novel therapeutic strategy to target NF1-associated peripheral nerve sheath tumors. Our results showcased the ability of CAP to impact cellular oxidative stress, paving the way for its potential application in treating tumors associated with cancer predisposition syndromes.

Glioblastoma (GBM) is the most common and malignant primary brain tumor. Despite maximal therapy prognosis remains poor. In the preceding decade, histological methods of categorization have begun to give way to molecular and genetic methods. Most prominently, the IDH1 mutation, when found in glioblastoma correlates with a better prognosis than those glioblastomas in which the mutation is not found. It is not known why this correlation exists. The mutant IDH1 enzyme is able to reduce alpha-ketoglutarate to create extremely high levels of 2-hydroxyglutarate but the tumorigenic significance of this reaction is unknown.

In this dissertation we use an extensive collection of IDH1mutant and IDH1wildtype patient-derived gliomaspheres to dissect genetic and metabolic differences that distinguish the IDH1mutant subgroup. We found that the IDH1mutant gliomas were readily identifiable by their slower growth and distinctive expression profile. However, pharmacologic and genetic inhibition of the IDH1mutant enzyme did not change this signature in fact it did not seem to have any effect on these cells other than to decrease the levels of 2-hydroxyglutarate.

The expression profile of IDH1mutant cells showed a large set of down-regulated genes in comparison to IDH1wildtype cells. One of the few genes to be up-regulated was Olig2 which upon further study we revealed to be essential for growth and a regulator of other important genes such as Tet2. Further, Olig2 expression may be targetable by the histone deacetylase inhibitor (HDACi) class of drugs in particular valproic acid.

When we subjected the expression profiles of IDH1mutant glioma cells to metabolic analysis we found that IDH1mutant cells were enriched for pathways involved in DNA repair following radiation while IDH1wildtype glioma cells were enriched for pathways involved in de novo DNA synthesis. Following these results we confirmed that IDH1mutant cells do show better recovery following radiation and IDH1wildtype cells show a greater utilization and dependence on the de novo pathway for DNA synthesis.

Historically, glioblastoma has been defined based on histologic criteria and while two glioblastoma samples may contain two mutually exclusive lists of mutations they would be considered minor variants within the larger fairly homogenous class. It is now becoming clear constitute a distinct group of brain tumors with different characteristics, prognosis and response to therapy and thus need to studied separately from IDH1wildtype glioblastomas.

Glioblastoma (GBM) is one of the most lethal cancers and new therapies are urgently needed. One key limitation in the development of therapeutics is the availability of appropriate in vitro models in which key molecular pathways can be identified and potential novel therapies can be tested. One model of GBM is through the growth of “tumorspheres”, patient derived GBM cultures grown in neurosphere conditions that generally grow as non-adherent spheres of cells. We assessed gene expression profiling, in vitro phenotypes, and patient survival from a large cohort of high-grade glioma tumorsphere cultures. Through interrogation of the tumorsphere model system, we discovered molecular determinants of proliferation (PTGR1, EFEMP2, LGALS8), effective combinatorial treatment targeting ERK and mTOR, and a mechanism of resistance to chronic mTOR inhibition through GSK3B and MAP1B.

Hematopoietic stem cells (HSCs) are capable of self-renewing to maintain the stem cell pool as well as differentiating into different mature blood cells to replenish the blood system. Genotoxic stress, such as chemotherapy and radiation, could induce DNA damage in the HSCs, increasing the risk of malignant transformation and decrease the normal function of HSCs. Therapies to promote DNA repair in HSCs after exposure to genotoxic stress remains not well developed. This dissertation reports that the epidermal growth factor receptor (EGFR) signaling promotes DNA repair in HSCs through activation of the non-homologous end-joining (NHEJ) pathway and regulates HSCs regeneration. Data in this dissertation demonstrates that epidermal growth factor (EGF) treatment reduces DNA damage in HSCs after radiation and chemotherapy. EGFR signaling preferentially enhances the activity of the NHEJ pathway, as indicated by NHEJ specific molecules such as DNA-dependent protein kinase, catalytic subunit (DNA-PKcs), Artemis, and Ku70. Mechanistically, EGF binds and activates EGFR, which subsequently activates Akt, further leading to the activation of DNA-PKcs. Pharmacological inhibition of Akt and DNA-PKcs confirmed the EGFR/Akt/DNA-PKcs pathway for DNA repair in HSCs in vivo. Systemic administration of EGF accelerated the hematopoietic recovery of irradiated or chemotherapy-treated mice without affecting the relapse of acute myeloid leukemia. Conditional suppression of EGFR in the hematopoietic stem and progenitor cells (HSPCs) impaired DNA repair and functional recovery, underlining the necessity of EGFR signaling in DNA repair in HSCs. Moreover, EGF treatment accelerated the recovery of irradiated human bone marrow HSCs shown by immunophenotyping in vitro and multilineage reconstitution in vivo. EGF treated human HSPCs also presented enhanced DNA repair. Whole-genome sequencing of HSPCs from irradiated EGF-treated mice revealed no significant difference in the coding regions in terms of mutation rate compared to irradiated control mice, despite increased intergenic copy number variant mutations. RNA sequencing of HSPCs from irradiated EGF-treated mice displayed no significant alterations of the transcription of leukemogenesis related genes. This thesis project uncovered the EGFR/Akt/DNA-PKcs pathway for NHEJ DNA repair in HSCs and explored the therapeutic potential of EGF to promote human HSCs regeneration.