| Glioblastomas (GBM), the most malignant primary brain tumors, are among the most biologically aggressive and therapeutically challenging cancers. Recently, the Cancer Genome Atlas (TCGA) project has compiled multidimensional analyses of the genome space from human GBM tumor samples to serve as an unprecedented framework for the molecular understanding of this deadly disease. TCGA GBM datasets have catalogued well-recognized genomic alterations, including amplification/mutation of EGFR (EGFRvIII) along with deletion/mutation/silencing of p16INK4a and PTEN tumor suppressors. The datasets also exhibit many more recurrent genomic events, suggesting the existence of additional novel genetic loci in GBM pathogenesis.;My lab previously generated an EIP (for EgfrvIII, Ink4a/Arf loss, and Pten deletion) GBM mouse model that harbors the most common human GBM aberrations but develops malignant glioma only at low penetrance. This dissertation depicts a comprehensive utilization of TCGA datasets to identify and characterize novel cancer genes that cooperate with the EIP background in gliomagenesis.;In this dissertation, I performed an in vivo tumor suppressor screen based on the most recurrent and focal deletions in the TCGA GBM aCGH (array Comparative Genomic Hybridization) dataset and identified novel tumor suppressor genes that cooperate synergistically with the EIP genetic lesions. Then, I identified Quaking (QKI) as a frequently and focally deleted gene in GBM from the TCGA datasets and characterized the mechanism of action of QKI as a novel GBM tumor suppressor gene. Specifically, I elucidated that QKI is transcriptionally activated by p53, and QKI associates with and stabilizes miR-20a, which in turn inhibits TGFbetaR2 to block gliomagenesis. Linkages of these pathway components are further supported by their relative expressions and genome status across TCGA GBM specimens. Lastly, I described the generation of a mouse model with loss of p18INK4c, a GBM tumor suppressor with a compensatory function to p16INK4a, to enhance tumorigenic potential of the EIP mouse model.;Taken together, this dissertation takes full advantage of the multidimensionality of TCGA datasets, which serves as a roadmap for the discovery, characterization, and mouse modeling of novel cancer genes. We hope these efforts will ultimately lead to novel drug development and preclinical systems for treatment of this devastating disease. |
