Glioblastoma (GBM), the most common primary malignant brain tumor, remains fatal and lacks effective treatment options. The influence of initiating mutations and cellular origin on GBM pathogenesis remains elusive. Previous studies utilizing genetically engineered mouse models (GEMM) have outlined some mutational and cellular requirements for GBM in different developmental contexts. However, the influence of particular mutations in differing cell types remains incompletely characterized. Groups such as The Cancer Genome Atlas (TCGA) have utilized high-throughput molecular analysis of human GBM and identified three frequently mutated core signaling pathways: the receptor tyrosine kinase (RTK), retinoblastoma (RB), and the P53 pathways. Leveraging this knowledge, we have designed conditional, inducible GEMMs to target commonly mutated signaling pathways in different cellular contexts in the adult mouse brain. In this model system, we outlined the genetic requirements of GBM initiation in glial fibrillary acidic protein (GFAP)+ astrocytes. We defined the resulting molecular heterogeneity, stochastic growth rates, and genotype-dependent survival effects. Also, we found tumor transcriptomes from our GEMM separate based on brain region, implicating the influence of regional heterogeneity on tumor subtype. Finally, we noted that tumor transcriptomes are reminiscent of purified neural cell types, suggesting that the initially mutated cell influences resulting tumor heterogeneity. GEMMs have been used to target different mutations to distinct cell populations to generate GBM. However, it remains unclear whether heterogeneity within a targeted cell population influences resulting tumor composition. Therefore, we targeted the same mutations to two different astrocyte subpopulations in adult conditional, inducible GEMMs and used genetic lineage tracing, fate mapping, and immunofluorescence to monitor transformation as well as cellular composition of the tumor microenvironment over time. We found that tumor growth rate and composition of the microenvironment varied depending on the initially transformed astrocyte population. Using GEMM, we and others have shown that terminally differentiated astrocytes can serve as a cell of origin, however the molecular mechanisms which allow for their de-differentiation to a more primitive cell type remain unclear. Using immunofluorescence, we found expression of the primitive transcription factor Sox2 in the adult murine astrocytes. To test whether Sox2 was important for astrocyte de-differentiation, we wounded the brains of adult mice with Sox2 deleted in astrocytes and observed an inappropriate wounding response by glial cell populations. Next, we engineered in a conditional Sox2 knockout into our GEMM tumor model and found tumorigenesis was unaffected. The results suggest that Sox2 could play a role in astrocyte reentry into the cell cycle under diverse pathological conditions, including traumatic injury and tumorigenesis. Overall this research explores the requirements for tumorigenesis in murine astrocytes and examines the molecular mechanisms that enables their response to pathological stimuli. The models we developed will be useful for future studies elucidating the roles of the transformed cell of origin on GBM pathogenesis, subtype specific preclinical modeling of GBM, and the role of tumor microenvironment in GBM.