Summary Glioblastoma multiforme (GBM) are the most aggressive type of gliomas and lead to poor prognosis. Despite modern multimodal therapy, relapse recurs in up to 90% of all cases and the survival of patients is around one year in average. GBM are highly heterogenous at the genomic, histopathological and disease progression level, which represents a great challenge for developing treatments. The aggressiveness and relapsing behavior of GBM partly relies on the presence of a population of cells with stem cell properties that are resistant to therapies and have the capacity to replenish the tumor bulk. These cancer stem cells (CSCs) share features with neural stem cells (NSCs) such as similar transcriptional regulatory pathways. In this study, we focused on the helix-loop-helix (i.e. ID proteins) and basic helix-loop-helix transcription factor network (i.e. E proteins and Olig2) to investigate its potential as target for therapeutic approaches. This transcriptional network regulates proliferation and differentiation of stem cells and is also involved in maintenance of tumor homeostasis. In this thesis, the aim was to disrupt the (b)HLH network in its ensemble with endogenous E proteins in order to induce anti-glioma effects.
The first chapter describes how overexpression of a dominant-negative E protein (dnE47) lacking its nuclear localization signal in human glioblastoma-derived cells led to sequestration of HLH and bHLH proteins in the cytoplasm. DnE47 overexpression induced apoptosis in adherent immortalized tumor cell lines and down-regulation of pro-proliferative as well as anti-apoptotic genes. Similar effects were detected in tumor-initiating cells derived from human GBM biopsies. In these later cells, overexpression of dnE47 impaired the sphere formation capacities of tumor-initiating cells and delayed the onset of clinical symptoms in a murine xenograft model. Chapter two describes the development and establishment of tools and methods necessary for the study presented in Chapter one. We engineered an inducible lentiviral vector and established the optimal conditions for its application in patient- derived tumor-initiating cells. We also present a more detailed histology of tumors induced in immunosuppressed mice as well as possible novel analytical approaches for characterizing the number and distribution of grafted cells within the mouse brain. In chapter three, we finally tested a novel xenograft model for assessing the ability of glioma-initiating cells to form tumors. We transplanted tumor-initiating cells in Drosophila melanogaster flies and monitored the survival, migration and tumor- formation of the xenografts.
To conclude, in this thesis I demonstrate the feasibility of using properties of endogenous proteins in order to disrupt a transcriptional network globally in the context of human glioblastoma and to achieve anti-glioma effects.