Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Nearly all the cells of an organism share the same DNA sequence or genome, and yet they show different phenotypes and carry out different functions. This diversity is made possible by a verity of molecular modifications acting on the DNA sequence that collectively define the cell's epigenome. Failure in the proper regulation of epigenome can lead to abnormal activation or inhibition of vital signaling pathways, which contributes to the initiation and progression of multiple diseases, including cancer. In this thesis, I used a combination of in silico, in vitro, and in vivo techniques to explore the roles of epigenetic dysregulation in tumor progression and metastasis. In the first part, I performed a systematic and unbiased investigation of cancer-associated DNA methylation and gene expression changes across human cancers. I designed a novel algorithmic approach (RESET) to identify aberrant DNA methylation and associated gene expression changes across >6,000 human tumors. We identified a DNA Methylation Instability (DMI) phenotype in tumors acquiring numerous hyper and hypomethylated transcription start sites, which is associated with mutations of chromatin remodeling factors and Wnt-signaling. We showed that silenced genes coalesced in specific pathways including apoptosis, transcriptional regulation, and cell metabolism. The majority of the enhanced genes belonged to cancer-germline antigens (CG), whose expression correlated with response to anti-PD-1 in melanoma patients. Finally, we demonstrated the potential of RESET to explore aberrant DNA methylation in pediatric tumors; pediatric Wilms tumors with diffuse anaplasia exhibit DMI, and specific silencing events predict a worse outcome in cases with favorable histology. These results established a new approach to explore pan-cancer epigenetic modifications and provided a resource of candidate oncogenic events for future functional and therapeutic studies. In the second part, I studied the dynamics of cancer progression and metastasis in Pancreatic Neuroendocrine Tumors (PanNET). This rare tumor is the second most common form of pancreatic cancer. Two principal subtypes of PanNET has been identified: insulinomas (IT) and metastasis-like primaries (MLP), corresponding to the low and high grade of human PanNETs, respectively. From the mouse model of PanNET (RT2), we profiled the single-cell, bulk mRNA and miRNA transcriptomes, and the proteomes of primary and metastasis specimens. We demonstrated that the tumor progression from IT to MLP follows the reverse embryonic and postnatal developmental path. Transcriptomic data from human patients showed that aggressive human PanNETs also follow the same reverse developmental trajectory of dedifferentiation. We established one possible mechanism underlying IT-to-MLP transition. Over-expression of the miR-181cd cluster in IT-like cancer cell-lines resulted in acquisition of the MLP phenotypes. miR-181cd mediated its effect through suppressing the expression of Meis2 and, indirectly, inducing the expression of Hmgb3 and Mycn transcription factors. Inhibiting the expression of Hmgb3 in MLP-like cancer cell-lines resulted in a significant growth decrease both in vitro and in vivo, demonstrating the importance of Hmgb3 in the maintenance of MLP-like cell state. These data presented dedifferentiation as a mechanism by which malignant neuroendocrine cancer cells acquire progenitor-like features, enabling them to become more aggressive and metastatic.
, ,