Colorectal cancer: the EMT transcription factor ZEB1, the interaction between myofibroblast and colorectal cancer cells and GMM analysis of cancer gene expression

Thesis

Colorectal cancer is one of the most prevalent types of cancer. Many factors are associated to the poor prognosis of colorectal cancer including genetic mutations, epigenetic changes, myofibroblast recruitment, and metastasis. Metastasis is the major cause of death in colorectal cancer patients. The most common sites are liver and lymph node metastasis with approximately 20-30% of patients with liver metastasis. Most literature discuss epithelial to mesenchymal transition (EMT) as the molecular mechanism tumor cells undergo to increase motility and invasiveness, ultimately leading to cancer metastasis.

This thesis is consisted of two major sections: computational and experimental work. The first aim of the thesis is to demonstrate our Python-implemented pipeline GMMchi developed to identify bimodal patterns in a mixture of normal and non-normal components (Chapter 3). GMMchi is further applied onto multiple large datasets to demonstrate its capability in pattern-based analysis (Chapter 4). In addition, we built a GMMchi-based scRNA-seq postprocessing pipeline efficient in separating clusters in a population of similar cell types (Chapter 7).

In the second aim of this thesis, we studied colorectal cancer metastasis in the context of EMT-like colorectal cancer cell lines and myofibroblast recruitment via co-cultured system consisting of colorectal cancer and myofibroblast cell lines. We showed a novel epigenetic regulation of ZEB1, the gene encoding an EMT transcription factor highly associated with our CRC cell lines exhibiting EMT markers, via CpG island methylation (Chapter 5). The second part of this section focused on understanding PDGF-C as an important ligand responsible for myofibroblast and CRC cell line interaction (Chapter 6).

In this thesis, we identified a novel mechanism for ZEB1 promoter methylation, revalidated myofibroblast recruitment phenotype previously observed in co-culture assays as well as developed and demonstrated our suite of tools for researchers to carry out pattern-based analysis.

Authors: Liu, T

• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: Gaussian mixture modeling; colorectal cancer; myofibroblast; pattern based analysis; single cell sequencing; EMT; big data
• Subjects: Oncology; Cancer; Colon (Anatomy)--Cancer; Computational biology