Computational analysis of DNA repair pathways in breast cancer

Abstract

The integrity of the human genome is constantly challenged by a variety of endogenous and exogenous factors such as ultraviolet radiation and cigarette smoke. To deal with these threats, five major DNA repair pathways, which are principally defined by the type of lesions they repair, have evolved. Defects in these repair pathways predispose individuals to a wide variety of cancers, and at the same time can be therapeutically exploited to target tumours with defective DNA repair. Dysregulation of these repair pathways are also frequently observed in cancer, presenting both opportunities and challenges for cancer therapy: downregulated repair pathways sensitise tumours to DNA-damaging therapies, while upregulated repair pathways cause resistance to these therapies. The primary aim of this thesis is to obtain a comprehensive and in-depth understanding of the mechanisms and roles of these major DNA repair pathways in the context of breast cancer. This will be beneficial for predicting response to radiation and chemotherapy, and for developing novel targeted therapies in this common type of malignancy. By careful literature search and consulting a domain expert, the research presented in this thesis started with a manual curation of six DNA repair pathways, including the five major repair pathways and the Fanconi anaemia pathway that is closely associated with breast cancer susceptibility. Six comprehensive pathway figures were generated, each for one repair pathway, describing in total 195 genes and 138 reactions with direct relevance to DNA repair. Moreover, to facilitate a deep understanding of the repair mechanisms, a detailed description for each reaction was given, importantly including the literature references used for curating the reaction. This curation work enables a mechanistic understanding of how cells respond to DNA damage, and provides a solid foundation for the subsequent computational analyses. In the second study of this PhD research, I performed a personalised pathway analysis to investigate the status of homologous recombination (HR) pathway dysregulation in individual sporadic breast tumours, its association with HR repair deficiency and its impact on tumour characteristics. Specifically, using the expression values of the HR genes curated in the previous study, I calculated an HR score for each tumour that quantifies the extent of HR pathway dysregulation in that tumour. Based on that score, I observed a great diversity in HR dysregulation between and within gene expression-based breast cancer subtypes. And by comparing to two published HR-defect signatures, I found HR pathway dysregulation reflects HR repair deficiency. Furthermore, I uncovered a novel association between HR pathway dysregulation and chromosomal instability (CIN): tumours with more-dysregulated HR tend to have higher CIN. Although CIN has long been considered to be a hallmark of most solid tumours, with recent studies highlighting its importance in tumour evolution and drug resistance, the molecular basis of CIN in sporadic cancer remains poorly understood. The novel association revealed in this study implies that HR pathway dysregulation is an important determinant of CIN in sporadic breast cancer, and thus helps pinpoint the causative factors of CIN in breast and other sporadic cancers. The third study is a multi-omics data analysis that aimed to dissect the underlying mechanisms of DNA repair dysregulation in breast cancer. Specifically, I assessed the contributions of DNA copy number alteration (CNA), DNA methylation at gene promoter regions, and expression changes of transcriptional factors (TFs) to the differential expression of individual DNA repair genes in breast tumour versus normal samples. These gene-specific results were summarised at pathway level to estimate whether different DNA repair pathways are influenced in distinct manner. In particular, TFs potentially associated with each differentially expressed DNA repair gene were identified using a regularised linear regression-based statistical framework developed in this study. The results suggest that CNA and expression changes of TFs are major factors for DNA repair dysregulation in breast cancer, and that a limited number of TFs with multiple targets in various repair pathway may exert a global impact on repair dysregulation in this malignancy. This study thus provides new insights into the underlying mechanisms of DNA repair dysregulation in breast cancer. These insights improve our understanding of the molecular basis of the DNA repair biomarkers identified thus far, and have potential to inform future biomarker discovery in this common cancer type.