Epigenetic silencing of genes via DNA promoter methylation is seen in multiple critical genes and pathways associated with ovarian cancer. Epimutations occur more frequently than genetic mutations and they can be detected readily in surrogate tissues such as plasma and ascites (Gifford, Paul et al. 2004). This means that methylation profiling can be utilised to identify key pathways which can then be investigated in more depth, by functional analysis of key genes.
Aberrant DNA methylation can potentially be used to diagnose cancer earlier or to identify groups of patients who may benefit from epigenetic and other targeted therapies(Gifford, Paul et al. 2004). In addition, for genes that acquire methylation during the course of chemotherapy it can highlight pathways and key genes that are important in ovarian cancer pathogenesis (Dai, Teodoridis et al. 2010). Epigenetics is being increasingly implicated in both embryonic and cancer stem cell initiation and maintenance and therefore study of DNA methylation may help to further unravel the answers to cancer initiation and maintenance (Rizzo, Hersey et al. 2011).
It appears that small changes in multiple genes may be of more functional consequence than large changes in one gene and this presents a real challenge when interpreting results from high throughput analyses. In addition, differentiating mutations that are fundamental – so called „driver (epi) mutations‟ from ones that are observed but not critical – so called „passenger mutations‟ will be the key to using these techniques to increase our knowledge. The field of epigenetics is becoming increasingly complex and the interplay between DNA promoter methylation, histone modifications and micro- and non-coding RNAs is not fully understood.
1.4.1 Thesis hypotheses to be tested
The aim of the experiments outlined in this thesis was to investigate whether increased methylation of genes in drug resistant ovarian tumour cell lines could identify novel biomarkers of acquired drug resistance in EOC. If this was the case one would predict that these candidates would show increased methylation in matched tumour pairs following chemotherapy if in vitro resistance correlates with in vivo resistance. In addition it could be hypothesised that changes in methylation would cluster to particular key pathways which are required for ovarian cancer initiation and progression. In order for methylation to have a phenotypic effect it would be predicted that an increase in methylation would correlate with a decrease in expression and that this could be reversed using demethylating agents. In order to address these questions bioinformatic approaches were combined with microarray analysis of methylation by differential methylation hybridistion (DMH) in order to identify differentially methylated genes in human ovarian cell line models of acquired drug resistance. Any genes identified in this manner were independently validated in
the laboratory using MSP or pyrosequencing of bisulphite modified DNA – in matched sensitive and resistant cell lines, primary ovarian cancer tumour pairs and matched samples from patients with ovarian cancer, pre- and post- relapse. The correlation between promoter methylation and gene expression was investigated using qRTPCR in cell lines that had been treated with or without decitabine, and in order to assess the phenotypic effect of knockdown or over-expression of key genes short interfering RNA (siRNA) and over-expression following gene reintroduction were performed respectively.
1.4.2 Outline of the aims of the experiments outlined in each
individual chapter
The aim of the experiments outlined in chapter 3 was to identify and validate novel DNA methylation markers for acquired drug resistance in ovarian cell lines and evaluate their relevance to acquired resistance in patient samples using prediction analysis for microarrays (PAM). We also aimed to validate pyrosequencing of bisulphite-modified DNA as a better technique for examining candidate loci in the laboratory, than MSP.
In Chapter 4, following a comparison of the A2780 sensitive and resistant cell lines, the loci identified using methylation linear discrimination analysis (MLDA) vs. PAM were characterised in order to assess whether this novel statistical method was useful as a means of detecting differentially methylated loci. Again loci identified biostatistically were characterised in the laboratory by both MSP and pyrosequencing.
In Chapter 5 the main aim of the experiments outlined was to assess whether loci identified by comparing in vivo derived cell lines (as opposed to the previous experiments which had all compared in vitro derived cell lines) could guide us to more clinically important genes or pathways.
The primary aim of Chapter 6 was to use functional assays to further assess the effect of over- and under-expression of two genes which had been identified in previous chapters. qRTPCR, siRNA knockdown and over expression using whole gene cloning were all utilised.
In work that was performed in parallel to these experiments, in Chapter 7 OGT customised arrays were used to hybridise both cell lines and primary tumour samples, using DMH. The preliminary results of loci identified from these experiments are discussed and in particular the loci which had been identified from the original 12k array, and characterised in earlier chapters of the thesis, were re-examined in this independent data set.