Low levels of methylation are targeted to aberrantly methylated CGIs in human colorectal cancer cells

Abstract

Disruption of DNA methylation patterns is a primary hallmark of cancer. One main disruption is CpG islands (CGIs) aberrant hypermethylation which is associated with transcriptional repression of tumour suppressor genes such as BRCA1, MLH1, and CDKN2A (p16/ARF). However, the mechanism(s) underpinning this process are still unknown. One hypothesis is that high de novo methylation activity is targeted to CGIs that get aberrantly methylated in cancer. To investigate this, I used human colorectal cancer cell line (HCT116) as a model. Firstly, I have ectopically integrated representative aberrantly methylated CGIs randomly into the genome and assessed their methylation at ectopic loci. I show that integrated CGIs rarely gained methylation at ectopic locations. This suggested that the sequence doesn’t program CGIs aberrant methylation at ectopic loci in HCT116 and that low de novo methylation is targeted to aberrantly methylated CGIs. Next, I wanted to confirm this result by integrating CGIs at a chosen genomic loci in order to exclude any effects from the CGI position on its methylation status. This was performed by recombinase mediated cassette exchange (RMCE). Successful targeting of a RMCE cassette to the desired chromosomal locus was achieved. However, isolating clonal cell lines with integrated CGIs was technically infeasible due to difficulty in using the thymidine kinase selection in HCT116. In order to investigate which CGIs have higher de novo methylation in HCT116 on a genome wide level, I restored DNMT3B expression in DNMTA1/DNMT3B double knockout (DKO) of HCT116. In agreement with previous reports, DNMT3B showed higher de novo methylation activity at CGIs overlapping H3K36me3. Most importantly, aberrantly methylated CGIs showed ~ 78.5% less methylation gain compared to normally methylated CGIs. To confirm results from this experiment, I treated HCT116 with 5-Aza-2′-deoxycytidine and then allowed cells to recover methylation. The rate of methylation recovery of CGIs was assessed. This showed that aberrantly methylated CGIs recover slower than normally methylated CGIs, confirming that lower de novo methylation is targeted to aberrantly methylated CGIs compared to normally methylated ones. The work carried out during the course of this thesis provides novel insights into the process of de novo methylating CGIs in cancer on a genome wide level. It suggests that low levels, rather than high levels, of methylation are targeted to aberrantly methylated CGIs in colorectal tumours.