Generalized Loss of Stability of Epigenetic Domains Across Cancer Types
Kasper Daniel Hansen, Johns Hopkins
The basis of phentypic variation between species is frequently thought of as genotype -based. However, if you consider two organs in the same individual, the tissues are clearly sharing genomes, but are phenotpyically different.
Cancer and DNA methylation. DNA methylation was first epigenetic changed in cancer. Focus, recently, has been on hypermethylation of CpG islands. We now see more on global hypomethylation, hypomethylation of selected genes (oncogenes)
CpG island Shores. Many changes are not in CpG islands, but in regions bordering CpG islands, termed CpG shores. (Irizarry et al, 2009)
Increased methylation variation across all cancers. The same regions that distinguish cancers from normals also distinguishes normal tissue types. [sorry, spilling water glass behind me distracted me for a minute.. missed a bit.]
Study design. Whole genome bisulfite sequencing on tumour and matched normals. 3 colon cancers, and normal mucosa and 2 adenomas.
Wrote a custom aligner, Merman, to process bisulfite sequencing data.
Global levels of methylation – it’s not particularly interesting, but is representative. some clustering between tumour types, normals.
Loss of methylation boundaries: There are sharp methylation boundaries in normals, but the boundaries are lost in cancers. in cancers, methylation appears constant across shores and islands in graph shown.
Boundary shifts also occur and novel hypomethylation exists as well.
Capture bisulfite – 40,000 capture regions. Great data, consistent across labs doing the work, across samples, etc.
Large blocks of hypomethyation observed in cancers. increased variability in cancer samples compared to normals. Consistent boundaries, cover more than 1/2 of genome. Related to structural conformation of the DNA in the nucleus. Some cell type specificity as well. [a lot is data is being presented here in graphical form, so if the above notes are confused, it’s because I’m having a bit of trouble following the flow.]
what predicts hypomethylation? Silenced genes are correleated?
Blocks are enriched for hyper-variables [hypervariable methylation status?]. Some of the genes are associated with tumour progression.
[And a very abrupt end to the talk! That’s it.]