Methylation of CpG Sites
Cecilia Lövkvist, Niels Bohr Institute, University of Copenhagen, Denmark.
DNA-methylation is a well-studied epigenetic phenomenon where cytosine is modified in the dinucleotide sequence CpG by addition of a methyl group. The human genome contains approximately 28 million CpG sites and a majority of them are methylated. Clusters of CpG sites that are overlapping promoter regions have been observed to be correlated with gene expression. In the literature, high density clusters of CpG sites are termed CpG-islands.
We propose to expand the definition of CpG-islands by instead using the spatial distribution of CpG sites in terms of distances and clustering. By analysing epigenomes from different experiments, we find that CpGs are not randomly distributed in the genome. CpG sites are instead organized in clusters where short (long) distances are followed by short (long) distances. Furthermore, we find that their spatial organization, e.g. the size of a given cluster, affects the distribution and bimodality of methylation, i.e. induces a switching between methylated states.
In this work, we propose a dynamic model consistent with the data which incorporates the existence of collaboration between CpGs. The model describes clusters of closely spaced CpG sites and their interaction with surrounding CpG sites and considers the spatial or- ganization of the sites.
From the data and our model we conclude that the length of CpG-islands affects the methylation distribution and we show that a crucial building block is collaborative distance dependency between the CpG sites.