The analysis of epigenomic evolution

Abstract

While the investigation of the epigenome becomes increasingly important, still little is known about the long-term evolution of epigenetic marks and systematic investigation strategies are still withstanding. Here, we systematically demonstrate the transfer of classic phylogenetic methods such as maximum likelihood based on substitution models, parsimony, and distance-based to interval-scaled epigenetic data (available at Github). Using a great apes blood data set, we demonstrate that DNA methylation is evolutionarily conserved at the level of individual CpGs in promotors, enhancers and genic regions. Our analysis also reveals that this epigenomic conservation is significantly correlated with its transcription factor binding density. Binding sites for transcription factors involved in neuron differentiation and components of AP-1 evolve at a significantly higher rate at methylation than at nucleotide level. Moreover, our models suggest an accelerated epigenomic evolution at binding sites of BRCA1, CBX2, and factors of the polycomb repressor 2 complex in humans. For most genomic regions, the methylation-based reconstruction of phylogenetic trees is at par with sequence-based reconstruction. Most strikingly, phylogenetic reconstruction using methylation rates in enhancer regions was ineffective independently of the chosen model. We identify a set of phylogenetically uninformative CpG sites enriching in enhancers controlling immune-related genes.