Profound flanking sequence preference of Dnmt3a and Dnmt3b mammalian DNA methyltransferases shape the human epigenome

Mammalian DNA methyltransferases methylate cytosine residues within CG dinucleotides. By statistical analysis of published data of the Human Epigenome Project we have determined flanking sequences of up to +/-four base-pairs surrounding the central CG site that are characteristic of high (5'-CTTGCGCAAG-3') and low (5'-TGTTCGGTGG-3') levels of methylation in human genomic DNA. We have investigated the influence of flanking sequence on the catalytic activity of the Dnmt3a and Dnmt3b de novo DNA methyltransferases using a set of synthetic oligonucleotide substrates that covers all possible +/-1 flanks in quantitative terms. Methylation kinetics experiments revealed a >13-fold difference between the preferred (RCGY) and disfavored +/-1 flanking base-pairs (YCGR). In addition, AT-rich flanks are preferred over GC-rich ones. These experimental preferences coincide with the genomic methylation patterns. Therefore, we have expanded our experimental analysis and found a >500-fold difference in the methylation rates of the consensus sequences for high and low levels of methylation in the genome. This result demonstrates a very pronounced flanking sequence preference of Dnmt3a and Dnmt3b. It suggests that the methylation pattern of human DNA is due, in part, to the flanking sequence preferences of the de novo DNA MTases and that flanking sequence preferences could be involved in the origin of CG islands. Furthermore, similar flanking sequence preferences have been found for the stimulation of the immune system by unmethylated CGs, suggesting a co-evolution of DNA MTases and the immune system.