Principles and challenges of genome-wide DNA methylation analysis

Key Points Methylation of cytosine residues at the carbon 5 position occurs naturally in many bacteria, archaea and eukaryotic species, in which it has various roles in protecting the genome from invading genomic parasites or in controlling the expression potential of regions of the genome. DNA methylation is established after DNA synthesis by dedicated enzymes with specific target sequence recognition sites. The uneven distribution of target sites and sample heterogeneity can result in complex DNA methylation patterns. The genomic distribution of DNA methylation encodes important biological information. Hence, techniques for comprehensively describing DNA methylation patterns have been developed. Many standard molecular biology techniques, such as cloning and PCR, erase DNA methylation information, and hybridization does not distinguish between methylated and unmethylated cytosines. There are three different initial treatments of DNA that can be used to reveal DNA methylation: endonuclease digestion, affinity enrichment and bisulphite conversion. The implementation of array hybridization techniques greatly facilitated genome-scale analysis of DNA methylation. Endonuclease-treated or affinity-enriched DNA methods are particularly well suited for array hybridization, whereas bisulphite conversion techniques are not. Next-generation sequencing allows for whole-genome single-base-pair resolution characterization of DNA methylation patterns, particularly as applied to bisulphite-converted DNA. No single technique excels in all aspects. Sample number and characteristics, as well as the desired accuracy, coverage and resolution, influence the choice of technique. DNA methylation is usually measured on a β-distributed absolute scale from 0 to 1, or 0 to 100%, rather than on an infinite scale of log ratios. The unique data distribution characteristics of DNA methylation will require the development of dedicated bioinformatics and computational tools. Single-molecule and nanopore sequencing approaches are likely to usher in the next revolution in high-throughput DNA methylation analysis.