Generally speaking, eukaryotic cells do not discard DNA as they differentiate. Cellular differentiation therefore has to be explained as the consequence of differential gene expression. So how are genes stably yet reversibly regulated? During the past 30 years, the direct modification of DNA by methylation has been shown to have a central role in repressing gene expression and transmitting the silenced state to daughter cells.
Among the founding papers in the field were the 1975 reviews by Arthur Riggs, and by Robin Holliday and John Pugh, who discussed the literature on DNA methylases in bacteria. Their models proposed that the properties of these enzymes — in particular, their preference for hemi-methylated substrates — made them ideally suited to establish stable differentiated states in the absence of genetic mutation. They further proposed that the sequence-specific binding of these enzymes would have a gene-regulatory role. This idea was not new, but its relevance to eukaryotic transcription had been in doubt, possibly owing to the apparent lack of DNA methylation in yeast, Caenorhabditis elegans and Drosophila melanogaster. Although not correct in all details, these influential syntheses focused attention on the potential role of methylation in gene expression.
By 1980, when a seminal paper by Peter Jones and Shirley Taylor was published, more substantial evidence of a role for DNA methylation in transcriptional repression had accumulated. Jones and Taylor introduced the use of cytidine analogues to prevent the methylation of cytosine residues in DNA, and directly linked changes in the patterns of methylation with changes in the differentiated state of the treated cells. The analogue 5-azacytidine was shown to be a potent inhibitor of DNA methylation, and is now used routinely to reactivate genes that are silenced by methylation. It has also entered the clinic as a treatment for myelodysplastic syndrome.
Adrian Bird and colleagues subsequently provided further evidence linking DNA methylation and gene expression. In 1985, Bird and co-workers characterized the small fraction of the mouse genome that is frequently cleaved by a methylation-sensitive restriction enzyme. These sequences, which would come to be known as CpG islands, are CpG-rich fragments with low or undetectable levels of methylation. The available literature indicated that CpG islands were typically located near the 5' ends of genes, and the authors predicted correctly that genes might be associated with "methylation-free zones near sequences of regulatory significance."
Although all of these data pointed to a role for DNA methylation in regulating gene expression, the underlying molecular mechanisms were not discovered until the late 1990s. Perhaps the most farsighted prediction made by Riggs was that DNA methylation might affect gene expression indirectly by changing the affinity of sequence-specific DNA-binding proteins for their target sites. Definitive evidence came from the laboratories of Bird and Alan Wolffe in 1998. Each group showed that Mecp2, which had been shown to bind to methylated DNA and repress transcription, does so by recruiting a histone deacetylase complex that alters chromatin structure. Although the exact nature of this crosstalk between DNA and histone epigenetic marks (see Milestone 22) is still being worked out, the importance of DNA methylation as a stable regulator of transcriptional repression has been firmly established.
