DNA methylation is generally thought of as a mark of inactive chromatin. The recent finding that the active copy of the human X chromosome is more highly methylated overall than the inactive copy therefore comes as something of a surprise, prompting a rethink of how we view DNA methylation patterns.

One well-known characteristic of X-chromosome epigenetics is that CpG islands at gene promoters are methylated on the inactive copy, contributing to gene silencing, and are unmethylated on the active X chromosome, allowing transcription. However, previous studies have hinted that, on a chromosome-wide scale, the active X chromosome might actually be more highly methylated.

To explore this possibility, Asaf Hellman and Andrew Chess developed a method for profiling the methylation status of sites across both the active and inactive X chromosomes (Xa and Xi). They made use of an array that allows 500,000 human single-nucleotide polymorphisms (SNPs) to be genotyped. By carrying out restriction-enzyme digestions that are sensitive to DNA methylation, followed by PCR and hybridization to the SNP array, the authors were able to determine allele-specific DNA methylation patterns across the X chromosome. Active and inactive copies were distinguished by gene-expression and DNA-replication analyses, and DNA methylation patterns were compared in clonal cell lines, derived from a given individual, that had inactivated either the maternal or the paternal copy of the X chromosome.

The results showed a strikingly higher level of allele-specific DNA methylation on the Xa than on its inactive counterpart, at a ratio of about 2.4:1. Most of this methylation occurs within the bodies of genes (the transcribed regions), rather than in upstream promoters or intergenic areas. The authors found no biases that give a clue to the function of this methylation; for example, methylation was not especially common at tissue-specific genes or repetitive elements.

Does this relative hypermethylation on the Xa result from active methylation on this copy, or from demethylation of the Xi? To answer this question, Hellman and Chess looked at a human embryonic stem cell line, which is presumed to represent a stage in development before X inactivation. Few of the Xa-specific methylation sites were methylated monoallelically in these cells, indicating that biallelic methylation is the initial state, with later demethylation of the Xi at gene bodies.

The authors suggest a model for this differential DNA methylation pattern on the two copies of the X chromosome. They propose that inactive, untranscribed regions, such as the Xi and intergenic regions, are more likely to undergo loss of methylation, resulting in higher methylation levels remaining at gene bodies on the Xa. Whether this is the case or not, this study alters our views of DNA methylation on the X chromosome, revealing marked differences outside the promoter regions that have been the focus of previous studies and setting the scene for other reassessments of DNA methylation patterns.