In 2009, Anjana Rao and her team from Harvard Medical School characterized an enzyme responsible for producing a modified DNA base, bringing the total number of different building blocks of DNA to six, the standard four bases, methylcytosine and now hydroxymethylcytosine (hmC).

Methylcytosine has been tightly linked to transcriptionally silent chromatin. It is removed in the course of development, and a longstanding question in the field has been whether active demethylation occurs on DNA. The enzyme Rao's team discovered, a hydroxylase of the TET family, converts methylcytosine to hmC and may be responsible for the first step in demethylation. But far from being a byproduct in an enzymatic cascade, hmC is receiving more and more attention for a role in its own right. It is abundant in embryonic stem cells, and recently a link between hematopoietic cancers and low levels of hmC was discovered.

The difficult question is, how to accurately profile the levels of hmC in a cell. In the last two years antibodies to hmC have been developed, which allow specific pull-downs, but the downside is that these antibodies are dependent on the density of hmC and do not work efficiently in genomic regions with only sparse hmC occurrence. In a recent Nature paper, Rao and colleagues introduced two new methods for hmC mapping (Pastor et al., 2011).

Suneet Agarwal, a former graduate student in the Rao group, developed an early idea for the first approach, glucosylation, periodate oxidation, biotinylation (GLIB). “I became very interested in the idea that the hydroxyl could be exploited to covalently tag the base,” he recalls. Using T4 phage glucosyltransferases the team added glucose and, following a method developed by William Pastor, a current graduate student in the Rao lab, oxidized the sugar and then created a reactive group to which biotin was added. Pulling down the biotin-tagged hmC with streptavidin beads allowed the identification of regions bearing hmC.

Conceptually, this approach is similar to a strategy recently developed by Song et al., which uses a synthetic glucose analog to label hmC (Song et al., 2011); the advantage of GLIB is in the fact that no custom-made reagents are needed and that it allows for a very stringent specificity control.

In parallel to the development of GLIB, Yun Nancy Huang, a postdoc in the Rao group, worked on the second approach, cytosine 5-methylenesulphonate (CMS) pulldown. Upon reaction of hmC with sodium bisulphate, CMS is created and can be isolated with specific antibodies. “Although the CMS pulldown still has some density dependence,” says Huang, “a substantial amount of low hmC-containing fragments can still be pulled down with very low background. GLIB and CMS complement each other.”

The team tested both approaches on mouse embryonic stem cells. They compared regions enriched for either methylcytosine or hmC and found overlap within transcribed regions. But only hmC was enriched in transcriptional start sites, 5′ untranslated regions and enhancers. Notably, hmC occurred predominantly at genes with bivalent chromatin marks, genes that are silent in embryonic stem cells but poised for activation during differentiation.

More tools to profile hmC will be invaluable for better characterization of this modification in the context of different cell types and developmental states, and this will help address the big outstanding questions in this field. The specific role of hmC during differentiation and malignant transformation still remains to be discovered. And the question of whether active demethylation actually occurs or whether hmC blocks re-methylation also still awaits answering.

Though all enrichment methods developed so far are very useful in providing a genome-wide picture of where regions enriched in hmC are, Agarwal calls it the “500-foot view”; the 'ground-level view' is still needed: which base is an hmC, and what are its neighbors? Such single-base resolution will require approaches that directly detect the modification on a stretch of DNA.