Methylation of histone residues can have various consequences for genetic regulation, such as flagging transcriptional repression or activation. While there was some evidence that histones could also become unmethylated, no one knew what was responsible. Then in 2004, Yang Shi at Harvard Medical School and his colleagues identified the first histone demethylase, a protein called LSD1 that removes one or two methyl groups from histone 3 lysine 4.
This discovery left a number of demethylating enzymes still at large, including those that demethylate lysine 9 and 36, and the enzymes that act upon trimethylated residues. Yi Zhang at the University of North Carolina was determined to find them, and in the first Hot Paper highlighted here his team identified a protein domain...
A pair of papers that appeared in Nature in 2002 showed that the bacterial AlkB protein could demethylate DNA. The biochemistry struck Zhang: "The reaction generally is the same as histone demethylation." In both cases the enzyme removes a methyl group from a lysine. So Zhang's lab went about searching for the protein that would do the same to a histone substrate, and they purified a protein that contained a jumanji (Jmj) domain.
The enzyme, called JHDM1, acts upon dimethylated lysine 36. The link between jumanji domains and demethylation "had been predicted," says David Bentley at the University of Colorado Health Sciences Center, and Zhang "did a terrific job nailing that connection."
Though Zhang was the first to publish on the jumanji demethylases, Harvard's Shi and others were already hot on the jumanji trail using genetic approaches. Within months of Zhang's paper, Shi's group published evidence of a Jmj demethylase that removes one methyl group from trimethylated lysine 9 and 36.
In 2007, Bentley's group identified another family of jumanji-containing demethylases, JARID1, which demethylates lysine 4.
Finding their purpose
When Julie Secombe, a postdoc at the Fred Hutchinson Cancer Research Center, read Zhang's 2006 Hot Paper in which he first identified the Jmj demethylase family, she was thrilled. She had been working on a protein called Lid that contained a jumanji domain. Lid is required for development and myc-induced cell growth. "We knew all this about Lid, but we were lacking completely [any information about] what Lid did," says Secombe. Zhang's discovery led her to ask whether Lid, because of its Jmj domain, is also a demethylase, and last year she found that it is indeed part of the JARID group of demethylases.
More hints on the functions of Jmj demethylases have emerged since Zhang's first Hot Paper. Colorado's Bentley points out that JARID1, for example, is overexpressed in breast cancers. In 2007, Shi found that JARID1C is implicated in X-linked mental retardation, and later that year Zhang found that JHDM2A is essential for spermatogenesis.
While the phenotype of some demethylase defects are known, the role they play in genetic regulation is still unclear. "Linking histone demethylation with some specific known step in the mechanics of gene regulation - that connection has not been made," Bentley says.
"My guess is these demethylases are going to play some role in fine-tuning methylated states," Harvard's Shi says. For example, he says, some demethylases have a preference for removing one methyl group from trimethylated lysine, and other demethylases may remove two. He continues to study Jmj demethylases to see how their structure influences their specificity.
Zhang says he's interested in looking in vivo to uncover the specific instances in which demethylases act. "Although methylation is reversible," he says, "it's still very difficult to remove" methyl groups, and the kinetics are not robust. "Only in specific situations does this happen. My bet is that it most likely will be in cell fate, when you need to reverse epigenetic markers." Zhang's in vivo studies are aimed toward discovering the roles of demethylases in differentiation. "There's a lot of biology still to be discovered," says Shi.