Histone Methylation Proves Reversible

It's the ultimate packing nightmare: Each of our cells must cram nearly 2 meters of genomic DNA into its nucleus, a compartment that measures a mere 6 µm across. To solve this problem, cells first wrap their DNA onto scaffolding proteins known as histones, which are then further compacted into a dense structure known as chromatin.

But because a variety of key cellular processes--including gene expression, DNA replication, and DNA repair--require access to DNA, cells must also have a way to momentarily "unpack" a given stretch of genomic DNA. One way to do so is through transient additions of acetyl or phosphate groups to the histones on which that stretch is spooled. When attached to specific amino acid side chains on a histone's tail, these chemical tags recruit proteins that change the local chromatin structure and, consequently, the accessibility of genomic DNA. The tags' removal reverses the effect.

Evidence is building that methylation should be added to the list of dynamic covalent modifications that control access to tightly packed genomic DNA. A number of enzymes transfer methyl groups to specific lysine and arginine side chains on histone tails. But after fruitless years of hunting for enzymes that remove these methyl groups, "some people suspected that histone methylation might be irreversible," says Ali Shilatifard, an associate professor of biochemistry and molecular biology at St. Louis University.

Such suspicions were put to rest late last year, when Yang J. Shi's group in the department of pathology at Harvard Medical School reported the first histone demethylase (Cell 2004, 119, 941). Shi, along with postdoc Yujiang Shi and graduate student Fei Lan, stumbled onto the demethylase while probing how a newly discovered protein suppresses gene expression. To their surprise, the protein does so by demethylating a specific histone methyllysine linked to active transcription.

THE DEMETHYLASE that Shi's group discovered (an amine oxidase they dubbed lysine-specific demethylase 1, or LSD1) does not directly cleave this methylated lysine's N–CH₃ bond. Instead, the researchers propose that the enzyme uses a flavin cofactor to oxidatively cleave the methylated lysine's C–H bond, forming an imine intermediate that is then hydrolyzed to give a carbinolamine that degrades into formaldehyde and lysine.

The discovery of this demethylase supports the long-held supposition that histone methylation is reversible, just as histone phosphorylation and acetylation are, Shilatifard points out.

In fact, earlier reports had hinted that cells might be able to reverse methylation. Last fall, Scott A. Coonrod of Cornell University's Weill Medical College and Yanming Wang, Joanna Wysocka, and C. David Allis of Rockefeller University reported that an enzyme called human peptidylarginine deiminase 4 (PAD4) converts methylarginine into citrulline, releasing the methyl group as methylamine in a process known as demethylimination (Science 2004, 306, 279; C&EN, Sept. 6, 2004, page 35). The researchers also noted then that PAD4 converts arginine into citrulline, too--an observation made concurrently by an independent team led by Graeme L. Cuthbert, Sylvain Doujat, and Tony Kouzarides of the University of Cambridge (Cell 2004, 118, 545).

Coonrod suggests that PAD4 serves to dismantle histone methylarginines. Kouzarides has other ideas: He proposes that PAD4 converts methylation-prone arginines to citrulline preemptively to "protect these residues against methylation." Both teams are now trying to determine whether the resulting citrulline residues are converted back to arginine or whether citrulline-bearing histones are simply replaced. The teams are also probing whether close relatives of PAD4 might remove histone methyl groups.

PAD4 can demethyliminate arginines bearing a single methyl group. But histone arginines may also be doubly methylated, either symmetrically (with a methyl group on each of its terminal amine groups) or asymmetrically (with two methyl groups on one of its terminal amine groups). It remains to be seen whether PAD4--or any other enzyme--can strip methyl groups from dimethylarginines.

Similarly, lysines can be loaded with up to three methyl groups. LSD1 can demethylate both mono- and dimethyllysines. But because LSD1 requires a protonated nitrogen to form the imine intermediate, this enzyme is unable to demethylate trimethyllysines. So how--and whether--histone trimethyllysines might be demethylated are questions that remain unanswered.

The importance of dynamic methylation may go far beyond modulating access to DNA in the chromatin. So-called histone methyltransferases "also methylate a broad range of other protein substrates," Coonrod explains, noting that arginine methylation is a particularly widespread modification. He suggests that histone demethylases are also likely to act on substrates other than histones. In support of this idea, he and Michael Stallcup of the University of Southern California recently reported that PAD4 demethyliminates methylated arginines on an acetyltransferase enzyme (Proc. Natl. Acad. Sci. USA 2005, 102, 3611). "I think we may find that dynamic methylation regulates a range of cellular processes," Coodrod says.