Methylation of the lysine residues of histone N-terminal tails is associated with local changes in chromatin structure and correlates with both gene activation and gene repression. The regulation of genes is, however, a dynamic process, and cells must therefore possess the ability to rapidly generate and erase the 'methyl mark'. Whereas several histone methyltransferases have been documented, the identification of an enzyme that removes the methyl group from histone tails has remained elusive, which has fuelled debate that histone methylation might constitute a permanent epigenetic modification. Now, Yang Shi and co-workers have brought us one step closer to resolving this question by the identification of a histone lysine demethylase, and they report this finding in Cell.

Shi and his team initially set out to characterize a protein, LSD1, that had been identified as a component of several transcriptional-repressor complexes. First, they showed that LSD1 repressed gene activity in a reporter assay and that this transcriptional repression was largely dependent on a region of the protein that had high sequence homology with amine oxidases — metabolic enzymes that are putative histone demethylases. This indicated that LSD1 might repress transcription by catalysing the demethylation of core histones.

Direct demethylation assays confirmed that LSD1 removed the methyl group from dimethyl and monomethyl H3-K4, but not from trimethyl H3-K4, H3-K9 or several other methylated lysine and arginine residues. H3-K4 methylation correlates with transcriptional activity, and so it is appropriate that a transcriptional repressor would erase this modification.

Because of its homology to the amine oxidases, the researchers predicted that an LSD1-mediated demethylation reaction would generate formaldehyde. Indeed, when LSD1 was incubated with dimethylated substrate in a spectrophotometric assay, large amounts of formaldehyde were detected. This finding provided further compelling evidence that LSD1 was a bona fide demethylase.

But does LSD1 repress endogenous genes by catalysing histone demethylation? The enzyme is found in the Co-REST complex — a transcriptional repressor that silences neuron-specific genes. LSD1-deficient HeLa cells that were generated by RNA interference aberrantly expressed several neuron-specific target genes, and chromatin-immunoprecipitation analyses detected both a loss of association of LSD1 and an increase in H3-K4 methylation at the promoters of these derepressed genes.

So, it seems that a lysine demethylase has at last been pinned down. It is important to note, however, that an enzyme that removes the methyl marks that identify heritably silenced chromatin still defies detection...