The most common forms of histone modification are acetylation and methylation, both linked to active and silenced transcription, respectively. The process of histone acetylation was very well known at the time of the publication of this paper, but the turnover of histone methylation was explained with two theoretical models, one was the removal of the complete histone tail or the replacement of the methylated histone, and the other was the existence of histone demethylases. Some authors proposed that amino oxidases could be function as demethylases, and a protein that could fulfill the picture was KIAA0601, because it copurified with corepressor complexes (such as Co-REST in non- neuronal cells), and showed significant sequence similarity with FAD- dependent amino oxidases. The authors first asked whether KIAA0601 functions as a repressor. To this end they expressed a Gal4- KIAA0601 fusion, and the reporter Gal4-luciferase promoter was strongly repressed; however, a C-terminal deletion construct, that lacks the amine oxidase homologous region, was not able to repress transcription. Therefore, the transcriptional function of this protein is linked to its enzymatic activity. KIAA0601 is known to bind FAD, and computational analyses predicted that it may catalyze oxidation reactions of amines. To answer this, they expressed a His-tag- KIAA0601 fusion, and the purified protein, that copurified with FAD, was incubated with either H3K4me2 or H3K9me2, and the methylation status was determined with antibodies. They observed that KIAA0601, but not the C-terminal deleted form of the protein, reduced the methylation levels in H3K4me2 but had no effect on H3K9me2 or unmethylated H3, which indicates specificity, and that the enzymatic activity is linked to its silencing function. This experiment was repeated with proteins isolated from cell lines, obtaining similar results. Then, they evaluated if this protein would have the same activity on the mono- or trimethylated forms of H3K4, and they found that the enzyme could reduce the methylation levels on the monomethyl K4 but not on the trimethylated form of the amino acid. Taken together with previous results, Shi et al. proposed a demethylation pathway for H3K4. In other similar assays (detected by Western blotting or by mass spectrometry), no other methylated residue was altered by this protein, further confirming the specificity, and the endogenous form of KIAA0601 showed the same activity and specificity that the recombinant fused form. As their model of reaction predicted the formation of formaldehyde, they used enzymatic assays to determine the presence of this metabolite, which resulted in a confirmation of their hypothetical pathway. The specificity of this protein, now termed as LSD1, for "lysine specific demethylase", was also confirmed with these biochemical assays. The formation of formaldehyde was also observed in mass spectrometry experiments. Finally, they asked whether LSD1 regulates transcription and histone methylation in vivo. They first knocked out LSD1 using an RNAi system, and observed that the Co-REST associated promoters were derepressed. ChIP experiments showed that LSD1 occupied the Co-REST associated promoters in vivo, and in the knocked out cells, the decrease of occupancy was accompanied with an increase of H3K4 dimethylation and promoter activity. They concluded that LSD1 is an specific histone lysine demethylase associated with epigenetic repression, and this paper is valuable because it was the initial characterization of an enzyme involved in processing H3K4 demethylation.
Shi Y. et al. 2004. Cell 119:941-953
epigeneticsMore comments can be found at the Faculty of 1000 website.
