News & Views | Published:

Molecular biology

No exception to reversibility

  • A Correction to this article was published on 27 October 2004

Histone proteins, which serve as scaffolds for packaging DNA, can be modified in numerous ways. It's been thought that one modification, methylation, is irreversible — but that view must now change.

Each of our cells contains about two metres or so of DNA, which must be packed down very tightly to fit into the cell nucleus. The compact form of DNA is known as chromatin, the basic unit of which consists of DNA wrapped around an octamer of histone proteins1. The histones are not merely architectural proteins, however: they also influence chromatin dynamics. One way in which they do so is through their covalent modification with certain chemical groups or small proteins — acetyl groups, phosphate groups, ubiquitin proteins or methyl groups2. Enzymes that catalyse the addition or removal of the first three modifications have been identified. But enzymes that remove methyl groups have been more elusive, raising the question of whether methylation is an exception to the rule: the only histone modification that is irreversible3. Two new papers, however — published in Science by Wang et al.4 and in Cell by Cuthbert et al.5 — reverse that view.

The histone octamer consists of two copies each of four ‘core’ histones: H2A, H2B, H3 and H4. Methylation occurs on certain lysine and arginine amino acids within H3 and H4, and is catalysed by distinct families of enzymes6; PRMT1 and CARM1, for instance, are enzymes involved in arginine methylation. Lysine residues can be mono-, di- or tri-methylated, whereas arginine residues can only be mono- or di-methylated.

Histone methylation has been linked to biological processes ranging from the regulation of gene transcription, to the inactivation of one copy of the X chromosome in females, to RNA-mediated gene silencing2,6,7. One way in which it works is by serving as a docking site for other proteins2. The nature of a specific methylation determines the protein that it recruits, which in turn dictates the biological outcome (see, for example, ref. 6).

Unlike other histone modifications, methylation has generally been regarded as stable3 — a notion that comes from early studies showing that histone proteins and methylated lysine or arginine residues within them have similar turnover rates8. But although a non-reversible methyl mark would fit with a role for histone methylation in long-term gene silencing, it is not compatible with situations in which rapid reversal of gene expression takes place. To solve this paradox, mechanisms including enzyme-catalysed demethylation, replacement of methylated histones by unmodified histones, and clipping of methylated histone ‘tails’ have been proposed3,9 (Fig. 1a) — but none has yet been demonstrated experimentally. Now, however, Wang et al.4 and Cuthbert et al.5 have found that the human enzyme peptidylarginine deiminase 4 (PAD4/PADI4) can catalyse the conversion of methylated arginines to citrulline, providing yet another mechanism by which histone methylation levels could be controlled (Fig. 1b).

Figure 1: Reversing methylation in histone proteins.

a, Three theoretical mechanisms: top, enzymatic demethylation, perhaps regulated by the HDM enzyme; centre, replacement of methylated histones by unmethylated variants; bottom, clipping of histone ‘tails’, by enzymes unknown. K, lysine; R, arginine. b, Wang et al.4 and Cuthbert et al.5 have discovered another mechanism: they show that methylated and unmethylated arginine amino acids in two histone proteins can be converted to citrulline. A possible cycle of events is shown. Methylation of arginines is accomplished by the PRMT1 or CARM1 enzymes, with SAM (S-adenosyl-L-methionine) as cofactor; AdoHcy (S-adenosyl-L-homocysteine) is released. Conversion to citrulline is achieved by PAD4/PADI4. It remains unknown what then happens to citrulline — whether it is converted back to arginine (by an aminotransferase enzyme, perhaps), or whether histones containing citrulline are replaced by unmodified versions.

How is it that two groups independently identified the same enzyme and came to similar conclusions? Previous studies10 established that arginine residues in other proteins can be converted to citrulline by enzymes of the peptidylarginine deiminase family. Of this family, only PAD4/PADI4 is found in the nucleus; moreover, its expression correlates with the appearance of citrulline in histones11. These facts made it a good candidate for a histone arginine demethylase.

So Wang et al. and Cuthbert et al. carried out direct tests of PAD4/PADI4 in vitro. They found that it ‘deiminated’ numerous unmethylated arginines — converted them to citrulline — in histones H3 and H4. It also decreased the amount of arginine methylation that is catalysed by PRMT1 and CARM1 in both histones. Notably, at the same time that the amount of this methylation decreased, the quantities of citrulline in both histones increased. The detection of methylamine as a released product4 supports the idea that methylated arginine is a genuine substrate of this enzyme. So, PAD4/PADI4 can catalyse both the deimination of unmethylated arginine and the ‘demethylimination’ of methylated arginine in vitro. As Wang et al. find, it can also do so in granulocyte cells.

What are the consequences of these reactions? First, as Cuthbert et al. show, the conversion of histone arginines to citrullines actually prevents histone methylation by CARM1. Second, Wang et al. find that demethylimination of methylated histone arginines reverses the effects associated with methylation. So, PAD4/PADI4 presumably antagonizes the functions of the methylating enzymes CARM1 and PRMT1. For instance, previous studies have linked histone arginine methylation by these enzymes to transcriptional activation by nuclear hormone receptors12,13,14,15 — so PAD4/PADI4 is likely to repress such transcription. Indeed, both groups link the recruitment of PAD4/PADI4 to an oestrogen-responsive gene, pS2, with the appearance of citrullinated histones and the downregulation of this gene.

These papers4,5 provide convincing evidence that the methylation of arginine amino acids in histone proteins can be reversed enzymatically. But, as ever, the findings raise questions. For example, it seems that PAD4/PADI4 has a very loose substrate specificity in vitro. It works on both methylated and unmethylated arginines. And the arginine residues it deiminates are not limited to the sites that are targeted by CARM1 and PRMT1; indeed, arginine 8 in histone H3, a site not known to be methylated by either enzyme, is the preferred target4. There might be an interplay between the citrullination of H3 arginine 8 and the methylation of H3 lysine 9, but for now the significance of this citrullination remains unknown.

The second issue worth noting is that, in vitro, PAD4/PADI4 cannot demethyliminate H4 or H3 peptides containing di-methylated arginines4,5. This presents a puzzle. Although mass-spectrometry analysis14,15 identified mono-methyl groups as the major methyl form on arginine 3 in histone H4, most arginines 17 and 26 in histone H3 become di-methylated after incubation with CARM1 in vitro16. How, then, can this methylation be removed? Although an unidentified enzyme might be required, the available evidence suggests that PAD4/PADI4 is responsible: for instance, in granulocytes, activation of this enzyme led to less methylation on H4 arginine 3 and H3 arginine 17, as analysed using site-specific antibodies that recognize di-methylated arginine4. The most likely solution is that PAD4/PADI4 can work only on intact chromatin substrates. Alternatively, it might need to work with a partner.

As well as revealing that arginine methylation can be reversed, the new papers4,5 show that a new form of histone proteins — containing citrulline residues — exists in cells. This finding, too, raises questions. How stable are citrullinated histones? Do they have any effects on chromatin structure? Can other proteins recognize and bind to them? And what is the fate of these histones? With regard to the last question, the transient presence of citrullines in the pS2 gene5 suggests that their rapid removal must be possible. The availability of site-specific antibodies against citrullinated histones will allow at least some of these issues to be addressed. And one final question: is the methylation of lysine residues in histone proteins also reversible?


  1. 1

    Kornberg, R. D. & Lorch, Y. Cell 98, 285–294 (1999).

  2. 2

    Jenuwein, T. & Allis, C. D. Science 293, 1074–1080 (2001).

  3. 3

    Bannister, A. J., Schneider, R. & Kouzarides, T. Cell 109, 801–806 (2002).

  4. 4

    Wang, Y. et al. Science doi:10.1126/science.1101400 (2004).

  5. 5

    Cuthbert, G. L. et al. Cell 118, 545–553 (2004).

  6. 6

    Zhang, Y. & Reinberg, D. Genes Dev. 15, 2343–2360 (2001).

  7. 7

    Grewal, S. I. & Moazed, D. Science 301, 798–802 (2003).

  8. 8

    Byvoet, P. Arch. Biochem. Biophys. 152, 887–888 (1972).

  9. 9

    Henikoff, S., Furuyama, T. & Ahmad, K. Trends Genet. 20, 320–326 (2004).

  10. 10

    Vossenaar, E. R., Zendman, A. J., van Venrooij, W. J. & Pruijn, G. J. BioEssays 25, 1106–1118 (2003).

  11. 11

    Nakashima, K., Hagiwara, T. & Yamada, M. J. Biol. Chem. 277, 49562–49568 (2002).

  12. 12

    Bauer, U. M., Daujat, S., Nielsen, S. J., Nightingale, K. & Kouzarides, T. EMBO Rep. 3, 39–44 (2002).

  13. 13

    Chen, D. et al. Science 284, 2174–2177 (1999).

  14. 14

    Strahl, B. D. et al. Curr. Biol. 11, 996–1000 (2001).

  15. 15

    Wang, H. et al. Science 293, 853–857 (2001).

  16. 16

    Schurter, B. T. et al. Biochemistry 40, 5747–5756 (2001).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.