Howard Hughes Medical Institute, Department of Biochemistry & Biophysics, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, North Carolina 27599-7295, USA. Correpondence to Y.Z. e-mail: yi_zhang@med.unc.edu doi:10.1038/nrm1761 SET DOMAIN A sequence motif (named after Su(var)3���9, Enhancer of Zeste, Trithorax) that is found in several chromatin-associated proteins, including members of both the Trithorax group (trxG) and Polycomb group (PcG). X INACTIVATION A process of dosage compensation in mammals that is achieved by the transcriptional silencing of one of the X chromosomes in XX females. THE DIVERSE FUNCTIONS OF HISTONE LYSINE METHYLATION Cyrus Martin and Yi Zhang1 Abstract | Covalent modifications of histone tails have fundamental roles in chromatin structure and function. One such modification, lysine methylation, has important functions in many biological processes that include heterochromatin formation, X-chromosome inactivation and transcriptional regulation. Here, we summarize recent advances in our understanding of how lysine methylation functions in these diverse biological processes, and raise questions that need to be addressed in the future. The DNA within our cells exists in the form of chro- matin. The basic building block of chromatin is the nucleosome, a structure consisting of an octamer of four core histone proteins around which 147 bp of DNA is wrapped1. The histone proteins (H2A, H2B, H3 and H4) are each composed of a globular domain and an unstructured tail domain. A striking feature of core histones is that they are subject to a large number of covalent modifications including acetylation, methylation, phosphorylation and ubiquitylation2. Histone methylation occurs on arginine and lysine residues and is catalysed by enzymes belonging to three distinct families of proteins ��� the PRMT1 fam- ily, the SET���DOMAIN-containing protein family, and the non-SET-domain proteins DOT1/DOT1L3���6. Here, we focus on histone lysine methylation because of its prominence and its array of important functions (for more information on arginine methylation, see several recent reviews6���8). Histone lysine methylation occurs on histones H3 and H4. So far, six lysine residues located on these two histones have been reported to be sites of methylation (FIG. 1a). Each of these lysine side chains can be mono-, di- or trimethylated. Unlike acetylation, which gener- ally correlates with transcriptional activation, histone lysine methylation can signal either activation or repression, depending on the sites of methylation6. For certain processes, such as X INACTIVATION, methyla- tion on the same site can lead to different outcomes depending on the number of methyl groups added. In the past few years, studies on different organisms have resulted in the identification of several enzymes that catalyse site-specific histone lysine methylation (FIG. 1a). The characterization of these enzymes has revealed important functions of histone methylation in many different biological processes that range from heterochromatin formation to transcription regula- tion. In this review, we summarize recent advances in this rapidly progressing field, including the discovery of histone demethylases, and outline questions to be addressed in the near future. Histone methylation and the ���histone code��� In its extended form, chromatin appears as an array of nucleosomes, but in the nucleus, the chromatin fibres that form chromosomes undergo several levels of fold- ing, resulting in increasing degrees of condensation9. It is known that the histone tails have an important role in this folding process10. Considering this and the fact that histone methylation primarily occurs in histone tails, it would be reasonable to propose that methyla- tion might function to regulate chromatin structure directly by affecting the higher-order folding of the chromatin fibre. This could have important implica- tions for chromatin-templated processes such as tran- scription and DNA repair, assuming that folding alters the accessibility of DNA to the proteins that mediate these processes. Although there is no evidence that lysine methylation directly affects chromatin dynamics, acetylation of lysine residues in histones 838 | NOVEMBER 2005 | VOLUME 6 www.nature.com/reviews/molcellbio R E V I E W S

Chd1 (H3 K4) HP1 (H3 K9) Pc (H3 K27) 53BP1 (H3 K79) Crb2 (H4 K20) WDR5 (H3 K4) Chromodomain Tudor domain WD40 repeat SUV39h1, SUV39hh2 ESET G9a GLP RIZ EZH2 NSD1 E(Z) Trithorax ASH1 Su(var)3-9 ASH1 Set1 (S.c.) Set2 (S.c.) Dot1L Dot1 (S.c.) Clr4 (S.p.) SUV4���20h1, SUV4���20h2 SET8/PR-SET7 NSD1 Su(var)4-20 ASH1 Set9 (S.p.) MES-2 SET1 MLL SET7/SET9 SMYD3 K9 K27 K36 K79 H3 H4 K20 K4 a b BROMODOMAIN A conserved acetyl-lysine binding domain found in several transcriptional regulatory proteins that are involved in gene activation. CHROMODOMAIN A conserved protein structure that is common to some chromosomal proteins. It interacts with chromatin by binding to methylated lysine residues in histone proteins. TUDOR DOMAIN A conserved protein domain that is found in several RNA- binding proteins and chromatin-associated proteins. Recent studies indicate this domain can bind to methyl- lysine or methyl-arginine. is known to antagonize the folding of chromatin in vitro9. As acetylation neutralizes the positive charge of lysine, it has been suggested that this modification might operate through an electrostatic mechanism. Since methylation of lysine residues does not alter their charge, any direct effect of lysine methylation on chromatin folding would have to occur through a non-electrostatic mechanism (for example, through hydrophobic interactions). An alternative hypothesis proposes that specific histone modifications, including lysine methylation, are binding sites for different proteins that mediate downstream effects11���13. Consistent with this hypoth- esis, it has been shown that BROMODOMAINS can recog- nize acetylated lysines14,15. Recent studies on histone methylation identified at least three protein motifs ��� the CHROMODOMAIN16���20, the TUDOR DOMAIN21,22 and the WD40���REPEAT DOMAIN23 ��� that are capable of specific interactions with methylated lysine residues (FIG. 1b). As described below, proteins that contain these motifs are recruited by specific methylated lysines, and this recruitment step seems to play an important part in the unique biological outcomes that are associated with different methylation events. However, additional levels of complexity exist: for example, as mentioned above, lysine residues can adopt one of three different meth- ylation states. Also, the binding affinity of a protein for a particular modification might be affected by another adjacent modification. If true, these added complexi- ties may support the ���histone code��� hypothesis, which posits that different combinations of histone modifica- tions mediate unique cellular responses11���13. From the time that the hypothesis was first pro- posed, the existence of a histone code has been ques- tioned24. Much of the debate seems to be rooted in semantics, as the existence of a code depends on the definition of the word ���code���. For example, Stephen Henikoff recently reviewed this topic and compared the histone code to a simple binary code, similar to a computer code consisting of permutations of 0 or 1 ���REF. 25���. On the basis of this strict definition, the three lysine residues (K4, K36 and K79) on each budding- yeast H3 molecule could generate eight possible meth- ylation permutations (for simplicity, the methylation state is not considered). Hypothetically, each of these permutations might be recognized by a different effec- tor protein, which would result in unique biological outcomes. Contrary to this expectation, recent data indicate that some histone modifications are closely correlated with each other26,27. This implies that the number of permutations that occur in vivo is limited, thereby diminishing the potential richness of a histone code. However, evidence for a certain level of ���cross- talk��� between different covalent histone modifications does exist28���30, although the generality of these obser- vations are unclear. In summary, on the basis of current evidence, it is difficult to reach a definitive conclusion concern- ing the existence of a strict histone code. Now that a more comprehensive list of histone modifications is available, correlative analyses of these modifications on a genome-wide scale will allow an estimate of the number of permutations that are used in nature. This, in turn, will allow conclusions to be drawn concerning the potential existence of a histone code. For the pur- pose of this review, however, it is sufficient to think of individual methylated lysines residues as simple dock- ing sites that recruit different effector proteins. Heterochromatin formation Heterochromatin has been historically defined as chromosomal regions, such as CENTROMERES and TELOMERES, that remain condensed throughout the cell cycle31. The proper formation of heterochromatin is biologically important: for example, centromeric heterochromatin formation is required for the proper segregation of chromosomes during mitosis32. In addi- tion, heterochromatin formation also plays a crucial role in recombination events that are associated with MATING���TYPE SWITCHING in fission yeast33 ���BOX 1���. Figure 1 | Histone lysine methyltransferases, their target sites and methyl-lysine binding domains. a | The lysine methyltransferases are grouped according to the specific lysine residue targeted for modification and colour coded according to their origin (yeast, red worm, yellow fly, pink mammalian, purple). Methyltransferases are indicated as S.c. and S.p. for Saccharomyces cerevisiae and Schizosaccharomyces pombe, respectively. For the mammalian enzymes the name for only one species is given. The globular domains of the histones are shown as ovals and the tails are represented by straight lines. b | Methyl-lysine binding proteins contain one of three methyl-lysine binding domains: the chromodomain, the tudor domain or the WD40-repeat domain. These can not only interact with methyl-lysine, but also seem to discriminate between different methylated lysines. For example, the chromodomain of HP1 interacts specifically with methyl-H3-K9, whereas that of Polycomb (Pc) interacts specifically with methyl-H3-K27. This probably explains why different methylated lysines within histones H3 and H4 can have different biological outcomes. NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 6 | NOVEMBER 2005 | 839 R E V I E W S