This Article Abstract Full Text (PDF) All Versions of this Article: 26/2/147    most recent Final Manuscript Author Manuscript Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lee, D. Y. Articles by Stallcup, M. R. Search for Related Content PubMed PubMed Citation Articles by Lee, D. Y. Articles by Stallcup, M. R.

Role of Protein Methylation in Regulation of Transcription

Departments of Biochemistry and Molecular Biology (D.Y.L., M.R.S.) and Pathology (C.T., M.R.S.), University of Southern California, Los Angeles, California 90089; and Department of Biochemistry and Biophysics (B.D.S.), University of North Carolina School of Medicine, Chapel Hill, North Carolina 27599

Correspondence: Address all correspondence and requests for reprints to: Michael R. Stallcup, Department of Pathology, HMR 301, University of Southern California, 2011 Zonal Avenue, Los Angeles, California 90089-9092. E-mail: stallcup{at}usc.edu

	Abstract

Top Abstract I. Introduction II. Lysine Methylation III. Arginine Methylation IV. Conclusions and Unanswered... References

 
In the last few years, the discovery of lysine and arginine methylation in histones and other proteins and the enzymes that carry out these posttranslational modifications has added a new dimension to the signal transduction field. In particular, there has been a huge surge in our understanding of how methylation of nucleosomal histones at specific lysine or arginine residues affects chromatin conformations and either facilitates or inhibits transcription from neighboring genes. It appears that the responsible methyltransferases can be targeted in some cases to specific genes and in other cases to broader regions of euchromatin or heterochromatin. Methylation of histones is mechanistically linked to other types of histone modifications, such as acetylation, phosphorylation, and monoubiquitylation; combinations of these modifications cooperate to regulate chromatin structure and transcription by stimulating or inhibiting binding of specific proteins. Although lysine methylation has thus far been observed almost exclusively on histones, arginine methylation has been observed on a variety of other proteins associated with gene regulation, including DNA-binding transcriptional activators, transcriptional coactivators, and many RNA binding proteins involved in RNA processing, transport, and stability. Thus, lysine and arginine methylation of proteins, like many other types of posttranslational modifications, are regulated steps of many specific signaling pathways.

I. Introduction

II. Lysine Methylation

A. Enzymes

B. Overview of histone modifications

C. Functional implications of methylation of individual lysine residues of histones

D. Consequences of histone lysine methylation

E. Lysine methylation of a basal transcription factor

III. Arginine Methylation

A. Enzymes

B. Effects of arginine methylation on protein function

C. Regulation of arginine methyltransferase activity

IV. Conclusions and Unanswered Questions

A. Protein methylation and endocrinology

B. Are arginine and lysine methylation reversible or stable marks?

C. Future directions

	I. Introduction

Top Abstract I. Introduction II. Lysine Methylation III. Arginine Methylation IV. Conclusions and Unanswered... References

 
PROTEIN METHYLATION CAN occur on arginine, lysine, histidine, proline, and carboxyl groups. Studies within the last decade have identified a wide variety of posttranslational modifications that occur on histones and other proteins involved in the regulation of transcription, including lysine- and arginine-specific methylation (1, 2). This review will discuss the roles of lysine and arginine methylation in regulating gene expression at a variety of levels, but will emphasize the role of histone methylation in the regulation of chromatin structure and transcription. Before 1999, evidence suggested that protein arginine methylation is involved in various signaling pathways (3) and that the methylation of some RNA binding proteins is involved in their nuclear-cytoplasmic shuttling (4). Similarly, lysine methylation of histones had been extensively documented, but the function of this modification on histones remained elusive (5). In 1999, two papers provided compelling evidence that lysine and arginine methylation of histones functions in the process of gene transcription (6, 7). Since then, an explosion of new information on this topic, as well as the development of a new field of chromatin biology, has occurred.

Chromatin structure plays an integral role in the control of gene expression. The basic repeating unit of chromatin is the nucleosome, in which 146 bp of DNA wraps around an octamer of core histones, consisting of pairs of H3, H4, H2A, and H2B (8). N-terminal tails of histones protrude out of the nucleosome and are subject to a variety of posttranslational modifications such as acetylation, phosphorylation, ubiquitylation, and lysine and arginine methylation. Acetylation was the first of these modifications to be linked with active transcription, and subsequently phosphorylation of histone H3 was found to cooperate with acetylation in transcriptional activation (9, 10). There are many sites of lysine and arginine methylation in histones, and they play a variety of important and, in some cases, essential roles in regulating chromatin structure and gene transcription. Some histone methylation events, e.g., methylation of Lys-4 and Arg-17 of histone H3 and Arg-3 of histone H4, have also been associated with transcription activation; in contrast, methylation of H3 Lys-9 has been correlated with gene silencing (11, 12). Although many correlations of specific histone modifications with active or inactive chromatin are generally valid, exceptions do exist; it is now widely held that individual histone modifications may not constitute clear signals by themselves, but rather multiple modifications probably function together as part of a histone code which states that sequential or concurrent combinations of modifications constitute signals that are read by other proteins (1, 2, 13).

Although histones (specifically H3 and H4) have so far been the stars of the saga of protein methylation in transcriptional regulation, recent work has shown that methylation regulates the activities of an increasing list of other components of the transcription machinery. In addition to addressing how methylation of nonhistone proteins contributes to transcriptional regulation, we will also briefly discuss how methylation of a variety of protein substrates contributes to regulation of various posttranscriptional levels of gene regulation and to the regulation of various cellular signal transduction pathways.

	II. Lysine Methylation

Top Abstract I. Introduction II. Lysine Methylation III. Arginine Methylation IV. Conclusions and Unanswered... References

 
A. Enzymes
1. Classification of histone lysine methyltransferases and their histone substrates.
Histone lysine methylation occurs on histone H3 at lysines 4, 9, 14, 27, 36, and 79 and on histone H4 at lysines 20 and 59 (1, 12, 14, 15, 16). Many of the enzymes that modify these particular residues have been isolated and characterized (Fig. 1A), and crystal structures have been determined for some of them (17, 18, 19, 20, 21, 22, 23). All of the lysine-specific histone methyltransferases (HMTs) except Dot1 share a SET [Su(var), Enhancer of zeste, trithorax] domain that is responsible for catalysis and binding of cofactor S-adenosyl-L-methionine (AdoMet). HMTs then add one or more methyl groups to the -amino group of lysine residues, resulting in mono-, di-, or trimethylated lysine (Fig. 1B). Methylation of lysine residues does not change the net positive charge but progressively increases the bulk and hydrophobicity and may disrupt intra- or intermolecular hydrogen-bond interactions of the -amino group or create new sites for proteins that bind preferentially to the methylated protein. N-C bonds of methyllysine are very stable, and so far no demethylases have been discovered, leading to questions about the reversibility of this modification. These issues will be discussed further in Section IV.

View larger version (24K): [in this window] [in a new window]   FIG. 1. Arg and Lys methylation of histones. A, Methylation sites of histones H3 and H4. The N-terminal amino acid sequences of histones H3 and H4 are shown along with positions of specific methylation sites and some of the HMT and PRMT enzymes responsible for methylation at each site. Arg methylation is shown above the sequence, and Lys methylation below. The names of the three enzymes for which the SET domains of Lys HMTs were named are indicated on the left. Lys residues known to be acetylated are shown in red, and Ser residues known to be phosphorylated are shown in blue. B, Lys methylation. The structures of mono-, di-, and trimethyllysine are shown. C, Arg methylation. Structures of monomethyl-, asymmetric dimethyl-, and symmetric dimethylarginine are shown, along with the type of PRMT enzyme responsible for synthesizing each species.

Although histones are by far the predominant substrates identified for HMTs, a few nonhistone substrates were previously identified: calmodulin, cytochrome c, and ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) from plants (19). In fact, the idea that Suv39h1 might have HMT activity arose from the sequence homology among the Suv39h1 SET domain and other proteins with SET domains, including plant Rubisco large subunit methyltransferases, which were known to methylate Lys-14 of Rubisco (29). Thus, it would not be surprising to find a whole range of other proteins that are methylated on lysine residues; a similar experimental trend occurred with protein acetylation that was initially characterized primarily on histones but later recognized as a posttranslational modification for many nonhistone proteins (30, 31, 32, 33).

B. Overview of histone modifications
Histones were long regarded as a passive packaging structure for DNA. Of course, now it is widely recognized that histones play a dynamic role in controlling chromatin structure and transcription. There are apparently many different states of chromatin compaction. The more compact chromatin, called heterochromatin, generally is characterized by late replication during S phase (due to its highly condensed nature), low gene density, and repetitive DNA sequences (34, 35). Euchromatin has a more open structure and contains genes that are active or potentially active. It is now clear that histone lysine methylation plays a major role in regulating the state of chromatin compaction, and thus the establishment and maintenance of heterochromatic and euchromatic regions in chromatin. In addition, histone lysine methylation also apparently plays central roles in regulating activation and repression of gene transcription within euchromatin.

In general, methylation of histone H3 at Lys-4, 36, and 79 is correlated with euchromatin and transcriptional activation, whereas methylation of histone H3 at Lys-9 and 27 and histone H4 at Lys-20 is associated with heterochromatin and transcriptional repression (Figs. 2–4). However, this is not always the case, and it should be noted that the specific functions of various histone modifications are still under intensive investigation; thus the information summarized here should be considered as a progress report with the understanding that many new insights are yet to come. Furthermore, the histone code appears to have some variations in certain organisms, particularly in budding yeasts. For example, methylation of histone H3 at Lys-9, the Suv39h1 class of HMTs responsible for that modification, and the heterochromatin protein 1 (HP1) that binds to histone H3 methylated at Lys-9 do not exist in Saccharomyces cerevisiae, which also lacks the large blocks of heterochromatin found in higher eukaryotes (7, 26, 27, 36). Instead, S. cerevisiae uses H3 Lys-4 and Lys-79 methylation to limit the extent of silent domains with the help of Sir silencing proteins (24). One reason for this difference may be that the budding yeasts have very little repetitive DNA compared with higher eukaryotes. Thus, the general correlation between each histone modification and its function may not be universally true among all species, and caution should be used about interpretation and generalization of findings across wide evolutionary gaps.

View larger version (25K): [in this window] [in a new window]   FIG. 2. Examples of transcriptional regulation mechanisms by histone lysine methylation. A, The molecules responsible for or involved in recruitment of the HMT. There are both global and gene-specific mechanisms for recruiting HMTs to broad regions of heterochromatin and euchromatin or to specific genes, respectively. Molecules responsible for global recruitment (see text) include: shRNA or short heterochromatic RNA, short double-stranded RNA transcribed from centromeric heterochromatin; Xist, RNA transcribed from the Xist gene and bound to the inactive X chromosome; and Paf1, the Paf1 protein complex associated with RNA pol II. DNA-binding transcriptional activator proteins or cofactor (coactivator or corepressor) proteins that can recruit specific HMTs to specific genes (see text) include: PRDI-BF1 transcription factor; Rb, the retinoblastoma protein, a corepressor that is recruited by specific transcription factors; GAGA transcription factors; EcR, the Drosophila ecdysone receptor, a nuclear hormone receptor family transcriptional activator protein. In addition, the Ser-5-phosphorylated CTD of RNA Pol II and monoubiquitylated histone H2B can recruit specific HMTs to active genes. B, HMT enzyme responsible for the modification. C, Specific histone H3 residue methylated. D, Protein that binds or is prevented from binding by the specific histone H3 methylation. Arrowheads indicate positive effects, whereas lines with flat ends indicate inhibition. HAT, p300 histone acetyltransferase. E, Effect on transcription. mam, Mammalian; d, Drosophila; sc, S. cerevisiae; sp, S. pombe.

View larger version (14K): [in this window] [in a new window]   FIG. 3. Distribution of histone H3 Lys methylation and key proteins responsible for formation and maintenance of heterochromatin and euchromatin. Similar patterns of Lys methylation have been found in fission yeast (S. pombe), budding yeast (S. cerevisiae), and humans (H. sapiens). Centromeric heterochromatin in the indicated organisms contains HP1 proteins, which bind to histone H3 methylated at Lys-9, and also contains the Lys-9 HMT Suv39h1, which is recruited by shRNA and by HP1. These findings suggest a mechanism for spreading of heterochromatin structure. Lys-4 and Lys-79 methylation in euchromatic regions is thought to limit the spreading of heterochromatin by facilitating the binding of proteins that promote euchromatin structure and inhibiting the binding of proteins that promote heterochromatin spreading (see Figs. 2 and 4). The centromeric and telomeric regions of budding yeast do not have the dense heterochromatin characteristic of fission yeast and more complex eukaryotes, but the chromatin in budding yeast telomeric regions shares some characteristics of the heterochromatin structure of the other organisms, including the presence of Sir proteins in the telomeric regions. The fact that different proteins bind to telomeric vs. centromeric heterochromatin suggests that there are different types of heterochromatin regulated by different mechanisms.

View larger version (20K): [in this window] [in a new window]   FIG. 4. Regulation of Lys methylation of histone H3 by monoubiquitylation of histone H2B, by the Paf1 protein complex, and by phosphorylation of the CTD of RNA pol II. Lys-4, Lys-36, and Lys-79 methylation of histone H3, which are generally associated with active transcription, are regulated by a series of prior events (shown on the left) and promote active transcription by influencing the binding or activity of specific proteins on histone H3 (shown on the right). Arrowheads indicate positive effects, and lines with flat ends represent negative effects. Enzymes or proteins responsible for the positive effects are in boxes associated with the arrows. Monoubiquitylation of histone H2B at Lys-123 by Rad6 (ubiquitin E2 ligase) and Bre1/Lge1 (ubiquitin E3 ligase) leads to methylation of histone H3 at Lys-4 and Lys-79. Set1 preferentially associates with the CTD of RNA pol II, which is phosphorylated at Ser-5. Set2 prefers to bind the CTD of RNA pol II phosphorylated at Ser-2. The Paf1 complex contributes to the recruitment of several of the key enzymes. Lys-4 methylation promotes remodeling of chromatin to the active state by attracting p300 histone acetyltransferases (HAT) and coactivator Isw1p and by inhibiting the binding of the NuRD corepressor complex. Lys-79 methylation in euchromatin inhibits the binding of repressive Sir proteins. The mechanism by which Lys-36 methylation contributes to active transcription is unknown.

It should also be noted that there may be separate mechanisms for using histone modifications to establish global vs. gene-specific patterns of chromatin structure. At our current level of understanding, some histone modifications appear to be associated with establishment of broad regions of chromatin structure (e.g., heterochromatin vs. euchromatin), whereas other histone modifications appear to be regulated at the level of the individual gene or even specific regions of a gene (e.g., promoter vs. coding region) and are involved with regulating transcription of a specific euchromatic gene.

C. Functional implications of methylation of individual lysine residues of histones
As discussed above, most histone modifications have been primarily associated with either activation or repression of transcription. However, for methylation at many lysine residues, there are indications of its involvement in both activation and repression, either in different regulatory situations or due to differences in the histone code among different species. Below, for each modification we discuss the evidence for its involvement in activation or repression or both. The downstream consequences of Lys methylation will be discussed in Section II.D.

1. H3 Lys-4 methylation
a. Activation
i. Global chromatin structure.
At the global chromatin level, overwhelming evidence supports association of histone H3 Lys-4 methylation with euchromatin. At the individual gene level, methylation of this residue is usually (but not always) correlated with active transcription. Several key observations support the first point. Methylation of histone H3 at Lys-4 has been correlated with transcriptionally active macronuclei of Tetrahymena (7) and with euchromatic regions in fission yeast Schizosaccharomyces pombe (37). This general association also holds true in vertebrates (38, 39). For example, the highly transcribed ß-globin locus of 10-d erythrocytes of chicken embryo contains dimethylation of H3 Lys-4 within the 30-kb ß-globin locus, whereas the adjacent 15 kb of condensed, inactive chromatin did not contain this modification (38). With regard to gene-specific regulation, deletion of the major histone H3 Lys-4 HMT Set1 in the budding yeast S. cerevisiae resulted in the repression of most genes (5059/6144 ORFs; 80%) as determined by microarray analysis (40). In addition, it was shown that dimethylation of H3 Lys-4 is associated with the coding regions of active genes in S. cerevisiae (41). A subsequent analysis using newly developed antibodies then showed that it was the trimethylated Lys-4 that is highly correlated with active genes, whereas dimethylated Lys-4 is a mark for both active and inactive euchromatin based on the analysis of seven different genes in S. cerevisiae (42). These results have suggested that Lys-4 dimethylation is a mark of transcriptional permissiveness that functions to demarcate euchromatic from heterochromatic regions, whereas the trimethylation event at this same residue, which is restricted primarily to the 5' end of genes, plays a direct transcriptional role. Unlike in S. cerevisiae, however, dimethylation of Lys-4 in metazoans is not found to occur broadly throughout gene loci, but rather is found to track similarly with trimethylation of Lys-4, which occurs predominately, but not exclusively, in the promoter and 5' end of genes (43, 44). Thus, both the di- and trimethylated Lys-4 forms in multicellular organisms appear to function in the activation process.

ii. Recruitment of Set1 HMT: roles of RNA polymerase II (pol II) phosphorylation and PAF1 complex.
How then is Set1 recruited to the 5' coding region of active genes and to euchromatin? So far, the mechanism does not appear to be directly dependent on transcriptional activators but instead depends on the RNA pol II elongation machinery, at least in S. cerevisiae (45, 46). The elongation process of transcription is regulated by phosphorylation of the C-terminal domain (CTD) of RNA Pol II and by elongation factors such as the Paf1 complex. The CTD of RNA Pol II consists of a long series of heptapeptide repeats, Tyr-Ser-Pro-Thr-Ser-Pro-Ser. The phosphorylation status of the CTD correlates with the stages of RNA pol II in the transcription process. Ser-5 phosphorylation of the CTD is important for facilitating the transition from transcription initiation to elongation and correlates with the promoter and the early phase of transcriptional elongation; Ser-2 phosphorylation of the CTD is associated with the late phase (i.e., downstream coding region) of transcription elongation. In S. cerevisiae, Ser-5 is phosphorylated by the TFIIH-associated Kin28, whereas Ser-2 is phosphorylated by Ctk1 (Fig. 4) (47, 48).

Several cooperative mechanisms are thought to contribute to the recruitment of Set1 to 5' coding regions of actively transcribed genes. First, Set1 specifically associates with Pol II when the CTD is phosphorylated at Ser-5 but not at Ser-2 (45) (Fig. 4). In S. cerevisiae, deletion of Kin28 leads to significant loss of recruitment of Set1 to the 5' coding region in the PYK1 gene; this suggests that the newly initiated polymerase acts as a signaling platform for the recruitment of Set1, resulting in Lys-4 trimethylation of the core promoter and early coding region (45). Second, components of the Paf1 transcription elongation complex interact with Set1 and are also required for recruitment of Set1 (46) (Figs. 2 and 4). Although the relationship between Ser-5 CTD phosphorylation and Paf1-Set1 recruitment is unclear, it is possible that the Paf1 elongation complex recruits Set1 in a manner that is stabilized by the binding of Set1 to the Ser-5-phosphorylated CTD. Although these mechanisms have not been examined in more complex eukaryotes, the observation that trimethylation of Lys-4 occurs primarily in the promoter and 5' coding regions of active genes (43, 44) suggests that the recruitment of Set1 via the Paf1 complex and Pol II will most likely be conserved.

iii. Recruitment of Set1 HMT: role of histone H2B monoubiquitylation.
Another mechanism controlling Lys-4 methylation was elucidated by the discovery that monoubiquitylation of histone H2B at Lys-123 regulates dimethylation of histone H3 at Lys-4 and Lys-79 in S. cerevisiae (49, 50, 51, 52) (Fig. 4). The E2 ubiquitin conjugating enzyme Rad6 is responsible for H2B monoubiquitylation, and its deletion results in the elimination of global H2B monoubiquitylation and dimethylation of histone H3 at Lys-4 and Lys-79, but not Lys-36. In contrast, mutation of histone H3 Lys-4 to Arg had no effect on H2B monoubiquitylation, demonstrating a unidirectional regulatory path for controlling the global levels of H3 methylation. These studies provide a new paradigm whereby modifications on one histone can regulate the outcome of modifications on a different histone, hence the term "trans-histone" regulation. Bre1, an E3 ubiquitin ligase, is found to associate with Rad6 and is responsible for the targeting of Rad6 to chromatin in S. cerevisiae. As expected, the deletion of Bre1 also leads to loss of monoubiquitylated H2B at Lys-123 and global loss of Lys-4 and Lys-79 dimethylation (52, 53).

A significant advance in our understanding of trans-tail histone regulation and Lys-4 methylation came from studies showing that components of the Paf1 complex are also required for global H2B Lys-123 monoubiquitylation and thus, Lys-4/Lys79 methylation (54, 55, 56). Like Set1, Rad6 associates with the elongating polymerase via the Paf1 complex, and Kin28 inactivation (i.e., the loss of CTD phosphorylation at Ser-5) results in elimination of H2B monoubiquitylation (Fig. 4). Thus, the association of Rad6 with Pol II appears to be essential for its catalytic activity (56).

Given that Rad6 and H2B ubiquitylation do not contribute directly to the recruitment of Set1, it is likely that H2B ubiquitylation functions to create an environment in chromatin that is permissive for Set1 and Dot1 methylation (45). Because Rad6 and H2B monoubiquitylation track with elongating Pol II, it has been suggested that H2B monoubiquitylation may disrupt the nucleosomes surrounding Pol II, thereby making them accessible to the cotraveling HMTs including Set1. In addition, a new study reveals that proteasomal ATPases are recruited to ubiquitylated H2B and are required for Lys-4 and Lys-79 methylation of H3 (57). These studies link proteasome function to the establishment of Lys-4 and Lys-79 methylation and suggest that chromatin remodeling is required for some HMTs to recognize their sites of methylation in chromatin.

iv. H3 Lys-4 methylation in transcriptional activation by nuclear receptors and other DNA-binding transcription factors.
Although Set1 is the only enzyme in yeast responsible for Lys-4 methylation, this enzyme belongs to the trithorax group of genes, and a number of homologs have been identified in other species (58, 59). In Drosophila, TRR, a trithorax-related SET domain protein that di- and trimethylates histone H3 at Lys-4, can be recruited to promoters by interacting with the ecdysone receptor, which is a DNA-binding transcription factor in the nuclear receptor family (60). In Drosophila S2 cells, TRR is recruited to the promoters of hedgehog and BR-C Z1 genes by the ecdysone receptor in an ecdysone-dependent manner. Furthermore, two different trr truncation mutants displayed significant decreases in the mRNA level and the promoter-trimethylation level of ecdysone-responsive genes. The human homolog of Drosophila trithorax, MLL/ALL1/HRX, is also recruited to the promoter of Hox c8 gene (61, 62), suggesting that other types of DNA-binding transcription factors may recruit this HMT to promoters. Finally, Herr and colleagues (63) identified the human Set1 homolog and showed it to be involved in mammalian cell proliferation. Other Lys-4 HMTs such as Set7/9 have been identified (64, 65), but their functions in chromatin are poorly defined. Nevertheless, all of the Lys-4 HMTs appear to play a significant role in gene activation.

b. Role of H3 Lys-4 methylation in repression in budding yeast: direct or indirect effects?
Although overwhelming evidence associates histone H3 Lys-4 methylation with gene activation, there are some cases in which this modification is implicated in gene repression, most notably in budding yeast. In fact, Set1 was originally identified as a protein important in gene silencing in S. cerevisiae (36, 66, 67). In budding yeast, three regions [mating type loci, telomeres, and ribosomal DNA (rDNA)] display heterochromatin-like behavior. Genes placed near or within these regions become silenced in a manner similar to the position effect variegation (PEV) phenomenon observed in Drosophila. Deletion of the SET1 gene in S. cerevisiae led to disruption of silencing of an artificial reporter gene integrated in telomeres (66, 67), mating type loci and rDNA (36, 68). In addition, deletion of RAD6 or mutation of H2B Lys-123 to Arg caused defects in silencing of a URA3 reporter gene in the telomeres (49, 69). Furthermore, deletion of SET1 or mutation of Lys-4 to Arg caused an increase in expression of an integrated reporter gene at the rDNA locus along with loss of Lys-4 dimethylation in the same region (36, 68). Thus, it appears that Set1 plays a complex role in gene activation and repression in budding yeast, and more work will be required to understand how it participates in both processes. However, another interpretation of the above data is that the loss of Lys-4 methylation may affect gene silencing by an indirect mechanism (70). This was suggested by the fact that the silenced regions in budding yeast appear to be hypo-methylated at Lys-4 (41, 71). If the distribution of Lys-4 methylation (low in silenced regions, high in potentially active regions of chromatin) helps to restrict Sir silencing proteins to specific (silent) chromatin regions, then global loss of Lys-4 methylation could lead to a redistribution of the Sir silencing proteins, thereby resulting in increased expression of genes in loci that are normally silenced.

So far, evidence implicating H3 Lys-4 methylation in gene silencing seems to be restricted to budding yeast. In fission yeast, H3 Lys-4 dimethylation was not present in the mating type or the centromere regions, and in contrast to the silencing defect observed in budding yeast, deletion of Set1 did not lead to loss of silencing of a reporter gene integrated in those regions (72). The association of Lys-4 methylation with repression in budding yeast may be explained by the fact that this organism lacks Lys-9 methylation (36). Thus, while Lys-4 methylation is clearly associated with active genes, silencing functions in budding yeast may also be controlled by Lys-4 methylation, directly or indirectly.

2. H3 Lys-79 methylation
a. Activation.
Dot1 is a unique HMT because it does not contain a SET domain and methylates Lys-79 of histone H3, which is located in the core rather than the tail of the nucleosome (24, 73, 74, 75). The distribution of H3 Lys-79 methylation is similar to that of H3 Lys-4 methylation at both global and gene-specific levels (Fig. 3). At the global level both modifications are associated with euchromatin. In S. cerevisiae, H3 Lys-79 dimethylation is present in the euchromatin but not in the heterochromatic rDNA, telomere, and silent mating type regions (76). In higher eukaryotes, active chromatin regions within the ß-globin, Ig heavy chain, and TCRß chain loci were enriched in H3 Lys-79 methylation in specific hematopoietic cell lineages (76, 77).

A recent model proposes that methylation of H3 Lys-79 functions in budding yeast by inhibiting binding of Sir2/3 proteins, which deacetylate histones and help to establish heterochromatin in silenced regions such as telomeres, rDNA, and mating type regions (Figs. 3 and 4). Deletion or overexpression of Dot1 results in mislocalization of Sir2/3 from silenced regions (24, 73, 78), suggesting a complex role for H3 Lys-79 methylation in regulating Sir2/3 localization. It is not clear how H3 Lys-79 methylation is restricted from heterochromatin and how this modification regulates Sir2/3 localization. Di- and trimethylation of H3 Lys-4 inhibits binding of the nucleosome remodeling and deacetylase (NuRD) corepressor complex (Figs. 2 and 4) (which also contains histone deacetylases) and also inhibits H3 Lys-9 methylation by Suv39h1 (65, 79), but whether methylation of H3 Lys-79 has similar consequences is not known. It is interesting to note that deletion of SET1 and DOT1 in S. cerevisiae synergistically reduces the occupancy of Sir2 at the telomere, suggesting cooperative mechanisms for Set1 and Dot1 in establishing and maintaining euchromatin and heterochromatin (54).

Mechanisms for recruitment of Dot1 to euchromatin are similar to Set1 at a global level. In budding yeast, H2B Lys-123 monoubiquitylation and components of the Paf1 complex control both Lys-79 and Lys-4 methylation of histone H3 (Fig. 4) (46, 52). At a gene-specific level, dimethylated Lys-79 is present at a similar level in the promoter and the coding regions of active genes in S. cerevisiae (76). Dot1 occupancy, however, was more enriched in the coding regions than the promoters (54).

It is interesting to note that the steady-state level of H2B monoubiquitylation at Lys-123 is very low, around 5%, whereas the levels of H3 Lys-4 and Lys-79 methylation are much higher, 34 and 90%, respectively in S. cerevisiae (49). The lower steady-state level of the ubiquitin modification may be due to the reversibility of this modification by deubiquitylating enzymes and suggests that the ubiquitin modification does not have to persist to maintain H3 Lys-4 and Lys-79 methylation (80, 81, 82). The higher levels of H3 Lys-4 and Lys-79 methylation suggest that they may not be reversible or at least that they have a longer half-life; the question of reversibility of lysine methylation is still unanswered. In any case, histone H2B monoubiquitylation may play a central role for establishing and/or maintaining euchromatin, and thus understanding how the Bre1-Rad6 complex is recruited and regulated will shed light on this process.

3. H3 Lys-36 methylation
a. Activation.
The Lys-36 residue of histone H3 lies at the junction between the histone tail and core domains (25). Because of its unique location, methylation of this residue could thus exert an effect by directly altering nucleosome structure or by promoting or inhibiting binding of chromatin remodeling proteins that recognize this mark. Currently, the role of H3 Lys-36 methylation in global euchromatin or heterochromatin formation is not clear. But, because methylation of H3 Lys36 does not occur at the telomeres and rDNA regions of S. cerevisiae, it is likely that this modification is associated with euchromatin (83).

At the individual gene level, this modification is associated with active genes, similar to H3 Lys-4 methylation (25, 83, 84, 85, 86, 87). In S. cerevisiae, the histone H3 Lys-36 HMT, Set2, preferentially binds to the Ser-2-phosphorylated vs. the unphosphorylated CTD of RNA Pol II (83), suggesting a mechanism for Set2 recruitment to the coding regions of the transcribed genes (Fig. 4). Furthermore, deletion of approximately 10 heptapeptide repeats of the CTD of RNA pol II resulted in a significant global loss of histone H3 Lys-36 methylation, while having no effect on the Lys-4 or Lys-79 methylation level (83, 87). Deletion of individual components of the Ctk complex also led to complete loss of H3 Lys-36 methylation at a global level, providing strong evidence that Ser-2 phosphorylation controls H3 Lys-36 methylation by providing a recruitment signal for Set2. As with Set1 recruitment, the Paf1 complex also plays an important role in the recruitment of Set2 (86). Thus, both the Paf1 complex and Ser-2 phosphorylation of CTD work together to target Set2 to coding regions of actively transcribed genes.

In contrast to the positively correlated relationship between H2B Lys-123 monoubiquitylation and H3 Lys-4 methylation, H2B monoubiquitylation appears to inhibit H3 Lys-36 methylation. Mutation of H2B Lys-123 to Arg in S. cerevisiae led to a dramatic increase in the dimethylation of H3 Lys-36 in the GAL1 promoter along with activated transcription (80). Set2 and H3 Lys-36 dimethylation occur on both the promoter and coding regions of several genes, although the relative levels of Set2 and Lys-36 methylation are higher in the coding regions (83, 86). However, because not all genes are methylated at H3 Lys-36, it appears that there is a gene selective targeting mechanism for Set2 (83).

b. Repression.
Similar to Set1, Set2 also was originally implicated in transcriptional repression in budding yeast (25, 88). Tethering of Set2 to the promoter of a reporter gene by fusing Set2 to a DNA-binding domain led to repression of reporter gene expression, and the repression was relieved partially by mutations in the SET domain (25). Furthermore, Set2 is required for maintaining low basal expression of the GAL4 gene in S. cerevisiae (88). Deletion of Set2, mutation of the catalytic site of Set2, or mutation of histone H3 Lys-36 to Arg led to elevation of the basal level of GAL4 gene expression. It is possible that methylation of H3 Lys-36 in promoter regions of genes is responsible for repression of transcription, whereas coding region methylation at H3 Lys-36 could be associated with active transcription (83). How Set2 and H3 Lys-36 methylation is directed toward specific promoters is not clear. Presumably, a different targeting mechanism is responsible for the histone H3 Lys-36 methylation in the promoter region and coding regions. Alternatively, the mono- and trimethyl forms of Lys-36, if they exist, may function to regulate different aspects of the transcription process than the dimethyl forms. Nevertheless, the dual roles of H3 Lys-36 methylation on transcription activation and repression may provide another example of the complexity of the histone code.

4. H3 Lys-9 methylation
a. Repression
i. Role of H3 Lys-9 methylation in PEV.
The phenomenon of PEV in Drosophila provided a critical entry point for beginning to dissect the role of histone lysine methylation in heterochromatin formation and maintenance (34, 35). PEV is a gene-silencing event that results from chromosomal rearrangements such as inversion. As a consequence, a euchromatic gene is brought near the pericentric heterochromatin and becomes silenced due to a change in chromatin structure to heterochromatin. The fact that not all Drosophila cells inactivate a euchromatic gene that has been newly juxtaposed to pericentric heterochromatin suggests that spreading of heterochromatin is a dynamic and regulated process. In support of this, the genetic evidence for many other genes that serve as modifiers of PEV further supports that conclusion. Modifiers of PEV can enhance or suppress gene inactivation of the repositioned euchromatic gene. Such modifier genes were identified by selecting for random mutations that affect PEV. Mutations in Su(var) genes, including Su(var)3–9 and su(var)2–5/HP1, suppress PEV and thus reduce gene inactivation normally associated with juxtaposing a euchromatic gene with heterochromatin (89). This implies then that the wild-type Su(var) proteins contribute to gene inactivation by PEV. Su(var)3–9 encodes a Lys-9 HMT that is homologous to the mammalian Suv39h1 and Suv39h2 proteins (29). Su(var)2–5/HP1 encodes the HP1 protein that is known to be involved in establishing heterochromatin (89, 90). HP1 contains a chromodomain and a related chromoshadow domain. A breakthrough in understanding of the mechanism of heterochromatin formation resulted from the demonstration that the chromodomain of HP1 can specifically recognize methylated Lys-9 of histone H3 (91, 92), indicating that a Lys-9 HMT and HP1 are mechanistically linked and act together to establish heterochromatin.

ii. Different HMTs for H3 Lys-9 methylation in euchromatin and heterochromatin.
Different degrees of Lys-9 methylation correlate with distinct chromatin regions, and it appears that the function and regulation of Lys-9 methylation is different in heterochromatin vs. euchromatin (93, 94, 95). In constitutive pericentric heterochromatin, Suv39h1/2 mediates trimethylation of H3 Lys-9, whereas in euchromatin, the HMT G9a mediates dimethylation of H3 Lys-9 in vivo (93, 94). It is interesting to note that in vitro, both Suv39h1 and G9a can convert a histone H3 peptide with dimethylated Lys-9 to the trimethyl form, whereas in vivo they display very different characteristics. In Suv39h1/2 double-null mouse embryo fibroblasts, trimethylation of H3 Lys-9 is abolished, whereas mono- and dimethylation were not significantly affected. In contrast, in G9a null mouse embryo fibroblasts, there was a complete disappearance of dimethylation of H3 Lys-9, a significant decrease in monomethylation, and no change at the trimethylation level. In addition, trimethylation of Lys-9 is a property of pericentric heterochromatin, whereas dimethylation is dispersed throughout the euchromatin, suggesting that mono-, di-, and trimethylation at Lys-9 are differentially regulated and can exert different functional outcomes, as with Lys-4 methylation. Thus, Suv39h1/2 is the major Lys-9 trimethylase in pericentric heterochromatin, and G9a is the major H3 Lys-9 dimethylase in euchromatin.

These two enzymes also display different chromosomal localization patterns, implying different modes of recruitment. Whereas Suv39h1 is associated with pericentric heterochromatin and colocalizes with HP1/ß, G9a is associated with euchromatin and does not colocalize with HP1/ß (96, 97). Whereas Suv39h1 is recruited to heterochromatin through its N-terminal chromodomain (98), G9a lacks a chromodomain; instead, G9a contains an ankyrin-repeat domain, which is also implicated in protein-protein interactions (96) and thus may play an important role in G9a targeting. In addition, whereas Suv39h1 is inhibited by H3 Lys-4 dimethylation in vitro, G9a is not (65). It is also interesting to note that some H3 Lys-4 HMTs, such as Set7/Set9 and MLL/ALL1, also are not inhibited by H3 Lys-9 dimethylation in vitro (62, 65). Because this implies that both modifications can coexist on the same histone tail, it is currently unclear how the antagonism between H3 Lys-4 methylation (which is generally thought to be associated with active or potentially active chromatin) and H3 Lys-9 methylation (which is generally thought to be associated with inactive chromatin or heterochromatin) is resolved.

iii. Recruitment of Suv39h1 HMT by short heterochromatic RNAs (shRNAs) in formation of pericentric heterochromatin.
How does the Suv39h1 HMT recognize which regions of cellular chromatin to methylate? Genetic evidence indicated that Suv39h1 lies upstream of HP1 action (91, 98), but how Suv39h1 is targeted to assemble heterochromatin was unclear until a surprising discovery implicated repetitive DNA elements and RNA interference (RNAi) machinery in recruiting Clr4 (the S. pombe equivalent of Suv39h1) to the centromeric heterochromatic region of S. pombe (99, 100, 101). Centromeric repeats are transcribed bidirectionally to produce noncoding double-stranded RNA, which is processed to small interfering RNA (siRNA, also called short heterochromatic RNA or shRNA) by the RNAi machinery (99). Deletion of any of the three components of RNAi machinery [RNAseIII helicase dicer (dic1), RNA-dependent RNA polymerase (rdp1), and Argonaute (ago1)] caused inappropriate activation of a reporter gene integrated within centromeric heterochromatin. In addition, loss of centromeric localization of Swi6 (the S. pombe equivalent of HP1) and H3 Lys-9-dimethylation were observed, along with increased H3 Lys-4 methylation of the centromeric region. These observations suggest that shRNA in heterochromatic regions helps to recruit Clr4, which establishes Lys-9 methylation that then recruits Swi6. Consistent with the knowledge that histone deacetylases facilitate the initial stages of assembly of heterochromatin, Clr3, which deacetylates H3 Lys-14, was partially required for the H3 Lys-9 methylation and recruitment of Swi6 to the centromere (98). Evidence that heterochromatic shRNA can directly recruit Suv39h1 is currently lacking. However, two different mutations in the chromodomain of Clr4 lead to loss of H3 Lys-9 methylation and Swi6 localization in vivo, whereas in vitro they do not affect Clr4 methyltransferase activity (98). One possible explanation for these findings is that the chromodomain of Suv39h1 may be important in the recognition of shRNA. The chromodomains of two other proteins (histone acetyltransferase MOF and male-specific lethal protein MSL3) have been shown to have RNA binding activity (102, 103).

Once HP1/Swi6 has been recruited to initiate heterochromatin formation, it also initiates the spreading of heterochromatin by self-association with other HP1 molecules and by using its chromoshadow domain to recruit additional Suv39h1, which further catalyzes Lys-9 methylation to attract more HP1, and so forth (37). How these events lead to gene silencing and how the spreading of heterochromatin is regulated are also important questions for future investigation. The role RNA plays in the formation of centromeric heterochromatin in S. pombe also seems likely to be true in higher eukaryotes. In permeabilized human cells, dimethylation of histone H3 at Lys-9 and recruitment of HP1 in pericentric heterochromatin were abolished by RNase treatment (95). Furthermore, addition of total or nuclear RNA restored the methylation and HP1 localization pattern, suggesting that RNA is an important component of the pericentric heterochromatin. The identity of the RNA has not been determined but is presumably similar to heterochromatic shRNA observed in S. pombe.

iv. Recruitment of H3 Lys-9 HMTs to euchromatic promoters by sequence-specific DNA-binding transcriptional repressor proteins.
Although methylation by Suv39h1 has been primarily associated with the establishment and maintenance of heterochromatin, there are examples of its involvement in repressing genes in mammalian euchromatin. The retinoblastoma (Rb) protein is part of a corepressor complex that binds to the E2F transcription factors to repress transcription of genes required for cell cycle progression. Phosphorylation of Rb at a particular stage of the cell cycle causes Rb to dissociate from E2F and thus allows cell cycle progression. The Rb corepressor complex contains histone deacetylases and also Suv39h1. Suv39h1 methylation of Lys-9 of histone H3 results in recruitment of HP1 to the cyclin E gene promoter and represses its transcription (Fig. 2) (104, 105). Similarly, KRAB-ZFP, which is a DNA sequence-specific transcriptional repressor protein, recruits the KAP1 corepressor that brings the H3 Lys-9 HMT SETDB1/ESET to promoters of specific genes and results in transcriptional silencing due to histone H3 Lys-9 methylation and HP1 deposition (106). Similarly, G9a is specifically targeted to the promoter of the interferon ß gene by the DNA-binding PRDI-BF1 repressor protein (107).

ESET, a SET domain protein that associates with an ets-related-gene transcription factor, is regulated by mAM, a murine activating transcription factor-associated modulator, which was biochemically identified as a protein tightly associated with ESET (108, 109). mAM increased the enzymatic activity of ESET and also changed its substrate specificity so that it produced trimethyllysine instead of only dimethyllysine at Lys-9 of histone H3. Furthermore, in transcription assays using chromatin templates, trimethylation of H3 Lys-9 at the promoter region by mAM/ESET leads to transcriptional repression, whereas dimethylation of H3 Lys-9 by ESET alone has only a modest effect (109). These findings suggest possible mechanisms for regulating the number of methyl groups on a specific histone lysine residue by regulating cellular levels and promoter localization of specific HMTs, or availability of HMT-associated proteins that modulate HMT activity. It also suggests putative mechanisms for physiological regulation of the activity and substrate specificity of at least some HMTs by intracellular signaling or cell context.

b. Activation: possible roles for H3 Lys-9 methylation in specific cases of transcriptional activation?
Although most H3 Lys-9 methylation appears to be involved in heterochromatin formation and gene repression, a few observations hint at possible selective involvement in gene-specific transcriptional activation. Ash1, a member of the trithorax group in Drosophila, is an unusual HMT because it can methylate histone H3 at Lys-4 and Lys-9 and histone H4 at Lys-20 in vitro (110); however, in vivo Ash1 is responsible for the majority of H3 Lys-4 methylation but not for the majority of H3 Lys-9 and H4 Lys-20 methylation (111). It is still possible, though, that Ash1 can mediate methylation of H3 Lys-9 and H4 Lys-20 at specific loci in vivo. Chromatin immunoprecipitation experiments demonstrated that dimethylation of Lys-4 and Lys-9 of histone H3 and Lys-20 of histone H4 is linked with transcriptional activation of Ash1 target genes, both an integrated reporter gene and the endogenous Ultrabithorax (Ubx) gene (110). HP1, on the other hand, was not present on the promoters of the integrated reporter gene or the endogenous Ubx when they were active. These observations demonstrate that dimethylation of H3 Lys-9 can be associated with activated transcription in the presence of H3 Lys-4 and H4 Lys-20 methylation, suggesting the possible importance of context specificity in dimethylated Lys-9 for its biological consequences.

Like H3 Lys-9 methylation, HP1 is usually associated with inactive chromatin. However, evidence in Drosophila indicates that HP1 is associated with active genes in at least a few specific cases. In larval salivary gland polytene chromosomes, HP1 is associated with ecdysone and heat-shock induced puffs of euchromatin, in addition to its localization at the heterochromatic regions (112, 113). In chromatin immunoprecipitation analysis of cultured Drosophila S2 cells, heat-shock treatment resulted in recruitment of HP1 to the coding region of the Hsp70 gene, but not the promoter. Whether H3 Lys-9 methylation is responsible for the recruitment of HP1 in these cases remains to be determined.

5. H3 Lys-27 methylation
a. Repression.
Methylation of histone H3 at lysine 27 displays two functional similarities to that of lysine 9. First, different degrees of methylation of both marks have different distributions in the chromatin. Monomethylation of Lys-27 is found in pericentric heterochromatin along with trimethylation of Lys-9 (93, 94) (Fig. 3). On the other hand, trimethylation of Lys-27 is characteristic of facultative heterochromatin of the inactive X chromosome during the initial stage of X inactivation (114, 115). The inactive X chromosome also displays dimethylated but not trimethylated Lys-9 (114, 116, 117, 118). There are specific functional consequences associated with the number of methyl groups on a specific histone lysine residue. HP1 colocalizes with Lys-9 trimethylation in pericentric heterochromatin, but not with dimethylated H3 Lys-9 in the inactive X chromosome (95, 116, 117). In addition, Suv39h double-null mouse embryonic fibroblasts still maintain Lys-9 dimethylation of inactive X, indicating that a different HMT is responsible for the methylation of the inactive X. EZH2, a mammalian homolog of Drosophila enhancer of zeste [E(z)] is the HMT that mediates methylation of H3 Lys-27 of inactive X chromosome; this enzyme also methylates H3 Lys-9 in vitro, but whether it does so on the inactive X chromosome in vivo is not clear (119, 120, 121, 122). E(z) belongs to the polycomb (Pc) group of proteins that are involved in long-term silencing of Hox genes. The Pc group contains two major complexes, ESC-E(z) and Pc repressive complex-1 (PRC1).

The second similarity is that both modifications create binding sites for the recruitment of specific effector proteins that contain chromodomains. Although Suv39h1-mediated trimethylation of histone H3 Lys-9 leads to recruitment of HP1 in mammals, ESC-E(z) complex mediated methylation of histone H3 Lys-27 creates a specific binding site for the recruitment of the PRC1 via Pc protein in Drosophila (123, 124) (Fig. 2). The chromodomain of Pc protein specifically recognizes trimethylated Lys-27 of histone H3.

Two different mechanisms exist for recruiting H3 Lys-27 HMTs to their targets. At the global level, EED-EZH2, the human ESC-E(z) complex, is recruited to inactive X chromosome via Xist RNA to trimethylate histone H3 at Lys-27; this is similar to the mechanism by which centromeric shRNA recruits Clr4 (equivalent of human Suv39h1) to heterochromatin in fission yeast (Fig. 2) (114, 115). Interestingly, recruitment of EED-EZH2 and trimethylation of H3 Lys-27 are transient, occurring only during the initial stage of X inactivation. At the individual gene level, Drosophila ESC-E(z) complex is targeted to Pc response elements via many DNA binding proteins such as GAGA factor, pleiohomeotic (Pho), and Zeste (119, 125, 126, 127, 128). For example, E(z), Pc and dimethylation of H3 Lys-27 are all targeted to the region of the Pc response element of the Drosophila Ubx gene to repress transcription (122). Furthermore, mutations in the SET domain of E(z) led to loss of repression of Ubx in wing imaginal discs in vivo, demonstrating that HMT activity is important for Pc group-mediated gene silencing (120).

6. H4 Lys-20 methylation.
Methylation of histone H4 at lysine 20 is mediated by the PR-Set7/Set8 HMT (26, 27). In Drosophila polytene chromosomes, this modification is associated with the chromocenter and euchromatic arms. Staining in the euchromatin does not significantly colocalize with dimethylated Lys-4 of histone H3, suggesting a role in the silent domains of euchromatin.

D. Consequences of histone lysine methylation
It is clear that histones, especially the N-terminal tails, provide a platform for convergence and integration of signals into a distinct biological outcome, in this case regulation of transcription. How does a methyl mark then lead to a distinct biological outcome? There are three possible mechanisms that may not be mutually exclusive. First, methylation of lysine residues could inhibit binding of proteins (or other nucleosomes) to histone tails or inhibit binding or activities of enzymes that make additional modifications of histone tails or other proteins. Second, methylation could create a binding site for the recruitment of specific proteins involved in chromatin remodeling or enhance activities of certain enzymes to make additional histone modifications. Third, histone methylation in strategic regions of the nucleosome could affect nucleosome conformation and thus its interaction with other proteins or nucleosomes. A few specific examples illustrating these mechanisms are summarized here (Figs. 2 and 4). H3 Lys-4 dimethylation, for example, inhibits binding of mammalian NuRD corepressor complex, which contains histone deacetylase and an ATP-dependent chromatin remodeling activity (79). In contrast, Lys-4 methylation enhances mammalian p300 HAT activity and stimulates the recruitment of coactivator Isw1p ATPases in budding yeast (64, 65, 79, 129), although the interaction is apparently not direct and the domain/protein that can recognize methylated H3 Lys-4 is still not clear. H3 Lys-4 methylation, in addition, inhibits enzymatic activity of the H3 Lys-9 HMT Suv39h1 in mammals by serving as a poor substrate (65). H3 Lys-9 and Lys-27 methylation, on the other hand, creates a high affinity binding site for the chromodomains of Drosophila HP1 and Pc, respectively (123, 124). Although Lys-9 and Lys-27 of histone H3 are both embedded in the same local sequence context, Ala-Arg-Lys-Ser, HP1 and Pc can discriminate between the two methylated lysines by reading the residues preceding the Ala-Arg-Lys-Ser sequence. HP1 primarily reads residues n–1 to n–3, where "n" is the methylated lysine. In contrast, Pc reads residues n–4 through n–7, contributing to differential protein recognition. As a consequence, Pc binds 25 times better to trimethylated Lys-27 than to trimethylated Lys-9. HP1, on the other hand, binds 13 times better to trimethylated Lys-9 than to Lys-27. Clearly, the identification of additional proteins that recognize specific sites of histone methylation will extend our understanding of the histone code and how it regulates transcription.

E. Lysine methylation of a basal transcription factor
SET9, which was originally characterized as a H3 Lys-4 HMT, was recently shown to methylate TAF10 (TAF_II30), a subunit of the TFIID complex (130). TFIID, which consists of the TATA box binding protein TBP and about 10 TBP associated factors or TAFs of varying sizes, is the basal transcription factor that binds TATA boxes and thus sets the transcription start site for TATA box-containing genes and helps to assemble the other basal transcription factors and RNA pol II on the promoter. Lysine methylation of TAF10 increases its affinity for RNA pol II. Furthermore, loss of SET9 methyltransferase activity or mutation of the TAF10 methylation site caused a reduction in transcriptional activation of specific transient reporter genes as well as specific endogenous genes (130). Because very few proteins have actually been examined for lysine methylation at this point, this report is presumably one of the first of what will be many reports in the next few years of lysine methylation of nonhistone proteins.

	III. Arginine Methylation

Top Abstract I. Introduction II. Lysine Methylation III. Arginine Methylation IV. Conclusions and Unanswered... References

 
A. Enzymes
1. The protein arginine methyltransferase (PRMT) family.
Methylation of arginine residues is a common posttranslational modification in eukaryotes. Two types of PRMTs transfer the methyl group from AdoMet to the guanidino group of arginines in protein substrates. Type I PRMT enzymes form monomethylarginine and asymmetric dimethylarginine products. Type II PRMT enzymes catalyze the formation of monomethylarginine and symmetric dimethylarginine (131, 132) (Fig. 1C). PRMTs may be universal to all eukaryotes, because one or more representatives are found in fungi, higher plants, invertebrates, and vertebrate animals (133). Seven mammalian PRMT genes have been identified: PRMT1, PRMT2, PRMT3, CARM1/PRMT4, JBP1/PRMT5, PRMT6, and PRMT7 (Fig. 5); but the yeast S. cerevisiae has only one member, Hmt1/Rmt1. PRMT5 is the only example of a type II enzyme, whereas the other PRMTs (except PRMT7) are all type I enzymes. PRMT 7 makes only monomethylarginine and contains two methyltransferase domains in a single polypeptide chain (134) and thus may represent a third class of PRMT.

View larger version (11K): [in this window] [in a new window]   FIG. 5. The PRMT family of proteins. The catalytic core regions common to all PRMTs consist of a highly conserved AdoMet binding region formed by a Rossman fold and two helices (orange), and a less conserved ß-barrel structure (yellow) that folds against the AdoMet binding region to form the protein substrate binding cleft. Some family members contain additional recognizable motifs such as zinc fingers and SH3 domains. PRMT7 contains two catalytic domains within the same polypeptide chain; the two halves of the protein are aligned separately in the figure (PRMT7 N-terminal and PRMT7 C-terminal).

2. Structure, catalytic mechanism, and substrate specificity
a. Protein substrate specificity.
PRMT proteins vary in length from 348 (S. cerevisiae Rmt1) to 637 (PRMT5) amino acids; PRMT7, with duplicate methyltransferase domains, has 692 amino acids. The methyltransferase activity resides in a highly conserved core region of approximately 310 amino acids. In addition to the conserved methyltransferase domain, each PRMT member has a unique N-terminal region that varies widely in length, and CARM1 also has a unique C terminus (Fig. 5). Despite the high degree of conservation within the methyltransferase domain, there is relatively little overlap in the protein substrate specificities of the seven PRMTs (11, 145). The mechanism of specific substrate recognition is still unclear, and in fact based on the known substrates, few clear consensus recognition sequences have emerged. For example, PRMT1 methylates many substrates in regions containing Arg-Gly-Gly repeats (146), but not all PRMT substrates have such sequences (147). CARM1 also appears to have more than one type of sequence immediately surrounding the methylation sites of its substrates (148). Although the crystal structures do not fully explain substrate specificity, they may provide some clues about this mystery and also provide insight to the general mechanism of catalysis by these enzymes. Structures of the conserved core region have been published for Hmt1 (149), PRMT3 (150), and PRMT1 (133), in some cases in complex with S-adenosylhomocysteine (AdoHcy; the product remaining after a methyl group is extracted from AdoMet) and/or some substrate peptides. On the surface of PRMT1 is a long, meandering but possibly continuous groove which provides multiple sites that can bind substrate peptides. Thus, it is conceivable that these multiple sites could accommodate several different types of peptide substrates and thus could explain the apparent ability of a single enzyme to recognize multiple consensus sequences, which could be at varying distances from the actual site of methylation.

b. Structure and catalysis.
In all three published structures, the core region consists of two domains folded together into an integral structure. The N-terminal half of the core, consisting of a typical Rossman fold and two -helices, is the AdoMet-binding domain, the most highly conserved region among PRMTs and also partially conserved in other types of AdoMet-dependent methyltransferases (150). The C-terminal half of the core forms a barrel-like structure, unique to the PRMT family, which folds against the N-terminal AdoMet-binding region. The resulting cleft provides a protein substrate binding site and the site of catalysis. The three-dimensional structure analysis also reveals a dimerization interface that is essential for the enzymatic activity. Indeed, dimer formation encloses the active sites into a hole between the two monomers. Mutation of dimer contact residues of Hmt1 (149) or deletion of the PRMT1 dimerization arm (133) eliminates the enzymatic activity. The dimerization may contribute to the formation of dimethylarginine by facilitating transfer of the monomethylated substrate from the active site of one monomer directly into the active site of the second monomer, adding the second methyl group without dissociation of the protein substrate from the dimeric enzyme. Some PRMTs also have the ability to form larger homomeric aggregates. Hmt1 (149) and PRMT1 (142) form hexamers in solution, and PRMT5 also forms homooligomers of more than two subunits (144). The role of the multimerization is not clear, although a PRMT5 multimer complex shows a high enzymatic activity, suggesting that the formation of homooligomers is important for the efficient catalytic activity of PRMT5 (144). The residues implicated in AdoMet binding, in catalysis, in intramolecular contacts between the AdoMet-binding region and the barrel-like domain, and in the putative dimer interface are conserved across the PRMT family, suggesting a common fold and catalytic mechanism (149, 150).

c. Functions of unique N- and C-terminal regions of PRMTs.
Whereas the functions of the unique N-terminal and C-terminal regions of PRMTs are still not clear, deletion analyses have provided some clues to their functions. Deletion of the N terminus of Hmt1 impairs the stability of the hexamer and decreases the methyltransferase activity, suggesting that multimer formation contributes to enzymatic activity (149). Similarly, an interaction between the N and C termini of PRMT5 appears to be involved in its homomeric complex formation (144). Deletion of the N-terminal part of PRMT3, which contains a zinc-finger motif, impairs its enzymatic activity and possibly alters protein substrate specificity, suggesting a role in protein substrate recognition (142, 151). The unique C terminus of CARM1, which contains a strong autonomous transcriptional activation activity, is required along with the methyltransferase activity for the transcriptional coactivator function of CARM1 (see Section III.B), and thus may interact with other components of the transcription machinery that participate in the process of transcriptional activation (152).

B. Effects of arginine methylation on protein function
For the most part, the specific changes in protein function resulting from protein arginine methylation have yet to be determined. A few specific cases where such information is available will be discussed below. More extensive previous studies on the functional effects of other protein modifications, such as phosphorylation, can provide models to guide our thinking about the functional ramifications and potential physiological roles of protein methylation and can suggest experimental strategies to test such ideas. While methylation should not affect the overall charge of an arginine residue, it can be expected to add bulk and hydrophobicity that can promote or inhibit intra- or intermolecular interactions (with proteins or other types of molecules) (Fig. 1C). Such altered interactions may change the shape and thus the function or stability of the methylated protein, or may serve to facilitate or interfere with intermolecular (e.g., protein-protein or protein-RNA) interactions or enzymatic activities that play important roles in specific signaling pathways.

1. Implication of arginine methylation in transcriptional regulation.
The arginine methyltransferases modify proteins that function at many different steps in cellular regulation, including cytoplasmic and nuclear signal transduction pathways, nuclear-cytoplasmic shuttling, transcriptional activation, and multiple posttranscriptional steps in gene expression. This section will focus on methylation of histones and nonhistone proteins involved in transcriptional regulation, followed by brief examples of protein methylation that may regulate posttranscriptional steps of gene expression. For perspective, subsequent sections will briefly discuss evidence implicating arginine-specific protein methylation in some cytoplasmic signaling pathways.

a. Histone methylation and chromatin remodeling
   i. PRMTs as transcriptional coactivators for nuclear receptors.
Activation of transcription of a specific protein-encoding gene generally involves binding of one or more transcriptional activator proteins to specific enhancer elements associated with the gene. The DNA-bound transcriptional activator protein recruits a variety of coactivator proteins that remodel chromatin in the promoter region into a more open conformation and recruit and activate RNA pol II and the rest of the basal transcription machinery. The large number of coactivators that are apparently involved suggests that the processes of chromatin remodeling and transcriptional activation are exceedingly complex and subject to regulation at many different steps and from many different interacting signaling pathways. Many of these coactivators catalyze a variety of posttranslational protein modifications, and in fact a dauntingly large number and variety of posttranslational modifications of histones and nonhistone proteins have been implicated in transcriptional regulation (14, 153). Studies with the nuclear receptor family of hormone-regulated transcriptional activator proteins (154, 155) have played a central role in elucidating the role of coactivators in general and, in particular, protein arginine methylation in transcriptional activation. Binding of the appropriate hormone to nuclear receptors causes a conformational change that facilitates binding of many coactivators and thus presumably allows the DNA-bound receptor to recruit key coactivators to the promoter. Of the many coactivators that have been implicated in nuclear receptor function (156, 157), only a few have been functionally well characterized (Fig. 6). For example, the SWI/SNF complex possesses ATP-dependent chromatin remodeling activity; histone acetyltransferases such as CREB-binding protein (CBP), p300, and p300/CBP-associated factor (pCAF) contribute to chromatin remodeling through a different mechanism; the TRAP/DRIP/mediator complex associates with and is apparently involved in the recruitment and activation of RNA pol II. Coactivators of the p160 family (SRC-1, GRIP1/TIF2, and pCIP/ACTR/RAC3/AIB1/TRAM1) bind only to the hormone activated form of the nuclear receptors and recruit additional coactivators, called secondary coactivators, including the protein/histone acetyltransferases CBP and p300 (158, 159) and the protein/histone arginine methyltransferases CARM1 and PRMT1 (6, 160). The discovery of CARM1 as a histone methyltransferase and a secondary coactivator for nuclear receptors (6) constituted the first implication of protein methylation in transcriptional activation. CARM1 methylates histone H3 at Arg-2, -17, and -26 (161) (Fig. 1A) and enhances transcriptional activation by nuclear receptors in transient transfection assays (6); both the methyltransferase activity and the association with p160 coactivators are essential for CARM1 coactivator function with nuclear receptors (162, 163). Moreover, in chromatin immunoprecipitation assays, CARM1 itself and methylation of histone H3 at Arg-17 were associated with the hormone-inducible promoters of stably integrated reporter genes and endogenous (i.e., native) genes in a hormone-dependent manner (164, 165). Thus, recruitment of CARM1 and arginine methylation of histone H3 are integral parts of the transcriptional activation process.

View larger version (44K): [in this window] [in a new window]   FIG. 6. Role of PRMTs in transcriptional activation by nuclear receptors. Hormone-activated nuclear receptors (NR) bind to a specific hormone-responsive enhancer element (HRE) sequence in DNA and recruit various coactivator complexes, which help to remodel chromatin structure and recruit and activate RNA pol II complex and its associated basal transcription machinery (including basal transcription factors TFIIB and TBP; TBP binds to TATA boxes in the basal promoter region). The SWI/SNF coactivator complex carries out ATP-dependent chromatin remodeling. The p160 coactivator complex is anchored to the nuclear receptors through the p160 protein (GRIP1 or SRC-1) and also contains histone acetyltransferases (CBP or p300 and pCAF) and histone-methylating PRMTs (PRMT1 and CARM1). CARM1 has also been found to associate with the SWI/SNF complex and thus may coordinate the activities of these two complexes. The TRAP/DRIP (also called mediator) complex helps to recruit and activate RNA pol II by association with the CTD of RNA pol II. Double-headed black arrows, recruitment of coactivators; single-headed black arrows, recruitment of basal transcription machinery by coactivators; red arrows, enzymatic modifications of histones and other proteins.

   ii. Functional interactions of histone acetylation and histone arginine methylation.
The fact that arginine methylation of histones occurs in the N-terminal tails among the sites for lysine methylation, lysine acetylation, and serine phosphorylation (Fig. 1A) strongly suggests functional relationships among histone modifications (1, 12, 170, 171). In fact, multiple coactivators with histone modifying activities were found to cooperate synergistically in transient transfection assays: CARM1 (arginine methylation of histone H3) cooperated with PRMT1 (arginine methylation of histone H4) (160); CARM1 (but not PRMT1, PRMT2, or PRMT3) cooperated with p300, CBP, and pCAF (acetylation of histones H3 and H4) (163). Both the PRMTs and p300/CBP are apparently recruited as secondary coactivators for nuclear receptors by interaction with a p160 protein. One (but by no means the only) possible explanation for the cooperative coactivator functions of these histone-modifying enzymes is that some of the histone modifications may be facilitated by the prior occurrence of the others. Indeed, methylation of free histone H4 by PRMT1 stimulated the subsequent acetylation of this histone by p300 (167) and acetylation of nucleosomes in p53-dependent transcription in vitro (172). Similarly, prior acetylation by p300 enhanced CARM1 binding and enzymatic activity on histone H3 tail peptides (173) and on assembled chromatin templates (172). On the contrary, PRMT5, when associated with the corepressor complex mSin3/histone deacetylase 2 and Brg1 (the hSWI/SNF ATPase subunit) could methylate hypoacetylated histones H3 and H4 more efficiently than hyperacetylated histones H3 and H4. In this case, PRMT5 is recruited to c-Myc target genes with the other components of the complex and appears to be involved in gene repression (174).

Thus, arginine-specific histone methylation is a part of the transcriptional activation process and occurs in cooperation with other histone modifications. Proof that these histone methylation events actually play important roles in transcriptional activation has been provided recently from in vitro transcription experiments using recombinant chromatin templates (172). The transcriptional activator p53 binds directly to p300, CARM1, and PRMT1 and recruits them to the target promoter, where they make the appropriate histone modifications in an ordered fashion (PRMT1, p300, CARM1; see above) and cooperate synergistically as coactivators. Moreover, reconstitution of the chromatin template with histones containing mutations that prevent acetylation or methylation by one of these enzymes abolishes the ability of that enzyme (but not the other two enzymes) to enhance p53-mediated transcription. This work confirms that arginine methylation of histones not only occurs, but is important for transcriptional activation.

   iii. PRMTs as coactivators for diverse types of transcriptional activator proteins.
Although much of the evidence for the involvement of protein