Report

Role of Histone H3 Lysine 9 Methylation in Epigenetic Control of Heterochromatin Assembly

See allHide authors and affiliations

Science  06 Apr 2001:
Vol. 292, Issue 5514, pp. 110-113
DOI: 10.1126/science.1060118

Abstract

The assembly of higher order chromatin structures has been linked to the covalent modifications of histone tails. We provide in vivo evidence that lysine 9 of histone H3 (H3 Lys9) is preferentially methylated by the Clr4 protein at heterochromatin-associated regions in fission yeast. Both the conserved chromo- and SET domains of Clr4 are required for H3 Lys9methylation in vivo. Localization of Swi6, a homolog ofDrosophila HP1, to heterochomatic regions is dependent on H3 Lys9 methylation. Moreover, an H3-specific deacetylase Clr3 and a β-propeller domain protein Rik1 are required for H3 Lys9 methylation by Clr4 and Swi6 localization. These data define a conserved pathway wherein sequential histone modifications establish a “histone code” essential for the epigenetic inheritance of heterochromatin assembly.

The organization of the higher order chromatin structure has been linked to the posttranslational modifications of histone tails, including acetylation, phosphorylation, and methylation (1). It has been suggested that distinct combinations of covalent histone modifications, also referred to as the “histone code,” provide a “mark” on the histone tails to recruit downstream chromatin-modifying proteins (2, 3). This is best illustrated by recent studies indicating that the conserved bromodomain of several transcriptional coactivators bind specifically to acetylated lysine residues on histone tails (4, 5). The mechanisms responsible for the establishment and maintenance of multiple covalent modifications within the same or different histone tail are not fully understood.

Modifications of histone tails have also been linked to heterochromatin assembly. Histones H3 and H4 are largely hypoacetylated in heterochromatic chromosomal regions in organisms as diverse as yeast, flies, and mammals (6–8). In fission yeast, hypoacetylation of histones is associated with the silent mating-type region and centromeres (9, 10), chromosomal domains that share many parallels with heterochromatic regions in higher eukaryotes (11). Centromeric regions comprising a central core of unique sequences surrounded by inner (imr) and outer (otr) repeats are assembled into silenced chromatin structures (12). Similarly, a large ∼15-kb chromosomal domain at the mating-type (mat2/3) region, including themat2 and mat3 loci and an interval between them, known as the K-region, is maintained in a silent epigenetic state (13, 14). Among the trans-acting factors that affect silencing at these regions, Clr3 and Clr6 belong to family of histone deacetylases (HDACs) (15). Swi6 and Clr4 proteins contain a chromodomain, an evolutionarily conserved motif initially identified in HP1 and Polycomb proteins (16–18). Recently, both Clr4 and its mammalian counterpart, SUV39H1, have been shown to have intrinsic histone H3-specific methyltransferase (HMTase) activity in vitro (19). However, it is not known whether histones are the physiological targets of these methyltransferases in vivo.

Consistent with previous findings, recombinant Clr4 (rClr4) was found to contain HMTase activity exclusively for histone H3 (Fig. 1B). To identify the specific residue of H3 methylated by rClr4, we used synthetic peptides derived from the NH2-terminus of H3 as substrates in an in vitro HMTase assay (20, 21). Clr4 preferentially methylated the H3 1-20 unmodified peptide but failed to methylate the H3 19-35 unmodified peptide (Fig. 1C), indicating that the target residue of Clr4 HMTase resides in the first 20 amino acids of H3. To determine this target residue, we developed a synthetic H3 1-20 peptide set that contained covalent modifications on different amino acids. With these peptides as substrates, only acetyl or methyl modifications on Lys9 effectively blocked rClr4 HMTase activity, indicating that Clr4, like its mammalian homolog SUV39H1 (19), selectively methylates Lys9 of H3. Furthermore, similar to SUV39H1, rClr4 HMTase activity was inhibited by phosphorylation of serine 10 (Fig. 1C). These results demonstrate that enzymatic features of the Su(var)3-9 protein family are evolutionarily conserved from fission yeast to humans.

Figure 1

Clr4 selectively methylates lysine 9 of histone H3. (A) Schematic representation of the Clr4 protein. Conserved domains and locations of the five mutations are shown. aa, amino acid. (B) Effect of mutation in Clr4 on its HMTase activity in vitro. rClr4 wild-type and mutant proteins were analyzed in an in vitro HMTase assay with chicken core histones as the substrate andS-adenosyl-[methyl-3H]-methionine as the methyl donor. Samples were separated by 15% SDS-PAGE and visualized by Coomassie staining (bottom). Fluorography indicates the relative HMTase activity of each of the rClr4 proteins (top). The asterisk (*) indicates a histone H3 proteolytic breakdown product. (C) rClr4 was used in an in vitro HMTase assay with the following covalently modified peptides from the first 20 amino acids of the NH2-terminal tail of H3 as substrates: unmodified, dimethylated at lysine 4, dimethylated at lysine 9, acetylated at lysine 9, acetylated at lysine 14, or phosphorylated at serine 10. An unmodified H3 19-35 was also assayed.

A recent study demonstrated that the conserved SET domain and two flanking cysteine-rich regions were required for SUV39H1 HMTase activity in vitro (19). To determine whether the conserved domains, the chromo, SET, and cysteine-rich regions, were also critical for Clr4 HMTase activity, we tested mutant Clr4 proteins for HMTase activity. Although mutations in the chromodomain [Trp31→ Gly (W31G) and Trp41 → Gly (W41G)] had little effect on Clr4 HMTase activity, mutations in the SET domain [Gly328 → Ser (G378S)] and both cysteine-rich regions [Arg320 → His (R320H) and Gly486 → Asp (G486D)] greatly reduced Clr4 HMTase activity, indicating that these three regions are critical for Clr4 HMTase activity in vitro (Fig. 1, A and B).

To test the hypothetical correlation between H3 Lys9methylation and silencing, we developed an H3 Lys9–methyl specific antibody. In an enzyme-linked immunosorbent assay, the H3 Lys9–methyl antibody specifically recognized the H3 1-20 Lys9–methyl peptide in a wide range of antibody dilution (22). Moreover, the H3 Lys9–methyl antibody did not detect recombinant histone H3 (rH3) alone compared with the HeLa core histone positive control but did detect rH3 selectively methylated by rClr4 (Fig. 2A), further demonstrating the specificity of this antibody. Using this antibody in chromatin immunoprecipitation (ChIP) experiments (10) (Fig. 2B), we found that the H3 Lys9–methyl modification is specifically localized at the silenced chromosomal regions. H3 Lys9methylation and Swi6 were preferentially enriched at a marker gene (Kint2::ura4+ ) inserted within the silenced mat2/3 chromosomal domain, compared with controlura4DS/E locus at the endogenous location (Fig. 2C). Similarly, H3 Lys9 methylation was also preferentially enriched at the ura4+ marker inserted within the highly repressed innermost repeat (imr1R::ura4+ ) and the outer repeat (otr1R::ura4+ ), but not at the weakly repressed central core (cnt1::ura4+ ) of cen1 (Fig. 2D). In addition, H3 Lys9methylation coincided with the presence of Swi6 at these regions (Fig. 2D) (23). These findings suggest that H3 Lys9–methyl modification and Swi6 are preferentially localized to silent chromosomal regions and that Swi6 localization is functionally dependent on H3 Lys9 methylation.

Figure 2

Presence of H3 Lys9 methylation at the mat and cen1. (A) Specificity of the H3 Lys9–methyl antibody. Recombinant histone H3 (rH3) was incubated alone or in the presence of rClr4 (rH3+Clr4) in an HMTase assay. The samples, including the HeLa core histone positive control, were separated by SDS-PAGE, transferred to a polyvinylidene difluoride (PVDF) membrane, and visualized by Ponceau staining (bottom). Western blot analysis with the H3 Lys9–methyl antibody is shown (top). (B) DNA isolated from immunoprecipitated chromatin (ChIP) or whole-cell extracts (WCE) was quantitatively analyzed with a competitive polymerase chain reaction (PCR) strategy, whereby one set of primers amplifies a 694–base pair (bp) and a 426-bp size products fromura4+ at silenced loci and the controlura4DS/E minigene at the endogenous location, respectively. PCR fragments were resolved on polyacrylamide gel and then quantified with phosphoimager. (C) A physical map of the mating-type locus indicating the Kint2::ura4+ insertion site (top). The cross-hatched box indicates a part of theK-region homologous to centromeric repeats. Results from ChIP analysis with the H3 Lys9–methyl (K9-Me) or Swi6 antibody are shown. The ratios of ura4+ and control ura4DS/E signals present in WCE were used to calculate relative precipitated fold enrichment shown underneath each lane. Serial dilution plating assay was used to measureura4+ expression (middle). (D) Schematic representation of cen1 and the correspondingura4+ insertion sites (top). Levels of H3 Lys9 methylation and Swi6 at the three insertion sites were determined by ChIP analyses (bottom). Expression ofura4+ markers was assayed with serial dilution plating assay.

We next sought to determine the biological effect of Clr4 on H3 Lys9 methylation and Swi6 localization at mat2/3region and centromeres. Compared with the relatively high levels of Swi6 and H3 Lys9 methylation at bothKint2::ura4+ andotr1R::ura4+ in wild-type cells, Swi6 and H3 Lys9 methylation were absent in a clr4Δstrain at both loci (Fig. 3, A and B). This result suggests that H3 Lys9 is the physiological target of Clr4 HMTase activity and that Clr4 appears to be the exclusive in vivo H3 Lys9–specific HMTase atmat and cen loci. In comparison with our in vitro result showing that only the SET domain is required for Clr4 HMTase activity (Fig. 1B), we found that both the chromo- and SET domains are required for H3 Lys9 methylation and Swi6 localization in vivo (Fig. 3, C and D). Taken together, these results indicate that the chromodomain is presumably required for targeting Clr4 to themat2/3 region and centromeres, whereas the SET domain and associated cysteine-rich regions of Clr4 constitute the catalytic site.

Figure 3

The effect of clr4 mutations on H3 Lys9 methylation and Swi6 localization at matand cen1. (A and B) Deletion ofclr4 abolishes H3 Lys9 methylation and Swi6 localization at mat and cen1. ChIP analyses were used to compare the levels of H3 Lys9–methyl or Swi6 in wild-type (WT) or Δclr4 strains. (C and D) Results from ChIP analyses to determine effect of mutant alleles on H3 Lys9 methylation and Swi6 localization (top) and silencing phenotypes (bottom) are shown. Mutations in clr4 have varying effects on H3 Lys9 methylation, Swi6 localization, and silencing atKint2::ura4+ andotr1R::ura4+ .

The Swi6 levels at mat and cen in differentclr4 mutant backgrounds were directly correlated with H3 Lys9 methylation levels (Fig. 3, C and D), further suggesting that Swi6 localization at silent chromosomal domains is functionally dependent on H3 Lys9 methylation. The importance of our in vivo analyses was further highlighted by observations that some mutations in Clr4 that decrease its HMTase activity in vitro do not substantially decrease H3 Lys9methylation and Swi6 localization in vivo. In addition, mutations in the SET domain and the NH2-terminal cysteine-rich regions of Clr4 (G378S and R320H) greatly reduce H3 Lys9methylation and Swi6 localization at the mat locus; however, these mutations have moderate or negligible effects at cen1(Fig. 3, C and D). These mutations also have weak effects on centromeric silencing compared with mating-type silencing. Our results are consistent with the notion that enzymatic defects displayed by recombinant monomeric proteins in vitro can be “rescued” by functioning in the context of a multisubunit complex in vivo. Moreover, the functional organization of the mat2/3 region and centromeres may differ, and an additional factor(s) may help promote Clr4 activity at centromeres.

Mutations in the clr3 HDAC, which specifically deacetylates H3 Lys14 (24), affects silencing at mat and cen (12,15). ChIP analysis demonstrated that a clr3-735mutant partially defective in H3 Lys14 HDAC activity (24) displayed a moderate decrease in H3 Lys9 methylation and Swi6 localization atotr1::ura4+ (Fig. 4A), coincident with the apparent reduction in its HDAC activity. This result suggests that H3 Lys14 acetylation inhibits Clr4 HMTase in vivo (22). To further investigate the functional interaction between Clr3 and Clr4, we created a double-mutant strain containing theclr3-735 and clr4R320H mutations, aclr4 mutation that had the least effect on H3 Lys9 methylation atotr1R::ura4+ (Fig. 3D). ChIP analysis of the double mutant demonstrated that H3 Lys9 methylation and Swi6 localization were nearly abolished when compared with the single mutants (Fig. 4A). These findings indicate that Clr3 acts synergistically with Clr4 to effectively localize Swi6 to heterochromatic domains. In other words, deacetylation of H3 Lys14 by Clr3 is required for H3 Lys9methylation by Clr4 and for Swi6 localization either indirectly, by altering Clr4 activity, or directly or both. These data also support the theory that residues neighboring Lys9, and potentially their modification states, play an important role in establishment of the appropriate H3 Lys9–methyl mark.

Figure 4

Functional Clr3 and Rik1, but not Swi6, are required for H3 Lys9 methylation by Clr4. (A) Clr3 and Clr4 act synergistically to silence cen1. Theclr3-735, clr4 R320H, andclr3clr4 double-mutant strains were used for ChIP analyses to determine H3 Lys9 methylation and Swi6 localization atotr1R::ura4+ . (B) A rik1mutation abolishes H3 Lys9 methylation and Swi6 localization at mat and cen1. Arik1-304 mutant strain was used for ChIP analysis to determine H3 Lys9 methylation and Swi6 localization atmat or cen1 loci. (C) A swi6mutation does not affect Clr4 methylation of H3 Lys9. A swi6-115 strain was used for ChIP analysis (top). Western blot analysis with the Swi6 antibody shows Swi6 protein level in wild-type or the swi6-115 mutant strain (bottom).

Previous studies have shown that rik1+ affects silencing as well as Swi6 localization at silent loci (12, 25). Computational analyses revealed that Rik1 contains β-propeller domains typically found within WD-40 repeat proteins and are theorized to participate in protein:protein interactions (26). A mutation in rik1completely abolished H3 Lys9 methylation and Swi6 localization at both mat and cen compared with wild type (Fig. 4B). WD-40 proteins are involved in many aspects of chromatin remodeling and histone metabolism, such as chromatin assembly and acetylation or deacetylation of histones (27). Therefore, the β-propeller domains of Rik1 may form a complex with Clr4 to recruit its HMTase activity to heterochromatic regions and may play a role in coupling other trans-acting factors, such as Swi6 and histone deacetylases.

The possible role of Swi6 on Clr4-dependent methylation of H3 Lys9 was also tested. Strains carrying swi6-115(W269R) mutation that severely reduced Swi6 protein levels were used. As expected, Swi6 localization at both mat andcen was abolished as demonstrated by ChIP analysis (Fig. 4C). The swi6-115 mutation did not cause any detectable change in H3 Lys9 methylation when compared with the wild-type strain. These data indicate that Swi6 is dispensable for Clr4 function and suggest that Swi6 acts downstream of Clr4 H3 Lys9 methylation.

Collectively, the above results allow us to define a temporal sequence of events leading to establishment of the silenced chromatin state with regard to the covalent modifications of the H3 NH2-terminal tail (Fig. 5). We propose that HDACs and HMTases act cooperatively to establish a “histone code” that is then recognized by Swi6. More specifically, we propose that the HDACs, Clr6 and/or Hda1 (11), deacetylate H3 Lys9, whereas Clr3 deacetylates H3 Lys14 before H3 Lys9 methylation by the Clr4/Rik1 HMTase complex. Swi6 binding to the H3 Lys9–methyl modification would then result in self-propagating heterochromatin assembly (10). Because the heterochromatin-binding domain of Swi6 was mapped to its chromodomain (28), it is most likely that this protein motif has evolved to recognize the H3 Lys9–methyl modification.

Figure 5

A stepwise model for heterochromatin assembly leading to epigenetic silencing in Schizoaccharomyces pombe. Green flags and red lollipops represent acetyl and methyl modifications, respectively. A nucleosome core particle is composed of DNA (black line) wrapped around a histone octamer (blue). Orange or green protrusions represent the NH2-terminal tails of histones with or without acetyl modifications, respectively.

It was recently shown that Swi6 remains associated with the mat2/3 region throughout the cell cycle where it acts as an important determinant of the epigenetic cellular memory, promoting inheritance of the silenced state (10). Because the mouse homolog of Swi6, M31, associates with Su(var)3-9 (29), a similar interaction between Clr4 and Swi6 is predicted. The close association of Clr4 enzymatic HMTase activity, followed by recruitment and binding of Swi6 to Lys9 methyl “marks” in H3 through its chromodomain, suggests a pathway of epigenetic inheritance. The extent to which the chromodomain of Clr4 recognizes H3 Lys9–methyl marks is unknown, but it would provide the enzyme a means to bind chromatin as it performs subsequent methylation events. On the basis of the conservation of Clr4/SUV39H1 and Swi6/HP1 proteins and the presence of H3 Lys9–methyl modification in higher eukaryotes (30) (Fig. 2A), we predict that a similar mechanism may be responsible for higher order chromatin assembly in organisms ranging from fission yeast to humans. Considering the parallels between transcriptional repression by Polycombgroup proteins in flies and mammals and silencing in fission yeast, it is likely that histone methylation coupled with histone deacetylation may help localize Polycomb in pathways that lead to the regulation of homeotic gene expression (22).

  • * To whom correspondence should be addressed. E-mail: grewal{at}cshl.org

REFERENCES AND NOTES

View Abstract

Navigate This Article