Report

Transitions in Distinct Histone H3 Methylation Patterns at the Heterochromatin Domain Boundaries

See allHide authors and affiliations

Science  10 Aug 2001:
Vol. 293, Issue 5532, pp. 1150-1155
DOI: 10.1126/science.1064150

Abstract

Eukaryotic genomes are organized into discrete structural and functional chromatin domains. Here, we show that distinct site-specific histone H3 methylation patterns define euchromatic and heterochromatic chromosomal domains within a 47-kilobase region of the mating-type locus in fission yeast. H3 methylated at lysine 9 (H3 Lys9), and its interacting Swi6 protein, are strictly localized to a 20-kilobase silent heterochromatic interval. In contrast, H3 methylated at lysine 4 (H3 Lys4) is specific to the surrounding euchromatic regions. Two inverted repeats flanking the silent interval serve as boundary elements to mark the borders between heterochromatin and euchromatin. Deletions of these boundary elements lead to spreading of H3 Lys9 methylation and Swi6 into neighboring sequences. Furthermore, the H3 Lys9methylation and corresponding heterochromatin-associated complexes prevent H3 Lys4 methylation in the silent domain.

In eukaryotes, chromosomes are partitioned into structurally and functionally distinct regions that help to separate independently regulated parts of the genome (1, 2). Specialized DNA elements, known as insulators or boundary elements, have been suggested to mark the borders between adjacent chromatin domains and to act as barriers against the effects of enhancer and silencer elements from neighboring regions (3–5). The best characterized example of such long-range chromatin effects comes from studies of position effect variegation, where chromosomal rearrangements of heterochromatin near or into areas of euchromatin result in the stable epigenetic silencing of these regions (6–8). The mechanism that prevents the spreading of heterochromatin into euchromatin in a natural chromosomal context is not well understood, but recent studies suggest that boundary elements may be involved (9, 10).

In fission yeast, the mating-type region contains three linked loci called mat1, mat2, and mat3 (Fig. 1). In contrast to the transcriptionally activemat1 locus, which determines the mating type of the cell, the mat2 and mat3 loci and the interval between them, known as the K-region, are subject to heterochromatin-mediated silencing and recombinational suppression (11–13). Although silencing is thought to extend into theL-region, the interval between mat1 andmat2, the exact boundaries between heterochromatin and euchromatin within this region are unknown (14). Recent in vivo evidence shows that the enzymatic methylation of Lys9 of histone H3 by Clr4 is required for localization of Swi6, a homolog of Drosophila HP1, and plays a critical and highly conserved role in heterochromatin assembly (15,16).

Figure 1

High-resolution mapping of Swi6 protein levels at the mating-type region of stable M andP strains. (A) A physical map of the mating-type region is shown (top). IR-L and IR-R inverted repeats residing on both sides of the mat2/3 interval are shown as black arrows. The green box represents a part of the K-region,cenH, sharing homology to centromeric repeats. Red or blue vertical arrows indicate location of marker gene insertions showing repressed or expressed state, respectively (11, 13, 14). ChIP analyses with antibody to Swi6 (7) were used to measure Swi6 levels throughout the mat locus. DNA isolated from immunoprecipitated chromatin (ChIP) or whole-cell crude extracts (WCE) was subjected to multiplex PCR to amplify DNA fragments from the mat locus, indicated by bars (1 to 80), as well as an act1 fragment serving as an internal amplification control. PCR fragments were resolved on a polyacrylamide gel and then quantified using a phosphoimager. (B) Relative precipitated enrichments of mat sequences were calculated; quantitation of these results is plotted in alignment with the map of themat locus. (C) The levels of Swi6 at a centromeric (otr2) repeat in M and Pcells. DNA fractions from ChIP analyses of Swi6 levels at mat were analyzed for enrichment of otr2sequences using multiplex PCR. (D) Comparison of Swi6 protein levels in different mating-type backgrounds, as assessed by Western blot analysis using antibody to Swi6 (top panel) or antibody to TAT1 for loading control (bottom panel) on whole-cell protein extracts from stable M, P, and switching-competent h90 cells orswi6Δ strain.

To determine the exact locations of heterochromatin within themat locus and to address whether boundary elements flank the silent domain, we created a 47-kb high-resolution map by means of chromatin immunoprecipitation (ChIP) (17) with the polymerase chain reaction (PCR). We designed primers to amplify overlapping fragments of ∼0.5 to 0.65 kb from the entiremat region. To amplify a part of the K-region well known to exhibit homology to centromeric repeats (11) (Fig. 1A), we selected primers from areas containing segmental insertions or substitutions, and we confirmed their specificity by using genomic DNA from a strain carrying mat2/3 region deletion. These primer sets were used in quantitative PCR on DNA prepared either from immunoprecipitated chromatin fractions or from whole-cell crude extracts used as an input control. To account for differences in loading or the variance in amplification by different primer sets, we used multiplex PCR to normalize the relative enrichment of the mat sequences to the act1 PCR product (Fig. 1).

The data show that H3 Lys9 methylation and Swi6 were both preferentially enriched throughout a 20-kb region of DNA that included and extended past the mat2/3 interval (Figs. 1A and2). Unexpectedly, we observed a marked decrease in H3 Lys9 methylation and Swi6 localization on both sides of the mat2/3 interval, which coincided precisely with two 2-kb identical inverted repeats unique to the matlocus (named IR-L and IR-R, respectively, for inverted repeat left and right) (Figs. 1A and 2). Interestingly, H3 Lys9 methylation and Swi6 were enriched at IR-L and IR-R but were not detected outside these repeats in surrounding euchromatic regions. Consistent with these results, expression of the marker genes inserted outside the IR-L toward mat1 was unaffected, but severe repression was observed on the other side (7, 14) (see also Fig. 1A). Together, these data suggest that the IR-L and IR-R repeats may serve as boundaries of the silent heterochromatin domain at themat2/3 interval.

Figure 2

Distribution of H3 Lys9 methylation across the mat locus. Results from the ChIP analyses with an antibody to H3 Lys9-methyl (15) are presented in alignment with the map of the mat locus. Quantitative measurements of H3 Lys9 methylation levels throughout themat locus were carried out as described in Fig. 1.

Silencing at the mating-type region in mat1-P cells is less stringently controlled than in mat1-M cells (14), which suggests that the P andM cells might differ in their chromatin organization at themat locus. Comparison between nonswitching mat1-Mand mat1-P strains using ChIP analysis revealed thatP cells contain relatively lower levels of Swi6 throughout almost the entire mat2/3 interval (Fig. 1A). The difference in Swi6 localization appears to be limited to the mat locus, because no such differences in Swi6 levels were observed at centromeric repeat sequences of M and P cells (Fig. 1C) and equal amounts of Swi6 protein were present in each cell type (Fig. 1D). Interestingly, ChIP analysis revealed no clear difference in H3 Lys9 methylation between the P and Mmating-type cells (Fig. 2). This suggested that additional factors, perhaps the information encoded by mat1-M or -Ploci, might account for these differences in Swi6 localization.

According to the histone code hypothesis, different posttranslational modifications of histones may have distinct biological implications (18, 19). In contrast to H3 Lys9, methylation of H3 Lys4 is associated with transcriptionally active regions in Tetrahymena(20). To compare these two sites, we next sought to map at high resolution the H3 Lys4 methylation pattern across the same locus. ChIP analysis revealed that H3 Lys4 methylation was enriched at transcriptionally competent regions surrounding the inverted repeats, as well as at the act1 locus, but no enrichment was observed throughout the heterochromatic region flanked by the IR-L and IR-R elements (Fig. 3A). The levels of H3 Lys4 methylation are higher in the regions containing genes than in intergenic regions (Fig. 3A). These findings suggest that H3 Lys4 methylation is associated with the euchromatic regions of the genome but is lacking at the heterochromatic domains. Indeed, ChIP analysis showed that H3 Lys4methylation was absent at the ura4 + marker inserted within the transcriptionally repressed centromeric (otr and imr) repeats, but was enriched at theura4DS/E endogenous euchromatic location (Fig. 3B). The lack of H3 Lys4 methylation at the otr andimr locations coincides with the presence of H3 Lys9 methylation and Swi6 (Fig. 3B). As expected, H3 Lys4 methylation was not detected at the central core (cnt) region, where the histone H3 variant CENP-A is localized (21). Although the biological significance of H3 Lys4 methylation is not yet clear, it is possible that this histone modification may serve to recruit transcriptional coactivators, similar to arginine methylation of histone H4 (22). Together, these results further suggest that the IR-L and IR-R repeats define the borders between euchromatin and heterochromatin, and that distinctive patterns of H3 methylation exist on opposite sides of these putative boundary elements.

Figure 3

Analysis of H3 Lys4 methylation at the mat and cen loci. (A) Localization of H3 Lys4 methylation at the matlocus. The physical map of the mating-type region indicating locations of the open reading frames (ORFs), including an essentiallet1 + gene, are indicated (top). Results from ChIP analysis with antibody to H3 Lys4-methyl (Upstate Biotechnology) are shown (bottom). Quantitation of H3 Lys4methylation levels at mat, relative to act1locus, is plotted. (B) Schematic representation ofcen1 and the corresponding ura4 +insertion sites (top). Levels of Swi6 and Lys4 (K4) or Lys9 (K9) methylation of H3 at the three insertion sites were determined by ChIP analyses (bottom). DNA from ChIP or WCE was analyzed using a competitive PCR strategy, whereby one set of primers amplifies different-sized products from theura4 + marker gene at cen1 and the control ura4DS/E minigene at the endogenous euchromatic location, respectively. The ratios of ura4 + and control ura4DS/E signals present in ChIP and WCE were used to calculate relative enrichment, shown beneath each lane.

If IR-L and IR-R repeats are bona fide boundary elements, deletions of either repeat should result in the spreading of heterochromatin into neighboring regions. For this purpose, strains containing a deletion of either IR-L or IR-R were constructed (23). Because spreading of heterochromatin is stochastic and depends on dosage of chromatin proteins, we used strains carrying three copies ofswi6 +, which is known to enhance silencing at the mat locus (7). Remarkably, the IR-L deletion resulted in the spreading of both H3 Lys9methylation and Swi6 localization to more than 8 kb of the adjoiningL-region, whereas no spreading was observed in IR-L+ cells (Fig. 4). Furthermore, the spreading of heterochromatin was associated with a decrease in H3 Lys4 methylation levels in the L-region. The IR-L deletion also caused growth defects, most likely due to the spreading of heterochromatin into the essentiallet1 + gene. Spreading of H3 Lys9methylation and Swi6, and a concomitant decrease in H3 Lys4methylation, was also observed in the IR-R deletion strain (Fig. 4), indicating that these repeats are indeed boundary elements that function to prevent heterochromatic spreading into neighboring euchromatic regions.

Figure 4

Deletion of IR-L or IR-R causes spreading of heterochromatin into flanking regions. Three strains (WT, IR-LΔ, and IR-RΔ) were used. Results from ChIP analyses with antibodies to Swi6, H3 Lys9-methyl, or H3 Lys4-methyl are shown aligned with the map of themat2/3 interval and surrounding regions. Numbered bars below the map indicate PCR fragments analyzed in each ChIP assay. Thick arrows indicate direction and distance of heterochromatin spreading that was caused by IR-LΔ or IR-RΔ.

Recently, it was suggested that insulators may pair in cis to form chromatin loops within the nucleus (24–26). Therefore, it was of interest to determine whether deletion of IR-L would affect IR-R boundary function, or vice versa. Although it remains a formal possibility that the IR-R or IR-L elements may interact with each other, we found that these repeat elements can operate independently of each other in preventing the spreading of Swi6 and H3 Lys9methylation to adjacent sequences on their respective sides (Fig. 4). That is, IR-LΔ did not affect IR-R boundary activity, and vice versa.

The results presented above indicate that the presence of repressive heterochromatin complexes at the mat2/3 interval may prevent H3 Lys4 methylation or, alternatively, that H3 Lys9 methylation may directly inhibit H3 Lys4methylation. To test this, we performed ChIP analysis on theKint2::ura4 + marker gene inserted at the K-region (Fig. 5A). In aswi6-115 mutant strain, which results in defects in Swi6 localization and loss of silencing, H3 Lys9 methylation was unchanged; however, there was an observed increase in H3 Lys4 methylation relative to wild-type background cells. Deletion of clr4, the histone methyltransferase responsible for H3 Lys9 methylation, resulted in a marked decrease of both H3 Lys9 methylation and Swi6 localization but a much more pronounced increase in H3 Lys4 methylation relative to the wild type and swi6-115 strains. Taken together, these data indicate that H3 Lys9 methylation, in combination with heterochromatin-associated proteins such as Swi6, prevents H3 Lys4 methylation at the mat2/3 region.

Figure 5

(A) Effect of H3 Lys9methylation and heterochromatin on H3 Lys4 methylation atmat2/3 region. The physical map of the mating-type locus indicates the Kint2::ura4 + insertion site (top). ChIP with competitive PCR analyses, as described in Fig. 3, was used to measure the effect of mutations in swi6(swi6-115) or clr4clr4) on Lys4 (K4) or Lys9 (K9) methylation of H3 and Swi6 at Kint2::ura4 +. Serial dilutions of indicated cultures were plated onto nonselective (N/S), medium lacking uracil (URA), or 5-fluoroorotic acid (FOA) medium to assay ura4 + expression (bottom). (B) Schematic model showing a proposed higher order chromatin organization at the mat locus. The IR-L and IR-R boundary elements (blue arrows) prohibit the spreading of Swi6 and H3 Lys9 methylation to the flanking euchromatic regions. In addition, we hypothesize that these elements might also tether themat locus to nuclear periphery, thereby creating a specialized domain.

Collectively, these results suggest that methylation of H3 at Lys4 or Lys9 is reciprocally associated with euchromatic and heterochromatic regions, respectively. We also provide direct evidence that the boundary elements that define the borders between euchromatin and heterochromatin protect against the encroachment of heterochromatin signals into neighboring euchromatic regions. Boundary or insulator sequences have been suggested to govern subnuclear organization of DNA (24). In this regard, we find that the mat locus is localized near the nuclear periphery (27). We hypothesize that the higher order chromatin organization at the mat2/3 interval involves (i) the methylation of H3 Lys9 and the subsequent localization of heterochromatin-associated proteins to establish a primary level of organization, and (ii) the binding of specialized proteins to boundary elements that then tether themat locus to the nuclear periphery (Fig. 5B). This specialized chromatin organization results in silencing and recombination suppression, and is likely to promote long-range interaction between mat1 and the opposite mating-type donor locus, which is required for efficient switching (11, 28). Further characterization of the boundary elements and their role in preserving distinct chromatin configurations will help us better understand the higher order organization of chromosomes.

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

REFERENCES AND NOTES

View Abstract

Navigate This Article