The transcription of genes located in subtelomeric regions of yeast chromosomes is repressed relative to the rest of the genome. This repression requires wild-type nucleosome levels but not the telomere silencing factors Sir2, Sir3, Sir4, and Rap1. Subtelomeric heterochromatin is characterized by the absence of acetylation or methylation of histone H3 lysine residues, but it is not known whether histone H3 hypoacetylation or hypomethylation is a prerequisite for the establishment of subtelomeric heterochromatin. We have systematically mutated the N-terminal tails of histone H3 and H4 in Saccharomyces cerevisiae and characterized the effects each mutant has on genome-wide expression. Our results show that subtelomeric transcriptional repression is dependent on the histone H3 N-terminal domain, but not the histone H4 N-terminal domain. Mutating lysine-4, lysine-9, lysine-14, lysine-18, lysine-23, and lysine-27 to glycine in histone H3 is also sufficient to significantly reduce subtelomeric gene repression. Individual histone H3 lysine mutations, however, have little effect on subtelomeric gene repression or genome-wide expression, indicating that these six lysine residues have redundant functions. We propose that acetylation and methylation of histone H3 N-terminal lysine residues act as redundant mechanisms to demarcate regions of euchromatin from heterochromatin.
THE packaging of DNA with histones into nucleosomes and higher-order chromatin structures has profound effects on transcription initiation (Horn and Peterson 2002; Narlikar et al. 2002; Grewal and Moazed 2003). Genomic chromatin structure is organized into regions of heterochromatin and euchromatin by a cadre of chromatin-modifying enzymes. Heterochromatin is characterized by a compact chromatin structure that is inaccessible to DNA-binding proteins (Grewal 2000), including components of the transcription machinery. Genes located in heterochromatin are transcriptionally silenced, irrespective of their own promoter sequences (Moazed 2001). Heterochromatin is also distinguished by distinct patterns of post-translational histone modifications (Jenuwein and Allis 2001; Richards and Elgin 2002). Each of the four core histone proteins (H2A, H2B, H3, and H4) is post-translationally modified at multiple residues in vivo. These modifications include acetylation of lysine residues, methylation of arginine and lysine residues, phosphorylation of serine residues, and ubiquitination of lysine residues (Iizuka and Smith 2003).
The link between histone modifications and heterochromatin has been best established at yeast telomeres. Sir2, Sir3, and Sir4 are recruited to telomeres by the telomere-binding proteins Rap1 and yeast Ku (Lustig 1998; Luo et al. 2002). Sir2 is a NAD-dependent histone deacetylase (HDAC) that deacetylates lysine-16 of histone H4, among other residues (Imai et al. 2000; Landry et al. 2000). Since Sir3 binds preferentially to unacetylated H4 lysine-16 (Carmen et al. 2002), the Sir2 HDAC activity allows the Sir complex to spread along chromatin up to 6–8 kb away from the telomere (Kimura et al. 2002; Suka et al. 2002). Sir complex spreading is opposed by the Sas2 and, to a lesser extent, the Esa1 histone acetyltransferases (HATs; Kimura et al. 2002; Suka et al. 2002). These HATs acetylate lysine-16 of H4 in euchromatin, so as to make Sir3 binding unfavorable.
Recent functional genomics studies have identified a novel heterochromatin domain adjacent to telomeres in Saccharomyces cerevisiae. These subtelomeric or HAST (Hda1-affected subtelomeric) domains extend 10–25 kb from the telomere end (Robyr et al. 2002). Genes located in these domains are transcriptionally repressed when yeast are grown in standard glucose media. DNA microarray analysis has revealed that the repression of subtelomeric genes is dependent upon local nucleosome structure, but is independent of Sir2, Sir3, Sir4, or Rap1 function (Wyrick et al. 1999). Acetylation microarray experiments have determined that these subtelomeric domains are characterized by hypoacetylation of histone H3 residues through the activity of the Hda1 HDAC (Bernstein et al. 2002; Robyr et al. 2002). However, deletion of HDA1 has a small but significant effect on subtelomeric gene repression (Bernstein et al. 2000; Hughes et al. 2000), so it is unclear if there is a functional requirement for histone H3 hypoacetylation in subtelomeric heterochromatin.
Global analysis of histone H3 lysine-4 methylation patterns has shown that yeast telomeric and subtelomeric heterochromatin is also preferentially hypomethylated relative to euchromatin (Bernstein et al. 2002). Deletion of the Set1 histone methyltransferase, which is required for methylation of histone H3 lysine-4 (Briggs et al. 2001; Roguev et al. 2001), results in loss of silencing at the rDNA locus and at telomeres (Nislow et al. 1997; Briggs et al. 2001; Bryk et al. 2002; Krogan et al. 2002). However, a direct role for lysine-4 methylation in telomeric, subtelomeric, or rDNA silencing has yet to be established (Bernstein et al. 2002).
To determine the functional role of histone hypoacetylation and hypomethylation in subtelomeric gene repression, we have profiled the genome-wide expression changes due to mutations in the histone H3 and H4 N-terminal tails. We have chosen this strategy because genome-wide expression analysis is the best available assay of subtelomeric gene expression and because it has proven to be highly successful in advancing our knowledge of how transcription is regulated by histones (Wyrick et al. 1999) and other factors (Holstege et al. 1998; Bernstein et al. 2000; Lee et al. 2000; Wyrick and Young 2002). Our results indicate that acetylated and methylated lysine residues in histone H3 have redundant functions in repressing the transcription of both subtelomeric genes and genes located elsewhere in the genome.
MATERIALS AND METHODS
Strains, media, and growth conditions:
A complete list of yeast strains is given in Table 1. For each genome-wide expression experiment, two mutant and two wild-type yeast cultures were grown in parallel to a final OD600 of 0.5–0.7 in yeast extract/peptone/dextrose (YPD) media and then harvested as described previously (Holstege et al. 1998).
Plasmid construction and site-directed mutagenesis:
The wild-type histone H3 (HHT2) and H4 (HHF2) genes were PCR amplified from plasmid pRM200 (Mann and Grunstein 1992) using primers OJLW17For (AAACTCGAGGGATTGCTAGTAAAAACAACTGG) and OJLW17Rev (AAATCTAGACAGTGACTCAAAATTCAAACG). The resulting 2.1-kb PCR fragment was digested with the XhoI and XbaI restriction enzymes (New England Biolabs, Beverly, MA) and cloned into the pRS316 shuttle vector (Sikorski and Hieter 1989) to give the plasmid pJL001. The wild-type histone H4 and mutant histone H3 (K9,14,18,23G) genes were PCR amplified using the same primers and cloned into the pRS316 vector to give plasmid pJL002.
The QuikChange kit (Stratagene, La Jolla, CA) was used to mutate histone H3 lysine-4 to glycine and lysine-27 to glycine in plasmids pJL001 and pJL002. Oligos OJLW1 (CTCCACAATGGCCAGAACTGGACAAACAGCTAGAAAATCC) and OJLW7 (GGATTTTCTAGCTGTTTGTCCAGTTCTGGCCATTGTGGAG) were used to mutate lysine-4 to glycine. Oligos OJLW1C (CCAAGGCTGCCAGAGGATCCGCCCCATCTAC) and OJLW8 (GTAGATGGGGCGGATCCTCTGGCAGCCTTGG) were used to mutate lysine-27 to glycine in pJL001; oligos OJLW18 (CCGGTGCTGCCAGAGGATCCGCCCCATCTA) and OJLW9 (TAGATGGGGCGGATCCTCTGGCAGCACCGG) were used to mutate lysine-27 to glycine in pJL002. Each mutation was confirmed by DNA sequencing using primers OJLW2 (GTTCTTTTATATAGGACC) and OJLW3 (AGATTGAAATCTCAAGTC). A subset of the mutant histone H3 alleles were cloned into the pRS412 shuttle vector (Brachmann et al. 1998) using the XhoI and NotI restriction enzymes.
Genome-wide expression profiling:
Total RNA was isolated from each yeast culture and used to prepare complementary DNA and biotin-cRNA following standard protocols (Affymetrix, Santa Clara, CA; Holstege et al. 1998). For the histone H3 Δ(4-30) and histone H4 K5,8,12,16G mutant and wild-type strains, the cRNA was hybridized to a set of four Ye6100 oligonucleotide arrays (Affymetrix) and scanned as described elsewhere (Wodicka et al. 1997). The cRNA from all other strains was hybridized to a single S98 genome oligonucleotide array and scanned following standard protocols. Intensities were captured using GeneChip software (Affymetrix) and a single raw expression level for each gene was determined. Complete data sets are available at http://wyrick.sbs.wsu.edu/histoneH3/.
For the Ye6100 arrays, the data from each chip were normalized and corrected as described previously (Holstege et al. 1998). A change in mRNA levels was deemed significant on the basis of the following criteria: (1) the average fold change (up or down) was >3-fold, (2) the fold change (up or down) in each replicate experiment was >1.5-fold, and (3) the change in absolute intensity values was above background values in both comparisons. See Wyrick et al. (1999) for more details.
The data from each S98 array were normalized using the GeneChip software (Affymetrix), since no global changes in mRNA levels were detected in any of the experiments. The same criteria as above were used to identify genes with significant changes, except that the average fold change (up or down) was required to be more than twofold instead of threefold. The data from the S98 arrays were of a higher quality and reliability than the data from the Ye6100 arrays, allowing us to use this less stringent threshold. Analysis of S98 array data from four wild-type strains using these criteria revealed that only 1 gene of 6064 was identified as significantly changed, indicating a very low false-positive rate.
Telomere-proximal gene analysis and statistics:
Genes within 40 kb of a telomere were pooled and ordered according to their distance from a telomere. For the Ye6100 array data [histone H3 Δ(4-30) and histone H4 K5,8,12,16G], genes with Affymetrix ‘_f’ calls were excluded from the analysis, as described previously (Wyrick et al. 1999). A sliding 50-gene window was moved along the ordered gene list in 10-gene steps. The fraction of genes upregulated in each mutant and the average distance from the telomere were plotted for each 50-gene window. The statistical significance of the enrichment of upregulated genes in subtelomeric regions (10–20 kb from a telomere end) for each data set was calculated using the hypergeometric probability distribution.
Statistical analysis of comparisons of array data:
A hypergeometric probability model was employed to determine the statistical significance of the overlap between the lists of upregulated genes in the various histone H3 mutants. The resultant P-values were numerically calculated using the Mathematica software package (Wolfram Research, Champaign, IL).
Histone H3 N-terminal domain is required for the repression of subtelomeric genes:
Whole-genome Affymetrix oligonucleotide arrays were used to investigate whether the N-terminal domain of histone H3 was required for the repression of subtelomeric gene transcription in S. cerevisiae. Analysis of replicate samples revealed that the mRNA levels of 275 genes were upregulated and the mRNA levels of 185 genes were downregulated in the histone H3 Δ(4-30) mutant strain relative to wild type. Chromosome display of these data revealed that many of the upregulated genes clustered in subtelomeric chromosomal regions. Several representative chromosomes are shown in Figure 1; the effects on all chromosomes can be seen at http://wyrick.sbs.wsu.edu/histoneH3/. About 17% of genes located in subtelomeric regions showed increased mRNA levels in the histone H3 Δ(4-30) strain, compared to a genome-wide average of only 4% (P = 5.0 × 10−6). Subtelomeric regions were defined by a location of 10–20 kb from each telomere end. This region was chosen on the basis of previous studies that had shown that gene expression of this region was repressed by histones but was independent of Sir function (Wyrick et al. 1999). The choice of different boundaries for the subtelomeric region (e.g., 8–20 kb or 1–25 kb) did not significantly change the results (data not shown).
The pattern of subtelomeric gene expression in the histone H3 Δ(4-30) mutant strain was compared to that found in a sir2Δ or a nucleosome-depleted strain (Wyrick et al. 1999). In Figure 2A, the fraction of genes that showed increased mRNA levels in each mutant strain was plotted vs. distance from a telomere end. The histone H3 Δ(4-30) mutant appears to have significant effects on the expression of genes up to 20 kb from a telomere end, similar to the effects seen in the nucleosome-depleted strain, although smaller in magnitude. Our analysis of the published data for the histone H3 Δ(1-28) mutant strain (Sabet et al. 2003) showed similar results (A. Martin and J. Wyrick, unpublished data).
As a comparison, the pattern of subtelomeric gene expression in a histone H4 mutant strain (K5,8,12,16G) was analyzed in a similar manner in Figure 2B. The histone H4 mutant strain appears to have major effects on the expression of genes within 10 kb of a telomere end, similar to the effects seen in the sir2Δ strain. This result is consistent with the current model that Sir2 acts to silence telomere-proximal genes by deacetylating lysine-16 in histone H4. The histone H4 mutant did not have significant effects on subtelomeric gene expression.
These results indicate that subtelomeric gene repression requires the histone H3 N-terminal domain, but does not require the four acetylated lysine residues in the histone H4 N-terminal domain. Intriguingly, other studies have shown that lysine-9, lysine-14, and lysine-18 in the histone H3 N-terminal domain are preferentially hypoacetylated in subtelomeric heterochromatin (Bernstein et al. 2002; Robyr et al. 2002). This correlation suggests that hypoacetylated lysine residues in the N-terminal domain of histone H3 may be required for subtelomeric gene repression.
Genome-wide expression analysis of histone H3 lysine mutants:
To test this hypothesis, oligonucleotide arrays were used to profile the genome expression changes in a histone H3 K9,14,18,23G mutant. Each of these lysine residues is acetylated in yeast (Roth et al. 2001; Suka et al. 2001), and mutation of all four lysine residues to glycine leads to loss of telomeric silencing in S. cerevisiae (Thompson et al. 1994). Analysis of replicate samples revealed that the mRNA levels of 44 genes were upregulated and the mRNA levels of 14 genes were downregulated in the histone H3 K9,14,18,23G mutant strain relative to wild type. Only two of the upregulated genes were located in the subtelomeric region. The overall pattern of subtelomeric gene expression was analyzed in Figure 2C. This analysis shows that the histone H3 K9,14,18,23G mutant does not have significant effects on subtelomeric gene expression, indicating that these four lysine residues alone are not required for the repression of subtelomeric genes.
Two other modified lysine residues are present in the histone H3 N-terminal tail. Lysine-27 has recently been shown to be acetylated by Gcn5 and Elp3 (Suka et al. 2001; Kristjuhan et al. 2003), and lysine-4 is methylated by Set1 (Briggs et al. 2001; Roguev et al. 2001). Whole-genome oligonucleotide arrays were used to examine the effects of mutating lysine-4 and lysine-27 to glycine, both singly and in combination with the K9,14,18,23G mutation. Figure 3 shows the complete list of mutations examined and the numbers of genes up- and downregulated in each mutant strain. To estimate the false-positive error rate in these data sets, data sets from four wild-type strains were analyzed in the same manner as the experimental data sets. Only 1 of 6064 genes represented on the oligonucleotide array was identified as changing significantly in the wild-type to wild-type comparisons, indicating a very low false-positive rate.
Overall, 312 genes were upregulated in at least one of the six mutant strains listed in Figure 3, and 62 genes were downregulated in at least one of the six mutant strains. The list of upregulated genes showed significant enrichment of genes involved in vitamin metabolism (P = 2.1 × 10−9), carbohydrate metabolism (P = 1.7 × 10−7), and carboxylic acid metabolism (P = 7.4 × 10−7). The list of downregulated genes showed significant enrichment of genes involved in pheromone response (P = 2.0 × 10−6).
Effects of histone H3 lysine mutants on subtelomeric gene expression:
To determine what effect these mutations had on subtelomeric gene transcription, we calculated what fraction of genes located 10–20 kb from a telomere end were upregulated in each mutant strain. As shown in Table 2, genes upregulated by the histone H3 K4,9,14,18,23,27G are highly enriched in subtelomeric regions (P = 8.1 × 10−9), similar to the enrichment seen in the histone H3 Δ(4-30) mutant. The histone H3 K27G, K9,14,18,23,27G, and K4,9,14,18,23G mutants also showed enrichment in subtelomeric regions, although to a lesser extent. The histone H3 K4G and K9,14,18,23G mutants showed no enrichment for genes upregulated in subtelomeric regions.
To further test the observation that the histone H3 K4,9,14,18,23,27G mutant and the histone H3 Δ(4-30) mutant had similar effects on subtelomeric gene expression, the genome expression data for each of these mutants were compared using a telomere distance plot. As shown in Figure 4, the profile of upregulated genes as a function of distance from the telomere is similar for the histone H3 Δ(4-30) and K4,9,14,18,23,27G mutants. This analysis confirms the observation that mutating lysine-4 and lysine-27 in combination with lysine-9, lysine-14, lysine-18, and lysine-23 in histone H3 leads to an induction of subtelomeric gene expression that is similar to the effect of deleting amino acids 4–30 in histone H3.
Comparison of genome-wide expression profiles for histone H3 lysine mutants:
To determine the contributions of the H3 lysine residues to genome-wide expression, we compared the sets of genes upregulated in each histone H3 mutant strain. As shown in Figure 5A, the sets of genes upregulated in the histone H3 K4G and K9,14,18,23G mutants are small subsets of the set of genes upregulated in the combined histone H3 K4,9,14,18,23G mutant. This analysis indicates that lysine-4 and one or more of lysine-9, -14, -18, and -23 have redundant roles in repressing the transcription of 181 genes. A similar pattern is seen in Figure 5B. The sets of genes upregulated in the histone H3 K27G and K9,14,18,23G mutants are primarily subsets of the genes upregulated in the combined histone H3 K9,14,18,23,27G mutant, indicating that lysine-27 and one or more of lysine-9, -14, -18, and -23 have redundant functions in repressing the transcription of 68 genes.
As shown in Figure 5C, there is a large degree of overlap between the genes upregulated in the histone H3 K4,9,14,18,23,27 and H3 K4,9,14,18,23G mutants. A total of 57% of the genes upregulated in the K4,9,14,18,23G mutant overlap with 72% of the genes upregulated in the K4,9,14,18,23,27G mutant (P = 2.1 × 10−161). The effect of mutating lysine-4 in combination with the other two sets of mutations is examined in Figure 5D. There is a highly significant degree of overlap (P = 7.6 × 10−75) between the sets of genes upregulated by the histone H3 K9,14,18,23,27G mutant and the histone H3 K4,9,14,18,23,27G mutant (64 genes). An additional 116 genes that are not affected in the histone H3 K9,14,18,23,27G mutant are upregulated in the histone H3 K4,9,14,18,23,27G mutant, showing (similar to Figure 5A) that lysine-4 and one or more of lysine-9, -14, -18, -23, and -27 have redundant functions in repressing the transcription of >100 genes.
Previous studies had demonstrated that lysine residues in the N-terminal domain of histone H3 are preferentially hypomethylated and hypoacetylated in subtelomeric heterochromatin (Bernstein et al. 2002; Robyr et al. 2002). We have now characterized the functional role of histone H3 lysine residues in regulating subtelomeric heterochromatin and genome-wide expression. First, we have shown that the histone H3 N-terminal domain is uniquely required for subtelomeric gene repression. Second, mutating H3 lysine-4, -9, -14, -18, -23, and -27 to glycine mimics the effect of the H3 N-terminal deletion on subtelomeric gene repression. Third, H3 lysine-4, -9, -14, -18, -23, and -27 have redundant functions in repressing the transcription of genes located in subtelomeric regions and elsewhere in the genome. These findings suggest a model for the role of histone H3 lysine residues in subtelomeric transcriptional repression, which is discussed later in this section.
Distinct roles for histone H3 and H4 in yeast heterochromatin:
We find that the mRNA levels of telomere-proximal genes are induced by N-terminal mutations in histone H3 or H4, in accord with previous studies (Aparicio et al. 1991; Thompson et al. 1994). In contrast, we find that mRNA levels of subtelomeric genes are induced by N-terminal mutations in histone H3 but not by N-terminal mutations in histone H4. These findings indicate that the histone H3 N-terminal domain plays a unique role in the repression of genes located in subtelomeric heterochromatin.
These results, taken together with previous studies of the silent mating loci, reveal that S. cerevisiae employs a surprising diversity of mechanisms to silence the transcription of genes in heterochromatin. Transcriptional repression of the silent mating loci requires the N-terminal domain of histone H4 but not histone H3 (Kayne et al. 1988; Morgan et al. 1991; Mann and Grunstein 1992); repression of subtelomeric genes requires the N-terminal domain of histone H3 but not histone H4; and repression of telomere-proximal genes requires both histone H3 and H4, although histone H4 appears to play a more important role. Each of these distinct heterochromatin domains appears to require a specific pattern of histone modifications for regulating gene expression, as specified by the histone code hypothesis.
We have also examined the role of the N-terminal tails of histone H2A and H2B in subtelomeric gene repression. Our results indicate that deletion of the histone H2A N-terminal tail [H2A Δ(4-20)] leads to a small but significant effect on subtelomeric repression, while deletion of the histone H2B N-terminal tail [H2B Δ(3-32)] has no effect on subtelomeric repression (M. Parra and J. Wyrick, unpublished data). Interestingly, these results appear to mirror the effects seen on telomeric silencing for each of these N-terminal deletions (Thompson et al. 1994; Wyatt et al. 2003).
Genome-wide analysis of histone H3 lysine mutations:
Our results indicate that mutating lysine-4, -9, -14, -18, -23, and -27 to glycine is sufficient to significantly reduce subtelomeric gene repression. The H3 K4,9,14,18,23,27G mutation also leads to the derepression of telomere-proximal genes, although not to the same extent as that seen for subtelomeric genes. Since the H3 K4,9,14,18,23,27G mutation does not affect the expression of any known silencing genes, it is likely that the effect on subtelomeric gene expression is a direct consequence of the histone H3 N-terminal mutations. Overall, the histone H3 K4,9,14,18,23,27G mutation resulted in an increase in the mRNA levels for 180 genes and a decrease in the mRNA levels for 33 genes, indicating that the six lysine residues in the histone H3 N-terminal domain function primarily to repress transcription. This finding is consistent with previously published microarray data for a histone H3 Δ(1-28) mutant strain (Sabet et al. 2003).
Lysine-4 appears to play a key role in subtelomeric gene repression. Only mutants that contained the K4G mutation showed maximal induction of subtelomeric gene expression and maximal induction of genome-wide expression. The histone H3 K4G mutation on its own, however, had no affect on subtelomeric gene expression and little affect on genome-wide expression. Only when lysine-4 was mutated in combination with lysine-9, -14, -18, and -23 was a significant effect seen on subtelomeric gene expression. These results indicate that lysine-4 and lysine-9, -14, -18, and -23 have redundant functions in repressing the transcription of genes located in subtelomeric regions and elsewhere in the genome.
Our data indicate that lysine-27 also plays an important role in subtelomeric heterochromatin. The histone H3 K27G mutation had a small but significant effect on subtelomeric gene expression. While the histone H3 lysine-27 residue is acetylated in yeast, in higher eukaryotes it is also methylated. Intriguingly, lysine-27 methylation has been shown to play a key role in X inactivation and heterochromatin formation in mammalian cells (Plath et al. 2003).
The histone H3 K9,14,18,23G mutation alone had little affect on subtelomeric or telomeric gene repression. This finding is at odds with a previous study that indicated that the H3 K9,14,18,23G mutation resulted in a substantial loss of telomeric silencing, as measured using a 5-fluoroorotic acid sensitivity assay (Thompson et al. 1994). Since the observed effect on telomeric silencing in the H3 K9,14,18,23G mutant strain was significantly less than the effects seen in a Sir2 deletion or histone H4 mutation (Thompson et al. 1994), it is possible that this silencing defect was not significant enough to be detected in the DNA array assay. In addition, the previous study measured the silencing effects for a single, artificially constructed telomere, while our results measured the effects on all 32 wild-type telomeres.
Functional redundancy between histone H3 lysine residues:
Our data demonstrate that individual lysine-to-glycine mutations (such as K4G and K27G) had little effect on genome-wide expression, while combinations of lysine-to-glycine mutations (such as K4,9,14,18,23G and K9,14,18,23,27G) had large and synergistic effects on genome-wide expression. This observation is reminiscent of the signaling network model of chromatin (Schreiber and Bernstein 2002).
The signaling network model of chromatin argues that the behavior of post-translational histone modifications resembles that of receptor tyrosine kinase signaling (Schreiber and Bernstein 2002). A prediction of this model is that individual histone modifications would have redundant effects on transcription initiation. This prediction is based in part on a study by Grunstein and colleagues (Durrin et al. 1991), which showed that single lysine mutations in the N-terminal domain of histone H4 had little effect on GAL1 transcription, while mutating three or all four acetylated lysine residues in histone H4 resulted in a significant decrease in GAL1 transcription. Our data indicate that lysine residues in the histone H3 N-terminal tail act redundantly to repress gene transcription and provide evidence in support of the signaling network model of chromatin.
A model for subtelomeric gene repression:
In higher eukaryotes, methylation of lysine-9 and -27 in histone H3 has been linked to heterochromatin formation (Lachner and Jenuwein 2002; Plath et al. 2003). S. cerevisiae, however, lacks these heterochromatin specific post-translational modifications. Our data indicate that lysine-4, -9, -14, -18, -23, and -27 in histone H3 are required for subtelomeric gene repression, yet none of these lysine residues is frequently post-translationally modified in subtelomeric heterochromatin. Lysine-4 is hypomethylated in subtelomeric heterochromatin and is hypermethylated in the coding regions of active genes (Bernstein et al. 2002). The other histone H3 lysine residues are hypoacetylated in subtelomeric heterochromatin (Bernstein et al. 2002; Robyr et al. 2002), but are hyperacetylated in euchromatin, particularly in promoter regions (Bernstein et al. 2002).
We propose that methylation and acetylation of histone H3 act as redundant mechanisms to demarcate regions of euchromatin from heterochromatin in S. cerevisiae. In this model, hypomethylated and hypoacetylated histone H3 lysine residues play a direct role in establishing repressive subtelomeric heterochromatin. While the mechanism for this repression is unknown, in Figure 6 we hypothesize that unacetylated and unmethylated lysine residues may function as binding sites for one or more repressor proteins. Methylation or acetylation of the target lysine residues would disrupt their interaction with the repressor proteins, just as acetylation of lysine-16 in histone H4 disrupts binding of Sir3 in telomeric heterochromatin (Carmen et al. 2002; Kimura et al. 2002; Suka et al. 2002). For this reason, the repressor protein would be restricted to chromosomal regions in which histone H3 is hypomethylated and hypoacetylated (e.g., subtelomeric regions). A similar model has been proposed to explain the role of histone H3 lysine-79 methylation in telomeric silencing (van Leeuwen and Gottschling 2002).
Mutating lysine residues to glycine completely removes the lysine side-chain and thus should eliminate any potential interaction with a repressor protein. However, since each of the unacetylated and unmethylated lysine residues in the histone H3 N-terminal tail can serve as a potential binding site for the repressor protein, all of these lysine residues would need to be mutated to glycine to eliminate subtelomeric gene repression. This prediction of the model fits nicely with our array data, since maximal loss of subtelomeric repression is seen only when we mutate H3 lysine-4, -9, -14, -18, -23, and -27 to glycine, and single mutations, such as K4G, have no effect on subtelomeric repression.
This model also predicts that deletion of the Set1 histone methyltransferase would lead to the derepression of subtelomeric genes. In a SET1 deletion strain, unmethylated lysine-4 would be present in euchromatin and would serve as a potential binding site for the repressor protein. According to the model, this would result in a redistribution of the repressor protein from subtelomeric regions to euchromatin, leading to the upregulation of subtelomeric genes, as has been reported (Bernstein et al. 2002).
An alternative model is that the histone H3 K4,9,14, 18,23,27G mutation might disrupt subtelomeric heterochromatin by altering the electrostatic charge of the histone H3 N-terminal tail. Previous studies have suggested that positively charged lysine residues may play an important role in histone-DNA and histone-histone interactions (Wolffe and Hayes 1999). Hence, disruption of these interactions through mutation of histone H3 lysine residues could directly disrupt subtelomeric heterochromatin. This model cannot completely explain our results since lysine-4 (which is methylated) is also required for depression of subtelomeric genes. Future studies will allow us to determine which of these models is correct and to determine the identity of the postulated repressor proteins. It will also be interesting to examine whether lysine-36 and lysine-79 in histone H3 play a role in subtelomeric gene repression.
We are grateful to Lisa Gloss, Michael Parra, Ray Reeves, and Michael Smerdon for helpful discussions and comments on the manuscript. We thank Michael Grunstein for the generous gift of yeast strains and plasmids. We thank Randall Morse for sharing data prior to publication. This project originated in Richard Young's laboratory and we thank him for his support, encouragement, and helpful suggestions. We thank R. Trace Sinclair and Monique Kohagura for software and web support. This work was supported by Research Scholar grant RSG-03-181-01-GMC from the American Cancer Society and by funds from an American Cancer Society Institutional Research grant.
Communicating editor: F. Winston
- Received January 21, 2004.
- Accepted April 2, 2004.
- Genetics Society of America