The formation and stability of epigenetically regulated chromatin is influenced by DNA replication and factors that modulate post-translational modifications on histones. Here we describe evidence that PCNA can affect silencing in Saccharomyces cerevisiae by facilitating deposition of H3 K56ac onto chromosomes. We propose that PCNA participates in this process through a pathway that includes replication factor C, the chromatin assembly factor Asf1p, and the K56-specific acetyltransferase Rtt109p. We show that mutation of POL30 or loss of K56-acetylation in rtt109 and histone H3 mutants enhances silencing at the crippled HMR locus HMRae** via restoring Sir binding and that pol30 mutants with silencing phenotypes have reduced levels of H3 K56ac. Although loss of acetylation on H3 K56 was generally compatible with silencing, mutations at this residue also led to defects in silencing an ADE2 reporter at HMR and abolished silencing when combined with cac1 or pol30-8. These silencing phenotypes are analogous to those in asf1 mutants or pol30-6 and pol30-79 mutants with defects in ASF1-dependent pathways. On the basis of these findings, we propose that mutations in DNA replication factors alter acetylation of H3 K56. We show that this defect, in turn, contributes to misregulation of epigenetic processes as well as of cellular responses to DNA damage.
THE foundation supporting epigenetically regulated chromatin is nucleosomal DNA. In Saccharomyces cerevisiae, the strength and stability of silent chromatin depends on the characteristics of histones present not only at silenced loci but also elsewhere throughout the genome. In yeast, transcribed loci are enriched in both acetylated and methylated histones whereas histones in silenced loci are hypoacetylated and hypomethylated (Braunstein et al. 1996; Suka et al. 2001; Bernstein et al. 2002; Bryk et al. 2002; Hoppe et al. 2002; Rusché et al. 2002; Ng et al. 2003; Santos-Rosa et al. 2004; Katan-Khaykovich and Struhl 2005; Rudner et al. 2005; Li et al. 2006). Loss of acetylation in silent chromatin is mediated by the NAD+-dependent histone deacetylase Sir2p (Rusché et al. 2003 and references therein). Sir2p enables the major structural components of silent chromatin, Sir2, -3, and -4 proteins, to bind preferentially to hypoacetylated histones and spread across the chromosome (Hoppe et al. 2002; Luo et al. 2002; Rusché et al. 2002). The resulting silent chromatin then blocks transcription and prevents histone-modifying enzymes from reaching their targets. Silencing defects at the HM loci and telomeres in yeast can occur upon overexpression or loss of histone-modifying enzymes. In these instances, inappropriately modified histones may prevent Sir proteins from interacting stably with silent loci. In addition, global loss of histone modifications is thought to result in the redistribution of Sir protein binding throughout the genome, thereby limiting the pool of Sir proteins available for forming silent chromatin at appropriate sites (Singer et al. 1998; van Leeuwen et al. 2002; Rusché et al. 2003).
Silencing defects are also observed in cells with altered expression of proteins that load histones onto DNA. Cells lacking the chromatin assembly factors CAF-I (composed of Cac1p, Cac2p, and Cac3p) and Asf1p exhibit silencing defects (Kaufman et al. 1997; Monson et al. 1997; Enomoto and Berman 1998; Tyler et al. 1999). These defects are more severe in asf1 cac1 and asf1 cac2 mutants relative to single mutants (Tyler et al. 1999; Sutton et al. 2001). Overexpression of Asf1p also leads to defects in silencing at both the HM loci and telomeres (Le et al. 1997; Singer et al. 1998). The silencing defects associated with cac1 and asf1 mutants may reflect both improper chromatin assembly and altered post-translational modifications on histones. Cells lacking CAC1 have reduced levels of histone H3 as well as Sir proteins at HMR and telomere VIR (Tamburini et al. 2006; Huang et al. 2007) and loss of CAC1 or ASF1 alters the topology of plasmids in vivo (Prado et al. 2004; Driscoll et al. 2007). Both CAF-1 and Asf1p physically interact with the histone H4-K16-specific acetyltransferase complex SAS-I (Meijsing and Ehrenhofer-Murray 2001; Osada et al. 2001). Deletion of genes encoding the SAS-I subunits SAS2, SAS4, or SAS5 alters the distribution of Sir proteins on chromosomes and results in silencing defects (Reifsnyder et al. 1996; Ehrenhofer-Murray et al. 1997; Xu et al. 1999a,b; Meijsing and Ehrenhofer-Murray 2001; Osada et al. 2001; Kimura et al. 2002; Suka et al. 2002).
Recently, a second histone acetyltransferase that interacts with Asf1p has been identified. This acetyltransferase, Rtt109p, acetylates lysine 56 on histone H3 (Driscoll et al. 2007; Han et al. 2007a; Tsubota et al. 2007). Acetylation of K56 on histone H3 by Rtt109p is cell cycle regulated and peaks during S phase (Masumoto et al. 2005; Xu et al. 2005; Celic et al. 2006; Maas et al. 2006; Recht et al. 2006; Zhou et al. 2006; Driscoll et al. 2007), implying that this modification may be linked to DNA replication. Acetylation of K56 on histone H3 by Rtt109p is stimulated by Asf1p and by a second Rtt109p-binding partner, Vps75p, in vitro (Han et al. 2007b,c; Tsubota et al. 2007). In vivo, both RTT109 and ASF1 are required for H3-K56 acetylation, whereas H3 K56 remains acetylated in vps75 mutants (Celic et al. 2006; Recht et al. 2006; Schneider et al. 2006; Adkins et al. 2007; Driscoll et al. 2007; Han et al. 2007a,c). Recent reports also indicate that modulation of H3 K56 acetylation can lead to defects in silencing (Hyland et al. 2005; Xu et al. 2007), suggesting that ASF1-dependent silencing phenotypes may be related to acetylation of K56 on histone H3 in addition to acetylation of K16 on histone H4. In addition, work from several groups has led to the identification of a pathway involving H3 K56 acetylation, RTT109, and ASF1 in maintaining genome integrity in response to DNA damage (Masumoto et al. 2005; Ozdemir et al. 2005; Celic et al. 2006; Maas et al. 2006; Recht et al. 2006; Schneider et al. 2006; Agez et al. 2007; Driscoll et al. 2007; Han et al. 2007a,b,c; Tsubota et al. 2007; Jessulat et al. 2008). The findings that (1) acetylation of K56 on histone H3 affects DNA damage sensitivity and silencing, (2) a DNA-damaging agent prevents silent chromatin formation (Kirchmaier and Rine 2006; see also Miller and Nasmyth 1984; Fox et al. 1997), and (3) checkpoint activation in response to DNA damage alters Sir localization and silencing (Martin et al. 1999; McAinsh et al. 1999; Mills et al. 1999; Sharp et al. 2005) have motivated us to explore the relationship between histone modification, silencing, DNA replication, and responses to DNA damage.
To understand the contribution of DNA replication to H3 K56 acetylation and silencing, we examined whether the DNA polymerase processivity factor PCNA could influence H3 K56 acetylation through its role in directing the activity of chromatin assembly factors. Defects in telomeric silencing in pol30-8 mutants have previously been linked to a CAF-1-dependent pathway (Zhang et al. 2000; Sharp et al. 2001), whereas silencing defects in pol30-6 and pol30-79 mutants are primarily defective in an Asf1p-mediated pathway (Sharp et al. 2001). Here, we describe evidence for a pathway linking PCNA to silencing via Asf1p, Rtt109p, and lysine 56 on histone H3. PCNA mutants have defects in H3 K56 acetylation, and mutations in H3 K56 or RTT109 result in silencing and DNA repair phenotypes that overlap with those of asf1 and pol30 mutants. Our experiments argue that acetylation of H3 K56 is associated with DNA replication.
MATERIALS AND METHODS
Plasmids and strains:
Strains used in this study are listed in Table 1. Yeast strains were generated using standard genetic techniques, including homologous recombination and genetic crosses. Deletions of noted open reading frames were generated by one-step gene conversion, and strains expressing histone or pol30 mutants were generated by plasmid shuffling (Stearns et al. 1990; Wach et al. 1994; Adams et al. 1997; Goldstein and McCusker 1999).
Plasmids used in this study are listed in Table 2. Plasmids containing histone mutants were generated by site-directed mutagenesis according to the Quick Change site-directed mutagenesis kit protocol (Stratagene, La Jolla, CA) using oligonucleotides listed in supplemental Table 1 and then confirmed by sequencing. Plasmid AK965 (H3 K56R H4) was generated from pMP3 (Kelly et al. 2000) using oligonucleotides oALK642 and oALK643; plasmid AK972 (H4 K16R) was generated from pMP3 using oALK593 and oALK594; plasmid AK973 (H3 K56Q H4) was generated from pMP3 using oALK691 and oALK692; plasmid AK981 (H3 K56R H4 K16R) was generated from AK965 using oALK593 and oALK594; and plasmid AK982 (H3 K56Q H4 K16R) was generated from AK973 using oALK593 and oALK594.
Total RNA was isolated from logarithmically growing cells, and yFR057w mRNA levels relative to control SCR1 levels were analyzed by quantitative real-time PCR as described previously and in Figure 3 and supplemental Figure 1 legends (Schmitt et al. 1990; Kirchmaier and Rine 2001; Yang and Kirchmaier 2006).
Extraction of chromatin-associated histones:
Histones were isolated from the chromatin fraction of yeast nuclei on the basis of previous methods (Edmondson et al. 1996; Strahl et al. 1999). To isolate chromatin-associated histones, 200-ml cultures of yeast were grown in rich medium (YPD) to an OD600 of 0.8–1.5. Cells were harvested and resuspended in 3 ml of spheroplasting buffer [1 m sorbitol, 50 mm potassium phosphate, pH 6.5, 14 mm β-mercaptoethanol plus 5 mg/ml final concentration yeast lytic enzyme (ICN)] and incubated at 30° for ∼30 min. Spheroplasted cells were washed with spheroplasting buffer, lysed in 5 ml of lysis buffer (18% Ficoll 400, 20 mm potassium phosphate, pH 6.8, 1.0 mm MgCl2, 0.5 mm EDTA, 1.0 mm PMSF, and 1.0 μg/ml each of leupeptin and pepstatin) using a Dounce homogenizer, and then nuclei were isolated by ultracentrifugation at 50,000 × g for 30 min. Nuclei were resuspended and lysed in 0.3 ml RIPA buffer (150 mm NaCl, 50 mm Tris–HCl, pH 7.5, 1% NP-40, 0.5% Na deoxycholate, and 0.1% SDS) and vortexed vigorously. The chromatin fraction was isolated by centrifugation at 16,000 × g for 10 min, washed twice with buffer A (10 mm Tris–HCl, pH 8.0, 0.5% NP-40, 75 mm NaCl) and twice with buffer B (10 mm Tris–HCl, pH 8.0, 400 mm NaCl), and then stored at −80°.
Protein blot analyses:
For analysis of histone modifications, chromatin fractions from each strain (Figure 1E and Table 3) or whole-cell extracts (Figure 1G and Table 4) (Rusché et al. 2002) were separated on 15% SDS–polyacrylamide gels and transferred to PVDF membranes (Bio-Rad, Hercules, CA). Membranes were blocked with a 1:1 dilution of Odyssey blocking buffer (Li-Cor Biosciences) and 1× phosphate-buffered saline (137 mm NaCl, 2.7 mm KCl, 5.4 mm NaH2PO4, 1.47 mm KH2PO4). Protein blots were probed with anti-acetyl-histone H3 (K56) antibodies (1:10,000; Upstate) using Alexa Fluor anti-rabbit IgG (H + L) as the secondary antibody (1:20,000; Molecular Probes, Eugene, OR). Membranes were stripped with 0.2 m NaOH at room temperature as described previously (Gallagher et al. 2004) prior to reprobing with anti-acetyl histone H4 (K16) antibodies (1:10,000; Upstate) or anti-histone H3 antibodies (1:20,000; Abcam). Antibodies were diluted in 50% Odyssey blocking buffer, 0.5× PBS, and 0.1% Tween 20 prior to use. Blots were analyzed using an Odyssey infared imager and Odyssey software v1.2 according to the manufacturer's instructions (Li-Cor Biosciences, Odyssey User Guide, version 1.2). Data were analyzed by the Wilcoxon rank-sum test with MSTAT v.2.6. (http://mcardle.oncology.wisc.edu/mstat).
Chromatin immunoprecipitation (ChIP) experiments were performed on yeast that had been grown logarithmically in rich medium (YPD), and samples were analyzed by real-time PCR using an ABI Prism 7000 as described previously (Rusché et al. 2002; Kirchmaier and Rine 2006; Yang and Kirchmaier 2006). Anti-Sir3 and anti-Sir2 protein antibodies and oligonucleotides used for real-time PCR have been described previously (Rusché et al. 2002; Kirchmaier and Rine 2006; Yang and Kirchmaier 2006).
Mating and colony color assays:
Patch mating assays were performed in duplicate and quantitative mating assays were performed in triplicate as described in van Leeuwen and Gottschling (2002) and in Yang and Kirchmaier (2006) and in Figure 1, Table 5, and supplemental Tables 2 and 3. Quantitative data were analyzed by the Wilcoxon rank-sum test. Images were taken with an Alpha Innotech imager and ChemiImager 5500 v2.02 software. Colony color assays were conducted using two independent yeast strains for each genotype and were performed as outlined in Figure 4 and as described previously (van Leeuwen and Gottschling 2002). To monitor silencing, logarithmically growing yeast containing ADE2 integrated between the E and I silencers at HMR were plated on YPD plates, incubated at 30° for 2 days, and stored at 4° for 3 days before acquiring images using a Leica MZ125 microscope and SPOT 4.1.1 imaging software. Red colonies indicate that ADE2 at HMR is silenced, white colonies indicate that ADE2 is expressed, pink colonies indicate a defect in maintaining or inheriting silencing of ADE2 at HMR, and sectored colonies indicate a defect in establishing silencing.
Analysis of sensitivity to DNA-damaging agents:
Yeast were grown logarithmically in rich medium (YPD) and diluted to ∼1 × 104 cells/μl. Then 3 μl of 10-fold serial dilutions of cultures were plated onto YPD, YPD containing either 0.02 or 0.005% methyl methanesulfonate (MMS), 50 mm hydroxyurea, or 1.5 milliunits/ml bleomycin or onto YPD and exposed to 50 J/m2 UV and then wrapped in foil and maintained in darkness. Yeast were incubated for 2 days at 30° prior to photographing.
Loss of histone H3 K56 acetylation restores silencing at HMRae**:
Several studies have recently implicated Rtt109p and its binding partner Asf1p in the acetylation of lysine 56 on histone H3 (Celic et al. 2006; Recht et al. 2006; Schneider et al. 2006; Adkins et al. 2007; Driscoll et al. 2007; Han et al. 2007a,b,c; Tsubota et al. 2007). Here we tested for links between acetylation of K56 on histone H3 and silencing by asking if loss of acetylation at this residue could mimic several ASF1-dependent silencing phenotypes.
We first examined silencing at HMRae**. HMRae** has a mutated E silencer that contains an ORC-binding site and point mutations in the Rap1p- and Abf1p-binding sites (Figure 1A) (Kimmerly et al. 1988). This crippled e** silencer was defective in recruiting Sir proteins to HMR (Figure 2A) and the intact HMR-I silencer does not serve as a site for Sir protein loading (Rusché et al. 2002). This silencing defect at HMRae** was suppressed in asf1Δ cells (Figure 1B and supplemental Table 2; see also Ehrenhofer-Murray et al. 1999; Meijsing and Ehrenhofer-Murray 2001). Similarly, cdc44-5 and the PCNA mutants pol30-6 (Figure 1B and supplemental Table 2; see also Ehrenhofer-Murray et al. 1999), pol30-8, and pol30-79 (data not shown) also restored silencing at HMRae**. Significantly, the proteins encoded by these genes are all known to interact with one another. CDC44 encodes Rfc1p, the large subunit of the PCNA clamp loading complex RF-C. RF-C physically interacts with Asf1p (Franco et al. 2005). And pol30-6 and pol30-79 have silencing defects that overlap with an ASF1-dependent pathway (Sharp et al. 2001). These results implied that a common mechanism(s) was responsible for restoring silencing at HMRae** in these mutants.
Previously, ASF1- and POL30-dependent effects on silencing have been linked to HIR1 (Sharp et al. 2001; Daganzo et al. 2003). The Hir1/HIRA family of proteins bind to Asf1p and function in replication-independent chromatin assembly (Sharp et al. 2001; Sutton et al. 2001; Ray-Gallet et al. 2002; Daganzo et al. 2003; Tagami et al. 2004). To determine whether loss of HIR1 also restored silencing at HMRae**, we monitored silencing at HMRae** in hir1Δ mutants by patch and quantitative mating assays (Figure 1B and supplemental Table 2). In contrast to asf1Δ and pol30 mutants, the silencing defect at HMRae** was not suppressed in hir1Δ mutants. These results implied that a HIR1-independent pathway had restored silencing at HMRae** in the asf1Δ and pol30 mutants. Therefore, we next monitored silencing in MATα HMRae** cells lacking RTT109 as acetylation of histone H3 on K56 occurs through a pathway requiring ASF1 and RTT109, but not HIR1 (Recht et al. 2006). As with asf1Δ, cdc44-5, and pol30 mutants, silencing was restored in rtt109Δ mutants (Figure 1C) that were defective in acetylating K56 on histone H3 (Figure 1G). Mutation of lysine 56 on histone H3 to arginine to mimic the hypoacetylated form of this residue similarly rescued silencing at HMRae** (Figure 1D). In contrast, mutation of lysine 56 to glutamine to mimic the hyperacetylated state could not rescue silencing (supplemental Table 3).
To assess whether mutations in POL30 that restored silencing at HMRae** could also affect acetylation levels of K56 on histone H3, we isolated chromatin-associated histones from POL30 and pol30 cells and monitored acetylation of K56 on histone H3 in these strains by quantitative protein blot analyses (Figure 1E and Table 3). H3 K56 acetylation levels were significantly reduced in pol30-8, pol30-6, and pol30-79 mutants relative to POL30 cells (P = 0.018 for each pol30 mutant relative to wild type). This reduction in acetylation was not likely caused by a smaller fraction of the population of pol30 cells being in S phase in the cultures. pol30-8 and pol30-79 mutants have similar flow cytometry profiles as wild type and the fraction in S phase in pol30-6 mutants is increased relative to wild type (Zhang et al. 2000). The level of acetylation of K56 on chromatin-associated histone H3 was also reduced in cells lacking CAC1 (Table 3) and H3 K56 acetylation was not detected in asf1Δ mutants in control experiments (Table 3; see also Celic et al. 2006; Recht et al. 2006; Schneider et al. 2006; Adkins et al. 2007; Driscoll et al. 2007; Han et al. 2007a). As transcription-coupled chromatin assembly of H3 K56ac by Asf1p occurs readily (Rufiange et al. 2007), this pathway was likely responsible for most of the chromatin-associated H3 K56ac remaining in the pol30 mutants.
We next examined histone H3 K56 acetylation in whole-cell extracts from wild-type cells vs. pol30, cac1Δ, and asf1Δ mutants (Table 4). Total H3 K56 acetylation levels were similar in both POL30 and pol30 cells. This observation, together with the data in Table 3, implied that histone H3 could be acetylated efficiently in the pol30 mutants, but the pol30 mutants had a defect in incorporating H3 K56ac into chromatin. In contrast, total H3 K56 acetylation levels were reduced in cac1Δ relative to wild-type cells (P = 0.018), consistent with a previous report that histone H3 copurifying in a CAF-1-histone complex is acetylated on K56 (Zhou et al. 2006) (see discussion). In the past, the use of a qualitative rather than a quantitative approach to monitor H3 K56 acetylation likely prevented the detection of a defect in acetylation in cac1 mutants (Recht et al. 2006). Finally, H3 K56 acetylation was not detected whole-cell extracts from asf1Δ mutants (Table 4; see also Celic et al. 2006; Recht et al. 2006; Schneider et al. 2006; Adkins et al. 2007; Driscoll et al. 2007; Han et al. 2007a). Together, these results implied that mutations in POL30 could lead to a defect in replication-coupled chromatin assembly of H3 K56ac.
Restoration of silencing at HMRae** has previously been linked to loss of acetylation at K16 of histone H4 in the absence of the H4 K16-specific histone acetyltransferase Sas2p (Figure 1F; see also Reifsnyder et al. 1996; Ehrenhofer-Murray et al. 1997; Meijsing and Ehrenhofer-Murray 2001). As both CAF-I and Asf1p physically interact with SAS-I (Meijsing and Ehrenhofer-Murray 2001; Osada et al. 2001), it was possible that acetylation of K16 on histone H4 could either influence or be influenced by the acetylation status of K56 on histone H3. Therefore, we compared the acetylation levels of H3 K56 and H4 K16 in wild-type cells to cells expressing either H4 K16R or H3 K56R and to cells lacking RTT109 using protein blots (Figure 1G). This analysis indicated that neither acetylation of K56 on histone H3 nor acetylation of K16 on histone H4 was required for acetylation of the other lysine residue.
The above observations implied that sas2 and rtt109 might be functioning separately to restore silencing. To test this possibility, we examined silencing at HMR and HMRae** in rtt109Δ cells expressing either wild-type histones H3 and H4 or several mutants of K56 on histone H3 and of K16 on histone H4 (Figure 1H and Table 5). Neither H3 K56R nor H4 K16R disrupted silencing in rtt109Δ cells. In contrast, H3 K56Q disrupted rtt109Δ-dependent silencing at HMRae** (Figure 1H; see also Figure 2B). Combining H4 K16R with H3 K56Q could not restore silencing at HMRae** in rtt109Δ mutants (Figure 1H). Quantitative mating assays indicated that H3 K56R and H4 K16R mutants could restore silencing at HMRae** to a similar degree and that mating was enhanced in H3 K56R H4 K16R relative to H3 K56R mutants (P = 0.025). Mating was also more efficient in rtt109Δ H4 K16R relative to H4 K16R mutants (P = 0.063), whereas no difference in mating was observed in rtt109Δ H3 K56R relative to H3 K56R mutants (P = 0.25) (Table 5; see also supplemental Table 3). Together, these results implied that restoration of silencing at HMRae** in cells lacking RTT109 occurred through loss of acetylation of K56 on histone H3.
Loss of histone H3 K56 acetylation restores Sir recruitment to HMRae**:
In the above experiments, mating would have been restored to MATα HMRae** cells if rtt109Δ, asf1Δ, and pol30-6 disrupted transcription from the a1 promoter at HMR in a Sir-independent manner or if the recruitment of Sir proteins to the crippled HMR locus had been rescued. In separate analyses, we have found that a1 mRNA was expressed efficiently in the mutants in the absence of Sir proteins (our unpublished observations). Here, we monitored Sir2p and Sir3p association at HMRae** in cells expressing H3 K56R or in rtt109Δ, asf1Δ, and pol30-6 mutants by chromatin immunoprecipitation (Figure 2B vs. Figure 2A). Expression of H3 K56R or deletion of RTT109 restored the recruitment of Sir proteins to HMRae**. In contrast, expression of H3 K56Q in rtt109Δ cells disrupted Sir protein association. This was consistent with loss of acetylation of K56 on histone H3 playing a key role in Sir protein recruitment (Figure 2B; see also Figure 1H). Sir protein recruitment to the crippled e** silencer was similarly restored in asf1Δ and pol30-6 mutants. However, Sir protein spreading across HMR was compromised in pol30-6 relative to rtt109Δ mutants. This result indicated that the pol30-6 mutation likely led to additional defect(s) that adversely affected silencing, such as a defect in a HIR1-related function of ASF1 (see Sharp et al. 2001; Daganzo et al. 2003; Huang et al. 2007). Together, our results revealed that the common mechanism by which these mutants restored mating to MATα HMRae** cells was through facilitating the localization of Sir proteins to HMR.
Histone H3 K56 mutants disrupt telomeric silencing:
Asf1p was originally identified by its ability to disrupt silencing when overexpressed (Le et al. 1997; Singer et al. 1998). To assess whether the Asf1p-overexpression phenotype was linked to the acetylation status of K56 on histone H3, we compared silencing of yFR057w at telomere VIR in cells expressing wild-type histone H3, H3 K56R, or H3 K56Q in the absence or presence of overexpression of ASF1 (Figure 3, A and B). In this analysis, yFR057w was partially derepressed in cells expressing H3 K56Q relative to cells expressing either wild-type histone H3 or H3 K56R, similar to the telomeric and rDNA silencing phenotypes observed in H3 K56Q, but not in H3 K56R mutants (Hyland et al. 2005). While this article was in preparation, Xu et al. (2007) reported telomeric silencing defects in H3 K56Q mutants and, in contrast to Figure 3A, also observed telomeric silencing defects in H3 K56R mutants. The reason for this difference between these studies is not clear. As expected, overexpression of ASF1 in cells expressing wild-type histones H3 and H4 partially disrupted silencing of yFR057w (Figure 3, A and B). yFR057w was further derepressed in cells overexpressing ASF1 in the presence of either H3 K56R or H3 K56Q. These results implied that, although K56 on histone H3 influenced telomeric silencing, loss of silencing upon overexpression of ASF1 was not exclusively due to neutralization of a positive charge at residue 56 on histone H3 and may have also been linked to Sas2p-dependent acetylation of K16 on histone H4 or to another function of ASF1.
Mutations in histone H3 K56 result in silencing defects at HMR∷ADE2:
To test the effects of mutations in K56 on histone H3 on silencing of an ADE2 reporter gene at HMR, we generated yeast lacking chromosomal copies of histone H3 and H4 genes and expressing either wild-type or mutant histones H3 and H4 from a plasmid. Cells expressing wild-type histone H3 and H4 were silenced at HMR∷ADE2 and grew as red colonies whereas cells expressing either H3 K56R or H3 K56Q formed pink colonies, indicating that these mutants had mild defects in maintaining silencing (Figure 4A). We then examined silencing in analogous strains lacking ASF1 or CAC1. asf1Δ cells expressing wild-type histones H3 and H4 were defective in maintaining silencing and grew as light pink colonies. cac1Δ cells expressing wild-type histone H3 and H4 also exhibited silencing defects and grew as red and white sectored colonies. In contrast to the genetic background analyzed here, asf1 HMR∷ADE2 yeast with two chromosomal copies of genes encoding histones H3 and H4 grow as dark pink colonies and those lacking cac1 grow as pink colonies (Tyler et al. 1999). The more severe silencing defects observed here are likely due to the reduced copy number of histone H3 and H4 genes in our strains. Mutations in lysine 56 on histone H3 to either arginine or glutamine did not further disrupt silencing in asf1Δ cells. In contrast, cac1Δ cells expressing either H3 K56R or H3 K56Q grew as white colonies as silencing of ADE2 at HMR was lost (Figure 4A). The silencing defects of cac1 cells expressing H3 K56 mutants were similar to those reported for asf1 cac1 mutants (Tyler et al. 1999) and implied that loss of CAC1 and mutation of K56 on histone H3 affected silencing of HMR∷ADE2 through different pathways.
We then examined synthetic interactions between histone H3 K56 mutants and pol30 mutants with silencing phenotypes. The pol30p mutants, encoded by pol30-6, pol30-8, and pol30-79, all have defects in both physically interacting with Cac1p and silencing HMR∷ADE2. But, the silencing defects at telomeres in pol30-6 and pol30-79 mutants have been primarily linked to an ASF1-dependent pathway whereas those of pol30-8 mutants are associated with a CAC1-dependent pathway (Zhang et al. 2000; Sharp et al. 2001). To determine whether the silencing defects at HMR∷ADE2 in the pol30 mutants were related to K56 on histone H3, we combined histone H3 K56 mutations with pol30-6, pol30-8, or pol30-79 and monitored silencing of ADE2 integrated at HMR (Figure 4B). As had been observed in Figure 4A, HMR∷ADE2 was partially derepressed in POL30 cells expressing either H3 K56R or H3 K56Q (Figure 4B). pol30-6 mutants expressing wild-type histone H3 formed uniform light pink colonies, indicating a defect in maintaining silencing. Yeast colonies remained light pink in pol30-6 cells expressing either H3 K56R or H3 K56Q, but pol30-6 cells expressing H3 K56R exhibited severe growth defects relative to either single mutant (see also Figure 5A). The colony colors of pol30-79 mutants expressing wild-type histone H3 varied from light pink to dark pink as did those of pol30-79 mutants expressing either H3 K56R or H3 K56Q. Like pol30-6, pol30-79 cells expressing H3 K56R also exhibited growth defects relative to pol30-79 mutants (see also Figure 5A). pol30-8 mutants expressing wild-type histone H3 grew as red and white sectored colonies, indicating that pol30-8 mutants existed in two semistable populations in which HMR was either silenced or expressed (see also Zhang et al. 2000). In contrast to pol30-6 and pol30-79 mutants, when pol30-8 was combined with either H3 K56R or H3 K56Q, yeast colonies were white, indicating that silencing of HMR∷ADE2 had been disrupted (Figure 4B). This observation was similar to previous reports of synergistic silencing defects observed in pol30-8 asf1Δ cells relative to pol30-8 mutants (Sharp et al. 2001) as well as those seen for the cac1Δ double mutants (Figure 4A).
We also monitored silencing at HMR∷ADE2 in cells lacking RTT109. We did not observe dramatic defects in silencing in rtt109Δ cells expressing wild-type or mutant histone H3 (Figure 4A). In contrast to pol30-8 cells expressing histone H3 mutants, silencing was similar in rtt109Δ pol30-8 mutants and in rtt109Δ POL30 cells. rtt109Δ pol30-6 cells, however, displayed more severe silencing and growth defects (Figure 4C; see also Figure 5B). Together, these results indicated that mutation of K56 on histone H3 directly or indirectly led to subtle structural changes to chromatin that were not compatible with silencing HMR∷ADE2, whereas identical alterations in chromatin structure may not have occurred in rtt109Δ mutants.
Effects of histone H3 K56 on silencing vs. sensitivity to DNA damage:
In addition to restoring silencing at HMRae**, deletion of ASF1 can lead to silencing defects (Tyler et al. 1999; Daganzo et al. 2003; Mousson et al. 2005; Figures 3C and 4) as well as hypersensitivity to DNA-damaging agents (e.g., Tyler et al. 1999). Previously, silencing phenotypes in asf1 mutants have been attributed to both SAS2- and HIR1-dependent pathways (Meijsing and Ehrenhofer-Murray 2001; Osada et al. 2001; Sharp et al. 2001; Krawitz et al. 2002; Daganzo et al. 2003). However, asf1-dependent defects in silencing do not always correlate with loss of H4 K16 acetylation (Huang et al. 2007) and, under some conditions, can be genetically separated from HIR1 (Sharp et al. 2001; Krawitz et al. 2002; Daganzo et al. 2003; Mousson et al. 2005; see also Figure 1B). HIR1-independent, ASF1-dependent acetylation of K56 on histone H3 K56 has been tied to cellular responses to DNA damage (e.g., Recht et al. 2006) and H3 K56 acetylation has been linked to silencing (Hyland et al. 2005; Xu et al. 2007; Figures 1–4⇑⇑). To assess whether the silencing phenotypes at HMRae** were related to those at telomeres or to sensitivity to DNA damage, we examined the effects of H3 K56 on telomeric silencing and DNA damage sensitivity in the asf1, cac1, rtt109, and pol30 mutants.
To assess whether K56 on histone H3 contributed to telomeric silencing defects in asf1Δ mutants, we compared silencing of yFR057w at telomere VIR in wild-type cells to asf1Δ, cac1Δ, and rtt109Δ mutants expressing wild-type histone H3, H3 K56R, or H3 K56Q (Figure 3, C and D). Both asf1Δ and cac1Δ mutants expressing wild-type histone H3 had mild silencing defects at yFR057w. Expression of H3 K56R in the asf1Δ mutants did not further disrupt silencing. In contrast, as at HMR∷ADE2, the defect in silencing yFR057w was more severe in cac1Δ mutants expressing H3 K56R relative to wild-type histone H3 (Figure 3C), analogous to the silencing defects of asf1 cac1 mutants (Tyler et al. 1999). Loss of RTT109 did not disrupt silencing at yFR057w in cells expressing wild-type histone H3 and did not alter the telomeric silencing defects of H3 K56Q mutants (Figure 3D). Thus, in the absence of other defects, the lack of RTT109-mediated acetylation on K56 of histone H3 was compatible with telomeric silencing as well as with silencing at HMRae**.
We then tested the sensitivity of the above strains to the DNA-damaging agents MMS, hydroxyurea (HU), bleomycin (BLM), and ultraviolet light (UV) (Figure 6A). MMS is an alkylating agent that induces single- and double-stranded breaks (Schwartz 1989), HU is an inhibitor of ribonucleotide reductase that leads to defects in replication fork progression (Krakoff et al. 1968), BLM induces double-strand breaks (Povirk 1996), and exposure to UV results in the formation of thymidine dimers (Beukers and Berends 1960). Confirming previous reports, mutation of histone H3 to either K56R or K56Q resulted in sensitivity to MMS, HU, and BLM, with H3 K56R being far more sensitive than H3 K56Q (Masumoto et al. 2005; Ozdemir et al. 2005; Maas et al. 2006; Recht et al. 2006; Schneider et al. 2006; Driscoll et al. 2007; Han et al. 2007a,c). Histone H3 K56R mutants also exhibited a mild growth defect upon exposure to UV (Figure 6A; see also Masumoto et al. 2005; Ozdemir et al. 2005; Han et al. 2007c).
asf1Δ mutants expressing either wild-type or H3 K56R exhibited growth defects on rich media and were hypersensitive to all forms of DNA damage tested. In contrast, H3 K56Q mutants could largely suppress asf1-dependent growth defects on rich medium as well as sensitivity to MMS, HU, BLM, and UV (Figure 6A; see also Recht et al. 2006; Agez et al. 2007). H3 K56Q mutants also partially suppressed rtt109-dependent growth defects in rich medium as well as sensitivity to MMS, HU, BLM, and UV (Figure 6B). Thus, whereas the restoration of Sir protein recruitment and silencing at HMRae** (Figures 1 and 2) and hypersensitivity to DNA-damaging agents (Figure 6) correlated well with defects in acetylation of K56 (Tables 4 and 5) in these mutants, the loss of telomeric silencing (Figure 3) did not.
We next assessed telomeric silencing and DNA damage sensitivity in the pol30 mutants. Relative to POL30 cells, pol30 mutants had mild telomeric silencing defects that could generally be enhanced by mutating H3 K56 (supplemental Figure 1, A and B). The telomeric silencing defect of pol30-6 and pol30-79 mutants, which have defects in an ASF1-dependent pathway (Sharp et al. 2001), could not be suppressed by expression of histone H3 K56Q (supplemental Figure 1B). And, similar to HMR∷ADE2 (Figure 4), the telomeric silencing defects of pol30 mutants were also not enhanced by deletion of RTT109 (supplemental Figure 1A).
The pol30 mutants were sensitive to DNA-damaging agents to varying degrees relative to POL30 cells (Figure 5, A and B; see also Ayyagari et al. 1995; Eissenberg et al. 1997; Zhang et al. 2000). This sensitivity was enhanced when combined with either histone or rtt109 mutations. pol30-6 mutants were more sensitive to MMS, HU, BLM, and UV than POL30 cells whereas pol30-8 and pol30-79 were primarily sensitive to high levels of MMS (Figure 5, A and B). Both pol30 mutants were, however, somewhat less sensitive to DNA-damaging agents than asf1 mutants (Figure 5 vs. Figure 6A). This may, in part, reflect residual activity of the ASF1-dependent pathway in pol30 mutants or continued replication-independent chromatin assembly of acetylated H3 K56 by Asf1p in these mutants (Figure 1E and Tables 3 and 4). Supporting this notion, deletion of RTT09, like expression of H3 K56R, in pol30-6 or pol30-79 mutants increased their sensitivity to MMS, HU, and BLM (Figure 5, B and A, respectively). pol30-6 and, especially, pol30-79 mutants expressing H3 K56Q were also more sensitive to MMS, HU, and BLM than either single mutant (Figure 5A). Like cac1Δ mutants (Figure 6A), pol30-8 cells expressing H3 K56Q were less sensitive to DNA-damaging agents than pol30-8 cells expressing H3 K56R (Figure 5A). However, under most conditions, the sensitivity to DNA-damaging agents could not be completely suppressed by mimicking the hyperacetylated state of K56 on histone H3. This implied an alteration in chromatin structure caused by the histone mutation or an inability to reverse the acetylated state may have interfered with normal repair processes. In summary, as in asf1Δ and rtt109Δ mutants (Figure 6), hypersensitivity to DNA-damaging agents in pol30 mutants (Figure 5) correlated with defects in K56 acetylation (Tables 4 and 5) and with the restoration of Sir protein recruitment and silencing at HMRae** (Figures 1 and 2), but not with telomeric silencing defects (supplemental Figure 1).
In this study, we provide evidence that PCNA can affect silencing by influencing the acetylation status of lysine 56 on histone H3. Maintaining a positive charge at this residue by mutation to arginine, deletion of RTT109, or mutation of POL30 rescued silencing at HMRae** (Figure 1, C and D; Table 5; supplemental Tables 2 and 3), revealing that the unacetylated state of K56 on histone H3 enhances Sir protein interactions with the silencer. However, mutations of K56 on histone H3 also led to defects in maintaining silent chromatin (Figure 3; Figure 4, A and B; supplemental Figure 1), indicating that this residue also plays an important role in the structural changes to chromatin that occur when silent chromatin is formed. By analyzing genetic interactions between histone H3 K56 mutants, pol30, and chromatin assembly factor mutants, we have defined a mechanism by which factors involved in DNA replication contribute to silencing (Figures 1 and 2) in addition to the cell's ability to respond to DNA damage (Figures 5 and 6). On the basis of these findings, we propose that DNA replication factors can differentially contribute to these processes by affecting acetylation of lysine 56 on histone H3.
Lysine 56 lies within the αN helix region of histone H3 at the junction between the N terminal tail and the histone fold domain (Luger et al. 1997; White et al. 2001). In the nucleosome core particle, neutralization of this positively charged residue likely influences interactions with DNA. Consistent with this function, the in vivo superhelical density of plasmids is altered in asf1 and rtt109 mutants that lack H3 K56 acetylation (Prado et al. 2004; Driscoll et al. 2007). Superhelical density of plasmids is also altered in histone H3 K56R and K56Q mutants (Masumoto et al. 2005). Our observations support a role of lysine 56 on histone H3 in determining chromatin structure at silenced loci.
Other studies have also recently reported a link between lysine 56 on histone H3 and silencing. Histone H3 is hypoacetylated at K56 in silent chromatin (Xu et al. 2007; B. Yang and A. L. Kirchmaier, unpublished results) and Sir2p can deacetylate this residue in vitro (Xu et al. 2007). As K56 on histone H3 is hypoacetylated in cells lacking RTT109, the nucleosomes in rtt109Δ mutants were largely compatible with silencing (Figure 1, C and H; Figure 3D; Figure 4, A and C; supplemental Figure 1). The side chain of arginine retains a positive charge as is found on lysine and, in our experiments, the H3 K56R mutation was also generally tolerated in silent chromatin (Figure 1H, Figure 3, and supplemental Figure 1). Silent chromatin was, however, sensitive to the subtle structural differences between arginine and lysine at this residue in the sensitized cac1Δ and pol30-8 genetic backgrounds (Figure 3C; Figure 4, A and B). Repair of DNA damage caused by BLM was also compromised in pol30-8 mutants expressing H3 K56R relative to pol30-8 rtt109Δ mutants (Figure 5, A and B). One prediction from these observations is that rtt109Δ or H3 K56R mutants will also have severe defects in silencing when combined with cac1 mutants that are defective in interacting with PCNA (Krawitz et al. 2002).
Mutation of K56 to Q on histone H3 can lead to defects in silencing at the HM loci, telomeres, and rDNA locus (Figures 3 and 4; supplemental Figure 1; Hyland et al. 2005; Xu et al. 2007). Although H3 K56Q does not disrupt Sir protein association at telomere VIR (Xu et al. 2007; B. Yang and A. L. Kirchmaier, unpublished results), H3 K56Q can alter both transcription activity and sensitivity to DAM methylase in this region (Figure 3, A and D; supplemental Figure 1; Xu et al. 2007). Interestingly, H3 K56 normally becomes highly acetylated in response to DNA damage (Celic et al. 2006; Maas et al. 2006; Thaminy et al. 2007). Together, our results support a link between H3 K56 acetylation, Rtt109p, Asf1p, and PCNA that is important for repair of DNA damage but is antagonistic to silencing. Future studies should reveal whether hyperacetylation of K56 on histone H3 during response to DNA damage directly leads to the disruption of telomeric silencing (Martin et al. 1999; McAinsh et al. 1999; Mills et al. 1999; Sharp et al. 2005).
Acetylation of K56 on histone H3 is also regulated in a cell-cycle-dependent manner. This occurs primarily by the opposing actions of cell-cycle-regulated expression of Rtt109p and the NAD+-dependent deacetylases Hst3p and Hst4p (Masumoto et al. 2005; Xu et al. 2005; Celic et al. 2006; Maas et al. 2006; Recht et al. 2006; Zhou et al. 2006). Although newly synthesized histone H3 can be acetylated on K56 and loaded onto chromosomes in G1 (Rufiange et al. 2007), bulk H3 K56 acetylation levels remain low in G1 and then rise and peak in S phase before decreasing again in G2/M via deacetylation by Hst3p and Hst4p. In addition to this pattern of regulation, several other observations argue that incorporation of H3 K56ac is, in part, coupled to DNA replication. These findings include (1) the reduction of the levels of H3 K56 acetylation in chromatin in pol30 mutants (Figure 1 and Table 3), (2) the requirement of a chromatin assembly factor for acetylation of K56 on histone H3, and (3) the localization of Asf1p and CAF-I to DNA via RF-C or PCNA and to DNA replication foci (Krude 1995; Shibahara and Stillman 1999; Zhang et al. 2000; Franco et al. 2005; Schulz and Tyler 2006).
A pathway linking POL30 to acetylation of K56 on histone H3 may function in two different ways. Rtt109p may acetylate histone H3 on lysine 56 when (H3/H4) is bound by Asf1p, but prior to Asf1p being recruited to the replication fork during DNA replication. In vitro, Rtt109p can acetylate histone H3 on K56 while bound by Asf1p in the absence of DNA replication, but not while H3 is in nucleosomes or in tetramers bound to DNA (Driscoll et al. 2007; Han et al. 2007b; Tsubota et al. 2007). In one scenario, defects in recruiting Asf1p bound to H3 K56ac to the replication fork in pol30 mutants would lead to a failure to load H3 K56ac onto DNA during replication-coupled chromatin assembly. Alternatively, Rtt109p may also acetylate K56 on histone H3 while Asf1p-(H3/H4) is associated with RF-C (and, potentially, CAF-I) at the replication fork, but before Asf1p unloads (H3/H4) onto DNA. In this case, the defect in acetylation of K56 on histone H3 in pol30 (and, potentially, cac1) mutants would have been caused by a defect in targeting Rtt109p to the replication fork during replication-coupled chromatin assembly.
Both mechanisms may ultimately contribute to the reduction in chromatin-associated H3 K56ac in pol30 mutants (Figure 1E and Table 3). Acetylation events analogous to the first mechanism described above likely permitted transcription-coupled, Asf1p-dependent incorporation of H3 K56ac into chromatin to continue in pol30 mutants. However, reduced levels of H3 K56ac in chromatin in pol30 mutants would not have been expected if transcription-coupled chromatin assembly was the only mechanism for loading H3 K56ac onto the chromosome. We are unaware of reports of functions of POL30 in transcription in yeast. We have found that acetylation of K56 on histone H3 continues to oscillate during the cell cycle in pol30 mutants in α-factor synchronization-release experiments (data not shown). Thus, Asf1p was still weakly associated with the replication fork in pol30 mutants, Asf1p-dependent chromatin assembly of H3 K56ac during transcription was subject to deacetylation by Hst3p and Hst4p during G2/M, or CAF-I also incorporated H3 K56ac during replication. The observations that histone H3 K56 acetylation levels in cac1Δ mutants were reduced (Tables 3 and 4) and that H3 K56ac copurifies with Cac1p support a role of CAF-1 in this process as well (Zhou et al. 2006).
Our observations reveal a pathway linking the restoration of silencing at HMRae** in replication factor mutants to acetylation of K56 on histone H3 (Figure 7). In this pathway (Figure 7A), Rtt109p acetylates K56 on histone H3 in the context of (H3/H4) bound to Asf1p either prior to or upon association with the replication fork. Asf1p bound to acetylated (H3/H4) localizes to the replication fork through interactions between Asf1p and the Rfc1p subunit of RF-C. Additional interactions between Asf1p and the Cac2p subunit of CAF-1 likely stabilize the association of chromatin assembly factors with PCNA and RF-C at the replication fork. Consistent with this notion, Asf1p binds Cac2p in vitro, and cac1p mutants with defects in binding to PCNA have severe silencing defects in the absence of ASF1 (Krawitz et al. 2002). Analogous pathways linking acetylation of K16 on histone H4 by Sas2p to CAF-I and Asf1p at the replication fork may also contribute to the silencing phenotypes of pol30 mutants. In the case of HMRae**, the normal route for Sir protein recruitment to the HMR locus (Figure 7B) has been disrupted through mutation of Rap1p- and Abf1p-binding sites at the E silencer (Figure 7C). We propose that mutations in factors associated with acetylation of K56 on histone H3 and the replication fork restore silencing by rescuing Sir protein recruitment (Figure 7D). This could occur through the creation of high-affinity binding sites for Sir proteins on nucleosomes at HMR, as has been proposed for mutants with defects in H4 K16 acetylation. Alternatively, defects in histone acetylation in these replication mutants may have altered ORC's interactions with the chromosome, thereby enhancing the binding of Sir proteins to ORC. Either possibility has the potential to result in the mislocalization of Sir proteins to other chromosomal loci throughout the genome. Future studies characterizing how chromatin is regulated during cellular responses to DNA damage should clarify how signals during DNA repair also influence epigenetic processes (Martin et al. 1999; McAinsh et al. 1999; Mills et al. 1999; Sharp et al. 2005; Kirchmaier and Rine 2006).
Together, the work presented here implies that nucleosomal DNA that serves as the foundation for epigenetically regulated chromatin can contain inappropriately modified histones when defects in replication-coupled chromatin assembly occur. These observations explain how defects in DNA replication factors can affect epigenetic processes when DNA replication itself is not required to form silent chromatin in yeast (Fox et al. 1997; Kirchmaier and Rine 2001; Li et al. 2001). These findings also raise the possibility that DNA damage phenotypes associated with other replication fork mutants may be related to perturbations in replication-coupled modifications on histones.
We thank Joe Ogas and Paul Kaufman and anonymous reviewers for helpful discussions and comments on this manuscript. We thank Peter Burgers, Paul Kaufman, Ann Ehrenhofer-Murray, and Jasper Rine for reagents, strains, and plasmids. This work was supported by U. S. Department of Agriculture Hatch grant IND053072 (A.L.K.) and the National Science Foundation (A.L.K.). This research was also supported by an American Cancer Society Institutional Research Grant and a Continuing Umbrella of Research Experiences Supplement (Cancer Center Support grant no. CA23168-29) to the Purdue Cancer Center. This is paper no. 2008-18316 from the Purdue University Agricultural Experiment Station.
Communicating editor: L. Pillus
- Received November 13, 2007.
- Accepted March 24, 2008.
- Copyright © 2008 by the Genetics Society of America