HST3/HST4-dependent Deacetylation of Lysine 56 of Histone H3 in Silent Chromatin

Abstract

The composition of posttranslational modifications on newly synthesized histones must be altered upon their incorporation into chromatin. These changes are necessary to maintain the same gene expression state at individual chromosomal loci before and after DNA replication. We have examined how one modification that occurs on newly synthesized histone H3, acetylation of K56, influences gene expression at epigenetically regulated loci in Saccharomyces cerevisiae. H3 K56 is acetylated by Rtt109p before its incorporation into chromatin during S phase, and this modification is then removed by the NAD⁺-dependent deacetylases Hst3p and Hst4p during G2/M phase. We found silenced loci maintain H3 K56 in a hypoacetylated state, and the absence of this modification in rtt109 mutants was compatible with HM and telomeric silencing. In contrast, loss of HST3 and HST4 resulted in hyperacetylation of H3 K56 within silent loci and telomeric silencing defects, despite the continued presence of Sir2p throughout these loci. These silencing defects in hst3Δ hst4Δ mutants could be suppressed by deletion of RTT109. In contrast, overexpression of Sir2p could not restore silencing in hst3Δ hst4Δ mutants. Together, our findings argue that HST3 HST4 play critical roles in maintaining the hypoacetylated state of K56 on histone H3 within silent chromatin.

INTRODUCTION

Newly synthesized histones are decorated with several posttranslational modifications that are altered upon their incorporation into chromatin after DNA replication. The changes acquired after chromatin assembly enable individual chromosomal loci to retain specific patterns of modifications on histones that are compatible with their preexisting states of gene expression. One such evolutionarily conserved modification that is subject to alterations is acetylation of K56 on histone H3.

H3 K56ac is found in diverse eukaryotes, including Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila, and Tetrahymena (Hyland et al., 2005; Masumoto et al., 2005; Ozdemir et al., 2005; Xu et al., 2005; Recht et al., 2006; Schneider et al., 2006; Zhou et al., 2006; Garcia et al., 2007; Xhemalce et al., 2007; Haldar and Kamakaka, 2008). H3 K56 is also conserved in mammals but only low levels of the acetylated form of this residue may be present in human cells (Xu et al., 2005; Garcia et al., 2007). In budding and fission yeast, H3 K56 is acetylated by the recently discovered acetyltransferase Rtt109p (Schneider et al., 2006; Collins et al., 2007; Driscoll et al., 2007; Han et al., 2007a,b,c; Tsubota et al., 2007; Xhemalce et al., 2007). In both organisms, H3 K56ac is regulated throughout the cell cycle by the combined actions of Rtt109p and the NAD⁺-dependent deacetylases Hst3p and Hst4p in S. cerevisiae and their orthologues encoded by rtt109⁺ and hst4⁺ in S. pombe. In budding yeast, H3 K56ac and expression of RTT109 peak during S phase. This modification is later removed in an HST3 HST4-dependent manner during G2 and M 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). HST3 and HST4 are also cell cycle regulated; expression of HST3 and HST4 peaks during G2/M, and M/G1, respectively (Maas et al., 2006). Expression of hst4⁺ and H3 K56ac in S. pombe during the cell cycle is similar (Recht et al., 2006; Xhemalce et al., 2007; Haldar and Kamakaka, 2008).

Rtt109p acetylates K56 on newly synthesized H3, while H3/H4 dimers are bound by the chromatin assembly factor Asf1p (Han et al., 2007b; Tsubota et al., 2007). Once assembled into nucleosomes, H3 K56 becomes positioned near the “entry-exit point” of the DNA double helix on the nucleosome (Luger et al., 1997), making this residue a potential regulatory site for changes in chromatin structure. Consistent with this notion, different charge states at this residue coincide with structural alterations in chromatin. Chromatin isolated from cells expressing H3 K56Q mutants, mimicking the acetylated state at this residue, is more sensitive to digestion by micrococcal nuclease than is chromatin from cells expressing wild-type histones or H3 K56R mutants mimicking the deacetylated state. And, plasmids from cells expressing H3 K56Q are less supercoiled in topology assays than those from cells expressing wild-type histones or H3 K56R (Masumoto et al., 2005), whereas plasmids from rtt109 mutants are more supercoiled than those from wild-type cells (Driscoll et al., 2007). These observations imply the acetylated form of this residue reflects a more “relaxed” state of chromatin.

Also, consistent with the model that this modification may be regulatory in nature, multiple studies have recently reported mutations leading to either a constitutively hypo- or hyperacetylated state at H3 K56 through either mutation of K56 on H3 or deletion of RTT109 or HST3 and HST4 result in defects in chromatin integrity and hypersensitivity to DNA damaging agents (Brachmann et al., 1995; Hyland et al., 2005; Masumoto et al., 2005; Ozdemir et al., 2005; Celic et al., 2006; Maas et al., 2006; Recht et al., 2006; Schneider et al., 2006; Collins et al., 2007; Driscoll et al., 2007; Han et al., 2007a,b,c; Thaminy et al., 2007; Xhemalce et al., 2007; Fillingham et al., 2008; Jessulat et al., 2008; Miller et al., 2008). In this study, we have explored the relationship between H3 K56 and epigenetic gene regulation in S. cerevisiae.

Across phyla, epigenetically silent chromatin is formed through varying processes sharing a common series of steps. These steps include recruitment of the structural components of silent chromatin to specific chromosomal sites, alterations in chromatin modifications to permit propagation of those structural components across large chromosomal regions, plus additional, often poorly understood, events that ultimately result in the heritable inactivation of gene expression. In budding yeast, genes adjacent to telomeres and at the silent mating-type loci are regulated epigenetically through a process called silencing. The structural components of silent chromatin found at the silent mating-type loci, HML and HMR, and at telomeres are the Sir proteins, Sir1-4p. When silent chromatin forms at the HM loci, the Sir proteins are recruited to silencers where they physically interact with DNA binding proteins, including Rap1p, Abf1p, and ORC. Sir proteins are recruited to telomeres in a similar manner, but primarily via multiple Rap1p binding sites (Strahl-Bolsinger et al., 1997; Lieb et al., 2001, and references within). Once recruited to a silencer or the end of a telomere, Sir proteins will propagate along the chromosome by binding nucleosomes (Hoppe et al., 2002; Luo et al., 2002; Rusché et al., 2002, 2003). During Sir spreading, histones H3 and H4 are deacetylated through the enzymatic action of Sir2p, a NAD⁺-dependent histone deacetylase that preferentially removes acetyl groups from K9 and K14 on histone H3 and K16 on histone H4 (Imai et al., 2000; Tanny and Moazed, 2001). Histone deacetylation is critical for silencing and cells expressing catalytically inactive mutants of Sir2p are defective in both Sir spreading and silencing (Imai et al., 2000; Hoppe et al., 2002; Luo et al., 2002; Rusché et al., 2002; Kirchmaier and Rine, 2006; Yang and Kirchmaier, 2006).

Several groups have reported links between H3 K56ac and silencing. In both S. cerevisiae and S. pombe, cells lacking HST3 and HST4 or hst4⁺, respectively, have silencing defects (Brachmann et al., 1995; Freeman-Cook et al., 1999; Grunweller and Ehrenhofer-Murray, 2002; Durand-Dubief et al., 2007). Mutations affecting the charge of H3 K56 also influence silencing (Hyland et al., 2005; Xu et al., 2007; Miller et al., 2008) and Sir2p can deacetylate K56 on histone H3 in vitro (Xu et al., 2007). In this study, we have explored the relationship between acetylation of K56 on histone H3 and silencing. Our findings revealed that HST3 HST4 are required for telomeric silencing and efficient deacetylation of K56 on H3 within silent chromatin.

MATERIALS AND METHODS

Yeast Strains and Plasmids

Yeast strains (Table 1) and plasmids (Table 2) used in this study. Yeast strains AKY4454, AKY4456, AKY4460, AKY4470, and AKY4472 containing URA3 and TRP1 at Tel VIIL were generated by integration of SalI to EcoR1 fragments of pVII-L URA3-TEL or pADTRP1(+) at ADH4 as described previously (Gottschling et al., 1990) and were confirmed by DNA blots (data not shown). pAK966 encoding histone H3 K9,14,56R/H4 K16R was generated from pAK923 (Yang and Kirchmaier, 2006) by site-directed mutagenesis using Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA) following the QuikChange site-directed mutagenesis kit protocol (Stratagene) using oALK642 5′-GAA GAT TCC AAA GAT CTA CTG AAC T-3′ and oALK643 5′-AGT TCA GTA GAT CTT TGG AAT CTT C-3′, then confirmed by sequencing. pAK974 was generated by digesting pAR14 (Holmes et al., 1997) with EcoRI/HindIII, and ligating the 6-kbp fragment from pAR14 containing SIR2 expressed from the GAL10 promoter to EcoRI/HindIII of pRS426 (Christianson et al., 1992). pAK975 was generated by cleavage of pAK974 with SalI and religation of the vector to remove SIR2. pAK996 was generated by ligating a 3-kbp SalI/SacI fragment of ICe72 (pCEN-URA3-HST3) (Celic et al., 2006) containing HST3 into SalI/SacI of pRS415 (Christianson et al., 1992). pAK998 was generated by ligating the same 3-kbp SalI/SacI fragment of ICe72 into XhoI/SacI of pRS426.

Table 1.

Yeast strains used in this study

Strain	Genotype	Source
JRY2726	MATa his4	P. Schatz
JRY2728	MATα his4	P. Schatz
W303	MATa or α ade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100	R. Rothstein
AKY1968	W303 MATα hht1-hhf1Δ::LEU2 hht2-hhf2Δ::HIS3 plus pPK189	Yang and Kirchmaier (2006)a
AKY944	W303 MATa hht1-hhf1Δ::LEU2 hht2-hhf2Δ::HIS3 plus pPK189	Yang and Kirchmaier (2006)a
AKY1101	W303 MATα LEU2::sir2-345 sir2Δ::KanMX hht1-hhf1Δ::LEU2 hht2-hhf2Δ::HIS3 plus pPK189	Yang and Kirchmaier (2006)a
AKY1103	W303 MATa LEU2::sir2-345 sir2Δ::KanMX hht1-hhf1Δ::LEU2 hht2-hhf2Δ::HIS3 plus pPK189	Yang and Kirchmaier (2006)a
AKY3857	AKY1968 rtt109Δ::KanMX	Miller et al. (2008)
AKY3991	AKY944 rtt109Δ::KanMX	This studya
AKY3855	AKY1101 rtt109Δ::KanMX	This studya
AKY3995	AKY1103 rtt109Δ::KanMX	This studya
PKY4220	W303 MATa bar1Δ::his5+hst3Δ::NatMX hst4Δ::KanMX	P. Kaufman
PKY4222	PKY4220 MATα	P. Kaufman
AKY4250	PKY4222 rtt109Δ::HphMX	This study
AKY4534	W303 MATα hht1-hhf1Δ::LEU2 hht2-hhf2Δ::HIS3 hst3Δ::NatMX hst4Δ::HphMX plus pPK189	This studya
AKY4496	W303 MATα LEU2::sir2-345 sir2Δ::KanMX hht1-hhf1Δ::LEU2 hht-2hhf2Δ::HIS3 hst3Δ::NatMX hst4Δ::HphMX plus pPK189	This studya
AKY4454	W303 MATα URA3-TELVII-L	This studya
AKY4456	W303 MATα sir2Δ::TRP1 URA3-TELVII-L	This studya
AKY4470	PKY4222 URA3-TELVII-L	This studya
AKY4460	W303 MATα URA3-TRP1-TELVII-L	This studya
AKY4472	PKY4222 URA3-TRP1-TELVII-L	This studya

^a Parental strains used in this study. See Table 2 for plasmids that were introduced into these strains for experiments described in text.

Table 2.

Plasmids used in this study

Plasmid	Description	Source
pPK189	HHT2 HHF2 ARS/CEN/URA3	P. Kaufman
pMP3	HHT2 HHF2 ARS/CEN/TRP1	Kelly et al. (2000)
pAK965	H3 K56R H4 ARS/CEN/TRP1	Miller et al. (2008)
pAK972	H3 H4 K16R ARS/CEN/TRP1	Miller et al. (2008)
pAK981	H3 K56R H4 K16R ARS/CEN/TRP1	Miller et al. (2008)
pAK923	H3 K9,14R H4 K16R ARS/CEN/TRP1	Yang and Kirchmaier, (2006)
pAK966	H3 K9,14,56R H4 K16R ARS/CEN/TRP1	This study
pFA6::kanMX4		Wach et al. (1994)
pAG32	pFA6::HphMX4	Goldstein and McCusker (1999)
pAG25	pFA6::NatMX4	Goldstein and McCusker (1999)
pRS415	ARS/CEN/LEU2	Christianson et al. (1992)
pRS426	2 μm/URA3	Christianson et al. (1992)
pAK975	GAL10p in pRS426	This study
pAK974	GAL10p SIR2 in pRS426	This study
pAK996	HST3 in pRS415	This study
pAK998	HST3 in pRS426	This study
ICe72	HST3 ARS/CEN/URA3	Celic et al. (2006)
pAR14	GAL10p SIR2 in YEp51	Holmes et al. (1997)
YEp351	2 μm/LEU2	Hill et al. (1986)
pLP349	SIR2 in YEp351	Sherman et al. (1999)
pVII-L URA3-TEL	URA3-TELVII-L	Gottschling et al. (1990)
pADTRP1(+)	URA3-TRP1-TELVII-L	Gottschling et al. (1990)

Chromatin Immunoprecipitation (ChIP)

ChIP experiments were performed as described previously (Rusché et al., 2002; Yang and Kirchmaier, 2006) by using anti-Acetyl-H4 (K16) antibodies (07-329; Millipore, Billerica, MA), anti-Acetyl-H3 (K56) antibodies (07-677; Millipore), anti-H3 antibodies (Ab1791-100; Abcam, Cambridge, MA), anti-Sir2p and anti-Sir3p antibodies (Axelrod, 1991), or normal rabbit serum (100 853; Roche Diagnostics, IN). ChIPs were analyzed by real-time polymerase chain reaction (PCR) by using an ABI Prism 7000 (Applied Biosystems, Foster City, CA) as described previously and in figure legends (Kirchmaier and Rine, 2006; Yang and Kirchmaier, 2006). Oligonucleotides used during analyses have been described previously (Yang and Kirchmaier, 2006). Statistical analyses were performed using the Wilcoxon rank sum test with MSTAT version 2.6. (http://mcardle.oncology.wisc.edu/mstat), and ChIP data are expressed as average ± SD, n = 3. The level of H3 K56ac (normalized to H3) at SSC1 was similar to those at three other control loci, ACT1, MAT, and 7.5 Kb from Tel VIR, in hst3 hst4 mutants, validating the use of SSC1 as an internal control for this study (data not shown).

RNA Analysis

Total RNA was extracted from logarithmically growth cells as described previously (Schmitt et al., 1990; Kirchmaier and Rine, 2001), and levels of a1, yFR057w, HST3, and SCR1 transcripts were analyzed as described previously by quantitative real-time PCR using an ABI Prism 7000 (Applied Biosystems) (Kirchmaier and Rine, 2006; Yang and Kirchmaier, 2006). Oligonucleotides used for the analyses of HST3 expression were oALK825 5′-CAGTCGATCGGGCTCAATGT-3′ and oALK826 5′-TTCATCGTCGGCATCAAGAC-3′. Statistical analyses were performed using the Wilcoxon Rank sum test with MSTAT version 2.6, and transcription data are expressed as average ± SD, n = 3.

Immunoblot Analyses

Immunoblot analyses were performed as described previously (Rusché et al., 2002; Miller et al., 2008). We separated 0.33 OD cell equivalents of whole cell lysates from logarithmically growing cells in 15% (for analysis of histone H3) or 7.5% (for Sir2p) SDS-polyacrylamide gel electrophoresis (PAGE) gels and transferred to polyvinylidene difluoride membranes (162-0177; Bio-Rad). Immunoblots in Figure 4 were probed with anti-acetyl-H3 (K56) antibodies (1:10,000; 07-677, Millipore) or anti-Sir2p (1:10,000; Axelrod, 1991) and with the secondary antibody Alexa Flour anti-rabbit IgG (H+L) (1:20,000; A21109, Invitrogen, Carlsbad, CA). Membranes were stripped (Gallagher et al., 2004) and reprobed with anti-H3 antibodies (1:10,000; 06-755, Millipore) or anti-PGK1 antibodies (1:10,000; A-6457, Invitrogen), respectively, and with the secondary antibodies, Alexa Flour anti-rabbit immunoglobulin G (IgG) (H+L) (as described above) or Alexa Flour anti-mouse IgG (H+L) (1:20,000; A21058, Invitrogen), respectively. Immunoblots were analyzed using an Odyssey Infrared Imager (Li-Cor Biosciences, Lincoln, NE) according to manufacturer's instructions.

Figure 4.

Overexpression of SIR2 does not reduce H3 K56ac or restore silencing in hst3Δ hst4Δ mutants. (A) Immunoblot analyses of Sir2p and H3 K56ac. Whole cell extracts of hst3Δ hst4Δ mutants containing a vector (pAK975) or a SIR2 overexpression plasmid (pAK974) were analyzed by immunoblotting using anti-H3 K56ac or anti-Sir2p antibodies plus anti-H3 or anti-Pgk1p antibodies (internal controls), respectively (see Materials and Methods). Cells were grown to OD 0.5 in SC-Ura with 2% glucose, washed, diluted in SC-Ura with 2% galactose, and incubated at 30°C for 4 h to induce SIR2. Two independent clones for each genotype are shown. (B and C) Overexpression of SIR2 does not reduce H3 K56ac at Tel VIR in hst3Δ hst4Δ mutants. H3 K56ac levels at 0.6 kb (B) and 1.2 kb (C) from the end of Tel VIR were monitored by ChIP in MATα hst3Δ hst4Δ yeast containing a vector or a SIR2 overexpression plasmid. The efficiency of coprecipitation of each locus with anti-H3 K56ac antibodies was determined relative to that with anti-H3 antibodies, then each locus was normalized to hst3Δ hst4Δ mutants containing the vector control, which was set to 1. Data were calculated as 2^{[(H3 K56ac C_T − H3 C_T)_vector − (H3K56ac C_T − H3 C_T)_{SIR2 O/E}]}. (D) Overexpression of SIR2 in hst3Δ hst4Δ mutants does not restore silencing at Tel VIR. Transcript levels of yFR057w relative to SCR1 were determined by quantitative real-time PCR for each strain and then expressed relative to wild type, which was set to 1. Data were calculated as shown in Figure 2. (E and F) Overexpression of SIR2 in hst3Δ hst4Δ mutants does not reduce H3 K56ac at HMR. H3 K56ac levels at the HMR E silencer (E) and a1 (F) were monitored by ChIP as described above.

Mating Assays

Patch mating assays were performed on two independent yeast strains for each genotype as described previously (van Leeuwen and Gottschling, 2002; Yang and Kirchmaier, 2006) and as outlined in Figure 3. In quantitative mating assays, the mating efficiency of each strain relative to wild-type was determined as follows: (colonies on YM plate with indicated tester strain/colonies on YM plate with supplements)_{indicated strain}/(colonies on YM plate with indicated tester strain/colonies on YM plate with supplements)_{wild type}.

Figure 3.

Loss of HST3 and HST4 leads to hyperacetylation of H3 K56 but does not disrupt silencing at HMR. (A) Map of HMR. The regions amplified at HMR are noted. (B) Mating analyses of silencing at HM loci in hst3Δ hst4Δ mutants. Haploid MATα or MATa hst3Δ hst4Δ mutants were grown for 1 d at 30°C on rich medium (YPD) and then tested for silencing at HML and HMR by replica plating to α or a mating-type tester lawns on minimal medium and incubated for 2 d at 30°C. Growth of diploids on minimal medium indicates that HML or HMR was silenced. Relative mating efficiencies from quantitative mating assays are noted on right. The efficiencies of mating of the MATa or MATα HST3 HST4 strains to tester strains JRY2728 (MATα) or JRY2726 (MATa), respectively, was determined relative to their plating efficiency on minimal medium plus supplements (for MATa, 100 ± 6.7%, n = 3; for MATα, 87 ± 17%, n = 3) and were set to 1. The mating efficiencies of hst3Δ hst4Δ mutants were determined similarly and are expressed relative to HST3 HST4 strains (see Materials and Methods), average ± SD, n = 3. (C–E) HST3 and HST4 are required for deacetylation of H3 K56 at HMR. H3 K56ac (C), Sir2p (D), and H4 K16ac (E) levels at HMR in wild-type and hst3Δ hst4Δ strains were monitored by ChIP. The efficiency of coprecipitation of DNA from the HMR E silencer, HMR a1 or SSC1 with anti-H3 K56ac, H4 K16ac, or anti-Sir2p antibodies was determined and data were calculated as shown in Figure 2.

Telomeric Silencing Assays

Yeast were grown logarithmically in minimal medium containing adenine, histidine, tryptophan, and uracil and containing or lacking leucine, and then they were diluted to ∼1 × 10⁴ cells/μl. Three microliters of 10-fold serial dilutions of yeast were plated onto complete synthetic medium; synthetic medium lacking leucine, uracil, leucine, and uracil, or leucine and tryptophan; or medium containing 5-fluorootic acid (5-FOA). Cells were incubated for 2 d at 30°C and then photographed using an Alpha Innotech (San Leandro, CA) imager and ChemiImager 5500 version 2.02 software. In wild-type cell populations, telomeric reporter genes are expressed or silenced in different fractions of the population. Deprepressed subpopulations will express telomeric TRP1 or URA3 reporter genes and grow in the absence of tryptophan or uracil, respectively. Expression of URA3 prevents growth in the presence of 5-FOA due to the conversion of 5-FOA into toxic 5-flurouracil by Ura3p.

RESULTS

Acetylation of K56 on Histone H3 Disrupts Telomeric Silencing

Recently, we have found that mutating K56 to Q on histone H3, which neutralizes the positive charge at this residue, disrupts telomeric silencing, whereas mutating H3 K56 to R, which maintains a positive charge at this residue, is largely compatible with silencing (Miller et al., 2008; see also Hyland et al., 2005; Xu et al., 2007). Consistent with acetylation of K56 on H3 influencing silencing, cells lacking the H3 K56-specific deacetylases encoded by HST3 and HST4 are defective in silencing a telomeric URA3 reporter gene adjacent to Tel IV (Brachmann et al., 1995). Similarly, we found that silencing of a URA3 reporter at Tel VIIL was disrupted in hst3Δ hst4Δ mutants by at least 1 × 10⁴-fold as measured by growth on 5-FOA plates (Figure 1). This silencing defect was similar in magnitude to defects observed in cells lacking SIR2 (Figure 1A). The silencing defect in hst3Δ hst4Δ mutants could be suppressed by ectopic expression of HST3 (Figure 1B) but not by overexpression of SIR2 (Figure 1, A and C). Sir2p has been reported to modulate silencing via deacetylation of H3 K56 (Xu et al., 2007). So, to clarify the role of HST3 HST4 in silencing, we assessed whether HST3 HST4-dependent silencing defects were also related to the acetylation status of H3 K56.

Figure 1.

Overexpression of SIR2 does not restore telomeric silencing in hst3Δ hst4Δ mutants. (A and B) Silencing of URA3 at Tel VII-L. For URA3-TRP1-VII-L, transcription of URA3 occurs toward the telomere. (C) Silencing of TRP1 and URA3 at TelVII-L. For URA3-TRP1-VII-L, transcription of URA3 and TRP1 occur toward the telomere. Plasmids in indicated strains include YEp351 or pLP349 (A and C) and pRS415 or pAK996 (B). Ten-fold serial dilutions of logarithmically growing yeast with the indicated genotypes were plated onto the indicated media and were incubated for 2 d at 30°C before photographing (see Materials and Methods).

To test the effects of H3 K56 acetylation on silencing, we first examined expression of yFR057w, a gene adjacent to Tel VIR (Figure 2A), in hst3Δ hst4Δ mutants by quantitative real-time PCR (Figure 2). yFR057w was derepressed in hst3Δ hst4Δ mutants relative to wild-type strains (Figure 2B), and this silencing defect in the hst3Δ hst4Δ mutants could be suppressed by expression of HST3 from a low copy plasmid (Figure 2B and Supplemental Figure 1). The severity of the silencing defect in hst3Δ hst4Δ mutants was similar to that found in cells expressing a catalytically inactive mutant of Sir2p (see below). These findings indicated HST3 and HST4 were required for silencing yFR057w.

Figure 2.

Loss of HST3 and HST4 disrupts telomeric silencing. (A) Map of Tel VIR. The regions amplified at Tel VIR are noted. (B) Transcription from yFR057w at Tel VIR in hst3Δ hst4Δ mutants. Transcript levels of yFR057w relative to SCR1 (internal control) were determined by quantitative real-time PCR for each strain and then expressed relative to wild type, which was set to 1. Data were calculated as 2^{[(locus C_T − SCR1 C_T)_wild-type − (locus C_T − SCR1 C_T)_a]}, where “a” is the indicated genotype. (C–E) HST3 and HST4 are required for H3 K56 hypoacetylation at Tel VIR. H3 K56ac (C), Sir2p (D), and H4 K16ac (E) levels at Tel VIR in wild-type and hst3Δ hst4Δ strains were monitored by ChIP. The efficiency of coprecipitation of DNA from 0.6 and 1.2 kb from the end of Tel VIR or SSC1 (internal control) with anti-H3 K56ac (C) or H4 K16ac (E) antibodies was determined relative to that with anti-H3 antibodies, and then each locus was normalized to SSC1 which was set to 1. Data were calculated as 2^{[(H3 K56ac or H4 K16ac C_T − H3 C_T)_SSC1 − (H3K56ac or H4 K16ac C_T − H3 C_T)_a]}, where “a” is the indicated locus. For anti-Sir2p ChIPs (D), the efficiency of coprecipitation of each locus with anti-Sir2p antibodies was determined relative to Input and then each locus was normalized to SSC1, which was set to 1. Data were calculated as 2^{[(Sir2p C_T − Input C_T)_SSC1 − (Sir2p C_T − Input C_T)_a]}, where “a” is the indicated locus (see Materials and Methods).

Because deacetylation of K56 on histone H3 is regulated by Hst3p and Hst4p (Celic et al., 2006; Maas et al., 2006) and the silencing defect of the hst3Δ hst4Δ mutants could be suppressed by deletion of RTT109 encoding the H3 K56ac-specific acetyltransferase (Figure 2B), we reasoned that this silencing defect must have been caused by hyperacetylation of H3 K56 at Tel VIR in the hst3Δ hst4Δ mutants. To test this possibility, we examined the acetylation status of K56 on histone H3 at Tel VIR in hst3Δ hst4Δ mutants by ChIP. H3 K56ac levels at Tel VIR were increased in hst3Δ hst4Δ mutants relative to wild-type cells (Figure 2C; see also Xu et al., 2007). Because telomeric silencing requires Sir proteins (Aparicio et al., 1991), it was possible that hyperacetylation of H3 K56 at Tel VIR had disrupted Sir localization in the above-mentioned experiments. To determine whether the silencing defect in hst3Δ hst4Δ mutants resulted from the loss Sir protein binding at Tel VIR, we compared Sir2p association at Tel VIR in wild-type and hst3Δ hst4Δ strains by ChIP (Figure 2D). Surprisingly, Sir2p binding at Tel VIR was similar in both wild-type and hst3Δ hst4Δ strains. And, in contrast to H3 K56ac, the Sir2p-specific substrate H4 K16 remained hypoacetylated at Tel VIR in both wild-type and hst3Δ hst4Δ strains (Figure 2E), despite the loss of silencing of yFR057w (Figure 2B).

hst3 and hst4 mutants have mild defects in silencing a URA3 reporter at HMR (Grunweller and Ehrenhofer-Murray, 2002). Therefore, we next examined silencing at the native HM loci in hst3Δ hst4Δ mutants (Figure 3). In contrast to telomeric silencing, silencing was maintained at the native HM loci in cells lacking HST3 and HST4, as measured by both patch and quantitative mating assays (Figure 3B). However, like at Tel VIR, H3 K56ac levels at HMR increased in hst3Δ hst4Δ mutants relative to wild-type cells (Figure 3C), despite the continued presence of Sir2p (Figure 3D), and H4 K16ac (Figure 3E) levels at HMR having remained similar in both hst3Δ hst4Δ mutants and wild-type cells. Together, these findings indicated HST3 and HST4 were required for efficient deacetylation of H3 K56 at both HMR and Tel VIR, and hyperacetylation of K56 on histone H3 resulted in the disruption of telomeric silencing. The differences in sensitivity of a1 and URA3 at HMR to loss of HST3 and HST4 argue sensitivity to disruption of silencing via hyperacetylation of H3 K56 may be promoter-dependent.

Overexpression of SIR2 Neither Restores Silencing Nor Decreases H3 K56ac in hst3 hst4 Mutants

Our data indicated that deletion of HST3 and HST4 increased the levels of H3 K56ac at both yFR057w and HMR, implying Hst3p and Hst4p were required to maintain the deacetylated state of H3 K56 within silent chromatin. This result was somewhat surprising as Sir2p remained physically present at both loci in the vicinity of H3 K56ac and was able to deacetylate H4 K16 at these sites efficiently (Figures 2 and 3). A possible explanation for this observation was that the amount of Sir2p present at HMR and yFR057w might somehow be limiting in the hst3Δ hst4Δ mutants, and this had adversely affected Sir2p's ability to deacetylate H3 K56. To test this possibility, we overexpressed SIR2 from the galactose inducible promoter GAL10 in hst3Δ hst4Δ mutants and monitored H3 K56ac. Overexpression of Sir2p did not alter the levels of H3 K56ac in whole cell extracts from hst3Δ hst4Δ mutants (Figure 4A). We then examined the acetylation status of H3 K56 at Tel VIR and HMR by ChIP. Overexpression of SIR2 in hst3Δ hst4Δ mutants did not lead to the deacetylation of H3 K56 at Tel VIR (Figure 4, B and C) and could not restore silencing at yFR057w (Figure 4D), despite the ability of Sir2p to deacetylate H3 K56 efficiently in vitro (Xu et al., 2007). Similarly, overexpression of SIR2 in hst3Δ hst4Δ mutants did not reduce H3 K56ac at the E silencer (Figure 4E) or a1 (Figure 4F) at HMR relative to control strains. In contrast, overexpression of either HST3 or HST4 from the galactose inducible promoter GAL1 in hst3Δ hst4Δ mutants resulted in hypoacetylation of H3 K56 at Tel VIR (data not shown). Together, these results plus the observation that loss of RTT109 had restored telomeric silencing in hst3Δ hst4Δ mutants (Figure 2B) argued that, although Sir2p can deaceatylate H3 K56 (Xu et al., 2007), Hst3p/Hst34p were playing critical roles in deacetylating this residue within silent chromatin.

Hypoacetylation of H3 K56 at HMR Requires Silencing

The above-mentioned experiments established that HST3 HST4 were required for efficient deacetylation of H3 K56 within silent chromatin. We next assessed whether hypoacetylation of K56 on histone H3 required silencing per se or simply the presence of Sir proteins by using SIR2 and sir2-345 yeast. sir2-345p is a catalytically inactive mutant of Sir2p that is stably expressed, can interact with other Sir proteins and is readily recruited to silencers in the presence of other Sir proteins (Imai et al., 2000; Rusché et al., 2002; Kirchmaier and Rine, 2006). However, Sir protein propagation along chromatin is defective in sir2-345 mutants (Rusché et al., 2002; Kirchmaier and Rine, 2006; see also (Hoppe et al., 2002; Luo et al., 2002). Sir spreading, but not silencing, can be restored in sir2-345 cells expressing histone mutants that mimic the hypoacetylated state of substrates of Sir2p (Yang and Kirchmaier, 2006; Yang et al., 2008). Because loss of acetylation of H4 K16 is both necessary and sufficient for Sir protein spreading (Yang et al., 2008), we monitored H3 K56ac levels in SIR2 or sir2-345 yeast expressing either wild-type histones or H4 K16R mutants. ChIP assays using anti-Sir2p, anti-H4 K16ac or anti-H3 K56ac antibodies indicated Sir2p was present at Tel VIR (Figure 5A), and both H4 K16 (Figure 5B) and H3 K56 (Figure 5C) were hypoacetylated in SIR2 cells expressing wild-type histones. In contrast, Sir2p was absent from Tel VIR (Figure 5A) and both H4 K16 (Figure 5B) and H3 K56 (Figure 5C) were hyperacetylated in sir2-345 mutants expressing wild-type histones. In SIR2 and sir2-345 cells expressing H4 K16R mutants, Sir2p and sir2-345p spread throughout Tel VIR (Figure 5A). And, although H3 K56 was hypoacetylated in SIR2 cells, H3 K56 remained hyperacetylated in sir2-345 mutants (Figure 5C), implying deacetylation of this residue was not critical for Sir spreading (see also Figures 2 and 3). Similar results were observed at HMR (Supplemental Figure 2). Thus, HMR and Tel VIR remained transcriptionally active (see Figures 6 and 7 and Supplemental Figures 3 and 4 below) and hyperacetylated at H3 K56 in sir2-345 mutants expressing H4 K16R in which Sir spreading had been restored. This hyperacetylated state of H3 K56 in sir2-345 mutants was consistent with the above-mentioned experiments, which collectively indicated Hst3p/Hst4p-dependent loss of H3 K56ac at HMR and Tel VIR was also silencing dependent.

Figure 5.

SIR protein spreading in sir2-345 H4 K16R mutants does not disrupt H3 K56ac at Tel VIR. (A–C) Sir2p association and H4 K16ac and H3 K56ac levels at Tel VIR in SIR2 or sir2-345 strains expressing wild-type histones or H4 K16R were monitored by ChIP. For each of the indicated genotypes, the efficiency of coprecipitation of DNA from 0.6 and 1.2 kb from the end of Tel VIR or SCC1 with anti-Sir2p (A), H4 K16ac (B), or H3 K56ac (C) antibodies was determined and data were calculated as shown in Figure 2.

Figure 6.

Derepression of yFR057w near Tel VIR in hst3Δ hst4Δ and hst3Δ hst4Δ sir2-345 mutants. (A) Loss of HST3 and HST4 disrupts telomeric silencing. Transcript levels of yFR057w relative to SCR1 were determined for each strain by quantitative real-time PCR and then normalized to sir2-345 mutants expressing wild-type histones, which was set to 100%. Data were calculated as 2^{[(locus C_T − SCR1 C_T)_{sir2-345 H3/H4} − (locus C_T − SCR1 C_T)_a]} × 100, where “a” is the indicated genotype. (B) Sir3p localization to Tel VIR. The efficiency of Sir3p association at Tel VIR in strains with the indicated genotypes was monitored by ChIP using anti-Sir3p antibodies and data were calculated as shown in Figure 2. (C and D) Loss of HST3 and HST4 results in hyperacetylation of H3 K56 at Tel VIR. For each strain, the efficiency of coprecipitation of DNA from 0.6 kb (C) and 1.2 kb (D) from the end of Tel VIR with anti-H3 K56ac antibodies was determined relative to that with anti-H3 antibodies, and then each locus was normalized to HST3 HST4 sir2-345 mutants expressing wild-type histones, which was set to 1. Data were calculated as 2^{[(H3 K56ac C_T − H3 C_T)_{sir2-345 H3/H4} − (H3 K56ac C_T − H3 C_T)_a]}, where “a” is the indicated genotype.

Figure 7.

Loss of H3 K56ac in rtt109 mutants does not disrupt silencing at Tel VIR. (A) Loss of histone H3 K56ac in rtt109 mutants does not disrupt Sir spreading. The efficiency of Sir2p association at Tel VIR in RTT109 and rtt109Δ strains with the indicated genotypes were monitored by ChIP using anti-Sir2p antibodies and data were calculated as shown in Figure 2. (B) Deletion of RTT109 neither disrupts telomeric silencing in SIR2 cells nor rescues telomeric silencing in sir2-345 cells expressing H4 K16R. Transcript levels of yFR057w relative to SCR1 were determined for each strain by quantitative real-time PCR and then normalized to sir2-345 mutants expressing wild-type histones, which were set to 100%. Data were calculated as 2^{[(locus C_T − SCR1 C_T)_{sir2-345 H3/H4} − (locus C_T − SCR1 C_T)_a]} × 100, where “a” is the indicated genotype. (C and D) Deletion of RTT109 disrupts H3 K56ac at Tel VIR. H3 K56ac levels at Tel VIR in each strain were monitored by ChIP. For each strain, the efficiency of coprecipitation of DNA from 0.6 kb (C) and 1.2 kb (D) from the end of Tel VIR with anti-H3 K56ac antibodies was determined relative to that with anti-H3 antibodies, and data were calculated as shown in Figure 6.

H3 K56ac Levels Are Similar in hst3Δ hst4Δ, sir2-345, and hst3Δ hst4Δ sir2-345 Mutants

To assess the relative contributions of Sir2p and Hst3p/Hst4p to the deacetylation of H3 K56, we next combined hst3Δ hst4Δ and sir2-345 mutants and assessed silencing, H3 K56ac and Sir3p association at yFR057w adjacent to Tel VIR and at HMR (Figure 6 and Supplemental Figure 3, respectively). yFR075w was derepressed in both sir2-345 mutants and hst3Δ hst4Δ mutants expressing wild-type histones relative to wild-type cells. However, yFR057w transcript levels were slightly lower in the hst3Δ hst4Δ mutants than in the sir2-345 mutants (Figure 6A). This reduction correlated with the presence of Sir3p adjacent to Tel VIR in hst3Δ hst4Δ mutants versus their absence in sir2-345 mutants (Figure 6B). Because a mild reduction in yFR057w transcript levels was also observed in hst3Δ hst4Δ sir2-345 mutants expressing H4 K16R in which Sir spreading, but not silencing had been restored (Figure 6, B and A, respectively), Sir protein spreading across yFR057w may have interfered with transcription in the absence of the formation of heritable silent chromatin in these mutants rather than residual HST3 HST4-independent telomeric silencing having occurred. Similar decreases in transcript levels occurred in sir2-345 mutants expressing hypoacetylated histone mutants that support Sir protein spreading (e.g., Figure 7B; see also Yang and Kirchmaier, 2006; Yang et al., 2008). Also consistent with this notion, silencing of a telomeric URA3 reporter gene had been disrupted to a similar degree in both hst3Δ hst4Δ and sir2Δ mutants (Figure 1). On loss of silencing, the increase in H3 K56ac adjacent to Tel VIR was similar in sir2-345, hst3Δ hst4Δ, and hst3Δ hst4Δ sir2-345 mutants expressing wild-type histones and in hst3Δ hst4Δ sir2-345 mutants expressing H4 K16R mutants relative to wild-type cells (Figure 6, C and D). Thus, the absence of HST3 HST4 in the context of the catalytically inactive sir2-345p did not lead to further increases H3 K56ac adjacent to Tel VIR. Similar patterns of Sir association and levels of H3 K56ac were observed at HMR, but, in contrast to yFR057w, HMR remained silenced in hst3Δ hst4Δ mutants (Supplemental Figure 3). In contrast to these findings, Xu et al. (2007) reported reduced telomeric H3 K56ac in hst3Δ hst4Δ sir2Δ mutants relative to hst3Δ hst4Δ mutants (Xu et al., 2007). The reason for this difference between the two studies is not clear. Together, our results indicate SIR2 and HST3/HST4 play distinct roles in defining telomeric silent chromatin.

Loss of H3 K56ac Does Not Disrupt Sir Protein Spreading and Silencing

Silencing has been proposed to occur when Sir2p deacetylates H3 K56, thereby facilitating the compaction of silent chromatin (Xu et al., 2007). In this scenario, deacetylation of H3 K56 might reflect a late step in silent chromatin formation that influences the efficiency of Sir spreading or silencing. To test the effect of the absence of H3 K56ac on silent chromatin formation, we compared Sir protein localization and silencing in SIR2 and sir2-345 strains containing or lacking RTT109 and expressing either wild-type histones or H4 K16R mutants (Figure 7 and Supplemental Figure 4). We first tested whether loss of H3 K56ac in rtt109Δ mutants affected Sir protein association with Tel VIR or HMR by ChIP. Similar patterns of Sir2p spreading were observed at both Tel VIR and HMR in SIR2 and sir2-345 cells containing or lacking RTT109 and expressing wild-type histones or H4 K16R (Figure 7A and Supplemental Figure 4A, respectively). Thus, the absence of H3 K56ac neither facilitated nor compromised Sir2p spreading. We next compared silencing in the same yeast strains. Deletion of RTT109 did not alter silencing at yFR057w adjacent to Tel VIR or at HMR in SIR2 cells expressing either wild-type or mutant histones (Figure 7B and Supplemental Figure 4B, respectively; see also Figure 8A). Therefore, the constitutively hypoacetylated state of K56 on H3 was compatible with silencing (see also Miller et al., 2008). Deletion of RTT109 does not alter the levels of yFR057w or a1 mRNA (Figure 7B and Supplemental Figure 4B, respectively) in sir2-345 cells expressing wild-type histones. Thus, the acetylated state of K56 on H3 was not required for transcription from the yFR057w and a1 promoters. Finally, deletion of RTT109 also did not rescue silencing at yFR057w or HMR (Figure 7B and Supplemental Figure 4B, respectively; see also Figure 8) in sir2-345 cells expressing H4 K16R mutants, despite the lack of H3 K56ac at Tel VIR or HMR in these mutants (Figures 7, C and D, and Supplemental Figure 4, C and D, respectively). The reduction in transcript levels seen in sir2-345 mutants expressing H4 K16R relative to wild-type histones is due to Sir protein spreading interfering with transcription rather than the formation of heritable silent chromatin in a subset of the population. Transcript levels of a1 in sir4Δ mutants expressing wild-type histones or H4 K16R is similar (Yang et al., 2008), and, as described below, mating was not restored in sir2-345 mutants expressing H4 K16R (see also Yang and Kirchmaier, 2006; Yang et al., 2008).

Figure 8.

Loss of H3 K56ac does not rescue HM silencing in sir2-345 mutants. (A) Patch mating analyses of HM silencing in SIR2 or sir2-345 rtt109Δ cells expressing the indicated histone mutants. (B and C) Patch mating analyses of HM silencing in SIR2 or sir2-345 RTT109 cells expressing the indicated histone mutants. Patch mating assays were conducted as shown in Figure 3.

To examine further the impact of H3 K56 on silencing at the HM loci, we tested whether loss of H3 K56ac by deletion of RTT109 or mutation of H3 K56 to R could restore mating in sir2-345 yeast expressing histone mutants that supported Sir protein spreading (Figure 8). Deletion of RTT109 did not restore HM silencing in sir2-345 mutants expressing either wild-type or mutant histones (Figure 8A). In contrast, MATα SIR2 cells lacking RTT109 and expressing either wild-type histones or various histone mutant combinations mated efficiently (Figure 8A), indicating silencing was intact at HMR. However, unlike MATα cells, MATa SIR2 rtt109Δ cells expressing H3 K9,14R H4 K16R were defective in mating (Figure 8A; compare also with RTT109 cells expressing H3 K9,14R H4 K16R in Figure 8C). Similarly, silencing at HML was lost in MATa SIR2 cells expressing H3 K9,14,56R H4 K16R (Figure 8C). Together, these results implied that the absence of H3 K56ac is largely compatible with silencing in SIR2 cells, except at HML under conditions in which the N-terminal tail of histone H3 is also hypoacetylated (e.g., H3 K56R H4 K16R vs. H3 K9,14,56R H4 K16R in Figure 8, B and C). Silencing at the native HML locus is more sensitive to disruption than is silencing at HMR under a variety of conditions (e.g., Ehrenhofer-Murray et al., 1997; Huang et al., 1997; Xu et al., 1999; Yang et al., 2008). Relocalization of Sir proteins to other loci throughout the genome due to the global hypoacetylated state of histone H3 and H4 may contribute to the observed silencing defects at HML. Loss of H4 K16ac causes Sir spreading beyond normally silenced loci (Kimura et al., 2002; Suka et al., 2002) and loss of acetylation at H3 K56 permits Sir recruitment to loci not normally bound by Sir proteins (Miller et al., 2008).

The absence of silencing in sir2-345 mutants in which Sir spreading has been restored (Yang and Kirchmaier, 2006; Figures 7 and 8 and Supplemental Figure 4) implies additional unknown substrates of Sir2p are normally deacetylated during silent chromatin formation. As defects in biological pathways can often be suppressed by overexpression of genes in overlapping or related pathways, we examined whether overexpression of HST3 could restore silencing in sir2-345 cells expressing wild-type or histone hypoacetylation mutants. Unlike overexpression of SIR2 (Yang and Kirchmaier, 2006), overexpression of HST3 (Figure 9A) did not restore silencing in sir2-345 mutants at HMR or HML (Figure 9B) or at yFR057w at Tel VIR (Figure 9C), even in the context of histone hypoacetylation mutants that support Sir spreading. Together, these results imply HST3/HST4 and SIR2 primarily regulate the acetylation status of different histone residues within silent chromatin and they generally cannot substitute for one another.

Figure 9.

Overexpression of HST3 does not rescue silencing in sir2-345 mutants. (A) HST3 overexpression. Transcript levels of HST3 relative to SCR1 in yeast containing a vector (pRS426) or HST3 expressed from its endogenous promoter on a 2μ plasmid (pAK998) were determined by quantitative real-time PCR and then expressed relative to SIR2 yeast containing a vector, which was set to 1. Data were calculated as 2^{[(HST3 C_T − SCR1 C_T)_vector − (HST3 C_T − SCR1 C_T)_{HST3 overexpression}]}. (B) HM silencing. Patch mating analyses of SIR2 or sir2-345 cells containing a vector or an HST3 overexpression plasmid were conducted as shown in Figure 3. (C) Telomeric silencing. Transcription from yFR057w at Tel VIR in MATα SIR2 or sir2-345 cells expressing the indicated histones and containing a vector or an HST3 overexpression plasmid were determined as in Figure 2. Data were calculated as 2^{[(locus C_T − SCR1 C_T)_vector − (locus C_T − SCR1 C_T)_a]} × 100, where “a” is the indicated genotype.

DISCUSSION

We previously assessed the impact of mutating H3 K56 on silencing in S. cerevisiae and found that telomeric silencing is disrupted in H3 K56Q mutants, but not in H3 K56R mutants (Miller et al., 2008; see also Hyland et al., 2005; Xu et al., 2007). Here, we show that Hst3p and Hst4p are required for efficient deacetylation of H3 K56 within silent loci (Figures 2, 3, 6, and Supplemental Figure 3). In the absence of Hst3p and Hst4p, hyperacetylation of H3 K56 at Tel VIR disrupted silencing without resulting in the dissociation of Sir proteins or an increase in H4 K16ac (Figure 2). This telomeric silencing defect in hst3Δ hst4Δ mutants could be suppressed and hypoacetylation of H3 K56 at Tel VIR could be restored by ectopic expression of HST3 or by deletion of RTT109 (Figures 1 and 2) but not by overexpression of SIR2 (Figure 4; see also Figure 1). These findings revealed the mechanism by which HST3 and HST4 contribute to silencing in budding yeast is through regulating the acetylation status of H3 K56.

Like in S. cerevisiae, the single HST3 HST4 orthologue in S. pombe, hst4⁺, is required for deacetylation of H3 K56 in vivo (Haldar and Kamakaka, 2008). Fission yeast lacking hst4⁺ also have telomeric silencing defects, are sensitive to multiple DNA-damaging agents, and are defective in chromosome maintenance (Freeman-Cook et al., 1999; Durand-Dubief et al., 2007; Haldar and Kamakaka, 2008). And, overexpression of S. pombe hst4⁺ suppresses telomeric silencing defects in S. cerevisiae hst3Δ hst4Δ mutants (Freeman-Cook et al., 1999), implying the role of these orthologues in silent chromatin is similar. Hst4⁺ has been localized to silent chromatin in fission yeast (Durand-Dubief et al., 2007). However, we have not observed preferential enrichment of Hst3p at silenced loci in budding yeast by ChIP (data not shown; see also Grunweller and Ehrenhofer-Murray, 2002) despite Hst3p/Hst4p-dependent hypoacetylation of H3 K56 at HMR and Tel VIR being silent chromatin dependent (Figures 5–7 and Supplemental Figures 2–4). It is possible that the epitope on Hst3p for our ChIP analysis was inaccessible in silent chromatin or that Hst3/Hst4p is present at silent loci only transiently during the cell cycle. In contrast to Sir proteins, which are relatively stable, Hst3p is rapidly turned over by cells (Dasgupta et al., 2004; Belle et al., 2006; Thaminy et al., 2007). This instability may result in short-lived interactions with other factors.

H3 K56ac and the Formation and Stability of Silent Chromatin

Both the stability of telomeric silent chromatin and the levels of H3 K56ac vary during the cell cycle, raising the possibility that these events are related. Sir proteins partially delocalize from telomeric foci during G2/M (Laroche et al., 2000), and the transcriptional activator Ppr1p can activate a telomeric URA3 reporter within silent chromatin during G2/M, but not during G0, G1, or early S phase (Aparicio and Gottschling, 1994). The presence of H3 K56ac at telomeres may contribute to this reduced stability of telomeric silent chromatin during G2/M. Consistent with this notion, mutations in H3 K56 increase the sensitivity of telomeric DNA to modification by dam methylase and of cleavage of DNA by enzymes (Masumoto et al., 2005; Xu et al., 2007). Mutation of H3 K56 also can alter the superhelicity of plasmids (Masumoto et al., 2005) and result in the derepression of silent chromatin (Hyland et al., 2005; Xu et al., 2007; Miller et al., 2008). Because replication-coupled assembly of newly synthesized H3 K56ac would occur near the end of S phase in late replicating regions such as the telomeres (Raghuraman et al., 2001), these regions likely still contain high levels of H3 K56ac upon entry into G2/M. Our ChIP analyses hint that minor defects in telomeric localization of Sir2p may occur in hyperacetylated hst3Δ hst4Δ mutants (e.g., Figure 2). Future studies using microscopy should clarify whether partial Sir delocalization and derepression of silenced genes during G2/M are a function of temporary structural changes at silenced regions caused by incorporation of H3 K56ac.

It will also be of interest to determine whether HST3 HST4-dependent deacetylation of H3 K56ac reflects a cell cycle requirement for establishing silencing under certain conditions. Passage through mitosis facilitates the assembly of silent chromatin at both telomeres and HM loci, but the mechanism involved is not known (Lau et al., 2002; Martins-Taylor et al., 2004; Katan-Khaykovich and Struhl, 2005; Matecic et al., 2006). Ultimately, however, the establishment of silencing will likely turn out to be more complex as H3 K56ac does not disrupt silencing at the HM loci except under sensitized conditions (Figures 3 and 8), and passage through S phase is sufficient to establish silencing at HMR, despite low levels of HST3 and HST4 during this period of the cell cycle (Miller and Nasmyth, 1984; Fox et al., 1997; Kirchmaier and Rine, 2001; Li et al., 2001; Lau et al., 2002).

Deacetylation by Sir2p, Hst3p/Hst4p, and Silencing

Sir2p has multiple critical functions in silencing. One key function is to deacetylate histones to permit Sir2-4p to bind to nucleosomes and spread along the chromosome. Sir proteins do not readily spread away from silencers along chromatin in cells expressing catalytically inactive sir2-345p unless histones residues normally deacetylated by Sir2p maintain a positive charge (Yang and Kirchmaier, 2006). However, Sir spreading alone is not sufficient for silencing (Figure 7 and Supplemental Figure 4), (Kirchmaier and Rine, 2006; Yang and Kirchmaier, 2006), implying Sir2p mediates one or more additional critical events during silent chromatin formation. Such events may include the deacetylation of novel key residues on histones to promote structural changes in chromatin needed for inactivating transcription or to generate 2′-O-acetyl-ADP ribose, a product of the deacetylation reaction that alters the conformation and stoichiometry of Sir2-4p complexes and Sir–nucleosome interactions in vitro (Liou et al., 2005; Onishi et al., 2007). Our findings indicate the production of 2′-O-acetyl-ADP ribose specifically through the deacetylation of H3 K56 by Sir2p or Hst3p/Hst4p was not required for silencing at the HM loci. Rather, deacetylation of a different unknown substrate by Sir2p may be needed for silencing. Consistent with this, hypoacetylation of H3 K56 by mutation or deletion of RTT109 did not restore silencing in sir2-345 mutants or disrupt silencing at HMR in SIR2 cells (Figures 7 and 8 and Supplemental Figure 4). The changes required for silencing are SIR2-dependent as overexpression of SIR2 (Yang and Kirchmaier, 2006) but not HST3 (Figure 9), in sir2-345 mutants restores silencing at both HMR and HML.

H3 K56ac is assembled into chromatin through both replication- and transcription-coupled pathways (Ozdemir et al., 2006; Schneider et al., 2006; Rufiange et al., 2007; Miller et al., 2008, and references within). Our data are consistent with a model in which hypoacetylation of H3 K56 at silenced loci is maintained by at least two methods. In the first method, Sir2p-mediated silencing prevents transcription, thereby ensuring H3 K56ac is not loaded onto DNA at silenced loci via transcription-coupled chromatin assembly (Figure 5) (Xu et al., 2007). In the second method, when H3 K56ac is loaded onto DNA at silenced loci each cell cycle via replication-coupled chromatin assembly, deacetylation of this residue is primarily regulated by the NAD⁺-dependent deacetylases Hst3p/Hst4p, although Sir2p may also assist with maintaining the hypoacetylated state of H3 K56 throughout the cell cycle (Xu et al., 2007). Future studies should clarify how each of these histone deactylases control H3 K56ac in silent chromatin.

Abbreviations used:

ChIP

chromatin immunoprecipitation.