Sir3 C-Terminal Domain Involvement in the Initiation and Spreading of Heterochromatin

Abstract

Heterochromatin is nucleated at a specific site and subsequently spreads into distal sequences through multiple interactions between modified histones and nonhistone proteins. In the yeast Saccharomyces cerevisiae, these nonhistone proteins include Sir2, Sir3, and Sir4. We have previously shown that loss of the C-terminal Rap1 domain containing Sir3 and Sir4 association sites can be overcome by tethering a 144-amino-acid C-terminal domain (CTD) of Sir3 adjacent to the telomere. Here, we explore the substructure and functions of the CTD. We demonstrate that the CTD is the minimum domain for Sir3 homodimerization, a function that is conserved in related yeasts. However, CTD heterodimers associate at only low efficiencies and correspondingly have low levels of tethered silencing, consistent with an essential role for dimerization in tethered silencing. Six missense alleles were generated, with ctd-Y964A producing the most extreme phenotypes when tethered to the LexA binding sites. Although ctd-Y964A is capable of dimerization, telomere silencing is abrogated, indicating that the CTD serves a second essential function in silencing. Chromatin immunoprecipitation analyses of wild-type and ctd-Y964A mutant cells indicate an association of the CTD with the deacetylated histone tails of H3 and H4 that is necessary for the recruitment of Sir3. The efficiency of spreading depends upon the apparent stoichiometry and stability during the initiation event. The predicted Cdc6 domain III winged-helix structure may well be responsible for dimerization.

A conserved property of telomeres is the formation of heterochromatin in telomere-adjacent regions. In Saccharomyces cerevisiae, telomeric heterochromatin confers both regional silencing, also termed telomere position effect (TPE), and late replication (13, 15, 54). Despite the unknown physiological function of these specialized regions, TPE serves as a sensitive indicator of the chromatin state. TPE can be measured quantitatively through the expression of a Pol II- or Pol III-transcribed gene positioned in subtelomeric regions (15, 19). Under these conditions, heterochromatin spreads continuously and unidirectionally from the telomere. In the natural telomere, TPE involves additional levels of regulation, including the presence of barriers and insulators typical of higher eukaryotes (14, 46).

Heterochromatin normally initiates at specific “silencer” sequences and subsequently spreads to nearby sequences in a process that involves intimate associations between structural proteins and specifically modified histones (18, 33, 37, 49). Recent studies support an ordered recruitment of factors to initiate and propagate telomeric silencing unidirectionally in the yeast Saccharomyces cerevisiae (18, 39, 48, 49). The current working model posits that telomeric silencing begins with recruitment of a Sir2/Sir4 structural complex by the telomere binding protein Rap1 (or the end-binding heterodimer yKu) through interactions between Rap1 and Sir4 in a process that is independent of both Sir2 deacetylase activity and Sir3 function (18, 32). Telomeres in yeast are nucleosome free. The structural characteristics of TPE initiation occurring at the telomeric/subtelomeric border are known to involve a loop-back structure that positions the telomere in close proximity to subtelomeric nucleosomal DNA (10). Recent data have suggested that Sir2/Sir4 complexes can bind to chromatin and partially deacetylate histone H3 at K9 and K14 and histone H4 at K16 (18, 48). A modified nucleosome initially placed adjacent to telosomal sequences can subsequently associate with the telomeric junction via associations with Sir3 and/or Sir4, both of which can associate only with the deacetylated forms of histones H3 and H4 (5, 16, 17, 28, 38). This gives rise to a stepwise unidirectional action of Sir2 deacetylation along the chromatin fiber that is stabilized by Sir3 and Sir4 interactions (17, 48). The spreading is greatly increased by overproduction of Sir3 in a process that is still poorly understood.

Two-hybrid, plasmon resonance, and in vitro studies have demonstrated interactions between Sir4 and Sir3, between Sir2 and Sir4, and of both Sir3 and Sir4 with deacetylated histones H3 and H4 (see reference 25). No interaction between Sir2 and Sir3 has been identified in vivo or in vitro in any context. Indeed, precipitation of the Sir complex by a Sir2 antibody is independent of the presence or absence of Sir3 (51).

While Sir2 and Sir4 appear to perform a significant function in TPE initiation through initial deacetylation, we have investigated the role of Sir3 in initiation and spreading (29, 41, 43). To achieve a better understanding of the capability of Sir3 to initiate TPE, we have used a LexA-tethered-silencing system (see Fig. 1A) previously developed in our lab as a model system for dissecting components of silencing (35, 43). This assay is performed with strains carrying the rap1-17 allele lacking the C-terminal 165-amino-acid region of Rap1 that is normally responsible for the recruitment of Sir3 and Sir4/Sir2 complexes to the telomere and hence for generating TPE. Tethering of LexA-Sir3 to multiple LexA binding sites at the telomeric/subtelomeric junction confers an in cis restoration of TPE, a process that we refer to as “tethered silencing” (43). This assay was established to bypass the Rap1-mediated initiation step and thereby test the requirements and chromatin structure of a putative Sir3-mediated initiation process.

FIG. 1.

(A) Tethered system as a measure of minimal requirements for silencing. The system measures the ability of LexA fusion proteins (small circles, LexA; large circles, fused protein) tethered at the telomere at LexA binding sites to restore silencing in a strain carrying the rap1-17 allele that is incapable of associating with Sir3 and Sir4. In a strain containing URA3 adjacent to the telomere (CLY3), silencing frequency is measured by determining the fractions of cells that grow on 5-FOA. 5-FOA allows the growth of Ura3⁻ but not Ura⁺ cells. (B) 5-FOA resistance (FOA^r) is elicited by the LexA-CTD. LexA represents the LexA protein alone (FOA^r frequency, <1.77 × 10⁻⁶; n = 14). LexA-Sir3 (FOA^r frequency, 0.007; n = 14), LexA-CTD (FOA^r frequency, 0.057; n = 35), LexA-Sir3(1-835) (FOA^r frequency, <3.5 × 10⁻⁶; n = 19), and LexA-Sir3N205(1-835) (FOA^r frequency, <2× 10⁻⁶; n = 18) represent LexA fusions with full-length Sir3, the CTD, Sir3(1-835), and Sir3N205(1-835). All strains were constructed in a CLY3 rap1-17 background. FOA^r values are the medians of the observed range of values. n, number of samples.

Recent studies have helped to define the interactions between regions of Sir3 and other factors (25). We previously demonstrated that a region encompassing the C-terminal 144 amino acids, termed the CTD (C-terminal domain of Sir3), is sufficient and necessary for tethered silencing in a Sir2-, Sir3-, and Sir4-dependent process (43). The interacting partners of CTD in this region in vivo remain uncertain, although both Sir3 and the histone H3-H4 interacting domains overlap with the CTD. However, in vivo functions have not been ascribed to these potential interactions.

In this study, we found that Sir3 amino acids 843 to 978 [Sir3(843-978)] comprise the minimal domain required for the restoration of tethered silencing. The CTD also undergoes dimerization, a process that is functionally conserved in related yeasts. The sequence of CTD resembles domain III of Cdc6, a winged-helix structure that is also predicted to be present in Orc1, a protein that confers low levels of tethered silencing. Further study revealed a series of severe ctd mutations that act independently from dimerization. Our results suggest that the CTD has an essential second function in the deacetylation of the N termini of histones H3 and H4. The spreading of heterochromatin appears to be related to the stability of the initiating complex and appears to be most dependent on Sir3. These data suggest that this putative domain may play a role in the function of the CTD.

MATERIALS AND METHODS

Yeast strains.

Growth conditions were as described previously (3). All Saccharomyces cerevisiae strains except for those used in two-hybrid analysis are isogenic to W303a. The strain used for the tethered-silencing assay was CLY3/rap1-17 (MATα rap1::LEU2 leu2-3,112 ade2-1 his3 trp1 ura3-1 VIIL::URA3LexAS3 pRS313/rap1-17) (43). The sir3 null strain for the silencing assay was ΔGK (MATα rap1::leu2 leu2-3,112 ade2-1 his3 trp1 ura3-1 sir3::ADE2 VIIL::URA3 pRS313/RAP1).

Two-hybrid interactions were measured in CTY10-5D (43). BL22-2b sir3-835VV, BL22-2b sir3-835GV, and BL22-2b sir3-835WV alleles were derived from BL22-2b (MATα leu2-3,112 ade2-1 his3 trp1 ura3-1 VIIL::URA3/ADE2) by two-step gene replacement. pRS304-sir3-835VV, pRS304-sir3-835GV, and pRS304-sir3-835WV were cut with AflII and transformed into BL22-2b (27). Gene replacements were selected on 5′-fluoroanthranilic acid (55) that allows the growth of Trp⁻ but not Trp⁺ cells. The presence of these sir3-835VV, sir3-835GV, and sir3-835WV alleles was confirmed by DNA sequencing.

Saccharomyces bayanus strains Y244 (MATα ura3 G418^r) and Y245 (MATa ura3 G418^r) were kindly provided by Jure Piskur (36). We selected cells for Y245 trp⁻ (MATa ura3 trp G418^r) strains by plating 10⁹ cells on 5′-fluoroanthranilic acid plates. Y245 Δsir3 (MATa ura3 trp sir3::URA3 G418^r) strains were generated by gene replacement using sir3::URA3 PCR products. Two primers, TCTAGATCTGTTTAGCAATTGGCAAGCATAGTTGGCTAGAGATTCGGTAATCTCCGAACA and CGTATATATACTAGCGCATGTTGTGGGGGCGGGATAAATTAAACACCGCAGGGTAATAACTG, were used to amplify the URA3 gene in pRS306. The resulting 1.1-kb PCR product was then transformed into Y245 trp⁻ strains to disrupt the S. bayanus SIR3 gene (Sbsir3). The mating tester strains for these studies were EMPY75 (MATα thr3 met⁻) and EMPY76 (MATa lys1) (26).

The sir4::KAN^r null alleles were generated by targeted PCR. The sir4::KAN^r fragment was amplified by PCR, using pFA6a-kanMX6 (31) as a template and two primers, 5′ GCTTCAACCCACAATACCAAAAAAGCGAAGAAAACAGCCACGGATCCCCGGGTTAATTAA 3′ and 5′ GGTACACTTCGTTACTGGTCTTTTGTAGAATGATAAAAAGGAATTCGAGCTCGTTTAAAC 3′, corresponding to the upstream and downstream sequences, respectively, of SIR4. The PCR product was transformed into CTY10-5d strains. The SIR4 disruptions were verified by both PCR and Southern analysis. Additionally, each of the sir4 strains failed to mate with the tester strains, EMPY75 and EMPY76.

Plasmids.

DNA sequencing and Western blotting confirmed all of the following constructs. pBTM-SIR3(1-835) and pBTM-SIR3N205(1-835) were reconstructed by site-directed mutagenesis to create a stop codon at amino acid 835. The first construct, pBTM-SIR3(1-835) (43), has now been confirmed by DNA sequencing and Western blotting to have a valine insertion mutation at position 835, so we have renamed the construct pBTM-sir3-835VV. pRS304-sir3-835VV was generated by SalI digestion of pBTM-sir3-835VV, followed by cloning in the SalI site of pRS304. pRS304-sir3-835GV and pRS304-sir3-835WV were constructed by cloning the 0.4-kb NruI/NdeI fragment of pBTM-sir3-835GV(525-978) and pBTM-sir3-835WV(525-978) containing the glycine and tryptophan insertion mutations into pRS304-sir3-835VV. Both pBTM-sir3-835GV(525-978) and pBTM-sir3-835WV(525-978) were created by site-directed mutagenesis. pBTM-SIR3(835-978) was constructed as described previously (43).

pBTM-SIR3(835-978)/KanMX6 was generated by replacing a TRP1 gene with KanMX6. KanMX6 fragments were generated by PCR to create PvuII and XbaI in the flanking sequence. After digestion with PvuII/XbaI, the KanMX6 fragment was cloned into the TRP1 site of pBTM-SIR3(835-978). All of the CTD missense mutations were generated by site-directed mutagenesis (Stratagene). Primer sequences are available on request. The internal minideletions within the CTD were generated by site-directed mutagenesis that creates specific flanking restriction sites. The mutated pBTM-CTD construct was digested with the relevant restriction enzyme, and the product was isolated and ligated to form a 15- to 20-amino-acid in-frame deletion. The Δ842-859 and Δ957-977 deletion constructs were created by generating two SacI sites at positions 841 and 859 or 956 and 977, respectively. The Δ858-875, Δ877-897, and Δ901-922 constructs were constructed by generating two KpnI sites at positions 857 and 875, 876 and 897, or 900 and 922, respectively. The Δ924-945 and Δ946-960 constructs were created by generating one XhoI site at position 923 or 960, respectively, with a preexisting endogenous XhoI site at position 945. The two N-terminal deletions, Δ835-842 and Δ835-860, were generated by creating SacI sites at positions 841 and 859, respectively. After digestion with SmaI and SacI, the pBTM-CTD fragments were filled in with T4 DNA polymerase and ligated.

pBTM-Saccharomyces bayanus CTD was constructed by PCR amplification of the C-terminal 700 bp of SbSIR3 with the primers CTGTATCATTCCAGGGGATCCTACCACGTT and GCCGACGCGAAGGGATCCGGAAAC, followed by BamHI digestion. The resulting fragments were then cloned into a BamHI site of pBTM116 and their orientation was determined by DNA sequencing.

pBTM-ORC1(738-914) was constructed by PCR amplification of the C-terminal 620 nucleotides of ORC1 using the primers GGAAAGACGGTTATTGAATTCGAAAATGAGGAGC, containing an EcoRI site, and GATAAATGCGCTACTCTGCAGGTATATGTATGTG, containing a PstI site, after digestion with both enzymes. The resulting fragments were cloned into the EcoRI/PstI site of pBTM116 after cleavage with both enzymes.

pGAD-CTD was generated after cleaving pBTM-SIR3 with Asp718 and SalI and after filling in with the Klenow fragment. The resulting C-terminal 1.5-kb fragment of SIR3 was cloned into blunt-ended pGAD10 after its XhoI ends were filled in with the Klenow fragment.

Genomic library screening of SbSIR3.

The Saccharomyces bayanus strain was purchased from ATCC (ATCC 8741). The Saccharomyces castellii strain and both the S. bayanus and the S. castellii genomic libraries were kindly provided by Vicki Lundblad. The genomic libraries consist of a Sau3AI partial digest of the total S. bayanus or S. castellii genomic DNA cloned into the BamHI site of YEp112. The S. bayanus DNA fragment, consisting of the C-terminal 839 nucleotides of SbSIR3, was PCR amplified using the primers GTCCTCTAATCAACCGTTTGG and CTATCTATCATTGGCGTTGGC and S. bayanus genomic DNA as the template to generate fragments permitting genomic library screening.

The S. castellii DNA fragment, corresponding to the C-terminal 505 nucleotides of S. castelli SIR3 (ScasSIR3), was generated by PCR amplification using two primers, GATGACCAAGTGGAAGAGCAAGAC and GGACTCAAAACACCAAAGTTACGC, and S. castelli genomic DNA as the template. The PCR fragment was subsequently used as a probe for genomic library screening. Of the more than 6,000 colonies that were screened, three clones with full-length SbSIR3 and two clones with full-length ScasSIR3 were isolated.

Telomeric-silencing assay conditions.

5-Fluoroorotic acid (5-FOA) assays to measure the silencing of a URA3-marked VIIL telomere and ADE2 color assays for telomeric silencing of a URA3/ADE2-marked telomere were previously described (34). The median values were determined from the results for at least seven independent colonies. The sample sizes and the ranges of values are indicated below (see Table 1). Statistical significance was determined using the Mann-Whitney rank sum test.

TABLE 1.

Tethered-silencing efficiency requires CTD dimerization^a

Fusion protein	FOAr frequencyb	β-Galactosidase activity (U/mg)c
LexA-CTD	5.70 × 10−2 (1.9-10.7; 35)	115.58 ± 35.98 (11)
LexA-CTD(Δ835-842)	1.70 × 10−2 (0.5-4.4; 21)	148.9 ± 60.4 (2)
LexA-CTD(Δ835-860)	<9.14 × 10−7 (14)	3.4 ± 0.0 (2)
LexA-CTD(Δ842-859)	<2.03 × 10−6 (12)	3.0 ± 0.0 (2)
LexA-CTD(Δ858-875)	<1.68 × 10−6 (12)	3.1 ± 0.0 (2)
LexA-CTD(Δ877-897)	<2.45 × 10−6 (12)	3.7 ± 0.2 (2)
LexA-CTD(Δ901-922)	<2.02 × 10−6 (12)	3.5 ± 0.2 (2)
LexA-CTD(Δ924-945)	<1.67 × 10−6 (12)	3.4 ± 0.4 (2)
LexA-CTD(Δ946-960)	<1.77 × 10−6 (12)	3.8 ± 0.8 (2)
LexA-CTD(Δ957-977)	<1.63 × 10−6 (12)	3.6 ± 0.5 (2)
LexA	<1.77 × 10−6 (14)	4.2 ± 0.5 (2)

^aThe strain used was CLY3/rap1-17.

^bFOA^r frequencies are the medians of observed values. The ranges of the observed values and/or the numbers of samples are shown in parentheses.

^cThe values in Miller units of the two-hybrid assay were derived from a β-galactosidase assay of CTY10-5d cells containing the indicated LexA fusion proteins and GAD-CTD. The β-galactosidase activities shown are mean± standard deviations. The numbers of samples are given in parentheses. Higher numbers indicate stronger interactions between two fusions.

Two-hybrid liquid assay conditions.

Two-hybrid assays were performed as described previously (7). pGAD fusions and LexA fusions were transformed into CTY10-5d cells. The mean values of β-galactosidase activity in Miller units were determined from the results for at least two independent transformants. The data are represented as the means ± standard deviations (SD), and the sample sizes are indicated (see Table 2).

TABLE 2.

CTD dimerization is conserved^a

Protein(s)	β-Galactosidase activity of GAD fused in frame with:b
	S. cerevisiae CTD	S. bayanus CTD
LexA	4.86 ± 0.87 (2)	5.85 ± 0.56 (2)
LexA-S. cerevisiae CTD	115.58 ± 35.98 (11)	13.03 ± 1.25 (2)
LexA-S. bayanus CTD	19.09 ± 0.23 (2)	299.18 ± 2.70 (2)

^aAll values are derived from β-galactosidase assays of CTY10-5D cell extracts containing both LexA and GAD fusion proteins.

^bValues represent the specific activities of β-galactosidase (U/mg) as means±SD. The numbers of samples are given in parentheses. Mean values below 6.0 indicate the absence of interaction and represent background values.

Western blotting.

Ten milliliters of cells was grown to an optical density at 600 nm (OD₆₀₀) of 1.0. The cells were then washed and resuspended in 100 μl of phosphate-buffered saline (PBS) with protease inhibitors (18 μg/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1.5 μg/ml leupeptin, and 5 μg/ml pepstatin A). Lysis was performed using glass beads for five cycles of a 1-min pulse and 1 min on ice. After the addition of 4× Laemmli sample buffer (0.1 M Tris-HCl [pH 6.8], 3% sodium dodecyl sulfate [SDS], 20% glycerol, 4% mercaptoethanol, 0.01% bromophenol blue), samples were boiled for 5 min, separated on a 10% SDS-polyacrylamide gel electrophoresis (PAGE) gel and transferred to a polyvinylidene difluoride (PVDF) membrane. Protein blots were probed using anti-LexA antibodies at a 1:2,000 dilution (Upstate Biotechnology) and anti-Sir3 antibodies at a 1:6,000 dilution (kindly provided by Lorraine Pillus). After incubation with a horseradish peroxidase-conjugated secondary antibody, the ECL Western blotting analysis system (Amersham Biosciences) was used for chemiluminescent detection.

ChIP assays.

The chromatin immunoprecipitation (ChIP) assay was performed as described previously, with some modifications (42). Briefly, 400 ml of cells was grown to an OD₆₀₀ of 0.5 to 1.0 and cross-linked with 1% formaldehyde for 1 h at room temperature. The cells were washed and resuspended in 20 ml of spheroplast buffer (18.2% sorbitol, 1% glucose, 0.2% yeast nitrogen base, 0.2% Casamino Acids, 25 mM HEPES [pH 7.4], 50 mM Tris, 1 mM dithiothreitol), and zymolyase (800 units) was added to generate spheroplasts. After being washed with 5 ml of ice-cold PBS buffer (10 mM KH₂PO₄, 40 mM K₂HPO₄, 150 mM NaCl, and 0.5 mM PMSF), HEPES-Triton X-100 buffer (0.25% Triton X-100, 10 mM EDTA, 0.5 mM EGTA, 10 mM HEPES [pH 6.5], 0.5 mM PMSF, 1 μg/ml pepstatin, and 1 μg/ml leupeptin), and HEPES-NaCl buffer (200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 10 mM HEPES [pH 6.5], 0.5 mM PMSF, 1 μg/ml pepstatin, and 1 μg/ml leupeptin), the spheroplasts were resuspended in 1 ml of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris [pH 8.1], 0.5 mM PMSF, 1 μg/ml pepstatin, and 1 μg/ml leupeptin) and sonicated to generate an average DNA fragment size of 0.5 to 1 kb. After centrifugation, approximately 1.1 ml of supernatant was added to 10 ml of IP dilution buffer (1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris [pH 8.1], 167 mM NaCl, 0.5 mM PMSF, 1 μg/ml pepstatin, 1 μg/ml leupeptin) to form the chromatin fraction. Immunoprecipitation was performed with 7 μl of anti-LexA antibodies (Upstate Biotechnology), 3 μl of anti-Sir2 antibodies, 5 μl of anti-Sir4 antibodies (kindly provided by Susan Gasser), 2 μl of anti-Sir3 antibodies (kindly provided by Rohinton Kamakaka), or 6 μl of anti-AcK16 of H4 (Upstate Biotechnology) per 700 μl of chromatin solution (∼20 OD equivalents of cells). The primers for LexA binding sites were CCAGTGGTTATATGTACAGGATCC and GGTTTAGATGACAAGGGAGACGC, and the primers for ACT1 were CCAATTGCTCGAGAGATTTC and CATGATACCTTGGTGTCTTG. PCRs were carried out in 50-μl reaction mixtures with 1/46 of the immunoprecipitates and 1/4,000 of input DNA, 0.1 mM deoxynucleoside triphosphates, and 0.1 mCi of [α-³²P]dCTP. PCR cycling was conducted at 95°C for 2 min, followed by 25 cycles of 95°C for 30 s, 56°C for 30 s, 72°C for 30 s, and finally 72°C for 5 min. Samples were subjected to electrophoresis on 6% polyacrylamide gels, and the fragments were quantified using a Fuji Bio-phosphorimager.

The ChIP data were quantified by the formula (IP_LexA/IP_ACT1)/(Input_LexA/Input_ACT1), where IP_LexA is the immunoprecipitated LexA product and Input_LexA is the input LexA strain. These values were normalized to the values from the LexA strain. The ChIP data were generated from at least four independent experiments. The statistical significance was determined by Student's t test.

In silico analysis of Sir3 and the crystallized Pyrabaculum aerophilum Cdc6 domain.

The Cdc6 conserved domain was detected by the protein BLAST search (http://www.ncbi.nlm.nih.gov/BLAST/). The Sir3(525-921) sequence was aligned with several proteins containing the Cdc6 domain, including Pyrobaculum aerophilum Cdc6 (Protein Data Bank number 1FNN), which has been X-ray crystallized. The CTD was aligned with domain III of P. aerophilum Cdc6 using the Cn3D program (http://www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml). The three-dimensional image of the P. aerophilum Cdc6 was regenerated by the UCSF chimera program (45).

RESULTS

Sir3(835-978) is essential for TPE.

We have previously developed a system to study the domains of the 978-amino-acid Sir3 protein that are required for the initiation of tethered silencing in rap1-17 cells (Fig. 1A) (35, 43). The rap1-17 allele encodes a Rap1 molecule lacking the C-terminal 165 amino acids that are essential for the recruitment of Sir3 and Sir4 (26). As a consequence, the rap1-17 allele completely disrupts TPE. By targeting LexA-Sir3 fusions at LexA binding sites positioned at the subtelomeric/telomeric junction, silencing can be restored. We term this form of silencing “tethered silencing” because it bypasses the Rap1-mediated initiation step, thereby serving as a model system for testing the ability of fusion proteins to initiate TPE. Previous studies of a dominant SIR3 allele, SIR3N205, that was thought to lack the terminal 144 amino acids suggested that the C terminus was dispensable for tethered silencing (22, 29, 43). Further studies of this allele revealed a nucleotide insertion that produced an in-frame full-length protein (data not shown). We therefore reexamined the necessity for the CTD, which encompasses amino acids 835 to 978, in the silencing of the LexA-Sir3(1-835) and LexA-Sir3N205(1-835) fusion proteins. In both cases, the CTD was absolutely required for tethered silencing (Fig. 1B).

Sir3(843-978) is the minimum domain sufficient for the restoration of tethered silencing.

In order to determine the region of the CTD that is responsible for the initiation of tethered silencing, a series of LexA fusion proteins containing small in-frame deletions that span the CTD were generated (Table 1). These deletion mutations were confirmed by both DNA sequencing and Western blotting (Fig. 2A; data not shown). With the exception of the deletion of amino acids 835 to 842, all deletions abrogated tethered TPE (Table 1), indicating that amino acids 843 to 978 of the CTD comprise its minimal functional domain.

FIG. 2.

Western blot analyses of LexA-CTD fusion proteins show comparable abundances under most conditions. The CLY3 rap1-17 strain was transformed with different LexA fusion proteins after lysis in PBS, followed by detection with monoclonal anti-LexA antibodies. The LexA fusion proteins with a series of in-frame mutations are expressed in strain CLY3 for microdeletions (A), the ctd-Y964A allele (B), and the remaining mutants obtained in this study (C). The wild-type CTD and CTD mutants are indicated at the tops of the panels. Lanes with a minus sign at the top show the CLY3 signal in the absence of LexA. Arrows indicate the locations of LexA fusion proteins. (B) The CTD and ctd-Y964A mutant are expressed with similar abundances.

Given that most small deletions in the CTD were unable to confer silencing, it is possible that deletions may disrupt the overall structure of the CTD. However, based on the results from Western blotting, the deletions do not affect abundance and proteolytic products are not observed (Fig. 2A). Of course, we cannot rule out the possibility that multiple redundant sites are disrupted in these deletions. Since random mutagenesis with ethylmethane sulfonate failed to generate mutations in the CTD, we turned to alanine-scanning mutagenesis to generate missense mutations spanning the CTD. We also generated mutations at potential phosphorylation sites as defined by the NetPhos algorithm (2). Each residue tested was mutated to alanine, and the resulting mutants were screened to determine which were defective in TPE. Mutants with any of six missense mutations, S852A, Y900A, E915A, E955A, Y964A, and K973A, were significantly defective in TPE. Three mutant alleles, ctd-S852A, ctd-E915A, and ctd-E955A, conferred a two- to threefold loss of silencing; two mutant alleles, ctd-Y900A and ctd-K973A, conferred a 15- to 30-fold decrease in silencing; and ctd-Y964A conferred the most severe phenotype (FOA^r frequency, 9.76 × 10⁻⁶) (Fig. 3; see Table 3). The ctd-Y964A allele acts as a recessive allele: expression of both LexA-CTD and LexA-CTD-Y964A fusion proteins results in the restoration of tethered silencing (data not shown). Since the Y964A mutation is recessive, we have used the term ctd-Y964A. In the context of the full-length protein, sir3Y964A did not reveal any defect (data not shown), suggesting redundancy within Sir3.

FIG. 3.

Alanine scanning was used to identify mutants with missense mutations that are defective in TPE. Six mutants with missense mutations that had defects in TPE of twofold or higher (black bars) were isolated; the mutations confer statistically different results from those of the wild type at the 95% confidence interval. Gray bars represent TPE values that are not significantly different from those of the LexA-CTD fusions. The error bars represent the 95% confidence interval. Asterisks indicate sites of homology between species. These six mutations were conserved, while all other mutations were conserved in only 9/13 cases. The median value for the ctd-Y964A mutant is 9.8 × 10⁻⁶, with a sample size of 14.

TABLE 3.

Interaction between LexA-CTD and GAD-CTD fusion proteins is independent of Sir4

GAD/LexA fusion protein	Silencing frequencya	Association inb:
		wt	Sir4::KanR
CTD/CTD	5.7 × 10−2 (1.9-10.7; 35)	115.6 ± 36.0 (11)	128.4 (1)
CTD/Y900A	4.0 × 10−3 (1.0-46; 14)	142.5 ± 49.5 (4)	NT
CTD/Y964A	9.8 × 10−6 (1.6-56.6; 14)	124.3 ± 23.6 (8)	136.4 ± 10.3 (2)
CTD/K973A	6.5 × 10−4 (1.4-380; 14)	130.0 ± 39.6 (2)	NT
CTD/LexA	<1.77 × 10−6 (14)	4.2 ± 0.5 (2)	3.1 (2)

^aSilencing frequency is expressed as the fraction of cells that grow on 5-FOA. The ranges of the observed values and the numbers of samples are given in parentheses.

^bAssociation is expressed as the β-galactosidase activity in Miller units (U/mg) in CTY10-5D cell extracts containing both LexA and CTD fusion proteins. The numbers of samples are given in parentheses. NT, not tested.

The CTD contains a Sir3 dimerization activity.

Previous two-hybrid and in vitro approaches have suggested that several proteins associate with the Sir3 CTD, including Sir3 and deacetylated histones H3 and H4 (17, 41). However, specific CTD-protein interactions in vivo have not yet been demonstrated.

To test for CTD-CTD interactions, we generated a fusion protein that has the yeast GAL4 activation domain fused in frame with the CTD (GAD-CTD). The level of interaction between GAD-CTD and a series of LexA fusion proteins was then tested. Interactions between LexA-CTD and GAD-CTD fusion proteins were robust (Table 2) and were independent of the presence of Sir3 or Sir4 (Table 3; data not shown). The ctdΔ835-842 allele reduced silencing threefold without any significant change in dimerization (Table 1). In contrast, all other small in-frame deletions completely disrupted dimerization and failed to support TPE (Table 1). Previous studies have been unable to detect direct interactions with Sir2 in any domain in a multitude of contexts (18, 51). The two-hybrid data from the minideletions correlate well with the ability to silence, suggesting that amino acids 842 to 978 of the CTD comprise a dimerization domain that may be required for the function of tethered silencing.

CTD dimerization is partially maintained in related yeast strains.

Comparative phylogenetic sequence analysis is a powerful tool for the identification of functional sequence elements that may be conserved in evolution (7, 8). The putative Candida glabrata SIR3 gene has been identified (11). The N-terminal 215-amino-acid domain of C. glabrata Sir3 shares 45% identity with S. cerevisiae Sir3. Indeed, only the C-terminal sequence from amino acid 588 to 883 can be aligned with 21% identity to ScSir3 and 22% identity to ScOrc1. It is thus impossible to distinguish between the two proteins (data not shown).

Given the apparent high rate of divergence, we compared SIR3 sequences from the closely related Saccharomyces species Saccharomyces paradoxus, S. bayanus, and S. castellii (listed in order of similarity from highest to lowest) (Fig. 4A) (8, 9, 23). The N-terminal 215 amino acids of Sir3, known as a BAH (bromo-adjacent homology) domain, is found in many transcriptional regulation-related proteins and is highly conserved among S. paradoxus, S. bayanus, and S. castellii, sharing 95%, 85%, and 63% identity, respectively, with ScSir3 (Fig. 4B, top graph) (4). In contrast, the C-terminal 460 amino acids of Sir3 are once again more divergent, displaying 79%, 60%, and 20% identity, respectively, with ScSir3 (Fig. 4B, top graph). A comparison of the Saccharomyces C-terminal sequences of Sir3 from these species with either ScSir3 or ScOrc1 revealed a pattern of increasing similarity to Orc1 as a function of evolutionary distance from S. cerevisiae (Fig. 4B, bottom graph). Indeed, ScasSIR3 has 20% identity to ScSIR3 but 30% identity to ScORC1, making the identification of Sir3 equivocal.

FIG. 4.

The Sir3 CTD is divergent in sequence but functionally conserved. (A) Diagram of the relationship among the yeast species used in this study (light gray). mya, million years ago. (B) Sir3 N-terminal sequence and C-terminal sequence were compared among a subset of the Saccharomyces species. The N terminus contains a highly conserved bromo-adjacent homology domain of 215 amino acids, and the more divergent C-terminal Cdc6 domain contains 453 amino acids. The Sir3 N-terminal sequence and the C-terminal sequence were compared among Saccharomyces species (top graph). The C-terminal sequences of Sir3s among Saccharomyces species were compared with either ScSir3 or ScOrc1 (bottom graph).

To investigate whether SbSIR3 retains Sir3-mediated silencing, we generated an Sbsir3 null allele and examined silencing of the cryptic HML mating type loci in S. bayanus (Fig. 5A). We found that deletion of SbSIR3 impairs the mating ability of S. bayanus (Fig. 5A), indicating that, like ScSir3, SbSir3 is required for silencing of the HML silent cassette. Conversely, SbSir3 restores TPE in an S. cerevisiae sir3 null strain carrying a URA3-marked VIIL telomere. TPE (as measured by FOA^r frequencies) was only twofold lower than in cells carrying ScSir3. In contrast, the more distantly related ScasSIR3 displayed a 50-fold decrease in silencing relative to ScSIR3 (Fig. 5B).

FIG. 5.

SbSIR3 and ScasSIR3 can functionally complement the Scsir3 null strain. (A) The Sbsir3 null strain is unable to mate due to the silencing of the cryptic HMLα mating type locus. The SbSIR3 strain is S. bayanus strain Y245 with an auxotrophic trp marker. Sbsir3 is the SbSIR3 null strain. Strains EMPY75 and EMPY76 are the mating tester strains. Two clones were tested in each case. Diploids were detected on plates lacking components for the auxotrophic marker. (B) Depiction of URA3-marked VIIL telomere and graph showing TPE values that each protein can confer. The mean FOA^r frequencies were as follows: for pRS314, <5.8 × 10⁻⁷ (n = 6); for ScasSir3, 0.006 (n = 7); for SbSir3, 0.15 (n = 19); for ScSir3, 0.29 (n = 14). n, number of samples.

We used the yeast two-hybrid system to assay CTD interactions between related species. We found that both the S. cerevisiae CTD and the S. bayanus CTD can homodimerize robustly (Table 2). The interspecies interaction between the S. cerevisiae CTD and the S. bayanus CTD is, however, 10-fold weaker than the homodimerization. The same result was obtained using the reciprocal tagged proteins GAD-S. bayanus CTD and LexA-CTD. Interestingly, the S. bayanus CTD, when present in S. cerevisiae, decreased TPE 1,000-fold relative to that of the wild type (data not shown), consistent with the importance of CTD dimerization in the initiation of tethered silencing.

CTD dimerization is insufficient to nucleate tethered silencing.

We initially hypothesized that the ctd-Y964A mutation might disrupt CTD interaction and hence abrogate silencing. However, Y964A-Y964A, Y964A-CTD, and CTD-CTD interactions do not significantly differ in two-hybrid assays (Table 3). Since Y964A-CTD interactions fail to restore tethered silencing, these data argue further that dimerization is not the sole function of the CTD. This result is consistent with the TPE results described above, indicating that interspecies dimerization is not sufficient to confer silencing. We note that interactions in the two-hybrid system may be saturating and therefore may mask small differences in abundance between mutant and wild-type CTDs that may contribute to the ctd-Y964A phenotype.

The CTD is able to recruit chromatin-bound Sir3.

Since ctd-Y964A maintains the ability to form oligomers in both two-hybrid systems and chromosomal contexts (see below) but is unable to restore tethered silencing, we wanted to understand the process perturbed by Y964A. Previous studies have shown that Sir2, Sir3, and Sir4 form a core telomeric heterochromatin complex (49). To address how targeting of the CTD to the telomere facilitates the assembly of Sir proteins at the LexA binding sites, we examined the binding of Sir2, Sir3, and Sir4 by using ChIP assays. These data were compared with results from ChIP studies of cells transformed with a plasmid encoding LexA as a control.

In this study, we investigated the LexA binding site on the modified left arm of chromosome VIIL (Fig. 6A). The ChIP data from ACT1 sites were also used as specificity controls, since LexA fusion proteins fail to associate with the ACT1 locus. The PCR fragments at LexA binding sites are present as a doublet because one of the primers is located on the repeated LexA binding sites, as confirmed by restriction enzyme analysis (data not shown).

FIG. 6.

The CTD is able to recruit Sir2, Sir3, and Sir4 at LexA binding sites and to spread Sir proteins to the 0.5-kb-distal site B. (A) Schematic representation of PCR products used for ChIP experiments. LexA binding sites are present as doublets since one primer contains the LexA repeat sequence and LexA binding sites are present in several copies. Restriction enzyme digestion confirmed the identity of the doublet (data not shown). ACT1 sequences were used as a background control. (B, C) ChIP experiments with specific antibodies for LexA, Sir3, Sir2, and Sir4. The LexA fusion proteins bind with high efficiency to LexA binding sites. The abundances of LexA fusions at the LexA binding sites are normalized by ACT1 sequences. (D) Results of panel C. The CTD recruits Sir2, Sir3, and Sir4 at LexA binding sites and is able to spread to site B (0.5 kb distal to the LexA binding sites), while spreading is diminished in ctd-Y964A strains. The abundances of Sir2 (white), Sir3 (gray), and Sir4 (dark gray) at LexA binding sites and at site B sites were normalized relative to LexA. All experiments were repeated independently at least four times. The error bar represents the 95% confidence interval. α, anti.

The anti-LexA antibody was used to immunoprecipitate cells that contain LexA, LexA-CTD, or LexA-ctd-Y964A (Fig. 6B). Cells containing tethered LexA-CTD and LexA-ctd-Y964A have two- to threefold-higher levels of precipitated LexA than do cells containing LexA alone, which itself forms dimers and tetramers (50). The increased signal for LexA-CTD and LexA-ctd-Y964A at the LexA site is most likely mediated through the formation of CTD and ctd-Y964A multimers in the context of chromatin, consistent with the dimerization results.

As expected, LexA alone was unable to recruit Sir2, Sir3, and Sir4 (Fig. 6C). By contrast, recruitment by LexA-CTD of Sir2, Sir3, and Sir4 to LexA binding sites was enriched 4.4-fold, 11-fold, and 3.5-fold, respectively, relative to that by LexA (Fig. 6C). However, recruitment of Sir3 by LexA-ctd-Y964A was only 3.1-fold higher than that by LexA alone, reflected in its inability to nucleate TPE.

These results suggest that the LexA-CTD dimer recruits Sir2, Sir3, and Sir4 at LexA binding sites to nucleate the Sir complex, a process that is abrogated in ctd-Y964 cells.

Tethering of CTD dimers to LexA binding sites diminishes acetylated histone H4-K16.

Previous studies have shown that telomeric heterochromatin is composed of deacetylated histones H3 and H4 at specific N-terminal tail residues (16, 21, 22). In particular, deacetylation of the histone N-terminal tails appears to be required for Sir3 association (20, 24, 52, 53). We hypothesized that the CTD can recruit deacetylated histones H3 and H4 to telomeric heterochromatin in vivo. To this end, we used ChIP analysis to determine whether LexA-CTD and LexA-ctd-Y964A associated with deacetylated histones (Fig. 7).

FIG. 7.

ChIP assay shows that the CTD confers a decrease in the abundance of acetylated histone H4-K16 (AcK16H4). The abundance of acetylated histone H4-K16 at LexA binding sites was normalized relative to LexA. The experiments were repeated independently at least five times. The error bar represents the 95% confidence interval. α, anti.

ChIP assays using antibodies directed against acetyl-K16 H4 were used to test whether a low abundance of LexA-CTD DNA was precipitated relative to that of LexA (Fig. 7). LexA-CTD is precipitated with twofold-less efficiency by anti-acetyl-K16 H4 than is LexA alone. This was the only appropriate antibody available to measure acetylation in a mixed population of repressed and derepressed cells, with the repressed form present at only 5% in wild-type cells. The twofold change in anti-acetyl-K16 H4 precipitation may be an underestimate due to the predominance of derepressed cells even under repressive contrast. In contrast, cells transformed with LexA-ctd-Y964A have an abundance of specifically acetylated histone H4-K16 similar to that of cells containing LexA. Taken together, these data suggest that both CTD dimerization and/or the Sir2-catalyzed histone acetylation state is critical for the recruitment of Sir3 to heterochromatin. Although the results for anti-deacetylated histone H3 were generally consistent with these results, a high background in experiments led to variable results (data not shown).

Heterochromatic spreading in CTD-initiated silencing.

The data presented here also relate to the process of spreading (Table 4; Fig. 6D). Clearly, Sir2, Sir3, and Sir4 at the LexA sites are able to spread to site B (0.5 kb distal from the LexA binding sites), consistent with their ability to silence URA3. We interpret this to indicate that the Sir complex is sufficiently stable and intact to promote the step-by-step spreading. A different series of results is obtained when ctd-Y964A cells are compared at the two sites. First, a stable Sir complex is not created by the low abundance of components. Second, only one component, Sir3, shows an ability to spread from the LexA sites to site B. Indeed, we have observed spreading of Sir3 in these cells up to 6 kb from the LexA sites (data not shown) in the absence of the other bound components. This effect has been observed before in the extended chromatin observed after overproduction of Sir3 (47) and may be mechanistically related. Hence, we must consider the possibility that the unique properties of Sir3 allow it to spread, requiring only a transient association with Sir2 or other histone deacetylase activities.

TABLE 4.

Spreading in CTD-initiated silencing

Fusion protein	Site	Distance spreada (kb)
		Anti-Sir2	Anti-Sir3	Anti-Sir4
LexA-CTD	LexA	4.401	11.05	3.7
	Site B	1.747	2.7	2.35
LexA-ctd-Y964A	LexA	1.45	3.12	1.12
	Site B	0.87	1.51	1.12

^aNormalized to a pBTM value of 1.

In silico structural algorithms predict that the CTD resembles a Cdc6 winged-helix domain.

In silico studies can often provide hints as to future directions by providing potential structural models that become powerful tools when combined with genetics. Interestingly, in silico analysis, as described in Materials and Methods, revealed a predicted structural similarity of Sir3(525-921) to the crystallized structure of the Cdc6 protein of Pyrobaculum aerophilum (Protein Data Bank number 1FNN; E = 6 × 10⁻³⁵) (30). Interestingly, Cdc6 domain III contains predicted structural homology within the CTD (amino acids 835 to 921) (Fig. 8). Domain III of Cdc6 belongs to the winged-helix family of folds, a configuration present in both transcription factors and DNA binding proteins (30). Among these is the C-terminal region of the replication protein Orc1 (1, 30).

FIG. 8.

The CTD resembles domain III of the Cdc6 motif. The crystal structure of Pyrobaculum aerophilum Cdc6 (Protein Data Bank number 1FNN) domain III is shown, with the significant structural relatedness predicted for the regions shown in aqua. The alignments between P. aerophilum Cdc6, the CTD, the corresponding CTD of S. cerevisiae Orc1, and the S. bayanus CTD were used to determine likely regions of structural similarity. The sequence alignment was based on a CDD v2.03 BLAST search as described in Materials and Methods. V835 represents wild-type Sir3. VV, GV, and WV represent the insertion mutations with valine, glycine, and tryptophan inserted upstream of V835.

To test initially the hypothesis that the maintenance of a structural domain may be responsible for partial function, we constructed, in strain CLY3 rap1-17, a LexA-Orc1(738-914) fusion protein that contains a Cdc6 domain III with low similarity to Sir3. Since the CTDs of Sir3 and Orc1 seem to share structural similarity, we reasoned that the fusion protein might retain some activity in tethered silencing despite their sequence divergence. Indeed, tethering of Orc1(738-914) at the telomere conferred a small but significant restoration of silencing (median FOA^r frequency, 5.01 × 10⁻⁶; range of observed values, 0.67 to 54.4; sample size, 21) that is reproducibly more than fivefold higher than the LexA limit of sensitivity (median FOA^r frequency, <1.77 × 10⁻⁶; sample size, 14; P < 0.001). That these represent true TPE events is indicated by their epigenetic behavior: all FOA survivors were capable of switching back to the Ura⁺ state after induction of the URA3 gene on uracil omission media (data not shown), uniquely characteristic of the switching between silencing states. These correlative data are also consistent with the hypothesis of a functional evolutionary and structural relationship between Sir3 and Orc1 proteins (1, 12).

Structural constraints influence CTD function.

One prediction of the in silico model was that the putative “unstructured” hinge would not be required for activity when separated from domain II within Sir3. Indeed, these amino acids constitute the only portion of the CTD that is dispensable for function (Table 1). To examine the effects of sequences adjacent to the CTD, we inserted a glycine, valine, or tryptophan upstream of amino acid 835 (sir3-835GV, sir3-835VV, or sir3-835WV allele, respectively) (Fig. 8). These mutant alleles were integrated into a wild-type Rap1 strain carrying a URA3/ADE2-marked elongated VIIL telomere (Fig. 9A). The elongated telomeres cause the hyperrepression of subtelomeric ADE2 genes that are visualized as red colonies or sectors, while white colonies or sectors indicate derepression (44).Cells with wild-type SIR3 confer 85% red colonies (Fig. 9B). By contrast, cells carrying sir3-835VV or sir3-835WV alleles result in fully white colonies, while the sir3-835GV allele results in a high frequency of pink colonies, which is usually indicative of a high rate of switching between repressed and derepressed states. Silencing at the telomere-distal URA3 locus was also observed in sir3-835WV, sir3-835VV, and sir3-835GV (in decreasing order of frequency), consistent with a structural perturbation.

FIG. 9.

Insertion mutations adjacent to amino acid 835 diminish TPE. (A) The telomeric silencing assay was performed with BL22-2b strains containing wild-type RAP1 and a URA3-ADE2-marked VIIL telomere. (B) The silencing effect was measured by the ADE2 color assay. For wild-type SIR3 strains, repression of ADE2 results in red colonies. In contrast, no red colonies are observed for sir3-835VV and sir3-835WV mutants. Sir3-835GV has the weakest effect, with only 31% of cells showing as pink colonies or sectors.

Since Sir3 abundance and telomere length are critical factors that influence telomeric silencing, we determined the expression of each allele by Western blotting and found that Sir3 abundances in the sir3-835VV and wild-type SIR3 alleles are comparable. Furthermore, both SIR3 and the sir3-835VV, sir3-835WV, and sir3-835GV alleles have identical telomere sizes of 1.1 kb, eliminating telomere size as a possible source of variation in silencing efficiency (data not shown). Hence, the structure of the CTD may regulate its interactions with target proteins that are critical for the Sir3 silencing function. Determining whether it is the structure of the CTD or its arrangement or interaction with neighboring domains that is critical for activity must await crystallization studies of the CTD. Regardless, these data confirm the importance of the CTD in the context of the full-length protein.

DISCUSSION

Several key findings of this study help to elucidate the interactions that take place during Sir3-CTD-mediated initiation and spreading of heterochromatin. First, we show that the region comprising amino acids 842 to 978 (within the CTD) is the minimum functional domain required for the restoration of tethered silencing, regardless of amino-terminal suppressor mutations. This domain is functionally conserved in S. bayanus and S. cerevisiae, where it participates in a critical role in silencing.

Second, we provide evidence linking CTD dimerization and tethered silencing that is likely to identify a Sir3 dimerization site. This conclusion is based on interspecies comparisons and mutational analyses of the CTD.

Third, we have found alleles that retain dimerization but not activity, suggesting a second distinct function for the CTD.

Fourth, ChIP analysis has revealed that the CTD can help to model a complex of Sir2, Sir3, and Sir4 with hypoacetylated H4-K16. CTD-Y964A helps to recruit Sir3 with lower levels of Sir2 and Sir4 and a loss in the depletion of acetylated H4-K16. These data suggest that efficient recruitment of Sir3 may involve CTD dimerization and interactions between the CTD and Sir2-mediated deacetylated forms of histones H3 and H4.

Fifth, these studies suggest that the stability of the initial silencing complex has a profound effect on the spreading of all components of the Sir complex, although Sir3 has the capacity to maintain spreading over longer distances, reminiscent of earlier studies of Sir3-dependent spreading (47).

Finally, using in silico analysis, we found that the CTD resembles the winged-helix domain of the Cdc6 motif, a structure found in many DNA binding proteins and transcription-related proteins. This appears to be functionally relevant, since tethering the corresponding Cdc6 motif of Orc1 at telomeres confers a small but reproducible increase in tethered silencing, raising the possibility of an evolutionary functional link between these proteins.

What is the function of dimerization in tethered silencing?

Using diverged species as a form of “evolutionary” mutagenesis, we have shown that the CTD contains a minimum Sir3 dimerization domain that is able to recruit Sir3 to chromatin LexA binding sites. This conclusion is consistent with the importance of CTD dimerization in TPE in these diverged species. Three findings support this assertion. First, a 10-fold reduction in S. cerevisiae CTD/S. bayanus CTD association concomitantly reduces tethered silencing 1,000-fold. Second, a similar situation is found in the structurally homologous Orc1, which supports minimal TPE and is unable to dimerize with the CTD (data not shown). Third, all minideletions except for the most N-terminal deletion failed to dimerize. At the speculative level, the winged-helix motif of the proposed model for the CTD may also be responsible for dimerization.

However, dimerization does not act alone in Sir3 recruitment to chromatin. One indication of this is that mutations that eliminate or significantly reduce silencing, while supporting both homo- and heterodimerization, have a substantially reduced capacity to recruit Sir3 to chromatin in the heterologous organism. Furthermore, while ctd-Y964A is able to oligomerize robustly, its ability to associate with Sir3 under these conditions remains low. These data suggest that a component or interaction necessary for the silencing promoted by the CTD is absent.

CTD mutations result in a defect in the recruitment and spreading of Sir3 as well as the abundances of deacetylated histones.

As noted, the recessive ctd-Y964A mutation results in a severe defect in tethered silencing but is able to confer robust Y964A-Y964A and Y964A-CTD interactions, demonstrating a second function for the CTD.

In contrast, the ChIP profiles reveal that ctd-Y964A is able to recruit Sir3 at LexA binding sites, albeit with lower abundance than the CTD. However, ctd-Y964A is deficient in forming a complex with Sir2 and Sir4 at LexA binding sites and is insufficient to establish silent chromatin. Previous two-hybrid and in vitro studies have shown that Rap1, Sir3, Sir4, and deacetylated N-terminal tails of histones H3 and H4 are able to associate with Sir3 (5, 6, 16, 17, 28, 29, 40, 41, 47, 48). Rap1 and Sir4 are not involved in the interaction with the CTD. The CTD is unable to interact with Sir4, both in the two-hybrid assays and in vitro (data not shown), and the product of rap1-17 is unable to interact with both Sir3 and Sir4. By contrast, LexA alone or ctd-Y964A has a high abundance of acetylated H4-K16. The recessive phenotype produced by ctd-Y964A suggests a significant loss of interaction between ctd-Y964A and deacetylated H4-K16. Hence, CTD binding to deacetylated histones is likely to contribute to the formation of stable Sir3 recruitment and silencing, as demonstrated by ChIP using anti-deacetylated histone H3 in the majority of experiments (data not shown). It is noteworthy that within the context of the full-length Sir3 protein, the Y964A mutant does not display a loss of silencing. One explanation is found in recent data demonstrating a second histone interaction domain (amino acids 623 to 762) that is likely to act redundantly with the CTD in histone H3 and H4 association and dimerization, eliminating the essentiality of Y964 for silencing (25).

The hallmark of the assembly of silent chromatin is the nucleation at the cis-acting silencer (in this case the LexA binding sites), followed by spreading of the heterochromatic region. This spreading step involves homo- or heterointeraction of Sir2/Sir4 with Sir3 and Sir3-deacetylated histone interaction. We have shown that the CTD is able to nucleate silent chromatin at LexA binding sites and to spread Sir2, Sir3, and Sir4 to more-distal sequences, thereby acting as an artificial silencer. By contrast, although ctd-Y964A is able to recruit Sir3 at the LexA binding sites, it is unable to recruit either Sir2 or Sir4 at LexA binding sites. Interestingly, Sir3 appears to be able to spread alone.

A model for the cooperative interaction of dimerization and deacetylated histones in CTD (and Sir3)-mediated silencing.

Because the association between ctd-Y964A and Sir3 is insufficient to nucleate silent chromatin, we propose that the nucleation of silent chromatin requires a cooperative interaction between the CTD and Sir3 and between the CTD and Sir2-catalyzed deacetylated histones H3 and H4 (Fig. 10). When either of these two interactions is defective, silencing is defective. In this model, the CTD has dual recruitment functions to establish silent chromatin, (i) between CTD and Sir3 and (ii) between CTD and deacetylated H4-K16. The cooperative interaction between the CTD, Sir3, and deacetylated H4-K16 is likely to stabilize Sir3-CTD interactions, since Sir3 can bind only to deacetylated N-terminal tails of histones H3 and H4. The first deacetylation event can be explained by the previous findings that Sir2/Sir4 complexes can form independently of Sir3 and form catalytically active heterodimers that associate with chromatin. Indeed, in vitro binding assays have demonstrated that self-association of Sir3 can increase its affinity for Sir2/Sir4 (25). However, mutations in SIR4 do not interfere with CTD oligomerization. It is likely, therefore, that Sir2 can deacetylate the first histone, merely facilitating Sir3 association with chromatin.

FIG. 10.

A model for initiation and spreading by the CTD. Components of the interacting factors are shown, including a Sir2/Sir4 complex, Sir3, LexA-CTD (associated with the LexA sites), and N-terminally acetylated histones H3 and H4 within the nucleosomes (shown in pink). The CTD initiates silencing after both dimerization and Sir2 catalysis of the deacetylation of K9 and K14 of histone H3 and K16 of histone H4 (black nucleosomes), thereby forming a tight association between the CTD and Sir3. Sir3 association permits further Sir4/Sir3 associations. The Sir2/Sir4 complex moves unidirectionally to the next acetylated histone that, once deacetylated, allows for additional interactions with Sir3 and Sir4.

These results also speak to the issue of spreading in general. First, 25% of the wild-type Sir3 that is initiated at the LexA sites by the CTD is able to spread to site B, 0.5 kb distal to the LexA sites. Neither Sir2 nor Sir4 displays a pronounced spreading phenotype to site B, although any signal may be below the level of sensitivity of the ChIP assay. Interestingly, ctd-Y964A retained 50% of the original Sir3 association at site B. This is again dissimilar from the characteristics of Sir2 and Sir4. These data give rise to the hypothesis that Sir3 and the association of Sir3 with specific sites on histones may be a major force driving silencing into an adjacent domain. If so, it would appear that the ctd-Y964A mutant does not influence spreading, given its similar characteristics to wild-type Sir3. Previous data have shown a lack of stoichiometric association of Sir2 and Sir4 with Sir3 in extended chromatin when Sir3 is overproduced. Assuming that Sir3 can play at least part of the structural role of Sir4 at the more-distant sites, these data raise the possibility that Sir2 may retain catalytic activity but may bind too transiently to produce a signal in ChIP analyses. Hence, extended chromatin may be a minimal form of silencing that does not require all of the structural components of the core heterochromatin. An alternative possibility is that enzymes other than Sir2 are capable of performing its function in vivo.