Using a screen for genes that affect telomere function, we isolated sir3-P898R, an allele of SIR3 that reduces telomeric silencing yet does not affect mating. While sir3-P898R mutations cause no detectable mating defect in quantitative assays, they result in synergistic mating defects in combination with mutations such as sir1 that affect the establishment of silencing. In contrast, sir3-P898R in combination with a cac1 mutation, which affects the maintenance of silencing, does not result in synergistic mating defects. MATa sir3-P898R mutants form shmoo clusters in response to α-factor, and sir3-P898R strains are capable of establishing silencing at a previously derepressed HML locus with kinetics like that of wild-type SIR3 strains. These results imply that Sir3-P898Rp is defective in the maintenance, but not the establishment of silencing. In addition, overexpression of a C-terminal fragment of Sir3-P898R results in a dominant nonmating phenotype: HM silencing is completely lost at both HML and HMR. Furthermore, HM silencing is most vulnerable to disruption by the Sir3-P898R C terminus immediately after S-phase, the time when new silent chromatin is assembled onto newly replicated DNA.
IN eukaryotes with large chromosomes that are easily analyzed in the light microscope, heterochromatin was originally defined as chromosomal regions that stain darkly and appear to remain condensed during interphase. Heterochromatin replicates late in S-phase, often localizes to the periphery of the nucleus during interphase, and is less accessible to enzymes and DNA binding proteins (reviewed in Pardue and Hennig 1991). In the budding yeast Saccharomyces cerevisiae, the silent mating loci (HM loci) and telomere-adjacent sequences are organized into specialized domains of the genome that share these characteristics of heterochromatin (Grunstein 1998) and have a number of features in common. The four core histones and the silent information regulator proteins Sir2p, Sir3p, and Sir4p are structural components of both telomeric and HM locus heterochromatin (reviewed in Grunstein 1998). The two HM loci (HML and HMR) and telomere-adjacent regions compete with each other for the same silent chromatin components, presumably because the available pool of these proteins is limiting (Buck and Shore 1995; Marcandet al. 1996).
The process of forming heterochromatin is thought to proceed by a similar process at both the HM loci and at telomeres (reviewed in Grunstein 1998; Lustig 1998). First, DNA binding proteins bind to “silencer” sequences adjacent to the HM loci or to TG1-3/C1-3A tracts at the telomeres. The HM silencers bind Rap1p, Abf1p, and the origin recognition complex (ORC). At telomeres, Rap1p binds the the terminal (TG1-3) repeats. Second, at the HM loci, but not at telomeres, Sir1p associates with proteins bound to the silencer sites (Triolo and Sternglanz 1996; Foxet al. 1997). Third, the DNA binding proteins recruit (at HM loci with the help of Sir1p) the Sirp complex, which is composed of Sir2p, Sir3p, and Sir4p (Chienet al. 1993; Morettiet al. 1994; Lustiget al. 1996; Marcandet al. 1996). Fourth, the Sirp complex propagates a higher-order chromatin structure along the DNA, rendering it inaccessible to transcription factors and other enzymes (Hechtet al. 1996; Strahl-Bolsingeret al. 1997).
Three mechanistic processes, establishment, maintenance, and inheritance, play important roles in silencing. Miller and Nasmyth (1984) first demonstrated that de novo silencing of a derepressed HM locus could occur if cells pass through S-phase. Pillus and Rine (1989) observed that sir1 cells exist in two epigenetically distinct subpopulations: one in which HM silencing is normal and the cells are mating competent and one in which the HM loci are derepressed and the cells are mating defective. The de novo repression of an HM locus in cells previously carrying a derepressed HM locus, termed the process of establishment, is very inefficient in sir1 cells. Furthermore, the silent, mating-competent state is heritable in sir1 cells that were previously carrying a silent HM locus. This ability to promote the transmission of the silent state from mother to daughter cells is defined as inheritance.
The maintenance of silencing is defined as the process required after the formation of silent chromatin that retains the silent state during the same cell cycle. Defects in the maintenance of silencing were first detected experimentally by exposing MATa HMLα cells to α-factor. Wild-type cells arrest and form mating projections (shmoos) in response to α-factor while cac1 cells alternately shmoo (respond to α-factor, indicating that they can establish the silent state) and then divide (fail to respond to α-factor, indicating that they are no longer in the silent state). This alternation between responding and not responding to α-factor gives rise to “shmoo clusters,” microcolonies of cells with a shmoo morphology that are indicative of a defect in the maintenance of silencing (Enomoto and Berman 1998). Defects in the inheritance of silencing have been studied by monitoring the state of excised chromosomal HM loci containing or lacking the adjacent silencer sequences (Holmes and Broach 1996; Bi and Broach 1997; Chenget al. 1998; Ansari and Gartenberg 1999). The inheritance of the silent state into the next generation (involving passage of S-phase) required functional silencers in cis (Holmes and Broach 1996; Chenget al. 1998). If silencers were excised from the silent DNA, HML silencing was maintained during arrest on α-factor (Holmes and Broach 1996) or in vitro (Ansari and Gartenberg 1999). HM silencing was not maintained well in cells moving through the cell cycle (Bi and Broach 1997; Chenget al. 1998). However, in a small fraction (5%) of the cells, the silent state was inherited in the next cell cycle despite the absence of silencer sequences (Holmes and Broach 1996). This was interpreted to indicate that it is the silent chromatin itself, rather than the silencer sequences, that is inherited. The silencer sequences, however, improve the efficiency of that inheritance.
Distinguishing between establishment, maintenance, and inheritance is complicated by the fact that these three aspects of silencing appear to be interdependent and partially redundant processes. One can easily imagine that, in the presence of a highly efficient maintenance and inheritance, a defect in establishment will not manifest as a silencing defect. Conversely, if maintenance is defective, there will be no silent structure to inherit, and silencing will be highly dependent upon strong establishment.
Role of Sir3p in silencing: Sir3p is an important structural component of silent chromatin that is required for silencing at both HM loci and at telomeres. Sir3p interacts physically with Sir4p, Rap1p, Rad7p, the N termini of histone H3 and histone H4, and with other molecules of Sir3p (reviewed in Stone and Pillus 1998). Most of these interactions occur via the Sir3p C terminus (Sir3-C). The Sir3p N terminus (Sir3-N) modulates the activity of Sir3-C: interactions of Sir3p with Sir3-C and with Sir4p are increased when the Sir3p N terminus is deleted (Morettiet al. 1994; Gottaet al. 1998; Parket al. 1998). Sir3p is post-translationally modified into multiple phosphorylated forms (Stone and Pillus 1996). The classic experiment done by Miller and Nasmyth (1984) used an allele of SIR3 (sir3-8ts) to demonstrate that Sir3p is required for the maintenance of silencing throughout the cell cycle and that the de novo establishment of silencing requires functional Sir3p during S-phase.
High-copy expression of full-length Sir3p leads to increased silencing and increased spreading of telomeric silencing: silent chromatin extends farther inward from the telomere (Renauldet al. 1993) and Sir3p is physically associated with more centromere-proximal chromatin than in strains expressing wild-type levels of the Sir proteins (Strahl-Bolsingeret al. 1997). On the other hand, high-level expression of Sir3p domains affects the nuclear distribution of the Sirp complex and influences silencing in different ways: high-copy expression of the N-terminal domain of Sir3p (Sir3-N) enhances telomeric silencing and redistributes more Sirp complex to the telomeres (Gottaet al. 1998); high-copy expression of the C-terminal fragment of (Sir3-C) decreases telomeric silencing by promoting the localization of the Sirp complex to the nucleolus (Gottaet al. 1998), where it acts to reduce the silencing of, and frequency of recombination between, rDNA repeats (Smithet al. 1998). However, high-copy expression of Sir3-C does not have an obvious effect on HM silencing, suggesting that the HM loci compete effectively with the nucleolus for the Sirp complex while the telomeres do not. Thus different domains of Sir3p have markedly different effects on the cellular distribution of the Sir proteins and on silencing at the HM loci and telomeres.
Role of Sirp complex localization in silencing: Rap1p, Sir3p, and Sir4p co-localize with telomeric DNA to a small number of punctate foci near the nuclear periphery of wild-type cells (Gottaet al. 1996). This localization pattern often, but not always, correlates with the silent state of the HM loci and telomeres (Konkelet al. 1995; Gottaet al. 1996). RLF (Rap1 localization factor) genes were isolated by their effect on the segregation of TEL + CEN plasmids (circular plasmids carrying both centromere and telomere sequences; Enomoto et al. 1994a,b) and their effect on the nuclear distribution of Rap1p. rlf mutants alter both the segregation of TEL + CEN plasmids and the localization of Rap1p (Enomoto et al. 1994b, 1997). RLF2 is identical to CAC1, which encodes the largest subunit of the chromatin assembly factor I complex and causes a significant loss of silencing at telomeres but no obvious loss of silencing at the mating loci (Enomotoet al. 1997; Kaufmanet al. 1997; Monsonet al. 1997). RLF4 is identical to NMD2/UPF2 (Lewet al. 1998), a component of the nonsense-mediated mRNA decay pathway. Both rlf2 and rlf4 mutants disrupt telomeric silencing, are mating competent, yet have subtle defects in the maintenance of HM silencing (Enomoto and Berman 1998; Lewet al. 1998).
In this report we characterized a rlf3 mutant, a mating-competent allele of SIR3 that is defective in telomeric silencing. Like rlf2 and rlf4 alleles, sir3rlf3 mutants are mating competent, yet they exhibit synergistic mating defects in combination with mutations in silencer sequences or with loss of SIR1 function. The relevant mutation in sir3rlf3 alters the proline codon at position 898 to an arginine codon. The C-terminal fragment of Sir3rlf3p exhibits a novel anti-Sir phenotype when overexpressed: it causes loss of HM silencing as well as the loss of telomeric silencing. Furthermore, HM locus chromatin is most vulnerable to this anti-Sir activity during and immediately after S-phase, the time when chromatin is assembled onto newly replicated DNA.
MATERIALS AND METHODS
Strains and plasmids: Yeast strains used in this study are listed in Table 1. The temperature-sensitive sir3-8 allele was introduced into the W303 strain background by digesting pSH135 (Holmes and Broach 1996) with NruI and performing two-step gene replacement (Rothstein 1991).
Plasmids used in this study are listed in Table 2. pSE615 was constructed by in vivo recombination of pSE562 and pLL550. pSE562 was constructed by inserting the EcoRI fragment of sir3rlf3 [amino acids (aa) 439–972] from pSE393 into pACT-II (Liet al. 1994). pLL550 was constructed by inserting the BglII-SalI fragment of sir3rlf3 (aa 307–979) from pSE393 into pGAD424.
pSE647 and pSE856 were constructed by gap repair of pSE615 digested with BsiW1 + NdeI and transformed into a SIR3 strain to replace the sir3-P898R allele. DNA sequencing confirmed that the wild-type allele replaced the mutant allele in pSE856 and that in pSE647 the insertion of an adenine immediately after codon 853 led to a frameshift mutation generating 35 additional amino acids prior to a termination codon. pSE853 was constructed by gap repair of pM393 (digested with BsiWI and NdeI; Morettiet al. 1994) in sir3rlf3 strain (YJB966). The presence of the mutant allele was confirmed by sequencing. pSE1072 was constructed by recombination between pSE1033 (digested with XhoI and SphI) and pSE438 (digested with HpaI) such that the sir3rlf3 allele was replaced by the wild-type SIR3 allele pSE438. pSE1071 was constructed by recombination between pSE1033 (digested with XhoI and SphI) and pSE425 (digested with HpaI) to ensure that the XhoI-SphI fragment was the wild-type SIR3 allele. pSE1033 was constructed by recombination between pAR16 (digested with KpnI and EcoRV; Holmes and Broach 1996) and pSE715 (digested with XbaI) such that the PGAL-SIR3-1-979 gene was inserted into the PGAL-SIR3-C (439–987) region of pSE715. pSE332 was constructed by recombination between pJR273 (digested with PvuII) and pJkmf (−) (digested with HindIII; Kirschman and Cramer 1988). pSE334 is an in vivo recombination product of pSE332 (digested with PvuII) and YCplac111 (digested with HindIII; Gietz and Sugino 1988) such that SIR3 was inserted into the lacZ gene in YCplac111. pSE338 was constructed by inserting the 3′ region of (EcoRI-SalI fragment) SIR3 from pSE332 into YIPlac128 (Gietz and Sugino 1988). pSE650 was made by inserting the pSE615 BamHI-PstI fragment into the BglII-PstI-digested pSE497. pSE497 is analogous to pRSET-C (Invitrogen, Carlsbad, CA) with a kanamycin resistance marker replacing the β-lactamase gene (S. Enomoto, unpublished results). pSE912 was made by ligation of pSE856 (digested with BamHI and PstI) and pSE497 (digested with BamHI and PstI). pSE715 was made by recombination of pSE650 (digested with XbaI) and YGALSET351 (digested with XhoI; Enomotoet al. 1998). pSE481 is an in vivo recombination product of pSE438 (digested with PvuII) and YIPlac211 (digested with HindIII; Gietz and Sugino 1988).
Isolation of rlf3 alleles: The TEL + CEN plasmid screen used to isolate sir3rlf3 and methods used to isolate complementing genes were described previously (Enomoto et al. 1994a,b, 1997; Lewet al. 1998). Quantitative mating, telomeric silencing, and HMR::TRP1 derepression assays were performed as previously described (Enomotoet al. 1997; Enomoto and Berman 1998). For the quantitative mating assays, four matings were performed for each strain. The median values were compared by the rank sum test (Snedecor and Cochran 1980).
Different sir3rlf3 alleles were subcloned by gap repair. pSE334 was digested with HpaI and transformed into a SIR3/sir3rlf3 strain (YJB276 × YJB497). The HpaI sites flank the SIR3 open reading frame (ORF) at nucleotide (nt) −206 and nt 3485 relative to the ORF. A total of 10 independent plasmids with the appropriate restriction map were recovered and functionally tested for complementation of the telomeric silencing phenotype in YJB1033 (sir3 null). Two classes of plasmids were obtained and we chose one of each type for further characterization. pSE392 restored telomeric silencing to wild-type levels; pSE393 did not restore telomeric silencing to wildtype levels and thus contained the sir3rlf3 allele.
Mapping the lesion in the sir3rlf3 alleles: To map the mutations within the sir3rlf3 allele, pSE393 was digested at nt 487 with ClaI or at nt 2502 with KpnI and cotransformed with pSE332 digested with different restriction enzymes whose sites span the SIR3 gene. The telomeric silencing phenotype was checked for several transformants for each plasmid pair. This analysis indicated that the lesion in sir3rlf3 was located between the NruI site (nt 2283) and the XhoI site (nt 2833).
To reintegrate sir3 alleles, pSE481 was linearized by digestion with either KpnI (at nt 2502) or XhoI (at nt 2833), and two-step gene replacement of these alleles was performed into wild-type SIR3 strains. Candidate strains were assayed for TEL + CEN antagonism [by crossing them to YJB499 carrying p49K (Enomoto et al. 1994a,b)] and for telomeric silencing by monitoring expression from URA3 inserted at the left end of chromosome VII. Strains for assaying telomeric silencing were generated by transformation with pVIIL-URA3-TEL (Gottschlinget al. 1990) or by genetic crosses to strains carrying chromosome VIIL-URA3-TEL.
Sequencing of sir3rlf3 alleles: DNA sequencing of the entire SIR3 gene in both pSE392 and pSE393 was performed by the University of Minnesota microchemical facility, using the following primers:
acaggagatggtaccacgct, agcgtggtaccatctcctgt, tttatgcggcgtcc aaaa,
gtaaatagtcatttccttc, ttccggattttgtattaa, agtttattttgggaagac, caaa ccggtctaaaatta,
tgcttcatcagaactttc, ggtgatgtgagcgcagaa, gtttgggttccatttcct, tag atctggcctgaattg,
Mating assays: For the mating establishment assay, cells were pregrown on synthetic complete medium lacking leucine (SC-Leu; Shermanet al. 1986) containing 2% glucose (to prevent expression of PGAL-SIR3 or PGAL-sir3-P898R) at 25° and shifted to 37° for 18 hr (to inactivate Sir3-8p). Cells were transferred to SC-leu containing 0.5% galactose and 2% raffinose (to induce PGAL-SIR3 or PGAL-sir3-P898R expression) that was preheated to 37° and maintained at 37° (to continue inactivation of Sir3-8p) for the times indicated in Figure 4A. All subsequent manipulations were performed on plates prewarmed to 37° and incubated at 37°. Mating competence was assayed by streaking these cells across tester strain B364B (Table 1) on rich medium containing glucose (YPAD; which represses PGAL-SIR3 or PGAL-sir3-P898R expression) and allowing the cells to mate for 18 hr. The efficiency of mating was determined by analyzing the growth of cells on SC-his medium containing glucose, which selected for diploids formed in the test cross.
For mating assays with Sir3-R898P-C, strain YJB905 (ade2 leu2) containing LEU2-marked plasmids pSE856, pSE647, pSE615, or pACT II was mated with tester strain YJB199 (ADE2 leu2) and the formation of diploids was determined by selection on SC-leu-ade.
β-Galactosidase assays: β-Galactosidase assays were performed using standard methods (Ausubelet al. 1989). A minimum of four independent transformants were assayed at least twice each. All measurement values were normalized to that of PADH-LEXBD-GAL4AD = 10,000 units. The rank sum test (Snedecor and Cochran 1980) was used to assess the statistical significance of the values obtained between wild-type and sir3-P898R strains.
GST pull-down assay: Sir3p, Sir3-P898Rp, and Rap1p were produced in vitro in the presence of [35S]methionine using the TNT-coupled transcription and translation system (Promega, Madison, WI). GST-β-globin1–123, GST-histone H31–46, and GST-H41–34 were produced in Escherichia coli essentially as described (Smith and Johnson 1988; Hechtet al. 1995). Briefly, recombinant protein production was induced in E. coli strain BL21 carrying the plasmids by the addition of isopropyl thiogalactoside to 1 mm and incubation at 25° for 1 hr. Cells were collected by centrifugation and lysed by sonication in TGD150 [20 mm Tris-Cl (pH 8.0), 150 mm NaCl, 1 mm EDTA, 0.1% Triton X-100, 1 mm glutamate, 1 mm DTT, 1 mm PMSF]. Cell debris was removed by centrifugation and the resulting supernatant was incubated with 40 μl of a 50% slurry of glutathione-agarose (Sigma, St. Louis) for 45 min at room temperature. The resin was washed twice with TGD150 and then incubated with 35S-labeled protein in 100 μl buffer for 1 hr at room temperature. The resin was washed with buffer three times before all proteins were eluted by the addition of 20 μl of SDS loading buffer. Proteins were separated by SDS-PAGE and detected by autoradiography.
Arrest/release anti-SIR shmooing assay: Cells were grown in SC-leu containing 2% glucose and arrested by the addition of α-factor (0.1 μg/ml), hydroxyurea (400 mm), or nocodazole (10 μg/ml) for 4 hr at 25°. Arrested cells were collected by centrifugation and resuspended in SC-leu medium containing 2% galactose and 2% raffinose and the appropriate cell cycle inhibitor and incubated for 18 hr. Cells were then washed three times with fresh medium containing glucose and after a 20-min recovery were placed on SC-leu medium containing α-factor. Arrest and cycling of the cells was monitored by examination of cell morphology.
Identification of a novel allele of SIR3: Circular plasmids carrying both telomeric and centromeric DNA (TEL + CEN plasmids) are highly unstable (Longtineet al. 1992), a phenotype termed TEL + CEN antagonism. Furthermore, these TEL + CEN plasmids are stabilized by mutations that alter telomeric chromatin structure and function (Enomoto et al. 1994a,b). A screen for mutants that lost TEL + CEN antagonism initially identified a large number of mutants that were defective in mating and could be restored to wild-type mating competence by transformation with a copy of SIR2, SIR3, or SIR4 (Enomoto et al. 1994a,b). We focused our attention on mutants that also perturbed the nuclear localization of Rap1p [from the punctate, perinuclear foci that co-localize with a majority of telomeric DNA foci (Gottaet al. 1996) to a more diffuse distribution of Rap1p]. These mutants, which segregated as single genes and altered the localization of Rap1p, were termed Rap1p localization factor mutants and the mutated genes were designated “RLF genes.” Analyses of RLF2, which is allelic with CAC1, and of RLF4, which is allelic with NMD2/UPF2, have been reported elsewhere (Enomotoet al. 1997; Enomoto and Berman 1998; Lewet al. 1998).
Here we describe the characterization of rlf3, a novel mating-competent allele of SIR3. Strains carrying the rlf3 allele are defective in TEL + CEN antagonism, Rap1p localization, and telomeric silencing (see below). To identify the gene mutated in the rlf3 strain, we used a low-copy (CEN) library and isolated clones that restored the TEL + CEN antagonism phenotype and the telomeric silencing of the rlf3 strain. We isolated >80 clones, all of which contained the SIR3 gene. Because an extra copy of SIR3 can complement some nonallelic mutations (Enomotoet al. 1998), the isolation of SIR3 was not sufficient to indicate that the rlf3 mutation is an allele of SIR3, especially since the rlf3 strain was mating competent (see below) and all other reported sir3 alleles were defective in HM silencing and mating (Rine and Herskowitz 1987). To test the allelism of rlf3 and SIR3, rlf3 strain YJB497 was crossed to strain YJB998, which contained a LEU2-marked SIR3 allele. In five complete tetrads, the SIR3-LEU2 allele always segregated away from the rlf3 mutation, detected using TEL + CEN antagonism assays (Enomoto et al. 1994a,b), consistent with the idea that rlf3 is allelic to SIR3.
The rlf3 allele of SIR3 (sir3rlf3) was cloned by gap repair of a wild-type, plasmid-borne copy of SIR3. The sequence of the mutant allele identified six different missense mutations, five of them near the BglII site and one 3′ of the KpnI site (Figure 1A). The gap repair strategy mapped the mutant phenotype to the single cytosine-to-guanine transversion at nt 2693, which causes a proline-to-arginine substitution at amino acid 898. We confirmed that the single P898R substitution is responsible for the restoration of the TEL + CEN antagonism phenotype when expressed from a plasmid in a sir3 null strain. This allele is called sir3-P898R. We refer to the original allele as sir3rlf3 and to the five missense mutations near the BglII site collectively as sir3-5μ.
To generate a strain carrying a genomic copy of the rlf3 allele, we performed two-step gene replacements using different restriction enzyme digests that targeted replacement of different domains of the wild-type SIR3 allele with domains of the rlf3 allele. Replacement of almost all of SIR3 with sequence encoding aa 1–945 from the rlf3 allele resulted in abrogation of telomeric silencing (at least 10,000-fold lower than wild-type levels), similar to that seen with the original rlf3 mutant strain (Figure 1C, sir3-P898R, 5μ). In contrast, replacement of SIR3 with aa 834–978 (the C terminus of Sir3p) from sir3rlf3 resulted in telomeric silencing levels 1000- to 10,000-fold lower than wild-type levels (Figure 1C, sir3-P898R). Replacement of the five upstream mutations created a strain that had a very modest reduction in telomeric silencing (Figure 1C, sir3-5μ). These replacement experiments indicated that the major telomeric silencing defect was due to the P898R mutation and that the 5μ mutations enhanced the telomeric silencing defect. The studies described below were conducted primarily with strains carrying the sir3-P898R allele.
sir3-P898R mutations have subtle mating defects: To ask if the different mutations within the alleles contributed differently to the mating competence of these strains, we performed quantitative mating assays. We observed no significant difference in mating competence between wild-type, sir3rlf3, sir3-R898P, sir3-5μ, and sir3-R898P, 5μ strains (Figure 1B). While subtle defects in HM silencing are not detected by these mating assays, the results confirm that all of these sir3 alleles are mating competent.
There are several genes that, when mutated, significantly reduce telomeric silencing and have only very subtle effects on HM locus silencing. For example, strains lacking CAF-I subunit genes (CAC1, CAC2, and MSI1/CAC3) do not have an obvious mating defect in quantitative mating assays, unless they are combined with sir1 mutations (Enomoto and Berman 1998; Kaufmanet al. 1998). We constructed both MATa and MATα strains carrying sir3-P898R alone or together with sir1 or cac1 mutations and analyzed the mating ability of the strains in patch mating assays (Figure 2A). Similar to results seen previously with cac1 sir1 strains, MATa sir3-P898R sir1 mutants were defective for mating and MATα sir3-P898R sir1 mutants exhibited reduced mating ability. In contrast, sir3-P898R cac1 double mutants (both MATa and MATα) mated as efficiently as wild-type and single-mutant strains (Figure 2A). These results suggest that the sir3-P898R mutation and the cac1 mutations may affect a similar aspect of silencing (e.g., the maintenance of silencing) that is distinct from SIR1 function [e.g., the establishment of silencing (Pillus and Rine 1989)].
sir3-P898R mutations enhance defects in the HMR silencer: To measure the effect of sir3-R898P at HMR, we utilized an HMR::TRP1 construct, in which the a1 and a2 genes at HMR were replaced by the TRP1 gene (Hardyet al. 1992). Assays that measure expression of HMR::TRP1 are more sensitive to low levels of HMR derepression than are mating assays. We constructed a series of isogenic strains carrying the sir3-R898P allele and either HMR::TRP1 (including the intact silencer) or derivatives missing binding sites for either ORC, Abf1p, or Rap1p and compared the ability of these strains to grow on medium lacking tryptophan with the growth of isogenic SIR3 strains (Figure 3). As previously reported (e.g., Sussel and Shore 1991), of the SIR3 strains, only the strain missing the HMR E Rap1p site grew on medium lacking tryptophan (Figure 3). Of the sir3-P898R strains, the strain with the wild-type HMR::TRP1 allele and the strain with the HMR e-ΔAbf1 did not grow in the absence of tryptophan. However, the sir3-P898R HMR e-ΔORC site strain grew in the absence of tryptophan. Similarly, the sir3-P898R HMR e-ΔRap1 site strain grew very well in the absence of tryptophan. In this case the TRP1 gene was more derepressed than in the corresponding SIR3 HMR e-ΔRap1 strain, as evidenced by the larger size of the Trp+ colonies as well as by the increased frequency of Trp+ colonies. Because we observed a synergistic loss of silencing when we combined sir3-P898R with mutations that affect the establishment of silencing [sir1 and HMR e-ΔORC site (Susselet al. 1993)], these results are consistent with the hypothesis that the sir3-P898R allele is defective in the maintenance, but not the establishment, of silencing.
sir3-P898R is not defective in the establishment of silencing: At the HM loci, the de novo establishment of silencing is critically dependent on Sir1p. In sir1 mutant cells, HML exists in one of two epigenetic states: silent (off) or derepressed (on; Pillus and Rine 1989). Furthermore, the process of restoring the silent state to derepressed cells is very inefficient, requiring >40 generations (Pillus and Rine 1989). Experimentally, the establishment of HML silencing can be observed by monitoring the α-factor response of MATa cells (Pillus and Rine 1989). Wild-type MATa cells arrest in response to α-factor, forming cells with a single shmoo. MATa cells carrying mutations that completely derepress HML (e.g., sir3Δ strains or sir3-8ts cells held at 37°) do not respond to α-factor, continue dividing, and form colonies of cells. MATa sir1 cells respond in one of two ways: those that are silent at HML form shmoos and those that are derepressed at HML form colonies. To ask if a strain carrying the sir3-P898R allele, like sir1 strains, is defective in the establishment of silencing, we analyzed the α-factor responses of a sir3-P898R strain and a sir1 sir3-P898R strain. Unlike sir1 strains [but similar to the response of cac1 strains (Enomoto and Berman 1998)], the entire population of sir3-P898R cells initially formed shmoos that subsequently gave rise to shmoo clusters (Figure 2B). This indicates that HML in the sir3-P898R cells cannot be considered to be in different epigenetic states. Furthermore, the shmoo cluster phenotype is indicative of a defect in the maintenance of the silent state, since cells initially form shmoos (and thus have a silent HML locus), but the silent state is not as persistent in the mutant cells as it is in wild-type cells (Enomoto and Berman 1998). In contrast, sir1 sir3-P898R double-mutant cells are completely derepressed at HML, growing as a relatively uniform population of α-factor resistant colonies (Figure 2B). This result also supports the idea that sir1 and sir3-P898R do not affect the same aspect of silencing.
We previously noted that limiting the amount of Sir3p alone resulted in increased levels of shmoo clusters in otherwise wild-type cells (Enomoto and Berman 1998). All of the sir3-P898R phenotypes could be explained if Sir3-P898Rp was less stable than wild-type Sir3p. Immunoblot analysis detected similar steady-state levels of Sir3p in strains carrying the wild-type and any of the other sir3rlf3 mutant alleles including sir3-P898R as shown in Figure 1 (M. McClellan and S. Enomoto, unpublished data). Thus, the silencing phenotypes observed in the sir3rlf3 mutants are not due to a significant reduction in the stability of the mutant Sir3 proteins.
A second way to examine the role of SIR1-dependent de novo silencing is to monitor the kinetics of the restoration of mating competence by providing Sir3p to cells that were derepressed because they lacked Sir3p function. SIR1-dependent de novo establishment of silencing must include the recruitment of Sir3p to the HML locus and propagation of a silent chromatin structure that includes Sir3p. If Sir3-P898Rp is defective in being recruited by Sir1p, then we would expect a delay in the de novo formation of silent chromatin in a strain expressing sir3-P898R. We used the sir3-8ts allele (Miller and Nasmyth 1984) to generate cells in which HML was completely derepressed. Wild-type or mutant forms of Sir3p were provided by inducing PGAL-SIR3 or PGAL-sir3-P898R expression on medium containing galactose. Cells were pregrown at 37° on glucose to eliminate all silencing (Figure 4A, time 0). The cells were then streaked to medium containing galactose (to induce the expression of PGAL-SIR3 or PGAL-sir3-P898R) held at 37° (to keep sir3-8ts inactive) for different pulse time periods. Then the restoration of mating competence was tested by streaking across a tester strain on glucose medium (to repress PGAL-SIR3 or PGAL-sir3-P898R expression) held at 37°. The crosses were then replica-plated to glucose medium held at 37° that was selective for diploids resulting from successful mating (Figure 4A).
SIR1 strains expressing either PGAL-SIR3 or PGAL-sir3-P898R displayed similar mating kinetics: they restored mating ability to the sir3-8ts strain within 1.5–2.5 hr (Figure 4A). The sir1 strains did not restore mating for up to 4 hr and the kinetics of the appearance of mating competence was similar in the strains expressing PGAL-SIR3 or PGAL-sir3-P898R. The efficiency of mating at the earliest time points was slightly lower for both of the sir3-P898R strains relative to the SIR3 strains. However, the kinetics of the appearance of some mating-competent cells were similar when either PGAL-SIR3 or PGAL-sir3-P898R was expressed (Figure 4A). Thus, Sir3-P898R, like wild-type SIR3, was sufficient to restore silencing to chromatin that was previously in a transcriptionally active state. The fact that the kinetics of the appearance of mating competence was similar in the SIR3 and sir3-P898R strains suggests that sir3-P898R is not defective in the ability to be recruited to HML and to initially form silent chromatin.
In the sir1Δ strains, silencing and mating competence only appeared after much longer periods of time (∞ in Figure 4A), indicating that we can detect the sir1-independent establishment of silencing in this assay. The amount of time required in our assay was similar to that required (~2 days, >30 generations) for the subpopulation of HML-derepressed sir1 cells to become silenced as monitored by arrest and shmooing in response to α-factor (Pillus and Rine 1989; S. Enomoto, unpublished data).
sir3-P898R strains are defective in the maintenance and/or inheritance of telomeric silencing: The inheritance of silencing is defined as the ability of silent mother cells to produce silent daughter cells. Inheritance can only be monitored if the silent state is maintained during the previous cell cycle. Cells that inherit silent chromatin do not need to assemble the silent chromatin de novo, in part because the components of silent chromatin are already present at the silent loci. Monson et al. (1997) examined the “inheritance” of the silent state of a telomere-adjacent URA3 gene by pregrowing cells on 5-fluoroorotic acid (5-FOA) to enrich for those that were silent. They then examined the proportion of cells that formed daughter cells in which the telomeric URA3 gene remained silent. Because 5-FOA is toxic to any cell that becomes derepressed, cells that continue to divide are those that both maintained and inherited silent chromatin. Based on the assumption that any cell that failed to maintain the silent state would die on 5-FOA prior to the establishment of a de novo silent state, this assay measures only establishment-independent contributions to silencing. It cannot, however, distinguish between defects in the maintenance or the inheritance of silencing.
To measure the effect of the sir3-P898R allele on the inheritance and/or maintenance of telomeric silencing, we pregrew a strain YJB1267 on 5-FOA, transferred the cells to a fresh 5-FOA plate, and examined the size of the microcolonies formed after 18 hr of growth. The size of the microcolonies formed by the sir3-P898R strain were smaller (~40–60 cells/microcolony in most cases and occasional appearance of microcolonies with ~100 cells/microcolony) than the microcolonies formed by the SIR3 strain (~100–>1000 cells/microcolony; Figure 4B). In fact, colonies formed by wild-type SIR3 cells were comparable in size to colonies formed by a ura3 strain (Figure 4B). Since the size of the micro-colonies is a function of either maintenance and/or inheritance, this indicates that the sir3-P898R allele is defective in at least one of these functions.
The shmoo cluster assay measures the ability of cells to maintain the silent state during a single cell cycle. This 5-FOA survival assay measures the ability of cells to maintain and/or inherit the silent chromatin state. While it is formally possible that strains carrying the sir3-P898R allele are defective in both the inheritance and the maintenance of silencing, a defect in maintenance alone can account for all of the observed silencing defects in strains carrying the sir3-P898R allele.
Interactions of Sir3-P898Rp with Rap1p, Sir3p, Sir4p, Rad7p, and histones H3 and H4: The C-terminal domain of Sir3p (including amino acid 898) interacts with Sir3p, Sir4p, Rap1p, and Rad7p in the yeast two-hybrid system (Morettiet al. 1994; Paetkauet al. 1994). We compared the ability of the C terminus (aa 307–978) of Sir3p and Sir3-P898Rp to interact with Rap1p, Sir4p, and Rad7p using two-hybrid constructs that were originally used to reveal Sir3p interactions. In all these cases, we detected no significant difference in the interactions between the proteins and either Sir3p or Sir3-P898Rp. However, when we analyzed the Sir3p-Sir3p interactions of the mutant and wild-type proteins we found a significant difference: the Sir3-P898Rp with Sir3-P898Rp interaction was increased relative to the wild-type Sir3p with wild-type Sir3p interaction (Table 3).
Coprecipitation experiments have also demonstrated that Sir3p binds the unacetylated N-terminal tails of histones H3 and H4 in vitro (Hechtet al. 1995). Because the sir3-P898R mutation maps within the histone H3/H4 interaction domain of Sir3p (Hechtet al. 1995), we asked if the sir3-P898R mutation affected interactions between Sir3p and histones H3 and H4. Coprecipitation experiments were performed with either Sir3p or Sir3-P898Rp and either GST-HHT (aa 1–46) or GST-HHF (aa 1–34) (Hechtet al. 1995) (Figure 5). Sir3p and Sir3-P898Rp were expressed as fusions with an N-terminal histidine6-T7-gene-10-epitope tag by in vitro transcription and translation (see materials and methods). As a negative control, we used a histidine6-T7-gene-10-epitope-tagged Rap1p (Enomotoet al. 1998). As expected, the wild-type Sir3p fusion protein coprecipitated with histones H3 and H4 and the Rap1p fusion protein did not coprecipitate with either of the histones (Figure 5). Like Sir3p, Sir3-P898Rp coprecipitated with histones H3 and H4 and the affinity of the wild-type and mutant Sir3 proteins for the histones was indistin-guishable (Figure 5). Thus, the sir3-P898R mutation did not alter the ability of the protein to interact, in vitro, with unacetylated N-terminal tails of histones H3 and H4.
The sir3-P898R C terminus confers a strong nonmating phenotype: Interestingly, during our analysis of Sir3-P898Rp interactions, we found that two-hybrid plasmids (both “binding domain” and “activation domain” constructs) carrying codons 307–978 of sir3-P898R interfered with the mating ability of the otherwise wild-type two-hybrid reporter strain (Figure 6). Expression of these plasmids in other MATa strains resulted in a similar nonmating phenotype (data not shown). The Sir3-P898R 307-978 plasmids also conferred a nonmating phenotype on MATα strains (data not shown). Similarly, a strain carrying TRP1 within the HMR locus was Trp+ when the Gal4-AD-sir3-P898R allele was expressed (data not shown). During the course of these studies, overproduction of the C terminus of wild-type Sir3p was reported to interfere with telomeric silencing (Leet al. 1997; Gottaet al. 1998; Parket al. 1998). We observed a similar effect with the C-terminal fragments of both Sir3p and Sir3-P898R (Figure 6). However, the mutation in sir3-P898R causes an increased level of derepression relative to the wild-type SIR3 allele: the Sir3-P898Rp C-terminal fragment (Sir3-P898R-C) caused a complete loss of mating competence in patch mating assays (Figure 6). These results indicate that the sir3-P898R mutation enhances the ability of the Sir3p C terminus to derepress silencing at both telomeres and the HM loci.
The C terminus of Sir3-P898Rp disrupts silencing specifically during late S-phase: To better understand how the C-terminal fragment of Sir3-R898Pp interferes with HM silencing, we asked whether the disruption of silencing by the mutant protein occurred during a particular stage of the cell cycle. One possibility was that Sir3-R898Pp could interfere with silencing at any stage of the cell cycle, perhaps by titrating away a component of the normal silent chromatin complex. Another possibility was that Sir3-R898Pp could interfere with silencing by being physically assembled into the silent chromatin complex in late S-phase. For these experiments we expressed the C-terminal portion of Sir3-R898P from the galactose-inducible GAL10 promoter. Cells were pregrown in glucose, which prevented PGAL-sir3-R898P expression; in these cells the HM loci were repressed as evidenced by their sensitivity to α-factor. Cells were then arrested in G1 by the addition of α-factor, in S-phase by the addition of hydroxyurea, or in M-phase by the addition of nocodazole. Four hours later, cells were shifted to medium containing the same cell cycle inhibitor plus galactose, to induce expression of sir3-R898P307–979 during the cell cycle arrest. Cells were held under these conditions for 18 hr, washed three times with fresh glucose medium, and released into glucose medium for a brief recovery period. α-Factor was then added to the medium to monitor the mating response of the MATa cells to α-factor in the subsequent cell cycle. Most of the cells that had been arrested in G1 with α-Factor or in M-phase with nocodazole during the induction of the C-terminal fragment of Sir3-R898P responded to α-factor in the subsequent cell cycle by arresting and forming a mating projection (Figure 7), indicating that the silent mating loci remained silent in these cells despite the presence of the sir3-R898P C terminus. In contrast, 66% of the cells that had been arrested with hydroxyurea were α-factor resistant, indicating that the HMLα silencing had been perturbed in the majority of these cells. Control cells carrying only the vector and arrested with hydroxyurea continued to respond to α-factor (S. Enomoto, unpublished data). These results indicate that Sir3-R898Pp interferes with the assembly of silent chromatin during or just after replication. This could occur either by being incorporated directly into the chromatin or by interfering with the assembly of another component required for the complete silencing of the HM loci.
Tethering wild-type Sir3p cannot bypass the sir3-P898R silencing defect: Telomeric silencing is thought to be nucleated by the binding of Rap1p at the telomere repeats, recruitment of the Sirp complex (by the Rap1p C terminus interacting with Sir3p and Sir4p), and propagation of the Sirp complex onto the telomere-adjacent DNA. In a rap1-17 strain, telomeric silencing does not occur because the Rap1-17p does not interact with Sir3p and Sir4p. However, tethering LEXA-Sir3p to telomere-adjacent lexA operator sites in a rap1-17 strain permits silencing of a telomeric URA3 gene (Lustiget al. 1996). If the major defect in sir3-P898R is in an early step of silencing, such as recognizing and binding to proteins like Rap1p, or being recruited to the Sirp complex, it might be possible to suppress this defect by tethering wild-type Sir3p, to weakly bypass early initiation steps. We compared the ability of wild-type Sir3p to generate silent chromatin in a rap1-17 SIR3 strain and a rap1-17 sir3-P898R strain (Figure 8). As previously reported, the rap1-17 SIR3 strain expressing pLEX-SIR3 produced FOA-resistant colonies at a frequency of 10−3 (Lustiget al. 1996). In contrast, tethering pLEX-SIR3 in the rap1-17 sir3-P898R strain did not bypass the rap1-17 silencing defect: no FOA-resistant colonies were observed (Figure 8). Thus, the silencing defect in sir3-P898R strains is seen both when tethered Sir3p initiates or when normal telomere sequences initiate silencing, implying that the defect in sir3-P898R cannot be in an early step in silencing. Rather, the major defect must be a problem in either the propagation/assembly of the silent chromatin/Sirp complex or in the stable maintenance of the complex. This assay does not allow us to distinguish between a defect in the assembly process, which would form an unstable Sirp complex structure, and a defect in Sirp complex structure itself.
sir3-P898R is an interesting allele of SIR3 that allows us to dissect some of the separable roles of Sir3p in the processes of establishing, maintaining, and inheriting silent chromatin. Sir3p is an important component of silent chromatin at both telomeres and at the HM loci. Here we identified and characterized rlf3, an allele of SIR3 that affects TEL + CEN antagonism, Rap1p localization, and telomeric silencing without an obvious effect on HM silencing. The original sir3rlf3 allele included a point mutation near the C terminus (R898P) that accounts for the majority of the phenotypes observed in the original allele. Five additional mutations slightly enhanced the telomeric silencing defect in the original sir3rlf3 allele.
We used several assays to analyze the silencing defect in sir3-P898R strains. In qualitative and quantitative assays, sir3-R898P did not cause any obvious defects in HML or HMR silencing (Figures 1B, 2A, and 3). Yet in combination with mutations that affect the establishment of HM silencing [e.g., sir1 or ORC site silencer mutations (Pillus and Rine 1989; Susselet al. 1993)], sir3-P898R caused a significant loss of HM silencing and mating competence (Figures 2A and 3). This synergistic decrease in silencing between mutations in sir1 (or silencer sites) and sir3-P898R can be interpreted in a number of ways. One possibility is that the two genes (e.g., SIR1 and SIR3) encode proteins that have similar, at least partially redundant, functions and when both are missing the function cannot be accomplished. An alternative interpretation is that the two genes encode proteins that have distinct functions that are dependent upon one another. We favor the latter interpretation because we do not have any data to support the idea that sir3-P898R affects the establishment of silencing. Using more sensitive assays, we found that sir3-P898R strains formed primarily shmoo clusters in response to α-factor (Figure 2B), indicating a defect in the ability to sustain the silent state of HML. This is very different from how sir1 mutations affect silencing: sir1 mutants either arrest or divide in response to α-factor and do not form shmoo clusters. In addition, the kinetics of the establishment of de novo silencing at HML was similar for Sir3-P898Rp and wild-type Sir3p (Figure 4A), indicating that initial steps of establishment were not defective in sir3-P898R strains, and thus implying that it is later steps in silencing that are affected. At a URA3-marked telomere, sir3-P898R strains grown on FOA had a defect in the maintenance and/or inheritance of the silent state (Figure 4B). Furthermore, while wild-type tethered Sir3p was able to initiate silencing (Lustiget al. 1996), it could not do so in a sir3-P898R strain (Figure 8), indicating that silencing in sir3-P898R was defective even if the initiation step was provided (by bypassing normal initiation via tethered Sir3p at the telomere). This implies that a step after the initiation of silencing is defective in sir3-P898R strains. The simplest explanation consistent with all of these results is that the primary defect in sir3-P898R strains is a defect in the maintenance of silencing. While we cannot rule out the possibility that Sir3-P898Rp has additional defects in the inheritance of a silent chromatin structure, a defect in the maintenance of silencing is sufficient to account for all of the results observed.
Sir3-R898P-C has a strong dominant negative effect on HM silencing, especially during S-phase: Sir3-C interacts with other proteins (e.g., Rap1p, Sir4p, Sir3p, and the N termini of histones H3 and H4) that form stable silent chromatin. Like wild-type Sir3-C, Sir3-R898P-C has a dominant negative activity that interferes with telomeric silencing. The assembly of silent chromatin requires passage through S-phase (Miller and Nasmyth 1984), presumably because the state of the chromatin is reformed following passage of the replication fork. We found that cells released from HU in the presence of Sir3-R898P-C did not remain in the silent state efficiently while cells released from either α-factor or nocodazole usually remained in the silent state despite the presence of Sir3-R898P-C. This result is consistent with the idea that Sir3-R898P-C affects HM silencing by being assembled directly into the chromatin following early S-phase, and likely after the replication of the silent chromatin. Alternatively, Sir3-R898P-C may interfere with silencing during S-phase by associating with some factor that is required for the formation of silent chromatin especially during S-phase. In either case, the interference with silencing must have occurred sometime between early S and G2, since expression of Sir3p-R898P-C during and after release from nocodazole did not have much effect on silencing. The interference in silencing may be due to the increased affinity of Sir3-P898R-C for other Sir3-P898R-C molecules as detected in the two-hybrid assays (Table 3).
These results differ slightly from the previous studies of silencing and cell cycle of Gottschling and colleagues, who found that silent chromatin is most accessible to transcription factors during arrest with nocodazole (which holds cells at G2/M; Aparicio and Gottschling 1994). An important difference between these two series of experiments is that here we analyzed the maintenance of silencing after release from a cell cycle arrest, while Aparicio and Gottschling (1994) monitored transcription factor accessibility in cells during cell cycle arrest. The dynamics of chromatin accessibility are likely to be different when cells are cycling than when cells are arrested. Support for this idea comes from studies that monitored the state of excised chromosomal HM loci containing or lacking the adjacent silencer sequences (Holmes and Broach 1996; Bi and Broach 1997; Chenget al. 1998; Ansari and Gartenberg 1999). The inheritance of the silent state into the next generation (involving passage of S-phase) requires functional silencers in cis (Holmes and Broach 1996; Chenget al. 1998). If silencers are excised from the silent DNA, HML silencing is maintained during arrest on α-factor (Holmes and Broach 1996), while chromatin is altered during arrest with nocodazole (Bi and Broach 1997). Yet HM silencing is not maintained well in cells moving through the cell cycle (Bi and Broach 1997; Chenget al. 1998), even if the DNA is not replicated (Chenget al. 1998).
sir3-P898R may interfere with silencing by increasing the affinity of Sir3p-Sir3p interactions: Sir3p interacts with many components of silent chromatin (reviewed in Stone and Pillus 1998) and many of these interactions are necessary for silent chromatin function. We envision three types of molecular interactions that Sir3p may use to contribute to silencing: nucleation, propagation, and stabilization.
Sir3p would contribute to the nucleation of silencing at both telomeres and the HM loci by interacting with proteins such as Rap1p (Morettiet al. 1994) and Sir4p (Moazedet al. 1997) that associate with the silencer sequences (Lustiget al. 1996). (In tethering experiments, this type of interaction is bypassed.)
Interactions between Sir3p and the other components of the silent chromatin, including Sir proteins and nucleosomes, may stabilize contacts between the Sirp complex and the silenced DNA (Hechtet al. 1996; Strahl-Bolsingeret al. 1997).
Finally, we propose that the relative strength of the interactions between Sir3p and proteins at the silencer site, proteins in the Sir complex, and proteins in the nucleosomes must be balanced so that silent chromatin is appropriately organized on the DNA. If any one of these interactions is too strong [which is likely the case for sir3-P898R-sir3-P898R interactions (Table 3)], an aberrant structure that is less effective in overall silencing would be formed.
Since the kinetics of establishment are similar between SIR3 and sir3-P898R strains (Figure 4A), our results suggest that the nucleation functions of Sir3p are not significantly affected in sir3-P898R mutants. The strong dominant negative effect of the Sir3-R898P-C allele on silencing is consistent with the fact that Sir3-R898P-C/Sir3-R898P-C interactions were stronger than Sir3C/Sir3C interactions in two-hybrid experiments. We propose that the stronger protein-protein interactions of Sir3-P898Rp relative to Sir3p perturb the stability of the Sirp complex, leading to a defect in the maintenance of silencing in sir3-P898R strains. That stronger protein-protein interactions can inhibit the function of complexes has been observed in other systems as well (Sandrocket al. 1997). Interestingly, rap1-12 mutants also have stronger interactions with Sir4p in the two-hybrid system. In the case of rap1-12, however, the increased interaction between Rap1-12p and Sir4p caused reduced silencing at the HM loci because of a limiting supply of Sir4p that was preferentially interacting with the Rap1-12p present at higher levels at the telomeres. In the case of sir3-P898R, the altered Sirp complex interactions affect the maintenance of a silent Sirp complex, especially in regions of the genome where the establishment of silencing is less efficient.
Why are telomeres more vulnerable to defects in the maintenance of silencing? sir3rlf3 strains have a dramatic reduction in telomeric silencing but only very subtle defects in HM silencing. Silencing at telomeres is less stable than HM silencing. This epigentic nature of telomeric silencing is likely due to less efficient establishment of silencing: tethering Sir1p to telomeres improves telomeric silencing (Chienet al. 1993), presumably by improving the establishment of silencing. In contrast, the establishment of silencing at HM loci is strong and partially redundant: two silencer sites, E and I, can nucleate silent chromatin, and Sir1p specifically improves the establishment of silencing at the HM loci. In fact, in sir1 cells, HML behaves much like telomere-adjacent sequences in wt cells: it is silent in some cells and is actively expressed in others. And, in sir1 sir3-P898R strains, HM silencing is dramatically reduced.
Strains carrying mutations in CAC1, which encodes the large subunit of CAF-I, have many phenotypes similar to those seen in sir3-P898R strains. Both sir3-P898R and cac1 mutations give rise to shmoo clusters and do not influence the kinetics of the de novo establishment of silencing (Figures 2B and 4A; Enomoto and Berman 1998), suggesting that they cause defects in the maintenance of silencing. Furthermore, sir3-P898R and cac1 mutations each reduce telomeric silencing dramatically and have no obvious mating defects, yet exhibit significant reduction in mating efficiency when combined with sir1 mutations (Figures 1 and 2). Other mutants exhibiting these characteristics include the eso mutants (Stoneet al. 2000). We propose that, like sir3-P898R, other mutations that affect telomeric silencing, but not HM silencing, may do so because they affect mechanisms involved in the maintenance of silencing and because the establishment of silencing is naturally weaker at telomeres. We predict that such mutations will, like sir3-P898R, also affect HM silencing if establishment is weakened by mutations in the silencers or by mutation of SIR1. This is indeed the case for mutations in CAC1/RLF2, CAC2, CAC3/MSI1, HHF1, and HHT1: strains carrying mutations in these genes exhibit reduced HM silencing in combination with sir1 mutations (Thompsonet al. 1994; Enomoto and Berman 1998) and give rise to the shmoo cluster phenotype (Enomoto and Berman 1998).
We thank Stan Fields, Lee Hartwell, and Art Lustig for providing strains; and Jim Broach, Daniel Gietz, Dan Gottschling, Michael Grunstein, Art Lustig, and David Shore for providing plasmids. Weare grateful to Elisa Stone and Lorraine Pillus for providing results prior to publication. We thank members of the Berman laboratory for helpful discussions and Mark McClellan and Lewis Lukens for technical assistance. This work was supported by National Institutes of Health GM38616 to J.B. S.D.J. was supported by the National Institute of General Medicine 1 F32 GM19065-01.
Communicating editor: F. Winston
- Received October 12, 1999.
- Accepted March 3, 2000.
- Copyright © 2000 by the Genetics Society of America