Sir1 establishes transcriptional silencing at the cryptic mating-type loci HMR and HML (HM loci) by recruiting the three other Sir proteins, Sir2, -3, and -4, that function directly in silenced chromatin. However, SIR1-independent mechanisms also contribute to recruiting the Sir2–4 proteins to the HM loci. A screen to elucidate SIR1-independent mechanisms that establish HMR silencing identified a mutation in YKU80. The role for Ku in silencing both HMR and HML was masked by SIR1. Ku's role in silencing the HM loci was distinct from its shared role with the nuclear architecture protein Esc1 in tethering the HM loci and telomeres to the nuclear periphery. The ability of high-copy SIR4 to rescue HMR silencing defects in sir1Δ cells required Ku, and chromatin immunoprecipitation (ChIP) experiments provided evidence that Ku contributed to Sir4's physical association with the HM loci in vivo. Additional ChIP experiments provided evidence that Ku functioned directly at the HM loci. Thus Ku and Sir1 had overlapping roles in silencing the HM loci.
TRANSCRIPTIONAL repression of the cryptic mating-type loci HMR and HML (the HM loci) in the yeast Saccharomyces cerevisiae occurs through a mechanism called transcriptional silencing. Silencing describes the formation of a specialized “silent” chromatin structure that is analogous on many levels to metazoan heterochromatin (reviewed in Rusche et al. 2003). Specifically, silencing causes heritable, position-dependent, long-range transcriptional repression that relies on interactions between nucleosomes, chromatin-modifying enzymes, and nonhistone chromatin-binding proteins. In addition, several proteins and chromatin modifications involved in silencing are conserved in metazoans and have roles in heterochromatin (reviewed in Perrod and Gasser 2003). An understanding of silencing requires identification of the proteins and protein–protein and protein–DNA interactions required to target silent chromatin formation to the appropriate domains in the genome. A challenge to obtaining a comprehensive description of silencing at the HM loci using genetic approaches is that several overlapping mechanisms contribute to a robustness in silent chromatin structure such that the functions of key proteins and molecular interactions can be masked. In this study we analyzed silencing compromised by a deletion of SIR1 (silent information regulator; sir1Δ) and discovered that the yeast Ku complex had a redundant role with Sir1 in silencing the HM loci.
Silencing is targeted and stabilized at the HM loci by small (∼150-bp) DNA elements called silencers (Loo and Rine 1995). HMR and HML are each flanked by two silencers that contain different combinations of binding sites for the nuclear proteins ORC (origin recognition complex), Rap1, and Abf1. The role of the silencer-binding proteins is to recruit the four Sir proteins, Sir1, -2, -3, and -4, through a number of distinct protein–protein interactions to silencers and “nucleate” silencing (reviewed in Rusche et al. 2003; Moazed et al. 2004; Fox and McConnell 2005). In current models, once the Sir2–4 proteins, which form a stable complex (Rudner et al. 2005), are positioned at a silencer, the enzymatic activity of Sir2 deacetylates nucleosomes that neighbor the silencer, which in turn promotes binding of an additional Sir2–4 complex. The process repeats until Sir2–4 protein complexes bind and modify the nucleosomes that compose the silent loci (reviewed in Rusche et al. 2003; Moazed et al. 2004). The binding of the Sir2–4 protein complexes may directly repress transcription and/or lead to higher-order looping and compaction that renders the underlying DNA less accessible (Vazquez and Schedl 1994; Cheung et al. 2000; Sekinger and Gross 2001; Valenzuela et al. 2008).
In this model, the Sir2–4 protein complexes are essential structural components of silent chromatin, whereas Sir1 has a different role. In particular, Sir1's role is confined to silencers where it enhances the establishment (i.e., the initial recruitment of a Sir2–4 protein complex to the silencer) and/or reduces the disassembly of silent chromatin (stabilizes a preassembled but meta-stable silent chromatin structure) (Pillus and Rine 1989, 2004; Xu et al. 2006). However, at natural HMR and HML a substantial level of silencing is retained even in the complete absence of Sir1 (i.e., in sir1Δ cells) (Ivy et al. 1986; Pillus and Rine 1989), indicating that the Sir2–4 protein complex can nucleate at the silencers and establish silencing fairly efficiently without the help of Sir1. In addition, defects in silencing caused by mutations in silencer-binding sites or SIR1 can be suppressed by many different mutations that modulate progress through the cell cycle (Axelrod and Rine 1991; Laman et al. 1995; Ehrenhofer-Murray et al. 1999), suggesting that a cell-cycle-regulated target(s) can modulate Sir2–4 protein interactions with silencers. In a recent study, we presented evidence for a pathway shared by the cell-cycle transcription factor Fkh1 and the S-phase cyclin Clb5 that modulates the establishment of Sir2–4 chromatin at HMRa in sir1Δ cells (Casey et al. 2008). Specifically, high-copy expression of FKH1 or a deletion of CLB5 (clb5Δ) partially bypasses the requirement for SIR1 in silencing HMR. Furthermore, genetic epistasis and molecular experiments reveal that FKH1 and CLB5 exert their effects on silencing through a common pathway yet do so without regulating each other's expression. The simplest model to explain these data posits that FKH1 is a positive regulator while CLB5 is a negative regulator of a common target or pathway that in turn contributes to silencing via a SIR1-independent mechanism (Figure 1). Identification of such a target should provide insights into the connection between the cell cycle and silencing. However, regardless of the specific mechanisms involved, this and other studies establish clearly that Sir1-independent mechanisms involving the Sir2–4 proteins contribute substantially to silencing of the HM loci.
In this study, a genetic screen was performed to identify genes that contributed to SIR1-independent pathways involved in HMRa silencing. This screen identified YKU80 that encodes one subunit of the evolutionarily conserved Ku70/Ku80 heterodimer, referred to as the Ku complex or Ku. Ku has a well-characterized role in silencing telomeres, but its role in silencing the HMR and HML loci has been unclear. In particular, although Ku and another protein, Esc1, work redundantly to maintain HMR's localization near the nuclear periphery (Gartenberg et al. 2004), genetic evidence has failed to establish that these proteins have a functional role in silencing HMR. Genetic and chromatin immunoprecipitation (ChIP) experiments described in this report provided evidence that the inability to detect a role for Ku in HM silencing was because Ku and Sir1 shared overlapping functions in recruiting Sir4. Additional ChIP experiments provided evidence that Ku functioned directly at the HM loci. Ku's role in silencing the HM loci was distinct from its overlapping role(s) with Esc1 since ESC1 did not contribute to silencing the HM loci even in sir1Δ cells. In addition, Ku's role in HMRa silencing in sir1Δ cells worked independently of and functioned additively with the FKH1/CLB5 pathway. Thus Ku acted in a previously uncharacterized SIR1-independent pathway for targeting Sir2–4 chromatin to the HM loci.
MATERIALS AND METHODS
Strains and plasmids:
Yeast strains and plasmids used in this study are listed in Tables 1 and 2, respectively. The strains used in this study were constructed using standard yeast molecular genetics and recombinant DNA techniques (Sambrook et al. 1989; Guthrie and Fink 1991). Cells were grown at 30° in standard rich medium (YPD), minimal medium supplemented with casamino acids (CAS) (to select for URA3-containing plasmids), or synthetic media (SC) lacking defined supplements to select for diploids and/or plasmids as appropriate. All cells were isogenic to W303-1A except the MATa (JRY19) and MATα cells (CFY617) used for mating lawns.
MATα sir1Δ HMR-SSa (CFY1649) cells harboring a 2μ plasmid that contained FKH1 (pCF480) were transformed with a mTn3-lacZ/LEU2 library that had been digested with NotI as described previously (Burns et al. 1994). Individual Leu+ colonies were picked and transferred to 96-well plates (U-bottom assay plates, Falcon) in a total volume of 100 μl of liquid CAS media, covered (Airpore tape sheets, Qiagen) and grown with agitation at 30° for three to four doublings (∼8 hr). Cells were precipitated by brief centrifugation at 3000 rpm and the cell pellets were resuspended in 50 μl of CAS. Five microliters of each cell suspension were plated on solid agar CAS medium (in rectangular plates, OmniTray, Nunc) using a multichannel pipetteman. An excess of MATa tester cells (CFY616) was added (15 μl of log-phase cells concentrated to 2 × 108 cells/ml) to the remaining cells in suspension. The cells were mixed, and 10 μl of this mixture was spotted onto diploid selective SC solid agar medium in rectangular plates. Cells were incubated at 30° for 2 days and then scored for mating and growth efficiencies. A total of 6006 Leu+ transformants, representing ∼2000 transformants each from three of the mTn3 pools (Burns et al. 1994), were screened in this manner. Isolates that grew efficiently on CAS but not on diploid selective media were subjected to further analysis.
A total of 209 nonmating Leu+ mutants were transferred from CAS agar to solid agar SC medium containing 5-fluoroorotic acid (5-FOA) to select for cells that lost the URA3-containing FKH1 plasmid. 5-FOA-resistant cells were subsequently transformed with URA3-containing CEN SIR1 (pCF99) or 2μ FKH1 (pCF480) using standard methods. Ura+ transformants were tested for mating efficiency. Only those isolates that were mating competent in the presence of SIR1 and mating defective in the presence of 2μ FKH1 were analyzed further. Fifty-three mutants met these criteria and were crossed to the parental strain to test for 2:2 segregation of LEU2. To determine that the LEU2 gene disruption was the cause of the desired mating phenotype, several MATα sir1Δ HMR-SSa Leu+ isolates from the cross were transformed with 2μ FKH1 (pCF480) and then retested for their ability to mate with MATa cells. If each of the the transformants tested failed to mate, then we concluded that the LEU2 insertion and desired phenotype were cosegregating.
Three Leu+ mutants met the above criteria, and the site of mTn3 insertion was amplified using vectorette PCR (C. Friddle, http://genomics.princeton.edu/botstein/protocols/vectorette.html) and identified by sequencing. Briefly, yeast total DNA isolated from mutant strains was digested with AluI and ligated to annealed anchor bubbles. The ligation reaction was used as the template for PCR amplification of the mTn3-yeast genomic DNA junction. Products of this reaction were sequenced using the M13/pUC sequencing primer (−40, New England Biolabs) that anneals to proximal lacZ sequences. The sequence obtained was compared to the S. cerevisiae sequence using BLASTn search on the Saccharomyces Genome Database (http://www.yeastgenome.org) to identify the genomic site of insertion.
Semiquantitative mating assay:
The MATα or MATa cells examined were grown in log phase for at least 24 hr before analysis. The most concentrated samples analyzed were 5 × 106 cells/ml in a total volume of 50 μl. Ten-fold serial dilutions of this concentration were generated in 50-μl final volumes. Five microliters of each dilution were analyzed per drop on either YPD or CAS solid agar medium, as appropriate, to determine cell plating efficiency. To the remainder of the cells, excess cells of the opposite mating type, either MATa (JRY19) or MATα (CFY617), were added (15 μl of log-phase cells concentrated to 2 × 108 cells/ml). Ten microliters of this mixture were plated to synthetic solid agar medium appropriately supplemented to select for diploids (and the retention of a URA3 plasmid, if appropriate).
Quantitative mating assay:
Cells were grown in log phase for at least 24 hr in YPD or CAS to select for plasmids when necessary. Cells were then diluted to obtain ∼100 colonies/plate and plated on the same medium to determine the total number of cells. Appropriate dilutions for testing mating were mixed with MATa cells (JRY19) and plated onto minimal medium supplemented appropriately for diploid selection to quantify the number of mating-competent cells. Mating efficiency was defined as the number of cells that mated per total number of cells.
Antibodies and protein immunoblots:
Anti-Fkh1 and anti-Orc2 polyclonal antibodies were described previously (Bose et al. 2004; Casey et al. 2008) . Anti-Ku antibodies were a gift from the laboratories of A. Tomkinson (Baltimore) and S. E. Lee (San Antonio, TX). Anti-Sir4 antibodies were a gift from the laboratory of D. Moazed (Boston).
To compare levels of Fkh1 expressed from a chromosomal copy of FKH1 and from a 2μ plasmid (FKH1hc) quantitatively (Figure 2), crude yeast extracts were prepared as described previously (Gardner et al. 1999) from wild-type cells transformed with either empty vector (pRS426) or a 2μ plasmid expressing FKH1 (FKH1hc; pCF480) and from fkh1Δ cells transformed with vector (pRS426). Ten microliters of crude cell extracts were analyzed per lane by SDS–PAGE and standard protein blotting methods. The blot was prestained with Poncea S prior to incubation with antibody to ensure that protein transfer to the blot was equivalent for all lanes. The blot was simultaneously incubated with the primary antibodies, anti-Fkh1 and anti-Orc2. Rabbit IgG was visualized using an HRP-conjugated secondary antibody and SuperSignal West Chemiluminescent HRP detection reagent (Pierce).
Reverse transcriptase–PCR analysis of a1 mRNA:
To determine the levels of a1, mRNA transcribed from HMRa total RNA was prepared from MATα cells grown to an OD of 0.6–0.8 using standard hot phenol extraction methods. Total RNA concentration was determined by spectroscopic analysis. For reverse transcription (RT), 5 μg of total RNA was used in a 20-μl reaction mixture using oligo(dT)12–18 primers (Invitrogen) and Superscript III (Invitrogen) following the manufacturer's protocol. cDNA was purified from each RT mixture using a QIAquick PCR purification kit (Qiagen) and eluted in 50 μl TE. Twofold serial dilutions were used in 25-cycle PCR reactions as follows: for measuring ACT1 expression, 1 and 2 μl of 1:10 diluted cDNA were used with ACT1-specific primers (caagaaatgcaaaccgctgc and ggtcaataccggcagattcc) or, for a1 mRNA, with a1-specific primers (ggcggaaaacataaacagaactctg and ccgactatgctattttaatcattgaaaacg). The PCR products were resolved on a 1% agarose gel stained with GelRed dye (Biotium). The bands were quantified using video densitometry analysis and Labworks analysis software (UVP).
ChIPs were performed as described (Strahl-Bolsinger et al. 1997) using anti-Ku or anti-Sir4 antibodies. IP and total DNAs were purified using a QIAquick PCR purification kit (Qiagen) and eluted in 50 μl TE. For PCR, twofold dilutions were used: 1 and 2 μl of undiluted IP DNA and 1 and 2 μl of 1:50 diluted total DNA (which represented 3.5% of the starting material). DNA was subjected to 26 or 27 cycles of PCR using primers specific for HMR-E (forward: ggtagagttccttgttgaacgtgataaccc and reverse: gctttggggacatcatgtataacagcg), X/Ya (Rusche et al. 2002), HMR-I (Rusche et al. 2002), HML-E [forward (Rusche et al. 2002) and reverse: gttagatttggcccccgaaatcga] HML-I [forward (Rusche et al. 2002) and reverse: gaacgtacatagtgtgcccagctt], Tel VI-R 0.5, and Tel VI-R 7.5 (Sharp et al. 2003). PCR products were separated on a 1% agarose gel containing GelRed dye (Biotium) and band intensities were quantified using video densitometry analysis and Labworks analysis software (UVP).
A yku80∷LEU2 insertion isolated in a screen designed to examine SIR1-independent pathways required for silencing HMR-SSa:
Although SIR1 has an established role in silencing the HM loci, a large body of evidence indicates that there exist SIR1-independent means for recruiting the Sir proteins to HMR and HML and establishing silencing (Ivy et al. 1986; Pillus and Rine 1989; Axelrod and Rine 1991; Laman et al. 1995; Reifsnyder et al. 1996; Ehrenhofer-Murray et al. 1999; Hollenhorst et al. 2000). In a recent study, we exploited the synthetic HMR-SSa silencer that absolutely depends on SIR1 for robust silencing (McNally and Rine 1991) to identify a SIR1-independent pathway that responds to both the forkhead transcription factor Fkh1 and the S-phase cyclin Clb5 (Casey et al. 2008). This work led to a model in which FKH1 and CLB5 have opposing regulatory roles on a common target(s) that in turn enhances silent chromatin assembly (Casey et al. 2008; Figure 1). This study also provides evidence that distinct SIR1-independent molecular interactions at silencers can lead to efficient recruitment of the Sir2–4 protein complex to HMR-SSa. To define such interactions, a genetic screen was performed to isolate genes necessary for FKH1's ability to act as a high-copy suppressor of an HMR-SSa silencing defect caused by a sir1Δ mutation (Hollenhorst et al. 2000). In theory, this screen could identify genes belonging to one of at least two SIR1-independent pathways that contributed to silencing HMR-SSa: (1) one category of genes could function as part of the SIR1-independent FKH1/CLB5 pathway previously described (Casey et al. 2008) and (2) a second category of genes could function as part of a SIR1-independent pathway that was distinct from the FKH1/CLB5 pathway but that functioned additively with it. Genes in this second category might be equivalent functionally to previously identified eso (enhancer of sir1) genes that caused synthetic silencing defects at HMLα in combination with a sir1Δ mutation (Reifsnyder et al. 1996; Stone et al. 2000).
MATα HMR-SSa sir1Δ cells, normally unable to mate because SIR1 is required to silence HMR-SSa, regain the ability to silence HMR-SSa and mate when transformed with a 2μ plasmid expressing FKH1 (FKH1hc) (Hollenhorst et al. 2000; Casey et al. 2008). To better understand how this SIR1-independent silencing was achieved, we screened for genes that prevented FKH1hc-dependent silencing. Yeast cells with the genotype MATα HMR-SSa sir1Δ and containing FKH1hc were transformed with a yeast genomic library that had been randomly mutagenized mTn3-lacZ/LEU2 (Burns et al. 1994). Individual Leu+ colonies were transferred to a 96-well plate and grown under conditions that selected for the 2μ FKH1 plasmid (Figure 2). An aliquot of cells in each well was transferred onto a growth control agar plate (Figure 2, plate 1, selected for plasmid-containing haploid cells). An excess of MATa cells was then mixed thoroughly with the remaining cells, and an aliquot of this mixture was transferred to a second plate that selected for diploids (Figure 2, plate 2, diploid selective media). Thus the MATα HMR-SSa sir1Δ FKH1hc cells had to mate efficiently with the MATa cells to grow on plate 2.
The goal was to identify genes that contributed to silencing HMR-SSa through a SIR1-independent pathway. Therefore, in a secondary screen the nonmating mutants were tested for their ability to mate in the presence of a SIR1 plasmid. Mutants that were silencing defective in the presence of FKH1hc but silencing competent in the presence of SIR1 were pursued further. One mutant contained an insertion in the YMR106C open reading frame that encodes yKu80, providing evidence that YKU80 contributed to a SIR1-independent pathway for silencing HMR-SSa.
Ku and FKH1hc contributed independently to silence HMR-SSa in sir1Δ cells:
Although previous studies provide evidence that the Ku complex helps maintain HMR's association with the nuclear periphery (Gartenberg et al. 2004), no functional role for Ku in silencing of HMR has been documented. However, isolation of a LEU2 insertion mutant in YKU80 provided evidence that Ku had a role in silencing HMRa, at least in the context of HMR-SSa, sir1Δ, and/or FKH1hc-dependent silencing. To test whether an independent and complete deletion of YKU80 would recapitulate the phenotype of the LEU2 insertion mutant isolated in the screen, MATa cells containing a complete unmarked deletion of YKU80 (yku80Δ) were crossed to MATα HMR-SSa sir1Δ cells, and silencing of several independent YKU80 and yku80Δ MATα HMR-SSa sir1Δ segregants from this cross were analyzed by a semiquantitative mating assay (Figure 3A). These experiments recapitulated the basic results from the screen (Figure 2) and revealed that yKu80 was required to detect robust FKH1hc-dependent silencing but not SIR1-dependent silencing.
One possible explanation for these data was that the Fkh1 protein was not sufficiently expressed in yku80Δ cells to achieve FKH1hc-dependent silencing. Therefore the levels of Fkh1 were determined in the YKU80 and yku80Δ cells by protein immunoblot analyses using Fkh1 antibodies (Figure 3B). An antibody recognizing Orc2, a subunit of the origin recognition complex, was also used to control for the amount of protein in each lane. Assuming that the levels of Fkh1 protein required for silencing do not vary with the YKU80 genotype, Fkh1 was expressed sufficiently in yku80Δ cells to achieve silencing of HMR-SSa.
To extend the analyses, we also tested the effect of yku70Δ on silencing HMR-SSa and quantified the silencing defects caused by both yku70Δ and yku80Δ by performing quantitative mating assays to measure mating efficiency. Quantitative mating assays determine the number of mating-competent cells within a population of viable cells, and hence in MATα cells can serve to determine the number of cells that have silenced HMRa. The ratio of mating-competent cells to total number of cells assessed was indicated on the y-axis, presented in log scale for each yeast genotype indicated. Consistent with the data presented in Figure 3A, yku80Δ substantially reduced silencing of HMR-SSa in MATα HMR-SSa sir1Δ cells expressing FKH1hc, but had no effect on the same cells expressing SIR1 (Figure 3C). In addition, yku70Δ produced the same phenotype as yku80Δ, as expected since the Ku complex functions as a heterodimer of Ku70 and Ku80. These data were fully consistent with the screen (Figure 2) and revealed that Ku functioned to silence HMR-SSa in sir1Δ cells.
If Ku was the sole or primary downstream target of the FKH1hc-mediated silencing pathway, then FKH1hc should fail to enhance silencing of MATα HMR-SSa sir1Δ yku80Δ cells. However, the quantitative data revealed clearly that Ku had a role in silencing HMR-SSa in the absence of FKH1hc and, conversely, that FKH1hc could improve silencing of HMR-SSa in the absence of Ku. FKH1hc improved silencing of MATα HMR-SSa sir1Δ yku80Δ cells by almost 100-fold (Figure 3C; compare FKH1hc yku80Δ or yku70Δ with vector yku80Δ or yku70Δ). This magnitude of enhancement was similar to that observed in comparable cells that were wild type for Ku (Figure 3C; compare FKH1hc and vector in cells that were wild type for Ku). Thus Ku contributed to silencing HMR-SSa at least in part through a SIR1-independent pathway that worked additively with FKH1hc.
YKU80 and CLB5 also made independent contributions to silencing of HMR-SSa in sir1Δ cells:
The quantitative mating data presented above (Figure 3C) provided evidence for FKH1 and Ku contributing independently to silencing HMR-SSa in sir1Δ cells. Like FKH1hc, clb5Δ is also established as an efficient suppressor of sir1Δ silencing defects (Laman et al. 1995), and our recent study provides evidence that FKH1 and CLB5 modulate silencing through a common pathway (Casey et al. 2008). A prediction of this model is that CLB5 and YKU80 should also contribute independently to modulate HMR-SSa silencing. Therefore we generated an isogenic set of MATα HMR-SSa sir1Δ cells that varied only in terms of their YKU80 or CLB5 genotypes and measured silencing of HMR-SSa by semiquantitative mating assays (Figure 4A). These experiments revealed that yku80Δ reduced the level of SIR1 bypass silencing that could be detected in clb5Δ cells, which was similar to what was observed for FKH1hc-mediated silencing (Figure 3A). Quantitative mating data (Figure 4B) revealed that CLB5 and YKU80 made independent contributions to silencing HMR-SSa in sir1Δ cells. Specifically, clb5Δ enhanced silencing in MATα sir1Δ cells regardless of whether these cells were wild type, YKU80, or yku80Δ. Thus YKU80 contributed substantially to silencing HMR-SSa in both CLB5 and clb5Δ cells. These data supported the conclusion that effective SIR1 bypass silencing of HMR-SSa required both the FKH1/CLB5 and the Ku pathways.
A surprising feature of FKH1/CLB5-dependent silencing is that it can use an ordinary replication origin as a silencer (Casey et al. 2008). Substitution of the HMR-E silencer with the early to mid-S-phase firing origin ARS1 (HMRΔE∷ARS1) reduces SIR1-dependent silencing, but has no effect on FKH1/CLB5-dependent silencing. Since the data presented above provided evidence that Ku functioned additively with the FKH1/CLB5 pathway to contribute to silencing at HMR-SSa in sir1Δ cells, we asked whether it also supported silencing of HMRa when this locus was under the control of the HMRΔE∷ARS1 silencer (Figure 4, A and B). As was observed for HMR-SSa, yKu80 contributed to silencing HMRΔE∷ARS1 in both CLB5 and clb5Δ cells. Thus silencing mediated by the unusual HMRΔE∷ARS1 silencer, although largely independent of SIR1 (Casey et al. 2008), relied on yKu80.
Ku functioned in silencing natural HMR and HML:
The above data indicated that Ku functioned in silencing at HMR-SSa and HMRΔE∷ARS1. These loci contain engineered versions of HMR-E designed to sensitize HMRa silencing for genetic studies. To ask whether Ku had a role in silencing either natural HMR or HML, we measured the effects of yku80Δ on HMRa and HMLα silencing in MATα and MATa strains, respectively. Consistent with previous findings, yku80Δ had no effect on silencing either locus in SIR1 cells (Figure 5A). Similarly, sir1Δ cells did not exhibit a substantial silencing defect (Ivy et al. 1986; Pillus and Rine 1989). However, yeast cells containing both mutations in SIR1 and YKU80 (sir1Δ yku80Δ) showed dramatic silencing defects at HMR and HML. As an independent measure for HMRa silencing, a1 mRNA levels were measured by reverse transcriptase PCR (RT–PCR) in MATα sir1Δ cells that varied only in terms of their YKU80 or SIR3 genotype (Figure 5B). These experiments revealed that the ratio of a1 mRNA to the ACT1 mRNA control increased ∼2.3-fold in yku80Δ cells relative to YKU80 cells (Figure 5B). This elevated a1/ACT1 ratio was similar to that observed for sir3Δ cells (Figure 5B). Thus, YKU80 became critical for silencing natural HMR in sir1Δ cells.
ESC1 and YKU80 functioned differently in HM silencing:
Ku binds telomeres and functions directly to link telomeres to the nuclear periphery and to silence telomeric regions. The inner nuclear membrane protein Esc1 also contributes to telomeric tethering and silencing, although to a lesser degree than Ku (Taddei et al. 2004). HMR is also tethered to the nuclear periphery (Laroche et al. 2000; Gartenberg et al. 2004), but in contrast to telomeres, this tethering requires both YKU and ESC1 (Gartenberg et al. 2004); release of HMR from the nuclear periphery into the nucleoplasm occurs only when both YKU70 and ESC1 are deleted (yku70Δ esc1Δ). Thus Ku and Esc1 have redundant roles in HMR tethering. Since our data revealed that Ku had a substantial role in silencing the HM loci in sir1Δ cells and that Ku and Esc1 shared roles in HMR tethering, we predicted that Esc1 might also have a role in HM silencing that would be revealed in sir1Δ cells (Figure 6).
However, genetic experiments revealed that ESC1 did not have a positive role in silencing the HM loci in sir1Δ cells (Figure 6). We first examined silencing of the sensitized HMR-SSa locus where a role for YKU80 was initially revealed (Figure 2). In contrast to cells with a yku80Δ (Figure 4A), cells with an esc1Δ exhibited no defect in silencing HMR-SSa when combined with a sir1Δ mutation (Figure 6A) and, in fact, exhibited enhanced silencing. The explanation for such enhancement is unknown, but perhaps in esc1Δ cells the Sir2–4 proteins were released from one region of the nuclear periphery into another region that was more relevant to HMR silencing. That is, perhaps ESC1 controls Sir2–4 pools within the nuclear periphery itself. To test whether ESC1 might have a role in FKH1/CLB5-dependent silencing, the effect of an esc1Δ on silencing of MATα HMR-SSa sir1Δ clb5Δ cells was measured. Again, esc1Δ did not reduce silencing of HMR-SSa, indicating that unlike YKU80, ESC1 did not function positively to silence HMR-SSa.
It was possible that ESC1-dependent elements had been removed during the construction of HMR-SSa such that a positive role for ESC1 in silencing the HM loci in sir1Δ cells went undetected. Therefore, silencing of natural HMR and HML was examined in sir1Δ cells. Unlike a yku80Δ (Figure 5), an esc1Δ did not reduce silencing of either HMRa or HMLα in sir1Δ cells (Figure 6A). Thus, although YKU80 and ESC1 have roles in tethering HMRa and telomeres to the nuclear periphery, YKU80 possessed a unique function in silencing the HM loci that was not shared with ESC1.
As stated above, Ku and ESC1 have overlapping roles in tethering HMR and telomeres to the nuclear periphery. However, although HMR, telomeres, and their associated Sir proteins are completely released from the nuclear periphery into the nucleoplasm in yku70Δ esc1Δ cells, HMR silencing remains unperturbed (Gartenberg et al. 2004). Thus tethering of HMR is not required for silencing, at least in SIR1 cells. One explanation for HMR silencing remaining intact in yku70Δ esc1Δ cells is that any silencing defect caused by redistribution of HMR to the nucleoplasm was compensated for by the higher concentration of Sir proteins that reside in the nucleoplasm in yku70Δ esc1Δ cells.
Given the tethering roles of YKU80 and ESC1 (Gartenberg et al. 2004), it was possible that a yku80Δ caused a change in Sir2–4 protein concentration at the nuclear periphery relative to the position of HMR that was insufficient for HMR silencing in sir1Δ cells. To test whether Sir2–4 protein distribution changes could explain the roles of ESC1 and YKU80 in HM silencing in sir1Δ cells, we compared HMRa silencing in sir1Δ cells containing an esc1Δ, yku80Δ or both mutations (esc1Δ yku80Δ) (Figure 6B). MATα sir1Δ cells that lacked both ESC1 and YKU80 (esc1Δ yku80Δ) showed a slight enhancement of silencing compared to cells that lacked only YKU80 (Figure 6B; compare esc1Δ yku80Δ to ESC1 yku80Δ). This enhancement was likely attributable to the release of HMR to the nucleoplasm, where higher levels of Sir2–4 proteins now reside, due to release of Sir2–4 proteins from telomeres. However, these esc1Δ yku80Δ cells remained substantially defective in silencing HMR compared to MATα sir1Δ ESC1 YKU80 cells (Figure 6B; compare esc1Δ yku80Δ to either ESC1 YKU80 or esc1Δ YKU80). That is, the double mutation did not rescue silencing of HMRa in MATα sir1Δ cells to “wild type” (i.e., ESC1 YKU80) levels. Therefore these epistatic analyses indicated that Sir redistribution relative to HMR positioning was unlikely the sole reason that Ku was required for HMR silencing in sir1Δ cells. Rather, these analyses were consistent with the idea that YKU80 had a distinct positive role in HMR silencing not shared with ESC1.
SIR4 modulation of HM silencing in sir1Δ cells was influenced by YKU genotype:
A direct interaction between Sir4 and Ku contributes to telomeric silencing (Roy et al. 2004). On the basis of our data, we hypothesized that a Sir4–Ku interaction might be important for HM silencing as well. Overexpression of Sir4 can stimulate ectopic silencing at nontelomeric sites (Maillet et al. 1996), including HMR (Andrulis et al. 1998), and rescue silencing defects at telomeres. In addition, high-copy expression of SIR4 can rescue the HMR silencing defect in MATα HMR-SSa sir1Δ cells (Hollenhorst et al. 2000 and Figure 7A). Therefore, we overexpressed Sir4 on a 2μ plasmid (SIR4hc) and measured HMR silencing in MATα sir1Δ cells that contained HMR-SSa, HMRΔE∷ARS1, or natural HMRa (Figure 7A). SIR4hc was able to rescue silencing in sir1Δ cells containing either HMR-SSa or HMRΔE∷ARS1 silencers. Since silencing of natural HMRa remained so efficient in sir1Δ cells, no obvious effect of SIR4hc was observed. However, in yku80Δ cells, SIR4hc-dependent silencing with any of these versions of HMRa, including natural HMRa, was reduced. In particular, SIR4hc silencing of HMRΔE∷ARS1 was quite robust in sir1Δ YKU80 cells but undetectable in sir1Δ yku80Δ cells.
To further address the idea that Ku shared functions with Sir1 to affect Sir4 at the HM loci, we examined the effect of the yku80-4 allele on silencing of HMRa and HMLα in sir1Δ cells (Figure 7B). The yku80-4 allele (encoding yKu80P437L) impairs the interaction between yKu80 and Sir4 (Taddei et al. 2004) but does not abolish telomeric silencing completely or the ability of yKu80 to mediate chromosomal tethering. If the Ku–Sir4 interaction was important for HM silencing in the absence of SIR1, then the yku80-4 allele should produce a similar phenotype as yku80Δ. We tested this idea by comparing the silencing of HMRa and HMLα in sir1Δ cells that were YKU80, yku80Δ, or yku80-4 (Figure 7B). These experiments revealed that yku80-4 reduced silencing of HMLα in sir1Δ cells substantially (Figure 7B). The effect of this mutation on silencing of HMRa in sir1Δ cells was more modest, suggesting that natural HMRa has both SIR1- and Ku-independent means of supporting SIR4 function in silencing that are less effectively used (or absent) at HMLα. Nevertheless, these data were consistent with Ku playing a redundant role with Sir1 in enhancing Sir4 function at the HM loci.
YKU80 contributed to Sir4's association with HML and HMR in sir1Δ cells:
On the basis of the available evidence, the simplest explanation for Ku's role in HM silencing is that it contributed the physical recruitment of Sir4 to each of the HM loci. Therefore, we compared Sir4 binding to the HML and HMR loci in wild type, yku80Δ, sir1Δ, and sir1Δ yku80Δ cells by ChIPs using anti-Sir4 polyclonal antibodies (Figure 8). As reported previously (Rusche et al. 2002), Sir4's association with HML and HMR is reduced but not abolished by a sir1Δ (Figure 8; compare SIR1 YKU80 to sir1Δ YKU80). Although a yku80Δ had no effect on Sir4's association with either HM locus in SIR1 cells (Figure 8; compare SIR1 YKU80 to SIR1 yku80Δ), this mutation abolished the level of Sir4 associated with either HM locus in sir1Δ cells (Figure 8; compare sir1Δ YKU80 to sir1Δ yku80Δ). Therefore Ku, like Sir1, contributed to Sir4's physical association with the HM loci in vivo.
Ku physically associated with HML and HMR:
These genetic data were consistent with Ku sharing an overlapping function with Sir1 at the HM loci. Sir1's role at the HM loci is direct. However, although Ku has been detected at telomeres and at HMLα during mating-type switching, there exists no evidence that Ku binds to HMLα or HMRa directly as part of normal silencing. Therefore, we tested whether Ku bound to the silencers of these loci by ChIP. A ChIP experiment using anti-yKu polyclonal antibodies provided evidence that Ku physically associated with the HML and HMR silencers, although the association with the HML silencers was clearly more robust (Figure 9). In these experiments, an internal site 7.5 kb from telomere VI-R (Tel 7.5) served as a negative control and a region 0.5 kb from the end of chromosome VI-R (Tel 0.5) served as a positive control (Sharp et al. 2003). We found significant enrichment of HML-I, HML-E, and HMR-I, providing evidence that the yeast Ku complex was physically associated with the HM loci (Figure 9).
The yeast Ku complex's involvement in telomeric silencing and in localizing the Sir2–4 proteins and associated telomeres to the nuclear periphery is well established (reviewed in Fisher and Zakian 2005). In addition, experimental data support a role for Ku in controlling the localization of the HM loci to the nuclear periphery (Gartenberg et al. 2004). Given these data, it is somewhat perplexing that an unambiguous functional role for Ku in silencing either HMR or HML has not been established. This study revealed that the explanation for an inability to detect a role for Ku in silencing of the HM loci was that Ku's function was completely masked by the specialized role of SIR1 at HMR and HML. Yeast cells harboring null mutations in either SIR1 or YKU80 showed little or no silencing defects at either HM locus, as expected (Ivy et al. 1986; Pillus and Rine 1989; Laroche et al. 2000; Gartenberg et al. 2004). However, cells with mutations in both genes (i.e., sir1Δ yku80Δ cells) were completely defective for silencing the HM loci, and Sir4-directed ChIP experiments revealed that this defect was likely due to an inability of the HM loci to bind Sir4. Thus Ku was required for the residual but substantial Sir4 association with the HM loci in sir1Δ cells. Genetic analysis suggested that Ku's HM silencing function was distinct from the overlapping role that it plays with the protein Esc1 in physically tethering the HM loci, telomeres, and associated Sir2–4 proteins to the nuclear periphery. Finally, studies of the effect of SIR4 dosage and Ku localization by ChIP provided evidence that Ku physically associates with the HM loci and helps to promote Sir4 function. Thus Sir1 and Ku contributed to independent and redundant pathways in the assembly of silent chromatin at the cryptic mating-type loci through recruitment of Sir4. We conclude that Ku, like Sir1, should be viewed as an integral and direct stabilizing force for silent chromatin at the HM loci.
Ku in HMR and HML silencing:
An important mechanistic role for Sir1 in silencing the HM loci is to stabilize association of Sir4 with silencers (reviewed in Fox and McConnell 2005). However, some Sir4 remains at natural HMR and HML even in the absence of SIR1 (Rusche et al. 2002), providing evidence that Sir4 recruitment represents a key Sir1-independent step required to nucleate silencing at HMR and HML. These observations are consistent with Sir4's pivotal role in nucleating telomeric silencing (Hoppe et al. 2002; Luo et al. 2002) that occurs completely independently of Sir1 (Aparicio et al. 1991). Taken together with the data reported here, a simple model for Ku's role at the HM loci is that it functions directly with Sir1 to recruit and/or stabilize Sir4.
Sir1 promotes Sir4 binding to the HM silencers by binding to silencer-bound ORC and directly interacting with Sir4 (Bose et al. 2004). Interestingly, this function of Sir1, as measured by studies of the HMR-E silencer, is aided by this silencer's Rap1-binding site (Casey et al. 2008), suggesting that stable binding of Sir4 to the HMR-E silencer requires that Sir4 interact with both Sir1 and Rap1 (Moretti et al. 1994; Cockell et al. 1995; Moretti and Shore 2001; Luo et al. 2002). Previous studies of Ku–Sir4 interactions in telomeric silencing combined with the SIR4 dosage and Sir4-directed ChIP data described in this study raise the possibility that Ku provides yet another direct molecular force that stabilizes Sir4 at the HM loci. Two possible and not exclusive mechanisms might explain how Ku performs this function. First, Ku could contribute to Sir4's stable association with silencers and subsequent nucleation of silent chromatin. This mechanism agrees with the idea that Ku's role in silencing is mechanistically redundant with Sir1's role: both proteins may help stabilize Sir4 at silencers, yet remain separated from the Sir2–4 chromatin that forms over the HM loci. A second alternative is that Ku binds with Sir4 throughout silent chromatin to stabilize Sir4 association with the HM loci. This mechanism is distinct from how Sir1 is thought to work with Sir4 in HM silencing. Additional and extensive ChIP analyses might help sort through these models. Regardless, both models describe a direct role for Ku in silencing the HM loci that extends beyond the proposed “indirect” role for Ku in HM silencing where Ku modulates HM silencing because it controls the relative localization of the HM loci and pools of Sir proteins in the nucleus (Gartenberg et al. 2004). In addition, the data reported here provided evidence that a Sir4–Rap1 interaction was not a key factor in Ku-dependent silencing at HMR. Indeed, the HMRΔE∷ARS1 silencer should create an HMR locus that lacks any Rap1-binding sites, yet is proficient in Ku-dependent silencing. In addition, Sir4–Rap1 interactions were insufficient for supporting robust silencing at the HM loci in the absence of SIR1 and Ku, even when SIR4 was overexpressed. This observation underscores a difference between telomeric and HM loci silencing since overexpression of SIR4 can rescue telomeric silencing in yku80Δ mutants (Moretti et al. 1994; Cockell et al. 1995; Moretti and Shore 2001; Luo et al. 2002). Perhaps the multiple Rap1 proteins present at telomeres provide a more robust platform for Sir4 than the few Rap1-binding sites present within the HM loci.
The Ku complex at internal chromosomal regions:
Although extensive studies of Ku in silencing have been performed, there is no report that Ku binds to either HML or HMR as part of normal silencing. In fact, a Myc-tagged Ku was not detected at HMLα except during mating-type switching (Ruan et al. 2005). Additionally, during the course of our experiments, we could not detect HA-tagged Ku proteins at the HM loci with anti-HA-directed ChIPs even though these ChIPs clearly demonstrated Ku association with the positive control telomere locus (E. E. Patterson, unpublished results). However, clearly a ChIP with a polyclonal antibody raised against the yeast Ku complex (Zhang et al. 2007) effectively detected the Ku complex at HML and to a lesser extent at HMR. We suspect that the disparity between ChIPs with antibodies against the HA epitope and ChIPs with a polyclonal antibody raised against Ku was due to differences between how Ku associates with telomeres vs. the HM loci. Ku binds to the ends of DNA with a high affinity and thus its association with telomeres likely involves extensive direct Ku–DNA contacts. In contrast, Ku's association with internal chromosomal regions such as the HM loci probably relies more on protein–protein interactions; direct Ku–DNA contacts may be limited or even nonexistent. Perhaps the polyclonal antibody was more effective at immunoprecipitating the small amount of Ku effectively crosslinked through protein–protein bridges to HM DNA. Alternatively, in the context of the internal HM loci, the HA epitopes may be masked within a protein complex and not available for interactions with the anti-HA antibody after crosslinking.
There is precedence for protein–protein interactions contributing to Ku's localization to an internal chromosomal region. In particular, Ku binds to the recombinational enhancer and this binding depends on Mcm1 (Ruan et al. 2005). Additionally and as stated above, this same study (Ruan et al. 2005) reported that Ku could be detected at HML but only during mating-type switching. The difference between these results and those reported here might be that the former study performed ChIPs with an epitope-tagged version of Ku. However, it is also possible that during mating-type recombination the conformation of Ku changes to expose the epitope tag more effectively and/or enhance Ku's contact with DNA, thus allowing Myc-tagged Ku's association with HML to be detected.
The key question arising from the observations reported here is, what interaction(s) recruits Ku to the HM loci? Obviously, a strong candidate is Sir4 itself. Perhaps Sir4 and Ku recognize some element(s) of silencers as a protein complex. But, interestingly, neither Ku nor Sir4 appear to be relying on Rap1 for their function at the HM loci since Ku-dependent silencing was observed with the HMRΔE∷ARS1 silencer that lacks a Rap1-binding site. Instead, Ku and Sir4 may be relying on some other silencer-recognition mechanism. In this regard, it is intriguing that the effects of yeast Ku on ORC binding in vitro have been reported (Shakibai et al. 1996) and that in mammalian cells Ku associates with replication origins and has been found in a complex with known replication proteins, including ORC2 (Novac et al. 2001; Matheos et al. 2002, 2003). In addition, recent two-hybrid analyses of Orc subunits report an interaction between Orc2 and Sir4 (Matsuda et al. 2007). Thus an intriguing possibility is that Ku and/or Sir4 may interact together with some feature of ORC.
Cell-cycle regulators in SIR1-independent silencing at HMR:
A major goal of the screen presented in this report was to identify genes that could help explain how the cell-cycle regulators Fkh1 and Clb5 contribute to silencing of HMR (Casey et al. 2008). In particular, we propose that HMR silencing in sir1Δ cells allows us to detect an effect on chromosomal architecture and/or the nuclear distributions of Sir2–4 proteins that is highly regulated and dynamic during the cell cycle and that is likely to have broad significance for chromosome biology. As described in this report, this screen revealed a role for Ku in silencing the HM loci. Ku's role in chromosomal dynamics during the cell cycle (Cosgrove et al. 2002; Taddei et al. 2004) would be an attractive candidate for explaining how cell-cycle regulators could have such a substantial impact on silencing. And although the genetic analyses provided evidence that Ku's role in SIR1-independent silencing is unlikely to be the primary target regulated by Fkh1 and Clb5, its unanticipated identification in this screen is a reminder that there are many aspects of HM silencing and its regulation that remain poorly understood.
We are especially grateful to Melissa Hefferin, Alan Tomkinson (Greenbaum Cancer Center, Baltimore), and Sang Eun Lee (University of Texas Health Science Center, San Antonio, TX) for sharing anti-Ku polyclonal antibody; Danesh Moazed (Harvard Medical School, Boston) for sharing Sir4 antibodies; and Rodney Rothstein (Columbia University, New York) for sharing unpublished yeast strains. We are also grateful to Marc Gartenberg (Robert Wood Johnson Medical School, Piscataway, NJ) and Erika Shor for comments on the manuscript and other relevant scientific interactions regarding this project. We thank Philipp Müller for help on Figure 3 and other members of the Fox lab for helpful discussions. E.E.P. was supported in part by a National Institutes of Health (NIH) Predoctoral Training Program in Genetics to the Laboratory of Genetics (5 T32 GM07133) and by an American Heart Association predoctoral fellowship (0615552Z). This work was supported primarily by a grant from the American Cancer Society (RSG-02-164-02-GMC to C.A.F.) and supplemented by a grant from the NIH (RO1 GM56890 to C.A.F.).
Note added in proof: A recent independent study also demonstrates a role for Ku in silencing of the HM mating-type loci [C. L. Vandre, R. T. Kamakaka and D. H. Rivier, 2008, The DNA end-binding protein Ku regulates silencing at the internal HML and HMR loci in Saccharomyces cerevisiae. Genetics (in press)].
Communicating editor: F. Winston
- Received May 19, 2008.
- Accepted July 9, 2008.
- Copyright © 2008 by the Genetics Society of America