Separation-of-function Mutants of Yeast Ku80 Reveal a Yku80p-Sir4p Interaction Involved in Telomeric Silencing*

The Saccharomyces cerevisiae Ku heterodimer comprising Yku70p and Yku80p is involved in telomere maintenance and DNA repair by the pathway of non-homologous end joining. It is also a key regulator of transcriptional silencing of genes placed in close proximity to telomeres. Here, we describe the identification of separation-of-function mutants of Yku80p that exhibit defects in silencing but not DNA repair and show that these mutations map to an evolutionarily conserved domain within Yku80p. Furthermore, we reveal that Yku80p interacts with the silent information regulator protein Sir4p and that this interaction is mediated by the N-terminal 200 amino acid residues of Sir4p. Notably, this interaction also requires the region of Yku80p that contains the sites of the silencing defective mutations. Finally, we show that these mutations impair the Yku80p-Sir4p interaction and recruitment of Sir3p to telomeric regions in vivo. Taken together with other data, these findings indicate that the Yku80p-Sir4p interaction plays a vital role in the assembly of telomeric heterochromatin.

The DNA of all organisms is continually damaged by the byproducts of physiological processes and by mutagens in the environment. Ionizing radiation and radiomimetic chemicals induce a variety of DNA lesions, the most cytotoxic of which is the DNA double strand break (DSB). 1 Consequently, eukaryotic cells have developed elaborate mechanisms to sense DNA DSBs and to mediate their effective repair (1). One key protein recognizing DNA DSBs is Ku, which binds to DNA ends in addition to other types of discontinuity in double-stranded DNA (2,3). Ku is a tightly associated heterodimer of ϳ70and ϳ80-kDa subunits that in humans acts together with the 470-kDa DNA-dependent protein kinase catalytic subunit (see Ref. 4) to repair DNA DSBs caused by physiological oxidation reactions, V(D)J recombination, ionizing radiation, and certain chemotherapeutic drugs (5). The Ku-dependent repair process called illegitimate recombination or non-homologous end joining (NHEJ) is a major DSB repair mechanism in mammalian cells (3). Although DSB repair by NHEJ is used less frequently in the budding yeast Saccharomyces cerevisiae, this organism possesses homologues of both Ku70 and Ku80 and these play crucial roles in NHEJ (3,6,7). Thus, yeast cells expressing a functional Ku heterodimer accurately and efficiently join cohesive ends of a transformed linearized plasmid, whereas this repair is severely impaired in cells lacking either Ku subunit (6,8).
S. cerevisiae Ku is also critically important for telomere maintenance, and disruption of either yeast Ku subunit results in the reduction of telomere repeat lengths (9,10). In addition, both YKU70 and YKU80 interact genetically with CDC13 whose product binds to the single-stranded telomeric overhang and is required for telomerase activity in vivo (11,12). Yku80p has also been found to help recruit telomerase to the telomeric end via interaction with a 48-nucleotide stem-loop component of telomerase RNA (13). Precisely how Ku is itself tethered to telomere ends is not yet clear.
Deletion of Ku also affects the transcriptional silencing of genes in close proximity to telomeres, a process called the telomeric position effect (TPE) (14,15). TPE is severely diminished in cells lacking Yku70p or Yku80p, but repression of the silent mating type loci, HMR and HML, is maintained (9,16,17). The silent information regulator (Sir) proteins, Sir2p, Sir3p, and Sir4p, are the major mediators of transcriptional silencing at both the telomeric and the silent mating-type loci (18) and are thought to bring about silencing by packaging DNA into a heterochromatin-like state (19). The Sir complex is recruited to telomeres through interactions with Rap1p (a telomere TG repeat-binding protein) and histones H3 and H4 (20). Sir4p plays a major role in the initiation of this assembly and recruits Sir2p and Sir3p. Deacetylation of histone H4 by Sir2p then promotes further recruitment of Sir4p (21)(22)(23). The cooperative action between the Sir proteins thus helps to spread silencing to more internal regions of the telomeres. Two other proteins, Rif1p and Rif2p, also bind to Rap1p, and it has been suggested that the Rif proteins compete with Sir4p for binding to Rap1p (24). Yeast Ku is thought to be recruited to telomere ends and to antagonize the effect of Rif proteins, hence facilitating the recruitment of Sir proteins (25). However, recent work has shown that Sir4p can also be recruited to telomeres independently of Ku (22). Furthermore, it has been suggested that S. cerevisiae Ku is not only bound to the most distal end of the chromosome but can also spread along the transcriptionally repressed subtelomeric regions (26).
To gain mechanistic insights into how Ku executes its varied functions, we have carried out screens to identify Ku mutants that separate its role in DSB repair from its involvement in silencing of telomere-proximal genes. Here, we describe a class of Yku80p mutants that is dysfunctional in telomeric silencing but not DNA repair and reveal that these mutants are impaired in a newly defined interaction between Yku80p and Sir4p. These results provide insights into how Ku influences TPE through the formation of telomeric heterochromatin.

MATERIALS AND METHODS
Media, Growth Conditions, and Antibodies-Yeast strains were grown using standard conditions (27). Assays for sensitivity to methyl methane sulfonate (MMS) and 5-fluoro-orotic acid (5-FOA) were done as described previously (6,25). Mouse monoclonal anti-MYC antibody and monoclonal anti-HA antibody were from Roche Applied Science.
Plasmids, Yeast Strains, and DNA Manipulations-General DNA manipulations were performed according to established methods. Plasmids used were as follows: pRS413, selectable marker HIS3; pBTM116, selectable marker TRP1; pSH18 -34, selectable marker URA3 with the ␤-galactosidase gene under the control of six LexA binding sites; pEG202, selectable marker HIS3 and expresses a LexA DNA binding domain; and pJG4 -5, selectable marker TRP1 and expresses a B-42 transcription activation domain with an SV40 nuclear localization sequence. Yeast strains used are listed in Table I. Bacterial plasmid DNA extraction was performed using a mini plasmid isolation kit (Qiagen). Genomic or plasmid DNA from yeast was isolated as outlined previously (28). Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene). Transformation of yeast strains was performed by the lithium acetate method (29). Plasmid repair assays were done as described previously (8).
PCR Mutagenesis and Gap Repair-yku80 mutants were generated by mutagenic PCR essentially as described previously (30). YKU80 was mutagenized using the primers YKu80 -2 (5Ј-ATGTCAAGTGAGTCAA-CAACTTTCATCGTG-3Ј) and YKu80 -3 (5Ј-TAATTATTGCTATTGTTT-GGACTTCCCCT-3Ј). PCR conditions were optimized to give 1-3 mutations in the final product. 10 independent PCR reactions were pooled, and the 1890-bp YKU80 band was gel-purified. Vector DNA was prepared by digesting pRS413 carrying full-length YKU80 (pYKU80FL) with SnaBI, which makes a single blunt-ended cut in the YKU80 open reading frame, 1 kb downstream of the start codon. The DNA was then dephosphorylated and gel-purified. PCR product (500 ng) was mixed with 250 ng of digested vector DNA (ϳ5:1 molar ratio of product to vector) and used to transform the strain UT80 bearing a yku80 deletion. Linearized pYKU80FL was recircularized by incorporation of the mutated YKU80 open reading frame by gap repair (31).
Yeast Two-hybrid Assay-The system used two reporter genes, lex-Aop-LEU2 and LexAop-lacZ (32). Strain EGY48 was transformed with plasmids expressing LexA fusion of full-length YKU80 or YKU80 mutants and/or B42 activation domain fusions of SIR4 (plasmids used were pEG202 and pJG4 -5, respectively) together with LexAop-lacZ reporter plasmid pSH18 -34. The level of interaction between Yku80p and Sir4p was quantitated by yeast ␤-galactosidase assay kit (Pierce). For qualitative analysis, six independent colonies were restreaked on Ura/Trp/His-deficient plates containing raffinose. After 1-3 days of growth at 30°C, a sample of each was resuspended in 0.5 ml of water and 5 l of this were spotted onto the indicated plates in the presence of glucose or galactose. The development of blue color in the presence of X-gal (5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside) was then monitored. Activation of the LEU2 reporter gene was monitored by plating on Ura/Trp/His/Leu-deficient plates in the presence of glucose or galactose.
Coimmunoprecipitation Assay for Yku80p-Sir4p Interaction-YKU70 was cloned under control of the ADH1 promoter into plasmid pAH resulting in pAH-YKU70. YKU80wt and the YKU80 mutants were cloned under control of the ADH1 promoter into plasmid pAU, resulting in pAU-YKU80wt, pAU-YKU80D, and pAU-YKU80PF. Sir4-  was cloned in-frame into plasmid pRS314-GAL-HA (containing the GAL1-10 promoter followed by a three HA epitope). The resulting plasmid pRS314-GAL-HA-Sir4p-(1-200) contains a HA(3)-Sir4p-  fusion under control of the galactose-inducible GAL promoter. For protein extract preparation, the yku80-deficient strain W303h2 was cotransformed with pRS-314-GAL-HA-Sir4p-(1-200), pAH-YKU70, and pAU-YKU80wt (or pAU-Yku80D/pAU-Ylu80PF as indicated). As a control, strain W303h2 was cotransformed with pAH-YKU70, pAU-YKU80wt, and pRS314-GAL-HA as vector control. After overnight growth in selective medium containing raffinose, strains were diluted into YP medium containing 2% raffinose and 2% galactose to A 600 ϭ 0.2 and grown for 4 -5 h to A 600 ϭ 1.0. After harvesting, cell pellets were washed in ice-cold water and disrupted using glass beads in a mini bead-beater. Cell disruption was performed in a buffer containing 25 mM HEPES (pH 7.5), 10% glycerol, 1 mM dithiothreitol, 2 mM EDTA, and 500 mM NaCl. To remove cell debris, extracts were centrifuged at 20,000 rpm for 20 min at 4°C. Supernatant was adjusted to 100 mM NaCl using disruption buffer without NaCl and stored at Ϫ80°C until use. Soluble protein extracts (1 mg) from the indicated strains were incubated with anti-HA antibody bound to protein G-Sepharose in a buffer containing 25 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM dithiothreitol, 10% glycerol, 1 mM EDTA, and 0.1% Nonidet P-40 for 2 h at 4°C. Protein G-Sepharose beads were washed twice with the same buffer containing 200 mM NaCl, twice with the buffer containing 0.2% Nonidet P-40 and 200 mM NaCl, twice in the buffer containing 0.25% Nonidet P-40, and 200 mM NaCl, and twice with buffer containing only 200 mM NaCl. Such extensive washing was necessary to remove the high nonspecific binding. The protein bound to the beads were eluted in sample buffer and resolved in a 10% SDS-PAGE. The proteins were transferred onto a polyvinylidene difluoride membrane and immunoblotted with a rat monoclonal antibody raised against Yku80p. Subsequently, the blot was stripped with 100 mM glycine (pH 2.9) and probed with anti-HA monoclonal antibody.
Chromatin Immunoprecipitation (ChIP) Analysis for Telomere Localization of MYC-Yku80p-ChIP was done as described previously (33,34). After sonication of extracts, a portion (5-10%) of each sample was kept for future analysis as INPUT. Immunoprecipitation with anti-MYC antibody and DNA extraction were carried out as described previously (33). The INPUT and immunoprecipitated (IP) samples were incubated at 65°C overnight to revert formaldehyde cross-linking prior to DNA purification. The extracted DNA was resuspended in water (100 l for each input samples and 40 l for each IP). A portion of each sample (10 l of input and 20 l of IP) was incubated at 95°C followed by 10 min on ice. Samples were transferred to Hybond-N, pre-equilibrated in 10ϫ SSC, using a dot-blot manifold. After baking at 80°C for 20 min, each side of the Hybond-N membrane was cross-linked by exposing to 800 J/m 2 UV-C before prehybridization in Church buffer (0.25 M sodium phosphate, 1 mM EDTA, 1% bovine serum albumen, and 7% SDS). A 300-bp EcoRI fragment containing yeast telomeric repeat sequence (35) or a 600-bp KpnI fragment derived from the telomere proximal side of subtelomeric YЈ-element (36) was used as a probe. A PCR product corresponding to the LCD1 open reading frame was used as a control probe. DNA fragments were radioactively labeled by ran-

ChIP Assay for HA(3)-Sir3p
Localization-ChIP assays were done as described above. Immunoprecipitation was performed with 6 g of anti-HA monoclonal antibody overnight at 4°C, and immunocomplexes were purified with protein G-Sepharose. Both INPUT and immunoprecipitated DNA were analyzed by semi-quantitative PCR in the presence of radiolabeled [␣-32 P]dATP. Primers used were specific for a 380-bp region at the chromosome VI right telomere and a 310-bp region at the HMR loci. PCR products were resolved on 10% polyacrylamide gel. Thereafter, the gel was dried and exposed to a phosphorimaging screen. PCR products were then quantitated by densitometric analysis of the bands using MacBas image analysis software.

Isolation of Separation-of-function Mutants of YKU80 -A
PCR-based procedure was used to generate mutations in the YKU80 open reading frame and to generate a library of yku80 mutants in pRS413. This was then transformed into yku80deficient strain UT80, which contains URA3 integrated into the subtelomeric region of chromosome VII. This URA3 gene is subject to TPE and is consequently transcriptionally silent in the presence of active YKU80 but is derepressed when the silencing functions of YKU80 are impaired. By measuring growth in medium containing 5-FOA, a uracil analogue that is lethal to cells able to metabolize uracil, it is therefore possible to determine the transcriptional status of the telomere-associated URA3 marker. To screen yeast clones containing PCRmutagenized YKU80 for defects in silencing, we plated them onto selective medium with or without 5-FOA and also tested them in parallel for their sensitivity to MMS, which causes DNA base alkylation damage that ultimately results in the generation of DNA DSBs (37). Of the 300 colonies screened, approximately 80% were indistinguishable from the wild-type YKU80 control (yku80-deficient strain containing low copy vector pRS413 carrying full-length YKU80 under the control of its own promoter), whereas ϳ15% behaved as yku80 null mutants (MMS hypersensitive and defective in TPE as evidenced by non-viability on 5-FOA). Such strains were not pursued further.
Instead, we focused on those remaining mutants that grew as well as the control wild-type YKU80 strain in the presence of MMS but were significantly impaired for growth on 5-FOA, suggesting a specific defect in TPE. The plasmids carrying mutated yku80 were rescued from this set of transformants and were then retransformed into strain UT80 to confirm their phenotypes. A closer examination of these mutants revealed that most also exhibited defects in DNA repair. However, two of the clones were consistently found to be dysfunctional only in TPE and displayed no discernable MMS hypersensitivity. Importantly, Western blot analysis revealed that each mutant Mutants display normal NHEJ activity as determined by an in vivo plasmid repair assay. ⌬yku80 cells transformed with empty vector shows a greatly reduced yield of transformants with EcoRI-linearized pBTM116 as compared with cells complemented with full-length YKU80. Both yku80PF and yku80D mutants had normal recircularization ability. Competent cells for indicated yeast strains were transformed in parallel with supercoiled pBTM116 and EcoRIlinearized pBTM116. For each mutant, the value plotted is the number of transformants obtained with EcoRI-linearized vector expressed as a percentage of the number obtained with supercoiled vector. The efficiency of repair for the wild-type cells was normalized to 100%. protein was expressed at a level comparable to that of wild-type Yku80p (data not shown). Sequence analysis revealed that one of the mutants had a Asp to Gly mutation at residue 441 (referred to as Yku80D), whereas the other contained the clustered double mutation Pro-Phe to Ala-Ala at residues 437 and 438 (referred to as Yku80PF). Strikingly, these mutations map FIG. 3. Localization of Yku80p mutants to the telomeres. a, Yku80PFp and Yku80Dp are capable of localizing to the distal ends of the telomere. Cells from either MYC-tagged or untagged wild-type and mutant yeast strains (⌬yku80 strain was also used as control) were used in a ChIP assay with anti-MYC antibody. The immunoprecipitates were analyzed by dot-blot with radioactively labeled telomeric TG (1)(2)(3) repeat DNA. b, Yku80p mutants showed reduced binding to subtelomeric region. Cells from indicated yeast strains were used in ChIP assay with anti-MYC antibody. The immunoprecipitates were analyzed by dot-blot with a radioactively labeled YЈ-subtelomeric DNA fragment 1.5 kb away from the telomere end. As a control, a strain was used where expression of Sir4p was driven by a GAL promoter. Anti-MYC immunoprecipitates of extracts prepared from galactose-induced GAL-SIR4 strain were enriched for the subtelomeric fragment. In each case, the blots were stripped and the immunoprecipitates were probed for LCD1 as a nonspecific control. to a region that is highly conserved in Ku80 and Ku70 (Fig. 1a) (see Ref. 38). Therefore, these data imply that this region of Yku80p plays a key role in telomeric silencing.
The phenotypes caused by the yku80D(441)G and yku80PF-(437,438)AA mutations were confirmed by introducing them into wild-type YKU80 by site-directed mutagenesis and then transforming the resulting mutant yku80 alleles into the strain UT80. Drop-test growth assays revealed that both mutants were indeed hypersensitive to 5-FOA but not to MMS (Fig. 1b) or to phleomycin (data not shown), confirming their separationof-function phenotypes. The yku80D mutant was, however, consistently more sensitive to 5-FOA than the yku80PF mutant, suggesting that the two mutants differ in the degree to which TPE is impaired.
Further Analysis of NHEJ in the yku80-silencing Defective Mutants-To fully ascertain the NHEJ phenotypes of the yku80D and yku80PF mutants, we employed a transformationbased in vivo plasmid repair assay (6,8). Thus, yku80-deleted strain UT80 expressing these mutants or wild-type YKU80 was transformed with pBTM116 that had been linearized with EcoRI. The number of resulting transformants was then ascertained and normalized to the number of colonies arising from a parallel series of transformations with supercoiled pBTM116. In line with previous findings (6,8,9), the ⌬yku80 strain displayed a dramatic decrease in end-joining efficiency compared with the ⌬yku80 strain complemented with full-length YKU80. Notably, however, the yku80D and yku80PF mutants each showed a high level of NHEJ that was essentially indistinguishable from that of the strain expressing wild-type Yku80p (Fig. 2). These results therefore confirm that the yku80D and yku80PF mutants are not impaired in NHEJ.
TPE-deficient Yku80p Mutants Localize to Terminal but Not Internal Telomeric Regions-Although the localization of yeast Ku to chromosome ends appears to be independent of other factors involved in telomeric silencing, its localization to more internal subtelomeric regions depends on Sir4p (26). To see whether the TPE-deficient Yku80p derivatives were impaired in any telomere interactions, we performed ChIP experiments using MYC-tagged derivatives of the mutants and wild-type Yku80p. Thus, after protein-DNA cross-linking of cells, the wild-type and mutant proteins were immunoprecipitated using an anti-MYC monoclonal antibody and the immunoprecipitates were then analyzed for the presence of telomeric DNA by dotblot hybridization with a radioactively labeled probe corresponding to the terminal telomeric TG (1)(2)(3) repeats. Perhaps surprisingly, telomeric DNA was detected in the anti-MYC immunoprecipitates from both mutants and the magnitude of the hybridization signal was similar to that obtained with MYC-tagged wild-type Yku80p (Fig. 3a). By contrast, when immunoprecipitations were performed in the absence of prior cross-linking, no hybridization to the telomeric probe was detected (data not shown). In addition, no telomeric DNA was retrieved from cells expressing untagged mutant or untagged wild-type Yku80p or from the ⌬yku80 strain (Fig. 3a). Furthermore, no enrichment of DNA from a control locus (LCD1) was observed in any of the immunoprecipitates. Quantification of telomere signals in immunoprecipitates relative to the input and to the signal from a nonspecific LCD1 locus further confirmed this result. Taken together, these data indicate that the TPE-defective Yku80p mutants are still recruited to the terminal telomeric region.
We next tested for the recruitment of wild-type and mutated Yku80p derivatives to internal subtelomeric regions by using a probe corresponding to the telomere-proximal side of the subtelomeric YЈ-element, which is ϳ870 bp from the start of the telomeric TG (1-3) sequences (36). As Sir4p is required for the recruitment of Yku80p to internal telomeric sequences (26), a control strain (DMR428D80) complemented with MYC-tagged wild-type Yku80p was used. In this strain, the genomic copy of SIR4 is driven by the galactose-inducible promoter. Consequently and as shown previously (26), we detected subtelomeric YЈ-DNA in anti-MYC immunoprecipitates from this strain when it had been grown in the presence of galactose but not when Sir4p expression was repressed by growth in the presence of glucose (Fig. 3b, left panel). Also consistent with previous results, subtelomeric YЈ-DNA was immunoprecipitated from strains expressing MYC-tagged Yku80p but not untagged strains. Interestingly, no enrichment of subtelomeric DNA was observed from cells expressing MYC-tagged derivatives of the TPE-deficient Yku80p mutants compared with the untagged control (Fig. 3b, right panel). This experiment was performed multiple times, and in each case, enrichment of subtelomeric DNA was observed with MYC-tagged wild-type Yku80p but not the MYC-tagged mutants when compared with their respective untagged controls. Therefore, these data imply that the TPEdeficient Yku80p mutants are defective in their recruitment to the internal subtelomeric sequences.
Yku80p Interacts with Sir4p-The Sir protein complex is essential for the establishment of transcriptional silencing at yeast telomeres (18). Sir4p plays a particularly important role in TPE by initiating the assembly of telomeric heterochromatin via direct interactions with Rap1p (22). However, the association of Sir4p with more internal telomeric sequences requires Ku and other heterochromatin assembly proteins (22,26). Previous work (39) revealed an interaction between Sir4p and Yku70p in the yeast two-hybrid system, suggesting a possible mechanism for Sir4p recruitment to internal telomeric sequences. When we performed two-hybrid screens with Yku70p or Yku80p as baits, in each case we retrieved clones encoding the N-terminal 600 amino acid residues of Sir4p. Notably, however, such SIR4 clones were identified much more frequently with Yku80p than with Yku70p as bait (29 times and once, respectively). Moreover, when we used full-length Sir4p FIG. 4. Yku80p interacts directly with Sir4p in a yeast twohybrid assay. Full-length YKU80 was fused to the LexA DNA binding domain in the plasmid pEG202 (pEG202-YKU80) and cotransformed with the activation domain plasmid pJG4 -5 with full-length SIR4 (pJG-SIR4FL) into the interaction trap strain EGY48-Wt. As a control, EGY48 was co-transformed with empty pEG202 and pJG-SIR4FL or pEG202-YKU80 and empty pJG4 -5. To test for Yku70p-independent interaction of Yku80p, Sir4p pEG-YKU80 and pJG-SIR4FL were cotransformed into the YKU70 deletion strain EGY48⌬70. The plasmid pSH18 -34 carrying the LexAop-lacZ reporter gene was also introduced in these strains. There was a large increase in the transcriptional activity of the reporter gene as shown by ␤-galactosidase activity in strains transformed with pEG-YKU80 and pJG-SIR4 compared with their interaction with the pJG4 -5 or the bait vector pEG202. A similar level of Yku80p-Sir4p interaction was detected in the strain deleted for Yku70p. Wt, wild type.
in such assays, we detected a strong interaction with Yku80p but were unable to demonstrate a significant interaction with Yku70p (data not shown). Quantification of ␤-galactosidase activity confirmed this strong interaction between Yku80p and Sir4p (Fig. 4). Importantly, a similar level of ␤-galactosidase activity was also detected when the Yku80p-Sir4p two-hybrid interaction assay was performed in a yku70-deleted strain or strain deleted for Sir2p or Sir3p (Fig. 4 and data not shown). Therefore, we conclude that the interaction between Yku80p and Sir4p is probably direct.
Mapping the Protein Regions Required for the Yku80p-Sir4p Interaction-To characterize the region(s) of Sir4p that interacts with Yku80p, a series of SIR4 deletion derivatives expressed as B42 activation domain fusions was tested in the two-hybrid assay. As depicted in Fig. 5a, interaction with Yku80p was markedly reduced by the removal of the N-terminal 50 amino acid residues from Sir4p and was completely abolished by the more extensive N-terminal truncation of residues 1-105. By contrast, interaction with Yku80p still took place when much of the C terminus of Sir4p had been deleted.
Indeed, removal of C-terminal sequences augmented the Yku80p interaction with the strongest interaction being observed with a derivative corresponding to the first 200 amino acid residues of Sir4p (Fig. 5a). Significantly, this interaction was reduced by further C-terminal truncation of Sir4p to residue 176 (derivative 1-175) and was essentially abolished when the C-terminal truncation extended to residue 151 (derivative 1-150). In addition, Yku80p could be coimmunoprecipitated using the HA epitope-tagged N-terminal 1-200 amino acid residues of Sir4p (HA-Sir4p-(1-200)) (Fig. 5b, lane 6). These results therefore establish that the N-terminal portion of Sir4p mediates interactions with Yku80p.
In an analogous manner, we investigated which region of Yku80p is necessary for interaction with Sir4p. The N-terminal boundary of this interaction domain was not defined because N-terminal truncations of Yku80p had intrinsic activity in the two-hybrid assay (data not shown). However, C-terminal truncations were more informative. Thus, truncation of 179 amino acid residues from the C terminus of Yku80p had no discernible effect on the interaction with Sir4p (Fig. 5a, derivative 1-450).  5) and immunoprecipitated with anti-HA antibody-coupled protein G-Sepharose beads. Proteins bound to the beads were eluted in sample buffer resolved together with 1% of the input materials on SDS-PAGE and immunoblotted with anti-Yku80p monoclonal antibody (top panel). The blot was stripped and probed with anti-HA antibodies (lower panel). c, yku80PF and yku80D mutant proteins show reduced interaction with Sir4p in the two-hybrid assay. Full-length YKU80 or mutants were fused to LexA in the plasmid pEG202 and cotransformed into the interaction trap strain EGY48 with the activation domain plasmid pJG4 -5 with or without full-length SIR4 together with LexAop-lacZ reporter plasmid pSH18 -34. ␤-Galactosidase activity was significantly decreased in strains expressing the Yku80 mutants transformed with pJG-SIR4 compared with the wild-type control.
However, deletion of a further 25 amino acid residues from Yku80p (derivative 1-425) led to an undetectable level of Yku80p-Sir4p interaction. Thus, the C-terminal boundary of the Sir4p interaction domain lies between residues 425 and 450 of Yku80p. As shown above, this region also contains the sites of the silencing defective mutations. Significantly, we found that each of the silencing defective Yku80p mutants was impaired in its ability to interact with Sir4p in the coimmunoprecipitation assay using HA-Sir4p-(1-200) (Fig 5b, lanes 7 and 8) as well as full-length Sir4p in the two-hybrid assay (Fig. 5c). Taken together, these results suggest that Yku80p interacts with Sir4p and thereby helps to recruit Sir4p to internal telomeric regions. In addition, the results suggest that the TPE defects of the yku80PF and yku80D mutants, at least in part, reflect impaired interactions with Sir4p.
Silencing defective Yku80p Mutants Lead to Reduced Binding of Sir3p to Telomeres-Sir4p binding to telomeres is not sufficient for the formation of telomeric heterochromatin. Binding of Sir4p is followed by recruitment of Sir2p and Sir3p, and this process results in the spreading of silencing. To investigate whether the mutants yku80PF and yku80D are affected in the recruitment of Sir3p to telomeres, we used the yeast strain FIG. 6. yku80PF and yku80D mutants show decreased Sir3p binding to telomeres in a ChIP assay. GA823D80 (HA(3)-Sir3 ku80::KanMX4) was transformed with full-length YKU80, mutants, or vector alone, and anti-HA immunoprecipitates were analyzed by PCR using telomere and HMR-specific primers. Simultaneously, the input was also processed for PCR. The products of PCR from input and IP were resolved on a 1% agarose gel. Densitometric quantitation of the telomere and HMR signal was performed using MacBas image analysis software. For each strain, input and IP values are the ratio of telomere signal over HMR signal and the cumulative ratio of input to IP gives a reflection of Sir3p binding to the telomere. There was a ϳ50% decrease in Sir3p binding to the telomere in both mutants as compared with the wild-type control.
GA823D80 in a ChIP-based assay. This strain contains a genomic copy of Sir3p endogenously tagged with three HA epitopes. Formaldehyde cross-linked protein extracts from the indicated strains were immunoprecipitated with anti-HA monoclonal antibody, and then the immunoprecipitates were tested for the presence of telomeric DNA by performing quantitative PCR with telomere-specific primers using radiolabeled [␣-32 P]dATP. As an internal control, the immunoprecipitates were also tested for HMR DNA because silencing at the HMR locus depends on Sir proteins but not on Ku (9,18). The PCR signals obtained were quantified, and the signal obtained at the telomere was normalized to that obtained at the HMR locus (Fig. 6). These results indicate that Sir3p recruitment to telomeres is severely affected in the yku80PF mutant and is also reduced with the yku80D mutant compared with wild-type control. These data therefore support a model in which the observed TPE defect reflects defective recruitment of Sir proteins to telomeric regions, leading to impaired assembly of telomeric heterochromatin. DISCUSSION S. cerevisiae Ku is involved in at least two different functions: NHEJ and transcriptional silencing of telomere proximal genes (6,8,9). Here, we have demonstrated that these two functions of Ku are distinct and can be dissociated from one another. Thus, through phenotypic screening of randomly generated YKU80 mutations, we have identified two separationof-function mutants, yku80PF(437,438)AA and yku80D(441)G, that are defective in TPE but proficient for DSB repair as measured by an in vivo plasmid repair assay and by survival in presence of MMS or phleomycin. Expression levels of these mutant proteins were similar to that of the wild-type protein, suggesting that their phenotypes reflect impaired protein function rather than reduced translation or protein stability. Fur-thermore, we have established that Yku80p interacts with Sir4p and have shown that this interaction involves the Nterminal 200 amino acid residues of Sir4p and residues 425-450 of Yku80p. Notably, the yku80PF(437,438)AA and yku80D(441)G mutations map within this 25 amino acid region and impair the Sir4p-Yku80p interaction as measured by twohybrid assays and coimmunoprecipitation. Taken together, these results indicate that the Sir4p-Yku80p interaction is likely to play a critical role in the establishment of TPE and suggest that the TPE defects of the yku80PF(437,438)AA and yku80D(441)G mutants are caused, at least in part, by impaired interactions with Sir4p.
The sites of our separation-of-function Yku80p mutants are essentially invariant in Ku80 orthologues from a range of species and map to a region that forms part of the structural core of the Ku heterodimer ( Fig. 1 and Refs. 38 and 40). In addition, this region is well conserved in the analogous part of Ku70. Although previous structural and mutational analyses have implicated these regions in mediating heterodimerization between Ku70 and Ku80 (40 -42), we did not detect any loss of interaction between Yku70 and the mutated Yku80p subunits in our assays (data not shown). Moreover, the fact that the yku80PF and yku80D mutants appear to be proficient in DSB repair corroborates their functional competence to form a heterodimer with Yku70p. When we mapped the location of the mutated residues on to the crystal structure of the human Ku heterodimer (40), residues Pro-437, Phe-438, and Asp-441 of Yku80p mapped to a region that is partly buried in the crystal structure of human Ku (data not shown), indicating that they are probably not accessible for direct proteinprotein interactions. Instead, it seems more likely that the yku80PF(437,438)AA and yku80D(441)G mutations compromise the conformational integrity of the protein. Therefore, we FIG. 7. A schematic model for the establishment of silencing at telomeres. In wild-type cells, Sir4p is constitutively bound to Sir2p and the N terminus of Sir4p interacts with its C terminus (44). This intramolecular interaction inhibits Sir3p from binding to Sir4p. At telomeres, Yku80p interacts with the N terminus of Sir4p, inducing a conformational change in Sir4p. This unmasks the Sir3p interaction domain on Sir4p and facilitates the recruitment of Sir3p. Thereafter, cooperative interactions between Sir proteins and the Ku heterodimer bring about the spreading of silencing. In the case of the mutant proteins Yku80PFp or Yku80Dp, Sir4p is recruited only to Rap1p (22). These mutants being debilitated for their interaction with Sir4p are unable to induce a conformational change within Sir4p; hence, Sir3p is not effectively recruited, thereby limiting the spreading of silencing speculate that these mutations specifically but indirectly influence the ability of the Ku heterodimer to interact with Sir4p and perhaps other proteins and thus bring about the phenotypic changes we see in yeast strains bearing yku80PF or yku80D mutations.
One explanation for the observed defect in TPE brought about by the P437A/F438A and D441G mutations is that these mutations lead to a reduction or loss in binding of the Ku heterodimer to the telomere end. However, when we analyzed the telomeric localization of the Yku80 mutants in chromatin immunoprecipitation experiments, we found that the yku80PF and yku80D mutant proteins were capable of binding to the distal end of the telomere as efficiently as wild-type Yku80p. Could it be then that the mutants were defective in some aspect of the subsequent spreading of silencing components from the telomere ends? The formation of telomeric heterochromatin is a complex procedure and is orchestrated by a concerted sequence of events in which the spreading of the silencing components including Ku and Sir proteins to subtelomeric regions is a prerequisite for the formation of telomeric heterochromatin and a fully active TPE. Having bound to the most distal end of the telomere, yeast Ku appears to spread along the repressed subtelomeric region for ϳ3 kb (26), the binding of yeast Ku and Sir4p to more internal sequences being mutually dependent on each other (22,26). In our ChIP experiments, we observed a severe reduction in the recruitment of Ku to the internal subtelomeric regions, indicating that the formation of telomeric heterochromatin is perturbed in these mutants. Therefore, the reduced Sir4p interaction observed in the two-hybrid assay with the Yku80p mutants might be the reason for impaired recruitment of these mutants to internal subtelomeric regions, thus leading to ineffective assembly of telomeric heterochromatin and thereby loss of TPE.
An important step in establishment of the silencing apparatus and its spreading is the recruitment of Sir3p and Sir2p (43). It has been suggested that Sir4p bound to Rap1p initiates the recruitment of Sir2p and Sir3p to the telomere. Although Sir2p has been shown to constitutively interact with Sir4p in crude yeast extract, Sir3p can only interact with Sir4p after removal of its N-terminal two-thirds, indicating that the Sir4p N-terminal domain somehow inhibits interaction of Sir3p with the Sir4p C terminus (44). An early step in Sir protein assembly might, therefore, be the unmasking of the Sir4p C terminus to allow its association with Sir3p.
Our results indicate that Yku80p interacts with the Sir4p N-terminal domain. An attractive model for how yeast Ku might help to establish TPE, therefore, is that the Yku80p-Sir4p interaction relieves the inhibitory intramolecular interaction within Sir4p that normally prevents Sir3p recruitment. In other words, Yku80p may induce a conformational change within Sir4p that facilitates the recruitment of Sir3p to the C terminus of Sir4p. An illustration of this model is depicted in Fig. 7. This model is supported not only by our finding that Sir3p recruitment to the telomere is significantly reduced in our Yku80 mutants, which do not interact efficiently with Sir4p, but also by the rescue of the silencing defect of our mutants by overexpression of Sir3p (data not shown).
In conclusion, we propose that the Yku80p-Sir4p interaction, by facilitating the recruitment of Sir3p to telomeres via Sir4p, is essential for the establishment of TPE. This model is probably not operative at the mating type loci where Ku is not required for the assembly of Sir proteins, and it would therefore be interesting to know what the driving factor for Sir3p recruitment is at these loci and how the assembly of silent heterochro-matin here differs from that assembled at the telomeres. In addition, further in-depth analysis of the Yku80p mutants described in this study should enable a better understanding of the involvement of Ku in telomere length maintenance as well as in the establishment of TPE.