A Quantitative Assay for Telomere Protection in Saccharomyces cerevisiae

Corresponding author: Daniel E. Gottschling, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave. North, MS A3-025, Seattle, WA 98109-1024., dgottsch{at}fhcrc.org (E-mail)

Communicating editor: L. PILLUS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Telomeres are the protective ends of linear chromosomes. Telomeric components have been identified and described by their abilities to bind telomeric DNA, affect telomere repeat length, participate in telomeric DNA replication, or modulate transcriptional silencing of telomere-adjacent genes; however, their roles in chromosome end protection are not as well defined. We have developed a genetic, quantitative assay in Saccharomyces cerevisiae to measure whether various telomeric components protect chromosome ends from homologous recombination. This "chromosomal cap" assay has revealed that the telomeric end-binding proteins, Cdc13p and Ku, both protect the chromosome end from homologous recombination, as does the ATM-related kinase, Tel1p. We propose that Cdc13p and Ku structurally inhibit recombination at telomeres and that Tel1p regulates the chromosomal cap, acting through Cdc13p. Analysis with recombination mutants indicated that telomeric homologous recombination events proceeded by different mechanisms, depending on which capping component was compromised. Furthermore, we found that neither telomere repeat length nor telomeric silencing correlated with chromosomal capping efficiency. This capping assay provides a sensitive in vivo approach for identifying the components of chromosome ends and the mechanisms by which they are protected.

THE work of H. J. Muller and Barbara McClintock defined telomeres as the protective ends of linear chromosomes. Muller coined the term "telomere" as he observed that X-ray-induced chromosomal inversions, fusions, and translocations in Drosophila never involved chromosome ends (MULLER 1938 ); he therefore reasoned that the native ends were protected. McClintock reached similar conclusions by following dicentric chromosomes that were torn apart during meiotic divisions (MCCLINTOCK 1941 , MCCLINTOCK 1942 ). From these breakage-fusion-bridge experiments, she found that broken chromosome ends are unstable and must be repaired, either by a fusion with another chromosome end or by a healing event that creates a stable chromosome end (by the addition of telomeric sequence repeats, as was later revealed). Thus, both of these pioneers were able to distinguish between broken DNA ends and native ends. Today we know that broken DNA ends are ideal substrates for recombination and DNA repair activities, whereas chromosome ends are not recognized as damaged DNA, do not stimulate a cell cycle arrest, and do not undergo recombination with the same high frequency as double-stranded breaks (DSBs; BLACKBURN and GREIDER 1995 ; PAQUES and HABER 1999 ).

The ends of most eukaryotic linear chromosomes terminate in telomeric DNA, repeats of 5–25 bp that vary in sequence among species and end in a 3' single-stranded overhang (WELLINGER and SEN 1997 ). The telomeric DNA sequences are synthesized by telomerase (GREIDER and BLACKBURN 1987 ), a protein-RNA complex, and the DNA repeats are bound by telomere-specific binding proteins that, with other proteins, form a telomeric complex (reviewed in BLACKBURN and GREIDER 1995 ). In the absence of telomerase, telomeric DNA is slowly lost with each round of DNA replication, which leads to a senescent phenotype that typically ends in cell death (LUNDBLAD and SZOSTAK 1989 ). During these dire circumstances, the telomeric structures are altered, and an inefficient mechanism of homologous recombination can be used to maintain the telomeric DNA (LUNDBLAD and BLACKBURN 1993 ; TENG and ZAKIAN 1999 ). Under normal circumstances, subtelomeric regions exhibit relatively low rates of recombination (LOUIS 1995 ; STAVENHAGEN and ZAKIAN 1998 ); the rates of interchromosomal recombination between telomeric repeats have not been analyzed directly.

The protection of the chromosome end arguably depends on the structure of the telomeric terminus—the absolute end of the chromosome. Many studies involving telomere biochemistry, telomeric length maintenance, and the transcriptional silencing of telomere-adjacent genes have identified several components that contribute to telomeric structure (MCEACHERN et al. 2000 ; BAILEY et al. 2001 ), but whether all these proteins are required for protection of the end is unclear. To date, the greatest number of telomeric components has been identified in the budding yeast Saccharomyces cerevisiae. The telomeric DNA repeats of the TG_1–3 sequence are bound by Rap1p, an essential protein that interacts with other telomeric components including Rif1p, Rif2p, and the silent chromatin proteins Sir3p and Sir4p (reviewed in DUBOIS et al. 2000 ; MCEACHERN et al. 2000 ). An essential single-stranded DNA-binding protein, Cdc13p, binds the 3' TG_1–3 overhang at the end of the chromosome (GARVIK et al. 1995 ; NUGENT et al. 1996 ; HUGHES et al. 2000 ) and participates in the recruitment of telomerase and DNA polymerases to the chromosome (NUGENT et al. 1998 ; EVANS and LUNDBLAD 1999 ; QI and ZAKIAN 2000 ). Interactions between Cdc13p and two other essential telomeric components, Stn1p and Ten1p, have been suggested to contribute to chromosome end protection (GRANDIN et al. 1997 , GRANDIN et al. 2001B ; PENNOCK et al. 2001 ).

In other species, structural details of the molecular interactions between telomere-binding proteins and the telomeric DNA repeats have been characterized. Ciliated protozoa have a heterodimeric protein complex that envelops the macronuclear chromosomal DNA end (FANG and CECH 1995 ; HORVATH et al. 1998 ). Mammalian telomeric repeats are bound by two proteins, TRF1 and TRF2, that in a TRF2-dependent manner, create t loop structures, in which the 3' single-stranded telomeric DNA overhang invades the upstream duplex telomeric DNA (GRIFFITH et al. 1999 ). T loops also have been detected at the ends of chromosomes in Trypanosoma brucei (MUNOZ-JORDAN et al. 2001 ) and in the polytene chromosomes of Oxytricha nova (MURTI and PRESCOTT 1999 ), which suggests that the sequestration of the DNA end within the t loop may have evolved as a form of end protection.

Several telomeric proteins are conserved evolutionarily: The TRF proteins are related to the fission yeast Schizosaccharomyces pombe telomere-binding protein Taz1 (LI et al. 2000 ), and Taz1 interacts with homologs of the S. cerevisiae Rap1p and Rif1p at S. pombe telomeres (KANOH and ISHIKAWA 2001 ). In addition, another S. pombe and human protein, Pot1, is related to the ciliate telomere-binding proteins (BAUMANN and CECH 2001 ), and human RAP1 binds to TRF2 (LI et al. 2000 ). Furthermore, the Ku protein complex is conserved among eukaryotes, including S. cerevisiae, S. pombe, mice, and humans. Paradoxically, Ku binds to both telomeres and DSBs; in the latter case, it participates in DNA repair via nonhomologous end-joining (NHEJ; reviewed in TUTEJA and TUTEJA 2000 ). At S. pombe telomeres, Ku is required to maintain normal telomeric DNA tract length and subtelomeric repeat stability (BAUMANN and CECH 2000 ; MANOLIS et al. 2001 ). In S. cerevisiae, Ku participates in telomere length maintenance (BOULTON and JACKSON 1996 ; PORTER et al. 1996 ), subnuclear localization of telomeres (LAROCHE et al. 1998 ), transcriptional silencing of telomere-adjacent genes (BOULTON and JACKSON 1998 ; NUGENT et al. 1998 ), and the inhibition of intrachromosomal recombination between elongated telomeric DNA repeats (POLOTNIANKA et al. 1998 ).

Several in vivo assays have addressed whether a telomere is protected or "capped," but they have led to different definitions of what constitutes a cap. Many assays have simply reflected the need for telomeric DNA repeats at the end of the chromosome. For instance, telomeres in telomerase-negative strains are said to be uncapped because the telomeric repeats continually shorten and can lead to end-to-end fusions (MCEACHERN and BLACKBURN 1996 ; HACKETT et al. 2001 ). A particularly striking example is in S. pombe, where cells that survive the loss of telomerase activity undergo stable circularization of all three chromosomes (NAKAMURA et al. 1998 ).

The possible "capping" functions of some of the aforementioned telomeric components have been investigated. For example, deletion of either of the telomere-binding protein genes taz1⁺ or pot1⁺ in S. pombe leads to end-to-end fusions between telomeres (BAUMANN and CECH 2001 ; FERREIRA and COOPER 2001 ), which suggests that these proteins normally protect the chromosome end. Similarly, end-to-end fusions have been visualized by immunofluorescence microscopy in Drosophila strains containing mutations in the heterochromatin protein HP1 (FANTI et al. 1998 ) and in mammalian cells with mutations in the TRF2 gene (VAN STEENSEL et al. 1998 ) or the Ku complex (BAILEY et al. 1999 ; GILLEY et al. 2001 ). In addition, the temperature-sensitive cdc13-1 and stn1-13 mutations in S. cerevisiae give rise to long tracts of single-stranded TG_1–3 DNA at the chromosome ends (GARVIK et al. 1995 ; GRANDIN et al. 1997 ; POLOTNIANKA et al. 1998 ), suggesting that Cdc13p and Stn1p normally protect the telomeric end against extended resection of the C_1–3A strand. Telomeric DNA repeat length has been another suggested indicator of compromised telomere caps. For example, chromosomal ends with telomeric repeat sequences of abnormal length that are created by an altered telomerase RNA template or by a mutant telomere-binding protein have been described as uncapped (KRAUSKOPF and BLACKBURN 1998 ; SMITH and BLACKBURN 1999 ; MCEACHERN and IYER 2001 ). Furthermore, a telomeric rapid deletion assay, in which artificially created long terminal telomeric DNA tracts are shortened, suggests that intrachromosomal recombination with the chromosome terminus may play a role in telomeric size control (BUCHOLC et al. 2001 ).

Telomere capping also has been inferred by recombination assays. For example, a gradient of mitotic recombination that is highest near the telomere is observed in S. cerevisiae strains with a cdc13-1 mutation (CARSON and HARTWELL 1985 ; HARTWELL and SMITH 1985 ). In addition, the telomeres of cdc13-1 checkpoint-deficient strains show increased levels of recombination with other telomeric sequences (GRANDIN et al. 2001A ). Similarly, other analyses in yeast have used recombination between terminal telomeric sequences and internal telomeric tracts on other chromosomes (located within subtelomeric elements) as evidence for loss of terminal protection (LUNDBLAD and BLACKBURN 1993 ; TENG and ZAKIAN 1999 ). Typically these events are detected in telomerase-deficient cells as cells escape senescence, and as a result, it is impossible to ascertain when these events occur or how often they occur. There is no selection for a single recombination event: Recombination is detected in cells that have undergone a large number of recombination events to survive senescence. Furthermore, these recombination events have been detected only by Southern analysis. Consequently, it has been difficult to use such an assay to assign a telomere-capping role to most telomeric components.

To better understand how chromosome ends are capped, we took advantage of the strengths of S. cerevisiae to develop an in vivo assay that would be amenable to evaluating one aspect of end protection at a time and be quantitative in nature. Furthermore we sought that it be adaptable for use in a genetic screen to identify additional factors that participate in the chromosomal cap or its regulation. The genetic assay we have developed quantitatively measures the ability of native chromosome ends to be protected from homologous recombination.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmid and DNA manipulations:
The following fragments were PCR amplified and used in the construction of the artificial chromosomes. The "HIS3 end" was PCR amplified from pHIS3.2000.TEL(+) with oligos lam-his-1 (YE4bT sequence in uppercase letters and HIS3 sequence in lowercase: AGCCTAGCCGGGTCCTCAACGACAGGAGCACGATCATGCGgcctcggtaatgattttcat) and lam-his-2 (CGAATTCCCCCTGCCACCAC) to generate an 3.4-kb fragment. pHIS3.2000.TEL(+) was constructed by S. Diede by inserting the 2-kb HpaI fragment from DNA into the NruI site of pAKHIS3.14. pAKHIS3.14 also was created by S. Diede by ligating a HIS3 PCR fragment into the blunted BglII and HindIII sites of pVZ1 (HENIKOFF and EGHTEDARZADEH 1987 ). The HIS3 PCR fragment was PCR amplified from pRS303 (SIKORSKI and HIETER 1989 ) using the 5' oligo (GGATCCTGCCTCGGTAATGATTTTC) and 3' oligo (GGATCCTCTCGAGTTCAAGAGAAAAAAAAAGAA).

The "URA3 fragment" was PCR amplified from pRS306 (SIKORSKI and HIETER 1989 ) with the oligos mchr-RS+ (ade3-2p sequence in uppercase and pRS sequence in lowercase: GCGGCCGTCCGCTTGCTTTTCCAACAATCTGTGCTTTAGCctgtgcggtatttcacaccg) and mchr-RS- (CEN4 sequence in uppercase and pRS sequence in lowercase: CTTCTTCCGCCTTTTTCTTTTTCGAACTTTTCTATATCGagattgtactgagagtgcac).

pSD243 was constructed by S. Diede from a ligation of the BglII/SpeI blunted fragment with MET15 from pGC3 (COST and BOEKE 1996 ) into the blunted HindIII/NruI sites of pAK1 (HUANG et al. 1997 ); MET15 is transcribed toward the telomere. To create pMET-HO-TELO, the oligos HO-1 (GATCCGCGGTTATACTGTTGCGCGAAGTAGTCCCATAAAAGATCT) and HO-2 (GATCAGATCTTTTATGGGACTACTTCGCGCAACAGTATAACCGCG) were annealed and ligated into the BamHI site of pSD243; the TGTT "overhang" (post-cleavage by HO endonuclease) is on the opposite strand of the telomere sequence. To create pMET-HO, the 139-bp BbsI/BsaAI fragment from pSC9 (MAT plasmid; ADAMS et al. 1998 ) was ligated into the blunted BglII/NotI sites of pMET-HO-TELO.

Oligos HOint1 (artificial chromosome sequence upstream of ARS1 in uppercase and pMET-HO-TELO sequence in lowercase: GTGAAGGAGCATGTTCGGCACACAGTGGACCGAACGTGGGcaaatggcacgtgaagctgtc) and HOint2 (artificial chromosome sequence downstream of TRP1 in uppercase and pMET-HO-TELO sequence in lowercase: GTCTGTTATTAATTTCACAGGTAGTTCTGGTCCATTGGTGcagctcattttttaaccaataggc) were used to PCR amplify both the "MET-HO-TG" fragment from pMET-HO-TELO and the "MET-HO" fragment from pMET-HO.

Yeast methods and strain construction:
Rich (YEP) and synthetic complete yeast (YC) media have been described (VAN LEEUWEN and GOTTSCHLING 2002 ) and are on the lab's website: http://www.fhcrc.org/labs/gottschling/homepage.html. Strains with artificial chromosomes were grown in dropout synthetic media (either YC-ura or YC-trp). Drug supplements to the YC media include 60 µg/ml canavanine (can; Sigma, St. Louis), 3 µg/ml cycloheximide (cyh; Sigma), and 1 mg/ml 5-fluoroorotic acid (FOA; Toronto Research), and YEP supplements include 200 µg/ml G418 (for kanMX gene) or 200 µg/ml hygromycin B (for hph gene). Yeast genomic DNA isolations and transformations have been described (ADAMS et al. 1998 ).

The artificial chromosome YE4bT from R. Wellinger (WELLINGER and ZAKIAN 1989 ), with yeast telomeres, a yeast centromere (CEN4), and a single origin of replication (ARS), was modified: The artificial chromosome was transferred into BY4705a [BY4705 (BRACHMANN et al. 1998 ) converted to MATa with pSC11 (ADAMS et al. 1998 ) by A. Stemm-Wolf] to create UCC2247. The URA3 gene of the artificial chromosome in UCC2247 was replaced with the HIS3 marker and sequence by transformation with the "HIS3 end" (below) to create UCC2249. UCC2249 was transformed with the "URA3 fragment" to create UCC2301, which has the URA3 gene inserted between ade3-2p and CEN4. UCC2301 was transformed with the "MET-HO-TG" fragment, and this artificial chromosome was transformed into UCC5843 to create UCC2311 and UCC2313 (two independent transformants). Similarly the "MET-HO" fragment was transformed into UCC2301, and then this artificial chromosome was transferred into UCC5843 to create UCC2317 and UCC2319 (two independent transformants). The "MET-HO-TG" (A) and the "MET-HO" (C) fragments replace the ARS1-TRP1 regions of YE4bT, resulting in the final set of artificial chromosomes (65 kb each). Strains UCC2311, UCC2313, UCC2317, and UCC2319 were then mated with YPH925 (a kar1 strain; SPENCER et al. 1994 ) and plated on YC-his-ura+cyh (to select for the transfer of the artificial chromosome into YPH925) to create UCC2340-A1, UCC2340-A3, UCC2340-C1, and UCC2340-C4, respectively.

All wild-type and single-mutant strains used in the assays (Table 1) are MATa-inc strains (SWEETSER et al. 1994 ) derived from UCC2325, which is a spore product from the diploid UCC2324, which itself came from the mating of BY4727 (BRACHMANN et al. 1998 ) and UCC5843. UCC5843 was a spore product from the mating of BY4705 (BRACHMANN et al. 1998 ) and DY3023 (SWEETSER et al. 1994 ). UCC2325 was transformed with pRS304-derived and pRS402-derived PCR products (BRACHMANN et al. 1998 ) to replace the entire CAN1 open reading frame (ORF) and produce UCC2330 and UCC2338, respectively. To generate disruptions of RAD1 and RAD52, UCC2330 was transformed with BamHI-digested pRR46 (REYNOLDS et al. 1987 ) and BamHI/PvuII-digested pBE77 (a gift from R. Esposito) to generate UCC2719 and UCC2335, respectively; similarly, UCC2421 and UCC2770 were made from UCC2338, and UCC2384 from UCC2332. To generate cdc13-1 and stn1-13 strains, pop-in-pop-out plasmids were used: UCC2330 was transformed with XhoI-digested pVL451 (a gift from V. Lundblad) and BstXI-digested YIplac211-stn1-13 (GRANDIN et al. 1997 ), Ura+ colonies were selected, and FOA-resistant colonies that exhibited temperature sensitivity at 37° and long telomeric repeat tracts were selected as UCC2356 and UCC2446, respectively; similarly, UCC2357 was constructed, starting with UCC2332. To generate UCC2842 and UCC2843, UCC2330 was transformed with PCR-based null mutations of HML and HMR and with HindIII-digested Be198 and BamHI-digested Be199 (IVY et al. 1986 ), respectively. Null mutations were generated using PCR-based gene knockouts via homologous recombination to replace the entire ORF (BRACHMANN et al. 1998 ; GOLDSTEIN and MCCUSKER 1999 ).

All double-mutant strains were constructed by sporulation of diploids that were heterozygous for the mutations of interest (Table 2). The diploids were created from matings of single-mutant haploid strains, all of which were derived from UCC2325, UCC2326, or UCC2327.

Telomere capture assays:
The following steps were performed with each yeast strain (wild type or mutant) shown in Fig 3 and Fig 4. To transfer the artificial chromosomes into each of these strains, two transformants (or two independent spores of each of the double-mutant strains) of each of these yeast strains were mated with four different kar1 strains (SPENCER et al. 1994 ) that harbor artificial chromosomes (two with TG_1–3, UCC2340-A1 and UCC2340-A3; two without TG_1–3, UCC2340-C1 and UCC2340-C4), which resulted in eight matings. After >8 hr, the eight mating mixes were spread onto YC-his-ura-trp + can plates (or YC-his-ura-ade + can for the double-mutant Ade⁺ strains) to select for the transfer of the artificial chromosome. At least 8 isolates of each mating mix were patched onto the same type of plate, grown for 2–3 days, and then replicaplated onto a panel of dropout media to check genetic markers. At least 8 isolates of each artificial chromosome-containing strain then were streaked for single colonies (resulting in at least 32 isolates per mutant strain for analysis: 16 from each of the two transformants/spores—4 isolates of the four artificial chromosomes). Colonies from these 32 isolates were used to inoculate YC-his-ura medium containing 3% raffinose and then grown 12–20 hr (until early log phase). A total of 10 µl was used to determine the cell density of each culture by plating dilutions onto YC plates. Then uracil and galactose were added to the liquid cultures to final concentrations of 60 µg/ml and 3%, respectively, and after 8 hr, cells from each liquid culture were plated individually onto YC-his + FOA plates to select for the HIS3 gene and against the URA3 gene (BOEKE et al. 1984 ). The colonies were counted and then replica plated to YC-his-met plates. The secondary colony counts (Met⁺ colonies) were subtracted from the YC-his + FOA counts to generate the number of events (His⁺Met^- FOA^R) for that culture. The counts for each of the two groups of 16 cultures (with and without TG_1–3) then were analyzed. Strains were grown at either 23° (cdc13-1, stn1-13, and all double-mutant strains) or 30° (all other single-mutant strains). Other assays were done with the cdc13-1 and stn1-13 single-mutant strains: instead of 8 hr at 23° after the addition of galactose, strains were incubated for 4 hr at either 30° or 37°, respectively, and another 4 hr at 23°.

View larger version (37K): [in this window] [in a new window] [Download PPT slide]   Figure 1. The capping assay detects both homologous recombination events (A) and nonhomologous end-joining (B) events at native telomeres. Both artificial chromosomes have a single centromere (CEN4), one origin of replication (ARS), and three genetic markers, URA3, MET15, and HIS3. The chromosomes differ by the sequence that flanks the HO site: One has an 81-bp tract of the TG1–3 sequence on the HIS3 side of the HO cleavage site (A), and the other artificial chromosome does not (B). Upon addition of galactose to the medium, the HO endonuclease is produced, and HO cleaves the artificial chromosome, releasing an 7.5-kb fragment that contains the HIS3 gene. After HO cleavage, the assay selects for the capture of the HIS3 fragment onto a native telomere. Capture may result from homologous recombination between the native telomeric repeats and the TG1–3 tract on the artificial chromosome (A) or a NHEJ event between the native chromosome end and the cleaved end of the HIS3 fragment (B). NHEJ events also can occur in A, but because the non-TG chromosome (B) cannot detect homologous recombination events, the combination of A and B allows us to evaluate the sequence dependency of the capture events. Both types of events lead to cells that may be genetically selected because they are His+ (which indicates the maintenance of the HIS3 gene), Met- (due to the loss of the MET15 gene), and FOA resistant (FOAR, by the loss of the URA3 gene). The frequencies of homologous (A) and nonhomologous (B) events in wild-type cells (UCC2330) are indicated in the bottom corners.

View larger version (39K): [in this window] [in a new window] [Download PPT slide]   Figure 2. The capping assay detects events that occur at telomeres. The analysis of isolates selected as His+ Met- FOAR cells from capping assays (as in Fig 1, Fig 3, and Fig 4) is shown. All strains in Fig 2 were passaged >60 generations before the assay, so telomere lengths should have stabilized. (A) If a native chromosome that contains a Y' element captures the HIS3 fragment, then PCR can amplify across the capture junction. Possible Y' PCR products using one primer in the Y' element (Y') and one in the HIS3 fragment (either "a" or "b") are shown. The sizes of the PCR products between Y' + a and Y' + b are shown with dashed lines because the telomeric repeat sequence can undergo internal recombination after the HIS3 fragment capture, making its size somewhat variable. (B) Y' PCR products, using either PCR set A or B, from six wild-type isolates are shown. Isolates were derived from assays with UCC2330, and all were derived from assays using the artificial chromosome with the TG1–3 tract. (C) Y' PCR products, using PCR set A, from five cdc13-1 isolates and five yku80 isolates are shown. Isolates were derived from assays with UCC2356 and UCC2334, respectively, and all were derived from assays using the artificial chromosome with the TG1–3 tract. (D) Six wild-type isolates (W1–W6, derived from assays with UCC2330), four from cdc13-1 cells (C1–C4, from UCC2356), and four from mre11 cells (M1–M4, from UCC2336) were analyzed by pulsed-field gel electrophoresis: the ethidium bromide (EtBr)-stained CHEF gel was then Southern blotted and probed for the HIS3 gene. One-half of the isolates were derived from assays using the artificial chromosome with the TG1–3 tract (+TG), and one-half without it (-TG). The lane marked "pre" represents a strain with the uncleaved (65 kb) precursor artificial chromosome (indicated with the arrowhead). Isolate W5 is one example of a NHEJ capture event that likely occurred at a nontelomeric location (see RESULTS).

View larger version (29K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Cdc13p, Ku, Tel1p, and Stn1p protect the telomere from homologous recombination events. The distributions of frequencies of telomere capture events are shown. The assay was performed as described in MATERIALS AND METHODS with each of the following yeast strains, using artificial chromosomes with or without TG1–3 tracts, to monitor homologous recombination (A) and NHEJ events (B), respectively. Strains used are as follows (complete genotypes are in Table 1): wild type (UCC2330), yku70 (UCC2333), yku80 (UCC2334), cdc13-1 (UCC2356), stn1-13 (UCC2446), tel1 (UCC2400), rif1 (UCC2459), rif2 (UCC2460), sir2 (UCC2842), sir3 (UCC2843), yku70 tel1 (UCC2745), yku80 tel1 (UCC2746), and cdc13-1 tel1 (UCC2747). The percentage of isolates that had no capture events is shown as "% with no events" for that strain. For strains that had one or more isolates without a capture event, the frequency for each isolate is shown in the scatterplot. The line within the scatterplot represents the median frequency; when the median equals zero (Table 3), the line is shown at 10-8. The box graphs are used to present the frequencies when all isolates of a strain had capture events. The dark line in the middle of each box represents the median frequency for each strain; the box shows the distribution of 25% of the frequencies above and below the median (75 and 25% quartiles, respectively), and the lines that extend vertically above and below the box indicate the maximum and minimum frequencies that were detected. The limit of detection in these assays is 4 x 10-8.

View larger version (61K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Epistatic relationships reveal differences between the capping mutant strains.(A) Frequencies of telomere capture events via homologous recombination are presented as graphs, as described in Fig 3. Several single-mutant strains and data are the same as those in Fig 3; other strains are as follows (complete genotypes are in Table 1 and Table 2): mre11 (UCC2336), rad50 (UCC2399), xrs2 (UCC2461), cdc13-1 mre11 (UCC2758), cdc13-1 rad50 (UCC2757), cdc13-1 xrs2 (UCC2759), tel1 mre11 (UCC2749), tel1 rad50 (UCC2748), tel1 xrs2 (UCC2750), rad52 (UCC2335), rad51 (UCC2430), rad1 (UCC2719), yku70 rad52 (UCC2391), yku70 rad51 (UCC2789), yku70 rad1 (UCC2809), yku80 rad52 (UCC2392), yku80 rad51 (UCC2790), yku80 rad1 (UCC2810), cdc13-1 rad52 (UCC2760), cdc13-1 rad51 (UCC2791), cdc13-1 rad1 (UCC2811), tel1 rad52 (UCC2744), tel1 rad51 (UCC2780), and tel1 rad1 (UCC2774). (B) The telomere lengths of viable double-mutant strains were analyzed by Southern analysis. Genomic DNAs were digested with XhoI, run on a 1.25% agarose gel, and hybridized with a probe to Y' elements. The strains on the blot are as follows (complete genotypes are in Table 1 and Table 2): wild type (UCC2330), cdc13-1 (UCC2356), yku70 (UCC2333), tel1 (UCC2400), mre11 (UCC2336), cdc13-1 mre11 (UCC2758), cdc13-1 rad50 (UCC2757), cdc13-1 xrs2 (UCC2759), cdc13-1 tel1 (UCC2747), tel1 mre11 (UCC2749), yku70 tel1 (UCC2745), and yku80 tel1 (UCC2746). Two different isolates of each of the double-mutant strains are shown. All of the strains were passaged for >30 generations before genomic DNA was isolated; additional passages of 60 generations showed no differences in telomere lengths from those illustrated in Fig 4B (data not shown).

Statistical analysis:
The data for each strain as generated by the assay are summarized for both homologous recombination and NHEJ capture events in Table 3. Median frequencies of capture events were determined for all strains. [Rates of capture events were not calculated because the underlying assumptions of computing rates from frequency measurements, as in a fluctuation analysis (LEA and COULSON 1948 ), were not upheld in our experiments. For instance, the rates of capture events are not constant due to our induction of the HO endonuclease and subsequent interruption of growth (ROSCHE and FOSTER 2000 ).] Because the limit of detection of the assay (4 x 10^-8) is close to the average frequency of several strains, these strains showed one or more isolates that had no capture events ("% with no events" in Fig 3 and Fig 4). The P values for pairwise comparisons between wild-type and mutant strain (shown in Table 3) and other pairs of mutant strains are generated using the Wilcoxon rank-sum test. Group comparisons were done using one-way ANOVA. Prism (Graphpad Software) was used to generate Fig 3 and Fig 4.

Molecular biology techniques:
Agarose gels were blotted onto Hybond N+ nylon membranes (Amersham, Buckinghamshire, UK) in 0.4 M NaOH, 1.5 M NaCl for >4 hr. Blots were probed with a digoxigenin-labeled Y' fragment (SINGER et al. 1998 ) or a HIS3 fragment (below) per manufacturer's instructions (Roche). The 303-bp HIS3 fragment used as a Southern probe was generated from a PCR off of pRS303 (SIKORSKI and HIETER 1989 ), using the oligos HIS3N⁺ (TGAGCAGGCAAGATAAACGAAGGC) and midHIS3 (GTGTGATGGTCGTCTATGTGTAAG).

Agarose-embedded DNA plugs, pulsed-field gel electrophoresis, and Southern blotting of contour-clamped homogeneous electric field (CHEF) gels were done according to the manufacturer's instructions (Bio-Rad; model CHEF-DRII). Gels were run at 6 V/cm for 27 hr with 7- to 170-sec switch times and alkaline blotted as above for 24 hr.

Y' PCR experiments used one primer within the Y' element (Y'XhoI: GATACGGTCTTTGTGGAAGCGCTCG) and one primer within the HIS3 fragment: either primer a (TTTCCCAGTCACGACGTTGT) or primer b (GAGTATACGTGATTAAGCAC) with 1 µl of a diluted 1/100 genomic DNA preparation as template in a 25-µl reaction. PCRs were run as follows: 94°/2 min, polymerase addition, and then 27 cycles of 94°/1 min, 57°/45 sec, 72°/1.5 min.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A genetic assay for detecting a compromised chromosomal cap:
To detect "uncapped" telomeres, we developed an assay based on the idea that an unprotected chromosome end would be vulnerable to recombination with an induced DSB end. The DSB is created in an artificial linear chromosome by an inducible HO endonuclease. On one side of the HO cleavage site are two selectable markers (URA3 and MET15), a centromere, and the only ARS (origin of replication) within the chromosome; on the other side is the HIS3 gene (Fig 1A and Fig B). After cleavage with HO, the HIS3 chromosomal fragment has a stable telomere on one end (the native end of the artificial chromosome) and the DSB end on the other. Because the HIS3 fragment lacks an ARS and is selected against rejoining with the other half of the artificial chromosome (Fig 1), the HIS3 fragment's only option for propagation is a capture event by a native chromosome. Only the DSB end of the HIS3 fragment is a substrate for a capture event because the opposite end is a telomere, and the rest of the HIS3 fragment lacks sequence homology with the yeast genome, which eliminates its opportunity for homologous recombination. Thus, the HIS3 fragment will most likely be captured by an unprotected native chromosome end.

By our definition, an unprotected chromosome end is susceptible to recombination events that we can detect by capture events in the assay. The capture events may be either sequence dependent or sequence independent. In the absence of telomerase in S. cerevisiae, telomeres and subtelomeric regions are maintained by homologous recombination (LUNDBLAD and BLACKBURN 1993 ; TENG and ZAKIAN 1999 ); a similar mechanism is likely used in some human immortalized and tumor cell lines (BRYAN et al. 1997 ). However, in some organisms (e.g., mammals and S. pombe; NAKAMURA et al. 1998 ; BAILEY et al. 1999 ) when telomeres are severely compromised, the chromosome ends typically undergo NHEJ events with one another or with other DNA ends.

Therefore, to detect both homologous recombination and NHEJ at telomeres, two artificial chromosomes were created for use in the capping assay. Homologous recombination at telomeres is detected with an artificial chromosome that has a stretch of the TG_1–3 sequence adjacent to the HO cleavage site (Fig 1A). This internal TG_1–3 sequence tract is oriented so that it cannot act as a functional telomere, but it does have the potential to recombine with the telomeric DNA of a native chromosome end, thus allowing the HIS3 fragment to be captured by the end of a chromosome (Fig 1A). (This recombination event is technically homeologous because of the heterogeneous nature of the telomeric DNA repeats of S. cerevisiae, but for simplicity, we describe these events as homologous.) The other artificial chromosome has no homology to the rest of the genome, so that capture of the HIS3 chromosomal fragment will most likely be by NHEJ or end-to-end fusion (Fig 1B). Fusion events also can occur in the case of the artificial chromosome with the TG_1–3 sequence tract (Fig 1A), but because the non-TG artificial chromosome cannot detect homologous recombination events, the combination of these two constructs allows us to evaluate the sequence dependency of the mechanism by which telomeres become compromised. Therefore, using both of the artificial chromosomes should permit detection of unprotected chromosome ends that succumb to either homologous recombination or NHEJ.

In a wild-type cell, telomeres are expected to be "uncapped" infrequently, if at all. Indeed, as measured in our assay, the capture of the HIS3 chromosomal fragment by homologous recombination or NHEJ occurred at a very low frequency in wild-type cells (Fig 1, Table 3), on the order of 10^-8. These low frequencies suggested that telomeric ends are normally well protected against recombination.

During the development of the assay, we also examined the frequencies of capture of the HIS3 fragment without HO cleavage of the TG_1–3-containing artificial chromosome. Because some of the mutant strains showed increased frequencies of telomeric capture after HO cleavage (described later), these data were compared to assays done without HO cleavage. In the absence of HO cleavage, there were fewer capture events, and the statistical distribution of the frequencies was broader than the distribution after HO cleavage (data not shown). This difference may be due to increased accessibility of the broken (HO-cut) end for recombination with a compromised telomeric end, which makes the capture of the HIS3 fragment a more reliable event. Alternatively, a compromised telomeric end may invade the TG_1–3 tract on the artificial chromosome during growth of the culture and, via a nonreciprocal event, capture the HIS3 fragment. Either of these scenarios represents a sequence-dependent event at a compromised chromosome end and is discussed in the context of homologous recombination. The higher reproducibility (narrower distribution of frequencies) of the assay done in the presence of HO led us to use HO cleavage as part of all subsequent assays that are presented.

Different chromosome ends can become compromised:
To test whether the capture events in this genetic assay were truly indicative of capture at chromosome ends, we tested three predictions. First, to examine whether the HIS3 chromosomal fragment was physically linked to a native telomere after capture, a PCR-based method was designed. Because many chromosomes in S. cerevisiae have subtelomeric Y' elements that are conserved in sequence (LOUIS and HABER 1992 ), primers were designed for PCR across the junction where the HIS3 fragment had been captured by a native, Y'-adjacent telomere (Fig 2A). By PCR, >90% of all capture events that occurred via homologous recombination (TG_1–3 sequence tract on artificial chromosome) were adjacent to a Y' element (Fig 2B and Fig C, and data not shown). Many, but not all, telomeres in S. cerevisiae have subtelomeric Y' elements immediately adjacent to the telomeric repeats (LOUIS 1995 ); thus, a lack of PCR product simply may be due to the absence of a Y' element on the telomere where the HIS3 fragment was captured. Several of the PCR products were sequenced and revealed that the remaining internal TG_1–3 sequence varied in length from 10 to 300 bp (data not shown). These data indicate that virtually all of the homologous recombination capture events occurred at chromosome ends.

However, only 15% of NHEJ capture events in wild-type cells produced a PCR product indicative of capture at a Y'-associated chromosome end (data not shown). It was possible that a lack of PCR product was due to the loss of the primer sites on the HIS3 fragment by degradation from the HO end after cleavage. However, when different primers located within the HIS3 coding sequence (which must be present because of the His⁺ selection) were used, PCR products indicative of capture at a Y' telomere were still not detected in >80% of the NHEJ capture events. Thus the absence of PCR product may be due to attachment to a non-Y' telomere or to capture of the HIS3 fragment at a nontelomeric genomic location. As a result of the uncertainty in the precise genomic locations of these capture events, we have chosen not to use these results as true indicators of compromised telomeres. However, we do use the non-TG data to compare with the homologous recombination data set; it serves as a measure of "background" capture events in a particular strain.

Second, if any telomere can become compromised and lose its protective cap, then the capture events should be random, and the HIS3 fragment should be captured by any chromosome end. Examination of chromosomes by pulsed-field gel analysis revealed that the HIS3 gene was distributed among many different chromosomes in the various isolates (Fig 2D and data not shown; >200 isolates examined).

Third, because the capture events occurred on several different chromosome ends, we examined the "new telomere" of the native chromosome after the capture of the TG_1–3-containing HIS3 fragment (Fig 2A). Indeed we found that all captured HIS3 fragments in >100 isolates had become the new chromosomal telomeres; when probed by Southern analysis, each isolate of a capture event produced a heterogeneous length band that is unique to telomeric DNA (data not shown). Similar analyses with NHEJ capture events also indicated that virtually all these HIS3 fragments had become the new telomeres.

Taken together, these data indicate that the assay using homologous recombination reliably detects capture events that occur at chromosome ends and that most, if not all, telomeric ends in the yeast genome can become compromised and participate in a capture event.

Cdc13p, Ku, and Stn1p protect the chromosome end:
The capping assay described above was applied to strains containing mutant versions of several genes that are associated with telomeres. The first set of genes included those encoding the telomeric DNA-binding proteins, Ku and Cdc13p, and the Cdc13p-associated protein, Stn1p. The yku70, yku80, cdc13-1, and stn1-13 strains showed dramatic increases in homologous recombination-based capture events at telomeres compared to wild-type cells (Fig 3A and Table 3). The yku70, yku80, and cdc13-1 mutations resulted in increases in the median frequencies of approximately three orders of magnitude over wild-type levels. By giving the cdc13-1 cells a 4-hr pulse at the restrictive temperature (30°), the frequency of homologous recombination at telomeres increased another 2-fold. Every isolate with one of these mutations (yku70, yku80, or cdc13-1) exhibited capture events in the assay. In contrast, at the permissive temperature (23°) the stn1-13 allele showed a small increase from wild-type levels, but 13% of the isolates exhibited no capture events. After a pulse at the restrictive temperature (37°), an increase in telomeric homologous recombination of >100-fold over wild-type levels was observed, and all isolates exhibited capture events. These data suggest that Ku and Cdc13p play roles in protecting the chromosome end. Stn1p plays a role as well, but it is unclear whether it is less important than its partner, Cdc13p, in capping the chromosome end or that the stn1-13 allele was not as penetrant as the yku and cdc13-1 alleles in the assay.

These mutations also were assayed for their abilities to protect the telomeric end against NHEJ events. None of the single-mutant strains showed large increases in frequencies of NHEJ (Fig 3B and Table 2); most of the strains had frequencies within threefold of wild-type levels, although the cdc13-1 strain had an increase of approximately fivefold over wild type. Strains with yku mutations showed a lower level of NHEJ events than did wild type, which likely reflects Ku's importance in NHEJ (reviewed in TUTEJA and TUTEJA 2000 ). The differences in these frequencies compared with those of the corresponding TG_1–3-mediated events offer further support that the TG_1–3-mediated events are due to homologous recombination and not to end-to-end fusions. Thus, the assay revealed that Cdc13p, Ku, and Stn1p have roles in protecting S. cerevisiae chromosome ends against homologous recombination.

Tel1 contributes to telomere protection, but shortened telomere length does not correlate with a loss of capping efficiency:
We next examined several telomere-associated mutations that affect the lengths of the telomeric DNA tracts. Like the yku alleles, null mutations of MRE11, RAD50, and XRS2, whose gene products comprise the MRX nuclease complex (PETRINI et al. 2001 ), and TEL1, which encodes a kinase similar to ATM (ataxia telangiectasia mutated; GREENWELL et al. 1995 ; MORROW et al. 1995 ; SMILENOV et al. 1997 ; MALLORY and PETES 2000 ), result in short telomeric DNA tracts in S. cerevisiae (LUSTIG and PETES 1986 ; KIRONMAI and MUNIYAPPA 1997 ; BOULTON and JACKSON 1998 ; NUGENT et al. 1998 ). The MRX complex may recruit telomerase to the telomere (TSUKAMOTO et al. 2001 ) or perhaps recruit or act as a 5'-3' exonuclease to create a single-stranded telomeric repeat template for binding of other telomere factors (NUGENT et al. 1998 ; DIEDE and GOTTSCHLING 2001 ). Tel1p may phosphorylate members of the MRX complex (RITCHIE and PETES 2000 ; MYUNG et al. 2001 ; USUI et al. 2001 ); a related ATM kinase has been shown to phosphorylate Nbs1, the mammalian homolog of Xrs2p (KASTAN and LIM 2000 ).

A tel1 mutation resulted in an increased frequency of capture via homologous recombination of 150-fold (Fig 3A and Table 3) but had little effect on NHEJ frequencies (Fig 3B and Table 3). In contrast, the null mutations of the MRX complex generated near-wild-type frequencies for homologous recombination (Fig 4A and Fig B, and Table 3). Because the capture events require recombination and the MRX genes are involved in recombination, it is possible that the mrx mutations prevented the appearance of capture events. However, when combined with cdc13-1 and tel1 mutations that did increase the frequency of capture events, the double-mutant strains with mrx mutations still led to increased frequencies (discussed later; Fig 4A and Table 3). These results suggest that TEL1 is critical to chromosomal cap integrity but that the MRX complex is not. Furthermore, although the yku and tel1 mutations result in both short telomeres and an increased frequency of homologous recombination, the mrx strains have short telomere repeat lengths yet retain their capping ability (Fig 4A and Fig B). Thus, a short average telomere length is not sufficient to compromise the chromosomal cap.

Two mutations with longer than normal telomeres, rif1 and rif2, also were examined. Rif1p and Rif2p interact at telomeres with Rap1p, the essential DNA-binding protein, and all are required for the maintenance of normal telomere length (HARDY et al. 1992 ; WOTTON and SHORE 1997 ). The frequency of capture events in rif1 or rif2 strains by homologous recombination was not dramatically different from wild type (Fig 3A and Fig B, and Table 3). Curiously, there was an increase in these mutants in the NHEJ capture events. By the homologous recombination assay, cells with extended telomeric DNA tracts do not inherently have a problem in capping their chromosomes.

Telomeric silencing does not correlate with telomere capping:
The YKU genes are also required for transcriptional silencing of telomere-proximal genes (BOULTON and JACKSON 1998; NUGENT et al. 1998 ), which is mediated by the Sir proteins. As part of their silencing function, the Sir proteins inhibit recombination at the silent mating cassette and rDNA loci (GOTTLIEB and ESPOSITO 1989 ; LOO and RINE 1995 ). Therefore we examined whether SIR2 or SIR3 might play a similar role at the telomere, specifically in the context of telomere capping. In sir2 or sir3 mutant strains, there was no difference, compared to wild type, in the frequency of capture events in either the homologous recombination or the NHEJ assay (Fig 3A and Fig B). [The sir mutant strains lacked HML and HMR to avoid the potential complications of losing silencing at these loci: the increased availability of HO cut sites at HML and HMR (HERSKOWITZ et al. 1992 ) and the changes in recombination that are regulated by concurrent a and gene expression (ASTROM et al. 1999 ; LEE et al. 1999 ).] Thus, it appears that telomeric silencing is not important for capping the end of the chromosome in S. cerevisiae.

Ku and Cdc13p define different telomere capping pathways:
Mutations in CDC13, YKU, and TEL1 created the most dramatic increases in frequencies of telomere capture events (Fig 3). To determine whether they acted in the same or different capping pathways, epistatic relationships among these genes were examined. In our strain background, the double-mutant combination of cdc13-1 yku senesced 25 generations after sporulation, although infrequent survivors did arise in culture; these observations are consistent with growth defects reported previously for this mutant combination (NUGENT et al. 1998 ; POLOTNIANKA et al. 1998 ). Each of the other double-mutant combinations was viable in our strain background.

The cdc13-1 tel1 strain showed high frequencies of telomeric capture by homologous recombination that were very similar to those of the cdc13-1 strain (Fig 3A and Table 3). In addition, there was no significant difference between the frequencies of capture events via NHEJ between cdc13-1 and cdc13-1 tel1 strains (Fig 3B and Table 3). Thus tel1 did not contribute any additional uncapping of telomeres in the cdc13-1 mutant strain, suggesting that Tel1p might normally affect telomere protection through Cdc13p.

In contrast, combinations of yku70 tel1 or yku80 tel1 showed the highest frequencies of telomeric capture via homologous recombination that were detected (Fig 3A and Table 3), and the frequencies were significantly higher than those of either of the single-mutant strains (P < 0.001 between yku tel1 and yku or tel1). These double-mutant strains exhibited a modest increase in the frequencies of NHEJ compared to wild-type or single mutants (Fig 3B and Table 3). The synergistic effect of these mutations on telomeric homologous recombination suggests that Ku and Tel1p act via different pathways to cap the telomere.

The combinations of these double-mutant strains suggest that chromosomal caps are created from contributions of two genetic pathways, one mediated by CDC13 and the other by YKU. When both are compromised, telomeres are unstable and yield the lethal/senescent phenotype. Tel1p may act through the CDC13 pathway, perhaps modulating Cdc13p capping activity.

Further differences between compromised caps are revealed with recombination mutations:
The large increases in telomere capture events detected in cdc13-1, yku, and tel1 mutations all occurred through the homologous recombination assay (Fig 3A and Table 3). However, we were curious whether there was a qualitative change in the cap for each mutant, which might then lead to distinct recombination pathways being used during a capture event. Therefore, using our capping assay with these three mutations, we tested the requirements for genetic pathways known to affect recombination at a DSB.

The first recombination component tested was Rad52p, a protein required for most homologous recombination in S. cerevisiae (PETUKHOVA et al. 2001 ). The rad52 mutation, in combination with yku70, yku80, cdc13-1, or tel1 mutations, resulted in significantly decreased frequencies of telomeric capture, although the frequencies were not reduced to wild-type levels (Fig 4A and Table 3; P < 0.001 for rad52 double-mutant vs. each single-mutant strain). The frequencies of NHEJ were not dramatically changed from the levels of wild-type or the single-mutant strains (Table 3; because all experiments examining NHEJ resulted in distributions that were similar to those seen in Fig 3B, analyses of NHEJ events in the mutant strains in Fig 4A are summarized only by the statistical data in Table 3). These results indicate that Rad52p contributes significantly, but not entirely, to the homologous recombination events at telomeres, as expected, but that another pathway is also involved.

Mutations in RAD1 were also examined. Rad1p binds Rad10p to form an endonuclease that cleaves the 3' single-stranded "flap" that is formed during recombination processing; the complex is required for some recombination events, such as single-stranded annealing, which occurs between repeated sequences (PAQUES and HABER 1999 ). All double-mutant strains with rad1 resulted in lower frequencies of telomeric capture via homologous recombination (Fig 4A and Table 3; P < 0.001 for rad1 double-mutant vs. each single-mutant strain). This suggests that Rad1p also is involved in telomere capture events, perhaps using a single-stranded annealing mechanism.

Rad51p is a homolog of the bacterial RecA protein; in S. cerevisiae, Rad51p initiates strand exchange after a double-stranded break and is required for some gene conversion events (PAQUES and HABER 1999 ; PETUKHOVA et al. 2001 ). Examination of the double-mutant strains with rad51 mutations reveals a difference between the capping proteins: yku rad51 and tel1 rad51 showed an increase in frequency of telomeric capture by homologous recombination over the single-mutant frequencies, whereas the frequencies for cdc13-1 rad51 decreased from the cdc13-1 levels (Fig 4A and Table 3; P < 0.001 for rad51 double-mutant vs. each single-mutant strain). The decrease in homologous recombination frequency observed in the cdc13-1 rad51 strain suggests that Rad51p is involved in the recombination events in cdc13-1 strains. In contrast, it appears that Rad51p was competing with, or otherwise inhibiting, a recombination mechanism in the yku and tel1 strains (discussed later). Taken together with the single-mutant data, the chromosome ends seem qualitatively different among these three telomere-capping mutations.

Finally, we examined contributions by the MRX genes, which play a role in recombination as well as telomere length regulation (mentioned above). Sporulations of several different diploids revealed that the double- or triple-mutant combinations of yku mrx were senescent shortly after their creation (25 generations; data not shown), although infrequent survivors did arise in the cultures. This effect is similar to the lethality observed with the yku cdc13-1 strain (described above) and may reflect the contribution of the MRX complex to the loading of factors at the telomere (NUGENT et al. 1998 ; DIEDE and GOTTSCHLING 2001 ; TSUKAMOTO et al. 2001 ). All of the other double-mutant combinations among cdc13-1, yku, tel1, and mrx were viable in our strain background, including our cdc13-1 mrx strains. While not identical, these results are similar to previous observations in other strain backgrounds, where yku mrx strains showed a synthetic growth defect (NUGENT et al. 1998 ; MOREAU et al. 1999 ) and cdc13-1 mrx strains were additionally temperature sensitive (NUGENT et al. 1998 ); senescence was not discussed in these other studies.

The tel1 mrx and cdc13-1 mrx strains were then tested in the capping assay. The tel1 mrx strains had a modest decrease in telomere capture frequency compared to the tel1 strain in the homologous recombination assay (Fig 4A and Table 3; P values vs. tel1: P < 0.0001 for tel1 rad50 and tel1 xrs2; P = 0.127 for tel1 mre11). On the other hand, compared to cdc13-1 alone, the cdc13-1 mrx combination showed no difference via homologous recombination (Fig 4A, Table 3; P value for the group = 0.249; P values vs. cdc13-1: P = 0.673 for cdc13-1 mre11 and cdc13-1 xrs2; P = 0.075 for cdc13-1 rad50). These results indicate that when Cdc13p is compromised, the MRX complex does not affect the protective cap nor the mechanism of capture via homologous recombination, but in the absence of Tel1p, the MRX complex seems to make a small contribution to the homologous recombination events. Thus, just as the analysis with rad51 revealed, combinations with rad50 (or mre11 or xrs2) mutations suggest that telomeres are qualitatively different among yku, tel1, and cdc13-1 strains.

Telomere lengths of double-mutant strains do not correlate with telomere recombination:
We examined the telomere lengths of several double-mutant strains as another method to identify relationships among these telomere components (Fig 4B). As previously shown, telomeres of yku70, tel1, and mre11 strains have telomeric DNA repeats significantly shorter than those of wild type (LUSTIG and PETES 1986 ; BOULTON and JACKSON 1998 ; NUGENT et al. 1998 ); thus, double-mutant strains of yku tel1 and tel1 mre11 also have short telomeric repeats (RITCHIE and PETES 2000 ). However, the telomeres of the cdc13-1 mrx and cdc13-1 tel1 strains show stable telomeric repeats of an intermediate length between the longer repeats of the cdc13-1 strain and the shorter ones of mrx and tel1 strains (Fig 4B). These results are inconsistent with the similarities observed in the capping assay between the cdc13-1 single- and double-mutant strains (Fig 3A and Fig 4A); the cdc13-1 double-mutant strains show the same increased frequency of homologous recombination events as the cdc13-1 strains. These results provide additional evidence that telomere length does not correlate with capping efficiency.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have developed a quantitative method to measure when a telomere becomes "uncapped." Because telomeric DNA is normally refractory to recombination, the assay was designed to detect rare events in which telomeric DNA is unprotected and can enter into DNA recombination events. We used the assay to determine whether a number of known telomeric components contributed to chromosomal capping. The assay revealed that the telomeric DNA-binding proteins Cdc13p and Ku provide protection against homologous recombination and that they define two pathways that are used to protect a chromosome end.

The protective role of Ku at telomeres appears to be evolutionarily conserved among eukaryotes. Ku was identified as a capping component in S. cerevisiae by the capping assay (Fig 3A) and in mammals by immunofluorescence microscopy (BAILEY et al. 1999 ). Ku also is required to maintain normal telomere length in both S. cerevisiae (BOULTON and JACKSON 1996 ; PORTER et al. 1996 ) and S. pombe (BAUMANN and CECH 2000 ). In contrast, when S. pombe cells are stressed by nitrogen starvation in the absence of the telomere-binding protein Taz1, Ku is required for the formation of chromosome end-to-end fusions (FERREIRA and COOPER 2001 ). The two roles of Ku at telomeres—in both protection and fusions—are seemingly at odds: How might this complex act at telomeres?

These opposing actions may reflect regulation of Ku that may occur as a result of cell cycle timing, competition with other capping activities, alternative binding partners—either DNA (blunt ends, single stranded, etc.) or protein (e.g., Taz1 or the DNA-dependent protein kinase)—or perhaps species-specific differences. Alternatively, Ku may change the telomeric structure, perhaps in preventing the formation of a single-stranded overhang (discussed later), so that the absence of Ku enables homologous recombination at telomeres. This latter idea is consistent with the capping assay, which revealed that telomeric homologous recombination events increased in yku strains, and NHEJ events were unaffected. Because the assay provides a recombination substrate for both homologous and NHEJ events, this difference could be detected, whereas immunofluorescence studies do not distinguish between these events; the absence of Ku may stimulate homologous events at mammalian telomeres as well. Another possibility is that the presence of telomerase inhibits NHEJ events at the telomere, whether or not Ku is present. The deletion of EST1, a telomerase-associated component, led to end-to-end fusion events in S. cerevisiae (HACKETT et al. 2001 ); perhaps a lack of telomerase activity creates a compromised chromosomal cap that leads to increased end-to-end fusions in some cell types.

Our assay also showed that Cdc13p, a telomeric DNA-binding protein, protects against homologous recombination at telomeres (Fig 3A). The cdc13-1 strain was defective in protection at the permissive temperature (23°), and the defect was enhanced after a pulse at 30°. Our results are in agreement with earlier characterization of the cdc13-1 mutation, which showed that it had higher levels of mitotic recombination near telomeres (CARSON and HARTWELL 1985 ; HARTWELL and SMITH 1985 ). Because of genetic and physical interactions between Cdc13p, telomerase, and Stn1p, Stn1p has been suggested to play a role both in the protection of the end and in telomere lagging-strand synthesis (GRANDIN et al. 2000 , GRANDIN et al. 2001B ; CHANDRA et al. 2001 ). In our assay we found that Cdc13p plays a crucial role in protection, but the importance of Stn1p is less clear. Although the Cdc13-1 mutant protein binds DNA (HUGHES et al. 2000 ) and may bind Stn1p (CHANDRA et al. 2001 , but see WANG et al. 2000 ), the telomeres in a cdc13-1 strain are still susceptible to homologous recombination at the permissive temperature (Fig 3A), which suggests that Cdc13p-mediated telomere protection may require something more than Stn1p.

The TEL1 gene, an ATM kinase homolog, also had a pronounced effect on capping the telomere (Fig 3). Tel1p's role in protection is likely regulatory, acting, at least in part, through Cdc13p. Our epistasis analysis revealed that a cdc13-1 tel1 strain exhibits a similar median telomeric homologous recombination frequency ^<