Telomere Binding of Checkpoint Sensor and DNA Repair Proteins Contributes to Maintenance of Functional Fission Yeast Telomeres

Corresponding author: Toru M. Nakamura, MB3, The Scripps Research Institute, 10550 N. Torrey Pines Rd., La Jolla, CA 92037., nakamut{at}scripps.edu (E-mail)

Communicating editor: G. SMITH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Telomeres, the ends of linear chromosomes, are DNA double-strand ends that do not trigger a cell cycle arrest and yet require checkpoint and DNA repair proteins for maintenance. Genetic and biochemical studies in the fission yeast Schizosaccharomyces pombe were undertaken to understand how checkpoint and DNA repair proteins contribute to telomere maintenance. On the basis of telomere lengths of mutant combinations of various checkpoint-related proteins (Rad1, Rad3, Rad9, Rad17, Rad26, Hus1, Crb2, Chk1, Cds1), Tel1, a telomere-binding protein (Taz1), and DNA repair proteins (Ku70, Rad32), we conclude that Rad3/Rad26 and Tel1/Rad32 represent two pathways required to maintain telomeres and prevent chromosome circularization. Rad1/Rad9/Hus1/Rad17 and Ku70 are two additional epistasis groups, which act in the Rad3/Rad26 pathway. However, Rad3/Rad26 must have additional target(s), as cells lacking Tel1/Rad32, Rad1/Rad9/Hus1/Rad17, and Ku70 groups did not circularize chromosomes. Cells lacking Rad3/Rad26 and Tel1/Rad32 senesced faster than a telomerase trt1 mutant, suggesting that these pathways may contribute to telomere protection. Deletion of taz1 did not suppress chromosome circularization in cells lacking Rad3/Rad26 and Tel1/Rad32, also suggesting that two pathways protect telomeres. Chromatin immunoprecipitation analyses found that Rad3, Rad1, Rad9, Hus1, Rad17, Rad32, and Ku70 associate with telomeres. Thus, checkpoint sensor and DNA repair proteins contribute to telomere maintenance and protection through their association with telomeres.

CHECKPOINT and DNA repair pathways are crucial to the progression of the normal cell cycle. Without them, cells cannot maintain a stable genome, and genetic instability can lead to cell death, cancer, and other genetic disorders (KHANNA and JACKSON 2001 ). The DNA replication checkpoint prevents mitosis until DNA is replicated completely. Similarly, the DNA damage checkpoint inhibits mitosis in response to damaged DNA. Checkpoint controls are highly conserved. Homologs to many checkpoint proteins that were originally identified in yeasts have now been found in multicellular eukaryotes, including humans. Checkpoint proteins form signaling cascades activated by a DNA replication block or DNA damage and subsequently generate the negative constraints on cell cycle progression (RHIND and RUSSELL 1998 ). One form of DNA damage recognized by checkpoint proteins is the double-strand break (DSB). DSBs may be created as a part of cell cycle events, such as DNA replication and meiotic recombination, or by exposure to genotoxic chemicals or ionizing radiation. DSBs are repaired by either homologous recombination or nonhomologous end joining (NHEJ).

Telomeres, the natural ends of linear chromosomes, are maintained by the specialized reverse transcriptase called telomerase. Many proteins bind telomeric DNA and protect it from degradation and recombination. Telomeres pose special challenges to the DNA repair machinery and checkpoint proteins because these DNA ends must be maintained, unlike other internal DSBs, which must be rejoined (BLACKBURN 2001 ). Therefore, one might expect the DNA repair and checkpoint machinery to be excluded from telomeres. Surprisingly, studies from the budding yeast Saccharomyces cerevisiae and mammalian cells suggest that proteins required for NHEJ, such as the Ku70-Ku80 heterodimer and the Mre11-Rad50-Xrs2 complex (Mre11-Rad50-Nbs1 in mammalian cells), are present at telomeres and are needed for normal telomere functions (NUGENT et al. 1998 ; RITCHIE and PETES 2000 ; ZHU et al. 2000 ; TSUKAMOTO et al. 2001 ). Likewise, studies in S. cerevisiae and in the fission yeast Schizosaccharomyces pombe suggest that rather than being denied access to telomeres, the checkpoint sensor proteins, such as budding yeast Mec1 and fission yeast Rad3, seem to recognize telomeres (DAHLEN et al. 1998 ; NAITO et al. 1998 ; MATSUURA et al. 1999 ; RITCHIE et al. 1999 ; LONGHESE et al. 2000 ). Yet, these checkpoint sensors are somehow prevented from activating checkpoints when they interact with the DNA ends at telomeres.

Using the fission yeast S. pombe as a model system, we wished to understand how this apparent alteration in the checkpoint signaling pathways at telomeres is achieved to allow the DNA structure checkpoint proteins to recognize telomeres as the unique DNA ends that should not be repaired. Advantages of the fission yeast system include well-characterized DNA damage responses with high structural and functional conservation to the mammalian system; amenability to genetic, biochemical, and cytological studies; and a small number of telomeres per cell. In addition, the ability of fission yeast to bypass the need for a functional telomere maintenance mechanism by circularizing all chromosomes (NAITO et al. 1998 ; NAKAMURA et al. 1998 ) provides flexibility, not available in any other organisms, in manipulating telomere-related genes without being hindered by cell lethality.

The DNA structure checkpoint responses in S. pombe require a group of six "checkpoint Rad proteins" (Rad1, Rad3, Rad9, Rad17, Rad26, and Hus1), which are thought to function as sensors of DNA replication arrest and DNA damage (CASPARI and CARR 1999 ). Rad1, Rad9, and Hus1 proteins have weak sequence similarity to proliferating cell nuclear antigen (PCNA) and form a ring-shaped complex (CASPARI et al. 2000 ; KAUR et al. 2001 ; GRIFFITH et al. 2002 ). Rad17 protein shows sequence similarity to replication factor C (RFC) proteins (GRIFFITHS et al. 1995 ) and associates with other RFC subunits (KAI et al. 2001 ). Since the RFC complex recruits PCNA onto DNA, it has been proposed that the Rad17 complex loads the Rad1-Rad9-Hus1 complex onto sites of DNA damage. Rad3 is related to human ATM and Rad3-related (ATR) and ataxia telangiectasia-mutated (ATM) protein kinases, the latter of which is mutated in patients with the cancer-prone genetic disorder ataxia-telangiectasia. Rad3 is stably associated with its regulatory subunit Rad26 (EDWARDS et al. 1999 ; WOLKOW and ENOCH 2002 ). Two additional proteins, Crb2 and Cut5, are also implicated in the DNA damage checkpoint (SAKA et al. 1997 ), but they appear to function downstream of the checkpoint Rad proteins. Mutations in rad1, rad3, rad17, and rad26 cause telomere shortening (DAHLEN et al. 1998 ). Therefore, checkpoint sensor proteins appear to have a positive role in maintaining telomere length.

In addition to Rad3, S. pombe cells have another protein kinase related to ATR and ATM called Tel1, and the phenotype of rad3 tel1 illustrates the importance of ATR and ATM family proteins in telomere maintenance. The double-mutant cells have dramatically shortened telomeres, and the cells often lose their telomeres completely and circularize all chromosomes (NAITO et al. 1998 ), much like survivors of a telomerase deletion mutant strain (trt1; NAKAMURA et al. 1998 ). Rad3 and Tel1 might be needed simply for a delay in the cell cycle to allow telomerase to complete telomeric DNA synthesis, with chromosomal fusions being the secondary consequence of the ensuing loss of telomeric repeats. Alternatively, Rad3, Tel1, and other checkpoint proteins might act directly at telomeres to allow cells to distinguish telomeres from other types of DNA ends or to recruit telomerase to telomeres. As the ATR and ATM family kinases also show sequence similarity to the catalytic subunit of DNA-protein kinase (DNA-PK), which binds and is activated by broken DNA ends, Rad3 and Tel1 might bind directly to telomeric DNA. Binding of DNA-PK to telomeric DNA has been reported in mammalian cells (D'ADDA DI FAGAGNA et al. 2001 ).

Studies of telomere length in cells carrying mutations in the DNA damage checkpoint downstream signal transducer proteins support a more direct role for the checkpoint Rad proteins in telomere length maintenance (DAHLEN et al. 1998 ; MATSUURA et al. 1999 ). In S. pombe, the Chk1 and Cds1 protein kinases work downstream of the checkpoint Rad proteins to transmit the signals created in response to DNA damage and replication blocks to the cyclin-dependent kinase Cdc2 regulators Cdc25 and Mik1 (RHIND and RUSSELL 1998 ). Despite the involvement of the checkpoint Rad proteins in telomere length maintenance, mutations of Chk1, Cds1, and another Cdc2 regulator Wee1 or overexpression of Cdc25 have no effect on telomere length; nor do mutations in Cdc2 affect telomere length (DAHLEN et al. 1998 ). Therefore, telomere length maintenance in S. pombe appears to be independent of cell cycle arrest that involves the regulation of Cdc2. Conceivably, functional telomeres might inhibit interactions between checkpoint Rad proteins and checkpoint downstream signaling transducers. The checkpoint Rad proteins might even have unidentified alternate downstream signal transducers that are used to ensure telomere replication.

To define the roles that checkpoint and DNA repair proteins play in telomere maintenance, we undertook epistasis analysis of various checkpoint and DNA repair mutants on the basis of steady-state telomere length in a series of multiple mutant combinations. From these studies, we conclude that Rad3/Rad26 and Tel1/Rad32 represent two independent functional pathways required for the maintenance of stable telomeres (Rad32 is an ortholog of the S. cerevisiae and mammalian Mre11 proteins). We also compared senescence rates upon telomerase trt1 deletion in various checkpoint mutant backgrounds and conclude that Rad3/Rad26 and Tel1/Rad32 pathways must also be important for functions other than the recruitment of telomerase to telomeres. In addition, we show that damage-induced phosphorylation of Rad32 is independent of both Rad3 and Tel1 kinases, and we thus implicate other unidentified kinase(s) in phosphorylation of Rad32. We also demonstrate specific association of checkpoint sensor and DNA repair proteins to telomeres by chromatin immunoprecipitation (ChIP) analyses. Through these studies we conclude that checkpoint sensor and DNA repair proteins contribute to maintenance and protection of telomeres through their binding to telomeres.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains and general methods:
The fission yeast strains used in this study were constructed by standard techniques (ALFA et al. 1993 ). Most strains used in this study are leu1-32 ura4-D18 his3-D1 ade6-M210 or ade6-M216, and detailed genotypes are listed in the supplementary Table S1 at http://www.genetics.org/supplemental. Sequences of PCR oligonucleotide primers used for strain construction are listed in the supplementary Table S2 at http://www.genetics.org/supplemental.

Mutations were previously described for rad1 (rad1::ura4⁺; SUNNERHAGEN et al. 1990 ), rad9 (rad9::ura4⁺; MURRAY et al. 1991 ), hus1 (hus1::LEU2; KOSTRUB et al. 1997 ), rad17 (rad17::ura4⁺; GRIFFITHS et al. 1995 ), rad3 (rad3::ura4⁺; BENTLEY et al. 1996 ), rad26 (rad26::ura4⁺; AL-KHODAIRY et al. 1994 ), crb2 (crb2::ura4⁺; SAKA et al. 1997 ), chk1 (chk1::ura4⁺; AL-KHODAIRY et al. 1994 ), cds1 (cds1::ura4⁺; BODDY et al. 1998 ), and trt1 (trt1-D2::his3⁺; NAKAMURA et al. 1997 ). Strains carrying nmt-HA-rad3 (MOSER et al. 2000 ) and taz1-HA (KANOH and ISHIKAWA 2001 ) were also described previously.

For taz1, a PCR-based method (BAHLER et al. 1998 ) was used to create taz1-D3::LEU2 (taz1-LEUT and taz1-LEUB primers) and taz1-D4::kanMX4 (taz1-MX4T and taz1-MX4B primers). In addition, a previously described taz1::ura4⁺ mutation (COOPER et al. 1997 ) was PCR amplified from genomic DNA (tazKO-LU and tazKO-RB primers). These PCR products were then used for taz1 gene deletion in various checkpoint and DNA repair mutant strain backgrounds.

A PCR-based method (BAHLER et al. 1998 ) was used to create rad32 (rad32-D1::kanMX4), using rad32-KO1 and rad32-KO2 primers; pku70 (pku70-D1::kanMX4), using pku70-KO1 and pku70-KO2 primers; and rad3 (rad3-D2::LEU2), using rad3-LEUT and rad3-LEUB primers.

For tel1 (tel1-D1::kanMX4), the carboxy-terminal untranslated region was amplified by PCR (tel1-T1 and tel1-B2 primers) and then cloned into pBluescript II SK(+) (Stratagene, La Jolla, CA) as a HindIII-XhoI fragment. The amino-terminal untranslated region was subsequently amplified (tel1-T3 and tel1-B4 primers) and cloned into the same plasmid as the SacII-XbaI fragment. This plasmid was then digested with BamHI and EcoRI to clone the BamHI-EcoRI kanMX4 fragment from the pFA6a-kanMX4 plasmid (WACH et al. 1994 ), creating pBS-tel1::kanMX4 plasmid. The SacII-SnaBI tel1::kanMX4 fragment from pBS-tel1::kanMX4 was then used for tel1 deletion. Another tel1 (te11-D2::LEU2) was created by a PCR-based method (BAHLER et al. 1998 ), using tel1-LEUT and tel1-LEUB primers.

A PCR-based method (BAHLER et al. 1998 ) was used to create carboxy-terminally tagged rad32-TAP and rad32-myc (rad32-tagT and rad32-tagB primers), pku70-myc and pku70-HA (BAM102 and BAM103 primers), rad9-myc (BAM84 and BAM85 primers), and hus1-myc (BAM88 and BAM89 primers). Another PCR technique (KRAWCHUK and WAHLS 1999 ) was used to create carboxy-terminally tagged rad1-myc (BAM36–BAM39 primers) and rad17-myc (BAM31–BAM34 primers).

HA-rad3 cells express the amino-terminally 3HA-tagged Rad3 fusion protein from the endogenous rad3⁺ promoter. It was created by transforming a strain with an integrated ura4⁺ marker 5' adjacent to the rad3⁺ gene with the plasmid carrying the 3HA-rad3 fusion construct and then selecting for 5-fluoroorotic acid (5-FOA)-resistant cells (GRIMM et al. 1988 ).

Pulsed-field gel electrophoresis:
For pulsed-field gel electrophoresis (PFGE), cells were suspended and lysed in agarose plugs as follows: Cells were washed twice in SP1 [50 mM citrate-phosphate (pH 5.6), 40 mM EDTA, 1.2 M sorbitol] and then incubated for 2–3 hr at 37° in SP1 with 0.6 mg/ml Zymolyase-100T (ICN Biomedicals). The cells were pelleted and resuspended at 6–7 x 10⁸ cells per ml in TSE [10 mM Tris-HCl (pH 7.5), 0.9 M sorbitol, 45 mM EDTA]. The cell suspension was warmed to 42°, and 1–1.5 volume of 1% low-melting agarose (Bio-Rad, Richmond, CA) in TSE was added. Aliquots were dispensed into plug molds and allowed to solidify. The gelled plugs were incubated at 55°, first for 90 min in 0.25 M EDTA, 50 mM Tris-HCl (pH 7.5), and 1% SDS and then for 48 hr in 1% lauryl sarcosine, 0.5 M EDTA (pH 9.5), and 1 mg/ml proteinase K. Plugs were washed three times in Tris-EDTA and stored at 4° in Tris-EDTA. For NotI-digested PFGE, plugs were preequilibrated 2–3 hr at 37° in NEB3 buffer [10 mM NaCl, 5 mM Tris-HCl, 1 mM MgCl₂, 0.1 mM dithiothreitol (pH 7.9 at 25°)] plus 100 µg/ml BSA and then digested with NotI restriction endonuclease at 37° overnight. Probes specific for telomeric NotI fragments (C, I, L, and M) were created as previously described (NAKAMURA et al. 1998 ).

Liquid culture growth curve:
Heterozygous diploid strains were sporulated and the resulting tetrads were dissected and germinated on yeast extract medium-supplemented (YES) plates (ALFA et al. 1993 ); genotypes of the resulting cells were then distinguished by growing them on selective minimal plates. Colonies derived from each spore were grown at 32° for 3 days and then picked and diluted to 5 x 10⁴ cells/ml in 20 ml YES. These cultures were grown for 24 hr at 32°, at which point the cell density was determined by counting in a hemacytometer, and the cells were diluted to a cell density of 5 x 10⁴ cells/ml in 20 ml fresh YES liquid medium and incubated at 32°. These procedures were repeated every 24 hr for 18 days.

Immunopurification and Western blot analysis:
Whole-cell extracts from rad32-TAP-tagged (RIGAUT et al. 1999 ) strains were purified with IgG Sepharose (Pharmacia, Piscataway, NJ). Purified material was analyzed by Western blot analysis using peroxidase anti-peroxidase (PAP) antibody (P 2026; Sigma, St. Louis). protein phosphatase (New England Biolabs, Beverly, MA) was used to perform phosphatase treatment.

ChIP assays:
ChIP assays were performed as described (STRAHL-BOLSINGER et al. 1997 ) with minor modifications. Cells were lysed in lysis buffer (50 mM Hepes-KOH pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 5 µg/µl aprotinin, 5 µg/µl leupeptin, 5 µg/µl pepstatin, 1 mM phenylmethylsulfonyl fluoride). The crude cell lysate was sonicated to yield 0.5–1 kb DNA fragments and clarified by centrifugation for 10 min at 16,000 x g. Prior to immunoprecipitation, 1/10 volume of the cell lysate was saved for an input control. Immunoprecipitations were performed with either monoclonal anti-myc antibody (9E10; Babco) or monoclonal anti-HA antibody (16B12; Babco). Immunoprecipitates were washed three times each with 1 ml lysis buffer and 1 ml lysis buffer/500 mM NaCl and two times each with 1 ml wash buffer (10 mM Tris-HCl pH 8, 0.25 M LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA) and 1 ml TE buffer (10 mM Tris-HCl, 1 mM EDTA pH 8). The samples were then processed as described (ORLANDO and PARO 1993 ). PCR reactions used the following primers to amplify the telomeric DNA (BAM136, 5' GTG TGG AAT TGA GTA TGG TGA A 3'; BAM137, 5' CGG CTG ACG GGT GGG GCC CAA TA 3') or the ade6⁺ DNA (BAM138, 5' AGG TAT AAC GAC AAC AAA CGT TGC 3'; BAM139, 5' CAA GGC ATC AGT GTT AAT ATG CTC 3'). To assure a quantitative analysis between individual samples, we first established that we assayed in the linear range of the PCR. PCR reactions were analyzed on 2% agarose gels in 1x TAE buffer.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Checkpoint sensor mutants all have shorter telomeres:
Previous studies in S. pombe reported that rad1, rad17, rad3, and rad26 mutant cells have shorter telomeres, while rad9 and hus1 mutant cells have normal telomere length (DAHLEN et al. 1998 ; NAITO et al. 1998 ; MATSUURA et al. 1999 ). Rad1, Rad9, and Hus1 share a PCNA-like motif (CAI et al. 2000 ; VENCLOVAS and THELEN 2000 ), form a heterotrimeric complex, and appear to function in a single pathway with regard to checkpoint responses (AL-KHODAIRY et al. 1994 ; KOSTRUB et al. 1997 ; CASPARI et al. 2000 ; KAUR et al. 2001 ). Therefore, it was surprising to find that rad1 mutant cells have shortened telomeres while rad9 and hus1 mutant cells have normal telomere lengths. However, many of the checkpoint mutant cells tested in previous studies were point mutations, and thus it was possible that some checkpoint proteins were erroneously found not to play a role in telomere maintenance. Therefore, we decided to retest telomere length in deletion mutants of checkpoint sensor proteins (Rad1, Rad9, Hus1, Rad17, Rad3, and Rad26) and their downstream effector proteins (Crb2, Chk1, and Cds1).

In contrast to previous reports, we found that rad1, rad9, and hus1 strains all had shorter telomeres (Fig 1A, lanes 2–4; Fig 2A). In addition, the rad17 mutant strain had shorter telomere length and the extent of shortening was similar to that of rad1, rad9, and hus1 strains (Fig 1A and Fig 2A). rad3 and rad26 cells had the shortest telomere lengths among the six checkpoint sensor mutants (Fig 1A, lanes 7 and 8; Fig 2A). Mutations in other checkpoint-related proteins (crb2, chk1, cds1, and chk1 cds1) that are thought to function downstream of the six checkpoint sensor proteins had little or no effect on telomere length (Fig 1A, lanes 9–13; Fig 2A).

View larger version (115K): In this window In a new window Download PPT slide   Figure 1. Telomere lengths of various checkpoint-, DNA repair-, and telomere-related protein mutant combinations. After strains were created by either genetic crosses or DNA transformations, they were restreaked at least five times on YES plates prior to preparation of genomic DNA to ensure telomere length equilibrium. After digestion with ApaI, genomic DNA was subjected to electrophoresis on 2% (A–F and H) or 1% (G) agarose gels, stained with ethidium bromide to confirm approximately equal loading in each lane, transferred to a nylon membrane, and hybridized to a telomeric DNA probe (NAKAMURA et al. 1997 ). The ApaI site is located in the telomere-associated sequence (TAS) 30–40 bp away from telomeric repeat sequences in both ends of chromosomes I and II and at least one end of chromosome III (SUGAWARA 1988 ), giving rise to a broad 300-bp telomere hybridization signal in the wild-type (wt) strain (marked as "telomeres"). Hybridization signals designated as "TAS & rDNA adjacent telomeres" come from cross-hybridization to TAS or hybridization to telomere(s) of chromosome III, which contain rDNA repeats directly adjacent to the telomeric repeat sequence and therefore lack the TAS-associated ApaI site directly adjacent to the telomeric repeat sequence.

View larger version (54K): In this window In a new window Download PPT slide   Figure 2. Phosphor-Imager analysis of telomere lengths for selected checkpoint and DNA repair mutants (A–D) and summary of telomere length analysis for all the mutant combinations tested in this study (E). Data from multiple Southern blots (Fig 1) were analyzed using ImageQuant software (Molecular Dynamics, Sunnyvale, CA) and signal intensities for telomere hybridization signals were normalized and plotted against DNA size. Graphs were shifted vertically to allow easier comparison of telomere lengths among different mutant cells. Dotted vertical lines indicate peaks of telomere hybridization signals.

We further analyzed telomere length in various double-mutant combinations among checkpoint sensor proteins (Fig 1B) and found that rad1 hus1, rad9 hus1, and rad17 hus1 mutant combinations have the same telomere length as the single mutants (Fig 1B, lanes 2–8; Fig 2B). These results suggest that rad1, rad9, hus1, and rad17 function in a single pathway for telomere maintenance, consistent with their function in the checkpoint response (AL-KHODAIRY and CARR 1992 ). Rad3 and Rad26 appear to function in the same pathway, as the rad3 rad26 double mutant had the same telomere length as the single mutants (Fig 1B, lanes 9–11; Fig 2B; MATSUURA et al. 1999 ). Mutations in rad3 and rad26 are epistatic to rad1, rad9, hus1, and rad17, since rad3 rad1, rad3 rad9, rad3 hus1, rad3 rad17, and rad26 hus1 all showed no additional telomere shortening compared to rad3 or rad26 single mutants (Fig 1B, lanes 10–17; Fig 2B). These results thus suggest that Rad1, Rad9, Hus1, Rad17, Rad3, and Rad26 contribute to telomere maintenance in a single pathway, but that Rad3 and Rad26 are more important in maintenance of telomeres in fission yeast.

Tel1 and Rad32 function in the same pathway for telomere maintenance:
We next examined how checkpoint proteins interact with Tel1 and Rad32 proteins in S. pombe. Rad32 is an ortholog of the S. cerevisiae and mammalian Mre11 proteins. Studies in S. cerevisiae have shown that the Mre11-Rad50-Xrs2 complex and Tel1 function in a single pathway for telomere maintenance (RITCHIE and PETES 2000 ; TSUKAMOTO et al. 2001 ).

We found that tel1 mutant cells had normal telomere length. We observed synergistic loss of telomeres in tel1 rad3 and tel1 rad26 cells (Fig 1C, lanes 7 and 8), in agreement with previous studies (NAITO et al. 1998 ; MATSUURA et al. 1999 ). Chromosome circularization in tel1 rad3 cells was previously reported, using PFGE analysis and microscopic observations (NAITO et al. 1998 ), while chromosome circularization in tel1 rad26 cells has not been reported. As shown in Fig 3B, we observed that both tel1 rad3 and tel1 rad26 cells have fused C, I, L, and M NotI telomeric fragments to generate I + L and C + M bands that are specific to circularized chromosome I and chromosome II, respectively (lanes 7 and 8), like trt1 telomerase mutant survivors (lane 11; NAKAMURA et al. 1998 ). In contrast, other upstream checkpoint sensor mutants (rad1, rad9, hus1, and rad17) showed only slight telomere shortening compared to single mutants when combined with tel1 mutation (Fig 1C, lanes 3–6). Mutants of downstream effectors of the checkpoint pathway (crb2, chk1, cds1, and chk1 cds1) showed wild-type telomere length even in combination with a tel1 mutation (Fig 1C, lanes 9–12).

View larger version (51K): In this window In a new window Download PPT slide   Figure 3. Pulsed-field gel electrophoresis fractionation and hybridization analysis of S. pombe chromosomal DNAs. (A) NotI restriction enzyme map of S. pombe chromosomes (vertical lines). The telomeric fragments C, I, L, and M are solid. Chromosome III lacks a NotI site. (B and C) NotI-digested S. pombe chromosomal DNAs were fractionated in a 1% agarose gel with 0.5x TBE buffer at 14°, using the CHEF-DR II system (Bio-Rad) at 6 V/cm (200 V) and a pulse time of 60–120 sec for 24 hr, transferred to nylon membrane, and hybridized to C-, I-, L-, and M-specific probes (NAKAMURA et al. 1998 ). Four telomeric fragments (C, I, L, and M) and fusion products (C + M and I + L) are marked on the left. The "I" fragment from pku70 tel1 rad17 cells migrates faster (marked with a triangle) since the rad17+ gene resides on the I fragment and the deletion mutation introduced an additional NotI site.

Rad32 mutant cells have previously been reported to have shorter than wild-type telomere length in S. pombe (WILSON et al. 1999 ; MANOLIS et al. 2001 ). We did not observe telomere shortening in our rad32 mutant (Fig 1D and Fig 2D). Since S. pombe Rad32 and Rad50 are expected to be in a complex analogous to the S. cerevisiae Mre11-Rad50-Xrs2 complex, both mutations might be expected to show similar effects on telomere length. S. pombe rad50 cells have also been reported to have short telomeres (HARTSUIKER et al. 2001 ). However, we observed normal telomere length for rad50 cells (data not shown), much like in rad32 cells. It was also suggested that rad32 mutation is synthetic lethal with rad3 mutation (TAVASSOLI et al. 1995 ), but we were able to generate double mutants and propagate them for many generations, although these cells are extremely sick and produce many dead cells with abnormal morphology. We do not know the exact cause for these discrepancies, but unsuspected suppressor mutation(s) or variations in growth conditions might have contributed to these disagreements. In this regard, we note that both rad32 and rad50 strains grow poorly and appear to accumulate DNA damage, as many cells appear to be arrested by the checkpoint.

When the rad32 mutation was combined with the tel1 mutation, we found that the rad32 tel1 double mutant still had normal telomere length (Fig 1D, lanes 1–4). When the rad32 mutation was combined with rad1, rad9, hus1, or rad17 mutations, double mutants showed only slight shortening of telomere lengths compared to single mutants in rad1, rad9, hus1, or rad17 strains (Fig 1D, lanes 5–8). Combination of the rad32 mutation with either rad3 or rad26, on the other hand, caused total loss of the telomere hybridization signal (Fig 1D, lanes 9 and 10). This is due to circularization of chromosomes, as PFGE analysis showed a shift of C, I, L, and M telomeric NotI fragments into two bands corresponding to I + L and C + M bands (Fig 3B, lanes 9 and 10). Combination of the rad32 mutation and mutations of downstream effectors of the checkpoint pathway (crb2, chk1, cds1, and chk1 cds1) showed wild-type telomere length (Fig 1D, lanes 11–16). Therefore, rad32 and tel1 mutations caused identical phenotypes in terms of telomere length in all checkpoint mutant backgrounds we tested. Taken together, these results are consistent with the idea that Tel1 and Rad32 function in the same pathway for telomere maintenance much like S. cerevisiae Tel1 and Mre11-Rad50-Xrs2. The above data also indicate that Rad3/Rad26 and Tel1/Rad32 represent two functional groups required for telomere maintenance in S. pombe.

Interaction between Ku70 and checkpoint proteins:
Next, we tested how telomere length is affected by combining the pku70 mutation with mutations in checkpoint genes tel1 and rad32. In S. pombe, pku70 makes telomeres shorter and the telomere-associated sequences (TAS) more recombinogenic (BAUMANN and CECH 2000 ; MANOLIS et al. 2001 ). We observed shorter and more heterogeneous telomere length for pku70 cells compared to wild-type cells. We also found that double mutants of pku70 and checkpoint sensor protein mutations have the telomere lengths of checkpoint sensor single mutants for rad1, rad9, hus1, rad17, rad3, and rad26 and also made telomere length more homogeneous compared to a pku70 strain (Fig 1E, lanes 1–11; Fig 2C and Fig D). Therefore, mutations in checkpoint sensor genes are epistatic to pku70 in maintenance of telomere length. For combinations of pku70 and downstream protein mutations (crb2, chk1, cds1, and chk1 cds1), telomere lengths were like that of the pku70 single mutant (Fig 1E, lanes 12–17). Telomere lengths in pku70 tel1 and pku70 rad32 cells were also the same as in the pku70 single mutant (Fig 1F, lanes 4–6; Fig 2D).

We also created triple mutants in which a checkpoint sensor was deleted along with tel1 and pku70. We hypothesized that the Rad3-Rad26 complex may contribute positively to telomere maintenance both through a pathway involving Ku70 and through another pathway involving the Rad1/Rad9/Hus1/Rad17 proteins since mutations in rad3 and rad26 were found to be epistatic to mutations in rad1, rad9, hus1, rad17, and pku70. If this were true, deletion of both pathways in combination with the tel1 mutation might cause chromosomes to circularize as they do in tel1 rad3 or tel1 rad26 cells. However, we found that pku70 tel1 rad1, pku70 tel1 rad9, pku70 tel1 hus1, and pku70 tel1 rad17 cells all maintained short but stable telomeres (Fig 1F, lanes 8–11). Telomere lengths in these triple-mutant cells were slightly reduced compared to single checkpoint mutant cells (rad1, rad9, hus1, rad17), pku70 checkpoint double-mutant cells (pku70 rad1, pku70 rad9, pku70 hus1, pku70 rad17), or pku70 tel1 cells (Fig 1F and Fig 2D). PFGE analysis found no evidence of chromosome circularization in those triple-mutant cells (Fig 3C, lanes 2–5). Therefore, the Rad3-Rad26 complex must have additional telomere-associated targets, outside the Rad1/Rad9/Hus1/Rad17 and Ku70 epistasis groups, which confer protection from chromosome circularization in tel1 and rad32 backgrounds.

We also tested the possibility that synergistic chromosome circularization observed in tel1 rad3 cells might be suppressed by pku70 mutation. This might be the case because pku70 cells were reported to have elevated TAS recombination (BAUMANN and CECH 2000 ) and therefore loss of Ku70 protein might allow either the recombinational machinery or the telomerase better access to telomeres, thus suppressing the telomere loss observed in the tel1 rad3 background. Alternatively, the Rad3 and Tel1 kinase pathways may be necessary to specifically inhibit the NHEJ pathway from fusing chromosome ends. In that case, elimination of NHEJ by removal of Ku protein may allow cells to avoid fusing their telomeres. Indeed, telomere fusions observed in nitrogen-starved taz1 cells can be suppressed by pku70 or lig4 mutation (FERREIRA and COOPER 2001 ). However, pku70 tel1 rad3 cells again completely lost telomeric hybridization (Fig 1F, lane 13) and have circular chromosomes (Fig 3C, lane 6). Therefore, pku70 mutation cannot suppress chromosome circularization in tel1 rad3 cells.

Interaction between Taz1 and checkpoint proteins:
In S. cerevisiae, the telomere shortening phenotype of a tel1 mutation is epistatic over the telomere elongation phenotype of the rap1-17 mutation (CRAVEN and PETES 1999 ; RAY and RUNGE 1999 ). Rap1 is a major telomere-binding protein in S. cerevisiae, and Rap1, through interaction with the Rif1 and Rif2 proteins, is thought to be involved in negative regulation of telomerase activity and telomeric recombination (HARDY et al. 1992 ; WOTTON and SHORE 1997 ). Rap1 interacts with Rif1 and Rif2 through its carboxy-terminal domain, and the rap1-17 mutation disrupts these associations. The fact that the tel1 rap1-17 double mutant has a short telomere length, much like the tel1 mutant, suggests that in S. cerevisiae telomerase recruitment/activation is still largely dependent on Tel1 kinase even in the absence of negative regulators of telomerase (CRAVEN and PETES 1999 ; RAY and RUNGE 1999 ).

As deletion of S. pombe Taz1 telomere-binding protein leads to extreme elongation of the telomere tract, which is reminiscent of the S. cerevisiae rap1-17 phenotype (COOPER et al. 1997 ), we tested whether telomere elongation in taz1 required Tel1, Rad32, Ku70, or other checkpoint proteins. We created double-mutant combinations by individually deleting the taz1 gene from single-mutant cells of rad1, rad9, hus1, rad17, rad3, rad26, crb2, chk1, cds1, chk1 cds1, tel1, rad32, and pku70. We used this sequential procedure to eliminate the possibility that starting with highly elongated taz1 telomeres would mask the effects of the checkpoint mutations. The resulting double-mutant cells were then restreaked multiple times on rich media to allow cells to achieve equilibrium telomere length. As shown in Fig 1G, we found that telomeres are still elongated in all double-mutant cells. In taz1 rad3 and taz1 rad26, telomere elongation was slightly reduced compared to taz1 cells, but they were still extremely elongated compared to wild-type telomere length (Fig 1G, lanes 1, 7, and 8). These results indicate that telomere elongation in the taz1 mutant is epistatic to mutations in the Tel1/Rad32, checkpoint sensors (Rad1/Rad9/Hus1/Rad17 and Rad3/Rad26), or Ku70 epistasis groups.

If telomere elongation in taz1 cells is independent of Tel1 and Rad3 kinases, the elongation induced by the taz1 deletion might be expected to suppress the rapid telomere loss and circularization of tel1 rad3 cells. To test this possibility, we also created taz1 tel1 rad3, taz1 tel1 rad26, taz1 rad32 rad3, and taz1 rad32 rad26 cells. These triple-mutant strains were created by deleting the tel1 or rad32 gene from the taz1 rad3 or taz1 rad26 cells. Therefore, these cells originally had highly elongated telomeres prior to the deletions. We found that the triple-mutant cells still completely lost their telomere hybridization signal (Fig 1H, lanes 2–5) and circularized their chromosomes (Fig 3C, lanes 9–12), indicating that even in the absence of Taz1 protein, telomeres cannot be maintained in tel1 rad3, tel1 rad26, rad32 rad3, or rad32 rad26 backgrounds. In contrast, chromosome circularization observed in telomerase trt1 mutants (Fig 3B, lane 11) was suppressed and the cells maintained stable linear chromosomes indefinitely if the trt1 gene was deleted in cells that were already deleted for taz1 (Fig 1H, lane 6; Fig 3C, lane 13; NAKAMURA et al. 1998 ). In taz1 trt1 cells, telomeres are presumably maintained by recombination (NAKAMURA et al. 1998 ). Therefore, the fact that the taz1 mutation could not overcome elimination of the Rad3/Rad26 and Tel1/Rad32 pathways suggests that these pathways are necessary for both telomerase-based and recombination-based maintenance of telomeres.

Rad3 and Tel1 kinases have additional roles other than recruitment of telomerase:
In S. pombe, careful analysis of how tel1 or checkpoint mutants affect the rate of senescence in telomerase mutant cells has not yet been carried out, nor has direct comparison of the rate of senescence for telomerase vs. tel1 rad3 mutants. Therefore, we undertook such analyses to gain insight into the contribution of Rad3/Rad26 and Tel1/Rad32 pathways to telomere maintenance.

We performed a series of growth curve experiments in which heterozygous diploid cells were sporulated and dissected, and then cultures of cells with appropriate genotypes were serially diluted (Fig 4). As previously reported (NAKAMURA et al. 1998 ; HAERING et al. 2000 ), the growth rate of trt1 cells gradually declined in a reproducible manner from day 2 to day 10 in independent liquid cultures (Fig 4A and data not shown). On the other hand, different trt1 cultures displayed different patterns of recovery in growth rate in the phase when survivor cells start to take over the cultures. We did not observe a delayed decline in growth rate for tel1 trt1 cells compared to trt1 cells (Fig 4B). For rad3 trt1 and rad26 trt1 cells, the rate at which growth rates declined among independent cultures became much less reproducible compared to trt1 cells (Fig 4C and Fig D). This effect presumably is related to the checkpoint-related functions of Rad3 and Rad26, as chk1 trt1 cells showed similarly wide-ranging variability in decline of growth rate among independent cultures (data not shown).

View larger version (39K): In this window In a new window Download PPT slide   Figure 4. Comparison of growth characteristics of wild-type (wt) and mutant cells after extended growth in liquid cultures. Various heterozygous diploid strains (see the supplementary Table S1 at http://www.genetics.org/supplemental) were sporulated and the resulting tetrads were dissected and germinated on YES plates. The resulting haploid cells with indicated genotypes were grown at 32° for 3 days and then picked and diluted to 5 x 104 cells/ml in YES. These cultures were grown for 24 hr at 32°, at which point the cell density was determined, and the cells were diluted into fresh YES liquid medium at 5 x 104 cells/ml. These procedures were repeated for 18 days, and cell densities from each day were plotted (see MATERIALS AND METHODS for more detailed procedures).

As trt1 cells undergo senescence, an increasingly large fraction of cells becomes highly elongated (NAKAMURA et al. 1997 ). These elongated cells were not observed in rad3 trt1, rad26 trt1, or chk1 trt1 cells, suggesting that senescing trt1 cells show checkpoint-dependent cell cycle arrest as the cells lose their telomeric DNA (data not shown). Interestingly, cds1 trt1 cells still elongated as they senesced, and growth rate decline was similar to that in trt1 cells and without wide-ranging variations among independent cultures (data not shown). Therefore, defective telomeres in trt1 cells appear to be recognized as DSBs and trigger G2 checkpoint cell cycle arrest.

We next compared growth characteristics among tel1 rad3, tel1 rad3 trt1, tel1 rad26, and tel1 rad26 trt1 cells after germination of meiotic spores from heterozygous diploid cells. We observed that both tel1 rad3 and tel1 rad26 reached the point of lowest viability much earlier (5 days) than trt1 cells did (10 days), and survivors grew more slowly than trt1 survivor cells (Fig 4E and Fig F). Moreover, tel1 rad3 trt1 and tel1 rad26 trt1 cells did not show any additional loss of growth rate compared to tel1 rad3 and tel1 rad26. Therefore, the presence of functional telomerase did not help to delay senescence. The accelerated senescence phenotype observed for tel1 rad3 and tel1 rad26 cells is reminiscent of those seen in pku70 trt1 cells and cells lacking the proposed telomere capping protein, Pot1 (BAUMANN and CECH 2000 2001 ). Therefore, the two pathways involving Rad3/Rad26 and Tel1/Rad32 must play roles not only in telomerase recruitment, but also in other functions such as telomere protection.

Phosphorylation of Rad32 is independent of Rad3 and Tel1:
Previous studies have shown that Rad32 is phosphorylated in a cell-cycle-dependent manner. Rad32 phosphorylation accumulates in S-phase and this phosphorylation is independent of Rad3 (WILSON et al. 1998 ). Our genetic analysis indicated that Rad32 and Tel1, but not Rad3, function in the same pathway to maintain telomere length in fission yeast. Therefore, we tested whether Rad32 phosphorylation might be crucial for Rad32 telomere function and whether Tel1 might be responsible for Rad32 phosphorylation.

Rad32 phosphorylation was detected by the appearance of a slow mobility species in SDS-PAGE that can be converted to a faster mobility species by treatment with phosphatase (Fig 5A). Asynchronous S. pombe cells showed a small amount of Rad32 phosphorylation, which is probably due to a small percentage of cells that are in S-phase (Fig 5B). In contrast, when cells were arrested in S-phase through the addition of hydroxyurea (HU) or when cells were exposed to the DNA-damaging agent methyl methanesulfonate (MMS), increased phosphorylation of Rad32 was observed. Phosphorylation of Rad32 was still observed in rad1, rad9, hus1, rad17, crb2, chk1, cds1, and chk1 cds1 cells (data not shown) and, surprisingly, in tel1, tel1 rad3 (Fig 5B), and tel1 rad26 cells (data not shown). These results showed that Rad3 and Tel1 are not the kinases responsible for the observed Rad32 phosphorylation and suggest that there must be other kinase(s) that can phosphorylate Rad32. However, it is possible that Rad3 or Tel1 carries out phosphorylation of Rad32 that does not alter its mobility on SDS-PAGE. Whether Rad32 phosphorylation is actually required to maintain telomeres has to be resolved. We observed more prominent phosphorylation of Rad32 in asynchronous tel1 rad3 and tel1 rad26 cells (Fig 5B and data not shown). These cells are extremely sick and have circular chromosomes (Fig 3B, lanes 7 and 8). We suggest that these cells have problems in either DNA replication or DNA segregation and therefore accumulate DNA damage, which may explain why these cells have elevated Rad32 phosphorylation.

View larger version (34K): In this window In a new window Download PPT slide   Figure 5. Rad32 phosphorylation is independent of Rad3 and Tel1. (A) TAP-tagged Rad32 was affinity purified from asynchronous rad32-TAP cells. Purified Rad32-TAP was either mock treated or phosphatase treated prior to Western blot analysis. (B) Wild-type (wt), rad3, tel1, and rad3 tel1 cells with rad32-TAP were either treated with 12 mM HU or 0.05% MMS for 3 hr or left untreated (AS). Extracts were prepared from these cells and Rad32-TAP affinity purifications were performed. Whole-cell extracts (top) and affinity-purified protein (bottom) were analyzed by immunoblotting. (*) phosphorylated form of Rad32-TAP.

Checkpoint sensor and DNA repair proteins are bound to telomeric DNA:
Recent studies in S. cerevisiae showed that Mec1, Ddc2, Rad24, Rad17, Ddc1, and Mec3 (homologs of S. pombe Rad3, Rad26, Rad17, Rad1, Rad9, and Hus1, respectively) are recruited to sites of DNA breaks upon induced DNA damage (KONDO et al. 2001 ; MELO et al. 2001 ; ROUSE and JACKSON 2002 ). Therefore, we tested if the checkpoint sensor proteins are physically bound to normal telomeres by performing ChIP assays. As controls we chose the known telomere-binding protein Taz1. Telomeric binding of Taz1 in vivo by ChIP assay has been reported recently (KANOH and ISHIKAWA 2001 ). We also tested telomeric binding of Ku70 by ChIP assay. Binding of Ku proteins to telomeres has been established in mammals and S. cerevisiae (GRAVEL et al. 1998 ; HSU et al. 1999 , HSU et al. 2000 ; D'ADDA DI FAGAGNA et al. 2001 ). On the other hand, binding of the S. pombe Ku70 to telomeres has not been investigated directly, and a recent report showed that S. pombe Ku70 is localized throughout the nucleus and not confined to telomeres (MANOLIS et al. 2001 ). Precipitated DNA was amplified by PCR with primers for the telomere-associated sequence TAS (Fig 6A) and the non-telomere-adjacent ade6⁺ gene. Telomeric DNA, but not ade6⁺ DNA, was specifically amplified from Taz1-HA, Ku70-HA, and Ku70-myc immunoprecipitates while no such enrichment was observed for a nontagged strain (Fig 6B and Fig C). This result shows specific binding of Ku protein to telomeres in fission yeast.

View larger version (41K): In this window In a new window Download PPT slide   Figure 6. Checkpoint sensor proteins Rad32 and Ku70 bind to telomeres in vivo. (A) Schematic diagram of the S. pombe telomere and TAS region. Primers used in the ChIP assay are indicated. (B) ChIP assay of Rad3, Taz1, and Ku. Untagged wild-type control strain or strains with indicated HA-tagged proteins were used. (C) ChIP assay of Rad17, Hus1, Rad9, Rad1, Ku70, and Rad32. Untagged wild-type control strain or strains with indicated myc-tagged proteins were used. (D) ChIP assay of Ku70 in various deletion strains. Untagged wild-type control strain and strains with pku70-HA in indicated mutant backgrounds were used. PCRs were performed on whole-cell extracts (Input) and on chromatin immunoprecipitates (ChIP) using primers to amplify a telomere-specific DNA (telo) and primers to amplify DNA from the ade6+ gene (ade6).

We were unable to detect HA-Rad3 at the telomere when it was expressed from its endogenous promoter, possibly because of its low abundance. On the other hand, HA-Rad3 overexpressed from the nmt promoter was able to specifically enrich telomeric DNA, but not the control ade6⁺ DNA, suggesting that Rad3 binds specifically to telomeres (Fig 6B). However, we cannot exclude the possibility that overexpressed Rad3 associates with telomeres in a nonphysiological manner. We also observed enrichment of telomeric DNA over ade6⁺ DNA in immunoprecipitates from Rad17-myc and to a lesser extent from Rad1-myc, Rad9-myc, and Hus1-myc (Fig 6C). Although the signals we obtained were weaker than those for Ku70-myc, they were reproducible. Differences in signal intensity are most likely due to differences in immunoprecipitation efficiency and protein abundance at the telomere. Taken together, these ChIP assays show that Rad3 and Rad17 and most likely Rad1, Rad9, and Hus1 bind to telomeres. We also obtained a low, but significant signal for telomeric DNA in immunoprecipitates from Rad32-myc cells (Fig 6B). Therefore, our data show that Rad32 also binds to telomeres.

Ku70 binding to telomeric DNA is independent of checkpoint sensor proteins but dependent on Taz1 protein:
In our genetic analysis we found that the mutations eliminating checkpoint sensor proteins are epistatic to pku70 in maintaining stable telomere length, indicating that these proteins may function in the same pathway. To investigate whether Ku70 binding to telomeres might be dependent on the checkpoint sensor proteins, we undertook ChIP analyses (Fig 6D). We observed no change in Ku70 binding in either rad17 or rad3 mutants, indicating that the checkpoint sensor proteins do not function through regulating binding of Ku70 to telomeres.

We also investigated Ku70 binding to telomeres in tel1 and taz1 mutants. Again, no change in telomere binding was found in the tel1 strain. In contrast, in the absence of Taz1 protein, Ku70 binding was greatly diminished. This datum is consistent with data from mammalian cells in which Ku70 is found to bind the Taz1 homologs TRF1 and TRF2 (HSU et al. 2000 ; SONG et al. 2000 ), and Ku heterodimers can be recruited to the circular DNA with internal telomeric repeat sequence via their association with TRF1 (HSU et al. 2000 ). Our data may therefore indicate that Ku70 binding to telomeres is facilitated through Taz1. However, we cannot exclude the possibility that Ku70 binds exclusively to the very termini of chromosomes and that the apparent loss of Ku70 binding is caused by telomere elongation in taz1 cells, since our ChIP assay is designed to detect proteins bound to sites close to TAS (500–1000 bp).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Checkpoint sensor proteins have alternative targets for telomere maintenance:
In this study, we extensively tested the relative contributions of S. pombe checkpoint and DNA repair proteins in telomere maintenance by creating cells carrying various mutant combinations and examining average telomere length and chromosome circularization in the resulting cells. Our results are summarized in Fig 2E. One of the conclusions we draw from such analyses is that downstream effectors of the checkpoint (Crb2, Chk1, and Cds1) are not important for telomere maintenance in S. pombe, even though checkpoint sensor proteins (Rad1, Rad9, Hus1, Rad17, Rad3, and Rad26) are required for proper telomere maintenance. Therefore, checkpoint sensor proteins must contribute to telomere maintenance through alternative telomere target(s) that are unrelated to these checkpoint effectors.

Previous studies in S. pombe also found Chk1 and Cds1 to be not important for telomere maintenance (DAHLEN et al. 1998 ; MATSUURA et al. 1999 ). On the other hand, one study suggested Crb2 is important for telomere maintenance since crb2 (rhp9) cells have shorter telomere length (WILSON et al. 1999 ). We do not know the cause of this discrepancy, but we note that our crb2 strains are generated by backcrossing a crb2 strain obtained from a laboratory different (SAKA et al. 1997 ) from that of the study that reported telomere shortening in crb2 cells. In comparison, S. cerevisiae rad53 (S. pombe Cds1 homolog) mutants have been reported to have short telomeres (LONGHESE et al. 2000 ), while S. cerevisiae rad9 (S. pombe Crb2 homolog) cells were variously reported to have short (VIALARD et al. 1998 ) or wild-type (LONGHESE et al. 2000 ) telomere length.

Checkpoint sensors Rad1, Rad9, Hus1, and Rad17 function in a single pathway for telomere maintenance and associate with telomeres:
Our studies indicate that checkpoint proteins with PCNA homology (Rad1, Rad9, and Hus1) as well as the RFC-like protein Rad17, which has been proposed to recruit the Rad1-Rad9-Hus1 complex to sites of DNA damage, function in the same pathway for maintenance of telomere length. This conclusion is based on the observation that mutant combinations among these proteins did not lead to additional telomere shortening and mutants lacking these four proteins showed identical telomere lengths under all different mutant backgrounds (tel1, rad32, pku70, pku70 tel1, taz1) that we tested.

Our results are consistent with results from previous studies for rad1 and rad17 mutants (DAHLEN et al. 1998 ; MATSUURA et al. 1999 ). However, our results disagree with a previous study for rad9 and hus1 mutants where S. pombe rad9-192 and hus1 mutations were found not to affect telomere length (DAHLEN et al. 1998 ). Since the previous study tested telomere length in rad9-192 cells and not rad9 cells, the difference between the two results may be explained by partial retention of function of the rad9-192 allele with respect to telomere length maintenance, although rad9-192 is as sensitive to UV and ionizing radiation as a rad9 mutant strain (MURRAY et al. 1991 ; LIEBERMAN et al. 1992 ). On the other hand, the previous study and our study used the same hus1::LEU2 deletion mutation; this latter disagreement cannot be easily explained.

In S. cerevisiae, rad17 (S. pombe rad1 homolog) and ddc1 (S. pombe rad9 homolog) cells were reported to have short telomeres, and they were considered to be in the same pathway, since rad17 ddc1 double-mutant cells showed no additional telomere shortening (LONGHESE et al. 2000 ). Curiously, S. cerevisiae mec3 cells (S. pombe hus1 homolog) were reported to have longer (CORDA et al. 1999 ; LONGHESE et al. 2000 ) or wild-type (GRANDIN et al. 2001 ) telomere length. Therefore, checkpoint proteins with the PCNA-like motif may not have completely equivalent functions in S. cerevisiae. It is also interesting to note that S. cerevisiae rad24 (S. pombe rad17 homolog) was reported to have wild-type telomere length (LONGHESE et al. 2000 ) even though it was recently shown that Rad24 was necessary to recruit Ddc1 to sites of DNA DSBs (KONDO et al. 2001 ; MELO et al. 2001 ).

Our ChIP analyses showed robust binding of S. pombe Rad17 to telomeres, while the PCNA-like checkpoint proteins (Rad1, Rad9, and Hus1) bound weakly. Therefore, we suggest that the checkpoint proteins with RFC and PCNA homology contribute to telomere maintenance through their binding to telomeres. As telomere shortening in this class of checkpoint proteins is also observed in S. cerevisiae and C. elegans (AHMED and HODGKIN 2000 ; LONGHESE et al. 2000 ), we suggest that this is a highly conserved mechanism. Perhaps PCNA-like checkpoint proteins may provide a launching pad for recruitment of factors that help prepare telomeres to be extended by telomerase, which is analogous to PCNA-dependent recruitment of DNA replication proteins.

The Rad3-Rad26 complex has additional roles that are independent of other checkpoint sensor proteins and that function through its association with telomeres:
Our data indicated that Rad3 kinase and its proposed regulatory subunit Rad26 together form a separate epistasis group for telomere maintenance from other checkpoint sensor proteins, as these two proteins had the shortest telomere lengths among checkpoint sensor proteins and the rad3 rad26 double-mutant cells had the same telomere length as single-mutant cells. Studies by other