|Home | About | Journals | Submit | Contact Us | Français|
The Saccharomyces cerevisiae CDC13 protein binds single-strand telomeric DNA. Here we report the isolation of new mutant alleles of CDC13 that confer either abnormal telomere lengthening or telomere shortening. This deregulation not only depended on telomerase (Est2/TLC1) and Est1, a direct regulator of telomerase, but also on the yeast Ku proteins, yKu70/Hdf1 and yKu80/Hdf2, that have been previously implicated in DNA repair and telomere maintenance. Expression of a Cdc13-yKu70 fusion protein resulted in telomere elongation, similar to that produced by a Cdc13-Est1 fusion, thus suggesting that yKu70 might promote Cdc13-mediated telomerase recruitment. We also demonstrate that Stn1 is an inhibitor of telomerase recruitment by Cdc13, based both on STN1 overexpression and Cdc13-Stn1 fusion experiments. We propose that accurate regulation of telomerase recruitment by Cdc13 results from a coordinated balance between positive control by yKu70 and negative control by Stn1. Our results represent the first evidence of a direct control of the telomerase-loading function of Cdc13 by a double-strand telomeric DNA-binding complex.
Telomeres, the ends of eukaryotic chromosomes, are critical for maintaining chromosome stability and genome integrity (2, 8, 60). Telomeres are composed of particular DNA sequences which are rich in TG and arranged in species-specific repeated motifs. Telomeres are capped by proteins that bind to these repeating DNA sequences (6, 20). This apparently serves at least two distinct purposes. First, some of these telomeric proteins presumably form complexes that regulate telomerase activity and, hence, the length of telomeric tracts (31, 43). Some telomeric proteins have also been implicated in the physical protection of chromosome ends (38), in preventing recombinational events that would otherwise frequently occur between repeating telomeric sequences (33, 34, 53), and in keeping off DNA repair enzymes (14). Indeed, telomeres represent naturally occurring DNA double-strand breaks that, contrary to those resulting from accidental damage, do not need to (and must not) be repaired. Surprisingly, however, yeast Ku proteins, as well as proteins of the Mre11-Rad50-Xrs2 complex, which have been implicated in DNA repair by nonhomologous end joining have also been implicated in telomere maintenance (3, 4, 7, 11, 12, 27, 28, 39, 44, 46, 47). Moreover, yKu70 and yKu80 have been found to localize at the telomeres (18, 37).
The repeating TG-rich telomeric DNA sequences are mostly double stranded. However, during S phase only, telomeres display a short (ca. 35- to 50-nucleotide) single-stranded DNA extension that marks the very end of the telomere (57, 58). Single-stranded telomeric DNA is thought to represent the site of anchoring of telomerase, which is composed of the evolutionary conserved Est2 reverse transcriptase enzyme and of the TLC1 RNA template (31, 42). However, recent experiments suggest that telomerase-dependent elongation of de novo ends does not appear to involve single strandedness and does not require significant degradation prior to addition of newly synthesized telomeric DNA (9). Est1 and Est3 represent two subunits of the telomerase complex (25, 29, 55), which although not required for in vitro telomerase catalytic activity (32), are nevertheless stable components of the enzyme and regulate its activity in vivo through physical association with Est2 and TLC1 (25, 61).
Cdc13 was the first identified single-strand DNA-binding telomeric protein in Saccharomyces cerevisiae and, consequently, its status as a candidate for most of telomeric functions has become prominent (5, 14, 30, 45). The isolation of the cdc13-2/est4-1 allele, which confers a strong deficit in telomerase activity (29), as well as the recent finding that a fusion protein made of Cdc13 plus Est1 could bypass telomerase-defective mutations in either protein, strongly suggests that interactions between Cdc13 and Est1 represent the mechanism by which a number of regulators can control telomerase recruitment (10). Indeed, physical association between Cdc13 and Est1 has been revealed recently (48). Cdc13 has also been shown to bind Pol1 in vivo, and it has been proposed that Cdc13 might coordinate regulation at the telomere ends of G-strand lengthening by telomerase, via Est1, and C-strand resynthesis by polymerase α (48). In addition, the observation that the temperature-sensitive cdc13-1 allele displays abnormal accumulation of single-stranded DNA specifically at telomeric regions of chromosomes argued that Cdc13 might be a major telomeric capping protein (14). It has also been observed that when Cdc13 was defective, in cdc13-1 cells, the absence of either yKu70 or yKu80, which was otherwise dispensable, impaired growth (44, 46). The yKu proteins, which bind to double-stranded telomeric DNA, have been proposed to be involved in establishing the proper terminal DNA structure of chromosomes in cooperation with telomerase (18, 46). In addition to its nonessential function in the recruitment of telomerase at telomere ends (45), Cdc13 has an essential function that has not yet been clearly defined. cdc13-1 mutant cells are temperature sensitive and present a first cell cycle arrest at restrictive temperatures of growth (14).
In the present study, we have isolated several new mutant alleles of CDC13 which confer either abnormal telomere elongation or, on the contrary, telomere shortening. Telomere elongation in these novel cdc13 alleles was found to be more affected by mutations in either YKU70 or YKU80 than by mutations in TEL1 or RAD50, therefore implicating the yeast Ku proteins in the telomerase-loading function of Cdc13. This was supported by observing telomere elongation as a direct result of the expression of a Cdc13-yKu70 fusion protein, which is comparable in length to that produced by a Cdc13-Est1 fusion. We also present evidence, based on overexpression of Stn1 or expression of Stn1-Cdc13 fusions, that Stn1, a protein that associates with Cdc13 by two-hybrid interaction (17), is an inhibitor of telomerase recruitment via Cdc13. We propose that Cdc13 is both a positive and a negative regulator of telomerase recruitment, which establishes differential interactions with other telomeric proteins, and that the balance between these two opposing effects principally relies on interactions with yKu70 or yKu80 and with Stn1.
General plasmids and media used in this study were as described previously (17). Disruptions of TLC1, EST1, RAD50, YKU70, or YKU80 in wild-type, cdc13-1, or cdc13-109i strains were achieved following transformation of linearized tlc1::LEU2 (52), est1::URA3 (55), rad50::hisG-URA3-hisG (41), yku70/hdf1::URA3 (47), or yku80::TRP1 (see below) DNA fragments. The correct disruptions were detected on Southern blots. In some cases, these disruption strains were further mixed with other mutations by genetic crosses. The tel1::kanMX4 (strain record number 3114; Research Genetics, Inc., Huntsville, Ala.) and rad52-7::LEU2 (strain record number XS560-1C-1D1; Yeast Genetic Stock Center, Berkeley, Calif.) disruption strains were backcrossed five times against the genetic background used in our lab (17). The cdc13::TRP1 disruption plasmid was constructed by inserting TRP1 at the BamHI site of CDC13, at nucleotide 1349, and the yku80::TRP1 disruption plasmid was constructed by inserting TRP1 between the XbaI and AccI sites (nucleotides 84 to 1762) of YKU80. The cdc13-109 integrated allele was constructed by the pop-in–pop-out method. To do this, cdc13-109 was cloned into a URA3 integrative plasmid (YIp211) which was then used to transform a wild-type strain. The Ura+ transformants were then grown on 5-fluoro-orotic acid (5-FOA) plates to counterselect for cells that, together with the URA3 marker gene, had lost one copy of the CDC13 gene (either the wild-type or the mutant copy). Telomere length was then monitored on Southern blots in several of these Ura− cells after a few generations of growth (see below). Only half of these colonies displayed elongated telomeres, with the other half exhibiting telomeres of wild-type size. Cells with elongated telomeres were selected, and the presence of only one copy of the CDC13 gene was verified by Southern blot. This suggested that cells with elongated telomeres had integrated the cdc13-109 allele at the CDC13 locus, while cells with wild-type telomeres had, on the contrary, evicted the cdc13-109-URA3 integrated construct. Integration of the cdc13-69 allele at the CDC13 locus was performed using the same methodology as for cdc13-109.
CDC13 open reading frame (ORF) plus 300 bp upstream of the ATG was amplified by PCR under mutagenic conditions, as described previously for STN1 (17) in standard PCR buffer supplemented with MnCl2, using standard Taq polymerase (Gibco-BRL). Following two rounds of PCR mutagenesis, using the products of the first reaction as a template for the second reaction, the PCR products were cleaned and used directly to transform a cdc13-1 strain, together with a single-copy, centromeric, plasmid, YCp111-LEU2 (15), made linear by digestion and carrying CDC13 flanking regions at each extremity (gap repair method). The 5′ fragment of these flanking regions consisted of the 964 bases before the start codon plus the 57 bases after, while its 3′ fragment comprised the 690 bases before the stop codon plus the 830 bases after. Cells were then plated onto leucine-lacking (Leu−) medium and incubated at 25°C until colonies of transformants developed. Because the objective of CDC13 mutagenesis was to uncover non-temperature-sensitive alleles deregulated in telomere length control, transformants were then replica plated on Leu− medium at 32 and 34°C, temperatures of growth restrictive for cdc13-1. This allowed us to select for mutagenized CDC13 plasmids capable of sustaining growth of the cells bearing them in the absence of functional endogenous Cdc13 protein, because the Cdc13-1 protein is inactivated at temperatures higher than 28°C in our genetic background (17). Among several thousands of such colonies growing at 32 or 34°C, 121 were picked out randomly and separately grown for further analysis of the length of their telomeric tracts (see below). The most interesting YCp111-borne cdc13 alleles, in terms of telomere length deregulation, were then recovered from the original cdc13-1 recipient strain and used to retransform the cdc13-1 strain. Genomic DNA from transformants grown for about 100 generations was then prepared, and the lengths of the telomeric tracts were analyzed by Southern blotting, as explained below.
The stn1-63 allele was generated by PCR mutagenesis, followed by gap repair, under conditions described previously (17). The stn1::TRP1 strain bearing the stn1-63 allele on a YCp111-LEU2 plasmid was selected (after eviction of the wild-type STN1 allele carried by a YCp33-URA3 plasmid on 5-FOA-containing medium) among several tens of other potential stn1 mutant strains on the basis of telomere length deregulation, as directly measured on Southern blots, as described below.
For sequence analysis of the cdc13 alleles, the ORFs of the mutant CDC13 genes, cloned into YEp195-GAL1 (an episomal, 2μ, URA3 plasmid), were digested with EcoRI, taking advantage of the presence of two natural EcoRI sites in the CDC13 sequence. This generated three pieces of CDC13 ORF roughly equal in size, which were then subcloned into pBluescript. DNA sequencing was performed in a semiautomated DNA sequencer (Applied Biosystems) using T3 or T7 primer as the sequencing primer.
Genomic DNA was prepared as described previously (17), digested with XhoI and separated by electrophoresis in a 0.9% agarose gel in Tris-borate-EDTA. After denaturation, DNA was transferred onto nitrocellulose membrane and immobilized by baking at 80°C for 1 h under a vacuum (1). The membrane was then prehybridized in 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% sodium dodecyl sulfate–1% nonfat milk and hybridized with a 270-bp TG1–3 32P-labeled probe representing S. cerevisiae telomeric sequences. Results were analyzed using a Storm PhosphorImager (Molecular Dynamics).
Control experiments in which DNA fragments after XhoI cutting were separated under denaturing conditions (1% agarose gel run in 40 mM NaOH–2 mM EDTA) were performed to ensure that both single-stranded and double-stranded modifications in telomere length were detected (see also reference 17).
Cdc13-Est1, Cdc13-69-Est1, Cdc13-yKu70, Cdc13-yKu80, Cdc13-Stn1-63, and Cdc13-231-Stn1 in-frame fusion proteins were constructed by cloning in a single-copy, centromeric, plasmid the entire CDC13 ORF (or the cdc13-69 or cdc13-231 ORFs) plus upstream promoter sequences in front of the EST1, YKU70, YKU80, or STN1 (wild-type or mutant) ORF (which included their natural stop codons). The yKu80-Cdc13 fusion protein was constructed by cloning in a single-copy plasmid the entire YKU80 ORF plus upstream promoter sequences in front of CDC13, so as to keep a continuous reading frame. The CEN plasmids used in the present study are of the YCplac series and are single-copy plasmids (15), just like the CEN plasmid, pRS415 (51), used by Evans and Lundblad (10). Telomere elongation by the Cdc13-Est1 fusion protein was not due to increased protein expression because overexpression of CDC13 or EST1 had no effect on telomere length (10, 17). Moreover, the functionality of the Cdc13-Est1 fusion protein was attested to by its ability to assume the essential function of Cdc13 (rescue of cdc13-1 cells at a restrictive temperature or of cdc13Δ cells) and to rescue the senescence phenotype of an est1::URA3 mutant.
The 10 cdc13 mutant DNAs corresponding to the mutant strains described in the present study were subcloned into a YEp195-GAL1 (2μ, URA3) plasmid under the control of the strong, inducible, GAL1 promoter. In this plasmid the natural stop codon of the cdc13 mutant genes was eliminated so as to obtain a continuous reading frame between the Cdc13 protein and a 2HA-6His epitope tag (16). This allowed visualization of the Cdc13 mutant proteins by Western blotting using monoclonal anti-hemagglutinin antibody (12CA5; Boehringer). The presence of this epitope tag also allowed purification of the Cdc13 mutant proteins by Ni chromatography directed against the His6 part of the tag, using Qiagen reagents. After transfer to nitrocellulose membrane (Schleicher & Schuell), proteins were detected using an enhanced chemiluminescence system (ECF; Amersham) coupled with detection in a Storm PhosphorImager (Molecular Dynamics).
For band shift experiments and detection by Western blotting, wild-type cells bearing, on a YEp195-GAL1-URA3 plasmid, wild-type CDC13 or a cdc13 mutant allele fused in 3′ with a 2HA-6His epitope were grown overnight, in liquid cultures, in glucose-based minimal medium lacking uracil. Cdc13 protein expression was then induced for 4 h at 30°C after shifting the cells to Ura− galactose-containing medium following four or five washes in galactose-based medium. Cells were then harvested by centrifugation, and extracts were prepared in band shift lysis buffer (50 mM NaH2PO4-Na2HPO4 buffer, pH 8.0; 30 mM NaCl; 10 mM Na imidazole) supplemented with protease inhibitors (1% phenylmethylsulfonyl fluoride and 10 μg each of aprotinin, leupeptin, and pepstatin per ml). Approximately 500 μg of the total proteins were then incubated with Ni-nitrilotriacetic acid beads (Qiagen) for 2 h at 4°C. Nickel beads were then washed three times in 50 mM NaH2PO4-Na2HPO4 buffer (pH 8.0), 20 mM Na imidazole, and 0.5% Tween 20 and then once in 50 mM NaH2PO4-Na2HPO4 buffer (pH 8.0) and 20 mM Na imidazole. Cdc13–2HA-6His proteins were then eluted in 50 μl of 50 mM NaH2PO4-Na2HPO4 buffer (pH 8.0)–250 mM Na imidazole. Binding reactions were performed in 50 mM Tris-HCl buffer (pH 7.5), 1 mM EDTA, 50 mM NaCl, 1 mM dithiothreitol, 1 μg of single-stranded poly(dI-dC) using 2 to 4 μl of eluted protein, and 1 ng of 32P-labeled (TG1–3)3 (TGTGTGGGTGTGTGGGTGTGTGGG) for 20 min at 30°C. A monoclonal anti-HA antibody (12CA5) was used at 0.5 μg per reaction. After the addition of 1 μl of 10% glycerol, the reactions were loaded on a 4.5% polyacrylamide gel that was run for 2 h at 4°C. Gels were then dried and analyzed using a Storm PhosphorImager (Molecular Dynamics).
To better understand the role of Cdc13 at telomeres, we set out to generate new alleles of CDC13 using PCR amplification under mutagenic conditions (see Materials and Methods) and reintroduce them into a conditional system provided by a cdc13-1 mutant strain (14). cdc13-1 cells stop growing at temperatures above 23 to 25°C (14) or above 27 to 28°C in our genetic background (17). Among several thousand transformants of cdc13-1 capable of growth at 32 or 34°C, 121 were picked randomly for further study. Genomic DNA was prepared, and the lengths of the telomeric tracts were measured by probing total XhoI-digested genomic DNA with a 32P-labeled TG1–3 telomeric probe, as described in Materials and Methods. Only mutant strains with substantial changes in telomere length were selected for further analysis, namely, cdc13-7, cdc13-109, cdc13-231, and cdc13-276, which are telomere-elongating alleles, and cdc13-23, cdc13-30, cdc13-69, cdc13-243, cdc13-273, and cdc13-280, which are telomere-shortening alleles (Fig. (Fig.1A).1A).
Since we suspected that the Cdc13-1 mutant protein might have some residual activity at 32 to 34°C, we next tested whether the isolated cdc13 alleles were capable of sustaining growth on their own. To this end, we attempted to replace the genomic copy of CDC13 by each one of these cdc13 alleles using the pop-in–pop-out method (see Materials and Methods). Concerning the telomere-elongating alleles (telomere-shortening alleles will be examined below), we were successful in recovering cells that had integrated a mutant copy of cdc13-109 (from now on referred to as cdc13-109i) at the CDC13 genomic locus. The Cdc13-109 mutant protein in the cdc13-109i strain could assume the essential function of Cdc13 while harboring long telomeres similar to those in the cdc13-1 strain bearing the cdc13-109 allele on a centromeric plasmid (Fig. (Fig.1A1A and B). In terms of growth and cellular morphology, the cdc13-109i strain was indistinguishable from a wild-type strain (data not shown). We were not successful at obtaining replacements with the other three selected telomere-elongating cdc13 alleles.
In another approach to characterize these telomere-elongating cdc13 alleles, we constructed a cdc13 disruption strain (cdc13::TRP1), that survived owing to a single-copy URA3 plasmid expressing wild-type CDC13, which was then transformed with one of the four selected cdc13 alleles carried by a single-copy plasmid. After eviction of the CDC13-URA3 plasmid on 5-FOA-containing medium, the cdc13::TRP1 YCp-cdc13-109 and cdc13::TRP1 YCp-cdc13-231 strains were viable and exhibited telomeres of a length similar to that in the respective original strains in a cdc13-1 background at 32–34°C (Fig. (Fig.1A1A and D), while the cdc13::TRP1 YCp-cdc13-7 and the cdc13::TRP1 YCp-cdc13-276 strains were not viable (data not shown).
To serve as a control for some of the experiments that have been performed in a cdc13-1 background (see below), we measured telomere length in the cdc13-109i strain transformed with cdc13-1 on an episomal plasmid. Importantly, at 32 or 34°C, telomeres in this strain were the same length as those in the cdc13-109i strain (Fig. (Fig.1E)1E) and as those in the cdc13-1 strain harboring cdc13-109 on a centromeric plasmid (Fig. (Fig.1A).1A). These control experiments establish that the presence of the Cdc13-1 mutant protein at 32 to 34°C has no effect on telomere length while, on the other hand, it may provide, at these temperatures, a residual Cdc13 activity allowing growth of strains harboring the cdc13-7 and cdc13-276 mutations which are otherwise unable to sustain growth on their own.
Because both telomerase-dependent and telomerase-independent mechanisms, in the latter case relying on homologous recombination, can regulate telomere size (33, 34), we next asked which one of these two mechanisms affected telomere regulation in the telomere-elongating cdc13 mutants described here. To do this, we introduced a null mutation in TLC1, the RNA subunit of telomerase essential for telomerase activity (52), into the cdc13-109i mutant (see Materials and Methods) and measured telomere size in the resulting double mutants. The typical telomere elongation observed in the cdc13-109i was not observed in the cdc13-109i tlc1Δ mutants (Fig. (Fig.1B).1B). These results were confirmed using the cdc13-1 YCp-cdc13-276 strain (Fig. (Fig.1B).1B). Because the homologous recombination mechanisms controlling telomere length are entirely dependent on Rad52, we then analyzed telomere size deregulation in cdc13-109i rad52Δ or cdc13-1 YCp111-cdc13-276 rad52Δ mutant strains and found that they displayed telomeres elongated to the same extent as that in the corresponding RAD52+ strains (data not shown).
It has been recently demonstrated that Cdc13 regulates telomerase recruitment via functional interaction with Est1 (10, 48, 55). We found that telomere elongation associated with the cdc13-109i and cdc13-276 mutations was suppressed in the absence of Est1 (Fig. (Fig.1C),1C), implying that Est1 is necessary for abnormal telomere lengthening in the cdc13-109i and cdc13-276 mutants. These results also suggest that regulation of telomerase recruitment by wild-type Cdc13 requires Est1.
To determine whether the telomere length phenotype could be merely due to a defect of the Cdc13 mutant proteins in binding telomeric DNA (5, 26, 30), band shift experiments were performed using Cdc13–2HA-6His mutant proteins purified by Ni chromatography (see Materials and Methods). The Cdc13-109 and Cdc13-231 proteins were still be able to bind telomeric DNA (Fig. (Fig.2A).2A). The specificity of these interactions was evidenced by visualizing supershifts when a monoclonal anti-HA antibody was added to the reaction (Fig. (Fig.2A).2A). Careful examination of the intensities of the DNA-protein bands revealed that the Cdc13-109 and Cdc13-231 mutant proteins were only partially competent in binding telomeric DNA compared to wild-type Cdc13 (Fig. (Fig.2).2).
We reasoned that the telomere-elongating Cdc13 mutant proteins might be deregulated in their interaction with another telomeric protein. To test this hypothesis, telomere-elongating alleles of CDC13 were introduced into mutant strains known to exhibit abnormally short telomeres: rad50, tel1, yku70/hdf1, and yku80/hdf2. Rad50 is part of the Mre11-Rad50-Xrs2 telomeric complex that has been previously shown to function in so-called DNA nonhomologous end-joining (NHEJ) and in telomere maintenance (21, 28, 44). Tel1 is a telomeric protein that has been implicated in telomere length regulation and shares homology with the human ATM proteins (19, 35, 49). The yeast Ku proteins have been implicated in NHEJ, in heterochromatin organization, in telomere silencing, and in telomere maintenance (4, 7, 11–13, 18, 22, 37, 39).
Genetic disruption of either RAD50, TEL1, YKU70, or YKU80 induced telomere shortening (Fig. (Fig.3),3), as previously demonstrated (3, 28, 35, 47). In the cdc13-109i yku70Δ and cdc13-109i yku80Δ double mutants, telomeres were basically of the same length as those in the yku70Δ and yku80Δ single mutants (Fig. (Fig.3A,3A, compare lane 7 to lane 8 and lane 14 to lane 15), whereas telomeres of the cdc13-109i rad50Δ double mutant were of an average length intermediate between those of each of the corresponding single mutants (Fig. (Fig.3A,3A, compare lane 10 to lanes 11 and 12). The effect of tel1Δ in these experiments was between that of yku70Δ/yku80Δ and that of rad50Δ and was therefore more difficult to interpret (Fig. (Fig.3A,3A, compare lane 1 to lanes 3 and 4). However, careful examination of the data revealed that the upper limit of the smear defining the average value of the bulk of telomere lengths was much higher in cdc13-109i tel1 cells than in cdc13-109i yku70 or cdc13-109i yku80 cells (Fig. (Fig.3A).3A). These observations were confirmed by experiments using the cdc13-276 mutant (Fig. (Fig.3B).3B).
Because the genetic interactions described above were only indicative of a potential functional relationship between Cdc13 and either yKu70 or yKu80, we decided to adopt a complementary approach. One of our hypotheses to explain these interactions relied on the existence of a physical interaction between either yKu70 or yKu80 and Cdc13. Because attempts to detect physical association between Cdc13 and yKu70 in a two-hybrid system failed (unpublished results), we decided to use a method recently applied with success in the analysis of interactions between Cdc13 and Est1 (10), which consists in the expression of fusion (hybrid) proteins. We first expressed a Cdc13-Est1 fusion protein from a single-copy centromeric plasmid under the control of the CDC13 promoter (see Materials and Methods) and observed that it produced telomere elongation in a CDC13+ EST1+ strain, an effect that was much more pronounced in a cdc13-1 background at 34°C or in cdc13::TRP1 than in a wild-type background (Fig. (Fig.4A,4A, lanes 2, 8, and 11), as described recently (10).
We then analyzed the effects on telomere length of expressing a Cdc13-yKu70 fusion protein from a single-copy plasmid under the control of the CDC13 promoter. The Cdc13-yKu70 fusion protein was fully functional because it rescued the nonviability of cdc13Δ cells as well as the growth defect of yku70Δ cells at 37°C (13) (data not shown). Expression of the Cdc13-yKu70 fusion resulted in telomere elongation in cdc13Δ and CDC13+ cells, as well as in cdc13-1 cells grown at 34°C (Fig. (Fig.4A,4A, lanes 5, 9, and 12). In all backgrounds (CDC13+, cdc13-1, and cdc13Δ), expression of the Cdc13-yKu70 fusion lengthened telomeres to a lesser extent than that of the Cdc13-Est1 fusion (Fig. (Fig.4A).4A). Importantly, the Cdc13-yKu70 did not cause telomere elongation in a strain disrupted for EST1 (Fig. (Fig.4B,4B, lane 6). This suggested that the artificially introduced Cdc13-yKu70 hybrid protein affected telomerase recruitment via Est1. On the other hand, telomere elongation due to the Cdc13-Est1 fusion still took place in a strain disrupted for YKU70 to the same extent as that in a YKU70+ strain (Fig. (Fig.4A,4A, compare lanes 3 and 4). In addition, expression of CDC13 or EST1 alone from a single-copy plasmid in a cdc13-1 yku70Δ strain (Fig. (Fig.4B,4B, lanes 1 and 3) provided controls for the experiments shown above.
Because yKu70 and yKu80 are active in DNA repair only as a heterodimeric complex (12, 39), we measured the effect of expressing a Cdc13-yKu80 fusion protein on telomere length. Surprisingly, expression of a Cdc13-yKu80 fusion protein expressed from a single-copy centromeric plasmid under the control of the CDC13 promoter, although it restored inviability of cdc13-1 cells at 34°C, did not rescue the yku80Δ-induced telomere shortening (data not shown). We therefore constructed another hybrid protein which this time consisted of a yKu80-Cdc13 fusion protein expressed from a single-copy centromeric plasmid under the control of the YKU80 promoter. The yKu80-Cdc13 fusion rescued the inviability of cdc13Δ cells but only partially the short telomere phenotype of yku80Δ (telomeres were of a heterogenous size, forming a smear whose upper limit reached the wild-type size and lower limit the yku80Δ size; data not shown). On the other hand, the yKu80-Cdc13 fusion produced only a moderate lengthening of telomeres in cdc13Δ cells (data not shown). Because of the lack of full functionality of the Cdc13-yKu80 and yKu80-Cdc13 fusions, one cannot conclude whether the absence of a drastic effect on telomere length of these fusions is real or not.
Finally, we asked whether the Cdc13-yKu70 fusion could produce telomere lengthening in the absence of yKu80. Interestingly, a yku80Δ cdc13Δ double mutant expressing the Cdc13-yKu70 under the conditions described above (see Fig. Fig.4A)4A) did not display any significant change in telomere length, unlike the YKU80+ cdc13Δ YCp-Cdc13-yKu70 strain, which clearly exhibited telomere elongation (Fig. (Fig.4C,4C, compare lanes 1 and 2). These experiments suggest that telomere elongation induced by the Cdc13-yKu70 fusion requires the presence of yKu80.
None of the telomere-shortening cdc13 alleles described in this study (see Fig. Fig.1A),1A), namely, cdc13-23, cdc13-30, cdc13-69, cdc13-243, cdc13-273, and cdc13-280, provoked senescence (29, 33, 34) in contrast to cdc13-2/est4-1 cells or tlc1Δ cells (data not shown). To further characterize these novel telomere-shortening Cdc13 mutant proteins, we performed band shift experiments to measure their ability to bind telomeric DNA (Fig. (Fig.5A).5A). All six telomere-shortening Cdc13 mutant proteins were severely defective in binding telomeric DNA (Fig. (Fig.5A).5A). Among these, Cdc13-69–2HA-6His Cdc13-243–2HA-6His, and Cdc13-273–2HA-6His retained some DNA binding activity, as confirmed by observing a supershifted band upon addition of anti-HA antibody during the reaction, while the Cdc13-23–2HA-6His, Cdc13-30–2HA-6His, and Cdc13-280–2HA-6His proteins were almost completely defective in binding telomeric DNA (Fig. (Fig.5A).5A).
Among the six telomere-shortening cdc13 alleles, cdc13-69 conferred the largest telomere shortening effect (Fig. (Fig.1A).1A). Integration of cdc13-69 at the CDC13 genomic locus (to create cdc13-69i) demonstrated that the Cdc13-69i protein could assume Cdc13's essential function (no apparent defect; data not shown) and conferred a short telomere phenotype similar in amplitude to that conferred by Cdc13-69 (compare lane 12 in Fig. Fig.1A1A to lane 2 in Fig. Fig.55C).
It has been shown that Cdc13-2, the only other Cdc13 mutant protein known to confer telomere shortening, was capable of binding telomeric DNA (45) and Est1 (48). It is important to note that a Cdc13-69–Est1 fusion protein conferred telomere lengthening, by a just slightly smaller degree than that produced by the Cdc13-Est1 fusion (Fig. (Fig.4A,4A, lane 16), thus suggesting that increased association between Cdc13-69 and Est1 could cure the telomere size regulation defect of the Cdc13-69 mutant protein. However, because Est1 is also a single-strand telomeric DNA-binding protein (55), it is not possible yet to decide whether the defect of Cdc13-69 is in its ability to bind telomeric DNA, a defect rescued by Est1-mediated recruitment to DNA, or rather lies in a putative physical interaction with Est1. Expression of a Cdc13-69–yKu70 fusion also rescued the short telomere phenotype conferred by the cdc13-69 allele (Fig. (Fig.4A,4A, compare lanes 15 and 17), thus supporting the model that yKu70 promotes the Cdc13-mediated recruitment of telomerase.
We have previously proposed that Stn1 might negatively regulate Cdc13 activity (17). In view of the potential positive modulation of Cdc13 activity by yKu70, we speculated that Stn1 might feed negative signals into the Cdc13-regulating machinery. In our previous studies on Stn1 (17), it had not been demonstrated that a loss of Stn1 function led to deregulation of telomerase recruitment. We now demonstrate that the telomeric defect of stn1-13, a mutant which exhibits a severe growth defect at the restrictive temperature of 37°C and telomere elongation at any temperature between 25 and 37°C (17), results from a deregulation in telomerase recruitment and/or activity (Fig. (Fig.6A).6A).
Because Stn1 loss of function leads to telomerase hyperactivation, Stn1, which physically associates with Cdc13 by two-hybrid interaction (17, 54), might be a negative regulator of Cdc13. If so, overexpression of STN1 might be able to modify telomere length regulation. Overexpression of STN1 from a multicopy (episomal, 2μ) plasmid under the control of its natural promoter produced no visible effect on telomere length in a wild-type strain (data not shown), as described previously (17). However, when STN1 was overexpressed in a mutant strain expressing the telomere-shortening cdc13-273 allele from a single-copy plasmid, a further increase in telomere shortening was observed (Fig. (Fig.6B,6B, compare lanes 2 and 3). Likewise, overexpression of STN1 in a mutant strain expressing the telomere-elongating cdc13-276 allele from a single-copy plasmid resulted in a noticeable slowing down of telomere elongation (Fig. (Fig.6B,6B, compare lanes 4 and 5). On the basis of these experiments, one can conclude that Stn1 behaves as an inhibitor of telomerase recruitment.
A possible mechanism accounting for the observations described above consists of direct titration of Cdc13 by Stn1. To test this prediction, we designed the following fusion protein experiments. Expressing a fusion protein made of Cdc13-231 and Stn1 in a cdc13Δ strain resulted in a dramatic suppression of telomere elongation conferred by the Cdc13 mutant protein (Fig. (Fig.7A,7A, compare lanes 2 and 3). Thus, it appears that the defect of the Cdc13-231 protein in telomere length regulation can be totally corrected by providing a more permanent association between wild-type Stn1 and the Cdc13-231 mutant protein. As an important control, we verified that YCp111-cdc13-231-induced telomere elongation was not compromised by a fusion with Est1 (data not shown). Interestingly, a fusion between wild-type Cdc13 and wild-type Stn1 provoked a small shortening of telomeres (Fig. (Fig.4A,4A, lane 13).
We next examined the consequences of fusing an Stn1 mutant protein conferring telomere elongation to wild-type Cdc13. To do this, we used stn1-63, a mutant that was generated using a PCR-based methodology (see Materials and Methods). The strain harboring stn1-63 carried on a centromeric plasmid (YCp111-LEU2) in an stn1 null background (stn1::TRP1) was selected on the basis of its deregulation in telomere length. The stn1-63 mutant strain had no visible morphological or growth defect at temperatures between 25 and 37°C (data not shown). The constitutive defect in telomere size of stn1-63 cells, namely, a very dramatic increase in telomere length (Fig. (Fig.7B,7B, lane 2), is comparable to that in stn1-13 cells growing at 34°C (17). As shown above for stn1-13 (Fig. (Fig.6A),6A), telomere lengthening conferred by stn1-63 was found to depend entirely on telomerase (data not shown). Introduction of the gene encoding the Cdc13–Stn1-63 fusion carried by a single-copy plasmid under the control of the CDC13 promoter into an stn1Δ strain almost totally suppressed telomere elongation conferred by Stn1-63 (Fig. (Fig.7B,7B, compare lanes 2 and 3). This observation suggests that the defect of the Stn1-63 protein, which as far as we know about Stn1 function (17; present data) might be a failure to properly regulate Cdc13, can be almost totally corrected by artificially increasing its association with Cdc13 by means of a fusion protein.
DNA sequencing revealed that all six sequenced cdc13 alleles (cdc13-7, cdc13-69, cdc13-109, cdc13-243, cdc13-273, and cdc13-276) contained multiple point mutations. In addition, a mutation in lysine 706 of Cdc13-276 introduced a termination codon that resulted in the truncation of the last 218 amino acids, while a frameshift in cdc13-7 sequence at lysine 702 led to the introduction of a premature stop codon at amino acid 721. We have not been able so far to identify the mutations responsible for the phenotypes of the corresponding mutants.
The survival of an organism relies on the proper duplication, segregation, and stability of its genome. Telomeres play an important role in maintaining chromosome structure because, for instance, chromosomes that lose a telomere are themselves eliminated from the cell (50). In yeast, mutations in telomerase components or regulators produce a gradual erosion of chromosomes that eventually leads to death by senescence due to chromosome instability (29, 34). Cdc13 has been previously implicated, together with Est1, as the main regulator of telomerase access to telomeric ends (10, 30, 45, 48, 61).
The present study extends our knowledge on the role of Cdc13 in telomerase control through the analysis of novel cdc13 alleles and their functional relationships with mutations in other telomeric proteins. Our data suggest that the yeast Ku proteins, previously implicated in DNA repair and telomere maintenance in yeast and humans (4, 7, 16, 24, 27, 37, 39), promote telomerase recruitment by Cdc13. Our results also demonstrate that regulation by yKu70 or yKu80 is opposite of that by Stn1, since we found that Stn1 negatively regulates Cdc13-mediated telomerase recruitment. The present study also provides telomere-shortening cdc13 alleles that do not confer senescence and may be useful in some biochemical or genetic assays.
Our observation that STN1 overexpression produces telomere shortening (Fig. (Fig.6B)6B) is reminiscent of the effects of overexpressing RIF1 or RIF2 (59). A striking parallel between these two situations is that Rif1 and Rif2 physically associate with Rap1, a master regulator of telomere length which binds double-stranded telomeric DNA (23, 36, 59), while Stn1 physically associates with Cdc13, a master regulator of telomere length which binds single-stranded telomeric DNA (5, 17, 26, 30, 45). However, a noticeable difference between the two situations was that STN1 overexpression did not affect telomere length in wild-type cells (17; the present data), whereas overexpression of RIF1 and RIF2 did (59). We propose that overproduction of Stn1 can affect telomere length only when Cdc13 function is already compromised, as explained below.
Experiments using fusion proteins further established the role of Stn1 as an inhibitor of Cdc13's telomerase loading function (Fig. (Fig.7).7). As argued from experiments using Cdc13-Est1 fusion proteins (10), the present data suggest the existence of physical interactions between Cdc13 and Stn1. Indeed, suppression of the cdc13-231-induced telomere elongation following consolidation of its natural interaction with Stn1, as well as suppression of the stn1-63-induced telomere elongation following consolidation of its natural interaction with Cdc13, strongly argue that a major control over Cdc13 activity operates through physical association with Stn1. In fact, because it is already known that Stn1 and Cdc13 associate in a two-hybrid system (17, 54), the experiments on Cdc13-Stn1 fusions presented here enhance interpretations made concerning experiments done with the Cdc13-yKu70 fusion (Fig. (Fig.4).4). It is noticeable that a fusion made of wild-type Cdc13 and wild-type Stn1 provoked a small shortening of telomeres (Fig. (Fig.4A,4A, lane 13), the interpretation of which is discussed below.
Disrupting either YKU70 or YKU80 had a larger suppressing effect on telomere elongation conferred by the Cdc13-276 or Cdc13-109 mutant proteins than disrupting RAD50 or TEL1 (Fig. (Fig.3).3). Although at first glance tel1Δ may appear to have an effect similar to that of yku70Δ or yku80Δ in diminishing the cdc13-109-associated telomere elongation, careful examination of the Southern blot revealed a difference between the two, which was confirmed in the cdc13-1 YCp-cdc13-276 strain. It could be argued that such experiments are difficult to interpret because, for instance, all of the telomere-shortening mutations considered here are in proteins known to be involved each in several pathways, including telomere maintenance. Moreover, since the two mutations present in a given strain affect telomere length in opposite directions, it is difficult to determine whether the resulting average telomere length corresponds to equilibrium between the two opposing effects or rather reflects actual genetic interaction between the two mutations. For these reasons, we were very cautious in interpreting these epistasis experiments. In the end, a noticeable result is that the yku70Δ and yku80Δ mutations totally suppressed the cdc13-109-induced telomere elongation (Fig. (Fig.3A),3A), while conferring wild-type length telomeres to strains bearing the cdc13-276 mutation (Fig. (Fig.3B).3B). The effects of rad50Δ and tel1Δ were smaller than those of yku70Δ and yku80Δ in either cdc13 strain.
Because we did not want to overinterpret the epistasis experiments discussed above, we used these results only as an indication and not as a conclusive argument. In fact, the indications provided by these epistasis experiments were further confirmed by the finding that expression of a Cdc13-yKu70 fusion protein clearly resulted in telomere lengthening (Fig. (Fig.4).4). Given that telomere elongation in the cdc13 alleles described here depends on Est1 (Fig. (Fig.1C),1C), altogether these findings suggest that yKu70 and yKu80 may regulate interactions between Cdc13 and telomerase. Importantly, the presence of yKu80 was necessary to mediate the effect of the Cdc13-yKu80 on telomere length (Fig. (Fig.4C).4C). A possible model is that yKu70 might help recruit telomerase through an interaction with Cdc13, as explained below, with yKu80 playing a crucial role in this mechanism as representing the Ku component attached to telomeric DNA.
Based on the experiments presented here and the data available in the literature (principally, references 4, 10, 14, 17, 18, 25, 26, 29, 30, 40, 44–48, 55, 57, and 61), we propose an improved model for the control of telomerase recruitment by Cdc13 that involves the existence of a balance between the effects of yKu70 and Stn1 on Cdc13 (Fig. (Fig.8).8). The top panel of Fig. Fig.88 depicts the situation in wild-type cells, while the bottom panels propose two hypotheses to account for the situation encountered in the cdc13-109i mutant strain.
According to our model (Fig. (Fig.8,8, top panel), one might expect Stn1 overproduction to shorten telomeres in wild-type cells, which was not observed experimentally. In fact, this result can be explained by assuming that Cdc13 is solidly anchored at telomere ends and that overproduced Stn1 cannot titrate it out or pull it away from the telomeres. Under such conditions, only the presence of Stn1, but not its amount, in close proximity to Cdc13 would be required for the precise tuning of telomere length control, according to our working model (Fig. (Fig.8,8, top panel). On the other hand, a Cdc13 mutant protein exhibiting an altered interaction with another telomeric protein (as depicted, for instance, for Cdc13-109 in Fig. Fig.8,8, bottom panels) or telomeric DNA (as is the case for Cdc13-69; see Fig. Fig.5A),5A), might see its overall stability affected, thus resulting in competitive inhibition by Stn1 for the affected binding sites and, hence, in telomere shortening (Fig. (Fig.6B).6B). Confirmation of this view must await experimental support.
We observed that expression of a fusion between wild-type Cdc13 and wild-type Stn1 resulted in a slight shortening of telomeres (Fig. (Fig.4A,4A, lane 13). It is difficult to know whether such a decrease is significant or not. If it is significant, then expressing a single copy of the Cdc13-Stn1 fusion gene would be more efficient on Cdc13 regulation than overexpressing STN1. This point of view is supported by the fact that overexpression of STN1 only slightly affected telomere length in cdc13 mutant strains (Fig. (Fig.6B),6B), whereas expressing a fusion between a Cdc13 mutant protein and wild-type Stn1 had a very dramatic effect on telomere length (Fig. (Fig.7A).7A). Although these effects are compatible with the model proposed here, it is too early to provide an accurate explanation of the molecular mechanisms involved, particularly since we know that the defects of the Cdc13 mutant proteins described here are still at the hypothetical stage (Fig. (Fig.8,8, bottom panels). In addition, full comprehension of these mechanisms may be complicated by the possible existence of still-unknown partners of Cdc13 and Stn1.
Because both the Cdc13-yKu70 and the Cdc13-Est1 fusions produced telomere elongation that mimicked telomere deregulation in the cdc13 mutants, the corresponding Cdc13 mutant proteins may be deregulated in their interaction with either yKu70 or Est1 (Fig. (Fig.8,8, bottom panels). On first analysis, a potential defect in the association of Cdc13 with yKu70 in these cdc13 mutants is the more plausible explanation (hypothesis 1, Fig. Fig.8,8, left bottom panel). Indeed, in the alternative hypothesis—a defect in the association of Cdc13 with Est1 in these cdc13 mutants—one should not observe a suppressing effect of the yku70Δ and yku80Δ mutations (Fig. (Fig.3)3) unless the yKu proteins directly interact with telomerase. However, a defect in Cdc13-Est1 interactions could fit with some other aspects of Ku functions (hypothesis 2, Fig. Fig.8,8, right bottom panel). Indeed, yeast Ku mutants have been shown to retain extended TG1–3 tails throughout the cell cycle (46), unlike wild-type cells which retained them only during late S phase (57). It has been proposed that the yeast Ku proteins are responsible for controlling the 5′-3′ processing of telomere ends (22, 46, 56). Therefore, the absence of the Ku proteins might structurally modify the telomeric ends so as to produce an alteration of Cdc13 positioning on the 3′ free end. Under such conditions, disruption of YKU70 or YKU80 in the cdc13-109i mutant might result in Cdc13-109 being physically displaced along the telomere end due to the presence of longer single-stranded telomeric sequences, which might then, by competitive interactions, abolish the higher than normal association between Cdc13-109 and Est1. These hypotheses should now be challenged by biochemical experiments.
Even though the details of the interactions described here have to be elucidated in future experiments, our data establish two major conclusions. First, expression of a Cdc13-yKu70 fusion promotes telomerase recruitment, which is supported by the finding that, in the absence of either Ku protein, the Est1-dependent recruitment of telomerase by Cdc13 is compromised. Second, Stn1 exerts a negative control over Cdc13-mediated telomerase recruitment.
Stn1 and Rif1 or Rif2 might exert two parallel and complementary levels of negative control of telomere length, taking place at two spatially different locations on the telomere. It has been suggested that the Rap1-dependent telomere length-sensing mechanism (36) might be mediated by a balance between opposing effects of Rif1 or Rif2 and those of Sir3 or Sir4 (59). In light of the present data, as well as of the recent demonstration of a competition between Rif1 or Rif2 and yKu70 in the recruitment of Sir proteins by Rap1 at the telomere (40) and of the localization of the Ku proteins potentially at the junction between double-stranded and single-stranded telomeric DNA (18, 37), the Ku proteins represent likely candidates for constituting a functional link between Rap1, Rif1, and Rif2 and Cdc13 that is capable of modulating telomerase recruitment. If our hypothesis is correct, the Ku proteins might contribute to control telomere length by their ability to detect changes in telomere structure, or even by an ability to modify telomere structure (46), parameters that would then be fed into the Cdc13 machinery according to mechanisms proposed above. On the other hand, Stn1 might perceive and convey other types of signals, possibly emanating from extratelomeric locations. The genetic system described here, based on the analysis of yeast Cdc13 mutant proteins and their interaction with other wild-type or mutant telomeric proteins, provides an excellent frame with which to study the mechanism of telomerase recruitment at the telomeres.
We thank James Haber, Sang Eun Lee, Victoria Lundblad, Thomas Petes, Patricia Greenwell, Ethelle Moustacchi, Daniel Gottschling, Miriam Singer, Leland Hartwell, and Eric Gilson for the gifts of strains and plasmids. We also thank Eric Gilson for discussion, Catherine Koering for technical advice concerning the band shift experiments, and Suzy Markossian and Armelle Roisin for operating the semiautomated DNA sequencer.
This work was supported by grants from the Association pour la Recherche contre le Cancer, the Centre National de la Recherche Scientifique, programme Génome, the Comités Départementaux de l'Ardèche, la Loire et la Haute-Savoie de la Ligue Nationale contre le Cancer, and the Région Rhône-Alpes, programme Apoptose et Vieillissement.