Main

Replication protein A (RPA) is a eukaryotic single-stranded DNA–binding protein involved in processes including replication, recombination, transcription, checkpoints, telomere maintenance and DNA repair1,2,3. Human and yeast RPA are each composed of three subunits of 70, 32 and 14 kDa4. RPA70 and RPA32/34 (in yeast Rfa1p and Rfa2p) become phosphorylated in response to DNA damage5,6. After DNA damage, phosphorylation of Rfa1p depends on the checkpoint proteins Mec1p, Mec3p, Rad9p and Rad53p whereas phosphorylation of Rfa2p depends mainly on Mec1p5,6. Rfa2p is also phosphorylated during the normal cell cycle—in a process that depends on the yeast homologs of the related PI3-like protein kinases ataxia-telangiectasia mutated (ATM) and ATM-Rad3-related (ATR) Mec1p and Tel1p (ref. 5)—and during meiosis7. Finally, we have shown that inactivation of the histone H3 methyltransferase Set1p induces a MEC1- and TEL1-independent hyperphosphorylation of Rfa2p in G1 phase in a process that depends on the checkpoint kinase Rad53p (ref. 8). Notably, in set1Δ cells, a truncated form of Rfa2p lacking 38 amino acids from the N terminus of Rfa2p (Rfa2pΔ40), is still phosphorylated in S phase by Mec1p but is not phosphorylated in G1 phase (ref. 8). The rfa2Δ40 mutation conveys checkpoint proficiency, increased viability of checkpoint mutants subjected to irradiation, and slightly slower DNA replication (a delay of 15 min as compared to wild type)8.

Telomeres are essential to the stability of chromosome ends. In most organisms, including yeast and vertebrates, telomeric DNA consists of a tandem array of short guanine-rich repeats9. The telomeric repeats are synthesized by a unique ribonucleoprotein reverse transcriptase called telomerase, which specifically extends the 3′ G-rich telomeric strand10. Synthesis of the complementary 5′ C-rich telomeric strand is thought to be driven by the DNA polymerase α–primase complex11. In S. cerevisiae, the 3′ ends of telomeric DNA consist of a G-rich single-stranded stretch that is detected only during S phase12. This G-tail associates with two single-stranded DNA–binding proteins, Cdc13p and Est1p, that are required for telomerase activity in vivo13,14,15. The action of telomerase at the telomere itself is cell-cycle restricted, indicating that a switch in telomere structure may be required for telomerase action16,17,18,19. This process is related to the interaction of the telomerase catalytic subunit (Est2p) with Cdc13p and Est1p (refs. 14,15,18). Est2p seems to be present at telomeres throughout the cell cycle, in contrast to Est1p, which binds telomeres mainly in S phase in a fashion that seems to be preserved in the presence of the defective cdc13-2 allele18. It has therefore been proposed that Est1p binds telomeres in late S phase and interacts with Cdc13p to activate telomere-bound Est2p (ref. 18).

Here we investigated the relationships between RPA and telomeres. We show that RPA is required to maintain telomere length and acts in a pathway that involves Est2p and the telomerase positive regulator Tel1p. Our results indicate that RPA regulates the action of telomerase during the cell cycle by specifically facilitating the binding of Est1p to telomeres during S phase.

Results

Binding of Rfa2p to telomeric DNA

We investigated whether Rfa2p is bound in vivo to telomeric DNA. We first used a one-hybrid assay for telomere-binding proteins20 in which Rfa2p was fused to a transcriptional activation domain and tested for its ability to activate transcription of a telomere-linked, promoter-defective HIS3 allele. We compared transcriptional activation of the telomeric HIS3 reporter by the Rfa2p fusion protein to transcriptional activation by similar fusion proteins of Rif2p (Rif2p binds telomeres through its interaction with the telomeric DNA binding protein Rap1p) and of Cdc13N (the N-terminal domain of Cdc13p, which does not contain the DNA-binding domain). The Rfa2p fusion protein induced transcriptional activation of the telomere-linked HIS3, indicating that Rfa2p is a telomere-binding protein (Fig. 1a). The Rfa2p fusion protein produced weaker transcriptional activation than the Rif2p fusion protein but stronger activation than the Cdc13N fusion protein. Unlike Rif2p, Rfa2p did not activate the HIS3 reporter gene when it was adjacent to an internal tract of telomeric sequences (Fig. 1a). This indicates that in vivo binding of Rfa2p may require the presence of a chromosome end.

Figure 1: Rfa2p binds to telomeric DNA in vivo.
figure 1

(a) Rfa2p fusion protein activates HIS3 at a telomere but not at an internal tract of telomeric DNA. Proteins fused to the B42 activation domain were expressed in strains (YM701 background) in which the HIS3 promoter–defective allele was inserted, either adjacent to an internal tract of telomeric DNA (HIS-Int-CA) or immediately adjacent to the left telomere of chromosome VII (HISTel)20. Cells were spotted onto test plates (−His+3AT) or control plates (−Trp). (b) Scheme showing the fragmented URA3 gene at the left arm of chromosome VII (left), the telomere at the right arm of chromosome VI (ChR-VI; right) and BDF1 at chromosome XII (top). Primer pairs used to amplify telomeric fragments pUra1 and pUra2, Chr-VI-R telomere fragment pTelVI and control nontelomeric fragment pBdf1 are indicated. The ChIP assay was carried out with antibodies to Myc (9E10) on chromatin from the indicated UCC1001 strains. After DNA purification, PCR was carried out with telomeric primers (pUra1, pUra2 and pTelVI) and nontelomeric control primers (pBdf1). (c) Association of Rfa2p with telomeric DNA increases in S phase. Right, FACS analysis of the synchronized UCC1001 cells after release from α-factor-induced G1 block. Left, results of ChIP assays on Myc-tagged Rfa2p, carried out on chromatin from cells harvested in exponentially growing (exp), G1 phase, S phase and G2 phase cells after release from an α-factor block. Relative enrichment of Rfa2-Myc-bound telomeric DNA (pUra2) over background (pBdf1) was calculated as indicated in Methods. The relative amount of telomeric DNA in the precipitates of exponentially growing cells was arbitrarily taken as 1.

Full size image

To confirm the association of Rfa2p with telomeric DNA, we carried out chromatin immunoprecipitation (ChIP) assays. We used cells carrying the URA3 reporter inserted next to the left telomere of chromosome VII and expressing a Myc-tagged version of Rfa2p (UCC1001 RFA2-MYC). We considered the Rfa2-Myc fusion protein to be functional: the RFA2-MYC cells behaved like wild-type cells with respect to growth, DNA replication, biological responses to DNA damage and telomeric position effects, although their telomeres were slightly (30 bp) longer than those of wild-type cells (data not shown). Chromatin from RFA2-MYC cells was cross-linked, immunoprecipitated and analyzed by PCR. Rfa2-Myc immunoprecipitates allowed amplification of the telomeric PCR products pUra1 and pUra2, or pTelVI, but not of the control pBdf1 PCR fragment (Fig. 1b). When cells without the Myc-tagged Rfa2p were subjected to the same ChIP assay, or when immunoprecipitation was done in the absence of in vivo formaldehyde cross-linking, we did not detect any PCR amplification (Fig. 1b), demonstrating the specificity of the ChIP.

Next we addressed whether Rfa2p showed a cell cycle–regulated association with telomeres. We released cells expressing the chromosomally encoded Rfa2-Myc18 from an α-factor-induced G1 block and monitored their progression through the cell cycle by FACS analysis (Fig. 1c). Cells in G1, S and G2 phases were subjected to ChIP (Fig. 1c). Amounts of DNA in the input samples and immunoprecipitates were measured by real-time PCR and normalized to the level of the nontelomeric pBdf1 fragment presents in the immunoprecipitates. PCR quantification analysis indicated that the amount of telomeric DNA bound to Rfa2p in cells harvested in S phase was four times higher than in cells harvested in G1 phase (Fig. 1c). Data showing enrichment at telomeres, as compared to a nontelomeric control of Rfa1p and Rfa2p during the cell cycle, is described below.

Rfa2p is involved in regulation of telomere length

We had previously constructed a mutant strain with a chromosomally encoded, truncated form of Rfa2p lacking 38 amino acids from the N terminus8. We introduced the mutated allele, rfa2Δ40, into yeast strain UCC1001, which carries a subtelomeric URA3 reporter gene, and examined the median length of the left telomere of the fragmented chromosome VII-L. Telomeres in the rfa2Δ40 strain were critically short (220 bp shorter than wild-type telomeres; Fig. 2a). We next asked whether this telomere shortening was accompanied by an increase in the amount of single-stranded G-strands in exponentially growing cells, as is the case for cells deficient for Ku (yku)21, a DNA-binding heterodimer that is crucial in DNA end protection. In wild-type cells, telomeric single-stranded extensions cannot be detected with telomeric probes in the native DNA of cells outside S phase12. This contrasts with yku mutants, which show long terminal extensions of the G-rich strand. We did not detect long G-tails in the rfa2Δ40 cells, although we detected them in yku70 mutants. Natural telomeres were also shortened in rfa2Δ40 cells (Fig. 2b).

Figure 2: Rfa2p is involved in telomere length regulation.
figure 2

(a) Comparison of the telomere length of haploid strains isogenic to UC1001 grown for more than 100 generations carrying a subtelomeric URA3 reporter gene on the left arm of chromosome VII. Median telomere length was calculated from two independent experiments using ImageQuant (Molecular Dynamics). (b) The rfa2Δ40 mutation does not lead to an increase in single-stranded telomeric G-rich DNA. Genomic DNAs isolated from the indicated strains were digested with XhoI and analyzed by nondenaturing hybridization (native) followed by denaturation of the DNA (denatured) in the same gel and rehybridization with the same telomeric C1–3A/TG1–3 probe as described21. nt, nucleotide; ss, single-stranded; WT, wild type.

Full size image

RPA acts in the telomerase pathway

The length of the single-stranded overhang has been proposed to correlate with the rate of telomeric DNA degradation22; thus, the absence of an expanded 3′ overhang in rfa2Δ40 suggested that this allele does not affect the rate of telomeric DNA degradation. To address this prediction directly, we compared telomere length dynamics in Rfa2p and Est2p mutants. We constructed a heterozygous diploid strain (JKM179 background) carrying the rfa2Δ40 N-terminal truncation allele and a deletion allele of EST2, est2Δ. The diploid strain RFA2/rfa2Δ40; EST2/est2Δ:NAT, ADH4/TELadh4:URA3 was sporulated, and the growth reduction with increasing generations and the rate of telomere shortening at the adh4:URA3 telomeres were measured in cell cultures established from different spores. The est2Δ/rfa2Δ40 double mutant showed a growth reduction over generations comparable to that of the est2Δ single mutant, whereas growth reduction of the single rfa2Δ40 mutant was less severe (Fig. 3a). We did not notice clear signs of senescence in rfa2Δ40 cells, which can grow indefinitely, although at a low growth rate. In agreement with this observation, the short mean length seemed stable in long-term culture. However, after several streakings we noticed a high frequency of rfa2Δ40 cells with recombined telomeres and higher growth rates. Thus, telomeres were not fully stable in rfa2Δ40 cells.

Figure 3: RFA2 acts in the telomerase pathway.
figure 3

(a) Viability of the est2Δ, rfa2Δ40 and est2Δ rfa2Δ40 strains with increasing number of generations (gen.). Shown are successive streak cultures on YPD medium of spores generated from a single tetrad. (b) Telomere degradation rates. Telomere length of the indicated yeast strains (JKM179 background) were plotted against generation and telomeric degradation rates were calculated from linear regression curves. (c) Deleting TEL1 does not aggravate telomere shortening of the rfa2Δ40 mutant. Telomere length of the indicated spores was measured after 3 streakings. Left, the URA3 genetically marked telomere was analyzed as in Figure 2a. Right, lengths of Y' telomeres. Telomere length was measured by reference to molecular weight markers (SmartLadder, Eurogentec).

Full size image

We next quantified the rate of telomere shortening in growing cells harboring the same three mutation combinations. As expected, in cultures from wild-type spores, telomere length remained constant (data not shown); in cultures from spores lacking EST2, median telomere length decreased linearly at a rate of roughly 3 bp per generation (Fig. 3b). Notably, rates of telomere shortening were identical in cultures from est2Δ spores with and without the rfa2Δ40 allele (Fig. 3b). Variation in initial telomere length between individual spore-derived cultures is likely to reflect differences in telomere length in the initial diploids or differences in the number of generations since the initial diploids. Initial length is not expected to influence the degradation rate16. We concluded that the rfa2Δ40 allele does not increase the rate of telomeric DNA degradation resulting from the loss of telomerase. Indeed, the rate of telomere shortening in the rfa2Δ40 single mutant was 2.3–2.4 bp per generation (Fig. 3b), substantially lower than the 3 bp per generation in cells lacking EST2. This, together with the fact that rfa2Δ40 cells show a reduction of the median telomere length rather than a total loss of telomeric DNA, indicates that telomerase action in the rfa2Δ40 mutant is reduced but not abolished. Overall, we conclude that RFA2 is important for telomerase action.

The results described above prompted us to analyze the genetic interactions between RFA2 and the ATM homolog TEL1, which has been proposed to regulate telomerase activity23. We first sporulated and dissected a diploid RFA2/rfa2Δ40; TEL1/tel1; ADH4/TELadh4:URA3 strain. After two streakings starting from a spore and an overnight culture, we compared telomere lengths in isogenic tel1/rfa2Δ40, tel1 and rfa2Δ40 mutants for both the genetically marked URA3 telomeres and telomeres carrying a conserved XhoI site in their Y' element (Fig. 3c). Although the sizes of the natural telomeres were similar in rfa2Δ40 and tel1 mutants, the URA3-marked telomeres were reproducibly shorter in rfa2Δ40 cells than in tel1 cells (by about 110 bp). Such differences between Y' telomeres and genetically marked telomeres have been reported elsewhere24. Most notably, the genetically marked and natural telomeres in rfa2Δ40 and rfa2Δ40 tel1 cells were almost identical in size (≈10 bp difference; Fig. 3c). We concluded that deleting TEL1 does not exacerbate the telomere shortening in rfa2Δ40 mutant cells.

Telomeric binding of RPA in wild-type and rfa2Δ40 cells

We investigated whether the binding of Rfa1p and Rfa2p to telomeres was altered in rfa2Δ40 cells. We carried out ChIP experiments at different stages of the cell cycle in RFA1-MYC and RFA1-MYC/rfa2Δ40 cells, using polyclonal antibodies to Rfa2 and monoclonal antibodies to Myc to monitor the binding of Rfa2p and a functional Rfa1-Myc fusion, respectively (Fig. 4). In the course of the cell cycle, the truncated version of Rfa2p bound to telomeres much like full-length Rfa2p. Like Rfa2p, Rfa1p bound to telomeres with a maximum in late S phase, and this was true in the rfa2Δ40 strain as well as in the wild type (Fig. 4). Thus, rfa2Δ40 does not alter RPA binding to telomeres. We also found that Rfa2p was coimmunoprecipitated with Rfa1p-Myc and vice versa, and that Rfa2Δ40p was also coimmunoprecipitated with Rfa1p-Myc and vice versa (data not shown). We therefore concluded that Rfa1p and Rfa2p both bind to telomeres with similar kinetics and that truncating the N terminus of Rfa2p does not affect the telomeric binding of Rfa2p and Rfa1p. Consistent with these results, Rfa2Δ40p still interacts with Rfa1p. Taken together, these data indicate that the role of RPA in telomere maintenance is most likely to involve the whole RPA heterotrimer. It is likely that an activity of RPA other than DNA-binding activity is responsible for the telomeric phenotype shown by rfa2Δ40 mutant cells.

Figure 4: Telomeric association of Rfa2p (or Rfa2Δ40p) and Rfa1p in wild-type (WT) and rfa2Δ40 cells.
figure 4

(a) FACS analysis of the synchronized UC1001 cells after release from α-factor-induced G1 block. (b) Amounts of telomeric DNA (pUra2 or Tel) determined by ChIP assays of chromatin from cells harvested every 15 min after release from an α-factor block. The amount of DNA was calculated by quantitative real-time PCR. Relative enrichment of Rfa2p- and Rfa1-Myc-bound telomeric DNA (pUra2/Tel) over background (pBdf1) is plotted against time. Amounts of Rfa2p, Rfa2Δ40p or Rfa1p-Myc immunoprecipitated protein at each time point are shown.

Full size image

Binding levels of Rfa1-Myc (anti-Myc ChIP) and Rfa2p (anti-Rfa2p ChIP) to telomeric DNA in G1 phase was five and ten times, respectively, that for the nontelomeric genomic control (Fig. 4). By comparison, under the same ChIP conditions, telomeric binding of Myc-tagged Est1p, Cdc13p and Est2p in G1 phase was 4, 4.6 and 36 times, respectively, that of the control (Fig. 5). We therefore believe that the amount of RPA bound to telomeres in G1 phase is functionally important though low. The difference in the telomeric binding of Rfa2p and Rfa1-Myc may reflect a slightly higher background binding when antibodies to Rfa2p, as compared to antibodies to Myc, are used.

Figure 5: The rfa2Δ40 allele impairs Est1p telomeric DNA binding during S phase.
figure 5

(a) FACS analysis of the indicated UC1001 cells after release from α-factor-induced G1 block. (b) Telomeric DNA binding of Est1-myc, Est2-myc, Cdc13-myc and yKu80-myc was monitored by ChIP assays (anti-Myc (9E10)) as described in Figure 4. Relative enrichment of Est1-myc, Est2-myc, Cdc13-myc and yKu80-myc-bound telomeric DNA over background is plotted against time. Immunoprecipitated proteins at each time point were analyzed by western blotting with anti-Myc (9E10).

Full size image

Rfa2p is required for telomeric binding of Est1p during S phase

We next assessed, using ChIP assays, whether the binding of Cdc13p, Est1p and Est2p to telomeric DNA is impaired in rfa2Δ40 cells. A Myc-tagged version of CDC13, EST1 or EST2 was introduced into wild-type and rfa2Δ40 strains bearing the URA3-marked telomere. Overall, the tagged proteins were functional. The tagged strains were synchronized in G1 phase with α-factor and released into the cell cycle. We monitored cell-cycle progression of both the wild-type and rfa2Δ40 strains by FACS analysis, collecting samples every 15 min for ChIP assays. The amounts of telomeric DNA bound to Cdc13p and Est2p did not differ notably between the strains (Fig. 5). In both strains, telomeric binding of Cdc13p and Est2p increased in S phase (Fig. 5), although a proportion of Est2p is already associated with the telomeres in G1 phase (ref. 18). By contrast, the increase in telomeric binding of Est1p that normally occurs in S phase was severely impaired in the rfa2Δ40 cells (Fig. 5). We concluded from these data that the rfa2Δ40 mutation results in a specific defect in Est1-Myc telomeric binding in S phase.

To assess whether the Est1p telomeric binding defect observed in rfa2Δ40 cells would not simply result from a shortened or perturbed telomere, we analyzed the telomeric binding of yKu80p by the same method using a yKu80-Myc fusion. A perturbation of yKu at the telomere might be predicted to lead to a loss of Est1p recruitment25. Similar yKu80p binding occurred in the wild type and the rfa2Δ40 mutant. In both strains, yKu80p bound to telomeres throughout the cell cycle, with a slight increase in S phase (Fig. 5). Thus, the defect in Est1p recruitment in rfa2Δ40 cells does not seem to be related to a perturbation of yKu80p localization to the telomere. This additive control supports a direct role of RPA in Est1p access to telomeres.

Following the same logic, we next analyzed whether the binding of Cdc13p, Est1p and Est2p would be affected in a yku70Δ mutant that has short telomeres (Fig. 6). We observed a slight decrease in the binding of Est2p to mutant telomeres in G1 and in S phase in the yku70Δ mutant, whereas binding of Cdc13p to telomeres in S phase was increased. Binding of Est1p to telomeres in S phase is slightly decreased in the yku70Δ mutant, whereas binding of Est1p to telomeres in S phase is severely decreased in the rfa2Δ40 mutant (compare Fig. 5b, wild-type and rfa2Δ40, with Fig. 6b). From these results, we believe that the decrease of Est1p binding to telomeres is not an indirect consequence of telomere shortening but rather results from a direct role of RPA in enabling access of Est1p to telomeres.

Figure 6: Association of Est1-myc, Est2-myc and Cdc13-myc in yKu70Δ cells.
figure 6

(a) FACS analysis of the indicated UC1001 cells after release from α-factor-induced G1 block. (b) Telomeric DNA binding of Est1-myc, Est2-myc and Cdc13-myc was monitored by ChIP assays as described in Figure 5. The wild-type control experiment is illustrated in Figure 5b.

Full size image

rfa2Δ40 telomere length when Est1p is tethered to the telomere

That rfa2Δ40 cells show a strong defect in the telomerase pathway of telomere length regulation and an altered binding of Est1p to telomeres indicates that RPA may facilitate loading of Est1p to promote telomerase action at telomeres. If this notion is correct, one can anticipate that the rfa2Δ40 mutation would be suppressed if telomeric binding of Est1p were forced by expression of a hybrid protein consisting of Est1p and the DNA-binding domain of Cdc13p. We transformed wild-type and rfa2Δ40 cells with the plasmid pVL1120, which directs the expression of such a hybrid protein, Est1p-DBDCDC13 (ref. 14). Efficient telomere lengthening occurred in the rfa2Δ40 mutant (>200 bp) (Fig. 7), and the growth defect of rfa2Δ40 cells was suppressed by the transformation with pVL1120 (data not shown). This occurred for every clone we analyzed, although the extent of telomere elongation varied between transformants. These results confirmed that Cdc13p retains the ability to bind to telomeric DNA in rfa2Δ40 mutants (as discussed above) and that the telomere length defect associated with the rfa2Δ40 mutation can be corrected by forcing the loading of Est1p onto the telomere.

Figure 7: Telomere lengthening by targeted Est1p in rfa2Δ40 cells.
figure 7

Telomere length was measured by Southern blotting using either ura3 (left) or Y' probe (right). pVL1120, which directs the expression of Est1-DBDCDC13 (ref. 14), derives from pRS314. Telomere length was measured by reference to molecular weight markers (SmartLadder, Eurogentec). WT, wild type.

Full size image

Discussion

We have shown that RPA is present at telomeres and that truncation of the N-terminal region of Rfa2p greatly reduces telomere length. Because RPA participates in most aspects of DNA metabolism as a single-stranded DNA-binding protein, binding of RPA to telomeres might seem to be related to the presence of single-stranded telomeric DNA (G-tail). Consistent with this hypothesis, during the S phase of the cell cycle there is a general coincidence between the maximal telomeric binding of RPA described here, G-tail formation and telomere elongation. Based on all these results, we propose that RPA facilitates telomerase action during S phase.

Our ChIP analysis showed no notable differences between wild-type and rfa2Δ40 cells in the amount of Cdc13p and Est2p associated with telomeres. In contrast, binding of Est1p in S phase is severely impaired in rfa2Δ40 cells. Although the question was not directly addressed in this study, it is likely that Cdc13p is also required for proper Est1p loading during telomerase activation14,15,18. This suggests that telomere shortening in the rfa2Δ40 mutant is not due to reduced binding of Cdc13p and Est2p to telomeric DNA, but rather reflects a decrease in the binding of Est1p to telomeres. This theory is in agreement with the fact that artificial tethering of Est1p to telomeres (via a Est1p-DBDCDC13 fusion protein) corrected the rfa2Δ40-associated defects. The binding of Est2p to telomeres seems to be independent of proper loading of Est1p. This is consistent with the idea that the role of Est1p is not to recruit Est2p to telomeres but rather to activate an already bound form of telomerase18. We have been unable to coimmunoprecipitate Est1p (or Cdc13p or Est2p) with Rfa2p (data not shown), indicating that control of telomerase activity by Rfa2p may not be mediated by a stable interaction between Rfa2p and Est1p. Either the putative interaction between RPA and Est1p is too infrequent, too transient or both to be detected, or RPA does not directly recruit Est1p but rather maintains single-stranded DNA in a state amenable to Est1p binding. For instance, RPA could act through the stimulation of a helicase activity, opening up the duplex portion of the telomeric DNA and thereby facilitating the binding of Est1p. Overall, we propose that RPA and Cdc13p cooperate to allow Est1p binding to telomeres and subsequently promote telomerase action (Fig. 8). The rfa2Δ40 allele seems to be specifically impaired in the Est1p loading step (Fig. 8).

Figure 8: Model of telomerase action during the cell cycle.
figure 8

Left, in G1 phase, a proportion of the catalytic subunit of telomerase (Est2p) is bound at the telomere in an inactive form18. In late S phase, the 3′ overhang lengthens12 and RPA, Est1p and Cdc13p become bound to the telomere18,19 (see text). Binding of Est1p depends on both Cdc13p14,15,18 and RPA, whereas binding of RPA and Cdc13p can occur independently (see text). The complex formed between the 3′ overhang, RPA, Cdc13p and Est1p is required for telomerase action at the chromosome end. The interactions between RPA, Est1p, Est2p and Cdc13p are likely to be transient, indirect or both, as the complex has not been identified in coimmunoprecipitation experiments. Notably, in G1, we found a basal level of RPA binding to telomeres (not shown in this model). Right, in the presence of the rfa2Δ40 mutation, the RPA complex (Δ40) binds to the 3′ overhang but is unable to load Est1p. Consequently, the telomerase stays in its inactive form despite the proper fixation of Cdc13p. Because residual telomerase activity can be detected in rfa2Δ40 cells, one can imagine that a weak loading of Est1p can still occur.

Full size image

Our results do not rule out roles for RPA in other aspects of telomere length regulation, including the control of telomeric DNA degradation and the coupling of telomerase activity to lagging-strand synthesis17,26. RPA has consistently been shown to interact with the DNA polymerase α–primase complex27. In addition, we found that the rfa1-M2 allele28 of the large subunit of RPA (RFA1) results in telomere elongation (data not shown), indicating a possible role of RPA in C1–3A strand synthesis by DNA polymerase α.

We have found that deleting TEL1 in rfa2Δ40 cells does not aggravate telomere shortening, indicating that RPA and Tel1p may act in the same pathway. Notably, the phenotypes of rfa2Δ40 cells show similarities with the telomeric phenotypes of cells devoid of Mec1p and Tel1p: (i) rfa2Δ40 cells and cells lacking Tel1p and Mec1p have residual telomerase activity (as described here and in refs. 29,30); (ii) the association of Cdc13p with telomeres occurs efficiently in tel1Δ cells and rfa2Δ40 cells (described here and in ref. 31); and (iii) forcing the Cdc13p-dependent recruitment of telomerase to the telomeres suppresses the growth defects of both rfa2Δ40 and mec1 tel1 cells (described here and in ref. 31). Another link between Rfa2p, Mec1p and Tel1p is provided by the fact that Rfa2p is phosphorylated during S phase mainly by Mec1p but also by Tel1p (ref. 5). We can speculate from these data that RPA could be one of the targets of these kinases at telomeres, although a very recent report indicates that substituting alanine residues for the two serines (Ser122 and Ser238) of Rfa2p that are in SQ motifs (preferred targets of the ATM kinase family) does not affect telomere length32. Notably, the same study showed that the rfa2-55 allele (encoding E90A and D91A substitutions produces short telomeres32,33.

The control of telomerase activity by two single-stranded DNA–binding proteins, Cdc13p and RPA, may provide a means to couple telomerase activity tightly to the formation of 3′ overhangs during telomere replication. Thus, RPA might be a crucial link between telomere homeostasis and the replication of chromosome ends. The role of RPA in the loading of Est1p at telomeres may be evolutionarily conserved, as both RPA and Est1p are present in human as well as yeast cells2,34,35.

Methods

Mutant strain construction and telomere experiments.

A deletion removing the RFA2 intron and the region extending from codon 3 to codon 40 of RFA2 (rfa2Δ40) was introduced as described8. By replacing the chromosomally encoded RFA2 with the mutated rfa2Δ40t allele, we generated an allele (rfa2-165) encoding a truncated protein extending from residue 1 to residue 165 that is coexpressed with Rfa2pΔ40. To get rid of potential effects of rfa2-165, this inviable, truncated allele was fully disrupted with a kanMX4 marker8. Rfa2p was tagged at its C terminus with 18 MYC epitope with pGG1 (ref. 8). Epitope-tagged Cdc13p, Est1p and Est2p were constructed to retain their endogenous promoters. Cdc13p and Est2p were 18× Myc epitope–tagged in their N-terminal portions using pVL1042 and pVL1001 linearized plasmids (kindly provided by V. Lundblad). Est1p was 13× Myc epitope–tagged in its C-terminal portion using one-step PCR. The functional RFA1-MYC18 allele was obtained from T. Tanaka36. We carried out silencing assays, telomere length analysis by Southern blotting, G-strand extension at telomeres and synchronization as described8,16,21. We carried out most of the experiments in yeast strains isogenic to UCC1001 (MATa ade2-101 his3Δ200 leu2Δ1 trp1Δ1 ura3-52 TELadh4:URA3).

We combined the rfa2Δ40 mutation with est2Δ as follows: we crossed the yeast strain JKM179 MATα TELadh4:URA3 est2Δ (grown for 25 generations after the disruption of EST2) to JKM139 MATa rfa2Δ40 to produce the diploid JKM179/JKM139 (MATα /MATa EST2/est2Δ RFA2/rfa2Δ40 ADH4/TELadh4:URA3 Δho/Δho Δhml:ADE1/Δhml:ADE1 Δhmr/Δhmr ade1-100/ade1-100 leu2-3,112/leu2-3,112 lys5/lys5 trp1:hisG/trp1:hisG ura3-52/ura3-52 ade3:GAL:HO/ade3:GAL:HO). Integration of a PCR product containing the nourseothricin-resistance (NAT) gene amplified from pAG25 (a kind gift from P. Goldstein and J. McCusker) disrupted EST2. We streaked the diploid strain several times in rich medium before sporulation, then identified spores with the appropriate genotypes and determined their telomere lengths as described16.

Southern blotting.

For Southern blots, EcoRV- and HindIII-digested genomic DNAs were resolved in 1.2% agarose gel and transferred to Hybond-N+ membrane (Amersham Biosciences). We probed the digested genomic DNAs with a radiolabeled URA3 DNA fragment. For Y' Southern blots, XhoI-digested genomic DNAs were resolved in 1.2% agarose gel and transferred to Hybond-N+ membrane and probed with a radiolabeled Y' DNA fragment.

Chromatin immunoprecipitations assays.

We carried out ChIP assays as described8 in yeast strains isogenic to UCC1001. Immunoprecipitation of cross-linked DNA was done with agarose-conjugated 9E10 monoclonal antibodies (to Myc; Santa Cruz Biotechnology) or with polyclonal antibodies to Rfa2p (kindly provided by S. Brill, Rutgers University, Piscataway, New Jersey and B. Stillman, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, respectively). The results obtained with anti-Rfa2p antibodies respect to the telomeric binding of Rfa2p are similar to those described obtained with the RFA2-MYC strain with anti-Myc 9E10 antibodies. We used primer pairs specific for the fragmented URA3 to amplify the 400-bp pUra2 and primer pairs specific for the telomeric TG1–3 repeats and URA3 to amplify the 230-bp pUra1. The chromosome VI-R telomere-specific primers (SG355 and SG356), which have been described previously37, produced a 400-bp fragment located 0.6 kb from the TG repeat. Control primers specific for the BDF1 control sequence produced a 100-bp PCR fragment. (All primer sequences are available on request.) We calculated the amounts of telomeric and nontelomeric DNAs in input and immunoprecipitated samples by quantitative PCR with a LightCycler PCR (Roche) as described8. Relative enrichment of telomeric DNA (pUra2) over background (pBdf1) in the immunoprecipitates was calculated as follows: (IPa pUra2/IPa pBdf1) × (inputa pBdf1/inputa pUra2); IPa and inputa represent the amount of PCR product in the immunoprecipitates and the input samples, respectively.

Yeast one-hybrid assay for telomere-interacting proteins.

Yeast strains were derived from YM701 (MATa ura3-52 his3-200 ade2-101 lys2-801 trp1-901 tyr1). Strains HIS-Int-CA and HIS-Tel and plasmids pJG4-5/RIF2 and pJG4-5/CDC13N were kindly provided by V. Zakian20. We cloned RFA2 into pJG4-5 as an XhoI DNA fragment. To monitor telomere interaction, we assayed cells on 3% Gal −His test plates containing 5 mM 3-amino-1,2,4-triazole (3-AT). Control plates were 3% Gal −Trp. Plates were incubated at 30 °C for 5 d.