The telomeric Cdc13 protein interacts directly with the telomerase subunit Est1 to bring it to telomeric DNA ends in vitro

Abstract

In Saccharomyces cerevisiae, a Cdc13–Est1 interaction is proposed to mediate recruitment of telomerase to DNA ends. Here we provide unique in vitro evidence for this model by demonstrating a direct interaction between purified Cdc13 and Est1. The Cdc13–Est1 interaction is specific and requires the in vivo defined Cdc13 recruitment domain. Moreover, in the absence of this interaction, Est1 is excluded from telomeric single-stranded (ss)DNA. The apparent association constand (K_d) between Est1 and a Cdc13-telomeric ssDNA complex was ∼250 nM. In G2 phase cells, where telomerase is active, Cdc13 and Est1 were sufficiently abundant (∼420 and ∼110 copies per cell, respectively) to support complex formation. Interaction between Cdc13 and Est1 was unchanged by three telomerase-deficient mutations, Cdc13^E252K (cdc13-2), Est1^K444E (est1-60), and Cdc13^S249,255D, indicating that their telomerase null phenotypes are not due to loss of the Cdc13–Est1 interaction. These data recapitulate in vitro the first step in telomerase recruitment to telomeric ssDNA and suggest that this step is necessary to recruit telomerase to DNA ends.

Telomeres, the specialized nucleoprotein structures that cap the ends of linear chromosomes, are essential for genome integrity and hence cell viability because they protect chromosome ends from fusions and degradation. They also counter the progressive loss of terminal sequences due to the inability of the conventional DNA polymerase to replicate the very ends of linear DNA molecules. In most eukaryotes, telomeres are maintained by the enzyme telomerase, a conserved reverse transcriptase that extends telomeric DNA using an integral RNA component as the template. In the absence of telomerase activity, most cells undergo progressive telomere shortening, eventually stop dividing, and senesce (1). In the budding yeast Saccharomyces cerevisiae, EST2 and TLC1 encode the telomerase reverse transcriptase subunit (2, 3) and RNA template (4), respectively. Additionally, EST1 and EST3, each of which encodes a telomerase accessory factor, and CDC13, which encodes the sequence-specific telomeric single-stranded (ss)DNA binding protein (5–7), are needed for telomere maintenance in vivo (8). Deletion or mutation of any of these genes leads to an ever-shorter telomeres phenotype (1, 8) that is shared with est2Δ and tlc1Δ mutants (4).

The replication of telomeres is tightly controlled during the cell cycle. After semiconservative replication of telomeric DNA, which takes place in late S phase, telomeres are processed to generate transient, long, single-stranded G tails (9). Telomerase action also takes place in late S/G2 phase (10, 11), coincident with an increase in telomere association of the essential telomerase components Cdc13, Est1, Est2, and Est3 (12–14). Among the essential components of telomerase, only Cdc13 is not part of the telomerase holoenzyme (15) but rather arrives at telomere ends presumably via its high affinity and selectivity for TG-rich ssDNA (5, 6). Cdc13 telomere association is independent of and presumably before all other telomerase components, as the chromatin immunoprecipitation (ChIP) signal for Cdc13 at telomere ends or at a de novo telomere addition site can be detected in a number of telomerase null strains [tlc1Δ (13, 16), cdc13-2 (13, 16), or est1-60 (17)], and Cdc13 levels at telomeres do not increase with increased telomere association of Est1 or Est2 (18). On the other hand, the late S/G2 phase assembly of the telomerase holoenzyme at telomere ends is strongly dependent on Est1, a cell cycle-controlled recruiter and activator of telomerase (16).

Even though telomerase action is restricted to late S/G2 phase in vivo (10, 11), telomerase activity can be detected in extracts from both G1 and G2/M phase cells in vitro (10). In addition, despite their critical importance in vivo, Est1, Est3, and Cdc13 are dispensable for telomerase activity in vitro (19). Moreover, Cdc13 and its functional counterpart in humans and Schizosaccharomyces pombe, POT1, inhibit telomerase activity in vitro (20–22). Given the high affinity of Cdc13 for telomeric ssDNA, Cdc13 may hamper telomerase action in vivo by competing with telomerase for binding to telomeric ssDNA. The current model to explain these data is that, in vivo, Est1 brings telomerase to Cdc13-occupied ssDNA. This hypothesis can explain the discrepancy between the in vivo and in vitro requirements for Cdc13 and Est1. Moreover, it is consistent with current models indicating that telomerase action is regulated at the level of telomere access (18).

Genetic evidence strongly implies that telomerase recruitment to telomeres is achieved by a specific interaction between Cdc13 and Est1. In vivo interaction between Cdc13 and Est1 has been demonstrated by yeast two-hybrid and coimmunoprecipitation analyses of overexpressed proteins (23–25). Fusion of Cdc13 and Est2 bypasses the need for EST1 to maintain stable telomeres (26). The in vivo defined Cdc13 recruitment domain (RD) is localized to amino acids 211–331 (27). Furthermore, a “charge-swap” mutant of Cdc13, cdc13-2 (Cdc13^E252K), a mutation within the RD, confers a telomerase-null phenotype on its own (6, 8) but is suppressed by a charge-swap allele of Est1, est1-60 (Est1^K444E) (27). These results suggest that interaction between Cdc13 and Est1 is supported by the electrostatic attraction of a specific Lys-Glu pair (27). Consistent with this interpretation, Est1 binding is low in cdc13-2 cells at both telomeres (16) and double-strand breaks (17). However, inconsistent with the aforementioned hypothesis, the in vivo biochemistry data showed that Est1 interacts equally well with both wild-type (WT) Cdc13 and Cdc13^E252K (23), and Est1^K444E binds telomeres as well as WT Est1 (28).

The yeast TEL1 and MEC1 checkpoint kinases, the homologs of ATM and ATR, respectively, are also involved in regulating telomerase action. Deletion of TEL1 or MEC1 leads to stably short (29) or near WT-length (30) telomeres, respectively, whereas the tel1Δ mec1Δ double mutant is unable to maintain telomeres (30). Fusion of Cdc13 to Est2 bypasses the need for TEL1 and MEC1 for telomere maintenance, suggesting that TEL1 and MEC1 function in telomerase recruitment (7). Indeed, TEL1 is required for efficient Est1 and Est2 telomere association (31) and preferential elongation of at least some short telomeres (18, 32, 33). Cdc13 is thought to be a Tel1/Mec1 target as both kinases can phosphorylate N-terminal fragments of Cdc13 in vitro. Moreover, simultaneous mutation of two of the Tel1/Mec1 sites in Cdc13, Ser-249, and Ser-255, to alanine, leads to cellular senescence, a phenotype that can be rescued by expressing a Cdc13–Est1 fusion (34). These findings suggest that telomerase recruitment is controlled by Tel1- or Mec1-dependent phosphorylation of Cdc13 in the RD, with phosphorylated Cdc13 being more favorable for interaction with Est1. However, contrary to the expectations of this model, Ser255 phosphorylation is undetectable in Cdc13 purified from yeast, and simultaneous mutation of all of the SQ sites in Cdc13 to SA, where SQ is the Tel1 consensus sequence, does not lead to telomere shortening (25).

In this report, we took in vitro approaches to examine the Cdc13–Est1 interaction, a step central to telomerase recruitment and regulation in vivo. We provide unique evidence for a direct interaction between Cdc13 and Est1 and show that this interaction can support the first step in telomerase recruitment to DNA ends in vitro. However, mutant proteins that are defective in telomerase recruitment in vivo, Cdc13^E252K, Est1^K444E, and Cdc13^S249,255D_, had WT levels of Cdc13–Est1 interactions in vitro. We also determined the in vivo concentrations of Cdc13 and Est1 in both G1 (when telomerase is not active) and G2 (when it is). Only in G2 phase cells are the concentrations of the two proteins sufficiently high to support productive complex formation. Our results confirm and extend the current models on the molecular mechanisms that are needed to recruit telomerase to yeast telomeres.

Results

Purification and Characterization of Recombinant Cdc13 and Est1.

The DNA binding activity of Cdc13 has been thoroughly characterized by several research groups using recombinant proteins purified from Escherichia coli or insect cells (5, 6, 35). The DNA binding domain of Cdc13, which maps to residues 497–694 (36), binds to telomeric ssDNA with high affinity and sequence specificity (36, 37). However, proteins obtained from a heterologous host will likely be devoid of posttranslational modifications that are important for their function and regulation. We therefore overexpressed and purified full-length Cdc13 (hereafter called Cdc13^FL) and the DNA binding domain, Cdc13^DBD (amino acids 445–694) from its native host, S. cerevisiae to near homogeneity (Fig. S1A). Purification was performed in the presence of phosphatase inhibitors to preserve phosphorylation, and the affinity tag was removed by protease cleavage after purification. We also purified the N-terminal 455-aa (Cdc13^{N ter}) polypeptide containing the predicted Est1 RD from E. coli (Fig. S1A). For all proteins used in this study, we confirmed protein identity and integrity by liquid chromatography-mass spectrometry (LC-MS) and MS/MS analysis.

Using a filter binding assay, we showed that purified Cdc13^FL and Cdc13^DBD exhibited indistinguishable DNA binding activity for an 11-mer minimal consensus sequence (TEL11) across various salt and magnesium conditions at affinities comparable to previously reported values for recombinant Cdc13^FL purified from E. coli (6) or insect cells (35, 36, 38) (Fig. S1 B and C). In contrast, Cdc13^{N ter} did not bind the TEL11 substrate or a 43-mer longer substrate to any appreciable extent under the same conditions (Fig. S1 B and D). Our results confirm previously published findings that the Cdc13 DNA binding domain (DBD) alone is sufficient for high-affinity telomeric ssDNA binding (35–37).

We also purified full-length, C-terminally strep-tagged Est1 from S. cerevisiae (Fig. S2A). The sequence of Est1 from a YPH499 strain differed at eight amino acid positions from that from the Saccharomyces Genome Database (S288c strain) (Fig. S2B). We purified and carried out experiments with both the YPH499 and S288c versions of Est1 and obtained identical results. For simplicity, we show only the results obtained with the YPH Est1.

Consistent with findings by others (28, 39), our purified Est1 bound TG-ssDNA, and its affinity for TG-ssDNA increased as the length of the ssDNA increased (Fig. S2C). Furthermore, the strep-tagged Est1 supported WT-length telomeres in vivo when expressed from a CEN plasmid and its native promoter (Fig. S2D). On electrophoretic mobility shift assay (EMSA), the strep-tagged Est1-ssDNA complex remained in the well. This behavior is likely due to the high pI value of the protein (9.59), as ssDNA bound by a myc9-strep–tagged Est1 (pI 7.66) purified using the same strategy migrated into the gel (Fig. S3A). Furthermore, in a glycerol gradient experiment, strep-tagged Est1 sedimented in a manner expected for a largely monomeric protein (Fig. S3B). Therefore, strep-tagged Est1 was not aggregated in the presence or absence of ssDNA. We conclude that our purified Cdc13 (and its derivatives) and Est1 are functional by previously established criteria with the exception that we did not observe a G4 DNA promoting activity of Est1 reported by a previous paper (40).

Cdc13 and Est1 Interact Directly to Form a 1:1 Complex in Vitro.

In vivo studies suggest that Cdc13 and Est1 interact and this interaction is important to recruit telomerase to telomeres (17, 23, 26, 27). To determine whether this interaction is direct, we tested purified untagged Cdc13^FL and C-terminally tagged Est1 for their ability to interact in vitro, using magnetic beads pull-down experiments (Fig. 1A). Varying amounts of Cdc13^FL in the presence or absence of Est1 were mixed with streptavidin-coated magnetic beads that capture the C-terminal affinity tag of Est1 (Fig. 1, lane 1). Cdc13^FL alone was not pulled down by the beads (Fig. 1, lane 9). However, in the presence of Est1, Cdc13^FL was bead associated, and the amount of bead-associated Cdc13^FL increased as the concentration of input Cdc13^FL increased (Fig. 1, lanes 2–8). The nonspecific control, BSA, was not pulled down by the beads either in the presence or in the absence of Est1.

Fig. 1.

Cdc13 interacts directly with Est1 to form a 1:1 stoichiometric complex that brings Est1 to telomeric ssDNA. (A) Magnetic beads pull-down experiment between purified strep-tagged Est1 and untagged Cdc13^FL. Each experiment contained 0.5 μM of Est1 (lanes 1–8) and varying amounts of Cdc13^FL (lanes 1–9). Upper, beads fractions (“beads”); Lower, 20% of the input materials shown as a control (“20% input”). Positions of Cdc13^FL, Est1, and BSA are indicated on the right. (B) Beads-associated Cdc13^FL in the presence of 0.5 μM (solid squares) or 1 μM (open squares) Est1 was quantified against Cdc13^FL concentrations in the input. In this and subsequent figures, error bars represent 1 SD from at least three independent experiments.

The amount of Cdc13^FL brought down by the beads saturated at ∼0.56 μM (Fig. 1B, solid squares), which was the approximate concentration of Est1 in the reaction. To establish the stoichiometric relationship between these two proteins, we increased Est1 concentration by twofold and observed that the saturating amount of Cdc13^FL on the beads also increased to ∼1.1 μM (Fig. 1B, open squares). Thus, Cdc13^FL and Est1 form a complex at an apparent 1:1 ratio.

Cdc13–Est1 Interaction Is Specific.

To further characterize the Cdc13–Est1 interaction, we subjected Cdc13^{N ter} and Cdc13^DBD to the same magnetic beads pull-down experiment used for Cdc13^FL (Fig. 2B and Fig. S4). Neither Cdc13^{N ter} nor Cdc13^DBD were bead associated in the absence of Est1 (Fig. 2B, lanes 3 and 4). However, in the presence of Est1, Cdc13^{N ter} was pulled down by the beads (Fig. 2B, lane 6) with similar efficiency to Cdc13^FL (Fig. 2B, lane 5), whereas Cdc13^DBD was not bead associated (Fig. 2B, lane 7). This in vitro result confirms the in vivo findings (17, 27) that the Est1 RD of Cdc13 resides in its N terminus.

Fig. 2.

Cdc13 interacts with Est1 via its N-terminal RD and the interaction is specific. (A) Schematic illustration of Cdc13 fragments used. (B) A total of 0.5 μM strep-tagged Est1 was incubated with 0.5 μM untagged Cdc13^FL, Cdc13^Nter, or Cdc13^DBD in a magnetic beads pull-down experiment. Upper, beads fractions (“beads”); Lower, 20% of the input material as controls (“20% input”). (C) A total of 1 μM of strep-tagged Est1 was incubated with 1 μM untagged Cdc13 fragments in the beads pull-down assay. The beads fractions and 15% of the input material are shown as “beads” and “15% input”, respectively. An Est1 degradation product is indicated by an asterisk (*). (D) A total of 1 μM of strep-tagged Est1 or Est3 was incubated with 1 μM untagged Cdc13^FL, RPA, or BSA. Experiments were performed as described in Materials and Methods, except BSA was substituted with 20 μg/mL aprotinin. The beads fractions and 15% of the input material are shown as “beads” and “15% input”, respectively. The same gels stained by SyproRuby for better detection are provided in Fig. S4.

The N terminus of Cdc13 contains two additional oligosaccharide-oligonucleotide binding (OB) folds in addition to the genetically defined RD (41). The first OB fold (OB1) is required for Cdc13 dimerization (41, 42) as well as Pol1 (the largest subunit of DNA polymerase α) interaction (23, 41). We therefore characterized Cdc13–Est1 interaction with a dimerization-defective mutant of Cdc13, L91R (41, 42) (Fig. S1E and Fig. 2C, lanes 4 and 5). Est1 interaction was not compromised in the Cdc13^L91R mutant, indicating that Cdc13 dimerization is not a perquisite for its interaction with Est1. We also tested two smaller fragments of Cdc13, Cdc13^1–340 (Fig. 2C, OB1-RD, lanes 6 and 7) and Cdc13^190–340 (Fig. 2C, RD, lanes 8 and 9) for Est1 interaction: Both variants were beads associated in the presence of Est1. This result suggests that the RD is necessary and sufficient for Est1 interaction.

To determine whether the interaction between Cdc13 and Est1 is specific, we compared it with the interaction between Est1 and RPA, the major cellular sequence nonspecific ssDNA binding protein (Fig. 2D). Because the largest subunit of RPA (Rfa1) migrates at the same position as BSA, we omitted BSA from the reaction. In separate control experiments, we showed that BSA was not bead associated either in the presence or in the absence of Est1 (Fig. 2D, lanes 4 and 7), whereas Cdc13^FL was bead associated at an ∼1:1 ratio of Cdc13 to Est1 (Fig. 2D, lane 5; ∼102% of the input levels). Reproducibly, a low level of Rfa1, the largest subunit of RPA, was bead associated in the presence of Est1 (Fig. 2D, lane 6; ∼18% of the input levels), suggesting a weak interaction between Est1 and RPA. Using a more sensitive stain, SyproRuby, we could also detect Rfa2 in the beads fraction (Fig. S4D), suggesting that Est1 interacts with the functional heterotrimeric complex of RPA.

As another test for specificity, we examined the ability of Cdc13^FL to interact with Est3, another telomerase subunit that, like Est1, is essential for telomerase action in vivo (8, 15). When E. coli-expressed strep-tagged Est3 was used to replace Est1 in the beads pull-down experiment (Fig. 2D, lanes 8 and 9), Cdc13^FL was not bead associated (Fig. 2D, lane 9). Using the same assay, purified Est3 interacted with Est1 (14), demonstrating that our purified Est3 is capable of forming specific interactions. Therefore, the interaction between Cdc13 and Est1 is specific.

Cdc13–Est1 Interaction Is Critical for Recruiting Est1 to Telomeric ssDNA.

We confirmed our observations on the specific Cdc13–Est1 interaction from the beads pull-down experiments with an EMSA (Fig. 3A). In this and subsequent EMSA experiments, we first formed a complex between Cdc13^FL and a 15-mer TG-ssDNA (TEL15) and then added purified Est1. We monitored formation of the Est1-Cdc13^FL-TEL15 tertiary complex by following the position of radioactively labeled TEL15 (Fig. 3A, Left, lanes 14–20). As reported previously (28, 39), Est1 did not bind efficiently to short oligonucleotides (28, 39) (K_d for Est1-TEL15 was 2.14 ± 0.31 μM) (Fig. 3A, Left, lanes 1–6 and Fig. 3A, Right, triangles). However, Est1 efficiently shifted the Cdc13^FL-TEL15 ssDNA complex, forming a tertiary complex in the well with a K_d of 252 ± 19 nM, ∼8.5 times lower than that for TEL15 alone (Fig. 3A, Right, squares). Thus, not only do Cdc13 and Est1 interact directly with each other, but also this interaction promotes Est1 association with telomeric ssDNA.

Fig. 3.

Cdc13–Est1 interaction is important for Est1 recruitment to the telomeric ssDNA end. (A) A total of 50 nM of TEL15, alone or complexed with Cdc13^DBD (150 nM) or Cdc13^FL (100 nM), was incubated with varying amounts of Est1 in an EMSA. Left, a representative gel; Right, quantification. (B) Cdc13^{N ter} outcompetes Est1 for binding to the Cdc13^FL–TEL15 complex. A total of 50 nM TEL15 was complexed with 100 nM Cdc13^FL without (lane 1) or with (lanes 2–8) 500 nM Est1 to form a tertiary complex. After 5 min, varying amounts of Cdc13^{N ter} were added to the reaction to disrupt the cocomplex. Left, a representative gel; Right, quantification.

To determine whether Est1 interaction with Cdc13-TEL15 is physiologically relevant, we asked whether this association requires the Cdc13-RD. In contrast to the efficient tertiary complex formation observed with Cdc13^FL-TEL15, Est1 did not form a tertiary complex with Cdc13^DBD-complexed TEL15 (Fig. 3A, Left, lanes 7–13, and Fig. 3A, Right, open circles). Moreover, when Cdc13^{N ter} was added to the Est1-Cdc13^FL-TEL15 tertiary complex, the tertiary complex dissociated to yield the Cdc13^FL-TEL15 complex in a Cdc13^{N ter} concentration-dependent manner (Fig. 3B, Right, solid squares). Cdc13^{N ter} did not effectively dissociate a preformed Est1-TEL15 complex (Fig. 3B, Right, open circles). Both of these results indicate that the Cdc13–Est1 interaction is essential for Est1 recruitment to Cdc13-bound ssDNA. Because Cdc13 is bound to telomere ends in vivo throughout the cell cycle (13), our results imply that it is the specific interaction between Cdc13 and Est1, not DNA binding by Est1, that brings Est1 to telomere ends.

Est1 Has Similar Affinity for WT and Mutant Cdc13 in Vitro.

The EST4/cdc13-2 telomerase-defective allele is a point mutation in CDC13 that changes Glu-252 to Lys (8). Purified Cdc13^E252K binds telomeric ssDNA as well as WT Cdc13 (6). However, genetic experiments suggest that Cdc13^E252K is defective in telomerase recruitment (6, 26, 27). In a cdc13-2 background, Est1 is still telomere associated, but the levels of association are significantly reduced (16). The EST1 (K444E) allele est1-60 also confers a telomerase-null phenotype but suppresses the cdc13-2 phenotype in an allele-specific manner (17, 27). The charge-swap nature of these mutations led to the hypothesis that the telomerase null phenotypes of the two single mutants are due to their loss of the Cdc13–Est1 interaction.

To test whether the Est1 interaction is compromised by the cdc13-2 mutation, we purified full-length Cdc13^E252K and subjected it to the magnetic beads pull-down experiment (Fig. 4A, “EK”). In this assay, Cdc13^E252K interacted with Est1 as well as WT Cdc13^FL. In the EMSA experiment where affinity was quantified, Est1 had a K_d of 268 ± 26 nM for Cdc13^E252K, which is not significantly different from that for WT Cdc13 (Fig. 4B). In addition, we observed no interaction difference between Est1 with WT Cdc13 or Cdc13^E252K when the experiments were carried out at 200 mM NaCl, a condition in which the Est1–Cdc13 interaction was weakened (Fig. S5). Likewise, purified Est1^K444E (encoded by the est1-60 allele) interacted with WT and Cdc13^E252K equally well in bead pull-down experiments (Fig. 4C), with an observed K_d of 241 ± 38 and 245 ± 30 for WT and Cdc13^E252K-complexed TEL15, respectively, in the EMSA experiment (Fig. 4D).

Fig. 4.

Est1 interacts with Cdc13 mutants at strengths indistinguishable from that of the WT protein. (A) A total of 0.5 μM Strep-tagged Est1 was incubated with 0.5 μM untagged Cdc13^WT (WT), Cdc13^E252K (EK), or Cdc13^S249,255D (DD) in a magnetic beads pull-down assay. Upper, beads fractions (beads); Lower, 20% of the input materials are shown as control (20% input). (B) A total of 50 nM of TEL15, complexed with 100 nM of FL-Cdc13-WT, EK, or DD, was subjected to cocomplex formation with Est1 in EMSA experiments. Upper, a representative gel; Lower, quantification. (C) A total of 0.5 μM Strep-tagged Est1^K444E (KE) was incubated with 0.5 μM untagged Cdc13^WT (WT) or Cdc13^E252K (EK) in a magnetic beads pull-down assay. Upper, beads fractions (beads); Lower, 20% of the input material is shown as controls (20% input). (D) A total of 50 nM of TEL15 complexed with 100 nM of Cdc13^WT or Cdc13^E252K (EK) was subjected to cocomplex formation with Est1^K444E (KE) in EMSA experiments. Left, a representative gel; Right, quantification.

It has also been proposed that the Cdc13–Est1 interaction is controlled by Tel1/Mec1 phosphorylation of residues S249 and S255 in the Cdc13 RD. Simultaneous mutation of both residues to unphosphorylatable alanine leads to an est phenotype (34). However, exhaustive mass spectrometry analysis did not detect phosphorylation of either residue in purified Cdc13^FL. Also, the Cdc13^{N ter} purified from E. coli is not expected to be phosphorylated. Thus, Cdc13 and Est1 can interact in the absence of phosphorylation within the Cdc13 RD.

To determine whether phosphorylation would enhance the Cdc13–Est1 interaction we detect in vitro, we purified full-length Cdc13^S249,255D in which the two serine residues are replaced by phosphomimetic amino acids. The mutant protein was subjected to the magnetic beads pull-down experiment with Est1 (Fig. 4A, “DD”). The Cdc13^S249,255D mutant interacted with Est1 as effectively as the WT Cdc13. In the EMSA experiment where the interaction was quantitatively measured (Fig. 4B), Est1 had a K_d of 269 ± 29 nM for the Cdc13^S249,255D mutant, which was not significantly different from that for WT Cdc13. Therefore, we conclude that the in vivo defects of the Cdc13^E252K (cdc13-2), Est1^K444E (est1-60), and Cdc13^S249,255D mutants do not reflect a change in Cdc13–Est1 interaction in vitro as measured by the assays used in this paper.

Cdc13 and Est1 Abundance Is Sufficient to Support Complex Formation in Vivo.

Cdc13 and Est1 are both low-abundance proteins in vivo: In asynchronous cells, Cdc13 and Est1 are present at, respectively, ∼320 (43) and ∼70 copies per cell (14). Using a quantitative Western blot approach (14), we determined Cdc13 and Est1 concentrations (Fig. 5). We used strains expressing either Cdc13-myc9 or Est1-myc9 from their endogenous promoters and native loci. The tagged proteins are fully functional as they support WT-length telomeres (7, 12). In asynchronous cultures, our methods yielded values of 324 ± 66 copies of Cdc13 per cell and 67.6 ± 23.7 copies of Est1 per cell, in excellent agreement with previous reports (14, 43). We then determined Cdc13 and Est1 levels in α-factor–arrested cells (late G1) when telomerase is not active and nocodazole-arrested (G2) cultures, when it is (10, 11). Cdc13 and Est1 were present at 288.9 ± 23.2 and 20.3 ± 6.9 copies per cell in G1 phase and 417.8 ± 67.2 and 109.1 ± 25.8 copies per cell in G2 phase cells, respectively. The cell cycle-dependent increase in G2 was statistically significant for both proteins (P = 0.03 and 0.003 for Cdc13 and Est1, respectively). We previously concluded that the increase in Cdc13 levels detected (but not quantified) by Western blotting is consistent with the increase in cell mass and DNA content as cells progress through the cell cycle, whereas the increase in Est1 abundance represents cell cycle-dependent up-regulation (13) via proteosome-dependent degradation (44). The volume of the yeast nucleus expands quickly in G1 phase but remains largely unchanged from late G1 to S phase, at ∼2.91 fL (45). Using this value for the volume of nuclei in both conditions, the in vivo concentrations for Cdc13 were 164.8 ± 13.2 nM in G1 and 238.4 ± 38.3 nM in G2, whereas Est1 concentration was 11.6 ± 4 nM in G1 and 60.3 ± 12.6 nM in G2. These estimates suggest that there are no more than ∼8 (4.5 nM) and ∼48 (27.6 nM) Cdc13–Est1 complexes in, respectively, G1 and G2 phase cells. All of these concentrations are below the K_d for Cdc13–Est1 interaction calculated in vitro (Fig. 1). Thus, unless there are additional interactions that stabilize the Cdc13–Est1 interactions on telomeric DNA, in vivo, most of the Cdc13 and Est1 is expected to exist as free proteins, rather than in complex form.

Fig. 5.

Cell cycle-specific in vivo concentration of Cdc13 and Est1. (A) A representative gel of quantitative Western blot analysis. Total cell lysates were extracted from asynchronous, α-factor–arrested, or nocodazole-arrested cultures expressing Est1-myc9 and an untagged strain. Lysates equivalent to 2 × 10⁷ cells were loaded alongside a myc9-tagged standard protein (10, 6, 4, 2, 1, and 0.5 fmol) mixed with extract from 2 × 10⁷ cells from the untagged control for anti-myc Western blot analysis. Right, Flow cytometry profiles are shown. (B) Quantification of in vivo concentrations of Cdc13 and Est1 from asynchronous, G1-phase, and G2-phase cells. Concentrations in nanomoles are shown in parentheses.

Discussion

Here we used full-length Cdc13 and Est1 proteins purified from their native host, S. cerevisiae, to provide unique in vitro evidence that Cdc13 and Est1 interact directly to form a 1:1 complex (Fig. 1) through the genetically defined RD of Cdc13 (Fig. 2C). This interaction, which was essential for Est1 recruitment to telomeric ssDNA in vitro (Fig. 3), mimics the in vivo roles of Cdc13 and Est1 as comediators for telomerase recruitment (26). Unexpectedly, the Cdc13–Est1 interaction was unchanged by charge-swap mutations in either Cdc13 (cdc13-2) or Est1 (est1-60) or by simultaneous phosphomimetic mutations of the potential Tel1/Mec1 phosphorylation sites on Cdc13 (cdc13-S249,255D) (Fig. 4). Additionally, we show that the in vivo concentrations of Cdc13 and Est1 in G2 phase cells are sufficiently high to support complex formation in vivo (Fig. 5).

Genetic experiments strongly suggest that a Cdc13–Est1 interaction is critical to recruit telomerase to telomeres in vivo (17, 26, 27). Previously, a weak Cdc13–Est1 interaction was demonstrated by yeast two-hybrid and in vivo coimmunoprecipitation experiments (23–25). However, it is not clear from these earlier experiments whether the Cdc13–Est1 interaction is direct or bridged by another cellular component, and there was no quantitative information on the strength of this interaction. Our in vitro results establish that these two proteins interact directly, and the Est1 recruitment domain of Cdc13 resides within a small region predicted from in vivo experiments (17, 27). The apparent K_d for the Cdc13–Est1 interaction was ∼250 nM, which falls within the range of that for other transient interactions between yeast nuclear proteins, such as the replication machinery components PCNA and Polη (∼100 nM) (46), the Pho80-Pho85 kinase and its inhibitor Pho81 (∼120 nM) (47), Pho80-Pho85 and its substrate Pho4 (400–800 nM) (48), and Cdc13 OB1 and a 36-aa fragment from Pol1 (3.8 μM) (41).

The abundance of Cdc13 and Est1 in vivo was extremely low, with Cdc13 levels close to the K_d value but the level of Est1 far below it (Fig. 5). On the basis of the determined in vivo concentrations and the K_d for the Cdc13–Est1 complex, we estimate that there are no more than 8 and 48 Cdc13–Est1 complexes in G1 and G2 phase cells, respectively. G tails in G1 phase are short, 12–14 nt in length (49), and can therefore accommodate only a single Cdc13 molecule. On the basis of these considerations, most (∼89%) of the ∼300 Cdc13 molecules will not be telomere associated in G1 phase cells [78% if Cdc13 forms a dimer (41, 42) on ssDNA]. Thus, even though the number of Cdc13–Est1 complexes is low in G1 phase cells, the number of telomere-associated complexes is even lower, ≤2 complexes per cell. However, G tails in S/G2 phase are 50–100 nt in length (9, 50), and therefore each of these G tails can bind more Cdc13. Assuming that four to six Cdc13 molecules (or two to three dimers) are bound to each of the 64 telomeres in a haploid G2-phase cell, ∼75% of the Cdc13 molecules could be telomere associated at the time in the cell cycle when telomerase is active. Thus, given the number of Cdc13–Est1 complexes that are predicted to be telomere associated, up to 57% of the telomeres in a G2-phase cell can potentially be elongated by telomerase.

Although we estimate that more than half of the telomeres in G2-phase cells could be Cdc13–Est1 associated, in vivo data demonstrate that telomere extension occurs at a much lower frequency, with only ∼7% of WT-length telomeres elongated in a given cell cycle (51). Therefore, productive Cdc13–Est1 interactions must occur at a much lower efficiency in vivo. There are multiple possible explanations for this discrepancy. For example, if the interaction between Cdc13 and Est1 is sufficiently transient, the complex could dissociate from telomeres before productive telomerase assembly or action, a process that could be hastened by the Pif1 DNA helicase, which removes active telomerase from DNA ends in vivo and in vitro (52). Productive Cdc13–Est1 complex formation is likely also reduced by the presence of other proteins that compete with Est1 for Cdc13 binding, such as Stn1 (53, 54) and Pol1 (DNA Pol α) (23). These considerations can explain why in vivo Cdc13–Est1 interactions are detected by coimmunoprecipitation only when the proteins are overexpressed (23, 24). Taken together, these results suggest that the Cdc13–Est1 interaction is necessary but insufficient to support stable telomere association of Est1 (and the telomerase holoenzyme). Other interactions such as Est1–ssDNA (28, 39), Est1–Est2 (28), or Est1–TLC1 (28, 55) may prevent dissociation of Est1 and/or the telomerase holoenzyme from telomeres. In vivo data support this interpretation as in the absence or destabilization of TLC1 or Est2, Est1 telomere association is greatly reduced (13, 16, 17), even though the Cdc13–Est1 interaction should still occur.

We examined two Cdc13 mutants, Cdc13^E252K (cdc13-2) and Cdc13^S249,255D, that show reduced Est1 recruitment in vivo, defects that are generally attributed to their presumed inability to interact with Est1 (27, 34). Surprisingly, we saw no significant change in Est1 affinity for these two mutants (Fig. 4). We also did not observe a change in the strength of the WT–Cdc13 interaction for Est1^K444E (est1-60). These findings are consistent with in vivo results showing that Est1 interacts with WT and Cdc13^E252K equally well by the criteria of yeast two hybrid and coimmunoprecipitation analyses (23), and Est1^K444E associates as well as WT Est1 with telomeres (28). These data provide additional support for our view that the Cdc13–Est1 interaction is not sufficient to support stable telomerase assembly at telomeres. We propose that, although the charge-swap or phosphorylation mutants support WT levels of Cdc13–Est1 interaction, the resulting Cdc13–Est1 complex is defective. For example, the Cdc13-Est1-K444E and Cdc13-E252K-Est1 complexes might be improperly oriented so that Est1 cannot interact stably with Est2, TLC1, or other telomerase subunits. These defects could lead to Est1 dissociation from the telomere and thus loss of Est2 binding and telomerase activity.

Both Cdc13 and Est1 proteins are reported to bind telomeric ssDNA specifically (6, 28, 39, 40, 56), and our results are consistent with these previous findings (Fig. 1–3). One might question why telomerase recruitment depends on a Cdc13–Est1 interaction when Est1 itself binds telomeric ssDNA. In vivo, Cdc13 telomere association is independent of other telomerase components (16, 17). The affinity of Est1 for telomeric DNA is not nearly as high as that of Cdc13 (6, 37), and the presence of large quantities of nuclear RNA may compete with ssDNA for Est1 binding, because DNA binding and RNA binding by Est1 are mutually exclusive (28). Therefore, in the absence of an Est1-interacting partner that confers high affinity for telomeric ssDNA, Est1 is expected to be excluded from telomeric ssDNA. Indeed, when ssDNA was bound by Cdc13^DBD, a protein that does not interact with Est1, Est1 was unable to bind telomeric ssDNA (Fig. 3). Taken together, we propose that it is the Cdc13 interacting ability of Est1, not its ssDNA binding activity, that is essential for telomerase recruitment.

It is intriguing that we observed a weak but reproducible interaction between RPA and Est1 (Fig. 2), given that certain mutations in RPA result in telomere shortening in both S. cerevisiae (57) and S. pombe (58, 59). A previous report suggested that interaction between Est1 and RPA helps telomerase recruitment (60). However, short telomeres that are preferred substrates for telomerase have very low levels of RPA binding, and even this binding can be explained as being due to semiconservative DNA replication (61). At this time, it is not possible to assess whether the RPA–Est1 interaction has functional significance.

In summary, our results provide the biochemical basis of Cdc13–Est1 interaction and insights into the molecular mechanism of telomere recruitment of telomerase. Our work recapitulates the first step of telomerase recruitment in vitro. How the Cdc13–Est1 interaction is regulated in vivo and the necessary components for stable assembly of telomerase at the telomere are areas for future investigation.

Materials and Methods

Strains.

Est1, Cdc13^FL, both WT and mutants, and all Cdc13 fragments except Cdc13^{N ter} were purified from yeast strain BCY123 carrying an arc1-K86R mutation. RPA was purified from yeast BJ2168. Cdc13^{N ter} (amino acids 1–455) and Est3 were purified from E. coli Rosetta2(DE3) (Novagen). Complementation test and quantitative Western blot analyses were carried out in a YPH499 background. Genotypes of the yeast strains are listed in Table S1.

Protein Purification.

Cdc13^FL, WT and mutants, and all Cdc13 fragments except Cdc13^{N ter} were cloned into a pYES2 vector fused to a carboxyl-terminal tag consisting of a Gly₈ linker, an HRV 3C cleavage site, 5× streptag II (Novagen), and a HAT tag (Clontech). Protein overexpression was induced with 2% galactose at 30 °C for 12 h. Cdc13^{N ter} was cloned into pGEX6P-1 fused to an amino-terminal GST tag. Fresh E. coli transformants were grown at 18 °C, and protein overexpression of Cdc13^{N ter} was induced with 0.2 mM isopropylthiogalactose for 16 h.

Cdc13^FL, Cdc13^OB1-RD, and Cdc13^RD were purified over 0.1% polyethyleneimine precipitation, 50% ammonium sulfate precipitation, streptactin agarose (Novagen), and Talon Metal Affinity resin (Clontech). The affinity tags were removed by HRV 3C protease (Novagen) digestion at 4 °C. Untagged Cdc13^FL or Cdc13^OB1-RD was concentrated and buffer exchanged to TDEG/100 buffer (25 mM Tris⋅Cl, pH 7.5, 0.1 mM DTT, 0.1 mM EDTA, 10% glycerol, 100 mM NaCl) on an Amicon Ultra-4 [molecular weight cut off (MWCO) 30 kDa] concentrator that retained Cdc13 but not the protease and the tag. The affinity tag on Cdc13^RD was removed on column and the untagged Cdc13^RD was separated from the resin and concentrated, and buffer was exchanged to TDEG/100 buffer on an Amicon Ultra-4 (MWCO 10 kDa). Cdc13^DBD was purified as described except the ammonium sulfate precipitation step was omitted and a GST-tagged HRV 3C was used to cleave the tag. The protease was then removed by an Amicon Ultra-4 (MWCO 50 kDa). Untagged Cdc13^DBD in the filtrate was then concentrated and buffer exchanged to TDEG/100 buffer on an Amicon Ultra-4 (MWCO 10 kDa) concentrator. Cdc13^{N ter} was purified using glutathione Sepharose (GE Healthcare), and the tag was removed on column by GST-HRV 3C protease digestion. Cdc13^{N ter} was separated from resins and concentrated buffer exchanged to TDEG/100 buffer on an Amicon Ultra-4 (MWCO 50 kDa) concentrator. Concentrations of Cdc13 fragments were determined using an extinction coefficient at 280 nm calculated on the basis of the amino acid composition.

Est1 and Est3 were purified as described (14) except that the affinity tag of Est3 was not removed postpurification. RPA was purified as described (62).

All proteins from the final step of purification were subjected to sequential in-gel endoproteinase digestion. Peptides were eluted, desalted, and then subjected to reversed-phase nano–LC-MS and MS/MS coupled to an LTQ-Orbitrap hybrid mass spectrometer (Thermo) to confirm protein identity and integrity (Princeton Mass Spectrometry Facility).

Oligonucleotide DNA Substrates.

DNA oligonucleotides were purchased from Integrated DNA Technologies (Table S2). DNA concentrations were determined using extinction coefficients provided by the vendor.

Magnetic Bead Pulldown.

Experiments were carried out as described in ref. 14 at indicated concentrations. The gels were first stained with Sypro Ruby (Invitrogen) and analyzed using a Storm 860 system (Molecular Dynamics). The relative amounts of bead-associated Cdc13 or RPA were measured by comparing Cdc13/Est1 or RPA/Est1 ratio in the input gel with proteins of known quantity. After quantification, the gels were restained with Coomassie brilliant blue for visualization.

EMSA.

Unless otherwise indicated, reactions were performed in 20 mM Tris⋅HCl, pH 8, 100 mM NaCl, 5 mM MgCl₂, 0.1 mg/mL BSA, 1 mM DTT, 5% glycerol, and 0.005% bromophenol blue at room temperature for 20 min before loading onto a 0.8% agarose gel in 1× TBE. Gels were run at 120 V for 2 h at 4 °C, dried on a DE81 paper (Millipore), and visualized and quantified using a Storm 860 system.

Quantitative Western Blot Analysis.

Early log-phase culture (OD 0.1–0.2) expressing myc9-tagged Cdc13 (YTSF32) or Est1 (YCTT373) was arrested in G1 or G2 by 10 ng/mL α-factor or 15 μg/mL nocodazole, respectively, at 24 °C for 3 h. Cell cycle profiles were determined by flow cytometry analysis, cells counted on a Coulter Z2 cell counter (Beckman), and quantitative Western blots were performed as described (14).