Report

Est1p As a Cell Cycle-Regulated Activator of Telomere-Bound Telomerase

See allHide authors and affiliations

Science  09 Aug 2002:
Vol. 297, Issue 5583, pp. 1023-1026
DOI: 10.1126/science.1074968

Abstract

In Saccharomyces cerevisiae, the telomerase components Est2p, TLC1 RNA, Est1p, and Est3p are thought to form a complex that acts late during chromosome replication (S phase) upon recruitment by Cdc13p, a telomeric DNA binding protein. Consistent with this model, we show that Est1p, Est2p, and Cdc13p are telomere-associated at this time. However, Est2p, but not Est1p, also binds telomeres before late S phase. The cdc13-2 allele has been proposed to be defective in recruitment, yet Est1p and Est2p telomere association persists in cdc13-2 cells. These findings suggest a model in which Est1p binds telomeres late in S phase and interacts with Cdc13p to convert inactive, telomere-bound Est2p to an active form.

In S. cerevisiae, telomeric DNA consists of irregular C1-3A/TG1-3 double stranded repeats as well as the G tail, a >30 nucleotide single-stranded overhang of TG1-3 DNA that is detectable only in late S phase (1, 2). Telomere length is maintained by telomerase, a specialized reverse transcriptase with an integral RNA template. In yeast, telomerase action requires at least five genes, deficiencies in any or all of which lead to the same est (ever shortening telomeres) phenotype, a progressive telomere shortening leading to eventual cell death (3). EST2 and TLC1 encode the reverse transcriptase and RNA subunit, respectively. Telomerase action in vivo also requires Est1p, Est3p, and Cdc13p, a protein that binds single-stranded TG1-3 DNA. A current model for telomerase regulation is that Cdc13p binds G tails in late S phase to recruit a telomerase holoenzyme consisting of Est1p, Est2p, Est3p, andTLC1 RNA (4–6). Consistent with this model, in vivo assays have shown telomerase action is cell cycle limited, occurring in late S phase or G2-M phase (the period between S and mitosis) but not in G1 (the prereplicative growth phase of the cell cycle) (7,8). Because Cdc13p and Est1p interact in biochemical and yeast two-hybrid assays, recruitment is thought to occur via direct binding between these two proteins (5). The requirement for Est1p in telomere maintenance can be bypassed by artificially fusing Cdc13p to Est2p (4). Also, the estphenotype of the cdc13-2 allele, which disrupts an electrostatic bond with Est1p, can be restored by an allele-specific reciprocal mutation (est1-60) (6). However, contrary to the expectations of a recruitment model, thecdc13-2 protein and Est1p interact normally by both biochemical and two-hybrid criteria, suggesting that the functional interaction between Cdc13p and Est1p that is lost in acdc13-2 strain occurs at a step other than recruitment (5).

We used chromatin immunoprecipitation (ChIP) to assess the temporal pattern and genetic dependence of Est1p, Est2p, and Cdc13p telomere association in vivo (9). Sheared chromatin was prepared from formaldehyde cross-linked cells and subjected to immunoprecipitation of Est1p, Est2p, Cdc13p, and, as a positive control, Rap1p, the major structural component of telomeric chromatin (10,11). To allow immunoprecipitation, the chromosomal loci ofEST1, EST2, and CDC13 were tagged with Myc epitopes (Fig. 1A) (12,13). Polyclonal antiserum was used to immunoprecipitate Rap1p. We used quantitative multiplex polymerase chain reaction (PCR) to monitor coimmunoprecipitation of three chromosomal loci with each protein: ARO, a control, nontelomeric fragment; ADH, a subtelomeric fragment ∼5 kilobase pairs (kbp) from telomere VII-L; and TEL, a unique fragment ∼20 base pairs (bp) away from telomere VII-L (Fig. 1B). Enrichment of ADH and TEL over the background amounts in a control Myc antibody precipitate from an untagged strain reflects subtelomeric and telomeric association in vivo, respectively (13). To control for spurious in vitro binding to telomeric DNA in the cell lysate, we also performed immunoprecipitations from mock–cross-linked cells.

Figure 1

Est1p, Est2p, and Cdc13p are telomere-associated in vivo. ChIP was performed in either wild-type (not tagged) or strains containing Myc-tagged Est1p, Est2p, or Cdc13p (13). Soluble chromatin from formaldehyde (w/crosslink) or mock treated (w/o crosslink) cells was immunoprecipitated with antibody to Myc, nonspecific serum, or polyclonal Rap1p antiserum. (A) Western blot probed with antibody to Myc of whole-cell extracts from the no tag and Myc epitope–tagged strains. Identical cell equivalents were loaded in each lane. (B) Schematic of TEL, ADH, and ARO chromosomal loci monitored by ChIP; solid black arrow represents telomere VII-L. Open arrows represent URA3,ADH4, and ARO1 open reading frames and their directions of transcription. (C) Multiplex PCR results of a representative ChIP experiment. Twofold serial dilutions of the input DNA delineate the linear range of PCR (lanes 1 to 5); the remaining results are from precipitated DNA. Immunoprecipitation was performed on the strains indicated using antibodies to Myc (lanes 6 to 9, 12 to 15, and 18 to 20), nonspecific serum (lanes 10 and 16), or antiserum to Rap1p (lanes 11 and 17). Parentheses refer to either empty vector or plasmids expressing Myc-tagged Cdc13p or a derivative lacking the DNA binding domain (Cdc13-ΔDBD). These plasmids were introduced into a wild-type background, so competition with endogenous untagged Cdc13p lowered TEL enrichment. A fraction of immunoprecipitated material was subjected to a Western blot using antibody to Myc to allow comparison of protein yields. (D) Data from (C) are expressed as x-fold enrichment of ADH (open bar) and TEL (solid bar) over background as described in (13). Samples showing substantial telomere association have fold enrichment written (numeral) above the bars. Open bars, ADH; solid bars, TEL. Error bars represent the standard error from multiple independent experiments. Results without error bars were observed in ≥two experiments.

In cycling cells, TEL was enriched ∼60-fold over background in immunoprecipitates of Rap1p antiserum but not a nonspecific serum (Fig. 1C, lanes 12, 16, and 17, and Fig. 1D). Consistent with previous results that showed Rap1p associates with subtelomeric regions, we also observed enrichment of the ADH sequence (14). Because enrichment of TEL was, for the most part, crosslinker-dependent, these data confirm previous findings that Rap1p binds telomeres in vivo. Similarly, specific, crosslinker-dependent TEL enrichment for Cdc13p that was dependent on its DNA binding domain (∼30-fold; lanes 9, 15, and 18 to 20 of Fig. 1, C and D) confirms previous findings that Cdc13p binds telomeres in vivo (12,15). Lastly, we observed specific, crosslinker-dependent TEL enrichment for Est1p and Est2p (∼3.5-fold and ∼16-fold; lanes 7, 8, 13, and 14 of Fig. 1, C and D). Therefore, each of the telomerase pathway components, Cdc13p, Est1p, and Est2p, associates specifically with telomeres in vivo.

We examined Cdc13p telomere binding through the cell cycle (9). The tagged Cdc13p strain was synchronized in G1 phase with α factor and released into the cell cycle, and samples were assayed by ChIP. Control untagged cells were synchronized simultaneously to determine background at each time point. Fluorescence-activated cell sorter (FACS) analysis of samples showed that progress through the cell cycle was indistinguishable for Myc-tagged and untagged cells; it also showed that S phase had begun at the 30-min time point, was almost complete at 45 min, and was finished at 60 min (fig. S1). The experiment was ended at 105 min, before mitotic cytokinesis. Western blot analysis of whole-cell extracts revealed that the level of Cdc13p increased ∼twofold over the course of the experiment (Fig. 2A), consistent with the doubling of mass as cells progressed through the cell cycle and formed buds. Cdc13p was telomere-associated in G1phase, and this association persisted through the beginning of S phase (Fig. 2B, lanes 5 to 10; additional G1 phase data for Cdc13p, Est2p, and Est1p are in fig. S2). A large increase in association was observed in late S phase, peaking at 60 min (∼70-fold; lanes 11 to 14 of Fig. 2B). Given that G tails form in late S phase and multiple copies of Cdc13p can bind the same DNA molecule, this increase in association is most consistent with Cdc13p multimerization on G tails (16). Later in the cell cycle, association decreased to a lower but still significant level (lanes 15 to 20 of Fig. 2B; this level was consistent with that seen in nocodazole-arrested cells, fig. S2).

Figure 2

Cell cycle regulation of Cdc13p, Est2p, and Est1p telomere association. Untagged and Myc-tagged strains were arrested in late G1 with α factor (0 min), and a portion of each was removed for ChIP using antibodies to Myc. α factor was then removed, allowing synchronous progression through the cell cycle. Samples were taken every 15 min. Early S phase occurred at 30 min; late S, at 45 and 60 min (fig. S2). (A) Using antibody to Myc, Western blot of Myc-tagged proteins in whole-cell extracts as well as immunoprecipitated protein over time. Identical cell equivalents are loaded. To allow quantitation of protein levels, twofold serial dilutions of the 105-min whole-cell extract are shown. (B) Representative cell cycle ChIP time courses. Multiplex PCR results from the no tag and tagged samples from each time point are shown side by side. The Myc tag–dependent fold enrichment of ADH (solid circle) and TEL (solid square) over background was determined independently for each time point (13). In each case, results were observed in ≥two experiments.

Cell cycle analysis for Est2p showed no changes in protein levels other than the gradual twofold increase caused by bud growth (Fig. 2A). Despite observations that telomerase does not act at this time, strong Est2p telomere association was observed in G1 phase cells (Fig. 2B, lanes 5 to 8; fig. S2). This high level of association was followed by a reduction in signal in early S phase (30 min), which was in turn followed by another strong peak of association in late S phase (45 min; lanes 9 to 12 of Fig. 2B). Lower levels of Est2p association were observed later in the cell cycle and also in nocodazole arrested cells (Fig. 2B, lanes 15 to 20; fig. S2). This pattern was reproducible in five of five independent experiments and was also observed for telomere VI-R (fig. S3) (17). Variations in association over time were not due to changes in Est2p immunoprecipitation efficiency (Fig. 2A). A second synchrony technique, in which cells were arrested in telophase using the temperature sensitive cdc15-2 allele and were released into the cell cycle, confirmed this pattern. After low Est2p telomere association in arrested M phase cells, two peaks of association were again observed, one in G1 phase and the other in late S (17).

Cell cycle analysis of Est1p revealed relatively low levels of protein in G1 phase cells, followed by a marked ∼two- to threefold increase once S phase was underway (30 min; Fig. 2A). After this point, protein levels continued to increase gradually, concomitant with the general increase in cell mass. The increase in Est1p levels in early S phase was in good agreement with previous work that showed transcription of EST1 is induced late in G1(18). Thus, Est1p abundance is cell cycle regulated. We observed no Est1p telomere binding in G1 and early S phase (Fig. 2B, lanes 5 to 10; fig. S2). However, a peak of Est1p telomere binding began in late S phase at 45 min and peaked at 60 min (Fig. 2B, lanes 11 to 14). Est1p binding at 45 min was concurrent with the late S phase peaks of Cdc13p and Est2p association. Thus, all three proteins, Est1p, Est2p, and Cdc13p, were telomere-associated in late S phase, a time when telomerase action has been observed in vivo.

The telomerase-null phenotype of the cdc13-2 allele has been interpreted as eliminating telomerase recruitment due to disrupted interaction between Cdc13p and Est1p (6, 19). This model predicts that neither Est1p nor Est2p will be telomere-associated in a cdc13-2 background. However, telomere association of both Est1p and Est2p was normal in cyclingcdc13-2 cells (Fig. 3B, lanes 6, 8, 17, and 19; fig. S4). As a control, we also monitored binding in a tlc1Δ background. The steady state level of Est2p in cycling tlc1Δ cells was reduced ∼twofold, but was not reduced in cdc13-2 cells (Fig. 3A). This effect was specific; a TLC1 deletion affected neither Cdc13p nor Est1p levels (Fig. 3A). Thus, TLC1 RNA is required for full stability of Est2p in vivo. In addition, TLC1 was required for Est2p but not Cdc13p telomere association (Fig. 3B, lanes 5-7 and 9-10). Est1p telomere association was also substantially decreased in the tlc1Δ strain (Fig. 3B, lanes 16 to 18; fig. S4).

Figure 3

Est2p, Cdc13p, and Est1p telomere association in tlc1Δ and cdc13-2 backgrounds. Cycling Myc-tagged Est2, Cdc13p, or Est1p cells with wild-type,tlc1Δ, or cdc13-2 backgrounds were subjected to ChIP. (A) Using antibody to Myc, Western blots of whole-cell extracts from cycling Myc-tagged Est2p, Est2p cdc13-2, Est2ptlc1Δ, Cdc13p, Cdc13p tlc1Δ, Est1p, Est1p tlc1Δ, and Est1p cdc13-2 strains. Identical (1×) cell equivalents were loaded except for wild-type Est2p, where 1× and 0.5× cell equivalents are shown. (B) Multiplex PCR results (for quantitation of these data, see fig. S4). Relative yields of precipitated proteins are shown. (C) Cell cycle ChIP time courses on Myc-tagged Est2p and Est1p cdc13-2cells. Experiments were performed as in Fig. 2. FACS profiles and protein levels at each time point were identical to wild-type (17). Representative fold TEL (solid square) and ADH enrichments (solid circle) above background at each time point are shown. In each case, results were observed in ≥two experiments.

We examined Est2p association throughout the cell cycle in synchronizedcdc13-2 cells. Est2p telomere association was similar to wild-type at all time points tested except in late S phase (45 min), when the peak of association observed in wild-type cells was missing in the mutant (Fig. 3C, fig. S5). The missing late S phase peak was not due to an absence of telomere-bound Cdc13p, as we observed no difference in telomere binding between cdc13-2 protein and wild type Cdc13p (17). For Est1p, cell cycle ChIP showed that, if anything, telomere binding was slightly enhanced at most time points in cdc13-2 cells (Fig. 3C, fig. S5). Thecdc13-2 allele is deficient in interaction with Stn1p (20). It is possible that the increase in Est1p telomere association in cdc13-2 cells reflects increased telomere accessibility due to a lack of steric hindrance from Stn1p. Regardless, from these data we conclude that Est1p is fully able to associate with telomeres in a cdc13-2 background.

Telomerase action is not observed in G1 phase cells, yet we observed strong TLC1-dependent telomere association of Est2p, the catalytic subunit, at that time point (Figs. 2 and 3, fig. S2). Access of the Est2p-TLC1 RNA complex to chromosome ends is, therefore, not sufficient for telomerase action in vivo. The Est2p signal was decreased, but not eliminated, in early S phase, reflecting either dissociation of Est2p from a subset of telomeres or decreased cross-linking efficiency, possibly through a change in conformation or in neighboring components of telomeric chromatin. We observed a second peak of Est2p telomere association in late S phase (Fig. 2B). The fact that this peak did not occur in the telomerase-defectivecdc13-2 background suggests that it reflects an essential process for telomerase action in vivo (Fig. 3C, fig. S5).Our findings suggest that two key events are concurrent with the late S phase Est2p peak: multimerization of Cdc13p on the G tail and Est1p binding. We propose that Est1p is a cell cycle regulated (Fig. 2A) activator of telomerase that binds to an inactive, telomere-bound Est2p-TLC1 RNA complex in late S phase and then interacts with one or more Cdc13p molecules arrayed on the G tail. This interaction changes the state of bound Est2p, either through binding of additional Est2p (e.g., dimerization) (21) or a conformational change that enhances cross-linking; this is manifested as the late S phase peak of Est2p telomere association and results in activation of telomerase for synthesis. The cdc13-2 mutation disrupts a functional interaction with Est1p (6). Because Est1p telomere association is robust in cdc13-2cells (Fig., 3, B and C), this functional interaction must be separable from and downstream of Est1p telomere binding. We propose the telomerase deficiency of the cdc13-2 allele reflects failure to interact properly with telomere bound Est1p to promote activation, resulting in persistence of inactive Est2p at the telomere and concurrent loss of the late S phase peak. We detected decreased but still considerable Est2p telomere association after S phase (Fig. 2B; fig. S2), suggesting that at least a subset of Est2p remains bound to telomeres for the remainder of the cell cycle, perhaps as a protective cap (21). Because the G1 peak of association followed low association in M phase (17), Est2p appears to load onto telomeres again at the beginning of the next cell cycle.

Supporting Online Material

www.sciencemag.org/cgi/content/full/297/5583/1023/DC1

Materials and Methods

Figs. S1 to S5

Table S1

  • * Present address: Department of Microbiology, College of Medicine, National Taiwan University, Taiwan.

  • To whom correspondence should be addressed. E-mail: vzakian{at}molbio.princeton.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article