The TPR-containing domain within Est1 homologs exhibits species-specific roles in telomerase interaction and telomere length homeostasis

Background The first telomerase-associated protein (Est1) was isolated in yeast due to its essential role in telomere maintenance. The human counterparts EST1A, EST1B, and EST1C perform diverse functions in nonsense-mediated mRNA decay (NMD), telomere length homeostasis, and telomere transcription. Although Est1 and EST1A/B interact with the catalytic subunit of yeast and human telomerase (Est2 and TERT, respectively), the molecular determinants of these interactions have not been elaborated fully. Results To investigate the functional conservation of the EST1 protein family, we performed protein-protein interaction mapping and structure-function analysis. The domain in hEST1A most conserved between species, containing a TPR (tricotetrapeptide repeat), was sufficient for interaction of hEST1A with multiple fragments of hTERT including the N-terminus. Two mutations within the hTERT N-terminus that perturb in vivo function (NAAIRS92, NAAIRS122) did not affect this protein interaction. ScEst1 hybrids containing the TPR of hEST1A, hEST1B, or hEST1C were expressed in yeast strains lacking EST1, yet they failed to complement senescence. Point mutations within and outside the cognate ScEst1 TPR, chosen to disrupt a putative protein interaction surface, resulted in telomere lengthening or shortening without affecting recruitment to telomeres. Conclusions These results identify a domain encompassing the TPR of hEST1A as an hTERT interaction module. The TPR of S. cerevisiae Est1 is required for telomerase-mediated telomere length maintenance in a manner that appears separable from telomere recruitment. Discrete residues in or adjacent to the TPR of Est1 also regulate telomere length homeostasis.


Background
In S. cerevisiae, telomeres are usually maintained by telomerase or RAD52-dependent recombination. The 'ever shorter telomere' genes EST1, EST2 (the telomerase reverse transcriptase, TERT), EST3, and TLC1 (the telomerase RNA) are essential for telomerase function because loss of any one gene results in progressive telomere shortening and senescence [1,2]. Mutation of CDC13 -an essential gene -elicits similar consequences [2,3]. Rare "survivors" can bypass this senescence by maintaining telomeres through recombination. The homologous recombination factor Rad52 is important for the generation of telomerase-independent Type I and Type II survivors in which telomeres are maintained by amplification of Y' elements or telomeric repeats, respectively; more rarely, survival can occur without Rad52 [4][5][6][7][8][9][10].
Cdc13 and Est1 are critical for the recruitment of the telomerase core complex (Est2-Tlc1) to telomeres in S phase [reviewed in [11][12][13][14]]. Cdc13 binds to singlestranded telomeric DNA [15,16] and associates with telomeres throughout the cell cycle, with a peak in association during S phase [17,18]. Est1 also binds to singlestranded telomeric DNA [19] and associates with telomeres in S phase [17,18,20]. Cdc13 and Est1 interact physically [21] and genetically, as evidenced by the unlinked complementation of cdc13-2 and est1-60 alleles, each of which affects telomere maintenance and viability [15,17,22,23]. Several findings suggest that Est1 recruits Est2 to the telomere in S phase by acting as an intermediary between Cdc13 and TLC1. For example, the telomere shortening and senescence that occur in the absence of EST1 are rescued by expression of a Cdc13-Est2 fusion protein [22]. Est1 binds to TLC1 [24][25][26][27][28], and the telomeric localization of Est2 in S phase is reduced when the region of TLC1 responsible for the Est1 interaction is deleted, or when EST1 is deleted [29].
In other organisms, Est1 homologs similarly associate with telomerase and regulate telomere length and stability, although in some instances their precise contributions to telomere function are still being uncovered. S. pombe Est1 associates with active telomerase in cell extracts, and est1-cells exhibit telomere shortening, senescence, and defects in chromosome end protection [30]. In C. albicans, telomere length in est1Δ cells fluctuates over serial passages, suggesting that Est1 may repress homologous recombination at telomeres [31]. In humans, three Est1 homologs, hEST1A/SMG6, hEST1B/ SMG5 and hEST1C/SMG7 (hereafter referred to hEST1A, hEST1B and hEST1C) interact with chromatin and bind preferentially at telomeres [32][33][34]. Human EST1A and EST1B associate with active telomerase in cell lysates and in vitro [32,33]. Like ScEst1, hEST1A binds single-stranded telomeric DNA [33]. Transient over-expression (or depletion) of hEST1A causes telomere uncapping/end-to-end fusion and apoptosis, and stable over-expression of hEST1A in telomerase-positive cell lines elicits telomere shortening that can be mitigated by co-expression of hTERT [32][33][34].
The complex functions of EST1 homologs are mirrored by their diversity in size and structure across species (Additional file 1, Figure S1). The region of highest homology among EST1 proteins includes tetratricopeptide repeat (TPR) consensus sequences [30][31][32][33] (Additional file 1, Figure S1). Typically, TPRs mediate protein-protein interactions [45]. The structure of hEST1C reveals a bona fide 14-3-3-like TPR comprised of alpha helices; several contiguous upstream alpha helices fold into a TPR-like structure despite the fact they lack a TPR repeat consensus sequence [46]. Upstream of the TPR, hEST1A contains an N-terminal DNA binding activity [ [33,42]; Jen Cruickshank and Lea Harrington, unpublished], and the N-terminus also interacts with hTR and other RNAs through an hTRinteraction domain (TRID) [47]. A region downstream of the hEST1B TPR resembles the DNA binding domain of ScEst1 [48], although DNA binding by hEST1B has not been described. Both hEST1A and hEST1B contain a C-terminal 'PilT N-terminal' (PIN) domain [32,33,36,49]. The PIN domain of hEST1A possesses a single-stranded RNA endonuclease activity and degrades PTC-containing mRNA [49,50]. The nuclease activity of the hEST1B PIN domain is greatly reduced likely due to the absence of critical residues in the active site [49]. This extensive cross-species complexity prompted us to conduct structure-function analysis of human and S. cerevisiae EST1 proteins using biochemical and genetic methods.

Mapping the interactions between hEST1A and hTERT
Although it was known that hEST1A interacts with hTERT in vitro independently of hTR [33], and that TPRs typically mediate protein-protein interactions [45], it was unknown whether the TPR was sufficient for the interaction between hEST1A and hTERT. Fragments of hEST1A spanning all or part of the TPR (502-824, 334-824, 114-749, 114-780) proved unsuitable for in vitro interaction analysis as they precipitated non-specifically ( Figure 1C, and data not shown). An hEST1A fragment spanning a.a. 114-631 did not exhibit non-specific precipitation but failed to exhibit a significant interaction with hTERT fragments ( Figure 1A, B). Increasing the C-terminal boundary to fully encompass the minimal TPR domain spanning a.a 695-761 (to a.a. 824) resulted in a reproducible enrichment of hEST1A(114-824) onto anti-FLAG agarose in the presence of FLAG-hTERT fragments spanning amino acids 1-200, 1-350, 201-560, 601-1132, but not fragments spanning 201-350 or 350-560, a control protein (Akt), or mock translation reactions containing no input cDNA ( Figure 1A-C). The hTERT-hEST1A interactions were not perturbed by micrococcal nuclease, indicating that they did not require nucleic acids (Figure 1A, B). Taken together, these data identify at least one interaction interface within hTERT(1-350) that includes residues beyond a.a. 200 that are important (but insufficient in the context of a.a. 201-350) for an interaction with hEST1A(114-824).
To further refine the hTERT protein interaction interface within the hEST1A TPR, we introduced various mutations within a.a. 551-785 of hEST1A(114-824) that we predicted to lie on the protein surface (see below),   and tested these fragments for an hTR-independent interaction with hTERT(1-350) or hTERT(601-1132) (Additional file 1, Figure S2). None of the 22 introduced mutations, including the 17 residues mutated between hEST1A a.a. 631 and 824, perturbed the interaction of hEST1A(114-824) with either hTERT fragment (data not shown) [51]. In addition, removal of a sequence in hEST1A (a.a. 708-724) corresponding to a putative disordered loop in hEST1C [46] did not disrupt the hEST1A-hTERT interaction ( Figure 1C). Thus, despite the existence of a putative hTERT interaction interface in hEST1A (631-824) that includes a TPR domain (695-761), we were unable to identify residues within the TPR essential for this interaction.

Non-conservation of EST1 TPR function between species
Recent data establish that hEST1C, of which the closest S. cerevisiae homolog is Ebs1, plays a similar role as Ebs1 in NMD [42]. Thus, we wished to determine if hEST1A or hEST1B might display functional similarity with ScEst1. Despite various strategies, we were unable to observe expression of full-length hEST1A, B, C in S. cerevisiae, which precluded analysis of whether human Est1 homologs might, alone or in combination, complement the senescence phenotype of strains lacking EST1.
An alternate strategy was to test whether a region of Est1 containing the TPR could be "swapped" between species. Using an in vivo gap-repair cloning strategy in S. cerevisiae [59], the TPR of ScEst1 was replaced with the TPR of hEST1A, hEST1B or hEST1C or GFP(S65T) (the latter representing a similarly-sized domain as the TPR) (Additional file 1, Figure S3). The boundaries of the TPR were estimated according to the structurebased sequence alignment of EST1 homologs [32,33,46]. Specifically, the first nine alpha helices of hEST1C, or the corresponding region of hEST1A or hEST1B, were integrated into the corresponding region of ScEst1.
We first verified that est1Δ and est1Δ rad52Δ haploid strains generated from a diploid strain (S288C EST1/ est1Δ::NAT RAD52/rad52Δ::KAN) underwent senescence as expected [1,8]. Next, we tested the ability of low-copy (pRS316) or high-copy (pRS426) hybrid Est1 expression plasmids (introduced into the EST1/est1Δ RAD52/ rad52Δ strain prior to haploid selection) to rescue senescence in two separately isolated est1Δ rad52Δ haploid spores ( Figure 2). Although the resultant TPR hybrid proteins were expressed ( Figure 2B), neither the GFP(S65T)-Est1 hybrid nor any of the human TPR-Est1 hybrids proved sufficient to rescue senescence or maintain telomere length when spores were passaged every two days (Figure 2A, 3A, Additional file 1, Figure S4). Extension of the time between colony propagation from two days to four days, which promotes the rare survival (approximately 4-8%) of populations lacking telomerase and RAD52 [6], also failed to permit complementation with the Est1 TPR hybrid proteins (Additional file 1, Figure S4). The TPR hybrids also failed to exert a dominant interfering effect on cell viability or telomere length maintenance in an EST1 rad52Δ strain ( Figure  2A, C, Figure 3B, Additional file 1, Figure S4). Therefore, replacement of the cognate TPR of ScEst1 with the TPR of hEST1A, B, or C neither complemented nor interfered with wild-type ScEst1.

Dissection of Est1 TPR function by mutagenesis
Since the TPRs of hEST1A, B, or C were unable to function in place of the TPR of ScEst1, we mutated residues within and just outside the ScEst1 TPR domain and assessed effects on function in vivo. Selected residues that we predicted to be exposed to the concave surface of the putative protein-protein interaction interface (according to the alignment of the ScEst1 sequence to the hEST1C TPR structure) were changed to alanine [46] (Additional file 1, Figure S2); all mutants complemented the viability of est1Δ rad52Δ strains over multiple generations and did not interfere with viability in rad52Δ strains ( Figure 4A, B, Additional file 1, Figure  S5). The average telomere length of cell populations expressing wild-type FLAG-tagged Est1 were compared with those expressing FLAG-tagged Est1 point mutants including Est1(F511S) -a mutant which confers viability in the absence of endogenous Est1 but maintains shorter-than-wild-type telomere lengths [19,60] ( Figure  4C, lanes [13][14][15][28][29][30]. Like cells expressing Est1 (F511S), cells expressing Est1(K84A/W87A/Q89A)-FLAG exhibited slightly shorter telomeres ( Figure 4C, lanes 19-21, 31-33). Unexpectedly, est1Δ rad52Δ cells containing certain Est1-FLAG mutants -namely E92A/ Q96A/W97A, R193A/N197A, S200A/F203A/Y204A, F243A/Q244A/K247A, N277A/N278A, or D281A/ T285A -exhibited longer telomeres than cells containing wild-type Est1-FLAG ( Figure 4C). Telomere lengths appeared to slightly increase upon successive passages with certain mutants such as N277A/N278A, E92A/ Q96A/W97A and S200A/F203A/Y204A ( Figure 4C, lanes 7-9, 22-24 or 35-36, and 25-27, respectively). The elongation phenotype was not due to overexpression of Est1, since telomere lengths in cells overexpressing wildtype Est1 (tagged or untagged) did not differ from strains lacking high-copy Est1 ( Figure 5B). The differences in telomere lengths relative to wild-type Est1 and to each other were reproducible with different strain isolates (data not shown). Although we did not test the Est1 hybrids do not rescue est1Δ rad52Δ strains or interfere with growth of rad52Δ strains. (A) Heterozygous diploid est1Δ rad52Δ yeast were transformed with pRS316(URA3) or pRS426(URA3) plasmids alone ('vector') or plasmids expressing yeast/human EST1 hybrids in which the TPR of ScEst1 was replaced with the TPR of hEST1A/B/C, or GFP(S65T). Haploid spores were isolated and passaged every two days on plates containing synthetic dropout media lacking uracil. Summary plates of growth from the 1 st passage to 6 th passage (as shown, counterclockwise) were prepared by re-streaking colonies from each passage onto one sector of a single plate (refer to Methods). Refer to Additional file 1, Figure S4 for summary of spore growth. (B, C) Yeast/human Est1 hybrid proteins were expressed. Western blot of cell lysates using anti-FLAG and anti-mouse IgG-HRP antibodies. Arrowheads indicate wild-type or hybrid Est1 proteins. A non-specific, cross-reacting band is indicated with an asterisk (*). Molecular mass (kDa) is indicated at the left.
protein expression levels of all point mutants, larger perturbations in Est1 (e.g., the TPR 'swaps') did not alter Est1 levels relative to wild-type Est1-FLAG ( Figure 2B). Thus, the different effects of TPR mutations on telomere length suggest that distinct residues in the vicinity or within the TPR domain function in telomere length maintenance, while other residues may function to limit telomere elongation.

Recruitment of Est1 point mutants to telomeric chromatin
To test whether mutation of the Est1 TPR affected recruitment to the telomere, we performed chromatin immunoprecipitations (ChIP) experiments. Yeast lysates were subjected to crosslinking and sonication to obtain chromatin fragments approximately 500 bp in length, followed by immunoprecipitation onto anti-FLAG M2 resin. PCR analysis was performed using primers designed to amplify a single-copy genomic locus (ARS607) or the subtelomeric region of the right arm of chromosome X (tel XR). After normalization to both input genomic DNA and the ARS607 amplification signal, wild-type Est1-FLAG ( Figure 5A, lanes 3, 6) exhibited a significant fold-enrichment at telomeric chromatin relative to same strain lacking ScEst1-FLAG ( Figure 5A, lanes 1, 4). We next tested three FLAGtagged Est1 mutants (K84A/W87A/Q89A, N277A/ N278A, F511S) for enrichment at telomeric chromatin in the est1Δ rad52Δ background. All three mutants exhibited telomeric enrichment similar to Est1-FLAG ( Figure 5A). Notably, mutants associated with telomere elongation (N277A/N278A) or shortening (F511S, K84A/W87A/Q89A) exhibited equivalent enrichment. The similar levels of enrichment also suggest that fluctuations in telomere length are not simply due to changes in the level of Est1 protein recruited to telomeric heterohromatin ( Figure 5B). These data suggest  that the effects on telomere length equilibrium conferred by the Est1 TPR mutants may involve steps downstream of recruitment to the telomere.

Species-specific differences in the TERT-EST1 interaction
Our observation that N-and C-terminal regions of hTERT can interact with hEST1A (114-824) independently of hTR extends previous findings [33,47]. Redon et al. showed that an hTR-dependent interaction occurs between hTERT(147-311) and hEST1A(212-381), and that hTERT(147-311) may interact with hEST1A fragments beyond a.a. 212-502 independently of hTR [47] (Additional file 1, Figure S6). Collectively, our results suggest that hTERT forms protein-protein contacts with the TPR of hEST1A. As yet, the specific amino acids responsible for these interactions have not been identified. That the hEST1A TPR exhibits multiple contacts with discrete hTERT fragments is not without precedent; the hTERT amino-terminus (hTEN) also exhibits specific interactions with itself and the hTERT C-terminus, suggesting that an hEST1A interaction could be bridged by intramolecular hTERT interactions in vivo [56,[61][62][63]. In addition, hTEN exhibits specific interactions with telomeric DNA and TPP1 that stimulate repeat addition processivity [64][65][66][67][68]. Taken together, these data underscore that multiple protein-protein and protein-DNA interaction interfaces serve to regulate hTERT function in vivo.
Like the interaction between hEST1A and hTERT, the interaction between Est1 and Est2 may involve both protein-protein and protein-RNA contacts. Est1 and Est2 interact with separate regions of Tlc1 [26][27][28]69], and Est1 does not co-purify with Est2 in a tlc1Δ strain, suggesting that Tlc1 is required to assemble the telomerase complex [60]. However, recent findings suggest that recombinant ScEst1 stimulates telomerase activity even when the Tlc1-Est1 interaction is compromised, and that Tlc1 stimulates but is not required for an Est1-Est2 interaction in RRL [70]. These results suggest that Est1 and telomerase may also interact through protein contacts.   [15,16]. All samples were resolved on the same gel, and irrelevant or blank lanes between lanes 3, 4 and 9, 10 were removed.
The TPR of ScEst1 may be involved in binding Tlc1. Mutations at residues that we predicted to reside on the convex surface of the domain (118/122/123 and 222/ 223/226) elicit telomere shortening and perturb association with Tlc1 [60]. None of the mutations that we introduced into the putative concave surface of the TPR of hEST1A reduced the interaction with hTERT. Mutagenesis of the convex surface of hEST1A TPR may yet identify an hTERT interaction interface.
A role for the ScEst1 TPR in curtailing telomere elongation downstream of telomere binding We found that mutations within the ScEst1 TPR that affected telomere length equilibrium did not reduce the association of Est1 with telomeric heterochromatin in asynchronous cultures. Since Est1 recruitment to telomeres peaks in S phase [17,18,20], further experiments in synchronized cultures may reveal further nuances of the influence of these mutations on the temporal regulation of telomere recruitment. The fact that the mutants showed a comparable enrichment to that of wild-type Est1 suggests that these ScEst1 TPR residues are nonetheless dispensable for an interaction with Cdc13, and is consistent with the notion that the perturbation of telomere length reflects a defect downstream of telomere recruitment.
While certain mutations predicted to disrupt the concave surface of the Est1 TPR elicited telomere shortening, other similarly predicted mutations elicited telomere elongation. The concave surface of the Est1 TPR may negatively regulate telomere length through an interaction with Ies3 -a component of the INO80 chromatin remodeling complex [74]. As demonstrated by yeast two-hybrid and co-immunoprecipitation experiments, Est1 and Ies3 interact, and deletion of Ies3 results in telomere lengthening [74]. Mutation of three amino acids within the TPR of Est1 (G199A/P236A/ N278A) abolishes the Ies3 interaction [74]. This triple mutation impairs Tlc1 binding and elicits telomere shortening, thus precluding a direct assessment of the Ies3-Est1 interaction in telomere length regulation [74]. In contrast, we observed that alanine replacement of N277/N278 in Est1 led to telomere elongation. Thus, it is possible that the N277A/N278A mutation might disrupt interaction with Ies3 without perturbing Tlc1 association, which would support a direct role for the Ies3-Est1 interaction in the negative regulation of telomere length by the INO80 complex [74]. Our results indicate that the TPR-containing region within ScEst1 positively and negatively regulates telomere length, and that the concave surface of the TPR domain appears important for Est1 function.

Functional specialization of key cellular processes within and between species
We observed that the TPRs of human EST1 proteins could not substitute for the ScEst1 TPR. Functional differences in other telomere-associated proteins have been noted across species. Two related 5'-3' helicases in S. cerevisiae, Pif1 and Rrm3, perform non-overlapping functions in inhibiting telomerase recruitment and promoting DNA fork progression, respectively [75], and both contribute to mitochondrial DNA stability in distinct ways [76,77]. In S. pombe, the single Pif1/Rrm3 ortholog Pfh1 is essential; some, but not all, of its diverse roles in mitochondrial, nuclear and telomere DNA replication are complemented by ScRrm3 [78][79][80][81]. In mammals, although the Pif1/Rrm3 ortholog interacts with telomerase, it is dispensable for viability, telomere length maintenance, and chromosome healing [82][83][84]. Similarly, Rif1 plays a critical regulatory role in telomere homeostasis in yeasts [reviewed in [14,[85][86][87][88][89], while in mammals Rif1 modulates genome stability in response to DNA damage and appears dispensable for telomere length regulation [90][91][92][93].
In S. cerevisiae, the NMD-related function of EST1 homologs may have been delegated primarily to Ebs1, which shares the most sequence similarity (particularly in the TPR) with hEST1C [42]. Strains lacking EBS1 possess short telomeres [26] and, unlike est1Δ strains, exhibit a defect in NMD [37,42]. Over-expression of hEST1C or Ebs1 in S. cerevisiae disrupts NMD [42]. In A. thaliana, EST1C/SMG7 is an essential gene with a role in NMD and meiotic cell cycle progression, but possesses no apparent role in telomere integrity or length regulation [94]. It will be interesting to determine whether mutations exist within the hEST1A TPR that selectively affect NMD or an interaction with hTERT. If a telomerase-specific interaction interface within hEST1A exists, it may represent an attractive target for telomerase inhibition in cancer since its short-term perturbation might spare other essential functions of EST1 related to telomere end protection or NMD.

Conclusions
A fragment of hEST1A of approximately 700 a.a. and encompassing the tetratricopeptide repeat (TPR, a.a. 695-761) was sufficient for nucleic acid-independent interactions with the N-terminus of hTERT (a. a. 1-350). The TPR of ScEst1 could not be complemented genetically by TPRs of the human Est1 homologs hEST1A, hEST1B, or hEST1C, suggesting that the former has evolved species-specific interactions that cannot be substituted by human EST1 TPRs. Mutagenesis within and just outside the ScEst1 TPR identified residues important for telomere length homeostasis but not recruitment to telomeric heterochromatin. Collectively, these data identify species-specific roles of the Est1 TPR domain in the recognition of hTERT and regulation of telomere length.

Methods
Co-immunoprecipitation of hTERT and hEST1A fragments and micrococcal nuclease treatment pCR3-FLAG-hTERT-FLAG and pcDNA3.1-HA2-hEST1A constructs were expressed in rabbit reticulocyte lysates (RRL) according to the manufacturer's instructions (Promega Corp.) in the absence of hTR. The inclusion of hTR RNA in hTERT translation reactions did not affect interactions with hEST1A (data not shown). Mock reactions contained no cDNA. Reactions were incubated separately for 90 min at 30°C. Fifteen to 20 μL of each reaction was mixed in a total volume of 500 μL with cold CHAPS buffer (0.5% w/v CHAPS, 10 mM Tris-Cl pH 7.5, 10% v/v glycerol, 100 mM NaCl, 1 mM MgCl 2 1 mM DTT, Roche EDTA-free Complete protease inhibitor cocktail) and agitated gently for 1 hr at 4°C. FLAG-tagged complexes were immunoprecipitated with pre-equilibrated anti-FLAG M2 agarose resin (Sigma). Anti-c-Myc agarose (BD Biosciences) was employed as a negative control resin. For micrococcal nuclease digestions, immunoprecipitates were washed twice with cold CHAPS buffer and treated with 40 U/μL micrococcal nuclease (New England Biolabs) at 30°C for 45 min in buffer containing 50 mM Tris-Cl pH 7.9, 5 mM CaCl 2 and 1 mg/mL bovine serum albumin, or mock-treated in buffer without micrococcal nuclease. Reactions were terminated by the addition of excess EGTA, and immunoprecipitates were washed twice in cold CHAPS buffer. For experiments not involving micrococcal nuclease, immunoprecipitates were washed three times in cold CHAPS buffer.
In Figure 1B, immunoprecipitation data from four independent replicates of the experiment shown in Figure 1A were quantified as follows. Firstly, the raw intensity of each hEST1A fragment (114-824 or 114-631) immunoprecipitated onto anti-FLAG resin was divided by the average intensity (n = 4) of the fragment in the mock immunoprecipitation lane (no added TERT). Secondly, the raw intensity of each hTERT fragment precipitated onto anti-FLAG resin was expressed as a ratio relative to the average intensity value of the same hTERT fragment (n = 4). The normalized hEST1A value was divided by the normalized hTERT fragment value in each experiment, and the mean and standard deviation of these ratios is illustrated in Figure 1B. Statistical significance was determined by ANOVA analysis of the normalized ratios, assuming non-paired samples, and applying the Tukey post-hoc test for pairwise comparisons where confidence levels exceeded 95%.

Generation of yeast/human EST1 TPR hybrids
The boundaries of the TPR were selected according to the structure-based alignment of EST1 sequences [32,33,46] (Additional file 1, Figure S3A). Plasmids encoding hybrid EST1 proteins in which the TPR of ScEst1 was replaced with the TPR of hEST1A, hEST1B, or hEST1C were generated by the in vivo gap-repair cloning method [59]. In the first step, pRS316(URA3)-Est1-FLAG was digested with NruI and PflMI which cleaved the TPR coding sequence at unique restriction sites in the plasmid (Additional file 1, Figure S3B). In parallel, the sequence encoding the TPR of hEST1 was amplified by PCR using hybrid primers. The 5' ends of the forward primers contained 45 nt of sequence identical to sense strand immediately upstream of ScEst1 codon 14 (the TPR sequence starts at codon 14). The 3' ends of the primers contained 30 nt of sequence identical to the sense strand of the TPR coding sequence of hEST1A (starting at codon 551), hEST1B (codon 25), or hEST1C (codon 2). GFP(S65T) (amino acids 1-237) was also engineered in place of the TPR of ScEst1 as a control for disruption of the domain. The forward primers were as follows: Est1p13/hEST1A551, 5'-(PO 4

Manipulation and propagation of yeast on solid media
Diploid yeast cultures were transformed according to a high-efficiency transformation protocol [98]. Diploids were sporulated on solid media, asci were digested with zymolyase, and haploid spores were dissected under a microscope using a microneedle (Singer, UK). Haploid spores of the desired genotype were streaked onto plates containing synthetic dropout media lacking uracil (SD-URA) to select for maintenance of plasmids expressing the URA3 gene. After two or four days of growth at 30°C (P1, passage 1 or approximately 20 generations), a single colony was re-streaked onto a second plate and incubated as described above (P2). Meanwhile, the P1 plate was stored at 4°C. This scheme was continued until P4 or P5 plates were obtained. To demonstrate growth from P1-6, one colony from each "passage" plate stored at 4°C (including the starting haploid spore from the tetrad dissection plate) was re-streaked onto a sector of a "summary plate" which was then incubated at 30°C for a final two or four days. Plates were imaged on a Bio-Rad Molecular Imager Gel Doc XR System under white light. Images were captured below the point of signal saturation.

Yeast protein extraction and western blotting
Protein extracts were prepared as described [98]. Proteins were boiled in SDS-PAGE buffer, resolved through 5% w/v 29:1 acrylamide:bisacrylamide gels, transferred to PVDF membranes, and subjected to western blotting using mouse anti-FLAG M2 antibody (Sigma). Sheep anti-mouse IgG-HRP (GE Healthcare) was employed as secondary antibody. Blots were developed with ECL Plus reagents, and fluorescent signals were captured using a Typhoon Trio variable mode imager (GE Healthcare).

Telomere length analysis by Southern blot
A single colony growing on solid media was inoculated into 5 mL SD-URA and grown overnight at 26°C. Cells were pelleted and genomic DNA was extracted and purified as described [99]. Genomic DNA (5 μg) was digested with XhoI overnight at 37°C. Restriction fragments were resolved through 0.7% w/v agarose gels at 45 V (2 V/cm). DNA was denatured in-gel by incubation with 0.5 M NaOH and 1.5 mM NaCl, and neutralized in buffer containing 1.5 M NaCl and 0.5 mM Tris-Cl pH 7.5. DNA was transferred to Hybond-N+ membranes in 20X SSC. Following transfer, membranes were rinsed in 2X SSC. Telomeric DNA was hybridized to a 32 P 5'-end-labeled (CACACCCA) 2 CC probe in Church buffer (0.5 M NaPO 4 pH 7.2, 1% w/v BSA, 7% w/v SDS, 1 mM EDTA), then washed with 1X SSC/0.1% w/v SDS. Membranes were exposed to a phosphorimager screen and scanned using a Typhoon Trio variable mode imager.

Mutagenesis of hEST1A and ScEst1 TPRs
The structure-based sequence alignment of the TPRlike/TPR of hEST1C/SMG7 and EST1 homologs [32,33,46] was used as a guide to select residues in hEST1A(551-785) and ScEst1(14-289) that may be surface-exposed. The design of PCR primers for site-directed mutagenesis was aided by The Primer Generator [100] or SiteFind [101] and primers were obtained from IDT or Operon. The incorporation of the intended mutations into the respective cDNAs (and the absence of unwanted mutations) was confirmed by DNA sequencing.

Chromatin immunoprecipitation
Yeast cells were grown to logarithmic phase in liquid culture to an optical density (O.D. 600 nm) of 1.0. Fifty mL of each culture was adjusted to 1% v/v formaldehyde (Sigma) and incubated at room temperature for 45 min. The crosslinking reaction was terminated by addition of glycine to a final concentration of 135 mM. Samples were washed three times in ice cold PBS at 4°C, and resuspended in 500 μL FA buffer (50 mM Hepes pH 7.6, 140 mM NaCl, 1 mM EDTA, 1% v/v Triton X-100, 0.1% v/v Na-deoxycholate, Roche EDTA-free Complete protease inhibitor cocktail) per 50 O.D. 600 cells. Samples were sonicated for four cycles of 10 s followed by 30 s on ice, using a 3.2 mm microtip (VWR). Sonication resulted in fragmentation of chromatin to an average size of 500 bp (range 300-800 bp) as determined by agarose gel electrophoresis. Two hundred μL sonicated lysate was incubated overnight at 4°C with anti-FLAG-M2 agarose resin (Sigma). After extensive washing with FA buffer, complexes were eluted from beads using 300 μg/mL FLAG peptide (Sigma). Cross-linked DNA was released by the addition of reverse cross-linking buffer (10 mM Tris pH 8, 4 mM EDTA, 100 mM NaCl) and incubation at 65°C for 16 hr. Samples were then incubated for 3 hr at 56°C after the addition of Proteinase K (0.3 mg/mL) and glycogen (0.5 mg/mL). DNA was extracted and PCR amplification was performed on approximately 10 ng template DNA using primers specific to a unique subtelomeric sequence of chromosome XR (XR F: 5'-TACCCTGTCCCATTTCATTATACC-3', XR R: 5'-TACAAGTGCAAGTGAGTATGGCAT-3') or to an internal non-telomeric region at ARS607 (ARS607 F: 5'-GTGGTGATATAAACACTACATTCGC-3', ARS6 07 R: 5'-GCTTTCTAGTACCTACTGTGCCG-3'). PCR products were separated on a 6% w/v polyacrylamide gel in 1X TBE. The relative intensity of the subtelomerespecific PCR product was calculated using ImageQuant software (GE Healthcare). A region of interest containing each PCR product was defined, and the local average background signal was subtracted. The resultant subtelomere-specific values were then divided by a value for the corresponding internal loading control (ARS607), and expressed as a percentage of the signal generated from input DNA. In Figure 5, the ANOVA analysis was performed on sample values normalized to the values for the vector control strain within each experiment to enable between-experiment comparisons. By varying the concentration of DNA (in separate experiments), PCR reactions for ChIP samples and input DNA were determined to be in the linear range of amplification. Statistical analysis was carried out using GraphPad Prism 5.0, employing one-way ANOVA and Tukey's post-hoc test.

Additional material
Additional file 1: Supplementary Figures. Figure S1. Schematic representation of human and S. cerevisiae EST1 homologs. Figure S2: Comparative sequence analysis of EST1 at positions selected for mutation. Figure S3: Schematic representation of yeast/human Est1 hybrid proteins. Figure S4: Schematic representation of colony growth in Figure 2. Figure S5: Schematic representation of colony growth in Figure  4. Figure S6: Summary of hTERT/hEST1A co-IP experiments.