Rif1 Supports the Function of the CST Complex in Yeast Telomere Capping

Telomere integrity in budding yeast depends on the CST (Cdc13-Stn1-Ten1) and shelterin-like (Rap1-Rif1-Rif2) complexes, which are thought to act independently from each other. Here we show that a specific functional interaction indeed exists among components of the two complexes. In particular, unlike RIF2 deletion, the lack of Rif1 is lethal for stn1ΔC cells and causes a dramatic reduction in viability of cdc13-1 and cdc13-5 mutants. This synthetic interaction between Rif1 and the CST complex occurs independently of rif1Δ-induced alterations in telomere length. Both cdc13-1 rif1Δ and cdc13-5 rif1Δ cells display very high amounts of telomeric single-stranded DNA and DNA damage checkpoint activation, indicating that severe defects in telomere integrity cause their loss of viability. In agreement with this hypothesis, both DNA damage checkpoint activation and lethality in cdc13 rif1Δ cells are partially counteracted by the lack of the Exo1 nuclease, which is involved in telomeric single-stranded DNA generation. The functional interaction between Rif1 and the CST complex is specific, because RIF1 deletion does not enhance checkpoint activation in case of CST-independent telomere capping deficiencies, such as those caused by the absence of Yku or telomerase. Thus, these data highlight a novel role for Rif1 in assisting the essential telomere protection function of the CST complex.


Introduction
Telomeres, the specialized nucleoprotein complexes at the ends of eukaryotic chromosomes, are essential for genome integrity. They protects chromosome ends from fusions, DNA degradation and recognition as DNA double-strand breaks (DSBs) that would otherwise lead to chromosome instability and cell death (reviewed in [1]). Telomeric DNA in the budding yeast Saccharomyces cerevisiae, as well as in nearly all other eukaryotes examined to date, comprise short TG-rich repeated sequences ending in a short single-stranded 39 overhang (G tail) that corresponds to the strand bearing the TG-rich repeats. The addition of telomeric repeats depends on the action of telomerase, a specialized reverse transcriptase that extends the TG-rich strand of chromosome ends. Recruitment/activation of this enzyme requires the Cdc13 protein that binds to the telomeric TG-rich single-stranded DNA (ssDNA) [2][3][4][5][6]. The direct interaction between Cdc13 and the Est1 regulatory subunit of telomerase is essential for telomerase recruitment, and it is disrupted by the cdc13-2 mutation that leads to gradual telomere erosion and accompanying senescence [2,4,7].
The average length of S. cerevisiae telomeric 39 overhangs is 12-14 nucleotides, although it can increase to ,50 nucleotides during the late S/G2 phase of the cell cycle [8][9][10]. While single-stranded telomeric G-tails can arise after removal of the last RNA primer during lagging-strand replication, the blunt ends of the leadingstrand telomere must be converted into 39 overhangs by resection of the 59 strand. This 59 to 39 nucleolytic degradation involves several proteins, such as the MRX complex, the nucleases Exo1 and Dna2 and the helicase Sgs1 [10,11]. Cyclin-dependent kinase activity (Cdk1 in S. cerevisiae) is also required for generation of the extended single-stranded overhangs in late S phase [12,13]. As Cdk1 activity is low in G1, telomere resection can occur only during S/G2 [8], coinciding with the time frame in which G-tails are lengthened and can serve to recruit telomerase.
Keeping the G tail in check is crucial to ensure telomere stability, and studies in budding yeast have shown that Cdc13 prevents inappropriate generation of ssDNA at telomeric ends [2,14,15]. This essential capping function depends on Cdc13 interaction with the Stn1 and Ten1 proteins to form the so-called CST (Cdc13-Stn1-Ten1) complex. This complex binds to telomeric ssDNA repeats and exhibits structural similarities with the heterotrimeric ssDNA binding complex Replication protein A (RPA) [16], suggesting that CST is a telomere-specific version of RPA. Loss of Cdc13 function through either the cdc13-1 temperature sensitive allele or the cdc13-td conditional degron allele results in telomere C-strand degradation, leading to activation of the DNA damage checkpoint [13,14,17,18]. Similarly, temperature sensitive mutations in either STN1 or TEN1 genes cause telomere degradation and checkpoint-mediated cell cycle arrest [19][20][21]. Interestingly, Stn1 interacts with Pol12 [22], a subunit of the DNA polymerase a (pola)-primase complex with putative regulatory functions, while Cdc13 interacts with the pola catalytic subunit of the same complex [7], suggesting that CST function might be tightly coupled to the priming of telomeric C strand synthesis. In any case, it is so far unknown whether the excess of telomeric ssDNA in cst mutants arises because the CST complex prevents the access of nuclease/helicase activities to telomeric ends and/or because it promotes pola-primase-dependent C strand synthesis.
In addition to the capping function, a role for the CST complex in repressing telomerase activity has been unveiled by the identification of cdc13, stn1 and ten1 alleles with increased telomere length [2,21,23,24]. The repressing effect of Cdc13 appears to operate through an interaction between this protein and the Cterminal domain of Stn1 [25,26], which has been proposed to negatively regulate telomerase by competing with Est1 for binding to Cdc13 [4,24].
A second pathway involved in maintaining the identity of S. cerevisiae telomeres relies on a complex formed by the Rap1, Rif1 and Rif2 proteins. Although only Rap1 is the only shelterin subunit conserved in budding yeast, the Rap1-Rif1-Rif2 complex functionally recapitulates the shelterin complex acting at mammalian telomeres (reviewed in [27]). Rap1 is known to recruit its interacting partners Rif1 and Rif2 to telomeric double-stranded DNA via its C-terminal domain [28][29][30]. This complex negatively regulates telomere length, as the lack of either Rif1 or Rif2 causes telomere lengthening, which is dramatically increased when both proteins are absent [30]. The finding that telomere length in rif1D rif2D double mutant is similar to that observed in RAP1 C-terminus deletion mutants [30] suggests that Rap1-dependent telomerase inhibition is predominantly mediated by the Rif proteins. However, Rif proteins have been shown to regulate telomere length even when the Rap1 C-terminus is absent [31], suggesting that they can be brought to telomeres independently of Rap1.
In addition to negatively regulate telomere length, Rap1 and Rif2 inhibit both nucleolytic processing and non homologous end joining (NHEJ) at telomeres [32][33][34]. Telomeric ssDNA generation in both rif2D and rap1DC cells requires the MRX complex [33], and the finding that MRX association at telomeres is enhanced in rif2D and rap1DC cells [33,35] suggests that Rap1 and Rif2 likely prevent MRX action by inhibiting MRX recruitment onto telomeric ends. Interestingly, the checkpoint response is not elicited after inactivation of Rap1 or Rif2, suggesting that either the accumulated telomeric ssDNA is insufficient for triggering checkpoint activation or this ssDNA is still covered by Cdc13, which can inhibit the association of the checkpoint kinase Mec1 to telomeres [36]. Notably, Rif1 is not involved in preventing telomeric fusions by NHEJ [32] and its lack causes only a slight increase in ssDNA generation at a de novo telomere [33]. These findings, together with the observation that Rif1 prevents telomerase action independently of Rif2, indicate that Rif1 and Rif2 play different functions at telomeres.
As both CST and the shelterin-like complex contribute to telomere protection, we asked whether and how these two capping complexes are functionally connected. We found that the viability of cells with defective CST complex requires Rif1, but not Rif2. In fact, RIF1 deletion increases the temperature sensitivity of cdc13-1 cells and impairs viability of cdc13-5 cells at any temperature. Furthermore, the rif1D and stn1DC alleles are synthetically lethal. By contrast, the lack of Rif2 has no effects in the presence of the same cdc13 and stn1 alleles. We also show that cdc13-1 rif1D and cdc13-5 rif1D cells accumulate telomeric ssDNA that causes hyperactivation of the DNA damage checkpoint, indicating that loss of Rif1 exacerbates telomere integrity defects in cdc13 mutants. By contrast, deletion of RIF1 does not enhance either cell lethality or checkpoint activation in yku70D or est2D telomere capping mutants. Thus, Rif1 is required for cell viability specifically when CST activity is reduced, highlighting a functional link between Rif1 and CST.

Results
Rif1, but not Rif2, is required for cell viability when Cdc13 or Stn1 activities are reduced Yeast cells harbouring the cdc13-1 temperature-sensitive allele of the gene encoding the essential telomeric protein Cdc13 are viable at permissive temperature (20-25uC), but die at restrictive temperature (26-37uC), likely due to accumulation of ssDNA at telomeres caused by the loss of Cdc13 capping functions [14]. As also the shelterin-like complex contributes to the maintenance of telomere integrity, we investigated its possible functional connections with Cdc13 by disabling either Rif1 or Rif2 in cdc13-1 cells. Deletion of RIF2 did not affect cdc13-1 cell viability in YEPD medium at any tested temperature ( Figure 1A). By contrast, cdc13-1 rif1D cells showed a maximum permissive temperature for growth of 20uC and were unable to grow at 25uC, where cdc13-1 single mutant cells could grow at almost wild type rate ( Figure 1A). The enhanced temperature-sensitivity of cdc13-1 rif1D cells was due to the lack of RIF1, because the presence of wild type RIF1 on a centromeric plasmid allowed cdc13-1 rif1D cells to grow at 25uC ( Figure 1B). The synthetic effect of the cdc13-1 rif1D combination was not uncovered during a previous genome wide search for gene deletions enhancing the temperature-sensitivity of cdc13-1 cells [37], likely because that screening was done at 20uC, a temperature at which cdc13-1 rif1D double mutants do not show severe growth defects ( Figure 1A). Our data above indicate that Rif1, but not Rif2, is required to support cell viability when Cdc13 protective function is partially compromised.
If the lack of Rif1 in cdc13-1 cells increased the temperaturesensitivity by exacerbating the telomere end protection defects of these cells, Rif1 overexpression might suppress the temperature sensitivity caused by the cdc13-1 allele. Indeed, high copy number plasmids carrying wild type RIF1, which had no effect on wild type cell viability, improved the ability of cdc13-1 cells to form colonies on synthetic selective medium at the semi-permissive temperature of 26-27uC ( Figure 1C).

Author Summary
Protection of chromosome ends is crucial for maintaining chromosome stability and genome integrity, and its failure leads to genome rearrangements that may facilitate carcinogenesis. This protection is achieved by the packaging of chromosome ends into protective structures called telomeres that prevent DNA repair/recombination activities. Telomeric DNA is bound and stabilized by two protein complexes named CST and shelterin, which are present in a wide range of multicellular organisms. Whether structural and functional connections exist between these two capping complexes is an important issue in telomere biology. Here, we investigate this topic by analyzing the consequences of disabling the two Saccharomyces cerevisiae shelterin-like components, Rif1 and Rif2, in different hypomorphic mutants defective in CST components. We demonstrate that Rif1 plays a previously unanticipated role in assisting the essential telomere protection function of the CST complex, indicating a tight coupling between CST and Rif1. As CST complexes have been recently identified also in other organisms, including humans, which all rely on shelterin for telomere protection, this functional link between CST and shelterin might be an evolutionarily conserved common feature to ensure telomere integrity.
The function of Cdc13 in telomere protection is mediated by its direct interactions with Stn1 and Ten1, leading to formation of the CST complex (reviewed in [38]). In addition to the capping function, the CST complex is implicated in repression of telomerase action [2,21,23,24]. This CST-dependent negative regulation of telomerase can be separated from CST capping function, as yeast cells either carrying the cdc13-5 allele or lacking the Stn1 C-terminus (residues 282-494) (stn1DC) display extensive telomere elongation but no or minimal growth defects [24][25][26]. We evaluated the specificity of the genetic interaction between rif1D and cdc13-1 by analysing the consequences of deleting RIF1 and RIF2 in cdc13-5 or stn1DC cells. Deletion of RIF1 turned out to reduce cell viability of cdc13-5 mutant cells at any temperatures, while deletion of RIF2 did not ( Figure 2A). Furthermore, meiotic tetrad dissection of stn1DC/STN1 rif1D/RIF1 diploid cells did not allow the recovery of viable stn1DC rif1D double mutant spores ( Figure 2B), indicating that rif1D and stn1DC were synthetic lethal. By contrast, viable rif2D stn1DC spores were found with the expected frequency after tetrad dissection of stn1DC/STN1 rif2D/ RIF2 diploid cells ( Figure 2C). The observed synthetic phenotypes suggest that both stn1DC and cdc13-5 cells have capping deficiencies and that the lack of Rif1 enhances their protection defects. Consistent with this hypothesis, cdc13-5 and stn1DC mutants were shown to accumulate telomeric ssDNA, although the amount of this ssDNA was not enough to invoke a DNA damage response [24,25]. We conclude that Rif1, but not Rif2, is required to support cell viability when a partial inactivation of CST capping function occurs.
A Cdc13 specific function that is not shared by the other subunits of the CST complex is its requirement for recruitment/ activation of telomerase at chromosome ends [2][3][4][5][6]. Cdc13mediated telomerase recruitment is disrupted by the cdc13-2 mutation, which leads to progressive telomere shortening and senescence phenotype [4]. We therefore asked whether RIF1 deletion influences viability and/or senescence progression of cdc13-2 cells. Viable cdc13-2 rif1D spores were recovered after tetrad dissection of cdc13-2/CDC13 rif1D/RIF1 diploid cells (data not shown), indicating that the lack of Rif1 does not affect the overall viability of cdc13-2 cells. When spores from the dissection plate were streaked on YEPD plates for 4 successive times, the decline in growth of cdc13-2 and cdc13-2 rif1D spores occurred with similar kinetics ( Figure 2D), indicating that RIF1 deletion did not accelerate the senescence phenotype of cdc13-2 cells specifically defective in telomerase recruitment. Taken together, these genetic interactions indicate that Rif1, but not Rif2, has a role in assisting the essential function of the CST complex in telomere protection.
The lack of Rif1 enhances the DNA damage checkpoint response in cdc13 mutant cells The synthetic effects of combining rif1D with cdc13 and stn1 mutations suggest that Rif1 might normally assist the Cdc13 and Stn1 proteins in carrying out their essential telomere protection functions. It is known that cdc13-1 cells undergo checkpointdependent metaphase arrest when incubated at the restrictive temperature [14]. Failure to turn on the checkpoint allows cdc13-1 cells to form colonies at 28uC [42,43], indicating that checkpoint activation can partially account for the loss of viability of cdc13-1 cells. We then asked whether the enhanced temperature sensitivity of cdc13-1 rif1D cells compared to cdc13-1 cells might be due to upregulation of the DNA damage checkpoint response. Deletion of the checkpoint gene RAD9, which partially suppressed the temperature sensitivity of cdc13-1 mutant cells, slightly improved the ability of cdc13-1 rif1D cells to grow at 23-25uC ( Figure 3A), indicating that the synthetic interaction between Rif1 and Cdc13 can be partially alleviated by checkpoint inactivation. Furthermore, when wild type, rif1D, cdc13-1 and cdc13-1 rif1D cell cultures were arrested in G1 with a-factor at 20uC (permissive temperature) and then released from G1 arrest at 25uC (non-permissive temperature for cdc13-1 rif1D cells), they all replicated DNA and budded with similar kinetics after release ( Figure 3B and 3C). However, most cdc13-1 rif1D cells then arrested in metaphase as large budded cells with a single nucleus, while wild type, cdc13-1 and rif1D cells divided nuclei after 75-90 minutes ( Figure 3D).
To assess whether the cell cycle arrest of cdc13-1 rif1D cells was due to DNA damage checkpoint activation, we examined the Rad53 checkpoint kinase, whose phosphorylation is necessary for checkpoint activation and can be detected as changes in Rad53 electrophoretic mobility. Rad53 was phosphorylated in cdc13-1 rif1D cells that were released from G1 arrest at 25uC, whereas no Rad53 phosphorylation was seen in any of the other similarly treated cell cultures ( Figure 3E).
RIF1 deletion caused a checkpoint-mediated G2/M cell cycle arrest also in cdc13-5 cells. In fact, exponentially growing cdc13-5 rif1D cell cultures at 25uC contained a higher percentage of large budded cells with a single nucleus than rif1D or cdc13-5 cell cultures under the same conditions ( Figure 3F). Furthermore, Rad53 phosphorylation was detected in these cdc13-5 rif1D cells, but not in the rif1D and cdc13-5 cell cultures ( Figure 3G). Thus, the lack of Rif1 results in DNA damage checkpoint activation in both cdc13-1 and cdc13-5 cells under conditions that do not activate the checkpoint when Rif1 is present.
The synthetic interaction between Rif1 and CST is independent of rif1D-induced telomere overelongation The lack of Rif1 is known to cause telomere overelongation [29]. Thus, we examined telomere length in cdc13-1 rif1D double mutant cells. The length of duplex telomeric DNA was examined after transferring at 25uC cell cultures exponentially growing at 20uC, followed by Southern blot analysis with a TG-rich probe of XhoI-digested genomic DNA prepared at different times after shift at 25uC ( Figure 4A). As expected [29], rif1D mutant cells had longer telomeres than wild type and cdc13-1 cells ( Figure 4A). Telomeres in cdc13-1 rif1D double mutant cells either at 20uC or after incubation at 25uC were longer than those of wild type and cdc13-1 cells, but undistinguishable from those of rif1D cells ( Figure 4A). Not only RIF1 deletion, but also the cdc13-5 mutation is known to cause telomere overelongation [24] ( Figure 4B). Interestingly, when telomere length was analyzed in cdc13-5 rif1D double mutant cells grown at 25uC, telomeres were longer in cdc13-5 rif1D double mutant cells than in cdc13-5 and rif1D single mutants ( Figure 4B), indicating that the cdc13-5 mutation exacerbates the telomere overelongation defect caused by the lack of Rif1.
The finding that telomeres in cdc13-1 rif1D double mutant cells at 25uC were longer than those of cdc13-1 cells, but undistinguishable from those of cdc13-1 rif1D cells grown at 20uC ( Figure 4A) suggests that the growth defects of cdc13-1 rif1D cells at 25uC are not due to rif1D-induced telomere overelongation. Telomere lengthening in rif1D mutant cells is telomerase-dependent [44] and requires the action of the checkpoint kinase Tel1 that facilitates telomerase recruitment [45,46]. To provide additional evidences that loss of viability in cdc13 rif1D mutants occurs independently of rif1D-induced alterations in telomere length, we asked whether RIF1 deletion was still deleterious in cdc13-1, cdc13-5 and stn1DC cells in a context where telomeres cannot be elongated due to the lack of Tel1 [45]. We found that TEL1 deletion did not alleviate the growth defects of cdc13-1 rif1D cells ( Figure 5A). Rather, cdc13-1 tel1D and cdc13-1 rif1D tel1D cells showed an enhanced temperature sensitivity compared to cdc13-1 and cdc13-1 rif1D cells, respectively, presumably due to the combined effects of loss of a telomere elongation mechanism and inability to protect telomeres from shortening activities. Furthermore, the growth defects of cdc13-5 rif1D double mutant cells were similar to those of cdc13-5 rif1D tel1D triple mutant cells ( Figure 5B). Finally, viable stn1DC rif1D tel1D mutant spores could not be recovered after meiotic tetrad dissection of stn1DC/STN1 rif1D/RIF1 tel1D/TEL1 diploid cells (data not shown), indicating that stn1DC and rif1D were synthetic lethal even in the absence of Tel1.
As telomere lengthening is dramatically increased when both Rif1 and Rif2 are absent [30], we also investigated whether the absence of Rif2 exacerbates cdc13-1 rif1D growth defects. As shown in Figure 5C, cdc13-1 rif1D rif2D cells formed colonies at the maximum temperature of 20uC and behaved similarly to cdc13-1 rif1D cells. We therefore conclude that the synthetic interaction between rif1D and cdc13 alleles is not due to rif1D-induced alterations in telomere length, but it is a direct consequence of Rif1 loss.
The lack of Rif1 causes generation of telomeric ssDNA in cdc13 cells It is known that cdc13-1 cells at 37uC accumulate telomeric ssDNA that triggers checkpoint-mediated cell cycle arrest [14]. Thus, we investigated whether cdc13-1 rif1D and cdc13-5 rif1D cells contained aberrant levels of single-stranded TG sequences at their telomeres that could be responsible for loss of viability in cdc13-1 rif1D and cdc13-5 rif1D cells at 25uC. The integrity of chromosome ends was analyzed by an in-gel hybridization procedure [9], probing for the presence of single-stranded TG sequences. Both cdc13-1 and rif1D single mutants either grown at 20uC ( Figure 6A, lanes 2 and 4) or incubated at 25uC for 3 hours ( Figure 6A, lanes 6  and 8) showed only a very slight increase in single-stranded TG sequences compared to wild type ( Figure 6A, lanes 1 and 5). By contrast, cdc13-1 rif1D double mutant cells contained higher  Figure 6A, lane 7). A similar telomere deprotection defect was observed also for cdc13-5 rif1D cells grown at 25uC ( Figure 6A, lane 11), which displayed an increased amount of telomeric ssDNA compared to similarly treated wild type and cdc13-5 cells ( Figure 6A, lanes 9 and 10).
Because the length of single-stranded G overhangs increases during S phase [8], the strong telomeric ssDNA signals observed in cdc13-1 rif1D cell cultures at 25uC ( Figure 6A) might be due to an enrichment of S/G2 cells. We ruled out this possibility by monitoring the levels of single-stranded TG sequences in cdc13-1 rif1D cell cultures that were arrested in G2 with nocodazole at 20uC and then transferred to 25uC in the presence of nocodazole for 3 hours ( Figure 6B). Similarly to what we observed in exponentially growing cell cultures, G2-arrested cdc13-1 rif1D cells at 20uC displayed increased amounts of ssDNA compared to each single mutant under the same conditions, and incubation at 25uC led to further increase of this ssDNA ( Figure 6B). Taken together, these findings indicate that the lack of Rif1 causes a severe defect in telomere protection when Cdc13 activity is partially compromised.
The lack of Exo1 counteracts DNA damage checkpoint activation and telomeric ssDNA accumulation in cdc13 rif1D cells If telomeric ssDNA accumulation contributes to checkpoint activation in cdc13-1 rif1D and cdc13-5 rif1D cells, then mutations reducing ssDNA generation should alleviate the arrest and relieve the lethality caused by the lack of Rif1 in cdc13-1 and cdc13-5 background. Because the Exo1 nuclease contributes to generate telomeric ssDNA in cdc13-1 cells [47], we examined the effect of deleting EXO1 in cdc13 rif1D cells. When G2-arrested cell cultures at 20uC were transferred to 25uC for 3 hours, cdc13-1 rif1D exo1D triple mutant cells contained significantly lower amounts of telomeric ssDNA than cdc13-1 rif1D cells ( Figure 6B). A similar behaviour of the triple mutant was detectable even when G2arrested cultures where kept at 20uC, although the quantity of telomeric ssDNA accumulated by cdc13-1 rif1D cells at this taken at the indicated times after release from a-factor for FACS analysis of DNA content (B), for determining the kinetics of bud emergence (C) and nuclear division (D), and for western blot analysis of Rad53 using anti-Rad53 antibodies (E). cyc, cycling cells. (F-G) Cell cultures were grown exponentially in YEPD at 25uC. (F) The frequency of cells with no, small or large buds was determined by analyzing a total of 200 cells for each strain. The percentage of large budded cells with one or two nuclei was evaluated by fluorescence microscopy. (G) Rad53 in cell cultures exponentially growing at 25uC was visualized as in panel E. Three independent cdc13-5 rif1D strains were analyzed. doi:10.1371/journal.pgen.1002024.g003 temperature was lower than at 25uC ( Figure 6B). Furthermore, EXO1 deletion partially suppressed both the temperature-sensitivity of cdc13-1 rif1D cells ( Figure 7A) and the loss of viability of cdc13-5 rif1D cells ( Figure 7B), further supporting the hypothesis that reduced viability in these strains was due to defects in telomere protection.
Exo1-mediated suppression of the cdc13 rif1D growth defects correlated with alleviation of checkpoint-mediated cell cycle arrest. In fact, when cell cultures exponentially growing at 20uC were incubated at 25uC for 3 hours, the amount of both metaphasearrested cells and Rad53 phosphorylation was reproducibly lower in cdc13-1 rif1D exo1D cells than in cdc13-1 rif1D cells ( Figure 7C and 7D). Similar results were obtained also with cdc13-5 rif1D exo1D cells growing at 25uC, which accumulated less metaphasearrested cells and phosphorylated Rad53 than similarly treated cdc13-5 rif1D cells ( Figure 7E and 7F). Thus, both cell lethality and checkpoint-mediated cell cycle arrest in cdc13 rif1D cells appear to be caused, at least partially, by Exo1-dependent telomere DNA degradation.
The lack of Rif1 does not enhance the checkpoint response to CST-independent capping deficiencies or to an irreparable DSB The lack of Rif1 might increase the lethality of cells with reduced CST activity just because it causes a telomere deprotection defect that exacerbates the inherent telomere capping defects of cdc13 or stn1 mutants. If this hypothesis were correct, RIF1 deletion should affect viability also of other non-CST mutants defective in end protection. Alternatively, Rif1-CST functional interaction might be specific, thus reflecting a functional connection between Rif1 and CST. To distinguish between these two possibilities, we analyzed the effects of deleting RIF1 in Yku70 lacking cells, which display Exo1-dependent accumulation of telomeric ssDNA, as well as checkpoint-mediated cell cycle arrest at elevated temperatures (37uC) [47,51]. Loss of Yku in est2D cells, which lack the telomerase catalytic subunit, leads to synthetic lethality, presumably due to the combined effects of telomere shortening and capping defects [48][49][50]52]. As expected [53], yku70D cells were viable at 25uC and 30uC, but they were unable   1-4) were incubated at 25uC for 3 hours (lanes 5-8). Wild type and otherwise isogenic cdc13-5 and cdc13-5 rif1D cell cultures were grown exponentially at 25uC (lanes 9-11). Genomic DNA was digested with XhoI and singlestranded telomere overhangs were visualized by in-gel hybridization (native gel) using an end-labelled C-rich oligonucleotide. The same DNA samples were separated on a 0.8% agarose gel, denatured and hybridized with the end-labeled C-rich oligonucleotide for loading and telomere length control (denatured gel).  to form colonies at 37uC ( Figure 8A). Similarly, yku70D rif1D double mutant cells grew well at 25uC and 30uC ( Figure 8A) and did not show Rad53 phosphorylation when grown at 25uC ( Figure 8B, time 0). Furthermore, similar amounts of phosphor-ylated Rad53 were detected in both yku70D and yku70D rif1D cell cultures that were kept at 37uC for 4 hours ( Figure 8B), indicating that loss of Rif1 does not enhance the telomere protection defects already present in yku70D cells. Consistent with a previous Figure 8. RIF1 deletion does not influence the checkpoint response to CST-independent telomere capping deficiencies or to an irreparable DSB. (A) Cell cultures with the indicated genotypes were grown overnight in YEPD at 25uC. Serial 10-fold dilutions were spotted onto YEPD plates and incubated at the indicated temperatures for 2-3 days. (B) Cell cultures exponentially growing in YEPD at 25uC were shifted to 37uC at time 0. Rad53 was visualized at the indicated time points as in Figure 3E. (C) Meiotic tetrads from a EST2/est2D RIF1/rif1D diploid strain were dissected on YEPD plates. After ,25 generations, spore clones from 20 tetrads were subjected to genotyping and to four successive streak-outs (1X to 4X), corresponding to ,25, ,50, ,75 and ,100 generations of growth, respectively. All tetratype tetrads behaved as the one shown in this panel. (D) After ,25 generations of growth on the dissection plates, est2D and est2D rif1D spore clones were propagated in liquid YEPD medium. At the indicated generations, protein extracts were subjected to western blot analysis with anti-Rad53 antibodies. (E-F) Checkpoint response to an irreparable DSB. (E) YEP+raf G1-arrested cell cultures of wild type JKM139 and its isogenic rif1D derivative strain were spotted on galactose-containing plates that were incubated at 28uC (time zero). At the indicated time points, 200 cells for each strain were analyzed to determine the frequency of single cells and of cells forming microcolonies of 2, 4 or more than 4 cells. (F) Galactose was added at time zero to cell cultures of the strains in panel E exponentially growing in YEP+raf. Protein extracts were subjected to western blot analysis with anti-Rad53 antibodies. doi:10.1371/journal.pgen.1002024.g008 observation [54], RIF1 deletion partially suppressed the temperature sensitivity ( Figure 8A) and the telomere length defect (data not shown) caused by the lack of Yku70, suggesting that the elongated state of the telomeres could be the reason why yku70D rif1D cells can proliferate at 37uC.
Checkpoint activation can also be induced during telomere erosion caused by insufficient telomerase activity [55,56]. Thus, we asked whether RIF1 deletion accelerated senescence progression and/or upregulated checkpoint activation in cells lacking the telomerase catalytic subunit Est2. Meiotic tetrads were dissected from a diploid strain heterozygous for the est2D and rif1D alleles, which are recessive and therefore do not affect telomere length in the diploid. After 2 days of incubation at 25uC (approximately 25 generations), spore clones from the dissection plate were both streaked for 4 successive times ( Figure 8C) and propagated in YEPD liquid medium to prepare protein extracts for Rad53 phosphorylation analysis at different time points ( Figure 8D). Similar to what was previously observed [44], RIF1 deletion did not accelerate senescence progression in est2D cells, as est2D rif1D clones showed a decline in growth similar to that of est2D clones ( Figure 8C). Furthermore, est2D and est2D rif1D cell cultures showed similar patterns of Rad53 phosphorylation with increasing number of generations ( Figure 8D). Thus, the lack of Rif1 does not enhance either DNA damage checkpoint activation or senescence progression during telomere erosion caused by the lack of telomerase.
Finally, because the telomerase machinery is known to be recruited to an unrepaired DSB [57], we ruled out the possibility of a general role for Rif1 in inhibiting checkpoint activation by examining activation/deactivation of the checkpoint induced by an unrepaired DSB. To this end, we used JKM139 derivative strains, where a single DSB can be generated at the MAT locus by expressing the site-specific HO endonuclease gene from a galactose-inducible promoter [58]. This DSB cannot be repaired by homologous recombination, because the homologous donor sequences HML or HMR are deleted. As shown in Figure 8E, when G1-arrested cell cultures were spotted on galactose containing plates, both wild type and rif1D JKM139 derivative cells overrode the checkpoint-mediated cell cycle arrest within 24-32 hours, producing microcolonies with 4 or more cells. Moreover, when galactose was added to exponentially growing cell cultures of the same strains, Rad53 phosphorylation became detectable as electrophoretic mobility shift in both wild type and rif1D cell cultures about 2 hours after HO induction, and it decreased in both cell cultures after 12-15 hours ( Figure 8F), when most cells resumed cell cycle progression (data not shown). Thus, Rif1 does not affect the checkpoint response to an irreparable DSB. Altogether these data indicate that Rif1 supports specifically CST functions in telomere protection.

Discussion
Both shelterin and CST complexes are present in a wide range of unicellular and multicellular organisms, where they protect the integrity of chromosomes ends (reviewed in [38]). Thus, the understanding of their structural and functional connections is an important issue in telomere regulation. We have approached this topic by analysing the consequences of disabling the shelterin-like S. cerevisiae proteins Rif1 or Rif2 in different hypomorphic mutants defective in CST components. We provide evidence that Rif1, but not Rif2, is essential for cell viability when the CST complex is partially compromised. In fact, RIF1 deletion exacerbates the temperature sensitivity of cdc13-1 mutant cells that are primarily defective in Cdc13 telomere capping functions. Furthermore, cells carrying the cdc13-5 or the stn1DC mutation, neither of which causes per se DNA damage checkpoint activation and growth defects [24,26], grow very poorly or are unable to form colonies, respectively, when combined with the rif1D allele. By contrast, RIF1 deletion does not affect either viability or senescence progression of cdc13-2 cells, which are specifically defective in telomerase recruitment. This Cdc13 function is not shared by the other CST subunits, suggesting that Rif1 is specifically required to support the essential capping functions of the CST complex.
Cell lethality caused by the absence of Rif1 in both cdc13-1 and cdc13-5 cells appears to be due to severe telomere integrity defects. In fact, telomeres in both cdc13-1 rif1D and cdc13-5 rif1D double mutant cells display an excess of ssDNA that leads to DNA damage checkpoint activation. Deleting the nuclease EXO1 gene partially restores viability of cdc13-1 rif1D and cdc13-5 rif1D cells and reduces the level of telomeric ssDNA in cdc13-1 rif1D cells, indicating that cell lethality in cdc13 rif1D cells is partially due to Exo1-dependent telomere DNA degradation and subsequent activation of the DNA damage checkpoint.
Although Rif1 and Rif2 interact both with the C-terminus of Rap1 and with each other [29,30], our finding that only Rif1 is required for cell viability when Cdc13 or Stn1 capping activities are reduced indicates that Rif1 has a unique role in supporting CST capping function that is not shared by Rif2. Earlier studies are consistent with the idea that Rif1 and Rif2 regulate telomere metabolism by different mechanisms [30,31,35]. Furthermore, while the content of Rif2 is lower at shortened than at wild type telomeres, the level of Rif1 is similar at both, suggesting that these two proteins are distributed differently along a telomere [59]. Finally, inhibition of telomeric fusions requires Rif2, but not Rif1 [32].
Noteworthy, although RIF1 deletion is known to cause telomere overelongation [29], the synthetic interaction between Rif1 and CST occurs independently of rif1D-induced alterations in telomere length. In fact, the lack of Tel1, which counteracts rif1D-induced telomere overelongation [45], does not alleviate the growth defects of cdc13 rif1D cells. Furthermore, deletion of RIF2, which enhances telomere elongation induced by the lack of Rif1 [30], does not exacerbate the synthetic phenotypes of cdc13 rif1D double mutant cells. Thus, loss of viability in cdc13 rif1D cells is not due to telomere overelongation caused by RIF1 deletion, but it is a direct consequence of Rif1 loss.
By analyzing the effects of combining RIF1 deletion with mutations that cause telomere deprotection without affecting CST functions, we found that the functional interaction between Rif1 and the CST complex is highly specific. In fact, the lack of Rif1 does not enhance the DNA damage checkpoint response in telomerase lacking cells, which are known to experience gradual telomere erosion leading to activation of the DNA damage checkpoint [55,56]. Furthermore, RIF1 deletion does not upregulate DNA damage checkpoint activation in yku70D cells, which display Exo1-dependent accumulation of ssDNA and checkpointmediated cell cycle arrest at 37uC [47][48][49][50][51]. This is consistent with previous observations that comparable signals for G strand overhangs can be detected on telomeres derived from yku70D and yku70D rif1D cells [54], indicating that RIF1 deletion does not exacerbate the end protection defect due to the absence of Yku. By contrast, the lack of Rif1 partially suppresses both temperaturesensitivity and telomere shortening in yku70D cells ( Figure 8A) [54], possibly because the restored telomere length helps to compensate for yku70D capping defects. Notably, although RIF1 deletion leads to telomere overelongation in cdc13-1 and cdc13-5 mutants, this elongated telomere state does not help to increase viability in cdc13-1 rif1D and cdc13-5 rif1D cells.
The simplest interpretation of the specific genetic interactions we found between Rif1 and CST is that a functional connection exists between Rif1 and the CST complex, such that Rif1 plays a previously unanticipated role in assisting the CST complex in carrying out its essential telomere protection function. Indeed, this functional interaction is unexpected in light of Rif1 and CST localization along a telomere. In fact, while CST is present at the very ends of chromosomes, Rif1 is thought to be distributed centromere proximal on the duplex telomeric DNA [59]. However, as yeast telomeres have been proposed to fold back onto the subtelomeric regions to form a ,3-kb region of core heterochromatin [60,61], this higher-order structure could place Rif1 and CST in close proximity, thus explaining their functional interaction.
The function of Rif1 in sustaining CST activity cannot be simply attributable to the Rif1-mediated suppression of ssDNA formation at telomeres, as rif1D cells show only a very slight increase in ssDNA at both native ( Figure 6) and HO-induced telomeres [33] compared to wild type. Furthermore, although deletion of Rif2 leads to increased amounts of telomeric ssDNA [33], cdc13-1 rif2D, cdc13-5 rif2D and stn1DC rif2D double mutants are viable and do not display growth defects. Finally, other mutants defective in telomere capping or telomere elongation (yku70D and est2D) are perfectly viable in the absence of Rif1.
One possibility is that Rif1 physically interacts, directly or indirectly, with the CST complex. Indeed, human Stn1 was found to copurify with the shelterin subunit TPP1 [62], suggesting the existence of CST-shelterin complexes in mammals. Unfortunately, we were so far unable to coimmunoprecipitate Rif1 with Cdc13 or Stn1, and further analyses will be required to determine whether Rif1 and the CST complex undergo stable or transient association during the cell cycle.
Indeed, not only 59-39 resection, but also incomplete synthesis of Okazaki fragments is expected to increase the size of the G tail during telomere replication. The yeast CST complex genetically and physically interacts with the pola-primase complex [7,22,25] and the human CST-like complex increases pola-primase processivity [63,64]. Furthermore, the lack of CST function in G1 and throughout most of S phase does not lead to an increase of telomeric ssDNA [13], suggesting that the essential function of CST is restricted to telomere replication in late S phase. Altogether, these observations suggest that CST may control overhang length not only by blocking the access of nucleases, but also by activating polaprimase-dependent C-strand synthesis that can compensate G tail lengthening activities. Based on the finding that Rif1 regulates telomerase action and functionally interacts with the pola-primase complex (Figure 2), it is tempting to propose that Rif1 favours CST ability to replenish the exposed ssDNA at telomeres through activation/recruitment of pola-primase, thus coupling telomerasedependent elongation to the conventional DNA replication process.
The recent discoveries that human TPP1 interacts physically with Stn1 [62] and that CST-like complexes exist also in S. pombe, plants and mammals [65][66][67][68] raise the question of whether functional connections between the two capping complexes exist also in other organisms. As telomere protection is critical for preserving genetic stability and counteracting cancer development, to address this question will be an important future challenge.

Southern blot analysis of telomeres and in-gel hybridization
Genomic DNA was digested with XhoI. The resulting DNA fragments were separated by electrophoresis on 0.8% agarose gel and transferred to a GeneScreen nylon membrane (New England Nuclear, Boston), followed by hybridization with a 32 P-labelled poly(GT) probe and exposure to X-ray sensitive films. Standard hybridization conditions were used. Visualization of singlestranded overhangs at native telomeres was done by in-gel hybridization [9], using a single-stranded 22-mer CA oligonuleotide probe. The same DNA samples were separated on a 0.8% agarose gel, denatured and hybridized with an end-labeled C-rich oligonucleotide for loading control.

Other techniques
For western blot analysis, protein extracts were prepared by TCA precipitation. Rad53 was detected using anti-Rad53 polyclonal antibodies kindly provided by J. Diffley (Clare Hall, London, UK). Secondary antibodies were purchased from Amersham and proteins were visualized by an enhanced chemiluminescence system according to the manufacturer. Flow cytometric DNA analysis was determined on a Becton-Dickinson FACScan on cells stained with propidium iodide.