A proto-telomere is elongated by telomerase in a shelterin-dependent manner in quiescent fission yeast cells

Strains and growth conditions

The fission yeast strains used in this study are prototrophic and listed in Supplementary Table S1. Telomerase was deleted by substituting the ter1 gene with kanamycin cassette by one-step homologous insertion using FwTer1-5’UTR and RevTer1-3’UTR primers from genomic DNA. For G0 experiments, strains were grown in Edinburgh minimal medium with sodium glutamate substituted for ammonium chloride (PMG) and 100 μg/ml hygromycin B (InvitroGen) at 32°C (or at 25°C for stn1-226 thermosensitive mutant) to a density of 6 × 10⁶ cells/ml, washed three times in minimal medium deprived from nitrogen (MM-N), and resuspended in MM-N at a density of 2 × 10⁶ cells/ml at 32°C, reaching after two rounds of cell division 8 × 10⁶ cells/ml. This density of cell bodies is stable for several weeks and was used as a source of G0 cells.

Telomere length analysis by Southern blotting

Genomic DNA was prepared from 2 × 10⁸ cells and digested with ScaI-HF restriction enzyme (New England Biolabs). The digested DNA was resolved on 1.5% agarose gel and blotted onto a Hybond-XL membrane (GE Healthcare), as previously described (39). After transfer, the membrane was cross-linked with UV and hybridized with ura4 and abo1 probes made with primers listed in Supplementary Table S2 (40). ³²P labeling of DNA probes was performed by random priming using Klenow fragment lacking exonuclease activity (New England Biolabs), in presence of [α-³²P]dCTP (Perkin Elmer) and hybridizations were performed in Church buffer at 60°C for ura4 and abo1 probes (40). Radioactive signals were detected using a Biorad molecular imager FX. The telomere length was determined using the BioRad software.

Telomere PCR and sequencing

Telomere PCR was performed as previously described (40). Briefly, 200 ng of genomic DNA was suspended in a total of 5 μl of deionized water and heated at 95°C for 5 min. Tailing reaction was performed in a final volume of 10 μl with final concentrations of 1 mM dCTP, 15 units of Terminal Deoxynucleotide Transferase (Promega) and 1X Terminal Transferase Buffer (100 mM cacodylate buffer, 1 mM CoCl₂, 0.1 mM DTT (pH 6.8)). Reaction mix was incubated at 37°C for 30 min, then at 70°C for 10 min to stop the reaction. The poly-dC-tailed DNA was then PCR amplified using AccuPrime™ GC-Rich DNA Polymerase (Invitrogen) with a primer specific for the ura4 region and a primer complementary to the poly-dC-tail listed in Supplementary Table S2. PCR program is: 3′ at 94°C, 45 cycles of 94°C 30’’; 55°C 30’’;72°C 30’’ followed by a final extension step of 5′ at 72°C. The PCR products were either purified by gel extraction (QIAquick Gel Extraction Kit) or using spin columns (QIAquick PCR purification Kit) and cloned into the pCRBLunt plasmid and sequenced using M13F primer.

ChIP

2 × 10⁹, 4 × 10⁸ or 7 × 10⁸ cells were crosslinked with 1% of Formaldehyde (Sigma Aldrich) at room temperature for 15 min for Est1-V5, Ccq1-flag Ku-myc ChIP and Taz1-GFP, respectively. Crosslinking was quenched with 125 mM Glycine for 5 min at room temperature and cells were washed twice with ice-cold TBS. Cell pellets were then resuspended in Lysis Buffer Low Sodium (50 mM HEPES–KOH pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-Deoxycholate, 2 mM of PMSF, protease inhibitors) and mechanically lysed with glass beads using 7 cycles of 20 s at speed 6.5. in fast-prep machine. Whole cell extracts were sonicated for 10 min (30 s ON, 30 s OFF) using the Diagenode Bioruptor. Sonicated lysates were cleared by centrifugation and standardized by nanodrop. Solubilized chromatin was prepared at equal DNA content before adding 2 μl of anti-V5 (R96025, Invitrogen), 2 μl of anti-myc (9E10 sc-40, Santa Cruz), 2 μl of anti-flag (M2, Sigma Aldrich) or 2 μl of anti-GFP (A11122 Invitrogen) to solubilized chromatin and incubated overnight at 4°C. 25 μl of Protein G Dynabeads (Thermofisher) were added for 3H at 4°C. Beads were then washed twice with ice-cold Lysis Buffer Low Sodium, twice with ice-cold Lysis Buffer High Sodium (50 mM HEPES–KOH pH 7.5, 500 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-Deoxycholate), twice with ice-cold sodium deoxycolate buffer (10 mM Tris–HCl pH 8, 250mM LiCl, 0.5% NP-40, 0.5% Na-Deoxycholate, 1mM EDTA) and once with ice-cold TE buffer (10 mM Tris–HCl pH 8, 1 mM EDTA). Beads were then eluted in TES (TE with 1% of SDS) and incubated for 15 min at 65°C. Samples were incubated 12 h at 65°C and then treated with Proteinase K during 4 h at 37°C. DNA was purified using innuPREP PCRpure purification column (Analytikjena) and samples were analyzed by qPCR.

qPCR and statistical analysis for ChIP

Input samples were diluted to 1/40. 2 μl of template DNA was added to 5 μl of TB Green Premix Ex Taq II (Takara), and primers were added to a final concentration of 0.4 μM. Each sample was run in triplicate on the same 96-well PCR plate in a C1000 Touch Thermal Cycler (CFX96 Biorad). PCR program was as followed: 95°C for 30 s followed by 55 cycles at 95°C for 10 s, 55 or 56°C (for 2R-48 bp or control 100 kb, respectively) for 10 s, 72°C for 10 s and a final extension step at 72°C for 5 min. Primers used for qPCR are listed in Supplementary Table S2. Corrected % of IP was calculated as followed: Corrected % IP = % IP 2R-48 bp – % IP 100 kb where with E the efficiency of qPCR,  and  the mean of triplicate for IP and Input (IT) respectively, DF the dilution factor. Statistical comparisons were performed using a Mann–Whitney test when variance is significantly different between the two analyzed data sets or unpaired two-tail t-test if not. * P-value < 0.05, ** P-value < 0.01, *** P-value < 0.005.

Western blot

Proteins were extracted with trichloracetic acid (TCA) from indicated strains, as previously described in Coulon et al. (41). Protein samples are loaded on 6% acrylamide gel and monitored using anti-flag (M2, mouse monoclonal antibody). Ponceau-S staining of the membrane was used as a loading control.

Telomerase elongates an eroded proto-telomere in G0

To investigate how eroded telomeres are processed in post-mitotic cells, we took advantage of the conditional telomere-repeat capped end system developed in the Runge laboratory (Figure 1A and B) (40). This artificial end, named proto-telomere, mimics in many aspects a single short telomere in cycling cells (42). Briefly, a single I-SceI cut site was integrated ∼47 kb away from the right telomere of chromosome II (C2-R) flanked by selectable markers (ura4⁺ and hph⁺). Upon induction of the endonuclease I-SceI with tetracycline (ahTet), a single I-SceI cut site is generated leading to the formation of the proto-telomere bearing either 48 bp of telomeric repeats or no repeats as a control (2R-48 bp and 2R-0 bp, respectively), marked with ura4⁺. Importantly, loss of the 47 kb of non-unique subtelomeric sequences in C2-R (containing an hph⁺ cassette) is dispensable for cell viability (40). In vegetative cells, the 2R-48bp is rapidly elongated by telomerase and reaches a final length of 250 bp after approximately 8 population doublings (PD), while the 2R-0 bp is degraded (40). To determine the fate of the proto-telomere in quiescent cells, we introduced this system in prototrophic fission yeast strains and induced I-SceI by adding ahTet once cells are arrested in G0 after nitrogen starvation. We collected cells before induction of I-SceI at day 0 (D0) and 1, 3, 5 and 7 days after induction (D1 to D7) (Figure 1C). For each experiment, we checked that the cell population was completely arrested in G0 by microscopic observation and FACS analysis (as exemplified by Supplementary Figure S1D). Genomic DNA was extracted for each time point, digested with ScaI, subjected to Southern blot analysis, and the 2R proto-telomere was revealed with an ura4 probe (Figure 1D). The uncut 1.9 kb ura4⁺::2R-48 bp::hph⁺ band was visible prior to induction of I-SceI and was replaced by the cut 0.4 kb ura4⁺::2R-48 bp band after induction. At D1, the elongation of the 2R-48 bp was already visible and it continued until D7. We reproducibly observed an elongation of approximately 80 bp (±5 bp) of the telomeric seed. Notably, the cut and uncut bands remained detectable during the time in quiescence (see below, Figure 2). To confirm that elongation was the result of telomerase action, we induced the 2R-48 bp in ter1Δ strain. No elongation was detected in the ter1Δ strain but a band just below the cut fragment appeared with time in G0 (Figure 1E). Importantly, we did not observe telomere elongation with the 2R-0 bp (Supplementary Figure S1A). Chromatin Immunoprecipitation (ChIP) of telomerase subunit Est1 showed significant enrichment at the 2R-48 bp after I-SceI induction (D1) (Figure 1F and Supplementary Figure S1B). Of note, ChIP experiments were performed in lig4Δ cells to avoid ligation of the cut product (see next paragraph). In addition, ChIP of Taz1-GFP confirmed the presence of Taz1 at the 2R-48 bp before the I-SceI cut and showed that Taz1 was enriched after the generation of the proto-telomere (Figure 1G). Sequencing of the elongated 2R-48bp fragments extracted from agarose gel revealed that an average of 63 bp of telomeric repeats were added to the proto-telomere (Figure 1H and Supplementary Figure S1C), slightly lower than that estimated by Southern blot. This could be explained by a bias during the telomere cloning step performed for telomere sequencing. Sequencing of the shortened 2R-48 bp fragments extracted from agarose gel in the ter1Δ strain revealed that the degradation of the I-SceI and polylinker sequence was observed in most cases, but importantly telomeric repeats were maintained (Figure 1I).

From these results, we concluded that telomerase is able to elongate the short proto-telomere generated by the I-SceI cut in quiescence, despite the absence of cell cycle-dependent control and the passage of the replication fork. Importantly, in quiescence the proto-telomere reached a final length of ∼130 bp after extension by the telomerase (48 bp from the seed plus the 80 bp of extension), a size significantly shorter than that produced by 2R-48bp extension in vegetative cells (250 bp) (40). Figure 1J shows that when cells undergo exit from G0 and cell cycle re-entry they further extend the 2R-48 bp up to the size of 180–250 bp, revealing a steady state size difference between G0 and cycling cells. This size difference might reveal the existence of a negative feedback control of telomerase action in G0 or might indicate that telomerase is less active in quiescent than in vegetative cells.

Notably, in most of our experiments we observed that a small fraction of cut fragment appeared before the induction of I-SceI by ahTet (Figure 1D). We confirmed that early cleavage occurred when cells enter into quiescence by monitoring ura4⁺-2R-48 bp directly after nitrogen starvation (Supplementary Figure S1D). This observed premature cleavage was likely caused by the negative control of adh1 promoter that occurred upon nitrogen starvation, leading to decreased level of tetracycline repressor (34). Indeed, in the absence of I-SceI no cleavage was observed (Supplementary Figure S1E). For this reason, we wanted to induce I-SceI expression as soon as cells are arrested, 6–8 h after starvation as previously described (43). FACS analysis confirmed that cells fully entered into quiescence after 8 hours (Supplementary Figure S1D), thus we decided to induce the cut after 8 hours of starvation in the following experiments.

Both NHEJ and telomerase act upon the 48 bp proto-telomere

In fission yeast arrested in G0, NHEJ is a prominent pathway to repair DSB (44–46). We wondered whether a competition exists between the elongation of the proto-telomere by telomerase and the repair of the I-SceI break by NHEJ. We thus analyzed the fate of the proto-telomere in cells lacking Lig4 and Ku70, two main players of the NHEJ pathway. In the absence of Lig4, the 2R-48bp was elongated and the ura4⁺::2R-48 bp was stabilized over the G0 period (Figure 2A). The uncut band completely disappeared at D1 in lig4Δ cells, in contrast to what observed in lig4⁺ cells (Figure 1D), indicating that NHEJ was efficiently acting upon the proto-telomere in quiescence. I-SceI digestion of the ura4⁺::2R-48 bp::hph⁺ PCR-amplified fragments revealed that a fraction of this fragment become resistant to I-SceI cleavage (Supplementary Figure S2A), indicating that error-prone NHEJ occurs in lig4⁺ cells, and explaining the persistence of the uncut fragment observed in Figure 1D (47). Despite the fact that I-SceI cleaved more efficiently the ura4⁺::2R-48bp::hph⁺ in lig4Δ cells, the extent of elongation by telomerase remained similar to that in lig4⁺ cells (Figure 2A and D). We next investigated the impact of Ku complex on the elongation of the 2R-48 bp (Figure 2B). In the absence of Ku70, the uncut fragment progressively shortens, likely due to exonuclease activity, indicating that NHEJ was inefficient similarly to that in lig4Δ cells. Importantly, the length of the extended 2R-48bp increased compared to pku70⁺ cells (Figure 2D) with the concomitant gradual degradation of the cut fragment (Figure 2B). In line with this, we also observed degradation of the 2R-0 bp (Supplementary Figure S2B). Taken together, these results suggest that the Ku heterodimer promotes NHEJ, protects the extremity of the proto-telomere, and prevents telomerase action. Sequencing of the shortest ura4+::2R-48 bp fragment extracted from agarose gel at D7 in pku70Δ cells confirmed the shortening of the telomeric seed (Figure 2E).

To confirm the presence of Ku at the 2R-48 bp, we performed ChIP of myc-tagged Ku70 protein at the 2R-48 bp in lig4Δ cells. We observed a significant enrichment of Ku70 after I-SceI induction (Figure 2F). Consistent with this result, telomerase was more enriched at the proto-telomere in the absence of Ku70 compared to WT strain (Figure 2G and Supplementary Figure S1B). These observations indicate that the Ku complex protects the proto-telomere from degradation and at the same time hampers telomere extension by telomerase. Deleting lig4⁺ in pku70Δ cells did not further modify the I-SceI cleavage pattern or the 2R-48 bp length (Figure 2C and D).

Overall, our results indicate that Ku is required to promote NHEJ that is the prominent pathway operating at the 2R-48 bp. Telomerase is also able to elongate the proto-telomere in a fraction of cells, however the absence of Ku does not increase the number of cells able to elongate the proto-telomere. This result indicates that Ku interferes with the processivity of telomerase engaged with the proto-telomere, but does not affect the initiation step. Based on these observations, we decided to monitor the fate of the proto-telomere in the absence of Lig4 in the following experiments to avoid the interference of the NHEJ.

Rad3^ATR controls telomerase activity through Ccq1 phosphorylation

In cycling cells, telomerase recruitment to telomeres is restricted to late S/G2 phase and telomere elongation is intimately linked to replication. Rad3^ATR and Tel1^ATM phosphorylate Ccq1 on threonine 93 (T93), Rad3^ATR being the prominent kinase involved in telomerase activation. In the absence of replication, the question of telomerase activity regulation arises. We found that the 2R-48bp was not elongated in ccq1-T93A cells (Figure 3A), indicating that Ccq1 phosphorylation is a crucial step for telomerase activation in quiescent cells. Next, we monitored elongation of the proto-telomere in mutants of rad3⁺ and tel1⁺ carrying a deletion of the kinase domain (rad3-kdΔ, tel1-kdΔ and rad3-kdΔ tel1-kdΔ double mutants). While the elongation of the 2R-48bp was not affected in tel1-kdΔ (Figure 3B), extension of the proto-telomere was reduced in rad3-kdΔ cells (Figure 3C), and completely abolished in the rad3-kdΔ tel1-kdΔ double mutant as in ccq1-T93A cells (Figure 3D). We concluded that both Rad3^ATR and Tel1^ATM are important to control the elongation of the proto-telomere in G0, Rad3^ATR being the prominent kinase such as in cycling cells (48,49).

Current models propose that ATR-ATRIP is recruited to DNA damage sites by interacting with replication protein A (RPA)-covered single-stranded DNA (ssDNA) (50,51). By analogy, Rad3^ATR-Rad26^ATRIP is thought to be recruited by RPA at telomeres to phosphorylate Ccq1 during telomere replication in fission yeast (48). Because the proto-telomere extension mainly relies on Rad3 in G0 cells, it suggests that RPA-coated ssDNA, generated by resection of the 2R-48 bp could be a key step for the proto-telomere extension. We were unable to demonstrate by ChIP the enrichment of Rpa1 at the 2R-48 bp after induction of the cut. However, we showed that Mre11, a member of the MRN complex that is important to initiate resection, likely by clipping Ku from double-stranded DNA (52,53), was indeed required for the proto-telomere extension in quiescent cells (Figure 3E). This suggests that Mre11-dependent resection of the 2R-48bp and the eviction of Ku are important to promote proto-telomere extension.

Taz1 and Rap1 are required for proto-telomere elongation in quiescent cells

In cycling cells, loss of Taz1 or Rap1 results in highly elongated telomeres, indicating that these factors inhibit telomere elongation by controlling telomerase activity (37,38). Hence, we investigated the involvement of these telomeric proteins in the 2R-48 bp elongation. Strikingly, the absence of Taz1 or Rap1 strongly impaired the extension of the proto-telomere in quiescent cells since only a limited extension of the proto-telomere was detected in a small fraction of cells (Figure 4A and B). This is in contrast with the situation in vegetative cells where Taz1 and Rap1 act as negative regulators of telomerase. We demonstrated that by showing that the proto-telomere was indeed overextended when the cut was induced in taz1Δ and rap1Δ cycling cells (Figure 4C). Therefore, the role that both Taz1 and Rap1 play during extension of the proto-telomere is different in vegetative and quiescent cells.

Because the proto-telomere lacks a canonical telomeric overhang, we speculated that the function of Taz1 is to promote the shelterin assembly at the cut site. To test this hypothesis, we analyzed the fate of the proto-telomere in rap1-I655R and tpz1-I200R mutants which are deficient for the interaction between Taz1 and Pot1, respectively (54,55). In both mutants, the extension of the 2R-48 bp is impaired (Figure 4D and E), showing that the shelterin bridge is crucial for elongation of the proto-telomere by telomerase.

To strengthen the role played by both Taz1 and Rap1 in promoting telomerase action in G0, we further monitored the recruitment of Ccq1 protein at the 2R-48bp (Figure 4F). After the induction of the I-SceI cut, a strong recruitment of Ccq1 protein was observed, which was significantly reduced in the absence of Taz1 and Rap1. These data confirm the role of Taz1-Rap1 in G0 for the proto-telomere elongation, and the need of the presence of Taz1-Rap1 to recruit Ccq1. Based on our data, we speculate that the role of Taz1 and Rap1 is to recruit the other telomeric elements to allow telomerase recruitment at the 2R-48 bp. Remarkably, Taz1 and Rap1 play different roles in the proto-telomere elongation in cycling and quiescent cells.

Stn1 and Rif1 negatively regulate proto-telomere extension

We next analyzed the role of the Stn1-Ten1 complex that inhibits telomerase action by promoting the synthesis of the complementary strand and whose recruitment to telomere occurs through its interaction with the SUMO-modified Tpz1 (56–59). The stn1-226 allele carries point mutations in the SIM domain of Stn1 that provokes the telomerase-dependent elongation of telomeres (57). In line with these results, we observed that in stn1-226 cells, the 2R-48bp was overelongated compared to stn1⁺ cells (Figure 5A). Our data indicate that inhibition of telomere elongation is similarly controlled by Stn1-Ten1 complex in quiescence.

Rif1 is also a negative regulator of telomerase activity (38). Thus, we further investigated the involvement of Rif1 in the control of telomere elongation in G0. Rif1 controls the timing of telomere replication by repressing the firing of late origins via its associated PP1 phosphatases (60–62). As a consequence, early replication of telomeres in cycling cells has been proposed to affect the length of telomeres (23). We found that the 2R-48 bp was also overelongated in rif1Δ quiescent cells (Figure 5B and D) in conditions in which replication origins are dormant. Moreover, in rif1-PP1 cells expressing a Rif1 mutant form unable to bind PP1 phosphatases (63), the 2R-48 bp was extended as in rif1Δ cells (Figure 5C and D). These results suggest that Rif1 regulates telomere-length in quiescence through its associated PP1 phosphatases. We thus hypothesized that Rif1-PP1 may control telomerase activity by dephosphorylating Ccq1. Hence, we monitored in vegetative and in quiescent cells the phosphorylation of Ccq1 in rif1Δ and rif1-PP1 mutants. As previously reported (39), we confirmed that phosphorylation of Ccq1 was stabilized in rif1Δ cells and also in the rif1-PP1 mutant (Figure 5E). Strikingly, we were also able to detect Ccq1 phosphorylation stabilization after induction of I-SceI in rif1Δ and rif1-PP1 quiescent cells (Figure 5F).

Ccq1 is phosphorylated at multiple sites including the T93 residue known to control telomerase recruitment (30,31). We could not establish whether Rif1/PP1 directly controls the dephosphorylation of Ccq1 T93 amino acid, however our results indicate that Rif1/PP1 is responsible for Ccq1 dephosphorylation in both vegetative and quiescent cells in which it controls the proto-telomere extension. Our results also point to the molecular mechanism by which Rif1 negatively regulates telomerase activity in fission yeast in general.