Rif2 protects Rap1-depleted telomeres from MRX-mediated degradation in Saccharomyces cerevisiae

Rap1 is the main protein that binds double-stranded telomeric DNA in Saccharomyces cerevisiae. Examination of the telomere functions of Rap1 is complicated by the fact that it also acts as a transcriptional regulator of hundreds of genes and is encoded by an essential gene. In this study, we disrupt Rap1 telomere association by expressing a mutant telomerase RNA subunit (tlc1-tm) that introduces mutant telomeric repeats. tlc1-tm cells grow similar to wild-type cells, although depletion of Rap1 at telomeres causes defects in telomere length regulation and telomere capping. Rif2 is a protein normally recruited to telomeres by Rap1, but we show that Rif2 can still associate with Rap1-depleted tlc1-tm telomeres, and that this association is required to inhibit telomere degradation by the MRX complex. We find that Rap1, Rif2, and the Ku complex work in parallel to prevent telomere degradation, and absence of all three at telomeres causes lethality. The partially redundant mechanisms may explain the rapid evolution of telomere components in budding yeast species.


Introduction 29
Telomeres, nucleoprotein complexes located at the ends of eukaryotic chromosomes, protect 30 chromosome ends from degradation, from telomere-telomere fusion events, and from being 31 recognized as double-stranded DNA breaks (Wellinger & Zakian, 2012;de Lange, 2018). In 32 most eukaryotic species, telomeric DNA consists of tandem G/C-rich repeats of double-33 stranded DNA with the G-rich strand extending to form a 3′ single-stranded overhang. These 34 repeats are bound by specialized proteins-some to the double-stranded region and others to 35 the 3′ overhang-which are important for proper telomere function. Telomere length is 36 maintained by a dynamic process of shortening and lengthening. Telomeres shorten due to 37 incomplete DNA replication and nucleolytic degradation, and are lengthened by the action of 38 a specialized reverse transcriptase called telomerase (Wellinger, 2014). At its core, 39 telomerase consists of a catalytic protein subunit and an RNA subunit, and extends telomeres 40 by iterative reverse transcription of a short G-rich sequence to the 3′ overhang, using the 41 RNA subunit as a template. 42 Rap1 is the main double-stranded telomeric DNA binding protein in the budding 43 yeast Saccharomyces cerevisiae (Buchman et al., 1988;Conrad et al., 1990), with important 44 roles in regulating telomere length (Lustig et al., 1990;Conrad et al., 1990;Marcand et al., 45 1997), transcriptional silencing of subtelomeric genes (Kyrion et al., 1993), and preventing 46 telomere-telomere fusions (Pardo & Marcand, 2005). Rap1 mediates these functions in part 47 by recruiting additional proteins (i.e. Rif1, Rif2, Sir3, and Sir4) via its C-terminal domain 48 (Hardy et al., 1992;Wotton & Shore, 1997;Moretti et al., 1994). Expression of a mutant 49 Rap1 lacking the C-terminal domain, which retains the ability to bind telomeric sequences, 50 mimics the deletion of the RIF and SIR genes in terms of telomere length regulation and 51 subtelomeric gene silencing (Wotton & Shore, 1997;Kyrion et al., 1993). However, the 52 (B) Telomere Southern blot analysis of TLC1 and tlc1-tm cells (three clonal isolates per genotype). Black arrowhead indicates a 1.8 kb DNA fragment generated from the BsmAI-digestion of plasmid pYt103 (Shampay et al., 1984). The major terminal restriction fragment is below the 1.8 kb control band.
(C) Southern blot analysis of the arti cial VII-L telomere, with either wild-type or tlc1-tm mutant sequence, using a probe to the adjacent URA3 gene. Multiple clones were examined, with each clone propagated for 1 to 4 passages (each passage corresponds to approximately 25 generations). A wild-type strain (lacking the arti cial VII-L telomere) was used as a control. (D, F, G) ChIP-qPCR analysis of the association of (D) Protein A-tagged Rap1, (F) Myc-tagged Cdc13, and (G) Myc-tagged Tbf1 to six Y' telomeres, the VII-L telomere, or to the non-telomeric ARO1 locus. Untagged wild-type and tlc1-tm strains were used as controls. The mean percentage of input +/-SEM is shown (n=3, ** p<0.01).
(E) EMSA of Rap1 protein incubated with radiolabeled oligonucleotides with either wild-type or tlc1-tm mutant telomeric sequence.
7 al., 2006). Thus, we assessed binding of Tbf1 to tlc1-tm telomeres. We find no change in the 128 levels of Tbf1 at tlc1-tm telomeres compared to wild-type telomeres by ChIP-qPCR ( Fig.  129 1G), consistent with neither wild-type nor mutant telomere sequences containing the RCCCT 130 Tbf1 consensus binding sequence (Preti et al., 2010), indicating that the loss of Rap1 is not 131 compensated by recruitment of Tbf1. The Tbf1 association observed at the native Y 132 telomeres is due to the presence of TTAGGG repeats at subtelomeric regions (Brigati et al., 133 1993), while the association observed at the artificial VII-L telomere is likely due to the 134 presence of several RCCCT motifs in the adjacent sequence ( Supplementary Fig. 1). 135 136 Loss of Rap1-mediated telomere length regulation at tlc1-tm telomeres 137 The decreased binding of Rap1 to tlc1-tm sequences likely explains the long, heterogeneous-138 sized telomeres in tlc1-tm strains, because Rap1 negatively regulates telomerase through 139 what is called the "protein counting" model. This model posits that Rap1, through its 140 recruitment of Rif1 and Rif2, inhibits telomerase; the longer a telomere is, the more Rap1, 141 Rif1, and Rif2 will be present, and the stronger the inhibition of telomerase will be (Marcand 142 et al., 1997;Levy & Blackburn, 2004). Reduced binding of Rap1 would cause tlc1-tm 143 telomeric sequences to not be recognized as telomeric in terms of Rap1-mediated telomere 144 length regulation. To test this hypothesis, we again modified telomere VII-L to generate a 145 telomere that would be seeded with either 84 bp or 300 bp of tlc1-tm telomeric sequence, but 146 in a TLC1 strain expressing wild-type telomerase so that the tip of the telomere would contain 147 wild-type telomeric sequences. In both cases, the size of the telomere VII-L terminal 148 restriction fragment increased in size, with the magnitude of the increase roughly equivalent 149 to the size of the tlc1-tm telomeric seed sequence, indicating that this sequence is not being 150 sensed by the Rap1 protein counting mechanism ( Fig. 2A).  Figure 2. Telomere length regulation is disrupted in tlc1-tm cells (A) Southern blot analysis of the arti cial VII-L telomere using a probe to the adjacent URA3 gene. The telomere was seeded with either wild-type or tlc1-tm mutant sequence of the indicated lengths in a strain expressing wild-type TLC1. Multiple clones of each strain were examined. A wild-type strain (lacking the arti cial VII-L telomere) was used as a control. (B) In vivo extension of tlc1-tm telomeres was examined using the iSTEX assay. Telomere VI-R was ampli ed and sequenced after the induction of wild-type telomerase. Each bar represents an individual telomere. The black and red portions of each bar represent wild-type and tlc1-tm sequence, respectively, that is identical in sequence and thus present before telomerase induction. The length of the black/wild-type sequence is 48 bp. Sequence that is divergent from the black and red sequence is shown in grey and green. Grey represents newly added wild-type sequence after the induction of telomerase. Green represents divergent tlc1-tm sequence, most likely a result of homologous recombination. Telomeres are sorted based on the length of the undiverged (black plus red) sequence. (C) Telomere VI-R sequences obtained from the iSTEX analysis in B were binned into groups of 10 nt in size according to telomere length before telomerase induction. iSTEX data for the extension of wild-type telomeres were taken from previous studies (Strecker et al., 2017;Stinus et al., 2017) and included for comparison. Groups containing less than four telomeres were excluded from this analysis. Frequency of extension and average telomere length before telomerase induction were calculated and plotted for each group. Logarithmic regression curves for each data set were also included in the plot. Telomeres shorter than 40 nt before telomerase induction, which are not e ciently recognized and extended by telomerase (Strecker et al., 2017), were removed from the regression analysis.

8
To examine telomere length regulation in tlc1-tm strains further, we monitored 152 telomerase-mediated telomere extension events at nucleotide resolution after a single cell 153 cycle using the iSTEX (inducible Single Telomere EXtension) assay (Strecker et al., 2017). 154 At wild-type telomeres, telomerase extends only a subset of telomeres in each cell cycle, with 155 a strong preference for the extension of short telomeres (Teixeira et al., 2004;Strecker et al., 156 2017). At tlc1-tm telomeres, we find that telomere extension frequency is dramatically 157 increased, with nearly all (92%) of telomeres being extended during a single cell cycle (  hypothesized that while decreased Rap1 at tlc1-tm telomeres may favor their extension by 169 telomerase, increased degradation may limit the length of tlc1-tm telomeres. If true, removal 170 of telomerase should trigger rapid entry into senescence. To test this idea, we sporulated 171 diploid strains that were heterozygous for EST2, which encodes the protein catalytic subunit 172 of telomerase (Lingner et al., 1997), or TLC1, with either wild-type or mutant telomeres (i.e. 173 est2Δ/EST2 versus est2Δ/EST2 tlc1-tm/tlc1-tm and tlc1Δ/TLC1 versus tlc1Δ/tlc1-tm), and 174 performed senescence assays with the haploid meiotic progeny. In the presence of wild-type 175 telomeres but absence of telomerase (est2Δ and tlc1Δ), telomeres shortened until the cells 9 senesced after 60 to 70 population doublings, as expected (Fig. 3A). A small subset of the 177 senescent population was then able to lengthen the telomeres by recombination-mediated 178 mechanisms, forming so-called survivors (Lundblad & Blackburn, 1993). We found that cells 179 containing mutant telomeres senesced rapidly, only ~40 population doublings after 180 telomerase loss (Fig. 3A, est2Δ tlc1-tm* and tlc1Δ*). We examined the telomere length of 181 cells that had undergone ~30 population doublings after isolation of the haploid spores and 182 found that, upon loss of telomerase, mutant telomeres became extremely heterogeneous and    Figure 1B.   Figure 4. Recruitment of Rif1 and Sir4, but not Rif2 nor Sir3, is significantly reduced at tlc1-tm telomeres (A) ChIP-qPCR analysis of the association of Myc-tagged Rif1, Rif2, Sir4, and Protein A-tagged Sir3 to six Y' telomeres, the VII-L telomere, or to the non-telomeric ARO1 locus. Untagged wild-type and tlc1-tm strains were used as controls. (B) Total RNA was reverse transcribed and TERRA from speci c telomeres (I-L and XV-L) was analyzed by qPCR. TERRA values were normalized to ACT1 levels, and to the respective wild type (TLC1). (C) Tenfold serial dilutions of strains with the indicated genotypes were spotted on SC plates without or with 5-FOA. (D) The expression of the subtelomerically integrated URA3 gene was measured in the indicated yeast strains by RT-qPCR. All data (except C) are shown as mean +/-SEM (n=3, *** p<10 -3 , **** p<10 -4 ).
In this study, we investigated the consequences of depleting Rap1 from S. cerevisiae 325 telomeres. Our findings suggest that impairing Rap1 binding does not significantly affect cell 326 proliferation. The depletion of telomeric Rap1 causes defects in telomere length regulation 327 and telomere capping. Surprisingly, Rif2 is still recruited to Rap1-depleted telomeres, and its 328 recruitment is dependent on the MRX complex. tlc1-tm telomeres become extensively 329 degraded in the absence of Rif2 or the Yku complex, and rif2∆ yku70∆ tlc1-tm triple mutants 330 are inviable. Thus, Rap1, Rif2, and the Yku complex perform separate tasks at telomeres that 331 are together essential for telomere capping and cell viability. 332 Telomere capping is essential for cell viability; yeast cells cannot survive the loss of a 333 single telomere (Sandell & Zakian, 1993). Telomere capping is executed by telomere-binding 334 proteins, so it was surprising to find that yeast cells can survive when Rap1, the main protein 335 that binds double-stranded telomeric DNA, is significantly reduced at telomeres. Previous 336 studies have reported that telomeres of S. cerevisiae expressing mutant telomerase that adds 337 TTAGGG repeats are also devoid of Rap1; however, the absence of Rap1 in this situation 338 appears to be compensated by the binding of Tbf1 to the TTAGGG sequence (Alexander & 339 Zakian, 2003;Brevet et al., 2003;Berthiau et al., 2006). In contrast, there is no change in 340 Tbf1 binding at tlc1-tm telomeres (Fig. 1G). The difference between yeast telomeres with 341 TTAGGG repeats and tlc1-tm telomeres is further highlighted by the fact that Rif2 is absent 342 at the former (Alexander & Zakian, 2003), but is crucial for telomere protection of the latter. Previously studied tlc1 template mutations that abolish Rap1 binding to the resulting 348 mutant telomere sequences generally fall into two categories; they either cause telomere 349 shortening or they cause rapid telomere elongation, which in some cases is accompanied by 350 extensive telomere degradation (Prescott & Blackburn, 2000;Lin et al., 2004), similar to that 351 seen in rif2∆ tlc1-tm mutants (Fig. 5B). The first category is likely due to a loss of telomerase 352 enzymatic activity, because most tlc1 template mutations result in a reduction in the 353 nucleotide addition processivity of telomerase (Förstemann et al., 2003). The extent of 354 telomere elongation and degradation in the second category appears to be correlated with the 355 decrease in Rap1 binding (Prescott & Blackburn, 2000). Rap1 association to fully mutant 356 tlc1-tm telomeres is reduced to approximately 13% compared to wild-type telomeres (Fig.  357   1D). We suspect that this level of Rap1 prevents the more extensive elongation and 358 degradation seen in several other tlc1 template mutants (e.g. tlc1-476A; Fig. 7A) that likely 359 have even less Rap1 telomere association (Prescott & Blackburn, 2000;Lin et al., 2004). 360 Consistent with this hypothesis, tlc1∆ cells derived from the sporulation of tlc1-tm/tlc1∆ 361 diploids with mutant telomeres also exhibit extensive degradation (Fig. 3B), which is likely 362 the result of further reducing Rap1 telomere association due to telomere shortening. 363 Our findings build upon previous work to show that Rap1 and Rif2 inhibit the MRX 364 complex to prevent telomere degradation. First, decreased telomere association of Rap1 in 365 tlc1-tm cells (Fig. 1D) is accompanied by an increase in Mre11 telomere association (Fig.  366 6A), which is consistent with a previous report showing that Rap1 binding inhibits Mre11 367 recruitment (Negrini et al., 2007). Second, the telomere degradation observed in rif2∆ tlc1-tm 368 cells is eliminated by deletion of RAD50 (Fig. 6B), which is consistent with previous studies The association of Rif2 to Rap1-depleted tlc1-tm telomeres was unexpected given the 390 well-characterized role of the Rap1 C-terminal domain in recruiting Rif2 to telomeres 391 (Wotton & Shore, 1997;Shi et al., 2013). We find that this Rap1-independent recruitment of 392 Rif2 is dependent on the MRX complex (Fig. 7D), which itself is increased at tlc1-tm 393 inverse is true of Rif2 at DSBs and telomeres. Thus, Sae2 appears to limit Rif2 function at 400 DSBs, and Rif2 may limit the role of Sae2 at telomeres. In wild-type cells, the prominent role 401 of Rif2 at telomeres can be explained by its recruitment by Rap1. However, it is unclear how 402 Rif2 maintains its prominent role at telomeres without Rap1-dependent recruitment. One 403 possibility is that there is preferential recruitment of Sae2 to DSBs compared to telomeres. 404 Further studies are needed to examine this hypothesis. Healthcare). The membrane was hybridized to telomere-specific digoxygenin-labeled probe 425 (wild type probe: 5′-CACCACACCCACACACCACACCCACA-3′; tlc1-tm mutant probe: 426 5′-ACCACACCACACCACACACACCACAC-3′). For detection of the artificial VII-L 427 telomere, a similar procedure was performed, except that 6 µg yeast genomic DNA was 428 digested with EcoRV and the membrane was hybridized at 42ºC with a digoxigenin-labeled 429 probe complementary to URA3 sequence. Unless otherwise mentioned (i.e. Fig. 1C and 3B), 430 all strains were propagated for at least 100 generations before Southern blot analysis. 431 432

Chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR) 433
ChIP-qPCR was performed essentially as previously described (Graf et al., 2017 The recombinant S. cerevisiae Rap1 full-length protein was expressed in E. coli BL21 cells 444 as previously described (Wahlin & Cohn, 2000). The ability of Rap1 to bind the tlc1-tm 445