Chromatin modifiers and recombination factors promote a telomere fold-back structure, that is lost during replicative senescence

Telomeres adopt a lariat conformation and hence, engage in long and short distance intra-chromosome interactions. Budding yeast telomeres were proposed to fold back into subtelomeric regions, but a robust assay to quantitatively characterize this structure has been lacking. Therefore, it is not well understood how the interactions between telomeres and non-telomeric regions are established and regulated. We employ a telomeric chromosome conformation capture (Telo-3C) approach to directly analyze telomere folding and its maintenance in S. cerevisiae. We identify the histone modifiers Sir2, Sin3 and Set2 as critical regulators for telomere folding, which suggests that a distinct telomeric chromatin environment is a major requirement for the folding of yeast telomeres. We demonstrate that telomeres are not folded when cells enter replicative senescence, which occurs independently of short telomere length. Indeed, Sir2, Sin3 and Set2 protein levels are decreased during senescence and their absence may thereby prevent telomere folding. Additionally, we show that the homologous recombination machinery, including the Rad51 and Rad52 proteins, as well as the checkpoint component Rad53 are essential for establishing the telomere fold-back structure. This study outlines a method to interrogate telomere-subtelomere interactions at a single unmodified yeast telomere. Using this method, we provide insights into how the spatial arrangement of the chromosome end structure is established and demonstrate that telomere folding is compromised throughout replicative senescence. Author summary Telomeres are the protective caps of chromosome ends and prevent the activation of a local DNA damage response. In many organisms, telomeres engage in a loop-like structure which may provide an additional layer of end protection. As we still lack insight into the regulation of the folded telomere structure, we used budding yeast to establish a method to measure telomere folding and then study the genetic requirements for its establishment. We found that cells require the homologous recombination machinery as well as components of the DNA damage checkpoint to successfully establish a folded telomere. Through the deletion of telomerase in budding yeast, we investigated how telomere folding was regulated during replicative senescence, a process that occurs in the majority of telomerase negative human cells. During senescence, telomeres gradually shorten and erode until cells stop dividing which is a potent tumor suppressor and prevents unscheduled growth of potential cancer cells. We found, that the folded telomere structure is compromised as part of the cellular senescence response, but not due to telomere shortening per se. We think, that an altered telomeric chromatin environment during senescence is important to maintain an open state – which may be important for signaling or for repair.


Abstract 23
Telomeres adopt a lariat conformation and hence, engage in long and short distance intra-chromosome 24 interactions. Budding yeast telomeres were proposed to fold back into subtelomeric regions, but a robust 25 assay to quantitatively characterize this structure has been lacking. Therefore, it is not well understood 26 how the interactions between telomeres and non-telomeric regions are established and regulated. We 27 employ a telomeric chromosome conformation capture (Telo-3C) approach to directly analyze telomere 28 folding and its maintenance in S. cerevisiae. We identify the histone modifiers Sir2, Sin3 and Set2 as 29 critical regulators for telomere folding, which suggests that a distinct telomeric chromatin environment 30 is a major requirement for the folding of yeast telomeres. We demonstrate that telomeres are not folded 31 when cells enter replicative senescence, which occurs independently of short telomere length. Indeed, telomere structure and function are controlled by a six-protein complex called Shelterin, that binds to 57 telomeres in a sequence specific manner (1). In yeast, these functions are executed by the CST (Cdc13-58 Stn1-Ten1) complex (2) together with Rap1, Rif1 and Rif2 (3-5). Although telomeres resemble a one-59 ended DNA double strand break (DSB) they are refractory to being acted upon by the DNA damage 60 response (DDR) (6). Hence, telomeres prevent illegitimate repair events that would cause chromosome 61 end-to-end fusions and lead to genome instability (6). This "end protection" property of telomeres has 62 largely been attributed to the associated protein binding complexes, but a telomeric loop structure 63 (t-loop) also appears to be critical. Telomeric lariat structures have been demonstrated in species ranging 64 from yeast to human (7-13), however, their properties may vary considerably between organisms. In 78 Human and mouse t-loops can be visualized via electron-and super resolution-microscopy (13,14,20).

79
The existence of looped structures could also be shown microscopically in K. lactis harboring 80 overelongated telomeres (11). The short length, and base composition, of S. cerevisiae telomeres prevent 81 such microscopy-based approaches, hence only genetic-and chromatin immunoprecipitation (ChIP) -82 based experiments have been employed to study telomere folding in yeast (12,15,16,19). A major 83 concern of the genetic approach is that the subtelomere is deleted and modified with a reporter  Fig 2A). Telo-3C analysis at PD 9 revealed that telomere folding of tlc1 183 mutants was comparable to wt telomeres, despite their telomeres being much shorter that wt telomeres 184 (Fig 2B, C). However, when tlc1 cells had critically short telomeres and approached the point of 185 telomeric crisis, telomere folding decreased significantly (PD 47, 56 and 64 in Fig 2C). In survivors, 186 however (PD 101 in Fig. 2C), telomeres re-acquired the ability to establish a fold-back structure. Using 187 a probe specific for telomere 1L, we confirmed that the length of telomere 1L approximately followed 188 the same shortening and recombination pattern as bulk telomeres (S1A Fig).  The fact that telomeres remain folded at PD 9 ( Fig 2C)  Auxin-inducible degron (AID*) (Morawska and Ulrich 2013) ( Fig 3A). As all of these complexes are 217 essential in yeast, this system allowed us to study their impact on telomere folding by degrading the

225
We tested the telomeric proteins Rif1 and Rif2 for their contribution to establishing telomere folding.

226
Both Rif1 and Rif2 were previously reported to regulate telomere structure as measured by the fold-227 back reporter gene (15). However, by Telo-3C we did not observe a telomere folding defect in either 228 rif1 or rif2 deletion mutants arguing that these two telomeric proteins might not contribute to the 229 establishment of telomere folding ( Fig 3C). This discrepancy between the two methods might arise from 230 the two different readouts employed, or due to the introduction of an artificial subtelomere with the 231 reporter gene. Additionally, the chromatin structure at telomeres of rif1 and rif2 mutants is altered which 232 affect natural and modified subtelomeres in different ways (4,5).  desilenced and TERRA levels were approx. 10x higher than wt ( Fig 4G). As previously reported the 281 increase in TERRA was heterogeneous in tlc1 cells, due to the stochastic appearance of critically short 282 telomeres ( Fig 4G) (66). TERRA levels were also slightly increased in sin3 and set2 mutants ( Fig 4G).

283
Therefore, the state of telomere folding may have effects on TERRA levels in the cell.

284
We show here, that Sir2, Sin3 and Set2 are involved in the regulation of telomere folding. The levels of 285 these histone modifiers are downregulated during senescence, which may contribute to the open state during this process. Interestingly, Sir2, Sin3 and Set2 protein levels were not downregulated in a folding-287 proficient tel1 mutant with very short telomeres (S5B Fig, Fig 2D,

382
In mutants defective for telomere folding we did not observe an increased incidence of telomere-383 telomere fusions. Also when telomerase was deleted in addition to the histone modifiers Sir2, Sin3 and 384 Set2, telomere-telomere fusion did not increase. This suggests that the non-homologous end joining 385 repair pathway cannot act on unfolded telomeres and that they remain protected from fusions ( Fig 5B). and permissive to repair events (Fig 5A, bottom).

395
In summary, we propose that the presence of the histone modifiers Sir2, Sin3 and Set2, as well as the 396 DDR and HDR factors, is required for the successful establishment of telomere folding (Fig 5)   desalted on a C18 StageTip as described (79) and analyzed by nanoflow liquid chromatography on an 532 EASY-nLC 1000 system (Thermo Scientific) coupled to a Q Exactive Plus mass spectrometer (Thermo 533 Scientific). The peptides were separated on a self-packed reverse phase capillary (75 µm diameter, 534 25 cm length packed with C18 beads of 1.9 µm (Dr Maisch GmbH). The capillary was clamped on an 535 electrospray ion source (Nanospray Flex™, Thermo Scientific). A 90 min gradient starting from 2 % -536 60 % gradient acetonitrile in 0.1 % formic acid was used at a flow of 225 nl/min. Data was collected in 537 data-dependent acquisition mode with one MS full scan followed by up to 10 MS/MS scan with HCD 538 fragmentation.

540
MS raw files were processed using the MaxQuant software (version 1.5.2.8) and the ENSEMBL 541 S. cerevisiae protein database (Saccharomyces_cerevisiae.R64-1-1.24). LFQ quantitation and match 542 between run options were activated. MaxQuant output files were analyzed using an in-house R script. 543 Known contaminants, reverse hits and protein groups only identified by site modification were excluded. 544 Identified protein groups (minimum 2 peptides, 1 of them unique) were further filtered to a minimum 545 of 2 quantification events per experiment. Missing values were imputed using a downshifted and 546 compressed beta distribution within the 0.001 and 0.015 percentile of the measured values for each 547 replicate individually. The LFQ intensities were log 2 transformed and a two sample Welch t-test was 548 performed. Volcano plots were generated by plotting -log 10 (p-values) and fold changes. The threshold 549 line for enriched proteins is defined with p-value = 0.05. Gene ontology analysis were performed with 550 the PantherDB.org (80) overrepresentation Test (Release 20190711) with the annotation database 551 released on 20190703. Fisher's exact test followed by FDR correction was applied to calculate the p-552 values. Heatmap for enriched proteins were generated using the "pheatmap" (version 1.0.12) or 553 "ggplot2" (version 3.2.1) package in R. Additionally the Venn diagram was generated using the "eulerr" 554 (version 5.1.0) package in R (version 3.5.1).

555
Telomere-telomere fusions analysis 556 Telomere-telomere fusions (T-TFs) were analyzed by semiquantitative and quantitative PCR analyses 557 as reported (39), with the modifications described in (40). Briefly, ~100 ng of Sau3A-treated genomic 558 DNA extracted by standard protocols from asynchronous cultures was PCR-amplified for 559 semiquantitative analyses using a primer from the X element of chromosome XV-L and a primer from 560 the Y' element of chromosome V-R. A DNA fragment from HIS4 was PCR-amplified as input control. 561 T-TFs and HIS4 were PCR-amplified using 35-40 or 20 cycles, respectively, under the conditions 562 previously reported, except for the annealing temperature (60 °C) and the extension time (30 s). 563 Quantitative analyses were performed by real-time PCR by using the same amount of Sau3A-treated 564 genomic DNA, the oligonucleotides used for semiquantitative analyses and the PCR conditions 565 described previously (Mieczkowski et al, 2003). The frequency of T-TFs per genome was calculated 566 with the formula: T-TFs/genome = 2 -N / N = Ct (T-TFs) -Ct (HIS4). Prior to applying this formula, the 567 curves representing the increasing amounts of DNA for the two products as a function of the number of 568 PCR cycles were confirmed to be parallel (i.e., the slope of the curve representing the log of the input 569 amount versus ∆Ct was < 0.1).

571
Strains carrying auxin-inducible degrons (AID*) for SMC complexes were created as described before 572 (81). Primers for tagging are listed in S2 Table. 573 The RAD53-AID*-9MYC construct was transferred from the strain published by Morawska and Ulrich 574 2013 into the S288C background. For this purpose, the c-terminal sequence of RAD53 including the tag 575 was amplified by PCR and integrated into a strain carrying leu2::AFB2::LEU2 (yKB244).

576
Acknowledgements 577 We thank the Luke lab members for support and discussions, the Media lab, Flow cytometry and Protein 578 production core facilities of the IMB. We thank Alex Orioli, Annika Müller, and Diego Bonetti in the 579 assistance with strain construction. BL's lab was supported by the Heisenberg Program of the DFG -580 LU 1709/2-1. Research in FP's lab was funded by grants from the Spanish government (BFU2015-581 63689-P).