Rap1‐mediated nucleosome displacement can regulate gene expression in senescent cells without impacting the pace of senescence

Abstract Cell senescence is accompanied, and in part mediated, by changes in chromatin, including histone losses, but underlying mechanisms are not well understood. We reported previously that during yeast cell senescence driven by telomere shortening, the telomeric protein Rap1 plays a major role in reprogramming gene expression by relocalizing hundreds of new target genes (called NRTS, for new Rap1 targets at senescence) to the promoters. This leads to two types of histone loss: Rap1 lowers histone level globally by repressing histone gene expression, and it also causes local nucleosome displacement at the promoters of upregulated NRTS. Here, we present evidence of direct binding between Rap1 and histone H3/H4 heterotetramers, and map amino acids involved in the interaction within the Rap1 SANT domain to amino acids 392–394 (SHY). Introduction of a point mutation within the native RAP1 locus that converts these residues to alanines (RAP1SHY), and thus disrupts Rap1‐H3/H4 interaction, does not interfere with Rap1 relocalization to NRTS at senescence, but prevents full nucleosome displacement and gene upregulation, indicating direct Rap1‐H3/H4 contacts are involved in nucleosome displacement. Consistent with this, the histone H3/H4 chaperone Asf1 is similarly unnecessary for Rap1 localization to NRTS but is required for full Rap1‐mediated nucleosome displacement and gene activation. Remarkably, RAP1SHY does not affect the pace of senescence‐related cell cycle arrest, indicating that some changes in gene expression at senescence are not coupled to this arrest.

secretion of factors that influence the function of tissues in which senescent cells reside (Campisi, 2013;Ritschka et al., 2017;Sapieha & Mallette, 2018;van Deursen, 2014). Although cell senescence plays beneficial roles early in life by contributing to tumor suppression, wound healing, and immunity, several lines of evidence suggest that it can also drive age-related pathologies through stem cell depletion Molofsky et al., 2006) or by disruption of tissue structure and function, apparently via secretion of factors such as proteases and inflammatory cytokines (Baar et al., 2017;Baker et al., 2016Baker et al., , 2011Childs et al., 2016;Schafer et al., 2017). Understanding the mechanisms underlying cell senescence, particularly those regulating altered gene expression, is thus of substantial interest.
One important driver of human cellular senescence is telomere shortening. Critically short (i.e., "uncapped") telomeres are recognized by the DNA damage response (DDR) machinery, leading to arrest and gene expression changes. Senescence driven by telomere shortening can be modeled in Saccharomyces cerevisiae. Yeast naturally expresses telomerase to maintain telomere length, but if telomerase is inactivated genetically, cells gradually lose telomeric DNA through rounds of division and eventually arrest-although rare survivors, which maintain telomeres via homologous recombination, eventually emerge from senescent populations. Many factors known to influence senescence in human cells have similar roles in telomerase-deficient yeast, including exonucleases, helicases, and DDR proteins (Herbig, Jobling, Chen, Chen, & Sedivy, 2004;IJpma & Greider, 2003;Johnson et al., 2001;Ritchie, Mallory, & Petes, 1999;Schaetzlein et al., 2007).
A key and conserved feature of senescence, and other types of aging-related biology, from yeast to humans is histone loss (Ivanov et al., 2013;Liu et al., 2013;O'Sullivan & Karlseder, 2012;Platt et al., 2013;Song & Johnson, 2018). Histone gene expression, global levels of all core histones, and nucleosome occupancy at particular genomic sites are all decreased in senescent telomerase-deficient yeast (Platt et al., 2013), and similar observations have been made in aged yeast mother cells (Feser et al., 2010;Hu et al., 2014). This loss is apparently closely linked to the altered gene expression observed in senescent cells, as highly similar gene expression patterns are seen when histone levels are artificially downregulated (Platt et al., 2013). In both yeast models, artificial overexpression of core histones promotes longevity. However, little is known about the mechanisms underlying histone-related changes in senescent cells.
Previously, we found that the telomeric protein Rap1 plays a major role in replicative senescence in telomerase-deficient yeast, including regulation of histone gene expression and site-specific nucleosome occupancy (Platt et al., 2013). Rap1 is conserved between yeast and humans, and the yeast protein binds directly to telomere repeat DNA in a sequence-specific fashion via two tandemly arranged Myb domains, where it plays roles in regulating telomere length, transcriptional silencing, and capping (Kyrion, Liu, Liu, & Lustig, 1993;Marcand, Wotton, Gilson, & Shore, 1997; Martínez, Gómez-López, Pisano, Flores, & Blasco, 2016; Moretti & Shore, 2001;Pardo & Marcand, 2005;Rai, Chen, Lei, & Chang, 2016;Vodenicharov, Laterreur, & Wellinger, 2010;Yang et al., 2017). It also functions to regulate transcription throughout the genome, in particular repressing expression of the silent mating-type loci, and activating expression of approximately ten percent of all yeast genes, particularly the highly expressed ribosomal protein and certain glycolytic enzyme genes. During replicative senescence, Rap1 relocalizes from shortened telomeres and subtelomeres to the promoters of hundreds of new genes, named NRTS (new Rap1 targets at senescence), which have lower affinity Rap1 binding sites than natural Rap1 targets. Among the NRTS are the genes that encode the core histone proteins, which are transcriptionally repressed by Rap1, thus contributing to the loss of histone proteins observed at senescence. In contrast to the histone genes, the majority of NRTS become activated by Rap1. This activation is associated with the displacement by Rap1 of nucleosomes from the promoters of these NRTS, but it is not known if nucleosome displacement causes NRTS activation (Platt et al., 2013). Furthermore, Rap1 drives the overall pace of senescence, because it is delayed by experimental diminishment of Rap1 levels. However, whether the function of Rap1 to repress global histone levels, or its function to locally displace nucleosomes and upregulate NRTS, might underlie its effect on the rate of senescence has not been tested.
It has long been known that Rap1 can bind nucleosomal DNA, and its ability to exclude nucleosomes from promoters is functionally similar to pioneer transcription factors (pTFs) in higher eukaryotes (Ganapathi et al., 2011;Knight et al., 2014;Koerber, Rhee, Jiang, & Pugh, 2009;Kubik et al., 2015;Lickwar, Mueller, Hanlon, McNally, & Lieb, 2012;Rhee & Pugh, 2011;Yan, Chen, & Bai, 2018;Yarragudi, Miyake, Li, & Morse, 2004;Yu, Sabet, Chambers, & Morse, 2001;Zaret & Carroll, 2011). However, the mechanisms by which Rap1 displaces histones have not been thoroughly explored. It is possible that direct contacts between Rap1 and histones are involved, because proteome-wide interaction screens in yeast and genome-wide split-YFP complementation assays in human cells suggest that Rap1 proteins may bind histones (Gilmore et al., 2012;Lee et al., 2011), although this has not been studied in detail. This possibility is of general interest because histone binding has been so far described for only two other pTFs, FoxO1 and FoxA (Cirillo et al., 2002;Hatta & Cirillo, 2007). In addition, we reasoned that if Rap1-histone contacts are involved in histone displacement by locally bound Rap1, but not in other Rap1 functions including histone gene repression, then a Rap1 mutant selectively deficient in histone contact would provide a tool to not only address the role of nucleosome displacement in NRTS activation, but also test the importance of NRTS upregulation in driving the rate of senescence.
Here we describe the creation of such a Rap1 mutant, identified based on its disruption of a direct interaction established between Rap1 and histone H3/H4 heterotetramers, as well as the effects of the mutation on the functions of Rap1 at senescence. We also describe a role for a histone H3/H4 chaperone, Asf1, in nucleosome displacement and NRTS upregulation by Rap1.

| Rap1 binds H3/H4 histone tetramers
As reviewed above, Rap1 is a nucleosome-displacing factor, functionally similar to pTFs in higher eukaryotes. Consistent with this function of Rap1, we previously reported that Rap1 displaces nucleosomes from the promoters of activated NRTS at senescence.
In contemplating underlying mechanisms, we considered evidence indicating potentially direct binding between histones and the yeast and human Rap1 proteins. This evidence comes from a proteomewide screen in yeast and a genome-wide split-YFP fluorescence complementation screen in human cells (Gilmore et al., 2012;Lee et al., 2011), but the apparent Rap1-histone interactions have not been investigated in any detail. Such an interaction, if mapped, could provide us with tools to manipulate Rap1 functions at senescence. Therefore, we decided to explore the possibility that nucleosome displacement by Rap1 might involve direct interactions with histone proteins.
We performed the histone binding assay by incubating 0.5 μM of GST-tagged Rap1 proteins bound to glutathione beads with 2 μM H2A/H2B dimers or H3/H4 tetramers. The beads were then washed under stringent conditions, and retained histones were examined by SDS-PAGE. We found that Rap1 bound to H3/ H4 tetramers, but not H2A/H2B dimers, including under conditions where the H3/H4 and H2A/H2B proteins were mixed with one another prior to binding (Figure 1b). Significant binding was observed in salt concentrations ranging from 150 to 750 mM.
Rap1 N did not bind to H3/H4, whereas the Rap1 DBD showed similar binding strength compared to full-length Rap1 (Figure 1c).
When increased to levels equimolar to the histones (2 μM each), full-length Rap1 and Rap1 CΔ showed more robust binding to H3/ H4 tetramers, whereas the C-terminal fragment also displayed weak binding (Figure 1d,f). However, even in 7.5-fold molar excess, Rap1 N failed to bind histones ( Figure S1). Taken together, our findings indicate Rap1 interacts directly with the H3/H4 histone tetramer, which involves relatively strong versus weak F I G U R E 2 Amino acids 392-394 (SHY) facilitate Rap1-histone interactions. (a) Location of alpha-helices within the SANT domain, redrawn from Konig et al. (1996). Triple alanine mutants were generated from amino acids 359-410. (b) Immunoblot analysis of in vitro GST pull-down assay of histones showing representative triple alanine mutants. Pull-down was performed with equimolar GST-SANT (0.5 μM) and histones (0.5 μM) at 400 mM NaCl. Bottom panel is the blot stained with Ponceau S as a loading control. (c) Quantitation of triple alanine mutants binding to H3/H4, normalized to Ponceau stain signal, and with WT SANT set to 1.0. Error bars for all quantitations indicate standard error of the mean (N = 2). Only mutant 12 (amino acids 392-394, SHY) showed a significant loss of H3 signal. (d) Two views of the SANT domain bound to DNA. Amino acids SHY side chains are colored in magenta. SHY is located immediately C-terminal to helix 2, with side chains facing away from the Rap1-DNA interaction surface. Image generated using Pymol (PDB ID: 3UKG). (e) Representative immunoblot analysis of GST pull-down histone binding assay with full-length Rap1 and Rap1 SHY . Pull-down was performed with 0.5 μM each Rap1 and H3/H4 at 400 mM NaCl. Rap1 SHY displays a ~50% loss of histone binding. Bottom panel is a loading control gel stained with Coomassie blue. (f) Quantitation of full-length Rap1 and Rap1 SHY binding to H3/H4 (N = 3). (g) Representative immunoblot analysis of GST pull-down histone assay using two truncated versions of Rap1 lacking the C-terminus, Rap1 CΔ and Rap1 643 Δ . Pull-down was performed with 2 μM Rap1 truncated constructs and 2 μM H3/H4 at 400 mM NaCl. Both truncated forms show a significant and similar loss of histone signal when amino acid SHY is mutated to AAA (rightmost two lanes). Bottom panel is Coomassie loading control. (h) Quantitation of g (N = 3). (i) Representative coimmunoprecipitation of histone H3 with immunoprecipitated HA-Rap1 and HA-Rap1 SHY . Input is 5% of the WCE, and Rap1 SHY shows a significant loss of histone binding in the extracts. (j) Quantitation of the ratio of co-immunoprecipated H3 to input H3 signals in i (N = 3) binding interactions between histones and the Rap1 DBD versus C-terminus.
The Rap1 DBD consists of two tandem Myb domains, the first of which is also classified as a SANT domain (amino acids

360-410). The SANT domain is a stretch of approximately 50
amino acids containing the helix-turn-helix motif and is typically involved in protein-protein interactions. Some SANT domains can bind to histone tails and have been proposed to function as histone interaction modules important for nucleosome remodeling (Boyer et al., 2002;Boyer, Latek, & Peterson, 2004;Grüne et al., 2003). Therefore, we predicted that the H3/H4 interaction seen in the DBD involves interaction surfaces within the SANT domain. To test whether it might be sufficient for binding, we fused GST to the Rap1 SANT domain and confirmed that it binds the H3/H4 tetramers under equimolar concentrations (2 μM each), with approximately half the affinity of the full-length protein ( Figure 1e,f). These findings indicate that the SANT domain contributes substantially to the capacity of Rap1 to bind histone H3/ H4 tetramers.

| Amino acids 392-394 (SHY) facilitates Rap1histone interactions
To identify residues within the Rap1 SANT domain required for binding H3/H4 tetramers, we generated GST-tagged triple ala- To address whether the SHY patch impacts binding of Rap1 to soluble histones in vivo, we immunoprecipitated HA-tagged Rap1 and Rap1 SHY from whole-cell extracts (WCE) and immunoblotted for H3. The extracts were treated with benzonase to prevent indirect DNA-mediated interactions between the proteins. Rap1 SHY showed a significant loss of H3 binding compared to WT Rap1 (Figure 2i,j). Together with the in vitro pull-down data, this strongly supports a physical interaction between Rap1 and H3/H4 that involves amino acids in the SHY patch of the DNA binding domain.

| Rap1 SHY is deficient in NRTS activation and histone displacement
Given the physical interactions observed between Rap1 and H3/H4 tetramers, we proceeded to investigate the functional effects of Rap1 SHY in vivo, in particular its effects on the different functions of Rap1 at senescence. We hypothesized that the compromised ability of Rap1 SHY to interact with the H3/H4 tetrameric core of nucleosomes would interfere with its roles in NRTS promoter clearance and gene activation.
As the SHY to AAA mutation is within the DNA binding domain, we first confirmed that Rap1 SHY did not compromise the ability of Rap1 to bind DNA. Electrophoretic mobility shift assays (EMSA) using a telomeric sequence, a natural Rap1 binding site within the TEF2 promoter, and a representative NRTS promoter demonstrated Rap1 and Rap1 SHY bound DNA with similar affinities ( Figure S3a-c). We next used a system of Rap1 overexpression in wild-type cells, which we showed previously recapitulates the selective binding of Rap1 to NRTS promoters, from which nucleosomes are displaced and gene expression is upregulated.
Wild-type cells were transformed with 2-micron based plasmids from which either HA-tagged Rap1 or Rap1 SHY expression is driven by the GAL1 promoter. Expression was induced with galactose for 130 min, which we reported previously is sufficient for local histone displacement at promoters by Rap1 but avoids potential secondary effects from toxicity manifesting as growth inhibition after eight hours of induction (Platt et al., 2013). Rap1 localization to NRTS promoters and histone displacement were measured by ChIP-qPCR, using antibodies against the HA-tag and H3, respectively. Total cellular levels ( Figure 3a) and lo-

| Rap1 SHY confers diminished NRTS activation at senescence without affecting the rate of senescence
To examine the effects of Rap1 SHY in the context of senescence, we introduced the SHY to AAA mutation within one of the endogenous RAP1 loci in a TLC1/tlc1Δ diploid. Upon sporulation and dissection of tetrads, the Rap1 SHY haploid spore products formed smaller colonies ( Figure S4a). Southern blot analysis using probes for Y' telomeric fragments showed similar telomeric lengths in RAP1 and RAP1 SHY strains, as well as in their respective telomerase deletion (tlc1Δ) strains at 50 population doublings after spore germination ( Figure S4b), suggesting a normal level of telomere capping and maintenance by Rap1 SHY . This is consistent with the similar colony sizes observed for RAP1 SHY tlc1Δ double mutants and tlc1Δ strains, at least for the ~20-25 divisions needed for colony formation from the germinated spores ( Figure S4a). for the slow growth of the mutants. In addition, Rap1 SHY functions normally to silence subtelomeres ( Figure S4d,e) and the silent mating-type loci ( Figure S4f).
We passaged both tlc1Δ and Rap1 SHY tlc1Δ cells to senescence by measuring the daily growth of liquid cultures seeded at a fixed starting concentration with cells obtained from the previous day of growth (see Methods). Taking senescence as the nadir of the growth curve before survivor formation, Rap1 SHY had no effect on the rate of senescence compared to WT Rap1 (Figure 4a). However, given the reduced NRTS activation observed when Rap1 SHY is overexpressed, we predicted that a similarly blunted NRTS profile would also be seen in Rap1 SHY at senescence. Indeed, this was confirmed by comparing relative mRNA expression in senescent and proliferating cells (Figure 4b). Interestingly, this suggests that the tested gene expression changes do not correlate with the rate of senescence.
Previously, we have reported that Rap1 relocalization at senescence represses histone gene expression, and that artificial overexpression of all core histones will delay the rate of senescence (Platt et al., 2013), suggesting that the rate of senescence may be related to global histone levels. Consistent with this, the degree to which expression of all eight core histone genes was repressed at senescence was similar for cells expressing Rap1 SHY versus normal Rap1 ( Figure 4c).  (Gunjan & Verreault, 2003).

| Asf1 contributes to Rap1-dependent NRTS activation and histone displacement
We found that although asf1Δ tlc1Δ mutants did not senesce at a rate significantly different from tlc1Δ (Figure 5f), deletion of Asf1 significantly blunts the upregulation of activated NRTS at senescence (Figure 5a). To test if this reduced NRTS activation is related to regulation by Rap1, we tested the Asf1dependence of NRTS activation when Rap1 was overexpressed in wild-type cells. Similar to its effects at senescence, deletion of ASF1 reduced Rap1-driven NRTS activation ( Figure 5b) and displacement of nucleosomes from NRTS promoters (Figure 5d).
However, Rap1 localization to the promoters did not depend on Asf1 (Figure 5c) by overexpressed Rap1 (Figure 5e), but deletion of cac1Δ, cac2Δ, and cac3 Δ had no effect ( Figure S5). This is consistent with the fact that cells arrest in G2/M at senescence and therefore would not be expected to utilize the replication-dependent pathway.
Furthermore, similar to ASF1 deletion, HIR1 deletion did not affect the rate of senescence of tlc1Δ cells (Figure 5g)

| Direct Rap1-histone interactions are involved in Rap1-mediated chromatin opening
There is long-standing evidence that Rap1 can bind to nucleosomal DNA both in vivo and in vitro (Koerber et al., 2009;Lickwar et al., 2012;Rossetti et al., 2001). Notably, single-nucleotide resolution ChIP-exo shows that not only does histone occupancy not interfere with Rap1 binding, but that high-affinity Rap1 binding sites and Rap1 occupancy are in fact more common in nucleosomal than nonnucleosomal regions of the genome (Rhee & Pugh, 2011 (Rossetti et al., 2001) and chromatin opening (Yu et al., 2001).
The SHY patch is located within the turn immediately C-terminal to the second helix in the three-helix bundle of the SANT domain.
Crystal structures of Rap1 bound to DNA (Konig et al., 1996;Matot et al., 2012) show that helices 2 and 3 form a helix-turn-helix motif that docks deep in the major groove of DNA, and that the side  (Chen et al., 2017;Li & Widom, 2004).
Such "breathing" not only allows exposure of DNA sequences for transcription factor binding, but also bares the tetrameric core for protein-protein interactions. Therefore, Rap1 binding to H3/ H4 tetramers may drive the dynamic equilibrium of wrapped and partially unwrapped nucleosomes toward the unwrapped state, possibly by preventing H2A/H2B reassembly. Alternatively, or in addition, it may also alter the conformation of the tetrasome in a fashion that facilitates full nucleosome disassembly by H3/H4 histone chaperones such as Asf1. This is similar to the functions of some histone PTMs such as H3K56ac, which destabilize nucleosomes and enable them to be more easily disassembled (Williams, Truong, & Tyler, 2008).
Rap1 is functionally similar to pTFs in eukaryotes. pTFs are the first to bind to target sites in compact chromatin and initiate the F I G U R E 5 Asf1 is required for NRTS activation and histone displacement. (a) NRTS mRNA levels at senescence, measured by qPCR, and normalized to ACT1 and nonsenescent strains. asf1Δ tlc1Δ double mutants have reduced NRTS activation compared to tlc1Δ strains (N = 5, p < .025). (b) Asf1 is required for NRTS activation in response to NOP1-driven Rap1 overexpression (Rap1 OE) (p < .02). (c) ChIP-qPCR of Rap1 in WT and asf1Δ strains with Rap1 OE driven by GAL1. Rap1 localization to promoters of activated NRTS is not affected by ASF1 deletion (p values insignificant). ChIP signals are normalized to noninduced cells. (d) ChIP-qPCR of histone H3 in WT and asf1Δ strains with Rap1 OE driven by GAL1. Histone displacement is diminished in asf1Δ strains (p < .05). ChIP signals are normalized to noninduced cells. (e) NRTS activation by Rap1 OE is blunted upon deletion of members of the HIR complex (p < .02). (f) asf1Δ does not affect the rate of senescence. Senescence assay with WT (n = 2), asf1Δ (n = 2), tlc1Δ (n = 5), and tlc1Δ asf1Δ (n = 5). (g) hir1Δ does not affect the rate of senescence. Senescence assay with WT (n = 2), hir1Δ (n = 2), tlc1Δ (n = 5), and tlc1Δ hir1Δ (n = 5). All error bars indicate the standard error of mean sequential binding of other factors, possibly through opening up local chromatin. Well-known pTFs such as FoxA have DNA binding domains consisting of helix-turn-helix motifs flanked by "wings" of polypeptides, allowing the motif to bind alongside one side of DNA without interfering with the binding of histones on the other side (Soufi et al., 2015;Zaret & Carroll, 2011). Such a DBD secondary structure and its orientation on DNA are very similar to those observed in Rap1. In addition, FoxA has a C-terminal domain that interacts directly with core histones H3 and H4 (Cirillo et al., 2002).
Interestingly, while the C-terminal region of Rap1 also contributes to histone binding abilities, the Rap1 DBD alone is able to open local chromatin (Yu et al., 2001), consistent with our observations that direct histone interactions in the SANT domain, via amino acids SHY, are important for Rap1's functions as a pTF.

| Gene expression changes can be uncoupled from the rate of senescence
Previously, we reported several functions of Rap1 at senescence: first, it represses histone gene expression and contributes to global downregulation of histones; second, it contributes to local nucleosome losses at promoters of upregulated NRTS (Platt et al., 2013); third, Rap1 drives the rate of senescence, probably through regulation of histone dynamics, as diminishment of Rap1 levels via destabilization of the Rap1 mRNA (DAmP allele) or artificial overexpression of core histones can delay the rate of senescence.
Here, we explore mechanisms of Rap1-mediated local nucleosome losses. We characterize an amino acid patch-residues SHY in the SANT domain-that contains one or more residues required for direct binding of Rap1 to histones. Mutation of SHY results in deficient nucleosome clearance at NRTS promoters and subsequently reduced activation of NRTS. However, RAP1 SHY affects neither the rate of senescence nor histone gene repression. This is consistent with our previous finding that histone gene repression at senescence does not involve nucleosome losses from the histone gene promoters (Platt et al., 2013).
Similarly, we found Rap1-mediated gene expression changes at senescence require the histone H3/H4 chaperone Asf1 and the HIR complex. Deletion of ASF1 results in blunted NRTS upregulation at senescence. Similar loss of activation was seen when ASF1 or genes encoding members of the HIR complex were deleted under settings of Rap1 overexpression, which we previously found was sufficient for selective NRTS upregulation. However, much like the RAP1 SHY mutation, despite causing substantial losses in NRTS activation, deletion of ASF1 or HIR1 also do not affect the rate of senescence.
This begs the question of how gene expression changes-in particular, wide-spread gene upregulation that has been observed in multiple senescent models-relate to the rate of senescence. As   Georgilis et al., 2018;Nacarelli et al., 2019;Tasdemir et al., 2016).
This is encouraging, as it indicates that negative aspects of cell senescence can be blocked without compromising its tumor-suppressive properties.

| Yeast strains and plasmids
All experiments are performed using BY4741/4742 background, and deletion strains are from the haploid yeast knockout collection or were constructed using standard gene replacement techniques.
Plasmids were made using Gateway cloning methods. Site-directed mutagenesis to generate Rap1 Escherichia coli expression plasmids with AAA mutations in the SANT domain was performed using QuickChange primer design (Agilent) and primer extension using Phusion HF to introduce the changes into pGST-SANT-6xHis (BSS48); all mutations were verified by sequencing. All strains and plasmids used are listed in Table S1, and primers used for mutagenesis are listed in Table S2.

| Expression and purification of Rap1 and Rap1 derivatives
All Rap1 proteins were N-terminally tagged with GST, and some were also C-terminally tagged with a 6x-His tag, as indicated in the text. Novex with the following modifications. Cell pellets were resuspended in 20 ml of native binding buffer (50 mM NaH 2 PO 4 , pH 8.0, 500 mM NaCl, 1% Triton X-100, 1 mg/ml lysozyme, benzonase 1,000 Units). Protein lysate, treated with benzonase and clarified by centrifugation as recommended, was loaded onto Ni-NTA resin. After incubation and washing, to ensure removal of any contamination by residual DNA, an additional on-column benzonase digestion was performed (10 mM Tris-Cl, pH 8.0, 100 mM NaCl, 1 mM DTT, 1.5 mM MgCl 2 , benzonase 500 Units) at room temperature for 15 min.
High-sensitivity measurements of DNA by Qubit showed minimal amounts of DNA (40 Rap1 molecules for every base pair of DNA).
Protein was eluted using 10 ml elution buffer (50 mM NaH 2 PO 4 , pH 8.0, 500 mM NaCl, 250 mM imidazole). Protein of interest was quantified using SDS-PAGE gel electrophoresis and BSA standards and then flash-frozen for storage.
For GST-only protein purifications, 200 ml of BL21(DE3) cells containing the various Rap1 expression plasmids were grown and induced as described above. Cell pellets were flash-frozen and stored at −80°C. The GST-Rap1 was purified using Methods described in Schäfer, Seip, Maertens, Block, and Kubicek (2015) with modifications. Clarified protein lysate treated with benzonase was loaded onto 2.5 ml of glutathione resin (GE Healthcare). An on-column benzonase digestion as described above was performed to remove residual DNA. Protein concentration was determined by boiling 10 µl of resin in 2× SDS-PAGE buffer and SDS-PAGE gel electrophoresis using BSA as standard.

| GST histone pull-down assay
Protein purified by Ni-NTA were thawed and diluted using 1 volume of binding buffer (50 mM Tris-Cl, pH 7.5, 1 mM BME, and 150-750 mM NaCl, as indicated in text), and incubated overnight with equilibrated glutathione resin (50 µl). GST-6X-His proteins were used as negative controls. After supernatant is removed, resin was washed twice in 500 µl of binding buffer. Proteins purified using GST protein purifications were used directly. Histones were purified as described in Ricketts et al. (2015). The pull-down assay was performed by incubating 2 µM purified histones and desired con- qPCR was performed as described in Platt et al. (2013). Statistical analyses were performed using two-tailed unpaired t-tests.

| Quantitation of mRNA analysis
mRNA expressions were quantified using methods described in Platt et al. (2013). All mRNA analyses for Rap1 or mutant overexpression were performed using protein expression driven by the NOP1 promoter. Signals were calculated using standard curves of pooled cDNA samples and normalized to ACT1. Error bars indicate standard error of mean. P-values were calculated using two-tailed unpaired t-tests.

| Electrophoretic mobility shift assays
Proteins were purified using Ni-NTA and GST resin and quantified using Coomassie blue staining with BSA standards. DNA probes were generated using polynucleotide kinase to 32 P-end label oligos, followed by annealing to their unlabeled complementary strands (Table S6). The fraction of active protein was similar for Rap1 and Rap1 SHY preparations (~85%), and was quantified by incubating 10 nM protein as measured by Coomassie blue staining with increasing amounts of TEF2 probe (0-100 nM), taking the fraction of protein-DNA complex formation at saturated DNA concentrations as a measure of active protein. For EMSAs, 0.5 nM 32 P-labeled duplexes were incubated with increasing concentrations of active protein in binding buffer (20 mM Hepes-KOH, pH 8, 100 mM KCl, 10 µg/ml BSA, 1 mM EDTA, 2 mM MgCl 2 , 5% glycerol) for 30 min at room temperature. Reactions were loaded on 6% DNA retardation gel (Invitrogen) and electrophoresis was conducted at 100V at 4°C. Radioactive signals were visualized using Typhoon FLA 7000.

| Integration of SHY to AAA mutation in the RAP1 locus
Genome editing was performed using the 50:50 method for PCRbased seamless genome editing in yeast (Horecka & Davis, 2014).
Forward and reverse primers encompassing SHY→AAA mutation and homologous to URA3 were used for amplification of URA3 from pRS306 by PCR. As RAP1 is an essential gene in S. cerevisiae, the mutation was introduced into diploid cells (TLC1/tlc1Δ), which were then sporulated and dissected. Haploids containing the mutation were confirmed via sequencing.

| Southern blotting
Telomere lengths were determined as described (Johnson et al., 2001), using XhoI digested DNA run on a 1% agarose gel, transferred to a Hybond-XL membrane, and probed using a radio-labeled telomere Y' fragment.

| Senescence assays
Senescence assays were performed as described in Platt et al. (2013).
In short, cells from the Yeast Knockout Library were mated with early generation tlc1Δ::LEU2 and the diploids were grown for 60 doublings to allow for equilibration of telomere lengths. Diploids were sporulated, dissected, and genotyped. All comparisons between different genotypes were derived from the same tetrad heterozygous from tlc1Δ deletion and other mutations of interest (e.g., RAP1/RAP1 SHY ) to ensure inheritance of similar telomere length. Spore products were grown in YPAD liquid media and passaged every 22 hr. For each passage, cells were counted using a Coulter counter and diluted to 10 6 cells/ml in 5 ml of liquid media. Cell counts were used to determine population doublings (PD), and the point of senescence was determined from the PD displaying the lowest level of growth. Cells for mRNA expression at senescence were obtained ~5 PDs prior to the nadir to avoid formation of survivors, and grown 2-3 more doublings to a density of 1 × 10 7 cells/ml in fresh medium before harvest.

ACK N OWLED G M ENTS
We would like to thank members of the Epigenetics and Aging P01