An interplay between multiple sirtuins promotes completion of DNA replication in cells with short telomeres

The evolutionarily-conserved sirtuin family of histone deacetylases regulates a multitude of DNA-associated processes. A recent genome-wide screen conducted in the yeast Saccharomyces cerevisiae identified Yku70/80, which regulate nonhomologous end-joining (NHEJ) and telomere structure, as being essential for cell proliferation in the presence of the pan-sirtuin inhibitor nicotinamide (NAM). Here, we show that sirtuin-dependent deacetylation of both histone H3 lysine 56 and H4 lysine 16 promotes growth of yku70Δ and yku80Δ cells, and that the NAM sensitivity of these mutants is not caused by defects in DNA double-strand break repair by NHEJ, but rather by their inability to maintain normal telomere length. Indeed, our results indicate that in the absence of sirtuin activity, cells with abnormally short telomeres, e.g., yku70/80Δ or est1/2Δ mutants, present striking defects in S phase progression. Our data further suggest that early firing of replication origins at short telomeres compromises the cellular response to NAM- and genotoxin-induced replicative stress. Finally, we show that reducing H4K16ac in yku70Δ cells limits activation of the DNA damage checkpoint kinase Rad53 in response to replicative stress, which promotes usage of translesion synthesis and S phase progression. Our results reveal a novel interplay between sirtuin-mediated regulation of chromatin structure and telomere-regulating factors in promoting timely completion of S phase upon replicative stress.


Introduction
Histone post-translational modifications influence chromatin structure and serve as recruitment platforms for diverse protein complexes [1]. Acetylation of histones on lysine residues is catalysed by histone acetyltransferases (HAT) and reversed by histone deacetylases (HDAC). Four HDAC classes are defined based on sequence identity and catalytic mechanism [2]. Class III HDACs are referred to as sirtuins because of their sequence homology to yeast Sir2. These enzymes deacetylate lysine residues in histone and non-histone proteins in a reaction that requires nicotinamide adenine dinucleotide (NAD + ) and releases nicotinamide and O-acetyl ADP ribose [3,4]. Sirtuins are evolutionarily conserved, and regulate several DNA-associated processes including gene silencing, DNA replication, and DNA repair [5].
The kinases Mec1 and Rad53 are activated during replicative stress to phosphorylate multiple substrates which cooperate to inhibit DNA replication origin activation, stabilize stalled replication forks, and increase dNTP pools [23]. Rad53 activation depends on the mediator protein Rad9, which is recruited to chromatin through interaction with phosphorylated serine 128 of histone H2A (γ-H2AX, a DNA damage-induced modification) and methylated histone H3 lysine 79 (H3K79me) [24][25][26][27], the latter being catalyzed by the methyltransferase Dot1 [28,29]. Recent data demonstrate that cells have evolved mechanisms that limit Rad53 activation upon replicative stress, as well as others that permit its progressive inactivation upon DNA lesion resolution [30][31][32]. While the precise consequences of Rad53 "hyperactivation" are incompletely characterized, its biological relevance is highlighted by the fact that it causes sensitivity to replicative stress-inducing drugs [31]. Interestingly, we and others have shown that limiting Rad53 activation via DOT1 deletion or histone gene mutations that inhibit H3K79me promotes resistance to DNA replication-blocking drugs in several yeast mutants, including hst3Δ hst4Δ cells, by elevating usage of error-prone translesion synthesis (TLS) [22,[33][34][35]. These observations emphasize the importance of the interplay between chromatin and DNA damage checkpoint signalling in regulating the cellular response to replicative stress.
In eukaryotes, DNA replication is initiated in a temporally ordered manner at genomic regions called "origins" that are activated in early, mid, or late S phase [36]. Genomic context influences the timing of origin activation (or "firing"); for example, in the yeast S. cerevisiae origins located near telomeres and within rDNA repeats are activated during late S, while those close to centromeres fire earlier [37][38][39][40]. Interestingly, the silent information regulator (SIR) HDAC complex, which comprises the Sir2-Sir3-Sir4 subunits and deacetylates telomeric/subtelomeric chromatin, has been shown to prevent early firing of telomeric origins [41]. Telomere length also influences origin activity; indeed, cells with short telomeres, such as those lacking the Yku70/80 complex, initiate DNA replication at telomeric and subtelomeric regions abnormally early during S phase [42][43][44]. While several telomere-associated factors have been shown to influence the timing of telomeric DNA replication origin [43][44][45][46], the functional significance of such regulation is poorly understood.
The Yku70/80 complex is present at chromosomal ends where it protects telomeres from nucleolytic degradation and promotes recruitment of telomerase. Cells lacking Yku70/80 heterodimers present short but stable telomeres harboring abnormally long stretches of ssDNA [47][48][49][50][51]. Yku70/80 is also involved in DNA double-strand break (DSB) repair by non-homologous end joining (NHEJ). This complex binds DSB ends where it recruits the DNA ligase machinery composed of Lif1-Dnl4 and Nej1, thereby promoting end ligation [47,52,53]. Interestingly, cells lacking Yku70/80 are sensitive to genotoxins that generate DNA replicationblocking lesions without directly causing DSBs, suggesting an NHEJ-independent role for this complex during replicative stress [54,55]. However, the extent to which the other cellular functions of Yku70/80, e.g., at telomeres, might influence the cellular response to DNA replication stress is unclear.
A genetic screen conducted by our group in S. cerevisiae identified yku70Δ and yku80Δ mutants as sensitive to pharmacological inhibition of sirtuin HDACs by nicotinamide (NAM) [35]. Since NAM causes replicative stress [35], we originally postulated that the Yku70/80 complex might influence DNA replication progression in the absence of sirtuin activity. Here, we reveal that a novel interplay between multiple sirtuins promotes completion of DNA replication in yku70Δ and yku80Δ cells, and that telomere shortening is the root cause of the sensitivity of these mutants to NAM-induced sirtuin inhibition. Our data further indicate that misregulation of replication origin firing at short telomeres, as well as modulation of DNA damage checkpoint kinase activity by chromatin structure, influence the resistance to NAMand genotoxin-induced replicative stress in cells with short telomeres.

An interplay between multiple sirtuins permits growth of cells lacking Yku70/80
S. cerevisiae yku70Δ and yku80Δ mutants are sensitive to NAM [35], a pan-sirtuin inhibitor [3,56]. To identify which among the five yeast sirtuins (Sir2, Hst1-4) are responsible for this phenomenon, single deletions of each sirtuin gene were combined with yku70Δ by mating, and double mutants isolated via tetrad dissection (S1A Fig). None of the double mutants displayed noticeable growth defects, suggesting that NAM-induced growth inhibition in yku70Δ mutants is likely due to concurrent inhibition of multiple sirtuins. We previously showed that deletion of the H3K56ac acetyltransferase RTT109 rescues the sensitivity of yku70Δ and yku80Δ mutants to NAM [35]. We therefore tested whether this reflects NAM-induced inhibition of the H3K56ac-deacetylases Hst3 and Hst4 [18]. Consistently, we found that yku70Δ hst3Δ hst4Δ cells displayed moderate but significant decrease in growth rate and doubling time compared to hst3Δ hst4Δ cells (Fig 1A and 1B), and that the H3K56A mutation, which prevents H3K56ac, allows proliferation of yku70Δ cells in NAM (S1B and S1C Fig). Nevertheless, growth of the yku70Δ hst3Δ hst4Δ mutant could still be significantly exacerbated by exposure to NAM (S2A Fig). Thus, while H3K56 hyperacetylation is an essential component of the NAM sensitivity of yku70Δ mutants, inhibition of sirtuins other than Hst3/4 also probably contribute to this phenomenon.
Mutations inhibiting H4K16 acetylation (H4K16ac), the levels of which are regulated by Sir2 and Hst1 in vivo [57,58], partially rescue certain phenotypes of hst3Δ hst4Δ cells [22]. Interestingly, deletion of SAS2, a gene encoding the catalytic subunit of the H4K16 acetyltransferase complex SAS-I [57,59,60], or mutation of H4K16 to alanine (H4K16A), rescued growth of yku70Δ cells in NAM (Fig 1C and 1D). We could not directly test whether reduced Sir2 activity exacerbates the growth defects of yku70Δ hst3Δ hst4Δ mutants since sir2Δ causes synthetic lethality when combined with hst3Δ hst4Δ [22,61]. Sir2 is recruited to rDNA repeats and HMR/HML/telomeres as part of either the RENT (Sir2/Cdc14/Net1) or SIR (Sir2/Sir3/Sir4) complexes, respectively [62][63][64]. We could not evaluate the impact of RENT subunit-encoding CDC14 or NET1 genes on hst3Δ hst4Δ cells since their deletion causes lethality. rDNA silencing defects arising from lack of Sir2 activity in the context of the RENT complex can be rescued by deletion of FOB1, which encodes a component of the rDNA replication fork barrier [10,65,66]. We found that deletion of FOB1 did not rescue the sensitivity of yku70Δ cells to NAM (S1D Fig); moreover, combining hst3Δ hst4Δ yku70Δ with either sir3Δ or sir4Δ did not cause synthetic growth defects (S1E Fig). Overall these results suggest that growth of yku70Δ cells depends on Hst3/4 and either i) on sirtuins other than Sir2, or ii) on Sir2-dependent processes that are not associated with the RENT or SIR complexes.
Published data indicate that Sir2 interacts with Sum1 to promote transcriptional silencing in the absence of Hst1 [67]. While deletion of SIR2 did not confer increased MMS sensitivity in yku70Δ mutants, the sir2Δ hst1Δ yku70Δ triple mutant was strongly sensitized to MMS compared to control double mutants ( Fig 1F, S2B Fig). Moreover, the MMS sensitivity of sir2Δ hst1Δ yku70Δ was similar to that of sum1Δ yku70Δ, and was not further increased in sir2Δ hst1Δ sum1Δ yku70Δ cells (Fig 1F). We also note that sir2Δ hst1Δ yku70Δ mutants remain sensitive to NAM (S2C Fig), consistent with the notion that concurrent inhibition of multiple sirtuins, i.e., Sir2, Hst1, Hst3 and Hst4, causes the sensitivity of yku70Δ cells to this agent.
We next tested whether preventing H4K16 acetylation alleviates other phenotypes of cells lacking Yku70/80. In addition to their MMS sensitivity, yku70Δ mutants exhibit growth and DNA replication defects at elevated temperatures [68]. We found that the H4K16A and sas2Δ mutations rescued the MMS sensitivity of yku70Δ and yku70Δ sum1Δ mutants (Fig 1G and  1H, S2D Fig). The temperature sensitivity of yku70Δ cells could also be suppressed by mutations that reduce H4K16ac levels ( Fig 1G and 1H, S2D Fig), and was exacerbated by sum1Δ in a H4K16ac-dependent manner (S2D Fig). Overall, our data indicate that the inability of yku70Δ mutants to grow in the presence of NAM results in part from lack of Sum1-Hst1/Sir2 activity, which leads to misregulation of H4K16ac levels and consequent sensitization to replicative stress caused by constitutive H3K56ac.

Short telomeres sensitize cells to NAM-induced sirtuin inhibition
The Yku70/80 complex is required for both DNA repair by non-homologous end joining (NHEJ) and telomere maintenance [47,48]. NHEJ-abolishing mutations (lif1Δ, nej1Δ and dnl4Δ) did not cause notable growth defects in NAM (Fig 2A), indicating that the sensitivity of yku70Δ/80Δ mutants to this chemical is unlikely to result from defective DSB repair by NHEJ. Cells lacking Yku70/80 present very short, but stable, telomeres [48,49]. Pre-senescent haploid cells lacking the telomerase subunit-encoding genes EST1 and EST2 also present very short telomeres [47,69], and are as sensitive to NAM as yku70Δ cells (Fig 2B), suggesting that reduced telomere length might cause NAM sensitivity. On the other hand, the Yku70/80 complex promotes telomerase intracellular trafficking and recruitment to telomeres by binding to the TLC1 telomerase RNA [51, [70][71][72][73], raising the possibility that defective recruitment of telomerase to telomeres, and not telomere length per se, might influence NAM sensitivity. Contrary to this notion, the yku80-135i mutation, which eliminates the TLC1-binding functions of Yku80, while only slightly reducing telomere length, did not sensitize cells to NAM (S3A and S3B Fig).
To further investigate the impact of telomere length on NAM sensitivity, we performed a time course experiment using a strain expressing an auxin-inducible degron (AID)-tagged YKU70 allele [74,75]. Auxin addition to the growth medium provoked rapid (within one hour) degradation of Yku70 and progressive telomere shortening over several days of cell growth, while auxin removal allowed rapid Yku70 re-expression and progressive telomere length recovery ( Fig 2C and 2D, S4 Fig). We reasoned that if Yku70 activity/presence within the cell is important for NAM resistance, sensitivity to this agent should increase within hours of auxin treatment. In contrast, if telomere length homeostasis promotes NAM resistance, progressive decrease in telomere length caused by Yku70 depletion should lead to a concomitant increase in NAM sensitivity over several days of growth. At every time point analyzed, we tested the capacity of cells to grow in NAM with or without auxin, i.e., with or without Yku70 re-expression during NAM exposure ( Fig 2E). Strikingly, NAM sensitivity correlated well with overall telomere length of the cell population, and re-expression of Yku70 during the growth assay (by omitting auxin in NAM-containing medium) did not reverse this trend. Further supporting the notion that telomere length, and not presence of Yku70/80 per se, is important to promote survival in NAM, expression of a Cdc13-Est1 chimera that bypasses Yku70/80 in telomerase-mediated telomere elongation [51,76] rescued growth of yku80Δ mutants in NAM (

Cells with short telomeres present Tel1-dependent DNA replication defects upon NAM exposure
Intriguingly, deletion of TEL1, despite rendering telomeres as short as those of yku70Δ or est1/ 2Δ mutants [77,78], did not provoke growth defects in NAM, and even rescued the NAM sensitivity of yku70Δ cells (Fig 3A). Telomeres of yku70Δ mutants, but not tel1Δ, are resected by the nuclease Exo1 and therefore accumulate ssDNA [49,79]. However, deleting EXO1 did not eliminate the growth defects of yku70Δ mutants in NAM (S3C Fig), indicating that elevated ssDNA at telomeres cannot explain the differential sensitivity of yku70Δ vs tel1Δ cells to NAM. In contrast to yku70Δ and yku80Δ mutants, cells lacking Tel1 activate their telomeric replication origins in late S; moreover, deletion of TEL1 restores late firing of telomeric origins in  . To explore the possibility that Tel1-dependent activation of telomeric origins in early S might influence the response to NAM-induced replicative stress in yku70Δ mutants, we first evaluated the impact of Tel1 on S phase progression in yku70Δ cells. Strikingly, NAM-treated yku70Δ mutants accumulated in early-mid S in a Tel1-dependent manner, indicating that Tel1 influences global dynamics of DNA replication in yku70Δ cells experiencing replicative stress (Fig 3B). Pre-senescent telomerase mutants (est1Δ and est2Δ) also presented NAM-induced S phase progression defects which became worse over cell generations (Fig 3C), strongly suggesting a link between progressive shortening of telomeres and compromised DNA replication. Interestingly, we did not detect significant increase in the frequency of Rad52-YFP and Rfa1-YFP foci in yku70Δ cells exposed to NAM compared to WT, indicating that replication defects in cells with short telomeres are unlikely to result from elevated induction of DNA lesions in these conditions (Fig 3D and 3E). We note that replication proceeds slowly in telomeric regions even in the absence of exogenous DNA damage, presumably because non-histone protein complexes impede replication fork (RF) progression at these loci [80]. Indeed, cells devoid of the Rrm3 helicase, which promotes DNA replication across genomic regions harboring chromatin-bound protein complexes, display a 10-fold increases in the number of stalled RFs at telomeres [80]. We found that deletion of RRM3 caused synthetic MMS sensitivity when combined with yku70Δ (Fig 3F), suggesting that an abnormal abundance of stalled RFs at telomeres might contribute to the phenotypes of cells lacking Yku70/80.
Our data indicate genome-wide reduction in replication progression in NAM-treated yku70Δ cells vs WT (Fig 3C), which cannot only reflect abnormal RF progression at telomeric/ subtelomeric regions since they represent a minor proportion of a cell's total DNA. Instead, the Tel1-dependent NAM sensitivity of yku70Δ mutants suggests that abnormal activation of telomeric origins in early S influence DNA replication dynamics genome-wide. Elevated origin activation at repetitive loci harboring replication origins, e.g., rDNA repeats in sir2Δ mutants, diminishes origin activity at unlinked loci by titrating replication initiation factors that are in limiting abundance [81,82]. Lack of YKU70 causes misregulation of a significant number of origins: yeast cells possess at least one origin per telomere (32 per haploid cell), and origin activation repression extends up to 40 kb inward from chromosome ends [44]. We hypothesized that origin activation in early S at telomeric regions, combined with that occurring at canonical early replicating loci, might i) generate an overwhelming number of stalled RFs in early S in response to NAM-induced replicative stress, leading to ii) sequestration of replication factors. This might compromise activation of origins throughout chromosomes and overall replication progression in mid/late S, and prevent rescue of stalled RF by converging forks. We first tested whether increasing the number of replication origins competing for limiting pools of factors sensitizes otherwise WT cells to replicative stress. Consistently, cells harboring 200-400 copies of a YEPFAT7.5 2μ plasmid [83], which depends on endogenous DNA replication factors for its propagation, were noticeably more sensitive to MMS and NAM than WT cells (S7A and S7B Fig). Conversely, we reasoned that increasing the availability of limiting DNA replication initiation factors might rescue replicative stress-induced growth defects in cells lacking Yku70/ 80. Cdc45, Sld3 and Sld7 (45/3/7) are required to trigger activation of licensed replication origins; moreover, these factors are limiting in abundance and their overexpression rescues genome-wide anomalies in DNA replication dynamics caused by early S activation of rDNA origins in sir2Δ mutants [81,84]. We found that 45/3/7 overexpression partially rescued growth of yku70Δ mutants in response to MMS or NAM (Fig 3G and 3H), as well as the strong synthetic sensitivity of rad52Δ yku70Δ mutants to MMS (Fig 3I). These results suggest that increasing the availability of limiting replication factors can mitigate DNA replication defects caused by short telomeres in yku70Δ cells.

H4K16ac influences the sensitivity to replicative stress of cells with short telomeres by modulating DNA damage-induced signaling
As was the case for tel1Δ, the H4K16A mutation rescued NAM-induced S-phase progression defects in yku70Δ cells (Fig 4A and 4B). The effect of sas2Δ was partial, which may reflect the fact that this mutation does not completely abolish H4K16ac, in contrast to H4K16A [22]. We first explored the possibility that mutations abolishing H4K16ac might suppress replication-associated phenotypes of yku70Δ mutants by reversing early activation of telomeric and subtelomeric origins. We synchronized cells in G1 using alpha factor and released them in early S in the presence of hydroxyurea (HU) and BrdU for 90 minutes, followed by BrdU immunoprecipitation and quantitative PCR to monitor dNTP incorporation at origins. We note that the amplitude of signals obtained by this technique has been shown to correlate well with origin activation/efficiency in early S phase [85][86][87][88]. We examined BrdU incorporation at two telomeric origins, ARS102 and ARS610, and one sub-telomeric origin, ARS522. As expected, these origins were more active when telomeres are shortened in the yku70Δ mutant, but not in tel1Δ cells (Fig 4C and 4D). We found that sas2Δ did not significantly influence the activity of telomeric/subtelomeric replication origins, alone or when combined with YKU70 deletion (Fig 4C and 4D). Importantly, neither sas2Δ or H4K16A significantly modulated telomere length (Fig 4E), indicating that H4K16ac impedes S phase progression of cells with short telomeres via other mechanisms.
We previously showed that mutations which prevent H4K16ac cause a reduction in tri-methylated H3K79 (H3K79me3) levels [22], an effect that we also observed in yku70Δ mutants ( Fig  5A). Interestingly, we found that abolishing H3K79 methylation via deletion of the histone methyltransferase-encoding gene DOT1 (Fig 5A) significantly rescued growth and S phase progression defects in yku70Δ cells exposed to NAM (Fig 4B, Fig 5B). This suggests that H4K16ac might influence the NAM sensitivity of yku70Δ cells at least in part by modulating H3K79 methylation levels. While the biological consequences and molecular mechanisms of co-reduction in H4K16ac and H3K79me3 are uncharacterized, we originally speculated that modulation of H4K16ac might influence the DNA damage response (DDR) by influencing the recruitment of the H3K79me3-binding DDR protein Rad9 to damaged chromatin and subsequent Rad53 activation [24][25][26][27]. We found that upon exposure to NAM, yku70Δ cells displayed increased Rad53 activation, which was strongly reduced by mutating either SAS2 or DOT1 (Fig 5C). Directly limiting the amplitude of DDR signaling, via RAD9 mutation or expression of a hypomorphic rad53-HA allele [33], rescued the growth of yku70Δ mutants in NAM (Fig 5D and 5E). rad9Δ also rescued DNA replication progression in yku70Δ cells upon NAM exposure (Fig 5F). Conversely, deletion of either PPH3 or SLX4, which cripples cellular pathways that act to limit Rad53 activation upon replicative stress [30,31], caused synthetic sensitivity to MMS when combined with yku70Δ, an effect which was found to depend on the H4K16ac acetyltransferase Sas2 (Fig  5G and 5H). Taken together, our results indicate that i) co-dependent H4K16ac and H3K79me3 promote DDR signalling in yku70Δ mutants, and ii) elevated Rad53 activity contributes significantly to the sensitivity of cells lacking Yku70 to MMS and NAM-induced replicative stress.

H4K16ac and H3K79me control translesion synthesis usage in yku70Δ mutants
Rad53 activation is known to inhibit DNA replication origin firing, thereby delaying S phase progression upon replicative stress [89,90]. This effect can be bypassed by preventing deletion does not prevent early activation of telomeric origins in yku70Δ cells. Cells were synchronized in G1 and released toward S phase in the presence of 200 mM hydroxyurea. 30 minutes before release, 400 ug/mL BrdU was added to cultures. Sonicated BrdU-labelled DNA was immunoprecipitated and recovered material from telomeric/subtelomeric origins was quantified by qPCR as described in materials and methods. Error bars: Standard error of the mean. (E) H4K16ac levels do not significantly influence telomere length. Telomere length was analysed by southern blotting using a probe that recognizes the telomeric TG 1-3 repeats. https://doi.org/10.1371/journal.pgen.1007356.g004 Sirtuins promote DNA replication in cells with short telomeres

Fig 5. Elevated DDR signalling is deleterious for growth of yku70Δ cells upon NAM-and MMS-induced replicative stress. (A)
Mutations that abolish H4K16ac cause a reduction in H3K79me3 levels. Protein samples from asynchronous cells were immunoblotted with indicated antibodies. Bar graph represents the ratio of H3K79me3 onto H4 signals as quantified by densitometry. Error bars: standard error of the mean (B) Deletion of DOT1 rescues growth of yku70Δ mutants in NAM (C) sas2Δ and dot1Δ mutations reduce Rad53 activation in yku70Δ cells. Cells were exposed to 20 mM NAM for 8 hours at 30˚C and samples were taken for Rad53 in situ autophosphorylation assays (see materials and methods for details). (D-E) rad9Δ (D) and RAD53-HA (E), which limit DDR signalling in response to replicative stress, rescue growth of yku70Δ mutants in NAM. Growth assay in 96-well plates (see materials and methods). Error bars: standard deviation. (F) Deletion of RAD9 rescues S phase progression defects of yku70Δ cells exposed to NAM. Asynchronous cells were incubated in YPD for 8 hours at 30˚C in the presence of 20 mM NAM.
Rad53-dependent phosphorylation of Dbf4 and Sld3, which are two key proteins of the origin activation cascade [90]. We found that yku70Δ yeast strains expressing non-phosphorylable alleles of Dbf4 and Sld3 are as sensitive to NAM as control strains (S8 Fig), suggesting that other consequences of Rad53 hyperactivation influence the ability of cells lacking Yku70/80 to proliferate in response to replicative stress. Limiting the activation of Rad53, by deleting DOT1 or expressing a hypomorphic rad53-HA allele, increases resistance to MMS-induced replicative stress by elevating lesion bypass via the translesion synthesis (TLS) pathway [33,34]. Our data indicate that the sensitivity to MMS of cells lacking Yku70 complex is exacerbated by mutating REV3, which encodes the catalytic subunit of TLS polymerase zeta required for the bypass of MMS-induced lesions (Fig 6A and 6B). In addition, we found that the rescue of the MMS sensitivity of yku70Δ by deletion of SAS2 or DOT1 depends on Rev3 (Fig 6A and 6B). TLS is intrinsically error-prone, and elevated usage on this pathway increases mutagenesis [91]. Concordantly, deletion of SAS2 or DOT1 in either WT or yku70Δ cells led to a statistically significant increase in MMS-induced CAN1 mutation frequency, an effect which was reverted by rev3Δ (Fig 6C, p-value < 0.05). Together, these results suggest that H4K16ac and H3K79me influence TLS-dependent bypass of replication blocking DNA lesions in cells lacking Yku70/ 80 by modulating DDR signalling.

Discussion
In the current study, we investigated the molecular basis of the sensitivity of cells lacking Yku70/80 to the pan-sirtuin inhibitor NAM. Our results reveal a novel interplay between telomere length and sirtuin-mediated regulation of chromatin structure in protecting cells against DNA replication stress. Indeed, we provide compelling evidence that in the absence of Yku70/ 80, telomere shortening, but not NHEJ deficiency nor other telomere-related defects, diminishes the ability of yeast cells to respond to replicative stress. Notably, we found that telomerase (est1/2Δ) mutants phenocopied yku70Δ both in terms of NAM-induced growth defects and S phase accumulation, and that manipulating telomere length via inducible Yku70 degradation or expression of a Cdc13-Est1 fusion modulates the sensitivity of cells to NAM. We note that sudden telomere shortening occurs naturally in vivo in WT yeast and is thought to result from replication fork collapse in telomeric tracts [92]. However, since such telomere shortening events are relatively infrequent and are expected to involve only one or a few telomeres in any given cell, the impact on DNA replication stress responses as described here is expected to be minor. This is in contrast with the situation for cells exhibiting uniform reduction in the length of all telomeres, such as telomerase and yku70Δ/yku80Δ mutants. Overall, our results are consistent with the fact that senescence caused by lack of telomerase activity is associated with induction of classical markers of the DNA damage response [93,94], and raise the interesting possibility that compromised responses to replicative stress might be a general phenomenon arising in mutants with short telomeres.
An important exception to the above involves the tel1ΔΔ mutation which, despite rendering telomeres extremely short, does not sensitize cells to NAM. Rather, we found that deletion of TEL1 rescued NAM-induced growth inhibition and DNA replication defects in yku70Δ cells. Previous reports established that telomere shortening caused by deletion of YKU70/80 provokes Tel1-dependent firing of telomeric origins in early S phase, as opposed to late S in WT cells [43][44][45][46]. Since RFs progress slowly within telomeric and subtelomeric regions [70,80,95], it is plausible that activation of telomeric origins in early S in yku70Δ cells, but not in yku70Δ Samples were taken at indicated time for flow cytometry-based DNA content analysis. (G-H) yku70Δ display synthetic sensitivity to MMS when combined with slx4Δ (G) and pph3Δ (H) mutants.
https://doi.org/10.1371/journal.pgen.1007356.g005 tel1Δ mutants, generates an overwhelming number of stalled RFs over a short period in the presence of replicative stress-inducing drugs. In addition, we provide evidence suggesting that RF stalling at telomeric regions in early S may compromise genome-wide replication dynamics by causing sequestration, and eventual exhaustion, of limiting replication factors, which in turn is expected to negatively impact further activation of origins later in S. We propose that such reduction in RF initiation events might prevent rescue of stalled RFs by converging forks in mid/late S, thereby contributing to the sensitivity of yku70Δ cells to replicative stress. We note however that while the above-described models rationalize our observations regarding the impact of Tel1 on the sensitivity of yku70Δ cells to NAM, we cannot exclude that consequences of telomere shortening other than misregulation of telomeric origins might also be implicated.
Consistent with the notion that cells with short telomeres exhibit an elevated number of stalled RFs during replicative stress, we found that the DDR kinase Rad53 is strongly activated upon NAM exposure in yku70Δ cells. Our data further indicate that such hyperactive DDR signalling contributes to the phenotypes of these mutants; indeed, we found that i) yku70Δ causes synthetic sensitivity to MMS when combined with either slx4Δ or pph3Δ, which are both known to limit Rad53 activity in response to DNA replication impediments [30][31][32], and ii) mutations that cripple Rad53 activation rescue the NAM sensitivity of yku70Δ cells. While the consequences of Rad53 hyperactivation during replicative stress are incompletely characterized, our published data and those of others [33,34] suggest that restricting DDR signalling improves cell survival and replication progression in response to genotoxins at least in part by promoting DNA damage tolerance via TLS. Overall, the results presented here are consistent with the above, and highlight the importance of mechanisms that dampen DDR signalling in promoting the survival of cells with short telomeres upon DNA replication stress.
In contrast to most other mutations causing NAM sensitivity that we analysed so far [35], deletion of YKU70 does not cause synthetic lethality when combined with hst3Δ hst4Δ, implying that other sirtuins are essential for survival of cells presenting short telomeres. Indeed, our genetic data support the notion that the redundant ability of Sir2 and Hst1 to deacetylate H4K16ac promotes resistance to replicative stress in cells lacking Yku70/80, whereas Hst3/ 4-dependent removal of H3K56ac mainly acts to limit the generation of endogenous DNA damage in this context. We note that even though the identity of the DNA lesions generated by constitutive H3K56ac is unknown, our results showing that the ability of Yku70/80 to promote NHEJ is not necessary for NAM resistance suggest that the predominant lesions caused by hyperacetylated H3K56 are unlikely to be DSBs. This is also consistent with the sensitivity of yku70Δ mutant to MMS, which produces few, if any, DSBs in yeast [55]. While other Sir2/ Hst1 targets might also contribute to this phenomenon, abolishing H4K16ac rescued the NAM, MMS, and temperature sensitivity of yku70Δ cells, indicating that this histone modification is a critical determinant of the phenotypes of these mutants. We note that the SIR complex was previously shown to suppress origin firing at telomeres [41] which, combined with our results, raised the possibility that this effect might depend on H4K16ac levels. However contrary to this idea, our data clearly indicate that reducing H4K16ac by deletion of the SAS2 acetyltransferase does not impact origin activity at short telomeres (Fig 4), consistent with a prior report indicating that telomeric origins in yku70Δ mutants are activated in early S independently of histone tail acetylation [44]. Our data highlight a novel role for H4K16ac regulation in limiting the activation of Rad53 upon replicative stress in yku70Δ mutants, presumably by modulating H3K79 trimethylation levels and Rad9 activity/recruitment to chromatin. This is noteworthy since H4K16ac is very abundant and is removed by sirtuins only at specific transcriptionally silent genomic loci, e.g., mating loci (HMR and HML), as well as at telomeric regions [29,57,96]. These regions present intrinsic impediments to DNA replication fork progression, often in the form of chromatin-bound protein complexes [80,[97][98][99]. In view of this, our results raise the intriguing possibility that cells may have evolved mechanisms to limit H4K16ac levels in these genomic regions in part to mitigate the deleterious consequences of unchecked DDR signalling arising from frequent RF stalling. We also note that senescent yeast cells have been shown to manifest reduced levels of Sir2, and consequently exhibit elevated H4K16ac at telomeric regions [100]. Moreover, SAS2 deletion was demonstrated to extend replicative life span [100]. It is tempting to speculate that misregulation of H4K16ac might contribute to certain phenotypes of senescent cells, e.g., elevated DNA damage, by promoting intense DDR signaling in response to spontaneous replicative stress arising at loci that are intrinsically difficult to replicate, including telomeric regions [101][102][103].

Yeast strains and growth conditions
Experiments were performed using standard yeast growth conditions. Yeast strains used in this study are listed in Table 1. To avoid frequent emergence of spontaneous suppressor mutations in cells with constitutive H3K56 hyperacetylation, hst3Δ hst4Δ strains used in this study were propagated with a URA3-harboring centromeric plasmid encoding Hst3. To evaluate the phenotypes caused by hst3Δ hst4Δ, cells were plated on 5-Fluoroorotic Acid (5-FOA)-containing medium immediately before experiments to select cells that spontaneously lost the plasmid, or during the experiment (spot assays on 5-FOA-containing plates). For experiments involving telomerase mutants (est1Δ or est2Δ), fresh haploid clones were obtained from tetrad dissection of heterozygous diploids to ensure that cells were not undergoing senescence during experiments. For spot assays, cells were grown to saturation in YEP with 2% glucose or 2% raffinose in a 96-well plate. Five-fold serial dilutions of these cultures with identical OD were then plated on indicated media and allowed to grow for 2 to 5 days. Growth assays in NAM were done as previously described [35]. Cells were diluted to OD 600 0.0005 in 100 μL of YPD with increasing NAM concentrations in a 96-well plate. OD 630 were acquired using a BioTek EL800 plate reader, and growth of each strain was normalized relative to an untreated control well. For doubling time assessments, cells were diluted to OD 600 0.01 in 100 μL of YPD in a 96-well plate and incubated at 30˚C in a BioTek EL808 plate reader for 48h. Every 30 minutes, plates were shaken for 30 seconds and OD 630 readings were acquired. Doubling times were derived from exponential regression of the resulting growth curve.

Telomere southern blot
Monitoring of telomere length by southern blotting was performed as described [107]. Briefly, genomic DNA was digested with XhoI (New England Biolabs) and run on a 1.2% agarose gel for 17 hrs in 1x TBE buffer. Telomeric repeats were detected with a TG 1-3 probe kindly provided by Dr Raymund Wellinger (Université de Sherbrooke).

DNA content analysis by flow cytometry
Cells were fixed in 70% ethanol, sonicated, treated with 0.4 ug/mL RNAse A in 50mM Tris-HCl pH 7.5 for 3 hours at 42˚C followed by treatment with 1mg/mL Proteinase K in 50mM Tris-HCl pH 7.5 for 30 minutes at 50˚C. DNA content was assessed by Sytox Green (Invitrogen) staining as previously described [109]. DNA content analysis was performed on a FACS Calibur flow cytometer equipped with Cell Quest software. Graphs were produced using FlowJo 7.6.5 (FlowJo, LLC).

Fluorescence microscopy
Cells expressing Rad52-YFP or Rfa1-YFP were fixed with formaldehyde as previously described [106,110] and stained with DAPI. Fluorescence was examined with a DeltaVision microscope equipped with SoftWorx version 6.2.0 software (GE Healthcare). Images were examined using a custom MATLAB script (version R2017a; MathWorks) to extract the number of cells with Rfa1-YFP or Rad52-YFP foci. Briefly, a mask was created based on DAPI signals to identify cell nuclei and count the number of cells within an image. A second mask was created with the YFP channel to mark foci by finding spots with elevated YFP fluorescence compared to surrounding regions. Nuclei with at least one focus were listed as cells with Rfa1-YFP foci.

in situ Rad53 autophosphorylation assays
Protein samples were prepared by trichloroacetic acid/glass beads lysis, separated on 10% acrylamide gels and transferred to a PVDF membrane. Autophosphorylation assays were carried out as previously described [111].

Auxin-induced degradation
Cells were maintained in logarithmic phase for the indicated number of days (see Fig 2) by dilution in fresh YPD ± 2 mM Auxin (3-indoleacetic acid, Sigma). For each time point, a growth assay was performed in YPD with increasing concentrations of NAM ± 2 mM Auxin. Growth was normalized to the untreated control.

Analysis of origin firing by BrdU immunoprecipitation
Cultures were synchronized in G1 with α-factor at 30˚C. 30 minutes prior to release, 400 ug/mL BrdU was added to cultures except for the control condition. Release was carried out by addition of 50 μg/mL pronase and 200 mM hydroxyurea and cells were incubated for 90 minutes at 30˚C. 0.1% sodium azide was added and cultures were incubated for 10 minutes on ice. Cells were centrifugated and pellets were washed once with TBS, transferred to screwcap tubes and frozen on dry ice. Immunoprecipitation (IP) was performed as previously described [112]. Quantitative PCR was done using 2x SYBR Green qPCR Master Mix (Bimake) per the manufacturer's guidelines. qPCR plates were analyzed on a ABI7500 real-time PCR system. Since replication at origins doubles the amount of DNA, normalizing on the input signal from probed origins might reduce BrdU signals from replicated regions, and interfere with quantification of changes in origin activity. To remove the contribution of replicated DNA from qPCRs quantifications, we instead normalized IP signals to a region that is expected to remain unreplicated throughout our experiments. We chose the ACT1 locus that is located 15 kb and 54kb away from the closest origin of replication (ARS603) and telomere, respectively. Replication forks are expected to travel 3-7.5kb away from origins in our experiments [113], and are therefore not expected to reach the ACT1 locus. Data from replicate experiments were further normalized to a highly efficient replication origin, ARS305, to account for differences in BrdU incorporation between strains. Data are represented as values relative to the WT strain. Primer pairs are listed in S1 Table. CAN1 mutagenesis assay 6 colonies of relevant genotype were grown to saturation for 48 hours in YPD containing 0.001% MMS. Cells were plated on YPD assess the number of colony forming units and on synthetic media containing 60 μg/mL canavanine to determine the number of CAN1 mutation events. The frequency of canavanine resistance was calculated as the ratio between colonies growing on canavanine plates and the initial number of plated viable cells (assessed by plating appropriate dilutions on YPD plates). The median CAN1 mutation frequency of the 6 clones was then determined and data is represented as the average of the median from several experiments. Statistical significance was determined using two-tailed student's T-tests.