Diverse roles in DNA metabolism have been envisaged for budding yeast and mammalian Rif1. In particular, yeast Rif1 is involved in telomere homeostasis, while its mammalian counterpart participates in the cellular response to DNA double‐strand breaks (DSBs). Here, we show that Saccharomyces cerevisiae Rif1 supports cell survival to DNA lesions in the absence of MRX or Sae2. Furthermore, it contributes to the nucleolytic processing (resection) of DSBs. This Rif1‐dependent control of DSB resection becomes important for DSB repair by homologous recombination when resection activities are suboptimal.
Rif1 promotes nucleolytic processing of DNA double‐strand breaks in S. cerevisiae via a pathway that is partially redundant with the one involving Sae2 and Exo1.
Saccharomyces cerevisiae Rif1 supports cell viability in the absence of the MRX complex.
Rif1 promotes resection of DNA double‐strand breaks.
Lack of Rad9 suppresses the resection defect of rif1∆ cells and the growth defect of rif1∆ mre11∆ cells.
Rif1 has been identified in Saccharomyces cerevisiae as a negative regulator of telomere length and transcriptional silencing , . Although Rif1 physically interacts with the telomeric proteins Rap1 and Rif2, these proteins regulate telomere metabolism by different mechanisms. In fact, Rap1 and Rif2 inhibit both nucleolytic processing and non‐homologous end joining (NHEJ) at telomeres, while Rif1 is not involved in these processes , , . Instead, Rif1 plays a unique role in supporting cells' viability ,  and in preventing nucleolytic degradation in situations where telomere protection is altered, such as in mutants affecting the CST (Cdc13‐Stn1‐Ten1) complex . On the other hand, both Rif1 and Rif2 prevent short telomeric ends from causing a checkpoint‐mediated cell cycle arrest by inhibiting the recruitment of the checkpoint proteins Rad9, Mec1 and Rad24 to these ends , .
Mammalian Rif1 is not part of the telomeric complex, while it is involved in the response to DNA double‐strand breaks (DSBs). DSBs can be repaired by homologous recombination (HR), which requires the formation of RPA‐coated single‐stranded DNA (ssDNA) that arises from 5′ to 3′ nucleolytic degradation (resection) of DNA ends . DSB resection in mammals is promoted by BRCA1, which forms a complex with CtIP and MRN (orthologs of S. cerevisiae Sae2 and MRX, respectively) . Rif1 has been recently shown to prevent DSB resection in G1 by blocking the accumulation of BRCA1 at the sites of damage , , , .
Whether budding yeast Rif1 functions exclusively at telomeres or it plays a role also at DSBs like its mammalian counterpart remains to be determined. Here, we show that S. cerevisiae Rif1 functions together with Sae2 and MRX in DSB repair by HR.
Results and Discussion
Rif1 supports cell viability in the absence of the MRX complex
In S. cerevisiae, the MRX (Mre11‐Rad50‐Xrs2) complex initiates DSB end resection by acting in concert with Sae2 . To investigate whether Rif1 is involved in the DNA damage response, we analysed the effects of its absence in cells either lacking Mre11 or Sae2 or carrying the nuclease‐defective mre11‐H125N allele. When meiotic tetrads from diploid strains heterozygous for the rif1∆ and mre11∆ alleles were analysed for spore viability on YEPD plates, all rif1∆ mre11∆ double‐mutant spores formed much smaller colonies than each single‐mutant spore (Fig 1A). The smaller colony size was due to the loss of viability, as rif1∆ mre11∆ spore clones contained a much lower number of colony‐forming units than each single‐mutant spore clone (Supplementary Fig S1). By contrast, RIF1 deletion did not significantly affect the size of the colonies formed by mre11‐H125N and reduced only slightly the size of the colonies of sae2∆ spores (Fig 1B,C).
It is known that mre11∆ cells suffer a more severe resection defect than sae2∆ or mre11‐H125N cells. In fact, the MRX complex, but not its nuclease activity, is required to recruit to DSBs the 5′–3′ exonuclease Exo1 that can substitute for MRX‐Sae2 nuclease function in resection . We then asked whether rif1∆ sae2∆ cells' viability depended on EXO1. When tetrads from the appropriate diploid were dissected on YEPD, all the rif1∆ sae2∆ exo1∆ triple‐mutant spores formed much smaller colonies than rif1∆ sae2∆ double‐mutant spores (Fig 1D). Thus, Rif1 appears to support cell viability when the MRX complex is not functional.
On the other hand, rif1∆ sae2∆ and rif1∆ mre11‐H125N double mutants were more sensitive to phleomycin (phleo) and methyl methanesulfonate (MMS) than each single mutant (Fig 1E,F). This Rif1 function reflects cooperation between Rif1 and Sae2‐MRX in repairing DNA lesions by HR. In fact, RIF1 deletion did not exacerbate the sensitivity to hydroxyurea (HU), phleomycin and camptothecin (CPT) of cells carrying the deletion of the HR genes RAD51 and RAD52 (Fig 1G), indicating that Rif1 acts in the same pathway of Rad51 and Rad52. This function in HR is specific for Rif1, as the lack of its interacting protein Rif2 did not affect the colony size of mre11∆ spores (Fig 1H) and did not exacerbate the sensitivity to DNA‐damaging agents of mre11∆ and sae2∆ mutants (Supplementary Fig S2A).
As the genetic interactions between Rif1 and MRX‐Sae2 might underlie a possible Rif1 involvement in DSB resection, we asked whether EXO1 overexpression suppressed the sick phenotype of rif1∆ mre11∆ cells and the hypersensitivity to DNA‐damaging agents of rif1∆ sae2∆ cells. A rif1∆ homozygous diploid strain that was heterozygous for the mre11∆ allele was then transformed with a 2 μ high copy number plasmid either empty or carrying the EXO1 gene or the nuclease‐defective exo1‐D171A allele. After sporulation and tetrad dissection of the transformed strains, all rif1∆ mre11∆ spores with high copy number EXO1 formed colonies of almost wild‐type size, whereas all rif1∆ mre11∆ spores carrying either the empty vector or high copy number exo1‐D171A allele still formed small colonies (Fig 1I,L). Furthermore, overexpression of EXO1, but not of the exo1‐D171A allele, partially suppressed the hypersensitivity to phleomycin and MMS of rif1∆ sae2∆ double mutants (Fig 1M). Altogether, these results suggest that Rif1 can cooperate with MRX‐Sae2 in DSB resection.
Rif1 promotes DSB resection
To further investigate the possible role of Rif1 in the generation of 3′‐ended ssDNA at a DSB, we deleted RIF1 in a haploid strain where a DSB can be generated at the MAT locus by inducing the expression of the HO endonuclease from a galactose‐inducible promoter . This strain cannot repair the HO‐induced DSB because the HML and HMR homologous donor sequences have been deleted. Because ssDNA is resistant to cleavage by restriction enzymes, we directly monitored the formation of ssDNA at the HO‐induced DSB by following the loss of SspI restriction sites by Southern blot analysis under alkaline conditions (Fig 2A). HO was induced by galactose addition to α‐factor or nocodazole‐arrested wild‐type and rif1∆ cells that were kept arrested in G1 or G2, respectively. To induce a persistent G1 arrest, all the strains carried the deletion of the BAR1 gene that encodes an α‐factor‐degrading protease. The quality and persistence of the cell cycle arrest was assessed by FACS analysis and by determining the percentage of unbudded and budded cells (Fig 2B). Although DSB resection occurs primarily in S/G2 when Clb‐Cdk1 activity is high , , a certain amount of ssDNA can be detected even in G1‐arrested wild‐type cells (Fig 2C,D). This finding is in agreement with previous studies that showed that resection in G1 is inhibited compared to G2, but not completely abolished , . In G1‐arrested cells, the HO‐cut DNA fragment was converted into the r2 resection product with similar kinetics in both wild‐type and rif1∆ (Fig 2C,D). However, the appearance of r3 resection product was delayed in rif1∆ cells compared to wild‐type (Fig 2C,D), indicating that the lack of Rif1 impairs resection beyond the SspI restriction site located 1.7 kb from the HO cutting site. By contrast, all the resection products accumulated with similar kinetics in G2‐arrested wild‐type and rif1∆ cells (Fig 2C,D).
As the long resection products are not easily detectable with the assay used above, we investigated the requirement of Rif1 in specific DNA repair pathways that are known to strictly rely on resection, in order to uncover a possible role of Rif1 in promoting extensive DSB resection in G2. Single‐strand annealing (SSA) is the main repair pathway of a DSB that is flanked by direct repeats and requires degradation of the 5′ DSB ends to reach the complementary DNA sequences that can then anneal (Fig 2E) . Subsequent nucleolytic removal of the protruding single‐stranded tails results in deletion of the intervening DNA sequence and one of the repeats (Fig 2E). We deleted RIF1 in YMV45 and YMV80 strains that carry tandem repeats of the LEU2 gene separated by 4.6 kb and 25 kb, respectively, with a recognition site for the HO endonuclease adjacent to one of the repeats (Supplementary Fig S3) . Because repair of this DSB can occur by either SSA or break‐induced replication (BIR), all the strains carried the RAD51 deletion, which abolishes BIR but does not affect SSA . HO was expressed by galactose addition to nocodazole‐arrested cells that were kept arrested in G2 with nocodazole. SSA‐mediated DSB repair showed equivalent efficiency in both wild‐type and rif1∆ strains carrying one of the flanking LEU2 repeats at 4.6 kb from the DSB (Fig 2F,G). However, when one of the flanking LEU2 repeats was located at 25 kb from the DSB, accumulation of the repair product was reduced in G2‐arrested rif1∆ cells compared to wild‐type (Fig 2F,G). These data indicate that the lack of Rif1 impairs extensive DSB resection even in G2, and this defect can reduce the efficiency of SSA‐mediated repair of a DSB between distant flanking homologous sequences.
We then asked whether Rif1‐mediated regulation of DSB resection in G2 was partially redundant with other resection activities. When a DSB occurs, MRX and Sae2 initiate resection of the 5′ strand, while extensive resection relies on two pathways, which depend on the nucleases Exo1 and Dna2, respectively, with the latter acting together with the helicase Sgs1 , . As shown in Fig 3, G2‐arrested rif1∆ sae2∆ (Fig 3A,B) and rif1∆ exo1∆ double mutants (Fig 3C,D) showed more severe resection defects than each corresponding single mutant. Furthermore, when EXO1 was deleted in rif1∆ YMV45 strain, G2‐arrested rif1∆ exo1∆ cells repaired the DSB by SSA less efficiently than rif1∆ and exo1∆ single mutants (Supplementary Fig S4).
Altogether, these results indicate that Rif1 promotes DSB resection not only in G1, but also in G2 by controlling a pathway that is partially redundant with the one involving Sae2 and Exo1. The epistatic relationships in resection between Rif1 and Sgs1/Dna2 could not be investigated because of the dramatic growth defects of rif1∆ sgs1∆ double‐mutant cells. Furthermore, the rif1∆ dna2∆ combination was synthetically lethal even when the essential function of Dna2 was bypassed by the pif1‐M2 mutation (Supplementary Fig S2B), which reduces the formation of long flaps that are substrates for Dna2.
Finally, we investigated whether Rif1 was recruited in the surroundings of the HO‐induced DSB. After HO induction by galactose addition, chromatin immunoprecipitation (ChIP) experiments detected Rif1 binding close to the cut site as early as 1 h after HO induction (Fig 4A), indicating that Rif1 is recruited to the DSB site. Consistent with a stronger effect of RIF1 deletion on DSB resection in G1 than in G2, Rif1 binding was higher in G1 than in G2 (Fig 4A).
The lack of Rad9 restores resection in G1‐arrested rif1∆ cells
We quantified by ChIP analysis the effect of the lack of Rif1 on the binding at the HO‐induced DSB of positive (Mre11, Exo1, Sgs1, Dna2, and Rpa1) and negative (Rad9) regulators of DSB resection , , , . Association of Mre11, Exo1, Sgs1, Dna2 and Rpa1 near the HO‐induced DSB was not impaired in exponentially growing rif1∆ compared to wild‐type cells (Fig 4B), which also showed similar amount of Exo1 bound to the DSB in G1 (Fig 4B). Thus, the resection defect of rif1∆ cells is not due to decreased association of these proteins with the DSB ends. Rather, Mre11 and Dna2 recruitment was even greater in rif1∆ cells than in wild‐type (Fig 4B). Mre11 association with the DSB was increased also in G1‐arrested rif1∆ cells compared to wild‐type (Fig 4B), whereas Dna2 association was very poor in both G1‐arrested wild‐type and rif1∆ cells, probably because Dna2 binds ssDNA, whose amount is reduced in G1. Interestingly, the amount of Rad9 bound near the DSB after HO induction was higher in both exponentially growing and G1‐arrested rif1∆ cells compared to wild‐type (Fig 4C). Detection of all the above proteins near the HO cut was not influenced by DSB resection, as all the ChIP signals were normalized for each time point to the corresponding input signal that decreased with similar kinetics in both wild‐type and rif1∆ cells at 1.8 kb from the DSB (Fig 4D).
After DNA damage, Rad9 binding to chromatin is promoted by interaction with histone H2A that has been phosphorylated at serine 129 (γH2A) by the Mec1 checkpoint kinase , , , . Formation of γH2A was still required in rif1∆ cells to promote Rad9 association with chromatin, as the substitution of H2A Ser129 with a non‐phosphorylatable alanine residue (hta1‐S129A) reduced Rad9 recruitment at the rif1∆ DSB to the extent observed in the hta1‐S129A single mutant (Fig 4E). The increased Rad9 binding to the DSB in rif1∆ cells might be due to an increased generation of γH2A at the damaged site by Mec1. Indeed, the binding of Mec1 (Fig 4F) and γH2A (Fig 4G) near the HO‐induced DSB was higher in rif1∆ cells than in wild‐type, suggesting that the increased Mec1 association at the DSB leads to more efficient γH2A generation, which in turn enhances Rad9 binding at DSBs.
Thus, the lack of Rif1 seems to increase the accessibility of some proteins to regions around the break site. Whether an excess of MRX and/or Dna2 association at the break site affects DSB resection is unknown. On the other hand, enhanced Rad9 association has been proposed to block resection in cells lacking the chromatin remodeler Fun30 , raising the possibility that robust Rad9 binding might be responsible for resection inhibition in rif1∆ cells. We therefore investigated whether RAD9 deletion was capable to suppress the resection defect of G1‐arrested rif1∆ cells. Consistent with a previous finding that the Rad9 inhibitory effect on DSB resection in G1 becomes apparent only in the absence of Yku, resection in G1‐arrested rad9∆ cells did not increase compared to wild‐type . However, RAD9 deletion abolished the resection defect of G1‐arrested rif1∆ cells (Fig 5A,B), indicating that Rad9 exerts its function in inhibiting DSB resection in rif1∆ G1 cells even in the presence of Ku. Rad9 binding at the DSB was increased not only in G1‐arrested but also in exponentially growing rif1∆ cells (Fig 4C), suggesting that this Rad9 excess in rif1∆ cells is sufficient to impair resection in G1, but not in G2, where DSB resection occurs much more efficiently than in G1. Notably, RAD9 deletion suppressed the growth defect of rif1∆ mre11∆ cells (Fig 5C), further supporting the hypothesis that their synthetic sickness is due to resection defects that impair DSB repair by HR. This suppression did not depend on the Rad9 checkpoint function, as the same growth defect was not suppressed by the deletion of MEC3 (Fig 5D), whose function is necessary for the checkpoint response to DSBs.
This Rif1 function on DSB resection recalls the role of Fun30, which has been shown to promote extensive resection probably by counteracting Rad9 . We found that RIF1 deletion exacerbated the resection defect of G2‐arrested fun30∆ cells (Fig 5E,F), indicating that Rif1 and Fun30 act in two different pathways.
In summary, we demonstrate a role for S. cerevisiae Rif1 in the response to DNA damage, where it promotes DSB nucleolytic processing possibly by limiting the action of the resection inhibitor Rad9. This Rif1 control on Rad9 loading becomes crucial for DSB repair when resection activities are suboptimal, such as in mre11∆, sae2∆ and exo1∆ mutants. This function is different from that of mammalian Rif1, which inhibits resection in G1 by excluding BRCA1 from the DSBs , , , . As budding yeast lacks a BRCA1 ortholog, mammalian Rif1 may have acquired this function during evolution to regulate resection in a BRCA1 context.
How Rif1 influences protein binding to DSBs remains to be determined. Budding yeast Rif1 blocks checkpoint activation at telomeres by limiting the association of checkpoint proteins , . Furthermore, Rif1 has been shown to negatively control the firing of a subset of replication origins by modulating the binding of replication factors in both yeast and mammals . Interestingly, both budding yeast and mammalian Rif1 localize at the nuclear periphery , , and also DSBs, replication origins and telomeres are clustered and tethered to the nuclear membrane . We therefore speculate that Rif1 may act at all these DNA regions possibly by directly controlling their chromatin structure and therefore their accessibility to regulatory proteins. In this view, modulation of chromatin accessibility might be the evolutionarily conserved function of Rif1.
Materials and Methods
Strain genotypes are listed in Supplementary Table S1. Strains JKM139, YMV45 and YMV80 were kindly provided by J. Haber (Brandeis University, USA). Strains carrying MEC1‐MYC allele have been constructed as described in . Cells were grown in YEP medium (1% yeast extract, 2% peptone) supplemented with 2% glucose (YEPD), 2% raffinose (YEPR) or 2% raffinose and 3% galactose (YEPRG). Synthetic dextrose plates lacking leucine (SD‐Leu) were used to maintain the selective pressure for the 2 μ LEU2 plasmids.
Double‐strand breaks end resection at the MAT locus was analysed on alkaline agarose gels as described in . Quantitative analysis of DSB resection was performed by calculating the ratio of band intensities for ssDNA and total amount of DSB products.
ChIP analysis was performed as described in . Data are expressed as fold enrichment at the HO‐induced DSB over that at the non‐cleaved ARO1 locus, after normalization of each ChIP signals to the corresponding input for each time point. Fold enrichment was then normalized to the efficiency of DSB induction.
MM, DB and MPL conceived and designed the experiments. MM, DB and MV performed the experiments. MM, DB, MV, GL and MPL analysed the data. MPL and GL wrote the paper.
Conflict of interest
The authors declare that they have no conflict of interest.
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Table S1
We thank J. Haber for strains, Arianna Lockhart for preliminary results and Michela Clerici for critical reading of the manuscript. This work was supported by grants from Associazione Italiana per la Ricerca sul Cancro (AIRC) (Grant IG11407) and Cofinanziamento 2010‐2011 MIUR/Università di Milano‐Bicocca to MPL.
FundingAssociazione Italiana per la Ricerca sul Cancro (AIRC) IG11407
- © 2014 The Authors