The mRNA export adaptor Yra1 contributes to DNA double-strand break repair through its C-box domain

Yra1 is an mRNA export adaptor involved in mRNA biogenesis and export in S. cerevisiae. Yra1 overexpression was recently shown to promote accumulation of DNA:RNA hybrids favoring DNA double strand breaks (DSB), cell senescence and telomere shortening, via an unknown mechanism. Yra1 was also identified at an HO-induced DSB and Yra1 depletion causes defects in DSB repair. Previous work from our laboratory showed that Yra1 ubiquitination by Tom1 is important for mRNA export. Here, we found that Yra1 is also ubiquitinated by the SUMO-targeted ubiquitin ligases Slx5-Slx8 implicated in the interaction of irreparable DSB with nuclear pores. We further show that Yra1 binds an HO-induced irreparable DSB in a process dependent on resection. Importantly, a Yra1 mutant lacking the evolutionarily conserved C-box is not recruited to an HO-induced irreparable DSB and becomes lethal under DSB induction in a HO-cut reparable system. Together, the data provide evidence that Yra1 plays a crucial role in DSB repair via homologous recombination. While Yra1 sumoylation and/or ubiquitination are dispensable, the Yra1 C-box region is essential in this process.


Introduction
Yra1 (Yeast RNA annealing protein 1) is an essential protein in S. cerevisiae, well characterized as an mRNA export adaptor involved in transcription elongation, 3' processing, and finally mRNA export together with the Mex67/Mtr2 export receptor and the poly(A) binding protein Nab2 [1].
Yra1 is evolutionarily conserved from yeast to human and belongs to the RNA and Export Factor (REF) family of hnRNP-like proteins [2][3][4][5]. REF proteins include a conserved domain organization with a central RNP-motif containing an RNA binding domain (RBD) and two highly conserved N-and C-terminal boxes (N-box and C-box). These domains are separated by two variable regions (N-var and C-var), rich in positively charged amino acids that mediate this event is followed by extensive resection driven by the exonuclease Exo1 as well as by the helicase Sgs1 and the nuclease Dna2 forming the 3' end ssDNA tails. The replication protein A (RPA) binds ssDNA and recruits proteins important for the DNA damage checkpoint that blocks the cell cycle, allowing DNA repair to occur [14]. A crucial component to start the HR process is the mediator Rad52 that displaces the RPA complexes to recruit the Rad51 recombinase on the 3' end ssDNA tails. Once formed, the Rad51 nucleoprotein filament drives DNA strand invasion on the homologous template through different mechanisms that allow HR to occur [14,16,17]. The HR process has to be tightly regulated to avoid aberrant genomic rearrangements. On site sumoylation of HR proteins induced under DNA damage is pivotal to ensure efficient and optimal DSB repair [18][19][20][21]. SUMO-targeted E3 ubiquitin ligases (STUbL), such as the Slx5-Slx8 complex in yeast, have also been shown to contribute to the maintenance of genome stability, although their targets have not been systematically identified [22].
In this work, we show that Yra1 is sumoylated by the SUMO ligases Siz1 and Siz2, desumoylated by the SUMO protease Ulp1 and ubiquitinated by the SUMO-dependent E3 ligases Slx5-Slx8, which are important for genome integrity [22,23]. Importantly, we find that Yra1 is recruited to DSBs in a resection-dependent process and identify the Yra1 C-box domain to be crucial for the binding and repair while Yra1 ubiquitination and/or sumoylation are not required in this process. Our results strengthen the importance of Yra1 in genome integrity and provide evidence for a critical role of Yra1 in DSB repair by homologous recombination.

Yeast strains and plasmids
The strains and plasmids used in this study are listed in S1 and S2 Tables. Primers are listed in supplementary S3 Table. The YRA1 shuffled strains were obtained by transformation of the YRA1 shuffle strain (yra1::HIS3, YCpLac33-URA3-YRA1WT, Cen) with YCpLac22-TRP centromeric plasmids encoding wild-type HA-tagged Yra1. The transformed strains were plated on 5-FOA to select against the WT YRA1 URA3 plasmid. The cells able to grow on 5-FOA contain only the YCplac22-TRP1-HA-YRA1 WT plasmid (YRA1 shuffled background). Single clones were analyzed for correct auxotrophic markers and checked for HA-Yra1 expression by Western blot with αHA antibodies.
The strains with integrated HA-YRA1 WT or HA-yra1 mutant were obtained by transformation of the W303 Mat-a/α diploid strain or FSY5073 (GA-6844 HO irreparable system [24] with a fragment containing the HA-tagged wild-type or mutant YRA1 sequences obtained by SmaI digestion of an engineered pUC18 construct. The pUC18 plasmids were obtained by Gibson assembly and contain a SmaI fragment consisting of the HA-tagged wild-type or mutant YRA1 sequences preceded by the YRA1 promoter and followed by the YRA1 3' UTR, a selective marker (URA3 or HIS3) and an additional 100 pb of YRA1 3' downstream sequences. Yeast transformants were plated on the relevant selective medium. Correct recombination and integration into the endogenous YRA1 locus was checked by PCR with a forward primer complementary to a sequence -600bp upstream of the YRA1 locus (OFS3118), not present in the plasmid sequence, and a reverse primer matching the HA-tag sequence present only in the plasmid-derived sequence (OFS3120). The W303 diploid strains containing the integrated HA-YRA1 WT or HA-yra1 mutant sequences were sporulated on K-acetate agar plates for 3 days at 25˚C and dissected. Single spores were analyzed for relevant auxotrophic markers; HA-YRA1 integration was confirmed by PCR as described above and expression of HA-Yra1 proteins was verified by Western blot.
The deletion strains were generated by homologous recombination of a cassette containing an auxotrophic marker flanked by sequences adjacent to the gene to delete. The pUG73::LEU2, pAG25::natMX4 or pUG6::kanMX6 cassettes were amplified by PCR using 80 nucleotides long forward and reverse primers (20 nt complementary to the plasmid and 60 nt complementary to the target sequences). PCR products were transformed into the YRA1 shuffle or WT W303 Mat-a/α diploid strains and correct insertion confirmed by PCR. The W303 Mat-a/α diploid strains containing the gene deletion were sporulated and single spores analyzed for auxotrophic markers. Haploid Mat-α WT W303 deletion mutants were crossed with haploid Mat-a strains containing integrated HA-YRA1 WT or mutant sequences obtained as described above. The diploid yra1 double mutants were sporulated to obtain haploid yra1 double mutants in W303 background. In the case of deletions in the YRA1 shuffle, the yra1 double mutants were obtained by plasmid shuffling as explained above.
The strains with integrated HA-YRA1 in FSY6881 (NA17 strain with HO reparable system) [25] were obtained after four back crosses between the integrated HA-YRA1 WT or HA-yra1  and HA-yra1allKR mutants in W303 and the NA17 strain. The sporulation, dissection and analysis of the strains was performed as described above. The presence of the cassette KanMX::HO-cs at URA3 and KanMX::ClaI at LYS2 was checked by PCR followed by digestion with the restriction enzymes BamH1 (near the HO site) and ClaI.

Media and culture conditions
If not specified, yeast strains were thawed on yeast extract-peptone-dextrose (YPD) plates and grown for two days at 25˚C. Cells were pre-cultured in 5 ml of liquid YPD to reach an OD 600 = 0.7-0.8 at 25˚C and diluted into 100 ml YPD overnight culture to reach OD600 = 0.8/1 at 25˚C in the morning.
For the protein stability assays using metabolic depletion of GAL-HA-YRA1 in presence of the endogenous wild-type YRA1 gene, cells expressing HA-Yra1 from the GAL promoter on a centromeric plasmid were grown over-night in selective medium containing 2% galactose. When reaching OD 600 = 0.3, cells were shifted to selective medium containing 2% glucose to repress GAL-HA-YRA1 and collected at time 0, 1h, 2h, 3h, 4h, 5h, 6h, and 7h following glucose addition.
To induce the HO endonuclease-mediated irreparable DSB, cells were grown over-night in SCLGg (SC lactate 2%/glycerol 2% containing 0.05% Glucose). Cells at OD = 0.4 were shifted to SCLGg medium containing 2% glucose for 2h (no cut induction) or to SCLGg medium containing 2% galactose to induce the HO endonuclease. Cells were collected at 30 minutes, 1h, 2h and 4h following galactose addition. To induce the HO endonuclease-mediated reparable DSB, cells were grown over-night in SCLGg (SC lactate 2%/glycerol 2% containing 0.05% Glucose). Exponentially growing cells were treated with 2% galactose to induce the HO endonuclease or not (control) for 2h. Serial dilutions of 200/100/50 cells were plated on SCLGg Glu 2%. In another related experiment, serial dilutions of exponentially growing cells in SCLGg medium were directly plated on SCLGg Gal 2% or SCLGg Gal 3%-Raf 1% to induce the HO cut, and on SCLGg Glu 2% to repress HO endonuclease expression.

Spot test
Cells grown in YPD medium to stationary phase were diluted to OD 600 = 1 and five 10-fold serial dilutions were prepared for spotting on agar plates. For each spot, 3μl were deposited on 2% glucose YPD plates in the presence or absence of drug (Zeocin 25 μg/ml, 50 μg/ml, and 100 μg/ml). Plates were incubated at 25˚C, 30˚C, 34˚C or 37˚C for 3 days.

Chromatin immunoprecipitation (ChIP) and quantitative real-time PCR
Cells grown to OD 600 = 1 were cross-linked with 1.2% of formaldehyde (Molecular Biology grade Calbiochem TM ) for 10 minutes at 25˚C under continuous gentle agitation, quenched with 250mM of glycine (Sigma) for 5 min at 25˚C and then on ice for at least 5 min, washed with PBS 1X and frozen at -20˚C. Pellets of 100 ml cultures at OD 600 = 1 were resuspended in 1ml of FA lysis buffer (10mM HEPES KOH pH 7.5, 140mM NaCl, 1mM EDTA pH 8, 1%Triton X-100, 0.1% sodium deoxycholate) containing a protease inhibitor cocktail (Complete tablets, Mini EDTA-free, Roche). Cells were mechanically broken with a magnalyser at 6500rpm for 30 seconds (4 times), and genomic DNA was sonicated for 20 cycles of 30 seconds ON/ OFF in presence of 0.5% SDS added before the sonication step. Samples were centrifuged at 13000rpm for 15 min at 4˚C, and chromatin (supernatant phase) was quantified by Bradford. For each IP, 1/10 of the total extract was kept as INPUT for final normalization. Chromatin extracts (500μg) were incubated at 4˚C o/n with a specific antibody. In parallel, magnetic beads (Dynabeads Magnetic, Thermo Fisher Scientific) were incubated with BSA 5 mg/ml at 4˚C o/n. The magnetic beads were washed twice with FA lysis buffer and resuspended with the same volume of FA lysis buffer containing a protease inhibitor cocktail (beads 50% v/v). The chromatin extracts with a specific antibody were incubated with 30μl of magnetic beads for 4h at 4˚C on a rotating wheel. The magnetic beads were then washed twice with FA lysis buffer, twice with FA 500 (50mM HEPES KOH pH 7.5, 500mM NaCl, 1mM EDTA pH 8, 1%Triton X-100, 0.1% sodium deoxycholate), once with Buffer III (20mM Tris-HCl pH 8, 1mM EDTA pH 8, 250mM LiCl, 0.5% NP40, 0.5% sodium deoxycholate) and once with TE 1X (100mM Tris-HCl pH 8, 10mM EDTA pH 8). DNA was eluted with 200μl of elution buffer (50mM Tris-HCl pH 7.5, 1% SDS) at 65˚C for 20 minutes. IP and INPUT DNAs were finally de-crosslinked with proteinase K (Roche) (0.4 μg/μl) for 2 hours at 42˚C, and o/n at 65˚C. The decrosslinked IP and INPUT DNAs were purified (Promega, Wizard Genomic DNA Purification Kit). IP and INPUT (2μl) were quantified by qPCR with SYBR Green PCR Master Mix (Applied Biosystems) using specific primers described in S3 Table. To check the HO cut induction, the corresponding locus was quantified by qPCR and the level was normalized to SCR1.
The following antibodies were used: a rabbit polyclonal αHA antibody (Enzo), a rabbit polyclonal αYra1 antibody and corresponding pre-immune (Stutz laboratory).
Cells lysates were spun at 13000 rpm for 20 min. Between 5-8 mg of protein from the supernatant was incubated with 100μl of Ni-NTA acid-agarose (Qiagen) for 2h at room temperature on a rotating wheel. Agarose beads were washed once with Guanidinium buffer and three times with Urea buffer (100 mM sodium phosphate at pH 6.8, 10 mM Tris-HCl, 8M urea, 20 mM imidazole, 0.2% Triton X-100, complete protease inhibitor mix [Roche]). His6-ubiquitinated and His6-SUMOylated proteins were eluted with 40 μl of Sample Buffer and boiled for 5 min at 95˚C. 20 μl samples were analyzed by Western blot with the relevant antibodies: αHIS for ubiquitinated proteins, αSUMO for SUMOylated proteins, αHA or αYra1 for ubiquitinated or SUMOylated HA-Yra1 and Yra1 proteins. Input samples were also precipitated with TCA 5%, the pellets resuspended with Sample Buffer and boiled 5 min at 95˚C to be analyzed by Western Blot with αHA for HA-Yra1, αYra1 for Yra1 and αPgk1 for Pgk1 as loading control.

Poly(A)+ RNA FISH experiments
The FISH experiments on the YRA1 shuffled strains deleted for various ubiquitin ligases were done essentially as described in [8], while the FISH experiments on the different HA-YRA1 integrated strains were performed as described in [28]. In the latter case, images were acquired using an Olympus BX61 wide field epi-fluorescence microscope with a 100X/1.35NA UPla-nApo objective. Samples were visualized using an X-Cite 120 PC lamp (EXFO) and the ORCA-R2 Digital CCD camera (Hamamatsu). Metamorph software (Molecular Devices) was used for acquisition. Z-sections were acquired at 200nm intervals over an optical range of 8.0 μm. In both cases Poly(A) + mRNA in situ hybridization was performed with a Cy3-labeled oligo-dT (50) probe.

Colony forming unit assay (CFU)
In the reparable HO cut system, three serial dilutions (200/100/50) of exponentially growing cells in SCLGg medium were plated on SCLGg Gal 2% or SCLGg Gal 3%-Raf 1% or SCLGg Glu 2% and incubated at 25˚C for 5 days. The percentage of colonies was determined as the relative number of Colony Forming Units (CFUs) in each strain plated on SCLGg Gal 2% or Gal 3%-Raf1% compared to the one plated on SCLGg Glu 2%. To normalize the variability in growth due to the different media condition, the CFUs of each strain transformed with pGal-HO endonuclease were normalized to the corresponding strain transformed with the empty vector (%CFU = (%CFU on SCLGg Gal 2% pGal-HO/EV)/ (% CFU on SCLGg Glu 2% pGal-HO/EV).
In the irreparable HO cut system, three serial dilutions of (10000/5000/2500) cells were plated on SCLGg Gal 2%-Raf 2% to induce the HO cut and three serial dilutions of 200/100/50 cells were plated on SCLGg-Raf 2% as control for the number of cells plated. Plates were incubated at 25˚C for 8 days. The % of colonies was expressed as CFUs growing on SCLGg-Gal2%-Raf2% relative to the CFU growing on SCLGg-Raf 2% used as reference for the number of cells plated on SCLGg Gal 2%-Raf2%.

Rad52 foci analysis
YRA1 WT and yra1 mutants strains were transformed with the YCpLac111-LEU2-RAD52-YFP construct. Cells exponentially growing on selective medium were treated or not with Zeocin (100 μg/ml for 2h) and fixed with 4% PFA. Images were taken with the LSM700 microscope using laserline 405 nm for DAPI detection and laserline 514 nm for YFP, taking 8 zstacks of 0.25 nm. Bright-field images were taken to define the cell cycle stage of the imaged cells. The Rad52-YFP foci were revealed and counted through the Z-stack images for each cell.

Yra1 is modified by the SUMO-targeted E3 ubiquitin ligase Slx5-Slx8
Previous work from our laboratory showed that Yra1 ubiquitination by Tom1 elicits Yra1 dissociation from mRNPs, presumably in the context of the nuclear pore complex (NPC), allowing proper mRNP export into the cytoplasm [8]. Intriguingly, Yra1 ubiquitination is not fully abrogated in the Δtom1 mutant, suggesting that other E3 ligases are involved in Yra1 regulation, possibly for other Yra1 functions. In view of the putative role of Yra1 in genome stability, we wondered whether this protein could be modified by SUMO-dependent ubiquitination.
The ubiquitination assay of HA-Yra1 in wild-type and in Δtom1, Δslx5, Δslx8, Δslx5Δslx8, Δslx5Δtom1, Δslx8Δtom1 mutant backgrounds showed that the Yra1 ubiquitination detected in the Δtom1 mutant was completely abrogated in the Δslx5Δtom1 and Δslx8Δtom1 double mutants (Fig 1A), indicating a role for both the Slx5-Slx8 and Tom1 E3 ligases in Yra1 regulation. Surprisingly, the ubiquitinated Yra1 levels are higher in Δslx5, but not in Δslx8 and in Δslx5Δslx8 mutants. This may reflect that Slx5 and Tom1 compete for Yra1 ubiquitination, and that loss of Slx5 thereby favors ubiquitination of Yra1 by Tom1, since the phenotype is lost in the Δslx5Δslx8 double mutant. Because, the Slx5-Slx8 E3 ligase complex activity is stimulated by substrate sumoylation [29], and in view of the reported identification of Yra1 as potentially sumoylated in a proteome-wide study [30], we wondered whether Yra1 was itself modified by SUMO. We showed that Yra1 is indeed sumoylated (Fig 1B and 1C). Both Siz1 and Siz2 SUMO E3 ligases are involved in this modification as Yra1 sumoylation is fully abrogated in the Δsiz1Δsiz2 double mutant background ( Fig 1C). Furthermore, Yra1 is de-sumoylated by the SUMO protease Ulp1 as Yra1 sumoylation increased in the ulp1 temperaturesensitive (ts) mutant (S1A Fig). These data support the hypothesis that Yra1 is regulated both by sumoylation and ubiquitination. In addition, HA-Yra1 ubiquitination was increased in the ulp1 ts mutant compared to a wild-type background, suggesting a possible stimulating effect of sumoylation on ubiquitination (S1B Fig). Conversely, sumoylation does not appear to depend on ubiquitination. Indeed, Yra1 can still be sumoylated in the Δslx8Δtom1 strain in which ubiquitination of Yra1 is prevented; the result is not as clear in Δslx5Δtom1 as the overall protein sumoylation is strongly reduced in this mutant (S1C Fig).
Known targets of Slx5-Slx8 are controlled by ubiquitin-dependent proteasomal degradation [31][32][33][34][35]. To define whether Yra1 ubiquitination by Slx5-Slx8 may target Yra1 to degradation, we used metabolic depletion to examine Yra1 turnover. Because YRA1 is essential, an HAtagged version of YRA1 was expressed from a galactose-inducible promoter on a plasmid transformed into a strain expressing a wild-type YRA1 gene. Switching cells from galactose to glucose-containing medium represses GAL-HA-YRA1 gene expression and allows following

HIS-Ubi proteins
HIS-SUMO proteins We previously proposed that Yra1 regulation by Tom1 is linked to the function of Yra1 in mRNP export [1,8]. Visualization of poly(A)+ RNA distribution by fluorescence in situ hybridization (FISH) in the Δslx5 and Δslx8 single mutants did not show any nuclear poly(A) + RNA retention while the Δslx5Δtom1 (32.3%) and Δslx8Δtom1 (26%) double mutants had mRNA export defects comparable to the Δtom1 mutant (30.8%) (S3A Fig). These observations suggest that Yra1 ubiquitination by Slx5-Slx8 may regulate a function of Yra1 distinct from mRNA export.

Loss of the Yra1 C-box sensitizes the genome to DSBs
Since our data indicate that Yra1 is modified by Slx5-Slx8, a STUbL important for genome stability [36], we examined whether the abrogation of Yra1 ubiquitination and sumoylation induces defects in genome integrity. For this purpose, we used the HA-yra1allKR mutant that cannot be ubiquitinated nor sumoylated since all the Lysines (K) are replaced by Arginines (R) (Fig 2A) [8].
We also used the HA-yra1(1-210) mutant which codes for a protein that is still ubiquitinated and sumoylated but lacks the highly conserved 16 C-terminal amino-acids (Fig 2A). Because Yra1 levels are maintained through splicing autoregulation, the intron was retained in both wild-type and mutant HA-YRA1 constructs to limit the potential toxic effect of Yra1 overexpression [4,11,[37][38][39]. Although the C-terminal domain has been implicated in splicing inhibition [4], the HA-yra1(1-210) protein is only mildly overexpressed compared to wildtype HA-Yra1; the HA-yra1allKR mutant is also slightly overexpressed indicating a potential role of post-translational modifications in splicing autoregulation (Fig 2A and 2B). Both mutants presents only a slight growth defect at 25˚C but are thermosensitive as shown by spot test analysis at different temperatures (25˚C, 30˚C, 34˚C and 37˚C) ( Fig 2C). Notably, the stronger thermosensitivity of the HA-yra1allKR mutant at 37˚C may be linked to the overexpression of the HA-yra1allKR protein after 2h and 5h of growth at 37˚C respectively of 1.75 and 1.98 fold increase (S3B Fig). Interestingly, additional spot test analyses in the presence of Zeocin indicated that the HA-yra1(1-210) but not the HA-yra1allKR mutant is sensitive to this genotoxic drug (Fig 2D). Importantly, the Zeocin treatment did not affect Yra1 ubiquitination and sumoylation (S4A Fig), nor Yra1 protein stability (S4B Fig); it also had no effect on the HA-Yra1 WT, HA-yra1(1-210) and HA-yra1allKR protein levels ( S4C Fig). Finally, the lack of nuclear poly(A)+ RNA retention scored in these mutants in the conditions used in this study (25˚C) was unchanged in the presence of Zeocin (S5A Fig). These observations indicate that the Yra1 C-box is important for genome stability in the presence of DNA double strand breaks (DSBs) while Yra1 ubiquitination and sumoylation are not. Importantly, this function of Yra1 is separable from its canonical role in mRNA export.
Double strand breaks cluster together in homologous recombination centers characterized by the co-localization with the repair factor Rad52. We monitored Rad52 foci formation in yra1 mutants with and without Zeocin treatment to define whether they accumulate sumoylated by Siz1/Siz2. Yra1 sumoylation assay in wild-type as well as Δsiz1, Δsiz1/siz2, Δsiz2 and mms21-11 mutant backgrounds was performed as described above. His-sumoylated proteins were affinity-purified and the sumoylated forms of Yra1 detected by Western Blot with αYra1 antibodies. One representative experiment of 2 is shown.

Yra1 is recruited to an irreparable DSB (HO cut)
To obtain more direct evidence for a possible role of Yra1 in the DNA damage response pathway (DDR), we induced an irreparable DSB at the MAT locus using a galactose-inducible HO endonuclease as previously described [22] (S6A Fig). Consistent with the irreparable nature of the induced HO cut, these strains do not grow on galactose (S7A Fig). Yra1 recruitment at the HO cut, examined by ChIP with an αYra1 antibody, was significant at regions close to the DSB after 2h of HO induction (Fig 3A).
Considering that the efficiency of the cut is nearly 100% after 30' of HO induction [22] (S8A Fig), the recruitment after 2h suggests it occurs following extensive resection.
To define whether the sensitivity to Zeocin of the HA-yra1(1-210) mutant may be due to its impaired recruitment to DSB loci, sequences encoding HA-tagged wild-type or mutant Yra1 (HA-YRA1 WT, HA-yra1  and HA-yra1allKR) were integrated into the irreparable HO DSB strain at the YRA1 locus; the recruitment of these different HA-Yra1 proteins at the HO cut was examined by ChIP using αHA antibodies after 2h in galactose (Fig 3B), which induces efficient HO cleavage in both wild-type and mutant strains (S8B Fig). These experiments show that the HA-yra1allKR protein is recruited to the HO cut site to similar levels as the HA-Yra1 WT in Galactose (HO cut) (Fig 3B). In contrast, although in this experiment the HA-yra1(1-210) protein is expressed to slightly higher levels than HA-Yra1 WT (S7B and S7C Fig), its binding to the HO site does not increase in galactose, suggesting that the Yra1 C-terminal region is important for Yra1 recruitment to the DSB.
Since Yra1 is recruited to the HO cut 2h after Gal induction, once there has been extensive resection, we asked whether RPA binding to the HO cut might vary in the different HA-yra1 mutants and whether Yra1 binding may depend on extensive resection. RPA association was not affected in the HA-yra1 mutants despite the lack of HA-yra1(1-210) recruitment (S7D Fig), suggesting that RPA binding is probably not dependent on Yra1 recruitment. Importantly, HA-Yra1 binding to the HO cut was still present in the Δsae2Δexo1 mutant (Fig 3C), in which the extensive resection is performed by the helicase Sgs1 [40,41]; however HA-Yra1 recruitment was compromised in the Δexo1Δsgs1 mutant (Fig 3C), in which the extensive resection is completely abrogated [40,41]. The HA-Yra1 protein level as well as the HO cut efficiency in the Δexo1Δsgs1 mutant are comparable to those measured in the WT background

The Yra1 C-box, but not its ubiquitination and sumoylation, is important for genome stability. (A)
Scheme of Yra1 mutants used in this study with corresponding ubiquitination and sumoylation assays. Left: Ubiquitination assay of shuffled HA-YRA1 WT, HA-yra1  and HA-yra1allKR mutants performed as described in Materials and Methods. Right: Sumoylation assay of shuffled HA-YRA1 WT, HA-yra1allKR and HA-yra1(1-210) mutants performed as described in Materials and Methods. One representative experiment of at least three is shown. Note that Yra1 is a very basic and charged protein and its non-modified forms tend to be retained on the Ni-NTA agarose beads. This is particularly striking with the HA-yra1allKR protein whose expression is increased in these strains. The mRNA export adaptor Yra1 contributes to DNA double-strand break repair through its C-box domain

Kb from HO cut
The mRNA export adaptor Yra1 contributes to DNA double-strand break repair through its C-box domain (S8C and S9A Figs). These observations support that Yra1 binding at the HO cut is dependent on extensive resection.

The Yra1 C-terminal region is important for DSB repair (HO cut)
Since irreparable DSBs relocate to nuclear pores in G1/S phase [23] within 2h after cut induction [22], one possibility is that Yra1 recruitment to irreparable DSB is the consequence of HO cut re-localization to pores. To exclude this possibility, we took advantage of an HO cut reparable system (S6B Fig) [25], since DSB repair occurs within the nuclear interior [42,43]. To define whether the HA-yra1 mutants may be defective in DSB repair, the HA-YRA1 WT, HA-yra1(1-210) and HA-yra1allKR sequences were integrated at the YRA1 locus of the reparable HO DSB strain and the percentage of cells surviving under HO cut induction was examined. Three serial dilutions of exponentially growing cells were plated on galactose 2% or galactose 3%-raffinose 1% to induce the HO cut, and on glucose 2% to repress HO endonuclease expression. CFUs were counted as indication of cells able to repair the DSB in the HA-YRA1 WT, HA-yra1(1-210), HA-yra1allKR strains transformed with a pGAL-HO endonuclease plasmid or Empty Vector; a No-Tag and a Δrad52 strain transformed with the Empty Vector only were used as controls as these two strains contain an endogenous pGAL-HO endonuclease sequence ( Fig 4A).
Interestingly, like the Δrad52 control strain, the HA-yra1(1-210) mutant was not able to grow on galactose when the reparable HO cut is induced, indicating that the Yra1 C-box is important for DSB repair. This effect was confirmed by spot test analysis (Fig 4B). Notably, although both HA-yra1 mutants have comparable cut efficiency after 2h in Galactose (S9B Fig), the HA-yra1allKR showed no growth defect under HO cut induction whether in the CFU assay or in the spot test, indicating that Yra1 ubiquitination and sumoylation are not required for DSB repair (Fig 4A and 4B).
DSBs can be repaired by NHEJ if they occur in the G1 phase and by HR if they occur in G2 and S phase [44]. During the G1 phase, the NHEJ pathway inhibits extensive resection crucial for the HR process [45]. We asked whether the HA-yra1(1-210) mutant that shows impaired DSB repair also has defects in NHEJ. To address this question we took advantage of the irreparable HO cut system (S6A Fig) since the HO cut cannot be repaired by HR and a low percentage of cells can survive thanks to the Break-Induced Replication (BIR) and the NHEJ pathways [46]. To define whether the HA-yra1 mutants may be defective in NHEJ, the percentage of cells surviving under HO cut induction was examined in the Δsae2Δexo1 background since cells lacking Sae2 promote the NHEJ pathway to repair the DSBs [47]. While the number of CFUs was clearly increased in the Δsae2Δexo1 background compared to WT, the HA-YRA1

Fig 3. Yra1 is recruited to an irreparable DSB HO cut site. (A) Yra1 recruitment at the HO cut site was defined by
ChIP with an αYra1 antibody after 0.5h, 1h, 2h and 4h of HO endonuclease induction with galactose using the GA6844 strain described in [22]. The 2h Glucose time point was used as no cut control. ChIP values are shown as percentage of input at 0.6Kb, 1.6Kb, 4.5Kb, 9.6Kb and 23Kb from the HO cut. The average of 6 experiments is shown with corresponding standard error of the mean. Two way ANOVA test was performed with multiple comparisons; P values < 0.05 ( � ), < 0.01 ( �� ), < 0.001 ( ��� ) that refer to Glu 2h (no cut) are shown. (B) Yra1 mutants are differentially recruited to the irreparable HO cut site. ChIP using αHA antibody of HA-Yra1 WT, HA-yra1(1-210), HA-yra1allKR, Yra1 WT (no tag) at 0.6 Kb from the HO cut site after 2h of HO induction with Galactose using the strains with HA-YRA1 WT or mutants integrated in strain GA6844 described in [22].   a 1 ( 1 -2 1 0 )  WT, HA-yra1  and HA-yra1allKR mutants showed a comparable number of colonies in both contexts, indicating that the yra1 mutants do not have defects in the NHEJ pathway and BIR (Fig 4C).
Overall these observations support the view that Yra1 is important for DSB repair in a process dependent on the 16 amino acids C-terminal region and resection. Absence of this domain may result in the inability to repair HO cuts possibly because of the reduced capacity of Yra1 to interact with the DSB after extensive resection, a key step for the DSB repair by HR.

Discussion
This study strengthens the importance of Yra1 in genome stability. In particular, our data provide evidence that the Yra1 C-terminal box is crucial for DSB repair. We have started to investigate the sensitivity of yra1 mutants to DNA damage based on the observation that Yra1 is not only sumoylated by Siz1-Siz2 but also ubiquitinated by Slx5-Slx8, a SUMO-dependent E3 ligase important for genome stability (Fig 1). However, our data indicate that Yra1 ubiquitination and sumoylation are not important for DSB repair since the HA-yra1allKR mutant that completely abrogates Yra1 ubiquitination and sumoylation (Fig 2A) [8] does not display sensitivity to the DSB agent Zeocin (Fig 2D) nor any defect in DSB repair (Fig 4).
To investigate the effect of Yra1 on genome stability, we rather took advantage of the HA-yra1(1-210) mutant that lacks the Yra1 C-box domain (Fig 2). This domain does not interact with RNA [7] and the HA-yra1(1-210) mutant does not show any obvious mRNA export defect (S5A Fig). Interestingly, our data show that the HA-yra1(1-210) mutant is sensitive to the DSB inducing genotoxic agent Zeocin (Fig 2D), a phenotype that is not due to a defect in mRNA export in this condition (S5A Fig). In line with these results, it was recently published that the DAmP allele of YRA1 is specifically sensitive to Zeocin [13]. Together, these observations suggest that lack of the Yra1 C-box either promotes DSBs or impairs DSB repair in a process independent of mRNA export activity.
A recent study revealed that Npl3, an RNA binding protein involved in mRNP biogenesis, contributes to DSB resection by ensuring efficient production of EXO1 mRNA [48]. While Npl3 was proposed to have an indirect role in repair, our observations indicate that Yra1 is recruited to an irreparable DSB after 2h of cut induction and therefore extensive resection, consistent with a direct role of Yra1 in DSB repair (Fig 3). Accordingly, Yra1 binding at the HO cut is impaired in the Δsae2Δexo1 mutant defective in the extensive resection event (Fig   Fig 4. Survival under persistant induction of a reparable HO cut. (A) The Yra1 C-box is important for DSB repair. NA17 strains [25] containing integrated HA-YRA1 WT (WT) , HA-yra1(1-210), HA-yra1allKR were transformed with pGAL-HO endonuclease containing plasmid or Empty Vector, as well as No-Tag (NA17) and Δrad52 strains containing endogenous pGAL-HO endonuclease were transformed with an Empty Vector. Diluted cells were plated on SCLGg Gal 2% or Gal 3%-Raf1% to constantly induce HO cut, and on SCLGg Glu 2% to repress HO endonuclease expression. The percentage colony forming units (CFUs) was determined as described in Materials and Methods. The average of 3 independent experiments for each condition SCLGg Gal 2%/ Glu 2% and SCLGg Gal 3%-Raf 1%/ Glu 2% is shown with corresponding standard error of the mean. One way ANOVA test was performed with multiple comparisons and P value < 0.001 ( �� ) is shown on the graph referring to HA-YRA1 WT. (B) Yra1 C-box is important for DSB repair. Spot test analysis on Leu-SCLGg Glu 2%, SCLGg Gal 2%, SCLGg Gal 3%-Raf 1%, at 25˚C of exponentially growing HA-YRA1 WT (WT) , HA-yra1(1-210), HA-yra1allKR (transformed with pGAL-HO endonuclease containing plasmid or Empty Vector); No-Tag and Δrad52 strains (containing endogenous pGAL-HO endonuclease and transformed with an empty vector) served as controls. One representative experiment out of 3 is shown. (C) yra1 mutant survival in the Δsae2Δexo1 background under persistant induction of an irreparable HO cut. Dilutions of exponentially growing cells of GA6844 strain [22] containing integrated HA-YRA1 WT (WT), HA-yra1 , HA-yra1allKR combined or not with Δsae2Δexo1 were plated on SCLGg Gal 2%-Raf2% to constantly induce the HO cut. Corresponding dilutions of cells were plated on SCLGg Raf2%. The percentage of Colony Forming Units (CFUs) was determined as described in Materials and Methods. The average of 2 independent experiments is shown with corresponding standard error of the mean. One way ANOVA test was performed with multiple comparisons and P value < 0.05 ( � ) is shown on the graph. https://doi.org/10.1371/journal.pone.0206336.g004 The mRNA export adaptor Yra1 contributes to DNA double-strand break repair through its C-box domain 3C). Importantly, the recruitment to an irreparable DSB does not depend on Yra1 ubiquitination and sumoylation but requires the conserved C-box, suggesting that this domain may be involved in repair, although it has no effect on RPA binding to the locus (Fig 3B and S7D Fig). The C-box could mediate Yra1 recruitment to DSBs by virtue of a direct interaction with resected DNA ends or DSB-associated proteins. However, we cannot fully exclude that Yra1 recruitment to irreparable DSBs may be the consequence of HO cut re-localization to the nuclear pore that occurs within 2h after cut induction [22].
Furthermore, we also examined whether the irreparable DSB can be repaired by alternative pathways such us Non Homologous End Joining (NHEJ) [14] or Break Induced Replication (BIR) [23] by inducing persistent irreparable HO cut in the HA-yra1 mutants and plating cells on Galactose. To specifically favour the NHEJ pathway [47], the HA-yra1 mutants were expressed in the Δsae2Δexo1 background. The YRA1 WT and yra1 mutants showed comparable colony formation ability both in a wild-type and Δsae2Δexo1 background, indicating that the Yra1 C-box and Yra1 ubiquitination/sumoylation do not contribute to alternative repair pathways such as NHEJ (Fig 4C).
To directly address DSB repair efficiency in the HA-yra1allKR and HA-yra1(1-210) mutants, we used the HO reparable system described in [25]. Unfortunately, we were unable to observe significant recruitment of Yra1 to this type of DSB by ChIP, probably because the HO reparable system is more dynamic. However, an independent recent study identified Yra1 at an HO-induced reparable DSB using ChAP-MS (Chromatin Affinity Purification with mass spectrometry) [13]. These data indicate that Yra1 is recruited to the DSB locus also when the HO cut is located within the nucleus [49]. Thus, the observed Yra1 binding at the irreparable HO cut (Fig 3) may be specific rather than the indirect consequence of DSB relocalization to the nuclear periphery.
Besides detecting Yra1 at reparable DSBs, the recent study by Wang et al. [13] also shows that a Yra1 DAmP hypomorph mutant has a defect in global DSB repair following Zeocin treatment comparable to that observed in the absence of the key repair protein Rad52. As discussed by the authors, this global effect probably results from the reduced expression of Rad51 due to defective mRNA biogenesis and export activity in the presence of low levels of Yra1. The same study investigated the importance of Yra1 in the repair of a single HO cut using the Yra1 anchor away system. These experiments were unable to demonstrate a role for Yra1 in this process probably because the depletion by anchor away was not optimal. Since irreparable DSBs lead to cell death [14], we addressed the critical role of Yra1 in DSB repair by defining the repair efficiency of the HA-yra1allKR and HA-yra1(1-210) mutants based on survival following induction of the reparable HO cut (Fig 4A and 4B). Interestingly, while the HA-yra1allKR mutant has not effect, the HA-yra1(1-210) mutant exhibits very poor survival, comparable to that observed in Δrad52 (Fig 4A and 4B). Since the HA-yra1(1-210) strain has no obvious mRNA export phenotype and exhibits normal Rad51 levels (S5A and S9C Figs), our data support the hypothesis that Yra1 may play a direct role in DSB repair by HR in a process that involves extensive resection and C-box-dependent recruitment of Yra1 to the resected damaged site (Fig 3). Since the resection at DSBs is paralleled by transcription inhibition of surrounding loci [50], it is likely that Yra1 binding to resected DNA ends is not linked to transcription. In conclusion, one possibility is that C-box-dependent Yra1 recruitment is important for repair by somehow favoring homologous recombination at the DSB. Eventhough we identified a link between the Yra1 C-box domain and the DBS repair process during extensive resection, our experiments did not reveal defects in the subsequent steps of HR such as RPA binding ( While our data show that Yra1 ubiquitination and sumoylation are not required for DSB repair, we cannot exclude that Yra1 modification by Slx5-Slx8 may facilitate relocalization of irreparable DSBs to nuclear pores [22,23]. The physiological relevance of irreparable DSB relocation to the nuclear periphery is still not fully clear. It has been speculated that it leads to proteasomal degradation of DSB-bound proteins targeted by the STUbL Slx5-Slx8 [22] to induce alternative repair pathways such us Break Induced Replication [23]. In that respect, our data show that ubiquitination and sumoylation do not lead to Yra1 degradation (S2 Fig). Furthermore, Yra1 ubiquitination and sumoylation are not required for cell growth after irreparable DSB induction suggesting that they are not important for non-canonical repair and NHEJ (Fig 4C).
In summary, this work indicates that at physiological expression levels, Yra1 is beneficial for genome stability by facilitating the repair of DSBs by HR in a C-box-dependent and sumoylation/ubiquitination-independent manner. Future studies should address how Yra1 recruitment to DSBs may contribute to repair through homologous recombination. showing the Gal-induced HO-mediated irreparable DSB described in [24]. The HO endonuclease is expressed in the presence of Galactose, inducing the HO cut at the Mat locus that cannot be repaired because of the deletion of HML and HMR. (B) Scheme showing the Gal-induced HO-mediated reparable DSB described in [25]. The HO endonuclease is expressed in the presence of Galactose, inducing the HO cut at the KanMx cassette next to the URA3 locus. The repair of the DSB at the HO cut is possible by HR thanks to the KanMX cassette at the LYS2 locus. If this occurs, the repair will result in an HO insensitive KanMX cassette at the URA3 locus as well as the loss of the short unique sequence surrounding the initial HO cut site. HA-yra1allKR expressed from copies integrated into the GA-6844 strain [22] after 2h in Glucose or Galactose to induce the irreparable HO cut. The levels of WT or mutant HA-Yra1 proteins remain quite constant between the different time points Glu 2h and Gal (0.5h, 1h, 2h). Values of HA-Yra1/Pgk1 are shown below the blot. One representative Western Blot is shown. Analysis of HO cut site levels in the GA6844 strain described in [22] after 0.5h, 1h, 2h and 4h of HO endonuclease induction with galactose. The HO cut genomic locus was quantified by qPCR and the level was normalized to SCR1. The average of 6 independent experiments is shown with corresponding standard error of the mean. (B) Analysis of HO cut site levels in the HA-YRA1 WT and HA-yra1 mutants integrated in GA6844 strain described in [22] after 2h of HO endonuclease induction with galactose or 2h in Glucose (no HO induction). The average of 3 independent experiments is shown with corresponding standard error of the mean.