At the anaphase onset, Esp1 cleaves Scc1, opening the cohesin ring and releasing sister chromatids from cohesion. Thereafter, Scc1 needs to be translated de novo since cleaved fragments are unstable and degraded by the proteasome (Rao et al., 2001). However, it is known that activation of Esp1 can be blocked by DDC kinases (Yam et al., 2020). In addition, DDC kinases also phosphorylate Scc1 to facilitate DI-cohesion at a post-replicative stage (Heidinger-Pauli et al., 2008). Thus, it is feasible that DSB generation in late mitosis can stabilize de novo Scc1 as well as rendering it cohesive. Noteworthy, Smc1 and Smc3 subunits remain attached to chromatin after segregation (Garcia-Luis et al., 2022; Renshaw et al., 2010; Tanaka et al., 1999).

To assess how cohesin subunits behave after DSBs in late mitosis, cdc15-2 strains that bear Scc1 tagged with myc epitopes and Smc1/3 tagged with HA epitopes were first blocked in telophase at 34°C for 3 hr (Tel) (Figure 1A). The cdc15-2 allele encodes a conditionally thermosensitive Cdc15 kinase, which drives telophase-to-G1 transition (Machín and Ayra-Plasencia, 2020). These strains also harbor an inducible HO endonuclease, which cuts the HO cutting site (HOcs) at the MAT locus in chromosome III. The HO gene is under the control of a promoter system that responds to β-estradiol (lexO4p:HO:plus LexA-ER-B112-T) (Gnügge and Symington, 2020; Ottoz et al., 2014). In wild-type strains, the HO DSB is repaired by HR with the intramolecular ectopic loci HMR and HML. The locus selection depends on the MAT allele (the MATa HO DSB repairs with HML) and results in a gene conversion (e.g., from MATa to MATα). In order to undoubtedly attribute the effects on cohesin to the DSB and not to downstream events of the repair process, we used a set of strains unable to repair the HO-mediated DSB (Δhml Δhmr double mutant). After the Tel arrest, the cultures were divided into three. The first subculture served as a mock control, the second was used to generate the single HO DSB (+HO), and the third was used to generate multiple random DSBs by means of the radiomimetic drug phleomycin (+phle). Noteworthy, these latter DSBs are repairable in principle. Incubation of the subcultures was prolonged at 34°C for 2 extra hours.

The presence or absence of Scc1-3myc, Smc1-6HA, and Smc3-6HA was checked by Western blot. As controls, we included asynchronous samples from the same strains, before the telophase arrest, as well as a parallel arrest at G2/M by the drug nocodazole (Nz). As expected, Smc1, Smc3, and Scc1 bands were detected in both cycling (asyn) cells and cells blocked in G2/M (Figure 1B–D). The Scc1 signal disappeared almost entirely when cells were arrested in telophase and remained low in the mock control 2 hr later. Strikingly, full-length Scc1 was restored after single and multiple DSBs generation (Figure 1B, E). Compared to the mock condition, phleomycin treatment increased Scc1 levels fivefold, reaching approximately half of the levels observed in G2/M. In the case of HO DSBs, Scc1 levels surpassed those of the mock by more than 15-fold and were even higher than in G2/M. In contrast, Smc1 and Smc3 levels remained more stable in both types of DSBs (Figure 1B–D).

The above findings suggest that the cohesin complex could be newly formed and loaded onto the breaks, promoting DI-cohesion, which could in turn ease the recombinational events. To assess whether cohesin subunits are reassembled after DSB induction in telophase, we performed a co-immunoprecipitation (co-IP) experiment using Smc1-6HA as the bait protein in a strain that also expresses Scc1-3myc (Figure 2A, Figure 2—figure supplement 1). In our experimental setup, in addition to the anti-HA and anti-myc antibodies, we included an antibody for the acetylated form of Smc3 at lysine residues K112 and K113 (acSmc3), which have been linked to the chromatin-bound pool of the Smc1–Smc3 dimer (Beckouët et al., 2010; Minamino et al., 2015; Rolef Ben-Shahar et al., 2008; Unal et al., 2008). In asynchronous cultures, both Scc1-3myc and acSmc3 co-immunoprecipitated with Smc1-6HA, indicating that we were able to pull down the chromatin-bound pool (acetylated Smc3) of the canonical Smc1–Smc3–Scc1 complex. A parallel co-IP using untagged Smc1 did not recover the other subunits, confirming the specificity of the interaction. As expected, Scc1 was not detected in the pull-down under the mock condition (no DSBs); however, it was recovered after generation of the HO DSB. Surprisingly, this was not the case when random DSBs were induced by phleomycin. Incidentally, acSmc3 appeared as two bands in the asynchronous culture but as a single upper band in all telophase conditions, suggesting a difference in posttranslational modifications at this cell cycle stage.

The reconstitution of the complex and the detection of the acetylated form of Smc3 prompted us to confirm whether the newly formed Scc1 was bound to chromatin, since the coexistence of two pools could not be ruled out based on the co-IP experiments alone. For instance, a soluble Smc1–Smc3–Scc1 complex could coexist with a chromatin-bound Smc1–acSmc3 dimer, which could represent up to 30% of the total Smc1–Smc3 in anaphase (Beckouët et al., 2010). To this end, we performed biochemical fractionation of the chromatin-bound proteins, observing an increase in all cohesin subunits in the chromatin fraction after the HO DSB (Figure 2B, Figure 2—figure supplement 2). The increase was approximately twofold for Smc1–acSmc3 and up to sevenfold for Scc1. These results confirmed that the new Scc1 binds to a reconstituted chromatin-bound cohesin complex following the HO DSB. In contrast, this chromatin recruitment was not observed after phleomycin treatment. The differences in co-IP and fractionation due to the source and nature of the DSBs are intriguing. In this experimental setup, the HO DSB is irreparable, whereas DSBs caused by phleomycin can eventually be repaired by anaphase regression and sister loci coalescence (Ayra-Plasencia and Machín, 2019). In addition, the HO DSB leaves clean ends that can be easily resected, while the DSBs after phleomycin are chemically modified and require extended processing.

Next, we used chromatin immunoprecipitation (ChIP) to investigate whether the HO DSB recruits more cohesin to the vicinity of the HOcs. This recruitment has been reported during DSBs that exert a G2/M arrest by the DDC (Scherzer et al., 2022; Ström et al., 2004). Because the level of Scc1 changes so dramatically after the HO DSB, we selected one of the subunits that does not for the ChIP. An Smc3-HA strain unable to repair the HO DSB (Δhml Δhmr double mutant) was arrested in both G2/M and telophase. We confirmed an increase in the Smc3-bound pool around the HOcs in G2/M after the DSB (Figure 2—figure supplement 3); however, such an increase was not observed in telophase (Figure 2C). We thus conclude that the core of the cohesin SMCs (Smc1–Smc3) is not further recruited to the HO DSB in late mitosis. This lack of local recruitment could be due to several non-exclusive scenarios, including the inability to increase the acSmc3 pools by Eco1, whose levels drop after G2/M (Lyons and Morgan, 2011), as well as condensin binding to chromosome arms during anaphase (Leonard et al., 2015), which might act as an eviction barrier for cohesin loading. However, reassembly of the full cohesin complex at residual Smc1–acSmc3 dimers may shift the organization of the surrounding chromosome regions and impinge on the HO DSB repair in telophase.

In a previous report, we showed from both genetic and cytological points of view that HR was still active in late mitosis (Ayra-Plasencia and Machín, 2019). Having observed that cohesin is reconstituted and chromatin-bound after the HO DSB in telophase, we next wondered about the functional role of the complex in this context.

We first tested whether the HO-induced DSB is efficiently resected into 3′ ssDNA tails in late mitosis. Resection is necessary to form the protruding 3′ Rad51 nucleoprotein filaments that invade the donor homologous sequence to restore the break by HR (Peng et al., 2021; Symington, 2016). To confirm and quantitate the formation of ssDNA flanking the HOcs, we used a qPCR approach whereby primers can only amplify a target DNA cut with the restriction enzyme StyI if the target has been rendered single-stranded by resection (Figure 3—figure supplement 1; Gnügge et al., 2018; Zierhut and Diffley, 2008). We used the Δhml Δhmr double mutant to hamper the normal HR flow from resection to invasion and thus boost ssDNA detection. Two different locations downstream of the HOcs were monitored to determine both the initial resection (0.7 kb from HOcs) and the kinetics further away (5.7 kb). In addition, cells arrested in either G2/M or telophase were directly compared to each other. Cells were first arrested for 3 hr, the HO was induced afterwards, and samples were taken every hour during a 4 hr time course while sustaining the HO induction (Figure 3A). As expected, cells in G2/M efficiently resected the HOcs ends. The time to reach half of the maximum resection (t_1/2) was ~1 hr at 0.7 kb and ~2.5 hr at 5.7 kb from the DSB, respectively. Resection in telophase was only slightly delayed at 0.7 kb (t_1/2 ~ 1.5 hr), whereas the delay was more pronounced at 5.7 kb (t_1/2 ~ 3.5 hr). The delay may be due to either the fact that CDK activity in late mitosis, though still high, is slightly inferior to that at G2/M or the different chromosome compaction degrees between the arrests. Together with the samples for the qPCR, samples for Western blot of Rad53 were taken. Rad53 is an effector kinase of the DDC that gets activated by hyperphosphorylation after being recruited to the 3′ ssDNA overhangs (Branzei and Foiani, 2006). In agreement with the resection profile, Rad53 hyperphosphorylation occurred in late mitosis, although with ~1 h delay (Figure 3B). This delay is also consistent with previous studies on Rad53 activation upon an unrepairable HO DSB (Pellicioli et al., 2001).

Once resection begins, cells are normally committed to recombine because long ssDNA fragments do not work as efficient substrates for NHEJ. In our haploid strain, HR at the MAT locus after the HO-driven DSB implies the gene conversion from the MATa to the MATα allele, whereas repair by NHEJ results in reconstitution of the MATa allele (Figure 3C; Yamaguchi and Haber, 2021). This gene conversion can be detected by Southern blot because of a restriction fragment length polymorphism for StyI. Cells that are in G1 can barely repair the HO DSB, whereas those in G2/M do so through HR (Ira et al., 2004). To check whether HR was efficient at the HO DSB in late mitosis, we performed an experimental setup similar to the one shown above, but with two important differences. On the one hand, the strain retains the HML locus, enabling HR if engaged, and on the other hand, the HO endonuclease was transiently induced by 1 h β-estradiol pulse, allowing time for the complete repair of the induced DSB. Samples were taken at the time of the arrest, after the pulse, and all through the repair window. We confirmed by Western blot that HO (HO tagged with the Flag epitope) is produced after the pulse and is rapidly degraded afterwards (Figure 3D, HO-Flag blot). Rad53 hyperphosphorylation followed the ~1 hr delay shown above and became prominent by the time the DSB inducer had been removed (Figure 3D, Rad53 blot), suggesting extensive resection even in this repairable setup. The Southern blot showed that (1) most MATa locus in the cell population has been efficiently cut after the HO pulse (Figure 3E, F; decrease of the MATa band intensity and rise of that of the cut HOcs); and (2) most DSBs are repaired afterwards exclusively through HR (Figure 3E, F; drop of the HO cut band intensity and rise of that of the MATα). No signs of DSB repair by NHEJ were observed (the remaining MATa band stays constant throughout). This was further confirmed by measuring the repair kinetics of a derivative strain deficient in NHEJ (yku70Δ; Figure 3—figure supplement 2A–C), and is consistent with our previous studies with a HR-deficient rad52Δ strain (Medina-Suárez et al., 2024). Interestingly, HR kinetics were not altered in cells unable to activate the DDC (rad9Δ; Figure 3—figure supplement 2D–F) or to tether the DSB ends (mre11Δ; Figure 3—figure supplement 2G–I). These results are also consistent with previous studies in G2/M cells, in which HR-driven repair of the HO DSB appeared more robust than other types of DSBs (Lazzaro et al., 2008; Moreau et al., 2001).

Overall, we concluded that HR is the chosen repair pathway in late mitosis. This is in agreement with predictions based on CDK activity and also fits well with the genetic and cytological data we have obtained before (Ayra-Plasencia and Machín, 2019; Machín and Ayra-Plasencia, 2020; Medina-Suárez et al., 2024). By contrast, this emphasizes the paradoxes of having HR in the context of mostly segregated sister chromatids and without the cohesin complex holding them together as expected. We have now explored the second paradox further as we have just shown that cohesin is reconstituted after DSBs in telophase, and DI-cohesion appears important for efficient DSB repair by HR (Hou et al., 2022; Phipps and Dubrana, 2022).

In cells arrested in cdc15-2, the Smc1–Smc3 pool mostly resides at centromeres where it brings together adjacent regions in both chromosome arms (Garcia-Luis et al., 2022). Of note, this configuration could favor the physical interaction in cis of the MAT and the HML loci (Piazza et al., 2021), which may in turn promote a rapid and efficient HR-driven MAT switching. Thus, we tested whether MAT switching was dependent on this residual centromeric cohesin. To this aim, we tagged the Smc3 subunit with the auxin-mediated degron system (aid*). To check that the system was working correctly, a serial dilution spot assay was carried out (Figure 4A). The strain expressing Smc3-aid* was plated on YPDA and YPDA plus 8 mM of the auxin indole-acetic acid (IAA). Alongside, we also plated a derivative strain where we had removed the preceptive F-box protein OsTIR1, which forms the functional ubiquitin ligase responsible for targeting the aid* for degradation upon IAA addition (Morawska and Ulrich, 2013). Since cohesin is an essential complex for vegetative growth, the lack of growth in the SMC3-AID* OsTIR1 strain strongly points out that Smc3-aid* is degraded efficiently.

Next, we used this strain to degrade Smc3 after cells have been arrested in late mitosis and before the HO DSB. After IAA addition, there was a clear decline in Smc3-aid* levels, including the chromatin-bound acSmc3-aid* pool (Figure 4—figure supplement 1A). After Smc3 depletion, a 1-hr pulse of HO expression was applied, which allowed us to follow both DSB generation and its subsequent repair without residual cohesin. More than 90% of the Smc3-aid* had disappeared by the time the HO promoter was shut down (Figure 4—figure supplement 1A). The addition of IAA drastically delayed gene conversion from MATa to MATα (Figure 4—figure supplement 1B,C), although the cut efficiency was lower than in the wild type in the absence of auxin (~50% MATa remained uncut in the SMC3:aid* OsTIR1 strain). Also, the detection of DNA damage through Rad53 phosphorylation seems fainter and slightly delayed (Figure 4—figure supplement 1). To unequivocally determine whether MAT switching kinetics were affected by Smc3 depletion, and not by an off-target effect of IAA, we compared strains that could either degrade or retain Smc3 in the presence of IAA. Four isogenic strains were analyzed, comprising all possible combinations of Smc3-aid* and OsTir1, as confirmed by Western blot (Figure 4B; Tel samples). The addition of IAA-induced Smc3-aid* degradation only in the strain co-expressing OsTIR1 (Figure 4B; +IAA samples). In the wild-type strain (no tagged Smc3 and no OsTIR1), the cutting efficiency of HOcs and the yield of gene conversion to MATα were both reduced in the presence of IAA, confirming that IAA itself negatively impacts MAT switching (Figure 4C, compared two leftmost lanes with Figure 3E). However, when the yield of the MAT conversion after the recovery time was normalized to the cut HOcs when HO was repressed, no significant differences were observed among the four strains (Figure 4C, D). We therefore conclude that the core of the cohesin complex is not required for the HR-driven repair of the HO DSB in telophase. Of note, it was previously reported that cohesin was not required for the MAT switching in G2/M-arrested cells, despite it being essential for efficient post-replicative DSB repair (Unal et al., 2004).

In this work, we have addressed whether HR was molecularly efficient in yeast cells arrested in late mitosis. We have used the well-established MAT switching system and found that HR is indeed working in late mitosis, complementing previous results that pointed in this direction both genetically and cytologically (Ayra-Plasencia and Machín, 2019). In addition, we have addressed the paradox of having HR in a cell mitotic stage with no sister chromatid cohesion. Our results suggest that residual cohesin, or at least residual chromatin-bound Smc1–Smc3 dimers, are complemented by incorporating de novo Scc1 (Figure 4E). This fact suggests that chromosome structure may undergo significant changes after late DSBs. The nature of these changes remains unexplored, and we cannot currently rule out that some kind of DI-cohesion could be established, perhaps globally. Alternatively, telophase chromatin may acquire new intramolecular loops as it occurs in G2/M-arrested cells (Piazza et al., 2021). Either of these may favor repair by HR with the intact sister or intramolecularly, although it seems still dispensable for MAT switching at this late cell cycle stage. Further research will be needed to confirm these hypotheses, as well as the whereabouts of cohesin and the consequences for late segregation after DSB repair.

Table 1 contains the S. cerevisiae strains used in this work. Genetic backgrounds are either W303 or YPH499. Strain construction was performed by lithium acetate-based transformation on pre-made frozen competent cells (Knop et al., 1999).

Strains were cultured overnight in the YPD medium (10 g·l⁻¹ yeast extract, 20 g·l⁻¹ peptone, and 20 g·l⁻¹ glucose) with moderate shaking (180 rpm). The day after, 10–100 μl of grown cells were diluted into a flask containing an appropriate YPD volume and grown overnight at 25°C again. Finally, the exponentially growing culture was adjusted to OD₆₀₀ = 0.5 to start the experiment.

In general, cells were first arrested in late mitosis by incubating the culture at 34°C for 3 hr (all strains carry the cdc15-2 thermosensitive allele). Then, the culture was split in two, one subculture remained untreated (mock), and DSBs were generated in the second one. When a G2/M arrest was required, 15 µg·ml⁻¹ nocodazole (Nz; Sigma-Aldrich, M1404) was added to the asynchronous culture, which was then incubated at 25°C for 3 hr (with a 7.5 µg·ml⁻¹ Nz shot at 2 hr). To conditionally degrade the Smc3-aid* variant, 8 mM of the auxin 3-indol-acetic acid (IAA; Sigma-Aldrich, I2886) was added to the arrested cells 1 hr prior to generating DSBs.

DSB generation was accomplished by incubating the culture for 1–3 hr with either 10 µg·ml⁻¹ phleomycin (random DSBs) or 2 μM of β-estradiol (HO-driven DSBs). For the HOcs DSB, the corresponding strain harbors an integrative system based on a β-estradiol inducible promoter (Ottoz et al., 2014). These strains were constructed by transforming with the pRG464 plasmid.

Western blotting was performed as reported before (Ayra-Plasencia and Machín, 2019). The trichloroacetic acid method was used for protein extraction. Total protein was quantified in a Qubit 4 Fluorometer (Thermo Fisher Scientific, Q33227). Proteins were resolved in 7.5% SDS–PAGE gels and transferred to PVDF membranes (Pall Corporation, PVM020C099). The membrane was stained with Ponceau S solution (PanReac AppliChem, A2935) as a loading reference. The following primary antibodies were used for immunoblotting: mouse monoclonal α-HA (1:1000; Sigma-Aldrich, H9658), mouse monoclonal α-myc (1: 5000; Sigma-Aldrich, M4439), mouse monoclonal α-Pgk1 (1:5000; Thermo Fisher Scientific, 22C5D8), mouse monoclonal α-miniaid (1:500; MBL, M214-3), mouse monoclonal α-Flag (1:5000; Sigma-Aldrich, F3165), mouse monoclonal α-Rad53 (1:1000; Abcam, ab166859), and mouse monoclonal α-acSmc3 (1:5000; a gift from Katsuhiko Shirahige). The secondary antibody was a horseradish peroxidase polyclonal goat anti-mouse (from 1:5000 to 1:10,000, depending on the primary antibody; Promega, W4021). Proteins were detected with the ECL chemiluminescence reagent (GE Healthcare, RPN2232), and visualized in a Vilber-Lourmat Fusion Solo S chamber.

For co-IP assays, approximately 100 OD₆₀₀ units from the corresponding conditions were collected. Cell disruption was performed by resuspending the frozen pellets in 200 μl of Buffer A (50 mM HEPES-Na, 150 mM KCl, 1.5 mM MgCl₂, 0.5 mM DTT, 0.5% Triton X-100, 1× EDTA-free protease inhibitor cocktail) together with glass beads. Samples were vortexed vigorously for 5 min, and additional lysis cycles were performed as needed, allowing 1 min incubation on ice between rounds. Lysates were transferred to a clean tube. 400  μl of Buffer A was used to wash the beads and pooled with the lysates. Samples were then centrifuged (12,000 rpm, 5  min, 4°C), and the clarified lysates were transferred to new tubes. An aliquot of 70 μl was saved as whole cell extract for Western blot analysis. The remaining lysate was incubated with pre-equilibrated Pierce Anti-HA Magnetic Beads (Thermo Fisher Scientific; 88836) for 2 hr at 4°C on a rotating wheel. After incubation, beads were placed on a magnetic rack and washed five times with 200 μl of IPP150 buffer (10 mM Tris-Cl pH 7.5, 150 mM NaCl, 0.5% Triton X-100).

For elution, beads were resuspended in 25 μl of 1× SR buffer (8% SDS, 0.5 M Tris-Cl pH 6.8) and incubated at 37°C for 4  min. The supernatant was collected using a magnetic rack and transferred to a new tube containing 8.3 μl of 4× SS buffer (20% sucrose, 0.05% bromophenol blue, 0.1% sodium azide) supplemented with β-mercaptoethanol. Samples were boiled for 2 min at 95°C, centrifuged (max speed, 30 s), and analyzed by SDS–PAGE and Western blotting.

Biochemical fractionation of cells was performed as described before with little modifications (Cuevas-Bermúdez et al., 2020). Briefly, cells were collected and resuspended in 400 µl of buffer 1 (20 mM HEPES pH 8, 60 mM KCl, 15 mM NaCl, 10 mM MgCl₂, 1 mM CaCl₂, 0.8% Triton X-100, 1.25 M sucrose, 0.5 mM spermine). Breakage of cell walls was done by adding ~200 mg of glass beads and vortexing for 4 min. After centrifuging at 500 × g for 5 min, supernatant was transferred to a new tube and 40 µl were kept as Input. Samples were centrifuged at 18,000 × g for 20 min, the supernatant was kept as S1, whereas the pellet was resuspended in 200 µl of Buffer 1 (50 µl were saved as P1). Samples were centrifuged again at 18,000 × g for 20 min, the supernatant was kept as S2, and the pellet was resuspended in 200 µl of buffer 2 (20 mM HEPES pH 7.6, 45 mM NaCl, 7.5 mM MgCl₂, 20 mM EDTA pH 8, 10% glycerol, 1% NP-40, 2 M urea, 0.5 M sucrose, protease inhibitors) and 50 µl were saved as P2. Finally, samples were centrifuged one last time at the same conditions, keeping the supernatant as S3, and the pellet was resuspended in 50 µl of Laemmli buffer, constituting the chromatin-enriched fraction (P3).

For ChIP–qPCR, a total of 100 OD₆₀₀ units of the treated cultures were collected and crosslinked with 1.4% formaldehyde for 15 min. Crosslinking was quenched with glycine (final concentration 125 mM) for 7 min. Cells were harvested by centrifugation (4000 rpm, 1 min), washed with PBS, transferred to screw-cap tubes, and frozen at –80°C.

Frozen pellets were resuspended in 300 μl of IP buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.5, 5 mM EDTA, 1% Triton X-100, 0.05% NP-40) supplemented with 1 mM PMSF and protease inhibitor cocktail (EDTA-free, Roche). Disruption was performed by bead beating with 500 μl of glass beads using a Bertin Precellys homogenizer for nine 20 s cycles at hard setting. Lysates were transferred to new tubes and supplemented with an additional 100 μl of IP buffer containing PMSF and protease inhibitors. Samples were clarified by centrifugation (15,000 rpm, 10 min, 4°C). Pellets were resuspended in 1 ml IP buffer with PMSF and protease inhibitors and sonicated in a Diagenode Bioruptor Plus (30 cycles: 30 s ON 30 s OFF, high power, 4°C). After sonication, samples were centrifuged (15,000 rpm, 10 min), and the supernatant was collected. A 200-μl aliquot was reserved as input. Input DNA was precipitated with 0.3 M sodium acetate and 2.5 volumes of cold ethanol, centrifuged (15,000 rpm, 30 min), washed with 70% ethanol, and air-dried.

For immunoprecipitation, 400 μl of the sonicated chromatin was incubated with 40 μg of anti-HA antibody (Roche, 11666606001) in the Bioruptor at low power for 30 min (30 s ON/30 s OFF cycles). Antibody–chromatin complexes were pelleted (13,000 rpm, 5 min), and the supernatant was incubated with 60  μl of Dynabeads Protein G (Invitrogen), pre-equilibrated in IP buffer, then incubated for 2 hr at 4°C on a rotating wheel, and finally washed five times with IP buffer using a magnetic rack.

Both input and IP samples were resuspended in de-crosslinking buffer (1× TE, 1% SDS, 10 μg·ml^–1 RNase A, 1 mg·ml^–1 proteinase K) and incubated overnight at 65°C and purified using High Pure PCR product purification kit (Roche) for the qPCR.

In the resection experiments, qPCR was performed on genomic DNA (gDNA) extracted by the glass beads/phenol method (Hoffman and Winston, 1987). Experimental details on the resection assay can be found in Gnügge et al., 2018. Briefly, each gDNA sample was divided into two, and one aliquot was digested with StyI-HF (NEB, R3500S). Then, qPCR reactions were mounted using the PowerUp SYBR Green Master Mix (Thermo Scientific, A25741) and run in a Bio-Rad CFX384 Real-Time PCR instrument (10 µl final volume in 384-well block plates).

Resection was calculated as a normalized fraction (f_resected) to that of the HOcs effectively cut by HO (f). These two parameters were calculated by their corresponding formulas (Gnügge et al., 2018).

In this case, gDNA was extracted by digesting with 50 U lyticase (Sigma-Aldrich, L4025) a cell pellet previously resuspended in 200 µl of digestion buffer (1% SDS, 100 mM NaCl, 50 mM Tris-HCl, and 10 mM EDTA). Then, the gDNA was phase separated by phenol:chloroform (PanReac AppliChem, A0944), precipitated with ethanol, resuspended in TE 1X with 10 μg·ml^–1 RNase A, precipitated again, and resuspended in TE 1X. Then, the gDNA was digested with StyI, the restriction fragments separated on a 1.2% low EEOO LS Agarose gel, and finally Southern blotted. Southern blot was carried out by a saline downwards transference onto a positively charged nylon membrane (Hybond-N+, Amersham-GE; RPN303B) (García-Luis and Machín, 2014). DNA probes against ACT1 and MAT loci were made using Fluorescein-12-dUTP Solution (ThermoFisher; R0101) and the Expand High Fidelity PCR System (Roche; 11732641001). Hybridization with fluorescein-labeled probes was performed overnight at 68°C. The following day, the membrane was incubated with an anti-fluorescein antibody coupled to alkaline phosphatase (Roche; 11426338910), and the chemiluminescent signal detected using CDP-star (Amersham; RPN3682). Blots were visualized in a Vilber-Lourmat Fusion Solo S chamber.

For band quantification, each individual band was normalized to the ACT1 signal in the lane. Then, a second normalization to the MATa band at the arrest was performed. To determine the yield of gene conversion to MATα specifically, the MATα band after recovery from the DSB (2–3 hr after β-estradiol removal) was normalized to the amount of the cut HOcs band immediately before recovery (1 hr after β-estradiol addition).

Three types of graphs were used to represent the data: bar charts, marker line graphs, and box plots. In box plots, the center line represents the medians, box limits represent the 25th and 75th percentiles, the whiskers extend to the 5th and 95th percentiles, and the dots represent outliers. Error bars in all bar and line charts represent the SEM of three independent biological replicates. Individual values are also represented as dots in the bar charts. GraphPad Prism 9 was used for generating the charts and for statistical analysis. Differences between experimental data points were generally estimated using the Mann–Whitney U test for box plots and one-way ANOVA with Tukey post hoc for bar charts.