Harmful R-loops are prevented via different cell cycle-specific mechanisms

ARTICLE Harmful R-loops are prevented via different cell cycle-specific mechanisms Marta San Martin-Alonso1, María E. Soler-Oliva 1, María García-Rubio1, Tatiana García-Muse 1✉& Andrés Aguilera 1✉ Identifying how R-loops are generated is crucial to know how transcription compromises genome integrity. We show by genome-wide analysis of conditional yeast mutants that the THO transcription complex, prevents R-loop formation in G1 and S-phase, whereas the Sen1 DNA-RNA helicase prevents them only in S-phase. Interestingly, damage accumulates asymmetrically downstream of the replication fork in sen1 cells but symmetrically in the hpr1 THO mutant. Our results indicate that: R-loops form co-transcriptionally independently of DNA replication; that THO is a general and cell-cycle independent safeguard against R-loops, and that Sen1, in contrast to previously believed, is an S-phase-specific R-loop resolvase. These conclusions have important implications for the mechanism of R-loop formation and the role of other factors reported to affect on R-loop homeostasis. https://doi.org/10.1038/s41467-021-24737-x OPEN 1Centro Andaluz de Biología Molecular y Medicina Regenerativa CABIMER, Universidad de Sevilla-CSIC-UPO, Seville, Spain. ✉email: tatiana.muse@cabimer. es;[email protected] NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecommunications 1 1234567890():,; R-loops, formed by a DNA−RNA hybrid and the displaced single-stranded DNA (ssDNA), are a natural source of genome instability1–3. Numerous factors have emerged that protect cells from harmful R-loops. They can regulate R-loop levels by preventing their formation, a paradigmatic example being the THO complex, or by resolving them via RNA cleavage or DNA−RNA unwinding, as is the case of RNH1 and Sen1/ SETX, respectively2,3. THO is critical for the formation of optimal mRNPs during transcription elongation and functions at the RNAPII-transcribed regions4,5. In yeast, this protein complex is formed by five interacting subunits: Tho2, Hpr1, Mft1, Thp2, and Tex16–8. Their mutations cause transcription elongation and RNA export defects, and increase recombination and DNA damage in association with high levels of R-loops in yeast9,10 and human cells11. Although evidence suggests that the main mechanism by which THO protects cells from R-loops relies on promoting an optimal assembly of an export-competent proteincoated mRNA, we now know that THO physically interacts with the Sin3A histone deacetylase and the UAP56/DDX39B DNA −RNA helicase, both of which also protect cells from R-loop accumulation11,12. Thus, THO may coordinate a complex cotranscriptional mechanism of DNA protection that relies on RNA packaging, localized chromatin compaction, and DNA-RNA unwinding. On the other hand, yeast Sen1 is an essential DNA −RNA helicase that participates in transcription termination of non-polyadenylated RNAs, including snoRNAs, snRNAs, and cryptic unstable transcripts13.sen1 mutants display high recombination rates associated with R-loops14. Interestingly, the human ortholog of Sen1, senataxin (SETX), a protein linked to DNA repair and neurodegeneration15, functions in transcription termination and R-loop control in association with BRCA116,17. Its DNA-RNA unwinding ability is required to remove hybrids formed at double-strand breaks (DSBs) to facilitate homologousdependent repair (HDR)18. Despite the different factors protecting cells from harmful R- loops, the mechanism by which these are formed and cause genome instability is still unclear. Evidence indicates that nucleases such as XPG and XPF can cleave the ssDNA of an R- loop19, but the main mechanism by which this leads to DNA damage is by blocking replication fork (RF) progression20. This has been shown in artificial systems10,21, and in yeast and human cells depleted of R-loop suppressor factors such as THO, Sin3A, Fanconi Anemia, Sen1/SETX, the UAP56/DDX39B helicase, or the bromodomain chromatin factor BRD45,11,12,22–24. Altogether, the data supports that R-loops associated with transcriptionreplication (T-R) conflicts are a source of DNA breaks, complying with the fact that transcription interferes with RF progression25 and promotes genome instability20,26,27. We showed in the past that T-R collisions cause genome instability in head-on (HO) but not in co-directional (CD) orientation in yeast cells28. A study in human cells29, extended the conclusions to an accumulation of R- loops seen at HO but not CD conflicts, leading to a model by which R-loops are formed at HO T-R conflicts. However, R-loops are detected in both HO and CD collisions when they are stabilized by an RNA binding factor, as it is the case of yeast Yra130. Moreover, a recent genome-wide analysis shows no difference in R-loop signals at HO versus CD conflicts31. Identifying whether R-loops are caused by T-R conflicts or vice versa is crucial not only to decipher the mechanisms of R-loop formation, but also to know how R-loops compromise genome integrity. This acquires additional relevance after noticing that DNA breaks facilitate DNA−RNA hybrid accumulation32, opening the possibility that breaks occurring at T-R conflicts could contribute to spontaneous R-loops. Here we demonstrate that R-loop formation is a co-transcriptional event that can occur independently of replication and that THO is a general safeguard against R-loop accumulation along the cell cycle. In contrast to previously believed, Sen1 is not a global DNA−RNA hybrid resolvase, but functions to remove DNA−RNA hybrids formed at T-R collision sites during S-phase. These conclusions have important implications for the mechanisms by which R-loop are originated and cause genome instability and invite to revisit our view on the role of factors reported to impact on R-loop homeostasis. Results Rapid harmful R-loop accumulation after Hpr1 and Sen1 depletion. Aiming at deciphering the mechanisms by which cells minimize R-loop and/or DNA−RNA hybrid accumulation, we generated hpr1 and sen1 conditional mutants with the auxininducible degron (AID) system, hpr1-aid, and sen1-aid. Westernblot analysis of cell extracts at different time points after auxin addition, showed elimination of Hpr1 (after 60 min) and Sen1 (after 30 min) proteins (Fig. 1a). Serial dilution assays revealed that, in contrast to wild-type (WT), hpr1Δ, and sen1-1 cells, both hpr1-aid and sen1-aid strains displayed a growth defect upon prompt Hpr1 or Sen1 depletion in the presence of 1 mM of auxin, a defect that was exacerbated in 2.5 mM (Supplementary Fig. 1a). To know whether the two conditional mutants exhibited the genetic instability described for hpr1Δand sen1-14,10,14,we measured Rad52-YFP foci as a readout of global DNA breaks33. Whereas auxin addition did not affect the percentage of WT cells showing Rad52 foci, both Hpr1 and Sen1 depletion significantly increased such a percentage (Fig. 1b and Supplementary Fig. 1b). Importantly, overexpression of RNase H1 fully suppressed this increase in Rad52 foci-containing cells after Hpr1 and Sen1 removal (Fig. 1b). Next, we assayed the effect on recombination using the L-LacZ system, consisting of direct repeats of two leu2 alleles flanking the bacterial lacZ ORF that lead to Leu+ recombinants by single-strand annealing (SSA). Leu+recombination frequencies increased in hpr1-aid and sen1-aid strains 20- and 10-fold after 1 mM auxin addition and 185- and 61-fold after 2.5 mM addition compared to the WT, which showed no significant changes with and without auxin (Fig. 1c). Overexpression of RNase H1 resulted in a significant reduction of this phenotype in hpr1-aid and sen1-aid strains (Fig. 1d), confirming that hyper-recombination relied on R-loops9,14. Next, we assessed whether this increase in DNA damage was linked to an increase in R-loops. This was assayed using the S9.6 antibody that recognizes DNA−RNA hybrids. Even though S9.6 detects DNA−RNA hybrids, regardless of whether or not part of an R-loop, studies with bisulfite mutagenesis suggest that in most cases hybrids correlate with R-loops34–37. Thus, here we will refer indistinctly to R-loops, even though in some cases the structure could be just a DNA−RNA hybrid. We performed immunofluorescence (IF) of chromosome spreads, to gain a broader view of DNA−RNA hybrid accumulation, and DNA−RNA hybrid immunoprecipitation (DRIP) to study the R-loop accumulation at specific genomic regions. First, chromosome spreads in asynchronous cultures after auxin-induced depletion revealed that cells with hybrids were 5% in the wild-type, but increased to 16% in hpr1-aid and to 19% in sen1-aid (Fig. 1e and Supplementary Fig. 1c). Both signal values were significantly reduced by in vivo RNase H1 overexpression (Fig. 1e and Supplementary Fig. 1c). Consistently, DRIP analysis after Hpr1 or Sen1 depletion in asynchronous cultures showed that both conditional mutants accumulated hybrids compared to WT at the GCN4 and PDR5 regions analyzed. The DRIP values were fully suppressed by in vitro RNase H1 treatment, confirming the specific detection of DNA−RNA hybrids (Fig. 1f and Supplementary Fig. 1d). ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x 2NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecom munications Altogether, these results prove that both proteins exert a direct role in preventing R-loop accumulation and R-loop-mediated recombinogenic DNA damage. Hpr1 and Sen1 depletion impairs replication and S-phase progression. We next analyzed whether the growth defect caused by sen1 and hpr1 depletion was linked to a possible defect in cell cycle progression. Thus, we monitored cell cycle progressioninWT,hpr1-aid, and sen1-aid cells in the presence or absence of auxin by fluorescence-activated cell sorting analyses (FACS) (Fig. 2a). Cells were grown up to an exponential phase, treated or not with auxin (to deplete Hpr1 or Sen1) and α-factor (to synchronize the cells in G1-phase) for 2 h, and then ac LEU2 Prm leu2∆3’ leu2∆5’LacZ Fold change hpr1-aid WT sen1-aid 1 10 100 1000 10000 x10.5 x20.8 x1 1.64 2.41 16.3 1.76 1.56 -aux +aux -/+ 1 mM aux Recombination frequency (x10-3) Recombination frequency (x10-3) d e f 0.05 0.10 0.15 0.20 S9.6 si g nal ( arb.units. ) 0 0.05 0.10 0.15 0.20 hpr1-aid sen1-aid WT RNH - + - + - + S9.6 si g nal ( arb.units. ) hpr1-aid sen1-aid WT RNH - + - + 0 Rad52-YFP +DAPI hpr1-aid sen1-aid Rad52-YFP +aux auxaux - +aux RNH - +RNH RNH - +RNH +DAPI b hpr1-aid sen1-aid aux RNH - + - + - + - + - - + + - - + + -/+ 2.5 mM aux 260 86 1.4 1.23 2.42 1.64 x61.5 x185.8 x1 Fold change 1 10 100 1000 10000 Fold change 1 10 100 1000 10000 x1 x100.5 x272.7 2.5 mM aux x1 x145.7 x28.5 43.6 484 223 1.77 1.53 hpr1-aid WT sen1-aid - RNH + RNH 178 Recombination frequency (x10-3) hpr1-aid WT sen1-aid p=0.04 p=0.03 p=0.04 p=0.04 p=0.008 p=0.03 p=0.008 p=0.03 p=0.005 p=0.001 p=0.017 p=0.03 p=0.042 p=0.043 p=0.047 p=0.020 p=0.038 p=0.019 p=0.001 p=0.004 p=0.002 p=0.005 p=0.001 p=0.046 p=0.017 - + GCN4 PDR5 0 5 10 15 20 % S-G2 Cells with Rad52-YFP foci -aux +aux 63 Protein levels (arb.units.) 0 30 60 90 0.0 0.5 1.0 1.5 2.0 Time after aux addition (min) Hpr1 Sen1 hpr1-aid sen1-aid WT RNH -+ -+ -+ 0 % Nuclei with detectable DNA:RNA hybrids p=0.026 p=0.0012 p=0.008 p=0.002 10 20 30 hpr1-aid Hpr1::aid -9myc TIR1 0306090 t (min) sen1-aid Sen1::aid -9myc 0 30 60 90 TIR1 t (min) 95 130 kDa 95 250 NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x ARTICLE NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecommunications 3 washed and released into fresh medium with auxin and without α-factor. Wild-type and conditional mutants progressed through the cell cycle similarly under control conditions in the absence of auxin. On the contrary, upon Hpr1 depletion by auxin, hpr1- aid cells showed a clear delay in reaching the S/G2 phase. A similar delay in reaching the S/G2 phase was observed for the sen1-aid cells after Sen1-depletion compared to WT (Fig. 2a), consistent with previous data22. These delays in S-phase progression could be explained by a defect in replication. To test this possibility, we performed ChIP analyses of cells treated with BrdU, a thymidine analog that can be incorporated into replicating DNA and whose incorporation can be monitored by ChIP in properly modified yeast strains (Fig. 2b). We analyzed four regions where replication collides with transcription in a HO orientation, as validated in previous analysis38, so that we could assay a proximal and a distal sequence with respect to each replication origin (ARS). A delay in BrdU incorporation was observed at the four regions tested in Hpr1- depleted cells (+aux) compared to the WT condition (−aux), demonstrating that replication is hampered in the absence of Hpr1 (Fig. 2b). In the case of sen1-aid, BrdU was incorporated in the two ARS proximal regions (1 and 3) with similar efficiency at early time points. However, in contrast to the WT condition (−aux) in which BrdU signals decrease, BrdU incorporation increases until it reaches a sustained plateau 10 min later (Fig. 2b). Instead, BrdU incorporation at the two ARS distal regions (2 and 4) was clearly delayed and less efficient (Fig. 2b). Altogether, these results demonstrate that depletion of either Hpr1 or Sen1 leads to replication impairment, but it seems to be of different nature in each case. Differential R-loop accumulation through the cell cycle. Since a rapid depletion of Hpr1 or Sen1 leads to R-loop-dependent genetic instability and replication impairment of different nature we asked whether R-loops are accumulated differently during the cell cycle, which we monitored by FACS analyses (Supplementary Fig. 2a). First, we performed chromosome spreads and DRIP experiments in G1-synchronized cultures. Notably, whereas chromosome spreads of hpr1-aid cells showed a clear and statistically significant increase in the number of cells with S9.6 signal (approx. 4-fold), sen1-aid cells showed no significant differences with respect to WT (Fig. 3a and Supplementary Fig. 2b). DRIP analyses confirmed this result. Whereas Hpr1 depletion in G1-arrested cells increased R-loops with respect to the WT, Sen1 depletion did not (Fig. 3b and Supplementary Fig. 2c). Second, we determined R-loop accumulation in S-phase. For this, hpr1-aid and sen1-aid G1-arrested cells were released in fresh media to allow proliferation, and samples were collected in S-phase. As observed, the change of cell-cycle stage did not affect the basal S9.6 signal in WT cells. As in G1, depletion of Hpr1 exhibited a significant increase of R-loops compared to WT, as detected by both IF of chromosome spreads and DRIP analysis (Fig. 3c, d and Supplementary Fig. 2b). Importantly, Sen1 depletion increased the percentage of cells with R-loop signal (approx. 7-fold), as determined by chromosome spreads (Fig. 3c and Supplementary Fig. 2b). This result was corroborated by DRIP-qPCR analysis, which showed that Sen1 depletion led to a significant DNA−RNA hybrid accumulation at the analyzed regions GCN4 and PDR5 (Fig. 3d). These data unambiguously indicate that DNA−RNA hybrids increase during the S-phase upon Sen1 depletion but not during G1, whereas they increase during both G1 and S-phases upon Hpr1 depletion. This correlates with the observation that mRNA and protein levels of Hpr1 are detected throughout the cell cycle (Supplementary Fig. 2d, e). We asked next whether DNA−RNA hybrids formed in S-phase, as is the case of sen1-aid cells, could pass unresolved to the following G1-phase. To test this, we performed a new experiment in which G1-synchronized cells were allowed to complete one cell cycle, and then we measured R- loops in the following G1-phase (hereafter called G1 2 ). As observed in Fig. 3e, R-loop levels detected by DRIP in hpr1-aid and sen1-aid cells in the two regions studied, were higher than in the WT in G1 2 . Thus, R-loops formed at one stage of the cell cycle could pass unresolved to another stage and to the next cell cycle. R-loop-accumulating mutants show increased DNA damage through the cell cycle. Although removal of Hpr1 or Sen1 leads to R-loop accumulation in a cell cycle phase-specific manner, both conditional mutants accumulate Rad52 foci. Therefore, we examined how R-loops associate with DNA damage in relation to cell cycle phases using phosphorylation of histone H2A at S129 (H2AP mark) as a readout of DNA damage39.Wefirst assayed H2AP levels by western blot, using protein extracts obtained from cultures in G1, S, and G1 2 -phases. In G1, only Hpr1 depletion exhibited elevated H2AP levels, whereas in S-phase depletion of either Hpr1 or Sen1 increased H2AP levels above WT (Fig. 4a, b). Again, in the G1 2 phase, only Hpr1 and not Sen1 depletion showed an increase of H2AP with respect to WT (Fig. 4a, b). To assay whether the high DNA damage detected was linked to R-loops, we assessed the presence of H2AP by ChIP in S-phase cultures overexpressing or not RNase H1. Rapid depletion of both Hpr1 and Sen1 significantly increased the H2AP signal at the GCN4 and PDC1 genes. Importantly, the signal was reduced, even though slightly, upon RNase H1 overexpression (Fig. 4c). Therefore, since R-loops formed in G1 or S lead to DNA damage, the ability of a harmful R-loop to result in DNA damage is independent on the cell cycle stage at which it is formed. Loss of Hpr1 and Sen1 lead to differential genome-wide DNA −RNA hybrid accumulation in S-phase. Next, we wondered how R-loops and DNA damage association occurs genome-wide, Fig. 1 Rapid Hpr1 or Sen1 depletion leads to R-loop-dependent genomic instability. a Immunoblotting with αmyc after 0, 30, 60, or 90 min of 1 mM auxin treatment of hpr1-aid and sen1-aid cells (with aid-degron and myc translational fusions to the HPR1 and SEN1 genes, respectively). AtTIR1-9Myc was used as loading control. Quantification (mean) of protein levels normalized to TIR1 (n=2 biologically independent samples). bRepresentative images and value of Rad52-YFP foci in hpr1-aid and sen1-aid strains with or without RNH1 and with or without auxin. Graph shows mean and SEM of independent experiments (n=3). A minimum of 100 cells were counted per experiment. cScheme of the L-lacZ direct-repeat recombination system. Recombination analysis in WT, hpr1-aid, and sen1-aid with and without the indicated auxin concentrations (n=3 biologically independent experiments) dRecombination analysis in strains as in (c) with or without RNH1 in the presence of auxin (n=3 biologically independent experiments). In (c,d), average and SEM are shown. ePercentage of positive DNA−RNA hybrids nuclei in chromosome spreads stained with the S9.6 antibody in WT, hpr1-aid, and sen1-aid asynchronous cells with or without RNH1. Data are presented as mean values +/−SEM (n=100 cells examined over three independent experiments). fDRIP analysis at GCN4 and PDR5 genes in cells as in (e), treated (+) or not (−) with RNH1 in vitro. Data are presented as mean values +/−SEM (n=4 biologically independent experiments). The Pvalues were calculated by the two-tailed unpaired Student t-test. Scale bar 5 µm. Data underlying this figure are provided as Source data file. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x 4NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecom munications hpr1-aid 1C 2C 1C 2C 0 10 15 20 25 30 35 40 50 t (min) 60 0 10 15 20 25 30 40 50 t (min) 1C 2C 1C 2C sen1-aid 1C 2C 1C 2C 0 10 15 20 25 30 35 40 50 asyn t (min) 60 WT SPF1 reg 1 reg 2 reg 3 reg 4 PDC1 TRX1 a b ARS508 ARS1211 35 -aux +aux asyn asyn -aux +aux +aux-aux reg 1 p=0.0349 sen1-aid -aux +aux 0 10 20 30 40 50 0 10 20 30 40 Time after G1 release ( min ) reg 3 p=0.0024 sen1-aid -aux +aux 0 10 20 30 40 50 0 20 40 60 80 100 Time after G1 release ( min ) reg 2 sen1-aid -aux +aux 0 10 20 30 40 50 0 5 10 15 20 25 Time after G1 release (min) Time after G1 release (min) reg 4 sen1-aid -aux +aux 0 10 20 30 40 50 0 5 10 15 20 25 hpr1-aid -aux +aux 0 10 20 30 40 50 0 2 4 6 8 10 Time after G1 release (min) reg 3 p=0.0001 hpr1-aid -aux +aux 0 2 4 6 8 10 Time after G1 release (min) 0 10 20 30 40 50 reg 2 p=0.0034 hpr1-aid -aux +aux 0 2 4 6 8 10 Time after G1 release ( min ) 0 10 20 30 40 50 reg 1 p=0.0002 BrdU-IP signal (arb.units.)BrdU-IP signal (arb.units.)BrdU-IP signal (arb.units.)BrdU-IP signal (arb.units.) BrdU-IP signal (arb.units.)BrdU-IP signal (arb.units.)BrdU-IP signal (arb.units.) 0 10 20 30 40 50 0 2 4 6 8 10 Time after G1 release (min) p=0.0002 reg 4 hpr1-aid -aux +aux BrdU-IP signal (arb.units.) Fig. 2 Quick depletion of Hpr1 or Sen1 impairs replication. a Analysis of cell cycle progression after alpha factor release of WT, hpr1-aid, and sen1-aid strains with and without auxin by FACS. bChIP analysis of BrdU incorporation into the DNA at the genomic regions and amplicons depicted in the schemes (top) in the hpr1-aid and sen1-aid degron strains with and without auxin. Mean +/−SEM is shown (n=3 biologically independent experiments). The Pvalues calculated by the two-tailed Wilcoxon signed-rank test are shown for each condition. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x ARTICLE NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecommunications 5 for which we focused on S-phase at which DNA−RNA hybrids can be detected in both mutants. To obtain the genome-wide R- loop distribution after depletion of Hpr1 or Sen1 in S-phase synchronized cells, we conducted DRIPc-seq40. We implemented this method by adding an RNase III treatment prior to S9.6 immunoprecipitation that permits the elimination of dsRNA, which is also recognized by S9.641,42. Genomic DNA−RNA hybrid profiles of both hpr1-aid and sen1-aid cells in S-phase show regions with DNA−RNA hybrids, and the overexpression of RNase H leads to undetectable levels of DNA−RNA hybrids establishing that the immunoprecipitated material corresponded to DNA−RNA hybrids (Supplementary Fig. 3a). Two replicates of DRIPc-seq were performed and were highly correlated with each other (Supplementary Fig. 3b). The genome-wide coverage of R-loop peaks was obtained using chromstaR43. Two representative genomic regions are shown in Fig. 5a. DNA−RNA hybrids accumulate along the whole genome in the WT and two conditional mutants, but both hpr1-aid and sen1-aid cells exhibit a peak enrichment (referred to peaks with higher levels than WT, whether or not pre-existing) of 5.7- and 4.3- times over the WT, respectively (Supplementary Fig. 3c). The data indicate that Hpr1 and Sen1 depletion causes a general R-loop- enrichment genome-wide. Accordingly, metaplot analyses of all enriched peaks show a higher DRIPc-seq signal for hpr1-aid and sen1-aid strains with respect to WT (Fig. 5b). Analysis of DRIPcseq signal locations revealed that 85% of hpr1-aid and 78% of sen1- aid DRIPc-seq peaks coincide with protein-coding genes (Supplementary Fig. 4a), in agreement with a role of both proteins in RNAPII-driven transcription4,13. Comparison of R-loops formed by sense and anti-sense transcripts reveals that DNA−RNA hybrid accumulates in both strands at gene bodies, but with a progressive increase towards the 3′region with a peak at the transcription termination site (TTS) in the sense strand that is not observed in the anti-sense orientation apart form a sharp peak coincident at the TTS (Fig. 5c). It is worth noting that genes showing increased DNA−RNA hybrids in the mutants are longer and expressed at higher levels than the genome average (Supplementary Fig. 3d). In addition, metaplot analysis reveals a specific increase over the WT at telomeres in hrp1-aid mutant not observed in sen1-aid (Fig. 5d), whereas at tRNAs and snoRNAs the increase is observed in sen1- aid and not in hpr1-aid (Fig. 5e and Supplementary Fig. 4a), consistent with their previously reported functions44,45.In addition, the volcano plot of hpr1-aid and sen1-aid peaks reveals a 17.3 and 23% of specific peaks, respectively (Supplementary Fig. 3c). Accordingly, the metaplot analysis of the peaks specificto hpr1 and sen1 shows higher DRIPc-seq signals (Supplementary Fig. 4b) and, in both mutants, shows a clear enrichment at gene bodies of both sense and anti-sense signals (Fig. 5f). For common peaks, the increase is similar for both mutants and also observed preferentially at gene bodies (Fig. 5f and Supplementary Fig. 4b). Altogether, these genomic data indicate that both proteins have a global but distinct role in R-loop control. Depletion of Hpr1 in G1-arrested cells increased R-loops with respect to the WT; therefore, we assessed if the S-phase DNA−RNA hybrids observed in hpr1-aid cells were originated in G1 or de novo. DRIPc-seq in WT and Hpr1-depleted G1-synchronized cultures revealed a genome-wide coverage of DNA−RNA hybrid peaks in both strains along the whole genome. Representative regions showing DRIPc-seq peaks of both DNA strands in G1 in hpr1-aid and WT (Supplementary Fig. 5a) or the joined data of both strands compared between G1 and S-phase are shown (Supplementary Fig. 5b). Consistent with the DRIP-qPCR results, R-loops were differentially enriched in the PDC1,PDR5, and GCN4 regions in G1 and S phase (SupplementaryFig.5c).Thereisa2.7-foldenrichmentofG1peaks in hpr1-aid cells over the WT (Supplementary Fig. 5d), pointing out to a widespread R-loop-enrichment also in G1 phase. Interestingly, after Hpr1 depletion, 75% of total peaks (1053 out of 1462) detected in G1-phase are observed in S-phase (Supplementary Fig. 5e, left), but 90% (347 out of 387) of the G1 hpr1-aid R-loop-gain peaks (those significantly enriched in G1 over the WT), are not enriched in S-phase (Supplementary Fig. 5e, right). Instead, 2051 new R-loop- gain peaks are detected in S-phase. Notably, metaplot analysis of all these peaks shows that hpr1-aid R-loop-gain peaks detected in G1 (clearly enriched in G1 over the WT in S-phase) either disappear (G1-specific, Supplementary Fig. 5f) or, if still enriched, the levels are much lower (G1-S common) (Supplementary Fig. 5g). Importantly, the level of hpr1-aid R-loop-gain peaks detected in S-phase (Supplementary Fig. 5h) is much lower than that of G1-enriched peaks in G1-phase (compare Supplementary Fig. 5h, bottom with Supplementary Fig. 5f, g, top). Altogether, the data indicate that Hpr1 protects from R-loops at both G1 and S phase, but its effect is much stronger during G1 phase. S-phase R-loops correlate with specific sites of DNA damage. Provided the differential impact on R-loop accumulation at G1 and S-phase and at distinct genomic regions, we wondered how this different R-loop control affects DNA damage. We extended the S-phase genome-wide analysis with a ChIP-seq of H2AP. We found a background level of H2AP along the genome in WT cells consistent with S-phase replication being the most vulnerable DNA process and therefore prone to DNA breaks (Supplementary Fig. 6a). Such enrichment of H2AP was also observed in both hpr1-aid and sen1-aid strains, with signal intensities at specific sites clearly above WT levels, sen1-aid exhibiting the highest signals. When subtracting the WT signals from those from each conditional mutant to focus on the signals specifically enriched after Hpr1 or Sen1 depletion, a differential group of regions sensitive to H2AP accumulation in both mutants is observed (Supplementary Fig. 6b). Next, we did a bioinformatic comparison between the S-phase DNA−RNA hybrid (DRIPc-seq) and H2AP (ChIP-seq) genomewide profiles for each strain considering the H2AP signal average of the 5 kb region around each R-loop peak. While the R-loops of WT cells show low H2AP signal in those 5 kb regions, both mutants show a significant H2AP accumulation spreading from the DNA−RNA hybrid sites (Fig. 6a, b). When we restricted the analysis of the H2AP profile to regions surrounding enriched DRIPc-seq peaks for each mutant, we found that in both hpr1-aid and sen1-aid mutants H2AP occupancy was higher than in the WT, as expected (Fig. 6c and Supplementary Fig. 6c). Remarkably, whereas this occupancy was symmetric toward each side of the R-loop peak in hpr1-aid, it was asymmetric in sen1-aid (Fig. 6c), a feature that can be seen even better around the sen1- aid specific R-loop sites (Supplementary Fig. 6c). Since it has been previously observed in triple mutants rnh1 rnh2 sen1 that persistent DNA−RNA hybrids show asymmetric recruitment of the DSB repair factor Rad52, we extended our comparative analysis to the available data from Rad52 ChIP- seq46. Metaplot analysis of Rad52 recruitment to DRIPc-seq peaks in WT cells did not show the pattern of H2AP, suggesting that only a fraction of sites detected by H2AP in WT cells may correspond to breaks that are repaired via Rad52 (Supplementary Fig. 6d). However, sen1-aid cells show a similar asymmetric pattern of Rad52 signal distribution around the sen1-aid specific R-loop peaks as the H2AP signal (Supplementary Fig. 6d, e), which confirms that the increase in H2AP reflects mainly DNA breaks that can be repaired via Rad52. Therefore, the difference in the pattern of H2AP symmetry between hpr1-aid and sen1-aid mutants must reflect a profound difference in the way R-loops generated in G1 (hpr1-aid) and S-phase (sen1-aid) impact on DNA break formation and repair. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x 6NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecom munications ab hpr1-aid sen1-aid WT RNH - + - + - + hpr1-aid sen1-aid WT RNH - + - + - + d e hpr1-aid sen1-aid WT RNH - + - + - + 0 1 2 3 4 5 hpr1-aid sen1-aid WT RNH - + - + - + 0 c % Nuclei with detectable DNA:RNA hybrids 0 10 20 30 40 hpr1-aid sen1-aid WT RNH - + - + - + 0 20 40 60 80 % Nuclei with detectable DNA:RNA h y brids hpr1-aid sen1-aid WT RNH - + - + - + 0.0 0.2 0.4 0.6 0.8 S9.6 si g nal ( arb.units. ) 0.0 0.2 0.4 0.6 0.8 S9.6 si g nal ( arb.units. ) hpr1-aid sen1-aid WT RNH - + - + hpr1-aid sen1-aid WT RNH - + - + - + 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 p=0.039 p=0.07 p=0.045 p=0.045 p=0.003 1 2 3 4 5 S9.6 si g nal ( arb.units. ) p=0.038 p=0.039 p=0.039 p=0.049 p=0.048 p=0.012 p=0.006 p=0.01 p=0.03 p=0.035 p=0.033 GCN4 PDR5 GCN4 PDR5 GCN4 PDR5 αF aux asyn. G1 120’ αF aux asyn. G1 120’ Release αF aux asyn. G1 160’ S aux Release αF aux asyn. G1 160’ S aux G12 Release αF aux asyn. G1 280’ S aux αF - + S9.6 si g nal ( arb.units. ) S9.6 si g nal ( arb.units. ) S9.6 si g nal ( arb.units. ) Fig. 3 Hpr1 or Sen1 depletion causes nuclear DNA−RNA hybrid accumulation differently through the cell cycle. a Percentage of positive nuclei for DNA −RNA hybrids in chromosome spreads stained with the S9.6 antibody in G1-phase WT, hpr1-aid, and sen1-aid cells with or without RNH1. Data are presented as mean values +/−SEM (n=100 cells examined over three independent experiments). bDRIP with S9.6 antibody at GCN4 and PDR5 genes in G1-phase WT, hpr1-aid, and sen1-aid cells treated (+) or not (−) with RNH1 in vitro. Data are presented as mean values +/−SEM (n=6and n=4 biologically independent experiments for GCN4 and PDR5, respectively). cPercentage of positive nuclei for DNA−RNA hybrids in chromosome spreads stained with the S9.6 antibody in S-phase cells as in (a). Data are presented as mean values +/−SEM (n=100 cells examined over three independent experiments). dDRIP with S9.6 antibody at GCN4 and PDR5 genes in S-phase cells as in (b), treated (+) or not (−) with RNH1 in vitro. Data are presented as mean values +/−SEM (n=3 biologically independent experiments). eDRIP with S9.6 antibody at GCN4 and PDR5 genes in G1-phase after one complete cell cycle (G1 2 ) cells as in (b), treated (+) or not (−) with RNH1 in vitro. Data are presented as mean values +/−SEM (n=3 biologically independent experiments). The Pvalues were calculated by the two-tailed unpaired Student t-test. A diagram of the experiment is represented above each plot. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x ARTICLE NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecommunications 7 Finally, we wonder whether this difference in replication defects in both mutants could be extended to regions containing replication origins (ARSs) (Supplementary Fig. 7a). Metaplots of H2AP signal distribution at early ARSs show a remarkable increased signal in the sen1-aid strain (Fig. 6d). The genes near early ARSs showed an increased H2AP signal in the sen1-aid mutant regardless of whether they were oriented HO or CD versus the RF while the hpr1-aid mutant did not show such an increase with respect to the WT (Supplementary Fig. 7b). This result suggests that transcription in the absence of Sen1 promotes T-R conflicts more prone to cause DNA breaks than in the absence of Hpr1. To address if the orientation in T-R conflicts is relevant in our conditional mutants, we grouped the ARS- proximal genes depending on their CD or HO orientation and calculated the percentage of those genes with common and S- phase hpr1- or sen1- specific R-loops. While the proportion of ARS-proximal genes that accumulate R-loops in hpr1 cells (either specific or common with sen1) is similar between HO and CD orientation, the genes that accumulate sen1-specific R-loops are mainly in the HO orientation (Fig. 6e). The results confirm the different nature and consequences of the R-loops prevented by each factor. Discussion THO is the paradigm of conserved factors that prevent R-loop formation9. If inactive, a suboptimal protein-coated nascent mRNA is generated prone to hybridize back with the DNA template. However, human THO works together the mSin3A deacetylase complex and the UAP56 DNA−RNA unwinding activity to prevent R-loop accumulation behind an ongoing RNAPII11,12, and inactivation of THO, mSin3A, and UAP56/ a c b WT hpr1-aid sen1-ai d - - - +++ RNHRNH 0 10 20 30 40 0 5 10 15 20 H2A-P IP/INPUT (arb. units) - WT hpr1-aid sen1-aid - - +++ p=0.003 p=0.001 p=0.018 p=0.021 p=0.007 p=0.009 Release αF aux asyn. G1 160’ S aux GCN4 PDR5 G1 S G12 1 10 100 H2A-P accumulation (Relative values) hpr1-aid sen1-aid WT A A A WT G1 S G1 2 G1 SG1 2 hpr1-aid sen1-aid G1 SG1 2 H2A-P TIR1 aid-9myc G12 Release αF aux asyn. G1 S aux αF 17 95 17 95 130 250 17 95 kDa kDa kDa H2A-P IP/INPUT ( arb. units ) Fig. 4 Hpr1 and Sen1 degron strains accumulate H2AP in S-phase. a Immunoblotting with αH2AP of WT, hpr1-aid, and sen1-aid cells in asynchronous (A), G1, S-phase, and G1-phase after one complete cell cycle (G1 2 ) cultures. AtTIR1-9Myc was used as loading control. bImmunoblotting quantification (means +/−SEM) of protein levels normalized to the WT (n=3 biologically independent samples). cH2AP ChIP analysis at the GCN4 and PDC1 genes in WT, hpr1-aid, and sen1-aid strains with or without RNH1. Data are presented as mean values +/−SEM (n=7 biologically independent experiments). The P values are calculated by the two-tailed unpaired Student t-test. A diagram of the experiment is represented above each experiment. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x 8NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecom munications ChrIV:1,234,290-1,248,197 a bc Watson Crick Watson Crick -2kb peak 2kb WT hpr1-aid sen1-aid 5’ telomeric rep. 3’ d 5’ tRNA 3’ e sense antisense hpr1-aid specific sen1-aid specific common f rep1 rep2 rep1 rep2 rep1 rep2 rep1 rep2 WT WT hpr1 -aid hpr1 -aid rep1 rep2 rep1 rep2 rep1 rep2 rep1 rep2 WT WT sen1 -aid sen1 -aid 1600 1000 400 DRIPc-seq signal (arb.units) 400 800 1200 antisense sense 1200 800 400 1200 800 400 -1Kb 1Kb -1Kb 1Kb 1600 1000 400 2000 1200 400 1000 800 600 -0.5Kb 0.5Kb 1600 1000 400 1600 1000 400 1Kb -1Kb 5’ gene body 3’ 5’ gene body 3’ 5’gene body3’ 1Kb -1Kb 5’gene body3’ 1Kb -1Kb 5’gene body3’ ChrXIII:134,391-151,176 DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) DRIPc-seq signal (arb.units) Fig. 5 DNA−RNA hybrids genome-wide accumulation after Hpr1 and Sen1 controlled depletion. a Representative screenshots of different genomic regions showing the DRIPc-seq signal of detected peaks profiles for WT (blue), hpr1-aid (red), and sen1-aid (green) mapped at Watson and Crick strand (n=2). bDRIPc-seq signal (average coverage) metaplot analysis of peaks ±2 Kb enriched in samples as in (a). cDistribution of antisense and sense DRIPcseq signal (average coverage) along a gene metaplot for cells as in (a). dDistribution of DRIPc-seq signal (average coverage) at telomeric repeats metaplot in samples as in (a). eDistribution of DRIPc-seq signal (average coverage) at tRNAs metaplot in samples as in (a). fDistribution of antisense and sense DRIPc-seq signal (average coverage) along specific enriched genes metaplot in samples as in (a). NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24737-x ARTICLE NATURE COMMUNICATIONS | (2021) 12:4451 | https://doi.org/10.1038/s41467-021-24737-x | www.nature.com/naturecommunications 9