DNA-RNA hybrids: The risks of DNA breakage during transcription

! ! 1! DNA-RNA hybrids: the risks of DNA breakage during 1! transcription 2! 3! Andrés Aguilera & Belén Gómez-González 4! Centro Andaluz de Biología Molecular y Medicina Regenerativa-CABIMER, 5! Universidad de Sevilla-CSIC-Universidad Pablo de Olavide, 6! Seville, Spain 7! [email protected] 8! 9! 10! ABSTRACT 11! Although R loops can occur at different genomic regions, a yet unsolved question is 12! how they form and at which frequency. Recent reports indicate that DNA breaks 13! stimulate DNA-RNA hybrid formation. We discuss the possibility that formation of 14! hybrids may be the inevitable risk of breaks occurring at transcribed DNA regions. 15! Such hybrids must be removed to allow repair, but it is unclear that they are 16! intermediates required for repair. 17! 18! 19! 20! 21! DNA-RNA hybrids can be made between complementary RNA and DNA strands in 22! vitro 1. When this occurs in a double-stranded DNA (dsDNA), the RNA displaces the 23! DNA strand of identical sequence and the whole structure formed by the DNA-RNA 24! hybrid plus the displaced ssDNA is referred to as an R loop. In vivo, very-short DNA-25! RNA hybrids are formed inside the RNA polymerases (RNAPs) during transcription as 26! well as with the RNA primers in Okazaki fragments, in both cases usually no more than 27! 10 nucleotides long. Apart from that, longer DNA-RNA hybrids have been shown to 28! form specifically during class switch recombination and during replication in 29! mitochondrial DNA and in the bacterial plasmid ColE1, in all cases presumably as R 30! loops (reviewed in 2). The length of such hybrids can exceed 1 kb, as shown for the 31! immunoglobulin (Ig) switch regions 3. Their formation during such specific DNA 32! metabolic processes, could in some cases be promoted by specialized proteins. Thus, 33! the mechanism of action of the CRISPR-Cas9 system that relies on the identification of 34! the target DNA site by the guide RNA that presumably form a DNA-RNA hybrid 4 is a 35! good example of this. A different question is how DNA-RNA hybrids form 36! ! 2! spontaneously throughout the genome likely without being mediated by a specialized 37! protein system, becoming a threat to genome integrity and gene expression. 38! 39! Preventing and dissolving R loops that compromise genome integrity 40! 41! The spontaneous formation of R loops was first evidenced by the in vivo identification 42! of DNA-RNA hybrids in yeast cells mutated in THO, a conserved complex that works at 43! the interface of transcription and mRNP biogenesis 5 (Fig. 1). This led to a model in 44! which specific RNA binding proteins (RBPs) protect the nascent RNA from hybridizing 45! back with the DNA template. This model has gained support by the implication of an 46! increasing number of RBPs in the prevention of R loop formation in yeast, C. elegans 47! and human cells 6-11. These R loops have typically been detected by their phenotypical 48! consequences, such as genome instability, which is a stochastic phenomenon, and by 49! in vivo detection methods that only provide measurements of relative increases rather 50! than absolute values (reviewed in 2,12-14). For this reason and despite well-established 51! methods developed to analyze R-loops formed at the Ig switch regions 3,15, the length 52! and frequency at which such spontaneous R loops are formed in the different mutant 53! conditions from yeast to human cells is still unknown. This limits our capacity to clearly 54! define the DNA features required for R loop formation during different DNA metabolic 55! processes. 56! At present, we know a number of specific DNA features that are associated with 57! a higher tendency to form R loops. Yeast and human cell lines show enrichment of 58! DNA-RNA hybrids in the rDNA as well as in some repetitive regions such as yeast 59! transposons, human L1 LINEs, telomeres and regions with chromatin modifications 60! characteristic of repetitive DNA 16-21, which suggests that R loop formation is not 61! necessarily random but can be stimulated by specific DNA characteristics. In this 62! sense, although R-loops in the Ig switch regions only form co-transcriptionally 22, we 63! cannot exclude that some RNAs might be able to re-anneal to the DNA in trans, once 64! they have been released from the transcription site, despite the low stability of RNA 65! due to its single strandedness 23. This possibility might be particularly relevant in the 66! case of small non-coding RNAs or RNAs generated from repetitive DNA sequences. 67! Cells possess different mechanisms to reduce the accumulation of harmful R 68! loops. In addition to specific RBPs, such as the THO complex, that would act as a first 69! option preventing the formation of DNA-RNA hybrids, specific enzymes such as RNase 70! H1 and likely some RNA-DNA helicases, would act subsequently as a second option to 71! remove the RNA once hybridized with the DNA (Fig. 1). Nevertheless, R loops can 72! constitute a roadblock not only for the oncoming RNA polymerases but also for the RF, 73! ! ! 3! thus being a major cause of transcription-replication conflicts that ultimately lead to 74! genetic instability and chromosome fragility 13,24-31. Consequently, a tight cellular 75! coordination between transcription and replication is required so that upon replication 76! fork (RF) stalling, cells do still have a third option to reduce R loops. The replication-77! related DNA repair machinery could contribute to R-loop dissolution (Fig. 1), as shown 78! for the Fanconi Anemia and for BRCA1 and BRCA2 repair factors 30,32-34. 79! In addition, R-loops have been reported to induce antisense transcription, 80! leading to the generation of dsRNA that in turn recruits the silencing machinery 35. The 81! antisense RNA could potentially counteract DNA-RNA hybrid formation by competing 82! with the DNA strand in the hybrid. In support of this, an accumulation of DNA-RNA 83! hybrids has been observed in the absence of DICER in S. pombe, although it has only 84! been attributed to a defect in transcription termination 36. Whether antisense RNA could 85! represent a global mechanism to remove DNA-RNA hybrids is yet to be addressed. 86! Anyhow, the existence diverse mechanisms to prevent, remove or dissolve DNA-RNA 87! hybrids after the RF encounters an R loop, supports their potential harmfulness (Fig. 88! 1). Indeed, R-loops might also be involved in neurological diseases and cancer 89! (reviewed in 37). 90! 91! DNA breaks at transcribing DNA as promoters of DNA-RNA hybrid formation 92! 93! Consistent with the high-energy cost that would support a single-stranded nucleic acid 94! invading a closed dsDNA, several in cis elements have been shown to favor DNA-RNA 95! hybridization (Fig. 2). The local negative supercoiling accumulated behind the 96! elongating RNAP facilitates DNA strand opening, thus favoring invasion of the nascent 97! RNA produced in cis, as can be inferred from genetic data in bacteria and yeast 26,38-40. 98! Also, DNA-RNA hybrids preferentially accumulate throughout the genome with a 99! different likelihood depending on their GC or AT richness 18,19 and the formation of DNA 100! secondary structures such as G quadruplexes formed at a nontemplate G-rich strand 101! can also favour R loops as shown in vitro 22,41. Indeed, given the complex and 102! demanding step of ssDNA invasion into a dsDNA during homologous recombination 103! (HR), which requires an available free DNA end and a number of proteins responsible 104! for strand exchange (reviewed in 42), it is unlikely that an internal sequence of a long 105! ssRNA could stably hybridize with an intact dsDNA template spontaneously. Although 106! a strand exchange protein does not seem to be required for R-loop formation, at least 107! in vitro, a free DNA end would relieve any topological tension and enable greater 108! breathing of the non-template DNA strand, so that the broken DNA would more easily 109! undergo unwinding. In this scenario, the nascent RNA could better interact with the 110! ! 4! template complementary DNA strand thus favouring the extension or stabilization of 111! any putative transient RNA-DNA hybrid. 112! Importantly, it has been shown that an ssDNA nick in the Ig switch region 113! transcribed from a T7 promoter strongly stimulates DNA-RNA hybrid formation 40. 114! Along this same line, camptothecin, which causes an ssDNA nick with its 3’-end 115! covalently bound to Topo I (reviewed in 43), also increases DNA-RNA hybrids 44,45. An 116! ssDNA nick in a transcribing double helix of DNA will release the topological 117! constriction of the DNA, allowing DNA opening and rotation and the emergence of free 118! ssDNA ends prone to anneal with the nascent RNA (Fig. 2). The relevance of DNA 119! cleavage as one driving force in DNA-RNA hybrid formation thus seems clear. 120! Supporting the formation of DNA-RNA hybrids after DNA damage at 121! transcription sites, the Calsou group detected a transcription-dependent accumulation 122! of a catalytically inactive version of the E. coli RNase H to human DNA after laser 123! micro-irradiation, which produces ssDNA nicks and DSBs 46. Furthermore, they 124! described the existence of a mechanism antagonizing R-loop formation after such 125! damage that is dependent on all main players in the DNA damage response: ATM, 126! ATR and DNA-PK. This mechanism relies on the exclusion of the RNA-bound pool of 127! RNA-binding proteins from the damaged DNA 46. Consistent with the general view that 128! cleavage of the DNA behind an elongating RNAP facilitates DNA-RNA hybrids, two 129! papers by the Godbout and Fisher groups have recently shown that DNA-RNA hybrids 130! also accumulate upon double-strand break (DSB) induction 47,48. 131! The Godbout group reported previously that the human RNA-unwinding protein 132! DEAD box 1 (DDX1) is an ATM target and forms foci at DSBs upon ionizing radiation 133! (IR) 49. Now, they have further shown that DDX1 depletion causes a delay in the repair 134! of IR-induced γH2AX foci, reduced levels of RAD51 foci and increased lethality after IR, 135! implying an involvement of DDX1 in the DNA damage response 47. Importantly, IR-136! induced DDX1 foci are sensitive to RNase H treatment 49 as well as to the transcription 137! inhibitors actinomycin D and cordycepin 47. Using DNA-RNA Immuno-Precipitation 138! (DRIP) as well as an in vitro assay based on the combined treatment with duplex-139! specific nuclease and/or RNase H, they have shown that DNA-RNA hybrids 140! accumulate around a DSB when DDX1 is depleted 47. Their results led them to propose 141! that transcription promotes DDX1 recruitment to remove DNA-RNA hybrids at DSBs in 142! human cells. The study constitutes new evidence that breakage of DNA during 143! transcription facilitates the nascent RNA to hybridize back with the template DNA. In 144! the same line, DDX1 foci are formed after RF breakage induced by long exposure to 145! cisplatin or UV 49, which might reflect a high incidence of transcription-replication 146! conflicts under such DNA damaging conditions. 147! ! ! 5! On the other hand, the Fisher group has reported that in S. pombe RNase H is 148! required for the repair of the DSBs induced by I-PpoI endonuclease as well as for cell 149! survival after cleavage 48. This is accompanied by an increase in DNA-RNA hybrids at 150! the site of DSB in rnh1∆ rnh2∆ cells. The results again strongly support that DNA 151! breakage can lead to DNA-RNA hybrids. In this sense, it has been shown that the 152! bacterial strand exchange protein RecA/Rad51 can catalyze RNA-DNA hybrid 153! formation in vitro and its yeast ortholog Rad51 can modulate DNA-RNA hybrid 154! formation in trans in S. cerevisiae cells 50-52. RecA/Rad51 might sporadically contribute 155! to make a cleaved DNA molecule more susceptible to hybrid formation, but this is 156! unknown. In any case, the recent S. pombe study also shows that RNH1 157! overexpression affects DSB repair and survival without increasing DNA-RNA hybrids 158! and leads to hyper-recombination between direct repeats when a DSB is induced 159! between the repeats 48. This suggests that the consequences of RNH1 action on RNA-160! DNA hybrids formed at broken DNA may not always be benign. Indeed, RNase H 161! overexpression cannot compensate for the removal of RNA-DNA hybrids at DSBs in 162! DDX1-depleted cells 47, implying that RNA removal does not always have straight and 163! predictable consequences. 164! 165! Possible impact of DNA-RNA hybrids on DNA repair 166! 167! In light of these two reports 47,48 and the previous work of the Lieber and Calsou labs 168! 40,46, we can conclude that ssDNA or dsDNA breakage, is a major driver of stable DNA-169! RNA hybrid formation and that DNA-RNA hybrids will have an impact on the repair of 170! such breaks (Fig. 2). This adds a new step to the repair of breaks occurring in the DNA 171! when this is being transcribed. In agreement with an important fraction of the genome 172! prone to form DNA-RNA hybrids upon breakage, DDX1 depletion increases not only 173! the sensitivity to IR but also to PARP inhibitors 47, a characteristic phenotype of cells 174! deficient in HR factors. It is thus expected that the eventual formation of DNA-RNA 175! hybrids at DSBs requires the removal of the RNA to allow DSB processing and repair. 176! Consistently, a recent report suggests that the persistence of R-loops compromises 177! DNA repair based on the fact that rnh1∆ rnh2∆ cells lead to a dramatic increase in 178! unrepaired Rad52 foci and lethality in the absence of Top1 mostly due to defective 179! break-induced replication at the repetitive rDNA region 53. In principle, this should be 180! specific to DNA regions undergoing transcription. 181! A breakage occurring at a transcribing DNA produces changes in local 182! supercoiling and chromatin structure that, together with the free DNA ends generated, 183! could facilitate hybridization of the nascent RNA with the ssDNA templates (Fig. 2). 184! ! 6! The formation of DNA-RNA hybrids at the DSB site might also be facilitated by the 5’-185! end resection step of HR, because this generates a free 3’-ended DNA strand 186! overhang that could easily hybridize with the nascent RNA (Fig. 3). In agreement with 187! the non-resected template DNA strand forming a DNA-RNA hybrid, the amount of 188! RPA-bound DNA after DSB induction is reduced in rnh1∆ rnh2∆ mutants and increased 189! upon RNase H overexpression 48. In this scenario, standard methods used to infer 190! resection based on the exposure of ssDNA, including RPA, may not be valid. 191! Alternatively, DNA-RNA hybrids might be formed before resection, displacing 192! the 5’-ended non-template strand, so that the major exposure of the 5’-ended ssDNA 193! could promote its degradation by nucleases. Since 5’-end DSB resection is the key 194! step channeling DSB repair towards HR 54, DNA-RNA hybrids at DSBs might therefore 195! promote HR versus NHEJ repair. Indeed, the alteration by the DNA-RNA hybrid of the 196! nature of the DSB end might impede its proper recognition by NHEJ factors (Fig. 3). 197! Interestingly, and consistent with this view, a genome-wide analysis of repair of AsiSI 198! nuclease-induced DSBs in human cells has shown that breaks at transcriptionally 199! active chromatin are preferentially repaired by HR, whereas those transcriptionally 200! inactive chromatin are repaired by NHEJ 55. It would be interesting to know whether 201! DNA-RNA hybrids contribute to this difference. As discussed above, for the repair to be 202! productive, the RNA should be removed by RNase H1 or different DNA-RNA helicases. 203! Interestingly, DDX1 seems to be required for DSB repair only after DNA resection 47. 204! 205! Do DNA-RNA hybrids have a regular role in DNA repair? 206! 207! The next question is whether DNA-RNA hybrids have a positive function in DSB repair. 208! Transcription of non-coding RNA has been proposed to be important in the DDR upon 209! DNA breakage in plants, human cells and flies 56-58. Furthermore, RBPs and 210! transcription factors have been shown in different studies to be enriched at damaged 211! DNA although their direct role in DNA repair remains to be defined (reviewed in 59). 212! However, there is no experimental evidence to our knowledge for the requirement of 213! DNA-RNA hybrids in the process of DSB repair. Indeed, the fact that DSBs can be 214! repaired in vitro without any RNA suggests that R-loops are not obligatory. The 215! detection of DNA-RNA hybrids at DSBs may just be a consequence of the nascent and 216! sticky RNA being an unavoidable guest of breaks occurring at active chromatin. 217! Examples of rare events that can happen during DSB repair include those cases in 218! which an RNA molecule serve as an homologous template for recombination 60 without 219! meaning that a template RNA is required in HR repair, or in which Ty and other mRNA-220! ! ! 7! derived cDNAs are used as intermediates in some DSB repair events 61,62, without 221! meaning that such sequences enhance the repair process itself. 222! As indicated above, an important unresolved question is the absolute frequency 223! at which R loops are formed at DSBs. A requirement for R loops in DSB repair would 224! imply that they should be obligatory formed at DSBs, something for which so far we do 225! not have evidence. In this sense, it is worth noting that the potential occurrence of 226! DNA-RNA hybrids at breaks could be pretty general given the increasing evidence that 227! supports that most of the genome in eukaryotes is transcribed at some moment, 228! leading to mRNAs, rRNAs, tRNAs, snRNAs, or ncRNAs, including lncRNAs, scRNAs, 229! CUTs, miRNAs or siRNAs. However, in an important fraction of the genome, 230! transcription is spurious so that the frequency at which DSBs occur during transcription 231! may vary greatly between chromosomal regions. 232! A certainly interesting observation is that RNAPII is enriched at DSB sites as 233! observed by ChIP in S. pombe 48. This does not necessarily mean that RNAPs are 234! recruited to DSB sites. ChIPs only provide information about occupancy, not 235! necessarily about active recruitment. Although speculative at this point, the DNA-RNA 236! hybrid or the supercoiling and chromatin modifications associated with DSBs might 237! block or retain a previously engaged RNAP at the vicinity of the breaks via different 238! means, leading to an enhanced detection by ChIP of an RNAP present at the site prior 239! to the break. In agreement, there are no changes in the transcript levels at the DSB 240! site 48. These and other intriguing questions would need to be solved in the near future. 241! In this context, given that rDNA sequences are responsible for almost 50% of all DNA-242! RNA hybrids in yeast 19, an important question would be whether the enrichment at 243! DSBs is restricted to RNAPII, or would it also be the case for RNAPI or RNAPIII 244! polymerases. 245! 246! In summary, RNA is a sticky molecule that can hybridize back with the DNA if 247! circumstances are favorable. DNA breakage at transcribed regions leads to such 248! favorable situations, leaving DNA defenseless with respect to the nascent RNA. Given 249! that RNA is highly abundant and that most of the genome can be transcribed, the 250! reports discussed here further support that understanding and identifying the factors 251! and mechanisms that either prevent or eliminate DNA-RNA hybrids is necessary to 252! have a complete understanding of genome dynamics, a subject of increasing relevance 253! in cancer and genetic disease. 254! 255! 256! ! 8! References 257! 258! 1. Thomas, M., White, R.L. & Davis, R.W. Hybridization of RNA to double-259! stranded DNA: formation of R-loops. Proc. Natl. Acad. Sci. USA 73, 260! 2294-8 (1976). 261! 2. Aguilera, A. & Garcia-Muse, T. R loops: from transcription byproducts to 262! threats to genome stability. Mol. Cell 46, 115-24 (2012). 263! 3. Yu, K., Chedin, F., Hsieh, C.L., Wilson, T.E. & Lieber, M.R. R-loops at 264! immunoglobulin class switch regions in the chromosomes of stimulated B 265! cells. Nat. Immunol. 4, 442-51 (2003). 266! 4. Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in 267! adaptive bacterial immunity. Science 337, 816-21 (2012). 268! 5. Huertas, P. & Aguilera, A. Cotranscriptionally formed DNA:RNA hybrids 269! mediate transcription elongation impairment and transcription-associated 270! recombination. Mol. Cell 12, 711-21 (2003). 271! 6. Li, X. & Manley, J.L. Inactivation of the SR protein splicing factor 272! ASF/SF2 results in genomic instability. Cell 122, 365-78 (2005). 273! 7. Paulsen, R.D. et al. A genome-wide siRNA screen reveals diverse 274! cellular processes and pathways that mediate genome stability. Mol. Cell 275! 35, 228-39 (2009). 276! 8. Castellano-Pozo, M., Garcia-Muse, T. & Aguilera, A. R-loops cause 277! replication impairment and genome instability during meiosis. EMBO 278! Rep. 13, 923-9 (2012). 279! 9. Wahba, L., Amon, J.D., Koshland, D. & Vuica-Ross, M. RNase H and 280! multiple RNA biogenesis factors cooperate to prevent RNA:DNA hybrids 281! from generating genome instability. Mol. Cell 44, 978-88 (2011). 282! 10. Stirling, P.C. et al. R-loop-mediated genome instability in mRNA 283! cleavage and polyadenylation mutants. Genes Dev 26, 163-75 (2012). 284! 11. Sridhara, S. et al. Transcription dynamics prevents RNA-mediated 285! genomic instability through SRPK2-dependent DDX23 phosphorylation. 286! Cell Rep. 18, 334-343 (2017). 287! 12. Skourti-Stathaki, K. & Proudfoot, N.J. A double-edged sword: R loops as 288! threats to genome integrity and powerful regulators of gene expression. 289! Genes Dev. 28, 1384-96 (2014). 290! 13. Sollier, J. & Cimprich, K.A. Breaking bad: R-loops and genome integrity. 291! Trends Cell. Biol. 25, 514-22 (2015). 292! 14. Santos-Pereira, J.M. & Aguilera, A. R loops: new modulators of genome 293! dynamics and function. Nat. Rev. Genet. 16, 583-97 (2015). 294! 15. Huang, F.T. et al. Sequence dependence of chromosomal R-loops at the 295! immunoglobulin heavy-chain Smu class switch region. Mol. Cell. Biol. 27, 296! 5921-32 (2007). 297! 16. Chan, Y.A. et al. Genome-wide profiling of yeast DNA:RNA hybrid prone 298! sites with DRIP-chip. PLoS Genet. 10, e1004288 (2014). 299! 17. El Hage, A., Webb, S., Kerr, A. & Tollervey, D. Genome-wide distribution 300! of RNA-DNA hybrids identifies RNase H targets in tRNA genes, 301! retrotransposons and mitochondria. PLoS Genet. 10, e1004716 (2014). 302! 18. Ginno, P.A., Lott, P.L., Christensen, H.C., Korf, I. & Chedin, F. R-loop 303! formation is a distinctive characteristic of unmethylated human CpG 304! island promoters. Mol. Cell 45, 814-25 (2012). 305! ! ! 9! 19. Wahba, L., Costantino, L., Tan, F.J., Zimmer, A. & Koshland, D. S1-306! DRIP-seq identifies high expression and polyA tracts as major 307! contributors to R-loop formation. Genes Dev. 30, 1327-38 (2016). 308! 20. Nadel, J. et al. RNA:DNA hybrids in the human genome have distinctive 309! nucleotide characteristics, chromatin composition, and transcriptional 310! relationships. Epigenetics Chromatin 8, 46 (2015). 311! 21. Balk, B., Dees, M., Bender, K. & Luke, B. The differential processing of 312! telomeres in response to increased telomeric transcription and RNA-DNA 313! hybrid accumulation. RNA Biol. 11, 95-100 (2014). 314! 22. Duquette, M.L., Handa, P., Vincent, J.A., Taylor, A.F. & Maizels, N. 315! Intracellular transcription of G-rich DNAs induces formation of G-loops, 316! novel structures containing G4 DNA. Genes Dev. 18, 1618-29 (2004). 317! 23. Lindahl, T. Instability and decay of the primary structure of DNA. Nature 318! 362, 709-15 (1993). 319! 24. Wellinger, R.E., Prado, F. & Aguilera, A. Replication fork progression is 320! impaired by transcription in hyperrecombinant yeast cells lacking a 321! functional THO complex. Mol. Cell. Biol. 26, 3327-34 (2006). 322! 25. Gan, W. et al. R-loop-mediated genomic instability is caused by 323! impairment of replication fork progression. Genes Dev. 25, 2041-56 324! (2011). 325! 26. Tuduri, S. et al. Topoisomerase I suppresses genomic instability by 326! preventing interference between replication and transcription. Nat. Cell 327! Biol. 11, 1315-24 (2009). 328! 27. Helmrich, A., Ballarino, M. & Tora, L. Collisions between replication and 329! transcription complexes cause common fragile site instability at the 330! longest human genes. Mol. Cell 44, 966-77 (2011). 331! 28. Houlard, M., Artus, J., Leguillier, T., Vandormael-Pournin, S. & Cohen-332! Tannoudji, M. DNA-RNA hybrids contribute to the replication dependent 333! genomic instability induced by Omcg1 deficiency. Cell Cycle 10, 108-17 334! (2011). 335! 29. Merrikh, H., Zhang, Y., Grossman, A.D. & Wang, J.D. Replication-336! transcription conflicts in bacteria. Nat. Rev. Microbiol. 10, 449-58 (2012). 337! 30. Schwab, R.A. et al. The Fanconi Anemia Pathway Maintains Genome 338! Stability by Coordinating Replication and Transcription. Mol. Cell 60, 339! 351-61 (2015). 340! 31. Garcia-Muse, T. & Aguilera, A. Transcription-replication conflicts: how 341! they occur and how they are resolved. Nat. Rev. Mol. Cell. Biol. (2016). 342! 32. Bhatia, V. et al. BRCA2 prevents R-loop accumulation and associates 343! with TREX-2 mRNA export factor PCID2. Nature 511, 362-5 (2014). 344! 33. Hatchi, E. et al. BRCA1 recruitment to transcriptional pause sites is 345! required for R-loop-driven DNA damage repair. Mol. Cell 57, 636-47 346! (2015). 347! 34. Garcia-Rubio, M.L. et al. The Fanconi Anemia Pathway Protects 348! Genome Integrity from R-loops. PLoS Genet 11, e1005674 (2015). 349! 35. Skourti-Stathaki, K., Kamieniarz-Gdula, K. & Proudfoot, N.J. R-loops 350! induce repressive chromatin marks over mammalian gene terminators. 351! Nature 516, 436-9 (2014). 352! 36. Castel, S.E. et al. Dicer promotes transcription termination at sites of 353! replication stress to maintain genome stability. Cell 159, 572-83 (2014). 354!