Insights into RNA-mediated genome instability by altering ALY expression and the use of new drugs

Insights into RNA-mediated genome instability by altering ALY expression and the use of new drugs Cristina Guillén Mendoza Tesis Doctoral Universidad de Sevilla 2023 2 3 4 5 6 7 A mis abuelas, por su esfuerzo y ejemplo 8 9 AGRADECIMIENTOS Después de tantos años y vivencias, solo me queda agradecer mucho a mucha gente. Pensando un poco, tendría para escribir otra tesis solo con agradecimientos. En primer lugar, quiero dar las gracias a mi director Andrés por haberme dado la oportunidad de realizar la tesis y seguir desarrollándome como profesional en su laboratorio. Es de agradecer que a pesar de todos los que somos y todas sus tareas siempre haya tenido un hueco para dedicarme. Ha sido un lujo poder compartir ciencia contigo y espero haber podido aprender mucho de ti. A mi codirectora tengo que agradecerle infinito. Rosa, has sido mi suerte en la tesis, de verdad que no sé cómo podré agradecerte tu esfuerzo, trabajo, dedicación, cariño, paciencia…Es una verdadera suerte poder decir que adoro a mi jefa desde el primer día. Eres una extraordinaria profesional y mejor persona. Además de con mis directores, en este tiempo he contado con todo un laboratorio repleto de compañeros dispuestos a ayudar y animar. Desde los despachos hasta las poyatas, desde compañeros veteranos hasta recién llegados, a todos, mi eterno cariño y agradecimiento. Me gustaría destacar a Emanuela, de la que aprendí toda la mecánica del laboratorio allá por mi época del máster y a Sonia Pi por confiar en mi y darme la oportunidad de hacer cosas nuevas que he disfrutado muchísimo. No solo tengo que agradecerle a mi laboratorio, sino a todo CABIMER, a los demás laboratorios y su gente, por tantos reactivos y momentos compartidos. A administración, a las unidades de genómica, microscopía, cultivos celulares, citometría, organismos modelo, cocina, informática y al equipo de limpieza y conserjería por vuestra ayuda en innumerables ocasiones. Me llevo buenos recuerdos, compañeros y amigos de estos inolvidables años. No quiero dejar pasar esta oportunidad para agradecer a los que siempre están conmigo y han hecho mucho más fácil estos años de tesis. A mis amigos, a los de siempre. Tan diferentes cada uno y tan unidos desde hace tantísimos años. Gracias por escucharme, por callarme cuando entro 16 Figure R18. Analysis of ALY and UAP56 at damage induced by photoablation laser ................................................................................................................. 98 Figure R19. Analysis of ALY-GFP recruitment at DBS .................................. 101 Figure R20. ALY mutant 10E is still recruited at damage sites ...................... 102 Figure R21. ALY recruitment at DSBs is completely dependent on parylation ....................................................................................................................... 103 Figure R22. ALY exclusion at DSBs does not depend on ATM, ATR or DNA-PK ....................................................................................................................... 105 Figure R23. ALY recruitment upon repair factors depletion ........................... 107 Figure R24. Analysis of recombination upon ALY and UAP56 depletion ....... 109 Figure R25. Survival assays after irradiation or Olaparib treatment .............. 111 Figure R26. Recruitment of endogenous ALY ............................................... 112 Figure R27. үH2AX repair upon ALY overexpression .................................... 112 Figure R28. Screening of epigenetic compounds that leads to R-loop accumulation .................................................................................................. 118 Figure R29. R-loop accumulation upon epigenetic compounds treatment .... 120 Figure R30. DNA damage analysis upon epigenetic compounds treatment .. 122 Figure R31. EU incorporation upon epigenetic compound treatment ............ 123 Figure R32. Analysis of genetic instability upon compounds’ targets knockdown ....................................................................................................................... 125 Figure R33. Treatment of siTHOC1 and siSETX with the selected epigenetic compound....................................................................................................... 127 Figure R34. R-loop and DNA damage upon double treatments .................... 128 Figure R35. Survival assays in patient cell lines deficient on R-loop factors upon treatment with epigenetic compounds ................................................... 129 DISCUSSION Figure D1. A model to explain how deregulated levels of ALY protein could lead to R-loop accumulation, DNA repair impairment and genetic instability. ........ 142 Figure D2. Epigenetic inhibitors inducing R-loops and their potential use in combined therapies. ....................................................................................... 147 17 APPENDIX Appendix 1. R-loop accumulation upon DNA damage induction by irradiation. ....................................................................................................................... 157 Appendix 2. RNase H1 hybrid-binding domain at sites of laser microirradiationinduced DNA damage. ................................................................................... 158 Appendix 3. R-loop accumulation after irradiation upon depletion of different factors. ........................................................................................................... 159 Appendix 4. Drugs used in the high-throughput immunofluorescence screening. ....................................................................................................... 159 18 ABBREVIATIONS 4-OHT 4-hydroxitamoxifen 53BP1 p53-binding protein 1 үH2AX Histone variant H2AX phosphorylated on Serine 139 AID Activation-Induced cytidine Deaminase alt-EJ Alternative non-homologous end joining AREX Alternative mRNA export complex ATM Ataxia Telangiectasia Mutated ATP Adenosine triphosphate ATR Ataxia Telangiectasia and Rad3-related protein A.U. Arbitrary Units BFP Blue fluorescent protein BLESS direct in situ breaks labeling, enrichment on streptavidin and nextgeneration sequencing bp base pairs BRCA1 Breast cancer 1 BRCA2 Breast cancer 2 ChIP-seq Chromatin Immunoprecipitation and sequencing CIN Chromosome instability CLIP UV crosslinking and immunoprecipitation c-NHEJ Canonical non-homologous end joining CNV Copy Number Variant CSR Class switching recombination DAPI 2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride DDR DNA damage response DiVA DSB inducible via AsiSI DNA Deoxyribonucleic acid DNA-PKcs DNA-dependent protein kinase catalytic subunit 19 dNTP Deoxyribonucleotide triphosphate Dox Doxycycline DRB 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole DRIP DNA-RNA immunoprecipitation DRIP-qPCR DRIP-quantitative polymerase chain reaction dsDNA Double-stranded DNA DSB Double-strand break dsRNA Double-stranded RNA EdU 5-Ethynyl-2’-deoxyuridine ER Estrogen receptor EU 5-ethynyluridine FA Fanconi Anemia FACS Fluorescence Activated Cell Sorting FBS Fetal bovine serum GCR Gross chromosomal rearrangements GFP Green fluorescence protein H3S10pho Histone H3 Phosphorylation at serine 10 HAT Histone acetyltransferase HBD Hybrid-binding domain HDAC Histone Deacetylase HnRNP heterogenous ribonucleoprotein HR Homologous Recombination HU Hydroxyurea IF Immunofluorescence Ig Immunoglobulin IP Immunoprecipitate IR Ionizing radiation Kb Kilobases KDa Kilodalton 20 LOH Loss of heterozygosity MIN Micro- and minisatellite instability MMEJ Microhomology-mediated end joining MOI Multiplicity of infection MRN MRE11-RAD50-NBS1 mRNA Messenger RNA mRNP Messenger ribonucleoprotein ncRNA Non-coding ribonucleic acid NHEJ Non-homologous end joining NTS Non-transcribed strand PAR Poly (ADP-ribose) PARG Poly (ADP-ribose) glycohydrolase PARP Poly (ADP-ribose) polymerase Parylation Poly (AD-ribosyl)ation PI3K Phosphoinositide-3-kinase PIKK Phosphatidylinositol 3-kinase-like protein kinase qPCR Quantitative polymerase chain reaction RBP RNA-binding protein RBD RNA-binding domain REF RNA export factor RF Replication fork RFB Replication fork barrier RGG Sequence enriched in arginine and glycine rRNA ribosomal RNA RNA Ribonucleic acid RNAP RNA polymerase RNase Ribonuclease RNH1 RNase H1 ROI Region of interest 21 RPA Replication protein A RRM RNA-recognition motif RT Room temperature siC Non-targeting siRNA control siRNA small interfering RNA SSA Single strand annealing SSB Single-strand break ssDNA single-stranded DNA STAT signal transducers and transcription activator TAM Transcription-associated mutation TAR Transcription-associated recombination TAGIN Transcription-associated genome instability TdT Terminal deoxynucleotidyl transferase TRC Transcription-Replication conflict TREX Transcription-Export complex tRNA transfer ribonucleic acid TS Transcribed Strand TSA Trichostatin A UBM UAP56-binding motif UV Ultraviolet WT Wild Type 22 RESUMEN 23 Las células han desarrollado una serie de mecanismos que utilizan de forma coordinada para mantener la integridad del genoma y permitir una correcta transmisión del material genético a su descendencia. Sin embargo, deben hacer frente a una amplia gama de daños en el genoma generados por agentes genotóxicos tanto internos como externos, y que en algunos casos pueden provocar roturas en el ADN. Las roturas de cadena sencilla son las más comunes, pero las de doble cadena son más citotóxicas. Todo ello puede desencadenar inestabilidad genética que se manifiesta en forma de mutaciones o reordenamientos cromosómicos. Si bien es cierto que puede suponer un motor para la evolución, la inestabilidad genética es una patología celular asociada al envejecimiento prematuro, el cáncer y patologías neurológicas. La transcripción es un proceso celular que permite copiar la información requerida desde el ADN al ARN. Este proceso es esencial para la supervivencia celular pero puede comprometer la integridad del genoma. Ello se debe en parte a cambios topológicos en el ADN que lo pueden hacer más vulnerable. Puede suponer un obstáculo para la replicación y aumentar los conflictos transcripciónreplicación que pueden desembocar en roturas de ADN. Los bucles R son una de las causas más estudiadas de la inestabilidad genética asociada a transcripción. Estas estructuras se componen de un híbrido de ADN-ARN y una cadena simple de ADN desplazada. Estos bucles R aunque pueden tener funciones fisiológicas, también pueden suponer un riesgo para la estabilidad del genoma. Por ello las células poseen factores y mecanismos para prevenir y/o resolver estas estructuras, entre las que se encuentran proteínas del metabolismo del ARN. Recientemente, se ha descrito la presencia de bucles R en las roturas de ADN, aunque si tienen o no alguna función de reparación del ADN no se ha resulto aún. Dada la conexión entre la transcripción y el ARN con la inestabilidad genética y el cáncer, nos preguntamos si factores previamente relacionados con el metabolismo del ARN tendrían un impacto en inestabilidad genética e incluso en la respuesta al daño en el ADN. Además, dada la emergente importancia del contexto de la cromatina en la formación de híbridos que pueden desencadenar en inestabilidad genética y cáncer decidimos buscar nuevos compuestos que incrementaran el nivel de estos bucles. 24 En esta tesis hemos mostrado la importancia de mantener un nivel adecuado de la proteína de unión al ARN ALY para prevenir que la formación de híbridos de ADN-ARN y así preservar la integridad del genoma. Las células con un reducido nivel de ALY obtenido mediante silenciamiento con siARN, así como las que sobreexpresan esta proteína, muestran un mayor nivel de híbridos y de daño en el ADN. Por otro lado, mostramos que la proteína ALY es reclutada específica y rápidamente al sitio de daño en el ADN inducido mediante experimentos de microirradiación. A pesar de que ALY es una proteína cuyas funciones principales son de metabolismo del ARN, su reclutamiento al sitio de daño en el ADN solo depende parcialmente de la presencia de ARN o híbridos ADN-AR. No obstante, dicha localización depende completamente de PARylación, que genera cadenas de poly (ADP-ribosa). Así, el silenciamiento y la sobreexpresión de ALY generan problemas en la señalización o la reparación del daño inducido por microirradiación, lo cual sugiere que la proteína ALY podría interferir en la respuesta al daño al ADN. En una segunda parte de este trabajo, hemos podido identificar nuevos compuestos cuyo tratamiento provoca la acumulación de híbridos ADN-ARN. Para ello hemos realizado un escrutinio de alto rendimiento con microscopía de fluorescencia con el anticuerpo S9.6, que reconoce híbridos de ADN-ARN, y una colección de inhibidores y activadores de enzimas relacionadas con procesos epigenéticos. Tras el escrutinio, titulación y validación de los candidatos obtenidos, se seleccionaron tres compuestos para identificar el daño en el ADN que producían. Finalmente, nos centramos en los inhibidores de quinasas cuyo uso además de una acumulación de híbridos ADN-ARN generó daños en el ADN. En este sentido, el silenciamiento del factor de transducción de señales y activador de la transcripción STAT3, diana de uno de los compuestos seleccionados, aumenta los híbridos ADN-ARN. Queda por determinar el mecanismo por el que se produce este fenotipo. Por último, cabe resaltar el posible potencial clínico de estos dos compuestos, tal y como reflejan ensayos de viabilidad de células deficientes en, FANCD2 y BRCA2, factores relacionados con inestabilidad genética. 25 INTRODUCTION 32 Figure I2. Transcription-associated genome instability. (A) The progression of transcription promotes the formation of local negative supercoiling behind the RNAP and positive supercoiling ahead the RNAP that make the DNA more susceptible to damage. (B) Behind the RNAP, the more relaxed DNA is not only more vulnerable to different mutagenic agents but also to form non- B structures such as hairpins, G4-quadruplexes or DNA-RNA hybrids or R loops. (C) Encounters between replication and transcription machineries cause transcription-replication collisions. For transcription machinery, RNAPII is represented whereas for replication machinery, the major replicative DNA helicase MCM and DNA polymerase II (DNA pol II) are represented in the RF. MCM, minichromosome maintenance complex; RF, Replication Fork; RNAPII, RNA polymerase; DNA pol II, DNA polymerase II. Figure adapted from (Gaillard et al., 2013). 33 1.3. R-loops as a source of genome instability R-loop is a three stranded structure formed by a DNA-RNA hybrid and the displaced ssDNA. These non-B DNA structures appear when the nascent RNA rehybridize with the DNA template outside the transcription bubble displacing the NT strand that remains as ssDNA (Figure I3). Therefore, R-loops are formed naturally and they have physiological roles but when they take place in a nonscheduled manner they can also be pathological being a source of genome instability (Brickner et al., 2022; Crossley et al., 2019; García-Muse & Aguilera, 2019). Figure I3. Schematic representation of an R-loop structure. R loops is a non-B DNA structure which is constituted by a three stranded nucleic acid structure that consists in a DNA-RNA hybrid and a displaced ssDNA. Genome-wide analysis has unveiled that exist a relevant percentage of R- loops in normal/WT cells reaching 8% and 5% genome coverage data in yeast and human cells, respectively (Sanz et al., 2016; Wahba et al., 2016). These structures accumulate preferentially at highly transcribed genes, including rRNA and tRNA loci, together with transposable elements, centromeres and telomeres, antisense-RNAs or ncRNAs regions (Y. A. Chan, Aristizabal, et al., 2014; El Hage et al., 2014). R-loops are detected in G-C rich sequences (L. Chen et al., 2017). Moreover, in mammalian cells R-loops also accumulate at promoter and termination regions and are prone to formed at CpG islands or G-quadruplexes (G4)-containing sequences (L. Chen et al., 2017; Ginno et al., 2012; Sanz et al., 2016). 1.3.1. Physiological roles R-loops are present in WT cells playing different roles. Programmed R-loops have been described in different physiological processes such as plasmid 34 replication in E. coli (Kogoma, 1997), mitochondrial DNA replication (B. Xu & Clayton, 1996), immunoglobulin (Ig) class switching recombination (CSR) in B- cells (Yu et al., 2003) or even in CRISPR-Cas9 gene editing (F. Jiang et al., 2016). The fact that R-loops accumulate at promoter and terminator regions have suggested a role in gene expression regulation (Niehrs & Luke, 2020). Indeed, Genome-wide analysis have revealed that these structures are linked to unmethylated CpG island which are present at the promoters of many genes suggesting that R-loops protect these regions against transcription silencing (Ginno et al., 2012; Grunseich et al., 2018). Moreover, genome-wide analysis have also shown an overlap between R-loops and termination sites and it has been suggested that the intrinsic ability of R-loops to stall transcription can help to pause RNAPII and mediate transcription termination (Niehrs & Luke, 2020). R- loops can induce antisense transcription over the 3’ ends of genes and it has been linked to transcription termination (Skourti-Stathaki et al., 2014). On the other hand, the presence of R-loops can affect the binding of different transcription factors, mediated in some cases by ncRNAs (Arab et al., 2019; Boque-Sastre et al., 2015; Cloutier et al., 2016). Although R-loops may have different physiological roles, when they are formed unscheduled or in excess they can interfere with transcription, replication or repair processes becoming a threat for genome integrity (Brickner et al., 2022; Crossley et al., 2019; García-Muse & Aguilera, 2019). 1.3.2. R-loops as genomic threats R-loops are structures that can lead to genome instability in form of mutations, recombination, chromosome rearrangements or chromosome loss (Andrés Aguilera & García-Muse, 2012). The three-stranded structure provide a vulnerable ssDNA displaced strand as a mutagenic substrate for enzymes such as activation-induced cytidine deaminase (AID) or nucleases (Gómez-González & Aguilera, 2007). As a consequence, ssDNA gaps, replication stress or DSBs can be generated. The presence of R-loops at DSBs is controversial and need further studies to elucidate whether they are just obstacles that need to be removed for accurate repair or if they can participate in DSB repair (Andrés 35 Aguilera & Gómez-González, 2017; Brickner et al., 2022; Marnef & Legube, 2021). R-loop-mediated genome instability also occurs in cycling S-G2 cells and the most relevant mechanism by which this happen is the R-loop ability to stall RF progression that can lead to T-R conflicts and fork breakage (García-Muse & Aguilera, 2019). At stalled forks, the replisome remains associated to be able to restart without major consequences once the obstacle has been removed but when the fork arrest persist it can be disassembled prior to breakage. Defective progression of RF can lead to ssDNA breaks or DSBs as a consequence of replication stress and the S-phase checkpoint can be activated through the binding of replication protein A (RPA, a protein that coats ssDNA) or by DSB formation to ensure genome integrity (Andrés Aguilera & García-Muse, 2013). RF progression impairment in R-loop-accumulating cells has been demonstrated from bacteria to human by different techniques such as 2D gel electrophoresis or DNA combing and with tools like the overexpression of the ribonuclease RNase H1 (RNH1) that degrades the RNA moiety of the R-loop (Andrés Aguilera & García-Muse, 2012; Gómez-González et al., 2011; Herrera-Moyano et al., 2014; Salas‐Armenteros et al., 2017). Although R-loops can play a natural role in replication and recombination, for instance in the bacteriophage T4, in mitochondrial DNA or in Ig CSR of B cells, they can also cause important cellular problems such as chromosome fragility, anaphase bridges, sister-chromatid exchange (SCE), hyper-recombination, GCRs or chromosome loss. It is noteworthy that fragile sites which are defined as metaphase chromosomal regions prone to breaks or mutations upon replication due to their DNA repeats are more vulnerable to R-loop-mediated replication problems (Andrés Aguilera & García-Muse, 2012, 2013; Chakraborty et al., 2021). 1.3.3. R-loop prevention and resolution Cells are exposed to different exogenous and endogenous agents that can provoke DNA damage and genome instability that if unproper repaired can lead to mutations, cell death or cancer (Figure I1). R-loops although with physiological roles are also a source of genome instability and the cell have developed different mechanisms to prevent or resolve them. 36 Among the mechanisms that prevent R-loop formation we can highlight RNA biogenesis factors such as the THO complex or topoisomerases (Figure I4). On one hand, topoisomerases are the enzymes in charge of relax the supercoils derived from transcription and replication machineries avoiding R-loop formation in the more open DNA caused by negative supercoiling (Y. Liu et al., 2021; Sarni et al., 2022; Sutormin et al., 2022). On the other hand, mRNA biogenesis factors are needed to ensure a proper transcription, packaging and export of mRNA avoiding the rehybridization of the nascent mRNA to the DNA. Therefore, these factors are not only important for gene expression but also for genome integrity maintenance (Bentley, 2014; Y. A. Chan, Hieter, et al., 2014; García-Muse & Aguilera, 2019). The connection between RNA biogenesis and R-loops was first shown in yeast THO/TREX (Huertas & Aguilera, 2003) (more details in section 3.1). Next, the relationship between RNA biogenesis and genome instability was extended to other factors involved in splicing such as SRF1 (serine/argine splicing 1) (X. Li & Manley, 2005) or others (Paulsen et al., 2009), in mRNA 3’ end processing and degradation (Gavaldá et al., 2013; Luna et al., 2005; Stirling et al., 2012), in transcription and export such as UAP56 or THSC/TREX2 (González-Aguilera et al., 2008; Pérez-Calero et al., 2020) among others, that highlight the importance of mRNA structure and fate in R-loop prevention. Among the mechanisms involved in R-loop resolution it is necessary to highlight ribonucleases (Figure I4), in particular RNase H widely used in this study. First of all, RNH specifically degrades the RNA moiety of DNA-RNA hybrids and there are two types: RNase H1 which is monomeric and removes DNA-RNA hybrids like Okazaki fragments, and RNase H2 which is monomeric in bacteria, but is composed of three subunits in eukaryotes and is involved in ribonucleotide excision repair (Cerritelli & Crouch, 2019). Although both can remove DNA-RNA hybrids, RNH1 is the main player removing co-transcriptional R-loops (Cerritelli & Crouch, 2009; Chon et al., 2013). Thus, RNase H1 overexpression is widely used to remove R-loops and suppress R-loop- dependent genome instability phenotypes. R-loops can also be resolved by DNA-RNA helicases (Figure I4) able to unwind these structures. DNA-RNA helicases are enzymes that use ATP to 37 unwind double-stranded DNA-DNA, RNA-RNA or DNA-RNA hybrids, what is essential for different processes involving nucleic acid metabolism (Singleton et al., 2007; Valentini & Linder, 2021). Although there are many examples of helicases related to R-loop homeostasis such as DDX23, DDX19, RECQL5, DDX5, DDX21, DDX18, DDX41 (García-Muse & Aguilera, 2019; Lin et al., 2022), here we highlight two of them: Sen1/SETX and Sub2/UAP56. Sen1/SETX is a conserved helicase (yeast/human gene) widely studied in relation with R-loop metabolism and cells deficient in this helicase show hyper-recombination phenotype (Mischo et al., 2011) and R-loop accumulation at transcription termination pause sites (Skourti-Stathaki et al., 2011). Human SETX resolves DNA-RNA hybrids at DSB in order to limit translocations and ensure cell viability (Cohen et al., 2018). Recently, genome-wide data indicate that Sen1 is an S- phase specific R-loop resolvase (San Martin-Alonso et al., 2021). The other example, the conserved RNA helicase Sub2/UAP56 that interacts with THO complex has been described as a major co-transcriptional DNA-RNA helicase that unwinds harmful R-loops genome-wide and its overexpression can rescue R-loop-mediated genome instability phenotypes (Pérez-Calero et al., 2020). Finally, there are other factors that indirectly contribute to resolve R-loops and most of them have a role in DNA repair and/or mediation of T-R conflicts. For instance, transcription-coupled nucleotide excision repair (TC-NER) factors may contribute to mitigate R-loop-induced DNA damage, specifically, the TC-NER nucleases XPG and XPF could excise R-loops being actively processed into DSBs (Sollier et al., 2014; Yasuhara et al., 2018). BRCA1 and BRCA2 DSB repair factors can regulate RNAPII transcription elongation preventing R-loop accumulation (Bhatia et al., 2014; Hatchi et al., 2015; Shivji et al., 2018; X. Zhang et al., 2017). Cells lacking Fanconi Anemia (FA) factors show an R-loop accumulation and genome instability phenotype (M. L. García-Rubio et al., 2015; Schwab et al., 2015). In addition, different screenings have been performed to identify new factors involved in R-loop metabolism (Y. A. Chan, Aristizabal, et al., 2014; Cristini et al., 2018; Marchena-Cruz et al., 2023; Sollier et al., 2014) (Figure I4). 38 1.3.4. R-loops and chromatin Recently, it has emerged the importance of chromatin factors and chromatin state in R-loop metabolism (Figure I4). Indeed, it is unclear if an R-loop structure is enough to provoke RF stalling or whether a more complex chromatin structured is involved. Harmful R-loops have been associated with a condensation chromatin mark H3S10P in yeast, C. elegans and human cells that opens the possibility that R-loops could trigger local compaction leading to replication stress and derived consequences (Castellano-Pozo et al., 2013; García-Pichardo et al., 2017). Although the relationship between chromatin structure and R-loop metabolism is still under study, the crosstalk between RNA-binding factors and chromatin modifiers with a major role in preventing R-loop formation and RNA- mediated genome instability seems to be clear. Specifically, the chromatin state is important during transcription allowing the binding of the transcription machinery but also allowing a closed chromatin structure that avoid nascent RNA to rehybridize and forming R-loops. Therefore, a dynamic crosstalk between transcription factors and chromatin-related factors is needed. This is possible for example due to the co-transcriptional interaction between the THO complex and the Sin3A histone deacetylase that achieve a co-transcriptional closed chromatin state (Salas‐Armenteros et al., 2017), the Tip600-p400 histone acetyltransferase complex that is associated with genes harboring promoter-proximal R-loops and controls the genome-wide occupancy of the PRC2 histone methyltransferase (P. B. Chen et al., 2015). Other proteins such the FACT (facilitates chromatin transcription) complex that reorganizes chromatin allowing the progression of transcription and replication machineries and nucleosome reassembly at T-R collisions with a major role at regions enriched in R-loops (Herrera-Moyano et al., 2014). In addition, there are some chromatin remodelers such as the ATRX chromatin remodeling factor that suppress R-loops at transcribed telomeres (D. T. Nguyen et al., 2017) or the SWI/SNF complex that helps to resolve R-loops at T-R conflicts and whose subunit BRG1 is the most frequently mutated Snf2-like ATPase in cancer (Bayona-Feliu et al., 2021). 39 Figure I4. Factors involved in R-loop prevention and resolution. Examples of different categories of proteins involved in R-loop prevention (upper panel) or resolution (lower panel) are shown. Figure adapted from (García-Muse & Aguilera, 2019). 1.3.5. R-loops and disease The importance of R-loops on cancer seems to be clear due to the fact that R- loops are a source of genome instability which is one hallmark of cancer cells. This correlation is reinforced by the fact that genes encoding different factors involved in R-loop metabolism are mutated in cancer, as BRG1 in different tumors (Bayona-Feliu et al., 2021), BRCA1/BRCA2 in breast cancer (Bhatia et al., 2014; X. Zhang et al., 2017), ORF57 protein expressed in Kaposi’s sarcoma associated herpervirus-infected cells (Jackson et al., 2014) and EWS-FLI fusion protein in Ewing’s sarcoma patient cells (Gorthi et al., 2018). In addition, mutations of R- loop and DNA repair related factors are also found, as proteins of Fanconi Anemia pathway (M. L. García-Rubio et al., 2015). Nevertheless, cancer has not been the only disease with which R-loops have been linked. R-loops have been also associated to several neurological disorders. More than 40 genetic disorders are caused by gene-specific repeat expansions like Huntington’s disease or myotonic dystrophy type 1 (Richard & Manley, 2017). The formation of R-loops at expanded trinucleotide DNA repeats 40 has been associated with heterochromatin formation and transcriptional repression of genes linked to neurological disorders including Friedreich’s ataxia and fragile X syndrome (Groh et al., 2014). Mutations in SETX has been also linked to multiple neurological disorders (Groh et al., 2017) and amyotrophic lateral sclerosis type 4 (ALS4) patient cells have a decreased R-loop levels due to a gain of helicase function in SETX that leads to gene silencing (Grunseich et al., 2018). In addition, other connections between (ALS) and R-loops has been described (Belzil et al., 2013; Giannini et al., 2020). R-loops have been also related to autoimmune diseases, specifically to Aicardi-Goutières syndrome (AGS) which is a rare inflammatory disorder that affects the brain, skin and immune system and these AGS cells show elevated DNA-RNA levels that can contribute to the pathology altering gene expression or reactivating retroelements (Cristini et al., 2022; Lim et al., 2015). 2. DSBs as a relevant DNA lesion There are numerous types of DNA lesions that can lead to genome instability. The most common DNA lesions are ssDNA breaks or nicks whereas DSBs, in which the phosphate of the two complementary DNA strands are broken simultaneously, are less frequent but extremely cytotoxic and can lead to cell death when they are not properly repaired. There are programmed DSB that arouse at non-random hotspots (Mehta & Haber, 2014) like the ones produced in mammals during immunoglobulin class switching (Soulas-Sprauel et al., 2007) or during meiosis to promote recombination and faithful segregation (Lam & Keeney, 2015; Mehta & Haber, 2014). However, non-programmed DSBs can occur being really dangerous for genome integrity and the cell has evolved different repair pathways to deal with these lesions (Andrés Aguilera & García- Muse, 2013; Tubbs & Nussenzweig, 2017). 2.1. DSBs formation DNA is continuously threated by exogenous and endogenous sources of DSBs. Among exogenous damaging sources are included Ultraviolet (UV) light, ionizing raditation and carcinogenic chemicals such as DNA-alkylating agent (e.g., methyl methanesulfonate (MMS)), cross-linking agents (e.g., mitomycin-C or cisplatin), 41 replication inhibitors (e.g., hydroxyurea (HU) or aphidicolin) and radiomimetic compounds (e.g., neocarzinostatin) which are potential sources of DSBs (Mehta & Haber, 2014; Taylor et al., 2002). Even in the absence of exogenously inflicted DSBs, DNA is vulnerable due to the cellular metabolism itself, where processes as replication, transcription or recombination can result in DSBs (Andrés Aguilera & Gaillard, 2014; Tubbs & Nussenzweig, 2017). During transcription, nicks can be converted into DSBs being replicationborn DSBs, what can be the reason why high number of spontaneous DSBs appear during replication whereas the most common DNA lesions are nicks (Lindahl & Barnes, 2000; Mehta & Haber, 2014). Replication process counts with different factors that can provoke DSBs. Among them we can find topoisomerases, that relieve torsional stress by transient breaks allowing the replisome advance but also when they fail, it leads to the accumulation of covalently DNA-linked cleavage complexes and DNA damage (Pommier et al., 2016). In addition, we can find helicases associated with the replisome that can bypass the lesion and allow replication-coupled repair overcome the obstacles (Cortez, 2019; Ivessa et al., 2003). Whether the problem persist, nucleases or the human deaminase APOBEC can act and lead to DSBs (Buisson et al., 2017; Cortez, 2015). During these cellular processes, there are regions along the genome more vulnerable to suffer damages and they are known as Common fragile sites (CFSs). CFSs are usually long active genes that replicate late and whose replication can persist in mitosis leading to unscheduled replication process called MiDAS (mitotic DNA synthesis) or even to ultrafine anaphase bridges that could break in mitosis. Thus, these regions are prone to suffer replicative stress and DSBs (K. L. Chan & Hickson, 2011; Gaillard et al., 2015; Marnef et al., 2017; Minocherhomji et al., 2015). In addition, the chromatin context has influence not only in the occurrence of R-loops (See section 1.3.4) but also in DSBs formation and repair. DSBs form preferentially at accessible chromatin such as active genes (Allen et al., 2003; Chaurasia et al., 2012; C. S. Lee et al., 2014) but although compacted DNA has lower number of lesions its chromatin state can interfere with signaling and repair factors accessibility, indeed, heterochromatin 48 et al., 2018) or I-sceI (L. Li et al., 2016). However, the possible role at DNA lesions is still a matter of debate, being considered an obstacle that need to be removed for proper repair or an intermediate with a role in repair (Andrés Aguilera & Gómez-González, 2017; Brickner et al., 2022; Crossley et al., 2019). Transcription of non-coding RNA has been proposed to be important for DDR in different organisms (Francia et al., 2012; Michalik et al., 2012; Wei et al., 2012) and RNA-binding proteins (RBPs) and transcription factors have been shown enriched at damaged sites (D’Alessandro & d’Adda di Fagagna, 2017; Sonia Jimeno et al., 2019). There are cases in which the RNA molecule serves as a homologous template for recombination (Keskin et al., 2014). Since 5’-end DSB resection is a key step toward HR, DNA-RNA hybrid might promote HR over NHEJ. However, there is no evidence supporting the need of hybrid formation for DSB repair, indeed, DSBs can be repaired in vitro in the absence of RNA. Moreover, the RNAPII enrichment seen by ChIP at DSBs does not necessarily mean that they have been recruited actively. On the other hand, there are evidences supporting the need of removing this hybrids at DSBs by helicases such as DDX1 (L. Li et al., 2016), senataxin (Cohen et al., 2018), DDX5(Sessa et al., 2021) or DDX18 (Lin et al., 2022). In order to de novo transcription induction at DSBs, it has been reported an active role facilitating DSB repair and its disruption inhibits DDR, the recruitment of BRCA1, BRCA2 and rad51 and thus, HR and NHEJ (D’Alessandro et al., 2018; Francia et al., 2012; S. Liu et al., 2021; W. T. Lu et al., 2018). Transcription at breaks favors resection and protect 3’ ssDNA from degradation (S. Liu et al., 2021). Taking all together, the possible role(s) of DNA-RNA hybrids at DSBs remains to be stablished. 2.3. New techniques to study DSBs-related processes Although DSB sources are abundant and their nature different, the type of DNA damage produced is similar (De Bont & van Larebeke, 2004). Therefore, different tools to provoke DSBs and study their repair have been used. Until recently, they were studied through imprecise methods such as ɣ IR or restricted to studying predefined genomic regions for example with systems containing endonuclease sites like I-SceI. However, new techniques for detecting DSBs with next- 49 generation sequencing on a genome-wide scale have emerged (Rybin et al., 2021). There are already many protocols to measure DSB indirectly or directly with lower resolution or with a nucleotide resolution. Among them, chromatin immunoprecipitation and sequencing (ChIP-seq) of DNA fragments associated with DNA repair machineries, such 53BP1 or үH2AX has been used as an indirect method to detect DSBs genome-wide. Other low-resolution techniques such DSB-seq, Break-seq, among others are based on the labeling of DSBs with biotinylated nucleotides using the terminal deoxynucleotidyl transferase (TdT) DNA polymerase, followed by immunoprecipitation and sequencing of the captured fragments with an average resolution of ~100-300 base pairs (bp). For nucleotide resolution DSB detection, several techniques have been developed as for example BLESS (direct in situ break labeling, enrichment on streptavidin and next-generation sequencing) (Crosetto et al., 2013). Finally, for this work it is worth to mention two more techniques. The first one is based on the AsiSI endonuclease restriction which is a common validation method used for genome-wide break mapping studies due to the high efficiency of this endonuclease with more than 1200 predicted cleavage sites in the human genome (Rybin et al., 2021). DSB Inducible via AsiSI system, also known as DIvA cells, relies on the expression of this restriction enzyme fused to an oestrogen receptor that allow its nuclear localization from the cytoplasm when 4- hydroxitamoxifen (4-OHT) is added and have been used to study DSBs and their repair genome-wide with statistical power (Clouaire et al., 2018; Iacovoni et al., 2010; Massip et al., 2010). The second one is based on the fact that most DSB repair factors accumulate at or in the vicinity of DSB sites, even forming discernible foci which is dissociated when the repair is completed. Microirradiation in cells with fluorescently tagged factors have been used as a tool to study in a site-specific manner the accumulation of these factors at the damaged area in real-time. Changes in the kinetics of the factors studied in response to DSBs can provide useful information about their repair pathway involvement, their interactions with other proteins and the spatio-temporal organization of repair factors in general. With this aim, a meta-analysis of DSB response kinetics was performed describing the accumulation kinetics of different 50 DSB repair proteins whose results were available, providing a temporal organization of them in mammalian cells (Kochan et al., 2017). 3. ALY and other mRNA related factors The generation of RNA from the genetic information stored as DNA is essential for cellular survival and cells have developed different tools to guarantee biogenesis and metabolism. The nascent messenger RNA (mRNA) needs to be properly processed and packaged into a competent messenger ribonucleoprotein (mRNP) and numerous RNA-binding proteins (RBPs) are recruited cotranscriptionally for this aim (Mitchell & Parker, 2014). The coupling of transcription with RNA processing and export is essential for gene expression in eukaryotes, thus mRNA capping, splicing and 3’-end formation are processes that take place in a co-transcriptional manner and are integrated with quality controls (Bentley, 2014; Peck et al., 2019). Failures in RNA metabolism have an impact on gene expression, but cumulative evidence show that are also linked to genome instability (Andrés Aguilera, 2005). Mutations in mRNP factors lead to genome instability such as DNA damage, recombination or GCRs, as reported in yeast and human cells (Luna et al., 2005; Paulsen et al., 2009; Stirling et al., 2012; Wahba et al., 2011). In some cases, this genome instability has been associated to the formation of R-loops (García-Muse & Aguilera, 2019; Huertas & Aguilera, 2003; X. Li & Manley, 2005). 3.1. THO complex, UAP56 and ALY Among the numerous important factors involved in the mRNP biogenesis we can highlight the THO complex (Figure I7). This is a conserved complex, first isolated in yeast as a four-subunit complex (Tho2, Hpr1, Mft1 and Thp2) (Chávez et al., 2000). THO has been also purified in human cells containing Hpr1/THOC1, Tho2/THOC2 and Tex1/THOC3, the most evolutionary conserved subunits, together with THOC5, THOC6 and THOC7 (Masuda et al., 2005; Rehwinkel et al., 2004). Although their sequences present no homology, structural data suggest that THOC5 and THOC7 could be the Thp2 and Mft1 functional homologues (Pühringer et al., 2020; Ren et al., 2017). THO is considered a structural and functional unit, since the core of the complex has been purified as a salt-stable complex in yeast, Drosophila and human cells, and depletion of the 51 subunits have been shown to confer similar phenotypes in yeast (Chávez et al., 2000; M. García-Rubio et al., 2008; Masuda et al., 2005; Rehwinkel et al., 2004) In addition, THO associates with the DEAD-box RNA helicase Sub2/UAP56 and with the mRNA export factor Yra1/ALY (yeast/human gene names) forming a larger complex called TREX (TRanscription-Export) (K. Sträßer et al., 2002) (Figure I7). The core of the THO complex is stable whereas the interactions with Sub2/UAP56 and Yra1/ALY are weaker (Masuda et al., 2005; Rehwinkel et al., 2004). The three of them are co-transcriptionally recruited to the transcribed region partially overlapping with RNAPII. It has been suggested that THO could facilitate the loading of Sub2/UAP56 and Yra1/ALY onto the mRNA (Rondón et al., 2010). THO complex is recruited to active chromatin by redundant mechanisms (H. Cheng et al., 2006; Gómez-González et al., 2011; Masuda et al., 2005) and its interactions with Sub/UAP56 and Yra1/ALY and other mRNPs facilitates the coupling of transcription with export and mRNA metabolism. Indeed, THO mutants show transcription-elongation impairment, export defects, increased mRNA instability and defects in 3’mRNA formation (Luna et al., 2019). In addition THO mutants show transcription-dependent hyper-recombination phenotypes, increased DNA damage, replication defects and increased DNA-RNA hybrid formation (Chávez et al., 2000; Domínguez-Sánchez, Barroso, et al., 2011; Huertas et al., 2006; S. Jimeno et al., 2002; Y. Li et al., 2005; Rondón et al., 2003; Viphakone et al., 2012). The role of this complex in the maintenance of genome integrity is conserved in yeast, C. elegans and human cells (Castellano-Pozo et al., 2013; Chávez et al., 2000; Domínguez-Sánchez, Barroso, et al., 2011; Huertas et al., 2006; Huertas & Aguilera, 2003). As mentioned before, the proposed model is that THO would be needed to ensure a proper transcription, packaging and export of mRNA avoiding the rehybridization of the nascent mRNA to the transcribed DNA (Luna et al., 2019). The THO interactor Sub2/UAP56 is an ATP-dependent RNA helicase (K. Sträßer et al., 2002) firstly identified as a splicing factor (Fleckner et al., 1997). However, it seems to play a more general role in mRNP biogenesis unwinding secondary structures (Katahira, 2015) and in mRNA export, being essential for 52 most mRNAs in D. melanogaster, S. cerevisiae, C. elegans and human cells (Herold et al., 2003; Kapadia et al., 2006; Macmorris et al., 2003). In addition, Sub2/UAP56 is a relevant protein preventing R-loop-mediated genome instability. Indeed, Sub2/UAP56 mutants lead to similar genome instability phenotypes than THO mutants and in yeast was seen that SUB2 overexpression partially suppress the hpr1Δ mutants of the THO complex (González-Aguilera et al., 2008; S. Jimeno et al., 2002). In human cells, UAP56 has been shown to be a major cotranscriptional DNA-RNA helicase that unwinds harmful R-loops genome-wide (Pérez-Calero et al., 2020). Interestingly, UAP56 also known as DDX39B has a paralog in humans called DDX39A or URH49. Both share ~90% of amino acid homology but URH49 is expressed in specific tissues whereas UAP56 shows a ubiquitous expression (Pryor et al., 2004). Both helicases seem to play similar roles in RNA export and in mammalian cells only the simultaneous depletion causes a substantial retention of mRNAs in the nucleus suggesting a redundant function of these proteins (Kapadia et al., 2006). Whereas UAP56 interacts with ALY and THO forming the TREX complex, URH49 interacts with THO core complex and with the RBP CIP29 in a different complex called AREX (alternative mRNA export complex) (Dufu et al., 2010). ALY, the other main THO-interacting protein, is a conserved mRNA export factor that will be described in the following sections. THO/TREX complex is a paradigm in the connection of mRNA transcription and biogenesis with genome instability, but factors involved in RNA metabolism have been also associated with genome instability highlighting the importance of proper mRNP assembly preventing RNA re-hibrydization to DNA template that could compromise genetic instability. Thus, different screenings have led to the identification of factors that could be associated with DNA-RNA hybrid accumulation, such as proteins involved in transcription, chromatin modification, nuclear export, RNA processing and surveillance (Y. A. Chan, Aristizabal, et al., 2014; Paulsen et al., 2009; Stirling et al., 2012; Wahba et al., 2011). 53 Figure I7. THO/TREX complex in mRNA biogenesis. THO complex interacts with UAP56 and ALY. These proteins are recruited to the transcribed chromatin and play a role in transcription elongation, 3’-end processing and mRNA export. ALY interacts with TAP export protein and allows the export of the mRNP. TAP accompanies the mRNP through the nucleopore from the nucleus to the cytosol whereas ALY and other factors remains inside the nucleus. In the cytosol, mRNA translation takes place. Adapted from (Katahira, 2012). 3.2. ALY is a conserved RNA binding protein Yra1/ ALY is an abundant protein that belongs to the RNA export factor (REF) family (Stutz et al., 2000) with a role in mRNA export that is conserved from yeast to human cells (Katahira et al., 2009; T. Sträßer & Hurt, 2001) . Yra1/ALY was shown to interact with the RNA helicase Sub2/UAP56 forming the conserved TREX complex, that serves as a link of transcription and export (Luna et al., 2019; K. Sträßer et al., 2002). In yeast, Yra1 contains two highly conserved motifs at the N- and C-terminal ends involved in Sub2 binding, a central RNA-recognition motif (RRM) and between them there are regions with sequences enriched in arginine and glycine (RGG boxes) that has been described in many RNA-binding proteins with variable lengths (N-vr; C-vr) (Figure I8A) (Burd & Dreyfusst, 1994; Ren et al., 2017; Stutz et al., 2000). Similar and conserved domains are found in human ALY protein with UAP56-binding motifs (UBM) at the N- and C-terminal ends, a central 54 RRM and RGG boxes between them which are defined as the RNA-binding domains (RBD) (Pühringer et al., 2020) (Figure I8A). RNA-binding ability of Yra1 lies in N-var and C-var regions rich in RGG, but not in the canonical RRM ALY (Ren et al., 2017; Zenklusen et al., 2002) whereas in human cells, at least in vitro, it has been described that neither the RGG-rich regions nor the RRM is able to bind RNA by itself and that ALY binding to UAP56 seems to be dispensable for its interaction with mRNA (Gromadzka et al., 2016). The association of this well-conserved protein with different cellular processes will be discussed in next sections. Recent THO/TREX structural data in yeast and human cells have revealed new insights of the functional conserved domains of these proteins (Pühringer et al., 2020; Ren et al., 2017; Schuller et al., 2020). Whereas the yeast THO complex in association with Sub2 forms a dimer (Figure I8C), human THO complex in association with UAP56 forms a tetramer of two asymmetric dimers (Figure I8D). Tetramerization may have evolved in response to an increased complexity in gene architecture and mRNP composition in human cells. This tetramerization could be relevant since mutations in THOC6, the human subunit in charge of generating the tetramer conformation, is known to be mutated in human diseases. As deduced from these structural analysis, a single Yra1/ALY protein could bind one UAP56 or bridge two UAP56 proteins, within one dimer or possibly multiple THO-UAP56 tetramers in the case of the human complex (Pühringer et al., 2020) (Figure I8C-D). Thus, shows conserved motifs and structural domains involved in RNA binding and interaction with THO and UAP56 quite similar to those of Yra1. 55 Figure I8. Yra1/ALY is a conserved protein from yeast to humans. A) Scheme of Yra1 and ALY orthologs from yeast and human cells, respectively. Both proteins show similar lengths with Sub2/UAP56-binding domains (UBM) in blue and a central RNA recognition motif (RRM) in green. Besides among them and in yellow are represented the RRG-rich regions which are called N-vr and C-vr in yeast and correspond to the RNA-binding domains of human ALY protein (RBD) (Gromadzka et al., 2016; T. Sträßer & Hurt, 2001). B) Front view of yeast TREX complex where it is possible to see the ability of a single Yra1 to interact with two molecules of Sub2. The model is based on the combined revised chimeric yeast THO-Sub2 complex crystal structure and the Sub2-RNA-Yra1 structure. Adapted from (Pühringer et al., 2020). C) Schematic model of human TREX complex where a single ALY could interact with one or two different UAP56 molecules. The model was obtained by superimposing the human homology model THO-UAP56 of the yeast Sub2-Yra1-RNA crystal structure. Adapted from (Pühringer et al., 2020). Biochemical and genetic evidences supports that Sub2/UAP56 is more tightly associated with THO than Yra1/ALY (Gómez-González et al., 2011; S. Jimeno et al., 2002; Masuda et al., 2005; Schuller et al., 2020). It has been shown that UAP56 mediates the association of ALY with the THO complex in an ATP- dependent manner (Dufu et al., 2010; M. J. Luo et al., 2001) suggesting that by remodeling the RNA, UAP56 could facilitate the action of ALY (Luna et al., 2019). Although ALY can bind RNA per se, having more preference for mRNAs than other RNAs species, the interaction with ATP-bound UAP56 allows a more efficient RNA binding (Taniguchi & Ohno, 2008). On the other hand, it is reported 56 that the presence of RNA and ALY stimulates the ATPase and RNase helicase activities of UAP56 (C. Te Chang et al., 2013; Ren et al., 2017; Taniguchi & Ohno, 2008). 3.3. ALY role in transcription and mRNA export ALY is an RNA-binding factor relevant for transcription and mRNA export and its main roles during mRNA biogenesis will be discussed in this section. Yra1/ALY acts as an export adapter, binding mRNA to be transferred to the heterodimer Mex67-Mtr2/TAP-p15, a general mRNA receptor that mediates RNA export through the nuclear pore (Figure I7) (Hautbergue et al., 2008; Viphakone et al., 2012). In human cells, ALY interacts with UAP56 and THO complex and colocalizes with splicing machinery at nuclear speckles (Masuda et al., 2005). Evidence suggest that pre-mRNA undergoes splicing in these nuclear domains, before their release by TREX components that mediates efficient export to the cytoplasm (Dias et al., 2010). Thus, ALY links different steps of RNA metabolism. The role of ALY as an adaptor for mRNA export is based on its ability to bind RNA (See section 3.2) and its interactions with the helicase UAP56 and the THO complex (Figure I8D) that could contribute to ALY recruitment and function in transcription-export process but also in the receptor TAP which is in charge of the mRNA export to the cytosol (Figure I7). Mutually exclusive interactions among factors could favor the transition from transcription to export, indeed, in yeast, the binding of Mex67 and Sub2 is mutually exclusive and the recruitment of Mex67 is thought to trigger the dissociation of Sub2 from the mRNP (Ren et al., 2017; K. Sträßer et al., 2002; T. Sträßer & Hurt, 2001). In human cells it has been shown that upon the formation of a ternary complex with TAP, ALY and RNA, the RNA is transferred from ALY to TAP (Hautbergue et al., 2008; Katahira, 2015; Taniguchi & Ohno, 2008). Finally, Mex67/TAP can bind to nucleoporin and shuttles to the cytoplasm allowing the RNA export (Figure I7) (Katahira, 2012). Although the importance of ALY in mRNA export is clear, in higher eukaryotes ALY knockdown only leads to a modest mRNA export reduction in contrast to the stronger mRNA export block shown by UAP56 or TAP knockdown 57 or the impact of reduced levels of Yra1 in yeast which causes an accumulation of polyA RNAs inside the nucleus. This phenomenon could be explained by additional export adaptors (Delaleau & Borden, 2015). Interestingly, ALY promotes the export of mRNAs that encode heat shock 70 kDa proteins (HSP70s) which are crucial factors in the response to heat stress (Katahira et al., 2009) and ALY also controls the nuclear export of transcripts encoding essential proteins for accurate genome duplication and DNA repair. Transcripts for RAD51, CHEK1 or FANCD2 involved in HR are included in the subset whereas transcripts from NHEJ are unaffected. However, the selective transport in observed with or without DNA-damaging agents exposure (Wickramasinghe & Laskey, 2015). Phosphoinositide signaling can regulate RNA export and interestingly, ALY is a physiological target of nuclear phosphoinositide-3-kinase (PI3K). The phosphorylation of ALY by this kinase is relevant for proper ALY location at nuclear speckles and function in mRNA export (Okada et al., 2008). Other posttranslational modifications of ALY can be also important, such as arginine methylation that reduces the RNA-binding ability of ALY in vitro and in vivo (Hung et al., 2010). The importance of ALY is not limited to its role in mRNA export, being also critical for gene expression at the level of transcription. Reduced levels of ALY leads to a diminished RNAPII occupancy and density of its elongation competent form finally causing lower transcription rates (Stubbs & Conrad, 2015). In addition, ALY can stabilize certain mRNAs as determined by UV crosslinking and immunoprecipitation (CLIP) technique. It was found that ALY stabilizes MYC mRNAs and that ALY is regulated by MYC at the transcriptional level suggesting an ALY-MYC positive feedback loop (J. Wang et al., 2021). Finally, it is noteworthy that ALY is the main reader of the m5C post-transcriptional RNA modification identified in stable and abundant tRNAs, rRNAs and mRNA whose main writer is the RNA methyltransferase NSUN2 (X. Yang et al., 2017). Thus, ALY as a reader of methylated RNA could link mRNA export and posttranscriptional regulation. 64 1. Effect of ALY decreased levels in human cells 65 66 To get further insight into the connection between mRNP biogenesis and genome instability, we decided to study the impact of ALY depletion in human cells that could help us to better understand how this mRNA binding protein contributes to the maintenance of genome integrity. 1.1. Analysis of cell cycle progression in ALY depleted cells It has been previously reported that the reduction of the levels of mRNA binding factor ALY results not only in transcription and mRNA export defects but also increases genome instability (Domínguez-Sánchez, Barroso, et al., 2011). To try to clarify the underlying mechanism, we decided to study whether an alteration in the cell cycle progression could contribute to the genome instability phenotype observed. To address this question, we carried out flow cytometry analysis in U2OS cells depleted of ALY via small interfering RNAs. First, we checked the efficiency of a siRNA pool against ALY by western-blot. A clear reduction of ALY protein levels was observed after 72 hours of transfection (Figure R1A). Then, we analyzed the cell cycle distribution of asynchronously growing siALY- transfected U2OS cells incubated with DAPI and 5-Ethynyl-2’-deoxyuridine (EdU), a thymidine analogue that is incorporated into the DNA of dividing cells. Figure R1. Cell cycle progression analysis in ALY-depleted cells. (A) Western blot analysis showing ALY expression levels of U2OS cells transfected with control and ALY siRNAs. Actin protein was used as a loading control. (B) FACS analysis of asynchronously growing U2OS cell cultures of control (siC) and ALY depleted cells (siALY) after 20 min of incubation with 20mM 5’- ethynyl-2-deoxyuridine (EdU). The percentages of cells in G1, S and G2 phases are depicted in upper panel and the EdU intensity from EdU positive cells is shown in the lower panel. Data are plotted as Mean ± SEM (n=3) (two-tailed paired t-test). 67 Any significant difference was observed among the percentages of cells in G1, S or G2, neither in the intensity of EdU labelling (Figure R1B). We reasoned that the cell cycle progression was not responsible for the genome instability observed in ALY-depleted cells. 1.2. Genome instability upon ALY depletion ALY depletion was previously shown to induce an increase in DNA damage as determined by үH2AX foci and single-cell gel electrophoresis (comet assay) in HeLa cells (Domínguez-Sánchez, Barroso, et al., 2011). Since ALY is a mRNA factor associated to THO, a complex that prevents R-loop formation and genome instability (Luna et al., 2019), we wondered whether the genome instability phenotype associated to ALY knockdown was due to co-transcriptional DNARNA hybrids. 1.2.1. ALY depletion leads to an R-loop-dependent genome instability phenotype To determine whether the genome instability caused by ALY depletion was mediated by R-loops we measured DNA damage in U2OS siALY-depleted cells overexpressing RNase H1 (RNH1), a ribonuclease that specifically degrades the RNA moiety of DNA-RNA hybrids. Immunofluorescence (IF) assays in siALY cells showed a significant decrease of үH2AX foci upon overexpression of RNH1 (Figure R2A). Since this decrease was not observed in siC cells, data suggest that DNA damage in siALY could be due to DNA-RNA hybrid accumulation, despite the lower number of үH2AX foci per cell observed respect to control cells. Indeed, an impairment of DNA damage signaling in U2OS upon deletion of ALY cannot be discarded. In order to explore better this phenotype, DNA damage was directly assessed by single-cell gel electrophoresis (Comet assay) by which, following DNA unwinding under alkaline conditions, broken DNA fragments (damaged DNA) migrate away from the nucleus. Comet analysis in U2OS cells revealed that ALY depletion increased DNA breaks, in agreement with previous published data in HeLa cells (Domínguez-Sánchez, Barroso, et al., 2011). Moreover, DNA breaks in siALY were partially suppressed by RNH1 overexpression (Figure R2B). 68 Figure R2. Analysis of DNA damage in siALY U2OS cells. (A) Representative images of immunostaining with үH2AX (red) antibody in U2OS cells upon ALY knockdown. Quantification of the number of үH2AX foci per cell in U2OS transfected cells with the indicated siRNAs and with or without RNase H1 overexpression is represented (n=3). Scale bars 25 µm. The black line in the plot indicates the median. Black stars denote significant increases, whereas red stars denote significant decreases. **P < 0.01 (Mann-Whitney U-test, two-tailed). (B) Representative images of single-cell alkaline gel electrophoresis (comet assay) of U2OS cells upon ALY depletion with and without RNase H1 overexpression. Data are plotted as mean of the medians ± SEM (n=7). *P < 0.05 (one-tailed paired t-test). (C) Representative images of immunostaining with FANCD2 (green) antibody in U2OS cells upon ALY knockdown with or without RNH1 overexpression. The graph shows the quantification of the percentage of cells containing > 5 FANCD2 foci. Around 100 cells were scored in each experiment. Data is plotted as Mean ± SEM (n=3) (two-tailed paired t-test). Cumulative evidence indicate that replication fork (RF) stalling is a major cause of R loop-dependent DNA damage (Andrés Aguilera & García-Muse, 2012; Hamperl et al., 2017). In order to study whether ALY depletion leads to RF 69 blockage, we examined the levels of FANCD2 foci by IF since Fanconi Anemia factors accumulate at RF blocked caused by R-loops (M. L. García-Rubio et al., 2015). ALY depletion in U2OS cells did not lead to an accumulation of FANCD2 foci (Figure R2C). We concluded that the DNA damage caused by ALY depletion is not associated to RF stalling. In order to corroborate the increase in DNA damage caused by ALY depletion in HeLa cells and to study the dependence on DNA-RNA hybrids we checked the number of үH2AX foci per cell by immunofluorescence with or without RNH1 overexpression. In agreement with prior data published, ALY depletion in HeLa cells caused an increase in the percentage of cells with more than 5 үH2AX foci and in addition we saw that it was significantly reduced with RNH1 overexpression (Figure R3). Altogether data indicate that ALY depletion in human cells leads to DNA damage that can be mediated by R-loops. Figure R3. Analysis of DNA damage in siALY HeLa cells. Representative images of immunostaining with үH2AX (green) antibody in siALY HeLa cells. Quantification of the percentage of cells with more than 5 үH2AX foci in HeLa transfected cells with the indicated siRNAs and with or without RNase H1 overexpression is represented. Data is plotted as Mean ± SEM (n=3) *P < 0.05 (two-tailed paired t-test). 1.2.2. ALY depletion leads to R loop accumulation We assayed whether R-loops were increased in U2OS siALY-depleted cells by immunofluorescence (IF) using the anti-DNA-RNA hybrid S9.6 monoclonal antibody. Since this antibody also recognizes dsRNA structures (Chédin et al., 2021; Silva et al., 2018; Smolka et al., 2021) an in vitro treatment with the ribonuclease RNaseIII, an enzyme that degrades dsRNAs was included. A significant enrichment of the nucleoplasm S9.6 signal was observed upon ALY 70 depletion. Importantly, this increase was suppressed by in vitro RNase H treatment, confirming that it corresponds to DNA-RNA hybrids (Figure R4A). Next, we studied R-loop accumulation at a molecular level by DRIP-qPCR analysis in the standard human genes (APOE and RPL13A) that have been previously validated for R-loop detection (Ginno et al., 2013; Herrera-Moyano et al., 2014) and in MIB2 gene, a validated region that shows lower basal levels of R-loops (Pérez-Calero et al., 2020). In vitro RNase H treatment was used in order to confirm the specificity of DNA-RNA hybrid immunoprecipitation. A significant DNA-RNA hybrid accumulation was observed in APOE and MIB2, and high levels of hybrids but no significant were detected in RPL13A (Figure R4B). We extended the analysis to HeLa cells. S9.6 IF also revealed a significant increase in siALY cells that was reduced by RNH1 treatment (Figure R4C). Therefore, we concluded that the genome instability phenotype caused by ALY depletion had a general impact on R-loop formation in both cell lines (U2OS and HeLa). 71 Figure R4. DNA-RNA hybrid accumulation in ALY depleted cell. (A) Representative images of immunostaining with S9.6 (red) and anti-nucleolin (green) antibodies in U2OS cells upon ALY knockdown. The median of S9.6 signal intensity per nucleus after nucleolar signal removal in siRNAs U2OS transfected cells and in vitro treated with RNase III and with or without RNase H is shown. Data from more than 200 total cells from three independent experiments is shown. ****P < 0.0001 (Mann-Whitney U-test, two-tailed). (B) Schematic diagrams of APOE, RPL13A and MIB2 genes. Exons are depicted as closed boxes, the arrows indicate the start of transcription and red lines indicate the regions where the primer pairs used for amplification were located. Relative DRIP-qPCR signal values in siALY U2OS at the indicated regions is shown in the lower panel. Samples were treated in vitro with RNase H prior immunoprecipitation where indicated. Signal values normalized with respect to the siC control are plotted as Mean ± SEM (n=5). *P < 0.05; **P < 0.01 (one-tailed paired t-test). (C) Representative images of immunostaining with S9.6 (green), anti-nucleolin (red) and anti-FLAG (cyan) antibodies in HeLa cells upon ALY depletion. Other details as in A. Data from 400 total cells from four independent experiments is shown. *P < 0.05; ****P < 0.0001 (Mann-Whitney U-test, one-tailed). 72 1.3. DNA damage response upon ALY depletion We have seen that the depletion of ALY leads to lower levels of үH2AX foci than siC cells (Figure R2A), which suggest a problem in DNA damage signaling. We wonder if ALY depletion had an impact in DNA damage repair. For this purpose, we studied үH2AX repair kinetics in siALY-depleted cells upon an exogenous DNA damage. Laser micro-irradiation was used to induce DNA double strand breaks (DSBs) and cells were collected at different time points. Next, immunofluorescence with antibodies against үH2AX and ALY proteins and DAPI stanning were used to measure DNA damage and ALY knockdown respectively. үH2AX intensity at the laser stripe was lower in siALY than in siC upon laser micro-irradiation (Figure R5). We also observed significant differences in үH2AX repair kinetics between siC and siALY U2OS transfected cells that could be due to a defect in үH2AX DNA damage signaling when ALY levels are reduced. Figure R5. үH2AX repair upon ALY depletion. DNA damage repair kinetics as determined by үH2AX IF (signal intensity) at laser stripe after laser photoablation in siC and siALY cells along time. Red stars denote significant decrease. Intensity fold change at the first time point after damage (15 minutes) has been normalized to 1 and data are plotted as Mean ± SEM (n=2 with 10 cells/experiment). **P < 0.01 (Two-way Anova). Next, we wondered whether in addition the recruitment of early factors involved in the main DNA damage repair pathways which are non-homologous end joining (NHEJ) and homologous recombination (HR) was altered. For this purpose, we depleted ALY and studied in vivo recruitment of two early factors, KU70 (NHEJ) and MRE11 (HR) measuring the location of GFP-tagged proteins at the photoablation laser stripe. KU70 is a subunit from the KU complex which is recruited at DSBs and considered essential for NHEJ (Abbasi & Schild-Poulter, 73 2019). We could not observe any difference of KU70-GFP recruitment at the laser stripe upon ALY depletion (Figure R6A). MRE11, a subunit of the MRN complex, is a single stranded DNA endonuclease and 3’→5’ exonuclease that functions in conjunction with other factors to generate the 3’ single-stranded DNA overhang necessary for strand invasion during HR. In addition, this factor also can participate in end processing, through its nuclease activity, in canonical NHEJ (cNHEJ) and in alternative NHEJ (a-NHEJ) (Anyong et al., 2009; Zha et al., 2009). The kinetics of MRE11-GFP showed that at the initial time upon DNA damage Mre11 was recruited at similar levels in siC and siALY, but later on MRE11-GFP seemed to be retained at DNA damage when ALY is knock-down, although the differences were not statistically significant (Figure R6B). From these results, we reasoned that KU complex was being recruited at DNA damage site and acting as expected, despite ALY depletion, whereas lower levels of ALY could have an impact increasing the persistence of the nuclease MRE11 that could alter DNA end processing. To address this question, we checked if ALY depletion has an impact on the initial and critical steps of resection that channel the repair to HR or NHEJ pathways. For this aim we measured RPA and CtIP recruitment at DNA damage site along time upon ALY depletion. We followed a similar procedure using laser microirradiation to induce DSBs and measured in vivo recruitment of GFP-tagged proteins. RPA coats the single stranded DNA that comes out at the resected ends at DNA damage repair (Galanty et al., 2012). When ALY was depleted, RPA70 was more recruited and maintained more time at damage site, although it was not significant (Figure R6C). CtIP is an endonuclease that cooperates with the MRN complex in DNA-end resection (Wijnhoven et al., 2015). We could not observe any difference of CtIP-GFP recruitment at the laser stripe upon ALY depletion (Figure R6D). These results indicate that ALY depletion could lead to an increase in single stranded DNA. 80 2.2. Analysis of genome instability upon ALY overexpression The overexpression of Yra1 in yeast causes DNA damage mediated by DNARNA hybrids (M. García-Rubio, Aguilera, et al., 2018; Gavaldá et al., 2016). Interestingly, ALY is highly detected in proliferative cells and overexpressed in a broad range of tumors suggesting an association of altered ALY expression with tumorogenesis (Domínguez-Sánchez, Sáez, et al., 2011). Since genome instability is a hallmark of cancer cells, we were interested in the study of the effect of ALY overexpression in genome instability in human cells. For this aim, we checked the levels of DNA damage in ALY-FLAG transiently transfected U2OS cells. After 24 hours of ALY overexpression we did not see an increase in DNA damage measured by immunofluorescence of үH2AX foci (Figure R9A) or in DNA breaks measured by single-cell gel electrophoresis (Figure R9B). Given the arrest in S phase observed in ALY overexpressing cells, we studied whether there are transcription-replication (T-R) conflicts upon ALY overexpression. T-R conflicts were analyzed by immunofluorescence of FANCD2 foci in ALY-FLAG transfected U2OS cells after 24 hours. Data showed a decrease of the percentage of cells with more than 5 FANCD2 foci respect to control cells (Figure R9C), probably linked to the replication impairment seen upon ALY overexpression. Altogether data indicate that transient ALY overexpression in U2OS cells did not lead to a genome instability phenotype, at least in the conditions assayed. 81 Figure R9. Genome instability analysis upon ALY overexpression. (A) Representative images of immunostaining with anti-үH2AX (red) and anti-FLAG (green) antibodies in U2OS cells overexpressing ALY after 24 hours transfection. Percentage of cells with more than 5 үH2AX foci is shown. Data are plotted as Mean ± SEM (n=4) (two-tailed paired t-test). (B) Representative images of single-cell alkaline gel electrophoresis (comet assay) of U2OS cells transfected for 24 hours with ALY-FLAG. Medians of comet-tail moments are plotted as Mean ± SEM (n=3) (twotailed paired t-test). (C) Representative images of immunofluorescence of U2OS overexpressing ALY with anti-FANCD2 (green) and anti-FLAG antibody (red) and DAPI staining. Percentage of cells with more than 5 FANCD2 foci is shown upon ALY overexpression. Data are plotted as Mean ± SEM (n=3). *P < 0.05 (two-tailed paired t-test). It has been described that in vitro, Yra1 binds to DNA-RNA hybrids and in vivo when it is artificially overexpressed can be recruited to chromatin in a hybrid dependent manner, stabilizing R-loops and converting them into replication obstacles (M. García-Rubio, Aguilera, et al., 2018). For this reason, we studied DNA-RNA hybrids levels upon ALY overexpression in U2OS. At a cellular level, we checked R-loops accumulation by IF with S9.6 antibody after in vitro treatment with RNase III and with or without RNase H in ALY-FLAG U2OS transfected cells. When ALY was overexpressed transiently in U2OS, we could see a general and 82 significant decrease in S9.6 signal that did not depend on RNase H (Figure R10A). In agreement with these cellular results, DRIP-qPCR analysis did not show R-loop accumulation at the standard gene APOE or in the nucleolar regions 18S and 28S in ALY overexpressing cells (Figure R10B). We reasoned that, contrary to what was expected, in U2OS cells the transient overexpression of ALY did not lead an increase in DNA-RNA hybrids. Figure R10. Analysis of DNA-RNA hybrids upon ALY overexpression. (A) Representative images and quantification of S9.6 immunofluorescence signal in U2OS transfected cells with FLAG or ALY-FLAG. All samples were treated with RNase III and with or without RNase H. The graph shows the median of S9.6 signal intensity in the nucleoplasm after nucleolar signal removal. At least 100 cells per condition were counted in each experiment (n=4). Red stars denote significant decrease. **P < 0.01; ****P < 0.0001 (Mann-Whitney U-test, two-tailed). (B) Schematic diagrams of APOE, 18S and 28S genes. Exons are depicted as closed boxes, the arrows indicate the start of transcription and red lines indicates the regions where the primer pairs used for amplification were located (above panel). DRIP-qPCR signal values in U2OS cells overexpressing ALY (pink). Samples were treated with RNase H in vitro prior immunoprecipitation where indicated. Signal values normalized with respect to the control (FLAG transfected cells) are plotted (n=3) as Mean ± SEM (paired t-test, two-tailed). Black stars denote significant increases, whereas red stars denote significant decreases. 2.3. Relevance of conserved ALY protein domains in genome instability ALY is an RNA binding protein that binds to the helicase UAP56 (Dufu et al., 2010). ALY protein has two UAP56 binding domains (UBM) and conserved RNA binding and RNA recognition domains (RBD and RRM) (Figure R11A). In vitro 83 experiments performed in the laboratory of Dr. Xiaoyu Xue with ALY purified proteins support that ALY binds not only to RNA, but also to different types of nucleic acids: ssDNA, dsDNA, DNA-RNA hybrids and R-loop structures (X. Xue personal communication). Moreover, they have evidence that ALY protein can stimulate the hybrid unwinding activity of the helicase UAP56. They show that mutant of the RBD domains (ALY 10E) is not able to bind R-loops, as expected. Interestingly the mutant lacking the N- and C-terminal UBM domains (ALY ΔNC), binds more efficiently R-loops, although it cannot interact with UAP56 and help to hybrid resolution (personal communication). Altogether these data lead us to study the effect of the different ALY protein domains on a possible genome instability. We took advantage of HeLa cells to check the effect of the mutant forms in genome instability and also to study whether ALY WT overexpression lead to a genome instability phenotype that could not be seen in U2OS cells. In collaboration with Dr. Xiaoyu Xue we analyzed whether this mutant ΔNC could exacerbate the genome instability phenotype of ALY overexpression. First, we cloned the WT and the two mutated forms in mammalian expression vectors under a CMV promoter and with the epitopes MYC and FLAG (Figure R11A). ALY expression was detected by immunofluorescence (Figure R11B) and western-blot (Figure R11C) being the levels of the mutant form 10E the lowest one. Immunofluorescence with anti-FLAG antibody was performed with and without pre-extraction what allowed us to realize that all the forms were located at the nucleus without pre-extraction, but the mutant 10E disappeared upon preextraction (Figure R11B). 84 Figure R11. Expression of ALY conserved domains and mutated forms. (A) Schematic representation of conserved domains of WT ALY protein and mutated forms called ΔNC and 10E. Aminoacidic point mutations R→E are represented as black stars. (B) Representative images of immunostaining with anti-FLAG antibody of HeLa cells after 24 hours transfection without preextraction (upper panel) or with prior pre-extraction (lower panel). (C) Western blot analysis of ALY-overexpressing cells with anti-FLAG (upper panel) and anti-ALY (lower panel) antibodies. Vinculin protein is used as a loading control. Once the expression was validated, we analyzed the genome instability phenotype caused by the expression of the WT and the mutant forms of ALY protein in HeLa cells. For this aim, DNA damage was measured by үH2AX foci per cell and to study its dependence on R-loops, in vivo RNase H1 overexpression was used. Immunofluorescence with antibodies against үH2AX and FLAG in HeLa cells allowed us to see a significant increase in DNA damage upon ALY-WT overexpression that was not seen when the mutant forms where expressed. Besides, the increase in DNA damage associated with ALY-WT overexpression was reduced when RNase H1 was overexpressed (Figure R12A). We concluded that the DNA damage observed was linked to DNA-RNA hybrids and for this reason we decided to check R-loop levels when HeLa cells were transfected with WT or mutant ALY forms. Cellular levels of R-loops were 85 measured by immunofluorescence of S9.6 antibody. All the conditions were treated with RNaseIII in vitro and with or without RNase H. The expression of the WT and the ΔNC ALY proteins caused a significant R-loop accumulation respect to the control (empty plasmid) that was higher in the case of the ALY-WT (Figure R12B). We concluded that in HeLa cells, unlike U2OS cells, ALY overexpression lead to genome instability, and that the expression of the mutant forms of ALY was not as harmful as the overexpression of the WT form, at least at the assayed conditions. Figure R12. Genome instability analysis upon WT or mutated ALY forms. (A) Representative images of immunostaining with antibodies against үH2AX (green) and FLAG (red) in HeLa cells after 24 hours transfection of ALY WT or mutated forms. Quantification of the number of үH2AX foci per cell is shown (n=3). (B) Representative images of HeLa cells immunostained with S9.6 (green), nucleolin (red) and FLAG (cyan) antibodies after transfection with ALY WT or mutated forms. The median of S9.6 signal intensity per nucleus after nucleolar signal removal in HeLa cells transfected with the indicated ALY form is shown (n=3). In both quantifications, the statistical significance of the difference was calculated with Mann-Whitney U-test two-tailed. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Black stars denote significant increases, whereas red stars denote significant decreases. 86 3. mRNA BINDING FACTOR ALY AT DNA damage sites 87 88 89 3.1. DNA-RNA hybrids and R-loop related factors at DSBs Cellular DNA is constantly damaged by numerous internal and external factors. Probably the most deleterious damage that the DNA could suffer is the doublestrand DNA breaks (DSBs) (Aleksandrov et al., 2020). Although R-loops can occur at different genomic locations, recently, their formation at damage sites has attracted much interest. There is some controversy as to whether or not these hybrids are formed as an inevitable risk of DNA breaks in active regions that need to be removed to allow the repair, or if they could have a potential role as repair intermediates (Andrés Aguilera & Gómez-González, 2017; Sonia Jimeno et al., 2019; Marnef & Legube, 2021). To go further into the role of these structures at DNA damage, we decided to study the enrichment of R-loop related factors at damaged sites. 3.1.1. R-loops accumulation upon DNA damage induction Proper detection of DNA-RNA hybrids was a crucial step to advance our knowledge about R-loops and factors related with them upon DNA damage. For this aim we followed three different approaches: In the first one, Dr. S. Silva studied R-loops accumulation by immunofluorescence with antibodies against үH2AX and S9.6 after ɣ irradiation (IR) in HeLa and U2OS cells. As expected, irradiation increased DNA-RNA hybrids at early timepoints and these results are shown in the appendix 1. In the second approach, we took advantage of the hybrid-binding domain (HBD) of the RNase H1 fused to GFP, a tool previously used in the laboratory, to analyze DNA-RNA hybrids (Bhatia et al., 2014). Specifically, U2OS cells were transiently transfected with a plasmid containing three copies of HBD tagged with GFP and expressed under the ubiquitin promoter for 24 hours. Then, the procedure followed was to induce DNA damage, performing the photoablation with a 355 nm confocal laser, and monitor HBD-GFP in vivo localization by cell imaging. Fluorescence analysis showed a clear and early recruitment of HBD domain at the laser stripe after microirradiation (Figure R13A). Nevertheless, since HBD-GFP contains only the binding domain of RNase H1, and not the catalytic one, once that recognizes the hybrids remains bound to these structures. 96 Next, we studied the recruitment of ALY and UAP56 at these five regions in DiVA cells. siALY and siUAP56 cells were used to study the specificity of the antibodies. ALY ChIPs showed a slight ALY enrichment in RBMXL1 and AGA regions upon DSB induction by OHT, whereas no increase was observed in regions without AsiSI cutting sites (APOE, SNRPN or the gene desert) (Figure R17A). UAP56 ChIPs did not show any preference of recruitment of this helicase to genes with DSBs compared to regions without DSBs (Figure R17B). Thus, although we have only observed a tendency, the results seem to indicate that ALY protein could be recruited to DNA damage sites, as shown with DiVA cells. 97 Figure R17. ChIP analysis of ALY and UAP56 to DSBs in DiVA cells. Chromatin immunoprecipitation of ALY (A) and UAP56 (B) in regions with DSB induction by OHT addition (RBMXL1 and AGA) (upper panel) and in regions without AsiSI cutting sites (APOE, SNRPN and a gene desert) (lower panel). siALY and siUAP56 are included as a control for immunoprecipitation antibody efficiency. The red dashed line marks the normalization value of 1. Relatives ChIP-qPCR signal values plotted as Mean ± SEM (at least n=3) (two-tailed paired ttest). 98 3.2. ALY recruitment to DSB sites DiVA system allowed us to see a differential enrichment to DSB sites of the RNA- binding factor ALY and the helicase UAP56. In order to analyze better whether these proteins were specifically recruited to a damaged site, we took advantage of laser micro-irradiation previously set up (section 3.1) for in vivo cell imaging of GFP-tagged proteins. 3.2.1. Early recruitment of ALY to DNA damage upon laser microirradiation For in vivo cell imaging, we transfected U2OS cells with vectors that carried ALYGFP and UAP56-GFP fusions, and GFP expressed under the ubiquitin promoter. 24 hours after transfection, a 355 nm photoablation confocal laser was applied to perform the DNA damage and in vivo cell images were taken along the time. The protein recruitment was quantified as an intensity fold change (for more details see Materials and Methods). Whereas GFP protein, included as a negative control, did not change its location at damage site along time, UAP56-GFP seemed to be a little excluded just after the ablation and was not recruited at all, while we could observe a rapid and clear recruitment of ALY-GFP (Figure R18). ALY was recruited to the damaged site at few seconds upon DNA damage induction reaching a maximum level of recruitment at 2 minutes, then it started to disappear reaching basal levels at around 7 minutes (Figure R18). Figure R18. Analysis of ALY and UAP56 at damage induced by photoablation laser. U2OS cells transfected with expression vector for ALY-GFP, UAP56-GFP and GFP proteins were microirradiated and live-imaged after 10 minutes. Representative images are presented. Yellow arrows indicate where the laser ablation was performed. Scale bars, 25µm. Quantification of the GFP, ALY-GFP and UAP56-GFP intensity fold changes. The intensity of the pre-laser point has been normalized to 1. Data are plotted as mean ± SEM (n=2, 10 cells/experiment) (Two-way Anova). *** P < 0.001 99 3.2.2. Factors and mechanisms controlling ALY recruitment to DSBs Since ALY is an RNA binding factor that works at the interface between RNA transcription and RNA export (Hautbergue et al., 2008; Stubbs & Conrad, 2015; Stutz et al., 2000) and we have observed R-loop accumulation and DNA-RNA hybrid genome instability-associated in siALY cells (section 1.2). We decided to study whether the recruitment of ALY was related with its role in transcription and/or R-loop metabolism. First, we wonder whether ALY was being recruited at DSBs in a transcription dependent manner. For this aim we used two different transcription inhibitors: cordycepin (3’ deoxyadenosine), a general polyadenylation chain elongation inhibitor (Chu & Edelman, 1972; Pérez-Calero et al., 2020) and DRB (5,6-dichloro-1-β-D-ribofuranosylbenzimidazole), an inhibitor of RNAPII transcription elongation (H. Chen et al., 2020) at concentrations previously used and confirmed to inhibit transcription. We saw that ALY-GFP was still recruited at DSBs upon transcription inhibitor treatments but at lower levels. The lower levels of recruitment of ALY-GFP at DRB and its vehicle DMSO condition can be related with the DMSO itself and the longer treatment (24-hour treatment) (Figure R19A- B). In the first chapters of results from this thesis we have demonstrated that ALY can play a role in R-loop-mediated genome instability. Considering this and that R-loops have been shown to be accumulated upon DNA damage (Britton et al., 2014)(and in section 3.1) we wonder if the recruitment of ALY-GFP at induced DSBs was related to DNA-RNA hybrids. To address this question, we used the U2OS RNH1 inducible stable cell line created in Andrés Aguilera’s lab by José Calderón. This cell line overexpressed RNH1 under doxycycline (doxy) addition. Therefore, at this condition the levels of R-loops would be lower because RNH1 degrades the RNA moiety of the DNA-RNA hybrid. We reasoned that if ALY was binding R-loops at DNA damage sites and R-loops were being removed by RNH1 overexpression, no recruitment of ALY protein would be expected. However, ALY-GFP was still recruited at DSBs upon RNH1 overexpression but at lower levels, so the recruitment of ALY could be only partially dependent on R-loops (Figure R19C). 100 ALY is considered an important mRNA export factor conserved from yeast to humans that interacts with the RNA receptor TAP/Mex67 (Hautbergue et al., 2008; Stutz et al., 2000). Besides, ALY is the main and specific reader of m5C, a post-transcriptional RNA modification identified in tRNAs, rRNAs and in mRNAs, which are mainly catalyzed by the RNA methyltransferase NSUN2 and that has a role in mRNA export and post-transcriptional regulation (X. Yang et al., 2017). In addition, it has been recently described that m5C modification serves as a DNA damage code to promote homologous recombination, but in this case, the RNA methyltransferase that catalyze it would be TRMDT1 and the main reader would be RAD52 protein (H. Chen et al., 2020). Since, ALY is a main reader of m5C mark in the context of RNA export, we wondered whether could be a reader of this mark also in the context of DNA damage. We carried out micro-irradiation experiments in cells depleted of the methyltransferase TRMDT1 by siRNA depletion for 72 hours and expression of ALY-GFP protein for the last 24 hours (Figure R19D) Cell imaging at different time points reveals that ALY recruitment after DSB induction was not altered by TRMDT1 depletion. So, we concluded that early ALY recruitment does not depend on m5C damage code (Figure R19D). 101 Figure R19. Analysis of ALY-GFP recruitment at DBS. (A) Quantification of ALY-GFP recruitment with or without Cordycepin treatment. Cells were treated for 4 hours with 50 µM cordycepin or its mock control (H2O) (n=2 with 10 cells/experiment). (B) Quantification of ALYGFP recruitment with and without DRB treatment. Cells were treated for 24 hours with 20 µM DRB or DMSO respectively (n=2 with 10 cells/experiment). (C) Quantification of ALY-GFP recruitment in RNH1-inducible stable U2OS stable cell line. Overexpression of RNH1 (+RNH1) is achieved by doxycycline treatment (n=2 with 10cells/experiment) (D) Quantification of ALY-GFP recruitment in TRMDT1-depleted U2OS cells. (n=3 with 10 cells/experiment). (A-D) Data are plotted as intensity fold change mean ± SEM where the pre-laser point has been normalized to 1 (Two-way Anova). Red stars denote significant increase. * P < 0.05. We took advantage of the mutant ALY 10E that does not bind to RNA (described in section 2.3. of this thesis) to analyze if the recruitment of ALY at DNA damage sites depends on its ability to bind RNA. For this aim, we transfected U2OS cells with a plasmid carrying ALY-WT or ALY-10E and 24 hours later, DNA damage was induced by photoablation and cells collected and fixed to be analyzed by immunofluorescence against ALY and үH2AX. We could detect ALT-WT and ALY-10E at laser stripe (labeled by үH2AX IF) (Figure R20). ALY was recruited to DSBs despite of the mutations in its RBDs, suggesting that these mutations are not enough to block the ability of the protein to be recruited at laser stripe upon DNA damage induction. 102 Figure R20. ALY mutant 10E is still recruited at damage sites. Immunofluorescence panel with antibodies against ALY (red) and үH2AX (green) and nuclei stained with DAPI. Both, WT and mutant form 10E of ALY are recruited to laser stripe. Yellow arrows indicate where was the laser ablation performed. Scale bars, 25µm. The recruitment of ALY to DSBs was not found to be closely related to its role in transcription, RNA export and R-loop metabolism, so we envisaged the possibility that ALY was recognizing something specific for DNA damage. Parylation is one of the earliest steps in signaling DSBs and is involved in different repair pathways (Ray Chaudhuri & Nussenzweig, 2017; Smith et al., 2019). It consists in the generation of Poly(ADP)ribose (also known as PAR) chains that signal DNA damage and also serve as an scaffold for different repair factors (F. Zhang et al., 2015). PARP1 (PAR polymerase 1) is the main enzyme that generates these PAR chains (Ray Chaudhuri & Nussenzweig, 2017) and PARG (PAR glycohydrolase) is the enzyme primarily responsible for PAR degradation (Figure R21A) once the damage is repaired (Mashimo et al., 2013). Interestingly, it has been described that ALY is parylated upon different genotoxic agents (Jungmichel et al., 2013) and highly ADP-ribosylated in all breast cancer cells (Zhen et al., 2017). Thus, we decided to use inhibitors against the enzymes involved in the parylation process after ALY-GFP transfection to see whether ALY recruitment was mediated by the formation of PAR chains. ALY was not recruited at all at DSBs when cells were treated with the PARP1 inhibitor Olaparib (Figure R21B). In contrast, ALY was even more recruited at DSBs (Figure R21C) and maintained more time (Figure R21D) in cells treated with a PARG inhibitor (PARGi). Data support that ALY recruitment is completely dependent on parylation, nevertheless it is not clear whether ALY is recognizing PAR chains and/or if ALY needs to be parylated to be recruited. Further experiments would be needed to explore these two possible scenarios. 103 Figure R21. ALY recruitment at DSBs is completely dependent on parylation. (A) Scheme of the formation of Poly (ADP-ribose) chains and the main enzymes involved in the process. (B) In vivo imaging panel showing the absence of recruitment of ALY-GFP upon Olaparib treatment compared to DMSO. The plot shows the quantification of ALY-GFP at the stripe upon Olaparib treatment 10 µM for 1 hour (C) In vivo imaging panel showing the intense and longer recruitment of ALY-GFP upon PARGi treatment. Quantification of ALY-GFP at the stripe upon PARGi treatment 0.3µM for 10 minutes. (D) Cell imaging analysis was extended up to 30 minutes. Quantification of ALY-GFP at the stripe upon PARGi treatment is plotted. Intensity fold change where the pre-laser point has been normalized to 1 is represented. Yellow arrows indicate where the laser ablation was performed. Scale bars, 25µm. (B-D) Data are plotted as mean ± SEM (n=2 with 10 cells/experiment). Black stars denote significant increase whereas red stars denote significant decrease. **P < 0.01; ***P < 0.001 (Two-way Anova). 104 3.3. ALY and DNA repair Our results indicate that mRNA factor ALY is recruited at DNA damage in a rapid and completely PAR-dependent manner, whereas RNA and R-loop seem not to be so relevant for ALY recruitment. However, the reason why ALY is being recruited is still unclear, so we studied DNA damage response (DDR) and DNA repair in ALY-GFP expressing cells 3.3.1. Analysis of the effect of DNA damage response on ALY recruitment In mammalian cells, DSBs triggers a DDR to overcome their toxicity and allow DNA repair maintaining genome stability (Britton et al., 2014). Another component of DDR is local transcription arrest triggered by DNA breaks. An expanding aspect of the DDR is its connection with RNA metabolism. Indeed, the DNA damage activates kinases ATM or ATR that phosphorylate numerous proteins involved in RNA metabolism. A link with the DDR has been established for several members of the heterogenous ribonucleoprotein (hnRNP) family, RNA-binding proteins or pre-RNA processing factors that are major mediators of genome stability and some of them prevent interactions between the nascent RNA and template DNA (R-loops). Indeed, some RNA binding proteins have been shown to be rapidly and transiently recruited upon DNA damage (Britton et al., 2014; Krietsch et al., 2012). Some of these proteins are later excluded from the damage site by a mechanism that depends on the activation of DDR (Britton et al., 2014). Interestingly, we have observed that ALY was first recruited and then excluded from the laser stripe in some cells after laser irradiation (Figure R22A). So, we decided to explore whether ALY could be excluded from the DNA damage site in a DDR-dependent manner. For this aim, in vivo cell imaging analysis were performed in U2OS cells transfected with ALY-GFP plasmid and treated one hour prior irradiation with inhibitors of kinases involved in DDR (ATMi, ATRi or DNA-PKi). We expected that if ALY exclusion depends on DDR when the signal transduction is inhibited, ALY recruitment signal should increase or stay for longer times. Any of the single inhibition treatments led to a significant change on ALY recruitment kinetics (Figure R22B) although it is worth to notice that upon ATMi or ATRi treatment, the recruitment of ALY appeared to be higher than 105 control cells at early times while with DNA-PKi appeared to be lower. The combination of the three inhibitors caused a higher decrease on ALY recruitment that could be linked to a problem with DDR because the inactivation of these three kinases. We concluded that the observed phenomena of ALY-GFP exclusion after its recruitment did not rely on ATM, ATR and/or DNA-PK. Figure R22. ALY exclusion at DSBs does not depend on ATM, ATR or DNA-PK. (A) In vivo imaging panel showing a pre-laser ALY-GFP positive cell (left panel), the recruitment of ALY-GFP to laser stripe after microirradiation (middle panel) and ALY-GFP exclusion (right panel). Yellow arrows indicate where the laser ablation was performed. Scale bars, 25µm (B) Quantification of ALY-GFP at the stripe upon single treatments with ATMi, ATRi or DNA-PKi or upon treatment with the three inhibitors. Intensity fold change at the pre-laser time point has been normalized to 1 and data are plotted as mean ± SEM (n=2 with 10 cells/experiment) (Two-way Anova). Our data indicate that ALY is recruited in a rapid and PAR-dependent manner at DSBs but the possible relevance of ALY presence is still unclear. To explore this, we wondered whether ALY interacts physically or functionally with factors involved in different repair pathways. For this purpose, we studied ALYGFP recruitment by in vivo cell imaging in cells depleted of early and key factors of the two main repair pathways. Specifically, we depleted the subunit KU80 of the KU complex and the MRE11 subunit of the MRN complex, crucial early complexes for NHEJ and HR respectively. RAD51, although is not an early factor, was also included because it is an essential protein for HR but not for NHEJ. U2OS cells were depleted of these factors by siRNAs for 72 hours and transfected with expression ALY-GFP plasmid in the last 24 hours prior photoablation and in vivo cell imaging. Following this procedure, we could see 112 Figure R26. Recruitment of endogenous ALY. (A) Immunofluorescence images with antibodies against ALY and үH2AX and DAPI nuclear staining after photoablation. Yellow arrows indicate where the laser photoablation was done. (B) Quantification of the cells with ALY stripe from the cells with үH2AX stripe from the total population. Data are represented as Mean ± SEM n=2. It is possible that ALY is recruited at DNA damage at basal levels contributing in some way to DNA repair, in contrast, our hypothesis was that high levels of ALY protein could be an obstacle for the proper repair. We analyzed whether ALY overexpression caused an impair in DNA damage repair. For this aim, we checked the intensity of үH2AX at photoablation laser stripe along time by immunofluorescence against үH2AX and ALY. We found that ALY- overexpressed cells showed a significant higher үH2AX signal intensity along the time than the control cells expressing GFP (Figure R27). These results support that overexpression of ALY can be interfering with the proper DNA repair. Whether this has to do with the overexpression of ALY in cancer cells is a matter for future research. Figure R27. үH2AX repair upon ALY overexpression. DNA damage repair kinetics as determined by үH2AX IF (signal intensity) at laser stripe after laser photoablation and ALY overexpression. Cells were fixed and IF carried-out at the indicated time points. Data represent 113 the percentage of үH2AX intensity normalized to the first point after damage (15 minutes). Data are shown as Mean ± SEM (n=2, 10 cells per experiment). *P < 0.05 (Two-way Anova). 114 4. Epigenetic compounds and R-loops 115 116 R-loops, as a source of genome-instability can be used as a marker of this phenotype, a hallmark of cancer cells. The link between R-loops and disease raises the possibility that manipulation R-loop levels could be could be used as drug target and exploited therapeutically. In this, we have performed a screening with epigenetic compounds, given the connection between R-loops and chromatin, and looked for drugs that increase R-loop levels. We reasoned that this search could help to unveil new genes related to R-loop metabolism, in the case where the target of the compound is well defined, but also could serve as a tool to open up new opportunities and drug combinations with clinical relevance, as some of these compounds are used in therapy. 4.1. Screening of epigenetic compounds We have performed a screening of epigenetic compound library (Selleckchem L1900) in U2OS cells by S9.6 immunofluorescence to look for compounds that increase R-loop levels (Figure R28A). This library contains 128 small molecules modulators (inhibitors and activators) structurally and mechanistically different. It includes inhibitors of epigenetic enzymes such as Histone deacetylases (HDACs), Sirtuins (SIRTs), Lysine demethylases, histone acetyltransferase (HATs), DNA methyltransferase (Dnmts) and sirtuin activators. The most representative categories of the main targets defined in the library for these compounds have been schematized as a part of a whole (Figure R28B). In addition, 14 compounds from an anticancer library (Selleckchem L300) were included because their ability to inhibit targets related with chromatin structure and genome instability, thus a total 142 compounds were screened. This high-throughput microscopy screening was performed in 96-well plates taking advantage of the facilities of Genomic Unit at CABIMER. U2OS cells were treated for 3 hours with each drug by duplicated and at two different concentrations (1µM and 10µM) increasing the possibilities to see a positive drug effect. After the treatment, we carried out immunofluorescence with S9.6, an antibody that recognizes DNA-RNA hybrids and with an antibody against nucleolin to exclude nucleolar signal and specifically measure the nucleoplasm 117 signal, following standard protocols. S9.6 antibody signal detection was acquired by automated microscopy of cells with IMAGE-Express (137239) (Materials and Methods). The histone deacetylase inhibitor TSA and siUAP56 cells (Pérez- Calero et al., 2020; Salas‐Armenteros et al., 2017) were included as positive controls and DMSO treated cells as mock control (see Appendix 4 and Materials and methods for more details). The protocol of screening was performed twice, and compounds whose treatment lead to an increase in S9.6 signal versus non-treated conditions (DMSO) in both duplicates, being at least in one of the duplicates (≥1.1), were selected as candidates. With this criteria, 36 out of 142 compounds tested were selected. These candidates were analyzed in a third round and 11 hits were obtained, taking into account the increase in S9.6 signal and the reproducibility between duplicates of the different experiments (for more details see Materials and methods section) (Figure R28C). The list of the hits, their target and other useful data are shown (Figure R28D). Interestingly, among them different HDAC inhibitors were found, in agreement with the relationship between histone deacetylation and R-loop formation (Salas‐Armenteros et al., 2017); in addition two compounds from the anticancer library, known to cause DNA damage and genome instability, came up as hits altogether supporting our screening (Figure R28D). We decided to focus on CX-6258, a PIM kinase inhibitor and on the histone demethylase inhibitor tranylcypromine, two of the top hits. We also included a JAK kinase inhibitor, WP1066, because although its reproducibility was lower (2/3 rounds), it also showed a high S9.6 signal. 118 Figure R28. Screening of epigenetic compounds that leads to R-loop accumulation. (A) Schematic workflow for the screening of epigenetic compound library. (B) Main targets represented in the epigenetic library. (C) Scheme of hits obtained in the different steps of the screening with the 128 epigenetic compounds and 14 anticancer compounds. (D) Useful information about the hits from the screening: name of the product, main target described in the library, reproducibility as number of rounds and concentration at which the compound has been considered as a hit. In grey there are the 2 hits from the 14 anti-cancer library compounds included in the screening. Highlighted in bold are the compounds selected for further analysis. 4.2. Genome instability upon compound treatments Once the candidates were selected, we moved to validate them and characterize the effect of the treatment with these epigenetic compounds on genome instability mediated by DNA-RNA hybrids. 4.2.1. R-loop accumulation upon compound treatments The three selected compounds were titrated to validate R-loop accumulation and to determine the optimal compound concentration. For this aim, U2OS cells were treated for 3 hours with the candidates WP1066, CX-6258 or Tranylcypromine in 6-well plates at the concentrations used in the screening 10µM and 1µM but also with the intermediate concentration of 5µM and lower concentrations of 100, 50 and 10nM. In addition, TSA at 250nM was added as a positive control (Salas‐ 119 Armenteros et al., 2017). After the treatment, immunofluorescence with S9.6 and nucleolin antibodies was performed. It was found that WP1066 at 10µM caused a huge increased in R-loop levels that was so high that we preferred to proceed with lower concentrations, 5µM and 100nM. In the case of CX-6258 and Tranylcypromine 50nM was found to be the best concentration (Figure R29A). Once the concentrations were properly titrated for R-loop accumulation, we decided to verify that the increased levels of S9.6 signal were due to DNARNA hybrids. For this aim, in vitro treatments with RNase III for all the samples and with or without RNAse H prior S9.6 IF were performed. The R-loop accumulation was validated for WP1066 and CX-6258, but not for Tranylcypromine, where the in vitro treatment of RNase H did not cause a decrease of the S9.6 signal (Figure R29B). In order to validate the R-loop accumulation at a molecular level, we performed DRIP in the standard genes APOE and RPL13 for R-loop detection (Ginno et al., 2013; Herrera-Moyano et al., 2014) in U2OS cells treated for 3 hours with the three epigenetic compounds. DRIP analysis did not show R-loop accumulation in cells treated with WP1066 and CX6258, at least at the regions assayed, whereas Tranylcypromine whose cellular R-loop accumulation was not reduced with RNH in vitro treatment showed a significant increase in both genes by DRIP-qPCR (Figure R29C). Thus, the three compounds were validated at least by one technical approach. 120 Figure R29. R-loop accumulation upon epigenetic compounds treatment. (A) Titration of epigenetic compound for R-loop increase in U2OS cells by immunofluorescence with S9.6 and nucleolin antibodies, and DAPI staining. DMSO and TSA were used as negative and positive controls, respectively. At least 300 cells were quantified (n=1). (B) Validation of cellular R-loop accumulation by S9.6 IF. The median of S9.6 signal intensity per nucleus after nucleolar signal removal in U2OS cells treated in vivo with the corresponding compounds. Samples were treated in vitro with RNase III and with or without RNase H. More than 150 cells were scored in each experiment) (n=3). Black stars denote significant increase whereas red stars denote significant 121 decrease. *P < 0.05; ***P < 0.001; ****P < 0.0001 (Mann-Whitney U-test, two-tailed). (C) Schematic diagrams of APOE, RPL13A and MIB2 genes. Exons are depicted as closed boxes, the arrows indicate the start of transcription and red lines indicate the regions where the primer pairs used for amplification were located. DRIP-qPCR signal values in U2OS treated with DMSO, WP1066, CX-6258, Tranylcypromine at the indicated regions are shown in the lower panel. Samples were treated in vitro with RNase H prior immunoprecipitation where indicated. Signal values are plotted as Mean ± SEM (n=3). *P < 0.05 (one-tailed paired t-test). Other details as in B. 4.2.2. Analysis of DNA damage induced by compound treatments Some compounds of the epigenetic library are at different stages of clinical trials and are used for anticancer treatments or for other diseases (WP1066 is at clinical phase I for metastatic or recurrent tumors and Tranylcypromine is at clinical phase 4 for Bipolar disorder). We were interested in knowing whether our candidates induced DNA damage and with this purpose we performed үH2AX IF in 3h-treated U2OS cells. WP1066 and CX-6258 caused a significant increase in үH2AX foci, whereas Tranylcypromine treatment did not cause an increase in DNA damage. Neocarzinostatin was used as positive control (Figure R30A-B). We wonder whether the DNA damage caused by the different compound treatments was mediated by DNA-RNA hybrids. For this aim, RNH1 was expressed prior the drug treatment. We found that the DNA damage caused by WP1066 was significantly rescued by RNH1 overexpression, whereas үH2AX foci associated to CX-6258 were not reduced. As expected, Tranylcypromine treatment, also included in these experiments, did not cause an increase in үH2AX foci and no reduction was observed upon RNH1 overexpression (Figure R30B). Therefore, we reasoned that only WP1066 caused DNA damage R-loop- dependent. 128 whether TSA caused also a synergistic effect in combination with our candidates WP1066 and CX-6258 on genome instability. For this aim, we carried-out combined treatments in U2OS cells and checked R-loop accumulation and DNA damage by S9.6 and үH2AX IF, respectively. The double treatments did not lead to an increase in R-loop levels, indeed, a decrease was seen for both combinations (Figure R34A). Regarding DNA damage phenotype, any significant synergistic effect was seen upon double treatments, probably due to the high levels of үH2AX foci caused by single treatments with WP1066 or CX-6258, whose effect appears to be epistatic to TSA treatment (Figure R34B). Figure R34. R-loop and DNA damage upon double treatments. (A) Quantification of S9.6 immunofluorescence in U2OS cells treated with DMSO, WP1066 or CX-6258 and with or without TSA. The black line represents the median of S9.6 signal intensity of each treatment and the red ones shows the median of S9.6 signal in the control with DMSO. (B) Quantification of the percentage of cells with more than 5 үH2AX foci in U2OS cells treated with DMSO, WP1066 or CX-6258 and with or without TSA as indicated. Data from one experiment is shown. 4.3.3. Effect of compounds on survival of patient cells deficient on R-loop-related factors Finally, we decided to test whether there was a synergistic effect on survival in patient cells with deficiencies in genes known to be related with R-loop accumulation. We tested the compounds in the cell lines PD20 and DLD-1 which are deficient for FANCD2 or BRCA2, and in their respective corrected cell lines. FANCD2 and BRCA2 prevent R-loop accumulation (Bhatia et al., 2014; M. L. García-Rubio et al., 2015) and BRCA2 also promotes the resolution of DNA-RNA hybrids at DNA damage by DDX5 helicase (Sessa et al., 2021). Both deficient cell lines (PD20-/- and BRCA-/-) and the corrrected ones (PD20+/+ and BRCA2+/+) were treated with WP1066 and CX-6258 in single and combined treatments. WP1066 did not cause a synergistic effect on cell survival in any of the cellular lines checked, whereas the CX-6258 inhibitor caused a significant reduction of 129 cell survival of FANCD2 deficient cell lines (PD20-/-) (Figure R35A). The combination treatment of both inhibitors in these cell lines leaded to a synergistic effect in colorectal adenocarcinoma cell line DLD-1 although it did not depend on BRCA2 defficiency because the synergy is seen in DLD-/- and DLD+/+ being significant in DLD+/+.//The combination treatment of both inhibitors (WP1066+CX- 6258) lead to a synergistic effect in colorectal adenocarcinoma cell lines DLD-/- and DLD+/+, probably not depend on BRAC2, but that could be relevant for the treatments of these adenocarcinomes (Figure R35B). Figure R35. Survival assays in patient cell lines deficient on R-loop factors upon treatment with epigenetic compounds. (A) Analysis of survival assays upon single treatment (WP1066 or CX-6258) or double treatment with both compounds of PD20-/- and DLD-/- and their respective corrected cell lines sorted by the treatment. (B) Analysis of the same survival assays but sorted by patient cell line. Red stars denote significant decrease. Data is shown as Mean ± SEM (n=3). *P < 0.05; ***P < 0.001 (two-way Anova). 130 DISCUSSION 131 132 In this thesis, we undertook two type of approaches to understand the origin of genome instability caused by R-loops by the identification and analysis of specific factors. First we studied the role of specific RNA-binding factors in R-loop control; second, we did a search for drugs that targeted cellular factors preventing R- loops. Firstly, we have unveiled the importance of the levels of the RNA-binding protein ALY in R-loop accumulation and genome integrity. We found that ALY is recruited to DNA breaks. This occur in a specific and rapid manner. Interestingly, even though ALY functions are linked to RNA metabolism, its recruitment to DNA breaks was only partially dependent on RNA or R-loops but completely dependent on parylation. We provide evidence that ALY levels could be relevant for DNA damage response (DDR), as suggested by an impairment in damage signaling and DNA repair upon depletion or overexpression of ALY. Secondly, we identified new drugs that cause R-loop accumulation by screening a collection of drugs known to act on enzymes involved in epigenetic processes, including HDACs, SIRTs, lysine demethylases, HATs or DNA methyltransferases. After screening, titration and validation steps, 3 compounds were identified from which we decided to focus on two, the kinase inhibitors WP1066 and CX-6258 whose use in cells also caused DNA damage. Interestingly, depletion of a protein target of WP1066, the transcription factor STAT3, led to an increase in R-loop levels, opening the possibility yet to be proved that this drug could increase R-loops by altering STAT3 activity. We provide evidence of the potential clinical relevance of WP1066 and CX-6258 in single or combination treatments as supported by their differential impact on cell viability in adenocarcinomas or FANCD2-deficient cells. 1. Role of mRNA factor ALY in genome instability Cumulative evidence supports that mRNP biogenesis is not only important for the stability and function of the mRNP but also for the maintenance of genome integrity. The current model is that a defective mRNP assembly could favor the hybridization of the nascent mRNA with the DNA template leading to the formation of R-loops that can impair transcription and replication processes and can finally cause DNA damage and genome instability (García-Muse & Aguilera, 2019). The first examples of this connection were the THO complex which is 133 involved in transcription elongation (Chávez et al., 2000) and the splicing factor/RNA processing factor SRSF1, formerly called ASF/SF2 (X. Li & Manley, 2005). In the last years, an emerging number of RNA-binding proteins such as splicing and 3’ mRNA processing proteins, among others, have been associated with the prevention of genome instability (Das & Krainer, 2014; Olazabal-Herrero et al., 2020; Stirling et al., 2012). Nevertheless, it is still unclear whether the phenotype associated with the deficiency of all the reported factors depends on the formation of DNA-RNA hybrids. In this thesis, we have focused on ALY, a factor that works at the interface between transcription and export and seems to be also relevant for genome integrity. Its absence leads to an increase in DNA damage and defects in RNA export in human cells (Domínguez-Sánchez, Barroso, et al., 2011) whereas its overexpression caused genome instability in yeast (Gavaldá et al., 2016). Moreover, ALY levels are altered in a wide variety of tumors (Domínguez-Sánchez, Sáez, et al., 2011; J. Wang et al., 2021). Taking together these data and the interaction of ALY with the THO complex and the RNA helicase UAP56 during transcription (K. Sträßer et al., 2002), here we studied the effect of different ALY protein levels in cellular homeostasis and genome integrity, as a way to define its role in genome integrity. We show that standard cell lines with reduced or increased levels of ALY caused an R-loop mediated genome instability phenotype (Figure R2-R12). On one hand lower levels of ALY would not bind enough the nascent RNA properly increasing the chance of rehybridization with the DNA template. This situation would increase R-loop formation, a three-stranded structure where the displaced ssDNA is more susceptible of being damaged. Supporting this model, we have observed an increase in DNA damage in siALY cells by comet assay that is suppressed by RNase H1 (Figure R2). On the other hand, an excess of ALY lead to an increase in R-loop and DNA damage (Figure R12). In yeast, it has been shown that Yra1 overexpression confers R-loop-dependent genome instability and replication impairment (Gavaldá et al., 2016). It has been proposed that this is caused by the ability of Yra1 to bind and stabilize R-loops structures potentiating their ability to block replication forks in cells when Yra1 is in excess (M. García-Rubio, Aguilera, et al., 2018). An important question is whether ALY overexpression in human cells could cause R-loop-dependent genome instability 134 in a similar manner as that reported by Yra1 overexpression. Interestingly, as is the case of its ortholog yeast Yra1, ALY can also bind DNA-RNA hybrids in vitro (X. Xue, personal communication). Thus, since ALY is recruited to the transcribed chromatin (J. Wang et al., 2021) when it is overexpressed it could bind and probably stabilize the hybrids. In agreement with this idea, overexpression of ALY in cells induced both R-loop accumulation and DNA damage (Figure R12). ALY interacts with the helicase UAP56 that unwinds harmful R-loops genome-wide (Pérez-Calero et al., 2020) and in vitro data indicate that ALY can stimulate UAP56 activity (Pühringer et al., 2020; Taniguchi & Ohno, 2008). So, we hypothesize that an ALY mutant lacking UAP56 binding motifs should show higher levels of genome instability and R-loops because ALY would be bound but the hybrid structure could not be removed. Importantly, in vitro data indicate that a mutant ALY ΔNC, lacking the referred domain, is more efficiently bound to R- loops but it does not stimulate its resolution (X. Xue, personal communication). Despite our hypothesis, ALY ΔNC mutant protein transiently expressed in cells did not have a higher impact than ALY WT on R-loops or DNA damage (Figure R12). So, it is possible that either ALY had additional functions independent of UAP56 or the effect of the mutant protein was not detected in our experiments because the endogenous ALY protein was still expressed in cells. Further experiments are needed to clarify this issue. Overexpression of ALY interferes with cell cycle progression showing an intra-S phase arrest with significantly lower levels of EdU incorporation (Figure R8B-C). This arrest could explain the lower levels of H3S10P, a mitotic condensation mark, because the cells are not entering in mitosis (Figure R8A). Data suggest that ALY levels are relevant for replication progression, although T- R conflicts as determined by FANCD2 foci were not detected at least at the condition assayed. Indeed, we observed lower levels of FANDC2 foci in ALY- overexpressing cells (Figure R9C) consistent with the conclusion that replication is diminished as deduced from the reduced levels of EdU incorporation and intra- S phase arrest (Figure R8B-C). Nevertheless, more specific replication assays such as DNA combing are needed to better study the effect on DNA replication. 135 Interestingly, ALY overexpression caused an alteration in the chromatin appearance as seen by DAPI nuclear staining that was more pronounced in cells expressing the highest levels of ALY (Figure R7B). This appearance was reminiscent of a pattern of condensation. However, H3S10pho as a marker of chromatin condensation and correlated with harmful R-loops and genome instability (Castellano-Pozo et al., 2013; García-Pichardo et al., 2017) was not associated with the appearance of cells with ALY overexpression (Figure R8A). Other histone modifications could be tested, such as H3K9me2, related with dsRNA, antisense transcription and R-loops leading to the recruitment of heterochromatin protein 1γᵞᵧℽ (HP1ɣ) (Skourti-Stathaki et al., 2014), H2BS14pho protein, which is linked to DSB response (Fernandez-Capetillo et al., 2004), somatic hypermutation and immunoglobulin class-switching recombination (Odegard et al., 2005) and with apoptosis by facilitating chromatin condensation (Cheung et al., 2003) or H2Bub1, related to chromatin compaction and oncogenesis (Jeusset et al., 2021). Although further analysis is needed, it is tempting to speculate that changes in chromatin structure promoted by R-loops could be behind the DAPI appearance of ALY overexpressing cells. It has been reported that transcription and DNA damage repair act in a coordinated manner. On one hand, active transcription promotes DSB repair due to a more open chromatin and faster DNA repair factors recruitment, evidence is shown that nascent RNA can be used as a template for HR repair and many DSBs are help in their repair by the presence of transcription. On the other hand, it is suggested that transcription repression could facilitate DNA repair by removing transcription factors that could interfere with DNA repair. Thus, transcription is repressed upon DNA damage through multiple mechanisms, for instance ATM-dependent ubiquitination of histone H2AK119 prevents RNAPII elongation at DSBs (Suzuki et al., 2019) or negative elongation factor complex (NELF) is recruited to DSBs nearby active genes in a PARP1-dependent manner to repress transcription and promote DSB repair (Awwad et al., 2017). Recently, DNA-RNA hybrids have also been found at DSBs, although their functional role at these sites is controversial (See Introduction 2.2.3). Recent data show that some RNA helicases locate at DSBs and contribute to DSB repair as 136 is the case of human SETX and its budding yeast ortholog Sen1, limiting R-loop accumulation and preventing genome instability (Cohen et al., 2018; Rawal et al., 2020) or DDX5 that acts in coordination with BRCA2 to repair DSBs (Sessa et al., 2021). It is possible that other helicases could also resolve R-loops at DNA damage sites as it is the case of DDX19, which is translocated from the cytoplasm to the nucleus upon DNA damage (Hodroj et al., 2017) or DHX9, which interacts with PARP1 preventing R-loop-associated DNA damage (Cristini et al., 2018). Nevertheless, further studies are needed to clarify a possible role of these helicases at DSBs. Following two different approaches, we found that ALY is recruited to DNA damage, whereas UAP56 is not (Figure R17-18). These results suggest that although UAP56 is a major co-transcriptional helicase that unwinds R-loops genome-wide (Pérez-Calero et al., 2020) may not be necessary to resolve hybrids at DSBs in contrast to other helicases (Cohen et al., 2018; Cristini et al., 2018; Hodroj et al., 2017; Sessa et al., 2021). Interestingly, UAP56 has been related with DSBs and proposed to ensure ssDNA formation and RAD51 accumulation at DSBs by maintaining BRCA1 levels (Z. Xu et al., 2020). Since we have not been able to detected UAP56 at DNA damage, we would need to see whether the reported effect of this helicase on DSB repair could be indirect and mediated by its role in gene expression. In the case of ALY we observed a rapid recruitment at DNA damage by laser microirradiation (Figure R18), whereas the ChIP enrichment of ALY at DSBs generated by I-sceI restriction enzyme was not significant (Figure R17A). These data can be explained if ALY did not persist at DSBs for long periods of time as it happens with other transient players in DSB repair as PARP1 (Mastrocola et al., 2013). Our data suggest that ALY has independent functions of UAP56, as deduced from experiments with ALY ∆NC mutant. In this scenario, ALY can be still recruited at DNA damage because it is an RNA-binding protein and DNARNA hybrids are shown to be present at DSB (Figure R13). Nevertheless, recruitment of ALY was shown to be only partially dependent on RNA and hybrids; when transcription was repressed or R-loops removed prior DNA damage induction, ALY was still recruited (Figure R19A-C). Consistently, ALY 137 mutant 10E, which contains point mutations in both RBDs and a reduced ability to bind RNA in vitro (X. Xue, personal communication), can be recruited at DSBs (Figure R20). Certain RNA modifications such as m5C are linked to RNA metabolism and recently has also been associated with DNA damage (H. Chen et al., 2020). ALY is its main reader in the context of RNA export but RAD52 has been described to be its reader in the context of DNA damage (H. Chen et al., 2020). In the study performed by Chen and colleges (2020) ALY was discarded as a m5C reader at DNA damage. Nevertheless, the experiment was performed 1 hour after light exposition. Our data indicate an early recruitment of ALY, but at early time points ALY was still recruited at DSBs even in the absence of the methyltransferase TRMDT1 in charge of m5C mark (Figure R19D), so we concluded that ALY binding to m5C is not related to DNA damage in agreement of the publish data. Interestingly, ALY recruitment at damaged DNA sites was completely dependent on parylation (See introduction 2.2.1.2), supported by opposite results obtained with Olaparib, an inhibitor of PARP enzymes, and PARGi, an inhibitor of PARG, a major enzyme responsible for the catabolism of PAR chains (Figure R21). It has been suggested that R-loop accumulation at DSBs could be a consequence of impaired DNA damage repair and PARP1 could have a potential role in controlling R-loop balance (Cristini et al., 2018). Besides, RNase H1 has been previously described to be recruited at DNA damage sites in a transcriptiondependent manner but parylation inhibition enhanced its recruitment, contrary to what we have seen with ALY. The relationship of this post-translational modification with R-loops and in our case with ALY remains to be clarified. However, it seems that parylation triggers by itself the recruitment of RNA- processing factors such as FUS, TAF15, EWSR1 and SAF-A to DNA damage (Britton et al., 2014). The first three proteins FUS, TAF15 and EWSR1 forms the FET family of RNA-binding proteins whose genes are usually rearranged with various transcription factor genes in sarcomas and rare hematopoietical and epithelial cancers causing oncogenic FET fusion proteins. The recruitment of these four RNA-processing factors to DNA damage show a rapid and transient parylation-dependent kinetics similar to those of ALY (Britton et al., 2014; 144 and other pathologies (Bondarev et al., 2021; Harris et al., 2019; Kaur et al., 2019; Y. T. Lee et al., 2020; Szyf, 2015). A previous work developed in our laboratory revealed a connection between the histone deacetylase inhibitor TSA and DNARNA hybrids in human cells (Salas‐Armenteros et al., 2017). Recent data linked the HDACi romidepsin with the accumulation of R-loops and defines that its role in DNA damage could contribute to a more effective cancer therapy design with HDAC inhibitors (Safari et al., 2021). In this thesis, we show a novel study aimed to identify drugs targeting epigenetic processes (“epigenetic drugs”) that lead to R-loop accumulation and could be used as a tool to induce DNA-RNA hybrids and genetic instability. The long-term goal would be to use them in combined therapeutic strategies for tumors. For this purpose we carried-out an high-throughput immunofluorescence screening based on the signal of the S9.6 antibody after the treatment with a library of epigenetic compounds (for more details see Materials and Methods section 5), some of them at different clinical stages. The protocol was set-up based on the experience of our laboratory in previous screenings with this antibody (Barroso et al., 2019). S9.6 immunofluorescence analysis detected 11 potential drug candidates (4 inhibitors of histone modifications such deacetylation, methylation and demethylation, 2 kinases inhibitors, 1 dual deacetylation and kinase inhibitor, 1 epigenetic reader domain and 2 genotoxic agents) (Figure R28D). It is worth to notice that 3 HDAC inhibitors were found in agreement with previous reports that link histone deacetylation with R-loops in human and yeast (Safari et al., 2021; Salas‐Armenteros et al., 2017; Wahba et al., 2011) providing all of them confidence to our results (Mortusewicz et al., 2016; Terasawa et al., 2014). Inhibitors of histone demethylases and methyltransferases were also identified, consistently with genome-wide data and other reports where different histone methylated marks were associated with regions prone to R-loop formation (Castellano-Pozo et al., 2013; Ginno et al., 2013; Sanz et al., 2016) We focused the work on 3 out of 11 candidates that were analyzed and further validated by S9.6 IF and DRIP assays The selected drugs WP1066, CX- 6258, whose main targets are JAK2 (Verstovsek et al., 2008) kinase and pan- 145 PIM kinases (Haddach et al., 2012) respectively, and Tranylcypromine, which is a clinically validated antidepressant (Frieling & Bleich, 2006) known to target monoamine oxidases A and B (Salsali et al., 2004) and histone demethylases LSD1 and LSD2 (Binda et al., 2010; Ji et al., 2017). Despite these three drugs have different targets, and it is known that they can have pleiotropic effects, interestingly they have in common a relationship with chromatin condensation and some histone modifications. In the next paragraph a brief summary of each drug is provided. JAK2 kinase belongs to the non-receptor tyrosine kinase, the Janus kinase (JAK) family that regulates signal transductions of hematopoietic cytokines and growth factors (Yamaoka et al., 2004). These kinases bind cytokine receptors and play essential roles in cell growth, survival and development by phosphorylation and activation of transcription factors (STAT signal transducers and transcription activator) (Brooks et al., 2014). JAK2 kinase has been linked with chromatin changes in several studies including an H3Y41 phosphorylation that excludes heterochromatin protein 1α (HP1α) from chromatin (Dawson et al., 2009) and this have been described to be negatively regulated by the H3K9 methylation (Son et al., 2012). Another example of the connection of JAK2 kinase with histone methylation is provided by the JAK2V617F mutation, a constitutively activated JAK2 form found in many patients with myeloproliferative neoplasm, that strongly phosphorylates the arginine methyltransferase PRMT5 reducing its ability to methylate histones and finally promoting myeloproliferation (F. Liu et al., 2011). CX-6258 inhibits more efficiently and at lower concentrations PIM1 which has been described to phosphorylate heterochromatin protein ɣ (HP1ɣ) increasing its capacity to bind H3K9me3 resulting in heterochromatin formation and proliferative gene suppression (Jin et al., 2014; Koike et al., 2000). On the other hand PIM1 kinase also phosphorylates H3S10 at the FOSL1 enhancer and increase Ser2-P RNAPII stimulating elongation step (Zippo et al., 2009). In addition, the phosphorylation of H3S10 in MYC target genes mediated by PIM1 kinase leads to transcriptional activation and oncogenic transformation (Zippo et al., 2007). Finally, Tranylcypromine inhibits the cytochrome P450 2A6 (CYP2A6) involved in the metabolism of xenobiotics in the body, but it also inhibits potently and irreversibly histone demethylase lysine-specific demethylase 1 (LSD1) 146 (Rudolph et al., 2013), an enzyme that is ubiquitously overexpressed in diverse cancers (Dai et al., 2020). Despite the pleiotropic functions of the targets, these three inhibitors could contribute indirectly to changes in chromatin structure or gene expression that finally could lead to R-loop formation. Independently from the mechanism of action, our goal was to identify compounds that lead to R-loop accumulation and genome instability. Tranylcypromine did not cause an increase in DNA damage so we decided to exclude it from future analysis. Therefore, we finally focused our analysis on WP1066 and CX6258 kinase inhibitors because both of them showed an increase in DNA damage, although only in the case of WP1066 it was R-loop-dependent (Figure R30B). The depletion with siRNA of the main targets of WP1066 and CX-6258, JAK2 and PIM1 respectively, did not lead to genome instability R-loop-dependent (Figure R32A-B). This data suggest that the drugs seem to have more pleiotropic effects than the solely inhibition of their main targets. Interestingly, we also checked the effect of the depletion of STAT3 which is a transcription factor that when phosphorylated by JAK2 undergoes dimerization and enters into the nucleus. WP1066 not only inhibits the phosphorylation of JAK2 and promotes its degradation, but also inhibits the phosphorylation of STAT3 (Ferrajoli et al., 2007) (Figure D2A). We found that the depletion of STAT3 caused an increased in R- loop accumulation that was reduced upon RNase H treatment (Figure R32C). Since STAT3 is a transcriptional regulator, that lead to the expression and repression of target genes involved in several cellular pathways, it would be necessary to confirm that this R-loop accumulation is not an indirect effect of a transcription activation. Interestingly, STAT3 phosphorylation can also induce PIM1 expression (Zemskova et al., 2008). This relationship between these two targets prompt us to test single and combinatory treatments (Figure R35). Double treatments with genotoxic and other chemical agents are usually used to find synergistic relationships that are relevant for their use on clinical assays, what is known as synthetic lethality phenotype. As described in results, we tested the combination of the best selected drugs of our screening with the histone deacetylase TSA known to induce R-loop accumulation (Salas‐ 147 Armenteros et al., 2017) (Figure R34) and also in cells depleted of protective factors against harmful R-loops (THOC1 and SETX) (Figure D2B). However, we could not find any synergistic effect on DNA damage or cell proliferation (Figure R33). Finally, we show some preliminary and promising data for a possible therapeutic potential of these compounds using them in single or combination treatments in cell lines derived of patients. Indeed, CX-6258 treatment cause a differential and significant effect on patient cells with FANCD2-depleted cells versus corrected ones and the combination of WP1066 and CX-6258 treatments also showed a synergistic effect in adenocarcinoma cells (Figure R35). Although adenocarcinoma cell lines with BRCA2 depletion was not specifically targeted by the combination of the drugs, this information could be useful for new combination treatments in recurrent and resistant tumors for adenocarcinomas. Treatments with WP1066 and/or CX-6258 need to be extended to other cell lines deficient in R-loop resolution or DNA repair, and to a set of cancer cell lines to validate its therapeutical potential (Figure D2B). Since these inhibitors have already a potential clinical relevance our work adds a new uncover mechanism of genotoxicity mediated by R-loop accumulation that could open new strategies for their use and improvement of treatments. Figure D2. Epigenetic inhibitors inducing R-loops and their potential use in combined therapies. (A) JAK2 kinase phosphorylates the signal transduction and transcription factor STAT3 allowing its dimerization and entrance to the nucleus. STAT3 regulates gene expression of different target genes, among them PIM1 kinase, contributing to different cellular pathways and cell proliferation. The compound WP1066 is able to inhibit and degrade JAK2 and also STAT3. 148 The compound CX-6258 is a PIM kinase inhibitor. These kinases are associated with MYC expression, cell proliferation and survival and are considered proto-oncogenes. (B) Scheme of the possible use of the single or the combination treatments with the selected compounds WP1066 and/or CX-6258 in order to increase R-loop levels, genome instability and reduce cell viability. These inhibitors could be used in combination with other factors such as genetic backgrounds deficient in R-loop prevention or resolution, DNA repair, as well other drugs looking for synthetic lethality and new combined therapeutic treatments. 149 CONCLUSIONS 150 151 1- Changes in the levels of the RNA-binding factor ALY causes R-loop accumulation and DNA damage, suggesting a key role of this factor in genome integrity. 2- ALY overexpression leads to replication impairment and an altered chromatin appearance. 3- ALY, but not its partner RNA helicase UAP56, is recruited at DNA breaks in a PARylation-dependent manner, majorly independent of transcription and R- loops. 4- Changes in ALY levels impair DNA damage signaling and repair, implying a role of ALY in DNA damage response. 5- ALY depletion leads to higher levels of ssDNA-binding protein RPA1 and nuclease MRE11, both involved in repair, suggesting an important role of ALY at the first stages of the DNA damage response 6- In a screening we identified three top drugs (WP1066; CX-6258; Tranylcypromine) targeting enzymes involved in epigenetic processes that cause R-loop accumulation. From these, kinase inhibitors WP1066 (JAK2 inhibitor) and CX-6258 (PIM inhibitor) also lead to DNA damage. 7- Depletion of STAT3, a transcription factor target of the compound WP1066 leads increase in R-loop levels. 8- Single and combination treatments with WP1066 and CX-6258 causes a significant effect on cell viability in cell lines deficient in FANCD2 and BRCA2, factors with a key role in genome instability. 152 CONCLUSIONES 153