Crosstalk between DNA end resection and RNA metabolism

Crosstalk between DNA end resection and RNA metabolism ! ! Mª Rosario Prados Carvajal Tesis Doctoral realizada en el departamento de Genética de la Universidad de Sevilla y en el Centro Andaluz de Biología Molecular y Medicina Regenerativa (CABIMER) para optar al título de Doctora por la Universidad de Sevilla 2019 Tutor y director: Pablo Huertas Sánchez “ El investigador sufre las decepciones, los largos meses pasados en una dirección equivocada, los fracasos. Pero los fracasos son también útiles, porque, bien analizados, pueden conducir al éxito. Y para el investigador no existe alegría comparable a la de un descubrimiento, por pequeño que sea…” Alexander Fleming Qué gran verdad esta frase del majestuoso Fleming. A lo largo de esta tesis he tenido muchos momentos de bajón por esos experimentos que no salían o no sabía qué significaban, pero también muchísimos momentos de alegría. Sin embargo, el pequeño gran descubrimiento que más me ha llenado ha sido el poder compartir estos años de una manera u otra con todos vosotros. Son tantas cosas y a tantos los que os tengo que agradecer que no se ni por donde empezar… Pablo gracias por acogerme en tu laboratorio cuando era solo una cría y hacer que me apasionara este mundo de la investigación. Ya son más de 8 años los que llevas aguantándome, que se dice pronto... Gracias por ser el jefe ideal, por preocuparte y apoyarnos, no solo en lo profesional sino también en la personal, por implicarte tanto con cada uno de nosotros y ayudar a que haya tan buen ambiente de trabajo. Gracias a todos mis PHS porque cada uno me habéis aportado algo y siempre os recordaré con muchísimo cariño. Sois muchos los que habéis pasado por el lab a lo largo de estos años, y quizás me olvide de alguno en este momento, sorry no es mi intención. Sin embargo, tenía que hacer mención especial a aquellos que habéis sido mi gran familia durante estos años. Ana, gracias por enseñarme casi todo lo que sé y estar siempre dispuesta a echar una mano, aunque fuera a las 12 de la noche. Sonia, gracias por tratarme como a una hija y cuidar de mi cuando más lo necesitaba. María Jesús, el lab fue un desastre tras tu partida, tu ausencia se hizo notar… Mi jefi bonita preciosa (Cristina), gracias por ser mi madre científica. Dani, mi gran confidente, gracias por cada consejo, no se qué hubiera hecho sin tus charlas… María, que pasara lo que pasara siempre tenía una sonrisa, gracias por animarnos como tú sólo sabes. A mi marido (Andrés), gracias por cada café y cada charla. Isa, tía lista donde las haya, siempre serás un gran referente. Ángeles, ¡de cuántos líos de papeleos nos libraste!, nos sabes cuánto te extrañamos ahora… Inés y Blanca, aunque estuvierais poco tiempo, os cogimos muchísimo cariño y ojalá que nuestros caminos se vuelvan a cruzar. Javi, siempre dispuesto a todo, qué encanto de persona. Mi Fernandillo y sus comentarios, el lab perdió a un gran compositor y humorista. Gracias también a mis niñas, ¿qué habría sido de mi sin vosotras?. A ti Cintia, por ser casi una hermana o más bien una gemela, por recorrer este largo camino codo a codo conmigo y estar siempre ahí. A mi Rosa, por transmitirme esa alegría constante que tiene y animarme en los momentos no tan buenos. Guille, menos mal que vuelves porque nos faltaba algo sin ti. Andresillo, a pesar de que nos riamos de tus peculiaridades, sé que llegarás muy lejos, ¡eres muy grande y te dejamos en buenas manos!. Fernando y ese buen rollo que transmite, ¡qué gran adquisición!. Sabrina, siempre dispuesta a ayudar a los demás. Néstor, vas a ser un gran jefe. Amador, aunque seas un pelín despistaillo, se te ve súper inteligente y sé que lograras todo lo que te propongas. Andrea, la niña más dulce que pudo aparecer por el lab. Mucho ánimo a todos, aunque sé que os irá genial, porque sois todos unos profesionales de los pies a la cabeza y lo más importante, unas personas increíbles. Gracias a todos de corazón porque habéis hecho que estos años hayan sido los mejores de mi vida, nunca os olvidaré PHS. A todo el personal de CABIMER, por ayudarme con los experimentos o simplemente por hacer el día más ameno saludando con una sonrisa cuando nos cruzábamos en los pasillos. Esos pequeños detalles hacen que una cambie la perspectiva en ciertos días. Tampoco podía olvidarme de mi segundo lab. Gracias a Neus y su laboratorio por hacer que mi estancia en Estocolmo, aunque fuese corta, fuera enriquecedora y me sintiera en todo momento como en casa. No pude tener una jefa mejor, gracias por todo Neus. Gracias también a Judit por recibirme con los brazos abiertos y hacer que todo fuera más fácil. A mi uruguayo, Martín, qué de horas hemos pasado juntos y qué de risas. Patri, gracias por acogerme en tu casa y ser una gran amiga. Jordi, nuestro escalador, ¡qué gran tío!. Shruti, thanks for being so patient and nice with me. I really enjoyed watching Indian movies or doing the flashmob. Thank you all guys for being my perfect family in Stockholm. I will never forget you! A mis padres y hermana por su amor incondicional, por apoyarme constantemente e incluso intentar entender en qué trabajaba. Esta tesis no hubiera sido posible sin vosotros, sin vuestras palabras alentadoras, sin vuestra ayuda y sobretodo sin vuestra paciencia. A Manolo, mi feo, gracias por aguantar más que nadie, por entenderme tan bien y saber cuando necesitaba un abrazo o unas palabras de ánimo. Por escuchar mis ensayos mil y una veces e intentar entender qué trataba transmitir. Por apoyarme y animarme a que persiga mis sueños por mucho cueste. Por ser la persona perfecta para mi…. Tampoco podía olvidarme de mis biólogos, Laura, Ana y Carlos, que a pesar de que hayan pasado varios años desde que acabamos la carrera y que nos separan varios miles de kilómetros, siguen siendo un pilar fundamental en mi vida. Espero que no me faltéis nunca y sigamos reuniéndonos muchos años más, ya sea en Sevilla, Londres, Suiza o allá donde el destino nos depare. A mi gente del baile, aparecisteis en el momento preciso para abrirme los ojos ante un mundo nuevo e increíble. Gracias por cada baile, cada cerveza, cada charla filosófica y cada muestra de cariño. Gracias por todos los momentos que hemos compartido y por todos los que quedan por vivir. ¡Me dais vida! Y por último, pero no menos importante, gracias a mis amigos, familiares y a todas aquellas personas que os habéis cruzado en mi camino para hacer mi vida aún más bonita. Gracias por preocuparos por mis celulitas y por transmitirme vuestros ánimos para que siga investigando. Gracias de todo corazón porque sin todos vosotros no hubiera sido posible. I did this work thanks to receiving a FPU fellowship from the Spanish Ministry of Education (FPU. Additionally, these projects were funded by two R+D+I grants from the Spanish Ministry of Economy and Competitivity (SAF2013-43255-P and SAF2016- 74855-P) and a Starting Grant from the European Research Council (DSBRECA). Abstract DNA suffers thousands of lesions from exogenous and endogenous sources. Nevertheless, double-strand breaks (DSBs) are the most dangerous form of DNA damage because if they are not properly repaired they can lead to mutations and genomic instability. In order to repair them, cells have developed two major pathways: Homologous Recombination and Non-Homologous End-Joining. CtIP is a key protein in this response since it promotes faithful repair of DSBs by licensing homologous recombination. Recently, a crosstalk between DNA Damage Response (DDR) and RNA metabolism at different levels has been discovered. Thus, in this Thesis we studied this relationship deeply. We described that the splicing complex SF3B is involved in in DNA damage repair. Although part of such function is through CtIP, we observed that SF3B plays additional roles in DNA double strand break repair that are independent of its functional relationship with this factor. Additionally, we demonstrated that this relationship is bidirectional, since CtIP is involved in DNA damage-dependent and independent alternative mRNA splicing of several genes. The careful study of the CtIP- dependent splicing of one of those genes, the one that codes the helicase PIF1, suggest that those events are important for the fine tuning of the DNA damage response. Furthermore, we reported a crosstalk between the DDR and RNA editing. RNA editing is, indeed, upregulated upon exposure to DNA damage in a fashion that requires CtIP and the checkpoint kinase ATR. Moreover, our data agree with the idea that the RNA editing enzymes ADARs have a role in the removal of DNA:RNA hybrids that facilitates DSB repair. Resumen El ADN sufre muchísimos tipos de daño debido a fuentes externas o internas. No obstante, los cortes de doble cadena son de los más peligrosos porque si no se reparan adecuadamente, pueden conducir a mutaciones e inestabilidad genómica. Para reparar este daño, las células han desarrollado principalmente dos vías: la recombinación homóloga o la unión de extremos no homólogos. CtIP es una proteína clave en esta respuesta al daño porque activa la reparación de los cortes de doble cadena vía recombinación homóloga. Recientemente se ha descrito una conexión a diferentes niveles entre la respuesta al daño en el ADN (DDR por sus siglas en inglés) y el metabolismo del ARN. Por eso, en esta Tesis hemos estudiado dicha relación en profundidad. Descubrimos que el complejo de procesamiento de intrones denominado SF3B está involucrado en la reparación del daño del ADN. Aunque parte de las funciones que realiza en este proceso son a través de CtIP, también hemos demostrado que SF3B tiene otro papel independiente de su relacional funcional con esta proteína. Además, hemos comprobamos que esta relación es bidireccional, ya que CtIP también esta involucrado en el procesamiento alternativo de intrones de numerosos genes, tanto de manera constitutiva como en respuesta a la presencia de daño en el ADN. El estudio cuidadoso del la maduración de intrones dependiente de CtIP de uno de esos genes, el que codifica la helicasa PIF1, sugiere que estos eventos permiten el ajuste fino de la respuesta al daño en el ADN. Asimismo, pudimos observar una interrelación entre la DRR y el proceso de edición del ARN. La edición del ARN, de hecho, aumenta como respuesta a la presencia de daño en el ADN en un proceso regulado por CtIP y la quinasa ATR. Además, nuestros datos están de acuerdo con la idea de que las proteínas de edición de ARN ADARs juegan un papel en la disolución de híbridos de ADN:ARN que facilita la reparación de cortes de doble cadena en el ADN. Indexes Indexes ! I! TABLE OF CONTENTS INTRODUCTION ........................................................................................................ 1 1. DNA damage ............................................................................................................ 3 2. DNA Damage Response (DDR) .............................................................................. 4 3. Double strand breaks repair pathways ...................................................................... 7 3.1. Non-homologous end-joining (NHEJ) .............................................................. 8 3.2. Homologous recombination (HR) ................................................................... 10 3.2.1. DNA end resection ................................................................................ 14 4. CtBP-interacting protein (CtIP) .............................................................................. 18 5. mRNA Splicing ...................................................................................................... 21 5.1. Alternative splicing ........................................................................................ 23 5.1.1. Alternative splicing in DNA repair ........................................................ 24 6. RNA editing ............................................................................................................ 24 OBJECTIVES ............................................................................................................ 29 RESULTS ................................................................................................................... 33 Chapter one: The SF3B role in DNA repair ........................................................... 35 1. The splicing complex SF3B interacts with CtIP .................................................... 35 2. The SF3B complex promotes DNA end resection ................................................ 38 3. SF3B is involved in homologous recombination ................................................... 42 4. The SF3B complex controls the abundance of homologous recombination proteins ................................................................................................................... 44 5. SF3B is required for both the pro- and antirecombination branches of the DDR ...................................................................................................................... 48 6. The SF3B complex controls CtIP recruitment to DSBs. ........................................ 51 7. CtIP overexpression rescues SF3B depletion phenotype in DNA resection .......... 53 Chapter two: A crosstalk between the DNA end resection machinery and DNA- damaged induced RNA alternative splicing ............................................................ 57 8. SF3B controls mRNA expression of many genes .................................................. 57 9. SF3B controls damaged-induced alternative RNA splicing ................................... 59 Indexes ! VIII! INDEX OF TABLES Table R1. Genes upregulated upon SF3B2 depletion .................................................. 58 Table R2. Genes downregulated upon SF3B2 depletion ............................................. 59 Table R3. Total splicing changes in response to IR and SF3B2 depletion .................. 60 Table R4. Specific splicing changes in response to IR and SF3B2 depletion ............. 60 Table R5. Genes upregulated by CtIP depletion ......................................................... 64 Table R6. Genes downregulated by CtIP depletion ..................................................... 64 Table R7. Total splicing changes in response to IR and CtIP depletion ..................... 65 Table R8. Specific splicing changes in response to IR and CtIP depletion ................. 65 Table R9. Genes whose splicing is affected in response to DNA damage and CtIP depletion ....................................................................................................................... 66 Table M1. Human cell lines used in this Thesis ........................................................ 113 Table M2. siRNAs used in this Thesis ...................................................................... 117 Table M3. Plasmids used in this Thesis ..................................................................... 118 Table M4. DNA primers used in this study ............................................................... 131 Table M5. Primary antibodies used in this study ...................................................... 139 Table M6. Secondary antibodies used in this study .................................................. 141 Indexes ! IX ABBREVIATONS alt-NHEJ Alternative non-homologous end-joining BFP Blue fluorescent protein BIR Break-induced replication BrdU Bromodeoxyuridine BS Branch site BSA Bovine serum albumin CDK Cyclin-dependent kinase CHX Cycloheximide CPT Camptothecin CTD Carboxyl-terminal domain DDR DNA damage response DDRNA Drosha and Dicer-dependent small RNA dilncRNA Damage-induced long non-coding RNAs diRNA DSB-induced small RNA D-loop Displacement loop DMEM Dulbecco’s Modified Eagle Medium DMSO Dimethyl sulfoxide DNA-PKcs DNA-dependent protein kinase catalytic subunit DSB Double-strand break dsRNA Double-strand RNA EDTA Ethylenediaminetetraacetic acid FISH Fluorescence in situ hybridization GFP Green fluorescent protein HBS HEPES buffered saline HR Homologous recombination ICL Interstrand cross-link IF Immunofluorescence IP Immunoprecipitation IR Ionizing radiation IRIF Ionizing-radiation induced foci Indexes ! X! LacO Lac operator lncRNA Long non-coding RNAs MES 2-(N-morpholino) ethanesulfonic acid MMEJ Microhomology-mediated end-joining MOI Multiplicity of infection MRN MRE11A-RAD50-NBS1 miRNA MicroRNA mRNA Messenger RNA ncRNA Non-coding RNA NHEJ Non-homologous end-joining NER Nucleotide excision repair NLS Nuclear localization sequence ORF Open reading frame PBS Phosphate buffered saline PDS Pyridostatin PIKK Phosphatidylinositol 3-kinase-like protein kinase PLA Proximity ligation assay PTM Post-transcriptional modification PVDF Polyvinylidene difluoride RFLTS Replication foci targeting sequence RFP Red fluorescent protein ROS Reactive Oxygen Specie RNP Ribonucleoprotein particles SDS Sodium dodecyl sulphate SDSA Synthesis-dependent strand annealing shRNA Short hairpin RNA siRNA Small interfering RNA SMART Single-molecule analysis of resection tracks snRNP Small nuclear ribonucleoprotein ss Splice site SSA Single-strand annealing SSB Single-strand break SSC Saline-Sodium Citrate Indexes ! XI ! ssDNA Single-stranded DNA TC-HR Transcription-coupled homologous recombination TLS Translesion synthesis UV Ultraviolet light Vp16 Etoposide UTR Untranslated regions WB Western blotting Introduction Introduction ! 3 1. DNA damage ! The DNA is the molecule that contains the genetic information required for the correct development and functioning of living organisms. Thus, one of the cell priorities is to maintain the stability of the genome and to guarantee the faithful transmission of genomic information across generations. However, every day DNA accumulates tens of thousands of lesions induced by either external agents or endogenous sources derived from the cellular metabolism (Jackson and Bartek, 2009). Spontaneous DNA alterations can occur during physiological processes like DNA transcription or replication, which can introduce nucleotide mismatches, insertions and deletions that change DNA sequence, etc. (Kim et al., 2007; Kim and Jinks-Robertson, 2012). Additionally, DNA strands breaks can be caused by abortive topoisomerase I and topoisomerase II activity (Kunkel & Bebenek, 2000) or by products of cellular metabolism like reactive oxygen compounds and reactive nitrogen species, endogenous alkylating agents, etc. (Lindahl, 1993; De Bont and van Larebeke, 2004). Otherwise, DNA can be damaged by exogenous sources such as ionizing radiation (IR), derived from cosmic radiation or radioactive compounds used in clinical protocols to treat diseases like cancer, X-rays, gamma rays; or ultraviolet light (UV) coming from sunlight (Cadet et al., 1997; Hoeijmakers, 2001). In addition, several chemotherapeutic agents used in cancer therapy as well as other chemical agents or cigarette smoke, can also generate DNA lesions (Ward, 1988; Cheung-Ong, Giaever and Nislow, 2013). If these DNA lesions are not repaired properly, they lead to the accumulation of mutations, chromosome aberrations and genome instability, which can causes human diseases, premature aging and predisposition to several types of cancer (Aguilera and Gomez-Gonzalez, 2008; So et al., 2017). Thus, in order to protect DNA from those lesions, cells have developed several mechanisms to detect, signal and repair DNA damage. As there are diverse types of DNA damage, there are different specific repair and signaling pathways. The most common lesions only affect one DNA strand, generating alterations in the canonical nucleotide structure. Examples of them are chemical modifications of either bases or nucleotides, mismatches in the DNA sequence, formation of pyrimidine dimers, bulky adducts formed by chemical compounds adhered to DNA, or covalent binding between nucleotides of the same or Introduction ! 4 the complementary DNA strands (intrastrand and interstrand crosslinks, respectively) (De Bont and van Larebeke, 2004; Noll, Mason and Miller, 2006). Although there are specialized repair pathways for each of those types of damage, all of them work in a similar fashion by recognizing the DNA damage with specific proteins, excising the damaged strand and generating a single-stranded DNA (ssDNA) gap that is repaired using the undamaged and complementary strand as a template (Hoeijmakers, 2009). However, there are also lesions that can break the DNA sugar-phosphate backbone. The most frequents are single-strand breaks (SSBs) in which only one strand is interrupted (Caldecott, 2008). In this type of lesion the complementary strand can still be used as template to repair the damage after some processing of the damaged strand. Nevertheless, if the break affects simultaneously both strands, the so-called double strand break (DSB), which interrupts the continuity of the DNA, there is not an intact strand that can be used as template. Thus, DSBs require more complex mechanisms to be repaired. For that reason, DSBs are the most cytotoxic form of DNA damage and its repair requires the intervention of multiple proteins, as well as, their coordination with several cellular events. This network of mechanisms is collectively called the DNA damage response (DDR) (Ciccia and Elledge, 2010)(see below). Moreover, the machinery implicated in DSB repair changes depending on multiple factors, including the type of break and the position on cell cycle in which it has been produced (Hartlerode and Scully, 2009; Huertas, 2010). 2. DNA Damage Response (DDR) ! The DDR is the process that cells use to detect DSBs and promote a cascade of events to repair them. This process modulates several aspect of the cellular metabolism, such as cell cycle, which is arrested or slowed down to increase the time available for DNA repair and prevent duplication or segregation of broken chromatin (Jackson and Bartek, 2009; Ciccia and Elledge, 2010). Moreover, the DDR promotes changes in DNA structure, replication and transcription, as well as, in the recruitment of proteins close to the breaks. Indeed, this process also activates post-transcriptional modifications (PTMs) in those proteins involved in DNA damage repair and, it is responsible of the induction of apoptosis when the load of damage is too high (Jackson and Bartek, 2009; Ciccia and Elledge, 2010; Polo and Jackson, 2011; Dantuma and van Attikum, 2016). Introduction ! 5 Figure I1. Double strand break signalling and response. When a DSB is produced (yellow lighting bolt), the MRN complex (MRE11-RAD50-NSB1) detects the damage and binds to DNA (1.I), what provokes the recruitment of the kinase ATM by interaction with NSB1. This kinase phosphorylates several substrates (marked with a “P”) as MRN or the histone variant H2AX in the vicinity of the break (1.II). Nucleosomes with canonical H2A histones (yellow) or the H2AX variant (orange) are represented in the figure. Theses phosphorylations serve for the recruitment of MDC1 to the sites of damage. MDC1 recruits more ATM and facilitates further H2AX phosphorylation. This attracts additional MDC1 molecules, creating a positive feedback loop and the spreading of the signal over the chromatin (1.III). BRCA1 and 53BP1 are also recruited to the breaks and contribute to damage signalling by stabilizing MRN-ATM interaction, being targets of post-translational modifications, such as sumoylations (marked with a “S”) or ubiquitinations (represented by “Ub”), and recruiting other proteins involved in DDR (1.III). Then, ATM phosphorylates and activates CHK2, a key checkpoint kinase that initiates a secondary cascade of phosphorylations (1.IV). Thus, CHK2 activates p53 (1.V) provoking the cell apoptosis (1.VI) or the cell cycle arrest (2,VI) to give time to the cell to repair the DNA damage properly (VII). Another important transducer is ATR. This kinase is recruited by ATRIP and detects ssDNA coated by RPA (red circle) formed during the processing of DSB by DNA resection (2.I-2.II). The full activation of ATR requires the recruitment of ETAA1 or TOBP1 (2.III), which promotes that ATR phosphorylates several substrates like γ H2AX or BRCA1, collaborating in that way with the DDR. Additionally, ATR activates CHK1 (2.IV), which phosphorylates CDC25 (2.V) that arrests the cell cycle (2.VI) or phosphorylates p53 with the possibility of inducing apoptosis (1.VII). ! Introduction ! 6 The DDR is extremely hierarchical. This process starts with the detection of the damage by sensor proteins. Then, transducer proteins are recruited to the break and activate a large number of mediator proteins, which are involved in the amplification of the signal. Finally, effector proteins carry out the cellular responses to repair DNA damage (Harper and Elledge, 2007). Interestingly, some effectors can act also like sensors, creating a feedback loop that maintains and spreads the signaling (Shiloh and Ziv, 2013). Although the DDR is conserved in evolution in all eukaryotes, there are differences in the nomenclature of proteins involved in diverse organisms. Our work has been carried out in mammals cells, thus in order to simplify the reading of this Thesis, all proteins will be named according to the human nomenclature (Arcas et al., 2014). The initiation of the DDR relies on the activation of key transducer proteins belonging to the family of phosphatidylinositol 3-kinase-like protein (PIKKs), namely Ataxia-Telangiectasia Mutated (ATM), Ataxia Telangiectasia and Rad3 related (ATR) and DNA-dependent Protein Kinase (DNA-PK). ATM and DNA-PK are activated directly by DNA breaks, but ATR is stimulated in the presence of single strand DNA (ssDNA) resultant of processed DNA breaks or replication stress (Lovejoy and Cortez, 2009). One of the main sensors in the DDR is the complex MRE11-RAD50-NBS1 (MRN), which tethers and processes the broken ends (Stracker et al., 2004) (Figure I1.1.I). Then, MRN recruits and retains ATM, through a direct interaction between this kinase and NBS1 (Uziel et al., 2003; You et al., 2005). ATM is considered the main transducer protein in the DDR that initiates and regulates the cascade of signals to repair DNA damage. Thereby, ATM phosphorylates serine-139 of the histone variant H2AX, known as γH2AX, on chromatin flanking DSBs (Savic et al., 2009) (Figure I1.1.II). Afterward, mediator proteins such as Breast Cancer 1 (BRCA1), p53 Binding Protein 1 (53BP1) or Mediator of DNA Damage Checkpoint 1 (MDC1) are recruited and stabilize this interaction (Lee et al., 2010) (Figure I1.1.III). Indeed, the phosphorylation of H2AX contributes to the recruitment and retention of MDC1, which attracts more ATM, therefore creating a feedback loop through phosphorylation of those proteins that spread and accelerates the recruitment of other DDR factors (Lou et al., 2006; Bartek and Lukas, 2007) (Figure I1.1.III-1.IV). Other important transducer is the kinase ATR, which detects ssDNA formed during the processing of DSB by DNA resection (Figure I1.2.I) (for specific details on Introduction ! 7 resection see below section 3.2.1). ATR is recruited by its partner protein ATR- Interacting Partner (ATRIP), which senses ssDNA coated by the Replicative Protein A (RPA) complex (Zou and Elledge, 2003; Namiki and Zou, 2006) (Figure I1.2.II). Indeed, the full activation of ATR requires the recruitment of another protein independently of ATR-ATRIP known as DNA Topoisomerase II Binding Protein 1 (TOPBP1) (Acevedo, Yan and Michael, 2016) (Figure I1.2.III). Recently, it has been identified a second ATR-activator, Ewing Tumor-Associated Antigen 1 (ETAA1), which is recruited to ssDNA by binding to RPA (Haahr et al., 2016) (Figure I1.2.III). ATR also share some substrates with ATM like γH2AX or BRCA1, collaborating in that way with the DDR (Marechal and Zou, 2013). Then, the cascade of signaling is so far extended thanks to two critical effector Checkpoint Kinases (CHK): CHK2, which is phosphorylated by ATM (Figure I1.1.IV) and CHK1, which is activated by ATR (Figure I1.2.IV). Finally, both kinases activate effector proteins like tumor suppressor protein p53 (Figure I1.1.V) and Cell Division Cell Cycle 25 (CDC25) (Figure I1.2.V). Those effectors promote apoptosis (Figure I1.1.VI) or cell cycle arrest (Figure I1.2.VI) to repair DNA damage, respectively (Figure I1.VII) (Matsuoka et al., 2007; Lorenza P. Ferretti, Lafranchi and Sartori, 2013). Additionally, phosphorylation and other post-transcriptional modifications like ubiquitylation, neddylation, sumoylation, acetylation, methylation or poly (ADP)- ribosylation are very important in the signaling cascade in the DDR to recruit or regulate proteins involved in this process, as well as to modify histones close to the breaks to regulate chromatin configuration (Figure I1.1.V) (Price and D’Andrea, 2013; Brown and Jackson, 2015). 3. Double strand breaks repair pathways ! As mentioned before, double strand breaks (DSBs) are the most cytotoxic DNA damage. For these reason, cells have developed several mechanisms to repair them, usually grouped in two major categories: Non-Homologous End-Joining (NHEJ) and Homologous Recombination (HR) (Hartlerode and Scully, 2009). Introduction ! 14 subpathways, with hyper-resection favoring mutagenic SSA over error-free gene conversion (Jimeno et al., 2015). 3.2.1. DNA end resection ! As mentioned above, DNA end resection is the first step in every homologous recombination pathway, and consists of the 5’-3’ degradation of one strand at each side of the break giving rise a ssDNA that is immediately protected by RPA complex (Huertas, 2010; Liu and Huang, 2016; Sharad C. Paudyal and You, 2016). Indeed, RPA binding to ssDNA activates the ATR-mediated checkpoint, leading to the regulation of cell cycle progression and also blocks KU binding, inhibiting NHEJ. It has been well established that DNA end resection is a two-step mechanism (Cejka, 2015; Symington, 2016). First, a slow and limited nucleolytic-processing of the ends or short-range resection, catalyzed by MRN complex and CtIP with the help of additional processivity factors such as BRCA1 (Anand et al., 2016) (Figure I4.I-II). As mentioned in section 2, MRE11 is one of the components of the MRN complex. This protein has many enzymatic abilities, including an endonuclease activity on ssDNA, a 3’-5’ exonuclease activity on dsDNA and a manganese-dependent 5’-3’exonuclease activity (Trujillo and Sung, 2001; Cejka, 2015). The current consensus is that MRE11 catalyze the initial DNA end resection through an endonucleolytic cleavage of the 5’-ended strand away from the DSB followed by 3’-5’ short digestion towards the DNA end (Myler et al., 2017) (Figure I4.III) possibly in conjunction with the Exonuclease 3′–5′ Domain-containing 2 (EXD2), which degrades DNA ends with its 3’-5’ exonuclease activity (Figure I4.IV) (Broderick et al., 2016). Interestingly, dimerization of MRE11 is important for binding to DNA and its interaction with RAD50 (Williams et al., 2008). RAD50 is an ATPase that maintains the conformation of MRN complex and binds DNA ends, holding them in close proximity (de Jager et al., 2001; Westmoreland et al., 2009). The third component of MRN is NBS1, which does not have enzymatic activity, but promotes DNA binding, the interaction with CtIP, the activation of ATM and therefore the activation of DDR (Williams et al., 2009). CtIP, among other functions (see below section 4), plays a key role activating the first step of DNA end resection, interacting with MRN and stimulating the endonuclease activity of MRE11A. This role is regulated by post-transcriptional modifications of CtIP, including its phosphorylation Introduction ! 15 at threonine 847 cyclin-dependent kinase, which limits resection to S and G2 phases (Huertas, 2010; Anand et al., 2016). ATR also controls CtIP recruitment to chromatin by its phosphorylation at threonine 859 (Peterson et al., 2013). Apart from the MRN complex and CtIP, proper DNA end resection requires the action of additional factors such as the tumor suppressor BRCA1 (Figure I4.II-I4.III). The interaction between BRCA1 and CtIP promotes DNA end resection processivity and is required to promote the removal of 53BP1 and Rap1-Interacting Factor 1 (RIF1) from DNA ends and to accelerate the speed of the DNA end resection process (Escribano-Diaz et al., 2013; Cruz-García, López-Saavedra and Huertas, 2014). This interaction is controlled by S/G2-CDK-mediated phosphorylation of CtIP at serine 327 (Yu and Chen, 2004; Cruz- García, López-Saavedra and Huertas, 2014). Initiation of DNA end resection by the short-range DNA end resection promotes the recruitment of other components involved in a second stage, known as long-range processing step of the DNA ends away from de DSB (Cejka, 2015). In this second step, resected DNA is extended by nucleases to facilitate homology search for homologous recombination. Two redundant pathways that involve either the exonuclease 1 (EXO1) or the helicase Bloom syndrome protein (BLM) together with the exonuclease activity of the DNA replication helicase/nuclease 2 (DNA2) are required for extensive resection (several kilobases) leading to the formation of extensive 3’ ssDNA overhangs (Nimonkar, Genschel and Kinoshita, 2011). However, recent studies suggest that the EXO1 resection branch is the preferred pathway in human cells, while BLM/DNA2 plays a secondary role (Tomimatsu et al., 2012; Myler et al., 2016). On one hand, EXO1 degrades the 5’ terminated strand to generate the extensive 3’ ssDNA tail (Figures I4.V and I4.VI, right side). Indeed, EXO1 resection function is positively regulated by MRN, BLM, RPA, Proliferating Cell Nuclear Antigen (PCNA) and the RAD9-HUS1-RAD1 complex (9-1-1) (Nimonkar, Genschel and Kinoshita, 2011). Furthermore, CtIP promotes the recruitment of EXO1 to the DNA damage, although it negatively regulates EXO1 nuclease activity (Eid et al., 2010; Sharad C Paudyal and You, 2016). On the other hand, the second long-range resection pathway is catalysed by DNA2 and BLM. DNA2 possesses ssDNA-specific nuclease activity that cleaves both 3’ and 5’- terminated ssDNA. BLM generates ssDNA tails that are engaged by RPA, which protects 3’strand against DNA2 and facilitates 5’ strand degradation, preventing RAD51 filament assembly (Figures I4.V and I4.VI, left side) (Nimonkar, Genschel and Introduction ! 16 Kinoshita, 2011; Zhou, Pourmal and Pavletich, 2015). Recent studies suggest that the activity of BLM is enhanced by the formation of the BLM-TOPOIIIa-RMI1-RMI2 (BTRR) complex (Wu et al., 2001; Sturzenegger et al., 2014). In addition, WRN, a BLM paralog, has been described to act epistatically with DNA2 to promote extensive resection (Pinto et al., 2016). CtIP also interacts with BLM and stimulates its helicase activity and the nuclease activity of DNA2 (Daley et al., 2017). Moreover, MRN recruits BLM to the break, increasing the effectiveness of this helicase (Nimonkar, Genschel and Kinoshita, 2011). Finally, and as mentioned in the previous section, RPA coats ssDNA resulting from resection and this intermediate serves as substrate for the formation of RAD51 filament and the activation of the ATR-mediated checkpoint, which slows or arrests cell cycle progression in order to repair the DNA damage properly (Cejka, 2015). Due to its importance, it is not surprising that DNA end resection is very regulated in order to ensure faithful DSB repair. Several factors modulate DNA end resection, including the chromatin status and the complexity of the break (Peterson et al., 2013). However, one of the key and more understood points of regulation is cell cycle progression. Indeed, resection is restricted to S and G2 by CDKs, when the sister chromatid is present. As explained before, the interaction between CtIP and BRCA1 is controlled by a CDK-mediated phosphorylation of CtIP at serine 327 in S/G2 cells (Yu and Chen, 2004; Cruz- García, López-Saavedra and Huertas, Figure I4. DNA end resection. When a DSB (yellow lighting bolt) is produced (I), the broken DNA molecule is recognized by MRN. Then, the complex CtIP-BRCA1 promotes MRN endonuclease activity (II) generating a nick on the 5’ strands and initiating a shortrange resection of the DNA in 3’-5’ direction (III). EXD2 helps during this process (IV). After so, the long-range resection is performed by two independent pathways: one mediated by EXO1 (right) and the other that involves the action of DNA2-BLM complex (left) (V). Finally, the extensive ssDNA tails generated after this process are immediately coated by RPA complex (red circle) for protection (VI). ! Introduction ! 17 2014). Additionally, phosphorylation of NBS1 at serine 432 (Falck et al., 2012) and CtIP at threonine 847 mediated by S/G2 specific CDKs are key events in the initiation of DNA end resection (Huertas and Jackson, 2009). Moreover, after resection initiation, CDK-mediated phosphorylation of CtIP at serine 276 and threonine 315 promotes its binding to the prolyl isomerase PIN (Steger et al., 2013). Other CDKs controls the resection length phosphorylating EXO1 and DNA2 to recruit them (Lorenza P. Ferretti, Lafranchi and Sartori, 2013) Another important regulatory point in the decision between NHEJ versus HR to repair DSBs, is the competition between 53BP1-RIF1, which promotes NHEJ, and BRCA1-CtIP, that initiates DNA end resection (Chapman et al., 2013; Escribano-Diaz et al., 2013; Feng et al., 2013; Zimmermann et al., 2013; Daley and Sung, 2014; Isono et al., 2017). 53BP1 is phosphorylated by ATM to facilitate the recruitment of other proteins that regulate positively NHEJ, like RIF1 and PAX Interacting Protein (PAXIP1 or PTIP) (Harding, Coackley and Bristow, 2011; Chapman et al., 2013). Both prevents DNA end resection, although acts in different ways. On the one hand, PTIP recruits Artemis to DNA damage to promotes NHEJ pathways as mentioned in section 3.1 (Figure I2) (Isono et al., 2017). On the other hand, RIF1 recruits Mitotic Arrest Deficient 2 Like 2 (MAD2L2 or REV7), which also inhibits DNA end resection (Boersma et al., 2015). Recently, it has been published that the complex shieldin formed by the proteins 53BP1-RIF1-REV7-SHLD1-SHLD2-SHLD3 localizes at the end of the breaks to mediate 53BP1-dependent DNA repair. Thus, the shieldin binds to ssDNA and occludes the access to resection nucleases (Mirman et al., 2018; Noordermeer et al., 2018; Setiaputra and Durocher, 2019). Additionally, shieldin recruits CST (CTC1– STN1–TEN1) a complex similar to RPA and also polymerase-α (Polα)–primase to DNA damage sites in a 53BP1-dependent manner for the purpose of filling in resected DSBs (Mirman et al., 2018). Moreover, RIF1 is an antagonist of BRCA1. Although it is unknow how RIF1 is able to oppose BRCA1 (Boersma et al., 2015), it has been suggested that it is not due to a simple competition role, but maybe RIF1 alters BRCA1 ability to recognize ubiquitinated chromatin (Escribano-Diaz et al., 2013). This antagonism contributes to the cell cycle regulation of resection. Indeed, CtIP and BRCA1 remove 53BP1 and its partner RIF1 from DSBs in S and G2, to facilitate processing of DNA ends, and switching in that way the repair to HR (Figure I5). Additionally, it has been published that BRCA1 promotes dephosphorylation of 53BP1 Introduction ! 18 perturbing RIF1-53BP1 complex interaction (Isono et al., 2017). Indeed, the accumulation of RIF1 is G1-restricted and is suppressed by BRCA1-CtIP in S/G2. Thus, although it is still unclear how the balance between both DSB repair machineries is achieved, CtIP has emerged as a critical factor for promoting DNA end resection and, therefore, for regulation of the DSB repair pathway choice (Huertas and Jackson, 2009; You and Bailis, 2010; Polato et al., 2014). 4. CtBP-interacting protein (CtIP) ! CtIP owes its name due to its interaction with the transcriptional corepressor C- terminal Binding Protein (CtBP) which binds to a PLDLS motif in CtIP located between residues 490 and 494 (Schaeper et al., 1998) (Figure I6). Additionally, it is also known as Retinoblastoma Binding Protein 8 (RBBP8) for its interaction with retinoblastoma (Rb) in the residues 153-156 (Figure I6) although this domain is not essential for that interaction (Fusco, Reymond and Zervos, 1998). CtIP gene is localized at chromosome 18q11.2 and encodes for a protein of 897 amino acids. Multiples studies have reported that CtIP is a major regulator of DNA metabolism, affecting DNA repair, cell cycle progression, checkpoint activation, replication and transcription (Liu and Lee, 2006; Wu and W.-H. Lee, 2006; Duquette et al., 2012; Moiola et al., 2012; Makharashvili and Paull, 2015). To perform all these functions, CtIP cooperates with multiples factors through direct interactions of specific domains (Figure I6). The domains needed for the interaction with the MRN complex are located in both the carboxyl and amino terminus (22-45 residues of the amino terminus and the last 108 residues of the carboxy-terminal tail) (Yuan and Chen, 2009). Moreover, in the carboxyl region it is located the domain (840-842 residues) needed for its interaction with Kelch-like protein 15 (KLHL15) (Figure I6), which is a substratespecific adaptor for Cullin-based E3 ubiquitin ligases (CUL3). KLHL15 mediates CtIP ubiquitylation promoting its proteasomal degradation. In addition to KLHL15-CUL3, Figure I5. Schematic model of the antagonist connection between NHEJ and HR proteins. Relationship between 53BP1-RIF1 and BRCA1-CtIP and how nonhomologous end-joing (NHEJ) and homologous recombination (HR) pathways are connected. ! Introduction ! 19 the Peptidyl-prolyl cis-trans isomerase NIMA-interacting 1 (PIN1) and Anaphase Promoting Complex/Cyclosome-adherin 1 (APC/C-CDH1) negatively regulate CtIP protein stability and, consequently, limit DNA end resection (Ferretti et al., 2016). In this same region, the domain needed to interact with the Cell Cycle and Apoptosis Regulator 2 (CCAR2) (Figure I6) is located. CCAR2 is an antagonist of CtIP that inhibits recombination by limiting the initiation and the extent of DNA end resection (Lopez-Saavedra et al., 2016). The amino-terminal tail has an oligomerization area that includes a coiled-coil domain (45-160 residues) (Figure I6), which is required for CtIP homodimerization (Dubin et al., 2004). More recently, it has been reported that two CtIP homodimers can bind to form a tetramer and this homotetramer structure is required for CtIP recruitment to DNA damage and its function in DSB repair (Andres et al., 2015). Additionally, there is a DNA binding domain in the central region of CtIP, which is also is required for CtIP recruitment to site of the damage (Figure I6). In this region the residues by which CtIP joins PCNA are present (Figure I6). This interaction is required to supervise the ongoing DNA replication and maintain genome stability (Gu and Chen, 2009). Moreover, close to the oligomerization region a nuclease domain is found, including residues from 180 to 350 (Makharashvili et al., 2014). The same region includes serine 327, by which CtIP interacts with BRCA1 (Cruz-García, López- Saavedra and Huertas, 2014), and the threonine 315, required for the interaction with PIN1 (Steger et al., 2013) (Figure I6). Furthermore, carboxy-terminal domain includes the previously mentioned threonine 847, a target for CDKs (Huertas and Jackson, Figure I6. Diagram of human CtIP protein. The known CtIP domains are represented as colour regions: the oligomerization domain (yellow), which includes the region by which interacts with MRN and RB; the nuclease domain (pink), where it is the residue S327 that is phosphorylated to mediates the interaction with BRCA1; the DNA binding region (green), where it is the dominion of interaction with PCNA; and the conserved region with Saccharomyces cerevisiae homologous Sae2 (blue) where the regions by which CtIP interacts with KLHL15, MRN or CCAR2 are located. The different mapped post-translational modifications are shown in black (CDK-mediated phosphorylation sites); in blue (ATM and/or ATR-mediated phosphorylation sites) and in red (sumoylation site). Brackets marks the regions or specific amino acids required for the interaction of CtIP with other proteins. Introduction ! 20 2009), the threonine 859, which can be phosphorylated by ATR (Peterson et al., 2013) (Figure I6), and lysine 896, constitutively sumoylated by Chromobox 4 (CBX4) (Soria- Bretones et al., 2017). All those post-translational modifications are essential for resection activation. Thereby, it is clear that this carboxy-terminal domain is very important for the regulation of CtIP. Indeed, mutations in this region promotes two genetic disorders in humans, the Seckel and Jawad syndromes (Qvist et al., 2011). Functional orthologs of CtIP are found in Saccharomyces cerevisiae (Sae2) (Lengsfeld et al., 2007; Huertas et al., 2008) Schizosaccharomyces pombe (Ctip1) (Limbo et al., 2007; Akamatsu et al., 2008), Caenorhabditis elegans (COM-1) (Penkner et al., 2007), Xenopus laevis (xCtIP) (You et al., 2009), Arabidopsis thaliana (AtGR1)(Deveaux et al., 2000), or mouse (Ahmed, Tsai and Lee, 2010). These proteins vary greatly in length and have diverged enormously in sequence. However, there are small conserved regions in the carboxyl terminus (Figure I6) (You and Bailis, 2010). In addition to the previously mentioned phosphorylations (see section 3.2.1), CtIP is exposed to several other post-translational modifications, including, ubiquitylation or sumoylation, which control the interaction with its partners and its multiple roles in DNA metabolism. For example, BRCA1 catalyzes CtIP ubiquitylation, which is important for its recruitment to damaged chromatin and its role in checkpoint activation (Yu et al., 2006). Moreover, CtIP requires a constitutive sumoylation at lysine 896 to be recruited to DSBs and act on resection (Soria-Bretones et al., 2017). As mentioned before, Rb acts as a transcriptional repressor by binding to the transcription factor E2F, which are mainly involved in the initiation of S phase (Harlow, 1999). CtIP binds to Rb at late G1, releasing its repressor function and promoting G1/S transition (Wu and W. Lee, 2006), which demonstrates that CtIP is a regulator of the cell cycle progression. Additionally, CtIP promotes the expression of genes that initiate S phase, such as Cyclin D1 and also activates its own promoter owing to they are regulated by RB-E2F pathway (Liu and Lee, 2006). Indeed, the role of CtIP in regulating gene expression is confirmed by its interaction with other transcriptional factors like Ikaros (IKZF1), Tribbles Pseudokinase 3 (TRIB3) and LIM domain transcription factor 4 (LMO4) (Koipally and Georgopoulos, 2002; Sum et al., 2002; Xu et al., 2007). Therefore, CtIP contributes to DNA damagedependent cell cycle arrest in S and G2 phases promoting p21 transcription (Li et al., 1999; Liu et al., 2014) and upregulating the expression of Growth Arrest and DNA- Introduction ! 21 Damage-inducible protein (GADD45A) forming a complex with the factors previously mentioned CtBP, Rb and E2F1(Liao et al., 2010). Moreover, CtIP facilitates DNA replication fork progression during S phase by its binding to the Proliferating Cell Nuclear Antigen (PCNA) through a replication foci targeting sequence (RFTS) localized at the central region of CtIP (540-546aa) (Gu and Chen, 2009) (Figure I6). In agreement with this data, it has been published that CtIP participates in the detection and resolution of replication stress caused by interstrand cross-link (ICL) DNA damage. ICLs are covalent bridges formed between complementary strands of DNA that block the progression of replication and transcription (Duquette et al., 2012). The repair of ICL lesions implies that Fanconi Anemia protein FANCM binds to the ICL stalled fork and triggers a nucleolytic cleavage at both sides of the ICL, which will be repaired by nucleotide excision repair (NER) and translesion synthesis (TLS) mechanisms (Ciccia and Elledge, 2010). CtIP is recruited to the ICL lesion in a FANCM-dependent manner to facilitate the generation of a ssDNA fragment that will activate the following repair by HR. Indeed, CtIP is required for the accumulation of DNA repair factors like RPA, ATR, FANCD2, ATM autophosphorylated on serine-1981 (ATM-pS1981), and γH2AX at ICL to repair the produced damage (Duquette et al., 2012; Murina et al., 2014). Thus, CtIP is an early factor of ICL detection that induces DSB and promotes the repair of this DNA damage facilitating HR and strand invasion to complete the repair of the ICL lesion. Furthermore, CtIP messenger RNA (mRNA) is constitutively targeted for degradation by some microRNAs (miRNAs). This includes degradation by miR-335, which is repressed after DSB induction, when ATM downregulates miR-335 to increase CtIP transcript levels and promote DSB repair (Martin et al., 2013). Additionally, CtIP mRNA can be degraded by miR-19a and miR-19b, which are activated by p53 pathway (Huhn et al., 2015).!! In summary, CtIP is implicated in multiple cell metabolism processes. However, some of those functions are still poorly understood and should be investigated. 5. mRNA splicing ! Recently, a crosstalk between the DNA damage response (DDR) and the RNA metabolism at different levels has been discovered. Indeed, the number of factors that Introduction ! 22 participate in the DDR has expanded considerably in recent years to include many RNA-related proteins, notably splicing and alternative splicing factors (Beli et al., 2012; Abbas et al., 2014; Prados-Carvajal et al., 2018). Splicing is the process by which introns are removed from a precursor (pre)-mRNA and exons are joined to produce a mature mRNA. In mammals, nearly all genes transcribed by RNA polymerase II contain introns and are, consequently, spliced (Jurica and Moore, 2003; Shkreta and Chabot, 2015). Splicing is performed by the spliceosome, a nuclear macromolecular complex that contains five small nuclear ribonucleoproteins (snRNPs) (U1, U2, U4, U5 and U6) and more than 150 accessory proteins (Matera and Wang, 2014). Spliceosome assembly is initiated by the recognition of splice sites (Figures I7.I and I7.II). On the one hand, the U1 snRNP recognizes the 5' splice site (5’ss) of exon A. On the other hand, the U2 snRNP interacts with an internal region, known as the branch point, and the U2 Auxiliary Factor (U2AF) interacts with an internal region called polypyrimide and the 3' splice site (3’ss) of the exon B. The interaction of those snRNPs forms the pre-spliceosome (Figure I7.II). Then, a preassembled U4/U6/U5 trimeric-snRNP is recruited (Figure I7.III) and, with the help of auxiliary proteins, the U1 and U4 snRNPs are displaced to allow U6 and U2 snRNPs to form a catalytically competent core that signal the branch point for the first of two cleavage steps (Figure I7.IV). This first step produces a free upstream exon and a lariat intron covalently linked to the downstream exon. The second cleavage step leads to the excision of the lariat intron and the ligation of both exons (Figure 17.V) (Will and Luhrmann, 2011). Interestingly, the spliceosome-assembly efficiency is higher when it is linked to transcription. This is most likely due, at least partially, because the carboxyl-terminal domain (CTD) of RNA polymerase II recruits spliceosome components to facilitate their deposition on the nascent pre-mRNA (Bentley, 2014). Additionally, the interaction between splicing factors and chromatin plays an important role in splicing regulation (Fu and Ares, 2014). Histone marks and chromatin remodeling factors might alter the speed of transcription elongation and the recruitment of splicing regulators to these sites (Naftelberg et al., 2015; Nieto Moreno et al., 2015). ! Introduction ! 23 5.1. Alternative splicing Alternative splicing refers to the ability of some pre-mRNA to generate different mRNAs by spliced in different ways to. This process is a main contributor to transcriptomic and proteomic diversity. The production of alternative version of proteins is highly controlled. Indeed, specific protein variants have been associated with diseases like cancer (Tazi, Bakkour and Stamm, 2009; Zhang and Manley, 2013). Figure I7. Spliceosome assembly. The spliceosome assembly starts with the recognition of splice sites by U1, which interacts with the 5’splicing site (5’ ss) of the exon A and U2FA (U2 auxiliary factor) that binds to 3’splicing site (3’ss) of the exon B and an internal region called polypyrimide tract. Additionally U2 interacts with another internal region known as branch point (I). Then, U1 and U2 interact approaching both exons and forming the pre-spliceosome (II). U4, U5 and U6 are recruited to help them (III) and the U1 and U4 factors are displaced to get that U2 and U6 form a core that signal the branch point for the first of two cleavage steps (IV). As result of both cleavages, the lariat intron is excised and both exons are ligated (V). Results Results - Chapter one 35 Chapter one: The role of the SF3B complex in DNA repair 1. The splicing complex SF3B interacts with CtIP As mentioned in the introduction, CtIP is an important protein involved in DSB repair and DNA end resection, amid other multiples functions. Its different roles are regulated through distinct interactions with specific proteins. As part of her PhD in Dr. Huertas laboratory, Ana Lopez-Saavedra isolated nine new partners of CtIP (Prados- Carvajal et al. 2018). Among them, she found three out of 7 subunits of the splicing complex SF3B; namely SF3B1, SF3B2 and SF3B3. Thus, I set up to try to understand the connection between such splicing complex and DSB repair. Initially, and in order to validate this interaction, we performed several immunoprecipitations. First, we immunoprecipitated proteins with a GFP-trap using samples from U2OS cells constitutively expressing GFP-CtIP or GFP as a control and then immunoblotted for SF3B1 and SF3B2. Although there was some unspecific binding of those proteins, we observed an enrichment of both in cells bearing GFP-CtIP compared with control cells (Figure R1A). Then, we carried out the reciprocal experiment and immunoprecipitated SF3B2 (Figure R1B) or SF3B1 (Figure R1C) using specific antibodies (see Table M5) from GFP-CtIP and GFP cells. In agreement with the previous result, both subunits interacted with tagged CtIP (Figures R1B and R1C black triangle) and, to a lesser extent, endogenous CtIP (Figures R1B and R1C white triangle). As explained in the introduction, BRCA1 is a known CtIP interactor (Yu et al. 1998), Strikingly, we realized that BRCA1 not only interacts with CtIP, but also with SF3B1 (Figure R1C). Additionally, BRCA1 has been related with the splicing process (Onyango, Lee, and Stark 2017; Savage et al. 2014). Taking all these data into account, we conjectured that BRCA1 might mediate the interaction between CtIP and SF3B. In order to test this hypothesis, we repeated the immunoprecitation using an anti-SF3B1 antibody in BRCA1-depleted cells. Nonetheless, depletion of BRCA1 did not affect the interaction between CtIP and SF3B1 (Figure R1D). In addition, both CtIP and SF3B complex were previously related with RNA metabolism, transcription and splicing respectively (see introduction). Thus, we wondered whether this interaction might be mediated or stabilized by RNA. However, when we treated the samples with RNAse prior immunoprecipitation we observed that CtIP-SF3B interaction was neither abolished nor severely reduced (Figures R1D and R1E). Results - Chapter one 36 In order to identify the subcellular localization of the interaction between SF3B and CtIP, we carried out a proximity ligation assay (PLA) in the presence or absence of DNA damage (10 Gy irradiation). We observed that CtIP interacts with SF3B2 (Figure R2A) and SF3B3 (Figure R2B) inside the nucleus and this interaction increased upon the exposure of the cultures to DNA damage. To assess the specificity of the technique, we depleted the cells using shRNA against CtIP, SF3B2 or SF3B3 to confirm that the PLA signal was abolished (Figures R2A and R2B). As mentioned before in the introduction, some subunits of the SF3B complex suffer DNA damage-dependent post-translational modifications, including an ATM-dependent phosphorylation of SF3B2 at serine 289 (Matsuoka et al. 2007). To study whether such modification affects CtIP-SF3B2 interaction, we performed a PLA using cells depleted for endogenous SF3B2 and bearing either FLAG-SF3B2-S289A (a mutant version that Figure R1. SF3B interacts with CtIP and BRCA1. A. Protein extracts from U2OS cells bearing GFP-CtIP construct or GFP as a control were used for immunoprecipitation using anti-GFP resin. Immunoprecipitated samples (IP) and protein inputs (Input) were resolved in SDS-PAGE and blotted with antibodies that recognize the indicated proteins. B. Protein samples from U2OS cells harbouring GFP-CtIP or GFP as control were used for immunoprecipitation with an antibody against SF3B2 (IP), resolved in SDS-PAGE and blotted with the indicated antibodies. White triangle marks endogenous CtIP, black triangle the exogenous and tagged CtIP version. C. Same as (B), but using an antibody against SF3B1 to perform the immunoprecipitation (B1) or a control IgG. D. Same as (C), but prior immunoprecipitation, 2 µg RNAse A was added or not as designated in the figure in cells depleted or not for BRCA1 using a siRNA (siBRCA1), as indicated. siNT denotes an siRNA with no target in the human genome. E. Same as (B), but cells were treated or with RNAse A as depicted in the figure. Input IP SF3B1 SF3B2 GFP-CtIP GFP-CtIP GFP + - - + + - - + E A CB GFP-CtIP GFP CtIP SF3B2 IP + - - + + - - + Input siNT Input GFP-CtIP SF3B1 +RNAse IP BRCA1 IgG siBRCA1 -RNAse siNT siBRCA1 IP IP D CtIP SF3B1 IP SF3B1 IgG Input BRCA1 Input +RNAse GFP-CtIP SF3B2 -RNAse IP IgG IgG IgG Results - Chapter one 37 cannot be phosphorylated at Ser 289), FLAG or FLAG-SF3B2 as control. As expected, when a tagged SF3B2 was expressed, the interaction with CtIP increased compared to control cells (Figure R2C). Interestingly, the mutation in the residue Ser 289 did not affect its interaction with CtIP, neither before nor after challenging the cells with DNA damaging agents (Figure R2C). Nonetheless, we did observe that the increased Figure R2. CtIP and SF3B Proximity Ligation Assay (PLA). A. PLA signal was scored using antibodies against CtIP and SF3B2 in unperturbed cells (undamaged) or cells exposed to 10 Gy of ionizing radiation and incubated for 1 hour (IR). To establish the specificity of the signal, the experiment was repeated with depletion of either CtIP or SF3B2. shNT denotes an shRNA with no target in the human genome. The quantification (right) and depictive images (left) of one representative experiment out of four with similar results, is shown. Cell nuclei are shown in blue and red foci correspond to the sites where CtIP and SF3B2 are in close proximity (indicated with white arrows). The median PLA foci number per cell is represented with a red line. B. Same as (A), but for SF3B3-CtIP interaction. C. PLA signal, relative to control (cells harbouring FLAG plasmid), was scored in cells transfected with plasmids expressing FLAG, FLAG-SF3B2 or the mutant version FLAG-SF3B2-S289A. Those cells were then irradiated (right) or not (left) with 10 Gy and incubated for 1hour. The average and standard deviation of the medians of at least four independent experiments are shown. For each replica, at least 200 cells were measured. Statistical significant between samples was analysed by one-way ANOVA. One, two or three asterisk(s) denotes statistical significance at p<0.05, p<0.01 and p<0.001, respectively. D. Same as (C), but PLA signal was scored in cells exposed to 10 Gy of irradiation (right; IR), cells unchallenged with exogenous damage (left; undamaged) and treated with 10 μM ATM inhibitor (ATMi; white bars) or DMSO as control (black bars). The data was relativized to control cells (undamaged and treated with DMSO) and statistical significant was analysed by two-way ANOVA test (* p<0.05, ** p<0.01, *** p<0.001). E. Representative image of one of the three PLA experiment performed using SF3B2 and CtIP antibodies in irradiated cells (10 Gy) harbouring a GFP-MDC1 construct to visualize DNA breaks. White arrows and red dots indicate places in which CtIP and SF3B2 are in close proximity, while green dots mark GFP-MDC1 signal. Results - Chapter one 38 interaction upon DNA damage (Figures R2A, R2B and R2C) was dependent on ATM activation (Figure R2D). Despite this, the majority of CtIP-SF3B interactions did not occur at the sides of the DNA breaks (Figure R2E), as the PLA signal was mostly absent of damaged chromatin, visualized with GFP-MDC1 foci. Thus, we can conclude that CtIP and SF3B complex are in close proximity in the nucleous and this interaction is stimulated by the presence of DNA damage in an ATM-mediated manner, although it mostly takes places outside the DSBs. 2. The SF3B complex promotes DNA end resection Considering the interaction between CtIP and the SF3B complex, we decided to characterize the role of such complex in DNA resection. To do so, we depleted several subunits using short hairpin RNA (shRNA) (Table M3) or small interfering RNA (siRNA) (Table M2). To be sure that the observed phenotypes were not caused by offtarget effects, we also complemented SF3B2 depletion with a siRNA-resistant FLAGSF3B2 construct. Knockdown efficiency was routinely analysed by western blot, and a representative image can be seen in Figure R3. Figure R3. Depletion of SF3B subunits. A. Representative western blot showing the expression levels of SF3B2, SF3B3, CtIP and β-actin in cells transduced with lentiviruses harbouring the indicated shRNAs. B. Same as (A), but protein samples from cells depleted for the indicated proteins using siRNA against them were resolved in SDS-PAGE and blotted with antibodies against SF3B1 and β-actin as loading control. C. Western blot show endogenous or exogenous, siRNA-resistant, flag-tagged SF3B2 and α-tubulin expression from samples of cells transfected with the plasmids FLAG, FLAG-SF3B2, FLAG-SF3B2-S289D or FLAG-SF3B2-S289A and also depleted for SF3B2 (siSF3B2) or not (siNT). Results - Chapter one 39 As mentioned in the introduction, RPA is the protein complex that coats the ssDNA tails formed during DNA end-resection for protection. Hence, the accumulation of this complex at DSBs in response to exposure to DNA by immunofluorescence was used as Results - Chapter one 40 a proxy for resection. The damage-induced phosphorylation of H2AX (γH2AX) was used as a marker of broken chromatin. Thus, first we depleted cells using shRNAs against SF3B2 or SF3B3 (Table M3) and analysed RPA foci formation upon exposure of 10 Gy ionizing irradiation (Figure R4A) or 10 µM etoposide (Figure R4B). We used shRNAs against a non-target sequence (shNT) and against CtIP as positive and negative controls, respectively (Table M3). In all cases, and in agreement with a role of the SF3B complex in resection, we observed that the depletion of these SF3B subunits caused a reduction in the percentage of cells positive for RPA (Figures R4A and R4B), although the effect was stronger if the DNA damage was caused by irradiation. Interestingly, simultaneous depletion of both SF3B2 and SF3B3 showed an epistatic effect in DNA resection, indicating they are in the same genetic pathway (Figure R4A). Moreover, the subunit SF3B1 was also involved in DNA resection, because its depletion with a siRNA also caused a decrement in RPA foci formation upon cell irradiation (Figure R4C). As expected, ectopic expression of SF3B2 was able to fully complement the resection impairment (Figure R4D). In order to understand better the role of SF3B2 in DNA resection, we analysed whether the phosphorylation of SF3B2 in serine 289 by ATM affects this process. We expressed siRNA-resistant mutant versions of the protein: FLAG-SF3B2-S289A, that cannot be phosphorylated and FLAG-SF3B2-S289D, in which the acidic residue might mimic its constitutive phosphorylation. In both cases, the mutants did not completely rescue the resection defect upon depletion of endogenous SF3B2 (Figure R4D), showing the importance of such serine in DNA resection. The fact that FLAG-SF3B2-S289D did not rescue the phenotype might, however, been interpreted in several ways: either constitutive phosphorylation of the protein also Figure R4. SF3B depletion impairs DNA end resection. A. Percentage of RPA foci positive cells detected by immunofluorescence in cells exposed to 10 Gy of ionizing radiation and incubated for 1 hour. Those cells were also depleted for SF3B2, SF3B3 or both using shRNAs. A non-target shRNA (shNT) and an shRNA against CtIP were used as negative and positive controls, respectively. In the right side cells from a representative experiment are shown. The graph (left) represents the average and standard deviation of at least four independent experiments. For each replicate, at least 200 cells were measured. One-way ANOVA was used to asses statistical significance (* p<0.05, ** p<0.01, *** p<0.001, n.s not significant). B. Same as (A), but cells were treated with 10 μM etoposide for 1 hour. C. Same as (A), but cells were depleted for SF3B1 using and siRNA against it. siNT denotes a siRNA with no target in the human genome. A Student’s t-test was used to denote statistical significance (** p<0.01). D. Same as (A), but cells depleted for SF3B2 and expressing the indicated versions of siRNA-resistant tagged SF3B2. E. Cell cycle distribution of cells upon transduction of lentivirus harboring the indicated shRNAs. The average and standard deviation of at least four independent experiments are plotted. Per each replicate, 10.000 cells were analysed using BD FACSAriaTM and FACSDiva v5.0.3 software. F. Same as (E), but cells were transfected with the indicated siRNAs. G. Same as (F), but cells were also transfected with the indicated plasmid to express the exogenous SF3B2 versions. Results - Chapter one 41 impairs resection or the acidic aminoacid fails to mimic the phosphorylation or the actual presence of a serine at position 289 is the key issue. As mentioned in the introduction, cell cycle is a major regulator of DNA resection. Hence, we verified that the effect of SF3B in resection was not due to a change in the cell cycle distribution (Figures R4E, R4F and R4G). There are two non-exclusive ways to influence DNA end resection: by affecting the number of cells that initiate resection, or by modulating the length of DNA resected. RPA foci formation is a good measurement of the first one, but it is not sensitive enough to estimate the last one. Hence, to test whether some subunits of SF3B complex also affects the length of resected DNA, we took advantage of a high-resolution technique that analyses DNA resection in single-molecules in vivo (Single-Molecule Analysis of Resection Tracks; SMART) (Huertas and Cruz-Garcia 2018). See Figure R5A for examples of DNA fibers obtained with this technique upon depletion of some SF3B subunits or CtIP as control. We confirmed the depletion of SF3B2 or SF3B3 not only decreases the number of cells that resect their DNA, but also mildly limits the extension of such resected DNA upon irradiation (Figure R5B) and etoposide (Figure R5C). Figure R5. SF3B complex also promotes the extension of DNA end resection. A. Representative images of ssDNA fibers obtained with SMART in the different samples. B. Length of resected DNA measured in individual DNA molecules using the SMART protocol in cells infected with lentiviruses bearing the indicated shRNAs, exposed to 10 Gy of ionizing radiation and collected one hour after irradiation. The plot shows the average and standard deviation of the medians of at least four independent experiments. For each replica, at least 300 individual ssDNA fibers were measured. The data were normalized to shNT. One-way ANOVA was performed to analyse statistical significance of the differences (* p<0.05, ** p<0.01) C. Same as (B), but cells were treated with 10 µM etoposide (Vp16) for 1 hour to induce DNA damage. IR (10 Gy) shNT shSF3B2 shSF3B3 shCtIP 0 10 20 30 * ** * Resection length (um) shNT shSF3B2 shSF3B3 shCtIP 0 20 40 60 80 100* * ** Vp16 (10 mM) Resection length (normalized to control) CB AshNT shSF3B2 shSF3B3 shCtIP Results - Chapter one 42 3. SF3B is involved in homologous recombination As efficient resection required the SF3B complex, we wondered whether downregulation of some subunits of the SF3B complex also reduced recombination. We took advantage of a family of reporter systems that score the number of cells able to repair DSB either by HR or NHEJ. First, we used U2OS cells expressing one single copy of the DR-GFP reporter (Sartori et al. 2007) that measures gene conversion by the formation of an active GFP gene upon induction of a DSB by the nuclease I-SceI, after depletion of several subunits of the SF3B complex using shRNAs. In agreement with a resection impairment, the absence of any of these proteins reduced classical recombination (Figure R6A). To extend our results, we transfected the same shRNAs into cells bearing the SA-GFP reporter, which allows quantification of another type of recombination, single strand annealing (SSA; see introduction section 3.2) as the Figure R6. SF3B2 and SF3B3 stimulate homologous recombination and impair NHEJ. A. Cells harbouring a single copy of the DR-GFP reporter were used to measure short-tract gene conversion for DSBs repair upon depletion of the indicated genes by shRNAs. A scheme of the reporter is shown on the top side. The efficiency of homologous recombination was calculated as the percentage of GFP positive cells. To facilitate the comparison between experiments, this percentage was normalized to control cells (shNT). The mean and standard deviation of at least four independent experiments are plotted on the bottom side. Statistical significance was analysed by one-way ANOVA (* p<0.05, ** p<0.01, and *** p< 0.001). B. Same as (A) but using cells harbouring a single copy of the SA-GFP reporter to measure single strand annealing efficiency. C. Same as (A), but cells expressing one copy of EJ5-GFP reporter were used to measure non-homologous end joining efficiency. Results - Chapter one 43 percentage of GFP positive cells after I-SceI-induced DSB formation. As seen in Figure R6B, in the absence of any of the subunits of SF3B complex, SSA was also reduced (Figure R6B). Thus, we concluded that SF3B complex has a role in all types of homologous recombination. Additionally, we studied the role of this complex in NHEJ using the EJ5-GFP reporter (Bennardo et al., 2008), which measures NHEJ as the number of GFP positive cells upon the appearance of an I-SceI-induced DSB. As seen in Figure R6C, downregulation of SF3B subunits, like CtIP, promotes NHEJ. In order to confirm that all subunits of SF3B complex have the same role in DNA repair, we also analysed these systems using a siRNA against SF3B1. The absence of this protein also caused an impairment of homologous recombination measured by DR-GFP reporter (Figure R7A). However, NHEJ that was measured with the EJ5- GFP reporter, was not significantly changed upon SF3B1 downregulation (Figure R7B), suggesting that maybe the main effect of the complex is through HR. Finally, to be sure that the observed phenotypes were not due to a non-target effect, we also depleted SF3B2 using a siRNA in stable cells expressing the tagged, siRNA-resistant, versions and bearing one copy of the DR-GFP or EJ5-GFP reporter. As expected, ectopic expression of FLAG-SF3B2 rescued the effect of SF3B2 downregulation for HR and NHEJ (Figures R8A and R8B). Nevertheless, the SF3B2 phosphomutant (FLAG-SF3B2-S289A) was still compromised for HR (Figure R8A), suggesting the role of SF3B2 in DNA repair is dependent on its phosphorylation by ATM. Taking all these findings together, we can conclude that SF3B complex is involved in the repair of DSBs, mainly by homologous recombination. Figure R7. SF3B1 depletion impairs homologous recombination. A. HR was measured as indicated in Figure R6A in cells harbouring one copy of DR-GFP reporter and transfected with the indicated siRNA. The plot shows the mean and standard deviation of at least four independent experiments. A Student’s t-test was performed (* p<0.05 and ** p<0.01). B. Same as (A), but cells bearing a single copy of EJ5-GFP to measure NHEJ. Results - Chapter one 50 (Figure R13A), but it was only significant upon SF3B3 downregulation. However, when we analysed 53BP1 foci formation, we realized that SF3B did not affect them (Figure R13B), supporting the previous observation (see section 3) that this complex is involved in DNA repair mainly by affecting HR. In addition, it suggests that SF3B might play some roles both in the pro- and antirecombination branch of the DDR, thus limiting NHEJ even in the absence of HR upon its downregulation. These results made us ponder a possible role of SF3B in checkpoint activation. For this reason, we studied the protein levels of CHK1 (see section 2 introduction) and its phosphorylated form (pCHK1) by western blot in cells downregulated for SF3B2 or SF3B3 (Figure R14A). First, we observed that SF3B2 depletion already affects unphosphorylated CHK1 levels (Figures R14A and R14B). In order to study the efficiency of CHK1 phosphorylation, pCHK1 levels were normalized to CHK1. Interestingly, only SF3B3 depletion rendered a significant reduction compared to the Figure R14. SF3B affects CHK1 levels and phosphorylation. A. Western blot of a representative experiment of protein extracts from U2OS cells transduced with lentivirus harbouring shRNAs against the indicated subunits of SF3B complex. The membrane was blotted with the depicted antibodies. B. CHK1 protein level quantification of the representative western blot (A) normalized to control cells. C. The plot shows the pCHK1/CHK1 ratio relativized to control cells from at least three independent experiments. Statistical significance between samples was analysed by one-way ANOVA (* p<0.05). D. Representative western blot of proteins extracts from cells expressing FLAG and siRNA resistant tagged SF3B2 and transfected with siRNA against non-target or endogenous SF3B2. E. Quantification of protein of the representative western blot (D) represented as ratio between pCHK1 and CHK1 using cells transfected with siRNA against SF3B2 (siSF3B2) or control cells (siNT) and expressing FLAG or FLAG-SF3B2. Results - Chapter one 51 control (Figure R14C). To clarify these results, we studied pCHK1 and CHK1 protein levels in cells depleted for SF3B2 and expressing FLAG-SF3B2 or FLAG as a control. As shown Figure R14D, the amount of pCHK1 and CHK1 strongly decreased in cells transfected with siRNA against SF3B2, but the ectopic expression of FLAG-SF3B2 rescued the phenotype. Indeed, the ratio between pCHK1 and CHK1 was strongly affected by SF3B2 downregulation with siRNA and the expression of exogenous SF3B2 only rescued the phenotype partially (Figure R14E). Thus, those data indicate that SF3B also affects proteins involved in DDR, like CHK1. 6. The SF3B complex controls CtIP recruitment to DSBs. Given the fact that a single mutation in the ATM-mediated phosphorylation site S289 in SF3B2 mimics the resection and homologous recombination phenotypes of SF3B2 depletion without affecting CtIP protein levels, we hypothesized that SF3B might play additional roles in DNA resection by affecting the functionality of CtIP in response to DNA damage. To test that, we studied CtIP recruitment to a specific DNA double strand break using a previously published reporter (Lemaître et al, 2014). On these cells, a single DSB can be created by the induction of I-SceI enzyme with the addition of doxycycline. This I-SceI cleavage site is flanked by tandem repeats of the lac operator (lacO), allowing visualization of the locus by Fluorescence In Situ Hybridization (FISH) using a specific probe, independently of the expression of I-SceI (Figures R15A, R15B and R15C). We confirmed DSB induction after doxycycline treatment by the detection of γH2AX at the lacO array (Figures R15B (red and green dots, respectively) and R15C). No differences in DSB appearance at the array were evident in cells downregulated for SF3B2 compared with control upon I-SceI induction, although we Figure R15. SF3B controls CtIP recruitment to DSBs. A. Schematic representation of the experimental system used to measure protein recruitment to DSBs in U2OS19ptight13. Cells were treated with 1 μg/ml doxycyline (+Dox) to create a single DSB by I-SceI or mock treated (–Dox). Fluorescence in situ hybridization (FISH) of the lacO tandem repeats located next to I-SceI cleavage site allows its visualization. B. Immuno-FISH representative confocal images of the lacO array in green and γH2AX in red in U2OS19ptight13. White arrows indicate the position of the lacO array. C. Quantification of the percentage of colocalization of the lacO array with γH2AX in cells treated as mentioned in (A) depleted or not for SF3B2, as indicated. The plot represents the average and SEM of three independent experiments. In each experiment, 100 cells were scored. Two-way ANOVA analysis was performed to compare the samples (* p<0.05, ** p<0.01, *** p<0.001). D. Same as (B), but using an antibody against CtIP instead of γH2AX. E. Same as (C), but showing the colocalization of the lacO array with CtIP. F. Same as (C) but using U2OS19ptight13 cells transfected with the indicated constructs. G. Same as (F), but showing colocalization of the lacO array with CtIP. Results - Chapter one 52 observed a background accumulation of DNA damage elsewhere upon SF3B2 depletion (Figures R15B and R15C). Strikingly, cells depleted for SF3B2 completely failed in the recruitment of CtIP to DSB, suggesting that not even the remaining protein is recruited to damaged chromatin (Figure R15D and R15E). To confirm this phenotype, FLAGSF3B2 was expressed to rescue the defect on CtIP recruitment (Figures R15F and Results - Chapter one 53 R15G). More importantly, the phosphomutant FLAG-SF3B2-S289A was unable to accumulate CtIP in damaged chromatin (Figure R15G), although DSB appeared normally (Figure R15F) and CtIP levels are not reduced in such genetic background (Figure R9G). Taking all these data together, we hypothesized that SF3B2 not only affects CtIP expression, but also has an additional role regulating CtIP recruitment or retention to DNA breaks, which is dependent on SF3B2 phosphorylation by ATM. 7. CtIP overexpression rescues SF3B depletion phenotype in DNA resection In order to study the contribution of CtIP in the phenotypes associated with SF3B downregulation, we analysed resection combining CtIP overexpression and SF3B depletion. First, we studied RPA foci formation. As seen in Figure R16A, the expression of a CtIP exogenous version from cDNA totally rescued the phenotype of SF3B2 or SF3B3 in resection. Performing SMART, we also confirmed that the defect in resection length was reverted with the CtIP overexpression (Figure R16B). In addition, we confirmed that CtIP and the SF3B complex work in the same genetic pathway, because the depletion of SF3B2 or SF3B3 had no further effect on cells already downregulated for CtIP (Figure R17). Thus, those data suggested that the resection defect observed upon SF3B depletion was dependent on CtIP. Figure R16. Exogenous CtIP expression reverts SF3B phenotype in resection. A. Cells overexpressing a GFP-tagged version of CtIP or the GFP-empty vector were infected with shRNAs against the indicated subunits of the SF3B complex, CtIP or a control sequence. Resection was calculated as the percentage of RPA foci-positive cells. The average and standard deviation of four experiments are shown. Statistical significance was measured with one-way ANOVA test (* p<0.05, ** p<0.01, *** p<0.001). B. Same as (A), but the resection was measured by SMART. The plot shows the fiber length normalized to control cells (cells expressing GFP and transduced with shNT). Results - Chapter one 54 After that, we wondered whether the role of SF3B in DNA repair is also dependent exclusively on CtIP. For this reason, we overexpressed CtIP in cells harboring DR-GFP reporter in order to measure HR. We used a FLAG-tagged CtIP instead of GFP-CtIP to avoid interference between the fluorescently tagged protein and the GFP expression that result of repair on this reporter system. We saw that stable cells expressing FLAG-CtIP rescued CtIP depletion, but, contrary to resection, it failed to suppress the recombination phenotype observed in cells downregulated for SF3B2 (Figure R18A). To be sure that CtIP did not rescued the defect of SF3B in HR, we studied SSA pathway. We depleted SF3B2 in stable cells expressing a single copy of SA-GFP reporter and exogenous CtIP. As seen in Figure R18B, FLAG-CtIP reverted the phenotype resulting of the CtIP depletion, but not SF3B2 downregulation. In addition, we analysed CHK1 and pCHK1 protein levels in cells depleted for SF3B2 and expressing exogenous CtIP. We observed that the expression of GFP-CtIP was unable to completely rescue the phenotype of SF3B2 depletion in DDR (Figure R19). Finally, we studied the survival upon exposure to DNA damage of cells depleted for Figure R18. Exogenous CtIP does not revert SF3B2 phenotype in homologous recombination. A. U2OS cells bearing a single copy of DR-GFP construct and expressing FLAG-CtIP or an empty vector (FLAG) were transfected with the indicated siRNAs. The homologous recombination (HR) efficiency for repairing I-SceI-induced DSBs was calculated as the percentage of GFP positive cells and normalized with control cells (siNT). The mean and standard deviation of at least four independent experiments are plotted. Statistical significance was analysed by two-way ANOVA (* p<0.05, ** p<0.01, and *** p< 0.001). B. Same as (A), but cells harbouring SA-GFP reporter to measure Single Strand Annealing (SSA) efficiency. Figure R17. CtIP and SF3B work in the same genetic pathway. Percentage of RPA foci formation in cells depleted for the indicated proteins. The average and standard deviation of four experiments are shown. Statistical significance was analysed by two-way ANOVA (* p<0.05 and ** p<0.01). Results - Chapter one 55 SF3B2. To do so, we treated them with etoposide and we studied mitochondrial activity performing a MTT assay. Interestingly, we observed that they were hyper-sensitive to the DNA damaging agent etoposide, and such effect was not rescued when GFP-CtIP was expressed (Figure R20). To confirm this result, we also studied cell growth in XCELLigance RTCA (see material and methods section 1.5.3 for details) in the presence of the drug. In agreement with the MTT assay, we observed that the depletion of SF3B2 caused sensitivity to etoposide, as the growth rate was halved, and the ectopic expression of FLAG-SF3B2 reverted such phenotype (Figure R21A). In addition, we depleted CtIP and SF3B2 at the same time and we observed that the cell growth decreased and the expression of exogenous SF3B2 reverted such phenotype (Figure R21B). Figure R19. Exogenous CtIP does not revert SF3B2 phenotype in the DDR. A representative western blot of protein extracts from cells expressing GFP-CtIP or empty vector (GFP), transfected with the indicated siRNAs and blotted with the depicted antibodies (left side). Quantification of protein abundance (right side) is shown in the plot as the ratio between pCHK1 and CHK1 relativized to control cells (GFP cells transfected with siNT). Figure R20. Exogenous CtIP does not rescue SF3B2 sensitive to etoposide. Cells bearing GFP- CtIP or GFP were transfected with the indicated siRNAs and treated for 48 hours with 0.5 μM etoposide or DMSO. Cell survival was calculated using a MTT assay, as described in section 1.5.2 of material and methods, by normalizing the MTT data obtained in the presence of etoposide with the controls exposed to DMSO and expressed as percentage. The average and standard deviation of at least three independent experiments are shown. Statistical differences were assessed by two-way ANOVA (* p<0.05 and ** p<0.01) Results - Chapter one 56 All these data suggest that the role of SF3B complex in DNA repair is not only through CtIP, but has additional contributions to HR and the cellular response to genotoxic agents. Figure R21. Cells downregulated for SF3B2 are sensitive to etoposide. A. Cells bearing siRNA- resistant FLAG-SF3B2 or empty vector (FLAG) and transfected with siRNA against SF3B2 were grown in XCELLigance RTCA (see material and methods section 1.5.3 for details) in the presence of 10 µM etoposide or DMSO as a control. The growth rate, measured as the slope of the curve, in each condition was calculated. The rate in the presence of etoposide was normalized with the slope obtained in cells growing in the presence of DMSO. A Student’s t-test was used to denotes statistical significance (* p<0.05). B. Same as (A), but cells were transfected with two siRNA at the same time as indicated in the plot. Statistical significance between samples was analysed by two-way ANOVA (* p<0.05 and ** p<0.01) Results - Chapter two ! 57 Chapter two: A crosstalk between the DNA end resection machinery and DNA-damage induced RNA alternative splicing 8. SF3B controls mRNA expression of many genes ! As mentioned in section 4, we discovered that SF3B controls the levels of several specific proteins involved in DNA resection, namely CtIP and BRCA1. Due to the role of SF3B in splicing, we wondered whether this control is at the level of RNA processing and whether it reacts to DNA damage. To test it, we decided to carry out a splicing microarray Transcriptome Arrays HTA & MTA using both damaged (6 hours after 10 Gy of irradiation) or untreated cells that were depleted or not for SF3B2 (see materials and methods for details). Such array allows the genome-wide study of RNA expression and RNA splicing simultaneously. First, we focused on total RNA levels. We found that SF3B2 depletion leads to the specific upregulation of 52 genes solely in undamaged conditions when compared with control cells (Figure R22A). Moreover, 26 genes were exclusively overexpressed in SF3B2 downregulated cells upon exposure to DNA damage (Figure R22A). Additionally, mRNAs abundance from 27 genes was increased in cells with reduced SF3B2 levels in both damaged and undamaged cells (Figure R22A). A list of those genes could be found in Table R1. On the other hand, SF3B2 depletion also reduced expression of 97 genes (Figure R22B and Table R2). Only 10 of those genes were downregulated specifically in unperturbed conditions, 82 in cells that were exposed to Figure R22. The role of SF3B2 in gene expression in response to DNA damage. A. Venn diagram depicting the distribution of genes upregulated in SF3B2 depleted cells compared with control cells. The number of genes which expression changes exclusively in undamaged conditions (green; left), 6 hours after irradiation with 10 Gy (IR; pink; right) or in both conditions (yellow; centre) is indicated. B, Same as (A), but genes downregulated in cells transfected with an shRNA against SF3B2 compared with shRNA control. ! Results - Chapter two ! 58 ionizing radiation and 5 in both conditions (Figure R22B). SF3B2 controls mRNA expression of several genes involved in multiple functions, including DDR factors like THOC7 (see Table R1; (Rehwinkel et al., 2004)) or FANCD2 (see Table R2; (Nepal et al., 2017)). How such expression changes contribute to the observed deregulation in the DNA repair pathways (see chapter one) and/or in the general metabolism of the cell has still to be studied. Table R1: Genes upregulated upon SF3B2 depletion TRAM1 HAVCR2 THEMIS CCDC30 RP4-553F4.6 NME8 SPATA9 RRH ZNF442 KLHL41 EIF5 JAKMIP1 ZNF571 RP11-545I10.2 RP11-386I8.6 CHKA WDR87 GABRR3 TSPY5P RP11-473L15.3 PDE4D SLFN12 ANKRD45 ANKRD18DP CTD-2651B20.3 SYNE1 ! Genes upregulated upon depletion of SF3B2 on damaged and undamaged conditions BET1 CBWD7 GMFB NQO1 TULP3 C12orf39 CDO1 KDM3A RP11-111F5.3 TXNDC2 CBWD1 CGRRF1 MITD1 RSAD2 ZNF844 CBWD2 CTD-2023N9.3 MMP13 SPHAR CBWD3 FAM133B MTHFD2L SRD5A3-AS1 CBWD6 FLVCR1 NPM3 STARD4 Genes upregulated upon depletion of SF3B2 exclusively on undamaged conditions ABCA5 CDKN1B IKZF3 PDIK1L SRPRB AC087073.1 CREBRF IL18R1 PPARGC1A SUN3 ANTXR2 DNAJC27 IL5 RP11-159G9.5 THOC7 ATG12 DTD2 ITPR2 RP11-213G2.2 TP53INP1 C14orf37 ENOX1 KANSL2 RP11-261C10.3 TRPM7 C1orf168 FBXW7 KIAA1467 RP11-46C20.1 USP54 C1orf63 GBP1 MAGOH RP13-996F3.4 ZMAT3 C21orf91 GDAP1 METTL13 S100A8 ZNF669 C9orf156 GTF2E2 NCAPG2 SCO1 CCDC66 HEBP1 PARP16 SESN1 CD274 HSD17B13 PCNXL2 SLC25A27 Genes upregulated upon depletion of SF3B2 exclusively on damaged conditions Results - Chapter two ! 59 Table R2: Genes downregulated upon SF3B2 depletion 9. SF3B controls damaged-induced alternative RNA splicing In terms of mRNA splicing, we confirmed that both depletion of SF3B2 and DNA damage induction affect such RNA processing globally. We first calculated the splicing index of genes on all four conditions and compared them in silico. The results are summarized in Table R3. As shown, both in control and SF3B2 downregulated cells a similar number of genes, around 900, altered their splicing upon DNA damage induction (Table 3; compare columns 2 and 3). Moreover, those genes suffered a similar amount of changes in splicing events (see Table R3, alternative exons row). Moreover, the pattern of gain (+) or loss (-) of specific events of alternative splicing was similar in both cases (Table R4; compare columns 2 and 3). As expected, due to its role in mRNA processing, depletion of SF3B2 changed the pattern of splicing when compared with control conditions (Tables R3 and R4; column 4). Strikingly, the role of SF3B2 in splicing was more pronounced in cells after Genes downregulated upon depletion of SF3B2 on damaged and undamaged conditions CAV1 EXOSC3 KDELC2 NID1 ST3GAL5 Genes downregulated upon depletion of SF3B2 exclusively on undamaged conditions GRAP LINC00482 PRSS22 SPINK14 VPS37C KRT27 OR6T1 SNCAIP Genes downregulated upon depletion of SF3B2 exclusively on damaged conditions AHCYL1 CCNDBP1 INPP5B MOCOS PRLR AL603965.1 CDC45 ITGB3 MYH16 QRFPR ALDH1L1 CHORDC1 KCNH7 NCEH1 RCAN2 ALPL CLMP KCNMA1 NDUFS3 SF3B2 ANO1 CSF2 KIAA1279 NEK7 SGK196 ANXA8 CTC-436P18.3 KLHDC3 NMD3 SLC22A4 ANXA8L1 CYB5R2 KRT20 PALM3 SLCO2A1 ARHGAP31 E2F1 KRT6A PAPOLA SSTR2 ATG5 FAM49A KRT6C PDE9A ST6GALNAC3 BCL2A1 FANCD2 KRT75 PEPD THBS1 BECN1 GALNT1 LGI2 PIEZO2 TNFRSF9 C12orf50 GINS1 LRFN5 PIR TOMM34 C14orf182 HSPB11 LRRC15 PLA2G7 TPK1 CCL2 IARS MBNL2 PLOD2 TPPP3 CCL7 ICAM1 MMP9 PPP4R4 WAC-AS1 Results - Chapter two ! 66 Considering the role of CtIP in the response to DNA damage, we were particularly interested in discover those mRNA which splicing was affected simultaneously by both CtIP presence and the exposure to ionizing radiation; i.e those genes with specific mRNA variants in response to DNA damage that are CtIP dependent. Thus, we carried out another analysis in which we compared all conditions in pairs: control cells without DNA damage (siNT) or exposed to IR (siNT6h), or depleted for CtIP in unperturbed (siCtIP) or damaged cells (siCtIP6h) (Figure R26 and Table R9). As shown Figure R26 in yellow, the splicing of 423 genes was changed specifically in response to DNA damage exclusively in control cells. A similar number of genes whose mRNA splicing was modified (494 genes) were exclusive of cells depleted for CtIP in response to DNA damage (Figure R26; in red). Only 107 genes were altered due to CtIP absence in irradiated cells, whereas 205 did so in undamaged cells (Figure R26; in green and blue respectively). Table R9: Genes whose splicing is affected in response to DNA damage and CtIP depletion Comparison Irradiated versus non irradiated samples in control cells Irradiated versus non irradiated samples in CtIP depleted cells CtIP depleted versus control cells in untreated conditions CtIP depleted versus control cells in irradiated conditions Genes with splicing modified 730 794 296 167 Thus, we observed that the alternative splicing of 423 genes changed in response to the exposure to DNA damage only when CtIP was present. Hence, we reasoning that CtIP might mediate, directly or indirectly, such DNA damage-dependent alternative mRNA processing. In order to validate those observations, we tested the abundance of ! Figure R26. The role of CtIP in mRNA splicing. Venn diagram represents the number of genes whose alternative splicing change upon CtIP depletion compared to control cells with or without DNA damage (10 Gy irradiation and 6 hours of incubation). ! Results - Chapter two ! 67 several of those genes as described for SF3B2 using qPCR and variant-specific primers. Again, we used a common isoform of each mRNA, present in all conditions, to normalize the results. As predicted by the array, we saw a decreased in the levels of a RANBP1 alternative spliced form when cells were damaged (Figure R27A). Similar results were observed for a RNF25 isoform mRNA expression level (Figure R27B). We also analysed isoforms whose splicing event occurs between two exons (junction event). The mRNA expression levels of CEP192 and CEP63 were relativized to undamaged cells and the common isoform. In both cases, there was an increase in mRNA expression upon irradiation, although the increase of CEP192 mRNA was greater (Figure R27C). Hence, these results validated the data from the microarray. ! ! 12. CtIP controls mRNA splicing of PIF1 In order to study deeply the crosstalk between CtIP, DNA end resection and RNA splicing, we decided to focus on the altered splicing of PIF1, whose processing was affected by DNA damage in a CtIP-mediated fashion. PIF1 is a helicase with activity 5’-3’ polarity. In humans there are only one PIF1 gene, but it was known to produce two well studied different transcripts. A short transcript (2295nt) produces the longer protein isoform (707aa) called PIF1ß, which is located in the mitochondria. On the other hand, the longer transcript (2688nt) is translated into a smaller protein variant named PIF1α (641aa) that is localized in the nucleus (Futami, Shimamoto and Furuichi, 2007). Figure R27. CtIP controls the splicing of some genes. A. The plot shows the RNA expression of one RANBP1 isoform CtIP- and DNA damage dependent normalized to the expression of mRNA RANBP1 ubiquitous isoform that is always expressed. IR indicates samples that cells were irradiated with 10 Gy and incubated for 6 hours. Non-damaged samples are indicated as “NO IR”. The average and standard deviations of the medians of at least four independent experiments are shown. Statistical significance was analysed by Student’s t-test (** p<0.01). B. Same as (A), but studying the mRNA expression of a splicing variant of RNF25. C. Same as (A), but the expression of the indicated mRNAs was also relativized to undamaged mRNA expression. Deviation from the dotted red line, set as 1, indicated changes in the normal splicing of the indicated mRNAs. !! Results - Chapter two ! 68 The difference between both proteins is the presence of a mitochondrial localization domain in PIF1ß, which also lacks the signal to translocate into the nucleus. We focus on PIF1α as its role in resection was discovered and its study carried out in parallel in our laboratory (Jimeno et al., 2018). First, we selected four different splicing events on the PIF1 mRNA that were CtIP- and DNA damage-dependent (Figure R28). On the one hand, we studied the junction of exon 2 and exon 3 (Figure R28.I) and exon 3 and exon 4 (Figure R28.II). On the other hand, we test whether exon 9 was jointed with the exon 8 and/or exon 10 (Figures R28.III and R28.IV). All these optional events are combinatorial, so different species of mRNA coexist in each condition. We studied the Figure R28. Splicing events in PIF1 mRNA. The image shows four splicing events dependent on CtIP and DNA damage that occurs in PIF1 mRNA. In pink is represented the junction of exon 2 with exon 3 (I) and exon 3 with exon 4 (II). In blue the junction of exon 8 with exon 9 (III) and exon 9 with exon 10 (IV). The arrows indicate where the primers were designed to study the expression of the different exons by qPCR (see Table M4). ! Figure R29. CtIP controls PIF1 splicing. A. The graph shows the presence of exon 3 in PIF1 mRNA upon exposure to IR (black bars) or not (white bars) and depleted or not for CtIP using a siRNA (siCtIP) as indicated. The data were normalized to the expression of the exon in cells not exposed to irradiation (no IR) and transfected with the control siRNA (siNT). Samples were obtained from U2OS incubated for 6 hours after irradiation or mock treated. The plot shows the mean and standard deviation of at least four independent experiments. Two-way ANOVA test was performed and one, two or three asterisk(s) denotes statistical significance at p<0.05, p<0.01 and p<0.001, respectively. B. Same as (A), but studying the presence of exon 4. C. Same as (A), but studying the expression of exon 9. D. Same as (A), but studying the presence of exon 10. ! Results - Chapter two ! 69 abundance of the different alternative events by qPCR. Inclusion of exon 3 was only slightly increased upon irradiation or CtIP depletion, but the changes were not significant (Figure R29A). A similar, but in this case significant, change was observed for exon 4 inclusion (Figure R29B). Exon 9 was strongly increased in response to irradiation in control cells, and both its inclusion and DNA damage-accrue was completely dependent on CtIP presence (Figure R29C). Exon 10 presence in the mRNA analysis rendered no statistically significant changes in any conditions (Figure R29D). These results suggested that the clearest CtIP-dependent alternative events in response to exogenous damage occur in exon 4 and, more specially, in exon 9 of PIF1 gene. In order to determine which activity of CtIP is involved in the splicing of PIF1, we carried out several qPCR to analyse the presence on the mRNA of specific PIF1 exons, but in cells bearing different mutated versions of CtIP. We used GFP-CtIP as control, the resection defective GFP-CtIP-T847A (Huertas and Jackson, 2009) and GFP-CtIP- E894A (Soria-Bretones et al., 2017), and lastly GFP-CtIP-S327A that does not interact with BRCA1 and still resects albeit at a slower pace (Cruz-García, López-Saavedra and Huertas, 2014). The data were normalized to the control GFP-CtIP, set as 1. Thus, changes on splicing specific of each mutant are depicted as a deviation from 1. As seen in Figure R30, GFP-CtIP-T847A (black columns) caused a consistent change in all the analysed splicing events. In agreement with our previous results (Figure R29), the clearer change occurred in exon 9. This did not happen for all the splicing events when GFP-CtIP-S327A (grey columns) or GFP-CtIP- E894A (white columns) cells were used, albeit some specific changes might be affected by those residues. This suggested threonine 847, a target site for CDK phosphorylation (see Figure I6) is affecting splicing of, at least, PIF1. This is likely resection independent, as the E894A mutant, a Figure R30. CtIP residues important for the splicing of PIF1. Presence of the indicated PIF1 exons in cells bearing different mutated versions of CtIP. The data was normalized to undamaged cells and control (GFP-CtIP), set as 1. Deviation from the dotted red line, set as 1, indicated changes in the normal PIF1 splicing. ! Results - Chapter two ! 70 sumoylation defective CtIP that is equally impaired in resection, did not share such phenotype. Thus, specific PTMs might also be required for specific changes. Taking together these results, we decided to study the relevance of those CtIP- and DNA damage-dependent splicing events in DNA repair process in human cells. Hence, we created two different PIF1 splicing variant cDNA constructs. First, a PIF1 containing all exons (“total PIF1”; tPIF1), which correspond to the canonical PIF1α. In contrast, we also created a PIF1 variant artificial cDNA that lacks both exon 4 and exon 9, the two exons which inclusion was more dependent on CtIP and DNA damage. We named it “variant PIF1” (vPIF1). Both genes were expressed from pCDNA. In order to detect the expression of either variant in cells, we tagged both isoforms with GFP. An empty pCDNA plasmid was also used as control in our experiments. We transfected each plasmid (pCDNA, GFP-vPIF1 or GFP-tPIF1) into U2OS in order to study the effect of either isoform in human cells in response to DNA damage. Expression of the proteins coded by those variants is shown in Figure R31. Initially, we decided to leave the expression of the endogenous forms of PIF1 unperturbed, as in our splicing experiments we could observe a mixture of different splicing variant, including always the canonical PIF1α. First, we studied the timely appearance/disappearance of the phosphorylated version of histone H2AX (γH2AX) upon irradiation as a proxy of the efficiency of the DDR activation and DNA repair (Figure R32). In agreement with previously reported results, this DNA damage dependent phosphorylation peaks early (30 minutes after irradiation) in response to checkpoint activation and then decreased as the breaks are repaired. Strikingly, at later time points (6 hours), the same timeframe in which we have tested splicing differences, cells expressing the tPIF1 seemed to repair slightly and significantly faster than those expressing vPIF1, which does not include the CtIP- and damage-dependent PIF1 exon 4 and 9 (Figure R32A). Indeed, expression of such alternative cDNA slows repair when compared with pCDNA transfected cells, suggesting vPIF1 presence hampers repair. In those same cells, overexpression of vPIF1 increased the intensity of γH2AX foci at later Figure R31. PIF1 isoforms. A representative western blot of samples from cells expressing pCDNA as control and the different variant of PIF1 gene. Black arrows indicate the size of the protein specifically produced by each isoform. ! Results - Chapter two ! 71 time points analysed upon irradiation (Figure R32B). Nevertheless, in cells expressing tPIF1, we observed that the intensity per foci was more increased in the hours after DNA damage than when vPIF1 was expressed (Figure R32B). However, the reduction of the signal was slower in cells expressing vPIF1 and at later time points (4-6h) the intensity per foci was higher compared to other cells (Figure R32B). Thus, in both cases, it seems that the intensity and number of γH2AX foci suffer a delay in cells expressing vPIF1 at later time points. Additionally, the size of the foci was bigger in cell expressing vPIF1 (Figure R32C). Taking these data together, it is likely that expression of tPIF1 increases the DDR at early times points, while vPIF1 delays the repair at later time points. Taking into consideration this finding and the recently published fact that PIF1 is required for resection (Jimeno et al., 2018), we decided to study the foci formation of Figure R32. The expression of PIF1 isoforms modifies γH2AX foci kinetics. A. The plot shows the number of γH2AX foci per cell relativized to control studied at difference times after irradiation (2 Gy) as indicated. U2OS cells expressing PIF1 isoforms or pCDNA as control was used. At least 200 cells were analysed per sample using MetaMorph Image Analysis Software. The mean and standard deviation of at least four independent experiments is shown. B. Same as (A), but studying the γH2AX intensity per foci. C. Same as (A), but studying the γH2AX foci size. ! Figure R33. The expression of PIF1 variants does not alter foci of recombination proteins. A. The plot shows RPA foci formation 1 hour after irradiation (10 Gy) in cells expressing the difference PIF1 isoforms or control plasmid (pCDNA). The mean and standard deviation of at least four independent experiments are shown. Statistical significance was analysed by one-way ANOVA. B. Same as (A), but studying BRCA1 foci formation. C. Cell cycle distribution of cells expressing PIF1 variants or the control plasmid (pCDNA). At least four independent experiments were studied and per each replicate, 10.000 cells were analysed on BD FACSAriaTM using FACSDiva v5.0.3 software. ! Results - Chapter two ! 72 some proteins involved in homologous recombination in response to irradiation (10 Gy) in cells expressing either PIF1 construct. As shown Figure R33, the overexpression of either isoform caused no effect in DNA resection (Figure R33A) or BRCA1 foci formation (Figure R33B). As a control, we also analysed the cell cycle in those cells to test whether such overexpression caused any change in the progress of cell cycle (Figure R33C). Then, we studied whether either isoform might affect NHEJ. To do so, we analysed 53BP1 foci formation 1 hour after irradiation (10 Gy). We observed that the expression of vPIF1, which lacks the CtIP- and DNA damage-dependent exons, slightly decreased the number of foci per cell (Figure R34A) and the intensity of each foci compared to control (Figure R34B). However, the expression of tPIF1 only decreased mildly the intensity of 53BP1 foci (Figure R34B). The size of foci was not modified in any case, expressing either PIF1 construct (Figure R34C). Those data agreed with the idea that vPIF1 expression might block repair by NHEJ. To study such effect of PIF1 variants on 53BP1 foci formation in more detail, we analysed the kinetics of its appearance/disappearance (Figure R35). As for γH2AX, the number of foci peaked at 30 minutes after irradiation and slowly decreased with time. However, we could not see any difference comparing cells expressing different PIF1 variants or cells transfected with pCDNA (Figure R35A), excepting the delay observed in cells expressing vPIF1. Moreover, the decrease in the intensity of 53BP1 foci were observed at early time points in cells expressing the vPIF1 isoform, whereas those changes were more evident at later time points in cells expressing tPIF1 variant (Figure R35B). Finally, 53BP1 foci size was slightly reduced in cells expressing either PIF1 variant compared with pCDNA-bearing cells (Figure R35C). Figure R34. The expression of the different PIF1 isoforms alters 53BP1 foci formation. A. Cells exposed to 10 Gy of irradiation, incubated for 1 hour, and expressing the indicated plasmids are used to study 53BP1 foci number per cell. The plot shows the data relative to control cells (pCDNA). The average and standard deviation of at least four independent experiments is shown. At least, 200 cells were studied for each replicate using MetaMorph Image Analysis Software. Statistical significance between samples was analysed by one-way ANOVA (* p<0.05 and *** p<0.001). B. Same as (A), but for signal intensity of 53BP1 foci. C. Same as (A), but studying the size of 53BP1 foci. ! Results - Chapter two ! 73 An important interactor of 53BP1 in NHEJ is RIF1 (see introduction). Thus, we also analysed RIF1 foci formation (Figure R36). Interestingly, the effect of the PIF1 variants in RIF1 foci mirrored the effect on 53BP1, with a significant decrease in the foci number in cells overexpressing vPIF1 (Figure R36A). These findings suggested that the expression of a specific PIF1 variant, namely the one dependent on CtIP and DNA damage, has a mild effect on the DDR and NHEJ pathway. Those results were obtained in a context in which the ectopic PIF1 variants coexisted with endogenous PIF1, in order to resemble what happens normally in which several isoforms are present. However, endogenous PIF1 will still be subjected to a normal splicing regulation. In order to have a clearer, less physiological but more extreme Figure R35. PIF1 isoforms modify 53BP1 foci formation after DNA damage. A. The plot shows a kinetic of the number of 53BP1 foci relative to control at the different times after DNA damage (2 Gy irradiation) in cells transfected with the indicated PIF1 variants or pCDNA. Point zero indicates undamaged cells. For each replicate, at least 200 cells were analysed using MetaMorph Image Analysis Software. B. Same as (A), but studying the intensity of 53BP1 foci. C. Same as (A), but representing the size of 53BP1 foci. ! Figure R36. RIF1 foci are altered when different PIF1 isoform are expressed. A. The plot represents the number of RIF1 foci formation relative to control (pCDNA) 1 hour after irradiation (10 Gy) in cells expressing the different PIF1 variants. The average and standard deviation of at least four independent experiments is shown. For each replicate, at least 200 cells were analysed using MetaMorph Image Analysis Software. Statistical significance between samples was analysed by oneway ANOVA (* p<0.05). B. Same as (A) but showing RIF1 foci average intensity. C. Same as (A) but representing RIF1 foci size. ! Results - Chapter two ! 74 phenotype, we decided to combine depletion of PIF1 using a siRNA with the expression of the different constructs (see Table M3). Unfortunately, such siRNA also affected our ectopic PIF1 variants, thus causing a partial depletion of the protein (Figure R37). This was more evident when wild type PIF1 was expressed, so the vPIF1 was expressed slightly higher. In any case, both isoforms were heavily overexpressed in those conditions (see Figure R37, where the endogenous PIF1 cannot be seen). Albeit the conditions were not ideal, we set up to repeat some of the studies on the DDR carried above in these new experimental conditions. First, we analysed RPA foci formation. In all cases, and in agreement with the role of PIF1 in resection (Jimeno et al., 2018), we saw a decrease upon PIF1 downregulation (Figure R38A). Moreover, we did not observe any difference between isoforms, and both only suppressed the PIF1 depletion poorly. Thus, we concluded that the constructs we were using for PIF1 complementation, which were partially depleted with the siRNA, did not fully compensate the loss of endogenous PIF1. Most likely, this was due to an inefficient splicing or the lack of additional signals. After that, we decided to analyse other proteins involved in recombination like RAD51 and BRCA1. Interestingly, when BRCA1 foci formation was analysed, we observed a decrease in control cells downregulated for PIF1, a phenomenon that was complemented by either PIF1 variant (Figure R38B). Thus, both species of PIF1, even if expressed at low levels or not completely functional, were competent in BRCA1 foci formation. Moreover, those results suggested that the reduction in BRCA1 foci formation caused by PIF1 depletion was not exclusively linked with a reduction in DNA end resection. Finally, and as shown in Figure R38C, there was a decrease in RAD51 foci formation in cells depleted for PIF1 compared to non-target cells. Furthermore, tPIF1 expression was not able to complement the phenotype observed upon endogenous PIF1 depletion, as a clear reflection of the defective DNA end resection on those samples (compare Figure 38A and 38C). But strikingly, when vPIF1 Figure R37. The expression of the different PIF1 variants. A representative western blot of stable cell lines expressing the different PIF1 isoforms or control plasmid (pCDNA) and transfected with siRNA against endogenous PIF1 or control (siNT). Black arrows show the size of each PIF1 isoform. ! Results - Chapter two ! 75 isoform was expressed in the presence of endogenous PIF1, RAD51 foci was reduced and, in this case, this was not changed upon endogenous PIF1 depletion (Figure R38C). Those phenotypes were not due to a delay in cell cycle (Figure R38D). Taking together all this data, we hypothesized that PIF1 isoform dependent on CtIP and DNA damage might have a role in the last steps of homologous recombination. Based on our previous observations of a defect in the DDR and, more specifically, in the 53BP1 branch, we studied 53BP1 and RIF1 foci formation in cells depleted for endogenous PIF1 and expressing PIF1 isoforms (Figures R39 and R40 respectively). In Figure R38. PIF1 isoform expression in homologous recombination. A. Percentage of RPA foci positive cells detected by immunofluorescence in cells exposed to 10 Gy of ionizing radiation and incubated for 1 hour. Those cells expressing the different PIF1 variants were depleted for endogenous PIF1 (siPIF1) or not (siNT). The graph represents the average and standard deviations of at least four independent experiments. For each replicate, at least 200 cells were measured. One-way ANOVA was used to statistical analysed (* p<0.05, ** p<0.01, *** p<0.001, n.s not significant). B. Same as (A) but representing BRCA1 foci formation. C. Same as (A) but studying RAD51 foci formation. D. Cell cycle distribution of stable cells expressing PIF1 variants or control plasmid (pCDNA) and transfected with the indicated siRNAs. Per each replicate, 10.000 cells were analysed by flow cytometry using BD FACSAriaTM and FACSDiva v5.0.3 software.! Results - Chapter three ! ! 82 DMSO (Figure R45). Moreover, a similar increase was apparent in cells exposed to an ATM inhibitor (Figure R45), indicating this accrual was not mediated by this kinase. However, ATR inhibition decreased RNA editing damaged and undamaged cells, although a mild increase upon irradiation was still observed (Figure R45). As both CtIP and ATR are involved in radiation-induced RNA editing, and CtIP is required for ATR activation upon DSB appearance, we wondered whether those two proteins indeed collaborate for this additional role. Hence, we studied whether the phenotype was enhanced when both proteins were compromised. As shown in Figure R46, the inhibition of ATR decreased editing regardless if they are irradiated and/or depleted for CtIP. Thus, this finding confirms that ATR has an important role in RNA editing, both in the presence or absence of ectopic DNA damage. CtIP seems required for DNA damage-induction of the RNA modification, but only when ATR is present (Figure R46). 14. ADAR is involved in DNA resection As mentioned before, ADAR proteins were positive hits in the screening carried out in our laboratory (Lopez-Saavedra et al., 2016). As we also found that CtIP is involved in RNA editing, we wondered whether this connection is bidirectional. Thus, we hypothesized that ADARs might also play a role in DNA end resection. To test this idea, we studied RPA recruitment to DBS induced by laser microirradiation. As control we used U2OS cells depleted for CtIP. As shown in Figure R47, the lack of ADAR2 or CtIP produced a defect in DNA resection, measured as a reduction of RPA recruitment to laser tracks. This involvement has been also confirmed by other members of the Figure R46. CtIP is required for RNA editing when ATR is present. Cells expressing RNAG reporter were transfected with the indicated siRNAs and also treated with 10 µM ATR inhibitor (ATRi) or DMSO as control for 2 hours. When indicated, cells were irradiated 10 Gy and incubated for 10 hours (black bars). The graph represents the average and standard deviation of at least four independent experiments. The percentage of green cells, was calculated from a total of 10.000 events that emit red fluorescence. Those cells were analysed on BD FACSAriaTM using FACSDiva v5.0.3 software and statistical significance was calculated using two-way ANOVA test (* p<0.05, ** p<0.01, and *** p< 0.001). ! Results - Chapter three ! 83 laboratory using other techniques and depleting other members of the ADAR family. We also tested whether ADAR binds to DNA damage to carry out its functions in DNA repair. However, we could not see either ADAR1 or ADAR2 recruitment at any time after DNA damage (Figure R48). 15. ADAR resection role depends on DNA:RNA hybrid resolution Recently, R-loop biology has been linked with DNA repair and the DDR. Indeed, R- loop accumulation has been shown to block resection (Bhatia et al., 2017; Gorthi et al., 2018) Moreover, it has been published that DNA repair and RNA metabolism proteins are involved in R-loops resolution (Kazuko Nishikura, 2010; Chakraborty, Huang and Hiom, 2018; Cristini et al., 2018; Makharashvili et al., 2018). In fact, ADAR proteins have been shown to edit sequences in the contest of DNA:RNA hybrids (Zheng, Lorenzo and Beal, 2017). Thus, we decided to test if ADAR phenotype in resection could be related with editing R-loops. Figure R47. Resection upon ADAR2 depletion. U2OS cells were transfected with the indicated siRNA. Representative images of RPA recruitment to DNA damage caused by laser microirradiation in the different samples are shown on the left side. As control, γH2AX recruitment and CtIP depletion (siCtIP) were used. The plot (right side) shows the percentage of cells showing RPA laser tracks and the average and standard deviation of at least four independent experiments. At least 20 cells per preparation were studied and the number of stripes were analysed using FIJI software. Statistical significance was analysed by one-way ANOVA (** p<0.01 and *** p<0.001). ! Figure R48. ADAR proteins recruitment to DSB. Representative images of the recruitment of ADAR2 (top), ADAR1 (bottom) or γH2AX, as a control of DNA damage, to double strand breaks created by laser microirradiation at the indicated times.! Results - Chapter three ! ! 84 To test this idea, we studied RPA foci formation in cells depleted for ADAR2 and overexpressing the Ribonuclease H1 (RNAseH1), which removes the RNA in the DNA:RNA hybrid (Cerritelli and Crouch, 2009). Interestingly, we saw that ectopic expression of RNAseH1 increased RPA foci formation even in control samples (Figures R49A, R49B and R49C), which indicated that DNA:RNA hybrids indeed block resection. As expected, cells downregulated for ADAR2 decreased RPA foci formation in U2OS cells (Figure R49A) and also in HeLa cells (Figure R49B). In addition, overexpression of RNaseH1 fully suppressed such phenotype, suggesting that the role of ADAR in resection is through DNA:RNA hybrid resolution. To confirm this finding, we studied the recruitment of RPA to chromatin damaged by laser microirradiation in HeLa cells depleted for ADAR2 and transfected with RNaseH1 plasmid. As before, cells downregulated for ADAR2 failed in the recruitment of RPA to DNA damage sites and the overexpression of RNAseH1 partially rescued such phenotype (Figure R49C). Figure R49. The expression of RNAseH1 reverts the resection defect caused by ADAR2 depletion. A. Percentage of RPA foci positive cells detected by immunofluorescence in U2OS cells exposed to 10 Gy of ionizing radiation and incubated for 1 hour. Those cells were transfected with a siRNA against ADAR2 or a non-target siRNA (siNT) as control and, also, they were transfected with a plasmid to express RNAseH1 (black bars) or a control plasmid (white bars). The graph represents the average and standard deviation of at least four independent experiments. For each replicate, at least 200 cells were measured. Two-way ANOVA was used to assess statistical significance (* p<0.05 and ** p<0.01). B. Same as (A), but using HeLa cells. C. HeLa cells were transfected with the indicated siRNAs and plasmids and microirradiated with a laser to induce DNA damage. The percentage of cells positive for RPA recruitment to DSB was plotted. The graph also shows the average and standard deviation of at least four independent experiments. At least 20 cells per replicated were studied and the number of stripes were analysed using FIJI software. D. Cell cycle distribution of U2OS cells transfected with the indicated siRNAs and plasmids. Per each replicate, 10.000 cells were analysed using BD FACSAriaTM and FACSDiva v5.0.3 software. E. Same as (D) but HeLa cells were used. F. A representative western blot shows the levels of the indicated proteins from U2OS cells depleted with the depicted siRNAs and transfected with RNAseH1 plasmid or not. Results - Chapter three ! 85 These changes in resection were not caused by a delay in cell cycle (Figure R49D for U2OS and Figure R49E for HeLa). An example of expression of RNaseH1 and depletion of ADAR2 in U2OS cells is shown in Figure R49F. In order to test whether this connection between DNA repair and DNA:RNA hybrid resolution is exclusively of ADAR, or a general theme on RNA metabolism proteins, we studied whether RNAseH1 also supressed the DNA resection defect in cells depleted for SF3B2 (see chapter one, section 4). Strikingly, overexpression of RNAseH1 also reverted such defect (Figure R50A) and it was not due to a fail in cell cycle progress (Figure R50B). An example of expression of RNaseH1 and depletion of SF3B2 is shown in Figure R50C. All these data indicated that DNA:RNA hybrids hamper DNA resection and different RNA-related proteins might mitigate such effect. As mentioned previously (chapter one, section 5), SF3B2 facilitates CtIP recruitment to DNA damage. As CtIP is functionally connected with RNA editing, we wondered whether ADAR proteins might also regulate CtIP accumulation at damaged chromatin. However, and in stark contrast with SF3B2 downregulation, the ADAR2 depletion did not affect CtIP recruitment (Figure R51), which indicated that the role of ADAR2 in resection is not through CtIP accumulation. Interestingly, RNAseH1 overexpression partially rescued SF3B2 phenotype (Figure R51). These data suggest that DNA:RNA hybrids might affect resection by different means in connection with several RNA-related factors. Figure R50. RNAseH1 partially reverts SF3B2 phenotype in resection. A. U2OS cells depleted for the indicated proteins using siRNAs were transfected with RNAseH1 or control plasmid. Those cells were irradiated 10 Gy and incubated for 1 hour. The plot shows the percentage of cells positive for RPA foci and the average and standard deviation of at least four independent experiments. For each replica, at least 200 cells were measured. Statistical significant between samples was analysed by one-way ANOVA. One or two asterisk(s) denote statistical significance at p<0.05 and p<0.01, respectively. B. Cell cycle distribution of cells upon transfection of the indicated siRNAs and plasmids. The average and standard deviation of at least four independent experiments are plotted. Per each replicate, 10.000 cells were analysed using BD FACSAriaTM and FACSDiva v5.0.3 software. C. Representative western blot showing the expression levels of SF3B2, RNaseH1 and β-actin in U2OS cells transfected with the indicated siRNAs and transfected with the depicted plasmids. ! Results - Chapter three ! ! 86 Another important protein involved in DNA:RNA hybrid resolution is the helicase Sentaxin (SETX). It is recruited to DSBs at transcriptionally active loci and is required to regulate γH2AX signaling and to promote a correct RAD51 loading (Cohen et al., 2018). Furthermore, it is published that BRCA1 is recruited to R-loops that are formed during transcription termination regions and collaborate with SETX in the removal of DNA:RNA hybrids during recombination (Hatchi et al., 2015; Zhang et al., 2017). Thus, if DNA:RNA hybrids hamper resection, SETX should show a phenotype similar to ADAR2 because in both cases there would be R-loop accumulation and consequently a resection block. In order to test this idea, we studied RPA foci formation in cells downregulated for SETX. As represented in Figure R52, SETX showed the same phenotype than ADAR2 and the depletion of either of those proteins (Figure R52A) Figure R52. SETX is involved in resection. A. A representative western blot showing expression levels of the depicted proteins in cells transfected with the indicated siRNAs. B. Cells were depleted for the indicated siRNAs, irradiated with 10Gy and incubated for 1 hour. The plot shows the percentage of cells positive for RPA and the average and standard deviation of at least three independent experiments. For each replica, at least 200 cells were measured. Statistical significant between samples was analysed by one-way ANOVA. One asterisk denotes statistical significance at p<0.05. ! Figure R51. CtIP recruitment to DSB. CtIP recruitment to DNA damage induced by laser microirradiation was scored using antibodies against CtIP or γH2AX, as control, using cells depleted for the indicated proteins using siRNAs. siNT denotes a siRNA with no target in the human genome. Those cells were also transfected with a plasmid to express RNASeH1 (black bars) or a control plasmid (white bars). The quantification (right) and depictive images (left) of one representative experiment out of four with similar results, is shown. ! Results - Chapter three ! 87 hampers DNA resection (Figure R52B). Moreover, another member of our laboratory, Sonia Jimeno, confirmed those results using the SMART technique. Thus, we can confirme that SETX is also involved in DNA resection. All those data together suggest DNA:RNA hybrid acumulation produces a resection and a homologous recombination deffect and, moreover, that the role of ADARs in DNA resection is through DNA:RNA hybrid biology. Discussion Discussion 91 DNA suffers thousands of damages every day that compromise the stability of the genome. For this reason, the accurate repair of those lesions is vital for the cells. DNA end resection is a critical step in this process because it controls the way the damage is going to be repaired (Symington, 2016). CtIP is a key protein involved in the initial steps of the resection process, so its regulation is vital for the appropriate DNA repair (You et al., 2009; Makharashvili and Paull, 2015). In order to understand better how CtIP role in the response to DNA damage is modulated, in our laboratory, Dr. Ana López-Saeevedra isolated nine new interactors of this protein, including two previously described CtIP partners, CCAR2 and hnRPU (Lopez-Saavedra et al., 2016). The majority of them were related with RNA metabolism including three subunits of the splicing complex SF3B (Prados-Carvajal et al., 2018). The number of RNA metabolism proteins known to participate in the DNA damage response (DDR) has increased significantly in recent years to include splicing and alternative splicing factors (Shkreta and Chabot, 2015). Indeed, mRNA processing factors are commonly enriched in many genome-wide screenings aimed to find new factors involved in the DDR and the maintenance of genomic integrity (Paulsen et al., 2009; Adamson et al., 2012; Lopez-Saavedra et al., 2016). Moreover, many RNA- related proteins have been detected as targets of the DNA damage-induced posttranslational modifications (Smolka et al., 2007; Jungmichel and Blasius, 2013). One clear connection between the resection machinery and the mRNA related proteins is the regulation of BRCA1 at many different levels. Hence, PRP19, an ubiquitin ligase involved in pre-mRNA splicing, regulates the protein levels of BRCA1 (Abbas et al., 2014). Additionally, PRP19 binds to RPA and localizes to DNA damage through this interaction, promoting DNA-damage-dependent RPA ubiquitylation, driving ATR activation (Marechal et al., 2014). Furthermore, EDC4, a known component of processing-bodies and regulator of mRNA decapping, seems to be part of the BRCA1- BRIP1-TOPBP1 complex. Indeed, EDC4 functionally phenocopies BRCA1, playing a role in HR-mediated DNA repair by regulation of end resection (Hernandez et al., 2018). Due to this close connection between DDR proteins and RNA metabolism, when we saw that DNA repair protein CtIP interacts with three subunits of the splicing complex SF3B, we decided to study the relationship between both factors. First, we tested the interaction between CtIP and some subunits of SF3B by several methods. We found that Discussion 92 in all cases it was constitutive and increased upon DNA damage (Figures R1 and R2). Nevertheless, this increase was abolished when checkpoint was inactivated by ATM inhibition (Figure R2D). It is reported several important proteins involved in DNA repair are phosphorylated by ATM/ATR in response to DNA damage (Matsuoka et al., 2007; Marechal and Zou, 2013). Strikingly, we discovered that some SF3B roles in resection rely on serine 289 in SF3B2, a predicted ATM/ATR phosphorylation site (Matsuoka et al., 2007), although its interaction with CtIP is not one of them (Figure R2C). One explanation would be that this interaction does not occur at double strand break (DSB) and therefore ATM does not have to phosphorylate SF3B and recruit it to damaged chromatin. These data are supported by the fact that we do not see SF3B2 recruitment either by the lacO system or by laser lines (data not shown). Additionally, the majority of CtIP-SF3B interaction does not colocalize with ionizing-radiation induced foci (IRIF) labelled with MDC1 (Figure R2E), a protein recruited early at DSB (Mochan et al., 2003; Shang, Bodero and Chen, 2003; Xu and Stern, 2003), indicating that it does not occur at DNA damaged sites. Additionally, we discovered that SF3B not only interacts with CtIP, but also with another resection protein, BRCA1 (Figures R1C and R1D). Moreover, and as mentioned before, we demonstrated that SF3B2 regulates BRCA1 mRNA splicing and expression by microarray and then by qPCR (Figures R24 and R11, respectively). Interestingly, it is reported that CtIP-BRCA1 interaction affects resection and, indeed, modulates its speed (Cruz-García, López-Saavedra and Huertas, 2014). Furthermore and in agreement with our data, Harkin’s lab published that BRCA1-SF3B1 physically interact (Savage et al., 2014). Taking together all those data, we thought that BRCA1 might be mediate CtIP-SF3B interaction. However, this hypothesis was rejected when BRCA1 was depleted and we still observed that CtIP and SF3B interact (Figure R1D). Moreover, and differently to the interaction with BRCA1, SF3B does not only contact CtIP upon DNA damage but constitutively (Figure R1D) (Prados-Carvajal et al., 2018). Furthermore, we saw that SF3B, as CtIP and BRCA1, is involved in DNA resection and DSB repair mainly through the homologous recombination pathway (Figures R4- R8). Interestingly, only part of this role is through CtIP (Figure R18). We also determined that SF3B controls CtIP-mediated resection at, as least, two independent levels: CtIP protein abundance (Figure R9) and CtIP ability to bind to DSBs (Figures R15 and R51). This data are in agreement with the fact that the inhibition of SF3B1 Discussion 93 upon irradiation does not affect the upstream proteins in DDR γH2AX and MDC1, but reduced ubiquitylation of damaged chromatin and impaired recruitment to DSBs of factors downstream of MDC1 like WRAP53β, RNF168, 53BP1, BRCA1 and RAD51 (Shkreta and Chabot, 2015). Thus, it could be possible that SF3B2 is acting in the same way and it is regulating the recruitment or retention to damaged chromatin not only of CtIP but also other factors involved in later steps of DDR. Indeed, it seems that SF3B2 requires a phosphorylation by ATM to be able to control CtIP recruitment, since the phosphomutant FLAG-SF3B2-S289A, which can no be phosphorylated for this kinase, is unable to recruit CtIP to DSB (Figure R15G). However, we cannot conclude that the repair defect observed upon SF3B depletion was entirely dependent on CtIP because we noted that the exogenous expression of CtIP did not rescue all phenotypes in resection caused by SF3B downregulation, as the impairment in homologous recombination (Figure R18), the low protein levels of CHK1 and its phosphorylated form activated upon irradiation (pCHK1) (Figure R19) and even the survival of cell exposed to etoposide (Figure R20). Thus, it suggests that SF3B complex has an additional role in DNA repair and cell response to genotoxic agents independent on CtIP. These additional effects could be also related with the control of the expression, splicing or recruitment of determinate repair genes. In fact, it has been just reported a double involvement of some RNA factors, which has simultaneous RNA-mediated and RNA-indirect roles in homologous recombination (Anantha et al., 2013). Not only we do not discard the possibility that the SF3B complex plays additionally more direct roles in DNA repair, but our microarray data demonstrate that SF3B controls basal and irradiation-induced splicing and expression of several DNA repair factors (see results; chapter two), including both CtIP and BRCA1 themselves. However, we did not observe a decrease in the protein stability (Figure R10), suggesting that SF3B complex controls CtIP abundance at the level of mRNA. It is possible that SF3B-CtIP interaction might be important for the CtIP mRNA splicing changes observed in the microarray and those changes in turn affect the abundance of mRNA and protein levels, explaining the low levels observed upon SF3B depletion (Figure R9). This might be similar to the previously observed regulation of FIR level and mRNA splicing by SF3B1, that requires the formation of a FIR-SF3B1 complex (Matsushita et al., 2012). A similar scenario might explain how PIF1 splicing is regulated by CtIP, as it has been published that CtIP directly interacts with PIF1 (Jimeno et al., 2018), but this idea should be tested Discussion 100 seems to be a bidirectional relationship because RPA stimulates RNAseH1 activity (Nguyen et al., 2017). Then, we discovered that the RNA editing complex ADAR also has a role in DNA resection and DNA repair, since its depletion caused an impairment of those processes (Figures R47 and R49). We first hypothesised that ADARs phenotype in resection could be due to an indirect effect because it was editing mRNA of DNA repair proteins. This will agree with the accrue of RNA editing as a response to DNA damage appearance. However, our colleague Dr. Mª Jesús Fernandez-Ávila did not see such interaction between ADARs and DNA repair genes mRNA performing RIP. Further investigation will clarify if this is somehow involved. On the other hand and as mentioned before, it is published that R-loops concentrate at the site of DSBs (Cohen et al., 2018) and the accumulation of such structures can block resection (Hamperl and Cimprich, 2014; Costantino and Koshland, 2018). Thus, we thought that ADARs phenotype in resection might be caused by a direct effect and ADARs were editing RNA in DNA:RNA hybrids facilitating R-loops resolution and therefore DNA end resection. In agreement with this hypothesis, recently, it has been published that ADARs edit RNA in DNA:RNA hybrids in vitro (Zheng, Lorenzo and Beal, 2017) and ADAR2 interacts with such structures (Damage et al., 2018). Thereby, if our hypothesis Figure D2. Different R-loops position causes diverse effects on DNA end resection. Depending on DNA strand paired with the RNA of DNA:RNA hybrids, those structures would have different impacts on DNA end resection. If the resected strand was paired with the RNA, such structure might block the activity of the resection machinery (I). On the other hand, if the resected strand was displaced by the hybrid, the resection machinery should access and process it (II). However, if structures like G-quadruplex were formed in the non-transcribed strand, they would impair DNA end resection processivity (III). Discussion 101 was correct and ADAR role on DNA resection is related with DNA:RNA hybrids removal, there were some predictions that we would expect to be fulfilled. First, the requirement for ADAR in DNA resection should disappear if R-loops are removed. Second, ADAR proteins might interact with the DNA:RNA hybrid removal machinery. Thirst, the depletion of the DNA:RNA hybrid removal machinery should hamper DNA end resection. As seen in Figure R49, the RNAseH1 overexpression suppressed the phenotype of ADAR2 in DNA resection corroborating the first prediction. This effect is not specific of RNA editing proteins because it also suppressed the phenotype in resection caused by SF3B2 (Figure R50). Indeed, RNAseH1 overexpression decrease CPT-induced γH2AX foci in cells depleted for DHX9 (Cristini et al., 2018), a protein that we have already seen is involved in DNA resection (Prados-Carvajal et al., 2018). This also agrees with the report that the depletion of CtIP stimulates the accumulation of stalled RNA polymerase and DNA:RNA hybrids and this phenotype is reduced overexpressing RNaseH (Makharashvili et al., 2018). Hence, all these data suggest that DNA:RNA hybrids resolution promotes DNA end resection and the role of ADAR in DNA repair is through the edition of DNA:RNA hybrids. As mentioned before, SETX is involved in DNA:RNA hybrid resolution. This protein is recruited to DSBs at transcriptionally active loci and is required to regulate γH2AX signalling and to promote a correct RAD51 loading (Cohen et al., 2018). Furthermore, BRCA1 is recruited to R-loops that are formed during transcription termination regions. It also mediates the binding of SETX and is essential for R-loop-induced damage repair (Hatchi et al., 2015; Zhang et al., 2017). Thus, it seems that BRCA1 and SETX collaborate in the removal of DNA:RNA hybrids during recombination. In our laboratory, Dr. Sonia Jimeno has demostrated that ADAR interacts with the helicase SETX and BRCA1 (data not shown), which corroborates the second prediction of our model. Finally, we found that SETX depletion impairs homologous recombination (data obtained by Dr. Fernando Mejías and Dr. Sonia Jimeno in our laboratory) and DNA end resection (Figure R52). Thereby, we confirm the thirst assumption of our model, the depletion of the DNA:RNA hybrid removal machinery (SETX) hampers DNA end resection. Thanks to all those data we hypothesize that BRCA1, SETX and ADARs interact and are involved in DNA:RNA hybrid removal and the depletion of one of these proteins accumulates R-loops and thereby it blocks resection and hamper HR. Discussion 102 Mechanistically, we propose that ADARs activity leads to the deamination of the RNA moiety of those DNA:RNA hybrids. The change from adenine to inosine caused by such editing will create a mismatch with the thymidine in the DNA strand. If a sufficient number of adenines were edited, the stability of the DNA:RNA pairing would be compromised, facilitating the dissolution of the hybrid. Furthermore, we found other evidences the relationship between DNA end resection and RNA editing. We observed that upon DNA damage, RNA editing activity increases (Figure R43). This might be relevant for the elimination of R-loops prior resection, but also as a way to spread the DDR signalling by changing multiple mRNAs sequences. Moreover, blocking the DDR checkpoint by ATR inhibition affects strongly RNA editing (Figures R45 and R46). On the contrary, depletion of proteins involved in DNA end resection and splicing regulation like BRCA1 or SF3B2, increase the percentage of editing in cells (Figure R43), suggesting that those two RNA related mechanisms are related. There are further evidences that confirm this connection, since they sometimes coincide in time and space (Raitskin et al., 2001; Bentley, 2014). First, it has been shown several mechanism by which RNA editing can modulate splicing (Goldberg et al., 2017; Hsiao et al., 2018). The creation or elimination of splice sites and branch points by RNA editing proteins can explain some editing-dependent splicing events (Rueter, Dawson and Emeson, 1999; Schoft, Schopoff and Jantsch, 2007). Indeed, ADAR2 autoregulates its own expression by alternative splicing i.e. ADAR2 edits its own pre-mRNA and creates an alternative 3’ splice site. This alternative spliced isoform contains a premature stop codon resulting in a dysfunctional truncated protein (Filippini et al., 2018). It is also published that ADAR1 performs the edition of the STAT3 intron and therefore it affects STAT3 alternative splicing (Goldberg et al., 2017). Another mechanism through RNA editing may influence splicing is by alteration of RNA secondary structures (Rieder and Reenan, 2012). The substitution of adenosines by inosines may alter the stability of the RNA structures and influence the accessibility of splice sites (Morse, Aruscavage and Bass, 2002; Levanon et al., 2004). This could be the case of R-loops. As mentioned, ADARs might deaminate RNA to remove DNA:RNA hybrids and thus promoting DNA end resection. On the other hand, RNA splicing efficiency also plays a key role in regulating RNA editing levels (Licht et al., 2016; Hsiao et al., 2018). Thus, the interaction between ADARs and splicing factors seems to improve editing efficiency (Tariq et al., 2013). Moreover, it has been shown Discussion 103 that decrease in ADAR expression induced global splicing changes in both Drosophila and human cells, although the mechanisms are still unknown (Solomon et al., 2013; St Laurent et al., 2013). All those findings show a clear relationship between RNA splicing and RNA editing. However, it is the first time that there have been evidences of the connection of DDR with RNA editing. Indeed, our data suggest that all those processes that we are studying during this Thesis are related one each other and there is a crosstalk between DNA end resection and RNA metabolism. We have observed that RNA splicing and DNA repair has a bidirectional relationship as we saw with SF3B and CtIP (results; chapter one) or CtIP with PIF1 (results; chapter two). Additionally, RNA editing is related with DNA repair because of the CtIP-ADAR relationship, the role of CtIP and other DNA repair proteins in RNA editing, the function of ADAR and SETX in resection and the rescue of phenotype caused by the depletion of DNA repair proteins upon RNAseH1 overexpression (results; chapter three). In order to organize these chromatin transactions to deal with various forms of genotoxic stress, cells would require rapid spatio-temporal coordination. In fact, it has been published mRNA splicing is subject to spatio-temporal control by transcribing polymerases, chromatin remodelers and histone marks (Luco et al., 2011). Since RNA splicing usually occurs in a transcription-coupled manner, splicing perturbations might affect transcription elongation and therefore might increase R-loop formation (Nguyen et al., 2018). On the other hand, it has been reported that the spliceosome maintains transcription of essential repair factors and also prevents the formation of R-loop structures in activated transcription sites, which promotes DNA repair (Tanikawa et al., 2016). Indeed, it is described that the binding of splicing proteins and ribonucleoprotein complexes to nascent transcripts avoid genomic instability by dissuading the formation of DNA:RNA duplexes (Shkreta and Chabot, 2015). Moreover, it is published that SF3B1 mutations promote an accumulation of R-loops and the expression of RNAseH1 alleviates these effects (Sorrells et al., 2018). These data are in agreement with our findings, since RNaseH1 also suppresses the phenotype in resection caused by SF3B2 depletion (Figure R49). Thus, those authors indicate that R-loop is an intermediate to impaired splicing and appears when splicing efficiency is limited but not completely abolished (Sorrells et al., 2018). In fact, they propose that diseases associated with splicing factor mutations could be treated regulating R-loop levels (Sorrells et al., Discussion 104 2018). Thus, in this Thesis, we have shed some light on the tight functional connection of three independent events: DNA resection, mRNA splicing and RNA editing. Our data reinforce the idea of an extensive crosstalk between those three mechanisms, in which CtIP seems to play a critical role, somehow controlling resection activation and DNA damage-dependent RNA splicing and RNA editing. This might clarify the pivotal role of CtIP for genome integrity. Furthermore, such connection could explain why these three fundamental processes are so relevant in cancer development. It is well established that some mutation in DNA repair genes increase the risk of cancer (Hosoya and Miyagawa, 2014; Paridah et al., 2016). For example, mutations in BRCA1 render tumours deficient in HR pathway, what increases genome instability and hypersensitivity to DNA damaging agents (Bianchini et al., 2016). Indeed, patients with a germ-line inherited defect in BRCA1 are at increased risk of breast, ovarian, prostate and other cancers (Paridah et al., 2016). Moreover, several years ago, CtIP mutations were found in breast, pancreas, colon, and ovarian carcinomas (Wong et al., 1998). In accordance with such results, a meta-analysis of gene expression data sets showed that lower levels of CtIP correlated with absence of endocrine receptors and higher risk of metastasis in breast tumours (Bianchini et al., 2016). Hence, in agreement with that, recently, in our laboratory, it was discovered that CtIP protein levels are strongly decreased in breast tumours, and that CtIP absence correlates with tumour aggressiveness and poor prognosis features (Soria-Bretones et al., 2013). This connection with cancer has been always explained as a defective DNA repair. However, considering the involvement of CtIP, described here, and other DNA repair factors in RNA splicing and editing, the cancer connection should be revisited on the light of those novel roles. Conversely, recent studies have detected recurrent mutations in components of the spliceosome in myelodysplastic syndromes (Shiozawa et al., 2018; Pollyea et al., 2019), renal cell carcinoma (Yang et al., 2017; Verma and Das, 2018), chronic lymphocytic leukemia (Agrawal et al., 2017; Maleki et al., 2019), lung adenocarcinoma (Kim et al., 2018; Mao et al., 2019), breast cancer (Gokmen-Polar et al., 2019; Zhao et al., 2019) or pancreatic cancer (Tian et al., 2015; Zhou et al., 2017). Along the same lines, alterations in expression of some splicing factors like SF3B can derive in various types of cancers (Goswami et al., 2014; Maguire et al., 2015; Alsafadi et al., 2016; Zheng et Discussion 105 al., 2018). This could be due to the fact the importance of splice variants in several important genes regulator-signalling pathways like apoptosis, metabolism, and angiogenesis (Grosso, Martins and Carmo-Fonseca, 2008). However, another tantalizing possibility, not yet analysed in detail, is that some of those connections of SF3B with cancer might be a consequence of its more direct role in DNA repair. Finally, RNA editing alterations are also a common phenomenon in cancer. Depending on the tissue of origin and disease stage, different cancers have been associated with induction or reduction of RNA editing levels (Paz et al., 2007; Han et al., 2015; Paul et al., 2017). In glioblastoma, ADAR2 inhibits tumour growth modulating the CDC14B- Skp2-p21-p27 pathway (Galeano et al., 2013). In colorectal cancer, a higher level of the Ras Homologue family member Q (RHOQ) RNA editing was observed (Han et al., 2014). In advanced stage of prostate cancer, ADAR1 and ADAR2 edit the androgen receptor mRNA contributing to impair the abilitiy of proteins to interacts with androgenic or anti-androgenic ligands (Martinez et al., 2008). Taking in account the data obtained during this Thesis, it would be interesting to investigate if the role of ADARs in DNA resection is relevant on those kinds of cancers. In summary, the findings presented in this Thesis show the important connection between DNA end resection and RNA metabolism. Thus, further studies of those processes will contribute to understand better their relationship with cancer appearance, and therefore will help to find early detection of such disease or new targets of therapy. Conclusions Conclusions ! 109 1. The SF3B complex regulates DNA resection and DNA double strand breaks repair through its connection with CtIP. Such modulation is achieved, at least, at two distinct levels: (I) affecting levels of CtIP and (II) controlling CtIP recruitment to damaged chromatin, which is dependent on Ser 289 of SF3B2, an ATM phosphorylation site. 2. Additionally, the SF3B complex plays other roles in DNA double strand break repair and cellular response to genotoxic agents independent of its functional relationship with CtIP. 3. SF3B controls RNA splicing and expression of several genes including DNA repair factors. Its role is particularly relevant regarding DNA damage-dependent alternative splicing of genes. 4. CtIP also regulates DNA damage-dependent RNA splicing of a set of specific genes. Among them, the helicase and DNA resection factor PIF1. CtIP and DNA damage alternatively regulated PIF1 isoforms play different roles in DNA repair; whereas vPIF1 variant affect mostly early events in the DDR, tPIF1 seems more relevant at later time points. Moreover, they are localised in different places in the cell. 5. RNA editing is induced by DNA damage. Such increase is dependent on ATR activity and CtIP, albeit independent of its function in DNA processing. 6. ADARs are involved in DNA end resection and homologous recombination. Such role relies on DNA:RNA hybrids resolution, as ADAR requirement in DNA resection disappears upon RNAseH1 overexpression. Depletion of other proteins involved in DNA:RNA hybrid removal, such as SETX, also hampers DNA end resection. Materials and Methods Materials and Methods ! 113 1. Cell culture procedures 1.1. Growth media and conditions In this Thesis, we used U2OS (human osteosarcoma), HEK293T (human embryonic kidney) and HeLa (human uterine cervical cancer) or cells lines derived from them stably harbouring the indicated constructs (Table M1) obtained by plasmid transfection or by infection with lentiviral particles (see section 1.3 for methodology). These cells were cultured in high-glucose Dulbecco’s Modief Eagle Medium (DMEM; Sigma, D6546), supplemented with 10% fetal bovine serum (Sigma, F7524), 2 mM L- glutamine (Gibco, 25030024), 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco, 15140122) at 37ºC in the presence of 5% CO2 in HEPA class 100 incubators (Thermo). Cells transfected with the different constructs were cultured in the same conditions and were selected by adding either 0.5 mg/ml G418 (Sigma, A1720) or 1 µg/ml puromycin (Sigma, P8833) to the culture medium, according to the selection marker of the plasmid inserted. The only exception was U2OS19ptight13 cells, which contain a doxycycline-inducible system (Table M3). For them, DMEM without phenol red (Sigma, D1145) and supplemented with 10% tetracycline-free FBS (Biowest, S181T) 0.5 mg/ml sodium pyruvate (Gibco, 11360070) and 0.8 mg/ml G418 was used. Table M1. Human cell lines used in this Thesis. Cell name Description Selection Source/Reference U2OS Human osteosarcoma - ATCC HTB-96 HEK293T Human embryonic kidney - ATCC CRL-11268 HeLa Human uterine cervical cancer - ATCC CCL-2 U2OS GFP U2OS cell line stably expressing GFP construct G418 Kind gift from Steve Jackson (Sartori et al., 2007) U2OS GFP-CtIP U2OS cell line stably expressing GFP-CtIP construct G418 Kind gift from Steve Jackson (Sartori et al., 2007) Materials and Methods ! ! 120 Opti-MEM (150 µl) were mixed with FuGENE (9 µl) in a 3:1 FuGENE:DNA ratio. Solution was incubated for 15 minutes at room temperature and finally was added dropwise to the plate with carefully rocking. 1.2.2.2. Calcium phosphate transfection Calcium phosphate transfection protocol was used to introduce the plasmids required for lentivirus production in HEK293T cells. For 150 mm plates, 7x106 cells were grown for 24 hours and the medium was changed at least 30 minutes before transfection. A mixture of the three plasmids required was prepared with a 3:2:1 ratio, i.e. 30 µg of the plasmid containing DNA of interest, 20 µg of the vector with virus capsid genes (p8.91) and 10 µg of the vector with virus envelope genes (pVSVG) (Table M3). The DNA mix was prepared in a volume of 1 ml containing 250 mM CaCl2, and was added dropwise while bubbling into 1 ml of 2x HEPES buffered saline (HBS; Sigma, 51558) using a glass Pasteur pipette. The mix was incubated 30 minutes at room temperature and added dropwise to the cells while carefully rocking the plate. The following day, the media was removed and exchanged with fresh complete DMEM to minimise cell death. Some experiments required the use of smaller plates, thus the volumes used were scaled accordingly in those cases. 1.3. Lentiviral production and transduction The other strategy used for protein downregulation was infection with lentiviral particles harbouring shRNAs vectors targeted against the gene that codes the protein of interest. Note that after transfection of the cells, all procedures were carried out in a P2 biological safety room. First, HEK293T cells transfected with the plasmids containing the specific shRNA (pLKO.1), p8.91 and pVSVG as described in the previous section (1.2.2.2) were grown to produce lentiviral particles. Two days after transfection, medium was collected and filtered using 0.45 µm polyvinylidene difluoride (PVDF) filters (Millex- HV, Millipore, SLHV035RS) to remove cellular debris and 8 µg/ml hexadimethrine bromide (Polybrene; Sigma, H9268) was added. Aliquots were stored at -80ºC until their use. For transduction, U2OS cells were seeded, and 24 hours later medium with lentiviruses was thawed and added to the plates. Lentiviruses produced in one 150 mm Materials and Methods ! 121 plate were used to infect four 100 mm plates. The medium was replaced 8 hours later to remove viral residues and hexadimethrine bromide. Knockdowns were validated by western blot. In addition, lentiviral transduction was also used to introduce pBFP-ISceI vector to express the restriction enzyme I-SceI and the blue fluorescent protein (BFP) in U2OS cells harbouring a repair reporter system (see section 1.7 for details). Lentiviruses produced with pBFP-ISceI were produced in HEK293T cells and harvested as indicated above. For large-scale production of viruses, the medium containing lentiviral particles was pelleted by centrifugation during 1.5 hours at 22.000 rpm at 4ºC. Supernatant was removed and viruses were resuspended in 500 µl DMEM for each 150 mm plate. Finally, samples were aliquoted and stored at -80ºC until use. Before infection, viral production was titrated by BFP expression using flow cytometer (see section 1.6.2.). For transduction, lentiviral particles were diluted in DMEM supplemented with 8 µg/ml hexadimethrine bromide (Sigma, H9268) in order to achieve a multiplicity of infection (MOI) of 10. Cells were washed the day after infection with new media to remove viral residues and hexadimethrine bromide. 1.4. Single clone stable cell line generation To generate FLAG, FLAG-SF3B2, FLAG-SF3B2 S289A or FLAG-SF3B2 S289D stable cell line clonally pure, U2OS cells were transfected to integrate a specific DNA construct using FuGENE® HD transfection reagent as described in section 1.2.2.1. The following day cells were diluted 1:50 and 1:100 and seeded in 150 mm plates and let grow to obtain colonies from single cells. When those isolated colonies had a suitable size (between 100 to 200 cells), they were picked using an automatic pipette under an inverted optical microscope. To do that, first, cells have to be partially detached from the plates, so we used the chelating agent ethylenediaminetetraacetic acid (EDTA) to weaken cell adhesion without completely detaching cells. Then, culture medium was removed and 5 mM EDTA diluted in 10 ml PBS were added to the plate. Each colony was picked up under the microscope by aspirating 150 µl of solution while scrapping the surface to detach the cells and was seeded in a well of 24-well plate, which contains 500 µl DMEM. The media was replaced 24 hours later and fresh DMEM containing G418 or puromycin according to Materials and Methods ! ! 122 the plasmid selection marker was added. At least 30 clones were picked up and were grown separately before freezing them. Finally, different clones were analysed by western-blot to check DNA insertion. 1.5. Cell survival assays To study survival of the different cells used in this Thesis after exposure to DNA damaging agents, three different assays were performed: 1.5.1. Clonogenic cell survival assay On the one hand, clonogenic assays with diverse DSB-inducing agents were performed to study the survival of cells bearing the different PIF1-isoforms. DSBs were produced by ionizing radiation (see section 1.8.) or by treatment with the topoisomerase I inhibitor camptothecin (CTP; Sigma, C9911). The next day, GFPPIF1-isoform, GFP-PIF1-total and GFP cells were seeded in 60 mm plates and depleted for endogenous PIF1 using siRNA against it (see Table M2). Twenty-four hours after siRNA transfection, cells were seeded in 6-well plates at two different concentrations in triplicates. For IR 250 and 500 cells and for CPT treatment 500 and 1.000 cells were seeded per well. The next day, cells were either irradiated with doses of 0, 2 or 4 Gy, or treated with 0.01 µM, 0.05 µM or 0.1 µM camptothecin to induce DNA damage. In the case of CPT, cells were incubated with drugs or DMSO as control for 1 hour and then washed twice with PBS. Afterwards, fresh medium was added to each well and cells were incubated at 37ºC for 7-14 days to allow colonies formation. Once colonies were big enough, cells were stained and visualized using crystal violet. To do so, culture medium was removed, cells washed with PBS and treated with a solution of 0.5 % crystal violet (Merck, 1.15940.0025) and 20% ethanol (Merck, 1.00983.1000). Finally, the plates were dried and the number of colonies per well was scored for each condition and normalized to the untreated condition for each cell type. 1.5.2. MTT cell viability assay Due to the high rate of death observed upon depletion of SF3B2, the study of cell Materials and Methods ! 123 survival in theses cells by standard clonogenic assay was unfeasible. Thus, the commercially MTT cell growth assay kit (Sigma, CT02) was used in order to evaluate cell viability in response to treatments with DNA damaging agents after SF3B2 delepletion in cells harbouring the GFP or GFP-CtIP plasmids. U2OS cells expressing the indicated plasmid were transfected with the SF3B2 or non-target siRNA (Table M2). The following day, 7000 cells were seeded into a 96-well plate and after 6 hours they were treated with 0.5 µM etoposide (Sigma, E1383) or DMSO as control. Fourty two hours after the depletion, the samples were incubated for 4 hours at 37°C upon MTT solution addition. After removing the medium and MTT solution, formazan was solubilized in 100 µl HCl with 0.4 N isopropanol (AppliChem 131090.1211). Finally, cell viability was calculated measuring absorbance at 570 nm and 630 nm as described in the manufacturer’s protocol in Varioskan Flash (Thermo Electron corporation) and normalized with control. 1.5.3. Cell proliferation assay We took advantage of X-CELLigence RTCA DP (ACEA Biosciences), which uses non-invasive electrical impedance monitoring to analyse the cell proliferation at real time. 6.000 cells were harvested in special 16-well plates (ACEA Biosciences; 2801032) and 24 hours after that, they were depleted using the indicated siRNA (Table M2). The next day, cells were treated with 10 µM etoposide or DMSO as control for 1 hour to induce DNA damage, washed twice with PBS and incubated in the machine with fresh medium. Cell proliferation was analysed continuously each hour during seven days since cells were transfected with the indicated siRNAs. Finally, data was analysed using XCELLigance RTCA Softaware pro. 1.6. DSBs repair assays in vivo To study the influence of the different proteins analysed in this studio on DSB repair pathways, we took advantage of different repair systems, which we described below in section 1.6.1. In all of them, the DSB is generated by the endonuclease I- SceI, whose 18 bp recognition sequence has been integrated in the sequence. The repair of the damage by a specific pathway results in the expression of a green fluorescent protein (GFP) that could be quantified by flow cytometry (see section Materials and Methods ! ! 124 1.8.3). 1.6.1. Description of repair systems used 1.6.1.1. DR-GFP To test DSB repair by homologous recombination, we used two different reporters: DR-GFP or SA‐GFP (see Figure M1, panels A and B respectively). DR-GFP (Pierce et al., 1999) system was used to measure homology-directed repair efficiency, specifically short tract gene conversion. This reporter is composed of a full-length GFP gene mutated at a BcgI restriction site to contain the recognition site for the I-SceI endonuclease. This mutation consists of a substitution of 11 bp of the wild-type GFP sequence that supplies two in‐frame stop‐codons, which terminates translation and inactivates the protein. Downstream of this gene and oriented as direct repeat, an 812 bp internal GFP fragment (iGFP) is integrated. Homologous sequences in the two mutated GFP genes (marked in grey) are separated along 3.7 kb by the puromycin N-acetyltransferase gene. When I-SceI‐induced DSB is produced, iGFP sequence could act as a donor of wild‐type sequence information to the broken gene for repairing by non-crossover gene conversion within the limited amount of homology. Thus, restoration of an intact GFP gene is achieved by a short tract gene conversion. There are other possible homologous recombination outcomes, including crossover recombination, long tract gene conversion or SSA pathway. However, these events retain only the 5’ fragment of the GFP gene that would encode a carboxyterminal truncation, so there will be no accumulation of fluorescence in these cells. 1.6.1.2. SA-GFP SA-GFP (Stark et al., 2004) consists of two GFP gene fragments (5’GFP and 3’GFP) that share 266 bp of homology (marked in grey) and are separated by 2.7 kb. I‐SceI site is in the downstream GFP fragment inside the region of homology. Repair of the I‐SceI-induced DSB by single-strand annealing gives rise to a functional GFP gene after annealing the complementary strands of both GFP fragments in an intramolecular event, followed by appropriate DNA-processing steps that produce a 2.7 kb deletion in the chromosome. This reporter could also be repaired by other Materials and Methods ! 125 homology-directed repair, but without restoration of a functional GFP gene, so there will be no accumulation of fluorescence in these cells. 1.6.1.3. EJ5-GFP Otherwise, to test DSB repair by non-homologous end-joining (NHEJ), we used the EJ5-GFP reporter (Bennardo et al., 2008; Figure M1, panel C). It contains a promoter that is separated from a GFP coding cassette by the puromycin (puro) gene flanked by two I‐SceI sites in the same orientation. Once the two DSBs are produced after I-SceI expression, they are repaired by NHEJ. Then, the puromycin gene is excised and the promoter is joined to the rest of the cassette, leading to restoration of GFP expression. The ligation of two DSB generated by I-SceI could regenerate or not the I‐SceI site, but in both cases, GFP expression is obtained, which means that this reporter could be used to study all types, mutagenic and not mutagenic, NHEJ. Figure M1. Schematic representation of the DSBs repair systems used. A. DR-GFP reporter that measures gene conversion. GFP gene is restored when I-SceI-induced DSB is repaired by short-tract gene conversion. B. SA-GFP reporter that measures SSA by the repair of DSB generated by I-SceI and gives rise to a functional GFP. C. EJ5-GFP reporter that measures NHEJ and GFP expression is restored by the ligation of two I-SceI-induced DSBs. I-SceI target sited are marked in blue and homology regions in grey. ! Materials and Methods ! ! 126 1.6.2. DSBs repair assays in vivo Cells harbouring a single copy integration of DR-GFP, SA-GFP or EJ5-GFP reporters were used to analyse the role of SF3B complex in the different DSB repair pathways. To carry out the repair assays, 60.000 cells per well were plated in 6‐well plates. The following day, the indicated proteins were downregulated by using different siRNAs or shRNAs against them, including CtIP and a non‐target sequence as positive and negative controls. The medium was changed after 6 hours and the appropiated plasmids were transfected when it was indicated. The day after that, cells were infected with lentivirus harbouring I-SceI and BFP with a MOI of 10. DMEM containing lentiviral particles was supplemented with 8 µg/ml hexadimethrine bromide to enhance transduction, as indicated in section 1.3. DMEM including hexadimethrine bromide but without lentivirus was added to another set of depleted cells as control of basal fluorescence. After 24 hours, cells were washed with fresh medium and maintained in culture during an additional day. Cells were harvested and fixed with 4% paraformaldehyde for 15 minutes at 4ºC, followed by the analysis by flow cytometry of the blue and green fluorescence, as described in section 1.6.3. The repair frequency for SA-GFP, DR-GFP and EJ5-GFP reporters was calculated as the percentage of green cells from 10.000 events positive for blue fluorescence, considering the background of green fluorescence obtained in the samples without infection with lentivirus harbouring pBFP-ISceI plasmid as previously described (Stark et al., 2004; Pierce et al., 1999; Bennardo et al., 2008). Finally, to facilitate the comparison between experiments, the percentage or the ratio calculated were normalized with the control siNT or shNT in each case. 1.7. RNA editing assays in vivo 1.7.1. Description of RNA editing systems used In order to study the role of some proteins in RNA editing, RNAG and RNWG reporters were used (Garncarz W et al. 2013). In these systems, an R/G site stem-loop, used as an editing substrate, was cloned between the ORFs of red and green fluorescent protein (RFP and GFP respectively). The transfection of these plasmid Materials and Methods ! 127 leads to constitutive expression of RFP. In the RNAG system, a stop codon in the stem-loop sequence prevents GFP expression unless RNA editing converts such codon from a stop to tryptophan coding. Thus, in the absence of editing, only RFP is expressed, but upon editing a RFP-GFP fusion is expressed (see Figure R43). On the contrary, RNWG reporter expresses a “pre-edited” version of the stop codon (UIG instead of UAG) that constitutively produces RFP and GFP expression. Thus, RNWG reporter was used as control of basal gene expression. 1.7.2. RNA editing assays in vivo Cells were seeded in 60 mm plates and 24 hours later, transfected with siRNAs (Table M2). The cells were irradiated with 10 Gy and incubated during 12 hours before harvesting. In the experiments performed with protein inhibitors, the medium was exchanged 2 hours before irradiation with fresh DMEM containing 10 µM ATMi (KU55933), 5 µM ATRi (ETP46464) or DMSO as control. The medium was replaced and cells were irradiated or not with 10 Gy and incubated 10 hours. Cells were then harvested with trypsin, spinned down at 500 g for 5 minutes and washed with PBS. After that, half of the cells were fixed with 4% paraformaldehyde for 15 minutes at 4ºC in the dark, and later rinsed and resuspended in 150 µl of PBS. Red and green fluorescence was measured on BD FACSAriaTM using FACSDiva v5.0.3 software as indicated in section 1.8.3. The rest of cells were collected to study RPF and GFP fluorescence by western blot. 1.8. Flow cytometry 1.8.1. Cell cycle analysis To analyse cell cycle, cells were harvested by trypsinization, centrifuged at 500 g for 5 minutes and washed with PBS. Cell pellet was resuspended in 200 µl ice-cold PBS, fixed by adding 3 ml of cold 70% ethanol dropwise while vortexing at low speed and incubated at 4ºC at least overnight. Afterwards, cells were centrifuged at 500 g for 5 minutes, washed with PBS and treated with 250 µg/ml RNAse A (Sigma, R6148) and 10 µg/ml propidium iodide (Fluka, 81845) diluted in PBS for 15-30 minutes at 37ºC in the dark. Then, cells were incubated at 4ºC untill their analysis with a BD FACSCalibur Flow Cytometer (BD Biosciences, 342975) and CellQuest Pro software. Materials and Methods ! ! 128 1.8.2. Titration of lentiviral production As mentioned before (section 1.3), production of lentiviruses harbouring pBFP- ISceI plasmid was titrated by measuring BFP expression using a flow cytometer. In order to do so, 25.000 U2OS cells were seeded in the wells of 12-well plates. Next day, culture medium was removed and 300 µl fresh DMEM supplemented with 8 µg/ml hexadimethrine bromide was added. Then, one aliquot of viruses was thawed and diluted 1:10 in DMEM, and 0, 2.5, 5, 10, or 20 µl of this solution were added to each well for infection. After 6 hours of incubation at 37ºC in a P2 biological safety room, the volume was completed with 700 µl fresh DMEM. The following day, cells were washed with DMEM and incubated 24 hours upon addition of fresh medium. Finally, cells were harvested by trypsinization and pelleted by centrifugation at 500 g for 5 minutes in FACs tubes. Samples were washed with PBS and resuspended in 150 µl PBS, keeping them at 4ºC and in the dark until they were further analysed. BFP expression in each cell was then measured by cytometry using BD FACSAriaTM (BD Biosciences) and FACSDiva v5.0.3 software. The highest dilution factor that still resulted in 25-50% of the cells positive for BFP signal was used to calculate viral titer. 1.8.3. Fluorescent proteins analysis by flow cytometry DNA double-strand breaks repair systems restore expression of a fluorescent protein when a specific pathway repairs the damage (see paragraph 1.7.1.). U2OS cells bearing a single copy integration of the reporters DR-GFP (gene conversion), SA-GFP (SSA), EJ5-GFP (NHEJ) (see Table M3) were used to analyse the different DSB repair pathways. Cells were collected as explained in section 1.6.2. To prevent losing fluorescence, cells were fixed with 4% paraformaldehyde for 15 minutes at 4ºC in the dark, and later rinsed and resuspended in 150 µl of PBS. Samples were kept on ice in the dark until they were analysed by flow cytometry using BD FACSAriaTM and FACSDiva v5.0.3 software. The percentage of green cells, was calculated from a total of 10.000 events that were BFP positive, i.e. harbouring the I-SceI nuclease. In the cases of RNA editing reporters, cells were collected as explained in section 1.7.2. Then, they were analysed by flow cytometry using BD FACSAriaTM and FACSDiva v5.0.3 software. The percentage of green cells where analysed from a total of 10.000 events that also emits red fluorescence. Materials and Methods ! 129 1.9. DSB induction by ionizing radiation In order to induce double-strand breaks in cells, an irradiator device that emits gamma rays (Gamma‐Service Medical GmbH, BIOBEAM GM 8000) was used. The damage was generated by exposure of the samples to an encapsulated radioactive source (Cs‐137). Cells were irradiated in plates containing culture medium, using a dose of 10 Gy, or 2 and 4 Gy in the case of clonogenic assays. The irradiator is adjusted to the dose rate on the cover surface does not exceed 3 µSv/h. 2. Molecular Biology procedures 2.1 DNA manipulations 2.1.1. Plasmid DNA amplification Plasmid DNA was amplified using competent DH5α Escherichia coli cells. To do so, bacteria were transformed using heat shock protocol and afterwards DNA was purified. Briefly, 100 µl competent bacteria were mixed with 100 ng plasmid DNA and incubated for 30 minutes on ice. Then, cells were placed at 42ºC for 1 minute and returned into ice for a heat shock. After 3-5 minutes, 1 ml LB-Broth (Luria-Bertani; 10g/l NaCl, 5g/l yeast extract, 10 g/l triptone) was added and the transformed cells were incubated at 37ºC for 30 minutes to allow the expression of the marker gene. Cells were harvested by centrifugation at 6000 rpm for 2 minutes and plated in LB supplemented with 100 µg/ml ampicillin (Sigma, A9518). DNA was purified from one single bacterial colony using PureYieldTM Plasmid Maxiprep System (Promega, A2393) or Midiprep (Promega, A2492) following the manufacturer’s instructions. DNA concentration in the sample was quantified by measuring the absorbance at 260 nm using a spectrophotometer ND-1000 (NanoDrop®). Materials and Methods ! ! 136 2.2 DNA electrophoresis DNA electrophoresis was performed on agarose gels (Pronadisa, 8014) in TAE 1X buffer (40 mM Tris pH 7.6, 20 mM acetic acid and 1 mM EDTA) containing RedSafeTM (Intron Biotechnology, 21141) for nucleic acid staining. Loading Buffer (Takara) was added to the DNA samples prior to loading in the gel, and 1 kb DNA ladder (gTPbio, GTPBM0002) was used to identify the approximate size of the bands. Stained DNA was visualized using an ultraviolet transilluminator (Bio-Rad) and the Quantity One software. 2.3 RNA extraction RNA extracts were obtained from cells using RNA isolation kit (NZYtech, MB13402) according to manufacturer’s instructions. RNA concentration was quantified by measuring 260 nm absorbance using a NanoDrop ND-1000 spectrophotometer, and the quality of the sample was checked by running a 1% agarose gel. 2.4 Reverse transcription To obtain complementary DNA (cDNA), 1 µg RNA was treated with DNAse enzyme (Promega, M6101) for 30 minutes at 37ºC, according to manufacturer’s instructions. Then, 1µl RQ1 DNAse Stop Solution was added and incubated for 10 minutes at 65ºC to inactivate the DNAse. After that, the RNA was subjected to reverse transcription reaction using Quantitect Reverse Transcription kit (Qiagen, 205311) or Maxima H Minus First Strand cDNA Synthesis Kit (Fisher scientific, K1652) according to manufacturer’s instructions. 2.5 Quantitative PCR (qPCR) Quantitative PCR from cDNA was performed to check siRNA-mediated knockdown of several proteins, the study of mRNA expression, the appearance of specific splicing forms or analyse RNA accumulation. For this, iTaq Universal SYBR Green Supermix (Bio-Rad, 172-5124) was used following manufacturer’s instructions. DNA primers Materials and Methods ! 137 used for qPCR are listed in Table M4. qPCR was performed in an Applied Biosystem 7500 FAST Real-Time PCR system or ViiA7 Real-Time PCR 384-well plates. The comparative threshold cycle (Ct) method was used to determine relative transcripts levels (Bulletin 5279, Real-Time PCR Applications Guide, Bio-Rad), using β-actin as internal control. Expression levels were determined with the formula 2-ΔΔCt (Livak and Schmittgen, 2001). 2.6 Microarray To analyse RNA splicing genome wide from cells depleted or not for SF3B2 or CtIP in damaged (10 Gy) and untreated conditions the splicing microarray Transcriptome Arrays HTA & MTA (Affymetrix) was used. RNA was obtained using an RNA isolation kit, as explanied in section 2.3. Transcripts were marked with biotin and annealed to a glass slide where probes against the human transcriptome were present. This array used 6 million oligonucleotides of 25-mer positioned in the slide using the photolithography technique on a quartz surface. The resulting images were processed with Affymetrix GeneChip Command Console 2.0 programm to check if the samples, the labelling process and hybridation were correct. Then, data was analysed by Gabriel Gutiérrez Pozo using AltAnalyze software (Emig et al., 2010). 3. Protein analysis 3.1. Protein extraction 3.1.1. Protein extraction under denaturing conditions Unless specified otherwise, protein extraction under denaturing conditions was carried out on plates by lysing the cells in Laemmli 2X buffer (125 mM Tris‐HCl pH 6.8, 4% SDS and 20% glycerol [Sigma, 49770]) using a plastic cell scraper (Sarstedt, 83.1830). The cell suspension was then transferred into a microcentrifuge tube. Chromatin was sheared by passing the solution through a syringe with a 0.5x16 mm needle (BD Plastipak) at least 10 times to reduce sample viscosity. Protein concentration was determined by the measurement of absorbance at 280 nm using a NanoDrop ND-1000 spectophotometer. Materials and Methods ! ! 138 3.1.2. Protein extraction under native conditions For immunoprecipitations, protein extracts were prepared under native conditions to keep protein interactions. Thus, protease (Roche, 11873580001) and phosphatase inhibitors cocktails (Sigma, P0044) were added into the lysis buffer. Cells growing in 150 mm plates were placed on ice, rinsed with cold PBS and scraped off the dish after adding cold lysis buffer. Lysis buffer used for maintaining native conditions had no SDS and a low detergent concentration, but its composition varied according to the requirements of each experiment (see section 3.5 for details). Cell suspension was passed through a syringe with a 0.5x16 mm needle (BD Plastipak) at least 10 times. Supernatant was collected and quantified using Bradford assay. Bovine serum albumin (BSA) dilutions were used in order to create a standard curve used to determine the protein concentration. Samples containing 0, 1, 2, 5, 10 and 20 µg/ml BSA (Sigma, A4503) diluted in milli-Q water up to a final volume of 100 µl were prepared, and 900 µl Bradford reagent (Sigma, B6916) was added. Then, our samples were prepared in mili‐Q water up to a final volume of 100 µl with a dilution factor that gave a concentration in the range of BSA dilutions, and 900 µl Bradford reagent was added. Optical density was measured at a wavelength of 595 nm using a spectrophotometer DU® 800 (Beckman Coulter). In order to create the standard curve, BSA concentration was plotted on the X‐axis and its absorbance on the Y-axis. Then, the equation that represents the graph was determined by linear regression and used to calculate protein concentration in our samples based on their absorbance. 3.2. SDS-PAGE electrophoresis Proteins samples were resolved in 29:1 acrylamide:bis-acrylamide gels prepared at different concentrations for each experiment, according to the molecular weight of the proteins of interest. Different samples were diluted to obtain similar protein concentration and then SDS-PAGE loading sample buffer 4X (250 mM Tris-HCl pH 6.8, 8% SDS, 40% glycerol, 20% β-mercaptoethanol and bromophenol blue) was added to a final concentration of 1X. Samples were incubated at 100ºC for 5 minutes and then loaded into the gels. Prestained protein ladder (gTPbio, GTPBM003) was also loaded as a molecular weight marker. Electrophoresis was performed in a Mini- Materials and Methods ! 139 PROTEAN Tetra Cell (Bio-Rad) with Running Buffer (25 mM Tris-HCl pH 8.3, 190 mM glycine, and 0.1% SDS) at 120-150 V. 3.3. Western blot analysis After electrophoresis, proteins were wet-transferred using Mini Trans-Blot system (Bio-Rad) for 1.5 hour at 400 mA in Transfer Buffer (25 mM Tris-HCl pH 8.3, 190 mM glycine, 20% methanol and 0.1% SDS) into PVDF membranes (Immobilon-FL Millipore, IPFL00010) previously rinsed in cold methanol for 1 minute and equilibrated in Transfer Buffer. Membranes were incubated in Odyssey Blocking Buffer (LI-COR Biosciences, 927-40000) during at least 1 hour at room temperature to reduce background signal with the antibodies and then they were incubated overnight at 4ºC with the appropriate primary antibodies (Table M5) diluted in Odyssey Blocking buffer. Then, membranes were washed three times with 0.1% Tween-20 in TBS, and incubated with IRDye secondary antibodies (Table M6) diluted in Odyssey Blocking Buffer for 1 hour at room temperature protected from ambient light. Afterwards, they were washed again three times with 0.1% Tween-20 in TBS in dark and dried at least for 2 hours before scanning. Image acquisition was performed in Odyssey CLx Imaging System (LI-COR Biosciences) at two infrared wavelengths (700 and 800 nm) for differential imaging of both anti-rabbit and anti-mouse secondary antibodies at the same time. ImageStudio v.2.1 software was used for scanning and analysis of the images. Table M5. Primary antibodies used in this study. WB, western blotting. IF, immunofluorescence. IF-FISH, immunofluorescence and fluorescence in situ hibridation. SMART, Single Molecule Analysis of Resection Tracks. IP, immunoprecipitation. PLA, Proximity Ligation Assay. Target protein Source Supplier/Reference Application Dilution CtIP Mouse R. Baer (14.1) WB, PLA, IFFISH 1:500, 1:200, 1:50 CtIP Mouse Active Motif (61141) IF (laser lines) 1:500 Materials and Methods ! ! 140 Target protein Source Supplier/Reference Application Dilution CtIP Goat Santa cruz (Sc-228328) WB 1:500 CtIP Rabbit Novus Biological (NBP1- 33098) WB 1:1000 FLAG Mouse Sigma (F3165) WB, IF-FISH 1:1000, 1:250 RFP Rabbit Abcam (ab62341) WB 1:500 GFP Rabbit Santa Cruz (sc-8334) WB, IF 1:1000, 1:500 GFP Mouse Roche 11 814 460 001 WB 1:2000 α-tubulin Rabbit Sigma (T9026) WB 1:50000 β-actin Rabbit Abcam (ab8227) WB 1:50000 RPA32 Mouse Abcam (ab2175) IF 1:500 RPA Mouse Merck Millipore (MABE285) IF (laser lines) 1:500 γ-H2AX Rabbit Cell Signaling (2577L) IF 1:500 γ-H2AX Mouse Millipore (05-636) IF, IF-FISH 1:1000, 1:500 γ-H2AX Rabbit Cell Signaling Technology (9718) IF (laser lines) 1:500 γ-H2AX Mouse Abcam (ab26350) IF (laser lines) 1:500 BrdU Mouse Amersham (RPN202) SMART 1:1000 RAD51 Mouse Abcam; ab213 IF 1:1000 BRCA1 Mouse Santa Cruz (sc-6954) WB, IF, IP 1:1000, 1:200, 1:50 53BP1 Rabbit NB100-304, Novus WB, IF 1:1000,1:500 RIF1 Mouse Bethyl Laboratories (A300-568A) WB, IF 1:500, 1:200 CHK1 Mouse Santa cruz (sc-8408) WB 1:1000 CHK1 Ser345 Rabbit Cell signaling (2348) WB 1:1000 Materials and Methods ! 141 Target protein Source Supplier/Reference Application Dilution Fluorescein conjugated biotin Goat Vector Lab (SP-3040) IF-FISH 1:250 SF3B1 Mouse MBL (D221-3) WB, PLA, IP 1:1000; 1:50; 1:50 SF3B2 Rabbit Proteintech (10919-1-AP) WB, PLA, IP 1:1000; 1:50; 1:50 SF3B3 Goat Abcam (ab3608) PLA, IP 1:50; 1:50 SF3B3 Goat Novus Biological (NB100-1077) WB 1:1000 PIF1 Mouse Santa cruz (sc-48377) WB 1:500 ADAR2 Mouse Santa cruz (sc-73409) WB, IF 1:500, 1:200 ADAR1 Rabbit Sigma, (HPA003890) WB, IF, 1:1000, 1:200 RNAseH1 Rabbit Proteintech, (156061-AP) WB 1:1000 SETX Rabbit Bethyl Laboratories (A301-105A) WB 1:1000 Table M6. Secondary antibodies used in this study. WB, western blotting. IF, immunofluorescence. IF-FISH, immunofluorescence and fluorescence in situ hibridation. SMART, Single Molecule Analysis of Resection Tracks. Antibody Supplier/Reference Application Dilution IRDye 680RD goat anti-mouse IgG (H+L) LI-COR (926-68070) WB 1:5000- 1:25000 IRDye 800CW goat anti-rabbit IgG (H+L) LI-COR (926-32211) WB 1:5000- 1:25000 IRDye 800CW donkey anti-goat IgG (H+L) LI-COR (926-32214) WB 1:5000- 1:25000 Alexa Fluor 594 goat anti-mouse Invitrogen (A11032) IF, SMART 1:1000 Alexa Fluor 488 goat anti-rabbit Invitrogen (A11034) IF 1:1000 Alexa Fluor 647 goat anti-mouse Invitrogen (A21235) IF-FISH 1:250 Materials and Methods ! ! 142 Antibody Supplier/Reference Application Dilution Alexa Fluor 588 goat anti-rabbit Invitrogen (A11011) IF-FISH 1:250 Alexa Fluor 568 goat anti-mouse Invitrogen (A11037) IF-FISH 1:250 Alexa Fluor 594 goat anti-mouse Jackson Immunorese (115-585-003) IF (laser line) 1:100 Donkey anti-rabbit-FITC Jackson Immunoresearch (711- 096-152) IF (laser line) 1:100 3.4. Cell fractionation U2OS cells stably expressing the different PIF1 isoforms were subjected to nuclear/cytoplasm fractionation to analyse the distribution of both PIF1 variants in the different cellular compartments. For this, cells stably expressing the different PIF1 isoforms from an integrated plasmid were transfected with siPIF1 to remove endogenous PIF1 and 48 hours later they were irradiated or not with 10 Gy ionizing radiation. One hour post-irradiation, cells were harvested in Resuspension Buffer (10 mM HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol, 1 mM DTT and protease inhibitors [Roche, 11 873 580 001]). To disrupt cellular membranes, Triton X-100 was added to cell suspensions to a final concentration of 0.1%, and samples were incubated on ice for 8 minutes. Cells were then centrifuged at 4ºC at 1300 g for 5 minutes to separate the cytoplasmic fraction from the nuclei pellet. Nuclei were then washed once with Resuspension Buffer and incubated in 100 µl of this buffer containing 90 U/ml Benzonase (VWR, 70746-4) and 0.1 mg/ml BSA for 1 hour on ice. Cytoplasmic and nuclear fractions were boiled in 1x SDS-PAGE loading sample buffer, loaded into a 4-20% SDS-PAGE polyacrylamide gel (Bio-Rad, 456- 1094) and analysed by western blotting (see section 3.3). 3.5. Protein immunoprecipitation 3.5.1. Immunoprecipitation using magnetic GFP beads U2OS cells overexpressing GFP or GFP-tagged proteins were seeded and treated Materials and Methods ! 143 accordingly to the requirements of the experiment. Protein extracts were prepared under native conditions (see section 3.1.2.) using 150 µl Lysis Buffer (50 mM Tris- HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, 0.2% Triton X-100, 1X protease inhibitors and 1X phosphatase inhibitor cocktail 3) containing 90 U/ml Benzonase. Protein extracts (1 mg) were mixed with 30 µl of magnetic anti-GFP resin (Chromotek, Gtm- 200) pre-equilibrated in the same buffer and incubated overnight at 4ºC with gentle rocking. Beads were then washed three times with Lysis Buffer and precipitated proteins were recovered by incubating the beads for 5 minutes at 100ºC in Lysis Buffer containing protein loading buffer. Finally, proteins were resolved by SDS-PAG and analysed by western blotting as previously described (see section 3.2 and 3.3) 3.5.2. Immunoprecipitation using protein A resin To immunoprecipitate endogenous CtIP, BRCA1 or SF3B2, U2OS cells were seeded in 150 mm plates and irradiated (10 Gy) when indicated. Proteins extracts were prepared under native conditions (see paragraph 3.1.2.) using Lysis Buffer (50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA, 0.5% NP-40, protease inhibitors and phosphatase inhibitor cocktail 3. To degrade DNA, 90 U/ml Benzonase was added to protein extracts and incubated 30 minutes on ice. 1 mg of protein extracts was then precleared to reduce non-specific binding by incubation with 20 µl pre-equilibrated magnetic protein A Dynabeads (Novex, 10002D) under gentle agitation at 4ºC for 30 minutes. Precleared samples were then incubated with 1 µg of the indicated antibody (Table M5) or 5 µg rabbit IgG (Sigma, I8140) or mouse IgG (Sigma, I8765) as a control for 1.5 hour at 4ºC. Dynabeads pre-equilibrated in the same buffer were added afterwards and the mixture was incubated over night at 4ºC with gentle agitation. The following day, beads were washed three times with Lysis Buffer, and the precipitate was eluted in 40 µl Lysis Buffer with 1X protein loading buffer to elute proteins complexes bound to the resin. Precipitated proteins were resolved in SDS-PAGE as described in sections 3.2 and 3.3. 3.6. Cycloheximide chase assay In order to analyse the effect of SF3B complex in CtIP and BRCA1 stability, cells were exposed to cycloheximide, an inhibitor of protein biosynthesis that prevents Materials and Methods ! ! 144 translational elongation. U2OS, FLAG, FLAG-SF3B2 or FLAG-SF3B2 S289A cells were depleted as indicated for SF3B1, SF3B2, SF3B3 or CtIP by shRNAs or siRNAs, and a non-target sequence was used as control (shNT and siNT). Then, cells were treated with fresh medium containing 150 µg/ml cycloheximide (Sigma, C7698) 48 hours post infection or transfection. Cells were collected 2, 4 and 8 hours after the cycloheximide addition and protein samples were prepared as described in section 3.1.1. Protein extracts from cells before the treatment were used as control of the initial amount of proteins. Finally, samples were resolved by SDS-PAGE, analysed by immunoblotting and protein quantification was made using Image Studio v.2.1 software. CtIP and BRCA1 expression was calculated normalizing the amount of these proteins by α-tubulin or β-actin signal. 4. Cell biology procedures. 4.1. Immunofluorescence Cells were knocked-down for different proteins using shRNA or siRNA. After 48 hours, cells growing on coverslips were irradiated (10 Gy), treated with etoposide (10 µM) or camptothecin (10 µM) or mock treated and incubated 1 hour (for RPA, 53BP1 and RIF1) or 3 hours (for RAD51) to allow foci formation. Then, cells were washed once with ice-cold PBS and collected. For RPA, BRCA1 and RAD51 foci, coverslips were treated for 5 minutes on ice with Preextraction Buffer (25 mM Tris-HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, 3 mM MgCl2, 300 mM sucrose and 0.2% Triton X-100) or Pre-extraction Buffer 2 (10mM PIPES pH 6.8, 300 mM sucrose, 50mM Nacl, 3mM EDTA, 25X proteinase inhibitor and 0.5% Triton X-100) in the case of BRCA1 foci to remove the nucleoplasmic and cytoplasmic proteins, leaving behind the chromatin-bound and matrix-associated proteins. Cells were then washed once with cold PBS, fixed with 4% paraformaldehyde (w/v) in PBS for 15 minutes on ice. For 53BP1 and RIF1 foci, cells growing on coverslips were treated for 10 minutes on ice with metanol and 30 seconds with acetone. In all cases, after the fixation, cells were washed three times with PBS and Materials and Methods ! 145 blocked for at least 1 hour with Blocking Solution (5% FBS diluted in PBS). Afterwards, cells were incubated for at least 2 hours at room temperature or overnight at 4ºC with the adequate primary antibodies (Table M5) diluted in Blocking Solution. Coverslips were washed three times with PBS, incubated with secondary antibodies (Table M6) for 1 hour at room temperature in the dark, washed again with PBS and dehydrated using ethanol if cells were not expressing GFP. Finally, cells were mounted with Vectashield Antifade mounting medium (H-1200, Vector Laboratories) containing 4’,6-diamidino-2-phenylindole (DAPI). At least 200 cells per sample and per experiment were analysed using Leica Fluorescence microscope and 63X objetive. The analysis of the number and intensity of 53BP1 and RIF1 foci formation was performed automatically using MetaMorph software. 4.2. Proximity ligation assay (PLA) To study if the different pairs of proteins localize close to each other in cells, a proximity ligation assay was performed using Duolink PLA kit (Sigma, DUO92008, DUO92002, DUO92004, DUO92003) according to the manufacturer’s protocol. To do so, U2OS cells growing on coverslips were irradiated (10 Gy) or not and 1 hour later they were fixed with cold metanol for 10 minutes followed by cold acetone for 30 seconds. Then, cells were washed three times with cold PBS and blocked with Duolink PLA blocking solution for 30 minutes at 37ºC. To investigate if the interaction between CtIP and SF3B colocalize at DSB, we used cells stably expressing a GFP-tagged version of MDC1, a protein that is recruited early to DNA damage. These cells were fixed using paraformaldehyde diluted in 4% PBS for 15 minutes at room temperature and treated with triton 0.1% in PBS for another 15 minutes. Cells were then washed three times with cold PBS and blocked with Duolink PLA blocking solution for 30 minutes at 37ºC in a wet chamber. In both cases, samples were then incubated with indicated primary antibodies (see Table M5) diluted in Duolink PLA antibody overnight at 4ºC, and incubated afterwards with the combination of MINUS and PLUS secondary PLA probes (Sigma, DUO92002, DUO92004, DUO92003) for 1 hour at 37ºC in a wet chamber. Detection was carried out with the Duolink Detection kit Red (Sigma, DUO92008). Afterwards, coverslips were dried and mounted into slides using Vectashield containing DAPI