Non-canonical DNA structures in Double Strand Break repair

DOCTORAL THESIS Non-canonical DNA structures in Double Strand Break repair MªRosa Camarillo Daza Non-canonical DNA structures in Double Strand Break repair CABIMER Memoria presentada por: MªRosa Camarillo Daza Para optar al grado de Doctora en Biología Molecular, Biomedicina e Investigación Clínica 2023 Director/Tutor Pablo Huertas Sánchez Directora Sonia Jimeno González Abstract DNA double-strand breaks (DSBs) pose a serious threat to genome stability, so their proper repair is crucial for cell viability. To repair them, the cell has evolved two main alternative mechanisms, non-homologous end-joining (NHEJ) and homologous recombination (HR). The correct choice between these two pathways is a key point to obtain a faithful restoration of the broken DNA sequence. One of best-known events regulating this decision is the generation of 3’ single-stranded DNA overhangs in a process known as DNA resection. Many different factors are involved in this process such as cell stemness, chromatin status or DNA conformation. In this Thesis, we investigated the role of G-quadruplexes, a noncanonical DNA structure, in DNA resection. First, we studied the impact of G-quadruplexes in this process and showed that DNA resection is impaired in the presence of these structures, both in vivo and in vitro. In addition, we demonstrated that PIF1α helicase activity unwinding G-quadruplexes promotes resection over these structures and that this process requires other factors such as BRCA1 and TOP2β. Furthermore, we took advantage of the key role of factors modulating DNA resection to regulate homologous recombination efficiency to search for new compounds that could affect CRISPR-Cas9 mediated homologous recombination efficiency. Resumen Los cortes que afectan a las dos cadenas del ADN suponen una seria amenaza a la estabilidad del genoma por lo que su reparación es crucial para la viabilidad celular. Para repararlos, la célula ha desarrollado principalmente dos mecanismos, la unión de extremos no homólogos (NHEJ) y la recombinación homóloga (HR). La correcta elección entre ambas rutas es un punto clave para conseguir restauran fielmente la secuencia de ADN dañada. Uno de los eventos más conocidos que regulan esta decisión es la generación de fibras de ADN de cadena sencilla colgantes en el extremo 3’ del corte de ADN en un proceso conocido como resección del ADN. Existen muchos factores involucrados en este proceso tal y como el grado de diferenciación de la célula, el estado de la cromatina o la conformación del ADN. En esta Tesis, se ha investigado el papel de los G-cuadruplexes, una estructura no canónica del ADN, en la resección del ADN. En primer lugar, se estudió el impacto de los G-cuadruplexes en este proceso y se ha mostrado que la resección del ADN se encuentra perjudicada en presencia de estas estructuras, tanto in vivo como in vitro. Además, se ha demostrado que la actividad helicasa de PIF1α en la resolución de los G- cuadruplexes promueve la resección a través de estas estructuras y que este proceso requiere de otros factores como BRCA1 y TOP2β. Asimismo, aprovechamos la importancia de la modulación de la resección del ADN por distintos factores en la regulación la eficiencia de la recombinación homóloga para buscar new compuestos que puedan influir en la eficiencia de la recombinación homóloga mediada por CRISPR-Cas9. Indexes Indexes 12  Tables Index  Table R1. Candidate compounds with a mean RPA foci formation efficiency over 2-fold or under 0.85-fold ............................................................................................................... 85 Table R2. Selected candidates description ........................................................................ 87 Table R3. Validated compounds selected for their impact on CRSPR-Cas9 efficiency ... 92 Table M1. Cell lines used in this Thesis .......................................................................... 146 Table M2. Plasmids used in this Thesis .......................................................................... 146 Table M3. siRNAs used in this Thesis ............................................................................ 147 Table M4. Primers used in this Thesis ............................................................................. 148 Table M5. Primary antibodies used in this Thesis ..................................................... 149 Table M6. Secondary antibodies used in this Thesis ....................................................... 150 Indexes 13  ABBREVIATIONS ROS Reactive Oxidative Species WRN Werner helicase UV Ultraviolet ligh t G4 G-quadruplex IR Ionizing Radiation PQS Putative Quadruplex Sequence SSB Single Strand Break ChIP-seq Chromatin- Immunoprecipitation highthroughput DNA sequencing BER Base Excision Repair SF DNA helicases superfamilies NER Nucleotide Excision Repai r PIF1 Petite Integration Frequency 1 MMR Mismatch Repai r BIR Break-Induced Replication DSB Double Strand Break TOPOI Topoisomerase I DDR DNA Damage Response TOPOII Topoisomerase II DNA-PK DNA-dependent protein kinase TOP1 Topoisomerase IB-1 PIKKs Phosphoinositide 3-Kinase (PI3K)-related Kinases family TOP2 Topoisomerase IIA-2 IRIF Ionizing Radiation Induced Foci TOP2cc TOP2 cleavage complex NHEJ Non-Homologous End Joining CRISPR Clustered Regularly Interspaced Palindromic Repeats HR Homologous Recombination Cas CRISPR-associated genes LIG4 Ligase IV crRNA CRISPR RNA CtIP CTBP Interacting Protein gRNA Guide RNA PARP1 Poly [ADP-Ribose] Polymerase 1 tracrRNA Trans-activating crRNA SSA Single-Strand Annealing PAM Protospacer Adjacent Motive SSTR Single Stranded Template Repair KO Knock-Out Indels Insertions and Deletions KI Knock-IN CDKs Cyclin-Dependent K inases ssODN ssDNA oligonucleotides ssDNA Single stranded DNA CAT-R Color Assay Tracing- Repai r BRCA1 Breast Cancer 1 BSA Bovine Serum Albumin EXO1 Exonuclease 1 PBS-T PBS-Tween Indexes 14  BLM BLOOM helicase BrdU 5-Bromo-2 Deoxyuridine RPA Replication Protein A SSR See-Saw Reporte r   I. Introduction    Introduction  19 1. DNA damage and genome stability The genetic information contained in the DNA is in charge of guiding all the processes that determine the behaviour of the cell and the characteristics of every given organism. Thus, it is of utmost importance the faithful preservation of this information and its proper transmission to the progeny. However, there is a large body of evidence demonstrating the constant exposure of the genomic eukaryotic DNA to genotoxic stress that induces numerous DNA insults. It is estimated that everyday each human cell suffers roughly 105 spontaneous DNA lesions that alter its DNA sequence and structure (Lam, 2022). This damage needs to be accurately repaired to ensure the stability of the genome and, for that matter, the cell has evolved multiple important repair pathways to solve these threats. The accumulation of mutations in genes responsible for sensing and repairing DNA damage leads to lesion accumulation, loss of genetic information, higher mutagenic rates and uncontrolled gene expression. These characteristics are considered hallmarks of cancer development and is the reason why DNA damage is a key event in neoplastic transformation (Negrini et al., 2010). There are multiple sources of DNA lesions but, mainly, they can arise endogenously during DNA transactions or be induced by exogenous agents (Fig.I1). Endogenous damage can be caused by the very process of DNA replication during cell division, for example, due to the incorrect nucleotide addition (the so-called mismatches), insertions or deletions during DNA polymerase activity (Budzowska and Kanaar, 2009). In addition, base alkylation, methylation, reactive nitrogen species and reactive oxidative species (ROS) can be generated as by-products during the normal functioning of the cell and interact with the DNA molecule altering its composition (Chatterjee and Walker, 2017). Exogenous damage, on the other hand, is induced when environmental, physical, and chemical agents alter the DNA. Examples of this type of DNA lesions include alkylating and crosslinking agents, ultraviolet light (UV), ionizing irradiation (IR), chemotherapeutic agents and other environmental stresses such as temperature, cigarette smoke or contamination (Mehta and Haber, 2014). Introduction 20  Figura I1Each of these distinct DNA threats impacts differently on the DNA molecule and, hence, encompasses diverse consequences in the genome (Chatterjee and Walker, 2017; Hoeijmakers, 2001). Most of the lesions that alter the DNA sequence such as mismatches, abasic sites, bulky adducts, pyrimidine dimers, covalent bounds between nucleotides or single strand breaks (SSB) usually affect just one of the DNA strands and can be repaired generally by BER (Base Excision Repair), NER (Nucleotide Excision Repair) or MMR (Mismatch Repair) (Fig.I1). Although the mechanism of repair in which the cell incurs to mend these alterations is specific to each kind of damage, the pathways followed when only one strand is disturbed share some similarities and can be safely carried out since the information from the complementary strand is present to act as a template. Basically, once the damage is recognized by specific proteins, the affected base, nucleotide or oligonucleotide is excised and replaced with the correct sequence (Caldecott, 2022). A different situation needs to be confronted when the damage affects both strands of the DNA since there is no longer an undamaged template to copy the information from. This is the Figure I1. Types of DNA damage DNA damage can be induced by several sources (exogenous and/or endogenous) that generate different kinds of lesions in the DNA sequence altering either one or both DNA strands. Each alteration requires a specific mechanism to be repaired. Modified from (Hoeijmakers, 2001). Ionizing radiation Oxygen radicals Alkylating agents Spontaneous reactions Ultra-violet light DNA crosslinkers Replication errors Ionizing radiation Genotoxic agents Meiosis Uracil Abasic site 8-Oxoguanine Altered base Single-strand break Bulky adduct Intrastrand crosslink (6-4) photoproduct Ciclobutane pyrimidine dimer A-G mismatch T-C mismatch Insertion Deletion Interstrand crosslink Double-strand break (DSB) Base Excision Repair (BER) Nucleotide Excision Repair (NER) MIsmatch Repair (MMR) Non-Homolohous End-Joining (NHEJ) Homologous Recombination (HR) DNA damaging sources DNA damage types DNA repair mechanisms Introduction  21 scenario for double-strand breaks (DSB), which implies a more complex and coordinated pathway of multiple proteins and cellular events to accurately restore the sequence (Ceccaldi et al., 2016). 1.1 DNA Double Strand Breaks (DSBs) DNA double strand breaks are considered one of the most cytotoxic and difficult to repair lesion on the DNA. Only one of these DSBs is sufficient to induce cell death or terminal cell cycle arrest. In all cases, the generation of a DSB implies the break of both strands of the double DNA helix through the cleavage of the phosphodiester linkages of the backbone. However, DSBs differ widely in their structure and in the mechanisms of their generation. While restriction endonucleases give rise to blunt or staggered clean ends, clastogens, such as radiomimetic chemicals, display a much more complex generation of DSBs, chemically modifying the strands termini (Bignold, 2009). Among the factors inducing DSBs, IR is a well-known one. IR targets DNA in two different ways. First, it can directly attack the DNA backbone inducing, mostly, DSBs. In this process IR might also generate water radiolysis which can result in the formation of reactive oxidative species (ROS) (Santivasi and Xia, 2014). These radicals can then attack macromolecules in the cell, such as DNA, forming single strand breaks (SSBs) that can turn into DSBs after DNA replication. Anticancer chemoterapeutic agents are also considered an important source of DSBs. Some examples are DNA-alkylating agents, cross-linking and radiomimetic compounds (Chen and Stubbe, 2005; Wyrobek et al., 2005). Finally, unusual chromatic structures such as G-quadruplexes, secondary hairpin loops and R-loops are another important source of genomic instability that leads to DSBs generation due to the stalling of the replication fork progression and the increased exposure of the DNA sequence to damaging agents (Aguilera and García-Muse, 2013; Lopes et al., 2011; Maffia et al., 2020; Rondón and Aguilera, 2019). However, DSBs are also necessary for some important cell biology processes (Oster and Aqeilan, 2020). First, this kind of damage is a great source of genetic variability due to the insertions, deletions and large chromosomal rearrangements that its unfaithful repair might generate (Hoff et al., 2018). In addition, programmed DSBs are also induced during meiosis to allow gene cross over and exchange to achieve an accurate chromosome Introduction 28  resection to undergo SSA has turned this pathway into one of the most common read-out of DNA resection. SSA requires longer resected DNA and longer homology sequences than alt-EJ and the annealing of the two complementary sequences is resolved by digesting the single stranded tails and, therefore, leading to loss of genetic information (Bhargava et al., 2016). Recently, since the upswing of genome editing, another RAD51-independent DSB repair mechanism has been described. This pathway uses a ssDNA template to repair a sitespecific DSB and is known as single stranded template repair (SSTR) (Gallagher and Haber, 2021). There is not much information about this process and most of it comes from studies using Cas9 in human cells. 1.4 DSB repair pathway choice Every time the cell faces a DSB, the DDR must activate a cascade of processes to correctly repair the damage. The choice of the pathway activated in response to the DSB is heavily regulated and determines getting either a faithful repair or an altered DNA sequence. This decision is influenced by several factors that the DDR considers to choose the correct repair mechanism. NHEJ is the fastest and predominant pathway to repair DSBs in human cells. As already mentioned, NHEJ is not limited to cell cycle phases and, therefore, can be employed whenever is needed. However, it is considered to be an error-prone pathway when the ends of the breaks need some processing (Burma et al., 2006). For example, when the DSB generated presents protein adducts, short overhangs or altered chemical structures they need to be removed prior to ligation of the ends. Then, the gaps generated after the ligation are filled by NHEJ associated polymerases. In these processes, some insertions and deletions (indels) might be generated originating and unfaithful repair (Rodgers and McVey, 2016). On the other hand, HR requires longer periods of time to repair the break and is limited to S/G2 phases of cell cycle, but the outcome is usually more accurate than NHEJ. Nevertheless, the accuracy of the repair by HR depends on the use of the correct template to restore the damaged DNA sequence. If the sister chromatid of the DNA sequence is present during the repair, its use as a template is promoted and the chances of having a correct repair are higher (Ciccia and Elledge, 2010; Heyer et al., 2010). Since the sister chromatid is only available when the DNA has been duplicated in the G2/S phases, it Introduction  29 is clearly explained why HR is restricted to these phases. In the case that this template is not present, or the HR is misadjusted, there are other substrates that can be used instead but, usually, with more deleterious outcomes. For example, the utilization of the homologous chromosome leads to non-allelic recombination and, therefore, loss of heterozygosity. In addition, non-identical repetitive sequences are another source of homology for HR that might generate an incorrect repair of the damage due to indels (Aguilera and Gómez- González, 2008; Rodgers and McVey, 2016). Taking into account all the detrimental consequences for the cell in case of undergoing the wrong repair pathway, there are many factors regulating the choice of the correct one. First and previously mentioned, cell cycle is critical in setting the right pathway. Cyclin-dependent kinases (CDKs) activated throughout the cell cycle have a key role in triggering one pathway or the other (Brandsma and van Gent, 2012; Dasika et al., 2000; Scully et al., 2019). Indeed, when a DSB is induced and G2/S phases are initiated, CDKs phosphorylates CtIP promoting DNA resection. By contrast, during G1 phase, proteasome degrades CtIP skewing the repair balance towards NHEJ (Brandsma and van Gent, 2012). Also, the cell type facing the DSB influences the choice. Proliferative cells, such as stem cells, promote HR to repair DSBs (Checa-Rodríguez et al., 2020; Mujoo et al., 2017; Swift et al., 2021) in opposition to post-mitotic or more differentiated cells which usually applies NHEJ. Other cell features influencing DNA repair pathway choice are epigenetic changes (Fernandez et al., 2021; Min et al., 2022), complexity of the break (Brandsma and van Gent, 2012), chromatin structure (Fernandez et al., 2021) and even the nuclear position of the damaged DNA (DNA in the periphery is prone to repair DSB by NHEJ over HR) (Lemaître et al., 2014). The deregulation of any of the elements described might result in genome instability and disease (Hosoya and Miyagawa, 2014; Majidinia and Yousefi, 2017). That is why DSB repair pathway choice needs to be a fine-tuned process. Despite all the factors already known to affect this process, the most decisive step guiding this resolution is DNA resection activation since its inhibition by KU70/80 binding or 53BP1, for example, guides the repair through NHEJ and inhibits HR (Symington and Gautier, 2011). Introduction 30  1.5 DNA resection DNA resection consists of the nucleolytic degradation of the 5’-terminated broken ends in the 5’-3’ direction, leaving 3’ ssDNA overhangs of up to 1 kb (Fig.I4). This process initiates and commits DSB repair to the HR pathway. Once resection is performed, ssDNA is rapidly coated by Replication Protein A (RPA) and homology-strand search and invasion by RAD51 is initiated. Although there are more steps downstream RAD51 strand-invasion to accomplish HR efficiently, they are not needed to be described in this Thesis and an extensive review on the matter can be found elsewhere (Krejci et al., 2012; San Filippo et al., 2008; Sun et al., 2020; Wright et al., 2018). DNA resection can be divided in two steps: short-range resection and long-range resection (Fig. I4). First, DNA resection is initiated by DNA recognition by MRN complex which is kept inactive until MRE11 activity is licensed by CtIP (Takeda et al., 2007). In the MRN complex, MRE11 is the core component and exhibits different activities, including 3’-5’ exonuclease and endonuclease activity, needed to perform the first short-range resection step (Garcia et al., 2011; Reginato and Cejka, 2020). Paradoxically, 3’ssDNA generation during resection is described to take place in the 5’-3’ direction, thus the resecting activity of MRE11 is proposed to require a two-steps bidirectional mechanism and the association with CtIP (Cannavo et al., 2019; Garcia et al., 2011). MRN-CtIP complex first makes a nick in the prone-to-be resected strand several using MRE11 endonuclease activity and then, the 3’-5’ exonuclease digest the DNA towards the DSB end producing 3’ long ssDNA tails (Garcia et al., 2011) (Fig. I4). CtIP is considered a critical additional cofactor of the MRN complex and, upon its phosphorylation by CDKs in a cell-cycle dependent manner, it regulates the endonucleolytic activity of MRE11 (Huertas and Jackson, 2009; You and Bailis, 2010; Zdravković et al., 2021). In addition, in humans, CtIP phosphorylation by ATM is required to elicit DNA resection (Makharashvili and Paull, 2015; You and Bailis, 2010). The complex originated by the binding of CtIP to MRE11 is considered to be the minimal requirement for DNA resection with CtIP being the bestknown regulator of this process and, therefore, of DSB repair pathway choice. Apart from CtIP, there are additional accessory factors involved in this first short-range resection such as the tumour suppressor Breast Cancer 1 (BRCA1). BRCA1 role in DNA resection is not Introduction  31 very understood but it is thought to help recruiting CtIP to DSBs by its interaction with both CtIP and the MRN complex (Densham and Morris, 2019; Rosen and Walle, 2013) and, therefore, promoting DNA resection. Moreover, BRCA1 has also been described to accelerate DNA resection processivity by binding MRN-CtIP complex (Cruz-García et al., 2014). Finally, it has been proposed that BRCA1 is required to overcome the anti-resection activity of the Shieldin complex (Dev et al., 2018). BRCA1also plays a role in several steps of DNA repair and cell cycle checkpoints activation, so it is important to maintain genomic integrity. Mutated versions of this gene are present in high prevalence in hereditary breast and ovarian cancer (Jiang and Greenberg, 2015; Rosen and Walle, 2013). Despite the importance of BRCA1, not all cells lacking a functional form of this protein undergo apoptosis. The initial short resection by MRN-CtIP is followed by a more extensive resection of the 3’ ends in what is known as long-range resection (Fig. I4). This second step is catalysed by two alternative pathways in human cells depending on the exonuclease used to resect the DNA. One of the two partially redundant mechanism is the 5’−3’ exonuclease 1 (EXO1) while the other mechanism involves DNA2 nuclease working in concert with BLOOM helicase (BLM) or Werner syndrome protein (WRN) (Huertas, 2010; Nimonkar et al., 2011). In both cases, the result is the generation of ~1Kb long ssDNA tracks that will be covered by RPA and will facilitate the homology search needed for homologous recombination (San Filippo et al., 2008). This quick coating of the DNA fiber by RPA makes it the golden read-out of resection and its quantification allows the study of DNA resection efficiency (Raderschall et al., 1999). The generation of this extensive ssDNA tails remove chemical adducts from the 5’ ends and displace KU70/80 heterodimers bound to the DNA, inhibiting NHEJ (Huertas, 2010). Being DNA resection such a key point in the balance between HR and NHEJ, there has been extensive study of the factor affecting and regulating DNA processing like the presence of DNA secondary structures or the presence of protein adducts covalently bound to the DNA. Introduction 32  Figure I4. DNA end resection DNA resection is triggered by the recruitment of CtIP and BRCA1 by the sensor protein comple x MRN to DSBs. This process consists of two steps, a short-range and a long-range DNA resection. Short-range resection is initiated by the endonucleolytic cleavage of the 5’-ended DNA strand by MRE11. Then, the exonuclease 3’-5’ activity of the MRN complex creates short 3’ overhangs o f ssDNA that are further resected in the long-range resection by either DNA2 and BLM (left) or EXO1 (right) in a 5’-3’ polarity. The 3’ssDNA overhangs are coated by RPA and facilitate the HR pathway. Alternatively, the first short resection can be used to undergo the alternative end joining (alt-EJ) pathway (dashed arrow). DSB 5’ 3’ 5 ’ 3 ’ CtIP CtIP BRCA1 BRCA1 BRCA1 BRCA1 CtIP CtIP MRN MRN MRN MRN MRN MRN DNA2 EXO1 RPA RPA Short-range DNA resection L ong-range DNA resection Alternative end-joining Introduction  33 2. DNA non-canonical secondary structures in DNA biology DNA is usually depicted in its canonical form, a right-handed B-DNA configuration of two anti-parallel linear strands that twist around the same axis giving rise to a double helix with a minor groove and a major groove (Fig.I5). This conformation is the most Figure I5. Schematic representation of non-canonical DNA structures On top left, Canonical B-form duplex DNA structure. On top right, representation of an R-loop structure formed by an RNA-DNA hybrid and the consequent displacement of the complementary DNA strand. On the bottom left, the representation of a putative quadruplex sequence (PQS) and the consequent G-quadruplex (G4) can be observed. On the bottom right, different conformations of G4s. Intramolecular (generated on one strand of the DNA) or intermolecular (generated by several strands of DNA). In each case, they can be parallel or antip arallel depending on the orientation of the strands. In the intermolecular box, a representation of an anti-parallel dimeric (two different strands) and a parallel tetrameric (four different strands) structures are represented. CANONICAL DNA B-form Minor groove Major groove ssDNA RNA ssDNA R-loops Anti-parallel Parallel Intramolecular Intermolecular Anti-parallel Parallel G-quadruplex G-quartet GG GG GG GG PQS (Dimeric) (Tetrameric) Introduction 34  common form in which DNA can be found in the cell, however, during different cell processes DNA can adopt others that are considered non canonical. The diversity of the different conformations of DNA is attributed to the different bonds and backbone configurations that the double helix can adopt. Their presence is enriched in areas with a higher rate of repetitive sequences such as direct repeats or short-handed repeats (Bansal et al., 2022). For example, Z-DNA requires an alternating purine-pyrimidine/GC rich sequences and cruciforms require inverted repeats to form. The presence of these non-canonical DNA structures is quite abundant in the human genome and has been associated to DNA damage, endogenous genome instability and disease (Pavlova et al., 2021; Puget et al., 2019; Rinaldi et al., 2021; Rondón and Aguilera, 2019). Interestingly, many hotspots for breakage sites and regulators of cancer-associated genes are located near to non-B DNA structures. One example is the promoter of the c- MYC gene that is abundant in guanine rich sequences that originate some of these structures (Siddiqui-Jain et al., 2002). Indeed, some diseases, such as Huntington’s disease or fragile X syndrome, are caused by trinucleotide repeats and expansion of repetitive sequences that weaken the genome and increase gene instability (Khristich and Mirkin, 2020; Möncke- Buchner et al., 2002). The most striking fact about non-canonical DNA structures is that, despite being considered an important source of genome instability, they are strongly conserved across evolution suggesting an additional role for them in cell biology. In fact, in the last years, there has been an upsurge of many studies describing the effects of non-canonical DNA structures in DNA metabolism. Apparently, these structures regulate many processes critical for cell biology not only by acting as physical barriers to the processivity of DNA machineries but also by operating as scaffolds to recruit proteins where needed (Arab et al., 2019; Boque-Sastre et al., 2015; De Magis et al., 2020). Among the wide variety of non-canonical conformations of DNA, R-loops and G- quadruplexes (G4s) are the most studied ones in relation with DNA metabolism. This is due to the broad scenarios where their presence influences the activity of the different proteins involved. R-loops are three-stranded non canonical structures formed by an RNA-DNA hybrid plus a displaced ssDNA (Fig.I5). They are expected to occur naturally as a result of Introduction  35 negative supercoiling during physiological cell procedures, such as replication and transcription. The progression of the transcription bubble and the synthesis of the DNA lagging strand promotes its formation in cis spanning lengths from 0.1 to 2 kb (Bader and Bushell, 2020; Malig et al., 2020; Massé et al., 1997; Roy and Lieber, 2009). Recently, there has been a lot of controversy regarding the role of R-loops in the cell biology. First, R-loops are described as inducers of DNA damage and genome instability since their occurrences correlate with blocking of the transcription bubble and DNA replication and cell stress (Cristini et al., 2019; Huertas and Aguilera, 2003; Lang et al., 2017). Indeed, they are emerging as potential sources of several diseases, such as cancer and neurodegeneration (Camarillo et al., 2021; Chen et al., 2004; Crow et al., 2006; García- Muse and Aguilera, 2019) However, the physiological presence of these DNA structures also function as regulators of class switch recombination of immunoglobulins (Yu et al., 2003), telomere maintenance (Tan et al., 2020), gene regulation (Grunseich et al., 2018; Sun et al., 2013) or in double-strand breaks (DSBs) repair (S. Liu et al., 2021). The turning point between these two opposite faces of R-loops relies in the efficiency of their regulation. In fact, their turnover is tightly controlled by different evolutionary conserved pathways. Among the factors regulating R-loops levels, RNA unwinding or degrading enzymes are the most known ones as it is the case of SETX or RNase H (Cerritelli and Crouch, 2009; Cohen et al., 2018). Nevertheless, other RNA processing factor involved in splicing, topoisomerases and different DNA repair pathways are also in charge of resolving these structures (Hatchi et al., 2015; Li et al., 2007; Massé et al., 1997). The deep study of R-loops has uncovered their coupled formation with another relevant non-canonical DNA secondary structure, the G4s. Indeed, the mapping of R-loops in the genome shows an overlapping location with G4 (Kuznetsov et al., 2018). As well as R-loops, G4s have dual roles in genome biology acting both as damage inducers and regulators of DNA metabolism. Introduction 36  2.1 G-quadruplexes (G4s) structures, mapping and characterization G4s are guanine rich sequences assembled by Hoogsteen hydrogen bonding into squared planar arrangements, known as G-quartets, that stack on top of each other generating these four-stranded helical structures (Fig.I5). A large variety of DNA (DNA G4) and RNA (RNA G4) sequences that can assembly into G4s have been described. Basically, all G4s contain at least four islands of at least two guanines separated by random nucleotides (loops) (Spiegel et al., 2020). Depending on the orientation or the nature of the strands conforming the G4, different types can be observed (Spiegel et al., 2020). In first place, depending on the nature of the nucleic acid adopting the G4 structure, there are DNA G4s and RNA G4s. In addition, depending on the number of strands involved in their formation, we can differentiate intermolecular (multimeric) G4s and intramolecular (monomeric) G4s (Fig. I5). Intermolecular structures can be dimeric, trimeric or tetrameric depending on the number of strands involved and these strands can be all DNA, RNA of a mix of RNA and DNA strands. In comparison, intramolecular G4s are considered more stable and with faster conformation dynamics than intermolecular G4s. Finally, the nucleotide strand/s can fold into parallel (all four G-strands in the same direction) or antiparallel (all adjacent G-strands antiparallel to each other) conformation (Frasson et al., 2022; Kim et al., 2003; Reshetnikov et al., 2010; Yang, 2019) (Fig. I5). The conformation that a sequence with potential to form a G4 (Putative Quadruplex Sequences, PQS) can adopt is difficult to predict and usually involves some experimental structure determination. G4s were first determined in vitro using a G4-specific antibody to visualise G4s in telomeres (Schaffitzel et al., 2001), but they also occur naturally in vivo. Certainly, several studies have mapped roughly 700000 G4s alongside the whole genome with special emphasis in determined regions such as promoters, splicing sites, telomeres and 5’ untranslated regions (Chambers et al., 2015; Huppert and Balasubramanian, 2007, 2005; Murat and Balasubramanian, 2014). Certainly, the presence of this structures in such key locations has increased the efforts to fully characterise them. To do so, several strategies have emerged. Among them, the most extended one is the use of the BG4 antibody that binds a wide variety of G4 structures with high affinity (Biffi et al., 2013). This antibody has been employed in many studies to map endogenous G4 structures along the genome and Introduction  37 to visualize them under the microscope. Although the BG4 antibody seems to recognize with high selectivity different G4s structures, it remains elusive if this antibody can bind a single G4 or if, on the contrary, it needs a high density to detect them. In addition, the number of G4s detected by BG4 was lower than the expected number predicted by nextgeneration sequencing (Biffi et al., 2013; Chambers et al., 2015). However, this could be explained by the dynamic and transient folding and unfolding of these structures in vivo. Other methods for G4s detection include Chromatin-Immunoprecipitation high-throughput DNA sequencing (ChIP-seq) using antibodies against proteins known to bind these structures (Gray et al., 2014) or proteins recognizing the DNA damage induced by G4s ligands (Rodriguez et al., 2012; Xu et al., 2017). As expected, all these approaches showed an enrichment of G4s in the aforementioned places, such as promoters and telomeres. The presence of G4s in these key places supports their regulatory role in gene expression and point them out as a novel strategy for disease treatment. 2.2 Physiological roles of G4 structures Despite many reports associating G4s to DNA damage and genome instability, their high conservation across evolution in mammalian species supports their role as regulatory elements. Indeed, several studies have outlined the many DNA transactions in which these non-canonical DNA structures are involved. One of the first processes associated to a physiological role of G4s was telomere biology. Telomeres are enriched in G-rich tandem repeats with high potential to form G4s. The presence of these structures in these areas of chromosomes have a dual role in telomere integrity both protecting telomeres from degradation by acting as a cap to telomeric DNA and regulating telomere synthesis by the telomerase thanks to the cell cycle dependent turnover regulation of telomeric G4s (Juranek and Paeschke, 2012; Smith et al., 2011). In fact, telomerase activity is inhibited by G4s stabilizing compounds leading to telomere shortening (Sun et al., 1997), although telomerase activity is not inhibited by every G4 ligand since it is able to process DNA through parallel G4s, but not antiparallel ones (Moye et al., 2015; Paudel et al., 2020; Zhang et al., 2010). Introduction 44  Apart from the aforementioned G4-driven genome instability mechanism, the most investigated pathway for G4 induced DNA damage is DNA replication. These noncanonical DNA structures constitute a threat to replication machinery interfering with the replication fork progression of both the lagging and leading strands (Lerner and Sale, 2019) and the function of DNA polymerases (Estep et al., 2019). Particularly, stalling of the replication fork due to an unresolved G4 induces DSBs, one of the major promoters of mutagenesis when it is not properly fixed. Besides, replication is the main mechanism of DSBs generation by G4s since the SSB produced can be converted into a DSB in the second round of replication. 2.5.1 G4 stabilising compounds and cancer treatment Evidence of the negative G4 impact on DNA integrity is validated by the deleterious effects of the action of small-molecule G4 ligands, which stabilize these structures and induce DNA damage by replication stress at PQS (Piazza et al., 2010; Rodriguez et al., 2012). Furthermore, the use of G4 ligands induce an increase in cell cytotoxicity in cells where HR-factors, such as BRCA1/2 and EXO1, are impaired (Stroik et al., 2020; Xu et al., 2017; Zimmer et al., 2016). To date, a mounting number of G4 interacting compounds have been developed and most of them can be found in the G4 Ligands Database 2.1 (http://www.g4ldb.com/). The common characteristics shared by most of these compounds is the presence of an aromatic core and the selectivity between G4 and duplexes topologies. Some of the best-known G4 ligands include pyridostatin, BRACO-19 and TMPyP4 (Kotsantis et al., 2020; Rodriguez et al., 2008). The finding of G4s in telomeres and promoters of oncogenes has led much attention to the use of G4 ligands as therapeutic compounds. This general concept was supported by the observation that cancer cells can be addicted to hyperactivated oncogenes and/or augmented activity of the telomerase preventing senescence. In consequence, specific downregulation of oncogenes expression or telomerase inhibition by G4s stabilization could inhibit cell proliferation and cause cell death resulting in anticancer activity. Despite the promising target that G4 ligands offer, not many of these compounds have entered phase II trials. This is explained by the low selectivity that G4 interacting molecules shows towards a certain G4 conformation (Zuffo et al., 2018) complicating their use as an specific anti- Introduction  45 carcer treatment. Nevertheless, efforts are being made to design new compounds without these drawbacks and, at the same time, new studies showing the efficacy of their use as novel strategies for cancer treatment keep arising (Amato et al., 2020; Gowan et al., 2002; Hu et al., 2020). 2.6 G4s and Topoisomerases Non-canonical DNA structures, such as G4s and R-loops, usually arise as consequence of negative supercoiling behind the transcription machinery, particularly in areas with a high transcription rate. To face these changes in the topology and to relax the supercoiled DNA areas, cells use topoisomerases (Yadav et al., 2014). Topoisomerases constitute a large family of ubiquitous nuclear enzymes that introduce transient breaks in the DNA backbone to maintain DNA topology maintenance during cellular transactions. In eukaryotic cells, Topoisomerases can be classified into two groups, topoisomerase I (TOPOI) and topoisomerase II (TOPOII), each of them being further divided in the subgroups A or B. Each group is characterised by different structural organization, enzymatic catalysis modalities and biological functions. TOPOI subfamily acts as monomers that cleaves transiently only one strand of the DNA to relax the DNA supercoiling, followed by passage of the intact strand through the break (type IA) or by controlled rotation of the helix around the break (type IB). Contrary, Type II topoisomerases are homodimers that covalently attach to 5’ ends and generate a transient DSB to carry out the passage of another duplex DNA. During the cleavage, a covalent bond is formed between the protein and the DNA generating catalytic intermediates known as topoisomerases cleavage complexes (TOPOcc). Once the topology threat is resolved, the DNA break is re-ligated. Due to their reaction mechanism, TOPOI can mediate winding and unwinding activities but is not able to resolve knots and tangles while TOPO II can modulate both activities (Champoux, 2001). Human cells possess six different topoisomerases with some redundant activities in transcription, replication and DNA repair: TOP1 and TOP1mt (type IB), TOP2α and TOP2β (two isoforms of type IIA) and TOP3α and TOP3β (type IA) (Pommier et al., 2016). Introduction 46  Despite TOP2 function being critical for the correct progressing of DNA transaction, its mechanism of action generates potentially deleterious DSBs if not properly re-ligated or if TOP2 activity is impaired or blocked. Indeed, topoisomerase inhibitors are described to induce DNA damage and genome instability and, in consequence, some of them are validated as anti-cancer treatments (Nitiss, 2009). To understand the mechanism of action of TOP2 inhibitors, it should be explained the catalytic cycle of TOP2 (Fig.I7). The cycle starts when TOP2 binds to an intact segment, the G (gate) segment, and then retains another DNA region, the T (transported) segment. Then, TOP2 experiences a conformational change using ATP from an open to a closed clamp form. In the presence of Mg2+, each TOP2 monomer strike a DNA phosphodiester bond four base apart on opposite strands of the G duplex and becomes covalently linked to the 5’ end of the broken DNA Figure I7. Topoisomerase 2 (TOP2) catalytic cycle ATP Mg 2+ TOP2cc ADP+Pi Merbarone G segment T segment Introduction  47 Figure I7. Topoisomerase 2 (TOP2) catalytic cycle The TOP2 homodimer binds two duplex DN fragments. First, the G segment (red) enters the clamp which is in the open conformation, followed by the interaction with a second DNA segment (T segment, black). When the T segment enters TOP2 complex, ATP provides the energy required to close the clamp, cleave the G segment and, in presence of Mg+2, covalently get linked to each strand of the G segment via a 5′-phospho-tyrosyl bond. forming the TOP2 cleavage complex (TOP2cc). Then, the T segment passes through cleaved G fragment to the central cavity and gets released followed by the relegation of the G segment. The hydrolysis of the remaining ATP and the dissociation of two ADP molecules finally releases the G-segment and prepares the TOP2 complex for an additional catalytic cycle. TOP2 can be inhibited or poisoned at several different points in the enzyme reaction cycle. For example, merbarone inhibits TOP2 cycle before DNA is cleaved. generating a cleavage complex intermediate (TOP2cc). Once the T segment is transferred through the induced break, the T segment is released and the broken ends of the G segment are resealed. The cycle finishes with the release of the G segment, that is released upon ATP hydrolysis that converts the complex back to its open clamp form, leaving the DNA product chemically intact from the initial backbone (Deweese and Osheroff, 2009). Taking TOP2 catalytic cycle into account, TOP2 inhibitors target different steps through this cycle and, thus, can be classified into two groups: TOP2 poisons and TOP2 catalytic inhibitors. On one hand, TOP2 poisons blocks the catalytic cycle after DNA is cleaved but prior to DNA relegation, leading to increased levels of DNA cleavage and, thus, DNA damage. In the other hand, catalytic inhibitors target TOP2 without stabilizing TOP2cc or increasing DNA cleavage. In this group, Merbarone is particularly interesting since this agent is able to block TOP2 cutting activity without affecting protein-DNA binding (Fortune and Osheroff, 1998; Nitiss, 2009; Pastor et al., 2012). Although the connection between topoisomerases and G4s has mostly been limited to the participation of the first ones in G4 resolution, recent studies show that they might also be involved in G4 mediated genomic instability. In fact, TOP2-induced DSBs seem to colocalise with DNA secondary structures and human TOPOI is able to bind G4 structures and even participate in their formation (Arimondo et al., 2000; Szlachta et al., 2020). Interestingly, novel data demonstrates that the cytotoxicity of some G4 ligands, such as pyridostatin, is mediated by TOP2 trapping (Bossaert et al., 2021; Bruno et al., 2020; Olivieri et al., 2020). Although these experiments show that TOP2 promotes cell-killing activity by some G4 stabilising agents, this contribution cannot be generalised to all of them Introduction 48  since, for example, PhenDC3 mediated instability is not promoted by TOP2 (Olivieri et al., 2020). Olivieri et al., 2020 indicated as well, that, despite the behaviour of pyridostatin being different to etoposide (a TOP2 poison) the outcome obtained was the same (TOP2cc levels increased). This intertwining between TOP2 ligands and G4 interacting compounds should be further addressed since this finding is not an isolated coincidence and other compounds have also showed a high activity as TOP2 inhibitors and G-quadruplex binders (Palluotto et al., 2016). 3. Clustered Regularly Interspaced Palindromic Repeats (CRISPR)–Cas9 The ability of modifying the DNA sequence to our desire at a precise location is of immense relevance for the molecular biology, biotechnology and even medicine and is known as genome editing. Genome editing has developed and improved impressively over the years from the discovery of restriction enzymes to clone a DNA sequence into a plasmid to the use of Clustered Regularly Interspaced Palindromic Repeats (CRISPR)–Cas system to accurately modify a single nucleotide in the genomic DNA. During this evolution, many techniques have been explored such as Cre-recombinases, meganucleases, ZINC-finger nucleases or TALEN, nicely reviewed in (Khalil, 2020; Perez-Pinera et al., 2012; Urnov et al., 2010; Zhang et al., 2019). However, the biggest advance in this field was made with the discovery of CRISPR-Cas system. CRISPR-Cas system for gene editing derives from the discovery of CRISPR in Prokaryotes. It is based in their adaptive immune system that prevents phage infection by storing small sequences of viral DNA, called spacers, in bacterial host chromosomes. These viral DNA fragments from different infections are separated by repetitive nucleotide sequences called direct repeats. These CRISPR elements were adjacent to multiple wellconserved genes known as CRISPR-associated (Cas) genes that encode nucleases. After a viral challenge, bacteria integrate new spacers derived from the infecting phage genomic sequence into its genome. These spacers are important for the system because, upon a reinfection, they get transcribed into some short guiding CRISPR RNA (crRNA) that lead Introduction  49 the targeting by Cas of the viral genome, thus, protecting the bacteria against the phage (Jore et al., 2012). The constant survival struggle of prokaryotes to confront the environmental challenges has promoted the evolution of different CRISPR-Cas systems. The system considered as classical mainly comprises six types of CRISPR-Cas that can be classified into two classes based on the structure of the Cas genes. Class I includes type I, II and IV systems and uses several Cas proteins, whereas Class II (type II, V and VI) possesses only one Cas protein with multiple domains, making it easier to engineer (Makarova et al., 2018). Due to its simplicity, the most widely used in genetic engineering is the type II CRISPR- Cas9 from Streptococcus pyogenes. Cas9 protein is an endonuclease with two domains, Ruvc and HNH. The HNH domain cleaves single DNA strands complementary to the sgRNA, while the Ruvc domain cleaves the opposite strand (Fig.I8) (Hryhorowicz et al., 2017). The basic components of the typical CRISPR-Cas9 system consist of a Cas9 nuclease and a single guide RNA (gRNA). The gRNA is formed by the synthetic fusion of the crRNA and the trans-activating crRNA (tracrRNA). To recognise its target sequence, crRNA encodes a 20 bases recognition sequence that binds specifically the target site with the help of the tracrRNA. In type II CRISPR-Cas9, this target site needs to be placed next to a protospacer adjacent motive (PAM) that can be either NGG or NAG, with N being either A, T, G, or C. Thus, the gRNA binds by its 5’end the Cas9 protein forming the sRNA/Cas9 complex and, by its 3’ end, it binds the target site, guiding the complex. Finally, upon target recognition, the gRNA/Cas9 complex cleaves the DNA 3-4 nucleotides downstream the PAM sequence inducing a DSB (Fig.I8) (Hryhorowicz et al., 2017). Introduction 50  3.1 Genome editing by CRISPR-Cas9 The site-specific DNA DSBs generated by Cas9 induce the cellular DDR to repair the break which can be exploited to engineer the genome. As already explained, DSBs can be repaired either by NHEJ or HR, both of which can be used to obtain the desired genome editing outcome (Fig.I8). In the absence of a proper template, the DNA ends of the DSB are re-ligated by the error-prone NHEJ pathway. As a consequence, the break can be repaired faithfully reconstituting the sequence or produce a diverse array of mutations, like indels. The generation of these indels in the coding sequence of a gene can produce frameshift mutations Figure I8. Clustered Regularly Interspaced Palindromic Repeats (CRISPR)-Cas9 system A guide RNA (gRNA) formed by the crRNA and the tracrRNA is the responsible for recognising the genomic sequence which precedes the protospacer adjacent motive (PAM). This recognition guides the Cas9 nuclease to the target sequence which generates a DSB 3-4 nucleotides upstream o f the PAM sequence by the two Cas9 nuclease domains (RuvC and HNH) and, therefore, activating the DSB repair machinery. In the absence of a homologous repair template, the break is repair by N HEJ which can result in indels disrupting the target sequence. Alternatively, the repair of the DSB by HR or SSTR can be promoted by the provision of a homologous donor template. crRNA tracrRNA DSB NHEJ HR/SSTR Indels Indels Donor C as9-RuvC Cas9-HNH 5’ 5’ 5’ 3’ 3’ 3’ PAM NNNNNNNNNNNNNNNNN NNNNGG NNNNNNNNNNNNNNNN NNN NNNNNNNNNNNNNNNN NNNNNCC N Introduction  51 or premature stop codons that leads to loss of the function of the gene and, thus, generate a knock-out (KO). Alternatively, the DSB can be repaired by HR introducing specific nucleotide changes that span from a single nucleotide variation to the insertion of large DNA fragments (Ran et al., 2013b). To promote Cas9-DSB repair by HR a template is used which can be endogenous (resulting in an unedited sequence) or exogenous. In general, there are two major types of exogenous templates to be used in the CRISPR-Cas9 system depending on the desired edition to make: a conventional dsDNA targeting constructs, such as a plasmid or a PCR product, with homology arms flanking the insertion sequence, or ssDNA oligonucleotides (ssODNs). dsDNA donor template repair occurs mainly through HR while ssODN repair is mediated by SSTR, similar to SSA (Gallagher and Haber, 2021). Although these repair templates have some differences, both must contain the desired edit as well as additional homologous sequence (homology arms) immediately upstream and downstream of the target sequence. While ssODNs provide a much more effective and simple method for making small edits in the genome, such as the introduction of singlenucleotide mutations. dsDNA constructs offer the possibility of introducing larger DNA fragments (a tag or a fluorophore) hence generating a knock-in (KI). Nevertheless, the DNA sequence provided in the template can introduce a mutation as well that leads to the KO of the gene (Renaud et al., 2016; Xue and Greene, 2021). 3.2 CRISPR-Cas9 applications Apart from generating KO and KI in genes, CRISPR-Cas flexibility as a gene manipulation tool have been exploited to widen the repertoire of their application. This is possible thanks to the ability of Cas enzymes to bind target DNA independently of their ability to cleave it. Therefore, the modification of the cleavage activity of Cas9 renders a new tool to modify gene expression or sequence without inducing DNA damage. The point mutations D10A and H840A of the endonuclease domains of the Cas9, Ruvc and HNH respectively, result in a catalytically nuclease dead Cas9 (dCas9) molecule that cannot cleave target DNA but can bind to it. Then, this dCas9 can be fused to transcriptional activators (CRISPRa) or repressors (CRISPRi) to bind gene promoters and regulate their expression (Gilbert et al., 2013; Maeder et al., 2013; Qi et al., 2013), to epigenetic regulators to create programmable epigenome regulators (Brocken et al., 2018) Introduction 52  or fluorescent tags to visualise the targeted genomic loci (Ye et al., 2017). Since dCas9 does not modify permanently the DNA sequence, this modulation is reversible. This creates a potential tool to investigate essential genes (Peters et al., 2016; Singh et al., 2016) or even regulate biosynthetic metabolic pathways in bacteria (Gao et al., 2018; Kim et al., 2017). In some organisms, the induction of Cas9-DSB is very toxic thus complicating the manipulation of their genome. To overcome this setback, dCas9 can be used by tethering it to DNA deaminases (Komor et al., 2016; Rees et al., 2017) to edit an specific loci without damaging the genome. Another strategy used in these situations includes the use o a partially inactive Cas9 in which just one of the endonucleases domains is inactive. This Cas9 only forms a non-lethal SSB hence is considered a nickase (nCas9) (Standage-Beier et al., 2015). The simplicity of gRNA design and production, as well as the ability to target almost any genomic locus, has pointed out CRISPR as the perfect system for large-scale genome-wide genetic screenings (Shi et al., 2015; Zhong et al., 2015). Currently, the most popular method for conducting genome-wide screens using CRISPR involves the use of pooled lentiviral CRISPR libraries where each lentiviral transfer vectors encodes an individual gRNA targeting a single gene (McDade et al., 2016). CRISPR-Cas9 have also raised interest for its potential use to correct genetic diseases. In cancer research, the applications of CRISPR-Cas9 are extensive and range from the development of in vivo and in vitro cancer models and the screening of novel oncogenic mutations and tumour suppressors to cancer gene therapy. The efficient and straightforward gene-editing capacity of CRISPR-Cas9 accelerates the process of obtaining a representative cancer model by mutating the gene involved in the under-study cancer (Kaltenbacher et al., 2022; H. Wang et al., 2013). In cancer treatment, CRISPR-Cas9 can be used to target specific sequences in the cancer genome or driver mutations associated with cancer progression. Doing so, only selective apoptosis of tumour cells would be induced, reducing adverse effects on normal tissue cells, which are usually associated to other clinical cancer treatments (Martinez-Lage et al., 2020). In addition, CAR-T cells, used in cancer immunotherapy, could be genetically modified by CRISPR-Cas9 to increase their effectiveness (Razeghian et al., 2021). Introduction  53 Nevertheless, CRISPR-Cas9 application in medicine should not be limited to cancer. Indeed, other genetic hereditary disease are in spot to be treated by CRISPR-Cas9. Cystic fibrosis (Graham and Hart, 2021) or Duchenne’s muscular dystrophy (Nelson et al., 2016; Zhang et al., 2022) are examples of them. Correcting the gene mutation that origins the disease could restore the altered genetic code. Infectious diseases like HIV (Herskovitz et al., 2021) or COVID-19 (Abbott et al., 2020) are another recent target of CRISPR-Cas. For instance, researchers have developed a method to selectively target and destroy the genetic material introduced by the virus in the cell, preventing infection. 3.3 Enhancing CRISPR-Cas9 technology The mounting number of studies employing CRISPR-Cas9 to such an extended number of applications is proof enough of all the future perspectives this genetic engineering tool may provide. CRISPR-Cas9 is cheap, easy to use, customise and produce, has a wide targeted sequence range and a robust activity in eukaryotes. Although CRISPR- Cas9 technology has constantly validated its significant advantages, it is undeniable that there are some major drawbacks in its use that limit its potential. The less concerning weakness of CRISPR-Cas9 might be the requirement of a specific PAM sequence next to the target DNA. While the PAM sequence has a critical role in CRISPR-Cas in bacteria immune system to avoid self-recognition, it constrains the selection of the ideal DNA target. To deal with this limitation, other Cas9 orthologs and variants are being optimised to provide a wider catalogue of PAM sequence that adapt better to the needs of the editing (Collias and Beisel, 2021). One of the major problems in CRISPR-Cas9 technology is nuclease cleavage at unintended genomic sites, caused by the nuclease tolerating mismatches between gRNA and “off-target” DNA (Fu et al., 2013). These induced DSBs in unprogrammed areas generates off-target effects that, among other consequences, impede CRISPR-Cas9 use in clinical application. Since this disadvantage has very relevant detrimental consequences for CRISPR-Cas9 development as the ideal genetic tool, there are many researchers focusing on smoothing these off-target effects out. One strategy for improving Cas9 specificity is to induce two colocalizing SSBs, one in each opposite strand, instead of directly generating a     III. Results Results-Chapter I  63  Chapter I: DNA resection over G4-prone sequences by PIF1α DNA resection is a complex process which involves a highly coordinated response of different factors. The research of the central machinery involved in this process has been the central topic of many studies, so it is well characterised. However, little is known about the additional factors necessary to overcome DNA structures that might impede DNA resection. Screening of helicases involved in DNA resection As in any other DNA transaction, an important factor in DNA resection are helicases. Helicases, by definition, unwind DNA so we hypothesised that, in a context of a barrier due to an abnormal DNA conformation, they might be a key factor to promote DNA resection. To have a better knowledge of helicases that may be involved in this key step of HR, we reviewed the effect of all human helicases in a previous genome-wide screening performed in the lab (López-Saavedra et al., 2016). In such study, more than 300 proteins that affect the balance between HR and NHEJ were described using the SeeSaw Reporter 2.0 (SSR 2.0). This reporter analyses the choice between HR and NHEJ at very early stages being particularly sensitive to changes in DNA resection velocity and/or processivity. With such reporter, the outcome of the repair of a single, inducible DSB, can be visualized by the appearance of green (cells that have repaired by NHEJ) or red (those that have completed HR, specifically SSA) cells (Fig. R1A). No helicases described to have a role in DNA resection were in this candidate shortlist, so the data was re-analyzed by lowering the threshold for positive candidates (Fig. R1B). In that revision, PIF1α depletion, among other helicases, was found to raise NHEJ over HR. Results-Chapter I 64  Figure R1. DNA helicases effect in DNA repair balance A) Schematic representation of the See Saw Reporter (SSR) 2.0. An I-SceI DSB is induced and repaired either by NHEJ, leading to GFP reconstitution, or by HR using RFP fragments, thus generating a functional RFP gene. B) Overview of the effect of the depletion of different DNA helicases in NHEJ/HR balance obtained from López-Saavedra et al. (2016). Helicase genes for which depletion produces an unbalance toward increased HR are marked in red. Genes that favours recombination (NHEJ increased when they were downregulated) are marked in green. The plo t represents the mean and SD of three independent experiments. A B BACH1 BLM CHD2 CHD7 ERCC2 ERCC3 HELLS INO80 MCM2 MCM3 MCM4 MCM5 MCM6 MCM7 MCM8 MCM9 MCM10 NCL PEO1 PIF1 RECQL RECQL4 RECQL5 RTEL1 RUVBL1 WRN 0.5 1 2 4 Increased NHEJ Increased HR NHEJ/HR ratio relative to control siLuc RWDD2A SeeSaw Reporter 2.0 I-SceI BFP infection Results-Chapter I  65  PIF1α is required for DNA resection So, PIF1α depletion seemed to favour NHEJ over HR. As stated before, the balance between both repair pathways is mostly regulated by DNA end resection, the initial step in HR. Indeed, as the SSR is specifically designed to detect factors affecting DNA resection, we investigated the role of PIF1α in HR through its involvement in that process. As explained in the introduction, it produces long single-stranded DNA (ssDNA) tails that are immediately coated by RPA protein complex for protection. The accumulation of RPA at the sites of the break produces foci that can be visualised by immunofluorescence and therefore can be used as a readout of DNA resection in mammalian cells. Figure R2. PIF1α is involved in DNA resection  * % RPA positive cells siNT siNT siPIF1α siPIF1α PIF1α HSP70 Ladder siNT siCtIP siPIF1α siNT siPIF1α 0 50 100 G1 S/G2 % Cells A D B C RPA γH2AX DAPI Results-Chapter I 66  Figure R2. PIF1α is involved in DNA resection A) Representative Western-blot image of siRNA-mediated depletion of PIF1α. Protein extracts were obtained from U2OS cells 48h after siRNA transfection. siNT represents a control, Non-Target, siRNA. HSP70 was used as a loading control. B) DNA resection proficiency measured as the percentage of RPA positive cells upon DSBs induction by ionizing radiation (10 Gy). The average and SD of three independent experiments are shown. Significance was determined by t-test Student. (*P < 0.05). C) Cell cycle distribution of U2OS cells 48h after transfection with the indicated siRNA. The mean and SD of three independent experiments are plotted. D) Median DNA resection length obtained by SMART after ionizing radiation (10 Gy) represented as the mean and SD of three different experiments (left) and a representative fluorescence microscopy image of the individual single stranded DNA fibers obtained (right). Significance was determined by t-test Student. (*P < 0.05, **P < 0.01). In order to test the effect of PIF1α in DNA resection, we depleted this protein by siRNA (Fig. R2A) and analysed its impact on RPA foci formation in response to a high dose of IR (10 Gy). The damage‐induced phosphorylation of H2AX (γH2AX) was used as a marker of DSBs. In agreement with a role favouring HR over NHEJ, PIF1α siRNAs rendered a defect in RPA foci formation (Fig. R2B). DNA resection and, in consequence, HR are limited to G2/S phases of cell cycle so any change in the percentage of cells going through each one of these phases have an impact in the DNA repair pathway undertaken. As we observed in Figure R2C, the RPA foci impairment associated to PIF1α depletion did not rely on any effect on the cell cycle since the knockdown of PIF1α did not affect cell cycle distribution. Despite RPA foci formation being the gold standard readout of resection, it is a lowresolution technique that does not provide information about the length or speed of the resection process. To overpass this deficiency, an alternative method named Single Molecule Analysis of Resection Tracks (SMART) (Cruz-García et al., 2014) was used to assess DNA resection in individual DNA fibers. Interestingly, not only the number of breaks resected was reduced upon PIF1α depletion, but the average length of ssDNA formed during resection was reduced in a similar extent to that caused by the depletion of the bonafide resection factor CtIP, as can be appreciated in the graph and representative in Figure R2D. This suggests that PIF1α might play a relevant role in DNA end resection. Results-Chapter I  67  PIF1α is recruited to DSBs In order to further study this potential role in DNA repair, we decided to check if PIF1α was recruited to DSBs. To do so, we first studied the ability of GFP-PIF1α to form foci after irradiation, a common feature of proteins involved in DNA repair. This analysis showed that, once damage was induced, the number of GFP-PIF1α foci raised, thus suggesting the recruitment of PIF1α to DSBs (Fig. R3A). To further study this recruitment Figure R3. PIF1α is recruited to DNA-damage induced foci Cells with GFP-PIF1α foci at the array (%) Time after DOX addition ** * ** * * Cells with γH2AX foci at the array (%) 0h 14 h 20 h 0 20 40 60 Time after DOX addition α -IR +IR Number of GFP-PIF1α foci A B CD Results-Chapter I 68  Figure R3. PIF1α is recruited to DNA-damage induced foci A) U2OS stably expressing GFP-PIF1α were used to study PIF1α foci formation. Quantification of the median number of foci per cell after mock treatment or 10 Gy of ionizing radiation (left) and representative images (right) of GFP-PIF1α foci (red), γH2AX (grey) and DAPI (blue) are shown. B) Schematic representation of the experimental system used to measure protein recruitment to DSBs (left). In U2OS19ptight13, I-SceI is expressed upon doxycycline addition to the medium and then the endonuclease produces a single DSB. Binding of a Cherry-lacI construct to the lacO tandem repeats located next to the I-SceI cleavage site allows its visualization. Representative fluorescence microscopy images of the lacO array in red, GFP-PIF1α in green and γH2AX in grey are shown (right). C) Quantification of the percentage of colocalization of the lacO array with γH2AX in U2OS19ptight13 cells expressing GFP-PIF1α after DSB induction upon Dox treatment. D) GFPPIF1α recruitment to DNA damage scored as the percentage of colocalization of the array with GFPPIF1α in cells treated as in C. Statistical significance as in Fig. R2D. of PIF1α to damaged chromatin we took advantage of a genetic tool in which a single DSB can be created by the endonuclease I-SceI after doxycycline induction (Lemaître et al., 2014). This I-SceI target site is flanked by 256 repeats of the lac operator (lacO) that can be visualized by the accumulation of Cherry-lacI (Fig. R3B). Therefore, after its induction with doxycycline, we created a single DSB that we can be visualize as a red spot where proteins involved in its repair would be recruited. DNA damage indicators, for example, γH2AX, will also co-localize with the Cherry-lacI. In order to analyse the recruitment of PIF1α to DSBs, we measured colocalization of GFP-PIF1α with the lacO array. First, we confirmed DSB induction after doxycycline treatment by the detection of γH2AX at the lacO array (Fig. R3C). Although there was some background binding of PIF1α before doxycycline addition, likely due to the structure created by lacO repeats, PIF1α showed an increased colocalization with the array after DSB induction (Fig. R3D). These results reinforce the hypothesis that PIF1α is recruited to DSBs. G-quadruplexes containing sequences act as roadblocks for DNA resection Among the different activities of PIF1α in DNA transactions, its role in the resolution of atypical DNA species seemed of particular interest for PIF1α requirement in DNA end processing. One of these noncanonical structures are G-quadruplexes (G4s). As these G4s act as a barrier to DNA replication machinery (Sarkies et al., 2010), we thought G4s could also affect DNA resection. To test this hypothesis, we used a G4 stabilizer, pyridostatin (Rodriguez et al., 2008) that is described to induce DNA damage (Rodriguez et al., 2012). This small molecule, on its own, produced a defect in DNA resection after Results-Chapter I  69  ionizing irradiation to a similar extent to the one produced by PIF1α depletion (Fig. R4A) and not linked to changes in cell cycle distribution (Fig. R4B). Furthermore, pyridostatin treatment was epistatic over PIF1 depletion, indicating a genetic connection between the helicase and the G4 dissolution. Figure R4. G4s stabilization impairs DNA resection BrdU G4 Merge % RPA positive cells * * * ** * ** * B A C D E -Pyridostatin -Pyridostatin +Pyridostatin +Pyridostatin siNTsiPIF1α RPA γH2AX DAPI Results-Chapter I 76  Figure R8. PIF1α participates in both short and long-range resection  A C B EFG D siNT siPIF1α * * * %MRE11 RNA levels relative to control %DNA2 RNA levels relative to control %EXO1 RNA levels relative to control %Cells Results-Chapter I  77  Figure R8. PIF1α participates in both short and long-range resection A) DNA resection proficiency measured as the percentage of RPA positive cells in cells cotransfected with the indicated siRNAs. Other details as in Fig. R2B. B) Cell cycle distribution of U2OS cells 48h after transfection with the indicated siRNA. The mean and SD of three independent experiments are plotted. C) Same as in A, but relativised to cell cycle distribution. D) mRNA level expression of PIF1α in U2OS cells depleted for the indicated proteins when compared to a control condition. Cells transfected with the corresponding siRNA were harvested 48 hours upon transfection. mRNA was extracted from cell pellets and reversed transcribed into cDNA. cDNA abundance was measured by qPCR with different primer pairs. The mean and standard error of three independent experiments is shown in the graph. E, F and G) Same as in D, but for the indicated proteins. G-quadruplex containing sequences influences DNA resection in vitro In an attempt to gain a neater knowledge of the impact of G4s structure in resection, we moved to in vitro DNA resection assays (Pinto et al., 2018). For that, we created a dsDNA substrate bearing two G4s motifs from the C-MYC promoter in the prone-to-be resected 5’- 3’ strand. In addition, two lacO cassettes were incorporated in the 3’ end of the plasmid to inhibit DNA resection from that end when lacI is added to the reaction (Fig. R9A, left). Next, we studied the resection efficiency in this linearised dsDNA substrate compared to a control without the G4 sequences (Fig. R9A, right) in the presence of different human proteins involved in DNA resection. First, we did a kinetics with DNA2 nuclease in combination with BLM helicase to analyse long-range DNA resection. The fall in the intensity of the substrate that remains unprocessed is considered the percentage of substrate that has been processed (% of substrate utilization), otherwise said, resected. Contrary to the expectations, the presence of the G4s structures in the substrate promoted substrate utilization compared to the control in the first 30 min of the reaction (Fig. R9B). A similar result was observed when WRN helicase was used instead of BLM, although in this case the increase was present in all the timepoints selected (Fig. R9C). In contrast, EXO1 resection of the substrates seemed unaltered by the presence of the G4s, probably due to a saturation of the reaction by the high concentration of EXO1 protein used (Fig. R10). Given the close proximity of the G4 structures to the simulated DSB, one plausible possibility to explain the increased DNA degradation observed in the G4s-bearing substrates could be due to a change in the melting conditions of the dsDNA and, in consequence, an easier access of the resection machinery to the DNA strand. To test this possibility, we Results-Chapter I 78  Figure R9. G4s presence favours DNA resection by DNA2-BLM and DNA2-WRN in dsDNA substrates A B C 30 nM hDNA2 20 nM hBLM -- -+ - + - + + + + + + + -- -+ - + - + + + + + + + 0 15 30 60 15 30 60 0 15 30 60 15 30 60 Time (min) pUC19x2LacO pUC19x2LacOx2G4 nontemplate hRPA 576 nM (300%) Unprocessed dsDNA ssDNA Resection products Unprocessed dsDNA ssDNA Resection products LACO LACO LACOLACO 5’ 5’3’ 3’ LACI LACI Resection Radioactive label Control DNA substrate2xG4s DNA substrate LACO LACO LACOLACO 5’ 5’3’ 3’ LACI LACI Resection Radioactive label Results-Chapter I  79  Figure R9. G4s presence favours DNA resection by DNA2-BLM and DNA2-WRN in dsDNA substrates A) Schematic representation of the dsDNA substrates used in the DNA resection in vitro assay. In the left, dsDNA substrate bearing 2 G4s sequences in tandem 44 bp from the 5’end of the linearized pUC19x2lacO plasmid and radioactively labelled with [α-32P]-dCTP in the 3’ end of the top strand. In the right, control substrate with the same characteristics as in the left, but without the G4 sequences. B) Quantification of substrate utilization kinetics by resection of dsDNA substrates bearing or not two G4 sequences by DNA2 (30 nM) and BLM (20 nM) (left). A representative agarose gel (1%) of the resection assay is shown (right). The mean and SD of three independent experiments are shown. Significance was determined by t-test Student. (*P < 0.05) C) Same as in B, but with DNA2 and WRN (50 nM). boiled the dsDNA to obtain a labelled ssDNA with the G4 motifs and without them (Fig. R11A), where no unwinding of the double helix would be necessary. Next, we repeated the resection assay with only DNA2, since no helicase is required. This experiment showed a decreased substrate utilization when the G4s were present (Fig. R11B), further supporting our previous in vivo findings that G4s inhibit DNA resection. However, further experiments with dsDNA substrates in which the G4s are located further inside the DNA will be required to clarify completely this point. Figure R10. DNA resection efficienc y b y EXO1 is not affected b y G4 structures in dsDNA substrates Quantification of substrate utilization by resection of dsDNA substrates bearing or not two G4 sequences by EXO1 (left). A representative agarose gel (1%) of the resection assay is shown (right). The mean and SD of three independent experiments are shown. Unprocessed dsDN A ssDNA Resection products hEXO1 (nM) 0 1 10 40 pUC19x2LacO pUC19x2LacOx2G4 nontemplate hRPA 576 nM (300%) 0 1 10 40 Results-Chapter I 80  Figure R11. G4 structures impair DNA resection by DNA2 in ssDNA substrates A) Schematic representation of the ssDNA substrates used in the DNA resection in vitro assay. On the left, ssDNA substrate bearing 2 G4s sequences in tandem 44 bp from the 5’end of the linearize d pUC19x2lacO plasmid and radioactively labelled with [α-32P]-dCTP in the 3’ end. On the right, control substrate with same characteristics as in the left, but without the G4 sequences. B) Quantification of substrate utilization by resection of ssDNA substrates bearing or not two G4 sequences by DNA2 (left). A representative agarose gel (1%) of the resection assay is shown (right). The mean and SD of three independent experiments are shown. Significance was determined by ttest Student. ( *P < 0.05 ) LACO LACO 5’ 3’ Resection Control DNA substrate2xG4 DNA substrate A B LACO LACO 5’ 3’ Resection Radioactive label Radioactive label hDNA2 (nM)0153060 pUC19x2LacO pUC19x2LacOx2G4 nontemplate hRPA 576 nM (300%) 0153060 dsDNA ssDNA Resection products Results-Chapter I  81  FAN1 nuclease cooperates with PIF1α in G4-associated DSBs In DSB repair, it is frequent the formation of a helicase/nuclease complex to process the damaged DNA. Being PIF1α a helicase with a possible role in DSB repair, we wondered which nuclease could be the associated with PIF1α activity. To do so, we checked FAN1, a known nuclease with a stablished role in DNA interstrand cross-link repair by HR and involved in G4s-induced fork collapse resolution (Porro et al., 2017), although its implication in DNA resection had not been defined. To study this possible role, we analysed the capacity of FAN1 downregulated cells to form RPA foci after DNA damage. In accordance with our expectations, we observed a fall in the percentage of cells able to recruit RPA and, interestingly, this effect was epistatic over PIF1α depletion (Fig. R12A) and independent of cell cycle (Fig. R12B). Since FAN1 depletion caused a defect in DNA resection initiation, measured by RPA foci formation, we wondered if it would also affect the resection length of the DNA fibers. Contrary to our expectations, the SMART assay of FAN1 depleted cells showed a rise in the length of the resected fibers that was abolished when G4s were stabilized by pyridostatin (Fig. R12C). This increase was also appreciated when we measured the percentage of fibers containing, at least, one G4 structure (Fig. R12D). However, in this case, the enhancement observed compensated the drop after pyridostatin treatment. Next, we wanted to check if PIF1α overexpression could rescue the phenotype obtained with FAN1 depletion. Using RPA foci formation as a readout, we observed a lessening after pyridostatin addition both in the control and in FAN1 depleted cells and this reduction disappeared when PIF1α was overexpressed (Fig. R12E). FAN1 downregulation was checked by qPCR (Fig. R12F). These findings support a role of FAN1 in DNA resection that is hampered by G4s and stimulated by PIF1α. Results-Chapter I 82  Figure R12. FAN1 is involved in DNA resection over G4 sequences A) DNA resection proficiency measured as the percentage of RPA positive cells as in Fig. R11A. B) Cell cycle distribution of U2OS cells 48h after transfection with the indicated siRNA. The mean and SD of three independent experiments are plotted. C) Median DNA resection length obtained by SMART after ionizing radiation (10 Gy) in cells depleted for FAN1 or for a control sequence (siNT) and treated either with H2O or pyridostatin (10 µM) for 2h. Other details as in Fig. R2D. D) Quantification of the percentage of ssDNA fibers containing at least one G4, other information as in C. E) Same as in A, but in U2OS cells stably expressing GFP, GFP-PIF1α or GFP-PIF1α E307Q and treated either with H2O or pyridostatin (10 µM) for 2h. F) mRNA level expression of FAN1 in U2OS cells depleted for either FAN1 or a Non-Target sequence when compared to a control condition. Other details as in Fig. R11D. A CB EF D %RNA levels relative to control %Cells GFP-PIF1αGFP *** Results-Chapter II  83  Chapter II: Searching new factors affecting CRISPR-Cas9 gene-editing efficiency As we have already stated in the introduction, CRISPR-Cas9 has been stablished as an efficient tool for genome editing. Despite its multiple advantages, one important setback of this technique is the low rates of KI efficiency obtained. That is the reason why many studies have focused on increasing these rates by different means. To participate in this search, we have carried out a screening to look for new drugs that might have an impact in the DNA repair pathway choice. A drug screening for compounds affecting DNA resection efficiency Since the generation of a KI sequence is mediated by HR, we hypothesized the drugs that would promote DNA resection, and therefore, HR, would also increase CRISPR- mediated KI efficiency. First, we optimized a system based on the detection of RPA foci formation in U2OS cells seeded in 96-well plates. These cells were pre-treated with three different drug libraries that contained either anti-cancer, anti-diabetic or natural compounds and were exposed to 10 Gy irradiation afterwards. Then, an immunofluorescence was performed on these plates and detected using a high-throughput microscope. A schematic layout of the process is represented in Figure R13A. To carry out the experiment, several controls were included: first, the vehicle of each drug (DMSO or water were used to normalize the results), second, as a positive control, the neddylation inhibitor drug (MLN2449) that we have previously demonstrated that increases DNA resection (Jimeno et al., 2015) and finally, as an additional positive control we also included an siRNA depleting the bona fide DNA resection factor CtIP. Each experiment was repeated independently at least twice. Using this system with the 552 drugs tested, we obtained several candidates (Table R1) with an effect in DNA repair balance. We detected 61 drugs that increased DNA end resection (green ellipse, see Fig. R13B) whereas 62 decreased DNA resection (red ellipse, Results-Chapter II 84  see Fig. R13B). The selected candidates (those with RPA foci formation efficiency over 1.5 or below 0.85 of the levels obtained with the mock treatment in at least 2 repeats) are listed in Table R2. The numeric results obtained for those candidates and of the controls indicated above are shown in Figure R14. Figure R13. Drug-based screening of compounds affecting RPA formation efficiency A) Schematic representation of the workflow of the screening. Previously seeded cells were treated for one hour with each one of the components of the indicated drug libraries, irradiated with 10 Gy and incubated for an additional hour. After performing an immunofluorescence for RPA, images were acquired and analysed. The compounds that skewed the balance over 2-fold or under 0.85-fold in this assay were selected as positive candidates to further progress in the validation of the screening and be, finally, tested as enhancers of CRISPR-Cas9 mediated KI efficiency. B) Representation o f the results of the screening. The mean of the number of RPA foci per cell of, at least, two independen t experiments relativised to their control is represented. A green ellipse circles the candidates with a p ossible enhancing effect on DNA resection. On the contrary, a red ellipse circles the ones apparently impairing DNA resection. B U2OS cells Drug treatment RPA foci immunofluorescence Drugs 1h 1h 10 Gy IR Anti-cancer library Natural products library Anti-diabetic library Positives candidates Impact on CRISPR-Cas9 KI efficiency? A 8,5 7,5 8 7 6,5 6 5,5 5 4,5 4 3,5 3 2,5 2 1,5 1 0,5 0,2 Number of RPA foci relative to control Drugs promoting DNA resection Drugs inhibiting DNA resection Results-Chapter II  85  Table R1. Candidate compounds with a mean RPA foci formation efficiency over 2-fold or under 0.85-fold COMPOUND REP1 REP2 REP3 MEAN Obatoclax mesylate (GX15-070) 13.0125564 2.94957302 7.9810647 IWP-2 8.23432425 4.17502986 6.20467706 BAY 11-7082 (BAY 11-7821) 3.82323012 4.46919164 4.14621088 Dexamethasone acetate 1.9498026 5.77044005 3.86012132 PCI-32765 (Ibrutinib) 2.49655107 5.18867904 3.84261506 GF109203X 2.24778224 5.23458376 3.741183 CNX-2006 4.21524092 3.09523808 3.6552395 Torin 2 2.06928353 4.86069152 3.46498752 GW4064 3.80631321 2.73721605 3.27176463 Dacomitinib (PF299804, PF- 00299804) 3.15494416 3.25471659 3.20483038 Sotrastaurin (AEB071) 3.76488391 2.57679034 3.17083712 AMG 900 1.97620712 4.32255819 3.14938265 VS-5584 (SB2343) 3.81607298 2.42160183 3.11883741 Dalcetrapib (JTT-705) 3.1424422 3.05643471 3.09943846 Sirtinol 2.92936847 2.93022639 2.92979743 Cabozantinib malate (XL184) 3.4683126 2.17929125 2.82380192 GSK2606414 3.20502611 2.2761858 2.74060595 PF-543 1.88168507 2.98427654 2.4329808 Pazopanib 1.2114414 3.64121678 2.42632909 Romidepsin (FK228, Depsipeptide) 2.14778303 2.54803132 2.34790717 TPCA-1 1.34172541 3.23327599 2.2875007 INK 128 (MLN0128) 1.51916431 3.04111025 2.28013728 GSK923295 2.04637396 2.48121909 2.26379653 (S)-10-Hydroxycamptothecin 1.81296816 2.61470275 2.21383546 AZ628 2.23723596 2.18266307 2.20994951 CHIR-99021 (CT99021) HCl 2.29811273 2.06721981 2.18266627 PFK15 3.10956441 1.22641502 2.16798971 Anagrelide HCl 2.09856588 2.20284128 2.15070358 Chrysophanic acid (Chrysophanol) 2.01230822 2.26439668 2.13835245 Z-VAD-FMK 1.39150183 2.77557096 2.0835364 PH-797804 1.64296655 2.51755736 2.08026195 Arbutin 2.65405405 1.33363513 2.18675722 2.0581488 Geldanamycin 1.376905 2.72147338 2.04918919 Lisinopril 1.0374315 1.57242102 3.43803056 2.01596103 Hordenine 0.95537822 2.1593602 2.87266553 1.99580132 Deforolimus (Ridaforolimus) 1.8135221 2.15314182 1.98333196 GW3965 HCl 1.21749876 2.66652345 1.94201111 Mycophenolic (Mycophenolate) 2.04737667 1.80782649 1.92760158 LY2603618 (IC-83) 1.78603927 1.87758293 1.8318111 Teniposide (Vumon) 2.0505313 1.53989941 1.79521535 DAPT (GSI-IX) 1.77945572 1.79313286 1.78629429 Sinomenine 0.89837838 1.85785423 2.57555178 1.77726146 Etoposide (VP-16) 0.55059795 2.9586515 1.75462472 IOX1 1.38800497 2.07733487 1.73266992 LDE225 (NVP-LDE225, Erismodegib) 1.20804726 2.24840419 1.72822572 Yohimbine HCl 2.16586824 0.84552163 2.07385399 1.69508129 Bosentan 1.41787661 1.96155843 1.68971752 (+)-Usniacin 1.68216216 0.87369851 2.41595925 1.65727331 Silymarin 0.57297297 1.82797646 2.51782683 1.63959209 Apocynin 1.37249807 1.8993725 1.63593528 Mitoxantrone HCl 1.79566353 1.44233315 1.61899834 Hematoxylin 1.08564125 1.6734282 2.08488964 1.61465303 Tie2 kinase inhibitor 1.61510669 1.59897191 1.6070393 Tetrahydropapaverine HCl 1.25837838 1.70574921 1.83531409 1.59981389 Pioglitazone (Actos) 1.894971 1.25733294 1.57615197 Silibinin 1.12864865 1.79538253 1.71816638 1.54739919 Nalidixic acid 0.38594595 1.61158895 2.63327674 1.54360388 Topotecan HCl 0.08614865 2.97206864 1.52910865 Cladribine 0.87884802 2.17700289 1.52792546 Osthole 0.51675676 1.78451788 2.25806452 1.51977972 Puerarin 0.7772973 1.89859665 1.85229202 1.50939532 Results-Chapter II 92  autophagy and apoptosis (Liu et al., 2013). (-)-Epigallocatechin gallate (PubChem CID 65064) is a phenolic antioxidant (found in green and black tea) that seems to impair DNA resection in our screening. It displays antineoplastic effects by inducing apoptosis and cell cycle arrest, inhibition of metastasis and tumour cell growth (Braicu et al., 2013; Han et al., 2011; Zan et al., 2019). Idarubicin HCl (PubChem CID 636362) is an anthracycline from the anti-cancer library that reduces DNA resection efficiency. It is considered a TOP2 poison, thus, impairing DNA replication and transcription (Bigioni et al., 1994; Dartsch and Gieseler, 2007). Irinotecan HCl trihydrate (PubChem CID 60837) is an anti-cancer compound that reduces RPA foci formation efficiency as well. It is the hydrochloride salt of camptothecin, a cytotoxic topoisomerase I inhibitor that stabilizes the cleavable complex between TOP1 and DNA, resulting in DNA breaks that inhibit DNA replication and trigger apoptotic cell death. Its cytotoxic activity is considered S-phase specific since it needs ongoing DNA synthesis (Thomas and Pommier, 2019). Merbarone (PubChem CID 4990817) is a TOP2 inhibitor from the anti-cancer library with DNA resection inhibiting activity. It is a TOP2 catalytic inhibitor that neither intercalate DNA nor stabilize TOP2cc. It is known to prolong cell cycle progression, retarding S phase and arresting G2 phase (Fortune and Osheroff, 1998; Pastor et al., 2012). Table R3. Validated compounds selected for their impact on CRSPR-Cas9 efficiency COMPOUND LIBRARY Hordenine Natural products Lisinopril Anti-diabetic Dimesna Anti-cancer AZ628 Anti-cancer GF109203X Anti-cancer IWP-2 Anti-cancer Irinotecan HCl Trihydrate Anti-cancer Idarubicin HCl Anti-cancer Merbarone Anti-cancer Torin2 Anti-cancer (-)-Epigallocatechin gallate Anti-cancer Results-Chapter II  93  Analysis of induced DNA damage signalling in the selected candidates Since some of the final candidate compounds selected were DNA damage inducers, we studied the intensity of the DNA damage marker γH2AX to rule out that the effect obtained in the RPA foci formation assay was due to a change in the DNA damage signalling (Fig. R17). As observed, from the candidates increasing DNA resection efficiency, only IWP-2 induced a significant uprise in the intensity of the DNA damage levels detected. Regarding the compounds that impaired DNA resection, Torin2 caused a dramatic fall in the levels of γH2AX. Being Torin2 an inhibitor of the DDR initial signalling step by blocking ATM/ATR/DNA-PK activity, this result was expected. Surprisingly, we detected a big drop as well with the treatment of TOP1 inhibitor Idarubicin HCl. Since this candidate induces SSBs that can potentially turn into DSB after DNA replication, this result came unexpectedly. In addition, we tested cell viability after the treatment with the selected candidate compounds using an MTT assay. Although with the conditions stablished a slight increase could be appreciated in some of the treatments, none of them rendered a significant change in cell viability. (Fig. R18). Figure R17. DNA damage signallin g levels induced b y the positive candidates from the screening In the left, γH2AX intensity levels in U2OS cells upon DSBs induction by ionizing radiation (10 Gy) and drug treatment. Other details as in Fig. R2B. In the right, representative images of the candidates from the left showing a significant change compared to the control. γ DMSO γH2AX DAPI Idarubicin HCl IWP-2 Torin2 Results-Chapter II 94  Analysis of the effect of DNA resection-modulating drugs in CRISPR-Cas9 efficiency The main aim of performing this drug screening was looking for new compounds that could enhance the low efficiency of CRISPR-mediated KI efficiency. To assess that, we used the HEK-293 CAT-R cell line. A more detailed explanation of this assay can be found in the Materials and Methods section. Briefly, the repair of a Cas9-induced DSB in a GFP gene linked to a mCherry sequence renders different outcomes depending on the repair pathway undertaken (Fig. R19A). In case of the repair pathway inducing indels, we would get a loss of the GFP fluorescence and, therefore, a KO. In addition, hyper-resection of the break could lead as well to the loss of fluoresce of GFP and mCherry, also generating a KO. On the contrary, if the DSB is repaired using the ssODN donor template added by SSTR, the repair would lead to the loss of GFP, but not mCherry, and the acquisition of BFP fluorescence, thus providing a measure of KI efficiency. Finally, error-free repair or untransfected cells would show both a GFP and mCherry positive profile. As controls of the experiment, DMSO, H2O and MLN2449 were included. As expected, treatment with the neddylation inhibitor MLN2449, which is known to favour DNA resection and, in consequence, HR, produced a significant increase in KI efficiency, as measured by the increase in the percentage of BFP positive cells. This result validates Figure R18. Positive candidates selected do not indue cell cytotoxicity A) MTT assay showing the viability of U2OS cells treated with 1 µM of the positive candidates from the screening for 2h, normalized to their vehicle (DMSO or water) and expressed as percentage. Statistical data as in Fig. R2B. Results-Chapter II  95  Figure R19. Drug effect on CRISPR-Cas9 efficiency measured by the CAT-R reporter A) Schematic representation of the CAT-R reporter showing the different possible DNA repai r outcomes after a Cas9-mediated DSB. The CRISPR/Cas9 target site is indicated at the eGFP locus. If the DSB is resolved through repair leading to small InDels generation, frameshift mutations will p roduce a KO of eGFP, and only the mCherry will be expressed; if the break is repaired with large deletions both mCherry and eGFP sequences will be lost and no fluorescence will be observed. Fo r simplicity, double negative population are referred as large deletions, but they might include additional classes of rearrangements as well. Finally, if the DSB is repaired by SSTR using an exogenous ssODN bearing the nucleotide change needed to turn eGFP into BFP, BFP positive cells will be produced. B) Quantification of the flow cytometry analysis for HEK293 CAT-R cell lines 72 h post-transfection with the synthetic gRNA and drug addition. In the box and whisker plots, centrelines mark the medians, dots indicate each replicate from the experiments, box limits indicate the 25th and 75th percentiles, and whiskers extend to minimum and maximum, showing all points. The red line indicates the level of the mock treated control. Data are derived from, at least, three independent experiments. Statistical significance was determined with an ANOVA test. *p<0.05, **p<0.01, ***p<0.001. B A mCherry P2A eGFP mCherry P2A eGFP mCherry BFP mCherry P2A eGFP mCherry P2A eGFP mCherry P2A eGFP mCherry P2A P2A eGFP DSB DSB repair 5’ 3’ ssODN donor template Small InDels Large Deletions Untransfected/ Error-free Single Strand Template Repair (SSTR) < 50-bp indel > 250-bp indel Cas9 target * * *** *** *** *** *** * Small Indels Large deletions SSTR Error-free/Untransfected % Population relative to control Results-Chapter II 96  the assay performed. Most of the candidates, namely merbarone, IWP-2, irinotecan HCl trihydrate, hordenine, AZ628, (-)-epigallocatechin gallate, lisinopril or bisindolylmaleimide I, did not produce any change in any of the repair products or the variability was too high to consider it reproducible (Fig. R19B). On the contrary, a shift in, at least, one of the fluorescent cell populations obtained was observed when cells were treated with dimesna, torin2 and idarubicim HCl. In the case of dimesna, there is a non-significant drop in the KI efficiency and KO efficiency by the generation of indels. This is in accordance with the decreased DNA resection efficiency observed in Figure R17. Nevertheless, it significantly either promotes the error-free repair of the DSB or hinders cell transfection of the gARN and the ssODN. Surprisingly, torin2 promoted BFP KI efficiency despite hampering DNA resection in our previous experiments, probably reflecting additional effects on repair through ATM/ATR independent of resection. In addition, it induced a fall in the percentage of cells undergoing large deletions due to the DSB repair. Finally, Idarubicin HCl decreased BFP positive and only Cherry positive cell populations, consistent with its observed role decreasing RPA foci formation efficiency. Furthermore, there was a raise in the induction of large deletions and error-free/untransfected cells. To further validate the influence of the compounds in CRISPR-Cas9 KI efficiency and to clear some contradictions between the effect in the RPA foci formation assay and the KI efficiency measured by the CAT-R system, we employed another assay to uniquely measure KI efficiency. This time a plasmid carrying both the Cas9 and gRNA were electroporated into U2OS cells with a GFP-bearing plasmid with homology arms complementary to the RAG1 gene, where the DSB would be induced. Then, the measurement of the different percentages of GFP positive cells would represent the efficiency of CRISPR-Cas9 mediated KI (Fig. 20A). Once more, MLN2449 was used as a positive control of the performing of the assay and, as anticipated, increased KI efficiency was observed. Contrary to our expectations, there was no reproducibility in the results obtained with the candidate compounds in this assay compared to the ones of CAT-R reporter (Fig. 20B). Here, lisinopril is the only candidate which increased KI efficiency while it was not evidenced as a possible influencer in the CAT-R system. Similarly, AZ628 and bisindolylmaleimide I also seemed to affect the integration efficiency of GFP but in the opposite direction to the previous observation, lowering the efficiency. The same effect was Results-Chapter II  97  obtained with torin2. It is worth noticing that idarubicin HCl could not be assessed with this assay since the concentration used in previous experiments was too cytotoxic for the new settings applied. So, we conclude that the selected drugs might affect DSB repair and KI efficiency in vivo, but the variability between replicas and reporters was too high to render conclusive results. This might be due to the pleiotropic effects of these drugs. So, we reasoned that further refinement was required to isolate drugs with potential. Merbarone inhibits DNA resection One striking observation from the group of candidates obtained from the screening is that is enriched in topoisomerase inhibitors, specifically, irinotecan HCl trihydrate (TOP1 inhibitor), idarubicin HCl (TOP2 inhibitor) and merbarone (TOP2 inhibitor). The inhibition of topoisomerase is of great interest for cancer treatment because it impedes the repair of the break induced and, thus, promotes DNA damage which, ultimately, would lead to cell Figure R20. Drug effect on CRISPR-Cas9 mediated KI efficiency A) Schematic representation of the assay used to assess KI efficiency by CRISPR-Cas9. A dono r plasmid carrying the GFP sequence flanked by homology arms to the RAG1 gene was coelectroporated into U2OS cells with another plasmid bearing the Cas9 nuclease and the gRNA targeting RAG1 gene. Then, cells were seeded and treated with the candidate drugs. 72h pos t transfection, the percentage of GFP positive cells was measured by flow cytometry. B) Quantification of the KI efficiency of GFP gene represented by the percentage of GFP positive cells measured by flow cytometry. Other details as in Fig. R2B. AB *** *** * * * RAG1 GFP + Drug Results-Chapter II 98  death. Many drugs used as chemotherapeutics work inhibiting DSB repair pathways, either NHEJ or HR, since they are usually mutated in many cancer cells (Chernikova et al., 2012; Luo et al., 2022; Patterson-Fortin et al., 2022; Xie et al., 2022). Considering the topoisomerase inhibitors found in our drug screening impaired DNA resection (Fig. R16), the initial step of HR, we decided to study the effect of topoisomerase inhibitors in this process in more detail. Since Irinotecan HCl Trihydrate gave the opposite result in the screening compared to the validation, this candidate was rejected. Turning to the TOP2 inhibitors idarubicin HCl and merbarone, both drugs block TOP2 activity using different mechanisms. While idarubicin HCl is considered a TOP2 poisons because it prevents TOP2 turnover after inducing the DSB (Bigioni et al., 1994), merbarone is a catalytic inhibitor of TOP2I and it blocks TOP2 activity before DNA cleavage is generated (Nitiss, 2009). So, in brief, idarubicin block TOP2 activity and creates DSBs, whereas merbarone inhibits the enzyme without breaking the DNA. Given that the generation of DSBs at the same time of TOP2 trapping could shade the real role of TOPOII in DNA resection, we decided to focus on merbarone as a DNA resection inhibitor. Merbarone effect in DNA resection is related to PIF1 helicase. As discussed in the introduction, topoisomerases have recently emerged as regulators of the homeostasis and function of G4s in human cells. In addition, a genetic CRISPR/Cas9 based multi-screen demonstrated that the cytotoxicity caused by pyridostatin, a G4-stabilising drug, is mediated by TOP2 trapping (Olivieri et al., 2020). Since in the previous Results chapter of this Thesis we showed that pyridostatin inhibited DNA resection in the same fashion as PIF1α depletion, we wondered if there was a connection between the effect of the TOP2 inhibitor merbarone in DNA resection and the role of G4s in this process. Thus, we decided to combine merbarone with the depletion of PIF1α helicase. As predicted, our results point to an epistatic effect in the RPA foci formation phenotype when both merbarone addition and PIF1α depletion are combined (Fig. R21A). From this data we can conclude that PIF1α and TOP2 are working in the same genetic pathway. Since both enzymes show a role in G4 resolution (Byrd and Raney, 2017; Yadav et al., 2014), we wondered if this was the common pathway in which they were collaborating to mediate Results-Chapter II  99  DNA resection. Indeed, the combined treatment of pyridostatin and merbarone was epistatic over the treatment with just one of them and in all the cases the overexpression of PIF1α suppressed the DNA resection defect caused by those treatments (Fig. R21B). To further support this genetic interaction, we performed a co-IP to test if TOP2 and PIF1α were physically interacting. Using a GFP-PIF1α fusion protein we were able to recover TOP2β Figure R21. Merbarone effect in DNA resection is related to PIF1α helicase activity A) Cells pre-treated with merbarone or DMSO as a control and transfected either with an siRNA against PIF1α or with a control siRNA (siNT) were irradiated and collected after 1 h and the presence of RPA foci was visualized by immunofluorescence. Other details as in Fig. R2B. B) Same as in A, but in cells transfected either with a plasmid carrying the GFP-PIF1α fusion or with a control plasmi d and treated or not with pyridostatin. Other details as in Fig. R2B. C) Anti-GFP Dynabeads were used to immunoprecipitate GFP-PIF1α and TOP2β from protein extracts of U2OS cells expressing GFPPIF1α or GFP. In the left, a representative image of the experiment is shown. The average and SD enrichment of two independent experiments are shown in the right. Results-Chapter II 100  in agreement with a physical interaction between both factors (Fig. R21C). All in all, we conclude that TOP2β activity was required to overcome some topological constrain that also requires PIF1α.      Discussion  108  functions of PIF1 in either the nucleus or the mitochondria (Chisholm et al., 2012; Li et al., 2021) and has been associated with an increased risk of breast cancer (Chisholm et al., 2012). Another DNA resection related protein with a high prevalence in breast cancer when mutated is BRCA1 (Jiang and Greenberg, 2015; Rosen and Walle, 2013). BRCA1 is a key protein in DNA resection and it is known to interact with other essential proteins involved in this process to recruit and activate them (Jiang and Greenberg, 2015; Yu et al., 2006). Interestingly, BRCA1 depletion has been associated with sensitivity to pyridostatin (Zimmer et al., 2016) thus, supporting a possible connection between this protein and G4s. We speculated that this connection could be through the activity of PIF1α, since they were both participating in DNA resection. In agreement with this shared role of BRCA1 and PIF1α in G4 unwinding, an increase in the immunoprecipitation efficiency of BRCA1 an PIF1α was detected after the pyridostatin treatment (Fig. R6A). Indeed, this interaction is direct as the use of BRCA1 recombinant protein also efficiently pulled down its recombinant PIF1α partner in an in vitro binding assay (R6C). In addition, further experiments to confirm the relationship between these two proteins show that they also interact genetically (Fig. R7A). As deduced from the study of PIF1α foci formation, BRCA1 depletion seems to impair PIF1α re-localization after IR, thus suggesting the requirement of BRCA1 to recruit this helicase to DNA damage. This is in agreement with a recent review that showed that BRCA1 haploinsufficiency might lead to defective G4 resolution (Groelly et al., 2022; Kim and Hwang, 2022) likely because PIF1α is no longer recruited to these structures. In accordance with the recruitment of PIF1α foci to pyridostatin target sites observed in Rodriguez et al., 2012, the pyridostatin treatment with no other source of DNA damage mildly increased the number of PIF1α foci. However, contrary to our expectations, in pyridostatin treated cells, PIF1α recruitment to DNA damage did not increase after irradiation (Fig. R7A). In fact, the effect seemed to be the contrary. The reason for this rather contradictory result is not yet completely clear, but a likely explanation could be the re-localization of PIF1α foci after irradiation. Initially, PIF1α might colocalize with G4s (Rodriguez et al., 2012) but, once we induce new DNA damage foci with irradiation, PIF1α have to re-localize to IR-induced foci. This is accordance with the significant raise in PIF1α foci detected in our IR experiments (Fig. R7A). However, in the presence of pyridostatin, PIF1α might be more tightly associated with stabilized G4s, hampering this re-localization. Discussion 109   It is possible that the timeframe of our experiment was not enough for the shift from stable G4s to the formation of new IR-induced PIF1α foci. This would result in a diminution of the number of foci observed. Further experiments with longer recoveries after IR treatment will clear this possibility. Up to this point, all the evidence confirms a role of PIF1α helicase in DNA resection. This DNA resection needed to undergo HR repair is not always initiated from broken DNA ends considered chemically “clean”. Indeed, DSB-DNA-ends may present proteins covalently attached to them, chemical modifications or secondary structures that might impede the initiation of the resection. Depending on the relative position of these roadblocks to the broken DNA ends, they may interfere with either the initial short-range resection or the following long-range resection. Since G4s are considered one of these roadblocks, their presence alongside the prone-to-be resected DNA sequence obstruct the resection. The requirement of their resolution to proceed with the resection satisfactorily explains the participation of PIF1α in both steps of the process as observed by the epistatic effect of the double depletion of PIF1α and factors involved in either long or short-range resection (Fig. R8). An important consideration when studying DNA resection is that it is regulated by cell cycle progression with many of the factors involved in this process being only expressed during S/G2 phases (Falck et al., 2012; Huertas et al., 2008; Tomimatsu et al., 2014). Thus, the significant cell cycle modifications produced by the depletion of MRE11 and DNA2 had to be taken into account. Therefore, we normalized the DNA resection efficiency with cell cycle profiles measured by the DNA content, considering the relative changes in the subpopulation of S/G2 cells compared to the control situation. Given that no significant change in the results was noticed after the normalization, we concluded that the phenotype observed is not attributed to any change in cell cycle distribution (Fig. R8C). In S.cerevisiae, Mre11 is indispensable to process “dirty” ends and initiate resection (Cejka, 2015). However, in our experiments performed in human cells, MRE11 downregulation did not show such an essential role for DNA resection initiation (Fig. R8C) suggesting the collaboration with accessory factors to overcome these impediments. Following this rationale, another explanation for the significant difference found in MRE11- PIF1α depleted cells could be the consideration of PIF1α as one of these accessory factors contributing to the processing of the ends by the removing of G4 structures. Discussion  110  The presence of G4s structures or PQS in the DNA sequence has long been proposed to act as barriers to many DNA transactions (see Introduction 2.2). Indeed it is well-known that they interfere with the progression of the replication machinery during DNA synthesis (Kaguni and Clayton, 1982; Kamath-Loeb et al., 2001). Surprisingly, the data obtained from our in vitro experiments pointed to the opposite direction. In fact, the introduction of two G4-forming sequences in the resected strand improved the DNA degradation efficiency of the DNA by both DNA2/BLM (Fig. R9B) and DNA2/WRN tandem (Fig. R9C). Remarkably, DNA resection by EXO1 was not promoted by the presence of the G4s compared to the control sequence (Fig. R10). This could be explained by the saturation of the reaction with high concentrations of EXO1 since the use of lower levels of this protein did suggest a stimulation of the process by G4s. The discrepancies observed with our initial hypothesis of G4s being a roadblock for resection can be attributed to a faster access of the resection proteins to the DNA due to the easier melting of the DNA duplex induced by the presence of the G4s close to the DNA ends. Indeed, other noncanonical DNA structures, such as DNA hairpins, seems to influence both the rate and efficiency of DNA hybridization by lowering the free energy required for their dissociation (Schreck et al., 2015). In line with this idea, no improvement was observed with DNA2/BLM after 60 min of reaction most likely because it was time enough for DNA2/BLM to unwind the DNA and equal the level of DNA degradation produced by the G4-bearing substrate (Fig. R9B). Further supporting this theory, when we removed the requirement of DNA duplex unwinding prior to DNA degradation by using a ssDNA substrate, it finally rendered the expected defect in DNA resection when G4s are present (Fig. R11B). It is noteworthy to mention that the formation of the G4 structure was not checked in these experiments. Therefore, another reasonable explanation of the promotion of DNA resection by G4s could be attributed to the presence of the PQS but not the structure itself, having this a different effect in the degradation efficiency. Additional experiments confirming the state of the potentially G4-forming sequences and positioning these sequences further from the DNA ends to avoid the influence on the opening of the duplex should be carried out to confirm this hypothesis. As mentioned in the Introduction, R-loops were described as another source of genome instability when not properly regulated, similarly to G4s. In fact, R-loops formation Discussion 111   is favoured by the presence of a G-rich sequence in the coding strand. Thus, the formation of an R-loop would generate a displaced G-rich strand with the potential to assemble a G4 structure. The concurrence of R-loops and G4s creates a novel structure that combines both of them and is called G-loop (Duquette et al., 2004). The current rationale indicates that the R-loop induced formation of a G4 stabilises in return the R-loops structure creating a very thermodynamically resistant structure (Lim and Hohng, 2020). In accordance to this data, both non-canonical DNA structures have parallel genome distributions (Kuznetsov et al., 2018) and even share similar cell functions. This evidence supports a synergistic activity between these two structures in their biological activity. In addition, stabilization of G4s using ligands increases R-loop formation in a similar fashion to G4 formation and endorse R-loop-mediated replication stress (Chappidi et al., 2020; De Magis et al., 2019; Kotsantis et al., 2020). One of the key roles of G4s is transcription regulation and, recently, the different outcomes of this regulation by G4s have been related to the formation of G-loops (Lee et al., 2020). Interestingly, the result obtained differed depending on which strand the G4 was located. Thus, if the G4 is present in the template strand, it inhibits transcription limiting the mRNA production whereas, if located in the coding strand, it induces R-loop formation in the template strand and, in consequence, promotes transcription. In line with this collaborative role of G4s and R-loops in transcription, it would be interesting to investigate if the observed effect in DNA resection by G4s is dependent on the strand in which they are located and if it is also mediated by R-loop formation. Indeed, an experiment performed in Camarillo et al., 2021 sheds some light in this regard. Apparently, the overexpression of RNase H1, an R-loop removing factor, seems to be enough to overcome the DNA resection defect produced by PIF1α depletion. Although our investigations in this area are still ongoing, this result suggests that R-loop resolution is enough to recover the defect attributed to G4 presence and, thus, support a combined activity of G4s and R-loops in DNA resection. Indeed, PIF1α is not only involved in G4 resolution but also in R-loops removal (Zhou et al., 2014). Thus, any of these structures, or the combination of both, can be the relevant structure that PIF1α must unwind for an efficient DNA resection. Discussion  112  Apart from PIF1α role in DNA resection unwinding complicated substrates that might interfere with this process, this helicase might be further involve in the DDR. This is supported by Prados-Carvajal et al., 2020 that uncovers the role of a variant of PIF1α obtained by alternative splicing. This variant lacks the helicase activity and it is not require for the early events of DNA repair but its required to promote survival upon exposure to damaging agents. Another remarkable characteristic of DNA resection is the conserved coordinated activity of a helicase with a nuclease to initiate HR. In Escherichia coli, the RecQ helicase and the RecJ exonuclease cooperate in DSB resection (Amundsen and Smith, 2003) and, the human equivalents, BLM helicase and DNA2 also need to act coupled to process DNA ends (Sturzenegger et al., 2014). The correlation of nucleases with their specific helicase is difficult to determine because there are many candidate proteins that could be eligible as the right partner. Since we demonstrated that PIF1α is a potential important helicase in DNA resection, its association with a nuclease to efficiently exert its role in DNA resection over G4s prone structures was considered logical. In a similar fashion to PIF1α, the nuclease FAN1 seemed to be involved in HR and G4 resolution (MacKay et al., 2010; Porro et al., 2017). FAN1 is a protein with endo and exonuclease activity important for the HR step of inter-strand-crosslink repair. Although previous studies showed that FAN1 depletion led to a delayed disappearance of RAD51 foci, they did not observe a role for this nuclease in DNA resection (MacKay et al., 2010). In contradiction with these findings, we detected a defect in DNA resection, measured by RPA foci formation, after FAN1 downregulation (Fig. R12A). Furthermore, this depletion seemed to be epistatic with PIF1α depletion since the double-downregulation of both proteins rendered the same defect level as the single depletion of them (Fig. R12A). This discrepancy could be explained by the differences in the experiment conditions used, such as the source of DNA damage (cisplatin versus IR) and the recovery time after the damage induction (1h vs 24 h). Indeed, from this data, it could be inferred that there is a delay in resection initiation and RPA removal from ssDNA when FAN1 is depleted. That would be the reason why there are less RPA positive cells after 1 h but, 24h latter, the percentage of RPA foci is higher. In addition, although the analysis of RPA foci formation is considered the golden-read out of DNA resection, it is a low resolution technique that gives information about the number of DSBs that initiate Discussion 113   resection but not about the extension of this resection (Cruz-García et al., 2014). Strikingly, using the SMART technique, which overcomes this throwback, we observed a rise in resected fibers length after FAN1 depletion (Fig. R12C). A similar phenotype is observed in Domingo-Prim et al., 2019 when EXOSC10 is depleted, inhibition of RPA foci formation and further extension of ssDNA resected fibers. A possible explanation might be that FAN1 is participating in both branches of DNA resection. In the short-range resection, it may act as a scaffold to recruit proteins during this first step but then, it is needed to stop the resection extension. Therefore, its removal from the DNA sequence promotes the extension of the resected fibers. More data elucidating the role of FAN1 in DNA resection should be collected from experiments widening the settings used. Maybe the comparison of different sources of DNA damage and different time points could clarify this matter. Despite our results not being solid enough to stablish the role of FAN1 in DNA resection, they do indicate certain relationship between PIF1α and FAN1 in this process. Apart from their co-downregulation being epistatic, as already mentioned, PIF1α overexpression seems to recover the resection initiation deficiency observed in cells lacking FAN1 (Fig. R12E). Although these results are preliminary, they might help to further elucidate how the cell deal with G4 structures during DNA resection. 2. New screening for new factors affecting DNA resection Given the relevance of CRISPR-Cas9 in genome editing, different approaches have been applied to cope with the low-efficiency of CRISPR-Cas9 driven HR. This, next to its hight off-target effect frequency, strongly affects its consideration as the ideal gene-editing tool. Indeed, this inefficiency hampers the generation of KI, limiting the modifications that can be introduced in the genome. As already discussed in the Introduction, some of the strategies developed to overcome this limitation include modifying CRISPR-Cas9 components, such as Cas9 variants, gRNA modifications or improved donors, or controlling the expression of certain proteins that might redirect the repair towards HR (Ghanta et al., 2021; Hu et al., 2018; Méndez-Mancilla et al., 2022). Nevertheless, despite these attempts greatly contributing to CRISPR-Cas9 development, they might be unpractical for a common Discussion  114  use in the lab since they require an additional adjustment of the original system. That is why the identification of small-molecule compounds that enhance CRISPR-Cas9 efficiency is a promising approach to easily and reversibly favour HR. Among the studied compounds to improve CRISPR-Cas9 HR, we can find a wide variety of targets. They can either modify the chromatin status to make the DNA ends more accessible, affect cell cycle progression to arrest cells in G2/S phases, when HR machinery is available, or inhibit NHEJ factors (Lin et al., 2014; B. Liu et al., 2020; Maruyama et al., 2015; Wienert et al., 2020). In general, the main aim of these compounds is to disrupt the balance between the two main DSB repair pathways towards HR. In line with this idea, our screening for factors that could be latter used to alter CRISPR-Cas9 efficiency was based on the search of compounds that could influence DNA resection. In addition, the screening was performed using three different libraries: anti-diabetic, natural and anti-cancer compounds. The majority of the chemotherapeutics agents are based on the fact that one of the hallmarks of cancer is the exacerbated genetic instability that increase the mutation rate. As a consequence, the loss-of-function mutations in key proteins for cell biology, such as the DNA repair machinery, would obligate the cell to find an alternative for the lost pathway. Thus, targeting these alternative repair pathways using DSB repair proteins inhibitors is becoming an interesting strategy to selectively kill cancer cells (Kaelin, 2005). That is why it did not come as a surprise the fact that most of the candidates obtained in the screening came from the anti-cancer library (Fig. R16). Some very-well known chemotherapeutic agents are TOPO inhibitors that induce DNA damage and block DNA repair (Hevener et al., 2018; Pommier et al., 2022; Sunter et al., 2010). Unexpectedly, three of our candidates were TOPO inhibitors that impaired DNA resection. The mechanism of action of these drugs in hindering DNA resection is latter further discussed. Contrary to our expectations for a group of compounds with a DNA-damage inducing activity, the treatment with any of these TOPO inhibitors candidates did not increase γH2AX intensity. In fact, the use of idarubicin HCl even rendered a reduced intensity of this histone variety , in agreement with the observed histone eviction induced by anthracyclines, such as doxorubicin, but not by other TOPO inhibitors, such as etoposide (Pang et al., 2013). Discussion 115   For the rest of the candidates, we could not find a common pathway that would relate their activity with their role in DNA resection. Despite each of them affecting different mechanisms, a general explanation for their effect in DNA repair was the inhibition of a pathway involved in HR. For example, IWP-2 inhibits Wnt signalling pathway. It has been described that this signalling upregulated LIG4 expression which is required for NHEJ (Chappidi et al., 2020). Therefore, its inhibition could be promoting resection initiation due to the absence of a proper NHEJ repair. Another example is the inhibition of mTOR signalling by Torin2 which seemed to impair DNA resection efficiency. Indeed, it has been previously shown that torin2 treatment affected CtIP foci formation and greatly delayed γH2AX foci formation and dissolution, in agreement with our results (Udayakumar et al., 2016) (Fig. R17). Regrettably, not all our candidates produced the same effect as described in the bibliography. This is the case of AZ628, a BRAF inhibitor. Apparently, abolishing BRAF reduced ATM and Chk1 activation which would lead to an inefficient DNA repair (Colomer et al., 2019). However, we observed the contrary effect as an increase in RPA foci formation was detected. Similarly, (-)-Epigallocatechin gallate was expected to favour DNA resection since, in yeast, was described to increase the expression of HR genes (Chong et al., 2019). Contrary to this research, we obtained a decrease in DNA resection efficiency in cells pretreated with this compound. For the rest of compounds, the process in which they participate have not been previously linked to an effect in DNA repair, such as Hordenine, Bisindolylmaleimide I or the anti-diabetic activity of Lisinopril. Regarding dimesna, its role in DNA resection could be attributed to its participation in acrolein metabolism. Acrolein is an environmental contaminant also produced endogenously by lipid peroxidation (Chung et al., 1996). This metabolite induces DNA damage and causes an inhibitory effect on DNA repair (Sarkar, 2019; Tang et al., 2011). Therefore, the alteration of its metabolism regulation could influence the DNA repair landscape, as observed by dimesna treatment. Further analysis will be required to define the molecular mechanisms and biological meaning behind these factors in DNA resection. Discussion  116  Regardless of the mechanism by which our drug candidates influence DNA resection, their effect in CRISPR-Cas9 efficiency was our real interest. The use of a colourbased system to trace the repair outcomes after the induction of a targeted DSB by CRISPR- Cas9 (Roidos et al., 2020) was employed to assess the influence in this gene editing tool. Unfortunately, although the use of MLN2449 as a control did produce the expected effect in CRISPR-Cas9 repair outcomes (increased KI efficiency), the results of the rest of the compounds obtained by this assay were not in line with the effects observed in the RPA foci formation assay (Fig. R19B). Indeed, only dimesna, torin2 and idarubicin HCl gave a significant phenotype in this reporter. Both dimesna and idarubicin HCl increased the population of cells which conserved GFP and mCherry expression. In other words, they increased error-free repair or impaired the transfection of the cells with the gRNA. Regarding the other parameters, torin2 and idarubicin HCl renders opposite results despite producing the same impaired DNA resection phenotype in the RPA foci formation assay. To explain this discrepancy, we should get a deeper understanding of the CAT-R reporter. Large deletions could be considered a product of DNA resection or other large rearrangements such as translocations, inversions or large insertions (Kosicki et al., 2018). This different causes leading to the loss of both GFP and mCherry signal cannot be differentiated with this reporter and thus, may explain why we obtain opposite effects. Torin2 could be blocking DNA resection but not influencing other kind of large deletions and, in consequence, reduce the large-deletion repair outcome. On the contrary, despite impairing the generation of resected tracks, the treatment with idarubicin HCl could be increasing the number of translocations or large inversions events leading to a net uprise of large deletions events. Indeed, the use of TOP2 inhibitors have been linked with increased chromosomal translocations and the TOP2-induced DSBs are also considered to be a source of genomic rearrangements (Gómez-Herreros et al., 2017; Stanulla et al., 1997). Further supporting this theory, we observe a decrease in the KI efficiency of the ssODN after the treatment with idarubicin HCl. This KI efficiency would represent the efficiency of SSTR, a type of DNA-resection dependent repair mechanism that is independent of RAD51 (Gallagher and Haber, 2021). This kind of repair is based in SSA and, therefore, is very sensitive to DNA resection. Changes in DNA resection efficiency would then be manifested in changes in SSTR, as the changes observed in idarubicin HCl treatment. Finally, we also Discussion 117   observe an increase in SSTR after torin2 treatment. This disagrees with our previous results and further data collection is required to discern the cause of this rise. Given that the focus of the study was on assaying the effect of some compounds in CRISPR-Cas9 mediated KI efficiency and the use of CAT-R reporter produced some confusing results, we validated the results with another reporter system. This time the DNA sequence template was provided by a plasmid, so the repair pathway measured was no longer SSTR but the classical HR mechanism. Once more, the use of MLN2449 confirmed the correct performance of the experiment since the use of this neddylation inhibitor increased the HR efficiency (Fig. R20B), as expected (Jimeno et al., 2015). In addition, in accordance with the results obtained by the RPA foci formation assay, lisinopril slightly increased CRISPR-Cas9 mediated KI efficiency, although the relevance of this increase is not very notorious. Similarly to previous results, torin2 treatment decreased HR efficiency which agrees with its role inhibiting DDR signalling (Liu et al., 2013). Surprisingly, AZ628 and Bisindolylmaleimide I decreased the integration efficiency of the template in opposition to what we expected from the increase in DNA resection efficiency observed after the treatment with these drugs. Since the drug treatment time is longer in this experiment compared to the RPA formation assay, some variations in the phenotypes were somehow expected and optimisation of the settings, such as lowering the concentration in compensation for the longer period of incubation with the compounds, should be performed. Here we stablished a protocol to search for new small molecule compounds that could influence the efficiency of CRISPR-Cas9 mediated genome editing. Although the settings stablished worked well in predicting the effect in this process by the controls, they were not able to render a clear reproducible result regarding the candidates selected. The selection of a different subset of candidates from the screening or further elucidating the working conditions of these compounds might give a more successful result in this aim.    VI. Materials and methods Materials and methods  127 1. CELL CULTURE PROCEDURES 1.1 Cell lines, growth media and conditions The main cell lines used in this thesis include the human osteosarcoma (U2OS), human embryonic kidney (HEK293) and several stable cell lines created from U2OS using different constructs (Table M1). HEK293 and U2OS cells were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM; Sigma, D6546), supplemented with 10% fetal bovine serum (FBS; Sigma, F7524), 2 mM L-glutamine (Gibco, 25030024), 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco, 15140122). Stable cell lines derived from the above-mentioned ones were cultured in the same conditions but adding either 0.5 mg/ml G418 (Sigma, A1720) or 1 μg/ml puromycin (Sigma, P8833) to the medium, depending on the selection marker of the inserted plasmid. For U2OS19ptight13 cells, which contain a doxycycline-inducible system (Table M2), DMEM without phenol red (Sigma, D1145) supplemented with 10% tetracycline-free FBS (Biowest, S181T) was used. HEPA class 100 incubators (Thermo) were used to maintained cell lines at 37ºC and 5% CO2. Trypsin-EDTA solution (Sigma, T4049) or accutase (Invitrogen, 00-4555-56) were used to detach the cells when required. To seed specific numbers of cells, an automatic cell counter was used (Z2 Coulter Counter, Beckman Coulter). For long-term preservation of the cells, they were harvested and pelleted by centrifugation at 500 g for 3 minutes. Cells were then resuspended in freezing solution (10% dimethyl sulfoxide [DMSO] in FBS), aliquoted in tubes and maintained at -80ºC for at least 24 hours. Then, they were transferred to liquid nitrogen-containing tanks for long-term storage. Materials and methods 128  1.2 siRNA transfection For transient protein knockdown, siRNA transfection was performed using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher, 13778500) and different plate formats, depending on the requirements of each specific experiment. Briefly, cells were seeded and grown for 24 hours. The day of transfection, medium was replaced by fresh DMEM without antibiotics and cells were incubated with a mix of siRNA and RNAiMAX diluted in Opti-MEM, according to manufacturer’s instructions. Cells were then incubated at 37ºC for 6 hours before replacing the media with fresh complete DMEM to minimise cell death. All siRNA-mediated knockdowns were validated 48 hours after transfection by western blot or quantitative RT-PCR. The list of all siRNAs used in this thesis can be found on Table M3. 1.3 Plasmid DNA transfection (Fugene) FuGENE HD Transfection Reagent (Promega, E2311) was used to introduce expression vectors (Table M2) and produce stable cell lines. For 60 mm plates, 300,000 cells were seeded the day before transfection. Then, 2 μg of plasmid DNA was mixed with FuGENE in a 3:1 FuGENE:DNA ratio and diluted in Opti-MEM (Gibco, 11058- 021). The mix was incubated for 15 min at room temperature and added dropwise to the plate with gentle rocking. 1.4 DSB induction by ionizing radiation To induce double strand breaks in the cells, an irradiator device emitting gamma rays (BIOBEAM GM 8000, Gamma-Service Medical GmbH) 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. The irradiator is adjusted to the dose rate so the cover surface does not exceed 3 μSv/h. Materials and methods  129 2. MOLECULAR BIOLOGY PROCEDURES 2.1 Nucleic acids manipulations 2.1.1 Plasmid DNA amplification Plasmid DNA was amplified by transforming competent DH5α Escherichia coli cells (Model Organism Service, CABIMER) using the heat shock protocol. Briefly, 0.1 ml of competent bacteria were mixed with plasmid DNA and incubated on ice for 30 min. Cells were then heat shocked by placing them at 42ºC for 35 seconds and incubated on ice for 5 min. Then, 1 ml of LB broth (Formedium, LB-Broth Lennox) was added, and the transformed cells were incubated at 37ºC for 30 min. Cultures were then harvested by centrifugation at 6,000 rpm for 2 min and plated in LB agar plates supplemented with 100 μg/ml ampicillin (Sigma, 9518) or 25 μg/ml kanamycin (Sigma, K4000) depending on the antibiotic resistance cassette of the transformed plasmid. Plasmid DNA was purified from one single colony using PureYield Plasmid Maxiprep (Promega, A2393), ZymoPURE II Plasmid Maxiprep Kit (Zymo research, D4203) or Midiprep (Promega, A2492) Systems, following the manufacturer’s instructions. DNA concentration was quantified by measuring 260 nm absorbance using a NanoDrop ND-1000 spectrophotometer. 2.1.2 DNA digestion with restriction enzymes For DNA cloning and plasmid checking, restriction endonucleases from Takara or New England Biolabs were used according to manufacturer's instructions. DNA fragments were resolved in agarose gels or non-denaturing polyacrylamide gels as described in the following sections. 2.2 Cloning strategies To produce the pUC19-2xLacO-2G4-nontemplate plasmid, two fragments bearing the c-myc promoter potential G-quadruplex sequence were cloned in tandem in the pUC19_2xLacO plasmid (Table M2). First, the G4 sequence was amplified from a previous plasmid from the lab (SA-GFP-G4_NruI, Table M2) by PCR using oligos containing either Materials and methods 130  the HincII and PstI sites as upstream and downstream primers, respectively, or the PstI and EcoRI sites. Then, both the pUC19 plasmid and the amplified G4 sequences were digested with HincII and PstI. The products of the restriction reaction were ligated by incubation with T4 DNA ligase (Takara, 2011A) at 15ºC overnight. To introduce the second G4 sequence in the plasmid containing already one G4, the same strategy was followed but using the PstI and EcoRI sites instead. 2.2.1 DNA electrophoresis in agarose gels DNA electrophoresis was performed on gels containing a variable percentage of agarose (Pronadisa, 8010.22) depending on the size of the bands to differentiate, and RedSafe (Intron Biotechnology, 21141) or Ethidium Bromide for DNA staining, both diluted in 1x TAE buffer (40 mM Tris-HCl pH7.6, 20 mM acetic acid, and 1 mM EDTA). Loading buffer (Takara) was added to DNA samples prior to loading in the gel, and 1 kb DNA ladder (gTPbio, GTPBM0002) was used for size estimation of the bands. Stained DNA fragments were visualized using an ultraviolet transilluminator (Bio-Rad) and analysed by Quantity One software. 2.2.2 DNA electrophoresis in non-denaturing polyacrylamide gel DNA electrophoresis to resolve fragments shorter than 200 bp was performed on 12% 29:1 acrylamide:bis-acrylamide gels under non-denaturing conditions. Loading buffer (1 mL 6X TrackIt Cyan/Orange Loading Buffer) was added to DNA samples prior to loading in the gel, and 50 bp DNA ladder (Thermofisher, 10416014) was used for size estimation of the bands. For DNA staining, the gel was incubated for 15 min in bromodeoxyuridine (BrdU) diluted in 1X TE buffer (10mM Tris-HCl containing 1mM EDTA), washed twice for 10 min with dH2O and visualized using an ultraviolet transilluminator (Bio-Rad) and analysed by Quantity One software. 2.2.3 Site-directed mutagenesis Directed changes in the DNA sequence of plasmids were performed using QuickChange Lightning Site-Directed Mutagenesis kits (Agilent Technologies, 210518) according to manufacturer's instructions. Briefly, mutagenesis was performed by PCR using Materials and methods  131 designed primers containing the desired mutation (Table M4). After the PCR, parental DNA was degraded by DpnI nuclease digestion. Finally, mutation-containing synthesized DNA was transformed into competent bacteria and amplified. Candidates were sent for DNA sequencing. 2.2.4 RNA extraction RNA extracts were obtained from cells using NZY Total 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. To remove trace amounts of DNA, 1 μg RNA was treated with RQ1 RNase-Free DNase (Promega, M6101) according to manufacturer's instructions. 2.2.5 Reverse transcription To synthesize complementary DNA (cDNA), 1 μg RNA was subjected to reverse transcription reaction using Maxima H Minus First Strand cDNA Synthesis kit (Thermo Scientific, K1652) according to manufacturer's instructions. 2.2.6 Quantitative PCR (qPCR) Quantitative PCR from cDNA was performed in triplicate to check siRNA- mediated knockdown of several proteins. For this, iTaq Universal SYBR Green Supermix (Bio-Rad, 172-5124) was used following manufacturer’s instructions. DNA primers used for qPCR are listed in Table M4. qPCR was performed in an Applied Biosystem 7500 FAST Real-Time PCR system. The comparative threshold cycle (Ct) method was used to determine relative transcripts levels (Bulletin 5279, Real-Time PCR Applications Guide, Bio-Rad), using the expression of the housekeeping gene β-actin as internal control. Expression levels relative to β-actin were determined with the formula 2-ΔΔCt (Livak and Schmittgen, 2001). 2.2.7 PCR PCR was performed using the indicated primers in each experiment (Table M4) and the Velocity polymerase (Bioline, BIO-21098) following manufacturer’s instructions. PCR Materials and methods 132  products were then purified using NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel, 740609.50). 2.3 Protein analysis 2.3.1 Protein extraction under denaturing conditions Protein extraction was carried out by lysing the cells in Laemmli buffer 2x (125 mM Tris-HCl pH 6.8, 4% SDS, and 20% glycerol) using a plastic cell scraper (Sarstedt, 83.1830). Alternatively, cells were harvested with trypsin, rinsed with PBS and resuspended in Laemmli 2X buffer. To reduce sample viscosity, protein extracts were passed through a syringe with a 0.5x16 mm needle (BD Plastipak, 303175) at least 10 times. Protein concentration was determined by the measurement of absorbance at 280 nm using a NanoDrop ND-1000 spectrophotometer. 2.3.2 Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) Protein samples were resolved by size in 29:1 acrylamide:bis-acrylamide gels prepared at different concentrations according to the molecular weight of the proteins to analyse. SDS-PAGE was performed according to previously described method (Laemmli, 1970). Samples were diluted to obtain similar protein concentration and then SDS-PAGE loading sample buffer 4x (250 mM Tris-HCl pH6.8, 8% SDS, 40% glycerol, 20% βmercaptoethanol (Sigma, M6250) and bromophenol blue) was added to a final concentration of 1x. Samples were incubated at 100ºC for 5 min and then loaded into the gels. Pre-stained protein ladder (gTPbio, GTPBM003) was also loaded as a molecular weight marker. Electrophoresis was performed in a Mini-PROTEAN Tetra Cell (Bio-Rad) with running buffer (25 mM Tris-HCl pH 8.3, 190 mM glycine, and 0.1% SDS) at 100-150 V. 2.3.3 Western blot analysis After electrophoresis, proteins were wet-transferred using Mini Trans-Blot system (Bio-Rad) for 2-3 h 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 activated in methanol for 1 min and equilibrated in transfer buffer. Materials and methods  133 Commercial Odyssey Blocking Buffer (LI-COR Biosciences, 927-40000) was used for blocking the membrane for at least 1 hour at room temperature to reduce background signal with the antibodies, and then, membranes were incubated overnight at 4ºC with the appropriate primary antibodies (Table M5) diluted in Odyssey Blocking Buffer containing 0.1% Tween-20. Then, membranes were washed three times for 5 minutes with 0.1% Tween-20 in TBS, followed by 1 hour incubation at room temperature protected from ambient light with the corresponding IRDye secondary antibodies (Table M6) diluted in blocking buffer containing 0.1% Tween-20. Afterwards, membranes were washed again three times with 0.1% Tween-20 in TBS and dried 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. 2.3.4 Protein A immunoprecipitation To immunoprecipitate endogenous BRCA1, U2OS cells were harvested in lysis buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.2% Triton X-100, 1X protease inhibitors (Roche), 1X phosphatase inhibitor cocktail 1 (Sigma)) and incubated with 100 U/ml Benzonase (VWR, 70746-4) 30 min on ice to degrade DNA. 1 mg of protein extracts was then precleared with 50 uL of washed magnetic protein A Dynabeads (Novex, 10002D) under gentle agitation at 4ºC for 1h. Precleared samples were then incubated with 1 μg anti-BRCA1 antibody (Table M5) or an equivalent amount of mouse IgG (Sigma, I8140) as negative a control for 30 min at 4ºC. The remaining dynabeads were added afterwards to the mixture of proteins and antibodies to be incubated together overnight at 4ºC with gentle agitation. Beads were then washed three times with lysis buffer, and the precipitate was eluted in Laemmli buffer and resolved in SDS-PAGE as described in the previous sections. 2.3.5 GFP immunoprecipitation (IP) U2OS cells expressing GFP or GFP-PIF1 were subjected to protein extraction under native conditions by scrapping with an immunoprecipitation lysis buffer (10 mM Tris-HCl Materials and methods 140  manufacturer’s protocol in Varioskan Flash (Thermo Electron corporation) and normalized with control. 3.4 Fluorescent protein analysis by flow cytometry 3.4.1 CAT-R reporter To study the effect of the compounds selected from the drug screening on CRISPR- mediated knock-out (KO) and knock-in (KI) efficiency, the HEK293 cell line bearing the CAT-R (Color Assay Tracing-Repair) reporter (Roidos et al., 2020) was used. 3.4.1.1 Description of the CAT-R reporter The CAT-R reporter (Fig. M2) consists of two coding sequences for the fluorescent proteins mCherry and eukaryotic green fluorescent protein (eGFP) linked with a selfcleaving P2A peptide. This reporter was then integrated at a single genomic locus in the HEK293 cell line to express doxycycline inducible Cas9. Basically, a DSB is induced by CRISPR-Cas9 using a guide RNA (gRNA, Table M4) targeting the eGFP locus located 355 bp downstream of the P2A peptide. The distinct pathways used to repair this DSB can generate three different populations that differs from each other in their fluorescent signals: 1) Small InDels repair-derived frameshift mutations lead to loss of GFP signal (mCherry+/GFP-). In most cases, this small InDels are expected to be due the action of c-NHEJ or alt-EJ. The percentage of cells with this outcome product could be considered as the frequency of KO efficiency. 2) Deletions or other rearrangements larger than ~420 base pairs (bps) lead to loss of both mCherry and eGFP signal (mCherry-/GFP-). The mCherry sequence is 420 bp away from the cutting site so the loss of its signals by large deletions are supposed to be products of DNA resection. 3) GFP-BFP conversion by single-strand template repair (SSTR) leads to gain of Blue Fluorescent Protein (BFP) signal by losing eGFP signal (mCherry+/BFP+/GFP-). eGFP can be converted to a BFP by a single amino acid change, allowing the measure of SSTR by providing a single-stranded oligo-deoxynucleotide (ssODN, Table M4) template together with the gRNA Materials and methods  141 targeting eGFP. The percentage of cells with this outcome product could be considered as the frequency of KI efficiency. 4) Untransfected cells/error-free repair (either by NHEJ or HR) leaves both mCherry and eGFP sequences intact (mCherry+/GFP+). The exact events that lead to the mCherry+/GFP+ population cannot be resolved with this reporter. 3.4.1.2 The CAT-R reporter in vivo Cells bearing a single copy integration of the CAT-R system were used to analyse the influence of a battery of drugs in the CRISPR-mediated KI and KO efficiency. To carry out the assay, 20000 cells/well were plated in U-bottom 96-well plates and the culture media was supplemented with 1 μg/ml Doxycycline (Sigma, D9891) to induce Cas9 expression. One day after seeding, cells were 60-80% confluent for transfection. The Alt-RR CRISPR- Cas9 crRNA and tracrRNA (IDT, 1072532) were used to form the guide RNA complex (gRNA, Table M4) targeting the eGFP sequence. To form the gRNA, each RNA oligo (Alt- RR CRISPR-Cas9 crRNA, tracrRNA) was resuspended in nuclease-free Duplex buffer (IDT, 11-01-03-01) to a final concentration of 100 μM. Then, the two RNA oligos were Figure M2. Schematic representation of the CAT-R (Color Assay Tracing-Repair) reporter In the CAT-R system, the different repair mechanisms of the CRISPR-induced DSB derives in different repair outcomes that can be traced by the fluorescent signal of the cell population. Small InDels lead to loss of GFP signal (mCherry+/GFP-). Deletions or other large rearrangements lead to loss of both mCherry and eGFP signal (mCherry-/GFP-). GFP to BFP conversion by single-strand template repair (SSTR) leads to gain of Blue Fluorescent Protein (BFP) signal by losing eGFP signal (mCherry+/BFP+/GFP-). Finally, untransfected cells or error-free repair leaves both mCherry and eGFP sequences intact (mCherry+/GFP+). mCherry P2A eGFP mCherry P2A eGFP mCherry BFP mCherry P2A eGFP mCherry P2A eGFP mCherry P2A eGFP mCherry P2A P2A eGFP DSB DSB repair 5’ 3’ ssODN donor template Small InDels Large Deletions Untransfected/ Error-free Single Strand Template Repair (SSTR) < 50-bp indel > 250-bp indel Cas9 target Materials and methods 142  mixed in equimolar concentrations to create a final duplex concentration of 3 μM in nuclease-free Duplex buffer. Finally, gRNA complex was heated at 95 °C for 5 min and then allowed to cool to room temperature (20-25 °C). We then transfected the gRNA targeting eGFP together with the single-stranded oligodeoxynucleotides (ssODNs, Table M4) as a template for GFP-BFP conversion. The ssODN was previously diluted in nucleasefree Duplex buffer to a final concentration of 100 µM. To transfect the cells Lipofectamine RNAiMAX transfection reagent (Thermo Fisher, 13778500) was used according to the user manual. The gRNA complex was combined with Lipofectamine RNAiMAX in a ratio of 2:1 in Opti-MEM Medium to a final concentration of 30 nM of the gRNA and 9 nM of ssODN. At the same time of the transfection with CRISPR gRNA, cells were incubated with 1 µM of the compounds selected (see Table R3) for three days to study their effect on CRISPR efficiency. Then they were analysed in a high-throughput FACS LSR Fortessa analyzer (BD Biosciences). These compounds were stored in −20 °C as 1 mM stocks in DMSO until needed. 3.4.2 GFP KI efficiency To specifically assess CRISPR-Cas9 mediated KI efficiency, U2OS cells were electroporated with CRISPR/Cas9-RAG1 and GFP donor-RAG1 plasmids using the Gene Pulser Xcell electroporation system (BioRad). First, 1.5x106 U2OS cells were harvested, washed with PBS and resuspended in 100 µL of electroporation buffer (2 mM HEPES pH 7.2, 250 mM D-Manitol, 1 mM MgCl2 and 15 mM K2HPO4). Then, 2.5 µg GFP donor- RAG1 and 0.6 µg CRISPR/Cas9-RAG1 were added to the mix of cells and electroporated (square-wavelength, 140 V, 6 pulses, 1.5 ms pulse length, 1.5 s interval length) using 0.2 mm cuvettes (Thermo Scientific, 10227570). Finally, approximately 280000 cells were seeded in 12-well plates and treated with 1 µM of the positive candidates from the screening or MLN2449, DMSO or H2O as controls. 72h after seeding cells were harvested and the percentage of GFP positive cells was detected using the FACS LSR Fortessa analyzer (BD Biosciences). Materials and methods  143 3.5 RPA/γH2AX foci immunofluorescence U2OS cells either downregulated or overexpressing different proteins or treated with a variety of compounds, as indicated in each specific case, were seeded on coverslips (Labolan, 20012). Cells growing on coverslips were irradiated (10 Gy) or mock treated, incubated for 1 hour to allow foci formation, washed once with cold PBS and collected. Coverslips were treated with pre-extraction 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) for 5 min on ice 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, washed three times with PBS and incubated for 1 h with blocking buffer (5% FBS in PBS). Afterwards, cells were co-stained with the appropriate primary antibodies (Table M5) in blocking buffer for 2 h at room temperature or at 4ºC overnight, washed three times with PBS and then co-immunostained with the appropriate secondary antibodies (Table M6) in blocking buffer for 1 hour at room temperature in the dark. Coverslips were washed again with PBS, dehydrated into increasing concentrations of ethanol, dried and mounted onto glass slides using Vectashield mounting medium containing 4',6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, H-1200). Samples were visualized using Leica AF6000 fluorescence microscope and a 63x objective. In each experiment, at least 200 cells were analysed. 3.6 Drug screening to affect DNA end resection In order to search for factor(s) affecting end resection, a screening of three different compounds libraries was performed. The three libraries used were:  Natural Product Screening Library (Selleckchem, L1400): 133 inhibitors  Anti-cancer Compound Library (Selleckchem, L3000): 386 bioactive compounds  Antidiabetic Compound Library (Selleckchem, L2900): 33 inhibitors Prior to start the immunofluorescence experiments, each library was manually diluted either in DMSO or Milli-Q water to a final stock of 20 µM in 96-well plates that were kept at -20ºC until use. To perform the screening, 6000 U2OS cells were plated in 96- Materials and methods 144  well plates (6005550, PerkinElmer). One day after seeding, control wells were transfected with CtIP or control siRNA in medium without glutamine. This medium was changed after 6 h. 72 h after seeding, cells were treated either with the indicated drugs (1 µM final concentration) from three previously mentioned Selleckchem libraries or with the corresponding controls DMSO, H2O or MLN2449. One hour after drug treatment, cells were irradiated with 10 Gy and kept at 37ºC for 1h. The protocol followed to perform the immunofluorescence experiment is described in the previous section (see section 3.5). Briefly, plates were washed once with PBS followed by treatment with pre-extraction 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) for 5 min on ice. Then, cells were fixed with 4% paraformaldehyde (Santa Cruz) for 15 min on ice. After two washes with PBS, cells were blocked with blocking solution (5% FBS in PBS) for 1h at RT. Primary antibodies against RPA and γH2AX (Table M5) were incubated for 2h at RT, washed with PBS and then incubated with secondary antibodies (Table M6) for 1h at RT in the dark. To visualize cell nuclei, cells were washed with PBS, stained with Hoechst 33258 (2 µg/mL, 861405, Sigma) for 10 min and washed with PBS again. The plates were sealed with parafilm and kept in the dark at 4ºC until image acquisition. To visualize the cells, plates were imaged using an ImageXpress Micro (Molecular Devices) at 40X magnification, and blue, green and Texas Red filters were used to detect nuclei, γH2AX and RPA signals respectively. Images were analysed, and the number of RPA foci per cell was quantified automatically with the Image MetaExpress Software (Molecular Devices). For this, at least 200 cells were scored per well. To facilitate the comparison between experiments, the median of the number of RPA foci per cell was normalized with each DMSO or H2O control. Conditions that skewed the balance towards increased HR repair resulted in a fold increase above 1. In contrast, a net decrease of this ratio (for example, values below 1) represented an imbalance towards NHEJ. Materials and methods  145 3.7 PIF1α foci at I-SceI DSBs To study the recruitment of PIF1 to DSB, U2OS19ptight13 cells (Lemaître et al., 2012) baring an I-SceI cleavage site is flanked by tandem repeats of the lac operator DNA sequences (lacO) were used. These cells were co-transfected with a Cherry-lacI and GFPPIF1α plasmids in a 10mm plates containing 2 coverslips and treated or not with doxycycline hyclate (Sigma, D9891) at a final concentration of 1 μg/ml for 14 h or 20 h, as indicated. Then, cells were washed and fixed with 4% paraformaldehyde (w/v) in PBS for 15 min, washed again twice with PBS and permeabilized with 0.5% Triton X-100 (Sigma Aldrich 9036-19-5) in PBS for 15 min at room temperature. Cells were then washed with PBS and blocked for 1h at room temperature using 3% BSA (Sigma Aldrich, A4503) diluted in PBS with 0.1% Tween 20 (Sigma Aldrich). Afterwards, coverslips were incubated for 1 hour at room temperature with primary antibodies against GFP and γH2AX (Table M5) prepared in blocking solution, washed three times with 0.1% Tween-20 in PBS, incubated for 1 h at room temperature with the appropriate secondary antibodies (Table M6) prepared in blocking solution and washed three times with PBS. Coverslips were mounted using Vectashield mounting medium (Vector Laboratories) containing DAPI. To visualize and acquire the images, a LEICA confocal microscope TCS SP5 was used with a HCX PL APO lambda blue 63X/ 1.4 OIL objective. 4. STATISTICAL ANALYSIS Unless otherwise specified in the figure legend, statistical significance was determined with a t-Student test using the PRISM software (Graphpad Software Inc.). Statistically significant differences were labelled with one, two or three asterisks if P < 0.05, P < 0.01 or P < 0.001, respectively. Materials and methods 146  5. TABLES Table M1. Cell lines used in this Thesis CELL LINE SELECTION SOURCE/REFERENCE U2OS No ATCC HTB-96 U2OS-GFP G418 This Thesis U2OS-GFPPIF1α G418 Kind gift from Stephen P. Jackson (Rodriguez et al., 2012) U2OS19ptight13 G418, puromycin Kind gift from Evi Soutoglou (Lemaître et al., 2014) HEK293-CATR G418 Kind gift from Balca R. Mardin (Roidos et al., 2020) Table M2. Plasmids used in this Thesis PLASMID DESCRIPTION SELECTION MARKER SOURCE/REFERENCE pEGFP-C1 Expression of EGFP Kanamycin/G418 Clontech (6084-1) GFP-PIF1α Expression of GFP fused to hPIF1α Kanamycin/G418 Kind gift from Stephen P. Jackson (Rodriguez et al., 2012) GFP-PIF1α E307Q Expression of GFP fused to hPIF1α containing E307Q mutation Kanamycin/G418 This Thesis pUC19 2xLacO Vector containing two lacO sequences to assay DNA resection Ampicilin Kind gift from Petr Cejka pUC19 2xLacO 2xG4 NT Vector containing two lacO sequences and two G4 sequences clones in tandem in the non-template strand to assay DNA resection Ampicilin This Thesis Materials and methods  147 Continuation of Table M2. Plasmids used in this Thesis PLASMID DESCRIPTION SELECTION MARKER SOURCE/REFERENC E SA-GFP G4 NruI Vector harbouring the SA-GFP reporter with a G4 sequence cloned next to the GFP sequence to analyse the influence of G4 in SSA efficiency Ampicilin/Purom ycin Generated in Pablo Huertas’ lab Cherry-LacI Expression of Cherry-lacI Kanamycin/G418 Kind gift from Evi Soutoglou CRISPR/Cas9- RAG1 Expression of Cas9 and gRNA targeting RAG1 gene Ampicilin Kind gift from Ron Jachimowicz and Toni Cathomen GFP Donor- RAG1 Donor of the GFP sequence to be integrated in RAG1 gene Ampicilin Kind gift from Ron Jachimowicz and Toni Cathomen Table M3. siRNAs used in this Thesis siRNA SENSE SEQUENCE (5’-3’) REFERENCE SUPLIER Nontarget (NT) pool UGGUUUACAUGUCGACUAA D-001810-10 Dharmacon CtIP GCUAAAACAGGAACGAAUC López- Saavedra et al., 2016) Sigma PIF1 3’UTR GGAUGUUCUCAGGUUGUAUUUAUTT This Thesis Sigma MRE11 GAAAGGCUCUAUCGAAUGU D-009271-02- 0005 Dharmacon EXO1 GAAGUUUCGUUACAUGUGU D-013120-01- 0005 Dharmacon DNA2 GCUAAACCGUGAAGCAAGA D-026431-01- 0005 Dharmacon FAN1 GUAAGGCUCUUUCAACGUA This Thesis Sigma BRCA1 CAGCUACCUUCCAUCAUA This Thesis Sigma Materials and methods 148  Table M4. Primers used in this Thesis PRIMER NAME SEQUENCE (5’-3’) USE PIF1α F/ CCCTGGATTGTGTGGAGATT R/ ACTCCAGACTGAGGCTCCTG RT-qPCR of PIF1α PIF1α F/CCTATGTGGCCCTTTCTCG R/GGTTTGGGTCCATGTTCTCC RT-qPCR of PIF1α MutE307Q Fw CGGTTGGTCATTGACCAGATC TCAATGGTGGAG Mutation of E307Q of PIF1α by QuickChange Multisite ki t MutE307Q Rv CTCCACCATTGAGATCTGGTC AATGACCAACCG Mutation of E307Q of PIF1α by QuickChange Multisite ki t GFPPIF1α Purexpress Fw GCGAATTAATACGACTCACT ATAGGGCTTAAGTATAAGGA GGAAAAAATATGGTGAGCAA GGGCGAGGAGC TGTTC In vitro production of GFP-PIF1α protein GFPPIF1α Purexpress Rv AAACCCCTCCGTTTAGAGAG GGGTTATGCTAGTTTCAGAGG ATTGGGTCCATGTTCTC In vitro production of GFP-PIF1α protein FAN Fw GATGAGCAGGAGAAGGGAAT TG RT-qPCR of FAN1 FAN Rv ACGGAATCTGAATTGTGGAC AC RT-qPCR of FAN1 EXO1 Fw CTGAAGTGTTTGTGCCTGAC RT-qPCR of EXO1 EXO1 Rv CCACAACTGCACCAC RT-qPCR of EXO1 DNA2 Fw GGAGAAGAGTGGCAGTT RT-qPCR of DNA2 DNA2 Rv TCTGTCACCTGCCATTAG RT-qPCR of DNA2 MRE11 Fw RT-qPCR of MRE11 MRE11 Rv RT-qPCR of MRE11 F1/F2 PstI G4 Fw TAAGCACTGCAGTAATGGGA ACCCGGGAGGGG Cloning of the c-Myc G4 sequence in pUC19 2xLacO F1/F2 PstI G4 Rv TGCTTACTGCAGTTATTCGCG ACCCCACCT Cloning of the c-Myc G4 sequence in pUC19_2xLacO F1 HincII G4 Fw TAAGCAGTTGACGTAATGGG AACCCGGGAGGGG Cloning of the c-Myc G4 sequence in pUC19 2xLacO F1 HincII G4 Rv TGCTTAGTCAACTTATTCGCG ACCCCACCTT Cloning of the c-Myc G4 sequence in pUC19_2xLacO Materials and methods  149 Continuation of Table M4. Primers used in this Thesis PRIMER NAME SEQUENCE (5’-3’) USE F2/F3 EcoRI G4 Rv TGCTTAGAATTCTTATTCGCG ACCCCACCT Cloning of the c-Myc G4 sequence in pUC19 2xLacO gRNA GFP CTCGTGACCACCCTGACCTAC GG crRNA sequence targeting GFP gene ssODN BFP CCTGAAGTTCATCTGCACCAC CGGCAAGCTGCCCGTGCCCT GGCCCACCCTCGTGACCACC CTGAGCCACGGGGTGCAGTG CTTCAGCCGCTACCCCGACCA CATGAAGCAGCACGACTTCT TCAAGTCCGCCATGCC Sequence complementary to GFP to convert GFP sequence to BFP sequence Table M5. Primary antibodies used in this Thesis ANTIBODY SPECIES REFERENCE APPLICATION (DILUTION) RPA32 Mouse Abcam (ab2175) IF (1:500) ϒH2AX Rabbi t Cell Signaling (2577L) IF (1:1000) GFP Rabbit Santa Cruz (sc-8334) IF (1:500), WB (1:1000) BrdU Mouse Sigma Aldrich (RPN202) IF (1:100) PIF1 Mouse Santa Cruz (sc-48377 IF (1:500), WB (1:500) Brca1 Mouse Santa Cruz (sc-6954) IF (1:500), WB (1:500) G4 Goat Absolute Antibody (Ab00174-24.1) IF (1:100) Anti-His Mouse GE Healthcare (27-4710-01) WB (1:500) TOP2β (H-286) Mouse Santa Cruz (sc-13059) WB (1:500) αTubulin Mouse Sigma (T9026) WB (1:50000)