Papel de factores de la transcripción y la replicación del ADN en el origen de la inestabilidad genómica.

Papel de factores de la transcripción y la replicación del ADN en el origen de la inestabilidad genómica Juan F. Lafuente Barquero Tesis Doctoral Universidad de Sevilla 2017 Papel de factores de la transcripción y la replicación del ADN en el origen de la inestabilidad genómica El doctorando, Sevilla, 2017 Juan F. Lafuente Barquero Andrés Aguilera López El director de tesis, Trabajo realizado en el Departamento de Genética, Facultad de Biología (Universidad de Sevilla), y en el Departamento de Biología del Genoma, CABIMER (Universidad de Sevilla-CSIC-UPO), para optar al grado de Doctor en Biología por el licenciado Juan F. Lafuente Barquero. TESIS DOCTORAL Las cosas no se pueden quedar pendientes... así que aqui estamos. Index of content INTRODUCTION 1 1. DNA metabolism and Genomic instability 3 1.1. REPLICATION 4 1.2. TRANSCRIPTION 8 2. Transcription-Replication collisions and DNA damage 13 2.1. Types of transcription-replication collisions 14 2.2. Sources of transcription-replication collisions 14 2.3. Preventing transcription-replication collisions 16 2.4. Resolving transcription-replication collisions 20 3. R-loops 22 3.1. Factors that remove R-loops. 24 3.2. Factors that prevent R-loop formation. 25 3.3. Good R-loops (Physiological functions) 27 3.4. Bad R-loops (Genome instability triggers) 29 3.5. R-loops: The Ugly (Impact on human health) 31 OBJECTIVES 33 RESULTS I 37 1. R-loops and NGS 39 1.1. RNA:DNA hybrid mapping in wild-type and hpr1∆ mutant yeasts 40 2. trans RNA as source of genomic instability 51 2.1. Impact of RNA in trans over the recombination frequency 53 2.2. The homologous recombination machinery is not needed for R-loop formation 59 2.3. Analysis of R-Loop in trans mediated recombination in a chromosomal system 62 RESULTS II 65 3. Search for novel factors involved in DNA:RNA hybrid metabolism 67 3.1. Selection of candidates 67 3.2. Screening 69 3.3. Effect of AID on the genetic instability of candidate mutants 72 3.4. R-Loop accumulation as detected by immunofluorescence 73 3.5. Effect on recombination and transcription dependency of the mutations selected 74 3.6. Characterization of the impact of the mph1∆ and dbp1∆ mutation on replication 76 3.7. Genetic interaction of mph1∆ with other DNA:RNA accumulating mutations. 77 3.8. Mph1 foci are induced in hpr1∆ cells 78 3.9. DNA:RNA hybrid detection by immunoprecipitation in mph1∆ mutants 80 4. Function of the human helicase RECQL5 in the model organism Saccharomyces cerevisiae 82 4.1. Expression of RECQL5 in yeast 82 4.2. RECQL5 recruitment to yeast chromatin 85 4.3. Impact of RECQL5 expression on the yeast genome stability 89 4.4. Effect of RECQL5 expression on the transcriptional process 94 4.5. Impact of RECQL5 helicase on cell cycle progression 103 4.6. Genome-wide distribution of RECQL5 through the S. cerevisiae genome 104 4.7. Yeast mutants resistant to RECQL5 expression 107 4.8. Genetic interaction of RECQL5 with yeast helicase mutants 111 DISCUSION 115 1. Genome wide study of RNA:DNA hybrids 117 1.1. Accumulation RNA:DNA hybrids towards the 3’ end of the gene 118 1.2. RNA:DNA hybrids in hpr1∆ 119 2. Prevalence of R-loops in trans 120 3. New helicases putatively involved in R-loop metabolism. 124 4. Understanding the role of RECQL5 at the interface between transcription and replication 128 CONCLUSIONS 135 MATERIALS & METHODS 141 1. Growth media and conditions 143 1.1. Yeast culture media 143 1.2. Bacteria culture media 143 1.3. Growth conditions 144 2. Antibiotics, drugs, inhibitors, enzymes and antibodies 144 2.1. Antibiotics 144 2.2. Drugs 144 2.3. Inhibitors 145 2.4. Enzymes 145 3. Antibodies 147 3.1. Antibody preabsorption 148 4. Strains 148 5. Plasmids 151 6. Yeast methodology 154 6.1. Yeast transformation 154 6.2. Cell cycle synchronization and FACS analysis 154 6.3. Genotoxic damage sensitivity assay 154 6.4. pLAUR transcription assay 155 6.5. Recombination assays 156 7. Detection of Rad52-YFP foci 157 8. Chromosome spreads Immunofluorescence 157 9. DNA analysis 158 9.1. Southern BLOT 158 9.2. Polymerase chain reaction (PCR) 158 10. Chromatin immunoprecipitation (ChIP) 159 10.1. Chromatin extraction 159 10.2. Immunoprecipitation 160 11. ChIP hybridized to Tilling Arrays (ChIP-Chip) 160 12. DNA:RNA Hybrid Immunoprecipitation (DRIP) 161 13. ChIP and DRIP data quantification and normalization 162 14. RNA analysis 162 14.1. Northern blot 162 14.2. Microarray analysis of gene expression 163 15. Statistical analyses 163 16. Primers and Probes 164 17. Protein analysis 166 17.1. Protein extraction 166 17.2. Sodium dodecyl sulfate Polyacrylamide gel electrophoresis (SDS-PAGE) 167 17.3. Western Blot Analysis 168 18. Miscellanea 169 APENDIX 171 BIBLIOGRAPHY 177 Index of Figures INTRODUCTION 1 Figure I1. Cell Cycle and replication 5 Figure I2. Transcription. 8 Figure I3. Transcription process 11 Figure I4. Types of transcription-replication collisions 14 Figure I5. Co-transcriptional mechanisms to suppress transcription-replication collisions. 16 Figure I6. Response pathways for transcription-replication collisions. 21 Figure I7. R-loop accumulation countermeasures. 27 RESULTS I 37 Figure R1. S9.6 signal profile over mitochondrial chromosome. 41 Figure R2. RNase H treatment for control. 42 Figure R3. Analysis of RNA:DNA hybrid positive ORFs in WT 45 Figure R4. RNA:DNA hybrid distribution over genes in WT 45 Figure R5. Distribution of hybrid positive regions at YRA1 locus. 47 Figure R6. Distribution of RNA:DNA hybrids across S. Cerevisiae genome. 48 Figure R7. Analysis of RNA:DNA hybrid positive features in WT and hpr1∆. 49 Figure R8. Signal distribution over hybrid-positive ORFs in wild-type and hpr1∆. 50 Figure R9. Schematic representation of the assay. 52 Figure R10. Analysis of the effect of RNA in trans on genetic recombination (I) 53 Figure R11. Analysis of the effect of RNA in trans on genetic recombination (II) 54 Figure R12. Analysis of the effect of RNA in trans on genetic recombination (III) 56 Figure R13. Analysis of the effect of RNA in trans on genetic recombination (IV) 57 Figure R14. Analysis of the effect of RNA in trans on genetic recombination (V) 58 Figure R15. Analysis of the effect of RNA in trans on genetic recombination (VI) 60 Figure R16. Analysis of Rad52 foci in hpr1∆ rad51∆ double mutants. 61 Figure R17. Analysis of the effect of RNA in trans on genetic recombination (VII) 63 RESULTS II 65 Figure R18. Genetic instability in different DNA and RNA helicase mutants. 70 Figure R19. Supression of Rad52-YFP foci by RNH1 overexpression in different helicase mutants 71 Figure R20. Effect of AID and Rnase H1 expression on genetic instability in selected mutants 72 Figure R21. Immunofluorescence of DNA:RNA hybrid accumulation in candidate strains. 73 Figure R22. Recombination analysis of selected helicase mutants. 74 Figure R23. Analysis of the interference with replication process of the selected helicase mutants. 75 Figure R24. Analysis of the genetic interaction between DNA:RNA hybrid accumulating mutants and mph1∆. 77 Figure R25. Accumulation of Mph1 foci in hpr1∆ cells. 79 Figure R26. DNA:RNA hybrid accumulation in mph1∆ deletion and helicase-dead mutants. 81 Figure R27. Expression of the human helicase RECQL5 in S. cerevisiae. 84 Figure R28. Interaction of the human helicase RECQL5 with yeast chromatin. 86 Figure R29. Effects of RECQL5 overexpression on the sensitivity to different chemical compounds. 88 Figure R30. Genetic instability caused by RECQL5 expression in S. cerevisiae. 90 Introduction Introduction 3 Since in 1953 two scientists working in Cambridge stormed into The Eagle pub announcing that they had discovered “the secret of life” a lot has happened (Watson & Crick, 1953). The so called molecule of life, the DNA, is widely known to hold the key for the survival of all living organisms, both thanks to its capacity to be copied and to its role as blueprint for the proteins. One could think that constituting such an important molecule, DNA would be strongly protected and maintained stable inside the cell. Much on the contrary, genomes, that comprehend all the DNA from an organism, are constantly subjected to a variety of dynamic processes, plus exogenous and endogenous agents can even damage them, generating approximately 105 lesions per day in mammalian genomes (Collins, 1999). All these factors lead to the appearance of genome instability, a process necessary for the generation of genetic variability and evolution, but also a hallmark of aging, hereditary genetic diseases and cancer-related disorders. This thesis adds to the knowledge on how the stability of genomes is maintained, focusing our study on the interface between the two most important biological processes that use the DNA as template, transcription and replication. In this introduction we will concisely cover the most important aspects about these processes, overviewing the mechanisms that control the interplay between them and the recent advances in the understanding of the impact derived from their miss-regulation. 1. DNA metabolism and Genomic instability The concept of genomic instability encompasses the range of genetic alterations derived from unresolved damages or modifications that the DNA can accumulate due to the action of exogenous or endogenous agents. Additionally, biological processes that take place on the DNA can also lead to the occurrence of negative effects for its integrity. Although such events may be harmful for the cell, they also contribute to the generation of genetic variability necessary for evolution. Some of these processes play specific roles since, for example, variability is essential for regulated developmental mechanisms like the immunoglobulin class- Ph.D. Thesis - Juan F. Lafuente Barquero 4 switching (Maizels, 2005). Nevertheless, genetic instability is more commonly associated with pathological disorders, and in humans can develop into premature ageing, inherited diseases and a number of cancer predispositions. Usually, genetic instability translates into two major problems, either the failure to properly replicate the genome, or the loss of information indispensable for cell surveillance. Both situations are directly consequence of the two main functions of the DNA molecule, that are: first, pass on the genetic information from a “mother” cell to a “daughter” cell through replication of its genome; and second, serve as template for the synthesis of the functional components of the cell through its transcription. Therefore, DNA replication and transcription are essential cellular processes that are mediated by complex machineries that copy genetic information into complementary DNA and RNA molecules, respectively. There are cases in which crosstalk between these processes and others that take place on the DNA, like recombination, have positive effects, as for example, in the case of transcriptioncoupled repair (Hanawalt & Spivak, 2008). However, the use of the same template by two different machineries is not an easy task and the encounter of two processes on the same genomic region may have negative consequences. This has been long proven for the case of replication and transcription, whose conflicts constitute a remarkable natural source of genomic instability (Zeman & Cimprich, 2013; Gaillard et al., 2015). 1.1. REPLICATION DNA replication is the process necessary for genome duplication, and so, for the maintenance of life. It is important to note that it occurs exactly once during the cell cycle, and in eukaryotic cells it takes place on a protein-DNA template called chromatin, not just on DNA. This makes replication fork (RF) movement considerably slower, being only about 1,000 to 3,000 bp per minute, compared to the rapid rate of 50,000 bp per min in prokaryotes. Accordingly, larger genomes take longer to replicate and so eukaryotic cells, like yeasts, have evolved to use many origins of replication per chromosome (Bell & Labib, 2016). Introduction 5 In S. cerevisiae there are about 250 to 400 replication origins distributed among its 16 chromosomes, almost all of these origins is also an autonomously replicating sequence or ARS. Their DNA sequence is not completely shared, but many have properties in common, like an A+T-rich core consensus sequence WAAAYATAAAW (W=A or T, Y=C or T). These stretches of nucleotides are binding sites for proteins involved in replication, mainly the origin recognition complex, or ORC. This complex, constituted by 6 subunits (ORC1 to ORC6, numbered from Adapted from Garcia-Muse & Aguilera, 2016 Replication Fork Chromatin assembly Leading strand Lagging strand Pol α–primase Supercoiling (+) RPA PCNA Replication cofactors MCM Pol δ Pol ε DNA synthesis G1 M G2 S Figure I1. Cell Cycle and replication (A) Temporal representation of S. cerevisiae ’s cell cycle. Replication only takes place once during a complete turn and it is on S (synthesis) phase. (B) Replication forks are complexes constituted by the minichromosome maintenance complex (MCM9 which opens the DNA helix, and DNA polymerases Epsilon and Gamma, which extend the leading and lagging strand respectively. Pol-alpha is required for the initiation of the synthesis. RPA coats the ssDNA that is left behind in the lagging strand until its complementary strand is synthesized. Many factors are required for the correct progression of the RF, including the proliferating cell nuclear antigen (PCNA). DNA unwinding generates positive supercoiling that requires of specialized topoisomerases to be alleviated. Also, nucleosomes need to be relocated in the DNA once the RF has passed. Arrows indicate the direction of synthesis by DNA polymerases. Ph.D. Thesis - Juan F. Lafuente Barquero 6 largest to smallest molecular weight), binds to origins of replication in an ATP- dependent manner. ORC complex is conserved to humans, and like its bacterial orthologue, DnaA, it is responsible of 3 functions at the origin: it binds to the specific DNA sequence at the origins, it participates in the DNA unwinding of the origin and it recruits other replication factors. After this, eukaryotic DNA replication process can be divided into three stages: Licensing: In yeasts, the ORC keeps stably bound to the origin even once it has been fired. The replication initiation stage involves the assembly of the pre-replicative complex, constituted by two hexamers of Mcm proteins (minichromosome maintenance complex proteins 2 - 7), which are loaded at the replication origins exclusively during G1 phase and require the presence of the 6 subunits of the ORC, the CDC6 (cell division cycle 6) ATPase and the CDT1 (chromatin licensing and replication factor 1). After its complete assembly during the end of G1, CDC6 and CDT1 are released and the ORC and MCM are retained on the DNA ready for next step. Origin Firing: The activation of licensed origins is triggered by the activities of cyclin-dependent kinases (CDK) and CDK-like kinases. This stage depends on the stable association of CDC45 and the DNA replication complex GINS (the name is an acronym for go-ichi-ni-san, the Japanese for 5-1-2-3, after the four subunits of the complex Sld5, Psf1, Psf2 and Psf3). Re-replication is avoided by the action of CDKs and the capture of CDT1, which prevents the reload of the MCM complex outside the G1 phase. Progression: Replication forks (RF) are driven by holo-enzymes that are proceeded by the replicative helicases present in the CDC45-MCM-GINS complex. Leading and lagging strands are extended by Pol e and Pol d, respectively, which interact with additional factors that promote RF progression, like the Mrc1/ CLASPIN factor or the replication pausing complex formed by Tof1-Csm3-Ctf4/ TIPIN-TIM-And1 which act in the regulation and crosstalk between the replicating DNA helicases and polymerases, especially during RF stalling and restart (Errico & Costanzo, 2012). Other replication factors include the Pol α-primase complex (Pol α-pri), which contains both DNA primase and DNA polymerase subunits and Introduction 7 is responsible of the synthesis of each new DNA molecule initiated thanks to a RNA primer, once for the leading strand and in each Okazaki fragment for the lagging; the Rrm3 helicase and the PCNA sliding clamp along with the PCNA clamp loader, RFC. Once DNA synthesis initiates, the new RFs start moving bidirectionally (in most cases) away from the origin. They will terminate when they meet opposing RFs from adjacent replicons. • Replication as source of breaks Although concisely explained here, such a complex process is commonly subjected to the occurrence of problems or failures, which lead to generation of replication stress. This is a source of DNA breaks and genome instability, and can be generated either by dysfunctions of the replication machinery or by blockage of the RF when it encounters an obstacle. Common barriers for replicating machineries are DNA bulky adducts, non B-DNA structures or tightly-bound DNA protein complexes. These blocks might force replication to be delayed or, in the worst cases, paused for a prolonged period of time in a phenomenon known as RF stalling. Eukaryotic cells have evolved mechanisms to restart and resume replication once the obstacles are removed, but blocked replisomes are susceptible to arrest or collapse, leading to the formation of DNA double strand breaks (DSBs) or fork regression, that would expose ssDNA stretches. Other significant source of DNA breaks is when RF encounters DNA lesions, which hamper its progression, or ssDNA gaps, which could be converted into DSBs during replication. Altogether, these kind of aggressions are liable of replication-associated genomic instability. Moreover, hyper-recombinant phenotypes of S. cerevisiae replication mutants, supports the notion that homologous recombination (HR) is the main repair pathway for these lesions (Hartwell & Smith, 1985; Aguilera & Klein, 1988; Yan et al., 1991). One of the common sources of replication stress are the collisions between the RF and the transcription machinery, which constitute an important source of genome instability. Since this mechanism of generation of endogenous damage will be the focus of this thesis we briefly present some details about the transcriptional process. Ph.D. Thesis - Juan F. Lafuente Barquero 8 1.2. TRANSCRIPTION Gene expression is a process that requires several steps, beginning with the transcription, which is the copying of the information from a DNA strand into a complementary sequence of RNA, and ending with the translation of that RNA into a protein in the cytoplasm. Eukaryotic transcription is evolutionarily conserved from yeast to human. It takes place inside the nucleus and the machinery required contains a core RNA polymerase that can be classified in three types depending on the nature of their products: Table I1. Eukaryotic RNA polymerases Polymerase Product Location RNA polymerase I (RNAPI) Larger ribosomal subunits (rRNA) Nucleus RNA polymerase II (RNAPII) Messenger RNA (mRNA), most small nuclear RNAs (snRNA), small interfering RNAs (siRNA) and micro RNAs (miRNA). Nucleus RNA polymerase III (RNAPIII) Transfer RNAs (tRNA), other small RNAs (snRNA & 5S rRNA), snRNA U6, signal recognition particle RNA (SRP RNA) and other stable short RNAs. Nucleus and nucleolus Due to RNAPII being responsible for transcription of all protein-coding genes, it has been studied in-depth and its function is well understood. In general, transcription can be divided into three stages, briefly detailed below. This process Adapted from Garcia-Muse & Aguilera (2016) Chromatin remodelling Supercoiling (+) RNA-processing factors RNAP Supercoiling (–) Promoter Topoisomerases Transcription elongation factors RNA Figure I2. Transcription. The transcription bubble comprehends a small portion of the DNA double helix that is unwind by the RNA polymerase (RNAP) to allow the transcription of that locus. Topoisomerases are required to alleviate the negative and positive supercoiling that the RNAP generates. Also, chromatin remodeling and RNA processing factors are recruited for a successful transcription. Introduction 9 has some specific differences depending on the polymerase type, thus we have focused on the transcription by RNAPII: Initiation Transcription initiation can be also separated in two steps. In the first step, the pre-initiation complex (PIC) is assembled. Promoter elements upstream of genes are recognized by general transcription factors, like TFIIA, TFIIB, TFIID, TFIIE and TFIIH (Orphanides et al., 1996; Saunders et al., 2006). This phase is tightly controlled by a wide number of signaling pathways that affect sequence-specific DNA-binding transcription factors, co-activator complexes and enzymes that regulate the accessibility of DNA by modulating chromatin structure (REF). Recruitment of the RNAPII marks the end of the pre-initiation complex assembly and starts the synthesis of the first phosphodiester bonds. The second step is constituted by promoter escape and initiation. It starts when an initial transcript reaches the 10 nucleotide threshold, entering the RNA exit channel. At the end of this stage, the C-terminal tail domain (CTD) of the polymerase is phosphorylated and the interactions with the promoter elements are broken. The CTD tail of the RNA polymerase II is comprised of repeats of the heptapeptide sequence “Tyr1-Ser2-Pro3-Thr4-Ser5- Pro6-Ser7”. Promoter-associated RNAPII is hypo-phosphorylated (referred as RNAPI- Ia) and elongating enzyme have a hyper-phosphorylated CTD (referred as RNAPIIo). All of the serine residues are susceptible of phosphorylation, and the differential patterns of CTD modification have been proposed to serve as docking platform for a variety of elongation factors and regulation proteins. These factors are required to ensure proper post-transcriptional processing of the nascent RNA as well as its export, and to help RNAPII deal with challenges due to the chromatin structure, difficult DNA sequences or lesions that the polymerase may encounter during elongation, usually causing transient pausing or enzyme arrest. Elongation After phosphorylation of CTD Ser5 and escaping the promoter, most of the initiation factors are substituted for new proteins required for the elongation and processivity of the polymerase (Saunders et al., 2006). Double stranded DNA (dsD- NA) enters from the front of the enzyme and is unzipped to provide the template Ph.D. Thesis - Juan F. Lafuente Barquero 16 Another kind of transcription-associated structures that can halt replisome and generate transcription-replication conflicts are co-transcriptional generated RNA:DNA hybrids (known as R-loops when formed outside the transcription bubble), which will be reviewed later in this introduction. 2.3. Preventing transcription-replication collisions On account of all these factors, collisions are not rare events and thus, cells have evolved mechanisms to prevent their occurrence. Bacteria have developed a bias towards genes located co-directional with replication origins (Merrikh et al., 2012). In eukaryotes this bias is not obvious, but they have alternative strategies to deal with head-on collisions. As for example, S. cerevisiae cells have established replication fork barriers (RFBs) surrounding the highly transcribed ribosomal DNA (rDNA). Although replication is initiated bidirectionally at the ARS located near the rDNA, these RFBs work polarizing the RFs to prevent head-on collisions by blocking Figure I5. Co-transcriptional mechanisms to suppress transcription-replication collisions. There are many mechanisms by which cells can resolve transcription roadblocks. Main proteins or factors involved in each pathway is shown either prokaryotic or eukaryotic. RECQL5 is a human helicase that controls transcription elongation rate to reduce conflicts with replication. The antibacktracking factors GreA/GreB and TFIIS help to resume transcription. There are pathways that lead to RNAP proteolytic degradation when it gets stalled at DNA damage sites by TC-NER or polyubiquitylation signaling. Transcription termination and resolution of R-loops is mediated by the Rat1/ XRN2 exonucleases and RNA:DNA helicases such as Sen1/SETX or AQR. Additionally, R-loops can by processed by the TC-NER endonucleases XPF/XPG. All of these pathways help to remove transcriptional blocks and minimize collisions with replication forks. Yellow star represents DNA lesion. Introduction 17 forks from progressing in the direction opposite to RNAPI transcription (Linskens & Huberman, 1988; Ivessa et al., 2000). Similar RF pauses have been described also at tRNA genes in yeast and at the ribosomal gene arrays of other organisms (Labib & Hodgson, 2007). Other strategies include the redistribution by the RNAP of replication initiation factors in G1 on chromatin prior to their activation (Gros et al., 2015). Evidence suggests that higher eukaryotes coordinate transcription and replication in the nucleus spatially and temporally. Reports indicate that, during S-phase, eukaryote cells compartmentalize replication so that DNA replication and transcription machineries occupy distinct nuclear territories and act at distinct times (Wei et al., 1998), and more recently, a nascent RNA capture assay showed a global anti-correlation between transcription and replication timing (Meryet- Figuiere et al., 2014) Cells also count with active mechanisms to minimize transcription-replication encounters. The transcription machinery itself is an important factor on this regard as it has been inferred from the studies of defective RNA polymerase strains. Some of these mutants are able to compromise the stability of transcription complexes, being able to suppress growth defects generated by the absence of DNA helicases that help resolve collisions, such as UvrD, Rep and DING in bacteria (Baharoglu et al., 2010), suggesting that more unstable transcription machineries suppose weaker obstacles for replication fork progression. Furthermore, other RNAPII mutants have been described in yeast to present elongation defects and exhibit important replication impairment, due to prolonged or tighter unions with the template DNA as determined by chromatin immunoprecipitation (ChIP) (Felipe-Abrio et al., 2014). These data suggest that the capability of an RNAPII to be evicted from the DNA can vary, and this characteristic influences on replisome advance. The same way the RNA polymerase holoenzyme represents an obstacle for the replication fork progression, other non-nucleosomal protein complexes can impede the replisome advance. In yeast, one such barrier is the protein Fob1, which is recruited to specific RFB (Replication Fork Blocking) sequences near the 35S rRNA gene. Replication progression through these RFB-Fob1 complexes requires Rrm3, and in its absence, cells present increased replication pauses at Ph.D. Thesis - Juan F. Lafuente Barquero 18 rDNA, DNA breaks and accumulation of excised rDNA circles (Torres et al., 2004). Mutants lacking Rrm3 also show increased fork pausing at tRNAs and telomeres, supporting its participation in other pause sites (Ivessa et al., 2002; Ivessa et al., 2003). However, although some highly transcribed genes also show RF pausing sites, Rrm3 does not accumulate in all of them, suggesting that other factors may have roles in the prevention or resolution of collisions. Consistent with this view, the E.coli transcription elongation factors GreA and GreB , which promote the release of backtracked and stalled RNAP seem to participate in the reduction of conflicts, as their absence in high transcription conditions leads to a complete block of replication (Tehranchi et al., 2010). In yeast, the elongation factor TFIIS holds a similar role, facilitating the catalytic activity of the RNAPII in the backtracking process (Dutta et al., 2015). Another arising important factor on this regard is the human helicase RECQL5 (RecQ-like dependent DNA helicase Q5). RECQL5 is a member of the RecQ family of ATP-dependent DNA helicases, which is one key player in genome maintenance and stability. Conserved from bacteria to humans, while lower organisms, like bacteria and yeast, express only one member of the family, most mammals possess five RecQ helicases: RECQL1, BLM, WRN, RECQL4 and RECQL5. Defects in BLM, WRN and RECQL4 are associated with monogenic disease syndromes characterized by genome instability, multiple cancer predispositions and/or premature ageing phenotype. Recent studies suggest links between mutations in the two other members, RECQL1 and RECQL5, with different types and predisposition to cancer development (Fu et al., 2017). RecQ helicases are involved in base excision repair (BER), DNA double strand break repair (DSBR), intra-strand cross link repair, recovery of stalled replication forks and telomere processivity and stability (Singh et al., 2011; Croteau et al., 2014). RECQL5 is of special interest owing to its particular versatility, forming protein complexes with a diverse set of cellular partners like the RNAPII, the PCNA complex or TOPII (Cohen et al., 2013), to coordinate various processes including replication, transcription and recombination. Up to date, RECQL5 is the protein with the best characterized active role in preventing transcription-replication conflicts. Indeed, it is the only RecQ helicase Introduction 19 found to interact with the RNAPII (Aygün et al., 2008). Using high throughput sequencing techniques it was described that overexpression of RECQL5 led to reduced transcription rates, and its depletion induced a general transcription upregulation increasing transcription pausing, arrest and backtracking (Saponaro et al., 2014). This supports the idea that RECQL5 has a negative regulatory impact on transcription elongation and that uncontrolled transcription is a source of DNA damage. RECQL5 also associates with the replicative sliding clamp PCNA (Kanagaraj et al., 2006), pointing to a direct participation in replication. Furthermore, DSBs accumulate in RECQL5-depleted cells during replication in a RNAPII-dependent way, since breaks locate in actively transcribing genes and transcription inhibitors suppressed their appearance (Hu et al., 2009). Recently, it has been shown that RECQL5 promotes TOP1 sumoylation, which is required for the efficient association of the topoisomerase with RNAPII and the recruitment of other mRNA-processing factors to transcriptionally active sites, as inferred by the accumulation of RNA:DNA hybrids in defective cells depleted of RECQL5 (Ramamoorthy et al., 2012). These data suggest for RECQL5 an active role in the maintenance of genome stability by preventing the occurrence of transcription-replication collisions, whether by limiting transcription itself or by reducing the accumulation of co-transcriptional RNA:DNA hybrids. Even though the role of chromatin has not been studied in relation with the regulation of transcription-replication collisions, both processes take place in the context of a highly structured template. Histone chaperones and chromatinremodeling complexes are required not only to allow the passage of polymerases, but also to revert the DNA to its original packaged state to avoid further damages. This has been clearly shown for the FACT complex (Facilitates Chromatin Transcription), which was found to have roles originally in transcription but also in replication. Cells lacking a functional FACT complex accumulate high levels of transcription-replication collisions, and exhibit RF progression impairment which correlates with higher levels of genome instability, both in yeast and human cells (Orphanides et al., 1996; Abe et al., 2011; Foltman et al., 2013; Herrera-Moyano et al., 2014). Ph.D. Thesis - Juan F. Lafuente Barquero 20 2.4. Resolving transcription-replication collisions The main consequence of transcription-replication collisions is their impact on genome stability. The resulting collapsed replication forks can induce the generation of DSBs and chromosomal breakage, therefore, collisions and their damage are expected to be sensed by the DNA damage response (DDR) and other DNA repair pathways. Although is not confirmed that DDR is able to sense transcription-replication collisions directly, the damage generated in such events is enough to activate the DNA damage checkpoint, as collisions may precede the uncoupling of the leading and lagging strands, generating long stretches of ssDNA (Lopes et al., 2006; Heller & Marians, 2006). These stretches of ssDNA are sensed by the Ataxia Telangiectasia and Rad3-related (ATR, Mec1 in budding yeast)-dependent replication checkpoint, which activates to protect the integrity of the replication fork. This is unless DSBs are generated, in which scenario the Ataxia Telangiectasia Mutated (ATM, Tel1 in budding yeast)-dependent checkpoint would be the pathway triggered (Maréchal & Zou, 2013). There is evidence of ATR/Mec1 role in regulating transcriptionreplication collisions by dismantling the pre-initiation complex at tRNAs (Nguyen et al., 2010) or by triggering the eviction of histones to facilitate the processing and release of RNAPII (Im et al., 2014; Poli et al., 2016). ATR/Mec1 activation starts a cascade that phosphorylates many downstream targets, including the tumor suppressor p53 in mammalian cells, which has also been recently involved in preventing transcription-replication collisions by reducing topological stress (Yeo et al., 2016). ATR-mediated replication stress response also triggers release from the nuclear pore of genes being transcribed. The phenomenon, known as ‘gene gating’, consists in the localization to the nuclear periphery of actively transcribed genes, presumably to facilitate the nuclear export of the transcripts, it is mediated by RNA binding proteins that coordinate the processing of the pre-mRNA and its transport through nuclear pores to the cytoplasm (Blobel, 1985; Cabal et al., 2006). Interestingly, mutations in some of these factors partially suppress the replication stress phenotypes in checkpoint mutants, suggesting that blocked replisomes at Introduction 21 genes located near the nuclear pore cannot be released in absence of a functional checkpoint to presumably allow supercoiling relief and replication restart (Bermejo et al., 2011; Bermejo et al., 2012), raising the possibility that this pathway regulates fork stability through control of transcription-coupled processes. Whether specific repair pathways are triggered at transcription-replication collisions sites after activation of the DDR by the DNA damage checkpoint is not clear yet. However, an increasing body of work suggests that transcription-replication blocks require the activity of factors involved in DNA recombination (Prado & Aguilera, 2005; Wellinger et al., 2006a; Gottipati et al., 2008). This is supported by recent findings that have linked the tumor suppressor BRCA1 and BRCA2, Figure I6. Response pathways for transcription-replication collisions. A stalled replication fork can activate the ataxia telangiectasia and Rad3 related (ATR; Mec1 in yeast)-dependent checkpoint to solve collisions and avoid their consequences. Specific factors are recruited to resolve or prevent the obstacle, including DNA helicases (such as Rrm3), mRNA-processing proteins or chromatin-remodeling complexes (such as FACT). In addition, the RNA polymerase (RNAP) could be released at transcription termination sites, with the help of BRCA1 and the helicase senataxin or the chromatin-remodeling complex INO80 and transcription factor complex PAF1C, as a way to avoid the collisions. Similarly, at tRNA transcription sites, RNA Pol III is directly evicted during S phase. Stabilization and resumption of stalled forks at transcribed DNA regions can occur via DNA repair factors, such as those of the Fanconi anaemia (FA) repair pathway, including the tumour suppressors BRCA1 and BRCA2. Finally, different DNA repair pathways can act at collision sites, if these degenerate into DNA lesions. Although direct involvement of the ATR-dependent checkpoint has been reported in some examples, the degree of its involvement in other cases is yet unknown. ASF1A, anti-silencing function protein 1 homologue A; HR, homologous recombination; MCM, minichromosome maintenance complex; Pol δ, DNA polymerase δ; Pol ε, DNA polymerase ε. MCM Transcription–replication collision Stalled replication fork Stalled replication fork Stalled replication fork Checkpoint activation? Replication fork stabilization RNAP eviction or transcription termination Obstacle removal, DNA repair and replication restart ATR (Mec1) • FA • BRCA1 • BRCA2 • FACT • Rrm3 • FA • HR • ASF1A • INO80 and PAF1C • BRCA1 and senataxin RNAP Pol δ Pol ε Ph.D. Thesis - Juan F. Lafuente Barquero 22 which are involved in DSB repair, with the prevention or resolution of RNA:DNA hybrids (Hill et al., 2014; Bhatia et al., 2014), providing a mechanistic connection between recombination and replication-stress induced by R-loops. For instance, BRCA1 was shown to interact with SETX to prevent damage at certain R-loop-prone termination regions (Hatchi et al., 2015), suggesting a role in R-loop-formation suppression or processing. In parallel, BRCA2 interaction with TREX-2 (Bhatia et al., 2014) reinforces the hypothesis that transcription factors work together with DNA repair proteins to reduce the number and/or the impact of transcription-replication collisions. BRCA2 is also member of the Fanconi Anemia (FA) DNA-repair pathway of interstrand crosslinks, which has recently been demonstrated to be activated by transcription-replication conflicts linked to the accumulation of RNA:DNA hybrids (Schwab et al., 2015; García-Rubio et al., 2015). Accordingly, the Fanconi Anaemia Complementation group M protein (FANCM, Mph1 in yeast) was shown to hold the capacity of RNA:DNA branch migration in vitro, which has been used to suggest a role for the translocase activity of FANCM to directly resolve the fork-blocking R-Loop (Schwab et al., 2015), but this has still to be demonstrated in vivo. Additional mechanisms that support the existence of cooperative pathways between transcription and replication have been found. For example, it has been described in yeast that the transcription factor Hog1 has the capacity to delay replication timing by influencing early-firing origins and replication fork progression itself through the interaction with Mrc1 (Claspin in humans), which is a replisome component (Duch et al., 2013) and a substrate for the replication checkpoint Mec1 (Alcasabas et al., 2001). These data suggest that there is a tight regulation of mechanisms to coordinate transcription and replication in special situations, helping the organisms overcome stressful situations that require transcriptional burst, without putting in danger their genome stability. 3. R-loops As mentioned before, RNA:DNA hybrids (more specifically R-Loops) are usually present at transcription-replication conflict sites, in some circumstances as Introduction 23 triggers for that collisions and in others as consequence of the blockage produced between the machineries. In the last two decades, a large quantity of data has been generated about the nature of this nucleic structure, on how they are formed and which tools have evolved the cells to regulate its presence (nicely reviewed by Santos-Pereira & Aguilera, 2015) RNA:DNA hybrids are naturally occurring structures formed during processes like replication, in which they prime DNA synthesis, or transcription, at the active site of the RNA polymerase during ribonucleotide triphosphate incorporation. These hybrids are short and are more stable than double-stranded DNA (dsDNA), adopting a conformation midway between B form of dsDNA and the A form of double-stranded RNA (dsRNA) (Roberts & Crothers, 1992). On the other hand, so called R-loops are longer RNA:DNA hybrids formed as result of the hybridization of a nascent RNA with its complementary DNA strand, leaving the other single-stranded DNA molecule displaced. Formation and/or stabilization of R-loops in vivo is favored by different DNA features, such as G-content or negative DNA supercoiling, for example (Santos-Pereira & Aguilera, 2015). Although considered rare events not so long ago, R-loops are found in vivo as natural intermediates in specific cellular processes, such as E. coli plasmid replication, mitochondrial replication or immunoglobulin (IG) class switching (Aguilera & García-Muse, 2012). However, recent studies have uncovered a wider presence than previously foreseen, and R-loops have been shown to be an important source of genome instability, a hallmark for cancer cells (Aguilera & García- Muse, 2012; Santos-Pereira & Aguilera, 2015) and to have an important impact on transcription (Skourti-Stathaki et al., 2014), chromatin structure (Castellano-Pozo et al., 2013) and replication (Chang & Stirling, 2017). Recent genome-wide approaches have permitted the detection of R-loops in many loci. An in silico tool was developed to identify R-loop-prone sites over the human genome and other model organisms (R-Loop DB - http://rloop.bii.a-star.edu.sg/), which identifies a high number of putative R-loop-forming sequences (250,000) in 59% of human genes (Jenjaroenpun et al., 2015). Also, a variety of techniques have been applied to genome wide detection of R-loops in human cells (DRIP followed Ph.D. Thesis - Juan F. Lafuente Barquero 24 by sequencing, DRIP-seq (Ginno et al., 2012; Ginno et al., 2013)) or in S. cerevisiae (Chromatin Immunoprecipitation followed by seq, ChIP-seq (El Hage et al., 2014) or DRIP followed by microarray, DRIP-chip (Chan et al., 2014) and DRIP treated with S1 nuclease followed by sequencing, DRIP-S1-seq, (Wahba et al., 2016)). All revealed numerous R-loop hotspots including the rDNA region, telomeres, Ty transposons, RNA Polymerase III-transcribed genes and highly expressed RNAPII-transcribed genes. With the discovery of this high incidence of R-loops, several cell mechanisms and pathways to regulate R-loop homeostasis are being uncovered that could be classified in two categories: those implicated in the removal of R-loops once they are formed, and those preventing their formation. 3.1. Factors that remove R-loops. R-loops can be removed by the RNase H enzymes (RNase H1 and RNase H2), which degrade the RNA moiety of an RNA-DNA hybrid molecule. In yeast, RNase H1 is encoded by the RNH1 gene, while the RNase H2 enzyme is a trimeric complex made up of RNH201 (the catalytic subunit), RNH202, and RNH203 (Cerritelli & Crouch, 2009). Another source of RNA:DNA hybrids results from the miss-incorporation of ribonucleoside triphosphates (rNTPs) into DNA by the replicating DNA polymerase (McElhinny et al., 2010). Ribonucleotide excision repair (RER) is performed primarily by the RNase H2 enzyme in combination with the endonuclease Fen1 (Sparks et al., 2012). Importantly, RNase H2 is competent to cleave both monoas well as consecutive –ribonucleotides, whereas RNase H1 requires stretches of at least 4 ribonucleotides (Cerritelli and Crouch, 2009). Alternatively, RNA:DNA hybrids can also be removed via displacement of the RNA strand, performed by RNA:DNA helicases which unwind the RNA:DNA hybrids or limit their formation. As for example, yeast Pif1 is involved in mitochondrial DNA maintenance, telomeric DNA synthesis, rDNA replication and Okazaki fragment maturation (Boulé & Zakian, 2009). The yeast Sen1 and its human homolog SETX have been also linked to the removal of R-loops in vivo since mutants in these genes accumulate high levels of RNA:DNA hybrids, especially at transcription Introduction 25 termination sites (Mischo et al., 2011; Skourti-Stathaki et al., 2011). More recently, the human aquiarius (AQR) protein, which is an RNA helicase of the same family as Sen1/SETX, has also been proposed to have a role in R-loop removal, but there is still more research needed to confirm it (Sollier et al., 2014). 3.2. Factors that prevent R-loop formation. Several studies in bacteria and eukaryotes have also shown an effect of supercoiling regulation by topoisomerases (TOP) in R-loop formation. Local unwinding produced by the negative supercoiling that follows an elongating RNA polymerase is a suitable substrate for the formation of co-transcriptional RNA:DNA hybrids. This has been proven in bacteria since topA rnhA mutants are not viable, and overexpression of RNase H1 in topA backgrounds suppresses its growth defect (Drolet et al., 1995), and supported in yeast, where deletion of both Top1 and Top2 increases R-loops accumulation at the rDNA locus, generating RNAPI stalling and defects in pre-rRNA synthesis (El Hage et al., 2010), or in humans, where TOP1- deficient cells show DNA breaks at active genes and replication defects that are suppressed by RNase H1overexpression (Tuduri et al., 2009). The other key element for the formation of the R-loops is the nascent mRNA, as deficiencies in the assembly of the messenger ribonucleoparticle (mRNP) can strongly stimulate R-loop formation. The THO complex was the first transcriptioncoupled RNA processing complex that was demonstrated to limit the formation of RNA-DNA hybrids in a co-transcriptional manner (Huertas and Aguilera, 2003) by binding to the nascent RNA and ensuring that it gets properly assembled into an mRNP that is efficiently exported. Many other transcription factors have been uncovered since then, like the splicing factor SRSF1 gene in chicken and human cells (Li et al., 2005), or the yeast ataxin 2 RNA-binding protein, PAB1-binding protein 1 (Pbp1), which prevents R-loop formation by interacting with non-coding RNAs (ncRNAs) generated at the intergenic spacers of the rDNA locus (Salvi et al., 2014). Additionally, control of the mRNAs half-life is also important to reduce the probability of R-loop formation, as it was shown in yeast cells lacking the Trf4 polyadenylation (poly(A)) polymerase, component in the TRAMP complex (Trf4- Ph.D. Thesis - Juan F. Lafuente Barquero 32 Besides, R-loops lead to genome instability and replication stress, which are hallmarks of pre-tumoral and tumoral cells (Gaillard et al., 2015), so R-loops could be potential drivers of cancer. Such connection is already supported by the results that link deficiencies of BRCA1 and BRCA2 to increased R-loop accumulation and DSBs that are partially reduced by RNase H1 (Bhatia et al., 2014). Additional connections have been stablished over the time, with many factors already known to lead to tumorigenesis now being associated with an increased presence of R-loops, which are deeply reviewed in (Santos-Pereira & Aguilera, 2015). Objectives Objectives 35 Objectives The main goal of this thesis is to advance in the knowledge about how R-loops are formed, what are their functions and how their presence is regulated to maintain the stability of the genome. For this purpose, we addressed the following specific objectives: 1. To analyze where R-loops are formed through the genome, in a wild-type strain and a hpr1∆ mutant, and study the differences. 2. To explore the possible formation of non-cotranscriptionally formed R-loops (R-loops in trans). 3. To search for new factors implicated in R-loop homeostasis maintenance. 4. To get new insights into the mechanisms of coordination between transcription and replication by the human helicase RECQL5, through its expression in the model organism S. cerevisiae. Results I Results I 39 1. R-loops and NGS The study of RNA:DNA hybrids distribution across the whole genome suppose a powerful approach to help shed some light on how these structures are generated in physiological or pathological environments and why their de-regulated presence is dangerous for the cell. Up to date, a number of similar studies have already been performed with more or less solidness in their results. Ginno et al. (2012) obtained a first genome-wide distribution of in vivo formed RNA:DNA hybrids by immunoprecipitation with the S9.6 antibody and subsequent DNA sequencing in mammal cells. The method used, called DRIP-seq (DNA-RNA hybrid immunoprecipitation followed by sequencing), allowed for the creation of the first map of RNA:DNA hybrid-prone forming regions in the human genome in vivo. Similar approaches have been accomplished over the years taking advantage of the model organism, Saccharomyces cerevisiae. Two preliminary studies (El Hage et al., 2014 and Chan et al., 2014) undertook the same task employing different techniques to analyse the complete genome, the first through chromatin immunoprecipitation and sequencing (ChIP-seq) and the second one by hybridization using tilling arrays (DRIP-chip). The conclusions dragged from these works showed only small differences and agreed with previous single-gene analyses, indicating that R-loops preferentially accumulated at highly transcribed genes and/or at high GC DNA regions and a major presence at the rDNA locus and retrotransposons. Recently, Wahba et al. (2016) improved the DRIP assay by adding a prior treatment with S1 nuclease that degrades the ssDNA displaced in the formation of the RNA:DNA hybrid. This step allows for the sonication of the samples without major disruption of the hybrids, leading to an important background reduction from the sequencing signal. Their results in wild-type strain yeasts stated that the work from Chan et al. (2014) overestimated positive regions and the study by el Hage et al. (2014) fell too short in their selection. However, a subset of the regions identified by Wahba et al. (2016) corroborated a number of findings from those previous DRIP-chip and/or ChIP-seq studies, including the occurrence of hybrids at the mitochondria, rDNA locus, telomeres, Ty elements, and tRNA genes. The new method allows a better Ph.D. Thesis - Juan F. Lafuente Barquero 40 detection of the differences. It is, thus, clear that the incorporation of new steps in the DRIP processing is important to get an accurate estimate of R-loop formation through the genome at different conditions. However, a concern with the new technique based on S1 nuclease is that a breakage in DNA may facilitate hybrids at regions where they do not spontaneously form. Further analysis with different techniques, likely not relying in the S9.6 antibody, might be necessary in the future. 1.1. RNA:DNA hybrid mapping in wild-type and hpr1∆ mutant yeasts Having our laboratory identified the first factor involved in the regulation of R-loop presence, we decided to incorporate the genome-wide approach of R-loop accumulation to our studies. Experiments were performed applying a protocol similar to that performed in human cells (Ginno et al., 2012) rather than those used in yeast (Chan et al., 2014; El Hage et al., 2014; Wahba et al., 2016). The technical details of the procedures are covered in the Material and Methods’ chapter. Briefly, we performed a very delicate chromatin extraction followed by digestion with a cocktail of endonucleases for the fragmentation of DNA. Afterward, samples were split and whether or not treated with RNase H before being subjected to immunoprecipitation with the S9.6 antibody. RNA:DNA hybrids were purified and amplified with a Whole Genome Amplification Kit before sending to the Genomics Laboratory for the sequencing with an Ion Torrent® platform. Inputs were used as controls. We performed a first experiment using a S9.6 antibody purified in-house from the original hybridoma (HB-8730TM, ATCC®) to immunoprecipitate R-loops formed in wild-type and hpr1∆ mutant strains with a W303 background. Sequencing results were quality checked, pre-processed and aligned to the S. cerevisiae sacCer3 reference genome build using BOWTIE. Generated BAM files were further analysed with a number of tools including MACS2 for the peak calling of positive regions and deepTools for the generation of profiles, among others. We used the data previously published from GSE53420 (El Hage et al., 2014), E-MTAB-2388 Results I 41 (Chan et al., 2014) and SRP071346 (Wahba et al., 2016) to compare our results and estimate the precision of our methodology. After generation of the coverage profiles created with the alignment of the sequencing reads, we started the evaluation of our results by comparing with available datasets. The only study that provided with data in a format that we could compare in this way was El Hage et al., (2014). We loaded their read coverage files (bigwig format) directly into IGV (Integrative Genomics Viewer v2.3.48, Broad Institute) along with ours, and used the diagram generated at the mitochondrial DNA of the yeast as reference to determine the efficiency of RNA:DNA hybrid immunoprecipitations. R-loops are necessary for mtDNA replication (Baldacci et al., 1984; Reyes et al., 2013), therefore the pattern of IP obtained in this region could be used as control. As can be seen in Figure R1 the pattern obtained was quite similar to the reference Next we evaluated the overall performance of the experiment by analysing the sensitivity of hybrid-forming regions to RNase H treatment. For this, we normalized our sequencing results employing the same methodology and scripts used by El Hage et al., (2014). Briefly, we measured the read depth in all intergenic regions (not annotated regions) according to the most updated version of sacCer3 table of features at SGD. The mean value of this measurement was used to normalize the number of reads in each sample, so they could be comparable. Figure R2 collects the profiles for INPUT samples (blue lines) and S9.6 IP samples pre-treated with Figure R1. S9.6 signal profile over mitochondrial chromosome. Coverage profiles over the mitochondrial chromosome. Wild-type and double rnh∆ strains data from GSE53420 (El Hage et al., 2014). Wild-type and hpr1∆ mutant (DRIP A) performed with in-house S9.6 antibodies, respectively. WT rnh1∆ rnh201∆ WT hpr1∆ El Hage. 2014 DRIP A Mitochondrial Chromosome Ph.D. Thesis - Juan F. Lafuente Barquero 48 it was previously published that from the common RNA:DNA hybrid accumulating mutants (hpr1∆, rnh1∆ rnh201∆, sen1-1) the HPR1 deficient mutant presents the lowest signal values, specifically shown for these two types of non-coding RNAs in the study by Chan et al. (2014). Focusing in the analysis of ORFs, about 22% of the genes identified to accumulate hybrids did so both in wild-type and in hpr1∆ strains. Deletion of the HPR1 gene leaded to the formation of hybrids in genes with different features to those accumulating spontaneous R-Loops in the wildtype. The main characteristics (Figure R7C) being that they were longer, presented a slight but significative reduction in their G+C content, and their expression levels according to published data (Beyer et al., 2004) were significantly lower compared to the median for the wild-type, but had not statistical difference compared to the median of the genome. Metagene analysis of all 457 hybrid-prone ORFs common to both wild-type and hpr1∆ strains indicated that hybrids distribution was similar independent of 0 100 Chr I Chr XVI 0 100 Chr XV 0 100 Chr XIV 0 100 Chr XIII 0 100 Chr XII 0 100 Chr XI 0 100 Chr X 0 100 Chr VIII 0 100 Chr VII 0 100 Chr VI 0 100 Chr V 0 100 Chr M 0 100 Chr IX 0 100 Chr IV 0 100 Chr III 0 100 Chr II 0 100 0 500000 1000000 1500000 Nº of Peaks S. cerevisiae genome: 12.105.075 bp Coverage (bp) % Genome hpr1∆ (Chan et al.)6247 3439053 28.3 WT (Chan et al.)6235 2860597 23.5 WT (el Hage et al.)3834 1611284 13.3 WT 1723 927666 7.6 WT (Wahba et al.)1782 803681 6.6 hpr1∆1744 635259 5.2 Figure R6. Distribution of RNA:DNA hybrids across S. Cerevisiae genome. Genomic view of RNA:DNA hybrids in WT(WRBb.9D) -blue- and hpr1∆ (HRBb.12) –redcells are shown. Representation of chromosomes with the value corresponding to Integer Score provided in the broadpeak file as result of peak calling using MACS2. The X-axis corresponds to coordinates in bp. Results I 49 the transcription factor deletion. The signal clusters spreaded along the length of transcription units with a slight peak in signal at the 5’ end and a little higher density towards the 3’ end of the genes (Figure R8). Table R2. Features mapped in WT and hpr1 The usefulness of DRIP profiling is widely demonstrated, and uncovers very interesting perspectives when the results are backcrossed with the huge quantity readily available about transcription and replication in the yeast and mutant strains. Therefore, although due to time limitations, this will be a future project of research ahead of this thesis. Figure R7. Analysis of RNA:DNA hybrid positive features in WT and hpr1∆. (A) Table R2. RNA:DNA hybrids positive features in WT (WRBb.9D) and hpr1∆ (HRBb.12) strains. Numbers represent the number of hits in each category. Percentages calculated respect to the total number of features in the genome for each category (B) Venn diagram showing the overlap between the WT and hpr1∆ positive genes (C) Statistical analysis of Length, G+C content and model-based expression values for genes mapped to be R-loop prone in WT and hpr1∆ strains. Median values are represented. Black line for the whole-genome median. *, P<0.05, **P<0.001, ****P<0.0001(Mann-Whitney’s U-test) Lenght Gene Lenght (bp) WT hpr1∆ 0 500 1000 1500 2000 ** 1071 **** 1413 **** 1583 0 2 4 6 10 8 Expression Values Expression Values WT hpr1∆ ** 7 7.2 6.4 G+C % GC content (%) WT hpr1∆ 10 20 30 40 50 0 39.7 **** 39.3 **** 38.9 * Genome WT hpr1∆ ORF 6604 1276 (19.3%) 1286 (19.5%) Centromere 16 0 (.%) 1 (6.3%) Telomere 32 17 (53.1%) 17 (53.1%) ARS 352 43 (12.2%) 43 (12.2%) Intron 377 41 (10.9%) 36 (9.5%) sn/snoRNA gene 83 28 (33.7%) 5 (6.%) tRNA gene 299 91 (30.4%) 58 (19.4%) rRNA gene 27 4 (14.8%) 4 (14.8%) ncRNA gene 18 4 (22.2%) 3 (16.7%) Transposon 91 83 (91.2%) 59 (64.8%) Total Hits 16455 3462 (21.%) 3191 (19.4%) 457 (21.7%) WT 1276 hpr1∆ 1286 ORF A B C 819 829 Ph.D. Thesis - Juan F. Lafuente Barquero 50 Average GeneIntergenic DownstreamUpstream Intergenic Signal 0.5 0.85 1.2 1.55 1.9 2.2 hpr1∆ WT Average GeneIntergenic DownstreamUpstream Intergenic Signal 0.5 0.85 1.2 1.55 1.9 2.2 WT (el Hage et al., 2014) WT A B (from Chan et al. 2014) Figure R8. Signal distribution over hybrid-positive ORFs in wild-type and hpr1∆. Metagene that compiles the signal median over all the hybrid-prone ORFs common to the WT (WRBb.9b) –orange line- and hpr1∆ (HRBb.12) –green linestrains. Results I 51 2. trans RNA as source of genomic instability The fact that the nascent RNA is able to induce hyperrecombination during the transcriptional process in absence of a fully functional THO complex has already been well stablished (Chávez & Aguilera, 1997; Huertas & Aguilera, 2003). It is also known that the deficiency of this complex and other RNA related factors also lead to the accumulation of mRNAs in the nucleus (Strässer et al., 2002; Jimeno et al., 2002; Paul & Montpetit, 2016). This supports that the co-transcriptional binding of many RNA processing and transcription factors is important to prevent hybrid formation by restricting the access of nascent RNA molecules to the DNA template at the site of transcription (Aguilera & García-Muse, 2012). However, it has been proven that these RNA biogenesis factors are not sufficient to prevent transient hybrid formation at some loci in wild-type budding yeast; rather, hybrids may form naturally but are removed by factors like the two endogenous RNase H enzymes and presumably Sen1p, an RNA:DNA helicase (Mischo et al., 2011; Wahba et al., 2011). Some recent studies have also proposed that mRNA molecules are not only susceptible of forming a DNA:RNA hybrid during the transcription process but also independent of it and even at different sites to where they were originally being produced, that is in trans (Wahba et al., 2013; Keskin et al., 2014). Therefrom, we wanted to further investigate the capability of RNA molecules to induce hyperrecombination in trans. In detail, our goal was to know if transcription of a DNA fragment is able to induce recombination of another homologous fragment not being actively transcribed at that moment. In order to achieve this, we devised a strategy based on the use of different inducible promoters to control the bacterial LacZ gene expression in S. cerevisiae cells. We initially cotransformed different yeast mutants with two plasmids (Figure R9): (a) The plasmid pCM190 bearing different sequences of the LacZ gene under the control of a tet promoter, which could be repressed by addition of doxycycline to the media. This transcriptional fusion would be the source of the mRNA in trans. Ph.D. Thesis - Juan F. Lafuente Barquero 52 (b) The plasmid pRS314-GL-LacZ, which carried the GL-LacZ recombination system (González-Barrera et al., 2002). This construct consisted in the sequence of the LacZ gene integrated between two truncated fragments of the LEU2 gene, which share 200bp of homology. This transcriptional fusion is controlled by a GAL1 promoter, activated by addition of galactose to the media or completely repressed in presence of glucose. Active transcription of this plasmid would produce mRNA in cis. With these plasmids, depending on the simultaneous state of both promoters, we could work with a total of four possible conditions to study the impact of RNA from two different sources on the recombination frequency of our system, measured as the frequency of Leu+ colonies. leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in trans mRNA in cis leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in trans leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in cis Transcription in cis: ON leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ Transcription in cis: OFF Transcription in trans: OFF Transcription in trans: ON tetp GALp tetp tetp GALp GALp tetp GALp (a) (b) Basal Cis Trans Cis + Trans Figure R9. Schematic representation of the assay. Schematic representation of the two mRNA sources used in the experiments and the four combinations of the promoters’ activation states. Results I 53 2.1. Impact of RNA in trans over the recombination frequency We started studying the capability of an mRNA molecule to induce recombination at a different homologous target in a physiological environment. For this, we cotransformed the wild-type strain W303.1A with the recombination plasmid pRS314-GL-LacZ and the plasmid pCM190 carrying the gene LacZ or not. Recombination frequencies for the GL-LacZ system in each of the four conditions tested are shown in Figure R10. When no transcript was generated, recombination levels were as expected for a wild-type strain. Generation of an in cis mRNA induced a significant increase of almost 3-fold in the recombination frequency, connected to the transcription-associated recombination (TAR) (Prado & Aguilera, 2005). However, when transcription of the LacZ gene was active in trans (Figure R10B, red bars) no significant difference could be observed in either condition, whether transcription in cis was active or not. These results suggested that recombination of a sequence induced by a homologous mRNA transcribed from a different locus do not represent a detectable source of genetic instability as measured by means of our system in wild-type cells. leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in trans mRNA in cis leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in trans leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ mRNA in cis Transcription in cis: ON leu2Δ3’ LacZ leu2Δ5’ GL-LacZ tet P:LacZ LacZ Transcription in cis: OFF Transcription in trans: OFFTranscription in trans: ON tetp GALp tetp tetp GALp GALp tetp GALp (a) (b) A Recombination frequency (x10 ) -4 2 tet P:LacZ LacZ tet p cis ON cis OFF B Trans - Trans + Figure R10. Analysis of the effect of RNA in trans on genetic recombination (I) (A) Schematic representation of the two mRNA sources used in the experiments and the four combinations of the promoters activation states. (1) cis OFF/ trans OFF (2) cis OFF/ trans ON (3) cis ON/ trans OFF (4) cis ON/ trans ON (B) Recombination analysis in WT (W303) strain carrying GL-LacZ plasmid system (pRS314-GL-LacZ) plus either the empty vector pCM190 (- Trans mRNA) or the same vector (+ Trans mRNA) carrying the LacZ gene. Average and SD of at least three independent experiments consisting in the median value of six independent colonies each are shown. ***, p≤0.001; ****, p≤0.0001. (Student’s t-test). A scheme of LacZ transcriptional fusion used in this experiment is shown on top. Ph.D. Thesis - Juan F. Lafuente Barquero 54 For the formation of a hybrid between DNA and RNA synthetized in a different location, a spontaneous annealing may be required to take place. Transcription is a tightly regulated process, in which the constitution of the ribonucleoparticle (mRNP) is essential to ensure the stability and export of the nascent mRNA. Therefore, we wondered if the accumulation of unprotected RNA molecules like those generated in THO mutants (Luna et al., 2012) could develop in an increase of spontaneous recombination events in trans. For this, we cotransformed a W303 wild-type strain and the congenic mft1∆ and hpr1∆ mutants. Both proteins are components of the THO complex, and their absence cause phenotypes related to the formation and accumulation of DNA:RNA hybrids, but with the first mutant presenting more moderate effects than the latter. Recombination frequencies obtained are shown in Figure R11. When both promoters are switched off (Figure R11A, blue bars), there is no LacZ mRNA being produced, and so, recombination levels stay at low values in the wild-type. The increase observed in the mutants can be attributed to the background instability phenotype of these mutations or to a leaky transcription AB WT Trans mRNA - Trans mRNA + Recombination frequency (x10 ) -4 mft1∆ hpr1∆ 1 10 10 4 10 3 10 2 * WT mft1∆ hpr1∆ 1 Recombination frequency (x10 ) -4 1 10 10 4 10 3 10 2 Trans mRNA - Trans mRNA + ** Transcription in cis: OFF Transcription in cis: ON tet P:LacZ LacZ tetp Figure R11. Analysis of the effect of RNA in trans on genetic recombination (II) Recombination analysis in WT (W303), mft1∆ (WMK.1A) and hpr1∆ (U678.4C) strains carrying GL- LacZ plasmid system (pRS314-GL-LacZ) plus either the empty vector pCM190 (- Trans mRNA) or the same vector carrying the LacZ gene (+ Trans mRNA). Average and SD of at least three independent experiments consisting in the median value of six independent colonies each are shown. *, p≤0.05. (Student’s t-test). A scheme of LacZ transcriptional fusion used in this experiment is shown on top. Results I 55 from the promoter. Under this condition, the presence of an mRNA in trans (Figure R11A, red bars) had no effect on recombination, only a minor statistically significant difference was observed in the mft1∆ strain alone. On the other hand, when transcription was activated in cis, the recombination frequency increased noticeably as it was expected from these mutations (Figure R11B, blue bars). Unexpectedly, presence of mRNA molecules in trans led to a moderate but significant suppression of hyper-recombination in both mutants (Figure R11B, red bars). In view of these results, we reasoned that failure in the formation of a functional ribonucleoprotein (mRNP), as it does occur in the THO mutants studied, can lead to the initiation of recombination events in homologous regions where transcription is not active. However, these events seem to take place at a very low frequency and we still would need to figure out, with a molecular assay such as a DRIP, whether the HR is due to the spontaneous formation of a DNA:RNA hybrid in trans or not. The suppression effect observed when both cis and trans mRNAs were produced at the same time could have different explanations. For example, the trans mRNA could interfere with R-loop formation by affecting the cis transcription, resulting in a reduction of RNA production. Alternatively, the trans mRNA could anneal with the displaced ssDNA in the cis transcribing locus, interfering in some way with the ability of the co-transcriptional R-loop to trigger HR. For this reason, we wanted to further elucidate the mechanism behind the capacity of the mRNA in trans to partially suppress the hyper-recombination phenotype of these mutants. For this, we cotransformed the plasmid pGL-LacZ carrying the recombination system together with a plasmid, in which the LacZ sequence had been cloned in the opposite direction of transcription (inverted LacZ, LacZi). As can be observed in Figure R12A, the wild-type strain did not manifest any effect generated by the production of a trans mRNA molecule. Transcription of an antisense transcript in trans did, however, led to a small but significant reduction in the recombination frequency of both mutant strains assayed, opposing to the result obtained when the trans was sense to the cis gene target in the recombination Ph.D. Thesis - Juan F. Lafuente Barquero 56 system (Figure R11A, red bars). Cis transcription activation led to the expected TAR-associated increase in the recombination frequency value when trans mRNA was absent (Figure R12B, blue bars). Then again, the presence of the antisense mRNA produced in trans when the in cis transcription was active produced a significant drop on the recombination levels (Figure R12B, red bars), but just in the THO mutant strains, as values for the wild-type remained unaltered. Up to this point, our data indicated that genomic instability associated to the production of trans mRNA did not have a observable impact on a wild-type strain assayed with our system. Additionally, the THO mutants assessed did show a difference when trans mRNA was being produced, and both behaved similarly, as expected for two components of the same complex. Because mft1∆ mutant grew better than hpr1∆, we decided to broaden the analysis substituting the Hpr1 deficient strain for a different one. Double RNase H mutants, rnh1∆ rnh201∆, are also prone to the accumulation of mRNAs, but present no defects in the transcriptional process (Huertas & Aguilera, 2003). For this experiment we cotransformed yeast cells with the plasmid pCM189 carrying the tetp::LacZ fusion and with the pGL-LacZ plasmid for the analysis of recombination. AB mft1∆ hpr1∆ WT 1 Recombination frequency (x10 ) -4 1 10 10 4 10 3 10 2 Trans mRNA - Trans mRNA + 1 Recombination frequency (x10 ) -4 WT mft1∆ hpr1∆ 1 10 10 4 10 3 10 2 Trans mRNA - Trans mRNA + ** * *** *** Transcription in cis: OFF Transcription in cis: ON tet P:LacZi LacZ tetp Figure R12. Analysis of the effect of RNA in trans on genetic recombination (III) Recombination analysis in WT (W303), mft1∆ (WMK.1A) and hpr1∆ (U678.4C) strains carrying GL- LacZ plasmid system (pRS314-GL-LacZ) plus either the empty vector pCM190 (- Trans mRNA) or the same vector carrying the inverted sequence of the LacZ gene (+ Trans mRNA). Average and SD of at least three independent experiments consisting in the median value of six independent colonies each are shown. *, p≤0.05; **, p≤0.01; ***, p≤0.001. (Student’s t-test). A scheme of LacZ transcriptional fusion used in this experiment is shown on top Results I 57 Blue bars in Figure R13 show that transcription in cis alone is responsible of a clear impact on recombination levels in all three strains assayed, with values for the double RNH mutant a little lower than that of the THO mutant, mft1∆. For this experiment, the values obtained when simultaneously expressing the trans mRNA show no effect in neither condition, with the cis transcription repressed (Figure R13A, red bars) or active (Figure R13B, red bars). We obtained consistent results with the wild-type strain, in which we could not observe any significant effect of the trans mRNA expression. Even though the mft1∆ mutant displayed a small statistically significant increase in instability in the first experiment when transcribing the trans mRNA and the cis target was repressed (Figure R11A), we could not replicate that result. This also occurred when we activated the cis transcription; the hyper-recombination phenotype could not be as effectively repressed as in previous assays by the trans mRNA molecule. Also, loss of RNH function didn’t seem to increase the odds for the trans mRNA to generate more DNA:RNA hybrids that led to an increase in recombination frequency as measured by this assay. 1 1 10 10 4 10 3 10 2 Trans mRNA - Trans mRNA + Transcription in cis: OFF AB Recombination frequency (x10 ) -4 WT mft1∆rnh1∆ rnh201∆ 1 Recombination frequency (x10 ) -4 WT mft1∆rnh1∆ rnh201∆ 1 10 10 4 10 3 10 2 Trans mRNA - Trans mRNA + * Transcription in cis: ON tet P:LacZ LacZ tetp Figure R13. Analysis of the effect of RNA in trans on genetic recombination (IV) Recombination analysis in WT (W303), rnh1∆ rnh201∆ (HRN2.1C) and mft1∆ (WMK.1A) strains carrying GL-LacZ plasmid system (pRS314-GL-LacZ) plus either the empty vector pCM190 (- Trans mRNA) or the same vector carrying the LacZ gene (+ Trans mRNA). Average and SD of at least three independent experiments consisting in the median value of six independent colonies each are shown. *, p≤0.05. (Student’s t-test). A scheme of LacZ transcriptional fusion used in this experiment is shown on top. Results II Results II 67 3. Search for novel factors involved in DNA:RNA hybrid metabolism DNA:RNA hybrid excess or its deregulated presence are a well-demonstrated source of genomic instability (Huertas & Aguilera, 2003; Castellano-Pozo et al., 2013; Herrera-Moyano et al., 2014; García-Rubio et al., 2015). However, they are natural occurring structures with a variety of functions, both in prokaryotic and eukaryotic cells (Aguilera & García-Muse, 2012). These facts lead to the idea that there must be many factors and proteins involved in the useful production and on time degradation of these nucleic hybrids. Many recent studies have proposed or shown the importance on these processes of transcription factors, replication regulators and other factors. (Gómez-González et al., 2011; Herrera-Moyano et al., 2014; Mischo et al., 2011). However, the physiological meaning and mechanisms of R-Loop processing are still not understood. In order to gain further insight into this problem, we decided to search for proteins that had not been previously related with R-loop metabolism. 3.1. Selection of candidates Rad52-YFP foci are formed at recombination centres in the nucleus of the cell (Lisby et al., 2004). Their abundance is directly related to the manifestation of instability problems. Thus, they constitute a sensitive tool to determine if mutants have difficulties in DNA repair or during the DNA replication process. Taking advantage of this visible and measurable phenotype, we designed our screening aiming at finding mutations that had increased Rad52-YFP foci formation that could be suppressed by the overexpression of RNase H. This procedure would let us screen for novel factors potentially implicated in R-loop homeostasis. We chose to focus our efforts in a selected number of proteins, members of the DEAD box superfamily of helicases. The list of candidates (Table R3) was elaborated by picking up mutants according to previous data generated in our laboratory or recently published by other researches related. Ph.D. Thesis - Juan F. Lafuente Barquero 68 Table R3. Dead Box Helicases analyzed Source Systematic Name Human homolog BRCA2 Co-IP YBR142W MAK5 DDX24 YHR065C RRP3 DDX47 YDL084W SUB2 DDX39A, DDX39B YDL160C DHH1 DDX6 YDR021W FAL1 DDX21, DDX50, EIF4A3 YDR243C PRP28 DDX23 YGL171W ROK1 DDX52 YLR276C DBP9 DDX56 YNL112W DBP2 DDX17, DDX5 YOR046C DBP5 DDX19A, YOR204W DED1 DDX3X, DDX1, YPL119C DBP1 DDX3X, DDX1, YBR237W PRP5 DDX23 YJL138C TIF2 EIF4A3 YKR059W TIF1 EIF4A3 YGL078C DBP3 DDX21, DDX50 HBD Co-IP YNL112W DBP2 DDX17, DDX5 YOR204W DED1 DDX3Y YPL119C DBP1 DDX3Y Stirling, 2012 YLR274W MCM5 MCM5 YPL235W RVB2 RUVBL2 YBL023C MCM2 MCM2 Chan, 2014 YDL084W SUB2 DDX39, UAP56 YNR038W DBP6 DDX51 YKR024C DBP7 DDX31 YLR424W SPP382 TFIP11 YLR430W SEN1 SETX YPL082C MOT1 BTAF1 Other YIL084C SDS3 BRMS1 YIR002C MPH1 FANCM YCL061C MRC1 Claspin YJL092W SRS2 RETL1 YMR190C SGS1 BLM YHR031C RRM3 PIF1 Results II 69 The specified factors were selected by the criteria detailed below: a. The BRCA2 candidates were selected among a list of positive interactors identified in a co-immunoprecipitation assay performed with protein extracts from HeLa cells (V. Bhatia, unpublished). b. Three DEAD-Box helicases were identified among other factors when immunoprecipitating DNA:RNA hybrids from HeLa’s chromatin extracts expressing the HBD-GFP construct (Bhatia et al., 2014). c. Three positive candidates related to de DEAD-Box helicase family, associated with DNA replication processes, were selected among the mRNA cleavage and polyadenylation factors involved in the maintenance of chromosome stability identified by Stirling et al., (2012). d. Four DEAD box helicases selected among a list of mutations described to confer chromosome instability phenotypes that increase R-loops as detected by means of immune-fluorescence using the S9.6 antibody in yeast chromosome spreads (Chan et al., 2014). e. Six factors including five helicases, involved in DNA repair or replication (SRS2, SGS1, RRM3, MRC1, MPH1 and SDS3). 3.2. Screening We transformed Saccharomyces cerevisiae cells with a plasmid bearing the Rad52-YFP construct under its endogenous promoter. The 25 mutants were divided in two groups, one for the strains from the KO deletion collection of non-essential genes (EUROSCARF, (Winzeler, 1999)) and the second batch with all the essential genes handpicked from the yeast Tet Hughes collection of essential genes (yTHC hereafter) (Mnaimneh et al., 2004), in which the endogenous promoters from selected genes had been substituted by a tetracycline controlled transactivator. The genetic background for all the mutants is derived from BY4741, and the cells were transformed using the lithium acetate method (Ito et al., 1983) adapted according to (Schiestl & Gietz, 1989). Three different transformants of each mutant were grown in different days to exponential phase in 3 ml of minimum selective media at 30ºC. Cells were fixed and Ph.D. Thesis - Juan F. Lafuente Barquero 70 visualized in a Leica Microscope (see M&M 4.9). From the initial round of selection (Figure R18) we could identify 6 mutants from the non-essential helicases and 7 from the essential ones that showed a spontaneous increase in the accumulation of Rad52 foci with respect to their wildtypes. Our values correlate quite well with previously published data for some of the strains, as for example, our positive control hpr1∆ shows an accumulation around 20% of S/G2 cells with Rad52-YFP foci equally observed by (Alvaro et al., 2007) giving consistency to our results. Once identified those helicases whose mutations affected the stability of the genome, we assayed whether such alterations could be derived from the accumulation of DNA:RNA hybrids. For this, selected strains were transformed again with a plasmid carrying the RNaseH gene (RNH1) under the control of a tetracycline promoter in the case of the KO mutants, and a GAPDH promoter for the yTHC strains. Only those mutants that showed a consistent reduction in the percentage of cells with Rad52 foci (Figure R19) were considered positive candidates for the screening, and are described in the following table (Table R4) Non Essential helicases WT hpr1 ∆ mph1 ∆ mrc1 ∆ dbp1 ∆ dbp3 ∆ dbp7 ∆ tif1 ∆ tif2 ∆ sds3 ∆ dhh1 ∆ sgs1 ∆ srs2 ∆ rrm3 ∆ 0 10 20 30 40 50 S/G2 cells with Rad52-YFP foci (%) * * * * * * * Essential helicases WT tet:SEN1 tet:DED1 tet:MAK5 tet:FAL1 tet:MOT1 tet:DBP6 tet:DBP9 tet:MCM5 tet:PRP5 tet:PRP28 tet:ROK1 tet:RVB2 tet:SSP382 0 10 20 30 40 50 S/G2 cells with Rad52-YFP foci (%) * * ** * * * * Figure R18. Genetic instability in different DNA and RNA helicase mutants. Spontaneous Rad52-YFP foci formation in WT and mutant candidates of the selected helicases. The WT used for the non-essential gene series is BY4741 strain and for the essential gene series is R1158. Average and SD of at least three independent experiments are shown. *, p≤0.05 (Student’s t-test). Results II 71 From these proteins, we discarded for further investigations the RNA binding protein Prp28. It was the only essential gene in the list and was not a helicase itself. The variety of different experimental conditions used to be able to obtain a reliable result suggested the need for further re-evaluation before considering it as a definitive candidate. Essential helicases 0 10 20 30 40 50 RNH - + RNH S/G2 cells with Rad52-YFP foci (%) WT tet:SEN1 tet:DED1 tet:MAK5 tet:PRP5 tet:PRP28 tet:RVB2 tet:SSP382 tet:ROK1 * * Non Essential helicases WT hpr1∆ mph1∆ mrc1∆ dbp1∆ sgs1∆ srs2∆ rrm3∆ 0 10 20 30 40 50 S/G2 cells with Rad52-YFP foci (%) RNH - + RNH * * * * Figure R19. Supression of Rad52-YFP foci by RNH1 overexpression in different helicase mutants Spontaneous Rad52-YFP formation in WT and selected candidates carrying the empty plasmid pCM189 (-RNH) or pCM189:RNH1 (+RNH) in the deletion collection strains and the empty plasmid pRS425 (-RNH) or pRS425:RNH1 (+RNH) in the yTHC strains. Average and SD of at least three independent experiments are shown. *, p≤0.05 (Student’s t-test) for each mutant comparing -/+ RNH conditions. Gene Name Function Mph1 YIR002C DNA helicase. Involved in translession and RF restart. Similar to FANCM Mrc1 YCL061C S-phase checkpoint protein required for DNA replication. Couples DNA helicase with polymerase. Dbp1 YPL119C Putative ATP-dependent RNA helicase. Required for translation of mRNA. Paralog of DED1. Prp28 YDR243C RNA binding protein; no helicase activity. Required for processing of pre-mRNA Table R4. Selected candidates Ph.D. Thesis - Juan F. Lafuente Barquero 72 3.3. Effect of AID on the genetic instability of candidate mutants Next, we assayed whether the instability phenotype observed was directly related to the accumulation of R-loops. To gain further insight on this regard, we expressed the human cytidine deaminase AID (Activation Induced Deaminse). AID uses the single stranded DNA displaced in the R-loop as target generating a greater instability defect in yeast strains prone to the accumulation of DNA:RNA hybrids (Gómez-González & Aguilera, 2007; Mischo et al., 2011). We expressed the AID gene under the tet promoter from the pCM189::AID plasmid in yeast cells expressing the RAD52-YFP fusion and measured the accumulation of Rad52 foci as indicated above. Interestingly, although overexpression of RNH1 consistently suppressed the increase in Rad52 foci, expression of AID did not significantly increase DNA damage above spontaneous levels (Figure R20) S/G2 cells with RAD52-YFP foci (%) mph1 ∆ dbp1 ∆ mrc1 ∆ 0 10 20 30 40 50 AID RNH Vector WT mph1 ∆ dbp1 ∆ mrc1 ∆ WT ** **** ** ** n.s. **** 7.5 7.2 9.8 22.8 11.5 21.2 19.2 19.0 28.2 30.6 30.8 9.5 Figure R20. Effect of AID and Rnase H1 expression on genetic instability in selected mutants Spontaneous Rad52-YFP foci accumulation in WT (BY4741), mph1∆ (YIR002C), dbp1∆ (YPL119C) and mrc1∆ (YCL061C) strains carrying the empty vector pCM189 or expressing RNase H1 from the plasmid pCM189:RNH1 (RNH) or AID from the plasmid pCM189:AID (AID). Average and SD of at least three independent experiments are shown. Representative microscope images are shown on the right. *, p≤0.05; **, p≤0.01; ***, p≤0.001; ****, p≤0.0001. (Student’s t-test). Results II 73 3.4. R-Loop accumulation as detected by immunofluorescence Suppression of Rad52 foci by RNH1 overexpression and their exacerbation by AID over-expression serve to indirectly assess the presence of DNA:RNA hybrids. To ascertain the accumulation of these nucleic structures in our mutant candidates we performed immunofluorescence assays using the S9.6 antibody on chromosome spreads from the selected candidate strains. The experiment was done including as control the known R-loop-accumulating strains such as rnh1∆ rnh201∆ and hpr1∆. Our results (Figure R21) show that the double RNase H mutant strain is the one that shows the highest signal accumulation value of all mutants, consistent with previous observations (Chan et al., 2014). Interestingly, our 3 selected candidates also show a significative increase over the wild-type levels, reassuring a novel connection of these proteins with DNA:RNA hybrids metabolism. hpr1∆ mph1∆ dbp1∆ mrc1∆ rnh1∆rnh201∆WT Cells with S9.6 signal (%) 0 20 40 60 80 100 ≥ 3 foci 2 foci 1 foci mph1∆ dbp1∆ mrc1∆ rnh1∆ rnh201∆ hpr1∆ WT 6.5 6.9 5.8 9.7 7.3 36.3 5.7 12.3 62.8 12.8 16.5 10.5 12.8 30.9 34.9 19.9 7.7 24.5 * * ** ** Figure R21. Immunofluorescence of DNA:RNA hybrid accumulation in candidate strains. Percentage of nuclei and chromosomal masses containing S9.6 signal as detected by indirect immunofluorescence on yeast chromosome spreads in WT (W303.1A), mph1∆ (WMPH1-2B), hpr1∆ (HPBAR1-R) and rnh1∆ rnh201∆ (RNH-R) strains. Representative chromosome spreads are shown: blue stain is DNA (DAPI), red foci are DNA:RNA hybrids detected with S9.6 antibody and white bars indicate 10μm scale. Data represent mean from at least three independent experiments. Statistical analysis calculated with total number of signals for each mutant respect to WT values. *, p≤0.05 (Student’s t-test). Experiments performed in collaboration with J. Kaplan Ph.D. Thesis - Juan F. Lafuente Barquero 80 3.9. DNA:RNA hybrid detection by immunoprecipitation in mph1∆ mutants To demonstrate that R-loops accumulate at a molecular level in the mph1∆ mutant we performed a direct immunoprecipitation assay using the S9.6 monoclonal antibody at different genes. DNA:RNA hybrids clearly accumulated in the mutant in all DNA regions tested (Figure R26A). We also analysed this increase in two point mutants of the MPH1 gene mutated at key residues of the helicase domain. The mutant E210Q, which has a replacement of the glutamic acid in the DEAH motif, and the mutant Q603D with a similar modification in the key residue of the helicase motif. These mutants have the helicase function impaired, but behave like mph1∆ for DNA damage sensitivity (Chen et al., 2009). As can be observed in Figure R26B, these mutants also show a significant increase in the accumulation of DNA:RNA hybrids in the locus analysed. Thus, the helicase function, already characterized to be essential for the Mph1 role in recombination (Schürer et al., 2004; Prakash et al., 2009; Banerjee et al., 2008) is necessary to prevent R-loop accumulation. Results II 81 A * * * RNH 0 2 4 6 8 +- +- mph1∆ WT 1.5 0.3 4.2 0.6 S9.6 signal (A.U.) GCN4 B GCN4 mph1∆ mph1-Q603D mph1-E210Q WT RNH 0 2 4 6 8 +- +- +- +- * * ** ** ** P=0.081 P=0.056 1.0 0.6 2.8 0.2 3.9 0.4 4.0 0.6 S9.6 relative signal PDC1TRX1ARS1211 mph1∆ mph1-Q603D mph1-E210Q WT RNH +- +- +- +- S9.6 relative signal 0 2 4 6 8 * ** * * ** 1 1.1 2.6 0.3 3.7 0.6 3.1 0.8 PDC1TRX1ARS1211 RNH 0 5 10 15 20 +- +- * * * 7.5 2.5 12.3 3.5 S9.6 signal (A.U.) mph1∆ WT Figure R26. DNA:RNA hybrid accumulation in mph1∆ deletion and helicase-dead mutants. S9.6 immunoprecipitation of DNA:RNA hybrids (DRIP) in (A) WT (W303), mph1∆ (WMPH1.1A) or (B) WT (W303), mph1∆ (WMPH1.1A), mph1-E210Q (T617), mph1-Q603D (T597-1) strains at the indicated regions (black line). Samples were treated with RNase H before immunoprecipitation. Mean and SD for at least three independent experiments are shown. *, p≤0.05; **, p≤0.01 (Student’s t-test). Ph.D. Thesis - Juan F. Lafuente Barquero 82 4. Function of the human helicase RECQL5 in the model organism Saccharomyces cerevisiae The human helicase RECQL5 is a protein member of the well-conserved RecQ family of helicases. Demonstrated genome caretakers (futher reviewed in (Chu & Hickson, 2009; Monnat, 2010; Bernstein et al., 2010; Singh et al., 2011), there are a total of five members present in most mammals, while there is usually just one in lower organisms. RECQL5 is the only one that has been shown to directly interact with the RNA polymerase II (RNAPII) (Aygün et al., 2008; Islam et al., 2010). This relationship makes this helicase unique among the family, since their roles in genome stability maintenance through the regulation of diverse DNA repair processes had been well stablished. The fact that RECQL5 is present at the transcription interface raises many questions about the regulation of the interactions between this process and others that use the DNA as template, such as the replication or the repair processes. Owing to the absence of assays to address these questions in human cell cultures and the potential of model organisms for the study of the DNA metabolic processes, we decided to assess the capacity of the yeast Saccharomyces cerevisiae to serve as a tool for the functional study of the human helicase RECQL5. 4.1. Expression of RECQL5 in yeast Human RECQL5 transcript presents three alternatively spliced forms: RECQL5α, RECQL5β and RECQL5g (Shimamoto et al., 2000). The second splice variant, hereinafter referred as RECQL5, comprises a total of 991 amino acids and it is the most abundant and better characterized. It is expressed ubiquitously and independently of cell-cycle phase and cell-type (Kitao et al., 1998; Kawabe et al., 2000). In the scheme in Figure R27A, it is shown the domain organization of the functional wild-type protein, which comprises (from left to right) the DEAD- Box helicase (yellow box), the RecQ C-terminal domain (RQC, blue box), which contains the structural features specific of the RecQ helicases and the KIX domain (dark green box), present in several transcriptional regulators (Parker et al., Results II 83 1996) and responsible for the interaction with both the hypo- (IIa) and the hyperphosphorylated (IIo) forms of the RNAPII (Islam et al., 2010). This domain has been mutated, as depicted in the lower scheme in Figure R27A, (RECQL5 ID, Interaction Domain mutant) with a 27-bp deletion in the sequence, that supposedly abolished its interaction with the polymerase, as previously shown (Aygün et al., 2009). Other important domains are the BRC-repeat variant (BRCv, light green box), a structural motif also found in the tumor suppressor BRCA2 (Islam et al., 2012), which localized to the Rad51 binding region and required for the anti-recombinase activity of RECQL5. The nuclear localization signal (NLS, brown box) and the SRI domain (Set2 Rpb1-interacting domain, orange box) present in the histone methyltransferase Set2 (Kizer et al., 2005), known to interact with the phosphorylated C-terminal domain of the RNAPII (pCTD). We transformed the wild-type strain W303.1A with multicopy yeast expression plasmids carrying both versions of RECQL5 under the control of a GAL1 promoter. Figure R27B and C illustrates that overexpression of both versions of RECQL5 in S. cerevisiae dramatically affected growth as shown in plates of synthetic media with galactose as the sole carbon source. Due to this severe effect, that did not let us pursue the experiments, we undertook two approaches to reduce the levels of RECQL5 in the cells. First, we added different amounts of glucose to the synthetic media containing 2% galactose, in order to reduce the activation level of the GAL1 promoter, in addition, we supplemented the cells with an alternative carbon source. We selected the media with 0.05% glucose and 2% galactose, in which cells showed reduced colony size that evidenced the expression of RECQL5, but were still able to grow. On the other hand, we cloned both versions of the human gene into a different vector backbone, in this case, either pCM184 (TRP1) or pCM189 (URA3). These are centromeric vectors that contain a tet chimeric promoter that is repressed by addition of doxycycline to the medium (Figure R27D). The new transcriptional fusion tetp::RECQL5 generated did not seem to have any visible effect in yeast viability. The correct expression at low levels of the RECQL5 protein was demonstrated by western blot (Figure R27E). Ph.D. Thesis - Juan F. Lafuente Barquero 84 Figure R27. Expression of the human helicase RECQL5 in S. cerevisiae. (A) Domain organization of the RECQL5 protein. Top scheme corresponds to the wild-type version (RECQL5). Bottom scheme shows a mutant version with a deletion of 9 aa in the KIX domain (RECQL5 ID). (B) Top, scheme showing the multicopy (2μ) and galactose inducible (GAL1p) vectors used, pYES2 (URA3) and pYES3 (TRP1). 10-fold serial dilutions of the W303.1A strain transformed with the pYES2 vector, or carrying RECQL5 or RECQL5 ID on minimum synthetic selective media with different carbon sources are shown. (C) Growth of wild-type yeast strain W303.1A transformed with plasmids either pYES3 (Vector), pYES3:RECQL5 (RECQL5) or pYES3:RECQL5 ID (RECQL5 ID) and streaked on synthetic media with different concentrations of sugars. Two different transformants (T1 and T2) are shown for the plasmids carrying genes. Photographs were taken after 3 days of growth at 30°C. (D) Scheme showing the centromeric (CEN) and doxycycline repressible (tetp) vectors used, pCM184 (TRP1) and pCM189 (URA3). 10-fold serial dilutions of the W303.1A strain transformed with the pCM184 vector or carrying RECQL5 or RECQL5 ID on selective media are shown. Photographs were taken after 3 days of growth at 30°C. (E) Western Blot analysis using the FLAG antibody to detect the tagged RECQL5 protein levels expressed from the or the tet promoter. The strain W303.1A was transformed with the indicated plasmids. Actin immunodetection is shown as loading control. DEAD Helicase domain RQC KIX BRCv NLS SRI RECQL5 (1-991) RECQL5 ID (542-552∆) RQC BRCv NLS SRI KIX DEAD Helicase domain RQC BRCv NLS SRI A FLAG 109 KDa 42 KDa Actin GAL:RECQL5 Empty Empty tet:RECQL5 GAL:RECQL5 ID tet:RECQL5 ID Vector RECQL5 T1 RECQL5 ID T1 RECQL5 T2 RECQL5 ID T2 GLU 2% GAL 2% GLU 0.05% GAL 2% B D C E GLU 2% Vector GAL 2% RECQL5 RECQL5 ID URA3/TRP1 RECQL5 WT/ID GAL1P 2μ pYES2/3 GLU 2% Vector RECQL5 RECQL5 ID TRP1/URA3 RECQL5 WT/ID tetP CEN pCM184/9 Results II 85 4.2. RECQL5 recruitment to yeast chromatin We have seen that the heterologous expression of the human helicase was clearly interfering with some cellular processes in the yeast, generating viability defects at higher doses. Our first objective, however, was to assay whether RECQL5 behaved in our model organism similarly as it had been proposed in human cells, by regulating transcription in order to avoid conflicts with replication (Aygün et al., 2009). For this, we performed co-immunoprecipitation (Co-IP) assays using an antibody against the FLAG epitope fused to our RECQL5 proteins and an antibody that recognised the Rpb3 subunit of the RNAPII. We transformed the W303 yeasts cells with pYES3 plasmids, either empty or carrying the RECQL5 or the mutant ID genes, and grew them for 3 hours in galactose until exponential phase. Through western blot of the corresponding immunoprecipitated protein extracts we demonstrated that RECQL5 was able to interact with the S. cerevisiae RNAPII (Figure R28A). Surprisingly, the mutant RECQL5 ID, which was supposed to have this capability abolished (Aygün et al., 2008), was also co-IP with RNAPII the same way as the wild-type helicase. We additionally performed ChIP (Chromatin Immunoprecipitation) to ensure that this interaction correlated with an active RNAPII, and not just with unspecific interaction. We transformed the W303.1A strain with the plasmid pCM184 either empty or carrying the wild-type RECQL5 or the mutant RECQL5 ID under the control of a tet promoter. Purified chromatin from these exponentially growing cells, either in glucose (repressed transcription) or galactose (active transcription), was subjected to IP using the antibodies described before. Results from this experiment (Figure R28B) also evidenced that both versions of RECQL5, wild-type and mutant, interacted with yeast’s chromatin and in a transcription-dependent manner. Our finding that the interaction of RECQL5 did not significantly affect the recruitment of RNAPII to the transcribing gene was opposed in to that observed in human cells by Saponaro et al. (2014), published later the completion of these experiments. This outcome was indicative that, although the interaction domains with the RNA polymerase could be conserved from yeast to humans, RECQL5 Ph.D. Thesis - Juan F. Lafuente Barquero 86 was not acting in the same exact way as when interacting with the human RNAPII in a physiological environment. In any case, our results prompted us to continue with the work, since the interaction between RECQL5 and the yeast RNAPII and its chromatin was confirmed, we could now study if this was the cause of the viability defects observed or there were other roles implicated that could help us learn more about the still unclear role of RECQL5. Vector RECQL5 RECQL5 ID RNApolII RECQL5-FLAG Input IP-Flag IP-RPB1 Input IP-Flag IP-RPB1 Input IP-Flag IP-RPB1 A B GAL1 (1.6Kb) RPB1 IP/INP Vector RECQL5 RECQL5 ID 0 2 4 6 8GLU GAL RecQL5 IP/INP Vector RECQL5 RECQL5 ID 0 2 4 6 8GLU GAL * Figure R28. Interaction of the human helicase RECQL5 with yeast chromatin. (A) Western Blot of different immunoprecipitations (IP) showing interactions between the two forms of the FLAG-tagged RECQL5 and RNAPII, using FLAG and RPB1 antibodies, respectively. Strain W303.1A was transformed with either pYES3 (Vector), pYES3:RECQL5 (RECQL5) or pYES3:RECQL5 ID (RECQL5 ID). (B) ChIP (Chromatin immunoprecipitation) of W303.1A cells transformed with either pCM184 (Vector), pCM184:RECQL5 (RECQL5) or pCM184:RECQL5ID (RECQL5 ID) plasmids. qPCR was performed in the GAL1 locus under conditions of active (GAL) or repressed (GLU) transcription. The black bar represents the amplified region. Results II 87 4.2.1. Sensitivity of RECQL5-expressing yeast cells to genotoxic agents The physiological role of RECQL5 in mammal cells had been associated with the maintenance of genome stability, presumably by acting as a regulating factor at the interface between transcription and replication. We wondered whether the expression of the human RECQL5 helicase in yeast could help the cells overcome the effect of the DNA damages produced by different genotoxic agents or, on the contrary, could interfere with the cellular response. For this, we transformed the wild-type strain W303.1A with the transcriptional fusions (both wild-type RECQL5 and the RECQL5 ID mutant) expressed from the GAL1p or the tetp promoters. By plating serial dilutions of exponential cultures in petri dishes we assayed the effect on viability when cells were grown on synthetic media supplemented with different concentrations of three different chemicals. We tested the genotoxic agents hidroxiurea (HU), which interferes with replication by depleting the amount of deoxyribonucleotide triphosphate (dNTPs); methylmethanesulfonate (MMS), an alkylating agent responsible of DNA methylation that leads to the formation of DSBs, presumably after the stalling of replication forks; and mycophenolic acid (MPA), a transcription elongation inhibitor that acts by interfering with the IMP dehydrogenase and, thus, the biosynthetic pathway of purine nucleotides impeding the transcription of newly synthesised mRNAs. The results obtained (Figure R29) evidenced that the presence of either the wild-type or the ID mutant version of the RECQL5 helicase, indistinctly, produced a clear impact on the maintenance of the yeast’s genome stability when cells were grown under genotoxic stress induced by HU or MMS, but not with the transcription elongation inhibitor MPA. Moreover, this effect was seen independently of the promoter strength, in both cases, either under the tetp or the GAL1p promoters. The presence of even low concentrations of HU or MMS in the media visibly affected the viability of the cells expressing either allele of the RECQL5 helicase in a similar way, compared to the yeast transformed with the empty vector, used as control in each condition. Surprisingly, MPA produced no effect in viability, no difference in growth rate was noticeable between the cells carrying the empty vector and those expressing RECQL5 gene at all MPA concentrations tested. This experiment Ph.D. Thesis - Juan F. Lafuente Barquero 88 Vector RECQL5 RECQL5 ID 10 mM 25 mM 50 mM 100 mM 0.002% 0.005% 0.01% 0.02% 25 μg/ml 12.5 μg/ml 50 μg/ml 100 μg/ml HU MMS MPA Glu 2% Glu 2%Gal 2% Glu 0.05% RECQL5 WT/ID tet P GAL P RECQL5 WT/ID Figure R29. Effects of RECQL5 overexpression on the sensitivity to different chemical compounds. Sensitivity of W303.1A strain transformed with indicated plasmids carrying the wild-type RECQL5 gene, or the mutated sequence of RECQL5 ID, or the empty vector, to different concentrations of genotoxic agents such as hidroxyurea (HU) and methylmetasulfonate (MMS) or the transcription elongation inhibitor mycophenolyc acid (MPA). 10-fold serial dilutions of exponentially growing cells in selective medium were plated on the indicated media. Photographs were taken after 3 days of growth at 30°C Results II 89 grounded the idea that heterologous expression of this protein was interfering with cell proliferation, possibly with some aspect of replication. Consistent with the sensitivity caused by HU and MMS, the absence of a detectable effect on cell viability after MPA does not allow us to conclude whether or not RECQL5 functionally interacts with the transcriptional process of the yeast. However, we cannot discard at this point that RECLQ5 mRNA synthesis was significantly diminished due to MPA action. 4.3. Impact of RECQL5 expression on the yeast genome stability In homologous recombination (HR), one of the first steps is the binding of Rad52 to the site of DNA breakage. This initial phase can be monitored with fluorescence microscopy and has been widely used to characterize the impact of mutations on genomic stability as a conssecuence of recombinogenic DSBs (Lisby et al., 2001; Alvaro et al., 2007). Taking advantage of this approach, we confirmed that overexpression of either RECQL5 or RECQL5 ID in a wild-type strain caused an increase in the accumulation of Rad52-YFP of about 4-fold above the control carrying the empty vector (Figure R30A). This increase was dependent on the expression level of the helicase, since the increment was only of 2-fold when RECQL5 was expressed from the tet promoter (Figure R30C, green bars). Taking into account the hypothesized roles of RECQL5 in a physiological environment, we next asked whether the helicase could be generating this phenotype of instability by means of a blockage in the replication machinery or a discoordination of transcription. Many transcription mutants, extensively studied in our laboratory such as hpr1∆, accumulate DNA damage due to errors in the replication process derived from transcription failures. One of the main sources of these instability events is the formation of DNA:RNA hybrids, whose presence is evidenced by a Rad52 foci accumulation that is suppressed by over expression of RNase H, which degrades the RNA strand of a DNA:RNA hybrid. Alternatively, hybrids can be inferred by expressing the human cytidine deaminase AID, which exploits the presence of the displaced single stranded DNA of an R-loop, which is used as substrate and will lead to further increase of RNase H-dependent Rad52 Ph.D. Thesis - Juan F. Lafuente Barquero 96 A Empty RECQL5 ID RECQL5 GAL1 mRNA (A.U.) 50 100 150 0 Time (min) 0 50 100 150 200 0 200 400 600 800 RT mRNA (A.U.) Time (min) 0 50 100 150 200 RT (4,3Kb) GAL1 (1,6Kb) rDNA GAL10GAL7KAP104 GAL1 FUR4 POA1 5’ 5’ Empty RECQL5 RECQL5 ID 0’ 15’ 40’ 100’170’ 0’ 15’ 40’ 100’170’ 0’ 15’ 40’ 100’ 170’ B 0 50 100 150 GAL10 mRNA (A.U.) Time (min) 0 50 100 150 200 0 50 100 150 RT mRNA (A.U.) Time (min) 0 50 100 150 200 RT (3,9Kb) GAL10 (2,1Kb) rDNA Empty RECQL5 0’ 15’ 40’ 100’170’ 0’ 15’ 40’ 100’170’ RECQL5 ID 0’ 15’ 40’ 100’170’ GAL10GAL7KAP104 GAL1 FUR4 POA1 5’ 5’ C 0 50 100 150 GAL10 mRNA (A.U.) Time (min) 0 50 100 150 200 RT mRNA (A.U.) 0 50 100 150 Time (min) 0 50 100 150 200 RT (3,9Kb) GAL10 (2,1Kb) rDNA Empty 0’ 15’ 40’ 100’170’ RECQL5 0’ 15’ 40’ 100’170’ RECQL5 ID 0’ 15’ 40’ 100’170’ GAL10GAL7KAP104 GAL1 FUR4 POA1 5’ 5’ Figure R33. RECQL5 expression impact on active transcription at the GAL locus. (A) Northern blot analysis of the GAL1 and (B) GAL10 mRNAs in W303.1A strain transformed with pYES2 expression vector carrying the indicated version of the RECQL5 gene. (C) Northern blot analysis of GAL10 mRNA in W303.1A strain transformed with pCM184 expression vectors carrying the indicated version of the RECQL5 gene. Schematic representation of the mRNAs detected and probe location (green bar) are shown. Graphs represent corresponding mRNA signal relative to rDNA. Mean of two experiments is shown. Results II 97 elongating polymerase fails to terminate properly entering into a co-transcribed downstream transcription unit, reducing its promoter activity. Transcriptional interference also has been documented to affect the proper use of both ARS and CEN elements inhibiting either replication or chromosomal segregation in mitosis. Our results evidence that both, the wild-type RECQL5 and the ID mutant, cause the same effect with comparable strength in the locus studied (Figure R33A-B), and this effect is also detected when the expression was driven from the tet promoter (Figure R33C). The GAL7 and GAL10 genes are part of the GAL gene cluster of S. cerevisiae genome, required for the metabolic pathway of galactose conversion into glucose- 6-phosphate (Fridovich-Keil & Jinks-Robertson, 1993). Both genes are arranged in tandem orientation and their high transcription rate facilitates that polymerases continue transcribing after the GAL10 poly(A) into GAL7 producing long readthrough transcripts, resulting in bi-cistronic mRNAs (Greger & Proudfoot, 1998). Promoter occlusion of GAL7 produced by this mechanism abolishes the synthesis of Gal7 protein, rendering the cells inviable since GAL7 is essential in the presence of galactose (Douglas & Hawthorne, 1964). For this reason, we wondered whether the viability defect observed when expressing RECQL5 from the GAL1 promoter was due to the read-through effect. As seen in Figure R34, wild-type cells simultaneously overexpressing Gal7 protein and RECQL5 grew better than those Vector RECQL5 GAL7 RECQL5+GAL7 SGAL SGAL + 100mM HU Figure R34. GAL7 deficit is not responsible for yeast sensitivity to genotoxic agents. Viability test with W303.1A strain transformed with GAL1 inducible vectors carrying the RECQL5 and/or the GAL7 genes. Serial dilutions (1:10) of exponentially growing cultures spotted on plates. Photographs were taken after 3 days of growth at 30°C. Ph.D. Thesis - Juan F. Lafuente Barquero 98 expressing only RECQL5 in a synthetic medium with galactose as the sole carbon source, demonstrating that the human helicase was interfering with the expression of the endogenous GAL7 gene, but this phenomenon was not responsible of diminishing the sensitivity to genotoxic agents like HU, as cell viability appeared compromised to the same degree in RECQL5-expressing yeast cells whether or not overexpressing GAL7. Aware that RECQL5 presence in S. cerevisiae cells had the capacity to alter the expression of some genes, we decided to perform a genome wide approach to search for transcripts whose levels would be affected and could explain the instability phenotypes observed. To achieve this, wild-type strain W303.1A was transformed with the plasmid carrying RECQL5 gene under the control of the tet promoter. RECQL5-expressing cells under this promoter only showed visible phenotypes in the presence of genotoxic agents (Figure R29), by studying genetic instability (Figure R30, Figure R31) or by molecular biology approaches (Figure R33). Therefore, we hypothesized that RECQL5 could be affecting the expression of some important genes to overcome stress, but unless under this situation, the changes were not enough to explain the effects on viability. To that end, cells were cultured in SC media and recovered in mid-log. Whole RNA extractions were performed in three independent RECQL5-expressing transformants and one as control transformed with the empty vector. Total mRNA was determined using the GeneChip Yeast Genome 2.0 Array (Affimetrix). Control values were calculated averaging the data from our empty vector transformed cells with two other similar wild-type analyses from GEO accession GSE22644 (Tu et al., 2011) (Pearson’s R correlation coefficient of 0.914). Gene expression profiles of the control and RECQL5-expressing cells were quite similar (Pearson’s R correlation coefficient of 0.972), suggesting that RECQL5 presence was not significantly impacting global expression levels in S. cerevisiae. To remove false positives, out of 5716 genes analysed in the microarray, we excluded as non-expressed those with values below the median of meiotic genes of the control experiments. Under this criteria, only 460 genes presented mRNA levels altered at least 1.5-fold respect to the control, and just 152 genes were Results II 99 selected for analysis after passing the FDR (False Discovery Rate) p-value<0.1. We found that the 58 genes that were up-regulated represented a group of significantly long genes (around 1kb longer than the genome median) and that their G+C content was slightly but significantly lower than the median value of the genome. Although up-regulated genes were overall highly expressed, this value had no statistical significance. On the other hand, the 98 down-regulated genes in RECQL5-expressing cells had very low transcription levels and were also characterized by a higher content in G+C respect to the genome median value, and slightly shorter in length than the genome median (Figure R35). Gene Ontology (GO) analyses of up-regulated transcripts, performed using the yeastmine.org service, evidenced an enrichment just in ribosome biogenesis related genes (18 out of 58 genes, P-value 0.02003, GO:0042254). On the other hand, down-regulated transcripts appeared to be enriched just in metabolic genes, particularly related to the lysine biosynthesis pathway (11 out of 98, p-value 3.5251e-15, GO:009085). Altogether, this analysis did not allow us to establish a direct association between the expression of RECQL5 in S. cerevisiae and a severe interference in the transcription process. The main reason behind this conclusion is the short number 0 500 1000 1500 2000 2500 DownUp 0 10 20 30 40 50 DownUp Length (bp) G+C content (%) Expression values 0.0 0.2 0.4 1000 1400 1800 2000 DownUp 1938 988.5 38.4 39.7 42.3 1071 1800 0.36 1420 **** **** * * F inal W TvsRQ Le ng th U P D OW NG en om e 0 1 00 0 2 00 0 3 00 0 F inal W TvsRQ Le ng th U P D OW N 0 1 00 0 2 00 0 3 00 0 F inal W TvsRQ GC U P D OW NG en om e 0 1 0 2 0 3 0 4 0 5 0 F inal W TvsRQ GC U P D OW N 0 1 0 2 0 3 0 4 0 5 0 F inal W TvsRQ Ex pr es si on 5 1 0 1 00 0 1 50 0 2 00 0 F inal W TvsRQ Ex pr es si on 5 1 0 1 00 0 1 50 0 2 00 0 Figure R35. Structural and functional features analysis of deregulated genes in RECQL5 expressing yeast. Statistical analysis of length, G+C content and model-based expression values of genes whose expression levels are significantly altered by RECQL5 expression in yeast. Median values are shown and line represents the genome median. *, p<0.05; **, p<0.01; ***, p<0.001. (Mann-Whitney’s U-test) Ph.D. Thesis - Juan F. Lafuente Barquero 100 of genes whose expression was significantly altered. However, by carefully analysing the lists of hits, we were able to find interesting genes with modified mRNA levels in RECQL5-expressing cells such as the ones compiled in Table R5. The fold change in none of the cases is enough to explain by itself the instability phenotypes, but could be indicative that RECQL5 affects transcription of a variety of genes, either as a response for possible damages that the helicase could be generating, or by directly impacting on their transcription. Table R5. Genes altered in RECQL5-exrpressing cells Up-Regulated genes Linear Fold Pvalue FDR ORF Alias Function 1.99 0.022 YOR346W REV1 Deoxycytidyl transferase; involved in repair of abasic sites and adducted guanines in damaged DNA by translesion synthesis (TLS); forms a complex with the subunits of DNA polymerase zeta, Rev3p and Rev7p; 1.62 0.061 YLR430W SEN1 Presumed helicase and subunit of the Nrd1 complex (Nrd1p-Nab3p-Sen1p); complex interacts with the exosome to mediate 3' end formation of some mRNAs, snRNAs, snoRNAs, and CUTs; homolog of Senataxin which causes Ataxia- Oculomotor Apraxia 2 and a dominant form of amyotrophic lateral sclerosis 1.6 0.054 YKR086W PRP16 DEAH-box RNA helicase involved in second catalytic step of splicing; exhibits ATP-dependent RNA unwinding activity; mediates the release of Yju2p and Cwc25p in the second step; in the absence of ATP, stabilizes the binding of Cwc25p to the spliceosome in the first catalytic step 1.56 0.047 YJL092W SRS2 DNA helicase and DNA-dependent ATPase involved in DNA repair and checkpoint recovery, needed for proper timing of commitment to meiotic recombination and transition from Meiosis I to II; blocks trinucleotide repeat expansion; affects genome stability 1.56 0.048 YDL031W DBP10 Putative ATP-dependent RNA helicase of the DEAD-box protein family, constituent of 66S preribosomal particles; essential protein involved in ribosome biogenesis Results II 101 4.4.1. RECQL5 expression does not interfere with RNAPII mutations As another approach to analyse the relationship of RECQL5 with the transcription of S. cerevisiae, we decided to study the possible interactions in the case that the RNAPII holoenzyme was mutated. Two mutants of the major subunit Rbp1p, rpb1-1 and rpb1S751F; another point mutant in the second subunit, rpb2-10 and a complete deletion of one of the 12 subunits, the rpb9∆ mutant (Felipe-Abrio et al., 2014). Because these four mutants are mainly characterized for presenting a variety of problems in the transcription elongation, we wondered if the effect of RNAPII stabilization observed in RECQL5-expressing cells as the accumulation of read-through transcripts (Figure R33) would help these strains to overcome their phenotypes. To answer this, we transformed these mutants with plasmids carrying or not the GAL::RECQL5 construction and plated serial dilutions of exponentially growing cultures on induction plates with galactose. Down-Regulated genes Linear Fold Pvalue FDR ORF Alias Function 3.42 0.0028 YML116W ATR1 Multidrug efflux pump of the major facilitator superfamily, required for resistance to aminotriazole and 4-nitroquinoline-N-oxide 1.91 0.029 YER139C RTR1 CTD phosphatase; dephosphorylates S5-P in the C-terminal domain of Rpo21p; has a cysteine-rich motif required for function and conserved in eukaryotes; shuttles between the nucleus and cytoplasm 1.64 0.087 YOL086W-A MHF1 Protein of unknown function; ortholog of human MHF1, and a component of the heterotetrameric MHF histonefold complex that in humans interacts with both DNA and Mph1p ortholog FANCM, a Fanconi anemia complementation group protein, to stabilize and remodel blocked replication forks and repair damaged DNA; mhf1 srs2 double mutants are MMS hypersensitive 1.62 0.0543 YOR230W WTM1 Transcriptional modulator involved in regulation of meiosis, silencing, and expression of RNR genes; required for nuclear localization of the ribonucleotide reductase small subunit Rnr2p and Rnr4p; contains WD repeats 1.56 0.0419 YBR095C RXT2 Subunit of the histone deacetylase Rpd3L complex; possibly involved in cell fusion and invasive growth Ph.D. Thesis - Juan F. Lafuente Barquero 102 After four days of incubation at 30ºC, the viability assay shown in Figure R36 revealed some moderate differences in the mutants response to the to the overexpression of RECQL5 compared to the wild-type strain. The heat sensitive rpb1-1 mutant showed a behaviour similar to that of the wild-type strain, revealing only a minor difference respect to the induction plate containing glucose, in which mutant cells expressing RECQL5 seemed less affected than wild-type cells. The mutant rpb2-10, either transformed with the empty plasmid or with the one expressing the RECQL5 helicase, grew similarly bad in all conditions, possibly as consequence of a poor transcription of the RECQL5 gene, this could also apply to the rpb9∆ mutant, which did not reflect any difference viability impact after RECQL5 expression. Notably, a slight improvement in the viability of the mutants rpb1S751F and rpb9∆ carrying the RECQL5 gene was observed respect to their control, maybe explained by a low expression of RECQL5 in glucose. rpb1-1 WT - + - + - + - + Glucose Galactose + 0.05% Glu Galactose - + rpb1S751F rpb2-10 rpb9Δ RecQL5 RECQL5 expression level Figure R36. Effect of RECQL5 expression in RNAPII mutants. Viability assay with WT (W303.1A), rpb1-1 (WRP1-12A), rpb1S751F (WSR8-5A), rpb2-10 (WRP2) and rpb9Δ (WRP9-3C) strains transformed with expression vector pYES2:RECQL5.10-fold serial dilutions of exponentially growing cultures plated in selective media with indicated concentrations of carbon source to regulate the expression of the protein. Photographs were taken after 4 days of growth at 30°C. Results II 103 4.5. Impact of RECQL5 helicase on cell cycle progression We know that transcription is one of the major sources of recombinogenic DNA damage (Gaillard & Aguilera, 2016). This threat is mostly derived from the encounter of transcriptional and replication machineries that usually lead to impairment of the latter process. The proposed role of the human helicase RECQL5 was to regulate transcription in the event of collisions with the replication machinery (Popuri et al., 2013; Saponaro et al., 2014). In our model, RECQL5- expressing yeasts showed elevated levels of genomic instability but subtle defects in the transcription, although we demonstrated a positive interaction with the transcription machinery. Therefore, we wondered whether RECQL5-associated impact in the S. cerevisiae genomic stability could be more related to replication fork stalling instead of transcription-associated defects. For this, We transformed Asyn. -2h +/- HU 0 + factor G1 -30 min 20 40 60 80 100 120 160 180 min factor Asyn. 0 20 40 60 80 100 120 140 160 180 Time after release (min) tetp:RECQL5 1C 2C Empty 1C 2C SC 1C 2C tetp:RECQL5 2C Empty 1C SC+HU (40 mM) Figure R37. RECQL5 expression does not lead to replication impairment. FACS analysis of S-phase progression of wild-type strain (W303.1A) transformed with indicated plasmids based on pCM189 backbone. Cells were synchronized in G1 with α-factor and released in presence (right) or absence (left) of 40 mM hydroxyurea by changing the culture medium. A diagram of the experiment is shown at the top. Ph.D. Thesis - Juan F. Lafuente Barquero 104 the wild-type strain W303.1A with the tet:RECQL5 construct or its empty backbone plasmid and monitored their S-phase progression in the presence or absence of HU by FACS (Figure R37). After α-factor synchronization, cells were released from G1 into fresh medium supplemented or not with 40mM of HU, in order to slow down replication progression and also, to slightly stress the cells in order to manage a clearer RECQL5-associated phenotype. As show in the result, we found that cells expressing RECQL5 did not present any delay in S-phase progression neither in the absence or the presence of the assayed concentration of HU. The only observable difference was at 180 minutes after α-factor release, when cells were under replication stress (+HU condition). At this point RECQL5-expressing yeasts showed a slight arrest in G2-phase compared to the strain transformed with the empty plasmid. Therefore, it seems that the instability phenotypes associated to RECQL5 expression in yeasts were not accompanied by a strong replication impairment. 4.6. Genome-wide distribution of RECQL5 through the S. cerevisiae genome To explore possible mechanisms of action for RECQL5 on S. cerevisiae, we performed ChIP-chip analysis using the RECQL5-FLAG tagged protein. The data were analysed and compared to Rpb3-HA and Hpr1-FLAG ChIP-chip results previously published (Gómez-González et al., 2011) (Figure R38). Statistical analysis of the IP signal allowed to identify regions or clusters where RECQL5 recruitment was statistically significant (genome regions where enrichment signal difference over the input had a p-value lower than 0.05). These positive intervals were distributed all over S. cerevisiae’s genome. Overall, the normalized signal values obtained for RECQL5 recruitment were slightly lower to those acquired for Rpb3 and Hpr1. Annotation of RECQL5 enriched regions (SGD_features.tab for sacCer3 genome version) showed that half (51.8%) of the yeast features where RECQL5 could be found were ORFs. This number was slightly below compared to that of yeast proteins directly associated with transcription like Rpb3 or Hpr1 (66.9% and 65.6% of total hits mapping to Results II 105 A Chromosome IV 2 0 -2 2 0 -2 2 0 -2 RECQL5-IP Rpb3-IP Hpr1-IP Signal log ratio 2 2 0 -2 2 0 -2 2 0 -2 Signal log ratio 2 RECQL5-IP Rpb3-IP Hpr1-IP GCN4 B YRA1 EFT2 UTP5 2 0 -2 2 0 -2 2 0 -2 Signal log ratio 2 RECQL5-IP Rpb3-IP Hpr1-IP C Figure R38. RECQL5 recruitment to the genome. Genomic views of RECQL5 (blue), Rpb3 (orange) and Hpr1 (yellow) recruitment. (A) Full view of chromosome IV. (B) GCN4 ORF and adjacent 10 kb of the surrounding region. (C) YRA1 locus and surrounding 30 kb are shown. Signal log2 ratio values are plotted. SGD ORFs are represented below as blue bars and white arrows according to the direction of transcription. Ph.D. Thesis - Juan F. Lafuente Barquero 112 RECQL5-expressing cells grew worse than the empty plasmid control in the galactose plates, and effect being exacerbated in the presence of HU. Interestingly, overexpression of SGS1 and BLM also impacted the growth capability of the yeast, although to a lower extent. We confirmed that the SGS1 expression plasmid was functional since it complemented the sensitivity conferred by sgs1∆ in the presence of HU. The effect of complementation was clear in the glucose plates with HU, where GALp was repressed or activated at very low levels. This observation is likely due to promoter leakage, so that low levels of expression in this condition were enough for complementation. In addition, we confirmed that SGS1 is a multicopy suppressor of srs2∆ mutant (Mankouri et al., 2002). Cells transformed with the GALp:SGS1 plasmid grew better than those with the empty plasmid in galactose plates. Interestingly, we observed a similar effect to that described previously with the sgs1∆ mutant in glucose plates. Again, low transcription of RECQL5 allowed the complementation in the srs2∆ mutant. However, overexpression of RECQL5 in the plates with galactose was highly deleterious, as expected. This new result BY4741 sgs1∆ srs2∆ Glucose 2% Galactose 2% Galactose 2% Glucose 0.05% Vector GAL::RecQL5 GAL::RecQL5 ID GAL::SGS1 GAL;;BLM GAL::WRN 100 mM HU -- -- --100 mM HU 150 mM HU Vector GAL::RecQL5 GAL::RecQL5 ID GAL::SGS1 GAL;;BLM GAL::WRN Vector GAL::RecQL5 GAL::RecQL5 ID GAL::SGS1 GAL;;BLM GAL::WRN Figure R42. RECQL5 complementation of yeast helicase mutants. RECQL5 overexpression in sgs1∆ (YMR190C) and srs2∆ (YJL092C). Ten-fold serial dilutions of WT (BY4741) and mutant strains transformed with plasmids carrying indicated human and yeast recq helicases, under the control of a GAL promoter, were plated on minimal selective medium with either glucose, galactose or a combination of both in the presence or not of hydroxyurea (HU). Photographs were taken after 3 days of growth at 30ºC Results II 113 broadened our view of the possible actions of RECQL5 in yeast. Next, taking into account that overexpression of SRS2 impairs cell growth (León Ortiz et al., 2011), we wondered whether although SRS2 is not a member of the RecQ family of helicases, it acted as an orthologue of the human RECQL5. Since high levels of expression from both these proteins seemed to affect the viability of the cells, we decided to test if they could affect the cell cycle progression in a similar way. For this, we transformed the wild-type strain W303.1A with the GALp expression vectors carrying genes RECQL5, SRS2 or none. We monitored the S-phase progression of the transformed yeast after overexpressing the indicated genes by FACS analysis (Figure R43). Cells were grown to exponential phase in synthetic media with raffinose, this sugar does not repress GALp as glucose does, and can be maintained in the media after induction with galactose to be used as an alternative source of carbon. After a first round of α-factor synchronization, galactose was added to the media to a final concentration of 2% and cells were incubated for Vector GAL:RECQL5 GAL:SRS2 240 Time (min) 0 180 150 120 90 60 40 20 Asyn. 1C 2C 1C 2C1C 2C S-Raffinose Asyn. -1h 0 + factor G1 -1h 20 40 60 90 120 150 180 240 min factor +GAL 2% Figure R43. Helicases overexpression impact on cell cycle. S-phase progression of wild-type strain (W303.1A) transformed with respective pYES2 vectors. Cells were synchronized in G1 with α-factor and released by changing the culture medium. A diagram of the experiment is shown at the top. Ph.D. Thesis - Juan F. Lafuente Barquero 114 an additional hour. After 2 hours of synchronization, cells were released from G1 by shifting to fresh media with galactose and raffinose and samples were taken every 20 minutes. FACS analysis of the samples revealed that both RECQL5 and SRS2 expressing cells presented a clear delay in entering S-phase, more acute in the SRS2 expressing ones. This indicates the presence of problems in the cell cycle progression, probably derived from impairment of replication. Further analyses are needed to understand the replication failure of cells expressing RECQL5 or SRS2. Our results support that there may exist shared functions between these proteins, and working with both RECQL5 and SRS2 as models could help to understand each other’s functions. Discusion Discusion 117 This thesis is focused on the study of transcription-replication collisions, natural events that occur inside the nucleus. This type of events is one of the main sources of replicative stress and increased genomic instability, which in turn are hallmarks of cancer and some human degenerative diseases (Gaillard et al., 2015). Specifically, we study R-loops and their role in the generation of genomic instability associated with transcriptional and replicative stress. We have studied different aspects such as where they are formed, how are they formed and which are the factors implicated in the regulation of their presence. 1. Genome wide study of RNA:DNA hybrids We studied the genome-wide distribution of RNA:DNA hybrids in budding yeast using DRIP, with the S9.6 antibody, followed by sequencing of immunopurified DNA fragments (DRIP-seq). Our results are related to data from similar previously published studies, which we use as reference for our analysis. We performed DRIP-seq on wild-type cells and a hpr1∆ mutant strain, and started analyzing our sequencing results by comparing them side by side with those from previous similar publications (Chan et al., 2014; El Hage et al., 2014; Wahba et al., 2016). We found that, although in our hands the protocol is still open to major optimization, the already available data about R-loop localization must be interpreted with discretion. The three works used as reference were performed with different methodologies, and the general conclusions dragged from these studies are essentially the same. All three studies propose that rDNA, transposons and telomeres constitute rich RNA:DNA hybrid-prone loci along the budding yeast’s genome. Additionally, they mostly agree about the ORFs prone of RNA:DNA hybrid formation to be highly transcribed and GC rich. However, Chan et al. (2014) suggested that around one third of the yeast genome is prone to hybrid formation, while the other two studies limit their incidence to a much more reduced number of regions, covering only 8% of the genome according to Wahba et al. (2016). This brings to light that the differences in the methodology may influence the general outcome of the analysis and limit the possibility of extracting specific conclusions Ph.D. Thesis - Juan F. Lafuente Barquero 118 and mechanistic hints that could potentially help understand R-loop physiological functions. We were able to detect R-loop enrichment in a wild-type strain at rDNA locus and telomeres, which are regions that had been identified initially by qPCR- based studies (El Hage et al., 2010; Balk et al., 2013), as well as in transposons and a subset of RNAPII-transcribed genes (Table R1). The analysis resulted in a low number of positive regions compared to previous works, but showed a high correlation with the values from the most recent study (Wahba et al., 2016), which was more astringent in their analysis thanks to revisiting the methodology from the precedent studies. We put special emphasis in the analysis of the open reading frames identified as hybrid-prone regions, since the actual literature widely supports the formation of R-loops as co-transcriptional by-products (further reviewed in Aguilera & García-Muse, 2012; Santos-Pereira & Aguilera, 2015). We took all the positive ORFs identified in each of the analysis and generated a list with 235 genes that showed DNA:RNA hybrid enrichment in all 4 datasets. These genes had in common to be long, GC-rich and highly expressed (Figure R3), which is consistent with previously reported features for genes prone of R-loop formation. 1.1. Accumulation RNA:DNA hybrids towards the 3’ end of the gene We also analized the signal distribution along all RNA:DNA hybrid-positive genes in our wild-type strain, compared to the data available for the wild-type used by El Hage et al., (2014). The profiles show a signal distribution that spreads along the length of the transcription unit of genes, with a slight enrichment towards the 3’ end (Figure R4). This is consistent with the generally accepted cotranscriptional nature of the R-loops, for which it has been speculated a function in the deceleration of the RNAPII to facilitate the co-transtricptional splicing process and the correct termination and poly-adenilation of the nascent mRNA (Wahba et al., 2016; Proudfoot, 2016). However, there are other interpretations that can not be ruled out yet with the data available. These include the accumulation of negative supercoiling behind the transcribing RNA polymerase, which could potentially favor the formation of an R-loop due to a higher probability of ssDNA able to Discusion 119 associate with the nascent mRNA as it has been observed in yeast top1 mutants (El Hage et al., 2010), or that failures in the formation of a functional mRNP could leave the RNA molecule unprotected increasing its chances to hybridize with the template DNA as observed in THO mutants (Gómez-González et al., 2011; Aguilera & García-Muse, 2012). 1.2. RNA:DNA hybrids in hpr1∆ We performed DRIP-seq on the hpr1∆ mutant. Cells lacking Hpr1 are defective in the formation of the THO complex, and consequently in the biogenesis of mRNPs (Aguilera, 2005), which leads to genomic instability associated with an elevated presence of R-loops (Huertas & Aguilera, 2003; Aguilera & García-Muse, 2012). Contrary to what could be initially expected, our results show little differences in the overall number of R-loop-prone regions compared to the wild-type (Table R2). This suggests the possibility that a higher or lower accumulation of R-loops is not critical in the origin of genomic instability, but likely the R-loop nature. Indeed other determinants could be their abnormal stabilization or other processes associated with R-loops, as for example chromatin changes as suggested in the study by Castellano-Pozo et al., (2013). According to our results, R-loop-accumulating ORFs are almost equally abundant in the wild-type as in the hpr1∆ mutant (1276 in wildtype vs 1286 in hpr1∆ strain). Interestingly, only one third of them are common to both strains. More biological replicates are needed to stablish conclusions, the results could imply that RNAPII-transcribed genes are not equally susceptible to form R-loops in the absence of the THO complex. Previous genome-wide studies (Gómez-González et al., 2011) showed that Hpr1 is preferentially recruited to RNAPII-transcribed genes that are long, C+G rich and highly expressed, and that Hpr1 binding follows a gradient that increases towards the 3’ end of ORFs. Similar features were found for the recruitment in different THO mutants of Rrm3, which is a DNA helicase required for the progression of the replication fork through obstacles in the DNA (Ivessa et al., 2000; Azvolinsky et al., 2009), suggesting that the absence of Hpr1 leads to such obstacles and, consequently, replicative stress (Gómez-González et al., 2011). Ph.D. Thesis - Juan F. Lafuente Barquero 120 Although hpr1∆ presents a lower RNA:DNA hybrid signal than the WT, similar results were previously shown by Chan et al. (2014). A posible explanation could be that DNA:RNA hybrids are more persistent in the hpr1∆ mutant. One critial feature could be R-loop length, so that this would make the hybrid more estable, thus increasing their half-life which in turn would facilitate the collapsing of incoming replication forks. This hypothesis would be consistent with the trasncription defects and the increase of Rrm3 clusters genome-wide associated to the absence of HPR1 (Gómez-González et al., 2011). Nevertheless, other unexplored possibilities may explain this phenomenon, such as the differences produced in chromatin structure. It will be interesting to compare results from studies on RRM3, histone marks and other chromatin associated factors with more definitive results obtained from DRIP-seq mapping of R-loops in the hpr1∆ strain, in order to stablish possible correlations with R-loop-prone regions over the genome. Such studies should open new possibilities to discover correlations woth the presence of different proteins at different sites over the genome. 2. Prevalence of R-loops in trans Our analyses support the idea that RNA:DNA hybrid presence is more common than previously foreseen before the application of genome-wide methodologies to the field. Most R-loop-prone sites are associated with regions in the genome were transcription is active (i.e. ORFs, tRNAs, rDNA and telomeres) (Chan et al., 2014; El Hage et al., 2014; Wahba et al., 2016; Ginno et al., 2012 and this study), supporting the hypothesis that most R-loops if not all are formed cotranscriptionally. However, recent studies have raised the possibility that R-loops can generate by other mechanisms non associated with transcription in cis (Wahba et al., 2013; Keskin et al., 2014). R-loops would form in trans when a RNA molecule anneals with a homologous DNA strand different from the locus where it was originally synthesized. We sought to assay whether the formation of these RNA:DNA hybrids form abundantly in the cell and if so, whether they have a measurable impact on genome instability. Henceforth our study stablishes a system that could Discusion 121 help us understand possible mechanisms underlying the generation of R-loops. To assay the formation of R-loops in trans we stablished an indirect system relying on the ability of R-loops to increase instability. The assay was based on the co-transformation of yeast cells with two plasmids. One plasmid carried the recombination system GL-LacZ which contains the whole LacZ ORF between the DNA repeats (Piruat et al., 1997) and the other one would carry different sequences of the LacZ gene expressed from a tet promoter, depending on the experiment performed. We used this approach in a wild-type strain and in R-loop accumulating mutants, including two mutants of the THO complex (mft1∆ and hpr1∆) and the double mutant rnh1∆ rnh201∆. The THO complex, conserved through eukaryotes, is responsible of proper mRNP formation and mRNA processing and export (Luna et al., 2012). Absence of different components of this complex leads to variable degrees of R-loop accumulation, defects in transcription elongation and increased levels of genomic instability (Huertas & Aguilera, 2003; García-Rubio et al., 2008), mostly associated with the nascent mRNA being unprotected and failures in its posttranscriptional processing. On the other hand, Rnh1 and Rnh201 are responsible for the degradation of most RNA:DNA hybrids that are generated in the cell (Cerritelli & Crouch, 2009). Therefore, their inactivation leads to a widespread signal of S9.6 antibody detecting R-loops throughout the nucleus as seen by cytological studies (Wahba et al., 2011; Chan et al., 2014). However, this effect does not impair growth and only reflects in minor genomic instability phenotypes (Huertas & Aguilera, 2003; Wahba et al., 2011). In general terms, our results did not show a measurable impact of the mRNA in trans on the genome stability of the wild-type strain in any of the conditions assayed (Figure R10 to Figure R14), suggesting that R-loops do not form easy in trans in the wild-type strain. We tested the capacity of different transcripts in trans to induce recombination in the LacZ gene (full length LacZ, inverted LacZ and 400bp fragment from the LacZ 3’ end). None of the mutants showed a consistent significant increase in recombination when the only source of LacZ mRNA was in trans (cis transcription OFF, trans transcription ON), further supporting the hypothesis that R-loops in trans are not frequent even under conditions that could Ph.D. Thesis - Juan F. Lafuente Barquero 128 rDNA, because converging replisomes do not exist to complete replication. In this regard, it is interesting that the SMC5/6 complex, the negative regulator of Mph1, has been proposed to play a particularly important role at sites of unidirectional replication (Murray and Carr, 2008). Additionally, Mph1 has been shown to interact with RPA and Mte1. RPA specifically binds to ssDNA that has been unwound at the fork and Mte1 binds to branched DNA structures and recruits Mph1 in S/G2 phase (Silva et al., 2016). Thus, we hypothesize that if Mph1 were to unwind RNA-DNA hybrids in a more general manner, we would expect Mph1 to also be essential in sen1-1 and THO-TREX mutants, but we did not see such genetic interactions. Together, here we identified Mph1 as a factor preventing DNA damage at RNA:DNA hybrids. We analyzed the effects of Mph1 absence in RNA:DNA hybrid turnover in more detail and found that Mph1 not only forms foci when RNA-DNA hybrids accumulate but it may also be important in the prevention of the replication fork stalling due to the presence of R-loops. 4. Understanding the role of RECQL5 at the interface between transcription and replication RECQL5 is one of the five members present in humans from the RecQ family of helicases. These proteins are evolutionarily well conserved and have critical roles in the maintenance of the genome integrity (Chu & Hickson, 2009; Croteau et al., 2014). RECQL5 is the only one to show direct interaction with RNAPII (Aygün et al., 2008; Islam et al., 2010) and it was proposed to have a regulatory function at the interface between transcription and replication trough the inhibition of transcription (Saponaro et al., 2014). To date, RECQL5 role in transcription hasn’t been linked to R-loop regulation in a straightforward way, but recent studies point to the direction that its presence is required for the coordination of different transcription factors, like TOP1, whose function is alleviating negative supercoiling behind the transcription machinery, contributing to the suppression of genome instability caused by R-loops (Li et al., 2015; García-Muse & Aguilera, 2016). In this Discusion 129 thesis we worked with S. cerevisiae as a model organism to further advance in the understanding of the biological role of RECQL5 as genome caretaker. We showed for the first time that the expression of the human RECQL5 or its mutant version, RECQL5 ID, in yeast cells generated a growth defect that was equivalent for both proteins (Figure R27). Both versions of the human helicase were found to interact with the S. cerevisiae RNAPII (Figure R28), despite being described by Aygün et al., (2009) that the KIX domain (mutated in RECQL5 ID) was essential for the interaction of the helicase with the human RNAPII in vitro. However, the posterior work by Islam et al., (2010) demonstrated the existence of a second RNAPII-interaction domain in the sequence of RECQL5, the SIR domain. The KIX domain is present in several other RNAPII-transcription factors (Parker et al., 1996), and the SRI domain takes its name from the Set2 Rpb1 interaction domain, found in the histone methyltransferase SetD2 which also regulates transcription (Kizer et al., 2005). The RNA polymerase II undergoes different phosphorylation states at its CTD (C-terminal domain) tail during transcription. Initiation state is associated with the absence of phosphorylation (RNAPIIa) while the transcription machinery is being fully assembled. Subsequent phosphorylations of the CTD mean the beginning of the elongation phase of the transcription, in which the RNAPII stays hyper-phosphorylated (RNAPIIo) until reaching the transcription termination step. RECQL5 associates to both RNAPIIa and RNAPIIo thanks to the KIX domain, while the SRI domain is only able to interact with the phosphorylated form of the polymerase (RNAPIIo). Islam et al. (2010) demonstrated that these associations are maintained between the human RECQL5 and the chicken RNAPII through their experiments with DT40 cells. Since RNAPII is known to be one of the most highly conserved proteins in all eukaryotes, we predicted that the human helicase was affectively interacting with the S. cerevisiae polymerase. We found that expressing RECQL5 in yeast generated high levels of genomic instability, measured by increases in recombination frequency (Figure R31), in the accumulation of Rad52 foci (Figure R30) and sensitivity to genotoxic agents (Figure R29). These phenotypes were dependent on dosage as a lower impact was produced when RECQL5 was expressed from a centromeric plasmid and placed Ph.D. Thesis - Juan F. Lafuente Barquero 130 under control of a tetp promoter. This was confirmed in the search for yeast mutants able to suppress the sensitivity to hidroxiurea when expressing RECQL5 that we performed, in which all initial positive candidates demonstrated to be impairing RECQL5 expression (Figure R41). The hyperrecombination phenotype is characteristic of yeast mutants for RNA metabolism factors like hpr1∆ (Prado et al., 1997), yra1∆ (Gavaldá et al., 2016) or npl3∆ (Santos-Pereira et al., 2013). This increase is largely derived from failures in the transcriptional process that lead to replication stress and DNA damage (Gaillard & Aguilera, 2016). In the case of RECQL5-expressing cells, we could not confirm the observed instability to be associated to the transcriptional state of the template (Figure R31). Moreover, we could not detect any impact on transcription itself, as cells expressing RECQL5 were able to properly transcribe the LacZ –URA construct (Figure R32), which was previously used as a tool to link transcription defects and genomic instability (Jimeno et al., 2002). In spite of this, by looking at transcription at the molecular level we found that expression of RECQL5 generated a visible defect on termination, with the accumulation of mRNA read-through at the GAL10 locus (Figure R33). We identified this to be the reason for the growth defect observed in cells expressing RECQL5 from a GALp, but not for the sensitivity to genotoxic agents (Figure R34). It is interesting though, that the physiological role in transcription of RECQL5 in mammalian cells has been described to be as transcription inhibitor (Saponaro et al., 2014), and the effect that we were observing in yeast cells closely resembles a termination bypass. We found that RECQL5 expressed in RNAPII mutants, rpb1S751F and rpb9∆, showed a subtle complementation with the leaky expression from the GALp in glucose. These mutants are specifically affected at transcription elongation, due to a diminished capability to avoid or repair ribonucleotide miss-incorporations into the nascent RNA (Strathern et al., 2013; Koyama et al., 2010). The phenotypes of these mutants have been shown to be at least partially suppressed by the overexpression of the transcription elongation factor TFIIS (DST1), which is recruited to the elongating RNAPII via the interaction with Rpb9 subunit and stimulates the backtracking and editing activity of Rpb1, which is affected in the rpb1S751F mutant (Koyama et al., Discusion 131 2007; Strathern et al., 2013). Recent in-vitro studies (Kassube et al., 2013) proposed that RECQL5 controls transcription regulation through structural mimicry with the TFIIS factor, blocking RNAPII by competing with TFIIS. Our results would partially support this similarity, but more experiments are needed to determine the levels of RECQL5 leaky expression to explain why RECQL5 seems to positively affect transcription elongation in S. cerevisiae, contrary to experiments performed invitro or with mammalian cell cultures (Aygün et al., 2009; Saponaro et al., 2014). To this point, an arising possibility is that RECQL5 could interfere with other yeast transcription factors that are recruited after elongation started, leading to failures in the proper termination process. Specific analyses are required to asses if the presence of RECQL5 was affecting the transcription rate of the yeast RNA polymerase, but to asses if the transcription defects are general and could affect mRNA production of other yeast genes we analyzed the expression profile of yeasts expressing RECQL5. Few genes were significantly affected, and most of them were just 1.5 to 2 fold compared to the control. We consider that these changes were not sufficient to explain the instability phenotypes observed. Analysis of genomewide recruitment of RECQL5 to chromatin in yeast cells supported the interaction with the RNAPII, as many of the hits corresponded with actively transcribed genes (52% of all positive regions). Interestingly, no hits mapped to the rDNA locus, confirming the interaction specificity with the RNAPII and not with RNAPI. Detailed examination of RECQL5 distribution along the positive ORFs revealed a pattern characterized by a low occupancy at the promoter region and the beginning of the ORF, followed by an increasing accumulation toward the 3’ end of the gene. This is consistent with previous analyses in human cells, where the interaction of RECQL5 is shown to be dependent on the phosphorylation of the CTD tail of an elongating RNAPII (Kanagaraj et al., 2010). In physiological conditions, RECQL5 is supposed to associate to RNAPII complex 1:1 (Aygün et al., 2008), such high accumulation at gene termination regions as observed in (Figure R39) could be explained by the accumulation of subsequently blocked RNA polymerases. However, this would dramatically impact on the viability of the cell caused by a great replication impairment and it is not the case. Accordingly, we were not able to find by FACS Ph.D. Thesis - Juan F. Lafuente Barquero 132 analysis a cell-cycle delay that could support the idea of RECQL5 causing a defect on replication progression. It would be of great interest though, to perform Rrm3- FLAG ChIP-chip analysis like it was done before to study the global defects of THO and TREX mutants in replication progression (Gómez-González et al., 2011; Santos-Pereira et al., 2014). Rrm3 is a helicase required for the progression of the replication fork through obstacles in the DNA and its accumulation at specific DNA sites has been used to identify replication fork pauses or stalls (Ivessa et al., 2000; Azvolinsky et al., 2009). Simultaneous analyses of Rrm3 accumulation or Rpb3 binding while RECQL5 is expressed could add relevant information to understand this phenomenon. On the other hand, RECQL5 was enriched in nine centromeres and was also present at telomeres. This could be indicative of RECQL5 interacting with other yeast proteins that lead to its recruitment to other regions of the genome different from RNAPII transcribed genes. It is known that RecQ helicases have roles in multiple DNA processes. For RECQL5, transcription is one of them due to its association with the RNAPII, but looking at other known interactors, like Rad51 (Hu et al., 2007), Topoisomerase 3α and 3b (Topo3α and Topo3b) (Shimamoto et al., 2000) proliferating cell agent (PCNA) (Kanagaraj et al., 2006), Mre11-Rad50-Nbs1 (MRN) complex (Zheng et al., 2009) it is evident that RECQL5 also participates in repair pathways as the other members of the RecQ family. Since we confirmed the interaction between the human RECQL5 and the yeast RNAPII, other yeast proteins homologous to human factors known to associate with RECQL5 could conserve this interaction in the model organism. We found that the human RECQL5 have the capability of suppressing the formation of Rad51 foci in yeast cells (Figure R30). Rad51 is a protein required in initial steps in all homologous recombination reactions. HR is an important mechanism for the maintenance of genome integrity, which functions to repair double-strand breaks (DSBs) and single-strand gaps formed during replication or created by DNA damaging agents or from processing DNA lesions. In addition, HR is implicated in the restart of damaged replication forks and functions in the telomere length maintenance in cells lacking telomerase. This result suggested that RECQL5-dependent instability phenotypes could be associated to an Discusion 133 interference in the DNA damage repair pathway. Nakayama et al. (2004) did a complementation work in S. cerevisiae with the RECQL5/QE from D. melanogaster. They concluded that the fly’s helicase was able to complement some functions of the yeast unique RecQ helicase, Sgs1. However, this complementation was not observed in our yeast system. Through the years it is more accepted that the likely human homolog of Sgs1 is the RecQ helicase BLM, although it is known that BLM and RECQL5 share some functionalities in mammal cells (Hu et al., 2005; Lu et al., 2011). Interestingly, we did observe that human RECQL5 could partially complement srs2∆ yeast mutants (Figure R42). A potent and well-studied mechanism to control recombination in yeast involves the Srs2 protein (Sung & Klein, 2006). Srs2 is an helicase with similarities to the bacterial UvrD/Rep helicase (Rong & Klein, 1993), which functions to restrict recombination by its ability to bind Rad51 (Krejci et al., 2003; Le Breton et al., 2008). Double mutants sgs1∆ srs2∆ are extremely sick (Gangloff et al., 2000), and overexpression of Sgs1 is able to partially rescue srs2∆ sensitivity to drugs but Srs2 does not complement sgs1∆ (Mankouri et al., 2002). Moreover, Srs2 overexpression is known to impair cell growth (Krejci et al., 2003; Mankouri et al., 2002) but not by means of dysregulation of HR but because of replication fork impairment through its interaction with PCNA (León Ortiz et al., 2011), which is also an identified partner of RECQL5. More experiments need to be performed in order to understand the effect of RECQL5 expression in cells. We have demonstrated that it is able to interact with the yeast RNAPII, that it does not interfere with transcription to a degree that explains the genomic instability phenotypes observed, but it does significantly affect the formation of Rad51 filaments. In agreement, it is able to complement for Srs2 deficiency at very low concentrations, establishing a possible functional homology. An interesting approach could be to validate more known interactors, like for example PCNA, or find new ones that could uncover new clues to understand the role of RECQL5. Conclusions Conclusions 137 Conclusions 1. The presence of R-loops in yeast wild-type strains as determined by DRIP-seq is widespread along the genome, with special enrichment found at rDNA, telomeres, transposons, tRNAs and specific RNAPII-transcribed ORFs, consistent with other reports. 2. R-loop-prone regions mainly correlate with genomic loci that are GC-rich and highly transcribed, consistent with the idea that transcription is the main source of R-loop formation. 3. Absence of Hpr1 does not lead to a significant increase in the number of R-loop- prone regions. However, these regions do not entirely correlate with the ones found in the wild-type. Still, our data suggest that there are technical issues yet to be resolved to stablish definitive conclusions. 4. We did not find evidence for the formation of R-loops in trans. Anyhow, if that were the case we show that is not a phenomenon with an impact on genomic instability. 5. Deletion of Rad51 in a hpr1∆ strain does not suppress the instability phenotype of the THO mutant, suggesting that there is no role for the recombinase in the formation of R-loops. 6. Absence of Mph1 or its helicase function leads to an increase in R-loop levels inside the nucleus, suggesting a role for this protein in the maintenance of RNA:D- NA hybrid homeostasis. 7. Mph1, which works with the replication fork, could potentially resolve R-loops encountered by the replication machinery, by a mechanism yet to decipher. 8. The human helicase RECQL5 can be expressed in yeast, leading to an accumulation of genomic instability that is not apparently dependent on transcription or on replication defects. 9. RECQL5 does interact with S. cerevisiae RNAPII and Rad51 and partially complements yeast srs2∆ mutants, suggesting that it may partially complement the Srs2 function, and act in yeast independent of transcription.