Analysis of the processing of complex recombination intermediates by structure-selective endonucleases

INTERNATIONAL DOCTORAL SCHOOL OF THE USC Raquel Carreira Rodríguez PhD Thesis Analysis of the processing of complex recombination intermediates by structureselective endonucleases Santiago de Compostela, 2023 Doctoral Programme in Molecular Medicine TESIS DE DOCTORADO ANALYSIS OF THE PROCESSING OF COMPLEX RECOMBINATION INTERMEDIATES BY STRUCTURESELECTIVE ENDONUCLEASES Raquel Carreira Rodríguez ESCUELA DE DOCTORADO INTERNACIONAL DE LA UNIVERSIDAD DE SANTIAGO DE COMPOSTELA PROGRAMA DE DOCTORADO EN MEDICINA MOLECULAR SANTIAGO DE COMPOSTELA 2023 DECLARACIÓN DEL AUTOR/A DE LA TESIS Dña. Raquel Carreira Rodríguez Título de la tesis: Analysis of the processing of complex recombination intermediates by structure-selective endonucleases Presento mi tesis, siguiendo el procedimiento adecuado al Reglamento y declaro que: 1) La tesis abarca los resultados de la elaboración de mi trabajo. 2) De ser el caso, en la tesis se hace referencia a las colaboraciones que tuvo este trabajo. 3) Confirmo que la tesis no incurre en ningún tipo de plagio de otros autores ni de trabajos presentados por mí para la obtención de otros títulos. 4) La tesis es la versión definitiva presentada para su defensa y coincide la versión impresa con la presentada en formato electrónico. Y me comprometo a presentar el Compromiso Documental de Supervisión en el caso que el original no esté depositado en la Escuela. En Santiago de Compostela, 12 de diciembre de 2022. AUTORIZACIÓN DEL DIRECTOR/TUTOR DE LA TESIS D./Dña. Miguel González Blanco En condición de: Tutor/a y director/a Título de la tesis: Analysis of the processing of complex recombination intermediates by structure-selective endonucleases INFORMA: Que la presente tesis, se corresponde con el trabajo realizado por Dña Raquel Carreira Rodríguez, bajo mi dirección/tutorización, y autorizo su presentación, considerando que reúne los requisitos exigidos en el Reglamento de Estudios de Doctorado de la USC, y que como director/tutor de esta no incurre en las causas de abstención establecidas en la Ley 40/2015. En Santiago de Compostela, 12 de diciembre de 2022 CONFLICTO DE INTERÉS Yo, Raquel Carreira Rodríguez, con DNI 33561062-Z, Declaro no tener ningún tipo de conflicto de interés, ni ninguna relación económica, personal, política, interés financiero ni académico que pueda influir en este trabajo. Así mismo, y con la presente, declaro la completa autoría de todas las figuras que se muestran en la tesis En Santiago de Compostela, 12 de diciembre de 2022 3.1.3 Bacterial proteins .................................................. p. 73 3.2 METHODS .................................................................................. P. 76 3.2.1 Yeast protein purification .................................... p. 76 3.2.1.1 Rad51 ......................................................... p. 76 3.2.1.2 RPA ............................................................ p. 77 3.2.1.3 Rad54 ......................................................... p. 78 3.2.1.4 Yen1 and Yen1ND ....................................... p. 79 3.2.1.5 Mus81-Mms4 and Mus81-Mms4ND ........... p. 80 3.2.2 Oligonucleotide purification and annealing into DNA substrates...................................................... p. 81 3.2.3 Plasmid purification............................................. p. 82 3.2.4 Endonuclease assays with oligonucleotide-based substrates ............................................................... p. 82 3.2.5 Endonuclease assays with plasmid-based Dloops ........................................................................ p. 84 3.2.5.1 Rad51/Rad54-made D-loops ...................... p. 84 3.2.5.2 RecA-made D-loops ................................... p. 85 3.2.5.3 Deproteinized D-loops ............................... p. 86 3.2.5.4 D-loops with a stable presynaptic filament p. 87 3.2.6 Mapping SSE cleavage sites on enzymaticallygenerated D-loops.................................................. p. 87 3.2.6.1 Mapping experiments using Sanger sequencing .................................................. p. 88 3.2.7 Detection of half-crossover precursors ............... p. 92 4. RESULTS ........................................................... p. 95 4.1 BOTH MUS81 AND YEN1 CAN CLEAVE OLIGONUCLEOTIDEBASED D-LOOPS ........................................................................... P. 95 4.2 YEN1 INCISIONS ON THE D3 D-LOOP ARE INDEPENDENT .. P. 102 4.3 MUS81 AND YEN1 CAN CLEAVE RAD51/RAD54-MEDIATED D-LOOPS ................................................................................. P. 106 4.4 MUS81 AND YEN1 CLEAVE ENZYMATICALLY-MADE DLOOPS COATED BY RPA ....................................................... P. 114 4.5 D-LOOP PROTEINS IMPAIR NUCLEASE CLEAVAGE AND RAD51 TURNOVER FACILITATES SSE ACTIVITY ................ P. 121 4.6 MUS81 AND YEN1 INCISIONS ON ENZYMATICALLY-MADE D-LOOPS MAP AT SIMILAR POSITIONS TO THOSE ON SYNTHETIC D-LOOPS ............................................................. P. 126 4.7 CONCURRENT CLEAVAGE OF MUS81 AND YEN1 ON A PLASMID-BASED D3 D-LOOP LEADS TO A HALFCROSSOVER PRECURSOR ...................................................... P. 142 5. DISCUSSION ...................................................p. 149 5.1 MUS81 AND YEN1 INCISIONS ON SYNTHETIC D-LOOPS REFLECT THE POLARITY OF THEIR RESPECTIVE NUCLEASE FAMILIES ................................................................................ P. 150 5.2 THE POTENTIAL BIOLOGICAL RELEVANCE OF THE INCISIONS PRODUCED BY MUS81 AND YEN1 ON D-LOOPS P. 152 5.3 THE ACTIVITY OF MUS81 AND YEN1 ON D-LOOPS IS INFLUENCED BY THE PRESENCE OF THE PROTEINS INVOLVED IN THEIR FORMATION ......................................... P. 154 5.4 HALF-CROSSOVER FORMATION THROUGH NUCLEOLYTIC PROCESSING OF D-LOOPS ..................................................... P. 159 6. CONCLUSIONS ..............................................p. 165 7. BIBLIOGRAPHY ............................................p. 167 INDEX OF FIGURES AND TABLES ...............p. 213 ANNEXES .............................................................p. 219 PAPERS CONTRIBUTED DURING THE PHD STAGE .............. P. 221 AWARDS RECEIVED DURING THE PHD STAGE .................... P. 222 ACKNOWLEDGEMENTS ........................................................ P. 223 SUMMARY IN GALICIAN LANGUAGE Estímase que o ADN de cada unha das nosas células sofre entre 10 e 20 roturas de dobre cadea o día. Estas roturas constitúen un dos tipos de lesións máis citotóxicas que existen, xa que, de non repararse axeitadamente, poden derivar na perda de información xenética. Para evitar as desastrosas consecuencias que isto podería supoñer, as células posúen dous mecanismos principais para a reparación das roturas de dobre cadea: a unión de extremos non homólogos, que permite un selado efectivo – aínda que potencialmente mutaxénico – da rotura, e a recombinación homóloga (Figura 1). Este último caracterízase pola utilización dunha molécula de ADN intacta de secuencia idéntica (cromátide irmá) ou case idéntica (cromosoma homólogo) para recuperar a información perdida na rotura. Por iso, adoitase considerar a recombinación homóloga como unha vía de reparación de alta fidelidade. As células somáticas utilizan preferentemente a cromátide irmá, mentres que en meiose, a molécula empregada como molde adoita ser o cromosoma homólogo, o cal pode presentar variacións naturais na súa secuencia. Por ende, a recombinación homóloga acontece de forma preferíbel durante as fases S e G2 do ciclo celular, cando a cromátide irmá está dispoñible. Aínda que historicamente considerouse a recombinación homóloga como unha reparación libre de erros, tamén se sabe que esta pode inducir inestabilidade no xenoma cando actúa de forma inadecuada ou non regulada. Por exemplo, se a reparación somática prodúcese entre homólogos, esta pode dar lugar á perda de heterocigosidade (evento potencialmente mutaxénico no que se perde unha das copias dun xene, especialmente perigoso cando está implicado un xene supresor de tumores). Da mesma forma, se a reparación acontece entre rexións repetitivas diverxentes en zonas non alélicas, esta pode dar lugar a eventos de translocación (Figura 1). Varias fontes de mutaxénese vinculadas á reparación do ADN por recombinación homóloga relacionáronse con diversas patoloxías humanas, dende mutacións a 20 pequena escala (mutacións puntuais e pequenas insercións e delecións) ata translocacións e reordenamentos cromosómicos masivos. A ruta de reparación de roturas de dobre cadea por recombinación homóloga está formada a súa vez por varias subrutas de reparación, incluíndo a hibridación de cadea sinxela, a hibridación de cadea dependente de síntese, a vía clásica de reparación de roturas de dobre cadea e a replicación inducida por rotura, as cales, difiren significativamente en termos de mecanismos de reparación e requisitos enzimáticos. Estudos empregando organismos modelo máis sinxelos estableceron os paradigmas que contribúen a entender os procesos equivalentes en células humanas. Dado que moitas das facetas da vía da recombinación homóloga estudáronse no fungo Saccharomyces cerevisiae, esta tese levouse a cabo empregando proteínas purificadas do devandito lévedo. Con todo, cabe destacar que a maioría dos mecanismos aquí referenciados están altamente conservados en todos os organismos eucariotas, o que pon de manifesto a relevancia da reparación por recombinación homóloga en todos os seres vivos. As catro subrutas de reparación previamente mencionadas comezan co mesmo evento, a degradación de 5' a 3' dos extremos do ADN escindido para xerar colas de ADN monocatenario, nun proceso nomeado resección dos extremos (Figura 2). Estas colas de ADN monocatenario con polaridade 3´ son rapidamente cubertas pola proteína de unión ao ADN monocatenario RPA (proteína de replicación A, SSB en bacterias), seguida da súa substitución pola recombinase altamente conservada Rad51 (RecA en bacterias). Este proceso acontece coa axuda das enzimas Rad52 e os parálogos de Rad51, Rad55-Rad57, dando lugar a unha estrutura coñecida como nucleofilamento ou filamento presináptico (Figura 4). Unha vez formado, o nucleofilamento realiza as reaccións características da vía da recombinación homóloga: procura de homoloxía e invasión dunha molécula de ADN intacta, a cal, será utilizada como molde para a reparación. Este evento xerará unha molécula conxunta formada polo apareamento da cadea invasora coa molécula de ADN molde, desprazando unha das cadeas do dúplex orixinal, dando lugar a unha estrutura coñecida como bucle de desprazamento (Figura 4). Este 21 intermediario representa a molécula central da reparación do ADN por recombinación homóloga. En S. cerevisiae, a proteína Rad54 é necesaria tanto para o evento de formación do bucle de desprazamento así como para o desprazamento da recombinase Rad51 do extremo 3´ do filamento presináptico. Isto permite o acceso da ADN polimerase δ ao extremo 3'-OH agora dispoñíbel, o cal, servirá como cebador para o comezo da síntese de ADN utilizando a molécula invadida como molde (Figura 5). Nas células somáticas, para evitar a formación de entrecruzamentos (eventos xenéticos que poden dar lugar a mutacións ou reordenamentos cromosómicos), unha vez copiada a información xenética, o bucle de desprazamento pode ser desestabilizado. Isto favorece que a cadea invasora, co ADN recentemente sintetizado, sepárese do bucle de desprazamento e hibride co outro extremo reseccionado presente na rotura. Este proceso representa unha das subrutas de reparación por recombinación homóloga, nomeada hibridación de cadea dependente de síntese (Figura 5). En S. cerevisiae, existen tres enzimas/complexos implicadas neste proceso: Sgs1-Top3Rmi1, Srs2 e Mph1. Este mecanismo de reparación conduce exclusivamente á formación de non entrecruzamentos, polo que, en células vexetativas, esta vía é favorecida fronte doutras que poden levar á xeración de entrecruzamentos. Ademais, se este mecanismo ten lugar entre cromátides irmás, darase lugar a unha conversión xenética equivalente. En mamíferos, esta conversión xenética entre cromátides irmás é a vía de reparación por recombinación homóloga predominante, xa que representa a subruta máis segura en termos de estabilidade xenética da vía da recombinación homóloga. Pola contra, se durante a proceso de síntese do ADN, a cadea desprazada no bucle alongado conecta co segundo extremo reseccionado da rotura, a ligadura da estrutura resultante dará lugar a unha molécula coñecida como dobre unión de Holliday (Figura 5). Esta estrutura, unida de forma covalente, representa o intermedio central doutra das subrutas de reparación por recombinación homóloga, a vía clásica de reparación de roturas de dobre cadea (Figura 6). Esta molécula pode ser 1) disociada topoloxicamente polo complexo Sgs1Top3-Rmi1, dando lugar exclusivamente a non entrecruzamentos, ou 2), pode ser escindida polas endonucleases selectivas de estrutura Slx1- 22 Slx4, Mus81-Mms4 e/ou Yen1, un tipo de enzimas especializadas no corte de estruturas ramificadas de ADN, xerando tanto entrecruzamentos como non entrecruzamentos (Figura 7). Estes mecanismos son de vital importancia para a supervivencia celular, xa que garanten a completa eliminación de calquera intermediario de recombinación remanente antes de que culmine a mitose. De non ser así, estes poderían interferir coa segregación cromosómica, dando lugar á perda de información xenética ou a xeración de aneuploidías. Por outra banda, eventos de reparación nos que non está presente este segundo extremo implicado na rotura poden ser reparados mediante outra subruta da recombinación homóloga, coñecida como replicación inducida por dano (Figura 8). Este tipo de lesións aparecen frecuentemente en contextos de furcos de replicación colapsados ou en telómeros erosionados. Os primeiros pasos desta vía da replicación inducida por dano son similares aos doutras subrutas da recombinación homóloga, dando lugar á formación dun bucle de desprazamento (Figura 8). Da mesma forma, esta vía tamén está preferentemente activa durante as fases G2/M do ciclo celular. Como neste tipo de roturas de dobre cadea non existe un segundo extremo ao que ligar, a síntese de ADN pode continuar ata o final do telómero, a non ser que sexa parada pola fusión cun furco de replicación converxente ou se deteña pola acción dunha nuclease. O trazo distintivo desta vía de reparación é que a síntese de ADN procede a través da migración dun bucle de desprazamento sen resolver, derivando na herdanza conservadora do novo ADN sintetizado (Figura 8). Este modo particular de síntese do ADN é un dos principais contribuíntes do alto nivel de inestabilidade xenética asociado con esta vía de reparación. A asincronía entre as febras líder e retrasada conduce á acumulación de ADN monocatenario por detrás do bucle de desprazamento. Este ADN monocatenario pode estar suxeito a danos, representando unha das fontes primarias de mutaxénese. O ADN en estado monocatenario é máis reactivo que cando se encontra en forma de dobre cadea, ademais de ter unha capacidade limitada de reparación de erros. En consecuencia, as lesións dunha soa febra adoitan dar lugar a mutacións que se transmiten á seguinte xeración. O patrón mutaxénico xerado 23 neste tipo de eventos é reminiscente aos eventos de kataegis descritos en células cancerosas, zonas de hipermutación rexional, nunha soa cadea, cunha taxa altísima de mutacións puntuais. Ademais, esta subruta de reparación tamén se relacionou con reordenamentos cromosómicos masivos. En primeiro lugar, o feito de que a síntese do ADN durante a replicación inducida por dano poida proceder ata o final do cromosoma converte a esta subruta nun fenómeno altamente perigoso para a estabilidade do xenoma. Se se está a utilizar o cromosoma homólogo en vez de a cromátide irmá como molde, esta vía de reparación pode producir unha extensa perda de heterocigosidade. Por outra banda, se a molécula invasora invadiu unha rexión repetitiva, isto pode derivar na xeración de translocacións non recíprocas. En segundo lugar, demostrouse que a replicación inducida por dano adoita acontecer a través de varias rondas de invasión (formación dun bucle de desprazamento) e desmantelamento, nun proceso denominado como cambio de molde. Este tipo de eventos poden representar una fonte de reordenamentos cromosómicos masivos cando acontecen entre secuencias repetitivas dispersas no xenoma. Ademais, describiuse que a resolución prematura dos intermedios xerados no contexto da replicación inducida por dano pode dar lugar a eventos de semi-entrecruzamento, os cales se caracterizan por presentar unha fusión entre o cromosoma doante e o receptor participantes na reparación. Estes eventos xeran tamén unha nova rotura de dobre cadea cun só extremo no cromosoma doante, o cal, pode perderse dando lugar a un evento de perda de información xenética, ou pode participar posteriormente na invasión doutro dúplex intacto. Isto podería conducir a rondas adicionais de formación de semi-entrecruzamentos, nun proceso coñecido como fervenzas de semi-entrecruzamentos, reminiscente dos ciclos de translocacións non recíprocas descritos en cancros humanos. Recentemente, propúxose un mecanismo parecido ás fervenzas de semi-entrecruzamentos para explicar outra fonte de inestabilidade xenética denominada reordenamentos mediados por multi-invasión, sendo ambos fenómenos posibles contribuíntes aos eventos de cromotripsis reportados nas células cancerosas. Cromotripsis é un fenómeno polo cal se producen unha serie de 24 reordenamentos cromosómicos masivos nun so cromosoma (ou nuns poucos) nun único evento celular catastrófico. Varios estudos xenéticos sinalaron ás endonucleases selectivas de estrutura, Mus81-Mms4 e Yen1, como enzimas involucradas na xeración dalgúns destes complexos reordenamentos cromosómicos no contexto da replicación inducida por dano, especialmente nos eventos de semi-entrecruzamento. O grupo da Dra. Symington demostrou que os semi-entrecruzamentos formados en ausencia de Polδ son parcialmente dependentes de Mus81-Mms4. Pardo e Aguilera (2012) identificaron un papel para Mus81-Mms4 e Yen1 tanto na promoción da replicación inducida por dano como na xeración de translocacións e semi-entrecruzamentos, probablemente escindindo o bucle de desprazamento implicado na reparación. Elango et al. (2017) ensinaron como a activación prematura de Yen1, mediante a expresión do mutante desregulado Yen1ON, diminúe a reparación por replicación inducida por dano ao mesmo tempo que conduce a un aumento dos eventos de perda cromosómica e semi-entrecruzamentos. Ademais, Piazza e colaboradores desvelaron a participación das endonucleases selectivas de estrutura nos reordenamentos mediados por multi-invasión, un fenómeno mecanisticamente similar ás fervenzas de semientrecruzamentos. A pesar de todos estes datos xenéticos, o substrato exacto e o mecanismo polo cal a acción destas enzimas podería explicar a xeración destes reordenamentos cromosómicos segue sen estar esclarecido. Os datos xenéticos suxiren fortemente que podería ser o bucle de desprazamento. Aínda que existen algúns datos bioquímicos sobre como Mus81-Mms4 procesa un bucle de desprazamento xerado pola hibridación de oligonucleótidos, nada se sabe sobre a capacidade de Yen1 de escindir este tipo de estruturas. Ademais, tradicionalmente, o estudo bioquímico destas endonucleases selectivas de estrutura levouse a cabo sobre substratos de ADN espidos, formados pola hibridación de oligonucleótidos de cadea sinxela (Figura 7). Pero, como xa se expuxo, os intermediarios de recombinación encóntranse interactuando con diversas proteínas que guían o proceso de reparación por recombinación homóloga e que poderían tamén modular a actividade das 25 endonucleases selectivas de estrutura sobre ditas moléculas. Porén, o obxectivo principal desta tese será analizar con maior precisión a actividade destas endonucleases selectivas de estrutura nun sistema que reflicta máis fielmente a complexidade dos intermediarios de recombinación in vivo e, así, poder recapitular algúns dos mecanismos subxacentes á reparación do ADN por recombinación homóloga. Para responder a todas estas preguntas, establecemos un enfoque bioquímico que nos permitiu investigar o procesamento de Mus81Mms4 e Yen1 sobre diferentes estruturas tipo bucle de desprazamento sintetizadas nun tubo de ensaio (Figura 9). Como punto de partida, comezamos xerando tres tipos distintos de bucles de desprazamento formados unicamente pola hibridación de oligonucleótidos de cadea sinxela. As tres estruturas que producimos representan diferentes estadios na formación deste tipo de intermediarios de recombinación: Bucle D3, o cal tenta mimetizar a estrutura equivalente trala invasión da cadea reseccionada na molécula de ADN molde (estrutura coñecida como bucle de desprazamento nacente); bucle D2, intermediario que aparecería tralo comezo da síntese do ADN e migración do bucle de desprazamento; e, por último, o bucle D1, estrutura amplamente empregada para os estudos deste tipo de intermediarios de recombinación in vitro. A marcaxe radioactiva de cada un dos oligonucleótidos que forman estas estruturas permitiunos analizar, con resolución dun nucleótido, o punto de corte exacto das endonucleases selectivas de estrutura nestes tres bucles de desprazamento. A análise dos resultados obtidos empregando esta estratexia experimental permitiunos, por un lado, confirmar que Mus81-Mms4 procesa o bucle tipo D3 na cadea complementaria á cadea invasora (Figura 16). Ademais, revelamos que Yen1 é capaz de escindir tanto a cadea desprazada como o oligonucleótido invasor, presentando tamén a capacidade de procesar o extremo 5´ da cadea complementaria ao oligonucleótido invasor (Figura 16). De xeito relevante, a combinación da incisión de Mus81-Mms4 xunto co corte producido por Yen1 na cadea desprazada nesta estrutura D3 (similar a un bucle de desprazamento nacente) podería explicar a xeración de eventos de semientrecruzamento. Ademais, a incisión de Yen1 sobre a cadea invasora, 1 INTRODUCTION 1.1 DNA DOUBLE-STRAND BREAKS AND GENOME STABILITY Tens of thousands of DNA lesions are constantly threatening genome integrity (Lindahl and Barnes 2000). The sources of these DNA lesions can be both endogenous (including metabolic reactions and replication stress) and exogenous (ionizing radiation, chemotherapeutics, etc.). Due to the nature of DNA double-strand breaks (DSBs), where the phosphodiester backbone of both strands is broken, they represent one of the most hazardous lesions to genome stability (Mehta and Haber 2014). Relevantly, the majority of spontaneous DSBs appear in the context of DNA replication, presumably when a replication fork encounters a single-stranded break while polymerising (Syeda et al. 2014). Failure to repair a DSB may lead to loss of genetic information and/or cell death but also its aberrant repair can cause irreversible genomic changes such as insertions, deletions, or translocations. These events are collectively known as genomic rearrangements, an important driver of oncogenesis and genomic disorders (Strathern et al. 1995; Malkova and Haber 2012; Yang et al. 2008; Elliott and Jasin 2002; Weinstock et al. 2006; Khanna and Jackson 2001; Rodgers and Mcvey 2016). In the complex task of dealing with a DSB, cells rely on two conserved mechanisms: non-homologous end-joining (NHEJ) and homologous recombination (HR). While NHEJ involves the direct ligation of the two broken ends, a process that can occur with little or no further modification at the breakpoint (Daley et al. 2005; Chang et al. 2017), HR is generally considered an error-free mechanism, since it takes advantage of the presence of a homologous sequence as a template for the repair (Figure 1). During somatic DNA repair, the common Introduction 34 template is the sister chromatid (Johnson and Jasin 2000; Hartwell 1992). Thus, HR preferentially takes place during Sand G2-phases of the cell cycle, when the sister chromatid is available (Aylon et al. 2004; Orthwein et al. 2015). In contrast, in the meiotic program, the preferred template is the homolog chromosome, which can contain natural sequence variations (Schwacha and Kleckner 1997). Although HR has historically been considered an error-free repair, it is also known that it can induce genome instability if acting inappropriately or unregulated (Elliott and Jasin 2002; Kolodner et al. 2002). For instance, if somatic cell repair occurs between homologs, it can result in loss of heterozygosity (LOH) and, if the repair occurs between repeats with diverged sequences at non-allelic regions, it can be the source of translocation events (Jinks-Robertson and Petes 1986; Moynahan and Jasin 1997; Ray et al. 1989; Weinstock et al. 2006; Anand et al. 2014; Tsaponina and Haber 2014). Several sources of mutagenesis have been linked to HR ranging from small-scale mutations, such as point mutations and small insertions and deletions (indels), up to translocations and gross chromosomal rearrangements (GCRs) (Figure 1) (Hicks et al. 2010; Malkova and Haber 2012). Four elements have been described to account for the mutagenesis associated with HR repair: donor selection, template integrity, polymerase choice and processivity (McVey et al. 2016). Genome instability was proposed as an enabling characteristic underlying hallmarks of cancer cells (Hanahan and Weinberg 2000, 2011) and it is a feature of almost all human cancers (Negrini et al. 2010). For example, chromothripsis, a GCR localized in space (one or a few chromosomes) and time (one or a few cell divisions) (Stephens et al. 2011; Zhang et al. 2015), as well as kataegis, characterized by clusters of strand-coordinated mutations (Chan and Gordenin 2015) were described in cancer cells (Chan and Gordenin 2015; CortésCiriano et al. 2020; Zack et al. 2013; Kloosterman et al. 2011; Roberts et al. 2012; Taylor et al. 2013). Both mutational phenomena have been proposed to be partially HR-related (Sakofsky et al. 2014; Savocco and Piazza 2021; Duda et al. 2016; Elango et al. 2019; Chen et al. 2010). Introduction 35 In the following sections, the mechanisms of DSB repair by HR are presented, highlighting the steps that may pose a threat to genome stability. Figure 1. Dangerous outcomes associated with DSB repair. Double-strand breaks (DSBs) can be repaired by non-homologous end-joining (NHEJ) or homologous recombination (HR). Both pathways can lead to the generation of mutations. Abbreviatures: CNVs, copy number variations; LOH, loss of heterozygosity; GCRs, gross chromosomal rearrangements. More information in section 1.1. Figure created by the author employing Adobe Illustrator. 1.2 MECHANISMS OF HOMOLOGOUS RECOMBINATION Several sub-pathways for the repair of DSB by HR have been described, including single-strand annealing (SSA), synthesis-dependent strand annealing (SDSA), double-strand break repair (DSBR), and breakinduced replication (BIR), which significantly differ in terms of repair Introduction 36 mechanisms and enzyme requirements (Kowalczykowski 2015). Studies employing simpler model organisms have established the paradigms to help understand the equivalent processes in humans. Since many of the facets of the HR pathway has been studied in the budding yeast Saccharomyces cerevisiae and the results presented in this thesis were obtained employing yeast proteins, the following chapters will focus on describing the machinery of HR in this organism. Nevertheless, it should be noted that most of the mechanisms are highly conserved across eukaryotes, highlighting the relevance of HR repair in all living organisms. 1.2.1 Nucleolytic resection of DNA ends The critical determinant of repair pathway selection is the 5´-3´ degradation of the broken DNA ends. This process, named end resection, generates 3´-single-stranded DNA (ssDNA) tails, an intermediate that commits the channelling of DSB repair towards HR. On the one hand, it creates the 3´-overhang that will be used for homology search. On the other, the generated ssDNA is a poor binding substrate for the Ku70-Ku80 (Ku) complex, essential for the repair of DSBs by NHEJ (Shibata et al. 2014; Cejka 2015; Chapman et al. 2012). As a crucial point for DNA repair pathway choice, end resection is strictly regulated. This containment is achieved by the cell cycle modulation of some of its key players, such as the endonuclease Sae2 (CtIP in mammals). Sae2 phosphorylation by cyclin-dependent kinases (CDKs, Cdc28 in S. cerevisiae) and other DNA damage response kinases during Sand G2/M-phases is required for the stimulation of the nuclease complex responsible for the initiating stages, MRX/N (Mre11-Rad50-Xrs2 (MRX) in S. cerevisiae; MRE11-RAD50-NBS1 (MRN) in mammals). Hence, end resection is limited to Sand G2/Mphases of the cell cycle, when the sister chromatid is available as a template for the repair (Cannavo et al. 2018; Huertas et al. 2008; Huertas and Jackason 2009; Ira et al. 2004; Cannavo and Cejka 2014; Aylon et al. 2004; Anand et al. 2016). Introduction 37 DNA end resection occurs by a two-step mechanism (Mimitou and Symington 2008; Gravel et al. 2008; Zhu et al. 2008). First, MRX is recruited to DSBs. Mre11, the nuclease responsible for the initiation of end resection, is endowed with both endonucleolytic and 3´-5´ exonuclease activity (Paull and Gellert 1998). This polarity of resection is the opposite of the expected one predicted by the DSB repair models, which indicates that resection occurs in 5´-3´direction. This conundrum was solved by the Cejka lab showing that Sae2 interaction with Mre11 within the MRX complex promotes the double-stranded DNA (dsDNA)-specific endonuclease activity of Mre11 (Cannavo and Cejka 2014; Anand et al. 2016). Thus, through Xrs2-mediated nuclear entry (Tsukamoto et al. 2005; Oh et al. 2016), phosphorylated Sae2 interacts with the ATPase Rad50 to activate Mre11 endonuclease activity to generate nicks at the 5´-terminated DNA, from 15 up to ~300 nucleotides (nt) away from the end (Cannavo and Cejka 2014; Neale et al. 2005; Garcia et al. 2011; Cannavo et al. 2019; Shibata et al. 2014). After MRX nicking, short and long-range resection can take place (Figure 2). During short-range resection, Mre11 utilises the nick as an entry site to resect in 3´-5´ direction back towards the DSB, employing its exonuclease activity (Figure 2A). This process is essential to remove DNA ends with noncanonical structures as DNA hairpins, chemical modifications, or protein adducts such as persistent Ku-bounded ends, aberrant topoisomerase-DNA complexes or Spo11-covalent attachment to induced DSBs during meiosis (Neale et al. 2005; Wang et al. 2017; Garcia et al. 2011; Langerak et al. 2011; Hartsuiker et al. 2009; Reginato et al. 2017; Mimitou and Symington 2010). During long-range resection, two partially redundant mechanisms extend the resection tracts (Zhu et al. 2008; Mimitou and Symington 2008). One depends on the Exo1 (EXO1) nuclease and the other on the dual enzyme Dna2 (DNA2) in conjunction with Sgs1 (BLM or WRN in mammals) and the accessory factors Top3-Rmi1 (TOPO IIIα-RMI1RMI2 in mammals) and replication protein A (RPA) (Figure 2B) (Cejka et al. 2010a; Szankasi and Smith 1995; Niu et al. 2010; Sturzenegger et al. 2014; Levikova et al. 2017). Introduction 38 Exo1 belongs to the XPG/Rad2 family of structure-selective endonucleases (SSEs). It is endowed with 5´-3´ exonuclease and 5´-flap (5´F) endonuclease activity. It degrades 5´-terminated dsDNA with a preference for DSBs with 3´-overhangs, although it can also process 5´- overhangs with lower efficiency (Tran et al. 2002; Cannavo et al. 2013). Importantly, it can begin DNA resection from a nicked dsDNA (Reginato et al. 2017; Wang et al. 2018) and it is stimulated by the MRX complex and Sae2 (Nicolette et al. 2010; Cannavo et al. 2013). The second mechanism requires the Sgs1-Dna2 complex. Dna2 is a dual enzyme with nuclease and 5´-3´ helicase activities, but this latter one is only active upon nuclease inactivation (Bae et al. 1998; Levikova et al. 2013). Unlike Mre11 and Exo1, Dna2 only degrades ssDNA, but with both polarities (Bae and Seo 2000; Kao et al. 2004). Therefore, it requires the RecQ-family 3´-5´ helicase Sgs1 to unwind the DNA. Sgs1 produces an ssDNA intermediate that is bound by RPA. This stimulates Sgs1 and the 5´ to 3´ nucleolytic capacity of Dna2, while preventing the degradation of the 3´-terminated strand, thus directing Dna2 nucleolytic activity towards the 5´-terminated strand. Dna2 is recruited by direct interaction with Mre11 and Top3-Rmi1, and the MRX complex stimulates end resection by Sgs1-Dna2 (Kasaciunaite et al. 2019; Cejka et al. 2010a; Xue et al. 2019; Zhu et al. 2008; Whelan and Rothenberg 2021; Niu et al. 2010; Sturzenegger et al. 2014). A model has been proposed in which MRX nicking activity serves as an initiating point for both 3´-5´ short-range resection (MRX complex) and 5´-3´ long-range resection (Exo1 and Sgs1-Dna2 complex), termed bidirectional resection model (Figure 2C) (Garcia et al. 2011; Shibata et al. 2014). This model argues that the generation of long 3´-ssDNA with a clean 3´-end is required for the downstream steps in HR. However, in unblocked DSB, Mre11 is dispensable for end resection (Llorente and Symington 2004) and, although less efficiently, Exo1 and Dna2-Sgs1 can initiate end resection in an MRX-independent manner (Zhu et al. 2008; Cejka et al. 2010a; Nicolette et al. 2010; Niu et al. 2010; Cannavo et al. 2013). As an additional level of complexity, it has been described that long-range resection is dispensable for Introduction 39 meiotic recombination and its absence has only a mild effect in vegetative cells (Zhu et al. 2008; Chung et al. 2010). Figure 2. Schematic representation of DNA end resection. (A) Short-range resection carried out by the MRX complex and stimulated by phosphorylated (P) Sae2. This process is essential for block removal at DNA ends. (B) Long-range resection pathway. MRX recruits the Sgs1-Dna2 complex or Exo1 independently of its nuclease activity. (C) Bidirectional model of end resection. MRX nicking activity serves as an initiating point for both shortand long-range resection. Protein names and cartoons are colour-coded. More information in section 1.2.1. Figure created by the author employing Adobe Illustrator. Introduction 40 1.2.2 Homology search and strand invasion After DNA end resection, the 3´-ssDNA tail is rapidly coated by the ssDNA-binding protein RPA (SSB in bacteria), protecting it from degradation and removing secondary structures (Chen and Wold 2014). In most cases, RPA is afterwards replaced by the highly conserved recombinase Rad51 (RecA in bacteria). Rad51 polymerises onto the ssDNA, assembling a nucleoprotein filament able to invade (intertwine with) a donor DNA molecule (also called template). After the invasion, the 3´-end of the ssDNA serves to prime repair DNA synthesis (Liu et al. 2011a). Therefore, recombinases trigger the pairing and shuffling of DNA sequences, through reactions that involve homologous DNA pairing and strand exchange. In the following subsections, these mechanisms will be developed but first, a Rad51-independent HR subroute will be briefly commented on. 1.2.2.1 Single-strand annealing Single-strand annealing (SSA) is considered as a HR subpathway due to its enzyme requirements, although it does not involve strand invasion (Ivanov et al. 1996; Bhargava et al. 2016). If repeated sequences are exposed during end-resection, the DNA binding protein Rad52 (RAD52), aided by Rad59, promotes the annealing of homologous sequences (Figure 3) (Sugawara et al. 2000; Mortensen et al. 1996). This results in DNA duplexes with two non-homologous 3´- tails (3´-flaps, 3´F) at the breaking point. These structures are removed by the structure-selective endonuclease Rad1-Rad10 (XPF-ERCC1 in mammals) in a complex that includes the mismatch repair factors Msh2-Msh3 and the scaffold proteins Slx4 and Saw1 (Ivanov and Haber 1995; Li et al. 2008; Sugawara et al. 1997; W.-L. Toh et al. 2010; Flott et al. 2007). It has been recently described that the timely and accurate position of Rad1-Rad10 at 3´-flaps is mediated by its interaction with Msh2-Msh3, Saw1, and RPA (Eichmiller et al. 2018; Seol et al. 2018). After 3´-flap removal, DNA synthesis fills the gaps followed by ligation to complete repair (Figure 3). This pathway is considered highly mutagenic because it implies the deletion of one copy Introduction 41 of the repeat plus the sequence between them, therefore contributing to genome instability associated with HR repair (Pâques and Haber 1999). Figure 3. Schematic representation of the single-strand annealing (SSA) pathway. After DNA end resection, Rad52 mediates the pairing of exposed repeat sequences. 3´-flaps are removed by Rad1-Rad10. Note that the repaired duplex is shorter than the original one due to nucleolytic removal of the protruding 3´-ssDNA ends. Protein names and cartoons are colour-coded. More information in section 1.2.2.1. Figure created by the author employing Adobe Illustrator. 1.2.2.2 Nucleofilament formation In contrast to SSA, the other three sub-pathways for HR repair, SDSA, DSBR, and BIR, require homology search and strand invasion of an intact duplex. For this task, the single-strand binding protein RPA must be replaced by the strand-exchange protein Rad51. Introduction 48 recombinase (Nimonkar et al. 2012). However, contrary to Dmc1, Rdh54 is also expressed in mitotically dividing cells (Lee et al. 2001) and its deletion leads to an increase in aneuploidy (Sharp et al. 2018), suggesting a role during somatic HR. Rdh54 was found to inhibit Rad51/Rad54-mediated D-loop formation and to limit D-loop length, independently of its motor activity (Shah et al. 2020; Piazza et al. 2019). Although they have distinct binding sites within the presynaptic filament (Crickard et al. 2020a), it was also reported that Rdh54 competes with Rad54 for its binding site, working as a roadblock to Rad54-mediated hDNA formation (Shah et al. 2020). In line with this, it has been recently proposed that it can stabilize Rad51 at the nascent D-loop structure, leading to a reduced rate of D-loop maturation (Keymakh et al. 2022). In addition, thanks to the development of a genetic assay to monitor D-loops in vivo, Piazza and collaborators proposed that Rdh54 delineates two independent D-loop reversal pathways defined by the extension of the hDNA (Piazza et al. 2019). However, further research needs to be done to clarify the discrete role of these proteins during HR repair, as well as their implication in template selection during mitotic and meiotic programs. 1.2.3 DNA synthesis and processing of recombination intermediates One of the key steps to ensure the fidelity of DSB repair by HR is DNA synthesis, since the accurate incorporation of nucleotides, as well as the length of the extended strand, is essential to precisely restore the genetic information lost in the lesion. 1.2.3.1 DNA repair synthesis After the formation of a nascent D-loop, the 3´-OH of the invading strand serves as a priming site for polymerases. They use the recipient strand as a template for DNA synthesis, extending the branched molecule and further displacing the dsDNA. This process leads to the formation of an extended D-loop (Figure 5). Introduction 49 DNA repair synthesis is executed by several components of the replication machinery (Wang et al. 2004). The high-fidelity polymerase (Pol) δ, which presents strand-displacement and proof-reading activities, is the primary polymerase that extends the free 3´-OH at the D-loop. The strand-displacement activity is needed to move the D-loop bubble to allow DNA synthesis, and the 3´-5´ exonuclease proofreading activity enhances its fidelity (Maloisel et al. 2008; Hicks et al. 2010; Sneeden et al. 2013; Sebesta et al. 2011; Wang et al. 2004). Polδ contains three subunits: the essential Pol3 (catalytic subunit) and Pol31, and the non-essential Pol32 (the human enzyme contains four subunits POLD1-4) (McVey et al. 2016). Extension of the invading strand also depends on the PCNA clamp and its loader RFC (replication factor C) (Wang et al. 2004; Sneeden et al. 2013; Li et al. 2009; Majka and Burgers 2004). PCNA is a homotrimeric complex that encircles DNA, preventing polymerase dissociation and, hence, enhancing its processivity and reducing template switching. Moreover, it serves as a docking site for other DNA replication and repair proteins (Moldovan et al. 2007). DNA synthesis is aided by RPA binding to ssDNA, both at the displaced strand (Eggler et al. 2002) and ahead of Polδ, preventing the formation of secondary structures and facilitating processive DNA synthesis (Sneeden et al. 2013; Osia et al. 2022). In vitro studies have shown that DNA synthesis can proceed until a topological block is reached, approximately after synthesizing 100-150 nt (Wright and Heyer 2014). Extension consumes one negative supercoil each helical turn (~10.5 base pair (bp)). In vivo, the length of the resected invading end is thought to relax the donor supercoiling. DNA extension produces positive supercoils that would block DNA synthesis. It has been suggested that one way to overcome this inhibition is by extruding the 5´-side of the invading strand, establishing a migrating bubble that prevents topological stalling (Wright et al. 2018). Introduction 50 Figure 5. Schematic representation of DNA repair synthesis, synthesis-dependent strand annealing (SDSA), and initiation of classical double-strand break repair (DSBR). After nascent D-loop formation, DNA polymerase δ (Polδ) with RFC-PCNA extends the invading end, displacing the original duplex. (A) During SDSA, the helicases Srs2, Mph1, and STR can dissociate the extended D-loop. Rad52 mediates the re-annealing of the two RPA-coated ssDNA ends, leading exclusively to noncrossover (NCO) recombinants. (B) In DSBR, Rad52 anneals the RPA-coated displaced strand with the second end of the DSB. Following DNA synthesis and ligation, a double Holliday junction (dHJ) intermediate is generated. Protein names and cartoons are colour-coded. More information in section 1.2.3. Figure created by the author employing Adobe Illustrator. Introduction 51 DNA extension must advance long enough so that when the invading strand is displaced, it contains sufficient sequence homology to allow its re-annealing with the second end of the DSB (Figure 5A). However, D-loops can also be disrupted before DNA synthesis begins. While disruption of nascent D-loops is an anti-recombinase mechanism, important to prevent excessive HR, disruption of extended D-loops represents the primary mechanism in somatic cells for the repair of DSBs. As this HR sub-pathway involves DNA synthesis followed by D-loop disruption and strand annealing, it is known as synthesis-dependent strand annealing (SDSA), detailed in the next section. It should be noted that, as long as DNA repair synthesis takes place, a gene conversion event will occur, defined as the nonreciprocal transfer of genetic information between two chromosomes in a repair event (Symington et al. 2014). 1.2.3.2 Synthesis-dependent strand annealing SDSA is the primary mechanism to avoid crossover (CO) formation in mitosis, as it prevents the generation of late recombination intermediates whose resolution can produce CO events (Ira et al. 2003; Bzymek et al. 2010). A CO is defined as the reciprocal exchange between the DNA duplexes that flank the breaking point (Petronczki et al. 2003). They are genetically silent events when they occur between identical sister chromatids, but they can also represent a source of deletions, duplications, chromosomal rearrangements, and LOH when they occur between repeats at non-allelic loci or between homologs with sequence variation (Malkova and Haber 2012). In fact, suppression of mitotic COs is critical to maintain genome integrity (Richardson et al. 1998; Shaw and Lupski 2004). In somatic cells, the majority of Dloops are dissociated by the anti-crossover helicases to minimize CO formation (Ira et al. 2003). In S. cerevisiae, three enzymes/complexes have been implicated in D-loop disruption: Srs2, Mph1, and Sgs1Top3-Rmi1 (STR) (Figure 5A) (Fasching et al. 2015; Piazza et al. 2019; Liu et al. 2017; Prakash et al. 2009). Introduction 52 Srs2 and Mph1 (FANCM in humans) were first genetically implicated and further biochemically demonstrated to be able to dissociate nascent and extended D-loops. They translocate in 3´-5´ direction, unwinding the D-loop structure using the energy from ATP hydrolysis (Liu et al. 2017; Sebesta et al. 2011; Prakash et al. 2009; Gangloff et al. 2000; Ira et al. 2003). In the case of STR (BLM-TOPO IIIα-RMI1-RMI2 (BTR) in humans), Sgs1 alone is able to dissociate deproteinized D-loop structures, but not enzymatically-reconstituted ones, as opposed to Top3 alone or the Top3-Rmi1 complex, which can only process Rad51-Rad54 mediated D-loops (Fasching et al. 2015). Surprisingly, this function is independent of the ATPase activity of Sgs1 but dependent on the relaxation activity of Top3 (Fasching et al. 2015). As mentioned in the previous section, it has been recently proposed that Rdh54 delineates two independent D-loop disruption pathways defined by the extension of the hDNA (Piazza et al. 2019). In this model, STR and Mph1 dissociate Rdh54-limited extended D-loops, while longer D-loops are exclusively processed by Srs2. In line with this hypothesis, it was also shown that Srs2 interaction with SUMOylated PCNA confers Srs2 a slight preference for the disruption of extended D-loops (Liu et al. 2017). These three helicases translocate in 3´ to 5´ direction over the Dloop to unwind the hDNA. This ejects the invading strand as ssDNA that is quickly bound by RPA. If there has been sufficient homology sequence synthesized, the two RPA-coated DSB ends can re-anneal (Figure 5A). In yeast, this reaction is mediated by Rad52 (Shi et al. 2009; Sugiyama et al. 2006; Nimonkar et al. 2009). After Rad52mediated re-annealing, DNA synthesis is required to complete the repair of the second end of the DSB (Figure 5A). The identity of the DNA polymerase implicated in this step remains controversial, although genetically, both Polδ and Polε have been involved in HR repair (Wang et al. 2004; McVey et al. 2016). Ligation of the remaining nicks restores the chromosome integrity in a non-crossover (NCO) configuration. If DNA synthesis exceeds the resection tract before Dloop dissociation, it can produce 3´-flaps that will require the SSE Introduction 53 Rad1-Rad10 to remove them, allowing further ligation of the two DNA ends (Lyndaker and Alani 2009). D-loop disruption pathways are proposed to contribute to genome integrity in several ways. First, they prevent CO generation by reducing double Holliday junction (dHJ) formation (see next section) (Mazón and Symington 2013); second, they limit the extension of HR-related DNA synthesis, which is associated with a ∼1000-fold increase in mutation frequency (Strathern et al. 1995); third, by limiting the extension of the hDNA, they also hamper CO formation, as longer extension tracts favour dHJ formation (Neuwirth et al. 2007; Aguilera and Klein 1989); and fourth, these activities hinder the formation of multi-invasion species, where one broken end may invade and establish hDNA with more than one donor sequence. These species have been described both in vivo and in vitro, and have the potential to create translocations and cascades of rearrangements in a tripartite mechanism involving the nucleolytic processing of nascent D-loops (Piazza et al. 2017; Wright and Heyer 2014; Piazza and Heyer 2019b). 1.2.3.3 Second-end capture and Holliday junction formation In some instances, after DNA repair synthesis and before extended D-loops get disrupted by the action of anti-crossover helicases, Rad52 mediates the annealing between the displaced strand of the D-loop and the second end of the DSB (Figure 5B) (Shi et al. 2009; Sugiyama et al. 2006; Nimonkar et al. 2009). Besides, it has also been proposed that the two resected ends can invade the same donor, which would follow simultaneous DNA synthesis (Wright et al. 2018). In both cases, ligation of the resultant nicked duplexes generates a double Holliday junction, the distinctive intermediate of the DSBR model (Szostak et al. 1983; Bzymek et al. 2010). A Holliday junction (HJ) is each one of the two four-way secondary DNA structures where two duplexes exchange a pair of single strands (Figures 5B and 6) (Holliday 1964). The DSBR model, also known as the dHJ model, was firstly suggested by Resnick and Martin (Resnick and Martin 1976) and later elaborated by Szostak and collaborators (Szostak et al. 1983). Due Introduction 54 to the covalent nature of this joint molecule, in which two DNA molecules are linked together, cells need to ensure their removal before cell division, since they could obstruct chromosomal segregation and lead to deleterious chromosomal rearrangements and aneuploidy (Matos et al. 2011; Wechsler et al. 2011; Chan and West 2018). It is important to note that HJs can also appear during DNA replication, as HR is also involved in replication fork reactivation and post-replicative repair (Giannattasio et al. 2014; Branzei and Szakal 2017; Falquet and Rass 2019; Prado 2018). These dHJs can be resolved either as COs or NCOs by the SSEs (developed in section 1.3) or dissolved by the STR complex, leading exclusively to NCO outcomes (developed in the next section) (Figure 6). 1.2.3.4 Dissolution As a last resort to avoid CO formation, cells can cope with the dHJs employing the dissolution pathway. The RecQ-family helicase Sgs1 (BLM in humans) promotes branch migration of the two HJs, fusing them into a hemicatenane intermediate. Then, the type Ia topoisomerase Top3 (TOPO IIIα) decatenates this intermediate. This Sgs1-Top3 complex has another partner, the OB-fold containing accessory protein Rmi1 (RMI1-2), which stimulates the dissolution process (Figure 6A). As aforementioned, these three proteins form a complex known as STR (BLM in humans) (Chang et al. 2005; Mullen et al. 2005; Wu and Hickson 2003; Gangloff et al. 1994; Cejka et al. 2012, 2010b; Xue et al. 2013). Mutations in BLM lead to the cancerprone disease Bloom syndrome (Chu and Hickson 2009). The hallmark of BLM mutant cells is that they present elevated levels of genome instability and sister chromatid exchanges, which occur as a result of COs between sister chromatids. Moreover, S. cerevisiae mutants for the STR complex show higher levels of CO recombinants (Ira et al. 2003). These observations indicate that dHJs that are not processed by the STR complex are further cleaved by SSEs, in a process known as resolution (see next section). Accordingly, deletion of the nucleases involved in resolution eliminates this CO increase (Wechsler et al. 2011; Wyatt et al. 2013; Sarbajna et al. 2014). Introduction 55 Figure 6. Schematic representation of double Holliday junction processing pathways. (A) Dissolution. Sgs1 promotes convergent migration of the single HJs, creating a hemicatenane that is further decatenated by Top3-Rmi1. This pathway generates NCOs exclusively. (B) Resolution. Resolvases can hydrolyse the double Holliday junction in two possible configurations. Incisions in the same orientation of the two single HJ result in an NCO outcome, whereas incisions in different orientations generate a CO outcome. Solid and shaded Pacmans represent the two possible cleavage orientations by the SSEs. Note that only two out of four possible resolution outcomes are depicted. Protein names and cartoons are colour-coded. More information in sections 1.2.3.4 and 1.2.3.5. Figure created by the author employing Adobe Illustrator. Introduction 56 This pathway, in conjunction with SDSA, handles the majority of the recombination intermediates during mitotic proliferation (Ira et al. 2003; Symington et al. 2014; Dayani et al. 2011). However, the exact mechanisms by which cells establish this hierarchy of action remains an open question in the field. Recent work from the Matos lab has shown that the activity of both Sgs1, and its human counterpart BLM, are cell-cycle regulated. This is accomplished by changes in their phosphorylation status that drives Sgs1/BLM stimulation during the mitotic Sand G2-phases (Pogliano et al. 2022; Grigaitis et al. 2020). Surprisingly, this regulatory control is achieved by the same cell-cycle kinases that also restrict resolvase activity until late phases of the cell cycle, Cdc28/CDK and Cdc5/PLK (see section 1.3) (Blanco et al. 2014; Eissler et al. 2014; Szakal and Branzei 2013; Gallo-Fernández et al. 2012; Matos et al. 2011). Overall, this suggests that cells are endowed with a system that ensures a bias towards NCO formation, where the SSEs act as a last resort for the elimination of persistent recombination intermediates, ensuring chromosome segregation even at the cost of generating some crossovers (Grigaitis et al. 2020). 1.2.3.5 Resolution The resolution pathway acts as a safeguard for the elimination of persistent HJs and other recombination intermediates that can interfere with cell division (Blanco and Matos 2015). For instance, single HJs, structures that cannot be dissolved by the STR complex, need to be eliminated by this pathway. Resolution is conducted by SSEs that have acquired the ability to introduce cuts on opposite strands of the HJ, generally termed resolvases. In somatic cells, there are three conserved nuclear SSEs implicated in the removal of branched DNA intermediates: Slx1-Slx4 (SLX1-SLX4/BTBD12/FANCP), Mus81Mms4 (MUS81-EME1/EME2), and Yen1 (GEN1) (Wyatt and West 2014). Depending on the orientation of the incisions across the HJ, CO or NCO products will be generated. If the resolvases cut the two HJs in the same orientation (i.e., the crossing or the non-crossing strands), ligation of the resultant nicks generates only NCOs (Figure 6B). Conversely, if they hydrolyse the crossing strands of one HJ and the Introduction 57 continuous strands of the other, the resultant product is a CO recombinant (Figure 6B). Additionally, MutLγ-Exo1, a nuclease that severs dHJs into COs exclusively, plays its mayor role during the meiosis program, ensuring chiasmata formation and accurate homologous chromosome segregation (Cannavo et al. 2020; Kulkarni et al. 2020; Zakharyevich et al. 2012, 2010). 1.3 RESOLVASES As stated previously, resolvases are structure-selective endonucleases. Besides HJ, these enzymes can recognize and process other branched DNA structures. This property allows them to act throughout the entire genome, contributing to genome integrity by removing secondary DNA structures that can impair important cellular processes such as DNA transcription, replication, repair, or chromosome segregation. However, this capacity also turns them into dangerous weapons that cells must be sure to keep under control (Wyatt and West 2014). For instance, anaphase bridges that impair chromosome segregation appear in the absence of these nucleases (Chan and West 2018; García-Luis and Machín 2014; Sarbajna et al. 2014), whereas their untimely activity can pose a threat to the replication progress (Duda et al. 2016; Szakal and Branzei 2013). Although it has been demonstrated that Slx1-Slx4 can cleave HJs in vitro (Svendsen et al. 2009; Fekairi et al. 2009; Fricke and Brill 2003; Muñoz et al. 2009), its role as resolvase in DSBR remains controversial. Contrary to sgs1Δmus81Δ mutants, the synthetic lethality of sgs1Δslx1Δ and sgs1Δslx4Δ mutants was not suppressed in the absence of HR, indicating that the toxic accumulation of recombination intermediates is not their primary cause of death (Fricke and Brill 2003; Fabre et al. 2002; Bastin-Shanower et al. 2003). Slx1-Slx4 was proposed to have a specific role in replication termination and rDNA maintenance as well as Slx4, but not Slx1, in the repair of replication-borne structures (Coulon et al. 2006; Schwartz and Heyer 2011; Fricke and Brill 2003; Deng et al. 2005; Kaliraman and Brill 2002; Munoz-Galvan et al. 2012). Therefore, the following sections will focus on the description of the Introduction 64 in chromosomal loss (CL) and half-crossovers (HC), two threatening events to genome stability (see section 1.4 section) (Elango et al. 2017). Human GEN1 also contains eight CDK consensus target sites and is phosphorylated at the M-phase. However, its biochemical activity seems relatively independent of its phosphorylation status. Instead, it is regulated by a nuclear export signal that keeps GEN1 outside the nucleus until mitosis, when the nuclear envelope breaks down (Chan and West 2014). Overall, it has become apparent that cells must carefully adjust the precise timing of resolvase activation to avoid genome instability. In Sphase, cells prioritize the STR dissolution pathway over resolution to prevent SSE-dependent catastrophic events. This is achieved by coupling the activation of resolvases to cell-cycle progression through post-translational modifications. In this way, Mus81-Mms4 and Yen1 are sequentially activated in G2/M and anaphase, respectively, well after the bulk of DNA replication is completed. Such strategy allows the safe removal of persistent joint molecules while minimizing their undesired effects on replication/early recombination intermediates (Wild and Matos 2016; Giaccherini and Gaillard 2021). 1.4 BREAK-INDUCED REPLICATION Break-induced replication is a HR sub-route specialized in dealing with frequently occurring one-ended DSBs, that is, when there is no second end to anneal (Figure 8). These lesions typically arise at broken replication forks (for instance, when they encounter a nick on the DNA while polymerising) or at eroded telomeres (Kockler et al. 2021). BIR has been profusely studied in S. cerevisiae, where two BIR pathways have been described: one is Rad51-independent and mediated by Rad52 strand annealing activity and the other is Rad51-dependent (Malkova et al. 1996, 2005; Davis and Symington 2004; Signon et al. 2001). As the Rad51-independent BIR pathway is poorly understood and the Rad51-dependent is the major BIR pathway in yeast, the following section focuses on the latter. Introduction 65 The first steps of the Rad51-dependent BIR pathway are similar to other sub-routes of HR, resulting in the formation of a D-loop structure (Davis and Symington 2004; Signon et al. 2001). Therefore, it is also preferentially active at G2/M. Since in these one-ended DSBs there is no second end to anneal, DNA synthesis can continue downstream to the end of the telomere, unless stopped by nuclease action or merged with a converging replication fork (Saini et al. 2013; Mayle et al. 2015). The hallmark of this pathway is that DNA synthesis is conservative, as lagging strand synthesis uses the leading strand as a template, leading to conservative inheritance of newly synthesized DNA (Figure 8) (Saini et al. 2013; Donnianni and Symington 2013). DNA synthesis proceeds through a bubble-migration mechanism that is driven by branch migration of an unresolved D-loop structure. Lagging strand synthesis is uncoupled from the leading one, resulting in the formation of long ssDNA (Figure 8) (Saini et al. 2013; Wilson et al. 2013; Donnianni and Symington 2013). Polα-primase complex initiates lagging strand synthesis and Polδ is responsible for both leading and lagging strand synthesis (Donnianni et al. 2019; Liu et al. 2021). The non-essential subunit Pol32 (POLD3) is essential for extensive BIR progression (Lydeard et al. 2007). In its absence, BIR can initiate but it cannot proceed longer than 15 kb (Piazza et al. 2019; Liu et al. 2021). Moreover, Pif1 (PIF1) is the main BIR helicase (Wilson et al. 2013; Buzovetsky et al. 2017) and when absent, BIR is interrupted within the first 5 kb, resulting in premature resolution of the DNA structure (Liu et al. 2021). Although the actual functions of Pif1 in BIR remain poorly studied, it has been proposed that it can unwind the DNA duplex to facilitate DNA synthesis and also that it can disengage the newly synthesized strand to prevent topological constraints (Saini et al. 2013; Wilson et al. 2013). In agreement with the idea that BIR intermediates are reminiscent of those formed during classical DSBR (Mehta et al. 2017; Štafa et al. 2014; Prakash et al. 2009), Mph1, one of the helicases involved in SDSA, opposes BIR progression, as its frequency increase when Mph1 is absent, probably because Mph1 can unwind migrating D-loops (Prakash et al. 2009; Sebesta et al. 2011). Introduction 66 Figure 8. Schematic representation of the Rad51-dependent break-induced replication (BIR) pathway. After nascent D-loop formation, long-range migrating bubble synthesis is executed by Polymerase δ (Polδ, with its essential subunit Pol32) and Pif1 helicase. Unless interrupted, DNA synthesis can proceed to the end of the chromosome. Polα-primase (Polα-prim) complex initiates lagging-strand synthesis using the leading strand as a template, leading to conservative inheritance of newly synthesized DNA. Protein names and cartoons are colour-coded. More information in section 1.4. Figure created by the author employing Adobe Illustrator. Introduction 67 It has been demonstrated that POLD3-dependent BIR also takes place in human cells (Costantino et al. 2014; Hu et al. 2019; Sotiriou et al. 2016) and the absence of human PIF1 also derives in shorter BIR extensions (Li et al. 2021). Moreover, and similarly to yeast, RAD51dependent and independent BIR pathways have also been identified (Li et al. 2021; Sotiriou et al. 2016). The particular mode of BIR synthesis is one of the main contributors to the high level of genetic instability associated with this pathway. Asynchrony between leading and lagging strands leads to the accumulation of ssDNA behind the migrating D-loop (Saini et al. 2013; Wilson et al. 2013). This ssDNA can be subject to DNA damage and is a primary source of mutagenesis. ssDNA is more fragile and has limited capacity for error-free repair. Consequently, single-strand lesions often lead to mutations (Saini and Gordenin 2020). It has been reported that the mutation pattern formed along the BIR tract is similar to the kataegis events described in cancer cells (Sakofsky et al. 2014; Elango et al. 2019). Furthermore, in addition to being protected by RPA (Ruff et al. 2016), this ssDNA could also be bound by Rad51 and invade other homologous templates (Sakofsky et al. 2014; Smith et al. 2007). In this context, the anti-recombinase activity of Srs2 becomes essential to dismantle presynaptic filaments assembled on the displaced strand, preventing subsequent invasions. In its absence, complex recombination intermediates are formed that impair the advance of the BIR bubble. Therefore, Srs2 activity is required to allow the progress of the migrating bubble (Elango et al. 2017). A recent work indicates that the template for leading strand synthesis, which also became singlestranded within a D-loop, can also be a source of mutagenesis associated with ssDNA (Osia et al. 2022). This low-fidelity mode of DNA synthesis has also been reported in mammalian cells during mitotic DNA synthesis (MiDAS), which often occurs at difficult-to-replicate genomic regions that remain underreplicated after S-phase and constitute chromosomal fragile sites. MiDAS requires MUS81-EME1, which is proposed to facilitate the process by cleaving the stalled replication fork (Minocherhomji et al. 2015; Ying et al. 2013; Bhowmick et al. 2016; Di Marco et al. 2017; Introduction 68 Naim et al. 2013). Another BIR-like mechanism in mammalian cells is the alternative lengthening of telomeres (ALT), a telomeraseindependent mechanism for telomere maintenance employed by 1015% of cancer cells (Roumelioti et al. 2016; Dilley and Greenberg 2015; Dilley et al. 2016; Bryan et al. 1997). Furthermore, BIR has also been linked with gross chromosomal rearrangements (GCRs). Firstly, the fact that BIR synthesis can proceed until the end of the chromosome can lead to extensive LOH if a nonsister chromatid is used as a template, or in non-reciprocal translocations (NRT) if dispersed repeats are involved (Costantino et al. 2014; Sakofsky and Malkova 2017; Smith et al. 2007). Secondly, it has been shown that BIR can occur through several rounds of strand invasion, in a process called templated switching (Smith et al. 2007; Štafa et al. 2014), which can be a source of GCRs when happening between dispersed repeats (Anand et al. 2014; Ruiz et al. 2009; Pardo and Aguilera 2012). Additionally, premature resolution of BIR intermediates may result in half-crossover (HC) events (Deem et al. 2008; Smith et al. 2009; Wilson et al. 2013), a fusion between the recipient and the donor chromosomes involved in the repair. These events also generate a new one-ended DSB on the donor chromosome, which can either be lost or subsequently involved in the invasion of another intact duplex. This could lead to additional rounds of HC formation, a process termed half-crossover cascades (HCCs) (Vasan et al. 2014; Sakofsky et al. 2014) reminiscent of the NRT cycles reported in human tumours (Sabatier et al. 2005). A mechanism similar to HCCs has been recently proposed to explain another source of genetic instability called multi-invasion mediated-rearrangements (MIR), being both possible contributors to chromothripsis events reported in cancer cells (Piazza et al. 2017; Piazza and Heyer 2019a; Savocco and Piazza 2021). Several studies have pointed out a role for Mus81-Mms4 and Yen1 in the generation of some of these complex chromosomal rearrangements in the context of BIR, especially half-crossover events. The Symington group demonstrated that half-crossovers formed in absence of Polδ are partially Mus81-dependent (Smith et al. 2009). Introduction 69 Pardo and Aguilera (2012) identified a role of Mus81-Mms4 and Yen1 in promoting Pol32-dependent BIR (as previously reported in (Ho et al. 2010)), but also causing translocations and HC presumably by cleaving the D-loop intermediate (Pardo and Aguilera 2012). As previously mentioned, it was shown that premature activation of Yen1 decreases BIR but also lead to an increase in chromosomal loss and half-crossover events (Elango et al. 2017). Moreover, Piazza and collaborators demonstrated a role for the SSEs in the DSB transfer onto the donor in their multi-invasion set-up, by a HC-like mechanism (Piazza et al. 2017), which they further proposed to be the driving mechanism of MIR (Piazza and Heyer 2019a; Savocco and Piazza 2021). 2 OBJECTIVE All available data regarding the involvement of the structure-selective endonucleases in the generation of complex chromosomal rearrangements come primarily from genetic approaches. Thus, the exact substrate(s) and mechanism(s) by which the actions of these enzymes account for the generation of the different recombination outcomes remain unaddressed. The genetic data strongly suggests that it could be the nascent D-loop structure. Despite some biochemical data on how Mus81 processes a synthetic oligonucleotide-based D-loop (Osman et al. 2003; Pepe and West 2014b; Ehmsen and Heyer 2009), nothing is known about the ability of Yen1 to process these structures. Moreover, these intermediates do not exist as naked DNA structures in vivo; instead, they are coated by all the proteins necessary for their formation, which are known to modulate or interfere with their processing by other enzymes (Ruff et al. 2016; Matulova et al. 2009; Fasching et al. 2015; Liu et al. 2017; Chavdarova et al. 2015). Therefore, the main goal of this doctoral thesis is to set-up a biochemical approach to investigate the potential processing of different D-loops by the SSEs Mus81-Mms4 and Yen1. This main objective can be subdivided into the following specific aims: Aim 1: Analyse the processing of different synthetic, oligonucleotide-based D-loop structures by Mus81-Mms4 and Yen1. Aim 2: Establish a system to analyse the cleavage of enzymatically-generated D-loops on plasmid substrates. Aim 3: In vitro reconstitution of the biochemical steps leading to the formation of a specific chromosomal rearrangement, the halfcrossover. 3 METHODOLOGY 3.1 MATERIALS All common laboratory reagents were purchased from Sigma Aldrich or Fisher Scientific at molecular biology grade or superior, unless otherwise stated. 3.1.1 Oligonucleotides The sequence and specifications of the oligonucleotides employed in this study are indicated in Table 1. 3.1.2 DNA substrates The composition of the DNA substrates generated in this study is listed in Table 2 and depicted in Figure 9. 3.1.3 Bacterial proteins Escherichia coli RecA was obtained from New England Biolabs (#M0249S) and E. coli SSB from Thermo Fisher Scientific (#70032Z500G). Methodology 80 cocktail tablets (Roche, #11873580001), and 1x Sigma yeast protease inhibitor cocktail (Sigma, #P8215). After clarification by ultracentrifugation (125000 g, 45 min), the lysate was incubated with 2.5-ml of anti-FLAG M2 agarose beads (Sigma), packed into glass econo-column (Bio-Rad), for 1 h. The proteins were eluted with 3x 1CV of A500 supplemented with 0.5 mg/ml 3xFLAG peptide. For Yen1 dephosphorylation, the combined eluates were supplemented with 1 mM MnCl2 final concentration and incubated with 12000 units (U) of Lambda phosphatase (λ-PPase, New England Biolabs, #P0753) for 1 h at room temperature. Next, imidazole was added to 5 mM final concentration, the proteins were mixed with 400 µl SuperFlow His-Pur Ni-NTA agarose beads (Fisher) in a dispensable column and incubated for 1 h. The beads were washed with 2x 10-CV of A500 buffer containing 5 and 20 mM imidazole, respectively. Finally, Yen1 was eluted with 3x 2-CV of A500 buffer supplemented with 300 mM imidazole. Yen1 was dialyzed against A500, flash-frozen in liquid nitrogen, and stored in 10 µl aliquots at -80ºC (Figure 11). 3.2.1.5 Mus81-Mms4 and Mus81-Mms4ND 3xFLAG-Mus81/10xHIS-2xSTREP-Mms4 and the catalytically inactive mutant 3xFLAG-Mus81/10xHIS-2xSTREPMms4ND (D414A, D415A, (Ehmsen and Heyer 2008)) were expressed and purified as described in (Princz et al. 2017). In short, 80 g of disrupted cells were resuspended in 2 volumes of A500 buffer (40 mM Tris-HCl, pH 7.5, 500 mM NaCl, 20% glycerol, 0.1% NP-40, 1 mM DTT) containing 5 mM NaF, 5 mM H2NaPO4, 5 mM βglycerophosphate, 1 mM PMSF, 1x EDTA-free protease inhibitor cocktail tablets (Roche), and 1x Sigma yeast protease inhibitor cocktail. After clarification by ultracentrifugation (125000 g, 45 min), the lysate was incubated with 2.5-ml of anti-FLAG M2 agarose beads (Sigma), packed into glass econo-column (Bio-Rad), for 1 h. The proteins were eluted with 3x 1-CV of A500 supplemented with 0.5 mg/ml 3xFLAG peptide. Then, imidazole was added up to 5 mM final concentration, the proteins were loaded onto a 400-µl SuperFlow His-Pur Ni-NTA agarose beads (Fisher) packed into a dispensable column and incubated Methodology 81 for 1 h. The beads were washed with 2x 10-CV of A500 buffer containing 5 and 20 mM imidazole, respectively. Finally, Mus81Mms4 was eluted with 3x 2-CV of A500 buffer supplemented with 300 mM imidazole. The proteins were dialyzed against A500, flash-freeze in liquid nitrogen, and stored in 10 µl aliquots at -80ºC. To confirm that Mus81-Mms4 was purified in a phosphorylated stated, 1 µg total protein was dephosphorylated with 100 U λ-PPase for 30 min at 30ºC according to standard procedures and analysed on a Coomassie-stained SDS-PAGE (Figure 11). Figure 11. SDS-PAGE analysis of purified yeast Yen1, Mus81-Mms4, and their nuclease-dead mutants. FTH-tagged Yen1ND (ND), Yen1WT (WT), Mus81-Mms4ND (ND), and Mus8-Mms4WT (WT) (treated or not with lambda-phosphatase (λ-PPase)) were purified, analysed by SDS-PAGE (0.5 µg each), and stained with Coomassie. More information in sections 3.2.1.4 and 3.2.1.5. Molecular weight marker (MWM) is indicated in kDa. 3.2.2 Oligonucleotide purification and annealing into DNA substrates The oligonucleotides employed in this study (Table 1) were obtained from Merck, PAGE-purified, and annealed as detailed in (Carreira et al. 2021). Shortly, labelled and unlabelled oligonucleotides were mixed in a 1:3 ratio, boiled in a water bath, and cooled down to Methodology 82 room temperature overnight. Fully ligated substrates were purified from 10% native polyacrylamide gels in Tris-Borate-EDTA (TBE) buffer (90 mM Boric acid, 90 mM Tris-base, 2 mM EDTA). For radioactive substrates, oligonucleotides were 5’-end-labelled with [γ-32P]-ATP (3000 Ci/mmol, Perkin Elmer) and T4 polynucleotide kinase (PNK; Thermo Fisher, #EK0031), according to standard procedures. For fluorescent substrates 5´- and/or 3´- Cy5 or Cy3 labelled oligonucleotides were used. When indicated, oligonucleotides containing three consecutive phosphorothioate (SP) linkages were employed. Importantly, the unpaired regions of these junctions are composed of heterologous sequences to prevent substrate dissociation by spontaneous branch migration. The strand composition of each substrate is detailed in Table 2 and depicted in Figure 9. 3.2.3 Plasmid purification The plasmid DNA (pBluescript SK (-), pBSK, 2958 bp) was purified using commercial DNA purification kits (GenElute™ HP Plasmid Miniprep kit, Merk, #NA0160), according to standard procedures. As D-loop reactions are sensitive to the topological state of the plasmid, requiring it to be negatively supercoiled (Van Komen et al. 2000; Petukhova et al. 2000), the following modifications were applied to minimize plasmid nicking: pellet resuspension by pipetting and centrifugation steps at 8000 g. 3.2.4 Endonuclease assays with oligonucleotide-based substrates For the experiments employing radioactively labelled oligonucleotide-based D-loops, 10 nM protein was incubated with ~1 nM 5'-32P end-labelled substrate in 25 µl of reaction buffer (Yen1 reaction buffer: 50 mM Tris-HCl pH 7.5, 0.5 mM MgCl2; Mus81-Mms4 reaction buffer: 25 mM Tris-HCl pH 7.5, 3 mM MgCl2, 100 mM NaCl, 0,1 mM DTT, 0.1 µg/ml BSA). In all reactions, enzymes represent 1/10 of the final volume (or enzyme storage buffer in mock reactions). After Methodology 83 10 min of incubation at 30ºC, 10 µl of each reaction were deproteinized by addition of 2 µl STOP solution (1.5% SDS, 10 mg/ml proteinase K (PK)) and incubation at 37ºC for 1 h, followed by incorporation of 0.2 vol 6x Ficoll loading buffer (15% Ficoll-400, 60 mM EDTA, 20 mM Tris-HCl pH 8.0, 0.5% SDS). Another aliquot of 10 µl was mixed with 1 vol of 2x denaturing loading buffer (TBE 1x, 80% formamide) and incubated at 99ºC for 3 min. The radiolabelled products were then separated by PAGE through 10% native or denaturing (7 M urea) gels in TBE buffer. After electrophoresis, gels were dried onto 3MM Whatman chromatography paper (GE Healthcare), exposed to a phosphor screen (Fujifilm), visualized in a Typhoon FLA9500 (GE Healthcare), and quantified by densitometry using ImageQuant software (GE Healthcare). For kinetics experiments employing fluorescently labelled oligonucleotide-based D-loops (D3, D3-SP and nicked D3), 20 nM protein was incubated with 10 nM 5´-Cy5 and 5´-Cy3 labelled substrate. Reactions were performed as described above. Aliquots were taken at the indicated times (0, 0.5, 1, 2, 4, 8, 16, 32, 64, 128 min) and analysed through 10% native and denaturing (7 M urea) PAGE in TBE buffer. Fresh gels were scanned in a Typhoon FLA9500 and quantified by densitometry using ImageQuant software. Each experiment was done three independent times. For experiments employing the long 3´-flap substrate, 2 µl at the indicated final concentration of RPA (25, 50, 100, 200, 400 and 600 nM) was pre-incubated with 40 nM 5´-Cy5-labelled 3´-flap for 10 min at 37ºC in 8 µl Mus81-D-loop buffer (see next section). Then, 1 µl of Mus81-Mms4 at 400 nM final concentration was added and incubated for another 10 min at 30ºC. Reactions were deproteinized by the addition of 2 µl STOP solution (1.5% SDS, 10 mg/ml proteinase K (PK)) and incubation at 37ºC for 1 h, followed by incorporation of 0.2 vol 6x Ficoll loading buffer. Reaction products were analysed through 10% native PAGE in TBE buffer. Fresh gels were scanned in a Typhoon FLA9500 and quantified by densitometry using ImageQuant software. For time-course experiments, 50 nM RPA was pre-incubated with 40 nM 5´-Cy5-labelled 3´-flap for 10 min at 37ºC. Then, 400 nM Mus81- Methodology 84 Mms4 was added and incubated at 30ºC. Aliquots were taken at the indicated times (0, 2, 4, 8, 16, 32 and 64 min). Reactions were performed, deproteinized, and analysed as described above. 3.2.5 Endonuclease assays with plasmid-based D-loops 3.2.5.1 Rad51/Rad54-made D-loops To form the different D-loop structures, 40 nM of the indicated fluorescently labelled molecule (single-stranded or partially singlestranded DNAs, fully complementary or not to a region of pBSK) was incubated with 2 µM Rad51 in D-loop buffer (35 mM Tris-HCl pH 7.5, 50 mM KCl, 1 mM DTT, 2 mM ATP, 20 mM creatine phosphate, 20 μg/ml creatine kinase, and 2.5 mM MgCl2 for Mus81-Mms4 reactions or 1.5 mM MgCl2 for Yen1 reactions) for 5 min at 37ºC. When appropriate, RPA or SSB (600 nM) were added to the nucleoprotein filament and incubated for another 4 min, followed by the addition of Rad54 (300 nM) and incubation at 23ºC for 3 min. Recombination proteins always represent 1/5 of the total final volume in the reaction (or enzyme storage buffer in mock reactions). D-loop formation was initiated by the addition of 2 µl pBSK (1/5 total volume, 64 ng/µl final concentration) to bring the final reaction volume to 10 μl and incubated for 5 min at 23ºC. After D-loop formation, 1 µl at the indicated concentration of Yen1, Mus81-Mms4, the nuclease-dead mutants (always at the highest concentration used for the catalytically active enzyme), or storage buffer (A500) were added to the reaction followed by incubation at 30ºC for the indicated times. Reactions were then deproteinized by the addition of 1.5 µl STOP solution (0.9 % SDS, 1 mg/ml PK) and incubation at 37ºC for 10 min, followed by the addition of 0.2 vol 6x glycerol loading buffer (66% Glycerol, 66 mM EDTA, 11 mM Tris-HCl pH 7.5). Reactions were analysed by electrophoresis in a 0.9% agarose gel in Tris-Acetate-EDTA (TAE) buffer (40 mM Tris Base, 1 mM EDTA, 20 mM acetic acid) developed at 90 V for 30 min (Figure 12). Fresh gels were imaged on a Typhoon FLA9500 and quantified by densitometry using ImageQuant software. Each experiment was done three independent times. Methodology 85 When D1-3SP invading molecule was employed (Figure 23), reactions were performed as previously described, but increasing the final reaction volume to 25 µl. Then, 10 µl of each reaction was deproteinized and analysed on agarose gels (as described). Another 10 µl was mixed with 1 vol of 2x denaturing loading buffer and incubated at 99ºC for 3 min. Reaction products were then separated by 16% PAGE through denaturing (7 M urea) gels in TBE buffer. Fresh gels were scanned on a Typhoon FLA9500 and quantified by densitometry using ImageQuant software. Figure 12. Experimental scheme for endonucleases assays with Rad51/Rad54made D-loop structures. Protein names and cartoons are colour-coded. The dashed line indicates the possible presence of a 5´-single-stranded or doubled-stranded overhang at the D-loop depending on which substrate/oligonucleotide is employed as an invading molecule (see Figure 9). More information in section 3.2.5.1. Figure created by the author employing Adobe Illustrator. 3.2.5.2 RecA-made D-loops RecA D-loops were generated in a similar way and in the same D-loop buffer as stated in the previous section, but with 15 mM MgCl2. 40 nM D1 oligonucleotide was incubated with 4 µM RecA for 5 min at 37ºC, followed by the addition of a single-stranded binding protein (100 mM SSB or RPA) when required. The reaction continued for an additional 5 min at 37ºC. Recombination proteins represent 1/5 of the total final volume in the reaction (or enzyme storage buffer in mock Methodology 86 reactions). D-loop formation was initiated by the incorporation of pBSK (1/5 total volume, 64 ng/µl final concentration), and, after 0.5 min, reactions were diluted 1:10 in Yen1 buffer (50 mM Tris-HCl pH 7.5) for Yen1 reactions or 1:5 in Mus81 buffer (25 mM Tris-HCl pH 7.5, 100 mM NaCl) for Mus81-Mms4 reactions, to reach, in both cases, the optimal MgCl2 concentration for nuclease activity. 1 µl of each SSE (A500 for control reactions) was added at the indicated concentration and incubated for 5 min at 30ºC. Reactions were then deproteinized by addition of 2.5 µl STOP solution (0.1 % SDS, 2 mg/ml PK) and incubation at 37ºC for 2 h. Analysis was done as described above. Each experiment was done three independent times. Figure 13. Experimental scheme for endonucleases assays with RecA-made D1 Dloops. Protein names and cartoons are colour-coded. More information in section 3.2.5.2. Figure created by the author employing Adobe Illustrator. 3.2.5.3 Deproteinized D-loops In deproteinized D-loop experiments, Rad51/Rad54-D1 Dloop formation was carried out as previously described. Before SSE incorporation, half of the reaction was deproteinized by the addition of 0.75 µl STOP solution (0.9 % SDS, 1 mg/ml PK) and incubation at 37ºC for 15 min. To remove PK and SDS, 1 vol of phenol-chloroform was added followed by DNA precipitation with 2 vol of 100% ethanol (EtOH), 0.1 vol of 3 M sodium acetate pH 5.2, and 0.02 vol of 5 mg/ml Methodology 87 glycogen. Pellet was resuspended at the original volume in which the D-loop reaction had taken place. Then, these deproteinized D-loops were employed in nuclease reactions that were carried out and analysed as described above. After being scanned to visualize the fluorescent products, gels were stained with ethidium bromide (5 µl of 10 mg/ml ethidium bromide (EtBr) solution in 100 ml TAE 1X) and imaged in a Gel Doc XR + System using the Image Lab software (Bio-Rad). Each experiment was carried out three independent times. 3.2.5.4 D-loops with a stable presynaptic filament For experiments employing the non-hydrolysable ATP analogue ATPγS, Rad51/Rad54-D1 D-loops were generated as described above, with the following modifications: 40 nM of D1 oligonucleotide was incubated with Rad51 (2 µM) in D-loop buffer without the ATP regeneration system (creatine phosphate and creatine kinase) and with 2.5 mM ATPγS for 9 min at 37ºC, followed by addition of Rad54 (300 nM), 2.5 mM ATP, and incubation at 23ºC for 3 min. D-loop formation was initiated by the incorporation of pBSK (640 ng) and incubated for another 10 min at 23ºC. After D-loop formation, reactions continued and were analysed as previously described. In control reactions with ATP, D-loop structures were produced in the absence of the ATP regeneration system and with 5 mM ATP. These experiments were carried out in triplicate. 3.2.6 Mapping SSE cleavage sites on enzymatically-generated D-loops To map the cleavage sites of the nucleases in the invading molecule, Rad51/Rad54-D-loop formation and treatment with the nucleases were done as previously described but increasing the reaction volume to 25 µl. After nuclease incubation, 10 µl of each reaction were deproteinized and analysed on agarose gels as stated above, to monitor D-loop formation and nuclease activity. Another aliquot of 10 µl was mixed with 1 vol of 2x denaturing loading buffer and incubated at 99ºC Methodology 88 for 3 min. Reaction products were then separated by denaturing PAGE (7 M urea) in TBE buffer. Fresh gels were scanned on a Typhoon FLA9500 and quantified by densitometry using ImageQuant software. To analyse the incision created by the nucleases in the donor molecule (pBSK), after Rad51/Rad54 D-loop formation (20 µl final volume) and cleavage with the indicated nuclease (as described), the plasmid was cut with 10 U of BspHI (New England Biolabs, #R0517S) and incubated at 37ºC for 20 min followed by dephosphorylation with 1 U Shrimp Alkaline Phosphatase (rSAP, New England Biolabs, #M0371S) and incubation for another 20 min. Then, half of each reaction was deproteinized and analysed on agarose gels, as outlined above, to check D-loop formation and nuclease activity. The other half was denatured at 99ºC for 3 min followed by labelling using 1 µl [γ32P]-ATP (3000 Ci/mmol) and 10 U T4 PNK for 1 h at 37ºC. The nonincorporated isotope was eliminated using G-25 columns (GE Healthcare) and the eluted DNA was precipitated with 2 vol of EtOH, 0.1 vol of 3 M sodium acetate pH 5.2 and 0.02 vol of 5 mg/ml glycogen. The pellet was resuspended in 10 µl 2x denaturing loading buffer and samples were analysed through denaturing PAGE, exposed to a phosphor screen, and visualized in a Typhoon FLA9500 (see scheme in Figure 35). 3.2.6.1 Mapping experiments using Sanger sequencing As a proof-of-concept, the ability to detect nicks on pBSK employing Sanger sequencing was tested by analysing the singlestranded break generated by the commercial nickase NtBspQ1 (New England Biolabs, #R0644S) on pBSK. It generates a nick after the cognate DNA sequence 5´-GCTCTTC-3´, which is present as a single copy in pBSK, from 1030 to 1037 bp (Figure 14A). Following manufacturer recommendations, 10 µg of pBSK was either nicked with NtBspQ1 or linearized with LguI (Thermo Fisher, #ER1931), a restriction enzyme that cuts the same strand at the same position as NtBspQ1 (Figure 14A). Both reactions were purified by DNA precipitation with 2 vol of EtOH and 0.1 vol of 3 M sodium acetate pH Methodology 89 5.2 (Figure 14B). Then, different proportions of nicked pBSK (50, 20, 10, 5, 2, and 1%) were mixed with supercoiled pBSK and 1 µg of supercoiled (SC), nicked (N), linear (L), and each mixture was sequenced using forward (NICK-FW, from 849 to 872 bp) and reverse (NICK-RV, from 1183 to 1202 bp) primers (Table 1). Sanger sequencing reactions were outsourced to Stabvida (STAB VIDA, LDA. Caparica, Portugal). The sequencing protocol employs Taq polymerase, which introduces an untemplated adenosine nucleotide on the 3´-end of the newly synthesized strands (A-tailing activity). This serves to localize the incision point made by the enzymes. Electropherograms were visualized and searched for untemplated adenosines using SnapGene viewer software (Figure 14C and D). To map the cleavages sites of Mus81-Mms4 and Yen1 on plasmidbased D-loops using Sanger sequencing, after Rad51/Rad54 D-loop formation in 25 µl final volume and incubation with the indicated nuclease (as described), each reaction was deproteinized by the addition of 3.75 µl STOP solution (0.9 % SDS, 1 mg/ml PK) and incubation at 37ºC for 30 min. Then, 5 µl were analysed on agarose gels to check Dloop formation and nuclease activity. 75 µl H2O was added to the rest followed by 1 vol of phenol-chloroform. After phenol-chloroform extraction, DNA precipitation was carried out with 2 vol of EtOH, 0.1 vol of 3 M sodium acetate pH 5.2, and 0.02 vol of 5 mg/ml glycogen. The pellet was resuspended in 22 µl 10 mM Tris-HCl pH 8.0. Finally, 10 µl was employed for sequencing using the forward primer (SEQFW, from 1767 to 1784 bp) and the other 10 µl for sequencing with the reverse primer (SEQ-RV, from 2162 to 2177 bp) (Table 1) in Stabvida laboratories (Figure 15). Results 96 incisions observed were dependent on Mus81 catalytic activity (Figure 16). In the case of Yen1, its incisions always lead to the disassembling of the structure: two DNA products when D3 synthetic D-loop is labelled on strand 1 or 2; three products when oligo 3 is marked, and complete loss of labelling of oligo 4 (Figure 16A, cleavage products are depicted on the right). The mapping of Yen1 incisions showed that it can cleave the four strands in the D3 D-loop, but with very different efficiency. The predominant cleavage appears at the displaced strand (oligo 2), from nt 51 to 53, with the main incision at position 51, 1 nt away to the 3´-side of the second branching point (Figures 16B and C). It can also nick the invading molecule (oligo 3), again, one, two, or three nt away to the first branching point. We also detected some nucleolytic activity on the 5´-end of the strand complementary to the invading oligo (oligo 4), which explains the loss of labelling shown in the native gel. Finally, minor incisions were observed on oligo 1, at positions 51 and 52, at the 3´-side of the first branching point (Figures 16B and C). None of these activities were detected when the nuclease-dead mutant Yen1ND (Y-ND) (Blanco et al. 2014) was employed. Results 97 Figure 16. Yen1 and Mus81 cleave a D3 oligonucleotide-based D-loop structure. (A) D3 synthetic D-loop (D3) was 5’-end-labelled with 32P (asterisk) on each strand (grey) and incubated with 10 nM Yen1WT (Yen1), Yen1ND (Y-ND), Mus81WT (Mus81), or Mus81ND (M-ND) for 10 min at 30ºC. (-) indicates no enzyme. Products were analysed by 10% native PAGE and phosphorimaging. Schemes for the substrate and cleavage products are depicted on the right. (B) Same reactions as A, but cleavage products were analysed by 10% denaturing PAGE. A mixture of 5'-32P end-labelled oligos of defined length was used as a molecular weight marker (MWM). Please note that MWM bands differ between lanes. (C) Schematic representation of Yen1 (purple) and Mus81 (green) incision sites on the oligonucleotide-based D3 D-loop. Arrowhead and number size are proportional to the relative efficiency of the cleavage. Results 98 Regarding Yen1 nucleolytic activity on the 5´-end of the oligo 4, we wondered if it could be dependent on the proximity of the 5´- phosphate group to the invasion point. To investigate this possibility, we generated the same D3 structure but containing a shorter oligo 4 (oligo 4s, Table 1), in which its 5´-end is located 5 nt away from the invasion point (sD3 synthetic D-loop). Analysis through native PAGE of Yen1 incisions on this structure revealed that, when oligo 4s is labelled, instead of loss of labelling, a short duplex product is observed, similar to that produced when oligo 3 is marked (Figure 17A). Further analysis in denaturing conditions confirmed that Yen1 activity on this strand is absent when this substrate is employed (Figure 17B). Additionally, Mus81 incisions in this structure are now closer to the invasion point, from nt 45 to 49, with one predominant incision at position 47, 3 nt away from first branching point (Figures 17B and C). Moreover, although very faint, Mus81 incisions at the displaced strand (oligo 2) could also be detected, 2 to 4 nt to the 3´-side of the first branching point (Figures 17B and C). Results 99 (on the previous page) Figure 17. Cleavage of the sD3 oligonucleotide-based Dloop by Yen1 and Mus81. (A) Short D3 synthetic D-loop (sD3) was 5’-end-labelled with 32P (asterisk) on each strand (grey) and incubated with 10 nM Yen1WT (Yen1), Yen1ND (Y-ND), Mus81WT (Mus81), or Mus81ND (M-ND) for 10 min at 30ºC. (-) indicates no enzyme. Products were analysed by 10% native PAGE and phosphorimaging. Schemes for the substrate and cleavage products are depicted on the right. (B) Same reactions as A, but cleavage products were analysed by 10% denaturing PAGE. A mixture of 5'-32P end-labelled oligos of defined length was used as a molecular weight marker (MWM). Please note that MWM bands differ between lanes. (C) Schematic representation of Yen1 (purple) and Mus81 (green) incision sites on the sD3 D-loop. Arrowhead and number size are proportional to the relative efficiency of the cleavage. Altogether, the main incision produced by Mus81 in combination with Yen1 cut at the end of the displaced strand on the D3 D-loop are compatible with the generation of a half-crossover outcome in the context of BIR repair. Moreover, Yen1 cleavage at the invading strand of a nascent D-loop structure could account for the generation of chromosomal loss events in a BIR scenario. Additionally, we generated two other D-loop structures: D2 and D1 (Figures 9, 18, and 19). The D2 D-loop is similar to D3, but employs a fully ssDNA invading oligo, mimicking the situation after D-loop bubble migration (Figure 8). Following the same experimental scheme as before, we found that while Yen cuts this structure at equivalent positions with respect to D3, Mus81 is unable to process it (Figure 18), in agreement with previous results employing its Schizosaccharomyces pombe ortholog Mus81-Eme1 (Osman et al. 2003). Results 100 Figure 18. Cleavage of the D2 oligonucleotide-based D-loop by Yen1. (A) D2 synthetic D-loop (D2) was 5’-end-labelled with 32P (asterisk) on each strand (grey) and incubated with 10 nM Yen1WT (Yen1), Yen1ND (Y-ND), Mus81WT (Mus81), or Mus81ND (M-ND) for 10 min at 30ºC. (-) indicates no enzyme. Products were analysed by 10% native PAGE and phosphorimaging. Schemes for the substrate and cleavage products are depicted on the right. (B) Same reactions as A, but cleavage products were analysed by 10% denaturing PAGE. A mixture of 5'-32P end-labelled oligos of defined length was used as a molecular weight marker (MWM). Please note that MWM bands differ between lanes. (C) Schematic representation of Yen1 (purple) incision sites on the D2 D-loop. Arrowhead and number size are proportional to the relative efficiency of the cleavage. Finally, we analysed the cleavage of the oligonucleotide-based D1 D-loop, a structure where the invading oligo does not present overhangs and that has been widely used to study these recombination intermediates in vitro (Figure 19) (Van Komen et al. 2006; Prakash et al. 2009; Fasching et al. 2015; Wilson et al. 2013; Buzovetsky et al. 2017; Wright and Heyer 2014; Sebesta et al. 2011). Analysis in native gels after nuclease incubation showed that both enzymes can cleave this structure, as in both cases a slower migrating band appears (Figure Results 101 19A). Analysis through denaturing PAGE indicated that, in this case, both enzymes introduce their incisions on the displaced strand. Mus81 nicked it at the first branching point, mainly at position 25, with 2 minor incisions at positions 26 and 27, while Yen1 introduced a single incision 1 nt to the 3´-side of the second branching point (Figures 19B and C). Additionally, Yen1 displayed a seemingly mild nucleolytic activity on the 5’-end of the invading molecule (Figures 19A, B, and C). Figure 19. Both Yen1 and Mus81 cleave a D1 oligonucleotide-based D-loop structure. (A) D1 synthetic D-loop (D1) was 5’-end-labelled with 32P (asterisk) on each strand (grey) and incubated with 10 nM Yen1WT (Yen1), Yen1ND (Y-ND), Mus81WT (Mus81), or Mus81ND (M-ND) for 10 min at 30ºC. (-) indicates no enzyme. Products were analysed by 10% native PAGE and phosphorimaging. Schemes for the substrate and cleavage products are depicted on the right. (B) Same reactions as A, but cleavage products were analysed by 10% denaturing PAGE. A mixture of 5'-32P endlabelled oligos of defined length was used as a molecular weight marker (MWM). Please note that MWM bands differ between lanes. (C) Schematic representation of Yen1 (purple) and Mus81 (green) incision sites on the D1 D-loop. Arrowhead and number size are proportional to the relative efficiency of the cleavage. Results 102 Collectively, these results showed that both Yen1 and Mus81 are able to process oligonucleotide-based D-loop structures according to their polarities. While Mus81 cuts the D3 D-loop at the same positions as described for other orthologs, it processes the D1 D-loop at the opposite strand. In the case of Yen1, it can efficiently cleave the three different structures. 4.2 YEN1 INCISIONS ON THE D3 D-LOOP ARE INDEPENDENT Yen1 is able to cut both the invading and the displaced strands of the D3 D-loop (Figures 16B, C, and 20D). Cleavage of the displaced strand by Yen1 combined with Mus81 incision at the donor strand would be compatible with the generation of a half-crossover, while Yen1 excision at the invading strand should suffice to produce a chromosomal loss event. Therefore, we wanted to determine if the two predominant Yen1 incisions in the D3 D-loop were co-dependent or if, alternatively, they could be uncoupled. To do so, we performed time-course analyses of Yen1 cleavage employing a D3 structure with a Cy3 5´-labelled displaced strand and a Cy5 5´-labelled invading oligo. Reaction products were analysed through native and denaturing PAGE (see Methodology section 3.2.4) (Figure 20). (on the next page) Figure 20. The two incisions produced by Yen1 on a D3 D-loop display different kinetics. (A) 10 nM synthetic D3 D-loop or (B) D3 containing 3 hydrolysis-resistant phosphorothioate linkages (D3-SP, SP linkages in oligo 2 between nt 50-51-52-53, indicated with a pink box) were incubated with 20 nM Yen1 the indicated times at 30ºC. DNA substrates were 5´-labelled with Cy3 in oligo 2 and 5´- labelled with Cy5 in oligo 3. Reaction products were analysed by 10% native and denaturing PAGE and scanned at 532 nm (Cy3) or 635 nm (Cy5) in a Typhoon FLA9500. Reaction products are depicted on the right. A representative gel is shown. (C) Quantification of oligonucleotide cleavage from denaturing PAGEs showed in A (solid lines) and B (dotted lines). Data represented as mean values ± standard deviation (SD) (n = 3). (D) Schematic representation of fluorescent D3 synthetic D-loop structure. Yen1 incision sites are indicated with purple arrowheads. Results 103 Results 104 Both native and denaturing gels indicate that Yen1 cleaves the displaced strand faster than the invading oligo (Figure 20A). Nicking of the displaced strand can be observed at 0.5 min of incubation, reaching the plateau after 4 min, while processing of the invading oligo becomes detectable only after 2 min of reaction, reaching the plateau at 1 hour (Figures 20A and C). This observation raises the question of whether the processing of the invading oligo depends on the prior excision of the displaced strand or both incisions can occur independently. To study this possibility, we generated a D3 D-loop with 3 hydrolysis-resistant phosphorothioate (SP) linkages between nt 50-53 in oligo 2 to inhibit Yen1 cleavage at the displaced strand (Figure 20B). Using this strategy, we could prevent ~50% of Yen1 activity on the displaced strand (only one of the two possible diastereomers at each SPlinkage prevents the phosphodiesterase activity) (Figures 20B and C). We observed that, on top of the ~50% reduction in the cleavage of the displaced strand, the incision on the invading oligo was reduced, but only a 25% (Figures 20A-C). As the strategy employing phosphorothioate linkages to prevent Yen1 cleavage on the displaced strand was not fully conclusive, we reasoned that if the cleavage at the invading oligo depends on the previous nicking at the displaced strand, then, pre-nicking this structure should accelerate invading strand cleavage. To check this hypothesis, we generated a nicked D3 structure (Figure 21C) and performed similar time-courses as described. As shown in Figure 21, the cleavage kinetics of the invading strand in a nicked D3 D-loop are very similar to those of the intact substrate. Taken together, these results indicate that the two incisions produced by Yen1 in the D3 D-loop are independent and suggest that the reduction seen when phosphorothioate linkages were employed could be due to conformational changes induced by the modified nucleotides. Results 105 Figure 21. Yen1 cleaves the invading strand of a nicked D3 D-loop with similar kinetics to those of the intact substrate. (A) 10 nM nicked D3 substrate was incubated with 20 nM Yen1 the indicated times at 30ºC. The substrate is 5´-labelled with Cy3 in oligo 2s and 5´-labelled with Cy5 in oligo 3. Reaction products were analysed by 10% denaturing PAGE and scanned at 532 nm (Cy3) or 635 nm (Cy5) in a Typhoon FLA9500. A representative gel is shown. (B) Quantification of denaturing PAGEs showed in A (dotted line) compared to intact D3 D-loop (solid line, data from Figure 20). Data represented as mean values ± SD (n = 3). (C) Schematic representation of fluorescent nicked D3 synthetic D-loop structure. Yen1 incision site is indicated with a purple arrowhead. Results 112 Since the D3 D-loop is a more physiological structure than the D1 D-loop, we wondered if these endonucleases could favour the processing of the D3 structure over the D1. To check this hypothesis, we performed a kinetic analysis of Yen1 and Mus81 cleavage on enzymatically-made D1 and D3 structures (Figure 26). As D-loop formation is a dynamic process and Rad54 was shown to dissociate Dloops at longer incubation times (Wright and Heyer 2014), the same experiment was performed with each nuclease-dead mutant (Figure 26, dotted lines). As shown in Figure 26, both Yen1 and Mus81 process D1 D-loops faster than D3s. Control experiments performed with the nuclease-dead mutants corroborate that the disappearance of the D-loop bands depends on the catalytic activity of both enzymes and not due to Rad54-mediated or spontaneous disassembling of the structure (Figure 26). (on the next page) Figure 26. Kinetic analysis of Yen1 and Mus81 processing of enzymatically-made D1 and D3 D-loops. (A) Experimental scheme of Rad51/Rad54mediated D-loop reactions. 40 nM end-labelled D1 or D3 molecule was incubated with 2 µM Rad51 at 37ºC for 9 min. 300 nM Rad54 was added to the reaction and incubated for 3 min at 23ºC. Then, supercoiled pBSK (640 ng) was added and incubated for another 5 min. 400 nM nuclease was incorporated and incubated at 30ºC. Aliquots were taken at 0, 0.5, 1, 2, 4, 8, 16, 32 and 64 min. After that, reactions were deproteinized and analysed on agarose gels. (B) Representative agarose gel from reactions with Yen1 (left) of Yen1ND (ND, right). (-) indicates no nuclease. Gels were scanned at 473 nm (6FAM) for D1 D-loops (top) or 532 nm (Cy3) for D3 D-loops (bottom) in a Typhoon FLA9500. (C) Quantification of D-loop cleavage by the SSEs. D-loops were normalized by setting the initial D-loop yield as 100%. Plotted are means ± SD from n = 3. Blue line: D1 D-loop; yellow line: D3 D-loop. Solid line: catalytic active protein; dotted line: nuclease-dead (ND) protein. (D) and (E) same as B and C, respectively, but using Mus81 and Mus81ND. Results 113 Results 114 In sum, these experiments demonstrate that the structure-selective endonucleases Yen1 and Mus81 can process plasmid-based D-loops generated with the proteins involved in their formation in vivo. 4.4 MUS81 AND YEN1 CLEAVE ENZYMATICALLY-MADE D-LOOPS COATED BY RPA Another protein implicated in D-loop formation in cells is the heterotrimeric ssDNA binding protein Replication Protein A (RPA). In vitro, RPA stabilises D-loops by binding to the displaced strand and preventing its reannealing to the template one (Eggler et al. 2002; Van Komen et al. 2002). To check if RPA could influence D-loop cleavage by the SSEs, we generated D1 D-loops in the presence or absence of RPA and employed them as substrates for the endonucleases, as described above (Figure 27A). As shown in Figure 27B, Yen1 is equally efficient in cleaving the D1 D-loop with or without RPA. For Mus81, rather than preventing it, RPA seems to enhance Mus81 cleavage of the D1 structure mildly (Figure 27C). Similar experiments with the D3 Dloop confirmed the absence of effect of RPA in Yen1 activity and showed an even milder effect in Mus81 activity on this structure (Figure 27D and E). (on the next page) Figure 27. Effect of RPA on D-loop cleavage by SSEs. (A) Experimental scheme of Rad51/Rad54-mediated D-loop reactions with RPA. 40 nM 5´-end-labelled D1 or D3 molecules was incubated with 2 µM Rad51 at 37ºC for 5 min. When appropriate, 600 nM RPA was added and incubated for 4 min at 37ºC. Then, 300 nM Rad54 was added to the reaction and incubated for 3 min at 23ºC followed by supercoiled pBSK (640 ng) incorporation and incubation at 23ºC for 5 min. The indicated SSE concentration was added, and reactions were incubated at 30ºC for 15 min for D1 D-loop or 60 min for D3 D-loops. (B) Representative agarose gel from reactions with Yen1 and enzymatic D1 D-loop with RPA. Gels were scanned at 473 nm (6FAM) in a Typhoon FLA9500. Graphs represent the quantification of Dloop cleavage by Yen1. D-loops were normalized by setting the initial D-loop yield as 100%. Plotted are means ± SD from n = 3. (C) Same as B but for Mus81. (D) and (E) Same as B and C, respectively, but for D3 D-loops. Gels were scanned at 532 nm (Cy3) in a Typhoon FLA9500. Results 115 We envisage two options to explain RPA effect on Mus81 cleavage: 1) RPA directly stimulates Mus81 catalytic activity, and the differences observed are due to technical variability, or 2), the presence of RPA makes the substrate more accessible for Mus81. In this sense, taking into account the cleavage position of Mus81 on the equivalent Results 116 oligonucleotide-based structures, the effect of RPA could be different depending on where Mus81 is processing these branched molecules. To determine if Mus81 is stimulated by RPA, we employed one of Mus81 preferred substrates, a 3´-flap (Ehmsen and Heyer 2008). We generated this structure with a long 80 nt ssDNA tail, the same length as the displaced strand in our D-loops (see Tables 1 and 2 and Figure 9). This 3´-flap was pre-incubated with increasing concentrations of RPA for 10 min before adding Mus81 to the reaction. Then, reaction products were analysed through native PAGE (see Methodology section 3.2.4 and Figure 28). As RPA concentration increases, Mus81 becomes less efficient in cleaving the 3´-flap structure. Importantly, when the same concentration of RPA as in D-loop reactions was used (600 nM), almost no cleavage was detected. This indicates that RPA not only does not stimulate Mus81 activity on one of its preferred substrates but displays a protective effect against its nuclease activity. Figure 28. Effect of RPA on Mus81 cleavage on a 3´-flap. 40 nM 3´-flap was preincubated with the indicated RPA concentrations for 10 min at 37ºC. 400 nM Mus81 was added to the reaction and incubated for 10 min at 30ºC. Reactions were deproteinized, analysed on a 10% native PAGE, and scanned at 635 nm (Cy5) in a Typhoon FLA9500. (-) indicates no enzyme. Cleavage products are depicted on the right. Results 117 As in the previous experiment there is a very minor increase in Mus81 cleavage when low RPA concentrations were used, we performed a time-course experiment to check Mus81 activity in the presence or absence of 50 nM RPA. As shown in Figure 29, there is no difference in the processing of 3´-flaps by Mus81 when RPA is present. Figure 29. Time-course analysis of the effect of RPA on Mus81 cleavage. 40 nM 3´-flap was preincubated with 0 or 50 nM RPA for 10 min at 37ºC. Then, 0 or 400 nM Mus81 was incorporated and incubated at 30ºC. Aliquots were taken at 0, 2, 4, 8, 16, 32 and 64 min. After that, reactions were deproteinized, analysed on 10% native PAGE, and scanned at 635 nm (Cy5) in a Typhoon FLA9500. (-) indicates no enzyme. Cleavage products are depicted on the right. Graphics represent quantification of native gels. Since we did not detect any direct RPA stimulation of Mus81 catalytic activity, we next addressed if the presence of other singlestranded binding protein on the D1 D-loop could remodel the substrate to make it more accessible to Mus81. To investigate this possibility, we performed similar D-loop cleavage assays, but employing the bacterial ortholog of RPA, E. coli SSB. As shown in Figure 30, neither Mus81 nor Yen1 ability to process D1 D-loops was altered by the presence of SSB. Results 118 Figure 30. Effect of SSB on the cleavage of D1 D-loops by SSEs. (A) Experimental scheme of Rad51/Rad54-mediated D-loop reactions with SSB. 40 nM 5´-6FAMlabelled D1 oligo was incubated with 2 µM Rad51 at 37ºC for 5 min. When appropriate, 600 nM SSB was added and incubated for 4 min at 37ºC. Then, 300 nM Rad54 was added to the reaction and incubated for 3 min at 23ºC followed by supercoiled pBSK (640 ng) incorporation and incubation at 23ºC for 5 min. The indicated SSE concentration was added, and reactions were incubated for 15 min at 30ºC. After that, reactions were deproteinized and analysed on agarose gels. (B) Representative agarose gel from D1 D-loop processing reactions with SSB and Yen1 (top) or Mus81 (bottom). Gels were scanned at 473 nm (6FAM) in a Typhoon FLA9500. Quantification of D-loop cleavage by the SSEs is shown on the right. D-loops were normalized by setting the initial D-loop yield as 100%. Plotted are means ± SD from n = 3. Finally, to check the possibility that the stimulatory effect of RPA might rely on the presence of the other yeast proteins involved in Dloop formation, we generated D1 D-loops with the Rad51 bacterial ortholog, RecA, and then, we incubated them with SSB or RPA, prior to nuclease incorporation (see Methodology section 3.2.5.2). These experiments were performed in collaboration with María Crugeiras (DNA Repair and Genome Integrity lab, CiMUS, USC). While both Mus81 and Yen1 were able to process D1 D-loops generated with Results 119 RecA, we did not observe overt differences when SSB or RPA were employed (Figure 31). Nevertheless, it is worth mentioning that for RecA D-loops, reactions had to be diluted 1:5 to reach a suitable MgCl2 concentration for Mus81 activity (i.e., there is 5 times less D-loop in the reaction). However, in order to detect cleavage, we had to employ the same Mus81 concentration as for the Rad51/Rad54 D-loops, in which the reaction is not diluted. Contrarily, Yen1 seems to process RecA and Rad51/Rad54-made D1 D-loops with similar efficacy, as RecA reactions were diluted 10 times, but we also used 10-fold less Yen1 protein, suggesting that RecA D-loops are more resistant to Mus81 cleavage than Rad51/Rad54-mediated ones. Results 120 Results 121 (on the previous page) Figure 31. Yen1 and Mus81 can process RecA-mediated D1 D-loops. (A) Experimental scheme of RecA-mediated D1 D-loop reactions with SSB or RPA. 40 nM 5´-6FAM -labelled D1 oligonucleotide was incubated with 4 µM RecA at 37ºC for 5 min. When appropriate, 100 nM SSB or RPA was added and incubated for 5 min at 37ºC. Then, supercoiled pBSK (640 ng) was added and incubated at 37ºC for 0.5 min in reactions with RecA, and 5 min in reactions with RecA + SSB or RPA. Dloop reactions were diluted 1:10 prior to Yen1 incorporation or 1:5 prior to Mus81 incorporation. The indicated SSE concentration was added, and reactions were incubated at 30ºC for 5 min. (B) Representative agarose gels from reactions with Yen1 and enzymatic RecA D1 D-loop alone (left), with SSB (middle), or with RPA (right). (-) indicates no enzyme. Gels were scanned at 473 nm (6FAM) in a Typhoon FLA9500. Quantification of D-loop cleavage by Yen1 is shown at the bottom. D-loops were normalized by setting the initial D-loop yield as 100%. Plotted are means ± SD from n = 3. (C) Same as B but with Mus81. All in all, these results indicate that neither the presence of RPA nor its bacterial ortholog SSB prevents nuclease action on the recombination intermediates generated in vitro. Furthermore, we have shown that, although with very different efficiency, both Yen1 and Mus81 can process RecA-mediated D-loop structures. 4.5 D-LOOP PROTEINS IMPAIR NUCLEASE CLEAVAGE AND RAD51 TURNOVER FACILITATES SSE ACTIVITY Given our previous results, we wanted to address if the same proteins required for D-loop formation could modulate the accessibility of the nucleases to the recombination intermediates. Hence, we tested the effect of deproteinization of Rad51/Rad54-coated D1 D-loops before Yen1 or Mus81 treatment (see Methodology section 3.2.5.3 and Figure 32A). Under conditions of identical nuclease concentration, total amount of DNA, and proportion of D-loop formation, both Yen1 and Mus81 displayed faster cleavage kinetics with deproteinized D-loops than with coated ones (Figures 32B and C). This indicates that Rad51 and Rad54 restrict the ability of the SSEs to access and/or cleave these structures. Results 128 Results 129 (on the previous page) Figure 35. Mapping of the cleavage sites of Yen1 and Mus81 on Rad51/Rad54-mediated D3 D-loops. (A) Scheme of the mapping strategy for the D3 structure. (B) To identify cuts on the invading molecule, after D3 D-loop formation, increasing concentrations of Yen1 (50, 100, 200, 400 nM) or 400 nM Yen1ND (ND) were added to the reaction and incubated for 1 h at 30ºC. (-) indicates no enzyme. Reaction products were analysed by 12% denaturing PAGE and scanned at 635 nm (Cy5) and 532 nm (Cy3) in a Typhoon FLA9500. A mixture of 5'-6FAM-endlabelled oligos of defined length was used as markers. (C) Same as B, but employing increasing concentrations of Mus81 (50, 100, 200, 400 nM). (D) To identify incisions on the plasmid, after D3 D-loop formation, 400 nM Yen1 was added to the reaction and incubated for 1 h at 30ºC. Then, pBSK was cut with 10 U BspHI at 37ºC for 20 min, followed by treatment with rSAP for another 20 min at 37ºC and heat inactivation at 99ºC for 5 min. Reaction products were labelled using T4 PNK and 32Pγ-ATP and analysed by 6% denaturing PAGE followed by phosphorimaging. A mixture of DNA fragments was radioactively labelled and used as marker. Arrows indicate products of expected sizes. (E) Same as D but employing 400 nM Mus81. (F) Schematic representation of Yen1 (purple) and Mus81 (green) incisions on the plasmid-based D3 D-loop. Open arrowheads indicate the other possible position for both cleavage sites. These mapping experiments were performed with the minimum set of proteins required to reconstitute D-loop formation in vitro (Rad51 and Rad54). The same experiments were performed including RPA to check if the presence of this single-stranded binding protein could modify any of the cleavage site positions produced by the SSEs. No changes were detected at any of the cleaving sites (Figure 36). Results 130 Figure 36. Mapping of the cleavage sites of Yen1 and Mus81 on RPA-coated D3 Dloops. All reactions in the presence of RPA were carried out as described in Figure 35. Same labels as in Figure 35. Results 131 A caveat about the mapping strategy to detect the incisions on the plasmid is that it cannot distinguish which strand is being cut, as in combination with BspHI, ssDNA fragments of the same size would be observed in denaturing gels. Therefore, we complemented these experiments with an alternative mapping approach by taking advantage of the Taq polymerase untemplated addition of one adenosine at the 3´- end of the newly synthesized strands in sequencing reactions (see Methodology section 3.2.6.1 and Figures 14 and 15). Therefore, when DNA from endonuclease-treated D-loop reactions was sequenced, the electropherograms reveal novel A peaks (or A peaks of increased intensity) only in the nicked strand, but not in the complementary (Figure 37A). Therefore, this strategy allows us to determine which strand is being cut by Yen1 and Mus81 with single nucleotide resolution. As expected, Yen1 cleaves the D3 D-loop at the end of the displaced strand, while Mus81 incises the template strand 4-5 nt away from the start of the homology region (Figure 37D). Experiments employing RPA confirmed that its presence at the D-loop does not alter nuclease cleavage sites (Figure 37D). Importantly, all control reactions without Rad51/Rad54 and/or without nuclease shown in Figures 37B and C were sequenced and analysed, showing no untemplated adenosine incorporation. Only nuclease-dead controls are shown for clarity in the Figure. As the region where Yen1 introduces its incisions on this D3 Dloop has several adenosines that obscure the precise identification of its cleavage sites, control experiments employing a different D3 D-loop structure (D3.2) were performed (Figure 38). In this case, the homologous region between the invading strand and pBSK was changed (from nt 1932-2012 to nt 1952-2022 of pBSK, see Tables 1 and 2), which resulted in the shifting of the A peaks to the equivalent positions as those described for the D3 D-loop (compare Figures 37 and 38). Results 132 Results 133 (on the previous page) Figure 37. Mapping by sequencing of Yen1 and Mus81 incisions on D3 D-loops. (A) Scheme of the mapping strategy employing DNA sequencing. After D-loop formation, 400 nM Yen1 or Mus81 was added and incubated for 1 hour at 30ºC followed by deproteinization. A sample was taken for analysis on agarose gels. The rest was purified by phenol-chloroform extraction followed by EtOH precipitation. Reaction products were sequenced with Taq polymerase employing forward and reverse primers. (B) Agarose gel analysis of Mus81 reactions prior to DNA purification and sequencing. Fresh gel was scanned at 532 nm (Cy3) in a Typhoon FLA9500. (-) indicates no enzyme. Labels: R51, Rad51; R54, Rad54. (C) Same as B but for Yen1 reactions. (D) Graphical representation of the incisions mapped on plasmid-based D3 D-loops. The nucleotide sequence near the branch point is shown. Electropherograms derived from sequencing using forward and reverse primers are depicted at the bottom. Left: reactions with Mus81. Green arrows indicate Mus81 cleavage sites. Adenine incorporation after the cleavage point is indicated with a green dotted box. Nuclease-dead mutant controls are shown (ND). Right: Reactions with Yen1. Purple arrows indicate Yen1 cleavage sites and adenine incorporation is pointed out with a purple dotted box. Nuclease-dead mutant controls are shown (ND). Figure 38. Mapping by sequencing of Yen1 and Mus81 incisions on the derivative D3.2 D-loop. All reactions were carried out as described in Figure 37. Same labels as in Figure 37. Note that both branched points are different from the previous ones. Only electropherograms of the region of interest are shown. Results 134 The same rationale and experimental set-up were applied to determine the incision sites created by Yen1 on a plasmid-based D2 Dloop (Figure 39A). On this structure, Yen1 activity is reminiscent of that on D3, showing multiple cuts on the invading strand and cleaving the displaced strand at the bubble distal end (Figures 39B, C, and F). As in the case of D3, RPA presence does not alter Yen1 cleavage positions (Figures 39D and E). (on the next page) Figure 39. Mapping of the cleavage sites produced by Yen1 on enzymatically-made D2 D-loops. (A) Scheme of the mapping strategy for the D2 structure. (B) To identify the cuts on the invading molecule, after D2 D-loop formation, increasing concentrations of Yen1 (50, 100, 200, 400 nM) or 400 nM Yen1ND (ND) were added to the reaction and incubated for 1 h at 30ºC. (-) indicates no enzyme. Reaction products were analysed by 12% denaturing PAGE and scanned at 635 nm (Cy5) in a Typhoon FLA9500. A mixture of 5'-6FAM-end-labelled oligos of defined length was used as markers. (C) To identify incisions on the plasmid, after D2 D-loop formation, 400 nM Yen1 was added to the reaction and incubated for 1 h at 30ºC. Then, pBSK was cut with 10 U BspHI at 37ºC for 20 min, followed by treatment with rSAP for another 20 min at 37ºC and heat inactivation at 99ºC for 5 min. Reaction products were labelled using T4 PNK and 32P-γ-ATP and analysed by 6% denaturing PAGE followed by phosphorimaging. A mixture of DNA fragments was radioactively labelled and used as marker. Arrows indicate products of expected sizes. (D) and (E) same as B and C but employing RPA-coated plasmid-based D2 Dloops. (F) Schematic representation of Yen1 (purple) incisions on the plasmid-based D2 D-loop. Open arrowheads indicate the other possible position for both cleavage sites. Results 135 Results 136 Congruently, when we employed the sequencing strategy to map Yen1 cleavage sites on the plasmid-based D2 D-loop, we confirmed that the incisions occurred also at the end of the displaced region and without changes due to the presence of RPA (Figures 40A-C). Again, a derivate D2 D-loop (D2.2, see Tables 1 and 2) in which the 3´-end of the invading strand was displaced (from position 2012 to 2007 of pBSK, see Tables 1 and 2), confirmed the specificity of the sequencing results (Figure 40D). (on the next page) Figure 40. Mapping by sequencing of Yen1 incisions on D2 Dloops. (A) Scheme of the mapping strategy employing DNA sequencing. After D-loop formation, 400 nM Yen1 was added and incubated for 1 hour at 30ºC followed by deproteinization. A sample was taken for analysis on agarose gels. The rest was purified by phenol-chloroform extraction followed by EtOH precipitation. Reaction products were sequenced with Taq polymerase employing forward and reverse primers. (B) Agarose gel analysis of Yen1 reactions prior to DNA purification and sequencing. Fresh gel was scanned at 635 nm (Cy5) in a Typhoon FLA9500. (-) indicates no enzyme. Labels: R51, Rad51; R54, Rad54. (C) Graphical representation of the incisions mapped on plasmid-based D2 D-loops. The nucleotide sequence near the branch point is shown. Electropherograms derived from sequencing using the reverse primer are depicted at the bottom. Note that only electropherograms of the region of interest are shown. Purple arrows indicate Yen1 cleavage sites and adenine incorporation is pointed out with a purple dotted box. Nuclease-dead mutant controls are shown (ND). (D) Same as C but employing a derivative D2.2 invading oligo. Note that the second branched point is different from the previous one. Results 137 Results 144 Figure 44. Analysis of a potential half-crossover intermediate in native gels. (A) Graphical experimental scheme to analyse the origin of the slowly-migration band. After D3 D-loop cleavage with Yen1 and Mus81, when appropriate, reaction products were ligated using T4 ligase (1 h, 23ºC) and/or cleaved with EcoRV (1 h, 37ºC). Cleavage by SSEs + EcoRV is expected to release two linear molecules of 1716 and 1242 bp, respectively. (B) EcoRV treatment releases two fragments of the expected size only when the D3 D-loop is cleaved by Mus81 and Yen1. Red arrowhead indicates the 1716 bp fragment (Cy5 labelled) and black arrowhead indicates the 1242 bp fragment (unlabelled). Cleavage products were analysed on an agarose gel and scanned at 635 nm (Cy5) on a Typhoon FLA9500 (left) followed by EtBr post-staining (right). Dashed line indicates removal of irrelevant lanes. Labels: N, Nicked; L, Linear; SC, Supercoiled. While the appearance of these products required the combined activity of Yen1, Mus81, and EcoRV, it did not require the presence of T4 ligase (Figure 44B, compare lanes 5 and 6 in both panels). It could hence be argued that since such intermediates are stable without the ligation of the invading strand to the plasmid, ligation may not have occurred in our reactions. Therefore, we repeated these experiments and analysed them through denaturing PAGE, which should reveal the presence of high molecular weight bands only if the invading strand is Results 145 ligated to the plasmid (Figure 45; 3058 nt expected size: 2958 from pBSK plus 100 nt of D2 oligo). Indeed, Cy5-labelled DNA can be observed in the wells of the gel only upon Yen1, Mus81, and T4 ligase treatment (Figure 45B, upper panel). Further digestion with EcoRV allows the entry of the DNA in the gel, despite its short migration (Figure 45B, upper panel, expected size: 1741 nt, compare lanes 5 and 6). To confirm that the expected molecule is being produced, we repeated the experiments with AhdI digestion, which cleaves closer to the ligation point and thus releases a smaller, 129 nt Cy5-labelled DNA fragment that migrates faster into the gel (Figure 45C, upper panel). The appearance of this band exclusively in the reaction with Yen1, Mus81, T4 ligase, and AhdI corroborates that combined action of Yen1 and Mus81 generate a product in which the invading strand is ligated to the donor molecule, one of the required steps to generate a half-crossover intermediate (see Figure 47). We also examined what happened to the other strand that should be involved in the formation of a halfcrossover, the 3´-Cy3 labelled oligonucleotide complementary to the invading one (Figure 45A). We scanned the same gels for Cy3 in order to see if we could also trace the transfer of the 3’-Cy3 labelling to the plasmid, but we could not detect it (Figure 45B, bottom panels). This is consistent with Mus81 cleaving this structure 4-5 nt away from the first branching point (Figure 37), thus leaving a 4-5 nt gap unsuitable for T4ligase reaction. Moreover, in Figure 45B it could be appreciated that the combined actions of Mus81 and Yen1 prevent Yen1 from cleaving the invading strand, indicated by the absence of short (approx. 20 nt) 5´-Cy5-labelled DNA fragments (Figure 45B, compare lanes 4-6 with 7-9). This incision would be necessary to generate a chromosomal loss event but would be incompatible with the generation of a half-crossover intermediate (see Figure 47). Results 146 Results 147 (on the previous page) Figure 45. Analysis of a potential half-crossover intermediate in denaturing gels. (A) Graphical scheme of the strategy to detect half-crossover precursors. After D3 D-loop cleavage with Yen1 and Mus81, when appropriate, reaction products were ligated using T4 ligase (1 h, 23ºC) and cleaved with EcoRV or AhdI (1 h, 37ºC). After analysis on denaturing PAGE, cleavage by SSEs + EcoRV should release a 1741 nt 5´-Cy5-labelled fragment and cleavage by SSEs + AhdI should release a 129 nt 5´-Cy5-labelled fragment. (B) Reaction products of Mus81 and Yen1 actions on a D3 D-loop followed by DNA ligation and cleavage with EcoRV were analysed on a 10% denaturing PAGE. Dual coloured arrows indicate the invading strand ligated to the plasmid backbone. Expected products are depicted on the right. Gels were scanned at 635 nm (Cy5) (upper panel) and at 532 nm (Cy3) (bottom panel) in a Typhoon FLA9500. Labels: Y, Yen1; M, Mus81; EV, EcoRV. (C) Same as B but employing AhdI restriction enzyme. Same labels as in B. AI, AhdI. Overall, these results demonstrate that the concurrent actions of Yen1 and Mus81 on a D3 D-loop, followed by ligation of the invading strand to the recipient plasmid, result in a nicked molecule that would represent a direct precursor of a half-crossover. 5 DISCUSSION Although HR has been traditionally considered an error-free pathway, it may promote genome destabilization through mutagenesis and chromosomal rearrangements (Malkova and Haber 2012). One dangerous toolkit involved in HR are the SSEs Mus81 and Yen1. Due to their ability to process branched DNA molecules according to their respective polarities, cells must control them in order to avoid genome instability (Dehé and Gaillard 2017). Genetic and biochemical studies indicate that the main function of these enzymes is to deal with branched DNA molecules remaining at the end of the cell cycle, structures that could compromise chromosome separation in anaphase (Wild and Matos 2016). Accordingly, these enzymes become activated at the end of the cell cycle, Mus81 in G2/M (Gallo-Fernández et al. 2012; Gritenaite et al. 2014; Matos et al. 2011, 2013; Princz et al. 2017; Szakal and Branzei 2013) and Yen1 at the metaphase-to-anaphase transition (Blanco et al. 2014; Eissler et al. 2014; Matos et al. 2011). Based on the recombination paradigm, it has been proposed that the main functions of these enzymes during DSB repair could be explained by their actions on nicked or fully ligated Holliday junctions, the last branched molecules generated in the classical DSBR model. However, biochemical approaches have shown that these enzymes can process many other branched structures as 3´-flaps, 5´-flaps, replication fork-like molecules, or Y-structures (Carreira et al. 2022; Ehmsen and Heyer 2008; Whitby et al. 2003; Bastin-Shanower et al. 2003; Ciccia et al. 2003; Osman et al. 2003; Gaillard et al. 2003; Taylor and McGowan 2008; Doe et al. 2002; Schwartz et al. 2012; Pepe and West 2014b; Blanco et al. 2014). Studies employing naked oligonucleotide-based DNA structures have helped to understand the reason behind the strict cell cycle regulation imposed over these enzymes as, when active, they Discussion 150 can process other physiologically essential branched structures like replication forks or early recombination intermediates (Kai et al. 2005; Duda et al. 2016; Szakal and Branzei 2013). It has been suggested that one of these molecules could be the displacement-loop, the central intermediate of the HR pathway (Figures 4, 5, and 8) (Heyer et al. 2003; Schwartz and Heyer 2011). 5.1 MUS81 AND YEN1 INCISIONS ON SYNTHETIC D-LOOPS REFLECT THE POLARITY OF THEIR RESPECTIVE NUCLEASE FAMILIES Here we show a comprehensive characterization of how S. cerevisiae Mus81 and Yen1 cleave three different oligonucleotide-based D-loop structures (Figures 16, 18, 19, and 46). The cleavage sites detected for Mus81 on the synthetic D3 D-loop (Figures 16 and 46) are in the same strand and almost at the same position as previously reported (Ehmsen and Heyer 2009), and as those described for the S. pombe ortholog Mus81-Eme1 (Osman et al. 2003) and the human counterpart MUS81EME1 (Pepe and West 2014b). S. cerevisiae Mus81 was previously shown to introduce a single incision at the strand complementary to the invading molecule, 4 nt away to the 5´-side of the first branching point (Ehmsen and Heyer 2009). S. pombe Mus81-Eme1 performs multiple incisions at the very same strand, 3-8 nt from the first branching point, with a predominant incision 3 nt away (Osman et al. 2003). In the case of MUS81-EME1, it cleaves the same strand, but 4 nt away from the first junction point (Pepe and West 2014b). In contrast, human MUS81EME2, which introduced multiple incisions at the same position as the other homologs, it can also cut the same strand but at the end of the second branching point and the invading strand at the first junction point (Pepe and West 2014b). This is consistent with MUS81-EME2 exhibiting a broader range of substrate specificity, as it was reported to be also able to process a 5´-flap structure (Pepe and West 2014b), an unusual activity for a member of the XPF-ERCC1 nuclease family (Ciccia et al. 2008). Regarding the synthetic D2 D-loop, structure that mimics a situation after D-loop bubble migration, we did not observe any activity Discussion 151 for Mus81 (Figures 18 and 46), as previously reported for its S. pombe ortholog Mus81-Eme1 (Osman et al. 2003) and in agreement with its catalytic requirement of a 5´-end near the junction point (BastinShanower et al. 2003; Ehmsen and Heyer 2009). Surprisingly, in the case of the D1 D-loop, which does not present any ssDNA tail, we detected that Mus81 can also process this structure, but on the opposite strand to that on D3 (Figures 19 and 46). We suspect that Mus81 is recognising this structure like a 3'-flap, and as such, cutting the 3´-single strand (Bastin-Shanower et al. 2003; Ehmsen and Heyer 2009). Similarly, the same activity, although very minor, was detected when the short D3 synthetic D-loop was analysed (Figure 17). This could stem from Mus81 activity on an incomplete annealing product, probably a bubble-like structure formed between oligos 1 and 2 (see Tables 1 and 2) (Ehmsen and Heyer 2008). Figure 46. Graphical overview of the cleavage sites identified for Yen1 and Mus81-Mms4 on the three oligonucleotideand enzymatically-based D-loop structures. Purple arrowheads indicate Yen1 incisions. Green arrowheads indicate Mus81-Mms4 incisions. Arrowhead and number size are proportional to the relative efficiency of the cleavage. Figure created by the author employing Adobe Illustrator. Discussion 152 In the case of Yen1, we characterized for the first time its ability to process a D-loop intermediate. The two main cleavage points of Yen1 on the analysed structures (at the end of the displaced strand and at the invading molecule) (Figure 46) are consistent with the expected ones for an endonuclease belonging to the Rad2/XPG family. Importantly, both incisions are independent, with the one at the displaced strand being much faster than the one at the invading oligo (Figures 20 and 21). We also detected three other unexpected activities: i) Nuclease activity on the 5´-end on the strand complementary to the invading oligo in the D3 D-loop (Figures 16 and 46), ii) nuclease activity on the 5´-end on the invading strand in the D1 D-loop (Figures 19 and 46) and iii) some minor activity on the strand complementary to the invading molecule, near to the first branching point (Figures 16, 18, 19, and 46). Regarding the first two activities, we have previously reported that Yen1 is endowed with a nick-specific 5´-3´ exonuclease activity (Carreira et al. 2022). Therefore, we speculate that Yen1 may be processing these two 5´-ends similarly to a nicked dsDNA. In line with this, when we separate the 5' end from the branching point in the D3 Dloop, this activity is gone (Figure 17). Nevertheless, we cannot rule out the possibility that, instead of being a 5'-3' exonuclease, these activities could stem from pseudo-5´-flaps created by DNA breathing that Yen1 may process endonucleolytically. Regarding the other minor endonuclease activities at the strand complementary to the invading molecule (Figures 16, 18, 19, and 46), we consider that they could arise from structural/conformational changes in the D-loop as a consequence of the main incisions produced by Yen1, and we cannot exclude that they could be an artefact of the in vitro experimental conditions. 5.2 THE POTENTIAL BIOLOGICAL RELEVANCE OF THE INCISIONS PRODUCED BY MUS81 AND YEN1 ON D-LOOPS A key question is whether these nuclease activities are of biological relevance. Significant genetic data with mus81Δ and yen1Δ mutants suggest that these enzymes may be able to process D-loops in vivo. It has been shown that MUS81 deletion causes an increase in BIR repair, increase that is even higher in the absence of YEN1 (Ho et al. 2010). Discussion 153 This observation is similar to what has been described for Mph1, a helicase that is able to dismantle D-loops in vitro (Prakash et al. 2009). As the D-loop molecule is the central intermediate of the BIR pathway, it has been described that Mph1 helicase activity over this intermediate opposes to the progression of this pathway (Štafa et al. 2014). Congruently, its deletion leads to an increase in BIR (Mehta et al. 2017). Our data may suggest that Mus81 and Yen1 can process D-loops and, by doing so, preventing the progression of the repair by BIR, limiting its associated mutagenesis. Indeed, employing a different genetic system to measure BIR contribution to the repair of broken replication forks, Mayle et al. (2015) demonstrated that converging forks and Mus81 activity limit the mutagenicity associated with replication restart by Pol32-dependent BIR. They also showed that both Mus81 and Yen1, as well as the key HR proteins Rad52 and Rad51, are required for the repair of this type of damage. This is consistent with the idea that the HR pathway is required for the restart of dysfunctional replication forks (Ait Saada et al. 2018). Later, employing the DNA topoisomerase I poison camptothecin (CPT) to block replication forks, it was shown that Sphase completion in response to CPT requires Rad52 and Rad51 but neither Mus81 nor Pol32, suggesting that Pol32-dependent BIR is not the primary pathway for the restart of blocked forks (Pardo et al. 2020), as previously reported for the repair of broken ones (Mayle et al. 2015). Pardo and colleagues also suggest that the HR proteins, rather than promote replication restart, might protect replication forks until the arrival of a convergent fork, reminiscent of the role of RAD51/BRCA2 in the protection of nascent DNA strands at damaged forks (Schlacher et al. 2011; Lemaçon et al. 2017; Halder et al. 2022). It has been proposed that this protection could be enforced through the formation of a D-loop structure (Jakobsen et al. 2019; Pardo et al. 2020), similar to the end-protection mechanism described for telomeres (Griffith et al. 1999). Since Mus81 is required for cell survival in response to CPT, they proposed that Mus81 is needed at G2/M to promote termination of DNA replication, presumably, by cleaving the single Holliday junction that would form upon merging the D-loop with the converging fork. In line with this, employing the same Flp-nick system as Mayle et al.