Impact of Pif1 translational mechanism on genetic interactions with YEN1ON and other DNA repair enzymes

INTERNATIONAL DOCTORAL SCHOOL OF THE USC Tomás Lama Díaz PhD Thesis Impact of Pif1 translational mechanism on genetic interactions with YEN1ON and other DNA repair enzymes Santiago de Compostela, 2023 Doctoral Programme in Molecular Medicine Tomás Lama Díaz 2 DOCTORAL THESIS IMPACT OF PIF1 TRANSLATIONAL MECHANISM ON GENETIC INTERACTIONS WITH YEN1 ON AND OTHER DNA REPAIR ENZYMES Tomás Lama Díaz INTERNATIONAL PHD SCHOOL OF THE UNIVERSITY OF SANTIAGO DE COMPOSTELA PHD PROGRAMMA IN MOLECULAR MEDICINE SANTIAGO DE COMPOSTELA AÑO 2023 Tomás Lama Díaz 4 DECLARACIÓN DEL AUTOR/A DE LA TESIS D./Dña. Tomás Lama Díaz Título de la tesis: Impact of Pif1 translational mechanism on genetic interactions with YEN1-ON and other DNA repair enzymes 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, 23 diciembre 2022. Tomás Lama Díaz 6 AUTORIZACIÓN DEL DIRECTOR/TUTOR DE LA TESIS D./Dña. Miguel González Blanco En condición de: Tutor/a y director/a Título de la tesis: Impact of Pif1 translational mechanism on genetic interactions with YEN1-ON and other DNA repair enzymes INFORMA: Que la presente tesis, se corresponde con el trabajo realizado por D/Dña Tomás Lama Díaz, bajo mi dirección/tutorización, y a utorizo su presentación, considerando que reúne l os r equisitos exigidos en el R eglamento de Estudios de Doctorado de la USC, y que como director/tutor de esta no incurre en las causas de abstención establecidas e n la Ley 40/2015. En Santiago de Compostela, 23 de diciembre de 2022 Tomás Lama Díaz 8 DECLARACIÓN DE CONFLICTO DE INTERESES Y AUTORÍA DE DATOS PRESENTADOS Yo, Tomás Lama Díaz, con DNI 50487181-L: Declaro no tener ningún tipo de conflicto de intereses, ni ninguna relación económica, personal, política, interés financiero o académico que pueda influir en este trabajo. Así mismo, declaro la completa autoría de todas las imágenes y figuras que se muestran en la tesis. Tomás Lama Díaz 16 INDEX AGRADECIMIENTOS ...................................................................................................................................... 13 INDEX ........................................................................................................................................................... 16 ABBREVIATIONS ........................................................................................................................................... 20 EXTENDED SUMMARY EN GALICIAN ............................................................................................................. 22 SUMMARY .................................................................................................................................................... 28 INTRODUCTION ............................................................................................................................................ 29 1. GENOME INTEGRITY AND DNA REPAIR ................................................................................................................... 29 1.1 DNA damage response ........................................................................................................................... 31 1.1.1 Direct reversal of DNA damage ........................................................................................................................ 31 1.1.2 Base excision repair .......................................................................................................................................... 31 1.1.3 Nucleotide excision repair ................................................................................................................................ 33 1.1.4 Mismatch repair ................................................................................................................................................ 35 1.1.5 DSB repair: NHEJ, HR and alternative pathways ............................................................................................... 37 1.1.5.1 NHEJ .......................................................................................................................................................... 37 1.1.5.2 Homologous recombination ..................................................................................................................... 39 1.1.5.3 Homology-mediated repair: MMEJ and SSA ............................................................................................ 41 1.1.6 Replication stress: DDT and fork reversal ......................................................................................................... 43 1.1.6.1 DDT pathways: DNA damage bypass at the fork and behind the fork. .................................................... 44 1.1.6.2 Fork reversal ............................................................................................................................................. 48 1.2 Coordination of DNA repair by DNA damage checkpoint and cell cycle kinases .................................... 50 1.2.1 DNA damage checkpoint .................................................................................................................................. 50 1.2.2 DNA repair throughout the cell cycle ............................................................................................................... 54 2. HOMOLOGOUS RECOMBINATION AND ITS CELL-CYCLE REGULATION .............................................................................. 55 2.1 DSB end resection and pathway choice .................................................................................................. 55 2.2 Nucleofilament assembly, strand invasion and D-loop formation ......................................................... 60 2.3 Extended D-loop disruption: BIR and SDSA ............................................................................................ 64 2.4 Second-end capture: Dissolution and Resolution of dHJs ....................................................................... 68 2.4.1 Dissolution of dHJs ............................................................................................................................................ 70 2.4.2 Resolution of dHJs ............................................................................................................................................ 72 2.4.3 Crossover regulation ......................................................................................................................................... 76 2.5 Modulation of SSE activity outside the NCO/CO balance ....................................................................... 81 3. PIF1 FAMILY OF HELICASES ................................................................................................................................... 87 3.1 Structural motifs and protein isoforms of Pif1-family helicases ............................................................. 88 3.2 Pif1-helicases in DNA replication through physical barriers ................................................................... 91 3.2.1 G-quadruplex structures ................................................................................................................................... 91 3.2.2 Highly transcribed genes and protein barriers ................................................................................................. 92 3.3 Pif1-helicases in Okazaki fragment maturation ..................................................................................... 93 17 3.4 Pif1-helicases in fork convergence during replication termination ........................................................ 94 3.5 Pif1-helicases in telomere length homeostasis ...................................................................................... 94 3.6 Pif1-helicases in DSB repair .................................................................................................................... 95 3.7 Pif1-helicases in mtDNA maintenance ................................................................................................... 96 CHAPTER 1: ANALYSIS OF YEN1 PREMATURE ACTIVATION ............................................................................ 98 1. OBJECTIVES .................................................................................................................................................... 100 2. RESULTS ......................................................................................................................................................... 101 2.1 Overexpression of YEN1 deregulated alleles results in lethality, cell cycle arrest and Rad53 phosphorylation ......................................................................................................................................... 101 2.2 The lethality at strain overexpressing deregulated YEN1 is not dependent on checkpoint activation or homologous recombination, ...................................................................................................................... 104 2.3 Expression of Yen1ON at endogenous levels displays genetic interactions with HR helicases .............. 105 2.4 Cells deficient in nuclear Pif1 become hypersensitive to HU upon expression of Yen1ON ..................... 109 3. DISCUSSION .................................................................................................................................................... 113 3.1 Cell-cycle regulation of Yen1 prevents genome instability. .................................................................. 113 3.2 Upon genotoxic stress, nuclear Pif1 may prevent the accumulation of DNA secondary structures that Yen1ON could hydrolyse. ............................................................................................................................. 114 3.3 Future plans ......................................................................................................................................... 115 CHAPTER 2: PIF1 TRANSLATIONAL MECHANISM .......................................................................................... 118 1. OBJECTIVES ................................................................................................................................................. 121 2. RESULTS ...................................................................................................................................................... 122 2.1 The pif1-m2 allele does not fully recapitulate the phenotypes derived from loss of Pif1 nuclear functions in pif1∆ mutants. ........................................................................................................................................ 122 2.2 A novel, fast-migrating isoform of Pif1 produced by alternative translation initiation ....................... 125 2.3 Pif1 isoforms are produced through ribosomal leaky scanning ........................................................... 128 2.4 Pif1107-859 is catalytically active and retains partial functionality in the nucleus. ................................. 133 2.5 Identification of the nuclear localization signal in Pif1 ........................................................................ 137 2.6 Generation of an improved nuclear-null PIF1 allele ............................................................................. 143 2.7 Residual mitochondrial activity is present in pif1-m1 mutants ............................................................ 146 3. DISCUSSION ................................................................................................................................................ 152 3.1 Variable phenotypic impact of the residual nuclear Pif1 activity in pif1-m2 mutants ......................... 152 3.2 Pervasive ribosomal leaky scanning of PIF1 mRNA facilitates ATI from AUG1, AUG40 and AUG107 ...... 153 3.3 The intrinsic ATI mechanism operating on PIF1 mRNA constrains the efficiency of pif1-m1 and pif1-m2 as separation-of-function alleles. ............................................................................................................... 153 3.4 On the novel Pif1 isoform and the functionality of the N-terminal domain ......................................... 156 3.5 pif1nuc and pif1mit as valuable tools for the DNA repair field. ............................................................... 158 3.6 Extensibility of our model to other genes ............................................................................................. 158 CONCLUSIONS ............................................................................................................................................. 166 MATERIALS ................................................................................................................................................. 168 Tomás Lama Díaz 18 1. BACTERIAL STRAINS .......................................................................................................................................... 168 2. YEAST STRAINS ................................................................................................................................................ 168 3. PLASMIDS ....................................................................................................................................................... 173 4. OLIGONUCLEOTIDES ......................................................................................................................................... 178 5. ANTIBODIES .................................................................................................................................................... 185 6. MEDIA ........................................................................................................................................................... 185 7. REAGENTS ...................................................................................................................................................... 187 METHODS ................................................................................................................................................... 192 1. MICROBIOLOGY ............................................................................................................................................... 192 1.1 Bacterial methods ................................................................................................................................ 192 1.1.1 Bacterial inoculation and stocks ..................................................................................................................... 192 1.1.2 Preparation and transformation of competent cells ...................................................................................... 192 1.2 Yeast methods ...................................................................................................................................... 193 1.2.1 Yeast cultures and stocks ................................................................................................................................ 193 1.2.2 Yeast mating, sporulation and tetrad dissection ............................................................................................ 193 1.2.3 Yeast transformation ...................................................................................................................................... 193 2. MOLECULAR BIOLOGY ....................................................................................................................................... 194 2.1 Isolation of plasmid DNA from bacteria ............................................................................................... 194 2.1.1 Isolation of plasmid DNA by alkaline lysis ....................................................................................................... 194 2.1.2 Isolation of plasmid DNA by non-ionic detergent (NID) boiling. ..................................................................... 195 2.2 Isolation of yeast genomic DNA ........................................................................................................... 195 2.3 Polymerase chain reaction (PCR) .......................................................................................................... 196 2.4 DNA purifications ................................................................................................................................. 196 2.4.1 PCR cassettes .................................................................................................................................................. 196 2.4.2 Gel extraction ................................................................................................................................................. 196 2.4.3 Ethanol precipitation ...................................................................................................................................... 196 2.5 DNA sequencing ................................................................................................................................... 197 2.6 Cloning and mutagenesis of DNA fragments ....................................................................................... 197 2.6.1 Cloning with restriction enzymes .................................................................................................................... 197 2.6.2 Gateway cloning ............................................................................................................................................. 198 2.6.3 Mutagenesis by inverse PCR ........................................................................................................................... 198 2.6.4 Gene editing of S. cerevisiae ........................................................................................................................... 198 2.6.4.1 Integration of plasmids in the host chromosomes ................................................................................. 198 2.6.4.2 Knockouts, promoter replacement and epitope tagging. ...................................................................... 199 2.7 Determination of telomere length by Southern Blot. ........................................................................... 199 2.8 Purification of synthetic DNA substrates .............................................................................................. 201 2.8.1 Purification of unlabelled and fluorescently labelled oligos ........................................................................... 201 2.8.2 Hybridization of oligonucleotides and purification of fluorescentlylabeled substrates ................................. 203 3. BIOCHEMICAL TECHNIQUES ................................................................................................................................ 205 3.1 Preparation and analysis of protein extracts ....................................................................................... 205 3.1.1 Preparation of denatured protein extracts with trichloroacetic acid (TCA) ................................................... 205 19 3.1.2 Total Protein detection by direct staining ...................................................................................................... 206 3.1.2.1 Ponceau staining .................................................................................................................................... 206 3.1.2.2 Coomassie staining ................................................................................................................................ 206 3.1.2.3 Total protein staining ............................................................................................................................. 206 3.1.3 Protein detection by Western blot ................................................................................................................. 206 3.2 Protein purifications ............................................................................................................................. 207 3.2.1 Purification of scPif1 in E. coli. ........................................................................................................................ 207 3.2.1.1 Optimization of induction conditions and solubility tests ..................................................................... 208 3.2.1.2 Purification of Pif1 .................................................................................................................................. 209 3.2.2 Purification of Yen1 ........................................................................................................................................ 210 3.3 Biochemical assays ............................................................................................................................... 213 3.3.1 ATPase assays ................................................................................................................................................. 213 3.3.2 DNA unwinding assays ................................................................................................................................... 214 4. CELL BIOLOGY TECHNIQUES ................................................................................................................................ 215 4.1 Fluorescence microscopy ..................................................................................................................... 215 4.2 Flow cytometry .................................................................................................................................... 216 4.3 Genotoxicity assays .............................................................................................................................. 217 5. IN SILICO ANALYSES ........................................................................................................................................... 218 BIBLIOGRAPHY ............................................................................................................................................ 219 INDEX OF FIGURES ...................................................................................................................................... 270 Tomás Lama Díaz 20 ABBREVIATIONS 3’- dRP 3’-deoxyribose phosphate 5’- dRP 5’-deoxyribose phosphate AP Apurinic or apyrimidinic ATI Alternative translation initiation ATP Adenosine triphosphate BER Base excision repair BIR Break-induced replication BSA Bovine serum albumin BTR BLM-Topoisomerase IIIaRMI1-RMI2 CDK Cyclin-dependent kinase CFSs Common fragile sites CO Crossover CPT Camptothecin CSM Complete supplement mixtures CTD C-terminal domain CV Column volume DDC DNA damage checkpoint DDK Dbf4-dependent kinase DDR DNA damage response DDT DNA damage tolerance D-loop Displacement loop dHJ Double Holliday junction DTT Dithiothreitol DIG Digoxigenin DSBR Double-strand break repair gDNA Genomic DNA HDR Homology-directed repair HJ Holliday Junction HR Homologous recombination HU Hydroxyurea IPTG Isopropyl βd-1-thiogalactopyranoside IRES Internal ribosomal entry sites IDLs Insertion or deletion loops JM Joint Molecule LB Lysogenic vroth LOH Loss of heterozygosity MMBIR Microhomology-mediated BIR MMEJ Microhomology-mediated end-joining MMR Mismatch repair 21 MMS Methyl methanesulfonate MTS Mitochondrial targeting signal NCO Non-crossover ND Nuclease-dead NER Nucleotide excision repair NHEJ Non-homologous end-joining nHJ Nicked Holliday junction NID Non-ionic detergent NLS Nuclear localization signal NTD N-terminal domain ORF Open reading frame PCNA Proliferating cell nuclear antigen PIC Pre-initiation complex PIKK Phosphatidylinositol-3 kinase-related kinase PIP PCNA-interacting protein Pold DNA polymerase d RDR Recombination-dependent restart RFB Replication fork barrier RFC Replication Factor C ROS Reactive oxygen species SC Synthetic complete media SDSA Synthesis-dependent strand annealing SP Salvage pathway SSA Single strand annealing SSB Single strand breaks SSE Structure selective endonuclease STR Sgs1-Top3-Rmi1 TAE Tris-acetate-EDTA TBE Tris-Borate-EDTA TCA Trichloroacetic acid TLC Thin-layer chromatography TLS Trans-lesion synthesis TS Template switching uORF Upstream uORF WGD Whole genome duplication WT Wild-type Tomás Lama Díaz 22 EXTENDED SUMMARY EN GALICIAN A reparación do ADN é fundamental para garantir a preservación da estabilidade xénomica e transmitir de forma fidedigna a información xenética á seguinte xeración. As células están expostas de maneira continuada a diversos axentes que comprometen a integridade do seu xenoma ao introducir modificacións que alteran a información quimicamente codificada polas súas febras. Entre os factores que constitúen un risco para o ADN inclúense axentes xenotóxicos exóxenos, como as especies reactivas de oxíxeno e diversos carcinóxenos, pero tamén procesos intrínsecos que inclúen transaccións entre moléculas ADN, como a replicación do xenoma. Un dos danos máis citotóxicos no ADN son as roturas de dobre febra (DSBs, Double strand breaks), que poden causar inestabilidade no xenoma se non se recupera a información xenética perdida na rotura. A reparación dos DSBs pode ser realizada por dúas vías principais: a recombinación homóloga (HR, homologous recombination) e a unión non homóloga de extremos (NHEJ, do inglés "non-homologous end joining"). A HR é un mecanismo de alta fidelidade que emprega unha molécula de ADN homóloga como modelo para a reparación dos DSBs. En particular, as células favorecen o uso da cromátide irmá como model de reparación, polo que esta ruta se encontra restrinxida ás fases S-G2-M do ciclo celular. Dentro do proceso da HR, existen varias rutas de reparación que aseguran a eliminación de todos os intermediarios de recombinación que xorden durante o proceso de copiado da información xenética perdida, incluíndo a hibridación de cadea dependente de síntese (SDSA, Synthesis-dependent strand-annealing), a vía clásica de reparación de roturas de dobre cadea (DSBR, Double Strand Break Repair) e a replicación inducida por rotura (BIR, do inglés Breakinduced replication). En todos os casos, o proceso de reparación comeza coa resección do DSB, un paso no que os extremos 5’ da rotura son degradados. Este paso é mediado pola acción coordinada de nucleasas e helicasas, incluíndo o complexo MRN (MRe11-Rad50-Xrs2), Exo1 e Dna2. A resección xera moléculas de cadea sinxela cun extremo 3’-OH que pode primar a síntese do ADN e, ademais, proporciona o substrato para a formación de filamentos nucleoproteicos. Os nucleofilamentos fórmanse pola unión da recombínase Rad51 sobre o ADN de cadea sinxela, iniciando a busca dunha secuencia homóloga na cromátida irmá ou no cromosoma homólogo que poida servir como molde para a reparación. Unha vez atopada a secuencia homóloga, o nucleofilamento de Rad51 é tamén capaz de invadir o dúplex de ADN homólogo, desprazando unha das cadeas do mesmo, o que resulta na formación dunha estrutura de ADN ramificada, denominada bucle de desprazamento (D-loop, Displacement loop). Os Dloops son os intermediarios centrais da HR, e poden ser procesados por helicasas (Mph1, Srs2) ou primar un evento de síntese a partir do 3’-OH da cadea invasora. A súa vez, o D-loop alongado pode ser procesado pola ruta de SDSA ou, se a cadea desprazada do D-loop híbrida co segundo estremo 3’-OH da rotura, madurar nun novo intermediario de recombinación, a dobre unión de Holliday (dHJ, Dobre unión de Holliday). Estas estruturas cruciformes conectan covalentemente as moléculas de ADN involucradas no proceso de reparación, polo que deben ser eliminadas antes de que aconteza a división celular, para evitar comprometer a segregación cromosómica. Dada a estabilidade diste intermediario, as células contan con dous mecanismos 23 diferentes para garantir a pronta eliminación das dHJ: a disolución e maila resolución. Durante a disolución, un complexo constituído por Sgs1-Top3-Rmi1 (STR) dirixe a migración converxente das unións de Holliday para orixinar un hemicatenano que é seguidamente decatenado por Top3. Por outro lado, a resolución das dHJ é levada a cabo por distintas endonucleases selectivas de estrutura (SSEs, do inglés Structure-selective endonucleases), entre as que se inclúen Mus81-Mms4, Slx1-Slx4 e Yen1. Dependendo da ruta de HR e das enzimas empregadas para a eliminación dos intermediarios de recombinación, a reparación dos DSB pola HR terá como produtos finais moléculas de ADN sen entrecruzamentos (CO, crossovers) ou con eles (NCO, non-crossovers). En particular, as rutas de SDSA ou disolución xeran exclusivamente NCOs, ao non producirense intercambios de dúplexes entre as moléculas donantes e receptoras a ambos lados da DSB. Pola contra, o corte das dHJ durante a resolución produce unha mestura entre CO é NCO, dependendo da orientación dos cortes requiridos para procesar as dHJs. Dado o potencial risco de perda de heterocigosidade asociada aos CO entre cromosomas homólogos, as células contan cun intricado sistema de regulación que permite canalizar os intermediarios da recombinación cara as rutas de SDSA e disolución. Durante a última década demostrouse como a maquinaria que dirixe a progresión do ciclo celular é tamén responsable de xerarquizar a activación das rutas de HR e controlar o seu acceso aos intermediarios de recombinación, incorporando unha dimensión temporal e espacial ao control da HR. O traballo de distintos grupos demostrou como quinasas e fosfatasas reguladoras da progresión do ciclo, a través de modificacións postraducionais, modulan a activación das rutas de procesamento de dHJs. En particular, os eventos de fosforilación e defosforilación das resolvasas Mus81-Mms4 e Yen1 pospoñen a súa activación ata a transición G2/M e fase M respectivamente, acadando así una activación secuencial que permite favorecer o procesamento de dHJs polas rutas SDSA e disolución. Ao pospoñer a activación das resolvasas ata as últimas etapas do ciclo celular, esta regulación limita a formación de COs e, ao mesmo tempo, acada a eliminación completa de tódalas dHJs antes da división celular. Así e todo, limitar os niveis de COs podería non ser a única causa pola que as células restrinxen a actividade das resolvasas aos últimos estadios do ciclo celular. Como se mencionou ao principio deste resumo, ademais da HR, os procesos de replicación e transcrición tamén involucran transaccións entre moléculas de ADN e estruturas ramificadas como forcas de replicación, estruturas G-cuádruplex, ramificacións 5’, R-Loops (híbridos entre ARN e ADN), etc. Ademais, a exposición de rexións de cadeas sinxelas de ADN e as condicións topolóxicas durante a replicación e a transcrición tamén favorecen que o ADN adquira estruturas que se desvían da conformación habitual, incluíndo G-cuádruplex ou H-DNA. Dado que a regulación de Yen1 a inactiva e exclúe do núcleo durante as fases S-G2, cando a maioría do ADN é duplicado, a súa activación tardía, ademais de reducir os niveis de entrecruzamentos, podería ter tamén unha función protectora para evitar a súa actividade sobre Tomás Lama Díaz 24 intermediarios de replicación, intermediarios de recombinación temperáns, ou de estruturas secundarias. En liña con esta hipótese, demostrouse en diferentes contextos como o procesamento por SSEs de intermediarios de replicación, reparación ou estruturas secundarias leva asociados riscos para á estabilidade xenómica das células, que polo tanto promoven o desmantelamento destes intermediarios por helicasas. Para testar a nosa hipótese, empregamos diversos alelos desregulado da nucleasa YEN1. Ao comezo da fase S, a fosforilación de Yen1 por Cdc28 (CDK) reduce a súa actividade catalítica e forza a súa exportación ao citoplasma ata o comezo da anafase, cando a fosfatasa Cdc14 reverte esta inactivación. Polo tanto, cara o final da fase M, Yen1 pode acceder ao núcleo nun estado activo para eliminar os intermediarios de recombinación máis persistentes. O coñecemento de esta regulación permitiu xerar un alelo desregulado baseado na mutación de nove serinas localizadas en lugares consenso de fosforilación para Cdk. Este alelo, denominado YEN1ON, codifica unha versión de Yen1 constitutivamente activa e que localízase no núcleo ao longo de todo o ciclo celular. Este alelo constitúe unha ferramenta de extrema utilidade, xa que permite desacoplar a resolución dos intermediarios recombinatorios da progresión do ciclo celular, para estudar as posibles consequencias deletéreas da activación temperán de Yen1. Unha aproximación inicial utilizando una versión do alelo YEN1ON inducible por galactosa demostrou que a sobreexpresión de YEN1ON é extremadamente prexudicial para as células, acadando a letalidade completa coa indución total do promotor GAL1, mentres que a sobreexpresión a niveis comparables do alelo silvestre non afecta a viabilidade. A maiores, obtivemos un resultado insospeitado co control catalíticamente inactivo da versión desregulada de YEN1, denominado YEN1ON-ND, xa que a sobreexpresión diste mutante tamén deriva na morte celular. Anque este resultado indicaría que a toxicidade de Yen1ON é independente da súa actividade nucleásica, propoñemos unha hipótese alternativa na que o efecto negativo dominante de YEN1ON-ND proviría da súa unión as estruturas de ADN ramificado obstaculizando o procesamento distas estruturas por outras nucleasas ou helicasas. Por lo tanto, as letalidades asociadas a sobreexpresión de Yen1ON e Yen1ON-ND terían unha orixe distinta en cada unha das versións desreguladas. En ambos casos, a morte celular vén acompañada dunha robusta activación do punto de control de dano ao ADN, tal e como se deduce pola fosforilación de Rad53 e a alteración do perfil normal do ciclo celular, cunha acumulación aparente de células na fase S. Isto podería ser compatible coa xeración de roturas de ADN ou a persistencia de longos tramos cadea sinxela nestes mutantes, o que concordaría coa súa capacidade para interferir na progresión normal da fase S. De importancia, un arresto prolongado do ciclo celular non é a principal causa de morte celular nestas estirpes, xa que a eliminación de RAD53 non pode suprimir esta letalidade. Cunha aproximación similar, testamos tamén se a letalidade puidera ser consecuencia do procesamento de intermediarios de recombinación temperáns, como os D-loops. Sen embargo, a toxicidade por sobreexpresión de Yen1ON e Yen1ON-ND non só non é suprimida no fondo xénico deficiente en recombinación rad52∆, senón que agrava o efecto tóxico da desregulación de Yen1. Isto 25 suxire que parte do dano asociado á desregulación e sobreexpresión de Yen1 pode ser reparado pola propia ruta de HR. Estas observacións, xunto cos resultados doutros membros do laboratorio, indicaron que a sobreexpresión de YEN1ON provoca unha inestabilidade xenómica masiva que non pode ser suprimida por mutacións noutros xenes. Polo tanto, co fin de obter máis información sobre as vías de metabolismo do ADN que poden verse afectadas pola actividade prematura da resolvase Yen1, decidimos empregar una nova aproximación, consistente na búsqueda de interaccións xenéticas entre YEN1ON expresada a niveis endóxenos e outros factores de reparación do ADN. Para iso, seleccionamos helicasas importantes para o procesamento de intermediarios de recombinación e replicación, cuxa deleción debería vir acompañada duns maiores niveis de potenciais substratos para Yen1ON. Con esta estratexia, descubrimos dous tipos principais de interaccións. Por un lado, a presenza de Yen1ON a niveis endóxenos parece ter un efecto beneficioso naquelas situacións onde a viabilidade celular se ve comprometida pola acumulación de intermediarios de recombinación sen resolver. Esta situación xorde como consecuencia da mutación simultánea de xenes de HR que levan a letalidade ou enfermidade sintética ou da combinación de mutacións individuais con axentes xenotóxicos, confirmando e ampliando resultados anteriores. Neste contexto, Yen1ON suprime eficazmente o mesmo conxunto de interaccións xenéticas que a eliminación de RAD52, demostrando co procesamento de intermediarios de recombinación por parte de Yen1ON rescata o fenotipo de morte por recombinación, pero non no contexto doutras letalidades sintéticas. Por outro lado, descubrimos que nalgúns contextos xenéticos a expresión de Yen1ON provoca unha maior hipersensibilidade a axentes xenotóxicos como MMS, HU ou CPT. En particular, a interacción entre pif1∆ é YEN1ON resulta especialmente interesante, xa que Pif1 é unha das principais helicasas para o procesamento de estruturas secundarias no ADN. Por tanto, decidimos afondar no mecanismo molecular subxacente desta interacción como unha forma de entender mellor a relevancia da regulación de Yen1. Non obstante, non moi avanzada esta liña de investigación, atopamos un obstáculo técnico que impediu o seu desenvolvemento posterior: os intentos de reproducir esta interacción xenética con calquera dos dous alelos clásicos de separación de función para PIF1, pif1-m1 (nuclear) e pif1-m2 (mitocondrial), fallaron dunha forma sistemática na recapitulación da interacción xenética entre pif1∆ e YEN1ON. Dados os efectos pleiotrópicos da eliminación completa, que combina defectos nucleares e mitocondriais, non puidemos descartar unha interacción indirecta debido a efectos metabólicos maiores causados polo fenotipo petite dos mutantes pif1∆, que poderían ser pouco informativos para os nosos propósitos. Aínda que conseguimos descartar a hipótese dunha interacción xenética entre YEN1ON e a condición petite, decidimos que, para entender as complexidades moleculares subxacentes a esta interacción, precisabamos as causas polas que pif1-m1 e pif1-m2 non reproducen a interacción pif1∆ YEN1ON. Os nosos resultados con pif1-m2 YEN1ON confirman outros estudos publicados nos que pif1tampouco recapitula totalmente os fenotipos pif1∆ atribuídos ás súas funcións nucleares, o que Tomás Lama Díaz 32 base to the sugar moiety, releasing the free base and leaving an AP site. Then, the AP site is either recognized and incised by AP endonucleases (Apn1, Apn2), yielding SSBs with a 5’ deoxyribose phosphate (5’-dRP) end, or by bifunctional glycosylases with AP-lyase activity, generating SSBs with a 3’-deoxyribosephosphate (3’-dRP) attached (Boiteux and Guillet, Figure 2. BER pathway. AP sites are generated by the incision of DNA N-glycosylases or by the spontaneous loss of a base (not depicted in the figure). Apn1 or Apn2 nick the DNA backbone to start the long-patch BER, generating a 5’-dRP that is removed by the sequential actions of Polδ and Rad27. Alternatively, in short-patch BER, the AP site is processed by AP lyases. The resulting 3’-dRP is trimmed by the diesterase activity of Apn1/Apn2, and the resulting gap filled by Polε. The final ligation of the compatible ends is mediated in both sub-pathways by the Cdc9 ligase 33 2004). These two different types of chemically modified breaks constitute the key determinant to channel these SSBs to long-patch BER (Figure 2. BER pathway, left branch) or short-patch BER (Figure 2. BER pathway, right branch), respectively. On the one hand, in long-patch BER, the presence of a 3’-hydroxyl group in SSBs with 5’-dRP primes DNA synthesis by Polδ, extending for 5-10 nucleotides. The resulting 5’-flap is processed by the 5’-flap endonuclease Rad27, leaving a nicked dsDNA molecule ready to be joined by Cdc9 ligase. On the other hand, in short-patch BER, the additional 3’-phosphodiesterase activity of AP endonucleases removes the blocked termini (3’-dRP) impeding DNA synthesis, leaving a 1-nt gap with a 3’-OH that is readily filled by Polε and subsequently ligated by Cdc9 (Boiteux and Guillet, 2004). 1.1.3 Nucleotide excision repair Nucleotide excision repair (NER) removes a variety of helix-distorting lesions produced by bulky adducts that can block the progression of DNA and RNA polymerases. In contrast to BER, NER mainly protects the DNA from the mutagenic effects of exogenous sources, such as UV-induced lesions and carcinogen-derived bulky adducts (Cadet et al., 2005; Friedberg, 2006), although it can also neutralize intramolecular crosslinks caused by endogenous oxidative agents (Kuraoka et al., 2000). Eventually, NER also constitutes an alternative to BER for the repair of AP sites and oxidized bases (Scott et al. 1999; Swanson et al. 1999). This pathway seems to function exclusively in the nucleus, since no NER activity has been detected in mitochondria (Canugovi et al., 2010). There are two well-established routes of NER in cells: global-genome NER (GG-NER) and transcription-coupled NER (TC-NER). After the initial steps of lesion recognition and recruitment of the NER machinery, both pathways converge into the same final steps, involving a dual-incision reaction in which a ssDNA fragment of 25-30 nt is released from the helix, followed by Pold-dependent gap-filling and ligation of the resulting nick (Boiteux and Jinks-Robertson, 2013). GG-NER (Figure 3) operates independently of the transcription status or chromatin structure, and begins with the opening of the double helix and the formation of a pre-initiation complex. First, the Rad4-Rad23-Rad33 complex recognises the distortion of the double helix by sensing thermodynamically unstable base pairs (Min and Pavletich, 2007). Next, the complex Rad7Rad16 remodels the chromatin through interactions with acetylating and ubiquitinating partners, while Rad4-Rad23-Rad33 recruits the transcription factor TFIIH to establish the preincision complex. The core repair complex is composed of 10 subunits, including the helicases Rad25 and Rad3, and three subunits with CAK kinase activity (Gibbons et al., 2012). The catalytic role of TFIIH during NER consists in unwinding the DNA flanking the lesion, which is driven by the opposing polarities of the ATPase/helicase activities of Rad3 and Rad25. After the DNA has been unwound, the pre-initiation complex is released and the single-strand binding protein RPA coats the exposed ssDNA to stabilize the pre-incision complex. The formation of the pre-incision complex is followed by the recruitment of two structure-selective endonucleases (SSEs): Rad2 and Rad1-Rad10. Rad2 and Rad1-Rad10 excise the damaged strand by dual-nicking near the branched ends of the repair bubble; Tomás Lama Díaz 34 Figure 3. NER pathway. Rad23-Rad4-Rad33 recognize the distortion in the double helix caused by bulky adducts, whereas Rad7-Rad16 remodels the chromatin. Rad23-Rad4-Rad33 recruits the TFIIH complex. Two helicases of the complex, Rad25 and Rad3 unwind the DNA flaking the lesion, with RPA coating the resulting ssDNA.Dual nicking by Rad2 and Rad1-Rad10 of the bubble excises the damaged strand. Repair is completed by DNA synthesis and ligation. TC-NER, not depicted in the figure, is coupled to RNA polymerase II. 35 Rad2 makes its incision 2–8 nt to the 3’ side of the lesion, while Rad1-Rad10 does it 15–24 nt to the 5’ side of the lesion (Evans et al., 1997). Once the damaged strand has been removed, the resulting gap is filled by either Polδ or Polε, and the nick ligated by Cdc9. 1.1.4 Mismatch repair DNA mismatch repair (MMR) is a highly conserved pathway with major roles during DNA replication, repair and recombination (Liu et al., 2017), as it promotes the correction of DNA non-Watson–Crick base pairs or small loops of extrahelical nucleotides. Mismatches arise during DNA synthesis due to misincorporation of nucleotides, while small loops and hairpins are usually a consequence of replication slippage by DNA polymerases; therefore, acting as a backup for the proofreading activity of DNA polymerase, MMR contributes to increase the fidelity of replication by several orders of magnitude, ensuring that daughter cells inherit a faithful copy of the parental genome (Jiricny, 2013). Mismatches can also occur within heteroduplex DNA during homologous recombination (HR); in that context, MMR monitors the identity between DNA interacting molecules during strand exchange and limits the use of non-identical sequences as repair templates (Jiricny, 2013). The MMR machinery detects the helical distortions caused by mismatches and excises a ssDNA tract in the newly-synthesized strands contaning the incorrect nucleotides, followed by gap fillin and ligation steps. In S. cerevisiae, two heterodimeric complexes - MutSa (Msh2-Msh6) and MutLa (Mlh1-Pms1) - repair the majority of replication errors, as detailed in Figure 4a. Mismatch binding and ATP promote the activation of MutSa and allow its interaction with MutLa, which is essential to nick the nascent strand. This provides an entry point for the 5’-3’ exonuclease Exo1 that will then remove a portion of the strand containing the mismatch. One of the key mechanistic steps for accurate repair is the correct discrimination between the nascent and template strands. When the mismatch occurs in the lagging strand, the discontinuity at the 5’ end of the Okazaki fragment allows the recognition of the nascent strand and enables Exo1 exonuclease activity; in fact, MMR removes more efficiently those mismatches appearing in the lagging strand (Lujan et al., 2012; Pavlov et al., 2003). However, the continuous synthesis on the leading strand requires a different discrimination strategy and the deliberate generation of a nick. In this case, the recruitment of the PCNA sliding clamp by MutSa and the endonuclease activity of MutLadependent on PCNA interaction - fulfil both these requirements (Kadyrov et al., 2006). Since PCNA is loaded asymmetrically on the fork by the replication factor C (RFC), it can promote MutLα incision only at the nascent strand in an ATPdependent manner (Pluciennik et al., 2010; Umar et al., 1996). The resulting nick will be used replication error after which repair will be corrected by new Polδ dependent-DNA synthesis The MMR machinery is not limited to MutSa and MutLa. MutSb (Msh2-Msh3) detects large insertion/deletion loops (IDLs) and, like MutSa, can also mark small IDLs for repair (Srivatsan et al., 2014) (Figure 4b). In addition, part of the MMR machinery has been adapted to recognize and cleave specific HR intermediates, like dHJ, during the meiotic cell cycle Tomás Lama Díaz 36 Figure 4. MMR pathway. (a) MutS (Msh2-Msh6) recognises the distortion of the double helix, and recruits PCNA and MutL (Mlh1-Pms1) to the site of mismatch. MutL creates entry points for Exo1 by nicking the damaged strand, resulting in a gap that is filled by Polδ or Polε. (b) MMR machinery is composed of different complexes specialized in several types of DNA damage recognition (MutS a and MutSb ), and with meiotic-specific functions (MutSγ and MutLγ). Additionally, budding yeast is endowed with a homodimeric complex for MMR in the mitochondria (Msh1-Msh1) 37 (Cannavo et al., 2020; Kulkarni et al., 2020; Rogacheva et al., 2014)(Figure 4b). Finally, MMR also promotes error-free replication of the mtDNA, with Msh1 being found exclusively in the yeast mitochondria (Reenan and Kolodner, 1992). While the role and activities of MMR nuclear complexes have been profusely characterized, much less is known about the putative role of this specific mitochondrial subunit (Figure 4b). The lack of a Msh1 homolog in mammalian cells could explain, at least partially, the higher mutation rates of human mtDNA (Reyes et al., 2015). 1.1.5 DSB repair: NHEJ, HR and alternative pathways DSBs are frequently categorised as one of the most cytotoxic types of DNA lesions, since the simultaneous disruption of the continuity of both DNA strands precludes the use of the complementary strand for accurate repair. Spontaneous DSBs are mostly pathological and may be caused by the actions of ROS species from the oxidative metabolism, environmental IR, rogue endogenous nucleases or the collision of a replisome with an unrepaired SSB (Chatterjee and Walker, 2017; Lieber, 2010). However, a small subset of DSBs fulfil physiological roles by triggering programmed recombination (or other types of repair) events, such as those in meiosis, budding yeast mating-type inter-conversion, or lymphocyte and neuronal development in mammalian cells (Haber, 2012; Keeney et al., 1997; Lieber et al., 2006). To avert the genome instability associated to DSBs, cells have evolved two main strategies: first, the direct ligation of the DNA by non-homologous end-joining (NHEJ); second, the use of homologous templates to restore genetic information by HR-dependent pathways. Additionally, two other pathways - single strand annealing (SSA) and microhomologymediated end-joining (MMEJ) - provide alternative solutions to repair DSBs based on short homology stretches, although in yeast these are quite restricted under physiological conditions, with DSBs being primarily channelled towards NHEJ or HR (Sfeir and Symington, 2015). Pathway selection is determined by two variables: the chemical status of the DSB ends and the availability of a suitable template for HR. DSBs can be classified as chemically “clean” or “dirty”, depending on their ability to be directly religated. Clean DSBs present the canonical 5’-phosphate and 3’-hydroxyl moieties required for ligation and are usually a consequence of the unscheduled activity of endonucleases. Conversely, dirty DSBs appear mostly as byproducts of radiation or abortive topoisomerase reactions. These may present chemically modified ends or protein blocks, thus requiring the end-processing activity of different enzymes before ligation (Symington and Gautier, 2011). 1.1.5.1 NHEJ Canonical NHEJ, the simplest mechanism to rejoin DSBs, involves a set of core proteins required for religation, which in S.cerevisiae includes the yKu70-yKu80 (yKu) heterodimer, the Mre11-Rad50-Xrs2 (MRX) complex, Dnl4 (DNA ligase IV), as well as the accessory factors Lif1 and Nej1 (Daley et al., 2005; Lieber, 2010). Additionally, other supporting factors aid in the end-processing of chemically modified DSBs to facilitate the activities of the core proteins. Repair starts with the binding of yKu to both DNA ends to protect them from degradation (Figure 5). Then, the MRX complex is recruited to tether both ends of the DSB (Chen et al., 2001). Singularly, MRX is also involved, unlike any of the other NHEJ core Tomás Lama Díaz 38 factors, in the early steps of HR (described in section 2.5.2). Once stabilized by yKu and MRX, the chemistry of the DNA ends will determine the ensuing repair stages. If the ends are chemically modified or blocked by a protein, an additional processing step that further trims the DNA ends is required prior to ligation. For that purpose, NHEJ is endowed – specially in human cells - with nucleases and polymerases able to act on a wide range of DNA end structures, conferring NHEJ a high degree of mechanistic flexibility (Lieber, 2010). In S. cerevisiae, this involves the activity of Tpp1 to remove 3’-phosphate groups; Tdp1, to hydrolyse the 3’ and 5’-phosphotyrosyl bonds in topoisomerase-protein adducts; the polymerase Pol4 for gap-filling and endonuclease Fen1 for the removal of 5’-flaps (Bahmed et al., 2010; Tseng and Tomkinson, 2004). However, S. cerevisiae lacks specific end-processing nucleases as part of this NHEJ toolkit and, therefore, its capacity to deal with dirty DSBs is relatively limited (Symington and Gautier, 2011). In the final step , yKu recruits Dnl4-Lif1Nej1 (Tseng and Tomkinson, 2004; Wilson et al., 1997) to catalyze the reformation of phosphodiester bonds. Interestingly, Cdc9 (DNA ligase I) cannot be used for ligation in NHEJ and, conversely, Dnl4 activity is restricted to this pathway, without any role in supporting DNA replication. Figure 5. DSB repair by NHEJ. yKu binds dsDNA ends and protects them from degradation by nucleases. Additionally, MRX tethers both ends together, and recruit the Dnl4 ligase with its accessory factors Lif1/Nej1. If the ends cannot be directly ligated, yeast is endowed with several enzymes to process the blocked ends, such as Pol4, Tdp1, Tpp1 or Rad27. 39 Although NHEJ has been traditionally described as highly mutagenic, this vision has been challenged during the last decade; NHEJ, which accounts for the majority of the repair of DSBs in metazoans, is not intrinsically inaccurate (Bétermier et al., 2014). Its probability to introduce mutagenic deletions or insertions is dictated by the structure of the DNA ends, and not by the intrinsic properties of core NHEJ machinery; while DSBs with compatible ends are faithfully rejoined, the ability of NHEJ to restore DNA integrity even when the DSBs display imperfect ends comes with the risk of mutagenic insertions or deletions at the break site. In conclusion, repair by NHEJ is fast, cell cycle independent and, if no end-processing is required, highly accurate. 1.1.5.2 Homologous recombination Homologous recombination (HR) is a high-fidelity DSB repair pathway that restores DNA continuity by retrieving any lost genetic information from a homologous donor, like the sister chromatid or the homologous chromosome. HR is tightly linked to DNA replication and the S/G2 phases of the cell cycle, since both of them cooperate to ensure faithful replication and repair. HR enables tolerance pathways to bypass damage, and participates in the protection and repair of stalled or broken forks (Heyer, 2015; Pâques and Haber, 1999). Lastly, the repair of programmed meiotic DSBs by HR generates chiasmata, the physical linkages between homologous chromosomes required for bipolar chromosome segregation during the reductional, first meiotic division (Hunter, 2015). As opposed to end-joining, HR is an intricate mechanism which entails the combined activities of various recombinases, helicases and nucleases that increase the accuracy of DSB repair at the expense of speed (Mao et al., 2008). To fulfil its different roles, HR comprises multiple sub-pathways for repairing both oneand two-ended DSBs. In all cases, HR begins with the nucleolytic degradation of DNA ends at the break to expose long tracts of 3’-OH ssDNA, a process termed end resection. The MRX complex binds to the DNA ends and initiates resection with an endonucleolytic incision to release a 5’-ending oligonucleotide. Resected ends are further extended by Exo1 and the Sgs1 helicase working together with Dna2 nuclease (Symington and Gautier, 2011) (Figure 6, step 1). This extended DNA resection commits DSB repair to HR, since resected ends become poor substrates for yKu binding and cannot be directly ligated, hence inhibiting the NHEJ pathway. Resection is followed by RPA coating to protect the ssDNA, Rad52 recruitment, and Rad52mediated displacement of RPA to allow the formation of the Rad51 nucleofilament, which scans the genome for homologous sequences (Figure 6, step 2). Eventually, this leads to the establishment of a synapse between the Rad51 nucleofilament and the homologous sequence, the invasion of the homologous duplex and the formation of a displacement loop (D-loop). In some cases, the Srs2 helicase can both disrupt Rad51 nucleofilaments and disengage D-loops to avoid unscheduled recombination events (Krejci et al., 2003; Veaute et al., 2003). Next, the invading strand primes DNA synthesis, extending the D-loop and copying the genetic information from the donor molecule (Figure 6, step 3). After this point, HR can follow three different routes: synthesis-dependent strand annealing (SDSA), break-induced replication (BIR) or classical double-strand break repair (DSBR) (Figure 6, step 4). Pathway selection will define the outcome of HR, which can be classified as a noncrossover (NCO) or a crossover Tomás Lama Díaz 40 Figure 6. DSB repair by HR. (1) Resection exposes the 3’-ssDNA required to invade homologous sequences and prime DNA synthesis. (2) Invasion is mediated by Rad51 and Rad52 recombinases, whereas Srs2 strips Rad51 filament to impede D-loop formation. (3) If a second end is available, the extended D-loop can be disrupted by (4) SDSA helicases to produce NCO. Alternatively, one-ended DSBs can restore the missing genetic information via BIR (5). Finally, second-end capture, DNA synthesis and ligation leads DSBR pathway (6), characterized by the formation of a dHJ. This metastable intermediate can be eliminated via dissolution (7) or resolution (8), producing NCOs or a mixture of NCOs/COs respectively. 41 (CO), depending on the absence or presence of reciprocal exchanges between the broken and donor molecules, respectively. Most of the extended D-loops in somatic cells are disrupted by the helicases Srs2 (Liu et al., 2017) or Mph1 (Ira et al., 2003; Prakash et al., 2009) and repaired via SDSA to yield exclusively NCOs. Alternatively, the second end of the broken duplex can be engaged to stabilize the D-loop (Figure 6, step 4) and ligated (Figure 6, step 6), leading to the formation of a new recombination intermediate, the double Holliday junction (dHJ) (Szostak et al., 1983). dHJs can be dissolved by the action of the STR complex to generate NCOs (Figure 6, step 7) (Cejka et al., 2010c; Ira et al., 2003, p. 2; Wu and Hickson, 2003), or undergo nucleolytic resolution to yield a mixture of NCO and CO products (Figure 6, step 8) (Wyatt and West, 2014). The third pathway, BIR (Figure 6, step 5), copes with one-ended DSBs, which appear mostly as a consequence of fork breakage during DNA replication. During BIR, the extended D-loop turns into a migrating D-loop that supports extensive recombinationcoupled DNA synthesis. Two factors have been identified as essential for BIR activity: the nonessential Polδ subunit Pol32 (Lydeard et al., 2007) and the Pif1 helicase (Wilson et al., 2013). BIR implies a type of conservative DNA replication where both the newly synthesized leading and lagging strands remain associated, leading to high levels of loss of heterozygosity (LOH). 1.1.5.3 Homology-mediated repair: MMEJ and SSA MMEJ and SSA are minor pathways that, in addition to HR and NHEJ, contribute to DSB repair. Both involve the alignment of microhomologies or homologies flanking the DSBs. Although to a different extent, they rely on the resection machinery to expose the internal homologies and, therefore, compete with HR for repair substrates. In this sense, MMEJ, SSA and HR can be grouped together as homology-directed repair (HDR) pathways. Even if MMEJ and SSA pathways are intrinsically mutagenic due to theit tendency to lose genetic information, these mechanisms could still promote a certain degree of faithful DSB repair in highly repetitive genomes (Chang et al., 2017; Sfeir and Symington, 2015). MMEJ (Figure 7) is an error-prone, alternative route to end-joining (Yu and Gabriel, 2003), independent of canonical NHEJ factors such as the yKu heterodimer, Dnl4 or Pol4. Instead, it requires the activity of the MRX complex or the long-resection pathway (Sgs1-Dna2 and Exo1) to expose complementary regions of 6-14 nt of homology before joining (Lee and Lee, 2007; Villarreal et al., 2012); due to its unique requirements, MMEJ is considered a pathway that stands between NHEJ and HR. Following resection, the homologous regions will be paired independently of recombinases, and the Rad1-Rad10 endonuclease will remove any heterologous flaps formed during the annealing, prior to ligation (Ma et al., 2003). Otherwise, if the resection machinery exposes ends longer than 15 nt without finding microhomology repeats, the resulting ssDNA stretches will be bound by RPA, which inhibits the annealing of microhomologies and channels repair towards SSA or HR (Deng et al., 2014). The physiological relevance of MMEJ remains poorly defined in S. cerevisiae. In this sense, its low efficiency suggests that it could simply act as a backup mechanism, with minimal functionality unless NHEJ or HR are compromised, given that RPA both inhibits MMEJ and promotes extensive resection by removing secondary structures (Sfeir and Symington, 2015). In mammalian cells, MMEJ is more prevalent than in yeast, which is reflected by the existence of Tomás Lama Díaz 48 depends on the generation of a Rad51 nucleofilament to invade the sister chromatid and bypass the damage, Srs2 is locally downregulated at damaged forks to relieve its inhibition on Rad51 nucleofilament formation. This is achieved by unloading SUMOylated PCNA and by targeting Srs2 for proteasomal degradation via the E3-ligase Slx5-Slx8 complex (Fumasoni et al., 2015). Finally, the SP is the last resort to bypass lesions during G2/M and operates independently of PCNA modifications. During a normal S phase, the presence of Srs2 inhibits this HR-dependent pathway. Moreover, even upon replication stress and downregulation of Srs2, the SP remains inactive due to an additional inhibitory layer mediated by Mgs1 that promotes the TS pathway (Jiménez-Martín et al., 2020). The stringent downregulation of this pathway in S phase suggests that toxic outcomes might ensue should SP be active during replication and points towards a specialized role in bypassing particularly persistent lesions (Branzei and Szakal, 2016). In conclusion, the DDT pathways ensure timely replication and increased resistance to genotoxic agents and replication stress. Notwithstanding, they may affect the fidelity of DNA replication. For that reason, the routes of TLS, TS and SP are tightly regulated to prevent undesired outcomes and minimize their potential to cause mutations. Since the DNA synthesized past lesions will be used as a template in the following rounds of replication, fidelity of lesion bypass and consequently, pathway selection, are essential to sustain the integrity of the genome. 1.1.6.2 Fork reversal Replication fork reversal requires the hybridization of the two newly-synthesized strands and the re-annealing of the parental strands, hence remodelling the replication fork (a three-way junction) into the HJ-like chicken-foot structure, with a fourth regressed arm (Neelsen and Lopes, 2015). In S. cerevisiae, like many HR-dependent transactions at the fork, fork reversal is considered mostly pathological (Neelsen and Lopes, 2015). In this sense, fork reversal is inhibited by checkpoint kinases to maintain fork stability in response to replication stress (Sogo et al., 2002) and disfavoured by the re-priming events downstream the lesion (Fumasoni et al., 2015). Therefore, fork reversal is not considered a general mechanism to protect stalled forks in yeast, in contrast to mammalian cells where reversed forks are detected even in an unperturbed S phase. However, fork reversal may still constitute a physiological pathway to assist DNA damage bypass (Branzei and Foiani, 2010; Saugar et al., 2014) or transiently stabilize stalled forks under special circumstances (Figure 11a,b) (Branzei and Szakal, 2016), namely: i) re-priming downstream the lesion is impaired (Fumasoni et al., 2015), ii) lesions blocking the progression of the replicative MCM helicase (Carr and Lambert, 2021); iii) the presence of excessive torsional stress (Ray Chaudhuri et al., 2012); or iv) low probability of a converging fork rescuing the 49 Figure 11. Possible DNA transactions during fork reversal. (a) Lesions blocking the replicative helicase (green rectangle) or the leading-strand polymerase (yellow star) may lead to fork reversal. (b) Fork reversal may transiently stabilize the replisome until the damage is removed. (c) DNA synthesis using lagging-strand as template can promote lesion bypass in lesions blocking the leading-strand polymerase. Reversed-fork breakage (d) or resection of nascent strand (e) can promote invasion to restart replication via BIR or RDR, respectively. Tomás Lama Díaz 50 stalled replisome (Figure 11a). These situations may arise when the replisome encounters DNA crosslinks, protein-DNA adducts or when it travels through highly transcribed genes, originpoor regions or long stretches of unidirectional replication (Carr and Lambert, 2021). Alternatively, the transient nature of reversed forks may account for the low levels detected and mask a more prevalent role of this reconfiguration (Neelsen and Lopes, 2015). In yeast, fork reversal is poorly understood compared to mammalian cells, but several factors have been implicated in this process under different conditions. Various helicases and DNA translocases can drive fork reversal in vivo and in vitro, including the SWI/SNF motor protein Rad5, the FANCM homolog Mph1 and the Pif1-family helicases Pif1 and Rrm3. Contrarily, fork reversal may be counteracted by the resection activity of nucleases like Exo1, the MRX complex and Dna2 (Meng and Zhao, 2016). After fork remodelling into a four-way junction, fork reversal may lead to different outcomes depending on the downstream molecular events and the type of lesion causing the blockage (Figure 11) (Meng and Zhao, 2016). First, reversal may have a transient, stabilizing effect, since stalled replication forks with long ssDNA regions are more susceptible to breakage (Figure 11b). Once the lesion has been removed, fork restart or rescue by a converging fork enable the completion of replication. Second, if the lesion stalls the leading strand polymerase, the nascent lagging strand can serve as a template for DNA synthesis, thus helping bypass the lesion on the parental strand (Figure 11c). Third, if the reversed fork is cleaved by a structure-selective endonuclease, further 5´-3´ resection of the one-ended DSB may recruit the HR recombinases to restart error-prone replication via BIR (Figure 11d). In the absence of fork breakage, recombination can still promote fork restart after resecting one of the nascent strands to allow binding of HR proteins and invasion ahead of the lesion. This pathway, commonly known as recombination-dependent restart (RDR) shares with BIR its dependence on Pif1 and Pol32 (Figure 11e). (Appanah et al., 2020). 1.2 Coordination of DNA repair by DNA damage checkpoint and cell cycle kinases DNA repair must be strictly regulated to coordinate pathway selection, avoiding substrate competition and promoting high-fidelity repair, while minimizing detrimental outcomes. Extensive research over the last thirty years has unequivocally established the prominent role of post-translational modifications, such as checkpointand cell cycle-dependent phosphorylation, SUMOylation and ubiquitylation, in the spatio-temporal regulation of DNA repair, as already hinted in the previous sections. 1.2.1 DNA damage checkpoint Coordination of the various DNA repair systems involves the crosstalk between the cell cycle and the DDC kinases; together, they modulate how different pathways interact to optimize the repair outcome, depending on the type of lesion and the cell-cycle-related substrate characteristics (Lanz et al., 2019). Eukaryotic cells respond to different types of DNA damage with the activation of the DDC, a signalling transduction pathway mediated by kinases that sense 51 Figure 12. Overview of DDC response in budding yeast. (a) The DDC recognises DNA structures that arise during DNA damage and replication, including broken ends and 3’-ssDNA stretches coated by RPA. Signalling is started by apical PIKK kinases, with adaptor proteins facilitating phosphorylation of downstream checkpoint kinases. Both apical and downstream kinases phosphorylate multiple targets to coordinate the cellular response against DNA damage which includes cell cycle arrest, repair of the lesions and the stability of replication forks, with Rad53 participating in the majority of phosphorylation events. (b) Recruitment of apical kinases and activation of downstream effectors via Mrc1 and Rad9 adaptors. Ddc2 recruits Mec1 to RPA-coated ssDNA at replication forks and 3’ ssDNA stretches after 5’-3’ resection of broken ends, whereas interaction with the MRX complex targets Tel1 to DSBs. Adaptor proteins Mrc1 and Rad9 mediate recruitment of downstream kinases to allow their activation by Mec1/Tel1s dependent phosphorylation. Tomás Lama Díaz 52 DNA damage and coordinate DNA repair events with otherrelevant cellular processes. The DDC can be activated at three different stages of the cell cycle in budding yeast. First, DDC activation in G1 prevents entry into S phase until DNA regains a pristine condition for replication (Siede et al., 1994). Second, during S phase, the DDC coordinates fork progression and DNA repair (Paulovich and Hartwell, 1995). Third, the existence of lesions in replicated DNA triggers the G2/M checkpoint, preventing mitotic entry until the damage has been fixed (Weinert and Hartwell, 1988). However, it is important to note that checkpoint activation does not arrest cell cycle progression indefinitely; after some time, in a process commonly referred as adaptation, cells will inactivate the checkpoint signal and continue with cell division without repairing persistent DNA lesions (Paulovich et al., 1997; Sandell and Zakian, 1993). In budding yeast, checkpoint activation is mediated by two different types of DDC kinases: the apical phosphatidylinositol-3 kinase-related kinases (PIKK) and effector checkpoint kinases (Figure 12a). As a consequence of DNA damage and/or replication stress, the apical PIKKs Mec1 and Tel1 relocate to sites containing structures like ssDNA stretches and broken DNA ends, respectively. Then they recruit and phosphoryled downstream checkpoint effector kinases (Rad53, Dun1 and Chk1) at consensus S/T-Q motives (Smolka et al., 2007), with the assistance of adaptor proteins like Mrc1 and Rad9. Mrc1 serves as a mediator during DNA replication (Alcasabas et al., 2001), whereas Rad9 has a more prominent role in G1 and G2 (Siede et al., 1993; Weinert and Hartwell, 1988). Once that apical and downstream kinases have been have been activated, they cooperate to phosphorylate key substrates for the spatiotemporal control of the checkpoint responses, including repair of DNA damage, cell-cycle arrest, regulation of dNTP levels, stabilization of replication forks, inhibition of origin firing and transcriptional control (Lanz et al., 2019). The apical PIKK kinases are recruited to damaged DNA by different mechanisms. On the one hand, Tel1 is targeted to DSBs by the MRX complex (Nakada et al., 2003). Reciprocally, Tel1 binding stabilizes MRX at the DNA ends to sustain DSB repair. On the other hand, Mec1 is loaded onto ssDNA by its integral partner Ddc2, which interacts with RPA (Rouse and Jackson, 2002); hence, the trigger for the Mec1-Ddc2 heterodimer recruitment is the presence of ssDNA stretches coated by RPA, which appear during resection of DSBs or replication stress (Figure 12a). Interestingly, Mec1 activation in S phase requires higher levels of RPA-coated ssDNA compared to G1 or G2. This specific increase in the threshold for Mec1 activation during DNA replication is most likely enforced to prevent unnecessary checkpoint activation by ssDNA generated at functional replication forks (Shimada et al., 2002; Tercero et al., 2003). Phosphorylation of downstream kinases by Mec1/Tel1 requires their physical proximity, a situation facilitated by the two checkpoint adaptors Rad9 and Mrc1 (Figure 12b). After recruitment to DNA damage sites, Mec1 and Tel1 phosphorylate proximal targets close to the lesion, like the histone H2A (Downs et al., 2000) or the Rad17-Mec3-Ddc1 (9-1-1) complex (Paciotti et al., 1998), respectively. This promotes Rad9 recruitment and subsequent phosphorylation by Mec1/Tel1 (Vialard et al., 1998) which, in turn, allows Rad53 binding (Gilbert et al., 2001; Schwartz et al., 2002) and activation by Mec1/Tel1-dependent phosphorylation (Sanchez et al., 1996; Sun et al., 1996). Alternatively, 53 Figure 13. Phosphorylation events involved in DDC activation in budding yeast. Schematic representation of cellular targets phosphorylated by DDC kinases to regulate cell-cycle arrest, fork stability and coordinate several DNA repair pathways. Tomás Lama Díaz 54 ssDNA accumulation due to replication stress does not require active mechanisms to assemble the Mrc1 adaptor at the fork, since Mrc1 is already a central component of the replisome and can be directly phosphorylated in situ. Analogously to Rad9, Mrc1 phosphorylation allows Mec1-dependent Rad53 phosphorylation to initiate the activation of downstream effectors (Katou et al., 2003; Osborn and Elledge, 2003). Notably, other kinases such as Cdc28 can also catalyse phosphorylation events involved in adaptor recruitment, although these events are often not induced by DNA damage. Once active, both the effector kinases Rad53, Chk1 and Dun1, as well as the apical kinases Mec1 and Tel1, can target an extensive network of substrates involved in different cellular responses, with Rad53 mediating the majority of checkpoint-related phosphorylation events in budding yeast (Sanchez et al., 1999). Relevant pathways and particular substrates phosphorylate by checkpoint kinases are summarized in Figure 13. In addition to phosphorylating several regulators of the cell cycle to mediate its arrest, DDC kinases also phosphorylate different substrates to mediate the protection and restart of stalled replication forks. Several components of the DNA repair machinery are also regulated by checkpoint kinases to bias pathway selection or enhance repair efficiency in a context-dependent manner (Lanz et al., 2019). 1.2.2 DNA repair throughout the cell cycle The mitotic cell cycle is defined as the ordered series of events occurring in distinct phases that ultimately results in the division of a cell into two daughters (Murray and Hunt, 1993). From a DNA-centric perspective, the cell cycle is defined by two paramount landmarks: the DNA replication in S phase, when the bulk of the nuclear genome is duplicated, and the chromosomal segregation in M phase, when the duplicated genome is condensed, sorted and equally distributed to the cellular offspring (Hustedt and Durocher, 2017; Murray and Hunt, 1993). Cell-cycle progression is orchestrated by the periodic activation and inactivation of cyclindependent Ser/Thr kinases (CDKs). Budding yeast encodes five CDKs: Cdc28, Pho85, Kin28, Ssn3 and Ctk1. However, only Cdc28 (CDK1) functions as a master regulator to drive the cell cycle (Mendenhall and Hodge, 1998). Accordingly, nine cyclins are classified with respect to the main cell-cycle stage in which they interact with Cdc28: the G1-phase cyclins (Cln1, Cln2 and Cln3), the S-phase cyclins (Clb5 and Clb6), and the G2/M-phase cyclins (Clb1, Clb2, Clb3 and Clb4) (Bloom and Cross, 2007). Specific cyclin-Cdc28 heterodimeric complexes phosphorylate multiple target proteins to impel cell cycle progression in an ordered manner. In addition to this canonical role, mounting evidence from the last two decades has demonstrated further roles for the cyclin-CDK complexes in transcription, cell metabolism and DNA damage repair (Hydbring et al., 2016). In this sense, integration of the peculiarities of different cell-cycle phases into the regulatory network that coordinates DNA replication, repair and segregation is essential to maintain genome stability. As chromatin engages in different transitions throughout the cell cycle, an accurate feedback between cell cycle progression and the various DNA repair machineries is required to adjust repair pathway selection to these different chromatin states in the G1-S-G2M phases (Hustedt and Durocher, 2017). 55 The primary sources of DNA damage during G1 are endogenous ROS, chemical agents, IR or UV, leading to damaged bases, bulky adducts, nicks or DSBs. Importantly, this damage must be repaired before the onset of replication to avoid replication stress, with BER, NER and NHEJ as the main actors at this stage. While BER and NER operate throughout the cell cycle, NHEJ is specific for G 1. Despite the superior fidelity of HR, the highly compacted chromatin and the absence of sister chromatids makes NHEJ the selected DSB repair pathway during G1, as employing the homologous chromosome as a template for recombination might lead to loss of heterozygosity. As cells enter S-phase and start replicating their genome, they face new challenges, including nucleotide misincorporations and polymerase slippage at repetitive sequences. Therefore, MMR acquires increased relevance in this phase to remove pairing mismatches and small loops due to insertions or deletions. NER and BER remain functional in this phase, with BER aiding in the removal of misincorporated uracils and ribonucleotides by replicative polymerases, a subpathway of BER known as RER (Ribonucleotide excision repair). Moreover, the S-phase is characterized by the accumulation of ssDNA nicks and gaps, fork stalling and even DSBs associated with replication stress. However, DNA replication also provides the most suitable substrate for accurate DNA repair: the sister chromatid. Therefore, during this phase, HRdependent pathways participate both in the repair of DSBs at broken forks, but also of lesions impeding fork progression or gaps behind the fork through TS or fork reversal. Gaps and DSBs left unattended during the S phase need to be solved before mitosis. Therefore, in G2/M, HR still mediates repair of the remaining lesions. Importantly, two additional DDT mechanisms, TLS and the salvage pathway, despite being more error-prone than HR, come into play at this stage to contribute to the removal of persistent lesions and completion of underreplicated regions. Finally, if DSBs occur arises during chromosome segregation, homology search may be impeded by the highly compact chromatin; under these circumstances, if the checkpoint does not arrest cell cycle, DSB repair will be mediated by NHEJ in the next cycle. 2. HOMOLOGOUS RECOMBINATION AND ITS CELL-CYCLE REGULATION 2.1 DSB end resection and pathway choice The exonucleolytic degradation of the DSB 5’ ends, known as end resection, results in the generation of the 3’-ssDNA overhangs that act as a landing pad for the HR machinery and initiate the search for homology (Gnügge and Symington, 2021). Beyond its role in creating the substrate for HR, end resection is crucial to regulate DSB repair pathway choice and, through RPA coating of the 3’ ssDNA ends, to trigger the Mec1-dependent DDC (Shiotani and Zou, 2009). Currently, end resection is considered a two-step mechanism, in which the complex -together with Sae2/CtIPcatalyses the step of DNA-end processing and resection (short-range resection). This initial step commits DSB repair to HR, whereas resected tracts are further extended in a second step mediated by the combined activities of Exo1 and Dna2-Sgs1 (long-range resection). Short resection, in contrast to long resection, is slower and restricted to the vicinity of the DNA ends. However, the short resection machinery is endowed with a high flexibility to process different types of DNA ends, including those with non-canonical MRX-N Tomás Lama Díaz 56 structures, such as secondary structures or protein adducts (Cejka and Symington, 2021). Due to its impact on pathway choice, the end resection machinery is tightly regulated by the DDC and cell-cycle kinases, which warrant an Sand G2-specific activation of HR, when the sister chromatid is available for repair. The MRX/N complex and its regulatory subunit Sae2/CtIP are highly conserved from yeast to humans. The MRX/N complex is formed by two subunits of each of its components: Mre11, Rad50 and Xrs2/NBS1. In vitro, Mre11 is a manganese-dependent exonuclease that degrades dsDNA with 3’-5’ polarity, the opposite of that expect for end-resection to yield the 3’-OH ssDNA ends capable of priming new DNA synthesis. Moreover, 5’-3’ resection combined with the 3’-5’ exonuclease activity of Mre11 would lead to the loss of genetic information, which has been demonstrated to be minimal in vivo. The solution to this apparent paradox came after the discovery of the importance of Mre11 cofactors in turning Mre11 into an endonuclease with affinity for 5’-terminated DNA strands. Structural studies have shown that Rad50 blocks Mre11 access to DNA but, upon ATP binding and hydrolysis, the complex undergoes a conformational change that displaces Rad50, allowing Mre11 to access DNA (Deshpande et al., 2014). In contrast, Xrs2 does not fulfil any catalytic function, but it is required to translocate the complex into the nucleus (Oh et al., 2016; Tsukamoto et al., 2005). However, its mammalian counterpart, NBS1, plays a more predominant role, since the physical interaction with NBS1 is required to activate MRE11 (Anand et al., 2019). Finally, the regulatory subunit Sae2/CtIP stabilizes the cutting state conformation of the MRX complex, stimulating Mre11 nuclease (Marsella et al., 2021) As described in section (1.1.5.1), MRX/N is one of the first factors recruited to DSBs, either by direct binding or, alternatively, by sliding along the DNA due to its capacity to bypass protein barriers, including nucleosomes (Myler et al., 2017). After reaching the DSB, MRX/N initiates end resection by nicking the 5’ strands upstream the DNA ends and resecting back towards the DSB with its 3’-5’ exonuclease activity (Garcia et al., 2011). The distance of to 5’ nick depends on the impediment blocking the ends, varying from 15-45 nt away in vitro to ~300-400 nt from the ends (Garcia et al., 2011; Reginato et al., 2017; Wang et al., 2017), and probably proceeds in a stepwise fashion (Cannavo et al., 2019). The first incision is critical in those DSBs with dirty ends; in fact, Mre11 5’ nicking activity can be stimulated in vitro by a variety of protein blocks, including RPA, nuclease-dead restriction enzymes or yKu (Reginato et al., 2017; Wang et al., 2017). Moreover, Mre11 itself can also stimulate the endonuclease activity of close MRX complexes (Cannavo et al., 2019). Contrary to S. cerevisiae, MRN activity is essential for shortrange resection of both dirty and clean ends in mammalian cells (Sartori et al., 2007). MRX 5’ nicking activity creates entry points for the long-resection machinery to further extend resection tracts (Figure 14). Long-resection is mediated by two partially redundant enzymatic activities, the 5’-3’ exonuclease activity of Exo1, and the endonuclease activity of Dna2, which works together with the helicase Sgs1 and its accessory factors (Top3, Rmi1). Exo1/EXO1 belongs to the Rad2/XPG nuclease family and preferentially degrades DNA with 3’-protruding ends, consistent with its role in the long-resection pathway (Tran et al., 2002). Alternatively, Dna2 is a bifunctional enzyme with RecB-like nuclease and superfamily I helicase domains that 57 only degrades ssDNA (Kao et al., 2004a). However, the nuclease activity overpowers the Dna2 helicase activity, which can be only detected in nuclease-deficient mutants (Levikova et al., 2013). Consequently, Dna2 needs a lead helicase to perform its role in DNA resection. All the known Dna2 partners in resection belong to the RecQ helicase family; Sgs1 in S. cerevisiae, and BLM or WRN in human cells (Cejka et al., 2010a; Gravel et al., 2008; Mimitou and Symington, 2008; Sturzenegger et al., 2014; Zhu et al., 2008). The activity of the lead helicase creates 5’ flaps that are quickly coated by RPA, which plays several roles in long resection. Figure 14. End-resection at blocked DSB ends. MRX-Sae2 nicks the DNA close to the ends, creating entry points needed to start resection, specially at DSBs blocked by protein adducts or secondary structures. Bidirectional resection proceeds via the exonuclease activity of the MRX complex in 3’-5’ direction up to the broken end, and via long-end resection in 5’-3’ direction. Long-end resection combines the exonuclease activity of Exo1, the endonuclease activity of Dna2, and chromatin remodelling by Fun30. The long 3’-OH ssDNA exposed tract is immediately coated by RPA to protect them from additional nucleolytic processing. Tomás Lama Díaz 64 2.3 Extended D-loop disruption: BIR and SDSA SDSA and BIR share their initial steps, involving strand invasion of one DSB end, establishment of a D-loop and DNA synthesis. In SDSA, after a short stretch of DNA synthesis, the newly synthesized strand is displaced from the D-loop so it can anneal to the second DSB end. Alternatively, BIR is characterized by the absence of a second DSB end, so cells resort to multiple rounds of invasion, D-loop formation and DNA synthesis that can reach the end of the chromosome. This creates a long ssDNA strand that is later converted into a double-stranded product in a conservative replication fashion (Saini et al., 2013; Stafa et al., 2014). Although the kinetics of strand invasion and leading strand synthesis is similar in BIR and SDSA, BIR requires additional processivity factors, probably due to the longer extension of the repair synthesis. The SDSA pathway produces exclusively NCO outcomes with minimal gene conversion, being the primary HR pathway during mitotic cell cycle. SDSA constitutes the primary HR pathway in somatic cells. Differently to the disruption of nascent D-loops, which is an inherently anti-recombination activity, disruption of extended Dloops promotes SDSA and NCO formation. SDSA starts with the previously described steps of strand invasion, formation of a nascent D-loop and extension by new DNA synthesis. Figure 17. Regulation of nascent D-loop stability in budding yeast. Rdh54 demarcates two pathways to regulate nascent D-loop formation, promoting disruption of nascent D-loops by the STR complex. The difference between Type I and Type II nascent D-loops and the mechanism by which Rad54 delimits both pathways are presently unknown. The thickness of the arrows represents pathway usage. 65 Figure 18. SDSA pathways. Srs2 and Mph1 helicases dismantle extended D-loops during SDSA to limit the extent of genetic information that is copied from the intact donor. Once the invading nucleofilament has been displaced, Rad52-Rad59 mediates re-annealing of the extended strand with its parental DNA molecule. Additional synthesis by Polδ-PCNA and ligation of the resulting nicks complete repair, yielding exclusively NCOs. Tomás Lama Díaz 66 DNA synthesis is initiated by Polδ, the main repair polymerase, although Polε can fulfil the same role in SDSA in the absence of Polδ (Lydeard et al., 2007). After extension, the D-loop is dismantled mainly by the 5’-3’ helicases Mph1 and Srs2. Mph1 accounts for the disruption of the majority of extended D-loops in budding and fission yeasts, as supported by genetic and biochemical experiments (Prakash et al., 2009; Sebesta et al., 2011; Sun et al., 2008). Srs2 has also been implicated in SDSA, playing a partially overlapping role with Mph1 (Mitchel et al., 2013; Prakash et al., 2009). Genetic profiles of mph1∆ and srs2∆ mutants using ectopic recombination systems demonstrate their ability to direct HR intermediates to NCO through SDSA, but with some differences: mph1∆ mutants shift repair from NCO to CO, whereas srs2∆ display a decrease in the number of NCO products with a concomitant defect in repair efficiency, without altering CO proportion (Ira et al., 2003; Mitchel et al., 2013). The repair outcomes in mph1∆ and srs2∆ strains suggest that, similarly to the disruption of nascent Dloops, Mph1 and Srs2 dismantle extended D-loops differently. Again, the particular characteristics of these two helicases may account for variation in length and structure between extended D-loops in vivo (Liu et al., 2017; Piazza and Heyer, 2019). Regarding one-ended DSBs, most of them arise during S-phase when the replisome encounters a nick. Alternatively, nucleolytic processing of stalled replisomes by SSEs and telomere erosion also contribute to their formation (Kockler et al., 2021). BIR was reported for the first time in the bacteriophage T4 (Kreuzer et al., 1995; Luder and Mosig, 1982), and subsequent studies in E.coli demonstrated its ability to restart collapsed replication forks (Asai et al., 1994; Heller and Marians, 2006) . BIR has also been extensively characterized in S. cerevisiae, employing genetic set-ups that unveiled the molecular mechanisms and particular enzymatic requirements of this pathway. Specifically, two BIR pathways have been described, commonly referred to as Rad51-dependent and Rad51-independent BIR. Both involve Rad52 activity, with Rad59 assisting Rad52 in strand annealing in the Rad51-independent pathway (Davis and Symington, 2004; Malkova et al., 2005, 1996). Rad51-dependent BIR is the major pathway in yeast and shares the same initial steps described for SDSA (Figure 19). DNA synthesis in BIR is mechanistically different from unperturbed. Similar to SDSA, Polδ is the main polymerase for the synthesis of the leading and lagging strands, with a rate approximately six times slower than S-phase replication (Deem et al., 2008; Lydeard et al., 2007; Smith et al., 2009). DNA synthesis proceeds through a migrating D-loop, where the newly synthesized strand is constantly displaced from the donor and accumulated as ssDNA (Figure 19). Moreover, BIR is asynchronous and conservative, unlike S-phase replication: eventually, the leading strand serves as a template for the synthesis of the lagging strand (Saini et al., 2013; Wilson et al., 2013). This unusual mechanism of synthesis plus the extent of the DNA that needs to be copied makes BIR reliant on additional factors. Pol32, a non-essential subunit of Polδ, supports the main role of this polymerase during BIR (Lydeard et al., 2007). Additionally, another factor essential for BIR, but dispensable for S-phase replication, is the 5’- 3’ helicase Pif1 (Saini et al., 2013; Wilson et al., 2013). Much like Pol32, the precise role of Pif1 is not clearly defined, with three non-mutually exclusive possibilities: i) to substitute for the Mcm2-7 replicative helicase in the unwinding of DNA ahead the migrating bubble; ii) to 67 Figure 19. Rad51-dependent BIR pathway. DNA synthesis during BIR proceeds via bubble migration. Like other steps of DNA synthesis during HR, Polδ is the main polymerase in this pathway. Long-range synthesis is supported by Polδ non-essential unit Pol32, and Pif1 helicase. Lagging strand synthesis is initiated by Polα-primase using the recently synthesized strand as template, resulting in conservative inheritance of newly synthesized DNA. Tomás Lama Díaz 68 unwind the newly synthesized leading strand from the donor and/or iii) to recruit Polδ to the BIR replisome. Finally, BIR is also dependent on the Polα-primase complex, which contributes to stabilize the synthesis of the leading strand. Recently, a new assay developed to monitor BIR synthesis has demonstrated that pol32∆ and pif1∆ cells are proficient in strand invasion and initiation of DNA synthesis, but the length of their BIR products is limited to 15 and 5 kb, respectively. Similarly, in cells with inactivated Polα-primase complex, BIR long-range synthesis is also affected, although synthesis can still be extended up to 25 kb (Liu et al., 2021). Interestingly, SDSA helicases Srs2 and Mph1 also play a role in BIR. Srs2 contributes to disrupting toxic intermediaries arising from the invasion of the ssDNA accumulated behind the migrating bubble (Elango et al., 2017), whereas Mph1 dissociates the invading strand, forcing BIR to proceed via multiple rounds of template switching and invasion (Stafa et al., 2014). Importantly, the basics of BIR are conserved in mammalian cells; POLD3 and PIF1, human orthologs of Pol32 and Pif1, are also essential to increase processivity during BIR replication (Costantino et al., 2014; Li et al., 2021). Although BIR provides a solution for the problem of one-ended DSBs, it is regarded as a source of genome instability that can lead to mutagenesis and genomic rearrangements. BIR errorproneness stems from its particular mechanism of DNA synthesis, which is highly susceptible to roadblocks and implies the persistence of long ssDNA tracts (Liu et al., 2021). Therefore, BIR-related sources of genome instability include: i) the long tracts of ssDNA exposed during leading strands synthesis, which are highly susceptible to chemical modifications; ii) frequent interruptions of DNA synthesis due to template switching events or upon encountering roadblocks, such as transcription bubbles; iii) prompt dissociation of newly synthesized strand from its template, hindering correction by MMR machinery and iv) resolution of the migrating D-loop by nucleases, which can trigger translocations and cascades of genome rearrangements. In particular, the frequency of frameshift mutations and base substitutions is greatly increased during BIR (Deem et al., 2011; Saini et al., 2013). If a second DSB end is available for repair, yeast cells employ several mechanisms to funnel extended D-loops to SDSA, diminishing BIR associated genetic instabilities. Mph1 promotes SDSA by unwinding the extended D-loop and enabling the annealing of the extended strand with the second DSB end. This step is mediated by the annealing activity of Rad52-Rad59 (Mehta et al., 2017; Pham et al., 2021). In addition, MRX contributes to sway extended D-loops towards SDSA by holding DSBs together and allowing synchronous resection of two ends (Pham et al., 2018). Cdk1-dependent phosphorylation also modulates pathway selection between BIR and SDSA. In this sense, phosphorylation of Srs2 by Cdk1 avoids its unscheduled SUMOylation and promotes its role in D-loop dismantlement for SDSA (Saponaro et al., 2010). Cdk1-dependent phosphorylation of Rad52 also assist in end-capture after D-loop formation by stimulating its strand annealing activity. 2.4 Second-end capture: Dissolution and Resolution of dHJs 69 When the extended D-loop is not disrupted by SDSA helicases, DNA synthesis can continue and, hence, either enlarge the D-loop or promote its migration. This allows the displaced strand to expose the sequence tract that is complementary to the second resected end. Then, Rad52/RAD52 facilitates the annealing between both strands, termed second-end capture, which further stabilises the D-loop (McIlwraith and West, 2008; Nimonkar et al., 2009; Shi et al., 2009; Sugiyama et al., 2006). Now, the 3’-OH group at the second DNA end can prime a new event of DNA synthesis using the displaced strand as a template that proceeds until the Figure 20. Second-end capture channels HR intermediates to DSBR. Rad52 stabilizes the D-loop by annealing the displaced strand to the second end of the DSB. Following second-end capture, DNA synthesis and ligation lead to the formation of a dHJ, the characteristic intermediate of DSBR. Tomás Lama Díaz 70 opposite end. The ligation of the two remaining nicks leads to the formation of the characteristic intermediate of DSBR, the dHJ (Figure 20) (Bzymek et al., 2010; Duckett et al., 1988; Holliday, 1964; Szostak et al., 1983). This four-way intermediate covalently links the two recombinant DNA molecules and must be processed prior chromosome segregation. Failure to disengage dHJs leads to severe chromosome abnormalities, aneuploidies and cell death, due to the persistence of physical connections between sister chromatids or homologous chromosomes (Wechsler et al., 2011). To prevent such deleterious effects, cells can eliminate dHJs by two different mechanisms: dissolution and resolution. 2.4.1 Dissolution of dHJs The sequence homology between the recombining molecules allows for the branch-migration of each single HJ. Although spontaneous branch-migration is limited, it can be stimulated by DNA helicases or translocases in ATP-hydrolysis-driven reactions (Tsaneva et al., 1992). Dissolution machinery takes advantage of dHJ mobility to disengage recombining molecules without reciprocal exchanges between them. In yeast, dissolution is mediated by the STR complex (Cejka et al., 2010c), while in humans it is carried out by its homologous counterparts BLM - Topoisomerase IIIa - RMI1-RMI2 (BTR) (Wu and Hickson, 2003; Xue et al., 2013). RecQ-family helicases Sgs1/BLM direct the convergent branch migration of the two HJs, with Top3/TopIIIa creating transient nicks to relax the positive supercoiling formed ahead of the migrating junctions (Chen et al., 2014). The two HJs advance towards each other until they fuse into a new type of recombination intermediate, a hemicatenane. Since the two DNA molecules are still physically linked, Top3/TOPOIIIalpha activity is required again to decatenate the junction and complete the repair (Figure 21). Absence of Top3 results in a pronounced slowgrowth phenotype caused by the accumulation of converged HJs; consistently, top3 mutants can be rescued by concomitant deletion of SGS1, allowing the endonucleolytic resolution of dHJs (Gangloff et al., 1994). Therefore, convergent branch migration and hemicatenane formation by Sgs1 impedes resolution and compromises repair by Top3 decatenation (Cejka et al., 2012). While Rmi1/RMI1 does not contribute significantly to the initial step of branch migration, it promotes Top3-dependent decatenation in vitro, although the precise role remains unknown (Bocquet et al., 2014). In line with this model, rmi1 deletion phenocopies top3∆ strains, suggesting the accumulation of hemicatenanes in these cells (Mullen et al., 2005). RMI2 is only present in higher eukaryotes, supporting the activity of BTR complex in more complex genomes by targeting BLM to different intermediates (Singh et al., 2008). Importantly, the singular mechanism of dHJ dissolution generates exclusively NCO products and, by promoting convergent migration of dHJs, impedes their resolution and hence suppresses CO formation and promotes genome stability (Ira et al., 2003; Wu and Hickson, 2003). This probably underpins why dissolution is the main DSBR sub-pathway to remove dHJs in somatic cells. Impairment of Sgs1/BLM activity in vivo results in higher CO levels and the concomitant increase of sister chromatid exchanges (Chaganti et al., 1974; Ira et al., 2003). 71 Figure 21. Schematic view of dHJ processing pathways during mitosis in S . cerevisiae . The STR complex disengages dHJs in a process termed dissolution to yield exclusively NCOs. Sgs1 promotes convergent migration of dHJs into a hemicatenane, which is further decatenated by Top3. Alternatively, dHJs can be processed by resolvases, producing a mixture of NCOs and COs, depending on the relative orientation of each pair of incisions. Only those resolved in the same orientation for both HJs yield a NCO configuration. Tomás Lama Díaz 72 2.4.2 Resolution of dHJs Occasionally, persistent dHJs that escape dissolution can be resolved by a specific type of SSEs, generally termed as resolvases (Blanco and Matos, 2015; Wyatt and West, 2014). Moreover, resolvases represent the only option for the elimination of single HJs, which may appear when one of the D-loop arms is cleaved prior to the second end capture, or upon fork converging during RDR or BIR (Mayle et al., 2015; Pardo et al., 2020). Initial studies with prokaryotic resolvases, such as RuvC in E.coli, set the paradigm for canonical resolution of HJs, and defined how the endonucleolytic activity of resolvases generates a mixture of COs and NCOs. RuvC is a homodimeric protein in solution that introduces two symmetrical and coordinated incisions at phosphodiester bonds in opposite HJ strands, producing two nicked duplexes that can be ligated without further processing (West, 1997). In the context of dHJs, if each pair of incisions occurs on the same plane for each single HJs, a NCO product is formed, whereas cleavage on orthogonal planes results in the reciprocal exchange of the flanking DNA duplexes that characterises CO outcomes (Figure 21)(West, 1997). In eukaryotes, the search for similar resolvases first revealed the existence of mitochondrial RuvC-like enzymes both in S. cerevisiae (Cce1) and S. pombe (Ydc2)(Lilley and White, 2001). Contrarily, nuclear resolvases with similar activities proved considerably more elusive, but work from several groups over almost two decades led to the identification of a series of SSEs with the ability to process HJs in vitro: Slx1-Slx4, Mus81-Mms4 and Yen1 in S.cerevisiae, as well as their orthologs in mammals (and also in other species) SLX1-SLX4, MUS81-EME1/EME2 and GEN1 (Figure 21)(Wyatt and West, 2014). As it is described below, the function of these SSEs is highly conserved from yeast to humans. Slx1 belongs to the GIY-YIG family of nucleases and represents the catalytic partner in the Slx1-Slx4 heterodimer (Figure 22a,b). Slx1-Slx4 displays a preference towards Y-structures, 5’-flaps and replication forks (Figure 22c). However, its role as a HJ resolvase has been questioned, given its low efficiency processing intact HJs. In this sense, Slx1-Slx4 introduces asymmetric nicks a few nucleotides away from the junction, producing a mixture of gapped and flapped products that cannot be directly ligated (Fekairi et al., 2009; Fricke and Brill, 2003; Guervilly and Gaillard, 2018; Munoz et al., 2009; Rass, 2013; Schwartz and Heyer, 2011; Svendsen and Harper, 2010; Wyatt et al., 2013). Its potential function in HJ resolution is further disputed by genetic studies demonstrating that the synthetic lethality of sgs1∆ slx1∆ or sgs1∆ slx4∆ double mutants cannot be suppressed by rad52∆, suggesting that the accumulation of unresolved recombination intermediates is not the primary cause of death in these mutants (Bastin-Shanower et al., 2003; Fabre et al., 2002; Fricke and Brill, 2003). Therefore, it has been postulated that, instead, the Slx1-Slx4 complex could operate in the processing of replication intermediates (Coulon et al., 2006, 2004; Deng et al., 2005). Mus81, the catalytic subunit of the Mus81-Mms4 complex, is a member of the XPF family of nucleases (Figure 23a,b) (Ciccia et al., 2008). Mus81-Mms4/MUS81-EME1 complexes, cleave efficiently a broad variety of structures, especially those containing a nick at the branched point, like replication forks, 3’-flaps and nicked HJs (Figure 23c) (Ehmsen and Heyer, 2009; Gaillard et al., 2003; Rass, 2013; Schwartz and Heyer, 2011). Contrarily, intact HJs constitute poor 73 substrates for these enzymes in vitro, being cleaved asymmetrically and thus yielding products that, akin to those of Slx1-Slx4, require further processing prior to ligation. (Ciccia et al., 2003; Ehmsen and Heyer, 2008; Gaillard et al., 2003; Gaskell et al., 2007; Kaliraman et al., 2001; Ogrunc and Sancar, 2003; Osman et al., 2003; Schwartz et al., 2012). Given that both Mus81Mms4 /MUS81-EME1/EME2 and Slx1-Slx4/SLX1-SLX4 complexes deviated from the RuvC paradigm of HJ resolution, they were dubbed as non-canonical HJ resolvases. Nonetheless, Mus81/MUS81 role in HJ resolution is supported by several in vivo studies. In this sense, mus81∆ and mms4∆ mutants, like slx1∆ and slx4, are essential to sustain the viability of cells lacking Sgs1 (Hickson and Mankouri, 2011). However, the lethality of mus81∆/mms4∆ sgs1∆ can be suppressed in a rad52∆ background, proving that, in those mutants, cell death ensues from the accumulation of toxic recombination intermediates, such as dHJs. Phenotypes of single deletions of mus81∆ and mms4∆ are also consistent with its role as HJ resolvase, including a reduction in CO formation and spore viability (Boddy et al., 2001; de los Santos et Figure 22. Domain structures of Slx1-Slx4/SLX1-SLX4 and in vitro DNA substrate specificities. (a,b) Slx1 is a small GIY-YIG family member with a zinc-finger motif at the C-terminus and, together with Slx4, forms a heterodimeric SSE. (c) Slx1-Slx4 cleaves multiple branched DNA substrates (blue arrowheads), including splayed arms, RF and 5’flaps. HJs are cleaved asymmetrically and inefficiently by Slx1-Slx4 complex in budding yeast, whereas in human cells the interaction between MUS81-EME1 and SLX4 promotes efficient HJ processing. Tomás Lama Díaz 80 consecutive waves of HJ resolution by Mus81-Mms4 and Yen1 from the G2/M transition to the end of the cell cycle (Figure 26). The high degree of functional overlapping between these three pathways may be a consequence of the inability of cells to trigger checkpoint activation and arrest cell-cycle progression once recombination intermediates mature into dHJs (Matos et al., 2011). Therefore, mitotic cells promote STR dissolution into NCOs to maintain genome stability for most of S and G2, whereas at the end of the cell cycle, Mus81-Mms4 and Yen1 late activation safeguards chromosome segregation at the expense of CO production and, therefore, the risk of increasing loss of heterozygosity and genome rearrangements. By backing up Mus81-Mms4, Yen1 might act as a checkpoint substitute under high levels of stress, ensuring the thorough elimination of abnormal levels of single or double HJs. Figure 26. Schematic representation of the sequential activation of HJ processing pathways biases DSBR output towards NCO. Early phospho-stimulation of Sgs1 by Cdk1 promotes dissolution of most dHJs into NCOs. Later on, Cdc5-dependent activation of Mus81-Mms4 at G2/M and dephosphorylation of Yen1 by Cdc14 during anaphase trigger two consecutive waves of HJ resolution to remove persistent JMs, at the expense of producing a mixture of NCOs and COs. Note: All enzymes have different levels of basal activity which cannot be compared in the same graph. P, prophase/prometaphase; M, metaphase; A, anaphase; T, telophase. (Adapted from Grigaitis et al,.2020) 81 2.5 Modulation of SSE activity outside the NCO/CO balance Nuclear eukaryotic resolvases are specialized flap endonucleases that recognize specific DNA secondary structures with 3’ or 5’ polarity, regardless of their specific sequences (Dehé and Gaillard, 2017). Resolvases -like most SSEscan process multiple types of branched DNA molecules matching their endonuclease family preferred polarity. The versatility of Slx1Slx4/SLX1-SLX4, Mus81-Mms4/MUS81-EME1/2 and Yen1/GEN1 debranching activities on threeand four-way DNA junctions allows them to fulfil roles beyond HJ resolution. For instance, during replication stress both the timely processing of recombination intermediates and fork cleavage to promote RDR can contribute to cell viability (Falquet and Rass, 2019). In this sense, two main roles have been proposed for Mus81 during replication stress. First, limiting BIR mutagenesis either by cleaving the migrating D-loop to reconstitute a functional fork, or by resolving the HJ formed after the D-loop is merged with a converging replication fork (Mayle et al., 2015). Second, Mus81 could also support replication at stalled replisomes, which become stabilized by HR factors that promote D-loop formation until the arrival of a converging fork. Mus81 would then resolve the resulting HJ, in a similar scenario to the one already described for BIR (Pardo et al., 2020). In S.pombe, Mus81-dependent chromosomal breaks correlate with replication restart (Hanada et al., 2007), suggesting that Mus81 promotes BIR by targeting stalled RFs and converting them into one-ended DSBs. In mammalian cells under different conditions of replication stress, MUS81 activity increases cell viability by enabling replication restart at CFS (Fugger et al., 2013; Lai et al., 2017; Murfuni et al., 2013; Naim et al., 2013; Pepe and West, 2014a; Regairaz et al., 2011; Shimura et al., 2008, 2013; Ying et al., 2013). Moreover, it has been hypothesized that MUS81-EME1/EME2 complex ability to process D-loops in vitro is consistent with a role analogous to Mus81-Mms4 in limiting BIR mutagenicity, although more data is required to prove this model (Pepe and West, 2014b). Yen1, given its later activation during the cell cycle, has been implicated in backing up Mus81 activities during replication stress, processing those HJs that could appear after D-loops and converging forks merge, with the additive effect of YEN1 deletion on mus81∆ phenotypes supporting this model (Mayle et al., 2015). Nonetheless, recent evidence indicates that Yen1 role is not limited to process HJs that escape Mus81 during DSBR or replication stress. In human cells, DNA2 promotes DNA replication and participates in the restarting of stalled replication forks (Thangavel et al., 2015); similarly, under-replicated DNA accumulates in Dna2 helicase-defective strains. In these mutants, further deletion of YEN1, but not MUS81, results in a strong synthetic sickness that is exacerbated by exposure to genotoxic agents and aggravated by RAD52 deletion (Olmezer et al., 2016). These results support a role for Yen1, which is not redundant with Mus81 functions, in the mitotic resolution of persistent replication intermediates by processing those DNA structures that either escape or arise in the absence of the helicase activity of Dna2 and which remain to be determined. In addition, the modelling of human CFS expression in budding yeast demonstrates that Yen1 also plays a protective role in the stability of these difficult-to-replicate regions (Kaushal et al., 2019). Tomás Lama Díaz 82 In view of this current picture where resolvases are paramount enforcers of replication completion and JM elimination, it is still quite striking that cells restrict their activities until the last stages of the cell cycle, even though under-replicated regions and persistent HJs constitute an enormous threat for chromosome segregation and viability. As described in section 2.4.3, it is well-established that the spatio-temporal regulation of resolvases is essential to bias HR outcome towards NCO pathways but, additionally, that timely activation of Mus81Mms4/MUS81-EME1 and Yen1/GEN1 safeguards the genome from the untimely processing of physiologically relevant secondary DNA structures (Wild and Matos, 2016). Following this rationale, the delayed activation of resolvases would minimise the exposure to these enzymes of the multiple branched molecules stemming from the normal DNA metabolism during S/G2, including replication (replication forks, reversed fork, 5’-flaps, HJs, etc), transcription (Rloops), repair (D-loops or HJs), protection of chromosome ends (T-loops), etc. Moreover, the number of possible substrates for SSEs cleavage is expanded by the potential of DNA to adopt different secondary structures that deviate from the canonical B conformation (Figure 27), like hairpins, cruciforms, G-quadruplexes and triplex DNA (Bochman et al., 2012). How DNA adopts these secondary structures depends on multiple factors, such as local base composition -symmetry, repetitiveness and GC contentand topology (Wang and Vasquez, 2017; Wickramasinghe et al., 2015), frequently arising in regions containing inverted, mirror and/or direct repeats and those with long tracts of guanines. Although these non-b DNA conformations can be readily induced in vitro on ssDNA, their in vivo relevance has been long under debate, since the duplex DNA conformation is energetically more favourable in cells. In fact, these structures are not continously present in the genome, but transiently appearing at certain genomic regions during DNA replication, repair, and transcription (Khristich and Mirkin, 2020; Krasilnikova and Mirkin, 2004), where the melting of dsDNA by helicases and the topological forces induced during such processes facilitate B-DNA to non-B DNA local transitions (Lerner and Sale, 2019). The conservation of these branched DNA molecules and secondary structures throughout eukaryotic genomes highlights their importance in multiple aspects of DNA dynamics. T-loops protect telomeres by avoiding chromosome ends from being recognized as damaged DNA, while reversed forks stabilize replisomes during replication stress and mediate fork restart (de Lange, 2004; León-Ortiz et al., 2014; Neelsen and Lopes, 2015). G-quadruplexes have been linked to telomere maintenance, replication origin specification in mammalian cells, transcriptional regulation, and immunoglobulin class switch recombination within the IgG locus (Bochman et al., 2012; Robinson et al., 2021). However, these very structures can also trigger genomic instability (Kaushal et al., 2019; van Wietmarschen et al., 2020; Zhang and Freudenreich, 2007; Zhao et al., 2018), especially due to their intrinsic capacity to hinder fork progression (Follonier et al., 2013; Gerhardt et al., 2016; Liu et al., 2012). In fact, local sequences predicted to form non-B DNA structures correlate with translocation breakpoints, DSBs, indels, copy number variations and point mutagenesis (Wang and Vasquez, 2022). In general terms, it appears that cells prioritise the helicase-mediated unwinding of replication and recombination intermediates, as well as secondary structures, to maintain fork progression 83 Figure 27. Schematic representation of secondary DNA structures that may serve as substrates for SSEs. (a) Replication and recombination give rise to branched intermediates that could be potentially cleaved by SSEs (b) Replication increases the probability of repetitive and GC rich sequences to fold into different secondary structures, such as hairpins, cruciforms, G-quadruplexes, and H-DNA. Helicases (black) and SSEs (red) that can target each structure are indicated. Black circles represent replication origins. Tomás Lama Díaz 84 and genome integrity, with the DDC extending the temporal window to remove those obstacles under situations of replication stress. However, if the intermediates or the secondary structures persist, SSEs play a critical role acting as a backup of helicase activity, albeit their nucleolytic activity is frequently associated with genomic rearrangement, insertions, deletions and telomere instability (León-Ortiz et al., 2014). The Srs2 and WRN helicases unwind hairpins in yeast and mammalian genomes, respectively. In WRN-deficient cells, MUS81-EME1-SLX4 cooperate to cleave the excess of harpins accumulated during DNA replication (Mengoli et al., 2023; van Wietmarschen et al., 2020). Similarly, Srs2 and RTEL unwind imperfect hairpins formed at trinucleotide repeats (Bhattacharyya and Lahue, 2005, 2004; Frizzell et al., 2014), preventing their cleavage by the MMR complex MutSb (Owen et al., 2005; van den Broek, 2002). Numerous helicases have been implicated in dismantling G-quadruplexes in vitro and in vivo, including the Pif1-family helicases (Paeschke et al., 2011; Sabouri et al., 2012; Sanders, 2010) and FANCJ (London et al., 2008; Sarkies et al., 2012; Schwab et al., 2013; Wu and Brosh, 2009). Absence of ScPif1 and Rrm3 in S.cerevisiae, or FANCJ in metazoans is associated with G4-dependent rearrangements and deletions of guanine-rich sequences (Cheung et al., 2002; Kruisselbrink et al., 2008; Lopes et al., 2011; Paeschke et al., 2013; Piazza et al., 2010; Ribeyre et al., 2009; Youds et al., 2006). Some of the RecQ-family helicases (Sgs1 in budding yeast; BLM and WRN in mammals) and RTEL are necessary for efficient replication of the G-rich telomeric ends, supporting their role in dismantling G4s at chromosome ends (Crabbe et al., 2004; Kamath-Loeb et al., 2001; Sfeir et al., 2009; Vannier et al., 2013, 2012). The action of the TLS polymerase REV1 provides an alternative solution to helicase unwinding to allow G4 bypass in vitro and in cultured vertebrate cells (Eddy et al., 2014; Sarkies et al., 2010; Schiavone et al., 2014). In contrast, rev1∆ mutants in S. cerevisiae do not exhibit increased levels of genome instability at G4 motifs (Lopes et al., 2011). It is tempting to speculate that nucleases could also backup helicase activity to resolve persistent G4s, as supported by the in vitro cleavage of G4 substrates by Mre11/MRE11, Exo1/EXO1 and Dna2/DNA2 (Lin et al., 2013; Masuda-Sasa et al., 2008). Moreover, the conversion of T-loops into dsDNA required for telomere replication during S phase is mediated by RTEL, which restores the lineal configuration of chromosome ends without affecting telomere length (Brenner and Nandakumar, 2022). In contrast, SLX4 recruitment to the telomere mediates nucleolytic processing of T-loops by SLX1, resulting in telomere shortening and the appearance of Tcircles in RTEL-deficient cells (Vannier et al., 2012; Wan et al., 2013). It could be argued that the downregulation of SSEs serves a general purpose of promoting helicase-dependent dismantling of branched DNA molecules and secondary structures as a first resource. In this sense, there are several examples in the literature showing how posttranslational modifications of SSEs or their interaction with specific proteins can modulate their functionality, from the direct modification of their catalytic properties to their targeting to precise locations or substrates (Dehé and Gaillard, 2017). For instance, the activity of the NER nucleases Rad2/XPG and Rad1-Rad10/XPF-ERCC1 must be tightly controlled, since each nuclease can potentially cleave damaged and undamaged 85 strands on opposite sides of the distorted DNA helix, resulting in a DSB. However, in human cells, proper positioning and recruitment of both nucleases is ensured by the scaffolding effect of XPA, RPA and TFIIH, with RPA also directing the incision of XPF-ERCC1 to the damaged strand (Laat et al., 1998; Staresincic et al., 2009). XPF-ERCC1 is also important in vertebrate cells for the removal of interstrand crosslinks. In this case, the scaffold SLX4 recruits XPFERCC1 to the lesions and imparts structural preferences towards replication-like structures over NER common substrates, such as stem-loops or bubbles (Douwel et al., 2017, 2014; Hodskinson et al., 2014). Another key SSE, Rad27/FEN1, interacts with PCNA to mediate its recruitment to replication sites during Okazaki fragment processing, as well as to stimulate its flap-endonuclease activity. Its timely interaction with PCNA relies on the concerted actions of two different PTMs. FEN1 methylation facilitates PCNA interaction by inhibiting phosphorylation at a nearby residue throughout S-phase. In late S phase and G2, CDK1-dependent phosphorylation both inhibits FEN1-PCNA interaction and reduces FEN1 catalytic activity. Phosphorylation also drives FEN1 SUMOylation which, in turn, allows FEN1 ubiquitination and targeting for proteasomal degradation (Guo et al., 2012, 2010). Whereas PTMs control the opportune activation of FEN1 during Okazaki fragment processing and its protein levels, additional mechanisms regulate FEN1 in the response to replication stress. In addition to its flap-endonuclease activity, FEN1 is also endowed with a gap-endonuclease activity necessary for the re-initiation of stalled replication forks. The balance between both activities is regulated by two different interactors: WRN, which promotes the gap endonuclease, and WDR4, which stimulates the flapendonuclease activity at unstressed forks (Cheng et al., 2016; Sharma et al., 2004). In order to understand the relevance of SSE regulation or, conversely, the consequences of their misregulation, two strategies are frequently employed: (1) the development of alleles to block or mimic the specific PTMs of the SSE of interest; (2) the overexpression of the SSE itself or the up/downregulation of its main regulators. For instance, this overexpression tactic has been employed to show that increased levels of various members of the Rad1and Rad2-families of nucleases in budding yeast lead to genome instability, checkpoint activation, impaired DNA replication, hypersensitivity to genotoxic agents or even cell death (Becker et al., 2018; Jimeno et al., 2017; Kang et al., 2010). Similarly, overexpression of a mutant version of FEN1 that cannot be degraded upon S/G2 transition causes cell-cycle delay and abnormalities during chromosome segregation (Guo et al., 2012). Similar approaches have also been applied to Mus81-Mms4/MUS81-EME1 and Yen1/GEN1, in order to understand how their stringent cellcycle control minimizes the risks associated to their unrestricted actions. Regarding Mus81-Mms4, which becomes activated upon phosphorylation of Mms4 by Cdk1 and Cdc5, the two main strategies that have been employed to anticipate the action of this complex are i) the overexpression of Cdc5 by placing it under the transcriptional control of the PGAL1 promoter (Matos et al., 2013) and ii) the generation of phosphomimetic mutants at CDK consensus sites present on Mms4, like mms4-S56E and mms4-S56E-S184D, which complement the sensitivity of mms4∆ strains without affecting their sensitivity to genotoxic agents (Szakal and Branzei, 2013). Both conditions led to a significant increase in mitotic COs, which could Tomás Lama Díaz 86 result in higher levels of LOH in diploid cells (Matos et al., 2013; Szakal and Branzei, 2013). It has been proposed that these higher CO levels were a direct consequence of increased interference by Mus81-Mms4 with at least two different repair pathways: dissolution of dHJs by the STR complex during HR, and the error-free DNA damage tolerance routes (Szakal and Branzei, 2013). However, it is worth mentioning that while Cdc5 overexpression led to the biochemical hyperactivation of immunoprecipitated Mus81-Mms4 in these studies (Matos et al., 2013), there is no direct evidence of increased biochemical activity of the Mus81-Mms4 phosphomimetics with respect to the wild-type enzyme. Therefore, the higher CO proportion observed in both type of strains might be a consequence of different types of perturbation to Mus81 regulation. As aforementioned, the ability of MUS81-EME1/2 to resolve dHJs in human cells is enhanced by the nickase activity of SLX1-SLX4 (Castor et al., 2013; Fekairi et al., 2009; Gritenaite et al., 2014; Matos et al., 2011; Munoz et al., 2009; Svendsen et al., 2009; Wyatt et al., 2013) and its targeting to replication intermediates also requires its interaction with SLX4 (Guervilly et al., 2015; Minocherhomji et al., 2015, Duda et al., 2016). In this sense several lines of evidence indicate that the genome instability caused by misregulation of MUS81-EME1/2 is due to its precocious interaction with SLX4. First, it has been shown that DNA damage induced by aberrant CDK1 levels or a defective DDC is MUS81-dependent (Beck et al., 2012; Dominguez-Kelly et al., 2011; Forment et al., 2011; Kim et al., 2013; Neelsen et al., 2013; Techer et al., 2016). Downregulation or inhibition of the WEE1 kinase, which repress CDK1, causes the breakage of replicating chromosomes, increasing DNA damage levels, checkpoint activation and S-phase cell cycle arrest (Beck et al., 2010, 2012; Dominguez-Kelly et al., 2011). This phenotype was also recapitulated by the transient overexpression of cyclin E and CDC25A phosphatase, which antagonizes WEE1 effect, except that in this case with arrest in the G2/M transition (Neelsen et al., 2013). While it was originally postulated that these results were due to a direct inhibitory effect of the WEE1 kinase on MUS81-EME1/2, it was later demonstrated that WEE1 inhibition triggers MUS81 activation indirectly. Anomalously high activity of CDK drives premature phosphorylation of SLX4 in S phase and the subsequent assembly of the hyperactive MUS81-EME1/SLX1-SLX4 complex, resulting in the cleavage of ongoing replication forks and chromosome pulverization (Duda et al., 2016). Similar phenotypes were described under conditions of checkpoint inactivation, since the CHK1 kinase directly inhibits CDC25 to avoid dephosphorylation of CDK while high levels of DNA damage persist (Beck et al., 2010, Forment et al., 2011, Syljuasen et al., 2005). In mouse embryonic cells, similar results have been obtained through the expression of a phosphomutant allele of CDK1 that cannot be inhibited by WEE1 and, indeed, the viability of the embryos harbouring this allele is severely compromised (Szmyd et al., 2019). Additionally, it has been shown that the S87D mutation in MUS81, which mimics a CK2-dependent phosphorylation event in G2/M, licenses untimely formation of the MUS81EME1/SLX1-SLX4 complex in vivo, promoting DSB formation in S phase and a mild chromosome pulverization phenotype (Palma et al., 2018). 87 For Yen1 and GEN1, the logic behind their cell cycle-dependent regulation implies active mechanisms to inhibit their ability to access their substrates until its last stages. This has provided an exciting opportunity to understand the consequences of the premature activation of resolvases in cleaner set-ups, as relatively simple mutations in these proteins uncouple their activation from cell cycle progression without altering any of the master regulators of the cell cycle (like CDK or checkpoint kinases). In this sense, a constitutively nuclear version of GEN1 (GEN1nuc) was generated by mutation of its nuclear export signal and additions of viral NLS at its C-terminus, overriding its main regulatory layer (Chan and West, 2014). Expression of GEN1nuc in mammalian cells produces an increase in sister-chromatid exchanges in undamaged and cisplatin-treated cells, but not hypersensitivity to genotoxic agents like hydroxyurea, camptothecin or cisplatin. Moreover, GEN1nuc is able to alleviate phenotypes associated to inactivation of MUS81 and BLM, like chromosomal defects and decreased viability (Chan and West, 2014). In budding yeast, Yen1 activity and localization is directly controlled by Cdk1-dependent phosphorylation. Mutation of all nine serines in CDK consensus sites to alanines resulted in a version of Yen1, Yen1ON, that is constitutively active and nuclear at all stages and, therefore, fully refractory to cell cycle control (Blanco et al., 2014). This mutant has already been widely employed to demonstrate the detrimental effects of the aberrant accumulation of recombination and replication intermediates by relieving the phenotypes of mutant strains with reduction or loss of function in genes like MUS81, SGS1, SRS2, MLH1/MLH3 or DNA2, in both mitotic and meiotic contexts (Alonso-Ramos et al., 2021; Arter et al., 2018; Bittmann et al., 2020; Elango et al., 2017; Grigaitis et al., 2020; Michel et al., 2017; Olmezer et al., 2016). However, its utility as a tool to understand the requirement for the stringent control of resolvase activation has been considerably less exploited. In mitotic cells, Yen1ON expression increases sensitivity to MMS, CO formation and total loss-of-heterozigosity (Blanco et al., 2014) while in meiosis it produces not only higher CO/NCO ratios, but also the loss of CO interference (Arter et al., 2018). Taking into account the plethora of different recombination/replication intermediates and other non-B DNA secondary structures that can potentially be hydrolysed by Yen1, it remains of great interest to determine which of them could be specifically compromised by its unrestricted activity in S/G2. Therefore, the concept driving the present PhD research project is to uncover the specific pathways or structures disrupted by Yen1 misregulation. To achieve this, we will combine previously described strategies, such as altering the expression levels of both wild-type and regulation-deficient YEN1 alleles with mutations in several candidate DNA helicases involved in genome integrity maintenance. Our focus will be on the Pif1 family of DNA helicases, given their crucial functions in DNA secondary structure metabolism in budding yeast. 3. PIF1 FAMILY OF HELICASES The Pif1 family, found in some bacteria, archaea, and virtually in all eukaryotes, belongs to the superfamily IB of 5’-3’ directed, ATP-dependent, P-loop helicases (Berger, 2008; Bochman et al., 2011). Saccharomyces cerevisiae PIF1 (ScPIF1), the founding member of the Pif1 family, was identified in a genetic screen searching for nuclear mutations affecting the recombination Tomás Lama Díaz 88 frequency of the mtDNA; three allelic mutants were discovered and assigned to the PIF1 locus (petite integration frequency) (Foury and Kolodynski, 1983). Further research showed that PIF1 encodes an enzyme with ssDNA-dependent ATPase and 5’-3’ helicase activities, and confirmed its predominant role in sustaining the integrity of the mtDNA (Foury and Lahaye, 1987; A. Lahaye et al., 1991; Lahaye et al., 1993a). Shortly after, PIF1 mutations were isolated in another screening designed to identify mutants influencing the loss of expression of subtelomeric genes (Schulz and Zakian, 1994a), initiating a long-lasting research line on the functions of Pif1 helicases in the maintainance of nuclear genome stability. S. cerevisiae harbours a second PIF1-like gene, named ribosomal DNA recombination mutation (ScRRM3), after its discovery in an independent screening for factors suppressing recombination at loci with tandem repeats, such as the rDNA and CUP1 genes. It was later identified as a member of Pif1 helicase family based on sequence similarity with ScPif1 (Ivessa et al., 2000; Keil and McWilliams, 1993). Whereas other fungal species, such as Candida albicans or Cryptococcus neoformans also contain two Pif1 family helicases, the distantly related fungi Schizosacchromyces pombe and metazoans contain only one. Importantly, the presence of two Pif1 helicases in S. cerevisiae is not related to the ancient whole-genome duplication (WGD) that occurred in the ancestor of the yeast WGD clade within the Saccharomycetaceae family (Dujon et al., 2004; Kellis et al., 2004). This suggests that one of the two Pif1 helicases was lost in a common ancestor of metazoans and in some fungal taxons. Phylogenetic analyses support that ScRrm3 may have evolved first (Bochman et al., 2010). 3.1 Structural motifs and protein isoforms of Pif1-family helicases Pif1-like helicases conserve the seven SFI family characteristic motifs (I, Ia, II, III, IV, V, and VI) and present three E. coli RecD motifs (A, B and C). In addition, they present a distinctive 21-aminoacid sequence, the Pif1 signature motif, located between regions II and III, which differentiates Pif1-family members from other helicases of the IB superfamily (Figure 28). This Pif1 signature motif is required for optimal ATPase and helicase activities in vitro and, consequently, for all in vivo functions requiring the catalytic activity of these proteins (Bochman et al., 2010; Geronimo et al., 2018; Mohammad et al., 2018; Zhou et al., 2016). As the founding member of the family, ScPif1 has been deeply characterized both in vitro and in vivo. In contrast, ScRrm3 research has been limited to in vivo approaches for decades, largely due to the difficulties to purify the full-length protein. ScRrm3 is expressed at low levels and forms insoluble aggregates when overexpressed in E. coli, yeast and Sf9 insect cells (Ivessa et al., 2002). Besides studies in budding yeast, SpPfh1 from S. pombe and PIF1 from Mus musculus and Homo sapiens (HsPIF1) have also been widely studied to interrogate the properties and functions of Pif1 helicases. ScPif1, ScRrm3, SpPfh1 and HsPIF1 present a central helicase domain flanked by N-terminal and C-terminal domains of variable length, hereinafter NTD and CTD, respectively (Andis et al., 2018; Bessler et al., 2001; Byrd and Raney, 2017) (Figure 28). The highly conserved helicase domain is required for both the ATP binding and hydrolysis as well as the ssDNA binding and translocation activities required to unwind duplex DNA, unfold G4s or displace non-nucleosomal protein complexes bound to DNA. The NTD and CTD domains are 89 intrinsically disordered and have diverged to fulfil specific roles in different Pif1 helicases by expanding their biochemical properties (Andis et al., 2018; Gu et al., 2008), mediating diverse protein-protein interactions (Syed et al., 2016) or reacting to specific post-translational modifications (Makovets and Blackburn, 2009a; Rossi et al., 2015). In this sense, the NTDs of ScPif1 and SpPfh1 also contain mitochondrial targeting signals (MTS), whereas in HsPIF1 the NLS and MTS localise at the NTD and CTD, respectively (Figure 28) (Futami et al., 2007). The NTD of ScPif1 is responsible for its preferential unwinding of RNA-DNA hybrids as well as for its specific interaction with the mitochondrial SSB Rim1. Acetylation of ScPif1 NTD and CTD enhances its helicase activity (Onyekachi E Ononye et al., 2020), while phosphorylation of both ScPif1 and ScRrm3 NTDs inhibits fork reversal during replication stress (Rossi et al., 2015). ScRrm3 NTD is indispensable for ScRrm3 functions in vivo and, like SpPfh1 NTD, contains a PIP box to interact with PCNA, as well as the residues that mediate ScRrm3 interaction with Orc5 (Schmidt et al., 2002; Syed et al., 2016). ScPif1 contains two PIP boxes that control its interaction with PCNA in the contexts of replication and DSB repair (Buzovetsky et al., 2017; Dahan et al., 2018), and its Mec1-dependent phosphorylation is necessary to recruit ScPif1 to DSBs (Makovets and Blackburn, 2009a). Similarly, the human HsPIF1 CTD is important to mediate physical interactions its post-translationally modified, while its NTD is required for strand annealing activity in vitro (Gu et al., 2008). All ScPIF1, SpPFH1 and HsPIF1 encode at least two different protein isoforms, allowing the dual-targeted localization required to fulfil their multiple roles in the nucleus and the mitochondria. In budding and fission yeast, the mitochondrial and nuclear isoforms are produced by Alternative Translation Initiation (ATI) at two in-frame AUG codons: AUG1 - AUG40 in S. cerevisiae and AUG1 - AUG21 in S. pombe. Importantly, in both cases, these AUG codons flank the sequence for an MTS (Sabouri et al., 2012; Schulz and Zakian, 1994a). If translation starts from AUG1, the MTS drives the mitochondrial import of the full-length isoform; instead, translation initiation from the second AUG produces shorter isoforms lacking the MTS. These are thus targeted to the nucleus, presumably through yet-uncharacterized NLSs. In contrast, HsPIF1 mitochondrial and nuclear isoforms are a consequence of mRNA alternative splicing, although the same rationale prevails: the presence of an MTS outcompetes the NLS, while the shorter isoform lacking the MTS is imported into the nucleus (Futami et al., 2007). Moreover, upon DNA damage, the nuclear isoform is replaced by a third HsPIF1 isoform, also generated by alternative splicing, that improves cell survival. Both the cytoplasmic location of Tomás Lama Díaz 96 As described in the 2.3 section, ScPif1, together with Pol32, is essential to promote long-track DNA synthesis during BIR. ScPif1 is recruited to the migrating D-loop through its interaction with PCNA, mediated by a non-canonical PIP domain located at ScPif1 CTD and mutation of this domain (Buzovetsky et al., 2017) or deletion of PIF1 in yeast, D. melanogaster or mammalian cells greatly impairs BIR completion (Kocak et al., 2019; Li et al., 2021; Wilson et al., 2013). It is considered that ScPif1 may promote BIR synthesis in two different ways: i) acting as the replicative helicase during D-loop-associated DNA synthesis and ii) releasing the newly synthesized strand to prevent topological stress and promote D-loop migration. Besides BIR, both mechanisms could also promote DNA synthesis during gene conversion. Indeed, ScPif1 activity is inhibited during meiosis to limit long-tract gene conversion, with its ability to support DNA synthesis in this context also relying on the PCNA interaction through its noncanonical PIP box (Vernekar et al., 2021). In contrast, rrm3∆ mutants do not exhibit any deficiency in repair by BIR (Muñoz-Galván et al., 2017; Wilson et al., 2013). Instead, ScRrm3 contributes to the repair of replication-dependent DSBs, consistent with its ability to move with the replisome (Azvolinsky et al., 2006; Schmidt et al., 2002). The repair of this subset of DSBs requires sister-chromatid exchange, but how ScRrm3 contributes to this process is currently unknown (Muñoz-Galván et al., 2017). 3.7 Pif1-helicases in mtDNA maintenance Pif1-family helicases are essential to maintain the integrity of mtDNA in most eukaryotic cells. In S.cerevisiae, mtDNA of pif1∆ strains is highly unstable due to fragmentation and mutagenesis, leading to the loss of a functional mitochondrial genome, depicted by the rhoor rho0 (commonly known as cytoplasmic petite phenotype) exhibited by these cells (Cheng et al., 2007; Foury and Van Dyck, 1985; O’Rourke et al., 2005). Like pif1∆ cells, strains harbouring the pif1-m1 allele have been reported to quickly acquire this petite phenotype and are more sensitive to genotoxic agents, such as EtBr, than wild-type cells (Cheng et al., 2007; O’Rourke et al., 2005; Schulz and Zakian, 1994a). PIF1-/- knockout mice develop mitochondrial myopathy with respiratory chain deficiency, accompanied by mutations of the mtDNA, supporting a conserved role for these enzymes in mtDNA maintenance in mammalian cells (Bannwarth et al., 2016). While the precise mechanisms by which these helicases operate in mitochondria is still unknown, several phenotypes support the notion that, like their nuclear isoforms, they could be involved in multiple steps of mitochondrial DNA replication and repair. ScPif1, together with the Ntg1 glycosylase/AP-lyase, protects mtDNA from oxidative damage (Doudican et al., 2005; O’Rourke et al., 2002). Similarly, murine PIF1 is also important to safeguard mtDNA from the action of ROS (Bannwarth et al., 2016). Moreover, ScPif1 has been also implicated in promoting mtDNA recombination and replication (Foury and Van Dyck, 1985) and scPif1 and SpPfh1 catalytic activity is required for efficient mtDNA replication (Cheng et al., 2009; Pinter et al., 2008). This suggests that akin to the nuclear roles of Pif1 helicases, both scPif1 and SpPfh1 may support mtDNA replication by displacing protein-DNA complexes or unwinding G4 structures. rrm3∆ mutants do not manifest visible mitochondrial defects, but the presence of a putative MTS in its sequence, and its identification -while not individually verifiedin a screen for 97 mitochondria-localized proteins were compatible with a role of ScRrm3 in the mitochondria. Indeed, RRM3 deletion partially suppresses the petite phenotype of pif1∆ cells. However, this suppression may be an indirect consequence of altered dNTP pools in these mutants, since the absence of ScRrm3 activates the DDC response via Rad53 phosphorylation, which, in turn, increases the activity of the ribonucleotide reductase. This leads to an increment of dNTPs concentration, which is known to promote replication under stress conditions; consistently, higher Rnr1 levels partially alleviate respiratory deficiency and mtDNA copy number in pif1∆ mutants to the same extent as RRM3 deletion (O’Rourke et al., 2005; Taylor et al., 2005). The lack of a direct role for ScRrm3 in mitochondria is also supported by experiments in fission yeast, where the heterologous expression of ScRrm3 can only compensate for the nuclear defects of pfh1 mutants (Pinter et al., 2008). Tomás Lama Díaz 98 CHAPTER 1: ANALYSIS OF YEN1 PREMATURE ACTIVATION The discovery of the late activation of Yen1 and its orthologs has prompted several lines of research to understand the benefits of its regulation to cell cycle progression. In addition to cellcycle dependent phosphorylation, Yen1 SUMOylation at K714 also contributes to restrain its activity by triggering SUMOylation-dependent ubiquitination and proteasomal degradation of chromatin-bound Yen1 during G1 (Figure 29a), adding another regulatory layer to minimize the presence of active Yen1 in the nucleus when S phase begins (Talhaoui et al., 2018)). Here, we will employ YEN1 alleles deficient for one or both types of regulatory PTMs to explore the potential detrimental outcomes of premature resolvase activity during S and G2 phases. As described in the introduction, abrogation of Cdk1-dependent phosphorylation by nine Ser to Ala mutations in CDK consensus sites (Figure 29b, Yen1ON) renders the protein nuclear and active through the cell cycle (Figure 29c), while a K714R mutation (Figure 29b, Yen1KR) prevents its proteasomal degradation in G1 when bound to chromatin. The use of the Yen1ON phosphomutant presents several advantages to establish genetic set-ups with well-defined variables in the context of premature HJ resolution activities if compared to systems based on Mus81-Mms4 or MUS81-EME1/EME2 deregulation. First, premature hyperactivation of Mus81/MUS81 is achieved either through unrestrained CDK1 activation during S phase (Beck et al., 2012; Duda et al., 2016; Forment et al., 2011; Neelsen and Lopes, 2015) or with phosphomimetic alleles (Palma et al., 2018; Szakal and Branzei, 2013). However, both experimental approaches present limitations that complicate the interpretation of genetic results. On the one hand, extensive CDK1 activity during replication can affect multiple cellular processes, given its role as a cell-cycle master regulator. On the other hand, serine and threonine phosphorylation is usually mimicked by their substitution with negatively charged amino acids, such as aspartate and glutamate, respectively. It has been proven that these acidic amino acids are rather poor pSer/pThr mimetics, since their charge (-1 vs -2 for pSer/pThr at physiological pH), shape and hydrogen-bond geometry differ considerably (Cerulli and Kritzer, 2020; Chen and Cole, 2015). Consequently, Asp/Glu mutations frequently fail to fully recapitulate the functional effects of phosphorylation events. Second, despite recent reports hinting at potential SUMO-mediated interactions of Yen1 with other proteins (Dorison et al., 2022), no stable interaction partners that may influence its activity have been detected for Yen1 (Wild et al., 2019), unlike for Mus81-Mms4 and MUS81-EME1/2. Therefore, a Yen1 phosphomutant refractory to cell cycle-dependent downregulation may more realistically imitate a fully active version of Yen1 in vivo during S and G2. We will focus on Yen1ON and, similarly, Yen1KR to investigate the protective role of Yen1 cell cycle regulation by applying the approaches described in the introduction, including alteration of their expression levels and/or interaction with DNA helicases to specifically increase the cellular load of secondary structures and recombination or replication intermediates. 99 Figure 29. Mutant variants of Yen1 to study the regulation of HJ resolution. (a) Schematic representation of Yen1 domains, depicting those residues modified in various Yen1 mutants: blue, serines in CDK consensus sites; green, residues required for catalytic activity; purple, residue required for SUMOylation-dependent proteasomal degradation. (b) Description of the Yen1 mutants employed in this work. (c) Illustrative summary of Yen1 cellcycle regulation by CDK-dependent phosphorylation (left panel), as opposed to the regulation-insensitive Yen1ON mutant (right panel), employed in this work as a tool to study the consequences of Yen1 premature activation. Tomás Lama Díaz 100 1. OBJECTIVES Aim 1. To analyse the consequences of the overexpression of YEN1ON on viability, cell-cycle progression and DDC activation. Aim 2. To search for genetic interactions between YEN1ON expressed at endogenous levels and helicases involved in DNA replication and repair. Aim 3. To characterise novel genetic interactions at the molecular level for the identification of potential substrates targeted by YEN1ON in vivo. 101 2. RESULTS 2.1 Overexpression of YEN1 deregulated alleles results in lethality, cell cycle arrest and Rad53 phosphorylation In our first approach to understanding the importance of an adequate control of HJ resolvases, we started by asking whether elevated levels of different YEN1 alleles could have a detrimental effect on cell viability. Hence, derivatives of the integrative plasmid pAG306GAL-YEN1-FTH were inserted at the endogenous ura3-52 locus for inducible overexpression of Yen1 by addition of galactose. Then, cell survival was tested by spotting serial dilutions of each strain on YPDA (repression conditions) or YPRaff+GAL (induction conditions) solid medium plates. While the overexpression of the wild-type and YEN1KR alleles did not cause cell mortality when compared to an unmodified strain, the strain harbouring YEN1ON was not viable under induction conditions (Figure 30a). This indicates that, even when overexpressed, wild-type Yen1 activity is efficiently curtailed by its regulatory network, as inactivation of this inhibitory system is lethal to cells. Since the overexpression of the YEN1KR allele did not cause any overt effects, we decided to combine this mutation with those of YEN1ON to create a new allele, YEN1ON-KR, which should be deficient in all its regulatory layers (catalytic, localization, degradation) and ask whether it can further enhance the toxic effects of YEN1ON. Indeed, YEN1ON-KR overexpression phenotype is slightly more severe that of YEN1ON, specially at lower galactose concentrations (Figure 30b). This indicates that all regulatory layers contribute to cellular resistance to YEN1 overexpression, even if at different extents. During the course of these experiments, we decided to generate catalytically inactive versions of all the described YEN1 alleles by mutating the catalytic glutamate residues E193 and E195 to alanines, generating nuclease-dead (ND) versions for each. Surprisingly, we found that the expression of YEN1ONND and YEN1-ON-KR-ND resulted in comparable levels of cell death with respect to their catalytically proficient counterparts (Figure 30b). To ensure that this result was not due to residual catalytic activity in these mutants, Yen1ON-ND was purified to near-homogeneity (Figure 30c) and compared to Yen1ON in in vitro nuclease activity assays employing various fluorescently-labelled DNA substrates. As opposed to Yen1ON, which readily digested both HJ and 5’-flap synthetic substrates, no traces of endonucleolytic activity could be detected with Yen1ON-ND (Figure 30d). As expected, a linear dsDNA substrate did not reveal any non-specific exonucleolytic activity for either enzyme. These results imply that we cannot formally rule out that the underlying mechanism driving cellular lethality in strains overexpressing YEN1ON or YEN1ON-KR is catalysis-independent. Notwithstanding, it is also reasonable to speculate that different mechanisms are responsible for cell death in the YEN1ON or YEN1ON-KR mutants as opposed to those harbouring YEN1ON-ND or YEN1ON-KR-ND. These observations prompted us to test whether the observed lethality phenotype would be equally accompanied by an activation of the DDC in our overexpression system. Therefore, we induced asynchronous cultures in YPRaff with 2% galactose and took samples at 0, 3 and 5h after induction. Western blot analysis of protein whole-cell extracts revealed that all strains Tomás Lama Díaz 102 Figure 30. Overexpression of catalytically active and nuclease-dead versions of deregulated Yen1 causes cell death in a dose-dependent manner. (a,b) Ten-fold serial dilutions of strains with the indicated genotypes were spotted on YPDA or YPRaff containing different galactose concentrations and photographed after 2 days of incubation at 30 ºC. (c) SDS-PAGE analysis of purified Yen1ON-ND stained with Coomassie. (d) Verification of the absence of nuclease activity for Yen1ON-ND on three different substrates, using Yen1ON as a positive control. After native PAGE, fresh gels were scanned for 5’-6FAM signal in a Typhoon FLA 9500. (-) No enzyme. 103 express relatively comparable amounts of their corresponding Yen1 version. Moreover, only the strains with YEN1 alleles that caused cell death in our previous assays (YEN1ON, YEN1ONKR, YEN1ON-ND and YEN1ON-KR-ND) displayed an activated checkpoint 5h after induction, as denoted by Rad53 phosphorylation (Figure 31a). Interestingly, mild phosphorylation of Rad53 can already be observed at 3h after induction in those strains harbouring catalytically active alleles, suggesting a faster checkpoint activation that relies on Yen1 nuclease activity. Figure 31. Overexpression of deregulated versions of Yen1 leads to checkpoint activation and disrupts cell cycle progression. (a) Protein extracts were prepared from liquid cultures of strains overexpressing the indicted YEN1 allele grown in YPRaff and induced for 3-5h by addition of 2% galactose. Abundance of different overexpressed versions of Yen1 was compared by western blotting using anti-Flag antibodies. Activation of the DDC response was assessed by phosphorylation of Rad53 with anti-Rad53 antibodies. Pgk1 was employed as a loading control. (b) FACS analysis of DNA content in samples collected at the indicated timepoints for the same strains employed in (a). Tomás Lama Díaz 104 Additionally, FACS analysis of DNA content in these strains indicated that the overexpression of lethal alleles disrupts normal cell-cycle progression, displaying a flattened profile without clear 1C and 2C peaks after 5h, and with more cells apparently stuck with an intermediate DNA content (Figure 31b). 2.2 The lethality at strain overexpressing deregulated YEN1 is not dependent on checkpoint activation or homologous recombination, Figure 32. Prolonged cell cycle arrest by DDC activation is not the main cause of cell death in strains overexpressing YEN1ON variants. (a) Strains expressing different YEN1 mutant alleles in sml1∆ or sml1∆ rad53∆ backgrounds were streaked and examined for growth on solid medium in the absence (YPDA) or presence of galactose (YPGAL). Pictures were taken after 3 days at 30 ºC. (b) Abundance of Yen1 variants in the strains from (a) was compared by western blotting using anti-flag antibodies. Protein extracts were prepared from liquid cultures grown in YPRaff and induced for 3 h by addition of 2% galactose. (c) Ten-fold serial dilutions of strains with the indicated genotypes were spotted on YPDA or YPRaff with the indicated concentrations of galactose and photographed after 2 days of incubation at 30 ºC. 105 Given the observation of Rad53 phosphorylation after overexpression of misregulated YEN1 alleles, we next asked whether checkpoint activation itself was the cause of cell death in these strains. Therefore, we created a set of rad53∆ strains in a sml1∆ background (deletion of SML1, coding for an inhibitor of the ribonucleotide reductase, suppresses the lethality of RAD53 deletion (Zhao et al., 1998) that carrying a similar set of integrative plasmids as described above. While all strains were viable under non-inducing conditions, overexpression of YEN1ON, YEN1ON-KR, YEN1ON-ND and YEN1ON-KR-ND led again to cellular death in both the sml1∆ and sml1∆ rad53∆ backgrounds (Figure 32a,b). This indicates that an aberrant activation of the checkpoint by Yen1 deregulation is not the main cause of lethality in these strains. Contrarily, induction with lower galactose concentrations demonstrated that, in fact, checkpoint abrogation led to a slight decrease in the survival of these strains, suggesting that an intact checkpoint may contribute to counter the toxic effects of a mild overexpression of YEN1ON, YEN1ON-KR, YEN1ONND and YEN1ON-KR-ND (Figure 32c). Taking into account that Yen1 can process a number of different HR intermediates (Blanco et al., 2014; Carreira et al., 2022; Ip et al., 2008)), we employed a similar approach to assess if the observed cause of lethality was the untimely processing or blockage of recombination intermediates. We thus deleted the RAD52 gene, required for most forms of homologous recombination, in the same set of strains employed in Figure 30, but abrogation of recombination did not prevent the lethality associated to the overexpression of the YEN1ON, YEN1ON-KR, YEN1ON-ND and YEN1ON-KR-ND alleles (Figure 33a). Additionally, akin to our observations with checkpoint mutants, the presence of Rad52 reduces the toxic effects of weak overexpression of these alleles (Figure 33b). Altogether these results indicate that an incorrect processing of physiological HR intermediates is not the fundamental cause of the death in these strains and that, most likely, recombination plays a role in alleviating the detrimental consequences derived from uninhibited Yen1 actions. 2.3 Expression of Yen1ON at endogenous levels displays genetic interactions with HR helicases The lethal effect of high overexpression of the indicated regulation-refractory YEN1 alleles seems to lead to an “overkill” situation that cells cannot easily adapt to. In this sense, on top of the recombinationand checkpoint inactivation experiments shown here, an unbiased screening of spontaneous suppressors of YEN1ON overexpression only found mutations that either directly inactivated YEN1ON or that impaired the galactose-dependent overexpression system (like GAL4 mutants), which drastically reduced the cellular levels of Yen1ON (Vanesa HurtadoNieves and Miguel G. Blanco, unpublished results). Therefore, in order to gather insight on the potential targets of unrestricted Yen1 activity in vivo, we decided to resort to a strategy based on combining a strain expressing YEN1ON from its endogenous promoter with deletions in a series of candidate DNA repair genes. To test the feasibility of this approach, we first decided to investigate whether unrestricted YEN1ON would display any synthetic genetic interaction with the deletion of various candidate helicases involved in the processing of HR intermediates. We thus generated diploid strains carrying homozygous YEN1 or YEN1ON alleles and heterozygous deletions for various combinations of the SGS1, SRS2 and MPH1 helicases, MUS81 and/or Tomás Lama Díaz 112 when we tested the hypersensitivity to HU of the strains combining YEN1ON with each of these alleles, neither pif1-m1 nor pif1-m2 recapitulated the synthetic interaction observed with pif1∆ (Figure 38b). This result, together with the known petite phenotype of pif1∆ strains, would be compatible with a scenario where the observed interaction with YEN1ON might be due to a general lack of mitochondrial respiratory functions, rather than the specific absence of Pif1. To test this hypothesis, we isolated spontaneous petites in the YEN1 and YEN1ON strains, where Pif1 is both present at nucleus and mitochondria, and compared their HU hypersensitivity to that of pif1∆ YEN1ON strains. As shown in Figure 38c, the petite YEN1ON strains grew comparably to wild-type cells in the presence of HU, unlike the pif1∆ YEN1ON mutants, indicating that this synthetic interaction is specific for the loss of Pif1. Quite remarkably, subjecting the same set of strains to growth in medium containing glycerol (a non-fermentable carbon source) instead of glucose, led to two important observations. First, while the pif1∆ and the spontaneous petite strains failed to proliferate in this medium due to their disabled respiratory functions, pif1-m1 strains were able to grow normally, even when they should also display a petite phenotype (Figure 38c). This pointed to an incomplete separation-of-function in this mutant, which will be further explored in the next chapter of this PhD Thesis. Second, the double mutant pif1∆ YEN1ON, despite its hypersensitivity to HU, quite unexpectedly showed an increased proliferation capacity in glycerol-containing medium compared to pif1∆, suggesting some compensatory effect by Yen1ON in cells without mitochondrial function or preventing its loss in the absence of Pif1. This observation and its potential implications are being explored by other members of the laboratory at the time of writing this PhD Thesis and will not be further discussed here. Both our own results during the generation and handling of strains carrying the pif1-m1 and pif1-m2 alleles, as well as a thorough revision of the Pif1 literature led us to the conclusion that the most likely explanation for the failure of these alleles to recapitulate the pif1∆ synthetic interaction with YEN1ON was that the pif1-m2 mutants still retained a phenotypically relevant level of nuclear Pif1 activity. Therefore, we decided to generate improved separation-offunction alleles in order to ascertain the precise molecular intricacies behind the strong pif1∆ YEN1ON interaction. The rationale employed to generate those refined alleles, now termed pif1nuc and pif1mit (Figure 39a), will be addressed in the next chapter of this PhD Thesis. Importantly, by combining YEN1ON with these new set of alleles, we could confirm that the hypersensitivity of pif1∆ YEN1ON strains to HU is specifically due to the loss of Pif1 nuclear functions, since only the pif1mit YEN1ON strain can phenocopy this deficiency (Figure 39b). Altogether, our findings support that downregulation of HJ resolvases during S phase prevents the aberrant processing of replication or repair intermediates. Further characterization of the phenotypes associated to YEN1ON overexpression, or its hypersensitivity to genotoxic agents when combined with loss of nuclear Pif1, may help to elucidate the precise DNA structures that could be targeted by premature Yen1 activation. 113 3. DISCUSSION 3.1 Cell-cycle regulation of Yen1 prevents genome instability. Work from several groups has established that downregulation of HJ resolvases during S-phase is an evolutionarily conserved strategy to preserve genome integrity, even if its mechanistic implementation for specific nucleases in different organisms may vary. A general interpretation of this observation is that these restrictions prevent the aberrant processing of replication or repair intermediates. However, the precise DNA structures that could be targeted by precocious HJ resolvases remains unclear. In this work, we have carried out a preliminary characterization of the impact of premature activation of the S. cerevisiae Yen1 resolvase by taking advantage of the regulation-refractory YEN1ON mutant, which is constitutively active and nuclear throughout the cell cycle. An initial approach using galactose-inducible constructs demonstrated that overexpression of YEN1ON induces cell death in a dose-dependent manner, reaching complete lethality with full induction of the GAL1 promoter (Figure 30a), whereas similar overexpression of the wild-type enzyme is inconsequential. Interestingly, we expected such phenotype to be dependent on Figure 39. Hypersensitivity of pif1∆ YEN1ON strains to HU is caused by the loss of Pif1 nuclear function. (a) pif1nuc and pif1mit represent improved versions of pif1-m1 and pif1-m2, respectively. (b) Ten-fold serial dilutions of strains with the indicated genotypes were spotted on YPD containing HU and photographed after 2 days of incubation at 30 ºC. Tomás Lama Díaz 114 Yen1ON catalytic activity, but much to our surprise, we found that overexpression of the catalytically inactive, but still nuclear and DNA binding-proficient, YEN1ON-ND mutant also derived in cell death (Figure 30b). Taken at face value, this result would indicate that Yen1ON toxicity is independent of its nuclease activity. However, we favour an alternative explanation in which lethality upon YEN1ON overexpression does depend on its nuclease activity to process yet-undetermined branched DNA structures aberrantly, while the dominant negative effect of YEN1ON-ND would stem from a “Dog in the Manger”-like behaviour, by tightly binding to physiologically relevant branched DNA structures and preventing their normal processing by other enzymes. In both cases, cell death is accompanied by robust checkpoint activation, as judged by Rad53 phosphorylation (Figure 31a), and disruption of the normal cell-cycle profile, with an apparent accumulation of cells in S phase (Figure 31b). This could be compatible with the generation of DNA breaks or the persistence of long ssDNA tracts in these mutants, which would be in agreement with their ability to interfere with normal S-phase progression. Importantly, a prolonged cell-cycle arrest is not the main cause of cell death in these strains, as deletion of RAD53 cannot suppress it (Figure 32a,c). Since Yen1 is not only able to process HJs, but also other branched structures that may appear during HR, like D-loops, we employed a similar simplistic rationale to verify if the primary cause of lethality in these cells was the unscheduled processing of recombination intermediates. However, overexpression of YEN1ON and YEN1ON-ND was not only still deleterious in a rad52∆ background (Figure 33a), but in fact HR seems to counter the toxic effect of Yen1 misregulation (Figure 33b). 3.2 Upon genotoxic stress, nuclear Pif1 may prevent the accumulation of DNA secondary structures that Yen1ON could hydrolyse. These observations, together with results from other members of the lab, indicated that YEN1ON overexpression leads to massive genome instability that cannot be suppressed by mutations in other genes. Therefore, in order to gain further information on the DNA metabolism pathways that may be disrupted by premature resolvase activity, we changed our approach and started to search for genetic interactions between YEN1ON expressed at endogenous levels and other DNA repair factors, unveiling two main types of interactions. On the one hand, the presence of Yen1ON at endogenous levels seems to have a beneficial effect in those situations where cellular viability is compromised by the accumulation of unresolved recombination intermediates. This situation arises as consequence of either the simultaneous mutations of HR genes leading to synthetic lethality or sickness (Figure 34) or the combination of single mutations with genotoxic agents (Figure 36 and Figure 37), confirming and extending previous results (Blanco et al., 2014; Elango et al., 2017). In this context, YEN1ON could efficiently suppress the same set of genetic interactions as RAD52 deletion, confirming that Yen1ON provides an alternative therapeutic route against the “sickness/death-by-recombination” phenotype, but not in the context of other synthetic lethalities (Figure 35). On the other hand, we found that in some genetic backgrounds expression of YEN1ON leads to increased hypersensitivity to genotoxic agents like MMS, HU or CPT (Figure 36 and Figure 37),). In particular, the YEN1ON pif1∆ 115 interaction (Figure 37c,d) was particularly exciting due to its strong synthetic effect and that the Pif1 helicase is one of the main actors in the handling of multiple secondary DNA structures. Therefore, we decided to further delve into the underpinning molecular mechanism of this interaction as a way to better understand the relevance of Yen1 regulation. However, not far into this line of research we encountered a technical obstacle that prevented its further development: attempts to recapitulate this genetic interaction with any of the two classical separation-of-function alleles for PIF1, pif1-m1 (nuclear) and pif1-m2 (mitochondrial), systematically failed to recapitulate the genetic interaction between pif1∆ and YEN1ON (Figure 38a,b). Given the pleiotropic effects of the full deletion, (nuclear defects and loss of mitochondrial DNA), we could not rule out an indirect interaction caused by the petite phenotype of pif1∆ mutants, which could be uninformative for our purposes. Although we managed to discard the hypothesis of a genetic interaction between YEN1ON and the petite condition (Figure 38a,b), we decided that, in order to understand the underlying molecular intricacies of this interaction, we needed to find out the reason behind the failure of pif1-m1 and pif1-m2 to recapitulate the YEN1ON pif1∆ interaction. This ultimately became the focus for the rest of my PhD Thesis, which will be laid out in Chapter 2. Importantly, once we had managed to generate improved separation-of-function alleles for nuclear and mitochondrial Pif1, we demonstrated that the increase in hypersensitivity to genotoxic agents in YEN1ON pif1∆ mutants is specifically connected to the loss of Pif1 nuclear function (Figure 39). The most intuitive explanation for such results would be that in the absence of nuclear Pif1 those DNA secondary structures that this helicase contributes to unwind persist for longer times, either becoming substrates themselves or indirectly contributing to the appearance of new substrates for Yen1ON activity. Alternatively, we cannot formally rule out that, as in the case of Rad52, nuclear Pif1 is important to avert the damage generated by Yen1ON under genotoxic stress. 3.3 Future plans While, due to time constraints, we could not reach further conclusions at this point, I would like to lay out some of our future plans to delineate the molecular mechanism that underpins the lethality of YEN1ON overexpression, as well as the hypersensitivity to DNA damaging agents of Yen1ON strains in the absence of nuclear Pif1. Since the still inceptive stage of this project complicates the elaboration of data-driven hypothesis, for the remaining part of this discussion I will focus on different experimental approaches that could be useful to address some relevant questions. Firstly, it would be of interest to investigate whether the overexpression of both YEN1ON and YEN1ON-ND is concomitant with the occurrence of DNA breaks that could explain the subsequent lethality under these conditions. For this, physical analysis of chromosomal breakage by pulsedfield electrophoresis would be employed, both in asynchronous and synchronous cultures, followed by general DNA staining and/or probing for particular loci. If the integrity of a specific chromosome is compromised, it could point towards a particular genomic region (like the rDNA array in chromosome XII (Petes, 1979) being affected by Yen1 unscheduled activity. Alternatively, if all chromosomes are affected, resembling a chromosome pulverization Tomás Lama Díaz 116 phenotype (Crasta et al., 2012; Duda et al., 2016), it would imply the unscheduled cleavage of widespread secondary structures arising during S phase, such as replication forks, hairpins or G4s. Similarly, bidimensional gel electrophoresis analysis followed by southern blotting could be employed to examine the accumulation and processing of ongoing replication intermediates upon premature activation of Yen1. Finally, it would be also interesting to tease out the route for checkpoint activation after YEN1ON or YEN1ON-ND overexpression. This could be determined by assessing the phosphorylation status of Rad53 in our assays in strains lacking the checkpoint apical kinases Mec1 and Tel1 or the Rad9 and Mrc1 mediators (Waterman et al., 2020) Ideally, genome-wide techniques could be applied to map the binding and cleavage sites of prematurely activated Yen1 in phases when it does not normally have access to the nucleus. In this sense, the lethality of overexpressed Yen1ON-ND suggests a very tight binding to its substrates, in line with the previously described dominant negative effect of YEN1ND expression in mus81∆ strains (Blanco et al., 2010). This “substrate-trapping” behaviour could be exploited to interrogate the in vivo binding sites of Yen1 in S phase by ChIP-seq, which would hint at particular secondary DNA structures targeted by this nuclease. Moreover, it would be also interesting to obtain a genome-wide profiling of Yen1-dependent incisions, taking advantage of the explosion of techniques to map DNA breaks that have been developed in the recent years, including BLESS (Crosetto et al., 2013), END-seq (Canela et al., 2016), DSB-capture (Lensing et al., 2016), BLISS (Yan et al., 2017), i-BLESS (Biernacka et al., 2018), qDSB-seq (Zhu et al., 2019), CNN-seq and sBLISS (Bouwman et al., 2020) for DSBs and SSiNGLe (Cao et al., 2019), GLOE-seq (Sriramachandran et al., 2020), Nick-seq (Cao et al., 2020) and TRAEL-seq (Kara et al., 2021) for mapping of SSBs (Cao et al., 2019; Kara et al., 2021; Sriramachandran et al., 2020). Importantly, some of these methodologies have already been successfully applied to the identification of cleavage sites by other SSEs, like Mus81, even in a quantitative manner… Algunas técnicas y han demostrado los cortes de Mus81, y variants cuantitivvas. Most of the techniques described to characterize the reason behind YEN1ON overexpressiondependent lethality could be similarly employed to unravel the underlying mechanism for the hypersensitivity of pif1mit YEN1ON strains to DNA damaging agents, such as pulse-field electrophoresis and mapping of binding sites with Yen1ON-ND. In the latter case, it would be interesting to employ for comparison purposes the pif1K264A allele, which renders a catalytically inactive, but binding proficient version of Pif1 that has already been employed in ChIP-seq experiments to map its binding sites genome-wide (Paeschke et al., 2011). Similarly, mapping of DNA breaks produced by Yen1ON in a pif1-mit background could also help to delineate the structures disrupted in pif1mit YEN1ON mutants upon HU treatment. Given the multiple factes of DNA metabolism where Pif1 plays a role, we will also take advantage of PIF1 mutants defective in pathway-specific protein-protein interactions, like pif1R3E, pif1-PIPmut or pif1-4A. pif1-R3E impairs the PCNA-Pif1 interaction required to promote DNA synthesis during BIR, whereas pif1-PIPmut is necessary for Pif1-assisted replisome progression through G4s (Buzovetsky et al., 2017; Dahan et al., 2018). Since the DDR phosphorylates Pif1 at is CTD to promote BIR and, at the same time, prevent telomerization of DSBs, the pif1-4A phosphomutant allele is both defective at BIR stimulation and inhibition of 117 telomerase activity at chromosomal breaks (Makovets and Blackburn, 2009a; Vasianovich et al., 2014). A specific interaction of YEN1ON with any of these alleles could also pinpoint the pathways affected by the premature activation of Yen1. However, it is important to consider that the interaction of YEN1ON with this type of alleles might not resemble that with pif1mit because of two main reasons: i) a complex/additive effect of YEN1ON through the disruption of multiple pathways in which nPif1 is involved under replicative stress, including the repair of YEN1ON-dependent DNA break; ii) incomplete separation of function of those alleles, as in the case of pif1-m2. Given such potential limitations, we also intend to combine YEN1ON with mutations in other genes that render cells deficient in specific pathways in which Pif1 plays a role. For instance, the lack of interaction between rrm3∆ and YEN1ON indicates that the underlying cause of the pif1mit YEN1ON phenotype can be attributed to Pif1-specific functions or those in which Rrm3 plays a backup role, such as fork progression through G4-rich regions, inhibition of telomerase, support of the RFB and BIR. To determine whether impairment of fork progression through G4s underlies the interaction of pif1mit YEN1ON, we propose several strategies. First, we will combine pif1mit allele with rrm3∆ and YEN1ON, considering Rrm3 back-up role in promoting genome stability at G4 motifs (Paeschke et al., 2013). Furthermore, we will combine YEN1ON with deletions of MMS1 and the hypomorphic RPA allele rfa1-D228Y. Mms1, a subunit of an E3 ubiquitin-ligase complex, binds G4 structures and recruits Pif1 to those regions (Wanzek et al., 2017), whereas RPA cooperates with Pif1 to remove G4-quadruplex structures. In particular, the rfa1-D228Y mutant reduces RPA ssDNA binding activity, especially to G-rich tracts (Audry et al., 2015; Maestroni et al., 2020). Finally, we will test the effects of G4 chemical stabilisers, such as Phen-DC3 (Lopes et al., 2011; Piazza et al., 2010) in YEN1ON and pif1mit YEN1ON strains, both in terms of cell viability and chromosomal integrity. To address if Pif1dependent telomere homeostasis or telomere healing at DSBs is altered in the presence of YEN1ON, the separation-of-function alleles cdc13-S306A and est2-up34 could be employed. Mec1 phosphorylates Cdc13 at S306 to prevent its accumulation at DSBs, and the S306A substitution increases telomere healing to levels comparable to PIF1-deficient strains (Zhang and Durocher, 2010). In contrast, est2-up34, a mutant version of the catalytic subunit of telomerase, prevents the Pif1-dependent downregulation of Est2 activity, leading to extended telomeres (Eugster et al., 2006). Similarly, a possible effect on the stability of the rDNA or the maintenance of the replication fork barrier, which is partially dependent on Pif1 could be assessed in fob1∆ mutants. Finally, a potential interference by YEN1ON with the repair of singleend DSBs by BIR, which requires Pif1 for extensive D-loop-associated DNA synthesis, could also be revealed in mutants of the POL32, the subunit of Polδ indispensable for this repair pathway (Lydeard et al., 2007; Wilson et al., 2013). We believe that with such results in hand we could compose a more precise picture of the physiological events that cells are shielding from the action of SSEs by delaying their activation until the late stages of the cell cycle. Tomás Lama Díaz 118 CHAPTER 2: PIF1 TRANSLATIONAL MECHANISM Pif1 has two different isoforms in S. cerevisiae, allowing the dual-targeted localization required to fulfil its roles in both nucleus and mitochondria. This is achieved by ATI of PIF1 mRNA from the first two in-frame AUG codons (AUG1 and AUG40), which flank a MTS. When translation starts from AUG1, the N-terminal MTS quickly directs the full-length precursor (98 kDa) to the mitochondria. Once imported into the matrix, the MTS is proteolytically removed, rendering the mature mitochondrial isoform (mPif1, 92 kDa), starting at arginine 46 (Lahaye et al., 1993b). Instead, if translation initiation starts at AUG40, a 93 kDa isoform lacking the MTS is produced and subsequently imported into the nucleus (nPif1) (Figure 40). Figure 40. ATI from AUG1 and AUG40 produces Pif1 mitochondrial and nuclear isoforms. Translation initiation from the first start codon (AUG1) renders the full-length precursor of mPif1. Alternatively, translation from the second in-frame AUG (AUG40) produces nPif1. After proteolytical cleavage of the mPif1 precursor before Arg46, both mature isoforms will differ only in 6 amino acids. mPif1, mitochondrial isoform of Pif1; nPif1, nuclear isoform of Pif1; MTS, mitochondrial targeting signal. 119 The described ATI model inspired the development of two separation-of-function alleles for the mitochondrial (pif1-m1) and nuclear (pif1-m2) functions of Pif1 (Schulz and Zakian, 1994a). In pif1-m1, the production of mPif1 is abolished by mutation of ATG1, whereas nPif1 can still be translated from the intact AUG40 and imported into the nucleus, presumably through a yetuncharacterized NLS (Figure 41a). Conversely, in pif1-m2, the mutation of ATG40 abrogates the production of nPif1 without affecting the translation of mPif1, resulting in nPif1-deficient strains (Figure 41b). Given the extensive research into the molecular mechanisms that ensure the stability of the nuclear genome, pif1-m2 has been widely exploited to interrogate the roles Figure 41. Schematic representation of PIF1 separation of function alleles with their predicted molecular weight and subcellular localization. (a) m1 indicates mutation of the first ATG codon. pif1-m1 produces exclusively a nuclear isoform starting at Met40. (b) m2 stands for mutation of the second ATG codon, at position 40. pif1-m2 mRNA is only translated as the mitochondrial Pif1 full-length precursor. Like in wild-type cells, mPif1 precursor is proteolytically cleaved upon mitochondrial import. mPif1, mitochondrial isoform of Pif1; nPif1, nuclear isoform of Pif1; MTS, mitochondrial targeting signal. Tomás Lama Díaz 120 of nPif1 in nuclear DNA replication and repair, avoiding the slow growth and petite phenotypes conferred by the loss of mitochondrial Pif1 activity (Bochman et al., 2010). Despite being frequently mislabelled in the literature as a nuclear-null allele, it was acknowledged since their initial description (Schulz and Zakian, 1994a) that pif1-m2 mutants could retain traces of nuclear protein, as inferred from their partial suppression of some of the pif1∆ phenotypes in different genetic set-ups. In this sense, published data indicate that pif1-m2 consistently behaves as a poor separation-of-function allele in three specific contexts. First, the extended telomere length of pif1∆ strains is only partially recapitulated by pif1-m2 mutants, suggesting a partial inhibition of telomerase by nuclear Pif1 (Schulz and Zakian, 1994a). Also related to telomeric phenotypes, presence of pif1-m2 inhibits the growth of exo1∆ cdc13-1 strains when compared to the full knockout or the expression of the helicase-dead version pif1K264A (Dewar and Lydall, 2010). Second, pif1-m2 can promote stability at G4 tracts. Replication time through G4 regions is significantly slower in pif1∆ than in pif1-m2 cells (Dahan et al., 2018), and the elevated rate of GCRs at the GC-rich CEB1 minisatellite decreases from 20-fold in pif1∆ to wild-type levels in pif1-m2 cells (Ribeyre et al., 2009). This has led other authors to acknowledge that G4-dependent GCRs are probably underestimated in pif1-m2 strains (Paeschke et al., 2013). Third, the reduction in BIR efficiency, with a concomitant increase in aberrant half-crossovers, is more pronounced in pif1∆ than in pif1-m2 cells (Buzovetsky et al., 2017). Likewise, whereas PIF1 deletion, in addition to half-crossovers, also leads to microhomology-mediated-BIR (MMBIR) upon BIR collapse, pif1-m2 cells minimize MMBIR by promoting longer-tracts of BIR synthesis (Sakofsky et al., 2015). Moreover, one screening employing pif1-m2 to search for novel interactions of PIF1 failed to report several of the previously described interactions of pif1∆; importantly, as pointed out in the same study, this could be explained either by remnants of nPif1 in pif1-m2 mutants, or because some synthetic phenotypes described for pif1∆ are caused by its mitochondrial defects (Stundon and Zakian, 2015a). Given the suspected presence of nPif1 derived from pif1-m2, alternatives like placing PIF1 under the control of the CLB2 promoter have been employed to suppress its expression during the meiotic cell cycle (Vernekar et al., 2021; Ziesel et al., 2022). However, when no alternative to pif1-m2 is available, pif1∆ is still widely employed, either on its own or with pif1-m2 to compare the phenotypes of both alleles. Our own results in the context of YEN1ON mutants described at the end of Chapter 1 (Figure 38) confirm those in the literature showing that the commonly employed pif1-m2 allele does not fully recapitulate the pif1∆ phenotypes ascribed to its nuclear functions, suggesting again that Pif1 nuclear activity is retained to some extent in those mutants. Since employing pif1∆ strains to investigate the roles of Pif1 or other DNA repair factors raises important caveats due to the pleiotropic effects of the full deletion, in this Chapter 2 we investigated the extent of separation of function of pif1-m2 in different genetic set-ups, the potential source of nuclear activity from the pif1-m2 alleles, and the basis of the molecular mechanism of ATI for Pif1. Finally, we took advantage of our better understanding of those questions to develop improved separation-offunction alleles for PIF1. 121 1. OBJECTIVES Aim 1. To compare pif1∆ phenotypes with those of pif1-m2 in different genetic set-ups. Aim 2. To search for potential novel isoforms of Pif1 with nuclear activity. Aim 3. To characterise the molecular mechanism controlling the production of Pif1 isoforms. Aim 4. To generate improved separation-of-function alleles for PIF1. Tomás Lama Díaz 128 isoform. However, the underlying mechanism controlling start codon selection in PIF1 mRNA, which could be relevant for the design of improved separation-of-function alleles, was still unknown. 2.3 Pif1 isoforms are produced through ribosomal leaky scanning Translation involves four main steps: initiation, elongation, termination and recycling. Here, we will focus on the initiation step to investigate how different AUG codons are employed in order to produce three Pif1 isoforms from the same mRNA. During the canonical cap-dependent translation, eukaryotic initiation factors (eIFs) interact with the mRNA to initiate the process (Figure 46). In particular, one member of the eIF4F complex, eIF4E, recognizes the m7G cap structure at the 5’ end of the mRNA. Separately, the ternary complex comprising eIF2 complex, initiator tRNA, and GTP binds the 40S ribosomal subunit together with eIF3, forming the 43S pre-initiation complex (PIC). Next, the interaction between another member of the eIF4F complex (eIF4G) and eIF2 recruits the PIC to the 5’end of the mRNA (Hinnebusch, 2011; Figure 46. Schematic representation of canonical cap-dependent translation initiation. The eIF4F complex recognizes the m7G cap and recruits the PIC to the 5’ end of the mRNA. The PIC scans the mRNA in 5’-3’ direction until the Met-tRNA recognizes the start codon, which promotes the assembly of the 60s subunit and beginning of protein synthesis. 129 Leppek et al., 2018). The scanning model postulates that the PIC moves along the mRNA in the 5’-3’ direction, aided by ATP-powered helicases (Figure 46). Once the first AUG codon is encountered, the Met-tRNA pre-loaded in the PIC complex recognizes it and halts the scanning, leading to the release and recycling of eIF2 and other initiation factors. The dissociation of initiation factors allows the engagement of the 60S large ribosomal subunit with the remaining of the PIC complex, resulting in a functional ribosome ready for protein synthesis (Figure 46). The elongation phases proceed with the addition of codon-dependent amino acids to generate a polypeptide chain. Finally, during termination and recycling, the polypeptide is released and the ribosomes are dissociated from the mRNA. Importantly, according to the “first-AUG rule” in the canonical cap-dependent mechanism, translation is initiated at the AUG codon closest to the 5’ end of the mRNA (Kozak, 1989; Kozak and Shatkin, 1978). Nonetheless, various mechanisms of alternative translation initiation in eukaryotes enable novel initiation events to occur downstream of the first AUG (Figure 47), including reinitiation, ribosomal shunting, internal ribosomal entry sites (IRES), and ribosomal leaky scanning (Hinnebusch, 2011a; Sriram et al., 2018). Reinitiation (Figure 47a) facilitates the translation of two or more consecutive open reading frames (ORF) from a single mRNA. It involves the recycling of the translational machinery upon termination of the upstream ORF (uORF) by reinitiation factors that allow the 40S subunit to remain bound to the mRNA and reacquire the initiator tRNA, thus resuming 5’-3’ scanning and enabling the translation of the downstream ORF (Hinnebusch, 2011a; Sriram et al., 2018). In yeast, GCN4 constitutes the best characterized example of how reinitiation operates to regulate protein expression. Reinitiation is generally inefficient, suggesting that uORFs are regulatory elements that limit production of the protein encoded by the downstream ORF. Consequently, this mechanism leads to the production of two different peptides, but it cannot generate two dual-targeted isoforms from a single mRNA that differ only in the N-terminal region. Ribosome shunting is another cap-dependent translation mode initially discovered in viruses. The PIC complex is recruited to the mRNA in a cap-dependent manner and starts scanning in 5’-3’ direction. Upon encountering a stem-loop structure (Figure 47b), the 40S subunit bypasses or shunts over the obstacle and lands at a downstream start codon to initiate translation (Kwan and Thompson, 2019; Sriram et al., 2018). Although ribosomal shunting is both rare and the less studied ATI mechanism in eukaryotic cells, it has been demonstrated that it can operate in budding yeast (Chappell et al., 2006)). However, we could not find any yeast mRNA that relies exclusively on this mechanism for the generation of alternative protein isoforms. Interestingly, ribosome shunting can be relevant in mRNAs with a short uORF, facilitating reinitiation by allowing the 40S subunit to shunt other short ORF or secondary structure to reinitiate at a downstream ORF. However, as a facilitator of reinitiation, ribosomal shunting raises the same caveats described above to produce dual targeted isoforms. IRESs are RNA sequences that can recruit 40S ribosomal subunits internally to the mRNA, independently of the 5’ cap-dependent initiation (Figure 47c). IRESs can be unstructured or contain stable secondary and tertiary structures, and their lengths vary from 9 to 1000 nt (Kwan Tomás Lama Díaz 130 and Thompson, 2019; Sriram et al., 2018). Like ribosome shunting, IRES were initially discovered in viruses, but they were also later identified in eukaryotic mRNAs (Johannes and Sarnow, 1998), where they mediate translation of different protein isoforms from important genes, such as TP53 (Ray et al., 2006) or DMD (Wein et al., 2014). In Pif1, an IRES located between AUG1 and AUG40 might recruit 40S subunit to initiate the translation of the nuclear isoform at AUG40, and mutation in pif1-m2 could result in ribosome scanning until the next AUG. However, IRES are specially relevant upon situations of cellular stress when 5’ capdependent translation is impaired; for example, in yeast, IRES are particularly relevant upon glucose starvation (Gilbert et al., 2007), whereas both Pif1 isoforms are constitutively produced in unperturbed conditions. Moreover the majority of experimentally validated IRES in yeast from IRESbase (Zhao et al., 2020) are longer than 150 bp, compared to the 114 bp between AUG1 and AUG40 in Pif1. Hence, it seems unlikely that an IRES is mediating isoform production in Pif1. Finally, leaky scanning (Figure 47d) involves the ribosomal skipping of the first AUG codon and initiation of translation at a downstream, in-frame AUG codon (Kozak, 2002). In leaky scanning, the efficiency of any given AUG for translation initiation depends on the sequence context around this codon, commonly known as the Kozak sequence. Interestingly, the AUG1 in PIF1 mRNA is flanked by a suboptimal Kozak sequence, which would increase the probability of the ribosome to bypass AUG1 (that produces mPif1) and reach AUG40 to start the translation of nPif1. Moreover, the context around AUG40 is also suboptimal, which suggests that another event of leaky scanning might allow a minor fraction of scanning ribosomes to reach AUG107. This scenario seems both compatible with the low amounts of Pif1107-859 produced in PIF1 and pif1-m1 cells and with the selective enrichment of this isoform in pif1m2 cells due to mutation of AUG40 (Figures 44 and 45). Therefore, we decided to investigate if indeed a leaky scanning mechanism is regulating the abundance of Pif1 isoforms. In addition to AUG107, there are other four in-frame AUG codons downstream AUG40 and upstream the Pif1 helicase domain, which begins at Phe255. The first cluster would include AUG107 itself, together with AUG113 and AUG127, while the second one comprises AUG187 and AUG196. Initiation of translation from any of these positions would result in a Pif1 isoform that retains an intact helicase domain (Figure 48a). Given the also suboptimal Kozak context of AUG107 and the proximity of downstream codons, we cannot rule out that a fraction of scanning ribosomes could also skip AUG107 and reach AUG113, specially in pif1-m2 cells. Importantly, as we aim to improve the pif1-m2 allele, the scanning model predicts that mutation of ATG107 in the pif1-m2 would result in ATI from the next downstream AUG. To test the leaky scanning model, we took two complementary approaches. First, we ectopically expressed PIF1 alleles with single conservative methionine-to-isoleucine mutations at positions 107, 113, 127, 187 and 196 and determined whether the expression of the fast-migrating isoform was affected, but none of these single mutations eliminated the fast-migrating band (Figure 48b). This could be coherent with the proposed model of leaky scanning, as mutations in a given AUG would simply enable the scanning ribosomes to reach the next downstream 131 Figure 47. Schematic overview of the main mechanisms of ATI from different AUGs. (a) Reinitiation after translation of the uORF creates two different peptides. (b) Shunting of the 40S subunit upon encountering a stable secondary structure can favour translation reinitiation at a distant downstream AUG. (c) An IRES located between the first two AUGs could lead to the production of two isoforms. Translation from the first AUG would produce a full-length isoform, whereas the IRES could recruit 40S subunits to the downstream AUG, producing shorter isoforms lacking N-terminal regions that could contain targeting sequences or other functional domains. (d) In leaky scanning, the suboptimal context surrounding the first AUG allows a fraction of ribosomes to skip it and continue scanning until a downstream AUG is reached, generating an N-terminal truncated isoform. Alternatively, some ribosomes start translation even in the presence of a poor Kozak sequence, producing the full-length isoform. Tomás Lama Díaz 132 methionine. In contrast, double Met-to-Ile substitutions in the first cluster revealed both changes in the relative amounts and mobility of the fast-migrating band, demonstrating that translation initiation may occur from any of these AUGs, albeit M107 appears as the most relevant, followed by M113 (Figure 48c). Crucially, the fast-migrating band became undetectable when these three methionines were simultaneously mutated, without further detection of smaller bands (Figure 48d). Therefore, the amount of scanning ribosomes reaching AUG187 or AUG196 in the triple mutant is not enough to produce a detectable band, in accordance with the reported decay of ribosomes after scanning ~150 triplets (Chappell et al., 2006). These results confirm the existence of, at least, a third cellular Pif1 isoform that is translated from M107 (hereinafter, Pif1107-859) and demonstrate that M113 and M127 can also function as alternative starting points upon mutation of upstream start codons. To verify that Pif1 translation is specifically under the control of a leaky scanning mechanism, we mutated the sequences surrounding the AUG1 and AUG40 codons to match the best Kozak consensus sequence in S. cerevisiae (Figure 49a). According to the leaky scanning model, improving the Kozak context should reduce the ribosomal skipping of AUG40, and, as a consequence, the translation of Pif1107-859. We mutated the Kozak contexts around AUG1 in PIF1 or AUG40 in both PIF1 and pif1-m1. For AUG1, we only optimized the upstream nucleotides at the highly conserved -4, -3 and -2 positions (Hamilton et al., 1987; Hernández et al., 2019), but we did not mutate the +4 residue since there is no conservative change for proline. Figure 48. ATI can initiate translation from a cluster of ATGs downstream ATG40. (a) Schematic representation of the 5’ region of PIF1, depicting the ATGs before the helicase domain, whose expression could give rise to N-terminal truncated -but still catalytically activeisoforms. (b,c,d) Western blot analyses of the changes in the abundance and electrophoretic mobility of the fast-migrating isoform in single (b), double (c) and triple (d) mutants of downstream methionines to isoleucines. Protein extracts were prepared from strains ectopically expressing different variants of pif1-m2 with the indicated Met to Ile substitutions. Red arrowheads in (c) and (d) point to changes in the electrophoretic mobility of the fast-migrating isoform. 133 With respect to AUG40, we designed synonymous changes in its two surrounding amino acids to match the Kozak consensus in the two key positions -3 and +4. Next, we ectopically expressed these PIF1 and pif1-m1 alleles and verified the expression of Pif1107-859 by westernblotting. Confirming our hypothesis, Pif1107-859 could not be detected in those cells harboring constructs with optimal Kozak contexts at AUG40 (Figure 49b). Also, optimization of AUG1 in PIF1 cells produced an apparent increase in mobility of the Pif1 doublet that would be consistent with increased proportion of mPif1 over nPif1. Overall, our data demonstrate that the mechanism controlling the alternative translation of Pif1 isoforms is ribosomal leaky scanning. 2.4 Pif1107-859 is catalytically active and retains partial functionality in the nucleus. After identifying the underlying mechanism of ATI in PIF1, we investigated if the increased abundance of Pif1107-859 accounts for residual Pif1 activity described for pif1-m2. Although the NTD is dispensable for Pif1 activity, its in vivo roles may be affected by an N-terminal truncation, given that the NTD, in addition to containing Pif1 MTS, is post-translationally modified to control Pif1 nuclear activities (Ononye et al., 2020; Rossi et al., 2015). Consistent with the lack of MTS, cells harbouring exclusively the pif1107-859 allele display a petite phenotype and are unable to grow on non-fermentable carbon sources (Figure 50a). Moreover, we integrated fusion constructs with various PIF1-eGFP alleles under the control of the inducible GAL1 promoter in a strain with endogenous, untagged pif1-m2, to assess the subcellular localization of pif1107-859 by fluorescence microscopy. pif1-m1 and pif1-m2 were employed as controls for nuclear and mitochondrial localization, respectively. As expected, Pif1-eGFP signal could be detected both in the nucleus and mitochondria in PIF1 cells, while in those with pif1-m1 or pif1107-859 a clear nuclear enrichment was observed, confirming that nuclear localization is not impaired in pif1107-859 (Figure 50b). In agreement with the presence Figure 49. Optimization of Kozak context surrounding AUG40 abrogates the expression of the fast-migrating isoform. (a) Kozak consensus sequence in S. cerevisiae according to Hamilton et al.,1987 (b) Changes in the expression of Pif1107-859 (red arrow) in PIF1 and pif1-m1 strains with improved Kozak sequences were assayed by western blotting, as described for previous Figures. The specific mutations introduced to optimize the Kozak context of AUG1 and AUG40 are shown in yellow. Tomás Lama Díaz 134 of increased levels of Pif1107-859 in pif1-m2 strains, these mutants displayed a mild accumulation of nuclear Pif1, in addition to the predicted mitochondrial localization (Figure 50b). Next, we investigated if the cause of the incomplete separation of function in pif1-m2 alleles is due to the increased levels of nuclear Pif1107-859. Since its NTD is relevant for Pif1 function in different contexts, we first decided determined the biological functionality of Pif1107-859 using Figure 50. In vivo characterization of Pif1107-859 localization, mitochondrial functionality, and toxicity by overexpression. (a) The indicated strains were streaked and examined for growth on solid medium containing fermentable (YPD, 2% glucose) or non-fermentable carbon sources (YPEGly, 3% ethanol, 3% glycerol). Pictures were taken after 4 days at 30 ºC. (b) Subcellular localization of Pif1 isoforms in live yeast cells overexpressing different PIF1-eGFP alleles. DNA was visualized with DAPI staining and cell contour by DIC microscopy. (c) Tenfold dilutions of strains with the indicated genotypes were plated on YPD or YPGAL plates and imaged after 2 days of incubation at 30 ºC. (d) Abundance of Pif1 isoforms in the strains from (c) was compared by western blotting using anti-GFP antibodies. Protein extracts were prepared from liquid cultures grown in YPRaff and induced for 3 h by addition of 2% galactose. Ponceau staining of the PVDF membrane was used as a loading control. 135 as a proxy the lethality caused by the overexpression of nuclear Pif1 (Chang et al., 2009; Lahaye et al., 1991; Ononye et al., 2020). For that, we employed the same PGAL1-PIF1-eGFP expressing constructs to assess cell viability under overexpression conditions of different PIF1 alleles in galactose-containing medium. While the overexpression of PIF1 or pif1-m1 derived in extensive cell death, overexpression of pif1107-859 caused no detectable reduction in cell viability when compared to an empty vector (Figure 50c), despite their comparable protein levels (Figure 50d). Since the lack of toxicity of the overexpression of Pif1107-859 could be indicative of a loss of function of this isoform, we decided to analyse its catalytic activity in vitro. For this purpose, we expressed Pif1, Pif1107-859 and the catalytically inactive Pif1KA (Zhou, 2000b) as Nterminally 6His-tagged fusion proteins in E. coli and purified them to compare their ATPase and helicase activities. ATP hydrolysis time-courses with the three Pif1 variants were analysed by thin-layer chromatography (TLC) and phosphorimaging to detect radioactively labelled ATP or free inorganic phosphate (Figure 51a,b). Helicase activity was measured by strand Figure 51. Functional characterization of purified Pif1107-859. a) ATP hydrolysis by Pif1 and Pif1107-859 was measured by TLC and phosphorimaging. The catalytically inactive Pif1KA mutant was employed as a negative control. (b) Quantification of ATPase activity assays as displayed in (a). Data are represented as mean values ± standard deviation (n = 3). (c) Native PAGE analysis of the unwinding of a 5’-6FAM-splayed arm substrate by Pif1 and Pif1107-859, using Pif1KA as a negative control. Gels were scanned for 6FAM signal in a Typhoon FLA 9500. (d) Quantification DNA unwinding activity as displayed in (c). Data are represented as mean values ± standard deviation (n = 3). H, heat-denatured substrate. (-), no enzyme. Dashed line indicates splices lanes from the same gel. Tomás Lama Díaz 136 displacement assays with fluorescently labelled splayed-arm substrates (Figure 51c,d). Pif1107859 displayed at least comparable (or higher) catalytic activity to Pif1 in both ATPase (Figure 51b) and helicase assays (Figure 51d), confirming that this isoform is fully functional in vitro. Given these results, we next addressed if Pif1107-859 could substitute for nPif1 specifically in those genetic interactions where the pif1-m2 allele did not recapitulate the pif1∆ phenotypes (Figure 42 and Figure 43). We included pif1-m1 as a control of Pif1 nuclear activity since the petite phenotype of cells harboring pif1107-859 complicates its comparison with nPif1 expressed from PIF1. First, tetrad dissection analyses showed that dia2∆ pif1107-859 mutants displayed reduced synthetic sickness compared to dia2∆ pif1∆ mutants, indicating that Pif1107-859 can support cell growth similarly to Pif1 in dia2∆ mutants (Figure 52a). Second, compared to the lethality of dna2∆ in PIF1 and pif1-m1 strains, dna2∆ pif1107-859 mutants are able to grow. However, they form smaller colonies than dna2∆ pif1∆ mutants, indicating that in this context Pif1107-859 is partially functional (Figure 52b). Third, top3∆ pif1107-859 strains failed to grow after tetrad dissection, similarly to top3∆ pif1∆ strains, suggesting that Pif1107-859 cannot substitute for Pif1 in these conditions (Figure 52c). Finally, pif1107-859 could not reduce the hypersensitivity to HU observed in rad3-102 pif1∆ mutants, unlike pif1-m1 (Figure 52d). Therefore, pif1107-859 resembles pif1-m2 in dia2∆, top3∆ and dna2∆ mutants, confirming that Figure 52. In vivo characterization of pif1107-859 . For simplicity, pif1107-859 is labelled as pif1107. Microdissection of diploid strains carrying heterozygous mutations for the indicated wild-type and mutant alleles of PIF1 and (a) DIA2, (b) DNA2 or (c) TOP3. Plates were incubated at 30 ºC for 2 days (a) or 4 days (b) and (c). (d) Ten-fold dilutions of strains with the indicated genotypes were plated in YPD with different HU concentrations, incubated for 2 days at 30 ºC and photographed. 137 the increased level of this isoform impinges on the separation of function of pif1-m2 alleles. In contrast, the hypersensitivity of pif1107-859 rad3-102 strains to HU is difficult to explain with our current model, in which the nuclear levels of Pif1107-859 in pif1-m2 should derive in mild to complete suppression of pif1∆ phenotype. The failure of our model to explain this result led us to explore different possibilities for this particular interaction, as described in section 2.7. Taken together, our data indicate that in pif1-m2 mutants there is an increase in the translation of the partially functional Pif1107-859 isoform, which is able to enter the nucleus and fulfil some of the roles of nuclear Pif1. 2.5 Identification of the nuclear localization signal in Pif1 Once we have shown that alternative nuclear Pif1 isoforms may lead to the misinterpretation of genetic results obtained with the classical pif1-m2 allele, we reasoned that a better separationof-function allele could be attained by blocking the nuclear import of all isoforms. Since the upper limit of protein diffusion through the nuclear pore complex is around 60 kDa (Wang and Brattain, 2007), the size of Pif1 suggests that it should contain an NLS that has not been identified so far. To begin with, we employed an in-silico analysis to search for potential NLSs in the Pif1 sequence. In addition, we also performed two extra in silico analyses, PSIPRED (Buchan and Jones, 2019; McGuffin et al., 2000) (data not shown) and DISOPRED (Ward et al., 2004) to obtain a structural profile of Pif1 (Figure 53a). The intrinsic disordered profile obtained with DISOPRED predicts three unstructured regions, including both the NTD and the CTD, as well as a sharp peak centered around position 600, in thePif1 helicase domain (Figure 53a). Given that the NLSs mediates nuclear import through karyopherins, they tend to locate in the surface of the protein at disordered regions. Therefore, we combined the information of DISOPRED and NLS mappers to filter different putative NLSs, selecting four for further analysis (Figure 53b): two located at the CTD and two around the most disordered region inside the helicase domain (Figure 53c). Next, we deleted in our PGAL-PIF1-eGFP vectors regions encompassing those putative NLSs, employing the structural information of PSIPRED to minimally disrupt the secondary structure of each region, thus reducing the probabilities to generate complete loss-of-function mutants (Figure 53d). As a first approach to test their functionality, we took advantage again of the lethality of PGAL1PIF1 overexpression, which is mostly linked to aberrant increases in the nuclear isoform. Therefore, blocking Pif1 nuclear import should preserve cell viability in this assay. We observed that overexpression of pif1∆814-834 is as deleterious as that of PIF1 and pif1-m1, whereas the absence of the 623-646 region only slightly alleviates the lethality of Pif1 overexpression. Interestingly, two different alleles (pif1∆650-663 and pif1∆775-803) resulted in a marked increase in survival compared to PIF1, pif1-m1 and even pif1-m2 (Figure 54a). These differences in cell viability are not likely due to gross variations in the expression of each Tomás Lama Díaz 144 wild-type protein. In both cases, Pif1nls displayed similar reaction kinetics to Pif1, indicating that the four mutations in the NLS do not interfere with its enzymatic activity. We then estimated the impact of PIF1 NLS disruption (Figure 59a) on its genetic interactions with top3∆, dia2∆, and dna2∆ mutants by tetrad dissection analyses. Our findings indicate that the disrupted nuclear import of Pif1nls leads to synthetic lethality with TOP3 deletion, which mimics the phenotype of both pif1∆ (Figure 59b) and pif1-m2 (Figure 43). Similar to the results obtained with pif1-m2 (Figure 42), pif1nls does not display such an overt synthetic sickness with dia2∆ as pif1∆ (Figure 59c) and produces only a modest suppression of dna2∆ lethality (Figure 59d), implying residual nuclear Pif1 activity like in pif1-m2 strains (Figure 43). These results demonstrate that even the disruption of the NLS in an otherwise wild-type allele diminished nuclear Pif1 only to levels functionally comparable to those of pif1-m2, rather than behaving like the full PIF1 deletion. Figure 59. Disruption of Pif1 NLS does not completely abolish nuclear function. (a) Schematic representation of pif1nls. (b) Microdissection of diploid strains carrying heterozygous mutations for the indicated wild-type and mutant alleles of PIF1 and (b) TOP3, (c) DIA2 and (d) DNA2. Plates were incubated at 30 ºC for 2 days (c) or 4 days (b) and (d). 145 Given such unexpected results with pif1nls, we set out to further minimize this residual nuclear Pif1 activity by combining the mutations in both pif1nls (impaired nuclear import) and pif1-m2 (reduced nPif1) into a new allele that we have termed pif1mit (Figure 60a). In this case, tetrad dissection analyses demonstrated that when combined with dia2∆, pif1mit phenocopies pif1∆ Figure 60. pif1mit recapitulates the nuclear phenotypes of pif1∆ . (a) Schematic representation of the pif1m2 and pif1mit alleles. (b) Tetrad microdissection analyses of diploid strains carrying heterozygous mutations for the indicated wild-type and mutant alleles of DIA2 and PIF1. Images were taken after 2 days incubation at 30 ºC. (c) As in (b), but for the indicated wild-type and mutant alleles of DNA2 and PIF1. Images were taken after 4 days incubation at 30 ºC. (d) Telomeric southern blot of different PIF1 alleles to assess the telomeric length. A probe targeting CDC15 probe was employed as a loading control. (e) Ten-fold serial dilutions of strains with the indicated genotypes were plated in YPD containing HU and photographed after 2 days at 30 ºC. Tomás Lama Díaz 146 synthetic sickness more closely than pif1-m2 (Figure 60b). Likewise, the suppression of dna2∆ lethality by pif1mit is similar to that of pif1∆, and much more robust than for pif1-m2 mutants (Figure 60c). Moreover, pif1mit strains also resemble pif1∆ in telomeric lengthening, exhibiting a slight increase with respect to pif1-m2 mutants (Figure 60d). These results indicate that pif1mit represents an improved separation-of-function allele that translates almost exclusively into mitochondrial Pif1, with almost no functional traces of nuclear Pif1. However, we were surprised that pif1mit was still unable to recapitulate the increased HU hypersensitivity conferred by pif1∆ to rad3-102 mutants (Figure 60e), an observation that will be further explored in the next section. 2.7 Residual mitochondrial activity is present in pif1-m1 mutants During the course of our studies, we noticed an anomalous behaviour of the pif1-m1 strains with respect to their expected loss of mitochondrial function. While haploid pif1∆ strains freshly generated by tetrad dissection of PIF1/pif1∆ diploids acquire an unmistakable petite phenotype by the time the colony is visible, pif1-m1 colonies arising from PIF1/pif1-m1 diploids do not. To formally validate this observation, we generated diploid PIF1/pif1∆ strains with one additional copy of the pif1-m1 allele under the control of its endogenous promoter, integrated at the ura3-52 locus (Figure 61a), and analysed their meiotic progeny by tetrad dissection. Confirming our previous observations, pif1∆ colonies displayed a clear petite phenotype as judged by their lack of pigmentation in YPD plates and inability to grow in a medium without a fermentable carbon source (Figure 61b,c). Contrarily, when pif1∆ co-segregated with the ectopic pif1-m1 allele, normal pigmentation and the ability to grow on YPEGly were restored. Figure 61. pif1-m1 retains functional mitochondria. (a) Schematic representation of the N-terminal MTS region in pif1-m1 integrated at the ura3-52 locus. (b) Tetrad microdissection of a diploid strain carrying heterozygous deletion of PIF1 and an ectopic pif1-m1 allele integrated at ura3-52. YPD plates were not supplemented with adenine to allow for colour discrimination of petite colonies. (c) Strains harbouring the indicated PIF1 alleles were streaked on rich medium with a fermentable (YPD, 2% glucose) or non-fermentable (YPEGly, 3% ethanol, 3% glycerol) carbon source. Plates were incubated at 30 ºC for 2 days and photographed. 147 Interestingly, colonies where pif1∆ and pif1-m1 co-segregate exhibit a rough border, suggesting that mitochondrial function could still be partially compromised in pif1-m1 cells (Figure 61b). These results suggested that the pif1-m1 allele may also be subject to some sort of genetic bypass. While this situation could seem reminiscent of the ATI by ribosomal leaky scanning observed for pif1-m2, there are no additional in-frame start codons in the vicinity of AUG1 that could initiate translation of isoforms with a functional MTS. However, there is an additional mechanism of ATI by leaky scanning involving the use of “near-cognate” triplets as start codons. In this sense, it is well established that ribosomes can initiate translation at near-cognate start codons, although these triplets function at frequencies of <1 to 10% compared to canonical AUGs, according to in vivo and in vitro studies (Clements et al., 1988; Ingolia et al., 2009; Kolitz et al., 2009; Peabody, 1989). To maximize the likelihood of translation initiation, nearcognates require to be embedded within a robust Kozak context (Chen et al., 2008; Kearse and Wilusz, 2017). In S. cerevisiae, mitochondrial isoforms of the glycyland alanyl-tRNA synthetases (Grs1 and Ala1, respectively) are generated upon translation iniation from nonAUG codons (Chang and Wang, 2004; Tang et al., 2004), illustrating the ability of this mechanism to produce mitochondrial isoforms in dual-targeted proteins. In the case of Pif1, such mode of ATI had not been previously considered when ATI mechanisms were described, since the existing model for Pif1 translation postulated that two consecutive AUG codons were the only responsible for ATI events in Pif1 mRNA. Consistent with this potential role of near cognates in Pif1 ATI, we identified six in-frame near-cognate codons in the vicinity of ATG1 that could give rise to Pif1 isoforms with potentially functional MTSs (Figure 62a). One is located upstream - ATC-5 (Ile) -, while the remaining five are downstream ATG1: AAG3 (Lys), ATA5 (Ile), TTG9 (Leu), ATT12 (Ile), ATA13 (Ile). According to the scanning model, the mutation of ATG1 in pif1-m1 would significantly increase the fraction of PICs reaching the downstream non-AUG codons. Indeed, computational analysis of the putative MTSs generated from some of this near-cognate codons predicts a high (ATA5) or very high (ATC-5, AAG3, ATA5) probability of MTS functionality, which coherently decreases with their distance from ATG1 (Figure 62b). Therefore, to challenge our model of residual mitochondrial activity in pif1-m1 mutants due to ATI from near-cognate start codons, we mutated some of these codons in a pif1-m1 construct, being as conservative as possible with respect to amino acid changes in those downstream ATG1. In particular, we replaced all the downstream near-cognates coding for leucine or isoleucine to TTA leucine codons, since all codons for isoleucine are near cognates (Figure 62c). We did not mutate AAG3 , since near cognates with purine substitutions at the second position cannot initiate translation (Clements et al., 1988; Kolitz et al., 2009) Next, we ectopically integrated the new alleles into heterozygous PIF1/pif1∆ strains to evaluate their respective mitochondrial phenotypes in pif1∆ backgrounds. Crucially, we carefully selected clones with expression levels equal to or higher than our pif1-m1 strain (Figure 62d), Tomás Lama Díaz 148 as our hypothesis suggests that minute mPif1 amounts, undetectable by western blot, could still support mitochondrial function. First, we employed pif1-m1I(-5)A allele to assess the influence of the upstream near-cognate codon for mitochondrial function (Figure 63a). When co-segregating with pif1∆, mutation of ATC-5 to GCA-5 in pif1-m1 had only a minor detrimental effect on mitochondrial proficiency (Figure 63b,c). While we did notice an increase in the rugged morphology of yeast colonies (Figure 63b), these mutants still sustained an apparently robust respiratory metabolism (Figure 63c). Given that the number of scanning ribosomes along the upstream near-cognates start codons should not be affected by the ATG1 mutation, this finding suggests that this particular ATC-5 triplet should contribute to produce minor fractions of mPif1 even in WT cells. Next, we evaluated the effect of the single TTG9 to TTA9 mutation (pif1-m1L9L; Figure 63d) , since among all the downstream near-cognate codons, this is the only one embedded in an optimal Kozak sequence (AACAUUGA). Consistent with our prediction, expression of this mutant as the only source of cellular Pif1 derived in the distinctive slow growth and white Figure 62. In-frame near-cognate start codons surrounding AUG1 could produce mitochondrial-targeted isoforms. ( a) ATG-like near-cognate codons upstream and downstream ATG1 in PIF1. (b) Probability of functional MTS in extended or truncated peptides arising from ATI at non-AUG codons. (c) Mutations introduced in nearcognate ATGs in various pif1-m1 alleles. (d) Western blot analysis of Pif1-flag in protein extracts from strains ectopically expressing the indicated pif1-m1 variants. Unt, untagged. Ponceau staining of the PVDF membrane was employed as a loading control. 149 colour of petite colonies (Figure 63e). Moreover, the ability of these strains to employ nonfermentable carbon sources was greatly reduced, although they exhibited a slightly superior growth compared to pif1∆ strains (Figure 63f). Figure 63. Translation initiation from near-cognates AUGs can sustain mitochondrial function in the absence of AUG1. (a) Schematic representation of the N-terminal part of the pif1-m1I(-5)A allele. Red colour designates the near-cognate ATG mutated in this allele. Blue colour indicates the near-cognate ATGs downstream ATG1. (b) Tetrad microdissection of a diploid strain carrying a heterozygous deletion of PIF1 and an ectopic pif1-m1I(-5)A allele integrated at the ura3-52 locus. (c) Strains harbouring the indicated PIF1 alleles were streaked in rich medium with a fermentable (YPD, 2% glucose) or non-fermentable (YPEGly, 3% glycerol + 3% ethanol) carbon source. Plates were incubated at 30 ºC for 2 days and photographed. (d) As in (a), but indicating the mutation of the pif1-m1L9L allele. (e) As in (b), but the diploid strain harbours an ectopic copy of pif1-m1L9L. (f) As in (c), but with pif1-m1L9L. Tomás Lama Díaz 150 Next, we attempted to fully recapitulate the pif1∆ mitochondrial phenotype through the mutation of all the downstream near-cognate codons that could still yield a functional MTS (pif1-m14L; Figure 62b,c). Importantly, we first ruled out a potential effect of some of these non-synonymous changes on the functionality of MTS itself by introducing the same set of mutations in an otherwise wild-type PIF1 allele (pif14L) and testing its mitochondrial functions (Figure 64a). pif14L strains exhibit the typical pink hue and morphology of wild-type cells after Figure 64. Mutations of four near-cognates in pif1-m1 phenocopies pif1∆ mitochondrial defects. (a) Schematic representation of the N-terminal part of pif14L allele. Red colour designates the near-cognate ATGs mutated in this allele. Blue colour indicates the near-cognate codon still present upstream ATG1. (b) Tetrad microdissection of a diploid strain carrying heterozygous deletion of PIF1 and an ectopic pif14L allele integrated at the ura3-52 locus. (c) Strains harbouring the indicated PIF1 alleles were streaked in rich medium with a fermentable (YPD, 2% glucose) or non-fermentable (YPEGly, 3% glycerol + 3% ethanol) carbon source. Plates were incubated at 30 ºC for 2 days and photographed. (d) As in (a), but indicating the mutations of the pif1-m14L allele. (e) As in (b), but the diploid strain harbours an ectopic copy of pif1-m14L. (f) As in (c), but with pif1-m14L. . 151 tetrad dissection analyses (Figure 64b), as well as a similar capacity to grow on YPEGly (Figure 64c), hence demonstrating that these mutations do not interfere with the mitochondrial import of mPif1. Importantly the combination of the pif14L and m1 mutations into the pif1-m14L allele (Figure 64d) not only results in typical petite colonies after tetrad dissection experiments (Figure 64e), but also completely abrogates the residual growth on non-fermentable carbon sources previously described for pif1-m1L9L (Figure 64f). This confirms the loss of Pif1 mitochondrial function in pif1-m14L strains. Collectively, these results demonstrate that translation initiation from near-cognate start codons in the vicinity of AUG1 can give rise to phenotypically relevant amounts of mitochondrial Pif1, suppressing the mitochondrial defects predicted for pif1-m1. Finally, we took advantage of these improved mitochondrial-deficient PIF1 alleles, pif1-m1L9L and pif1-m14L, to revisit the genetic interaction between pif1∆ and rad3-102, which could not be recapitulated by the nuclear-deficient pif1mit allele. Therefore, we decided to test whether the hypersensitivity of pif1∆ rad3-102 mutants to HU derives from the loss of mitochondrial Pif1. Indeed, both pif1-m14L or pif1-m1L9L expression in rad3-102 mutants resulted in similar HU sensitivity to the pif1∆ rad3-102 strain (Figure 65), demonstrating a counterintuitive mitochondrial connection between Pif1 and Rad3 in their response to genotoxic agents. These results underscore the advantages of the improved PIF1 separation-of-alleles generated as part of this PhD Thesis to potentially disambiguate previously misascribed interactions between DNA repair/replication genes and the nuclear or mitochondrial roles of the pleiotropic Pif1 helicase. Figure 65. Improved, mitochondrial-deficient pif1-m1L9L and pif1-m14L alleles phenocopy the HU hypersensitivity of pif1∆ in rad3-102 mutants. Ten-fold serial dilutions of strains with the indicated genotypes were plated on YPD containing different concentrations of hydroxyurea (HU) and imaged after 2 days of incubation at 30 ºC. Tomás Lama Díaz 152 3. DISCUSSION This PhD Thesis constitutes the first comprehensive characterization of the translational mechanism controlling start codon selection in PIF1 mRNA. Soon after Pif1 identification in yeast, it was shown that its dual targeting to mitochondria and nucleus is dependent on ATI of its mRNA from AUG1 and AUG40, resulting in two different isoforms: mPif1 or nPif1. However, the underlying leaky scanning mechanism that dictates the choice between the two start codons involved remained unproven. This refinement in the model for Pif1 ATI led us to the discovery of a new nuclear Pif1 isoform translated from AUG107 that explains the incomplete loss of Pif1 nuclear function of pif1-m2 strains. To our knowledge, we have also provided the first evidence of pif1-m1 defective behaviour as a mitochondrial-null separationof-function allele by revealing its inability to recapitulate some of the mitochondrial phenotypes of pif1∆. In contrast to pif1-m2, the residual mitochondrial activity of pif1-m1 derives from multiple ATI events starting at non-AUG codons. Importantly, this deeper understanding of the intricacies of the leaky scanning model for PIF1 mRNA, together with our identification of Pif1 NLS, allowed us to develop improved versions of the classical pif1-m1 and pif1-m2 separationof-function alleles, which we have successfully employed to disambiguate pif1∆ genetic interactions with YEN1ON and rad3-102. Still, we consider that some of our observations as well as remaining open questions merit further discussion in the following sections. 3.1 Variable phenotypic impact of the residual nuclear Pif1 activity in pif1-m2 mutants Our experiments with pif1-m2 alleles indicate that whatever the proportion of nuclear Pif1 in these mutants, it produces a considerably different phenotypic outcome depending on the biological context (Figure 42 and Figure 43). In some genetic set-ups, pif1-m2 behaves like a genuine nuclear-null allele, as exemplified by its recapitulation of the synthetic lethality of pif1∆ top3∆ strains (Figure 43c). In other cases, pif1-m2 resembles a hypomorphic allele, exhibiting intermediate phenotypes due to the reduced levels nPif1, such as its incomplete suppression of dna2∆ lethality or its intermediate telomeric lengthening compared to PIF1 and pif1∆ strains (Figure 42b and Figure 43a). Lastly, as described in Chapter 1 for YEN1ON (Figure 38b) and in Chapter 2 for dia2∆ (Figure 42c), pif1-m2 can completely mask genetic interactions previously observed with pif1∆. Such variations on the impact of pif1-m2 between the different genetic set-ups tested indicate that a reduced concentration of nuclear Pif1 impairs to a different extent the multiple functions of this helicase in the nucleus. Thus, Pif1 nuclear levels in pif1-m2 would be sufficient to fulfil some of the nPif1 roles, but not enough for others, thus distinctly impinging on the genetic interactions of pif1-m2 with other genes. Consistent with this hypothesis, a constitutively overexpressed PADH1-pif1-m2 allele restores both the viability of top3∆ and the lethality of dna2∆ mutants, mimicking the phenotypes of strains harbouring PIF1 or pif1-m1 alleles (Figure 43b,d). 153 3.2 Pervasive ribosomal leaky scanning of PIF1 mRNA facilitates ATI from AUG1, AUG40 and AUG107 In this work, we have determined that ribosomal leaky scanning is the specific ATI mechanism responsible for the generation of mPif1 an nPif1. Under this model, there is a high probability for the scanning of the 43S PIC to skip the first AUG codon in PIF1 mRNA due to the suboptimal Kozak sequence around it, hence enabling translation from downstream AUG codons. Accordingly, two lines of evidence support that the suboptimal Kozak context surrounding AUG1, responsible for the translation of mPif1, allows a significant fraction of the scanning ribosomes to bypass AUG1 and reach AUG40, resulting in the translation of nPif1. First, various western blots in this PhD Thesis (Figure 44) and other published works point at a roughly similar abundance of mPif1 and nPif1 (Crider et al., 2012; Vega et al., 2007). If we assume similar translational efficiencies for ribosomes engaging at AUG1 or AUG40, as well as similar protein turnover rates for mPif and nPif1, this alone would imply that fewer than 50% of total ribosomes initiate translation at AUG1. Second, the optimization of the AUG1 Kozak context in PIF1 strains could be leading to a slight increase in mPif1, as judged by a subtle change in the electrophoretic migration of the tight nPif1/mPif1 doublet (Figure 49b, compare second and third lanes). Moreover, the Kozak motif surrounding AUG40 is also permissive enough to allow a minor fraction of scanning ribosomes to read through and traverse to the next in-frame start codon (AUG107), producing detectable amounts of the shorter Pif1107-859 isoform even in wild-type cells (Figure 44, Figure 45, Figure 46, Figure 48 and Figure 49). Again, in agreement with the leaky scanning model, mutations in the preceding start codon, as those in the pif1-m2 allele, enable a higher proportion of ribosomes reaching AUG107 and increase the abundance of Pif1107-859 (Figure 44). Conversely, optimization of the Kozak context surrounding AUG40 results in increased engagement of ribosomes at this start codon, rendering this isoform undetectable (Figure 49). In fact, the commonality of suboptimal Kozak sequences at PIF1 AUG codons implies that the sole mutation of AUG107 is not sufficient to abrogate the generation of a fast-migrating isoform. Instead, the complete disappearance of these smaller isoforms requires the concomitant mutation of the clustered AUG107, AUG113 and AUG127 codons, further substantiating the intrinsic tendency of PIF1 mRNA to undergo ATI by leaky scanning (Figure 48). 3.3 The intrinsic ATI mechanism operating on PIF1 mRNA constrains the efficiency of pif1-m1 and pif1-m2 as separation-of-function alleles. Under the light of our results, we argue that the incomplete separation of function observed for the classical pif1-m1 and pif1-m2 alleles is a consequence, at least partially, of the ATI mechanism by ribosomal leaky scanning mediating translational start from AUGs and nearcognates AUGs on PIF1 mRNA. In the case of pif1-m1 strains, a priori lacking mPif1 due to AUG1 elimination, the characteristic petite phenotype of pif1∆ mutants was not reproduced, as cells maintained both their ability to grow on non-fermentable carbon sources and the typical red pigmentation of W303 strains in YPD, both proxies for respiration-proficient mitochondria (Figure 61). Given the absence of any in-frame AUG codon that would allow the translation of MTS-containing Pif1 isoforms (Figure 62), we demonstrated that near-cognate start codons in Tomás Lama Díaz 256 Sancar, A., Lindsey-Boltz, L.A., Ünsal-Kaçmaz, K., Linn, S., 2004. Molecular Mechanisms of Mammalian DNA Repair and the DNA Damage Checkpoints. Annu. Rev. Biochem. 73, 39–85. https://doi.org/10.1146/annurev.biochem.73.011303.073723 Sanchez, Y., Bachant, J., Wang, H., Hu, F., Liu, D., Tetzlaff, M., Elledge, S.J., 1999. Control of the DNA Damage Checkpoint by Chk1 and Rad53 Protein Kinases Through Distinct Mechanisms. Science 286, 1166–1171. https://doi.org/10.1126/science.286.5442.1166 Sandell, L.L., Zakian, V.A., 1993. Loss of a yeast telomere: Arrest, recovery, and chromosome loss. Cell 75, 729–739. https://doi.org/10.1016/0092-8674(93)90493-A Sanders, C.M., 2010. Human Pif1 helicase is a G-quadruplex DNA-binding protein with Gquadruplex. Biochem. J. 430, 119–128. https://doi.org/10.1042/BJ20100612 Saponaro, M., Callahan, D., Zheng, X., Krejci, L., Haber, J.E., Klein, H.L., Liberi, G., 2010. Cdk1 targets Srs2 to complete synthesis-dependent strand annealing and to promote recombinational repair. PLoS Genet. 6, e1000858. https://doi.org/10.1371/journal.pgen.1000858 Sarkies, P., Murat, P., Phillips, L.G., Patel, K.J., Balasubramanian, S., Sale, J.E., 2012. FANCJ coordinates two pathways that maintain epigenetic stability at G-quadruplex DNA. Nucleic Acids Res. 40, 1485–1498. https://doi.org/10.1093/nar/gkr868 Sarkies, P., Reams, C., Simpson, L.J., Sale, J.E., 2010. Epigenetic Instability due to Defective Replication of Structured DNA. Mol. Cell 40, 703–713. https://doi.org/10.1016/j.molcel.2010.11.009 Sartori, A.A., Lukas, C., Coates, J., Mistrik, M., Fu, S., Bartek, J., Baer, R., Lukas, J., Jackson, S.P., 2007. Human CtIP promotes DNA end resection. Nature 450, 509–514. https://doi.org/10.1038/nature06337 Sass, E., Blachinsky, E., Karniely, S., Pines, O., 2001. Mitochondrial and Cytosolic Isoforms of Yeast Fumarase Are Derivatives of a Single Translation Product and Have Identical Amino Termini *. J. Biol. Chem. 276, 46111–46117. https://doi.org/10.1074/jbc.M106061200 Sass, E., Karniely, S., Pines, O., 2003. Folding of Fumarase during Mitochondrial Import Determines its Dual Targeting in Yeast *. J. Biol. Chem. 278, 45109–45116. https://doi.org/10.1074/jbc.M302344200 Sausen, C.W., Rogers, C.M., Bochman, M.L., 2019. Thin-Layer Chromatography and RealTime Coupled Assays to Measure ATP Hydrolysis, in: Balakrishnan, L., Stewart, J.A. (Eds.), DNA Repair, Methods in Molecular Biology. Springer New York, New York, NY, pp. 245–253. https://doi.org/10.1007/978-1-4939-9500-4_15 Saxena, S., Zou, L., 2022. Hallmarks of DNA replication stress. Mol. Cell 82, 2298–2314. https://doi.org/10.1016/j.molcel.2022.05.004 Schiavone, D., Guilbaud, G., Murat, P., Papadopoulou, C., Sarkies, P., Prioleau, M.-N., Balasubramanian, S., Sale, J.E., 2014. Determinants of G quadruplex-induced epigenetic instability in REV1-deficient cells. EMBO J. 33, 2507–2520. https://doi.org/10.15252/embj.201488398 257 Schmidt, K.H., Derry, K.L., Kolodner, R.D., 2002. Saccharomyces cerevisiae RRM3, a 5′ to 3′ DNA Helicase, Physically Interacts with Proliferating Cell Nuclear Antigen *. J. Biol. Chem. 277, 45331–45337. https://doi.org/10.1074/jbc.M207263200 Schulz, V.P., Zakian, V.A., 1994a. The saccharomyces PIF1 DNA helicase inhibits telomere elongation and de novo telomere formation. Cell 76, 145–155. https://doi.org/10.1016/0092-8674(94)90179-1 Schulz, V.P., Zakian, V.A., 1994b. The saccharomyces PIF1 DNA helicase inhibits telomere elongation and de novo telomere formation. Cell 76, 145–155. https://doi.org/10.1016/0092-8674(94)90179-1 Schwab, R.A., Nieminuszczy, J., Shin-ya, K., Niedzwiedz, W., 2013. FANCJ couples replication past natural fork barriers with maintenance of chromatin structure. J. Cell Biol. 201, 33–48. https://doi.org/10.1083/jcb.201208009 Schwartz, E.K., Heyer, W.-D., 2011. Processing of joint molecule intermediates by structureselective endonucleases during homologous recombination in eukaryotes. Chromosoma 120, 109–127. https://doi.org/10.1007/s00412-010-0304-7 Schwartz, E.K., Wright, W.D., Ehmsen, K.T., Evans, J.E., Stahlberg, H., Heyer, W.-D., 2012. Mus81-Mms4 functions as a single heterodimer to cleave nicked intermediates in recombinational DNA repair. Mol. Cell. Biol. 32, 3065–3080. https://doi.org/10.1128/MCB.00547-12 Sebesta, M., Burkovics, P., Haracska, L., Krejci, L., 2011. Reconstitution of DNA repair synthesis in vitro and the role of polymerase and helicase activities. DNA Repair 10, 567–576. https://doi.org/10.1016/j.dnarep.2011.03.003 Seong, C., Colavito, S., Kwon, Y., Sung, P., Krejci, L., 2009. Regulation of Rad51 recombinase presynaptic filament assembly via interactions with the Rad52 mediator and the Srs2 anti-recombinase. J. Biol. Chem. 284, 24363–24371. https://doi.org/10.1074/jbc.M109.032953 Sfeir, A., Kosiyatrakul, S.T., Hockemeyer, D., MacRae, S.L., Karlseder, J., Schildkraut, C.L., de Lange, T., 2009. Mammalian Telomeres Resemble Fragile Sites and Require TRF1 for Efficient Replication. Cell 138, 90–103. https://doi.org/10.1016/j.cell.2009.06.021 Sfeir, A., Symington, L.S., 2015. Microhomology-Mediated End Joining: A Back-up Survival Mechanism or Dedicated Pathway? Trends Biochem. Sci. 40, 701–714. https://doi.org/10.1016/j.tibs.2015.08.006 Sharma, S., Otterlei, M., Sommers, J.A., Driscoll, H.C., Dianov, G.L., Kao, H.-I., Bambara, R.A., Brosh, R.M., 2004. WRN Helicase and FEN-1 Form a Complex upon Replication Arrest and Together Process Branchmigrating DNA Structures Associated with the Replication Fork. Mol. Biol. Cell 15, 734–750. https://doi.org/10.1091/mbc.e03-080567 Sherman, F., 2002. Getting started with yeast, in: Guthrie, C., Fink, G.R. (Eds.), Methods in Enzymology, Guide to Yeast Genetics and Molecular and Cell Biology - Part B. Academic Press, pp. 3–41. https://doi.org/10.1016/S0076-6879(02)50954-X Tomás Lama Díaz 258 Shi, I., Hallwyl, S.C.L., Seong, C., Mortensen, U., Rothstein, R., Sung, P., 2009. Role of the Rad52 Amino-terminal DNA Binding Activity in DNA Strand Capture in Homologous Recombination. J. Biol. Chem. 284, 33275–33284. https://doi.org/10.1074/jbc.M109.057752 Shibata, A., Moiani, D., Arvai, A.S., Perry, J., Harding, S.M., Genois, M.-M., Maity, R., van Rossum-Fikkert, S., Kertokalio, A., Romoli, F., Ismail, A., Ismalaj, E., Petricci, E., Neale, M.J., Bristow, R.G., Masson, J.-Y., Wyman, C., Jeggo, P.A., Tainer, J.A., 2014. DNA Double-Strand Break Repair Pathway Choice Is Directed by Distinct MRE11 Nuclease Activities. Mol. Cell 53, 7–18. https://doi.org/10.1016/j.molcel.2013.11.003 Shim, E.Y., Hong, S.J., Oum, J.-H., Yanez, Y., Zhang, Y., Lee, S.E., 2007. RSC Mobilizes Nucleosomes To Improve Accessibility of Repair Machinery to the Damaged Chromatin. Mol. Cell. Biol. 27, 1602–1613. https://doi.org/10.1128/MCB.01956-06 Shimada, K., Gasser, S.M., 2012. DNA replication: Pif1 pulls the plug on stalled replication forks. Curr. Biol. CB 22, R404-405. https://doi.org/10.1016/j.cub.2012.04.015 Shimura, T., Ochiai, Y., Noma, N., Oikawa, T., Sano, Y., Fukumoto, M., 2013. Cyclin D1 overexpression perturbs DNA replication and induces replication-associated DNA double-strand breaks in acquired radioresistant cells. Cell Cycle 12, 773–782. https://doi.org/10.4161/cc.23719 Shimura, T., Torres, M.J., Martin, M.M., Rao, V.A., Pommier, Y., Katsura, M., Miyagawa, K., Aladjem, M.I., 2008. Bloom’s Syndrome Helicase and Mus81 are Required to Induce Transient Double-strand DNA Breaks in Response to DNA Replication Stress. J. Mol. Biol. 375, 1152–1164. https://doi.org/10.1016/j.jmb.2007.11.006 Shiotani, B., Zou, L., 2009. Single-Stranded DNA Orchestrates an ATM-to-ATR Switch at DNA Breaks. Mol. Cell 33, 547–558. https://doi.org/10.1016/j.molcel.2009.01.024 Siede, W., Friedberg, A.S., Dianova, I., Friedberg, E.C., 1994. Characterization of G1 checkpoint control in the yeast Saccharomyces cerevisiae following exposure to DNAdamaging agents. Genetics 138, 271–281. https://doi.org/10.1093/genetics/138.2.271 Siede, W., Friedberg, A.S., Friedberg, E.C., 1993. RAD9-dependent G1 arrest defines a second checkpoint for damaged DNA in the cell cycle of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. 90, 7985–7989. https://doi.org/10.1073/pnas.90.17.7985 Singh, T.R., Ali, A.M., Busygina, V., Raynard, S., Fan, Q., Du, C., Andreassen, P.R., Sung, P., Meetei, A.R., 2008. BLAP18/RMI2, a novel OB-fold-containing protein, is an essential component of the Bloom helicase-double Holliday junction dissolvasome. Genes Dev. 22, 2856–2868. https://doi.org/10.1101/gad.1725108 Slusher, L.B., Gillman, E.C., Martin, N.C., Hopper, A.K., 1991. mRNA leader length and initiation codon context determine alternative AUG selection for the yeast gene MOD5. Proc. Natl. Acad. Sci. 88, 9789–9793. https://doi.org/10.1073/pnas.88.21.9789 Smirnov, M.N., Smirnov, V.N., Budowsky, E.I., Inge-Vechtomov, S.G., Serebrjakov, N.G., 1967. Red pigment of adenine-deficient yeast Saccharomyces cerevisiae. Biochem. 259 Biophys. Res. Commun. 27, 299–304. https://doi.org/10.1016/S0006-291X(67)800962 Smith, C.E., Lam, A.F., Symington, L.S., 2009. Aberrant Double-Strand Break Repair Resulting in Half Crossovers in Mutants Defective for Rad51 or the DNA Polymerase δ Complex. Mol. Cell. Biol. 29, 1432–1441. https://doi.org/10.1128/MCB.01469-08 Smolka, M.B., Albuquerque, C.P., Chen, S., Zhou, H., 2007. Proteome-wide identification of in vivo targets of DNA damage checkpoint kinases. Proc. Natl. Acad. Sci. 104, 10364– 10369. https://doi.org/10.1073/pnas.0701622104 Snow, B.E., Mateyak, M., Paderova, J., Wakeham, A., Iorio, C., Zakian, V., Squire, J., Harrington, L., 2007. Murine Pif1 interacts with telomerase and is dispensable for telomere function in vivo. Mol. Cell. Biol. 27, 1017–1026. https://doi.org/10.1128/MCB.01866-06 Sogo, J.M., Lopes, M., Foiani, M., 2002. Fork Reversal and ssDNA Accumulation at Stalled Replication Forks Owing to Checkpoint Defects. Science 297, 599–602. https://doi.org/10.1126/science.1074023 Solinger, J.A., Kiianitsa, K., Heyer, W.-D., 2002. Rad54, a Swi2/Snf2-like Recombinational Repair Protein, Disassembles Rad51:dsDNA Filaments. Mol. Cell 10, 1175–1188. https://doi.org/10.1016/S1097-2765(02)00743-8 Song, B., Sung, P., 2000. Functional Interactions among Yeast Rad51 Recombinase, Rad52 Mediator, and Replication Protein A in DNA Strand Exchange *. J. Biol. Chem. 275, 15895–15904. https://doi.org/10.1074/jbc.M910244199 Spingola, M., Grate, L., Haussler, D., Ares, M., 1999. Genome-wide bioinformatic and molecular analysis of introns in Saccharomyces cerevisiae. RNA 5, 221–234. https://doi.org/10.1017/S1355838299981682 Sriram, A., Bohlen, J., Teleman, A.A., 2018. Translation acrobatics: how cancer cells exploit alternate modes of translational initiation. EMBO Rep. 19, e45947. https://doi.org/10.15252/embr.201845947 Sriramachandran, A.M., Petrosino, G., Méndez-Lago, M., Schäfer, A.J., Batista-Nascimento, L.S., Zilio, N., Ulrich, H.D., 2020. Genome-wide Nucleotide-Resolution Mapping of DNA Replication Patterns, Single-Strand Breaks, and Lesions by GLOE-Seq. Mol. Cell 78, 975-985.e7. https://doi.org/10.1016/j.molcel.2020.03.027 Srivatsan, A., Bowen, N., Kolodner, R.D., 2014. Mispair-specific Recruitment of the Mlh1Pms1 Complex Identifies Repair Substrates of the Saccharomyces cerevisiae Msh2Msh3 Complex *. J. Biol. Chem. 289, 9352–9364. https://doi.org/10.1074/jbc.M114.552190 Stafa, A., Donnianni, R.A., Timashev, L.A., Lam, A.F., Symington, L.S., 2014. Template switching during break-induced replication is promoted by the Mph1 helicase in Saccharomyces cerevisiae. Genetics 196, 1017–1028. https://doi.org/10.1534/genetics.114.162297 Tomás Lama Díaz 260 Staresincic, L., Fagbemi, A.F., Enzlin, J.H., Gourdin, A.M., Wijgers, N., Dunand-Sauthier, I., Giglia-Mari, G., Clarkson, S.G., Vermeulen, W., Schärer, O.D., 2009. Coordination of dual incision and repair synthesis in human nucleotide excision repair. EMBO J. 28, 1111–1120. https://doi.org/10.1038/emboj.2009.49 Stein, I., Peleg, Y., Even-Ram, S., Pines, O., 1994. The single translation product of the FUM1 gene (fumarase) is processed in mitochondria before being distributed between the cytosol and mitochondria in Saccharomyces cerevisiae. Mol. Cell. Biol. 14, 4770–4778. https://doi.org/10.1128/mcb.14.7.4770-4778.1994 Steinacher, R., Osman, F., Dalgaard, J.Z., Lorenz, A., Whitby, M.C., 2012. The DNA helicase Pfh1 promotes fork merging at replication termination sites to ensure genome stability. Genes Dev 26, 594–602. https://doi.org/10.1101/gad.184663.111 Stelter, P., Ulrich, H.D., 2003. Control of spontaneous and damage-induced mutagenesis by SUMO and ubiquitin conjugation. Nature 425, 188–191. https://doi.org/10.1038/nature01965 Storici, F., Resnick, M.A., 2006. The Delitto Perfetto Approach to In Vivo Site‐Directed Mutagenesis and Chromosome Rearrangements with Synthetic Oligonucleotides in Yeast, in: Methods in Enzymology. Elsevier, pp. 329–345. https://doi.org/10.1016/S0076-6879(05)09019-1 Stundon, J.L., Zakian, V.A., 2015a. Identification of Saccharomyces cerevisiae Genes Whose Deletion Causes Synthetic Effects in Cells with Reduced Levels of the Nuclear Pif1 DNA Helicase. G3 Bethesda Md 5, 2913–2918. https://doi.org/10.1534/g3.115.021139 Stundon, J.L., Zakian, V.A., 2015b. Identification of Saccharomyces cerevisiae Genes Whose Deletion Causes Synthetic Effects in Cells with Reduced Levels of the Nuclear Pif1 DNA Helicase. G3 Bethesda Md 5, 2913–2918. https://doi.org/10.1534/g3.115.021139 Sturzenegger, A., Burdova, K., Kanagaraj, R., Levikova, M., Pinto, C., Cejka, P., Janscak, P., 2014. DNA2 cooperates with the WRN and BLM RecQ helicases to mediate long-range DNA end resection in human cells. J. Biol. Chem. 289, 27314–27326. https://doi.org/10.1074/jbc.M114.578823 Sugawara, N., Haber, J.E., 1992. Characterization of double-strand break-induced recombination: homology requirements and single-stranded DNA formation. Mol. Cell. Biol. 12, 563–575. https://doi.org/10.1128/mcb.12.2.563-575.1992 Sugiyama, T., Kantake, N., Wu, Y., Kowalczykowski, S.C., 2006. Rad52-mediated DNA annealing after Rad51-mediated DNA strand exchange promotes second ssDNA capture. EMBO J. 25, 5539–5548. https://doi.org/10.1038/sj.emboj.7601412 Sun, T., Yu, Y., Wu, X., Acevedo, A., Luo, J.-D., Wang, J., Schneider, W.M., Hurwitz, B., Rosenberg, B.R., Chung, H., Rice, C.M., 2021. Decoupling expression and editing preferences of ADAR1 p150 and p110 isoforms. Proc. Natl. Acad. Sci. 118, e2021757118. https://doi.org/10.1073/pnas.2021757118 Sun, W., Nandi, S., Osman, F., Ahn, J.S., Jakovleska, J., Lorenz, A., Whitby, M.C., 2008. The FANCM Ortholog Fml1 Promotes Recombination at Stalled Replication Forks and 261 Limits Crossing Over during DNA Double-Strand Break Repair. Mol. Cell 32, 118– 128. https://doi.org/10.1016/j.molcel.2008.08.024 Sung, P., 1997. Yeast Rad55 and Rad57 proteins form a heterodimer that functions with replication protein A to promote DNA strand exchange by Rad51 recombinase. Genes Dev. 11, 1111–1121. https://doi.org/10.1101/gad.11.9.1111 Sung, P., 1994. Catalysis of ATP-Dependent Homologous DNA Pairing and Strand Exchange by Yeast RAD51 Protein. Science 265, 1241–1243. https://doi.org/10.1126/science.8066464 Sung, P., Stratton, S.A., 1996. Yeast Rad51 Recombinase Mediates Polar DNA Strand Exchange in the Absence of ATP Hydrolysis *. J. Biol. Chem. 271, 27983–27986. https://doi.org/10.1074/jbc.271.45.27983 Surova, O., Zhivotovsky, B., 2013. Various modes of cell death induced by DNA damage. Oncogene 32, 3789–3797. https://doi.org/10.1038/onc.2012.556 Suzuki, Y., Holmes, J.B., Cerritelli, S.M., Sakhuja, K., Minczuk, M., Holt, I.J., Crouch, R.J., 2010. An Upstream Open Reading Frame and the Context of the Two AUG Codons Affect the Abundance of Mitochondrial and Nuclear RNase H1. Mol. Cell. Biol. 30, 5123–5134. https://doi.org/10.1128/MCB.00619-10 Svendsen, J.M., Harper, J.W., 2010. GEN1/Yen1 and the SLX4 complex: Solutions to the problem of Holliday junction resolution. Genes Dev. 24, 521–536. https://doi.org/10.1101/gad.1903510 Syed, S., Desler, C., Rasmussen, L.J., Schmidt, K.H., 2016. A Novel Rrm3 Function in Restricting DNA Replication via an Orc5-Binding Domain Is Genetically Separable from Rrm3 Function as an ATPase/Helicase in Facilitating Fork Progression. PLOS Genet. 12, e1006451. https://doi.org/10.1371/journal.pgen.1006451 Symington, L.S., Gautier, J., 2011. Double-Strand Break End Resection and Repair Pathway Choice. Annu. Rev. Genet. 45, 247–271. https://doi.org/10.1146/annurev-genet110410-132435 Szakal, B., Branzei, D., 2013. Premature Cdk1/Cdc5/Mus81 pathway activation induces aberrant replication and deleterious crossover. EMBO J. 32, 1155–1167. https://doi.org/10.1038/emboj.2013.67 Szostak, J.W., Orr-Weaver, T.L., Rothstein, R.J., Stahl, F.W., 1983. The double-strand-break repair model for recombination. Cell 33, 25–35. https://doi.org/10.1016/00928674(83)90331-8 Talhaoui, I., Bernal, M., Mullen, J.R., Dorison, H., Palancade, B., Brill, S.J., Mazon, G., 2018. Slx5-Slx8 ubiquitin ligase targets active pools of the Yen1 nuclease to limit crossover formation. Nat. Commun. 9, 5016. https://doi.org/10.1038/s41467-018-07364-x Tang, H.-L., Yeh, L.-S., Chen, N.-K., Ripmaster, T., Schimmel, P., Wang, C.-C., 2004. Translation of a Yeast Mitochondrial tRNA Synthetase Initiated at Redundant nonAUG Codons *. J. Biol. Chem. 279, 49656–49663. https://doi.org/10.1074/jbc.M408081200 Tomás Lama Díaz 262 Tay, Y.D., Wu, L., 2010. Overlapping roles for Yen1 and Mus81 in cellular Holliday junction processing. J. Biol. Chem. 285, 11427–11432. https://doi.org/10.1074/jbc.M110.108399 Taylor, S.D., Zhang, H., Eaton, J.S., Rodeheffer, M.S., Lebedeva, M.A., O’rourke, T.W., Siede, W., Shadel, G.S., 2005. The conserved Mec1/Rad53 nuclear checkpoint pathway regulates mitochondrial DNA copy number in Saccharomyces cerevisiae. Mol Biol Cell 16, 3010–8. https://doi.org/10.1091/mbc.e05-01-0053 Thangavel, S., Berti, M., Levikova, M., Pinto, C., Gomathinayagam, S., Vujanovic, M., Zellweger, R., Moore, H., Lee, E.H., Hendrickson, E.A., Cejka, P., Stewart, S., Lopes, M., Vindigni, A., 2015. DNA2 drives processing and restart of reversed replication forks in human cells. J. Cell Biol. 208, 545–562. https://doi.org/10.1083/jcb.201406100 Torres, J.Z., Bessler, J.B., Zakian, V.A., 2004. Local chromatin structure at the ribosomal DNA causes replication fork pausing and genome instability in the absence of the S. cerevisiae DNA helicase Rrm3p. Genes Dev. 18, 498–503. https://doi.org/10.1101/gad.1154704 Touriol, C., Bornes, S., Bonnal, S., Audigier, S., Prats, H., Prats, A.-C., Vagner, S., 2003. Generation of protein isoform diversity by alternative initiation of translation at nonAUG codons. Biol. Cell 95, 169–178. https://doi.org/10.1016/S0248-4900(03)00033-9 Tran, P.L.T., Pohl, T.J., Chen, C.-F., Chan, A., Pott, S., Zakian, V.A., 2017. PIF1 family DNA helicases suppress R-loop mediated genome instability at tRNA genes. Nat. Commun. 8, 15025. https://doi.org/10.1038/ncomms15025 Tran, P.T., Erdeniz, N., Dudley, S., Liskay, R.M., 2002. Characterization of nucleasedependent functions of Exo1p in Saccharomyces cerevisiae. DNA Repair 1, 895–912. https://doi.org/10.1016/S1568-7864(02)00114-3 Tsaneva, I.R., Müller, B., West, S.C., 1992. ATP-dependent branch migration of holliday junctions promoted by the RuvA and RuvB proteins of E. coli. Cell 69, 1171–1180. https://doi.org/10.1016/0092-8674(92)90638-S Tseng, H.-M., Tomkinson, A.E., 2004. Processing and Joining of DNA Ends Coordinated by Interactions among Dnl4/Lif1, Pol4, and FEN-1 *. J. Biol. Chem. 279, 47580–47588. https://doi.org/10.1074/jbc.M404492200 Tsukamoto, Y., Mitsuoka, C., Terasawa, M., Ogawa, H., Ogawa, T., 2005. Xrs2p Regulates Mre11p Translocation to the Nucleus and Plays a Role in Telomere Elongation and Meiotic Recombination. Mol. Biol. Cell 16, 597–608. https://doi.org/10.1091/mbc.e0409-0782 Tubbs, A., Nussenzweig, A., 2017. Endogenous DNA Damage as a Source of Genomic Instability in Cancer. Cell 168, 644–656. https://doi.org/10.1016/j.cell.2017.01.002 Ubersax, J.A., Woodbury, E.L., Quang, P.N., Paraz, M., Blethrow, J.D., Shah, K., Shokat, K.M., Morgan, D.O., 2003. Targets of the cyclin-dependent kinase Cdk1. Nature 425, 859–864. https://doi.org/10.1038/nature02062 Ulrich, H.D., Jentsch, S., 2000. Two RING finger proteins mediate cooperation between ubiquitin-conjugating enzymes in DNA repair 10. 263 Ulrich, H.D., Walden, H., 2010. Ubiquitin signalling in DNA replication and repair. Nat. Rev. Mol. Cell Biol. 11, 479–489. https://doi.org/10.1038/nrm2921 Umar, A., Buermeyer, A.B., Simon, J.A., Thomas, D.C., Clark, A.B., Liskay, R.M., Kunkel, T.A., 1996. Requirement for PCNA in DNA Mismatch Repair at a Step Preceding DNA Resynthesis. Cell 87, 65–73. https://doi.org/10.1016/S0092-8674(00)81323-9 van den Broek, W.J.A.A., 2002. Somatic expansion behaviour of the (CTG)n repeat in myotonic dystrophy knock-in mice is differentially affected by Msh3 and Msh6 mismatch-repair proteins. Hum. Mol. Genet. 11, 191–198. https://doi.org/10.1093/hmg/11.2.191 van Wietmarschen, N., Sridharan, S., Nathan, W.J., Tubbs, A., Chan, E.M., Callen, E., Wu, W., Belinky, F., Tripathi, V., Wong, N., Foster, K., Noorbakhsh, J., Garimella, K., CruzMigoni, A., Sommers, J.A., Huang, Y., Borah, A.A., Smith, J.T., Kalfon, J., Kesten, N., Fugger, K., Walker, R.L., Dolzhenko, E., Eberle, M.A., Hayward, B.E., Usdin, K., Freudenreich, C.H., Brosh, R.M., West, S.C., McHugh, P.J., Meltzer, P.S., Bass, A.J., Nussenzweig, A., 2020. Repeat expansions confer WRN dependence in microsatelliteunstable cancers. Nature 586, 292–298. https://doi.org/10.1038/s41586-020-2769-8 Vannier, J.-B., Pavicic-Kaltenbrunner, V., Petalcorin, M.I.R., Ding, H., Boulton, S.J., 2012. RTEL1 dismantles T loops and counteracts telomeric G4-DNA to maintain telomere integrity. Cell 149, 795–806. https://doi.org/10.1016/j.cell.2012.03.030 Vannier, J.-B., Sandhu, S., Petalcorin, M.IR., Wu, X., Nabi, Z., Ding, H., Boulton, S.J., 2013. RTEL1 Is a Replisome-Associated Helicase That Promotes Telomere and GenomeWide Replication. Science 342, 239–242. https://doi.org/10.1126/science.1241779 Vasianovich, Y., Harrington, L.A., Makovets, S., 2014. Break-induced replication requires DNA damage-induced phosphorylation of Pif1 and leads to telomere lengthening. PLoS Genet. 10, e1004679. https://doi.org/10.1371/journal.pgen.1004679 Veaute, X., Jeusset, J., Soustelle, C., Kowalczykowski, S.C., Le Cam, E., Fabre, F., 2003. The Srs2 helicase prevents recombination by disrupting Rad51 nucleoprotein filaments. Nature 423, 309–312. https://doi.org/10.1038/nature01585 Vega, L.R., Phillips, J.A., Thornton, B.R., Benanti, J.A., Onigbanjo, M.T., Toczyski, D.P., Zakian, V.A., 2007. Sensitivity of yeast strains with long G-tails to levels of telomerebound telomerase. PLoS Genet 3, e105. https://doi.org/10.1371/journal.pgen.0030105 Vernekar, D.V., Reginato, G., Adam, C., Ranjha, L., Dingli, F., Marsolier, M.-C., Loew, D., Guérois, R., Llorente, B., Cejka, P., Borde, V., 2021. The Pif1 helicase is actively inhibited during meiotic recombination which restrains gene conversion tract length. Nucleic Acids Res 49, 4522–4533. https://doi.org/10.1093/nar/gkab232 Vilenchik, M.M., Knudson, A.G., 2003. Endogenous DNA double-strand breaks: Production, fidelity of repair, and induction of cancer. Proc. Natl. Acad. Sci. 100, 12871–12876. https://doi.org/10.1073/pnas.2135498100 Tomás Lama Díaz 264 Villarreal, D.D., Lee, K., Deem, A., Shim, E.Y., Malkova, A., Lee, S.E., 2012. Microhomology Directs Diverse DNA Break Repair Pathways and Chromosomal Translocations. PLOS Genet. 8, e1003026. https://doi.org/10.1371/journal.pgen.1003026 Wach, A., Brachat, A., Pöhlmann, R., Philippsen, P., 1994. New heterologous modules for classical or PCR-based gene disruptions in Saccharomyces cerevisiae. Yeast 10, 1793– 1808. https://doi.org/10.1002/yea.320101310 Waga, S., Stillman, B., 1994. Anatomy of a DNA replication fork revealed by reconstitution of SV40 DNA replication in vitro. Nature 369, 207–212. https://doi.org/10.1038/369207a0 Wagner, M., Price, G., Rothstein, R., 2006. The absence of Top3 reveals an interaction between the Sgs1 and Pif1 DNA helicases in Saccharomyces cerevisiae. Genetics 174, 555–573. https://doi.org/10.1534/genetics.104.036905 Waizenegger, A., Urulangodi, M., Lehmann, C.P., Reyes, T.A.C., Saugar, I., Tercero, J.A., Szakal, B., Branzei, D., 2020. Mus81-Mms4 endonuclease is an Esc2-STUbL-Cullin8 mitotic substrate impacting on genome integrity. Nat. Commun. 11, 5746. https://doi.org/10.1038/s41467-020-19503-4 Wallgren, M., Mohammad, J.B., Yan, K.-P., Pourbozorgi-Langroudi, P., Ebrahimi, M., Sabouri, N., 2016. G-rich telomeric and ribosomal DNA sequences from the fission yeast genome form stable G-quadruplex DNA structures in vitro and are unwound by the Pfh1 DNA helicase. Nucleic Acids Res 44, 6213–31. https://doi.org/10.1093/nar/gkw349 Wan, B., Yin, J., Horvath, K., Sarkar, J., Chen, Y., Wu, J., Wan, K., Lu, J., Gu, P., Yu, E.Y., Lue, N.F., Chang, S., Liu, Y., Lei, M., 2013. SLX4 assembles a telomere maintenance toolkit by bridging multiple endonucleases with telomeres. Cell Rep. 4, 861–869. https://doi.org/10.1016/j.celrep.2013.08.017 Wang, G., Vasquez, K.M., 2022. Dynamic alternative DNA structures in biology and disease. Nat. Rev. Genet. https://doi.org/10.1038/s41576-022-00539-9 Wang, G., Vasquez, K.M., 2017. Effects of Replication and Transcription on DNA StructureRelated Genetic Instability. Genes 8, 17. https://doi.org/10.3390/genes8010017 Wang, H., Rosidi, B., Perrault, R., Wang, M., Zhang, L., Windhofer, F., Iliakis, G., 2005. DNA Ligase III as a Candidate Component of Backup Pathways of Nonhomologous End Joining. Cancer Res. 65, 4020–4030. https://doi.org/10.1158/0008-5472.CAN-04-3055 Wang, L., Wang, Q.-M., Wang, Y.-R., Xi, X.-G., Hou, X.-M., 2018. DNA-unwinding activity of Saccharomyces cerevisiae Pif1 is modulated by thermal stability, folding conformation, and loop lengths of G-quadruplex DNA. J. Biol. Chem. 293, 18504– 18513. https://doi.org/10.1074/jbc.RA118.005071 Wang, R., Brattain, M.G., 2007. The maximal size of protein to diffuse through the nuclear pore is larger than 60kDa. FEBS Lett. 581, 3164–3170. https://doi.org/10.1016/j.febslet.2007.05.082 265 Wang, W., Daley, J.M., Kwon, Y., Krasner, D.S., Sung, P., 2017. Plasticity of the Mre11– Rad50–Xrs2–Sae2 nuclease ensemble in the processing of DNA-bound obstacles. Genes Dev. 31, 2331–2336. https://doi.org/10.1101/gad.307900.117 Wang, W., Daley, J.M., Kwon, Y., Xue, X., Krasner, D.S., Miller, A.S., Nguyen, K.A., Williamson, E.A., Shim, E.Y., Lee, S.E., Hromas, R., Sung, P., 2018. A DNA nick at Ku-blocked double-strand break ends serves as an entry site for exonuclease 1 (Exo1) or Sgs1-Dna2 in long-range DNA end resection. J. Biol. Chem. 293, 17061–17069. https://doi.org/10.1074/jbc.RA118.004769 Wang, Y., Lyu, Y.L., Wang, J.C., 2002. Dual localization of human DNA topoisomerase IIIα to mitochondria and nucleus. Proc. Natl. Acad. Sci. 99, 12114–12119. https://doi.org/10.1073/pnas.192449499 Wanzek, K., Schwindt, E., Capra, J.A., Paeschke, K., 2017. Mms1 binds to G-rich regions in Saccharomyces cerevisiae and influences replication and genome stability. Nucleic Acids Res. 45, 7796–7806. https://doi.org/10.1093/nar/gkx467 Ward, J.J., McGuffin, L.J., Bryson, K., Buxton, B.F., Jones, D.T., 2004. The DISOPRED server for the prediction of protein disorder. Bioinformatics 20, 2138–2139. https://doi.org/10.1093/bioinformatics/bth195 Warner, J.R., 1999. The economics of ribosome biosynthesis in yeast. Trends Biochem. Sci. 24, 437–440. https://doi.org/10.1016/S0968-0004(99)01460-7 Waterman, D.P., Haber, J.E., Smolka, M.B., 2020. Checkpoint Responses to DNA DoubleStrand Breaks. Annu. Rev. Biochem. 89, 103–133. https://doi.org/10.1146/annurevbiochem-011520-104722 Waters, L.S., Minesinger, B.K., Wiltrout, M.E., D’Souza, S., Woodruff, R.V., Walker, G.C., 2009. Eukaryotic Translesion Polymerases and Their Roles and Regulation in DNA Damage Tolerance. Microbiol. Mol. Biol. Rev. 73, 134–154. https://doi.org/10.1128/MMBR.00034-08 Watson, J.D., Crick, F.H.C., 1953. Genetical Implications of the Structure of Deoxyribonucleic Acid. Nature 171, 964–967. https://doi.org/10.1038/171964b0 Wechsler, T., Newman, S., West, S.C., 2011. Aberrant chromosome morphology in human cells defective for Holliday junction resolution. Nature 471, 642–646. https://doi.org/10.1038/nature09790 Wein, N., Vulin, A., Falzarano, M.S., Szigyarto, C.A.-K., Maiti, B., Findlay, A., Heller, K.N., Uhlén, M., Bakthavachalu, B., Messina, S., Vita, G., Passarelli, C., Passarelli, S., Bovolenta, M., Neri, M., Gualandi, F., Wilton, S.D., Rodino-Klapac, L.R., Yang, L., Dunn, D.M., Schoenberg, D.R., Weiss, R.B., Howard, M.T., Ferlini, A., Flanigan, K.M., 2014. Translation from a DMD exon 5 IRES results in a functional dystrophin isoform that attenuates dystrophinopathy in humans and mice. Nat. Med. 20, 992–1000. https://doi.org/10.1038/nm.3628 Tomás Lama Díaz 272 Figure 53. Putative NLS in Pif1. ............................................................................................ 138 Figure 54. Importing of Pif1 nuclear isoform is impaired in Pif1∆775-803 ............................... 139 Figure 55. Mutation of 781KKRK784 basic patch located at position 781-784 suppress PIF1 overexpression toxicity. ......................................................................................................... 140 Figure 56. 781KKRK784 is necessary to mediate import of Pif1 nuclear isoforms. ................ 141 Figure 57. 781KKRK784 is sufficient to drive the nuclear accumulation of eGFP. ................. 142 Figure 58.Functional characterization of purified Pif1nls. ...................................................... 143 Figure 59. Disruption of Pif1 NLS does not completely abolish nuclear function. ............... 144 Figure 60. pif1mit recapitulates the nuclear phenotypes of pif1∆. .......................................... 145 Figure 61. pif1-m1 retains functional mitochondria. .............................................................. 146 Figure 62. In-frame near-cognate start codons surrounding AUG1 could produce mitochondrialtargeted isoforms. ................................................................................................................... 148 Figure 63. Translation initiation from near-cognates AUGs can sustain mitochondrial function in the absence of AUG1. ......................................................................................................... 149 Figure 64. Mutations of four near-cognates in pif1-m1 phenocopies pif1∆ mitochondrial defects. ................................................................................................................................................ 150 Figure 65. Improved, mitochondrial-deficient pif1-m1L9L and pif1-m14L alleles phenocopy the HU hypersensitivity of pif1∆ in rad3-102 mutants. .............................................................. 151 Figure 66. Integrated model of Pif1 isoforms production mediated by leaky scanning in different PIF1 alleles. ........................................................................................................................... 154 Figure 67. Analysis of the Kozak context in dual-targeted genes in S. cerevisiae. ............... 161 Figure 68. Analysis of the Kozak context in dual-targeted genes in H. sapiens. .................. 162 Figure 69. Isoform p110 of ADAR1 is produced by both alternative splicing and leaky scanning. ................................................................................................................................................ 163 Figure 70. hPIF1α produces two isoforms by ATI with dual-targeted localization. ............. 165 Figure 71. Coomassie staining of all Pif1 variants purified in this study. Pif1, the ATPase-dead version Pif1KA, Pif1nls and the truncated isoform Pif1107-859 were purified, analysed by SDSPAGE and stained with Coomassie. ...................................................................................... 210 Yen1 resolution activity is restrained by Cdk1 until the onset of anaphase. To understand the biological consequences of this regulation, we analyzed strains harbouring a constitutively active allele of YEN1 in different genetic backgrounds. Here, we show that deletion of PIF1 helicase in YEN1ON strains results in a dramatic reduction of its viability under genotoxic stress. Further characterization with classical PIF1 separationof-function alleles failed to recapitulate such genetic interaction. This prompted us to delve into the translational mechanism of Pif1, leading us to the re inement of the molecular mechanism of alternative translation initiation for PIF1 mRNA, the discovery of a new nuclear Pif1 isoform and the development of the first bonaide mitochondrial and nuclear-null Pif1 alleles.