La recombinaison homologue est essentielle à la réparation des lésions de l’ADN ainsi qu’à la ségrégation correcte des chromosomes en méiose. Une étape importante de la recombinaison méiotique est la formation des crossovers (CO). Au cours de ma thèse, j’ai mis en évidence un nouveau mécanisme de régulation de la recombinaison méiotique. J'ai montré que les cycles d'activation et de désactivation des cullin-RING ligases (CRL) sont absolument nécessaires à la recombinaison méiotique. Les CRL sont activées par neddylation et désactivées par la deneddylation. De plus, elles peuvent aussi être inhibées par la séquestration via la protéine CAND1. Mon travail a démontré que ces trois niveaux de régulation des CRL jouent des rôles cruciaux dans la recombinaison homologue méiotique chez A. thaliana. J’ai montré qu'AXR1, un composant clé de la machinerie de neddylation, est nécessaire à la localisation correcte des CO méiotiques et à la recombinaison homologue somatique. J’ai aussi prouvé que le processus de deneddylation médié par CSN5A est nécessaire à la formation des CO. J'ai obtenu des données montrant que cette régulation de la localisation des CO agit à travers la régulation d’un complexe CRL4. Enfin, j’ai pu montrer que l'inhibiteur des CRL, CAND1, est requis pour la formation de plus de 90 % des CO. En utilisant des outils génétiques et cytologiques, j'ai montré que CAND1 agit probablement sur la régulation du biais inter-homologue. L’ensemble de ces données, met l’accent sur un nouveau mécanisme de la régulation de la recombinaison homologue, connectant pour la première fois la méiose et l’ubiquitination via les cullin-RING Ligases. / Homologous recombination is essential to all living organisms in order to repair DNA damages. In addition, a large majority of organisms use homologous recombination in meiosis to ensure proper chromosome segregation. A main step of meiotic recombination is crossover (CO) formation. During my PhD, I was able to highlight a new pathway controlling meiotic recombination. I showed that cycles of activation and deactivation of cullin-RING ligases (CRLs) are absolutely required for correct meiosis. CRLs are activated by neddylation, and deactivated by deneddylation. In addition, they can also be inhibited by sequestration by the CAND1 protein. My work demonstrated that these three levels of CRL regulation play crucial roles in meiotic homologous recombination in A. thaliana. First, I showed that AXR1, a key component of the neddylation machinery, is required for the correct localisation of meiotic COs and for somatic homologous recombination. Second, I showed that the deneddylation process mediated by CSN5A is also necessary for normal CO formation. I obtained evidence that this regulation of CO position is likely to be mediated by a CRL4 complex. Last, I could show that the CRL inhibitor, CAND1, is required for the formation of up to 90% of the COs. Using genetic and cytological tools, I showed that CAND1 probably acts on the regulation of the inter-homolog bias. Considering all these data, my work draws the attention to a new mechanism regulating meiotic homologous recombination, connecting for the first time meiosis to CRL-mediated ubiquitylation.

Évolution de la recombinaison méiotique: variation et fonction du gène Prdm9 chez la souris. L’espèce M. musculus est divisée en trois sous-espèces dont deux en Europe, M. m. musculus et M. m. domesticus. Lorsque certaines lignées de souris venant de ces différentes sous-espèces (PWD de M. m. musculus et C57BL/6 de M. m. domesticus) sont croisées, on observe une stérilité chez les descendants mâles, due à un arrêt en prophase de la première division de méiose. Cette stérilité est liée à des incompatibilités génétiques qui impliquent une combinaison spécifique d’allèles du gène Prdm9, une hétérozygotie génomique ainsi qu’une région du chromosome X. Le gène Prdm9, qui est l’un des gènes évoluant le plus rapidement chez les rongeurs et les primates, a ainsi été proposé être un gène impliqué dans la spéciation. Au niveau moléculaire, Prdm9 spécifie la localisation des événements de recombinaison méiotique en des sites qui peuvent différer selon les allèles. Lors de ma thèse, j’ai entrepris de tester l’hypothèse du rôle de Prdm9 dans la spéciation en évaluant la généralité des incompatibilités dues à Prdm9. J’ai testé la fertilité de 32 croisements réalisés à partir de 8 lignées de souris issues de la nature, 4 de chaque sous espèce et portant des allèles de Prdm9 distincts. L’analyse du poids testiculaire et de la spermatogénèse par histologie et immuno-cytochimie a révélé un phénotype sauvage chez tous ces hybrides. Par contre, en croisant la lignée C57BL/6 avec les 4 lignées de M. m. musculus utilisées précédemment, j’ai observé une corrélation entre la présence d’un allèle spécifique de M. m. musculus (PWD) et la stérilité, avec des variations des phénotypes indiquant que les facteurs impliqués n’ont pas été fixés.J’ai donc démontré que la stérilité hybride est restreinte à des combinaisons spécifiques d’allèles Prdm9 et de fonds génétiques, ce qui définit de manière précise le contexte dans lequel Prdm9 pourrait jouer un rôle dans la spéciation. Compte tenu du phénomène d’érosion des sites de PRDM9 par conversion génique et que leur hétérozygotie peut conduire à des défauts de recombinaison et à la stérilité, mes données conduisent également à des prédictions sur la fréquence et l’activité de différents allèles de Prdm9 sur les génomes de M. m. musculus et M. m. domesticus.En parallèle, j’ai également initié une étude portant sur 300 souris sauvages d’origines géographiques et phylogénétiques variées : par capture et séquençage de régions génomiques d’intérêt, nous chercherons à déchiffrer les modes de sélection régissant l’évolution de la région génomique du gène Prdm9. Lors de ma thèse, j’ai entrepris de tester l’hypothèse du rôle de Prdm9 dans la spéciation en évaluant la généralité des incompatibilités dues à Prdm9. J’ai testé la fertilité de 32 croisements réalisés à partir de 8 lignées de souris issues de la nature, 4 de chaque sous espèce et portant des allèles de Prdm9 distincts. L’analyse du poids testiculaire et de la spermatogénèse par histologie et immuno-cytochimie a révélé un phénotype sauvage chez tous ces hybrides. Par contre, en croisant la lignée C57BL/6 avec les 4 lignées de M. m. musculus utilisées précédemment, j’ai observé une corrélation entre la présence d’un allèle spécifique de M. m. musculus (PWD) et la stérilité, avec des variations des phénotypes indiquant que les facteurs impliqués n’ont pas été fixés.J’ai donc démontré que la stérilité hybride est restreinte à des combinaisons spécifiques d’allèles Prdm9 et de fonds génétiques, ce qui définit de manière précise le contexte dans lequel Prdm9 pourrait jouer un rôle dans la spéciation. Compte tenu du phénomène d’érosion des sites de PRDM9 par conversion génique et que leur hétérozygotie peut conduire à des défauts de recombinaison et à la stérilité, mes données conduisent également à des prédictions sur la fréquence et l’activité de différents allèles de Prdm9 sur les génomes de M. m. musculus et M. m. domesticus. / Meiotic recombination evolution: Prdm9 gene variation and evolution in miceHouse mouse Mus musculus is divided into three sub species, two of which are found in Europe, M. m. musculus and M. m. domesticus. Male hybrids from some mouse strains derived from these sub-species (PWD from M. m. musculus and C57BL/6 from M. m. domesticus) are sterile with a meiotic arrest during the first meiotic prophase. This sterility is due to genetic incompatibilities involving heterozygosity at Prdm9 and in the genome as well as a locus on the X chromosome. Prdm9, which one of the fastest evolving gene in primates and rodents, was thus proposed to be a speciation gene. At the molecular level, Prdm9 is known to specify the sites of meiotic recombination which can be distinct in different Prdm9 alleles.During my thesis, I tested the hypothesis of the role of Prdm9 in speciation by evaluating the generality of incompatibilities due to Prdm9. I tested the fertility of 32 hybrids made by crossing 8 wild-derived mouse strains from both sub-species and carrying distinct Prdm9 alleles. Based on the analysis of testis weight and of spermatogenesis by histological and cytological analysis, I observed a wild type phenotype in all hybrids. Instead when C57BL/6 was crossed with the four M. m. musculus strains used in previous crosses, I observed a correlation between the presence of the PWD allele and sterility, with variations in the strength of the phenotype.I have thus shown that hybrid sterility is limited to specific combinations of Prdm9 alleles together with specific genomic backgrounds which define contexts in which Prdm9 could act as a speciation gene. Given that PRDM9 sites are eroded by gene conversion and that heterozygosity at PRDM9 sites can lead to defects in meiotic recombination and to sterility, my observations lead to predictions about Prdm9 allele frequencies and Prdm9 historical activities on the genomes of M. m. musculus and M. m. domesticus.Simultaneously, I initiated a study on nearly 300 mice of various geographic and phylogenetic origins: using genomic sequence capture technology, we are aiming to decipher selection forces affecting the evolution of Prdm9 gene genomic region.

Repair of double-strand DNA breaks (DSBs) by the homologous recombination (HR) pathway results in crossovers (COs) required for a successful first meiotic division. DSB resection is the nucleic degradation of DSB ends to expose 3’ single strand DNA (ssDNA), an intermediate required for HR. To investigate genes involved in meiosis, a forward genetic screen was performed to search for novel genes or informative new mutant alleles of known genes. Mre11 is one member of the MRX/N (Mre11-Rad50-Xrs2/Nbs1) complex required for meiotic DSB formation and for resection in budding yeast. In Caenorhabditis elegans, evidence for the MRX/N’s role in DSB resection is limited. We isolated the first separation of function allele in C. elegans , mre 11(iow1), isolated from our forward genetic screen. The mre-11(iow1) mutants are specifically defective in meiotic DSB resection but not in DSB formation. The mre 11(iow1) mutants display chromosomal fragmentation and aggregation in late prophase I. Recombination intermediates and crossover formation is greatly reduced in mre 11(iow1) mutants. Irradiation induced DSBs during meiosis fail to be repaired from the early to middle prophase I in mre 11(iow1) mutants. Our data suggest that some DSBs in mre 11(iow1) mutants are repaired by the non homologous end joining (NHEJ) pathway because removing NHEJ partially suppresses some meiotic defects conferred by mre 11(iow1). In the absence of NHEJ and a functional MRX/N, meiotic DSBs are channeled to an EXO 1 dependent form of recombination repair. Overall, our analysis supports a role for MRE-11 in the resection of DSBs in early to middle meiotic prophase I and in blocking NHEJ. A reverse genetic screen and a yeast two hybrid screen were performed to search for genes with genetic and/or physical interactions with mre-11. The reverse genetic screen isolated a novel meiotic gene, nhr-2, as a partial suppressor of the meiotic defects conferred by mre-11(iow1). The yeast two hybrid screen identified kin-18 interacting with mre-11. KIN-18 is the C. elegans homolog of mammalian Thousand And One kinase (TAO) kinase. KIN-18/TAO is MAPK kinase kinase whose meiotic role was unknown. We have found that KIN-18 is essential for normal meiotic progression as kin-18 mutants exhibit accelerated meiotic recombination, ectopic germ cell differentiation, and enhanced levels of germline apoptosis. In C.elegans MPK-1 activation in late pachytene is required for physiological apoptosis (nuclei removed by apoptosis serve as nursing cells for oocytes) and oocyte differentiation. The kin-18 mutants also showed absence of MPK-1 activation and aberrant MPK-1 activation that includes ectopic activation in the wrong regions in the germline or more than one time of activation. The progression defects in kin-18 mutants are suppressed by inhibiting an upstream activator, KSR-2, of the canonical MPK-1 signaling. Our data suggest KIN-18 affects meiotic progression by modulating the timing of MPK-1 activation. This regulation ensures the proper timing of recombination and normal apoptosis, which is required for the formation of functional oocytes. Meiosis is a conserved process; revealing that KIN-18 is a novel regulator of meiotic progression in C. elegans will motivate hypothesis for TAO kinase’s role in the germline development in higher eukaryotes. Meiosis is a crucial for sexually reproducing organisms to maintain ploidy level from one generation to the next. Accurate chromosome segregation in the meiosis requires meiotic recombination between homologous chromosomes. Failure in recombination can lead to abnormal segregation of chromosomes in meiosis, which leads to aneuploidy. Anueploidy is a leading cause of miscarriages and attributes to chromosomal related birth defects. Meiotic recombination starts with programmed DNA double strand breaks (DSBs), followed by repair of these DSBs by homologous recombination (HR) pathway. One key step in HR is resection, a process to covert DSB ends into single strand DNA (ssDNA). To broaden our understanding of meiotic DSB resection, we used a nematode, C. elegans, as a model to investigate genes in DSB resection. We have isolated a specific mutant allele of a meiotic gene, mre-11. Our data suggest meiotic DSB resection in C. elegans requires collaboration of mre-11 and another gene exo-1; efficient resection of DSB ends is important to safeguard repair of DSB by HR against other illegitimate repair pathway. In addition, we identified a gene kin-18 by looking for genes interacting with mre-11. Characterization of kin-18 show meiotic recombination is tightly coordinated with germ cell progression. Our analysis provides significant improvement in the understanding of meiotic recombination in C. elegans. Given the high conservation of the two genes, mre-11 and kin-18, our finding may be applied to other organisms.

Previous evidence has shown that the simple sequences microsatellites and poly-purine/poly-pyrimidine tracts (PPTs) could be both a cause, and an effect, of meiotic recombination. The causal link between simple sequences and recombination has not been much explored, however, probably because other evidence has cast doubt on its generality, though this evidence has never been conclusive. Several questions have remained unanswered in the literature, and I have addressed aspects of three of them in my thesis. First, what is the scale and magnitude of the association between simple sequences and recombination? I found that microsatellites and PPTs are strongly associated with meiotic double-strand break (DSB) hotspots in yeast, and that PPTs are generally more common in human recombination hotspots, particularly in close proximity to hotspot central regions, in which recombination events are markedly more frequent. I also showed that these associations can't be explained by coincidental mutual associations between simple sequences, recombination and other factors previously shown to correlate with both. A second question not conclusively answered in the literature is whether simple sequences, or their high levels of polymorphism, are an effect of recombination. I used three methods to address this question. Firstly, I investigated the distributions of two-copy tandem repeats and short PPTs in relation to yeast DSB hotspots in order to look for evidence of an involvement of recombination in simple sequence formation. I found no significant associations. Secondly, I compared the fraction of simple sequences containing polymorphic sites between human recombination hotspots and coldspots. The third method I used was generalized linear model analysis, with which I investigated the correlation between simple sequence variation and recombination rate, and the influence on the correlation of additional factors with potential relevance including GC-content and gene density. Both the direct comparison and correlation methods showed a very weak and inconsistent effect of recombination on simple sequence polymorphism in the human genome.Whether simple sequences are an important cause of recombination events is a third question that has received relatively little previous attention, and I have explored one aspect of it. Simple sequences of the types I studied have previously been shown to form non-B-DNA structures, which can be recombinagenic in model systems. Using a previously described sodium bisulphite modification assay, I tested for the presence of these structures in sequences amplified from the central regions of hotspots and cloned into supercoiled plasmids. I found significantly higher sensitivity to sodium bisulphite in humans in than in chimpanzees in three out of six genomic regions in which there is a hotspot in humans but none in chimpanzees. In the DNA2 hotspot, this correlated with a clear difference in numbers of molecules showing long contiguous strings of converted cytosines, which are present in previously described intramolecular quadruplex and triplex structures. Two out of the five other hotspots tested show evidence for secondary structure comparable to a known intramolecular triplex, though with similar patterns in humans and chimpanzees. In conclusion, my results clearly motivate further investigation of a functional link between simple sequences and meiotic recombination, including the putative role of non-B-DNA structures.

To study hybrid sterility our laboratory uses mouse strains PWD/Ph (PWD), derived from Mus musculus musculus wild mice and the common laboratory strain C57BL/6J (B6) mostly of Mus musculus domesticus origin as a model. Crossing between PWD female and B6 male results in sterile male progeny. F1 hybrid males carry defects in the repair mechanisms of asymmetric double-strand DNA breaks (DSBs). Functional interplay of SPO11 and PRDM9 proteins in the meiotic prophase I is necessary for repairs. Its defect leads to incorrect synapse formation between homologous chromosomes, leading to halt in spermatogenesis and thus male sterility. The formation of DSBs and their subsequent repair is essential for first meiotic division. The working hypothesis stems from the findings in yeast model, where supposed antirecombinatorial mechanism of mismatch repair genes Msh6 and Msh2 prevents DSBs repairs during meiosis. Despite the functional mechanism of these two genes is not explicitly known, existence of similar repair system in mice is presumed. Variety of methods was implemented in this thesis. The effects of Msh6 deletion on meiotic prophase I and sperm maturation were performed by designing guide RNAs for CRISPR/Cas9 for creation of three knock-outs in B6 mice. The PCR was used to amplify regions adjacent to the...

L’agriculture moderne doit assurer notre sécurité alimentaire dans le contexte d’un changement climatique qui entraîne une baisse des rendements. Une meilleure compréhension des facteurs contrôlant la recombinaison méiotique ouvrirait la voie à la modification du nombre et de la répartition des crossing-overs, ce qui permettrait une localisation plus précise des facteurs génétiques contrôlant les caractères d’intérêt agronomique et faciliterait le pyramidage d’allèles favorables au sein d’un même génotype élite. Pendant ma thèse, j’ai développé une méthode de mesure à haut débit de la recombinaison chez la levure Saccharomyces cerevisiae pour étudier la diversité de la recombinaison et de l’interférence au sein d’une collection de 24 souches représentatives de la diversité de l’espèce ainsi qu’au sein d’un dispositif di-allèle à cinq parents. Les résultats montrent un nombre moyen de crossing-overs par méiose compris entre 24 et 61, ce qui est plus élevé que chez la majorité des autres espèces. Plus particulièrement, les profils de recombinaison diffèrent entre souches, atteignant un écart d’un facteur 9 dans certaines régions. Les souches originaires d’habitats peu stables n’ont cependant pas un niveau de recombinaison plus élevé que les souches originaires d’environnements stables. En outre, la plupart des souches montrent de l’interférence dont la force est corrélée positivement avec le niveau de recombinaison. L’étude de la relation entre niveau de recombinaison et similarité de séquence entre homologues, à différentes échelles locales ou globales, indique que la recombinaison est contrôlée à la fois par des éléments cis et des facteurs trans. Par ailleurs, l’hétérozygotie chez les hybrides a un effet négatif sur le niveau de recombinaison, mais les homozygotes ont aussi un niveau de recombinaison réduit par un effet de dépression de consanguinité. Ce travail permettra maintenant d’étudier la réponse de la recombinaison à la sélection et de détecter les QTL de nombre de crossing-overs, afin d’identifier des gènes qui contrôlent la recombinaison. / Modern agriculture must ensure food security in a context of climate change that will lower yields. A better understanding of the factors controlling meiotic recombination could pave the way to modifying the number and distribution of crossing-over, which would allow a more precise localization of genetic factors controlling agronomic traits, and facilitate gene pyramiding in selection programs. During my thesis, I developed a method for high-throughput measurement of recombination rates in the yeast Saccharomyces cerevisiae. This allowed me to study the diversity of recombination and interference in a collection of 24 strains representing most of the diversity of the species, as well as within a five-parent di-allele design. The results show an average number of crossovers per meiosis ranging between 24 and 61, higher than in the majority of other species. Furthermore, recombination patterns differ between strains, and ratios of local recombination rates show 9-fold differences in some regions. Strains from unstable habitats, however, do not have a higher level of recombination than those from stable environments. In addition, most strains show interference whose strength is positively correlated with the level of recombination. The study of the relationship between recombination rate and sequence similarity between homologs at different scales (from local to global) indicates that recombination is controlled by both cis elements and trans factors. Lastly, heterozygosity in hybrids has a negative effect on crossing-over, but homozygotes also have a reduced level of recombination due to inbreeding depression. This work will now be used to study the response of recombination to selection and to detect QTL of crossover number in order to identify genes controlling recombination.

Meiosis is a specialized cell division that produces haploid cells (gametes) from diploid progenitors. During meiosis parental chromosomes (homologs) need to pair, synapse and eventually segregate. Faithful chromosome segregation depends on chromosome recombination. In the beginning of prophase I programmed double strand breaks (DSBs) are introduced in meiotic cells by SPO11 enzyme. DSBs are positioned at hotspot sites that are specified by that action of DNA-binding histone methyltransferase PRDM9. Specific enzymes act at the site of breaks to create 5’ single stranded DNA ends. With the assistance of the strand exchange proteins DMC1 and RAD51 these ends invade homologous DNA sequence and DSB repair is initiated. DSB repair can be completed either as a crossover (reciprocal exchange of DNA) or as a non-crossover. Crossover events lead to the formation of chiasmata between homologs and ensure proper segregation during the first meiotic division. An interesting feature in male meiosis is the XY chromosomes. The shared region between sex chromosomes is short and is called pseudoautosomal region (PAR). Due to their large non synapsed region, XY chromosomes need to be transcriptionally silenced. Thus they are covered with the phosphorylated histone variant H2AX (γH2AX) forming the so called sex body. PAR region has higher density of DSBs than autosomes and it had been shown that sex chromosomes undergo delayed homologous pairing. Nevertheless little is known how meiotic recombination is regulated in PAR region of sex chromosomes. In close proximity with sex body it has been found a structure named dense body (DB). There are few reports suggesting that DB contains RNAs/proteins but no DNA. Its role in meiosis was unclear because no structural component had been described. In the present thesis the role of two meiotic expressed genes is described. In our group after performing RNA screens we identified several genes that are highly expressed during meiotic prophase I. Based on the expression profile we selected polycomb-related sex comb on midleg like 1 (Scml1) gene and the ankyrin repeat domain 31 (Ankrd31) to study their role in mammalian meiosis.:List of figures i List of abbreviations ii 1. Introduction 1 1.1 Gametogenesis 1 1.2 Meiotic prophase I 2 1.2.1 Meiotic recombination 4 1.2.2 Regulation of meiotic recombination 7 1.2.2.1 Meiotic recombination hotspots and PRDM9 activity 7 1.2.2.2 Meiotic surveillance mechanisms 8 1.3 Unique properties of XY recombination 9 1.4 Sex chromatin associated structure: The dense body 10 1.5 Aim of the thesis 11 2. Publications 12 3. Discussion 92 4. Summary 98 5. References 102 Acknowledgements 108 Declarations 109

La recombinaison méiotique initiée en prophase I de méiose génère soit des crossing-over (COs), qui sont des échanges réciproques entre segments chromosomiques, ou des conversions géniques non associées aux COs (NCOs). Les deux types d'événements se produisent dans de petites régions (moins de 10 kilobases) appelées points chauds, qui sont distribuées de manière non homogène le long des chromosomes. L'objectif de ma thèse était la caractérisation d'un point chaud de recombinaison méiotique (nommée 14a) chez Arabidopsis thaliana (i) dans différentes accessions (ii) dans le mutant msh4, un gène impliqué dans la formation des COs. Dans les deux hybrides ColxLer et ColxWs (i) 14a a un taux très élevé de COs 0,85% et 0,49%, respectivement (ii) Les COs sont regroupés dans deux petites régions de quelques kilobases, 14a1 et 14a2 avec une distribution de type gaussienne observée aux points chauds décrits dans d'autres espèces (iii) 14a1 est aussi un point chaud de NCO avec un taux aussi élevé que celui des COs (0,5%) dans ColxLer (iv) un biais de l'initiation de recombinaison a été trouvé dans 14a1 aussi bien pour les COs que les NCOs dans le fond génétique ColxLer.Une réduction de la fréquence de CO a été observée dans le mutant msh4 dans le fond génétique ColxLer à 14a1 et 14a2 (6,4% et 18,7% par rapport au sauvage). Cela confirme le rôle précédemment connu de la protéine MSH4 impliqué dans la formation de CO. La fréquence de NCO à 14a1 est similaire à celle observéedans le fond sauvage. Le rôle des H3K4 histones trimethyltransferase d’Arabidopsis dans la recombinaison méiotique (comme précédemment observé comme Set1 chez S. cerevisiae ou PRDM9 chez les mammifères) a également été étudiée. Aucun des dix gènes d’histones méthyltransférase étudié n'a montré de rôle dans la méiose. Cela pourrait être dû à (i) une forte redondance de la fonction entre les protéines (ii) une autre histone méthyltransférase en charge de l'étiquetage des points chauds de recombinaison méiotique (plus de 29 putatif histone méthyltransférase ont été identifiés dans le génome d'Arabidopsis!) (iii) contrairement à S. cerevisiae, les souris et l'homme, un autre mécanisme de contrôle épigénétique de la recombinaison méiotique. / Meiotic recombination initiated in prophase I of meiosis generates either crossovers (COs), which are reciprocal exchanges between chromosome segments, or gene conversion not associated to crossovers (NCOs). Both kinds of events occur in narrow regions (less than 10 kilobases) called hotspots, which are distributed non-homogenously along chromosomes. The aim of my PhD was the characterization of a hotspot of meiotic recombination (named 14a) in Arabidopsis thaliana (i) across different accessions (ii) in msh4 mutant, a gene involved in CO formation. In both ColxLer and ColxWs hybrids (i) 14a had a very high rate of COs 0.85% and 0.49%, respectively (ii) COs clustered in two small regions of a few kilobases, 14a1 and 14a2 with typical Gaussian curve distribution observed in other organisms (iii) 14a1 was also a hotspot of NCO with high rate (0.5%) in ColxLer (iv) a bias of recombination initiation at 14a1 CO and NCO hotspot was found in ColxLer. A reduction of CO frequency was observed in msh4 mutant in ColxLer background at 14a1 (6.4%) and 14a2 (18.7%) compared to wild type. This confirmed previously known role of MSH4 protein in CO formation. Frequency of NCO at 14a1 was similar to wild type. The role of putative Arabidopsis histone H3K4 trimethyltransferase in meiotic recombination as previously observed like Set1 in S.cerevisiae or PRDM9 in mammals (mice and human) was also studied. None of ten putative histone methyltransferase genes was involved in meiosis. This could be due to (i) a strong redundancy of function between gene products (ii) another histone methyltransferase in charge of labeling meiotic recombination hotspots (more than 29 putative histone methyltransferase have been identified in the Arabidopsis genome!) (iii) contrary to S. cerevisiae, mice and humans, another mechanism for epigenetic control of meiotic recombination

Saccharomyces cerevisiae HOP1, which encodes a component of the synaptonemal complex, plays an important role in both gene conversion and crossing over between homologs, as well as enforces the meiotic recombination checkpoint control over the progression of recombination intermediates. The zinc-finger motif (Znf) 348CX2CX19CX2C374) of Hop1 is crucial for its function in meiosis, since mutation of conserved Cys371 to Ser in this motif results in a temperature-sensitive phenotype, which is defective in sporulation and meiosis. The direct role for Hop1 or its ZnF in the formation of joint molecules and checkpoint control over the progression of meiotic recombination intermediates is unknown. To understand the underlying biochemical mechanism, we constructed a series of recombination intermediates. Hop1 or its ZnF were able to bind different recombination intermediates. Interestingly, the binding affinity of Hop1 and its ZnF was much higher for the Holliday junction as compared to other recombination intermediates. The complexes of Hop1 or its ZnF with the Holliday junction were stable and specific as shown by NaCl titration and competition experiment. Hop1 and its ZnF blocked BLM helicase-induced unwinding of the Holliday junction, indicating that the interaction between Hop1 and its ZnF with the Holliday junction is specific. DNase I footprinting experiment showed that Hop1 or its ZnF bind to the center of the Holliday junction. 2-aminopurine fluorescence and KMnO4 experiments showed that Hop1 or its ZnF can distort the Holliday junction in a 2-fold symmetrical manner. The molecular modeling study showed that Hop1 ZnF folded into unique helix-loop-helix motif and bound to center of the Holliday junction. In summary, this study shows that Hop1 protein or its ZnF interact specifically with the Holliday junction and distort its structure. Taken together, these results implicate that Hop1 protein might coordinate the physical monitoring of meiotic recombination intermediates during the process of branch migration and that Hop1 ZnF acts as a structural determinant of Hop1 protein functions.

La recombinaison méiotique joue un rôle essentiel dans la ségrégation des chromosomes pendant la méiose et dans la création de nouvelles combinaisons du matériel génétique des espèces. Ses effets cause une déviation du principe de l'assortiment indépendant de Mendel; cependant, les mécanismes moléculaires impliqués restent partiellement incompris jusqu'à aujourd'hui. Il s'agit d'un processus hautement régulé et de nombreuses protéines sont impliquées dans son contrôle, dirigeant la recombinaison méiotique dans des régions génomiques de 1 à 2 kilobases appelées « hotspots ». Au cours des dernières années, l'apprentissage profond a été appliqué avec succès à la classification des séquences génomiques. Dans ce travail, nous appliquons l'apprentissage profond aux séquences d'ADN humain afin de prédire si une région spécifique d'ADN est un hotspot de recombinaison méiotique ou non. Nous avons appliqué des réseaux de neurones convolutifs sur un ensemble de données décrivant les hotspots de quatre individus non-apparentés, atteignant une exactitude de plus de 88 % avec une précision et un rappel supérieur à 90 % pour les meilleurs modèles. Nous explorons l'impact de différentes tailles de séquences d'entrée, les stratégies de séparation des jeux d'entraînement/validation et l’utilité de montrer au modèle les coordonnées génomiques de la séquence d'entrée. Nous avons exploré différentes manières de construire les motifs appris par le réseau et comment ils peuvent être liés aux méthodes classiques de construction de matrices position-poids, et nous avons pu déduire des connaissances biologiques pertinentes découvertes par le réseau. Nous avons également développé un outil pour visualiser les différents modèles afin d'aider à interpréter les différents aspects du modèle. Dans l'ensemble, nos travaux montrent la capacité des méthodes d'apprentissage profond à étudier la recombinaison méiotique à partir de données génomiques. / Meiotic recombination plays a critical role in the proper segregation of chromosomes during meiosis and in forming new combinations of genetic material within sexually-reproducing species. For a long time, its side effects were observed as a deviation from the Mendel’s principle of independent assortment; however, its molecular mechanisms remain only partially understood until today. We know that it is a highly regulated process and that many molecules are involved in this tight control, resulting in directing meiotic recombination into 1-2 kilobase genomic pairs regions called hotspots. During the past few years, deep learning was successfully applied to the classification of genomic sequences. In this work, we apply deep learning to DNA sequences in order to predict if a specific stretch of DNA is a meiotic recombination hotspot or not. We applied convolution neural networks on a dataset describing the hotspots of four unrelated male individuals, achieving an accuracy of over 88% with precision and recall above 90% for the best models. We explored the impact of different input sequence lengths, train/validation split strategies and showing the model the genomic coordinates of the input sequence. We explored different ways to construct the learnt motifs by the network and how they can relate to the classical methods of constructing position-weight-matrices, and we were able to infer relevant biological knowledge uncovered by the network. We also developed a tool for visualizing the different models output in order to help digest the different aspects of the model. Overall, our work shows the ability for deep learning methods to study meiotic recombination from genomic data.