Les bactéries sont continuellement exposées à un danger, la prédation par des bactériophages. Pour se défendre, elles ont développé une grande variété de mécanismes. Parmi ceux-ci, on retrouve CRISPR-Cas (« clustered regularly interspaced palindromic repeats »), un système adaptatif que possèdent environ 45% des bactéries. Une caractéristique unique du système CRISPR-Cas est qu’il constitue en quelque sorte la mémoire de l’hôte. Par exemple, le système peut emmagasiner des petits fragments d’un génome viral, appelés espaceurs, et les introduire dans son CRISPR. Cette mémoire lui permet de se défendre contre une réinfection par le même virus ou un virus hautement apparenté. Par contre, malgré que l’acquisition de nouveaux espaceurs semble fréquente dans la nature, ce phénomène n’est que très rarement observé en conditions de laboratoire. Néanmoins, quelques bactéries font exception à la règle et l’une d’entre elles est Streptococcus mutans. Dans le cadre de cette étude, l’interaction entre la souche S. mutans P42S et le bactériophage virulent M102AD a été analysée en détail. De plus, certaines applications potentielles du système CRISPR-Cas ont également été approfondies. Le premier objectif de cette thèse était de caractériser le système CRISPR-Cas de S. mutans P42S au niveau moléculaire et de déterminer son rôle dans les interactions phage-bactérie. Le deuxième objectif était d’établir le potentiel de la protéine Cas9 de S. mutans P42S (SmutCas9) comme nouvel outil d’édition génomique. S. mutans P42S possède un système CRISPR-Cas de type II-A. Bien que ce type de système soit probablement le plus étudié, celui de S. mutans P42S présente plusieurs caractéristiques uniques lui permettant de se démarquer. En effet, ce dernier reconnaît un PAM différent de ce qui était auparavant connu pour cette espèce bactérienne, l’acquisition simultanée de multiples espaceurs semble fréquente, ce qui est probablement dû au phénomène de « priming ». Malgré le rôle de CRISPR-Cas dans la défense antivirale, S. mutans P42S dispose d’autres mécanismes de défense contre les phages. Des cellules mutantes sont résistantes aux phages en empêchant l’adsorption de particules virales à la cellule ont notamment été observées. D'autres mécanismes sont assurément impliqués dans la défense antivirale de S. mutans. Finalement, SmutCas9 s’est montrée efficace dans l’édition de génomes viraux et elle apparaît comme une candidate à explorer pour cette application. / Bacteria are exposed to the constant threat of viral predation. To defend themselves, bacteria have developed a wide variety of different mechanisms. One of these mechanisms is CRISPR-Cas (clustered regularly interspaced palindromic repeats), an adaptive immune mechanism found in approximately 45% of bacteria. A unique feature of CRISPR-Cas systems compared to other antiviral defence mechanisms is that it has a memory. The system is capable of remembering previous viral encounters and protects the bacterial host from re-infection by the same or highly-related viruses. This memory is due to the acquisition of virus-derived genome fragments called spacers. Despite common acquisition of novel spacers in nature, and thereby the emergence of new immunity, acquisition of new spacers under laboratory conditions has been rarely observed. One of the few exceptions is Streptococcus mutans. In this study, the interactions between S. mutans strain P42S and its virulent bacteriophage M102AD are investigated in detail. In addition, possible applications of the CRISPR-Cas system are analysed. The first objective of this thesis was to characterise the CRISPR-Cas system of S. mutans P42S on the molecular level and to determine its role in antiviral defence. The second objective was to determine the potential of the Cas9 protein of S. mutans P42S (SmutCas9) in genome editing. S. mutans P42S possesses a type II-A CRISPR-Cas system. Although this is arguably the best studied system, the one found in the strain S. mutans P42S has several features that makes it stand out. It recognises a PAM different from what was known for this species, multiple spacer acquisitions are frequent, and this appears to be partially due to priming. Although CRISPR-Cas plays a role in antiviral defence, there are additional antiviral defence mechanisms that protect S. mutans against phages. Adsorption resistance is one of them, although additional unidentified antiviral defence mechanisms are likely involved. Finally, SmutCas9 has been shown functional in editing of viral genomes and appears to be a candidate for human genome editing.

Titre de l'écran-titre (visionné le 15 mai 2023) / La métagénomique virale a permis l'identification de milliers de nouvelles séquences de phages dans une variété d'écosystèmes. Leur caractérisation est limitée par leur grande diversité, puisque la majorité des nouvelles séquences virales ne ressemble à aucun phage connu et répertorié. Pour faire face à ce défi, de nouveaux outils de la bio-informatique doivent être développés permettant de mieux comprendre les interactions phages-bactéries et ainsi élucider le rôle des phages dans leur écosystème. Une approche pour étudier les interactions phages-bactéries repose sur l'exploitation des systèmes CRISPR-Cas. Chez les bactéries et les archées, ces systèmes servent de mécanismes de défense contre le matériel génétique envahisseur comme le génome des phages. Grâce à de courtes séquences appelées espaceurs, incorporées dans leur locus CRISPR, les bactéries qui portent ce système peuvent reconnaître rapidement un génome de phage, le couper et ainsi bloquer l'infection. Les espaceurs sont archivés et continuent d'être ajoutés au fil des infections, rendant le locus CRISPR très dynamique. Pour les analyses bio-informatiques, les espaceurs ont plusieurs utilités, notamment pour le typage de souches bactériennes et pour la prédiction des hôtes bactériens de séquences virales. Cependant, les quelques outils existants ont été développés avant l'accélération massive dans la découverte de nouvelles séquences de phages offerte par la métagénomique virale. Ainsi, beaucoup de travail répétitif, manuel et non standardisé était requis pour utiliser les espaceurs CRISPR dans le contexte d'études de métagénomique. Les deux premiers chapitres de cette thèse sont dédiés au développement d'outils bio-informatiques exploitant les loci CRISPR. Dans un premier temps, le logiciel CRISPRStudio a été développé pour automatiser la création de figures représentant les loci CRISPR. Ce logiciel offre une grande polyvalence, tout en accélérant la production de figures. Dans un deuxième temps, un second logiciel a été créé pour prédire les hôtes bactériens des phages. Celui-ci s'appuie sur une base de données d'espaceurs de plus de 11 millions de séquences et d'une série de critères fondés sur la biologie des systèmes CRISPR-Cas pour sélectionner l'hôte bactérien le plus probable d'un phage. Cet outil a permis d'améliorer les performances, la standardisation et la facilité d'utilisation des approches de prédiction de l'hôte utilisant les espaceurs CRISPR. Puis, les deux outils ont été mis à profit dans le troisième chapitre pour l'étude spécifique des interactions phages-bactéries entre Escherichia coli et ses phages, dans le contexte du microbiote intestinal. La caractérisation des loci CRISPR a permis d'élucider le rôle probable du système CRISPR-Cas chez cette espèce, soit un mécanisme anti-prophages. À ma connaissance, il s'agit également de la première étude identifiant une aussi grande proportion des cibles des espaceurs, en trouvant l'origine de 60 % des proto-espaceurs. / Viral metagenomics has allowed the identification of thousands of new phage sequences in various ecosystems. Yet, their characterization is still limited by their great diversity, as the majority of new viral sequences resembles no known phages in reference databases. To tackle this challenge, new bioinformatic tools are needed to better understand phage-bacteria interactions and to elucidate the role of phages in their ecosystem. One approach to study phage-bacteria interactions consists in taking advantage of CRISPR-Cas systems. In bacteria and archaea, these systems act as defense mechanisms against invading genetic material, such as phage genomes. Thanks to short sequences called spacers incorporated in their CRISPR locus, bacteria that carry this system can rapidly recognize a phage genome and block its infection, analogous to an adaptive immune system. Spacers are archived and continue to be added upon phage infection, which makes the CRISPR locus highly dynamic. In bioinformatics analyses, spacers can be utilized to perform strain typing and to predict the bacterial hosts of phages. However, the few existing tools were developed before the metagenomics era and the massive discovery of new phage sequences. Thus, exploiting CRISPR loci for viral metagenomics projects requires repetitive, manual, and non-standardized work. The first two chapters of this thesis are dedicated to developing bioinformatics tools exploiting CRISPR loci. First, a software was developed to automatize the creation of figures representing CRISPR loci. CRISPRStudio offers versatility, is user-friendly and accelerates the figure production. Second, another software was created to predict the bacterial hosts of phages. It relies on a spacer database containing more than 11 million sequences and on a set of criteria inspired by CRISPR-Cas biology to select the most likely bacterial host for a phage genome. This tool improved the performances, the standardization, and the ease of use of host prediction approaches using CRISPR spacers. In the third chapter, the two tools were used to specifically study phage-bacteria interactions between Escherichia coli and its phages, present in the gut microbiota. CRISPR characterization allowed us to uncover the probable role of the CRISPR-Cas system in this species, being an anti-prophage mechanism. To my knowledge, this is the first study that identified a large proportion of spacer targets, by finding the origin of 60% of the protospacers.

<p dir="ltr">Anti-CRISPR (Acr) proteins are produced by phages to deactivate CRISPR–Cas systems in bacteria and archaea, thus expanding the CRISPR toolbox for genome editing. In this study, we present the structure and function of AcrIF24, an Acr protein that inhibits the type I-F CRISPR–Cas system in Pseudomonas aeruginosa. AcrIF24 forms a homodimer that binds to two surveillance complexes (Csy), preventing CRISPR RNA from hybridizing with target DNA. Additionally, AcrIF24 acts as an anti-CRISPR-associated (Aca) protein, suppressing the transcription of the acrIF23-acrIF24 operon. Whether alone or in complex with Csy, AcrIF24 binds to the acrIF23-acrIF24 promoter DNA with nanomolar affinity. The 2.7 Å structure of the Csy–AcrIF24–promoter DNA complex reveals how transcriptional suppression occurs. Our findings demonstrate that AcrIF24 functions as an Acr-Aca fusion protein and enhance our understanding of the varied mechanisms employed by Acr proteins.</p><p dir="ltr">In the ongoing evolutionary struggle between bacteria and bacteriophages, the emergence of CRISPR and anti-CRISPR systems has shaped host-pathogen interactions significantly. Bacteriophages exert intense selective pressure on bacteria, driving the evolution of defense mechanisms such as restriction enzymes and the CRISPR-Cas system. Conversely, bacteriophages have evolved anti-CRISPR proteins (Acrs) to counteract CRISPR-Cas-mediated targeting. Here, we investigate the interactions and regulatory mechanisms within co-encoded Acrs, focusing on AcrVA1-5 from a prophage within Moraxella bovoculi. Our findings reveal that AcrVA1 and AcrVA2 form a stable complex capable of inhibiting Cas12a-mediated DNA cleavage, with AcrVA1 regulating the activity of AcrVA2. Additionally, AcrVA4 and AcrVA5 form a complex that modulates Cas12a activity by inhibiting DNA binding and lysin acetylation, respectively. Structural and biochemical analyses uncover a complex regulatory network governing the function of co-encoded Acrs, highlighting their role in downregulating DNA targeting in response to Cas12a presence and aiding the survival of both phage and host bacteria during infection.</p>

Myoclonus dystonia (MD) is a rare autosomal-dominant combined dystonia movement disorder characterised by quick, involuntary muscle jerks (myoclonus) paired with sustained muscular contraction (dystonia). Although known to be genetically heterogeneous, the most common genetic factor is mutations within SGCE, the gene encoding ε-sarcoglycan, accounting for approximately 45% of cases. Previous linkage analyses conducted on a family displaying inherited MD without SGCE mutations lead to the identification of another critical region, DYT15. Preliminary data suggested that mutations within the long non-coding RNA (lncRNA) CD108131, found within the DYT15 locus, resulted in decreased expression of both the SGCE transcript, as well as the SGCE protein. Validation of the remaining variants of interest yielded no new candidate genes. A low coverage area coinciding with the entire sequence of TMEM200C was discovered, however subsequent sequencing data revealed no potential disease-causing variants. Therefore, to further characterise the relationship between CD108131 and SGCE suggested by the preliminary data, a CRISPR-Cas9 knockout was developed in HEK293 cells using a double-cut strategy that allowed for complete excision of the CD108131 gene. Stable CD108131 knockout mutant cell lines were examined for differences in gene expression. QRT-PCR analysis was conducted and revealed a significant decrease in SGCE expression in the absence of CD108131. Additionally, expression also trended towards a decrease for ZBTB14, however ARHGAP28 and RPPH1 were not significantly altered. This data demonstrates that the lncRNA CD108131 is likely to have a regulatory effect on SGCE, and perhaps ZBTB14, transcription.

Recent whole-exome sequencing studies of hundreds of high-risk neuroblastoma (hNB) patients have identified Anaplastic Lymphoma Kinase (ALK) as the only directly ‘druggable’ target with a significant mutation rate (9%). ALK is a receptor tyrosine kinase whose dysregulation has been implicated as the driver lesion in a variety of cancer types, including Non-Small Cell Lung Cancer (NSCLC) and various paediatric malignancies. As a kinase normally only expressed during early development in the foetal brain, ALK is an ideal therapeutic target and it has proven relatively simple to target therapeutically. However, resistance to ALK-targeted therapy, particularly in ALK+ NSCLC patients has frequently been observed. The majority of the acquired resistance mechanisms noted in NSCLC patients rely on bypass signalling pathways, which are tissue-context dependent. To proactively identify and develop strategies to counter these varied yet expected resistance mechanisms in other ALK-driven tumours, we must gain a better insight of the bypass-track mechanism(s) in a tumour-specific manner. The present study aimed to functionally identify putative resistance mechanisms against ALK inhibitors via extensive CRISPR/Cas9-based genome-wide knockout (GeCKO) or overexpression screens (SAM) in the human neuroblastoma cell line, SHSY-5Y, to develop novel therapeutic strategies for ALK mutant NBs. The GeCKO screen identified a total of 39 genes and miRNAs, and the SAM overexpression screen identified 25 genes that induce resistance to ALK inhibitors. These putative resistance-inducing candidates were then aligned with a publicly available expression dataset of hNB patients (n = 476) to identify those with prognostic significance (Kaplan-Meier event-free survival analysis), specifically those that are indicative of relapse risk. Furthermore, all candidates identified from the screen were individually validated in vitro. Two of the candidates, one from each of the knockout and overexpression screens, were further investigated. Inhibition of hsa-miR-1304-5p, identified from GeCKO screen, induced resistance to ALK inhibitors. Interestingly, interference of has-miR-1304-5p, in the absence of ALK inhibitors, also enabled enhanced cell viability whilst the transfection of its mimic led to a significant reduction of viability across 17 distinct NB cell lines. Through genome-wide cDNA microarrays, in silico predictions, and UTR-luciferase assays, this study identified hsa-miR-1304-5p to be a major regulator of the Ras/MAP Kinase pathway. Overexpression of PIM1, identified from the SAM screen, in NB cell lines induced resistance to ALK inhibitors and this phenotype could be reversed on transducing cells with RNAi against PIM1. Interestingly, inhibition of PIM1 in wild-type cell lines via RNAi or pharmacological compounds led to substantially enhanced potency of ALK inhibitors suggesting PIM1 inhibitors as combinatorial agents with ALK inhibitors for the therapy of treatment-naive hNB. Through protein analysis of all identified downstream targets of PIM1, this study revealed NB-specific actions of the PIM1 oncoprotein that include the inactivation of the pro-apoptotic protein BAD. In summary, this study has identified mechanisms of resistance to ALK inhibitors as well as novel front-line therapeutic strategies for hNB patients that should be implemented clinically.

CRISPR/Cas9 is the gene editing tool of choice in basic research and poised to become one in clinical context. However, current studies on the topic suffer from a number of shortcomings. Mutagenesis is often assessed using bulk methods, which means rare events go undetected, unresolved or are discarded as potential sequencing errors. Many of the genotyping methods rely on short-range PCR, which excludes larger structural variants. Other methods, such as FISH, do not provide basepair resolution, making the genotype assessment imprecise. Furthermore, it is not well understood how Cas9 delivery format influences the dynamics of indel introduction. Finally, many studies of on-target activity were conducted in cancerous cell lines, which do not accurately model the mutagenesis of normal cells in the therapeutic context. In my thesis, I have investigated on-target lesions induced by Cas9 complexed with single gRNAs and no exogenous template. I have followed the time dynamics of Cas9-induced small indels as a function of reagent delivery methods, established an assay for quantification of Cas9-induced genomic lesions that are not small indels ("complex lesions") and used this assay to isolate and genotype complex lesions, many of which would be missed by standard methods. I found that DNA breaks introduced by single guide RNAs frequently resolved into deletions extending over many kilobases. Furthermore, lesions distal to the cut site and cross-over events were identified. Frequent and extensive DNA damage in mitotically active cells caused by CRISPR/Cas9 editing may have pathogenic consequences.

In adult mammals, the gastrointestinal (GI) epithelium exhibits the highest turnover rate among the endodermal tissues. The harsh luminal environment of the GI tract necessitates replenishment of epithelial cells to maintain organ structure and function during routine turnover and injury repair. This delicate balance between gain and loss of cells is called tissue homeostasis, and multipotent tissue specific adult stem cells serve as the continuous source of self-renewal. Due to their important contribution to homeostatic maintenance the proliferative capacity of the stem cells needs to be tightly controlled, as an imbalance can result in diverse pathologies such as cancer or insufficient injury repair. Despite the crucial role for regulatory processes the molecular mechanisms and the genes governing these processes remain poorly understood. Rnf43 and its paralogue Znrf3 (RZ) act as tumour suppressors in the intestine, but their role in the gastric epithelium has not been previously investigated. Using a novel unpublished stomach specific CreERT2 expressing mouse line I found that simultaneous knockout of RZ (RZ DKO) result in gastric hyperplasia of the corpus epithelium. Gastric RZ DKO organoids show independence from the essential growth factor Rspondin-1 but require exogenous Wnt. A similar exogenous Wnt dependence was identified in a human gastric cancer cell line harbouring homozygous Rnf43 inactivating mutations. Thus, Wnt secretion inhibition might provide a new treatment paradigm for a subset of patients carrying Rnf43 mutations. The prominent role of the E3s Rnf43 and Znrf3 in the intestinal and gastric epithelial led to the question of whether other E3s either closely related to RZ or specifically expressed in stem or niche cells could play a role in homeostatic regulation, specifically in the small intestine. Using a retroviral overexpression screen I identified Rnf24 and Rnf122, two E3s that rendered intestinal organoids insensitive to withdrawal of the BMP inhibitor Noggin. Moreover, potential substrate candidates located at the cell surface membrane were identified and the generation of in vivo models initiated to provide a basis for further studies investigating the role of these E3s. In trying to address the function of the abovementioned genes using in vitro functional genetics I identified gaps in the current technology for organoid genetic engineering. I therefore developed two gene editing methods; a gRNA concatemer system allowing simultaneous knockout of multiple genes and CRISPR-FLIP enabling generation of conditional gene knockouts In summary, this thesis describes the first stomach specific knockout of Rnf43 and Znrf3 in the gastric epithelium, showing that it results in gastric hyperplasia located to the corpus epithelium. The dependence of the Rnf43 and Znrf3 knockout epithelium on exogenous Wnt signalling provides a potential treatment strategy for a subset of patients harbouring Rnf43 mutations. Next, it identifies Rnf24 and Rnf122 as E3 ubiquitin ligases involved in intestinal stem cell regulation and provide preliminary data and a basis for future studies. Finally, it describes the establishment of two advanced genetic engineering approaches which can be applied to various in vitro culture systems such as 3D organoids, mouse embryonic stem cells and conventional cell lines. Collectively this work and the developed methods will contribute to our understanding of the mechanisms regulating adult stem cell homeostasis.

Cilia are polarized extensions of the cells microtubule-based cytoskeleton dedicated to sensory, signaling and motility-related functions. In mammals, there are two main types of cilia, immotile and motile, where motile cilia generate/modulate fluid flow at the embryonic node, in respiratory airways, cerebral ventricles and the oviduct in addition to sperm propulsion via the flagellum. Defects in cilia motility cause a rare genetic disorder called Primary Ciliary Dyskinesia (PCD). In this thesis, I present functional and molecular characterisation of two PCD causing genes HEATR2 and ZMYND10. Core cilia genes are transcriptionally activated by members of the winged-helix transcription factors of the RFX family. The forkhead transcription factor FOXJ1, additionally activates motility genes such as the ones encoding components of axonemal dynein motors which transfer the chemical energy released from ATP hydrolysis to kinetic motion necessary for ciliary motility. I present data in this thesis which show that Heatr2 and Zmynd10 are both targets of the RFX3-FOXJ1 transcriptional module which co-operatively switches on genes required to make motile cilia Mutations in both HEATR2 and ZMYND10 cause the same subtype of PCD (loss of inner and outer arm dyneins in cilia). I characterise a human PCD causing mutation in HEATR2 in this thesis. Additionally, using genetic null mouse models generated using the CRISPR technology, I describe the phenotypic effects of complete loss of Zmynd10 in mice. Zmynd10 mutant mice display characteristic PCD-like features. Adding to my functional studies, I present proteomic data to propose mechanisms by which HEATR2 and ZMYND10 proteins control cilia motility. Mass spectrometry and protein interaction studies support distinct roles for HEATR2 and ZMYND10 in intracellular transport and pre-assembly of axonemal dynein motors. The multisystem approaches described in this thesis to characterise the roles of HEATR2 and ZMYND10 highlight the molecular complexity underlying the assembly and delivery of axonemal dyneins to motile cilia and provide novel functional and molecular insights into the pathophysiology of PCD.

In 2006, Kazutoshi Takahashi and Shinya Yamanaka demonstrated the ability of four transcription factors; Oct4, Sox2, Klf4 and c-Myc to 'reprogram' differentiated somatic cells to a pluripotent state. This technology holds huge potential in the field of regenerative medicine, but reprogramming also a model system by which to the common regulators of all forced cell identity changes, for example, transdifferentiation. Despite this, the mechanism underlying reprogramming remains poorly understood and the efficiency of induced pluripotent stem cell (iPSC) generation, inefficient. One powerful method for elucidating the gene components influencing a biological process, such as reprogramming, is screening for a phenotype of interest using genome-wide mutant libraries. Historically, large-scale knockout screens have been challenging to perform in diploid mammalian genomes, while other screening technologies such as RNAi can be disadvantaged by variable knockdown of target transcripts and off-target effects. Components of clustered regularly interspaced short palindromic repeats and associated Cas proteins (CRISPR-Cas) prokaryote adaptive immunity systems have recently been adapted to edit genomic sequences at high efficiency in mammalian systems. Furthermore, the application of CRISPR-Cas components to perform proofof- principle genome-wide KO screens has been successfully demonstrated. I have utilised the CRISPR-Cas9 system to perform genome-wide loss-of-function screening in the context of murine iPSC reprogramming, identifying 18 novel inhibitors of reprogramming, in addition to four known inhibitors, Trp53, Cdkn1a, Jun, Dot1l and Gtf2i. Understanding how these novel reprogramming roadblocks function to inhibit the reprogramming process will provide insight into the molecular mechanisms underpinning forced cell identity changes.

Eukaryotes use proteins to carefully package and compact their genomes to fit into the nuclei of their individual cells. Nucleosomes are the primary level of compaction. Nucleosomes are formed when DNA wraps around an octamer of histone proteins and a nucleosome's position can limit access to genetic regulatory elements. Therefore, nucleosomes represent a basic level of gene regulation. DNA and its associated proteins, called chromatin, is usually classified as euchromatin or heterochromatin. Euchromatin is transcriptionally active with loosely packed nucleosomes while heterochromatin is condensed with tightly packed nucleosomes and is transcriptionally silent. In order to become active, heterochromatin must first be remodeled. We have studied the effects of nucleosome positioning on transgene expression in vivo using Caenorhabditis elegans as a model. We show that both location and polarity of the DNA sequence can influence transgene expression. We also discuss some considerations for working with CRISPR/Cas9. A major reason for doing in vitro nucleosome reconstitutions is to determine the effects of DNA sequence on nucleosome formation and position. It has previously been implied that nucleosome reconstitutions are stochastic and not very reproducible. We show that nucleosome reconstitutions are highly reproducible under our reaction conditions. Our results also indicate that a minimum depth of 35X sequencing coverage be maintained for maximal gains in Pearson's correlation coefficients. Communicating science with others is an important skill for any researcher. The rising generation of scientists need mentors who can teach them how to be independent thinkers who can carry out scientific experiments and communicate their finding to others. With this goal in mind, we have devised a scaffolding pedagogical method to help transform undergraduates into confident independent thinkers and researchers.