La plupart des virus sont connus pour leur capacité à causer des maladies symptomatiques chez l’Homme et chez les autres animaux. Certains d’entre eux sont des grands virus à ADN nommés Virus à Grand ADN Nucléo-Cytoplasmique (NCLDV), rapportés comme infectant les cellules eucaryotiques. Au début du XXI ème siècle, quatre familles ont été définies par Iyer et al. comme ayant une origine commune (groupe monophylétique) : Asfarviridae, Phycodnaviridae, Irido-Ascoviridae et Poxviridae.En 2003, la description d’Acanthamoeba polyphaga mimivirus a cassé un paradigme dans le monde des virus. Par leur taille de particule (450nm), par leur longueur de génomes(supérieure à 1Mb) et leur contenu génique, leur découverte a changé la définition traditionnelle des virus (Lwoff). Depuis 2013 et notamment par les isolements successifs de Pandoravirus,Pithovirus et Mollivirus, ces virus ont été décrits comme possédant de nouvelles propriétés.Leur découverte a été rendue possible grâce à la méthode de co-culture utilisant des protistes, notamment des cellules du genre Acanthamoeba. Cette méthode a été de nombreuses fois modulée par différentes équipes. Dans notre cas, nous avons combiné différentes stratégies appliquées à notre co-culture : la co-culture a été couplée à la cytométrie en flux pour détecter la lyse des protistes. De plus, la cytométrie a été utilisée avec un marqueur à ADN dans le but d’identifier de façon putative le virus et de discriminer les différentes populations virales. Enfin,nous sommes capables de séparer ces populations en utilisant un appareil FACS trieur.L’ensemble de ces techniques a permis l’isolement de nouveaux virus. / Most viruses are known for their ability to cause symptomatic diseases in humans andother animals. Some of them are large DNA viruses named Nucleo-cytoplasmic Large DNAviruses (NCLDV), known for infecting eukaryotic cells. At the beginning of the 21st centuryfour families were defined by Iyer et al. as having a common origin (monophyletic group):Asfarviridae, Phycodnaviridae, Irido-Ascoviridae and Poxviridae.In 2003, the description of Acanthamoeba polyphaga Mimivirus broke this paradigmin the virus world. Because of their particles size (450 nm), their genome size (up to 1Mb),and their gene contents, their discovery changed the traditional definition of viruses (Lwoff).Since 2013 and the successive isolations of Pandoravirus, Pithovirus and Mollivirus; theseviruses have been characterized as possessing various novel properties.Their discoveries have been possible thanks to the co-culture method using protistnotably Acanthamoeba genus cells. This method went through multiple improvements and isemployed by different teams in different ways. In our case and in order to enhance thismethod we combined strategies applied in our co-culture. Indeed, this method consists inusing flow cytometry to detect lysis of protist cells (after all steps of co-culture enrichment).In addition, the flow cytometry was used with a DNA marker in order to identity viruses anddiscriminate viral populations. Then, we were able, using a FACS sorter device, to separatedifferent viral populations from our supernatants.Altogether these techniques have permitted the isolation of new viruses.

Les virus sont ubiquitaires et abondants dans l’environnement. Ils influent fondamentalement sur l’écologie de l’ensemble des écosystèmes et du microbiote humain. Dès 2002, avec la découverte de virus géants d’amibes, la virologie s’est complexifiée. Les Megavirales (nouvel ordre au sein des grands virus nucléocytoplasmiques) ont 10% de gènes homologues aux cellules eucaryotes, et ont la caractéristique singulière d’être infectés par des virophages.Avec l’avènement de la métagénomique, le nombre de métagénomes produits ne cesse de croître de manière exponentielle. C’est ainsi que la virologie a connu un nouvel essor et a pu mieux être étudiée en s’affranchissant des difficultés de culture et d’isolement des virus dans les conditions artificielles de laboratoire. La métagénomique permet d’étudier les communautés microbiennes mais également de découvrir de nouveaux microbes. La bioinformatique est devenue incontournable dans le domaine de la biologie et essentielle pour les biologistes afin de traiter les masses de données et en extraire toute la richesse de l’information biologique nécessaire. La première partie de cette thèse consiste en une revue de la littérature décrivant la bioinformatique au service de la métagénomique virale. La deuxième partie présente la création d’un nouvel outil « MG-Digger » dédié à l’analyse rapide et automatisée de séquences d’intérêts spécifiques dans les métagénomes. La dernière partie se concentre sur l’utilisation de cet outil sur des données issues de projets métagénomiques afin de répondre à des questions biologiques précises, notamment sur les données de l’expédition scientifique TARA à travers les océans. / Viruses are ubiquitous and abundant in the environment and can influence all ecosystems ecology and the human microbiota.Since 2002, with the discovery of giant viruses of amoeba, virology has become more complex and the definition of virus has been called into question, not only because of their phenotypic sizes similar to those of bacteria but also their genomic content exceeding thousand genes. Megavirales, also known as nucleocytoplasmic large DNA viruses, have 10% homologous genes to eukaryotic cells and interestingly can be infected by virophages. With the advent of metagenomic, the number of metagenomes produced is exponentially increasing as well as our understanding of virology which has been studied. Metagenomics studies showed an efficient way to study microbial communities and identify novel viruses without the difficulties of culture and isolation of viruses in artificial laboratory conditions.Metagenomic requires considerable computational and storage resources (Big data processing). Therefore, bioinformatics has become an integral part of research and development in the biomedical sciences by providing tools that handle complex datasets and finally giving the necessary biological information. The first part of this thesis consists of an exhaustive review of the literature describing bioinformatics and viral metagenomics. The second part presents a new "MG-Digger" tool dedicated to the rapid and automated analysis of specific interest sequences in metagenomes. The third part focuses on the use of this tool on metagenomic data to answer to specific biological questions, including data from the TARA scientific expedition across the oceans.

Marseilleviridae est une famille de virus géants isolés initialement à partir de prélèvements environnementaux, dont Marseillevirus est le membre fondateur. La présence des marseillevirus chez l’Homme a été démontrée dans quelques études. Les objectifs sont de mieux documenter la présence des marseillevirus chez l’Homme, de modéliser l’infection par Marseillevirus chez la souris, et enfin, de décrire les génomes des marseillevirus. Nous rapportons un cas d’infection par Marseillevirus chez une patiente atteinte d’un cancer des ganglions, soulevant la question d’un éventuel lien entre Marseillevirus et cancer, à l’instar de l’association existant entre d’autres virus et les cancers. L’infection des souris par Marseillevirus montre que celui-ci persiste un mois au niveau des «amygdales», confirmant le portage pharyngé chronique observé chez un deuxième patient. Enfin, nous identifions deux nouveaux groupes au sein de la famille, soulignant l’importante diversité génétique de la famille. / Marseilleviridae is a new family of giant viruses primarily isolated from environmental samples and whose Marseillevirus is the founding member. The presence of marseilleviruses in humans has been demonstrated in few studies. The aims are to better document the presence of marseilleviruses in humans, to develop a model of infection of mice with Marseillevirus, and to describe the genomes of marseilleviruses. We report a first caes of infection by Marseillevirus in apatient with a lymph nodes cancer, raising the question of a potential link between Marseillevirus and cancer, as the well established association between some viruses and cancers. The infection of miceshows that Marseillevirus persist one month in the “tonsils”, confirming the chronic pharyngeal carriage reported in a second patient. Finally, we identify two new subgroups in the family, highlighting the considerable genetic diversity of the family.

Since their 'discovery' at the turn of the 21st century, giant viruses of the amoeba have captured the fascination of virologists. They have raised a plethora of questions regarding their evolution and ecological significance and have greatly defied a century's old definition of viruses. By now, it is understood that a handful of giant viruses enter the amoeba via the phagosomal pathway. This thesis chooses to focus on the giant Melbournevirus (MelV) regarding its entry and uncoating pathway. We now conclude that the initial attachment between MelV and amoeba cells is built upon glycan interactions based on evidence that mannose competitively inhibits MelV binding. This attachment likely entails an approximately 70 kDa mannose containing glycoprotein on the MelV. Mannose and other glycans induce secretion of proteins including phagosomal enzymes from amoeba. Based on these findings, it is hypothesised that the mannose-induced phagosomal enzymes could play a role in the uncoating of the MelV. The results further reveal isolated phagosomes, also to some extent the glycan-induced protein secretions, to have high levels of proteins involved in cell redox homeostasis. This implies that the highly oxidative environment of the phagosome may be an overlooked physiological factor when regarding the uncoating of the MelV. A deeper understanding of the physiological uncoating conditions can be used for studying internal structures of giant viruses, such as the enigmatic Large and Dense Body (LDB) of the MelV particle.

Viruses hold our attention for the horrific impact they have on human health and welfare. However, viruses are a critical part of our ecosystem and facilitate the cycling of carbon and other important nutrients. The cycle of virus infection, followed by host resistance and the subsequent evolution of new strains enables adaptation to changing hosts and the environment. Giant viruses, some with particle sizes large enough to be visible in light microscopes and their bewildering array of accessory genes, have captivated scientists and the general public since their discovery two decades ago. Giant viruses are part of the Nucleocytoviricota (NCV) whose members include both harmful agents (such as the causative agents of smallpox and swine hemorrhagic fever) and beneficial ones (such as those that provide biocontrol of insects, mitigation of toxic algal blooms and enzymes for biotechnology). Most of the giant viruses discovered, to date, are from marine and freshwater ecosystems where their hosts are abundant. In terrestrial soils, very few giant viruses have been revealed because of challenges in shifting through the astounding microbial genetic diversity in soil to assemble genomes from metagenomic data. Currently there is a lack of knowledge about abundance and genetic diversity of giant viruses in terrestrial soils, knowledge about their hosts and their influence on biogeochemical cycling. In 2018, giant viruses were discovered in the Barre Woods experimental warming plots at Harvard Forest. (Schulz et al 2018 Nature Communications). A novel environmental genomics approach involving filtration and fluorescence activating cell-sorting (FACS) was used to discover 16 Nucleocytoviricota (NCVs) in just a few grams of Harvard Forest soil. All these newly discovered viruses represent distinct lineages (new species, genera, and families). This experiment involved just two soil cores (1 warming and 1 control) and a single time point at Harvard Forest. There is much to learn about the terrestrial giant virus genetic biodiversity as these same viruses have not yet been discovered at other sites around the world. My research will focus on a genus of giant viruses with only three known representatives, all from Harvard Forest. They are Hyperionvirus (with the world’s 2nd largest virus genome at 2.4 MBp), Terrestrivrus the 10 th largest genome at 1.8 MBp), and Harvfovirus (the 15 th largest genome at 1.6 MBp). In the experimental warming plots the relative amount of bacteria to fungi has increased. We hypothesize that the relative increase in bacteria has led to an increase in protists, which feed on the bacteria, which in turn has led to an increase in giant viruses, which infect the protists. Because of the high genetic diversity in viruses and the lack of ribosomal genes, it is not possible to create primers that span the entire Nucleocytoviricota phylum or even at the family level. To test our hypothesis, we designed degenerate PCR primers that detect and quantify members of the genus containing the 3 giant Harvard Forest viruses. DNA was extracted from soil samples the soil (stored at -80C) from the 2017 temperature toggle experiment at Barre Woods in which the power to the warmed plots was turned off from late May until early September were used. The giant viruses were originally discovered in the sample just prior to turning off the power. We used 4 time points spanning the experiment with 8 samples from each the warmed and control plots (4 x 16 = 16 samples total). The primers were designed based on five hallmark genes that are present in most members of the Nucleocytoviricota. After amplification, the amount of DNA would be quantified and normalized. We expect to better understand the genetic diversity of this genus of giant viruses in the soil including the possibility of detecting new species in this genus.

The discovery of large DNA viruses has challenged the traditional perception of viral complexity due to their enormous genome size and physical dimensions. Previously, viruses were considered small, filterable agents until the discovery of large DNA viruses. Among large DNA viruses, the phylum Nucleocytoviricota and its members, which are often called "giant viruses" have large genome sizes (up to 2.5 Mbp) and virion sizes (up to 1.5 um). Due to having large virion and genome sizes, these viruses were often excluded from viral surveys and remained understudied for years. Luckily, the advancement of metagenomic analysis has facilitated the study of large DNA viruses by analyzing them directly from their environment without cultivating them in the lab, which could be challenging for viruses. In the first chapter of the thesis, I investigated 11 metagenome-assembled genomes (MAGs) of giant viruses previously surveyed from Station ALOHA in the Pacific Ocean. St. ALOHA is located near Hawaii and represents oligotrophic gyres which the majority of the ocean is made of them. I focused on 11 MAGs of giant viruses to get insight into their phylogenetic characteristics, genomic repertoire, and global distribution patterns. Despite the fact that metagenomic analysis has facilitated the study of genetic materials of microbes and viruses on a huge scale, it is essential to benchmark the performance of metagenomic tools and understand the associated biases, particularly in viral metagenomics. In the second chapter, I evaluated the performance of metagenomic tools (contigs assembler and binning tool) in recovering viral genomes using annotated dataset. We used a metagenome simulator (CAMISIM) to generate simulated short reads with known composition to assess these processes. Moreover, I emphasized the importance of binning contigs for viral genomes to fully recover the genomes of viruses along with discussing how diversity metrics were differed for contigs, bins populations. / Master of Science / Viruses are generally thought to be small biological agents with small genome (genetic material) sizes and tiny physical structures; for instance, the genome length of a Human Immunodeficiency Virus (HIV) is around 10 kilobase pair (a unit for measuring genetic material in an organism), and the virion size (physical dimension of a virus) can go up to 120 nm. The discovery of large DNA viruses has challenged the idea of considering viruses as small biological entities, as their genome sizes and physical dimensions can be up to 2.5 megabase pairs and 1500 nm, respectively. Famous members of large DNA viruses from the phylum Nucleocytoviricota are often known as "Giant Viruses'' because they have enormous genome sizes and physical dimensions. Due to having large viral particles, these viruses may usually be excluded from viral surveys. For instance, in field studies, samples must be filtered through a fraction (e.g., 0.2 um) to eliminate bacterial and archaeal genomes and cellular debris, which also results in excluding larger viruses. Since these viruses remain understudied for several years because of biases associated with having large viral particles, there is a solid need to discover and investigate more about them. Growing and cultivating viruses in the laboratory may be challenging, as they need specific hosts to be dependent on to produce more viral progeny and some specific laboratory environments. Luckily, with the advancement of biotechnology, scientists could find ways to evade the need for cultivating viruses in the lab and study them with computational tools such as metagenomic analysis and bioinformatic tools. Metagenomics analysis helps to study the genetic materials of microbial or viral populations directly from their habitat without growing them in a laboratory. In short, metagenomic analysis has multiple steps, including collecting and filtering samples, fragmenting DNA within the samples, generating short DNA sequences (short-read sequences) with NGS (Next Generation Sequencing) technology, assembling short-read sequences into large DNA fragments which can be contigs (contiguous DNA fragments) and metagenome-assembled genome (MAGs). With metagenomic analysis, we can recover the genome of multiple organisms, and we name the recovered genome as metagenome-assembled genome (MAGs) as it is generated through metagenomic processes. The metagenomic analysis will allow us to study microbes and viruses in their environment and gain insight into their taxonomic details, genomic content, and how widespread they are. In the first chapter, I studied 11 MAGs of giant viruses previously surveyed from St. ALOHA, Hawaii. St. ALOHA is a good field site for examining microbial processes and diversity and a good representative of oligotrophic waters (low in nutrients). I examined 11 MAGs of giant viruses to investigate their taxonomic characteristics to clarify which order they belong to within their phylum, their genomic content, and their global distribution pattern. Although studies have successfully recovered the genome of large DNA viruses from their habitats and then analyzed them, all these metagenomic processes need to be evaluated so the results will be valid to consider as the genome of our interested organisms. In the second chapter, I developed a workflow for viral metagenomic analysis to assess metagenomic tools' performance in recovering reliable viral genomes, particularly for large DNA viruses. Most of these benchmarking workflows are done for bacterial and archaeal genomes, and in this thesis, I used these metagenomic tools and applied them to recover large DNA viruses genomes. Also, I emphasized the importance of using binning tools to fully recover large DNA viruses genomes, as due to their large genome size, their genomes might remain fragmented into different contigs, which are longer sequences than reads but shorter than MAGs.

Nucleocytoplasmic large DNA viruses (NCLDVs) have genome sizes that range from around 100 kilobases (kb) to up to 2.5 megabases, and virion sizes that can reach up to 1.5 μm. Their large size in both of these contexts is atypical and defies the traditional view that viruses are streamlined, "filterable infectious agents". NCLDVs include many diverse groups, including Poxviruses, Asfarviruses, Iridoviruses, Mimiviruses, and Marseilleviruses. Poxviruses are perhaps the most well-studied; these viruses have 135-360 kbp genomes with about half of the genes encoding essential replication genes and the other half encoding genes related to host-virus interactions. Many of the genes involved in host-virus interactions are involved in immunomodulatory processes and have homology to proteins encoded by the host. These viral genes, often referred to as "mimics", are therefore believed to be the result of host-to-virus gene transfer. In this study I sought to examine if common poxvirus immunomodulatory genes were found in other NCLDV lineages, and if so, to analyze the evolutionary history of these genes. I identified 5 protein families of immunomodulatory genes that were found in both poxviruses and other NCLDV lineages, and I used phylogenetic tools to compare viral immunomodulatory genes of NCLDVs to their eukaryotic orthologs to evaluate the number of times different NCLDV lineages have acquired these genes. Our phylogenetic analyses showed that several viral immunomodulatory genes were acquired multiple times by different NCLDV lineages, while others appear to have been transferred between viral groups. Interestingly, some NCLDV genes clustered together with homologs from the unrelated Herpesviridae family, suggesting that inter-viral gene exchange can traverse vast evolutionary distances. The vast diversity of hosts infected by different NCLDV lineages suggests that these immunomodulatory genes play key roles that are useful to viruses in a variety of contexts. This research provides insight into how giant viruses acquire host genes, which contribute to their large genome size, and how those genes evolved to subvert antiviral defenses. / Master of Science / Giant viruses are a relatively recent discovery, from the beginning of this century. Nucleocytoplasmic large DNA viruses (NCLDVs) are a classification of multiple giant virus families. These viruses have large genomes from around 100 kilobases to 2.5 megabases of DNA. For reference, the genome size of the flu virus is approximately 13 kilobases. Most viruses cannot be seen by the human eye, even with microscopes, but giant viruses can get as big as bacteria, which can be seen with microscopes. It is unknown how or why these viruses get so large. One explanation is that they steal genes from their host and those genes evolve to work against the host. In this thesis, I explored some of the genes that these viruses have picked up. I curated a set of 49 previously characterized viral genes to analyze in this context. These genes have to do with modulating the host immune system and are known as "immunomodulatory genes". Viral immunomodulatory genes are often mimics of the host genes which function to help the immune system. However, a virus evolves faster than a host and the virus mimic gene can evolve to work against the immune system. This change can be visualized using phylogenetic tools; the viral genes will be more similar to each other than to the host genes and cluster separately on a phylogenetic tree. About half of the genes of Poxviruses, a giant virus family that has viruses that infect humans, are related to virus-host interactions, and include viral mimic genes. Poxviruses have been far better studied than other NCLDV families because of their public health importance. Variola virus, the virus that causes smallpox, is a poxvirus. Other NCLDV infect animals, algae, and amoeba. Though their hosts are different, their genomes have similar features. I set out to discover whether some of these previously characterized viral immunomodulatory genes that exist in poxviruses also exist in other NCLDV families. I utilized phylogenetic tools and a database of giant virus sequences to figure out which genes are being picked up by which family of NCLDV. I also sought to determine whether the individual NCLDV families have their own acquired immunomodulatory gene or have a gene very similar to all other families, suggesting an ancient acquisition. If the gene is very similar, it suggests that an ancestor of the NCLDV acquired the gene and it has stuck around as the group diverged into families. It is also interesting if different families stole the same type of gene multiple times because that indicates the importance of that gene in subverting the antiviral immune system for viral replication. This work provides insight into how giant viruses acquire host genes, which contribute to their large genome size, and how they evolved those genes to subvert antiviral defenses.

Les virus sont présents dans tous les écosystèmes, et sont les entités les plus abondantes dans le milieu marin. Bien que nous associons systématiquement virus aux maladies, la plupart d'entre eux coexistent cependant en équilibre avec leur hôte. Les virus sont associés à tous les règnes de la vie, même les virus qui affectent d'autres virus(virophages). La définition aujourd'hui d'un virus chez les virologues, c'est qu'un virus est un parasite génétique qui utilise des systèmes cellulaires pour sa propre réplication. Les hôtes les plus couramment utilisés par les virus que nous avons étudiés sont principalement des protozoaires. Ainsi, les Amoebozoa font l'objet de nombreuses études et sont utilisés pour isoler de nouvelles espèces intracellulaires( virus, bactéries). Ces espèces ont évolué de manière à résister aux effets consécutifs à la phagocytose ou à l'ingestion dans des vacuoles, et restent viable dans le cytoplasme de l'amibe, et ont le potentiel de se multiplier dans les parasites. Dans cette étude, nous avons dans un premier temps étudier les diverses interactions existantes entre virus Acanthamoeba polypaghaga Mimivirus(APMV) et des bactéries au sein de l'amibe. Pour cela, nous avons choisi un système original basé sur la co-culture de l'APMV, soit seul ou en combinaison avec deux autres microorganismes isolés individuellement à partir de l'amibe. Il s'agit d'une bactérie intracellulaire stricte(BABL1) et le virophage de APMV (Sputnik). Cela nous a permis de mettre en évidence, d'une part la capacité du virophage à moduler la virulence d'APMV tout en révélant, d'autre part, la bataille qui a eu lieu entre eux au cours de l'infection de l'hôte. dans un deuxième temps, nous avons examiné l'activité virucide des biocides couramment utilisés en pratique clinique pour la désinfection des équipements hospitaliers. APMV et Marseillevirus montrent une grande résitance aux biocides chimiques, en particulier l'alcool. Seule la température de 75°C et le glutaraldéhyde ont réussi à réduire les titres d'APMV et Marseillevirus à des niveaux indétectables. Après dessiccation ou exposition aux rayonnements ultraviolets, APMV et marseillevirus ont démontré leur stabilité durable. Précédent le pré-traitement des échantillons de l'environnement par l'éthanol à 70°C, a permis la disparition des contaminants bactériens sans réduire la charge virale, permettant leur isolement sur amibe, sans avoir besoin d'utiliser des antibiotiques, qui peuvent avoir un effet délétère su les amibes. / In this study, we first examined the various interactions taking place between the virus Acanthamoeba polyphaga Mimivirus (APMV) and bacteria within the amoeba. We chose an original system based on a co-culture of APMV either alone or in combination with two other organisms isolated from amoeba, i.e a strict intracellular bacterium (BABL1) and the virophage of APMV (Sputnik). This allowed us to highlight, on the one hand, the possibility to modulate the virulence of APMV while revealing, on the other hand, the battle which occurs between them during the infection of the host. We then examined the virucidal activity of biocides commonly used in clinical practice for the disinfection of hospital equipment. APMV and Marseillevirus show high resistance to chemical biocides, especially to alcohol. Only a temperature of 75°C or glutaraldehyde were able to reduce APMV and Marseillevirus titres to undetectable levels. Whether dried or under ultraviolet, APMV and Marseillevirus demonstrated their lasting stability. Previous pre-treatment of environmental samples by ethanol 70° allowed disappearance of bacterial contaminating bacteria without reducing giant virus load allowing their isolation on amoeba without need the use of antibiotic that may have a deleterious effect on amoebae.

Cette thèse menée à l'interface entre deux écoles doctorales, a permis d'étudier la diversité virale et de proposer l'utilisation de biomarqueurs pour répondre à des questions environnementales. La première partie présente l'étude et la caractérisation de familles de virus géants isolées au laboratoire Information Génomique et Structurale. La première, les Pandoravirus, découverte il y a 2 ans, remet en question la définition de virus, leur origine et leur mode d'évolution. Ce virus possède un génome dépassant 2,5 Mb codant pour plus de 2500 protéines, dont 90% complètement inédites. Une autre famille de virus, les Marseilleviridae, m'a permis de mesurer leur vitesse d'évolution. Les particularités de cette famille sont (i) des génomes extrêmement conservés et (ii) la séparation de la famille en différentes lignées en fonction de leur pourcentage d'identité nucléotidique. J'ai pu étudier l'évolution de cette famille, montrant que la majorité des gènes de ces virus sont conservés entre les lignées et sous pression de sélection, donc nécessaires à leur réplication. La deuxième partie avait pour but de valider l'utilisation de marqueurs biologiques afin de répondre à une question environnementale. Il s'agissait d'identifier l'origine de l'eau saumâtre d'une rivière souterraine. Pour cela nous avons recensé les communautés de procaryotes de différentes sources d'eau douce et d'eau de mer dans le but de les comparer aux communautés de la rivière souterraine. Cette démarche nous a permis de conclure sur l'origine à la fois marine et terrestre de l'eau de l'exsurgence souterraine. Des hypothèses sur le mode d'acheminement de ces eaux vers l'exsurgence sont également proposées. / This thesis which takes place between two doctoral schools, allowed us to study viral diversity and to suggest the use of a biomarker to answer environmental questions. The first part presents a work on the characterization of two giant virus families isolated in the IGS laboratory. The first family, Pandoraviruses, questions the definition of virus, their origin and the way they evolve. This virus’ genome is bigger than 2,5 Mb, which codes for more than 2,500 proteins, of which 90% were unknown before. The isolation of new members of this family could allow us to study their evolution. Another virus family, Marseilleviridae, allowed me to study their evolution speed. This family features are (i) to have highly conserved genomes and (ii) the family is separated in three lineages, according to their nucleotide identity percentage. I thus studied the evolution of this family, showing that most genes of these viruses are conserved between lineages and under selection pressure, therefore necessary for their replication. The second part describes a work to validate the use of a biologic marker in order to respond an environmental question. We tried to identify the origin of underground river brackish water flowing from a submarine karstic spring in Port-Mio, Cassis. For that we initiated a comparison of prokaryotic community from various springs of fresh water, sea water, and the underground river. This method, coordinated to biogeochemical data allowed us to identify biomarkers which are specific of each sample. We can then conclude that this brackish water finds an origin in both fresh and sea water. Hypothesis on the way these waters flows in the spring are also proposed.

La découverte des virus géants il y a une dizaine d’années a véritablement bouleversé notre perception du monde viral. Cette découverte a ouvert un débat sur l’origine et l’histoire évolutive de ces virus, et ravivé celui portant sur la nature des virus : peuvent-ils être considérés comme vivants ?J'ai caractérisé un nouveau Marseilleviridae, isolé à partir d’un échantillon prélevé en Nouvelle-Calédonie, appelé Noumeavirus. Les Marseilleviridae sont des grands virus à ADN, qui possèdent des particules à symétrie icosaédrique d’environ 200 nm de diamètre, et des génomes à ADN double-brin de plus de 300 kb. Différentes approches de génomique, protéomique, ainsi que l’étude du cycle infectieux ont montré que le cycle infectieux de ces virus n’était pas indépendant du noyau cellulaire mais recrutait transitoirement les fonctions nucléaires à l’usine virale.J'ai également caractérisé un nouveau Pandoravirus, isolé à partir d’un échantillon prélevé en Nouvelle-Calédonie, appelé Pandoravirus neocaledonia. Les Pandoravirus possèdent une morphologie unique au sein des virus, ainsi qu’un génome colossal à ADN double-brin de 2.5 Mb. Des études comparatives avec d’autres Pandoravirus ont été réalisées en combinant plusieurs approches, afin de mieux comprendre les caractéristiques de ces virus inédits. La morphologie étonnante de ces virus nous a poussés à étudier la nature de leur enveloppe, constituée d’un réseau de fibres formant des structures lamellaires. Se pourrait-il que les Pandoravirus, contrairement aux autres virus, détournent la machinerie cellulaire pour construire leurs particules ? Dans ce cas, quelle serait leur place dans l’histoire évolutive des virus ? / The discovery of giant viruses about a decade ago has truly shaken our perception of the viral world. This discovery has initiated a debate on the origin and evolutionary history of these viruses, and it revived the debate on their nature: are viruses alive?I characterized a new Marseilleviridae, isolated from a sample collected in New-Caledonia, named Noumeavirus. The Marseilleviridae are large DNA viruses that have icosahedral particles of about 200 nm in diameter, and double-stranded DNA genomes of more than 300 kb. Various approaches, such as genomics, proteomics and the study of the infectious cycle, allowed us to reveal that the infectious cycle of these viruses was not independent from the cell nucleus as we thought, but was transiently recruiting nuclear functions to the viral factory.I also characterized a new Pandoravirus, isolated from a sample collected in New-Caledonia, named Pandoravirus neocaledonia. Pandoraviruses have a unique morphology and a gigantic double-stranded DNA genome of about 2.5 Mb. Comparative studies with other Pandoraviruses were performed using several approaches to better understand the characteristics of these original viruses. The astonishing morphology of these viruses led us to investigate the nature of their envelope, which is made of a mesh of fibers forming lamellar structures. Is it possible that Pandoraviruses, unlike other viruses, hijack the cellular machinery to build their particles? In this case, what would be their place in the evolutionary history of viruses?