This dissertation presents applications of machine learning and statistical approaches to infer protein-DNA bindings in the presence of epigenetic modifications. Epigenetic modifications are alterations to the DNA resulting in gene expression regulation where the structure of the DNA remains unaltered. It is a heritable and reversible modification and often involves addition or deletion of certain chemical compounds to the DNA. Histone modification is an epigenetic change that involves alteration of the histone proteins – thus changing the chromatin (DNA wound around histone proteins) structure – or addition of methyl-groups to the Cytosine base adjacent to a Guanine base. Epigenetic factors often interfere in gene expression regulation by promoting or inhibiting protein-DNA bindings. Such proteins are known as transcription factors. Transcription is the first step of gene expression where a particular segment of DNA is copied into the messenger-RNA (mRNA). Transcription factors orchestrate gene activity and are crucial for normal cell function in any organism. For example, deletion/mutation of certain transcription factors such as MEF2 have been associated with neurological disorders such as autism and schizophrenia. In this dissertation, different computational pipelines are described that use mathematical models to explain how the protein-DNA bindings are mediated by histone modifications and DNA-methylation affecting different regions of the brain at different stages of development. Multi-layer Markov models, Inhomogeneous Poisson analyses are used on data from brain to show the impact of epigenetic factors on protein-DNA bindings. Such data driven approaches reinforce the importance of epigenetic factors in governing brain cell differentiation into different neuron types, regulation of memory and promotion of normal brain development at the early stages of life. / Doctor of Philosophy / A cell is the basic unit of any living organism. Cells contain nucleus that contains DNA, self replicating material often called the blueprint of life. For sustenance of life, cells must respond to changes in our environment. Gene expression regulation, a process where specific regions of the DNA (genes) are copied into messenger RNA (mRNA) molecules and then translated into proteins, determines the fate of a cell. It is known that various environmental (such as diet, stress, social interaction) and biological factors often indirectly affect gene expression regulation. In this dissertation, we use machine learning approaches to predict how certain biological factors interfere indirectly with gene expression by changing specific properties of DNA. We expect our findings will help in understanding the interplay of these factors on gene expression.

Résumé indisponible / Résumé indisponible

Historically, sperm has been considered merely as a carrier of genetic material at fertilisation. However, it is known that sperm supports embryonic development better than other cell types, suggesting that it might also have additional important, non-genetic contributions to embryonic development. The work described in this dissertation focuses on identifying the molecular determinants of developmental programming of sperm. First, the development of embryos derived from sperm and spermatids, immature precursors of sperm was compared. Sperm-derived embryos developed significantly better than spermatid-derived embryos. Further research aiming to identify the reasons for the developmental advantage of sperm led to the identification of proteins that are present specifically in sperm and not in spermatids. Moreover, egg factors which are preferentially incorporated into the sperm, but not into the spermatid chromatin were identified with the use of egg extracts, suggesting that the chromatin of sperm could be programmed to interact with the components of the egg. Subsequently, the reasons for developmental failure of spermatid-derived embryos were investigated. By comparing the sperm with spermatids it was shown that the programming of sperm to support efficient development is linked to its special ability to regulate expression of developmentally-important embryonic genes, and not to its ability to support DNA replication or rRNA production. Further characterisation of the sperm and spermatid chromatin with the use of genome-wide sequencing allowed me to link the correct regulation of gene expression in the embryo with a certain combination of epigenetic marks in the sperm, but not in the spermatid chromatin. Finally, it is shown that enzymatic removal of epigenetic modifications at fertilisation leads to misregulation of gene expression. This therefore suggests that epigenetic information contained in parental genomes at fertilisation is required for a proper regulation of embryonic transcription. My results support the hypothesis that the sperm is not only a carrier of genetic material, but also provides the embryo with epigenetic information for regulation of transcription after fertilisation. I believe that these findings advance our current understanding of the nature and mechanisms of sperm programming for embryonic development, and are important contributions to the emerging field of transgenerational inheritance of epigenetic traits in general.

La leucémie est une forme de cancer qui affecte les cellules du système hématopoïétique. Selon la lignée cellulaire affectée et la vitesse de développement du cancer, la leucémie peut être myéloïde ou lymphoïde, aiguë ou chronique, respectivement. Chez les enfants, elles sont souvent caractérisées par la présence de translocations chromosomiques, impliquant notamment le gène KMT2A. L'impact biologique de ces fusions de gènes, connues pour être des perturbateurs épigénétiques, est encore mal compris. Afin d’étudier spécifiquement les conséquences de la présence de fusion impliquant le gène KMT2A, un modèle leucémique humain chez la souris a été mis en place. Le modèle utilisé consiste à induire de manière rétrovirale l’expression d’une fusion oncogénique dans des cellules souches hématopoïétiques et progénitrices d’un unique donneur sain. Ces cellules sont ensuite injectées dans des souris immunodéficientes pour produire une leucémie aiguë myéloïde ou lymphoïde après quelques semaines. L’utilisation de ce modèle leucémique vise à définir les gènes qui sont régulés de manière épigénétique et essentiels dans le processus de leucémogenèse médié par une translocation chromosomique faisant intervenir le gène KMT2A. La première partie des travaux cartographie les changements génétiques et épigénétiques à chacun des stades de la leucémogénèse causée par la fusion KMT2A-MLLT3. Nous avons cartographié les changements épigénétiques tels que la méthylation de l’ADN (Methyl-seq), les modifications des histones (ChIP-seq) et l’accessibilité de la chromatine (ATAC-seq), puis les avons corrélés avec les niveaux d’expression des gènes (RNA-seq). Nous avons observé que les leucémies myéloïdes aiguës présentent un phénotype global d'hypométhylation tandis que les changements d'expression après l'addition de la fusion ont mis en évidence l’inactivation de gènes associés aux cellules souches et des altérations dans d'autres gènes impliqués dans la leucémogenèse tels que S100A8/9. Nos données d’ATAC-seq ont montré qu'il y avait relativement peu de changements spécifiques à la leucémie myéloïde aiguë et que la grande majorité correspondait à des régions de chromatine ouvertes et à des régions contenant des motifs pour des facteurs de transcription précédemment observés dans d'autres types de cellules sanguines. L’analyse des marques d’histones associées à des promoteurs actifs suggère également un potentiel rôle du récepteur CCR1 et de son ligand spécifique CCL23. Finalement, nos résultats suggèrent que la transformation leucémique par la fusion KMT2A-MLLT3 implique des modifications épigénétiques minimes qui requièrent également la coopération des réseaux transcriptionnels utilisés dans les cellules sanguines normales. La deuxième partie de cette thèse s’intéresse à la fusion de gènes KMT2A-MLLT4, une translocation chromosomique peu étudiée mais pour laquelle le pronostic vital des patients est connu pour être défavorable et pire que celui des patients porteurs de la fusion KMT2A-MLLT3. L’extension de notre modèle à la fusion KMT2A-MLLT4 nous permet d’appliquer les mêmes approches que précédemment et de détailler les différences génétiques et épigénétiques entre ces deux fusions, jusqu’à maintenant jamais caractérisées. Nous avons pu observer une baisse globale d’expression dans un groupe de gènes intervenant dans les processus ribosomaux et traductionnels. Par ailleurs, PROM1 (CD133) fait office de potentiel candidat biomarqueur permettant la distinction entre ces deux translocations chromosomiques tandis que le gène LPL pourrait jouer un rôle dans la leucémogenèse médiée par la fusion de gènes KMT2A-MLLT4. En conclusion, l’étude des mécanismes à chacun des stades du développement leucémique nous a fourni une meilleure compréhension des changements épigénétiques intervenant dans le processus de leucémogenèse causé par des réarrangements de type KMT2A. Une meilleure caractérisation de la pathophysiologie de la leucémie pourrait permettre d’explorer des avenues thérapeutiques plus ciblées. / Leukemia is a form of cancer that affects blood cells. Depending on the affected cell lineage and the rate at which the cancer grows, leukemia can be myeloid or lymphoid, or acute or chronic, respectively. In children, they are often characterized by the presence of chromosomal translocations, in particular involving the KMT2A gene. The biological impact of these gene fusions, known to be epigenetic disruptors, is still poorly understood. To study the consequences of the presence of gene fusions involving KMT2A, we have developed a human leukemia model. The model consists of transducing hematopoietic stem and progenitor cells (CD34+) from a single healthy donor with a retrovirus bearing an oncogenic fusion. These cells are injected into immunodeficient mice to produce acute myeloid or lymphoid leukemia after a few weeks. By using this model, we aim to define genes that are epigenetically regulated and essential in the process of leukemogenesis mediated by KMT2A gene fusions. The first part of this thesis characterized the genetic and epigenetic changes at each step of leukemogenesis caused by KMT2A-MLLT3 gene fusion. We investigated epigenetic changes such as DNA methylation (Methyl-seq), histone marks (ChIP-seq), and chromatin accessibility (ATAC-seq) and correlated these with expression changes (RNA-seq). We observed that acute myeloid leukemias exhibit a profound hypomethylation phenotype while expression changes after addition of the fusion highlighted the loss of stem cell associated genes and alterations in other genes implicated in leukemogenesis such as S100A8/9 in the early stages of leukemic transformation. Our ATAC-seq data showed that there were relatively few changes specific to acute myeloid leukemia and that the vast majority corresponded to open chromatin regions and clusters of transcription factors previously seen in other types of blood cells. Examination of ChIP-seq data for active histone marks revealed that leukemia specific expression of the chemokine CCL23 can enable autocrine signalling through its cognate receptor, CCR1. Our results suggest that KMT2A-MLLT3 induces minimal changes in the epigenome while co-opting the normal transcriptional machinery to drive leukemogenesis. The second part of this thesis focuses on KMT2A-MLLT4 gene fusion, another chromosomal translocation for which the vital prognosis of patients is known to be worse than that of patients carrying the KMT2A-MLLT3 fusion. The extension of our model to the KMT2A-MLLT4 fusion allows us to apply the same approaches and to characterize the genetic and epigenetic differences between these two different leukemias. We were able to observe a dramatic decrease in the expression level of a group of genes involved in ribosomal and translational processes. Furthermore, PROM1 (CD133) acts as a potential biomarker candidate which might be used to make the distinction between these two leukemias. LPL gene might play a role in leukemogenesis mediated by KMT2A-MLLT4 gene fusion. In conclusion, studying the mechanisms at each stage of leukemic development has provided us with a better understanding of the epigenetic changes involved in the process of leukemogenesis mediated by KMT2A rearrangements. A better characterization of the pathophysiology of leukemia could make it possible to eventually develop more targeted therapeutic treatments.

Nitrogen (N), an essential plant macronutrient, is among the most limiting factors of crop yield. To sustain modern agriculture, N is often amended in soil in the form of chemical N fertilizer, a major anthropogenic contributor to nutrient pollution that affects climate, biodiversity and human health. To achieve agricultural sustainability, a comprehensive understanding of the regulation of N response in plants is required, in order to engineer crops with higher N use efficiency. Recently, epigenetic mechanisms, such as histone modifications, have gained increasing importance as a new layer of regulation of biological processes. However, our understanding of how epigenetic processes regulate N uptake and assimilation is still in its infancy. To fill this knowledge gap, we first performed a meta-analysis that combined functional genomics and network inference approaches to identify a set of N-responsive epigenetic regulators and predict their effects in regulating epigenome and transcriptome during plant N response. Our analysis suggested that histone modifications could serve as a regulatory mechanism underlying the global transcriptomic reprogramming during plant N response. To test this hypothesis, I applied chromatin immunoprecipitation-sequencing (ChIP-Seq) to monitor the genome-wide changes of four histone marks (H3K27ac, H3K4me3, H3K36me3 and H3K27me3) in response to N supply in tomato plants, followed by RNA-Seq to profile the transcriptomic changes. To investigate the organ specificity of histone modifications, I assayed shoots and roots separately. My results suggest that up to two-thirds of differentially expressed genes (DEGs) are modified in at least one of the four histone marks, supporting an integral role of histone modification in regulating N response. I observed a synergistic modification of active histone marks (H3K27ac, H3K4me3 and H3K36me3) at gene loci functionally relevant to N uptake and assimilation. Surprisingly, I uncovered a non-canonical role of H3K27me3, which is conventionally associated with repressed genes, in modulating active gene expression. Interestingly, such regulatory role of H3K27me3 is specifically associated with highly expressed genes or low expressed genes, depending on the organ context. Overall, I revealed the multi-faceted role of histone marks in mediating the plant N response, which will guide breeding and engineering of better crops with higher N use efficiency