Dans cette thèse, nous nous intéressons à la modélisation statistique de données biologiques, et plus particulièrement à l'étude de l'information génétique et protéique.Dans un premier volet, nous avons amélioré un modèle statistique des données immunologiques existant chez la souris, que nous avons transposé à l'homme, afin d'étudier les différentes recombinaisons qui apparaissent au sein du thymus, à la fin de la vie embryonnaire, entre les segments des gènes de la portion V(D)J du chromosome 14 humain, appelées recombinaisons V(D)J.Dans un deuxième volet, nous avons étudié l'information génétique par le biais des réseaux de régulations génétique, celui de la maladie familiale « atrésie biliaire », ainsi que dans les réseaux de contrôle du système immunitaire, que nous avons appelés « Immunetworks ».Dans un troisième volet, nous proposons une nouvelle approche de la compression des données biologiques, qui intègre une étape de modélisation des processus dynamiques qui leur ont donné naissance : nous avons appelé cette approche la transformée Dynalet et nous l'appliquons, entre autres, à des signaux de spectrométrie RMN (Résonance Magnétique Nucléaire) des protéines et acides nucléiques. Cette méthode consiste à convertir les signaux de spectrométrie en sons, afin de construire une lutherie anharmonique permettant de reproduire les pics de relaxation périodisés, issus des spectres RMN des 20 acides aminés, ainsi que de ceux des 4 bases nucléiques azotées. / In this thesis, we focus on statistical modeling of biological data, and more particularly on the study of genetic and protein information.First, we have improved a statistical model of existing immunological data in mice, we have transposed it to human, in order to study the various recombinations that occur within the thymus, at the end of the embryonic life, between segments of genes from the portion V (D) J of the human chromosome 14, called recombinations V (D) J.Secondly, we studied the genetic information through genetic regulatory networks, for example in the case of a family illness called the "biliary atresia," as well as in the immune system control networks, which we have called "Immunetworks".In a last part, we propose a new approach for biological data compression, which includes a step of modeling the dynamic processes that gave rise to them: we call this approach the "Dynalet" transform and we apply it, among others, to NMR spectrometry signals, i.e., Nuclear Magnetic Resonance spectra of proteins and nucleic acids. This method consists in converting the peaks of the spectrometric signals into sounds, in order to construct an anharmonic instrument capable to reproduce periodized relaxation peaks from the NMR spectra of the 20 amino acids, as well as those of the 4 nucleic bases.

RAGs (Recombination Activating Genes) are responsible for generation of antigen receptor diversity in case of B-cells and T-cells, through the process of combinatorial joining of different V (variable), D (diversity) and J (joining) gene segments. Each of these segments are flanked by recombination signal sequences (RSS), which consist of a conserved heptamer and nonamer separated by a less conserved spacer of 12 or 23 bp. RAGs recognize and cleave at the 5’ end of heptamer, leading to the formation of hairpin coding ends and blunt signal ends. The coding ends are joined through the process of no homologous DNA end joining (NHEJ), leading to the rearrangement of variable region of antigen receptors. Apart from its physiological property, RAGs can also act as a structure-specific nuclease. Previously, it has been shown that inadvertent action of RAGs on cryptic RSS and non B-DNA structures can lead to the generation of genomic instability and cancer. A very coordinated expression of RAGs has been observed in pro- and pre-B cells of the lymphoid system, which overlaps with the window of productive rearrangement during V(D)J recombination. Besides, studies by us and others have shown that RAG cleavage at altered DNA structures and cryptic RSS leads to chromosomal translocations resulting into cancer. However, several questions related to regulation of RAG expression and its activity in lymphoid cells remains to be answered. Previous studies have suggested regulation of RAG expression at different levels, such as methylation, ubiquitination, phosphorylation and by coordinate action of various transcription factors. In the present study, we evaluate the potential role of miRNAs in the regulation of RAG expression and its function in lymphoid cells. miRNAs are small, single-stranded non-coding RNAs, which play an important role in the regulation of gene expression. They play a critical role in the regulation of different cellular functions. Although there are miRNAs identified to play critical role during development of immune system, several key questions such as its role in the regulation of RAGs is yet to be addressed. In the current study, we have used bioinformatics approach to extract potential miRNAs that bind to 3’UTR of RAG1 and RAG2. miRNA expression datasets were downloaded from NCBI SRA database and extensive evaluation was done using various bioinformatics tools such as Bowtie, Sam tools, Bam tools, Bed tools and R package. We screened the miRNA expression profile across different stages of B-cell development (pro, pre, immature and mature B-cells), which overlap with the narrow window of RAG expression. The shortlisted miRNAs were further analyzed using miRNA databases such as miRBase, Targetscan and EMBL. Results showed that 33 miRNAs were specific to RAG1, among that one (miRNA1) followed RAG expression profile in B-cells. Besides miRNA2, which is a novel miRNA, was selected only on the basis of RAGs expression profile in a stage specific manner and the complementarity of the seed sequence of miRNA2 to the 3’UTR of RAG1 was checked manually. Interestingly, we observed that RAG1 expression was significantly down regulated in the presence of these miRNAs. However, there was no significant difference in the levels of other genes analysed. Further, semi-quantitative RT-PCR analysis confirmed the endogenous processing of pre-miRNA into mature miRNA using the cellular machinery. Besides, enrichment of 3’UTR of seed region of these miRNAs, enhanced the expression level of RAG1. Importantly, the enhancement in RAG1 expression level was limited in case of mature B-cells, where RAG expression is normally not observed. Further, transfection of lymphoid cells with miRNA inhibitors, specific to the miRNAs under study, showed the enhancement in RAG1 expression in lymphoid cells. In addition to this, specificity of selected miRNAs was confirmed by performing 3’UTR reporter assays, where enhanced luciferase expression was observed in case of mutant 3’UTR, while it was minimal in case of wild type constructs. Endogenous expression levels of selected miRNAs were evaluated in both lymphoid and nonlymphoid normal tissues and cancer cells using RT-PCR. Interestingly, we observed inverse correlation of expression levels of miRNA and RAG expression in all the cells tested. Besides, miRNA expression levels were less in pre-B cells and T-cells, owing to the increased expression of RAGs. Apart from this, recombinogenic potential of candidate miRNAs was assessed using episomal based V(D)J recombination assays. Interestingly we observed significant decrease (2-4 fold) in the V(D)J recombination efficiency when miRNA1 or 2 constructs were transfected in Nalm6 cells, as compared to that of controls, where no miRNAs were used. However, in case of Reh cells upon transfection with miRNA1construct, the decrease in recombination potential was upto 9 fold. Hence, we identify two miRNAs that can play an important role in the regulation of RAG1 expression and its physiological activity. Further, studies are being carried out to confirm their role in the regulation of RAG1 during different developmental stages of lymphoid cells in mice. As stated above, in addition to the sequence-specific activity, RAG possesses structure-specific nuclease activity as well. It has been shown that RAGs can cleave different types of altered DNA structures. Studies from our laboratory showed that even when RAGs act as a structure-specific nuclease there is a sequence bias. Presence of cytosine and thymine at the single-stranded region of heteroduplex DNA is important for RAG nicking and double-strand break (DSB) formation. In addition, proximity of a nonamer to bubble structures can enhance RAG cleavage. However, the role of immediate flanking sequences in the RAG mediated cleavage at heteroduplex regions is not understood. We investigated the role of flanking double-stranded DNA sequences in the regulation of RAG cleavage on non-B DNA structures. We found that RAG binding and cleavage on heteroduplex DNA is dependent on the length of double-stranded flanking region. Besides, immediate flanking regions of the heteroduplex DNA affected the RAG binding and cleavage in a sequence dependent manner. Interestingly, we also observed that the cleavage efficiency of RAGs at heteroduplex region was influenced by the phasing of DNA. Thus, our results suggest that sequence, length and phase positions of the DNA can affect the efficiency of RAG cleavage when it acts as a structure-specific nuclease. These findings provide novel insights into regulation of the pathological action of RAGs. Previous studies have shown that in addition to formation of coding and signal joints during V(D)J recombination, nonstandard V(D)J recombination products known as hybrid joints and open-shut joints may be formed, particularly in certain aberrant conditions such as defective NHEJ machinery. Interestingly, the hybrid and open-shut joints closely resemble the transposition mechanisms associated with transposons oretroviruses. Studies have also shown that RAGs possess structural similarity with integrases in domain organization. Both the proteins have Zinc Finger Binding domain (ZFB) which helps in multimerization of the protein, a central catalytic core domain comprising three acidic amino acids D, D and E essential for enzymatic activity and C-terminal domain (CTD) responsible for nonspecific binding to the DNA. Previous studies from our laboratory showed that, Elvitegravir, an inhibitor of integrase could interfere with the biochemical functions of the RAGs in vitro. Specifically, it inhibited the RAG binding and cleavage at RSS, hairpin formation, post-cleavage complex formation involving 12RSS and 23RSS. Using the episomal assay system that mimics signal joints (pGG49) and coding joints (pGG51), we show that Elvitegravir can inhibit V(D)J recombination inside cells. Interestingly we observed 3-6 fold decrease in the recombination frequency in signal ends joining, when treated with increasing concentrations (100, 500 and 1000 nM) of Elvitegravir. A 5-8 fold decrease in coding joints formation was also observed upon treatment with the inhibitor. The presence of recombination was confirmed by restriction digestion followed by sequencing analysis. Further analysis of recombination junctions revealed extensive deletion before joining in the case of Elvitegravir treated samples. Insertions or substitutions near to the recombination junctions were also prominent in treated samples. In depth analysis of sequenced junctions showed the presence of sequence having the features to form hairpins both upstream and downstream to the RSS sequences and was the site of cleavage in cases were higher deletion was observed. The analyzed recombinants did not show any signal joints or coding joints formation in treated samples. This suggests that Elvitegravir affects the physiological function, the V(D)J recombination of RAGs inside the cells. Thus, in the present study, we show that RAGs can be regulated by specific miRNAs. We have identified two potential miRNAs, which can regulate the RAG expression as well as its function in different stages of B- and T-cell development. Further, we also identify a novel regulatory mechanism for the structure-specific activity of the RAG complex. In addition to this, we find that integrase inhibitor, Elvitegravir, affects V(D)J recombination within B-cells, indicating its potential deleterious impact in HIV patients, which needs to be further evaluated.

Ce travail de thèse porte sur la diffusion des porteurs de charge en excès dans deux semiconducteurs à large bande interdite: l’alliage ZnMgO et le diamant 13C. Il est basé sur l’étude d’hétérostructures ZnMgO/ZnO/ZnMgO et 13C/12C/13C à puits de collecte ZnO ou 12C. Sur leurs sections transverses et avec la résolution nanométrique en excitation par cathodoluminescence (CL), nous avons étudié l’évolution de l’intensité de l’émission issue du puits en ZnO ou 12C en fonction de la distance entre l’impact de l’excitation et le puits. Cela nous a permis de mesurer directement les longueurs de diffusion effectives dans ZnMgO et le diamant.Dans ZnMgO, la valeur de 55 nm à 300 K, mesurée sur section transverse clivée, est proche de celle du matériau massif. Elle correspond à une diffusion mixte excitons/porteurs libres. Avec l’utilisation de lames minces érodées par faisceau d’ions, une diminution de a été observée jusqu’à 8 nm dans les parties les plus fines. Cet effet est attribué aux recombinaisons non radiatives de surface. Les lames minces apparaissent alors d’un grand intérêt pour améliorer la résolution spatiale des images CL.Dans le diamant, la diffusion excitonique à basse température montre une faible dépendance de avec l’énergie incidente des électrons. Cela indique que ≈ 15 µm à 20 K dans le diamant massif 13C. Une diminution de jusqu’à 3,3 µm à 118 K est observée en fonction de la température.Enfin, nous avons mis en évidence la formation de polyexcitons dans le diamant en augmentant la densité des paires électron-trou, soit par la puissance d’excitation, soit par le confinement spatial des excitons dans des puits de diamant 12C de faible épaisseurs. / This work focuses on the determination of the carrier diffusion length in two wide bandgap semiconductors: the ternary alloy ZnMgO and diamond. This determination has been achieved by using of ZnMgO/ZnO/ZnMgO and 13C/12C/13C heterostructures containing ZnO or 12C collecting wells. Their transverse section was scanned by CL spectroscopy with a nanometer scale resolution in excitation. The effective excess carrier diffusion length is deduced from the evolution of the well emission intensity with the distance between the excitation impact and the well.In ZnMgO, the value at 300 K is 55 nm, obtained from a cleaved cross section. It is close to the bulk material diffusion and is attributed to a mixed diffusion of excitons/free carriers. A decrease of down to 8 nm is observed in the thinnest portions of cross sections shaped by focused ion beam (FIB). This effect is attributed to non-radiative surface recombinations. These thin slabs appear of great interest to enhance the spatial resolution of CL images.In diamond, the exciton diffusion at 20 K exhibits a slight dependence on the incident electron energy. This indicates that the exciton diffusion length is around 15 µm in 13C bulk diamond. The values decrease down to 3.3 µm at 118 K.Finally, we highlighted the formation of polyexcitons in diamond by increasing the electron-hole pairs density either by the excitation power, or by the spatial confinement of excitons in thin 12C wells.