Molecular association is part of many important processes in living cells. Computational methods for calculating binding free energies allows for a quantitative examination of biomolecular structures and hypotheses drawn from biochemical experiments. Here, binding free energy calculations for tRNAs and release factors binding to mRNA codons on the ribosome, sugars binding to lectins and purine analogs binding to the purine riboswitch are presented. The relative affinities between cognate and non-cognate tRNAs for different states involved in codon reading on the ribosome were determined. The calculations show that tRNA discrimination varies between different conformations of the 30S subunit, where the existence of both low and high selectivity states provides an efficient common mechanism for initial selection and proofreading. The simulations reveal a desolvation mechanism for the 30S conformational switch with which the accuracy of peptide bond formation can be amplified. When an mRNA stop codon (UAA, UAG or UGA) is located in the ribosomal A-site release factors bind to the ribosome and the synthesized protein is released. RF1 is specific for UAA and UAG whereas RF2 is specific for UAA and UGA. The free energy calculations and an analysis of the performed simulations show the mechanisms for how RF1 and RF2 are able to read the stop codons with different specificities. Also mitochondrial release factors were investigated. Vertebrate mitochondria have four stop codons, UAA, UAG, AGA and AGG and two release factors mtRF1 and mtRF1a. The calculations show how the specificities of both mtRF1 and mtRF1a agree with RF1 and that none of them are likely to read the non-standard stop codons AGA and AGG. The linear interaction energy method has also been examined for the RSL and PA-IIL lectins and for the purine riboswitch. The standard parameterization of the method works well for RSL, but fails for PA-IIL and the purine riboswitch due to compositions of the active sites in these systems. The development of new parameterizations to overcome these problems leads to a better understanding of both the method and the binding mechanisms in these systems.

DNA sequence alignments are usually not homogeneous. Mosaic structures may result as a consequence of recombination or rate heterogeneity. Interspecific recombination, in which DNA subsequences are transferred between different (typically viral or bacterial) strains may result in a change of the topology of the underlying phylogenetic tree. Rate heterogeneity corresponds to a change of the nucleotide substitution rate. Various methods for simultaneously detecting recombination and rate heterogeneity in DNA sequence alignments have recently been proposed, based on complex probabilistic models that combine phylogenetic trees with factorial hidden Markov models or multiple changepoint processes. The objective of my thesis is to identify potential shortcomings of these models and explore ways of how to improve them. One shortcoming that I have identified is related to an approximation made in various recently proposed Bayesian models. The Bayesian paradigm requires the solution of an integral over the space of parameters. To render this integration analytically tractable, these models assume that the vectors of branch lengths of the phylogenetic tree are independent among sites. While this approximation reduces the computational complexity considerably, I show that it leads to the systematic prediction of spurious topology changes in the Felsenstein zone, that is, the area in the branch lengths configuration space where maximum parsimony consistently infers the wrong topology due to long-branch attraction. I demonstrate these failures by using two Bayesian hypothesis tests, based on an inter- and an intra-model approach to estimating the marginal likelihood. I then propose a revised model that addresses these shortcomings, and demonstrate its improved performance on a set of synthetic DNA sequence alignments systematically generated around the Felsenstein zone. The core model explored in my thesis is a phylogenetic factorial hidden Markov model (FHMM) for detecting two types of mosaic structures in DNA sequence alignments, related to recombination and rate heterogeneity. The focus of my work is on improving the modelling of the latter aspect. Earlier research efforts by other authors have modelled different degrees of rate heterogeneity with separate hidden states of the FHMM. Their work fails to appreciate the intrinsic difference between two types of rate heterogeneity: long-range regional effects, which are potentially related to differences in the selective pressure, and the short-term periodic patterns within the codons, which merely capture the signature of the genetic code. I have improved these earlier phylogenetic FHMMs in two respects. Firstly, by sampling the rate vector from the posterior distribution with RJMCMC I have made the modelling of regional rate heterogeneity more flexible, and I infer the number of different degrees of divergence directly from the DNA sequence alignment, thereby dispensing with the need to arbitrarily select this quantity in advance. Secondly, I explicitly model within-codon rate heterogeneity via a separate rate modification vector. In this way, the within-codon effect of rate heterogeneity is imposed on the model a priori, which facilitates the learning of the biologically more interesting effect of regional rate heterogeneity a posteriori. I have carried out simulations on synthetic DNA sequence alignments, which have borne out my conjecture. The existing model, which does not explicitly include the within-codon rate variation, has to model both effects with the same modelling mechanism. As expected, it was found to fail to disentangle these two effects. On the contrary, I have found that my new model clearly separates within-codon rate variation from regional rate heterogeneity, resulting in more accurate predictions.

The human prion protein (PrP) gene (PRNP) codon 129 (M/V) polymorphism is a susceptibility factor for variant Creutzfeldt-Jakob Disease (vCJD) and a major determinant of clinico-pathological phenotype in sporadic CJD. The role of codon 129 in defining susceptibility and strain transmission properties has been investigated in three lines of transgenic mice that express human PrP. The human PRNP gene has directly replaced the murine version, by gene targeting, and variation at codon 129 has given the three genotype lines (HuMM, HuMV, and HuVV). The genetics of these three mouse lines are otherwise identical, and therefore differences in transmission properties can be directly attributable to the codon 129 genotype. vCJD inoculation has shown that all three codon 129 genotype mice are susceptible with a ranking of transmission efficiency of HuMM>HuMV>HuVV. HuMM mice develop the most widespread neuropathology with features similar to human vCJD. Subclinical infection was noted in each mouse line. These data suggest that the vCJD strain is transmissible to humans of each of the three codon 129 genotypes, implying that non-MM cases of human infection with bovine spongiform encephalopathy (BSE) may exist but with long subclinical incubation periods. Inoculation of material from blood transfusion associated vCJD showed no change in transmission properties suggesting that the threat of a future epidemic of human-to-human vCJD infection has not been increased by adaptation of the vCJD strain. However the route of infection, for example via blood transfusion or surgery, may be more efficient that the original oral route of BSE infection. sCJD is classified into six subgroups according to clinico-pathological features, and defined by codon 129 genotype and electrophoretic mobility type (1 or 2) of disease associated PrPSc (MM1, MM2, MV1, MV2, VV1, VV2). Typical cases from each subgroup have shown specific transmission properties suggesting that the subgrouping is defining separate disease strains. The commonest subgroup (MM1) was the most transmissible and the HuVV mouse line the most susceptible host. These data outline the transmission risk from all sCJD types to recipients of each codon 129 genotype should an infection event occur, and show the significant role of recipient codon 129 genotype in defining the clinical or subclinical state and the success or failure of transmission. This is important for determining individual risk following known exposure, and for modelling the potential of iatrogenic infection from sCJD patients.

Translation initiation in the hepatitis C virus (HCV) occurs through a cap- independent mechanism that involves an internal ribosome entry site (IRES) capable of interaction with and utilization of the eukaryotic translational machinery. We focused on the structural configuration of the different HCV-IRES domains and the impact of IRES primary sequence variations on secondary structure conservation and function. For this purpose we introduced into our laboratory, methods such as denaturing gradient and temperature gradient gel electrophoresis for screening the degree of heterogeneity and total amount of HCV-IRES variability accumulated in HCV infected patients over a period of time. The selected samples showed variable migration pattern of the HCV-IRES (from all the patients) visualized in DGGE and TGGE, were sequenced and evaluated for translation efficiency using flow cytometry. In some cases, we discovered that multiple mutations, even those scattered across different domains of HCV-IRES, led to restoration of the HCV-IRES translational activity, although the individual occurrences of these mutations were found to be deleterious. We propose that such observation may be attributed to probable long- range inter- and/or intra-domain functional interactions. We established a large-scale HCV-IRES...

Arsenic contamination of groundwater remains a serious health concern in many areas of the world. Developing countries such as Bangladesh and Nepal are particularly affected because access to high quality water infrastructure is low. Since the 1970s, most water in these countries is sourced from shallow tube wells installed to reduce the spread of diseases associated with poor water hygiene. In this goal they were successful, however by the mid 1990s it became apparent that many of these wells were contaminated by arsenic and that these countries’ rural poor were being slowly poisoned. No simple, cheap, and reliable test for arsenic exists, and efforts to mitigate arsenic contamination have been severely limited by this over the past two decades. Government backed well-testing efforts using commercially available field kits have many issues with reliability, safety, rigour, and transparency, and have lost their urgency over the past decade, while the expensive field test kits remain out of the reach of most ordinary people in these areas. Synthetic Biology offers the technology to develop a new class of biosensor by exploiting bacteria’s natural ability to sense and respond to levels of arsenic considerably lower than commercially available kits which are based on analytical chemistry. In order to reach this goal, we must first develop our understanding of the natural response to arsenic in our chosen host, B. subtilis. Although we have a reasonably good qualitative understanding of the operon responsible for arsenic sensing, very little quantitative analysis has been carried out, and a robust system for ratiometric characterisation has not been established in the bacteria. In this work, a robust platform for rapid ratiometric characterisation is established in B. subtilis. A rigorous mathematical model of the ars operon is developed and analysed before being verified experimentally. This new knowledge is then used to explore synthetic permutations to the natural system aimed at improving the sensor properties of the system. Finally, a biological architecture for an easily tunable biosensor with good characteristics is recommended.

La traduction des ARNm en protéine est un processus finement régulé grâce aux mécanismes développés par la cellule pour en contrôler l’efficacité et la fidélité. En effet, les ARNm sont sujets à diverses erreurs au cours de leur transcription et leur maturation. En particuliers, les erreurs entrainant l’apparition de codons stop précoces peuvent conduire à la synthèse de protéines tronquées à effet néfaste sur la cellule. C’est pour cela que de tels ARNm sont rapidement dégradés grâce à un mécanisme régulateur appelé la NMD (Nonsence mediated mRNA Decay). Chez la levure Saccharomyces cerevisiae, cette voie est régie par l’action coordonnée des protéines Upf1, Upf2 et Upf3 formant le complexe de surveillance, mais elle fait également intervenir les facteurs de terminaison classique eRF1 et eRF3, ainsi que d’autres facteurs peu caractérisés tels que la protéine Ebs1. Par ailleurs, la dégradation de ces ARNm défectueux est accélérée par la dégradation rapide de la coiffe ou « decapping ». Au cours de ce travail, nous avons caractérisé des domaines fonctionnels de protéines impliquées dans la détection et la dégradation de ces ARNm. En particuliers, nous nous sommes intéressés à l’étude structurale de la protéine Upf2 qui constitue l’élément central du complexe de surveillance. Nous avons également caractérisé un domaine de la protéine Pat1, puissant activateur du « decapping ». Cette étude nous a permis de mieux comprendre le rôle de ces protéines dans le contrôle qualité et la dégradation des ARNm. / MRNA translation process is finely tuned thanks to the regulatory mechanisms evolved by the cell controlling its rate, efficiency and fidelity. Indeed, mRNAs are often subjected to transcription and maturation errors. In particular, mRNA harboring premature stop codons (PTC) in their open reading frames could be translated into truncated proteins with a deleterious impact on the cell. Thus, such mRNAs are rarely detected in the cell as they are rapidly degraded thanks to the NMD (Nonsence mediated mRNA Decay) pathway. In yeast Saccharomyces cerevisiae, this process is governed by the Upf1, Upf2 and Upf3 proteins forming the “surveillance complex”, the termination factors (eRF1 and eRF3) as well as some other poorly characterized factors like Ebs1 protein. In addition, degradation of such mRNAs is enhanced by rapid degradation of the 5’ cap or decapping. In this work, we focused on the characterization of some proteins involved in this process. In particular, we addressed the structural characterization of Upf2 protein, the central component of the surveillance complex. In addition, we characterized a functional domain of Pat1 protein, a strong decapping enhancer. This study allowed us to give a new insight into the role of these proteins in mRNA quality control and decay.

Le code génétique est la table de correspondance entre codons (unité structurelle d'un gène) et acides aminés (brique élémentaire des protéines). Le code génétique est (1) universel, tous les êtres vivants ou presque partagent le même code; (2) univoque, chaque codon spécifie un seul acide aminé et (3) dégénéré, les acides aminés peuvent être codés par plusieurs codons. Ce code dégénéré est donc utilisé par l'ensemble du vivant mais pas de la même manière, certains codons synonymes étant utilisés préférentiellement chez des espèces et pas d'autres. Pour comprendre l'émergence des biais d'usage du code (BUC) génétique entre espèces, je me place dans un contexte évolutif.Dans ce manuscrit, je présente mes travaux de recherche en quatre parties. La première partie introductive décrit la mise en évidence et les propriétés du code génétique, son biais d'usage et les diverses caractéristiques de précédents modèles de codons. La deuxième partie présente le modèle d'évolution de codons SENCA pour Sites Evolution at the Nucleotides, Codons and Amino-acids layers que j'ai développé durant ma thèse. SENCA prend en compte la structure du code génétique. Je valide sa paramétrisation par des simulations numériques et une étude sur des espèces bactériennes ou archées. La partie suivante décrit deux extensions de SENCA qui modélisent plusieurs hypothèses d'origines évolutives du BUC et une application de SENCA sur les conséquences génomiques d'adaptations environnementales. La dernière partie étudie les origines de variations de BUC le long du génome humain par une approche de génomique comparative / In this manuscript, I introduce my doctoral research in four parts. The first introductive part highlights the properties of the genetic code and its usage bias but also the caracteristics of previous published codons models. The second part presents an evolutionary codons models named SENCA for Sites Evolution at the Nucleotides, Codons and Amino-acids layers that I developped. SENCA takes into account the genetic code structure. I perform simulations and study prokaryotes species to confirm its parametrization. The following part provides two extensions of SENCA to test the hypotheses concerning the evolutive origins of CUB and an application of SENCA to study the genomic consequences of an environmental adaptation. The last part studies the origins of CUB variation within the human genome using a comparative genomic strategy

Submitted by Erika Demachki (erikademachki@gmail.com) on 2014-10-14T20:46:44Z No. of bitstreams: 2 Dissertação - Jeany Camelo Santos - 2013.pdf: 1042379 bytes, checksum: 0e6b3b41192470d2c48f4809f1e9d49b (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Jaqueline Silva (jtas29@gmail.com) on 2014-10-16T18:23:59Z (GMT) No. of bitstreams: 2 Dissertação - Jeany Camelo Santos - 2013.pdf: 1042379 bytes, checksum: 0e6b3b41192470d2c48f4809f1e9d49b (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2014-10-16T18:23:59Z (GMT). No. of bitstreams: 2 Dissertação - Jeany Camelo Santos - 2013.pdf: 1042379 bytes, checksum: 0e6b3b41192470d2c48f4809f1e9d49b (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2013-03-27 / The CML is a expansion clonal of cells progenitors hematopoietics and is associated to an specific genetic lesion, known as the Philadelphia chromosome, product of the reciprocal translocation t(9, 22)(q34, q11) that causes the oncogene BCR-ABL. The TP53 is a tumor suppressor gene located on the chromossome 17p13.1 coding for phosphoprotein TP53. The polymorphism arises from the exchange of G for C at codon 72, resulting the genotypes Arg/Arg, Arg/Pro and Pro/Pro. This study aims to determine the allelic and genotypic frequencies of TP53 polymorphism at codon 72 in CML patients and to correlate with the response to imatinib therapy. The work had the participation of 85 CML patients treated at the Clinic of Hematology, at Hospital das Clínicas – UFG in Goiânia city, state of Goiás for the diagnosis and control of disease. To investigate the allelic and genotypic frequencies, DNA samples were isolated from peripheral blood for analysis of PCR reactions. For genotyping, forward and reverse primers were used for each variant allelic. The study had the participation of 85 CML patients, which 69 were in chronic phase, eight in accelerated phase and only one in blast crisis. The mean age was 51 years and eight months. The frequency of genotypes Pro/Pro, Arg/Pro and Arg/Arg was 11% (4/35) 43% (15/35) 46% (16/35) for patients resistant to imatinib treatment (group case) and 16% (8/50) 62% (31/50) and 22% (11/30) for patients with response to imatinib (control group), respectively. The population in this study is in Hardy-Weinberg Equilibrium (x2 = 1, 12, P> 0, 05). Regarding age, gender, disease stage, and score Sokal not observed an association of the disease with the response or resistance to treatment (P= 0,36; P = 0.82, P = 0.47 and P = 0.72), respectively. When we evaluated the genotypes with respect to the Score Sokal (High x Intermediate/Low), it was observed that Pro/Pro genotype was significantly lower in the high Sokal Score group than Intermediate/low (P = 0.017, OR = 8, 19). For criterion, age over 40 years old at diagnosis, by analyzing the Fisher’s test, we found that patients carrying the Arg/Arg genotype are four times more susceptible to produce any resistance to imatinib therapy. When we evaluated the variables age, gender, disease phase, genotype and Sokal Score in logistic regression showed that only the variable genotype was significant (P = 0.0159). Our results are not according to the previous studies, in which suggest that the Pro/Pro genotype and the Pro allele can check risk of developing disease or resistance to imatinib treatment. Our findings suggest that patients carrying the Arg/Pro and Pro/Pro genotypes responded well to treatment and that the Pro/Pro genotype represented an indicator for a good prognosis. Genotype Arg/Arg represented a risk factor in genetic susceptibility in CML’s pathogenesis, contributing for a worse outcome. / A LMC caracteriza-se por uma expansão clonal de células progenitoras hematopoiética e está associada a uma alteração citogenética, conhecida como o cromossomo Philadelphia (Ph+), produto de uma translocação recíproca t(9q34;22q11), gerando a proteína híbrida BCR-ABL. O gene TP53 é um gene supressor tumoral está localizado no braço curto do cromossomo 17 (17p13.1) que codifica uma proteína fosfonuclear a TP53. O polimorfismo desse gene, envolve uma troca de uma única base guanina (G) por uma citosina (C) no códon 72, originando os genótipos Arg/Arg, Arg/Pro e Pro/Pro. Este trabalho tem como objetivo determinar as frequências alélicas e genotípicas do polimorfismo de TP53 no códon 72 em pacientes com LMC e correlacionar com a resposta ao tratamento com imatinibe. O trabalho contou com a participação de 85 pacientes com LMC atendidos no serviço do Ambulatório de Hematologia do HC (Hospital das Clínicas) da UFG, na cidade de Goiânia, Goiás, para o diagnóstico e controle da doença. A idade, o sexo, a fase da doença, o Índice Sokal e resposta e resistência ao tratamento, foram levados em consideração. Para investigar as frequências alélicas e genotípicas, amostras de DNA foram isoladas do sangue periférico para posterior análise em reações de PCR. Para a amplificação das regiões de interesse, primers específicos forward e reverse foram utilizados. Dos 85 pacientes portadores de LMC, 69 pacientes estavam na fase crônica, oito na fase acelerada e um na crise blástica. A média de idade foi de 51 anos e oito meses. A frequência dos genótipos Pro/Pro, Arg/Pro e Arg/Arg foi 11% (4/35), 43% (15/35) e 46% (16/35) para pacientes com resistência ao tratamento com imatinibe (grupo caso) e 16% (8/50), 62% (31/50) e 22% (11/30) para pacientes com resposta ao tratamento (grupo controle), respectivamente. A população do presente estudo está em Equilíbrio de Hardy Weinberg (x2 = 1, 12; P > 0, 05). Quanto à idade, sexo, fase da doença e índice Sokal não se observou associação com a resposta e resistência ao tratamento (P= 0,36; P= 0,82; P=0,47 e P=0,72), respectivamente. Quando avaliou-se os genótipos com relação ao Índice Sokal (Alto x Intermediário/baixo), observou-se que Pro/Pro, foi significativamente menor no grupo com Índice Sokal alto (P=0,017; OR=8,19). Para o critério idade acima de 40 anos no diagnóstico da doença, através da análise do Teste de Fisher, observou-se que pacientes homozigotos para o genótipo Arg/Arg são quatro vezes mais susceptíveis a apresentar algum tipo de resistência ao tratamento com imatinibe. Quando avaliou-se as variáveis: idade, sexo, fase da doença, genótipos e Índice Sokal na regressão logística, observou-se que somente a variável genótipo foi significativa (P= 0,0159). Nossos resultados divergem dos dados apresentados pela literatura sobre a LMC, que sugerem que o genótipo Pro/Pro ou alelo Pro, pode conferir risco de desenvolver a doença ou resistência ao tratamento com imatinibe. Nossos achados sugerem que pacientes Arg/Pro e Pro/Pro, responderam bem ao tratamento e que o genótipo Pro/Pro, representou um indicador para um bom prognóstico. O genótipo Arg/Arg representou um fator de risco na susceptibilidade genética à patogênese da LMC, contribuindo para um pior prognóstico da doença.

Au cours de mon travail de thèse j’ai étudié la relation entre les facteurs participant à la terminaison de la traduction et ceux participant à la dégradation des ARNm chez S. cerevisiae.D’une part, je me suis intéressée au facteur Tpa1, caractérisé pour son rôle dans la terminaison de la traduction et la stabilité des ARNm chez S. cerevisiae et dont l’homologue chez S. pombe, Ofd1, participe au contrôle de la réponse hypoxique. Je me suis basée sur la structure de ce facteur, établie par nos collaborateurs pour comprendre plus précisément la fonction moléculaire de Tpa1 et rechercher des similitudes avec sa fonction chez S. pombe.Tpa1 est composée de deux domaines de type DSBH dont le premier, contenant le site catalytique, présente des homologies structurales avec la famille des prolyl-hydroxylases.Nous avons reproduit l’effet de la protéine Tpa1 sur la translecture in vivo et montré que son site catalytique prédit, ainsi que la présence des deux domaines étaient nécessaires pour cette activité. Nous avons aussi observé que Tpa1 inhibait par un mécanisme inconnu le facteur de transcription Hap1, qui régule des gènes en fonction de la quantité d’oxygène. Basé sur l’existence d’un inhibiteur d’Ofd1 chez S. pombe, nous avons ensuite montré qu’Ett1 (l’homologue de cet inhibiteur chez S. cerevisiae) avait un rôle similaire à Tpa1 dans la translecture. Une étude structurale collaborative d’Ett1 a mis en évidence une région conservée, se liant à une molécule de sulfate et à un ligand inconnu. Cette région est importante pour la translecture. Cependant, le substrat de Tpa1 reste pour l’instant inconnu comme les rôles précis de Tpa1 et Ett1 dans la terminaison de la traduction et dans la réponse à l’hypoxie.D’autre part, j’ai étudié le processus de NMD, particulièrement en me focalisant sur le mécanisme de discrimination entre un codon stop précoce (PTC) et un codon stop normal, et en analysant également la modification post-traductionnelle d’un facteur central du NMD, Upf1. Nous avons mis en évidence, qu’en plus de la région aval, la région en amont du PTCparticipait à sa reconnaissance. Nous avons testé plusieurs hypothèses sur le rôle de cette région, qui ont confirmé son rôle sans permettre de démontrer un mécanisme définitif. En parallèle, l’étude de la protéine Upf1 s’est concentrée sur ses modifications posttraductionnelles, particulièrement par phosphorylation. En effet, une telle modification est importante chez son homologue humain. Nous avons pu confirmer l’existence d’une forme modifiée et démontrer que celle-ci était localisée entre les acides aminés 153 et 971. Cette modification s’est avérée être très labile ce qui n’a pas permis de confirmer qu’il s’agissait d’une phosphorylation, ni de la cartographier plus précisément. / During my PhD thesis, I analyzed the relation between factors that participate intranslation termination and those participating in mRNA decay in yeast S. cerevisiae.First, I focused on Tpa1, that had been proposed to participate in translationtermination and mRNA decay in S. cerevisiae, and whose homologue in S. pombe, Ofd1,participates to the control of hypoxic response. Based on the structure of Tpa1, established byour collaborators, I performed functional analysis to understand more precisely the molecularfunction of Tpa1 and similarities with its role in S. pombe. Tpa1 is composed of two DSBHdomains; the first, which contains the catalytic site, has structural homologies with the familyof prolyl-hydroxylase. We could reproduce the effect of Tpa1 on stop codon readthrough invivo and we showed that the predicted catalytic site and the presence of the two domains ofTpa1 were necessary for its activity. We also showed that Tpa1 inhibited one factor, Hap1,implicated in regulation of gene expression by oxygen. The existence of an inhibitor of Ofd1in S. pombe, allowed the identification of Ett1 (its homologue in S. cerevisiae). We showedthat Ett1 has a role similar to the one of Tpa1 in translational readthrough. A collaborativestructural and functional study of Ett1 revealed a conserved region, which binds a sulfate ion,and an unknown ligand. This region is important for the readthrough. However, thesubstrate(s) of Tpa1 remain(s) for the moment unknown, and the precise roles of Tpa1 andEtt1 in translation termination and in response to hypoxia remain to be deciphered.I also analyzed the NMD process by focusing more particularly on the mechanism thatallows the discrimination between a normal stop and a PTC (premature termination codon)and on the analysis of the post-translational modification of an important factor for the NMD,Upf1. This study revealed that, not only the region downstream of the PTC but also theupstream region participates to its recognition. We have tested several hypotheses on the roleof this upstream region, which confirmed its implication but did not reveal a definitivemechanism. In parallel, we started the study of the post-translational modifications of Upf1,and more particularly by phosphorylation. Indeed, the phosphorylation of Upf1 in human isvery important for the NMD process. We could confirm the presence of a modified form ofyeast Upf1 and we have demonstrated that it was localized between amino acids 153 and 971.This modification appeared to be highly labile. This prevented us to confirm definitively thatit was really a phosphorylation and to cartography precisely its location.

Le NMD (nonsense-mediated mRNA decay) est un mécanisme qui reconnaît et dégrade les ARNm portant un codon stop prématuré afin d'empêcher la synthèse de protéines tronquées qui pourraient avoir des effets néfastes pour la cellule ou tout simplement être non fonctionnelles. Cependant, dans un certain nombre de cas, selon la position du codon stop prématuré, la protéine tronquée qui serait synthétisée si le NMD n'existait pas, pourrait remplir complètement ou partiellement la fonction de la protéine sauvage. Il faut noter qu'un codon stop prématuré est retrouvé dans le gène responsable d'une pathologie dans un tiers des maladies génétiques et de nombreuses formes de cancer. Dans la plus grande majorité des cas, la maladie se développe non pas parce qu'une protéine tronquée non fonctionnelle ou instable est synthétisée, mais plutôt parce que le gène muté n'est pas exprimé du fait de l'intervention du NMD sur l'ARNm qui en dérive. Une nouvelle approche thérapeutique de ces maladies serait d'inhiber le NMD afin de permettre la synthèse de protéines tronquées fonctionnelles et sauver le phénotype clinique. Nous avons donc décidé de rechercher des inhibiteurs du mécanisme du NMD parmi des petites molécules chimiques. Pour cela, nous avons mis au point un système de criblage en culture cellulaire reliant l'efficacité du NMD dans une cellule avec une activité luciférase mesurable directement sur les cultures cellulaires, au moyen d'un luminomètre. A partir d'un premier criblage d'environ 1500 composés chimiques, nous avons identifié une nouvelle molécule capable d'inhiber efficacement le NMD. De façon intéressante, cette nouvelle molécule est capable également d'induire la synthèse de protéines entières à partir d'un ARNm portant un codon stop prématuré. Nous avons utilisé cet inhibiteur dans des expériences pour déterminer son potentiel thérapeutique sur des modèles cellulaires de maladies génétiques tels que la dystrophie musculaire de Duchenne, la mucoviscidose et le cancer. Nos résultats démontrent que l'inhibition du NMD peut être en effet envisagée comme une nouvelle approche thérapeutique pour des maladies causées par l'apparition d'une mutation non sens. Nous avons aussi identifié une autre molécule chimique capable d'inhiber le NMD et permettant de faire un lien entre efficacité du NMD et intégrité du cytosquelette.