Search

Aux fondements de la vie, on retrouve les protéines qui agissent telles des nanomachines à l’échelle de la cellule. En effet, elles effectuent une immense diversité de tâches vitales à tout organisme vivant, allant de la transcription/traduction de l’ADN, au transport membranaire en passant par le métabolisme énergétique, etc. Les protéines sont des macromolécules composées de combinaisons des 20 acides aminés classiques. Sous l’effet des lois fondamentales de la physique, cette chaîne va se replier localement (structure secondaire) puis globalement (structure tertiaire). C’est cette dernière qui donne la fonction aux protéines et son étude se révèle donc primordiale. Les méthodes numériques offrent un complément indispensable à l’expérience pour la réalisation de cet immense défi. Ce mémoire s’articule autour de deux axes principaux. Le premier est l’étude du N-terminal de la protéine Huntingtine à l’aide de simulations numériques. Le second est l’optimisation et le développement de la famille de potentiel gros-grain OPEP. La protéine Huntingtine est à l’origine du développement de la maladie d’Huntington et son N-terminal est crucial pour son agrégation et son interaction avec la membrane. Suite à la proposition d’un premier modèle expérimental du N-terminal de Huntingtine, nous avons déterminé son ensemble structurel, en solution et en membrane, à l’aide de simulations tout-atome et de techniques d’échantillonnage avancées. Les motifs structurels et interactions dominantes observés sont mis en relation avec les modèles détaillés de son agrégation et de son ancrage dans la membrane. Finalement, les potentiels OPEP sont des potentiels gros-grain dont l’application s’est révélée un succès pour l’étude des protéines amyloïdes et pour la prédiction de structure tertiaire avec PEPFOLD. Le potentiel sOPEP fut optimisé sur de vastes ensembles de protéines avec comme résultat une amélioration de 25% au niveau des inégalités. La paramétrisation de sOPEP fut aussi testée à l’aide de simulation sur neuf protéines. En parallèle, nous développons le potentiel aaOPEP, qui portera la philosophie des potentiels OPEP en régime tout-atome. Ces potentiels sont intégrés à un code de dynamique moléculaire flexible qui permettra leur distribution., Proteins are the vital machinery of the cell and therefore are fundamental to life. They play a wide variety of roles from the transcription/translation of DNA to membrane transport to energy metabolism and many more. In terms of structure, proteins are macromolecules composed by the combination of 20 basic amino acids. Under the fundamental laws of physic, this chain will fold itself locally (secondary structure) and then globally (tertiary). This three dimensional structure gives rise to the protein’s function thus making the prediction of three dimensional structure a crucial aspect of protein study. Numerical techniques provide an essential tool to complement experiments in the study of proteins. The following thesis is divided into two main research axes. The first one is the study, via numerical techniques, of the N-terminal of the Huntingtin protein. The second one is the optimization and the development of the coarse grained forcefield family OPEP. First, we present our study of Huntingtin’s N-terminal. The Huntingtin protein is associated with the Huntington disease and its N-terminal is crucial for the protein aggregation and anchoring inside the membrane. Following the proposition of a first experimental model of the N-terminal, we determined its structural ensemble in aqueous solution and membrane environment. To do so, we used powerful sampling techniques in the all-atom regime. Our discoveries are linked to Huntingtin’s aggregation and membrane association models. Finally, the last chapter discusses the optimization of the sOPEP and aaOPEP forcefield. The coarse grained OPEP forcefields were successfully applied to multiple studies, most notably for the simulation of amyloid peptides and the tertiary structure prediction using the PEPFOLD methodology. Vast protein ensemble were generated and used in the optimization of sOPEP resulting in 25% of improvement compared to the previous version on our decoys set. The parametrization is also validated by short molecular dynamic simulations on nine proteins. In parallel, we develop the new aaOPEP, a forcefield that will use the OPEP philosophy but in the all-atom regime. Those potential are included in a flexible molecular dynamic code that will be available to everyone when completed.

The transformation of prion protein (PrP) into its insoluble amyloid form correlates with neurodegeneration in transmissible spongiform encephalopathies. PrP is connected to the neuronal membrane by a covalently-linked glycosylphosphatidylinositol (GPI) anchor. The current study demonstrates that phosphatidylinositol and phosphatidylethanolamine in low concentrations (0.5-50 mu M) stimulate rapid unlimited aggregation of PrP. At a higher concentration (500 mu M), lipid particles prevent the formation of large PrP aggregates and induce an increase in the beta-sheet structure content of PrP protein. Thus, the liberation of PrP from the membrane and its direct interaction with its own GPI moiety, as well as with membrane lipids. can promote the formation of aggregated structures of PrP. The phospholipids studied are also able to upregulate the aggregation of oligomeric PrP forms (12-mers and 36-mers), the neurotoxicity of which has been reported recently. Low phosphatidylinositol concentrations induce these oligomers to form aggregates of smaller size when compared with aggregates formed directly from monomers. The inhibition of extensive aggregation observed at a high concentration of phosphatidylinositol (500 mu M) results in both the formation of amyloids from PrP monomers and the interaction of protein molecules with lipid micelles. Thus, phospholipids are not only involved in the aggregation of prion monomers and their amyloidogenic conversion, but also regulate the aggregative status of prion oligomers already formed. Consequently, depending on their micellar status, phospholipids can either promote amyloidogenic conversion and conserve neurotoxic oligomeric forms (lipid micelles) or mediate the formation of large-size amorphous aggregates (non-micellar phospholipids).