Cette thèse porte sur le développement de nouvelles approches par marquage de spin suivi par spectroscopie RPE. Cette technique est bien adaptée pour suivre la dynamique structurale des protéines. Son principe repose sur l'insertion d'un radical nitroxyde, en un (ou plusieurs) site(s) choisi(s) d'une protéine et permet de sonder localement la structure de la protéine étudiée grâce aux différentes techniques de RPE (en onde continue et impulsionnelle).Dans une première partie, cette technique a été appliquée à la caractérisation de la dynamique structurale de l'IF1 de levure, un peptide inhibiteur de l'ATP-synthase. L'utilisation des spectroscopies de RPE et de dichroïsme circulaire a permis de montrer qu'IF1 de levure dimérise par sa partie médiane et que la partie C-terminale est désordonnée.La seconde partie est plus méthodologique et a pour but d'étudier et de caractériser un marqueur nouvellement synthétisé afin d'élargir les potentialités du marquage de spin. En effet, cette technique est notamment limitée par la faible diversité spectrale offerte par les sondes disponibles (trois raies). Le nouveau marqueur donne un spectre RPE à six raies grâce à la présence d'un noyau magnétique dans l'environnement du radical. Greffé sur une protéine modèle, nous avons montré que ce nouveau marqueur est tout autant capable de rendre compte de variations structurales qu'un marqueur classique. La superposition des signatures spectrales (trois raies + six raies) montre qu'il est possible de différencier les deux signatures spectrales et de sonder simultanément deux sites d'une protéine et de son partenaire. / This thesis focuses on the development of new approaches for site-directed spin labeling followed by EPR spectroscopy. This technique is well suited to monitor the structural dynamics of proteins. The insertion of a nitroxide radical, in one (or several) selected site(s) of a protein, allows probing the structure of the protein using different EPR spectroscopy approaches (continuous wave and pulsed).In a first part, this technique has been applied to characterize the structural dynamics of the yeast IF1, an inhibitory peptide of the ATP-synthase. Using EPR and circular dichroïsm spectroscopies we showed that yeast IF1 dimerizes by its central part and that the C-terminal part remains disordered.The second part is more methodological and the aim is to study and characterize a newly synthesized spin label in order to expand the potential of site-directed spin labeling. In particular, the technique is limited by the poor spectral diversity offered by the available labels (three lines). The new label gives a six lines EPR spectrum thanks to the presence of a magnetic nucleus in the environment of the radical. Grafted on a model protein, we demonstrated that this new label is as able as classical ones to report on structural variations. The superposition of the spectral signatures (three lines + six lines) showed that it is possible to differentiate the two spectral signatures and to probe two sites of a protein and its partner simultaneously.

Analyses of protein sequences from diverse genomes have revealed the ubiquitous nature of multi-domain proteins. They form up to 70% of proteomes of most eukaryotic organisms. Yet, our understanding of protein structure, folding and evolution has been dominated by extensive studies on single-domain proteins. We provide quantitative treatment and proof for prevailing intuitive ideas on the strategies employed by nature to stabilize otherwise unstable domains. We find that domains incapable of independent stability are stabilized by favourable interactions with tethered domains in the multi-domain context. Natural variations (nsSNPs) at these sites alter communication between domains and affect stability leading to disease manifestation. We emphasize this by using explicit all-atom molecular dynamics simulations to study the interface nsSNPs of human Glutathione S-transferase omega 1. We show that domain-domain interface interactions constrain inter-domain geometry (IDG) which is evolutionarily well conserved. The inter-domain linkers modulate the interactions by varying their lengths, conformations and local structure, thereby affecting the overall IDG. These findings led to the development of a method to predict interfacial residues in multi-domain proteins based on difference evolutionary information extracted from at least two diverse domain architectures (single and multi-domain). Our predictions are highly accurate (∼85%) and specific (∼95%). Using predicted residues to constrain domain–domain interaction, rigid-body docking was able to provide us with accurate full-length protein structures with correct orientation of domains. Further, we developed and employed an alignment-free approach based on local amino-acid fragment matching to compare sequences of multi-domain proteins. This is especially effective in the absence of proper alignments, which is usually the case for multi-domain proteins. Using this, we were able to recreate the existing Hanks and Hunter classification scheme for protein kinases. We also showed functional relationships among Immunoglobulin sequences. The clusters obtained were functionally distinct and also showed unique domain-architectures. Our analysis provides guidelines toward rational protein and interaction design which have attractive applications in obtaining stable fragments and domain constructs essential for structural studies by crystallography and NMR. These studies enable a deeper understanding of rapport of protein domains in the multi-domain context.