The work leading to this thesis has provided additional information and novel knowledge concerning structure-function relationship in the potato epoxide hydrolase. Epoxide hydrolases are enzymes catalyzing the hydrolysis of epoxides to yield the corresponding vicinal diols. The reaction mechanism proceeds via a nucleophilic attack resulting in a covalent alkylenzyme intermediate, which in turn is attacked by a base-activated water molecule, followed by product release. Epoxides and diols are precursors in the production of chiral compounds and the use of epoxide hydrolases as biocatalysts is growing. The promising biocatalyst StEH1, a plant epoxide hydrolase from potato, has been investigated in this thesis. In paper I the active site residue Glu35, was established to be important for the formation of the alkylenzyme intermediate, activating the nucleophile for attack by facilitated proton release through a hydrogen bond network. Glu35 is also important during the hydrolytic half reaction by optimally orienting the hydrolytic water molecule, aiding in the important dual function of the histidine base. Glu35 makes it possible for the histidine to work as both an acid and a base. In paper II a putative proton wire composed of five water molecules lining a protein tunnel was proposed to facilitate effective proton transfer from the exterior to the active site, aiding in protonation of the alkylenzyme intermediate. The protein tunnel is also proposed to stabilize plant epoxide hydrolases via hydrogen bonds between water molecules and protein. Enzyme variants with modified enantiospecificity for the substrate (2,3-epoxypropyl)benzene have been constructed by in vitro evolution using the CASTing approach. Residues lining the active site pocket were targeted for mutagenesis. From the second generation libraries a quadruple enzyme variant, W106L/L109Y/V141K/I155V, displayed a radical shift in enantioselectivity. The wild-type enzyme favored the S-enantiomer with a ratio of 2:1, whereas the quadruple variant showed a 15:1 preference for the R-enantiomer.

Les protéines fluorescentes (PF) homologues d’AvGFP (Green Fluorescent Protein de la méduse Aequorea victoria) sont des outils incontournables de l’imagerie des processus de la cellule vivante. Leurs performances conditionnent la précision de l’analyse quantitative des signaux de fluorescence. Le développement de nouvelles PF demande donc à la fois de parvenir à une forte brillance tout en contrôlant la réponse de la protéine aux variations des paramètres physico-chimiques de la cellule en fonction de la question biologique étudiée. A ce jour, les PF jaunes disponibles montrent une forte sensibilité au pH. Afin d’élaborer des mutants moins sensibles, deux approches ont été considérées : une première consiste à mieux appréhender l’incidence de la dynamique du réseau de liaisons hydrogène entourant le chromophore sur son équilibre acido-basique. La seconde vise à identifier les facteurs structuraux à l’origine de la brillance particulièrement élevée de nouvelles PF jaunes et jaune-vert provenant d’un ver marin, Branchiostoma lanceolatum.J’ai d’abord mis au point un algorithme recherchant l’ensemble des liaisons hydrogène présentes au sein d’une protéine et qui étudie leur dynamique au cours de simulations par dynamique moléculaire. Il permet leur agrégation en réseaux, l’identification des réseaux connectés à un atome d’intérêt ainsi que le suivi de leur dynamique. Pour validation, cet algorithme a été appliqué à la recherche des réseaux de liaisons hydrogène présents au sein de différents mutants d’AvGFP pour lesquels un transfert de proton à l'état excité a été étudié expérimentalement. Cet algorithme pourra également servir à comprendre de façon dynamique le mécanisme d’autres systèmes biologiques dont la fonction repose sur le transfert de protons.D’autre part, j’ai résolu la structure de la protéine fluorescente jaune naturelle lanYFP de Branchiostoma lanceolatum, particulièrement brillante mais à la structure quaternaire tétramérique. Cette protéine a été rendue monomérique par évolution dirigée, ce qui a donné la protéine mNeonGreen à la fluorescence jaune-vert, protéine désormais étalon dans cette gamme spectrale, et dont j’ai également résolu la structure. Mon étude a permis de rationaliser a postériori l’ensemble des mutations introduites au cours de l’évolution. Enfin, j’ai réalisé une étude du dégât d’irradiation spécifique des rayons X permettant de comprendre le changement remarquable de couleur observé sur les cristaux de mNeonGreen après collecte de données de diffraction.L’ensemble des résultats obtenus au cours de ma thèse permet de proposer un cadre de compréhension à la fois théorique et expérimental des déterminants contrôlant les propriétés de fluorescence des PF jaunes. / Fluorescent Proteins (FPs) homologous to AvGFP (Green Fluorescent Protein from the jellyfish Aequoria victoria) are versatile tools used in live cell imaging. The amount of information that can be derived from the fluorescence signals depends on the spectroscopic performances of the FP. The development of new FPs should focus on both brightness increase and control of the protein response to physicochemical parameter variations within the cell. Current yellow FPs exhibit a strong sensitivity to pH. In order to engineer less sensitive variants, two complementary approaches have been used: the first one consists in studying the influence of the hydrogen bond network dynamics around the chromophore on its protonation state. In the second one, I have sought to identify the structural determinants of the particularly high brightness of newly discovered yellow FPs from a sea worm, Branchiostoma lanceolatum.First, I wrote an algorithm that can identify all hydrogen bonds within a protein and analyse their dynamics along molecular dynamics simulations. It allows for their clustering in networks, the identification of networks connected to a given atom and the monitoring of their dynamics. The method was validated by using the algorithm on various AvGFP mutants for which excited state proton transfer has been experimentally studied. This algorithm should also be useful for the study of other biological systems whose function is based on proton transfer.Besides, I solved the structure of the natural yellow FP lanYFP from Branchiostoma lanceolatum, which is particularly bright, but presents a tetrameric arrangement. This protein was monomerized by directed evolution, which led to the yellow-green FP mNeonGreen, now a benchmark in this spectral range. I also solved the structure of mNeonGreen, which allowed me to rationalize a posteriori the mutations that have been introduced during the evolution process. Finally, I performed a specific radiation damage study in order to explain the remarkable change in colour of mNeonGreen crystals upon X-ray data collection. Altogether, the results of my PhD work provides a theoretical and experimental framework of the determinants that drive the fluorescence properties of yellow FPs.