Una de les àrees més prometedores en recerca biomèdica i farmacèutica és el disseny molecular computacional, que intenta establir relacions entre propietats físico-químiques i activitat biològica. L'èxit d'aquestes tècniques depen críticament de la qualitat de la descripció molecular. En aquest sentit, metodologies basades en potencials d'interacció molecular (MIP) són eines útils per la comparació de compostos que presenten comportaments biològics semblants. Aquest projecte desenvolupa eines per comparar molècules basades en la caracterització de llurs MIPs. El programa de similaritat molecular MIPsim ha estat desenvolupat i aplicat a diferents problemes biològics. Aquesta tesi consisteix en quatre estudis científics que mostren l'ús del MIPSim en aliniament molecular, catalisi enzimàtica, en acoratge de molècules dins el lligand i en estudis 3D-QSAR. / One of the most promising areas in biomedical and pharmaceutical research is computer assisted molecular design, which tries to stablish relationships between physicochemical properties and biological activity. The success of these techniques depends critically on the quality of the molecular description. In this sense, methodologies based on molecular interaction potentials (MIP) are useful tools for the comparison of compounds displaying related biological behaviours. This project aims to develop tools to compare 'molecules based on the characterization 'of their MIPs. To this end, the molecular similarity program MIPSim has been further developed and applied to different biological problems. This thesis consists on four scientific studies showing the use of MIPSim for molecular alignment, enzymatic catalysis, ligand-protein docking and 3D-QSAR analyses.

Macromolecular interactions play very important roles in regulation of all levels of biological processes. Aberrant macromolecular interactions often result in diseases. By applying a combination of spectroscopy, calorimetry, computation and other techniques, the protein-protein interactions in the system of the Shaw2 Kv channel and the protein-RNA interactions in West Nile virus RNA-cellular protein TIAR complex were explored. In the former system, the results shed light on the local structures of the key channel components and their potential interaction mediated by butanol, a general anesthetic. In the later studies, the binding modes of TIAR RRM2 to oligoU RNAs and West Nile virus RNAs were investigated. These findings provided insights into the basis of the specific cellular protein–viral RNA interaction and preliminary data for the development of strategies on how to interfere with virus replication

Ce manuscrit présente une étude fonctionnelle et structurale du cytochrome P450 2J2 humain (CYP2J2), enzyme exprimée dans les tissus cardiovasculaires, dont les rôles biologiques sont mal connus. En utilisant la terfénadone comme base structurale, qui est un composé oxydé régiosélectivement par le CYP2J2, plusieurs composés ont été synthétisés se sont révélés être des inhibiteurs affins et sélectifs du CYP2J2. En particulier, un inhibiteur très affin et compétitif (Ki = 160 nM) et deux substrats suicides efficaces du CYP2J2 (kinact/Ki 3000 L/mol/s) ont été mis en évidence. L'étude de l'oxydation de ces composés par le CYP2J2 a révélé une régiosélectivité surprenante, en faveur d'une position chimiquement moins réactive vis-à-vis des oxydations. La caractérisation du site actif du CYP2J2 et l'identification de résidus importants pour la reconnaissance des dérivés de terfénadone a pu être réalisée en construisant un modèle par homologie 3D de cette enzyme et par le docking de certains dérivés dans le site actif du CYP2J2. Enfin, une étude préliminaire des rôles biologiques possibles du CYP2J2 a été réalisée en étudiant les effets inhibiteur de ce P450. En conclusion, ce travail a permis de caractériser les premiers outils biochimiques d'étude des rôles biologiques du CYP2J2, de proposer une première topologie du site actif du CYP2J2, et d'affiner les rôles biologiques du CYP2J2.

To be able to make informed descicions regarding the research of new drug molecules (ligands), it is crucial to have access to information regarding the chemical interaction between the drug and its biological target (protein). Computer-based methods have a given role in drug research today and, by using methods such as molecular docking, it is possible to investigate the way in which ligands and proteins interact. Despite the acceleration in computer power experienced in the last decades many problems persist in modelling these complicated interactions. The main objective of this thesis was to investigate and improve molecular modelling methods aimed to estimate protein-ligand binding. In order to do so, we have utilised chemometric tools, e.g. design of experiments (DoE) and principal component analysis (PCA), in the field of molecular modelling. More specifically, molecular docking was investigated as a tool for reproduction of ligand poses in protein 3D structures and for virtual screening. Adjustable parameters in two docking software were varied using DoE and parameter settings were identified which lead to improved results. In an additional study, we explored the nature of ligand-binding cavities in proteins since they are important factors in protein-ligand interactions, especially in the prediction of the function of newly found proteins. We developed a strategy, comprising a new set of descriptors and PCA, to map proteins based on their cavity physicochemical properties. Finally, we applied our developed strategies to design a set of glycopeptides which were used to study autoimmune arthritis. A combination of docking and statistical molecular design, synthesis and biological evaluation led to new binders for two different class II MHC proteins and recognition by a panel of T-cell hybridomas. New and interesting SAR conclusions could be drawn and the results will serve as a basis for selection of peptides to include in in vivo studies.

Rheumatoid arthritis (RA) is an autoimmune inflammatory disease that leads to degradation of cartilage and bone mainly in peripheral joints. In collagen-induced arthritis (CIA), a mouse model for RA, activation of autoimmune CD4+ T cells depends on a molecular recognition system where T-cell receptors (TCRs) recognize a complex between the class II MHC Aq protein and CII259-273, a glycopeptide epitope from type II collagen (CII). Interestingly, vaccination with the Aq/CII259-273 complex can relieve symptoms and cause disease regression in mice. This thesis describes the use of modified glycopeptides to explore interactions important for binding to the Aq protein and recognition by autoimmune T-cell hybridomas obtained from mice with CIA. The CII259-273 glycopeptide was modified by replacement of backbone amides with different amide bond isosteres, as well as substitution of two residues that anchor the glycopeptide in prominent pockets in the Aq binding site. A three-dimensional structure of the Aq/glycopeptide complex was modeled to provide a structural basis for interpretation of the modified glycopeptide’s immunological activities. Overall, it was found that the amide bond isosteres affected Aq binding more than could be explained by the static model of the Aq/glycopeptide complex. Molecular dynamics (MD) simulations, however, revealed that the introduced amide bond isosteres substantially altered the hydrogen-bonding network formed between the N-terminal 259-265 backbone sequence of CII259-273 and Aq. These results indicated that the N-terminal hydrogen-bonding interactions follow a cooperative model, where the strength and presence of individual hydrogen bonds depended on the neighboring interactions. The two important anchor residues Ile260 and Phe263 were investigated using a designed library of CII259-273 based glycopeptides with substitutions by different (non-)natural amino acids at positions 260 and 263. Evaluation of binding to the Aq protein showed that there was scope for improvement in position 263 while Ile was preferred in position 260. The obtained SAR understanding provided a valuable basis for future development of modified glycopeptides with improved Aq binding. Furthermore, the modified glycopeptides elicited varying T-cell responses that generally could be correlated to their ability to bind to Aq. However, in several cases, there was a lack of correlation between Aq binding and T-cell recognition, which indicated that the interactions with the TCRs were determined by other factors, such as presentation of altered epitopes and changes in the kinetics of the TCR’s interaction with the Aq/glycopeptide complex. Several of the modified glycopeptides were also found to bind well to the human RA-associated DR4 protein and elicit strong responses with T-cell hybridomas obtained from transgenic mice expressing DR4 and the human CD4 co-receptor. This encourages future investigations of modified glycopeptides that can be used to further probe the MHC/glycopeptide/TCR recognition system and that also constitute potential therapeutic vaccines for treatment of RA. As a step towards this goal, three modified glycopeptides presented in this thesis have been identified as candidates for vaccination studies using the CIA mouse model.

Le triméthoprime (TMP) est un antibiotique communément utilisé depuis les années 60. Le TMP est un inhibiteur de la dihydrofolate réductase (DHFR) bactérienne chromosomale. Cette enzyme est responsable de la réduction du dihydrofolate (DHF) en tétrahydrofolate (THF) chez les bactéries, qui lui, est essentiel à la synthèse des purines et ainsi, à la prolifération cellulaire. La résistance bactérienne au TMP est documentée depuis plus de 30 ans. Une des causes de cette résistance provient du fait que certaines souches bactériennes expriment une DHFR plasmidique, la DHFR R67. La DHFR R67 n'est pas affectée par le TMP, et peut ainsi remplacer la DHFR chromosomale lorsque celle-ci est inhibée par le TMP. À ce jour, aucun inhibiteur spécifique de la DHFR R67 est connu. En découvrant des inhibiteurs contre la DHFR R67, il serait possible de lever la résistance au TMP que la DHFR R67 confère aux bactéries. Afin de découvrir des inhibiteurs de DHFR R67, les approches de design à base de fragments et de criblage virtuel ont été choisies. L'approche de design à base de fragments a permis d'identifier sept composés simples et de faible poids moléculaire (fragments) inhibant faiblement la DHFR R67. À partir de ces fragments, des composés plus complexes et symétriques, inhibant la DHFR R67 dans l'ordre du micromolaire, ont été élaborés. Des études cinétiques ont montré que ces inhibiteurs sont compétitifs et qu'au moins deux molécules se lient simultanément dans le site actif de la DHFR R67. L'étude d'analogues des inhibiteurs micromolaires de la DHFR R67 a permis de déterminer que la présence de groupements carboxylate, benzimidazole et que la longueur des molécules influencent la puissance des inhibiteurs. Une étude par arrimage moléculaire, appuyée par les résultats in vitro, a permis d'élaborer un modèle qui suggère que les résidus Lys32, Gln67 et Ile68 seraient impliqués dans la liaison avec les inhibiteurs. Le criblage virtuel de la librairie de 80 000 composés de Maybridge avec le logiciel Moldock, et les essais d'inhibition in vitro des meilleurs candidats, a permis d'identifier quatre inhibiteurs micromolaires appartenant à des familles distinctes des composés précédemment identifiés. Un second criblage virtuel, d'une banque de 6 millions de composés, a permis d'identifier trois inhibiteurs micromolaires toujours distincts. Ces résultats offrent la base à partir de laquelle il sera possible de développer iv des composés plus efficaces et possédant des propriétés phamacologiquement acceptables dans le but de développer un antibiotique pouvant lever la résistance au TMP conféré par la DHFR R67. / Trimethoprim (TMP) is a common antibiotic which is used since the 60's. TMP is an inhibitor of the bacterial chromosomal dihydrofolate reductase (DHFR). This enzyme catalyses the reduction of the dihydrofolate (DHF) to tetrahydrofolate (THF) which is essential to the biosynthesis of purines thus to cellular proliferation. Bacterial TMP resistance is documented since about 30 years. One of the cause of this resistance comes from the fact that certain bacteria express a plasmidic DHFR, the R67 DHFR, which confers TMP resistance. The R67 DHFR is not inhibited by TMP and can replace the chromosomal DHFR when the latter is inhibited by TMP. The discovery of R67 DHFR inhibitors would allow to break the trimethoprim resistance granted by R67 DHFR. In order to discover R67 DHFR inhibitors, fragment based design and virtual screening approaches were selected. By fragment based design, seven simple compounds with a low molecular mass which inhibited weakly R67 DHFR (fragments) were identified. From these fragments, more complex and symmetrical compounds inhibiting R67 DHFR in the micromolar range were identified. Kinetic studies showed these inhibitors were competitive and at least two molecules bind simultaneously to the active site of the R67 DHFR. Test of the micromolar inhibitors analog showed that the presence of carboxylate, benzimidazole and the length of the molecule all have an effect on the potency of the inhibitors. Molecular docking of the inhibitors, supported by in vitro data, were used to develop a model which suggest that residue like Lys32, Gln67 and Ile68 would be involved in the binding of the inhibitors to the R67 DHFR. Virtual screening of the 80 000 compound Maybridge library with Moldock software, followed by in vitro test of the best candidate, identified four micromolar inhibitors which are chemically distinct from the inhibitor beforehand identified. A second virtual screening of a 6 million compounds bank identified three micromolar inhibitors which are also distinct from the inhibitor beforehand identified. vi These results offer a basis which will allow further development of more potent inhibitors with more acceptable pharmacologic properties in order to develop an antibiotic which would break the TMP resistance granted by the R67 DHFR.

Thrombosis is the most important pathological process underlying many cardiovascular diseases, which are responsible for high mortality worldwide. In this theses the computer-aided design of new anti-thrombotic agents able to inhibit two types of receptors located on the surface of the platelets has been applied. The first one - αIIbβ3 - is responsible for the interaction of activated platelets with fibrinogen to form clots, whereas the second one - thromboxane A2 - is responsible for platelet activation by one of agonists excreted by activated platelets. To achieve this, different types of models have been developed using experimentally available information and structure of protein-ligand complexes. This concerns: QSAR models, structure-based and ligand-based 3D pharmacophore models, 2D pharmacophore models, shape-based and molecular field-based models. The ensemble of the developed models were used in virtual screening. This study resulted in suggestion of new potential antagonists of αIIbβ3 and thromboxane A2 receptors. Suggested antagonists of αIIbβ3 able to bind either open or closed form of the receptor have been synthesized and tested experimentally. Experiments show that they display high activity; moreover some of theoretically designed compounds are more efficient than Tirofiban - the commercialized drug molecule. The recommended antagonists of thromboxane A2 receptor have been already synthesized but biological tests have not been completed yet.

La chémoinformatique et la bioinformatique sont des disciplines devenues indispensables à la découverte de médicaments. De nos jours, les industries pharmaceutiques consacrent près de 10% de leur budget de recherche et développement, à la recherche de médicaments assisté par ordinateur (Kapetanovic 2008). Cette émergence peut s'expliquer à la fois par le développement des architectures de calculs mais aussi par le faible coup qu'engendrent des analyses in silico par rapport à des tests in-vitro.Les essais biologiques qui ont été menés depuis des décennies afin d'identifier des médicaments potentiels, commencent à former une source très importante de données et plusieurs bases de données commencent à les répertorier. La disponibilité de ce type de données a favorisé le développement d'un nouvel axe de recherche appelé la "chémogénomique" et qui s'intéresse à l'étude et à l'identification des associations possibles entre plusieurs molécules et plusieurs cibles. Ainsi, la chémogénomique permet de déterminer le profil biologique d'une molécule et nous renseigne sur sa capacité à devenir une touche intéressante mais aussi à identifier ses possibles effets indésirables. Des méthodes de chémoinformatique permettent d'utiliser ces sources de données à des fins d'apprentissage et établir des modèles prédictifs qui permettront par la suite de faire des prédictions pour connaitre l'activité d'une molécule.Cette thèse a porté sur le développement et l'utilisation de méthodes de prédictions d'association protéine-ligand. La prédiction d'une association est importante en vue d'un criblage virtuel et peut s'effectuer à l'aide de plusieurs méthodes. Au sein du laboratoire, on s'intéresse plus particulièrement au profilage de bases de données de molécules (chimiothèques) contre une série de cibles afin d'établir leur profil biologique. J'ai donc essayé au cours de ma thèse de mettre au point des modèles prédictifs d'association protéine-ligand pour un grand nombre de cibles, valider des méthodes de criblage virtuel récentes à des fins de profilage mais aussi établir un protocole de profilage automatisé, qui décide du choix de la méthode de criblage la plus adaptée en s'appuyant sur les propriétés physico-chimiques du ligand à profiler et de l'éventuelle cible.

Cette thèse a pour but de développer le pendant in silico des étapes clés du Fragment-Based Drug Design (FBDD), et ce dans le cadre plus général du développement de l'outil S4MPLE. Le FBDD génère des ligands drug-like à partir de petites molécules (fragments). Après une étape de validation de S4MPLE et de sa fonction d'énergie, un recentrage autour du FBDD est réalisé, à travers le docking puis l'optimisation virtuelle de fragments par growing ou linking (G/L). Cette stratégie reposesur 1) la création d'une chimiothèque focalisée en connectant un ou deux fragment(s) avec des linkers pré-générés, et 2) l'échantillonnage avec S4MPLE des composés chimères dans le site avec des contraintes. Des simulations de G/L plus ou moins ambitieuses (site flexible, ajout de H2O libres) permettent de valider cette approche avec des études rétrospectives basées sur des données expérimentales. La dernière phase de la thèse a consisté à appliquer ce protocole in silico à un projet de l'entreprise.

Directed evolution is generally regarded as a useful approach in protein engineering. By subjecting members of a mutant library to the power of Darwinian evolution, desired protein properties are obtained. Numerous reports have appeared in the literature showing the success of tailoring proteins for various applications by this method. Is it a one-way track that protein practitioners can only learn from nature to enable more efficient protein engineering? A structure-and-mechanism-based approach, supplemented with the use of reduced amino acid alphabets, was proposed as a general means for semi-rational enzyme engineering. Using human GST A2-2*E, the most active human enzyme in the bioactivation of azathioprine, as a parental enzyme to test this approach, a L107G/L108D/F222H triple-point mutant of GST A2-2*E (thereafter designated as GDH) was discovered with 70-fold increased activity, approaching the upper limit of specific activity of the GST scaffold. The approach was further experimentally verified to be more successful than intuitively choosing active-site residues in proximity to the bound substrate for the improvement of enzyme performance. By constructing all intermediates along all putative mutational paths leading from GST A2-2*E to mutant GDH and assaying them with nine alternative substrates, the fitness landscapes were found to be “rugged” in differential fashions in substrate-activity space. The multidimensional fitness landscapes stemming from functional promiscuity can lead to alternative outcomes with enzymes optimized for other features than the selectable markers that were relevant at the origin of the evolutionary process. The results in this thesis suggest that in this manner an evolutionary response to changing environmental conditions can readily be mounted. In summary, the thesis demonstrates the attractive features of the structure-and-mechanism-based semi-rational directed evolution approach for optimizing enzyme performance. Moreover, the results gained from the studies show that laboratory evolution may refine our understanding of evolutionary process in nature.