L’homme possède ~400 gènes codant pour des récepteurs aux odorants (ROs) qui sont différentiellement activés par un espace virtuellement infini de molécules. Le code combinatoire qui résulte de cette activation permettrait au nez humain de discriminer plus de mille milliards de stimuli olfactifs différents. Mais comment le percept est-il encodé dans la structure d’une molécule ? Pour comprendre comment notre nez décrypte la structure des molécules odorantes, des modèles numériques ont été utilisés pour étudier les principaux protagonistes de l’olfaction : les ROs et les odorants. Ici, l’apprentissage automatique est utilisé pour explorer et exploiter les données déjà existantes sur les ROs. D’autre part, la modélisation moléculaire est employée pour comprendre les mécanismes qui sous-tendent la reconnaissance moléculaire. Dans cette thèse j’ai passé en revue les relations structure-odeur du point de vue d’un chimiste. J’ai ensuite développé un protocole d’apprentissage automatique, qui a été validé pour prédire de nouveaux ligands pour quatre ROs. La modélisation moléculaire a été utilisée pour comprendre la reconnaissance moléculaire des ROs. Notamment, l’existence d’un site vestibulaire conservé dans une classe de ROs a été mis en évidence et le rôle de la cavité de liaison orthostérique dans les ROs a été étudiée. L’application de ces techniques permet de moderniser la déorphanisation guidée par ordinateur. Dans sa globalité, mes travaux ont aussi permis de préparer le terrain pour tester de façon virtuelle le code combinatoire des odeurs, et pour prédire la réponse physiologique déclenchée par ces molécules. Dans son ensemble, ce travail ancre la relation structure-odeur dans l’ère post-génomique, et souligne la possibilité de combiner différentes approches computationnelles pour étudier l’olfaction. / Humans have ~400 genes encoding odorant receptors (ORs) that get differentially activated by a virtually infinite space of small organic molecules. The combinatorial code resulting from this activation could allow the human nose to discriminate more than one trillion different olfactory stimuli. But how is the percept encoded in the structure of a molecule? To understand how our nose decrypts the structure of molecules, numerical models were used to study the main protagonists of olfaction: ORs and odorants. These approaches included machine-learning methods to explore and exploit existing data on ORs, and molecular modeling to understand the mechanisms behind molecular recognition. In this thesis I first review the structure-odor relationships from a chemist's point of view. Then, I explain how I developed a machine learning protocol which was validated by predicting new ligands for four ORs. In addition, molecular modeling was used to understand how molecular recognition takes place in ORs. In particular, a conserved vestibular binding site in a class of human ORs was discovered, and the role of the orthosteric binding cavity was studied. The application of these techniques allows upgrading computer aided deorphanization of ORs. My thesis also establishes the basis for testing computationally the combinatorial code of smell perception. Finally, it lays the groundwork for predicting the physiological response triggered upon odorant stimulation. Altogether, this work anchors the structure-odor relationship in the post-genomic era, and highlights the possibility to combine different computational approaches to study smell.

Les récepteurs couplés aux protéines G (RCPG) constituent la plus grande famille de protéines membranaires du génome humain. Ils transmettent les signaux extracellulaires provenant de plusieurs stimuli comme les odeurs, les ions, les hormones et les neurotransmetteurs, à l'intérieur des cellules. En se liant aux RCPGs, ces molécules contribuent à la stabilisation des changements conformationnels activateurs qui se propagent jusqu'au domaine intracellulaire des récepteurs. Ces derniers engagent ensuite un ou plusieurs effecteurs, comme les protéines G hétérotrimériques et les β-arrestines (βarrs), qui activent une cascade d'événements moléculaires menant à la réponse cellulaire.Récemment, la publication de structures cristallines de RCPGs liant des ligands diffusibles a offert une opportunité de raffiner à une résolution atomique les modèles des mécanismes de transduction des signaux. Dans la première partie de cette thèse, nous avons donc exploré les déterminants de la signalisation du récepteur prototypique β2-adrénergique (β2AR), induite par les β-bloqueurs. En ne tenant compte que de leur efficacités sur le β2AR dans les voies de l'adénylate cyclase (AC) et des protéines kinases activées par les facteurs mitogéniques (MAPK), les β-bloqueurs peuvent être classés en 3 groupes distincts (agoniste inverse AC / agoniste MAPK, antagoniste neutre AC / agoniste MAPK et agoniste inverse AC / agoniste inverse MAPK). Afin de déterminer le lien entre leur efficacité et leur mode de liaison, nous avons réalisé des expériences d'arrimages moléculaires in silico entre des β-bloqueurs de chacun des groupes et la structure cristalline du β2AR liée au carazolol. De manière intéressante, les ligands à l'intérieur d'un groupe partagent un mode de liaison, alors que ceux des ligands entre les groupes divergent, suggérant que le mode de liaison des β-bloqueurs pourrait être utilisé pour prédire leur l'efficacité. En accord avec cette hypothèse, nous avons prédit et confirmé l'efficacité agoniste MAPK du carazolol, un inverse agoniste AC du β2AR se liant au récepteur de manière similaire au groupe inverse agoniste AC / agoniste MAPK. De manière intéressante, le groupement aryl des ligands agonistes inverses agonistes AC / agoniste MAPK, le seul groupement chimique variable de ce groupe, est prédite pour lier la région des 3e et 5e hélices transmembranaires (TM3 et TM5). Nous avons donc émis l'hypothèse que cette région pourrait être un déterminant de l'efficacité de ces ligands. En accord avec cette dernière, la mutation de 2 résidus (T118I, S203A) localisés proches du site de liaison des groupements aryls des β-bloqueurs, prévient l'efficacité agoniste inverse de l'ICI-118551 sur la voie de l'AC sans affecter l'efficacité d'un agoniste, indiquant que cette région est importante pour la transmission de l'effet agoniste inverse, du moins sur la voie de l'AC. Les βarrs sont des protéines d'échafaudage qui coordonnent la formation de complexes avec plusieurs dizaines d'effecteurs de signalisation. Originalement identifiées pour leur rôle dans la désensibilisation et l'internalisation des RCPGs, elles sont aussi d'importants effecteurs de la signalisation des RCPGs indépendante des protéines G hétérotrimériques. Cependant, contrairement aux protéines G hétérotrimériques, il n'existe que peu d'outils pour les étudier. Ainsi, la deuxième partie de la thèse est dédiée au développement d'outils pour l'étude des βarrs. À cette fin, nous avons d'abord tenté de transposer une méthode de mesure de l'interaction entre 2 protéines par la technologie de transfert d'énergie de bioluminescence par résonance (BRET) en microscopie et chez des souris transgéniques afin de mesurer de manière subcellulaire et dans un contexte natif l'engagement de la βarr à des RCPGs. Ainsi, nous avons établi les preuves de principe que le BRET peut être utilisé pour localiser l'interaction entre la βarr et le récepteur de la vasopressine de type 2 (V2R) sur une cellule au microscope et pour détecter l'interaction entre la βarr et le β2AR sur des tissus de souris transgéniques exprimant ces protéines fusionnées avec des partenaires BRET. Finalement, il n'existe aucun inhibiteur pharmacologique ciblant les βarrs. Ainsi, grâce à la combinaison d'approches de criblage virtuel sur un modèle de la structure des βarrs et d'essais de validation cellulaire, nous avons développé un inhibiteur pharmacologique des βarrs. À l'aide de cet outil, nous avons confirmé l'implication des βarrs dans l'activation des MAPK par le V2R, mais aussi montré un nouveau rôle des βarrs dans le recyclage du β2AR. Les connaissances et outils développés dans cette thèse permettront de mieux comprendre les déterminants moléculaires de la signalisation des RCPGs et entre autres, grâce à des nouvelles approches pour étudier le rôle cellulaire et physiologique des βarrs. / G Protein-Coupled Receptors (GPCR) are members of the largest family of membrane protein in the human genome. They transduce the signal from a variety of stimuli like odors, ions, hormones and neurotransmitters, inside the cells. By binding directly to the receptors, these molecules stabilize activating conformational changes that are allosterically propagated through transmembrane to intracellular domains. Effectors like heterotrimeric G protein and β-arrestins (βarrs) are then engaged by activated receptors and trigger a cascade of signalling events leading to a cellular response. Recently, the resolution of the crystal structure of GPCR that bind to freely diffusible ligands provided the opportunity to refine at an atomic level the models describing the mecanisms of receptor signal transduction. In the first section of this thesis, we have explored the determinants of the prototypical β2-adrenergic receptor (β2AR) signalling induced by β-blockers. Given their efficacy on Adenylate Cyclase (AC) and Mitogen-Activated Protein Kinase (MAPK) pathways, β-blockers can be classified within 3 signalling groups (AC inverse agonist / MAPK agonist, AC neutral antagonist / MAPK agonist and inverse agonist for AC and MAPK). In order to gain insight on the relation between their efficacy and binding mode, we performed in silico binding experiments between β-blockers from each group and the β2AR crystal structure bound to carazolol. Interestingly, ligands within a group share similar binding mode in contrast to those of different groups, suggesting that β-blockers binding mode could be used to predict their efficacy. In accordance to this hypothesis, we have predicted and confirmed that carazolol, an AC inverse agonist that bind to β2AR in a similar way than the AC inverse agonist / MAPK agonist group, is indeed an agonist for MAPK pathway. Moreover, aryl chemical function from AC inverse agonist / MAPK agonist ligands, barely the only variable structure feature of this group, was predicted to bind β2AR nearby the transmembrane helices 3 and 5 (TM3 and TM5). We thus have predicted that this region would be a determinant of the AC inverse agonist / MAPK agonist ligand efficacy. Accordingly, we found that mutation of 2 residues (T118I, S203A) close to the aryl moiety binding site prevents inverse agonist efficacy of ICI-118551 on AC pathway, without affecting agonist efficacy, indicating that this receptor region is important for the efficacy of these group of β-blockers, at least on AC inverse agonism.βarrs are scaffolding proteins that coordinate protein complex formation with dozen of signalling effectors. First identified for their role on GPCR desensitization and internalization, βarrs are also an important heterotrimeric G protein independent GPCR signalling effectors. However, in contrast to heterotrimeric G protein, only a few tools are available for their study. Thus, the second section of this thesis aim at developing tools for the study of βarrs. For this purpose, we had attempted to transpose a method to measure protein-protein interaction that use Bioluminescence Resonance Energy Transfer (BRET) technology, in microscopy and in transgenic mice, in order to detect subcellular localization and in a native context the engagement of βarr to RCPGs. Thus, we have established a proof of principle that BRET can be combined with microscopy to locate an interaction between βarr and the type 2 vasopressin receptor (V2R) within a cell. Moreover, we have established a second proof of principle that we can detect βarrs recruitment to β2AR on cells extracted from tissues of transgenic mice expressing these proteins fused to BRET partner. Finally, there is no pharmacological inhibitor of βarrs. Thus, using a combination of virtual screening and cellular validation approches, we have developed the first pharmacological βarrs inhibitor. With this novel tool, we have confirmed the implication of βarrs in V2R-mediated MAPK activation, but also showed a new role of βarrs in β2AR recycling.The finding and the tools presented in this thesis should allow to better understand the molecular determinants of GPCR signalling, and among other things, by proposing new tools to study βarrs cellular and physiological roles.

No description available.

This work was aimed at developing novel tools that utilize HINT, an empirical forcefield capable of quantitating both hydrophobic and hydrophilic (hydropathic) interactions, for implementation in theoretical biology and drug discovery/design. The role of hydrophobicity in determination of macromolecular structure and formation of complexes in biological molecules is undeniable and has been the subject of research across several decades. Hydrophobicity is introduced, with a review of its history and contemporary theories. This is followed by a description of various methods that quantify this all-pervading phenomenon and their use in protein folding and contemporary drug design projects – including a detailed overview of the HINT forcefield. The specific aim of this dissertation is to introduce our attempts at developing new methods for use in the study of antibacterial drug resistance and antiviral drug discovery. Multidrug efflux is commonly regarded as a fast growing problem in the field of medicine. Several species of microbes are known to have developed resistance against almost all classes of antibiotics by various modes-of-action, which include multidrug transporters (a.k.a. efflux pumps). These proteins are present in both gram-positive and gram-negative bacteria and extrude molecules of various classes. They protect the efflux pump-expressing bacterium from harmful effects of exogenous agents by simply evacuating the latter. Perhaps the best characterized mechanism amongst these is that of the AcrA-AcrB-TolC efflux pump. Data is available in literature and perhaps also in proprietary databases available with pharmaceutical companies, characterizing this pump in terms of the minimum inhibitory concentration ratios (MIC ratios) for various antibiotics. We procured a curated dataset of 32 β-lactam and 12 antibiotics of other classes from this literature. Initial attempts at studying the MIC ratios of β-lactam antibiotics as a function of their three dimensional topology via 3D-quantitative structure activity relationship (3D-QSAR) technology yielded seemingly good models. However, this methodology is essentially designed to address single receptor-ligand interactions. Molecules being transported by the efflux pump must undoubtedly be involved in multiple interactions with the same. Notably, such methods require a pharmacophoric overlap of ligands prior to the generation of models, thereby limiting their applicability to a set of structurally-related compounds. Thus, we designed a novel method that takes various interactions between antibiotic agents and the AcrA-AcrB-TolC pump into account in conjunction with certain properties of the drugs. This method yielded mathematical models that are capable of predicting high/low efflux with significant efficiency (>93% correct). The development of this method, along with the results from its validation, is presented herein. A parallel aim being pursued by us is to discover inhibitors for hemagglutinin-neuraminidase (HN) of human parainfluenza virus type 3 (HPIV3) by in silico screening. The basis for targeting HN is explored, along with commentary on the methodology adopted during this effort. This project yielded a moderate success rate of 34%, perhaps due to problems in the computational methodology utilized. We highlight one particular problem – that of emulating target flexibility – and explore new avenues for overcoming this obstacle in the long run. As a starting point towards enhancing the tools available to us for virtual screening in general (and for discovering antiviral compounds in specific), we explored the compatibility between sidechain rotamer libraries and the HINT scoring function. A new algorithm was designed to optimize amino acid residue sidechains, if provided with the backbone coordinates, by generating sidechain positions using the Dunbrack and Cohen backbone-dependent rotamer library and scoring them with the HINT scoring function. This rotamer library was previously used by its developers previously to design a very successful sidechain optimization algorithm called SCWRL. Output structures from our algorithm were compared with those from SCWRL and showed extraordinary similarities as well as significant differences, which are discussed herein. This successful implementation of HINT in our sidechain optimization algorithm establishes the compatibility between this forcefield and sidechain rotamer libraries. Future aims in this project include enhancement of our current algorithm and the design of a new algorithm to explore partial induced-fit in targets aimed at improving current docking methodology. This work shows significant progress towards the implementation of our hydropathic force field in theoretical modeling of biological systems in order to enhance our ability to understand atomistic details of inter- and intramolecular interactions which must form the basis for a wide variety of biological phenomena. Such efforts are key to not only to understanding the said phenomena, but also towards a solid basis for efficient drug design in the future.

Computationally driven drug design/discovery efforts generally rely on accurate assessment of the forces that guide the molecular recognition process. HINT (Hydropathic INTeraction) is a natural force field, derived from experimentally determined partition coefficients that quantifies all non-bonded interactions in the biological environment, including hydrogen bonding, electrostatic and hydrophobic interactions, and the energy of desolvation. The overall goal of this work is to apply the HINT-based atomic level description of molecular systems to biologically important proteins, to better understand their biochemistry – a key step in exploiting them for therapeutic purposes. This dissertation discusses the results of three diverse projects: i) structural modeling of human sphingosine kinase 2 (SphK2, a novel anticancer target) and binding mode determination of an isoform selective thiazolidine-2,4-dione (TZD) analog; ii) structural modeling of human cytomegalorvirus (HCMV) alkaline nuclease (AN) UL98 (a novel antiviral target) and subsequent virtual screening of its active site; and iii) explicit treatment of interfacial waters during protein-protein docking process using HINT-based computational tools. SphK2 is a key regulator of the sphingosine-rheostat, and its upregulation /overexpression has been associated with cancer development. We report structural modeling studies of a novel TZD-analog that selectively inhibits SphK2, in a HINT analysis that identifies the key structural features of ligand and protein binding site responsible for isoform selectivity. The second aim was to build a three-dimensional structure of a novel HCMV target – AN UL98, to identify its catalytically important residues. HINT analysis of the interaction of 5’ DNA end at its active site is reported. A parallel aim to perform in silico screening with a site-based pharmacophore model, identified several novel hits with potentially desirable chemical features for interaction with UL98 AN. The majority of current protein-protein docking algorithms fail to account for water molecules involved in bridging interactions between partners, mediating and stabilizing their association. HINT is capable of reproducing the physical and chemical properties of such waters, while accounting for their energetic stabilizing contributions. We have designed a solvated protein-protein docking protocol that explicitly models the Relevant bridging waters, and demonstrate that more accurate results are obtained when water is not ignored.

Les séquences d’ADN et d'ARN riches en Guanines peuvent adopter des conformations inhabituelles connues sous le nom de G-quadruplexes (G4). Les topologies et les formes de ces structures fascinantes sont très diverses. Les G4 sont stabilisés par la présence de cations monovalents et des liaisons Hydrogène de type Hoogsteen. De petites molécules contribuent également à la formation de formes stables, principalement par des interactions d'empilement π - π. Bien que les G4 soient connus depuis des décennies, l'intérêt de la communauté scientifique a été stimulé par la découverte de leur effet potentiellement inhibiteur sur la télomérase, une transcriptase inverse impliquée dans la transformation maligne de la plupart des cellules cancéreuses. En ce qui concerne la télomérase, le cancer et G4, plusieurs groupes ont été impliqués dans la découverte de nouveaux stabilisateurs G4 qui peuvent indirectement inhiber l'enzyme. Des centaines de ligands ont été identifiés par ce biais au cours de la dernière décennie et c'est encore un domaine très actif. Prenant en compte les avantages et la facilité qu'offre l'identification de nouvelles structures à l'aide de techniques de calcul grâce à des modèles mathématiques simples et reproductibles, nous avons entrepris un criblage à haut débit et à faible coût de calcul afin d’identifier de nouveaux ligands G4. Avec l'utilisation de la modélisation QSAR nous pouvons prédire l’IC50 d'un ensemble de composés congénères. Nous avons également été en mesure de relier les descripteurs moléculaires qui apparaissent dans nos modèles avec des caractéristiques structurales que les études de la littérature scientifique et SAR ont rapportés dans les études précédentes, pour un ensemble de ligands congénères. En outre, nous avons construit des modèles différents utilisant des ensembles non congénères de composés en appliquant une stratégie de consensus et pu identifier six ligands approuvés par la FDA qui stabilisent les structures G4. Par la suite, en appliquant des techniques non linéaires et un processus pour le traitement de la base de données que nous avons contruite à partir de publications antérieures, nous avons effectué un criblage virtuel de plus de 500 000 ligands d'une base de données commerciale de composés. Nous avons pu identifier de nouveaux ligands avec une puissance plus forte que les précédentes, qui peuvent également stabiliser d’autres structures G4 impliqués dans les processus liés au cancer. Ces observations ouvrent un spectre large de possibilités à explorer. Malgré les limites des techniques de modélisation QSAR explorées tout au long de ce travail, nous considérons qu'elles peuvent être combinées et utilisées avec soin pour répondre à la recherche de nouveaux stabilisateurs G4. / DNA and RNA G-rich sequences can adopt unusual arrangements that are known as G-quadruplexes (G4). The topologies and forms of these fascinating structures are very diverse. G4 are stabilized by the presence of monovalent cations and Hoogsteen Hydrogen bonds. Small molecules also contribute to the formation of stable forms mainly via π-π stacking interactions. Although G4s are known for decades, interest in this field started with their potential effect on inhibition of telomerase enzyme, a Reverse Transcriptase involved in the malignant transformation of most cancer cells. With regards to telomerase, cancer and G4, several groups have been involved in the discovery of new G4 stabilizers that would indirectly inhibit the enzyme. Most of the G4 ligands were identified following this paradigm. Hundreds of ligands have been identified during the past decade and this is still a very active field in science. Taking into account the advantages and easiness that offers the identification of new structures using computational techniques we built single and reproducible mathematical models with high screening capacity and low computational cost in order to use them on the identification of G4 ligands. With the use of QSAR modelling we can predict the telIC50 of a congeneric set of compounds. We have also been able to relate the molecular descriptors that appear in ours models with some structural features that scientific literature and SAR studies have reported in previous studies as appropriated for describing the above mentioned activity, also for congeneric set of ligands. Moreover, we built different models using non congeneric sets of compounds applying a consensus strategy and could identify six FDA approved ligands that stabilize G4 structures. Subsequently, by applying nonlinear techniques and a process for the cure of the database proposed for us in previous publications, we have performed a virtual screening of more than 500 000 ligands from a commercial database of compounds, followed of structure-based model in order to reduce the number of candidates. We were able to identify new ligands with stronger potency than the previous ones, which can also stabilize other G4 structures involved in processes related to cancer. These observations open a wide-ranging spectrum of possibilities to be explored. Despite the limitations of the QSAR modelling techniques explored along this work, we consider they can be combined and used carefully to address the search for new G4 stabilizers.

Os receptores dos hormônios da tireóide (TRs) são fatores de transcrição envolvidos na diferenciação celular, metabolismo e funções fisiológicas da maioria dos tecidos. Muitos estudos mostram que diversos efeitos farmacológicos mediados pelos TRs podem ser benéficos na farmacoterapia, especialmente aqueles mediados pelo TR que podem ser úteis em condições médicas importantes como obesidade, hipercolesterolemia e diabetes. Além disso, a descoberta que o TR é a isoforma predominante no coração, mediando a maioria dos efeitos cardiovasculares prejudiciais, estimulou a pesquisa por ligantes seletivos para o TR que poderiam ser utilizados em quadros clínicos importantes com perfil de segurança aceitável. Foi realizado um estudo das relações quantitativas entre a estrutura e atividade (QSAR) de um conjunto de compostos com atividade biológica descrita para TR e TR, que gerou modelos de Holograma QSAR com elevada consistência interna e externa, apresentando bom poder de correlação e predição das propriedades biológicas. Também foi realizado um minucioso estudo de triagem virtual, que propiciou a seleção de 7 compostos que foram adquiridos para terem suas atividades biológicas avaliadas. Ensaios de transfecção e gene repórter foram estabelecidos e utilizados na avaliação da atividade biológica dos compostos selecionados pelo ensaio virtual. Finalmente, um ensaio utilizando ressonância plasmônica de superfície (SPR) foi desenvolvido e utilizado para avaliar a atividade agonista desses compostos, e que pode ser útil para avaliar a atividade de novos ligantes. A técnica de SPR também foi empregada em um cuidadoso estudo da interação do TR com seus correguladores, que incluiu estudos cinéticos e termodinâmicos, propiciando a determinação das taxas cinéticas e parâmetros termodinâmicos para a interação do complexo TR-T3 com peptídeos derivados de dois de seus correguladores. Os resultados obtidos são relevantes e devem ser considerados no planejamento de futuros experimentos utilizando o LBD de TR e agonistas. / The thyroid hormone receptors (TRs) are transcriptional factors involved in cell differentiation, development, metabolism and physiological function of most tissues. Many lines of evidence show that several pharmacological actions of TRs might be beneficial in medical therapy, specially those mediated by TR that target important medical conditions like obesity, hypercholesterolemia and diabetes. Additionally, the findings that TR is the predominant isoform in the heart and mediates most of the TRs deleterious cardiovascular effects, stimulated the research for selective TR ligands which could address important medical needs with an acceptable safety profile. In this PhD thesis, studies of the quantitative structure-activity relationships (QSAR) of a dataset of compounds with reported biologic activity for both TR and TR were performed, and statistically significant Hologram QSAR models with good predictive ability for untested compounds were created. In parallel, a careful virtual screening procedure was executed, leading to the selection of 7 compounds which were purchased for the evaluation of their biological activities. Cell transfection and reporter gene assays were developed, validated and used to evaluate the biological activities of these compounds. Finally, a surface plasmon resonance (SPR) assay was developed and used to assess the agonistic activity of these compounds. The SPR technique was also employed in a careful study of the interaction between the ligand binding domain of TR and peptides derived from its coregulators, which included the determination of the kinetic and thermodynamic parameters for this interaction. The results suggest that flexibility plays an important role in the interaction between the receptor and its coregulators, and point out important aspects of experimental design that should be addressed when using TR LBD and its agonists. Furthermore, the methodology described here may be useful for the identification of new TR ligands.

No description available.

Les biothérapies (anticorps monoclonaux, récepteurs solubles) ciblant les cytokines IL-6 etTNF-alpha pour le traitement des maladies inflammatoires chroniques ont constitué un succèsmajeur de l’industrie pharmaceutique. Elles présentent néanmoins des inconvénientsimportants : résistances, mode d’administration contraignant, coût élevé.Notre équipe travaille sur l’identification de petites molécules inhibant directement cescytokines, afin d’élargir l’offre thérapeutique existante. Administrées par voie orale, ellesconstitueraient une alternative particluièrement favorable aux patients.Durant ma thèse, j’ai réalisé le criblage expérimental (tests cellulaires et tests biochimiquesde liaison) des meilleurs composés identifiés par criblage virtuel d’un grande chimiothèque dediversité, ainsi que de composés dérivés de pyridazine issus d’une chimiothèque médicinale. J’aiainsi pu identifier plusieurs inhibiteurs directs du TNF-alpha et de l’IL-6. De plus, mon travail apermis d’affiner les procédures de criblage du Laboratoire.Ces travaux ouvrent de nouvelles pistes pour le développement de médicaments anti-cytokines. / Anti-cytokine biologics (monoclonal antibodies, soluble receptors) targeting TNF-alpha and IL-6in chronic inflammatory diseases have been a major success for pharmaceutical industry.However, they exhibit several drawbacks : resistance, difficult administration, high costs.Our team works on the discovery of small molecule inhibitors of cytokines suck as TNF-alphaand IL-6, in order to widen the range of therapeutic drugs. Orally active drugs would represent ahighly beneficial alternative for patients.During my PhD, I have performed an experimental screening (using cellular and biochemicalbinding testings) of the best compounds identified through virtual screening of a large chemicallibrary, and on pyridazine compounds of a medicinal chemical library. I have been able toidentify several small molecules inhibiting the interaction of TNF-! and IL-6 with their receptor.Moreover, my work will have an impact on the laboratory screening strategies.Overall, this work opens new avenues for anti-cytokine drug discovery.

No description available.