It is predicted that by 2050 there will be 10 million deaths annually due to super-resistant bacterial infections. Antimicrobial resistance (AMR) is already responsible for nearly 5 million deaths a year. Ribosomes serve as an ideal drug target being frequently targeted by antibiotics and having a highly conserved structure with few options for resistance. However, computer aided drug design (CADD) using ribosome crystal structures presents several challenges and is underutilized in the field. In this work we establish a successful protocol for antibiotic redocking and docking within the high interest sites of the peptidyl transferase center (PTC). Molecular visualization and interaction mapping were used to atomistically delineate binding patterns in the ribosomal PTC that could be used for CADD. Eleven ribosome crystal structures were validated for computational testing, which revealed derivative binding patterns in the A-site and P-site that can be used to increase antibiotic efficacy. Ribosome overlays revealed high interaction frequency nucleotides that were widely conserved throughout the different species and could be used to inform bidentate design to target two pockets at once. This work serves as a basis for methods to computationally explore drug optimization on ribosome targeting antibiotics to help combat the rapid expansion of AMR. / Master of Science in Life Sciences / Antimicrobial resistance (AMR) to antibiotics by bacteria is a rapidly increasing problem. Current trends predict that there will be more death due to super-resistant bacterial strains than cancer by 2050. Ribosomes are essential cellular machinery for bacteria and make an ideal antibiotic target. Using computational tools to optimize antibiotics with available ribosome crystal structural data presents several challenges and is underutilized throughout the field. In this work we establish a successful protocol for determining and exploiting antibiotic binding patterns within the functional center of the ribosome, the peptidyl transfer center (PTC). Nearly a dozen ribosome crystal structures were validated for computational testing, and binding patterns were revealed within the PTC that allowed antibiotic derivatives with increased efficacy to be developed. Ribosome validation also helped inform new drug class design so that multiple drug sites could be targeted at once, which were docked sharing high frequency nucleotide interactions with both parent antibiotics. This work serves as a basis for methods to computationally explore drug optimization on ribosome targeting antibiotics to help combat the rapid expansion of AMR.

<p dir="ltr">The technological advancements in marine robotics led to the expansion of the autonomous underwater vehicle (AUV) fleet. Depending on the applications, the type of the AUV ranges across various shapes and sizes. It seeks a solution for the issue of limited power capacity, often in terms of underwater docking systems. Underwater docking poses a significant challenge for AUVs, especially when considering the diverse shapes and sizes of these vehicles. Existing solutions usually are task specific, and do not address the idea of scalable underwater docking system design.<br>This thesis investigates the adaptability of the specific docking system design, previously validated for torpedo-shaped AUVs, to boxed-shaped AUVs in a nonlinear open water environment. In order to achieve this goal, the scalability of the docking system design of choice was tested in an open water non-linear underwater environment and validated. The scalability of the robust docking system was adapted to the box-shaped AUV, encompassing path planning, path following, and docking maneuver. The adapted docking system was based on the optic methods for docking station detection and subsequent docking. Additionally, the simulated environment was developed for the AUV model, for testing and debugging purposes. In the simulation, a custom PID controller was developed along with integrating the navigation and guidance package, to fully simulate the real life behavior of the AUV. </p><p dir="ltr">Furthermore, this work introduces a recurrent neural network-based architecture for investigating temporal dependencies of the sequential data input. The proposed architecture is based on CNN for spatial feature extraction and LSTM/GRU for temporal feature detection. The dataset collection is based on the simulation environment, by enhancing the artificial images with imposed realism. The dataset was gathered on different levels of turbidity and the collection process was automated.</p>

Dans le système olfactif des moustiques, les protéines liants les molécules odorantes ou odorant binding proteins (OBPs) interviennent dans les toutes premières étapes permettant d'aboutir à la reconnaissance de leurs hôtes et font l'objet d'un intérêt croissant dans les recherches sur la transmission des maladies infectieuses par ces insectes. Le travail présenté a pour objet d'approfondir les connaissances sur ces OBPs dans trois génomes de moustiques, tous vecteurs de maladies infectieuses : Anopheles gambiae, Aedes aegypti et Culex quinquefasciatus. Une analyse à l'échelle de ces génomes a été réalisée et a permis d'identifier un nombre important de nouveaux gènes d'OBPs notamment chez les espèces de moustiques Aedes aegypti et Culex quinquefasciatus. Complétée par une étude phylogénétique du répertoire complet de ces gènes dans les trois génomes étudiés, cette analyse a permis d'établir une nouvelle classification des sous familles des OBPs. Ce résultat démontre l'extraordinaire multiplicité et diversité des gènes impliqués dans l'olfaction chez ces espèces de moustiques tout en mettant en lumière certaines propriétés des séquences des OBPs qui sont hautement conservés chez les moustiques. Grâce à la disponibilité de certaines structures d'OBPs de moustiques ou d'autres insectes apparentées, des modèles structuraux de tous les OBPs de la sous famille dites Classic dans les trois génomes, soit au total 137 structures, ont été construits. Ces structures ont servi de base pour le criblage à grande échelle par docking moléculaire d'une chimiothèque de 126 molécules odorantes connues pour leurs propriétés attractives ou répulsives vis-à-vis des moustiques. Ces résultats fournissent pour la première fois, les bases structurales et fonctionnelles pour la compréhension au niveau moléculaire de l'efficacité de certains agents répulsifs tout comme de l'attractivité de certains agents provenant des émanations humaines. Par simulation de dynamique moléculaire, les changements qui s'opèrent dans une de ces OBPs lorsque celle ci, liée à une molécule odorante, se retrouve dans des conditions de pH modifiée ont été caractérisée et un mécanisme probable par lequel ces OBPs participeraient à la reconnaissance et la libération des molécules odorantes est proposée. Cette thèse fournit des éléments de réponses importants quant à la caractérisation génomique, structurale et fonctionnelle des OBPs de moustiques et peut servir de base de départ pour des recherches expérimentales plus approfondies sur ces aspects. / The role of odorant binding proteins in the olfaction of mosquitoes, the primary mechanism of human host recognition, has been an important focus of biological research in the field of infectious disease transmission by these insects. This thesis provides an in depth knowledge of these proteins in three mosquito species Anopheles gambiae, Aedes aegypti and Culex quinquefasciatus. A large scale analysis on these genomes has been carried out towards the identification of the odorant binding proteins in the mosquito genomes. Identification of many new OBP members, in particular in the Aedes aegypti and Culex quinquefasciatus species, and an extensive phylogenetic analysis presenting a novel classification of the OBP subfamilies of these mosquito species has been proposed. This results further demonstrates the extraordinary multiplicity and diversity of the OBP gene repertoire in these three mosquito genomes and highlights the striking sequence features that are nevertheless highly conserved across all mosquito OBPs. Owing to the availability of homologous structures from mosquitoes or related species, the 3D structure modelling of all the Classic OBPs from the three genomes (representing in total 137 structures) has been performed. This was completed by large scale docking studies on these structures by screening a large set of compounds that are known to be mosquito attractants or repellents. These provide many exciting new insights into the structural and functional aspects towards understanding the efficacy of some repellents and of some attractants from human emanations. Through molecular dynamics simulation, the structural changes observed in an OBP bounded to an odorant when pH conditions are modified were characterized and the probable mechanism of ligand binding and release is presented. This work provides the first insights to many of the long awaited questions on the genomic, structural and functional characterization of mosquito OBPs and can be viewed as a reliable starting point for further experimental research focussed on these aspects.

A hipercolesterolemia familial (HF) é uma doença autossômica dominante considerada como uma das formas mais graves de hiperlipidemia, assim como, a principal causa de morbi-mortalidade por ser o principal fator desencadeante da aterosclerose. A alteração primária e mais freqüente da HF incide no gene do receptor da LDL (LDLr), sabe-se que mais de 1600 mutações são descritas na literatura e a principal consequência dessas alterações resultam no comprometimento da remoção da LDL, aumentando a concentração plasmática. Atualmente, o ultrasequenciamento genômico permite gerar muitos dados, que podem identificar novas mutações gênicas de forma eficiente, reprodutiva e rápida. No entanto, somente a validação da nova mutação por atividade funcional pode realmente estabelecer a associação com a doença. O presente estudo tem como objetivo realizar a análise da atividade do receptor da LDL, identificadas através do sequenciamento de alto rendimento, no gene LDLr realizado pelo nosso grupo de pesquisa e correlacionar com dados clínicos, in vitro, in silico e estrutural. Para cumprir esta meta, os linfócitos T dos portadores de HF foram isolados do sangue periférico, cultivados e submetidos a estímulo para a expressão de receptores da LDL, incubados com LDL marcada para avaliação de ligação e interiorização pelas células de cada paciente. Dos 30 pacientes selecionados para esse estudo, 63% apresentaram mutação no LDLR, sendo que quase todas as variantes (p.Gly373Asp, p.Asp601His, p.Ile488Thr, p.Gly549Asp, p.Gly592Glu e Gly681Asp) são localizadas no segundo domínio entre os éxons 7 ao 14. De acordo com o docking molecular a variante p.Gly592Glu (rs137929307), que já foi identificada na população polonesa, espanhola e brasileira, já relacionada com a HF, pode aumentar a interação do LDLr com a ApoB e consequentemente o modo de interação entre as proteínas, no estudo in vitro foi possível notar um aumento tanto na média de fluorescência da ligação e da ligação e interiorização em relação a quantidade de LDLr na superfície celular. / Familial hypercholesterolemia (HF) is an autosomal dominant disease considered as one of the most severe forms of hyperlipidemia, as well as the main cause of morbidity and mortality because it is the main triggering factor for atherosclerosis. The primary and more frequent alteration of the HF affects the LDL receptor gene (LDLr), it is known that more than 1600 mutations are described in the literature and the main consequence of these alterations results in the compromise of the LDL removal, increasing the plasma concentration. Nowadays, genomic ultrasequencing allows the generation of many data, which can identify new gene mutations efficiently, reproductively and rapidly. However, only the validation of the new functional activity mutation can actually establish association with the disease. The aim of the present study was to analyze LDL receptor activity, identified by high-throughput sequencing, in the LDLr gene performed by our research group and to correlate with clinical, in vitro, in silico and structural data. To meet this goal, the T lymphocytes from the HF carriers were isolated from the peripheral blood, cultured and challenged for the expression of LDL receptors, incubated with labeled LDL for binding assessment and internalization by the cells of each patient. Of the 30 patients selected for this study, 63% had a mutation in LDLR, and almost all variants (p.Gly373Asp, p.Asp601His, p.Ile488Thr, p.Gly549Asp, p.Gly592Glu and Gly681Asp) are located in the second domain between exons 7 to 14. According to the molecular docking the variant p.Gly592Glu (rs137929307), which has already been identified in the Polish, Spanish and Brazilian population, already related to HF, can increase the interaction of LDLr with ApoB and consequently the mode of interaction between proteins, in the in vitro study it was possible to note an increase in both the mean fluorescence of binding and binding and internalization in relation to the amount of LDLr on the cell surface.

Os insetos são os mais numerosos animais encontrados no mundo, com mais de 675 mil espécies conhecidas. Pertencentes à ordem Hymenoptera, da superfamília Apoidea, as abelhas são encontradas distribuídas em aproximadamente 20 mil espécies. No Brasil estima-se que existam 1.700 espécies. Uma das principais espécies é a Apis mellifera, com ocorrência cosmopolita. A Apis mellifera, popularmente conhecida como abelha africanizada, é agressiva, enxameia várias vezes ao ano e utiliza uma grande variedade de locais para nidificar. Esse comportamento aumenta o contato direto entre o inseto e a população, aumentando o número de acidentes. Os acidentes com abelhas representam um problema de saúde pública em diversos países do mundo pela freqüência com que ocorrem e pela mortalidade que ocasionam. O presente estudo propõe a busca por inibidores naturais contra o veneno de abelhas. Um sistema e uma base de dados foram desenvolvidos para a integração entre dados de plantas medicinais antivenenos e os venenos de abelhas. As atividades anti-hemorrágica, anti-proteolítica, anti-miotóxica, antifosfolipase e anti-edema de plantas medicinais antiveneno foram analisadas por meio de ensaios farmacológicos. As possíveis interações entre as toxinas Melitina e Fosfolipase A2 com inibidores foram avaliadas, através do docking virtual. O banco de dados, denominado Bee Venom, foi implementado e os dados de bancos de dados públicos foram inseridos no sistema. O sistema foi liberado para acesso público no endereço eletrônico http://gbi.fmrp.usp.br/beevenom/. Durante a análise da proteína Melitina foram encontradas as regiões da proteína em que os possíveis inibidores devem interagir e identificadas as propriedades químicas que os inibidores devem possuir para interagir corretamente com a Melitina. Nas análises in silico foi possível identificar 10 possíveis inibidores que interagiram corretamente com o sítio ativo da Fosfolipase A2. Algumas espécies do Banco de Germoplasma da FMRP/USP foram obtidas e utilizadas nos experimentos de atividade fosfolipásica indireta e de Edema, sendo possível observar inibição do veneno total e da proteína Fosfolipase A2. Os compostos sintéticos e inibidores avaliados não causaram inibição em todos os experimentos avaliados. Já as plantas obtidas no laboratório de Toxinas Animais e Inibidores Naturais e Sintéticos causaram inibição do veneno total e da proteína Fosfolipase A2. / Insects are the most numerous animals worldwide, with more than 675 thousand known species. Belonging to Hymenoptera order, Apoidea, superfamily, bees are found distributed in approximately 20 thousand species. In Brazil there are about 1,700 species. One of the major species is Apis mellifera, with cosmopolitan occurrence. Apis mellifera, popularly known as Africanized bee, is aggressive, swarm several times per year and uses a great variety of locals to nidificate. This behavior raises the contact between the insect and the population, increasing the accidents numbers. Bee accidents represent a public health problem in many countries because of their frequency and mortality. The present study proposes to search for natural inhibitors of bee venom. A system and a data base have been developed to integrate anti-venom medicinal plants data and bee venoms. Plants activities against venom have been evaluated by farmacological assays, such as anti-hemorraghic, anti-proteolitic, anti-myotoxicity, anti-Phospholipase and anti-edema. The possible interactions between Melittin and Phospholipase A2 toxins with inhibitors have been evaluated by virtual docking. The data base, denominated Bee Venom, was implemented and the data from public data bases have been inserted in the system. The system was released to public access in the following address http://gbi.fmrp.usp.br/beevenom/. In Melittin analysis the protein regions which the inhibitors may act have been found and also the chemical properties that the inhibitors must have to interact with Melitina have been identified. During in silico analysis it was possible to identify 10 possible inhibitors that interacted well with Phospholipase A2 active site. Some plants species from FMRP/USP Germoplam Bank have been obtained and used in the indirect Phospholipase activity and edema, being possible to observe inhibitions of total venom and Phospholipase A2 protein. The synthetic compounds and inhibitors evaluated did not cause inhibition in any experiments. However, the plants obtained on Animals Toxins and Natural and synthetic Inhibitors laboratory have caused inhibition of total venom and Phospholipase A2 protein.

Vários problemas do mundo real podem ser analisados como problemas de otimização. Na bioinformática, em especial, como exemplos podem ser citados o alinhamento múltiplo de sequências, a filogenia, a predição de estruturas de proteínas e RNA, entre outros. As Meta-heurísticas Populacionais (MhP) são técnicas baseadas em interações de conjuntos de soluções candidatas, como elementos de uma população, utilizadas na otimização de funções. Seu uso é especialmente interessante na otimização de problemas onde há conhecimento parcial ou nenhum do espaço de busca. O objetivo deste trabalho é investigar o uso de auto-organização da população de um sistema imunológico artificial (AIS) a fim de aplicá-lo no problema de docking, que pode ser visto como um problema de otimização multimodal complexo. O AIS é um tipo de MhP inspirado na microevolução do sistema imunológico adaptativo de organismos complexos. Neste, as soluções candidatas representam células do sistema imunológico que busca se adaptar para a eliminação de um patógeno. O desenvolvimento do algoritmo foi baseado no opt-aiNet, que utiliza dos princípios das teorias de seleção clonal e maturação de afinidade para realizar a otimização de funções. Adicionalmente, o opt-aiNet, inspirado na teoria de redes imunológicas, realiza uma etapa de supressão, que busca eliminar soluções semelhantes, aumentando assim a diversidade populacional. Esta etapa é computacionalmente custosa, dado que é feito o cálculo da distância entre todos os possíveis pares de células (soluções) afim de eliminar aquelas próximas de acordo com um dado critério. A proposta deste trabalho é o desenvolvimento de um algoritmo de supressão auto-organizável, inspirado no fenômeno da criticalidade auto-organizada, buscando diminuir a influência da seleção de parâmetros e a complexidade da etapa de supressão. O algoritmo proposto foi testado em um conjunto de funções contínuas conhecidas e comumente utilizadas pela comunidade de computação evolutiva. Os resultados obtidos foram comparados com aqueles de uma implementação do opt-aiNet. Em adição, foi proposta a utilização de operadores de mutação com distribuição q-gaussiana nos AISs desenvolvidos. O algoritmo foi também aplicado no problema de docking rígido baseado em complementaridade de superfícies e minimização de colisões, especificamente no docking de proteínas. Os resultados foram comparados com aqueles de um algoritmo genético, resultando em um melhor desempenho obtido pelo algoritmo proposto. / Many real world problems can be described as optimization problems. In bioinformatics in special, there is multiple sequence alignment, filogeny and RNA and Protein structure prediction, among others. Population based metaheuristics are techniques based in the interaction of a set of candidate solutions as elements of a population. Its use is specially interesting in optimization problems where there is little or no knowledge of the search space. The objective of this work is to study the use of self-organization of population in an artificial imune system for use in the docking problem, considered a complex multimodal optimization problem. The artificial imunme system is a type of population based methaheuristics inspired in the microevolution of the adaptive immune system of complex organisms. Candidate solutions represent cells of the immune system adapting its antibodies to eliminate a pathogen. The development of the algorithm was based in the opt-aiNet, based in the principles of clonal selection and affinity maturation for function optimization. Additionally, the opt-aiNet, inspired in theories of immune network, makes a suppression stage to eliminate similiar solutions and control diversity. This stage is computationally expensive as it calculates the distance between every possible pair of cells (solutions) eliminating those closer than a threshold. This work proposes a self-organized suppression algorithm inspired in the self-organized criticality, looking to minimize the influence of parameter selection and complexity of the suppression stage in opt-aiNet. The proposed algorithm was tested in a set of well-known functions in the evolutionary computation community. The results were compared to those of an implementation of the opt-aiNet. In addition, we proposed a mutation operator with q-Gaussian distribution for the artificial immune systems. The algorithm was then applied in the rigid protein docking problem based in surface complementarity and colision avoidance. The results were compared with a genetic algorithm and achieved a better performance.

Cadeias de suprimentos podem ter operações seguindo diferentes estratégias de distribuição e a utilização de cada uma dessas estratégias pode resultar em diferentes operações e custos. A estratégia de cross-docking auxilia na redução dos custos de distribuição de produtos, consolidando cargas, e a redução de tempo e custos de armazenamento, uma vez que o tempo máximo de estoque permitido pela estratégia é de cerca de 24 horas. O objetivo deste trabalho é apresentar um modelo para o problema de cross-docking, em que cargas são entregues e reorganizadas de forma a atender a outras cargas que são coletadas e garantir que as janelas de tempo para início das operações sejam atendidas. Devido à falta de instâncias para o problema disponíveis na literatura, buscou-se gerar um benchmark e disponibilizá-las à comunidade científica. Uma vez que o problema é de difícil solução exata, um método heurístico para a resolução do problema foi desenvolvido. Os resultados mostraram que o modelo proposto resulta em boas soluções quando comparado ao modelo da literatura. O estudo de calibração do software IBM CPLEX mostrou que a calibração dos parâmetros pode resultar em melhores soluções e, por fim, a matheurística se mostrou competitiva com o CPLEX, principalmente para cenários em que a proporção de entregas e coletas diverge. / Supply chains may have operations which follow different distribution strategies and each one of these strategies may result in different operations and costs. The Cross-docking strategy helps to reduce the products distribution costs by consolidating loads and reducing storage costs as the maximum inventory time is approximately 24 hours. The aim of this research is to present a model for the cross-docking problem where loads are delivered and reorganized so as to cater for other loads that are collected and ensure that time windows are respected. Due to the lack of instances available in the literature, a benchmark was generated and was made available to the scientific community. As the problem is difficult to obtain the exact solution, a heuristic method was developed. The results showed that the proposed model has good solutions when compared to the literature model. A study of the IBM CPLEX software showed that tuning can result in better solutions and the matheuristcs was competitive with the software, mainly in scenarios where deliveries and pickups are very different.

Com o objetivo de descobrir moléculas com atividade inibitória contra enzimas alvo de tripanosomatídeos, as estruturas cristalográficas da enzima gliceraldeído-3-fosfato desidrogenase em complexo com dois análogos de 1,3-bisfosfoglicerato (compostos 30 e 33) foram determinadas por difração de raios X, estudos de modelagem molecular foram realizados e o gene xprt (xantina fosforibosiltransferase) de Leishmania major foi clonado e super-expresso em Escherichia coli, e a enzima correspondente foi purificada e caracterizada cinéticamente. O complexo gGAPDH-33 foi determinado até 2,5A e revelou como esse análogo do intermediário tiocetal se liga na enzima. O modelo final da proteína com o inibidor foi refinado utilizando um conjunto de dados com 97,5% de completeza, com um R final de 0,20. Essa estrutura cristalográfica fornece a primeira evidência experimental do mecanismo flip-flop, que descreve como o substrato se desloca do sítio de ligação do fosfato inorgânico para o sítio do fosfato orgânico. O complexo gGAPDH-30 foi determinado até 2,75A de resolução, a partir de um conjunto de dados com 92,4% de completeza e revela o modo de interação dessa classe de inibidores com a gGAPDH. O modelo final apresenta R igual a 0,19. Essa estrutura foi utilizada para estudos de modelagem molecular que explicam a diferença de atividade dessa classe de inibidores entre a gGAPDH de Trypanosoma cruzi e de Trypanosoma brucei. Com relação a XPRT de L. major, essa enzima apresenta uma grande afinidade por hipoxantina, quando comparada a enzima homóloga de L. donovani. Com a finalidade de tentar entender esse comportamento, estudos de modelagem por homologia estão sendo realizados. / Aiming at discover molecules with good inhibitory activity against tripanosomatides enzymatic targets, the crystallographic structures of glyceraldehydes-3-phosphate dehydrogenase in complex with 1,3 bisfosfoglyceric acid analogues (30 and 33)were solved, molecular modeling studies were undertaken and xprt (xanthine phosphorybosil transferase) gene from Leishmania major was cloned and over-expressed in Escherichia coli. The enzyme thus obtained was purified and kinetically characterized. .gGAPDH-33 complex, up to 2,5A resolution revealed the tioketal intermediate binding mode. The final model was refined to R 0.20 from a 97,5% completeness dataset. The crystallographic structure gives, for the first time, experimental evidence for the flip-flop mechanism, which describes how the substrate goes from inorganic-phosphate binding site to substratephosphate binding site. gGAPDH-30 complex, solved to 2,75A resolution, revealed the inhibitor binding mode. The final model has R= 0,19 and was refined from a 92,4% completeness dataset. This structure was used as the framework upon which modeling studies were performed. Modeling results suggest why these inhibitors show a different inhibitory profile against Trypanosoma brucei and Trypanosoma cruzi. L. major XPRT shows a high affinity for hypoxanthine, an alternative substrate, when compared to L. donovani XPRT, aiming at understand this behavior homology modeling studies are currently under progress.

Os receptores nucleares constituem uma superfamília de fatores de transcrição regulados pela interação com hormônios. Esta superfamília inclui, por exemplo, os receptores de hormônio tireoidiano, estrogênio, androgênio, glicocorticóide e mineralocorticóide. Neste trabalho, empregamos técnicas de biologia estrutura e bioinformática para estudar as interações entre alguns dos membros da família de receptores nucleares e seus respectivos ligantes. Para o receptor de hormônio tireoidiano, foi demonstrado, através da análise das estruturas cristalográficas das duas isoformas do receptor ligados aos tiromimético Triac, que os componentes entálpicos visíveis nas estruturas não explicam a seletividade do ligante. Dados de dinâmica molecular confirmaram que a seletividade do hormônio tem um importante componente entrópico. Empregando a técnica de dinâmica molecular, estudamos a ligação do receptor de mineralocorticóide humano à aldosterona, ao cortisol, à espironolactona e à cortisona e simulamos ainda o efeito da mutação S810L, conhecida por converter a atividade antagonista da cortisona e da espironolactona em agonista. A análise das simulações revelou um perfil de ligações de hidrogênio similar na ligação do receptor selvagem ao cortisol e à aldosterona. A cortisona perde, por conta da inserção de uma hidroxila na posição 11, uma ligação de hidrogênio importante com a Asn770 e, por isso, tem menor energia potencial de ligação. A espironolactona perde a mesma ligação de hidrogênio ao mesmo tempo em que aumenta o número de contatos de van der Waals pela inserção do grupo tioacetil na posição 7. A mutação S810L simulada no complexo com cortisona, cortisol e espironolactona não interfere no padrão de ligações de hidrogênio estabelecidas entre o receptor e os ligantes, mas altera a mobilidade de uma das regiões propostas como rota de dissociação. Propomos, portanto, que a mutação interfere na cinética de dissociação dos ligantes e não no padrão de interações estabelecidas no equilíbrio. Simulações de dissociação induzida do ligante confirmam esta proposição. Na última etapa, utilizamos os modelos experimentalmente determinados para o receptor ativado por proliferadores peroxissomais gama para a busca de novos ligantes através da técnica de docking molecular. Neste trabalho, utilizamos uma base de dados com aproximadamente um milhão de compostos. Destes, quatro foram selecionados após o docking molecular e testados experimentalmente. Um dos compostos testados se mostrou ativo neste receptor, apresentando uma atividade de 60-70% da atividade da rosiglitazona, conhecido agonista total do PPARg. / Nuclear receptors are a superfamily of hormone-regulated transcriptional factors. This superfamily includes, for example, the receptor for thyroid hormone, estrogen, androgen, gluco and mineralocorticoid. In this work, we used structural biology and bioinformatic tools to study the interactions between some members of the nuclear receptor superfamily and its respective ligands. We showed by the analysis of the crystal structures of both thyroid hormone receptor isoforms bound to the thyromimetic Triac that the enthalpic components visible in the structures do not explain the ligand selectivity. Molecular dynamics simulation data confirmed later that the hormone selectivity has an important entropic component. Using the molecular dynamics simulation, we studied, in a second stage, the interaction between the human mineralocorticoid receptor bound to aldosterone, cortisol, spironolactone and cortisone and also simulated the effects of the mutations S810L, known to convert the antagonist properties of spironolactone and cortisone in an agonist activity. The analysis of the simulations showed a similar profile in hydrogen bonds established between the wild type receptor bound to cortisol and aldosterone. Cortisone looses an important hydrogen bond with Asn770 because of the insertion of a carbonyl group in the 11 position and shows a decreased binding potential energy. Spironolactone loses the same interaction but has an increased number of van der Waals contacts because of the insertion of a tioacetyl group in the 7 position. The mutant S810L simulated in complex with cortisol, cortisone and spironolactone showed that the mutation do not interfere with the hydrogen bond profile established between the receptor and the ligands but changes the mobility of a region in the receptor previously proposed as a ligand dissociation route. Ligand unbinding simulations through steered molecular dynamics (SMD) confirm that aldosterone and cortisol unbind differentially and the mutation S810L alters the unbinding profile. We then propose that the mutation changes the kinetics of ligand association/dissociation without changing the profile of the interactions established in the equilibrium. In the last stage, we used the experimentally determined structural model of the peroxissome proliferator-activated receptor gamma to search for novel ligands using the molecular docking technique. For this work, we used a database containing about 1 million compounds. Among those, four compounds were selected after the docking computation and experimentally tested. One of these compounds was found to be active in the receptor, showing about 60-70% of the agonistic activity of rosiglitzone, a known PPARg total agonist.

Les polykétides sont des composés naturels. Ces composés possèdent des rôles thérapeutiques variés tels que antifongiques, antibiotiques, anticancéreux, immunosuppresseurs ou encore anticholestérolémiques. Par conséquent, la recherche de nouvelles structures possédant des bioactivités diverses se révèlent être intéressante. Une stratégie prometteuse pour créer des nouveaux polykétides est l’ingénierie génétique des enzymes synthétisant ces molécules, les polykétide synthases modulaires (PKS), une approche désignée sous le terme de « biologie synthétique ». Pour ce faire, il faut comprendre de façon détaillée le fonctionnement de ces systèmes multienzymatiques. Plusieurs points restent à éclaircir, dont : i) le contrôle de la stéréochimie du polykétide ; et ii) l’interaction des sous-unités composant la PKS. Lors de ma thèse, j’ai identifié deux kétoréductases (KR) qui, introduites dans un contexte modulaire intrinsèquement non-épimérisant, sont capables d’épimériser le méthyle en Cα de façon efficace. Cependant, la modification de la stéréochimie du polykétide ne dépend pas exclusivement des propriétés intrinsèques de la KR mais aussi du contexte modulaire. J’ai également contribué à la réalisation d’un second projet, pour lequel notre équipe a mis en évidence une nouvelle classe de domaine de docking de PKS de type trans-AT présentant une nouvelle topologie. L’un des DD étudié est une protéine intrinsèquement désordonnée dont le repliement est induit par son partenaire. Nous avons caractérisé l’interface complète entre deux sous-unités de PKS de type trans-AT, révélant une chambre de réaction protégée dans laquelle les chaînes de polykétide peuvent croître / Polyketides are natural products which exhibit a variety of therapeutic activities, including anti-fungal, antibiotic, anticancer, immunosuppressant and anti-cholesterolemic properties. Given their medical and economic importance, there is significant interest in identifying new structures with new biological activities. A promising strategy to create such analogues is to genetically engineer the enzymes responsible for synthesizing these molecules, the modular polyketide synthases (PKSs), an approach referred to as ‘synthetic biology’. However, in order to increase the efficacy of this approach, we must understand in detail how the PKS multienzymes work. A number of issues remain to be clarified, including: i) polyketide stereocontrol, ii) the interaction of the component subunits PKS. During my thesis, I identified two ketoreductase (KR) domains which when introduced into an intrinsically non-epimerizing modular context, were able to efficiently epimerise at the Cα of a model polyketide. I also showed that the modular context in which the KR functions has an influence on the ultimate stereochemical outcome. I also made essential contributions to a second project, in which the group identified a novel family of docking domains (DD) in the trans-AT type of PKS which present a novel topology. One of the two model DDs studied is an intrinsically disordered protein whose folding is induced by its partner. Finally, we were able to visualize a complete intersubunit interface within a trans-AT PKS, revealing a protected reaction center in which polyketide chains can grow.