Thesis (Ph. D.)--Florida State University, 2003. / Advisor: Dr. Kenneth A. Taylor, Florida State University, College of Arts and Sciences, Institute of Molecular Biophysics. Title and description from dissertation home page (viewed 5/4/04). Includes bibliographical references.

Diss. (sammanfattning) Stockholm : Karol. inst., 2005. / Härtill 4 uppsatser.

Thesis (Ph. D.)--University of Virginia, 1998. / Spine title: Cryo-EM of Ca²⁺-ATPase from SR. Includes bibliographical references (p. 149-159). Also available online through Digital Dissertations.

Les ADN topoisomérases de type 2 (Top2) sont des protéines universelles et essentielles à la vie cellulaire. Elles ont pour fonction de réguler de manière fine l’équilibre topologique de l’ADN. Chez l’homme, les Top2 (HsTop2) sont des cibles thérapeutiques de première importance en oncologie. Des molécules telles que l’étoposide et la doxorubicine font partie des traitements anticancéreux les plus utilisés en clinique à l’heure actuelle, elles agissent en stabilisant les complexes Top2/ADN/inhibiteurs qui sont transformés en lésions permanentes dont l’accumulation entraîne la mort cellulaire. Mon travail de thèse porte sur la caractérisation fonctionnelle et structurale des isoformes des HsTop2 entières en complexe avec des composés thérapeutiques. Mon projet de thèse comporte trois axes principaux et complémentaires : (i) La production et la caractérisation fonctionnelle in vitro des isoformes recombinantes. Nous avons optimisé et simplifié les protocoles de surexpression des deux isoformes dans la levure et avons également mis au point leur expression dans les cellules de mammifère. (ii) une étude structurale par cryo-microscopie électronique sur les HsTop2 entières en complexe avec des composés thérapeutiques et des oligonucléotides. Ces premiers travaux ont permis l’obtention d’une carte tridimensionnelle de l’enzyme entière qui servira de base pour l’étude de l’architecture des HsTop2 au sein de complexes protéiques de plus grande taille. (iii) des expériences de protéomique chimique permettant de mettre à jour les cibles secondaires potentielles de l’étoposide et de la doxorubicine dans des extraits cellulaires de lignées cancéreuses. / The type 2 DNA topoisomerases are universal proteins essential for cell survival. These enzyme fine tune the topological equilibrium of the DNA in cells. The human proteins are major targets of therapeutics drugs used in oncology. Molecules like etoposide and doxorubicin are among the most effective anticancer drugs used in chemotherapy treatments; they form stable Top2/DNA/drugs complexes which are then transformed in permanent DNA damages. This thesis project focuses on the functional and structural characterization of these complexes using 3 main strategies:(i) the production and the in vitro functional characterization of the recombinant isoforms. We optimized and simplified the production and purification procedures and overexpressed both isoforms in mammalian cells. (ii) a structural study by cryoelectron microscopy on the full enzyme complex with DNA and therapeutic drugs. We obtained a 3D density map that will be used for further studies on the architecture of large HsTop2 associated complexes. (iii) ChemoProteomic experiments on cancer cell lines to highlight the potential etoposide and doxorubicin secondary targets.

Orientadores: Fernando José Von Zuben, Rodrigo Villares Portugal / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-25T22:36:38Z (GMT). No. of bitstreams: 1 Righetto_RicardoDiogo_M.pdf: 5898819 bytes, checksum: c98b9e2b61390aa847a4a6040d3f550b (MD5) Previous issue date: 2014 / Resumo: Análise de Partículas Isoladas é uma técnica que permite o estudo da estrutura tridimensional de proteínas e outros complexos macromoleculares de interesse biológico. Seus dados primários consistem em imagens de microscopia eletrônica de transmissão de múltiplas cópias da molécula em orientações aleatórias. Tais imagens são bastante ruidosas devido à baixa dose de elétrons utilizada. Reconstruções 3D podem ser obtidas combinando-se muitas imagens de partículas em orientações similares e estimando seus ângulos relativos. Entretanto, estados conformacionais heterogêneos frequentemente coexistem na amostra, porque os complexos moleculares podem ser flexíveis e também interagir com outras partículas. Heterogeneidade representa um desafio na reconstrução de modelos 3D confiáveis e degrada a resolução dos mesmos. Entre os algoritmos mais populares usados para classificação estrutural estão o agrupamento por k-médias, agrupamento hierárquico, mapas autoorganizáveis e estimadores de máxima verossimilhança. Tais abordagens estão geralmente entrelaçadas à reconstrução dos modelos 3D. No entanto, trabalhos recentes indicam ser possível inferir informações a respeito da estrutura das moléculas diretamente do conjunto de projeções 2D. Dentre estas descobertas, está a relação entre a variabilidade estrutural e manifolds em um espaço de atributos multidimensional. Esta dissertação investiga se um comitê de algoritmos de não-supervisionados é capaz de separar tais "manifolds conformacionais". Métodos de "consenso" tendem a fornecer classificação mais precisa e podem alcançar performance satisfatória em uma ampla gama de conjuntos de dados, se comparados a algoritmos individuais. Nós investigamos o comportamento de seis algoritmos de agrupamento, tanto individualmente quanto combinados em comitês, para a tarefa de classificação de heterogeneidade conformacional. A abordagem proposta foi testada em conjuntos sintéticos e reais contendo misturas de imagens de projeção da proteína Mm-cpn nos estados "aberto" e "fechado". Demonstra-se que comitês de agrupadores podem fornecer informações úteis na validação de particionamentos estruturais independetemente de algoritmos de reconstrução 3D / Abstract: Single Particle Analysis is a technique that allows the study of the three-dimensional structure of proteins and other macromolecular assemblies of biological interest. Its primary data consists of transmission electron microscopy images from multiple copies of the molecule in random orientations. Such images are very noisy due to the low electron dose employed. Reconstruction of the macromolecule can be obtained by averaging many images of particles in similar orientations and estimating their relative angles. However, heterogeneous conformational states often co-exist in the sample, because the molecular complexes can be flexible and may also interact with other particles. Heterogeneity poses a challenge to the reconstruction of reliable 3D models and degrades their resolution. Among the most popular algorithms used for structural classification are k-means clustering, hierarchical clustering, self-organizing maps and maximum-likelihood estimators. Such approaches are usually interlaced with the reconstructions of the 3D models. Nevertheless, recent works indicate that it is possible to infer information about the structure of the molecules directly from the dataset of 2D projections. Among these findings is the relationship between structural variability and manifolds in a multidimensional feature space. This dissertation investigates whether an ensemble of unsupervised classification algorithms is able to separate these "conformational manifolds". Ensemble or "consensus" methods tend to provide more accurate classification and may achieve satisfactory performance across a wide range of datasets, when compared with individual algorithms. We investigate the behavior of six clustering algorithms both individually and combined in ensembles for the task of structural heterogeneity classification. The approach was tested on synthetic and real datasets containing a mixture of images from the Mm-cpn chaperonin in the "open" and "closed" states. It is shown that cluster ensembles can provide useful information in validating the structural partitionings independently of 3D reconstruction methods / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica