The aim of this thesis is to investigate applicability of regulariza- tion by Krylov subspace methods to discrete inverse problems arising in single particle analysis (SPA). We start with a smooth model formulation and describe its discretization, yielding an ill-posed inverse problem Ax ≈ b, where A is a lin- ear operator and b represents the measured noisy data. We provide theoretical background and overview of selected methods for the solution of general linear inverse problems. Then we focus on specific properties of inverse problems from SPA, and provide experimental analysis based on synthetically generated SPA datasets (experiments are performed in the Matlab enviroment). Turning to the solution of our inverse problem, we investigate in particular an approach based on iterative Hybrid LSQR with inner Tikhonov regularization. A reliable stopping criterion for the iterative part as well as parameter-choice method for the inner regularization are discussed. Providing a complete implementation of the proposed solver (in Matlab and in C++), its performance is evaluated on various SPA model datasets, considering high levels of noise and realistic distri- bution of orientations of scanning angles. Comparison to other regularization methods, including the ART method traditionally used in SPA,...

Pseudomonas aeruginosa est une bactérie Gram-négative qui présente une grande résistance aux antibiotiques, lui permettant de sévir dans le milieu hospitalier en infectant plus particulièrement les patients immunodéprimés. Cette résistance est principalement due au système d’efflux membranaire MexAB-OprM, capable d’exporter les antibiotiques en dehors de la cellule. Cette pompe à efflux est composée de trois protéines, MexA, MexB et OprM, incorporées dans les membranes internes et externes de la paroi bactérienne. Les structures de MexA, OprM et AcrB -une protéine présente chez E. coli, homologue de MexB- ont été déterminées individuellement par cristallographie des rayons X. Cependant, la structure du complexe entier, regroupant les trois protéines en interaction, ainsi que le mécanisme de cette pompe font toujours défaut. Le renforcement de nos connaissances structurales et fonctionnelles est donc capital pour lutter plus efficacement contre ces bactéries, par de nouvelles stratégies médicamenteuses. Ce travail porte sur l’étude de la structure et de la stœchiométrie de l’assemblage des protéines OprM et MexA au sein d’une membrane lipidique. La caractérisation du complexe OprM/MexA a été réalisée à l’aide de nouvelles techniques de caractérisation physico-chimique des surfaces, telle que la Microbalance à Cristal de Quartz avec Mesure de Dissipation (QCM-D), et par des méthodes d’imagerie, telles que la Cryo-Microscopie Electronique en Transmission (CryoMET) et la Cryo-Tomographie Electronique (CryoTE). En QCM-D, les mesures d’interaction entre OprM et MexA ont été réalisées sur support solide en contrôlant l’orientation d’OprM placée dans un environnement lipidique. Après ajout de la protéine MexA, la formation de complexes OprM/MexA a été mise évidence. Pour comprendre l’organisation de ce complexe, nous avons procédé à une étude comparative de l’organisation des protéines OprM, MexA et du complexe OprM/MexA incorporés dans une membrane lipidique, par CryoMET. Trois types d’organisation, respectivement spécifiques d’OprM, de MexA et du complexe OprM/MexA, ont été mis en évidence. Une analyse structurale de ces trois différents assemblages, pris en sandwich entre deux membranes lipidiques, a été menée par CryoTE. La reconstitution de la protéine OprM conduit à la formation de protéoliposomes, dû à des interactions intervenant entre les protéines OprM au niveau de leurs hélices périplasmiques. La protéine MexA s’organise sous forme d’une structure annulaire de 13 nm de hauteur au sein des membranes lipidiques, et d’une structure plus complexe de 26 nm de hauteur, résultant de l’empilement tête-bêche de deux structures annulaires de 13 nm. Ce travail révèle les dimensions exactes de l’assemblage formé par MexA, et permet de localiser à proximité des membranes les domaines non résolus dans la structure cristallographique. La reconstitution du complexe OprM/MexA révèle une disposition régulière des deux protéines dans les membranes lipidiques. Au sein des complexes, les protéines OprM sont présentes sous forme de trimères. Dans la membrane opposée, à l’aplomb d’une molécule d’OprM, MexA ne forme pas une structure annulaire similaire à celle décrite précédemment, indiquant un état d’oligomérisation différent de celui observé dans les assemblages MexA. Les densités de MexA sont compatibles avec la présence de quelques molécules de MexA. Cependant des structures annulaires de MexA, positionnées à l’aplomb de trois trimères d’OprM sont visibles. Notre étude montre que MexA adopte des structures oligomériques spécifiques en fonction de ses interactions avec les membranes lipidiques ou avec son partenaire OprM. / The structure determination of membrane protein in lipid environment can be carried out using cryo electron microscopy combined with the recent development of data collection and image processing. We describe a protocol to study assemblies or stacks of membrane protein reconstitued into a lipid membrane using both cryo electron tomography and single particle analysis which is an alternative approach to electron crystallography for solving 3D structure. We show the organization of the successive layers of OprM molecules revealing the protein-protein interactions between OprM molecules of two successive lipid bilayers.

The transcription initiation is the first step in gene expression and an important regulation step in all living organisms. In bacteria, it has been proposed that DNA bending and its wrapping on the surface of E. coli RNAP might facilitate the opening of the transcription bubble, which is necessary for the initiation of gene transcription. In this work, it is shown the first structural study to evaluate a DNA wrapping model, including its length and the relative position in the bacterial transcription initiation complex (RP complex), assembled between RNA polymerase-σ70 holoenzyme (RNAP) and a λPR promoter (-100 to +30 wild type). RP complex was prepared and negatively stained with 2% uranyl acetate on a thin-carbon coated grid and the data acquisition of 500 images was performed in a JEM-2100 (JEOL, Japan) microscope equipped with an F-416 CMOS camera (TVIPS, Germany). Single particle analysis of 16,015 particles, grouped in 666 class-averages, was conducted using IMAGIC 4D software (Image Science, Germany) to obtain a three-dimensional model of the RP complex at 20Å resolution. After the rigid-body fitting of the RNAP crystallographic structure (PDB 4YG2) and the modeled DNA promoter, it was observed that the regions 1.2 and 4.2 of the σ70 subunit interacts with the consensus zones, -10 and -35 hexamers of the promoter. Furthermore, it was possible to observe that αCTDs (C-terminal domain) in both alpha subunits would be oriented to facilitate the interaction with the first and second UP-elements regions, respectively (centered around –50 and -75 positions in the promoter). These was enabled by the presence of the characteristics motifs helix-hairpin-helix in these domains. In addition, the downstream DNA, from the transcription bubble, appears to be inside the protein main channel, oriented in a way to enable interactions with the RNAP clamp and jaws. Finally, it was observed that the DNA wrapping has ~32 nm of total length and involves a promoter bent of ~255° around the RNAP surface. The 3D-model obtained in this study is the very first direct structural confirmation of the DNA promoter wrapping in a bacterial transcription initiation complex. / A iniciação da transcrição é o primeiro passo na expressão gênica e importante ponto de regulação em todos os organismos vivos. Em bactérias, foi proposto que o enovelamento do DNA na superfície da RNAP de E. coli pode facilitar a abertura da bolha de transcrição, necessária para o início da transcrição gênica. Neste trabalho, é apresentado o primeiro estudo estrutural direto para avaliar o comprimento do enovelamento do DNA e sua posição no complexo de iniciação da transcrição bacteriana (complexo RP), montado entre a holoenzima RNA polimerase-σ70 (RNAP) e um promotor λPR (-100 para +30, tipo selvagem). Amostras do complexo RP foram preparadas e contrastadas negativamente com 2% de acetato de uranila em uma grade com filme fino de carbono e a aquisição de 500 imagens foi realizada em um microscópio JEM-2100 (Jeol, Japão) equipado com uma câmera CMOS F-416 (TVIPS, Alemanha). A análise de partículas isoladas de 16.015 partículas, agrupadas em 666 médias de classe, foi conduzida usando o software IMAGIC 4D (Image Science, Alemanha) para obter um modelo tridimensional do complexo RP, a 20Å de resolução, estimado pelo critério de ½ bit. Após o ajuste de corpo rígido da estrutura cristalográfica da RNAP (PDB 4YG2) e do promotor de DNA modelado, observou-se que as regiões 1.2 e 4.2 da subunidade σ70 interagem com as zonas de consenso, hexâmeros -10 e -35, do promotor. Além disso, foi possível observar que os αCTDs (domínio C-terminal) em ambas as subunidades alfa estariam orientados para facilitar uma possível interação com a primeira e segundas regiões dos elementos UP, respectivamente (centradas em torno das posições –50 e -75 do promotor). Estas seriam possíveis devido à presença de alguns motivos de características hélice-grampo-hélice nesses domínios. Além disso, a região do promotor, downstream da bolha de transcrição, parece estar dentro do canal principal da proteína, orientado de forma a possibilitar interações com o clamp e jaw da RNAP. Finalmente, foi observado que o comprimento total do enovelamento de DNA envolve cerca de 32 nm e 255° de rotação do DNA ao redor da superfície da RNAP. Portanto, este modelo 3D é a primeira confirmação estrutural direta do enovelamento de DNA em um complexo bacteriano de iniciação da transcrição.

The overall structure of photosynthetic pigment-protein complexes and thylakoid membranes of various photosynthetic organisms was studied using electron microscopy.

The KdpFABC complex from E. coli functions as a high affinity K uptake system and belongs to the superfamily of P-type ATPases. So far, no information is available about the orientation of the subunits within the complex as well as its oligomeric state. By chemical crosslinking, gel filtration, electron transmission microscopy and single particle FRET analysis this study shows that the KdpFABC complex occurs as a homodimer with a dissociation constant between 30 to 50 nM. Furthermore, by means of single particle analysis of transmission electron micrographs, the solution structure of the complex at 1.9 nm resolution could be solved, thus providing the first structural analysis resolving all subunits of the holoenzyme. Based on crystal structures, it is generally assumed that P-type ATPases undergo large domain movements during catalysis. However, these conformational changes have never been shown directly. By use of single molecule FRET with alternating laser excitation, distance changes could be measured directly within KdpB during ATP hydrolysis. With this technique, distances and dwell times were determined for three conformational states in the working enzyme as well as in the orthovanadate- and the OCS-inhibited state.

L'apparition croissante de bactéries pathogènes multi-résistantes à la plupart des antibiotiques disponibles apparaît comme un problème mondial de santé publique. Malheureusement, un usage excessif à la fois en médecine humaine et animale a conduit à l’apparition de souches multi-résistantes à la plupart des antibiotiques disponibles sur le marché. Il est donc urgent de mieux comprendre les mécanismes mis en place par les bactéries pour résister aux antibiotiques afin de trouver des solutions pour combattre les souches multi-résistances.Dans ce contexte, le projet de la thèse vise à mieux comprendre les bases moléculaires de l’efflux actif de drogues chez Pseudomonas aeruginosa, qui est un des plus importants mécanismes utilisés par la bactérie pour lutter contre l’action de plusieurs antibiotiques. Les systèmes d’efflux forment des complexes protéiques situés dans la paroi de la bactérie et expulsent de manière active les antibiotiques avant même qu’ils aient pu atteindre leur cible intracellulaire, les rendant ainsi inactif.L’étude structurale se focalise sur le système RND (Resistance-Nodulation and cell Division) MexA-MexB-OprM qui est constitutivement exprimé chez la bactérie sauvage et est surexprimé chez les souches résistantes. Ce complexe tripartite est composé d'un transporteur inséré dans la membrane interne, d'une protéine canal insérée dans la membrane externe et d’une protéine adaptatrice périplasmique qui relie les deux autres protéines pour former un conduit étanche traversant le périplasme. En l’absence de la connaissance de la structure du complexe tripartite, l’objectif de la thèse a été de développer une stratégie originale pour reconstituer in vitro le complexe entier dans un environnement lipidique à partir des trois composants natifs produits séparément.L’assemblage du complexe tripartite est réalisé en mélangeant MexB et OprM en Nanodisque avec MexA mimant les deux bicouches lipidiques. La structure de ce complexe tripartite a été obtenu en combinant la cryo microscopie électronique et à l’approche dite ‘des particules isolées’. La structure tridimensionnelle du complexe calculée à une résolution inférieure à 4 Å a permis de construire un modèle atomique du complexe tripartite assemblé entre deux Nanodisques.Le complexe tripartite est composé d’un trimère d’OprM, d’un trimère de MexB et d’un hexamère de MexA entourant MexB et en interaction avec OprM. Ces données ont permis de résoudre la structure complète de MexA dans le complexe dont la partie N-terminale jusqu’alors inconnue car trop flexible et décrivent pour la première fois l’ancrage de MexA dans une membrane lipidique. Les changements conformationnels sont observés sur OprM et MexB lorsqu’ils sont engagés dans le complexe avec l’ouverture de l’extrémité périplasmique d’OprM et le basculement d’une boucle de MexB permettant d’établir un contact supplémentaire avec MexA.Pour replacer cette structure tripartite dans le cycle d’efflux de l’antibiotique, celle-ci décrit un état qui s’apparente probablement à un état au repos, sachant qu’aucun ligand spécifique n’a été ajouté au cours de l’assemblage. De plus, le complexe forme un canal ouvert à son extrémité extracellulaire, fournissant le conduit pour évacuer les drogues transportées par MexB qui utilise la force protomotrice comme source d’énergie.Ce travail ouvre la perspective à des études structurales d’autres états conformationnels du système d’efflux en condition « énergisé » pour compléter la compréhension du mécanisme du cycle d’efflux. Par ailleurs, la connaissance de cette première structure du complexe natif tripartite constitue le premier pas vers le développement de molécules capables de bloquer l’assemblage du complexe à des fins thérapeutiques. En effet, de telles molécules inhiberaient l’efflux actif et restauraient l’efficacité perdue des antibiotiques actuels. / The increasing appearance of multi-drug-resistant pathogenic bacteria to most available antibiotics is emerging as a global public health problem. Unfortunately, excessive use in both human and animal medicine has led to the emergence of multi-drug-resistant strains for most antibiotics available on the market. It is therefore urgent to better understand the underlying mechanisms by which bacteria resist to antibiotics to combat multi-resistance strains. In this context, this work aims at better understanding the molecular basis of active drug efflux in Pseudomonas aeruginosa, which is one of the most important mechanisms used by the bacterium to resist to several antibiotics. Efflux systems form protein complexes in the bacterial wall and actively expel antibiotics even before they reach their intracellular target, rendering them inactive. The structural study focuses on the MexA-MexB-OprM RND (Resistance-Nodulation and cell Division) system that is constitutively expressed in wild-type bacteria and is over-expressed in resistant strains. This tripartite complex is composed of a transporter inserted into the inner membrane, a channel protein inserted in the outer membrane and a periplasmic adapter protein that connects the other two proteins to form a sealed conduit through the periplasm. In the absence of knowledge of the structure of the tripartite complex, the aim of the thesis was to develop an original strategy to reconstitute the whole complex in vitro in a lipid environment from the three native components produced separately.The assembly of the tripartite complex is made by mixing MexA with MexB and OprM in Nanodisc mimicking the two lipid bilayers. The structure of this tripartite complex was obtained by combining cryo electron microscopy and the so-called 'isolated particles' approach. The three-dimensional structure of the complex, calculated at a resolution of less than 4 Å, was used to build an atomic model of the tripartite complex assembled between two Nanodiscs. The tripartite complex is composed of an OprM trimer, a MexB trimer and a MexA hexamer surrounding MexB and interacting with OprM. We solve the complete structure of MexA whose N-terminal part hitherto unknown because of a high flexibility and describe for the first time the anchoring of MexA in a lipid membrane. The conformational changes are observed on OprM and MexB when they are assembled in the complex with the opening of the periplasmic end of OprM and the spatial re-orientation of a MexB loop to establish additional contact with MexA.To integrate this tripartite structure into the antibiotic efflux cycle, it describes a state that is probably a resting state, knowing that no specific ligand was added during assembly. In addition, the complex forms an open channel at its extracellular end, providing the conduit to evacuate the drugs carried by MexB that uses the proton motive force as a source of energy. This work opens new perspective for structural studies of other conformational states of the efflux system in "energized" conditions to fulfill our understanding of the efflux cycle mechanism. Moreover, the knowledge of this first tripartite native complex structure constitutes the first step towards the development of molecules capable of blocking the assembly of the complex for therapeutic uses. Indeed, such molecules would inhibit active efflux and restore the lost efficiency of current antibiotics.

<p>Human noroviruses (HuNoVs) are the leading cause of an acute form of non-bacterial gastroenteritis, where strains belonging to genogroup (G) II are dominant. Upon expression with the baculovirus culture system, virus-like particles (VLPs) of HuNoVs are expected to assemble into T = 3 icosahedral capsid particles resembling the structure of the infectious virion particles. However, some strains were found to assemble into either T = 1 or T = 4 capsids, or a combination of two different capsid forms. In this study, we showed that VLPs of the Virginia 1997 387 (VA387) GII.4 outbreak strain assembled into T = 1, T = 3, and T = 4 capsids upon expression in insect cell culture, the first case for a naturally occurring HuNoV strain to assemble into all three capsid states. TEM analysis revealed that T = 1 icosahedral particles were the most abundant in purified samples, which contrasts previous findings where either T = 3 or T = 4 were the most abundant. We resolved the cryo-EM structures of the T = 1 shell (S) domain, T = 3, and T = 4 particles to 2.24, 2.44, and 3.43 Å, respectively, making them the most resolved norovirus (NoV) structures to date. Single particle cryo-EM 3D analysis showed that the protruding (P) domain of T = 1 and T = 4 VLPs are highly dynamic. Additionally, we showed that T = 3 VLPs are resistant while T = 1 and T = 4 VLPs are sensitive to digestion in the presence of trypsin. This suggested that T = 1 and T = 4 capsids are less stable among the VLPs, which is consistent with the highly dynamic P domain inferred from our cryo-EM 3D analysis. During infection, HuNoVs travel through the gastrointestinal (GI) tract where they encounter a broad range of variable conditions that include pH, ionic strength, and host defenses (e.g., proteases). Our analyses suggest that virions are T = 3 particles as they can survive the GI tract upon exiting the host. We determined the first cryo-EM structure of T = 3 VLPs in complex with the known HuNoV host cell receptor, histo-blood group antigen (HBGA), to a resolution of 2.51 Å, demonstrating that NoV VLPs can serve as a platform in the structural characterization of small ligand molecules. Lastly, we identified a histidine residue retained in the S domain of all identified caliciviruses critical in the assembly of capsids. Our structures and their characterization will contribute to the development of therapeutic agents to combat noroviruses. </p>