[pt] Sistemas multi-input multi-output (MIMO) massivos livres de células são uma variante de sistemas MIMO multi-celulares que consideram a ausência de células. Desta forma, a interferência entre as células é minimizada e a capacidade de cobertura do sistema é melhorada devido à menor distância entre os pontos de acesso (APs) e os usuários. É uma solução de comunicação MIMO massiva multi-usuário que envolve um número estendido de APs que podem ser equipados com tecnologia MIMO para fornecer serviço a usuários simultaneamente. Os APs são controlados por uma unidade central de processamento (CPU) para garantir a coordenação dentro da rede e para processamento e decodificação de informação. Possíveis arranjos para a arquitetura livre de células incluem esquemas centralizados e descentralizados. Para a configuração centralizada, os APs enviam todas as suas estimativas de canal e informações recebidas para a CPU por meio de enlaces de transporte frontais para processamento e detecção de sinais. Além disso, na arquitetura centralizada, os APs atuam como repetidores na rede. Outro nível de cooperação para sistemas MIMO massivos livre de células é o esquema descentralizado. Nesta proposta de tese, a arquitetura dos sistema MIMO massivos livres de células no canal reverso é estudada para as implementações centralizadas e descentralizadas. Em particular, estuda-se o desempenho de técnicas de mitigação de interferência para essas redes supondo-se conhecimento perfeito de canal e usando técnicas de detecção lineares e não lineares, seleção de APs, e esquemas iterativos de detecção e decodificação com códigos LDPC para melhorar o desempenho do sistema e reduzir a carga de sinalização. Para o caso em que há falta de compartilhamento de informações sobre os canais, o uso de pilotos para obter estimativas de canais é considerado e explorado. / [en] Cell-free massive multiple-input multiple-output (CF-mMIMO) is an advanced variant of network multiple-input multiple-output (MIMO) which considers absence of cell boundaries. Thus, the interference between cells in cellular systems is greatly minimised and the system s coverage capacity is improved due to the shorter distances between the access points (APs) and the users. It is a multi-user massive MIMO communications solution that involves an extended number of APs that can either be equipped with MIMO or single antennas to provide service to users simultaneously. The APs are controlled by a central processing unit (CPU) to ensure coordination within the network and for information processing and decoding. Possible arrangements for the CF-mMIMO architecture include, but are not limited to: centralized and decentralized schemes. In this thesis, the uplink of a CF-mMIMO system architecture is studied for the centralized and decentralized implementations. In particular, we study the performance of interference mitigation techniques for CF-mMIMO networks using iterative detection and decoding (IDD) schemes. The performance of the system is studied assuming perfect and imperfect channel state information (CSI). Access point selection based on the effective channel gain to make the network more practical and scalable are devised. The use of low-density parity check (LDPC) codes that adopt message passing has been investigated. Furthermore, log likelihood ratio (LLR) refinement strategies have been proposed to improve decentralized processing for CF-mMIMO networks. Finally, the performance of the considered schemes is analyzed theoretically and simulations are used to assess the performance in terms of BER, number of fronthaul signaling, and computational cost.

La NADPH oxydase du phagocyte est une enzyme impliquée dans la défense immunitaire contre les pathogènes. Après activation du phagocyte, cette enzyme produit des ions superoxyde par réduction du dioxygène par le NADPH. Elle est constituée de quatre sous- unités cytosolubles (p47phox ; p67phox ; p40phox et Rac), et deux membranaires (gp91 ; p22phox). Son activation fait intervenir un processus complexe qui met en jeu des changements d’interaction entre les protéines la constituant et qui permet l’assemblage des six sous- unités. Afin d’obtenir des informations sur les processus d’assemblage et d’activation, j’ai reconstitué le complexe dans un système cell free à l’aide de protéines recombinantes pour pouvoir contrôler tous les paramètres. Dans ce travail nous avons comparé les modes d’activation de p47phox par phosphorylation, par mutation substitutionelle sérine - aspartate en position S303,S304 et S328 pour mimer la phosphorylation et enfin par addition d’acide arachidonique (AA) activateur connu de l’enzyme in vitro mais aussi in vivo. Bien qu’il ai été montré que ces trois méthodes ouvrent la protéine vers une conformation ayant des propriétés similaires, nous avons trouvé que les effets de ces méthodes d’activation sont significativement différents. Ainsi, les changement de conformation observés par dichroisme circulaire, sont dissemblables. Pour p47phox, l’addition de AA déstructure la protéine. La phosphorylation induit un déplacement bathochrome des bandes de CD qualitativement similaire, alors que les mutations S-D de p47phox provoquent un déplacement opposé. Pour le complexe p47phox-p67phox l’addition d’AA destructure le mélange tandis que la mutation induit relativement peu de changement. Nous avons mesuré les constantes de dissociation Kd du complexe p47phox-p67phox. Alors que pour les protéines « sauvages », le Kd est faible (4±2 nM), les mutations de p47phox ainsi que l’addition d’AA augmentent cette valeur jusqu’à environ 50 nM, montrant une diminution de l’affinité entre p47phox-p67phox. De même, sur le complexe entier, l’effet de la phosphorylation de p47phox est différent de la mutation. Nous avons mesuré les valeurs de EC50 relatives à p67phox pour les différentes formes de p47phox. L’activation de p47phox par phosphorylation diminue l’EC₅₀, alors que les doubles ou triple mutations augmentent sa valeur. Nous avons confirmé que la phosphorylation et la mutation sont insuffisantes pour activer l’enzyme. La présence de AA est indispensable pour le fonctionnement du complexe. L’ordre de fixation des sous unités cytosoliques semble indifférent mais il faut que tous les composants soient présents lors de l’ajout de AA. Enfin, la délétion de p47phox dans la partie C-terminale (aa 343 à 390, domaine d’interaction avec p67phox) il n’y a plus de formation du dimère mais l’enzyme fonctionne normalement. Ces résultats apportent des éléments nouveaux sur le rôle de la dimérisation p47 phox-p67 phox, non indispensable à l’activité du système et sur le rôle mineur de la phosphorylation dans l’activation de la NADPH oxydase in vitro. / The NADPH oxidase of phagocytes is an enzyme involved in the innate defense of organisms against pathogens. After phagocyte activation, this enzyme produces superoxide ions by reduction of dioxygen by NADPH. It is constituted of four cytosolic sub-units (p47phox ; p67phox ; p40phox et Rac) and two membrane proteins (gp91 ; p22phox). Its activation takes place through a complex process that involves protein-protein interaction changes leading to assembly and functionning of the catalytic core. In order to obtain information on this process, I have reconstituted the enzyme in a cell free systeme using recombinant proteins, to be able to fully control all the measurement conditions. In this work, we have compared different activation modes of p47phox i) phosphorylation; ii) substitution serine - aspartate by mutations at positions S303, S304 and S328 to mimic phosphorylation; iii) addition of arachidonic acid (AA), a well known activator molecule in vitro. It has been shown that these three activating methods transform p47phox to an open configuration with similar characteristics. However, we have found that the effects of these methods are significantly different. Indeed, the conformational changes observed by circular dichroism are different. For p47phox, the addition of AA destructures the protein. Its phosphorylation induces a bathochromic displacement of the bands, whereas the mutations S-D lead to an opposite displacement. For the dimer p47phox-p67phox , the addition of AA destructures the proteins while mutations induce hardly no changes. We have measured the dissociation constant Kd of the complex p47phox-p67phox. For wild type proteins, Kd value is low (4±2 nM), while mutations of p47phox as well as addition of AA increase its value up to 50 nM, showing a decrease of affinity between p47phox and p67phox. Moreover, on the whole complex, the effect of phosphorylation of p47phox is different from mutations. We have shown that the EC50 values relative to p67phox are sensitive to the various modifications of p47phox. Phosphorylation of p47phox decreases EC₅₀, while double or triple mutations increase its value. We have confirmed that phosphorylation and mutation are not sufficient to activate the enzyme. The presence of AA is a prerequisite for the functionning of the complex, i.e. production of superoxide. The binding order of the cytosolic proteins seems random but it is necessary that all the components be present during the activation by AA. Finally, deletion of the C terminal part of p47phox (aa 343 to 390, interaction domain with p67phox) leads to the absence of dimer formation but does not affect the enzyme activity. These results bring new information on the role of dimerisation of p47-p67 and on that of phosphorylation in the activation of NADPH oxidase in vitro.

La production hétérologue des protéines membranaires reste difficile, peut-être parce que l’insertion dans la membrane de la cellule hôte constitue une étape limitante de la production. Afin de tourner cette difficulté, deux modes de synthèse ont été envisagés: la synthèse de protéines dans un système a-cellulaire, en l’absence de membrane mais en présence de détergent, ou l’adressage forcé de la protéine vers les corps d’inclusion dans le cas d’une expression plus classique en bactérie entière. La réalisation des deux stratégies repose sur l’utilisation de protéines de fusion possédant une séquence d’entraînement en amont du gène d’intérêt, soit qu’elles améliorent la traduction du transcrit en limitant le repliement spatial de ce dernier, soit qu’elles favorisent la production de la protéine d’intérêt en corps d’inclusion. La porine OmpX et le peptide T7 ont été choisis en cas d’expression dans les systèmes bactériens. La protéine SUMO est utilisée pour la production dans un lysat eucaryote. Les différentes approches ont été testées sur la production de la pannexine1 humaine (Px1).Si les séquences d’entraînement OmpX et le peptide T7 sont correctement produites in vitro, aucune des deux, en revanche, ne favorise la production de la Px1. Seul l’entraîneur SUMO est efficace. En effet, nous avons observé que cette protéine augmente la production de la Px1 dans un lysat eucaryote de germe de blé. Par ailleurs OmpX, connue pour être largement produite in vivo dans les corps d’inclusion, n’entraîne pas la localisation de la Px1 dans ces structures. Contre toute attente, l’étiquette T7 dirige la Px1 dans les corps d’inclusion. L’étude électrophysiologique de la Px1 a donc été effectuée à partir de la protéine produite in vivo (T7his-Px1) après renaturation, ou produite sous forme soluble in vitro (his6-Px1) dans le lysat eucaryote. Dans le cas de la protéine T7his-Px1 renaturée, une activité canal qui rappelle celle qui est observée après expression dans l’ovocyte de Xénope, a été détectée en patch-clamp, mais dans trois cas seulement. Dans le cas de la protéine his6-Px1, aucune activité canal n’est clairement détectée. Dans une deuxième partie de ce travail on examine le rôle de la boucle périplasmique dans la sensibilité à la pression du MscL, un canal mécanosensible bactérien devenu un système modèle dans l’étude de la mécanosensibilité. Presque toutes les études fonctionnelles sur ce canal ont été réalisées sur le canal de E.coli, alors que la structure a été obtenue à partir de l’homologue de M. tuberculosis. Une étude fonctionnelle a montré que le MscL de M. tuberculosis est difficile à ouvrir : son ouverture requiert l’application d’une pression double de celle qui est nécessaire chez E.coli. Les deux homologues diffèrent principalement par la longueur de leur unique boucle périplasmique. De manière à examiner le rôle de la boucle, on a comparé l’activité du canal MscL de E.coli, celle du canal de M. tuberculosis et celle d’une protéine chimère constituée de la protéine de M. tuberculosis dans laquelle la boucle a été changée pour celle de la protéine de E.coli. De manière inattendue, nous avons constaté que les canaux de E.coli et de M. tuberculosis ont la même sensibilité à la pression. La protéine chimère n’avait pas d’activité canal. Si ce travail ne permet pas de conclure quant au rôle de la boucle, il montre sans ambigüité que contrairement à ce qui a été rapporté les canaux MscL de E.coli et de M. tuberculosis ne diffèrent pas sensiblement sur le plan fonctionnel / The production of heterologous membrane protein is notoriously difficult; this might be due to the fact that insertion of the protein in the membrane host is a limiting step. To by-pass this difficulty, two modes of synthesis were tested: 1) production in a cell-free system devoid of biological membrane but supplemented with detergent or liposomes, 2) production in bacteria, with targeting of the membrane protein to inclusion bodies. Both strategies were tested for the production of the human pannexin 1 channel (Px1). The gene coding the protein was fused with an “enhancer” sequence resulting in the addition of a peptide or short protein at the N terminus of the protein of interest. This enhancer sequence which is well produced in vitro or in vivo is supposed to facilitate the translation of the protein of interest. Three enhancer sequences were chosen: 1) the small porin OmpX of E. coli, which, in addition, should target the protein to inclusion bodies when the protein is expressed in bacteria 2) a peptide of phage T7 for expression in E.coli lysate or E.coli cells 3) the small protein SUMO for production in a wheat germ cell-free system. In a bacterial cell-free system, neither OmpX nor T7 promoted Px1 production. Px1 is only produced when the SUMO enhancer sequence is used in the wheat germ system. In bacteria, OmpX, known to form inclusions bodies did not promote the targeting of the fusion protein to inclusion bodies. Unexpectedly, the peptide T7 was able to do it.Px1 obtained from inclusion bodies (T7his-Px1) was renatured and reconstituted in liposomes. Similarly his6-Px1 produced in wheat germ system was reconstituted in liposomes. Both preparations were used for electrophysiological studies (patch-clamp and planar bilayers). With the refolded T7his-Px1, channel activity reminiscent of that observed with Px1 expressed in Xenope oocyte (Bao et al., 2004) could be detected, but only in three cases. In the case of his6-Px1, no clear channel activity could be observed. The second part of this work deals with the involvement of the periplasmic loop of the bacterial mechanosensitive channel MscL in its sensitivity to pressure. Mscl has become a model system for the investigation of mechanosensisity. Nearly all functional studies have been performed on MscL from E.coli while the structure of the protein has been obtained from the Mycobacterium tuberculosis homologue. In one functional study it was shown that MscL from M. tuberculosis is extremely difficult to open, gating at twice the pressure needed for E.coli MscL The periplasmic loop is the most variable sequence between the two homologues, being longer in E.coli than in M. tuberculosis. In order to assess the role of the periplamic loop in the sensitivity to pressure, we compared the activity of the E.coli and M. tuberculosis MscL and of a chimeric protein made of the M. tuberculosis protein in which the periplasmic loop has been exchanged for that of the E. coli channel. Unexpectedly, M. tuberculosis and E .coli MscL were observed to gate at a similar applied pressure. The chimeric protein had no functional activity. In conclusion, this study does not allow any conclusion as to the role of the loop in the sensitivity to pressure, but it shows clearly that, in contrast to the results of a previous study, there is no functional difference between E. coli and M. tuberculosis MscL.

Synthesizing proteins for use in therapeutics is restrained by, in part, contaminants in in vivo expression systems and limited production capacity of in vitro systems. Cell free expression (CFE) systems have emerged as a potential alternative for protein expression because of the inherently lower contents of contaminants, and their flexible modular design that allows the addition of factors that aid in synthesis of complex products. Here, we investigate and establish an in-house Escherichia coli-based cell free protein synthesis (CFPS) system, explore different CFPS commercial kits, develop assays to test performance of these systems and identify potential rules that dictate expression levels. Using CFE, we were able to test different vectors and conditions of system, as well as scale-up protein synthesis reactions. In conclusion, this work shows that CFPS is a functional and easy-to-use platform and can potentially meet the requirements for the synthesis of therapeutics.