Le travail de cette thèse vise à contrôler la formation d’une structure fortement ordonnée à l’échelle supramoléculaire d'un liquide associé, par dilution dans un liquide aprotique, ou par confinement dans un silicate mésoporeux (MCM-41, SBA-15). Dans le volume, une forte perturbation du mélange idéal représentée en termes de regroupement entre les espèces similaires est observée par la diffusion de neutrons aux petits angles. Sous confinement, un phénomène remarquable de nanoségrégation des mélanges dans les pores conduit à des structures concentriques tubulaires de type coeur-gaine, ce qui est surprenant pour les systèmes binaires macroscopiquement homogène et entièrement miscibles. La dynamique moléculaire de ces nanostructures originales est étudiée par diffusion quasiélastique de neutrons dans le liquide et les états vitreux, révélant différents types de modes de relaxation sous confinement et étant fortement dépendant de la concentration. L’étiquetage isotopique HD des composants des mélanges fournit un point de vue global unique sur les propriétés de ces liquides binaires confinés. Ceci permet d’établir une corrélation directe entre les hétérogénéités dynamiques spatialement distinctes et la formation de nanostructures d'enveloppe de base d'origine induites par des interactions préférentielles. / The objective of this thesis is tuning the formation of ordered supramolecular structures of an H-bonded liquid alcohol, either by dispersion in an aprotic solvent or by confinement in mesoporous silicates (MCM-41 and SBA-15). In the bulk, a strong perturbation from ideal mixing depicted in terms of clustering between similar species is observed by small angle neutron scattering. Under confinement, a remarkable nanosegregation phenomenon of the mixture is observed in the pores, leading to concentric tubular structures of core-shell type, which are striking for macroscopically homogeneous and fully miscible binary systems. The molecular dynamics of these original glassforming nanostructures is studied by quasielastic neutron scattering in the liquid and vitreous states, revealing different types of relaxation modes under confinement with nontrivial concentration dependencies. Isotopic HD labelling of the mixtures components provides a unique comprehensive viewpoint on the properties of these confined binary liquids, establishing a direct correlation between spatially segregated dynamical heterogeneities and the formation of original core shell nanostructures induced by preferential interactions.

Fe-Cr based steels, i.e. stainless steels, possessing a combination of excellent corrosion resistance and good mechanical properties, have indispensable applications ranging from low-end cooking utensils, to sophisticated components for nuclear power plants. However, the bcc/bct phase containing stainless steels which have a miscibility gap (MG) suffer from the so-called “475 oC embrittlement” leading to hardness increase and toughness deterioration. It occurs due to demixing of Fe and Cr leading to the formation of Fe-rich (α) and Cr-rich (α′) regions in bcc/bct phases. The demixing is referred to as phase separation (PS). The goal of this work was to study PS in ferrite containing stainless steels mainly by small-angle neutron scattering (SANS). Firstly, the application of different experimental techniques for the study of phase separation in Fe-Cr based steels was reviewed and supplemented by new measurements. SANS was shown to be very sensitive to the nanostructure change caused by PS and capable of characterizing the early stages of PS in Fe-Cr alloys. However, atom probe tomography and transmission electron microscopy are complementary to SANS. Therefore, in order to have a more complete view of the microstructure, the combination of these techniques should be pursued. Secondly, the factors affecting the initial microstructure prior to aging treatment and the effect of the resulted initial microstructure on PS were systematically investigated using binary Fe-Cr model alloys. The critical temperature of the MG was determined to be located between 560 and 580 oC in binary Fe-Cr. The results indicate that the solution treatment temperature above the MG and the cooling rate after solution treatment have significant effects on the initial microstructure and thus on PS during subsequent aging. The mechanisms responsible for the changed aging behavior are Cr clustering, quenched-in vacancy and decomposition during cooling. Therefore, computational simulations should take into account these factors and the initial microstructure to make predictions that are more accurate. Thirdly, the study was extended to PS in commercial duplex stainless steels (DSSs) which are of practical importance in various industries, e.g., nuclear power. It is found that alloying elements have an important effect on PS in DSSs. The grade 2507 (25 %Cr, 7 %Ni) experiences stronger PS than grade 2205 (22 %Cr, 5 % Ni) for the same heat treatment. Moreover, the fracture mechanisms as well as the mechanical properties depend on the extent of PS.  Finally, the fundamental aspects regarding the neutron scattering behavior for Fe-Cr alloys were examined. The results show that the nuclear and magnetic scattering of neutrons depend on the evolution of the nanoscale compositional fluctuation in Fe-Cr alloys. The ratio of the magnitude of nuclear scattering versus magnetic scattering varies with the extent of PS. / Stål baserade på Fe-Cr systemet, det vill säga rostfria stål, som har en kombination av utmärkta korrosionsegenskaper och bra mekaniska egenskaper, har många tillämpningar; allt från köksredskap, till sofistikerade komponenter för kärnkraftverk. Rostfria stål som innehåller Bcc / bct-fasen och som således har en blandningslucka, är känsliga för den så kallade "475 °C försprödningen" som leder till en hårdhetsökning men kraftigt försämrad slagseghet. Detta uppstår på grund av en uppdelning av Fe och Cr som leder till bildandet av Fe-rika (a) och Cr-rika (a’) regioner i bcc / bct-fasen. Denna uppdelning brukar kallas fasseparation. Målet med detta arbete var att studera fasseparationen i ferrit-innehållande rostfria stål främst genom lågvinkel-spridning av neutroner (SANS). Till att börja med studerades och jämfördes olika experimentella tekniker för undersökning av fasseparation i Fe-Cr-baserade stål med nya SANS- mätningar. SANS visade sig vara mycket känslig för förändringar på nano-skala orsakad av fasseparation och tekniken visade sig även kapabel att karakterisera de tidiga stadierna av fasseparation i Fe-Cr-legeringar. För att få en mer fullständig bild av mikrostrukturen efter fasseparation, bör emellertid en kombination av SANS och komplementära tekniker, såsom atomsond och transmissions-elektronmikroskopi, användas. Vidare undersöktes de faktorer som påverkar den ursprungliga mikrostrukturen före åldringsbehandling, och effekten av den initiala mikrostrukturen på fasseparation studerades systematiskt med användning av binära modell-legeringar av Fe-Cr. Den kritiska temperaturen för blandningsluckan i Fe-Cr bestämdes vara belägen mellan 560 och 580 °C. Resultaten indikerar att temperaturen för upplösningsbehandling ovanför blandningsluckan och kylhastigheten har en signifikant inverkan på den initiala mikrostrukturen och därmed på fasseparationen under efterföljande åldring. Mekanismerna som är ansvariga för det förändrade åldringsbeteendet är: Cr-klustring, insläckta vakanser och fasseparation under kylning. Simuleringar av fasseparationen bör därför ta hänsyn till dessa faktorer och den ursprungliga mikrostrukturen för att göra mer exakta förutsägelser av hur mikrostrukturen utvecklar sig med åldringstiden. Fasseparationen i kommersiella duplexa rostfria stål (DSS), som är av stor praktisk betydelse i olika branscher, t ex kärnkraft, studerades också med SANS. Det visade sig att mängden av olika legeringselement har en viktig effekt på graden av fasseparation i DSS. Legeringen 2507 uppvisade en tydligare fasseparation jämfört med legering 2205 för samma värmebehandling. Brottmekanismerna såväl som de mekaniska egenskaperna visade sig bero på omfattningen av fasseparationen. Slutligen undersöktes de grundläggande aspekterna hos neutronspridnings-beteendet för binära Fe-Cr-legeringar. Resultaten visade att kärn- och magnetisk spridning av neutroner beror på utvecklingen av sammansättningsfluktuationerna på en nanoskala i Fe-Cr-legeringar. Förhållandet mellan magnetisk- och kärnspridning varierar med omfattningen av fasseparationen. / <p>QC 20171117</p>

L’utilisation de la mousse en récupération assistée du pétrole (Enhanced Oil Recovery, EOR) présente un avantage indéniable par rapport à l’injection du gaz seul pour pallier les problèmes de ségrégation gravitaire et de digitations visqueuses. Son utilisation systématique en ingénierie du réservoir nécessite des connaissances plus approfondies sur son comportement en milieu poreux. La littérature montre deux types d’approches expérimentales basées soit sur des études pétrophysiques effectuées sur des systèmes poreux 3D et basées sur des mesures de pressions intégrées sur l’ensemble du milieu poreux, soit sur des études micro-fluidiques qui permettent une visualisation directe de l’écoulement mais qui sont limitées à des systèmes modèles dans des géométries à 1 ou 2 dimensions. L’objectif de cette thèse est de faire le pont entre ces deux approches. La stratégie proposée consiste à caractériser in situ l’écoulement de la mousse dans des milieux poreux 3D à différentes échelles, en utilisant des techniques complémentaires permettant d’accéder à une large gamme de résolutions spatiale et temporelle. Un environnement instrumenté donnant accès aux mesures pétro-physiques classiques a été développé puis couplé à différentes cellules d’observation conçues spécifiquement pour chaque instrument de caractérisation. Dans un premier temps, un scanner X a été utilisé pour décrire et visualiser les écoulements de la mousse à l’échelle de la carotte. La rhéologie de la mousse à cette échelle a pu être étudiée en fonction des conditions d’injections comme la vitesse interstitielle du gaz et la qualité de mousse. Dans un deuxième temps, la technique de diffusion des neutrons aux petits angles (SANS) a permis de sonder la texture de la mousse en écoulement sur trois ordres de grandeurs en taille. Des informations in situ sur la texture de la mousse en écoulement (taille et densité des bulles et des lamelles) ont pu être mesurées pour différentes qualités de mousse puis en fonction de la distance au point d’injection. Une comparaison avec les caractéristiques géométriques du milieu poreux a également été effectuée. Dans un troisième temps, la micro-tomographie X rapide haute résolution sur Synchrotron a été utilisée pour visualiser la mousse en écoulement à l’échelle du pore. Cette technique a permis de confirmer de visu certaines caractéristiques de la mousse mesurées par SANS et de décrire en sus les effets d’intermittence du piégeage de la mousse. Cette étude constitue une étape importante de la caractérisation multi-échelle de l’écoulement des mousses en milieux poreux 3D et apporte des éléments de réponse à certaines hypothèses admises. / Foam has long been used as a mobility control agent in Enhanced Oil Recovery (EOR) processes to enhance sweep efficiency and overcome gravity segregation, viscous fingering and gas channeling, which are gas-related problems when the latter is injected alone in the reservoir. However, the systematic use of foam in reservoir engineering requires more in-depth knowledge of its dynamics in porous media. The literature shows two types of experimental approaches based either on petrophysical studies carried out on 3D porous systems and based on pressure measurements, or on microfluidic studies that allow direct visualization of foam flow but are limited to 1D or 2D model systems. The research investigated in this thesis aims to bridge the gap between these two approaches. The proposed strategy is to characterize in situ the foam flow in 3D porous media with techniques providing a wide range of temporal and spatial resolutions. A coreflood setup giving access to classical petro-physical measurements was developed and then coupled to different observation cells designed specifically for each characterization instrument. First, an X-ray CT scanner was used to describe and visualize the foam flow at the core scale. The rheological behavior of foam on this scale was studied as a function of the injection conditions such as gas velocity and foam quality. Secondly, Small Angle Neutron Scattering (SANS) was used to probe the foam structure in situ during the flow, on a wide length scale, up to three orders of magnitude in size. In situ foam texture (size and density of bubbles and lamellae) was measured for different foam qualities and at different propagation distances from the injection point. A comparison to the geometric characteristics of the porous medium was also realized. Thirdly, High Resolution Fast X-ray Micro-tomography on a Synchrotron was used to visualize the foam flow at the pore scale. This allowed to confirm visually some foam characteristics measured with SANS and to investigate on local intermittent gas trapping and mobilization. This study is an important step in the multi-scale characterization of foam flow in 3D porous media and provides some answers to certain generally accepted assumptions.

Indiana University-Purdue University Indianapolis (IUPUI) / Large protein molecules are abundant in biological cells but are very difficult to study in physiological conditions due to molecular disorder. For large proteins, most structural information is obtained in crystalline states which can be achieved in certain conditions at very low temperature. X-ray and neutron crystallography methods can then be used for determination of crystalline structures at atomic level. However, in solution at room or physiological temperatures such highly resolved descriptions cannot be obtained except in very few cases. Scattering methods that can be used to study this type of structures at room temperature include small-angle x-ray and neutron scattering. These methods are used here to study two distinct proteins that are both classified as glycoproteins, which are a large class of proteins with diverse biological functions. In this study, two specific plasma glycoproteins were used: Fibrinogen (340 kDa) and Alpha 1-Antitrypsin or A1AT (52 kDa). These proteins have been chosen based on the fact that they have a propensity to form very large molecular aggregates due to their tendency to polymerize. One goal of this project is to show that for such complex structures, a combination of scattering methods that include SAXS, SANS, and DLS can address important structural and interaction questions despite the fact that atomic resolution cannot be obtained as in crystallography. A1AT protein has been shown to have protective roles of lung cells against emphysema, while fibrinogen is a major factor in the blood clotting process. A systematic approach to study these proteins interactions with lipid membranes and other proteins, using contrast-matching small-angle neutron scattering (SANS), small angle x-ray scattering (SAXS) and dynamic light scattering (DLS), is presented here. A series of structural reference points for each protein in solution were determined by performing measurements under osmotic stress controlled by the addition of polyethylene glycol-1,500 MW (PEG 1500) in the samples. Osmotic pressure changes the free energy of the molecular mixture and has consequences on the structure and the interaction of molecular aggregates. In particular, the measured radius of gyration (Rg) for A1AT shows a sharp structural transition when the concentration of PEG 1500 is between 33 wt% and 36 wt%. Similarly, a significant structural change was observed for fibrinogen when the concentration of PEG 1500 was above 40 wt%. This analysis is applied to a study of A1AT interacting with lipid membranes and to a study of fibrinogen polymerization in the presence of the enzyme thrombin, which catalyzes the formation of blood clots. The experimental approach presented here and the applications to specific questions show that an appropriate combination of scattering methods can produce useful information on the behavior and the interactions of large protein systems in physiological conditions despite the lower resolution compared to crystallography.

Yes / Understanding the structure-property relationships of drug delivery system (DDS) components is critical for their development and the prediction of bodily performance. This study investigates the effects of introducing polyethylene oxide (PEO) homopolymers, over a wide range of molecular weights, into Pluronic injectable smart hydrogel formulations. These smart DDSs promise to enhance patient compliance, reduce adverse effects and dosing frequency. Pharmaceutically, Pluronic systems are attractive due to their unique sol-gel phase transition in the body, biocompatibility, safety and ease of injectability as solutions before transforming into gel matrices at body temperature. This paper presents a systematic and comprehensive evaluation of gelation and the interplay of microscopic and macroscopic properties under both equilibrium and non-equilibrium conditions in controlled environments, as measured by rheology in conjunction with time-resolved Small Angle Neutron Scattering (SANS). The non-equilibrium conditions investigated in this work offer a better understanding of the two polymeric systems’ complex interactions affecting the matrix thermo-rheological behaviour and structure and therefore the future release of an active pharmaceutical ingredient from the injectable DDS.

No description available.

Le but du travail est d'étudier la structure supramoléculaire de mélanges de tensioactif hydrophile, n-octyl-beta-glucoside (C8G1), et d'un extractant d'ions métalliques hydrophobe, le tributyl-n-phosphate (TBP), en présence d'eau, d'huile et de sels. Les systèmes classiques d'extraction ionique (composés d'une phase aqueuse, d'huile et d'extractant dont le but est d'extraire un soluté de la phase polaire sont passés en revue. L'aspect colloïdal et les transitions de phases que l'on retrouve dans ces systèmes sont souvent décrits singulièrement. Nous avons transposé l'approche « diagramme de phases » issue de la physico-chimie des systèmes moléculaires organisés à ces systèmes d'extractant afin d'orienter globalement l'analyse de ces systèmes complexes. La discussion est basée sur des considérations géométriques. Un modèle thermodynamique a été développé en considérant les contraintes d'empilement des ces extractants dans le film moléculaire formant les micelles inverses d'extractant dans l'huile. Ce modèle a permis de prédire la solubilité de l'eau au sein de ces micelles inverses ainsi que leurs tailles obtenues expérimentalement. Dans une deuxième partie, le comportement physico-chimique des phases aqueuses et organiques composées respectivement d'eau/C8G1 et de TBP/huile/eau ont été étudiées, en s'intéressant particulièrement aux effets de sels, par des techniques de diffusion de rayons X aux petits angles, diffusion dynamique de la lumière et de spectroscopie UV-visible. Dans la dernière partie la description complète de la microémulsion en faisant varier la balance hydrophile-hydrophobe du mélange C8G1 et TBP a été obtenue en combinant des mesures de diffusion de neutrons aux petits angles et d'analyse chimique (Karl-Fischer, Carbone Organique Total, ICP-OES…). Le comportement co-surfactant du TBP a été déterminé par comparaison aux co-surfactants classiques que sont les n-alcools (4<n<8). Les compositions de films moléculaires mixtes de C8G1/TBP et de C8G1/n-hexanol, obtenues expérimentalement, ont été confirmées par un modèle basé sur des paramètres géométriques moléculaires. Nous avons tenté d'exploiter les propriétés interfaciales de ces molécules pour le contrôle des cinétiques d'extraction liquide-liquide d'ion et la séparation d'ion « sans solvant » par flottation. / The presented work describes the supramolecular structure of mixtures of a hydrophilic surfactant n-octyl-beta-glucoside (C8G1), and the hydrophobic metal ion extractant tributylphosphate (TBP) in n-dodecane/water as well as in the presence of salts.In the first part, basic solvent extraction system, composed of water, oil and extractant, will be introduced. The focus, however, lies on the extraction of multivalent metal ions from the aqueous phase. During this extraction process and in the following thermodynamic equilibrium, aggregation and phase transition in supramolecular assemblies occur, which are already described in literature. Notably, these reports rest on individual studies and specific conclusions, while a general concept is still missing. We therefore suggest the use of generalized phase diagrams to present the physico-chemical behaviour of (amphiphilic) extractant systems. These phase diagrams facilitated the development of a thermodynamic model based on molecular geometry and packing of the extractant molecules in the oil phase. As a result, we are now in the position to predict size and water content of extractant aggregates and, thus, verify the experimental results by calculation.Consequently, the second part presents a systematic study of the aqueous and organic phase of water/C8G1 and water/oil/TBP mixtures. The focus lies on understanding the interaction between metal ions and both amphiphilic molecules by means of small angle x-ray scattering (SAXS), dynamic light scattering (DLS) and UV-Vis spectroscopy. We confirmed the assumption that extraction of metal ions is driven by TBP, while C8G1 remains passive. In the third and last part, microemulsions of C8G1, TBP, water (and salt) and n-dodecane are characterized by small angle neutron scattering (SANS), and chemical analytics (Karl Fischer, total organic carbon, ICP-OES,...). The co-surfactant behaviour of TBP was highlighted by comparison to the classical n-alcohol (4<n<8) co-surfactants. The compositions of the C8G1/TBP and C8G1/n-hexanol interfacial mixed films obtained experimentally were confirmed by the prediction of a model based on the molecular geometrical parameters. We furthermore exploit the interfacial properties of these molecules to control the kinetics of liquid-liquid extraction and attempt a “solvent free” ion separation using flotation.

Cette thèse présente une étude des solutions aqueuses de trois complexes de polyélectrolytes (PECs) modèles. Les PECs résultent de la complexation de deux polyélectrolytes linéaires de charge opposée: un polycation (le PDADMA) et trois polyanions de longueur de persistance non électrostatique distinctes: le polystyrene sulfoné (PSS), le poly(α-méthyl styrène sulfoné) (PαMSS) et l’acide hyaluronique (HA). En plus de l’influence de la rigidité intrinsèque des polyanions sur la formation et la structure des PECs, les effets de la force ionique et de la méthode de préparation des solutions aqueuses de PECs ainsi que l’influence de la concentration des solutions binaires de polyélectrolytes initiales ont également été abordés. Nous suggérons que le processus de complexation entre polycations et polyanions en régimes semidilué et concentré est analogue à une gélification. Il y a toutefois une différence entre les deux complexes qui est reliée à la différence de rigidité intrinsèque. / This PhD thesis presents a study of the aqueous solutions of three model polyelectrolyte complexes (PECs). PECs were formed between hydrophilic and highly charged linear macrocations of poly(diallyldimethyl ammonium) (PDADMA) and linear macroanions of distinct intrinsic persistence lengths: sulfonated polystyrene (PSS), sulfonated poly(α-methyl styrene) (PαMSS) and hyaluronate (HA). In addition to the effect of the macroion stiffness on the PEC formation and structure, those of the ionic strength and the way of preparing the PEC aqueous solutions as well as that of the concentration regimes of the initial PE aqueous solutions were also tackled. We suggest the complexation between macrocations and macroanions in the semidilute and concentrated regimes can be described as a universal gelation process. A difference between PDADMA-PSS and PDADMA-HA complexes is related to the primary self-assembling process and is associated with the distinct structural models for PECs.

The EcoR124INT restriction-modification (R-M) system contains the genes HsdS3, HsdM and HsdR. S3 encodes the N-terminal domain of the wild-type S subunit and has been shown to dimerise in solution (Smith et al., 1998). Following purification of the subunits of the EcoR124INT R-M system, complexes of the methyltransferase S3/M and restriction endonuclease S3/M/R were formed and shown to have activity in vitro, methylating and hydrolysing a symmetrical DNA recognition sequence, respectively. The DNA mimic OCR (overcome classical restriction) protein inhibited the methyltransferase activity in vitro, with maximum inhibition at a 1: 2 molar ratio of (S3/M)2 to an ocr dimer. Dynamic light scattering (DLS), sedimentation equilibrium (SE) and sedimentation velocity (SV) experiments showed S3 to exist as a dimer and S11 (the central conserved domain of S) to exist as a tetramer in solution. M was found to be dimeric in solution, whilst the R protein was monomeric. A complex of S3/M was found to have a stoichiometry (S3/M)2 and a complex of S3/M/R had a stoichiometry of S3/M/R1, even when a 2: 1 molar ratio of R to S3/M, was added. Small angle neutron scattering (SANS) experiments provided values for the radius of gyration (Rg), which for S3 was comparable to that calculated for the recently published crystal structure of the S subunit from Methanococcus jannaschii (Kim et al., 2005). These experiments also showed a decrease in the Dmax in the presence of the 30 bp DNA recognition sequence from 200A to 140A, suggesting a similar conformational change in the positioning of the subunits as has been detected for the wild-type M. EcoR124I and a related type 1 1/2 system AhdI. This change following DNA binding was also observed by SV experiments. Furthermore ab initio modelling from the SANS data has provided a low-resolution structure for the EcoR124INT MTase and its complex with DNA.

Les rhabdovirus, dont les virus de la stomatite vésiculaire (VSV) et de la rage (RAV) constituent des prototypes, sont des virus enveloppés dont le génome est constitué d'une seule molécule d'ARN simple brin de polarité négative qui font partie de l'ordre des Mononegavirales (MNV). La machinerie de transcription/réplication de ces virus est constituée de l'ARN génomique et de trois protéines qui sont communes à tous les virus de l'ordre des MNV, la (N) qui encapside le génome viral, la grande sous-unité de l'ARN polymérase ARN dépendante (L) et la phosphoprotéine (P) qui est un cofacteur non-catalytique de la L et sert de chaperonne à la N. Le premier objectif de mon travail de thèse consistait à déterminer la structure cristallographique du domaine de dimérisation de la phosphoprotéine du virus de la rage. La P des rhabdovirus est une protéine modulaire qui contient deux régions intrinsèquement désordonnée, un domaine central responsable la dimérisation et un domaine C-terminal responsable de la fixation sur la matrice N-ARN. Le modèle atomique obtenu à une résolution de 1.5A montre que la structure est très différente de celle du domaine correspondant chez VSV. Le second objectif de mon travail était la caractérisation structurale de la grande sous-unité L de la polymérase du virus de la stomatite vésiculaire. Cette enzyme de 2109,aa, possède six régions conservées. Le domaine conservé III comprend les régions impliquées dans l'activité de polymérisation et les domaines V et VI sont responsables de la formation de la coiffe des ARNm. Plusieurs stratégies ont été envisagées successivement. (1) Sur la base de prédictions de structures secondaires et de prédictions de désordre, nous avons essayé d'exprimer différents fragments en système d'expression bactérien. Les constructions testées se sont avérées insolubles et certaines d'entre elles fixaient GroEL, indiquant un problème de repliement. (2) Nous avons alors essayé d'exprimer la L seule ou en complexe avec la P en système d'expression eucaryote. La purification s'est avérée impossible, la protéine L restant toujours associées à des protéines cellulaires visibles par coloration au bleu de Coommassie. (3) Finalement nous avons réussi à purifier la polymérase à partir de virus entier. La préparation de la polymérase était très homogène et a permis d'entreprendre une caractérisation par microscopie électronique. Une classification d'images a permis de construire un premier modèle à basse résolution. Le modèle révèle la présence d'un domaine annulaire avec plusieurs domaines structurés attachés au coeur de la polymérase. La cryo-microscopie électronique et la tomographie permettront d'obtenir plus de détails sur cette protéine. / Rhabdoviruses, including vesicular stomatitis virus (VSV) and rabies virus (RAV), are enveloped viruses which genome is made of a single molecule of negative-sense RNA and are classified in the order Mononegavirales (MNV). The transcription/replication machinery of these viruses consists of the genomic RNA and of three proteins, which are common to all other viruses of the order MNV, a nucleoprotein (N) that encapsidates the viral genome, a large subunit of the RNA-dependent RNA polymerase (L) and a phosphoprotein (P) that acts as a non-catalytic cofactor of L and a chaperone of N. The first goal of my research project was to determine the crystallographic structure of the dimerization domain of the rabies virus phosphoprotein. The P protein of the rhabdoviruses is a modular protein, which contains two intrinsically disordered regions, a central dimerization domain and a C-terminal domain involved in binding to the N-RNA template. The atomic model obtained at a resolution of 1.5 A showed that the structure is different from that of the corresponding domain of VSV. The second goal was the structural characterization of the large subunit L of VSV polymerase. The enzyme of 2109 aa has six conserved regions. Conserved region III includes the residues involved in the RNA synthesis activity, whereas domains V and VI are involved in mRNA capping formation. Three strategies were successively developed: (1) On the basis of secondary structure and disorder predictions, we tried to express different fragments in bacterial expression systems. These constructions appeared to be insoluble and some of them bound GroEL suggesting a folding problem; (2) We tried to express L alone or co-express it with P in eukaryotic expression system. The purification appeared to be impossible, the L protein always remaining associated with host-cell proteins in amounts detectable by Coommassie staining; (3) We succeeded in purifying the L protein from the virus. The L samples were homogenous and allowed a characterization by electron microscopy. Image classes allowed the reconstruction of a first low-resolution model. This model revealed the presence of a large ring-like domain and several globular domains. Cryo-electron microscopy and tomography should lead to a more detailed description of this protein.