Septinas constituem uma família conservada de proteínas de citoesqueleto pertencentes à superclasse das P-loop GTPases. Tais proteínas estão envolvidas em vários processos celulares. Em humanos, algumas septinas também estão relacionadas a casos de patologia. Suas seqüências são divididas em três domínios: domínio N-terminal, domínio GTPase e domínio C-terminal, que geralmente possui predição de coiled coil. A principal característica da família está na capacidade de seus membros formarem filamentos compostos por septinas diferentes. Em 2007, Sirajuddin et al apresentaram a primeira e única estrutura cristalográfica de um complexo de septinas, formado pelas septinas 2, 6 e 7. Embora os domínios C-terminal estivessem presentes, eles não apresentaram densidade eletrônica. Assim, a estrutura não trouxe informação estrutural sobre tais domínios. Atualmente, existem quatro estruturas de septinas depositadas no PDB: complexo 2/6/7 e três estruturas SEPT2 sem o C-terminal. Dada a inexistência de informações estruturais a nível atômico para os domínios C-terminal e baixa qualidade das poucas estruturas existentes, propusemos a obtenção de informações bioquímicas e estruturais dos domínios C-terminal de septinas humanas e a obtenção da estrutura cristalográfica de SEPT3-GC (GTPase mais C-terminal). Vale ressaltar que SEPT3 pertence ao único grupo de septinas que possui predição de não apresentar um coiled coil no C-terminal e para o qual não há nenhuma estrutura disponível. Expressamos e purificamos os domínios C-terminal de SEPT2, SEPT6 e SEPT7 (SEPT2-C, SEPT6-C e SEPT7-C). Mostramos que eles formam homo dímeros e que SEPT6-C e SEPT7-C formam um hetero dímero (KD 15,8 nM), nomeado por SEPT67-C. Tanto SEPT6-C quanto SEPT7-C tendem a precipitar, ao passo que SEPT67-C e SEPT2-C (KD 4 μM) são estáveis à altas concentrações. Tentamos, sem sucesso, cristalizar SEPT2-C e SEPT67-C. Via ressonância magnética nuclear, vimos que SEPT2-C possui duas regiões dinamicamente diferentes, uma central, em α-hélice, e duas extremidades desestruturadas. Neste ponto, planejamos construções para as regiões centrais dos domínios C-terminal, nomeadas SEPT2-CC, SEPT4-CC, SEPT6-CC e SEPT7-CC, referente às septinas 2, 4, 6 e 7. Obtivemos cristais para SEPT2-CC, SEPT4-CC e SEPT6-CC. Contudo, resolvemos apenas a estrutura de SEPT4-CC, mostrando que a construção forma um coiled coil anti-paralelo. Então, propusemos, pela primeira vez, um possível mecanismo de formação de ligações cruzadas entre filamentos de septinas. Por outro lado, obtivemos cristais para uma construção contendo os domínios GTPase e C-terminal de SEPT3 (SEPT3-GC) e resolvemos sua estrutura (2,9 Å). Vimos que SEPT3-GC forma filamentos no cristal, utilizando as mesmas duas interfaces já descritas anteriormente para as outras estruturas. Comparamos a estrutura obtida com estruturas da literatura e observamos diferenças significativas em algumas regiões, além de diferença em relação à orientação das duas subunidades adjacentes. Deve ser notado que a estrutura de SEPT4-CC é a primeira estrutura de um coiled coil de septina e que a estrutura de SEPT3-GC é a primeira estrutura de uma septina do grupo I. Em conclusão, o presente trabalho apresentou um conjunto de resultados os quais auxiliará no entendimento desta intrigante família de proteínas, inclusive em relação à formação de filamentos e as interações entre estes. / Septins are a conserved family of cytoskeletal proteins belonging to the superclass of the Ploop-GTPases, involved in various cellular processes. In humans, some septins are also linked to cases of pathology. Their sequences are divided into three domains: an N-terminal domain, a GTPase domain and a C-terminal domain, which usually is predicted to form a coiled coil. The main feature of the family is the ability of its members to form filaments composed of different septins. In 2007, Sirajuddin et al presented the first and only crystal structure of acomplex of septins, formed by septins 2, 6 and 7. Although the C-terminal domains were present, they showed no electron density, indicating disorder. Thus, this structure provides little structural information concerning their nature. Currently, there are four septin structures deposited in the PDB: the 2/6/7 complex and three structures of SEPT2 lacking the Cterminal domain . Based on the absence of structural information at atomic resolution about the C-terminal domains and the low quality of the few existing structures we set out to characterize both biochemically and structurally the C-terminal domains of selected human septins as well as to obtain the crystal structure of SEPT3-GC (a construct containing the GTPase and C-terminal domains). It is noteworthy that SEPT3 belongs to the only group of septins that are predicted to lack the C-terminal coiled coil and for which no crystal structure is available. We have expressed and purified the C-terminal domains of SEPT2, SEPT6 and SEPT7 (SEPT2-C, SEPT6-C and SEPT7-C). We show that they form homo-dimers and that SEPT6-C and SEPT7-C form a hetero-dimer (KD 15.8 nM), SEPT67-C. Both SEPT6-C and SEPT7-C were unstable, but SEPT67-C and SEPT2-C (KD 4 μM) were both stable at high concentrations. NMR measurements showed that SEPT2-C has two dynamically different regions, a central one which is -helical, and two extremities which are unstructured. Constructs were designed for the central regions of the C-terminal domains of septins 2, 4, 6 and 7 (SEPT2-CC, SEPT4-CC, SEPT6-CC and SEPT7-CC). We have obtained crystals of SEPT2-CC, SEPT4-CC and SEPT6-CC and have solved the structure of SEPT4-CC which unexpectedly is observed to form an anti-parallel coiled coil. We use this observation to propose, for the first time, a possible mechanism for the formation of cross-links between septin filaments. Crystals were obtained for SEPT3-GC its structure solved at 2.9 Å. We observe that SEPT3-GC forms filaments in the crystal, employing the same two interfaces previously described for other structures. We compare the structure obtained with those from the literature and observe significant differences in some regions of the molecule as well as differences in the relative orientation of the subunits. It should be noted that the structure of SEPT4-CC is the first structure of a septin coiled coil and that the structure of SEPT3-GC is the first structure of a septin from group I. In conclusion, this study presentsm results which will assist in the understanding of this intriguing protein family, including observations pertinent to filament formation and cross-linking.

The Rab family of small GTPases orchestrates intracellular endomembrane transport through the recruitment of diverse effector proteins. Since its first discovery in 1987, almost 70 Rab proteins have been identified in humans to date and their perturbed function is implicated in several hereditary and acquired diseases. In this Ph.D. thesis, I systematically characterize cell type expression and subcellular localization of all Rab proteins present in Drosophila melanogaster utilizing a genetic resource that represents a major advance for studying membrane trafficking in vivo: the ’Drosophila YRab library’. This collection comprises 27 different D. melanogaster knock-in lines that harbor YFPMyc fusions to each Rab protein, referred to as YRab. For each YRab, I present a comprehensive data set of quantitative and qualitative expression profiles across six larval and adult tissues that include 23 annotated cell types. The whole image data set, along with its annotations, is publicly accessible through the FLYtRAB database that links to CATMAID for online browsing of tissues. I exploit this data set to address basic cell biological questions. i) How do differentiating cells reorganize their transport machinery to perform cell type-specific functions? My data indicates that qualitative and quantitative changes in YRab protein expression facilitate the functional specialization of differentiated cells. I show that about half of the YRab complement is ubiquitously expressed across D. melanogaster tissues, while others are missing from some cell types or reflect strongly restricted cell type expression, e.g. in the nervous system. I also depict that relative YRab expression levels change as cells differentiate. ii) Are specific Rab proteins dedicated to apical or basolateral protein transport in all epithelia? My data suggests that the endomembrane architecture reflects specific tasks performed by particular epithelial tissues, rather than a generalized apicobasal organization. I demonstrate that there is no single YRab that is similarly polarized in all epithelia. Rather, different epithelial tissues dynamically polarize the subcellular localization of many YRab compartments, producing membrane trafficking architectures that are tissue- and stage-specific. I further discuss YRab cell type expression and subcellular localization in the context of Rab family evolution. I report that the conservation of YRab protein expression across D. melanogaster cell types reflects their evolutionary conservation in eukaryotes. In addition, my data supports the assumption that the flexible deployment of an expanded Rab family triggered cell differentiation in metazoans. The FLYtRAB database and the ’Drosophila Rab Library’ are complementary resources that facilitate functional predictions based on YRab cell type expression and subcellular localization, and to subsequently test them by genetic loss-of-function experiments. I demonstrate the power of this approach by revealing new and redundant functions for Rab23 and Rab35 in wing vein patterning. My data collectively highlight that in vivo studies of endomembrane transport pathways in different D. melanogaster cell types is a valuable approach to elucidate functions of Rab family proteins and their potential implications for human disease.

Septinas constituem uma família conservada de proteínas de citoesqueleto pertencentes à superclasse das P-loop GTPases. Tais proteínas estão envolvidas em vários processos celulares. Em humanos, algumas septinas também estão relacionadas a casos de patologia. Suas seqüências são divididas em três domínios: domínio N-terminal, domínio GTPase e domínio C-terminal, que geralmente possui predição de coiled coil. A principal característica da família está na capacidade de seus membros formarem filamentos compostos por septinas diferentes. Em 2007, Sirajuddin et al apresentaram a primeira e única estrutura cristalográfica de um complexo de septinas, formado pelas septinas 2, 6 e 7. Embora os domínios C-terminal estivessem presentes, eles não apresentaram densidade eletrônica. Assim, a estrutura não trouxe informação estrutural sobre tais domínios. Atualmente, existem quatro estruturas de septinas depositadas no PDB: complexo 2/6/7 e três estruturas SEPT2 sem o C-terminal. Dada a inexistência de informações estruturais a nível atômico para os domínios C-terminal e baixa qualidade das poucas estruturas existentes, propusemos a obtenção de informações bioquímicas e estruturais dos domínios C-terminal de septinas humanas e a obtenção da estrutura cristalográfica de SEPT3-GC (GTPase mais C-terminal). Vale ressaltar que SEPT3 pertence ao único grupo de septinas que possui predição de não apresentar um coiled coil no C-terminal e para o qual não há nenhuma estrutura disponível. Expressamos e purificamos os domínios C-terminal de SEPT2, SEPT6 e SEPT7 (SEPT2-C, SEPT6-C e SEPT7-C). Mostramos que eles formam homo dímeros e que SEPT6-C e SEPT7-C formam um hetero dímero (KD 15,8 nM), nomeado por SEPT67-C. Tanto SEPT6-C quanto SEPT7-C tendem a precipitar, ao passo que SEPT67-C e SEPT2-C (KD 4 μM) são estáveis à altas concentrações. Tentamos, sem sucesso, cristalizar SEPT2-C e SEPT67-C. Via ressonância magnética nuclear, vimos que SEPT2-C possui duas regiões dinamicamente diferentes, uma central, em α-hélice, e duas extremidades desestruturadas. Neste ponto, planejamos construções para as regiões centrais dos domínios C-terminal, nomeadas SEPT2-CC, SEPT4-CC, SEPT6-CC e SEPT7-CC, referente às septinas 2, 4, 6 e 7. Obtivemos cristais para SEPT2-CC, SEPT4-CC e SEPT6-CC. Contudo, resolvemos apenas a estrutura de SEPT4-CC, mostrando que a construção forma um coiled coil anti-paralelo. Então, propusemos, pela primeira vez, um possível mecanismo de formação de ligações cruzadas entre filamentos de septinas. Por outro lado, obtivemos cristais para uma construção contendo os domínios GTPase e C-terminal de SEPT3 (SEPT3-GC) e resolvemos sua estrutura (2,9 Å). Vimos que SEPT3-GC forma filamentos no cristal, utilizando as mesmas duas interfaces já descritas anteriormente para as outras estruturas. Comparamos a estrutura obtida com estruturas da literatura e observamos diferenças significativas em algumas regiões, além de diferença em relação à orientação das duas subunidades adjacentes. Deve ser notado que a estrutura de SEPT4-CC é a primeira estrutura de um coiled coil de septina e que a estrutura de SEPT3-GC é a primeira estrutura de uma septina do grupo I. Em conclusão, o presente trabalho apresentou um conjunto de resultados os quais auxiliará no entendimento desta intrigante família de proteínas, inclusive em relação à formação de filamentos e as interações entre estes. / Septins are a conserved family of cytoskeletal proteins belonging to the superclass of the Ploop-GTPases, involved in various cellular processes. In humans, some septins are also linked to cases of pathology. Their sequences are divided into three domains: an N-terminal domain, a GTPase domain and a C-terminal domain, which usually is predicted to form a coiled coil. The main feature of the family is the ability of its members to form filaments composed of different septins. In 2007, Sirajuddin et al presented the first and only crystal structure of acomplex of septins, formed by septins 2, 6 and 7. Although the C-terminal domains were present, they showed no electron density, indicating disorder. Thus, this structure provides little structural information concerning their nature. Currently, there are four septin structures deposited in the PDB: the 2/6/7 complex and three structures of SEPT2 lacking the Cterminal domain . Based on the absence of structural information at atomic resolution about the C-terminal domains and the low quality of the few existing structures we set out to characterize both biochemically and structurally the C-terminal domains of selected human septins as well as to obtain the crystal structure of SEPT3-GC (a construct containing the GTPase and C-terminal domains). It is noteworthy that SEPT3 belongs to the only group of septins that are predicted to lack the C-terminal coiled coil and for which no crystal structure is available. We have expressed and purified the C-terminal domains of SEPT2, SEPT6 and SEPT7 (SEPT2-C, SEPT6-C and SEPT7-C). We show that they form homo-dimers and that SEPT6-C and SEPT7-C form a hetero-dimer (KD 15.8 nM), SEPT67-C. Both SEPT6-C and SEPT7-C were unstable, but SEPT67-C and SEPT2-C (KD 4 μM) were both stable at high concentrations. NMR measurements showed that SEPT2-C has two dynamically different regions, a central one which is -helical, and two extremities which are unstructured. Constructs were designed for the central regions of the C-terminal domains of septins 2, 4, 6 and 7 (SEPT2-CC, SEPT4-CC, SEPT6-CC and SEPT7-CC). We have obtained crystals of SEPT2-CC, SEPT4-CC and SEPT6-CC and have solved the structure of SEPT4-CC which unexpectedly is observed to form an anti-parallel coiled coil. We use this observation to propose, for the first time, a possible mechanism for the formation of cross-links between septin filaments. Crystals were obtained for SEPT3-GC its structure solved at 2.9 Å. We observe that SEPT3-GC forms filaments in the crystal, employing the same two interfaces previously described for other structures. We compare the structure obtained with those from the literature and observe significant differences in some regions of the molecule as well as differences in the relative orientation of the subunits. It should be noted that the structure of SEPT4-CC is the first structure of a septin coiled coil and that the structure of SEPT3-GC is the first structure of a septin from group I. In conclusion, this study presentsm results which will assist in the understanding of this intriguing protein family, including observations pertinent to filament formation and cross-linking.

L'épithélium intestinal se renouvelle tous les 4 à 6 jours chez les mammifères grâce aux cellules souches localisées au fond des cryptes. Le renouvellement dépend des signaux émis par le microenvironnement et requiert une phase de prolifération des cellules souches, de différenciation et d'apoptose/desquamation des cellules épithéliales. La signalisation Wnt, joue un rôle majeur dans l'homéostasie intestinale par l'action de deux gradients inversés le long de l'axe crypte/lumière ; la signalisation Wnt canonique, active au fond des cryptes, contrôle la prolifération alors que la signalisation non-canonique, active vers le haut des cryptes contrôle la différenciation. Il a été montré que ces deux voies contrôlent l'activité de la GTPase atypique RhoU/Wrch1. RhoU fait partie des GTPases qui s'activent spontanément, son activité est donc directement proportionnelle à son niveau d'expression dans la cellule. Enfin, cette GTPase atypique est sous exprimée dans de nombreuses tumeurs gastriques et colorectales.Compte tenu de ces données, nos objectifs étaient donc de caractériser les changements morphologiques induits par l'invalidation conditionnelle de RhoU dans l'épithélium intestinal murin et d'en déterminer les mécanismes d'action. Nos résultats montrent que la déplétion de RhoU n'est pas létale, cependant elle a induit une augmentation de 20% de la densité cellulaire et une désorganisation de la structure de l'épithélium dans le haut des cryptes du colon. Cette augmentation concerne aussi bien les lignages sécrétoires et absorptifs, cependant, l'absence de RhoU a induit une sur-représentation du lignage sécrétoire. Dans la lignée de tumeur colorectale DLD-1, nous avons montré que l'absence de RhoU mime le phénotype d'augmentation de la densité cellulaire observé chez la souris. L'invalidation de RhoU ne modifie pas la distribution des phases du cycle cellulaire ni de celle de la mitose, cependant, elle réduit le nombre des cellules en apoptose dans le colon des souris et dans les cellules DLD-1. L'invalidation de RhoU a réduit la signalisation Hippo et a altéré la contractilité cellulaire via une augmentation de la phosphorylation de la protéine MLC2. Des travaux récents ont montré que la diminution du niveau MLC2 phosphorylée est nécessaire pour l'activation des caspases par un stimulus apoptotique. Ceci suggère que cette perturbation de la contractilité peut être à l'origine de cette diminution de l'apoptose qui est la cause majeure responsable de ce phénotype. En conclusion, RhoU est un régulateur de l'homéostasie intestinale chez la souris via son rôle modérateur de la mort cellulaire. / In Mammals, the intestinal epithelium is renewed every 4-6 days through the stem cells located at the bottom of crypts. The renewal depends on signals from the micro-environment and requires a proliferation phase of stem cells, then a differentiation and apoptosis/desquamation phases of epithelial cells. Wnt signaling plays a major role in intestinal homeostasis by the action of two reversed gradients along the axis crypt/ lumen: canonical Wnt signaling, active in the bottom of crypts, control proliferation while non canonical signaling, active in the top of the crypts control cell differentiation. It was shown that these two pathways are regulator of the atypical GTPase RhoU/Wrch1. The RhoU protein activates spontaneously, its activity is directly proportional to its expression level in the cell and is expressed as in gastric and colorectal tumors. In view of these informations, our objectives were therefore to characterizethe morphological changes induced by conditional invalidation of RhoU in the intestinal epithelium of mice and to determine the mechanisms of action. Our results show that RhoU depletion is not lethal. However, it induces an increase of cell density (+20%) and a disruption of the epithelium structure in the top of the colonic crypts. This increase affects both absorptive and secretory lineages. However, the absence of RhoU induced over-representation of secretory lineage. In colorectal tumor cell line DLD-1, we have shown that the absence of RhoU mimics the phenotype of cell density increase observed in mice. RhoU invalidationdid not change the distribution of cell cycle phases and mitosis, however, it reduces the number of apoptotic cells in the colon of mice and in the DLD-1 cells. RhoU invalidation reduced Hippo signaling and altered cell contractility via the increase of the protein MLC2 phosphorylation. Recent work has shown that the reduction of MLC2-P level is necessary for the caspase protein activation by an apoptotic stimulus. Suggesting that the perturbation of contractility may be the cause of this apoptosis decrease which is the main cause responsible of this phenotype. Finally, RhoU is a regulator of the intestinal homeostasis in micevia its moderating role of cell death.

Rab35 est une petite GTPase de la superfamille des Ras protéines. Chez l’Homme, plus de 60 Rabs jouent un rôle clef « d’interrupteur moléculaire » dans la régulation du trafic membranaire. Des progrès significatifs ont été réalisé sur le rôle de certaines Rabs, cependant il reste encore un travail important de compréhension fonctionnelle et du mécanisme d’action de celles-ci au sein de la cellule. L’actuel projet a pour but d’étudier Rab35, un régulateur essentiel d’une voie de recyclage vésiculaire vers la membrane plasmique, aussi impliqué dans la régulation de l’actine lors de la dernière étape de la division cellulaire. Rab35 agit via le recrutement et la régulation de protéines spécifiques appelées effecteurs. Chez l’humain, des mutations chez les partenaires de Rab35 entrainent de rares maladies connues sous le nom du syndrome de Birt-Hogg-Dube et syndrome de Lowe. Ce projet en étroite collaboration avec le laboratoire d’Arnaud Echard (Institut Pasteur) consiste à mieux comprendre le rôle cellulaire et fonctionnel de deux nouveaux effecteurs de Rab35 : MICAL1 et MiniBAR. Lors de la division cellulaire, MICAL1 pourrait réduire le niveau d’actine avant l’abscission car il possède un domaine catalytique monooxygénase démontré comme étant un facteur de désassemblage de l’actine. MICAL1 agit directement sur les filaments d’actine et oxyde des résidus spécifiques permettant la dépolymérisation de ces derniers. Mais il est actuellement inconnue si MICAL1 joue un rôle dans la division cellulaire. Par ailleurs, nos collaborateurs ont découvert un nouvel effecteur de Rab35 appelé MiniBAR. Ils ont récemment décrit MiniBAR comme étant un partenaire spécifique de Rab35 et contenant un domaine BAR (connue pour lier et courber les membranes). Ce domaine, adjacent au domaine d’interaction de Rab35, lie spécifiquement Rac1 une autre GTPase bien connue pour son rôle dans la régulation de l’actine. De ce fait MiniBAR pourrait être un lien entre les deux GTPases coordonnant le processus de remodelage de l’actine dans la cellule.Le laboratoire de Motilité Structurale dirigé par Anne Houdusse est spécialisé dans l’étude structurale des moteurs moléculaire et GTPases qui contribuent aux différentes fonctions dans la cellule par l’interaction avec leur différents cargos et effecteurs. L’élaboration de ce projet consiste à caractériser d’un point de vue structural et fonctionnel les complexes entre Rab35 et ses effecteurs. Laboratoire spécialisé dans la détermination de structure à haute résolution par cristallographie aux rayons X, ce travail mènera à la résolution de complexes macromoléculaires afin de désigner des mutants spécifiques pour des études fonctionnelles dans la cellule en étroite collaboration avec l’équipe d’Arnaud Echard, leader dans l’étude cellulaire de Rab35. / Rab35 is an essential regulator of a recycling pathway back to the plasma membrane, that is also required for the post-furrowing terminal steps during cytokinesis that are associated with F-actin depolymerisation [1]. Rab35 performs its role in the cell via recruitment and regulation of specific effector proteins. Recently the lab of our collaborator Arnaud Echard (Pasteur institute, Paris) has identified and is currently studying by cell biology approaches two novel Rab35 effectors - MICAL1 and MiniBAR proteins. MICAL1 may restrict actin levels before cytokinesis abscission, because it harbours a monooxygenase catalytical domain and has been shown to be an F-actin-disassembly factor [2]. But it is unknown whether MICAL1 has a function in cell division. Another effector of Rab35 currently uncharacterized was identified and named MiniBAR by our collaborators. They recently described that MiniBAR specifically binds to active Rab35, that it contains an unnoticed, putative BAR domain (known to sense membrane curvature) and that this domain, adjacent to the Rab35-binding domain, binds specifically to GTP-bound Rac1 - a well known actin remodelling regulator. So, MiniBAR may function as a linker between the two small GTPases coordinating actin-remodelling processes in the cell. The aim of the project is to perform extensive structural/functional characterization of complexes between the Rab35 GTPase and its interacting effector proteins.References:[1] - Dambournet and al, Nat Cell Biol, 2011. Rab35 GTPase and OCRL phosphatase remodel lipids and F-actin for successful cytokinesis.[2] - Giridharan SS and Caplan S, Antioxid Redox Signal, 2014. MICAL-family proteins: Complex regulators of the actin cytoskeleton.

Les cellules multiciliées (MCC) possèdent à leur surface apicale des centaines de cils mobiles générant un flux directionnel liquidien nécessaire par exemple pour le nettoyage des voies respiratoires. La fabrication de ces cils (multiciliogénèse) requiert une séquence d’évènements cellulaires dont un arrêt du cycle cellulaire, une réorganisation du réseau apical d’actine, une multiplication massive des centrioles suivie de leur migration au pôle apicale et de leur maturation en corps basal, à partir desquels les cils s’allongent.Mon laboratoire d’accueil a mis en évidence le rôle conservé de la famille de microARN miR-34/449 dans le contrôle de la multiciliogénèse en inhibant la voie de signalisation Notch ainsi qu’en induisant un arrêt du cycle. Au cours de ma thèse, j’ai mis en évidence un nouveau niveau de régulation de ces microARN par lequel ils contrôlent la réorganisation apicale du cytosquelette d’actine, en modulant l’expression et l’activité de certaines petites GTPases. Par ailleurs, j’ai identifié et caractérisé des séquences variantes des miR-34/449 canoniques, appelées isomiR. Tandis que ces isomiR partagent des fonctions semblables à celles de leurs homologues canoniques, ils apportent également une complémentarité d’action en modulant des transcrits cibles spécifiques. Enfin, le dernier axe de mon travail a permis d’identifier le rôle de la voie de signalisation BMP dans la multiciliogénèse ainsi que d’élucider certains des mécanismes moléculaires par lesquels elle contrôle ce phénomène. L’ensemble de nos découvertes offre une opportunité inédite pour développer des stratégies thérapeutiques dans le traitement de maladies associées à des désordres ciliaires / Vertebrate multiciliated cells (MCC) project hundreds of motile cilia at their apical surface which coordinately beat to generate a directional fluid flow necessary for many biological functions including airway cleansing. Biogenesis of multiple cilia (multiciliogenesis) follows different key cellular steps corresponding to a cell cycle arrest, a massive multiplication of centrioles which then migrate to the apical surface to form basal bodies, from which cilia elongate. In 2011, my host laboratory evidenced that the miR-34/449 family of microRNAs control vertebrate multiciliogenesis by inducing the cell cycle arrest and by repressing the Notch pathway. My thesis work has revealed a new role of miR-34/449 by demonstrating that they modulate expression and activity of small GTPases to drive the apical reorganization of the actin network, a prerequisite for basal body anchoring. Besides, I have identified and characterized variant sequences of canonical miR-34/449 family, named isomiRs. Whereas these isomiRs share common biological functions with canonical miR-34/449 miRNAs, they may also contribute to a complementary effect by targeting specific transcripts. Finally, the last part of my work has contributed to the identification of the conserved role of the BMP pathway in the control of multiciliogenesis. I have evidenced some molecular mechanisms by which the BMP signal controls this phenomenon. Importantly, I demonstrated that BMP inhibition promotes regeneration of tracheal MCC in vivo in an asthmatic mouse model. Overall, our findings offer an unprecedented opportunity to develop novel therapeutic strategies to treat diseases associated with ciliary disorders

The small GTPases function as molecular switches to control diverse signaling cascades. The mammalian Rit and Rin, along with Drosophila Ric, comprise an evolutionarily conserved subfamily of the Ras-related GTPases. Previous studies using cultured cell models suggested that Rit was involved in the control of cell proliferation, transformation, neuronal differentiation, morphogenesis, and cell survival, but the principal physiological function of Rit remained uncharacterized. To address this outstanding question, we employed a genetic approach, engineering a Rit knockout mouse. Using this animal model, we demonstrate a central role of Rit in governing cell survival in a p38-dependent fashion. Primary mouse embryonic fibroblasts (MEFs) derived from Rit-/- mice display increased apoptosis and selective disruption of MAPK signaling following oxidative stress. These deficits include a reduction in ROS-mediated stimulation of a novel p38-MK2-HSP27 signaling cascade, which appears to act upstream of the mTORC2 complex to control Akt-dependent cell survival. In the adult brain, proliferation of stem cells within the subgranular zone (SGZ) of the hippocampal dentate gyrus (DG), provide a lifelong supply of new neurons. Adult neurogenesis appears critical for learning and memory and is altered in animal models of brain injury and neurological diseases. Thus, a greater understanding of the regulation of adult neurogenesis will provide insight into its myriad physiological roles but also to the development of therapeutic strategies for the treatment of injury and the progression of brain diseases. Here we find that Rit plays a central role in governing the survival of hippocampal neurons in response to oxidative stress. Importantly, using a controlled cortical impact model of traumatic brain injury (TBI), we show that Rit acts to protect newborn immature neurons within the SGZ of the DG from apoptosis following TBI. Finally, studies indicate that Rit plays a significant role in directing IGF-1 signaling, a key neurotrophin known to promote neurogenesis and to protect neurons against apoptotic stress. Together, these studies establish Rit as a critical regulator of a p38 MAPKdependent signaling cascade that functions as an important survival mechanism for cells in response to oxidative stress, including the survival of newborn hippocampal neurons in the traumatically injured brain.

<p>Protein biosynthesis is a core process in all living organisms. Assembly of the protein chain from aminoacids is catalysed by the ribosome, ancient and extremely complex macromolecular machine. Several different classes of accessory molecules are involved in translation, and one set of them, called translational GTPases (trGTPases), was in the focus of this work. </p><p>In this thesis properties of two trGTPases– EF-G and eRF3 - were studied by means of direct biochemical experiments. EF-G is a bacterial trGTPase involved in two steps of translation: translocation and ribosomal recycling. Translocation is a process of the ribosomal movement along the mRNA, and recycling as the step when upon completion of the protein ribosome is released from the mRNA via splitting in two ribosomal subunits. We found that off the ribosome EF-G has similar affinities to GDP and GTP, and thus given the predominance of the latter in the cell, EF-G should be present mostly in the complex with GTP. However, binding to the ribosome increases factors affinity to GTP drastically, ensuring that it is in the GTP-bound state. GDP can not promote neither translocation, not recycling, and GDPNP can not promote recycling. It can, however, promote translocation, but in so doing it results in an intermediate ribosomal state and translocation process can be reversed by addition of GDP, which is not the case for the EF-G•GTP-catalyzed reaction.</p><p>The second trGTPase we investigated is eukaryotic termination factor eRF3. This protein together with another factor, eRF1, is involved translation termination, which is release of the synthesized protein from the ribosome. We demonstrateed, that eRF3 alone has basically no propensity to bind GTP and thus resides in the GDP-bound state. Complex formation between eRF1 and eRF3 promotes GTP binding by the latter, resulting in the formation of the ternary complex eRF1•eRF3•GTP, which in turn is catalyzing the termination event.</p><p>Experimental investigations of trGTPases where rationalized within a generalized thermodynamical framework, accommoding the existent experimental observations, both structural and biochemical. </p>

Rad/Rem/Rem2/Gem/Kir (RGK) proteins are Ras-like GTPases with diverse (and expanding) functions including: regulating cytoskeleton dynamics, cell proliferation, synaptogenesis, and inhibition of high voltage-dependent calcium (CaV) channels. Furthermore, they have tissue-specific distribution with Rem and Rad most highly expressed in the heart. Indeed, the importance of Rem and Rad in the cardiovascular system is underscored by a number of studies linking them to disease states including cardiac hypertrophy, cardiac fibrosis, and inflammation. A hallmark feature of RGK proteins is their ability to inhibit current through CaV channels (ICa) and in fact, they are recognized as the most potent endogenous inhibitors of ICa. However, how RGK proteins are regulated and what their physiological role is are unknown. Understanding these points is critical for defining the patho-physiological roles of RGK proteins. My thesis work contributes towards the RGK field on two fronts: First, we demonstrate that RGK proteins are non-canonical G-proteins in the context of their ability to undergo nucleotide regulation and second, we reveal a novel paradigm of RGK-mediated inhibition on CaV channels. In Chapters 2 and 3, we show that Rem and Rad are are non-canonical G-proteins with respect to the regulatory role of their guanine nucleotide binding pocket (GNBP). Canonical Raslike G-proteins contain a conserved G-domain that encompass a GNBP and is important for guanine nucleotide binding and hydrolysis. Since RGK proteins also possess a G-domain and GNBP as well as demonstrate bona fide nucleotide binding, it was initially thought that they were regulated in a manner similar to other Ras proteins. However, subsequent studies suggested that RGK proteins may not obey such a classical model and as a result, the regulatory role of their GNBP in the G-domain was unclear. By using a wide range of functional measurements (CaV1.2 currents, Ca2+ transients, β-subunit binding), we demonstrate that RGK proteins Rem and Rad are non-canonical G-proteins. Utilizing point mutants that abolish GTPbinding and prevent GTPase activity (RemT94N and RadS105N), we show that only some cellular functions are dependent on an operational nucleotide binding pocket while others are unperturbed. Specifically, Rem- and Rad-mediated inhibition of ICa is independent of guanine nucleotide regulation whereas protein interactions with the b-subunit of CaV channels (CaVβ) and protein stability are sensitive to nucleotide regulation. We also discover skeletal and cardiac actin to be novel binding partners of Rem. And lastly, we observe differences between the effects of Rem and Rad on their degree of ICa inhibition in cardiac myocytes. Thus, Rem and Rad are non-canonical G-proteins with respect to the regulatory role of their GNBP. In collaboration with a close colleague, Akil Puckerin, Chapter 4 reveals a novel mechanism behind RGK-mediated inhibition of ICa. Together, we show RGK proteins display different modes of inhibition against specific CaV channels and that we can utilize this property to design calcium channel blockers which inhibit CaV channels in an isoform specific manner. We demonstrate this by designing Rem and Rad mutants which have diminished CaVβ capacity, termed Rem-βNULL and Rad-βNULL, respectively. Characterization of these mutants using wholecell patch clamp experiments revealed that Rem-βNULL inhibits only CaV1.2 whereas Rad-βNULL inhibits only CaV1.2 and CaV2.2. Thus, our results describe the first genetically encoded calcium channel blocker that can selectively distinguish amongst L-type channels. Altogether, this thesis work contributes towards our understanding of RGK protein regulation function and the underlying mechanisms by which they inhibit ICa. These findings advance the field both from a mechanistic and physiological standpoint, and will be of great importance towards investigating the patho-physiological role of RGK proteins.