Respiratory syncytial virus (RSV) is a globally circulating, non-segmented, negative sense (NNS) RNA virus that causes severe lower respiratory infections. This study explored several avenues to ultimately expand upon our understanding of RSV pathogenesis at the protein level. Evaluation of RSV intrinsic protein disorder increased the relatively limited description of the RSV structure-function relationship. Global proteomics analysis provided direction for further hypothesis-driven investigation of host pathways altered by RSV infection, specifically the interaction between the RSV NS2 protein and the host ubiquitin system. NS2 primarily acts to antagonize the innate immune system by targeting STAT2 for proteasomal degradation. The goal was to identify NS2 residues important for interaction with the host ubiquitin system, as well as describe the mechanism by which NS2 induces host protein ubiquitination. Bioinformatics analysis provided a platform for development of loss-of-ubiquitin-function NS2 mutants. Combining critical mutations as double or triple NS2 ubiquitin mutants displayed an additive effect on reducing NS2-induced ubiquitination. Recombinant RSV (rRSV) containing NS2 ubiquitin mutations maintained their effect on ubiquitin expression during infection in addition to limiting STAT2 degradation activity. NS2 ubiquitin mutants decreased rRSV growth and increased levels of innate immune responses, indicating a correlation between NS2’s ubiquitin function and antagonism of type I IFN to enhance viral replication. Finally, several proteomics strategies were employed to identify specific cellular proteins ubiquitinated by NS2 to further define host-pathogen interactions during RSV infection. This study demonstrates an effective approach for limiting viral protein function to enhance immune responses during infection.

Mamalian protein ubiquitination and SUMOylation are reversible post translational modifications, which are involved in a multitude of important complex regulatory processes within the cell. Current mass spectrometry approaches that involve bottom-up proteomics to comprehensively analyse these modifications, have proved to be problematic. In this work, analytical approaches are carried out to improve and enhance the comprehensive analysis of these modifications. Tryptic proteolysis of ubiquitinated proteins results in the generation of isopeptides bearing adi-glycine (GG) remnant. Current mass spectrometry approaches used to identify these isopeptides are predominantly reliant on detecting the signature mass shift of the GG remnant (114.043 Da). The lack of sequence information from the GG remnant post MS/MS acquisition results in database search algorithms falsely identifiying these isopeptides. Reductive methylation chemistry was employed to derivatize these isopeptides. Upon collision induced dissociation of the isopeptides two robust ions were released from the iso-N-terminus of the GG remnant ; i) an a1’ ion at m/z 62.09, corresponding to the G of the remnant and ii) a b2’ ion at m/z 147.11, corresponding to the full GG remnant. Post-acquisition data extraction of these unique diagnostic ions demonstrated enhanced selectivity towards identifying these isopeptides. Tryptic proteolysis of SUMOylated proteins results in the generation of isopeptides bearing a substantial iso-C-terminal SUMO remnant. The CRA(K) (Consecutive Residue Addition tolysines (K)) approach combined independant use of proteolytic enzymes and unbiased consecutive residue addition of amino acids pertaining to these iso-C-terminal SUMOremnants, on all lysine residues. This approach enabled the identification of analytically useful novel wildtype isopeptides derived from the proteolysis of SUMO(1/2/3)ylated proteins, bearing GG, TGG and QTGG remnants. The analytically useful isopeptides derived from proteolysis of SUMO(2/3)ylated proteins lacked robust diagnostic information from their iso-C-terminal bearing TGG and QTGG remnants. Reductive methylation chemistry was utilised to derivatize these isopeptides and enabled diagnostic a’ and b’ ions to be released from their iso-N-termini; i) a1’ (m/z 133.13),b2’ (m/z 262.17) and b4’ (m/z 376.22) ions, corresponding to the QTGG remnant and ii) (m/z106.10), b2’ (m/z 191.14) and b3’ (m/z 248.14) ions, corresponding to the TGG remnant. Post-acquisition data extraction of these unique diagnostic ions, enabled comprehensive structural elucidation of these isopeptides and enhanced selectivity towards identification.

We have previously shown that the insulin-like growth factor 1 receptor (IGF-1R) translocates to the cell nucleus, where it binds to enhancer-like regions and increases gene transcription. Further studies have demonstrated that nuclear IGF-1R (nIGF-1R) physically and functionally interacts with some nuclear proteins, i.e. the lymphoid enhancer-binding factor 1 (Lef1), histone H3, and Brahma-related gene-1 proteins. In this study, we identified the proliferating cell nuclear antigen (PCNA) as a nIGF-1R-binding partner. PCNA is a pivotal component of the replication fork machinery and a main regulator of the DNA damage tolerance (DDT) pathway. We found that IGF-1R interacts with and phosphorylates PCNA in human embryonic stem cells and other cell lines. In vitro MS analysis of PCNA co-incubated with the IGF-1R kinase indicated tyrosine residues 60, 133, and 250 in PCNA as IGF-1R targets, and PCNA phosphorylation was followed by mono- and polyubiquitination. Co-immunoprecipitation experiments suggested that these ubiquitination events may be mediated by DDT-dependent E2/E3 ligases (e.g. RAD18 and SHPRH/HLTF). Absence of IGF-1R or mutation of Tyr-60, Tyr-133, or Tyr-250 in PCNA abrogated its ubiquitination. Unlike in cells expressing IGF-1R, externally induced DNA damage in IGF-1R-negative cells caused G(1) cell cycle arrest and S phase fork stalling. Taken together, our results suggest a role of IGF-1R in DDT.

In many vertebrate and invertebrate species mediators of innate immunity include antimicrobial peptides (AMPs) such as peptide fragments of histones and other proteins with previously ascribed different functions. Shark AMPs have not been described and this research examines the antibacterial activity of nurse shark (Ginglymostoma cirratum) peripheral blood leukocyte lysates. Screening of lysates prepared by homogenizing unstimulated peripheral blood leukocytes identified muramidase (lysozyme-like) and non-muramidase antibacterial activity. Lysates were tested for lysozyme using the lysoplate assays, and antibacterial (AB) activity was assayed for by a microdilution growth assay that was developed using Planococcus citreus as the target bacterium. Fractionation of crude lysates by ion exchange and affinity chromatography was followed by a combination of SDS-PAGE with LC/MS-MS and/or N-terminal sequence analysis of low molecular weight protein bands (kDa). This yielded several peptides with amino acid sequence similarity to lysozyme, ubiquitin, hemoglobin, human histones H2A, H2B and H4 and to antibacterial histone fragments of the catfish and the Asian toad. Not all peptide sequences corresponded to peptides potentially antibacterial. The correlation of a specific protein band in active lysate fractions was accomplished by employing the acid-urea gel overlay assays in which AB activity was seen as zones of growth inhibition on a lawn of P. citreus at a position corresponding to that of the putative AB protein band. This study is the first to describe putative AMPs in the shark and their potential role in innate immunity.

Expanded polyglutamine proteins are known to be the causative agents of a number of human neurodegenerative diseases but the molecular basis of their cytoxicity is still poorly understood. Polyglutamine tracts may impede the activity of the proteasome, and evidence from single cell imaging suggests that the sequestration of polyglutamine proteins into inclusion bodies can reduce the proteasomal burden and promote cell survival, at least in the short term. The presence of misfolded protein also leads to activation of stress kinases such as p38MAPK, which can be cytotoxic. The relationships of these systems are not well understood. We have used fluorescent reporter systems imaged in living cells, and stochastic computer modeling to explore the relationships of expanded polyglutamine proteins, p38MAPK activation, generation of reactive oxygen species (ROS), proteasome inhibition, and inclusion body formation. In cells expressing a polyglutamine protein, inclusion body formation was preceded by proteasome inhibition but cytotoxicity was greatly reduced by administration of a p38MAPK inhibitor. Computer simulations suggested that without the generation of ROS, the proteasome inhibition and activation of p38MAPK would have significantly reduced toxicity. Our data suggest a vicious cycle of stress kinase activation and proteasome inhibition that is ultimately lethal to cells. There was close agreement between experimental data and the predictions of a stochastic computer model, supporting a central role for proteasome inhibition and p38MAPK activation in inclusion body formation and ROS-mediated cell death.

Intranuclear rodlets (INRs) are poorly understood intranuclear bodies originally identified within neuronal nuclei on the basis of their unique morphology. Their mechanism of formation, biochemical composition and physiological significance are largely unknown. To gain insight into the molecular regulators of INR formation, mice with a conditional adult β cell-specific knockout of the master regulator of β-cell metabolism, Lkb1 protein kinase (LABKO mice) were studied. The proportion of beta cells containing INRs was significantly reduced in LABKO mice. Further examination ruled out mTOR and Mark2 as downstream effectors of Lkb1 knockout INR phenotype. Instead it identified the mTOR pathway as an independent regulator of INR formation. To investigate INR changes in a pathophysiological context, β cell INRs were examined in two models of human metabolic syndrome: (1) mice maintained on a high-fat diet and (2) leptin-deficient ob/ob mice. Significant INR reduction was observed in both models. Taken together, our results support the view that INR formation in pancreatic β cells is a dynamic and regulated process. The substantial depletion of INRs in LABKO and obese diabetic mice suggests their relationship to β cell function and potential involvement in diabetes pathogenesis. The significance of these findings was further underscored by the demonstration of INRs in human endocrine pancreas, suggesting their potential relevance to the development of metabolic syndrome in humans. The existence of biochemically distinct subtypes of INRs has been suggested by previous reports of differential immunological staining of INRs in neurochemically distinct neuronal populations. Here, a novel variant of INR has been identified that is immunoreactive for the 40kDa huntingtin associated protein and ubiquitin; and evidence was provided for the existence of additional INR subtypes sharing ubiquitin immunoreactivity as a common feature. Selective association of these INRs with melanin concentrating hormone and tyrosine hydroxylase immunoreactive neurons of the hypothalamus and the locus coeruleus was described. It was also demonstrated for the first time that biochemically distinct INR subtypes can co-exist within a single nucleus where they engage in non-random spatial interactions. These findings highlight the biochemical diversity and cell type specific expression of these enigmatic intranuclear structures. On the basis of these findings and previous literature a hypothesis is proposed as to the overall functional significance of INRs in the cell nucleus.

HDAC6 est une histone déacétylase hautement conservée, principalement cytoplasmique. Contrairement à d'autres désacétylases, HDAC6 a une spécificité de substrat unique pour les protéines non - histones . Outre les domaines de désacétylation, HDAC6 contient également un domaine de liaison à l'ubiquitine , qui relie HDAC6 de la voie ubiquitine / protéasome .L’atrophie du muscle squelettique est une condition sévère de perte progressive de masse musculaire au cours de certaines maladies telles le cancer, le diabète, le SIDA ou également immobilizations prolongées. Le contrôle de la masse musculaire est sous la dépendance d’un équilibre entre les processus anaboliques et cataboliques. L’atrophie se caractérise par une augmentation substantielle de la dégradation des protéines par le système ubiquitine-protéasome, causée par l'expression d'une série de gènes spécifiques, les atrogenes . Un des atrogenes induits plus spectaculaire est le muscle spécifique de l'ubiquitine ligase E3 MAFbx/Atrogin-1, qui prend soin de la dégradation de MyoD et de eIF3 -f. La dégradation de ces deux protéines inhibe l'expression de gènes et la traduction myotrophiques empêchant le remplacement de protéines dégradées.Récemment, nous avons identifié l’Histone Deacetylase 6 (HDAC6) comme un nouvel atrogène. L’expression de HDAC6 augmente au cours de l’atrophie musculaire, à la fois chez la souris et l’homme, à travers un mécanisme FOXO3 -dépendante. La déplétion de cet enzyme in vivo (electroporation de l’shRNA contre HDAC6 dans des muscle squelettiques de souris ou analyse de souris invalidées pour ce gène) protège contre l’atrophie. De plus, l’inhibition de HDAC6 après déclenchement de l’atrophie peut aussi atténuer le phénotype. Lors de la caractérisation du mécanisme d’action de HDAC6, nous avons montré que HDAC6 intéragit avec MAFbx et que elle est nécessaire pour l’ubiquitination de MyoD par MAFbx. Nos résultats montrent que la surexpression d’un mutant MyoD resistant à la degradation par MAFbx protège contre l’atrophie provoqué par la denervation.. De plus, certaines données préliminaires indiquent une implication de HDAC6 dans la dégradation de eIF3-f et dans le processus de autophagy dans le tissu musculaire , révélant une double rôle de HDAC6 dans le muscle squelettique .Ces preuves suggèrent que HDAC6 représente potentiellement une cible utile pour des traitements curatifs. / HDAC6 is a highly conserved histone deacetylase, mostly cytoplasmic. Unlike other deacetylases, HDAC6 has unique substrate specificity for non-histone proteins. Besides the deacetylation domains, HDAC6 also contains an ubiquitin-binding domain, which links HDAC6 to the ubiquitin/proteasome pathway. Skeletal muscle atrophy is a severe condition of muscle mass loss occurring during aging or in many clinical disorders as cancer, diabetes and AIDS. The maintenance of muscle mass is subtly controlled by an equilibrium between catabolic and anabolic processes. Muscle atrophy results as a partial suppression of protein synthesis and a substantial increase of protein breakdown by the ubiquitin-proteasome system, caused by the expression of a series of specific genes, the atrogenes. One of the atrogenes induced more dramatically is the muscle specific E3 ubiquitin ligase MAFbx/Atrogin-1, which takes care of the degradation of MyoD and of eIF3-f. Degradation of those two proteins inhibits expression of myotrophic genes and translation preventing the replacement of degraded proteins.We identified HDAC6 as a new atrogene. HDAC6 expression is up regulated during muscle atrophy in mouse and human through a mechanism FoxO3-dependent. In vivo depletion of this enzyme by shRNA electroporation or homologous recombination gives protection against atrophy and its inhibition during atrophy can partially reverse the muscle wasting phenotype. HDAC6 can interact with MAFbx and is required for MAFbx-mediated degradation of MyoD. According to our results, forced expression of a MyoD mutant resistant to HDAC6 and MAFbx dependent degradation prevents muscle wasting induced by denervation. Furthermore, some preliminary data show an involvement of HDAC6 in the degradation of eIF3-f and in the autophagy process in muscle tissue, revealing a double role of HDAC6 in skeletal muscle.These evidences suggest that HDAC6 potentially represents a valuable target for curative treatments.

Pour faciliter l’initiation de la transcription par l’ARN Polymérase II, le complexe co-activateur de la transcription SAGA possède une activité d’acétylation des histones H3 et une activité de déubiquitination des histones H2B, catalysée chez l’homme par l’enzyme USP22. Mon travail de thèse a porté sur l’étude de la régulation de cette activité de déubiquitination.Au sein de SAGA, USP22 interagit fortement avec trois protéines pour former un module structural appelé module de déubiquitination (DUBm). Nous avons montré que la formation d’un tel module était requise pour activer USP22. D’autre part, deux sous-unités du DUBm humain, ATXN7 et ATXN7L3, contiennent un domaine SCA7. Nos résultats montrent que le repliement structural adopté par ces deux doigts de zinc n’avait pas encore été décrit. Nous avons démontré que le domaine SCA7 de ATXN7 peut interagir avec un nucléosome in vitro et que cette interaction participe à la régulation fine de l’activité de déubiquitination de SAGA. Nous proposons qu’en interagissant avec le nucléosome, le domaine SCA7 de Sgf73 ou de ATXN7 pourrait positionner le DUBm de façon optimale par rapport à son substrat. / The SAGA complex is one of the most studied transcriptional co-activator complexes. To facilitate transcription by RNA Polymerase II, SAGA presents a modular organization and harbours two enzymatic activities. In human cells, these two enzymes are called GCN5 and USP22 and they can respectivelly acetylate histones H3 and deubiquitinate histones H2B. During my PhD thesis, I have worked on the regulation of SAGA deubiquitination activity. In the SAGA complex, USP22 interacts strongly with three other subunits to form a structural and functionnal module, named deubiquitination module (DUBm). We have shown that the free recombinant USP22 enzyme is not active, but that the formation of a stable DUBm triggers a strong stimulation of USP22 catalytic activity. Secondly, in human cells, two subunits of the DUBm, ATXN7 and ATXN7L3, contain a domain, called SCA7, that is not found in any other protein. Our results show that the new structural fold adopted by these two domains is specific to these zinc-fingers. These two SCA7 domains share a common structural heart, but their atomic structures reveal also differences, especially in the spatial organization of secondary structure elements. Indeed, we have shown that ATXN7 SCA7 domain can interact in vitro with a nucleosome which is not the case of ATXN7L3 SCA7 domain. Finally, I could show that in vivo the SCA7 domain of Sgf73, the ortholog of ATXN7 has a role in fine tunning SAGA deubiquitination activity. We hypothesize that the interaction between a nucleosome and the SCA7 domain of ATXN7 or Sgf73 would regulate SAGA deubiquitination activity by an optimal positionning of the module to its substrate.

Das Ubiquitin-Proteasom-System (UPS) spielt eine zentrale Rolle bei kardialen Remodeling-Prozessen infolge von Kardiomyopathie und schwerer Herzinsuffizienz. Das UPS ist für die Protein-Homöostase zuständig, indem es fehlerhafte und fehlgefaltete Proteine durch eine gezielte Markierung mit Ubiquitin dem proteasomalen Abbau zuführt. Die Ultima-Ratio-Therapie von Kardiomyopathie und Herzinsuffizienz ist bis heute die Herztransplantation. Aufgrund der geringen Verfügbarkeit von Spenderherzen ist die Suche nach Alternativen von außerordentlicher Wichtigkeit. Um mögliche neue Therapiestrategien für Kardiomyopathie und schwere Herzinsuffizienz zu entwickeln, ist es deshalb von zentraler Bedeutung das UPS im humanen Myokard zu charakterisieren, um die Zusammenhänge zwischen dem UPS und der Herzmechanik genauer zu verstehen und auf diese therapeutisch einwirken zu können. Die vorliegende Arbeit untersuchte erstmalig das UPS bei fortgeschrittener ischämischer (ICM) und dilatativer Kardiomyopathie (DCM), den beiden häufigsten Indikationen für eine Herztransplantation. Vor diesem Hintergrund wurde in der vorliegenden Arbeit die Funktionsweise des UPS anhand spezifischer Marker genauer untersucht. Zu den untersuchten UPS-Komponenten zählen die Aktivität des Proteasoms, die Ubiquitinierung und die E1-E2-E3-Enzymkaskasde, über die eine Ubiquitinierung erfolgt. Für die Untersuchungen wurden Myokardproben von n=24 ICM- und n=28 DCM-Patienten verwendet, welche im Rahmen der Implantation eines linksventrikulären Herzunterstützungssystems gewonnen wurden. Als Kontrolle (n=12) diente Septumgewebe aus Morrow-Resektionen von Patienten mit Aortenstenose jedoch ohne diagnostizierte hypertrophe Kardiomyopathie und ohne Subaortenstenose. Mittels Western-Blot-Analysen wurde die Ubiquitinierung von Proteinen, die Aktivität der E3-Ligasen Muscle RING-finger protein-1 (MuRF-1) und Muscle atrophy F-box (MAFbx) sowie der Ubiquitin-Linker K48 (Proteasomaler Abbau) und K63 (Autophagie-assoziiierte Prozesse) untersucht. Weiterhin wurde die Aktivität des Trypsin- und Chymotrypsin-ähnlichen katalytischen Zentrums des Proteasoms quantifiziert. Die Aktivität der Nicotinamidadenindinucleotidphosphat (NADPH)-Oxidase wurde ermittelt, um Rückschlüsse auf die Entstehung von reaktiven Sauerstoffspezies (ROS) ziehen zu können. Des Weiteren wurde die Ubiquitinierung sowie das Auftreten möglicher apoptotischer Vorgänge durch den Nachweis des Apoptose-induzierenden Faktors (AIF) analysiert. ICM- und DCM-geschädigtes Myokard wies deutliche Unterschiede bezüglich verschiedener Komponenten des UPS auf. In ICM- und DCM-Gewebe konnte eine Abnahme des zellulären, freien Ubiquitins im Vergleich zur Kontrollgruppe nachgewiesen werden (p=0,05). ICM-geschädigtes Myokard wies darüber hinaus einen geringen Anteil von Zellen mit Ubiquitin-positiven Ablagerungen, sogenannten Deposits, auf (p=0,03). Die Multi-Ubiquitinierung von Proteinen war im Myokardgewebe von DCM-Patienten deutlich verringert (p=0,02), die Proteinubiquitinierung über K48 und K63 jedoch zwischen den Gruppen vergleichbar. Eine Quantifzierung der verschiedenen katalytischen Zentren des Proteasoms zeigte nur einen statistischen Trend für eine verringerte Aktivität des Trypsin-ähnlichen katalytischen zentrums in ICM-geschädigtem Myokard. Die Expression der E3-Ligasen MAFbx und MuRF-1 unterschied sich nicht zwischen den Gruppen. Im Vergleich zur Kontrolle wiesen die Myokardproben der ICM-Patienten eine verringerte NADPH-Oxidase-Aktivität auf (p=0,01), was auf eine verringerte ROS-Produktion hinweist. Jedoch unterschied sich der Anteil AIF-positiver Zellen und damit die Caspase-unabhängige Apoptose nicht zwischen den Gruppen. Die vorliegende Arbeit zeigt erstmalig, dass sich die Aktivität und Expression verschiedener Komponenten des UPS in ICM und DCM unterscheiden. Deshalb liefern die vorliegenden Daten verschiedene Ansatzpunkte für die Entwicklung neuer therapeutischer Interventionen für die Herzinsuffizienz, um den Bedarf an Spenderherzen zukünftig senken zu können:Inhaltsverzeichnis 1 Einleitung 1 1.1 Kardiomyopathien 1 1.1.1 Dilatative Kardiomyopathie 4 1.1.2 Ischämische Kardiomyopathie 5 1.2 Kardiales Remodeling 6 1.3 Das Ubiquitin-Proteasom-System 8 1.3.1 Das Proteasom 9 1.3.2 Ubiquitin 10 1.3.3 E1-E2-E3-Enzymkomplex 10 1.3.4 Deubiquitinasen (DUBs) 12 1.4 Therapie 13 2 Zielsetzung 14 3 Material 15 4 Methoden 23 4.1 Probengewinnung 23 4.1.1 Probengewinnung bei LVAD-Implantation 23 4.1.2 Probengewinnung im Rahmen von Morrow-Resektionen 24 4.2 Verarbeitung der Proben 25 4.3 Proteinextraktion und Bestimmung der Proteinkonzentration 26 4.4 Auftrennung von Proteinen durch SDS-Polyacrylamidgelelektrophorese 27 4.5 Western-Blot 28 4.6 Immunhistologische Färbungen 30 4.6.1 Ubiquitin-Färbung 30 4.6.2 Färbung des Apoptose-induzierenden Faktors AIF 31 4.6.3 Hämatoxylin/Eosin-Färbung 31 4.7 Proteasomaktivitätsmessung 32 4.8 NADPH-Oxidase-Aktivitätsmessung 33 4.9 Statistik 34 5 Ergebnisse 35 5.1 Translation von Proteinen 35 5.2 Proteasomaktivität 36 5.3 Expression der E3-Ligasen 38 5.4 Ubiquitinierung 39 5.5 Oxidativer Stress und Apoptose 45 6 Diskussion 49 7 Zusammenfassung 60 8 Literaturverzeichnis 61 9 Abkürzungsverzeichnis 67 10 Abbildungsverzeichnis 69 11 Tabellenverzeichnis 70 Eigenständigkeitserklärung 71 Lebenslauf 72 Danksagung 73

La GTPase Rac1 est une protéine de signalisation intracellulaire qui joue notamment un rôle clé dans la prolifération cellulaire. Notre laboratoire a montré que la toxine CNF1, produite par les Escherichia coli pathogènes, catalyse l’activation de Rac1. Nous avons également identifié le rôle de la E3 ubiquitine-ligase HACE1, un suppresseur de tumeur avéré, dans la régulation par ubiquitylation de Rac1 actif. S’il est prouvé que la forme activée de Rac1 est une cible d’HACE1, le mode d’interaction de ces deux protéines reste à définir ainsi que le rôle de ces interactions dans l’infection et le cancer. L’objectif de mon travail a été de caractériser les interactions moléculaires entre HACE1 et Rac1. Nous avons testé l’hypothèse que des mutations ponctuelles d’HACE1 identifiées dans les cancers pourraient interférer avec son interaction avec Rac1 et sa capacité de contrôle de la croissance cellulaire. J’ai ainsi pu mettre en évidence que 13 mutations somatiques d’HACE1 issues de tumeurs séquencées altèrent sa fonction de contrôle de la croissance cellulaire. De plus, l’étude de ces mutations nous a permis d’identifier un groupe d’acides aminés, situés sur les ankyrin-repeats 5 à 7 d’HACE1, qui contrôle l’interaction d’HACE1 avec Rac1 et de ce fait son ubiquitylation. Enfin dans cette étude nous précisons le rôle du domaine intermédiaire d’HACE1 (MID) dans la spécificité d’interaction de la ligase avec la forme active de Rac1. In fine, la caractérisation de mutants d’interaction entre HACE1 et Rac1 ainsi que l’effet de la toxine CNF1 sur cet axe de signalisation doit nous renseigner sur l’importance de cette voie de régulation dans le cancer et l’infection. / The small GTPase Rac1 plays a key role in various intracellular signaling pathways including cell proliferation. Our laboratory has shown that the CNF1 toxin, produced by pathogenic Escherichia coli, catalyzes the activation of Rac1. We also identified the role of the E3 ubiquitin-ligase HACE1, a tumor suppressor, in the regulation by ubiquitylation of active Rac1. If the activated form of Rac1 is proved to be a target of HACE1, the mode of interaction between these two proteins remains to be define as well as the role of these interactions in infection and cancer. The aim of my work was to characterize the molecular interactions between HACE1 and Rac1. We tested the hypothesis that HACE1 point mutations identified in cancers could interfere with its interaction with Rac1 and its ability to control cell growth. We showed that 13 cancer-associated somatic mutations of HACE1, led to a defective control of cell proliferation. Moreover, the study of these mutations allowed us to identify a group of amino acids, located on the ankyrin-repeats 5 to 7 of HACE1, which controls the interaction of HACE1 with Rac1 and thus its ubiquitylation. We also identified a role for the intermediate domain of HACE1 (MID) in conferring the specificity of association of HACE1 to the active form of Rac1. Ultimately, the characterization of interaction mutants between HACE1 and Rac1 as well as the effect of the CNF1 toxin on this signaling axis will give us more insight on this regulatory pathway in cancer and infection.