Les modifications post-transrationnelles avec les molécules ubiquitine et ubiquitine-like sont des mécanismes essentiels pour la régulation des fonctions protéiques et des voies de signalisations. Les ubiquitine-like molécules comme par exemple NEDD8 jouent un rôle majeur dans la fonction des protéines et sont impliqués dans des maladies humaines comme par exemple les maladies neuro dégénératives. NEDD8 est essentiel pour la survie, la croissance et le développement. C’est pour ceci, l’élucidation de leurs mécanismes d’action est très importante.Dans notre laboratoire, nous nous intéressons à l’ubiquitine-like molécule NEDD8. Le mécanisme de modification avec cette protéine est dit NEDDylation et il nécessite une intervention enzymatique médiées par E1, E2 et E3. Le rôle majeur de NEDD8 est la régulation de l’activité des Cullins, des E3 ubiquitine ligases. Les Cullins sont des scaffolds dont le rôle est de faciliter la dégradation des substrats en assemblant tout le complex enzymatique. La dérégulation de l’activité des Cullins a été montrée de contribuer à l’oncogenèse suite à l’accumulation des oncoprotéines ou suite à la dégradation excessive des suppresseurs de tumeurs. Récemment, la voie de NEDDylation a été ciblée pour des interventions thérapeutiques ; MLN4924 (Provenodistat) (Millenium Pharmaceuticals) bloque la voie de NEDDylation en formant une liaison covalente avec l’enzyme E1 de la voie de NEDDylation. L’identification des mécanismes de régulation des fonctions protéiques par NEDD8 fournira plus d’informations sur le mécanisme d’action de NEDD8 d’un point de vu moléculaire, ce qui permettra l’identification de nouvelles cibles pour de nouvelles interventions thérapeutiques. Dans 50% des tumeurs, le gène p53 est muté. Dans les 50% qui restent, le gène p53 est intact mais la protéine est non fonctionnelles dû à des dérégulations des voies de signalisations. Nous avons trouvé que l’inhibition de la NEDDylation via MLN4924 bloque l’invasion des cellules de mélanomes métastatiques. D’un point de vu moléculaire, MLN4924 provoque la dégradation de Mdmx, l’homologue de Mdm2, tous les deux sont des majeurs régulateurs négatifs de p53. Ceci est dû à la formation d’un complexe entre Mdmx, Mdm2 et Cullin4A inactif. On a démontré que la dégradation de Mdmx est essentielle pour l’activation de la voie RhoA/ROCK capable de bloquer l’invasion tout en modulant le cytosquelette d’actine.De plus, Mdmx a été identifié en tant que régulateur de l’activité de RhoA. Mdmx se lie à la forme inactive de RhoA (GDP-RhoA) et contrôle son activation. Durant les travaux de recherche, nous avons aussi essayé de développer une méthode nous permettant d’utiliser MLN4924 d’une façon spécifique dans le but de minimiser les effets secondaires normalement accompagnant les traitements de chimiothérapies. MLN4924 comme tout agent chimio-thérapeutique, est toxique envers les cellules normales. Dans le but d’utiliser MLN4924 d’une manière spécifique nous permettant de cibler uniquement les cellules cancéreuses sans affecter les cellules normales, nous avons testé cette molécule dans une approche basée sur le statut de p53 appelé « cyclotherapie ».Nos résultats montrent que l’activation de p53 avec des concentrations minimes d’Actinomycine D, induit un arrêt du cycle cellulaire dans la phase G1, protégeant ainsi les cellules de l’effet toxique de MLN4924. Ceci n’est pas le cas pour les cellules cancéreuses dont p53 est déficient ou sans p53. Celles-ci progressent en phase S où elles seront éliminées par apoptose. Cette combinaison a été aussi testée in vivo, plus spécifiquement en utilisant les zebrafish comme model, ce qui nous a permis de confirmer l’effet protecteur de l’Actinomycine D contre l’effet toxique de MLN4924. Nos résultats constituent une nouvelle possibilité de combinaison de MLN4924 basé sur le statut de p53. / Post-translational modifications with ubiquitin and ubiquitin-like molecules (ubls) are essential regulatory mechanisms of protein function and signalling networks. Ubiquitin-like molecules, such as NEDD8 have emerged as major mechanisms of protein function regulation and implicated in human diseases including cancer and neurodegeneration. It is essential for viability, growth and development. Therefore, the elucidation of their mechanism of action is critical. In our lab, research is focused on NEDD8. The process called NEDDylation, involves the three enzymatic activities (E1, E2 and E3). The well-known role of NEDD8 is the regulation of the activity of E3 ubiquitin ligases called Cullin Ring Ligases through modification of Cullins. Cullins are scaffolds to facilitate E3 ligase complex assembly and mediated substrate degradation. Deregulation of Cullins activity has been shown to contribute to oncogenesis through the accumulation of oncoproteins or excessive degradation of tumour suppressors. Recently, the NEDDylation pathway has been targeted for therapeutic intervention; MLN4924 (Pevonedistat) (Millenium Parmaceuticals) forms a covalent adduct with the NEDD8 E1 enzyme and blocks the NEDDylation cascade. Identification of mechanisms of protein function regulation by NEDDylation will provide the molecular basis for the action of the NEDD8 inhibitors in clinic and potentially identify novel targets for therapeutic intervention. In 50% of tumours the TP53 is mutant. In cancer cells where TP53 is WT, upstream or downstream signalling pathways are deregulated. We found that inhibition of NEDDylation by MLN4924 dramatically blocks invasion of metastatic melanoma cells. At the molecular level, MLN4924 induces proteasomal degradation of the Mdmx oncogene product, a key negative regulator of the p53 tumor suppressor. Mechanistically, MLN4924 induces the complex formation of Mdmx with Mdm2 and inactive Cullin4A that promote Mdmx degradation. We found that the degradation of Mdmx is required for the activation of the RhoA/ROCK pathway which blocks cell invasion through changes in the cell cytoskeleton. Interestingly, Mdmx was identified as regulator of RhoA activity through binding with inactive RhoA (GDP- bound RhoA). Another question was addressed which is how to use MLN4924 in a specific manner, thus reducing side effects usually resulting from chemo treatments. MLN4924 as all chemo-agents, is toxic towards healthy cells. In order to use MLN4924 in a specific manner targeting cancer cells specifically leaving healthy cells unharmed, a p53 based cyclotherapy was tested. Our results show that low activation of wild type p53 by low doses of Actinomycin D, causes a G1 cell cycle arrest and protects normal cells from MLN4924 treatment leaving cancerous cells with mutant p53 or no p53 to progress to S-phase and selectively commit apoptosis upon MLN4924 treatment. Our protocol was tested in vivo, in zebrafish model system confirming the protective effect of LDACTD against the cytotoxic effect of MLN4924. Our results provide a possible combination therapy for MLN4924 based on the p53 status.

<b>Background:</b> Aryl Hydrocarbon Receptor Interacting Protein-Like 1 (AIPL1) interacts with NUB1 and restricts the entry of NUB1 protein into the nucleus. The interferon-induced NEDD8 ultimate buster (NUB1) protein causes degradation of neddylated and FAT10ylated proteins through the ubiquitin proteasome system. We observed AIPL1 were frequently down-regulated in various cancers compared to normal tissues. The mechanistic roles of AIPL1 and NUB1 protein in cancer cell cycle regulation remain unexplored. <b>Results:</b> Meta-analysis of cancer databases revealed that expression transcripts of chaperones, including AIPL1, were down-regulated in lung, pancreatic cancer and breast cancer relative to the adjacent normal tissues. Opposite levels of both AIPL1 and NUB1 transcripts were observed in the breast cancer. So it triggers the in vitro experiments using breast cancer cells. METABRIC breast cancer clinical cohort highlighted that patients with low NUB1 transcripts had poor survival in the ER-negative subgroup (but not in ER-positive) of breast cancer patients: hazard ratio (HR)=0.66, 95% confidence interval (CI)=0.5-0.87, p=0.003 and triple negative subgroup of breast cancer patients: HR=0.67, 95% CI=0.47-0.96, p=0.028. NUB1 silencing significantly inhibits in vitro cell growth in MDA-MB-231 and MCF7 under hypoxia. AIPL1 protein forms multimers in cancer cells. NUB1 protein moved into the nucleus in hypoxia (0.1% O2 48hrs) with final confluency at 80-90%. p21 (marker of senescence) &amp; p27 (marker of cell cycle arrest) accumulated in NUB1-silent MDA-MB-231 and RCC4 cells. It suggested that low NUB1 nuclear localisation in hypoxia cause cancer cell cycle arrest. In MDA-MB-231 cell, upon hypoxia, neddylation inhibitor (MLN4924) treated and siNUB1 transfected cells showed decreased CUL1 and further accumulated p21 &amp; p27. The evidence suggested lower neddylated CUL1 and reduced NUB1 cooperatively stabilise p21 and p27 as the substrate of CUL1-ubiquitin ligase. The neddylation inhibitor MLN4924 treated and NUB1 knockdown group exhibited more cells in sub-G1 stage as compared to the control group. In connection to higher p21/p27, it is associated with prolonged arrested cellular aging with depletion. After silencing of NUB1, the increases in cell death of cancer cells upon hypoxia happen through the neddylation-dependent CUL1-p27-p21 and CUL2-VHL axis. We then demonstrated that HIF1&alpha; protein could be both neddylated and FAT10ylated upon reoxygenation. In a tissue microarray study of breast cancer, lower cytoplasmic expression (n=57) had worse overall survival than higher cytoplasmic expression (n=57): HR=1.779, 95% CI=1.006-3.346, p=0.048. <b>Conclusions:</b> AIPL1 and NUB1 proteins exert a role in cell cycle regulation in breast cancer. Low cytoplasmic NUB1 levels are observed in the G₁-S transition of cancer cells. NUB1 depletion causes G₀/G₁ phase arrest due to CUL1 and CUL2 ubiquitin E3 ligase-dependent pathways.

The ubiquitination pathway regulates virtually all cellular processes such as cell cycle control and immune surveillance in eukaryotes, and is thus highly regulated through a variety of means. For instance, the Cullin-RING ubiquitin E3 ligases are regulated by neddylation through the action of a newly identified protein Dcn1. In chapter two, I describe an X-ray crystal structure of yeast Dcn1, encompassing an N-terminal ubiquitin association (UBA) domain and a C-terminal domain of unique architecture, which I termed the PONY (POtentiating NeddYlation) domain. I describe the identification of the reciprocal, conserved binding surfaces on both Dcn1 and yeast cullin Cdc53. In collaboration with Dr. Matthias Peter’s group (ETH Zurich), we show that Dcn1 is necessary and sufficient for cullin neddylation in a purified recombinant system. Together, our data identify Dcn1 as the long sought-after Nedd8 E3 ligase for cullin neddylation. As a modulator of immune surveillance and inflammatory responses, the ubiquitin system serves as an attractive target for subversion by pathogens. In chapter three, I present a structural and functional analysis of a newly identified bacterial ubiquitin E3 ligase IpaH, present in various pathogenic and commensal bacteria. I demonstrate that the leucine-rich repeat (LRR) substrate recognition domains of different IpaH enzymes auto-inhibit the enzymatic activity of the adjacent catalytic domain by two distinct but conserved structural mechanisms. Auto-inhibition is required for the biological activity of two IpaH enzymes in a yeast model system. Retro-engineering of auto-inhibition into a constitutively active IpaH enzyme from Yersinia demonstrates that most of the infrastructure required to support auto-inhibition is evolutionarily conserved. In brief, my research provides insights into the mechanism of action of two newly identified protein ligases in the ubiquitination pathway, namely the Nedd8 E3 ligase Dcn1 and bacterial ubiquitin E3 ligase IpaH.

The ubiquitination pathway regulates virtually all cellular processes such as cell cycle control and immune surveillance in eukaryotes, and is thus highly regulated through a variety of means. For instance, the Cullin-RING ubiquitin E3 ligases are regulated by neddylation through the action of a newly identified protein Dcn1. In chapter two, I describe an X-ray crystal structure of yeast Dcn1, encompassing an N-terminal ubiquitin association (UBA) domain and a C-terminal domain of unique architecture, which I termed the PONY (POtentiating NeddYlation) domain. I describe the identification of the reciprocal, conserved binding surfaces on both Dcn1 and yeast cullin Cdc53. In collaboration with Dr. Matthias Peter’s group (ETH Zurich), we show that Dcn1 is necessary and sufficient for cullin neddylation in a purified recombinant system. Together, our data identify Dcn1 as the long sought-after Nedd8 E3 ligase for cullin neddylation. As a modulator of immune surveillance and inflammatory responses, the ubiquitin system serves as an attractive target for subversion by pathogens. In chapter three, I present a structural and functional analysis of a newly identified bacterial ubiquitin E3 ligase IpaH, present in various pathogenic and commensal bacteria. I demonstrate that the leucine-rich repeat (LRR) substrate recognition domains of different IpaH enzymes auto-inhibit the enzymatic activity of the adjacent catalytic domain by two distinct but conserved structural mechanisms. Auto-inhibition is required for the biological activity of two IpaH enzymes in a yeast model system. Retro-engineering of auto-inhibition into a constitutively active IpaH enzyme from Yersinia demonstrates that most of the infrastructure required to support auto-inhibition is evolutionarily conserved. In brief, my research provides insights into the mechanism of action of two newly identified protein ligases in the ubiquitination pathway, namely the Nedd8 E3 ligase Dcn1 and bacterial ubiquitin E3 ligase IpaH.

NEDD8, molécule de la famille de l’ubiquitine est essentielle au développement, à la croissance et à la viabilité d’un organisme, de plus c’est une cible prometteuse en thérapeutique. Nous avons découvert que l’inhibiteur spécifique de la NEDDylation, MLN4924 altère la morphologie sans fragmentation et augmente la surface du nucléole de cellules humaines et de noyaux de la lignée germinale de Caenorhabditis elegans. Une approche de protéomique quantitative (SILAC) combiné à l’analyse de la production des ARNr et des ribosomes montrent que MLN4924 change la composition protéique du nucléole sans affecter l’activité transcriptionnelle de l’ARN pol I. Notre analyse montre que MLN4924 active p53 par la voie RPL11/RPL5-Mdm2 caractéristique d’un stress du nucléole. Cette étude identifie le nucléole comme une cible intéressante dans l’utilisation d’inhibiteurs de la NEDDylation et apporte un nouveau mécanisme d’activation de p53 par inhibition de la voie NEDD8.Dans une seconde étude nous avons adapté la méthode de FLIM-FRET (« Fluorescence Lifetime Imaging Microscopy – Förster Resonance Energy Transfer ») à l’étude de la compaction de la chromatine à l’échelle du nanomètre dans un organisme vivant. Le nématode Caenorhabditis elegans s’est révélé être un modèle de choix. Au sein des chromosomes méiotiques, nous avons identifié différentes régions de compaction, de niveau variable par mesure du FRET entre histones fusionnées à des protéines fluorescentes. Par une approche originale d’ARN interférence et injection d’un « extra-chromosome » nous avons défini l’architecture à une nano-échelle de différents états de l’hétérochromatine et montré que cette organisation est contrôlée par les protéines HP1 « Heterochromatin Protein 1 » et SETDB1, une protéine « H3-Lysine 9 methyl transferase ». Nous avons également montré que la compaction de l’hétérochromatine est dépendante des condensines I et II et plus particulièrement la condensine I contrôle l’état faiblement compacté de la chromatine.Nos travaux ont confirmé que C. elegans est un modèle d’intérêt majeur pour l’étude des compartiments nucléaires et parfaitement adapté pour des études pré-clinique. / The ubiquitin-like molecule NEDD8 is conserved and essential for viability, growth and development; its activation pathway is a promising target for therapeutic intervention. We found that the small molecule inhibitor of NEDDylation, MLN4924, alters the morphology and increases the surface size of the nucleolus in human cells and Caenorhabditis elegans germ cells in the absence of nucleolar fragmentation. Through SILAC proteomic analysis and rRNA production, processing and ribosome profiling, we show that MLN4924 changes the composition of the nucleolar proteome but does not inhibit RNA Pol I transcription. Further analysis demonstrates that MLN4924 activates the p53 tumour suppressor through the RPL11/RPL5-Mdm2 pathway, with characteristics of nucleolar stress. The study identifies the nucleolus as a target of the NEDDylation pathway and provides a mechanism for p53 activation upon NEDD8 inhibition.Then we adapted a quantitative FRET (Förster resonance energy transfer)-based fluorescence lifetime imaging microscopy (FLIM) approach to assay the nano-scale chromatin compaction in a living organism, the nematode Caenorhabditis elegans. By measuring FRET between histone-tagged fluorescent proteins, we visualized distinct chromosomal regions and quantified the different levels of nanoscale compaction in meiotic cells. Using RNAi and repetitive extrachromosomal array approaches, we defined the heterochromatin state and showed that its architecture presents a nanoscale-compacted organization controlled by Heterochromatin Protein-1 (HP1) and SETDB1 H3-lysine-9 methyl-transferase homologs in vivo. Next, we functionally explored condensin complexes. We found that condensin I and condensin II are essential for heterochromatin compaction and that condensin I additionally controls lowly compacted regions. Our data show that, in living animals, nanoscale chromatin compaction is controlled not only by histone modifiers and readers but also by condensin complexes.We confirm that C. elegans is an interesting model to study nuclear signalling and perfectly adapt to be a platform for pre-clinical studies.

Le but de cette thèse était d’identifier de nouveaux substrats au complexe FANC Core,déficient dans l’Anémie de Fanconi, une pathologie génétique rare. Cette maladie estcaractérisée par un phénotype hétérogène associant une pancytopénie à des malformationscongénitales et une prédisposition accrues aux leucémies myéloïdes aigues.L’anémie de Fanconi est causée par la mutation biallélique dans un des seize gènesFANC. Les protéines produites par ces gènes participent à une même voie moléculaireimpliquée dans la signalisation des dommages de l’ADN. Huit de ces protéines forment lecomplexe FANC Core, une E3 ubiquitine ligase, dont les seuls substrats à ce jour sontFANCD2 et FANCI.Dans le but d’identifier de nouveaux substrats du complexe FANC Core, j’ai réalisé uneanalyse protéomique après immunoprécipitation des peptides modifiés par l’ubiquitine ou parles ubiquitin-like NEDD8 et ISG15. Cette expérience a été faite dans des cellules déficientespour la voie FANC, mutées sur les gène FANCA ou FANCC et comparée à des cellulescorrigées par l’expression de ces gènes.Cette analyse révèle que FANCD2 et FANCI sont les seules cibles du complexe FANCCore en réponse à des dommages de l’ADN.Néanmoins, je montre l’existence d’autres protéines qui sont modifiées d’une manièreFANCA dépendante. Ces protéines sont pour la grande majorité des protéines membranairesou associées aux membranes cytoplasmiques. Parmi celles-ci, j’ai pu déterminer que lerécepteur aux chimiokines, CXCR5, était modifié d’une manière FANCA dépendante parl’ubiquitin-like NEDD8. Cette modification impacte sur la localisation de la protéine à lamembrane et à des conséquences sur la migration des cellules.J’ai aussi montré que FANCA participe d’une manière similaire à la régulation de lalocalisation membranaire d’autres protéines comme APLP2.Ainsi, il est proposé par ce travail un rôle de la protéine FANCA en dehors du complexeFANC Core et en dehors de la réparation des dommages à l’ADN. Comment la protéineFANCA participe à la régulation du trafic des protéines membranaires reste un point nonrésolu à ce jour. / The aim of this thesis was to find new substrates of the E3-ubiquitin ligase activity of theFANC Core complex, mutated in the rare genetic disorder Fanconi Anemia. This disease ischaracterized by bone marrow failure, developmental abnormalities and predisposition tocancer. Eight of the 16 known FANC proteins participate in the FANCcore nuclear complex,which has E3 ubiquitin-ligase activity and monoubiquitinates FANCD2 and FANCI inresponse to replication stress.In this thesis, I used mass spectrometry to compare cellular extracts from FANC Corecomplex deficient FA-A and FA-C cells to their ectopically corrected counterparts after agenotoxic stress.FANCD2 and FANCI appear to be the only true direct targets of the FANCcore complex.However, I also identified other proteins that undergo post-translational modifications throughFANCA- or FANCC-specific direct or indirect mechanisms that are independent of theFANCcore complex. The majority of these potential FANCA or FANCC target proteinslocalize to the cell membrane.Finally, I demonstrated that (a) the chemokine receptor CXCR5 is a neddylated protein; (b)FANCA, surprisingly, appears to modulate CXCR5 neddylation through a currently unknownmechanism; (c) CXCR5 neddylation is involved in the targeting of this receptor to the cellmembrane; and (d) CXCR5 neddylation stimulates cell migration/motility.I also confirmed that the role of FANCA in neddylation is not restrict to CXCR5 but also to,at least, one other protein, APLP2.My work has uncovered a new signaling pathway that is potentially involved in the rarehuman syndrome Fanconi Anemia and in cell motility and has highlighted a potential newfunction for the FANCA protein independant of the FANC Core complex and of a genotoxicstress.

Oncolytic viruses (OVs) have made tremendous strides as a viable cancer therapeutic in recent years; however, variable infectivity rates have since limited clinical efficacy. Residual type 1 interferon (IFN-1) responses are integral to the tumour’s innate antiviral defense and confer resistance to OVs. To combat this, small molecules with viral sensitizing ability can be used in combination to transiently knockdown IFN-1 responses, allowing OVs to gain a foothold for increased infectivity and therapeutic efficacy. Accordingly, we hypothesize that some chemical or genetic manipulations of cellular processes can indirectly antagonize antiviral IFN-1 responses and modulate pro-inflammatory pathways to potentiate oncolytic virotherapy. In this thesis, we identify several avenues to modify cell signalling events to increase OV therapeutic efficacy through IFN-1 inhibition. Firstly, with respect to the demonstrated OV-enhancing effects of vanadium, a pan-phosphatase (PP) inhibitor, we elucidate that its IFN-1 suppressing activity involves activation of the epidermal growth factor receptor (EGFR) pathway via STAT1/2 and NF-κB. Pharmacological inhibition of EGFR abrogated vanadium’s viral sensitizing ability in vivo. Secondly, using high-throughput screening methodology, we identify protein phosphatases that inherently regulate the IFN-1 response as targets for oncolytic vesicular stomatitis virus (VSV∆51) potentiation. Indeed, cloning interfering RNA against one of these PP targets, acid phosphatase 2 (ACP2), into the VSV∆51 platform demonstrated superior infectivity and cancer cell cytotoxicity compared to the non-targeting VSV∆51 control. Thirdly, we characterize pevonedistat, a first in-class neddylation activating enzyme inhibitor, to potentiate OV therapeutic efficacy across several in vitro and in vivo contexts. We demonstrate pevonedistat’s ability to inhibit IFN-1 signalling and pro-inflammatory cytokine production using both neddylation independent and dependent mechanisms. Taken altogether, we dissect multiple signaling mechanisms by which the IFN-1 response can be modulated for the purposes of improving OV therapeutic efficacy. This knowledge can subsequently be directly translated into designing optimized OV strategies for clinical testing.

E2F-1 has a central role in cell cycle orchestration, and its activity is tightly regulated. One of the ways E2F-1 activity is controlled is by direct modification by post translational modifications such as acetylation, ubiquitination and phosphorylation. Here it was demonstrated that E2F-1 is targeted by two novel modifications, namely methylation by Set7/9 and NEDDylation, both within the DNA binding and heterodimerisation domain of the protein. NEDDylation and methylation of E2F-1 both decrease the stability and diminish the transcriptional activity of E2F-1. Lysine residues in E2F-1 involved in NEDDylation are also targeted by methylation, allowing the potential for interplay between these modifications. Methylation of E2F-1 was demonstrated to be a prerequisite for its NEDDylation and the multi-domain protein UHRF1 implicated in mediating this effect. The results define a new level of control on E2F-1 and suggest a protein code with pleiotropic effects involved in E2F-1 regulation.