Advances in protein topology mapping methods are urgently needed to complement the wealth of interactome data that is presently being generated at a rapid pace. Chemical crosslinking followed by mass spectrometry (MS) has evolved over the last decade as an attractive method for protein topology and interface mapping, and holds great promise as a counterpart to modern interactome studies in the field of proteomics. Furthermore, stabilization of proteins and protein complexes with crosslinking offers many advantages over high-resolution structural mapping methods, including the ability to study protein topologies in vivo. The reliance on direct detection of crosslinked peptides, however, continues to pose challenges to protein topology and interface mapping with chemical crosslinking plus MS. The present body of work aimed to develop a novel generic methodology that utilizes chemical crosslinking, cyanogen bromide (CNBr) cleavage and MS for the low-resolution mapping of protein topologies and interfaces. Through such low-resolution mapping of crosslinked regions, this novel strategy overcomes limitations associated with the direct detection of crosslinked peptides. Following optimization of various steps, the present method was validated with the bacterial DNA-directed RNA polymerase core complex and was subsequently applied to probe the tetrameric assembly of yeast Skp1p-Cdc4p heterodimers. Further improvements were made through the enrichment of crosslinked CNBr-cleaved protein fragments prior to their identification via MS. Two enrichment strategies were developed which depended upon the conjugation of tags to CNBr-cleaved peptide C-termini followed by either tandem affinity purification or tandem reversed-phase HPLC purification. These strategies were successfully applied for the efficient purification of disulfide-linked peptides from peptide mixtures. It is expected that the potential to achieve sensitive mapping of topologies and interfaces of multi-subunit protein complexes in vivo, in combination with further enhancements to permit studies on complex protein samples, will extend the utility of this method to complement large-scale interactome studies.

La variabilité non maîtrisée de la tendreté de la viande bovine est un problème majeur pour la filière industrielle, cette qualité étant très recherchée par les consommateurs. Depuis de nombreuses années, différents programmes de recherche ont identifié des marqueurs ADN, ARN et protéines de la tendreté de la viande bovine, dans des contextes différents. Aux Etats-Unis et en Australie, des recherches ont abouti à la conception de tests efficaces de prédiction de la qualité de la viande. Ces tests ne fonctionnent cependant pas sur les élevages français. La filière française est donc demandeuse de tests de prédiction simples permettant de mesurer la tendreté sur l'animal vivant et la carcasse. De tels tests sont inexistants à l'heure actuelle. L'objectif de la thèse est de valider ou non une liste de potentiels marqueurs protéiques de la tendreté, et de développer des équations de prédiction de cette qualité, afin d'envisager la conception de tests immunologiques de prédiction à destination de la filière sur l'animal vivant et la carcasse, dans un avenir proche. Une nouvelle technique pour quantifier les protéines d'un échantillon de muscle bovin a été développée, le Dot-Blot, et permet de disposer de prototype pour de futurs tests de phénotypage de la tendreté. L'utilisation de cette technique a permis de quantifier 24 protéines sur 111 échantillons de deux muscles de boeufs et de taurillons de race Charolaise. Ces travaux ont identifié en premier lieux des effets biologiques liés au type de muscle et d'animal sur l'abondance des protéines. Puis, trois analyses différentes, dont les résultats sont concordants, ont validé des marqueurs de tendreté et mis en lumière les principaux mécanismes cellulaires impliqués dans la tendreté, générant des équations de prédiction de la tendreté. Des outils de bioinformatique ont été construits à partir de données expérimentales de la thèse sur les protéines de la tendreté et de bases de données humaines, afin de mieux comprendre les mécanismes biologiques impliqués dans la mise en place de la tendreté. En conclusion, ce travail de thèse a développé un nouvel outil d'analyse et les premières équations de prédiction de la tendreté de la viande bovine fiables sur un système centré sur les mâles, supports indispensables au développement de tests de prédiction de la tendreté sur l'animal vivant et la carcasse. De plus, ce travail a permis de mieux comprendre et caractériser les mécanismes moléculaires impliqués dans la mise en place de la tendreté.

Protein complexes are critical to virtually every biological process performed by living organisms. The cellular “interactome”, or set of physical protein-protein interactions, is of particular interest, but no comprehensive study of human multi-protein complexes has yet been reported. In this Thesis, I describe the development of a novel high-throughput profiling method, which I term Fractionomic Profiling-Mass Spectrometry (or FP-MS), in which biochemical fractionation using non-denaturing high performance liquid chromatography (HPLC), as an alternative to affinity purification (e.g. TAP tagging) or immuno-precipitation, is coupled with tandem mass spectrometry-based protein identification for the global detection of stably-associated protein complexes in mammalian cells or tissues. Using a cell culture model system, I document proof-of-principle experiments confirming the suitability of this method for monitoring large numbers of soluble, stable protein complexes from either crude protein extracts or enriched sub-cellular compartments. Next, I document how, using orthogonal functional genomics information generated in collaboration with computational biology groups as filters, we applied FP-MS co-fractionation profiling to construct a high-quality map of 622 predicted unique soluble human protein complexes that could be biochemically enriched from HeLa and HEK293 nuclear and cytoplasmic extracts. Our network is enriched in assemblies consisting of human disease-linked proteins and contains hundreds of putative new components and novel complexes, many of which are broadly evolutionarily conserved. This study revealed unexpected biological associations, such as the GNL3, FTSJ3, and MKI67IP factors involved in 60S ribosome assembly. It is my expectation that this first systematic, experimentally-derived atlas of putative human protein complexes will constitute a starting point for more in depth, hypothesis-driven functional investigations of basic human molecular and cellular biology. I also note that my generic FP-MS screening approach can, and is currently, being applied by other members of the Emili laboratory to examine the global interactomes of other mammalian cell lines, tissues, sub-cellular compartments, and diverse model organisms, which should expand our understanding of proteome adaptations and association networks associated with cell physiology, animal development and molecular evolution.

Made available in DSpace on 2015-03-26T13:42:30Z (GMT). No. of bitstreams: 1 texto completo.pdf: 2281509 bytes, checksum: 8e44475ccbcd12f91301ea5e4fda8830 (MD5) Previous issue date: 2013-03-07 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Plants are engaged in a continuous co-evolutionary struggle for dominance with their pathogens and the outcomes of these interactions are of particular importance to human activities, as they can have dramatic effects on agricultural systems (Dodds & Rathjen, 2010). Recently, the convergence of molecular studies of plant immunity and pathogen infection strategies is revealing an integrated picture of the plant pathogen interaction (Mukhtar et al., 2011) in which the pathogen effectors interaction converge onto highly connected subgroups of proteins, named hubs. A well-defined hub form plant immune system network corresponds to CSN5A protein, a catalytic subunit of the COP9 signalosome acting as a key regulator in several basic cellular processes. Consistent with the prediction that different effectors from different pathogens target similar connections in plant-pathogen interaction network, it has been shown, independently, that the protein C2 from geminivirus, a DNA virus that infects a wide variety of agronomic crops, interacts to CNS5A (Lozano- Duran et al, 2011). Additionally, it was shown that NIG and the immune receptor NIK, both targets of geminivírus NSP, interact to CSN5A (Machado, 2011; Mukhtar et al., 2011). Based on this information, it is expected that the hub CNS5A is a functional element in the geminivirus-host interaction network. Recently, it was reported that NIG, a cellular partner of CSN5A, also interacts with a unknown function protein, encoded by the locus AT2G41020 in yeast (Machado, 2011). As a possible component from geminivirus-host interaction network converging to CSN5A, AT2G41020 may interact directly or indirectly with virulence factors in defense response or compatibility. The objectives of this research involved biochemical characterization of the protein encoded by the locus AT2G4102, designated AtWWP1, and identification of its possible interactions with viral proteins and host factors. In silico analysis of tWWP1 predicted structure revealed the presence of two WW domains, and a C-terminal domain highly conserved between homologous in plant and animals. Furthermore, it has been shown that AtWWP1 is a nuclear protein capable of forming nuclear bodies via the conserved C-terminal domain. Coimmunoprecipitation and BiFC assays demonstrated that AtWWP1 interacts in vivo with the cytoplasmic protein NIG, redirecting it to nuclear bodies. In order to explore the formative activity of nuclear bodies AtWWP1, the interaction between AtWWP1 and a second protein partner AtMBD2 (methyl CG binding domaincontaining protein) was characterized in vivo. The ability to form nuclear bodies as interaction with AtMBD2 were mapped AtWWP1 occuring via its domain and C-terminal conserved, substantiating the argument that this region of AtWWP1 is responsible for the formation of nuclear bodies. Colocalization assays have shown that nuclear bodies contained in AtWWP1 are distinct from those formed by proteins involved in RNA splicing, but colocalized with nuclear bodies containing CDKC2. Furthermore, it was demonstrated that AtWWP1 does not bind to RNA, but exhibits a binding activity to DNA. These characteristics imply that AtWWP1 should be involved with basic nuclear functions. As a component of a functional hub in geminivirus-host interaction network, it is important to assess whether the viral infection would affect the nuclear bodies formed by AtWWP1. / Estudos moleculares envolvendo o sistema imune de plantas e a infecção por patógenos revelaram um panomara integrado de interações plantapatógeno em que as interações dos efetores de virulência convergem para subconjuntos de proteínas do hospedeiro altamente interconectadas e designadas hubs. Um hub funcional e bem definido do sistema imune de plantas corresponde a interconexões convergentes para a proteína CNS5A que constitui a subunidade catalítica do complexo COP9 signalosome, um regulador chave de diversos processos celulares básicos. Consistente com a previsão de que efetores de diferentes patógenos devem alvejar similares conexões na rede de interações planta-patógeno, foi demonstrado, independentemente, que a proteína C2 de geminivírus, um vírus de DNA que infecta uma grande variedade de culturas agronômicas, interage com a proteína CNS5A. Além disso, foi também demonstrado que tanto a proteína NIG, quanto o receptor imune NIK, ambos alvos da proteína NSP de geminivírus, também interagem com CNS5A. Baseado nestas informações, prevê-se que a interconexão (hub) representada por CNS5A seja um elemento funcional na interação geminivírus-hospedeiro. Recentemente, foi identificado que, além de interagir com CSN5A, a proteína NIG também interage com uma proteína de função desconhecida codificada pelo locus AT2G41020, em leveduras . Como possível componente da rede de interações geminivírus-hospedeiro que converge em CNS5A, AT2G41020 pode interagir direta ou indiretamente com fatores de virulência em resposta de defesa ou de compatibilidade. Sendo assim, os objetivos principais dessa investigação envolveram caracterização bioquímica da proteína codificada pelo locus AT2G41020 e identificação de possíveis interações com proteínas virais e fatores do hospedeiro. Análise in silico da estrutura predita da proteína codificada pelo lócus At2G41020, designada AtWWP1, revelou a presença de dois domínios WW e um domínio C-terminal altamente conservado entre proteínas homólogas de espécies vegetais e animais. Além disso, foi demonstrado que a proteína AtWWP1 é uma proteína nuclear capaz de formar corpos subnucleares via o domínio C-terminal conservado. Ensaios de coimunoprecipitação e BiFC demonstraram que AtWWP1 interage in vivo com a proteína citoplasmática NIG promovendo o seu redirecionamento para corpos nucleares. Com a finalidade de explorar a atividade formadora de corpos nucleares de AtWWP1, a interação entre AtWWP1 e uma segunda proteína parceira AtMBD2 (proteína contendo um domínio de interação com CG metilado) foi caracterizada in vivo. Tanto a capacidade de formar corpos nucleares quanto a interação com AtMBD2 foram mapeadas em AtWWP1 e ocorrem via seu domíno C-terminal conservado, substanciando o argumento de que esta região de AtWWP1 é responsável pela formação de corpos subnucleares. Ensaios de co-localização demonstraram que os corpos nucleares contidos em AtWWP1 são distintos daqueles formados por proteínas envolvidas em splicing do RNA; porém co-localizam com corpos nucleares contendo CDKC2. Além disso, foi demonstrado que AtWWP1 não liga a RNA, mas exibe uma atividade de ligação ao DNA. Estas características implicam que AtWWP1 deve estar envolvida com funções nucleares básicas. Como componente de um hub funcional na interação geminivírus-hospedeiro, torna-se relevante avaliar se a infecção viral afetaria os corpos nucleares formados por AtWWP1.

Orientador: Jörg Kobarg / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Biologia / Made available in DSpace on 2018-08-25T14:15:32Z (GMT). No. of bitstreams: 1 Basei_FernandaLuisa_D.pdf: 43645148 bytes, checksum: 4cb7da11417bc57aea32c2db81dddc18 (MD5) Previous issue date: 2014 / Resumo: As Neks são proteínas quinases similares a NIMA, proteína que é indispensável para a entrada em mitose de células de Aspergillus nidulans. Em humanos foram identificadas 11 Neks (1-11) e, estudos crescentes vêm demonstrando a participação destas em diversas funções celulares além do controle do ciclo e divisão celular. A Nek4 é um dos maiores membros dessa família e sua relação com a manutenção ciliar e resposta ao DNA danificado já foi demonstrada. Contudo, seus parceiros de interação e substratos são ainda desconhecidos. Para melhor compreender o papel da Nek4 foi realizado um estudo de interatoma para identificar novos processos biológicos com os quais a Nek4 está envolvida. Inicialmente foi identificada uma nova isoforma para a Nek4 e assim, realizou-se o estudo de interatoma para as duas isoformas com a finalidade de comparar o perfil de interação das duas proteínas. As duas isoformas da Nek4 foram expressas em células humanas e após imunoprecipitação seguida de identificação por espectrometria de massas, foram identificadas 474 proteínas que interagem com a isoforma 1 da Nek4, Nek4.1 e 149 para a isoforma 2, Nek4.2. Dentre as proteínas identificadas, 102 interagem com ambas isoformas da Nek4. Nossos resultados confirmam o envolvimento da Nek4 com a resposta ao DNA danificado, função ciliar, estabilização dos microtúbulos e ainda sugerem o envolvimento da Nek4 em funções completamente novas, como processamento de RNAm, resposta ao estresse, controle de qualidade das proteínas e apoptose. Entre os parceiros de interação encontramos importantes proteínas como TRAP-1, Whirlin, PCNA, 14-3-3?, Btf, PARP-1, SRSF1, PAI-RBP1 e KAP-1. As duas isoformas compartilham funções que não foram ainda descritas para os membros da família Nek e a isoforma 1 ainda apresenta funções que já foram descritas para outros membros da família. Aliado ao resultado da imunoprecipitação ainda foram realizadas imunofluorescências que permitiram verificar a localização da Nek4 em diferentes estruturas celulares, como os speckles nucleares e a mitocôndria, corroborando com a função no processamento de RNAm e apoptose. O experimento de imunoprecipitação seguido de identificação por espectrometria de massas também apontou para a possibilidade de autofosforilação e dimerização da Nek4. Além disso, foi possível observar diferenças entre o perfil de interação das duas isoformas da Nek4, sendo que a isoforma 1 interage com proteínas que mantém funções biológicas similares a outras Neks, que a isoforma 2 não apresenta / Abstract: Neks are serine-threonine kinases that are similar to NIMA, a protein found in Aspergillus nidulans which is essential for cell division. In humans there are eleven Neks (1-11) which are involved in different biological functions besides the cell cycle control. Nek4 is one of largest members of the Neks family and has been related to the primary cilia formation and in DNA damage response. However, its substrates and interaction partners are still unknown. Thus in an attempt to better understand the role of Nek4 we performed an interactomics study to find new biological processes in which Nek4 is involved. Besides, we described here a novel Nek4 isoform and compared the interactomics profile of these two Nek4 proteins. Isoform 1 and isoform 2 of Nek4 were expressed in human cells and after an immunoprecipitation followed by mass spectrometry, 474 interacting proteins were identified for isoform 1 and 149 for isoform 2 of Nek4. 102 proteins are common interactors between both isoforms. Our results confirm Nek4 involvement in the DNA damage response, cilia maintenance and microtubules stabilization, and raise the possibility of new functional contexts including mRNA processing, apoptosis signaling, stress response, translation and protein quality control. Among the interaction partners, we found important proteins such as TRAP-1, Whirlin, PCNA, 14-3-3?, Btf, PARP-1, SRSF1, PAI-RBP1 and KAP-1. We could observe that both isoforms share functions that are new to the Nek family, and isoform 1 apparently has also maintained functions which have already been established to other Nek family members. From our immunoprecipitation followed by mass spectrometry experiment a possible site for Nek4 autophosphorylation and dimerization was identified. This study provides new insights into Nek4 functions, identifying new interaction partners, localization to new cellular compartment and further suggests an interesting difference between isoform 1 and the novel isoform 2 of Nek4. Nek4 isoform 1 may have maintained similar roles compared to other Neks and these roles are not related to isoform 2 / Doutorado / Bioquimica / Doutora em Biologia Funcional e Molecular

Orientador: Kleber Gomes Franchini / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Biologia / Made available in DSpace on 2018-08-25T14:22:48Z (GMT). No. of bitstreams: 1 Cordeiro_IsabelleBezerra_D.pdf: 17466566 bytes, checksum: de71f83d2b4d3c8e0f0269f5d61d332d (MD5) Previous issue date: 2014 / Resumo: A proteína Focal Adhesion Kinase (FAK; PTK2) participa de vários processos celulares. A identificação das proteínas parceiras de FAK tem contribuído para o entendimento de suas múltiplas funções celulares. Utilizando um sistema de indução de FAK fusionada a uma cauda de FLAG, combinada com análises por espectrometria de massas (MS), identificamos proteínas associadas à FAK em células HEK293. Um total de 153 proteínas foram repetidamente identificadas com alta resolução por experimentos de MS. Além de confirmar interações já previamente descritas como Paxillin (PXN), Heat shock protein Hsp90 (HSP90) e Transforming growth factor beta-1-induced transcript 1 protein (TGF?1I1), as análises por MS revelaram um novo conjunto de proteínas associadas à FAK, incluindo proteínas envolvidas nas vias de síntese de proteínas, expressão gênica, crescimento celular, proliferação, morte e sobrevivência celular. Além disso, análises de bioinformática das vias indicaram que a FAK se associa com um conjunto de 30 proteínas envolvidas com a via de modificação pós-transcricional do RNA, incluindo a proteína Serine/arginine-rich splicing factor 2 (SC-35; SRSF2), que é um marcador de speckles nucleares. Esse conjunto de proteínas está associado ao spliceossomo e um conjunto similar de proteínas também foi identificado com os experimentos que utilizaram somente o domínio FERM da FAK. Validações por Western Blotting e imunocitoquímica demonstraram que FAK se associa e co-localiza com a proteína SC-35 no núcleo celular. Ainda identificamos um papel funcional da FAK no splicing do mRNA. Demonstramos que FAK pode modificar o padrão de splicing de um gene repórter E1A ao ser superexpressa em células HEK. Nosso trabalho propõe novas funções da FAK no núcleo, indicando que esta pode estar envolvida em eventos relacionados à função do RNA, como a regulação do splicing do mRNA / Abstract: Focal adhesion kinase (FAK; PTK2) has roles in many cellular processes. The identification of protein partners for FAK has greatly contributed to our understanding of its multiple function. Using inducible FLAG-tagged FAK combined with affinity/purification mass spectrometry (MS) approach, we identified proteins associated with FAK in HEK293 cells. A total of 153 proteins were repeatedly detected in high-resolution MS measurements. Beyond the well characterized partnering with Paxillin (PXN), Heat Shock protein 90 (HSP90) and Transforming growth factor beta-1-induced transcript 1 protein (TGFB1I1), analysis of MS data uncovered novel sets of proteins that associate with FAK, playing a role in protein synthesis, gene expression, cellular growth, proliferation, death and survival. In addition, the network analysis established the unexpected finding that a module of 30 proteins linked to RNA post-transcriptional modifications are recruited by FAK, including Serine/arginine-rich splicing factor 2 (SC-35; SRSF2), which is a marker of the nuclear speckles. Indeed, this module is found to be enriched by proteins associated with spliceosome. Remarkably, a similar set of proteins was also recruited by FAK N-terminal FERM domain. Biochemical and imunofluorescence validations established that FAK associates and co-localizes with SC-35 protein at the cell nuclei. We further pinpoint a functional role of FAK in mRNA splicing. We showed that FAK can modify the splicing site selection of the adenoviral E1A minigene in a dose-dependent manner. Our work provides new insights into the molecular function of FAK in the nucleus, indicating that it may be involved in events related to RNA function, such as pre-mRNA splicing / Doutorado / Bioquimica / Doutora em Biologia Funcional e Molecular

Le cancer du poumon est la première cause de mortalité liée au cancer en France et dans le monde. Il s’agit d’un cancer de mauvais pronostique, diagnostiqué généralement aux stades tardifs III ou IV et avec une survie à 5 ans faible, respectivement de 6 et 1%. Ce cancer comprend différents types histologiques, parmi lesquels les adénocarcinomes sont les plus fréquents et comptent pour 40% des diagnostics. L’analyse génétique par séquençage des tumeurs des patients en stade IV permet de mettre en évidence des mutations récurrentes, et généralement mutuellement exclusives, au niveau des gènes KRAS, EGFR, BRAF et ALK. L’identification de ces mutations permet dans le cadre de tests compagnons de décider de l’éligibilité du patient aux thérapies ciblées lorsqu’elles sont disponibles.Bien que ces thérapies ciblées représentent une véritable révolution dans la prise en charge des patients, une majorité de patients ne peuvent bénéficier de ces traitements car ils ne présentent pas au niveau tumoral de mutation activatrices actionnables (absence de mutation EGFR, BRAF et ALK, présence d’une mutation KRAS…). De plus, dans l’éventualité où ils peuvent être traités par une molécule orale, le bénéfice observé reste malheureusement de courte durée et tous finiront à terme par échapper au traitement, c’est le cas par exemple lors de mutations de l’EGFR.Ainsi, afin de mieux comprendre quels sont les mécanismes moléculaires associés aux phénomènes de résistances thérapeutiques observés en clinique, et dans l’objectif de proposer de nouvelles voies thérapeutiques pour les patients non éligibles aux thérapies ciblées, j’ai étudié, au niveau protéique, l’impact des mutations sur les réseaux d’interactions protéiques en utilisant la technique du BioID (proximity-dependent biotinylation identification). Plus particulièrement, je me suis intéressé aux mutations activatrices de l’EGFR, KRAS, BRAF et ALK. Les protéines de la famille RAS (HRAS, NRAS, KRAS) étant mutées dans plus de 30% des cancers, je me suis également intéressé aux réseaux d’interactions protéiques portés par ces trois isoformes afin de mettre en évidence des interactions protéiques spécifiques à l’isoforme KRAS.Au cours de ma thèse, j’ai montré que les réseaux d’interaction caractérisés par le BioID sont bien plus denses que ceux obtenus par la technique plus conventionnelle de purification par affinité, et qu’ils permettent de mettre en évidence des interacteurs spécifiquement dérégulés par l’apparition de mutations activatrices. Par ailleurs les modèles cellulaires HEK293 (cellules humaines embryonnaires de rein) et BEAS2B (cellules pulmonaires d’origine non-cancéreuses) ont montré un fort recouvrement des interacteurs identifiés, ce qui suggère que la stratégie employée est pertinente pour identifier des interacteurs spécifiques de mutations. Ce travail de thèse a permis d’identifier une série de plusieurs interacteurs spécifiques des formes mutante de KRAS, d’EGFR, de BRAF, de NRAS et de la fusion EML4-ALK. Treize interacteurs spécifiques de la forme mutée de KRAS ont été validés fonctionnellement dans des modèles cellulaires de cancers du poumon. Enfin, en utilisant les données de BioID, un modèle préliminaire de résistance de l’EGFR aux thérapies ciblées a pu être élaboré, lequel fait apparaitre les protéines CBL et IGF2R comme des protéines potentiellement impliquées dans l’acquisition des résistances aux thérapies ciblées.Ce travail de thèse propose ainsi de nouvelles perspectives quant à la détermination des mécanismes de résistance spécifiques aux thérapies ciblées et à l’identification de nouvelles cibles thérapeutiques chez les patients présentent des mutations activatrices. / Lung cancer is the leading cause of cancer-related death in France and in the world. It is a cancer of poor prognosis, diagnosed at the late stage III or IV, with a 5-years survival of 6% and 1%, respectively. This cancer encompass several histological types, and among them adenocarcinoma account for 40% of the diagnosis. Genetic sequencing of stage IV tumors highlights redundant mutations, and generally exclusives from each other’s, of KRAS, EGFR, BRAF and ALK genes. The identification of these mutations enable, within companion test, to make eligible patients for targeted therapies when molecules are available.Even though these targeted therapies represent a true revolution in patient’s care, the majority of them cannot benefit from these treatments because their tumors do not harbor activating mutations that are targetable (e.g. absence of EGFR, BRAF and ALK mutation, presence of KRAS mutation). Additionally, when they can be treated using an oral molecule, the benefit observed is unfortunately poor in terms of period of time, and all the patients will escape from the treatment. This is for example the case with EGFR mutations.To better understand the molecular mechanisms associated with the resistance events observed in the clinic, and to propose new therapeutics for patient not-eligible for targeted therapies, I studied at the proteome level, the impact these mutations on protein networks, using the BioID technology (proximity-dependent biotinylation identification). More particularly, I have been interested in the activating mutation of EGFR, KRAS, BRAF and ALK. Considering that proteins from the RAS family (HRAS, NRAS and KRAS) are mutated in around 30% of cancers, I have been also interested in the protein network of these proteins to highlight interaction specific to the KRAS isoform.During my thesis, I showed that the protein networks characterized using BioID are much more dense compared to those identified with the more conventional technic of AP-MS (Affinity-purification and mass spectrometry), and that they enable to identify interactors specifically deregulated upon activating mutation. Additionally, the HEK293 cell model (Human Embryonic Kidney) and BEAS2B cell model (non-cancerous lung cell line) showed a high overlapping degree of the interactors identified, suggesting that the strategy used is relevant to identify interactors specific to mutations. This thesis enabled to identify several interactors specific to the mutant KRAS, EGFR, BRAF, NRAS and EML4-ALk fusion. Thirteen interactors specific to the mutated-KRAS have been functionally validated in lung-cancer cell lines models. Finally, using BioID data I have been able to propose a model of EGFR resistance to targeted therapies. This model shows that CBL and IGH2R might be the EGFR partners responsible for therapeutic escape.Altogether, this thesis propose new perspective to determine resistance mechanisms and to identify new therapeutic targets for KRAS-mutated patients.

Le noyau est une innovation évolutive majeure caractéristique des organismes eucaryotes. Ces dernières années de nombreux travaux se sont intéressés à l’organisation de la chromatine dans l’espace nucléaire lors de l’interphase. Les protéines associées à la périphérie nucléaire ou ancrées dans la membrane nucléaire interne ont suscité un intérêt majeur due à leur contribution dans l’organisation spatiale de la chromatine. Chez les animaux, les lamines qui forment des filaments à la périphérie nucléaire et le complexe LINC, un complexe protéique reliant la membrane externe et interne du noyau sont connues pour interagir avec la chromatine, influencer l’organisation de cette dernière et moduler la régulation transcriptionnelle. Chez la plante modèle Arabidopsis thaliana utilisée dans ce travail, le complexe LINC est conservé, par contre les lamines ne le sont pas et seraient remplacées par d’autres acteurs spécifiques du règne végétal. Le travail détaillé dans ce manuscrit porte sur la mise en évidence d’un nouveau réseau d’interaction protéique localisé à la périphérie nucléaire et sur l’impact de ces protéines dans la morphologie du noyau et l’organisation de la chromatine. Mes travaux se sont concentrés sur les protéines à domaine SUN, l’une des composantes du complexe LINC et sur les protéines CRWN et KAKU4 présentes à la périphérie du noyau. Des cribles double hybride chez la levure m’ont permis d’identifier 24 partenaires protéiques potentiels dont plus d’un tiers sont des facteurs de transcription L’étude plus précise du facteur de transcription MaMYB pour lequel nous avons créé un allèle nul par la méthode CRISPR montre qu’il joue un rôle plus spécifique dans la formation des racines. L’étude de mutants combinatoires pour les gènes SUN, CRWN et KAKU4 montre des anomalies développementales notamment des tissus reproductifs. Enfin, une étude plus détaillée de la protéine KAKU4 suggère sa participation au maintien de la morphologie du noyau et au rapprochement de l’hétérochromatine vers la périphérie nucléaire. En résumé, mes travaux ont mis en évidence l’existence d’un réseau de facteurs de transcription recrutés à la périphérie nucléaire par les protéines SUN, CRWN et KAKU4. Ce réseau d’interaction protéine-protéine participerait à un mécanisme de séquestration de certains facteurs de transcription et/ou d'un rapprochement à la périphérie nucléaire de certains domaines de chromatine afin d’activer ou de réprimer leur transcription. / The nucleus is a major evolutionary innovation characteristic of eukaryotic organisms. In recent years, numerous studies have focused on the organization of chromatin in nuclear space during interphase. Proteins associated with the nuclear periphery or anchored in the inner nuclear membrane have been particularly studied for their contribution to the spatial organization of chromatin. In animals, the lamina that forms filaments at the nuclear periphery and the LINC complex, a protein complex linking the outer and inner membrane of the nucleus, are known to interact with chromatin, to influence its organization and to modulate transcriptional regulation. In the model plant Arabidopsis thaliana used in this work, the LINC complex is conserved, but not the lamina constituents, which are replaced by other specific actors of the plant kingdom. The work detailed in this manuscript identified a new protein interaction network located on the nuclear periphery and studied the impact of these proteins on nuclear morphology and chromatin organization. My work focused on SUN-domain proteins, one of the components of the LINC complex, and on the CRWN and KAKU4 proteins at the periphery of the nucleus. Double hybrid screens in yeast allowed me to identify 24 potential protein partners, more than a third of which are transcription factors. The more precise study of the transcription factor MaMYB for which we created a null allele using the CRISPR method, shows that it plays a more specific role in root formation. The study of mutant combinations for SUN, CRWN and KAKU4 genes reveals developmental abnormalities, particularly in reproductive tissue. Finally, a more detailed study of the role of the KAKU4 protein suggests that it contributes to the morphology of the nucleus in maintaining heterochromatin at the nuclear periphery. In summary, we propose the existence of a transcription factor network recruited to the nuclear periphery by SUN, CRWN and KAKU4 proteins. This protein-protein interaction network would participate in the sequestration of certain transcription factors and/or the localization of certain chromatin domains to the nuclear periphery in order to activate or suppress their transcription.

<p>Molecular interactions lie at the heart of myriad biological processes. Knowledge of molecular recognition processes and the ability to model and predict interactions of any biological molecule to any chemical compound are the key for better understanding of cell functions and discovery of more efficacious medicines.</p><p>This thesis presents contributions to the development of a novel chemo-bioinformatics approach called proteochemometrics; a general method for interaction space analysis of biological macromolecules and their ligands. In this work we explore proteochemometrics-based interaction models over broad groups of protein families, evaluate their validity and scope, and compare proteochemometrics to traditional modeling approaches.</p><p>Through the proteochemometric analysis of large interaction data sets of multiple retroviral proteases from various viral species we investigate complex mechanisms of drug resistance in HIV-1 and discover general physicochemical determinants of substrate cleavage efficiency and binding in retroviral proteases. We further demonstrate how global proteochemometric models can be used for design of protease inhibitors with broad activity on drug-resistant viral mutants, for monitoring drug resistance mechanisms in the physicochemical sense and prediction of potential HIV-1 evolution trajectories. We provide novel insights into the complexity of HIV-1 protease specificity by constructing a generalized IF-THEN rule model based on bioinformatics analysis of the largest set of HIV-1 protease substrates and non-substrates.</p><p>We discuss how proteochemometrics can be used to map recognition sites of entire protein families in great detail and demonstrate how it can incorporate target variability into drug discovery process. Finally, we assess the utility of the proteochemometric approach in evaluation of ADMET properties of drug candidates with a special focus on inhibition of cytochrome P450 enzymes and investigate application of the approach in the pharmacogenomics field.</p>

La matrice extracellulaire est constituée d'un réseau tridimensionnel de protéines et de polysaccharides complexes, les glycosaminoglycanes. Elle apporte un support structural aux tissus et aux cellules dont elle est capable de moduler la prolifération, la migration et la différenciation. Nous avons créé une base de données d'interactions extracellulaires protéine-protéine et protéine-glycosaminoglycane, MatrixDB, qui est disponible sur le Web (http://matrixdb.ibcp.fr). Nous avons intégré des données expérimentales, des données issues de l’analyse de la littérature et des données issues de bases de données d'interactions publiquement disponibles. Nous avons respecté les standards de curation et d’échange de données du consortium IMEx dont fait partie MatrixDB. MatrixDB permet la construction et la visualisation de l'interactome extracellulaire entier et de plusieurs types de réseaux d'interactions, spécifiques d'une molécule, d'un tissu, d'une pathologie ou d'un processus biologique. Nous avons ainsi caractérisé le réseau d’interactions extracellulaire associé au vieillissement et mis en évidence le rôle important des glycosaminoglycanes et du calcium dans ce réseau. Nous avons construit le réseau d'interactions d'une matricryptine anti-angiogénique et anti-tumorale, l'endostatine, qui est issue du collagène XVIII. Les analyses structurales et fonctionnelles de ce réseau ont montré que les partenaires de l’endostatine sont majoritairement impliqués dans l’adhésion cellulaire et que les domaines EGF (Epidermal Growth Factor) sont surreprésentés. Cette propriété nous a permis d'identifier expérimentalement d'autres partenaires de l'endostatine possédant un ou plusieurs domaines EGF et de nouvelles fonctions de l’endostatine. Nous avons modélisé les complexes formés par l'endostatine avec deux de ses partenaires pour identifier les sites d'interactions. Ces prédictions, associées aux données expérimentales, ont permis de déterminer des interactions susceptibles d'être établies simultanément par l'endostatine. L'intégration de ces données et des paramètres cinétiques et d'affinité dans le réseau d'interactions de l'endostatine sera utilisée pour proposer un modèle de son mécanisme d'action qui reste mal connu / The extracellular matrix is composed of a tridimensional network of proteins and complex polysaccharides called glycosaminoglycans. It provides a structural support to tissues and modulates cell proliferation, migration and differenciation. We have created a database of protein-protein and proteinglycosaminoglycan extracellular interactions, MatrixDB (http://matrixdb.ibcp.fr). We have integrated experimental data, data issued of the literature curation and data from interaction databases publicly available. We have respected the curation and exchange standards of the IMEx consortium that includes MatrixDB. MatrixDB allows the construction and the visualization of the entire extracellular network and other types of interaction networks specific of a molecule, a tissue, a disease or a biological process. We have characterized the aging-related extracellular interaction network and underlined the important role of glycosaminoglycans and calcium in this network. We have constructed the interaction network of an antitumoral and anti-angiogenic matricryptin, endostatin, issued from collagen XVIII. Functional and structural analysis of their network showed that partners of endostatin are mostly involved in cell adhesion and that EGF domains are overrepresented. This has allowed us to to identify experimentally other partners of endostatin possessing one or more EGF domains and to propose new functions of endostatin. We have modelled complexes formed by endostatin with two of its partners to identify the binding sites.These predictions, associated with experimental data, allowed us to determine interactions able to be established simultaneously by endostatin. Integration of these data and of kinetics and affinity parameters in the interaction network of endostatin will be used to build a model of its mechanism of action that is not fully elucidated