Current treatments for prostate cancer (PCa) are marred with high relapse frequency and development of progressively aggressive cancers; developing new treatment options for PCa remains crucial. In this project, a series of synthetic compounds based on natural products will be screened to identify inhibitors for Aurora-A kinase (Aur-A). Aur-A facilitates centrosome separation and bipolar spindle formation during mitosis. Aur-A is overexpressed in metastatic PCa cells, and is a good candidate for targeted therapies. Compound libraries are designed using natural compounds that contain simple structural elements as starting points for developing drug like libraries. High-throughput screening of these libraries will be used to identify potent antimitotic agents that selectively affect cancer cells but not normal cells. A combination of in vitro protein assays – quantifying protein activity – cell-based assays – measuring cell growth and proliferation – and cell-reporter assays – to determine which metabolic pathway the compound affects – were used to identify potential inhibitors. Through these methods, we have identified several compounds, with special consideration to thiazole piperazine compounds, to successfully inhibit proliferation of metastatic PCa cells.

Mutações ativadoras no gene KRAS são prevalentes em cancer de pulmão e a as vias de sinalização de RAS estão aumentadas em células iniciadoras de tumor (CITs), que são definidas como células autorrenováveis capazes de iniciar a formação tumoral, sustentar o crescimento tumoral e promover a disseminação tumoral. Entretanto, terapias direcionadas a RAS não foram efetivas até hoje e a identificação de alvos de KRAS que contribuam para o fenótipo oncogênico é necessária. Como a quinase Aurora A (AURKA) já foi implicada, tanto na oncogênese induzida por KRAS, quanto em promover a função das CITs, nós hipotetizamos que a inibição das vias de AURKA seria detrimental para a função de CITs pulmonares portadoras de KRAS oncogênica, desta forma diminuindo o comportamento maligno do câncer de pulmão. Para avaliar a função das CITs, nós usamos ensaios de crescimento de tumoresferas que permitem o crescimento seletivo de CITs in vitro. As linhagens pulmonares positivas para KRAS H358 e A549 formaram tumoresferas em cultura de baixa aderência e, quando comparadas às linhagens parentais, às células oriundas de tumoresferas apresentaram maior capacidade clonogênica in vitro e maior tumorigenicidade in vivo. Além disso, uma análise por qPCR revelou que as células oriundas de tumoresferas possuem expressão aumentada de fatores de células tronco, uma característica de CITs. Em seguida, nós inibimos a AURKA nas linhagens pulmonares positivas para KRAS H358 e A549 por interferência de RNA (RNAi) ou com um inibidor das quinases Aurora (AI II). A inibição de AURKA diminuiu a formação de tumoresferas e o crescimento destas em culturas seriadas, além de reduzir a capacidade clonogênica das células oriundas de tumoresferas. Estes resultados indicam que a AURKA é importante para a autorrenovação e a oncogenicidade de CITs, e que a AURKA induz o fenótipo tronco-tumoral, o que é corroborado pelo achado de que a inibição de AURKA nas tumoresferas reduz a expressão de fatores de célula tronco. Um destes fatores regulados por AURKA é o marcador de superfície de célula tronco CD24. De fato, quando comparadas às células cultivadas de forma aderente, as células oriundas de tumoresferas apresentam maior número de células positivas para CD24 (CD24+) e estes números são reduzidos pelo tratamento com AI II. Finalmente, nós purificamos células H358 CD24+ por citometria de fluxo e mostramos que, quando comparadas às células negativas para CD24, as células CD24+ apresentam maior capacidade de formar tumoresferas em culturas seriadas, e o tratamento com AI II inibe preferencialmente a capacidade de células CD24+ de formarem tumoresferas. Nossos resultados sugerem que uma terapia baseada na inibição de AURKA pode reduzir o número e função de CITs pulmonares portadoras de KRAS oncogênica e, portanto, pode representar uma estratégia terapêutica atraente para reduzir a recidiva e metástase no câncer de pulmão induzido por KRAS. / Activating mutations in KRAS are prevalent in lung cancer and RAS sinaling is enhanced in cancer initiating cells (CICs), which are defined as self-renewing tumor cells able to initiate tumor formation, sustain tumor growth and drive tumor dissemination. However, therapies targeted to oncogenic RAS have been ineffective to date and identification of KRAS targets that impinge on the oncogenic phenotype is warranted. Because Aurora kinase A (AURKA) has been implicated both in RAS oncogenesis and in promoting CIC function, we hypothesized that targeting AURKA pathways would impair KRAS-positive lung CIC function, thereby decreasing lung cancer malignant behavior. To evaluate CIC function, we used tumorsphere assays that allow selective growth of CICs in vitro. KRAS positive lung cancer H358 and A549 cells formed tumorspheres under low attachment conditions, and, when compared to the parental cell lines, sphere-forming cells had increased clonogenic ability in vitro and increased tumorigenicity in vivo. In addition, qPCR analysis revealed that tumorsphere cells displayed increased expression of stem cell factors, a hallmark of CICs. Next, we targeted AURKA in KRAS positive lung cancer H358 and A549 cells by RNA interference (RNAi) or with an Aurora inhibitor (AI II). AURKA targeting decreased tumorsphere formation and growth in serial cultures and reduced clonogenic growth of tumorsphere-forming cells. These results indicate that AURKA is important for CIC selfrenewal and oncogenicity and that AURKA induces a CIC phenotype, which is further underscored by the finding that AURKA targeting in tumorspheres decreases expression of stem cell factors. One such factor shown to be regulated by AURKA is the stem cell surface marker CD24. In fact, when compared to adherent cultures, A549 and H358 tumorspheres display increased numbers of CD24-positive (CD24+) cells and these numbers are reduced by AI II treatment. Finally we purified H358 CD24+cells by flow cytometry and showed that, when compared to CD24-negative cells, CD24+ cells have increased ability to form tumorspheres in serial cultures, and AI II treatment preferentially reduced the ability of CD24+ cells to form tumorspheres. Our results suggest that AURKA inhibition therapy can reduce the number and function of KRAS-positive lung CICs, and, therefore might be an attractive therapeutic strategy to reduce recurrence and metastasis in KRAS-induced lung cancer.

Durante a divisão celular as Aurora-quinases (AURKA e AURKB) participam da formação e controle das fibras do fuso mitótico enquanto as isoformas proteicas (DIDO1, DIDO2 e DIDO3), originadas do splicing alternativo do gene DIDO, auxiliam na junção dos microtúbulos aos cinetócoros. Portanto, ambas são relevantes na regulação do ciclo celular. Interessantemente, a superexpressão (ou o ganho de função) das AURKs ou a baixa expressão (ou perda de função) das isoformas de DIDO estão ambos associados com amplificação dos centrossomos e à instabilidade cromossômica (CIN), com consequente aneuploidia. Dentre as doenças hematológicas com registros de CIN, a leucemia linfoide crônica (LLC) pode apresentar amplificação dos centrossomos e alteração nos níveis de expressão das AURKs acarretando aneuplodias. Apesar disso, não existem estudos avaliando a potencial associação destes genes com CIN na LLC. Avaliando seus níveis de expressão gênica em amostras de LLC de pacientes com ou sem aberrações cromossômicas, mostramos que o aumento dos níveis de AURKA e AURKB e, inversamente, a redução dos níveis das variantes de DIDO, são significativamente associados com ganhos cromossômicos e com aumento da contagem de glóbulos brancos (WBC). Claramente, amostras de LLC sem qualquer anormalidade citogenética apresentam níveis de expressão semelhantes às amostras que contêm aberrações não-numéricas. O achado de que níveis de expressão de AURKs e variantes de DIDO são completamente opostos, mostrando um padrão discreto de inter-relação, levou-nos a investigar o potencial mecanismo regulatório por trás disso. Tendo em vista que outros, anteriormente, mostraram que o cluster oncogênico miR-17~92 é significativamente hiper-regulado em células de pacientes com LLC purificadas expressando genes IGHV não mutados (em comparação com células mutadas de pacientes) e, que o miR-17 é expresso em níveis significativamente mais elevados em células IGHV não mutadas ou ZAP-70 positivas (mau prognóstico geralmente associada à CIN), resolvemos investigar o potencial de regulação negativa dos microRNAs deste cluster sobre as variantes de DIDO. Além disso, com base no mecanismo regulatório já descrito pelo qual a superexpressão de AURKA induz a transcrição do cluster miR-17~92, mediada por E2F1 (com uma correlação entre as expressões de ambas as proteínas em diferentes tipos de câncer), decidimos investigar este eixo regulatório em LLC. Notavelmente, todas as variantes de DIDO apresentam-se preditas como fortes alvos de vários microRNAs deste cluster oncogênico. Mostramos, então, que amostras de LLC com baixa expressão de DIDO, além dos já mencionados níveis elevados de AURK, exibiram níveis significativamente mais elevados do fator de transcrição E2F1 e de seu alvo transcricional, o transcrito primário do miR-17~92 (MIR17HG). Além disso, por meio do uso da linhagem de celular NTERA-2, como modelo experimental, mostramos que o siRNA nocaute para AURKA (nos níveis transcricional e proteico, como confirmado por qPCR e western blot) é acompanhada por uma significativa redução de E2F1 e também de MIR17HG. Ainda, a transfecção de células NTERA-2 com sintéticos microRNAs miméticos do cluster miR-17~92 (ou seja, 19a-miR, miR-20a e miR-92a) resultou em uma clara e significativa redução dos níveis de transcrição de todas as variantes de DIDO. Por fim, a inibição do siRNA especifico para a variante DIDO3 (mas não às outras variantes) levou a uma redução significativa dos níveis de transcrição de todas as variantes de DIDO, indicando um mecanismo adicional contribuindo para a downregulação dos transcritos de DIDO. Ao todo, nossos resultados demonstram a existência de um potencial mecanismo regulatório interconectado entre AURK e DIDO, associado à CIN e maior contagem de WBC na LLC. Mais importante, os níveis de expressão elevada de AURKs e os baixos níveis associados das variantes de DIDO são especificamente relacionados com anormalidades citogenéticas apresentando ganhos cromossomais, com destaque para o mecanismo celular específico, subjacente à CIN, observado neste grupo distinto LLC. Dado o papel central da CIN na gênese e progressão do câncer, esses achados provavelmente terão um impacto importante no prognóstico ou tratamento da LLC. / During cell cycle division Aurora kinases (AURKA and AURKB) participate in the formation and control of mitotic spindle fibers, while, protein isoforms (DIDO1, DIDO2 and DIDO3), derived by alternative splicing of the DIDO gene, assist at the junction of microtubules to kinetochores. Thus, both are relevant to cell cycle maintenance. Interestingly, overexpression (or gain of function) of AURKs or low expression (or loss of function of DIDO) are both associated with centrosomal amplification and chromosomal instability (CIN), leading to aneuploidy. Among hematological diseases with CIN records, chronic lymphocytic leukemia (CLL) can display centrosome amplification and changes in AURKs expression levels leading to aneuploidy. Despite this, there are no studies evaluating the potential association of these genes with CIN in CLL. By evaluating their gene expression levels in CLL samples from patients with or without chromosomal aberrations, we show that increased levels of AURKA and AURKB and, conversely, reduced levels of DIDO variants, are both significantly associated with chromosomal gains and with increased white blood cell (WBC) counts. Clearly, CLL samples without any cytogenetic abnormality had expression levels similar to samples mostly harboring non-numerical aberrations. The finding that the expression levels of AURKs and DIDO variants are completely opposed, showing a discrete inter-related pattern, led us to investigate the potential regulatory mechanism behind this. Given that other have previously shown that the oncogenic miR-17~92 cluster is significantly upregulated in purified CLL patient cells expressing unmutated IGHV genes (as compared to mutated patient cells), and that miR-17 is expressed at significantly higher levels in unmutated or ZAP-70 high cases (bad prognostic cases generally associated with chromosomal instability), we investigated the potential negative regulation of DIDO variants by microRNAs from this cluster. In addition, based on the already described regulatory mechanism by which AURKA overexpression induces the E2F1-mediated transcription upregulation of the miR-17~92 cluster (with an observed expression correlation of both proteins in cancer specimens); we decided to investigate this regulatory axis in CLL. Notably, we found that all DIDO variants are predicted to be heavily targeted by several miRs of this oncogenic cluster. We show that CLL samples with low DIDO expression, in addition to the already mentioned AURK high levels, displayed significant higher levels of the transcription factor E2F1 and of its transcriptional target, the miR-17~92 primary transcript (MIR17HG). Moreover, by using the NTERA-2 cell line as a model, we show that siRNA knockdown of AURKA (at the transcript and protein level, as confirmed by qPCR and western blot) is accompanied by a striking significant reduction of E2F1 and also of MIR17HG. Furthermore, transfection of NTERA-2 cells with synthetic mimics of the miR-17~92 cluster (namely, miR-19a, miR-20a and miR-92a) results in a clear and significant reduction in the transcript levels of all DIDO variants. Finally, specific siRNA inhibition of the DIDO3 variant (but not the others) led to a significant reduction in the transcript levels of all DIDO variants, indicating an additional mechanism contributing to the downregulation of DIDO transcripts. Altogether, our results demonstrate the existence of a potential interconnected regulatory mechanism between AURK and DIDO, associated with CIN and higher WBC counts in CLL. More importantly, the high expression levels of AURKs and the associated low levels of DIDO variants are specifically associated with cytogenetic abnormalities presenting chromosomal gains, highlighting the specific cellular mechanism underlying the CIN observed in this distinct CLL group. Given the central role of CIN in cancer genesis and progression, these findings will likely have an important impact on prognosis or treatment of CLL.

Les microtubules constituent l’un des réseaux du cytosquelette des cellules eucaryotes. Ils jouent un rôle central dans de multiples fonctions comme la division cellulaire, les trafics intracellulaires et la morphogenèse cellulaire. Chez les plantes supérieures, les microtubules (MTs) forment différents réseaux qui s'assemblent au cours du cycle cellulaire. Cette spécificité nécessite un recrutement régulé des complexes de nucléation des MTs à l’enveloppe nucléaire, au cortex et au niveau de MTs préexistants, qui sont des sites de nucléation caractérisés. L'équipe d’A.C. Schmit (IBMP, CNRS, Strasbourg), dans laquelle j'ai effectué mon travail de thèse, se focalise sur la caractérisation des complexes de nucléation des MTs (γ-TuRCs) et la régulation de l'assemblage du fuseau mitotique chez les plantes. Deux nouvelles protéines associées au γ-TuRC ont été mises en évidence par une interaction directe avec l'un de ses composants AtGCP3. Ces protéines, AtGIP1 et AtGIP2 (GCP3 Interacting Protein 1 et 2), sont très conservées au cours de l'évolution, mais leur fonction reste totalement inconnue. Mon travail a été consacré à la caractérisation de cette nouvelle classe de protéines dans le but de comprendre leur rôle. Nos résultats suggèrent que l'association des protéines GIPs aux γ-TuRCs participe à la régulation de leur activité et à la formation d'un fuseau mitotique robuste. Le profil de localisation des protéines GIPs au cours du cycle cellulaire et les phénotypes observés chez les mutants "perte de fonction" gip1gip2 indiquent que ces protéines interviennent dans le recrutement des γ-TuRCs, la nucléation des MTs, l’assemblage du fuseau mitotique, le déroulement du cycle cellulaire et l'organisation des méristèmes. L’étude des mécanismes de régulation de cette famille de protéines a été initiée. Nos résultats ont permis d’identifier GIP1comme un substrat de la kinase Aurora1 in vitro. Les résultats d’expérience de complémentation avec des phosphomutants GIP1 indiquent que la/les fonction(s) des GIPs pourrai(en)t être dépendante(s) de la phosphorylation par la kinase Aurora1, qui est un régulateur avéré du cycle cellulaire. L’ensemble de mes travaux a ainsi contribué à la caractérisation de nouveaux acteurs du cytosquelette microtubulaire. Une meilleure connaissance de leur réseau d'interaction (interactome) ainsi que l’étude de leur homologue humain pourraient ouvrir de nouvelles perspectives de recherche dans le contrôle de la division cellulaire et la lutte contre le cancer. / Microtubules (MTs) constitute one of the cytoskeletal networks in eukaryotic cells. They are involved in various processes such as cell division, intracellular transport and cell morphogenesis. In higher plants, MTs can be organized into dynamic structures, which undergo continual assembly and disassembly during the cell cycle. This specificity requires the recruitment of the nucleation complexes of the MTs to the nuclear envelope, to the cortex and to pre-existing MTs. The work of A. C. Schmit’s team (IBMP, CNRS, Strasbourg), in which I did my thesis, focuses on the characterization of MT nucleation complexes (γ-TuRCs) and the regulation of mitotic spindle assembly in plants. We have identified small proteins interacting with Gamma-tubulin Complex Protein 3 (GCP) and named GIP1 and GIP2 (GCP3-Interacting Proteins). The aim of these studies was to characterize this new class of proteins in order to understand their role. It shows that GIPs are conserved among eukaryotes and suggests that their association with the γ-TuRC participates in the regulation of their activity and the formation of a robust mitotic spindle. The localization of GIPs during the cell cycle and the phenotypes observed in T-DNA insertional gip1gip2 double mutants indicatethat GIPs are required for the recruitment of γ-TuRCs, MT nucleation, spindle assembly, cell cycle regulation and stem cell maintenance. Likewise, in vitro assays showed that GIP1 is a novel substrate for Aurora kinase1, which is a well known cell cycle regulator. The results of complementation experiments with GIP1 phosphomutants indicate that the phosphorylation of GIPs may be required for their function(s). Altogether, our results have contributed to the characterization of a new class of proteins involved in MT nucleation/organization and functions. The study of the interaction network (interactome) of GIPs and oftheir homologues could open new ways of research in the control of cell division and in the fight against cancer.

Durante a divisão celular as Aurora-quinases (AURKA e AURKB) participam da formação e controle das fibras do fuso mitótico enquanto as isoformas proteicas (DIDO1, DIDO2 e DIDO3), originadas do splicing alternativo do gene DIDO, auxiliam na junção dos microtúbulos aos cinetócoros. Portanto, ambas são relevantes na regulação do ciclo celular. Interessantemente, a superexpressão (ou o ganho de função) das AURKs ou a baixa expressão (ou perda de função) das isoformas de DIDO estão ambos associados com amplificação dos centrossomos e à instabilidade cromossômica (CIN), com consequente aneuploidia. Dentre as doenças hematológicas com registros de CIN, a leucemia linfoide crônica (LLC) pode apresentar amplificação dos centrossomos e alteração nos níveis de expressão das AURKs acarretando aneuplodias. Apesar disso, não existem estudos avaliando a potencial associação destes genes com CIN na LLC. Avaliando seus níveis de expressão gênica em amostras de LLC de pacientes com ou sem aberrações cromossômicas, mostramos que o aumento dos níveis de AURKA e AURKB e, inversamente, a redução dos níveis das variantes de DIDO, são significativamente associados com ganhos cromossômicos e com aumento da contagem de glóbulos brancos (WBC). Claramente, amostras de LLC sem qualquer anormalidade citogenética apresentam níveis de expressão semelhantes às amostras que contêm aberrações não-numéricas. O achado de que níveis de expressão de AURKs e variantes de DIDO são completamente opostos, mostrando um padrão discreto de inter-relação, levou-nos a investigar o potencial mecanismo regulatório por trás disso. Tendo em vista que outros, anteriormente, mostraram que o cluster oncogênico miR-17~92 é significativamente hiper-regulado em células de pacientes com LLC purificadas expressando genes IGHV não mutados (em comparação com células mutadas de pacientes) e, que o miR-17 é expresso em níveis significativamente mais elevados em células IGHV não mutadas ou ZAP-70 positivas (mau prognóstico geralmente associada à CIN), resolvemos investigar o potencial de regulação negativa dos microRNAs deste cluster sobre as variantes de DIDO. Além disso, com base no mecanismo regulatório já descrito pelo qual a superexpressão de AURKA induz a transcrição do cluster miR-17~92, mediada por E2F1 (com uma correlação entre as expressões de ambas as proteínas em diferentes tipos de câncer), decidimos investigar este eixo regulatório em LLC. Notavelmente, todas as variantes de DIDO apresentam-se preditas como fortes alvos de vários microRNAs deste cluster oncogênico. Mostramos, então, que amostras de LLC com baixa expressão de DIDO, além dos já mencionados níveis elevados de AURK, exibiram níveis significativamente mais elevados do fator de transcrição E2F1 e de seu alvo transcricional, o transcrito primário do miR-17~92 (MIR17HG). Além disso, por meio do uso da linhagem de celular NTERA-2, como modelo experimental, mostramos que o siRNA nocaute para AURKA (nos níveis transcricional e proteico, como confirmado por qPCR e western blot) é acompanhada por uma significativa redução de E2F1 e também de MIR17HG. Ainda, a transfecção de células NTERA-2 com sintéticos microRNAs miméticos do cluster miR-17~92 (ou seja, 19a-miR, miR-20a e miR-92a) resultou em uma clara e significativa redução dos níveis de transcrição de todas as variantes de DIDO. Por fim, a inibição do siRNA especifico para a variante DIDO3 (mas não às outras variantes) levou a uma redução significativa dos níveis de transcrição de todas as variantes de DIDO, indicando um mecanismo adicional contribuindo para a downregulação dos transcritos de DIDO. Ao todo, nossos resultados demonstram a existência de um potencial mecanismo regulatório interconectado entre AURK e DIDO, associado à CIN e maior contagem de WBC na LLC. Mais importante, os níveis de expressão elevada de AURKs e os baixos níveis associados das variantes de DIDO são especificamente relacionados com anormalidades citogenéticas apresentando ganhos cromossomais, com destaque para o mecanismo celular específico, subjacente à CIN, observado neste grupo distinto LLC. Dado o papel central da CIN na gênese e progressão do câncer, esses achados provavelmente terão um impacto importante no prognóstico ou tratamento da LLC. / During cell cycle division Aurora kinases (AURKA and AURKB) participate in the formation and control of mitotic spindle fibers, while, protein isoforms (DIDO1, DIDO2 and DIDO3), derived by alternative splicing of the DIDO gene, assist at the junction of microtubules to kinetochores. Thus, both are relevant to cell cycle maintenance. Interestingly, overexpression (or gain of function) of AURKs or low expression (or loss of function of DIDO) are both associated with centrosomal amplification and chromosomal instability (CIN), leading to aneuploidy. Among hematological diseases with CIN records, chronic lymphocytic leukemia (CLL) can display centrosome amplification and changes in AURKs expression levels leading to aneuploidy. Despite this, there are no studies evaluating the potential association of these genes with CIN in CLL. By evaluating their gene expression levels in CLL samples from patients with or without chromosomal aberrations, we show that increased levels of AURKA and AURKB and, conversely, reduced levels of DIDO variants, are both significantly associated with chromosomal gains and with increased white blood cell (WBC) counts. Clearly, CLL samples without any cytogenetic abnormality had expression levels similar to samples mostly harboring non-numerical aberrations. The finding that the expression levels of AURKs and DIDO variants are completely opposed, showing a discrete inter-related pattern, led us to investigate the potential regulatory mechanism behind this. Given that other have previously shown that the oncogenic miR-17~92 cluster is significantly upregulated in purified CLL patient cells expressing unmutated IGHV genes (as compared to mutated patient cells), and that miR-17 is expressed at significantly higher levels in unmutated or ZAP-70 high cases (bad prognostic cases generally associated with chromosomal instability), we investigated the potential negative regulation of DIDO variants by microRNAs from this cluster. In addition, based on the already described regulatory mechanism by which AURKA overexpression induces the E2F1-mediated transcription upregulation of the miR-17~92 cluster (with an observed expression correlation of both proteins in cancer specimens); we decided to investigate this regulatory axis in CLL. Notably, we found that all DIDO variants are predicted to be heavily targeted by several miRs of this oncogenic cluster. We show that CLL samples with low DIDO expression, in addition to the already mentioned AURK high levels, displayed significant higher levels of the transcription factor E2F1 and of its transcriptional target, the miR-17~92 primary transcript (MIR17HG). Moreover, by using the NTERA-2 cell line as a model, we show that siRNA knockdown of AURKA (at the transcript and protein level, as confirmed by qPCR and western blot) is accompanied by a striking significant reduction of E2F1 and also of MIR17HG. Furthermore, transfection of NTERA-2 cells with synthetic mimics of the miR-17~92 cluster (namely, miR-19a, miR-20a and miR-92a) results in a clear and significant reduction in the transcript levels of all DIDO variants. Finally, specific siRNA inhibition of the DIDO3 variant (but not the others) led to a significant reduction in the transcript levels of all DIDO variants, indicating an additional mechanism contributing to the downregulation of DIDO transcripts. Altogether, our results demonstrate the existence of a potential interconnected regulatory mechanism between AURK and DIDO, associated with CIN and higher WBC counts in CLL. More importantly, the high expression levels of AURKs and the associated low levels of DIDO variants are specifically associated with cytogenetic abnormalities presenting chromosomal gains, highlighting the specific cellular mechanism underlying the CIN observed in this distinct CLL group. Given the central role of CIN in cancer genesis and progression, these findings will likely have an important impact on prognosis or treatment of CLL.

Mutações ativadoras no gene KRAS são prevalentes em cancer de pulmão e a as vias de sinalização de RAS estão aumentadas em células iniciadoras de tumor (CITs), que são definidas como células autorrenováveis capazes de iniciar a formação tumoral, sustentar o crescimento tumoral e promover a disseminação tumoral. Entretanto, terapias direcionadas a RAS não foram efetivas até hoje e a identificação de alvos de KRAS que contribuam para o fenótipo oncogênico é necessária. Como a quinase Aurora A (AURKA) já foi implicada, tanto na oncogênese induzida por KRAS, quanto em promover a função das CITs, nós hipotetizamos que a inibição das vias de AURKA seria detrimental para a função de CITs pulmonares portadoras de KRAS oncogênica, desta forma diminuindo o comportamento maligno do câncer de pulmão. Para avaliar a função das CITs, nós usamos ensaios de crescimento de tumoresferas que permitem o crescimento seletivo de CITs in vitro. As linhagens pulmonares positivas para KRAS H358 e A549 formaram tumoresferas em cultura de baixa aderência e, quando comparadas às linhagens parentais, às células oriundas de tumoresferas apresentaram maior capacidade clonogênica in vitro e maior tumorigenicidade in vivo. Além disso, uma análise por qPCR revelou que as células oriundas de tumoresferas possuem expressão aumentada de fatores de células tronco, uma característica de CITs. Em seguida, nós inibimos a AURKA nas linhagens pulmonares positivas para KRAS H358 e A549 por interferência de RNA (RNAi) ou com um inibidor das quinases Aurora (AI II). A inibição de AURKA diminuiu a formação de tumoresferas e o crescimento destas em culturas seriadas, além de reduzir a capacidade clonogênica das células oriundas de tumoresferas. Estes resultados indicam que a AURKA é importante para a autorrenovação e a oncogenicidade de CITs, e que a AURKA induz o fenótipo tronco-tumoral, o que é corroborado pelo achado de que a inibição de AURKA nas tumoresferas reduz a expressão de fatores de célula tronco. Um destes fatores regulados por AURKA é o marcador de superfície de célula tronco CD24. De fato, quando comparadas às células cultivadas de forma aderente, as células oriundas de tumoresferas apresentam maior número de células positivas para CD24 (CD24+) e estes números são reduzidos pelo tratamento com AI II. Finalmente, nós purificamos células H358 CD24+ por citometria de fluxo e mostramos que, quando comparadas às células negativas para CD24, as células CD24+ apresentam maior capacidade de formar tumoresferas em culturas seriadas, e o tratamento com AI II inibe preferencialmente a capacidade de células CD24+ de formarem tumoresferas. Nossos resultados sugerem que uma terapia baseada na inibição de AURKA pode reduzir o número e função de CITs pulmonares portadoras de KRAS oncogênica e, portanto, pode representar uma estratégia terapêutica atraente para reduzir a recidiva e metástase no câncer de pulmão induzido por KRAS. / Activating mutations in KRAS are prevalent in lung cancer and RAS sinaling is enhanced in cancer initiating cells (CICs), which are defined as self-renewing tumor cells able to initiate tumor formation, sustain tumor growth and drive tumor dissemination. However, therapies targeted to oncogenic RAS have been ineffective to date and identification of KRAS targets that impinge on the oncogenic phenotype is warranted. Because Aurora kinase A (AURKA) has been implicated both in RAS oncogenesis and in promoting CIC function, we hypothesized that targeting AURKA pathways would impair KRAS-positive lung CIC function, thereby decreasing lung cancer malignant behavior. To evaluate CIC function, we used tumorsphere assays that allow selective growth of CICs in vitro. KRAS positive lung cancer H358 and A549 cells formed tumorspheres under low attachment conditions, and, when compared to the parental cell lines, sphere-forming cells had increased clonogenic ability in vitro and increased tumorigenicity in vivo. In addition, qPCR analysis revealed that tumorsphere cells displayed increased expression of stem cell factors, a hallmark of CICs. Next, we targeted AURKA in KRAS positive lung cancer H358 and A549 cells by RNA interference (RNAi) or with an Aurora inhibitor (AI II). AURKA targeting decreased tumorsphere formation and growth in serial cultures and reduced clonogenic growth of tumorsphere-forming cells. These results indicate that AURKA is important for CIC selfrenewal and oncogenicity and that AURKA induces a CIC phenotype, which is further underscored by the finding that AURKA targeting in tumorspheres decreases expression of stem cell factors. One such factor shown to be regulated by AURKA is the stem cell surface marker CD24. In fact, when compared to adherent cultures, A549 and H358 tumorspheres display increased numbers of CD24-positive (CD24+) cells and these numbers are reduced by AI II treatment. Finally we purified H358 CD24+cells by flow cytometry and showed that, when compared to CD24-negative cells, CD24+ cells have increased ability to form tumorspheres in serial cultures, and AI II treatment preferentially reduced the ability of CD24+ cells to form tumorspheres. Our results suggest that AURKA inhibition therapy can reduce the number and function of KRAS-positive lung CICs, and, therefore might be an attractive therapeutic strategy to reduce recurrence and metastasis in KRAS-induced lung cancer.

Aurora-A (AurA) est une sérine/thréonine kinase jouant un rôle majeur dans le cycle cellulaire. Elle est connue pour son rôle oncogène et les compagnies pharmaceutiques développent des inhibiteurs ciblant son activité kinase. Cependant, il a été montré chez différentes espèces qu’Aurora-A possède des rôles indépendants de son activité kinase et agit également comme suppresseur de tumeur quand son activité kinase est altérée. Ceci pose donc un problème dans le développement des inhibiteurs car cibler l’activité kinase d’Aurora-A pour traiter le cancer pourrait mener à l’effet inverse. Pour résoudre ce dilemme, j’ai étudié en détail les phénotypes de mutants AurA nul et hypomorphe chez Drosophila melanogaster. J’ai étudié à la fois les défauts de développement en me basant sur le temps de pupation des larves et le rôle de suppresseur de tumeur en me basant sur les neuroblastes du cerveau central. Dans ce modèle, une caractéristique des suppresseurs de tumeur est leur capacité à induire la formation de neuroblastes supplémentaires dans le cerveau central conduisant à une surcroissance du cerveau. Chez les mutants AurA, la taille du cerveau est plus petite jusqu’à 96h de développement larvaire. Cependant, la pupation arrivant normalement entre 96h et 120h de développement larvaire est retardée chez le mutant et les larves ont une taille plus importante. Chez les mutants en retard de pupation le cerveau devient plus gros que ceux du contrôle. Le cerveau des mutants AurA a une importante augmentation du nombre de cellules positives pour Deadpan, un marqueur spécifique des neuroblastes et ce, avant que le cerveau des mutants AurA devienne plus grand que celui du contrôle. De plus, les disques imaginaux d’ailes et la glande annulaire sont clairement plus petits que ceux du contrôle à 96h de développement larvaire et les larves mutantes atteignent les stades L2 et L3 plus tôt. En conclusion, les mutants AurA montrent 1) une avance dans leur développement précoce certainement reliée au défaut de croissance de la glande annulaire ; 2) un retard de pupation ressemblant à celui observé en cas de défauts dans la voie de l’ecdysone, certainement dû à des défauts de croissance des disques imaginaux d’ailes ; 3) une surcroissance du cerveau à mettre en lien à la fois avec une augmentation du nombre de pseudo-neuroblastes et avec le retard de pupation. / Aurora-A (AurA) is a major kinase playing various roles in cell cycle. It’s a well-known oncogene and companies are developing drugs inhibiting its kinase activity. However, it has been shown in different species that AurA can have a kinase independent role or act as a tumor suppressor when its kinase activity is altered. This represents a problem for drugs development as inhibiting AurA kinase activity only could lead to life threatening phenotypes. To address this dilemma, we carefully deciphered phenotypes of AurA null and AurA hypomorph mutants in Drosophila melanogaster using the pupation as readout for development timing and larval central brain neuroblasts as model for tumorigenic study. One readout to define a tumor suppressor in this model is a brain overgrowth phenotype associated to central brain neuroblasts over-proliferation. In AurA mutants, brain size appears slightly smaller until 96h of larval development. However, pupation occurring normally between 96 and 120h of larval development is delayed in AurA mutants and larvae have an increased size. In this “delayed” mutant larvae, brains are eventually bigger than wild-type controls. Furthermore, AurA mutant central brains show a huge increased number of cells positive for deadpan, a marker of neuroblast identity, even before the appearance of brain over-growth phenotype. Additionally, wing discs and ring glands are clearly smaller in AurA mutants at 96h compared to control and mutant larvae reach L2 and L3 developmental stage earlier than control. In conclusion, AurA mutants have: 1) a precocious developmental advance certainly related to ring gland growth defect; 2) a pupation delay which resembles Ecdysone pathway timing defects certainly due to wing discs growth defect; 3) an enlarged brains phenotype due to an increased of the number of neuroblast-like cells and the pupation delay.

La mitose assure la répartition égale du génome. La kinase mitotique Aurora B y joue un rôle majeur en contrôlant la fidélité de la ségrégation des chromosomes de par sa localisation aux centromères et aux microtubules, qui nécessite son ubiquitination par CUL3. Cependant, le mécanisme conduisant la forme ubiquitinée d’Aurora B sur ces structures mitotiques reste à déterminer. Dans ce contexte, j’ai pu identifier la protéine UBASH3B, qui contient un domaine de liaison à l’ubiquitine (UBD) comme un régulateur essentiel de la ségrégation chromosomique, agissant comme un récepteur de l’ubiquitine pour Aurora B. UBASH3B interagit directement avec Aurora B et cette interaction est dépendante de la modification d’Aurora B par l’ubiquitine ainsi que de CUL3. UBASH3B ne régule pas le niveau d’expression d’Aurora B. En revanche, UBASH3B se localise aux fuseaux mitotiques et est à la fois nécessaire et suffisant pour transférer Aurora B aux microtubules. De plus, la redistribution d’Aurora B des centromères vers les microtubules contrôle le déroulement et la fidélité de la ségrégation des chromosomes et donc le contenu correct du matériel génétique des cellules. Ainsi, mes résultats expliquent comment la modification par l’ubiquitine régule la localisation et la fonction d’Aurora B, reliant une voie de signalisation impliquant un récepteur à l’ubiquitine à la mitose. / Mitosis ensures equal segregation of the genome. The major mitotic kinase Aurora B controls fidelity of chromosome segregation by its localization to centromeres and microtubules, which requires CUL3-mediated ubiquitylation. However, it remains unknown how ubiquitylated Aurora B is targeted to mitotic structures. Here, I identify ubiquitin-binding domain (UBD) protein UBASH3B that critically regulates chromosome segregation, acting as ubiquitin receptor for Aurora B. UBASH3B directly binds Aurora B, and this interaction is dependent on CUL3 and on ubiquitin recognition. UBASH3B does not regulate protein levels of Aurora B. Instead, UBASH3B localizes to the mitotic spindle and is both required and sufficient to transfer Aurora B to microtubules. Moreover, redistribution of Aurora B from centromeres to microtubules controls timing and fidelity of chromosome segregation and thereby euploidy of cells. Thus, my findings explain how ubiquitin attachment regulates localization and function of Aurora B, linking receptor-mediated ubiquitin signaling to mitosis.

La kinase Aurora A est une protéine essentielle au cycle cellulaire et plus particulièrement lors de la mitose. En effet, Aurora A est nécessaire à l'entrée en mitose et joue un rôle dans la maturation des centrosomes. Elle participe à l'assemblage du fuseau mitotique et est nécessaire à la réussite de la cytodiérèse. Elle est également nécessaire à l'égale répartition des mitochondries dans les cellules filles et joue un rôle dans l'épissage alternatif des ARNm de facteurs apoptotiques. Au-delà de ses fonctions mitotiques, plusieurs études récentes indiquent qu'Aurora A présente des fonctions supplémentaires dans les cellules en interphase. Elle est notamment essentielle au désassemblage du cil primaire et joue un rôle dans la dynamique des microtubules et la migration cellulaire. Enfin, une dérégulation de son expression, de sa stabilité et/ou de son activité perturbe le déroulement du cycle cellulaire ce qui conduit à la transformation des cellules et favorise l'apparition de cancers. Ses fonctions normales ainsi que ses fonctions lors de la carcinogenèse sont conduites à travers les nombreux partenaires protéiques qui entrent en interaction avec elle. Ils modulent son activité, sa localisation et sa stabilité. En retour Aurora A phosphoryle un bon nombre d'entre eux régulant ainsi leur activité, localisation et stabilité. Cependant, l'analyse des interactions déjà connues d'Aurora A ne permet pas d'expliquer tous les phénotypes observés lors de sa dérégulation. Afin de mieux comprendre les fonctions d'Aurora A, les mécanismes qui la régulent et mettre en évidence ses multiples rôles au sein de la cellule, j'ai construit puis analysé un interactome d'Aurora A généré à partir d'une méthode de purification d'affinité couplée à la spectrométrie de masse en tandem. J'ai identifié 477 partenaires potentiels dont 180 présentant une forte probabilité d'être des partenaires directs de la kinase. L'analyse bioinformatique approfondie de cet interactome a permis de révéler les partenaires associés à des mécanismes liés à la mitochondrie et l'épissage des ARN messagers mettant en évidence une implication potentielle d'Aurora A dans ces mécanismes. Pour valider cet interactome, j'ai choisi d'étudier plus précisément deux partenaires identifiés dans cette étude : les protéines WDR62 et CEP97. J'ai montré que ces deux partenaires co-localisent avec Aurora A et sont phosphorylés par la kinase. Ainsi, ce travail de thèse a permis de mettre en évidence un nombre important de nouveaux partenaires d'Aurora A associés à de nouvelles fonctions. L'étude de ces nouvelles fonctions liées aux mitochondries et à l'épissage des ARN, constitue deux nouveaux projets actuellement menés par des collaborateurs au sein de notre institut. / The kinase Aurora A is an essential mitotic cell cycle protein. Aurora A is necessary for mitotic entry and for the maturation and separation of centrosomes. It participates in mitotic spindle assembly and chromosome biorientation, and it is essential for the completion of cytokinesis. Furthermore, Aurora A activity is necessary for the equal distribution of mitochondria to daughter cells and, through its role in the alternative splicing of mRNA of apoptotic factors, it provides a link between cell cycle control and apoptosis. Beyond its mitotic functions, several recent studies suggest that Aurora A is also important during interphase. Notably, it influences microtubule dynamics, promotes cell migration and polarity control and is essential for primary cilia disassembly. Reflecting the fact that Aurora A is found to be up-regulated in many cancers, deregulation of Aurora A activity can result in an aberrant cell cycle, ultimately leading to malignant transformation of cells. The crucial regulation of Aurora A’s numerous functions is achieved through its interaction with several protein partners, which modulate its activity, localisation and stability. Aurora A in turn phosporylates a number of them, thus regulating their activity, localisation and stability. However, the known interactions of Aurora A cannot explain all the phenotypes that have been described of its deregulation.To better understand the functions of Aurora A, the regulation mechanisms governing it, and to expose its multiple roles in the cell, I have built and analysed an Aurora A interactome using tandem affinity purification coupled with mass spectrometry. This resulted in the identification of 477 potential interacting partners, of which, 180 were determined to have a high probability of interacting directly with the kinase.In-depth bioinformatic analysis of this interactome has revealed the associated partners to be related to mitochondria and mRNA splicing, highlighting the potential involvement of Aurora A in these mechanisms. To validate the interactome, two of the proteins identified in this study, WDR62 and CEP97, were examined in detail. Here I show that these two proteins colocalise with Aurora A, and are phosphorylated by the kinase.WDR62 is implicated in microcephaly and is deregulated in certain cancers. I have shown that Aurora A phosphorylates WDR62 during mitosis, and that this phosphorylation is necessary for its localisation to the centrosomes. CEP97 is a poorly charactarised protein of the primary cilium, abnormalities of which are associated with ciliopathies. I have shown that Aurora A phosphorylates CEP97 in vitro, and that the inhibition of Aurora A activity in vivo perturbs the localisation of CEP97 to cilia and centrosomes.This study has identified a number of new Aurora A-interacting proteins, implicating the kinase with novel functions. These functions, related to mitochondria and mRNA splicing have opened up a new area for further investigation.

A Leucemia Mielóide Crônica (LMC) é uma neoplasia mieloproliferativa resultante da expansão clonal de células mielóides positivas para o cromossomo Philadelphia. A patogênese da LMC está associada à expressão do oncogene BCR-ABL1, que codifica a proteína Bcr-Abl com constitutiva atividade da tirosina quinase, promovendo a mieloproliferação exacerbada e a resistência à apoptose das células leucêmicas. Os pacientes com LMC são tratados principalmente com inibidores de tirosina quinase (TKI), mas a resistência aos inibidores e a refratariedade tem sido relatada em alguns pacientes na fase crônica e na maioria dos pacientes em fases avançadas da doença. Assim sendo, continua a ser de suma importância a elucidação da patogênese da LMC e a busca de novos alvos terapêuticos, como os membros da via de sinalização Hippo e reguladores do ciclo celular, da família Aurora quinase. O presente estudo quantificou o nível de expressão de genes que codificam componentes da via de sinalização Hippo (MST1, MOB1B, MOBKL1B, LATS1, LATS2, YAP e TAZ) e Aurora quinases A e B em: 1) pacientes com LMC em diferentes fases da doença, resistentes ou sensíveis à terapia com mesilato de imatinibe (MI), em indivíduos saudáveis e 2) linhagens celulares HL-60, HL-60.Bcr- Abl tratadas com TKI (imatinibe, dasatinibe e nilotinibe), KCL22 e LAMA84 resistentes e sensíveis ao MI. Os níveis de expressão dos genes alvo foram correlacionados com o índice de prognóstico de Sokal. Os principais resultados revelaram que há alteração nos genes MST1, MOB1B, MOBKL1B, LATS1, LATS2, TAZ, AURKA e AURKB em pacientes com LMC em relação aos controles. Não houve correlação entre o índice de Sokal e a expressão gênica dos genes da via Hippo, MST1, MOB1B, MOBKL1B, LATS1, LATS2 e TAZ, assim como os genes Aurora quinases A e B. Pacientes com LMC em fases avançadas apresentaram maiores valores de expressão dos genes TAZ e AURKB, comparado aos pacientes na fase crônica. Os pacientes resistentes ao TKI apresentaram as expressões dos genes MST1, TAZ e AURKB, significativamente mais elevadas, comparado aos pacientes sensíveis ao MI. Os resultados dos estudos em linhagens celulares indicaram principalmente que a expressão do gene LATS1 pode ser modulada pela atividade de tirosina quinase Bcr-abl e que o oncogene BCR-ABL1 induz a expressão de AURKA, AURKB, LATS1 e TAZ. Em conjunto os dados obtidos revelam que a alteração da expressão dos genes da família Aurora quinase, A e B, e dos genes que codificam proteínas da via Hippo contribui para a patogênese e progressão da LMC. O desenvolvimento de fármacos e/ou a identificação de marcadores tumorais para a via de sinalização Hippo e família Aurora quinase, podem otimizar o tratamento da LMC, aumentando a susceptibilidade das células leucêmicas a apoptose e levando a um melhor prognóstico da doença / Chronic myeloid leukemia (CML) is a myeloproliferative neoplasm resulting from clonal expansion of myeloid cells positive for the Philadelphia chromosome. The CML pathogenesis is associated with BCR-ABL1 oncogene expression, which encodes the Bcr-Abl protein with a constitutive tyrosine kinase activity, leading to leukemic cell high proliferation and resistance to apoptosis. CML patients are mainly treated with tyrosine kinase inhibitors (TKI), but some of CML patients in chronic phase are resistant and in advanced phases are refractory to TKI. Thus, it is still relevant to elucidate the CML pathogenesis and seek to new therapeutic targets, including the Hippo signaling pathway members and cell cycle regulatory genes such as those encoding the Aurora kinase family. The present study quantified the RNA expression level of genes encoding components from the Hippo cell signaling pathway (MST1, MOB1B, MOBKL1B, LATS1, LATS2, YAP, and TAZ) and Aurora kinase A and B in: 1) CML patients at different stages of the disease, in CML patients resistant or sensitive to imatinib mesylate therapy, healthy individuals and 2) in cell lines HL-60, HL-60.Bcr-Abl treated with TKI (imatinib mesylate, dasatinib and nilotinib), KCL22 and LAMA84 resistant and sensitive to IM. The RNA expression levels of the target genes were also correlated to the CML Sokal\'s prognostic score values. The main results revealed that there are alterations in the genes MST1, MOB1B, MOBKL1B, LATS1, LATS2, TAZ, AURKA and AURKB in patients with CML in relation to the controls. There was no correlation between the Sokal index and the gene expression of the Hippo, MST1, MOB1B, MOBKL1B, LATS1, LATS2 and TAZ genes, as well as the Aurora kinase genes A and B. Patients with advanced phase CML had higher values of expression of the TAZ and AURKB genes, compared to the patients in the chronic phase. Patients resistant to TKI had significantly higher MST1, TAZ and AURKB gene expression compared to MI-sensitive patients. The results of the studies in cell lines indicated primarily that the expression of the LATS1 gene can be modulated by the Bcr-abl tyrosine kinase activity and the BCR-ABL1 oncogene induces the expression of AURKA, AURKB, LATS1 and TAZ. Together, the data show that altered expression of Aurora kinase family genes, A and B, and genes coding for Hippo pathway proteins contribute to the pathogenesis and progression of CML. The development of drugs and/or identification of tumor markers for the Hippo signaling pathway and the Aurora kinase family can optimize CML treatment by enhancing the susceptibility of leukemic cells to apoptosis and leading to a better disease prognosis