A identidade na segmentação do corpo de diversos organismos, durante o desenvolvimento, é devida, em grande parte, à ação das proteínas homeóticas. Em especial, dois grupos de proteínas, Trithorax (trxG) e Polycomb (PcG) têm um papel fundamental na manutenção, respectivamente, da ativação e da repressão da transcrição gênica, associando-se à cromatina. A importância das PcG nos estimulou a buscar a caracterização das proteínas humanas ortólogas ao \"Enhancer of Polycomb\" (Epc) de Drosophila, até então não descritas no genoma humano. Para tanto, buscamos: - obter a sequência completa e mapear o cDNA do novo gene humano homólogo ao \"Enhancer of Polycomb\" de Drosophila; - analisar sua expressão em tecidos fetais, adultos e tumorais e fazer estudos buscando sua caracterização funcional. Encontramos, mapeamos e obtivemos a seqüência completa de dois genes humanos, ortólogos de Epc1 (10p11-22) e de Epc2 (2q21-23) de camundongo, publicando estes dados em 2001 (Camargo et al., 2001). Ambos os genes são bastante conservados entre várias espécies, sendo que o cDNA de hEPC2 humano, por exemplo, é 94% idêntico ao Epc2 de camundongo e possui 96% de identidade ao nível de proteína, sugerindo que a função do gene deve ter sido mantida durante a evolução. No entanto, as seqüências protéicas de hEPC1 e hEPC2 humanos possuem apenas 68% de identidade entre si. Portanto, é provável que após a duplicação dos parálogos, estes tenham divergido funcionalmente. A expressão de ambos os genes foi avaliada utilizando \"dot-blots\" contendo 76 mRNAs de amostras de tecidos fetais, adultos e tumorais, mostrando-se fraca e ubíqua. Análises in silico sugeriram a existência de 4 isoformas de splicing para hEPC2, as quais foram validadas por RT-PCR ou \"Northern blots\". Uma das isoformas (de 2.7 Kpb) se mostrou mais abundante em todas as linhagens tumorais estudadas através de análises de \"Northern blot\", principalmente nas linhagens de linfoma de Burkitt\'s Raji e na linhagem de leucemia pró-mielocítica HL-60. Esta isoforma é gerada através de um sítio alternativo de poli-adenilação, que reduz sua porção 3\'UTR, retirando 4 dos 5 \"elementos ricos em adenilatos e uridilatos\" (AREs), envolvidos com a degradação de mRNAs lábeis que codificam proteínas regulatórias. Estes resultados se encontram em um manuscrito recentemente submetido à publicação (anexo à tese). Interação entre hEPC2 e SMADs e sua modulação por TGF-β. Durante a montagem da seqüência completa de hEPC2, verificamos que duas ESTs patenteadas mostravam alta identidade com o gene. Estas seqüências foram descritas como sendo parte de uma nova proteína de interação com as proteínas da família SMAD, envolvidas com transdução de sinais desencadeados por TGF-β. Esta citocina por sua vez, regula a proliferação, diferenciação e morte celular. Partimos para a avaliação da possível interação entre hEPC2 e as SMADs, em colaboração com o grupo do Dr. Aristidis Moustakas, do Ludwig Institute for Cancer Research de Uppsala, Suécia. Os resultados de co-imunoprecipitação sugeriram que as SMADs 2, 3, 4, 7 e 8 interagem com hEPC2, sendo que a interação entre as SMAD2, SMAD3, SMAD4 e hEPC2 nas células tratadas com TGF-β1, mostraram uma redução na co-imunoprecipitação. Este resultado sugere que TGF-β1 modula negativamente a interação entre essas proteínas. Da mesma maneira, foi observada uma redução na interação de hEPC2 com SMAD8 após o tratamento com BMP-7. Esse resultado é ainda mais destacado para as SMADs 2 e 3. Estes dados foram observados para ambas as construções de hEPC2, o que sugere fortemente a veracidade da interação entre estas proteínas. A localização celular de hEPC2, e também sua co-localização com SMAD2 foram investigadas através de imunofluorescência indireta e confirmaram a predição do programa PSORTII, de que hEPC2 se localiza no núcleo. No entanto, não foi possível observar a co-localização entre hEPC2 e SMAD2. É possível que hEPC2 não se ligue diretamente ao DNA, necessitando se associar como parceiro de um fator de transcrição. Esta foi uma das hipóteses para a atuação de hEPC2, como um co-fator que se associe com uma das SMADs e se ligue a um elemento específico de ligação a SMAD (SBE). Para investigar essa hipótese um ensaio de gene repórter foi feito utilizando uma construção de um repórter contendo 12 repetições da seqüência CAGA (seqüência específica de ligação das SMADs 2,3 e 4) fusionado com o gene da luciferase. No entanto, este ensaio não demonstrou que a transcrição de SMAD2 é dependente de hEPC2 e o experimento deverá ser repetido. Para confirmar a interação entre hEPC2 e as SMADs, será feito um experimento de \"pull-down\". Para tal o cDNA de hEPC2 foi clonado no vetor pET-32A de expressão indutível em bactérias. A proteína recombinante já foi produzida, tendo sido induzida e posteriormente purificada em condições desnaturantes. Apesar de dezenas de genes PcG terem sido caracterizados em Drosophila, poucos destes genes foram estudados em mamíferos. Portanto, a descrição do gene hEPC2 e seus transcritos alternativos, contribui para o conhecimento de PcG humanos, indicando a associação de maior expressão de uma de suas isoformas em linhagens celulares tumorais. Em relação à interação de hEPC2 com as SMADs, é interessante observar que nenhuma outra proteína foi descrita por possuir a particularidade de interagir com as SMADs de diferentes categorias. Talvez este seja um dado importante, que indique o papel singular de hEPC2 na sinalização de TGF-β1. / The identity of body segmentation in several organisms during development is, to a large extent, due to the action of the homeotic proteins. In particular, two groups of proteins, the Trithorax (trxG) and Polycomb (PcG), have a major role in maintenance of respectively, transcription activation and repression, when associated to the chromatin. The importance of PcGs has motivated us to pursue the isolation and characterization of two new human proteins that are orthologs of the \"Enhancer of Polycomb\" (Epc) of Drosophila. To achieve this goal we undertook the task of the cloning and mapping of complete cDNA sequence of the novel genes hEPC1 and hEPC2, analyzing its expression in fetal, adult and tumoral tissues and functionally characterizing the hEPC2 protein. In 2001, we published the mapping and cloning of the complete cDNA sequences of both genes, as being orthologs of the mouse Epc1 (10p11-22) and Epc2 (2q21-23), together with the strategy used to obtain the full-length cDNAs (Camargo et al., 2001). Both genes are shown to be highly conserved among several species. Thus, the human hEPC2 cDNA is 94% identical to the mouse Epc2 and displays 96% identity at the protein level, suggesting maintenance of its function during the evolution. However, the protein sequences of the human hEPC1 and hEPC2 display only 68% identity. Therefore, it is likely that they have undergone a functional divergence after their duplication. The expression of both genes was evaluated using \"dot-blots\" containing 76 mRNAs samples from fetal, adult and tumoral tissues and is shown to be weak and ubiquitous. \"In silico\" analysis suggested the existence of 4 hEPC2 splicing isoforms that were validated by RT-PCR and/or Northern-blots. One of the isoforms (of 2.7 Kbp) is shown to be more abundant in all of the tumoral cell lines evaluated using Northern-blot analysis, mainly in the Burkit\'s Raji lymphoma and in the promyelocytic leukemia HL-60. This isoform results from the use of an alternative polyadenylation site that reduces the 3\'UTR, abolishing 4 of 5 \"adenylates and urilates rich elements\" (AREs), involved in the degradation of labile mRNAs that codify to regulatory proteins. These results have been recently submitted to publication (manuscript attached to this thesis). Interaction between the hEPC2/SMADs and its modulation by TGF-β. During the assembly of the hEPC2 full-length cDNA sequence, we found two patented ESTs that tagged a portion of the gene. These sequences were described as partial sequences of a \"new SMAD interacting protein\", involved in signal transduction of TGF-β, a cytokine that regulates cell proliferation, differentiation and death. To evaluate this putative interaction between hEPC2 and the SMADs proteins, we begun a collaboration with the TGF-β signalling group of the Dr. Aristidis Moustakas, from the Uppsala Ludwig Institute for Cancer Research, Sweden. The results of co-imunoprecipitation assays suggested that SMADs 2, 3, 4, 7 e 8 interact with hEPC2. Moreover, the interaction among SMAD2, SMAD3, SMAD4 and hEPC2 in cells treated with TGF-β1 showed decreased co-imunoprecipitation. This result suggests that TGF-β1 negatively modulates the interaction of these proteins. Likewise, we observed a reduction in hEPC2 interaction with SMAD8 upon BMP-7 treatment. This effect was even more dramatic for SMADs 2 and 3. These data were observed for both hEPC2 plasmid constructs, which strongly suggest the veracity of these proteins interaction. The cell localization of the hEPC2 protein, as well as its co-localization with the SMAD2, were investigated through indirect immunofluorescence assay, confirming the predicted localization of hEPC2 in the cell nucleus using the PSORTII program. However, we were not able to confirm the co-localization of hEPC2 and SMAD2. It is possible that hEPC2 does not bind directly to the DNA, requiring an association with a partner such as a transcription factor. This raises the hypothesis of hEPC2 having a role as a co-factor associated to one of the SMADs and binding to a \"SMAD binding element\" (SBE). To investigate this hypothesis, gene reporter assays were undertaken using a reporter construct containing 12 CAGA sequence repetitions (specific binding sequence of the SMADs 2, 3 and 4) fused to the luciferase gene. However, this assay could not demonstrate that the transcription of the SMAD is dependent on hEPC2. This experiment must be repeated. To confirm the interaction of hEPC2 and SMADs, a pull-down assay will be performed. To this end, the coding region of hEPC2 was cloned into the pET-32A bacterial inducible expression vector. The recombinant protein was already produced, having been induced and purified under denaturing conditions. Despite the dozens of PcG genes that were described in Drosophila, only a few of these genes have been characterized in mammals. Therefore, the description of the hEPC2 and its alternative transcripts is a contribution to better knowledge of the human PcGs. Regarding the hEPC2 and SMADs interaction, it\'s it is noteworthy that this is the first protein described to interact with SMADs of distinct categories. This may be an important indication of a unique role for hEPC2 in the TGF-β1 signaling pathway.

Des études présentées dans cette thèse ont permis de démontrer que le gène du groupe Polycomb (PcG) Bmi1 est essentiel à l’auto-renouvellement des progéniteurs rétiniens immatures et pour le développement rétinien après la naissance. Ce travail illustre chez l’embryon que Bmi1 est hautement enrichie dans une sous-population de progéniteurs rétiniens exprimant le marqueur de surface SSEA-1 et différents marqueurs de cellules souches. À tous les stades de développement analysés, l’absence de Bmi1 résulte en une diminution de la prolifération et de l’auto-renouvellement des progéniteurs immatures. Pour mieux comprendre la cascade moléculaire en absence de Bmi1, nous avons inactivé p53 dans les colonies Bmi1-/-. Cette inactivation a permis une restauration partielle du potentiel d’auto-renouvellement. De plus, en absence de Bmi1, la prolifération et la maintenance de la population de progéniteurs rétiniens immatures localisés dans le corps ciliaire sont aussi affectées après la naissance. Bmi1 permet donc de distinguer les progéniteurs immatures de la population principale de progéniteurs, et est requis pour le développement normal de la rétine. Nous avons également démontré que l’oncogène Bmi1 est requis dans les neurones pour empêcher l’apoptose et l’induction d’un programme de vieillissement prématuré, causé par une baisse des défenses anti-oxydantes. Nous avons observé dans les neurones Bmi1-/- une augmentation des niveaux de p53, de la concentration des ROS et de la sensibilité aux agents neurotoxiques. Nous avons démontré ainsi que Bmi1 contrôle les défenses anti-oxydantes dans les neurones en réprimant l’activité pro-oxydante de p53. Dans les neurones Bmi1-/-, p53 provoque la répression des gènes anti-oxydants, induisant une augmentation des niveaux de ROS. Ces résultats démontrent pour la première fois que Bmi1 joue un rôle critique dans la survie et le processus de vieillissement neuronal. / The studies presented in this thesis establish that the Polycomb Group (PcG) gene Bmi1 is required for the self-renewal of immature retinal progenitor cells (RPCs) and for postnatal retinal development. Work performed in mouse embryos reveals that Bmi1 is highly enriched in a RPC subpopulation expressing the cell surface antigen SSEA-1 and different stem cell markers. Furthermore, at all developmental stages analysed, Bmi1 deficiency resulted in reduced proliferation and self-renewal of immature RPCs. To better understand the molecular cascade leading to this phenotype, we inactivated p53 in Bmi1-deficient colonies. p53 inactivation partially restored RPCs self-renewal potential. Moreover, the proliferation and the postnatal maintenance of an immature RPC population located in the ciliary body was also impaired in absence of Bmi1. Thus, Bmi1 distinguishes immature RPCs from the main RPC population and is required for normal retinal development. We have also shown that the oncogene Bmi1 is required in neurons to prevent apoptosis and the induction of a premature aging-like program characterized by reduced antioxidant defenses. We observed in Bmi1-deficient neurons an increased p53 and ROS levels, and a hypersensitivity to neurotoxic agents. We demonstrated that Bmi1 regulate antioxidant defenses in neurons by suppressing p53 pro-oxidant activity. In Bmi1-/- neurons, p53 induces antioxidant genes repression, resulting in increased ROS levels. These findings reveal for the first time the major role of Bmi1 on neuronal survival and aging.

The epithelial to mesenchymal transition (EMT) is a highly conserved cellular program that allows well-­‐differentiated epithelial cells to convert to motile mesenchymal cells. EMT is critical for appropriate embryogenesis and plays a crucial role in tumorigenesis and cancer progression. At this regard, it has become increasingly evident that, in addition to genetic alterations, tumour development involves the alteration of gene expression patterns owing to epigenetic changes. Taking this into account, this thesis mainly addresses the description of new molecular epigenetic mechanisms underlying one of the hallmark processes governing EMT, the Snail1-­‐mediated E-­‐cadherin repression. Indeed, our results demonstrate that both Polycomb group (PcG) proteins and the LOXL2 protein are involved in this process. Apart from providing novel insights into the significance of these proteins in tumor progression, our work uncovers the characterization of a new epigenetic modification carried out by LOXL2; H3K4 deamination. / La transició epiteli-­‐mesènquima (EMT) és un programa cel·lular molt conservat que permet a les cèl·lules epitelials convertir-­‐se en cèl·lules mesenquimals indiferenciades. La EMT és un procés crucial pel desenvolupament embrionari i per la progressió tumoral. A aquest respecte, ha esdevingut cada cop més evident que el desenvolupament tumoral no només està associat a alteracions genètiques, sinó també a l'alteració de l’expressió gènica causada per canvis epigenètics. Tenint això en compte, aquesta tesi es centra en la descripció de nous mecanismes moleculars en l’àmbit de l’epigenètica associats a un dels processos clau en la EMT, la repressió de la E-­‐ cadherina mitjançada pel factor de transcripció Snail1. De fet, els nostres resultats demostren que tant les proteïnes del grup Polycomb (PcG) com la proteïna LOXL2 estan implicades en aquest procés. A part de proporcionar nova informació respecte la importància d'aquestes proteïnes en la progressió tumoral, la nostra feina ha permès la caracterització d'una nova modificació epigenètica duta a terme per la proteïna LOXL2; la deaminació de H3K4.

A identidade na segmentação do corpo de diversos organismos, durante o desenvolvimento, é devida, em grande parte, à ação das proteínas homeóticas. Em especial, dois grupos de proteínas, Trithorax (trxG) e Polycomb (PcG) têm um papel fundamental na manutenção, respectivamente, da ativação e da repressão da transcrição gênica, associando-se à cromatina. A importância das PcG nos estimulou a buscar a caracterização das proteínas humanas ortólogas ao \"Enhancer of Polycomb\" (Epc) de Drosophila, até então não descritas no genoma humano. Para tanto, buscamos: - obter a sequência completa e mapear o cDNA do novo gene humano homólogo ao \"Enhancer of Polycomb\" de Drosophila; - analisar sua expressão em tecidos fetais, adultos e tumorais e fazer estudos buscando sua caracterização funcional. Encontramos, mapeamos e obtivemos a seqüência completa de dois genes humanos, ortólogos de Epc1 (10p11-22) e de Epc2 (2q21-23) de camundongo, publicando estes dados em 2001 (Camargo et al., 2001). Ambos os genes são bastante conservados entre várias espécies, sendo que o cDNA de hEPC2 humano, por exemplo, é 94% idêntico ao Epc2 de camundongo e possui 96% de identidade ao nível de proteína, sugerindo que a função do gene deve ter sido mantida durante a evolução. No entanto, as seqüências protéicas de hEPC1 e hEPC2 humanos possuem apenas 68% de identidade entre si. Portanto, é provável que após a duplicação dos parálogos, estes tenham divergido funcionalmente. A expressão de ambos os genes foi avaliada utilizando \"dot-blots\" contendo 76 mRNAs de amostras de tecidos fetais, adultos e tumorais, mostrando-se fraca e ubíqua. Análises in silico sugeriram a existência de 4 isoformas de splicing para hEPC2, as quais foram validadas por RT-PCR ou \"Northern blots\". Uma das isoformas (de 2.7 Kpb) se mostrou mais abundante em todas as linhagens tumorais estudadas através de análises de \"Northern blot\", principalmente nas linhagens de linfoma de Burkitt\'s Raji e na linhagem de leucemia pró-mielocítica HL-60. Esta isoforma é gerada através de um sítio alternativo de poli-adenilação, que reduz sua porção 3\'UTR, retirando 4 dos 5 \"elementos ricos em adenilatos e uridilatos\" (AREs), envolvidos com a degradação de mRNAs lábeis que codificam proteínas regulatórias. Estes resultados se encontram em um manuscrito recentemente submetido à publicação (anexo à tese). Interação entre hEPC2 e SMADs e sua modulação por TGF-β. Durante a montagem da seqüência completa de hEPC2, verificamos que duas ESTs patenteadas mostravam alta identidade com o gene. Estas seqüências foram descritas como sendo parte de uma nova proteína de interação com as proteínas da família SMAD, envolvidas com transdução de sinais desencadeados por TGF-β. Esta citocina por sua vez, regula a proliferação, diferenciação e morte celular. Partimos para a avaliação da possível interação entre hEPC2 e as SMADs, em colaboração com o grupo do Dr. Aristidis Moustakas, do Ludwig Institute for Cancer Research de Uppsala, Suécia. Os resultados de co-imunoprecipitação sugeriram que as SMADs 2, 3, 4, 7 e 8 interagem com hEPC2, sendo que a interação entre as SMAD2, SMAD3, SMAD4 e hEPC2 nas células tratadas com TGF-β1, mostraram uma redução na co-imunoprecipitação. Este resultado sugere que TGF-β1 modula negativamente a interação entre essas proteínas. Da mesma maneira, foi observada uma redução na interação de hEPC2 com SMAD8 após o tratamento com BMP-7. Esse resultado é ainda mais destacado para as SMADs 2 e 3. Estes dados foram observados para ambas as construções de hEPC2, o que sugere fortemente a veracidade da interação entre estas proteínas. A localização celular de hEPC2, e também sua co-localização com SMAD2 foram investigadas através de imunofluorescência indireta e confirmaram a predição do programa PSORTII, de que hEPC2 se localiza no núcleo. No entanto, não foi possível observar a co-localização entre hEPC2 e SMAD2. É possível que hEPC2 não se ligue diretamente ao DNA, necessitando se associar como parceiro de um fator de transcrição. Esta foi uma das hipóteses para a atuação de hEPC2, como um co-fator que se associe com uma das SMADs e se ligue a um elemento específico de ligação a SMAD (SBE). Para investigar essa hipótese um ensaio de gene repórter foi feito utilizando uma construção de um repórter contendo 12 repetições da seqüência CAGA (seqüência específica de ligação das SMADs 2,3 e 4) fusionado com o gene da luciferase. No entanto, este ensaio não demonstrou que a transcrição de SMAD2 é dependente de hEPC2 e o experimento deverá ser repetido. Para confirmar a interação entre hEPC2 e as SMADs, será feito um experimento de \"pull-down\". Para tal o cDNA de hEPC2 foi clonado no vetor pET-32A de expressão indutível em bactérias. A proteína recombinante já foi produzida, tendo sido induzida e posteriormente purificada em condições desnaturantes. Apesar de dezenas de genes PcG terem sido caracterizados em Drosophila, poucos destes genes foram estudados em mamíferos. Portanto, a descrição do gene hEPC2 e seus transcritos alternativos, contribui para o conhecimento de PcG humanos, indicando a associação de maior expressão de uma de suas isoformas em linhagens celulares tumorais. Em relação à interação de hEPC2 com as SMADs, é interessante observar que nenhuma outra proteína foi descrita por possuir a particularidade de interagir com as SMADs de diferentes categorias. Talvez este seja um dado importante, que indique o papel singular de hEPC2 na sinalização de TGF-β1. / The identity of body segmentation in several organisms during development is, to a large extent, due to the action of the homeotic proteins. In particular, two groups of proteins, the Trithorax (trxG) and Polycomb (PcG), have a major role in maintenance of respectively, transcription activation and repression, when associated to the chromatin. The importance of PcGs has motivated us to pursue the isolation and characterization of two new human proteins that are orthologs of the \"Enhancer of Polycomb\" (Epc) of Drosophila. To achieve this goal we undertook the task of the cloning and mapping of complete cDNA sequence of the novel genes hEPC1 and hEPC2, analyzing its expression in fetal, adult and tumoral tissues and functionally characterizing the hEPC2 protein. In 2001, we published the mapping and cloning of the complete cDNA sequences of both genes, as being orthologs of the mouse Epc1 (10p11-22) and Epc2 (2q21-23), together with the strategy used to obtain the full-length cDNAs (Camargo et al., 2001). Both genes are shown to be highly conserved among several species. Thus, the human hEPC2 cDNA is 94% identical to the mouse Epc2 and displays 96% identity at the protein level, suggesting maintenance of its function during the evolution. However, the protein sequences of the human hEPC1 and hEPC2 display only 68% identity. Therefore, it is likely that they have undergone a functional divergence after their duplication. The expression of both genes was evaluated using \"dot-blots\" containing 76 mRNAs samples from fetal, adult and tumoral tissues and is shown to be weak and ubiquitous. \"In silico\" analysis suggested the existence of 4 hEPC2 splicing isoforms that were validated by RT-PCR and/or Northern-blots. One of the isoforms (of 2.7 Kbp) is shown to be more abundant in all of the tumoral cell lines evaluated using Northern-blot analysis, mainly in the Burkit\'s Raji lymphoma and in the promyelocytic leukemia HL-60. This isoform results from the use of an alternative polyadenylation site that reduces the 3\'UTR, abolishing 4 of 5 \"adenylates and urilates rich elements\" (AREs), involved in the degradation of labile mRNAs that codify to regulatory proteins. These results have been recently submitted to publication (manuscript attached to this thesis). Interaction between the hEPC2/SMADs and its modulation by TGF-β. During the assembly of the hEPC2 full-length cDNA sequence, we found two patented ESTs that tagged a portion of the gene. These sequences were described as partial sequences of a \"new SMAD interacting protein\", involved in signal transduction of TGF-β, a cytokine that regulates cell proliferation, differentiation and death. To evaluate this putative interaction between hEPC2 and the SMADs proteins, we begun a collaboration with the TGF-β signalling group of the Dr. Aristidis Moustakas, from the Uppsala Ludwig Institute for Cancer Research, Sweden. The results of co-imunoprecipitation assays suggested that SMADs 2, 3, 4, 7 e 8 interact with hEPC2. Moreover, the interaction among SMAD2, SMAD3, SMAD4 and hEPC2 in cells treated with TGF-β1 showed decreased co-imunoprecipitation. This result suggests that TGF-β1 negatively modulates the interaction of these proteins. Likewise, we observed a reduction in hEPC2 interaction with SMAD8 upon BMP-7 treatment. This effect was even more dramatic for SMADs 2 and 3. These data were observed for both hEPC2 plasmid constructs, which strongly suggest the veracity of these proteins interaction. The cell localization of the hEPC2 protein, as well as its co-localization with the SMAD2, were investigated through indirect immunofluorescence assay, confirming the predicted localization of hEPC2 in the cell nucleus using the PSORTII program. However, we were not able to confirm the co-localization of hEPC2 and SMAD2. It is possible that hEPC2 does not bind directly to the DNA, requiring an association with a partner such as a transcription factor. This raises the hypothesis of hEPC2 having a role as a co-factor associated to one of the SMADs and binding to a \"SMAD binding element\" (SBE). To investigate this hypothesis, gene reporter assays were undertaken using a reporter construct containing 12 CAGA sequence repetitions (specific binding sequence of the SMADs 2, 3 and 4) fused to the luciferase gene. However, this assay could not demonstrate that the transcription of the SMAD is dependent on hEPC2. This experiment must be repeated. To confirm the interaction of hEPC2 and SMADs, a pull-down assay will be performed. To this end, the coding region of hEPC2 was cloned into the pET-32A bacterial inducible expression vector. The recombinant protein was already produced, having been induced and purified under denaturing conditions. Despite the dozens of PcG genes that were described in Drosophila, only a few of these genes have been characterized in mammals. Therefore, the description of the hEPC2 and its alternative transcripts is a contribution to better knowledge of the human PcGs. Regarding the hEPC2 and SMADs interaction, it\'s it is noteworthy that this is the first protein described to interact with SMADs of distinct categories. This may be an important indication of a unique role for hEPC2 in the TGF-β1 signaling pathway.

Les protéines du groupe Polycomb (PcG) sont des protéines essentielles et conservées, qui forment deux complexes principaux, PRC1 et PRC2, qui sont recrutés au niveau de la chromatine et qui répriment stablement l’expression génique. Chez Drosophila melanogaster, les complexes Polycomb sont recrutés à des éléments d’ADN appelés éléments de réponse Polycomb (PREs) pour réprimer la transcription. PREs sont des éléments mémoires permutables qui peuvent maintenir la répression ou l’expression génique. Malgré des dizaines d’années d’étude, des questions fondamentales sur le fonctionnement du système PcG subsistent. 1) Comment les protéines PcG sont recrutées aux PREs uniquement lors du contexte développemental approprié, et comment les PREs peuvent conduire à la fois à l’activation et à la répression stable. 2) Comment les complexes PcG répriment la transcription, et si cela implique de nouvelles activités biochimiques et interactions. 3) Comment la répression dépendante des PcG peut-elle être propagé à travers le cycle cellulaire. La recherche de nouvelles activités biochimiques pour les complexes PcG pouvant répondre à ces questions fait l’objet de cette thèse. Les PREs sont transcrits en ARN qui pourraient donner la spécificité de contexte pour recruter les protéines PcG. Nous avons supposé que des R-loops puissent se former aux PREs, et être reconnues par les complexes Polycomb, ce que vous avons testé dans le chapitre 2. Nous avons identifié les séquences formant des R-loops dans des embryons et une lignée cellulaire de Drosophila melanogaster, et nous avons trouvé que ~30% des PREs forment des R-loops. Nous avons découvert que les PREs ayant formé des R-loops ont une plus forte probabilité d’être liés par les protéines PcG in vivo et in vitro. PRC2 lie des milliers d’ARN in vivo, mais aucune fonction claire n’y a été associée. En utilisant des expériences in vitro, nous avons identifié une activité d’invasion de brins pour PRC2 qui induit la formation d’hybride ARN-ADN, la partie principale d’une R-loop. Dans ce chapitre, nous avons trouvé que les PREs forment des R-loops, et sont impliquées dans le recrutement des protéines PcG qui induisent la répression génique stable. Nous avons découvert une activité d’invasion de brins pour PRC2 qui pourrait impliquer ce complexe Polycomb dans la formation de R-loops in vivo. Dans le chapitre 3, nous avons identifié une activité similaire à celle de la topoisomérase I associée avec PRC1 et la région C-terminale de sa sous-unité PSC (PSC-CTR). PRC1 et PSC-CTR peuvent relaxer un plasmide surenroulé négativement et ajouter des supertours négatifs à un plasmide relaxé en absence de topoisomérase. Cette activité suggère que la régulation de la topologie de l’ADN puisse être un nouveau mécanisme utilisé par les complexes PcG. PRC1 peut résoudre les R-loops formées sur un ADN négativement surenroulé in vitro. Une fonction possible pour cette activité de topoisomérase peut être la régulation des R-loops, dont la stabilité dépend à la fois de la séquence d’ADN et de la topologie de l’ADN environnant, in vivo. Dans cette thèse, nous avons identifié de nouvelles activités chez les complexes PcG : une activité d’invasion de brins pour PRC2 et une activité similaire à celle des topoisomérases pour PRC1. Ces deux activités impliquent les complexes PcG dans la formation et la résolution de R-loops. De plus, ces deux complexes peuvent reconnaitre les R-loops et sont recrutés aux PREs ayant formé ces structures. En conclusion, nous avons identifié de nouvelles activités pour les complexes Polycomb PRC1 et PRC2 qui les lient à la formation, la reconnaissance et la résolution de R-loops. / Polycomb group (PcG) proteins are conserved, essential proteins, which assemble in two main complexes, PRC1 and PRC2, which are targeted to chromatin and stably repress gene expression. In Drosophila melanogaster, Polycomb complexes are targeted to DNA elements called Polycomb response elements (PREs) to repress transcription. PREs are switchable memory elements that can maintain either gene repression or gene activation. Despite decades of study, fundamental questions about how the PcG system functions remain. These include: 1) how PcG proteins are targeted to PREs only in the appropriate developmental context, and how PREs can mediate both stable activation and repression; 2) how PcG complexes repress transcription, and whether it involves novel biochemical mechanisms and interactions; 3) how PcG repression can be propagated through cell cycles. The search for new biochemical activities for PcG complexes that may answer these questions is the topic of this thesis. PREs are transcribed into RNAs which may give the context specificity to recruit PcG proteins. We hypothesized that R-loops may form at PREs, and be recognized by PcG complexes, which we tested in Chapter 2. We identified R-loop forming sequences in Drosophila melanogaster embryos and tissue culture cells, and found that ~30% of the PREs form R-loops. We found that PREs which have formed R-loops are more likely to be bound by PcG proteins both in vivo and in vitro. PRC2 binds to thousand RNA in vivo but no clear activity has been associated with it. Using in vitro assays, we identified a strand exchange activity for PRC2 which induces the formation of RNA-DNA hybrid, the main part of an R-loop. In this chapter, we have found that PREs form R-loops and are involved in the targeting of PcG proteins which induce stable gene repression. We have discovered an RNA strand exchange activity for PRC2 which may involve this Polycomb complex in the formation of R-loops in vivo. In Chapter 3, we identified a type I topoisomerase-like activity associated with PRC1 and the C-terminal region of its subunit PSC (PSC-CTR). PRC1 and PSC-CTR can relax a negatively supercoiled plasmid and add negative coils to a relaxed plasmid in absence of topoisomerase. This activity suggests regulation of DNA topology may be a novel mechanism used by PcG complexes. PRC1 can resolve R-loops formed on negatively supercoiled DNA in vitro. One role for the topoisomerase-like activity may be to regulate R-loops, whose stability of depends on both the DNA sequence and the topology of the surrounding DNA, in vivo. In this thesis, we identified new activities for Polycomb group complexes: an RNA strand exchange activity for PRC2 and a topoisomerase-like activity for PRC1. Both activities link PcG complexes to the formation and resolution of R-loops. In addition, both complexes can recognize R-loops and are recruited to PREs which have formed these structures. In conclusion, we have identified new nucleic acid-based activities for the Polycomb complexes PRC1 and PRC2, which link them to the formation, recognition and resolution of R-loops.

Les modifications post-traductionnelles telles que la phosphorylation, l’OGlcNAcylation et l’ubiquitination jouent des rôles critiques dans la coordination des fonctions protéiques et par conséquent influencent grandement de nombreux processus cellulaires. Il est à noter que ces modifications sont hautement dynamiques et finement regulées. Par exemple, l’ubiquitination peut être réversible via l’action des déubiquitinases comme le suppresseur de tumeurs BAP1. Parmis les gènes codant pour les déubiquitinases, BAP1 est la plus souvent mutée dans le cancer. Des études récentes ont démontré l’importance des dynamiques de modifications post-traductionnelles dans la régulation du complexe BAP1. En plus, BAP1 forme un complexe multi-protéiques contenant plusieurs régulateurs transcriptionnels comme la protéine polycomb OGT et les facteurs de transcription FOXK1 et FOXK2. OGT est une enzyme unique qui catalyze l’ajout d’un groupement O-GlcNAc sur ses substrats afin d’en moduler l’activité enzymatique, les interactions protéines-protéines et leur localisation cellulaire. Cette modification est aussi liée au métabolisme puisque son substrat donneur, l’UDP-GlcNAc, est dérivé de la voie biosynthétique des hexosamines. Parallèlement, FOXK1/2 ont aussi été démontrés comme étant critiques à des processus métaboliques telles que la myogenèse et l’autophagie. Lors de nos études, nous avons identifié FOXK1 comme un nouveau substrat d’OGT. De plus, les niveaux d’O-GlcNAcylation de FOXK1 fluctuent lors de l’entrée/sortie du cycle cellulaire. En outre, nous avons identifié l’importance de FOXK1 dans l’adipogenèse et observé que l’interaction FOXK1/BAP1 est affectée par le métabolisme cellulaire. En résumé, nos études ont révélé l’importance d’OGT dans la régulation de certaines composantes du complexe BAP1, ce qui aidera à la compréhension de l’effet suppresseur de tumeur de BAP1 ainsi que son mécanisme d'action dans différents processus tel que le remodelage de la chromatine. / Post-translational modifications such as phosphorylation, O-GlcNAcylation and ubiquitination play critical roles in coordinating protein function and are therefore involved in diverse cellular processes. Of relevance here, ubiquitination may be removed by deubiquitinases such as the tumour suppressor BAP1, which represents the most mutated deubiquitinase gene in the human genome. Recent studies have revealed that important and dynamic post-translational modifications regulate several functions of the BAP1 complex. Indeed, BAP1 has been shown to form a multi-protein complex with several transcriptional regulators including the polycomb group protein OGT and the transcription factors FOXK1 and FOXK2. OGT is a unique enzyme that catalyzes the addition of an O-GlcNAc moiety to target proteins, which impacts protein function including enzymatic activity, protein-protein interactions and subcellular localization. This modification is also highly linked to cellular metabolism, as the donor substrate for the reaction, UDP-GlcNAc, is derived from the hexosamine biosynthesis pathway. Similarly, FOXK1 and FOXK2 have been shown to be implicated in metabolic processes such as myogenesis and autophagy. During our studies, we identified FOXK1 but not FOXK2 as a novel substrate of OGT. Further, we found that this OGlcNAcylation is modulated during the entry/exit of cell cycle. We also found that FOXK1 is critical for adipogenesis and that the interaction between FOXK1/BAP1 is compromised during nutrient starvation. Thus, our studies have revealed that OGT selectively modulates and regulates components of the BAP1 complex which may impact different cellular processes, notably chromatin remodelling and could help understanding how BAP1 acts as a tumor suppressor.

Mouse embryonic stem cells (mESCs) are an excellent model to study epigenetics and chromatin structure, owing to their self-renewal capabilities and tolerance of dynamic changes to DNA and histone modifications. Culturing conditions impact on the ability of mESCs to effectively recapitulate in vivo developmental states, and this is exemplified by refined culture conditions (termed 2i) that promote a pluripotent ground state. 2i-cultured mESC populations are homogeneous, naïve, and distinct from conventional (serum/LIF-cultured) cells, which exist as a metastable population. Remarkably, 2i-cultured mESCs also display global DNA hypomethylation, with methylation patterns more comparable to the cells of the E3.5 pre-implantation blastocyst. This is distinct from conventional serum-cultured cells, which display DNA methylation profiles that resemble later-stage E6.5 post-implantation epiblasts. The ability to transition between 2i- and serum-culture states is an attractive model for studying the dynamic role of DNA methylation in a variety of processes. DNA hypomethylation has been linked with depletion of the Polycomb-mediated repressive histone mark H3K27me3 from its normal target loci. Polycomb repressive complexes (PRC1 and PRC2) are important developmental regulators that maintain the repression of lineage-specific genes through generating compact higher-order chromatin structures. Polycomb target sites are primarily unmethylated CpG islands (CGIs). However, under conditions of DNA hypomethylation, new (previously methylated) binding sites are unveiled, and Polycomb is redistributed from its normal CGI target regions to intragenic regions. Thus, shifting mESCs to ground state conditions results in both DNA methylation and Polycomb patterns that are quite distinct from their serum-cultured counterparts. In my PhD, I sought to investigate the effect of DNA hypomethylation and Polycomb redistribution on higher-order chromatin structure in the ground state. I used a targeted, single-locus approach (FISH) as well as a genome-wide approach (Hi-C) to analyse differences in chromatin structure between conventionally cultured and ground state mESCs. My work suggests that chromatin structure is globally altered in hypomethylated 2icultured mESCs, with a similar state present in E3.5 mouse blastocysts. Using mESC lines in which DNA methylation levels can be directly manipulated, I was able to dissect the molecular mechanism driving higher-order structure changes in 2i medium, and showed the importance of DNA methylation in directing Polycomb-mediated chromatin compaction. My results may be important in considering the impact of DNA-methylation mediated reprogramming in multiple developmental, disease and regenerative medicine contexts.

La formation des trois feuillets embryonnaires primordiaux et leur régionalisation selon les axes embryonnaires sont des étapes cruciales au cours du développement précoce. Dans ce contexte, nous avons montré que XDSCR6, un inducteur du mésoderme et des axes embryonnaires qui présente des propriétés dorsalisantes, interagit physiquement et fonctionnellement avec le facteur de transcription XSTAT3. Au cours des étapes précoces du développement, XSTAT3 est active pendant la gastrulation et l'activation anormale de cette protéine dans la région dorsale induit la ventralisation des tissus embryonnaires. Par ailleurs, nous avons montré que XDSCR6 et XSTAT3 présentent des rôles antagonistes in vivo au cours de la mise en place des axes embryonnaires. Cet antagonisme peut être expliqué par le fait que XDSCR6 régule négativement l'activité transcriptionnelle de XSTAT3 en modulant sa méthylation sur les résidus lysine. L'ensemble de nos résultats a permis de déterminer l'importance cruciale de cette modification post-traductionnelle dans les propriétés ventralisantes de XSTAT3. Par ailleurs, nous avons montré que la méthyltransférase XEZH2 méthyle et active XSTAT3 in vivo. Des travaux antérieurs de l'équipe avaient montré que les propriétés dorsalisantes de XDSCR6 reposent sur sa capacité à inhiber l'activité épigénétique répressive de XEZH2 sur les gènes du mésoderme dorsal. Ainsi, nos travaux suggèrent que XDSCR6 est un modulateur transcriptionnel situé à l'interface entre certains régulateurs chromatiniens et des facteurs de transcription pendant la mise en place des axes embryonnaires. / One of the most challenging questions in developmental biology is to understand how a totipotent zygote differentiates into an organism containing all cell lineages. The formation of the three germ layers and the establishment of embryonic axis are fundamental events during early development. In this context, we demonstrated that XDSCR6, a mesoderm and embryonic axis inducer that exhibits dorsalizing properties, physically and functionally interacts with the transcriptional factor XSTAT3. During early development, XSTAT3 is active throughout gastrulation step and its abnormal activation in dorsal region leads to embryonic tissues ventralization. Furthermore, we showed that XDSCR6 and XSTAT3 have antagonistic roles in vivo during axis formation. This antagonism can be explained by the fact that XDSCR6 negatively regulates the transcriptional activity of XSTAT3 by interfering with its methylation on lysine residues. Moreover, this post-translational modification plays a crucial role in the ventralization abilities of XSTAT3. In a previous study, it has been shown that XDSCR6 negatively regulates the XEZH2 repressive epigenetic activity on dorsal mesoderm genes. Thus, we propose that XDSCR6 is a transcriptional modulator acting between epigenetic regulators and transcriptional factors during embryonic axis formation.

<p>In multicellular organisms all cells in one individual have an identical genotype, and yet their bodies consist of many and very different tissues and thus many different cell types. Somehow there must be a difference in how genes are interpreted. So, there must be signals that tell the genes when and where to be active and inactive, respectively. In some instances a specific an expression pattern (active or inactive) is epigenetic; it is established and maintained throughout multiple rounds of cell divisions. In the developing <i>Drosophila</i> embryo, the proper expression pattern of e.g. the homeotic genes <i>Abd-B</i> and <i>Ubx</i> is to be kept active in the posterior part and silenced in the anterior. Properly silenced homeotic genes are crucial for the correct segmentation pattern of the fly and the Polycomb group (Pc-G) proteins are vital for maintaining this type of stable repression.</p><p>As part of this thesis, <i>Suppressor of zeste 12 (Su(z)12)</i> is characterized as a <i>Drosophila</i> Pc-G gene. Mutations in the gene cause widespread misexpression of several homeotic genes in embryos and larvae. Results show that the silencing of the homeotic genes <i>Abd-B</i> and <i>Ubx</i>, probably is mediated via physical binding of SU(Z)12 to Polycomb Response Elements in the BX-C. <i>Su(z)12</i> mutations are strong suppressors of position-effect-variegation and the SU(Z)12 protein binds weakly to the heterochromatic centromeric region. These results indicate that SU(Z)12 has a function in heterochromatin-mediated repression, which is an unusual feature for a Pc-G protein. The structure of the <i>Su(z)12</i> gene was determined and the deduced protein contains a C2-H2 zinc finger domain, several nuclear localization signals, and a region, the VEFS box, with high homology to mammalian and plant homologues. <i>Su(z)12 </i>was originally isolated in a screen for modifiers of the zeste-white interaction and I present results that suggests that this effect is mediated through an interaction between <i>Su(z)12 </i>and <i>zeste</i>. I also show that <i>Su(z)12</i> interact genetically with other Pc-G mutants and that the SU(Z)12 protein binds more than 100 euchromatic bands on polytene chromosomes. I also present results showing that SU(Z)12 is a subunit of two different E(Z)/ESC embryonic silencing complexes, one 1MDa and one 600 kDa complex, where the larger complex also contains PCL and RPD3. </p><p>In conclusion, results presented in this thesis show that the recently identified Pc-G gene, <i>Su(z)12</i>, is of vital importance for correct maintenance of silencing of the developmentally important homeotic genes.</p>

In multicellular organisms all cells in one individual have an identical genotype, and yet their bodies consist of many and very different tissues and thus many different cell types. Somehow there must be a difference in how genes are interpreted. So, there must be signals that tell the genes when and where to be active and inactive, respectively. In some instances a specific an expression pattern (active or inactive) is epigenetic; it is established and maintained throughout multiple rounds of cell divisions. In the developing Drosophila embryo, the proper expression pattern of e.g. the homeotic genes Abd-B and Ubx is to be kept active in the posterior part and silenced in the anterior. Properly silenced homeotic genes are crucial for the correct segmentation pattern of the fly and the Polycomb group (Pc-G) proteins are vital for maintaining this type of stable repression. As part of this thesis, Suppressor of zeste 12 (Su(z)12) is characterized as a Drosophila Pc-G gene. Mutations in the gene cause widespread misexpression of several homeotic genes in embryos and larvae. Results show that the silencing of the homeotic genes Abd-B and Ubx, probably is mediated via physical binding of SU(Z)12 to Polycomb Response Elements in the BX-C. Su(z)12 mutations are strong suppressors of position-effect-variegation and the SU(Z)12 protein binds weakly to the heterochromatic centromeric region. These results indicate that SU(Z)12 has a function in heterochromatin-mediated repression, which is an unusual feature for a Pc-G protein. The structure of the Su(z)12 gene was determined and the deduced protein contains a C2-H2 zinc finger domain, several nuclear localization signals, and a region, the VEFS box, with high homology to mammalian and plant homologues. Su(z)12 was originally isolated in a screen for modifiers of the zeste-white interaction and I present results that suggests that this effect is mediated through an interaction between Su(z)12 and zeste. I also show that Su(z)12 interact genetically with other Pc-G mutants and that the SU(Z)12 protein binds more than 100 euchromatic bands on polytene chromosomes. I also present results showing that SU(Z)12 is a subunit of two different E(Z)/ESC embryonic silencing complexes, one 1MDa and one 600 kDa complex, where the larger complex also contains PCL and RPD3. In conclusion, results presented in this thesis show that the recently identified Pc-G gene, Su(z)12, is of vital importance for correct maintenance of silencing of the developmentally important homeotic genes.