Les tumeurs corticosurrénaliennes unilatérales sont fréquentes (prévalence de 2 à 9% de la population). Il s’agit le plus souvent d’adénomes. Le cancer de la corticosurrénale, ou corticosurrénalome, a un pronostic sombre, la survie ne dépassant pas 40 % à cinq ans. Le diagnostic de malignité de ces tumeurs, actuellement basé sur l’histologie, peut être difficile. Le pronostic des corticosurrénalomes est hétérogène et peu prévisible. Enfin, la prise en charge thérapeutique est encore limitée. Ces difficultés diagnostiques, pronostiques et thérapeutiques s’expliquent entre autres par la connaissance limitée de la physiopathologie de ce cancer. Le génome a un rôle central dans le développement des cancers en général. La survenue d’altérations génomiques (mutations, anomalies de nombre de copies, pertes d’hétérozygotie, translocations, anomalies de méthylation) va aboutir à la surexpression d’oncogènes et à la répression de gènes suppresseurs de tumeurs. La génomique, en ouvrant la voie à la caractérisation moléculaire à l’échelle du génome entier des tumeurs, est devenue incontournable pour étudier la physiopathologie des cancers. Les études de transcriptome des tumeurs corticosurrénaliennes ont montré que le profil d’expression génique discrimine adénomes et corticosurrénalomes, et identifie deux groupes distincts de corticosurrénalomes avec des pronostics différents. Dans le sillage de ces travaux précédemment réalisés par l’équipe, je me suis intéressée pendant ma thèse aux anomalies de nombres de copies d'ADN et aux anomalies de méthylation des corticosurrénalomes. Dans une première partie j’ai étudié le génome de 38 adénomes et 21 corticosurrénalomes par puce d’hybridation génomique comparative (puces CGH). Le transcriptome de 54 de ces tumeurs était déjà disponible. Le génome des corticosurrénalomes est très altéré contrairement à celui des adénomes (44% du génome est perdu ou gagné, versus 10% pour les adénomes, p=2.10-10). Dans les adénomes, la région 9q34 (locus de SF-1) est fréquemment gagnée et ce gain est associé à une surexpression de SF-1. Pour les corticosurrénalomes, les évènements récurrents concernent les gains des chromosomes 5, 7, 12, 16, 19, 20 et les pertes des chromosomes 13 et 22. Les gènes situés dans les régions minimales communes gagnées ou perdues ont été filtrés en fonction de leur expression. La liste des gènes à la fois gagnés et surexprimés inclut des oncogènes comme FGFR4, CDK4, CCNE1 ; et la liste des gènes avec perte de matériel et sous-expression inclut des gènes suppresseurs de tumeurs (LATS2, ST13). Un outil diagnostique basé sur la mesure en PCR quantitative de 6 loci permet de séparer les corticosurrénalomes des adénomes dans une cohorte de validation indépendante de 79 tumeurs, avec une sensibilité de 100% et une spécificité de 83%. Le nombre d’altérations chromosomiques n’a pas de valeur pronostique, mais une technique de classification hiérarchique non supervisée permet de séparer les corticosurrénalomes en deux groupes de pronostic différent, et a été validée sur une cohorte indépendante de 25 tumeurs. Dans une deuxième partie, j’ai étudié les anomalies de méthylation des promoteurs des gènes de 51 corticosurrénalomes et 81 adénomes par puce Infinium HumanMethylation27 (Illumina). Les données d’expression étaient disponibles pour 87 tumeurs. Les corticosurrénalomes sont globalement hyperméthylés par rapport aux adénomes. La classification hiérarchique non supervisée sépare les corticosurrénalomes en 3 groupes : un groupe non-hyperméthylé, un groupe modérément hyperméthylé, et un autre très hyperméthylé. Cette classification a été confirmée par MS-MLPA. L’hyperméthylation est associée à un mauvais pronostic (p=0,02 en modèle de Cox). La corrélation entre niveau de méthylation et expression identifie 1741 gènes (sur les 12250 étudiés) corrélés négativement (…) / Unilateral adrenocortical tumors are common (prevalence : 2 to 9% of the population). Most of these tumors are adenomas. Adrenocortical cancer (ACC) has a poor prognosis, with a 5-yr survival rate not exceeding 40% in most series. Pathological diagnosis of these tumors relies on several histological features and can be difficult. The prognosis of adrenocortical carcinomas is heterogeneous and unpredictable. Knowledge of the pathophysiology of these tumors is also limited. The genome has a central role in the development of cancers in general. The occurrence of genomic alterations (mutations, abnormal copy number, loss of heterozygosity, translocations, abnormal methylation) will lead to the overexpression of oncogenes and repression of tumor suppressor genes. Genomic approaches became essential to study the pathophysiology of cancer. Transcriptome studies of adrenocortical tumors have shown that the gene expression profile discriminate ACC and adenomas, and identified two distinct groups of ACC with different prognoses. In addition to gene expression, genomics now covers a large spectrum of alterations, including chromosomal alterations, DNA sequence modifications, and epigenetic alterations. I studied chromosomal alterations and DNA methylation abnormalities in ACC during my thesis.In the first part the genome of 38 adrenocortical adenomas and 21 ACC were identified by comparative genomic hybridization arrays. The transcriptome of 54 of these tumors was already available. A larger proportion of the genome is altered in carcinomas compared with adenomas (44 % of the genome is lost or gained, versus 10% for adenomas , p = 2.10-10 ) . In adenomas, the 9q34 region, which includes the steroidogenic factor 1locus, is commonly gained and associated with an overexpression of steroidogenic factor 1 (SF-1).In carcinomas, recurrent gains include chromosomes 5, 7, 12, 16, 19, and 20 and recurrent losses chromosomes 13 and 22. Filtering the genes from these regions according to their expression profile identified genes potentially relevant to adrenocortical tumorigenesis. A diagnostic tool was built by combining DNA copy number estimates at six loci (5q, 7p, 11p, 13q, 16q, and 22q). This tool discriminates carcinomas from adenomas in an independent validation cohort (sensitivity 100%, specificity 83%). In carcinomas, the number of chromosomal alterations was not associated with survival (Cox p=0.84). A prognostic tool based on tumor DNA was designed with a clustering strategy and validated in an independent cohort. In the second part, methylation patterns of CpG islands in promoter regions of 51 adrenocortical carcinomas and 84 adenomas were studied by the Infinium HumanMethylation27 Beadchip (Illumina) . Gene expression data were available for 87 tumors. Methylation was higher in carcinomas than in adenomas (t test : p=3.1x10-9). Unsupervised clustering of DNA methylation profiles identified two groups of carcinomas, one with an elevated methylation level, evoking a CpG island methylator phenotype (CIMP). The subgroup of hypermethylated carcinomas was further divided in two subgroups, with different levels of methylation (CIMP-high and CIMP-low). This classification could be confirmed by methylation-specific multiplex ligation-dependent probe amplification. Hypermethylation was associated with a poor survival (Cox model p= 0.02). The transcriptome/methylation correlation showed 1741 genes (of 12250) negatively correlated; among the top genes were H19 and other tumor suppressors (PLAGL-1, G0S2, and NDRG2). The subgroups identified by the transcriptome adrenocortical have different levels of methylation. In conclusion, genomic alterations discriminate carcinomas from adenomas and contain prognostic information. The subgroups identified by the adrenocortical transcriptome profiles of different genomic alterations. Chromosomal alterations and abnormal methylation alter the expression of genes important for tumorigenesis.

Brain tumours account for the most frequent type of solid tumours among children. Despite advances in surgery and chemotherapy, brain tumours are still the main cause of cancer deaths in children. Furthermore, little is known about DNA methylation changes in paediatric astrocytoma. Recent investigations suggest that many tumours are initiated not only by genetic abnormalities, but also caused by epigenetic changes. DNA methylation is a key epigenetic mechanism that controls the regulation of gene expression. Interestingly, unlike DNA mutations, epigenetic abnormalities are reversible. The reversibility of epigenetic abnormalities upon pharmacological unmasking has prompted interest in developing epigenetic therapy with the crucial goal of restoring the expression of aberrantly silenced genes. The focus of this study was to utilise a combination of different microarray strategies to develop an integrative candidate gene approach to identify several novel frequently methylated genes in a cohort of paediatric HGA (High grade glioma) samples. In addition, to investigate the potential of therapeutic efficacy of a DNA methyltransferase inhibitor, 5-Aza-dC in paediatric HGA. There were 147 genes commonly identified to be potentially methylated in IN699 cells using the two different array strategies integration; re-expression array and Illumina Infinium Human Methylation 450k array. Furthermore, using two complementary microarray strategies including methylation 450k array and expression array, this work identified 55 genes that were both methylated and under-expressed in these HGA cultures. Following validation with CoBRA and RT-PCR coupled with the response of hypermethylated promoters to the demethylating agent 5-Aza-dC, six novel genes (CXCL14, PRR5L, ELTD1, ITGA2, KRT8 and NTM) that are frequently silenced in paediatric astrocytoma were identified. This study suggests that re-expression of ii CXCL14 inhibited the colony formation and cell growth and reduces the migration rate significantly in IN699 short term culture and likely have functional significance in the development of paediatric HGA and an excellent candidate gene for further analysis. In parallel, the efficacy of 5-Aza-dC treatment was examined in paediatric HGA aiming to introduce this epigenetic therapy as a potential mechanism in management of this tumours. This study demonstrated that, relatively low dose of 5-Aza-dC sharply reduced the colony formation and inhibited proliferation and not through the apoptotic effect. It is likely that this reduction in proliferation without cell death is due to using relatively low doses that do not acutely kill cells, thus, allow the sustained alterations in both gene expression patterns and appearance of a new phenotype to emerge. Taken together, this work contributes to a more detailed understanding of the effect of epigenetic silencing on paediatric HGA. This investigation also demonstrated the use of epigenetic drug, 5-aza-dC to reverse the gene silencing for the potential treatment of paediatric HGA.

La chromatine est le support de l’information génétique et sa structure, ainsi que son activitétranscriptionnelle, peuvent être modulées par des modifications épigénétiques. Le maintien des marquesrépressives telles que la méthylation de l’ADN et des histones, hors du corps des gènes, est nécessairepour le bon développement de la plante. Chez Arabidopsis thaliana, le mutant sg1 présente des défautsdéveloppementaux sévères caractéristiques de mutants affectés dans des mécanismes épigénétiques. Nousavons montré que le phénotype de sg1 est causé par une hyperméthylation CHG et H3K9me2 dans denombreux gènes. En effet, SG1 contrôle la transcription de l’histone déméthylase IBM1 et lesmodifications de l’épigénome observées chez sg1 sont dues à une dérégulation de IBM1. Nous avonsidentifié sept protéines partenaires de SG1, dont certaines se lient aux marques chromatiniennes. Nousavons réalisé un crible suppresseur qui a permis d’identifier FPA, une protéine régulant lapolyadénylation de certains transcrits, comme acteur impliqué dans le contrôle des cibles de SG1, dontIBM1. Nos résultats montrent que le complexe SG1 régule la transcription de ses cibles en influençant,par un mécanisme encore inconnu, le choix du site de polyadénylation, en lien avec les marqueschromatiniennes présentes aux locus cibles. D’autre part, certaines épimutations induites par la mutationsg1 peuvent être maintenues pendant plusieurs générations. Pour rechercher un lien entre méthylation desgènes et conséquences phénotypiques, nous avons caractérisé des épimutations liées à un défaut dedéveloppement de la fleur et identifié un certain nombre de gènes candidats potentiellement responsablesdu phénotype. Les résultats obtenus au cours de ma thèse ont contribué à préciser le rôle joué par lecomplexe SG1 et à comprendre le lien entre celui-ci et les marques épigénétiques. / Chromatin is known to contain the genetic information and its structure and transcriptionalstate can be regulated by epigenetic modifications. Repressive marks such as DNA and histonesmethylation needs to be kept away from gene bodies to enable the proper development of the plant. InArabidopsis thaliana, sg1 mutants show a range of severe developmental defects similar to thoseobserved in mutants affected in epigenetic pathways. We have shown that sg1 mutant phenotype iscaused by an increase of CHG and H3K9me2 methylation in many gene bodies. Indeed, SG1 regulatesthe histone demethylase IBM1 transcription and the impairment observed in sg1 mutant epigenomes iscaused by IBM1 misregulation. We found seven proteins interacting with SG1, among which somepartners are able to bind chromatin marks. Through a suppressor screen we identified FPA, alreadyknown to regulate the polyadenylation of some transcripts, as a player involved in SG1 targetsregulation, including IBM1. Our results show that the SG1 complex regulates target genes transcriptionby affecting polyadenylation site choice, in a way that remains to be determined, in a chromatin marksdependent manner. We also found that some of the sg1-induced epimutations can be maintained throughseveral generations. To investigate the link between gene body methylation and phenotypicconsequences, we have characterized epimutations related to a defect in floral development andidentified some candidate genes potentially responsible for the floral phenotype. Thus, our resultscontributed to clarify the role of SG1 and to understand its connection with epigenetic marks.

DNA methylation and histone modifications have emerged as key mechanisms in transcriptional regulation. The target of methylation-induced silencing 1 (TMS1) is a bipartite protein. Recent studies have indicated that methylation-associated silencing of TMS1 occurs in many cancers. However, whether and how TMS1 is regulated by epigenetic mechanisms in cancers remains unknown. In this study we showed that methylation of the TMS1 promoter occurred in five of six hepatocellular carcinoma (HCC) cell lines. TMS1 expression was reduced in four HCC cell lines and correlated with methylation status. Furthermore, the TMS1 promoter was completely methylated and mRNA expression was undetectable. TMS1 expression could be restored by 5-aza-2′-deoxycitidine (5-Aza-dC) (a DNA methyltransferase inhibitor) or trichostatin A (TSA) (a histone deacetylase inhibitor) alone and the promoter methylation. was partially reversible. TSA was more efficient than 5-Aza-dC in inducing TMS1 expression, and the combination of 5-Aza-dC and TSA resulted in markedly synergistic reactivation of the gene and completely reversed promoter methylation. Interestingly, TMS1 promoter methylation-associated gene silencing was accompanied by histone H3 Lysine 9 (H3K9) hypoacetylation and trimethylation. 5-Aza-dC and/or TSA also had some effect on conversion of methylated to acetylated H3K9 in restoring TMS1. This conversion was dynamic at the TMS1 promoter and a decrease in H3K9 trimethylation preceded an increase in H3K9 acetylation after 5-Aza-dC and/or TSA treatment. Our results thus suggest that epigenetic inactivation of TMS1 expression is regulated by promoter hypermethylation and H3K9 modifications in a coordinated way.

Indiana University-Purdue University Indianapolis (IUPUI) / CXXC finger protein 1 (CFP1) is an epigenetic regulator of H3K4 and cytosine methylation. Due to its role in establishing and maintaining methylation patterns, CFP1 determines whether DNA is found in its euchromatin or heterochromatin state and as such whether genes are transcriptionally active or inactive. In stem cells, deficiency of CFP1 results in inability to differentiate and in murine embryos it results in periimplantation death. Despite the demonstrated importance in developing tissue, the role of CFP1 in mature tissues, such as the liver, has yet to be elucidated. This study examined the role of CFP1 in maintaining liver homeostasis under conditions involving hepatocellular stress by examining liver regeneration, pregnancy-induced hepatomegaly, and non-alcoholic steatohepatitis (NASH) disease progression. The liver’s ability to recover was analyzed through liver:body mass ratios, blood serum analysis, liver histology, and qualitative observations. Deficiency of CFP1 in the livers of animals subjected to partial hepatectomies (PH) resulted in decreased liver regeneration capacity with liver mass restoration becoming significantly different starting at 48H post-PH and remaining so until 10D post-PH. This decreased regeneration appeared to be the result of reduced hepatocyte mitosis. Mouse dams lacking hepatic CFP1 mated with males expressing CFP1 displayed a proclivity for dystocia. Mice subjected to a fast food diet resulting in NASH while lacking hepatic CFP1 experienced decreased weight gain and hepatic lipid accumulation compared to their CFP1 expressing counterparts. Through these three studies, the critical role of CFP1 for the maintenance of liver homeostasis was demonstrated.

京都大学 / 0048 / 新制・課程博士 / 博士(医科学) / 甲第21028号 / 医科博第89号 / 新制||医科||6(附属図書館) / 京都大学大学院医学研究科医科学専攻 / (主査)教授 斎藤 通紀, 教授 浅野 雅秀, 教授 玉木 敬二 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

No description available.

DNA methylation plays a key role in disease analysis, especially for studies that compare known large scale differences in CpG sites, such as cancer/normal studies or between-tissues studies. However, before any analysis can be done, data normalization and preprocessing of methylation data are required. A useful data preprocessing pipeline for large scale comparisons is Functional Normalization (FunNorm), (Fortin et al., 2014) implemented in the minfi package in R. In FunNorm, the univariate quantiles of the methylated and unmethylated signal values in the raw data are used to preprocess the data. However, although FunNorm has been shown to outperform other preprocessing and data normalization processes for these types of studies, it does not account for the correlation between the methylated and unmethylated signals into account; the focus of this paper is to improve upon FunNorm by taking this correlation into account. The concept of a bivariate quantile is used in this study as an attempt to take the correlation between the methylated and unmethylated signals into consideration. From the bivariate quantiles found, the partial least squares method is then used on these quantiles in this preprocessing. The raw datasets used for this research were collected from the European Molecular Biology Laboratory - European Bioinformatics Institute (EMBL-EBI) website. The results from this preprocessing algorithm were then compared and contrasted to the results from FunNorm. Drawbacks, limitations and future research are then discussed. / Thesis / Master of Science (MSc)

No description available.

DNA methylation is an epigenetic mechanism critical for tissue development, cell specification and cellular function. Mammalian brains consist of millions to billions of neurons and glial cells that can be subdivided into many distinct types of cells. We hypothesize that brain methylomes are heterogeneously methylated across different types of cells and the transcription factors play key roles in brain methylome programming. To dissect brain methylome heterogeneity, in Chapter 2, we first focused on the identification of cell-subset specific methylated (CSM) loci which demonstrate bipolar DNA methylation pattern, i.e., hypermethylated in one cell subset but hypomethylated in others. With the genome-scale hairpin bisulfite sequencing approach, we demonstrated that the majority of CSM loci predicted likely resulted from the methylation differences among brain cells rather than from asymmetric DNA methylation between DNA double strands. Importantly, we found that putative CSM loci increased dramatically during early stages of brain development and were enriched for GWAS variants associated with neurological disorder-related diseases/traits. It suggests the important role of putative CSM loci during brain development, implying that dramatic changes in functions and complexities of the brain may be companied by a rapid change in epigenetic heterogeneity. To explore epigenetic regulatory mechanisms during brain development, as described in Chapter 3, we adopted unbiased data-driven approaches to re-analyze methylomes for human and mouse frontal cortices at different developmental stages. We predicted Egr1, a transcriptional factor with important roles in neuron maturation, synaptic plasticity, long-term memory formation and learning, plays an essential role in brain epigenetic programming. We performed EGR1 ChIP-seq and validated that thousands of EGR1 binding sites are with cell-type specific methylation patterns established during postnatal frontal cortex development. More specifically, the CpG dinucleotides within these EGR1 binding sites become hypomethylated in mature neurons but remain heavily methylated in glia. We further demonstrated that EGR1 recruits a DNA demethylase TET1 to remove the methylation marks at EGR1 binding sites and activate downstream genes. Also, we found that the frontal cortices from the knockout mice lacking Egr1 or Tet1 share strikingly similar profiles in both gene expression and DNA methylation. Collectively, the study in this dissertation reveals EGR1 programs the brain methylome together with TET1 during postnatal development. This study also provides new insights into how life experience and neuronal activity may shape the brain methylome. / Ph. D. / DNA methylation is a widespread epigenetic mark on DNA, serving as a “switch” to turn on or off gene expression. It plays essential roles in cellular functions, tissue development. Mammalian brains contain millions to billions of neurons and glial cells, which can be further divided into many different types of cells. We hypothesize that brain cells have different methylation profiles across the genome, and transcriptional factors play important roles in programming methylation in the mammalian brain genome. To study the diversity of methylation profiles across the genomes of different brain cells, in Chapter 2, we first focused on the identification of cell-subset specific methylated (CSM) genomic regions which show bipolar DNA methylation pattern, i.e., hypermethylated in one type of cell but hypomethylated in others. By applying a technique called the genome-scale hairpin bisulfite sequencing to mouse frontal cortices, we demonstrated that the majority of CSM genomic regions predicted likely resulted from the methylation differences among brain cells, rather than from methylation differences between DNA double strands. Surprisingly, we found that these predicted CSM genomic regions increased dramatically during early stages of brain development and were enriched for GWAS variants associated with neurological disorder-related diseases/traits. It suggests the importance of predicted CSM genomic regions, implying that dramatic changes in brain function and structure may be companied by a rapid change in DNA methylation diversity during brain development. To explore underlying epigenetic mechanisms during brain development, as described in Chapter 3, we re-analyzed methylomes for human and mouse frontal cortices at different developmental stages, and predicted Egr1, a transcriptional factor with important roles in neuron maturation, synaptic plasticity, long-term memory formation and learning, plays an essential role in brain methylome programming. We found thousands of EGR1 binding sites showed cell-type specific methylation patterns, and were established during postnatal frontal cortex development. More specifically, the methylation level of these EGR1 binding sites was low in mature neurons but pretty high in glial cells. We further demonstrated that EGR1 recruits a DNA demethylase TET1 to remove the methylation marks at EGR1 binding sites and activate downstream genes. Also, we found that the frontal cortices from the Egr1 knockout or Tet1 knockout mice show strikingly similar profiles in both gene expression and DNA methylation. Collectively, the study in this dissertation reveals EGR1 works together with TET1 to program the brain methylome during postnatal development. This study also provides new insights into how life experience and neuronal activity may shape the brain methylome.