p16INK4a (p16) traps the cell at the restriction point of the cell cycle by binding to cyclin-dependent kinase 4/6 thus preventing the phosphorylation of the retinoblastoma protein (pRB). As p16 accumulates the cell stops dividing and becomes senescent. This study investigates the modulation of p16 function by the putative membrane protein ABCB5 and a group of five putative oncogenic microRNAs (oncomiRs). ABCB5 is a poorly characterised member of the B-subfamily of human ATP Binding Cassette transporters. ABCB5 is reportedly transcribed into four transcripts, one of which could potentially encode a full-length transporter (ABCB5fl) whilst a second could encode a half-transporter (ABCB5β). The other two transcripts (ABCB5α and ABCB5γ) could only encode short polypeptides. Exogenous expression of ABCB5fl and ABCB5β was achieved in HEK293T cells, but the recombinant protein expressed poorly and localised to the endoplasmic reticulum. Point mutations introduced into the ATP catalytic domain failed to improve expression levels suggesting that protein function was not deleterious to the cell. Exogenous expression in HEK293T cells also allowed commercial antibodies purportedly raised against ABCB5 isoforms to be tested. Several were found not to recognise ABCB5 necessitating re-interpretation of published data. However, one antibody recognised both ABCB5fl and ABCB5β, and was subsequently used to evaluate protein expression levels in other cell types.siRNA knockdown of ABCB5 in human mammary epithelial cells (HMECs) caused a concomitant reduction in p16 expression and an increase in cellular proliferation. Differential siRNAs and RT-qPCR analyses demonstrated ABCB5β to be the relevant transcript with respect to the reduction in p16 expression; however, no native ABCB5β protein was detected in HMECs. Together these data lead to the hypothesis that the ABCB5β transcript may act as a long noncoding RNA to regulate p16. Exogenous expression of each of five distinct putative oncomiRs in HMECs was found to increase cellular proliferation and, surprisingly, increase p16 expression. These results mirror a phenotype commonly observed in p16-positive basal-like breast cancer (BLBC), an aggressive form of breast cancer with poor prognosis and few treatment options. Bioinformatic analysis of the predicted target genes for these oncomiRs identified multiple transcriptional regulators of pRB. These predictions, together with the work performed in a cellular model of p16-positive BLBC, suggest that the oncomiRs may cause unrestricted cell proliferation by indirectly reducing transcription of the pRB gene, RB1. In the absence of pRB, p16 expression is induced via a previously reported oncogeneinduced senescence-like positive feedback loop. These data, and previously published observations, suggest that a similar mechanism may explain the basis of p16-positive BLBC.

Changes in the epigenetic landscape are widespread in neoplasia, with de novo methylation and histone repressive marks commonly occurring in association with gene silencing. However, understanding the dynamics of epigenetic changes is often hindered due to the absence of adequate in vitro model systems that accurately reflect events occurring in vivo. Human mammary epithelial cells (HMECs) grown under standard culture conditions enter a growth arrest termed selection, but a subpopulation is able to escape from arrest and continue to proliferate. These cells, called post-selection cells, have many of the hallmarks seen in the earliest lesions of breast cancer, including transcriptional silencing and hypermethylation of the p16INK4A tumour suppressor gene. The overall aim of my thesis was to use post-selection HMECs as model system to identify and dissect the mechanism involved in early epigenetic aberrations. Firstly, using a microarray approach, I found that multiple members of the TGF-β signalling pathway were concordantly suppressed in post-selection cells, and this was associated with functional disruption of the TGF-β pathway. Interestingly, concordant gene suppression was not associated with aberrant DNA methylation, but with repressive chromatin remodelling. Secondly, to further understand the mechanism underpinning epigenetic silencing, I demonstrated using laser capture technology, that p16INK4A silencing is a precursor to DNA methylation and histone remodelling. Thirdly, I found that individual post-selection HMEC strains during the early passages shared a common 'wave' pattern of regional-specific methylation within the p16INK4A CpG island. Interestingly, the 'wave' pattern of early de novo methylation correlated with the apparent footprint of nucleosomes within the p16INK4A CpG island. Lastly, to further characterise the properties of the HMEC culture system, I demonstrated that post-selection cells do not possess a natural tumour-inducing property when transplanted into the mammary fat pad of immunocompromised mice. However, post-selection HMECs were associated with high expression of a variety of stem/progenitor markers, as well as stem/progenitor associated polycomb genes, thus demonstrating that these cells share some common features of stem/progenitor cells. The research presented in this thesis demonstrate that epigenetic changes occur early in the growth of post-selection HMECs and many of these changes are common in breast cancer.

Changes in the epigenetic landscape are widespread in neoplasia, with de novo methylation and histone repressive marks commonly occurring in association with gene silencing. However, understanding the dynamics of epigenetic changes is often hindered due to the absence of adequate in vitro model systems that accurately reflect events occurring in vivo. Human mammary epithelial cells (HMECs) grown under standard culture conditions enter a growth arrest termed selection, but a subpopulation is able to escape from arrest and continue to proliferate. These cells, called post-selection cells, have many of the hallmarks seen in the earliest lesions of breast cancer, including transcriptional silencing and hypermethylation of the p16INK4A tumour suppressor gene. The overall aim of my thesis was to use post-selection HMECs as model system to identify and dissect the mechanism involved in early epigenetic aberrations. Firstly, using a microarray approach, I found that multiple members of the TGF-β signalling pathway were concordantly suppressed in post-selection cells, and this was associated with functional disruption of the TGF-β pathway. Interestingly, concordant gene suppression was not associated with aberrant DNA methylation, but with repressive chromatin remodelling. Secondly, to further understand the mechanism underpinning epigenetic silencing, I demonstrated using laser capture technology, that p16INK4A silencing is a precursor to DNA methylation and histone remodelling. Thirdly, I found that individual post-selection HMEC strains during the early passages shared a common 'wave' pattern of regional-specific methylation within the p16INK4A CpG island. Interestingly, the 'wave' pattern of early de novo methylation correlated with the apparent footprint of nucleosomes within the p16INK4A CpG island. Lastly, to further characterise the properties of the HMEC culture system, I demonstrated that post-selection cells do not possess a natural tumour-inducing property when transplanted into the mammary fat pad of immunocompromised mice. However, post-selection HMECs were associated with high expression of a variety of stem/progenitor markers, as well as stem/progenitor associated polycomb genes, thus demonstrating that these cells share some common features of stem/progenitor cells. The research presented in this thesis demonstrate that epigenetic changes occur early in the growth of post-selection HMECs and many of these changes are common in breast cancer.

Le cancer du sein est une pathologie hétérogène au plan clinique et au moins 5 sous-types moléculaires ont pu être définis sur la base de différences d’expression ARNm. Ces sous-types présentent des différences de profils d’anomalies génomiques et de méthylation des cytosines. Ces différences génétiques et épigénétiques s’expliqueraient par des types cellulaires d’origines distincts au sein de l’épithélium mammaire, toutefois, ceci n’a pas été confirmé clairement à ce jour. Alternativement, il a été proposé que l’activation de voies oncogéniques différentes pouvait avoir un impact significatif sur les modifications génétiques ou épigénétiques. Dans ce travail nous avons voulu vérifier cette hypothèse en l’appliquant à un modèle de cellules épithéliales mammaires normales primaires humaines, que nous avons isolé des à partir de glandes mammaires. Ces cellules ont été transformées en deux étapes par transduction avec (i) un shARN ciblant TP53, (ii) un oncogène. Nous avons sélectionné 3 oncogènes qui activent des voies de signalisations distinctes CCNE1, HRAS-v12 et WNT1. Nous avons établi un modèle de transformation tumorale en trois étapes, cellules normales, immortalisées et transformées, permettant de suivre les modifications moléculaires associées à chaque étape et de vérifier si l’activation de voies oncogéniques distinctes produisait des profils d’anomalies différents. Les différents modèles ont été analysés par CGH-array, RRBS, transcriptome et miRNA à des temps de culture définis.Nos résultats montrent que l’activation de la voie RAS aboutit à des profils d’anomalies génétiques et de méthylation des CpG radicalement différents de ceux obtenus après surexpression des gènes CCNE1 et WNT1. Ces différences apparaissent très rapidement après transduction des oncogènes alors que les profils des cellules CCNE1 et WNT1 divergent plus tardivement. Enfin, l’inactivation de p53 n’induit pas par elle-même une instabilité élevée, mais produit un contexte de plasticité favorable aux modifications génétiques et épigénétique.Par ailleurs, nous avons noté des différences phénotypiques entre les HMEC RAS (mésenchymateuses) et les HMEC CCNE1 et les HMEC WNT1 (épithéliales). Dans ce travail, je montre que les HMEC shp53 immortalisées présentent une plasticité phénotypique, une partie des cellules entrant en EMT spontanément, l’autre restant épithéliales. J’ai montré que la transduction RAS sélectionnait les cellules ayant effectué une EMT, alors que la transduction de CCNE1 ou WNT1 sélectionnait les cellules épithéliales. J’ai cherché à identifier les déterminants de ces changements phénotypiques et mes résultats suggèrent qu’ils résultent d’une balance entre une signalisation TGFB1/BMP1, qui favorise l’EMT, et BMP4/WNT7 qui favorise la MET. / Breast cancer is a heterogeneous pathology. Based on the differences of mRNA expression, at least five molecular subtypes have been defined. These subtypes show differences in profiles of genomic abnormalities and CpG methylation. These molecular subtypes are thought to originate from different cell lineages in the mammary gland. However, this has not yet been clearly demonstrated. Alternatively, it has been proposed that the activation of different oncogenic pathways could have a significant impact on genetic or epigenetic changes.We wanted to verify this hypothesis by applying it to a normal primary human mammary epithelial cells (HMEC) model, which we isolated from normal mammary explants. These cells were transformed in two step process by sequential transduction of (i) a shRNA targeting TP53, (ii) an oncogene. We selected 3 oncogenes that activate distinct signaling pathways CCNE1, HRAS-v12 and WNT1. Our tumor transformation model was established in three-step, normal, immortalized and transformed cells, allowing us to monitor the molecular changes associated with each step and to verify whether the activation of distinct oncogenic pathways produced different profiles of genetic and epigenetic modifications. The different models were analyzed at defined culture times by CGH-array, RRBS, transcriptome and miRNA. Our results show that genetic abnormalities and CpG methylation profiles are different between cells where the RAS pathway was activated and cells overexpressing WNT1 or CCNE1. These differences appear rapidly after oncogene transduction, whereas the profiles of the CCNE1 and WNT1 cells diverged later. Finally, inactivation of p53 by itself does not induce high instability, but produces a context of plasticity favorable to genetic and epigenetic changes.In addition, we noted phenotypic differences between HMEC RAS (mesenchymal) and HMEC CCNE1 and HMEC WNT1 (epithelial). In this work, I show that the immortalized HMEC shp53 exhibit a phenotypic plasticity, where some cells enter a spontaneous EMT and the others remain epithelial. I showed that RAS transduction selected cells that are undergoing an EMT, whereas transduction with CCNE1 or WNT1 selected the epithelial cells. I have sought to identify the determinants of these phenotypic changes and my results suggest that a balance exists between TGFβ1 / BMP1 signaling, which promotes EMT, and BMP4 / WNT7a which promotes TEM.

No description available.

Rapidly growing tumor cells outgrow their blood supply resulting in a microenvironment with reduced oxygen and nutrients. Using an in vitro transformation model we found that cancer cells expressing the SV40 ST antigen (+ST cells) are more resistant to glucose deprivation-induced cell death than cells lacking the SV40 ST antigen (−ST cells). Mechanistically, we found that the ST antigen mediates this effect by activating a nutrient-sensing kinase, AMP-activated protein kinase (AMPK). We further show that AMPK mediates its effects, at least in part, by inhibiting mTOR (mammalian target of rapamycin), thereby shutting down protein translation, and by inducing autophagy as an alternate energy source. Resistance to anoikis upon anchorage-deprivation is yet another form of stress tolerated by both normal stem/progenitor cells of various tissues in our body and by cancer cells. Using mammospheres as a model to enrich for stem/progenitor cells we found that mammosphere formation is accompanied with increased activation of AMPK. Concomitant with AMPK activation, we detected increased phosphorylation of the anti-apoptotic protein PED/PEA15. We further demonstrate that AMPK directly interacts with and phosphorylates PEA15 at Ser116, thus establishing PEA15 as a new AMPK target. Thus, our study has identified AMPK-PEA15 signaling as a key component of sphere formation by both normal and cancerous breast tissues. During metastasis, epithelial cells lose attachments to their neighbors, acquire a mesenchymal-like morphology, a process termed as epithelial-mesenchymal transition (EMT) and become motile. Our results indicate that AMPK regulates EMT by both transcriptional and post-translational modification of EMT-inducing transcription factor, Twist. Thus, our study has identified a role for AMPK in nutrient deprivation, anchorage-independent growth, and epithelial-mesenchymal transition involved in metastasis. In addition, we have identified two novel substrates of AMPK, PEA15 and Twist, that may play key roles in cancer progression. Thus, our study suggests that targeting AMPK, or its newly identified substrates, can be explored as possible anti-cancer mechanisms.

No description available.