BACKGROUND: Medulloblastoma (MB) is the most common malignant brain tumor in children, accounting for 15-20% of all pediatric brain tumors. In patients with MB, prognosis depends heavily on the molecular makeup of the tumor. New genomic approaches over the last decade have enabled researchers to sub-classify MB based on differences in the transcriptome: WNT, Sonic hedgehog (SHH), Group 3 (MYC-amplified), and Group 4 (heterogeneous). SHH tumors represent a third of all MB cases, and small-molecule inhibitors have already been developed that target SHH signaling. Most notably, vismodegib has shown great promise in the treatment of MB and other SHH-driven cancers by targeting Smoothened (SMO), an upstream regulator of GLI activity. However, most patients who had initially responded to the drug quickly acquired point mutations in SMO that led to treatment resistance. In addition, patients who harbored mutations downstream of SMO had no response to treatment and were found to be intrinsically resistant. Although most patients with SHH-MB can be cured, current treatments often require broad base therapies, such as radiation and chemotherapy, which can have harmful and long-lasting side effects. These observations underscore the need for less toxic, more targeted therapies that act at the level of the GLI family of transcription factors themselves. However, as transcription factors are generally considered undruggable, Dr. Robbins’ group at the University of Miami Miller School of Medicine sought to address this need by using focused screens of siRNAs or small molecules that target epigenetic GLI regulators. They identified several candidates that act as readers, writers, and/or erasers of protein acetylation and methylation and showed that a subset of these candidates act downstream of SMO to attenuate GLI signaling (data not yet published). Bromodomain- and Plant Homeodomain-linked Zinc Finger-containing Protein 1 (BRPF1) was one of these candidates and further analysis revealed that its knockdown reduced Gli1 expression by more than 50%. Recent studies link BRPF1 to cerebellar development and tumor formation in SHH-MB and may be suggestive of its role as a negative regulator. OBJECTIVES: We sought to compare basal levels of Brpf1 expression in normal versus MB in mice; to characterize Brpf1 knockdown versus overexpression in SHH cell lines; and to determine if BRPF1 merits further investigation as a candidate for future drug targeting therapies in MB and other SHH-driven cancers. METHODS: We used RT-qPCR and immunoblotting analysis to look at Brpf1 expression in Ptch+/- and adult wild-type mice. cDNA and protein samples were donated by colleagues in the lab. We also grew and maintained SHH Light2 cells in culture and then used these cells to carry out siRNA and plasmid DNA transfections. RNA extraction, RT-PCR, and RT-qPCR were used to examine transfection efficiency and its effect on Gli1 expression. RESULTS: Brpf1 levels were higher in SHH-MB compared to normal cerebellum. However, BRPF1 proteins were not detected in either normal or tumor samples. Brpf1 knockdown in Light2 cells correlated with an overall decrease in Gli1 expression while overexpression had no obvious affect on Gli1 expression. CONCLUSIONS: Our findings suggest that BRPF1 may function as a positive regulator of GLI activity. Recent studies verify this claim at least partially stating that BRPF1 acts as both a positive and negative regulator of gene expression depending on the context. Thus, before we can draw any final conclusions, more research is needed to look at BRPF1 in the specific context of the SHH pathway and developing cerebellum.

La déficience intellectuelle (DI) est une trouble du neuro développement caractérisée par une extrême hétérogénéité génétique, avec plus de 700 gènes impliqués dans des formes monogéniques de DI. Cependant un nombre important de gènes restent encore à identifier et les mécanismes physiopathologiques de ces maladies neuro développementales restent encore à comprendre. Mon travail de doctorat a consisté à identifier de nouvelles causes génétiques impliquées dans la DI. En utilisant différentes techniques de séquençage de nouvelle génération, j’ai pu augmenter le taux de diagnostic chez les patients avec DI et identifié plusieurs nouvelles mutations (dans AUTS2, THOC6, etc) et nouveaux gènes (BRPF1, NOVA2, etc) impliqués dans la DI. Pour les moins caractérisés, j'ai effectué des investigations fonctionnelles pour valider leur pathogénicité, caractériser les mécanismes moléculaires qu'ils affectent et identifier leur rôle dans cette maladie. Mes travaux de doctorat permettront d’améliorer et d’accélérer la possibilité d’obtenir un diagnostic moléculaire qui donnera accès à un meilleur suivi et à une meilleure prise en charge pour les patients. Cela permettra également de mieux comprendre les mécanismes physiopathologiques impliqués dans ces troubles neuro développementaux. Ces connaissances aideront éventuellement à identifier de nouvelles cibles thérapeutiques. / Intellectual disability (ID) is a group of neurodevelopmental disorders characterized by an extreme genetic heterogeneity, with more than 700 genes currently implicated in Mendelian forms of ID but still some are not yet identified. My PhD project investigates the genetic causes of these monogenic ID by using and combining different NGS techniques. By using this strategy, I reached a relative high diagnostic yield and identified several novel mutations (in AUTS2, THOC6) and genes (BRPF1, NOVA2, etc) involved in ID. For the less characterized ones, I performed functional investigations to prove their pathogenicity, delineate the molecular mechanisms altered and identify their role in this disease. Overall, this work improved and provided new strategies to increase the molecular diagnosis in patients with ID, which is important for their healthcare and better management. Furthermore, the identification and the characterization of novel mutations and genes implicated in ID better delineate the implicated pathophysiological mechanisms, opening the way to potential therapeutic targets.

Tumors are composed of proliferating cells that invade healthy tissue and grow over time. Even though it is still unclear, it is a common opinion that the cells of origin should possess a proliferative capacity (Blanpain, 2013; Visvader, 2011). Particularly for brain cancers, the transition of neural progenitors to differentiated postmitotic neurons is considered irreversible in physiological and pathological conditions. Therefore, postmitotic neurons have not been considered as suitable cell of origin for brain cancer. Here, we show that neurons reprograming may occur upon Shh activation leading to medulloblastoma (MB) formation in vivo. Human SHH medulloblastoma (MB) is a brain tumor affecting adults and infants that is thought to originate from cerebellar granule neuron progenitors. Notably, several groups have shown that Shh pathway activation (SmoM2 overexpression) in mouse granule neuron progenitors is able to induce Shh MB (Schuller et al., 2008; Z.-J. Yang et al., 2008). These progenitors are present in infants and newborn mice, but they seem to be absent in adult humans and mice (Biran, Verney, & Ferriero, 2012; Marzban et al., 2014; Z.-J. Yang et al., 2008). Furthermore, it was recently discovered that the two different forms of SHH MB are distinguished by different transcriptome/methylome levels suggesting that the adult SHH MB may originate from a different cell of origin (Cavalli et al., 2017; Kool et al., 2014). Relying on these data, we take advantage of a conditional Cre-Lox recombination system to recapitulate the human adult medulloblastoma pathogenesis in mice, demonstrating that cerebellar post-migratory mature granule neurons upon SmoM2 overexpression can dedifferentiate and give rise to SHH MB in vivo. Moreover, human adult patients present inactivating mutations of the chromatin reader BRPF1 that are associated with SMO mutations and absent in pediatric and adolescent patients. Here we found that truncated BRPF1 protein, as found in human adult patients, is able to induce medulloblastoma in adult mice upon SmoM2 activation. Notably, gene expression profiling on our samples allowed to associate “cerebellar granule progenitors-derived MB” with the human infant form while “truncated BRPF1-induced tumors” clustered with human adult SHH MB. Furthermore, as previously described by Kool et al., 2014, human adult SHH MB is characterised by the copresence of p-AKT and p-S6, compared to the human infant SHH MB that are positive for either p- AKT or p-S6 and always in a mutually exclusive way. Truncated BRPF1-induced tumors are double positive for p-AKT and p-S6, similarly to adult patients, while cerebellar granule progenitors derived MB present only p-S6. Furthermore, to define the contribution of chromatin changes in granule neurons dedifferentiation in response to Shh activation, we profiled changes in chromatin accessibility by ATAC-seq analysis on mice cerebella. SmoM2 overexpression changed the epigenetic landscape of granule neurons, enriching the number of open chromatin regions 12 associated with stem/progenitor-like genes. Moreover, the cooperation between truncated BRPF1 and SmoM2 in reshaping the chromatin arrangement of granule neurons was explored applying ATAC-seq on differentiated human cerebellar neurons derived from neuroepithelial cells. ATAC-seq analysis pointed out a synergistic mechanism between SmoM2 and truncated BRPF1 in modifying the epigenetic landscape of postmitotic neurons, increasing the chromatin accessibility of super-enhancers, associated with stemness and chromatin organization/modification genes. Our novel model of cancer development could explain the human SHH medulloblastoma onset in adult individuals where granule neuron progenitors are no more present. For these reasons, we strongly believe that our model configures as an important starting point for a new field in cancer and stem cell biology focusing on the study of mechanisms driving tumorigenesis in postmitotic cells.