Neuroblastoma is a heterogeneous pediatric malignancy originating from the developing sympathetic nervous system, with poor long-term survival for high-risk patients (~40%). About half of advanced neuroblastomas harbor high-level amplification of the MYCN gene, and these tumors show few, if any, additional driver lesions. Despite significant increase in the body of knowledge of genetics in neuroblastoma, all the high-risk patients follow similar therapeutic procedures and little advancement has been made on molecular target based therapies. The major challenge is to dissect the complexity and heterogeneity of these tumors to find driver genes and activated pathways that are essential for the survival of these cancer cells. We used an integrated systems biology approach to define the core regulatory machinery responsible for maintenance of an aggressive neuroblastoma phenotypic state. In the first part of the thesis, I will discuss our computational approach to decipher the tumor heterogeneity by subtype classification, followed by identification of master regulator protein modules for three distinct molecular subtypes of high-risk neuroblastomas, which were validated in a large independent cohort of cases. We propose that such modules are responsible for integrating the effect of mutations in upstream pathways and for regulating the genetic programs and pathways necessary for tumor state implementation and maintenance. The second part of the thesis is focused on experimental validation of putative master regulators in the subtype of neuroblastomas associated with MYCN amplification. By using RNAi screening followed by experimental and computational analyses to elucidate the interdependencies between the top master regulators, we identified TEAD4-MYCN positive feedback loop as a major tumor maintenance mechanism in this subtype. While MYCN regulates TEAD4 transcriptionally, TEAD4 regulates MYCN through transcriptional and post-translational mechanisms. Jointly, MYCN and TEAD4 regulate 90% of inferred MR proteins and causally orchestrate 70% of the subtype-specific gene expression signature. TEAD4 gene expression was associated with neuroblastoma patient survival independently of age, tumor stage and MYCN status (P=2.1e-02). In cellular assays, MYCN promoted growth and repressed differentiation, while TEAD4 activated proliferation and DNA damage repair programs, the signature hallmarks of MYCN-amplified neuroblastoma cells. Specifically, TEAD4 was shown to induce MYCN-independent proliferation by transactivating key genes implicated in high-risk neuroblastoma pathogenesis, including cyclin-dependent kinases, cyclins, E2Fs, DNA replication factors, checkpoint kinases and ubiquitin ligases. The critical role of the core master regulator module in controlling tumor cell viability, both in vitro and in vivo, and its clinical relevance as a prognostic factor highlights TEAD4 as a novel and highly effective candidate target for therapeutic intervention. In this thesis, we demonstrate that interrogation of tumor specific regulatory networks with patient-derived gene expression signatures can effectively elucidate molecular subtypes as well as the core transcriptional machinery driving subtype specific hallmarks. This approach enables identification of oncogenic and non-oncogenic dependencies of high-risk neuroblastoma and is applicable to other tumor subtypes.

Neuroblastoma is a childhood solid tumor of a unique propensity to either regress spontaneously or grow relentlessly. Emerging evidence indicated that neuroblastoma contains heterogeneous populations of cells, and commitment of these cells to neuronal lineage may result in aggressive progression in patients, whereas to fibromuscular lineage may give a favorable outcome. However, mechanism(s) controlling the lineage commitment of neuroblastoma cells remains to be identified. Our preliminary data suggested that Kr?ppel-Like Factor 4 (KLF4) might promote neuroblastoma regression. KLF4 is a transcription factor regulating a variety of cellular functions, including proliferation and cell cycle progression. Recent studies have demonstrated that KLF4 may act as both tumor suppressor and oncogene in a cell-context dependent manner. Importantly, our preliminary data showed that low KLF4 expression is highly associated with poor clinical outcomes of the neuroblastoma patients. In addition, we found that overexpression of KLF4 suppresses neuroblastoma cell growth accompanied with loss of tumorigenicity. Morphologically, KLF4 overexpressing cells changed their morphologies to become epithelial-like, strongly substrate-adherent and expressing smooth muscle marker. Therefore, we hypothesized that KLF4 exerts its effects through two ways, it may (i) function to inhibit cell growth and reduce tumorigenicity; and (ii) promote differentiation of the neuroblastoma cells to the non-tumorigenic, fibromuscular-like cells. RT-PCR data revealed the differential expression of KLF4 in 11 neuroblastoma cell lines. In particular, a modest expression was found in Be(2)C, a cell line which was formerly demonstrated to differentiate and form tumor in mice xenograft assay. It was therefore chosen as the study model. To assess the effects of KLF4 knockdown on tumor growth, stable knockdown clones from Be(2)C cells were established by lentiviral transduction of KLF4-targeting shRNA. In parallel, clones that stably expressed non-target shRNA were used as controls. After the transduction, two stable knockdown clones showing significant KLF4 downregulation were isolated from single colony (monoclonal stable clones) and a pool of cells (polyclonal stable clones) respectively. The cell proliferation and growth rate of the stable clones were then measured by 5-bromo-2’-deoxyuridine (BrdU) proliferation assay and growth curve assay. The results have indicated that both monoclonal and polyclonal stable KLF4 knockdown clones grow faster than the control clones. In order to examine the tumorigenicity in vivo, the stable clones were xenotransplanted to severe combined immunodeficient mice. The stable KLF4 knockdown clones showed a significant higher growth rate and formed a larger tumor. The stable clones were also treated with BrdU for four weeks for differentiation towards fibromuscular lineage. As anticipated, the control clones showed fibromuscular features, like more flattened and epithelial-like morphology. In contrast, the stable KLF4 knockdown clones failed to present the fibromuscular features after treatment. In addition, immunocytochemistry staining of SMA and quantitative analysis of the immunocytochemistry further confirmed that only the control clones showed higher SMA expression after BrdU treatment, while there is no change in the SMA expression in the stable KLF4 knockdown clones. These results demonstrated that KLF4 functioned by inhibiting neuroblastoma cell proliferation and growth, reducing the tumorigenicity, and it was required for fibromuscular differentiation. / published_or_final_version / Surgery / Master / Master of Philosophy

Despite the identification of MYCN amplification as an adverse prognostic marker in neuroblastoma, no drugs that target MYCN have yet been developed. Here, by combining a whole genome shRNA library screen and Master Regulator Inference Algorithm (MARINa) analysis, we identified Transcription Factor Activating Protein 4 (TFAP4) as a novel synthetic lethal interactor with MYCN amplification in neuroblastoma. Silencing TFAP4 selectively inhibits MYCN amplified neuroblastoma growth both in vitro and in xenograft mice models. TFAP4 expression is inversely correlated with patient survival in MYCN-high neuroblastoma. Mechanistically, silencing TFAP4 induces neuroblastoma differentiation, as seen by increased neurite outgrowth, and up-regulation of neuronal markers. TFAP4 regulates a downstream signature similar to the signature of the oncogene anaplastic lymphoma kinase (ALK). Taken together, our results validate TFAP4 as an important master regulator in MYCN amplified neuroblastoma and a novel synthetic interactor with MYCN amplification. Thus, TFAP4 may be a novel drug target for neuroblastoma treatment.

The development of resistance to chemotherapeutic drugs is the main obstacle to the successful treatment of many cancers, including childhood neuroblastoma, the most common solid tumour of infants. One factor that may play a role in determining response of neuroblastoma tumours to therapeutic agents is the p53 tumour suppressor gene. A number of previous studies have suggested that this tumour suppressor protein is inactive in neuroblastoma due to its cytoplasmic sequestration. This thesis therefore has examined the functionality of p53 and its role in determining drug response of neuroblastoma cells. An initial study was undertaken that characterised an unusually broad multidrug resistance (MDR) phenotype of a neuroblastoma cell line (IMR/KAT100). The results demonstrated that the MDR phenotype of the IMR/KAT100 cells was associated with the acquisition of mutant p53. To explore the role of p53 in drug resistance further, p53-deficient variants in cell lines with wild-type p53 were generated by transduction of p53-suppressive constructs encoding either shRNA or a dominant-negative p53 mutant. Analysis of these cells indicated that: (i) in contrast to previous reports, wild-type p53 was fully functional in all neuroblastoma lines tested, as evidenced by its activation and nuclear translocation in response to DNA damage, transactivation of target genes and control of cell cycle checkpoints; (ii) inactivation of p53 in neuroblastoma cells resulted in establishment of an MDR phenotype; (iii) knockdown of mutant p53 did not revert the drug resistance phenotype, suggesting it is determined by loss of wild-type function rather than gain of mutant function; (iv) p53-dependent cell senescence, the primary response of S-type neuroblastoma cells to DNA damage, is replaced, after p53 inactivation, by mitotic catastrophe and subsequent apoptosis. In contrast to neuroblastoma, p53 suppression had no effect or increased drug susceptibility in several other tumour cell types, indicating the importance of tissue context for p53- mediated modulation of tumour cell sensitivity to treatment. Taken together, these data provide strong evidence for p53 having a role in mediating drug resistance in neuroblastoma and suggest that p53 status may be an important prognostic marker of treatment response in this disease.

The development of resistance to chemotherapeutic drugs is the main obstacle to the successful treatment of many cancers, including childhood neuroblastoma, the most common solid tumour of infants. One factor that may play a role in determining response of neuroblastoma tumours to therapeutic agents is the p53 tumour suppressor gene. A number of previous studies have suggested that this tumour suppressor protein is inactive in neuroblastoma due to its cytoplasmic sequestration. This thesis therefore has examined the functionality of p53 and its role in determining drug response of neuroblastoma cells. An initial study was undertaken that characterised an unusually broad multidrug resistance (MDR) phenotype of a neuroblastoma cell line (IMR/KAT100). The results demonstrated that the MDR phenotype of the IMR/KAT100 cells was associated with the acquisition of mutant p53. To explore the role of p53 in drug resistance further, p53-deficient variants in cell lines with wild-type p53 were generated by transduction of p53-suppressive constructs encoding either shRNA or a dominant-negative p53 mutant. Analysis of these cells indicated that: (i) in contrast to previous reports, wild-type p53 was fully functional in all neuroblastoma lines tested, as evidenced by its activation and nuclear translocation in response to DNA damage, transactivation of target genes and control of cell cycle checkpoints; (ii) inactivation of p53 in neuroblastoma cells resulted in establishment of an MDR phenotype; (iii) knockdown of mutant p53 did not revert the drug resistance phenotype, suggesting it is determined by loss of wild-type function rather than gain of mutant function; (iv) p53-dependent cell senescence, the primary response of S-type neuroblastoma cells to DNA damage, is replaced, after p53 inactivation, by mitotic catastrophe and subsequent apoptosis. In contrast to neuroblastoma, p53 suppression had no effect or increased drug susceptibility in several other tumour cell types, indicating the importance of tissue context for p53- mediated modulation of tumour cell sensitivity to treatment. Taken together, these data provide strong evidence for p53 having a role in mediating drug resistance in neuroblastoma and suggest that p53 status may be an important prognostic marker of treatment response in this disease.

The combinatorial code of cPcdh isoforms creates a diversified cell-surface molecular signature for cell-cell recognition in neural networks. This genetic architecture, combined with a regulated expression pattern and trans-homophilic binding properties, provides insights into cell specialization and signaling. Anomalies in cPcdhs, which include genetic mutations, epigenetic modifications, structural variations, and altered gene expression profiles, are associated with several neurological, neuropsychiatric, and systemic conditions, highlighting the importance of cPcdh investigations. This study focuses on the transcriptional regulation of the Pcdhα gene cluster. Each neuron displays a specific Pcdhα alternate exon repertoire, necessitating an understanding of the transcriptional dynamics. Using the SK-N-SH human neuroblastoma cell line and methodologies such as cRNA-seq and Start-Seq, these dynamics are examined. The application of CRISPR-Cas9 gene editing and a dCas9-VPR gain-of-function assay in the HEK293T cell line reveals the role of as-lncRNA and its interaction with DNA methylation within the Pcdhα gene cluster. This study identifies the role of noncoding as-lncRNA in RNA transcription and provides information on CTCF binding and Pcdhα promoter activation. The research also examines the gastrointestinal domain, as cPcdhs are linked to various diseases. Shifting focus from the canonical realm of the CNS, the research embarks on a preliminary yet pivotal exploration of the gastrointestinal domain. As cPcdhs intersect with a plethora of diseases, an incisive understanding of their expression could yield revelations into tissue susceptibilities with potential disease ramifications. Employing a novel single-domain antibody technique coupled with immunohistochemistry, the endeavor casts a precise lens into the gastrointestinal expression dynamics of Pcdhα and Pcdhγ. These insights not only fortify the understanding of cPcdh within neural structures but also beckon a deeper inquiry into their multifaceted biological roles.