The Yes-associated protein 1 (YAP1) is an oncogenic transcriptional co-activator, which is negatively regulated by the Hippo signalling pathway. If the Hippo pathway is deregulated, YAP1 can translocate to the nucleus where it interacts with various transcription factors to drive transcription. Suppressing YAP1 as a therapeutic strategy has attracted considerable interest, especially since YAP1 and oncogenic RAS have been shown to interact in different tumour models. I evaluated the role(s) of YAP1 in transforming non-tumourigenic epithelial cells along the epithelial-mesenchymal transition (EMT) spectrum. Using a tetracycline-inducible expression system, I found that induced YAP1 overexpression in non-tumourigenic mouse Eph4 cells resulted in the upregulation of the mesenchymal markers, demonstrating a partial EMT. As a comparison, H-RAS overexpression in Eph4 cells resulted in E-CADHERIN relocalization away from the cell-cell junctions, also demonstrating a partial EMT. Co- expression of H-RAS and YAP1 resulted in a transition further along the EMT spectrum. However, YAP1 overexpression alone did not enhance cell migration or proliferation, whereas single overexpression of H-RAS did. Therefore, mutations which lead to overexpression of oncogenic RAS can be considered ‘driver’ mutations as they confer a significant tumourigenic potential to cancer cells. In contrast, although the upregulation of mesenchymal markers by YAP1 may also confer a survival advantage, YAP1 overexpression is not sufficient to trigger E-CADHERIN relocalization. Thus, mutations leading to YAP1 overexpression are neither ‘driver’ or ‘passenger’ mutations. Instead mutations leading to YAP1 overexpression are likely to represent an intermediate between a ‘driver’ and ‘passenger’ mutation. Hence I am referring to them as a ‘co-pilot’ of tumourigenesis.

Thousands of genetic and epigenetic variants have been identified for many common diseases including cancer through genome-wide association studies (GWAS) and epigenome-wide association studies (EWAS). To advance the complex interpretation of both GWAS and EWAS results, I developed new software tools (FORGE2 and eFORGE) for the analysis and interpretation of GWAS and EWAS data, respectively. Both tools determine the cell type-specific regulatory component of a set of target regions (either GWAS-identified genetic variants or EWAS-identified differentially methylated positions). This is achieved by detecting enrichment of overlap with histone mark peaks or DNase I hypersensitive sites across hundreds of tissues, primary cell types, and cell lines from the ENCODE, Roadmap Epigenomics, and BLUEPRINT projects. Application of both tools to publicly available datasets identified novel disease-relevant cell types for many common diseases, a stem cell-like signature in cancer EWAS, and also demonstrated the ability to detect cell-composition effects for EWAS performed on heterogeneous tissues. To complement these bioinformatics efforts and validate selected variants predicted by FORGE2, eFORGE and additional analyses, I performed conformation capture using 4C-seq to fine-map the 3D context of the genomic regions involved, uncovering novel interactions for autoimmunity-associated variants and IKZF3.

BACKGROUND: Adamantinomatous Craniopharyngiomas (ACPs) are clinically challenging sellar region tumours, known to be characterised by mutations in CTNNB1. ACPs are often histologically complex, with different morphological cell types and surrounded by a florid glial reaction. Murine models have been generated through activating β-catenin and support a critical role for nucleo-cytoplasmic accumulating β-catenin cell clusters (‘clusters’) in driving tumorigenesis. AIMS: To phenotype in detail the 3D growth patterns of human and murine ACP; To characterise the genomic and transcriptomic landscape of human and murine ACP, including of clusters; To characterise therapeutically targetable molecular pathways and perform pre-clinical therapeutic trials. METHODS: Human ACP samples underwent micro-focus-CT scanning, whole genome sequencing, targeted next generation sequencing and RNA sequencing, both with, and without, laser capture microdissection. The growth dynamics of murine ACP was characterised by serial MRI and a cohort of murine ACPs, at various stages, underwent RNA and exome sequencing. A pre-clinical murine trial using a Sonic Hedgehog (SHH) pathway inhibitor was performed. RESULTS: CTNNB1 mutationsin human ACP were confirmed as clonal within tumour epithelia. Gene expression signatures corresponding to tumour epithelia, reactive glia and immune infiltrate were derived and novel ACP genes were identified (e.g. BCL11B). A relationship between human and murine ACPs with the developing tooth was also established, in particular the similarity of clusters to the enamel knot. Further molecular dissection identified a complex interplay between tumour cell compartments demonstrating a role for paracrine signalling. Inhibition of the SHH pathway in the pre-clinical murine trial resulted in a decrease in median survival from 33 weeks to 11.9 weeks (p=0.048). A signature of inflammasome activation in ACP was also identified in solid and cystic components of ACP. CONCLUSIONS: ACPs have clonal mutations in CTNNB1 and exhibit complex signalling interplay between different cell compartments. Expression analysis reveals a new molecular paradigm for understanding ACP tumorigenesis as an aberrant copycat of natural tooth development, with inflammation driven by activation of inflammasomes. Caution is recommended in the use of SHH pathway inhibitors in patients with ACP.

This study aimed to explore the impact of cancer on young adults’ evolving sense of self and identity, by listening to their stories over one year from the time of diagnosis. Data were collected using a range of methods: in depth, free association narrative interviews at three time points; photographs taken by participants; and extensive reflexive field notes. Forty interviews were conducted with eighteen young adults, sixteen to thirty years of age. Eight of the participants took part in three interviews, six participants in two interviews, five participants took photographs. Eight longitudinal cases were analysed in depth, visual images were analysed from discussion in the narrative text. Through memoing, coding and comparison themes were developed across all cases and all participant’s data. Five cases have been reported as longitudinal stories to illustrate the interplay between the internal and external over the year from diagnosis. Higher order themes across all the data demonstrate the renegotiation of self over time, both developmental and in terms of ‘cancer time’, through the core components of: the inner world, (psyche, emotion and coping); self as embodied; self as relating to others, and self as relating to place. The study offers new insights into the experience of young adults with cancer and the value of basing care on individual experience beyond age but situated within biography and identity. The findings demonstrate the intensity of the juxtaposition of cancer and developmental stage and highlight the importance of visual and oral narratives and a psychosocial lens in both research and practice. The study suggests the value of narrative as a prospective intervention in health care to support sense making, identity renegotiation and revision; to give patients a voice. The study also highlights the importance professionals to be supported in ‘being with’ and in walking alongside people going through life changing illness.

Ageing is the major risk factor for all age-related diseases such as cancer, diabetes, cardiovascular disorders, and neurodegenerative diseases. Moreover, the rate of ageing is controlled, at least to some extent, by genetic pathways and biochemical processes conserved in evolution. With age, there is a general loss of histones, global and local change in DNA methylation and an imbalance of activating and repressive histone modifications. The use of Drosophila melanogaster as a model to study ageing is well established and posses many advantages, such as a rich genetic toolbox. Therefore, I investigated whether and how chromatin status is associated with ageing in Drosophila, in specific through histone lysine methylation. In order to do so, I carried out an RNAi screen of many of the most relevant histone lysine methyltransferases and demethylases in flies. From this screen, I identified five longevity genes that, when downregulated by RNAi, can extend the lifespan. Thus, I successfully established a link between these specific chromatin modifiers with epigenetic changes and lifespan in Drosophila. Moreover, among these five genes, I further characterized two, Lsd1 an H3K4me1/me2 demethylase and Set1, and an H3K4me2/me3 methyltransferase. In addition, I examined the long-lived flies for resistance to stresses such as starvation and oxidative stress, and measured their health through fecundity assays. Furthermore, I observed by western blotting that there was no change in global H3K4 methylation levels in long-lived Lsd1 heterozygous flies, nor in long-lived flies in which Set1 has been downregulated (~60%) in the intestine/gut fat body, suggesting that lifespan extension might result as a consequence from changes in specific loci and/or alterations in non-histone targets. Insulin/IGF-1 signalling (IIS), mTOR and RAS-ERK-ETS are highly evolutionary conserved longevity pathways in which many of the anti-ageing interventions converge. By ascertaining the levels of known markers of these pathways by western blotting in different tissues, I found that Lsd1 and Set1 longevity do not converge on the IIS, mTOR, or Ras-Erk-ETS pathways. Therefore, downregulation of Lsd1 and Set1 constitute a novel anti-ageing intervention in Drosophila, which could be combined with inhibitors of the aforementioned pathways for a more pronounced lifespan extension. Finally, I tested whether any relationship was discernible between the lifespan extension mediated by Lsd1 and Set1 and genomic integrity. Interestingly, downregulation of Lsd1 and Set1 exhibited increased bleomycin resistance. Therefore, lifespan extension that results from downregulation of Lsd1 and Set1 might arise from changes in specific loci and/or alterations in non-histone targets, which might play a part in the DNA damage response triggered by double-strand breaks.

Mouse and human embryonic stem cells (ESCs) are derived from the inner cell mass of blastocysts however are distinct in their molecular and biological properties. Historically these discrepancies have been attributed to speciesspecific differences. Recent work capturing human ESC-like cells from postimplantation mouse embryos instead suggests that they represent distinct developmental stages of the pluripotent epiblast. Mouse ESCs resemble the preimplantation epiblast where cells are “naive” and have no differentiation bias. In contrast human ESCs represent a developmental stage postimplantation “primed” for lineage specification. Here a systematic approach was taken to identify conditions to induce and maintain naive human ESCs by screening for small molecules that support self-renewal based on the maintenance of endogenous OCT4 distal enhancer activity, a molecular signature of naive pluripotency. Iterative chemical screening identified a combination of five kinase inhibitors, 5i (MEK, GSK3, BRAF, SRC, ROCK) that induce and maintain OCT4 distal enhancer activity when applied directly to conventional human ESCs. Furthermore, these conditions allowed the capture of naive pluripotent cells from human embryos and somatic cells through reprogramming. The inhibitors described herein capture a population of human pluripotent cells in which transcription factors specific to ground state pluripotency are highly upregulated and bivalent chromatin domains are depleted. Comparison to published reports of naive human ESCs suggest that previous attempts to isolate naive human pluripotent cells fall short of recapitulating the transcriptional foundations of the naive state. In contrast the conditions presented in this thesis capture a distinct state of human pluripotency that very closely resembles mouse ESCs. Given the stringency of the reporter used and the systematic approach taken, this study should provide a framework for future work in characterizing and consolidating naive human pluripotency.

The advent of checkpoint immunotherapy has revolutionised the treatment of solid cancers, resulting in durable responses in a number of patients. However the majority of patients do not respond to treatment, underscoring the need to better understand the mechanisms that underlie the immunological response to cancer. T cell receptor (TCR) repertoire analysis and immune checkpoint mapping are powerful tools to study the anti-tumoural immune response. This thesis explores the TCR repertoire and immune checkpoint landscape in muscle invasive bladder cancer (MIBC), non-small cell lung carcinoma (NSCLC) and metastatic melanoma. In patients with MIBC, immune checkpoint mapping and TCR repertoire analysis revealed similarities between lymphocytes in the urine and bladder tumour microenvironment. Urine-derived lymphocytes may therefore provide a non-invasive immune biomarker to track the evolution of the immune landscape in MIBC. The lung TRACERx study is a prospective study exploring the cancer genome evolution of early stage NSCLC. The TCR repertoire of multi-region tumour specimens was found to be distinct to non-tumour lung and PBMC. TCR repertoire heterogeneity amongst TCRs preferentially expanded in the tumour, was correlated to intratumoural genomic heterogeneity. Intratumoural TCR expansion enriched for a tumour reactive T cell phenotype. These observations suggest a dynamic intra-tumoural T cell response related to the mutational landscape of NSCLC. The immune checkpoint landscape of treatment naïve and anti-PD-1 treated metastatic melanoma patients was assessed by multiparametric flow cytometry. The checkpoint phenotype of PD-1 expressing tumour infiltrating lymphocytes (TILs) in treatment naïve samples was heterogeneous. Drug bound effector TILs in anti-PD-1 treated melanoma co-expressed multiple immune checkpoint molecules that may have contributed to treatment resistance. Immune checkpoint mapping of CD8+ and CD4+FoxP3- cells and CD4+FoxP3+ (regulatory T cells) revealed differences in the expression level and frequencies of key co-inhibitory and co-stimulatory molecules. TCR repertoire analysis and T cell immune checkpoint mapping provide valuable and complementary tools for analysis of the immune response to solid cancers.

Cancer is the 2nd most common cause of death in our society and is associated with high morbidity and costs. The word ‘cancer’ amalgamates the complex interplay between cells which have acquired genetic alterations leading to uncontrolled proliferation, i.e. the malignant cells, and genetically ‘normal’ host cells, i.e. stromal cells, vascular cells and inflammatory cells which all acquire modified biological phenotypes in the presence of malignant cells. This community of cells and their secreted proteins defines the tumour microenvironment. Stromal cells in the tumour microenvironment display characteristic biological changes which promote cancer growth. Little is known on the underlying regulatory mechanisms defining this phenotype. Epigenetics describes inheritable changes not encoded by the nucleic acid sequence. Epigenetic regulation has been described to occur in stromal cells in the tumour microenvironment, but little is known about its role on myofibroblasts. In this work I describe how oesophageal cancer derived stromal cells, i.e. cancer associated myofibroblasts (CAMs) accelerate tumour growth in vivo. I observed that CAMs not only affect the local tumour microenvironment but might also accelerate tumour growth at a distant site. I also show how myofibroblasts play an important role in early tumour niche formation in xenograft models and describe their disappearance and replacement by murine stromal cells during tumour progression. Oesophageal CAMs were shown to be epigenetically distinct from matched adjacent tissue myofibroblasts (ATMs). They exhibited a global DNA hypo- methylation compared to ATMs. We identified distinct DNA methylation signatures between oesophageal cancer CAMs and ATMs with the use of the Illumina 450k bead chip methylation array. The methylation array data showed altered methylation signatures of genes implicated Wnt/β-catenin signalling pathway. The transcription factor paired like homeodomain (PITX) 2 and the regulatory protein secreted frizzled like protein (SFRP) 2 both showed altered methylation signatures and expression patterns between oesophageal cancer CAMs and ATMs. I found that upregulation of SFRP2 in myofibroblasts induces angiogenesis and I hypothesise that epigenetic modification regulates myofibroblasts-derived SFRP2 expression which may play an important role in tumour neovascularisation. Based on these findings I conclude that ATMs and CAMs are epigenetically distinct and altered protein expression is at least partially regulated by altered DNA methylation. This work also presents a model for epigenetic modification of tumour stroma cells: exposure of myofibroblasts to the DNA methyl transferase inhibitor 5’Aza-3’deoxycytosine (DAC) lead to a mild decrease of global DNA methylation and induced persistent biological changes in myofibroblasts. These epigenetically modified myofibroblasts induced an accelerated xenograft growth when injected together with oesophageal cancer cells. Based on these experiments I conclude that DAC epigenetically modifies myofibroblasts which induces an activation of normally silenced genes leading to a biologically more active cell.

Background: Screening is offered to individuals who have an identified increased risk of pancreatic cancer. Such risks may be increased because of a family history or a known genetic mutation which has been shown to confer increased risk. Amongst those being screened intraductal papillary mucinous neoplasm (IPMN) of the pancreas are increasingly common entities being detected incidentally; their risk of malignancy can be as high as 85%. Consensus guidelines exist for the management of these lesions – the morbidity associated with pancreas resection is as high as 50%. We set out to identify a marker which could be used to identify those IPMN which should be resected and those which may be safely observed. Methods: Individuals were identified from the European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer (EUROPAC) or patients identified in Liverpool or Hedielberg with cystic lesions and/or pancreatic cancer. Cancer screening was performed by imaging and with molecular analysis (including mutation analysis of TP53) of pancreatic juice obtained by endoscopic retrograde cholangiopancreatography (ERCP). Matched tissue sections and frozen sections of pancreatic tissue from 73 patients who underwent resection for IPMN were assessed histologically and for mutations in TP53 status using a novel limiting dilution Next Generation Sequencing technique. Results were assessed relative to clinical outcomes. Results: Amongst those 29 individuals who were screened with ERCP, 11 had IPMN, 7 IPMN with cancer (IPMC) and 3 pancreatic ductal adenocarcinoma (PDAC). The remaining cases were benign neoplastic or inflammatory conditions. Kaplan-Meier survival analysis at 5 years found that the presence of p53 mutation in the tissue was a better prognostic marker of survival than the histological diagnosis alone (p=0.0152 vs. p=0.0819). Sensitivity and specificity of p53 mutation as a predictor of survival was calculated as 0.89 and 0.95 respectively. There was 100% correlation between the p53 mutational status of the resected tissue and the pancreatic juice obtained at ERCP. When ERCP was assessed as a method for screening, however, there was found to be an unacceptably high incidence of post-ERCP pancreatitis (PEP) 7 cases of PEP in 16 ERCPs (44%). This rate was shown to be significantly reduced to 15% (6/40) with the use of pancreatic stent and diclofenac, but the overall prevalence of PEP was 23.2% over 14 years. There were no cases of PEP amongst those individuals being screened because of hereditary pancreatitis. Of 27 IPMN cases with frozen tissue 23 individuals had TP53 mutations. Seven cases died of pancreatic cancer after resection. Kaplan-Meier survival analysis revealed that one mutation p.L264R predicted survival regardless of histology (p= < 0.0001). The mutation was present in 6 of the 7 cases who died and in none of those who survived to 5 years. Mutation specific PCR was used to validate results showing that p.L264R discriminated between survivors and IPMN cases who died of cancer (AUC = 0.79). Conclusions: IPMN continues to cause concern and uncertainty among those individuals being screened for cancer who are largely well and asymptomatic. The p.L264R mutation could be used to differentiate those IPMN which result in poor survival to facilitate potentially curative surgery. The mutation may be present in pancreatic juice which can be collected endoscopically as a screening tool. The use of prophylactic measures to reduce PEP may be considered sufficient to bring the risk of complications to an acceptable level when compared to the relative certainty of prognosis afforded by a positive test for p.L264R.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers with an overall 5-year survival rate < 5%, a rate that has not improved for a long time. The dismal prognosis for PDAC is part due to late detection, often at advanced stages of the disease where patients have developed distant metastases, but also due to chemotherapeutical resistance. Patients eligible for surgical resection of the pancreatic tumour has the best prognosis, but even when resection is successful, patients often relapse with distant metastases within 2 years after surgery. Macrophages promote tumourigenesis and enhance metastasis in many cancer types; however, the role of macrophages in PDAC metastasis is poorly understood. Using an experimental mouse model of liver metastasis, we find that PDAC liver metastasis critically depends on the early recruitment of granulin-secreting metastasis-associated macrophages (MAMs) to the liver that promote metastatic colonisation. Mechanistically, we demonstrate that granulin secretion by MAMs activates resident hepatic stellate cells (HSTCs) into myofibroblasts that secrete extracellular matrix components, including periostin, resulting in a fibrotic microenvironment that sustains metastatic tumour growth. Disruption of MAM recruitment in PI(3)Kγ-deficient mice, chemical depletion of MAMs from established lesions or genetic ablation of granulin reduces HSTC activation and liver metastasis. Adjuvant CTX is standard for patients after surgical resection to eliminate any residual cancer cells, and it improves survival for patients after resection. We find that 45% of mice with metastatic lesions respond to gemcitabine treatment with a reduction in metastatic burden. The metastatic lesions in these mice are characterised by a reduction in αSMA+ myofibroblasts. HSTCs do not seem to promote cancer cell survival in the presence of chemotherapy, but rather constitutes a protective niche that promotes relapse by promoting the growth of pancreatic cancer cells after gemcitabine treatment.