Introduction Oesophageal cancer has a poor prognosis. Early diagnosis and the use of chemotherapy and surgery for local disease are key to improving survival. This study was designed to see if plasma and tissue metabolic profiles could be used to identify oesophago-gastric malignancy, indicate the presence of unstable pre-malignant (Barrett’s) epithelium or predict response to chemotherapy. Methods Patients were recruited from University Hospitals Birmingham between May 2009 and March 2010. Nuclear Magnetic Resonance (NMR) metabolomics was performed on filtered plasma and extracted tissue samples. Results Some 258 participants were recruited. NMR metabolomics discriminated between normal, Barrett’s and neoplastic epithelium. Tissue levels of hypoxanthine were highest in oesophageal adenocarcinoma compared to adjacent normal mucosa. Levels in Barrett’s mucosa in the presence of cancer fell between normal and neoplastic mucosa. 3-hydroxybutyrate levels were elevated both in cancer tissues and plasma compared to controls. Plasma levels of 3-hydroxybutyrate were higher in patients with node positive and full thickness tumours compared to those who were node negative with early local disease. Conclusion NMR metabolomics identified metabolic profiles that characterized different histologic tissue types. Metabolites involved in oesophageal carcinogenesis might influence diagnostic and management strategies in these patients.

EBV is a lymphotropic herpesvirus that establishes lifelong persistence in the memory B cell compartment of the human host. It is still unclear, however, whether the virus infects memory B cells directly, or first targets naive B cells, driving them to differentiate into memory B cells via germinal centre (GC) transit. Using ex vivo analysis of sorted B cell subsets, we have found that whilst EBV preferentially colonises isotype-switched memory B cells, the virus was excluded from tonsillar GC B cells. Furthermore we found substantial viral loads in non-switched memory B cell populations, whose origins are likely to be germinal centre-independent. Using in vitro infection experiments, we showed that enzymes AID, UNG and pol-η, which are associated with the GC processes somatic hypermutation (SHM) and class switch recombination (CSR) were upregulated in EBV-infected B cells. Indeed following in vitro transformation, SHM of immunoglobulin (Ig) genes was induced in a proportion naive B cells. Whilst these cells did not undergo CSR following EBV infection alone, additional cytokine stimulation together with CD40L was able to induce isotype switching. EBV infection in vitro together with the provision of appropriate signals was therefore able to induce genotypic and phenotypic memory B cell characteristics in naive B cells, in a non-GC environment. Taken together our findings suggest that GC transit is not an essential requirement for EBV colonisation of the memory B cell population.

Pituitary tumor transforming gene1 (PTTG1) - binding factor (PBF) is a ubiquitously expressed transmembrane glycoprotein that is significantly upregulated in a variety of human tumours. In thyroid cancer, PBF is implicated in early tumour recurrence, distant metastasis and poor oncological outcome. Indeed, PBF induces cell transformation in vitro and tumourigenesis in vivo. PBF interacts with multiple protein-binding partners and modulates their activities and functions, including cortactin, Src, NIS and p53. According to the COSMIC and TCGA databases, 27 PBF mutations have now been reported in human cancer. The studies in this thesis aimed to establish the basic biochemical characteristics of the first 10 reported missense PBF mutations, including protein glycosylation, dimerisation and stability. Following initial functional studies, the impact of two in vivo mutations, C51R and R 140W, upon WT PBF function in modulating cellular proliferation, cell invasion, cell migration, radioiodine uptake and transforming ability were investigated in more depth. The data in this thesis demonstrate that although the reported mutations preserve some critical functions of PBF, it is unlikely that PBF mutations represent distinct driver events, and that upregulation of PBF expression in human cancer may be more important in terms of driving tumourigenesis than sequence changes.

Protein kinases are important mediators of cell signalling that are often implicated in disease when deregulation occurs. The catalytic kinase domain is highly conserved with 518 identified members in the super family. Kinase structural studies are mainly based on relatively static crystal structures. However protein kinases are inherently dynamic entities in solution. Several Ser/Thr protein kinases were evaluated by NMR in order to obtain an amenable target for solution structure and functional characterisation. Subsequently a calcium calmodulin dependent protein kinase dubbed CaMK1D was identified as the optimal system. CaMK1D normally mediates intracellular signalling downstream of chemokines. It is amplified in breast cancer, and induces cell proliferation, migration and invasion. Here we report the backbone resonance assignments for the 38 kDa human autoinhibited CaMK1D in its free state, encompassing a canonical bi-lobed kinase fold and autoinhibitory and calmodulin binding domains. These assignments allowed us to probe the binding mode of CaMK1D with small molecule ligands and refine the crystal structure via dihedral angle restraints for a more complete structure. Furthermore we investigated the solution structure of the CaMK1D∙Ca\(^{2+}\)/CaM complex and propose a model of the activation mechanism and establish a key residue implicated in complex formation.

Recent studies have suggested an increasingly important and therapeutically relevant role for CD4+ T cells as direct effectors in immunotherapy. It has also become clear that antigens can access the MHC-II pathway via non-classical endogenous routes such as autophagy. The aims of this project were to investigate endogenous MHC-II processing of tumour antigens and explore methods that, by enhancing such processing, could increase T cell recognition of tumour cells. The frequency and diversity of T cell responses to the tumour antigen WT1 in healthy individuals and cancer patients was investigated. T cell responses were rare and only one epitope-specificity was identified in these donors. The Epstein Barr Virus Nuclear Antigen 1 (EBNA1), a protein that becomes more efficiently processed by autophagy when relocalised from the nucleus into the cytoplasm, was also investigated. When EBNA1-positive cells were treated with the CDK inhibitor Roscovitine, previously reported to relocalise EBNA1 into the cytoplasm, presentation of an autophagy-dependent indicator epitope was decreased. Roscovitine did not affect autophagy activity but decreased cellular levels of several autophagy adaptor proteins. siRNA Knock down of each autophagy adaptor showed CALCOCO2, NBR1 and OPTN were involved in the generation of an autophagy-dependent epitope from EBNA1; a control epitope was unaffected. Finally, efforts were made to define fundamental rules determining why only a subset of potential MHC-II epitopes are generated from an antigen by autophagy. Resistance to degradation by lysosomal peptidases was shown to be an important factor determining whether an MHC-II epitope can be generated by autophagy.

Latent membrane protein 1 (LMP1) is an oncogene expressed in a subset of germinal centre (GC)-derived lymphomas including Hodgkin’s lymphoma (HL) and diffuse large B cell lymphoma (DLBCL). However, LMP1 shares functional homology with CD40, a receptor required for normal GC B cell development. Dissecting how LMP1 functions differently from CD40 in GC cells is central to a better understanding of lymphomagenesis and is the subject of this thesis. In Chapter 3, I show that GC B cells can be successfully isolated from normal human tonsils and that these cells retain a GC phenotype upon short-term culture. In Chapter 4 I explore how the transcriptional programmes of LMP1 and CD40 differ in GC B cells and identify a subgroup of genes regulated by LMP1 but not by CD40, which are also concordantly regulated in primary HL cells from which I focus on sphingosine-1-phosphate receptor 2 (S1PR2). I confirm that S1PR2 is an LMP1 target in GC B cells and show that it is not expressed in the tumour cells of the majority of cases of HL and DLBCL. In DLBCL, S1PR2 loss is associated with LMP1 expression. I also provide preliminary evidence that the over-expression of S1PR2 can inhibit the HL cell migration. In Chapter 5, I report my initial attempts to optimise a method for the measurement of the activity of transcription factors in GC B cells which can be used to delineate those pathways activated by LMP1, but not by CD40, in GC B cells.

Tumour angiogenesis is a vital process in the pathology of tumour development and metastasis. Targeting markers of tumour endothelium provide a means of targeted destruction of a tumours oxygen and nutrient supply via destruction of tumour vasculature, which in turn ultimately leads to beneficial consequences to patients. Although current anti-angiogenic and vascular targeting strategies help patients, more potently in combination with chemo therapy, there is still a need for more tumour endothelial marker discoveries as current treatments have cardiovascular and other side effects. For the first time, the analyses of in-vivo biotinylation of an embryonic system is performed to obtain putative vascular targets. Also for the first time, deep sequencing is applied to freshly isolated tumour and normal endothelial cells from lung, colon and bladder tissues for the identification of pan-vascular-targets. Integration of the proteomic, deep sequencing, public cDNA libraries and microarrays, delivers 5,892 putative vascular targets to the science community. These analyses identify Endothelial Specific Molecule 1 as a pan vascular target and lysyl oxidase-like 2 as putative novel vascular target. It is envisioned vascular targets and angiogenesis genes in this data will destroy or inhibit tumour vessel growth without the side effects manifest with current clinical regimens.

It has been postulated that bladder cancer impacts health-related quality of life (HRQoL), but research is limited. This research is particularly important as survival for non-muscle invasive bladder cancer (NMIBC) is high, and these patients will need to live with acceptable HRQoL for many years. I investigated HRQoL in 1258 muscle invasive bladder cancer (MIBC) and NMIBC patients participating in the Bladder Cancer Prognosis Programme. Although I found no major difference in HRQoL around time of diagnosis between NMIBC and MIBC patients, I did find that different parts of HRQoL seemed to influence survival of NMIBC and MIBC. Also, patients prefer an invasive but accurate cystoscopy over a hypothetical non- invasive but less accurate urinary biomarker. Additionally, the European Organization of Research and Treatment of Cancer (EORTC) developed a quality of life questionnaire specifically for NMIBC (EORTC NMIBC-24). In this thesis, I have strengthened the structure of this questionnaire that was previously published by another UK research group. Finally, I found only a small difference in physical health between patients with incontinent and continent urinary diversion in a meta-analysis. These findings were discussed in each of the results chapters, and put into wider context in the general discussion chapter.

Survivors of childhood, teenage and young adult cancer are at increased risk of developing adverse health outcomes. This thesis aims to address the gaps in knowledge regarding the most severe adverse health outcomes. The Teenage and Young Adult Cancer Survivor Study (TYACSS) provides 200,945 survivors of cancer diagnosed aged 15-39 years. The PanCare Childhood and Adolescent Cancer Survivor Care and Follow-Up Studies (PanCareSurFup) provides 69,460 survivors of cancer diagnosed aged < 20 years. Within the TYACSS cohort 1) cancer survivors had increased risk of developing subsequent primary neoplasms, particularly in previously irradiated sites; 2) cancer survivors who likely received cranial irradiation had increased risk of a cerebrovascular event; and 3) central nervous system tumour survivors experienced premature mortality due to neoplastic and nonneoplastic causes. Within the PanCareSurFup cohort 1) the excess number of subsequent softtissue sarcoma was low, except leiomyosarcoma after retinoblastoma; and 2) the excess number of subsequent breast cancers remained elevated beyond 40 years of age among survivors of Hodgkin lymphoma, Wilms tumour and sarcoma. This thesis focuses on the most severe adverse health outcomes among childhood, teenage and young adult cancer survivors and provides evidence for developing clinical follow-up guidelines aimed at reducing such adverse health outcomes.

In the course of this study it was established that CtlP associates in vivo with, and directly binds to, the DNA damage mediator proteins containing BRCT domains: 53BP1, MDC1, TopBP1 and NBS1. The binding sites on all of these proteins have been mapped establishing which regions of CtlP interact directly with 53BP1, MDC1, TopBP1 and NBS1 and vice versa. This implies that CtlP is involved in the DNA damage response on many levels interacting with proteins that mediate different aspects of DNA damage sensing and signalling. CtIP's nuclear localization before and after DNA damage was determined and its colocalization with the other DNA damaged responsive proteins examined. It was found that CtIP associates in vivo with DNA damage sensor proteins such as the MRN complex (Mre11, Rad50 and NBS1) and RPA70, signal transducer proteins such as the PIKK kinases ATM, ATR and SMG1 and the mediator proteins 53BP1 and MDC1. All of those proteins are involved in detecting and repairing double stranded or single stranded lesions. Moreover CtIP has a major influence on phosphorylation events induced during repair processes and thus could be an inducer of kinase activity. The deletion of CtlP causes impairment of phosphorylation of important proteins that take part in the DNA damage response pathways such as NBS1, ATM, 53BP1, Chk1 and RPA70. However, impaired phosphorylation only occurs in response to single stranded DNA lesions or the collapse of replication forks. This implies that it is the ATR pathway that is malfunctioning in CtIP depleted cells. CtIP was previously reported to be phosphorylated on Ser664 and Ser745 by the ATM kinase in response to the DNA double stranded breaks. I have established two possible new phosphorylation sites: Ser506 and Ser555. They could be potentially involved in CtlP's redistribution to the DNA damage sites but this remains to be elucidated.