The angiogenic growth of blood vessels and lymphatic vessels plays important roles during embryonic development as well as in disease conditions such as cancer. Previous work has shown that ephrin-B2, a transmembrane protein and ligand for Eph family receptor tyrosin kinases, regulates the growth of these endothelial networks. In mice, inactivation of ephrin-B2 in endothelial cells (ECs) results in similar vascular defects and embryonic lethality as in global Efnb2 knockouts, indicating that the main function of ephrin-B2 is in the endothelium. Despite its importance, the functional roles of ephrin-B2 in vascular endothelium are incompletely understood. To gain further insight into the function of ephrin-B2 in ECs in mice, I have used EC-specific and temporally controlled loss-of-function and gain-of-function genetic approaches, and analysed the function of ephrin-B2 in different vascular beds. My data show that ephrin-B2 has two distinct functional roles in the blood vasculature. In arteries, the domain with the highest expression of ephrin-B2, the ligand is critical for arterial remodelling. Outside arteries, in the microvessels of capillary beds, ephrin-B2 is essential for endothelial motility and invasiveness during angiogenic sprouting. Furthermore, ephrin-B2 is an important regulator of angiogenic lymphatic endothelial cell (LEC) behaviour, which may be associated with VEGFR3-mediated signalling. In addition, ephrin-B2 is essential for the maintenance of LEC differentiation and, at least during embryonic development, normal Prox1 expression. Moreover, my data indicates that ephrin-B2 is critical for the postnatal remodeling of the primary dermal lymphatic plexus, which appears to be mainly mediated through ephrin-B2 reverse signalling. The sum of my findings demonstrates that ephrin-B2 is a key regulator of EC motility during the angiogenic growth of blood vessels and lymphatic vessels. Ephrin-B2 plays multiple indispensable and vascular domain-specific roles in the angiogenic growth of endothelial networks.

The tumour suppressor Merlin is mutated in the familial cancer syndrome Neurofibromatosis Type 2 (NF2) and in various sporadic tumours. Several lines of evidence show a role for Merlin in mediating contact-dependent inhibition of proliferation. Contact inhibition is essential for normal tissue homeostasis and its deregulation is a hallmark of cancer. Despite extensive efforts the mechanism underlying tumour suppression by Merlin remains elusive. A proteomics approach led to the identification of novel Merlin interacting proteins several of which are involved in gene expression, in particular transcription elongation, RNA processing and histone modifications, and are also deregulated in cancer. Among those are the RNA PolymeraseII-associated factor1 complex (Paf1C), CHD1, TAT-SF1 and spliceosome components. These interactions have been validated and mapped to Merlin’s FERM domain and loss-of-function Merlin mutations found in tumours invariably disrupt these interactions. The interaction of Merlin with the Paf1C is modulated by cell density and is required for Merlin’s tumour suppressor function. We find a remarkable case of tumour suppressor gene hypersensitivity in Merlin-deficient cells that correlates with Merlin’s ability to regulate gene expression. Using genome-wide expression profiling we identify a gene signature associated with growth arrest by Merlin expression that is consistent with Merlin’s role in mediating contact inhibition and suggests a role for Merlin in innate immunity and communication with the microenvironment. By integrating re-expression and knockdown gene signatures a core Merlin signature has been defined that correlates with NF2 mutational status and suggests differential drug sensitivities and potential therapeutic targets for treatment of NF2-related tumours. Using genome-wide occupancy analysis we identify a subset of genes where Merlin expression regulates association of the Paf1C with chromatin of coding regions. Some of these regulated genes are known target genes of YAP, the transcriptional co-activator of the Hippo tumour suppressor pathway. Although YAP expression rescues growth arrest by Merlin, Merlin can regulate expression independently of YAP. A model is proposed where Merlin functions parallel to YAP to regulate gene expression at the elongation level through the Paf1C.

The principal objectives of the work in this thesis were: (a) to investigate cancer cell invasion in squamous cell carcinoma (SCC) by examining the role of ubiquitin-linked pathways in the regulation of the actin cytoskeleton and (b) to establish a human tissue bank of SCCHN specimens matched to adjacent non-cancerous tissue. An siRNA screen was performed with the premise that genes that regulate the actin cytoskeleton would be involved in the invasion of cancer cells. 289 E3 ubiquitin ligases and 137 de-ubiquitinating enzymes (DUBs) were screened yielding 3 gene hits (RKHD2, TRAF6 and ZA20d1) capable of regulating both the F-actin cytoskeleton of SCC cells as well as their collective invasion in a three-dimensional organotypic invasion system. A secondary morphological and functional screen of 20 E3 ligases and 9 DUBs in carcinoma-associated fibroblasts demonstrated that 5 genes (MKRN2, RKHD2, TRAF6, ZA20d1 and USP6) influenced the ability of stromal fibroblasts to promote SCC invasion. Following transcriptional analysis using qPCR, the E3 ligase TRAF6 and its complementary DUB ZA20d1 emerged as candidate genes for detailed study. TRAF6 was shown to influence NF-ĸB signalling in SCC cells by modulating the levels of IĸBα. Given the importance of NF-ĸB activation in the pathogenesis of SCC, I investigated its significance in our organotypic system. Using TNFα, IL-1α and several IKK inhibitors, I demonstrated that SCC invasion was dose-responsive and NF-ĸB dependent. Furthermore, infliximab (a monoclonal antibody to TNFα in clinical use) inhibited carcinoma invasion at a therapeutic dose. Ethical and R&D approval for the establishment of a human tissue bank was obtained and SCCHN tumour samples with patient matched control tissue are now being banked. Analysis of these specimens and commercial tissue microarrays indicates that TRAF6 is overexpressed in a subset of SCC.

The epithelial restricted αvβ6 integrin is known to have minimal expression in healthy tissue and to be upregulated in cancer and healing wounds. This thesis explores the role of αvβ6 in cancer and tests the hypothesis that αvβ6 has a prognostic and therapeutic utility in cancer. Using immunohistochemistry, increased αvβ6 expression was found in nonmelanoma skin cancers (NMSC), particularly in morphoeic type basal cell carcinoma. In cell culture experiments, αvβ6 was found to activate TGF-β and promote myofibroblast differentiation, producing a tumour stroma rich in smooth muscle actin (SMA). These findings prompted a study of αvβ6 and SMA as prognostic indicators in oral squamous cell carcinoma (OSCC). A study of 282 cases of OSCC found that although αvβ6 was not a prognostic marker, patients with high SMA levels had a highly significant increased risk of disease specific mortality (HR 3.06 [CI 1.65-5.66], p<0.001). Next, the utility of αvβ6 as a target was explored through the development of a single chain antibody fragment (scFv) specific for αvβ6. The scFv was tested for the delivery of targeted magnetic fluid hyperthermia (MFH), an experimental cancer treatment based on the generation of heat by magnetic nanoparticles when placed within an alternating magnetic field. The αvβ6-specific scFv (B6.3) was manufactured and high ligand specificity confirmed on ELISA and FACS analysis. B6.3 was successfully conjugated to two alternative iron nanoparticles. In-vitro studies demonstrated increased cellular uptake of scFv-nanoparticle complexes and greater cellular toxicity on exposure to MFH compared to nanoparticles alone. In conclusion, αvβ6 is a potential target for therapy in NMSC and OSCC. SMA is found to be an independent prognostic marker in OSCC and αvβ6 identified as a proinvasive factor in morphoeic BCC. Finally, the production αvβ6 specific scFvs and use for in-vitro MFH potentiates the development of αvβ6 targeted MFH cancer therapy.

Osteosarcoma (OS) is a rare malignant mesenchymal neoplasm, but the third most common cancer in adolescents. Understanding the molecular pathogenesis of this disease is essential to design new, effective therapeutic strategies to improve patient survival. The recent report that LKB1-deficient mice develop bone-forming tumours begs the question whether loss of this tumour suppressor gene plays a role in human OS pathogenesis. We found that LKB1 protein expression was reduced in 20 of 26 (77%) human OS cases by Western blot and 153 of 259 (59%) by immunohistochemistry. Downstream, the mTOR pathway was activated in 137 of 158 cases (87%) and this activation was correlated to LKB1 loss, providing new insights into potential treatments. No copy number loss of the LKB1 region was identified in 93 OS cases by interphase fluorescent in situ hybridization. Four of 12 informative cases had concomitant loss of one parental allele at the locus of one single nucleotide polymorphism and reduced protein expression; one of them possessed only one LKB1 copy by qPCR. Direct sequencing of 21 cases failed to detect LKB1 mutations and all these cases expressed similarly high levels of LKB1 mRNA by qRT-PCR, irrespective of their protein expression. It suggested that LKB1 is regulated post-transcriptionally in OS. In silico analysis of our OS microRNA data showed that this cannot be accounted for by microRNA directly targeting LKB1 mRNA and a preliminary study could not exclude the involvement of SIRT1. The knock-down of LKB1 by shRNA in the osteoblast cell line HOB had only subtle effects on cell proliferation and survival. Its role in OS pathogenesis was confirmed by knock-in in LKB1-deficient OS cell lines, which induced reduced cell proliferation. To conclude, LKB1 protein expression is reduced in a subset of OS by a post-transcriptional mechanism, leading to increased cell proliferation in the tumour.

Angiogenesis plays a key role in the pathogenesis of haematological malignancies such as acute myeloid leukaemia (AML), multiple myeloma (MM) and lymphoma (NHL). The evidence supporting this theory includes the finding of increased bone marrow microvessel density and increased levels of plasma and urinary pro-angiogenic cytokines in patients with these malignancies as well as encouraging results from preclinical and clinical studies using antiangiogenic therapies. A significant limiting factor in most of these studies has been the short half-life of most antiangiogenic compounds which has necessitated the use of cumbersome administration protocols resulting in poor compliance and an inability to administer the drug long enough to result in its desired therapeutic effect. The aim of this thesis therefore is to investigate the therapeutic effects of stable antiangiogenic expression in xenograft models of AML, NHL and MM. We investigated the antitumour effects of stably blocking the vascular endothelial growth factor (VEGF) pathway, the matrix metalloproteinase pathway and the use of interferons, a family of cytokines with significant antitumour activity resulting not only from their antiangiogenic properties but also from their anti-proliferative, anti-apoptotic and immunomodulatory effects. Murine models of AML, NHL and MM were developed by engrafting a number of well characterized tumour cell lines – HL-60, Raji and KMS12-BM respectively into β2mnull NOD/SCID immunodeficient mice, resulting in a tumour phenotype resembling the human disease. Blockade of the VEGF pathway was achieved by stably expressing the soluble form of the VEGF receptor (sFlk-1) using adeno associated viral vectors or by using an antibody directed against VEGF but neither strategy resulted in antitumour activity in AML or NHL xenograft models. We were however able to show that stable expression of tissue inhibitor of metalloproteinase-3 (TIMP-3) resulted in significant antitumor activity against MM cells in vivo. Finally we were able to demonstrate for the first time that Interferon-β has profound antileukaemic activity in the setting of a xenograft model and that even low level expression when stable can significantly inhibit primary AML engraftment in vivo.

The WNT pathway plays a key role in development and disease. In addition to the better studied ß-catenin dependent pathway, WNT ligands can also activate the separate ‘non-canonical’ or Planar Cell Polarity (PCP) pathway. Perturbations in the PCP pathway contribute to the pathogenesis of a variety of diseases including cardiac and neural tube defects, and to the invasiveness of cancer cells. R-RAS subgroup GTPases share many of the properties of classical RAS GTPases including the ability to behave as oncogenes. However, they also have distinct functions of their own and how signalling and biological specificity is achieved is not fully understood. Using a proteomic approach to identify novel R-RAS subgroup effectors led to the identification of VANGL1, a WNT/PCP protein, demonstrated to be a novel R-RAS interacting protein. In this thesis, I have shown that VANGL1 functions as an effector of R-RAS and TC21. Using proteomic approaches, multiple VANGL1 interacting proteins have been identified and R-RAS, as well as selected Frizzled (FZD) receptors and the tyrosine kinase ROR2 can modulate at least some of these VANGL1 interactions. Furthermore, VANGL1 leads to the protein degradation of PRICKLE by a mechanism that remains to be determined, and R-RAS GTPases are able to inhibit this effect. Using RBD pulldown assays, I was able to show that WNT ligands and ROR2 can lead to the activation of R-RAS, and that R-RAS and TC21 are key mediators of RHO activation by WNT5a. Finally, I demonstrated that R-RAS/TC21 and VANGL1 are critically required for directed migration. The identification of R-RAS activation by WNT ligands and its interaction with VANGL1 provides an exciting new link between the R-RAS subgroup of the RAS family and the WNT/PCP pathway.

This thesis describes the need for imaging the tumour pathophysiological microenvironment in order to understand response to treatment. Specifically looking at tumour vascularisation in in vivo murine xenograft models of disease, response to treatment with vascular disruption is assessed via photoacoustic tomography (PAT) and magnetic resonance imaging (MRI). Photoacoustic imaging is a novel imaging modality based on the detection of ultrasound waves created by the absorption of nano-second pulsed laser energy within tissue chromophores. It has the spectral specificity of optical techniques whilst also achieving the high resolution of ultrasound. Haemoglobin is the main chromophore found in biological tissue and this modality is therefore ideally suited to imaging tumour vascularisation. Using a Fabry-Perot interferometer this thesis demonstrates for the first time the feasibility of using PAT for re-clinical research and the characterisation of typical tumour vascular features in a non-invasive non-ionising manner. Response to different concentrations of a vascular disrupting drug is then demonstrated, with novel insights in to how tumours recover from vascular damage observed. MRI of response to vascular disruption is also presented. As MRI is widely used in the clinic it can serve as a translational tool of novel imaging biomarkers, and serves to further understand the differences in response of pathologically vascularised of tumours. This thesis looks at markers associated with disruption of haemodynamics, using apparent diffusion (ADC) to elucidate onset of necrosis, increase in haemoglobin concentration (R2*) as indication of impaired flow, and arterial spin labelling (ASL) as a marker of tumour blood perfusion. This is shown in both subcutaneous and clinically relevant liver metastasis models. Taken as whole, the results from this thesis indicate that whilst understanding the response of the tumour vasculature to pharmacological intervention is complex, novel imaging techniques can provide invaluable translational information on the pathophysiology of tumours.

Compelling data has demonstrated the fundamental role of regulatory T cells (Tregs) in immuno-regulation. In vivo animal models indicate that recipient’s CD4pos CD25pos T cells adoptively transferred can prevent or control graft versus host disease (GvHD) and allograft rejection. Furthermore, it has been demonstrated that CD4pos CD25pos T cells from a third party could also be used as an adoptive cell therapy to ameliorate allo-responses. Compelling data supports that CD45RApos Tregs represent the most homogenous population among the overall Treg pool. The objective of this study is to test the "Third party approach" using CB Tregs to suppress alloresponses. Herein is shown that CB Tregs were mainly CD45RApos CD31pos (>80%), which specifically depicts RTE cells that confer a wide TCR repertoire. This study shows an optimized one-step isolation method using anti-CD25 microbeads that achieves high purity (90%) for CD4pos CD25high CD127low T cells and decreases substantially the level of effector T cells (<9% of CD4pos CD127high). Due to the low frequency of Tregs (1% from overall lymphocytes), the modality of pooling mismatch CB units for Tregs isolation was tested. The pooled CB Tregs showed constitutively potent suppression ability in vitro. Interestingly, in 40% of the cases a better suppression was seen with pCB Tregs compared to individual CB Tregs suggesting a “synergetic effect”. In summary, this study suggests that CB units fulfill the optimal properties for the isolation of bona fide Tregs, which can be isolated with the highest purity using a single step isolation method under GMP standards. Moreover, this study suggests that CB Tregs can be intentionally pooled and tailored under the required HLA matches for clinical settings.

Graft-versus-host disease (GvHD) is a major barrier to the successful outcome of allogeneic haematopoietic stem cell transplantation. This thesis investigates four novel strategies for managing GvHD: the role of extracorporeal photopheresis (ECP), the use of economic data to allow for funding of new treatments, the impact of a new dedicated GvHD clinic and the writing of national guidelines. The main hypothesis is that fortnightly ECP would improve signs and symptoms of chronic GvHD. A single centre retrospective analysis of 82 patients showed that this ECP schedule was effective with 94% of patients who completed 6 months of treatment achieving a complete or partial response. The information derived from this study was subsequently used to design a study to test this hypothesis in a prospective cohort of patients. 70% of patients who completed 6 months of ECP had a complete or partial response. Analysis of quality of life in a subset of patients showed a significant improvement in chronic GvHD symptom score (22 v 36, p=0.012) and dermatology quality of life index (3.4 v 6.9, p=0.009). A second hypothesis is that patients with GvHD have a higher readmission rate and economic burden than patients without GvHD. A retrospective review of 187 consecutive patients was undertaken which showed that the overall readmission rate was higher in patients with GvHD (86% (101/118) v 59% (41/69), p<0.001). The mean cost of readmission was higher in GvHD patients (£28860) than in non-GvHD patients (£13405; p=0.002) and in patients with grade III/IV GvHD (£40012) compared to those patients with grade I /II GvHD (£24560; p=0.038). This information was used to obtain funding for on-site ECP for acute GvHD. In addition, an exploratory evaluation of a new dedicated GvHD clinic is reported demonstrating the impact on referral patterns, quality of life and morbidity and mortality and the process of writing three national clinical practice guidelines is described.