In ER+ breast cancer initial responses to antihormones are variable, complete responses are rare and resistance is eventually acquired by many patients. It is important to model these events to discover predictive markers of antihormone outcome and so targeted strategies can be developed to maximise antihormone effectiveness. To date, most studies have employed the MCF-7 cell line which fails to represent the variability of ER+ disease. Focusing on Faslodex, the thesis objective was to use 4 cell lines in vitro encompassing ER+/HER2- (MCF-7/T47D) and ER+/HER2+ (BT474/MDA-MB-361) disease to (i) characterise the magnitude of initial antihormone response, (ii) monitor the onset of resistance by prolonged treatment and (iii) detail gene expression changes during Faslodex treatment. All models were initially growth-inhibited by Faslodex, with superior responses in HER2- lines. Microarray analysis revealed gene cohorts affected by Faslodex treatment differed between HER2+ and HER2- models. While MCF-7, BT474 and MDA-MB-361 cells acquired Faslodex resistance, this failed to develop in the T47D line, providing a model of complete-response. A filtering process identified genes involved in the varying Faslodex responses and clinical relevance was determined using the NEWEST Faslodex clinical trial dataset. Of interest was the Faslodex-induction of CXCR4, as a potential mediator of acquired resistance, while suppression of the RET signalling pathway related to improved initial response in the ER+/HER2- setting. Importantly up-regulation of DCN by Faslodex was associated with improved Faslodex response in T47D cells and also with proliferation (Ki67) fall in the NEWEST clinical trial. shRNA knockdown of DCN reduced the sensitivity of T47D cells to Faslodex and enabled development of resistance. This thesis has successfully identified novel elements of Faslodex response and resistance and further work is now required to clarify the importance of these mediators and to determine if DCN could prove a useful clinical biomarker of Faslodex response.

The anti-viral acyclic nucleoside monophosphate compound Cidofovir has shown efficacy in treatment of Human Papillomavirus (HPV) associated genital intraepithelial neoplasia; however, the mechanism of action of Cidofovir in this setting has not been determined. This investigation focused on modifying nucleoside analogue compounds to increase their efficacy in HPV positive cell models of disease, in addition to determining the molecular mechanism of action of Cidofovir in premalignant HPV associated intraepithelial neoplasia. ProTide modification increases the efficacy of nucleoside analogue compounds by increasing their cellular permeability. Cidofovir was not amenable to ProTide manipulation; however, ProTide derivatives of its sister compounds, Adefovir and Tenofovir, were synthesized. Parent Adefovir and Tenofovir and a range of their respective ProTide modified daughter compounds were examined for inhibition of cell growth and effect on cell size and morphology in HPV positive and negative transformed cell lines. The most effective compounds were further examined for dose response in normal HPV negative untransformed Human Epidermal Keratinocytes (HEKs) and naturally HPV immortalized short term (NHIST) cell lines cloned from vulval and vaginal intraepithelial neoplasia biopsies. ProTide analogues displayed striking increased efficacy in comparison to their parent compounds; however, they did not show specificity to transformed or HPV positive cell lines. Cidofovir did not show specificity to HPV positive cells when examined for growth inhibitory effect in HPV positive and negative cell models. A variety of molecular processes were examined to determine the mechanism by which Cidofovir inhibits cell growth in validated NHIST cell lines and HEK cells. At the concentrations investigated, Cidofovir did not cause apoptosis in HPV positive or negative cells and its growth inhibitory effect appeared likely to be associated with cell cycle arrest or senescence. The effects of radiation on the molecular response induced by Cidofovir were also evaluated as previous studies suggested Cidofovir can function as a radiosensitizer. Cidofovir combined with gamma radiation did not result in apoptosis but was associated with an augmented molecular response in NHIST cell lines. On the contrary, Cidofovir combined with gamma radiation caused a major apoptotic response in HPV negative HEKs, suggesting such a combination could result in disadvantageous effects on healthy tissue if it were used in vivo.

In this study a CD31-expressing co-culture system was used to establish whether differential CD31/CD38 signalling may contribute to the poor prognosis associated with CD38 expression in CLL. Using western blot analysis, a PKB phospho-substrate antibody was used in combination with phospho-specific antibodies to identify ribosomal protein S6 and GSK3β as key signalling molecules that were augmented following short-term CD31-expressing co-culture. CD31-expressing co-culture did not alter the phosphorylation of STAT6. However, the addition of IL-4 to the cultures was a potent mediator of this signalling pathway. This highlights the specificity of signalling molecules to different external stimuli. Multi-colour flow cytometry was employed to quantify the expression of cell surface activation markers as well as intracellular phospho-proteins. The CD31-expressing co-culture led to a significant up regulation of the activation markers CD38, CD49d and CD69. Selective pharmacological inhibition of the phosphorylation of S6, STAT6 and ERK resulted in the down regulation of activation markers. Furthermore, the inhibition of p-STAT6 and p-ERK resulted in increased levels of apoptosis, which indicates that these signalling pathways are directly involved in CLL cell survival. Multi-colour flow cytometry was also used to quantitate the levels of phospho proteins, p-S6 and p-ERK. Similar to the results observed by antibody detection following western blotting, basal and inducible levels of p-S6 and p-ERK were elevated in primary CLL cells expressing high levels of CD38. Taken together, the work carried out in this project highlights the importance of using co-culture systems to stimulate CLL cells in vitro in order to mimic some of the key stimuli encountered in vivo. The dissection of the signalling pathways activated as a result of CD31/CD38 interactions provides a rational for the poor prognosis associated with elevated CD38 expression in this disease and identified candidate therapeutic targets that might particularly benefit this group of patients.

In this thesis, we aimed to identify genetic factors that influence the risk of colorectal cancer (CRC). We also sought alleles that contribute to the likelihood of extreme adverse reactions to treatment. We validated five previously identified low penetrance variants using our training phase cohort, consisting of 2,186 advanced CRC (aCRC) from the COIN and COIN-B trials and 2,176 geographically matched controls. Using this cohort we also identified a variant in RAD1 that was significantly associated with risk (X2=13.51, P=2x10-4). However, we failed to replicate these findings in an aCRC validation cohort consisting of 1,053 cases and 1,397 geographically matched controls (X2=2.76, P=0.1), potentially as a result of a lack of power due to insufficient sample numbers. We identified ten patients from the COIN trial with severe peripheral neuropathy associated with oxaliplatin (PNAO) treatment. Through exome resequencing we identified a novel stop gain variant (Ser613X) in the nucleotide excision repair gene (NER), ERCC4. Following analysis of 54 additional patients from the COIN trial with PNAO, we identified three rare nonsynonymous variants (Pro379Ser, Arg576Thr and Glu875Gly) that were predicted to interfere with protein function. Consistent with the rare variant hypothesis of common disease, two of these variants were seen to collectively contribute to the risk of the phenotype (7/63 [11.11%] of patients with PNAO compared to 86/1763 [4.88%] of patients without PNAO; X2=4.89, P=0.03). Using the fission yeast, Schizosaccharomyces pombe, we sought to elucidate functional effects of these variants in ERCC4 by creating a model system. Using cre recombinase mediated cassette exchange, we introduced the variants of interest into the ERCC4 homolog, rad16. Following treatment with a range of DNA damaging agents, we observed an increased sensitivity following introduction of the novel stop gain, indicating a defect in the NER pathway. Additionally, there was a clear pattern of oxaliplatin-specific sensitivity of strains with the introduced rare nonsynonymous variants, suggesting a defect of XPF in other repair processes associated with interstrand crosslinks.

B-Cell Lymphoma 3 (Bcl-3) is a proto-oncogene modulating the nuclear factor κB (NF-κB) signalling pathway, which had been first identified at the site of a t (14,19) translocation in B-cell chronic lymphocytic leukemia. Deregulated Bcl-3 expression has been reported in various tumours and in breast tumours Bcl-3 was found to be significantly elevated compared to the normal adjacent tissue. Activated NF-кB signalling is detected in 86% of ER-negative and HER2+ breast tumours. The currently used monoclonal antibody against HER2 (trastuzumab), produces remissions in 10-15 % of HER2+ breast cancers, thus the identification of novel therapeutic targets is essential. Previous in vivo studies used ErbB2 (MMTV/neu) mouse models deficient in Bcl-3 to understand the role and underlying mechanism of Bcl-3 in mammary tumour formation and progression. Even though Bcl-3 deficiency in vivo did not affect the primary tumor growth, the occurrence of developed lung metastases was reduced by 40%. Despite these effects on tumour progression, no effect on normal mammary gland function was observed. Based on the this evidence, the inhibition of Bcl-3 seems a promising therapeutic strategy, especially for HER2+ breast cancers, mainly in the prevention of secondary-tumour seeding and spread at distal sites. Using a Bcl-3 binding mutant we have shown that Bcl-3 function can be inhibited by disruption of binding to its cognate partners, p50 and p52 from the NF-κB family. Using molecular modelling we constructed a model of the Bcl-3-p50 complex, identified a novel protein-protein interaction domain and performed initial virtual screening for potential small molecule inhibitors. Candidate compounds were evaluated in human tumorigenic (MDA-MB-231) and non-tumorigenic cell lines (HEK-293) in cell based assays determining the effect on disruption of Bcl-3—p50 binding, as well as the change in NF-κB activity and migration ability. A lead compound was identified with low nM activities in all three cell based assays. Analogues of the lead compound were designed and synthesized in order to evaluate the structure-activity relationships (SAR). Important moieties in the structure were identified and the structural analogue 15f had almost 10 fold increase in the activity over the lead compound. More structural analogues will be analysed in the near future to further explore the SAR of the lead compound. Meanwhile in vivo xenografts experiments are evaluating the efficacy of the lead compound and its analogue 15f on the ability to suppress the spread and growth of secondary tumours.

Urothelial cell carcinoma (UCC) of the bladder is a common malignancy worldwide, causing considerable morbidity and mortality. It is unique among epithelial carcinomas in respect of the fact that it has divergent pathways of tumourigenesis. Low-grade papillary tumours which frequently recur, but seldom convert to muscle invasive tumours harbour mutations that activate the MAPK pathways, as a consequence of oncogenic mutations in FGFR3 or HRAS. However, in contrast, the high-grade muscle invasive tumours that readily metastasise have been shown to have defects in the p53 and retinoblastoma (RB) protein pathways Transgenic mice have allowed us to analyse the molecular basis of initiation, invasion and progression of many human cancers. These mouse models increase our understanding of the disease process as well as providing targets for developing novel therapeutic approaches. In UCC there has been a paucity of models that readily mimic the human disease. Although deregulation of the Wnt signalling pathway has been implicated in urothelial cell carcinoma (UCC), the functional significance is unknown. Recent studies have demonstrated the importance of this pathway in UCC progression, thus I endeavoured to test its importance both as a “driver” mutation, as well as a “progressor” mutation in more established UCC mutations. Thus I targeted expression of an activated form of β-catenin to the urothelium of transgenic mice using Cre-Lox technology. Expression of this activated form of β-catenin led to the formation of localised hyperproliferative lesions by 3 months, which did not progress to malignancy. Furthermore expression in UroIICRE+ β-cateninexon3/+ mice showed marked upregulation of the PTEN tumour suppressor protein that appears to be a direct consequence of activating Wnt signalling in the bladder. I therefore combined PTEN deficiency with β-catenin activation, which resulted in rapid formation of papillary UCC by 6 months. These tumours had increased pAKT signalling and were dependent on mTOR. Importantly in human UCC, there was a significant correlation between high levels of β-catenin and pAKT (and low levels of PTEN) (p<0.01, n=80). Taken together, these data suggest that deregulated Wnt signalling plays a role in driving UCC, and human UCC that have high levels of Wnt and PI3 kinase signalling may be responsive to mTOR inhibition. I next expressed oncogenic K-Ras or H-Ras in the urothelium alone, and in urothelial cells expressing an activated β-catenin. Although Ras activation was not sufficient to drive tumourigenesis, Ras activation combined with β-catenin activation in UroIICRE+ β-cateninexon3/exon3 K-RasG12D/+ and UroIICRE+ β-cateninexon3/exon3 H-RasQ61L mice rapidly developed UCC. These tumours had upregulation of pERK1/2 with minimal levels of pAKT. Importantly in human UCC, there was a significant correlation between high levels of β-catenin and pERK1/2 (p<0.01, n=80). This data further supports the role of deregulated Wnt signalling and its co-operation with Ras in bladder carcinogenesis. I observed upregulated p21 expression in our UroIICRE+ β-cateninexon3/+urothelial lesions and thus postulated that p21 may be acting as a block to tumourigenesis in the lesion. It was highly relevant to observe tumour formation in the double mutant UroIICRE+ β-cateninexon3/exon3 p21-/- mice. FGFR3 gene is frequently mutated in superficial urothelial cell carcinoma (UCC). To test the functional significance of FGFR3 mutations as a “driver” of UCC, expression of mutated Fgfr3 was targeted to the murine urothelium using UroIICRE+ promoter. These FGFR3 mutations (K644E and K644M) had no effect on bladder homeostasis or tumourigenesis up to 18 months of ages. Even when these mutations were combined with β-Catenin or Ras activating mutations, no urothelial dysplasia/hyperplasia or UCC was observed. This suggests that other alterations are required that can cooperate with FGFR3 activation to cause UCC. Interestingly, however, due to sporadic ectopic Cre recombinase expression in the lung and skin of these mice, FGFR3 mutation caused skin papilloma and promoted lung tumourigenesis in cooperation with K-Ras and β-Catenin activation, respectively. This confirmed that the mutant Fgfr3 was functional and that FGFR3 cooperates with other genetic events involving Ras and Wnt pathways to promote tumourigenesis in a context dependent manner and support the hypothesis that activation of FGFR signaling contributes to human cancer.

It has previously been demonstrated that p53 loss and Moloney murine leukaemia virus (MMLV) infection are weakly collaborative in T cell lymphomagenesis. The work described in this thesis aimed to investigate the role of p53 in MMLV-induced lymphoma and explore the basis of their relatively weak collaboration. The effects of p53 loss in combination with MMLV were explored first in primary fibroblasts. While no viral cytopathology was observed in these cells, it was found that the virus conferred a growth advantage on infected cells that was additive with the effects of p53 loss. MMLV has been reported to induce a preleukaemic phase of apoptosis in the thymus of infected mice and the role of p53 in this phenomenon was investigated next. The response was found to be short-lived, with peak onset varying according to host mouse strain. No evidence of direct induction of p53 was observed in thymus tissue in vivo, but evidence on the p53 dependence of the MMLV-induced apoptosis was equivocal, yielding only provisional conclusions. The role of p53 was then examined at the later stages of MMLV disease and tumours were analysed for loss of heterozygosity of the wild type p53 allele and expression of functional p53 protein with prolonged in vitro and in vivo passage. Two mouse models were used; a cohort of MMLV-infected mice and also a Runx2/Myc surrogate model which develop tumours after a significantly reduced latent period. The data suggested that whilst p53 loss is obligatory for in vitro culture of tumour cells, it is not required for the growth of tumours passaged in vivo. The sparing effect was not absolute, however, and there was some evidence of p53 pathway inactivation and allele loss in progressing tumours. In summary, I conclude that the weak synergy observed between MMLV and p53 loss is principally due to functional overlap, largely driven by activation of MMLV target genes such as Myc and Runx2 that collaborate to suppress p53-dependent pathways in vivo.

This thesis describes how both LIM Kinase 1 and LIM Kinase 2 are both important regulators of cell invasion. Chapter 3 presents data that shows that inhibition of LIMK function blocks the collective invasion of MDA MB 231 breast carcinoma cells in a three-dimensional matrix. Although LIMK was not required for cell motility in two dimensions, a novel role for LIMK in both extracellular matrix degradation and deformation activities was shown in three dimensions in Chapter 4. Consistent with matrix remodeling being a requirement for path generation by leading cells in collective invasion, LIMK activity was also shown to be required by leading cells in MDA MB 231 collective invasion. However, it was also discovered that LIMK activity was not required for path following MDA MB 231. The importance of Cofilin activity as a conduit of LIMK activity during invasion was investigated in Chapter 5, a well as potential novel protein interactions of Cofilin. The identification of novel substrates of LIMK was attempted in Chapter 6, leaving prospective routes of investigation to further elucidate the roles of LIMK1 and LIMK2 in cells. The main findings presented in this thesis reveal a requirement for LIMK activity in the path generation function of leading cells in collective invasion. Given that individual invading cells must generate their own paths, these results lend support to the continued development of LIMK inhibitors to counter tumor cell invasion and metastasis.

Rab25 is a small GTPase of the Rab11 family that is known to control endosomal trafficking and the return of internalised receptors to the plasma membrane. Recent reports that Rab25 can act as either a promoter or suppressor of tumourigenesis highlight the need for greater understanding of how it controls integrin trafficking. In this thesis I show data indicating that Rab25 permits the sorting of ligand-occupied, active conformation α5β1 integrin to late endosomes/lysosomes. Novel photoactivation and biochemical approaches show that integrins thus sorted are not degraded, but rapidly recycled from late endosomes/lysosomes to the plasma membrane. This requires the Chloride Intracellular Channel 3 (CLIC3); a protein resident in late endosomes/lysosomes of previously unknown function, that is upregulated in Rab25-expressing cells. Here I show that CLIC3 exists in close proximity with active α5β1 in late endosomes/lysosomes and is required for cancer cell invasion into fibronectin–containing three-dimensional matrices. Clinical data indicate that CLIC3 is expressed in ovarian cancers, pancreatic ductal adenocarcinoma (PDAC) and pancreatic intraepithelial neoplasea (PanIN), but is absent from normal pancreas and tumour-associated stroma. Moreover, Kaplan-Maier analyses demonstrate a strong correlation between high levels of CLIC3 (at both mRNA and protein levels) and poor prognosis in operable cases of PDAC. Taken together, the findings presented in this thesis identify CLIC3 both as a regulator and component of a novel vesicular transport route that returns lysosomally-targetted integrins to the plasma membrane, and as an independent prognostic indicator in pancreatic cancer.

The development of new biomarkers for cancer patients would be advantageous in population screening for the early detection of cancers, pathological diagnosis, assessment of prognosis, tailoring treatment to individuals, and assessment of treatment response. With this in mind different proteomic approaches were used to identify biomarkers which could potentially aid prognosis and predict response in gastrointestinal and ovarian cancer. Raf Kinase Inhibitor Protein (RKIP) was originally purified from bovine brain extracts and named phosphatidylethanolamine-binding protein (PEBP). It has subsequently been shown to be a widely expressed and highly conserved protein. Several recent studies have suggested that RKIP may suppress metastasis in melanoma, prostate, and breast cancer, as reduction or loss of RKIP expression was observed in metastatic cell lines and metastatic tissue. In this part of the project RKIP expression was assessed by immunohistochemistry in tissue microarrays (TMA) from patients with colorectal and ovarian cancer. The results confirmed the findings of earlier studies and suggest that the level of RKIP expression is significantly and inversely associated with metastatic disease and can predict the risk of metastatic relapse in patients with no evidence of metastases at presentation. The level of RKIP expression as a prognostic factor was independent of sex, age, tumour site, mitotic index, lymphovascular invasion and tumour stage. Cytokeratin 18 (CK18) is an epithelial-specific cytokeratin that undergoes cleavage by caspases during apoptosis. Measurement of caspase-cleaved (CK18-NE) or total cytokeratin 18 (CK18) from epithelial-derived tumours could be a simple, non-invasive way to monitor or predict responses to treatment. Soluble plasma CK18-NE and CK18 were measured by ELISA from 73 patients with advanced gastrointestinal adenocarcinomas before treatment and during chemotherapy, as well as 100 healthy volunteers. Both CK18-NE and total CK18 plasma levels were significantly higher in patients compared to the healthy volunteers (p=0.015, p<0.001). The total CK18 baseline plasma levels prior to treatment were significantly higher (p=0.009) in patients who develop progressive disease than those who achieve partial response or stable disease and this correlation was confirmed in an independent validation set. The peak plasma levels of CK18 occurring in any cycle following treatment were also found to be associated with tumour response, but peak levels of CK18-NE did not reach significance (p=0.01, and p=0.07, respectively). A surface-enhanced laser desorption-ionisation mass spectrometry (SELDI-MS) pilot study on serum from 8 oesophageal cancer patients and 8 healthy volunteers revealed a novel biomarker, ~4kDa, downregulated in patients (p=0.012). An expanded 30 tumour/normal study was performed for validation which confirmed the down-regulation of this potential biomarker (p<0.0001). Attempts to identify tentatively suggested that the peptide may be inter-alpha-trypsin inhibitor heavy chain H4 precursor, which was interesting as a cleavage fragment of inter-alpha -trypsin inhibitor heavy chain H4 had been previously found to be up-regulated in patients with ovarian cancer, and down-regulated in patients with breast cancer. However, it was not possible to confidently confirm this identification. In a further part of this study, haptoglobin was found to be significantly more abundant in the serum from patients with oesophageal cancer compared to healthy volunteers. It was straightforward to isolate and identify and would be amenable to immunoassay as there are good antibodies available for confirmation. In conclusion, with the current lack of effective markers of metastatic relapse in colorectal cancer, a straightforward test like RKIP expression in the primary tumour may be a very cost-effective way to identify which patients may derive greater benefit from adjuvant treatment and closer post-operative surveillance, and in patients with advanced gastrointestinal malignancy levels of plasma CK18 are a potential marker of tumour response.