Pancreatic Ductal Adenocarcinoma (PDAC) is one of the five leading causes of cancer-related deaths in the West, and this, largely, is due to metastatic disease. In order to better understand PDAC metastatic spread and identify novel therapeutic targets, we analysed the proteome of primary tumours and matched lymph node (LN) metastases. As frozen specimens of metastatic lesions are scarce, we examined formalin-fixed paraffin-embedded (FFPE) tissues. Whilst such tissue is in routine diagnostic use, the cross-linkages induced by fixation have, in the past, precluded proteomic investigation for research purposes. Recent technological advances have, however, overcome this technical limitation. Using laser capture microdissection (P.A.L.M system), we isolated malignant epithelia from seven FFPE primary PDAC tumours and matched LN metastases. Following dissection, samples were analysed in duplicate using Multidimensional Protein Identification Technology (MudPIT); this resulted in the identification of 1504 proteins, 854 of which were common to all samples analysed. Comparison of the obtained proteins with data from previous proteomics studies on pancreatic tissue, pancreatic juice, serum and urine resulted in a less than 30 % overlap, indicating that our study has expanded the current database of proteins expressed in this malignancy substantially. Statistical analysis further showed that 115/854 proteins (13.5%) were significantly differentially expressed (g-value ≥ 3.8). Two proteins, S100P and 14-3-3 sigma, with highly significant g-values were confirmed to be significantly differentially expressed (S100P: p = 0.05 and 14-3-3 sigma: p < 0.001) 4 in a larger series of 55 cases of matched primary PDAC and LN metastases using immunohistochemistry. We chose to investigate further the roles of S100P in lymphatic invasion in vitro and in vivo. By co-culturing a Panc1 S100P-overexpressing clone (S5L), or a vector control clone (V3L), with human dermal lymphatic endothelial cells (HDLEC), we were able to show that different receptors mediate S5L adhesion to resting and activated HDLEC as opposed to V3L; and that the presence of S5L cells in these co-cultures significantly increased permeability at one (p = 0.02), four (p = 0.002) and eight (p = 0.007) hours post-seeding, and significantly increased translymphatic endothelial migration at 72 hours (p = 0.006). Using the V3L and S5L cell lines, which were transduced to express luciferase, we also created an orthotopic mouse model of PDAC, as well as experimental metastatic mouse models, in CD1 nude mice. These models were used to evaluate the effects of S100P on primary tumour growth, metastasis and site-specific growth. S100P was only found to significantly increase primary tumour growth in this model (n = 10 animals/group), both by bioluminescence (p = 0.002) and tumour weight (p = 0.01). No metastases (spontaneous and/or experimental) were seen however. Thus, this model can be used to evaluate the anti-tumour efficacy of novel therapies to S100P in the future.

Pancreatic ductal adenocarcinoma (PDAC) is the most common type of pancreatic cancer, and the 5th most common cause of cancer death in the UK. One of the peculiarities of this malignancy is its ability to invade nerves, a process called perineural invasion (PNI). PNI is found in almost 100% of PDAC, and is associated with poor prognosis, tumour recurrence and generation of pain. However, the molecular bases of PNI remain largely unknown. We investigated the molecular alterations underlying the neuro-epithelial interactions in PNI using one and two dimensional liquid chromatography – mass spectrometry (1D and 2D LC-MS) of laser microdissected PNI and non-PNI cancer from formalin fixed, paraffin embedded PDAC tissues. We also performed 1D LC-MS analysis of invaded and non-invaded nerves from the same cases. In addition, we developed an in vitro model of PNI using a co-culture system comprising PC12 cells, a rat pheochromocytoma cell line, as the neuronal element and PDAC cell lines. The overall proteomic profiles of PNI and non-PNI cancer appeared largely similar; of very few deregulated proteins, we have validated the up-regulation of antiapoptotic protein Olfactomedin 4 in PNI cancer using immunohistochemistry. In contrast, nerve samples demonstrated widespread molecular alterations characteristic of neuronal plasticity upon invasion by cancer cells. Immunohistochemistry confirmed the up-regulation of VGF in nerves from PDAC and chronic pancreatitis (CP) specimens compared to normal pancreas, as well as in invaded compared to non-invaded nerves in PDAC tissues. Furthermore, VGF expression strongly correlated with pain in CP; similar analysis in PDAC cases is still pending. Using the in vitro co-culture model, several PDAC cell lines were able to induce PC12 cells neuronal plasticity including survival, neurite extension as well as VGF expression, recapitulating thus the changes observed in human tissues. PDAC-induced PC12 plasticity was not mediated via NGF, a neurotrophin acting upstream of VGF and thought to be involved in the neuro-epithelial interactions. The induction of VGF expression was shown not to be necessary for PC12 cell survival, however, it contributed to the neurite extension induced by PDAC cell lines. In summary, based on proteomics analysis and in vitro modelling, we show the complex and intricate involvement and crosstalk of both tumoral and neural elements that are activated during perineural invasion in pancreatic cancer.

Replication-selective oncolytic adenoviruses represent a promising anticancer approach with proven efficacy in cancer cell lines and tumour xenografts in vivo. Anti-tumour efficacy, both in preclinical studies and clinical trials, was significantly improved in combination with chemotherapeutics in numerous cancers, including prostate cancer. It has been established that expression of the viral E1A gene is essential for the enhancement of cell killing in combination with cytotoxic drugs. The overall goal of this project is to identify specific E1A gene regions involved in the sensitization to the cytotoxic drugs mitoxantrone and docetaxel, the current standard of care for late stage prostate cancers, to enable the development of improved anti-cancer therapies. Specific regions in the E1A proteins bind to numerous cellular factors to regulate the host cell function and the viral life cycle, including the p300, p400 and pRb family proteins. This work was aimed at determining the mechanisms involved in the synergistic cell killing in prostate cancer cells in response to the combination of the replication-selective (oncolytic) mutant AdΔΔ with cytotoxic drugs. Previous findings suggested an enhancement of drug-induced apoptosis. I found that the small E1A12S protein, unable to induce viral replication, is sufficient to sensitize the prostate cancer cells, 22Rv-1 (AR+), and PC-3 and DU145 (AR-), to drugs. The non-replicating AdE1A12S-mutant AdE1A1104 (defective in p300-binding) could not sensitize the cells while mutants with intact E1A-p300 binding (AdE1A12S, AdE1A1102, AdE1A1108) and defective in p400- (AdE1A1102) or pRb-binding (AdE1A1108) potently sensitized all tested cell lines. In fact, all mutants except AdE1A1104 potently synergised with mitoxantrone and docetaxel to kill the prostate cancer cells. When comparing the non-replicating E1A12S mutants with the corresponding replicating E1A-deletion mutants (expressing E1A12S and 13S) synergy was demonstrated with all replicating mutants except dl1104, which caused an additive effect with mitoxantrone. We hypothesised that the synergistic cell killing is the result of pathway convergence through E1A-p300 and mitoxantrone-activated DNA-damage/apoptosis events. To address this I employed an extensive miRNA array screen to identify potential pathways. Several miRNAs were found to be differentially regulated in response to the combination of AdE1A12S with mitoxantrone compared to each single agent treatment. The majority of these miRNAs are reported to be part of cell death and survival pathways (e.g. apoptosis and autophagy) and to be differentially regulated in prostate cancer. To further investigate the role of these pathways, I determined changes in expression levels of key proteins that had previously been suggested to be targeted by the identified miRNAs, thereby preventing translation of the respective mRNAs. The greatest changes in protein levels in response to AdE1A12S and mitoxantrone were observed for Bcl-2, p-Akt, LC3BII and p62. Finally, I verified similar mechanisms of action when the oncolytic AdΔΔ was combined with mitoxantrone under synergistic conditions. These findings will direct future investigations aimed at dissecting the mechanisms of action for virus-induced sensitization to cytotoxic drugs and may aid in the development of improved therapies for prostate cancer by design of novel oncolytic mutants and combination strategies and/or identification of targets for small molecules inhibitors.

Androgen receptor (AR) cell signalling is active in most castration-resistant prostate cancer (PCa) tumours and suppression is hypothesized to impede cell proliferation. Hey1, a corepressor of AR is being investigated as a therapeutic transgene for late-stage PCa. A replication-defective recombinant adenovirus deleted for E1 and E3 and expressing Hey1 under a CMV promoter was constructed (Ad5Hey1). A dual luciferase reporter system demonstrated that Ad5Hey1 repressed AR activity in a dose dependent manner in miboleronestimulated 22Rv1 cells. Ad5Hey1 was cytotoxic in both AR-positive 22Rv1 and LNCaP and AR-negative DU145 cells. The doses required to kill 50% of cells (EC50) were comparable to those of AdE1A12S expressing the cytotoxic E1A12S gene from an identical vector. The mechanisms of Ad5Hey1-induced cell killing were investigated in 22Rv1 and DU145 cells. Using RNA interference towards AR or p53 in 22Rv1 cells we concluded both proteins were required for optimal cell killing by Ad5Hey1. In DU145 cells, with non-functional p53, Ad5Hey1 decreased levels of phospho- STAT3 and total STAT3 suggesting Ad5Hey1 might inhibit STAT3 signalling while the JAK1/2 inhibitor, AZD1480 was ineffective at sensitising DU145 cells to Ad5Hey1. Preliminary data therefore suggests Ad5Hey1 may interfere with JAK/STAT signalling in these cells. Cell-killing efficacy with Ad5Hey1 in combination with cytotoxic drugs currently used in the clinic for the treatment of late-stage PCa, mitoxantrone and docetaxel, resulted in a synergistic enhancement of cell death in 22Rv1 and DU145 cells. LNCaP cells were also sensitised to the drugs. Characterisation of the mode of cell killing demonstrated augmented mitochondrial membrane depolarisation and caspase-3 activation when combined with docetaxel in all cell lines and with mitoxantrone in 22Rv1 and LNCaP cells, typical of apoptotic death. In DU145 cells, the combination of Ad5Hey1 with mitoxantrone decreased the proportion of apoptotic cells suggesting cells are dying by alternative cell death mechanisms. In this thesis I have demonstrated that Ad5Hey1 potently eliminates PCa cells both in the presence and absence of functional AR or p53, and that cell killing is 6 improved in combination with cytotoxic drugs. I demonstrate that the mechanisms by which Ad5Hey1 acts as a cell death enhancer is mainly through cooperation with drugs on apoptotic pathways while other factors such as inhibition of survival are also involved. In conclusion, these data suggest that it is feasible to develop a future replication-selective adenovirus expressing Hey1 as a cytotoxic transgene to improve antitumour efficacy in vitro and in vivo, especially in combination with apoptosis-inducing drugs.

Oncolytic vaccinia virus has great potential in the treatment of cancer and two engineered strains have entered clinical trials. As the advent for oncolytic vaccinia virus as an approved therapy beckons, it is critical to consider some of the barriers that may hinder this progress. These include suboptimal delivery of the virus to tumour sites, incomplete destruction of the tumour mass, and a lack of full understanding of the way in which oncolytic vaccinia kills its target cells. This thesis attempts to address these issues, with a particular focus on ovarian cancer. As ovarian cancer is generally restricted to the peritoneal cavity, intraperitoneal delivery may be preferable over intravenous delivery. Here, it is shown that Lister-dTK, an engineered vaccinia strain, is able to selectively replicate in ovarian tumours, including metastases to the liver following intraperitoneal delivery. To determine whether Lister-dTK could potentially be used in combination with current therapies for ovarian cancer, the effect of cisplatin and Lister-dTK together was assessed in vitro but showed no improvement in overall cell death. In an attempt to further improve the anti-tumour efficacy of Lister-dTK, the extracellular matrix protein (ECM) decorin was expressed from the virus. Decorin interacts with various signalling pathways and is proposed to enhance virus spread. However, abrogation of EGFR and TGFβ signalling could not be demonstrated in vitro, nor could improved virus spread. In an intraperitoneal model of ovarian cancer, Lister-mDCN did not demonstrate enhanced efficacy over a control virus. To determine the mechanisms of ovarian cancer cell death induced by Lister-dTK, the roles of apoptosis, autophagy and necrosis were investigated. Whilst some features of both apoptosis and autophagy were observed, inhibition of these pathways did not attenuate Lister-dTK. It is proposed that necrosis is the primary cause of cell death but that this process may occur in a regulated manner.

Approximately 50% of all malignant pleural mesotheliomas (MPM) are deficient in argininosuccinate synthetase (ASS1), the rate-limiting enzyme in arginine biosynthesis, and are sensitive to arginine deprivation. This discovery in MPM has been translated into the clinic using the arginine depletor pegylated arginine deiminase (ADI-PEG20), which showed a halving in the risk of disease progression in a randomised phase II study. However, unstudied to date, stromal resistance to ADI-PEG20 may reduce its efficacy. Here, I studied the effect of macrophages, abundant in mesothelioma, on the tumour cytotoxicity of ADI-PEG20. A distinct pro-inflammatory cytokine gene expression signature involved in macrophage recruitment and activation was identified and validated in ADI-PEG20-treated ASS1 negative MPM cell lines. In vivo induction of pro-inflammatory cytokines was also seen in ADI-PEG20-treated patient plasma. Notably, in vitro co-culture experiments demonstrated a significant increase in ASS1 negative MPM cell viability upon co-culture with macrophages in the presence of ADI-PEG20. This was accompanied by a significant increase in ASS1 expression in co-cultured macrophages, with a corresponding increase in argininosuccinate lyase (ASL) expression in co-cultured tumour cells and a doubling in levels of the arginine precursor, argininosuccinate, in cell supernatant. The addition of argininosuccinate to tumour cell media rescued ASS1 negative MPM cells from ADI-PEG20 cytotoxicity, while the macrophage-mediated resistance to ADI-PEG20 was abrogated following ASL knockdown in MPM cells. Finally, xenograft studies demonstrated a significant reduction in tumour volume in mice treated with ADI-PEG20 in combination with macrophage depletion, compared with ADI-PEG20 alone. Collectively, the data indicate that as a result of metabolic 'cross-talk' between macrophages and ASS1 negative MPM cells, macrophages mediate MPM resistance to ADI-PEG20 via the provision of argininosuccinate. My studies provide a rationale for combining ADI-PEG20 with an inhibitor of macrophage recruitment in the treatment of ASS1-deficient mesothelioma.

The DNA Mismatch repair (MMR) pathway is responsible for the repair of base-base mismatches and insertion/deletion loops that arise during DNA replication. MMR deficiency is currently estimated to be present in 15-17% of colorectal cancer cases and 30% of endometrial cancers. MLH1 is one of the key proteins involved in the MMR pathway. MMR deficient tumours are often resistant to standard chemotherapies, therefore there is a critical need to identify new therapeutic strategies to treat MMR deficient disease. This study demonstrates that MLH1 deficient tumours are synthetically lethal with the mitochondrial-targeted agent Parthenolide which is known to induce reactive oxygen species (ROS) as one of its main mechanisms of action. Upon functional analysis, I show for the first time that loss of MLH1 is associated with deregulated mitochondrial function evidenced by a reduction in complex I expression and activity, reduced basal oxygen consumption rate and reduced spare respiratory capacity. This mitochondrial phenotype in the MLH1-deficient cell lines is accompanied by a reduction in mitochondrial biogenesis as evidenced by down regulation of pgc1β and decreased mitochondrial copy number. Furthermore, MLH1-deficient cancer cells have a decreased antioxidant defence capacity with reduced expression of the antioxidant genes NRF1, NRF2, Catalase, Glutathione peroxidase and SOD1 as well as increased ROS production when treated with Parthenolide. I further demonstrate that both MSH2- and MSH6-deficient cell lines also display deficiencies in complex I compared to their MMR-proficient counterparts. Taken together, the results of this study show a novel role for MLH1 in mitochondrial function and biogenesis. The MMR proteins MSH2 and MSH6 are also likely to have a role in the mitochondria. My results suggest that targeting the mitochondria may be a potential therapeutic strategy for the treatment of MMR and specifically MLH1 deficient disease.

Mutations in the tumour suppressor gene Adenomatous polyposis coli (APC) are found in 80 % of sporadic colorectal cancer (CRC) tumours and are also responsible for the inherited form of CRC, Familial adenomatous polyposis (FAP). In order to identify novel therapeutic targets for the treatment of APC mutated CRC, we have generated an in vitro model of APC mutant CRC using CRISPR-cas9 gene editing. Using the APC wildtype colorectal carcinoma cell line RKO, we targeted the cells with guide RNA (gRNA) targeting exon 2 or exon 15 (encodes 80 % of APC) of the APC gene. We generated isogenic cell lines which differed in the expression of APC, the controls were APC wildtype and the APC mutant (APC Lys736fs) cell lines expressed a truncated ~80 kDa APC protein. We used these cell lines to perform an siRNA screen against 720 kinases and kinase-related genes. We selected seven genes to investigate further, unfortunately none of the potential hits validated. Additionally, we performed an FDA-approved compound screen targeting over 1000 compounds. From this, we identified a group of HMG-CoA reductase (HMGCR) inhibitors known as statins, which selectively cause a greater loss in cell viability in the APC mutated cell lines, compared to the APC wildtype cells. Mechanistically, our data suggests this synthetic lethal relationship is due to a greater decrease in the anti-apoptotic protein survivin. We propose this is due to statins altering the localisation of Rac1, reducing Pak1 activation and reducing the level of Wnt signalling. This results in the reduction of the Wnt target gene survivin. We have successfully identified an FDA-approved family of compounds, which show synthetic lethality with the APC mutation in our in vitro model.

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第21686号 / 医博第4492号 / 新制||医||1036(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 妹尾 浩, 教授 小川 誠司, 教授 武田 俊一 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

The DNA Mismatch repair (MMR) pathway is responsible for the repair of base-base mismatches and insertion/deletion loops, formed during DNA replication. Mutations in MMR genes significantly increase the predisposition to cancer with MMR deficiency estimated to be present in 15-17 % of all colorectal cancers. 5-fluorouracil is the main treatment for advanced colorectal cancer however the majority of studies suggest that MMR deficient tumours are more resistant to 5-fluorouracil than MMR proficient tumours. Therefore, there is a critical clinical need to identify novel therapeutics to treat these tumours. To this end, we have performed a high-throughput compound screen, to identify compounds that cause selective lethality in MMR deficient cell lines. We identified the potassium-sparing diuretic drug, Triamterene, as selectively lethal in vitro and in vivo in MMR deficient cell lines. Our data suggest that this selectivity is through its antifolate activity, leading to the accumulation of reactive oxygen species and DNA double strand breaks in MMR deficient cells. Interestingly, we identified a requirement, for thymidylate synthase expression, the only de novo enzyme for dTTP synthesis for the Triamterene cytotoxicity. NRF2 and NRF2-induced antioxidants were regulated upon Triamterene treatment and thymidylate synthase silencing, therefore suggesting a role for the antioxidant response in Triamterene toxicity. Taken together, our results suggest Triamterene as a promising novel therapeutic for the treatment of MMR deficient cancers. In order to identify novel therapeutics to treat MMR deficient tumours, we have also performed a high-throughput siRNA screen, to identify genes that cause selective lethality in MMR deficient cell lines. We identified AURKA gene as synthetically lethal in MSH6 deficient cell lines which suggests AURKA as a promising novel therapeutic target for the treatment of MMR deficient cancers. Taken together, in this PhD thesis we have identified two novel therapeutic strategies for the treatment of MMR deficient cancers.