Medulloblastoma (MB) is the most common malignant pediatric brain tumor. Out of the four molecular subgroups (WNT, SHH, Group 3 and Group 4), Group 3 patients face the highest incidence of leptomeningeal spread and overall patient survival of less than 50%. Current clinical trials for recurrent MB patients based on genomic profiles of primary, treatment-naïve tumors, provide limited clinical benefit since recurrent metastatic MBs are highly genetically divergent from their primary tumors. The paucity of patient matched primary and recurrent MB samples has contributed to the lack of molecular targets specific to medulloblastoma recurrence, limiting relapsing MB patients to palliation. Our previous in silico analyses revealed enriched expression of many stem cell self-renewal regulatory genes in Group 3 MB. In this work, I have set out to investigate whether by identifying genes contributing to self-renewal of Group 3 MB cells, we can characterize a population of cells responsible for therapy evasion and subsequent tumor relapse. Initially, we have adapted the existing COG (Children’s Oncology Group) protocol for children with newly diagnosed high-risk MB for treatment of immuno-deficient mice intracranially xenografted with human MB cells. Cell populations recovered separately from brains and spines mice during the course of tumor development and therapy were comprehensively profiled for gene expression analysis, stem cell and molecular features to generate a global, comparative profile of MB cells through therapy. Additionally, we have investigated therapeutic potential of small molecules targeting BMI1, a known self-renewal regulating gene. In the setting of recurrent Group 3 MB, pharmacological inhibition of BMI1, led to a remarkable decrease in cell proliferation and self-renewal in vitro as well as reduction of local and spinal metastatic disease in vivo. Finally, by combining the established therapy-adapted patient-derived xenograft mouse model and BMI1 inhibitor, PTC-596, we have demonstrated an additive effect of two modalities and provided the pre-clinical data for the upcoming Phase I trial. Biological investigations into the drivers of MB recurrence will lead to development of new therapeutic options for children who are frequently limited to palliation. Clinically relevant mouse models of MB recurrence can serve as platforms for pre-clinical testing and validation of new treatments aimed to provide therapeutic intervention rather than palliation. / Thesis / Doctor of Philosophy (PhD) / Medulloblastoma is the most common type of brain cancer that affects children. Out of the four main subgroups of medulloblastoma, tumors in Groups 3 and 4 are the most aggressive and are associated with a low overall survival in children diagnosed with this type of brain cancer. These two subtypes of medulloblastoma also account for the largest number of patients in which gold standard therapies fail and no additional therapies are available. Several studies have shown the existence of few cells within the tumor that alone can drive tumor growth. The aggressive behavior of these cells has in part been attributed to dysregulation of genes involved in cell replication and division. Further studies that will focus on understanding the significance of genes that regulate cell growth and replication can help discover a population of cells that is capable of evading therapy and contribute to tumor relapse. The identification and characterization of such population can lead to development of novel treatments for the children affected with aggressive medulloblastoma. In my thesis, I have developed a mouse model that replicates the aggressive therapy given to the medulloblastoma patients in order to study cells capable of escaping the harsh treatment and drive tumor comeback. Next, by profiling the gene expression and functional attributes of those cells, we identified genes that contribute to regulation of cell division and growth. The effects of both increasing and decreasing the activity of those genes were then tested in cells grown in the dish. Subsequently, the most promising results were verified in the established mouse models. The main objective of my thesis was to discover new opportunities in treatments the most aggressive type of brain cancer affecting children, and thus not only improve the quality of treatment but also the overall survival of patients with medulloblastoma.

The purpose of this thesis was to characterize cancer metabolism in vitro using epithelial ovarian cancer as a model on an untargeted, systems-level, basis with particular attention paid to the difference between cancer stem cell metabolism and cancer cell metabolism. Two-dimensional gas chromatography coupled to mass spectrometry was used to measure the metabolite profiles of the ovarian cancer and cancer stem cell lines under normal baseline conditions and also under chemotherapeutic and environmental perturbations. These two cell lines exhibited significant metabolic differences under normal baseline conditions and results demonstrated that metabolism in the ovarian cancer stem cell line was distinct from that of more differentiated isogenic cancer cells, showing similarities to stem cell metabolism that suggest the potential importance of metabolism for the cancer stem cell phenotype. Glucose deprivation, hypoxia, and ischemia all perturbed ovarian cancer and cancer stem cell metabolism, but not in the same ways between the cell types. Chemotherapeutic treatment with docetaxel caused metabolic changes mostly in amino acid and carbohydrate metabolism in ovarian cancer cells, while ovarian cancer stem cell metabolism was not affected by docetaxel. Overall, these metabolic differences between the two cell types will deepen our understanding of the metabolic changes occurring within the in vivo tumor and will help drive development of cancer stem cell targeted therapeutics.

Small cell lung cancer (SCLC) is a highly aggressive malignancy with extreme mortality and morbidity. Although initially chemo- and radio-sensitive, almost inevitable recurrence and resistance occurs. SCLC patients often present with metastases, making surgery not feasible. Current therapies, rationally designed on underlying pathogenesis, produce in vitro results, however, these have failed to translate into satisfactory clinical outcomes. Recently, research into cancer stem cells (CSCs) has gained momentum and form an attractive target for novel therapies. Based on this concept, CSCs are the cause of neoplastic tissue development that are inherently resistant to chemotherapy, explaining why conventional therapies can shrink the tumour but are unable to eliminate the tumour completely, leading to eventual recurrence. Here I demonstrate that SCLC H345 and H69 cell lines contain a subset of cells expressing CD133, a known CSC marker. CD133+ SCLC sub-population maintained their stem cell-like phenotype over a prolonged period of culture, differentiated in appropriate conditions and expressed the embryonic stem cell marker Oct-4 indicating their stem-like phenotype. Additionally, these cells displayed augmented clonogenic efficacy, were chemoresistant and tumorigenic in vivo, distinct from the CD133- cells. Thus, the SCLC CD133 expressing cells fulfil most criteria of CSClike definition. The molecular mechanisms associated with CD133+ SCLC chemoresistance and growth is unknown. Up-regulated Akt activity, a known promoter of resistance with survival advantage, was observed in CD133+ SCLC cells. Likewise, these cells demonstrated elevated expression of Bcl-2, an anti-apoptotic protein compared to their negative counterpart explaining CD133+ cell chemoresistance phenotype. Additionally, CD133+ cells revealed greater expression of neuropeptide receptors, gastrin releasing peptide (GRP) and V1A receptors compared to the CD133- cells. Addition of exogenous GRP and arginine vasopressin (AVP) to CD133+ SCLC cells promoted their clonogenic growth in semi-solid medium, illustrating for the first time neuropeptide dependent growth of these cells. A novel peptide (peptide-1) was designed based on the known structure of the substance P analogues that have shown benefit in animal models and in early clinical trials. This compound inhibited the growth of SCLC cells in in vitro with improved potency and stability compared to previous analogues and reduced tumorigenicity in vivo. Interestingly, peptide-1 was more effective in CD133+ cells due to increased expression of neuropeptide receptors on these cells. In conclusion, my results show that SCLC cells retain a sub-population of cells that demonstrate CSC-like phenotype. Preferential activation of Akt and Bcl-2 survival pathways and enhanced expression of neuropeptide receptors contribute to CD133+ SCLC chemoresistance and growth. Therefore, it can be proposed that CD133+ cells are the possible cause of SCLC development, treatment resistance and disease recurrence. Despite being chemoresistant, CD133+ cells demonstrated sensitivity to peptide-1. The identification of such new analogue that demonstrates efficacy towards resistant CD133+ SCLC cells is a very exciting step forward in the identification of a potential new therapy for resistant disease.

Like with many cancers, a single ovarian tumor can display remarkable diversity in genetics, epigenetics, expression profiles, microenvironment and cell differentiation and plasticity. This so-called intratumoral heterogeneity (ITH) is thought to greatly increase mortality by enabling tumors to adapt quickly to therapy, metastasize, and recur, thus the study of ITH holds great clinical significance. Clonal evolution and cancer stem cell (CSC) theory are two models for the initiation and propagation of a tumor, which offer differing views on the way that ITH is developed and maintained. In the clonal evolution model, cancer arises from a single cell and, through genetic instability, proliferates into a diverse population of daughter cells, which develop additional mutations and undergo Darwinian selection under the influence of the tumor microenvironment. Each cell of the clonal evolution model may be capable of initiating a tumor independently. In CSC theory, cancer arises from the transformation of a stem cell that has the capacity to self-renew and differentiate into a diverse population of daughter cells. Each cell is NOT capable of tumorigenesis as most are terminally differentiated and do not harbor self-renewing capabilities. According to CSC theory, small, rare subpopulations of CSCs persist throughout chemotherapy and are responsible for repopulating the heterogeneous tumor post-treatment. The hypothesis that CSCs may play a role in ovarian cancer progression is the subject of this thesis. Many studies have detected the presence of stem cell markers and dysregulated stem cell signaling pathways in ovarian cancer, but doubts remain as to the existence of ovarian CSCs; critics have pointed out inherent flaws in experimental designs meant to identify and characterize CSCs. For example, the presence of cancer cells which express the stem cell marker CD133 has been correlated to both positive and negative impacts on prognosis. Further challenging the study of ovarian CSCs is the lack of consensus on the true cell of origin for ovarian cancer - whether it be from the fallopian tube epithelium or ovarian surface epithelium, or elsewhere in the peritoneal cavity - this will have important implications for the identification and characterization of tumorigenic ovarian CSCs. Advocates of clonal evolution theory have put forth incredible effort to reveal the extent of inter and intra-tumoral heterogeneity in ovarian cancer, and from these data there has arisen a general consensus that cancer cell populations do evolve in a step-wise fashion, accumulating additional mutations over time. The involvement of cancer stem cells in this progression and how exactly they fit in (as a cell of origin or arising from genetic mutations), as well as their significance for different cancer types, is a question worth answering. Despite the challenges facing the study of ovarian CSCs, the clinical impact of cells with stem-like properties has been repeatedly demonstrated, especially with regard to metastatic processes and chemoresistance. Moreover, new drugs which target stem cell pathways have proven effective in the treatment of ovarian cancer. The existence of a rare subset of cells that have enhanced tumor-initiating properties is apparent in ovarian cancer, and more work is needed to characterize the unique identifiers and behavior of these cells in vivo. Future experiments involving lineage tracing promise to deepen our understanding of the nature of ovarian CSCs and address whether normal stem cells might serve as the cell of origin.

C-X-C chemokine receptor type 4 (CXCR4) is known to regulate lung, pancreatic and prostate cancer stem cells. In breast cancer, CXCR4 signalling via stromal cell-derived factor-1 (SDF-1) has been reported to be a mediator of metastasis, and is linked to poor prognosis. However its role in normal and malignant breast stem cell function has not been investigated. Anoikis-resistant (AR) cells were collected from mammosphere culture from 2 immortalised (MCF10A, 226L) and 3 malignant (MCF7, T47D, SKBR3) breast cell lines. For all cell lines, AR cells had a significantly higher mammosphere forming efficiency (MFE) than unsorted cells. The AR cells of the normal cell lines also demonstrated increased formation of 3D structures using the Matrigel assay. In vivo, MCF7 and T47D AR cells demonstrated increased capacity to form tumours compared with unsorted cells. This suggests that AR cells are enriched for normal and malignant breast stem cells. We performed an Agilent custom gene microarray and demonstrated up-regulation of CXCR4 mRNA expression in AR cells. CXCR4 protein expression was also higher in AR cells, shown by flow cytometry. The effects of AMD3100 (CXCR4 antagonist) and SDF-1 (CXCR4 ligand) on stem cell activity were investigated in the mammosphere assay. In the normal cell lines, SDF-1 significantly reduced MFE and this decrease was rescued by AMD3100. Incubation with AMD3100 increased MFE in the estrogen receptor positive breast cancer cell lines (MCF7 and T47D) and patient-derived metastatic tumour samples. AMD3100 reduced the self-renewal of T47D cells, as assessed by second generation mammospheres. MCF7 cells were retro-virally transfected to over-express CXCR4 or sorted for CXCR4 cell surface expression. Mammosphere formation was significantly increased in CXCR4+ and CXCR4 over-expressing cells compared with CXCR4- and parental cells. There was a greater reduction in self-renewal following AMD3100 treatment in the CXCR4 over-expressing cells compared with parental cells. AMD3100 has been shown to have an agonistic effect on the novel chemokine receptor CXCR7, a scavenging receptor for SDF-1. All cell lines demonstrated cell surface expression of CXCR7, measured by flow cytometry and mRNA expression. Potential interactions between CXCR4, CXCR7 and SDF-1 must be considered in future investigation of the role of CXCR4 signalling. Our data establish that CXCR4 signalling has contrasting effects on normal and malignant breast stem cell activity. CXCR4 influences self-renewal of malignant stem cells which may account for its role in tumorigenesis. CXCR4 signalling may be important in tumour formation at the sites of metastases as well as in cell migration. Its role in stem cell migration merits further investigation. In conclusion, CXCR4-targeted therapy, alongside current standards of care, may improve breast cancer outcomes.

Cancer is a stem cell disease and populations of cancer stem cells (CSCs) are evident in many cancer types. CSCs exhibit similarities to normal embryonic and adult stem cells: they are able to self-renew and have the potential to give rise to a diverse array of differentiated progeny. CSCs are responsible for driving tumourigenesis and metastasis, and are inherently resistant to chemotherapy and radiotherapy. This cell population is enriched after treatment and, as a result of their tumourigenic capability, can re-populate tumour growth resulting in patient relapse, often with increased chemotherapeutic resistance. Increasing evidence supports that only by targeting this population of cells will a cure for cancer be possible. Hence, it is essential to identify pathways within CSC populations that can be targeted therapeutically. Cyclooxygenase-2 (COX-2) is an enzyme associated with inflammation and disease, and is upregulated in many cancers types. The COX-2 / prostaglandin E2 (PGE2) signalling pathway is associated with increased tumour growth, metastasis, immune evasion and overall worse patient prognosis. Recent evidence has identified that COX-2 is further upregulated in CSC populations isolated from cancer cell lines. Previously, we have shown that inhibition of COX-2 reduces CSC sphere-forming ability, a characteristic of stem cell self-renewal, suggesting a role for COX-2 in maintaining CSC populations. This work was carried out in both human and canine osteosarcoma cell lines with similar results. Cancer in dogs is a major health concern among an aging pet population. Many cancer types exhibit similarities between these species, suggesting that naturally occurring canine cancer may be a potential model for the human disease. The aim of this PhD project was to investigate the role of COX-2 in CSCs in a comparative cancer study. CSCs that express stem cell markers have been isolated from a panel of canine and human cancer cell lines including, mammary carcinoma and transitional cell carcinoma of the urinary bladder. CSCs over-express COX-2 compared to non-CSCs, therefore to determine the role of COX-2 in CSC biology the selective COX-2 inhibitor mavacoxib, a non-steroidal anti-inflammatory drug currently licenced for treating osteoarthritis in dogs, was utilised. Our results demonstrate that inhibiting COX-2 has a multifaceted impact on CSC biology, including reducing self-renewal capacity, clonogenicity, proliferation, migration, invasion and in vivo tumourigenicity. To confirm that mavacoxib is mediating these CSC-specific effects via inhibition of COX-2 rather than through unknown off-target effects, we generated canine specific-small interfering RNA to specifically reduce gene expression of COX-2. Our results confirm that mavacoxib exerts its anti-tumour effects via inhibition of COX-2. This project has highlighted a plethora of CSC-specific COX-2 effects, and to gain further insight we compared the global gene expression profiles of CSCs compared to non- CSCs isolated from a canine bladder carcinoma cell line. This data revealed that both mavacoxib and COX-2 specific siRNA target similar pathways within the two cell populations, confirming that mavacoxib exerts its effects in a COX-2 dependent manner. Interestingly, mavacoxib reduced the expression of a number of stemness related genes in the CSC population, including NOTCH and Wnt, suggesting that mavacoxib can inhibit CSC related pathways. Our overall results are comparable between canine and human cancer cell lines supporting the concept of naturally occurring tumours in dogs as models for the human disease. In conclusion, COX-2 plays an important role not only in maintaining CSC populations but also in their function, and targeting COX-2 in CSCs may provide therapeutic benefit.

Chronic myelogenous leukemia is a myeloproliferative hematopoietic stem cell disease resulting from a reciprocal translocation that gives rise to BCR-ABL, a constitutively active tyrosine kinase. Imatinib and other tyrosine kinase inhibitors are currently standard therapy; however, point mutations often lead to drug resistance and disease relapse often occurs due to the persistence of quiescent leukemia stem cells that are shielded by stromal factors within the bone marrow microenvironment. In an effort to develop new therapies capable of eradicating these elusive cells, a novel approach has been proposed in which the biochemical properties of cancer cells are targeted. It has been established that one such property is oxidative stress due to the increased production of reactive oxygen species, which makes cancer cells especially dependent on their antioxidant systems to maintain redox homeostasis. Recent studies demonstrate that chaetocin, a mycotoxin produced by Chaetomium species fungi, possesses potent and specific antimyeloma activity due in part to its ability to inhibit thioredoxin reductase-1, a central oxidative stress remediation enzyme. In this study, the effectiveness of chaetocin against leukemia stem cells has been investigated using in vitro and in vivo murine chronic myelogenous leukemia models. Our results indicate that: chaetocin and imatinib function synergistically in decreasing cell viability, inducing apoptosis, and inhibiting the colony formation of chronic myelogenous leukemia cells in vitro; that chaetocin in combination with imatinib reduces leukemia stem cell frequency in vivo; that chaetocin increases intracellular reactive oxygen species levels; and that chaetocin does not disrupt the proliferation and differentiation of normal murine hematopoietic stem cells. Surprisingly, our results also show that while bone marrow stromal factors inhibit the activity of imatinib, they potentiate the activity of chaetocin, indicating that chaetocin could potentially be used to target leukemia stem cells within the bone marrow niche.

The primary treatment for colorectal cancer is surgery. Radiotherapy and chemotherapy, sometimes combined, are also frequently used to diminish recurrence risk. In response to radiation exposure, several cellular signaling cascades are activated to repair DNA breaks, prevent apoptosis and to keep the cells proliferating. Several proteins in the radiation response and cell survival pathways are potential targets to enhance the effects of radiation. The epidermal growth factor receptor (EGFR), which is frequently upregulated in colorectal cancer and exhibits a radiation protective function, is an attractive target for treatment. EGFR is activated by radiation which in turn activates numerous signaling pathways such as the PI3 kinase/AKT cascade, the RAS/RAF/ERK pathway and STAT leading to tumor cell proliferation. EGFR is also believed to interact with proteins in the DNA repair process, such as DNA-PKcs and MRE11. The cytotoxic effect of an affibody molecule (ZEGFR:1907)2, with high affinity to EGFR,  in combination with radiation produced a small, but significant, reduction in survival in a KRAS mutated cell line. However, not in the BRAF mutated cell line. The next step was therefore to target proteins downstream of EGFR such as AKT. There was an interaction between AKT and the DNA repair proteins DNA-PKcs and MRE11 and both AKT1 and AKT2 were involved in the radiation response. The knockout of both AKT isoforms impaired the DNA double strand break rejoining after radiation and suppression of DNA-PKcs increased the radiations sensitivity and decreased the DNA repair further. The AKT isoforms also affected the expression of cancer stem cell markers CD133 and CD44 which are associated with the formation of metastasis as well as radiation and drug resistance. The CD133 expression was associated with AKT1 but not AKT2, whereas the CD44 expression was influenced by the presence of either AKT1 or AKT2. AKT was also involved in cell migration, cell-adhesion and metabolism. Overall, these results illustrate the complexity in response to radiation and drugs in cells with different mutations and the need for combining inhibitors against several targets such as EGFR, AKT, DNA-PKcs, CD133 or CD44.

Self-renewing, chemoresistant cancer cells that contribute to cancer metastasis and patient relapse have properties similar to those of stem cells, and have been termed "cancer stem cells" (CSCs) in the literature. The identification of signaling pathways that regulate CSC development and/or function is an important step towards understanding why patients relapse, and towards development of novel therapeutics that specifically target CSC vulnerabilities. Recent studies have identified a role for the aryl hydrocarbon receptor (AHR), an environmental carcinogen receptor implicated in cancer initiation, in normal tissue-specific stem cell self-renewal. These studies inspired the hypothesis that the AHR plays a role in CSC development. To test this hypothesis, AHR activity in Hs578T triple negative and SUM149 inflammatory breast cancer cells was modulated with AHR ligands, shRNA, or AHR-specific inhibitors and their phenotypic, genomic, and functional CSC characteristics were evaluated. Aldehyde dehydrogenase (ALDH) was used as an epithelial stem cell marker for flow cytometry. Results demonstrate that: 1) ALDHhigh cells express elevated levels of Ahr and the AHR-driven gene that encodes cytochrome p450 isoform 1b1 (Cyp1b1), 2) AHR knockdown reduces ALDH activity, 3) AHR hyper-activation significantly increases ALDH1 activity, expression of stem cell- and invasion/migration-associated genes, and accelerates cell migration, 4) a highly significant correlation between Ahr or Cyp1b1 expression (as a surrogate marker for AHR activity) and expression of the CSC- and invasion/migration-associated gene sets was seen with genomic data obtained from 79 human breast cancer cell lines and over 1850 primary human breast cancers, 5) the AHR interacts directly with the transcription factors Sox2 and Runx1, and AHR ligands increase this interaction, 6) AHR knockdown inhibits tumorsphere formation in low adherence conditions, 7) AHR inhibition blocks the rapid migration of ALDHhigh cells and reduces ALDHhigh cell chemoresistance, and 8) AHR knockdown inhibits tumor growth and reduces tumor Aldh1a1, Sox2, and Cyp1b1 expression in orthotopic xenografts. These data suggest that the AHR plays an important role in development of CSCs in a large fraction of human breast cancers and that environmental AHR ligands may exacerbate breast cancer by enhancing expression of CSC-like properties.

It is well established that the Wnt pathway is associated with tumorigenesis in a wide range of human cancers, including a majority of breast cancers. However, due to diverse roles of Wnt signalling, therapeutic targeting has not yielded consistent results and underlying mechanisms remain unclear. Here, I show that breast cancer cell lines with high E-cadherin expression are resistant to TCF4 inhibitors and develop cancer stem cell characteristics. Conversely, cells with low levels of E-cadherin are very susceptible to cell death with the same treatment. My results suggest that breast cancer cells in an epithelial-like state, but not mesenchymal-like state, will be more responsive to therapeutic targeting of the Wnt/TCF pathway. Importantly, E-cadherin high cells show robust Akt activation, whereas E-cadherin low cells do not. Thus, combinational inhibition of both Wnt and Akt signalling is needed to effectively target breast cancer cells in both the epithelial and mesenchymal states.