Breast cancer remains the most common malignant disease in women worldwide. Despite the advantages of early detection and improved treatments, studies into the mechanisms that initiate and drive breast cancer progression are still required. Recent studies have identified RUNX1, which is an essential transcription factor for hematopoiesis, is one of the most frequently mutated genes in breast cancer patients. However, the role of RUNX1 in the mammary gland is understudied. In this dissertation, we examined the role of RUNX1 in both normal mammary epithelial and breast cancer cells. Our in vitro studies demonstrated that RUNX1 inhibits epithelial to mesenchymal transition (EMT), migration, and invasion, reflecting its tumor suppressor activity, which was confirmed in vivo. Moreover, RUNX1 also contributes significantly to inhibition of the phenotypes of breast cancer stem cells (CSC), which is responsible for metastasis and tumor relapse. We showed that Runx1 overexpression reduces the tumorsphere formation and cancer stem cell population. Overall, our studies provide mechanistic evidence for RUNX1 repression of EMT in mammary cells, anti-tumor activity in vivo and regulation of CSC-like properties in breast cancer. Our results highlight crucial roles for RUNX1 in preventing epithelial to mesenchymal transition and tumor progression in breast cancer. This RUNX1 mediated mechanism points to novel intervention strategies for early stage breast cancer.

Autophagy is a cellular process that delivers cytoplasmic materials for degradation by the lysosomes. Autophagy-related (Atg) genes were identified in yeast genetic screens for vehicle formation under stress conditions, and Atg genes are conserved from yeast to human. When cells or animals are under stress, autophagy is induced and Atg8 (LC3 in mammal) is activated by E1 activating enzyme Atg7. Atg8-containing membranes form and surround cargos, close and mature to become the autophagosomes. Autophagosomes fuse with lysosomes, and cargos are degraded by lysosomal enzymes to sustain cell viability. Therefore, autophagy is most frequently considered to function in cell survival. Whether the Atg gene regulatory pathway that was defined in yeast is utilized for all autophagy in animals, as well as if autophagy could function in a cell death scenario, are less understood. The Drosophila larval digestive tissues, such as the midgut of the intestine and the salivary gland, are no longer required for the adult animal and are degraded during the pupal stage of development. Cells stop growing at the end of larval development, and proper cell growth arrest is required for midgut degradation. Ectopic activation of the PI3K/Akt signaling induces cell growth and inhibits autophagy and midgut degradation. Down regulating PI3K/Akt pathway by Pten mis-expression activates autophagy. In addition, mis-expression of autophagy initiator Atg1 inhibits cell growth and knocking down autophagy restore PI3K/Akt activity. Together, these results indicate that autophagy and growth signaling mutually inhibit each other. Midgut destruction relies on the autophagy gene Atg18, but not caspase activation. The intestine length shortens and the cells undergo programmed cell size reduction, a phenomenon that also requires Atg18, before cell death occurs during midgut destruction. To further investigate whether cell size reduction is cell autonomous and requires other Atg genes, we reduced the function of Atg genes in cell clones using either gene mutations or RNAi knockdowns. Indeed, many Atg genes, including Atg8, are required for autophagy and cell size reduction in a cell autonomous manner. Surprisingly, Atg7 is not required for midgut cell size reduction and autophagy even though this gene is essential for stress-induced autophagy. Therefore, we screened for known E1 enzymes that may function in the midgut, and discovered that Uba1 is required for autophagy, size reduction and clearance of mitochondria. Uba1 does not enzymatically substitute for Atg7, and Ubiquitin phenocopies Uba1, suggesting Uba1 functions through ubiquitination of unidentified molecule(s) to regulate autophagy. In conclusion, this thesis describes: First, autophagy participates in midgut degradation and cell death. Second it reveals a previously un-defined role of Uba1 in autophagy regulation. Third it shows that the Atg genes are not functionally conserved and the requirement of some Atg genes can be context dependent.

The Insulin Receptor Substrate (IRS) proteins IRS-1 and IRS-2 are cytoplasmic adaptor proteins that organize and propagate intracellular signaling downstream of specific growth factor receptors, including the Insulin and Insulin-Like Growth Factor-1 Receptors (IR and IGF-1R, respectively). Despite sharing a high level of homology and the ability to stimulate Phosphotidylinositol-3-Kinase (PI3K) and Mitogen-Activated Protein Kinase (MAPK) signaling, IRS-1 and IRS-2 play distinct roles in mammary tumor progression. Specifically, IRS-1 promotes growth and proliferation, whereas IRS- 2 promotes motility, invasion, survival, aerobic glycolyis, and metastasis. To further understand the differences between IRS-1 and IRS-2, I investigated the mechanistic basis of IRS-2-mediated PI3K activation. I identified tyrosines in IRS-2 that mediate its recruitment and activation of PI3K in response to insulin and IGF-1 stimulation. Using a PI3K-binding deficient IRS-2 mutant, I demonstrated that IRS-2-dependent PI3K signaling promotes aerobic glycolysis through its ability to selectively regulate the phosphorylation of the Akt effector Glycogen Synthase Kinase-3β (Gsk-3β). I also performed a rigorous comparison of IRS-1 and IRS-2 signal transduction and their ability to regulate functions associated with tumor progression. These studies required the generation of a novel model system where IRS-1 and IRS-2 function could be compared in a genetically identical background. Using this model, I confirmed a role for IRS-1 in growth regulation and IRS-2 in tumor cell invasion, as well as expanded the understanding of differential IRS protein function by showing that IRS-2 more vi effectively promotes Akt activation. The model system I have established can be used for further characterization of IRS-1 and IRS-2-specific functions.

During regeneration, cells must coordinate proliferation and differentiation to rebuild tissues that are lost. Understanding how source cells execute the regeneration process has been a longstanding goal in regenerative biology with implications in wound healing and cell replacement therapies. Melanocytes are pigment-producing cells in the skin of vertebrates that can be lost during hair graying, injury and disease-related depigmentation. Melanoma is an aggressive skin cancer that develops from melanocytes, and it is hypothesized that melanoma cells have properties that are similar to melanocyte stem cells. To gain insight into melanocyte regeneration we set out to identify the source of regeneration melanocytes in adult zebrafish and the path through which progenitor cells reconstitute the pigment pattern. Using targeted cell ablation and single cell lineage-tracing analyses we identified that a majority of regeneration melanocytes arise through direct differentiation of mitfa-expressing progenitor cells. Concurrently, other mitfa-expressing cells divide symmetrically to generate additional mitfa-positive progenitors, thus maintaining regeneration capability. Using reporter assays and drug studies, we found that Wnt signaling gets turned on in progenitor cells during regeneration and Wnt inhibition after melanocyte ablation blocks regeneration. Based on our finding that Wnt signaling is active in differentiated melanocytes but not in the progenitor cells, we explored the role of Wnt signaling in tumor initiation. We found that approximately half of the melanomas are Wnt silent, and overexpression of dkk1b, a negative regulator of canonical Wnt signaling, accelerates melanoma onset. This work defines an unappreciated contribution by direct differentiation in melanocyte regeneration and suggests a broader role for this process in the maintenance of epithelial sheets. This study also identifies a shared pathway between melanocyte progenitors and melanoma cells, which could be applicable to other cancers.

The integrin α6β4 (referred to as β4) is expressed in epithelial cells where it functions as a laminin receptor. Integrin β4 is important for the organization and maintenance of epithelial architecture in normal cells. Particularly, β4 is shown to be essential for mammary gland development during embryogenesis. Integrin β4 also plays important roles in tumor formation, invasion and metastasis in breast cancer. However, the mechanism of how integrin β4 mediates breast tumor formation has not been settled. A few studies suggest that integrin β4 is involved in cancer stem cells (CSCs), but the mechanism is not clear. To address this problem, I examined the expression of β4 in breast tumors and its potential role involved in regulating CSCs. My data shows that β4 is expressed heterogeneously in breast cancer, and it is not directly expressed in CSCs but associated with a basal epithelial population. This work suggests that β4 can regulate CSCs in a non-cell-autonomous manner through the interactions between β4+ non-CSC population and β4- CSC population. My data also shows that β4 expression is associated with CD24+CD44+ population in breast tumor. To further study the role of β4 in breast cancer progression, I generated a β4 reporter mouse by inserting a p2A-mCherry cassette before ITGB4 stop codon. This reporter mouse can be crossed with breast tumor models to track β4+ population during tumor progression.

Breast cancer is the second most common malignancy in the world, accounting for over 1.7 million new diagnoses and an estimated 500,000 deaths per year (1). Overexpression of the receptor tyrosine kinase ErbB2, also known as Her2 or Neu, occurs in over 30% of breast cancers and correlates with metastasis, poor prognosis, and decreased survival (1, 2). Although therapeutics targeting ErbB2 show clinical efficacy, many patients display no initial response or develop drug resistance over time (2). A deeper understanding of the molecular basis of ErbB2-driven tumorigenesis is thus required for the development of improved therapeutic strategies. In vitro experiments suggest that activation of the c-Jun NH2-terminal kinase (JNK) pathway, a mitogen-activated protein kinase pathway, promotes proliferation, cellular invasion, and stem cell expansion in ErbB2-driven breast cancer (3, 4). Furthermore, unpublished data from our lab using mammary epithelial cells expressing activated ErbB2 show that JNK is required for acinus formation in in vitro 3D cultures. In contrast to these studies showing a tumorigenic role for the JNK pathway, other data from our lab show that JNK loss results in accelerated breast tumor growth, suggesting a tumor suppressive role (5, 6). However, these studies were performed in p53 knockout mice with or without a Kras mutation, where the latter required extensive aging and genomic instability to occur before differences in tumor growth were observable. To date, limited in vivo studies exist to confirm the role of JNK in more biologically relevant breast tumor models, such as in ErbB2-mediated cancer, which accounts for over 30% of all human breast cancers. In addition, the molecular mechanisms by which JNK signaling promotes ErbB2-driven tumorigenesis remains poorly understood. To address the discrepancy in JNK function between the in vitro ErbB2-driven breast cancer data and the in vivo p53 knockout tumor data, I began the development of an in vivo murine model to confirm the role of JNK in ErbB2-driven breast cancer. This mouse model will also allow us to test a potential mechanism by which JNK regulates tumorigenesis. Studies show that ErbB2-mediated secretion of the inflammatory cytokine IL6 promotes transformation and tumor growth by activation of the STAT3 transcription factor, triggering an IL6/STAT3 autocrine signaling loop (7,8). A major regulator of Il6 gene expression includes activator protein 1 (AP-1), a transcription factor composed of downstream JNK targets in the Jun protein family (9). In vitro experiments using ErbB2-overexpressing mammary epithelial cell lines show that chemical inhibition of JNK suppresses secreted IL6 protein levels, supporting a role for the JNK pathway in IL6 regulation (7). Thus, I hypothesize that JNK drives ErbB2-driven breast cancer by promoting IL6-mediated tumor progression. Addressing this will increase our understanding of the role of JNK in ErbB2-driven breast cancer and reveal a potentially new mechanism by which JNK functions in tumor progression. Additionally, I began the development of a mouse model that will allow us to investigate the role of JNK in macrophage polarization as an alternative mechanism by which JNK regulates ErbB2-driven breast cancer. In addition to promoting STAT3-dependent tumor growth, IL6 can indirectly drive tumorigenesis by promoting expression of the IL4 receptor in macrophages, triggering STAT6-mediated macrophage polarization towards the pro-tumorigenic M2 phenotype (10, 11). Unlike classically activated M1 macrophages, which promote inflammation and anti-tumor immunity, alternatively activated M2 macrophages function in immunosuppression and metastasis and correlate with advanced stages of breast cancer (12, 13). Further evidence supporting a role for the JNK pathway in macrophage polarization includes a recent study suggesting that JunB, a downstream JNK target and component of the AP-1 complex, plays a crucial role in the induction of M2 macrophage polarization in human alveolar macrophages (13). I hypothesize that activation of the JNK signaling pathway induces IL6-dependent macrophage polarization towards the pro-tumorigenic M2 phenotype. Addressing this hypothesis will determine for the first time whether JNK functions in regulating macrophage polarization within the tumor microenvironment, offering a potentially new mechanism by which JNK can promote ErbB2-driven breast cancer. Determining the role of JNK in ErbB2-mediated breast cancer will have direct therapeutic relevance, as targeting JNK has the potential to inhibit ErbB2-driven breast cancer and other IL6-mediated diseases. Investigating the underlying mechanisms by which JNK functions in ErbB2-positive breast cancer can also offer new molecular targets and further contribute to effective drug design.

Indiana University-Purdue University Indianapolis (IUPUI) / Network-based modeling and analysis have been widely used for capturing molecular trajectories of cellular processes. For complex diseases like cancers, if we can utilize network models to capture adequate features, we can gain a better insight of the mechanism of cancers, which will further facilitate the identification of molecular vulnerabilities and the development targeted therapy. Based on this rationale, we conducted the following four studies: A novel algorithm ‘FFBN’ is developed for reconstructing directional regulatory networks (DEGs) from tissue expression data to identify molecular features. ‘FFBN’ shows unique capability of fast and accurately reconstructing genome-wide DEGs compared to existing methods. FFBN is further used to capture molecular features among liver metastasis, primary liver cancers and primary colon cancers. Comparisons among these features lead to new understandings of how liver metastasis is similar to its primary and distant cancers. ‘SCN’ is a novel algorithm that incorporates multiple types of omics data to reconstruct functional networks for not only revealing molecular vulnerabilities but also predicting drug targets on top of that. The molecular vulnerabilities are discovered via tissue-specific networks and drug targets are predicted via cell-line specific networks. SCN is tested on primary pancreatic cancers and the predictions coincide with current treatment plans. ‘SCN website’ is a web application of ‘SCN’ algorithm. It allows users to easily submit their own data and get predictions online. Meanwhile the predictions are displayed along with network graphs and survival curves. ‘DSCN’ is a novel algorithm derived from ‘SCN’. Instead of predicting single targets like ‘SCN’, ‘DSCN’ applies a novel approach for predicting target combinations using multiple omics data and network models. In conclusion, our studies revealed how genes regulate each other in the form of networks and how these networks can be used for unveiling cancer-related biological processes. Our algorithms and website facilitate capturing molecular features for cancers and predicting novel drug targets.

Cancers that are deficient in BRCA1 or BRCA2 are thought to be hypersensitive to genotoxic agents because they cannot prevent or repair DNA double strand breaks, but observations in patients suggest this dogma may no longer agree with experiment. Here, we propose that single stranded DNA underlies the hypersensitivity of BRCA deficient cancers, and that defects in double strand break repair and prevention do not. Specifically, in BRCA deficient cells, ssDNA gaps developed because replication was not effectively restrained in response to stress. In addition, we observed gaps could be suppressed by either restored fork restraint or by gap filling, both of which conferred therapy resistance in tissue culture and BRCA patient tumors. In contrast, restored double strand break repair and prevention did not confer therapy resistance when gaps were present. Critically, double strand breaks were not detected after therapy when apoptosis was inhibited, supporting a framework in which double strand breaks are not directly induced by genotoxic agents, but instead are created by cell death nucleases and are not fundamental to genotoxic agents. Together, these data indicate that ssDNA replication gaps underlie the BRCA cancer phenotype, "BRCAness," and we propose are fundamental to the mechanism-of-action of genotoxic chemotherapy.

One out of eight women in the United States will develop breast cancer during their lifetime. Ninety percent of cancer related deaths are due to metastasis. Metastasis is the biological process where individual or aggregate cancerous cells break away from the primary tumor site and colonize distant, non-adjacent locations throughout the body. It is my objectives to study how mechanical, topographical and biochemical cues affect metastatic breast cancer metastasis at an early developmental stage. ECM components have previously been shown to affect cell motility via ligand-receptor interactions, and physical cues, such as matrix stiffness and protein density. The primary tumor site significantly stiffens during tumor progression. The ability cells have to sense and respond to these matrix features influences and facilitates cell invasion. It is now widely accepted that mechanical properties of the ECM can regulate cell migration; however, presently, tissue modulus and stiffness have been used interchangeably. It is unknown if cell responses are sensitive to a bulk tissue modulus or stiffness on the geometric length scale of the cell. It is my objective to create tunable biomaterials from known materials to independently parse the roles of stiffness and modulus upon the migration of breast cancer cells. I have created a variety of tunable biomaterials which I can parse the roles of mechanical properties and observe their affect upon cell mechanosensing. All systems were coated with collagen I, which is the most abundant ECM protein during tumor development. I was able to quantify the migration along with other parameters of the metastatic breast cancer cell line MDA-MB-231. My results show that the highly metastatic MDA-MB-231 is stiffness sensitive among all biomaterial models. Cells maximum cell speeds are at high concentrations of collagen I on the polymer microlenses and show a biphasic response dependent on stiffness. On poly (ethylene glycol)- 2-Methacryloyloxyethyl phosphorylcholine (PEG-PC) hydrogels cells favor intermediate modulus and show stiffness dependency at low protein concentrations. Cells on Cd/Se and polydimethylsiloxane (PDMS) samples are influenced by the topographical cue more so than the stiffness or modulus of the material. By controlling mechanosensing via force transduction signaling pathways, and determining the appropriate length-scale by which mechanical properties regulate cancer metastasis, I hope to eventually uncover novel therapeutics to block cell invasion.

Breast cancer is the second leading cause of death of women in the United States, warranting further investigation into preventative therapies. It has been well documented that early pregnancy results in a lifetime decreased risk of breast cancer in humans and mounting evidence suggests that the retinoic acid pathway may play an important role in this protective effect. Cellular retinol binding protein-1 (CRBP1) is an essential component of the retinoic acid pathway and we propose that it plays an important role in pregnancy-induced protection against breast cancer. In order to investigate the role of CRBP1 in parity-induced protection against breast cancer, we utilized both mouse and human mammary epithelial cells. We examined the effect that pregnancy has on CRBP1 expression, how CRBP1 is regulated by growth promoting and inhibiting agents, if loss of CRBP1 is essential for the induction of the apoptotic pathway, and how CpG methylation of key breast cancer genes relates to known risk factors for the disease. Based on our study, CRBP1 is persistently upregulated in response to pregnancy in the mouse mammary gland at both the RNA and protein levels. Using a cell culture model, we established that CRBP1 is regulated by chemical agents that both promote and inhibit cellular growth. Utilizing CRBP1 knockout mice, we demonstrated that CRBP1 is not essential for induction of radiation induced apoptosis in parous mice. Finally, through methylation analysis, we examined how known breast cancer risk factors correlate to CpG methylation of three important genes for breast cancer and noted interesting trends that warrant future study.