Melanoma, the most lethal form of skin cancer, arises from altered cells in the melanocyte lineage, but the mechanisms by which these cells progress to melanoma are unknown. To understand the early cellular events that contribute to melanoma formation, we examined melanocytes in melanoma-prone zebrafish strains expressing BRAFV600E, the most common oncogenic form of the BRAF kinase that is mutated in nearly 50% of human melanomas. We found that, unlike wild-type melanocytes, melanocytes in transgenic BRAFV600Eanimals were binucleate and tetraploid. Furthermore, melanocytes in p53-deficient transgenic BRAFV600Eanimals exhibited 8N and greater DNA content, suggesting bypass of a p53-dependent arrest that stops cell cycle progression of tetraploid melanocytes. These data implicate tetraploids generated by increased BRAF pathway activity as contributors to melanoma initiation. Previous studies have used artificial means of generating tetraploids, raising the question of how these cells arise during actual tumor development. To gain insight into the mechanism by which BRAFV600E generates binucleate, tetraploid cells, we established an in vitro model by which such cells are generated following BRAFV600E expression. We demonstrate thatBRAFV600E-generated tetraploids arise via cytokinesis failure during mitosis due to reduced activity of the small GTPase RhoA. We also establish that oncogene-induced centrosome amplification in the G1/S phase of the cell cycle and subsequent increase in the activity of the small GTPase Rac1, partially contribute to this phenotype. These data are of significance as recent studies have shown that aneuploid progeny of tetraploid cells can be intermediates in tumor development, and deep sequencing data suggest that at least one third of melanomas and other solid tumors have undergone a whole genome doubling event during their progression. Taken together, our melanoma-prone zebrafish model and in vitro data suggest a role for BRAFV600E-inducedtetraploidy in the genesis of melanomas. To our knowledge, this is the first in vivo model showing spontaneous rise of tetraploid cells that can give rise to tumors. This novel role of the BRAF oncogene may contribute to tumorigenesis in a broader context.

Hippo signaling is a tumor suppressive signaling pathway that controls organ size by regulating cellular proliferation, apoptosis, and differentiation during development, regeneration, and homeostasis. The Hippo pathway inhibits transcriptional co-activators and Hippo pathway effectors YAP/TAZ, activation of which is often seen in cancer. Within the adult mammalian intestine, homeostasis of which requires intricate reciprocal interaction between the gut epithelium and adjacent mesenchyme, the Hippo-YAP pathway is crucial for intestinal epithelial homeostasis and regeneration. However, its role in adult mesenchymal homeostasis remains poorly understood. Here, I genetically dissect the role of mesenchymal Hippo-YAP signaling in adult intestinal homeostasis. I find that deletion of core kinases LATS1/2 or YAP activation in mesenchymal progenitor cells, but not terminally differentiated cells, disrupts signaling in the stem cell niche and mesenchymal homeostasis by inducing mesenchymal overgrowth and suppressing smooth muscle actin expression. Furthermore, inhibition of Hippo signaling in Gli1+ mesenchymal progenitors, the main source of Wnt ligands within the stem cell niche, stimulates Wnt ligand production and subsequent epithelial Wnt pathway activation, thereby driving epithelial regeneration following DSS-mediated injury as well as exacerbating APC-mediated tumorigenesis. Altogether, our data reveal a previously underappreciated requirement and the underlying mechanism for stromal Hippo-YAP signaling in adult intestinal homeostasis.

Glioblastoma multiforme (GBM) tumors are highly malignant in nature and despite an aggressive therapy regimen, long–term survival for glioma patients is uncommon as cells with intrinsic or acquired resistance to treatment repopulate the tumor. This creates the need to investigate new therapies for enhancing GBM treatment outside of the standard of care, which includes Temozolomide (TMZ). Our lab focused on two novel strategies to overcome resistance in GBMs. In our first approach, the cellular responses of GBM cell lines to two new TMZ analogues, DP68 and DP86, are reported. The efficacy of these compounds was independent of DNA repair mediated by Methyl Guanine Methyl Transferase (MGMT) and the mismatch repair (MMR) pathway. DP68 or DP86 treated cells do not give rise to secondary spheres, demonstrating that they are no longer capable of self-renewal. DP68-induced damage includes interstrand DNA crosslinks and exhibits a distinct S-phase accumulation before G2/M arrest; a profile that is not observed for TMZ-treated cells. DP68 induces a strong DNA damage response and suppression of FANCD2 expression or ATR expression/kinase activity enhanced the anti-GBM effects of DP68. Collectively, these data demonstrate that DP68, and to a lesser extent DP86, are potent anti-GBM compounds that circumvent TMZ resistance and inhibit recovery of cultures. Our second approach stems from a previous discovery in our lab which demonstrated that the combination of TMZ with Notch inhibition, using a gamma secretase inhibitor (GSI), enhances GBM therapy. Efficacy of TMZ + GSI treatment is partially due to GBM cells shifting into a permanent senescent state. We sought to identify a miR signature that mimics the effects of TMZ + GSI as an alternative vi approach to enhance GBM therapy. MiR-34a expression was highly upregulated in response to TMZ or TMZ + GSI treatment. Exogenous expression of miR-34a revealed that it functions as a tumor suppressor and mimicked the in vitro effects of TMZ + GSI treatment. Additionaly, miR-34a overexpression leads to the downregulation of Notch family members. Together these two studies contribute to our understanding of the complex mechanisms driving resistance in GBM tumors and suggest strategies to develop more effective therapies.

The p53 tumor suppressor protein is upregulated in response to DNA damage and other stress signals. The upregulation of p53 involves freeing it from negative regulation imposed by Mdm2 and MdmX (Mdm4). Accumulating evidence indicates that phosphorylation of Mdm proteins by different stress-activated kinases such as ATM or c-Abl significantly impacts p53 functions. We have previously shown that ATM phosphorylation of Mdm2 Ser394 is required for robust p53 stabilization and activation following DNA damage. This dissertation describes in vivo examination of the mechanism by which Mdm2 Ser394 phosphorylation impacts p53 activities and its contribution to suppression of oncogene and DNA damage-induced tumors. We determine that phosphorylation of Mdm2 Ser394 regulates p53 activity by modulating Mdm2 stability and paradoxically delays Myc-driven lymphomagenesis while increasing lymphomagenesis in sub-lethally irradiated mice. c-Abl phosphorylates the residue neighboring Mdm2 Ser394, Mdm2 Tyr393. This dissertation describes the generation of a novel Mdm2Y393F mutant mouse to determine if c-Abl phosphorylation of Mdm2 regulates p53-mediated DNA damage responses or tumor suppression in vivo. Mdm2Y393F mice develop accelerated spontaneous and oncogene-induced tumors, yet display no defects in p53 stabilization and activity following acute genotoxic stress. Furthermore, the effects of these phosphorylation events on p53 regulation are not additive, as Mdm2Y393F/S394A mice and Mdm2S394A mice display similar phenotypes. The studies presented herein further our understanding of the mechanisms by which DNA damage-associated kinases stabilize and activate p53, and influence p53-dependent responses and tumor suppression. A better understanding of the in vivo effects of Mdm2 phosphorylation may facilitate the development of novel therapeutics capable of stimulating p53 anti-tumor activity or alleviating p53- dependent toxicities in non-malignant tissues.

The nuclear matrix (NM) is a fibrogranular network of ribonucleoproteins upon which transcriptional complexes and regulatory genomic sequences are organized. A hallmark of cancer is the disorganization of nuclear architecture; however, the extent to which the NM is involved in malignancy is not well studied. The RUNX1 and RUNX2 proteins form complexes within the NM to promote hematopoiesis and osteoblastogenesis, respectively at the transcriptional level. RUNX1 and RUNX2 are both expressed in breast cancer cells (BrCCs); however, their genome-wide BrCC functions are unknown. RUNX1 and RUNX2 activate many tumor suppressor pathways in blood and bone lineages, respectively, including attenuation of protein synthesis and cell growth via suppression of ribosomal RNA (rRNA) transcription, which appears contrary to Runx-expression in highly proliferative BrCCs. To define roles for RUNX1 and RUNX2 in BrCC phenotype, we examined the involvement of RUNX1 and RUNX2 in rRNA transcription and generated a genome-wide model for RUNX1 and RUNX2-binding and transcriptional regulation. To validate gene expression patterns identified in our screen, we developed a Real-Time qPCR primer design program, which allows rapid, high-throughput design of primer pairs (FoxPrimer). In BrCCs, RUNX1 and RUNX2 regulate genes that promote invasiveness and do not affect rRNA transcription, protein synthesis, or cell growth. We have characterized in vitro functions of Runx proteins in BrCCs; however, the relationships between Runx expression and diagnostic/prognostic markers of breast cancer (BrCa) in patients are not well studied. Immunohistochemical detection of RUNX1 and RUNX2 in BrCa tissue microarrays reveals RUNX1 expression is associated with early, smaller tumors that are ER+ (estrogen receptor), HER2+, p53-, and correlated with androgen receptor (AR) expression; RUNX2 expression is associated with late-stage, larger tumors that are HER2+. These results show that the functions and expression patterns of NM-associated RUNX1 and RUNX2 are context-sensitive, which suggests potential disease-specific roles. Two functionally disparate genomic sequence types bind to the NM: matrix associated regions (MARs) are functionally associated with transcriptional repression and scaffold associated regions (SARs) are functionally associated with actively expressed genes. It is unknown whether malignant nuclear disorganization affects the functions of MARs/SARs in BrCC. We have refined a method to isolate nuclear matrix associated DNA (NM-DNA) from a structurally preserved NM and applied this protocol to normal mammary epithelial cells and BrCCs. To define transcriptional functions for NM-DNA, we developed a computational algorithm (PeaksToGenes), which statistically tests the associations of experimentally-defined NM-DNA regions and ChIP-seq-defined positional enrichment of several histone marks with transcriptome-wide gene expression data. In normal mammary epithelial cells, NM-DNA is enriched in both MARs and SARs, and the positional enrichment patterns of MARs and SARs are strongly associated with gene expression patterns, suggesting functional roles. In contrast, the BrCCs are significantly enriched in the silencing mark H3K27me3, and the NM-DNA is enriched in MARs and depleted of SARs. The MARs/SARs in the BrCCs are only weakly associated with gene expression patterns, suggesting that loss of normal DNA-matrix associations accompanies the disease state. Our results show that structural preservation of the in situ NM allows isolation of both MARs and SARs, and further demonstrate that in a disorganized, cancerous nucleus, normal transcriptional functions of NM-DNA are disrupted. Our studies on nuclear organization in BrCC, show that the disorganized phenotype of the cancer cell nucleus is accompanied by deregulated transcriptional functions of two constituents of the NM. These results reinforce the role of the NM as an important structure-function component of gene expression regulation.

Although cure rates have significantly improved for children with T-cell acute lymphoblastic leukemia (T-ALL), 20-30% undergo induction failure or relapse with most succumbing to disease. Leukemia-initiating cells (L-ICs) are hypothesized to be resistant to conventional chemotherapy and radiation and are thereby responsible for disease recurrence. Using an in vivo limiting dilution assay, we previously showed that the murine T-ALL L-IC is quite rare, with only 0.003-0.05% of cells capable of initiating disease, and demonstrated that the L-IC is a subset of the leukemic DN3 thymic progenitor population. Work described in this thesis validates the L-IC assay using two transplantation methods to rule out effects of homing and/or microenvironment on T-ALL L-IC survival and maintenance. Using this assay, we demonstrate that sustained Notch1 signaling is required for T-ALL initiation in vivo and show that treatment with a Notch1 inhibitor reduces or in some cases eliminates the L-IC population. We further analyze the effects of inhibiting c-Myc, a Notch1-regulated gene, on L-IC frequency and uncover an essential role for c-Myc in L-IC survival and expansion. Suppressing c-Myc by using specific shRNAs or a c-Myc inhibitor reduces the L-IC population and interferes with leukemia initiation. Together, these findings reveal a critical role of the Notch1-c-Myc pathway in T-ALL initiation and suggest that therapeutics targeted at this pathway could be used to treat and/or prevent disease relapse in patients.

Gain of function NOTCH1 mutations are common in both patients with T-ALL and in mouse models of the disease. Inhibiting the Notch pathway in T-ALL cell lines results in growth arrest and/or apoptosis in vitro, suggesting a requirement for Notch signaling in T-ALL. Therefore, we sought to examine the role of Notch1 signaling in both premalignancy and in the maintenance of leukemic growth. Using a murine model of T-ALL, in which expression of the Tal1 and Lmo2 oncogenes arrests thymocyte development, our preleukemic studies reveal that Notch1 mutations are early events that contribute to the clonal expansion of DN3 and DN4 progenitors. We also demonstrate that progenitors are maintained within the tumor and are enriched in leukemia-initiating cell (L-IC) activity, suggesting Notch1 may contribute to L-IC self-renewal. By studying the effects of Notch signaling in murine T-ALL cell lines, we also demonstrate that Notch1 promotes the proliferation and survival of leukemic blasts through regulation of Lef1 and the Akt/mTOR pathways. Given that T-ALL cell lines are dependent on Notch signaling in vitro, we investigated the effects of Notch inhibition in vivo. We provide evidence that Notch1 can be successfully targeted in vivo and that Notch inhibition, with γ-secretase inhibitors (GSIs), significantly extends the survival of leukemic mice. We also demonstrate that administration of GSIs in combination with rapamycin inhibits human T-ALL growth and extends survival in a mouse xenograft model. Given that NOTCH1 may be required to maintain both L-IC and bulk leukemic growth, targeting NOTCH1 may prove to be an efficacious targeted therapy for T-ALL patients with aberrant NOTCH1 activation.

In the mammalian genome, 5‟-CpG-3‟ dinucleotides are frequently methylated, correlating with transcriptional silencing. Genome-wide waves of demethylation are thought to occur only twice during development, in primordial germ cells and in the pre-implantation embryo. They are followed by de novo methylation, setting up a pattern that is inherited throughout development. No global methylation changes are thought to occur during further somatic development, although methylation does alter at gene-specific loci, contributing to tissue-specific patterns of gene expression. Here we studied DNA methylation in differentiating mouse erythroblasts in vivo using several approaches including genomic-scale, reduced representation bisulfite sequencing (RRBS). Surprisingly, demethylation at the erythroid-specific β-globin locus was coincident with a wave of global DNA demethylation at most genomic elements, including repetitive elements and genes silenced in erythropoiesis. Over 30% of total methylation is irreversibly lost during erythroid differentiation. Demethylation occurred through a passive mechanism, requiring the rapid DNA replication triggered with the onset of erythroid terminal differentiation. Global loss of DNA methylation was not associated with a global increase in transcription, as determined by GeneChip analysis. We propose that global demethylation is a consequence of cellular mechanisms required for the rapid demethylation and induction of β-globin and other erythroid genes. Our findings demonstrate that, contrary to previously held dogma, DNA demethylation can occur globally during somatic cell differentiation, providing a new experimental model for the study of global demethylation in development and disease.

For 30 years, the tumor suppressor p53 has been a subject of intense research in nearly every discipline of scientific inquiry. While numerous surprising roles for p53 in health and disease are uncovered each year, the central role of its activation in preventing neoplastic transformation has been and will remain at the forefront of p53 research as investigators work to address an unexpectedly complex question—precisely how does p53 integrate upstream stress signals to coordinate activation of its target genes in response to stress? One manner in which to address this question is at the level of transcription initiation—after upstream signals converge on p53 and produce a number of pools of post-transcriptionally modified p53, how exactly are specific target promoters activated in such a sensitive, context-specific manner? The work presented herein aims to address the role of histone acetylation at the p21 promoter—a critical mediator of G1/S arrest—by the P/CAF acetyltransferase in response to a variety of p53-activating stresses. We show that depletion of P/CAF strongly inhibits p21 expression in response to a variety of stresses, despite normal stabilization of p53 and recruitment to target promoters. This defect in p21 expression correlates closely with abrogation of stress-induced cell-cycle arrest. Strikingly, a p53 allele lacking putative P/CAF acetylation sites was still able to direct p21 expression, which was still dependent upon P/CAF. We show further that histone acetylation at H3K14 at the p21 promoter following stress is dependent upon P/CAF. Rescue of p21 expression with wild-type P/CAF or a ∆HAT point mutant indicates that P/CAF requires an intact HAT domain, suggesting that histone acetylation at H3K14 is catalyzed by P/CAF HAT activity, not the molecular bridging of a heterologous HAT by P/CAF. Furthermore, RNA polymerase II (RNAP II) was present at the p21 proximal promoter under all basal and stress conditions, but elongation of RNAP II after stress required the presence of P/CAF. These data indicate that H3K14 acetylation by P/CAF closely correlates with the activation status of the p21 promoter, and may be necessary for activation of a larger subset of p53-responsive promoters. In addition to its critical role in p21 expression, we noted that p53 stabilization and cell-cycle arrest in response to p14ARF, but not other p53-stabilizing stresses, were also dependent on P/CAF. Cell-cycle arrest induced by p16INK4A was intact after P/CAF ablation, indicating a role for P/CAF in cell-cycle arrest specific to p14ARF-p53 signaling. Basal MDM2 levels were unaffected by P/CAF knockdown, as were p53- MDM2 and ARF-MDM2 complexes. A preliminary analysis of MDM2 localization was inconclusive, due to vastly different quantities of MDM2 in different conditions making analysis of subcellular localization difficult; however, the role of P/CAF in the relocalization of MDM2 to the nucleolus by p14ARF could potentially explain the defect in p53 stabilization, and should be explored further. These observations, underscored by recent reports that P/CAF undergoes loss of heterozygosity in several tumor types, suggest that P/CAF plays a critical role in p53-mediated cell-cycle arrest through multiple, independent mechanisms. Further study should clarify whether P/CAF is lost in tumors maintaining wild-type p53, and whether its reintroduction into these tumors confers any potential therapeutic benefit.

The development of RNAi based technologies has given researchers the tools to interrogate processes as diverse as cancer biology, metabolism and organ development. Here I employ genome-wide shRNA screens to discover the genes involved in two different processes in carcinogenesis, oncogene-induced senescence [OIS] and epigenetic silencing of tumor suppressor genes [TSGs]. OIS is a poorly studied yet significant tumor suppressing mechanism in normal cells where they enter cell cycle arrest [senescence] or programmed cell death [apoptosis] in the presence of an activated oncogene. Here I employ a genomewide shRNA screen and identify a secreted protein, IGFBP7, that induces senescence and apoptosis in melanocytes upon introduction of the oncogene BRAFV600E. Expression of BRAFV600E in primary cells leads to synthesis and secretion of IGFBP7, which acts through autocrine/paracrine pathways to inhibit BRAF-MEK-ERK signaling and induce senescence and apoptosis. Apoptosis results from IGFBP7-mediated upregulation of BNIP3L, a proapoptotic BCL2 family protein. Recombinant IGFBP7 has potent pro-apoptotic and anti-tumor activity in mouse xenograft models using BRAFV600E-postive melanoma cell lines. Finally, IGFBP7 is epigenetically silenced in human melanoma samples suggesting IGFBP7 expression is a key barrier to melanoma formation. Next I investigated the factors involved in epigenetic silencing in cancer. The TSG p14ARFis inactivated in a wide range of cancers by promoter hypermethylation through unknown mechanisms. To discover p14ARF epigenetic silencing factors, I performed a genome-wide shRNA screen and identified ZNF304, a zinc finger transcription factor that contains a Krüppel-associated box [KRAB] repressor domain. I show that ZNF304 binds to the p14ARF promoter and recruits a KRAB co-repressor complex containing KAP1, SETDB1 and DNMT1 for silencing. We find oncogenic RAS signaling to promote the silencing of p14ARF by USP28-mediated stabilization of ZNF304. In addition I find ZNF304 to be overexpressed in human colorectal cancers and responsible for hypermethylation of over 50 TSGs known as Group 2 CIMP marker genes. My findings establish ZNF304 as a novel oncogene that directs epigenetic silencing and facilitates tumorigenicity in colorectal cancer.