We discuss a strategy for visual recognition by forming groups of salient image features, and then using these groups to index into a data base to find all of the matching groups of model features. We discuss the most space efficient possible method of representing 3-D models for indexing from 2-D data, and show how to account for sensing error when indexing. We also present a convex grouping method that is robust and efficient, both theoretically and in practice. Finally, we combine these modules into a complete recognition system, and test its performance on many real images.

Ovarian cancer is one of the most common causes of gynecological cancer related deaths in women. In 2014, the estimated number of deaths due to ovarian cancer is 14,270 with occurrence of over 22, 240 new cases (National Cancer Institute, http://seer.cancer.gov/statfacts/html/ovary.html). Despite improvement in treatment strategies, the 5-year survival rate is still below 50% mainly due to chemoresistance and relapse. Amplification of chromosomal region 3q26 is a common characteristic in various epithelial cancers including ovarian cancer. This region harbors various oncogenes including the TGFβ signaling mediators EVI1 and SnoN/SkiL, PKCι and PIK3CA amplified at 3q26.2 and 3q26.3, respectively, in ovarian cancers. Previous studies indicate that these genes can exhibit cooperative oncogenicity by cross-regulating one another and facilitating cancer development. Our earlier studies demonstrated that treatment of ovarian cancer cells with arsenic trioxide (As2O3) promotes cytoprotective autophagy regulated by induction of SnoN to antagonize the cytotoxic effects of As2O3. Since exact mechanisms underlying As2O3-induced SnoN expression and cytoprotective responses were unclear, we hypothesized that SnoN may be regulated by signaling pathways involving genes amplified at the 3q26 locus. Phospholipid scramblase 1 (PLSCR1) is located at 3q23 proximal to the amplified 3q26 region. It had been implicated in disruption of plasma membrane asymmetry by mediating phospholipid scrambling, a process critical for cellular events such as blood coagulation and apoptosis. However, recent findings have led to more investigations on the role and regulation of PLSCR1 in cancer development and immune responses. PLSCR1 expression is regulated by various stimuli including growth factors (EGF, G-CSF, and SCF), cytokines (IFN), and differentiation-inducing agents (ATRA). Despite these studies, transcriptional regulation of PLSCR1 remains incompletely understood. Numerous studies have suggested a critical role for PLSCR1 in the pathophysiology of various cancers including leukemia, ovarian cancer, colorectal cancer, and metastatic liver cancer. However, the precise contribution of PLSCR1 and its regulation in ovarian cancer development is unclear. Since PLSCR1 (at 3q23) is located in close proximity to SnoN/SkiL (at 3q26.2), we hypothesized that PLSCR1 expression in ovarian cancer cells could be regulated by SnoN. Herein, we present studies that primarily focus on understanding the role and regulation of SnoN/SkiL (a TGFβ pathway regulator) and PLSCR1 (an interferon-regulated gene), which are located at 3q26.2 and 3q23, respectively, in epithelial ovarian cancer. In Chapter 3, we determined that activation of the PI3K signaling pathway mediates SnoN expression and cytoprotective responses upon stimulation of ovarian cancer cells with As2O3. We first identified that As2O3 stimulation leads to activation of EGFR and its downstream signaling mediators as well as modulates its interaction with the adaptor proteins, ShcA and Grb2. Interestingly, while treatment with a general SFK inhibitor (PP2), reduced the As2O3-induced EGFR activation and SnoN induction, a more specific inhibitor SU6656 did not alter SnoN expression. Further, via studies utilizing specific inhibitors and siRNA targeting PI3K, we determined that inhibition of PI3K signaling pathway decreases SnoN induction and increases apoptosis in ovarian cancer cells in response to As2O3. This suggests that PI3K (PIK3CA) activity is required for the As2O3-mediated SnoN induction and the cell survival responses in ovarian cancer cells. Finally, we determined by siRNA-mediated knockdown that EGFR and MAPK1 alter As2O3-induced cell death response independently of SnoN induction. In Chapter 4, via bioinformatic analyses, we identified that PLSCR1 DNA copy number and mRNA expression is elevated in ovarian cancer patients and cell lines relative to immortalized (Tag/hTERT) normal ovarian surface epithelial (OSE) cells. Interestingly, altered PLSCR1 DNA and mRNA levels were correlated with SnoN in ovarian cancers. We next identified that SnoN knockdown leads to a significant (~35%, P2O3 transcriptionally downregulates PLSCR1 in a ROS-independent mechanism. Furthermore, PLSCR1 knockdown, similar to SnoN knockdown increases ovarian cancer cell sensitivity to As2O3. PLSCR1 knockdown increases cleaved PARP (marker of apoptosis) with a consequent reduction in LC3-II levels (marker of autophagosomes). Collectively, these studies implicate PLSCR1 in the pathophysiology of ovarian cancers and in altering the chemotherapeutic responses in ovarian cancer cells. PLSCR1 is an IFN-regulated gene and mediates antiviral/immune responses. More recent studies in plasmacytoid dendritic cells have implicated PLSCR1 in regulating TLR9 signaling upon stimulation with CpG ODN. However, whether PLSCR1 could mediate the innate immune responses upon stimulation with dsDNA remained unclear. In Chapter 5, we identified that stimulation of normal ovarian and mammary epithelial cells with dsDNA (empty plasmid) markedly induces PLSCR1 consequent with activation of IRF3, a downstream mediator of TLR signaling that transcriptionally regulates the expression of type 1 IFNs. Interestingly, IRF3 knockdown ablates the dsDNA-induced PLSCR1 expression suggesting that PLSCR1 induction in response to dsDNA could be mediated by IRF3. Additionally, we have determined that dsDNA stimulation induces nucleic acid sensing TLRs, TLR9 and TLR4 as well as IFN-α and IFN-β mRNAs. Interestingly, dsDNA stimulation did not induce PLSCR1 or IRF3 activation in ovarian cancer cells suggesting that the mechanisms of IRF3 activation and PLSCR1 induction in response to dsDNA might be dysregulated in ovarian cancers. Collectively, our studies demonstrate a possible synergistic role of SnoN and PLSCR1 in ovarian cancer pathophysiology and suggest a potentially dysregulated role of PLSCR1 in the dsDNA-induced immune responses of malignant epithelial cells relative to normal epithelial cells. These studies could potentially lead to development of a novel combinatorial therapeutic strategy that targets both these molecules for improving treatment of patients with ovarian carcinoma.

Genomic aberrations are primary contributors to the pathophysiology of cancer [11]. Dysregulated expression of genes located within these aberrations are important predictors of chemoresistance, disease prognosis, and patient outcome [12]. This dissertation is focused on understanding the regulation and/or functions of specific genes located at dysregulated genomic regions such as 3q26 and 14q32 in the biology of ovarian and renal cancer, respectively. Serous epithelial ovarian cancer (EOC) manifest amplification at the 3q26.2 locus [2], an observation consistent with the cancer genome atlas (TCGA) [13]. The most amplified gene in this region is EVI1 which has been extensively studied in hematological malignancies [2]. However, its contribution to the pathophysiology of solid cancers remains unknown. We hypothesized that dysregulated EVI1 and SnoN/SkiL expression (located at the 3q26.2 amplicon) leads to the altered cellular functional response, thereby contributing to the pathophysiology of ovarian cancer. Our group has previously shown that EVI1 splice forms may exhibit altered subcellular localization and functional properties relative to the wild type form [14]. In Chapter 3 of this dissertation, we identified that EVI1 splice forms could modulate epithelial-mesenchymal transition. Our findings indicate that siRNA construct targeting the splice junction between exon 2 of MDS1 to exon 2 of EVI1, (reduces the expression of MDS1/EVI1 and EVI1Del190-515 splice forms) increases epithelial cell markers while decreasing mesenchymal markers and reducing migratory potential of ovarian and breast cancer cells. SnoN/SkiL, another gene overexpressed at the 3q26 is reported by our group to be induced upon As2O3 treatment in ovarian cancer cells via unknown mechanisms [15]. This induction of SnoN opposes the apoptotic cell death pathway induced by the drug treatment [15]. We have previously identified that the PI3K/AKT pathway (also dysregulated in ovarian cancer [16]) contributes to the up-regulation of SnoN upon treatment with As2O3 [17]. However, SnoN is regulated via multiple mechanisms including post-translational modifications [18]. Additionally, c-Ski (a homolog of SnoN) is regulated post-transcriptionally by numerous miRNAs in cancer cells [19-22]. In Chapter 4, we attempted to identify potential miRNAs that could regulate SnoN expression post-transcriptionally. We discovered that miR-494 reduces both SnoN mRNA and protein levels. Our experimental outcomes also demonstrate that miR-494 further sensitizes ovarian cancer cells to drug treatment. Interestingly, miR-494 is located at the 14q32 region which has been shown to be down-regulated in renal cancers [23]. Several reports indicate miR-494 to be involved in tumor suppressive responses including apoptosis and cell cycle arrest in various cancers [24-26]. However, its role in renal cancer biology remains unknown. We hypothesized that miR-494 elicits a tumor suppressive response in renal cancer cells. Through our studies in Chapter 5, we demonstrate that miR-494 reduces cell viability and increases apoptotic response in renal cancer cells. We also show that miR-494 increases LC3B mRNA and protein levels. A 3’UTR luciferase assay indicated that LC3B may be a potential target of miR-494. Intracellular lipid droplets (LDs) increased in miR-494 expressing in a LC3B-dependent manner. This was accompanied with reduced intracellular cholesterol content, increased mitochondrial structural disorganization, and altered Drp1 localization. The outcome of our findings have improved our understanding of the regulation and functional response of these genes/miRNAs (EVI1, SnoN, and miR-494) in ovarian and renal cancers. The studies reported in Chapter 5 identified a novel function of miR-494 in increasing LDs and reducing renal cell survival. However, additional studies are warranted to fully understand the underlying mechanism of increased LDs formation in miR-494 expressing cells and the implication of miR-494 and other miRNAs at the 14q32 region in renal cancer biology. In future, these studies will aid in the development of better treatment strategies which will contribute towards the management of cancer.

This paper reports work in progress on an experimental text editor called CREF, the Cross Referenced Editing Facility. CREF deals with chunks of text, called segments, which may have associated features such as keywords or various kinds of links to other segments. Text in CREF is organized into linear collections for normal browsing. The use of summary and cross-reference links in CREF allows the imposition of an auxiliary network structure upon the text which can be useful for "zooming in and out" or "non-local transitions." Although it was designed as a tool for use in complex protocol analysis by a "knowledge Engineer's Assistant," CREF has many interesting features which should make it suitable for a wide variety of applications, including browsing, program editing, document preparation, and mail reading.