Files in this item

FilesDescriptionFormat

application/pdf

application/pdfMOODY-DISSERTATION-2019.pdf (8MB)Restricted Access
(no description provided)PDF

Description

Title:Genome-wide cross-cancer analysis reveals the importance of bimodal microRNA in predicting patient survival and drug response
Author(s):Moody, Laura
Director of Research:Pan, Yuan-Xiang
Doctoral Committee Chair(s):Chen, Hong
Doctoral Committee Member(s):Helferich, William; Arthur, Anna
Department / Program:Nutritional Sciences
Discipline:Nutritional Sciences
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):microRNA
cancer
bimodal
gene expression
drug response
survival
personalized medicine
Abstract:While genetic dysregulation is characteristic of cancer, specific mutational and transcriptomic signatures are quite variable between tumors. Many tumorigenic genes are only aberrantly expressed in a few patients and thus follow a bimodal distribution, having two modes of expression within a single population. In this study, we conducted an analysis of bimodal microRNA (miRNA) in order to find consistent sources of heterogeneity across cancers and demonstrate their clinical importance. First, we improved on previous methods and developed a new mixture modeling approach for finding bimodal genes. We used RNA-seq data from breast tumors to demonstrate the superior ability of our model to detect tumorigenic genes with non-trivial expression and evenly partitioned components. Given the extensive regulatory role of miRNA, we then examined bimodal miRNA and demonstrated their importance as biomarkers of cancer development. We applied our method to produce bimodal miRNA profiles in nine types of tumors. Tumors consistently displayed greater bimodality than normal tissue, and several miRNA were bimodally expressed across multiple cancer types. To validate the clinical relevance of the identified miRNA, we showed that concurrently-expressed modules of bimodal miRNA could be used to stratify patients based on overall survival. In liver and lung cancers, we found high expression of miR-105 and miR-767 to be associated with poor survival prognosis. Functional pathway analysis revealed that targets of miR-105 and miR-767 were enriched for genes in the phosphoinositide-3-kinase (PI3K) pathway, and a survey of over 200 cancer drugs found that cell lines with high miR-105 and miR-767 levels responded best to drugs targeting the same pathway. To mimic high miR-105 and miR-767-expressing tumors, the two miRNA were overexpressed in A549 lung cancer cells. High miRNA conditions were shown to reduce cell viability and increase apoptosis in response to treatment with the class I PI3K inhibitor, ZSTK474. We then investigated the molecular mechanisms by which miR-105 and miR-767 improved drug sensitivity. First, we measured the expression of integral proteins in the PI3K pathway but found no miRNA-mediated changes in PI3K subunits or AKT isoforms. Instead, we quantified proteins that mediate negative feedback and crosstalk with closely related pathways. We found that miR-105 and miR-767 reduced levels of forkhead box O3 (FOXO3) and insulin receptor substrate 1 (IRS1). Because of the activating role of FOXO3 and IRS1 in mitogen-activated protein kinase (MAPK) signaling, levels of phosphorylated extracellular signal-regulated kinase (p-ERK) were also measured. Drug induction of p-ERK was partially mitigated by miR-105 and miR-767 overexpression. Finally, RNA immunoprecipitation revealed FOXO3 and IRS1 as direct targets of miR-105 and miR-767. Thus, we propose that high miR-105 and miR-767 expression may downregulate crosstalk between the PI3K and MAPK pathways in order to reduce off-target drug effects and facilitate drug sensitivity. Overall, we demonstrate that high levels of bimodal miRNA expression are characteristic of cancer. While cancer is marked by considerable genetic heterogeneity, there is between-cancer concordance regarding the particular miRNA that are more susceptible to variability. The current study provides a framework for identifying novel bimodal miRNA biomarkers that can be used to stratify patients based on survival time and drug response.
Issue Date:2019-11-25
Type:Text
URI:http://hdl.handle.net/2142/106453
Rights Information:Copyright 2019 Laura Moody
Date Available in IDEALS:2020-03-02
Date Deposited:2019-12


This item appears in the following Collection(s)

Item Statistics