Developing Gene-Expression-Based Approaches for DrugDiscovery and Cancer Therapy

• Main Authors: Chen-Tsung Huang, 黃振綜
• Other Authors: 阮雪芬
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/2tr9af

博士 === 國立臺灣大學 === 生醫電子與資訊學研究所 === 107 === My thesis contains three chapters that were dedicated to cancer drug discovery using gene expression. This was made possible by analyzing a large compendium of publicly accessible perturbational gene-expression profiles obtained from the Library of Integrated Network-based Cellular Signatures (LINCS), a project initiated by the US National Institute of Health. The first chapter introduces a new gene expression similarity metric that surpasses other state-of-the-art metrics in drug clustering and repurposing. The second chapter presents an approach that uses small-molecule signatures to target tumor dependencies for combinatorial drug discovery. The third chapter describes an integrated transcriptome analysis in high-risk neuroblastoma, leading to identification of effective drug treatments. Compared with the traditional drug-development paradigm whereby “a single drug should bind to a single target”, the phenomenon of drug promiscuity, or polypharmacology, whereby a single drug can bind to multiple targets, has been recently attracting much attention in the medical community. This polypharmacology is probably even indispensable for effective treatment in some medications. We investigated the complex polypharmacological interactions mirrored in the compound-perturbed gene expression profiles to explore new opportunities for combinatorial drug therapy and repurposing, which will help to substantially reduce the cost and time spent on drug research and development. In the first chapter, we developed a gene expression similarity metric that directly emphasizes the genes exhibiting the greatest changes in expression in response to a perturbation. This metric was proved to outperform other state-of-the-art and commonly used metrics in a clustering task of given known drugs with diverse mechanisms of action. We then applied this metric to systematically compare thousands of small-molecule perturbations across 10 cell types and further investigated an anthelmintic and a loop diuretic as potential topoisomerase inhibitors for anticancer therapy. Along with the dependency on driver mutations that confer growth advantage, cancer cells can also develop an addiction to certain genes that are themselves not oncogenic but whose functions are required for maintenance of the tumorigenic state. These needs of both oncogenes and non-mutated genes for cancer cell survival are coined as oncogene and non-oncogene addictions, respectively. In the second chapter, we systematically analyzed a large compendium of compound-perturbed data to uncover several perturbational gene-expression signatures that are highly correlated with cancer hallmark and drug sensitivity. We then developed a computational approach that uses these small-molecule signatures to target non-oncogene tumor dependencies for combinatorial drug discovery and experimentally confirmed two unexpected drug pairs with synergistic killing. This chapter provides an alternative drug discovery strategy from non-oncogene addiction and has potential clinical applicability to guide future combination therapy in precision medicine. Neuroblastoma is a rare pediatric malignancy, whose heterogeneous mutational spectrum has restricted the development of targeted therapies. Despite intensive treatment, survival for high-risk neuroblastoma still remains below 40%. To address this unmet need, we performed, in the third chapter, an integrative transcriptomic analysis of nearly a thou-sand patients with primary neuroblastoma obtained from multiple Gene Expression Omnibus (GEO) datasets to identify potential drugs that target non-oncogene dependencies in high-risk neuroblastoma. Among these predictions, we demonstrated the in vivo efficacy of niclosamide, an anthelmintic drug approved by the US FDA to treat tapeworm infections, and further investigated its mechanism of action through proteomics. Collectively, these chapters present an alternative insight from gene-expression perspective to identify novel therapeutic options for complex diseases, particularly useful for tumors with few druggable molecular targets or acquired resistance to standard therapies. The methods and results from this collection represent important and significant advances in drug research and cancer therapy, achieved with gene expression analysis.