단백질들의 세포 내 위치 예측과 이를 이용한 병인 연구 : Prediction of protein subcellular localizations and its application to the pathogenesis

다국어 초록 (Multilingual Abstract)

Characterizing the subcellular localization of proteins provides a key clue for understanding protein function. Each organelle has unique chemical and physical environment which cluster functionally related proteins. Many bioinformatics approaches have developed to identify proteins subcellular localization using various features from primary sequences. However, the various predictors have different performances on predicting subcellular localization. Therefore, using only a single predictor is not sufficient to get comprehensive subcellular localization information of proteins.
Disease-causing mutations alter the synthesis of the gene product or change the protein targeting to proper subcellular localizations. Abnormal protein localizations are known to lead the loss of functional effects in diseases. Therefore, proper identification of protein subcellular localization is useful to identify disease-associated proteins. However, the direct connection between diseases and subcellular localization has not been studied systematically. In this study, I elucidated the impact of subcellular localization to understand the disease progression based on the novel consensus localization prediction method.
First, I developed a consensus localization prediction method called ConLoc based on a large-scale, systematic integration of 13 available programs. It focused on five major subcellular localizations, which are the cytosol, extracellular, mitochondria, nucleus, and plasma membrane. ConLoc showed higher prediction performance compared to other programs. It achieved the highest prediction accuracy of 0.96 and Matthew’s correlation coefficient (MCC) of 0.86 on the localization prediction of human proteins. I validated this method on multi-localized proteins and observed the equivalent accuracy (0.94) on the prediction of multi-localized proteins compared to that (0.96) of single-localized proteins, when I considered at least one localization correctly matched. Furthermore, I built a localization-guided functional interaction network of the human proteome and mapped known disease associations within this network. I found that proteins associated with same disease have a tendency to localize to the same subcellular compartment. This consensus localization prediction method could be an effective way to identify novel disease associated genes using comprehensive subcellular localization information.
Second, I investigated the relationship between disease-associated proteins and their subcellular localizations. I hypothesized that proteins associated with phenotypically similar diseases are more likely to be connected via subcellular localization. From the analysis of 1284 genetic diseases, I constructed a human Disease-associated Protein and subcellular Localization (DPL) matrix and found that diseases have their unique subcellular localization profiles. Furthermore, I explored the impact of proteins subcellular localization on comorbidity. I found that there is a positive correlation between subcellular localization similarity and comorbidity tendency. Using the subcellular localization, I explained 7,584 disease pairs which had not been identified using other molecular connections such as shared genes or protein-protein interactions (PPIs). These results demonstrated a direct correlation between protein subcellular localization and disease associations and suggested novel approach to understand the mechanism of human disease progression.
Finally, I proposed a novel data-integrative approach to discover human genes required for the virus pathogenesis, specifically hepatitis C virus (HCV). To do this, I integrated information from HCV-human protein interactions, co-expression with human genes related to HCV infection, and association with the protein interaction network of tight junction-tetraspanin web. Moreover, I experimentally validated the newly identified proteins by effects of knocking down the expression of the genes on HCV infection and confirmed five novel genes are involved in the early steps of HCV infections by knocking down their expression using small interfering RNA. These results demonstrated the usefulness of our data-integrative approach combined with experimental validation in the identifying genes that involve in the viral infection process.

목차 (Table of Contents)

• ABSTRACT 1
• CONTENTS 4
• ABBREVIATION 6
• LIST OF FIGURES 7
• Chapter 1. General Introduction 9
• 1-1. Protein localization analysis 9
• 1-2. Disease gene identification through protein subcellular localization 9
• 1-3. Molecular connections of phenotypically related disease pairs 10
• 1-4. Systems biology approach to infectious diseases 11
• Chapter 2. Construction of functional interaction networks through consensus localization predictions of the human proteome 12
• 2-1. INTRODUCTION 12
• 2-2. RESULTS AND DISCUSSION 14
• 2-3. CONCLUSIONS 21
• 2-4. MATERIALS AND METHODS 22
• 2-5. FIGURES 28
• Chapter 3. Protein localization as a principal feature of the etiology and comorbidity of genetic diseases 36
• 3-1. INTRODUCTION 36
• 3-2. RESULTS 38
• 3-3. DISCUSSION 45
• 3-4. MATERIALS AND METHODS 49
• 3-5. FIGURES 54
• Chapter 4. Discovery of cellular proteins required for early step of HCV infection using integrative genomics 69
• 4-1. INTRODUCTION 69
• 4-2. RESULTS 71
• 4-3. DISCUSSION 77
• 4-4. MATERIALS AND METHODS 80
• 4-5. FIGURES 84
• REFERENCES 93
• SUMMARY IN KOREAN 104
• CURRICULUM VITAE 106
• ACKNOWLEDGEMENT 108