View : 586 Download: 0

Full metadata record

DC Field Value Language
dc.contributor.advisor최선-
dc.contributor.author이진희-
dc.creator이진희-
dc.date.accessioned2016-08-26T03:08:45Z-
dc.date.available2016-08-26T03:08:45Z-
dc.date.issued2011-
dc.identifier.otherOAK-000000068298-
dc.identifier.urihttps://dspace.ewha.ac.kr/handle/2015.oak/204500-
dc.identifier.urihttp://dcollection.ewha.ac.kr/jsp/common/DcLoOrgPer.jsp?sItemId=000000068298-
dc.description.abstractPart I. Molecular modeling studies for the discovery of selective and potent A₃ adenosine receptor (AR) modulators Adenosine receptors (ARs) belong to G-protein-coupled receptors (GPCRs) family, which consist of four subtypes A1, A2A, A2B, and A₃. Among them, A₃ is the most recently identified adenosine receptor and modulates several physiopathological conditions. A₃ AR agonists are known for the treatment of ischemia and cancer; and its antagonists are being investigated as anti-asthma, anti-glaucoma, and anti-inflammatory agents. In order to discover potent and selective A₃ AR modulators, I constructed the classification models using some machine learning methods and performed the docking studies into homology models. The classification models were built by Laplacian-modified naïve Bayesian, recursive partitioning (RP), and support vector machine (SVM) to distinguish agonists from antagonists of A₃ AR. The models showed excellent results with high accuracy, sensitivity, and specificity. In addition, the 3D structures of A₃ AR were built by homology modeling based on the agonist- and antagonist-bound A2A AR crystal structures. To select the best model for each agonist and antagonist, I performed multiple receptor conformations (MRC) docking with representative agonists and antagonists. Furthermore, induced fit docking was carried out using the selected agonist and antagonist models to understand their interactions at the binding site of A₃ AR. Through the combination of classification modeling and docking study, important insights were obtained for the agonists and antagonists with A₃ AR at the molecular level. These results could be utilized for the design of novel A₃ AR agonists and antagonists. Part II. In silico prediction of CYP inhibition and BBB permeability Adverse side effects of drug-drug interactions induced by human cytochrome P450 (CYP) inhibition and distribution over the blood-brain barrier (BBB) for central nervous system (CNS) drugs are the important in drug discovery. In this study, I built the classification models to predict the inhibition of five major CYP isoforms and BBB permeability using four machine learning methods. Laplacian-modified naïve Bayesian, random forest (RF), recursive partitioning (RP), and support vector machine (SVM) are frequently used as a computational approach owing to their high speed and accuracy. Data sets were collected from the public domain and randomly divided into training and test sets in the ratio of 7 to 3. The classification models were built using VolSurf+, ADRIAND.Code and fingerprints descriptors. The quality of each model was evaluated by the accuracy, sensitivity, specificity, area under the receiver operating characteristic (ROC) scores, and matthews correlation coefficient (MCC), and most of our models successfully classified inhibitors and noninhibitors for each CYP isoform as well as BBB permeability. These models could be useful to predict inhibitors for five major CYP isoforms and BBB permeability in early stage of drug discovery. Part III. Structural insights into Transient Receptor Potential Vanilloid Type 1 (TRPV1) from homology modeling, flexible docking, and mutational studies The transient receptor potential vanilloid subtype 1 (TRPV1) is a non-selective cation channel composed of four monomers with six transmembrane helices (TM1-TM6). TRPV1 is found in the central and peripheral nervous system, and it is an important therapeutic target for pain relief. I describe here the construction of a tetrameric homology model of rat TRPV1 (rTRPV1). My collaborators experimentally evaluated by mutational analysis the contribution of residues of rTRPV1 contributing to ligand binding by the prototypical TRPV1 agonists, capsaicin and resiniferatoxin (RTX). I then performed docking analysis using our homology model. The docking results with capsaicin and RTX showed that our homology model was reliable, affording good agreement with our mutation data. Additionally, the binding mode of a simplified RTX (sRTX) ligand as predicted by the modeling agreed well with those of capsaicin and RTX, accounting for the high binding affinity of the sRTX ligand for TRPV1. Through the homology modeling, docking and mutational studies, I obtained important insights into the ligand-receptor interactions at the molecular level which should prove of value in the design of novel TRPV1 ligands. ;Part I. Molecular modeling studies for the discovery of selective and potent A₃ AR modulators Adenosine receptors은 GPCR superfamily에 속하며, 4개의 subtype (A1, A2A, A2B, A3)으로 구성되어있다. 그 중 가장 최근에 밝혀진 A₃ AR은 다양한 질병치료의 타겟 단백질로 생각되고 있다. A₃ 효능제는 허혈성 심질환과 암 등의 치료제로서 가능성을 가지고 있고, 길항제는 천식, 녹내장, 염증 치료제로 연구 중에 있다. 선택적이며 효능이 뛰어난 A₃ AR의 조절자들을 개발하기 위해, 기계학습(machine learning)방법을 이용한 분류기 모델(classification model)을 만들고 상동수용체 모델을 이용한 도킹 연구를 하였다. Laplacian-modified naïve Bayesian, recursive partitioning, support vector machine의 세 가지 방법을 통한 분류기 모델은 정확성, 민감도, 특수성에서 뛰어난 결과를 보여주었다. 또한, 효현제와 길항제가 결합된 A2A AR 결정구조를 바탕으로 A₃ AR의 상동 수용체 모델들을 만들었다. 효현제와 길항제를 위한 최적의 모델을 선정하기 위해 multiple receptor conformations (MRC) 도킹을 하였다. 선정된 최적의 모델에 대한 대표 효현제와 길항제의 Induced fit 도킹을 통해 A₃ AR의 결합자리에서의 상호작용을 이해하였다. 분류기 모델링과 도킹 연구의 조합을 통해 분자수준에서 A₃ AR과 효현제 및 길항제 간의 중요한 통찰을 얻을 수 있었으며 본 연구 결과는 새로운 A₃ AR 효현제와 길항제의 설계에 이용될 것이다. Part II. In silico prediction of CYP Inhibition and BBB permeability Cytochrome P450 저해로 인해 발생한 약물 상호작용에 따른 부작용과 중추신경계약물에 대한 혈액뇌관문(BBB)의 분포는 신약개발에서 중요하다. 본 연구에서는 주요한 5개의 CYP isoform에 대한 저해 예측 및 BBB 통과를 예측하는 분류기 모델을 Laplacian-modified naïve Bayesian, recursive partitioning, random forest 및 support vector machine의 세 가지 방법을 이용하여 만들었다. Public domain으로부터 data set을 수집하여 training과 test set을 7 대 3으로 나누어 VolSurf+, ADRIANA.Code와 fingerprints descriptor들을 이용하여 모델을 수립하였다. 각각의 모델들은 정확도, 민감도, 특수성 및 ROC, MCC 수치를 이용하여 평가하였다. 본 연구에서는 CYP 저해와 BBB 통과 예측을 위한 모델을 성공적으로 만들었으며 이 모델들은 신약개발의 초기단계에서 유용하게 활용될 것이다. Part III. Structural insights into Transient Receptor Potential Vanilloid Type 1 (TRPV1) from homology modeling, flexible docking, and mutational studies Transient receptor potential vanilloid subtype 1 (TPRV1)은 비선택적 양이온 채널로 6개의 transmembrane(TM1-TM6)을 가진 4개의 monomer로 이루어져 있다. TRPV1은 중추와 말초신경계에서 존재하며 통증완화에 대한 중요한 치료 표적이다. 본 연구에서는 rat TRPV1(rTRPV1)의 tetrameric 상동수용체 모델을 수립하였다. 또한 TRPV1 효현제인 capsaicin과 resiniferatoxin(RTX)의 결합에 기여하는 rTRPV1의 중요 잔기들에 대한 역할을 돌연변이 분석(mutational analysis)을 통해 평가하였다. 본 연구에서 수립한 상동수용체 모델을 이용하여 capsaicin과 RTX의 도킹연구를 수행하였으며, 이들의 도킹 결과는 mutation data와 일치하며 상동수용체 모델이 의미 있음을 뒷받침한다. 더 나아가, 모델링을 통해 예측된 simplified RTX(sRTX)의 결합모드는 capsaicin 및 RTX 유사하였으며, 이는 TRPV1에 대한 sRTX의 높은 결합력을 설명할 수 있다. 상동수용체 모델링, 도킹 및 돌연변이 연구를 통해서 리간드와 수용체간의 상호작용을 분자적 수준에서 이해할 수 있었으며, 새로운 TRPV1 리간드 설계에 기여할 것이다.-
dc.description.tableofcontentsPart I. Molecular modeling studies for the discovery of selective and potent A 3 adenosine receptor (AR) modulators 1 I. INTRODUCTION 3 II. METHODS 9 A. Data set preparation and descriptor selection 9 B. Machine learning methods 10 1. Laplacian-modified nave Bayesian 10 2. Recursive partitioning 10 3. Support vector machine 11 4. Classification model validation 11 C. Homology modeling 13 D. Multiple receptor conformations (MRC) 14 E. Molecular docking 15 F. Induced fit docking (IFD) 16 III. RESULTS AND DISUCSSION 18 A. In silico classification models 18 B. Homology modeling of human A3 AR 36 C. Multiple receptor conformations docking 37 D. Comparison of conformations for agonist- and antagonist-based models 44 E. Induced fit docking 46 IV. CONCLUSIONS 49 V. REFERENCES 50 Part II. In silico prediction of CYP inhibition and BBB permeability 53 I. INTRODUCTION 55 II. METHODS 57 A. Data preparation 57 B. Molecular Descriptors 59 1. VolSurf+ descriptors 59 2. Functional-class fingerprints (FCFPs) 60 3. ADRIANA.Code descriptors 60 C. Machine learning methods 61 1. Laplacian-modified nave Bayesian 61 2. Random forest 61 3. Recursive partitioning 62 4. Support vector machine 62 5. Model evaluation 63 D. PCA and PLS analysis 63 E. Docking study 64 III. RESULTS AND DISCUSSION 65 A. In silico prediction of CYP inhibition 65 1. Single-category classification models 66 2. Multiple-category classification model 72 3. Fingerprint analysis 77 4. Docking study 82 B. In silico classification models of BBB permeability 86 1. In silico classification models 86 2. In silico models using VolSurf+ approach 90 IV. CONCLUSIONS 94 V. REFERENCES 96 Part III. Structural insights into Transient Receptor Potential Vanilloid Type 1 (TRPV1) from homology modeling, flexible docking, and mutational studies 99 I. INTRODUCTION 101 II. METHODS 106 A. Constructs 106 B. Cell culture and transfection 106 C. [3H]RTX Binding Assay 107 D. 45Ca uptake measurement 108 E. Data analysis 109 F. Homology modeling 109 G. Molecular docking 111 III. RESULTS AND DISCUSSION 113 A. Mutation and biological studies 113 B. Homology modeling of rTRPV1 115 C. Flexible docking studies 121 IV. CONCLUSIONS 131 V. REFERENCES 132 국문 초록 135 Publication list 138 ACKNOWLEDGMENTS 139-
dc.formatapplication/pdf-
dc.format.extent3944527 bytes-
dc.languageeng-
dc.publisher이화여자대학교 대학원-
dc.titleMolecular modeling studies for the discovery of potent A₃ AR and TRPV1 modulators and in silico prediction of CYP inhibition and BBB permeability-
dc.typeDoctoral Thesis-
dc.format.pagexv, 140 p.-
dc.identifier.thesisdegreeDoctor-
dc.identifier.major대학원 생명·약학부약학전공-
dc.date.awarded2011. 8-
Appears in Collections:
일반대학원 > 생명·약학부 > Theses_Ph.D
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

BROWSE