建立電腦虛擬中藥篩選系統及其相關應用研究

曾宇鳳
國立台灣大學生醫電子與資訊學研究所
本計畫目的在建立適合中藥新藥開發之電腦藥物篩選系統，以加速中藥新藥開發。我們將以目前中藥篩選之重點第二型糖尿病及阿茲海默症為例，透過已有之中藥結構活性相關研究，模擬中藥活性分子與受體之作用關係，建立定量構效關係模型(QSAR model)，利用這些模型在相關資料庫中進行虛擬中藥篩選。對於現有受體結構不全之部分，本計畫擬用生物資訊學之結構生物資訊學方法，配合分子動態模擬技術以輔助定量構效關係模型進行虛擬中藥篩選。
本研究中藥篩選的兩大受體目標為(一)針對可用來治療第二型糖尿病的甲型葡萄糖水解酶抑制物(二)可用於治療阿茲海默症之免疫親和素FK506結合蛋白12(FKBP12)為標靶來發展其抑制劑。我們會藉由從文獻中蒐集的80組具甲型葡萄糖?有抑制力的自然有機化合物及由協同主持人所提供經實驗證實具抑制效果之九種樟科楨楠屬植物葉部萃取物作為電腦篩選系統的訓練初始集合，利用其蛋白質結構進行模擬嵌合運算，了解其可能之藥效基團(Pharmacophore)，利用分子動態模擬以及他們的四維結構指紋建立其定量構效關係模型，並使用該模型對天然產物檢索資料庫(DNP)、ZINC自然產物資料庫以及TCMD傳統中藥資料庫進行快速高通量虛擬篩選，找出對於甲型葡萄糖水解酶有效之中草藥先導化合物。就開發阿茲海默症藥物的部分，我們將利用蛋白質嵌合和分子動態模擬的方法確認AICD在FK506結合蛋白12的結合位置，接著將針對這個位置收集會相關結合分子，並延續與開發甲型葡萄糖水解酶抑制物的步驟，建立模型進行快速高通量虛擬篩選，以此二例子確立建立適合中藥新藥開發之電腦藥物篩選系統，以加速中藥新藥開發。
關鍵字：中藥；甲型葡萄糖水解酵素 ；FK506結合蛋白12；DNP天然產物檢索資料庫；ZINC自然產物資料庫；TCMD傳統中藥資料庫；高通量模擬篩選；四維結構指紋；四維定量構效模型

A Virtual Highthroughput Screening System for the Traditional Chinese Medicine

Y. Jane Tseng
Graduate Institute of BEBI, National Taiwan University
The main purpose of this proposal is to develop a virtual highthroughput screening system optimized for Traditional Chinese Medicine (TCM) development. We will use two major disease targets, α-Glucosidase for type II Diabetes mellitu and FKBP12, an immunophilin for Alzheimers disease as example to set up the platform as well as testing the validity for TCM drug screening and development. We will utilize current known TCM structure and activity relationship for the two targets identified to understand the target-ligand interaction to build up quantitative activity relationship models (QSAR model). We will use those models to run virtual high throughput screening to select potential hits for each targets. We will verify and modify our QSAR model and virtual screening platform with the assay results. For current protein without good crystallographic structures on receptors or receptor-ligand complexes, we will utilize structural bioinformatics techniques along with molecular dynamic simulation techniques to simulate the complexes and the protein functions to further assist virtual high throughput screening. 
For α-Glucosidase target, we have collected 80 natural inhibiors for α-Glucosidase and the leaves of 9 Machilus plants from our co-PI with proven inhibition ability as QSAR training set. QSAR models will be used for the high throughput screening on DNP, ZINC, and TCMD database. For Alzheimer screening, we will investigate the binding mechanism of FKBP12 with AICD with docking and molecular dynamic simulation. TCM with stronger bind affinity will be collected. Same computational strategies will be applied as α-glucosidase inhibitors. We will develop and validate our TCM screening systems with these two cases and be certain to provide optimized TCM screening system for TCM drug development.
關鍵字：Computer-aided drug design；Virtual high throughput screening；Traditaional Chinese Medicine (TCM) ；α-Glucosidase, FKBP12；DNP；ZINC；TCMD；CAM； 4D-Fingerpints；4D-QSAR