- 學位論文
電腦虛擬篩選人類腫瘤治療藥物
In silico Therapeutic Drug Screening for Human Tumor based on the Gene Expression Profiles
摘要
藥物開發是一件花錢又費時的工作。過去二十年，由美國FDA認可的藥物平均不到30個，但是新藥開發所需經費卻與日俱增。為了縮短開發時程，我們以老藥新用的模式，搭配生物資訊技術，進行上市藥物新適應症的探索。 Connectivity Map (CMap) 是一個以基因圖譜為基礎的藥物篩選平台，利用疾病基因標記與藥物基因圖譜，探索疾病與藥物間的關係，並藉由統計方法推論潛力治療藥物。此研究中，我們藉由不同的基因標記選取方法進行肺腺癌潛力藥物預測，並以生物實驗驗證。從CMap的藥物基因圖譜中，比較有效藥物與肺腺癌病人間的基因標記，共找到89個逆轉基因。此外，由於眾多證據顯示癌幹細胞與腫瘤發生、復發及抗藥性有密不可分的關係，因此我們以胚胎幹細胞及癌幹細胞基因標記進行藥物探索，成功找到可扭轉幹細胞基因標記的潛力藥物trifluoperazine。經實驗證實，trifluoperazine可有效抑制腫瘤生長並克服癌幹細胞之抗藥性。 此外，我們設計一個以基因演算法為基礎的合併用藥篩選平台。此研究以藥物作用標的為基礎，計算藥物、疾病間基因表現的變化，提供一個系統化藥物組合評估策略。我們以人類三陰性乳癌為藥物篩選標的，藉由篩選找到一組最佳的藥物組合，並以文獻探討此藥物組合的可行性。未來，將進行生物實驗驗證，並以實驗結果調整參數，提升系統預測準確率。 最後，由於中草藥開發在華人地區已成為重要的研究方向。但目前中草藥缺乏系統化科學驗證與作用機轉評估，因此在品質驗證上存在相當大的挑戰。我們嘗試以基因圖譜探討中草藥品質的穩定性，並以中草藥基因標記進行生物反應途徑分析。分析黃耆萃取物（血寶）生物反應途徑，有近八成與免疫相關。另外，我們還發現血寶可提昇化療的敏感性，增進化療療效。
並列摘要
Drug development is an expensive and time-consuming process. Over the past two decades, the expense of drug development grows annually, but new drugs approved by FDA each year remain at less than 30. In order to reduce development time, we adopted the drug-repurposing strategy in combination with bioinformatics to discover new indications for old drugs. Connectivity Map (CMap) is a gene expression based in silico drug screening platform. Using disease signatures and drug expression profiles, CMap determines connections between disease and drug and predicts potential drugs by statistical methods. In this study, we employed several gene signatures to discover potential drugs for lung adenocarcinoma patients. In addition, we have validated the anticancer effects in lung cancer cells. By comparing the signature of effective drugs with those of lung adenocarcinoma patients, 89 differentially expressed genes were identified that produced a reverse signature. Furthermore, several lines of evidence show that cancer stem cells (CSCs) are associated with tumor initiation, disease relapse, and drug resistance. Therefore, we explored anti-cancer stem cell drugs by using embryonic stem cells (ESCs) and CSCs signature. We have identified trifluoperazine as an anti-lung CSC agent to inhibit tumor growth and overcome chemotherapy resistance. Moreover, we designed a drug combination prediction system using genetic algorithm. Based on the interaction of drug targets, we provided a systematic evaluation strategy for combinatorial drug therapy. We used the triple-negative breast cancer (TNBC) as our target disease for the prediction of possible drug combinations and discuss the effectiveness of the results by literature review. In the future, we will use experimental results to improve the prediction accuracy. Finally, Chinese herbal medicine (CHM) has become an important research field in the Ethnic Chinese community. Due to the lack of systematic scientific evidence and evaluation mechanism, there is a considerable challenge for performing quality assurance on CHM. We investigated quality consistency of CHM by using gene expression profiles and pathway analysis. Pathway analysis of PG2 shows that approximately 82% of the affected pathways are immune-related pathways. In addition, we discovered that PG2 enhances doxorubicin sensitivity in leukemia cancer cells.