利用微陣列基因晶片建構癌症細胞株之全方位比對系統

細胞株是一種由生物細胞分離出來作為實驗用模式系統的重要工具，其基因表現圖譜理論上和其來源的生物細胞相同，但根據近年的研究發現，細胞株為了要能夠持續不斷地分裂出新的細胞，會使得與細胞增殖相關的基因高度被表現，造成與臨床組織有著不同的表現型，因此科學家在選擇使用最合適的細胞株當作實驗材料之前，必須檢測細胞株與臨床檢體之間基因型和表現型的相似性是很重要的工作。微陣列生物晶片是一項高通量檢測基因表現量的工具，利用此技術可以同時獲得大量基因於細胞株之間的表現差異。但是在現有的線上細胞株資料庫中，大都僅提供細胞株的資料查詢和不同細胞株之間基因表達的比較，缺乏和臨床樣本的基因表現做比較及連結。本研究主要是建構一個以細胞株和臨床組織樣本之生物晶片資料為基礎，進行癌症細胞株的線上基因圖譜比較系統－Carkinos。其三大功能特色有：生物晶片資料檢索、細胞株相似性評估、以及基因標誌(gene signature)瀏覽。生物晶片資料檢索功能提供了在732 個不同的細胞株與2,158 個臨床檢體樣本中，查詢特定基因在不同細胞株及臨床檢體中的表現量，其中作為正規化依據的內部控制參考值，除了透過常見的ACTB 與 GAPDH 二種管家基因之表現量作為正規化的參考值外，並提供了利用最小變異係數作為統計定義上相對穩定的基因當作正規化參考依據。細胞株相似性評估提供了一個量化細胞株相似性的量化模型，藉著將微陣列生物晶片的資料以奇異值分解轉換後降低維度至150 度，在此維度空間中以尤拉距離的方式量化細胞株之間的相似性，並計算其幾何中心，作為使用者挑選細胞株時的參考依據。基因標誌瀏覽透過t 檢定或是單因子變異數分析(one-way ANOVA)取得二群或是多群細胞株及臨床檢體之間的差異比 iii 較。本研究使用的資料和數學模型經過個案討論和文獻的交互參照，確定有相當的準確度：使用生物晶片資料檢索能夠成功辨識屬於乳癌分子亞型（Luminal A）的細胞株T47D 和MCF7，並預測出可能屬於此分子亞型的細胞株EFM-19，同時提出基因RPL41 作為新的管家基因；利用細胞株相似性評估找出分類錯誤的細胞株MDA-MB-435 及OVCAR8，並提出可能分類錯誤的細胞株名單；最後，利用基因標誌瀏覽的功能，提出大細胞肺癌細胞株在建立的過程中產生的基因圖譜改變，提供細胞株和臨床檢體的差異作為研究參考。此外，與現有的線上細胞株資料庫比較，Carkinos 不只有更短的運算時間以及更直覺的操作環境，更將細胞株的資料和活體組織樣本的資料經過整理、連結和比較，是一個作為挑選細胞株以及研究細胞株之差異性的有用工具。

關鍵字

癌症；細胞株；活體細胞樣本；基因表現圖譜比較；細胞株相似性

並列摘要

Cell lines are one of the most important materials served as model systems in the studies of biology, with their characteristics similar or identical to the cells which they were derived from. However, previous studies reported that gene expression profiles of cell lines may change during their establishments, which results in the inappropriate using of cell lines to represent tissues in living organisms. For this reason, it is necessary for scientists to check genetic similarities between cell lines and clinical tissues before deciding the most suitable cell lines as materials, and microarray information helps to distinguish this problem by examining thousands of genes’ expressions at the same time. There are several microarray databases containing cell line information, such as Cancer Cell Line Encyclopedia and Oncomine. However, existing microarray databases about cell line information only provide genetic information of cell lines, without mentioning the relationship between cell lines and clinical samples, and the rigid operation interfaces bring difficulties in acquiring information. By collecting available microarray data of cell lines and clinical tissue samples published online, a web-based comprehensive comparison system named Carkinos is established. Carkinos is composed of microarray data from both cell lines and clinical tissue samples, and then achieves the comparison by three main functions: searching of gene expressions in cell lines and clinical tissue samples, similarity assessments of cell lines to cell lines or tissues, and gene signature explorer. Searching of gene expression in cell lines and clinical tissues helps identify the expansiveness or specificity of genes, in which gene expression values in 732 different cell lines and 2,158 different clinical tissue samples can be seen, and it supports three internal controls, including two commonly used housekeeping genes: ACTB and GAPDH, and one statistical-defined most stable gene calculated by the smallest coefficient of variation from the selected arrays, to normalize the expression values. Similarity assessment provides a quantitative method to measure the resemblance of cell lines by using singular value decomposition transformation of microarray data, decomposing to 150 dimensions and calculating each Euclidean distance as a conjunct geometric center to represent the tissue expression profiles. Gene signature explorer is helpful in comparing two or more groups of cell lines and clinical tissue samples by Student’s t-test or one-way ANOVA, espectively. The quality of data in Carkinos and the precision of these three functions have been tested by several case studies, and the outputs of comparisons are endorsed with published papers. By using the function of searching gene expression in cell lines, breast cancer cell lines of the luminal A subtypes, such as T47D and MCF7 were verified while another cell line EFM-19 was discovered and suggested of the same subtype, and gene RPL41 was found to be a candidate of housekeeping gene; similarity assessment helps to recognize misclassified cell lines such as MDA-MB-435 and OVCAR8; gene signature explorer successfully demonstrated the genetic change during cell line establishment of lung large cell carcinoma. Moreover, in contrast to other existing cell line databases, Carkinos provides not only faster computing speed and more straightforward operation interface, but also the uniqueness in combining and integrating information of cell lines and clinical samples.

並列關鍵字

Cell line ； clinical tissue sample ； microarray ； gene expression profile ； similarity of cell lines

參考文獻

41. Cailleau, R., M. Olive, and Q.V. Cruciger, Long-term human breast carcinoma

1. Freshney, R.I., Culture of Animal Cells: A Manual of Basic Technique, 5th Ed.

2. Ross, D.T., et al., Systematic variation in gene expression patterns in human

thermoradiotherapy according to expression profiles of 35 genes selected by

5. Liotta, L. and E. Petricoin, Molecular profiling of human cancer. Nat Rev Genet,

國際替代計量

利用微陣列基因晶片建構癌症細胞株之全方位比對系統

全文下載

主題瀏覽