肝癌長期位居我國十大死因的前三名，原位癌開始發生移轉導致癌症的惡化與病人的死亡。本研究中取得了大量的肝癌細胞與正常肝細胞的基因表現模式並經由基因表現分析鎖定616個在有轉移現象的肝癌細胞上大量表現的gene。本研究利用遺傳因子演化保留性的特性，設計了一個方法來檢驗可能的調控因子與被它調控的基因之間的相關聯性。在七組已知的調控因子與其下游基因的樣本群上都能有效被檢驗出來。應用此一方法於616個T3階段有高度表現的基因群中，發現並確認SOX4為最可能的調控因子，並且同時預測在其中有41個基因為其下游被調控基因。經由生物實驗驗證後確認其中31個基因的確被SOX4所調控。由於這31個被SOX4調控的基因在抑制SOX4表現的實驗中在不同的肝癌細胞株中的表現出不一致的行為，本研究進而深入探討SOX4與其他調控因子的組合調控行為。利用SOX4調控目標基因在不同肝癌細胞株中的差異表現以及所有己知轉錄調控因子的接合點與SOX接合點位置上的關係，使用鑑別分析法建立預測模型，並以效能較佳的模型來找尋能有效鑑別不同細胞株的轉錄調控因子並依其在不同細胞株的差異表現上的關聯性進行分類。此方法預測出9個可能的SOX4協力調控因子。其中七個為己知與各種癌症發生相關，在針對尚未被確認與癌症相關以及與SOX4有協力調控關係的Pou2F1的實驗中。確認了SOX4與Pou2F1共同調控了NRP1，並且確認同時抑制SOX4與Pou2F1的表現能最有效的抑制肝癌細胞株的成長。証明此研究中的方法能有效的預測出轉錄調控模組及辦識其調控目標。 In this Dissertation, we propose an analysis flow combines genome databases, analysis tools and methods to identify transcription factor, regulation targets and relationship between transcription factors and targets. We collect gene expression profiles of hepatocellular carcinoma and normal cells. And focus on 616 genes that specific highly expressed in intrahepatic metastasis liver tumor cell. SOX4 is identified as a main factor in liver tumor metastasis. This Dissertation proposes an approach to examine dependency between a transcription factor and feasible regulation targets. Seven positive samples contains transcription factor and its known regulation targets are successful identified. This system can also report a group of genes potentially regulated by the transcription factor. 31 putative targets were verified by biological experiments in this Dissertation. Regulation targets in this Dissertation show differentially activity after SOX4 was knockdown in two HCC cell-lines. An approach is developed to identify regulation partners that cause of the differentially regulation in different cell-lines. Using discriminate analysis to build classification models by differential expression patterns of SOX4 targets in different liver tumor cell-lines and location correlation between known transcription factors binding sites and SOX4 binding sites. And select discriminate transcription factor from well, efficient discriminate models to be the candidates of SOX4 regulation partner. 9 putative SOX4 cofactors are predicted. Seven putative cofactors are known related to various cancers. Pou2F1 is confirmed cooperate with SOX4 to regulate NRP1 and the capability to regulate growth of liver tumor cell-line are also confirmed by biological experiments.
