利用蛋白質交互作用網路配合高轉移動物模式基因表現值探討乳癌轉移機制

Abstract

乳癌為女性常見的癌症，全球每年約有一百萬案例發生。臨床上，若乳癌患者發生遠端轉移，則五年存活率將低於15%。為了探討乳癌轉移成因，目前研究已開發出透過心臟及尾靜脈注射之乳癌轉移動物模型，利用血液循環將癌細胞送至其他器官，但相關研究指出，此方法會省略部分轉移重要步驟(如癌細胞由乳腺侵入血管)。因此，我們發展出乳癌自主性轉移動物模型(由台北醫學大學李嘉華教授開發)，並結合實驗室開發的蛋白質與蛋白質交互作用網路探討乳癌轉移機制。 我們利用乳腺注射(將MDA-MB-231注射至小鼠乳房脂肪墊)建立乳癌自主性轉移動物模式，後續收集乳腺部位的原位腫瘤細胞(稱為primary tumor cells, PTC)，及注射後12週發現的遠端肺轉移腫瘤(稱為lung metastasis-derived mammary tumors, LM1st)。進一步，我們將一代肺轉移細胞注射至其他小鼠乳腺脂肪墊，觀察8週後發現肺轉移，將此肺部轉移細胞取出並稱為二代肺轉移細胞(簡稱LM2nd)。為了探討乳癌轉移機制及各代腫瘤細胞之生物功能差異，我們利用這些細胞之基因表現差異及蛋白質交互作用網路分析。相較於PTC之基因表現，LM1st及LM2nd之共同顯著高表現(基因表現差異2.5倍以上)基因參與在調控細胞發育(TGF-beta signaling pathway)生化途徑，而共同顯著低表現基因則參與免疫反應(Cytokine-cytokine receptor pathway) 生化途徑。進一步透過系統生物角度，發現LM2nd額外調控Gap junction，Calcium signaling pathway，Focal adhesion等轉移相關的生化途徑。 此外，LM2nd對化療藥物(doxorubicin and paclitaxel)有較高的抗藥性，可能源自於UGT2B7及UGT2B10之基因顯著高表現且參與外源性物質降解與代謝相關生化途徑(drug metabolism - cytochrome P450)。我們透過The Cancer Genome Atlas(TCGA)收集14種癌症的臨床資訊以及RNA-seq資料發現，UGT2B7及UGT2B10在TCGA 14種癌症中皆屬基因低表現或不表現，但在我們高轉移動物模行為顯著高表現，此外，我們發現UGT2B7具顯著高表現之病患的五年存活率較差(Log-Rank p=0.005)。根據上述結果，我們相信利用此乳癌自主性轉移動物模型並結合實驗室開發的蛋白質交互作用網路，可以找出造成乳癌轉移的關鍵基因及機制。

Breast invasive carcinoma (BRCA) is one of the most common cancers in women. Clinically, the five-year overall survival rate is less than 15% for the patients with distant metastasis. To understand the metastasis mechanisms in BRCA, several studies have developed the BRCA metastasis models, which introduced the tumor cells into the circulation by intracardiac and tail vein injection. However, previous investigations showed that these models using blood stream inoculation bypass the first key step (i.e. invasion and intravasation). Therefore, we developed a novel spontaneous metastasis mouse model (developed by Dr. Chia-Hwa Lee in Taipei Medical University), and integrated protein-protein interaction networks with gene expressions of this model for revealing the metastasis mechanisms of BRCA. In our model, we injected MDA-MB-231 into the mouse mammary fat pad, and collected cancer cells in the orthotopic position (called primary tumor cells, PTC) and in the lung tissue (called lung metastasis-derived mammary tumors, LM1st) after 12 weeks. Then, we injected LM1st into mammary fat pad of another mouse, and lung metastasis was observed after 8 weeks and tumor cells (called LM2nd) were collected. To explore metastasis mechanisms and compare the differences between PTC, LM1st, and LM2nd, we further integrated those gene expression data with protein-protein interaction network and KEGG pathways. In comparison with PTC, the significantly changed genes (≥2.5-fold change) of LM1st and LM2nd are up-regulated in upstream and downstream of the TGF-beta signaling pathway, respectively, but down-regulated (i.e. LTA and TNF) in the cytokine-cytokine receptor pathway. Furthermore, LM2nd rather than LM1st exhibits up-regulation in the gap junction, calcium signaling pathway, and focal adhesion. In addition, we found that drug resistance for doxorubicin and paclitaxel in LM2nd is significantly higher than PTC, and UGT2B7 and UGT2B10 may play key roles in drug metabolism - cytochrome P450. Based on RNA-Seq data and clinical information of clinical samples in 14 cancer types of The Cancer Genome Atlas, UGT2B7 and UGT2B10 are non-significantly changed or down-regulated between tumor samples and corresponding normal samples, but highly up-regulated in LM2nd. Moreover, the patients with up-regulation (≥2-fold change) of UGT2B7 or UGT2B10 in BRCA exhibited a relatively poor prognosis than the other patients in five-year survival (Log-Rank p=0.005). We believe that integrating protein-protein interaction networks with gene expressions of high-metastatic tumor model can provide the new insights to reveal metastasis mechanisms and develop target therapy of BRCA metastasis.