自發性肝細胞癌小鼠模型產生腫瘤進展之轉錄體特徵分析

Abstract

原發性肝癌一直是臨床上難以處理的疾病，不僅異質性高，惡性血管大量新生分布也是一大特點，導致復發率高、容易轉移，雖然已經有不少標靶藥物問世，但因為癌細胞本身致癌分子機制可能有差，加上影響的腫瘤微環境不盡相同，不同病人對單一種標靶藥物反應仍有限。隨著次世代定序技術的發展，轉錄體分析就變得重要，透過分析一群細胞內所有表現的RNA，能反映出當下該組織或生物體表達哪些基因，而推測受到甚麼樣的環境刺激，或是尋找感興趣的分子做為實驗目標。小鼠模型是在醫學研究中不可或缺的實驗材料，透過對基因方面和生理方面的操作觀察，更能模擬人體內真實情況。而先前我們實驗室就發展出了能模擬原發性肝癌的自發性小鼠模型，利用臨床上常見致癌突變基因：NRAS、PTEN、TP53，誘發小鼠肝臟細胞癌化。 因此，搭配轉錄體分析，我將針對該模型在腫瘤生長過程，分析腫瘤細胞和血管內皮細胞的基因表達，探討哪些腫瘤微環境的特徵在不同的癌化進程有什麼樣的變化。 透過RNA表達量的分析，我發現了肝臟細胞與血管內皮細胞在三組時期的血管生成訊號路徑、趨化因子、細胞外基質、主要組織相容性複合體、免疫檢查點、其它致癌基因、缺氧誘導因子、第二型干擾素相關的發炎特徵，這些腫瘤微環境有關基因的不同變化。腫瘤細胞於癌化早期會著重表達特定生長路徑，使得其它功能基因減少表達量，但進入晚期後則會開始表現大量基因，其中趨化因子Cxcl9與免疫檢查點配體PD-L1的表現上升，可能代表腫瘤內部有許多免疫細胞，但處於免疫耗竭的狀態。血管內皮細胞在正常組別與癌化早期組別基因表現相近，但到了晚期則會表達大量血管生成訊號路徑、細胞外基質、免疫檢查點相關基因、缺氧誘導因子，代表免疫抑制環境的形成、血管過度生長和大量纖維堆積。其中我再針對腫瘤形成有關的兩個特徵，缺氧與第二型干擾素反應，的相關基因進行分析，來探討兩種細胞在三個時間點的各別變化趨勢。我利用基因集富集分析(GSEA)中的相關基因集(gene set)和前沿子集(leading-edge subset)，整理出腫瘤細胞的缺氧及趨勢都是先下降後上升；而血管內皮細胞的趨勢則是第二型干擾素反應會先上升後下降，對比缺氧反應是先下降後上升。而在這些趨勢中，我也整理出了不但相關而且顯著表現的一群關鍵基因，如Hspa5、Ldha和Cxcl9。最後，使用相似模板預測(Nearest template prediction)，加上臨床上原發性肝癌的基因型分類，我預測出我們的小鼠模型和其中一類人類肝癌最相近，該類的特徵有預後差、細胞週期活躍、容易血管侵襲、免疫耗竭等。 透過轉錄體分析，我分析由特定基因突變的肝癌小鼠模型中，癌症環境特徵的基因變化，並找出高度相關的基因，而這些特徵變化和臨床上肝癌相關，未來可做為實驗目標，應用在更有效的療法開發。

Hepatocellular carcinoma (HCC) is a deadly disease which has caused severe health issue worldwide, contributed by its heterogeneity of essence and aggressive vascular patterns. For high rate of recurrence and metastasis from HCC, there is an urgent need for more efficient treatment to clinically tackle this notorious foe, though numbers of targeted therapies still lack of expectation due to low responsiveness in other patients. As such, molecular mechanism of HCC is worthy of further investigation to develop more efficient treatment, since miracle of target therapies are based on gene or protein expressed in the tumor microenvironment (TME). Transcriptomic analysis is a powerful approach that determines what genes or physiological pathways may occur in a tissue under certain circumstances, and the information can be utilized to make a new hypothesis or target molecule to be further explored. Among multiple tools in transcriptomic analysis, RNA-sequencing (RNA-seq) is the most widely used in biomedical research. Disease animal models have convenience and authenticity, meaning they can simulate the clinical situation on human more realistically than cell lines do. Our lab has genetically developed a brand-new mouse HCC model which targeted on 3 oncogenic genes, namely NRAS, PTEN and TP53, to induce carcinogenesis of hepatic cells and formation of TME in mice spontaneously. To meet the need of deciphering the complex HCC TME, I have analyzed the RNA-seq data from our HCC mouse model in different cell population related to TME across 3 different stages of HCC progression, aiming to find what genes, pathways or features our model may have for further HCC characterization and research. Through analysis of RNA expression levels, I identified distinct temporal changes in TME–related genes in hepatocytes and vascular endothelial cells across three stages, including angiogenic signaling pathways, chemokines, extracellular matrix components, major histocompatibility complex molecules, immune checkpoints, other oncogenes, hypoxia-inducible factors, and type II interferon–tumor inflammatory signature. Tumor cells in the early stage of carcinogenesis preferentially expressed specific growth-related pathways, accompanied by reduced expression of genes involved in other cellular functions. In contrast, during the late stage, a broad range of genes became highly expressed. Notably, increased expression of the chemokine Cxcl9 and the immune checkpoint ligand PD-L1 suggested the presence of abundant yet exhausted immune cells in TME. Vascular endothelial cells exhibited similar gene expression profiles between the normal and early tumor stages. However, in the late stage, they showed upregulation of angiogenic signaling pathways, extracellular matrix components, immune checkpoint–related genes, and hypoxia-inducible factors, indicating the establishment of an immunosuppressive microenvironment, excessive vascular proliferation, and extensive fibrotic accumulation. Focusing on two key tumor-associated features, namely hypoxia and type II interferon responses, I further analyzed their related genes to investigate the dynamic changes in both cell types across the three stages. Using multiple related gene sets and leading-edge subsets from gene set enrichment analysis (GSEA), I found that hypoxia-related trend in tumor cells decreased initially and then increased in the advanced stages, as well as type II interferon responses did. In endothelial cells, type II interferon responses increased first and subsequently declined, whereas hypoxia responses showed an opposite pattern. From these trends, I identified a set of key genes that were both relevant and significantly expressed, including Hspa5, Ldha, and Cxcl9. Eventually, using nearest template prediction (NTP) coupled with clinical HCC molecular subtypes, I found that our mouse model most closely resembled a specific human HCC subtype characterized by poor prognosis, active cell cycle, high vascular invasion, and immune exhaustion. Through transcriptomic analysis of a genetically defined HCC mouse model, this study reveals alterations of TME–associated gene and feature trends that are highly relevant to clinical HCC, providing potential experimental targets for developing more effective therapeutic strategies.