利用轉錄基因體分析探究B型肝炎病毒與宿主間之交互作用-長期病毒量追蹤研究

Abstract

B型肝炎病毒(hepatitis B virus，HBV)感染為臨床上重要的持續感染。同為HBV帶原者中，HBV病毒複製程度在人與人間卻有高度異質性存在。HBV病毒量活躍複製會增加罹患肝臟相關疾病的危險性。HBV是一種非細胞致病性的病毒，在持續病毒感染過程中，宿主的免疫反應在肝損傷和病毒控制中扮演關鍵作用。然而，不同的免疫細胞亞群的作用以及各種臨床結果和病毒控制的潛在機制尚不清楚。最近的研究顯示血液基因轉錄體可以系統性探討與病毒感染相關之複雜的免疫反應機轉。本研究以過去所建立的病毒量與臨床指標病例世代研究為基礎，探討慢性B型肝炎病毒自然感染狀況下之免疫細胞組成與長期病毒量趨勢的關聯性。藉由全基因轉錄體了解周邊免疫細胞和其功能間的關係，並找尋和長期病毒量趨勢相關的免疫機轉。本論文共分三部份。

(I) 免疫細胞組成與長期病毒量趨勢 目的: 探究在慢性B型肝炎帶原者中，血液之免疫細胞組成和長期病毒量趨勢間的關聯性。 研究對象與方法: 本研究起始於1989-1992年所建立的B型肝炎表面抗原陽性男性之長期病毒量追蹤世代研究。並於2010-2011年持續追蹤，共有173名研究對象加入研究。 利用流式細胞儀計算周邊血液單核細胞之B細胞，CD4+ T細胞，CD8+ T細胞，自然殺手（NK）細胞，NKT細胞和調節性T細胞（Treg）等六種免疫細胞的比例。利用群組化軌跡建模以了解病毒量隨時間變化的趨勢。相對危險性(OR)和其95%信賴區間(CI)則由羅吉斯迴歸模式估計。 結果: 173個研究對象在1990-2011年追蹤期間，共有1,537個病毒量測量時點，利用群組化軌跡模型將其分成111名(64.2%)持續低病毒量和62名(35.8%)持續高病毒量兩組。持續高病毒量組在在NK細胞的百分比低於持續低病毒量組(中位數為22.52% vs. 29.09%，p=0.06)。調整年齡和HBV基因型等共變項， NK細胞>= 26.6%相較於NK細胞<26.6%的HBV帶原者，持續高病毒量的調整後OR=0.57（95％CI：0.30-1.09; p = 0.0881）。此外，比較高和低NK細胞比例兩組發現，追蹤期間至少有一次ALT異常（≥40U / L）的調整後OR=0.46（95％CI = 0.25-0.84）。然而，持續高和低病毒量兩組間在其他免疫細胞類型的百分比並沒有顯著差異。 結論: 研究結果顯示，慢性HBV感染可能會影響NK細胞的比例。NK細胞比例低與持續高病毒量有關，並可能導致肝病進展。

(II) 周邊免疫細胞之功能基因組分析 目的: 在持續HBV感染中，找尋能夠區分不同免疫細胞組成和長期病毒量軌跡間的基因表達標記。 研究對象與方法: 本研究包含60名研究對象，這些研究對象根據第一部份之173位受試者的長期病毒量資料分成持續低(n=36)和高(n=24)兩組。全基因轉錄體分析使用Affymetrix Human Gene 1.1 ST基因平台。藉由控制年齡的複線性回歸模式來找尋低(相對於高)NK細胞百分比的顯著差異表現基因探針。利用260個模組之模組轉錄譜分析以表徵免疫反應。 結果: 我們發現共有2,183個基因在比較低與高NK細胞百分比後呈現有顯著差異的基因探針。模組轉錄譜分析顯示，這些基因標記在特異的Cytotoxic/ NK Cell相關是有差異之基因模組(M3.6和M8.46)，且大多是上調基因(up-regulated)表現。此外，Cytotoxic/ NK Cell相關的五個基因: FCRL6（p = 0.0237），GPR114（p = 0.0341），KLRF1（p = 0.0175），GZMB（p = 0.0391）和KIR3DL1（p = 0.0256）也和長期病毒量之軌跡模式有關聯性。 結論: 本研究從基因模組方法暗示在持續HBV感染中，NK細胞可能牽涉到抗病毒的機轉。

(III) 長期病毒量趨勢相關基因共基因表現網絡分析 目的: 找尋和持續HBV感染之潛在的免疫反應機轉。 研究對象與方法: 利用複線性回歸模式控制年齡後，找出在持續高(n=24)和低(n=36)病毒量組間有顯著差異表現之基因。基因組富集分析被使用來檢視生物功能路徑。共基因表現模組用來偵測共表現模組和病毒量感染相關的表型之相關性。 結果: 本研究在比較長期病毒量趨勢後，有1,679個顯著差異表現基因被定義出。這些和長期病毒量相關的基因進行基因組富集分析，發現主要富集在在干擾素-γ(Interferon-γ，IFN-γ)(p=0.0181)和單核球(p=0.0039)基因組中。根據共基因表現模組分析，長期病毒量趨勢相關基因可歸類出六個特徵模組。這些特徵模組各別和四個和病毒量感染相關的表型(NK細胞、B細胞、HBV病毒量和B型肝炎自然史階段分期)呈現顯著相關性。其中，和B型肝炎自然史階段分期相關的模組基因最顯著富集於IL-8的訊息傳遞路徑(FDR q value=0.0027)。 結論: 本研究顯示在慢性HBV持續感染者中，和長期病毒量趨勢相關的免疫反應包含IFN-γ, 單核球, NK細胞, 和IL-8訊息傳導路徑。

Hepatitis B virus (HBV) causes clinically important persistent infection. There is remarkable variation in viral replication activity among HBV carriers, and high viral load is a strong predictor associated with hepatitis B progression. It has long been known that HBV is a noncytolytic virus, and host immune response plays a critical role in liver injury and virus control in persistent HBV infection. However, the role of different immune cell subsets and the underlying mechanisms for various clinical outcomes and virus control are not fully understood. Recent studies have shown that blood tanscriptomics can provide insight into the complexity of mechanisms of immune response related to virus infection. Based on an established cohort database of HBV carriers with longitudinally collected viral load and clinical data, this study aimed to examine the association between immune cell profile and the pattern of dynamics of peripheral blood viral load .We also performed a systems biology study using transcriptome analysis to detail the relationship between immune cell profile and function, and to characterize immune pathways for dynamics of viral load. This thesis has three parts.

(I) Relative proportion of immune cell types and dynamic pattern of viral load Specific aim: To examine the association between proportion of immune cell types in blood and dynamic pattern of viral load in chronic HBV carriers. Study subjects and methods: A total of 173 hepatitis B surface antigen-positive men aged 30-65 years old were recruited from a longitudinal viral-load study conducted from 1989 to 1992, with follow-up through 2010 to 2011. Flow cytometry from the peripheral blood mononuclear cells was used to determine proportion of B cells, CD4+ T cells, CD8+ T cells, natural killer (NK) cells, NKT cells, and regulatory T (Treg) cells. Group-based trajectory modeling with longitudinal data was performed to identify patterns of change of viral load over time. Odds ratios (ORs) and their 95% confidence intervals (CIs) were estimated by logistic regression. Results: A total of 173 study subjects contributed 1,537 viral load measurements between 1990 and 2011 and demonstrated high and low trajectory patterns of viral load in 35.8% and 64.2% of study subjects, repsectively. Reduction in the percentage of NK cells was found in high viral load group as compared with low viral load group (median of NK cell frequency: 22.5% vs. 29.1% for high vs. low viral load group; p=0.0608). The adjusted OR of persistent high viral load was 0.57 (95% CI: 0.30-1.09; p=0.0881) for HBV carriers with a NK cell percentage of >= 26.6 as compared with those with a lower percentage of NK cells. In addition, the adjusted OR of subjucts with at least one time abnormal ALT (≥40 U/L) during follow-ups was 0.46 (95%CI=0.25-0.84) between high vs. low NK cell percentage. There were no significant differences between high and low viral load groups with regard to percentages of other immune cell types. Conclusions: The data indicate that chronic HBV infection may have adversely affected the proportion of circulating NK cells which is correlated with persistent high viral load, leading to progression of hepatic disease.

(II) Functional genomic analysis for distinct composition of immune cell subsets Specific aim: To characterize gene expression signature for distinct composition of immune cell types that was associated with longitudinal trajectories of viral load in persistent HBV infection. Study subjects and methods: The study included 60 study subjects who represented low (n=36) and high (n=24) extreme of longitudinal trajectories of viral load in the 173 study subjects with immune cell profiling data. Microarray data were generated using Affymetrix Human Gene 1.1 ST array. Multivariate linear regression model controlling for age was used to identify low (vs. high) NK cell percentage associated differentially expressed genes. A modular transcriptional repertoire of 260 modules was used for analysis to characterize immune response. Results: Comparison of blood transcriptomes between low vs. high NK cell percentage identified 2,183 differentially expressed genes. The modular repertoire analysis revealed that the gene signature associated with two differentially active modules (M3.6 and M8.46) comprising exclusive cytotoxic/NK cell function-related genes, with 46.3% or more of the transcripts forming the module being expressed differentially in low vs. high NK cell percentage. In addition, cytotoxic/NK cell function-related genes FCRL6 (p=0.0237), GPR114 (p=0.0341), KLRF1 (p=0.0175), GZMB (p=0.0391), and KIR3DL1 (p=0.0256) were associated with longitudinal trajectories of viral load. . Conclusions: The findings from a gene module approach provide mechanistic insights into a role of NK cells involving antiviral function in persistent HBV infection. (III) Gene co-expression analysis of differentially expressed genes associated with dynamics of viral load Specific aim: To explore potential pathways of immune response related to gene expressions of persistent HBV infection. Study subjects and methods: We conducted multivariable linear regression model controlling for age to identify the differentally expressed genes (DEGs) between high (n=24) and low (n=36) trajectory patterns of viral load. Gene set enrichment analysis were used to identify the biological pathway.Weighted gene co-expression network analysis (WGCNA) to detect co-expression modules that correlate with potential phenotypes related to virus infection. Results: A total of 1,679 DEGs associated with pattern of long-term viral load were identified. These long-term viral load-related genes showed enrichment for IFN-γ (p=0.0181) and monocyte (p=0.0039) gene sets. By using WGCNA, a total of six co-expression modules were significantly associated with four phenotypes, including NK cell, B cell, HBV viral load and phases of HBV infection. The phase-related module genes are enriched in IL-8 signaling pathways (FDR q value=0.0027). Conclusions: Our findings revealed long-term viral load-related genes associated with IFN-γ, monocyte, NK cell, and IL-8 signaling pathways in persistent HBV infection.