系統性探索躁鬱症躁期發作的潛在轉錄體生物標記

Abstract

躁鬱症為一高度遺傳之精神疾患，於全球疾病負擔排行中位居前位，具有高頻率的反覆發作以及慢性進程的特性。目前對於病人急性發作之機制尚未明瞭，因此仍缺乏穩定的生物標記，來幫助患者進行發作的提早偵測，並提供治療效果的評估。尤其針對「躁症發作」與「緩解期」之間，病人狀態之基因表現變化差異的研究相當稀少。近年來，由於次世代定序的發展以及非編碼核糖核酸的發現，讓轉錄組及表基因組研究領域皆邁向新的時代。此博士論文是針對躁鬱症患者進行一單純病例研究，當病患急性「躁期發作」時開始進行追蹤直到「緩解」，同時於此兩時間點收集周邊血液、藥物資料及相關臨床資訊，並利用基因表現微陣列、核糖核酸定序以及簡化代表性重亞硫酸鹽定序多種跨平台之技術來針對同一人之兩時間點的「轉錄組」及「表基因組」資料進行分析比較，並找出與「躁期發作」相關之潛在「疾病狀態標記」。本論文之相關結果大致上可分為三部分，分別如下所述。於第一部分的轉錄組分析結果中，先利用「差異表達基因分析」，我們不僅發現了與躁期發作的多個「編碼核糖核酸」，更發現了更多種類之「非編碼核糖核酸」，其中尤以「長鍊非編碼核糖核酸」居多。我們更利用過去躁鬱症全基因體關聯分析所發現之訊號進行「富集分析」用以證明，初步所發現與「躁期發作」相關之「疾病狀態標記」與「疾病特徵標記」並不相同。最後我們使用「網絡分析」結合「生物路徑分析」來找出與「躁期發作」相關之生物網絡。於結果中可發現，同時結合編碼及非編碼的資訊時，可發現「躁期發作」與「免疫系統」調控相關，尤以「細胞激素與受體之間的交互作用」為主。 第二部分，我們則要針對第一部分所發現的「非編碼核糖核酸」來預測其可能具有的分子功能及調控目標。我們先以過去躁鬱症全基因體關聯分析所發現之訊號資料重新進行針對「非編碼核糖核酸」的基因註釋，接著利用多種生物資訊學及線上資料庫進行電腦預測流程之建立，並以實驗驗證完成後，將之套用於第一部分發現的結果。其流程包含：利用「非編碼核糖核酸」於腦部及其他人體組織的表現量以及腦部發展過程中的表現軌跡進行篩選、「長鏈非編碼核糖核酸」以及「微小核糖核酸」的交互作用、潛在的轉錄因子調控以及共表現網絡分析中的功能路徑預測。經細胞培養及相關實驗驗證後，證明了此系統之可行性。而針對第一部分所發現的56個「非編碼核糖核酸」進行預測後，我們發現共有10個「非編碼核糖核酸」可通過表現量之篩選，且其中2個「長鍊非編碼核糖核酸」可顯著與一「微小核糖核酸」hsa-miR-147a產生交互作用。其潛在調控目標基因也都與免疫功能的調控相關，呼應了第一部分所發現的結果。. 第三部分我們則會針對第一部分所發現的編碼核糖核酸並結合「表基因組」的資料來進行分析，並嘗試控制「細胞異質性」以及潛在的「藥物影響」。我們於結果中可發現於「躁期發作」期間，基因體上並沒有出現明顯的「甲基化」區域，而經過控制「細胞異質性」後，「差異表達基因」以及「差異甲基化基因」所影響的生物路徑不盡相同。「差異表達基因」經過校正「細胞異質性」後，多與「能量代謝」及「高基氏體」相關生物路徑相關；而「差異甲基化基因」則多與「訊號傳遞」及「鈣離子訊號」生物路徑相關。而結合了兩者之資訊之後，顯著的生物路徑變為「提升細胞溶質的鈣濃度」。而在控制潛在藥物影響方面，我們在眾多藥物資料中，僅發現抗精神疾病藥物「氟哌啶醇」所影響之基因，與發現的「差異甲基化基因」中有多個重複，且與「DR3、DR4/5死亡受體調控的細胞凋亡」相關，間接應證了「氟哌啶醇」可能分子機制及分子毒性。 綜合以上所述，我們利用單純病例追蹤研究，結合多種平台之技術及合適的分析，以減少個體間差異所造成的干擾，來發掘與「躁期發作」相關之生物標記，並屏除藥物可能之影響。我們相信上述之結果，將能夠更進一步讓我們了解躁鬱症躁期之發作機制，並提供了一嶄新的角度，來發掘「躁鬱症」新型治療方法及躁期發作前評估的可能性，且作為轉譯精神醫學的重要基礎。

Bipolar disorder (BD) is a highly heritable psychiatric disorder. BD is among the top list of disease burden worldwide and is well known for its recurrent episodes and long-term chronic courses. So far, we do not know the exact mechanism of BD episodes onset. Therefore, there are no reliable biomarkers for the prediction of manic episode or treatment evaluation. Very few studies focused on the disease course of manic episode. Recently, with the development of next generation sequencing (NGS) and the findings of non-coding RNAs, transcriptome and epigenome research have been entered a new era. This dissertation aimed to conduct a BD case-only study with the follow-up between the two time-points, the acute mania and remission of BD patients. We collected medications, clinical information and peripheral blood of the two time-points for intra-individual comparisons for BD patients. We used a combination of techniques from different platforms, including microarray, RNA-Sequencing and reduced representation bisulfite sequencing, to analyze the manic-state related transcriptome and epigenome data for the exploration of potential “state-markers” that are related to manic episodes. The contents of the dissertation could be separated into three parts as follows. In the first part, we performed transcriptome analysis to identify differentially expressed gene (DEG) as the state-markers. We found not only numerous mania-related coding genes (ncRNAs) but also lots of non-coding RNAs, especially long non-coding RNAs (lncRNAs). We further used the BD-associated genome-wide association studies signals for enrichment analysis and demonstrated that these “state-markers” of mania were different from the “trait-markers” that were associated with BD diagnosis. Lastly, with network and pathway analyses, we found that these mania-related coding- and non-coding genes were enriched in immune-related pathways, especially cytokine-cytokine receptor interaction. In the second part, we focused on the potential molecular functions and regulatory targets of the manic-state related ncRNAs from the results of the first part. We first used a set of BD-associated GWAS for ncRNAs re-annotation. We then constructed an in silico pipeline, composed of multiple bioinformatics algorithms and public databases, for ncRNAs functional prediction, following with experimental validation. Moreover, we used this pipeline to annotate and predict ncRNAs functional for the results reported in the first part. The processing steps in the pipeline included to filter the ncRNAs with gene expression values across various tissues, exploit the gene expression trajectories during brain development, examine the potential interaction between lncRNAs and microRNAs, and conduct transcription-factors enrichment analysis and pathway analysis based on co-expression networks. Results from this pipeline was validated with cell-based assay experiments, and provide the proof-of-principle for the feasibility of using this pipeline in complex traits. In brief, we used 56 non-coding genes from results of the first part, and 10 lncRNAs were selected based on pre-set criteria. Two of them showed significant interaction with hsa-miR-147a. The potential targets and regulatory pathways involved with candidate manic ncRNAs were also immune-related, which are in corresponding to the findings of the first part. In the third part, we focused on the manic-state related coding genes with the combination of methylome data to infer their regulation mechanisms, while adjusted for cell heterogeneity and potential medication effects. We found that there were no global methylation changes across the genome during manic episodes to remission. On the other hand, we reported that the regulatory pathways of the manic-related DEGs and the differentially methylated genes (DMGs) were different. DEGs were especially enriched with energy-related pathways and the Golgi apparatus, after adjusted for cell heterogeneity. DMGs were enriched with signal transduction pathways and calcium signaling. If we focused on the common DMG-DEG pairs, the significantly enriched pathways became elevation of cytosolic calcium levels. Furthermore, among the genes induced by various categories of medications, only the genes induced by the antipsychotic: haloperidol had overlapping genes with DMGs, and were enriched with the induction of apoptosis through DR3 and DR4/5 death receptors, which are potentially corresponded to the pharmacology and potential toxicity effect of the haloperidol. In summary, we explored the potential mania related state-markers under a case-only follow-up study to diminish potential confounding effects caused by inter-individual differences and medication usage. With the combined analyses using data from multiple platforms and resources, our results of the dissertation bring insights about the dynamic genomic alterations for manic episodic status, and provide clues for developing future novel treatments and therapeutic approaches for BD. We hope these results will form important groundwork towards translational psychiatry in the near future.