開發樣品濃度校正與鑑別性標誌分析方法於代謝體研究

Abstract

近年來代謝體研究已被廣泛地運用在尋找疾病或是外在環境刺激下的生物指標分子，並且已經成功地用於不同領域上如疾病研究、藥物療效評估、營養介入等。然而，代謝體研究的過程中往往會受到目前仍無法解決，來自於樣品本身以及分析儀器造成的誤差而導致於得到有偏差的分析結果和不正確的資料詮釋。有鑑於此，本研究透過開發樣品濃度校正與鑑別性標誌分析方法分別來達到降低源自樣品本身以及分析儀器造成的誤差來改善代謝體資料的完整性與正確性。 我們首先開發利用基質效應導致的離子抑制現象(matrix-induced ion suppression, MIIS)搭配流動注射與電噴灑離子源質譜儀 (flow injection-electrospray ionization mass spectrometry, FIA-ESI-MS)濃度校正法。我們將等量離子抑制指示劑(ion suppression indicator, ISI)加入樣品基質當中，再觀察不同稀釋倍數的樣品中其ISI訊號與稀釋倍數的關係，並建立檢量線來推算未知濃度之樣品。我們將MIIS 方法做進一步的方法確校，包括線性、精密度與準確度，並將開發的方法應用在研究乳癌病人尿液代謝物的變化與癌細胞生理學。MIIS方法提供便利與準確性能夠提升尿液與細胞的代謝體研究之正確性。 為了實現更有效率的代謝體分析，我們亦建立了利用鑑別性標誌法(differential labeling)針對脂肪酸之分析方法，此一策略能在同一樣品內提供較多資訊而能減少樣品注射針數，有效縮短分析時程並且減少儀器變動對分析結果之影響。我們分別將D0與D3-甲醇與脂肪酸標準品與萃取物在高溫酸性條件下衍生形成D0與D3-脂肪酸甲酯並進一步用氣相層析質譜儀分析之。此一方法可運用在病例對照研究(case-control)上，提供快速的掃瞄兩組間脂肪酸的變化量。為了能夠讓鑑別性標誌法能應用在多組間比較脂肪酸表現量，我們透過使用不同同位素的乙醇(D0, D3 和D5)來衍生出不同同位素的脂肪酸乙酯 (isotopes of fatty acid ethyl ester, iFAEE)。鑑別性標誌法能夠快速且準確地找出有潛力的脂肪酸生物指標並且為一個符合經濟效益的方法。我們透過此一方法來研究抗黴菌藥物voriconazole所引發的肝毒是否與特定脂肪酸的變化有相關。 我們開發出來的MIIS與鑑別性標誌法能夠大幅的減少生物樣品間與實驗儀器過程中所產生的變異，同時亦能縮短分析的時間提升效率。我們期待開發出來的方法可以提升目前各種代謝體研究的品質。

Metabolomics, the latest omics science, has been used to explore biomarkers of external stimuli and applied on a variety of fields. The variations in metabolomics, which might come from either biological samples or instrumental operations, usually lead to biased results and incorrect interpretations are not yet solved. In this dissertation, we will discuss some approaches to improve metabolomics data quality and lower these types of variation for more extensive applications by using sample normalization method and differential labeling approach to minimize biological variation and instrumental variations, respectively. We first reported a matrix-induced ion suppression (MIIS)-based method to normalize concentrations using flow injection analysis coupled with electrospray ionization mass spectrometry (FIA-ESI-MS) and applied it to investigate urinary metabolomics and cellular metabolomics. An ion suppression indicator (ISI) was spiked into the sample matrix, and the intensity of the extracted ion chromatogram (EIC) for ISI in a sample matrix was subtracted by EIC for a blank solution and used to calculate the extent to which the signal was reduced by the matrix. A serial dilution of pooled urine samples or reference cell extracts was used to correlate the concentration and level of ion suppression for ISI. A regression equation was used to estimate the relative concentration of unknown samples. The MIIS method was validated for linearity, precision and accuracy. This study demonstrated that the MIIS method is simple, accurate and can contribute to data integrity in urinary and cellular metabolomics studies and reduce biological variations in metabolomics. Differential labeling techniques could provide more informative results and reduce the number of sample injection that decrease the probability of MS source contamination and decline the instrumental variations. We took the analysis of fatty acids as a demonstration case for differential labeling. The analytical platform of fatty acid analysis mainly relies on complicated chemical derivatization with GC-MS. We developed an effective and accurate comparative fatty acid analysis method using differential labeling of D0- and D3- fatty acid methyl ester (FAMEs) to speed up the metabolic profiling of fatty acids in case-control studies. Consequently, for time-course experiments, we also developed a differential labeling on D0-, D3-, D5- isotopes of fatty acid ethyl ester (iFAEEs). In this part, the differential labeling of fatty acid profiling technique was determined to be fast and accurate and allowed the discovery of potential fatty acid biomarkers in a more economical and efficient manner. The approach was applied to the study of drug-induced hepatotoxicity, which revealed a potential toxicity mechanism and the possibility of using fatty acids as surrogate toxicity markers for drug induced liver injury. Conclusively, sample normalization by the MIIS method could greatly decline the biological variations caused metabolomic analytical bias. Differential labeling introduces sharing of the same matrix which could reduce variations contributed by instrument analysis and greatly reduce analytical time. We anticipate that the developed methods could improve data integrity for various metabolomics studies.