第一部分：開發以LC-MS定量芳香族胺基酸衍生之腸道菌代謝物的分析方法並應用於帕金森氏症；第二部分：開發以LC-MS定量人類糞便腸道菌代謝物的方法

Abstract

人體的腸道菌落複雜且多樣，會隨著飲食習慣與疾病狀態而變化。腸道菌代謝物（Gut microbial metabolites）不僅存在於人體的腸道中，也會分布於人體的血液循環。過去的研究發現腸道菌的組成與它們的代謝產物常常與疾病的發展有關，也因此越來越多的研究利用液相層析串聯質譜儀（Liquid Chromatography-Mass Spectrometry, LC-MS）分析人體血液與糞便中的腸道菌代謝物。 苯乙醯谷氨醯胺（Phenylacetylglutamine, PAGln）與對甲酚（Para-cresol, p-C）分別是苯丙胺酸（Phenylalanine）與酪胺酸（Tyrosine）的下游代謝物，過去被報導是帕金森氏症（Parkinson’s disease）的危險因子。但是這類非標靶代謝體學（Untargeted metabolomics）的研究無法針對特定代謝物做準確定量，亦無法聚焦於特定的代謝途徑做機轉研究。由於過去未有針對芳香族胺基酸（Aromatic amino acid, AAA）代謝途徑與帕金森氏症的研究，因此在本論文的第一部分，我們開發了定量分析血漿中苯丙胺酸、酪胺酸以及色胺酸（Tryptophan）之代謝產物的LC-MS方法。因為芳香族胺基酸代謝物的結構類似，在層析上會需要使用添加物（Additives）來增加分離效果，但添加物的存在又會對苯乙酸（Phenylacetic acid）、苯丙酸（Phenylpropionic acid）以及對甲酚這三種代謝物造成強烈的離子抑制（Ion suppression）。為了解決上述問題我們藉由調整移動相組成和梯度來最佳化層析效果。在一系列的確效後，我們將此方法應用於測量250位帕金森氏症病人的血漿代謝物濃度，並發現苯丙胺酸到苯乙醯谷氨醯胺的整個代謝途徑皆與帕金森氏症相關，更進一步發現苯乙醯谷氨醯胺與葡萄糖醛對甲酚（Para-cresol glucuronide, p-CG）與帕金森氏症病人的動作障礙嚴重度有顯著正相關。基於我們的研究發現，未來將有利於對帕金森氏症機轉的探討。 吲哚（Indole）與膽酸（Bile acids）是過去最常被討論的腸道菌代謝物之一。腸道菌代謝物於腸道中被製造卻會隨著血液循環去影響人體的各個部位，過去也有許多研究顯示同一種代謝物於血漿與糞便中的分布截然不同。由於過去已有定量血漿中吲哚與膽酸的方法，因此在本論文的第二部分，我們開發了同時定量分析人類糞便中的吲哚代謝物以及膽酸的LC-MS方法。糞便是不均質檢體且糞便中代謝物的濃度差異非常大，這對樣品前處理造成了非常大的挑戰。另外腸道菌的存在使得糞便檢體中的代謝物會持續反應，讓檢體的保存更加不易。為了克服以上困難，我們評估了糞便檢體的乾濕型態、取樣量、稀釋倍數與取樣誤差，做了一系列的最佳化後得出一個標準化的糞便前處理流程。我們利用凍乾法去除糞便水分，不但能校正不同檢體間的水分影響，也更容易讓檢體均質化。20毫克的糞便乾粉足以在一次分析中定量高濃度與大部分低濃度的代謝物，省去了二次稀釋上機的時間，並且不會有太大的取樣誤差。我們的安定性試驗顯示原始糞便必須在一小時內儲存於-20度C以降低定量誤差。 總結本論文開發了定量血漿芳香族胺基酸代謝物的方法，並應用於帕金森氏症的臨床研究；本論文更進一步最佳化糞便檢體的前處理與保存運送流程，讓未來的研究者於定量糞便中的吲哚代謝物以及膽酸時能有一個比較基準。

The human gut microbiota is complex and diverse, and it undergoes changes based on dietary habits and disease states. Gut microbial metabolites exist in the intestinal tract and distribute in the bloodstream. Previous studies have shown that the composition of gut microbiota and their metabolites are often associated with disease development. Therefore, an increasing number of studies have utilized liquid chromatography-mass spectrometry (LC-MS) to analyze gut microbial metabolites in human blood and feces. Phenylacetylglutamine (PAGln) and para-cresol (p-C) are downstream metabolites of phenylalanine and tyrosine, respectively, and have been reported as risk factors for Parkinson's disease (PD). However, untargeted metabolomics studies cannot accurately quantify specific metabolites or investigate specific metabolic pathways. Since there has been limited research on the aromatic amino acid (AAA) metabolic pathway in relation to PD, the first part of this thesis focuses on developing an LC-MS method for the quantitative analysis of phenylalanine, tyrosine, and tryptophan metabolites in human plasma. The structural similarity of AAA metabolites requires additives for enhanced separation; however, these additives can cause strong ion suppression for phenylacetic acid, phenylpropionic acid, and para-cresol. To address this issue, we optimized the chromatographic conditions by adjusting the mobile phase composition and gradient. Following a series of validation experiments, we applied the method to measure the concentrations of AAA metabolites in the plasma of 250 PD patients. We found that the entire pathway from phenylalanine to PAGln is associated with PD, and further analysis showed significant positive correlations between PAGln and para-cresol glucuronide (p-CG) with the severity of motor dysfunction in PD patients. Based on our findings, future studies can further explore the mechanisms of PD. Indole and bile acids are among the most discussed gut microbial metabolites. While these metabolites are produced in the intestine, they can impact various parts of the human body through the bloodstream. Previous studies have shown distinct differences in the distribution of the same metabolites between plasma and feces. As quantitative methods for indole and bile acids in plasma have already been established, the second part of this thesis focuses on developing an LC-MS method for the simultaneous quantification of indole metabolites and bile acids in human feces. Fecal samples pose challenges due to their heterogeneous nature and significant variations in metabolite concentrations. The presence of gut microbiota leads to continuous reactions of metabolites in fecal samples, such as dihydroxylation, oxidation, and epimerization, making sample storage more difficult. To overcome these challenges, we evaluated the effects of sample types (wet or dry), sampling weight, dilution factors, and sampling errors, and developed a standardized fecal sample preparation process. By using lyophilization to remove water in feces, we were able to correct for variations in water content between samples and achieve better sample homogenization. A 20 mg aliquot of lyophilized fecal powder was sufficient for the quantitative analysis of high and most low concentration metabolites in a single run, eliminating the need for second dilution and minimizing sampling errors. Our stability test showed that the crude wet feces must be stored at -20°C within one hour to minimize quantitative errors. In summary, this thesis developed a quantitative method for analyzing plasma AAA metabolites and applied it to clinical research in PD. It further optimized the sample preparation, storage, and transportation processes for fecal samples, providing a protocol for future researchers quantifying indole metabolites and bile acids in feces.