開發質譜分析流程以探討多重N型及O型醣蛋白定點精準分析技術

Development and applications of advanced mass spectrometry-based analytical workflows for multiple N- and O-glycosylated glycoproteins

簡瑜君(Yu-Chun Chien)

指導教授 ： 邱繼輝

國立臺灣大學/生命科學院/生化科學研究所/博士(2022年)

https://doi.org/10.6342/NTU202200381

摘要

以質譜分析蛋白特異性位點的醣基化，目前仍有許多尚待克服的技術面困難點。就N型醣蛋白而言，最常遇到的問題是軟體錯誤鑑定；O型醣蛋白則是難以精準定位；而多重N型及O型醣蛋白更是缺乏一個定點精準分析的流程，因此本研究致力於研發多重醣基化胜肽的質譜分析方法，並著重於數據深度分析，以利未來有效應用於醣蛋白質體學。
近年來，由新型冠狀病毒（SARS-CoV-2）引起的嚴重特殊傳染性肺炎（COVID-19）擴散於世界各地，目前已知新型冠狀病毒的棘蛋白結構、抗原性、功能等會受特異性醣基化影響；而貓冠狀病毒中的貓傳染性腹膜炎病毒 (Feline infectious peritonitis virus；FIPV)也會引起致死性貓傳染性腹膜炎 (FIP)。先前已利用軟體 (Byonic)分析過新型冠狀病毒及貓傳染性腹膜炎病毒的棘蛋白所產生的醣胜肽，為了進一步了解特異性位點的醣基化，本研究導入不同原理的分析軟體 (pGlyco3)以建立更值得信賴的定性及定量流程。藉由門檻設定篩選後，兩種軟體皆鑑定到的醣胜肽最為可信，由其中一種軟體鑑定到的醣胜肽次之，而利用滯留時間分組可更進一步校正醣的異構物以降低偽陽性的鑑定結果；另外，大部分的定量分析結果一致，但微量的醣胜肽在兩種分析軟體並不完全一致，則需進一步驗證。
先前已知蛋白酪氨酸磷酸酶A型受體 (Receptor-like protein-tyrosine phosphatase alpha；PTPRA) 的膜外區域只有122個胺基酸，卻是個高度N型醣基化的蛋白。缺乏PTPRA的成纖維細胞會減少其訊號傳導、細胞遷移及轉移的能力，但此能力是否受醣基化影響，且其N型特異性位點及是否有高度O型醣基化位點仍未知。為進一步了解PTPRA的特異性醣位點，在此利用PTPRA-Fc融合蛋白建立多重N型及O型醣蛋白的分析方法。首先，利用去N型及去O型醣蛋白來評估不同軟體對於多重N型及O型醣胜肽的分析能力。從去O型醣蛋白的醣胜肽分析結果發現Byonic與 pGlyco3 分別鑑定到的非重複醣胜肽總數目相近，但 Byonic 可提供較多的匹配結果，且只有Byonic可以鑑定到N型醣胜肽的N型共有序列上的O型醣基化修飾。相較於Byonic， pGlyco3及O-pair這兩種數據分析軟體則可以鑑定到更多源自去N型醣蛋白的O型醣胜肽，結果顯示突變的去N型醣蛋白有更多的O型醣基化修飾，而以同樣的樣品處理方式製備PTPRA-Fc，只能鑑定到少數的醣胜肽片段。因此，接下來以不同蛋白酶降低醣胜肽的長度及複雜度。在胰蛋白酶及絲氨酸蛋白酶的雙作用下，總共鑑定到四個N型及八個O型特異性醣基化位點，但只有其中一個O型位點具有唾液酸修飾；以胰蛋白酶及O型蛋白酶 (OpeRATOR) 雙作用則能找到更多唾液酸修飾的N型醣基化修飾並額外找到19個O型醣基化位點，且得以驗證O型蛋白酶的作用位點傾向在帶有core 1的O型醣基化絲氨酸及蘇氨酸。此外，分析結果也顯示醣胜肽的滯留時間會受到O型醣基化的數量影響。
為了減少偽陰性鑑定，本研究進一步開發一Fishing策略，利用相同胜肽序列會有相近的滯留時間特性，由可信的醣胜肽為基準，找出未鑑定到的醣異構物修飾。由此方法，可找到更多唾液酸化的 PTPRA醣胜肽，雖然因缺乏證據性的斷片離子，無法得知特異性位點，但可藉由滯留時間的特性，推斷出有幾個O型醣基化位點。最後，將成功建立的分析流程應用於全長的 PTPRA，總共可鑑定到兩段不同的胜肽序列，並由酵素的特性推斷出此全長的醣蛋白相較於融合蛋白可能有更多Tn、core 2或唾液酸化 core 1的O型醣基化修飾。
總結而言，由於目前內生性膜蛋白的醣基化修飾分析仍有技術上的高度困難，可藉由蛋白如PTPRA-Fc先建立分析方法，並評估各醣蛋白數據分析軟體的優勢及限制，以利往後更有效分析目前不易鑑定到的多重N型及O型醣胜肽。

關鍵字

醣胜肽

並列摘要

Mass spectroemtry-based protein site-specific glycosylation determination is still technically challenging with many problems unsolved despite recent progress in glycoproteomics, including several newly available computational tools to aid data analysis. Identification of N-glycosylation sites by commonly used software often results in incorrect glycan assignment. Confident site-specific localization of O-glycosylations is difficult, whereas identification of glycopeptides carrying both N- and multiple O-glycans still lacks a proper analytical pipeline. This thesis work aims to capitalize on these advances to develop a better analytical workflow addressing several problems associated with analyzing heavily glycosylated proteins.
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes the Coronavirus disease 19 (COVID-19) pandemic. Site-specific glycosylation of its spike (S) protein is known to impact significantly its structure, antigenicity and functions. Feline infectious peritonitis (FIP), caused by the FIP virus (FIPV), is another highly lethal coronavirus disease in cats. The glycopeptides derived from both the SARS-CoV-2 and FIPV S proteins produced by HEK293 cells have been extensively analyzed by Byonic. In this work, a dual search strategy is applied to the acquired datasets to investigate the feasibility and advantages in establishing a more reliable site-specific glycosylation localization and quantification that was inadequately afforded by the commonly used Byonic software alone. It was shown that glycopeptides identified by both Byonic and the recently upgraded pGlyco3 are the most reliable, whereas those identified by either one only are less so, despite stringent criteria. Grouping of the putatively identified glycopeptides based on their retention time could rectify the correct glycoform and decrease the false-positive results. Most of the abundant glycopeptides identified could be consistently quantified by both the Byologic module in Byos and pGlyco3, while those low-abundant ones yielded a more varied results, which required further manual verification.
The rather small ectodomain of receptor-like protein-tyrosine phosphatase alpha (PTPRA) of 122 amino acids is known to be heavily N-glycosylated but whether its Ser/Thr/Pro-riched sequence is also heavily O-glycosylated has not been investigated. PTPRA deficiency reduces the signaling transduction, invasion, and migration in fibroblast, indicating that PTPRA participates in fibroblast signaling. However, whether the glycans on its ectodomain contribute to any of these and other functions is unclear. To characterize the site-specific glycosylation on PTPRA, an analytical workflow was first established for N- and mutiply O-glycosylated glycopeptides derived from recombinant PTPRA fused with Fc domain. First, the de-N- and de-O-glycopeptides produced by genetic mutation and/or enzyme treatments were used to evaluate the reliability of site-specific glycosylation determination by different software. From de-O-glycosylated peptides analysis, Byonic was found to provide more positive spectral matches than pGlyco3, but the unique peptides identified are comparable in both software. However, only Byonic can identify the glycopeptides with both N- and O-glycan, the latter of which is carried on the consensus NXS/T site. From de-N-glycopeptide analysis, more O-glycopeptides could be identified and site-localized by pGlyco3 and another recently introduced software, O-pair search with MetaMorpheus. Compared to PTPRA de-N-glycosylated by PNGase F, more multiple core 1 O-glycans were found in sequon-mutated non-N-glycosylated PTPRA, consistent with their respective released glycan profiles. Relatively fewer glycopeptides could be similarly identified from fully glycosylated PTPRA. To decrease its complexity, different combinations of proteases were investigated. With trypsin and Glu-C, 4 N- and 8 O-glycosylation sites were found, but only one O-site could be localized with sialylated O-glycans. With trypsin and OpeRATOR, more complex type N-glycans with sialic acids and 19 additional O-glycosylation sites were identified and localized. It can be inferred from the results that the O-protease used preferred S/T sites carrying unmodified core 1 and may not cleave at sites with single GalNAc, core 2, and/or sialylated core 1. Additionally, this work shows that the elution time of the glycopeptide would be shifted much earlier with each additional O-glycan.
To reduce false negatives, a fishing strategy was additionaly developed to expand the identified glycoforms based on the glycopeptides reliably identified in the first attempt. More sialylated glycopeptides could thus be found but not site-localized due to lack of critical fragment ions. Some of these could be attributed to carrying additional O-glycans based on their significantly shifted retention time. Finally, applying all the different strategies established to full length PTPRA led to identification of a few N- and O-glycosylated peptides from the first two tryptic peptides, the O-glycans of which were deduced to comprise more Tn, core 2 or sialylated core1 structures, compared to the ectodomain-Fc constructs. It is concluded that the latter is a valid model to investigate the extent of glycosylation on endogenous membrane receptor, which remains technically difficult.
The concerted analytical workflow developed here including the use of multiple software and a critical appraisal of their respectvie strengths and limitations, will be equally applicable to a wide range of multiply N- and O-glycosylated peptides, which remain under-detected in current glycoproteomics.

並列關鍵字

glycopeptide

參考文獻

Al Barashdi, M. A., Ali, A., McMullin, M. F., Mills, K. (2021). Protein tyrosine phosphatase receptor type C (PTPRC or CD45). J Clin Pathol, 74 (9), 548-552. doi:10.1136/jclinpath-2020-206927

Andersen, J. N., Elson, A., Lammers, R., Romer, J., Clausen, J. T., Moller, K. B., Moller, N. P. (2001). Comparative study of protein tyrosine phosphatase-epsilon isoforms: membrane localization confers specificity in cellular signalling. Biochem J, 354 (Pt 3), 581-590.

Apweiler, R., Hermjakob, H., Sharon, N. (1999). On the frequency of protein glycosylation, as deduced from analysis of the SWISS-PROT database. Biochim Biophys Acta, 1473 (1), 4-8. doi:10.1016/s0304-4165 (99)00165-8

Ardini, E., Agresti, R., Tagliabue, E., Greco, M., Aiello, P., Yang, L. T., . . . Sap, J. (2000). Expression of protein tyrosine phosphatase alpha (RPTPalpha) in human breast cancer correlates with low tumor grade, and inhibits tumor cell growth in vitro and in vivo. Oncogene, 19 (43), 4979-4987. doi:10.1038/sj.onc.1203869

Aschner, Y., Khalifah, A. P., Briones, N., Yamashita, C., Dolgonos, L., Young, S. K., . . . Downey, G. P. (2014). Protein tyrosine phosphatase alpha mediates profibrotic signaling in lung fibroblasts through TGF-beta responsiveness. Am J Pathol, 184 (5), 1489-1502. doi:10.1016/j.ajpath.2014.01.016