In-depth N-glycoproteomic analysis of human blood plasma proteins

Abstract

One of the most common post-translational modifications of proteins is glycosylation. This modification enzymatically attaches a sugar chain, called glycan, to a protein. Glycosylated proteins represent around 45% of the proteins in the eukaryotic domain and about 60% of the proteins secreted to the human blood plasma. Regarding their biosynthesis, glycoproteins have a genetically determined component (the protein) and a variable component (the glycan). The biosynthesis of the latter is highly influenced by biological and environmental factors. There are three protein glycosylation types named N-, O- and C-glycosylation, of which N-glycosylation is the most frequent type. N-linked glycans are more extended than O- and C-glycans, can show various compositions, and are attached to a conserved sequon (Asn-X-Ser/Thr, where X is any amino acid except Pro). As a result, the structural diversity of N-glycans is extraordinary and the number of N-glycosylation sites per protein is variable. Thus, protein N-glycosylation increases significantly the proteoform heterogeneity of a biological system. While the analysis of individual N-glycans generates meaningful information, only the site-specific analysis of N-glycans can reveal their micro-heterogeneity per protein N-glycosylation site. Further, the characterization of site-specific N-glycosylation changes across human traits can provide critical insights, for example about a pathological state. Mass spectrometry (MS) is the preferred tool to achieve the site-specific elucidation of N-glycans via bottom-up glycoproteomics. The two typical approaches for bottom-up glycoproteomics are matrix-assisted laser desorption/ionization time-of-flight tandem mass spectrometry (MALDI-TOF-MS/MS) and liquid chromatography with tandem mass spectrometry (LC-MS/MS). While MALDI-TOF-MS/MS is a high-throughput technology, LC-MS/MS is convenient for analysis of complex protein mixtures. The application of an adequate sample preparation method and the data analysis workflow is essential for the correct site-specific elucidation of N-glycans of a complex protein sample, such as human blood plasma. Human blood plasma or serum is a fluid that networks organs and tissues. In-depth exploration of the composition of human blood glycoproteins is important, as the relation between alterations of N-glycosylation in health and disease has been demonstrated in many instances, such as autoimmune diseases, cancer, congenital disorders, and neurodegenerative disorders. Several strategies have been developed for the site-specific N-glycan characterization of human blood plasma proteins. The primary scope is to use the low-abundant glycoproteome as a source of therapeutic, prognostic, or diagnostic biomolecules for treatment decision-making. However, due to various limitations, analytics struggles to identify the low-abundant human blood plasma N-glycoproteome. In this thesis, a sample preparation workflow was established and applied on a standard human blood plasma sample (pool of normal citrated plasma of at least 20 donors) followed by a newly developed data analysis workflow, to achieve in-depth N-glycoproteomic analysis of the low-abundant human blood plasma glycoproteome. The established sample preparation workflow features two fractionation strategies for expanding the detection range: first, high-abundant protein depletion and second, a protein fractionation by molecular weight. Next, protein fractions are digested to obtain (glyco-)peptides, and the glycopeptides enriched to enhance the chance for their identification. Two high-resolution MS methods are applied for data collection, featuring different collision fragmentation energies. The stepped fragmentation energy gives glycan and peptide composition information, while the low fragmentation energy provides additional structural information. This MS-based measurement strategy follows methodologies previously established in this research group. The developed data analysis workflow comprises three phases: validation, integration of structural features, and re-annotation. First, data validation, based on a newly proposed decision tree, evaluates the coherence between the observed evidence and the proposed glycopeptide spectra match (gPSM). Second, the integration of structural features incorporates N-glycan information observed in the MS spectra acquired at a low fragmentation energy and enables the revision of incorrect gPSM. Finally, the data analysis focuses on re-annotating incorrect gPSMs. Application of this sample preparation workflow not only allows the identification of a high number of N-glycan compositions per N-glycosylation site, but also a significant expansion of the detection range of human blood plasma glycoproteins. While the analysis of untreated blood plasma allowed only the detection of glycoproteins in the range from 1·106 to 1·109 pg/mL, the lower observed protein concentration is 6·103 pg/mL with the newly established sample preparation method. In a few instances even the detection of glycoproteins at a lower abundance, like the identification of N-glycopeptides derived from cysteine-rich secretory protein LCCL domain-containing 1 (8.2 pg/mL), was achieved. After data validation, a total of 7,867 gPSMs were classified and condensed to 1,929 N-glycopeptides referring to 942 different N-glycosylation sites that belong to 805 human glycoproteins. Moreover, the data analysis workflow uncovers frequent pitfalls causing incorrect gPSM assignment, such as the identification of two fucoses instead of one neuraminic acid (NeuAc), which needs attention in further method development. By the integration of MS spectra acquired at low fragmentation energy, rare structures (e.g., N-glycans holding sialic acid-N-acetylhexosamine linkage on one antenna plus an extended N-acetyllactosamine at the second antenna), are additionally identified. Other example of incorrectly annotated gPSMs, which were successfully re-annotated using the developed workflow, are N-glycopeptides bearing three rare N-glycan building blocks (presumably monosaccharides as rare capping sugars). Based on mass accuracy it can be inferred that one of these unknown capping sugars is likely glucuronic acid. However, the two other rare N-glycan building blocks, detected on human prothrombin N-glycans attached to the N121 site, are still unknown. The presence of these rare N-glycan building blocks has been overlooked in earlier studies. Overall, most of the rare N-glycans identified are reported here for the first time. The developed in-depth N-glycoproteomic approach was also applied for the analysis of human serum IgA, a clinically relevant glycoprotein, in order to identify rare N-glycans. Several of these rare N-glycans were reported by glycomic analyses but overlooked using glycoproteomic methods, which hinders the understanding of their role in IgA N-glycosylation. Using the developed strategies, the site-specific identification of sulfated and other rare N-glycans, such as O-acetylated ones, was performed. Optimization of the data analysis setup was conducted with fractionated IgA, and also analyzing two commercial IgA samples with a simple and universal sample preparation method. A key finding of the optimized IgA glycoproteomic analysis is the identification of seven sulfated and four O-acetylated sialylated N-glycans, which includes seven N-glycans not reported before by glycomic analyses. In total eleven rare IgA N-glycans were identified for the first time in a site-specific manner for human serum IgA. Also, this study achieved a micro-heterogeneity analysis of IgA N-glycosylation that now includes O-acetylated and sulfated N-glycans. This analysis reflected that sulfated N-glycans are predominantly at the IgA tailpiece site, while O-acetylated N-glycans are detected ten times less abundant than sulphated N-glycans. In addition, the MS data obtained from the two IgA samples were also screened for other modifications using oxonium marker ions observed with the in-depth N-glycoproteomic analysis of human blood plasma. This allowed the site-specific identification of rare N-glycans bearing hexuronic acid (likely glucuronic acid) as capping attached to contaminating proteins in both commercial IgA samples. In summary, the developed workflow is suitable for expanding our tools to describe the N-glycosylation micro-heterogeneity by the identification of rare N-glycans, elucidation of N-glycan structural features, and detection of glycoproteins in the middle- to low-concentration range of blood plasma proteins. In the future, key features of the data analysis workflow can be transferred to glycoproteomics software tools to generate a more accurate interpretation of MS-based N-glycoproteomic measurements. To broaden the possibilities for glycoproteomics research, the workflow presented can be adapted to better characterize other biological samples, such as glycoproteins isolated from different organisms or other body fluids, such as urine, milk, saliva, or mucosal secretions.