Exploring Site-Specific N‑Glycosylation Microheterogeneity of Haptoglobin Using Glycopeptide CID Tandem Mass Spectra and Glycan Database Search

Glycosylation is a common protein
modification with a significant
role in many vital cellular processes and human diseases, making the
characterization of protein-attached glycan structures important for
understanding cell biology and disease processes. Direct analysis
of protein N-glycosylation by tandem mass spectrometry of glycopeptides
promises site-specific elucidation of N-glycan microheterogeneity,
something that detached N-glycan and deglycosylated peptide analyses
cannot provide. However, successful implementation of direct N-glycopeptide
analysis by tandem mass spectrometry remains a challenge. In this
work, we consider algorithmic techniques for the analysis of LC–MS/MS
data acquired from glycopeptide-enriched fractions of enzymatic digests
of purified proteins. We implement a computational strategy that takes
advantage of the properties of CID fragmentation spectra of N-glycopeptides,
matching the MS/MS spectra to peptide-glycan pairs from protein sequences
and glycan structure databases. Significantly, we also propose a novel
false discovery rate estimation technique to estimate and manage the
number of false identifications. We use a human glycoprotein standard,
haptoglobin, digested with trypsin and GluC, enriched for glycopeptides
using HILIC chromatography, and analyzed by LC–MS/MS to demonstrate
our algorithmic strategy and evaluate its performance. Our software,
GlycoPeptideSearch (GPS), assigned glycopeptide identifications to
246 of the spectra at a false discovery rate of 5.58%, identifying
42 distinct haptoglobin peptide-glycan pairs at each of the four haptoglobin
N-linked glycosylation sites. We further demonstrate the effectiveness
of this approach by analyzing plasma-derived haptoglobin, identifying
136 N-linked glycopeptide spectra at a false discovery rate of 0.4%,
representing 15 distinct glycopeptides on at least three of the four
N-linked glycosylation sites. The software, GlycoPeptideSearch, is
available for download from http://edwardslab.bmcb.georgetown.edu/GPS.

糖基化（Glycosylation）是一类常见的蛋白质修饰方式，在诸多关键细胞进程与人类疾病中发挥着重要作用，因此对蛋白质结合聚糖结构的表征，对于理解细胞生物学与疾病发生发展过程至关重要。通过对糖肽（glycopeptides）进行串联质谱（tandem mass spectrometry）分析以直接探究蛋白质N-糖基化，有望实现N-聚糖微观异质性的位点特异性解析——这是游离N-聚糖与去糖基化肽分析无法达成的。然而，依托串联质谱直接开展N-糖肽分析的有效实施仍面临诸多挑战。本研究针对从纯化蛋白质酶解产物的糖肽富集组分中获取的液相色谱-串联质谱（LC-MS/MS）数据，探讨了适配的分析算法技术。我们开发了一套计算策略，该策略利用N-糖肽的碰撞诱导解离（CID）碎裂谱特性，将MS/MS谱图与蛋白质序列及聚糖结构数据库中的肽-聚糖对进行匹配。尤为关键的是，我们还提出了一种全新的错误发现率（false discovery rate）估计方法，用于估算并管控错误鉴定的数量。我们选用经胰蛋白酶（trypsin）与GluC酶解、并通过亲水相互作用色谱（HILIC chromatography）富集糖肽的人源标准糖蛋白——结合珠蛋白（haptoglobin）进行LC-MS/MS分析，以此验证我们的算法策略并评估其性能。我们开发的软件糖肽搜索工具（GlycoPeptideSearch, GPS）以5.58%的错误发现率为阈值，为246条谱图分配了糖肽鉴定结果，在结合珠蛋白的四个N-连接糖基化位点上分别鉴定出42种不同的肽-聚糖对。我们进一步通过分析血浆来源的结合珠蛋白验证了该方法的有效性，以0.4%的错误发现率鉴定出136条N-连接糖肽谱图，对应四个N-连接糖基化位点中至少三个位点上的15种不同糖肽。本研究开发的糖肽搜索工具（GlycoPeptideSearch）可从http://edwardslab.bmcb.georgetown.edu/GPS下载获取。