Recent Advances in Labeling-Based Quantitative Glycomics: From High-Throughput Quantification to Structural Elucidation

Abstract

Glycosylation, a crucial posttranslational modification (PTM), plays important roles in numerous biological processes and is linked to various diseases. Despite its significance, the structural complexity and diversity of glycans present significant challenges for mass spectrometry (MS)-based quantitative analysis. This review aims to provide an in-depth overview of recent advancements in labeling strategies for N-glycomics and O-glycomics, with a specific focus on enhancing the sensitivity, specificity, and throughput of MS analyses. We categorize these advancements into three major areas: (1) the development of isotopic/isobaric labeling techniques that significantly improve multiplexing capacity and throughput for glycan quantification; (2) novel methods that aid in the structural elucidation of complex glycans, particularly sialylated and fucosylated glycans; and (3) labeling techniques that enhance detection ionization efficiency, separation, and sensitivity for matrix-assisted laser desorption/ionization (MALDI)-MS and capillary electrophoresis (CE)-based glycan analysis. In addition, we highlight emerging trends in single-cell glycomics and bioinformatics tools that have the potential to revolutionize glycan quantification. These developments not only expand our understanding of glycan structures and functions but also open new avenues for biomarker discovery and therapeutic applications. Through detailed discussions of methodological advancements, this review underscores the critical role of derivatization methods in advancing glycan identification and quantification.

Keywords: N‐glycan; O‐glycan; derivatization; glycan structure elucidation; sialylated glycans.

FIGURE 1

Mass spectrometry (MS)‐based quantitative glycomics strategies. Left: MS‐based glycomics workflow. Icons were created with Biorender.com. Right: Different isotope labeling methods and the stages of stable isotope introduction. The blue and pink boxes represent the experimental conditions, and the horizontal line represents the pooling of samples from different conditions for subsequent analytical steps. Dashed lines highlight stages where experimental variations may introduce quantification errors.

SCHEME 1

Overview of methods for multiplex glycan quantification. (A) Label‐free quantification involves the comparison of glycan signal intensities across multiple runs. This method often suffers from run‐to‐run variability, which can impact quantitative accuracy. (B) Precursor ion‐based quantification employs stable isotope labeling to introduce a mass difference between light and heavy labeling. This allows for relative quantification of labeled glycans within the same run by comparing extracted ion chromatograms (XIC). Precursor labeling methods include metabolic, enzymatic, and chemical labeling. (C) Mass defect precursor ion‐based quantification utilizes small mDa level mass differences achieved by stable isotopes to distinguish between light and heavy‐labeled glycans in high‐resolution MS1 spectra. This method effectively reduces spectral complexity. The zoomed‐in view highlights how subtle mass differences can be resolved with sufficient MS resolution. (D) Product ion‐based quantification uses isobaric labeling where differently labeled glycans coelute in MS1 with the same nominal mass. Upon fragmentation in MS/MS, distinct reporter ions are released, enabling the quantification of different samples based on reporter ion intensities. This method reduces spectral complexity and allows for high‐throughput multiplexing.

FIGURE 2

Glycan derivatization sites and reagents. (A) Overview of glycan functional groups that serve as reaction sites for derivatization, including the amine group on the glycosylamine, the aldehyde group on the reducing end, the hydroxyl groups, and the carboxylic acid groups on sialic acids. (B) Summary of derivatization reagents and their corresponding reactions, categorized by reaction sites. Examples of commonly used reagents for each reaction are provided.

FIGURE 3

(A) Workflow of matrix‐assisted laser desorption/ionization (MALDI‐MS) analysis of the O‐glycome using superbase releasing and isotopic Girard's reagent P labeling. (B) Scheme representing the O‐glycan release and labeling process. Adapted from Li et al. (2024) with permission. (C) Workflow of the relative quantification method using d0/d5‐BOTC probe labeling for N‐ and O‐glycans analysis in serum based on Pronase E digestion. Adapted from Li et al. (2023) with permission.

FIGURE 4

Graphical representation of different data acquisition strategies. The dotted boxes represent ions selected for tandem MS (MS/MS) isolation. The colored peak represents glycan‐related peaks while the gray peaks represent the background peaks. Panel (A) Data‐dependent acquisition (DDA) shows the selection of the most intense peaks from an MS1 spectrum for subsequent MS/MS analysis. Panel (B) Targeted acquisition illustrates the use of a predefined target list to guide MS/MS acquisition, focusing on specific m/z values over time. Panel (C) Mass‐signature guided acquisition features triplex labeling of glycans, with mass defect signatures guiding the selection of specific peaks in a high‐resolution MS1 spectrum for MS/MS. Panel (D) Data‐independent acquisition (DIA) captures all ions within predefined m/z ranges, with subsequent MS/MS analysis and deconvolution to identify glycan structures, aiming for comprehensive coverage and quantification.

FIGURE 5

Extracted ion chromatograms (EICs) of RapiFluor‐MS (RFMS) labeled glycans released from fetuin separated on a C18 nanocolumn (A–E). Panel (F) depicts the full mass spectrometry (MS) spectrum corresponding to the peak at 53.2 min, and the retention time of the tri‐sialylated glycan is shown in panel E. Adapted from Zhou et al. (2017) with permission. This figure demonstrated that the loss of sialic acid or the GlcNAc‐Gal‐Neu5Ac fragment is a commonly existing phenomenon in MS analysis of reducing end‐labeled glycans. Also, in‐source fragmentation can generate fragment ions with the same m/z as inherent N‐glycan ions, leading to potential misidentification.

FIGURE 6

Reaction scheme for the selective derivatization of sialic acids and subsequent GirP labeling. Adapted from Lageveen‐Kammeijer et al. (2019) with permission. (A) The α2,6‐linked sialic acid reacts to form an ethyl ester in the first step and remains stable throughout the reaction. (B) The α2,3‐linked sialic acid initially dehydrates to form a lactone. Upon the addition of ammonia, the lactone ring opens, resulting in the formation of a stable amide. (C) The reducing ends of all N‐glycans are labeled with Girard's reagent P.

FIGURE 7

Experimental workflow for capillary electrophoresis–mass spectrometry (CE‐MS)‐based N‐glycan profiling of single mammalian cells and blood‐derived isolates. Adapted from Marie et al. (2024) with permission. (A) Schematic representation of the analytical platform developed for in‐capillary sample processing and CE‐MS analysis of N‐glycans released from single mammalian cells and blood‐derived isolates (including model serum proteins, whole plasma, and plasma‐derived EVs). Individual mammalian cells are manually injected using the height difference between the inlet and outlet ends of the CE capillary. The single‐cell plug is positioned between two plugs (1 nL each) of a PNGase F digestion solution. After incubation with PNGase F for 30 min (for blood‐derived isolates) or 1 h (for mammalian cells), the CE and MS electrospray voltages are applied for label‐free CE‐MS analysis of released N‐glycans. (B) Representative image of a HeLa cell suspension droplet used for single‐cell loading (n = 5 technical replicates). (C) Overlay of bright‐field and fluorescence images showing a single HeLa cell (top view) and three HeLa cells (bottom view) loaded into the CE capillary (n = 3 technical replicates). (D) Comparative bright‐field (left) and fluorescence (right) images displaying cell integrity before (T0) and after (T1) the in‐capillary deglycosylation step with PNGase F (n = 3 technical replicates).

FIGURE 8

Predicting glycan structure using deep learning. Adapted from Urban et al. (2024) with permission. (A, B) Overview of the curated dataset of glycomics liquid chromatography coupled with tandem mass spectrometry (LC‐MS/MS), categorized by glycan class (A) and source (B). Diagonal bars represent data obtained in positive ion mode. The numbers indicate spectra with annotations. (C) Schematic diagram of the CandyCrunch model architecture. (D) Workflow for curating glycan predictions, from raw data files to the final output table. (E) Evaluation of Top‐1 accuracy on the independent test set at various levels of resolution. (F) Visualization of learned representations of all spectra in the test set using t‐distributed stochastic neighbor embedding (t‐SNE), colored by glycan class. Examples include illustrations of their glycan structures. (G) Excerpt from a sample prediction output. (H) Proportional Venn diagram comparing CandyCrunch and Glycoforest, with the latter not used for training CandyCrunch but used in its development. The diagram shows topologies that match those identified by a human annotator for each model. Glycoforest does not output full structures.